Petrochronology Methods and Applications by Matthew J. Kohn

REVIEWS in MINERALOGY
and GEOCHEMISTRY
Volume 83 2017
Petrochronology:
Methods and Applications
EDITORS
Matthew J. Kohn
Boise State University, USA
Martin Engi & Pierre Lanari

University of Bern, Switzerland
Cover image: Thorium compositional maps of monazite (X-ray counts, cps) and allanite (atoms per formula
unit, apfu). Left: Monazite crystal from a Greater Himalayan Sequence orthogneiss, central Nepal. Ellipses
show locations of ion probe Th–Pb analyses. Core shows oscillations, probably formed during igneous
crystallization. An age of ca. 27 Ma likely reflects prograde metamorphic overprinting (23 Ma age straddles
two chemical domains). An age of 10–11 Ma reflects hydrothermal replacement; based on Corrie and Kohn
(2011). Right: Allanite from the Cima d’Asta pluton, southern Alps (NE Italy). Magmatic allanite core
with oscillatory zoning (dated at 275.0 ± 1.7 Ma (2s) by quadrupole LA-ICP-MS) formed by breakdown
of early-magmatic monazite, preserved as relics (282.3 ± 2.3 Ma). During hydrothermal alteration Th-rich
allanite rim (267.8 ± 4.7 Ma) formed, and its partial breakdown produced a new generation of monazite
(267.0 ± 4.7 Ma); based on Burn (2016), compare Chapter 12, this volume.
Series Editor: Ian Swainson

MINERALOGICAL SOCIETY of AMERICA
GEOCHEMICAL SOCIETY
Reviews in Mineralogy and Geochemistry, Volume 83
Petrochronology: Methods and Applications
ISSN 1529-6466 (print)
ISSN 1943-2666 (online)
ISBN 978-0-939950-05-8
Copyright 2017
The MINERALOGICAL SOCIETY of AMERICA
3635 Concorde Parkway, Suite 500
Chantilly, Virginia, 20151-1125, U.S.A.
www.minsocam.org
The appearance of the code at the bottom of the first page of each chapter in this volume
indicates the copyright owner’s consent that copies of the article can be made for personal
use or internal use or for the personal use or internal use of specific clients, provided
the original publication is cited. The consent is given on the condition, however, that
the copier pay the stated per-copy fee through the Copyright Clearance Center, Inc. for
copying beyond that permitted by Sections 107 or 108 of the U.S. Copyright Law. This
consent does not extend to other types of copying for general distribution, for advertising
or promotional purposes, for creating new collective works, or for resale. For permission
to reprint entire articles in these cases and the like, consult the Administrator of the
Mineralogical Society of America as to the royalty due to the Society.
Petrochonology:
Methods and Applications
83 Reviews in Mineralogy and Geochemistry 83
iii
FROM THE SERIES EDITOR
It has been a pleasure working with the volume editors and authors on this 83rd volume of
Reviews in Mineralogy and Geochemistry. Several chapters have associated supplemental figures
and or tables that can be found at the MSA website. Any future errata will also be posted there.
Ian P. Swainson, Series Editor
Vienna, Austria
PREFACE
“Thy friendship makes us fresh”
Charles, King of France, Act III, Scene III
(Henry VI, Part 1, by William Shakespeare)
Friendship does indeed make us fresh—fresh in our enthusiasm, fresh in our creativity, and
fresh in our collaborative potential. Indeed, it is the growing friendship between petrology and
geochronology that has given rise to the new field of petrochronology. This, in turn, has opened a
new array of methods to investigate the history of the geologic processes that are encoded (oh, so
tantalizingly close!) in rocks, and to develop a broad new array of questions about those processes.
All friendships have their initiations and growth periods, and the origins and evolution
of petrochronology are discussed in some detail in the Introduction. In brief, petrochronology
has been practiced for many decades, but was first labeled in 1997. The seeds for this specific
volume were planted in 2013, watered in 2015, and (we hope) will thrive as a resource through
the coming decades. Indeed, it is hard to envision any future work involving the geochronology
of igneous or metamorphic rocks in the context of tectonics and petrogenesis that is not
somehow petrochronologic.
Our overall goal in this volume is to capture a high-resolution image of the state of
petrochronology during its ascendance, not simply for historical purposes, but rather to
provide a solid foundation for the future. We have striven to corral the very best practitioners
in the field in hopes that their wisdom can help train new generations of petrochronologists,
and inspire them to greater enthusiasm and more diverse research directions. The high quality
of each chapter suggests that we might just have succeeded in this endeavor!
We thank all the authors for their immense investment of time and resources to pull off
the writing of this book. Similarly, the reviewers worked overtime to temper the sometimes
soft metal of each chapter (often on regrettably short notice from the editors…). Ian Swainson
rapidly turned around our manuscripts, hardly giving us rest between submission of final
versions and editing proofs. We appreciate his remarkable attention to detail and unflagging
patience. We also thank our governmental, corporate, society, and university sponsors who
helped support the accompanying short courses: the US National Science Foundation,
Cameca & Nu Instruments, ESI, Selfrag, the European Association of Geochemistry, The
European Geosciences Union, the Geochemical Society, Société Française de Minéralogie
et Cristallographie, Boise State University, and the University of Bern. Last, but not least, we
thank our families and close friends for somehow managing to put up with us over the long
two years that it took to bring about this book.
Matthew J. Kohn, Boise, Idaho, USA
Martin Engi, Bern, Switzerland
Pierre Lanari, Bern, Switzerland
March 2017
1529-6466/17/0083-0000$00.00 (print) http://dx.doi.org/10.2138/rmg.2017.83.0
1943-2666/17/0083-0000$00.00 (online)
Petrochronology
83 Reviews in Mineralogy and Geochemistry 83
TABLE OF CONTENTS
1 Significant Ages—An Introduction to Petrochronology

Martin Engi, Pierre Lanari, Matthew J. Kohn
INTRODUCTION AND SCOPE...............................................................................................1
SIGNIFICANCE OF AGE DATA..............................................................................................2
PETRO-CHRONO-LOGICAL APPROACH AND AMBITION..............................................3
An example: P–T–t path for geodynamic and tectonic modeling..................................4
Methods of choice, choice of methods...........................................................................6
EVOLUTION OR REVOLUTION?..........................................................................................9
ACKNOWLEDGMENTS........................................................................................................10
REFERENCES........................................................................................................................11
2 Phase Relations, Reaction Sequences and Petrochronology

Chris Yakymchuk, Chris Clark, Richard W. White
INTRODUCTION...................................................................................................................13
MAJOR MINERALS...............................................................................................................14
Garnet...........................................................................................................................15
Plagioclase....................................................................................................................15
ACCESSORY MINERALS.....................................................................................................16
Epidote..........................................................................................................................16
Titanite..........................................................................................................................16
Rutile............................................................................................................................16
Zircon...........................................................................................................................18
Monazite.......................................................................................................................19
Xenotime......................................................................................................................21
PHASE EQUILIBRIA MODELLING....................................................................................21
Bulk compositions........................................................................................................22
Computational Methods...............................................................................................22
Modelled P–T paths......................................................................................................23
Modelling suprasolidus zircon and monazite dissolution............................................24
SUBSOLIDUS PHASE RELATIONS AND REACTION SEQUENCES..............................25
Metapelite.....................................................................................................................25
Greywacke....................................................................................................................28
v
Petrochronology ‒ Table of Contents
SUPRASOLIDUS PHASE RELATIONS AND REACTION SEQUENCES.........................30

Metapelite.....................................................................................................................30
Greywacke....................................................................................................................34
Average mid ocean ridge basalt....................................................................................38
Summary of reaction sequence modelling...................................................................40
COMPLICATING FACTORS.................................................................................................41
Changes in effective bulk composition.........................................................................41
Bulk composition and the suprasolidus behaviour of zircon and monazite.................41
Effects of open system behaviour on accessory minerals............................................43
Inclusion/host relationships..........................................................................................44
CONCLUDING REMARKS...................................................................................................44
REFERENCES........................................................................................................................45
3 Local Bulk Composition Effects on Metamorphic Mineral Assemblages

Pierre Lanari, Martin Engi
INTRODUCTION AND SCOPE.............................................................................................55
THEORETICAL BASIS AND LIMIT OF FORWARD THERMODYNAMIC ANALYSIS.56
Gibbs free energy minimization...................................................................................57
The concept of chemical equilibrium and its application to metamorphic rocks.........58
Bulk rock composition vs reactive bulk composition...................................................60
Open questions on bulk composition effects and the size of equilibration volume.....60
PORPHYROBLAST GROWTH.............................................................................................61
Crystal nucleation and growth......................................................................................63
Models of equilibrium and transport control................................................................63
Fractionation of the reactive bulk composition during porphyroblast growth.............65
Equilibrium crystallization models vs fractional crystallization models......................66
Automated fractional crystallization models designed to retrieve P–T paths
from chemical zoning................................................................................................70
Crystal resorption and implications on fractional crystallization models....................70
Timing of porphyroblast growth...................................................................................71
BULK ROCK COMPOSITION EFFECTS ON MELT PRODUCTION
AND ACCESSORY MINERALS.........................................................................................75
LOCAL REACTIONS AND FORMATION OF DOMAINAL ROCKS...............................77
Evidence of local reactions in discrete textural domains.............................................78
Chemical potential gradients and element transfer between domains.........................79
Chemical potential gradient within domains................................................................79
Size of the equilibrium volume versus scale of the model...........................................79
QUANTITATIVE MAPPING OF THE LOCAL BULK COMPOSITION
AS A BASIS FOR MODELING...........................................................................................80
Quantitative X-ray mapping.........................................................................................81
Strategy to derive local bulk composition from X-ray maps using XMapTools........81
Gibbs free energy minimization for the local bulk composition..................................83
Advantages of the micro-mapping approach................................................................83
Potential artifacts affecting the local bulk composition estimates...............................85
Toward systematic quantitative trace element micro-mapping to address
petrochronological problems.....................................................................................89
CONCLUSIONS AND PERSPECTIVES...............................................................................90
vi
APPENDIX 1—LOCAL BULK COMPOSITIONS

FROM OXIDE WEIGHT PERCENTAGE MAPS...............................................................92
REFERENCES........................................................................................................................93
4 Diffusion: Obstacles and Opportunities in Petrochronology

Matthew J. Kohn, Sarah C. Penniston-Dorland
INTRODUCTION AND SCOPE...........................................................................................103

PART 1: DIFFUSION THEORY AND CONTROLS ON DIFFUSIVITY...........................104
Fick’s laws..................................................................................................................104
Multicomponent diffusion..........................................................................................105
Crystal-chemical controls on diffusion......................................................................106
Dependence of D on defects.......................................................................................109
PART 2: PETROCHRONOLOGIC CONCEPTS..................................................................111
Closure temperature...................................................................................................111
Geospeedometry.........................................................................................................113
Diffusion in porous media: fluid–rock interaction.....................................................115
Major element thermometry.......................................................................................116
Trace element thermometry........................................................................................119
Reaction rates.............................................................................................................121
Diffusive resetting of isochrons..................................................................................123
PART 3: EXAMPLES............................................................................................................125
Natural constraints on D.............................................................................................125
Geospeedometry from major element cation zoning.................................................130
Geospeedometry from trace element zoning..............................................................131
Geospeedometry from chronologic zoning................................................................132
Reaction rates.............................................................................................................135
Timescales of magmatic processes from zoning in crystals.......................................137
Magma ascent rates from zoning in glass..................................................................138
Duration of metamorphism........................................................................................140
Fluid–rock interactions...............................................................................................141
FUTURE DIRECTIONS.......................................................................................................142
Boundary conditions...................................................................................................144
Empirical diffusivities................................................................................................144
Controls on diffusivities.............................................................................................144
Multiple element / isotope comparisons......................................................................144
Tectonics.....................................................................................................................145
Thermometry..............................................................................................................145
ACKNOWLEDGMENTS......................................................................................................145
REFERENCES......................................................................................................................145
5 Electron Microprobe Petrochronology

Michael L. Williams, Michael J. Jercinovic, Kevin H. Mahan, Gregory Dumond
INTRODUCTION.................................................................................................................153
REACTION DATING............................................................................................................154
COMPOSITIONAL MAPPING, TRACE ELEMENT ANALYSIS,
AND DATING BY ELECTRON MICROPROBE.............................................................155
Trace-element analysis by electron microprobe—analytical considerations.............159
vii
Analytical strategy......................................................................................................160
Analytical protocol.....................................................................................................162
APPLICATIONS AND EXAMPLES OF EPMA PETROCHRONOLOGY........................165
Textural and compositional correlation between accessory and major phases..........165
Monazite–garnet–yttrium connection—
Example 1: Legs Lake shear zone, Saskatchewan..................................................165
Monazite–garnet–Y connection. Example-2: dating deformation.............................168
Other compositional/textural linkages with silicate assemblages..............................168
EPMA PETROCHRONOLOGY COMBINED WITH ISOTOPIC ANALYSIS...................172
LOW-GRADE METAMORPHISM AND FLUID–ROCK INTERACTION.......................173
FUTURE TRENDS IN EPMA PETROCHRONOLOGY.....................................................174
Electron microprobe instrumental aspects.................................................................174
Thermochemical aspects............................................................................................175
REFERENCES......................................................................................................................177
6 Petrochronology by Laser-Ablation Inductively Coupled

Plasma Mass Spectrometry
Andrew R. C. Kylander-Clark
INTRODUCTION.................................................................................................................183
VIRTUES OF LA-ICPMS ....................................................................................................184
INSTRUMENTATION..........................................................................................................185
Multi-Collector (MC) ICPMS....................................................................................185
Single-collector (SC) ICPMS.....................................................................................185
Laser...........................................................................................................................186
TREATMENT OF UNCERTAINTIES..................................................................................187
LASER ABLATION SPLIT STREAM (LASS)....................................................................187
Plumbing for LASS....................................................................................................191
APPLICATIONS....................................................................................................................192
LASS of metamorphic zircon.....................................................................................193
Single-shot LASS (SS-LASS) analysis of thin metamorphic zircon rims.................193
Depth profiling of rutile..............................................................................................193
Petrochronology of detrital zircon..............................................................................193
Monazite petrochronology.........................................................................................194
Titanite Petrochronology............................................................................................194
CONCLUDING REMARKS.................................................................................................196
REFERENCES......................................................................................................................196
7 Secondary Ionization Mass Spectrometry Analysis in Petrochronology

Axel K. Schmitt, Jorge A. Vazquez

INSTRUMENTATION AND SAMPLE PREPARATION....................................................201
Large-magnet radius ion microprobes for petrochronology, and complementary
instrumentation...........................................................................................................201
viii
Large magnet radius instruments for geochronology and petrochronology...............201

Sample preparation for petrochronology....................................................................206
QUANTITATIVE SIMS ANALYSIS....................................................................................209
Relative sensitivity and instrumental mass fractionation factors...............................209
Two- and three-dimensional relative sensitivity calibrations.....................................210
Specific analytical consideration and strategies for petrochronology........................212
Data reporting for U–Th–Pb geochronology..............................................................217
COMPARING SIMS TO OTHER TECHNIQUES...............................................................220
CASE STUDIES FOR SIMS PETROCHRONOLOGY.......................................................221
OUTLOOK............................................................................................................................223
Improved detection.....................................................................................................223
Ion source developments............................................................................................223
Sample holder design and automation........................................................................224
Final considerations....................................................................................................224
REFERENCES......................................................................................................................225
8 Petrochronology and TIMS

Blair Schoene, Ethan F. Baxter
INTRODUCTION.................................................................................................................231
A BRIEF REVIEW OF TIMS GEOCHRONOLOGY..........................................................232
U–Pb ID-TIMS PETROCHRONOLOGY.............................................................................234
Workflows in petrochronology...................................................................................234
Limits on sample size and precision...........................................................................234
Linking textures with dates in ID-TIMS U–Pb geochronology.................................240
Linking geochemistry with U–Pb ID-TIMS geochronology.....................................242
SAMARIUM–NEODYMIUM ID-TIMS PETROCHRONOLOGY.....................................246
Why ID-TIMS Sm–Nd for Petrochronology?............................................................246
Sample Preparation for Sm–Nd TIMS Petrochronology............................................248
Sm–Nd Age Precision................................................................................................252
THE FUTURE.......................................................................................................................255
REFERENCES......................................................................................................................256
9 Zircon: The Metamorphic Mineral

Daniela Rubatto
INTRODUCTION.................................................................................................................261
PREAMBLE: THE MANY FACES AND NAMES OF METAMORPHIC ZIRCON..........261
PETROGRAPHY OF ZIRCON.............................................................................................264
Textural relationships and inclusions.........................................................................264
Internal zoning............................................................................................................266
Deformation................................................................................................................267
MINERAL CHEMISTRY.....................................................................................................268
Th/U systematics........................................................................................................268
ix
Rare earth elements....................................................................................................271

Ti-in-zircon thermometry...........................................................................................275
ISOTOPE SYSTEMATICS...................................................................................................276
U–Pb isotopes.............................................................................................................277
Lu–Hf isotopes...........................................................................................................279
Oxygen isotopes.........................................................................................................280
PETROGENESIS...................................................................................................................284
Diagenesis and low-T metamorphism........................................................................284
Metamorphic zircon and fluids at sub-solidus conditions..........................................284
Metamorphic zircon and melts...................................................................................286
Zircon forming reactions............................................................................................287
CONCLUSIONS AND OUTLOOK......................................................................................288
REFERENCES......................................................................................................................289
10 Petrochronology of Zircon and Baddeleyite in Igneous Rocks:

Reconstructing Magmatic Processes at High Temporal Resolution
Urs Schaltegger, Joshua H.F.L. Davies
INTRODUCTION.................................................................................................................297
WORKFLOW FOR ZIRCON PETROCHRONOLOGY......................................................298
WORKFLOW FOR BADDELEYITE PETROCHRONOLOGY.........................................301
PETROCHRONOLOGY OF ZIRCON
IN INTERMEDIATE TO FELSIC SYSTEMS..................................................................301
Crystallization of zircon in intermediate-felsic, calc-alkaline melts..........................301
What does zircon chemistry tell us about magmatic processes?................................308
Incremental assembly of magma batches in the upper crust—
wrapping up what we have learned......................................................................314
PETROCHRONOLOGY OF BADDELEYITE AND ZIRCON
IN MAFIC SYSTEMS........................................................................................................316
Crystallization of zircon in mafic (tholeiitic) melts....................................................316
Chemical characteristics of zircon in mafic magmas.................................................317
Baddeleyite geochronology........................................................................................319
OUTLOOK............................................................................................................................322
REFERENCES......................................................................................................................323
11 Hadean Zircon Petrochronology

T. Mark Harrison, Elizabeth A. Bell, Patrick Boehnke
INTRODUCTION.................................................................................................................329
WHY STUDY HADEAN ZIRCONS?..................................................................................330
Modes of investigation...............................................................................................332
JACK HILLS ZIRCONS.......................................................................................................335
Isotopic results............................................................................................................335
Inclusions in zircon....................................................................................................339
Zircon geochemistry...................................................................................................342
x
OTHER WESTERN AUSTRALIAN HADEAN ZIRCON OCCURRENCES.....................347

Mt. Narryer.................................................................................................................347
Churla Wells...............................................................................................................347
Maynard Hills.............................................................................................................347
Mt Alfred....................................................................................................................348
NORTH AMERICAN HADEAN ZIRCON OCCURRENCES............................................348
Northwest Territory, Canada.......................................................................................348
Greenland...................................................................................................................348
ASIAN HADEAN ZIRCON OCCURRENCES...................................................................348
Tibet............................................................................................................................348
North Qinling.............................................................................................................348
North China Craton....................................................................................................348
Southern China...........................................................................................................348
SOUTH AMERICAN HADEAN ZIRCON OCCURRENCES............................................349
Southern Guyana........................................................................................................349
Eastern Brazil.............................................................................................................349
OTHER PROPOSED MECHANISMS FOR FORMING
HADEAN JACK HILLS ZIRCONS...................................................................................349
Icelandic rhyolites......................................................................................................349
Intermediate igneous rocks.........................................................................................350
Mafic igneous rocks....................................................................................................350
Sagduction..................................................................................................................351
Impact melts...............................................................................................................351
Heat pipe tectonics.....................................................................................................352
Terrestrial KREEP......................................................................................................352
Multi-stage scenarios..................................................................................................353
Summary....................................................................................................................353
A LINK TO THE LATE HEAVY BOMBARDMENT?........................................................354
BROADER IMPACTS OF HADEAN ZIRCONS.................................................................355
The role of Hadean zircons in geochemical innovation.............................................355
The role of Hadean zircons in scientific thought........................................................355
SUMMARY...........................................................................................................................356
REFERENCES......................................................................................................................356
12 Petrochronology Based on REE-Minerals:

Monazite, Allanite, Xenotime, Apatite
Martin Engi

REE-minerals.............................................................................................................365
CRYSTAL CHEMISTRY AND CONSEQUENCES............................................................366
Monazite and xenotime..............................................................................................366
Apatite........................................................................................................................369
Allanite.......................................................................................................................369
Sector zoning..............................................................................................................371
Radiation damage.......................................................................................................371
Diffusion and closure temperature.............................................................................371
GEOTHERMOBAROMETRY..............................................................................................373
xi
Monazite and xenotime thermometry.........................................................................373

Monazite–melt thermobarometry...............................................................................374
TRACE ELEMENT GEOCHEMISTRY AND PETROGENESIS.......................................375
Magmatic and partial melting range...........................................................................375
Allanite.......................................................................................................................377
Subsolidus petrogenesis.............................................................................................378
Monazite and allanite.................................................................................................380
Effects of hydrous fluids.............................................................................................386
CHRONOLOGIC SYSTEMS................................................................................................387
Spatial resolution versus age resolution.....................................................................388
U–Th–Pb....................................................................................................................388
Apatite........................................................................................................................389
Allanite.......................................................................................................................389
CASE STUDIES....................................................................................................................392
Selection, purpose......................................................................................................392
Single REE mineral species.......................................................................................392
Diversity of monazite.................................................................................................393
Apatite: testing old ideas in new light........................................................................393
Monadic monazite......................................................................................................395
Combining strengths: petrochronology from monazite plus allanite.........................399
Monazite at extreme conditions..................................................................................401
Allanite: a hot finale...................................................................................................405
REFERENCES......................................................................................................................408
13 Titanite Petrochronology
Matthew J. Kohn

CRYSTAL CHEMISTRY OF TITANITE.............................................................................420
Crystal structure and chemical substitutions..............................................................420
Thermometry and barometry......................................................................................421
Sector zoning..............................................................................................................422
IGNEOUS TITANITE...........................................................................................................423
Petrogenesis................................................................................................................423
Trace element geochemistry.......................................................................................425
METAMORPHIC TITANITE................................................................................................428
Petrogenesis................................................................................................................428
CHRONOLOGIC SYSTEMS................................................................................................428
U–Pb ..........................................................................................................................428
Sm–Nd .......................................................................................................................430
Diffusional biases ......................................................................................................430
EXAMPLES..........................................................................................................................432
Temperature–time histories from single rocks...........................................................432
Temperature–time histories from multiple rocks.......................................................434
Pressure–time histories...............................................................................................435
Igneous processes ......................................................................................................436
xii
Diffusivities ...............................................................................................................437
Chemical domain structure ........................................................................................437
Models .......................................................................................................................437
Rutile activity ............................................................................................................438
Common Pb................................................................................................................438
REFERENCES......................................................................................................................438
14 Petrology and Geochronology of Rutile

Thomas Zack, Ellen Kooijman

RUTILE OCCURRENCE......................................................................................................443
MICROSTRUCTURES OF RUTILE IN METAMORPHIC ROCKS..................................445
ZIRCONIUM-IN-RUTILE THERMOMETRY
AND RUTILE–QUARTZ OXYGEN ISOTOPE THERMOMETRY.................................449
NIOBIUM AND Cr DISTRIBUTION AS SOURCE ROCK INDICATORS.......................452
URANIUM–LEAD GEOCHRONOLOGY...........................................................................453
Uranium, Th and common Pb distribution in rutile...................................................453
Analytics.....................................................................................................................454
Uranium–lead systematics in rutile............................................................................455
Cooling vs formation ages..........................................................................................457
Comparison with U–Pb titanite ages..........................................................................459
CASE STUDY: THE IVREA ZONE.....................................................................................459
CONCLUDING REMARKS AND RECOMMENDATIONS..............................................462
REFERENCES......................................................................................................................463
15 Garnet: A Rock-Forming Mineral Petrochronometer

E.F. Baxter, M.J. Caddick, B. Dragovic
INTRODUCTION.................................................................................................................469
PETRO- OF GARNET..........................................................................................................471
Textures of garnet—provider of tectonic context.......................................................472
Garnet as a pressure and temperature sensor..............................................................474
Garnet as a tracer of reaction pathways and fluid–rock interaction...........................483
CHRONO- OF GARNET......................................................................................................492
Garnet geochronology................................................................................................493
The development of zoned garnet geochronology.....................................................509
Geospeedometry with garnet......................................................................................510
EXAMPLES OF PETRO-CHRONOLOGY OF GARNET..................................................512
Petrochronology of garnet: High pressure crustal metamorphism.............................512
Petrochronology of garnet: Geospeedometry and the timescales of
granulite facies metamorphism................................................................................514
Petrochronology of garnet: Timescales of lower crustal melting...............................514
Petrochronology of garnet: Subduction zone dehydration.........................................516
xiii
Petrochronology of garnet: Collisional tectonics

and inverted metamorphic gradients........................................................................517
OUTLOOK............................................................................................................................518
REFERENCES......................................................................................................................518
16 Chronometry and Speedometry of Magmatic Processes

using Chemical Diffusion in Olivine, Plagioclase and Pyroxenes
Ralf Dohmen, Kathrin Faak, Jon D. Blundy
INTRODUCTION.................................................................................................................535
Types of chronometers...............................................................................................536
Basic approach of diffusion chronometry..................................................................538
OLIVINE...............................................................................................................................540
Diffusion coefficients.................................................................................................541
Determination of magma residence times using Fe–Mg, Ca, Ni, and Mn diffusion.. 541
Growth vs. diffusive zoning: stable isotopes as diffusive fingerprints.......................544
Determination of cooling rates from Fe–Mg zoning in olivine..................................545
Determination of cooling rates from Ca zoning in olivine.........................................548
Determination of cooling rates to constrain thermal structure of a crustal segment..550
Determination of magma ascent rates using H diffusion...........................................550
PLAGIOCLASE....................................................................................................................554
Diffusion coefficients and specifics for trace element diffusion in plagioclase.........555
Determination of magma residence times using Mg or Sr diffusion..........................558
Determination of cooling rates using Mg diffusion...................................................559
PYROXENES........................................................................................................................561
Diffusion coefficients.................................................................................................561
Determination of magma residence times using Fe–Mg diffusion in Opx and Cpx..563
SYNOPSIS/PERSPECTIVES...............................................................................................566
REFERENCES......................................................................................................................568
RiMG Series
HISTORY OF RiMG.............................................................................................................577
HOW TO PUBLISH IN RiMG..............................................................................................577
xiv
Reviews in Mineralogy & Geochemistry
Vol. 83 pp. 1-12, 2017 1
Copyright © Mineralogical Society of America
Significant Ages—An Introduction

to Petrochronology
Martin Engi and Pierre Lanari
Institute of Geological Sciences
University of Bern
Baltzerstrasse 3
3012 Bern
Switzerland
engi@geo.unibe.ch
pierre.lanari@geo.unibe.ch
Matthew J. Kohn
Department of Geosciences
Boise State University
Boise, Idaho, 83725
U.S.A.
mattkohn@boisestate.edu
INTRODUCTION AND SCOPE

Question: Why “Petrochronology”? Why add another term to an already cluttered
scientific lexicon?
Answer: Because petrologists and geochronologists need a term that describes the unique,
distinctive way in which they apply geochronology to the study of igneous and metamorphic
processes. Other terms just won’t do.
Such evolution of language is natural and well-established. For instance, “Geochronology”
was originally coined during the waning stages of the great Age-of-the-Earth debate as a
means of distinguishing timescales relevant to Earth processes from timescales relevant to
humans (Williams 1893). Eighty-eight years later, Berger and York (1981) coined the term
“Thermochronology,” which has evolved as a branch of geochronology aimed at constraining
thermal histories of rocks, where (typically) the thermally activated diffusive loss of a radiogenic
daughter governs the ages we measure. Thermochronology may now be distinguished from
“plain vanilla” geochronology, whose limited purpose, in the words of Reiners et al. (2005),
is “…exclusively to determine a singular absolute stratigraphic or magmatic [or metamorphic]
formation age, with little concern for durations or rates of processes” that give rise to these rocks.
Neither of these terms describes what petrologists do with chronologic data. A single date is
virtually useless in understanding the protracted history of magma crystallization or metamorphic
pressure–temperature evolution. And we are not simply interested in thermal histories, but in
chemical and baric evolution as well. Rather, we petrologists and geochronologists strive to
understand rock-forming processes, and the rates at which they occur, by integrating numerous
ages into the petrologic evolution of a rock. It is within this context that a new discipline,
termed “Petrochronology”, has emerged1. In some sense petrochronology may be considered
the sister of thermochronology: petrochronology typically focuses on the processes leading up
1
Several parts of this introduction are taken from a discussion that took place in the forum GEO-METAMORPHISM in June, 2013.

1943-2666/17/0083-0001$05.00 (online)
2 Engi, Lanari & Kohn
to the formation of igneous and metamorphic rocks—the minerals and textures we observe and
the processes that formed them—whereas thermochronology emphasizes cooling processes
in the wake of igneous, metamorphic, and tectonic events. Typically petrochronology is
“hot”, thermochronology is “cold”. While each field has its unique features, and while their
disciplinary boundaries overlap, each complements the other.
Any rock sample we study, whether igneous, sedimentary, or metamorphic, results from
the transformation of one or more previous rocks. Petrologists and geochemists have found
that such transformation rarely erases a rock’s memory completely, instead most samples
contain relics from more than one stage of their evolution. Whether and how these affect an
age determination is essentially a question of resolution—both spatial and chronometric—i.e.,
of isotopic and chemical analysis. Analytical efforts in petrochronology typically find that
several stages or generations of mineral formation are evident in any single rock sample, in
which case we conclude that such a rock does not have, sensu stricto, one age.
In fact, one is led to wonder what the term “age” may signify in everyday geologic usage.
It might seem clear what is meant, for example, by the age of a basaltic lava flow: the time of
deposition or solidification. But what is the age of a meta-basalt? Does it refer to the point on
the prograde path when its mineralogy and texture would define it as “metamorphic”, and no
longer igneous? The pressure peak? The maximum temperature? The point on the retrograde
path where mineralogy and chemistry no longer change measurably? And by what methods
can that singular metamorphic age be measured? Actually, defining “an” age of a volcanic
rock presents its own problems. How do we choose among the ages of initial melting, magma
movement or rejuvenation, crystallization of antecrysts and duration of residence in a magma
chamber, eruption, or solidification? And what does an age mean for a clastic rock, where
each grain may have a slightly different parent, and materials may be reworked. The concept
of “an age” really makes sense only within a defined petrogenetic context.
This recognition leads us to a practical definition: Petrochronology is the branch of Earth
science that is based on the study of rock samples and that links time (i.e., ages or duration) with
specific rock-forming processes and their physical conditions. Petrochronology is founded in
petrology and geochemistry, which define a petrogenetic context or delimit a specific process,
to which chronometric data are then linked.
SIGNIFICANCE OF AGE DATA

Chronometers essentially rely on the behavior of specific elements, more specifically of
ions or isotopes, in a mineral (or group of minerals) chosen for dating. To assess chronometric
data, two criteria need to be combined: transport properties (diffusivity), and analytical quality.
Diffusion of the elements relevant to the chronometric system in the chosen mineral, say Pb
in zircon or Ar in feldspar, sets basic limits on how a chronometer can be applied and what
meaning the age data have: Where diffusion is relatively fast, the age of a mineral refers to the
time when radiogenic daughter material started to accumulate in that mineral. Such systems
are used for thermochronology, and the resulting cooling age specifies the amount of time that
mineral has remained below a particular temperature, called the closure temperature (Tc). This
age will correspond to the time of formation only if that mineral crystallized (and remained)
at temperatures below Tc or was somehow shielded from diffusive loss of the daughter isotope.
By contrast, in systems where diffusion is extremely slow, commonly for the U–Pb, Th–Pb,
Lu–Hf, or Sm–Nd decay systems, minerals remain closed up to high temperature, and radio-
isotopic dating commonly returns formation ages, i.e., the time of crystallization. For igneous
rocks, this may reflect a pulse of crystallization, catalyzed by transport or degassing events
or possibly related to magma recharge. For metamorphic rocks, it may represent a particular
mineral reaction, an event of fluid infiltration, or deformation-induced recrystallization.
Significant Ages—An Introduction to Petrochronology 3
Typically, spatially resolved age data are combined with compositional data of accessory
and/or major phases to recover specific P–T–t or T–t points and paths. In many cases, data for
individual growth zones of minerals are required and/or several chronometers are combined
to secure the interpretation. Although we commonly focus on collecting ages with high spatial
and chronologic resolution, it is not actually the age that matters most to geologic investigations
but rather the process itself and our ability to characterize rates.
Chronologic quality is a delicate subject in petrochronology because analytical accuracy and
the behavior of elements involved in the system used are both important, and element behavior
is difficult to quantify. Traditionally the quality of an age determination, say for garnet using the
Sm–Nd system, focused on the MSWD or statistical likelihood that all data conform to a single
crystallization event. Today, in part through vastly improved analytical capabilities, we recognize
that protracted growth of garnet should rarely produce a good MSWD (Kohn 2009). Our ability
to measure age differences among different growth zones means that data scatter should nearly
always exceed analytical uncertainties. While this failure of traditional tests of chronologic
significance might undermine some past investigations (there is no single age that we can
assign to such a garnet), it opens a vast new spectrum of scientific inquiry related to the growth
dynamics of minerals, and consequently the processes that drive mineral growth. So, instead of
asking what the age of a pluton is, we might instead ask how rapid is magma transport, what
catalyzes degassing and consequent melt crystallization, and how these are related to magma
recharge. Similarly, in metamorphic petrology and tectonics, we can now ask how rapid was
heating/cooling or burial/exhumation, how are these related to thrusting or extensional processes,
and what does this history imply about large-scale deformation of the crust? Addressing these
types of questions means that a mineral age must be linked to some other petrogenetic indicator,
whether it be other associated minerals of petrogenetic or textural significance, major element
chemistry, trace element patterns, or independently estimated temperature or pressure.
PETRO-CHRONO-LOGICAL APPROACH AND AMBITION

A basic and generic petrochronological approach may be formulated in five steps:
1. Identify one or more specific stage(s) of the metamorphic, magmatic and/or
structural evolution in a given sample based on textural criteria such as overgrowths,
inclusions, fabric alignments, etc. In deformed rocks, it is helpful and often necessary
to work with oriented samples and sections, such that fabric characteristics of local
assemblages can be related to observed fabrics in hand specimens and to meso- to
megascopic observations.
2. Document the phase relations among minerals based on images and compositions as
determined using electron beam methods (scanning electron microscopy: SEM; electron
probe micro-analysis: EPMA), laser ablation–inductively coupled plasma–mass
spectrometry (LA-ICP-MS) and/or secondary ion mass spectrometry (SIMS) analysis.
3. Attempt to relate one or more specific growth zones of a suitable robust chronometer
(e.g., U–Pb, Th–Pb, Lu–Hf, Sm–Nd) to each stage. This step often requires support
from trace element data to verify coexistence.
4. Use thermobarometric techniques such as multi-equilibrium, isochemical phase
diagrams2, Thermoba-Raman-try, and empirical thermobarometry to constrain the
Pressure–temperature (P–T) conditions of local equilibria.
5. If steps 1–4 were successful, use a microdating technique to analyze the isotopic
ratios in each suitably large growth zone. This can be done by in situ analysis or
microdrilling for ID-TIMS.
2
Also known as “pseudosections“, but this term may be misunderstood. Spear FS, Pattison DRM, Cheney JT (2016) The metamorphosis of
metamorphic petrology. Geological Society of America Special Papers 523 doi:10.1130/2016.2523(02)
Several research groups have applied this type of approach. Ambitious applications have
attempted to reconstruct pressure–temperature–deformation-time (P–T–D–t) paths in units
that experienced a complex tectono-metamorphic evolution. As an example of the potential
of petrochronology, it is instructive to summarize results obtained for a specific area, the Dora
Maira terrane in the Western Italian Alps. A variety of techniques have helped elucidate its
evolution, with implications for other fields, from tectonic modeling to experimental petrology.
An example: P–T–t path for geodynamic and tectonic modeling
Prescient researchers realized long ago that P–T–t paths of rocks from different structural
positions in an orogenic belt can constrain models of large-scale tectonic processes (Thompson
and England 1984; Shi and Wang 1987). Theoretical P–T–t paths from numerical models started
appearing before they could be reconstructed from rock samples. Numerical models were first
restricted to one-dimension problems (Oxburgh and Turcotte 1974; England and Thompson
1984), but they rapidly evolved to complex geometry in two-dimensions (Peacock 1990; Ruppel
and Hodges 1994). At the dawn of the 21st century, dynamic thermo-mechanical models were
developed and intensively used to predict theoretical P–T–D–t trajectories of rock units involved
in various scenarios of subduction and collision zones (e.g., Beaumont et al. 2001; Gerya et al.
2002; van Keken et al. 2002, and many others inpired by these). While much progress has been
made since then in investigating geodynamic processes by means of numerical models (Gerya
2011 and references therein), stark differences occur between many model P–T–t trajectories
and the natural record, especially for continental collisions (Kohn 2008) and for prograde
subduction paths at pressures below 2 GPa (Penniston-Dorland et al. 2015).
An excellent example of how P–T–D–t paths may be determined from the study of specific
samples comes for the ultra-high pressure (UHP) slice of Dora-Maira located in the Western
Alps of Italy. This massif is renowned as one of two localities where coesite inclusions were
first discovered in continental rocks, specifically in garnet porphyroblasts (Chopin 1984). As
shown in Figure 1, this terrain now sports one of the best constrained P–T–D–t paths for a
UHP continental fragment in the world. Detailed field studies of the Dora-Maira UHP slice
document it as a coherent piece of continental crust (1 km thick, exposed over 10 by 5 km),
composed of granitic gneiss with intercalated (K- or Na-bearing) whiteschists, metabasites
(calcsilicate eclogites and eclogite boudins), and marbles (Vialon 1966; Chopin et al. 1991).
Several studies have demonstrated that this tectonic slice was subducted to depths greater than
~100 km during the early stages of Alpine collision (e.g., Chopin et al. 1991; Chopin and
Schertl 1999; Rubatto and Hermann 2001; Hermann 2003). P–T conditions based on several
prograde and retrograde reactions and local assemblages were derived using thermodynamic
modeling of phase equilibria for various rock types (Fig. 1). In addition, Hermann (2003)
compared the observed phase assemblages and compositions in K-bearing whiteschists with
experimentally determined petrogenetic grids and calibrated fluid-absent equilibria, which
reduced uncertainties in P–T conditions. Dating of accessory minerals included zircon,
monazite, and rutile, with results delimiting a restricted time span of 3.2 Ma for the UHP event
(e.g., Gebauer et al. 1997; Rubatto and Hermann 2001; Gauthiez-Putallaz et al. 2016). The
vertical burial and exhumation rates for Dora Maira are > 3.5 to 4 cm/yr, much faster than the
subduction rate expected from the horizontal velocity of 1.5 cm/yr inferred by plate motion
paths (Handy et al. 2010). The tightly constrained petrochronologic datasets for this terrane
have triggered the curiosity and imagination of modelers who are keen on understanding
geodynamic processes. So far, most of the available thermo-mechanical models predict both
lower peak temperatures and much lower burial and exhumation rates (e.g., Fig. 7 in Yamato et
al. 2008). The lower temperatures may reflect the neglect of frictional heating or heat advection
from fluids in the numerical experiments (Yamato et al. 2008; Penniston-Dorland et al. 2015).
Figure 1. Synthetic P–T–t path for the UHP unit of Dora Maira. Mineral reactions among quartz, coesite,
graphite and diamond were calculated using the thermodynamic database of Berman (1988). The reaction
Ky + Tc → Grt + Coe + Liq is from Gauthiez-Putallaz et al. (2016). P–T estimates for natural mineral
assemblages refer to the following lithologies: [1] K-bearing whiteschist (Schertl et al. 1991; Gauthiez-
Putallaz et al. 2016); [2] Na-bearing whiteschist (Compagnoni et al. 1995); [3] marbles (Castelli et al.
2007); [4] calcsilicate eclogites (Rubatto and Hermann 2001); [5] eclogite boudins (Groppo et al. 2007).
Experimental data [6] are from samples of K-bearing whiteschist (Hermann 2003). Select ages a1 and a2
are from Gauthiez-Putallaz et al. (2016), a3 from Gebauer et al. (1997) and a4, a5, a6, a7 from Rubatto
and Hermann (2001). Note that prograde ages (a1, a2), while nominally younger than the pressure peak
(a3, a4), are not distinguishable within their 2σ-uncertainties. These ages delimit a maximal time span of
0.7 Ma for burial of the Dora Maira UHP massif from 2.2–2.5 GPa to 3.5 GPa. Inset: burial and exhumation
rates from Rubatto and Hermann (2001) and Gauthiez-Putallaz et al. (2016). Abbreviations: Coe: coesite;
Dia: diamond; Gr: graphite; Grt: garnet; Ky: kyanite; Liq: silicate melt; Qz: quartz; Tc: talc.
Some models can reproduce the Alpine subduction P–T trajectories (Gerya et al. 2008;
Butler et al. 2014), but only for such slow subduction (which induces higher temperatures)
that the duration of the HP–UHP stages is overestimated by a factor of up to 2.5 (Fig. 2).
This large discrepancy has promoted alternative proposals that the pressure recorded by
the assemblages may exceed lithostatic pressure, and the observed (U)HP parageneses may
have formed at shallower depths (Gerya et al. 2008; Reuber et al. 2016). Such proposals
might reconcile durations of metamorphism, but would imply that slab-top geotherms in
Figure 2. Comparison between the synthetic pressure–time path of Figure 1 (dark) and the representative
P–T paths (light) from model V1.5 (Butler et al. 2014) and model 1 (Gerya et al. 2008) for the Dora Maira
UHP massif. These two thermal–mechanical models were selected because they show the highest burial and
exhumation rates. The duration of the HP–UHP stages suggested by the petrochronological data is shown
between the first recorded prograde stage (P ≥ 2 GPa) to the exhumation to crustal levels (P < 1 GPa). The
mismatch between durations as recorded by rocks (brief, ~3 Ma) vs. models (long, > 8 Ma) implies that natu-
ral subduction occurred faster than models are capable of explaining, i.e., rocks can be faster than models.
subduction zones are even hotter than rocks indicate, because the metamorphic temperature
recorded by a rock would occur at a shallower depth than the apparent pressure implies. If
tectonic overpressures were high, this would exacerbate the already stark discrepancy in P–T
conditions between samples (thermobarometry) and models: rocks record generally hotter
conditions, models predict generally colder conditions (Penniston-Dorland et al. 2015).
So far, the models that favor overpressure by introducing a mechanically heterogeneous
crust have not succeeded in reproducing the P–T–t path of the Dora Maira unit. Nor has
thermobarometry in this terrane discovered significant pressure-discrepancies, such as one
would expect from rock types with such different rheologies as those investigated. Indeed, the
range of pressures documented from weak and strong lithologies indicates no significant over-
and underpressures, thus thermobarometry does not appear to have registered appreciable
deviations from lithostatic pressure. Clearly, alternative geodynamic processes are needed to
understand continental subduction (see Guillot et al. 2009 for a review).
Methods of choice, choice of methods
The three sections of this volume present the state of the art and progress regarding
• BASICS: conceptual approaches used in petrochronology (3 chapters);
• METHODS: developments of analytical techniques (4 chapters);
• MINERALS: specific potential and use of mineral groups (8 chapters).
Numerous approaches may be used, so there is no best and certainly no unique approach
one could recommend for any specific research ambition or focus. Consequently, current and
future studies must make choices. Ideally, petrochronologists will be able to select tools in
combination, and in practice much will depend on the specific expertise of the investigators
and the available hardware in analytical labs, etc.
However, we must emphasize that the likelihood of obtaining convincing results depends
on nothing so much as finding the “right” samples. All of the approaches presented in this
book are time-intensive, and they promise success only when applied to suitable samples. It
is seldom evident in the field whether a sample will have the right combination of promising
compositions and textures. Experience shows that while many samples may interview for
the job of petrochronologic investigation, few are worth hiring, so extensive sampling is
recommended. The best machines in the world cannot replace sample selection and detailed
documentation. Critical requirements include good structural control in the field, careful
microscopy and then accurate chemical / isotopic analysis.
A brief walk through the present volume will expose some of the developments in
petrochronology and the main principles and purposes behind them.
Thermodynamic modeling of mineral assemblages and compositions (Chapters 2–3).
Quantifying conditions of rock formation is an essential requirement of petrochronology. This
task has become quite manageable in metamorphic rocks because thermodynamic models
and software based on these are readily available. It is now standard procedure to compute
phase diagrams for the approximate bulk composition of a sample under investigation, which
allows multi-equilibrium thermobarometry, supports the interpretation of microscopically
visible phase relations and reaction textures, all useful to design strategies for in situ age dating
(Yakymchuk et al. 2017). This approach has proven very helpful, but phase diagrams depict
equilibrium relations, and many rock samples contain disequilibrium features (e.g., zoned
minerals). Lanari and Engi (2017) emphasize the importance of choosing the appropriate bulk
composition if the approach used is based on equilibrium phase relations. That chapter presents
successful methods, based on X-ray maps, to accommodate compositional heterogeneity, such
as arise where porphyroblasts grow or partial melting occurs. However, the interpretation of
ages measured on minerals that grew out of equilibrium, e.g., due to reaction overstepping
(Spear et al. 2014), remains a concern.
The fundamentals of modeling igneous rocks are not specifically covered in this
volume, but the most prevalent models of melting and solidification are MELTS (Ghiorso
and Sack 1995), PETROLOG3 (Danyushevsky and Plechov 2011), and MELTS derivatives
(Ghiorso et al. 2002; Gualda et al. 2012). While these models appear to work reasonably
well for mafic and some rhyolitic compositions, thermodynamic models for compositionally
intermediate rocks are lacking, in part because adequate mixing models for amphibole have
not yet been assembled. Peraluminous rocks also pose modeling challenges. Still, MELTS
and PETROLOG3 provide good starting points, at least for mafic rocks. When combined with
accessory mineral solubilities (e.g., Watson and Harrison 1983; Montel 1993; Stepanov et al.
2012; Boehnke et al. 2013), the growth of monazite and zircon can also be modeled.
Diffusion can set limits and open new doors (Chapter 4) As discussed in Kohn and
Penniston-Dorland (2017), diffusion can homogenize pre-existing chemical heterogeneities,
biasing certain types of petrologic calculations, but can also induce chemical zoning that may
be useful for determining rates of geologic processes. One aspect is that diffusivity limits the
temperature range in which specific thermometers and chronometers can be trusted, a major
worry in some types of chronology (e.g., 40Ar/39Ar dating). However, quite a different aspect
is the door that diffusion opens for delimiting the duration of thermal events by modeling
the time-dependent flux of components in response to composition gradients. Diffusion
chronometry also is a central topic in Chapter 16 (Dohmen et al. 2017).
Analytical methods for mineral chronometry (Chapter 5–8). Petrochronological work
has continually stretched and exploited the range of analytical techniques available. Spatial
resolution is a concern to itself, but linking chronometric data with geochemical analyses
remains a key challenge. Consequently, much effort has focused on obtaining both types of
data using a single instrument, and many chapters emphasize not only the advantages but also
the disadvantages of each technique. Nearly all petrologic studies rely on the electron probe
microanalyzer (EPMA) to characterize major and minor element chemistry of minerals, which
is required for modeling (Chapters 1–2; Yakymchuk et al. 2017; Lanari and Engi 2017).
From the perspective of combined chronologic and compositional analysis:
•EPMA (Chapter 5, Williams and Jercinovic 2017) offers unparalleled spatial resolution for
polished sections and grain mounts (ca. 1 µm), but has relatively poor compositional resolution
(typically a few tens of ppm) and age resolution, in part because it cannot distinguish isotopes.
•Laser-ablation inductively-coupled plasma mass spectrometry (LA-ICP-MS;
Chapter 6, Kylander-Clark 2017) and secondary ion mass spectrometry (SIMS or ion
microprobe; Chapter 7, Schmitt and Vazquez 2017) offer superior compositional resolution
(sub-ppm), but with only moderate resolution of ages (a few percent, absolute) and position
(typically > 10 µm). However, depth profiling into crystal faces can permit sub-µm spatial
resolution (e.g., Grove and Harrison 1999; Carson et al. 2002; Cottle et al. 2009; Kohn and
Corrie 2011; Smye and Stöckli 2014).
•Thermal ionization mass spectrometry (TIMS; Chapter 8, Schoene and Baxter 2017)
offers the highest chronologic precision and accuracy, but the poorest spatial resolution (whole
grains or partial grains). Nonetheless, zoning in large crystals, such as garnet, can provide
direct estimates of mineral growth rates that can be linked directly to tectonic processes.
Preferred clocks and their role in petrochronology (Chapters 9–16). Accessory minerals
have dominated much of classical chronometry, whereas major rock-forming minerals (and
reactions among these) have played key roles in determining petrogenetic conditions. Over the
past decade, the boundaries between these two camps have gradually crumbled, and combined
thermometry, trace element fingerprinting, and chronometry are based increasingly on one and
the same mineral grain (or subgrain). From the perspective of published chronologic utility:
•Zircon (Chapter 9, Rubatto 2017; Chapter 10, Schaltegger and Davies 2017; Chapter 11,
Harrison et al. 2017) remains the most popular chronometer because of its high Pb retentivity
(Cherniak and Watson 2001), preservation of multiple growth zones (Corfu et al. 2003),
resistance to abrasion during transport as a clastic grain, and use of trace elements for
thermometry (Watson et al. 2006) as well as chemical fingerprinting (Rubatto 2002; Whitehouse
and Platt 2003). Zircon petrochronology from all three basic rock types is discussed in this
volume. Metamorphic zircon requires in situ analysis by LA-ICP-MS or SIMS because of its
complex zoning but sub-wt% U contents. It provides a key monitor of P–T–t paths and fluid
processes (Rubatto 2017), even if it remains difficult to pinpoint precisely the conditions,
especially of pressure, at which zircon formed. Igneous zircon records magmatic processes
(Schaltegger and Davies 2017), where short timescales commonly necessitate high-precision
TIMS analysis (Schoene and Baxter 2017). Detrital zircon preserves chemical, isotopic,
mineralogical (inclusion) and chronologic records of past Earth processes; Harrison et al.
(2017) illustrate the power and impact of this information most comprehensively through their
review of Hadean zircons, which are commonly analyzed using SIMS or LA-ICP-MS.
•REE minerals (Chapter 12, Engi 2017), especially monazite and allanite, are increasingly
popular targets of petrochronology, owing to their high Th and U contents, low cation diffusion
rates (high Pb retentivity; e.g., Cherniak et al. 2004), and, for monazite, low initial Pb (Parrish
1990). Multiple chemically distinct domains in each grain require in situ analysis, and can
in favorable cases be tied to specific mineral reactions and hence to P–T conditions. While
monazite can be dated using EPMA, LA-ICP-MS, and SIMS, allanite’s high common Pb
content requires isotopic methods of analysis, typically LA-ICP-MS and SIMS because of
intra-crystalline zoning. Xenotime and apatite have received less scrutiny, and Engi (2017)
discusses how more research could explore better their petrochronologic value.
•Titanite (Chapter 13, Kohn 2017) has long been analyzed for U–Pb ages, but until
recently its high closure temperature for Pb and other cations was underappreciated, and
ages were assumed to reflect cooling (see Frost et al. 2000). That is, geochronologists linked
titanite’s utility to thermochronology rather than petrochronology. Like allanite, multiple
domains with relatively high common Pb require in situ isotopic analysis via LA-ICP-MS or
SIMS, and, like zircon, a direct thermometer—the Zr-in-titanite thermometer (Hayden et al.
2008)—links temperature to age. Examples of applications commonly focus on P–T–t paths,
but new research is developing its use in understanding igneous petrogenesis.
•Rutile (Chapter 14, Zack and Kooijman 2017) is not so commonly analyzed as other
accessory minerals because it contains virtually no Th, and U contents are typically < 10 ppm.
These disadvantages are offset by extremely low common Pb contents, such that high-U rutile
can be analyzed in situ using LA-ICP-MS or SIMS. Zack and Kooijman (2017) emphasize
rutile’s utility in understanding crystallization conditions for low-temperature metamorphic
rocks, early stage cooling for high-temperature metamorphic rocks, and rutile’s ability to
recover temperatures directly through Zr-in-rutile thermometry. Textural context can be of
paramount importance, as it is for most petrochronology involving accessory minerals.
•Garnet (Chapter 15, Baxter et al. 2017) is, of course, a major mineral in many rocks, and
it is presented in all its glory for its ability to constrain petrogenetic processes through combined
major and minor element zoning plus Lu–Hf and Nd–Sm isotopic dating. Generally low parent-
daughter ratios and extremely low decay rates require TIMS analysis for accurate dating. While
inclusions in garnet commonly can be very helpful in relating garnet growth to specific P–T
stages or deformation fabrics, certain inclusions jeopardize garnet dating. These challenges are
reviewed by Baxter et al (2017), as are the techniques used to minimize the effects on ages.
•Major igneous minerals, including olivine, plagioclase and pyroxenes (Chapter 16,
Dohmen et al. 2017), round out this volume and provide useful petrochronologic constraints.
Although U-series dating can be applied (with care) to igneous systems, much emphasis is
placed on diffusion profiles. These allow the duration and rates of various processes to be
estimated, e.g., cooling rates for volcanic bodies and magma chamber time scales, such as
crystal growth rates and magma residence time.
EVOLUTION OR REVOLUTION?
The concept of petrochronology was originally proposed (Fraser et al. 1997) to unify
the fields of metamorphic petrology and geochronology. These authors demonstrated that the
breakdown of a major phase such as garnet or amphibole can trigger zircon growth, and thus
geochronological data can be linked to petrologically derived pressure–temperature conditions,
helping piece together a P–T–time path.
Previously, we outlined linkages among different branches of science that contribute to the
field of petrochronology. As a snapshot of the evolution thus far, we compare how the evolution
in each field or technique is reflected in citations over the past years (Fig. 3). For the fields
of “Petrology” and “Geochronology + Petrology” the citation frequencies increased linearly
from 1970 to ~2002, then growth rates doubled; “Petrochronology” was a term barely cited ten
years ago, taking off only after 2010—but look at the rate! Citations of “Thermobarometry”
seem to have been growing linearly since the early 80s, as have references to petrological
software, i.e., Thermocalc (since the late 80s), “PerpleX”, and “Theriak-Domino” (both since
~2005). However, exponential growth is evident for the chronometric techniques. Citations of
SIMS (including SHRIMP) lead the pack, but LA-ICP-MS is catching up.
Figure 3. Citations per year, found in Google Scholar (February 2017), using the following criteria,
A: “Petrology”, “Geochronology + Geochemistry”; B: “Petrochronology”; C: “Thermobarometry”,
“Thermocalc”, “Perple_X”, “Theriak-Domino”; D: “SIMS or SHRIMP + Geochronology + Petrology”,
“LA-ICPMS or LA-ICP-MS + Geochronology + Petrology”, “ID-TIMS + Geochronology + Petrology”.
The field of petrochronology is evidently young, yet spectacular progress has been made
already (Kohn 2016), as evident in the literature and reviewed in this volume. Depending on
the scope of applications, studies cover a wide range of scales, from microns to mountain belts.
Our understanding has thus deepened, from small-scale processes in the evolution of magmas
and rocks, their duration and tempo, all the way to the scale of orogeny, lithospheric tectonics,
the nature of Early Earth, and the dynamics of continental recycling.
For these reasons, the time to dedicate a RiMG volume to petrochronology is ripe, for the
field is burgeoning, combining many diverse directions, aiming to elucidate time and tempo
of processes that have shaped the Earth (and sister planetary bodies) at all scales. We trust that
this RiMG volume, compiling current petrochronological methods and reviewing applications,
will stimulate and nourish future work in this field. It is with this intent that the authors of each
chapter have outlined current limits and identify potentially useful directions for study.
ACKNOWLEDGMENTS
It is a pleasure to thank Ian Swainson, the Editor of the RiMG Series. He has provided
expedient and unfailing support in producing this volume. We also thank Frank Spear for his
perceptive comments that helped us improve this Introduction. This work was funded by US-
NSF grants EAR-1321897, -1419865, and -1545903 to MJK and by the Swiss Nationalfonds
grants 200021-117996/1, 200020-126946, and -146175 to ME.
REFERENCES
Baxter EF, Caddick MJ, Dragovic B (2017) Garnet: A rock-forming mineral petrochronometer. Rev Mineral
Geochem 83:469–533
Beaumont C, Jamieson RA, Nguyen M, Lee B (2001) Himalayan tectonics explained by extrusion of a low-
viscosity crustal channel coupled to focused surface denudation. Nature 414:738–742
Berger GW, York D (1981) 40Ar/39Ar dating of the Thanet gabbro, Ontario: looking through the Grenvillian
metamorphic veil and implications for paleomagnetism. Can J Earth Sci 18:266–273
Berman RG (1988) Internally consistent thermodynamic data for minerals in the system Na2O–K2O–CaO–MgO–
FeO–Fe2O3–Al2O3–SiO2–TiO2–H2O–CO2. J Petrol 29:445–522
Butler JP, Beaumont C, Jamieson RA (2014) The Alps 2: Controls on crustal subduction and (ultra)high-pressure rock
exhumation in Alpine-type orogens. J Geophys Res: Solid Earth 119:5987–6022 doi:10.1002/2013JB010799
Castelli D, Rolfo F, Groppo C, Compagnoni R (2007) Impure marbles from the UHP Brossasco–Isasca Unit
(Dora–Maira Massif, western Alps): evidence for Alpine equilibration in the diamond stability field and
evaluation of the X(CO2) fluid evolution. J Metamorph Geol 25:587–603
Chopin C (1984) Coesite and pure pyrope in high-grade blueschists of the Western Alps: a first record and some
consequences. Contrib Mineral Petrol 86:107–118 doi:10.1007/bf00381838
Chopin C, Schertl H-P (1999) The UHP unit in the Dora-Maira massif, Western Alps. Int Geol Rev 41:765–780
Chopin C, Henry C, Michard A (1991) Geology and petrology of the coesite-bearing terrain, Dora-Maira massif,
Western Alps. Eur J Mineral 3:263–291
Compagnoni R, Hirajima T, Chopin C (1995) Ultra-high-pressure metamorphic rocks in the Western Alps. In:
Ultrahigh Pressure Metamorphism. Coleman RG, Wang X (eds) Cambridge University Press, Cambridge,
p 206–243
Dohmen R, Faak K, Blundy JD (2017) Chronometry and speedometry of magmatic processes using chemical
diffusion in olivine, plagioclase and pyroxenes. Rev Mineral Geochem 83:535–576
Engi M (2017) Petrochronology based on REE-minerals: monazite, allanite, xenotime, apatite. Rev Mineral
Geochem 83: 365–418
England PC, Thompson AB (1984) Pressure–temperature–time paths of regional metamorphism I. Heat transfer
during the evolution of regions of thickened continental crust. J Petrol 25:894–928
Fraser G, Ellis DJ, Eggins S (1997) Zirconium abundance in granulite-facies minerals, with implications for zircon
geochronology in high-grade rocks. Geology 25:607–610 doi:10.1130/0091–7613(1997)025<0607:ZAIGF
M>2.3.CO;2
Gauthiez-Putallaz L, Rubatto D, Hermann J (2016) Dating prograde fluid pulses during subduction by in situ U–Pb
and oxygen isotope analysis. Contrib Mineral Petrol 171:15 doi:10.1007/s00410-015-1226-4
Gebauer D, Schertl HP, Brix M, Schreyer W (1997) 35 Ma old ultrahigh-pressure metamorphism and evidence for
very rapid exhumation in the Dora Maira Massif, Western Alps. Lithos 41:5–24
Gerya T (2011) Future directions in subduction modeling. J Geodynam 52:344–378 doi:10.1016/j.jog.2011.06.005
Gerya TV, Stöckhert B, Perchuk AL (2002) Exhumation of high-pressure metamorphic rocks in a subduction
channel: A numerical simulation. Tectonics 21:1056, doi:1010.1029/2002TC001406
Gerya TV, Perchuk LL, Burg J-P (2008) Transient hot channels: perpetrating and regurgitating ultrahigh-pressure,
high temperature crust–mantle associations in collision belts. Lithos 103:236–256
Groppo C, Lombardo B, Castelli D, Compagnoni R (2007) Exhumation history of the UHPM Brossasco-Isasca
Unit, Dora-Maira Massif, as inferred from a phengite–amphibole eclogite. Int Geol Rev 49:142–168
doi:10.2747/0020–6814.49.2.142
Guillot S, Hattori K, Agard P, Schwartz S, Vidal O (2009) Exhumation processes in oceanic and continental
subduction contexts: A review. In: Subduction Zone Geodynamics. Lallemand S, Funiciello F (eds) Springer
Berlin Heidelberg, Berlin, Heidelberg, p 175–205
Handy MR, Schmid SM, Bousquet R, Kissling E, Bernoulli D (2010) Reconciling plate-tectonic reconstructions
of Alpine Tethys with the geological–geophysical record of spreading and subduction in the Alps. Earth-Sci
Rev 102:121–158 doi:10.1016/j.earscirev.2010.06.002
Harrison TM, Bell EA, Boehnke P (2017) Hadean zircon petrochronology. Rev Mineral Geochem 83:329–363
Hermann J (2003) Experimental evidence for diamond-facies metamorphism in the Dora–Maira massif. Lithos
70:163–182 doi:10.1016/S0024-4937(03)00097–5
Kohn MJ (2016) Metamorphic chronology—a tool for all ages: Past achievements and future prospects. Am
Mineral 101:25–42
Kohn MJ (2017) Titanite petrochronology. Rev Mineral Geochem 83:419–441
Kohn MJ, Penniston-Dorland SC (2017) Diffusion: Obstacles and opportunities in petrochronology. Rev
Mineral Geochem 83:103–152
Kylander-Clark ARC (2017) Petrochronology by laser-ablation inductively coupled plasma mass
spectrometry. Rev Mineral Geochem 83:183–198
Lanari P, Engi M (2017) Local bulk composition effects on metamorphic mineral assemblages. Rev Mineral
Geochem 83:55–102
Oxburgh ER, Turcotte DL (1974) Thermal gradients and regional metamorphism in overthrust terrains with special
reference to the Eastern Alps. Schweiz Mineral Petrogr Mitt 54:642–662
Peacock SM (1990) Numerical simulation of metamorphic pressure–temperature–time paths and fluid production
in subducting slabs. Tectonics 9:1197–1211
Penniston-Dorland SC, Kohn MJ, Manning CE (2015) The global range of subduction zone thermal structures
from exhumed blueschists and eclogites: Rocks are hotter than models. Earth Planet Sci Lett 428:243–254
doi:10.1016/j.epsl.2015.07.031
Reiners PW, Ehlers TA, Zeitler PK (2005) Past, present, and future of thermochronology. Rev Mineral Geochem
58:1–18
Reuber G, Kaus BJP, Schmalholz SM, White RW (2016) Nonlithostatic pressure during subduction and collision
and the formation of (ultra)high-pressure rocks. Geology 44:343–346 doi:10.1130/g37595.1
Rubatto D, Hermann J (2001) Exhumation as fast as subduction? Geology 29:3–6
Rubatto D (2017) Zircon: The metamorphic mineral. Rev Mineral Geochem 83:261–295
Ruppel C, Hodges KV (1994) Pressure–temperature–time paths from two-dimensional thermal models: Prograde,
retrograde, and inverted metamorphism. Tectonics 13:17–44 doi:10.1029/93TC01824
Schaltegger U, Davies JHFL (2017) Petrochronology of zircon and baddeleyite in igneous rocks: Reconstructing
magmatic processes at high temporal resolution. Rev Mineral Geochem 83:297–328
Schertl H-P, Schreyer W, Chopin C (1991) The pyrope–coesite rocks and their country rocks at Parigi, Dora Maira
Massif, Western Alps: detailed petrography, mineral chemistry and P–T path. Contrib Mineral Petrol 108:1–21
Schmitt AK, Vazquez JA (2017) Secondary ionization mass spectrometry analysis in petrochronology. Rev
Schoene B, Baxter EF (2017) Petrochronology and TIMS. Rev Mineral Geochem 83:231–260
Shi Y, Wang C-Y (1987) Two-dimensional modeling of the P–T–t paths of regional metamorphism in simple
overthrust terrains. Geology 15:1048–1051 doi:10.1130/0091–7613(1987)15<1048:tmotpp>2.0.co;2
Spear FS, Thomas JB, Hallett BW (2014) Overstepping the garnet isograd: a comparison of QuiG barometry and
thermodynamic modeling. Contrib Mineral Petrol 168:1–15
Spear FS, Pattison DRM, Cheney JT (2016) The metamorphosis of metamorphic petrology. Geol Soc Am Spec
Papers 523 doi:10.1130/2016.2523(02)
Thompson AB, England PC (1984) Pressure–temperature–time paths of regional metamorphism II. Their inference
and interpretation using mineral assemblages in metamorphic rocks. J Petrol 25:929–955
van Keken PE, Kiefer B, Peacock SM (2002) High-resolution models of subduction zones: Implications for
mineral dehydration reactions and the transport of water into the deep mantle. Geochem Geophys Geosystem
3:1–20 doi:10.1029/2001GC000256
Vialon P (1966) Etude géologique du massif cristallin Dora–Maira, Alpes cottiennes internes, Italie. Thèse d’état,
Université de Grenoble
Williams HS (1893) The making of the geological time-scale. J Geol 1:180–197
Williams ML, Jercinovic MJ, Mahan KH, Dumond G (2017) Electron microprobe petrochronology. Rev
Yamato P, Burov E, Agard P, Le Pourhiet L, Jolivet L (2008) HP–UHP exhumation during slow continental
subduction: Self-consistent thermodynamically and thermomechanically coupled model with application to
the Western Alps. Earth Planet Sci Lett 271:63–74 doi:10.1016/j.epsl.2008.03.049
Yakymchuk C, Clark C, White RW (2017) Phase relations, reaction sequences and petrochronology. Rev
Zack T, Kooijman E (2017) Petrology and geochronology of rutile. Rev Mineral Geochem 83:443–468
Vol. 83 pp. 13-53, 2017 2
Phase Relations, Reaction Sequences

and Petrochronology
Chris Yakymchuk
Department of Earth and Environmental Sciences
University of Waterloo
Waterloo, Ontario
Canada
N2 L 3G1
cyakymchuk@uwaterloo.ca
Chris Clark
Department of Applied Geology
Curtin University
Perth
Western Australia
Australia
6102
C.Clark@curtin.edu.au
Richard W. White
Institute of Geoscience
University of Mainz
Mainz
Germany
D-55099
rwhite@uni-mainz.de
INTRODUCTION
At the core of petrochronology is the relationship between geochronology and the
petrological evolution of major mineral assemblages. The focus of this chapter is on outlining
some of the available strategies to link inferred reaction sequences and microstructures in
metamorphic rocks to the ages obtained from geochronology of accessory minerals and
datable major minerals. Reaction sequences and mineral assemblages in metamorphic rocks
are primarily a function of pressure (P), temperature (T) and bulk composition (X). Several
of the major rock-forming minerals are particularly sensitive to changes in P–T (e.g., garnet,
staurolite, biotite, plagioclase), but their direct geochronology is challenging and in many
cases not currently possible. One exception is garnet, which can be dated using Sm–Nd and
Lu–Hf geochronology (e.g., Baxter et al. 2013). Accessory mineral chronometers such as
zircon, monazite, xenotime, titanite and rutile are stable over a relatively wide range of P–T
conditions and can incorporate enough U and/or Th to be dated using U–Th–Pb geochronology.
Therefore, linking the growth of P–T sensitive major minerals to accessory and/or major
mineral chronometers is essential for determining a metamorphic P–T–t history, which is itself
critical for understanding metamorphic rocks and the geodynamic processes that produce them
(e.g., England and Thompson 1984; McClelland and Lapen 2013; Brown 2014).
1943-2666/17/0083-0002$05.00 (online)
14 Yakymchuk, Clark & White
Linking the ages obtained from accessory and major minerals with the growth and
breakdown of the important P–T sensitive minerals requires an understanding of the
metamorphic reaction sequences for a particular bulk rock composition along a well-
constrained P–T evolution. Fortunately, the phase relations and reaction sequences for
the most widely studied metamorphic protoliths (e.g., pelites, greywackes, basalts) can be
determined using quantitative phase equilibria forward modelling (e.g., Powell and Holland
2008). Comprehensive activity–composition models of the major metamorphic minerals in
large chemical systems (e.g., White et al. 2014a) allow the calculation of phase proportions
and compositions for a given rock composition along a metamorphic P–T path. For accessory
minerals, subsolidus growth and breakdown can be modelled in some cases using phase
equilibria modelling (e.g., Spear 2010; Spear and Pyle 2010). Suprasolidus accessory mineral
behaviour can be investigated by coupling phase equilibria modelling with the experimental
results of accessory mineral solubility in melt (Kelsey et al. 2008; Yakymchuk and Brown
2014b). This technique provides a basic framework for interpreting the geological significance
of accessory mineral ages in suprasolidus metamorphic rocks.
In this chapter, we use phase equilibria modelling techniques to investigate the reaction
sequences for three common rock types (pelite, greywacke, MORB) along several P–T paths and
explore how these sequences relate to accessory mineral growth and dissolution with a particular
focus on zircon and monazite and to a lesser extent apatite. First, we review the important major
rock-forming minerals that are used to link metamorphic reaction sequences to trace element
chemistries in accessory minerals. Second, we summarize the current understanding of the controls
of accessory mineral growth and breakdown in metamorphic rocks during a P–T evolution. Third,
we use phase equilibria modelling to examine the reaction sequences for common rock types
along several schematic P–T paths and discuss the implications for petrochronology. Finally, we
examine some of the complicating factors for reconciling the behaviour of accessory minerals in
natural systems with the predictions from phase equilibria modelling.
MAJOR MINERALS
Understanding the growth and consumption of the major rock-forming minerals is
important in accessory mineral petrochronology for four reasons. First, major minerals may
contain significant quantities of the essential structural constituents of accessory minerals
commonly used as chronometers. In some cases, accessory minerals can grow directly from the
breakdown of major minerals. Examples of this are zircon growth from the release of Zr during
the breakdown of garnet (Fraser et al. 1997; Degeling et al. 2001) and ilmenite (Bingen et al.
2001). Second, these minerals may represent important repositories for trace elements, and thus
the growth and breakdown of major minerals will influence the availability of these elements for
incorporation by growing accessory minerals. Third, the major minerals are important hosts for
inclusions of accessory minerals. The breakdown of the major minerals may liberate included
accessory minerals into the reaction volume of the rock or alternatively may sequester these
minerals away allowing their preservation when they would otherwise be consumed in a reaction
sequence (e.g., Montel et al. 2000). Fourth, the microstructural relationships between accessory
and major minerals provide context for delineating the P–T history of a metamorphic rock.
Garnet and plagioclase are the major minerals most commonly used in petrochronology
because of their distinctive trace element behaviour. Linking their growth and breakdown to ages
obtained from accessory mineral chronometers requires an understanding of the bulk composition
controls on their stabilities. Below, we outline the controls on the growth and breakdown of garnet
and plagioclase, which are of particular importance for petrochronology studies. Our focus is on
linking mineral growth to metamorphic reactions and we do not discuss the minerals that are
important for thermochronology studies (e.g., amphibole, biotite, muscovite, K-feldspar).
Phase Relations, Reaction Sequences and Petrochronology 15
Garnet
Garnet is one of the most useful minerals for constraining metamorphic grade and its high
density and strong partitioning of cations forms the basis of many useful thermobarometers (e.g.,
Spear 1995; Caddick and Kohn 2013). It is a common metamorphic mineral for many different
protoliths (pelites, mafic rocks, ultramafic rocks, calc-silicates) and has been extensively used
in petrochronology studies (e.g., Vance and O’Nions 1990; Vance and Mahar 1998; Harris et
al. 2004). Coupling P–T estimates from garnet with monazite U–Pb geochronology can also be
used to constrain P–T–t paths (e.g., Foster et al. 2000; Gibson et al. 2004; Dragovic et al. 2016).
Garnet can also be directly dated using Sm–Nd and Lu–Hf geochronology (e.g., Baxter et al.
2013, 2017, this volume). Although linking P–T information from garnet-bearing assemblages
with garnet geochronology is very powerful (e.g., Mulcahy et al. 2014; Dragovic et al. 2012,
2015), direct dating of garnet is challenging and is not yet extensively used.
An important control on the stability of garnet during metamorphism is the bulk rock MnO
content (e.g., Symmes and Ferry 1992; Mahar et al. 1997; White et al. 2014b). For pelites,
higher bulk rock Mn concentrations stabilize garnet at lower temperatures in the greenschist
facies and at lower pressure in the amphibolite facies (White et al. 2014b). The ratio of Fe/Mg
is also an important control on garnet growth. For example, relatively magnesium-rich pelites
have restricted garnet stability fields (e.g., White et al. 2014a) and in very Mg-rich bulk rock
compositions garnet may not even grow along common P–T paths (e.g., Fitzsimons et al.
2005). Therefore, depending on the bulk composition of the rock, garnet can be first stabilized
in the greenschist facies, the amphibolite facies or not at all.
Garnet is the most important major mineral sink for the heavy rare earth elements (HREE)
and yttrium in metamorphic rocks (Bea et al. 1994). Accessory minerals such as zircon,
monazite, xenotime, apatite, epidote and allanite are also important repositories for the HREE
and Y. Thus, the growth and breakdown of garnet plays an important role in the HREE and Y
budgets in rocks and links the trace element chemistry of accessory mineral chronometers to
the P–T information obtained from garnet-bearing metamorphic assemblages (Pyle et al. 2001;
Foster et al. 2002; Rubatto 2002; Kohn and Malloy 2004; Rubatto and Hermann 2007; Taylor
et al. 2015). In general, the equilibrium distribution coefficients of the HREE between garnet
and zircon are close to unity for Gd to Lu at high temperature (Taylor et al. 2015). For Yttrium,
concentrations in monazite are ~1.5 orders of magnitude higher than in garnet (Bea et al.
1994). Xenotime has Y concentrations that are ~2 orders of magnitude higher than in monazite
(Pyle et al. 2001). However, in most cases, garnet is substantially more modally abundant
than monazite and xenotime and thus the breakdown or growth of garnet will exert first-order
controls on the Y and HREE budget of the rock. Zircon and monazite crystallization during
garnet growth will result in relatively low concentrations of HREE and Y in these minerals
because these elements are partitioned into garnet. By contrast, garnet breakdown during
accessory mineral growth can result in HREE- and Y-enriched trace element concentrations
of the new growth zones of accessory minerals (e.g., Foster et al. 2002, 2004; Rubatto 2002).
Plagioclase
Plagioclase is extremely common in metamorphic rocks and is stable over a wide range
of P–T conditions. Plagioclase breakdown with increasing pressure is an important part of
the reaction sequences for high-pressure metamorphic rocks and eclogites. Similar to garnet,
plagioclase is commonly a key mineral in thermobarometers that provides important P–T
information in many different rock types (Ghent 1976; Molina et al. 2015; Wu 2015). For
petrochronology, plagioclase (and to a lesser extent K-feldspar) strongly partitions Eu and
Sr over the other major rock-forming minerals (e.g., Gromet and Silver 1983). The growth
and breakdown of plagioclase can be tied to Eu anomalies in rare earth element patterns of
accessory minerals (Rubatto et al. 2013; Holder et al. 2015; Regis et al. 2016). Strontium
concentrations have been used to link accessory mineral growth to the timing of plagioclase
breakdown; this is a particularity important relationship in ultrahigh pressure metamorphic
rocks (Finger and Krenn 2007; Kylander-Clark et al. 2013; Holder et al. 2015).
ACCESSORY MINERALS
Like the major minerals, accessory minerals are involved in the reaction sequence
experienced by metamorphic rocks. The behaviour of some accessory minerals used in
petrochronology such as epidote, titanite and rutile can be quantified along a P–T evolution
using phase equilibria modelling. However, the behaviour of zircon and monazite—the most
commonly used mineral chronometers—is more difficult to quantify with current phase
equilibria modelling techniques (e.g., Spear and Pyle 2010; Kelsey and Powell 2011) as they
contain key elements commonly not considered in model chemical systems.
Epidote
Epidote is common in low–medium pressure metabasites, metamorphosed intermediate
rocks and calc-silicates (e.g., Grapes and Hoskin 2004) as well as in high- to ultrahigh-pressure
metamorphic rocks (e.g., Enami et al. 2004). The chemical controls on epidote stability are
mainly bulk rock concentrations of Ca and Al as well as fO2 (e.g., Enami et al. 2004). The
epidote-group minerals can be important repositories for Zr (e.g., Frei et al. 2004; Kohn et al.
2015) and LREE in metamorphic rocks (Frei et al. 2004; Janots et al. 2008). Therefore, the
breakdown of Zr-rich epidote may have the potential to generate metamorphic zircon. Coupled
allanite and epidote breakdown may produce new monazite (e.g., Janots et al. 2008) if there is
also a source of phosphorus such as xenotime or apatite.
Titanite
Titanite is common in mafic rocks and in some low-grade metasedimentary rocks and it
can be used as a barometer in metabasites (e.g., Kapp et al. 2009). An important factor in the
stability of titanite is the relative activities of Ca and Al as influenced by the bulk rock ratio
of Ca to Al and the accompanying mineral assemblage. High Ca activities favour titanite over
ilmenite and high Al activities favour anorthite (and ilmenite) over titanite (Frost et al. 2001).
Consequently, titanite is a common mineral in low-Al compositions such as calc-silicate and
mafic rocks whereas aluminous metasedimentary rocks generally have ilmenite as the main
Ti-bearing mineral above the greenschist facies.
Titanite can be directly dated using U–Pb geochronology, but there is some uncertainty
about the closure temperature for Pb, which is generally considered to be around 600 °C
(Warren et al. 2012; Spencer et al. 2013; Stearns et al. 2015; Kirkland et al. 2016), depending
on cooling rate. Diffusion profiles of trace elements in titanite can be used to determine the
timing and duration of cooling (e.g., geospeedometry) and this is covered in more detail
in Kohn (2017, this volume). Titanite thermometry uses the Zr concentration of titanite
to estimate temperature (Hayden et al. 2008). However, application of this thermometer
requires an estimate of titania activity (aTiO2), which, in the absence of rutile, is difficult to
constrain and can vary along a metamorphic evolution depending on the reaction sequences
and mineral assemblages as discussed below.
Rutile
Rutile is commonly stable in relatively reduced bulk rock compositions (e.g., Diener and
Powell 2010) and is particularity useful in high-pressure metamorphic rocks (Zack and Kooijman
2017, this volume). Similar to titanite, rutile can be directly dated with U–Pb geochronology (e.g.,
Mezger et al. 1989; Clark et al. 2000; Zack et al. 2011; Ewing et al. 2015) and it can be used as
a geospeedometer (Smye and Stockli 2014). One of the main uses of rutile is as a thermometer
that uses the Zr concentration of rutile to estimate temperature (Zack et al. 2004; Watson et al.
2006; Ferry and Watson 2007; Tomkins et al. 2007; Hofmann et al. 2013; Taylor-Jones and Powell
2015). These applications are discussed in more detail by Zack and Kooijman (2017, this volume).
Metamorphic reaction sequences with or without rutile have important implications for
applying the Ti-in-quartz thermometer (e.g., Chambers and Kohn 2012; Ashley and Law
2015) and the Ti-in-zircon thermometer (e.g., Ferry and Watson 2007), both of which use
aTiO2 as a variable. In rutile-bearing systems aTiO2 is buffered at 1.0. In rutile-absent systems,
application of the thermometer requires an estimate of aTiO2 that is less that 1.0. Figure 1
shows the difference of calculated temperatures using the Ti-in-quartz thermometer (Wark
and Watson 2006) and the Ti-in-zircon thermometer (Ferry and Watson 2007) as a function
of temperature for a range of aTiO2 values.
(a) Ti-in-quartz thermometer
aTiO = 0.
2 9
-20
aTi = 0.8
O2
-40 aTi = 0.7

O2
a
-60 TiO
2 =0
.6
∆T (°C)
-80 a
TiO
2 =0
.5
-100
-120
-140
400 500 600 700 800 900 1000
Temperature (°C)
(b) Ti-in-zircon thermometer

0
aTiO = 0.
-10 2 9
aTi = 0.8
-20 O2
-30 aTi
O2 = 0.7
-40
a
TiO
2 =0
-50 .6
∆T (°C)
-60 a
TiO
2 =0
.5
-70
-80
-90
-100
600 650 700 750 800 850 900 950 1000
Temperature (°C)
Figure 1
Figure 1. Underestimation of metamorphic temperature when the bulk rock value of aTiO2 is less than 1.0
(the true value is shown by the contours), but a value of 1.0 is assumed in: (a) the Ti-in-quartz thermometer
of Wark and Watson (2006), and (b) the Ti-in-zircon thermometer of Ferry and Watson (2007).
For example, consider an amphibolite-facies rock at 650 °C with a true aTiO2 value 0.6. If
an aTiO2 value of 1.0 is assumed for Ti-in-quartz thermometry, the result would underestimate
the true temperature by ~50 °C (Fig. 1a). Similarly, using an aTiO2 value of 1.0 when the true
value is 0.6, the Ti-in-zircon thermometer would underestimate the true temperature by ~40 °C
(Fig. 1b). Phase equilibria modelling provides one method of determining aTiO2 in metamorphic
rocks at different stages of a P–T evolution in rutile-absent rocks (e.g., Ashley and Law 2015).
We explore this in more detail during our discussion of particular reaction sequences.
Zircon
Zircon is the main repository of Zr in most igneous and metamorphic rocks and is the
most widely used accessory mineral for U–Pb geochronology (Hoskin and Schaltegger 2003;
Rubatto 2017, this volume). Many metamorphic rocks contain relict zircon that may be
detrital or igneous in origin and may also be inherited from a previous metamorphic event. For
petrochronology, metamorphic zircon can be used to date different portions of a P–T evolution
depending on the growth mechanisms. The main processes that increase the mode of zircon in
metamorphic rocks include solid-state growth and the crystallization from anatectic melt. In
some cases, zircon may precipitate from hydrothermal fluids (e.g., Schaltegger 2007).
Solid-state zircon growth can occur from the breakdown of other Zr-rich minerals such
as garnet (Fraser et al. 1997; Degeling et al. 2001), ilmenite (Bingen et al. 2001), rutile
(Ewing et al. 2014) and possibly amphibole (Sláma et al. 2007). This zircon can be used to
date major metamorphic assemblage changes during prograde (e.g., Fraser et al. 1997) or
retrograde metamorphism (e.g., Degeling et al. 2001) depending on when the Zr-rich mineral
breaks down during the reaction sequence. However, major minerals such as garnet, rutile and
hornblende can accommodate more Zr as temperature increases and, in general, zircon will be
consumed and prograde zircon growth is expected to be limited (Kohn et al. 2015).
The crystallization of anatectic melt typically drives new zircon growth in suprasolidus
metamorphic rocks (Watson 1996; Roberts and Finger 1997; Schaltegger et al. 1999; Vavra et
al. 1999; Hermann et al. 2001; Kelsey et al. 2008; Yakymchuk and Brown 2014b). Experimental
studies show that the concentration of Zr in melt needed to maintain equilibrium with zircon
increases with temperature (e.g., Harrison and Watson 1983; Gervasoni et al. 2016) and the
compositional parameter M, which is the cation ratio of [Na + K + 2 Ca]/ [Al × Si] (Watson and
Harrison 1983; Boehnke et al. 2013). Generally, the concentration of Zr needed for saturation
of intermediate melts (high M values) is higher than for more felsic melts (low M values). In
suprasolidus metamorphic rocks, the most important factor is the amount of anatectic melt
present in the system (Kelsey et al. 2008, 2011; Yakymchuk and Brown 2014b). As the fraction
of melt increases during prograde metamorphism, zircon is expected to break down to maintain
Zr saturation of the melt in an equilibrated system. Consequently, prograde zircon growth is
expected to be limited above the solidus. In general, melt crystallization during cooling from
peak T is expected to be the main mechanism for zircon growth in suprasolidus metamorphic
rocks. This is supported by ranges of concordant ages that reflect protracted zircon growth during
melt crystallization from peak T in migmatites (e.g., Korhonen et al. 2013b, 2014).
New zircon growth can also occur at the expense of pre-existing zircon with no change in
the mode of zircon in the rock. Recrystallization of metamict zircon in the presence of a fluid
may be an important factor for some prograde zircon (e.g., Rubatto and Hermann 2003; Hay
and Dempster 2009). Ostwald ripening has been proposed as a mechanism that could produce
prograde growth of zircon in suprasolidus metamorphic rocks (Vavra et al. 1999; Nemchin et
al. 2001; Kawakami et al. 2013). Ostwald ripening (or second phase coarsening) is a process
whereby small solids are preferentially dissolved and precipitate on existing larger solids to
reduce the total surface free energy (e.g., Tikare and Cawley 1998). While this process has
been studied and debated for the major minerals in metamorphic rocks (Miyazaki 1991, 1996;
Carlson 1999, 2000) it has not been as extensively studied for accessory minerals—an exception
being Nemchin et al. (2001). Nonetheless, it is a possible mechanism for prograde growth of
zircon in suprasolidus metamorphic rocks, though involving no net modal increase in zircon.
Zircon can also be used as a thermometer in igneous and metamorphic rocks. The Ti-in-
zircon thermometer has been increasingly applied to high-temperature metamorphic rocks
over the last decade. Kelsey and Hand (2015) compiled Ti-in-zircon temperatures from UHT
rocks and noted that 62% of the results fall below the UHT threshold of 900 °C. One of the
possible reasons for this discrepancy is that some UHT rocks do not contain rutile and are
therefore undersaturated in TiO2 (aTiO2 < 1.0). Values of aTiO2 can range from 1.0 when rutile
is in equilibrium with growing zircon to values as low as 0.6 (e.g., Hiess et al. 2008). Using a
value of 1.0 when rutile is absent will underestimate the true temperature. Again, the presence
and absence of rutile during a reaction sequence has important consequences for applying
mineral thermometers in metamorphic rocks.
Monazite
Monazite is generally more reactive than zircon in subsolidus metamorphic rocks and has
been extensively used for U–Pb geochronology in aluminous bulk compositions (e.g., Parrish
1990; Engi 2017, this volume). Monazite can be detrital in origin (Smith and Barreiro 1990;
Kingsbury et al. 1993; Suzuki et al. 1994; Rubatto et al. 2001), produced through solid-state
reactions (e.g., Rubatto et al. 2001; Wing et al. 2003), precipitated from a fluid (e.g., Ayers et
al. 1999) or can crystallize from anatectic melt (e.g., Stepanov et al. 2012).
Solid-state monazite growth during prograde metamorphism occurs from the breakdown
of LREE-rich precursors, which can include: allanite for high-Ca bulk compositions (Spear and
Pyle 2010; Wing et al. 2003; Finger et al. 2016), LREE-rich clays (Copeland et al. 1971), Th
or LREE oxides and hydrous phosphates (Spear and Pyle 2002). The most studied solid-state
reaction is the growth of monazite at the expense of allanite in bulk compositions with sufficient
Ca and LREE to grow allanite at lower grade (e.g., Janots et al. 2008). In natural examples, this
reaction has been spatially correlated with the garnet-in isograd (Catlos et al. 2001; Foster et al.
2004), the staurolite-in isograd (Smith and Barriero 1990; Kohn and Malloy 2004; Corrie and
Kohn 2008) and the kyanite or sillimanite-in isograds (Wing et al. 2003; Štípská et al. 2015).
Spear (2010) used phase equilibria modelling to examine the P–T conditions of the
allanite to monazite transition in several bulk rock compositions relative to the average pelite
of Shaw (1956). The reaction boundaries for monazite-in at the expense of allanite for these
various compositions are summarized in Figure 2. A higher bulk rock CaO concentration
allows allanite to persist to higher temperatures whereas a higher Al2O3 concentration stabilizes
monazite at lower temperatures. However, the results of this modelling of monazite growth in
rocks with variable bulk rock Al2O3 concentrations by Spear (2010) are different to those
observed in some natural examples. For example, Wing et al. (2003) found that pelites from
New England with relatively elevated Ca and/or Al concentrations can preserve allanite to
higher temperatures. Gasser et al. (2012) found no link between bulk rock concentrations of
CaO and Al2O3 and the timing of monazite growth. Therefore, the bulk compositional controls
of monazite growth at the expense of allanite are not always clear, but it appears that monazite
can grow at a range of temperatures and pressures along a prograde P–T path.
The relative amounts of iron and magnesium in a bulk rock composition may also play
an important role in the reactions that produce subsolidus monazite. For example, Fitzsimons
et al. (2005) showed that monazite in pelitic schists from Western Australia was generated
during garnet-breakdown to staurolite and was interpreted to record the timing of peak
metamorphism. However, samples that were too magnesium-rich to grow garnet or too iron-
rich to grow staurolite at the expense of garnet yielded older monazite ages. Fitzsimons et al.
(2005) interpreted these older ages to represent greenschist-facies monazite growth.
12
Shaw (1956)
+ Allanite
10
1.25 x Al2O3
Pressure (kbar)
3
l2 O
3
2O
xA
l
xA
5
0.7
5
1.
O
Ca
2x
4
+ Monazite
350 450 550 650 750 850

Temperature (°C)
Figure 2. Compilation of the range of P–T conditions for the allanite to monazite reaction for variable bulk
rock pelite compositions. Curves are from Spear (2010).
Figure 2
In suprasolidus metamorphic rocks, monazite breakdown and growth is controlled
mainly by dissolution into and crystallization from anatectic melt. Experimental studies
demonstrate that monazite dissolution into melt is a function of temperature, pressure, and
the bulk composition of the melt (Montel 1986; Rapp and Watson 1986; Rapp et al. 1987;
Skora and Blundy 2012; Stepanov et al. 2012; Duc-Tin and Keppler 2015). The solubility of
monazite increases with higher H2O concentrations in melt (Stepanov et al. 2012) and with
decreasing phosphorus concentrations in the melt (Duc-Tin and Keppler 2015). Studies that
couple experimentally-determined solubility equations of monazite with phase equilibria
modelling show that monazite dissolution increases during prograde metamorphism above the
solidus and that for UHT metamorphism most monazite will be completely consumed along the
prograde path (Kelsey et al. 2008; Yakymchuk and Brown 2014b). Based on these theoretical
models of monazite behaviour, no prograde monazite growth is expected above the solidus.
This contrasts with studies of natural rocks that record prograde suprasolidus monazite growth
(e.g., Hermann and Rubatto 2003; Hacker et al. 2015). Johnson et al. (2015) suggest that apatite
dissolution during prograde metamorphism may have contributed to LREE saturation of the
melt and resulted in prograde suprasolidus monazite crystallization. However, the role of apatite
breakdown and monazite growth in suprasolidus metamorphic rocks is still poorly understood.
Finally, similar to zircon, Ostwald ripening could produce prograde monazite growth in
suprasolidus metamorphic rocks (e.g., Nemchin and Bodorkos 2000). However, this has not
been extensively studied and may only apply to monazite growth just above the solidus where
the modal proportion of anatectic melt is relatively low.
Xenotime
Xenotime is a common phosphate in low- to high-grade metamorphic rocks (e.g., Franz et
al. 1996; Bea and Montero 1999; Pyle and Spear 1999; Spear and Pyle 2002). It can be directly
dated using U–Pb geochronology (e.g., Parrish 1990; Rasmussen 2005; Sheppard et al. 2007;
Janots et al. 2009; Crowley et al. 2009) and can be used as a thermometer when in equilibrium
with monazite and garnet (Pyle et al. 2001). Some of the proposed mechanisms of metamorphic
xenotime growth are: (1) growth due to dissolution/reprecipitation of detrital or diagenetic zircon
(Dawson et al. 2003; Rasmussen et al. 2011), (2) growth during to the breakdown of detrital
zircon (Franz et al. 2015), (3) growth during the breakdown of allanite and/or monazite (Janots et
al. 2008), (4) crystallization from anatectic melt (Pyle and Spear 1999; Crowley et al. 2009), and
(5) growth during the breakdown of Y-rich garnet (e.g., Pyle and Spear 1999).
There are three important controls on the stability of xenotime in metamorphic rocks.
First, higher bulk rock Yttrium concentrations allow xenotime to persist to higher pressures and
temperatures (Spear and Pyle 2010). Second, similar to monazite and zircon, partial melting will
result in xenotime consumption in order to contribute to REE and P saturation of the melt (e.g.,
Wolf and London 1995; Duc-Tin and Keppler 2015). Third, xenotime will be consumed during
the growth of garnet, which partitions Y and HREE (Pyle and Spear 1999, 2000; Spear and Pyle
2002). Consequently, xenotime in the matrix of high-grade garnet-bearing rocks is relatively rare
although xenotime inclusions may be present in garnet (e.g., Pyle and Spear 1999). By contrast,
xenotime may be present in the matrix at various metamorphic grades in garnet-absent rocks.
PHASE EQUILIBRIA MODELLING

Phase assemblages in metamorphic rocks generally change through continuous or
multivariant reactions rather than discontinuous or univariant reactions (e.g., Stüwe and
Powell 1995; Kelsey and Hand 2015). Petrogenetic grids display discontinuous reactions and
may yield the false impression that reactions occur over narrow intervals along a P–T path
(e.g., Vernon 1996). In reality, mineral (and melt) modes and compositions are continuously
changing along a P–T evolution and these variations represent the reaction sequence of the
rock. The particular reaction sequence experienced by a metamorphic rock is dependent
on bulk composition, the P–T path and whether the system is open or closed (e.g., White
and Powell 2002; Brown and Korhonen 2009; Yakymchuk and Brown 2014a). The phase
assemblages expected for a particular bulk rock composition over a range of P–T conditions
are commonly depicted using P–T pseudosections, phase diagrams constructed for a fixed
bulk composition (e.g., Hensen 1971; Spear et al. 2016). More detailed information about the
reaction sequences can be determined by using the intersection of a P–T path with isopleths
(or contours) of different variables (e.g., mineral modes and compositions) on a pseudosection
or by constructing mode–temperature or mode–pressure diagrams (e.g., White et al. 2011).
Here, we present pseudosections and mode-box diagrams (P/T–mode plots) for different
bulk chemical compositions to investigate the reaction sequences along common P–T paths
and we discuss the consequences for interpreting the ages obtained from accessory mineral
geochronology. The effects of fractionation of growing porphyroblasts and changes in bulk
composition are discussed later. Because we use phase equilibria modelling, it is assumed
that there are no kinetic controls (e.g., nucleation barriers or sluggish diffusion) that impact
the growth or dissolution of major and accessory minerals. Although this assumption may
not be strictly valid in some circumstances (Watt and Harley 1993; Pattison and Tinkham
2009; Gaides et al. 2011; Pattison and Debuhr 2015), the modelling here provides a general
framework for investigating reaction sequences.
Bulk compositions
Three bulk chemical compositions are modelled that represent the most common protoliths
investigated during most metamorphic studies. For the subsolidus and suprasolidus P–T paths, this
includes an average amphibolite-facies metapelite (Ague 1991) and an average passive margin
greywacke (Yakymchuk and Brown 2014a). These two compositions were chosen because they
are expected to dominate passive margin turbidite sequences that are involved in orogenesis
at convergent plate boundaries. A mid-ocean ridge basalt (MORB) from Sun and McDonough
(1989) was also used to investigate a suprasolidus reaction sequence for a typical mafic protolith.
The modelled bulk compositions are summarized in Table 1. Note that we do not consider melt
loss nor fractionation of minerals away from the reacting volume in the modelling here, both of
which can modify the composition of the equilibrium volume during a metamorphic evolution
(e.g., White and Powell 2002; Evans 2004; Guevara and Caddick 2016; Mayne et al. 2016).
Computational Methods
Forward phase equilibria modelling is used to evaluate the changes to metamorphic
mineral assemblages along several schematic P–T paths. There are two main approaches to
phase equilibria modelling. The first approach is to use Gibbs free energy minimization to
determine the stable mineral assemblage at a given P–T condition and there are two commonly
used software packages available for this approach, including Perple_X (Connolly and Petrini
2002) and Theriak–Domino (de Capitani and Brown 1987; de Capitani and Petrakakis 2010).
A second approach is to determine the solution of simultaneous non-linear equations to build
up an array of points and lines that make up a metamorphic phase diagram (Thermocalc:
Powell and Holland 1988, 2008; Powell et al. 1998). Both of these approaches require an
internally consistent thermodynamic database derived from the results of experimental
studies. There are several commonly-used databases available: Berman (1988), Holland and
Powell (1998) and most recently Holland and Powell (2011). Finally, the calculations require
activity–composition models that relate end-member proportions to end-member activities for
the solid-solution minerals as well as for complex fluids and melt.
Here, calculations were performed using Thermocalc v.3.40 (Powell and Holland
1988) using the internally consistent dataset of Holland and Powell (2011). For metapelite
and greywacke compositions, modelling was undertaken in the MnO–Na2O–CaO–K2O–
FeO–MgO–Al2O3–SiO2–H2O–TiO2–Fe2O3 (MnNCKFMASHTO) chemical system using the
activity–composition relations in White et al. (2014a,b). An average MORB composition was
investigated in the NCKFMASHTO chemical system using the activity-composition models
from Green et al. (2016). Phases modelled as pure end-members are quartz, rutile, titanite,
aqueous fluid (H2O), kyanite and sillimanite. Mineral abbreviations are from Holland and
Powell (2011) with the exception of titanite (ttn).
Table 1. Bulk compositions used for phase equilibria modelling (mol%)

Figures H2O SiO2 Al2O3 CaO MgO FeO K2O Na2O TiO2 MnO O
subsolidus
metapelite 3,4 + 64.58 13.65 1.59 5.53 8.03 2.94 2.00 0.91 0.17 0.60
greywacke 5,6 + 77.62 8.20 1.15 3.85 4.21 1.26 2.85 0.52 0.35 0.32
suprasolidus
metapelite 7,8,13 6.24 60.55 12.80 1.49 5.18 7.52 2.76 1.88 0.85 0.16 0.60
greywacke 9,10,13 2.61 75.35 7.96 1.12 3.74 4.09 1.22 2.77 0.51 0.34 0.31
MORB 11,12 6.19 42.58 10.45 12.27 14.76 9.68 0.22 2.62 0.66 – 0.58
Note: + H2O in excess, – not considered
Phase equilibria modelling of subsolidus and suprasolidus systems requires different

approaches to approximate the bulk H2O content of the system. For subsolidus phase equilibria
modelling, the amount of H2O in each bulk composition was set to be in excess, such that
aH2O = 1. By contrast, suprasolidus rocks are generally not considered to have excess H2O as
the very small amount of free H2O at the solidus is partitioned into anatectic melt (Huang and
Wyllie 1973; Thompson 1982; Clemens and Vielzeuf 1987; White and Powell 2002; White et
al. 2005). Along the solidus, the solubility of H2O in anatectic melt increases with pressure.
Therefore, for phase equilibria modelling of suprasolidus rocks, the amount of H2O in the bulk
composition must be set, and was adjusted so that the melt is just saturated with H2O at solidus
at 8 kbar. If the modelled prograde path crossed the solidus at lower or higher pressures, the
quantity of melt produced will be slightly overestimated and underestimated, respectively.
Modelled P–T paths
Several simplified typical P–T paths for subsolidus and suprasolidus systems are
modelled to investigate the reaction sequences, changes in mineral modes, amount of
melt generated and/or consumed, variations in bulk rock aTiO2 and the consequences for
interpreting the results of accessory mineral geochronology. Two subsolidus and two
suprasolidus P–T paths were investigated for the metapelite and greywacke compositions
that represent different tectonic evolutions. We also model one suprasolidus P–T path for a
MORB composition applicable to high-pressure mafic granulites.
For the subsolidus systems, the first P–T path has a clockwise trajectory and is
representative of collisional orogenesis (e.g., England and Thompson 1984; Thompson and
England 1984). An important feature of this P–T path is that peak P occurs before peak T,
which is consistent with a relatively long residence time in the core of an orogenic belt. The
model path (‘clockwise’ in Figs 3, 5) contains four segments: (1) heating and burial up to
10 kbar and 550 °C, (2) isobaric heating up to 650 °C, (3) decompression and minor heating to
8 kbar and 660 °C, and (4) decompression and cooling to 6.2 kbar and 610 °C.
The second subsolidus P–T path has a ‘hairpin’ trajectory where peak P and peak T nearly
coincide (e.g., ‘hairpin’ in Figs 3, 5). This style of P–T path is also common in collisional
orogenesis (e.g., Brown 1998; Kohn 2008) and reflects relatively short residence times at depth
and can represent crustal thickening in an accretionary zone of a propagating orogen (Jamieson
et al. 2004). The model hairpin P–T path contains three segments: (1) heating and burial up to
7 kbar and 610 °C, (2) heating and decompression to 6 kbar and 650 °C, and (3) decompression
and cooling to 4 kbar and 600 °C. Peak metamorphism is at the same temperature for both
subsolidus P–T paths, but the pressures at peak T are different.
Two P–T paths for suprasolidus rocks that are typical of granulite-facies metamorphism
are investigated. First, a clockwise P–T path is modelled (e.g., ‘clockwise’ in Figs 7, 9) that is
associated with crustal thickening and heating and is typical of many granulite terranes (e.g.,
Clark et al. 2011). This P–T path contains four segments: (1) increase in pressure and heating
across the solidus up to 9 kbar at 850 °C, (2) isobaric heating up to 900 °C, (3) isothermal
decompression to 7 kbar, and (4) cooling and decompression to 5 kbar at 750 °C.
Second, a suprasolidus counterclockwise (or anticlockwise) P–T path is considered (e.g.,
‘counterclockwise’ in Figs 7, 9) and is common in high temperature–low pressure terranes
(e.g., Clarke et al. 1987; Collins and Vernon 1991) and some ultrahigh temperature terrains
(e.g., Korhonen et al. 2013a, 2014). This P–T path reflects heating prior to thickening (usually
at peak T) followed by near-isobaric cooling. In some cases, this has been attributed to
the inversion of a hot back-arc basin during collisional orogenesis (Clark et al. 2014). The
modelled counterclockwise P–T path contains four segments: (1) isobaric heating up to 830 °C
and 5 kbar, (2) heating and an increase in pressure up to 860 °C and 6 kbar, (3) cooling and an
increase in pressure to 850 °C and 7 kbar, and (4) isobaric cooling to 750 °C.
A high-pressure clockwise path was also chosen to model the reaction sequences for high-
pressure mafic granulites (e.g., O’Brien and Rötzler 2003) and to investigate the growth and
breakdown of garnet for these rocks. This P–T path contains four segments: (1) heating and an
increase in pressure up to 12 kbar and 850 °C, (2) isobaric heating up to 950 °C, (3) isothermal
decompression to 8 kbar, and (4) cooling and decompression to 800 °C and 6 kbar. The
decompression segments of this P–T path are also compatible with the decompression segments of
some ultrahigh-pressure metamorphic rocks (Hacker et al. 2010; Chen et al 2013; Xu et al 2013).
Modelling suprasolidus zircon and monazite dissolution
For the suprasolidus P–T paths, the growth and dissolution of zircon and monazite are
modelled using the method of Kelsey et al. (2008). Note that the bulk composition used in
this modelling is fixed without fractionation of elements into growing porphyroblasts (e.g.,
zirconium into garnet) or loss of melt. These factors are discussed later. First, the saturation
concentrations of the melt in ppm are calculated as follows. The major element composition
of the anatectic melt is calculated at a specified P–T. This composition of the melt is combined
with the solubility equations of Boehnke et al. (2013) for zircon and Stepanov et al. (2012)
for monazite, and the stoichiometric concentrations of Zr in zircon (497,664 ppm Zr) and
LREE in monazite (566,794 ppm LREE) to determine the saturation concentrations of Zr and
LREE in ppm (Kelsey et al. 2008). These initial calculations do not account for the proportion
of anatectic melt or the bulk-rock concentrations of Zr and LREE. Second, the saturation
concentrations of Zr or LREE (ppm) are multiplied by the proportion of anatectic melt in
the system (in oxide mole proportion where each phase is normalized to a one oxide basis—
this is approximately equivalent to vol%) at P–T to arrive at concentrations in ppm that are
required to saturate the melt in the equilibration volume of the rock. Finally, these values
are divided by the assumed bulk-rock chemical concentrations of Zr and LREE. The Zr and
LREE in pelites and the greywackes are generally similar to each other and to the values
of metasedimentary migmatites; a value of 150 ppm was chosen here (e.g., Yakymchuk and
Brown 2014b). For MORB, the Zr concentration was set at 103 ppm, which is an average of
global MORB compositions (White and Klein 2014). The sensitivity of these calculations
to bulk rock concentrations of Zr and LREE and the assumptions and limitations of this
methodology are discussed by Kelsey et al. (2008) and Yakymchuk and Brown (2014b).
The result of the calculations is the proportion of zircon or monazite dissolution required to
saturate the anatectic melt in Zr and LREE in the equilibration volume of the rock. This value
is subtracted from 100% and the results are reported as the percent of zircon and monazite
remaining relative to the amounts existing at the fluid-present solidus for each P–T path.
One important limitation of the monazite saturation equation used here is that it does
not account for phosphorus in anatectic melt, which can affect monazite stability (e.g., Duc-
Tin and Keppler 2015). The main repository of phosphorus in most metamorphic rocks is
apatite, which is expected to break down during partial melting (Wolf and London 1994).
Phosphorus saturation in melt is a function of temperature, SiO2 concentration (Harrison
and Watson 1984) and the aluminum saturation index of the melt (ASI = molar [Al2O3] /
[K2O + Na2O + CaO], Wolf and London 1994). The solubility of apatite in melt increases with
rising ASI, which is common during prograde partial melting in migmatites (e.g., Johnson
et al. 2015). For example, an increase in the ASI from 1.1 to 1.2 increases the solubility of
apatite by an order of magnitude (Wolf and London 1994). The breakdown of LREE-rich
apatite during prograde metamorphism may contribute to LREE saturation of anatectic melt
and monazite crystallization (e.g., Johnson et al. 2015; Rocha et al. 2016). Here, we use melt
ASI as a qualitative tool to investigate the behaviour of apatite during partial melting and the
consequences for prograde monazite growth in suprasolidus metamorphic rocks.
SUBSOLIDUS PHASE RELATIONS AND REACTION SEQUENCES

Phase relations and reaction sequences for two subsolidus P–T paths for two bulk
compositions are discussed below for a closed system where bulk composition does not
change. Fractionation of porphyroblasts away from the reacting volume is not considered in
the calculations here.
Metapelite
Phase relations. The pseudosection for the amphibolite-facies metapelite is shown in
Figure 3. Biotite is stable at T > 450–460 °C with increasing pressure. Garnet is stable across most
of the diagram except at low P and low T and at P < 5 kbar just below the solidus. The staurolite-in
+ H2O + mu + q
Average Pelite g bi pl hem ru g bi pl ilm hem ru
12 1 6
g chl bi pa g chl bi pa 7 g bi
ep ttn ep ilm hem ru pl ilm ru
2 8
soli
ru
g bi pa g bi
g chl bi
d
ep ilm ru pa pl ilm ru
11 pa ab ep
us
ttn ru g bi pa pl
g chl bi
g chl bi pa ep ilm ru g bi pl ilm ru
pa ep ru
3 ep ilm ru
g bi g ky bi
10 pl ilm pl ilm ru
g bi m
pa ep ilm g bi
l ep il
pa pl ilm
g
pa p
c
ttn pa e l
ru p
g ky st bi pl ilm
Pressure (kbar)
ab g ch
ab hl b g ky bi
9 ep i p g chl bi
g bi
ru a pl ilm
m
pa ep ilm g bi pa l il
ma pl ilm
ap
e
n ru
ip
is
tb
kw g st bi
ep tt
gs
8 oc g bi pa ma pl ilm
cl
ep tt ab
n ru
g ky bi
ep ilm
g chl bi ab
l
pl ilm
g ch
g chl bi pa mt
14
ab ep ilm ru g st bi g ky st bi
15 pl ilm mt pl ilm mt
g chl bi
7
u
g ch ilm
ab e
r
ab ep ru
ilm
4 g bi pa ma
l bi p
p
13 pl ilm mt
ep
m g ch
ab
mt
ap lb
12 l il i p
l bi
g c ep
5
m
m a
ab
l il
hl ilm
g ch
m
6 9
ap
t
bi
ip
gc ap t
tb
t
m
g sill bi
m m
hl l
gs
11
ilm
ilm
bi
pl ilm
10 16
pl
g sill bi
m
bi
p il
st
pl ilm mt
u
g chl bi
5
mr
l
sil
be
chl bi
in pl ilm mt
g
irp
p il
bi a
ab ep ru
19 ha 21
be
mt
chl
sill bi sill bi
bi a
17 l ilm pl ilm pl ilm

ip
20 st b
chl
18
4
450 500 550 600 650 700
Temperature (˚C)
1) g chl bi pa ep ttn 6) g chl bi pa ep hem ru 11) g chl bi pl ep ilm mt 16) g st chl bi pa pl ilm mt
2) g chl bi pa ab ep ttn 7) g bi pa ep hem ru 12) g chl bi pa pl ep ilm 17) chl bi pl ab ep ilm mt
3) g chl pa ab ep ttn 8) g bi pa ep ilm hem ru 13) g chl bi pa ma ep ilm 18) chl bi pl ab ep ilm
4) chl ab ep ttn ru 9) g chl bi pa pl ab ep ilm 14) g bi pa ma ep ilm mt 19) chl bi pl ep ilm mt
5) chl bi ab ep ttn ru 10) g chl bi pl ab ep ilm 15) g bi pa ma pl ep ilm mt 20) chl bi pl ilm mt
21) g st chl bi pl ilm mt
Figure 3. P–T pseudosection for the subsolidus metapelite. The solidus is shown by the heavy dashed line.
field boundary extends from 550 °C at 4 kbar up to 650 °C at 10 kbar. The stability field of the
aluminosilicate minerals (kyanite and sillimanite) ranges from T > 580 °C at 4 kbar to T > 660 °C
at 10 kbar. Rutile is stable at low P at low temperatures and at P > 10.5 at T > 550 °C.
Reaction sequence for the clockwise P–T path. A mode-box for the reaction sequence for
the clockwise P–T path is shown in Figure 4a. During the prograde segment of the clockwise
P–T path, chlorite is progressively consumed to produce biotite (starting at 455 °C) and then
garnet (starting at 460 °C). Chlorite is exhausted by 580 °C. Rutile is completely consumed
by ~530 °C. Epidote breaks down along the prograde path and the epidote-out field boundary
is encountered at 620 °C. The complete breakdown of epidote at 620 °C may result in minor
zircon crystallization or more likely the growth of new rims on pre-existing zircon.
The staurolite stability field is encountered during the heating and decompression segment
at 9 kbar and 650 °C. In natural examples, prograde monazite growth has been associated with
garnet, staurolite and kyanite/sillimanite growth in metapelites (Catlos et al. 2001; Smith and
Barriero 1990; Wing et al. 2003; Kohn and Malloy 2004; Fitzsimons et al. 2005; Corrie and
Kohn 2008), although kyanite/sillimanite is not encountered along the P–T path modelled here.
(a) Metapelite: clockwise P–T max T
ttn ru max P ilm prograde retrograde mt

normalized molar proportion and aTiO2
1.0
a TiO2
q
0.8
ab
pa pl
0.6
mu
0.4
0.2 bi
ep
chl g st
450 500 550 600 650 610
T (°C)
(b) Metapelite: hairpin P–T max P max T
prograde retrograde mt
1.0
ilm
q
a TiO2
0.8
sill
0.6 ma pl
pa
mu
0.4 mu
0.2 bi
chl g st
500 520 540 560 580 600 620 640 640 620 600
T (°C)
Figure 4. (a) Mode-box diagram and aTiO2 for the subsolidus clockwise P–T path for the metapelite. (b)
Mode-box diagram and aTiO2 for the subsolidus hairpin P–T path.
Figure 4.
In general, monazite could grow at different points along the prograde P–T path. Because
garnet growth is predicted to be continuous up to the staurolite-in field boundary, new monazite
grown during this segment of the P–T path is expected to have low concentrations of HREE
and Y; these elements are expected to partition into garnet. Staurolite growth is at the expense
of garnet and occurs during the decompression portion of the prograde path (e.g., Florence and
Spear 1993). Monazite formed during this segment is expected to have elevated HREE and Y
concentrations due to the breakdown of garnet and the age would record the initial stages of
decompression from peak P. By contrast, zircon is expected to undergo only minor changes to
its mode and will be slightly consumed during subsolidus prograde metamorphism (e.g., Kohn
et al. 2015). Two possible exceptions are minor zircon growth during the breakdown of Zr-rich
epidote and/or rutile. However, at low temperatures, rutile is not expected to be zirconium rich.
The stable mineral assemblage at the metamorphic peak of 660 °C and 8 kbar includes garnet,
staurolite, biotite, plagioclase, ilmenite, muscovite and quartz. In general, the mode of monazite
in metapelites is expected to reach a maximum at peak P–T and monazite is not expected to grow
during cooling (e.g., Spear and Pyle 2010). An additional factor is that monazite is susceptible
to dissolution/reprecipitation in the presence of fluids (e.g., Williams et al. 2011). While fluid-
mediated dissolution/reprecipitation should not increase the net mode of monazite, it can occur at
almost any point along the P–T evolution. Therefore, it is possible that monazite in a subsolidus
metapelite can record a range of ages that vary from monazite-in up to the metamorphic peak
and dissolution/precipitation may result in ages that record retrogression (Harlov et al. 2005;
Williams et al. 2011). On the other hand, zircon is expected to be relatively unreactive and may
not record prograde to peak metamorphism in subsolidus metamorphic rocks.
After the exhaustion of rutile during the heating and burial segment the value of aTiO2
decreases to 0.95 initially and remains nearly constant until the decompression and heating
segment, where aTiO2 drops to 0.8. During the subsequent decompression and cooling segment
the aTiO2 value remains stable at 0.8. After rutile is exhausted, any growth of new zircon or
modification of existing zircon will occur when aTiO2 < 1.0; this needs to be considered when
applying the Ti-in-zircon thermometer.
Reaction sequence for the hairpin P–T path. A mode-box for the reaction sequence
for the hairpin P–T path is shown in Figure 4b. The prograde reaction sequence includes
garnet-in at 520 °C and 4.5 kbar, staurolite-in at 590 °C and 6.5 kbar and sillimanite-in at
640 °C and 6 kbar. Garnet both grows and is consumed during the prograde portion of the
reaction sequence. The value of aTiO2 decreases from 0.88 at 500 °C to 0.79 at peak T. For the
decompression and cooling segment, a small amount of biotite grows at the expense of garnet;
the modes of the other minerals are little changed and the aTiO2 value remains close to 0.79.
As with the clockwise P–T path, monazite growth may occur at various times during
the prograde segment of the hairpin P–T path. Because the mode of garnet increases and
decreases multiple times during the prograde reaction sequence, linking Y and HREE in
monazite would be challenging in the absence of microstructural context, such as monazite
in coronae surrounding resorbed garnet. For example, HREE and Y-rich monazite could be
generated during: (1) the heating and burial segment during garnet breakdown to staurolite,
(2) the heating and decompression segment during garnet breakdown to biotite near the
metamorphic peak, or (3) garnet breakdown during the retrograde segment of the P–T path.
Furthermore, xenotime breakdown during prograde metamorphism would also contribute to
the growth of Y and HREE-rich monazite with or without a contribution from garnet (e.g.,
Spear and Pyle 2010). Therefore, for the hairpin P–T path modelled here, HREE- or Y-rich
monazite can record burial and/or exhumation.
Greywacke
Phase relations. The pseudosection for the greywacke is shown in Figure 5. Garnet is
stable across the entire diagram. Titanite is stable at T < 460–470 °C with increasing pressure.
Biotite is stable across the diagram except at the high-P–low-T corner. Chlorite is unstable
above 600 °C at 7 kbar and unstable above 530 °C at 12 kbar. Albite is stable below 565 °C at
4 kbar and below 620 °C at 11 kbar. White mica (muscovite and/or paragonite) is stable up to
the solidus at P > 10 kbar and at T < 570 °C at 4 kbar. There are two important sets of quasi-
linear field boundaries with positive slopes in P–T space. First is the set of boundaries that
range from 480 °C at 4 kbar up to 640 °C at 12 kbar represent the breakdown of epidote to
produce plagioclase. A second set of fields extends from 570 °C at 4 kbar to the solidus with
increasing P that represent the consumption of paragonite to produce sillimanite or kyanite.
Average passive margin greywacke + H2O + q

12 g bi mu
g ep bi
mu chl hem pa
pa ab g ep bi ab ru
ttn hem chl g ep bi mu hem pa ab ru
pa ab ru
11 g ep bi g pl bi
mu chl g ep bi 10 mu hem
pa ab ru hem pa g ep pl bi mu
solidu
hem pa pa ru
g ep mu ttn ab ru 12
ab ru
chl pa ab 11
ttn 13
10
s
g ep bi
1 chl pa
ab ru 5 14
4 7
g ep bi 8 9 15
mu chl pa
9 ab ru g ep bi 6 g pl bi mu ilm pa
2 ilm pa
ab ru
Pressure (kbar)
g ep e 16
mu chl is g ep bi
ab kw chl hem g pl bi
8 ttn loc pa ab ru ilm pa
C ab
g pl bi ilm pa g pl bi
mu ilm ky
24
7
23
g ep bi g pl bi chl g pl bi
3 mu chl ab 20 ilm pa ab chl ilm
ru 19 pa g pl bi
21 ilm sill
6 22
27
26
g ep bi mu g pl bi 28
chl ab ru ttn mu chl ilm
ab
5 29 g pl bi ilm mt sill
25
18
irpin
ha
17 30 g pl bi cd ilm mt sill
4
450 500 550 600 650 700
Temperature (˚C)
1. g ep mu chl pa ab ru ttn 11. g pl bi mu ilm hem pa ab ru 21. g pl ep bi mu chl ilm pa ab ru

2. g ep mu chl ab ru ttn 12. g pl bi mu ilm hem pa ru 22. g pl bi mu chl ilm pa ab
3. g ep bi mu chl ab ttn 13. g pl bi mu ilm pa ru 23. g pl ep bi chl ilm pa ab
4. g ep bi chl ilm hem pa ab ru 14. g pl bi mu ilm pa ru ky 24. g pl ep bi chl ilm pa ab ru
5. g ep bi ilm hem pa ab ru 15. g pl bi mu ilm ru ky 25. g pl bi chl ilm pa mt sill
6. g pl bi ilm pa ab ru 16. g pl bi mu ilm pa ky 26. g pl bi chl ilm pa mt
7. g ep pl bi mu ilm hem pa ab ru 17. g pl ep bi mu chl ab ru 27. g pl bi ilm pa mt
8. g pl bi mu ilm pa ab ru 18. g pl ep bi mu chl ilm ab 28. g pl bi ilm pa sill
9. g pl bi mu ilm pa ab 19. g ep bi mu chl ilm ab ru 29. g pl bi ilm pa mt sill
10. g pl bi mu hem pa ab ru 20. g ep bi mu chl ilm pa ab ru 30. g pl bi chl ilm mt sill
Figure 5. P–T pseudosection for the subsolidus greywacke. The solidus is shown by the heavy dashed line.
Figure 5
Reaction sequence for the clockwise P–T path. A mode-box for the assemblage sequence
for the clockwise P–T path is shown in Figure 6a. For the clockwise P–T path the important
changes along the prograde reaction sequence include: (1) the growth of rutile commencing
at 460 °C resulting in aTiO2 increasing from 0.9 at 450 °C to 1.0 at 460 °C, (2) the complete
consumption of titanite by 475 °C, (3) the growth of biotite and garnet at the expense of chlorite
and muscovite from 450 °C to 550 °C, which results in the complete consumption of chlorite
by 550 °C, (4) the breakdown of epidote to produce 2 mol% plagioclase at ~605 °C, (5) the
consumption of ~0.5 mol% garnet to produce plagioclase and hematite from 605 °C to 620 °C,
and (6) the growth of 0.5 mol% garnet from 620 °C to 650 °C. Similar to the metapelite, the
breakdown of Zr-rich epidote may result in minor zircon growth.
(a) Greywacke: clockwise P–T

max T
ttn max P
prograde retrograde
1.0
ru, ilm, hem ilm
a TiO2
0.8 q
0.6
ky
ab
0.4
pl
pa
mu mu pa
0.2
ep bi
chl
g
450 500 550 600 650 610
T (°C)
(b) Greywacke: hairpin P–T max P max T

prograde retrograde
1.0
ilm
a TiO2
0.8
q
0.6 ky sill
ab
0.4 pl
mu pa
0.2 mu
bi g
chl
500 520 540 560 580 600 620 640 640 620 600
T (°C)
Figure 6. (a) Mode-box diagram and aTiO2 for the subsolidus clockwise P–T path for the greywacke com-
position. (b) Mode-box diagram and aTiO2 for the subsolidus hairpin P–T path.
Decompression and heating from 10 kbar at 650 °C to 8 kbar at 660 °C results in: (1)
the disappearance of rutile at 9.8 kbar and a subsequent drop in aTiO2, and (2) the breakdown
of paragonite and garnet to produce kyanite, biotite and plagioclase at 8.8 kbar. The peak
metamorphic assemblage at 660 °C and 8 kbar includes garnet, kyanite, muscovite, biotite,
plagioclase and ilmenite. Further decompression and cooling results in the growth of biotite at
the expense of garnet and a gradual decrease in aTiO2.
Like the metapelite, garnet breakdown occurs during a portion of the heating and burial
segment of the P–T path (at ~600 °C) and during the heating and decompression segment
immediately before the metamorphic peak. Therefore, Y- and HREE-rich monazite associated
with garnet breakdown could record the timing of burial and/or exhumation.
Reaction sequence for the hairpin P–T path. A mode-box for the reaction sequence for the
hairpin P–T path is shown in Figure 6b. The hairpin P–T path has three notable differences in
its reaction sequence compared with the clockwise path for the greywacke. First, the P–T path
does not intersect the epidote stability field. Second, the value of aTiO2 is always less than one and
decreases from 0.95 to 0.80 along the P–T path. And third, white mica is completely consumed
by the end of the P–T path. In contrast to the multiple garnet growth/consumption segments
along the clockwise P–T path, the mode of garnet increases during burial and decreases during
decompression for the hairpin path. Therefore, Y- and HREE-depleted monazite is predicted
to record burial and monazite enriched in these elements would document garnet-breakdown
during decompression and/or cooling and may also record xenotime breakdown.
SUPRASOLIDUS PHASE RELATIONS AND REACTION SEQUENCES

Phase relations and reaction sequences for three suprasolidus scenarios are discussed
below for a closed system where bulk composition does not change and the water content
is fixed to just saturate the rock in H2O at 8 kbar at the solidus. This approach provides
important first-order constraints on the reaction sequences for the modelled compositions;
it does not take into account melt loss and the associated effects on rock fertility, solidus
temperature and zircon and monazite stability. The consequences of open system behaviour
on the reaction sequence and accessory mineral stability are discussed later.
Metapelite
Phase Relations. The pseudosection for the metapelite is shown in Figure 7. Ilmenite and
plagioclase are stable across the entire diagram. Garnet is stable at high pressures across the
diagram except at T < 700 °C at P < 5.5 kbar. Cordierite is stable at high-T–low-P conditions.
Orthopyroxene is not stable in this diagram. Rutile is restricted to P > 10–11.5 kbar across the
modelled temperature range. The three important partial melting reactions for the metapelite
include: (1) the consumption of any free aqueous fluid to produce melt at the wet solidus, which
ranges from 660 °C at low pressures to 710 °C at high pressure, (2) the incongruent breakdown
of muscovite to produce K-feldspar, which is represented by a narrow low-variance field that
extends from the wet solidus at low P to T > 780 °C at P > 12 kbar, and (3) the progressive
breakdown of biotite at temperatures above the muscovite stability field to produce either
garnet at higher pressure or cordierite ± garnet at lower pressure. Biotite is exhausted by 850 °C
at P > 7 kbar and melting progresses via the consumption of quartz and feldspar.
Reaction sequence for the clockwise P–T path. The mode-box and titania activity
for the clockwise assemblage sequence are shown in Figure 8a. The predicted amount of
monazite and zircon remaining as well as melt ASI are shown in Figure 8b. During the
Average Pelite + liq + ilm

12 2
1 g ky bi ksp g ky ksp 3
g ky bi mu pl 4
pl ru q pl ru q
ru q g sil ksp pl
ru q
11
g ky bi mu
ksp pl q
g ky bi
10 g ky bi mu ksp pl
pl q q
g sill ksp
g sill bi
pl q
ksp pl q
9
g sill bi ksp
Pressure (kbar)
pl q
8 e
ck wis
clo
7
g sill bi 15 16
g sill bi cd
mu pl 18
ksp pl q
q 17
14 19
6 g sill bi 13
g sill cd
ksp pl
ksp pl q
mt q
5 6 8 9 10 g cd ksp
sp pl g cd sp
5 counterclockwise g cd g cd pl
g cd bi ksp pl 11
sill bi ksp pl
ksp pl mt q
ksp pl q 7 mt 12
mt q
4
650 700 750 800 850 900
Temperature (°C)
1: g bi mu pl ru q 8: g cd bi ksp pl q 15: g sill ksp pl

2: g ky bi mu ksp pl ru q 9: g cd ksp pl q 16: g sill pl
3: g sill ksp ru q 10: g cd ksp pl
17: g sill cd pl
4: g sill ksp q 11: g cd ksp sp pl mt
5: sill bi mu pl q 12: g cd sp pl mt 18: g sill sp pl
6: sill ksp bi mu pl q 13: g sill cd ksp sp pl 19: g sill cd sp pl
7: g sill cd bi ksp pl mt q 14: g sill cd ksp pl
Figure 7. P–T pseudosection for suprasolidus metapelite. The solidus is shown by the heavy dashed line.
prograde segment of the clockwise P–T path, the metapelite begins to melt at the wet solidus
at ~680 °C through the consumption of any free H2O as well as quartz and plagioclase. The
amount of aqueous fluid at the solidus is expected to be small given the limited porosity of
high-grade metamorphic rocks (e.g., Yardley and Valley 1997). Therefore, the amount of
melt produced at the wet solidus is expected to be limited. For the modelled metapelite, the
amount of melt produced is 4 mol%, although this is mostly an artefact of the modelling
which assumes H2O saturation of the solidus at 8 kbar.
Metapelite: clockwise P–T

H 2O ilm prograde retrograde mt
1.0
(a) q
q a TiO
0.8 2
proportion and aTiO2
ksp sill
normalized molar
sill g
pl
0.6 cd
pl
0.4 mu liq bi
bi
0.2
(b) 100 1.25

80
zircon ASI
melt
% remaining
1.20
ASI of melt
60
solidus
monazite
40
1.15
20
0 1.10
650 700 750 800 850 9 8 7 850 800 750
Pressure (kbar)
Temperature (°C) Temperature (°C)
@ 900°C
Metapelite: counterclockwise P–T

H2O mt prograde retrograde ilm
(c) 1.0
q q
0.8 ksp sill
sill pl
normalized molar
0.6 a TiO2 cd
pl
g
0.4 mu bi
bi
0.2
liq
100 1.25
(d)
80
% remaining
1.20
ASI of melt
60 monazite
solidus
40 ASI
melt zircon
1.15
20
1.10
650 700 750 800 850 850 800 750
Temperature (°C)
Figure 8. (a) Mode-box diagram and aTiO2 for the suprasolidus clockwise P–T path for the metapelite.
(b) Amount of zircon and monazite remaining relative to the amount at the solidus and the ASI value of
Figure 8
melt. (c) Mode-box diagram and aTiO2 for the counterclockwise P–T path for the metapelite. (d) Amount
of zircon and monazite remaining relative to the amount at the solidus for the counterclockwise P–T path.
After any free H2O is consumed, partial melting continues with increasing temperature
through the consumption of muscovite up until 728 °C. This produces an additional 3 mol%
melt with a generally constant melt ASI. At 728 °C and 7.5 kbar, melting proceeds via the
incongruent breakdown of muscovite to generate K-feldspar over a narrow (~2 °C) temperature
range. This narrow field produces an additional 3 mol% melt. After muscovite is exhausted,
the rock contains ~9 mol% melt and approximately 8% of the zircon and 12% of the monazite
that was present at the solidus is predicted to have been consumed. A minor amount of apatite
breakdown is expected in order to saturate the anatectic melt in phosphorus.
After muscovite is completely consumed, partial melting continues through the

consumption of biotite, plagioclase and quartz to produce melt, garnet, K-feldspar and ilmenite.
This produces an additional 32 mol% melt. Because biotite is an important host of accessory
mineral inclusions (e.g., Watson et al. 1989), biotite breakdown may release inclusions of
monazite and zircon into the reaction volume of the rock. The liberation of these accessory
minerals may contribute to LREE and Zr saturation of the anatectic melt. On the other hand,
some zircon and monazite may be included in growing garnet and will be sequestered away
from the reaction volume. These minerals will be shielded from dissolution and are more
likely to preserve inherited (or detrital) ages as well as any subsolidus to early suprasolidus
prograde metamorphic ages. The inclusion of accessory minerals in major minerals effectively
reduces the Zr and LREE available to the system and proportionally more zircon and monazite
dissolution will be required to maintain melt saturation in these elements (e.g., Yakymchuk and
Brown 2014b). Therefore, the sequestration of zircon and monazite in stable peritectic minerals
such as garnet will promote the dissolution of accessory minerals along grain boundaries in the
matrix of the rock with increasing T.
Titania activity is predicted to reach its highest value of 0.92 at 850 °C. At this point,
all of the monazite and 85% of the zircon are expected to be consumed. However, during
this interval, the ASI of the melt increases from 1.10 to 1.15, which may result in enhanced
apatite dissolution. For example, Pichavant et al. (1992) estimate that a similar increase in
ASI at 800 °C and 5 kbar would change the P2O5 concentration of melt from 0.50 to 0.75
wt%. If apatite is LREE rich, the dissolution of apatite may delay the complete dissolution of
monazite to higher temperatures or, in extreme cases, may even promote prograde monazite
crystallization (e.g., Johnson et al. 2015).
After the complete consumption of biotite, the hydrous minerals have been exhausted and
the residuum is essentially composed of anhydrous minerals. Melting continues through the
continued consumption of quartz and K-feldspar up to the modelled peak temperature of 900 °C.
The ASI of the melt reaches 1.18 at peak T and significant apatite dissolution is likely. At this
point, the metapelite contains 43 mol% melt and both monazite and zircon are absent. After
zircon and monazite are completely consumed, further anatexis is expected to generate melt that
is undersaturated in Zr and LREE. Therefore, even with the breakdown of LREE-rich apatite,
prograde monazite crystallization is not expected above ~850 °C.
Isothermal decompression from 9 to 7 kbar produces an additional 11 mol% melt at the
expense of quartz and K-feldspar. At this point the rock contains the maximum amount of melt
of 54 mol%. During this decompression segment, melt ASI increases from 1.18 to 1.23 and aTiO2
decreases from 0.86 to 0.74. Approximately 1 mol% garnet is consumed during decompression,
which would liberate some HREE and Y into the reaction volume. If significant apatite dissolution
occurs and promotes monazite crystallization during decompression, this may be reflected as
elevated HREE concentrations in monazite. However, the melt is undersaturated in LREE given
the complete exhaustion of monazite at 850 °C so new monazite growth is unlikely unless the
apatite is very enriched in LREE or the modal proportion of apatite is high.
Decompression and cooling from peak T results in: (1) melt crystallization, (2) the growth
of K-feldspar and quartz until the cordierite stability field is reached at ~880 °C and 6.7 kbar,
(3) garnet and cordierite consumption to produce biotite, (4) new zircon and monazite growth,
(5) a decrease in melt ASI and aTiO2. Melt crystallization and cordierite growth are concomitant
with the consumption of garnet and sillimanite; this is a common reaction sequence for high-
temperature decompression in migmatites. Zircon or monazite produced over this reaction
interval is expected to be enriched in HREE and Y (e.g., Yakymchuk et al. 2015). Protracted
monazite and zircon growth will occur from peak T to the solidus. U–Pb zircon and monazite
ages that spread down Concordia have been interpreted to record protracted growth during
cooling and melt crystallization in various migmatite terranes with clockwise P–T evolutions
(Korhonen et al. 2012; Reno et al. 2012; Morrisey et al. 2014; Walsh et al. 2015).
Reaction sequence for the counterclockwise P–T path. The mode-box and titania
activity for the counterclockwise assemblage sequence are shown in Figure 8c. The amount of
monazite and zircon remaining as well as melt ASI are shown in Figure 8d. During isobaric
heating at 5 kbar, partial melting begins at the wet solidus at ~670 °C and continues through
the breakdown of muscovite to produce a total 8 mol% melt by 685 °C. Melting continues
via the progressive breakdown of biotite to produce peritectic garnet and K-feldspar. By
765 °C, 17 mol% melt is present and roughly 80% and 60% of the initial amount of zircon and
monazite, respectively, remains. From the solidus up to 765 °C, the aTiO2 value of the system
has decreased from 0.9 to 0.6 and the ASI of the melt has increased from 1.13 to 1.18. An
increase in melt ASI increases the solubility of apatite; this may liberate some LREE and P that
could contribute to minor monazite crystallization because the melt is predicted to be saturated
with respect to the LREE. Similar to the clockwise P–T path, the breakdown of biotite may
liberate zircon and/or monazite that was sequestered away from the reacting volume.
Cordierite enters the phase assemblage at ~770 °C and melting continues through
the breakdown of biotite, sillimanite and plagioclase to produce K-feldspar, garnet and
cordierite. Sillimanite is completely consumed by 780 °C and biotite is exhausted by 800 °C.
During the interval from 770 °C to 800 °C, approximately 15 mol% melt is generated and
melt ASI decreases slightly from 1.18 to 1.16. Although the change in ASI would decrease
the solubility of apatite, this may be counteracted by the additional melt generation during
biotite breakdown and apatite growth is not expected.
In the absence of biotite, melting continues through the breakdown of quartz,
plagioclase and K-feldspar, which results in a progressively drier melt. Monazite and zircon
are predicted to be completely consumed by 820 °C and 850 °C, respectively. The increase
in pressure near the metamorphic peak results in minor cordierite consumption to produce
garnet. At the metamorphic peak, the metapelite has generated ~44 mol% melt.
Isobaric cooling at 6.5 kbar from 860 °C to 750 °C results in: (1) melt crystallization;
(2) new zircon and monazite growth commencing at 855 °C and 825 °C, respectively;
(3) a decrease in melt ASI, which would decrease the solubility of apatite and contribute
to apatite crystallization; (4) the retrogression of cordierite and garnet to biotite; (5) the
consumption of K-feldspar and the growth of plagioclase; and (6) a decrease in aTiO2. The
concurrent breakdown of garnet and growth of zircon, monazite and plagioclase may result
in elevated HREE and Y concentrations and more pronounced negative Eu anomalies in
newly crystallized accessory minerals. New zircon growth occurs when the aTiO2 value of
the system ranges from 0.6 to 0.7. Applying the Ti-in-zircon thermometer assuming an
aTiO2 value of 1.0 would underestimate the true temperature by ~50 °C (Fig. 1b). Protracted
retrograde monazite and zircon growth is expected during melt crystallization along the
isobaric cooling segment of the counterclockwise P–T path (e.g., Korhonen et al. 2013b).
Greywacke
Phase relations. The pseudosection for the greywacke is shown in Figure 9. When
compared with the metapelite, the greywacke composition is less fertile (e.g., Clemens
and Vielzeuf 1987; Thompson 1996; Vielzeuf and Schmidt 2001; Johnson et al. 2008;
Yakymchuk and Brown 2014a), contains different mineral assemblages, and yields different
reaction sequences. For the greywacke, quartz, ilmenite, plagioclase and garnet are stable
across the entire diagram. Cordierite becomes stable at pressures of 5–7 kbar with increasing
temperature. Orthopyroxene is stable at P < 6 kbar and T > 800 °C. Rutile is stable at
P > 9–11 kbar. Muscovite is stable at P > 7–12 kbar with increasing temperature. Similar to
the metapelite, partial melting of the greywacke begins at the wet solidus at temperatures
of ~670–700 °C. A narrow low-variance field representing the breakdown of muscovite to
produce K-feldspar is restricted to P > 10.8 kbar; this contrasts with a similar field in the
metapelite pseudosection that extends from < 4 to > 12 kbar (Fig. 7). For the greywacke
Average Passive Margin Greywacke + q + ilm + liq

12
g pl ksp
g pl bi ksp ky ru
g pl bi mu ky ru
ky ru
11
g pl ksp
sill ru
10 g pl g pl
bi mu bi ky
ky
g pl bi ksp
g pl bi sill ru
9 mu ky
H 2O g pl ksp sill
g pl bi
P (kbar)
e
wis
8 ksp sill
ck
clo g pl ksp g pl
g pl bi cd
g pl bi sill cd sill cd sill
ksp sill
7 g pl sill
g pl bi
sill H2O
g pl ksp bi
g pl ksp cd
cd sill g pl cd
6
g pl ksp
bi cd g pl ksp
5 counterclockwise opx cd
g pl ksp g pl opx cd
g pl bi bi cd mt g pl bi ksp
cd sill opx cd
4
650 700 750 800 850 900 950
Temperature (˚C)
Figure 9. P–T pseudosection for the suprasolidus greywacke. The solidus is shown by the heavy
dashed line.
composition, biotite breakdown generates garnet at high pressure, cordierite at low pressure,
and orthopyroxene at low pressure and high temperature.
Reaction sequence for the clockwise P–T path. The mode-box and titania activity for the
clockwise assemblage sequence are shown in Figure 10a. The predicted amount of monazite
and zircon remaining along with the ASI of melt are shown in Figure 10b. Along the clockwise
P–T path, partial melting begins at ~675 °C at the wet solidus and muscovite is not stable.
Therefore, further melting proceeds through the breakdown of biotite and sillimanite to
produce garnet. Similar to the metapelite, growing garnet has the potential to capture inherited
or prograde monazite and zircon allowing their preservation to higher temperatures. K-feldspar
becomes stable at 800 °C and 8.2 kbar and biotite is completely consumed by 850 °C. Up to
this point, the greywacke has produced 14 mol% melt, which is less than the 34 mol% melt
generated by the pelite for the same P–T path. Up to 850 °C, melt is produced gradually for
the greywacke composition whereas melting of the metapelite occurs as a pulse in the narrow
muscovite–K-feldspar field followed by more gradual melting due to biotite breakdown.
Greywacke: clockwise P–T

prograde retrograde
1.0
(a) H2O ilm mt
0.8 a TiO2
q
normalized molar
q
0.6
sill
pl sill
0.4 pl ksp
g
0.2 bi
bi cd
liq
(b) 100 1.25

80 zircon
A SI
melt
% remaining
1.20
ASI of melt
60
solidus
monazite
40 1.15
20
0 1.10
650 700 750 800 850 900 900 850 800 750
9 8 7
Temperature (°C) Pressure (kbar) @ 900°C Temperature (°C)
Greywacke: counterclockwise P–T

prograde retrograde
(c) 1.0 ilm
H 2O
q
0.8 q
normalized molar
a TiO2
0.6 sill
sill
0.4 ksp pl
pl
0.2 opx g bi
cd
bi
liq
100
(d) zircon 1.25
80
1.20
% remaining
ASI of melt
60 elt ASI
solidus
meltmASI monazite
40
1.15
20
0 1.10
650 700 750 800 850 850 800 750
Temperature (°C)
Figure 10. (a) Mode-box diagram and aTiO2 for the suprasolidus clockwise P–T path for the greywacke.
(b) Amount of zircon and monazite remaining relative to the amount at the solidus and the ASI value of
melt. (c) Mode-box diagram and aTiO2 for the counterclockwise P–T path for the greywacke. (d) Amount
of zircon and monazite remaining relative to the amount at the solidus for the counterclockwise P–T path.
Beyond the loss of biotite from the stable assemblage, melting continues through the
breakdown of K-feldspar, plagioclase and quartz and an additional 4 mol% melt is produced
by 900 °C. Although monazite dissolution is modelled to continue from 850 to 900 °C,
some prograde monazite crystallization may be possible in this temperature range for three
reasons. First, the melt is saturated with respect to LREE due to progressive monazite
dissolution. Second, after the exhaustion of biotite the melt becomes progressively drier,
which decreases the solubility of monazite. Third, melt ASI increases, which increases the
dissolution of apatite and may liberate enough LREE to support new monazite growth.
However, unless this monazite is sequestered away in a growing peritectic mineral (such as
garnet) it is expected to be consumed during further heating and partial melting.
All of the monazite and ~65% of the zircon are consumed by 900 °C. Isothermal
decompression from 9 kbar to 7 kbar produces an additional 4 mol% melt and results in minor
(<1 mol%) garnet consumption. Melt ASI increases from 1.18 to 1.23 during decompression,
which enhances the solubility of apatite. However, the melt is undersaturated with respect to
LREE and new monazite growth is not expected during decompression. Approximately 10%
of the initial amount of zircon present at the solidus is consumed during decompression and
titania activity decreases from 0.95 to 0.85.
Cooling and decompression from peak T to 750 °C results in: (1) melt crystallization,
(2) the consumption of garnet and sillimanite to produce cordierite commencing at 890 °C
and biotite starting at 840 °C, (3) a decrease in aTiO2 from 0.85 to 0.65, (4) a decrease in
melt ASI from 1.23 to 1.16, (5) the growth of new zircon (likely as overgrowths on existing
zircon), and (6) the crystallization of monazite starting at ~885 °C. Zircon and monazite
growth occurs during garnet breakdown and these accessory minerals are expected to
have elevated Y and HREE concentrations. The application of Ti-in-zircon thermometry
should use an aTiO2 value ranging from 0.85 to 0.65. Assuming an aTiO2 value of 1.0 would
underestimate temperatures by up to 50 °C (Fig. 1b).
Reaction sequence for the counterclockwise P–T path. The mode-box and titania
activity for the counterclockwise assemblage sequence are shown in Figure 10c. The amount
of monazite and zircon remaining as well as melt ASI are shown in Figure 10d. Similar to
the clockwise P–T path for the greywacke, the prograde segment of the counterclockwise
P–T path generates melt gradually in contrast to the more pulsed melting in the metapelite.
Melting commences at the wet solidus and proceeds via the breakdown of biotite to produce
cordierite and garnet followed by K-feldspar at 760 °C. Orthopyroxene enters the phase
assemblage at 825 °C and biotite is completely consumed by 830 °C. The value of aTiO2
decreases from ~0.80 to 0.60 up to the orthopyroxene-in field boundary and then increases
for the remainder of the prograde path. The increase in pressure near peak T results in the
breakdown of orthopyroxene to produce garnet (e.g., White et al. 2008). By the end of the
heating segment, ~50% of the zircon and ~70% of the monazite that was present at the
solidus has been consumed. Melt ASI varies between 1.16 and 1.11 during the prograde
path. As with the clockwise P–T path, melting above the biotite-stability field (in this case
from 830–860 °C) has the potential to generate some prograde monazite if LREE-rich apatite
is consumed. A total of 18 mol% melt is predicted to be generated during heating, which is
significantly less than the 44 mol% produced along the same P–T path for the metapelite.
Consistent with the metapelite for the same P–T path, isobaric cooling results in melt
crystallization and monazite and zircon growth. The mode of garnet decreases by 7 mol% and
K-feldspar breaks down to sillimanite in the presence of melt at 815 °C. Newly crystallized
monazite and zircon are predicted to have elevated Y and HREE due to the breakdown of
garnet. Zircon growth occurs when the aTiO2 value of the system is ~0.7.
Average mid ocean ridge basalt

Phase relations. The P–T pseudosection for an average MORB composition is shown
in Figure 11. The wet solidus has a negative slope from 4 to 11.5 kbar and a positive slope at
T > 11.5 kbar. The temperature of the wet solidus ranges from 620 to 700 °C over the modelled
P–T range. Garnet is stable from 10–14 kbar with decreasing temperature. Orthopyroxene
enters the assemblage at temperatures of 800 °C to 900 °C with increasing P and is not stable
above 10 kbar at high T. Garnet and orthopyroxene are only stable together at pressures of
9.5–10 kbar and at T > 900 °C. Rutile is stable at T > 790 °C at P > 7 kbar.
Reaction sequence for the clockwise P–T path. The mode-box and titania activity for
the reaction sequence as well as the amount of zircon remaining and melt ASI are shown in
Figure 12. Melting starts at the wet solidus at 630 °C and 9.7 kbar and generates ~ 2 mol%
melt and ~ 2 mol% clinopyroxene at the expense of epidote and biotite. A minor amount
(~2 mol%) of zircon is expected to be consumed in order to saturate the melt in Zr. Epidote
can be an important source of Zr (e.g., Kohn et al. 2015) and the breakdown of Zr-rich
epidote may result in Zr saturation and minor zircon crystallization at this stage. After the
exhaustion of biotite at ~650 °C, melting continues via the breakdown of hornblende, titanite
and quartz to produce an additional 4 mol% melt as well as peritectic clinopyroxene by
800 °C and 11.4 kbar. During this portion of the prograde path, aTiO2 increases from 0.6 at the
wet solidus up to 0.9 at 800 °C. The amount of zircon remaining is ~80 mol% of the amount
present at the solidus. Melt ASI increases slightly from 0.99 to 1.00.
Garnet becomes stable at ~800 °C and melting proceeds through the breakdown of
hornblende and titanite. Garnet growth may include zircon grains that would be isolated from
the reacting volume of the rock and zircon could be preserved to higher T. Rutile becomes
stable at 818 °C and titanite is completely consumed by 821 °C. The peak pressure of 12 kbar
is reached at 850 °C and at this point the system contains ~13 mol% melt, 10 mol% garnet
and 65% of the zircon has been consumed. Although zircon is expected to be consumed
during partial melting, the breakdown of Zr-rich amphibole (e.g., Sláma et al. 2007) may
yield enough Zr to oversaturate the melt and grow new zircon; this zircon may be relatively
depleted in HREE in response to the presence of garnet in the rock.
The isobaric heating segment of the P–T path at peak P produces garnet at the expense
of hornblende and quartz. Quartz is exhausted at 905 °C. This has implications for applying
the Ti-in-zircon thermometer, which uses aSiO2 as a variable. Zircon is completely consumed
by 880 °C and the melt produced at higher T is expected to be undersaturated in Zr. A
consequence of this is that prograde zircon growth at T > 880 °C is unlikely because any
excess Zr due to the breakdown of other minerals (e.g., amphibole) will be incorporated
into the Zr-undersaturated melt. At the metamorphic peak of 950 °C, the system contains
~28 mol% melt, ~20 mol% garnet and ~14% hornblende.
Isothermal decompression from 12 to 10 kbar results in significant garnet consumption
(from 20 to 9 mol%), hornblende growth (from 14 to 24 mol%) and minor melt consumption
(from 28 to 27 mol%). Zirconium liberated from garnet breakdown is expected to be partitioned
between the Zr-undersaturated melt and hornblende. Consequently, zircon crystallization
is unlikely. Orthopyroxene enters the assemblage at 10.1 kbar and grows at the expense
of garnet, which is exhausted by 9.7 kbar. If garnet is completely consumed, any liberated
zircon may become available to the reacting volume and will likely be consumed into the
Zr-undersaturated melt. Rutile is completely consumed by 9.1 kbar and aTiO2 decreases with
further decompression. At the end of the isothermal decompression segment of the P–T path
the amount of orthopyroxene reaches 5 mol%. No new zircon growth is expected during
the isothermal decompression segment because: (1) there is a <1 mol% melt change during
isothermal decompression and the melt is already significantly undersaturated in Zr, (2)
the M value of the melt (cation ratio of [Na + K + 2Ca]/ [Al × Si]) increases from 1.6 to 1.7,
Average MORB + liq

14
hb cpx
mu ep hb
sph q cpx mu ep g hb
13 H2O ttn q
p cpx pl ttn q g hb
ue cpx pl ru q g hb cpx pl ru
p lm
ep px
q
x q
12 cp ttn
ttn
pl hb c
hb
hb cpx
11 pl bi ep
ttn q
ru
x cpx pl
P (kbar)
10 hb
n ru q
hb
cpx pl cpx pl ru q g hb op
ttn q
x pl tt
9
hb cp
hb opx
hb c
pl i x cpx
cpx pl
8
px p
q
ilm ru
lm
op
hb
l bi t
hb cpx
hb
pl bi cpx pl hb
7
tn q
sph q ilm q opx cpx pl

H2O ilm
6
q
ttn
ilm
5
l
di p
hb pl bi
sph q H2O
hb
4
600 650 700 750 800 850 900 950 1000
T (°C)
Figure 11. P–T pseudosection for the suprasolidus MORB. The solidus is shown by the heavy dashed line.
Figure 11
MORB: clockwise P–T

H2O ru prograde retrograde ilm
(a) 1.0
ttn q q
pl pl
0.8 ep
a TiO2
bi
normalized molar
0.6
cpx hb
hb
0.4
g
opx
0.2
liq
(b) 100 1.05
80
% zircon remaining
solidus
ASI of melt
zircon
60
ASI 1.00
40
20
0 0.95
600 650 700 750 800 850 900 12 11 10 9 950 900 850 800
Pressure (kbar)
Temperature (°C) Temperature (°C)
@ 950°C
Figure 12. (a) Mode-box diagram and aTiO2 for the suprasolidus clockwise P–T path for the MORB.
(b) Amount of zircon remaining relative to the amount at the solidus and the ASI value of melt.
Figure 12
which results in an increases the Zr required to saturate the melt (Harrison and Watson 1983;
Boehnke et al. 2013), (3) the mode of hornblende increases from 14 to 23 mol%, likely
accommodating significant amounts of Zr at high temperature (e.g., Kohn et al. 2015).
Cooling and decompression from 950 °C at 8 kbar to 800 °C at 6 kbar results in the
consumption of 17 mol% melt and the complete breakdown of orthopyroxene by ~830 °C.
Zirconium saturation of the melt is reached at ~870 °C and zircon begins to crystallize. At
~850 °C, approximately 20% of the amount of zircon originally present at the solidus has
grown back and is expected to be enriched in HREE because there is no garnet present in the
system. Zircon growth at this stage occurs in a system with an aTiO2 value of 0.9 and in the
absence of quartz, which is important for the application of Ti-in-zircon thermometry.
Summary of reaction sequence modelling
Linking the trace element concentrations of accessory minerals to the key major minerals
(e.g., garnet and plagioclase) requires an understanding of the reaction sequences for a particular
bulk composition. Key minerals like garnet and plagioclase can grow or be consumed multiple
times along a P–T path during heating and cooling or burial and exhumation (Figs 3–10).
For example, garnet is predicted to grow in the subsolidus greywacke composition during
heating and burial and garnet breakdown occurs during cooling and exhumation for the hairpin
P–T path (Fig. 6b). By contrast, garnet growth and consumption occurs multiple times along
the heating and increasing pressure segment of the P–T paths for the subsolidus metapelite
composition (Fig. 4). Therefore, high-Y and high-HREE zones in monazite that can be linked
to garnet breakdown may reflect the burial and/or exhumation portions of a P–T path.
Titania activity also varies along the modelled P–T paths and this needs to be taken into
account when applying Ti-in-zircon or Ti-in-quartz thermometers. In general, new zircon
growth in subsolidus rocks is expected to be limited because major minerals such as hornblende
and garnet can accommodate more Zr with increasing temperature. One exception to this may
be minor zircon growth through the breakdown of Zr-rich epidote. For the suprasolidus P–T
paths for the metapelite and greywacke, new zircon growth is expected during cooling and
melt crystallization; this growth is predicted to occur when bulk rock aTiO2 is less than one for
all of the modelled P–T paths. If the Ti-in-zircon thermometer is applied with an aTiO2 value
of one, then the result will be an underestimation in peak metamorphic temperatures by up to
~40 °C. For the MORB composition, new zircon growth occurs at aTiO2 < 1.0 as cooling and
melt crystallization occurs outside the stability field of rutile (Fig. 12).
In suprasolidus metamorphic rocks, zircon and monazite are expected to be consumed
along the prograde path and new growth is generally predicted to occur along the cooling
path. Zircon and monazite dissolution is non-linear and the rate increases with increasing
temperature. For the clockwise P–T path for the metapelite and greywacke, an increase in
the melt ASI leads to more apatite dissolution. Because the melt is saturated in LREE with
respect to monazite up to 820 °C, apatite breakdown may liberate enough LREE to promote
new monazite growth. However, after the exhaustion of monazite, the melt is expected to
be undersaturated in LREE and prograde monazite growth at the expense of LREE-rich
apatite is unlikely. For the counterclockwise P–T path for the metapelite and greywacke,
melt ASI increases and decreases during the P–T evolution and monazite growth from apatite
breakdown will be more complex to interpret.
The metapelite is expected to lose most of the inherited or subsolidus prograde monazite
and zircon during heating above the solidus (unless these mineral are sequestered away from
the reaction volume) whereas a larger proportion of these minerals remains in the greywacke
composition except for the clockwise P–T path where monazite is completely consumed for
the greywacke. The difference reflects the fertility of the two rocks; pelites generate more melt
than the greywacke and require more Zr and LREE to saturate this melt. Therefore, less fertile
compositions, such as greywackes, are more likely to preserve subsolidus zircon and monazite.
COMPLICATING FACTORS
Changes in effective bulk composition
Phase equilibria modelling requires an assessment of the effective bulk composition of a
system that is used to model the P–T phase relations for a rock (e.g., Stüwe 1997). The effective
bulk composition approximates the composition ‘available’ to the rock in which equilibrated
mineral assemblages and reaction sequence develop. In natural systems, the effective bulk
composition may change along a P–T path (e.g., Guevara and Caddick 2016), which has subsequent
implications for modelling reaction sequences in both subsolidus and suprasolidus metamorphic
rocks. There are two main mechanisms that can change the effective bulk composition of a
metamorphic system: fractionation of elements into growing porphyroblasts and melt loss.
Growing porphyroblasts can fractionate certain elements into their cores that become
unavailable to the reacting system in the remainder of the rock. For example, chemical zoning
in garnet is commonly preserved in metamorphic rocks because cation diffusion in garnet is
relatively slow (e.g., Chakraborty and Ganguly 1992). The preferential partitioning of elements
into garnet cores reduces their effective composition in the reactive volume of the rock (e.g.,
Lanari and Engi 2017, this volume). Some of the consequences for using phase diagrams to
infer metamorphic conditions considering garnet fractionation include reduced stability fields
for garnet (Gaides et al. 2008) and other minerals (Zuluaga et al. 2005; Moynihan and Pattison
2013) as well as changes in mineral compositional isopleths for garnet (Evans 2004; Gaides et
al. 2006) and plagioclase (Moynihan and Pattison 2013). While crystal fractionation needs to be
assessed on a case-by-case basis, it is generally most important to consider for greenschist- and
amphibolite-facies metamorphic assemblages where mineral chemistries are particularity useful
for determining P–T conditions. At higher grades, the use of mineral composition isopleths is
generally less effective due to retrograde exchange reactions (e.g., Spear and Florence 1992;
Kohn and Spear 2000; Pattison et al. 2003). For zircon, the fractionation of Zr into growing
garnet (e.g., Kohn et al. 2015) may reduce the Zr budget available for zircon growth.
In high-grade metamorphic rocks that underwent anatexis the preservation of peritectic
minerals and lack of extensive retrogression supports melt drainage during prograde
metamorphism (Fyfe 1973; Powell 1983; White and Powell 2002; Guernina and Sawyer
2003; Reno et al. 2012). Melt loss produces progressively more refractory bulk compositions,
which results in elevated solidus temperatures in the residual rocks. In migmatites that have
undergone melt loss, suprasolidus zircon and monazite growth is generally expected to occur
during cooling from peak T to the elevated solidus (e.g., Kelsey et al. 2008; Spear and Pyle
2010; Yakymchuk and Brown 2014b). Therefore, rocks that have experienced identical P–T
histories but variable amounts of melt loss and have different solidus temperatures should
record a range of ages (e.g., Korhonen et al. 2013b).
Bulk composition and the suprasolidus behaviour of zircon and monazite
For the modelled suprasolidus reaction sequences, the preservation of subsolidus (e.g.,
prograde or inherited) zircon and monazite is mainly related to the fertility of the rocks. The
metapelite generates more melt along the same P–T path than the greywacke. Consequently, zircon
and monazite are completely consumed for the metapelite composition along both P–T paths (Figs.
8a, b) whereas some zircon and/or monazite can survive in the greywacke. An additional factor is
the bulk rock content of Zr for zircon and LREE and phosphorus for monazite (e.g., Kelsey et al.
2008; Yakymchuk and Brown 2014b). Both of these factors are explored together in Figure 13.
The temperature–composition diagrams in Figure 13 illustrate the variation in melt
mode and the stability of zircon and monazite for compositions ranging linearly from the
metapelite (left side) to that of the greywacke (right side). The diagrams are isobaric and
were calculated at 7 kbar because this pressure intersects the main melt producing reactions
for both bulk compositions. The common reaction to both compositions is the breakdown of
biotite to produce cordierite at ~840–845 °C (Fig. 13a). The amount of melt in the metapelite
(a) T–X pseudosection (b) melt isopleths

950 950 55 50 45 40
65 60 35 30
g pl sill ilm liq
g pl ksp sill ilm liq g pl sill q ilm liq
25
900 900 50
g pl ksp sill q ilm liq
Temperature (C)
Temperature (C)
20
10
40
850 g pl ksp cd sill q ilm liq 850
30 15
g bi pl ksp cd sill q ilm liq
10
800 800 20
10
g bi pl ksp sill q ilm liq
5
750 750 10
g bi pl mu ksp sill q ilm liq
g bi pl sill q ilm liq
700 g bi pl mu sill q ilm liq 700

g bi pl mu sill q ilm liq H2O
subsolidus subsolidus
(c) % dissolution of zircon (Zr = 150 ppm) (d) % dissolution of monazite (LREE = 150 ppm)
950 950
zircon out
100% 90% 80% 70%
60%
50%
40% 900
900
30%
monazite out
20%
Temperature (C)
Temperature (C)
850 850 100% 90%

10% 80% 70%
60% 50%
40%
800 800 30%
20%
10%
750 750
700 700
solidus solidus
(e) Complete dissolution of zircon (f) Complete dissolution of monazite

950 m 950
200 pp m
150 pp 500 ppm
450 ppm
m 400 ppm
100 pp 350 ppm
900 900 300 ppm
m 250 ppm
50 pp
200 ppm
850 850 150 ppm

Temperature (C)
Temperature (C)
100 ppm
800 800 50 ppm
750 750
700 700
solidus solidus
650 650
0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1.0 0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1.0
Average metapelite Average Greywacke Average metapelite Average Greywacke
Figure 13. (a) T–X pseudosection calculated at 7 kbar showing the change in mineral assemblages for
compositions ranging from the metapelite (left side) to the greywacke (right side). (b) T–X diagram with
Figure 13
melt isopleths (mol%). (c) T–X diagram with the calculated amount of zircon dissolution assuming a bulk
Zr concentration of 150 ppm. (d) Calculated amount of monazite dissolution assuming a bulk LREE con-
centration of 150 ppm. (e) T–X diagram with contours for the complete dissolution of zircon for a range of
bulk rock Zr concentrations. (f) T–X diagram with contours for the complete dissolution of monazite for a
range of bulk rock LREE concentrations.
composition is roughly twice the amount in the greywacke composition over the range of
modelled temperatures (Fig. 13b). For a bulk composition of 150 ppm Zr, zircon is completely
consumed in the metapelite composition by 925 °C whereas only 30% of the zircon has been
consumed in the greywacke composition at the same temperature (Fig. 13c). For a bulk
composition of 150 ppm LREE, monazite is completely consumed by 830 °C for the metapelite
and 880 °C for the greywacke (Fig. 13d). The slopes of the dissolution contours are steeper for
zircon than for monazite in Figures 13c and 13d. This indicates that zircon dissolution is more
sensitive to bulk composition (metapelite vs. greywacke) than monazite.
The sensitivity of zircon and monazite dissolution to bulk composition of Zr and

LREE is explored in Figures 13e and 13f. The contours represent the complete dissolution
of zircon and monazite for various bulk rock Zr and LREE contents. For low-Zr pelites
with concentrations of 50 ppm, zircon is completely consumed by 860 °C, which is 65 °C
lower than for a closer-to-average Zr concentration of 150 ppm. Greywackes with low
concentrations of Zr (e.g., 50 ppm) can still preserve zircon up to UHT conditions due to
their lower fertility. Monazite dissolution contours in Figure 13f have shallower slopes than
those for zircon (Fig. 13e), which again suggests that monazite dissolution is less sensitive to
bulk composition. A metapelite with low concentrations of LREE (e.g., 50 ppm) is predicted
to lose monazite by 780 °C and monazite is expected to be completely consumed in the
greywacke composition by 820 °C. Preserving subsolidus monazite to UHT conditions
requires very LREE-rich compositions of > 400 ppm for the metapelite and >180 ppm for
the greywacke. Therefore, a strategy for finding subsolidus prograde zircon and monazite in
migmatites is to choose samples with the highest concentrations of Zr and LREE.
The breakdown of LREE-rich apatite is a potential mechanism to promote prograde
suprasolidus monazite growth (e.g., Johnson et al. 2015). This is most likely to occur when
the anatectic melt is saturated with respect to monazite in LREE. After the exhaustion of
monazite, the melt is expected to be undersaturated and LREE liberated from the breakdown
of apatite is not predicted to generate new prograde monazite growth. In principle, the
higher the bulk rock LREE the longer monazite will persist during a suprasolidus heating
path and the longer the melt will remain saturated with respect to LREE (Fig. 13f). A further
consideration is that bulk compositions with low phosphorus concentrations may lose
apatite during the prograde path and would result in melt that is undersaturated in P. This
may promote further monazite dissolution instead of monazite growth. Therefore, LREE
and P-rich bulk compositions should be targeted for accessory mineral geochronology to
constrain the timing of suprasolidus prograde monazite growth.
Effects of open system behaviour on accessory minerals
Dissolution/re-precipitation of accessory minerals in metamorphic rocks due to the
infiltration of an externally derived fluid has been documented in experiments and studies of
natural samples (e.g., Tomaschek et al. 2003; Crowley et al. 2008; Harlov and Hetherington
2010; Blereau et al. 2016). The careful integration of petrography with the chemistries of these
minerals can be used to provide information on the timing of fluid infiltration (e.g., Williams
et al. 2011) and on fluid chemistry (e.g., Harlov et al. 2011; Taylor et al. 2014; Shazia et
al. 2015). However, the timing of fluid ingress relative to the metamorphic history and the
chemistry of these fluids are highly variable and should to be assessed on a case-by-case basis.
For migmatites that have lost melt, the extraction of melt saturated in Zr or LREE will also
change the effective composition of the residuum (e.g., Rapp et al. 1987), which has subsequent
consequences for zircon and monazite dissolution. For example, consider the muscovite to
K-feldspar melting reaction for the metapelite along the clockwise P–T path at ~730 °C and
7.5 kbar and consider bulk rock values of 150 ppm for LREE and Zr. This reaction involves a
large positive volume change (e.g., Powell et al. 2005), which may be accommodated by melt
extraction from the rock. When muscovite is exhausted the rock contains ~9 mol% melt. The
saturation concentrations of LREE and Zr in the melt at this point are 195 ppm and 137 ppm,
respectively. Assuming 8 mol% melt (approximately equivalent to vol%) is extracted (leaving
1 mol% in the rock along grain boundaries), mass balance can be used to determine the amount
of Zr and LREE left in the system. For this example, the effective concentration of LREE in
the rock decreases to 146 ppm and the effective concentration of Zr increases to 151 ppm.
Although the changes to the bulk composition are minor in this example, monazite is predicted
to be completely consumed at lower T and zircon may persist to higher T than for closed
system (Fig. 13). For rocks with very low concentrations of Zr and LREE, melt extraction can
have a more significant impact on the effective bulk concentrations of these elements and the
stability of zircon and monazite (e.g., Yakymchuk and Brown 2014b).
Inclusion/host relationships
An important consideration for accessory mineral reactivity is their inclusion in the
major rock-forming minerals (Watson et al. 1989; Bea et al. 2006). Watson et al. (1989)
showed that for a Himalayan migmatite sample roughly 78% of the zircon mass is located
along grain boundaries and the remaining 28% is included in major minerals (predominately
biotite and garnet), though how representative this is of typical migmatitic gneiss is unknown.
Consequently, the breakdown of major minerals may liberate accessory minerals into the
reacting volume of the rock that would otherwise be sequestered away.
Inclusions of zircon and monazite that are isolated from the reaction volume may
also reduce the effective bulk rock concentration of Zr and LREE (Yakymchuk and Brown
2014b). For example, consider a bulk rock composition of LREE with half of the monazite
sequestered away from the reaction volume as inclusions. For a metapelite with a bulk rock
LREE composition of 300 ppm and considering that half of this is unavailable, the effective
concentration of LREE is 150 ppm. The complete dissolution of monazite is modelled to occur
at 830 °C in contrast to 870 °C for the scenario where all monazite is available for reaction
(Fig. 13f). Therefore, applying the models in Figure 13 to natural examples requires an estimate
of the amount of zircon or monazite sequestered away from the reaction volume as inclusions as
well as an estimate of the amount of Zr and LREE locked away in the major minerals.
The heterogeneous distribution of melt and minerals in high-grade metamorphic rocks
also has implications for the dissolution and preservation of accessory minerals. Even for
an initially homogenous protolith, in situ melt may be spatially associated with peritectic
minerals in isolated patches; this produces a heterogeneous melt framework (e.g., White et
al. 2004). Zircon and monazite proximal to the zones of incipient melting and in chemical
communication with this melt are more likely to contribute to Zr and LREE saturation of
the melt whereas more distal grains may not. Consequently, detrital, inherited or prograde
(subsolidus or early suprasolidus) zircon and monazite are more likely to be preserved in
domains away from incipient melting whereas post-peak and retrograde zircon and monazite
may be spatially associated with in situ leucosome.
CONCLUDING REMARKS
One important facet of petrochronology is to link the ages of accessory mineral chronometers
to the P–T information obtained from major rock-forming minerals in metamorphic rocks.
The growth and consumption of major minerals is important because these minerals: (1) may
contain the necessary essential structural constituents to promote accessory mineral growth
directly from their breakdown, (2) are repositories of the trace elements used to link accessory
mineral chronometers to P–T conditions (e.g., Sr and Eu related to stability of plagioclase as
well as Y and HREE reflecting the growth/consumption of garnet), (3) are important hosts for
accessory mineral inclusions, and (4) play a role in controlling the component activities (e.g.,
aTiO2) along a P–T evolution. The main controls on accessory mineral behaviour differ between
subsolidus and suprasolidus metamorphic conditions.
For subsolidus metamorphism, zircon is generally unreactive and monazite can grow
during the prograde and retrograde segments. Linking ages from these accessory mineral to a
metamorphic history requires an understanding of the major mineral reaction sequence as well
as the behaviour of accessory minerals like xenotime, apatite and allanite. Major minerals such
as garnet or plagioclase may experience growth and breakdown stages at any point along a P–T
path: linking their behaviour to the trace element chemistries of accessory minerals requires an
assessment of the reaction sequence for a particular rock along a well-constrained P–T path.
For suprasolidus metamorphism, phase equilibria modelling predicts that both zircon and
monazite will be consumed during prograde metamorphism and grow during cooling and melt
crystallization. However, this contrasts with some studies that have convincingly showed evidence
for suprasolidus prograde zircon and monazite growth. For monazite, apatite dissolution may
have contributed to minor prograde monazite growth if the anatectic melt is saturated in LREE.
For zircon, solid-state breakdown of major minerals that contain appreciable quantities of Zr
may facilitate prograde zircon growth. Ostwald ripening may also play a role in the prograde
growth of both zircon and monazite, but this mechanism is still incompletely understood.
ACKNOWLEDGMENTS
We thank Mark Caddick and Dave Waters for thorough and perceptive reviews and
Pierre Lanari for his patient editorial work. Nonetheless, the authors are responsible for any
misinterpretations or omissions that persist. CY was partially funded by a National Sciences
and Engineering Research Council of Canada Discovery Grant.
REFERENCES
Ague JJ (1991) Evidence for major mass transfer and volume strain during regional metamorphism of pelites.
Geology 19:855–858
Ashley KT, Law RD (2015) Modeling prograde TiO2 activity and its significance for Ti-in-quartz thermobarometry
of pelitic metamorphic rocks. Contrib Mineral Petrol 169:1–7
Ayers JC, Miller C, Gorisch B, Milleman J (1999) Textural development of monazite during high-grade
metamorphism: Hydrothermal growth kinetics, with implications for U, Th–Pb geochronology. Am Mineral
84:1766–1780
Baxter EF, Caddick MJ, Dragovic B (2017) Garnet: A rock–forming mineral petrochronometer. Rev Mineral
Geochem 83:469–533
Baxter EF, Scherer EE (2013) Garnet geochronology: timekeeper of tectonometamorphic processes. Elements
9:433–438
Bea F, Montero P (1999) Behavior of accessory phases and redistribution of Zr, REE, Y, Th, and U during
metamorphism and partial melting of metapelites in the lower crust: an example from the Kinzigite Formation
of Ivrea-Verbano, NW Italy. Geochim Cosmochim Acta 63:1133–1153
Bea F, Pereira MD, Stroh A (1994) Mineral/leucosome trace-element partitioning in a peraluminous migmatite (a
laser ablation-ICP-MS study). Chem Geol 117:291–312
Bea F, Montero P, Ortega M (2006) A LA–ICP–MS evaluation of Zr reservoirs in common crustal rocks:
implications for Zr and Hf geochemistry, and zircon-forming processes. Can Mineral 44:693–714
Berman RG (1988) Internally-consistent thermodynamic data for minerals in the system Na2O–K2O–CaO–MgO–
Bingen B, Austrheim H, Whitehouse M (2001) Ilmenite as a source for zirconium during high-grade metamorphism?
Textural evidence from the Caledonides of Western Norway and implications for zircon geochronology. J
Petrol 42:355–375
Blereau E, Clark C, Taylor RJM, Johnson TJ, Fitzsimons I, Santosh M (2016) Constraints on the timing and
conditions of high-grade metamorphism, charnockite formation and fluid–rock interaction in the Trivandrum
Block, southern India. J Metamorph Geol 34:527–549.
Boehnke P, Watson EB, Trail D, Harrison TM, Schmitt AK (2013) Zircon saturation re-revisited. Chem Geol
351:324–334
Brown M (1998) Unpairing metamorphic belts: P–T paths and a tectonic model for the Ryoke Belt, southwest
Japan. J Metamorph Geol 16:3–22
Brown M (2014) The contribution of metamorphic petrology to understanding lithosphere evolution and
geodynamics. Geosci Front 5:553–569
Brown M, Kothonen FJ (2009) Some remarks on melting and extreme metamorphism of crustal rocks. In: Physics
and Chemistry of the Earth’s Interior. Gupta AK, Dasgupta S (eds) Springer, p 67–87
Caddick MJ, Kohn MJ (2013) Garnet: Witness to the evolution of destructive plate boundaries. Elements 9:427–432
Carlson WD (1999) The case against Ostwald ripening of porphyroblasts. Can Mineral 37:403–414
Carlson WD (2000) The case against Ostwald ripening of porphyroblasts: Reply. Can Mineral 38:1029–1031
Catlos EJ, Harrison TM, Kohn MJ, Grove M, Ryerson FJ, Manning CE, Upreti BN (2001) Geochronologic
and thermobarometric constraints on the evolution of the Main Central Thrust, central Nepal Himalaya. J
Geophys Res B: Solid Earth 106:16177–16204
Chakraborty S, Ganguly J (1992) Cation diffusion in aluminosilicate garnets: experimental determination in
spessartine-almandine diffusion couples, evaluation of effective binary diffusion coefficients, and applications.
Contrib Mineral Petrol 111:74–86
Chambers JA, Kohn MJ (2012) Titanium in muscovite, biotite, and hornblende: Modeling, thermometry, and rutile
activities of metapelites and amphibolites. Am Mineral 97:543–555
Chen Y-X, Zheng Y-F, Hu Z (2013) Synexhumation anatexis of ultrahigh-pressure metamorphic rocks: petrological
evidence from granitic gneiss in the Sulu orogen. Lithos 156:69–96
Clark DJ, Hensen BJ, Kinny PD (2000) Geochronological constraints for a two-stage history of the Albany–Fraser
Orogen, Western Australia. Precambrian Res 102:155–183
Clark C, Fitzsimons ICW, Healy D, Harley SL (2011) How does the continental crust get really hot? Elements
7:235–240
Clark C, Kirkland CL, Spaggiari CV, Oorschot C, Wingate MTD, Taylor RJ (2014) Proterozoic granulite
formation driven by mafic magmatism: An example from the Fraser Range Metamorphics, Western Australia.
Precambrian Res 240:1–21
Clarke GL, Guiraud M, Powell R, Burg JP (1987) Metamorphism in the Olary Block, South Australia: compression
with cooling in a Proterozoic fold belt. J Metamorph Geol 5:291–306
Clemens JD, Vielzeuf D (1987) Constraints on melting and magma production in the crust. Earth Planet Sci Lett
86:287–306
Collins WJ, Vernon RH (1991) Orogeny associated with anticlockwise PTt paths: Evidence from low-P, high-T
metamorphic terranes in the Arunta inlier, central Australia. Geology 19:835–838
Connolly JAD, Petrini K (2002) An automated strategy for calculation of phase diagram sections and retrieval of
rock properties as a function of physical conditions. J Metamorph Geol 20:697–708
Copeland RA, Frey FA, Wones DR (1971) Origin of clay minerals in a Mid-Atlantic Ridge sediment. Earth Planet
Sci Lett 10:186–192
Corrie SL, Kohn MJ (2008) Trace-element distributions in silicates during prograde metamorphic reactions:
Implications for monazite formation. J Metamorph Geol 26:451–464
Crowley JL, Brown RL, Gervais F, Gibson HD (2008) Assessing inheritance of zircon and monazite in granitic
rocks from the Monashee Complex, Canadian Cordillera. J Petrol 49:1915–1929
Crowley JL, Waters DJ, Searle MP, Bowring SA (2009) Pleistocene melting and rapid exhumation of the Nanga
Parbat massif, Pakistan: Age and P–T conditions of accessory mineral growth in migmatite and leucogranite.
Earth Planet Sci Lett 288:408–420
Dawson GC, Krapež B, Fletcher IR, McNaughton NJ, Rasmussen B (2003) 1.2 Ga thermal metamorphism in the
Albany–Fraser Orogen of Western Australia: consequence of collision or regional heating by dyke swarms?
J Geol Soc London 160:29–37
de Capitani C, Brown TH (1987) The computation of chemical equilibrium in complex systems containing non-
ideal solutions. Geochim Cosmochim Acta 51:2639–2652
de Capitani C, Petrakakis K (2010) The computation of equilibrium assemblage diagrams with Theriak/Domino
software. Am Mineral 95:1006–1016
Degeling H, Eggins S, Ellis DJ (2001) Zr budgets for metamorphic reactions, and the formation of zircon from
garnet breakdown. Mineral Mag 65:749–758
Diener JFA, Powell R (2010) Influence of ferric iron on the stability of mineral assemblages. J Metamorph Geol
28:599–613
Dragovic B, Samanta LM, Baxter EF, Selverstone J (2012) Using garnet to constrain the duration and rate of water-
releasing metamorphic reactions during subduction: an example from Sifnos, Greece. Chem Geol 314:9–22
Dragovic B, Baxter EF, Caddick MJ (2015) Pulsed dehydration and garnet growth during subduction revealed by
zoned garnet geochronology and thermodynamic modeling, Sifnos, Greece. Earth Planet Sci Lett 413:111–122
Dragovic B, Guevara VE, Caddick MJ, Baxter EF, Kylander-Clark ARC (2016) A pulse of cryptic granulite-
facies metamorphism in the Archean Wyoming Craton revealed by Sm–Nd garnet and U–Pb monazite
geochronology. Precambrian Res 283:24–49
Duc-Tin Q, Keppler H (2015) Monazite and xenotime solubility in granitic melts and the origin of the lanthanide
tetrad effect. Contrib Mineral Petrol 169:1–26
Enami M, Liou JG, Mattinson CG (2004) Epidote minerals in high P/T metamorphic terranes: Subduction zone
and high-to ultrahigh-pressure metamorphism. Rev Mineral Geochem 56:347–398
Engi M (2017) Petrochronology based on REE–minerals: monazite, allanite, xenotime, apatite. Rev Mineral
Geochem 83:365–418
England PC, Thompson AB (1984) Pressure–temperature–time paths of regional metamorphism I. Heat transfer
during the evolution of regions of thickened continental crust. J Petrol 25:894–928
Evans TP (2004) A method for calculating effective bulk composition modification due to crystal fractionation in
garnet-bearing schist: implications for isopleth thermobarometry. J Metamorph Geol 22:547–557
Ewing TA, Rubatto D, Hermann J (2014) Hafnium isotopes and Zr/Hf of rutile and zircon from lower crustal
metapelites (Ivrea–Verbano Zone, Italy): implications for chemical differentiation of the crust. Earth Planet
Sci Lett 389:106–118
Ewing TA, Rubatto D, Beltrando M, Hermann J (2015) Constraints on the thermal evolution of the Adriatic margin
during Jurassic continental break-up: U–Pb dating of rutile from the Ivrea–Verbano Zone, Italy. Contrib
Mineral Petrol 169:1–22
Ferry JM, Watson EB (2007) New thermodynamic models and revised calibrations for the Ti-in-zircon and Zr-in-
rutile thermometers. Contrib Mineral Petrol 154:429–437
Finger F, Krenn E (2007) Three metamorphic monazite generations in a high-pressure rock from the Bohemian
Massif and the potentially important role of apatite in stimulating polyphase monazite growth along a PT
loop. Lithos 95:103–115
Finger F, Krenn E, Schulz B, Harlov D, Schiller D (2016) “Satellite monazites” in polymetamorphic basement
rocks of the Alps: Their origin and petrological significance. Am Mineral 101:1094–1103
Fitzsimons IC, Kinny PD, Wetherley S, Hollingsworth DA (2005) Bulk chemical control on metamorphic monazite
growth in pelitic schists and implications for U–Pb age data. J Metamorph Geol 23:261–277
Florence FP, Spear FS (1993) Influences of reaction history and chemical diffusion on PT calculations for staurolite
schists from the Littleton Formation, northwestern New Hampshire. Am Mineral 78:345–359
Foster G, Kinny P, Vance D, Prince C, Harris N (2000) The significance of monazite U–Th–Pb age data in
metamorphic assemblages; a combined study of monazite and garnet chronometry. Earth Planet Sci Lett
181:327–340
Foster G, Gibson H, Parrish RR, Horstwood MSA, Fraser J, Tindle A (2002) Textural, chemical and isotopic
insights into the nature and behaviour of metamorphic monazite. Chem Geol 191:183–207
Foster G, Parrish RR, Horstwood MS, Chenery S, Pyle J, Gibson HD (2004) The generation of prograde P–T–t
points and paths; a textural, compositional, and chronological study of metamorphic monazite. Earth Planet
Sci Lett 228:125–142
Franz G, Andrehs G, Rhede D (1996) Crystal chemistry of monazite and xenotime from Saxothuringian-
Moldanubian metapelites, NE Bavaria, Germany. Euro J Mineral:1097–1118
Franz G, Morteani G, Rhede D (2015) Xenotime-(Y) formation from zircon dissolution–precipitation and HREE
fractionation: an example from a metamorphosed phosphatic sandstone, Espinhaço fold belt (Brazil). Contrib
Fraser G, Ellis D, Eggins S (1997) Zirconium abundance in granulite-facies minerals, with implications for zircon
geochronology in high-grade rocks. Geology 25:607–610
Frei D, Liebscher A, Franz G, Dulski P (2004) Trace element geochemistry of epidote minerals. Rev Mineral
Geochem 56:553–605
Frost BR, Chamberlain KR, Schumacher JC (2001) Sphene (titanite): phase relations and role as a geochronometer.
Chem Geol 172:131–148
Fyfe WS (1973) The granulite facies, partial melting and the Archaean crust. Philos Trans R Soc London, Ser A
273:457–461
Gaidies F, Abart R, De Capitani C, Schuster R, Connolly JAD, Reusser E (2006) Characterization of
polymetamorphism in the Austroalpine basement east of the Tauern Window using garnet isopleth
thermobarometry. J Metamorph Geol 24:451–475
Gaidies F, De Capitani C, Abart R (2008) THERIA_G: a software program to numerically model prograde garnet
growth. Contrib Mineral Petrol 155:657–671
Gaidies F, Pattison DRM, De Capitani C (2011) Toward a quantitative model of metamorphic nucleation and
Gasser D, Bruand E, Rubatto D, Stüwe K (2012) The behaviour of monazite from greenschist facies phyllites to
anatectic gneisses: an example from the Chugach Metamorphic Complex, southern Alaska. Lithos 134:108–122
Gervasoni F, Klemme S, Rocha-Júnior ERV, Berndt J (2016) Zircon saturation in silicate melts: a new and improved
model for aluminous and alkaline melts. Contrib Mineral Petrol 171:1–12
Ghent ED (1976) Plagioclase–garnet–Al2SiO5–quartz: a potential geobarometer–geothermometer. Am Mineral
6:710–714
Gibson HD, Carr SD, Brown RL, Hamilton MA (2004) Correlations between chemical and age domains in
monazite, and metamorphic reactions involving major pelitic phases: an integration of ID-TIMS and
SHRIMP geochronology with Y–Th–U X-ray mapping. Chem Geol 211:237–260
Grapes RH, Hoskin PWO (2004) Epidote group minerals in low–medium pressure metamorphic terranes. Rev
Green ECR, White RW, Diener JFA, Powell R, Palin RM (2016) Activity–composition relations for the calculation
of partial melting equilibria in metabasic rocks. J Metamorph Geol, doi: 10.1111/jmg.12211
Gromet LP, Silver LT (1983) Rare earth element distributions among minerals in a granodiorite and their
petrogenetic implications. Geochim Cosmochim Acta 47:925–939
Guernina S, Sawyer EW (2003) Large-scale melt-depletion in granulite terranes: An example from the Archean
Ashuanipi Subprovince of Quebec. J Metamorph Geol 21:181–201
Guevara VE, Caddick MJ (2016) Shooting at a moving target: phase equilibria modelling of high-temperature
metamorphism. J Metamorph Geol 34:209–235.
Hacker BR, Andersen TB, Johnston S, Kylander-Clark ARC, Peterman EM, Walsh EO, Young D (2010) High-
temperature deformation during continental-margin subduction & exhumation: The ultrahigh-pressure
Western Gneiss Region of Norway. Tectonophysics 480:149–171
Hacker BR, Kylander-Clark ARC, Holder R, Andersen TB, Peterman EM, Walsh EO, Munnikhuis JK (2015)
Monazite response to ultrahigh-pressure subduction from U–Pb dating by laser ablation split stream. Chem
Geol 409:28–41
Harlov DE, Hetherington CJ (2010) Partial high-grade alteration of monazite using alkali-bearing fluids:
Experiment and nature. Am Mineral 95:1105–1108
Harlov DE, Wirth R, Förster H-J (2005) An experimental study of dissolution–reprecipitation in fluorapatite: fluid
infiltration and the formation of monazite. Contrib Mineral Petrol 150:268–286
Harlov DE, Wirth R, Hetherington CJ (2011) Fluid-mediated partial alteration in monazite: the role of coupled
dissolution–reprecipitation in element redistribution and mass transfer. Contrib Mineral Petrol 162:329–348
Harris NBW, Caddick M, Kosler J, Goswami S, Vance D, Tindle AG (2004) The pressure–temperature–time path
of migmatites from the Sikkim Himalaya. J Metamorph Geol 22:249–264
Harrison TM, Watson EB (1983) Kinetics of zircon dissolution and zirconium diffusion in granitic melts of variable
water content. Contrib Mineral Petrol 84:66–72
Harrison TM, Watson EB (1984) The behavior of apatite during crustal anatexis: equilibrium and kinetic
considerations. Geochim Cosmochim Acta 48:1467–1477
Hay DC, Dempster TJ (2009) Zircon behaviour during low-temperature metamorphism. J Petrol 50: 571–589.
Hayden LA, Watson EB, Wark DA (2008) A thermobarometer for sphene (titanite). Contrib Mineral Petrol
155:529–540
Hensen BJ (1971) Theoretical phase relations involving cordierite and garnet in the system MgO–FeO–Al2O3–
SiO2. Contrib Mineral Petrol 33:191–214
Hermann J, Rubatto D (2003) Relating zircon and monazite domains to garnet growth zones: age and duration of
granulite facies metamorphism in the Val Malenco lower crust. J Metamorph Geol 21:833–852
Hermann J, Rubatto D, Korsakov A, Shatsky VS (2001) Multiple zircon growth during fast exhumation of
diamondiferous, deeply subducted continental crust (Kokchetav Massif, Kazakhstan). Contrib Mineral Petrol
141:66–82
Hiess J, Nutman AP, Bennett VC, Holden P (2008) Ti-in-zircon thermometry applied to contrasting Archean
metamorphic and igneous systems. Chem Geol 247:323–338
Hofmann AE, Baker MB, Eiler JM (2013) An experimental study of Ti and Zr partitioning among zircon, rutile,
and granitic melt. Contrib Mineral Petrol 166:235–253
Holder RM, Hacker BR, Kylander-Clark ARC, Cottle JM (2015) Monazite trace-element and isotopic signatures
of (ultra) high-pressure metamorphism: Examples from the Western Gneiss Region, Norway. Chem Geol
409:99–111
Holland TJB, Powell R (1998) An internally consistent thermodynamic data set for phases of petrological interest.
J Metamorph Geol 16:309–343
Holland TJB, Powell R (2011) An improved and extended internally consistent thermodynamic dataset for phases
of petrological interest, involving a new equation of state for solids. J Metamorph Geol 29:333–383
Hoskin PWO, Schaltegger U (2003) The composition of zircon and igneous and metamorphic petrogenesis. Rev
Huang WL, Wyllie PJ (1973) Melting relations of muscovite-granite to 35 kbar as a model for fusion of
metamorphosed subducted oceanic sediments. Contrib Mineral Petrol 42:1–14
Jamieson RA, Beaumont C, Medvedev S, Nguyen MH (2004) Crustal channel flows: 2. Numerical models with
implications for metamorphism in the Himalayan–Tibetan orogen. J Geophys Res B: Solid Earth 109:2156–2202
Janots E, Engi M, Berger A, Allaz J, Schwarz JO, Spandler C (2008) Prograde metamorphic sequence of REE
minerals in pelitic rocks of the Central Alps: implications for allanite–monazite–xenotime phase relations
from 250 to 610 C. J Metamorph Geol 26:509–526
Janots E, Engi M, Rubatto D, Berger A, Gregory C, Rahn M (2009) Metamorphic rates in collisional orogeny from
in situ allanite and monazite dating. Geology 37:11–14
Johnson TE, White RW, Powell R (2008) Partial melting of metagreywacke: a calculated mineral equilibria study.
Johnson TE, Clark C, Taylor RJM, Santosh M, Collins AS (2015) Prograde and retrograde growth of monazite in
migmatites: An example from the Nagercoil Block, southern India. Geosci Front 6:373–387
Kapp P, Manning CE, Tropper P (2009) Phase-equilibrium constraints on titanite and rutile activities in mafic
epidote amphibolites and geobarometry using titanite-rutile equilibria. J Metamorph Geol 27:509–521
Kawakami T, Yamaguchi I, Miyake A, Shibata T, Maki K, Yokoyama TD, Hirata T (2013) Behavior of zircon in
the upper-amphibolite to granulite facies schist/migmatite transition, Ryoke metamorphic belt, SW Japan:
constraints from the melt inclusions in zircon. Contrib Mineral Petrol 165:575–591
Kelsey DE, Powell R (2011) Progress in linking accessory mineral growth and breakdown to major mineral
evolution in metamorphic rocks: a thermodynamic approach in the Na2O–CaO–K2O–FeO–MgO–Al2O3–
SiO2–H2O–TiO2–ZrO2 system. J Metamorph Geol 29:151–166
Kelsey DE, Hand M (2015) On ultrahigh temperature crustal metamorphism: phase equilibria, trace element
thermometry, bulk composition, heat sources, timescales and tectonic settings. Geosci Front 6:311–356
Kelsey DE, Clark C, Hand M (2008) Thermobarometric modelling of zircon and monazite growth in melt-bearing
systems: examples using model metapelitic and metapsammitic granulites. J Metamorph Geol 26:199–212
Kingsbury JA, Miller CF, Wooden JL, Harrison TM (1993) Monazite paragenesis and U–Pb systematics in rocks of
the eastern Mojave Desert, California, USA: implications for thermochronometry. Chem Geol 110:147–167
Kirkland CL, Spaggiari CV, Johnson TE, Smithies RH, Danišík M, Evans N, Wingate MTD, Clark C, Spencer
C, Mikucki E, McDonald BJ (2016) Grain size matters: Implications for element and isotopic mobility in
titanite. Precambrian Res 278:283–302
Kohn MJ (2008) PTt data from central Nepal support critical taper and repudiate large-scale channel flow of the
Greater Himalayan Sequence. Geol Soc Am Bull 120:259–273
Kohn MJ, Malloy MA (2004) Formation of monazite via prograde metamorphic reactions among common
silicates: implications for age determinations. Geochim Cosmochim Acta 68:101–113
Kohn MJ, Spear F (2000) Retrograde net transfer reaction insurance for pressure–temperature estimates. Geology
28:1127–1130
Kohn MJ, Corrie SL, Markley C (2015) The fall and rise of metamorphic zircon. Am Mineral 100:897–908
Korhonen FJ, Brown M, Grove M, Siddoway CS, Baxter EF, Inglis JD (2012) Separating metamorphic events in
the Fosdick migmatite–granite complex, West Antarctica. J Metamorph Geol 30:165–192
Korhonen F,J Brown M, Clark C, Bhattacharya S (2013a) Osumilite–melt interactions in ultrahigh temperature
granulites: phase equilibria modelling and implications for the P–T–t evolution of the Eastern Ghats Province,
India. J Metamorph Geol 31:881–907
Korhonen FJ, Clark C, Brown M, Bhattacharya S, Taylor R (2013b) How long-lived is ultrahigh temperature
(UHT) metamorphism? Constraints from zircon and monazite geochronology in the Eastern Ghats orogenic
belt, India. Precambrian Res 234:322–350
Korhonen FJ, Clark C, Brown M, Taylor RJM (2014) Taking the temperature of Earth’s hottest crust. Earth Planet
Sci Lett 408:341–354
Kylander-Clark ARC, Hacker BR, Cottle JM (2013) Laser-ablation split-stream ICP petrochronology. Chem Geol
345:99–112
Geochem 83:55–102
Mahar EM, Baker JM, Powell R, Holland TJB, Howell N (1997) The effect of Mn on mineral stability in
metapelites. J Metamorph Geol 15:223–238
Mayne MJ, Moyen JF, Stevens G, Kaisl Aniemi L (2016) Rcrust: a tool for calculating path-dependent open system
processes and application to melt loss. J Metamorph Geol 34: 663–682.
McClelland WC, Lapen TJ (2013) Linking time to the pressure–temperature path for ultrahigh-pressure rocks.
Elements 9:273–279
Mezger K, Hanson GN, Bohlen SR (1989) High-precision U–Pb ages of metamorphic rutile: application to the
cooling history of high-grade terranes. Earth Planet Sci Lett 96:106–118
Miyazaki K (1991) Ostwald ripening of garnet in high P/T metamorphic rocks. Contrib Mineral Petrol 108:118–128
Miyazaki K (1996) A numerical simulation of textural evolution due to Ostwald ripening in metamorphic rocks: A
case for small amount of volume of dispersed crystals. Geochim Cosmochim Acta 60:277–290
Molina JF, Moreno JA, Castro A, Rodríguez C, Fershtater GB (2015) Calcic amphibole thermobarometry in
metamorphic and igneous rocks: New calibrations based on plagioclase/amphibole Al–Si partitioning and
amphibole/liquid Mg partitioning. Lithos 232:286–305
Montel, J-M Kornprobst, J Vielzeuf, D (2000) Preservation of old U–Th–Pb ages in shielded monazite: example
from the Beni Bousera Hercynian kinzigites (Morocco). J Metamorph Geol 18:335–342
Montel J-M (1986) Experimental determination of the solubility of Ce-monazite in SiO2–Al2O3–K2O–Na2O melts
at 800 C, 2 kbar, under H2O-saturated conditions. Geology 14:659–662
Morrissey LJ, Hand M, Raimondo T, Kelsey DE (2014) Long-lived high-T, low-P granulite facies metamorphism
in the Arunta Region, central Australia. J Metamorph Geol 32:25–47
Moynihan DP, Pattison DRM (2013) An automated method for the calculation of P–T paths from garnet zoning,
with application to metapelitic schist from the Kootenay Arc, British Columbia, Canada. J Metamorph Geol
31:525–548
Mulcahy SR, Vervoort JD, Renne PR (2014) Dating subduction-zone metamorphism with combined garnet and
lawsonite Lu–Hf geochronology. J Metamorph Geol 32:515–533
Nemchin AA, Bodorkos S (2000) Zr and LREE concentrations in anatectic melt as a function of crystal size
distributions of zircon and monazite in the source region. Geol Soc Am, Abstracts and Programs: 52286
Nemchin AA, Giannini LM, Bodorkos S, Oliver NHS (2001) Ostwald ripening as a possible mechanism for zircon
overgrowth formation during anatexis: theoretical constraints, a numerical model, and its application to pelitic
migmatites of the Tickalara Metamorphics, northwestern Australia. Geochim Cosmochim Acta 65:2771–2788
O’Brien PJ, Rötzler J (2003) High-pressure granulites: formation, recovery of peak conditions and implications for
tectonics. J Metamorph Geol 21:3–20
Parrish RR (1990) U–Pb dating of monazite and its application to geological problems. Can J Earth Sci 27:1431–1450
Pattison DRM, Tinkham DK (2009) Interplay between equilibrium and kinetics in prograde metamorphism of
pelites: an example from the Nelson aureole, British Columbia. J Metamorph Geol 27:249–279
Pattison DRM, Chacko T, Farquhar J, McFarlane CRM (2003) Temperatures of granulite-facies metamorphism:
constraints from experimental phase equilibria and thermobarometry corrected for retrograde exchange. J
Petrol 44:867–900
Pattison DRM, DeBuhr CL (2015) Petrology of metapelites in the Bugaboo aureole, British Columbia, Canada. J
Metamorph Geol 33:437–462
Pichavant M, Montel J-M, Richard LR (1992) Apatite solubility in peraluminous liquids: Experimental data and an
extension of the Harrison-Watson model. Geochim Cosmochim Acta 56:3855–3861
Powell R (1983) Processes in granulite-facies metamorphism. In: Proceedings of the Geochemical Group of the
Mineraological Society: Migmatites, melting and metamorphism. Atherton, MP, Gribble, CD (eds). Shiva,
Nantwich, p 127–139
Powell R, Holland TJB (1988) An internally consistent dataset with uncertainties and correlations: 3. Applications
to geobarometry, worked examples and a computer program. J Metamorph Geol 6:173–204
Powell R, Holland TJB (2008) On thermobarometry. J Metamorph Geol 26:155–179
Powell R, Holland T, Worley B (1998) Calculating phase diagrams involving solid solutions via non-linear
equations, with examples using THERMOCALC. J Metamorph Geol 16:577–588
Powell R, Guiraud M, White RW (2005) Truth and beauty in metamorphic phase-equilibria: conjugate variables
and phase diagrams. Can Mineral 43:21–33
Pyle JM, Spear FS (1999) Yttrium zoning in garnet: coupling of major and accessory phases during metamorphic
reactions. Geol Mat Res 1:1–49
Pyle JM, Spear FS (2000) An empirical garnet (YAG)–xenotime thermometer. Contrib Mineral Petrol 138:51–58
Pyle JM, Spear FS, Rudnick RL, McDonough WF (2001) Monazite–xenotime–garnet equilibrium in metapelites
and a new monazite–garnet thermometer. J Petrol 42:2083–2107
Rapp RP, Watson EB (1986) Monazite solubility and dissolution kinetics: implications for the thorium and light
rare earth chemistry of felsic magmas. Contrib Mineral Petrol 94:304–316
Rapp RP, Ryerson FJ, Miller CF (1987) Experimental evidence bearing on the stability of monazite during crustal
anaatexis. Geophys Res Lett 14:307–310
Rasmussen B (2005) Radiometric dating of sedimentary rocks: the application of diagenetic xenotime
geochronology. Earth Sci Rev 68:197–243
Rasmussen B, Fletcher IR, Muhling JR (2011) Response of xenotime to prograde metamorphism. Contrib Mineral
Petrol 162:1259–1277
Regis D, Warren CJ, Mottram CM, Roberts NMW (2016) Using monazite and zircon petrochronology to constrain
the P–T–t evolution of the middle crust in the Bhutan Himalaya. J Metamorph Geol
Reno BL, Piccoli PM, Brown M, Trouw RAJ (2012) In situ monazite (U–Th)–Pb ages from the Southern Brasília Belt,
Brazil: constraints on the high-temperature retrograde evolution of HP granulites. J Metamorph Geol 30:81–112
Roberts MP, Finger F (1997) Do U–Pb zircon ages from granulites reflect peak metamorphic conditions? Geology
25:319–322
Rocha BC, Moraes R, Möller A, Cioffi CR, Jercinovic MJ (2016) Timing of anatexis and melt crystallization in the
Socorro–Guaxupé Nappe, SE Brazil: Insights from trace element composition of zircon, monazite and garnet
coupled to U–Pb geochronology. Lithos
Rubatto D (2002) Zircon trace element geochemistry: partitioning with garnet and the link between U–Pb ages and
metamorphism. Chem Geol 184:123–138
Rubatto D, Hermann J (2003) Zircon formation during fluid circulation in eclogites (Monviso, Western Alps):
implications for Zr and Hf budget in subduction zones. Geochim Cosmochim Acta 67:2173–2187
Rubatto D, Hermann J (2007) Experimental zircon/melt and zircon/garnet trace element partitioning and
implications for the geochronology of crustal rocks. Chem Geol 241:38–61
Rubatto D, Williams IS, Buick IS (2001) Zircon and monazite response to prograde metamorphism in the Reynolds
Range, central Australia. Contrib Mineral Petrol 140:458–468
Rubatto D, Chakraborty S, Dasgupta S (2013) Timescales of crustal melting in the higher Himalayan crystallines
(Sikkim, Eastern Himalaya) inferred from trace element-constrained monazite and zircon chronology.
Schaltegger U (2007) Hydrothermal zircon. Elements 3:51–79

Schaltegger U, Fanning CM, Günther D, Maurin JC, Schulmann K, Gebauer D (1999) Growth, annealing and
recrystallization of zircon and preservation of monazite in high-grade metamorphism: conventional and in-
situ U–Pb isotope, cathodoluminescence and microchemical evidence. Contrib Mineral Petrol 134:186–201
Sheppard S, Rasmussen B, Muhling JR, Farrell TR, Fletcher IR (2007) Grenvillian-aged orogenesis in the
Palaeoproterozoic Gascoyne Complex, Western Australia: 1030–950 Ma reworking of the Proterozoic
Capricorn Orogen. J Metamorph Geol 25:477–494
Shaw DM (1956) Geochemistry of pelitic rocks. Part III: Major elements and general geochemistry. Geol Soc Am
Bull 67:919–934
Shazia JR, Harlov DE, Suzuki K, Kim SW, Girish-Kumar M, Hayasaka Y, Ishwar-Kumar C, Windley BF, Sajeev
K (2015) Linking monazite geochronology with fluid infiltration and metamorphic histories: Nature and
experiment. Lithos 236:1–15
Skora S, Blundy J (2012) Monazite solubility in hydrous silicic melts at high pressure conditions relevant to
subduction zone metamorphism. Earth Planet Sci Lett 321:104–114
Sláma J, Košler J, Pedersen RB (2007) Behaviour of zircon in high-grade metamorphic rocks: evidence from Hf
isotopes, trace elements and textural studies. Contrib Mineral Petrol 154:335–356
Smith HA, Barreiro B (1990) Monazite U–Pb dating of staurolite grade metamorphism in pelitic schists. Contrib
Smye AJ, Stockli DF (2014) Rutile U–Pb age depth profiling: A continuous record of lithospheric thermal
evolution. Earth Planet Sci Lett 408:171–182
Spear FS (1995) Metamorphic phase equilibria and pressure–temperature–time paths. Mineral Soc Am. Washington
Spear FS (2010) Monazite–allanite phase relations in metapelites. Chem Geol 279:55–62
Spear FS, Florence FP (1992) Thermobarometry in granulites: pitfalls and new approaches. Precambrian Res
55:209–241
Spear FS, Pyle JM (2002) Apatite, monazite, and xenotime in metamorphic rocks. Rev Mineral Geochem 48:293–335
Spear FS, Pyle JM (2010) Theoretical modeling of monazite growth in a low-Ca metapelite. Chem Geol 273:111–119
Spear FS, Pattison DRM, Cheney JT (2016) The metamorphosis of metamorphic petrology. Geol Soc Am Sp Pap
523:SPE523-502
Spencer KJ, Hacker BR, Kylander-Clark ARC, Andersen TB, Cottle JM, Stearns MA, Poletti JE, Seward GGE
(2013) Campaign-style titanite U–Pb dating by laser-ablation ICP: implications for crustal flow, phase
transformations and titanite closure. Chem Geol 341:84–101
Stearns MA, Hacker BR, Ratschbacher L, Rutte D, Kylander-Clark ARC (2015) Titanite petrochronology of the
Pamir gneiss domes: Implications for middle to deep crust exhumation and titanite closure to Pb and Zr
diffusion. Tectonics 34:784–802
Stepanov AS, Hermann J, Rubatto D, Rapp RP (2012) Experimental study of monazite/melt partitioning with
implications for the REE, Th and U geochemistry of crustal rocks. Chem Geol 300:200–220
Štípská P, Hacker BR, Racek M, Holder R, Kylander-Clark ARC, Schulmann K, Hasalová P (2015) Monazite
Dating of Prograde and Retrograde P–T–d paths in the Barrovian terrane of the Thaya window, Bohemian
Massif. J Petrol 56:1007–1035
Stüwe K (1997) Effective bulk composition changes due to cooling: a model predicting complexities in retrograde
reaction textures. Contrib Mineral Petrol 129:43–52
Stüwe K, Powell R (1995) PT paths from modal proportions: application to the Koralm Complex, Eastern Alps.
Sun S-S, McDonough W (1989) Chemical and isotopic systematics of oceanic basalts: implications for mantle
composition and processes. Geol Soc London, Sp Pub 42:313–345
Suzuki K, Adachi M (1994) Middle Precambrian detrital monazite and zircon from the Hida gneiss on Oki-Dogo
Island, Japan: their origin and implications for the correlation of basement gneiss of Southwest Japan and
Korea. Tectonophysics 235:277–292
Symmes GH, Ferry JM (1992) The effect of whole-rock MnO content on the stability of garnet in pelitic schists
during metamorphism. J Metamorph Geol 10:221–237
Taylor RJM, Clark C, Fitzsimons ICW, Santosh M, Hand M, Evans N, McDonald B (2014) Post-peak, fluid-
mediated modification of granulite facies zircon and monazite in the Trivandrum Block, southern India.
Taylor RJM, Harley SL, Hinton RW, Elphick S, Clark C, Kelly NM (2015) Experimental determination of REE
partition coefficients between zircon, garnet and melt: a key to understanding high-T crustal processes. J
Taylor-Jones K, Powell R (2015) Interpreting zirconium-in-rutile thermometric results. J Metamorph Geol 33:115–122
Thompson AB (1982) Dehydration melting of pelitic rocks and the generation of H2O-undersaturated granitic
liquids. Am J Sci 282:1567–1595
Thompson AB (1996) Fertility of crustal rocks during anatexis. Geol Soc Am Spec Pap 315:1–10
Thompson AB, England PC (1984) Pressure—temperature—time paths of regional metamorphism II. Their
inference and interpretation using mineral assemblages in metamorphic rocks. J Petrol 25:929–955
Tikare V, Cawley JD (1998) Application of the Potts model to simulation of Ostwald ripening. J Am Ceram Soc
81:485–491
Tomaschek F, Kennedy AK, Villa IM, Lagos M, Ballhaus C (2003) Zircons from Syros, Cyclades, Greece—
recrystallization and mobilization of zircon during high-pressure metamorphism. J Petrol 44:1977–2002
Tomkins HS, Powell R, Ellis DJ (2007) The pressure dependence of the zirconium-in-rutile thermometer. J
Vance D, Mahar E (1998) Pressure–temperature paths from PT pseudosections and zoned garnets: potential,
limitations and examples from the Zanskar Himalaya, NW India. Contrib Mineral Petrol 132:225–245
Vance D, O’Nions RK (1990) Isotopic chronometry of zoned garnets: growth kinetics and metamorphic histories.
Vavra G, Schmid R, Gebauer D (1999) Internal morphology, habit and U–Th–Pb microanalysis of amphibolite-to-
granulite facies zircons: geochronology of the Ivrea Zone (Southern Alps). Contrib Mineral Petrol 134:380–404
Vernon RH (1996) Problems with inferring P–T–t paths in low-P granulite facies rocks. J Metamorph Geol 14:143–153
Vielzeuf D, Schmidt MW (2001) Melting relations in hydrous systems revisited: application to metapelites,
metagreywackes and metabasalts. Contrib Mineral Petrol 141:251–267
Walsh AK, Kelsey DE, Kirkland CL, Hand M, Smithies RH, Clark C, Howard HM (2015) P–T–t evolution of
a large, long-lived, ultrahigh-temperature Grenvillian belt in central Australia. Gondwana Res 28:531–564
Wark DA, Watson EB (2006) TitaniQ: a titanium-in-quartz geothermometer. Contrib Mineral Petrol 152:743–754
Warren CJ, Grujic D, Cottle JM, Rogers NW (2012) Constraining cooling histories: rutile and titanite chronology
and diffusion modelling in NW Bhutan. J Metamorph Geol 30:113–130
Watson EB (1996) Dissolution, growth and survival of zircons during crustal fusion: kinetic principles, geological
models and implications for isotopic inheritance. Geol Soc Am Sp Pap 315:43–56
Watson EB, Harrison TM (1983) Zircon saturation revisited: temperature and composition effects in a variety of
crustal magma types. Earth Planet Sci Lett 64:295–304
Watson EB, Vicenzi EP, Rapp RP (1989) Inclusion/host relations involving accessory minerals in high-grade
metamorphic and anatectic rocks. Contrib Mineral Petrol 101:220–231
Watson EB, Wark DA, Thomas JB (2006) Crystallization thermometers for zircon and rutile. Contrib Mineral
Petrol 151:413–433
Watt GR, Harley SL (1993) Accessory phase controls on the geochemistry of crustal melts and restites produced
during water-undersaturated partial melting. Contrib Mineral Petrol 114:550–566
White RW, Powell R (2002) Melt loss and the preservation of granulite facies mineral assemblages. J Metamorph
Geol 20:621–632
White RW, Powell R, Halpin JA (2004) Spatially-focussed melt formation in aluminous metapelites from Broken
Hill, Australia. J Metamorph Geol 22:825–845
White RW, Pomroy NE, Powell R (2005) An in situ metatexite–diatexite transition in upper amphibolite facies
rocks from Broken Hill, Australia. J Metamorph Geol 23:579–602
White RW, Powell R, Baldwin JA (2008) Calculated phase equilibria involving chemical potentials to investigate
the textural evolution of metamorphic rocks. J Metamorph Geol 26:181–198
White RW, Stevens G, Johnson TE (2011) Is the crucible reproducible? Reconciling melting experiments with
thermodynamic calculations. Elements 7:241–246
White RW, Powell R, Johnson TE (2014a) The effect of Mn on mineral stability in metapelites revisited: New a–x
relations for manganese-bearing minerals. J Metamorph Geol 32:809–828
White RW, Powell R, Holland TJB, Johnson TE, Green ECR (2014b) New mineral activity–composition relations
for thermodynamic calculations in metapelitic systems. J Metamorph Geol 32:261–286
White WM, Klein EM (2014) 4.13—Composition of the Oceanic Crust. In: Treatise on Geochemsitry (Second
Edition). Holland, HD, Turekian, K (ed). Elsevier, Oxford, p 457–496
Williams ML, Jercinovic MJ, Harlov DE, Budzyń B, Hetherington CJ (2011) Resetting monazite ages during fluid-
related alteration. Chem Geol 283:218–225
Wing BA, Ferry JM, Harrison TM (2003) Prograde destruction and formation of monazite and allanite during
contact and regional metamorphism of pelites: petrology and geochronology. Contrib Mineral Petrol
145:228–250
Wolf MB, London D (1994) Apatite dissolution into peraluminous haplogranitic melts: an experimental study of
solubilities and mechanisms. Geochim Cosmochim Acta 58:4127–4145
Wolf MB, London D (1995) Incongruent dissolution of REE-and Sr-rich apatite in peraluminous granitic liquids:
Differential apatite, monazite, and xenotime solubilities during anatexis. Am Mineral 80:765–775
Wu CM (2015) Revised empirical garnet–biotite–muscovite–plagioclase geobarometer in metapelites. J
Xu H, Ye K, Song Y, Chen Y, Zhang J, Liu Q, Guo S (2013) Prograde metamorphism, decompressional partial
melting and subsequent melt fractional crystallization in the Weihai migmatitic gneisses, Sulu UHP terrane,
eastern China. Chem Geol 341:16–37
Yakymchuk C, Brown M (2014a) Consequences of open-system melting in tectonics. J Geol Soc London 171:21–40
Yakymchuk C, Brown M (2014b) Behaviour of zircon and monazite during crustal melting. J Geol Soc London
171:465–479
Yakymchuk C, Brown M, Clark C, Korhonen FJ, Piccoli PM, Siddoway CS, Taylor RJM, Vervoort JD (2015)
Decoding polyphase migmatites using geochronology and phase equilibria modelling. J Metamorph Geol
33:203–230
Yardley BWD, Valley JW (1997) The petrologic case for a dry lower crust. J Geophys Res B: Solid Earth
102:12173–12185
Zack T, Moraes R, Kronz A (2004) Temperature dependence of Zr in rutile: empirical calibration of a rutile
thermometer. Contrib Mineral Petrol 148:471–488
Zack T, Stockli DF, Luvizotto GL, Barth MG, Belousova E, Wolfe MR, Hinton RW (2011) In situ U–Pb rutile dating
by LA-ICP-MS: 208Pb correction and prospects for geological applications. Contrib Mineral Petrol 162:515–530
Zuluaga CA, Stowell HH, Tinkham DK (2005) The effect of zoned garnet on metapelite pseudosection topology
and calculated metamorphic PT paths. Am Mineral 90:1619–1628
Vol. 83 pp. 55–102, 2017 3
Local Bulk Composition Effects

on Metamorphic Mineral Assemblages
Pierre Lanari and Martin Engi
University of Bern
Baltzerstrasse 3
CH-3012 Bern
Switzerland
pierre.lanari@geo.unibe.ch
martin.engi@geo.unibe.ch

Plate tectonic forcing leads to changes in the physical conditions that affect the lithosphere.
In response to such changes, notably the local temperature (T) and pressure (P), rocks evolve
dynamically. Processes mostly involve mineral transformations, i.e., solid-state reactions, but
(hydrous) fluids are often involved, and partial melting may occur in the Earth’s middle and
lower crust. While these chemical reactions reflect the tendency of natural systems to reduce
their Gibbs free energy, metamorphic rocks commonly preserve textural and mineralogical
relics, such as compositionally zoned minerals. Where relics are present, thermodynamic
equilibrium clearly was not attained during the evolution of the rock.
Petrochronology seeks to establish a temporal framework of petrologic evolution, and for
this purpose it is essential to determine the P–T conditions prevailing at several stages. When
analyzing a rock sample it is thus critical:
(a) to recognize whether several stages of its evolution can be discerned,
(b) to document the minerals that formed or were coexisting at each stage, and
(c) to estimate at what physical conditions this happened.
If (and only if) a chronometer then can be associated to one of these stages—or better yet several
chronometers to different stages—then the power of petrochronology becomes realizable.
This chapter is concerned with a basic dilemma that results directly from steps (b) and (c)
above: P–T conditions are determined on the basis of mineral barometers and thermometers,
which mostly rest on the assumption of chemical (or isotopic) equilibrium, yet the presence of
relics is proof that thermodynamic equilibrium was not attained. One way out of the dilemma is to
analyze reaction mechanisms and formulate a model based on non-equilibrium thermodynamics
and kinetics (Lasaga 1998). While this can be fruitful for understanding fundamental aspects of
metamorphic petrogenesis, there are more direct ways to address the limited scope needed for
petrochronology. The alternative pursued here seeks to define the scale(s) of equilibration, i.e., to
determine spatial limits within which chemical equilibration can reasonably be assumed.
The past fifty years have seen rapid developments in forward thermodynamic models
aimed at retrieving the conditions of equilibration from local assemblages. These models are
rooted in chemical equilibrium theory—the concept that rocks successively re-equilibrate along

1943-2666/17/0083-0003$05.00 (online)
56 Lanari & Engi
at least part of their evolution or remain at least close to chemical equilibrium. Since the advent
of electron probe micro-analysis (EPMA), mineral chemical data have been combined with
textural observations in thin section, and in many rocks different local mineral assemblages
have been documented. The question is what caused these differences: Do they represent
temporally distinct stages of a single chemical system or coeval domains of locally different
chemical composition? This question is known as the N-dimensional tie-line problem
(Greenwood 1967): Can two mineral assemblages be related to one another by a balanced
chemical reaction or must the observed differences in their mineralogy be attributed to differing
bulk compositions? A rigorous answer is possible,1 provided the frozen-in assemblages
reflect chemical equilibrium at the time of formation—a hypothesis to be tested. In any case,
equilibrium assemblages reflect the chemical composition of rocks, and for forward chemical
modeling of assemblages that composition must be known. Textural evidence in polymineralic
rocks typically suggests some chemical heterogeneity, as the modal distributions of the
minerals (i.e., their volumetric abundance) commonly varies at centimeter- to millimeter-scale,
or less. Chemical heterogeneity can be inherited, e.g., from interlayered strata, or it may form
during metamorphism, by metamorphic differentiation (Orville 1969). Either way, the effects
of chemical heterogeneity should be considered in forward modeling, especially in clearly
domainal rocks. As petrochronological studies focus on dating minerals in their local context,
petrogenetic conditions ought to be reliably determined by modeling at similarly small scale.
This papers examines some of the major causes for and consequences of chemical
heterogeneity in rocks, which can affect the scale of equilibration and thus the modeled mineral
assemblages. The central question addressed is: how can a realistic bulk composition be found,
such that model results can be compared in detail to a documented sample? A completely
systematic treatment is beyond the scope of this paper, but as some fundamental aspects are
involved in applying equilibrium thermodynamics to metamorphic rocks, the necessary basics
are presented in the first section. Some of the case studies discussed in the second half of this
paper also invoke kinetic aspects. These are cited where necessary but are not reviewed here;
several resources treat the kinetics of metamorphic processes (see for example the book of
Lasaga 1998). In the following sections we analyze specific cases such as the porphyroblastic
growth of garnet that fractionates the reactive bulk composition and the link with mineral
inclusion chronology. We also develop in a more general way the formation and evolution of
textural domains in rocks, including the chemical subsystems and potential gradients. In the last
section, we review analytical methods to estimate local bulk compositions using standardized
X-ray maps (and the program XMapTools, Lanari et al. 2014b). We then show some applications
to petrochronological analysis, and we outline perspectives for future research.
THEORETICAL BASIS AND LIMITS

OF FORWARD THERMODYNAMIC ANALYSIS
Based the pioneering work of Bowen (1913) and Goldschmidt (1911), equilibrium
thermodynamics has proven a powerful conceptual framework to develop forward chemical
modeling. Recent progress in the accuracy and efficiency of such techniques has had a major
impact on the evolution of metamorphic petrology (Spear et al. 2016). It now is straightforward
to model at least some of the main transformational stages experienced by crystalline rocks
(Yakymchuk et al. 2017, this volume and references therein). Many of the previously daunting
technical obstacles, such as the need for sophisticated solution models in complex systems
involving solid and fluid phases, now appear much less evident, as the available software is
making use of thermodynamic databases. However, despite periodic updates of such databases,
1
Greenwood showed that linear algebraic examination is sufficient to determine whether two samples (or
assemblages) overlap in composition space. If they do, yet they show different assemblages, this necessarily
reflects a difference in physical conditions (e.g., of pressure or temperature).
Local Bulk Composition Effects on Metamorphic Mineral Assemblages 57
models for some mineral systems remain imperfectly calibrated. For many bulk compositions
predicted phase relations are reliably known, but for others the models may be partly flawed, and
this is easily overlooked. Furthermore, predictions based on classical thermodynamics are limited
to systems in equilibrium, and the application to metamorphic rocks is not self-evident where they
contain non-equilibrium phenomena. In particular, the definition of an appropriate equilibration
volume suitable for modeling can be a challenging task, as natural rocks typically record several
stages of the metamorphic history through compositional zoning, mineral relics and textural
relationships. Nevertheless, our current understanding of the metamorphic processes in the Earth’s
crust has much improved due to the comparison of model predictions with the observed mineral
assemblages and compositions. Such predictions rely on the application of forward chemical
modeling based on the bulk rock composition of natural samples. In a recent review, Powell
and Holland (2010) concluded that “the success of using forward equilibrium model to study
metamorphic rocks provides, at least in part, an a posteriori justification of the assumption” [of
equilibrium]. On the other hand, some authors have warned the community about the limits of
equilibrium models and have demonstrated, for instance, that kinetic impediments to reaction may
prevent metamorphic rocks from attaining rock-wide chemical equilibrium along their prograde
crystallization paths (Carlson et al. 2015 and references therein). It is also increasingly recognized
that nucleation and growth can be kinetically inhibited (Gaidies et al. 2011; Spear et al. 2014), and
so models based on reaction affinity rather than equilibrium thermodynamics may be required.
Gibbs free energy minimization
The equilibrium phase assemblage in a chemically closed system for any specified set of
conditions is classically determined using the principle of Gibbs free energy minimization.
The chemical equilibrium condition of any system of fixed composition is:
minimize Gsystem = ∑nk gk (1)

k
where nk is the number of moles and gk the molar Gibbs free energy of phase k. The interested
reader would note that this chemical equilibrium condition is subject to two additional
constraints: (1) all nk ≥ 0 with ∑ nk =
1 and (2) the mass balance equations (e.g., ∑ xki nk =i
xsystem
i
for component i with xk the composition in phase k) are satisfied. The computation of the
equilibrium phase assemblage requires values for the molar Gibbs free energy of pure phases
and a formulation of the relation between composition and activity for each solution phase.
Consequently Gibbs free energy minimization relies on thermodynamic data for the standard
state thermodynamic properties, PVT-equations of state, and solution models for solids and
fluids. Successful application of forward equilibrium models requires an internally consistent
thermodynamic database with accurate solution models and a robust estimate for the bulk
composition of the system.
The development and successive improvements of internally-consistent thermodynamic
databases for solid and fluids (Berman 1988; Holland and Powell 1988, 1998, 2011; Gottschalk
1996; Miron et al. 2016) are one of the greatest advances in the field of metamorphic petrology
in the last three decades. Using the Gibbs free energy minimization principle, these databases
facilitate the creation of phase diagrams that describe which mineral phases are stable as a function
of temperature (T), pressure (P), H2O and CO2 activity (aH2O, aCO2), oxygen fugacity ( f O2) or
chemical potential of a component i (µi). A significant move occurred in the late 80’s with the
development of modeling programs including Thermocalc (Powell et al. 1998; Powell and Holland
2008), Perple_X (Connolly 1990, 2005) and Theriak-Domino (De Capitani and Brown 1987; de
Capitani and Petrakakis 2010). Such programs can be used to generate isochemical equilibrium
phase diagrams (elsewhere called ‘pseudosection’, but this term may engender confusion as noted
by Spear et al. 2016), which map assemblages predicted at minimum free energy for the chemical
system of interest (Eqn. 1). It is owing to the remarkable power and availability of these programs
that isochemical equilibrium phase diagrams have become so widely used.
58 Lanari & Engi
The concept of chemical equilibrium and its application to metamorphic rocks

A volume of rock is said to be in chemical equilibrium when the quantities and
compositions of the phases involved do not change in time without an external influence.
Equilibrium is a macroscopic concept that is defined in terms of macroscopic variables such
as P, T and chemical potential (or activity or composition) in the considered system, which
is commonly known as the equilibration volume. It is important to note that the macroscopic
variables are large-scale average quantities, which are subject to fluctuations at a microscopic
scale. Equilibrium thermodynamics requires that, in response to any change in P–T conditions,
the equilibration volume (with a given composition vector x) will adjust its phase assemblage,
i.e., the modes and compositions of all minerals, in an attempt to reach or maintain chemical
equilibrium. By definition, the principles of chemical equilibrium predict the final state of a
system, independent of the path by which the system arrived at its present state (see below).
A schematic thin section of metapelite consisting of Grt + Chl + Ms + Bt + Pl + Qz + H2O
(mineral abbreviations are from Whitney and Evans 2010) at ~525 °C and 6 kbar is shown in
Figure 1a. The mineral phases are assumed to be homogeneous in composition and coexist
at chemical equilibrium for the given P–T conditions. Any increase in P or T would require
Figure 1. Schematic thin section consisting of garnet, chlorite, muscovite, biotite in a quartz–plagioclase
matrix. (a) Chemical equilibrium model: all mineral phases are homogeneous and coexist in equilibrium
at the P–T conditions of interest. (b) Grain boundary equilibrium model and assemblage re-equilibration
under higher temperature conditions through the reaction Chl + Ms + Qz → Grt + Bt + H2O. (c) Two equi-
librium volumes are shown, one for an element (dark gray) with low diffusivity, a second one (light gray)
assuming fast transport. (d) The local equilibrium volume involves only the rim of the zoned minerals (here
garnet) and a homogenous domain (matrix) of a section. The reactive bulk composition is the composition
of the local equilibrium volume, which excludes the domains shown in black.
adjustments through the reaction Chl + Ms + Qz → Grt + Bt + H2O (Fig. 1b) to attain a new
equilibrium state. As the solid-state transformations are achieved by dissolution, precipitation
and transport of the elemental species through the intergranular medium, the problem can be
reduced to a grain boundary equilibrium model. This concept is a step forward in the application
of chemical equilibrium theory to analyze metamorphic rocks but it requires an approximation
of the bulk composition of the intergranular medium. However some fundamental concerns
appear as soon as we abandon this idealized scenario. Most of them are related to the size
and the composition of the equilibration volume. From a theoretical point of view, transport
and reaction kinetics can establish and maintain chemical potential gradients in the system
and avoid the achievement of chemical equilibrium, i.e., an equilibration volume cannot be
defined. An additional concern comes from the observation that a metamorphic rock is seldom
truly homogeneous and often contains evidence of disequilibrium, even at grain-scale, such as
chemical zoning or mineral relics. In this case it can be challenging to define an appropriate
equilibration volume to be modeled. Those cases are separately addressed below.
Reaction kinetics. Equilibrium thermodynamic models in closed systems are based on
the equilibrium condition (Eqn. 1), and they neglect the fact that solid-state reactions cannot
strictly proceed to equilibrium (Carlson et al. 2015 and references therein) simply because
the driving force—chemical potential gradients between reactants and products—would be
eliminated. Consequently, equilibrium thermodynamics ignores (1) the time-courses of mineral
assemblage transformations, (2) the specific mechanism by which rocks crystallize, and (3)
the rate at which that crystallization occurs. Any evolution of texture from metastable states
to a stable state is a transient feature, and intermediate states are completely obliterated if the
rock reaches equilibrium. From a microscopic point of view, the successive transformations
are achieved via re-equilibration, e.g., dissolution and precipitation or pseudomorphic
replacement of mineral phases and by inter- and intragranular diffusion of the chemical
components (Fig. 1b). The driving forces for diffusion are the chemical potential gradients
that are established between minerals such as the reactants and products of a metamorphic
reaction or between different groups of minerals, such as between domains (Fisher 1973).
In this context, equilibration is attained when the chemical potentials of each the chemical
component in the system of interest is equalized across a rock. Several studies on equilibrium
have demonstrated that deformation and fluid influx can help to achieve equilibration (Brodie
and Rutter 1985; Wintsch 1985; Foster 1991; Erambert and Austrheim 1993).
Frozen chemical potential gradients. The modeling of rock volumes that attained
chemical equilibrium, such as shown in the idealized scenario of Figure 1a, is relatively
straightforward using Equation (1) and an appropriate thermodynamic database. However,
nature provides abundant examples of diffusion-controlled structures such as coronas, which
typically result from incomplete metamorphic transformation due to low rates of elemental
movement by intragranular diffusion. Indeed chemical zoning in coronas may also reflect
chemical potential gradients, which must be distinguished from zoning due to changes in P–T
conditions (Indares and Rivers 1995). Diffusion-controlled structures can be identified in thin
section because they typically have a strong spatial organization with textural mineral zones
showing sharp changes in compositions at zone boundaries. Reaction fronts are arranged
in an orderly sequence of increasing or decreasing chemical potential (Fisher 1973). Non-
equilibrium thermodynamics (also known as non-classical or irreversible thermodynamics)
has been used to analyze such inhomogeneous systems; their study requires the knowledge
of the rates of reactions (Fisher 1973; Foster 1977, 1986, 1999; Fisher and Lasaga 1981;
Johnson and Carlson 1990; Carlson and Johnson 1991; Lasaga 1998).
Mineral compositional zoning and mineral relics. Growth related compositional zoning
and mineral relics are clear evidence of disequilibrium in metamorphic rocks. It was recognized
early after the introduction of the EPMA to the geosciences that garnet porphyroblasts can
60 Lanari & Engi
freeze compositional zoning (Hollister 1966). Such textures reflect changes in P–T conditions,
indicating that the crystal as a whole did not equilibrate with the matrix during growth (Atherton
1968; Tracy et al. 1976). In such cases some part of the minerals that formed earlier (e.g.,
garnet) is effectively removed from the reactive part of the rock (the effects of metamorphic
fractionation are intensely discussed below, in the section related to garnet porphyroblast
growth). A few years later, Tracy (1982) reported a list of 18 metamorphic minerals showing
similar compositional zoning. As most metamorphic rocks exhibit compositionally zoned
minerals, equilibrium at best was established at a local scale only (Fig. 1c). Hence, to apply
equilibrium thermodynamics, more sophisticated models are required.
The principle of local equilibrium. The principle of local equilibrium (or mosaic
equilibrium) restricts the investigation to subsystems that are small enough that the attainment
of an equilibrium state can be assumed (Korzhinskii 1936, 1959; Thompson 1955, 1959, 1970).
Local equilibrium is considered only at a scale over which variations in chemical potential—
and all physical variables—are negligible. Since chemical heterogeneity in a sample is primarily
evident in solid solutions, local equilibria address only those situations where the minerals
(or individual zones thereof) are chemically uniform. The principle of local equilibrium
significantly extends thermodynamic modeling to cases that would otherwise require a kinetic
description (White et al. 2008). This includes, as we shall see, the domain in which chemical
potential gradients were frozen in and the mineral relics (e.g., cores) were isolated from the
reactive part of the rock. The concept of local equilibrium is thus useful for forward equilibrium
models, but it demands that we define the equilibration volume to be investigated.
Bulk rock composition vs reactive bulk composition
In the literature of metamorphic petrology, the term bulk rock composition generally
refers to the average chemical composition of a whole-rock sample analyzed, for example by
X-ray fluorescence spectrometry (XRF). In the framework of local equilibrium, the analysis
can be restricted to a specific region or domain—its size often is less than thin-section scale—
in a rock, chosen for its uniform chemical composition (or mineral modes). In this case we
refer to the concept of local bulk composition. Note that bulk rock compositions and local
bulk compositions are measured quantities and may or may not be relevant for modeling, i.e.,
representative of the equilibrium volume to be investigated.
Various studies over the past 15 years struggled to define a relevant bulk composition for a
given sample situation, and this question typically was raised when models using the bulk rock
composition rendered unrealistic results (e.g., Warren and Waters 2006). As discussed above,
the relevant bulk composition of a rock for equilibrium models is the composition of the
(presumed) equilibration volume at a specific stage of the evolution, and it may exclude certain
refractory or inert minerals that are observed in a domain, but may be shielded from reactions.
The composition of the equilibration volume is known as the effective bulk composition
or reactive bulk composition (Fig. 1d). The term effective bulk composition was originally
introduced by Tracy (1982) and reused in several studies (Hickmott et al. 1987; Spear 1988b;
Stüwe 1997; Marmo et al. 2002; Evans 2004; Tinkham and Ghent 2005a), but in the following
we decided to use the term reactive bulk composition as it is most descriptive in a modeling
framework. As will be demonstrated below in the section on garnet porphyroblast growth,
the reactive bulk composition changes along a P–T trajectory because of compositional
fractionation (Tracy 1982; Spear 1988b; Spear et al. 1990).
Open questions on bulk composition effects and the size of equilibration volume
Forward equilibrium models based on the assumption of local equilibrium require the
use of reactive bulk compositions, which in some favorable cases can be approximated by
local bulk compositions or calculated using fractional crystallization models. The choice of an
appropriate or representative rock or domain composition can be critical, as it may significantly
affect the results of modeling. Aware of the perils, Kelsey and Hand (2015) warned that “this
topic is in some ways like the elephant in the room”. The question is: How sensitive are
predictions made by equilibrium models to variations in the bulk rock composition? Detailed
investigations have been conduced at the regional scale (Tinkham et al. 2001; White et al.
2003), but it appears that only a few studies have explicitly addressed this question at the sample
scale. The paucity of knowledge is due in part to the difficulty of estimating uncertainties in the
bulk rock composition and then propagating their effect.
Are realistic equilibration volumes impossible to identify exactly, as claimed by Stüwe
(1997)? Despite many studies over more than 25 years (e.g., Powell and Downes 1990; Vance
and Mahar 1998; Brown 2002; Marmo et al. 2002; Powell et al. 2005), it remains intriguing
which parameters control the size of the equilibration volume. Some progress has been made in
accounting for phases that should be included or excluded from the equilibration volume (Lanari
et al. 2017). If an intergranular medium (specifically a fluid) is present in a rock, transport
distances of dissolved species in this medium are critical (Carlson 2002); but what if mobility
was limited to volume diffusion? Orders of magnitude difference in transport result from this
uncertainty, especially in highly deformed rocks, since strain can increase the dislocation
density, reactivity, and transport rates. In any case, diffusion rates are temperature-dependent,
so equilibration volume is expected to increase during heating (and drastically decrease upon
cooling, as rocks tend to “dry up”), and as shown in Figure 1c, it may differ substantially for
different chemical species (Carlson 1989; Spear and Daniel 2001). This is particularly critical for
some of the trace elements relevant to petrochronology (e.g., Gatewood et al. 2015). An essential
parameter controlling the extent of local (dis)equilibrium is the rate at which local reactions
proceed towards equilibrium compared to the rate of change in physical parameters (P–T–X,
e.g., the heating rate). If kinetics are favorable, the system can respond rapidly by adjusting the
mode (i.e., the volumetric abundance) and composition of each phase in the assemblage to the
P–T–X conditions of the equilibration volume; if so, all the minerals in that domain will be stable
at the same reactive bulk composition. If not, either a smaller domain size (or component space)
must be chosen, or a kinetic approach is required. More work is needed to quantify the size of
the equilibration volume, and future studies should take into account the rate at which rocks
“travelled through P–T space”, i.e., to link petrogenetic analysis to chronology.
PORPHYROBLAST GROWTH
Porphyroblastic growth implies a continuous supply of mineral constituents and a
relatively high activation energy barrier for nucleation hence growth of existing grains is
favored over formation of new nuclei. The most common example is garnet, whose unique
chemical and mechanical properties can record evidence of a potentially complex path that
the host rock experienced during a petrogenetic cycle (Spear et al. 1991; Vance and Mahar
1998; Caddick and Kohn 2013; Ague and Axler 2016). Though diffusion can bias the record
at high temperatures (Anderson and Olimpio 1977; Tracy 1982; Caddick et al. 2010; Ganguly
2010; Kohn and Penniston-Dorland 2017, this volume), chemical zoning is commonly
preserved. In metapelites for instance, garnet commonly exhibits clear compositional zoning
with a systematic, bell-shaped decrease of Mn from core to rim (spessartine fraction, Xsps in
Fig. 2a). But garnet may also provide clues about polycyclic metamorphism, where core and
rim relate to two distinct metamorphic cycles (Fig. 2b). Garnet is quite resilient to mechanical
and chemical weathering. In fact, metamorphic overgrowths on detrital garnet (Manzotti and
Ballèvre 2013) have been recognized (Fig. 2c). Strong compositional zoning in both mono-
and polycyclic garnet demonstrates its refractory nature; it does not, in general, completely
re-equilibrate with the matrix during growth (Atherton 1968), though outside its stability field
garnet readily interacts with hydrous fluid or melt to form secondary phases, for example,
62 Lanari & Engi
Figure 2. High-resolution end-member compositional maps of garnet grains from the Western Alps. (a)
Typical bell-shaped zoning of spessartine of garnet from a metasediment of the Mont Emilius Klippe
(Burn 2016); Xsps = Mn/(Mn + Ca + Fe + Mg). (b) Grossular zoning in a garnet crystal from an eclogitic
micaschist of the Sesia Zone (Giuntoli 2016; Lanari et al. 2016); Xgrs = Ca/(Mn + Ca + Fe + Mg). This garnet
porphyroblast recorded two distinct cycles of metamorphism: a Permian LP–HT core (Grt1) and three Up-
per Cretaceous (Alpine) HP–LT rims (Grt2, Grt3, Grt4). (c) Grossular zoning in polycyclic garnet from the
Zone Houllière with a LT overgrowth of spessartine-rich hydrothermal garnet (Grt2); the core (Grt1) is a
detrital fragment that preserves pre-depositional internal zoning (Dupuis 2012).
retrograde chlorite. But where preserved, garnet serves as an archive of the P–T–X conditions
of formation, as long as intracrystalline diffusion remained negligible (Spear et al. 1984;
Florence and Spear 1991; Caddick et al. 2010; Kohn 2014). This fortunate ability to record
and preserve chemical zoning is the main reason why garnet compositions have been so
extensively used as an indicator of P–T conditions throughout its wide stability range. A
method referred to as isopleth thermobarometry for garnet is based on Duhem’s theorem,
which simply states that P–T conditions determine the composition and modal abundance
of all phases in a closed system at equilibrium; hence the intersection of garnet isopleths
for 3–4 endmember components in garnet (Fe, Mg, Ca, Mn) in P–T space indicate whether
equilibrium was closely approached and, if so, under what P–T conditions. Various modeling
techniques have been specially designed to be applied to garnet thermobarometry (Spear and
Selverstone 1983; Spear et al. 1984, 1991; Evans 2004; Zeh 2006; Gaidies et al. 2008, 2011;
Konrad-Schmolke et al. 2008; Schwarz et al. 2011; Moynihan and Pattison 2013; Vrijmoed
and Hacker 2014; Lanari et al. 2017). However, determining accurate P–T conditions by
modeling garnet zoning profiles demands some understanding of the interplay between
chemical equilibrium (e.g., Spear and Daniel 2001; Gaidies et al. 2008), reaction kinetics
(e.g., Gaidies et al. 2011; Pattison et al. 2011; Schwarz et al. 2011), and post crystallization
intragranular diffusion (e.g., Ganguly et al. 1996; Caddick et al. 2010).
Several issues are separately addressed in the next paragraphs: (1) We first recall some
basics of nucleation and growth. (2) Porphyroblast growth is then addressed, i.e., the successive
alienation of core parts, as they typically become isolated from the reactive part of the rock,
thus segregating components. In terms of the reaction volume in the matrix, this process causes
and progressively enhances chemical fractionation of the reactive bulk composition. (3) The
effects of fractionation on isopleth thermobarometry are then quantitatively illustrated for two
examples: A pelitic schist that experienced Barrovian metamorphism (Moynihan and Pattison
2013), and a typical MORB that underwent eclogite facies transformation (Konrad-Schmolke et
al. 2008). (4) Different automated strategies are currently in use to retrieve P–T information from
garnet zoning based on equilibrium thermodynamics. These are presented and briefly discussed,
then garnet resorption is addressed for those models. (5) Finally, the concept of distinct growth
stages is reviewed based on some studies that link such metamorphic stages to age data.
Crystal nucleation and growth

Four main processes are involved in the nucleation and growth of a porphyroblast crystal:
(1) Dissolution of source material provides potential nutrient material; (2) Nutrients generated
at various locations and / or from different sources are transported through the intergranular
medium to nucleation sites. (3) Nucleation occurs at the atomic scale where chemical species
from the transport medium and possibly local reactant phases rearrange into a cluster of the
product phase(s) of sufficient size to be thermodynamically stable. (4) Further precipitation
onto an existing surface is termed crystal growth. The reaction interface encompasses both
the detachment / dissolution at the reactant site and the attachment / precipitation of nutrients at
the product site. In this conceptual model, kinetics may affect a metamorphic reaction during
nucleation, during intergranular transport, and at the reaction interface. The slowest one of these
three processes usually dominates the overall reaction kinetics (Pattison et al. 2011). Detachment
and attachment are surface-processes whereas element transfer is a transport-process. Nucleation
and growth can be affected by microstructures (favorable site model) and by reaction overstepping.
Favorable site model. Microstructural features may catalyze nucleation by increasing
the local free energy, which amounts to reducing the critical energy barrier. Such is the case,
for example, in crenulation hinges (Bell et al. 1986) or at grain boundaries, dislocations and
cracks (Gaidies et al. 2011). Several lines of evidence support the interpretation that garnet
growth is more rapid along triple-grain intersections than along boundaries separating only
two grains (Spear and Daniel 2001). The heterogeneous distribution of porphyroblasts in
a rock may reflect effects of the local bulk composition and / or deformation. For example,
dissolution may occur along shear-dominated zones such as the crenulation limbs, whereas
nucleation occurs at the crenulation hinges (Bell et al. 1986; Vernon 1989; Williams 1994).
Several local processes may thus be involved in the overall transformation process leading
to porphyroblast growth (Carmichael 1969).
Reaction overstepping. Even if a product phase is nominally part of the thermodynamically
stable assemblage of the reaction volume, it may fail to nucleate (Waters and Lovegrove 2002). A
metamorphic reaction proceeds only if and when the mechanical or chemical energy overcomes
the kinetic barriers for nucleation. This delay can be viewed as a disequilibrium state required to
gain the amount of energy required to initiate the interface between reactants and products that will
initiate the reaction. The so-called interfacial energy (Gaidies et al. 2011 and references therein)
controls the departure from equilibrium required before a phase (in our case garnet) nucleates,
and the free energy of the system can attain a lower energy state. Reaction overstepping can be
estimated as the difference in Gibbs free energy between the thermodynamically stable, but not
yet crystallized products and the metastable reactants. This quantity is known as the reaction
affinity (e.g., Fisher and Lasaga 1981; Lasaga 1986; White 2013). P–T–reaction affinity maps
computed with Theriak-Domino have been used to predict metamorphic reaction overstepping
(Pattison et al. 2011). Such models demonstrate, for example, that thermal overstepping may
have some dependence on the heating rate (Gaidies et al. 2011; Pattison et al. 2011).
Models of equilibrium and transport control
In the framework of modeling garnet growth we distinguish two end-member models.
Equilibrium control. Chemical equilibrium is maintained at the rim of all growing
phases at all times if transport rates are faster than the rates of surface processes. Equilibrium
thermodynamics can be used to predict the distribution of all components among the various
phases of the system. This grain boundary equilibrium model (Fig. 1b) implies that the rim
composition of all the crystallizing phases will be identical and dictated by P–T conditions and
other intensive variables such as fluid composition and fO2.
64 Lanari & Engi
Transport control. If transport is the limiting factor, gradients in chemical potential

are established (at least for some elements) and may evolve via the intergranular medium.
Consequently, the composition of the growing phase will be controlled by the flux of these
elements to the reaction interface. As thermodynamics predicts the macroscopic equilibrium
state only, a kinetic description must be used to model how and how fast equilibrium will be
approached at a specific reaction site.
In rocks, the attainment of chemical equilibrium and departures from it depend on
scale, in both space and time. Several considerations are necessary to decide what modeling
approach is most appropriate for a given situation; strong cases can be made for combining the
equilibrium control and transport control approaches (e.g., Carlson et al. 2015). When the goal
is to analyze and quantify petrogenesis of rock samples as a basis for detailed chronology, we
regard the following considerations as essential:
• Equilibrium control may occur on different scales for different elements (Spear and
Daniel 2001). Specifically, highly charged or large ions often are far less mobile than
smaller or less highly charged ions (Fig. 1c).
• The concept of “chemical homogeneity” is useless beyond upper and lower spatial limits;
these limits essentially depend on rock texture (e.g., layering, grain size), which evolves
during petrogenesis. The spatial scale of chemical uniformity is bound to change by
rock-forming processes, e.g., as a few components are sequestered by porphyroblasts or
simply as static Ostwald ripening or dynamic recrystallization proceeds, by mechanical
grain size reduction (grinding), or differentiation induced by plastic deformation, partial
melting or metasomatic processes involving fluids.
• Trace elements contribute little in terms of the overall free energy of a rock. Equally
minor is the contribution to the driving force towards equilibrium that chemical potential
gradients in the trace elements may add.
• A few accessory minerals (e.g., monazite, zircon, titanite, allanite, apatite) largely
sequester certain trace elements of particular interest to petrochronology. Their transport
(by diffusion) is typically slow because they include large, highly charged ions, such as U4+,
Th4+, and REE3+, and they are involved in heterovalent coupled exchange mechanisms,
but their mobility may be enhanced by deformation, radiation damage, and reactive fluids
(e.g., Tropper et al. 2011). Modelers should be aware that accessory minerals commonly
occur as relics (often armored) that appear to have metastably survived high temperatures,
even partial melting.
• In general, some parts of a rock’s reaction path may be kinetically sensitive while others
are not.
Taken together, these considerations indicate that sober modeling should recognize the
potentials and limits of equilibrium and transport models, and possible combinations (Gaidies
et al. 2011; Schwarz et al. 2011). A conceptual division of samples may be most appropriate in
practice. For instance, it remains a largely open question to what extent trace element mobility
will allow chemical potential gradients to be retained (or to disappear) at the temporal scale
of rock-forming processes.2 Reaction-driven exchanges of these trace elements between major
rock-forming minerals and accessory phases, possibly mediated by fluids, may or may not be
sufficient to even out concentration gradients. Current understanding of transport properties is
quite insufficient for the case of highly charged minor and trace elements. The extent to which
small and sparsely present grains are able to communicate deserves careful study in the future,
as this promises to tighten the links between accessory mineral chronometry and petrogenetic
interpretations. For example, relations between zircon equilibration and Zr-in-rutile thermometry
are well documented (Meyer et al. 2011; Taylor-Jones and Powell 2015; Kohn et al. 2016).
2
Note that these time scales could well be much shorter than could be resolved by mineral chronology
(say < 103–105 years).
With our present goal of quantifying petrogenetic conditions, we concentrate on

equilibrium control in this chapter with the assumption of local equilibrium that reduces the
need to consider transport. We show that the power of thermodynamics can be enhanced and
several limitations reduced by restricting the analysis to local spatial domains.
Fractionation of the reactive bulk composition during porphyroblast growth
Some component of growth zoning in garnet such as Mn are largely controlled by reactive
bulk composition change generated by crystal fractionation. Mn distribution in garnet often
shows a progressive decrease with a typical “bell-shaped” zoning profile from the center
toward the rim (Fig. 2a, 3a). In this case the segregation of Mn during garnet growth is directly
controlled by the progressive depletion of Mn in the surrounding matrix. Many studies
explained such observation by a fractionation process following a Rayleigh’s (or Pfann’s)
model (Hollister 1966; Atherton 1968; Cygan and Lasaga 1982; Evans 2004). Based on those
results, a simple method was proposed by Evans (2004) to generate composition versus modal
proportion curves for garnet. This method models the effects of the crystal fraction on the
MnO content of the reactive bulk rock using the Rayleigh fractionation equation:
K d (1 − wgrt )
K d −1
=MnO
Cgrt MnO
Cbulk (2)
MnO MnO
where Cgrt and Cbulk are the concentrations of MnO (in oxide wt%); Kd is the bulk distribution
MnO MnO
coefficient (by mass) for the element between garnet and matrix (Cgrt / Cmtx ) and wgrt is the
mass fraction of garnet in the rock. For K d > 1, the species partitions into the porphyroblast;
a characteristic profile across a grain should be “bell shaped” (Fig. 3a), whereas for K d < 1
the species partitions into the matrix, i.e., growing porphyroblasts are depleted relative to
the matrix and show a U-shaped profile. If neither P–T conditions nor equilibrium phase
relations change drastically during early garnet growth, Kd can be approximated by the ratio
MnO MnO
between Cgrt in the earliest garnet nucleus and Cbulk (Gaidies et al. 2006). As partitioning
among minerals depends on temperature, and because changes in T drive mineral growth, Kd
is expected to change as the mineral grows (Hollister 1966; Kohn 2014).
To quantify the effects of fractionation of the reactive bulk composition, a case study is
selected for which the prograde P–T conditions of garnet growth are well constrained. Moynihan
and Pattison (2013) analyzed garnet porphyroblasts in a garnet- and staurolite-bearing schist
(sample DM-06-128) from the southern Omineca belt of the Canadian Cordillera that underwent
Barrovian metamorphism peaking at middle amphibolite facies during the Early Cretaceous. The
evolution involved heating and burial along a linear P–T trend, followed by a heating-dominated
stage accompanied first by exhumation, then renewed burial. Garnet porphyroblasts grew along a
P–T path from 500 °C, 5 kbar to 570 °C, 7 kbar. From the zoning profile reported in Moynihan and
Pattison (2013), we obtained a Kd value of 62 using the approximation of Gaidies et al. (2006).
The corresponding MnO composition of garnet was calculated using Equation (2) and plotted
against the volume percentage of garnet produced (Fig. 3c). Assuming a single population of
garnet with ~5 mm diameter for a total amount of 6 vol% of garnet in the rock (as predicted by
the thermodynamic model) the zoning profile matches Rayleigh fractionation with Kd = 62 ± 10,
reproducing to a first order the chemical zoning in MnO for this rock. In the matrix Mn is
distributed between chlorite and ilmenite (Moynihan and Pattison 2013).
This technique of Evans (2004) has been successfully used in various cases to model garnet
fractionation based on MnO zoning profiles (e.g., Gaidies et al. 2006; Sayab 2006; Groppo and
Castelli 2010; Vitale Brovarone et al. 2011; Cheng and Cao 2015). Zoning profiles of other
elements may reflect more complicated fractionation, with non-Rayleigh behavior and variable
Kd during garnet growth. This is expected to occur along a P–T path that has successively involved
more than one garnet-producing reactions as for most amphibolite- to upper amphibolite-facies
66 Lanari & Engi
Figure 3. Garnet compositional zoning from a schist that experienced Barrovian metamorphism (modified
from Moynihan and Pattison 2013) against (a) the radius, (b) the fractional volume of a single garnet por-
phyroblast, (c) the garnet volume fraction assuming a mode of 6 vol% garnet and a single garnet size popu-
lation. The Rayleigh fractionation model for Kd = 62 ± 10 is shown by the black curve in the gray domain.
rocks. For both simple and complex growth histories, quantitative models based on Gibbs free
energy minimization must be favored to approximate the changes in the reactive bulk composition.
Equilibrium crystallization models vs fractional crystallization models
Two models incorporating or ignoring chemical fractionation effects during porphyroblast
growth are presented and compared here. For garnet we show that (1) fractionation has a strong
effect on the amount of garnet predicted, and (2) isopleth thermobarometry systematically
produces erroneous P–T estimates if fractionation is ignored.
Equilibrium crystallization models (ECM). These classic models rely on isochemical P–T
equilibrium phase diagrams for the given bulk rock composition and isopleth thermobarometry
(Spear 1988a). Such a model cannot ‘predict’ chemical zoning because all the phases are
assumed to re-equilibrate at any P–T conditions. In theory, the use of ECM models must be
restricted to well-equilibrated mineral assemblages showing no relics, no chemical zoning in
the mineral phases, and no partial local re-equilibration. In practice, such diagrams are often
used to reveal first-order compositional and modal trends that help to interpret metamorphic
mineral growth and composition changes (Kohn 2014), but the results may be inaccurate.
Fractional crystallization model (FCM). These models assume chemical fractionation

of some mineral phases (usually garnet) that are produced at each step of a P–T path, where
the grains produced are immediately isolated from the reactive part of the rock (Spear 1988b;
Spear et al. 1991; Gaidies et al. 2008; Konrad-Schmolke et al. 2008). This is equivalent to
removing the fractionating phase from the system. Step-wise growth modeling is applied,
with the composition and volume fraction of the fractionating phases from the previous step(s)
being isolated or removed from the reactive system.
The first major difference between FCM and ECM is the modal amount of the phase produced
along a given P–T path. In their study, Konrad-Schmolke et al. (2008) documented this effect of
garnet fractionation under eclogite facies conditions using a typical MORB composition. Garnet
is predicted to be stable along a prograde trajectory from 525 °C, 15 kbar to 750 °C, 35 kbar using
ECM (Fig. 4 in Konrad-Schmolke et al. 2008). By contrast, garnet growth ends at 650 °C and
25 kbar using FCM because the matrix is strongly depleted due to garnet fractionation (Fig. 5 in
Konrad-Schmolke et al. 2008). Above that limit, garnet is no longer stable in the new reactive
bulk composition. There will be a significant difference between two tectono-metamorphic
scenarios based on the results from FCM and ECM models. In the present example, the FCM
shows that garnet does not ‘record’ the pressure peak reached by the rock. It is important to note
that the volume of garnet predicted by ECM is systematically higher than the volume predicted
by FCM (Fig. 4). Accurate modeling of mineral modes along any P–T trajectory requires the use
of FCM as soon as compositional zoning is observed in a sample.
To evaluate the consequences of garnet crystallization on the reactive bulk composition
of typical metamorphic rocks, we again analyze the two examples of garnet growth in (1) a
metapelitic schist along a typical LP–HT trajectory (Moynihan and Pattison 2013) and (2) a
metabasite with a typical MORB composition along a HP-LT trajectory (Konrad-Schmolke et
al. 2008). The changes in the reactive bulk rock compositions predicted using a FCM are plotted
in Figure 5 against the volume percentage of garnet produced. Note that in the metapelitic schist
only 6 vol% of garnet is produced along the P–T trajectory. For the metabasite, ~13 vol%
is produced along the selected P–T trajectory. For the metapelitic schist, all the components
Figure 4. Volume of garnet (in vol%) predicted by ECM and FCM for a typical MORB composition along
the P–T path used by Konrad-Schmolke et al. (2008). Model calculated for the system SiO2–Al2O3–FeO–
Fe2O3–MgO–CaO–Na2O using the thermodynamic dataset tc55.txt (Holland and Powell 1998), oxygen
fugacity controlled by hematite-magnetite buffer.
68 Lanari & Engi
Figure 5. Evolution of the reactive bulk rock composition using a fractional crystallization model (FCM)
for garnet growth in pelitic schist (Moynihan and Pattison 2013) and MORB (Konrad-Schmolke et al.
2008) along a LP–HT and a HP–LT path, respectively. Note that along the respective P–T paths only
6 vol% of garnet is produced in the schist, whereas in meta-MORB the garnet mode reaches 40 vol% garnet
(shown here to 12 vol% only, but the same trend continues). (e) Relative differences of the reactive bulk
compositions to the original bulk rock composition for the two samples; note that oxides are color-coded.
The gray band outlines relative differences of < 5%.
(except Al2O3 and MgO) show significant variation during garnet growth (Fig. 5e). As previously
discussed, the matrix quickly becomes depleted in MnO, i.e., the matrix loses > 90% of the
initial MnO once the garnet fraction reaches 2 vol%. To a lesser extent CaO and FeO decrease
and finally reach 60 and 70% of their initial value once 6 vol% garnet formed. Fractionation is
less pronounced for the metabasite (Fig. 5). Still, FeO and MgO are significantly affected once
the garnet proportion reaches 6 vol%. From these two examples we can conclude that garnet
growth systematically affects the reactive bulk composition.
What is the effect of FCM on equilibrium phase diagrams? For small fraction of garnet
produced, FCM has a minor effect on the major phase relations (Zuluaga et al. 2005; Groppo
et al. 2007; Moynihan and Pattison 2013). However, the compositional differences of garnet
predicted stable by ECM and FCM at given P–T conditions are systematic and large enough
to affect isopleth thermobarometry (Spear 1988b; Evans 2004; Gaidies et al. 2006; Chapman
et al. 2011). To evaluate this, the program GrtMod is useful because it searches the P–T
conditions for which the model composition of garnet best matches the measured composition
(Lanari et al. 2017), and it can be used to check differences in garnet isopleth thermobarometry
between FCM and ECM model. For the schist DM-06-128 (Moynihan and Pattison 2013),
this comparison was made along the original P–T path (500–650 °C, 3–12 kbar), taking three
steps (at 2, 4, and 6 vol% garnet produced). Results are reported in Table 1. It is evident that
ECM does not find satisfactory isopleth intersection in the P–T space for the three selected
cases. In the first case (2 vol%) a ‘best match’ is found, but with a relatively high residual (Co
value in Lanari et al. 2017), and it is predicted at higher pressure. Deviations are significant for
almandine and grossular components (modeled: 0.74 and 0.16; measured: 0.78 and 0.14). Such
deviations are due to the overestimation of CaO and FeO in the bulk rock composition, and
these affect the composition of phases coexisting with garnet (Fig. 5). For the second and third
cases (4 and 6 vol% of garnet), the deviations between modeled and observed compositions
become even larger, i.e., they show much higher residuals. In both cases GrtMod found two
distinct local minima but the lowest Co values always occur at higher pressure. Deviations of
0.10 and 0.04 are systematically observed in almandine and grossular components.
From these sensitivity tests we can conclude that FCM must always be used to model
porphyroblasts growth if compositional growth zoning is observed. Note that the compositional
changes become significant as soon as porphyroblasts represent > 2 vol% in pelitic systems
and > 4 vol% in mafic systems (Fig. 5). In the following FCM are used to model growth and
resorption of garnet porphyroblasts.
Table 1. Differences in P–T results between fractional crystallization models (FCM) and equilibrium
crystallization model (ECM) for three stages of garnet growth (2, 4 and 6 vol% of garnet growth) along the
prograde P–T path of Moynihan and Pattison (2013). Co is the residuum calculated by GrtMod. A value
of Co < 0.3 generally indicates a good fit between model and measured compositions (Lanari et al. 2017).
T (°C) P (bar) Co XPrp XGrs XAlm XSps
Case 1: 2 vol% of garnet produced
Reference 554 6916 0.055 0.141 0.779 0.024
FCM 554 6916 0.0001 0.055 0.141 0.779 0.024
ECM - S1 555 11996 0.0489 0.073 0.156 0.737 0.034
Reference 571 6427 0.081 0.086 0.831 0.002
FCM 571 6427 0.0002 0.081 0.086 0.831 0.002
ECM - S1 572 11994 0.1017 0.103 0.126 0.744 0.027
ECM - S2 584 7690 0.1092 0.110 0.122 0.737 0.031
Reference 583 6797 0.104 0.082 0.814 0.000
FCM 583 6797 0.0001 0.104 0.082 0.814 0.000
ECM - S1 581 11996 0.0869 0.121 0.115 0.739 0.025
ECM - S2 591 7731 0.0920 0.126 0.112 0.734 0.029
70 Lanari & Engi
Automated fractional crystallization models designed to retrieve P–T paths from

chemical zoning
Examples of simplified fractionation modeling based on BSE or X-ray images and
electron microprobe analyses can be found in many studies (e.g., Marmo et al. 2002; Tinkham
and Ghent 2005a; Zuluaga et al. 2005; Zeh 2006; Caddick et al. 2007). Four methods presented
here are based on forward equilibrium models and require no manual alteration of the reactive
bulk composition to account for material sequestered in garnet crystal cores.
• The first automated method models garnet zoning and the composition of coexisting
phases along an arbitrarily selected P–T path using Gibbs free energy minimization.
The generated compositional profiles are then compared against data along high-
resolution zoning profiles analyzed by electron microprobe (Konrad-Schmolke et al.
2008; Hoschek 2013; Robyr et al. 2014). These models merely test if the chosen P–T
trajectory is in accordance with the observed zoning (or vice versa). In fact, the three
studies cited found some mismatch between model compositions and observed zoning
profiles, in some cases with systematic and large discrepancies (> 0.08 in the end-
member proportions). Is the P–T trajectory at fault or the assumption of grain boundary
equilibrium? The two following approaches aim to derive the optimal P–T trajectory
and proposed numerical methods to improve its selection.
• Moynihan and Pattison (2013) provide a MATLAB©-based program linked to Theriak
(de Capitani and Brown 1987; de Capitani and Petrakakis 2010) that uses an inverse
modeling strategy to derive the “best” P–T trajectory by minimizing a misfit parameter,
i.e., the weighted differences between measured and model compositions. For any garnet
spot analysis, the best P–T conditions are found by matching the model against the
measured composition, and the procedure is successively applied to all analyzed points
from core to rim. Using this approach, Moynihan and Pattison (2013) successfully
modeled the observed zoning profile (see their Fig. 6a). The match is excellent and
demonstrates that in favorable cases, equilibrium thermodynamics in the context of
fractional crystallization can be successfully used to model porphyroblast growth,
provided the P–T path is part of the model optimization.
• Similarly, Vrijmoed and Hacker (2014) provided a MATLAB©-script linked to Perple_X
(Connolly 1990, 2005). It uses a brute-force inverse computational method to determine
the best P–T trajectory by minimizing the differences between predicted and the entire
measured garnet compositional profiles along different trajectories. This routine
examines a multitude of paths from an unspecified starting P–T to a predetermined
maximum P–T point. The best match is selected such that all endmember mole fractions
of the fractionated garnet show least discrepancy between data and model. Several tests
based on published zoning profiles data show that brute force can pay off.
• Finally, a fourth approach is of interest: Program Theria_G (Gaidies et al. 2008) allows
the simulation of porphyroblast nucleation and growth using a FCM for any given P–T
trajectory, and it takes into account further possible modifications driven by intragranular
multi-component diffusion. Theria_G uses the Gibbs free energy minimization routine
of Theriak and simulates the formation of a garnet population with variable grain size
that can be compared with observations. The P–T trajectory can also be part of the model
optimization as shown by Moynihan and Pattison (2013).
Crystal resorption and implications on fractional crystallization models
The models presented above take into account the fractionation due to garnet growth and
how this affects the reactive bulk composition. However, garnet may also be affected by local
resorption (de Béthune et al. 1975; Kohn and Spear 2000; Ague and Axler 2016), in some
extreme cases leading to atoll or mushroom garnets (Cheng et al. 2007; Faryad et al. 2010;
Robyr et al. 2014). More commonly, the production of staurolite and biotite at the expense
of garnet and chlorite causes the resorption of garnet and a step in the zoning profile (Spear
1991). Garnet resorption has been invoked to explain the peripheral increase in MnO (dubbed
near-rim kick-up) observed in some grains. Typically this results from partial consumption
or dissolution of garnet crystals during cooling and exhumation, which returns Mn to the
reactive matrix (Kohn and Spear 2000), and Mn can show retrodiffusion features in garnet
(de Béthune et al. 1975). The amount of Mn in the near-rim kick-up has been used as a semi-
quantitative measure of the volume of garnet dissolved. Kohn and Spear (2000) estimated that
45% of garnet was dissolved in an amphibolite facies metapelite from the central Himalaya
of Nepal. Based on forward equilibrium models (see below) and different types of evidence
Lanari et al. (2017) estimated that 50% of garnet in eclogitic micaschist was dissolved during
HP metamorphism before a new episode of garnet growth. As shown above garnet chemical
fractionation has a strong impact on the reactive bulk composition. Fractional crystallization
drives the reactive bulk composition away from the garnet composition. Thus, resorption of
garnet drives the reactive bulk composition back towards the garnet composition.
The effects of garnet resorption on forward equilibrium models were investigated by Lanari
et al. (2017), who proposed a numerical strategy and a MATLAB©-based program (GrtMod)
linked to Theriak. It allows numerical simulation of the evolution of garnet based on compositions
of successive growth zones, optimization of successive local reactive bulk compositions,
accommodating resorption and / or fractionation of previously crystallized garnet. Each growth
stage here is defined as an interval during which garnet grows at fixed P–T–X conditions while
in equilibrium with the same stable matrix assemblage. This approach uses compositional maps
(instead of profiles) to define the growth zones and to estimate average compositions.
Timing of porphyroblast growth
In porphyroblast petrochronology, two different ways are generally used to link
metamorphic age (t) with temperature and pressure: (1) in-situ dating of the porphyroblast
or (2) textural correlation between the porphyroblast and dateable accessory minerals either
trapped as inclusions or chemically correlated to distinct zones within a porphyroblast (e.g.,
Pyle and Spear 1999; Regis et al. 2014).
In situ dating of garnet porphyroblasts is the most attractive method to constrain the time
interval of growth because porphyroblasts commonly contain inclusions that help constrain
P–T, as can the chemical zones that recorded successive growth stages. Several potentially
useful isotopic systems (Sm–Nd, Rb–Sr, Lu–Hf) are extensively discussed in Baxter et al.
(2017, this volume). Rb–Sr dating of garnet is restricted to a few cases because it is not always
realistic to assume that the whole-rock or matrix compositions adequately reflect the reactive
bulk composition at the time garnet grew (Sousa et al. 2013). Sm–Nd is most commonly used
because Sm and Nd generally are uniformly distributed in garnet (Kohn 2009), whereas Lu
and Hf may be restricted towards early garnet growth (because Lu strongly fractionates into
the core, depleting the matrix in Lu). In addition, much more material is required per analysis
to obtain ages of specific zones using Lu / Hf. Sm / Nd is most powerful for dating specific
garnet zones (e.g., Pollington and Baxter 2010), since improved analytical techniques now
permit analysis of very small volumes (Harvey and Baxter 2009; Dragovic et al. 2012), such
as single growth zones mapped by electron microprobe (Gatewood et al. 2015). Ages are
calculated based on isochrons for each garnet zone and the surrounding matrix, and such ages
rest on assumptions about the extent of trace element homogenization.
Instead of applying direct garnet dating many studies have used textural correlations
between the porphyroblast and dateable accessory minerals trapped as inclusions. An
example of monazite is presented in the following paragraphs. Other examples are given in
Williams et al. (2017, this volume).
72 Lanari & Engi
Textural correlation is conceptually simple (e.g., Kohn 2016) and obviously requires in
situ dating of accessory phases to preserve textural relations. In particular, U–Th–Pb dating of
monazite has become one of the primary tools for constraining the timing of moderate to high-
grade metamorphism (Harrison et al. 2002). Of course, the youngest inclusions only set an upper
age limit to the enclosing garnet growth zone. If an accessory formed just before it got trapped
in a growing porphyroblast, it presumably was isolated from the matrix and thus protected
from later re-equilibration, recrystallization or overgrowth. Such an (idealized) scenario should
generate an age gradient among inclusions from the core of the porphyroblast to its rim.
Examples. Mottram et al. (2015) dated monazite included in garnet from pelitic schist
of the Lesser Himalayan Sequence. Monazite trapped in garnet cores gave (common-Pb and
Th-corrected) 238U / 206Pb ages of 20.7 ± 2.2 Ma (from 4 single-spot analyses), inclusions in the
mantle surrounding the core show ages of 17.9 ± 0.5 Ma (3 analyses), and inclusions in the
rim 15.8 ± 1 Ma (4 analyses). Three distinct stages of garnet growth with different inclusion
patterns are visible in chemical maps. As xenotime was absent in this sample, one might
expect monazite dissolution during garnet growth (Spear and Pyle 2010). The process invoked
by Mottram et al. (2015) assumes that monazite grains present in the matrix underwent
continuous and complete re-equilibration during garnet growth. However, it is not clear by
what process such re-equilibration happened nor whether there merely has been overgrowth.
Inclusions trapped by the growing garnet were shielded, since diffusion of U, Th and Pb in
garnet is slow. Many other studies have reported evidence that garnet porphyroblasts provided
such shielding and found monazite inclusions trapped in garnet porphyroblasts to be older
than in the matrix (Foster et al. 2000, 2004; Martin et al. 2007; Hoisch et al. 2008). Yet, to
derive robust estimates on the time and tempo of garnet growth, careful documentation of
the inclusion location is required, e.g., reporting the distance of the inclusion from the core
(in equatorial sections). In addition, inclusions that seem connected to the matrix by hairline
fractures or microcracks must be avoided, as these may have been subject to interaction with
the matrix or re-equilibration (Montel et al. 2000; Martin et al. 2007). For instance, Martin et
al. (2007) compiled 196 in situ Th–Pb dates of monazite inclusions obtained by LA-ICP-MS
and found that microcracks allowed communication between the interior of the garnet and the
matrix, facilitating dissolution, recrystallization, and intergrowth formation.
Assuming that continuous re-equilibration of matrix monazite does happen, as suggested
by Mottram et al. (2015), is it the rule or the exception? Some studies have indicated
incomplete re-equilibration to interpret the observed age distribution. For instance, monazite
ages in a garnet–biotite schist from Bhagirathi valley in Garhwal Himalaya (Foster et al.
2000) range between 40 and 25 Ma. The age of monazite m1 included in garnet cores is
~39.5 Ma, whereas monazite m2 in garnet rims spread in age between 36 and 41 Ma (Fig. 6).
However, instead of invoking partial re-equilibration, we contend that the data indicate partial
dissolution of m1-monazite, and discrete overgrowth by m2. In the matrix, no m1 and only
one survivor of m2 was found (Fig. 6), suggesting strong dissolution of monazite m1 and m2
before the formation of a new generation m3 at 30–27 Ma. So, we suggest partial replacement
rather than “re-equilibration”. To ascertain whether this (or some alternative) process was
responsible for monazite growth, geochemical tracers for monazite should be analyzed, and
readily interpretable ones are available: Th and Y contents and their zoning (Kohn et al. 2005;
Williams et al. 2017, this volume) are most helpful.
A model of partial replacement is also supported by the comprehensive dataset of spatially
distributed ages reported by Hoisch et al. (2008). Monazite inclusions were dated (SIMS, Th–Pb)
in several garnet grains from upper amphibolite facies pelitic schist in the northern Grouse
Creek Mountains, Utah. Compositional maps exhibit three successive growth zones with distinct
Figure 6. Monazite ages from grains trapped as inclusions in garnet (circles) and grains in the matrix
(squares) of a biotite schist from Bhagirathi valley, Garhwal Himalaya (modified after Foster et al. 2000).
compositions from a core to a mantle and a rim (inset in Fig. 7a). The age of inclusions from
different grains (gm1b-e and gm3h) correlated with chemical zoning in the three zones, as shown
in plots against the fractional garnet volume (Fig. 7). Ages generally decrease toward garnet rims,
and Hoisch et al. (2008) interpreted the monazite ages as decreasing linearly with increasing mode
(volume) of garnet. Kohn (2016) pointed out that this interpretation implies that analytical errors
were underestimated. Certainly the monazite inclusion ages in the three growth zones show scatter
beyond the analytical error (Fig. 7a). An alternative interpretation is to regard each inclusion
age as a maximum age estimate of a single growth stage (Fig. 7b). Thus partial preservation of
older monazite ages in discrete, successive growth zones is visible and suggestive of a process of
incomplete and continuous replacement of monazite, along the lines previously discussed.
Is partial or complete replacement the only process to explain the age trend recorded by
monazite inclusions? Monazite textures give some indications: Grains from a sillimanite-
bearing metapelite from the Hunza Valley in Pakistan studied by Foster et al. (2004) exhibit
complex zoning patterns with four distinct growth zones (Fig. 8), suggesting a succession of
partial resorption and precipitation stages. Monazite m1 grew slightly prior to garnet and was
then partially dissolved during growth of the garnet core (Fig. 8). Monazite m2 grew after the
garnet core and was then captured as inclusions in garnet rims. The enrichment in Y found in
rim garnet was interpreted as the prograde breakdown or consumption of xenotime (Fig. 8).
Because monazite m3 is not observed as inclusions in garnet, it is likely that it grew coevally
with xenotime during the garnet resorption stage. Finally the last monazite m4 grew locally in
the matrix, in equilibrium with xenotime. Some monazite grains in the matrix exhibit the four
growth zones, but with clear resorption features separating them. The preservation of several
growth zones of monazite in this sample is well supported by both textures and chemical zoning
(Fig. 8). Resorption of monazite appears to be correlated to garnet growth and provides some
constraints on timing. Based on the textural relationships (Fig. 8), we can conclude that the
garnet core grew between 80 and 68 Ma and garnet rims between 60 and 58 Ma. A break of
8 Ma occurred between the two stages of growth of garnet (core and rim). This dataset with its
extensive documentation of the textural and chemical relationships also provided a fairly precise
estimate (2.4 ± 1.2 °C / Ma) of the heating rate experienced by the sample (Foster et al. 2004).
74 Lanari & Engi
Figure 7. Ages from monazite inclusions (2σ) trapped in three garnet crystals (gm1b, gm3h, gm1e) from
upper amphibolite facies pelitic schist in the northern Grouse Creek Mountains, Utah plotted against the
fractional volume of garnet (modified after Hoisch et al. 2008). The garnet crystals were divided into three
successive growth zones (see inset); ages are plotted in corresponding colors. (a) Average age and weighted
uncertainty of inclusion ages from the three growth zones. (b) Alternative interpretation of the age distribu-
tion, assuming partial re-equilibration of monazite grains during growth or overgrowth.
Any petrochronological interpretation demands good documentation of both inclusion

texture and chemical zoning. From the selected examples of textural correlation, several main
conclusions can be drawn: (1) Age patterns can be complex because of partial replacement
(for example, by dissolution–reprecipitation, Putnis 2009; Grand’Homme et al. 2016). (2) P–T
modeling based on the concept of FCM is required, incorporating an inverse numerical strategy
to obtain the most likely P–T trajectory (Moynihan and Pattison 2013; Vrijmoed and Hacker
2014; Lanari et al. 2017). This is the price to be paid if we want to obtain robust P–T–t estimates
for successive growth zones. If, in addition, inclusion ages are combined with host (here garnet)
dating, this adds certainty, especially to the significance of the earliest inclusion age.
Figure 8. Monazite petrochronology on inclusions and matrix grains from a sillimanite bearing metapelite
from the Hunza Valley in Pakistan (modified after Foster et al. 2004). The ages are reported together with
schematic volume fractions of garnet, monazite, and xenotime through time and the corresponding tex-
tures. Abbreviations: G, growth; R, resorption, mnz, monazite; xtm, xenotime; grt, garnet. The four mona-
zite zones (m1…m4) correspond to the different generations discussed by Foster et al. (2004).
BULK ROCK COMPOSITION EFFECTS ON MELT PRODUCTION

AND ACCESSORY MINERALS
Many studies have pointed out the strong effect bulk rock composition has on major and
accessory mineral stability and on melt production (e.g., Spear 1993; Stevens et al. 1997;
Pickering-Witter and Johnston 2000; Tinkham et al. 2001; Evans and Bickle 2005; Janots et
al. 2008; Kelsey and Hand 2015; Yakymchuk et al. 2017, this volume; Zack and Kooijman
2017, this volume). To illustrate this effect in a petrochronological framework, we present two
case studies of the regional metamorphic aureole in the Mt Stafford area, central Australia.
A metamorphic field gradient from low-pressure greenschist- to granulite-facies is exposed,
and Greenfield et al. (1996) divided the terrain into five metamorphic zones (Fig. 9a). The
Proterozoic metasedimentary rocks of the Mt Stafford area comprise aluminous metapelites
interbedded with metapsammite layers and cordierite or amphibolite granofels on a centimeter
to meter scale. In the migmatite zones the interbedded metapelite and metapsammite form
bedded migmatites preserving sedimentary features (Greenfield et al. 1996) that suggest in situ
melting (caused by a series of biotite breakdown reactions) without substantial migration of
melt (Vernon and Collins 1988). Bulk rock compositions are constant across the various zones
(Greenfield et al. 1996), confirming the absence of significant melt mobilization. This result
was essential to use the bulk rock composition as representative of the equilibration volume.
Based on this assumption, White et al. (2003) reconstructed the melt production history across
the successive zones (Fig. 9b) using equilibrium phase diagrams computed for the bulk rock
compositions of Greenfield (1997). Metapelite produced more melt at lower temperatures,
whereas metapsammite experienced additional major melt-production at higher temperatures
(zone 5 in Fig. 9). This example shows the control of the bulk rock composition on the melt
production history. It is important to note that this approach would not work in case of large-
scale melt migration that would change the reactive bulk composition.3
3
In such cases a melt extraction step could be added in modeling along a P–T loop. For example, 6 vol%
of melt could be extracted once the proportion reaches the melt connectivity transition at 7 vol%, and a new
reactive bulk composition would be calculated for use in the next steps.
76 Lanari & Engi
Figure 9. Petrochronological investigation of a metamorphic aureole from the Mt Stafford area in central
Australia. (a) Metamorphic zones and mineral modes from Greenfield et al. (1996). (b) Melt fractions for
two metapelite and one metapsammite samples from White et al. (2003). Sample numbers refer to those
with age data for accessory phases (Rubatto et al. 2006). (c) Ages of monazite and zircon. Dates for SGP3
are not considered here because of possible excess 204Pb, as discussed by Rubatto et al. (2006).
Taking the scenario and quantification of the melt history from White et al. (2003), age data
(Rubatto et al. 2006) for these same rocks can be put into context. The metamorphic behavior
of monazite and zircon depends on host-rock composition (Wing et al. 2003; Fitzsimons et
al. 2005; Rubatto 2017, this volume), and metapelites and metapsammites affect inherited
zircon and monazite in different ways. In the case of the Mt Stafford aureole, monazite cores
show a less pronounced Eu anomaly than the rims, and both are interpreted as due to prograde
growth, with the amount of coexisting potassic feldspar increasing (Rubatto et al. 2006). In
the higher-grade samples, monazite rims may have formed later, as they show an increase in
Gd / Lu, which reflects strong fractionation of Lu into garnet cores. Therefore, the monazite
rims most likely grew during a limited period of garnet resorption close to peak metamorphic
conditions. This scenario is supported by the compositional zoning of garnet that exhibits clear
evidence of core resorption (Rubatto et al. 2006). Trace element maps (Th and Y) of the last
monazite growth zone would help in interpreting how the last monazite formed. However, the
solubility of monazite as well as the melt volume increase with temperature (Montel 1993;
Stepanov et al. 2012), one would expect resorption not growth (Kelsey et al. 2008). Chemical
characterization of monazite—ideally based on compositional maps (Mahan et al. 2006;
Williams et al. 2007, 2017, this volume)—is best done prior to chronological microanalysis
(Kohn et al. 2005). Other studies have reported that monazite rims may also form during
cooling from crystallizing melt (Pyle and Spear 2003; Kohn et al. 2004; Johnson et al. 2015).
In migmatites and residual granulites new metamorphic growth of zircon occurs because
the concentration of Zr and light rare earth elements (LREE) increases in the melt during cooling
(Kohn et al. 2015). The corresponding age data thus reflect high-temperature retrogression
(Roberts and Finger 1997; Whitehouse and Platt 2003; Kelsey and Powell 2011; Yakymchuk and
Brown 2014; Kohn et al. 2015), and indeed zircon ages from the Mt Stafford aureole are slightly
younger than monazite ages (Fig. 9c), supporting growth upon cooling and melt crystallization.
In this set of samples, we observe that (1) the ages between monazite (core and rim)
and zircon are slightly different, and (2) a systematic shift occurs between the weighted
mean ages of accessory minerals from higher-grade samples of metapsammitic versus
metapelitic compositions. For instance, monazite ages in the former are slightly older than
those in the latter (Fig. 9), though this shift in age is almost within analytical error. The
apparent age differences are most likely due to migmatites reaching the solidus at different
temperatures for the different local bulk rock compositions (Yakymchuk and Brown 2014).
This example can be used to demonstrate that if chemical and age differences do exist
between two metasedimentary layers with different bulk rock compositions, successive
P–T–t investigations should be made for the distinct layers. The same applies to distinct
domains occurring within a single layer, but in the case of Mt Stafford these have not been
investigated. Implications for domainal rocks are discussed in the next section.
LOCAL REACTIONS AND FORMATION OF DOMAINAL ROCKS

Chemical differentiation processes are essential to the formation of many metamorphic
rocks (e.g., schist, hornfels, banded gneiss, migmatite). While differentiation occurs at various
spatial scales, we focus here on evidence in single rock specimens and on processes that lead
to chemical segregation in these, thus producing compositional domains. Once more, our
goal is essentially to “read rocks”, i.e., to understand samples petrogenetically and quantify
the conditions at which they formed. We showed above that models rooted in chemical
thermodynamics remain powerful, if they are cleverly applied to analyze rocks that have not
experienced strong chemical differentiation (e.g., melt loss) or chemical segregation (zoned
porphyroblasts). In general, complexity levels increase if we consider rocks with textural and
chemical heterogeneity—both commonly evident in interesting samples. Clearly such rocks
did not reach anything like rock-wide chemical equilibrium, but documenting local domains
allows us to investigate them and paves the way to understand their petrogenesis.
In this section we examine further evidence for and consequences of domain formation
from various rock types. Touching briefly on isolated segregations, such as corona structures
or pseudomorphs after porphyroblasts, we mostly concentrate on analyzing spatially more
extensive and organized domains, such as typically occur in migmatites.
78 Lanari & Engi
Evidence of local reactions in discrete textural domains

Domain formation requires an initial step of physical segregation of material (DeVore
1955) to create spatial heterogeneity in composition. Such initial heterogeneity may form, for
example, by mass transfer during a period of interaction with a reactive fluid (e.g., Beinlich
et al. 2010) or by metamorphic differentiation (Fletcher 1977; Foster 1981, 1999), enhanced
by solid-state transformations (e.g., Brouwer and Engi 2005; Lanari et al. 2013), deformation
(e.g., Mahan et al. 2006; Goncalves et al. 2012; López-Carmona et al. 2014), or partial melting
(Milord et al. 2001; Kriegsman and Nyström 2003).
As shown above, prograde growth of garnet porphyroblasts generates strong local
heterogeneity in the rock composition, and this effect is not limited to garnet; indeed it may
be even more pronounced for porphyroblastic lawsonite or kyanite if they are later involved
as reactants in replacement reactions. Such reactions may respond to these unusually Ca- and
Al-rich domains, which are commonly identified because they preserve the original shape of
the porphyroblast (as pseudomorphs), and they generate new mineral assemblages very distinct
from the rest of the rock matrix (Carmichael 1969; Selverstone and Spear 1985; Foster 1986;
Carlson and Johnson 1991; Elvevold and Gilotti 2000; Ballevre et al. 2003; Brouwer and Engi
2005; Zhang et al. 2009; Verdecchia et al. 2013; López-Carmona et al. 2014). In some cases,
such domains can be modeled as a closed system. For example, Brouwer and Engi (2005)
combined backscatter electron images with electron microprobe spot analyses to obtain local
bulk compositions of four domains in retrogressed kyanite-bearing eclogite from the Central
Swiss Alps. The models account for the development of plagioclase symplectites involving
very Al-rich phases like corundum, hercynite, and staurolite, which are not normally expected
in rocks of basaltic bulk composition. This technique can be generalized to any textural domain
for which the local bulk compositions are calculated from mineral compositions and estimated
mineral modes (Tóth et al. 2000; Korhonen and Stout 2005; Mahan et al. 2006; Riel and Lanari
2015; Cenki-Tok et al. 2016; Guevara and Caddick 2016) assuming limited chemical interaction
among the domains (see below). In delimiting domains, a certain amount of judgment is needed,
and some assumptions are implied (Elvevold and Gilotti 2000; Kelsey and Hand 2015).
Fluid influx can cause metamorphic differentiation as well, and it may lead to significant
changes in the local reactive composition (Putnis and John 2010). Using an example from the
subduction complex of the Tianshan mountains (China), Beinlich et al. (2010) reported the
transitional conversion from blueschist to eclogite accompanied by a change in composition (from
Ca-poor to Ca-rich) along a profile sampled perpendicular to a vein (10–15 cm thick). Equilibrium
phase diagrams computed for the two distinct bulk rock compositions indicate that both mineral
assemblages were stable at the same peak P–T conditions (21 ± 1.5 kbar and 510 ± 30 °C). Here
the fluid affected the local composition and probably played a significant role to achieve chemical
equilibrium. However, it has long been debated (e.g., Ridley 1984) why blueschists and (low-
temperature) eclogites coexist in some areas; differences in bulk composition (Brovarone et al.
2011) or in P–T conditions (Davis and Whitney 2006) may be responsible.
The presence of textural domains and compositional differences among these may or may not
imply chemical potential gradients. If chemical exchange between domains was very inefficient,
each domain may be regarded as essentially a closed system, and thermodynamic modeling
poses no problem. If transport was very limited for only a few chemical components (e.g., of
trace elements), but efficient for the others, then chemical potential gradients between domains
may have been essentially zero for most components, i.e., (partial) equilibrium was maintained
between the domains. However, wherever local mineral assemblages within textural domains
indicate substantial chemical potential gradients between them, transport (of these components)
between the domains evidently happened, but at rates too low to attain overall equilibrium.
Chemical potential gradients and element transfer between domains

If two systems A and B are in thermal equilibrium and are open with regard to a mobile
component i, then µi must be the same in both systems, otherwise there is a tendency for transfer
of i between the two systems. When such a distribution has been attained, the free energy of
the total system is at minimum with regard to the distribution of component i (Thompson
1955). This model of mobile component has been termed selective chemical interaction. An
example of this principle concerns migmatites, which are domainal rocks with a leucosome
and a melt-depleted melanosome (here considered a residue). The segregation of solids and
melt and the subsequent evolution of their mineral assemblages can be modeled using T–X
residue–melt equilibrium phase diagrams (White et al. 2001), these are binary diagrams
displaying the evolution of the phase assemblage for reactive bulk composition lying between
a residue (melanosome or restite) and a melt-rich domain (leucosome). Of interest here is
the role that diffusion of H2O can play between leucosome and residue in the crystallization
history of (segregated) melt in contact with the residue. If the physical separation of melt from
its residue is purely mechanical and happens at chemical equilibrium, there are no gradients in
chemical potential. However, subsequent cooling results in chemical re-equilibration of now
separate domains of different reactive bulk compositions. Now chemical potential gradients
will be established, in particular µ(H2O) will rise upon crystallization of the melt-rich domain.
Where in contact with restite, equilibration between the two domains requires diffusion of
H2O from the leucosome to the melanosome. White and Powell (2010) used this model of
selective chemical interaction to show that the diffusive interaction of H2O between residue
and segregated melt—aiming to equalize µ(H2O)—promotes the crystallization of anhydrous
quartzofeldspathic products in the leucosome and hydration of the residue. Further studies
are warranted to identify evidence of interaction for other elements between the two domains,
possibly with different length scales. It is crucial to understand such scales if equilibrium
thermodynamics (based on local bulk composition, see below) may be applied or not.
Chemical potential gradient within domains
Frozen-in chemical potential gradients can also be observed within single domains provided
that diffusion-controlled structures such as coronae are preserved. An interesting approach
based on equilibrium thermodynamics and quantitative chemical potential diagrams has been
proposed by White et al. (2008) who reconstruct the chemical potential gradients preserved
in the final corona structure. The same approach was applied by Schorn and Diener (2016)
to investigate coronae developed at the expense of magmatic plagioclase and orthopyroxene
during the gabbro-to-eclogite transformation. So far, such models are restricted to simple
systems with some general assumptions on components in excess or assumed to be immobile.
Size of the equilibrium volume versus scale of the model
As discussed earlier, the size of the equilibrium volume is controlled by the diffusion
rate of the elements in the intergranular medium. Knowing diffusion rates is essential as they
may set a spatial limit within which rocks can maintain global chemical equilibrium as they
evolve. Because experimental data are sparse or may not apply, few reliable estimates on
rates of intergranular diffusion are as yet available; they have instead been extracted from
natural examples (Carlson 2002 and references therein). Because it is difficult to constrain the
duration of intergranular diffusion tightly, these rates are not very well known, but Carlson
(2010) succeeded in estimating the rate of intergranular diffusion of Al and its dependence
on temperature for H2O-saturated and fluid-undersaturated media (Fig. 10). Data for Al
diffusion are essential because many studies found the diffusive transport of Al to exercise the
dominant control on overall rates of reaction for aluminosilicates (Carmichael 1969; Foster
1977, 1981, 1983; Carlson 1989, 2002, 2010; White et al. 2008). To the extent that this may
apply, the diagram in Figure 10 sets limits on the length scales over which reactions and
80 Lanari & Engi
Figure 10. Characteristic intergranular diffu-

sion length scale for Al in a medium saturated
in hydrous fluid (blue curves) and undersatu-
rated in fluid (black dashed curves); modified
from Carlson (2010).
chemical equilibration can be expected for a range of temperature and typical duration of
metamorphic transformations. For example, the data indicate that at upper amphibolite facies
conditions (say 600–700 °C), domains 0.5 mm apart will require 1 Ma to equilibrate if grain
boundaries are not fluid-saturated, whereas domains 5 mm apart should equilibrate in 100 ka
when saturated with hydrous fluid. But do fluids ever persist that long?
In any case, this sort of analysis may serve to indicate whether we may rely on chemical
equilibrium models to infer petrogenetic conditions from domainal rocks. Where valid, it
is useful to analyze P–X or T–X phase diagrams, calculated for a range in local bulk rock
compositions (X1…X2), corresponding to different domains. Individual textural domains can
be analyzed to determine the local bulk compositions (Tóth et al. 2000; Brouwer and Engi
2005; Riel and Lanari 2015; Cenki-Tok et al. 2016) and then the equilibrium assumption can
be tested by comparing predictions from forward thermodynamic models against observed
phase relations. To support such comparisons, we propose an analytical method to quantify
variations in the local bulk composition based on standardized X-ray maps.
QUANTITATIVE MAPPING OF THE LOCAL BULK COMPOSITION

AS A BASIS FOR MODELING
It is possible to correct for the effects of chemical fractionation and chemical
differentiation described in the previous sections using quantitative compositional maps
(expressed in oxide wt%) and to extract local compositions from such a map. The first
application (to our knowledge) that used quantitative compositional maps to obtain the
local bulk composition was published by Marmo et al. (2002) who focused on two eclogite
samples showing zoned garnet porphyroblasts. The authors examined the effect of garnet
core and mantle fractionation on the bulk rock composition using compositional maps with
a standardization based on the semi-empirical Bence and Albee (1968) matrix correction
algorithm (Clarke et al. 2001). Major progress since then involved a density correction,
clarifications on the choice of domain boundaries, and a correct extrapolation from 2D to
3D. This last point is tricky; for example it is easy to overestimate the contribution of garnet
cores, as surface fractions need to be properly converted to volume fractions, but equally
easy to underestimate cores—simply by mapping a grain sectioned off-center.
The influence of various assumptions on the local bulk composition estimates is shown
below, but an adequate mapping strategy is definitely required to produce high-quality
standardized maps.
Quantitative X-ray mapping
Since the first X-ray “spot maps” measured using an electron probe micro-analyzer
(Cosslett and Duncumb 1956), both the instruments and techniques have been greatly improved.
Although quantitative methods (Kohn and Spear 2000; Clarke et al. 2001; De Andrade et al.
2006) and computer programs (Tinkham and Ghent 2005b; Lanari et al. 2014c) are available
to obtain oxide wt% maps, many studies still combine uncorrected ‘semi-quantitative’ X-ray
maps with profiles of high-precision point analyses.
A quantitative method requires a correction called ‘analytical standardization’ that
transforms the number of collected photons (i.e., X-ray intensity) using either a semi-
empirical Bence and Albee (1968) matrix correction (Clarke et al. 2001) or high-resolution
spot analyses as internal standard (De Andrade et al. 2006). The technique of internal
standardization provides accurate compositional maps and has recently been integrated into
the software XMapTools (Lanari et al. 2014c). Quantitative compositional mapping allows
measuring the natural variability of the mineral phases at the scale of a thin section and
thus is useful in petrogenetic analysis (e.g., Kohn and Spear 2000). Compositional mapping
has been successfully combined with multi-equilibrium thermobarometry (Vidal et al. 2006;
Yamato et al. 2007; Fiannacca et al. 2012; Lanari et al. 2012, 2013, 2014a,b; Pourteau et al.
2013; Grosch et al. 2014; Loury et al. 2015, 2016; Trincal et al. 2015; Scheffer et al. 2016)
and forward thermodynamic modeling (Abu-Alam et al. 2014; Lanari et al. 2017) or may be
used to extract mineral modes (Cossio et al. 2002; Martin et al. 2013) and local compositions
(Centrella et al. 2015; Riel and Lanari 2015; Mészaros et al. 2016).
Strategy to derive local bulk composition from X-ray maps using XMapTools
Local bulk composition can be easily estimated based on quantitative maps, for which the
composition at each pixel is expressed in oxide wt%. The analytical procedure is detailed in
Appendix 1. The local bulk composition of any desired part of the section mapped is spatially
integrated, using the (oxide wt%) map and applying a density correction for each mineral phase
identified. Density values are estimated using Theriak. This procedure is coded as a function
available in XMapTools 2.3.1 and allows extracting the composition (in oxide wt%) of any
domain directly from the compositional maps by polygonal boundaries selected by the user.
To illustrate the simplicity of this procedure, we selected an orthogneiss from the
Glacier–Rafray klippe in the Western Italian Alps (Burn 2016). X-ray maps were acquired
using a beam current of 100 nA and dwell time of 70 ms. These maps were standardized
using XMapTools 2.3.1 and the method described in Lanari et al. (2014c). The map size
is 10.24 × 10.24 mm (corresponding to 1024 × 1024 pixels with a pixel size of 10 µm; the
total measurement time is ~42h). This orthogneiss contains 38 vol% quartz, 33% albite, 21%
phengite, 4% K-feldspar, 2% epidote, and 2% actinolite. The local bulk composition LBC1
was extracted from a square domain of ~100 mm2 (dashed line in Fig. 11). To estimate the
uncertainty in the composition resulting from the domain selection, a sensitivity test was
performed using a Monte-Carlo simulation that randomly changed the position of the corners
and thus the shape of the selected domain (Fig. 11). In this example 100 permutations were
used with a displacement of each corner of ±10 pixels (1σ assuming a Gaussian distribution).
The composition of LBC1 is shown in Table 2 with the associated standard deviations and its
relative uncertainty. The sensitivity test shows that there is very little variation (<1% for all
the elements) in the composition due to this local domain selection.
82 Lanari & Engi
Figure 11. Estimation of local bulk composition for an orthogneiss from the Western Alps. Variably sized
areas were considered, one delimitation shown by dashed black outline, gray lines in inset show variations
used in sensitivity test performed by Monte-Carlo simulation (see text). Analysis based on XMapTools; the
composite compositional map shown is color-coded in an RGB image displaying variations in SiO2–Na2O–
FeO, but any combination of oxides or element ratios could be chosen.
Table 2. Local bulk composition LBC1 of an orthogneiss from the Glacier-Rafray Klipee in the
Western Alps (see text) calculated using XMapTools (see Fig. 11). Abbreviations: Stdev. Standard
deviation; Unc. Relative uncertainty.
Mean Stdev. (2σ) Unc. (%)

SiO2 75.420 0.057 0.076
Al2O3 13.230 0.033 0.249
FeO 1.230 0.006 0.488
MgO 0.940 0.004 0.426
CaO 0.810 0.008 0.942
Na2O 3.360 0.018 0.522
K2O 3.090 0.017 0.551
Total 98.08
Gibbs free energy minimization for the local bulk composition

A similar approach is used to propagate this relative uncertainty through the forward
thermodynamic models. Computations were made using program Bingo-Antidote designed to
estimate the optimal P–T conditions using equilibrium models (Lanari and Duesterhoeft 2016).
The thermodynamic database JUN92.bs (Berman 1988 and subsequent updates) was selected
for this test in the system SiO2–Al2O3–FeO–Fe2O3–MgO–CaO–Na2O–K2O–H2O. Excess
oxygen (0.09 mol%) in the bulk composition was added to stabilize the observed amount of
epidote. Negligible amounts of hematite (0.3 vol%) are predicted by the model (and ignored
in the following). Mineral modes and compositions extracted from the domain LBC1 (Fig. 11)
were modeled at 475 °C and 12.6 kbar (Fig. 12a), the estimated conditions of equilibration of
the assemblage quartz + albite + phengite + K-feldspar + epidote. The dispersion (±2σ) of the
mineral modes predicted by the equilibrium model was estimated from 2000 permutations and
is less than 1.6% (relative) for quartz, albite, phengite, and epidote, and 2.4% for K-feldspar
(Fig. 12b). The corresponding information can also be extracted for mineral compositions. For
example the model phengite composition is Si4+ = 3.481 ± 0.002 atoms per formula unit (2σ) and
XMg = 0.874 ± 0.008. This technique allows propagating the relative uncertainty in composition
through the equilibrium models. As mentioned previously, few such sensitivity analyses have
been reported (Kelsey and Hand 2015), and thus modeled modes have rarely been compared
to observed mineral phases that contribute to the bulk composition (Warren and Waters 2006).
In the present example, an arbitrary ~10 × 10 mm2 domain was mapped and used to estimate
the local bulk composition that was then taken as a reactive bulk composition for modeling.
The overall quality of the model was excellent, which would seem to support the choice of
this domain as an equilibrium domain. But is such a comparison sufficient to conclude that
equilibrium was established at this spatial scale for the major elements? The answer is probably
yes—at least for the elements of interest. However, for petrochronological studies the scale
of equilibration for trace elements would be particularly relevant to verify whether slowly
diffusing species attained uniform concentrations via an intergranular medium. To pursue this
topic, the approach should be extended to use LA-ICP-MS quantitative maps. In this case a
FCM would be needed as the accessory phases almost certainly act like porphyroblasts and
trace constituents would not be part of the equilibration volume or reactive bulk composition.
Advantages of the micro-mapping approach
The approach described above can be used to test several textural domains from the same
map dataset. Using Bingo-Antidote, it is thus possible to select a polygon and directly calculate
the corresponding model assemblage, modes, and compositions for any P–T conditions.
84 Lanari & Engi
Figure 12. Gibbs free energy minimization using local bulk compositions (based on Fig. 11) and the
program Bingo-Antidote. (a) Comparison between observed (left) and modeled (right) mineral modes. (b)
Results of a sensitivity test (2000 permutations) performed to evaluate the effect of the domain selection on
predicted mineral modes. Relative uncertainty in mode reported at 2σ level.
This strategy provides appropriate remedies to refine the estimation of the composition
of the reactive part of a rock. For instance, most accessory phases are generally ignored in
equilibrium phase diagrams, and this can lead to an erroneous bulk rock composition. Apatite is a
good example because it contains > 50 wt% CaO, and if a simplified system does not incorporate
phosphorous, the amount of CaO from the bulk rock composition available for the other phases
should be corrected for the CaO stored in apatite. However, correcting CaO based on P2O5
contents would introduce error if much monazite is present in the sample. This problem is easily
solved using the micro-mapping approach: Apatite pixels can simply be ignored (a software
option), and thus the CaO value of the local bulk composition is directly suitable for modeling. A
similar approach can be used if any mineral present is considered non-reactive.
Potential artifacts affecting the local bulk composition estimates
To illustrate potential effects of some artifacts on the local bulk composition estimate and the
equilibrium models, we use compositional maps of a mafic eclogite boudin from the Atbashi Range
in the Kyrgyz South Tien Shan (Loury et al. 2015). The sample contains garnet porphyroblasts
(with quartz and omphacite inclusions) in a matrix of omphacite, rutile, and ilmenite.
Geometric effects. Geometric effects arise as a consequence of sectioning a 3D texture
and are most consequential if the surface fractions of successive growth zones of crystals
cannot be linearly correlated with their volume fractions. For example, garnet typically has
a dodecahedral crystal habit that can be approximated by a spherical geometry (Fig. 13b).
If such a garnet grain is cut near its crystal center (see for example Fig. 1), any bulk crystal
composition extracted from 2D compositional maps will overestimate the contribution of the
core at the expense of the rim. If the cut is not equatorial, the section visible may not be
representative of the zoning at all. The effect of such geometric bias is explored by comparing
the composition of a single garnet grain extracted by first averaging from the 2D-map (Fig. 13a)
or by converting the composition of each 2D-annulus to a 3D-hollow sphere (Fig. 13b) and
then integrating these. This effect is generally neglected for local bulk composition estimates
(Marmo et al. 2002) but for spherical domains it is judicious to use a 3D extrapolation
(Mészaros et al. 2016). In the case of garnet from the mafic eclogite of the Atbashi Range,
MnO is enriched in garnet core, and the average MnO composition of garnet is overestimated
by 36% if integration is based on the 2D surface of a single grain (Fig. 13a). Similar deviations
are observed for other elements (MgO −29%; and FeO +7%). For a map containing many
grains, such geometric effects are partially compensated by stochastic sampling of different
sections (Fig. 13c). In our example, integrating all of the garnet composition in 2D cuts the
discrepancy in half, but it remains + 16% in MnO, −17% in MgO, and + 4% in FeO. It is also
possible to evaluate the average garnet composition using a single crystal that is cut near the
center (to be established with a diagnostic element such as Mn) and the spherical correction.
Then this average garnet composition can be mixed with a complementary fraction of matrix
for which an average composition is extracted from the maps.
Chemical equilibrium and the arbitrary choice of domains. Choosing or delimiting
an appropriate local domain is obviously problem-dependent and usually non-trivial. For
equilibrium thermodynamics to be applicable, the spatial domain selected for modelling
should be no larger than the equilibration volume. (Of course any smaller domain might
be selected). The problem is that the size limit is not known. Modellers often just hope
that the domains are uniform in composition, i.e., that the mobility of the (slowest)
components considered in their model was sufficient to equilibrate compositions over the
chosen domain size. Generally, decisions should be based on (1) textural criteria such as
the apparent homogeneity and representativeness of the domain compared to the entire
specimen, or (2) independent transport criteria such as the maximum length scale of
chemical equilibration predicted by the diffusion rate of Al (see Fig. 10).
86 Lanari & Engi
Figure 13. Average composition of garnet single grain estimated using XMapTools. (a) Method 1: 2D
averaging of all the garnet pixels of a single crystal. (b) Method 2: 3D extrapolation using seven ellipsoids
(details of procedure available in user guide of program, http://www.xmaptools.com). The 2D technique
overestimates MnO (absolute: 0.16 wt%; relative difference (∆) = 36%) and FeO (1.9 wt%; ∆ 7%), under-
estimation of Cao (0.13 wt%; ∆ 2%) and MgO (1.35 wt%; ∆ 29%). (c) Method 3: 2D averaging of garnet
pixels in all grains results in overestimation of MnO (0.07 wt%; ∆ 16%) and FeO (1.0 wt%; ∆ 4%) and
underestimation of Cao (0.23 wt%; ∆ 3%) and MgO (0.8 wt%; ∆ 17%).
For a Kyrgyz mafic eclogite from Loury et al. (2015), the predicted length scale of intergranular
diffusion of Al in 1 Ma, assuming a hydrous fluid present, is about 2 mm (Fig. 10). As garnet in
this case crystallized at the expense of chlorite and lawsonite, it is likely that the intergranular
medium was saturated in hydrous fluid, though perhaps only episodically, not for the entire 1 Ma.
It is interesting to note that this spatial scale corresponds to the minimum distance between the
centers of neighboring porphyroblasts observed in the section (Fig. 14). However the grains show
remarkably regular chemical zoning in Ca and XMg (Mg2+ / (Mg2+ + Fe2+) at the thin section scale
(Fig. 14), suggesting that chemical equilibrium was maintained at the periphery of growing phases
at all times (equilibrium control growth model assuming grain boundary equilibrium), which
implies a much larger scale for chemical equilibration of the other major elements.
Figure 14. Compositional maps of garnet grains of a mafic eclogite boudin from the Atbashi Range in
Kyrgyz South Tien Shan (Loury et al. 2015). Color code is for structural formula (apfu: atoms per formula
unit); (a) Ca and (b) XMg.
Another strategy to ensure that the local bulk composition estimate is robust is to check
the sensitivity of the composition and model predictions on the polygon selection (Fig. 15).
The compositional map of the Kyrgyz mafic eclogite was sampled using a floating rectangular
window, the size of which was increased from 6.8 × 6.8 mm2 to 20.5 × 12 mm2. The local
bulk composition was successively calculated after each increment, and results are shown in
Figure 15b. This experiment shows that the local bulk composition can be very sensitive to
the area selected. The fluctuations are visible in the variable proportion of garnet (from 32 to
45 vol%, Fig. 15). Once some 80% of the area is integrated, the composition no longer changes
significantly (< 5% for all elements), as shown in the sensitivity tests.
Seeing such variations in the local bulk compositions (Fig. 15b), the critical question
is: how sensitive are model predictions and P–T estimates? To appreciate the effect of
this, the compositions of the four steps used in the above test (i.e., 18%, 50%, 80% and
100% of the area, Fig. 15) were used at 550 °C and 25 kbar, with Bingo-Antidote and the
thermodynamic dataset tc55.txt (Holland and Powell 1998) in the system SiO2–TiO2–Al2O3–
FeO–Fe2O3–MnO–MgO–CaO–Na2O; the oxygen fugacity was controlled by the QFM buffer.
The modeled mineral modes are in line with the mineral modes observed in the four areas
(Fig. 15c), indicating that the predicted modes are correct. By contrast, the modeled garnet
compositions are slightly different for the four models, with deviations of 0.03 in almandine
88 Lanari & Engi
Figure 15. Estimated local bulk composition of a mafic eclogite boudin from the Atbashi Range in the
Kyrgyz South Tien Shan (Loury et al. 2015). (a,b) The sensitivity test is based on different areas of a thin
section (from 18% to 100% of the map surface). Note variation in predicted garnet mode. Compositional
map shown for CaO (wt%); small inset: Grt: garnet, Omp: omphacite, Qz: quartz, Rt: rutile. (c) Modeled
and observed mineral modes for the four cases shown in (a): 18%, 50%, 80% and 100% modeled at 550 °C
and 25 kbar (see text).
and pyrope contents (Fig. 15c). To explore the bias of such differences to P–T estimates,
the best isopleth intersection of a reference garnet core composition (Grtref corresponding
to garnet predicted stable for the entire map composition; Alm67Grs15Prp17Sps1; case 4 in
Fig. 15) was calculated for successive local bulk rock compositions using GrtMod (Fig. 16).
Note that this example is used essentially to illustrate the concept: only one composition
corresponding to a hypothetical garnet core is modeled, without expanding the case to FCM.
Figure 16. P–T conditions of Grtref

(see text) for the four bulk composi-
tions extracted from the areas shown
in Figure 14 (Cases 1, 2, 3 and 4) com-
puted with GrtMod. The errors bars
show the relative uncertainty of the
equilibrium model (i.e., isopleth posi-
tion) resulting from typical uncertain-
ties in garnet composition.
Grtref is predicted stable at lower temperature using the local composition of a small area
(480 ± 25 °C) compared to larger areas (550 ± 18 °C). The errors bars reported in Figure 14
indicate the relative uncertainty of the equilibrium model (i.e., the isopleth position) resulting
from typical uncertainties in garnet composition. The results presented here show that the
differences in local bulk composition, depending on the domain size integrated, definitely can
affect the garnet isopleths and hence P–T estimates. While somewhat technical and problem-
dependent, this selection has significant consequences on P–T estimation.
The illustration example of the Kyrgyz mafic eclogite was selected because it contains several
limits that have been intensively discussed in this review. First, it involves garnet porphyroblasts
that are compositionally zoned, and an optimized strategy would require a FCM to obtain the
P–T segment along which garnet grew. Second, the shape of garnet porphyroblasts requires the
use of 3D extrapolation to avoid geometric effects on the local bulk composition estimates.
If a domain is assumed to be in chemical equilibrium at given P–T conditions, any smaller
subdomain should yield a similar P–T estimate. Figure 16 shows substantial P–T differences
for the domains shown in Figure 15, because garnet compositional zoning (disequilibrium)
has a strong effect on the local bulk compositions. Apart from this, it is important to note that
differences among thermodynamic data sets (standard state properties and solution models;
not tested here) may also affect the P–T predictions.
Toward systematic quantitative trace element micro-mapping to address
petrochronological problems
Quantitative mapping of diagnostic trace elements is crucial for many petrochronological
applications. Over the past two decades, EMPA has been quite effectively used to produce such
maps (e.g., Spear and Kohn 1996; Chernoff and Carlson 1999), but recent improvements in
LA-ICP-MS analysis and software support (Paul et al. 2012; Rittner and Müller 2012) have
fostered applications using quantitative maps of trace elements and isotopes (Becker et al. 2007;
Woodhead et al. 2007; Stowell et al. 2010; Duval et al. 2011; Netting et al. 2011; Šelih and van
Elteren 2011; Peng et al. 2012; Paul et al. 2014; Gatewood et al. 2015; Ubide et al. 2015).
90 Lanari & Engi
For instance a garnet from the Peaked Hill shear zone, Reynolds Range, central
Australia, has been mapped by this technique (Raimondo et al. 2017), with successive raster
images of the focused laser beam stitched together to form a 2D representation of the trace
element distribution. Quantification was initially performed using Iolite (Woodhead et al. 2007;
Paton et al. 2011), and subsequent image processing and manipulation by XMapTools (Lanari
et al. 2014c). The maps reveal significant decoupling between the major and trace element
zoning patterns, with smooth radial zoning in Fe, Mg, Ca and Mn juxtaposed against a discrete
annular structure and successive satellite peaks for REEs and Cr (Fig. 17). Importantly, the
growth and dissolution of accessory phases can be directly linked to garnet evolution through
the superposition of sharp satellite peaks in Zr and HREE (zircon) and P, Th and LREEs
(monazite). Such micron-scale features are clearly resolved by trace element mapping, but
easy to miss with more coarsely spaced spot analyses. Thus the increased spatial resolution
and mass range of LA-ICP-MS mapping offer a powerful means to place garnet growth in a
specific paragenetic context and integrate it with temporal constraints provided by accessory
phase geochronometers or direct Lu–Hf / Sm–Nd dating.
CONCLUSIONS AND PERSPECTIVES

The examples presented in this review indicate that care is needed when modeling a rock
with evident compositional heterogeneity or textural domains. Textural evidence in rocks—
notably mineral zoning—shows that rocks adapt to changes in conditions during their evolution,
but that equilibration is limited by kinetics of transport. Nonetheless, equilibrium models—and
especially thermodynamic forward models—remain powerful for petrogenetic analysis and
allow the retrieval of P–T conditions, provided that partial and local assemblages are analyzed.
Zoned porphyroblasts can significantly fractionate the reactive bulk composition, and this
effect must be included in setting up FCM’s. In the case of garnet, it has been shown that
ECM’s significantly overestimate the predicted modes and affect the P–T estimates made by
isopleths thermobarometry. Dating porphyroblast growth remains attractive, based on textural
correlation and in situ dating techniques. Correlation efforts to link ages with porphyroblast
growth (or resorption) conditions can profit from quantitative element mapping of both
major and trace elements. Documenting how concentrations in REE, Y, Th, and U evolve
in successive growth zones of the host and in the included accessories should be useful for
testing whether or not host and inclusions are in exchange equilibrium (i.e., may or may not
be regarded as coeval). More fundamentally, such combined analysis promises much needed
insight into the mobility of trace elements relative to major elements.
A bulk rock composition determined by XRF analysis of an entire sample can be very
far from the composition of a local assemblage, hence thermobarometry based on isopleths
will be flawed, even observed phase relations may compare poorly with predictions. Instead,
the reactive bulk composition must be approximated by taking account the textural evidence
and preserved local mineral compositions. X-ray maps yield a solid basis to estimate the local
bulk composition, if the spatial domain to be modeled is carefully delimited in thin section.
Predictions from local equilibrium models must be critically evaluated, checking the textural
and mineralogical record preserved in a sample against the computed phase relations for it.
Using a program such as Bingo-Antidote also allows the sensitivity of predictions to the chosen
domain boundaries to be tested. In our experience the local composition needs to be iteratively
refined, or it may turn out that different chemical subsystems should be chosen to analyze
different parts of a rock’s evolution. The goal must be to gain a fairly detailed understanding of
the petrogenetic stages visible in a sample, which ensures that P–T conditions estimated using
isopleths are robust. Such a basis allows a meaningful strategy to select datable (sub)grains for
in situ analysis, which we consider critical to trustworthy petrochronology.
Figure 17. Minor and trace element compositional maps of a garnet grain from the Peaked Hill shear zone,
Reynolds Range (Raimondo et al. 2017), central Australia. (a) Mn in wt%; (b) Y in ppm; (c) Cr in ppm; (d)
Composite RGB image; (d) Chondrite-normalized REE patterns of garnet along the profile drawn in a, b,
c and d (one curve in (e) per pixel). Concentrations given are for the element of interest, isotope numbers
show which isotope was used to monitor element abundance.
Some elements are far less mobile than others, and gradients in concentration (and thus
in chemical potential) may be preserved that render equilibrium modeling inappropriate,
i.e., relatively immobile species (including some trace elements) may have to be excluded.
92 Lanari & Engi
At present, there is limited knowledge regarding the mobility of many trace elements, notably
for actinides, Pb and REE that are critical to chronology. It remains a challenge to develop
approaches that integrate local equilibrium models for domains in which these elements show
evidence of limited transport. To improve our understanding, concentration gradients, both
within and among domains, should be documented for elements with very different transport
properties (ionic charge and size of species dominant for transport). Such maps provide integrals
over the relative mobility and may allow us to distinguish chemical subsystems for which local
equilibrium modeling is reliable vs. those that are sensitive to kinetics. To reconstruct local
chemical potential gradients, it would be intriguing to invert the model, i.e., the compositional
gradient that can be observed between two textural domains. Extending earlier studies (Carlson
2002), the characterization of the frozen-in gradients will help quantify extents of disequilibrium.
ACKNOWLEDGMENTS
We thank M. Burn, F. Giuntoli, F. Guillot, C. Loury and T. Raimondo for providing sample
information and compositional maps used in Figures 2 (MB, FGi, FGu), 11 (MB), 13–14–15
(CL), 17 (TR). We acknowledge stimulating discussions with R. Berman, C. de Capitani, E.
Duesterhoeft, J. Hermann and D. Rubatto. Reviews from H. Stowell and G. Clarke, as well as
helpful comments from M. Kohn that led us to improve this paper are gratefully acknowledged,
as is Matt’s editorial handling. The Swiss National Science Foundation (Project 200020-146175)
and the Faculty of Science of the University of Bern have supported work presented here.
APPENDIX 1—LOCAL BULK COMPOSITIONS FROM OXIDE WEIGHT

PERCENTAGE MAPS
Let us consider a domain of rock composed of three mineral phases Min1, Min2 and
Min3, each homogeneous in composition C1i , C2i and C3i of the oxides of the element i.
C is expressed in oxide wt%. This is convenient here because chemical analyses of silicate
minerals are commonly reported in wt% of the oxides determined. The local bulk composition
of this domain CLB can be calculated as:
CLB = w1C1i + w2C2i + w3C3i (A1)
With w1, w2 and w3 the mass fractions of the mineral phases Min1, Min2 and Min3. This
relation can be generalized for a map of a given domain containing n pixels:
n
CLB = ∑w jC ij (A2)
j =1
wj and C ij are the mass fraction and composition in oxide weight percentage of pixel j. The
use of Relation (A2) is not straightforward, as it requires the knowledge of the mass fraction of
every pixel that may belong to different phases, each with a different molar mass.
On the other hand, the pixel fraction of a phase k is a good approximation of the surface
covered by this phase and can be extrapolated to a volume fraction. To a first approximation,
it is often assumed that:
vk = s k (A3)
In metamorphic petrology, this relation may be reasonable if (i) the sample was sectioned
perpendicular to the foliation or schistosity, (ii) the compositional map is acquired on an
unaltered rock surface devoid of local compositional heterogeneities, (iii) the size of the map is
sufficient to ensure good sampling, (iv) the resolution of the map is high enough to avoid issues
with the smaller grain size population, and (v) 3D effects are negligible. Possible pitfalls and
issues are discussed under ‘Potential artifacts affecting the local bulk composition estimates’.
The relationship between the mass fraction and the volume fraction is:
ρk
wk = vk (A4)
ρmixture
with rk being the density of the phase k and the average density rmixture of the domain.
Integrating the density correction in Equation (A2) leads to a more convenient expression of
the local bulk composition of the domain:
n
ρk
CLB = ∑ v jC ij (A5)
ρ
j =1 mixture
From this relationship it is possible to extract the local bulk composition of a domain using the
average density of every phase involved.
In case of multi-phase assemblages the density correction is required to predict accurate
local bulk compositions. For example, most of the studies discussed in this review did not
correct for density differences between the considered minerals (M. Tóth et al. 2000; Marmo et
al. 2002; Brouwer and Engi 2005; Cenki-Tok et al. 2016), and this can lead to large discrepancies
for elements sequestered by dense mineral phases (e.g., Fe in magnetite or garnet).
Bulk rock compositions can also be extracted from oxide weight percentage maps. In this
case, a domain with as many grains as practical should be selected. Based on the two examples
discussed in this paper, we suggest that the width of the selected surface must be at least
8–10 times the ‘average’ size of the largest grains. This criterion (i.e., > 65 times the ‘average’
grain surface) ensures that the composition is fairly representative of the entire composition of
the domain, as long as the section is uniform in mineral assemblage.
REFERENCES
Abu-Alam TS, Hassan M, Stüwe K, Meyer SE, Passchier CW (2014) Multistage tectonism and
metamorphism during Gondwana collision: Baladiyah Complex, Saudi Arabia. J Petrol 55:1941–1964,
doi:10.1093 / petrology / egu046
Ague JJ, Axler JA (2016) Interface coupled dissolution–reprecipitation in garnet from subducted granulites and
ultrahigh-pressure rocks revealed by phosphorous, sodium, and titanium zonation. Am Mineral 101:1696–
1699, doi:10.2138 / am-2016-5707
Anderson DE, Olimpio JC (1977) Progressive homogenization of metamorphic garnets, South Morar, Scotland;
evidence for volume diffusion. Can Mineral 15:205–216
Atherton MP (1968) The variation in garnet, biotite and chlorite composition in medium grade pelitic rocks from
the Dalradian, Scotland, with particular reference to the zonation in garnet. Contrib Mineral Petrol 18:347–
371, doi:10.1007 / bf00399696
Ballevre M, Pitra P, Bohn M (2003) Lawsonite growth in the epidote blueschists from the Ile de Groix
(Armorican Massif, France): a potential geobarometer. J Metamorph Geol 21:723–735, doi:10.1046 / j.1525–
1314.2003.00474.x
Geochem 83:469–533
Becker JS, Zoriy M, Becker JS, Dobrowolska J, Matusch A (2007) Laser ablation inductively coupled plasma mass
spectrometry (LA-ICP-MS) in elemental imaging of biological tissues and in proteomics. J Anal At Spectrom
22:736–744, doi:10.1039 / B701558E
Beinlich A, Klemd R, John T, Gao J (2010) Trace-element mobilization during Ca-metasomatism along a major
fluid conduit: Eclogitization of blueschist as a consequence of fluid–rock interaction. Geochim Cosmochim
Acta 74:1892–1922, doi:10.1016/j.gca.2009.12.011
94 Lanari & Engi
Bell TH, Rubenach MJ, Fleming PD (1986) Porphyroblast nucleation, growth and dissolution in regional
metamorphic rocks as a function of deformation partitioning during foliation development. J Metamorph
Geol 4:37–67, doi:10.1111 / j.1525–1314.1986.tb00337.x
Bence AE, Albee AL (1968) Empirical correction factors for the electron microanalysis of silicates and oxides. J
Geol 76:382–403
Berman RG (1988) Internally consistent thermodynamic data for minerals in the system Na2O–K2O–CaO–MgO–
Bowen NL (1913) The melting phenomena of the plagioclase feldspars. Am J Sci Ser 4 35:577–599, doi:10.2475 / ajs.
s4-35.210.577
Brodie KH, Rutter EH (1985) On the relationship between deformation and metamorphism, with special reference
to the behavior of basic rocks. In: Metamorphic Reactions: Kinetics, Textures, and Deformation. Thompson
AB, Rubie DC (eds). Springer New York, p.138–179
Brouwer FM, Engi M (2005) Staurolite and other high-alumina phases in Alpine eclogite: Analysis of domain
evolution. Can Mineral 43:105–128
Brovarone AV, Groppo C, Hetenyi G, Compagnoni R, Malavieille J (2011) Coexistence of lawsonite-bearing
eclogite and blueschist: phase equilibria modelling of Alpine Corsica metabasalts and petrological evolution
of subducting slabs. J Metamorph Geol 29:583–600, doi:10.1111 / j.1525–1314.2011.00931.x
Brown M (2002) Retrograde processes in migmatites and granulites revisited. J Metam. Geol 20:25–40
Burn M (2016) LA-ICP-QMS Th–U / Pb allanite dating: methods and applications. PhD thesis University of Bern
Caddick MJ, Kohn MJ (2013) Garnet: Witness to the evolution of destructive plate boundaries. Elements 9:427–
432, doi:10.2113 / gselements.9.6.427
Caddick MJ, Bickle MJ, Harris NBW, Holland TJB, Horstwood MSA, Parrish RR, Ahmad T (2007) Burial and
exhumation history of a Lesser Himalayan schist: Recording the formation of an inverted metamorphic
sequence in NW India. Earth Planet Sci Lett 264:375–390
Caddick MJ, Konopásek J, Thompson AB (2010) Preservation of garnet growth zoning and the duration of
prograde metamorphism. J Petrol 51:2327–2347, doi:10.1093 / petrology / egq059
Carlson WD (1989) The significance of intergranular diffusion to the mechanisms and kinetics of porphyroblast
crystallization. Contrib Mineral Petrol 103:1–24
Carlson WD (2002) Scales of disequilibrium and rates of equilibration during metamorphism. Am Mineral
87:185–204, doi:10.2138 / am-2002-2-301
Carlson WD (2010) Dependence of reaction kinetics on H2O activity as inferred from rates of intergranular
diffusion of aluminium. J Metamorph Geol 28:735–752, doi:10.1111 / j.1525–1314.2010.00886.x
Carlson WD, Johnson CD (1991) Coronal reaction textures in garnet amphibolites of the Llano Uplift. Am Mineral
76:756–772
Carlson WD, Pattison DRM, Caddick MJ (2015) Beyond the equilibrium paradigm: How consideration of kinetics
enhances metamorphic interpretation. Am Mineral 100:1659–1667, doi:10.2138 / am-2015-5097
Carmichael DM (1969) On the mechanism of prograde metamorphic reactions in quartz-bearing pelitic rocks.
Cenki-Tok B, Berger A, Gueydan F (2016) Formation and preservation of biotite-rich microdomains in
high-temperature rocks from the Antananarivo Block, Madagascar. Int J Earth Sci 105:1471–1483,
doi:10.1007 / s00531-015-1265-0
Centrella S, Austrheim H, Putnis A (2015) Coupled mass transfer through a fluid phase and volume preservation
during the hydration of granulite: An example from the Bergen Arcs, Norway. Lithos 236–237:245–255,
doi:10.1016 / j.lithos.2015.09.010
Chapman AD, Luffi PI, Saleeby JB, Petersen S (2011) Metamorphic evolution, partial melting and rapid exhumation
above an ancient flat slab: insights from the San Emigdio Schist, southern California. J Metamorph Geol
29:601–626, doi:10.1111 / j.1525–1314.2011.00932.x
Cheng H, Cao D (2015) Protracted garnet growth in high-P eclogite: constraints from multiple geochronology and
P–T pseudosection. J Metamorph Geol 33:613–632, doi:10.1111 / jmg.12136
Cheng H, Nakamura E, Kobayashi K, Zhou Z (2007) Origin of atoll garnets in eclogites and implications for
the redistribution of trace elements during slab exhumation in a continental subduction zone. Am Mineral
92:1119–1129, doi:10.2138 / am.2007.2343
Chernoff CB, Carlson WD (1999) Trace element zoning as a record of chemical disequilibrium during garnet
growth. Geology 27:555–558
Clarke GL, Daczko NR, Nockolds C (2001) A method for applying matrix corrections to X-ray intensity maps
using the Bence–Albee algorithm and Matlab. J Metamorph Geol 19:635–644, doi:10.1046 / j.0263–
4929.2001.00336.x
Connolly JAD (1990) Multivariate phase diagrams: An algorithm based on generalized thermodynamics. Am J
Sci 290:666–718
Connolly JAD (2005) Computation of phase equilibria by linear programming: a tool for geodynamic modeling
and its application to subduction zone decarbonation. Earth Planet Sci Lett 236:524–541
Cossio R, Borghi A, Ruffini R (2002) Quantitative modal determination of geological samples based on X-ray
multielemental map acquisition. Microsc Microanal 8:139–149, doi:10.1017 / S1431927601020062
Cosslett VE, Duncumb P (1956) Micro-analysis by a flying-spot X-ray method. Nature 177:1172–1173
Cygan RT, Lasaga AC (1982) Crystal growth and the formation of chemical zoning in garnets. Contrib Mineral
Petrol 79:187–200, doi:10.1007 / bf01132887
Davis PB, Whitney DL (2006) Petrogenesis of lawsonite and epidote eclogite and blueschist, Sivrihisar Massif,
Turkey. J Metamorph Geol 24:823–849
de Andrade V, Vidal O, Lewin E, O’Brien P, Agard P (2006) Quantification of electron microprobe compositional
maps of rock thin sections: an optimized method and examples. J Metamorph Geol 24:655–668,
doi:10.1111 / j.1525–1314.2006.00660.x
de Béthune P, Laduron D, Bocquet J (1975) Diffusion processes in resorbed garnets. Contrib Mineral Petrol
50:197–204, doi:10.1007 / bf00371039
de Capitani C, Petrakakis K (2010) The computation of equilibrium assemblage diagrams with Theriak / Domino
software. Am Mineral 95:1006–1016, doi:10.2138 / am.2010.3354
DeVore GW (1955) The role of adsorption in the fractionation and distribution of elements. J Geol 63:159–190,
doi::10.1086 / 626242
Dragovic B, Samanta LM, Baxter EF, Selverstone J (2012) Using garnet to constrain the duration and rate of
water-releasing metamorphic reactions during subduction: An example from Sifnos, Greece. Chem Geol
314–317:9–22, doi:10.1016 / j.chemgeo.2012.04.016
Dupuis M (2012) Macro- et microstructure de l’éventail briançonnais. MSc thesis Université de Lille 1
Duval M, Aubert M, Hellstrom J, Grün R (2011) High resolution LA-ICP-MS mapping of U and Th isotopes in an
early Pleistocene equid tooth from Fuente Nueva-3 (Orce, Andalusia, Spain). Quat Geochronol 6:458–467,
doi:10.1016 / j.quageo.2011.04.002
Elvevold S, Gilotti JA (2000) Pressure–temperature evolution of retrogressed kyanite eclogites, Weinschenk Island,
North–East Greenland Caledonides. Lithos 53:127–147, doi:10.1016 / S0024-4937(00)00014–1
Erambert M, Austrheim H (1993) The effect of fluid and deformation on zoning and inclusion patterns in poly-
metamorphic garnets. Contrib Mineral Petrol 115:204–214, doi:10.1007 / bf00321220
Evans TP (2004) A method for calculating effective bulk composition modification due to crystal fractionation in
garnet-bearing schist; implications for isopleth thermobarometry J Metamorph Geol 22:547–557
Evans KA, Bickle MJ (2005) An investigation of the relationship between bulk composition, inferred reaction
progress and fluid flow parameters for layered micaceous carbonates from Maine, U.S.A. J Metamorph Geol
23:181–197
Faryad SW, Klápová H, Nosál L (2010) Mechanism of formation of atoll garnet during high-pressure metamorphism.
Mineral Mag 74:111–126, doi:10.1180 / minmag.2010.074.1.111
Fiannacca P, Lo Pò D, Ortolano G, Cirrincione R, Pezzino A (2012) Thermodynamic modeling assisted by
multivariate statistical image analysis as a tool for unraveling metamorphic P–T-d evolution: an example from
ilmenite–garnet-bearing metapelite of the Peloritani Mountains, Southern Italy. Mineral Petrol 106:151–171,
doi:10.1007 / s00710-012-0228-4
Fisher GW (1973) Nonequilibrium thermodynamics as a model for diffusion-controlled metamorphic processes.
Am J Sci 273:897–924, doi:10.2475 / ajs.273.10.897
Fisher GW, Lasaga AC (1981) Irreversible thermodynamics in petrology. Rev Mineral Geochem 8:171–207
Fitzsimons ICW, Kinny PD, Wetherley S, Hollingsworth DA (2005) Bulk chemical control on metamorphic
monazite growth in pelitic schists and implications for U–Pb age data. J Metamorph Geol 23:261–277,
doi:10.1111 / j.1525–1314.2005.00575.x
Fletcher RC (1977) Quantitative theory for metamorphic differentiation in development of crenulation cleavage.
Geology 5:185–187, doi:10.1130 / 0091–7613
Florence FP, Spear FS (1991) Effects of diffusional modification of garnet growth zoning on PT path calculations.
Foster CT (1977) Mass transfer in sillimanite-bearing pelitic schists near Rangeley, Maine. Am Mineral 62:727–746
Foster CT (1981) A thermodynamic model of mineral segregations in the lower sillimanite zone near Rangeley,
Maine. Am Mineral 66:260–277
Foster CT (1983) Thermodynamic models of biotite pseudomorphs after staurolite. Am Mineral 68:389–397
Foster CT (1986) Thermodynamic models of reactions involving garnet in a sillimanite / staurolite schist. Mineral
Mag 50:427–439
Foster CT (1991) The role of biotite as a catalyst in reaction mechanisms that form sillimanite. Can Mineral
29:943–963
Foster CT (1999) Forward modeling of metamorphic textures. Can Mineral 37:415–429
181:327–340
96 Lanari & Engi
Foster G, Parrish RR, Horstwood MSA, Chenery S, Pyle J, Gibson HD (2004) The generation of prograde P–T–t
Sci Lett 228:125–142, doi:10.1016 / j.epsl.2004.09.024
Gaidies F, Abart R, de Capitani C, Schuster R, Connolly JAD, Reusser E (2006) Characterization of
thermobarometry. J Metamorph Geol 24:451–475, doi:10.1111 / j.1525–1314.2006.00648.x
Gaidies F, de Capitani C, Abart R (2008) THERIA_G: a software program to numerically model prograde garnet
growth. Contrib Mineral Petrol 155:657–671, doi:10.1007 / s00410-007-0263-z
Gaidies F, Pattison DRM, de Capitani C (2011) Toward a quantitative model of metamorphic nucleation and
growth. Contrib Mineral Petrol 162:975–993, doi:10.1007 / s00410-011-0635-2
Ganguly J (2010) Cation diffusion kinetics in aluminosilicate garnets and geological applications. Rev Mineral
Geochem 72:559–601
Ganguly J, Chakraborty S, Sharp TG, Rumble D (1996) Constraint on the time scale of biotite-grade metamorphism
during Acadian orogeny from a natural garnet–garnet diffusion couple. Am Mineral 81:1208–1216
Gatewood MP, Dragovic B, Stowell HH, Baxter EF, Hirsch DM, Bloom R (2015) Evaluating chemical equilibrium
in metamorphic rocks using major element and Sm–Nd isotopic age zoning in garnet, Townshend Dam,
Vermont, USA. Chemical Geology 401:151–168, doi:10.1016 / j.chemgeo.2015.02.017
Giuntoli F (2016) Assembly of continental fragments during subduction at HP: Metamorphic history of the central
Sesia Zone (NW Alps). PhD thesis University of Bern
Goldschmidt VM (1911) Die Kontaktmetamorphose im Kristianiagebiet. In Kommission bei J. Dybwad, Kristiania
Goncalves P, Oliot E, Marquer D, Connolly JAD (2012) Role of chemical processes on shear zone formation: an
example from the Grimsel metagranodiorite (Aar Massif, Central Alps). J Metamorph Geol 30:703–722,
doi:10.1111 / j.1525–1314.2012.00991.x
Gottschalk M (1996) Internally consistent thermodynamic data for rock-forming minerals in the system SiO2–TiO2–
Al2O3–Fe2O3–CaO–MgO–FeO–K2O–Na2O–H2O–CO2.EurJMineral9:175–223,doi:10.1127 / ejm / 9/1 / 0175
Grand’Homme A, Janots E, Seydoux-Guillaume A-M, Guillaume D, Bosse V, Magnin V (2016) Partial resetting
of the U–Th–Pb systems in experimentally altered monazite: Nanoscale evidence of incomplete replacement.
Geology 44:431–434
Greenfield JE (1997) Migmatite formation at Mt Stafford, central Australia. PhD thesis, University of Sidney
Greenfield JE, Clarke GL, Bland M, Clark DJ (1996) In-situ migmatite and hybrid diatexite at Mt Stafford, central
Australia. J Metamorph Geol 14:413–426, doi:10.1046 / j.1525–1314.1996.06002.x
Greenwood HJ (1967) The N-dimensional tie-line problem. Geochimi Cosmochim Acta 31:465–490
Groppo C, Castelli D (2010) Prograde P–T evolution of a lawsonite eclogite from the Monviso meta-ophiolite
(Western Alps): Dehydration and redox reactions during subduction of oceanic FeTi-oxide gabbro. J Petrol
51:2489–2514, doi:10.1093 / petrology / egq065
Groppo C, Lombardo B, Rolfo F, Pertusati P (2007) Clockwise exhumation path of granulitized eclogites
from the Ama Drime range (Eastern Himalayas). J Metamorph Geol 25:51–75, doi:10.1111 / j.1525–
1314.2006.00678.x
Grosch EG, McLoughlin N, Lanari P, Erambert M, Vidal O (2014) Microscale mapping of alteration conditions
and potential biosignatures in basaltic–ultramafic rocks on early earth and beyond. Astrobiology 14:216–228,
doi:10.1089 / ast.2013.1116
Guevara VE, Caddick MJ (2016) Shooting at a moving target: phase equilibria modelling of high-temperature
metamorphism. J Metamorph Geol 34:209–235, doi:10.1111 / jmg.12179
Harrison TM, Catlos EJ, Montel J-M (2002) U–Th–Pb dating of phosphate minerals. Rev Mineral Geochem
48:524–558, doi:10.2138 / rmg.2002.48.14
Harvey J, Baxter EF (2009) An improved method for TIMS high precision neodymium isotope analysis of very
small aliquots (1–10 ng). Chem Geol 258:251–257, doi:10.1016 / j.chemgeo.2008.10.024
Hickmott DD, Shimizu N, Spear FS, Selverstone J (1987) Trace-element zoning in a metamorphic garnet. Geology
15:573–576, doi:10.1130 / 0091–7613(1987)15 < 573:tziamg > 2.0.co;2
Hoisch TD, Wells ML, Grove M (2008) Age trends in garnet-hosted monazite inclusions from upper amphibolite
facies schist in the northern Grouse Creek Mountains, Utah. Geochimica et Cosmochimica Acta 72:5505–
5520, doi:10.1016 / j.gca.2008.08.012
Holland TJB, Powell R (2011) An improved and extended internally consistent thermodynamic dataset for
phases of petrological interest, involving a new equation of state for solids. J Metamorph Geol 29:333–383,
doi:10.1111 / j.1525–1314.2010.00923.x
Hollister LS (1966) Garnet zoning: an interpretation based on the Rayleigh fractionation model. Science 154:1647–1651
Hoschek G (2013) Garnet zonation in metapelitic schists from the Eclogite Zone, Tauern Window, Austria:
comparison of observed and calculated profiles. Eur J Mineral 25:615–629, doi:10.1127 /
0935–
1221 / 2013 / 0025–2310
Indares A, Rivers T (1995) Textures, metamorphic reactions and thermobarometry of eclogitized metagabbros; a
Proterozoic example. Eur J Mineral 7:43–56
Janots E, Engi M, Berger A, Allaz J, Schwarz J-O, Spandler C (2008) Prograde metamorphic sequence of REE-
from 250 to 610 °C. J Metamorph Geol 26:509–526, doi:10.1111 / j.1525–1314.2008.00774.x
Johnson CD, Carlson WD (1990) The origin of olivine–plagioclase coronas in metagabbros from the Adirondack
Mountains, New York. J Metamorph Geol 8:697–717, doi:10.1111 / j.1525–1314.1990.tb00496.x
Johnson TE, Clark C, Taylor RJM, Santosh M, Collins AS (2015) Prograde and retrograde growth of monazite in
migmatites: An example from the Nagercoil Block, southern India. Geosci Front 6:373–387, doi:10.1016 / j.
gsf.2014.12.003
Kelsey DE, Hand M (2015) On ultrahigh temperature crustal metamorphism: Phase equilibria, trace element
thermometry, bulk composition, heat sources, timescales and tectonic settings. Geosci Front 6:311–356,
doi:10.1016 / j.gsf.2014.09.006
SiO2–H2O–TiO2–ZrO2 system. J Metamorph Geol 29:151–166, doi:10.1111 / j.1525–1314.2010.00910.x
Kelsey DE, Clark C, Hand M (2008) Thermobarometric modeling of zircon and monazite growth in melt-bearing
Kohn MJ (2009) Models of garnet differential geochronology. Geochim Cosmochim Acta 73:170–182,
doi:10.1016 / j.gca.2008.10.004
Kohn MJ (2014) 4.7 - Geochemical zoning in metamorphic minerals A2. In: Treatise on Geochemistry (Second
Edition). Turekian KK, (ed) Elsevier, Oxford, p 249–280
Kohn MJ (2016) Metamorphic chronology — a tool for all ages: Past achievements and future prospects. Am
Mineral 101:25–42, doi:10.2138 / am-2016-5146
Kohn MJ, Spear FS (2000) Retrograde net transfer reaction insurance for pressure–temperature estimates. Geology
28:1127–1130
Kohn MJ, Wieland MS, Parkinson CD, Upreti BN (2004) Miocene faulting at plate tectonic velocity in the
Himalaya of central Nepal. Earth Planet Sci Lett 228:299–310
Kohn MJ, Wieland MS, Parkinson CD, Upreti BN (2005) Five generations of monazite in Langtang gneisses:
implications for chronology of the Himalayan metamorphic core. J Metamorph Geol 23:399–406,
doi:10.1111 / j.1525–1314.2005.00584.x
Kohn MJ, Penniston-Dorland SC, Ferreira JC (2016) Implications of near-rim compositional zoning in rutile for
geothermometry, geospeedometry, and trace element equilibration. Contrib Mineral Petrol 171:78
Kohn MJ, Penniston–Dorland SC (2017) Diffusion: Obstacles and opportunities in petrochronology. Rev Mineral
Geochem 83:103–152
Konrad-Schmolke M, O’Brien PJ, De Capitani C, Carswell DA (2008) Garnet growth at high- and ultra-high
pressure conditions and the effect of element fractionation on mineral modes and composition. Lithos
103:309–332
Korhonen FJ, Stout JH (2005) Borosilicate- and phengite-bearing veins from the Grenville Province of Labrador:
evidence for rapid uplift. J Metamorph Geol 23:297–311, doi:10.1111 / j.1525–1314.2005.00577.x
Korzhinskii DS (1936) Mobility and inertness of components in metasomatosis. Izv Akad Nauk SSSRm Ser Geol
1:58–60
Korzhinskii DS (1959) Physicochemical Basis of the Analysis of the Paragenesis of Minerals. Consultants Bureau,
New York
Kriegsman LM, Nyström AI (2003) Melt segregation rates in migmatites: review and critique of common
approaches. Geological Society, London, Special Publications 220:203–212
Lanari P, Duesterhoeft E (2016) Thermodynamic modeling using BINGO-ANTIDOTE: A new strategy to
investigate metamorphic rocks. Geophys Res Abstr 18:EGU2016-11363
Lanari P, Guillot S, Schwartz S, Vidal O, Tricart P, Riel N, Beyssac O (2012) Diachronous evolution of the alpine
continental subduction wedge: evidence from P–T estimates in the Briançonnais Zone houillere (France—
Western Alps). J Geodyn 56–57:39–54
Lanari P, Riel N, Guillot S, Vidal O, Schwartz S, Pêcher A, Hattori KH (2013) Deciphering high-pressure
metamorphism in collisional context using microprobe mapping methods: Application to the Stak eclogitic
massif (northwest Himalaya). Geology 41:111–114, doi:10.1130 / g33523.1
Lanari P, Wagner T, Vidal O (2014a) A thermodynamic model for di-trioctahedral chlorite from experimental and
natural data in the system MgO–FeO–Al2O3–SiO2–H2O: applications to P–T sections and geothermometry.
Contr Mineral Petrol 167–968, doi:10.1007/s00410-014-0968-8
98 Lanari & Engi
Lanari P, Rolland Y, Schwartz S, Vidal O, Guillot S, Tricart P, Dumont T (2014b) P–T–t estimation of deformation
in low-grade quartz–feldspar-bearing rocks using thermodynamic modelling and 40Ar / 39Ar dating techniques:
example of the Plan-de-Phasy shear zone unit (Briançonnais Zone, Western Alps). Terra Nova 26:130–138,
doi:10.1111 / ter.12079
Lanari P, Vidal O, De Andrade V, Dubacq B, Lewin E, Grosch EG, Schwartz S (2014c) XMapTools: A MATLAB©-
based program for electron microprobe X-ray image processing and geothermobarometry. Comp Geosci
62:227–240, doi:10.1016 / j.cageo.2013.08.010
Lanari P, Giuntoli F, Burn M, Engi M (2017) An inverse modeling approach to obtain P–T conditions of metamorphic
stages involving garnet growth and resorption. Eur J Mineral in press, doi:10.1127/ejm/2017/0029-2597
Lasaga AC (1986) Metamorphic reaction rate laws and development of isograds. Mineral Mag 50:359–373
Lasaga AC (1998) Kinetic Theory in the Earth Sciences. Princeton University Press, Princeton, New Jersey, 822 p.
López-Carmona A, Abati J, Pitra P, Lee JKW (2014) Retrogressed lawsonite blueschists from the NW Iberian
Massif: P–T–t constraints from thermodynamic modelling and 40Ar / 39Ar geochronology. Contrib Mineral
Petrol 167:1–20, doi:10.1007 / s00410-014-0987-5
Loury C, Rolland Y, Guillot S, Mikolaichuk AV, Lanari P, Bruguier O, Bosch D (2015) Crustal-scale structure of
South Tien Shan: implications for subduction polarity and Cenozoic reactivation. Geol Soc, London, Spec
Publ 427, doi:10.1144 / sp427.4
Loury C, Rolland Y, Cenki-Tok B, Lanari P, Guillot S (2016) Late Paleozoic evolution of the South Tien
Shan: Insights from P–T estimates and allanite geochronology on retrogressed eclogites (Chatkal range,
Kyrgyzstan). J Geodyn 96:62–80, doi:10.1016 / j.jog.2015.06.005
Mahan KH, Goncalves P, Williams ML, Jercinovic MJ (2006) Dating metamorphic reactions and fluid flow:
application to exhumation of high-P granulites in a crustal-scale shear zone, western Canadian Shield. J
Metamorph Geol 24:193–217, doi:10.1111 / j.1525–1314.2006.00633.x
Manzotti P, Ballèvre M (2013) Multistage garnet in high-pressure metasediments: Alpine overgrowths on Variscan
detrital grains. Geology 41:1151–1154
Marmo BA, Clarke GL, Powell R (2002) Fractionation of bulk rock composition due to porphyroblast growth:
effects on eclogite facies mineral equilibria, Pam Peninsula, New Caledonia. J Metamorph Geol 20:151–165,
doi:10.1046 / j.0263–4929.2001.00346.x
Martin AJ, Gehrels GE, DeCelles PG (2007) The tectonic significance of (U,Th)/Pb ages of monazite inclusions
in garnet from the Himalaya of central Nepal. Chem Geol 244:1–24, doi:10.1016 / j.chemgeo.2007.05.003
Martin C, Debaille V, Lanari P, Goderis S, Vandendael I, Vanhaecke F, Vidal O, Claeys P (2013) REE and Hf
distribution among mineral phases in the CV–CK clan: A way to explain present-day Hf isotopic variations
in chondrites. Geochim Cosmochim Acta 120:496–513, doi:10.1016 / j.gca.2013.07.006
Mészaros M, Hofmann BA, Lanari P, et al. (2016) Petrology and geochemistry of feldspathic impact-melt breccia
Abar al’ Uj 012, the first lunar meteorite from Saudi Arabia. Meteorit Planet Sci 51:1830–1848
Meyer M, John T, Brandt S, Klemd R (2011) Trace element composition of rutile and the application of Zr-in-rutile
thermometry to UHT metamorphism (Epupa Complex, NW Namibia). Lithos 126:388–401
Milord I, Sawyer EW, Brown M (2001) Formation of diatexite migmatite and granite magma during anatexis
of semi-pelitic metasedimentary rocks: an Example from St. Malo, France. J Petrol 42:487–505,
doi:10.1093 / petrology / 42.3.487
Miron GD, Wagner T, Kulik DA, Heinrich CA (2016) Internally consistent thermodynamic data for aqueous species
in the system Na–K–Al–Si–O–H–Cl. Geochim Cosmochim Acta 187:41–78, doi:10.1016 / j.gca.2016.04.026
Montel J-M (1993) A model for monazite / melt equilibrium and application to the generation of granitic magmas.
Chemical Geology 110:127–146
Montel J-M, Kornprobst J, Vielzeuf D (2000) Preservation of old U–Th–Pb ages in shielded monazite: example
from the Beni Bousera Hercynian kinzigites (Marocco). J Metamorph Geol 18:335–342
Mottram CM, Parrish RR, Regis D, Warren CJ, Argles TW, Harris NBW, Roberts NMW (2015) Using U–Th–Pb
petrochronology to determine rates of ductile thrusting: Time windows into the Main Central Thrust, Sikkim
Himalaya. Tectonics 34:2014TC003743, doi:10.1002 / 2014TC003743
with application to metapelitic schist from the Kootenay Arc, British Columbia, Canada. J Metamorph Geol
31:525–548, doi:10.1111 / jmg.12032
Netting A, Payne J, Wade B, Raimondo T (2011) Trace element micro-analytical imaging via laser ablation
inductively coupled plasma mass spectrometry (LA-ICP-MS). Microsc Microanal 17:566–567,
doi:10.1017 / S1431927611003709
Orville PM (1969) A model for metamorphic origin of thin layered amphibloites. Am J Sci 267:64–86
Paton C, Hellstrom J, Paul B, Woodhead J, Hergt J (2011) Iolite: Freeware for the visualisation and processing of
mass spectrometric data. J Anal At Spectrom 26:2508–2518, doi:10.1039 / C1JA10172B
Pattison DRM, De Capitani C, Gaidies F (2011) Petrological consequences of variations in metamorphic reaction
affinity. J Metamorph Geol 29:953–977, doi:10.1111 / j.1525–1314.2011.00950.x
Paul B, Paton C, Norris A, Woodhead J, Hellstrom J, Hergt J, Greig A (2012) CellSpace: A module for creating
spatially registered laser ablation images within the Iolite freeware environment. J Anal At Spectrom 27:700–
706, doi:10.1039 / C2JA10383D
Paul B, Woodhead JD, Paton C, Hergt JM, Hellstrom J, Norris CA (2014) Towards a method for quantitative
LA-ICP-MS imaging of multi-phase assemblages: mineral identification and analysis correction procedures.
Geostand Geoanal Res 38:253–263, doi:10.1111 / j.1751-908X.2014.00270.x
Peng S, Hu Q, Ewing RP, Liu C, Zachara JM (2012) Quantitative 3-D Elemental Mapping by LA-ICP-MS of
a Basaltic Clast from the Hanford 300 Area, Washington, USA. Environ Sci Technol 46:2025–2032,
doi:10.1021 / es2023785
Pickering-Witter J, Johnston AD (2000) The effects of variable bulk composition on the melting systematics of
fertile peridotitic assemblages. Contrib Mineral Petrol 140:190–211, doi:10.1007 / s004100000183
Pollington AD, Baxter EF (2010) High resolution Sm–Nd garnet geochronology reveals the uneven pace of
tectonometamorphic processes. Earth Planet Sci Lett 293:63–71, doi:10.1016 / j.epsl.2010.02.019
Pourteau A, Sudo M, Candan O, Lanari P, Vidal O, Oberhänsli R (2013) Neotethys closure history of Anatolia:
insights from 40Ar–39Ar geochronology and P–T estimation in high-pressure metasedimentary rocks. J
Metamorph Geol 31:585–606, doi:10.1111 / jmg.12034
Powell R, Downes J (1990) Garnet porphyroblast-bearing leucosomes in metapelites: mechanisms, phase
diagrams, and an example from Broken Hill, Australia. In: High-temperature metamorphism and crustal
anatexis. Springer, p 105–123
Powell R, Holland TJB (2008) On thermobarometry. J Metamorph Geol 26:155–179, doi:10.1111 / j.1525–
1314.2007.00756.x
Powell R, Holland T (2010) Using equilibrium thermodynamics to understand metamorphism and metamorphic
rocks. Elements 6:309–314, doi:10.2113 / gselements.6.5.309
Powell R, Holland T, Worley B (1998) Calculating phase diagrams involving solid solutions via non-linear
equations, with examples using THERMOCALC. J Metamorph Geol 16:577–588, doi:10.1111 / j.1525–
1314.1998.00157.x
Powell R, Guiraud M, White RW (2005) Truth and beauty in metamorphic phase-equilibria: conjugate variables
and phase diagrams. Can Mineral 43:21–33
Putnis A (2009) Mineral replacement reactions. Rev Mineral Geochem 70:87–124
Putnis A, John T (2010) Replacement processes in the earth’s crust. Elements 6:159–164, doi:10.2113 /
gselements.6.3.159
reactions. Geol Mater Res 1:1–49
Pyle JM, Spear FS (2003) Four generations of accessory-phase growth in low-pressure migmatites from SW New
Hampshire. Am Mineral 88:338–351
Raimondo T, Payne J, Wade B, Lanari P, Clark C, Hand M (2017) Trace element mapping by LA-ICP-MS:
assessing geochemical mobility in garnet. Contrib Mineral Petrol, in press.
Regis D, Rubatto D, Darling J, Cenki-Tok B, Zucali M, Engi M (2014) Multiple metamorphic stages within an
eclogite-facies terrane (Sesia Zone, Western Alps) revealed by Th–U–Pb petrochronology. J Petrol 55:1429–
1456, doi:10.1093 / petrology / egu029
Ridley JR (1984) The significance of deformation associated with blueschist facies metamorphism on the Aegean
island of Syros. In: The Geological Evolution of the Eastern Mediterranean. Vol 17. Dixon JE, Robertson
AHF (eds) Geological Society, 545–550
Riel N, Lanari P (2015) Techniques, méthodes et outils pour la quantification du métamorphisme. Géochroniques
136:53–60
Rittner M, Müller W (2012) 2D mapping of LA-ICPMS trace element distributions using R. Comput Geosci
42:152–161, doi:10.1016 / j.cageo.2011.07.016
Roberts MP, Finger F (1997) Do U–Pb zircon ages from granulites reflect peak metamorphic conditions? Geology
25:319–322, doi:10.1130 / 0091–7613(1997)025 < 0319:dupzaf > 2.3.co;2
Robyr M, Darbellay B, Baumgartner LP (2014) Matrix-dependent garnet growth in polymetamorphic rocks of the
Sesia zone, Italian Alps. J Metamorph Geol 32:3–24, doi:10.1111 / jmg.12055
Rubatto D, Hermann J, Buick IS (2006) Temperature and bulk composition control on the growth of monazite
and zircon during low-pressure anatexis (Mount Stafford, Central Australia). J Petrol 47:1973–1996,
doi:10.1093 / petrology / egl033
Sayab M (2006) Decompression through clockwise P–T path: implications for early N–S shortening orogenesis
in the Mesoproterozoic Mt Isa Inlier (NE Australia). J Metamorph Geol 24:89–105, doi:10.1111 / j.1525–
1314.2005.00626.x
Scheffer C, Vanderhaeghe O, Lanari P, Tarantola A, Ponthus L, Photiades A, France L (2016) Syn- to post-
orogenic exhumation of metamorphic nappes: Structure and thermobarometry of the western Attic-Cycladic
metamorphic complex (Lavrion, Greece). J Geodyn 96:174–193, doi:10.1016 / j.jog.2015.08.005
100 Lanari & Engi
Schorn S, Diener JFA (2016) Details of the gabbro-to-eclogite transition determined from microtextures and
calculated chemical potential relationships. J Metamorph Geol, doi:10.1111 / jmg.12220
Schwarz J-O, Engi M, Berger A (2011) Porphyroblast crystallization kinetics: The role of the nutrient production
rate. J Metamorph Geol 29, doi:10.1111 / j.1525–1314.2011.00927.x
Šelih VS, van Elteren JT (2011) Quantitative multi-element mapping of ancient glass using a simple and robust
LA-ICP-MS rastering procedure in combination with image analysis. Anal Bioanal Chem 401:745–755,
doi:10.1007 / s00216-011-5119-8
Selverstone J, Spear FS (1985) Metamorphic P–T Paths from pelitic schists and greenstones from the south-west
Tauern Window, Eastern Alps. J Metamorph Geol 3:439–465, doi:10.1111 / j.1525–1314.1985.tb00329.x
Sousa J, Kohn MJ, Schmitz MD, Northrup CJ, Spear FS (2013) Strontium isotope zoning in garnet: implications
for metamorphic matrix equilibration, geochronology and phase equilibrium modelling. J Metamorph Geol
31:437–452, doi:10.1111 / jmg.12028
Spear FS (1988a) The Gibbs method and Duhem’s theorem: The quantitative relationships among P, T, chemical
potential, phase composition and reaction progress in igneous and metamorphic systems. Contrib Mineral
Petrol 99:249–256, doi:10.1007 / bf00371465
Spear FS (1988b) Metamorphic fractional crystallization and internal metasomatism by diffusional homogenization
of zoned garnets. Contrib Mineral Petrol 99:507–517, doi:10.1007 / bf00371941
Spear FS (1991) On the interpretation of peak metamorphic temperatures in light of garnet diffusion during
cooling. J Metamorph Geol 9:379–388, doi:10.1111 / j.1525–1314.1991.tb00533.x
Spear FS (1993) Metamorphic Phase Equilibria and Pressure–temperature–Time Paths. Monograph Mineral Soc
Am, Washington, D.C.
Spear FS, Daniel CG (2001) Diffusion control of garnet growth, Harpswell Neck, Maine, USA. J Metamorph Geol
19:179–195
Spear FS, Kohn MJ (1996) Trace element zoning in garnet as a monitor of crustal melting. Geology 24:1099–1102
Spear FS, Selverstone J (1983) Quantitative P–T paths from zoned minerals: Theory and applications. Contrib
Spear FS, Selverstone J, Hickmott D, Crowley P, Hodges KV (1984) P–T paths from garnet zoning: A new
technique for deciphering tectonic processes in crystalline terranes. Geology 12:87–90
Spear FS, Kohn MJ, Florence FP, Menard T (1990) A model for garnet and plagioclase growth in pelitic
schists: implications for thermobarometry and P–T path determinations. J Metamorph Geol 8:683–696,
doi:10.1111 / j.1525–1314.1990.tb00495.x
Spear FS, Kohn MJ, Florence FP, Menard T (1991) A model for garnet and plagioclase growth in pelitic schists:
Implications for thermobarometry and P–T path determinations. J Metamorph Geol 8:683–696
thermodynamic modeling. Contrib Mineral Petrol 168:1–15
Papers 523, doi:10.1130 / 2016.2523(02)
Stepanov AS, Hermann J, Rubatto D, Rapp RP (2012) Experimental study of monazite / melt partitioning
with implications for the REE, Th and U geochemistry of crustal rocks. Chem Geol 300–301:200–220,
doi:10.1016 / j.chemgeo.2012.01.007
Stevens G, Clemens JD, Droop GTR (1997) Melt production during granulite-facies anatexis: experimental data from
“primitive” metasedimentary protoliths. Contrib Mineral Petrol 128:352–370, doi:10.1007 / s004100050314
Stowell H, Tulloch A, Zuluaga C, Koenig A (2010) Timing and duration of garnet granulite metamorphism in
magmatic arc crust, Fiordland, New Zealand. Chem Geol 273:91–110, doi:10.1016 / j.chemgeo.2010.02.015
reaction textures. Contrib Mineral Petrol 129:43–52, doi:10.1007 / s004100050322
Thompson J (1959) Local equilibrium in metasomatic processes. In: Researches in Geochemistry 1, Abelson PH
(Ed) Wiley, New York, p. 427–457
Thompson JB (1955) The thermodynamic basis for the mineral facies concept. Am J Sci 253:65–103,
doi:10.2475 / ajs.253.2.65
Thompson JB (1970) Geochemical reaction and open systems. Geochim Cosmochim Acta 34:529–551,
doi:10.1016 / 0016–7037(70)90015–3
Tinkham DK, Ghent ED (2005a) Estimating P–T conditions of garnet growth with isochemical phase-diagrams
sections and the problem of effective bulk-composition. Can Mineral 43:35–50, doi:10.2113 / gscanmin.43.1.35
Tinkham DK, Ghent ED (2005b) XRMapAnal: A program for analysis of quantitative X-ray maps. Am Mineral
90:737–744, doi:10.2138 / am.2005.1483
Tinkham DK, Zuluaga CA, Stowell HH (2001) Metapelite phase equilibria modeling in MnNCKFMASH: The
effect of variable Al2O3 and MgO/(MgO + FeO) on mineral stability. Geol Mater Res 3:1–42
Tóth M, Grandjean V, Engi M (2000) Polyphase evolution and reaction sequence of compositional domains in
metabasalt: A model based on local chemical equilibrium and metamorphic differentiation. Geol J 35:163–183
Tracy RJ (1982) Compositional zoning and inclusions in metamorphic minerals. Rev Mineral Geochem 10:355–397
Tracy RJ, Robinson P, Thompson AB (1976) Garnet composition and zoning in the determination of temperature
and pressure of metamorphism, central Massachusetts. Am Mineral 61:762–775
Trincal V, Lanari P, Buatier M, Lacroix B, Charpentier D, Labaume P, Muñoz M (2015) Temperature micro-
mapping in oscillatory-zoned chlorite: Application to study of a green-schist facies fault zone in the Pyrenean
Axial Zone (Spain). Am Mineral 100:2468–2483, doi:10.2138 / am-2015-5217
Tropper P, Manning CE, Harlov DE (2011) Solubility of CePO4 monazite and YPO4 xenotime in H2O and H2O–
NaCl at 800 °C and 1 GPa: Implications for REE and Y transport during high-grade metamorphism.Chem
Geol 282:58–66, doi:10.1016 / j.chemgeo.2011.01.009
Ubide T, McKenna CA, Chew DM, Kamber BS (2015) High-resolution LA-ICP-MS trace element
mapping of igneous minerals: In search of magma histories. Chem Geol 409:157–168, doi:10.1016 / j.
chemgeo.2015.05.020
Vance D, Mahar E (1998) Pressure–temperature paths from PT pseudosections and zoned garnets: potential,
limitations and examples from the Zanskar Himalaya, NW India. Contrib Mineral Petrol 132:225–245
Verdecchia SO, Reche J, Baldo EG, Segovia-Diaz E, Martinez FJ (2013) Staurolite porphyroblast controls on
local bulk compositional and microstructural changes during decompression of a St–Bt–Grt–Crd–And
schist (Ancasti metamorphic complex, Sierras Pampeanas, W Argentina). J Metamorph Geol 31:131–146,
doi:10.1111 / jmg.12003
Vernon RH (1989) Porphyroblast-matrix microstructural relationships: recent approaches and problems. Geol Soc,
London, Spec Publ 43:83–102, doi:10.1144 / gsl.sp.1989.043.01.05
Vernon RH, Collins WJ (1988) Igneous microstructures in migmatites. Geology 16:1126–1129, doi:10.1130 / 0091–
7613(1988)016 < 1126:imim > 2.3.co;2
Vidal O, De Andrade V, Lewin E, Munoz M, Parra T, Pascarelli S (2006) P–T–deformation–Fe3+/Fe2 + mapping
at the thin section scale and comparison with XANES mapping: application to a garnet-bearing metapelite
from the Sambagawa metamorphic belt (Japan). J Metamorph Geol 24:669–683, doi:10.1111 / j.1525–
1314.2006.00661.x
Vitale Brovarone A, Beltrando M, Malavieille J, Giuntoli F, Tondella E, Groppo C, Beyssac O, Compagnoni R
(2011) Inherited Ocean–Continent Transition zones in deeply subducted terranes: Insights from Alpine
Corsica. Lithos 124:273–290, doi:10.1016 / j.lithos.2011.02.013
Vrijmoed JC, Hacker BR (2014) Determining P–T paths from garnet zoning using a brute-force computational
method. Contrib Mineral Petrol 167:1–13, doi:10.1007 / s00410-014-0997-3
Warren CJ, Waters DJ (2006) Oxidized eclogites and garnet-blueschists from Oman: P–T path modelling in the
NCFMASHO system. J Metamorph Geol 24:783–802, doi:10.1111 / j.1525–1314.2006.00668.x
Waters DJ, Lovegrove DP (2002) Assessing the extent of disequilibrium and overstepping of prograde metamorphic
reactions in metapelites from the Bushveld Complex aureole, South Africa. J Metamorph Geol 20:135–149,
doi:10.1046 / j.0263–4929.2001.00350.x
White RW, Powell R (2010) Retrograde melt–residue interaction and the formation of near-anhydrous leucosomes
in migmatites. J Metamorph Geol 28:579–597, doi:10.1111 / j.1525–1314.2010.00881.x
White RW, Powell R, Holland TJB (2001) Calculation of partial melting equilibria in the system Na2O–CaO–
K2O–FeO–MgO–Al2O3–SiO2–H2O (NCKFMASH). J Metamorph Geol 19:139–153, doi:10.1046 / j.0263–
4929.2000.00303.x
White RW, Powell R, Clarke GL (2003) Prograde Metamorphic Assemblage evolution during partial melting of
metasedimentary rocks at low pressures: Migmatites from Mt Stafford, Central Australia. J Petrol 44:1937–
1960, doi:10.1093 / petrology / egg065
the textural evolution of metamorphic rocks. J Metamorph Geol 26:181–198, doi:10.1111 /
j.1525–
1314.2008.00764.x
White WM (2013) Geochemistry. Wiley-Blackwell
Whitehouse MJ, Platt JP (2003) Dating high-grade metamorphism—constraints from rare-earth elements in zircon
and garnet. Contrib Mineral Petrol 145:61–74, doi:10.1007 / s00410-002-0432-z
Whitney DL, Evans BW (2010) Abbreviations for names of rock-forming minerals. Am Mineral 95:185–187,
doi:10.2138 / am.2010.3371
Williams ML (1994) Sigmoidal inclusion trails, punctuated fabric development, and interactions between
metamorphism and deformation. J Metamorph Geol 12:1–21, doi:10.1111 / j.1525–1314.1994.tb00001.x
Williams ML, Jercinovic MJ, Hetherington CJ (2007) Microprobe monazite geochronology: Understanding
geologic processes by integrating composition and chronology. Ann Rev Earth Planet Sci 35:137–175,
doi:10.1146 / annurev.earth.35.031306.140228
Williams ML, Jercinovic MJ, Mahan KH, Dumond G (2017) Electron microprobe petrochronology. Rev Mineral
Geochem 83:153–182
145:228–250
102 Lanari & Engi
Wintsch RP (1985) The possible effects of deformation on chemical processes in metamorphic fault zones. In:
Metamorphic Reactions: Kinetics, Textures, and Deformation. Thompson AB, Rubie DC (eds). Springer
New York, New York, NY, p 251–268
Woodhead JD, Hellstrom J, Hergt JM, Greig A, Maas R (2007) Isotopic and elemental imaging of geological
materials by laser ablation inductively coupled plasma-mass spectrometry. Geostand Geoanal Res 31:331–
343, doi:10.1111 / j.1751-908X.2007.00104.x
Yakymchuk C, Brown M (2014) Behaviour of zircon and monazite during crustal melting. J Geol Soc 171:465–
479, doi:10.1144 / jgs2013-115
Yakymchuk C, Clark C, White RW (2017) Phase relations, reaction sequences and petrochronology. Rev Mineral
Geochem 83:13–53
Yamato P, Agard P, Burov E, Le Pourhiet L, Jolivet L, Tiberi C (2007) Burial and exhumation in a subduction
wedge: Mutual constraints from thermomechanical modeling and natural P–T–t data (Schistes Lustrés,
western Alps). J Geophys Res: Solid Earth 112:B07410, doi:10.1029 / 2006JB004441
Zeh A (2006) Calculation of garnet fractionation in metamorphic rocks, with application to a flat-top, Y-rich
garnet population from the Ruhla Crystalline Complex, Central Germany. J Petrol 47:2335–2356,
doi:10.1093 / petrology / egl046
Zhang L, Wang Q, Song S (2009) Lawsonite blueschist in Northern Qilian, NW China: P–T pseudosections and
petrologic implications. J Asian Earth Sci 35:354–366, doi:10.1016 / j.jseaes.2008.11.007
Zuluaga CA, Stowell HH, Tinkham DK (2005) The effect of zoned garnet on metapelite pseudosection topology
and calculated metamorphic P–T paths. Am Mineral 90:1619–1628, doi:10.2138 / am.2005.1741
Vol. 83 pp. XXX-XXX, 2017 4
Diffusion: Obstacles and Opportunities

in Petrochronology
Matthew J. Kohn
Boise, ID 83725
USA
Sarah C. Penniston-Dorland
Department of Geology
University of Maryland
College Park, MD 20742
USA
sarahpd@umd.edu

Many of the approaches in petrochronology are rooted in the assumption of equilibrium.
Diffusion is an expression of disequilibrium: the movement of mass in response to chemical
potential gradients, and isotopes in response to isotopic gradients. It is extremely important
that we be aware of how the effects of diffusion can place obstacles across our path towards
petrochronologic enlightenment. Conversely the effects of diffusion also provide opportunities
for understanding rates, processes, and conditions experienced by rocks. The enormity of the field
does not permit us to provide a comprehensive review of either the mathematics of diffusion or
quantitative data that have been obtained relevant to the interpretation of diffusive processes in
rocks and minerals. Many resources cover these topics, including RiMG volume 72 (Diffusion in
Minerals and Melts; Zhang and Cherniak 2010) and several textbooks (Crank 1975; Glicksman
2000). Particularly relevant to the discussion of petrochronology are summaries of the theory
and controls on diffusion (Brady and Cherniak 2010; Zhang 2010), as well as diffusion rates
in feldspar (Cherniak 2010a), accessory minerals (Cherniak 2010b), garnet (Ganguly 2010),
mica, pyroxene, and amphibole (Cherniak and Dimanov 2010), and melts (Zhang and Ni 2010;
Zhang et al. 2010). Rather than duplicate that material, our goal is to explore the obstacles
and opportunities presented by the effects of diffusion as they inform the rates of petrologic
processes. To achieve this goal, we emphasize key principles and illustrative examples.
Quantitative interpretation of the effects of diffusion assumes predictability of numerous
factors that may affect chemical or isotopic transport, including temperature, initial and
boundary conditions, water and oxygen fugacities, activities of other components, multiple
mechanisms of diffusion, and crystal chemistry (‘coupling’ of the substitution of elements
into different crystallographic sites). Additionally, the extraction of meaningful ages, durations
of events, and temperatures requires considerable petrologic and contextual information to
place a diffusive interval correctly into the petrologic history of a rock. Bulk-rock scale
interpretations of mass transport occurring through an intergranular medium require additional
assumptions about metamorphic porosity and tortuosity, the nature of the intergranular

1943-2666/17/0083-0004$05.00 (online)
104 Kohn & Penniston-Dorland
medium, the diffusivity of the element or isotope through that medium, the host minerals of
the element / isotope, and whether host minerals precipitate, dissolve, recrystallize, or remain
inert except for diffusion during mass transfer.
In this chapter we highlight mineralogical and petrologic factors influencing diffusion and
the interpretation of diffusion profiles. The chapter is divided into 3 sections:
• Part 1. Diffusion theory and controls on diffusivity, including the effects of crystal
chemistry on inhibiting or enhancing the mobility of atoms within minerals.
• Part 2. Models of diffusive processes relevant to petrochronology to predict how boundary
conditions and mechanisms of diffusion may affect composition profiles.
• Part 3. Examples of natural systems, emphasizing how assumptions about diffusive
processes, particularly boundary conditions, affect interpretations for a variety of
applications.
We explore examples where the effects of diffusion impede our understanding of the rock
record as well as examples where we can exploit the process of diffusion to illuminate the
duration and rates of geological processes. We emphasize concepts that have guided our own
research, in hopes that these principles in turn guide future petrochronologic studies. One theme
that runs through the examples is that understanding how chemical or isotopic compositions
have changed on a mineral surface (the boundary condition) proves singularly challenging, and
that differences in assumptions about boundary conditions can give rise to radically different
interpretations of petrogenetic and tectonic processes. Getting the boundary condition right is
central to accurate petrochronologic interpretations. Of similar importance is understanding the
initial conditions, especially temperature. Because diffusion is thermally activated, an error in
temperature of only 25 °C can easily change calculated timescales by a factor of 2.
PART 1: DIFFUSION THEORY AND CONTROLS ON DIFFUSIVITY

This section reviews principles of diffusion as well as atomic-scale processes that influence
diffusion rates: Fick’s laws, multicomponent diffusion, and the dependencies of diffusion rates
on crystal-chemical controls and defects (including the influence of oxygen fugacity, water
fugacity, and strain).
Fick’s laws
Fick’s first and second laws (written here in simple one-dimensional form) underlie the
quantitative interpretation of the effects of diffusion in minerals and rocks:
∂Ci (1)
J i = − Di
∂x
∂Ci ∂ 2C (2)
= Di 2 i
∂t ∂x
These equations relate the flux and chemical concentration of species i (Ji, Ci), time (t),
diffusivity (Di) and distance (x). Diffusion of elements or isotopes within mineral grains can
result in loss (or gain) of these constituents, creating smoothly varying compositions across
mineral grains that represent ‘frozen-in’ diffusion profiles. The spatial scale of these features
increases over time as constituents diffuse. Measuring compositions along profiles across such
grains can provide constraints on the duration and rates of metamorphic and igneous processes
(e.g., Ganguly 2002; Costa et al. 2008; Chakraborty 2008).
Diffusion: Obstacles and Opportunities in Petrochronology 105
Ultimately, estimates of diffusivity, whether experimental or empirical, rely on solutions

to Equations (1) and (2). Strictly speaking, chemical fluxes respond to a chemical potential
gradient (dµ / dx), which can differ from concentration gradient (dC / dx) due to non-ideal
thermodynamic interactions (activity coefficients), and to cross-coupling among multiple
interdiffusing species. Ganguly (2002) develops requisite theory more comprehensively, and
his work serves as an example of how Equations (1) and (2) may be modified for chemically
complex, thermodynamically non-ideal mineral systems. Indeed, chemical potential can be
important in designing experiments and understanding experimental results (Watson 1982),
especially if it controls defect densities (Jollands et al. 2014). Fortunately, many questions
in petrochronology involve diffusion of trace elements, which should follow Henry’s law
(constant activity coefficient). Thus, commonly the chemical potential and concentration
gradients are proportional to each other (Watson and Baxter 2007; however see also Costa et
al. 2003). How chemical potential influences defect densities in natural crystals presents a new
direction for future research (Jollands et al. 2014).
The Arrhenius relationship describes the diffusivity (D) as a function of temperature (T, in K):
 −E  (3)
D = D0 ·exp  
 RT 
where D0 is a pre-exponential constant, E is activation energy (which is commonly taken to be
constant, but can be slightly pressure dependent), and R is the gas constant (see Zhang (2010)
for a review of the empirical and theoretical foundations for this relationship). Although
different temperature intervals may be characterized by different D0–E values for a specific
ion in a specific mineral, reflecting the influence of intrinsic vs. extrinsic defects, most data
are interpreted using a single Arrhenius expression, i.e., single D0 and E values. Brady and
Cherniak (2010) tabulate experimental estimates of D0 and E for a wide range of minerals
and diffusing species. The exponential dependence of D on T means that accurate estimates of
temperature are needed for accurate interpretations.
Multicomponent diffusion
Diffusion of an element within a mineral can be affected by diffusive properties of other
elements within the same mineral. When interdiffusion occurs among at least 3 elements, a
substantially different D for one element can cause so-called “uphill” diffusion, or diffusion
against a concentration gradient. This counterintuitive process can occur even when components
mix ideally, as a consequence of cross-linkages among diffusing species—decreases in
concentration of one species must be balanced by increases in another (e.g., Lasaga 1979;
Ganguly 2002, 2010; Krishna 2015). In three-component systems, Equation (1) expands to a
series of two coupled equations:
 
 J1   D11
D12    (4)
  = −   
 J2   D21
D22   C2 
 
where the D’s are diffusion coefficients that take into account all chemical components in the
system of interest, including ideal and non-ideal interactions. A flux equation is not needed
for component 3 because the fluxes must sum to zero, so J3 is a dependent variable. Taking
Fe–Mg–Ca interdiffusion in garnet, with Fe as component 1, Mg as component 2, and Ca
as the dependent component, −DFeFe·dCFe/dx represents the diffusive response of Fe to its
own concentration gradient while −DFeMg·dCMg/dx represents the diffusive response of Fe to
the concentration gradient of Mg. Different expressions have been proposed for the D’s in
Equation (4), as referenced to either self-diffusion or interdiffusion coefficients. Again taking
Fe–Mg-Ca interdiffusion in garnet, Equation (4) can be expressed in terms of Maxwell–Stefan
interdiffusion coefficients1 (D’s, Krishna 2015) as:
1
Maxwell–Stefan interdiffusion coefficients in Equations (5) and (6) differ from the Fick matrix D’s in Equation (4) and should
not be equated. For example, Maxwell–Stefan D12 = D21 whereas in general Fick matrix D12 ≠ D21 (see example in Eqn. 7).
 dX Fe 
 J Fe  1 DFeCa  X FeDMgCa + (1 − X Fe ) DFeMg  X FeDMgCa DFeCa − DFeMg   
  dx  (5)
 = −
 J Mg  d X MgDFeCa DMgCa − DFeMg  DMgCa  X MgDFeCa + (1 − X Mg ) DFeMg    dX Mg 
  
 dx 
where X is mole fraction and:
d = X MgDFeCa + X FeDMgCa + X CaDFeMg (6)
Although the diagonal terms, D11 and D22 in Equation (4), are always positive (a component
tries to diffuse down its own concentration gradient), the off-diagonal terms, D12 and D21,
can be negative, as they are in garnet because DFeCa and DMgCa are smaller than DFeMg (e.g.,
Ganguly et al. 1998; Vielzeuf et al. 2007; Vielzeuf and Saúl 2011). This means that for
garnets that contain Ca, a steep gradient in Fe would tend to drive backwards diffusion of
Mg, while a steep gradient in Mg would tend to drive backwards diffusion of Fe. Some garnet
interdiffusion experiments do show clear uphill diffusion (Vielzeuf et al. 2007), and for one of
these (Fig. 1A), Vielzeuf and Saúl (2011) inferred coefficients for Equation (4) of:
 dX Fe 
 J Fe   5.86 x10 −19 −1.02 x10 −19   dx 
  = −   (7)
−19
 J Mg   −5.50 x10 1.18 x10 −19   dX Mg 
 dx 
Uphill diffusion of Mg in this experiment resulted from a large initial gradient in Fe

[abs(dXFe /dx)  0] and a small initial gradient in Mg [abs(dXMg / dx) ~ 0], such that the off-
diagonal term, DMgFe (= −5.50 × 10−19 m2/s), dominated Mg diffusion. Diffusion of Mg also
affected Fe (DFeMg < 0), but the large initial Fe gradient and small initial Mg gradient mean
that the effect is not obvious (Fig. 1A). More generally, uphill diffusion is apparent when
strong gradients occur in two components with different D’s (in this case Ca and Fe), and a
third component has a very small initial composition gradient (Mg). Carlson (2006) discusses
a natural example where Ca in garnet may have diffused against its concentration gradient.
Gradients in major element compositions can also influence diffusion rates of other
elements, including trace elements, because major elements dominate chemical potentials
(Costa et al. 2003). In plagioclase, for example, isothermal models of the diffusive interaction
across a step function in anorthite content (inset, Fig. 1B) and a corresponding step function
in trace Mg predict uphill diffusion of Mg (Fig. 1B). This unusual diffusive behavior occurs
for Mg even though, under dry conditions, the anorthite content is essentially inert to diffusive
modification (e.g., Grove et al. 1984), and Mg2+ simply exchanges for Ca2+. The anorthite
content influences the activity of the Mg-component in plagioclase, which in turn affects the
off-diagonal terms in Equation (4) (Costa et al 2003). Modeling that does not account for the
zoning in anorthite content (lower inset, Fig. 1B) shows quite different profiles. Consequently,
any interpretation of Mg zoning in plagioclase would have to account for these non-ideal terms.
Crystal-chemical controls on diffusion
Fully understanding cation diffusivities requires close consideration of crystal chemistry
and cation substitution mechanisms. While some elements of chronologic or petrologic
significance can diffuse via simple cation exchange (e.g., Pb2+ ⇔ Ca2+ in titanite, U4+ ⇔ Zr4+
in zircon, Fe2+ ⇔ Mg2+ in garnet, etc.), others require coupled exchanges (e.g., VIIISm3+ VIMg2+
⇔ VIIICa2+ VIAl3+ in garnet, U4+ Ca2+ ⇔ 2Ce3+ in monazite, Yb3+ P5+ ⇔ Zr4+ Si4+ in zircon,
etc.). As a very general rule of thumb, low valence cations in larger crystallographic sites
diffuse faster than high valence cations in small sites (e.g., Brady and Cherniak 2010). For a
Fig. 1 Kohn and Penniston-Dorland
60
A Uphill diffusion
Mg
XGrs, XAlm, XPrp
40 Fe
Garnet
(experiment) Figure 1. Examples of multicomponent dif-
20 Ca fusion phenomena. (A) Experimental garnet
diffusion couple, showing uphill diffusion of
Mg due to strong initial gradients in Ca and
0 Fe. Original Mg composition shown by dashed
-4.0 -2.0 0.0 2.0 4.0 lines. Modified from Vielzeuf and Saúl (2011).
Distance (µm) (B) Models of Mg diffusion in plagioclase. An
initial step function in anorthite (upper inset)
is correlated with an initial step function in
400 80 Mg (t0). Relaxation of the Mg profile (t1, t2)
B Plagioclase towards equilibrium (t∞) develops oscillations
(model) 40 and shows uphill diffusion because of ther-
300 An (%) modynamically non-ideal interaction between
t0
Mg (ppm)
the anorthite component and Mg. Lower inset

shows how diffusion would proceed without
200 Uphill effect of zoning in anorthite component. Mod-
t1
diffusion ified from Costa et al. (2003).
t2
100
t∞ t0 Mg 300
0
-30 -20 -10 0 10 20 30 100
Rim
Distance (µm) t∞
fixed valence, smaller cations generally diffuse faster than larger cations (Brady and Cherniak
2010), although such a simple relationship does not always hold for melts (Zhang 2010).
Chronologically useful radioactive isotopes and their daughters tend to have different ionic
radii and different ionic charges than the dominant cation for the site in which they substitute.
So, diffusion of these elements tends to be different than expected from ionic radius and site
characteristics alone. For example, REE3+ diffuse more slowly in garnet than divalent cations,
not necessarily because ionic radii are so different (Ca2+ = 1.12 Å, Fe2+ = 0.92 Å, Lu3+ = 0.98 Å,
Nd3+ = 1.11 Å), but because trivalent cations in the cubic site must be charge coupled to
3 +
other cations, e.g., VIIIREE + VIIINa ⇔ VIIIMg2+ VIIIMg2+, VIIIREE3+ VIMg2+ ⇔ VIIIMg2+ VIAl3+,
VIII 3+ IV 3+ VIII 2+ IV 4+
or REE Al ⇔ Mg Si (Fig. 2A; Carlson 2012). Charge coupling involving
vacancies may also influence diffusivity (Jollands et al. 2014).
Charge coupling readily explains the slow diffusion of Li in garnet at a rate comparable
to Y and Yb (Cahalan et al. 2014). Ordinarily, one might expect a monovalent cation to diffuse
faster than divalent or trivalent cations in the same site (Brady and Cherniak 2010). Cahalan et al.
(2014), however, show that natural diffusion profiles of Li and REE in garnet are indistinguishable
in form and ~6 times shorter (so D is lower) than for Fe, Mg, Mn and Ca (Fig. 2B). Coupled
substitution in the dodecahedral site of Li+ and REE3+ for 2(Mg, Fe, Mn, Ca)2+ may limit its
diffusivity, so that while Li+ diffusion might be fast when exchanging with other monovalent
cations, charge coupling to the slow-diffusing REE may constrain DLi to equal DREE (Cahalan et
al. 2014). In contrast, charge coupling should not impede diffusion of Li isotopes because 7Li can
exchange by a simple mechanism with 6Li. Consequently, Li isotopic gradients may be erased
at high temperature, even while Li concentration gradients are preserved. Slow diffusion due to
Figure 2. (A) Crystal structure and

chemistry of garnet projected along
a-axis showing tetrahedral, octahe-
dral, and distorted cubic sites. Detail
increases from left (tetrahedral sites
only, showing 4-fold screw axes)
to right (all 3 sites). Cation sizes il-
lustrated around two cubic sites
on right-hand side. Dashed lines
delineate unit cell. Geochemically
Mg, Fe2+, Mn, Ca, (Na, REE) Proposed REE Substitutions important cation substitutions and
REE3+ + VIIINa+ = 2VIIIMg2+
VIII
REE substitution mechanisms are
Al, Fe3+, (Mg)
REE3+ + VIMg2+ = VIIIMg2+ + VIAl3+
VIII
listed. Crystal structure image based
Si, (Al) REE3+ + IVAl3+ = VIIIMg2+ + IVSi4+
VIII
on http://www.uwgb.edu/dutchs/
A Four-fold screw axes petrology/Garnet Structure.HTM.
(B) Evidence for slow diffusion of
3.0 Li and REE (short compositional
76 B gradients; coupled substitutions) vs.
Li, Y (moles x 1000)
Garnet fast diffusion of divalent Fe (long

74 compositional gradient; simple sub-
2.0 stitution) in natural garnet from the
72
Makhavinekh Lake Pluton contact
XAlm
70 aureole, Labroador. Dashed lines

Fe 1.0 show assumed initial compositions.
Li Modified from Carlson (2006) and
68
Y Cahalan et al. (2014).
66
0.0
80 60 40 20 0
Core Distance (µm) Rim
charge coupling is also well known from studies of feldspars. At 800 °C interdiffusion of alkali-
site Na+ for Ca2+ in plagioclase (charge coupled to tetrahedral site Si4+ and Al3+) is ~10 orders
of magnitude slower than simple Na+ and K+ interdiffusion in alkali feldspar (Brady 1995).
Thus, in plutonic plagioclase, Ca isotopes might reset to low-temperature (through diffusional
self-exchange with other Ca atoms), while the anorthite content is virtually invariant and retains
information about high-temperature crystallization conditions.
Some species may be distributed among different sites, which may influence their diffusive
behavior. For example, Li in olivine appears to occupy both an octahedral site and an interstitial
position (Dohmen et al. 2010). Interpretation of Li diffusion profiles in olivine therefore requires
more complex models of diffusive transfer. Diffusion of interstitial Li is faster than octahedral
Li (Dohmen et al. 2010), and published data on Li distributions in olivine (e.g., Jeffcoate et
al. 2007; Parkinson et al. 2007; Rudnick and Ionov 2007) suggest that the slower Li diffusion
mechanism dominates in nature. This diffusion rate is, on average, an order of magnitude higher
than diffusion of Fe and Mg in olivine (Dohmen et al. 2010), so the diffusion of Li provides
information about shorter duration events and possibly processes occurring at lower temperature
than diffusion of Fe and Mg. Similarly, a two-species model of lithium diffusion appears to
operate in clinopyroxene (Richter et al. 2014), such that Li isotopic compositions take longer to
homogenize than Li concentrations. Other elements also have multiple substitution mechanisms
(e.g., Si in rutile, Golden et al. 2015) that presumably convey complex diffusive behavior.
Dependence of D on defects
Ultimately, diffusion depends on defects in the crystal structure, including thermal defects,
defects arising from chemical substitutions, and strain defects or dislocations. To illustrate the
importance of defects, Chakraborty (2008) draws an analogy to a parking lot: a fully packed
parking lot allows no movement of automobiles, whereas at least one vacancy (defect) allows
movement. Thermal defects are point defects including vacancies and interstitialcies (atoms
that occupy positions interstitial to the ideal crystallographic sites). Point defects might
involve multiple neighboring sites or interstitial positions, but do not extend into lines or
planes. Thermal defects form spontaneously and increase in abundance at higher temperature,
causing an increase in diffusion rates. Referring to the parking lot analogy, a greater number of
open vacancies, coupled with more vigorously active drivers, allows greater mobility. Thermal
defects are referred to as intrinsic defects because their presence is intrinsic to any compound,
no matter how chemically pure. Substitutional defects are associated with substitution of
aliovalent ions2, for example Cr3+ substitution for Mg2+ in olivine, Fe3+ substitution for Ti4+ in
rutile, etc. Such defects are also point defects and can depend on parameters such as f H2O or
fO2, which affect interstitial(cy) concentrations (e.g., H+ or OH−) and valence state, especially
Fe3+. Substitutional defects are referred to as extrinsic defects because their concentrations are
dependent on the chemical potentials of the substituting atoms or compounds. Strain defects,
more commonly called dislocations, are distortions to the crystal lattice caused by deformation
independent of crystal chemistry. Dislocations occupy lines or planes within a crystal and can
close upon themselves, forming dislocation loops.
fO2 dependence. Minerals whose cation diffusion rates appear to depend on fO2 include
olivine (see summaries of Dohmen et al. 2007; Dohmen and Chakraborty 2007; Chakraborty
2010), rutile (Sasaki et al. 1985), pyroxenes (see Cherniak and Dimanov 2010; Dohmen et
al. 2016), garnet (Chakraborty and Ganguly 1992) and sodic feldspar (see Cherniak 2010a).
These studies reveal two fundamental principles. First, oxidation state can influence defect
densities by altering the valence of transition metals. The oxidation of Fe2+ to Fe3+ with
increasing fO2 can create octahedral site vacancies, whereas the reduction of Ti4+ to Ti3+ with
decreasing fO2 can create oxygen vacancies (although interstitial cations appear to dominate
diffusion rates in rutile for several elements; Sasaki et al. 1985; Nowotny et al. 2008). Thus,
high fO2 for olivine, pyroxene, garnet, etc. and low fO2 for rutile are expected to induce higher
defect densities and faster diffusion. This concept helps reconcile otherwise disparate cation
diffusion rates determined experimentally in garnet (Chakraborty and Ganguly 1992), olivine
(Dohmen et al. 2007), and pyroxene (Dohmen et al. 2016). Second, the importance of fO2-
induced defects diminishes whenever the substitution of aliovalent cations produces a greater
proportion of total vacancies. Substitutional defects are expected to dominate diffusion in
ferro-magnesian silicates at low fO2 (low Fe3+ concentration), or in rutile at high fO2 (low
Ti3 + concentration). A change in diffusion rates with fO2 was proposed (Chakraborty 1997)
and observed experimentally (Dohmen et al. 2007), with faster diffusion dominated by defects
resulting from fO2-dependent transition metals and slower diffusion dominated by defects
resulting from fO2-independent aliovalent cations.
Exceptions to these principles include diffusion rates of Fe in labradorite (Behrens et al.
1990) and Eu in orthopyroxene (Cherniak and Liang 2007), which are both lower at higher
fO2. This phenomenon probably reflects slower diffusion for more highly charged ions (Fe3+
and Eu3+ rather than Fe2+ and Eu2+), where oxidation of a minor or trace constituent does
2
An aliovalent (a.k.a. heterovalent) ion has a different charge than the ion for which it substitutes in the
endmember compound. Substitution of aliovalent ions forms defects only when charge balance is achieved
through vacancies rather than coupled substitution of cations on other sites. For example, substitution of aliova-
lent Ca2+ for Na+ in albite does not normally induce vacancies because it is charge balanced by substitution of
Al3+ for Si4+ to form the plagioclase solid solution.
not change defect densities appreciably. No fO2 dependence was observed for Zr and Hf
diffusion in natural rutile (Cherniak et al. 2007a), for O and Si in forsterite (Jaoul et al.
1981, 1983), or for Fe–Mg interdiffusion in pyroxene (Müller et al. 2013) likely because
substitutional defects dominate. In this respect, just because a ferro-magnesian mineral
crystallizes at high fO2 does not mean it will necessarily have a different diffusion rate than
the same mineral that crystallizes at low fO2. This is because, for crystallization at high
fO2, charge balance likely occurs due to coupled substitution involving cations rather than
vacancies. For example, in pyroxenes, substitution of M1-site Fe3 + for Mg2+ at higher fO2
would likely be charge balanced by M2-site substitution of Na+ for Ca2+ or Mg2+, or by
T-site substitution of Al3+ for Si4+, not by vacancies. Nonetheless, as fO2 changes during
cooling along an fO2 buffer, in situ oxidation-reduction of transition metals should change
defect densities and diffusion rates. Thus, diffusion parameters for minerals should be
reported and applied along an fO2 buffer (e.g., Chakraborty and Ganguly 1992).
fH2O dependence. Oxygen diffusion in minerals commonly shows much larger D’s and
smaller E’s in water- or mixed-fluid-present experiments at 0.05–1 GPa than when PH2O < 1
bar (e.g., Zhang 2010; Brady and Cherniak 2010; Farver 2010). In general, these different
D’s are referred to as “wet” vs. “dry”. Enhanced D at high PH2O or f H2O is usually ascribed
to rapid diffusion of trace molecular H2O in the mineral as interstitial defects combined
with exchange between the H2O diffusant and lattice oxygen (Doremus 1969; Zhang et al.
1991; Zhang 2010; Farver 2010). Materials with higher molecular H2O contents and H2O
diffusivities show higher overall O diffusivities (Zhang et al. 1991). Peck et al. (2003) and
Bowman et al. (2011) proposed that the physical presence of an aqueous phase is necessary
to enhance D. However, if an equilibrium reaction can be written between molecular H2O
in a fluid phase and in a mineral, e.g., H2O(aqueous) = H2O(mineral), Henry’s Law would require
that a mineral’s trace molecular H2O content should be proportional to f H2O. Even if a fluid
is not present, most mineral assemblages contain hydrous minerals that should buffer f H2O to
high values and impart relatively high H2O contents in minerals (Kohn 1999). By analogy,
f H2O controls mineral “water” content (actually H+ and OH−) in nominally anhydrous
minerals through thermodynamic equilibria (e.g., Keppler and Bolfan-Casanova 2006), even
when water is not present, for example in experiments investigating partitioning between
nominally anhydrous minerals and basaltic melt (Aubaud et al. 2004).
Cation diffusion rates can also increase slightly with increasing f H2O. Silicon diffusion in
olivine and wadsleyite appears to occur via Si-vacancy complexes, and the abundance of these
vacancies increases with increasing H+ and f H2O (Costa and Chakraborty 2008; Shimojuku
et al. 2010; Fei et al. 2013). Increasing H+ concentration at higher f H2O also increases Fe–Mg
interdiffusion rates in olivine (Wang et al. 2004; Hier-Majumder et al. 2005) and in Fe-Mg oxide
(Demouchy et al. 2007) because interstitial protons cause an increase in octahedral site vacancies.
Strain dependence. Logically, if higher defect concentrations increase diffusion rates, and
deformation increases defect density, deformation should increase diffusion rates. If so, two
different states may be considered (Cohen 1970): dislocation-assisted diffusion occurs when
prior deformation has created dislocations, whereas strain-enhanced diffusion occurs when
dislocations form simultaneously with diffusion. Whether deformation significantly enhances
diffusion rates has proved surprisingly difficult to demonstrate. Oxygen diffusion in albite is
~5 times faster in pre-deformed vs. non-deformed crystals (Yund et al. 1981) whereas ongoing
deformation does not measurably affect Na–K interdiffusion in albite-adularia diffusion couples
(Yund et al. 1989). However, disordering rates of tetrahedral Al–Si in feldspar do increase with
increasing strain and strain rate (Yund and Tullis 1980; Kramer and Seifert 1991).
The effect of strain on trace elements in zircon has recently received close scrutiny.
Commonly, zircons are first imaged using cathodoluminescence and electron back-scatter
diffraction to map out chemical patterns and to identify sub-grain misorientations, which
result from deformation. Ion probe or atom probe analysis then allows comparison of the
chemistry of deformed vs. non-deformed regions or of dislocations themselves. Most studies
show increased concentrations of REE, U, and Th, and decreased concentrations of Ti and Pb
in regions containing dislocations (Reddy et al. 2006, 2016; Timms et al. 2006, 2011; Piazolo
et al. 2012, 2016; MacDonald et al. 2013). Such compositional effects will bias trace element
patterns, U–Th–Pb ages, and trace element thermometers, so understanding mechanisms of
chemical change is potentially important to petrochronologic research. Atom probe studies
now demonstrate that anomalously high and low concentrations spatially correlate with
dislocations, especially along sub-grain boundaries (Reddy et al. 2016; Piazolo et al. 2016).
Why would trace elements accumulate at dislocations? For reasons discussed in detail
elsewhere (see chapters 5 and 7 of Sutton and Balluffi 2003), diffusion of solutes from the crystal
interior to grain surfaces typically decreases free energies. This behavior can be understood either
from the macroscopic thermodynamics of grain boundaries (Gibbs 1874–1878; Cahn 1979) or
from the perspective that grain boundaries can readily accommodate a wider range of cations
than a crystal lattice, which has only a few specific sites. Consequently, a trace constituent that
fits poorly in the crystal lattice (higher free energy) might find a more compatible bonding
position along a grain boundary (lower free energy). Thus, creation of new sub-grain boundaries
during deformation should scavenge trace elements (just as grain coarsening might liberate trace
elements to form new accessory mineral grains; Corrie and Kohn 2008).
How do trace elements accumulate at dislocations? Accumulation could occur either as
dislocations sweep through a crystal and coalesce into sub-grain boundaries, or via diffusion
through the crystal lattice after dislocations become stationary (Piazolo et al. 2016; Reddy
et al. 2016). For interstitial cations, elastic modeling predicts that they will diffuse towards
dislocations (Cottrell and Bilby 1949), forming a “Cottrell atmosphere” of enhanced trace
element concentration. Examples in zircon may include interstitial Be, Mg, and Al (Reddy
et al. 2016). This process tends to pin dislocations, increasing the resistance of the solid to
further deformation. However, a Cottrell atmosphere does not explain high concentrations of
substitutional cations at dislocations, such as Y, so other factors have been invoked, including
diffusion of combined oxygen vacancies plus aliovalent cations (Reddy et al. 2016).
After migration of a linear or planar dislocation has ceased, fast diffusion along the
dislocation may continue, resetting trace element thermometers and U–Th–Pb ages in the
immediately adjacent host crystal (Piazolo et al. 2016). Fast diffusion along grain boundaries
also implies that some trace elements derive from the matrix, not solely from the host crystal.
Conversely, matrix minerals must serve as the host for trace elements whose concentrations
decrease in the host crystal, such as Ti in zircon.
PART 2: PETROCHRONOLOGIC CONCEPTS

This section reviews principles of diffusion that are directly relevant to petrochronology:
closure temperature, geospeedometry, diffusion through porous media (fluid–rock interactions),
major and trace element thermometry, reaction rates, and resetting of isochrons.
Closure temperature
Closure temperature (Tc) is usually defined in reference to thermochronologic data—there
is a transition temperature, Tc, above which virtually all radiogenic daughter (d*) is lost through
diffusion, and below which virtually all radiogenic daughter is retained; the age of a mineral
reflects the time (t) at which it passed through that transition temperature (Dodson 1973; Fig. 3A).
Following more closely the derivation of Dodson (1973; Fig. 3A), closure temperature reflects
the fact that, due to an exponential decrease in D with decreasing T (Eqn. 3), a mineral transitions
from essentially maintaining equilibrium with an imposed boundary condition throughout
the entire crystal, to adopting a composition that is independent of changes to the boundary
8.0
Tc(Qtz) Final Qtz
Qtz (35%)
Temperature
Pl (21%) composition
Tc Ms(8%)
St(9%)
Grt(6%) Qtz Metapelite model
Bt(21%)
6.0
Pl
δ18O (‰)
Diffusivity
Ms
4.0
St
Grt
Final Bt
composition
concentration
2.0
Daughter
Bt
d* Bulk δ18O
lk
Bu
Surface δ18O Tc(Bt)
B
0.0 Surface d* A 0.0
600 500 400 300 200 100
high-T low-T
time Temperature (°C)
C Bt
concentration
re
co
Daughter
δ18O (‰)
core
rim
ar-
ne near-rim
0.0 Surface d* D
high-T low-T high-T low-T
time time
Figure 3. Closure temperature concepts and models. Black arrows show forward (normal time) progression
of temperature, diffusivity and composition. Red arrows show backwards projections. (A) Radiogenic iso-
topes. At high temperature, D is large, and no radiogenic daughter accumulates—the bulk crystal maintains a
concentration of 0, equivalent to the fixed boundary condition. At low temperature, D is small, and radiogenic
daughter accumulates according to the age equation (approximately linearly for short times), independent of
the boundary condition. Tc is defined as the projection of the trace of accumulated d* vs. time to the bound-
ary condition (red arrow). Modified from Dodson (1973). (B) Stable isotope model of a typical metapelite
(modified from Kohn and Valley 1998). Model cooling rate was 5 °C/Ma, grain sizes were 1 mm diameter,
and diffusion rates under hydrothermal conditions were assumed. In this example, staurolite (St) and garnet
(Grt) crystallize below their closure temperatures, whereas quartz (Qtz), plagioclase (Pl), muscovite (Ms),
and biotite (Bt) crystallize above their closure temperatures and their oxygen isotope compositions evolve
during cooling. At high temperature the bulk d18O values of these minerals (excepting staurolite and garnet)
track their respective boundary conditions. At low temperature, isotope compositions become invariant. Tc
for each mineral is defined as the point on the boundary condition at which surface composition matches
final composition [i.e., the projection of the final composition (dots on each mineral trajectory) to the surface
trace; red arrows illustrate the concept for quartz and biotite]. These points are mathematically equivalent to
the analytical expression of closure temperature defined by Dodson (1973), and do not depend on mineral
assemblage. (C, D) Position-dependent closure temperature for radiogenic and stable isotope systems. Cores
of crystals close earlier and reflect higher Tc’s than positions closer to rims.
condition. Geometrically, Tc is then defined as the projection of the bulk mineral composition vs.
t onto the trace of its boundary composition vs. t (Dodson 1973; Kohn and Valley 1998), with a
conversion from time to temperature according to the cooling history (Fig. 3A). From this broader
perspective, the thermochronologic Tc represents the projection of the accumulation of daughter
(d*) vs. t onto the boundary condition, d* = 0 (Fig. 3A). This projection corresponds with the
measured age, t, of the mineral. The advantage of the broader definition of Tc is that it applies
to all diffusion-controlled systems, irrespective of diffusing species, mineral assemblage, etc.
(Fig. 3B; Kohn and Valley 1998). This definition also more meaningfully reveals the significance
of boundary conditions in defining closure for minerals. Closure temperature has also been used
to describe the apparent stable isotope temperature recorded between two minerals, but normally
this is referred to as an apparent temperature, or Tapp.
Mathematically, Tc for an entire crystal is (Dodson 1973):
E/R
Tc =
 ART 2 D / a 2  (8)
ln  c 0 
 E ⋅ dT 
 dt 
where R is the gas constant, A is a constant reflecting the geometry of diffusion (8.7 for a plane
sheet, 27 for a cylinder, and 55 for a sphere), a is the radius or half-width of the crystal, and dT / dt
is the cooling rate. The equation must be solved iteratively, and Tc is fixed for a given geometry,
crystal size, cooling rate, E and D0. Although Equation (8) was developed for radiogenic systems,
it applies even when boundary conditions exhibit complex behavior, including stable isotopic
systems (Kohn and Valley 1998; Fig. 3B)—closure depends on the intrinsic ability of a mineral
to respond to changes in its boundary condition, not on how that boundary condition changes. For
example, although muscovite and plagioclase boundaries can exhibit compositional inflections
and even oscillations, the closure temperature as evaluated numerically is indistinguishable from
the closure temperature as calculated using Equation (8) (Fig. 3B; Kohn and Valley 1998).
In principle, the composition of a mineral can be used to constrain cooling rate because Tc
and hence composition depend on cooling rate (Eqn. 8). For example, d18O of quartz in a schist
will be higher in a more slowly cooled rock (lower Tc) and lower in a more rapidly cooled rock
(higher Tc; Giletti 1986). Because cooling rate appears as a logarithmic term in Equation (8),
however, cooling rates estimated this way carry extremely large uncertainties—a large change
in cooling rate normally does not change Tc or composition much.
Geospeedometry
Closure can also be considered as a condition that initiates in the core of a crystal and
sweeps outward as temperature decreases with time. Because the core is farthest from the
boundary condition, it closes earliest and at the highest temperature whereas points closer to
the rim of a crystal close later and at lower temperatures (Fig. 3C–D). Dodson (1986) modeled
the position-dependent closure temperature for slow cooling:
E/R
Tc ( x ) =
 γ RT 2 D / a 2  (9)
ln  c 0  + 4 S2 ( x )
 E·dT 
 dt 
where x is fractional position relative to the crystal center (0 at the center, 1 at the rim), g is
the exponential of Euler's constant, and 4 S2 (x) is a position-dependent term (smaller towards
the core, larger towards the rim). The bulk Tc for a crystal (Eqn. 8) is simply Equation (9)
integrated over the volume of the crystal. Equation (9) implies that age or compositional
zoning in a mineral can provide useful information about the cooling history, and forms a
conceptual foundation for geospeedometry, or the determination of cooling rates.
One assumption in deriving Equation (9) is that initial temperatures are sufficiently high and
cooling rates sufficiently low that the core of a crystal is able to equilibrate with the boundary
condition imposed at the crystal surface. The parameter M describes this characteristic:
= DT0 τ / a 2
M (10)
where DT0 is the diffusivity at maximum temperature, t is the time required for D to drop by
a factor of E (which depends on the cooling rate), and a is the characteristic length scale of
the crystal. Thus M1 / 2 is the characteristic diffusion distance at T0 divided by the characteristic
length scale of the crystal. Large values of M (> 1) lead to Equation (9). Ganguly and Tirone
(1999) developed an analytical expression, evaluated numerically, to describe systems with
M ≤ 1, where the core of the crystal remains relatively inert to diffusive resetting:
E/R
Tc ( x ) =
 γRT 2 D / a 2  (11)
ln  c o  + 4 S2 ( x ) + g ( x )
 E·dT 
 dt 
and g(x) is a correction term for small M.
Although analytical expressions have been developed to invert zoning profiles in minerals
so as to calculate cooling histories and relate them to tectonic processes (e.g., Lasaga 1983;
Ganguly et al. 2000), many researchers employ numerical methods. Most use a Markov-chain
Monte Carlo (MCMC) approach, which differs from classic Monte Carlo (totally random)
approaches in requiring that temperature continually decrease for each time increment. That
is, for each new time step (ti + 1), a new temperature (Ti + 1) is chosen randomly between the
temperature of the previous time step (Ti at ti) and zero, i.e., Ti > Ti + 1 > 0. A model diffusion
profile is calculated for each random cooling history, compared with the observed profile, and
goodness of fit criteria calculated. An assemblage of T–t paths is then contoured for quality of
fit (e.g., Ketcham 2005; Gallagher 2012).
A key outcome of modeling and sensitivity tests is that inversion of chronologic zoning is
inherently more precise for constraining cooling rates than inversion of temperature-sensitive
chemical zoning (Kohn 2013, 2016; see also Lindström et al. 1991). Here we show this using
Equation (9) to model chemical and chronologic diffusion profiles at constant 25 °C / Ma cooling,
then re-calculate temperatures and times resulting from perturbations in E commensurate with
typical diffusion parameter uncertainties (Fig. 4). We assume E = 250 kJ and D0 = 1 × 10−9 m2 / s,
yielding a Tc of ~ 700 °C in the core of a 100 µm-radius spherical crystal. These values are similar
to Pb diffusion in rutile (Cherniak 2000). An uncertainty in both E and log10(D0) of ±10% (2s)
is assumed, well within typical 2s variations between ~ ±4% to ~ ±40% (e.g., see Brady and
Cherniak 2010). Because D0 correlates almost perfectly with E, a 10% increase or decrease in E
is assumed to increase or decrease log10(D0) by 10%.
For the chronologic zoning model, each point in a crystal has a measured age that
corresponds to the position-dependent closure temperature of the radiogenic daughter. That
is, we know Dt from direct chronologic measurements on at least 2 different positions in the
crystal, and we use Equation (9) with different values of E (and D0) to solve for DTc at the
same positions. Even accounting for correlated changes to D0, calculated Tc’s shift upward for
higher E, and downward for lower E. Thus, variations in E (and D0) shift the entire cooling
history to higher or lower temperatures (vertical double-arrow in Fig. 4), but with virtually no
change to either DTc or the cooling rate (24–26 °C / Ma). That is, chronologic zoning retains
information about cooling rate, but with some ambiguity about the specific temperature interval.
For chemical zoning, each point in a crystal has a measured composition that directly
defines Tc. For example, if diffusion zoning of Zr in a rutile crystal reflects resetting of the
Zr-in-rutile thermometer, each position-dependent Zr concentration represents a position-
dependent Tc. But Tc depends on cooling rate, too (Eqn. 9). So, if we change E (and D0), we
must change the cooling rate to obtain the same (measured) Tc within a crystal: cooling must
be slower with a higher E to obtain the same Tc and faster with a lower E (horizontal double-
arrow, Fig. 4). Even modest variations in E and D0 require very large changes to cooling
rate to obtain the same Tc (between ~2 and ~300 °C / Ma). This sensitivity occurs because
dT/dt appears in the logarithm of Equation (9). Whereas chemical zoning retains precise
information about temperature, it poorly constrains cooling rate (Lindström et al. 1991; Kohn
2016). A similar result was obtained from numerical models of major element zoning in
800
E=
Chronologic t1
275
zoning
t2
700
kJ
E=2
Temperature (°C)
E = 225 kJ 300 °C/Ma
26° C
75 k
J 2
°C /Ma
/Ma
600
Chemical
E= zoning
225
kJ
T1
500 T2
24 °
C/M
Input T-t path

E = 250 kJ
a
400
0 10 20 30 40 50
Time (Ma)
Figure 4. Sensitivity of cooling history inversions (geospeedometry) to uncertainty in diffusivities based
on chronologic vs. chemical zoning. Hexagons show schematic crystals with diffusion zoning in either
ages (t1, t2) or temperatures (T1, T2). Input T–t path is 25 °C/Ma, with an activation energy for diffusion
of 250 kJ. Perturbations of E by ±10% (±25 kJ, with commensurate changes to D0) shift cooling history
for chronologic zoning to higher or lower temperatures (vertical double-arrow), but retain basic T–t
shape with cooling rates between 24 and 26 °C/Ma. Same perturbations to E for chemical zoning shift
temperatures to younger or older ages (horizontal double-arrow), and cause radical changes to T–t shape
with cooling rates between ~2 and ~300 °C/Ma.
garnet, specifically an uncertainty in E of ~±15% (with corresponding error in D0) propagates

to an uncertainty in dT/dt of ~±1 order of magnitude (Spear 2014).
Diffusion in porous media: fluid–rock interaction
Mass transport through porous media, including diffusion, presents a large literature
which we can only briefly discuss here. For more comprehensive development of these
concepts, see Bear (1988). In geological applications, smoothly varying stable isotope
profiles at the hand sample to outcrop scale across lithologic discontinuities may reflect
chemical transport through rocks via both advection and diffusion, and provide constraints
on the duration of metamorphic fluid–rock interactions (e.g., Bickle and McKenzie 1987;
Bickle and Baker 1990; Baumgartner and Ferry 1991). Chemical transport between rocks
likely occurs through an intergranular medium because volume diffusion through the solid in
many cases is too slow at moderate metamorphic temperatures (Bickle and McKenzie 1987;
Cartwright and Valley 1991). The type of intergranular medium (dry grain boundaries, fluid,
or melt) will affect the diffusivity of the element or isotope, so it is important to consider
petrologic evidence for the type of medium likely involved in such transport.
Chemical and isotopic profiles collected across compositional boundaries can be modeled
using a one-dimensional solution to the mass continuity equation (Bickle and McKenzie 1987):
∂C ∂C ∂ 2C (12)
K e + νφ = De 2
∂t ∂x ∂x
where C is concentration, x is distance from the contact, t is time, v is the fluid velocity, f is the
porosity, De is the effective diffusivity, and Ke is the effective partition coefficient. Equation (12)
balances the time variation of concentration in the solid (1st term) and the effects of advection
(2nd term) against diffusion (3rd term). The effective partition coefficient takes into account the
porosity (f) of the rocks as well as the fractionation of the elements or isotopes into the solid or
fluid (Bickle and McKenzie 1987; Cartwright and Valley 1991, Baumgartner and Valley 2001):
ρs K c
=
Ke (1 − φ ) + φ (13)
ρf
where ρs and ρf are the density of the solid and fluid respectively and Kc is the partition
coefficient between the mineral and fluid (isotopic exchange is assumed to occur between
solid and an intergranular medium). In multi-mineralic rocks, a bulk fluid–rock partition
coefficient is determined by weighting each fluid–mineral partition coefficient according to
the abundance of that mineral and summing over all minerals. For most stable isotopes (e.g.,
oxygen, carbon, sulfur) the delta notation (e.g., d18O) may be used instead of the concentration
over relatively small variations (see Baumgartner and Rumble 1988; Baumgartner and Valley
2001). Equation (13) links composition profiles to the solid–fluid partition coefficient and
porosity (Fig. 5A, B). The effective diffusivity De is defined as (Bickle and Baker 1990):
De =φDτ (14)
where τ is the tortuosity of the system and D is the diffusivity of the element or isotope in
the medium (aqueous fluid or melt). Equation (14) links composition profiles to porosity,
diffusivity of solute in the fluid, and tortuosity (Fig. 5B–D). Modeled profiles are matched to
the data to solve for the diffusive distance, De ∆tK e−1 . The duration (Dt) of a metamorphic
event is then determined for given values of the mineral and fluid densities, mineral–fluid
partition coefficient, porosity, tortuosity and intergranular diffusivity.
Major element thermometry
Diffusive resetting and other retrograde processes occurring in a rock affect temperatures
recorded in mineral compositions. Here we discuss the differential impact of exchange
reactions, which simply exchange elements between two minerals without changing modes
(e.g., Fe–Mg exchange), as contrasted with net transfer reactions, which drive significant
changes to the mode of one or more minerals. In general, net transfer reactions cause changes
in concentration for numerous elements (e.g., Fe, Mg, Ca and Mn in garnet), whereas exchange
reactions do not (e.g., Fe and Mg only).
Major element thermometry commonly emphasizes Fe–Mg exchange between garnet
and other silicates, for example biotite. Models of diffusional resetting of apparent garnet–
biotite temperatures illuminate the impact of Retrograde Exchange Reactions (ReER’s) vs.
Retrograde Net Transfer Reactions (ReNTR’s; Spear 1991; Spear and Florence 1992; Kohn
and Spear 2000) on calculated temperatures. In nearly all models, matrix biotite is assumed
to homogenize its composition, whereas the rim of the garnet is allowed to equilibrate with
the matrix, inducing diffusive fluxes of Fe and Mg into and out of the garnet, and producing
a diffusion profile on the rim of the garnet. If only ReER’s occur, matrix biotite becomes
more Mg-rich, and garnet–biotite apparent temperatures are always equal to or less than the
peak metamorphic temperature (Fig. 6A). The garnet core paired with matrix biotite most
closely approximates the peak metamorphic temperature, but always underestimates it, at least
slightly, even if the mode of biotite is large (Fig. 6A). These models implicitly underlie many
interpretations of metamorphic temperatures and diffusion profiles: if exchange reactions
alone occur, calculated temperatures are always minima (lower T’s for smaller garnets), and
boundary conditions are modeled using simple Fe–Mg exchange equilibria.
Partition coefficient (Kc)

high low Tracer concentration
high
l ow
Kc
Kc
Distance
A
low high
crystal
Porosity (φ)
low
hig
Distance
hφ
B
De φD
Tortuosity (τ)
high
low
Distance
τ
C
D e τD
Diffusivity (D)
low
hig
Distance
D h
Figure 5. Factors influencing diffusion in an intergranular medium and bulk rock tracer concentration
profiles. Schematic diagram for each factor illustrates mineral grains surrounded by intergranular me-
dium (shaded region). Arrows indicate distance of diffusion for each scenario. Graphs to the right show
the expected differences in diffusion profiles. (A) Partition coefficient between minerals and fluid. Inset
shows flux from intergranular medium to crystal. A higher partition coefficient results in a shorter diffusion
distance. (B) Porosity. A higher porosity increases effective D and results in a greater diffusion distance
(De ∝ ΦD). (C) Tortuosity. A higher tortuosity results in a shorter diffusion depth. (D) Diffusion coefficient
in the intergranular medium. A higher diffusion coefficient results in a greater diffusion distance.
Unlike ReER’s, if ReNTR’s occur (specifically growth of biotite at the expense of garnet,
e.g., in the reaction garnet + K-feldspar + melt = biotite + sillimanite + plagioclase + quartz), matrix
biotite becomes more Fe-rich (Fig. 6B). Pairing a garnet core with matrix biotite then yields a
temperature that exceeds the peak metamorphic temperature (Fig. 6B). In fact, smaller garnet
crystals, which give lower apparent temperatures with ReER models, give higher apparent
temperatures in their cores with ReNTR models (500 µm vs. 5 mm radius garnets, Fig. 6A, B).
Nearly all high-grade garnets show Fe–Mg diffusion profiles near their rims, but this fact alone
does not justify a particular model. Rather, identifying whether ReNTR’s occurred during cooling,
for example by considering textures or zoning in other Fig. chemical
6 Kohn andconstituents or mineralogical
Penniston-Dorland
evidence for ReNTR’s, is crucial to correct thermometric interpretation (see Spear and Florence
1992; Kohn and Spear 2000; Spear 2004). In detail, the shapes of the diffusion profiles also
differ, which impacts geospeedometry calculations based on Fe–Mg zoning in garnet.
A Retrograde Exchange B Retrograde Net Transfer

Al2O3 Al2O3
B2
B3
B2 G2
B3 G3
G2 G G1 G1
1 G3 G1
B1 B2 B3 B1
FeO MgO FeO MgO
10 °C/Ma
Fe Fe
Concentration
Concentration
500 µm
in garnet
in garnet
Mg Mg
Mn Mn
Ca Ca
Core Rim Core Rim

900 900
10 °C/Ma 10 °C/Ma
Temperature (°C)
Temperature (°C)
800 800 5 mm
5 mm
700 700
Initial 500 µm Initial
600 T=725 °C 600 T=725 °C
500 500 500 µm

400 400
Core Rim Core Rim
Figure 6. Impact of diffusion and reaction type on major element thermometry, showing differences between
retrograde exchange reactions (ReER’s) and retrograde net transfer reactions (ReNTR’s). Models assume a
cooling rate of 10 °C/Ma, an initial temperature of 725 °C, and a homogeneous original garnet composition.
ReER model assumes a volume ratio of garnet/biotite of 0 (infinite biotite, so biotite composition does not
change); ReNTR model assumes a volume ratio of garnet/biotite of 0.2. Dashed lines indicate original com-
positions. Based on Spear (1991) and Spear and Florence (1992). (A) Exchange reactions cause divergence of
garnet (“G”) and biotite (“B”) compositions, such that matrix biotite becomes more Mg-rich. Zoning in garnet
occurs only in Fe and Mg, and apparent temperature is always at or below peak temperature. (B) Net transfer
reactions drive zoning in all garnet components, while biotite shifts towards more Fe-rich compositions. Ap-
parent temperature is low on the rim of the garnet, but exceeds peak temperature in garnet core.
Trace element thermometry

The diffusion of trace elements within minerals also responds to matrix–mineral equilibria.
Here we focus on Zr-in-rutile thermometry because it has been investigated extensively
and exhibits curious features that have spawned increasing debate about rutile’s diffusional
resistance to resetting and about crystal-chemical controls on trace element mobility more
generally. These issues may have broad applicability to many accessory minerals.
The Zr content of rutile in equilibrium with quartz and zircon is temperature-dependent
(Zack et al. 2004; Watson et al. 2006), and experiments indicate that the diffusivity of Zr in
rutile should be sufficiently fast that diffusive resetting is expected at temperatures above 700–
750 °C (Cherniak et al. 2007a). For example, Zr-in-rutile thermometry from matrix grains
yields temperatures of ~700 °C in rocks of the Greater Himalayan Sequence in central Nepal
that likely reached temperatures of 750–800 °C (Kohn 2008) and ~735 °C from granulites
from East Antarctica that likely reached temperatures > 900 °C (Pauly et al. 2016). However,
two perplexing questions have been raised recently about Zr-in-rutile thermometry: why does
zircon or baddeleyite sometimes precipitate in rutile interiors (Kooijman et al. 2012; Ewing
et al. 2013; Pape et al. 2016), and why are very high (up to 900 °C) Zr-in-rutile temperatures
sometimes recorded if Zr diffusion is so fast? One hypothesis is that slow Si diffusion could
explain high Zr-in-rutile temperatures through a resulting elevation of µSiO2 (Taylor-Jones and
Powell 2015). This hypothesis assumes that silicon concentrations increase with increasing
temperature, and that Si diffuses more slowly than Zr (Taylor-Jones and Powell 2015). While
neither assumption is as yet verified, the implications can be considered qualitatively. Here,
we discuss the importance of local equilibrium on diffusive reequilibration and the formation
of Zr and Si diffusion profiles (see also Kohn et al. 2016).
To understand variations in the chemical potential of Zr and Si, we first assume that both
Si and Zr homogenize if initial temperatures are sufficiently high (Fig. 7A). During cooling,
Zr and Si concentrations are predicted to decrease at rutile rims to maintain equilibrium with
matrix quartz and zircon, inducing diffusion loss profiles (Fig. 7B). Within rutile, Si and Zr each
diffuse in response to their respective chemical potential gradients (µSiO2 and µZrO2 respectively),
which for trace elements in a nearly pure mineral are proportional to the natural logarithm of
their concentration gradients via Henry’s Law. Because substitution of Si and Zr into rutile is
not crystal-chemically coupled, and because the interior of the grain contains neither quartz
nor zircon that can buffer chemical potentials, each element diffuses independently of the
other. Therefore, slow diffusion of Si does not impede Zr diffusion, each trace element simply
develops a different profile reflecting different diffusivities and chemical potential gradients.
As Si and Zr contents in the rutile interior progressively deviate from rim concentrations,
the interior of the rutile becomes supersaturated with quartz and zircon. If zircon does not
precipitate, decreasing temperature and ZrO2 content at the rutile rim increases the diffusion
gradient and the magnitude of core ZrO2 supersaturation (Fig. 7C), possibly crossing the
baddeleyite saturation threshold and catalyzing baddeleyite growth. If zircon does nucleate as
an inclusion in the rutile interior, both Zr and Si contents decrease adjacent to the inclusion,
inducing new diffusion loss profiles (Fig. 7D). Outward fluxes of Zr and Si produce matrix
zircon, whereas inward fluxes produce zircon inclusions. Because Zr solubility in rutile likely
exceeds Si solubility (Taylor-Jones and Powell 2015), Si concentrations at the interface with
a zircon inclusion drop below quartz saturation (so aSiO2 < 1.0), and Zr concentrations are
buffered at higher values than on the rutile rim to maintain equilibrium with zircon (Fig. 7D).
The Zr concentration at the zircon inclusion and matrix interfaces, however, is always lower
than the maximum Zr elsewhere in the rutile interior (Fig. 7D).

1.0
To
Zrn+Qtz Qtz 1.0
XSiO2
XZrO2
aZrO2
A
aSiO2
Zr
0.0 0.0
T1
1.0
Qtz 1.0
XSiO2
XZrO2
B Zrn+Qtz
aSiO2
aZrO2
0.0 0.0
Bdl
T2 No Zrn precipitation
Sat’n
1.0
XSiO2
XZrO2
C Qtz 1.0
aZrO2
Zrn+Qtz
0.0
aSiO2
0.0
T2 Zrn precipitates
Zrn
Zrn
1.0
XSiO2
XZrO2
D Qtz 1.0
aZrO2
Zrn+Qtz Low
aSiO2
High a(SiO2)
a(ZrO2)
0.0 0.0
Rim Core Rim Core
Distance Distance
Figure 7. Impact of diffusion on trace element thermometry, illustrated schematically for the evolution of
Zr and Si concentrations and activities in rutile (red lines and curves) during cooling from high temperature
(based on Kohn et al. 2016). Dashed lines are concentrations of Zr and Si in equilibrium with zircon (Zrn)
and quartz (Qtz) in the matrix, darker vs. lighter shading within circle indicates higher vs. lower concentra-
tion of Zr within rutile. Activities are relative to quartz and baddeleyite. (A) At high initial temperature,
T0, rutile has homogeneous Zr and Si concentrations in equilibrium with quartz and zircon; aZrO2 < 1. (B)
With decreasing temperature (T1), rim concentrations decrease to maintain equilibrium with matrix quartz
and zircon, inducing diffusive loss of Zr and Si; in the rutile interior, aSiO2 > 1 and aZrSiO4 > 1 (i.e., quartz
and zircon are supersaturated). (C) With further decreasing temperature (T2), quartz and zircon become
increasingly supersaturated, and the rutile core may become supersaturated in baddeleyite (bdl sat’n). (D)
If zircon precipitates in the rutile interior (gray region on right), Zr and Si will diffuse toward the zircon
to reach equilibrium with zircon (without quartz). If Si concentrations are low, Zr at the zircon inclusion
boundary will be higher than at the rutile rim, but lower than elsewhere in the rutile grain. Si concentrations
will be lower at the zircon inclusion boundary because, although zircon is a saturating phase, quartz is not.
Although the concepts illustrated in Figure 7 help explain formation of zircon and baddeleyite
inclusions inside high-T rutile, as well as 700–750 °C Zr-in-rutile temperatures in rocks that were
once hotter, they do not explain how unusually high Zr-in-rutile temperatures can be preserved.
Such high temperatures require that boundary concentrations did not maintain equilibrium
with matrix quartz and zircon (Kohn et al. 2016). That is, high Zr-in-rutile temperatures do not
necessitate slow Zr diffusion, as some have argued (e.g., Kooijman et al. 2012) but instead require
shutdown of mineral equilibration via chemical transport along grain boundaries. This topic is
explored further in the discussion of natural examples of trace element diffusion profiles.
Reaction rates
Commonly, diffusion profiles are modeled and interpreted assuming that the physical position
of a crystal surface is fixed. Reacting minerals, however, also form and preserve diffusion profiles.
These situations are generally referred to as moving boundary problems, and have different
solutions than fixed boundary problems. If specific mass balance and reaction rate criteria are
met, diffusion profiles can be used to infer mineral reaction rates (Jackson 2004; Lucassen et
al. 2010; Cruz-Uribe et al. 2014). Here we discuss more generally how reacting minerals might
develop diffusion profiles and how those profiles may inform rates of petrogenetic processes.
Understanding the time evolution of reaction (Fig. 8A) helps predict compositional
trends for different scenarios (Figs. 8B–E). In each case, material a transforms to material
b at a constant rate. The boundary moves at a constant velocity, v, which is proportional to
the reaction rate. Thus, to obtain the reaction rate, we must estimate v. At equilibrium, a and
b exhibit constant partitioning, k = Cb/Ca, where Ca is the concentration of the diffusant in a
and Cb is the concentration in b. For example, a and b might be olivine melt and crystalline
olivine, and the diffusant could be any major, minor or trace element with k ≠ 1, e.g., Fe / Mg,
Ni, Li, etc. Our illustrations assume k < 1 (the element prefers reactant phase a), but for k > 1,
our figures would simply be inverted vertically. Diffusion in b is assumed to be slow, so that
it preserves composition profiles, but infinitely fast diffusion in b does not change the main
conclusions regarding profiles in a. Compositional evolution of a and b depends on whether the
a–b interface is wholly isolated from equilibrating with the matrix or melt (“closed system”), or
whether the matrix participates as a source or sink of the diffusant (“open system”), as well as
whether partitioning is preserved at the a–b interface.
The closed system model has a unique analytical solution for the diffusion profile in
phase a, which is derived by transforming Equation (2) from a fixed coordinate system (x) to
a moving coordinate system (z) where z = x − vt:
 1  − vz / D )
C ( z ) =+
C0 C0  − 1  e( (15)
k 
where D and C0 are the diffusion coefficient and original concentration of the diffusant in
phase a (Fig. 8B; Jackson 2004). At steady state, a transforms to b at a constant rate, and both the
diffusion profile defined by Equation (15) and the boundary between a and b move at constant
velocity, v. To conserve mass, the steady state concentration of any new b must be C0, fixing the
steady-state boundary condition on the surface of a at C0 / k where k is the partition coefficient
as described above (Fig. 8B). Even if a and b are not compositionally comparable, the same
principles apply but require a scaling factor to conserve the mass of the diffusant (Lucassen et al.
2010). For example, if titanite (CaTiSiO5) grows at the expense of rutile (TiO2), rutile contributes
only ~40% of the total mass of titanite. So, on a mass basis, the concentration of a trace element
(e.g., Zr, Nb) in the steady-state titanite should be ~40% of the concentration in the original
rutile, assuming the matrix does not serve as a source or sink of the trace element.
Initial state Early transient Late transient Final/steady state

vt
A t1 t2 t3 t4 β α
z
Distance x
Closed system, equilibrium α - β partitioning Co/k

1
C=Co+Co( k -1)e(-vz/D)
Concen-
Equal areas
α α α α
tration
B Co Co Co Co
β β
β=k·Co β
k·Co Steady state
Open system, equilibrium β - matrix, disequilibrium α - matrix
α α α α
Concen-
Co Co Co Co
tration
C
β β β β
No steady state
Open system, equilibrium α - β - matrix partitioning
Concen-
tration
D α
Co
α
Co
α
Co
α
Co
β β β β Possible steady state
Open system, rapid reaction Reaction stops, then
Concen-
Profile forms
tration
E α
Co
α
Co
α
Co
α
Co
β β β β
No steady state
Figure 8. Models of reaction rates and induced diffusion profiles. (A) Time series, showing movement of
boundary between α and β (the same progression of the boundary is shown in Figs. 8B–E). Phase β replaces
phase α at a constant rate, and coordinate system can be transformed from fixed (x) to moving (z). The veloc-
ity (v) is proportional to the reaction rate. (B) Closed system with equilibrium partitioning of major, minor or
trace elements between α and β (Jackson 2004). During initial growth of phase β, a deficit in diffusant occurs
in β that is balanced by an excess of diffusant at the rim of α (shaded regions). During this transient phase the
excess diffusant in α increases, while the concentration in β increases to maintain equilibrium with α. This
excess forms a diffusion profile in α, which at steady state moves at the same rate as the boundary between
α and β. (C–E) Open system in which some diffusant enters matrix. Mass excess in α does not match mass
deficit in β: shaded area in β (equivalent to shaded area in α) is smaller than the total area in β. (C) Matrix
maintains equilibrium with β, but not with α, so boundary composition of α and α–β partitioning change with
time. Diffusion profile in α never reaches steady state. (D) Matrix maintains equilibrium with both α and β.
Steady state is possible with same solution as Equation (15). (E) Matrix maintains equilibrium with both α
and β, but rapid reaction prevents diffusion profile from developing in α until after reaction ceases.
The closed-system steady-state model further predicts that the first-formed b has
concentration k·C0 (a consequence of the a–b equilibrium condition), and that the mass
deficit that develops in b balances the mass excess present in the diffusion profile in a (a
consequence of mass conservation; shaded regions, Fig. 8B). Thus, if D is low in phase b,
the process by which the diffusion profile develops in phase a simultaneously produces a
complementary profile in b (Fig. 8B; Jackson 2004). In addition, after the system achieves
steady state, phase a should exhibit concentration C0 / k at the interface between a and b,
while the concentration ratio of b to a should be k (Fig. 8B).
What happens if mass is not conserved? For example, perhaps phase equilibria involving
matrix minerals control the composition of phase b, rather than partitioning with phase a. If so,
the composition of phase b is fixed, and the mass deficit in b will not match the mass excess in
a (Fig. 8C). Such a scenario might occur if some of the diffusant escapes the local a–b reaction
region, traversing phase b to form or enter some other phase in the matrix (e.g., Lucassen et
al. 2010). Depending on the rate of diffusant loss to the matrix, the formation of a steady-state
diffusion profile in a is not assured (Fig. 8C). Indeed, as phase b progressively armors phase a,
loss of diffusant from a probably decreases, and the mass of diffusant in the diffusion profile in
a gradually increases. No steady state exists, so no reaction rate can be calculated.
Alternatively (Figs. 8D, E), perhaps equilibrium with matrix minerals establishes not only
a constant composition for product phase b, but also a constant boundary condition for phase a,
independent of the mass balance that Equation (15) requires. A steady state diffusion profile
can form in a, mathematically comparable to Equation (15), but with a different maximum
concentration (Fig. 8D). It is unclear, however, whether steady state can ever be assumed
for such a system. Relaxing mass balance constraints opens alternate interpretations for how
diffusion profiles might form. For example, reaction might be too fast to form a diffusion
profile in a, but the fixed boundary condition forms a diffusion profile after the reaction has
ceased (Fig. 8E). The form of the resulting diffusion profile (an error function), is not easily
distinguishable from the exponential form of Equation (15) (Lucassen et al. 2010).
In sum, steady-state closed-system reaction (Fig. 8B) is the only scenario that permits
unequivocal reaction rate estimates and makes several testable predictions: an increasing
concentration in the product phase (b) towards the a–b boundary, a concentration in b that
is either the same as the interior of a (for 1:1 replacement; Jackson 2004) or in proportion
to the amount of reactant a that is present in b (Lucassen et al. 2010), and equilibrium
partitioning between a–b at their interface. Conversely, a constant concentration profile
in b, mass deficits between phases a and b, or disequilibrium at the a–b interface would
challenge a steady-state reaction model and its analytical solution. Such data might place
limits on the extraction of information about other kinetic processes, such as cooling rate
from the shape of the diffusion profile (Fig. 8E), but not on the extraction of reaction rates.
Diffusive resetting of isochrons
Differential diffusivity of parent and daughter isotopes in a mineral can change
chronologically useful isotopic ratios after a mineral has crystallized and produce ages that no
longer correspond to the crystallization age. Commonly, geochronologists assume that resetting
responds solely to diffusion of the daughter isotope, and that the parent isotope is inert, so that
the measured age is younger than the crystallization age. A large and well-trodden literature is
devoted to this phenomenon, especially 40Ar loss (see Dodson 1973; McDougall and Harrison
1999). While preferential resetting of the daughter isotope generally holds true for U–Pb and
K–Ar because U has a higher charge than Pb, and Ar is not electrostatically bonded in a crystal,
preferential diffusion of the parent can also occur. This phenomenon has been recognized and
quantitatively characterized for the Lu–Hf system in garnet (e.g., Kohn 2009; Bloch et al. 2015;
Bloch and Ganguly 2015), and leads to unusual chronologic behavior. In this system, the parent,
176
Lu3+, diffuses about an order of magnitude faster than the daughter, 176Hf 4+ (Bloch et al. 2015)
and whereas Hf is generally immune to diffusive resetting except for ultrahigh-temperature
rocks or sub-mm diameter garnets, strong zoning in Lu is subject to diffusional relaxation, even
at moderate temperatures (~750 °C; Kohn 2009; Bloch and Ganguly 2015).
Differential uptake of Lu but not Hf during garnet growth forms a bell-shaped Lu
concentration profile, while Hf concentrations are uniformly low (Fig. 9A; Lapen et al.
2003, Kohn 2009). Strong Lu concentration gradients drive outward diffusion from garnet
cores, reducing 176Lu /177Hf ratios; if resorption occurs, Lu can diffuse inward at garnet
rims, increasing 176Lu / 177Hf (Fig. 9B; Kohn 2009). Where metamorphic temperatures
are high enough (> 800 ºC), grain sizes small enough (< 1 mm), and / or the duration of
metamorphism long enough (tens of Ma), Lu may be completely homogenized, however
in most metamorphic environments these criteria are not met (Bloch and Ganguly 2015).
Consequently, intracrystalline diffusion of Lu, if it occurs after significant ingrowth of 176Hf,
should produce a counterclockwise rotation of isochrons, potentially yielding ages that are
older than the crystallization age of the garnet (Fig. 9B; Kohn 2009).
800 80
A Mn B Core
Yttrium
d
600 60 ed d ifie
Lu, Hf (ppm)
lax mo
Re Un
Y (ppm)
Hf/177Hf
400 1 mm 40
Trough Lutetium Trough Core Unmodified

Relaxed
176
200 20
Rim [Lu] Rim
Hafnium
0 0
0.0 0.5 1.0 1.5 2.0 2.5 176
Lu/177Hf
Distance (mm)
10.0 0.36 30
C D
8.0 2 20
0.34
15
Lu diffusion
Age offset (Ma) 10 Bulk Age

20 Hf
Lu/177Hf
Hf/177Hf
6.0 0.32
Age of
30 0
10 Nucleation
4.0 20 0.30
176
1020
176
-10
2.0 0.28 -20
0.0 0.26 -30

0.00 0.05 0.10 0.15 0.20 0.25 0.00 0.05 0.10 0.15 0.20 0.25
Core Rim Core Rim
Distance (mm) Distance (mm)
60
E 850ºC, 0.5 mm
40
800ºC, 0.5 mm
Age offset (Ma)
20
700ºC, 0.25 mm
0 700ºC, 0.5 mm
800ºC, 0.5 mm (MB)

-20
900ºC, 0.5 mm
-40
10 20 30 40 50 60
Duration of Garnet Growth (Ma)
Figure 9. Diffusive resetting of isochrons, modified from Kohn (2013) and Bloch and Ganguly (2015). (A)
Trace element zoning across garnet from a ~750 °C rock in the Himalaya, showing bell-shaped distributions
of Y and Lu, but increases towards rim (forming troughs), indicating resorption and back-diffusion during
cooling. Inset shows Mn X-ray map of garnet from same rock. (B) Schematic illustration of how Lu diffu-
sion causes 176Lu / 177Hf to decrease in garnet core and increase at garnet rim, rotating isochrons to older ages.
Inset illustrates Lu zoning. (C) Fully quantitative model of Lu zoning in garnet through a metamorphic cycle.
Each curve represents the Lu profile at specific times (in Ma) since garnet nucleation. Garnet grows from
500 °C to 750 °C over a 20 Ma interval, so its radius progressively lengthens up to 20 Ma, then cools at a rate
of 2 °C / Ma, ultimately achieving an age of 1020 Ma. Diffusion of Lu during and after growth drives down Lu
concentration in garnet core. Resorption and Lu increase at rim were not modeled. (D) Age offset (difference
between modeled age for garnet growth with and without diffusion)generously provided by E Bloch. Loss of
Lu from garnet cores leads to unsupported 176Hf and spuriously older ages. Younger ages towards rim reflect
position-dependent closure. (E) Age offset as a function of garnet growth rate, peak temperature, and final
garnet radius. Solid curves are for metapelitic compositions, dashed curve (MB) is for metabasaltic composi-
tion. The curve for 900 °C develops opposite trend because Hf diffusion becomes significant.
Fully quantitative numerical simulations of Lu and Hf uptake and diffusion (Bloch and
Ganguly 2015; Fig. 9C–E) demonstrate that garnet cores can lose Lu during both prograde and
retrograde metamorphism (Fig. 9C), leading to garnet core ages that exceed the nucleation age
by over 20 Ma (Fig. 9D). Even the bulk apparent age can exceed the nucleation age by ~10
Ma (Bloch and Ganguly 2015; Fig. 9D). Garnets from metabasaltic rocks may not show the
same age bias, however, because Lu is not so strongly fractionated in garnet cores (Lapen et
al. 2003) so the extent of diffusional bias is reduced (Bloch and Ganguly 2015). The Sm–Nd
system behaves differently, because parent and daughter have almost identical diffusivities
(Tirone et al. 2005; Carlson 2012), and because neither becomes strongly zoned during garnet
growth (Lapen et al. 2003; Kohn 2009; Bloch and Ganguly 2015). Clearly care must be taken
in interpreting Lu–Hf garnet ages, especially for relatively small (<~1 mm), zoned garnets that
grew between 700 and 900 ºC (Fig. 9E; Bloch and Ganguly 2015). Despite these problems,
Lu–Hf ages can be used to constrain the duration of prograde metamorphism if factors such
as peak temperature and cooling rate are determined independently, and if Lu–Hf dating is
combined with Sm–Nd dating, which experiences different but predictable age bias (Lapen et
al. 2003; Kohn 2009; Bloch and Ganguly 2015).
PART 3: EXAMPLES
This section focuses on examples of the effects of diffusion that are relevant to
petrochronology: natural contraints on diffusion rates, geospeedometry (from major element,
trace element, and radiogenic isotope zoning), reaction rates, timescales of magmatic processes
and magma ascent rates, duration of metamorphism, and fluid–rock interaction.
Natural constraints on D
Natural data are commonly interpreted using experimentally determined diffusion
parameters to characterize geological processes. But the inverse exercise—using natural data
from well-characterized geological systems to constrain diffusivities—can help refine E and
D0 by extending the ranges of intensive parameters that control D, especially temperature,
beyond what is possible in experiments. Such comparisons are needed to validate diffusion-
based petrochronologic applications in other settings. The vast majority of natural data are
consistent with experimental data, but in a few instances natural and experimental D’s are not
reconcilable. These instances may provide important insight into mechanisms of diffusion,
or recommend alternative experimental methods more relevant to natural systems. Here we
discuss examples illustrating consistency between natural data and high-T experiments, but
also a few exceptional cases where natural and experimental data appear in conflict—in
these cases natural data typically imply much lower D’s than experiments.
Before discussing these examples, we emphasize that the most difficult problem in using
natural rocks to estimate D’s is that solutions to the diffusion equation all involve the combined
parameter D·Dt, where Dt is a characteristic time interval; because D depends exponentially
on temperature (Eqn. 3), Dt is generally approximated with the duration (time-interval) the
rock spends near maximum T. Although maximum T may also be subject to considerable
error (Spear 2014), robustly constraining Dt often proves even more difficult. In metamorphic
rocks, there are few a priori constraints on how long peak metamorphism “should” last. For
example, some studies have argued for relatively short metamorphic events (e.g., Dt  1 Ma in
the Barrovian type locality; Ague and Baxter 2007), including rapid metamorphism of rocks
via movement through thermal gradients at plate tectonic velocities (e.g., European Alps:
Rubatto and Hermann 2001; Himalaya: Kohn et al. 2004; Kokchetav massif: Stepanov et al.
2016) or via thin-skinned duplexing at mid- to deep-crustal depths (northern Appalachians:
Spear et al. 2012; Spear 2014). Subduction models of metamorphism similarly imply peak
metamorphic Dt ~ 1–2 Ma (e.g., Gerya et al. 2002; Warren et al. 2008). Yet other studies
argue for more protracted high temperatures (Dt ~ 20 Ma; e.g., Himalaya: Kohn and Corrie
2011; Dabie-Sulu: Liou et al. 2012), as do other types of thermal-mechanical models (e.g.,
Beaumont et al. 2001, 2004). Commonly, we know the maximum duration of metamorphism
better than the minimum. For example, Dt is always smaller than the difference between
depositional and late-stage cooling ages. Because D is determined inversely to Dt, a minimum
D (maximum Dt) may be readily constrained, whereas a maximum D (minimum Dt) is more
difficult to estimate. In several studies, the duration of metamorphism is modeled using
thermal models of cooling plutons (e.g., garnet: Carlson 2006, 2012; monazite: McFarlane
and Harrison 2006). Otherwise Dt is determined from regional chronologies. With typical
analytical uncertainties of 1–2% for ages, errors in Dt for Archean and Proterozoic orogens
(e.g., Grenville) can easily exceed 10 or 20 Ma. In contrast, for younger orogens (Himalaya,
Alps), even large chronologic errors (e.g., 5–10%) result in much smaller errors in Dt.
As an illustration of potential ambiguities in determining D, Peck et al. (2003) argued
for slow oxygen diffusion in zircons from the Adirondack Mountains, in part based on the
preservation of isotopic differences on length scales of ~50 µm across zircon overgrowths.
Many other data also suggest slow oxygen diffusion in zircon, so these interpretations are not
unusual. However, Peck et al. (2003) assumed a peak temperature of ~600 °C based on regional
thermobarometry (Bohlen et al. 1985) and estimated Dt to be 25–50 Ma based on a broad
regional chronology for the Ottawan orogenic event (dated at ~1000 Ma). The chronology, in
turn, was determined from conventional U–Pb analyses of zircon, monazite, titanite, and rutile
separates from rocks distributed over large distances exceeding 100 km (Mezger et al. 1991).
Subsequent work on some of the same metamorphic rocks demonstrates that the zircon
overgrowths actually formed ~1150 Ma during an altogether different orogeny (the Shawinigan
event; Peck et al. 2010) and that temperatures locally reached at least 800 °C (Storm and Spear
2005, 2009). The duration of the Shawinigan event is not well constrained, but is associated
with anorthosite intrusions (e.g., McLelland et al. 2010). Arguably, if a short-lived thermal pulse
during the Shawinigan catalyzed formation of metamorphic zircon overgrowths, and Ottawan
overprinting was either low-temperature or brief (Dt as low as ~1 Ma, Page et al. 2010), D’s for
oxygen diffusion could be higher, perhaps by as much as 1–3 orders of magnitude. Conversely,
higher peak metamorphic temperatures would require even lower D’s than implied by the
original assumed temperature of 600 °C, perhaps by as much as 5 orders of magnitude. In the
absence of robust estimates of temperature and petrogenetically based chronologic data with
small absolute errors, preferably from the same outcrops, calculations of diffusivities based on
natural samples and comparison to experimental data can carry large uncertainties.
Garnet. Divalent cation diffusion in garnet has received close scrutiny, both
experimentally and in natural systems (see summary of Ganguly 2010), and illustrates
internal consistency for several elements. Early experiments yielded somewhat discrepant
results until the effects of composition and oxygen fugacity were fully understood. Now a
diverse set of experiments can be assembled into a self-consistent set of diffusion coefficients
(Chakraborty and Ganguly 1992; Ganguly et al. 1998; Ganguly 2010). Among empirical
estimates, Carlson (2006) is most comprehensive in attempting to reconcile experiments with
empirical analysis at two localities: the Llano uplift in central Texas and the Makhavinekh
Lake Pluton contact aureole in Labrador. Some questions have been raised about Carlson’s
estimates of pressure corrections and compositional dependencies (Ganguly 2010), and also
temperatures assumed for the Makhavinekh contact aureole are higher than reported from
mineral equilibria ( McFarlane and Harrison 2006), which might introduce a systematic error
of as much as 4% in 1 / T. Although it is difficult to account quantitatively for all sources of
error when using natural data, such that corresponding errors in T and D may be large, there
is still remarkable agreement among experimental and empirical data for Mg, Fe and Mn
(Fig. 10A). Internally consistent constraints have also been derived for diffusion of trivalent
cations (Carlson 2012). Experimental and empirical estimates of Ca diffusivities remain
unusually discrepant, however (~4 orders of magnitude at 1000 °C; Fig. 10B). Data scatter for
Ca diffusivity overlaps data for Mg, Fe, and Mn, but natural data have long demanded much
lower D’s for Ca than for Fe, Mg and Mn, in fact likely more similar to trivalent cations and
oxygen (e.g., Spear and Kohn 1996; Kohn 2004; Vielzeuf et al. 2005). But exactly how much
lower remains unknown. Regardless, even where absolute Fig.values
10ofKohn
D cannot be constrained,
and Penniston-Dorland
relative D’s can sometimes be inferred (Qian et al. 2010; but see also Till et al. 2015).
C
C
C
C
C
C
°
°
°
°
00
00
00
00
0°
0°
0°
0°
15
80
60
15
80
60
11
11
-14 -14
Garnet, Mg Garnet, Ca
Perchuk et Mg
al. (2009) Freer and
Edwards (1999)
-18 Borinski et -18
D(m2/sec)
D(m2/sec)
Ganguly et al. (1998)
al. (2012) Perchuk et al. (2009)
Ganguly et Vielzeuf
Kohn (2004)
Carlson
al. (1998) et al. (2007) (2006)
-22 -22 Schwandt
et al. (1996)
Carlson
(2006)
-26
A -26
B
5.0 7.5 10.0 12.5 5.0 7.5 10.0 12.5
10000/T(K) 10000/T(K)
°C
°C
C
C
C
C
0°
0°
00
00
0°
0°
0°
0°
0
0
15
80
60
15
80
60
11
11
-16 -16
Cherniak Titanite Zircon
(2006)
Sr
Watson and
Ta Cherniak (1997)
Nb Cherniak
-20 Cherniak (1993) -20
D(m2/sec)
D(m2/sec)
(2015) O, dry O, wet

DΔt≥7.5x10-10
Cherniak
Δt≥~150 ka
(1995) Nd Zr Pb
DΔt≤7.5x10-10
-24 Pb,U Dabie Pb,U WGR -24 Δt≤~6 Ma Pb
Pb Nepal B11
Zr,Pb,U, Cherniak and
0.5 Ma
Th Pamir Watson (2001) P03
5 Ma
P07
-28
C -28
D 50 Ma
5.0 7.5 10.0 12.5 5.0 7.5 10.0 12.5

10000/T(K) 10000/T(K)
Figure 10. Constraints on diffusivities from experiments (circles) and natural data (other symbols). (A–B) Mg
and Ca diffusion in garnet normalized to 1 GPa and fO2 at the C–CO buffer, showing good internal consistency
for Mg and large discrepancies for Ca. Borinski et al. (2012) data include reinterpretation of high-T experi-
mental data from Ganguly et al. (1998), and Ganguly et al. (1998) correct low-T data of Cygan and Lasaga
(1985) and Chakraborty and Rubie (1996) for fO2 and pressure. Solid lines show general trends of all data (two
illustrative lines for Ca). Maximum D for Ca must be lower than Mg D (Kohn 2004). (C) Diffusion data for
titanite, highlighting discrepancies between experimental vs. natural data for Zr and Pb (Kohn 2017). Lines
show trends of experimental data, dashed where extrapolated. (D) Diffusion data for Pb and O in zircon. Lines
show trends of experimental data, dashed where extrapolated. Large triangles for Pb show constraints on DDt
values and implications for minimum and maximum durations of peak metamorphism for Kokchetav and
Bohemian massifs. Boxes show constraints on oxygen D for different durations of metamorphism (50, 5 and
0.5 Ma). P03 = Peck et al. (2003; temperatures adjusted for data in Storm and Spear 2005, 2009), P07 = Page
et al. (2007; temperature adjusted for Bohlen et al. 1985), B11 = Bowman et al. (2011).
Titanite. In many rocks, titanite U–Pb ages are younger than zircon U–Pb ages, prompting
most early researchers to conclude that Pb diffusion is sufficiently fast that titanite records only
a cooling age, not a crystallization age (e.g., Mattinson 1978; Mezger et al. 1991; Heaman and
Parrish 1991). Experimental data also suggest relatively high D’s, implying typical closure
temperatures of ~600 °C (Cherniak 1993). Nearly every subsequent chronologic study has
assumed Pb diffusion is fast, and that ages are reset. Curiously, the high diffusivity for Pb contrasts
starkly with the low diffusivity of Sr (Cherniak 1995), which is also divalent, substitutes into the
same crystallographic site, and has a nearly identical ionic radius (1.21 Å for Sr and 1.23 Å for
Pb; Shannon 1976). An increasing body of data from diverse orogens now suggests that Pb
diffuses much more slowly, likely as slowly as Sr (Fig. 10C; see summary in this volume of
Kohn 2017). In fact, diffusivities of Pb, Zr, U and Th all appear comparable to Sr and Nd, and
2–4 orders of magnitude lower than experimental estimates for Pb and Zr (Fig. 10C; Kohn 2017).
The consistency of empirical Pb and experimental Sr diffusivities accords with their respective
charges and radii, whereas low empirical diffusivities for Zr, U and Th are consistent with their
high ionic charges and substitution either into a small crystallographic site (Zr) or via coupled
chemical exchanges (U and Th; Brady and Cherniak 2010; Cherniak 2010b).
Explanations for the discrepancies between experimental and natural data are lacking. Many
of the Pb diffusion data were collected using a Pb implantation method, which damages the crystal
lattice, but the damage quickly anneals and experiments using a PbS exchange medium yielded
indistinguishable D’s (Cherniak 1993). The Zr diffusion experiments show no dependence of D
on Zr source or fO2. Still it would appear the experiments tap a fast-diffusion pathway not seen
(yet) in nature. New experiments with different starting materials or methods seem warranted.
Zircon. Zircon is commonly assumed to be extremely resistant to diffusional
modification, but is this rule of thumb always true? Whereas experimental data for Pb and O
diffusion (dry) indeed suggest extremely low D’s, wet experiments suggest higher D’s for O
(Watson and Cherniak 1997; Cherniak and Watson 2001), and modeling of H2O fugacity in
metamorphic rocks during cooling suggests that oxygen should diffuse in most minerals at
rates commensurate with wet diffusion experiments (Kohn 1999). Do natural data show the
expected widespread resetting of O, but retention of Pb?
Although few rocks retain such high temperatures for sufficiently long times that significant
Pb diffusion is expected, resetting of Pb is possible for zircons from the Kokchetav and Bohemian
massifs, depending on assumed durations at near-peak temperatures. In evaluating these data, a
useful benchmark is that 90% diffusive resetting of the center of a sphere will occur at DDt / r2 ≥ 0.3
(r = radius, Dt = duration at peak metamorphic T; Crank 1975), or DDt ≥ 7.5×10−10 m2 for the core
of a 50 µm radius crystal. For both massifs, assuming D from experiments, we can identify the
timescales over which U–Pb ages will be preserved vs. erased. In the Bohemian massif, core,
mantle and rim domains in zircon, interpreted as prograde, peak, and high-T retrograde growth,
show no age differences (Kotková et al. 2016), possibly indicating resetting. Temperatures
reached ~1100 °C at UHP conditions, and retrograde overprinting occurred at ≥ 1050 °C in
the granulite facies (Haifler and Kotková 2016). At 1100 °C, using D’s from Cherniak and
Watson (2001; “dry” experiments), 90% resetting would occur for durations of only ~150 ka.
Models of subduction and UHP metamorphism suggest that maximum temperatures may be
retained for 1–2 Ma (Gerya et al. 2002; Warren et al. 2008), so resetting of zircon U–Pb ages
in the Bohemian massif is therefore expected. In the Kokchetav massif, peak temperatures of
1000 °C were insufficient to completely reset prograde and peak metamorphic zircon domains
(Hermann et al. 2001; Katayama et al. 2001; Stepanov et al. 2016), implying ≤ 90% Pb loss.
If experimental D’s are correct, this requires Dt ≤ ~6 Ma, which can be reconciled with the
1–2 Ma peak metamorphic duration that some subduction models imply. Thus, retention of
U–Pb ages at Kokchetav is possible. Whereas a shorter duration at peak temperature would
not constrain diffusivities, a longer duration would imply slower diffusion (D < ~10−23 m2 / s at
1000 °C, Fig. 10D). In sum, we know of no natural data that directly constrain Pb diffusivities,
but additional petrochronologic analysis in a few exceptional complexes such as the Kokchetav
and Bohemian massifs might provide limits. Diffusion profiles in other elements might be
compared with U–Pb ages to determine diffusion rates relative to Pb.
Attempts to reconcile experimental vs. empirical oxygen diffusion rates in zircon have
proved difficult. Microanalytical data reveal diffusion gradients in isotopic composition on
scales as small as a few µm (Adirondacks: Peck et al. 2003; Page et al. 2007; Superior
Province: Bowman et al. 2011), suggesting very slow diffusion. Unfortunately, durations
at peak metamorphic conditions are known poorly for the areas studied. An example of
ambiguity for the Adirondacks is discussed above. For the Superior Province, the duration
near peak granulite-facies conditions is constrained only to ≤ 27 ± 3 Ma, based on bracketing
ages of deposition (2667 ± 2 Ma) and of dikes that exhibit amphibolite-facies margins
(2640 ± 2 Ma; Krogh 1993). Regardless of these details, Bowman et al. (2011) report data that
seem irreconcilable with wet diffusion experiments: the duration of regional metamorphism
would have to be unrealistically brief, 500–5000 years (Fig. 10D).
Some studies have proposed that a hydrous fluid must be present to produce oxygen
isotope resetting, and that oxygen diffusion may be slow in fluid-free but high-f H2O rocks
(Peck et al. 2003; Bowman et al. 2011). However, the compositional systematics of retrograde
resetting of oxygen isotope compositions in quartz, feldspar, micas and oxides suggests that
high “wet” diffusion rates apply, even in granulites that likely lacked a free fluid phase (Kohn
1999). That is, rocks may be physically dry during cooling but are thermodynamically buffered
to relatively high f H2O and experience resetting as if they were wet. In addition, dehydration
during prograde metamorphism should produce a fluid during the heating portion of each rock’s
P–T path. Unless heating rates exceeded ~100 °C / Ma, zircon d18O should be reset. A simpler,
albeit more radical, interpretation is that the wet diffusion experiments are somehow biased.
Growth of thin rims on zircon, which would lengthen apparent profiles and increase apparent
D, is not supported (E. B. Watson, pers. comm. 2016). As suggested for titanite, perhaps the
experiments tap a fast-diffusion pathway that is less effective in nature. That is, although f H2O
does control oxygen resetting in cooling rocks (Zhang et al. 1991; Kohn 1999), natural wet
diffusivities for oxygen in zircon are simply lower than measured experimentally, much as
was inferred for Zr and Pb in titanite. One possible resolution to this quandary would be to
re-investigate oxygen diffusion rates in zircon using fluid-absent, but high-f H2O experiments.
Dolomite–ankerite. Dolomite cores with ankerite overgrowths provide a natural diffusion
couple for investigating the duration of biotite- to garnet-grade metamorphism in the northern
Appalachiansm, USA (Ferry et al. 2015). Regional petrology suggests that ankerite rims grew
before peak metamorphism, so the concentration profiles of Fe / Mg between carbonate cores
and rims provide a measure of the maximum duration near peak metamorphic conditions.
Experimentally determined diffusion rates for Fe and Mg in carbonates (Müller et al. 2012)
imply durations at peak metamorphic conditions of ~1 year, which is clearly implausible. One
complicating factor is fO2, which was up to 20 orders of magnitude lower during metamorphism
than in experiments (Ferry et al. 2015). Most diffusion rate fO2-dependencies are fairly small:
D ∝ fOa2 where the exponent a is typically ~1 / 6 (Chakraborty and Ganguly 1992; Dohmen et
al. 2007, 2016), but can be as high as 1 / 5 to 1 / 3.2 (see Chakraborty 2010). Thus, corrections
would increase calculated metamorphic durations by 3–6 orders of magnitude or between ~1 ka
and ~1 Ma, within the range of other estimates in the same region (≤ ~1 Ma; Spear et al. 2012;
Spear 2014). More accurate estimates of the duration of metamorphic events will require further
investigation of Fe–Mg diffusion rates and their fO2 dependence in ankerite and dolomite.
Geospeedometry from major element cation zoning

Major element zoning within garnet rims from the Valhalla Metamorphic Complex, British
Columbia, Canada, has been used to estimate cooling rates in this high-grade metamorphic
core complex (Spear and Parrish 1996; Ducea et al. 2003; Spear 2004). In these rocks,
Fe / Mg increases towards the rims of garnets due to retrograde diffusional reequilibration
with the matrix, and this zoning has been used to interpret cooling rates between ~825 and
~550 °C (Figs. 11A–B; Ducea et al. 2003; Spear 2004). These data exemplify petrology’s
central role in constraining models that are used to estimate cooling rates: differences in the
retrograde reactions that are assumed to produce the zoning (ReER’s vs. ReNTR’s) radically
change estimated rates and consequent tectonic interpretations of these rocks. In all models,
the change in composition (mole fraction) with respect to time (dX / dT) was determined
by varying cooling rate (dT / dt), and by using a temperature–composition path (dX / dt) to
establish the equilibrium boundary condition on the rim of the garnet. The thermodynamics of
the assumed governing reaction—whether ReER or ReNTR—define dX / dT. In both studies,
the diffusivities of Chakraborty and Ganguly (1992) were used in a finite difference diffusion
model to extract the change in composition over distance within the garnet.
XPrp A XPrp B
0.3 0.8 0.3 0.8
XAlm,Fe/(Fe+Mg)
XAlm,Fe/(Fe+Mg)
ReER ReNTR
XPrp,Sps,Grs
XPrp,Sps,Grs
0.2 5°C/Ma Fe/(Fe+Mg) 0.7 0.2 150°C/Ma Fe/(Fe+Mg) 0.7
X Alm X Alm
0.1 0.6 0.1 0.6
XGrs XGrs
increased Mn XSps increased Mn
XSps
0.0 0.5 0.0 0.5
500 400 300 200 100 0 1250 1000 750 500 250 0
Rim Rim
Distance (µm) Distance (µm)
900 900
Mnz
ReNTR cooling history

Temperature (°C)
Ttn ReER cooling history

800
600
R
NT
Hbl K-Ar Ms Rb-Sr

Re
ReER
700
Ms K-Ar
300 Bt K-Ar
600
Apatite fission track
500 C 0
D
1.0 0.8 0.6 0.4 0.2 0.0 70 60 50 40 30 20
Fe/(Fe+Mg) Age (Ma)
Figure 11. Example of geospeedometry using major element zoning. (A–B) Data and models for two
garnets from the same sample from the Valhalla metamorphic complex, British Columbia, analyzed and
interpreted in two different studies: Ducea et al. (2003) and Spear (2004). Towards garnet rims, XAlm and
Fe/(Fe + Mg) increase dramatically, XPrp decreases dramatically, XSps increases, and XGrs either remains flat
or increases slightly. Prp = pyrope, Sps = spessartine, Grs = grossular, Alm = almandine. (A) A ReER model
with a cooling rate of 5 °C / Ma (dashed black line) fits data (Ducea et al. 2003). (B) A ReNTR model with
a cooling rate of 150 °C / Ma fits a similar set of data (Spear 2004). (C) Temperature vs. composition curves
for different equilibria: Fe/(Fe + Mg) changes rapidly with temperature when governed by ReNTR equi-
librium and slowly when governed by ReER equilibrium. Modified from Spear (2004). (D) Temperature
vs. time paths determined from garnet zoning models and from independent petrochronologic and ther-
mochronologic data. Modified from Spear (2004); most chronologic data from Spear and Parrish (1996).
Mnz = monazite, Ttn = titanite, Hbl = hornblende, Ms = muscovite, Bt = biotite.
Ducea et al. (2003) calculated relatively slow cooling of ~5 °C /

Ma, assuming a
simple Fe–Mg ReER between matrix biotite and garnet rims (Fig. 11A). However, close
examination of minerals found at garnet rims, chemical zoning within garnets, and textural
examination of garnets demonstrate operation of a ReNTR (Spear 2004). For example, the
mineral assemblage sillimanite + biotite + plagioclase + quartz occurs along garnet rims,
and is likely a retrograde product after the reactants garnet +
melt
+ K-feldspar (Spear
2004). Increases in Mn towards the rims of garnets (Fig. 11A–B; Ducea et al. 2003; Spear
2004) further indicate operation of a ReNTR because the garnet–biotite ReER affects only
Fe–Mg, not Mn (Spear and Florence 1992; Spear 2004). Assuming that the retrograde reaction
garnet + melt + K-feldspar = sillimanite + biotite + plagioclase + quartz defined the boundary
condition at the garnet rim, an initial cooling rate of ~150 °C / Ma for < 1 Ma was calculated
(Fig. 11B; Spear 2004; the reported range of possible cooling rates was 100–500 °C / Ma).
The main reason calculated cooling rates differ so substantially between studies is that
the ReNTR yields a radically different equilibrium temperature-composition path (much larger
dX / dT in garnet) than the ReER (Fig. 11C). This T–X sensitivity requires that the measured
change in garnet rim compositions occurred over a much smaller temperature interval and
at higher temperature: DT ~ 75 °C between 825 and 750 °C for the ReNTR vs. DT ~ 275 °C
between 825 and 550 °C for the ReER (Fig. 11C; Spear 2004). Recalling that the length of a
diffusion profile generally scales as (D·Dt)1/2, and that D increases exponentially with increasing
temperature (Eqn. 3), average D for a high-temperature ReNTR must be far larger than for a
more protracted ReER (Fig. 11C). A larger D implies a smaller Dt and higher cooling rate. This
result does not reflect any complications resulting from a moving boundary such as resorption
of the garnet rim. Although resorption must have occurred (different profiles are not identical),
dX / dT for the ReNTR is simply larger (Fig. 11C), driving a higher calculated cooling rate.
Modeling of Fe–Mg exchange between biotite inclusions and garnets refines the later-stage
cooling history. Although faster initial cooling during operation of the ReNTR produced most
of the Fe–Mg–Mn profile observed at garnet rims, the ReER continued at lower temperature.
Compositions of biotite inclusions within garnet are sensitive to this later, lower temperature
cooling and Fe–Mg exchange. These data imply lower cooling rates of ~25 °C / Ma for several
Ma between ~600 and ~500 °C (Fig. 11D; Spear and Parrish 1996; Spear 2004).
A wealth of chronologic data permits comparison of different cooling rate models to an
independently determined T–t history (Fig. 11D). Spear and Parrish (1996) summarize most data,
which we updated for muscovite and titanite closure temperatures (Jenkin 1997; Harrison et al.
2009; Kohn 2017); Spear (2004) also reevaluated monazite ages in a petrochronologic context.
Altogether, these independent constraints on cooling history agree better with the ReNTR model
for garnet rim zoning than with the ReER model (Fig. 11D; Spear 2004). As discussed in Spear
(2004), rapid cooling is the expected consequence of rapid thrusting along a ramp. Thrust rates
of cm / yr along ramps of ~20° slope should cool rocks at rates of ~100 °C / Ma or greater. Such
rapid cooling may be more common in orogens than ReER-based geospeedometry would imply.
Geospeedometry from trace element zoning
A key conceptual outcome of Spear (2004) is that defining the temperature dependence of
reactions and other boundary conditions governing rim concentrations plays a crucial role in
interpreting data—different petrogenetic models can lead to radically different cooling rates.
The same cautionary principles apply to trace element zoning—new data from amphibolite-
facies rutile crystals from Catalina Schist amphibolites, California, suggest that different crystals
of the same mineral in a single rock experience different boundary conditions during cooling
(Kohn et al. 2016). Because rock-wide equilibrium is not maintained, a cooling rate cannot be
calculated. Rather, diffusion profiles reflect highly localized equilibria and transport phenomena.
Kohn et al. measured near-rim trace element zoning in LA-ICP-MS depth profiles from
separated rutile grains (Fig. 12A). Although profiles for a specific element can be similar,
many profiles show boundary concentrations and trends that deviate strongly from theoretical
expectations. For example, nearly all models of diffusive resetting follow Dodson (1973, 1986) in
assuming the crystal surface equilibrates with the matrix throughout cooling. If so, Zr concentrations
and Zr-in-rutile temperatures should trend towards zero. Instead only one profile conforms with
this expectation (profile a in sample A14-57b, Fig. 12A), and all other profiles within the same
sample and in other samples show markedly flatter trends (Fig. 12A, B). Moreover, crystals that
have essentially indistinguishable Zr profiles (Fig. 12B) can have completely different Nb profiles
(Fig. 12C), implying that Zr and Nb boundary conditions vary on small scales. Clearly if a rock is
fully equilibrated during cooling, profiles should be consistent among grains.
A single model of diffusive resetting cannot explain all data. For example, crystal orientation
(parallel vs. perpendicular to the c-axis) or spot proximity to crystal tips might explain some
disparities (e.g., profiles a and d, Fig. 12A), but some profiles imply completely different
cooling rates, e.g., ~15 °C / Ma for profile a vs.  100 °C / Ma for profile e (Fig. 12D). The
rock experienced only one cooling history, so if different profiles imply different cooling rates,
cooling rate is not reliably retrievable from these data. Kohn et al. (2016) proposed that flux
limitations could explain many of the disparities among crystals. They modeled two scenarios—
that the matrix supports only a maximum flux throughout its cooling history (defined in terms of
the percent of the total flux experienced for an equilibrium model; Fig. 12E), and that the matrix
not only undergoes high-T flux limitations, but also shuts down all equilibration at some cutoff
temperature, arbitrarily taken as 550 °C (Fig. 12F). These models explain relatively flat profiles
better (profiles c and e, Fig. 12D–F), as well as the trend towards a non-zero Zr concentration
and Zr-in-rutile temperatures ≥ 600 °C on crystal rims (nearly all profiles: Fig. 12A–B).
Two results from this work warrant particular emphasis. First, the average profile says
little about equilibrium behavior or cooling rates. Arguably, only profile a in sample A14-57b
could be modeled to infer cooling history—it is the only one that undoubtedly trends towards
zero Zr concentration, so can be modeled theoretically with the fewest assumptions. None of
the profiles in sample A15-10 conform to simple theory, and instead require (as yet arbitrary)
assumptions about proximal mass fluxes (Fig. 12E). Thus, the improved statistics of averaging
more data would not improve confidence in cooling rate estimates. Second, with one exception
(profile a) Zr data consistently indicate that there is a lower temperature at which equilibration
among matrix minerals simply shuts down. Carlson (2012) concluded similarly—flattening
of trace element zoning in the near-rim region of garnet crystals from the Makhavinekh Lake
Pluton contact aureole is best explained if rim reequilibration ceased (trace element flux dropped
to zero) at some elevated temperature. The two temperatures inferred in these studies differ
considerably (~550 vs. ~770 °C), possibly reflecting physical differences in f H2O (“wetter” for
Catalina, “drier” for Makhavinekh Lake) or rates of cooling (~25 °C / Ma for Catalina, up to
50 °C / Ma for Makhavinekh Lake). Clearly mineral rims do not always maintain equilibrium,
and the circumstances under which they depart from equilibrium deserve further study.
Geospeedometry from chronologic zoning
Grove and Harrison (1999) were the first to attempt to quantify cooling histories using
chronologic zoning. Their analysis of Th–Pb age zoning in Himalayan monazite using ion probe
depth profiling implied cooling of ~100 °C / Ma, assuming diffusion rates from then-current
experimental data (Smith and Giletti 1997). Subsequent experimental (Cherniak et al. 2004;
Gardés et al. 2006, 2007) and empirical (McFarlane and Harrison 2006) investigations of Pb
diffusion rates in monazite, however, suggest much lower diffusivities. The monazite age profile in
these rocks more likely reflects late-stage growth rather than diffusional resetting during cooling.
600 600 800 4000

A14-57b Zr A15-10 Zr 1
A B Temperature (°C) A15-10 Nb C
e 700
400 400 3000
d b 600 2
Similar Similar
200 b 200 Zr Boundary Initial Zr 2000
c Different Similar
a a Concentrations Concentrations 500
0 15 3 Nb Boundary Initial Nb
Distance (µm)
80 µm Raw data Ave data Concentrations Concentrations
Concentration (ppm)
Concentration (ppm)
0 0 0
1000
600 600 100%
600
Cooling Rate (°C/Ma) D High-T flux limit E Low-T zero flux cutoff F
Flux limit
c 10% 10%
Zr Flux
400 100 400 10% 10% 50% 400
c % 10%
% 50%
25 100 50 50
10 0% 0%
0% 5
25 10 100% 10
% 200 100%
200 10 200 100 Zero-flux
Cutoff
Zr Flux
25°C/Ma 50% 25°C/Ma
Concentration (ppm)
Concentration (ppm)
10% 0
0 0
0 5 10 15 0 5 10 15 700 400 0 5 10 15
Temperature (°C)
Distance (µm) Distance (µm) Distance (µm)
Figure 12. Trace element zoning and models for amphibolite-facies rutile crystals from the Catalina subduction complex, California (Kohn et al. 2016). Crystals were sepa-
rated using conventional crushing, and only clean crystal faces were analyzed. Data were collected using LA-ICP-MS depth profiling (photomicrograph in (A)) Illustrative
profiles are shown with darker symbols. (A–C) Some profiles show good internal consistency, but disparities in other profiles from the same sample require different boundary
conditions during cooling. Even crystals with essentially equivalent profiles in one element (Zr) can show radically different profiles in another element (Nb). Inset [between
(B) and (C)] shows temperature profile for averaged data; temperature does not approach zero at rim. (D–F) Models of Zr concentration vs. depth for sample A14-57b showing
that no one model fits all data [profiles a c, d and e in (A)]. Diffusion parallel to the c-axis [lower curves in (D)] is about 10 times faster than diffusion perpendicular to c [up-
per curves in (D)]. Independent chronologic data suggest a cooling rate of ~25 °C / Ma for ca. 10 Ma, but high-T cooling could have been faster or slower. Insets [between (E)
Diffusion: Obstacles and Opportunities in Petrochronology
and (F)] illustrate flux-limitation models. A minority of rutile grains were not flux limited, and their rim concentrations tend towards 0 ppm Zr (e.g., profile a) – these profiles
correspond to most models of diffusional resetting. A majority of grains were flux limited and their concentration profiles are flatter (e.g., especially profile e).
133
A more recent example of diffusion-induced age zoning shows decreasing U–Pb ages
towards the rims of rutile crystals separated from a high-grade gneiss from the Ivrea Zone,
southern European Alps (Fig. 13; Smye and Stockli 2014). Multiple depth profiles were
collected using LA-ICP-MS. All depth profiles show ages that increase from ~150 Ma near the
rim to 170–200 Ma towards cores (Fig. 13A). Averaged data were inverted using a (Markov
chain) Monte Carlo method with goodness of fit assigned using metrics of Ketcham (2005).
Pooled inversions yielded a best-fit T–t history and confidence intervals, suggesting rapid
cooling between ~185 and ~175 Ma, with slower cooling thereafter (Fig. 13B).
Although not emphasized in the original study, depth profiles from different rutile grains
actually yield statistically distinct age trends, much as was observed for trace element zoning
in Catalina Schist rutile. For example, although all profiles converge towards rim ages of
~150 Ma, some profiles between ~10 and ~30 µm display ages of ~200 Ma vs. ~170 Ma without
any age overlap (Fig. 13A). Taken individually, these profiles imply quite different cooling
histories, well outside the statistical bounds inferred for the averaged data (Fig. 13B). Because
rutile crystals were collected from the same rock, they should all yield indistinguishable age
profiles and T–t histories, whereas clearly they do not. Either some crystals were growing
during cooling (so the profiles do not reflect diffusion alone), or the boundary conditions
for Pb diffusion were not identical among different grains. Collecting trace element data,
especially T-sensitive Zr, simultaneously with U–Pb ages might help resolve some of these
ambiguities. Although Zr and Pb do not diffuse at the same rate (Cherniak 2000, Cherniak et
al. 2007a), differences in boundary conditions or possible rim growth during cooling might be
identifiable from the shapes of profiles. For example, a broader and younger profile in U–Pb
age might correspond with lower Zr concentrations, suggesting late-stage, lower-T growth.
225 Core ave.~185 Ma 700

Average Cooling
Max.
T-t Bounds
Temperature (˚C)
Ave.
U-Pb Age (Ma)
Maximum
Minimum
600
e
175
Averag
profile
profile
profile
Min.
laser Average
Raw data Best-fit T-t
125 Wtd ave. 500
core
A Mount
B
75 400
0 10 20 30 130 150 170 190 210
Distance (µm) Time (Ma)
Figure 13. Example of geospeedometry using chronologic zoning. Age profiles and T–t histories determined
via depth profiling of rutile crystals (each grain represented by different colored symbols) from a single rock
collected from the Ivrea Zone, southern European Alps. Raw data from Smye and Stockli (2014). Modified
from Kohn (2016) and Kohn et al. (2016). (A) U–Pb age profiles show decreasing ages towards rim, but sev-
eral profiles are analytically distinct. Although rim ages appear to converge, boundary conditions could not
have been equivalent among grains during initial cooling. Inset schematically illustrates how depth profiling
data are collected using LA-ICP-MS. Min = minimum, Max = maximum, Wtd ave = weighted average. (B)
Average cooling history (dark curve) and 95% confidence limits (dashed curve) calculated using a Monte
Carlo approach and diffusion rates of Cherniak (2000). From Smye and Stockli (2014). Maximum and mini-
mum T–t histories are more schematic and are based on bounding profiles from Figure 13A.
Reaction rates
Lucassen et al. (2010) introduced the geosciences to the mathematics of steady-state
reaction as a basis for determining reaction rates (Jackson 2004), and provided data from an
amphibolitized eclogite that can be used to test model predictions. Rutile porphyroblasts grew
under eclogite-facies conditions in these rocks (~700 °C, 2 GPa), then titanite overgrew rutile
in the amphibolite facies (~700 °C, 1 GPa; Ravna and Roux 2006; Fig. 14A, B). Determination
of reaction rates assumes equilibrium between titanite and rutile (Fig. 14C, see also Fig. 8B,
D), but the titanite and rutile may not have been in equilibrium with the matrix (forming
chemically zoned titanite: Jackson 2004; Fig. 8B), or they could both have equilibrated with
the matrix (forming homogeneous titanite; Fig. 8D). We refer to these different models as
“closed system” when Ti and trace elements are wholly conserved between rutile and titanite
and do not exchange with the matrix (Fig. 8B), and “open system” when titanite and the rutile
grain boundaries completely equilibrate with the matrix (Fig. 8D). The open system model
implies that the matrix can serve as a source or sink of major and trace elements. The molar
volume of rutile is approximately 1/3 that of titanite, and the mass ratio is ~40%.
The closed system model predicts that a Zr compositional minimum should form in
titanite 1/3 of the distance from the rutile–titanite interface to the edge of the titanite. This
composition reflects that of the first titanite to form in equilibrium with rutile at the original
position of the rutile crystal edge. If rutile has an initial concentration C0, first-formed titanite
should have a concentration k·C0 (Fig. 8B), where k is the trace element partition coefficient
between titanite and rutile (at T ~ 700 °C, k ~ 0.08 for Zr as measured in ppm by weight;
Watson et al. 2006, Hayden et al. 2008). To maintain mass balance, steady-state titanite should
have a trace element concentration 0.4·C0 (Fig. 14C), and the rutile crystal edge should have
concentration 0.4·C0 / k (or, for Zr, ~0.4·C0 / 0.08 = 5·C0). A compositional well develops as
titanite progressively grows inward and outward from the original titanite–rutile interface,
with a composition that evolves from k·C0 (~0.08·C0 for Zr) to 0.4·C0.
Under open system conditions with whole-rock equilibration, titanite trace element
concentrations should be homogeneous and consistently reflect the temperature of reaction,
whereas the rim of the rutile should adopt a commensurate composition (Fig. 14C). Because
Zr contents of rutile increase with decreasing pressure (Tomkins et al. 2007), a rutile rim
in equilibrium with the matrix should adopt a higher Zr concentration than in the core,
inducing a diffusion profile.
Data from rutile crystals indeed show that trace element concentrations increase
towards rutile–titanite interfaces (Fig. 14D, E), as expected from either closed- or open-
system equilibrium models (Fig. 14C). Such increases are not limited to the more abundant
trace elements (Zr and Nb), and include low-abundance geochemically similar elements
(Hf and Ta). In detail, however, several observations contradict the closed-system model:
titanite does not exhibit a compositional well, Zr concentrations are 2–3 times too high,
and Nb concentrations are slightly too low (Fig. 14D, E). Lucassen et al. (2010) recognized
the problems with mass conservation, noting that there was insufficient titanite developed
adjacent to rutile to balance Ti, and that Zr concentrations in titanite were far too high to
conform with Equation (15). Rather, Ti must have been exported to the matrix to produce
additional titanite, while Zr may have been imported from zircon breakdown or possibly
pyroxene to support the high Zr content of the titanite and produce the Zr diffusion profile
(Lucassen et al. 2010). Major and trace element import–export is unsurprising because the
reaction zone must import Ca and Si to form titanite from rutile. Significantly, the data
also do not meet expectations of open system equilibration among rutile, titanite and matrix
minerals. Zr-in-titanite and Zr-in-rutile temperatures (at the edge of the rutile) cannot be
reconciled, even accounting for potentially lower aSiO2 and aZrSiO4 at the reaction interface.
Amphibolite
matrix
Polycrystalline B Relict boundary
titanite
Relict
Rutile
Rutile
Rutile Titanite overgrowth

megacryst
Titanite Closed system, equilibrium Open system, equilibrium
Trace eleme megacryst Titanite 0.4·Co/k Titanite T = Rxn T
nt profile
boundary
boundary
Profile 2 Co Co
Relict
Relict
Rutile Rutile
Broken
0.4·Co
Polycrystalline T = Rxn T
titanite 1 mm A C k·Co
800 800
Polycrystalline D E Moving interface
titanite Titanite
Nb, Zr (ppm)
600 600 Fixed boundary
Nb (ppm)
Rutile megacryst megacryst Measured data
400 400
Nb
Co
200 Average 200
Zr Profile 2 0.4·Co
0 0
425±25 ppm = 680±5 °C

Ta, Hf (ppm)
20 400
Ta
Zr (ppm)
300
10 200
Average
Hf Profile 2 100 Co
0.4·Co 90±10 ppm = 738±6 °C
0 0
0 2000 4000 6000 8000 0 1000 2000 3000 4000
Distance (µm) Distance (µm)
Figure 14. Data and models relevant to steady-state reaction and induced diffusion profiles. All data and
model fits from Lucassen et al. (2010). (A) Sketch of coarse rutile with titanite overgrowths, amphibo-
litized eclogite, Tromsø Nappe, Norway. (B) Sketch illustrating growth of titanite at the expense of rutile,
with associated volume change due to the difference in molar volumes. Small arrows show direction of
titanite overgrowth. (C) Schematic models of reaction-induced diffusion profiles for rutile with titanite
overgrowths. Equilibrium is maintained at the rutile–titanite interface, but may be closed system (does
not equilibrate with the matrix; Jackson 2004; Fig. 8B) or open system (equilibrates with the matrix; Fig.
8D). For open system equilibration, trace element concentrations at the titanite–rutile boundary should cor-
respond to equilibrium concentrations for the temperature of reaction. (D) Data from trace element profiles
in Figure 14A, showing steep gradients near rutile rims, and relatively flat profiles in titanite. The spatial
distribution of analyses for Profile 2 was not reported so an average and 2 s.e. is shown for comparison with
model predictions. Different size data points were collected with different spot sizes and spatial resolutions.
(E) Detail of left-hand sides of profiles in Figure 14D, fitted to a moving interface solution (Eqn. 15) or a
fixed boundary condition (error function). Concentrations of Nb and Zr in titanite do not conform with ex-
pectations of a closed-system model, implying that the matrix served as a source or sink of trace elements.
Interface concentrations of Zr do not yield the same temperature, implying disequilibrium partitioning.
Overall, no steady-state reaction model appears completely reconcilable with these data.
While a steady-state solution fits the rutile diffusion profile reasonably well, so, too, does a
fixed boundary (error function) solution (Fig. 14E). Physically, this means that reaction rates
could have been so fast that no diffusion occurred in rutile during reaction, and only afterwards
did a fixed boundary condition induce a diffusion profile (Fig. 8E). Thus it is impossible to
know whether the diffusion profiles developed during the reaction of rutile to form titanite, or
were superimposed subsequently (Lucassen et al. 2010). Additional profiles collected for several
other examples of titanite overgrowths on rutile, including rocks of the Tromsø nappe, Norway
(Cruz-Uribe et al. 2014), show similar trends and incompatibilities with steady-state models.
The consistent shape of diffusion profiles documented by Lucassen et al. (2010) and Cruz-Uribe
et al. (2014) does suggest some consistent petrologic process, but despite good fits to diffusion
profiles using Equation (15), calculations of reaction rates should be viewed with caution.
Timescales of magmatic processes from zoning in crystals

Major and trace element profiles across igneous phenocrysts, including olivine, feldspar,
and quartz, constrain timescales of magmatic processes including phenocryst residence times,
duration of magma mixing events, and the time between a magma rejuvenation event and
eruption. Costa et al. (2008) summarize theory, applications, and potential pitfalls in using
diffusion profiles to constrain timescales of magmatic processes. Recent investigations have
taken advantage of in situ methods, especially NanoSIMS. This increasingly high chemical
and spatial resolution affords new insight into very short-duration events.
Minerals often develop oscillatory zoning as they grow from a melt, sometimes because of
extrinsic factors like magma convection, mixing, or rejuvenation, but also because of localized
undercooling and intrinsic mineral growth kinetics (see summary of Shore and Fowler 1996).
Diffusional smoothing of potentially sharp oscillatory boundaries allows calculation of crystal
residence times. The scale of zoning can be quite small, however, so quantifying the original
compositional step and measuring compositional gradients accurately can pose a major
analytical challenge. Oscillatory-zoned quartz phenocrysts from a Jurassic rhyolite of the El
Quemado Complex (Chon Aike large igneous province), Patagonia, provide an opportunity
to constrain crystal residence times in a silicic system (Fig. 15A, B). Although Ti diffusion in
quartz is not strongly anisotropic (Cherniak et al. 2007b), for consistency Seitz et al. (2016)
first imaged crystals using microtomography, then sectioned all crystals perpendicular to their
c-axes. NanoSIMS measurements of Ti concentrations along traverses that cross boundaries
between oscillatory bands reveal smoothly varying changes in Ti over scales of less than ten
microns (Fig. 15B; Seitz et al. 2016). Diffusion modeling of several profiles consistently
suggests residence times of four to six years for these quartz crystals in the magma.
Overgrowths on igneous phenocrysts can also be used to calculate timescales of magmatic
processes, especially the time between magma rejuvenation and eruption, but with potential
complications. Some common crystals, such as olivine, exhibit diffusional anisotropy (diffusion
of divalent cations parallel to the c-axis is ~6 times faster than parallel to the a- and b-axes;
Chakraborty 2010) while the 3-dimensional morphology of other crystals, such as plagioclase,
can influence diffusion profiles (Costa et al. 2003). When diffusion profiles develop on scales
approaching a significant proportion of a crystal, 2- and 3-dimensional models of diffusion may
become necessary as errors in applying 1-dimensional models can range from factors of 0.1–25
(Shea et al. 2015a). Examples of 2- and 3-dimensional modeling of Fe–Mg zoning in olivine
indicate timescales of months to years between magma rejuvenation and eruption (Costa and
Dungan 2005; Shea et al. 2015b). The study of Shea et al. (2015b) is novel in using P zoning
(P is virtually inert to diffusion) to identify the original core–rim boundary, which is needed for
modeling, but difficult to identify from the distribution of fast-diffusing Fe and Mg alone.
At a finer scale, zoning in Ba, Sr, and Mg was measured by NanoSIMS across core-
rim overgrowths from sanidine crystals of the Scaup Lake rhyolite, Yellowstone caldera
(Fig. 15C–D; Till et al. 2015). Like the study of Shea et al. (2015b), analysis of multiple elements
with different diffusivities confers significant modeling and interpretational advantages.
Specifically, Ba diffuses about 1000 times more slowly than Mg, and about 70 times more
slowly than Sr (Cherniak 1996, 2002; LaTourette and Wasserburg 1998). Thus, if overgrowths
formed with sharp compositional steps in all elements, the steepness of the compositional
gradient should increase from Mg to Sr to Ba, and diffusion modeling of the gradient should
yield the same apparent duration. Not so, however! Assuming an initial compositional step,
gradients imply quite different timescales, ranging from ~1000 years (Ba) and ~15 years (Sr)
to ~0.8 years (Mg; Fig. 15D). If diffusion occurred for 1000 years (to match the Ba profile),
Mg and Sr profiles should be considerably flatter, whereas if diffusion occurred for 0.8 years
(to match the Mg profile), Ba and Sr profiles should be considerably steeper (Fig. 15D).
Evidently, concentrations of these elements changed continuously, not discontinuously, as
overgrowths formed and the melt evolved. Crystal growth produced compositional gradients,
A B 1.0
33.0
Quartz 0.8
5.6±2.2 yr 32.0
Ti(ppm)
Ti/29Si
0.6
0.4
48
100 µm 0.2 29.0
±1σ
28.0
0.0
0 5 10 15 20 25
Distance (µm)
7000
Sanidine C D Ba (ppm) 0.8
Growth 15 1000
4000
120
45 µm Sr (ppm) 0.8
15
profile Growth 1000
60
8
Mg (ppm) 0.8
500 µm 15 1000
Growth?
4
20 25 30 35
Core Rim
Distance (µm)
Figure 15. Timescales of magma chamber processes from zoning in crystals as measured using NanoSIMS.
(A) Cathodoluminescence image of quartz phenocryst, El Quemado Complex rhyolite, Patagonia, showing
oscillatory zoning. Crystal is sectioned perpendicular to its c-axis. Small vertical lines show locations of
NanoSIMS traverses where Ti / Si was measured. Modified from Seitz et al. (2016). (B) NanoSIMS data
showing calculated error function fit assuming an initial ~2 ppm step in Ti content. Width of symbols is equal
to the spatial resolution of analyses. Modified from Seitz et al. (2016). (C) Sketch of backscattered electron
image of sanidine phenocryst from Yellowstone caldera rhyolite, showing distinct rim that has overgrown
core during magma rejuvenation. Tiny line within circle shows location of NanoSIMS traverse where Ba,
Sr and Mg concentrations were measured. Thin lines illustrate compositional banding. Based on Till et al.
(2015). (D) Trace element profiles along core–rim interface and diffusion models for 1000, 15, and 0.8 yr
durations (numbers next to models). Although each profile can be fit assuming diffusional relaxation of a
compositional step, similar compositional gradients when D differs by as much as 3 orders of magnitude
imply irreconcilable timescales. The core–rim compositional change for Sr and Ba reflects growth, whereas
the Mg profile probably reflects a combination of growth and diffusional relaxation (i.e., the model for Mg
provides a maximum estimate of 0.8 Myr for the duration of diffusion). Data from Till et al. (2015).
not steps, and in fact the profiles for Sr and Ba must be virtually unmodified by diffusion.
Each element provides a maximum time limit for diffusion—< 1000 years for Ba, < 15 years
for Sr, and < 0.8 years for Mg—but the limit from Mg is most strict. Clearly if only Ba or Sr
had been measured, a significantly different timescale would have been inferred. The resulting
interpretation reveals much shorter timescales between the magma rejuvenation events and
eruption (< 0.8 years; Till et al. 2015) and implies that geophysical evidence for increasing
melt fraction beneath volcanoes may necessitate remarkably fast response to mitigate hazards.
Magma ascent rates from zoning in glass
Several studies have used diffusion profiles of volatile species in glasses to constrain
rates of magmatic processes (e.g., Castro et al. 2005; Liu et al. 2007; Humphreys et al. 2008;
Lloyd et al. 2014). Melt embayments in igneous phenocrysts provide an unusual opportunity
to constrain magma ascent rates because they can trap melt, which will then attempt to expel
volatile components such as H2O and CO2 during magma ascent and decompression. Bubble
nucleation and dispersion in the melt outside the phenocryst should promote melt–bubble
equilibrium in the magma body. For melt that is trapped in an embayment, however, volatiles
must diffuse through the embayed melt to the nearest bubble to achieve equilibrium. Slow
magma ascent allows the interior of embayed melt to equilibrate its volatile content, whereas
rapid ascent does not. Thus, the H2O and CO2 content of the interior of embayed glass and the
shape of the H2O and CO2 composition profiles within it can potentially constrain ascent rates
(Liu et al. 2007; Humphreys et al. 2008; Lloyd et al. 2014).
Lloyd et al. (2014) measured composition profiles in major elements and volatiles (F, Cl,
S, H2O and CO2) in basaltic glasses that were trapped in embayments in olivine recovered from
basaltic tephra, Volcán de Fuego, Guatemala (Fig. 16A–B). During ascent, a bubble formed
at the outlet of each embayment, establishing a boundary condition of steadily decreasing
concentrations of H2O and CO2 on the margin of the embayed melt as the magma ascended
and degassed. Decreasing trends of H2O and CO2 in the glass (Fig. 16B) were interpreted to
reflect diffusion of volatiles towards the melt–bubble
Fig.interface.
16 Kohn Major element compositions
and Penniston-Dorland
also change systematically toward the embayment outlet, likely because of concurrent
crystallization of olivine and pyroxene proximal to the bubble.
Modeling the H2O and CO2 profiles requires establishing initial H2O and CO2 contents
of the original melt, modeling the pressure dependence of the volatile content of the melt
in equilibrium with a gas bubble, and establishing the diffusivity of H2O and CO2. Contents
of S, H2O and CO2 broadly correlate in melt inclusions from the same eruptive rocks
(Lloyd et al. 2013), so the S content of the interior of the embayed glass, where it appears
unaffected by diffusion, and the melt inclusion correlations for S–H2O and for S–CO2 were
used to constrain the original H2O and CO2 contents. The boundary condition was modeled
isothermally (1030 °C) as a function of pressure. Diffusivities were modeled at a fixed
temperature but varied systematically with composition because H2O diffusivity depends
somewhat on major element composition (the glass is chemically zoned), and the diffusivity
of CO2 depends on H2O content (Zhang et al. 2007).
5.0 500
A CO2 initial B
4.0 400
Olivine H2O initial
Phenocryst
CO2 (ppm)
H2O (wt%)
1-stage:
3.0 11.3 m/s, 12.6 min 300
1- or 2-stage
Ion probe al
CO2, H2O
2.0 fin 200
Bubble O l
e H2 fina
Electron prob CO 2
1.0 2-stage: 100
Glass embayment 2.1 m/s, 25.8 min
100 µm 10.6 m/s, 7.3 min
0.0 0
0 50 100 150 200
Bubble wall Olivine
Distance (µm)
Figure 16. Timescales of magma ascent from zoning in glass (modified from Lloyd et al. 2014) (A)
Sketch of backscattered electron image of olivine phenocryst with basaltic glass trapped in an embay-
ment that opens to a gas bubble. Squares and dots show locations of ion probe and electron probe compo-
sitional analyses. Inset shows larger view of olivine crystal and its embayment. (B) Composition profiles
of H2O and CO2 in embayment glass, showing gradual decreases towards embayment edge (bubble).
Initial H2O and CO2 concentrations are based on melt S content in the context of melt inclusion composi-
tions. H2O profile can be fit with either a 1-stage or 2-stage ascent history, but CO2 profile implies a more
complex history with accelerating ascent.
In general, H2O and S profiles can be explained with either a single-stage or multi-stage
ascent history, whereas the low CO2 content in the embayment interior but steep gradient near
the bubble implies accelerating ascent. Some second-order effects that could be considered
in future studies include loss of H2O to the olivine crystal (although this should be much
slower than diffusion through the melt; Gaetani et al. 2012) and concurrent changes in melt
composition during ascent. Use of a fixed, chemically evolved profile, as is observed in the
major element profile in the glass, probably underestimates ascent rates slightly because
diffusivities decrease in more silicic melts (Zhang and Ni 2010). Because of the tradeoff
between D and Dt in solutions to the diffusion equation, a smaller D assumed for the fully
evolved melt leads to a slightly larger Dt and lower calculated ascent rate.
Duration of metamorphism
Once diffusion coefficients are known for minerals, the duration of metamorphic processes
can be constrained by evaluating the variations in the elemental or isotopic composition along
profiles across mineral grains. Ague and Baxter (2007) modeled variations in Sr concentrations
in apatite from the classic Barrovian zones of northeast Scotland (Fig. 17A–B), as well as
major-element zoning in garnet (Fig. 17C–D), to constrain the duration of thermal events
during the regional metamorphism that accompanied the ~465 Ma Grampian orogeny. For
both minerals, a compositional step was assumed, establishing an initial diffusion couple, and
diffusion profiles were fitted to measured compositional gradients to estimate the duration of
peak metamorphism. For apatite, inclusions in garnet show the same type of zoning as matrix
grains, implying that rims formed during prograde metamorphism, and that compositional
steps must have experienced the full extent of peak metamorphism. For garnets, a simple
compositional step cannot be demonstrated, but gradual compositional changes during
growth would lead to shorter estimates of the duration of peak metamorphism. The best fits to
these diverse data range from < 100 ka to ~650 ka (Fig. 17A, 17D), regardless of metamorphic
grade, suggesting that a very rapid thermal pulse (or pulses) was responsible for producing
the observed regional metamorphism (Ague and Baxter 2007).
As Ague and Baxter (2007) emphasize, the data do not uniquely constrain the shape
of the T–t path, rather they reflect the integral of the D–t curve. For example, T–t histories
involving a single square pulse (model a, Fig. 17B), multiple instantaneous heat pulses
with thermal relaxation (model b, Fig. 17B), or a more gradual heating-cooling history
(model c, Fig. 17B) provide identical fits to the data (Fig. 17A), because their integrals
of D·Dt are identical. Traditionally, crustal thickening has been thought to produce much
longer timescales of metamorphism ranging from several Ma to several tens of Ma (e.g.,
England and Richardson 1977; England and Thompson 1984). Thus, many workers infer a
contact metamorphic origin based on petrochronologic and thermochronologic data (Ague
and Baxter 2007; Viete et al. 2013). This conclusion would be highly ironic insofar as
“Barrovian metamorphism” is virtually synonymized with crustal thickening, not contact
metamorphism, but would also accord with Barrow’s original view that igneous intrusions,
not crustal thickening, provided the heat to metamorphose these rocks (Barrow 1893;
see also Harte and Hudson 1979). As an alternative to the view that brief thermal pulses
require plutons, inversion of diffusion profiles from quartz inclusions (Ti) and garnet rims
(Fe–Mg–Mn) suggests similarly brief durations of regional metamorphism in the northern
Appalachians where intrusions are wholly lacking (Spear et al. 2012, Spear 2014). These
results imply that brief regional metamorphic heating might result from discrete, tectonically
driven events, e.g., from thin thrust slices or duplexing. Disentangling thin-skinned thrusting
from other types of thermal pulses will require further petrochronologic investigations.
0.25 Core rim

A ?
25 µm
600
B 4.0
0.20 500 Model a 2.0
Diffusivity x 1016 (m2/s)

s
Temperature
le
Temperature (°C)
ofi
Pr
Diffusivity
,c
0.15 600 4.0
Sr (wt%)
,b 230 ka
la
Model b
de
Data 500 2.0

Mo
0.10
Profile
600 4.0
0.05
Initial 500 Model c 2.0
0.00
0.0 20 40 60 80 0.0 0.5 1.0 1.5 2.0 2.5 3.0
Distance (µm) Time (Ma)
0.06 0.10
Assumed Assumed
0.04 boundary boundary
XSps
XPrp
Model 0.08 1 Ma
0.02 region
C D 68 ka
0.00 0.06
0.6 1.0 1.4 1.8 2.2 1.2 1.3 1.4 1.5 1.6
interior Distance (mm) rim Distance (mm)
Figure 17. Diffusion profile-based estimates of the duration of metamorphism determined from zoning
in apatite and garnet from Barrovian type region, Scotland. Modified from Ague and Baxter (2007). (A)
Strontium zoning in apatite with 3 diffusionally equivalent models that correspond with different T–t his-
tories. Backscattered electron image shows the core–rim boundary (dashed where inferred) and location
of line traverse. (B) Models of duration of thermal pulses. Several different T–t histories yield equivalent
extents of diffusional resetting. (C) Manganese zoning in garnet showing typical decrease outward from
core, and increase at rim. Dashed line is location of assumed original compositional step. Gray = modeled
region; Sps = spessartine. (D) Magnesium zoning in garnet from the modeled region, showing good fit for a
duration at peak metamorphic conditions of ~68 ka, and poor fit for 1 Ma duration. Prp = pyrope.
Fluid–rock interactions
The duration and physical scale of fluid–rock interactions during metamorphism can be
assessed by investigating whole-rock variations of fluid-mobile elements and isotopes. Most
studies emphasize geochemical traverses that cross features such as veins or reaction zones
that are associated with fluid–rock interaction. Such investigations focus on the record left
behind by diffusion through a porous rock medium. Recent research emphasizes Li because
it is highly soluble in metamorphic fluids, and its low valence and small ionic radius allow it
to diffuse readily in a variety of materials. Lithium also enjoys the unusual characteristic that
the large relative mass difference between its isotopes causes significant differences in 6Li
and 7Li diffusivities. This leads to kinetic fractionation of Li isotopes, such that incomplete
or unidirectional processes (e.g., diffusion along a chemical gradient, evaporation, etc.)
transport 6Li more rapidly than 7Li and produce isotopic anomalies. The difference in diffusion
coefficients between 6Li and 7Li is characterized by an empirical term b, defined as:
β
D6Li m 
=  7Li  (16)
D7Li m
 6Li 
(Richter et al. 1999), where m is the mass of each isotope. In experiments, kinetic isotopic
fractionations occur where 6Li diffuses up to 3% faster than 7Li in both silicate melt and water,
causing fractionations of up to tens of permil (Richter et al. 2003, 2006). This difference in
diffusion rates would not be detected in a concentration profile, but can be detected in profiles
of measured isotopic ratios (d7Li). Thus, unlike more traditionally studied elements whose
concentration gradients in rocks may be small (e.g., O), or whose isotopes diffuse at virtually
the same rate (e.g., Sr), Li develops two different profiles—Li concentration and d7Li that
permit simultaneous quantification of transport rates or durations.
Whole-rock Li concentrations and d7Li across fluid-related features such as veins (e.g.,
John et al. 2012), contacts with igneous rocks (Fig. 18A; e.g., Teng et al. 2006, Marks et al.
2007, Liu et al. 2010, Ireland and Penniston-Dorland 2015), or rock features diagnostic of
fluid infiltration (e.g., relatively “dry” eclogite altered to “wet” blueschist, Penniston-Dorland
et al. 2010; Fig. 18B–E) show isotopic fractionations up to ~30‰ and indicate transport
distances up to tens of meters. An unusual, key feature of some data is that preferential
transport of 6Li means that d7Li values can shift more than the initial range between two rocks.
For example, d7Li values of amphibolites next to the Tin Mountain pegmatite (Black Hills,
South Dakota) approach −20‰, even though the initial amphibolite and pegmatite values were
0 to +10‰ respectively (Fig. 18A; Teng et al. 2006). Diffusion through an interconnected
intergranular fluid, rather than through minerals, dominates transport of Li concentrations
and isotopic compositions, which depends on time, effective diffusivity, permeability,
temperature, and tortuosity (Fig. 5). Thus, if physical factors can be estimated, Li can be
used as a geospeedometer to record the duration of fluid–rock interactions. Correlation of the
variations in isotopic composition with petrologic indicators of fluid–rock interaction aids
in the interpretation of such profiles. For example, in the alteration of eclogite to blueschist
(Penniston-Dorland et al. 2010), the replacement of anhydrous omphacite and garnet in the
eclogite by hydrous pumpellyite and chlorite in the blueschist (Fig. 18C–E) was associated
with variations in Li concentrations and isotopic compositions.
Durations of fluid infiltration events inferred from Li concentration and isotope
measurements may be quite short, for example only 10s to 100s of years during subduction
metamorphism (Fig. 18B; Penniston-Dorland et al. 2010; John et al. 2012). Such pulsed
flow may have implications for formation of melts in arcs, where melt transport may be
pulsed on comparable timescales (Turner et al. 2001). A direct comparison of variations in
oxygen and lithium isotopic compositions across a contact between the Bushveld Complex
and metasedimentary rocks of the Phepane Dome demonstrates much faster diffusion of Li
relative to O (Ireland and Penniston-Dorland 2015). Along this traverse, O isotopes vary over a
distance < 1 m, while Li isotopes vary over a distance of ~100 m. These distances translate into
estimates of a duration for the emplacement and cooling of the Bushveld Complex of 5 kyr to
5 Myr, depending on choices of diffusivities and porosities.
FUTURE DIRECTIONS
Ever improving in situ technologies have dramatically refined our ability to investigate
diffusive processes in nature and constrain rates of petrogenetic processes. For example,
NanoSIMS has increased spatial resolution (Fig. 15; Schmitt and Vazquez 2017) while LA-
ICP-MS now allows composition profiles to be measured in ~1 minute (data in Fig. 12;
Kylander-Clark 2017). Undoubtedly such analytical improvements will continue to advance
petrochronologic research in the near future. Progress on the following topics would also help
promote more accurate petrochronologic interpretations that invoke diffusive processes:
A 6
Li diffuses faster 10
δ7Li (‰)
0
600 (Dt)1/2 = 3.1m
-10
β=0.12
400 -20
Li (ppm)
Li Initial Li
δ7Li Initial δ7Li
200 model
[Li] Increases
0
0 5 10 15 20 300
Distance from the contact (m)
6.0
B
(Det/Ke)1/2 = 0.13cm
4.0
β=0.12
Eclogite
δ7Li (‰)
δ7Li = - 0.2‰
2.0
0
Blueschist
δ7Li = 4.4‰
-2.0
30
Li concentration (ppm)
Distance (cm)
25 Blueschist
18.4 ppm
Eclogite
24.4 ppm
20
15
-2.5 -2 -1.5 -1 -0.5 0 0.5 1 1.5
Distance (cm)
Eclogite Blueschist
C D E
50 mm 50 mm 500 mm
Figure 18. Constraints on fluid–rock interaction from whole-rock Li concentration and isotope profiles.
(A) Data from amphibolites adjacent to Tin Mountain (South Dakota, USA) pegmatite along with dif-
fusion model curves (modified from Teng et al. 2006). The different curves illustrate how the initial
isotopic composition of the amphibolite does not strongly affect the model outcome. (B) Data across
blueschist-eclogite contact in subduction-related tectonic mélange of the Franciscan Complex (Califor-
nia; modified from Penniston-Dorland et al. 2010). Constant Li and δ7Li within blueschist layer indi-
cate fluid infiltration parallel to the contact that altered the original eclogite to blueschist. Dashed lines
show best-fit pinned boundary diffusion model. (C–E) Mineral textures support fluid-assisted alteration
of eclogite. Eclogite retains relatively pristine minerals, including idioblastic garnet (not shown), but
the blueschist contains altered omphacite (o), and chlorite replaces garnet. a = amphibole, chl = chlorite,
e = epidote, p = phengite, t = titanite.
Boundary conditions
Accurately constraining boundary conditions is the single most important task if we are to
obtain useful petrochronologic information from diffusion profiles. This problem can be framed
in terms of the reactive reservoir of a rock: if mineral rims equilibrate with a reservoir, we must
accurately identify its bounds (whole rock vs. proximal minerals; Lanari and Engi 2017) and
composition (all minerals vs. some minerals, entire grains vs. partial grains, etc.; Baxter and
DePaolo 2002; Sousa et al. 2013). In some instances (e.g., Nb in rutile), we do not know the
reservoir and cannot even predict whether a trace element concentration should increase or
decrease at a mineral rim during cooling. Yet many profiles are quite systematic, lending hope
that we can usefully invert profiles and quantify petrologic processes. Developing methods
for identifying the reactive bulk composition and for evaluating how it changes as a rock
evolves will be key for interpreting such systematic changes in concentration. Although some
theoretical models have been developed for specific problems during prograde metamorphism,
for example Sr isotope evolution (Baxter and DePaolo 2002; Sousa et al. 2013), specific tests
of assumed boundary conditions must be identified, such as the use of Mn profiles in garnet
to assess ReER’s vs. ReNTR’s (Spear 1991; Spear and Florence 1992; Kohn and Spear 2000).
Empirical diffusivities
Diffusivities as measured experimentally and as determined from natural samples do not
always correspond, e.g., Pb and Zr in titanite, O in zircon, and Fe–Mg in dolomite and ankerite.
There is a continuing need to improve comparisons of natural and experimental systems to identify
whether any experiments should be revisited. Errors in natural systems will always be large (only
the most obvious discrepancies will be identifiable), but can be reduced by determining accurate
timescales of metamorphism that are carefully evaluated in an appropriate petrologic context.
Controls on diffusivities
Applying diffusion models to constrain petrogenetic processes requires close understanding
of the numerous factors that control D. The advent of trace element microanalysis makes this
effort imperative because many trace elements have different valences than the major cations for
which they substitute (e.g., REE3+ for divalent cations). Diverse mechanisms can accommodate
the resulting charge imbalance, and D must depend on which other crystallographic sites are
involved, and whether vacancies play a role. For example, DLi in olivine depends on whether Li
is sited octahedrally or interstitially (Dohmen et al. 2010). The diffusivity of trivalent REE in
garnet probably depends on whether charge imbalance is accommodated through substitution
into the cubic site (probably highest D), octahedral site (moderately low D), or tetrahedral site
(lowest D). The proportionation of these substitutions probably depends on P–T conditions and
bulk composition (mineral assemblage), so a full model of REE diffusion may well have to
consider the full P–T–t and mineralogical evolution. How other intensive variables (e.g., fO2 and
chemical potential) affect defect densities and D is also poorly known.
Multiple element / isotope comparisons
Once we develop tight constraints on diffusivities for numerous elements and isotopes,
measurement of poly-elemental profiles should constrain petrogenetic processes and their
rates with increasing accuracy. Studies of igneous phenocrysts (Shea et al. 2015a,b; Till
et al. 2015) illustrate the power of this approach, in that some elements closely reflect
changes to magma chemistry while others more closely reflect rates of cooling and ascent.
Analogously, if a cooling rate is proposed based on the diffusion zoning of one element or
isotope in a mineral, diffusion profiles of all elements in that mineral should give the same
cooling rate (within uncertainty), regardless of their diffusivities.
Tectonics
A major goal of diffusion-based petrochronology is to determine cooling histories and
link them to tectonic processes. Several recent studies have inferred unexpectedly high
cooling rates (Spear 2004, 2014; Ague and Baxter 2007; Spear et al. 2012), not all of them
plausible (Ferry et al. 2015). Some cases implicate inadequate understanding of diffusion
coefficients in natural systems (Ferry et al. 2015), but others are interpreted as resulting
from brief thermal pulses, either magmatic (Ague and Baxter 2007) or tectonic (Spear 2004,
2014; Spear et al. 2012). A key goal of future research will be to determine whether the
interpretations of brief pulses are real, rather than artifacts of poorly understood diffusive
processes, and to examine their tectonic implications. Rapid thermal pulses imply either
that magmas play a much greater role in heating the crust than petrologists have assumed,
or that thin-skinned thrust processes are more common than structural geologists have
inferred. The mounting conflicts between diffusion-based vs. thermal model-based rates of
metamorphism suggest we may be on the verge of a scientific revolution (Kuhn 1962) that,
when resolved, will provide a much better understanding of how crust is assembled.
Thermometry
Because diffusion is thermally activated (Eqn. 3) with an exponential dependence on
temperature, tightly constraining maximum temperatures is crucial for accurate interpretations.
The advent of trace element thermometry has improved the precision with which temperatures
may be determined, but there are still lingering concerns about scales of equilibration and
how much a thermometer may be reset during cooling. Further efforts are needed to hone the
methods by which temperatures are determined.
ACKNOWLEDGMENTS
This research was funded by NSF grants EAR-1321897 and EAR-1419865 to MJK,
NSF grant EAR-1419871 to SPD, and NSF grant EAR-1545903 to MJK and SPD. We
thank Elias Bloch for sending model output for Lu and Hf diffusive resetting in garnet,
Bruce Watson for helpful discussions, and Frank Spear, Rick Ryerson, and Martin Engi
for especially insightful and helpful reviews. Lukas Baumgartner and the University of
Lausanne are thanked for hosting SPD during much of the writing. We thank Frank for
reminding us that we should never be in aReERs on our ReNTR’s insurance.
REFERENCES
Ague JJ, Baxter EF (2007) Brief thermal pulses during mountain building recorded by Sr diffusion in apatite and
multicomponent diffusion in garnet. Earth Planet Sci Lett 261:500–516
Aubaud C, Hauri EH, Hirschmann MM (2004) Hydrogen partition coefficients between nominally anhydrous
minerals and basaltic melts. Geophys Res Lett 31, doi:10.1029 / 2004gl021341
Barrow G (1893) On an intrusion of muscovite–biotite gneiss in the southeastern Highlands of Scotland and its
accompanying metamorphism. Quart J Geol Soc London 49:330–358
Baumgartner LP, Ferry JM (1991) A model for coupled fluid-flow and mixed-volatile mineral reactions with
applications to regional metamorphism. Contrib Mineral Petrol 106:273–285
Baumgartner LP, Rumble D, III (1988) Transport of stable isotopes: I: Development of a kinetic continuum theory
for stable isotope transport. Contrib Mineral Petrol 98:417–430
Baumgartner LP, Valley JW (2001) Stable isotope transport and contact metamorphism. Rev Mineral Geochem
43:415–467
Baxter EF, DePaolo DJ (2002) Field measurement of high temperature bulk reaction rates I: Theory and technique.
Am J Sci 302:442–464
Behrens H, Johannes W, Schmalzried H (1990) On the mechanisms of cation diffusion processes in ternary
feldspars. Phys Chem Miner 17:62–78
Beaumont C, Jamieson RA, Nguyen MH, Lee B (2001) Himalayan tectonics explained by extrusion of a low-
Beaumont C, Jamieson RA, Nguyen MH, Medvedev S (2004) Crustal channel flows: 1. Numerical models
with applications to the tectonics of the Himalayan–Tibetan orogen. J Geophys Res 109, doi:B06406
10.1029 / 2003jb002809
Bear J (1988) Dynamics of Fluids in Porous Media. Dover Publications, New York
Bickle MJ, Baker J (1990) Advective–diffusive transport of isotopic fronts: An example from Naxos Greece. Earth
Planet Sci Lett 97:78–93
Bickle MJ, McKenzie D (1987) The transport of heat and matter by fluids during metamorphism. Contributions to
Mineralogy and Petrology 95:384–392
Bloch E, Ganguly J (2015) 176Lu–176Hf geochronology of garnet II: numerical simulations of the development
of garnet–whole-rock 176Lu–176Hf isochrons and a new method for constraining the thermal history of
metamorphic rocks. Contrib Mineral Petrol 169:1–16, doi:10.1007 / s00410-015-1115-x
Bloch E, Ganguly J, Hervig R, Cheng W (2015) 176Lu–176Hf geochronology of garnet I: experimental determination
of the diffusion kinetics of Lu3+ and Hf 4+ in garnet, closure temperatures and geochronological implications.
Contrib Mineral Petrol 169, doi:10.1007 / s00410-015-1109-8
Bohlen SR, Valley JW, Essene EJ (1985) Metamorphism in the Adirondacks: I. Petrology, pressure and temperature.
J Petrol 26:971–992
Borinski SA, Hoppe U, Chakraborty S, Ganguly J, Bhowmik SK (2012) Multicomponent diffusion in garnets
I: general theoretical considerations and experimental data for Fe–Mg systems. Contrib Mineral Petrol
164:571–586, doi:10.1007 / s00410-012-0758-0
Bowman JR, Moser DE, Valley JW, Wooden JL, Kita NT, Mazdab FK (2011) Zircon U–Pb isotope, d18O and trace
element response to 80 m.y. of high temperature metamorphism in the lower crust: Sluggish diffusion and
new records of Archean craton formation. Am J Sci 311:719–772, doi:10.2475 / 09.2011.01
Brady J (1995) Diffusion data for silicate minerals, glasses and liquids. In: Mineral Physics and Crystallography:
A Handbook of Physical Constants, TJ Ahrens (ed) Am Geophys Union, p 269–290
Brady JB, Cherniak DJ (2010) Diffusion in minerals: an overview of published experimental diffusion data. Rev
Mineral Geochem 72:899–920, doi:10.2138 / rmg.2010.72.20
Cahalan RC, Kelly ED, Carlson WD (2014) Rates of Li diffusion in garnet: Coupled transport of Li and Y+ REEs.
Am Mineral 99:1676–1682, doi:10.2138 / am.2014.4676
Cahn JW (1979) Thermodynamics of solid and fluid surfaces. In: Interfacial segregation. Johnson WC, Blakely J,
(eds). American Society for Metals, Metals Park, OH, p 3–23
Carlson WD (2006) Rates of Fe, Mg, Mn, and Ca diffusion in garnet. Am Mineral 91:1–11
Carlson W (2012) Rates and mechanism of Y, REE, and Cr diffusion in garnet. Am Mineral 97:1598–1618
Cartwright I, Valley JW (1991) Steep oxygen-isotope gradients at marble–metagranite contacts in the Northwest
Adirondack Mountains, New York, USA; products of fluid-hosted diffusion. Earth Planet Sci Lett 107:148–163
Castro JM, Manga M, Martin MC (2005) Vesiculation rates of obsidian domes inferred from H2O concentration
profiles. Geophys Res Lett 32, doi:10.1029 / 2005gl024029
Chakraborty S (1997) Rates and mechanisms of Fe–Mg interdiffusion in olivine at 980°–1300°C. J Geophys Res
102:12317–12331
Chakraborty S (2008) Diffusion in solid silicates: a tool to track timescales of processes comes of age. Ann Rev
Earth Planet Sci 36:153–190
Chakraborty S (2010) Diffusion coefficients in olivine, wadsleyite and ringwoodite. Rev Mineral Geochem
72:603–639, doi:10.2138 / rmg.2010.72.13
Chakraborty S, Ganguly J (1992) Cation diffusion in aluminosilicate garnets; experimental determination
in spessartine-almandine diffusion couples, evaluation of effective binary, diffusion coefficients, and
applications. Contrib Mineral Petrol 111:74–86
Chakraborty S, Rubie DC (1996) Mg tracer diffusion in aluminosilicate garnets at 750–850 degrees C, 1 atm. and
1300 degrees C, 8.5 GPa. Contrib Mineral Petrol 122:406–414
Cherniak DJ (1993) Lead diffusion in titanite and preliminary results on the effects of radiation damage on Pb
transport. Chem Geol 110:177–194
Cherniak DJ (1995) Sr and Nd diffusion in titanite. Chem Geol 125:219–232
Cherniak DJ (1996) Strontium diffusion in sanidine and albite, and general comments on strontium diffusion in
alkali feldspars. Geochim Cosmochim Acta 60:5037–5043
Cherniak DJ (2000) Pb diffusion in rutile. Contrib Mineral Petrol 139:198–207
Cherniak DJ (2002) Ba diffusion in feldspar. Geochim Cosmochim Acta 66:1641–1650
Cherniak DJ (2006) Zr diffusion in titanite. Contrib Mineral Petrol 152:639–647
Cherniak DJ (2010a) Cation diffusion in feldspars. Rev Mineral Geochem 72:691–733, doi:10.2138 / rmg.2010.72.15
Cherniak DJ (2010b) Diffusion in accessory minerals: zircon, titanite, apatite, monazite and xenotime. Rev Mineral
Geochem 72:827–869
Cherniak DJ (2015) Nb and Ta diffusion in titanite. Chem Geol 413:44–50, doi:10.1016 / j.chemgeo.2015.08.010
Cherniak DJ, Dimanov A (2010) Diffusion in pyroxene, mica and amphibole. Rev Mineral Geochem 72:641–690,
doi:10.2138 / rmg.2010.72.14
Cherniak DJ, Liang Y (2007) Rare earth element diffusion in natural enstatite. Geochim Cosmochim Acta
71:1324–1340, doi:10.1016 / j.gca.2006.12.001
Cherniak DJ, Watson EB (2001) Pb diffusion in zircon. Chem Geol 172:5–24
Cherniak DJ, Watson EB, Grove M, Harrison TM (2004) Pb diffusion in monazite: a combined RBS / SIMS study.
Geochim Cosmochim Acta 68:207–226
Cherniak DJ, Manchester J, Watson EB (2007a) Zr and Hf diffusion in rutile. Earth Planet Sci Lett 261:267–279
Cherniak DJ, Watson EB, Wark DA (2007b) Ti diffusion in quartz. Chem Geol 236:65–74
Cohen M (1970) Self-diffusion during plastic deformation. Trans Japan Inst Metals 11:145–151
implications for monazite formation. J Metamorph Geol 26:451–464, doi:10.1111 / j.1525–1314.2008.00769.x
Costa F, Chakraborty S (2008) The effect of water on Si and O diffusion rates in olivine and implications for transport
properties and processes in the upper mantle. Phys Earth Planet Int 166:11–29, doi:10.1016 / j.pepi.2007.10.006
Costa F, Dungan M (2005) Short time scales of magmatic assimilation from diffusion modeling of multiple
elements in olivine. Geology 33:837, doi:10.1130 / g21675.1
Costa F, Chakraborty S, Dohmen R (2003) Diffusion coupling between trace and major elements and a model
for calculation of magma residence times using plagioclase. Geochim Cosmochim Acta 67:2189–2200,
doi:10.1016 / s0016-7037(02)01345–5
Costa F, Dohmen R, Chakraborty S (2008) Time scales of magmatic processes from modeling the zoning patterns
of crystals. Rev Mineral Geochem 69:545–594, doi:10.2138 / rmg.2008.69.14
Cottrell AH, Bilby BA (1949) Dislocation theory of yielding and strain ageing of iron. Proc Phys Soc A 62:49–62
Crank J (1975) The Mathematics of Diffusion. Oxford University Press, London
Cruz-Uribe AM, Feineman MD, Zack T, Barth M (2014) Metamorphic reaction rates at ~650–800°C from diffusion
of niobium in rutile. Geochim Cosmochim Acta 130:63–77, doi:10.1016 / j.gca.2013.12.015
Cygan RT, Lasaga AC (1985) Self-diffusion of magnesium in garnet at 750° to 900 °C. Am J Sci 285:328–350
Demouchy S, Mackwell SJ, Kohlstedt DL (2007) Influence of hydrogen on Fe–Mg interdiffusion in (Mg,Fe)O and
implications for Earth’s lower mantle. Contrib Mineral Petrol 154:279–289, doi:10.1007 / s00410-007-0193-9
Dodson MH (1973) Closure temperature in cooling geochronological and petrological systems. Contrib Mineral
Petrol 40: 259–274
Dodson MH (1986) Closure profiles in cooling systems. Mater Sci Forum 7:145–154
Dohmen R, Chakraborty S (2007) Fe–Mg diffusion in olivine II: point defect chemistry, change of diffusion
mechanisms and a model for calculation of diffusion coefficients in natural olivine. Phys Chem Mineral
34:409–430, doi:10.1007 / s00269-007-0158-6
Dohmen R, Becker H-W, Chakraborty S (2007) Fe–Mg diffusion in olivine I: experimental determination between
700 and 1,200°C as a function of composition, crystal orientation and oxygen fugacity. Phys Chem Mineral
34:389–407, doi:10.1007 / s00269-007-0157-7
Dohmen R, Kasemann SA, Coogan L, Chakraborty S (2010) Diffusion of Li in olivine. Part I: Experimental
observations and a multi species diffusion model. Geochim Cosmochim Acta 74:274–292, doi:10.1016 / j.
gca.2009.10.016
Dohmen R, Ter Heege JH, Becker H-W, Chakraborty S (2016) Fe–Mg interdiffusion in orthopyroxene. Am
Mineral 101:2210–2221, doi:10.2138 / am-2016-5815
Doremus R (1969) The diffusion of water in fused silica. In: Reactivity of Solids. Mitchell J, Devries R, Robers R,
Cannon P (eds) Wiley, New York, p 667–673
Ducea MN, Ganguly J, Rosenberg EJ, Patchett PJ, Cheng W, Isachsen C (2003) Sm–Nd dating of spatially
controlled domains of garnet single crystals: a new method of high-temperature thermochronology. Earth
England PC, Richardson SW (1977) The influence of erosion upon the mineral facies of rocks from different
metamorphic environments. J Geol Soc London 134:201–213
England PC, Thompson AB (1984) Pressure–temperature–time paths of regional metamorphism, Part I: Heat
transfer during the evolution of regions of thickened continental crust. J Petrol 25:894–928
Ewing TA, Hermann J, Rubatto D (2013) The robustness of the Zr-in-rutile and Ti-in-zircon thermometers during
high-temperature metamorphism (Ivrea-Verbano Zone, northern Italy). Contrib Mineral Petrol 165:757–779
Farver JR (2010) Oxygen and hydrogen diffusion in minerals. Rev Mineral Geochem 72:447–507,
doi:10.2138 / rmg.2010.72.10
Fei H, Wiedenbeck M, Yamazaki D, Katsura T (2013) Small effect of water on upper-mantle rheology based on
silicon self-diffusion coefficients. Nature 498:213–215, doi:10.1038 / nature12193
Ferry JM, Stubbs JE, Xu H, Guan Y, Eiler JM (2015) Ankerite grains with dolomite cores: A diffusion chronometer
for low- to medium-grade regionally metamorphosed clastic sediments. Am Mineral 100:2443–2457,
doi:10.2138 / am-2015-5209
Freer R, Edwards A (1999) An experimental study of Ca-(Fe,Mg) interdiffusion in silicate garnets. Contrib Mineral
Petrol 134:370–379
Gaetani GA, O'Leary JA, Shimizu N, Bucholz CE, Newville M (2012) Rapid reequilibration of H2O and oxygen
fugacity in olivine-hosted melt inclusions. Geology 40:915–918
Gallagher K (2012) Transdimensional inverse thermal history modeling for quantitative thermochronology. J
Geophys Res: Solid Earth 117:B02408, doi:10.1029 / 2011jb008825
Ganguly J (2002) Diffusion kinetics in minerals: principles and applications to tectono-metamorphic processes.
EMU Notes in Mineralogy 4:271–309
Geochem 72:559–601
Ganguly J, Tirone M (1999) Diffusion closure temperature and age of a mineral with arbitrary extent of diffusion;
theoretical formulation and applications. Earth Planet Sci Lett 170:131–140
Ganguly J, Cheng W, Chakraborty S (1998) Cation diffusion in aluminosilicate garnets; experimental determination
in pyrope–almandine diffusion couples. Contrib Mineral Petrol 131:171–180
Ganguly J, Dasgupta S, Cheng W, Neogi S (2000) Exhumation history of a section of the Sikkim Himalayas,
India: records in the metamorphic mineral equilibria and compositional zoning of garnet. Earth Planet Sci
Lett 183:471–486
Gardés E, Jaoul O, Montel JM, Seydoux-Guillaume A-M, Wirth R (2006) Pb diffusion in monazite: an experimental
study of Pb2+ + Th4+ ⇔ 2Nd3+ interdiffusion. Geochim Cosmochim Acta 70:2325–2336
Gardés E, Montel J-M, Seydoux-Guillaume A-M, Wirth R (2007) Pb diffusion in monazite: New constraints from
the experimental study of Pb2+ ⇔ Ca2+ interdiffusion. Geochim Cosmochim Acta 71:4036–4043
channel: A numerical simulation. Tectonics 21:doi:10.1029 / 2002TC001406
Gibbs JW (1874–1878) On the equilibrium of heterogeneous substances. Trans Connecticut Acad Arts Sci 3:108–
248, and 343–524
Giletti BJ (1986) Diffusion effects on oxygen isotope temperatures of slowly cooled igneous and metamorphic
rocks. Earth Planet Sci Lett 77:218–228
Glicksman ME (2000) Diffusion in Solids. Field Theory, Solid-State Principles, and Applications. John Wiley and
Sons, Inc., New York
Golden EM, Giles NC, Yang S, Halliburton LE (2015) Interstitial silicon ions in rutile TiO2 crystals. Phys Rev B
91:134110
Grove M, Harrison TM (1999) Monazite Th–Pb age depth profiling. Geology 27:487–490
Grove TL, Baker MB, Kinzler RJ (1984) Coupled CaAl–NaSi diffusion in plagioclase feldspar: experiments and
applications to cooling rate speedometry. Geochim Cosmochim Acta 48:2113–2121
Haifler J, Kotková J (2016) UHP–UHT peak conditions and near-adiabatic exhumation path of diamond-bearing
garnet–clinopyroxene rocks from the Eger Crystalline Complex, North Bohemian Massif. Lithos 248–
251:366–381, doi:10.1016 / j.lithos.2016.02.001
Harrison TM, Célérier J, Aikman AB, Hermann J, Heizler MT (2009) Diffusion of 40Ar in muscovite. Geochim
Cosmochim Acta 73:1039–1051
Harte B, Hudson NFC (1979) Pelite facies series and temperatures and pressures of Dalradian metamorphism in E
Scotland. Geol Soc Spec Publ 8:323–337
155:529–540
Heaman L, Parrish R (1991) U–Pb geochronology of accessory minerals. In: Applications of Radiogenic Isotope
Systems to Problems in Geology. Short Course Handbook. Vol 19. Heaman L, Ludden JN (eds) Mineral
Assoc Canada, Toronto, Canada, p 59–102
diamondiferous, deeply subducted continental crust (Kokchetav Massif, Kazakhstan). Contrib Mineral Petrol
141:66–82
Hier-Majumder S, Anderson IM, Kohlstedt DL (2005) Influence of protons on Fe–Mg interdiffusion in olivine. J
Geophys Res 110, doi:10.1029 / 2004jb003292
Humphreys MCS, Menand T, Blundy JD, Klimm K (2008) Magma ascent rates in explosive eruptions: Constraints
from H2O diffusion in melt inclusions. Earth Planet Sci Lett 270:25–40, doi:10.1016 / j.epsl.2008.02.041
Ireland RHP, Penniston-Dorland SC (2015) Chemical interactions between a sedimentary diapir and surrounding
magma: Evidence from the Phepane Dome and Bushveld Complex, South Africa. Am Mineral 100:1985–
2000, doi:10.2138 / am-2015-5196
Jackson K (2004) Kinetic Processes: Crystal Growth, Diffusion, and Phase Transitions in Materials. Wiley-VHC,
Weinheim
Jaoul O, Poumellec M, Froidevaux C, Havette A (1981) Silicon diffusion in forsterite: a new constraint for
understanding mantle deformation. In: Anelasticity in the Earth. Stacey FD, Paterson MS, Nicholas A, (eds).
Am Geophys Union, Washington, D.C., p 95–100
Jaoul O, Houlier B, Abel F (1983) Study of 18O diffusion in magnesium orthosilicate by nuclear microanalysis. J
Geophys Res 88:613–624
Jeffcoate AB, Elliott T, Kasemann SA, Ionov D, Cooper K, Brooker R (2007) Li isotope fractionation in peridotites
and mafic melts. Geochim Cosmochim Acta 71:202–218, doi:10.1016 / j.gca.2006.06.1611
Jenkin GRT (1997) Do cooling paths derived from mica Rb–Sr data reflect true cooling paths? Geology 25:907–910
John T, Gussone N, Podladchikov YY, Bebout GE, Dohmen R, Halama R, Klemd R, Magna T, Seitz H-M
(2012) Volcanic arcs fed by rapid pulsed fluid flow through subducting slabs. Nat Geosci 5:489–492,
doi:10.1038 / ngeo1482
Jollands MC, O’Neill HSC, Hermann J (2014) The importance of defining chemical potentials, substitution
mechanisms and solubility in trace element diffusion studies: the case of Zr and Hf in olivine. Contrib
Mineral Petrol 168, doi:10.1007 / s00410-014-1055-x
Katayama I, Maruyama S, Parkinson CD, Terada K, Sano Y (2001) Ion micro-probe U–Pb zircon geochronology
of peak and retrograde stages of ultrahigh-pressure metamorphic rocks from the Kokchetav Massif, northern
Kazakhstan. Earth Planet Sci Lett 188:185–198
Keppler H, Bolfan-Casanova N (2006) Thermodynamics of water solubility and partitioning. Rev Mineral
Geochem 62:193–230, doi:10.2138 / rmg.2006.62.9
Ketcham RA (2005) Forward and inverse modeling of low-temperature thermochronometry data. Rev Mineral
Geochem 58:275–314
Kohn MJ (1999) Why most “dry” rocks should cool “wet”. Am Mineral 84:570–580
Kohn MJ (2004) Oscillatory- and sector-zoned garnets record cyclic (?) rapid thrusting in central Nepal. Geochem
Geophys Geosyst 5:10.1029 / 2004gc000737, doi:Q12014 10.1029 / 2004gc000737
Kohn MJ (2008) P–T–t data from central Nepal support critical taper and repudiate large-scale channel flow of the
Greater Himalayan Sequence. Geol Soc Am Bull 120:259–273, doi:10.1130 / b26252.1
Kohn MJ (2009) Models of garnet differential geochronology. Geochim Cosmochim Acta 73:170–182
Kohn MJ (2013) Geochemical zoning in metamorphic minerals. In: Treatise on Geochemistry, volume 3, The
Crust. Rudnick R (ed) Elsevier, p 229–261
Mineral 101:25–42
Kohn MJ, Corrie SL (2011) Preserved Zr-temperatures and U–Pb ages in high-grade metamorphic titanite:
evidence for a static hot channel in the Himalayan orogen. Earth Planet Sci Lett 311:136–143
28:1127–1130, doi:10.1130 / 0091–7613(2000)028 < 1127:rntrif > 2.3.co;2
Kohn MJ, Valley JW (1998) Obtaining equilibrium oxygen isotope fractionations from rocks: theory and examples.
Himalaya of central Nepal. Earth Planet Sci Lett 228:299–310, doi:10.1016 / j.epsl.2004.10.007
Kohn MJ, Penniston-Dorland SC, Ferreira JSC (2016) Implications of near-rim compositional zoning in rutile
for geothermometry, geospeedometry, and trace element equilibration. Contrib Mineral Petrol 171:78 DOI
10.1007 / s00410-016-1285-1
Kooijman E, Smit MA, Mezger K, Berndt, J (2012) Trace element systematics in granulite facies rutile: implications
for Zr geothermometry and provenance studies. J Metamorph Geol 30:397–412
Kotková J, Whitehouse M, Schaltegger U, D’Abzac F-XD (2016) The fate of zircon during UHT–UHP
metamorphism: isotopic (U / Pb, d18O, Hf) and trace element constraints. J Metamorph Geol 34:719–739 doi:
10.1111 / jmg.12206
Kramer MJ, Seifert KE (1991) Strain enhanced diffusion in feldspars. In: Diffusion, Atomic Ordering, and Mass
Transport. Ganguly J, (ed) Springer-Verlag, New York, p 286–303
Krishna R (2015) Uphill diffusion in multicomponent mixtures. Chem Soc Rev 44:2812–2836,
doi:10.1039 / c4cs00440j
Krogh TE (1993) High precision U–Pb ages for granulite metamorphism and deformation in the Archean
Kapuskasing structural zone, Ontario: implications for structure and development of the lower crust. Earth
Planet Sci Lett 119:1–18, doi: 10.1016 / 0012-821X(93)90002-Q
Kuhn TS (1962) The structure of scientific revolutions. University of Chicago Press, Chicago
Kylander–Clark ARC (2017) Petrochronology by laser–ablation inductively coupled plasma mass spectrometry.
Rev Mineral Geochem 83:183–198
Geochem 83:55–102
Lapen TJ, Johnson CM, Baumgartner LP, Mahlen NJ, Beard BL, Amato JM (2003) Burial rates during prograde
metamorphism of an ultra-high-pressure terrane: an example from Lago di Cignana, western Alps, Italy.
Lasaga AC (1979) Multicomponent exchange and diffusion in silicates. Geochim Cosmochim Acta 43:455–469
Lasaga AC (1983) Geospeedometry: An extension of geothermometry. In: Kinetics and Equilibrium in Mineral
Reactions. Saxena SK (ed) Springer-Verlag, New York, p 81–114
LaTourrette T, Wasserburg GJ (1998) Mg diffusion in anorthite: implications for the formation of early solar system
planetesimals. Earth Planet Sci Lett 158:91–108
Lindström R, Viitanen M, Juhanoja J, Holtta P (1991) Geospeedometry of metamorphic rocks; examples in the
Rantasalmi-Sulkava and Kiuruvesi areas, eastern Finland; biotite–garnet diffusion couples. J Metamorph
Geol 9:181–190
Liou JG, Zhang R, Liu F, Zhang Z, Ernst WG (2012) Mineralogy, petrology, U–Pb geochronology, and geologic
evolution of the Dabie-Sulu classic ultrahigh-pressure metamorphic terrane, East-Central China. Am Mineral
97:1533–1543, doi:10.2138 / am.2012.4169
Liu Y, Anderson AT, Wilson CJN (2007) Melt pockets in phenocrysts and decompression rates of silicic magmas
before fragmentation. J Geophys Res 112, doi:10.1029 / 2006jb004500
Liu X-M, Rudnick RL, Hier-Majumder S, Sirbescu M-LC (2010) Processes controlling lithium isotopic
distribution in contact aureoles: A case study of the Florence County pegmatites, Wisconsin. Geochem
Geophys Geosystem 11:Q08014, doi:10.1029 / 2010gc003063
Lloyd AS, Ruprecht P, Hauri EH, Rose W, Gonnermann HM, Plank T (2014) NanoSIMS results from olivine-
hosted melt embayments: Magma ascent rate during explosive basaltic eruptions. J Volcanol Geothermal
Res 283:1–18
Lucassen F, Dulski P, Abart R, Franz G, Rhede D, Romer RL (2010) Redistribution of HFSE elements during rutile
replacement by titanite. Contrib Mineral Petrol 160:279–295, doi:10.1007 / s00410-009-0477-3
MacDonald JM, Wheeler J, Harley SL, Mariani E, Goodenough KM, Crowley Q, Tatham D (2013) Lattice
distortion in a zircon population and its effects on trace element mobility and U–Th–Pb isotope systematics:
examples from the Lewisian Gneiss Complex, northwest Scotland. Contrib Mineral Petrol 166:21–41,
doi:10.1007 / s00410-013-0863-8
Marks MAW, Rudnick RL, McCammon C, Vennemann T, Markl G (2007) Arrested kinetic Li isotope fractionation
at the margin of the Ilímaussaq complex, South Greenland: Evidence for open-system processes during final
cooling of peralkaline igneous rocks. Chem Geol 246:207–230, doi:10.1016 / j.chemgeo.2007.10.001
Mattinson JM (1978) Age, origin, and thermal histories of some plutonic rocks from the Salinian Block of
California. Contrib Mineral Petrol 67:233–245, doi:10.1007 / bf00381451
McDougall I, Harrison TM (1999) Geochronology and thermochronology by the 40Ar / 39Ar method. Oxford
University Press, Oxford
McFarlane C, Harrison TM (2006) Pb-diffusion in monazite: Constraints from a high-T contact aureole setting.
Earth Planet Sci Lett 250:376–384, doi:10.1016 / j.epsl.2006.06.050
McLelland JM, Selleck BW, Bickford ME (2010) Review of the Proterozoic evolution of the Grenville Province,
its Adirondack outlier, and the Mesoproterozoic inliers of the Appalachians. Geol Soc Am Mem 206:21–49
Mezger K, Rawnsley C, Bohlen S, Hanson G (1991) U–Pb garnet, sphene, monazite and rutile ages: Implications for
the duration of high grade metamorphism and cooling histories, Adirondack Mts., New York. J Geol 99:415–428
Müller T, Cherniak D, Bruce Watson E (2012) Interdiffusion of divalent cations in carbonates: Experimental
measurements and implications for timescales of equilibration and retention of compositional signatures.
Geochim Cosmochim Acta 84:90–103, doi:10.1016 / j.gca.2012.01.011
Müller T, Dohmen R, Becker HW, ter Heege JH, Chakraborty S (2013) Fe–Mg interdiffusion rates in clinopyroxene:
experimental data and implications for Fe–Mg exchange geothermometers. Contrib Mineral Petrol 166:1563–
1576, doi:10.1007 / s00410-013-0941-y
Nowotny MK, Sheppard LR, Bak T, Nowotny J (2008) Defect chemistry of titanium dioxide. Application of defect
engineering in processing TiO2-based photocatalysts. J Phys Chem C 112:5275–5300
Page FZ, Ushikubo T, Kita NT, Riciputi LR, Valley JW (2007) High-precision oxygen isotope analysis of picogram
samples reveals 2 µm gradients and slow diffusion in zircon. Am Mineral 92:1772–1775
Page FZ, Kita NT, Valley JW (2010) Ion microprobe analysis of oxygen isotopes in garnets of complex chemistry.
Chem Geol 270:9–19
Pape J, Mezger K, Robyr M (2016) A systematic evaluation of the Zr-in-rutile thermometer in ultra-high
temperature (UHT) rocks. Contrib Mineral Petrol 171, doi:10.1007 / s00410-016-1254-8
Parkinson I, Hammond S, James R, Rogers N (2007) High-temperature lithium isotope fractionation: Insights
from lithium isotope diffusion in magmatic systems. Earth Planet Sci Lett 257:609–621, doi:10.1016 / j.
epsl.2007.03.023
Pauly J, Marschall HR, Meyer H-P, Chatterjee N, Monteleone B (2016) Prolonged Ediacaran–Cambrian
metamorphic history and short-lived high-pressure granulite-facies metamorphism in the H.U. Sverdrupfjella,
Dronning Maud Land (East Antarctica): evidence for continental collision during Gondwana assembly. J
Petrol 57:185–228, doi:10.1093 / petrology / egw005
Peck WH, Valley JW, Graham CM (2003) Slow oxygen diffusion rates in igneous zircons from metamorphic rocks.
Am Mineral 88:1003–1014
Peck WH, Bickford ME, McLelland JM, Nagle AN, Swarr GJ (2010) Mechanism of metamorphic zircon growth
in a granulite-facies quartzite, Adirondack Highlands, Grenville Province, New York. Am Mineral 95:1796–
1806, doi:10.2138 / am.2010.3547
Penniston-Dorland SC, Sorensen SS, Ash RD, Khadke SV (2010) Lithium isotopes as a tracer of fluids in a
subduction zone melange: Franciscan Complex, CA. Earth Planet Sci Lett 292:181–190
Perchuk AL, Burchard M, Schertl HP, Maresch WV, Gerya TV, Bernhardt HJ, Vidal O (2009) Diffusion of divalent
cations in garnet: multi-couple experiments. Contrib Mineral Petrol 157:573–592, doi:10.1007 / s00410-
008-0353-6
Piazolo S,Austrheim H, Whitehouse M (2012) Brittle–ductile microfabrics in naturally deformed zircon: Deformation
mechanisms and consequences for U–Pb dating. Am Mineral 97:1544–1563, doi:10.2138 / am.2012.3966
Piazolo S, La Fontaine A, Trimby P, Harley S, Yang L, Armstrong R, Cairney JM (2016) Deformation-induced

trace element redistribution in zircon revealed using atom probe tomography. Nat Commun 7:10490,
doi:10.1038 / ncomms10490
Qian Q, O’Neill HSC, Hermann J (2010) Comparative diffusion coefficients of major and trace elements in olivine
at 950 °C from a xenocryst included in dioritic magma. Geology 38:331–334, doi:10.1130 / g30788.1
Ravna EJK, Roux MRM (2006) Metamorphic evolution of the Tonsvika eclogite, Tromso Nappe—Evidence for a
new UHPM Province in the Scandinavian Caledonides. Int Geol Rev 48:861–881
Reddy SM, Timms NE, Trimby P, Kinny PD, Buchan C, Blake K (2006) Crystal-plastic deformation of zircon: A
defect in the assumption of chemical robustness. Geology 34:257, doi:10.1130 / g22110.1
Reddy SM, van Riessen A, Saxey DW, Johnson TE, Rickard WD, Fougerouse D, Fischer S, Prosa TJ, Rice
KP, Reinhard DA, Chen Y (2016) Mechanisms of deformation-induced trace element migration in zircon
resolved by atom probe and correlative microscopy. Geochim Cosmochim Acta 195:158–170, doi:10.1016 / j.
gca.2016.09.019
Richter FM, Liang Y, Davis AM (1999) Isotope fractionation by diffusion in molten oxides. Geochim Cosmochim
Acta 63:2853–2861
Richter FM, Davis AM, DePaolo DJ, Watson EB (2003) Isotope fractionation by chemical diffusion between
molten basalt and rhyolite. Geochim Cosmochim Acta 67:3905–3923, doi:10.1016 / s0016-7037(03)00174–1
Richter FM, Mendybaev RA, Christensen JN, Hutcheon ID, Williams RW, Sturchio NC, Beloso AD (2006) Kinetic
isotopic fractionation during diffusion of ionic species in water. Geochim Cosmochim Acta 70:277–289,
doi:10.1016 / j.gca.2005.09.016
Richter F, Watson B, Chaussidon M, Mendybaev R, Ruscitto D (2014) Lithium isotope fractionation by diffusion in
minerals. Part 1: Pyroxenes. Geochim Cosmochim Acta 126:352–370, doi:10.1016 / j.gca.2013.11.008
Rudnick R, Ionov D (2007) Lithium elemental and isotopic disequilibrium in minerals from peridotite xenoliths
from far-east Russia: Product of recent melt / fluid–rock reaction. Earth Planet Sci Lett 256:278–293,
doi:10.1016 / j.epsl.2007.01.035
Sasaki J, Peterson NL, Hoshino K (1985) Tracer impurity diffusion in single-crystal rutile (TiO2−x). J Phys Chem
Solids 46:1267–1283
Schmitt AK, Vazquez JA (2017) Secondary ionization mass spectrometry analysis in petrochronology. Rev Mineral
Geochem 83:199–230
Schwandt CS, Cygan RT, Westrich HR (1996) Ca self-diffusion in grossular garnet. Am Mineral 81:448–451
Seitz S, Putlitz B, Baumgartner LP, Escrig S, Meibom A, Bouvier A-S (2016) Short magmatic residence times
of quartz phenocrysts in Patagonian rhyolites associated with Gondwana breakup. Geology 44:67–70,
doi:10.1130 / g37232.1
Shannon RD (1976) Revised effective ionic radii and systematic studies of interatomic distances in halides and
chalcogenides. Acta Crystallogr Sect A A32:751–767
Shea T, Costa F, Krimer D, Hammer JE (2015a) Accuracy of timescales retrieved from diffusion modeling in
olivine: A 3D perspective. Am Mineral 100:2026–2042, doi:10.2138 / am-2015-5163
Shea T, Lynn KJ, Garcia MO (2015b) Cracking the olivine zoning code: Distinguishing between crystal growth and
diffusion. Geology 43:935–938, doi:10.1130 / g37082.1
Shimojuku A, Kubo T, Ohtani E, Nakamura T, Okazaki R (2010) Effects of hydrogen and iron on the silicon
diffusivity of wadsleyite. Phys Earth Planet Int 183:175–182, doi:10.1016 / j.pepi.2010.09.011
Shore M, Fowler AD (1996) Oscillatory zoning in minerals: a common phenomenon. Can Mineral 34:1111–1126
Smith HA, Giletti BJ (1997) Lead diffusion in monazite. Geochim Cosmochim Acta 61:1047–1055
Smye AJ, Stockli DF (2014) Rutile U–Pb age depth profiling: a continuous record of lithospheric thermal evolution.
Sousa JL, Kohn MJ, Schmitz MD, Northrup CJ, Spear FS (2013) Strontium isotope zoning in garnets: Implications
for metamorphic matrix equilibration, geochronology, and phase equilibrium modeling. J Metamorph Geol
31:437–452
Spear FS (1991) On the interpretation of peak metamorphic temperatures in light of garnet diffusion during
cooling. J Metamorph Geol 9:379–388
Spear FS (2004) Fast cooling and exhumation of the Valhalla metamorphic core complex, southeastern British
Columbia. Int Geol Rev 46:193–209
Spear FS (2014) The duration of near-peak metamorphism from diffusion modelling of garnet zoning. J Metamorph
Geol 32:903–914, doi:10.1111 / jmg.12099
Spear FS, Florence FP (1992) Thermobarometry in granulites: Pitfalls and new approaches. J Precambrian Res
55:209–241
Spear FS, Parrish RR (1996) Petrology and cooling rates of the Valhalla complex, British Columbia, Canada. J
Petrol 37:733–765
Spear FS, Ashley KT, Webb LE, Thomas JB (2012) Ti diffusion in quartz inclusions: implications for metamorphic
time scales. Contrib Mineral Petrol, doi:10.1007 / s00410-012-0783-z
Stepanov AS, Rubatto D, Hermann J, Korsakov AV (2016) Contrasting P–T paths within the Barchi-Kol UHP
terrain (Kokchetav Complex): Implications for subduction and exhumation of continental crust. Am Mineral
101:788–807, doi:10.2138 / am-2016-5454
Storm LC, Spear FS (2005) Pressure, temperature and cooling rates of granulite facies migmatitic pelites from
the southern Adirondack Highlands, New York. J Metamorph Geol 23:107–130, doi:10.1111 / j.1525–
1314.2005.00565.x
Storm LC, Spear FS (2009) Application of the titanium-in-quartz thermometer to pelitic migmatites from the
Adirondack Highlands, New York. J Metamorph Geol 27:479–494, doi:10.1111 / j.1525–1314.2009.00829.x
Sutton AP, Balluffi RW (2003) Interfaces in Crystalline Materials. Clarendon Press
Taylor-Jones K, Powell R (2015) Interpreting zirconium-in-rutile thermometric results. J Metamorph Geol 33:115–
122, doi:10.1111 / jmg.12109
Teng F-Z, McDonough WF, Rudnick RL, Walker RJ (2006) Diffusion-driven extreme lithium isotopic fractionation
in country rocks of the Tin Mountain pegmatite. Earth Planet Sci Lett 243:701–710, doi:10.1016 / j.
epsl.2006.01.036
Till CB, Vazquez JA, Boyce JW (2015) Months between rejuvenation and volcanic eruption at Yellowstone caldera,
Wyoming. Geology 43:695–698, doi:10.1130 / g36862.1
Timms NE, Kinny PD, Reddy SM (2006) Enhanced diffusion of uranium and thorium linked to crystal plasticity
in zircon. Geochem Trans 7:10, doi:10.1186 / 1467-4866-7-10
Timms NE, Kinny PD, Reddy SM, Evans K, Clark C, Healy D (2011) Relationship among titanium, rare earth
elements, U–Pb ages and deformation microstructures in zircon: Implications for Ti-in-zircon thermometry.
Chem Geol 280:33–46, doi:10.1016 / j.chemgeo.2010.10.005
Tirone M, Ganguly J, Dohmen R, Langenhorst F, Hervig R, Becker H-W (2005) Rare earth diffusion kinetics in
garnet: Experimental studies and applications. Geochim Cosmochim Acta 69:2385–2398
Turner S, Evans P, Hawkesworth C (2001) Ultrafast source-to-surface movement of melt at island arcs from Ra-
226–Th-230 systematics. Science 292:1363–1366
Vielzeuf D, Saúl A (2011) Uphill diffusion, zero-flux planes and transient chemical solitary waves in garnet.
Contrib Mineral Petrol 161:683–702, doi:10.1007 / s00410-010-0557-4
Vielzeuf D, Veschambre M, Brunet F (2005) Oxygen isotope heterogeneities and diffusion profile in composite
metamorphic-magmatic garnets from the Pyrenees. Am Mineral 90:463–472
Vielzeuf D, Baronnet A, Perchuk AL (2007) Calcium diffusivity in alumino-silicate garnets: an experimental and
ATEM study. Contrib Mineral Petrol 154:153–170
Viete DR, Oliver GJH, Fraser GL, Forster MA, Lister GS (2013) Timing and heat sources for the Barrovian
metamorphism, Scotland. Lithos 177:148–163, doi:10.1016 / j.lithos.2013.06.009
Wang ZY, Hiraga T, Kohlstedt DL (2004) Effect of H + on Fe–Mg interdiffusion in olivine, (FeMg)2SiO4. Appl
Phys Lett 85:209–211
Warren CJ, Beaumont C, Jamieson RA (2008) Formation and exhumation of ultra-high-pressure rocks during
continental collision: Role of detachment in the subduction channel. Geochem Geophys Geosystem
doi:10.1029 / 2007GC001839
Watson EB (1982) Basalt contamination by continental crust: some experiments and models. Contrib Mineral
Petrol 80:73–87
Watson EB, Baxter EF (2007) Diffusion in solid-Earth systems. Earth Planet Sci Lett 253:307–327
Watson EB, Cherniak DJ (1997) Oxygen diffusion in zircon. Earth Planet Sci Lett 148:527–544
Petrol 151:413–433
Yund RA, Tullis J (1980) The effect of water, pressure, and strain on Al / Si order / disorder kinetics in feldspar.
Yund RA, Smith BM, Tullis J (1981) Dislocation-assisted diffusion of oxygen in albite. Phys Chem Miner 7:185–189
Yund RA, Quigley J, Tullis J (1989) The effect of dislocations on bulk diffusion in feldspars during metamorphism.
Zhang Y (2010) Diffusion in minerals and melts: theoretical background. Rev Mineral Geochem 72:5–59
Zhang Y, Cherniak DJ (2010) Diffusion in Minerals and Melts. Mineral Soc America, Washington, D. C.
Zhang Y, Ni H (2010) Diffusion of H, C, and O components in silicate melts. Rev Mineral Geochem 72:171–225
Zhang Y, Stolper EM, Wasserburg GJ (1991) Diffusion of a multi-species component and its role in oxygen and
water transport in silicates. Earth Planet Sci Lett 103:228–240
Zhang Y, Xu Z, Zhu M, Wang H (2007) Silicate melt properties and volcanic eruptions. Rev Geophys 45,
doi:10.1029 / 2006rg000216
Zhang Y, Ni H, Chen Y (2010) Diffusion data in silicate melts. Rev Mineral Geochem 72:311–408,
doi:10.2138 / rmg.2010.72.8
Vol. 83 pp. 153–182, 2017 5
Electron Microprobe Petrochronology

Michael L. Williams, Michael J. Jercinovic
University of Massachusetts
Amherst MA 01003
USA
mlw@geo.umass.edu
mjj@geo.umass.edu
Kevin H. Mahan
Department of Geological Sciences
University of Colorado
Boulder, CO 80302
USA
mahank@colorado.edu
Gregory Dumond
University of Arkansas
Fayetteville, AR 72701
USA
gdumond@uark.edu
INTRODUCTION
The term petrochronology has increasingly appeared in publications and presentations over
the past decade. The term has been defined in a somewhat narrow sense as “the interpretation of
isotopic dates in the light of complementary elemental or isotopic information from the same
mineral(s)” (Kylander-Clark et al. 2013). Although complementary isotopic and elementary
information are certainly a central and critical part of most, if not all, petrochronology
studies, the range of recent studies that might use the term covers a much broader scope.
The term “petrochronology” might alternatively be defined as the detailed incorporation of
chronometer phases into the petrologic (and tectonic) evolution of their host rocks, in order
to place direct age constraints on petrologic and structural processes. As noted by Kylander-
Clark et al. (2013), the linkage between geochronology and petrology can involve a variety
of data including mineral textures and fabrics, the distribution of mineral modes or volume
proportions, compositional zoning, mineral inclusion relationships, and certainly major
element, trace element, and isotopic composition of the chronometer and all other phases.
Electron probe micro-analysis (EPMA) has a central and critical role to play in establishing
the linkage between chronometer phases and their host assemblage. The basic instrument
is an electron microscope which can be used in either scanning or fixed beam modes, with
integrated wavelength dispersive spectrometers (WDS), energy dispersive spectrometers
(EDS), electron detectors (to image secondary and backscattered signals) a light optical system,
and optionally cathodoluminescence (CL) detection. The electron microprobe is used to
investigate the distribution, composition, and compositional zonation of all mineral phases, the
1943-2666/17/0083-0005$05.00 (online)
154 Williams, Jercinovic, Mahan, & Dumond
data that underpin thermobarometric analysis and modeling of P–T histories. The microprobe,
with mm-scale spatial resolution, can also characterize compositional zonation in very small
accessory phases including monazite, xenotime, zircon, allanite, titanite, apatite, and others.
This, as discussed below, can be a critical step in linking geochronology to petrology. Finally, in
many circumstances, the microprobe can be used to determine or constrain the age of domains
within certain chronometers, especially monazite and xenotime. Where the compositional
domains are small (< 5 µm), the microprobe may be the only feasible tool that can constrain the
age. Narrow rim domains are commonly key constraints on the petro-tectonic history.
This chapter is focused on the role of EPMA in petrochronological studies. The early parts
of the chapter highlight analytical considerations in using the electron microprobe: first, for
compositional characterization and for establishing the linkage between chronometer phases
and the petro-tectonic history, and second, for precise trace element analysis and dating.
Although the process is relatively straightforward, special analytical methods must be adopted
for high-precision trace element analysis by electron microprobe. The later parts of the chapter
provide examples and illustrations of the different roles that the electron microprobe can play
in petrochronological studies. Many of the examples, and much of the recent research, concern
monazite and xenotime in deformed and metamorphosed rocks. Although zircon has been
widely used to constrain age or provenance, and more rarely metamorphic history, monazite
and xenotime can record, with high fidelity, multiple stages in the igneous and metamorphic
history and can also provide some key constraints on the deformational history. Finally, it
should be noted that neither the techniques nor the electron microprobe instrument itself have
reached their ultimate potential for petrochronological analysis. We hope to shine some light
on future directions and challenges that suggest the ability to more efficiently extract, and to
more tightly constrain, the P–T–time–deformation history of rocks.
REACTION DATING
One of the major goals of igneous or metamorphic analysis is determining the sequence of
minerals or mineral assemblages that were present during the evolution of the rocks of interest,
and ultimately interpreting the sequence of chemical reactions that relate the minerals or
assemblages. The characterization of mineral assemblages and reactions, based on petrographic,
petrologic, and microstructural analysis allows the construction of P–T–t (±D = deformation)
histories that are central to most petrologic investigations. Although in some previous studies,
workers have used resetting of geochronometers during metamorphism as a means to constrain
the “age of metamorphism”, the very sluggish rate of diffusion in high-temperature chronometers
such as zircon, monazite, or xenotime (Cherniak et al. 2004; Cherniak and Pyle 2008; Cherniak
2010), make resetting generally unlikely. Instead, new mineral grains (neocrystals), or new
domains within chronometer minerals, probably grow during the reaction history. This is
particularly true of monazite, which can serve as a source or sink of a wide variety of elements,
including Rare Earths, actinides, and others, that are minor and trace elements in most silicates.
As such, the essence of petrochronology involves integrating trace elements and trace-element-
bearing chronometer phases into the silicate reaction history. Trace element partitioning may
place additional constraints on the metamorphic equilibria (Hickmott and Spear 1992; Bea et al.
1997; Pyle and Spear 1999; Yang et al. 1999; Spear and Pyle 2002), and in the present context,
dating specific generations of chronometer phases that have been associated with a particular
reaction allows a date to be directly associated with a reaction. This process could be described
as “reaction dating” and we note that it is increasingly becoming the focus of many researchers
around the world (for example Larson et al. 2011; Dumond et al. 2015; Regis et al. 2016;
and many others). Focusing on the term, “reaction dating” deemphasizes the goal of “dating
metamorphism” and instead, emphasizes the goal of dating or constraining as many prograde
and retrograde reactions as possible, in order to characterize the timing and hopefully duration
of metamorphic or igneous events.
Electron Microprobe Petrochronology 155
Most of the examples and illustrations in this chapter focus on petrochronology involving
monazite and to a lesser extent xenotime and zircon. For many reasons, monazite may be the
ideal chronometer for EPMA petrochronology. First, monazite has a wide stability field from
diagenesis to high-grade metamorphism, and it occurs in a range of metamorphic and igneous
rock compositions (Overstreet 1967; Williams et al. 2007; Catlos 2013). Further, its broad
compositional range implies that monazite can, to some degree, participate in a variety of silicate
reactions, serving as a source or sink of minor components that are also present in the silicate
minerals. There are many examples of monazite or xenotime reacting to (or from) other phosphate
or REE-bearing minerals (Finger et al. 1998; Janots et al. 2008; Budzyn et al. 2011), but as
discussed below, monazite compositions can also be modified during many silicate reactions
because of minor components either liberated or consumed during the reaction(s). Finally,
monazite is particularly amenable to electron microprobe analysis. In addition to U, Th, and Pb,
many components, including light REEs and actinides, are abundant enough to be analyzed by
EPMA. The current analytical protocol at the University of Massachusetts includes 25 elements
in a standard monazite analysis. Many other minerals can provide key petrochronological data
including apatite, allanite, titanite, thorite, uraninite, and others (see also Vance et al. 2003). The
methods described below are applicable to varying degrees to all of these phases. Examples of
some representative studies are highlighted, but many other examples could certainly be included.
Monazite has been an important geologic chronometer since the classic study of Parrish
(1990). However, starting in the late 1990s, a large number of studies documented multiple
generations of metamorphic monazite (and xenotime) and to varying degrees attempted to
integrate monazite or xenotime generations with silicate reactions. They illustrate the steadily
evolving logic used to correlate monazite generations and silicate reactions and assemblages.
Early studies tended to focus on Y-concentration because of the inverse relationship with garnet
abundance (Pyle and Spear 1999, 2003; Wing et al. 2003; Foster et al. 2004; Gibson et al. 2004;
Berger et al. 2005; Kohn et al. 2005; McFarlane et al. 2005; Yang and Pattison 2006). Along with
heavy Rare Earth Elements, Y is strongly partitioned into garnet, and as such, Y in monazite
(and the overall abundance of xenotime) typically decreases during garnet growth and increases
on garnet break-down. This Y–garnet–monazite–xenotime connection is still one of the most
powerful and widely used petrochronological tools, especially for microprobe-based studies.
More recently, workers have considered other trace components and proposed increasingly
more specific reactions and mechanisms linking accessory phases with silicate assemblages
(i.e., Rubatto et al. 2006; Buick et al. 2010; Dumond et al. 2015; Regis et al. 2016; Rocha
et al. in press). In addition, new experimental data and studies of well-constrained natural
assemblages have provided constraints on partitioning between accessory phases and silicate
assemblages and melts (Hermann and Rubatto 2003; Krenn and Finger 2004; Rubatto and
Hermann 2007; Stepanov et al. 2012), updating and expanding the classic work of Bea and
coworkers (Bea et al. 1994; Bea and Montero 1999). One ultimate goal is to better constrain
the thermodynamic properties of accessory phases such that the accessory chronometers can
be incorporated into phase diagrams, particularly isochemical phase diagram models (see
below). Several important steps have been made in modeling monazite, xenotime, and zircon
abundance and composition in metamorphic rocks and in melt–silicate systems (Kelsey et al.
2008; Spear 2010; Spear and Pyle 2010; Kelsey and Powell 2011).
COMPOSITIONAL MAPPING, TRACE ELEMENT ANALYSIS, AND DATING

BY ELECTRON MICROPROBE
Compositional mapping for petrochronology
Compositional maps have been used for many years to explore zoning in single minerals
(Tracy 1982; c.f. Kohn 2013). Early maps made by contouring point analyses were time
consuming and generally of low resolution. Modern maps made by rastering the electron
microprobe (beam or stage) or by mosaicking a grid of smaller SEM images are simple and
efficient to collect. Resolution can be as high as 1 mm per pixel (or better), although time
constraints may limit the resolution for larger maps.
Full-thin-section or large-area compositional maps are much less common in the literature,
but can be invaluable for petrochronological analysis (Fig. 1). Maps of major element
abundances, commonly Mg, Ca, K etc. can show the distribution of major phases and also
significant compositional zoning within the larger minerals. Maps of selected elements (Ce for
monazite, Y for xenotime, Zr for zircon) can be used to locate accessory chronometer phases
(see also Williams et al. 2006; Larson et al. 2011). Depending on the number of elements
required and number of available spectrometers or compositional channels, multiple mapping
acquisitions of the same area may be necessary in order to identify and characterize the major
and accessory phases of interest. Background maps (maps collected in a background position
for the desired element) or calculated backgrounds based on mean atomic number (Donovan
et al. 2016) can be subtracted from peak-position compositional maps in order to perform a
background correction. Also, peak pixel values can be calibrated in order to constrain phase
compositions (see also Kohn and Spear 2000; Clarke et al. 2001; De Andrade et al. 2006),
and image math can be used to calculate hybrid maps such as age maps (Goncalves et al.
2005). Commonly, a single set of maps can be used to successfully identify the major phases
and to locate and evaluate the chronometer phases. Other instruments (and methods) besides
electron microprobe WDS mapping are available for large-area compositional mapping,
including scanning electron microscopy (SEM) with EDS (sometimes also WDS) and
electron backscatter diffraction (EDS and EBSD) capabilities, and the related electron beam
techniques offered by quantitative evaluation of minerals by scanning electron microscopy
(QEMSCAN), and mineral liberation analysis (MLA). QEMSCAN in particular is capable of
efficiently capturing compositional information from minerals in large-area maps, although
these instruments, to date, have mainly been used in economic geology applications.
Plg + Qz
Bt
Grt
Crd
Bt
Grt
Figure 1. Full-thin-section Mg Kα compositional map of sample S32D. Lighter grey tone corresponds to
great Mg content. The large central subhedral garnet has been partly replaced by biotite (outside) and cor-
dierite on fractures within the crystal. Abbreviations from Whitney and Evans (2010). White stars show the
location of Monazite grains; larger stars are larger grains. The stars represent (i.e., are placed on) high-Ce
pixels (or clusters of pixels) on the full-section map. The map was run at 300 nA, 25 ms/pixel with a 35 mm
step size and 25–30 mm beam. Modified from Mahan et al. (2006).
Large-area maps showing the location of chronometer phases are, in themselves, useful
tools for illuminating the petrogenesis of the major and the accessory phases. For example
xenotime grains specifically situated around the rim of garnet crystals commonly represent
late-stage xenotime growth during garnet resorption (Fig. 2). Zircon in leucosome domains
can represent zircon formed during melt crystallization (Flowers et al. 2006a). These maps
are particularly useful when they are generated early in the petrographic/petrologic analysis
cycle rather than later after the petrographic analysis and interpretation has been completed.
High-resolution (small-area) maps of chronometer phases, including WDS,
cathodoluminescence, backscattered electron, etc., have been used extensively to characterize
zoning in chronometer phases in order to plan a strategy for dating, and to aid in interpreting
geochronologic results (Williams et al. 2006, 2007; Larson et al. 2011; Peterman et al. 2016;
and many others). When placed into the context of full-section compositional maps, the high-
resolution maps can be an even more powerful tool for petrochronology and reaction dating.
One method for integrating the maps involves placing the high-resolution maps around or on
the full-section image with links to the actual grain locations (Fig. 3a). This allows the zonation
within a high-resolution map to be interpreted in the context of its setting within the thin
section. Recent work has shown that even the most subtle compositional zoning in monazite
can reflect the local setting in the thin section (i.e., porphyroblast inclusion relations, nearby
phases, local structures, fractures). It is particularly important to process the high-resolution
compositional maps with the same look-up table such that similar compositions have similar
intensities on the images (Williams et al. 2006). One simple technique using Adobe software is
summarized in Appendix-1 (deposited with MSA and available from the authors).
The combination of high-resolution and low-resolution (large area) compositional
maps allows chronometer phases to be integrated into the petrographic analysis process.
In the case of monazite, zircon, or xenotime, it is commonly possible to identify several
key generations of the chronometer phase(s) and to relate the generations, at least in a
qualitative way, to the silicate phases/assemblages and to fabric generations. This also
changes the nature of the geochronologic analysis strategy. Rather than dating a number
of grains and interpreting ages, the analytical strategy involves first, verifying that
chronometer populations identified on compositional maps actually represent specific
a b
Grt
Grt
Figure 2. (a) Ca Kα compositional map of garnet from Vermont showing xenotime crystals concentrated
on the margin of resorbed garnet. Modified from Gatewood et al. (2015). (b) Mg Ka map showing resorbed
garnet from the Park Range, Colorado. Xenotime is concentrated in sinistral strain shadows of the garnet.
The xenotime is interpreted to have been stabilized during garnet breakdown and release of Y.
Figure-2
(a)
(b) (c)
Figure 3. (a) Full-thin-section Mg Kα compositional map of sample S32D (Fig. 1) with superimposed
high-resolution Y Lα monazite grain maps. (b) Close-up of central garnet. Note inclusions in inner core
contain monazite generations Mz1 and Mz2. Outer inclusions are dominated by generation Mz3. (c) High-
resolution Y Lα map of Mz14 showing all five generations of monazite. See text for discussion. Modified
from Mahan et al. (2006) and Villa and Williams (2012).
Figure 3
compositional and geochronological populations and second, dating (“sampling”) each
population in order to constrain the age of the particular assemblages or fabrics.
Trace element maps of silicate phases can add additional insights. Trace-element maps
can be acquired by EPMA, and are particularly straightforward in minerals such as garnet and
olivine that are stable under high sample current (Spear and Kohn 1996; Pyle and Spear 1999;
Goodrich et al. 2013; Kohn 2013). Trace element mapping has also recently been done by LA-
ICP-MS (Kylander-Clark 2017; Lanari and Engi 2017). With several notable exceptions, this
is a largely untapped opportunity for relating chronometer phases to silicate assemblages. Pyle
and Spear (1999) used high-current mapping of Y in garnet and documented the close coupling
between Y in garnet and xenotime in metamorphic rocks over a range of metamorphic grades
from garnet nucleation to partial melting. The xenotime–garnet relationships provide insights
into parts of the P–T history that are not well constrained by the major elements and silicate
phases alone; several thermometers have been calibrated for Y partitioning between garnet and
xenotime, and the xenotime itself can be dated, commonly by electron probe, in order to place
timing constraints on parts of the prograde and retrograde P–T path.
It is worth noting that many geochronological studies involve dating a large number
of chronometer grains/domains, plotting dates on an “age histogram” and then interpreting
populations. This “top-down” approach is useful for detrital mineral analysis, but is less
appropriate for analysis of metamorphic or igneous rocks. The “bottom-up” approach (Williams et
al. 2006) involves: (1) establishing populations, based on composition and texture; (2) developing
hypotheses about chronometer-forming reactions and relative timing; and (3) constraining the
age of the populations using the most appropriate analytical technique. For chronometers such as
monazite and xenotime, narrow rim domains can be very important for constraining the petrologic
history. For these domains, electron microprobe total Pb dating may be the only analytical option.
Trace-element analysis by electron microprobe—analytical considerations
The electron microprobe was initially developed for, and most applications involved, rapid
non-destructive major and minor element analysis. However, even during the early development,
some workers (e.g., Goldstein and Wood 1966; Goldstein 1967) recognized that the instrumentation
had great utility in the trace element realm (below 1 wt.%). In the modern age, where LA-ICP-MS
and Ion Microprobe instruments can analyze a broad suite of trace elements with low detection
limits (e.g., Kylander-Clark 2017; Schmitt and Vazquez 2017), EPMA still has numerous
applications in the measurement of minor and trace components, especially when non-destructive
analysis and high spatial resolution are important. Further, during the past several decades, there
have been a number of significant improvements in hardware, software, and analytical procedures
that have increased EPMA precision and accuracy for trace element analysis.
U–Pb dating (i.e., total-Pb dating) by EPMA is a relatively new application (Parslow
et al. 1985; Bowles 1990; Suzuki and Adachi 1991; Asami et al. 1996; Montel et al. 1996;
Cocherie et al. 1998; Williams et al. 1999). It is based on the assumption that common Pb
is insignificant compared to radiogenic Pb in U- and Th-rich (and Ca-poor) phases like
monazite, xenotime, zircon, uraninite, or thorite. As such, accurate measurement of the total
amount of Pb, U, and Th can be used to calculate a date from very small domains in the
chronometer phases. As noted above, many compositional domains, especially rim domains
and core domains, can be extremely small or narrow, but are critical for constraining P–T
histories. Although Th can be present in monazite at the major or minor element level, U and
Pb are almost always trace element measurements whose emission lines lie in a complicated
part of the X-ray spectrum, particularly when REEs are present.
One lesson learned over the past decade is that it is not adequate to apply the major element
analytical protocol to trace elements and simply increase the current and count time. Every
aspect of the analytical routine, from sample (and standard) preparation to data processing,
must be specifically designed for the trace elements involved and for the particular application
(e.g., Scherrer et al. 2000; Jercinovic et al. 2008). Many aspects of the analytical problems
and potential solutions have been described in other publications (Pyle et al. 2002; Williams
et al. 2006; Jercinovic and Williams 2005; Jercinovic et al. 2008, 2012; Spear et al. 2009). The
following paragraphs summarize some of the most recent improvements and some of the major
Table 1. Analytical Steps in EPMA Geochronology.

Procedure Instrument Setup
1 Find the accessory minerals in thin section (monazite, xenotime, 15 kV, 300 nA, 25 ms/pixel, 35 mm
zircon, thorite, uraninite, etc.) by full thin section mapping. Ce, beam diameter. 35 mm step, stage
Y, Zr, Mg, Ca. raster.
2 Process full section maps to overlay indicator elements on base 15 kV, 200 nA, 80 ms/pixel.
map (reveals accessory mineral grain locations in relation to
microstructure).
3 Micromap the accessory minerals to define compositional focused beam. 0.5 mm step size.
domains (e.g., Monazite: Y, U, Th, Ca, N).
4 Process maps: simultaneous and individual; superimpose maps
onto full-section map for analysis.
5 Select grains and domains for analysis based on compositional
and microstructural significance.
6 Recoat for quantitative analysis. Al coat 25 nm, followed by C-coat
8 nm.
7 Quantitative analysis for full chemistry and age. Trace element 15 kV, 200 nA. Focused beam.
methodology for Pb, U, and Th, as well as low concentration
elements relevant to key reactions (see Table 3).
lessons learned especially with respect to trace element analysis for petrochronology. Key steps
in the overall preparation and analytical procedure are summarized in Table 1.
Analytical strategy
An obvious goal in designing hardware, software, and analytical procedures for trace
element analysis is to obtain the smallest possible detection limits within a practical setup/
analytical timeframe. As formalized by Ziebold (1967), the propagated precision on the k-ratio
is given by:
 
N + N ( B) N s + N s ( B)
=σ2k k 2  +  (1)
 2
( ) ( )
2
 n N − N ( B) n′ N s − N s ( B) 

and similarly, the variance of the concentration is given by:
 
 × 1 − ( a − 1) C 
2
N + N ( B) N s + N s ( B)
=σC2 C 2  +   (2)
 2
( ) ( )
2
n N − N ( B) n′ N s − N s ( B)   a 
 
where σk = standard deviation, N = sample peak counts, N(B) = sample bkg counts, Ns = std.
peak counts, Ns(B) are the std. bkg counts, k = k-ratio = (N – N(B))/(Ns – Ns(B)) for a
pure element standard, n and n’ = number of measured points on sample and standard,
σc = concentration standard deviation, C = concentration, and a = correction factor that
relates the k-ratio to concentration.
The relationships in Equations (1) and (2) describe the precision of the acquisition
based on the Poisson statistics of X-ray emission. The equations also suggest where minor
inaccuracies can strongly influence the result, particularly in the case of low-net-intensity
unknowns such as trace elements. One clear implication is that the standard concentration for
trace element analysis should be as high as possible to diminish the second term in equation 1.
Then, for cases of diminishing peak/background, the precision of the result depends critically
on: (1) maximizing peak counts (by increasing current, voltage, counting time, spectrometer
collection efficiency, increasing the number of analysis points in an acquisition), and (2) the
precision of the background, which ultimately determines the signal/noise level to be overcome
in order for an element to be considered detectable. Each of these will be briefly discussed
below. See also Ziebold (1967), Goldstein (1967), Ancey et al. (1978), Merlet and Bodinier
(1990), and Lifshin et al. (1999) for formalisms of detection limit calculations.
The precision of an analysis in EPMA is clearly dependent on optimizing the total
counts collected at the characteristic wavelength (or energy) for the element of interest.
The simplest approach in the acquisition of more counts is to increase beam current and/or
counting time, but as with many analytical variables, there are important trade-offs. High beam
current with a small beam size results in high beam power density (Jercinovic et al. 2012)
and subsequent beam damage and contamination. Long count time will also exacerbate time-
dependent manifestations of beam damage and contamination, and also increase the potential
for instrumental drift or internal charge effects. Another option is to acquire a population
of peak acquisitions within a homogeneous compositional domain. However, the potential
gain diminishes as the number of points increases above approximately 6. Jercinovic et al.
(2012) showed that increasing the number of analysis points from 7 to 10 for a trace element
analysis of Pb resulted in an increase the sensitivity of only 2 ppm. Other approaches are also
very useful, including the use of multiple spectrometer simultaneous counting of the same
line, with subsequent integration of counts. In this way, quite high sensitivity is possible with
reasonable counting times, for example, a detection limit of 2–3 ppm is possible for Ti in
quartz by integrating 5 spectrometers in 960 s acquisitions (Donovan et al. 2011). Additionally,
the types of spectrometers employed can make a substantial difference, including the use of
large and very-large monochromators (and carefully adjusted counter parameters, specific to
the element being analyzed), and/or smaller radius focusing geometries (sacrificing spectral
resolution). Higher voltage will also increase the ionization efficiency, but with the obvious
trade-off of decreased spatial resolution and the possible introduction of increased spectral
complexity resulting from higher energy ionizations (see Jercinovic et al. 2008).
One realization has been evident for decades—trace element sensitivity (lowering
detection limit) results from the ability to characterize background at high precision as
much as it does from maximizing peak counts (Reed 1993). Simple off-peak, two-point
interpolation of background will not adequately account for curvature, particularly if
positions are limited in complex matrices (Jercinovic et al. 2008, 2012). Background shape,
a convolution of Bremsstrahlung emission vs. wavelength and spectrometer efficiency, must
be directly assessed, and must also include the recognition of even the slightest component
of interference in the background measurement regions of the spectrum. The combination of
background curvature and interferences can produce disastrously inaccurate trace element
results (Fig. 4) if not properly accounted for (Merlet and Bodinier 1990; Jercinovic et al.
2008). Further, because of the complexity of the spectrum in minerals such as monazite,
xenotime, or zircon, is it extremely difficult to recognize small background interferences
even on relatively high-resolution wavelength scans (see Williams et al. 2006; Jercinovic
et al. 2008). Detailed WDS scanning is essential, but not completely satisfactory unless
the precision of the WDS acquisition matches the precision of the quantitative analysis.
As discussed below, errors due to background interference are systematic and reproducible
for the domain being analyzed. The error could be severe even if secondary standards give
acceptable results, and particularly if only a single secondary standard has been evaluated.
80
70 VLPET With Ce+Th

interference on Bkg
60 position
VLPET Curvature only

% error of net intensity
50
40
500 ppm 500 ppm
30 PETH VLPET
20 100 ppm
PETH
10
100 ppm
0 VLPET
0.00 0.05 0.10 0.15 0.20
net intensity (cps/nA)
Figure 4. Growth of error (in %) on the net intensity (= Peak cps/nA − background cps/nA) as a function of
net intensity for Pb Mα in monazite if a two-point linear interpolation is used. One bkg point was placed
between Pb Mα and Pb Mβ, and the other at a suitable wavelength above the Pb Mα analytical line (above
the interferences from first order Th Mζ2, second order La Lα, and first order S Kα lines). There are two
components to the error, one arises from the curvature of the background itself (dashed curves show this
component only), and the other from the subtle interference that exists between the Pb Mα and Pb Mβ
wavelengths. Two different monochromators are used for this estimation: PETH is the JEOL high intensity
spectrometer, and VLPET is the Cameca very large PET monochromator + extended width detector. Note
that at about 1500 ppm Pb, the error is a few percent, but below 500 ppm, the error on the estimate of the
net intensity grows rapidly, approaching 100% at the 10 ppm level. Modified from Jercinovic et al. (2008)
[Used by permission of Elsevier from Jercinovic et al. (2008) Chemical Geology, v, 254, Fig. 18, p. 213].
One approach to background characterization involves collecting a high-resolution

wavelength scan of a broad region around the peak of interest and then selecting regions that
are approximately background and regressing the background value at the peak position (see
Williams et al. 2006). It is important to collect a very high-resolution scan and to avoid any
fast-scan mode so that per-step count times are compatible with peak counts. This method
has been successful when used for all reference materials and unknowns. However, there is a
certain degree of subjectivity in choosing regions to be included in the regression analysis. An
alternative method, called “multipoint background acquisition” (Allaz et al. 2011), involves
acquisition of background intensities at multiple wavelength positions above and below the
peak position. These points are exponentially regressed and evaluated for goodness of fit. When
the regression meets certain criteria, the background value at the peak position is calculated.
This method has the advantage of being objective and reproducible, and the regression statistics
and background intensity can be evaluated and modified retroactively if necessary.
Analytical protocol
The protocol for quantitative trace element analysis is dynamic, depending on the questions
being asked (see Pyle et al. 2005; Williams et al. 2006). Count times, background acquisition
strategy, coating materials, etc. must be varied as the focus on key elements shifts to address
particular reactions. Beam damage at high beam power density is problematic (Jercinovic et
al. 2012), therefore the tradeoffs of beam diameter (plus scattering dimensions), beam current,
and coating thickness/material all must be weighed. Table 2 lists the steps involved in the full
quantitative EPMA characterization of phosphates such as monazite or xenotime. Additional
comments and suggestions are included in Appendix 2 (deposited with MSA).
Table 2. Analytical Protocol for EPMA Trace Element Analysis.
Procedure Setup
1 Coating. Samples and standards should be coated Plasma clean, then apply 20 nm Al,
simultaneously for the best possible accuracy. followed by 8 nm C.
2 Calibration. U, Th, Pb and, other major, minor, and interference 15 kV, 80 nA, 5 mm beam diameter.
calibrations as needed. After initial setup, routine recalibration
is less routinely necessary as long a high quality secondary
standards are available.
3 Define the setup, including the use of multipoint background
modeling, interference corrections, time-dependent intensity
corrections, multiple spectrometer integration, and blank
corrections as necessary. All elements should be acquired at
each point, with count times adjusted to maximize precision
for key elements.
4 Analysis of secondary standards. This may include blanks to 15 kV, 200 nA, focused beam.
the protocol.
5 Analyze unknown domain. A single homogeneous domain is 15 kV, 200 nA, focused beam. Single
analyzed at a time. Measure background (multipoint or WDS background acquisition per domain.
scanning). Accumulate peak measurements (typically 5–6
points) around the background location. Calculate weighted
mean age, uncertainty, and MSWD.
6 Reanalyze consistency standard during and after session.
The analytical philosophy for high-precision EPMA trace element analysis involves first,
using compositional mapping to define compositionally homogeneous domains and second,
independently constraining peak and background counts (count rates) at the highest possible
precision. Because compositionally homogeneous domains are determined in advance by
mapping key elements, it is possible to decouple peak and background measurement. For
example, it has been noted that Th/LREE tends to control the relative background intensity
in monazite domains, so a Th map is crucial in defining domains in monazite for background
acquisition (Williams et al. 2007). Peak and background measurement strategy and protocol
are developed independently for maximum precision considering sample stability. Optimally,
background is measured using a multipoint regression scheme during a similar time window
as peak measurements. Multiple peak measurements are made adjacent to the background
position until uncertainty on the age calculations stabilizes at a minimum value (typically
5–7 peak measurements). The single background measurement is used with all peak
measurements in order to calculate a single “date” for the domain. The MSWD (Mean Squared
Weighted Deviation, Wendt and Carl 1991) value for the weighted mean of the multiple peak
measurements provides a test of the compositional homogeneity; values significantly greater
than unity generally indicate compositional heterogeneity within the analytical domain.
Error assessment
Propagated counting statistics in EPMA account for only a portion of the total uncertainty
on a trace element analysis, or calculated physical parameter based on trace concentrations (i.e.,
temperature, pressure, or age, etc.). Systematic errors, arising particularly from instrumental
factors or nonrandom analytical factors are particularly serious as they may not be detected
by comparisons with secondary standards. Pyle et al. (2005) highlighted many of the factors
contributing to the overall error, and in particular, suggested ways to minimize systematic errors.
Williams et al. (2006) suggested that one might distinguish three components of error: (1) short-
term random error—primarily counting statistics; (2) short-term systematic error—primarily from
background methodology (regression models, etc.), but also coating variation and conductivity
issues; (3) long-term systematic error—quality of standards and calibrations, interference
correction algorithms, matrix corrections, current measurement, dead-time corrections, etc.
The first component includes primarily the propagated count statistical error. This error is
readily calculated and is typically apparent from scatter in results. The third component produces
errors that are fully reproducible when comparing results from the same chronometer measured
on the same instrument. However, these errors may be reflected in systematic differences between
the results of different dating techniques or instrument types. The second error component,
as noted above, is much more problematical. It includes errors resulting from compositional
effects such as peak or background interferences. They may seriously compromise results from
one compositional domain but not from another, and unknowns may give erroneous results
when secondary standards are accurately dated. As shown by Jercinovic et al. (2008), errors of
this type could be on the order of tens of millions of years or several percent of the calculated
age. Currently, relatively small, but significant, background interferences cannot be predicted
from knowledge of the electromagnetic spectrum or sample composition. Thus, background
estimation based on wavelength scanning or multipoint regression of all unknowns is essential
for recognizing and accommodating this type of error. Careful background measurement by
regression is commonly also necessary in measurements for overlap corrections on standards.
Results of EPMA geochronology have been published or presented in many venues. As
with all types of geochronology, estimates of the precision and accuracy of the dates must be
evaluated in light of the analytical protocol and specific components incorporated in the error
estimation. In particular, systematic errors associated with background estimation (type-2
above) can be very large and are generally not included in published results. Background
estimations based on any two-point interpolation have the potential for large systematic
interference-related error that must be incorporated into estimates of accuracy or even
comparison of age determinations. In addition, small uncertainties calculated simply by taking
the mean of large numbers of measurements must be viewed with caution as this tends to yield
a misleading average with inappropriate uncertainty, especially when the total range of dates
is much larger than twice the calculated error estimate for a homogeneous dataset.
Because of the strong dependency on phase composition and analytical methodology, it is
difficult to provide generalizations about the precision of EPMA geochronology. Uncertainties
increase dramatically in U- or Th-poor chronometers and also increase in young samples
because of low Pb abundance. Because trace element analysis by EPMA is essentially a
measurement of minor signals above background, as peak heights diminish, minor inaccuracies
in background can produce large errors in net intensity and calculated concentrations and ages.
Spear et al. (2009) suggested that 2% uncertainties might be a best estimate for EPMA ages
using standard hardware (and this estimate does not include background-related systematic
error). Using optimized hardware, such as the SX-Ultrachron microprobe, and incorporating
background regression procedures (i.e., Allaz et al. 2011), errors approaching 1% (and for
older samples errors less than 1%) are possible and have been demonstrated using multiple
standards analyzed by multiple techniques in multiple facilities.
The examples below show the preferred manner of presenting EPMA monazite data (see
Fig. 7). Each probability density function (PDF) represents results from one compositional
domain, that is, one background analysis and 5–7 peak analyses. The width of the PDF plot
(the calculated 2-sigma error on the population of measurements) reflects the propagated error
from peak and background analysis. The magnitude of uncertainty reflects the abundance of
U, Th and Pb in the monazite domain, but large deviations from expected PDF uncertainty
commonly indicate compositional heterogeneity within the domain. Weighted means can be
calculated from individual dates, aggregating results from a particular generation of monazite.
APPLICATIONS AND EXAMPLES OF EPMA PETROCHRONOLOGY

Textural and compositional correlation between accessory and major phases
Compositional mapping by electron microprobe, at multiple scales, is an essential part of
petrochronologic analysis regardless of the instrument or technique ultimately used for dating.
As noted above, the mapping can provide the critical linkage between silicate assemblages and
chronometers. For example, the linkage between Y in monazite and Y in garnet is exceedingly
powerful and has been extensively used in petrochronology, especially EPMA-based studies
(Pyle and Spear 2003; Foster et al. 2004; Mahan et al. 2006; Williams et al. 2007). This
is partly because Y is abundant enough in monazite and garnet to be readily measured by
electron probe. Also, in medium and high-grade metamorphic rocks (those without xenotime),
monazite and garnet are the major hosts for Y, and thus, changes in the Y content of monazite
commonly reflect modal changes in garnet, and because garnet is one of the key petrologic
index minerals, constraining the timing of garnet growth and break-down events commonly
provides some of the most critical timing constraints on the tectonic history.
Monazite–garnet–yttrium connection—Example 1: Legs Lake shear zone, Saskatchewan
One example of integrated compositional mapping and petrologic analysis, and the power
of the Y–monazite–garnet connection, comes from Mahan et al. (2006) but has been expanded
for use here. Mahan et al. (2006) investigated the timing and significance of the Legs lake shear
zone, Athabasca Granulite terrane, Saskatchewan. The shear zone separates granulite facies
(~1.1 GPa) rocks from amphibolite facies (~0.5 GPa) rocks. Kinematic analysis indicates
oblique thrust-sense shearing of the high-grade rocks over lower-grade rocks. The granulite
facies rocks contain evidence for two high-P–T tectonic events, one at ca. 2.6–2.55 Ga and
one at ca 1.9 Ga. Thrusting/shearing occurred at ca. 1.85 Ga, and is interpreted to have been
associated with exhumation of the granulite facies rocks. Mahan et al. (2006) calculated
isochemical phase diagrams for rocks outside of and within the shear zone, and interpreted
the P–T history and reaction history of the rocks during their evolution from dry deep-crustal
conditions to hydrous mid-crustal conditions (Fig. 5).
(a) (b)
1.1 A
0.9
0.7
P(GPa)
0.5 D
0.3
0.1
600 700 800 900 T ( C) 0 0.2 0.4 0.6 0.8 µΗ20
Figure 5. a) P–T isochemical phase diagram (pseudosection) for sample S32D including 1% H2O. Black
arrow shows inferred path during Legs Lake shear zone thrusting and exhumation based on peak and
retrograde assemblages and reaction relationships (Mahan et al. 2006). b) P–MH2O pseudosection sample
S32D. Black arrow shows path of decompression and hydration during Legs Lake shear zone thrusting.
Thick black line marks H2O saturation. Monazite population-5 is interpreted to reflect reaction B→C (Grt
+ Kfs + Pl1 + H2O = Bt + Sil + Pl2 + Mz5 + Ap) [Used by permission of John Wiley and Sons from Mahan et
al. (2006), Journal of Metamorphic Geology, v24, Figs. 6, 7, p.203,205].
Sample S32D was collected from within the Legs Lake shear zone. It preserves a
particularly complete record of the metamorphic history including hydration associated
with retrograde metamorphism. Figure 1 is a full-thin-section Mg Ka compositional map of
sample S32D. The map shows several garnet porphyroblasts that have been fractured and
partially replaced by cordierite (inside) and by biotite (outside). Stars mark the location of
monazite grains, identified by full-thin section compositional mapping. Figure 3a shows the
full-section map with high-resolution Y La maps of monazite grains superimposed. The maps
were processed simultaneously so that intensities are comparable from grain to grain. Five
monazite populations can be distinguished based on monazite composition (Fig. 6). Note that
populations are here termed Mz1 through Mz5; Mahan et al. 2006 used the abbreviation pop-1
through pop-5. Monazite inclusions in garnet have been described by Mahan et al. (2006) and
by Williams and Jercinovic (2012). Inclusions within innermost garnet cores are dominated
by Mz1 and Mz2. Mz1 has high Y (up to several weight percent) and high Th content. Mz2
is characterized by distinctly lower Y content, reduced by at least 70%. Monazite inclusions
in the outermost portions of garnet porphyroblasts are dominated by low-Y, low-Th Mz3 (Y
contents of several hundred PPM or less). The very high Y content of Mz1 (comparable to
monazite present in garnet absent rocks) suggests that this population probably grew before
significant garnet growth. Because of the strong partitioning of Y into garnet, growth of even
small amounts of garnet are associated with significant decreases of Y in monazite. The
stepwise drop in Y from Mz1 to Mz2 to Mz3 (Fig. 6) is interpreted to represent two period(s)
of garnet growth in the rock. This and the spatial separation of Mz2 and Mz3 within garnet
suggests that these two populations may represent two distinct monazite producing reactions
associated with two different garnet growth reactions. Mz3 contains the lowest Y content seen
in the rock and is interpreted to represent the time of maximum garnet mode (i.e., volume
percent). Th has been seen to increase from Mz2 to Mz3 in some monazite grains and to
decrease in others. This may indicate the presence of limited amounts of partial melt where
local access to melt yielded higher Th and isolation from melt yielded lower Th.
Most matrix monazite grains are relatively unzoned. Although rare cores of earlier
monazite generations (Mz1 or Mz2) are locally present, most matrix grains are dominated by
low-Y Mz4. Locally, monazite grains display high-Y rims (Mz5). These rims are generally
restricted to monazite grains that are within several hundred mm of a garnet porphyroblast,
106
Mean concentration and range (ppm)
S32D Monazite
1 2 3 4 5
105
+ Melt
Th
- Melt
104
predated U
Grt growth
Major
stepwise Grt
103 Grt growth breakdown
Y
102 Max Grt mode
Monazite Generation
Figure 6. Plot showing variation in Y, Th and U among five monazite populations (Mz1–Mz5) in retrograde
felsic granulite S32D from the Legs Lake shear zone, Athabasca area, Canada. [Used by permission of John
Wiley and Sons from Mahan et al. (2006) Journal of Metamorphic Geology, v. 24, Fig. 8, p 208].
probably attesting to the limited mobility of Y in the matrix. In addition, monazite grains
that are included in dynamically recrystallized plagioclase do not have the high-Y rims. This
suggests that dynamic recrystallization of matrix plagioclase occurred after growth of Mz4 but
before significant garnet break-down. That is, Mz4 monazite was trapped within recrystallizing
plagioclase, and was shielded from later Mz5 overgrowths by the enclosing plagioclase.
Two monazite grains, of the 50 grains that were mapped, have all five populations (m14
and m48—Fig. 3). No other grain has more than three of the populations. We refer to these
grains as “Rosetta-stone” grains (Dahl et al. 2005) because they are particularly important
for petrochronology, i.e., linking chronology to tectonic events. Interestingly, the two grains
are not among the largest monazite grains in the section, nor do they occupy the same
microstructural setting. Monazite 14 occurs within the garnet rim and monazite 48 is in the
matrix. Because of their small size, neither grain would probably have been recovered in a
traditional mineral separation. Based on experience mapping a large number of samples from
many localities, it is common to have a monazite population dominated by relatively simple
one- or two-generation grains and then to have a small number of Rosetta-grains that capture
important details of the history. This underscores the need to map a relatively large number
of chronometer grains in a structurally and petrologically important sample.
Table 3 summarizes a set of observations that can be made from monazite compositional
relationships in sample S32D. Some interpretations are rather speculative but they establish a set
of hypotheses that can be tested and refined by quantitative monazite analysis and by analysis
of other rocks. The next step is to document the composition and compositional variation within
each monazite population and in the process, confirm that the map-defined populations are indeed
homogeneous compositional populations. Then, the final step is to develop an analytical strategy
to determine the age or age range for each population. This can be thought of as sampling each
population, using the most appropriate geochronologic technique, in order to constrain the age.
Mahan et al. (2006) presented dates for monazite from the five populations using earlier
EPMA techniques, and some new dates have been determined using techniques summarized
above. Mz1, Mz2, and Mz3 are Archean (2.6–2.55 Ga) and confirm the Archean age of
most garnet in the sample. Mz4 is Proterozoic (ca. 1.9 Ga), and corresponds to the second
metamorphic event in the area. The presence of Mz4 inclusions in some garnet rims suggests
that there may have been some Proterozoic garnet growth. Further, the very low Y content
of both Mz3 and Mz4 suggests that there may not have been significant decompression (i.e.,
exhumation) between Mz3 and Mz4, i.e., between the Archean and the Proterozoic events.
Garnet consumption associated with decompression is typically associated with growth of
new Y-richer monazite. This is one piece of evidence supporting prolonged residence of the
terrane in the deep crust, possibly from ca. 2.55 Ga to 1.9 Ga. Finally, high-Y Mz5 monazite
constrains thrust-sense shearing and exhumation associated with the Legs Lake shear zone to
have occurred between 1.9 and 1.85 Ga. The significantly increased Y in Mz5 is interpreted to
reflect garnet breakdown during exhumation and decompression.
Sample S32D is an example of a “Rosetta-Stone” sample that preserves a particularly
long and complete record of the tectonic history. Typical of many studies, once such a
sample is discovered and characterized, other samples can fill in missing details of the
P–T–t–D history. For example, the Legs Lake shear zone is locally cut by one other shear
zone, the Grease River shear zone. Within this zone, the Legs Lake assemblages once
again have been reequilibrated. Biotite plus cordierite pseudomorphs after garnet such as
those in sample S32D (Fig. 1) have been flattened and deformed into the new shear-related
foliation, and replacement of cordierite by chlorite and epidote documents greenschist facies
conditions. One new population of monazite has developed in the late-stage structures,
postdating Mz5 and constraining the age of this latest shearing event, and of greenschist
facies metamorphism, to ca. 1.80 Ga (Mahan et al. 2006; Dumond et al. 2013).
Table 3. Interpretive Petrochronologic History of Sample S32D.

1. Garnet growth occurred in at least two phases, probably from two garnet-producing reactions.
Monazite Mz1 predates significant garnet growth. Mz2 postdates the first garnet growth
phase and Mz3 postdates the second growth phase.
2. Mz1 and Mz2 are abundant and closely spaced in all garnet cores, but few Mz1 or Mz2
remnants are present outside of the inner cores of garnet. Thus, early monazite was consumed
or recrystallized before growth of Mz3. This may represent a stage of melting as suggested
by the synchronous growth of new garnet and high-Th nature of monazite Mz3.
3. The efficient removal of early monazite may also indicate a deformation phase between Mz2
and Mz3.
4. Little new garnet growth occurred after Mz3. Mz3 records the lowest Y (and Gd) contents of
all monazite populations.
5. The similar Y-content of Mz3 and Mz4 suggests that little garnet consumption occurred
between these two monazite populations.
6. Matrix monazite is dominated by Mz4, with few cores or remnants of earlier populations. This
suggests a second period of efficient monazite consumption after Mz3 and after garnet growth.
7. Dynamically recrystallized plagioclase wraps around garnet and contains only Mz4 monazite.
Deformation and recrystallization occurred during or after Mz4 and before Mz5. Plagioclase
is unzoned and thus the composition equilibrated with Mz4–Mz5 phases.
8. Mz5 was associated with garnet consumption as indicated by the major increase in Y. Mz5
rims are not aligned with dominant fabric; many occur on perpendicular sectors. This
the dominant matrix deformation had terminated before Mz5. This is consistent with the
persistence of abundant early monazite.
9. Cordierite in garnet fractures contains inclusions of xenotime. Thus, at least the later stages
of cordierite replacement of garnet occurred below the monazite ⇒ xenotime reaction.
Monazite–garnet–Y connection. Example-2: dating deformation

Figure 7 is one example of EMPA reaction dating following the protocol described above
and modified from Williams and Jercinovic (2012). The basic question concerned the timing
of folding and of garnet growth in a metamorphosed sediment from the Northwest Territories,
Canada. Inclusion fabrics suggest that garnet grew during folding of an earlier fabric (S1)
that was sub-parallel to compositional layering. Based on Y and Th content, three generations
of monazite were distinguished, one pre-garnet, one syn-garnet, and one post garnet. Once
the timing hypothesis was formulated, each generation was analyzed several times and dates
calculated. Figure 7b show results for pre and post-garnet domains and Figure 7c shows syn-
garnet results superimposed. Taken together the results show that garnet growth and folding
occurred at ca. 1900 Ma. and retrogression began soon afterward. As noted above, Figure 7
also shows the preferred manner of presenting EPMA monazite data.
Other compositional/textural linkages with silicate assemblages
Recently, a broader suite of trace elements and element ratios have been used to link
monazite with silicate assemblages in EPMA-based and also in LA-ICP-MS-based studies.
Th content in monazite may be particularly useful as a monitor of melting reactions. Th is
partitioned into the melt phase relative to the residual phases especially for haplogranite melts
(Keppler and Wyllie 1990; Stepanov et al. 2012), but Th is partitioned into monazite relative
to haplogranite melt, i.e., the monazite/melt partitioning coefficient is greater than unity (Rapp
et al. 1987; Xing et al. 2013). In addition, the solubility of monazite in granitic melt has
been interpreted to decrease with increasing pressure (Dumond et al. 2015). This suggests
(a)
07W-032B
m11 m12
m10
m13
m9
m6
m20 m5
m7
m14
m19 m4
m8
m18
m1
m2 m3
m17
m16
m15
Ce
(b) 1800 (c) 1800

07W-032-All 07W-032-All
Monazite Age (m.y.)
1900 1900
2000 2000
Figure 7. Monazite geochronology from west of Snowbird Lake, Northwest Territories, Canada. (a) Ca
Kα WDS full-thin-section compositional map showing folded early compositional layering and axial plane
S2 cleavage. High-resolution Y Lα compositional maps superimposed. Note, some grains are parallel to
S1 ⁄ S0 (also defined by inclusions in garnet), and some grains are parallel to S2 (upper left to lower right).
Zoning in monazite is much more pronounced in Grt-rich layers. (b) Microprobe monazite geochronology.
Each probability distribution plot represents one monazite date, including one background and 5–8 peak
analyses (see text for discussion). Green curves are grains that are early with respect to S2 cleavage and
garnet growth. Red curve is late with respect to S2 cleavage. (c) Same plot as (b) with addition of dates
for syn-S2 and syn-garnet monazite. S2 cleavage and peak metamorphism (garnet growth) are Fig. 5
constrained
to be ca. 1900±8 Ma. Early stage of exhumation and garnet consumption are constrained to 1893±12 Ma.
Modified from Williams and Jercinovic (2012). [Used by permission of John Wiley and Sons, Williams and
Jercinovic (2012) Journal of Metamorphic Geology, v30, Fig 5, p. 749].
that at higher pressures, monazite may provide key constraints on the timing and character
of melt reactions. For example, Dumond et al. (2015) used phase equilibria to characterize
melting reactions, adjust bulk compositions for melt loss, and interpret P–T–t–D histories for
high-T granulite facies migmatites from the Athabasca Granulite Terrane (Fig. 8). They used
multi-scale compositional mapping to integrate monazite into the reaction and melting history.
The rocks contain multiple generations of monazite. The earliest generation is interpreted to
represent detrital/inherited monazite. High-Th, low-Y monazite was interpreted to have been
produced during biotite dehydration melting. Sharp drops in Th in later monazite generations
were interpreted to reflect melt loss from the system, i.e., significant drops in bulk-rock Th
content (Fig. 8). Thus, Th in monazite may not only serve as a monitor of melting but also of
melt-extraction events in high-grade metamorphic rocks (Dumond et al. 2015).
More speculatively, Dumond et al. (2010, 2015) suggested that elevated Eu (and positive
Eu anomalies) in monazite may be related to plagioclase break-down and to some degree, to
sequestering of heavy REEs into garnet. In addition, light REE (especially Ce and La) enrichment
has been suggested to be related to feldspar loss or recrystallization (Dumond et al. 2015; also
see Bea 1996; Villaseca et al. 2003). These observations suggest a linkage between monazite
and feldspar that may further allow monazite generations to be integrated into silicate reactions.
Because of the limited mobility of Rare Earth elements, especially in relatively dry rocks, it
may be particularly useful to distinguish and compare different compositional layers or domains
in rocks. For example, in the eastern Adirondack Mountains, monazite in garnet-rich gneiss
has high-Y rims while in strongly recrystallized and lineated K-feldspar-rich layers monazite
rims are subtly enriched in U. The increased U is interpreted to have been hosted by feldspar
before recrystallization. The garnet rich and K-feldspar rich layers share a common fabric,
(a) (b)
Deposition of “White Gneiss” protolith (c. 2.61 Ga)
Temperature (°C)
1 Early high-Y detrital? monazite 0 200 400 600 800 1000
0
deposited after c. 2.64-2.61 Ga
Approximate Depth (km)

Grt + Opx + Kf
domain 1 0.2 And
Mnz -10
Bt + Pl + Q
0.4
Sil
10 µm
s+L
-20
tz
2 Dissolution of high-Y monazite

0.6
Grt1 cores
Y during partial melting synchronous 0.8
Ky
2.61-2.55 Ga -30
with growth of pertitectic garnet
1.0
Re-heating and Grt2 -40
Bt + Pl + Qtz + Mnz 1 Grt + Kfs + Opx + Melt 1.2 1.92-1.90 Ga
or -50
Bt + Pl + Sil + Qtz + Mnz 1 Grt + Kfs + Melt 10 µm 1.4 Grt1 annuli
domain 2
Pressure (GPa)
Mnz geotherm age (Ga) Q* (mW m-2) D (km) 2.59-2.52 Ga

1.6 2.61-2.55 60 30 -60
3 2.59-2.52 20 60
Growth of Y-depleted, Th-rich pre 1.92 20 40
Th Y monazite with melt + garnet 1.8
1.92-1.90 30 40 -70
2.0
c. 2.61-2.55 Ga prograde melting
Bt + Pl + Qtz Grt + Kfs + Opx + Melt + Mnz 2

or
domain 3
Bt + Pl + Sil + Qtz Grt + Kfs + Melt + Mnz 2 50 µm Mnz
4 Episodic melt loss and growth of
Th, Ca Eu (Th+Ca)-depleted, high Eu-anomaly monazite
with high-grossular garnet annuli
c. 2.55-2.54 Ga culmination of high-Grs

UHT-HP metamorphism Grt1
5 Melt-absent growth of Th-depleted,

20 µm
[La+Ce]-enriched monazite with
La, Ce second generation of garnet during
feldspar recrystallization low-Grs
Grt2
c. 1.92-1.90 Ga HP-metamorphism
and recrystallization during domain 4
Mnz
dextral transpressive strain
Grt2 + Mnz intergrowth
Figure 8. (a) Interpreted monazite reaction history and geochemical linkages with garnet generations and
melting in high-T granulite facies migmatites from the Athabasca Granulite Terrane, Saskatchewan. (b)
Interpreted P–T history based on petrologic modeling and monazite geochronology. See text for discus-
sion. Modified from Dumond et al. (2015). [Used by permission of John Wiley and Sons from Dumond et
al. (2015) Figures 15, 16, v.33, p 755–756].
mineral lineation, and kinematic shear sense, and both types of monazite rims have similar ages
(Wong et al. 2012). Taken together, the two layer types suggest that garnet consumption (i.e.,
decompression), K-feldspar recrystallization, and extensional shearing were all synchronous.
This allows the earliest phases of extensional shearing to be constrained and tied to early garnet
resorption and exhumation in the eastern Adirondack Mountains (Wong et al. 2012).
In some samples, the limited mobility of trace elements is particularly apparent. An Al-
rich gneissic sample from the Adirondack Mountains is distinctly layered with garnet-rich
and garnet-poor layers. Monazite composition and dates are dramatically different from
layer to layer. At least three distinct layer types can be recognized (Fig. 9). Layer-1 preserves
only relatively young monazite that largely reflects the early stages of exhumation. This
layer may have experienced more intense, late-stage strain, removing older monazite and
promoting growth or recrystallization of new monazite. Layer-2 preserves older monazite
(1180–1150 Ma). These are present as inclusions in garnet and also as matrix grains. It may
be that these garnet-rich layers were stronger and thus less deformed during the late-stage
deformation events. Layer-3 preserves a heterogeneous assemblage of monazite with dates
that are more difficult to interpret. Some of these grains bear the unmistakable signature of
dissolution-reprecipitation (see below) suggesting that this layer was affected by fluids of
the right composition to interact with monazite (Harlov and Hetherington 2010; Williams
et al. 2011). Taken together, monazite in this sample provides a very complete record of the
petro-tectonic history from early high-grade metamorphism to late extensional collapse,
but without careful multiscale mapping and in-situ analysis, accurate analyses would be
impossible and the results would be very difficult to interpret.
Y-Maps
Mg-Full-section
m18
m4 m13
m6
Layer-3
m20 m22
m1
m21 m14
m23
m15
Layer-2 m2
m9 m8
m16
m7
Layer-1
m5
m11
m12 m3 m19
m17
m10
Figure 9. Full-section Mg Ka compositional map of Sample 85-1 (Adirondack Mountains) with Y La
maps of monazite grains. Three different layer types have distinct monazite compositions and ages.
Layer-1 grains contain exclusively 1050 Ma cores and 1030 Ma rims. Layer-2 grains have 1180 Ma core
and 1150 Ma rims. Layer-3 grains have 1030 Ma domains and 980 Ma domains and complex textures char-
acteristic of fluid-related recrystallization.
EPMA PETROCHRONOLOGY COMBINED WITH ISOTOPIC ANALYSIS

The electron microprobe also plays a major role in petrochronologic analysis in studies
where the electron probe cannot be used for dating. Monazite and xenotime in relatively
young metamorphic or igneous rocks (younger than approximately 100 m.y.) is generally
not amenable to EPMA dating because of low Pb abundance. However, electron microprobe
mapping and analysis of the chronometer phases and evaluation of the local reaction context
are critical to drawing conclusions about the P–T–t–D history. Many studies in relatively young
rocks have involved electron microprobe mapping and analysis followed by isotopic analysis
by ion microprobe (i.e., see references in Vance et al. 2003; Kohn et al. 2005) or LA-ICP-MS
(i.e., Larson et al. 2011; Regis et al. 2016). One of the disadvantages of this type of analysis is
that many early core domains and late rim domains in chronometer phases are very small or
narrow even for a focused laser or ion microprobe. Compositional mapping of a larger number
of chronometer phases and identification of generations allow relatively larger domains to be
dated and then correlated with smaller domains in distinctive structural or petrological setting.
Titanite is known to contain significant amounts of common Pb and thus cannot be readily
dated using a total-Pb method such as electron microprobe dating. However, titanite compositions
can vary widely and can be integrated into silicate reactions. Titanite is certainly amenable to the
type of multi-scale compositional mapping and textural analysis as presented above for monazite.
For example, Wintsch et al. (2005) used compositional mapping to identify several generations
of titanite in granodioritic orthogneiss in Connecticut. The authors interpreted a reaction
involving replacement of metastable magmatic K-feldspar, hastingsite, magnetite, and titanite
by new metamorphic biotite, epidote, quartz and metamorphic titanite. Specifically, titanite was
interpreted to have been produced by a reaction such as clinozoisite = plagioclase + Al-dominant
titanite (Wintsch et al. 2005), and comparisons of REE composition of titanite and epidote were
used to distinguish titanite generations. EPMA and SEM images were used to select Al-richer
and Al-poorer titanite dating targets. Then, SHRIMP U–Pb dating was used to constrain ages.
Although this is an example of petrochronology involving the electron microprobe, it is suspected
that even more insight into the nature and relative timing of titanite producing reactions would
result from integrated multiscale image analysis such as that described above.
One additional petrochronology application that involves an important component of
EPMA analysis and also has great future potential is 40Ar–39Ar dating of mica, hornblende,
and other K-bearing phases. The electron microprobe has been used to identify and
distinguish compositional populations of micas or amphiboles and to generate phase
diagrams incorporating the stability of different generations of these phases. Then, ages have
been obtained by one of several possible methods: (1) in-situ laser techniques, (2) carefully
micro-sampling distinct structural/petrologic domains, or (3) by interpreting distinct ages
from step-heating results (e.g., Villa et al. 2000; Müller et al. 2002; White and Hodges 2003;
Condon et al. 2006; Flowers et al. 2006b; Wells et al. 2008; Growdon et al. 2013; Schneider
et al. 2013; Lanari et al. 2014; Chafe et al. 2014; Villa et al. 2014). One important strength
of this approach is the possibility to exploit the 37Ar–38Ar–39Ar correlation diagrams to
fingerprint phases, and so to integrate deformation stages, as defined by mica fabrics and
heterochemical generations, with P–T histories (Villa and Williams 2012).
Because K-feldspar, biotite, and especially white mica and amphibole can preserve
microstructural relicts and the associated isotope inheritance even under medium and high-T
conditions, all while new heterochemical mineral generations are growing, K–Ar chronometers
are not substantially different from U–Pb chronometers such as monazite (Villa and Williams
2012), as both are “Class II” mineral chronometers (Villa 2016). 40Ar–39Ar chronology, in
combination with detailed petrological characterization and dating of other geochronometer
systems, has been particularly useful for constraining the retrogression history all the way from
upper amphibolite facies to sub-greenschist facies. As new micas and feldspars tend to grow
during retrograde events that take place relatively late, they have been used to constrain the
exhumation stages of tectonic histories (i.e., White and Hodges 2003; Flowers et al. 2006b).
LOW-GRADE METAMORPHISM AND FLUID–ROCK INTERACTION

Many of the examples discussed above involve medium and high-grade metamorphism
where reactions involving silicates and chronometer phases have been interpreted. Another
rich avenue for petrochronologic study, and EPMA petrochronology, is in low-grade
metamorphic rocks and hydrothermal processes. A number of reactions have been interpreted
during diagenesis or low-grade metamorphism including: alteration of detrital monazite and
growth of allanite, alteration of detrital monazite and growth of new monazite, growth of
xenotime on detrital zircon, and growth of new generations of monazite, xenotime, or zircon
during fluid infiltration events (Rasmussen 1996; Vallini et al. 2002; Rasmussen and Muhling
2007, 2009; Janots et al. 2007; Rasmussen et al. 2007; Allaz et al. 2013). In each case, detailed
textural and compositional analysis of the chronometer phases have shed light on the nature
and timing of digenetic and low-grade metamorphic reactions. Commonly, the authigenetic,
digenetic, or low-grade metamorphic chronometer phases are very fine grained (less than 10
mm). The electron microprobe or SEM can be essential for simply finding and identifying the
minerals and for quantifying compositions and ages. One of the most common applications of
petrochronology in low-grade rocks is in dating and characterizing economic mineral and ore
deposits (i.e., Rasmussen et al. 2006; Muhling et al. 2012; Zi et al. 2015).
Diffusional resetting of monazite, zircon, or xenotime ages is considered unlikely under
most geological conditions because of extremely slow diffusion (Cherniak et al. 2004; Cherniak
and Pyle 2008; Cherniak 2010). However, experimental work and empirical observations suggest
that it is possible to reset zircon, monazite, and xenotime through dissolution-precipitation
mechanisms (Seydoux-Guillaume et al. 2012; Villa and Williams 2012; Didier et al. 2013; Ruiz-
Agudo et al. 2014), and that resetting may occur over a range of metamorphic grades (Harlov
and Forster 2002; Harlov et al. 2007, 2011; Hetherington et al. 2008; Harlov and Hetherington
2010; Williams et al. 2011; Kelley et al. 2012). These observations have important implications
for petrochronology because reset (recrystallized) chronometers may not be in equilibrium with
the associated silicate assemblage and thus, may lead to incorrect timing constraints on the
petrologic history (see Villa and Williams 2012). However, they also suggest a new avenue in
petrochronology, that is, the accessory phases can place constraints on composition, character,
and especially the timing of fluid–rock interactions (Grand’Homme et al. 2016).
The electron microprobe also has a role to play in identifying and characterizing the
products of these dissolution-precipitation reactions. First, it is critical to recognize domains
in zircon or monazite that have undergone dissolution-precipitation. The regions can have
remarkably straight and crystallographically controlled boundaries and can be easily mistaken
for overgrowth domains (see Harlov et al. 2011; Williams et al. 2011; Villa and Williams
2012). However, high-resolution compositional mapping and backscattered electron imaging
can show narrow reset domains that follow inclusion trails, or cracks, and locally form delicate
fingers intruding the core domains. Further, reset domains commonly have a distinctive
composition, typically characterized by a more nearly end-member monazite, xenotime, or
zircon composition than the host crystal (see discussion in Williams et al. 2011; also Didier et
al. 2013). The electron microprobe can be used to investigate the composition of very small
altered domains and for older chronometers to constrain the age of the alteration.
FUTURE TRENDS IN EPMA PETROCHRONOLOGY

Electron microprobe instrumental aspects
Electron Probe Micro-Analysis (EPMA) is a time-honored electron beam technique for
the non-destructive quantitative analysis of micro-volumes in-situ. Since its inception in 1950,
hardware, software, and physical theory have all advanced considerably, and continue to
evolve to this day. Many advances are the result of technological improvements that have been
inspired by the needs of the scientific community, and have primarily been aimed at improving
sensitivity, accuracy, spatial resolution, and efficiency. For high spatial resolution trace
element analysis and geochronology, a number of developments (some already in progress)
can increase the sensitivity, spatial resolution, and especially, the efficiency of compositional
mapping and quantitative analysis that are important for petrochronologic studies.
One recent emphasis centers around the improvement and refinement of low energy
X-ray detection and quantification in order to exploit the low scattering volumes attainable at
low accelerating voltage, resulting in higher analytical spatial resolution (e.g., McSwiggen et
al. 2011; Armstrong et al. 2013; Hombourger and Outrequin 2013; Gopon et al. 2013; Susan
et al. 2015). This development has been inspired in part by improvements in application
of higher brightness sources (rare earth hexaboride and Schottky field emitters). However,
the challenges are formidable particularly in trace element analysis, in which case high
beam current and/or long counting time can compromise accuracy at high current density
(Jercinovic et al. 2012), and more generally in evaluating soft X-ray spectra resulting from
valence band transitions (Burgess et al. 2014; MacRae et al. 2016a,b). We expect significant
progress over the next decade in the quantification of these difficult spectral emission
regions, and regardless of difficulties in quantification, high spatial resolution mapping is
extremely valuable. Quantification of WDS/EDS map data is also developing rapidly, even
becoming useful in trace element applications (e.g., Donovan et al. 2016). The approach to
petrochronologic studies with EPMA is mapping centric, and improvements in mapping
efficiency, performance, signal integration (e.g., MacRae et al. 2016a) or processing will be
advantageous, including rapid simultaneous processing (see Williams et al. 2006) of large
sets of maps from multiple samples in characterizing geologic terrains.
The spectrometers currently deployed on WDS instruments cover just about any need,
but there is room for improvement in terms of collection efficiency, specifically optimized
monochromators, or detector gases and pressures for certain elements or element ranges and/
or suppression of high order diffraction effects. These issues involve both high energy and
low energy detection, for example, there are reasons to explore more efficient detection of
the uranium L series using LIF [220], topaz [303], or α-quartz [2023], particularly in low Th
monazite, xenotime, or zircon analysis. As discussed above, there is also renewed emphasis on
low energy X-ray detection, including parallel beam optics (e.g., LEXS [EDAX, Inc.]), grating
optics with CCD photon counters (Terauchi et al. 2012; MacRae et al. 2016b; Robertson and
McSwiggen 2016), adaptation of silicon drift detector (SDD) technology to WDS (Moran
and Wuhrer 2016), which may help both in improving collection efficiency as well as spectral
resolution (e.g., Hombourger and Outrequin 2013; Robertson and McSwiggen, 2016).
Obviously continued improvement in the energy resolution in energy dispersive detectors, and
refinement of energy dispersive spectrometry for full integration into WDS based systems will
be important particularly in improving efficiency for rapid analysis of many chronometer phases
from multiple samples. So-called hyperspectral mapping datasets that combine signals for phase
identification/classification offer a powerful approach to large scale mapping projects involving
multiphase materials significant to petrochronology (e.g., MacRae et al. 2016c). This approach
will undoubtedly gain significance in routine EPMA sample evaluation. In addition, EDS may
be routinely employed in determination of actual accelerating potential in low kV systems,
particularly where sample biasing is used to lower landing voltage and matrix corrections may
be compromised at such low overvoltage if beam voltage is not accurately known.
Dual or multi-beam systems, and continued development of high performance sources
will likewise be important in EPMA-based petrochronologic studies. Electron-laser dual beam
integration may become tenable, allowing initial mapping and elemental analysis by EPMA,
followed by high sensitivity LA mass spectrometry for further trace element evaluation and
isotopic analysis. Continued development of electron sources for higher brightness at higher,
more stable current, along with continued improvements in lowering energy spread for lowering
chromatic aberration and improved performance at low kV. Multi-beam characterization machines
are already finding use in life sciences, and may find utility in extremely high resolution geologic
sample characterization as well, allowing detailed compositional analysis not accessible by
X-ray tomography, and a scale not practicable by tomographic atom probe (Eberle et al. 2015).
As mentioned above, the very high current density required for high sensitivity—high
spatial resolution trace element analysis is problematic. Therefore, continued exploration of
surface conductive coatings and anti-contamination are an important frontier. In particular, as
beam landing energy is lowered, more of the interaction volume involves interaction with the
conductive coating along with enhanced contamination effects (e.g., Gopon et al. 2013). Advanced
materials based on nanostructures such as graphene or stanine for high electrical and thermal
conductivity may be developed that provide reliable surfaces for microanalysis (Park et al. 2016).
Low contamination vacuum systems have improved remarkably over the past few decades,
and improvements in vacuum technology will continue to provide even cleaner environments.
Likewise, improvements in mounting media may replace epoxies that release carbon under
vacuum and beam exposure. Anticontamination systems are already commonplace, and continue
to be evaluated and improved. Advancements such as plasma cleaners are increasingly common
in SEMs for low kV characterization, and are now being considered for EPMA instruments, as
could other methods for sputtering or actively removing contaminants.
EPMA instrumentation includes electron detectors, specifically separate secondary and
backscattered detectors. However, electron microscopy has benefited greatly from newer
detection systems to better allow interrogation of the full electron spectrum at tunable energies
to enhance phase contrast. These include in-lens designs that can offer energy discrimination
as well as having the advantage of being less exposed to contaminants in the sample chamber.
For petrochronology, any advancement in phase imaging is potentially useful.
Software development always plays an important role in advancing analytical technologies,
and EPMA continues to see many exceptional improvements affecting all areas of data acquisition
and processing. Improvements in mapping software that permit very rapid acquisition of a broad
array of X-ray and other signals (electron, IR, visible, UV, EBSD, etc.), automatic identification
and location of accessory phases, and even automated mapping of these phases at high resolution
(e.g., MacRae et al. 2016c). Such mapping could potentially also take advantage of techniques
devised to limit beam exposure to the grains of interest only (polygon or other shape-limited
beam or stage motion), and potentially improve analytical efficiency significantly.
Thermochemical aspects
One major goal of future petrochronology research is to establish the thermodynamic
properties and phase relationships of accessory chronometer phases. That is, to
quantitatively incorporate accessory phases into phase diagrams. Thermodynamic modeling
with modified thermodynamic databases such as that done by Spear and coworkers (Spear
and Pyle 2010; Spear 2010) or Kelsey (i.e., Kelsey et al. 2008; Kelsey and Powell 2011)
are pioneering studies in this context. In addition, experimental studies constraining phase
relationships and trace-element partitioning (i.e., Rubatto and Hermann 2007; Stepanov
et al. 2012) are needed to provide critical thermodynamic data and stability relationships.
New dissolution-precipitation experiments, such as those carried out by Harlov and

coworkers (Harlov et al. 2007, 2011) are needed in order to better illuminate the controls on
accessory phase recrystallization and replacement and importantly, the composition of the
reactive fluids. Subtle differences in fluid composition can have a significant effect on the
degree of recrystallization (Harlov and Hetherington 2010). This certainly can complicate
petrochronological studies but also will provide new avenues for constraining fluid–rock
interactions (Rasmussen et al. 2007; Rasmussen and Muhling 2009; Peterman et al. 2016).
A first step toward better integrating accessory phases into petrologic phase diagrams
involves constraining, even qualitatively, the reactions that produce chronometer phases. The
electron microprobe can quantify the more abundant trace elements in chronometers including
monazite, titanite, xenotime, and zircon. However, the trend is clearly toward incorporating
the broadest suite of trace elements possible. We see a future involving a combination of
high-spatial resolution EPMA characterization with LA-ICP-MS analysis of larger domains
for greater compositional resolution. It is important to stress that the electron microprobe is
unsurpassed as a tool for characterizing spatial/compositional relationships by X-ray mapping
in situ, the absolutely critical first step in petrochronology. Trace elements in associated silicate
phases will generally need to be evaluated by LA-ICP-MS or ion microprobe. The classic
studies of Bea are seminal (Bea et al. 1994; Bea 1996), but many more are needed, particularly
those in which pre- and post-reaction compositions of silicate phases can be characterized. In
all cases, these trace element studies need to be closely integrated with in-situ characterization
and compositional mapping of accessory and silicate phases. As noted above, trace element
mapping of silicate phases holds great potential, and in certain minerals, can be readily made
by electron microprobe. For less abundant trace elements and less stable silicates, LA-ICP-
MS (i.e., Koening 2010; Ubide et al. 2015; Kylander-Clark 2017; Lanari and Engi 2017), ion
microprobe, or even synchrotron X-ray mapping (i.e., Dyl et al. 2014) may be necessary.
One major challenge for integrating accessory and silicate phases involves the different
scales (and degrees) of equilibrium for trace elements and for major elements and even for
different trace elements (Berger et al. 2005). Compositions of accessory phases may reflect a
much smaller equilibrium domain and a more restricted assemblage compared to the silicate
phases. Bulk-rock analyses from hand samples or thin-section chips have generally been
extensively used for forward modeling of silicate assemblages. For trace element modeling,
evaluation of local compositional subdomains may be more important (for example see
Chernoff and Carlson 1999). Multiscale compositional mapping, including trace-element
mapping, and analysis by electron microprobe is ever more essential for evaluating equilibrium
domains and for selecting chronometer domains that reflect specific reactions (i.e., Fig. 9).
New dating techniques and instrumentation will undoubtedly be developed allowing smaller
and smaller domains to be dated more and more precisely. Because of its spatial resolution
and increasing efficiency in major and trace element capability it is also clear that the EPMA
will play a central role in petrologic and petrochronologic analysis for the foreseeable future.
ACKNOWLEDGMENTS
The authors sincerely than Harold Stowell, Philippe Goncalves, Igor Villa, and an
anonymous reviewer for their careful and helpful review comments. In addition, we sincerely
thank Pierre Lanari for excellent review comments and editorial handling. Jeffrey Webber and
Sean Regan are thanked for their contributions to this research and for reading an earlier draft
of the manuscript. The research presented in this manuscript was partly supported by NSF
grants NSF/EAR-1419843 and NSF/EAR-1419876 to M.L. Williams for research in norther
Saskatchewan and the Adirondack Mountains respectively.
REFERENCES
Allaz J, Williams ML, Jercinovic MJ, Donovan J (2011) A new technique for electron microprobe trace element
analysis: the multipoint background method. EMAS Annual Meeting, Angers (France)
Allaz J, Selleck B, Williams ML, Jercinovic MJ (2013) Dating fluid events through the microprobe dating of
detrital monazite from the Potsdam Sandstone, NY, USA. Am Mineral 98:1106–1119
Ancey M, Bastenaire F, Tixier R (1978) Application des methodes statistiques en microanalyse. In: Microanalyse,
Microscopie Eléctronique à Balayage. Maurice F, Meny L, Tixier R, (eds). Les Exlitions du Physicien, Orsay,
p 323–347
Armstrong JT, McSwiggen P, Nielsen3 C (2013) A thermal field-emission electron probe microanalyzer for
improved analytical spatial resolution. Microsc Anal 27:18–22
Asami M, Suzuki K, Adachi M (1996) Monazite ages by the chemical Th-U-total Pb isochron method for pelitic
gneisses from the eastern Sor Rondane Mountains, East Antarctica. Proc NIPR Symp Antarct Geosci 9:49–64
Bea F (1996) Residence of REE, Y, Th and U in granites and crustal protoliths; Implications for the chemistry of
crustal melts. J Petrol 37:521–552, doi:10.1093/petrology/37.3.521
of Ivrea-Verbano, NW Italy. Geochim Cosmochim Acta 63:1133–1153, doi:10.1016/S0016-7037(98)00292–0
Bea F, Pereira MD, Stroh A (1994) Mineral/leucosome trace-element partitioning in a peraluminous migmatite (a
laser ablation–ICP-MS study). Chem Geol 117:291–312, doi:10.1016/0009–2541(94)90133–3
Bea F, Montero P, Garuti G, Zacharini F (1997) Pressure-Dependence of Rare Earth Element Distribution in
Amphibolite- and Granulite- Grade Garnets. A LA-ICP-MS Study. Geostand Newslett 21:253–270,
doi:10.1111/j.1751-908X.1997.tb00674.x
Berger A, Scherrer NC, Bussy F (2005) Equilibration and disequilibration between monazite and garnet: indication
from phase-composition and quantitative texture analysis. J Metamorph Geol 23:865–880, doi:10.1111/
j.1525–1314.2005.00614.x
Bowles JFW (1990) Microanalytical methods in mineralogy and geochemistry age dating of individual grains
of uraninite in rocks from electron microprobe analyses. Chem Geol 83:47–53, doi:10.1016/0009–
2541(90)90139-X
Budzyń B, Harlov DE, Williams ML, Jercinovic MJ (2011) Experimental determination of stability relations
between monazite, fluorapatite, allanite, and REE-epidote as a function of pressure, temperature, and fluid
composition. Am Mineral 96:1547–1567, doi:10.2138/am.2011.3741
Buick IS, Clark C, Rubatto D, Hermann J, Pandit M, Hand M (2010) Constraints on the Proterozoic evolution
of the Aravalli–Delhi Orogenic belt (NW India) from monazite geochronology and mineral trace element
geochemistry. Lithos 120:511–528, doi:10.1016/j.lithos.2010.09.011
Burgess S, Li X, Holland J, Statham P, Bhadare S, Birtwistle D, Protheroe A (2014) Development of soft X-ray
microanalysis using windowless SDD technology. Microsc Microanal 20 (suppl 3):1–2
Catlos EJ (2013) Versatile Monazite: resolving geological records and solving challenges in materials science.
Generalizations about monazite: Implications for geochronologic studies. Am Mineral 98:819–832,
doi:10.2138/am.2013.4336
Chafe AN, Villa IM, Hanchar JM, Wirth R (2014) A re-examination of petrogenesis and 40Ar/39Ar systematics in
the Chain of Ponds K-feldspar: “diffusion domain” archetype versus polyphase hygrochronology. Contrib
Mineral Petrol 167:1010, doi:10.1007/s00410-014-1010-x
Cherniak DJ (2010) Cation diffusion in feldspars. Rev Mineral Geochem 72:691–733
Cherniak DJ, Pyle JM (2008) Th diffusion in monazite. Chem Geol 256:52–61
Cherniak DJ, Watson EB, Grove M, Harrison TM (2004) Pb diffusion in monazite: a combined RBS/SIMS study.
growth. Geology 27:555–558, doi:10.1130/0091–7613(1999)027<0555:tezaar>2.3.co;2
Clarke GL, Daczko NR, Nockolds C (2001) A method for applying matrix corrections to X-ray intensity maps
using the Bence–Albee algorithm and Matlab. J Metamorph Geol 19:635–644, doi:10.1046/j.0263–
4929.2001.00336.x
Cocherie A, Legendre O, Peucat JJ, Kouamelan A (1998) Geochronology of polygenetic monazites constrained
by in situ electron microprobe Th–U–Total Pb determination: implications for Pb behavior in monazite.
Condon DJ, Hodges KV, Alsop GI, White A (2006) Laser ablation 40Ar/39Ar dating of metamorphic fabrics in the
Caledonides of North Ireland. J Geol Soc London 163:337–345, doi:10.1144/0016-764904-066
Dahl PS, Hamilton MA, Jercinovic MJ, Terry MP, Williams ML, Frei R (2005) Comparative isotopic and chemical
geochronometry of monazite, with implications for U–Th–Pb dating by electron microprobe; an example
from metamorphic rocks of the eastern Wyoming Craton (USA). Am Mineral 90:619–638
De Andrade V, Vidal O, Lewin E, O’Brien P, Agard P (2006) Quantification of electron microprobe compositional
maps of rock thin sections: an optimized method and examples. J Metamorph Geol 24:655–668, doi:10.1111/
j.1525–1314.2006.00660.x
Didier A, Bosse V, Boulvais P, Bouloton J, Paquette J-L, Montel J-M, Devidal J-L (2013) Disturbance versus
preservation of U–Th–Pb ages in monazite during fluid–rock interaction: textural, chemical and isotopic
in situ study in microgranites (Velay Dome, France). Contrib Mineral Petrol 165:1051–1072, doi:10.1007/
s00410-012-0847-0
Donovan JJ, Lowers HA, Rusk BG (2011) Improved electron probe microanalysis of trace elements in quartz. Am
Mineral 96:274–282, doi:10.2138/am.2011.3631
Donovan JJ, Singer JW, Armstrong JT (2016) EPMA method for fast trace element analysis in simple matrices.
Am Mineral 101:1839–1853
Dumond, G. PG, Williams ML, Jercinovic MJ (2010) Subhorizontal fabric in exhumed continental lower crust
and implications for lower crustal flow: Athabasca granulite terrane, western Canadian Shield. Tectonics 29
TC2006, doi:10.1029/2009TC002514
Dumond G, Mahan KH, Williams ML, Jercinovic MJ (2013) Transpressive uplift and exhumation of continental
lower crust revealed by synkinematic monazite reactions. Lithosphere 5:507–512, doi:10.1130/l292.1
Dumond G, Goncalves P, Williams ML, Jercinovic MJ (2015) Monazite as a monitor of melting, garnet growth and
feldspar recrystallization in continental lower crust. J Metamorph Geol 33:735–762, doi:10.1111/jmg.12150
Dyl KA, Cleverley JS, Bland PA, Ryan CG, Fisher LA, Hough RM (2014) Quantified, whole section trace element
mapping of carbonaceous chondrites by synchrotron X-ray fluorescence microscopy: 1. CV meteorites.
Geochim Cosmochim Acta 134:100–119, doi:10.1016/j.gca.2014.02.020
Eberle AL, Mikula S, Schalek R, Lichtman J, Tate MLK, Zeidler D (2015) High-resolution, high-throughput
imaging with a multibeam scanning electron microscope. J Microsc 259:114–120, doi:10.1111/jmi.12224
Finger F, Broska I, Roberts MP, Schermaier A (1998) Replacement of primary monazite by apatite-allanite-epidote
coronas in an amphibolite facies granite gneiss from the Eastern Alps. Am Mineral 83:248–258, doi:10.2138/
am-1998-3-408
Flowers RM, Bowring SA, Williams ML (2006a) Timescales and significance of high-pressure, high-temperature
metamorphism and mafic dike anatexis, Snowbird tectonic zone, Canada. Contrib Mineral Petrol 151:558–581
Flowers RM, Mahan KH, Bowring SA, Williams ML, Pringle MS, Hodges KV (2006b) Multistage exhumation
and juxtaposition of lower continental crust in the western Canadian Shield; linking high-resolution
U/Pb and 40Ar/39Ar thermochronometry with pressure–temperature–deformation paths. Tectonics 25,
doi:10.1029/2005TC001912
Sci Lett 228:125–142
Vermont, USA. Chem Geol 401:151–168, doi:10.1016/j.chemgeo.2015.02.017
Gibson HD, Carr SD, Brown RL, Hamilton MA (2004) Correlations between chemical and age domains in
monazite, and metamorphic reactions involving major pelitic phases: an integration of ID-TIMS and
SHRIMP geochronology with Y–Th–U X-ray mapping. Chem Geol 211:237–260
Goldstein JI, Wood F (1966) Experimental procedures for the determination of trace elements by electron probe
microanalysis. First National Electron Probe Meeting, College Park, Md
Golostnn JI (1967) Distribution of germanium inthe metallic phases of some iron-meteorites. J Geophys Res
72:4689–4696
Goncalves P, Williams ML, Jercinovic MJ (2005) Electron-microprobe age mapping of monazite. Am Mineral
90:578–587
Goodrich CA, Treiman AH, Filiberto J, Gross J, Jercinovic M (2013) K2O-rich trapped melt in olivine in the
Nakhla meteorite: Implications for petrogenesis of nakhlites and evolution of the Martian mantle. Meteorit
Planet Sci 48:2371–2405, doi:10.1111/maps.12226
Gopon P, Fournelle J, Sobol PE, Llovet X (2013) Low-voltage electron-probe microanalysis of Fe–Si compounds
using soft X-rays. Microsc Microanal 19:1698–1708, doi:10.1017/S1431927613012695
Grand’Homme A, Janots E, Seydoux-Guillaume A-M, Guillaume D, Bosse V, Magnin V (2016) Partial resetting
Geology 44:431–434, doi:10.1130/g37770.1
Growdon ML, Kunk MJ, Wintsch RP, Walsh GJ (2013) Telescoping metamorphic isograds; evidence from
40
Ar/39Ar dating in the Orange-Mulford Belt, southern Connecticut. Am J Sci 313:1017–1053
Harlov DE, Foerster H-J (2002) High-grade fluid metasomatism on both a local and a regional scale; the Seward
Peninsula, Alaska, and the Val Strona di Omegna, Ivrea-Verbano Zone, northern Italy; Part II, Phosphate
mineral chemistry. J Petrol 43:801–824
Harlov DE, Wirth R, Hetherington CJ (2007) The relative stability of monazite and huttonite at 300–900 ºC and
200–1000 MPa: Metasomatism and the propagation of metastable mineral phases. Am Mineral 92:1652–1664
dissolution–reprecipitation in element redistribution and mass transfer. Contrib Mineral Petrol 162:329–348,
doi:10.1007/s00410-010-0599-7
granulite facies metamorphism in the Val Malenco lower crust. J Metamorph Geol 21:833–852, doi:10.1046/
j.1525–1314.2003.00484.x
Hetherington CJ, Jercinovic MJ, Williams ML, Mahan KH (2008) Understanding geologic processes with xenotime:
Composition, chronology, and a protocol for electron probe microanalysis. Chem Geol 254:133–147
Hickmott DD, Spear FS (1992) Major- and trace-element zoning in garnets from calcareous pelites in the NW
Shelburne Falls Quadrangle, Massachusetts: Garnet growth histories in retrograded rocks. J Petrol 33:965–1005
Hombourger C, Outrequin M (2013) Quantitative analysis and high resolution X-ray mapping with a field emission
electron microprobe. Microsc Today 21:10–15
Janots E, Brunet F, Goffé B, Poinssot C, Burchard M, Cemič L (2007) Thermochemistry of monazite-(La) and
dissakisite-(La): implications for monazite and allanite stability in metapelites. Contrib Mineral Petrol 154:1–
14, doi:10.1007/s00410-006-0176-2
Janots E, Engi M, Berger A, Allaz J, Schwarz JO, Spandler C (2008) Prograde metamorphic sequence of REE
from 250 to 610 °C. J Metamorph Geol 26:509–526, doi:10.1111/j.1525–1314.2008.00774.x
Jercinovic MJ, Williams ML (2005) Analytical perils (and progress) in electron microprobe trace element analysis
applied to geochronology: Background acquisition interferences, and beam irradiation effects. Am Mineral
90:526–546
Jercinovic MJ, Williams ML, Lane ED (2008) In-situ trace element analysis of monazite and other fine-grained
accessory minerals by EPMA. Chem Geol 254:197–215
Jercinovic MJ, Williams ML, Allaz J, Donovan JJ (2012) Trace analysis in EPMA. IOP Conf Ser: Mater Sci Eng
32:012012
Kelly NM, Harley SL, Möller A (2012) Complexity in the behavior and recrystallization of monazite during high-T
metamorphism and fluid infiltration. Chem Geol 322–323:192–208, doi:10.1016/j.chemgeo.2012.07.001
SiO2–H2O–TiO2–ZrO2 system. J Metamorph Geol 29:151–166, doi:10.1111/j.1525–1314.2010.00910.x
systems; examples using model metapelitic and metapsammitic granulites. J Metamorph Geol 26:199–212,
doi:10.1111/j.1525–1314.2007.00757.x
Keppler H, Wyllie PJ (1990) Role of fluids in transport and fractionation of uranium and thorium in magmatic
processes. Nature 348:531–533
Koenig AE (2010) Methodology for detailed trace element mapping of garnet by laser ablation ICP-MS; a look
at unraveling zoning and inclusions. Abstracts with Programs–Geological Society of America 42:627–628
Kohn M (2013) Geochemical zoning in metamorphic minerals. In: Treatise on Geochemistry, vol 3: The Crust.
Rudnick R, (ed) Elsevier, p. 249–280
28:1127–1130, doi:10.1130/0091–7613(2000)28<1127:rntrif>2.0.co;2
Kohn MJ, Wieland MS, Parkinson CD, Upreti BN (2005) Five generations of monazite in Langtang gneisses;
implications for chronology of the Himalayan metamorphic core. J Metamorph Geol 23:399–406
Krenn E, Finger F (2004) Metamorphic formation of Sr-apatite and Sr-bearing monazite in a high-pressure rock
from the Bohemian Massif. Am Mineral 89:1323–1329
345:99–112, doi:10.1016/j.chemgeo.2013.02.019
Geochem 83:55–102
Lanari P, Rolland Y, Schwartz S, Vidal O, Guillot S, Tricart P, Dumont T (2014) P–T–t estimation of deformation
in low-grade quartz–feldspar-bearing rocks using thermodynamic modelling and 40Ar/39Ar dating techniques:
example of the Plan-de-Phasy shear zone unit (Briançonnais Zone, Western Alps). Terra Nova 26:130–138,
doi:10.1111/ter.12079
Larson KP, Cottle JM, Godin L (2011) Petrochronologic record of metamorphism and melting in the upper
Greater Himalayan sequence, Manaslu–Himal Chuli Himalaya, west-central Nepal. Lithosphere 3:379–392,
doi:10.1130/l149.1
Lifshin E, Doganaksoy N, Sirois J, Gauvin R (1999) Statistical considerations in microanalysis by energy-
dispersive spectrometry. Microsc Microanal 4:598–604
MacRae CM, Wilson NC, Torpy A, Bergmann J, Takahashi H (2016a) Holistic mapping—first results combining
SXES, Windowless-SDD and CL spectrometry in an EPMA. MAS EPMA 2016 Topical Conference Program
Guide with Abstracts 82–83
MacRae CM, Wilson NC, Torpy A, Bergmann J, Takahashi H (2016b) Collecting and analyzing 1.6 eV–20 keV
emission spectra in an EPMA. Microsc Microanal 22 (Suppl 3):410–411
MacRae CM, Torpy A, Glenn AM, Pownceby MI, Grey IE, Wilson NC, Pundas P (2016c) Zircon zonation and
trace chemistry characterized by mapping and analysis. MAS EPMA 2016 Topical Conference Program
Guide with Abstracts 110–111
Mahan KH, Goncalves P, Williams ML, Jercinovic MJ (2006) Dating metamorphic reactions and fluid flow:
application to exhumation of high-P granulites in a crustal-scale shear zone, western Canadian Shield. J
McFarlane CRM, Connelly JN, Carlson WD (2005) Monazite and xenotime petrogenesis in the contact aureole of
the Makhavinekh Lake Pluton, northern Labrador. Contrib Mineral Petrol 148:524–541
McSwiggen P, Mori N, Takakura M, Nielsen C (2011) Improving analytical spatial resolution with the JEOL field
emission electron microprobe. Microsc Microanal 17 (Suppl 2):624–625
Merlet C, Bodinier J-L (1990) Electron microprobe determination of minor and trace transition elements in silicate
minerals: A method and its application to mineral zoning in the peridotite nodule PHN 1611. Chem Geol
83:55–69, doi:10.1016/0009–2541(90)90140–3
Montel J, Foret S, Veschambre M, Nicollet C, Provost A (1996) Electron microprobe dating of monazite. Chem
Geol 131:37–53
Moran K, Wuhrer R (2016) Current state of combined EDS–WDS quantitative X-ray mapping. Microsc Microanal
22 (Suppl 3):92–93
Muhling JR, Fletcher IR, Rasmussen B (2012) Dating fluid flow and Mississippi Valley type base-metal
mineralization in the Paleoproterozoic Earaheedy Basin, Western Australia. Precambrian Res 212–213:75–
90, doi:10.1016/j.precamres.2012.04.016
Müller W, Kelley SP, Villa IM (2002) Dating fault-generated pseudotachylytes: comparison of 40Ar/39Ar stepwise-
heating, laser-ablation and Rb–Sr microsampling analyses. Contrib Mineral Petrol 144:57–77, doi:10.1007/
s00410-002-0381-6
Overstreet WC (1967) The geologic occurrence of monazite. U S Geological Survey Professional Paper
Park JB, Kim YJ, Kim SM, Yoo JM, Kim Y, Gorbachev R, Barbolina II, Kim SJ, Kang S, Yoon MH, Cho SP (2016)
Non-destructive electron microscopy imaging and analysis of biological samples with graphene coating. 2D
Mater 3:045004
Parrish RR (1990) U–Pb dating of monazite and its application to geological problems. Can J Earth Sci 27:1431–
1450, doi:10.1139/e90-152
Parslow GR, Brandstatter F, Kurat G, Thomas D (1985) Chemical ages and mobility of U and Th in anatectites of
the Cree Lake Zone, Saskatchewan. Can Mineral 23:543–551
Peterman EM, Snoeyenbos DR, Jercinovic MJ, Kylander-Clark A (2016) Dissolution–reprecipitation
metasomatism and growth of zircon within phosphatic garnet in metapelites from western Massachusetts.
Am Mineral 101:1792–1806, doi:10.2138/am-2016-5524
Pyle JM, Spear FS, Wark DA (2002) Electron microprobe analysis of REE in apatite, monazite and xenotime;
protocols and pitfalls. Rev Mineral Geochem 48:337–362
Pyle JM, Spear FS, Wark DA, Daniel CG, Storm LC (2005) Contributions to precision and accuracy of monazite
microprobe ages. Am Mineral 90:547–577, doi:10.2138/am.2005.1340
Rapp RP, Ryerson FJ, Miller CF (1987) Experimental evidence bearing on the stability of monazite during crustal
anaatexis. Geophys Res Lett 14:307–310, doi:10.1029/GL014i003p00307
Rasmussen B (1996) Early-diagenetic REE-phosphate minerals (florencite, gorceixite, crandallite, and xenotime)
in marine sandstones; a major sink for oceanic phosphorus. Am J Sci 296:601–632, doi:10.2475/ajs.296.6.601
Rasmussen B, Muhling JR (2007) Monazite begets monazite: evidence for dissolution of detrital monazite and
reprecipitation of syntectonic monazite during low-grade regional metamorphism. Contrib Mineral Petrol
154:675–689, doi:10.1007/s00410-007-0216-6
Rasmussen B, Muhling JR (2009) Reactions destroying detrital monazite in greenschist-facies sandstones from
the Witwatersrand basin, South Africa. {Stepanov, 2012 #1783}Chem Geol 264:311–327, doi:10.1016/j.
chemgeo.2009.03.017
Rasmussen B, Sheppard S, Fletcher IR (2006) Testing ore deposit models using in situ U–Pb geochronology
of hydrothermal monazite: Paleoproterozoic gold mineralization in northern Australia. Geology 34:77–80,
doi:10.1130/g22058.1
Rasmussen B, Fletcher IR, Muhling JR (2007) In situ U–Pb dating and element mapping of three generations
of monazite: Unravelling cryptic tectonothermal events in low-grade terranes. Geochim Cosmochim Acta
71:670–690, doi:10.1016/j.gca.2006.10.020
Reed SJB (1993) Electron Microprobe Analysis. Cambridge University Press, Cambridge
Regis D, Warren CJ, Mottram CM, Roberts NMW (2016) Using monazite and zircon petrochronology to constrain
the P–T–t evolution of the middle crust in the Bhutan Himalaya. J Metamorph Geol 34:617–639, doi:10.1111/
jmg.12196
Robertson VE, McSwiggen P (2016) Low Voltage, high spatial resolution, field emission SEM imaging coupled
with new low energy X-ray spectrometers, a novel new technique—successes and challenges. MAS EPMA
2016 Topical Conference Program Guide with Abstracts:86–87
Rocha BC, Moraes R, Moller A, Cioffi CR, Jercinovic MJ (2017) Timing of anataxis and melt crystallization in the
Socorro-Guaxupe Nappe, SE Brazil: Insights from trace element composition of zircon, monazite, and garnet
coupled to U–Pb geochronology. Lithos 277:337–355, doi: 10.1016/j.lithos.2016.05.020
Rubatto D, Hermann J (2007) Experimental zircon/melt and zircon/garnet trace element partitioning and implications
for the geochronology of crustal rocks. Chem Geol 241:38–61, doi:10.1016/j.chemgeo.2007.01.027
Rubatto D, Hermann J, Buick IS (2006) Temperature and bulk composition control on the growth of monazite and
zircon during low-pressure anatexis (Mount Stafford, central Australia). J Petrol 47:1973–1996
Ruiz-Agudo E, Putnis CV, Putnis A (2014) Coupled dissolution and precipitation at mineral–fluid interfaces. Chem
Geol 383:132–146, doi:10.1016/j.chemgeo.2014.06.007
Scherrer NC, Engi M, Gnos E, Jakob V, Liechti A (2000) Monazite analysis; from sample preparation to microprobe
age dating and REE quantification. Schweiz Mineral Petrogr Mitt 80:93–105
Schneider S, Hammerschmidt K, Rosenberg CL (2013) Dating the longevity of ductile shear zones: Insight from
40
Ar/39Ar in situ analyses. Earth Planet Sci Lett 369–370:43–58, doi:10.1016/j.epsl.2013.03.002
Seydoux-Guillaume A-M, Montel J-M, Bingen B, Bosse V, de Parseval P, Paquette J-L, Janots E, Wirth R (2012)
Low-temperature alteration of monazite: Fluid mediated coupled dissolution–precipitation, irradiation
damage, and disturbance of the U–Pb and Th–Pb chronometers. Chem Geol 330–331:140–158, doi:10.1016/j.
chemgeo.2012.07.031
Spear FS (2010) Monazite-allanite phase relations in metapelites. Chem Geol 279:55–62, doi:10.1016/j.
chemgeo.2010.10.004
Spear FS, Pyle JM (2010) Theoretical modeling of monazite growth in a low Ca metapelite. Chem Geol 273:111–
119, doi:10.1016/j.chemgeo.2010.02.016
Spear FS, Pyle JM, Cherniak D (2009) Limitations of chemical dating of monazite. Chem Geol 266:218–230,
doi:10.1016/j.chemgeo.2009.06.007
implications for the REE, Th and U geochemistry of crustal rocks. Chem Geol 300–301:200–220,
doi:10.1016/j.chemgeo.2012.01.007
Susan D, Grant RP, Rodelas JM, Michael JR, Maguire MC (2015) Comparing field emission electron microprobe
to traditional EPMA for analysis of metallurgical specimens. Microsc Microanal 21 (Suppl 3):2107–2108
Suzuki K, Adachi M (1991) Precambrian provenance and Silurian metamorphism of the Tsunosawa paragneiss
in the South Kitakami terrane, Northeast Japan, revealed by the chemical Th–U–total Pb isochron ages of
monazite, zircon and xenotime. J Geochem 25:357–376
Terauchi M, Takahashi H, Handa N, Murano T, Koike M, Kawachi T, Imazono T, Koeda M, Nagano T, Sasai
H, Oue Y (2012) Ultrasoft X-ray emission spectroscopy using a newly designed wavelength-dispersive
spectrometer attached to a transmission electron microscope. J Electron Microsc 61:1–8
Tracy RJ (1982) Compositional zoning and inclusions in metamorphic minerals. Rev Mineral 10:355–397
Ubide T, McKenna CA, Chew DM, Kamber BS (2015) High-resolution LA-ICP-MS trace element mapping of
igneous minerals; in search of magma histories. Chem Geol 409:157–168, doi:10.1016/j.chemgeo.2015.05.020
Vallini D, Rasmussen B, Krapež B, Fletcher IR, McNaughton NJ (2002) Obtaining diagenetic ages from
metamorphosed sedimentary rocks: U–Pb dating of unusually coarse xenotime cement in phosphatic
sandstone. Geology 30:1083–1086, doi:10.1130/0091–7613(2002)030<1083:odafms>2.0.co;2
Vance D, Müller W, Villa IM (eds) (2003) Geochronology: linking the isotope record with petrology and
textures
Villa IM (2016) Diffusion in mineral geochronometers: Present and absent. Chem Geol 420:1–10, doi:10.1016/j.
chemgeo.2015.11.001
Villa IM, Williams ML (2012) Geochronology of metasomatic events. In:Metasomatism and the Chemical
Transformation of Rock. Harlov DE, Austrheim H (eds). Springer-Verlag, Heidelberg, p. 171–202
Villa IM, Hermann J, Müntener O, Trommsdorff V (2000) 39Ar−40Ar dating of multiply zoned amphibole
generations (Malenco, Italian Alps). Contrib Mineral Petrol 140:363–381, doi:10.1007/s004100000197
Villa IM, Bucher S, Bousquet R, Kleinhanns IC, Schmid SM (2014) Dating polygenetic metamorphic assemblages
along a transect across the Western Alps. J Petrol 55:803–830, doi:10.1093/petrology/egu007
Villaseca C, Martı́n Romera C, De la Rosa J, Barbero L (2003) Residence and redistribution of REE, Y, Zr, Th
and U during granulite-facies metamorphism: behaviour of accessory and major phases in peraluminous
granulites of central Spain. Chem Geol 200:293–323, doi:10.1016/S0009-2541(03)00200–6
Wells ML, Spell TL, Hoisch TD, Arriola T, Zanetti KA (2008) Laser-probe 40Ar/39Ar dating of strain fringes:
Mid-Cretaceous synconvergent orogen-parallel extension in the interior of the Sevier orogen. Tectonics 27:
TC3012, doi:10.1029/2007TC002153. doi:10.1029/2007TC002153
Wendt I, Carl C (1991) The statistical distribution of the mean squared weighted deviation. Chem Geol; Isotope
Geosci Sect 86:275–285
White AP, Hodges KV (2003) Pressure–temperature–time evolution of the Central East Greenland Caledonides:
quantitative constraints on crustal thickening and synorogenic extension. J Metamorph Geol 21:875–897,
doi:10.1046/j.1525–1314.2003.00489.x
Whitney DL, Evans BW (2010) Abbreviations for names of rock-forming minerals. Am Mineral 95:185–187,
doi:10.2138/am.2010.3371
Williams ML, Jercinovic MJ (2012) Tectonic Interpretation of metamorphic tectonites: integrating
compositional mapping, microstructural analysis, and in-situ monazite dating. J Metamorph Geol
30:739–732, doi:10.1111/j.1525–1314.2012.00995.x
Williams ML, Jercinovic MJ, Terry M (1999) High resolution “age” mapping, chemical analysis, and chemical
dating of monazite using the electron microprobe: A new tool for tectonic analysis. Geology 27:1023–1026
Williams ML, Jercinovic MJ, Goncalves P, Mahan KH (2006) Format and philosophy for collecting, compiling,
and reporting microprobe monazite ages. Chem Geol 225:1–15
Williams ML, Jercinovic MJ, Hetherington CJ (2007) Microprobe monazite geochronology: understanding
geologic processes by integrating composition and chronology. Ann Rev Earth Planet Sci 35:137–175
Williams ML, Jercinovic MJ, Harlov DE, Budzyn B, Hetherington CJ (2011) Resetting monazite ages during fluid-
related alteration. Chem Geol 283:218–225, doi:10.1016/j.chemgeo.2011.01.019
contact and regional metamorphism of pelites; petrology and geochronology. Contrib Mineral Petrol
145:228–250
Wintsch RP, Aleinikoff JN, Yi K (2005) Foliation development and reaction softening by dissolution and
precipitation in the transformation of granodiorite to orthogneiss, Glastonbury Complex, Connecticut, U.S.A.
Can Mineral 43:327–347, doi:10.2113/gscanmin.43.1.327
Wong MS, Williams ML, McLelland JM, Jercinovik MJ, Kowalkoski J (2012) Late Ottawan extension in the
eastern Adirondack Highlands: Evidence from structural studies and zircon and monazite geochronology.
Geol Soc Am Bull 124:857–869, doi:10.1130/b30481.1
Xing L, Trail D, Watson EB (2013) Th and U partitioning between monazite and felsic melt. Chem Geol 358:46–
53, doi:10.1016/j.chemgeo.2013.07.009
Yang P, Pattison D (2006) Genesis of monazite and Y zoning in garnet from the Black Hills, South Dakota. Lithos
88:233–253, doi:10.1016/j.lithos.2005.08.012
Yang P, Rivers T, Jackson SJ (1999) Crystal-chemical and thermal controls on trace-element partitioning between
coexisting garnet and biotite in metamorphic rocks from western Labrador. Can Mineral 37:443–468
Zi J-W, Rasmussen B, Muhling JR, Fletcher IR, Thorne AM, Johnson SP, Cutten HN, Dunkley DJ, Korhonen
FJ (2015) In situ U–Pb geochronology of xenotime and monazite from the Abra polymetallic deposit in
the Capricorn Orogen, Australia: Dating hydrothermal mineralization and fluid flow in a long-lived crustal
structure. Precambrian Res 260:91–112, doi:10.1016/j.precamres.2015.01.010
Ziebold TO (1967) Precision and sensitivity in electron microprobe analysis. Anal Chem 36:858–861
Vol. 83 pp. 183–198, 2017 6
Petrochronology by
Laser-Ablation Inductively Coupled Plasma
Mass Spectrometry
Andrew R. C. Kylander-Clark
Department of Earth Science
University of California, Santa Barbara
Santa Barbara, CA 93106-9630
U.S.A.
kylander@geol.ucsb.edu
INTRODUCTION
Petrochronology is a field of Earth science in which the isotopic and / or elemental
composition of a mineral chronometer is interpreted in combination with its age, thus yielding
a more synergistic combination of petrology and chronology that can be used to interpret
geologic processes. It has recently attracted renewed interest as technologies for mineral
analysis have improved. Examples are many, and continue to grow, from the early adoption
of U / Th ratios in zircon as an indicator for magmatic vs. igneous crystallization (e.g., Ahrens
1965), to using the Nd isotopic composition in titanite to track source contribution over time
(see Applications; B. R. Hacker, personal communication). Age and chemical information can
be obtained by a variety of techniques: electron microprobe (age; major and minor elements;
see Williams et al. 2017), secondary ion mass spectrometry (SIMS; age; trace elements;
isotopic ratios; see Schmitt and Vazquez 2017), and laser-ablation inductively coupled plasma
mass spectrometry (LA-ICPMS; age; trace elements; isotopic ratios).
Laser-ablation ICPMS instrumentation and techniques, the focus of this chapter, have
been employed as a petrochronologic tool for decades, starting with separate analyses of ages
and elemental and / or isotopic compositions, which were then combined and interpreted. For
example, Zheng et al. (2009) employed LA-ICPMS to analyze the trace-element (TE) chemistry,
Hf isotopic composition, and age of zircons from kimberlites by using three spots on each zircon
grain, one for each type of analysis. This work was relatively time consuming and expensive,
given the required number of analytical sessions, but yielded far better confidence in the
conclusions, because of the link between physical conditions (petrology) and time (chronology).
Instrumentation and techniques which employ LA-ICPMS have continued to improve,
particularly in the ease with which petrochronologic data can be obtained. A single LA-
ICPMS instrument can now measure both the age and TE composition of one spot in a grain.
Alternatively, the laser aerosol can be split for simultaneous measurement of U–Pb age or
isotopic composition on one instrument and age and / or elemental concentration on another.
One of the first examples of the latter was given by Chen et al. (2010), in which—using the
first laser-ablation split-stream (LASS) setup, described by Yuan et al. (2008)—the authors
distinguished 4 different types of metamorphic zircon (re)crystallization, based on linked U–Pb
age, Hf isotope ratio and TE composition analyses from ultrahigh-pressure rocks in China.
The purpose of this paper is to describe the analytical setup and measurement of age
and chemistry of mineral chronometers used for petrochronology. LA-ICPMS techniques
that measure only age or chemistry have been previously described in detail; the focus

1943-2666/17/0083-0006$05.00 (online)
184 Kylander-Clark
herein is on recent analytical developments specifically related to petrochronology that

have occurred within the last decade. Extensive reviews of the basics of LA-ICPMS
instrumentation and analytical techniques can be found in Jackson et al. (1992), Jackson et
al. (2004), Longerich (2008), Schaltegger et al. (2015) and references therein.
VIRTUES OF LA-ICPMS
Laser ablation has advantages and disadvantages. First, analysis is relatively fast,
inexpensive and straightforward. Preparation of samples requires only a moderately well-
polished grain mount or thin section (grains or rock fragments can also be mounted); the laser
is much more forgiving of surface topography than SIMS, though pits and µm-scale scratches
can negatively affect precision and accuracy. Improvements in the technique have increased
the accuracy for U–Pb spot dating—now limited to ~ 1–2%—and most LA-ICPMS systems
can achieve this result with ≤ 60 second analyses (e.g., Košler et al. 2013); TE analyses are
similar in duration, but the ppm-level precision necessary for most isotopic analyses requires
analyses of up to 2 minutes on advanced systems (e.g., Kylander-Clark et al. 2013).
This short analysis time (an order of magnitude faster than SIMS), and the lower cost
of instrumentation, yields data with unprecedented ease, but is not without drawbacks.
First, a significant amount of material is removed from the sample—unlike electron-probe
microanalysis (EPMA) sampling, and much more than SIMS. Typical spot sizes for U–Pb age
analyses can be as small as 10 µm wide by a few µm deep (~1 ng of material), for samples with
ppm levels of radiogenic Pb. The most recent advances in reducing ablated material show that
even a single pulse of the laser (~100 nm depth) can yield U–Pb dates with a precision of 5%
and yield ages that are within 2% of accepted values (Cottle et al. 2009a, 2012; Viete et al.
2015). Isotopic tracer analyses such as for Hf require larger spots with a ~100 ng minimum
of ablated material (e.g., a 50 × 20 µm spot). Second, although the limit of detection for TE
is much better than for EPMA (sub-ppm vs. 10s of ppm respectively, depending on element,
spot size, etc.), ICPMS precision and accuracy is generally worse than EPMA. Thus, when
performing analyses by LA-ICPMS, it is desirable to perform all EPMA major and minor
element analyses required for interpretation of the data prior to ablation of the sample,
and, if sample conservation is critical, the extra time and cost of SIMS may be appropriate.
Nevertheless, use of LA-ICPMS for petrochronology has increased at a near-exponential rate
for the last 20 years and is arguably becoming the primary means by which to acquire isotopic
and elemental data on mineral chronometers (Fig. 1).
2000
publications
1800
SIMS
1600 LA-ICPMS
1400
1200 Figure 1. Publications involving petrochro-
1000 nology, with SIMS and LA-ICPMS, per year,
summed over 5 year intervals. LA-ICPMS
800 continues to grow at an exponential rate.
Source: Google Scholar.
600
400
200
year
0
1995 2000 2005 2010 2015
Figure 1. Publications involving petrology and

geochronology, with SIMS and LA-ICPMS, per year.
Petrochronology by LA-ICP-MS 185
INSTRUMENTATION
Petrochronologic analyses by LA-ICPMS generally consist of two types: age + TE
concentration, or age + isotopic composition. Three different mass spectrometers are generally
employed for such analyses, and each instrument has its benefits and drawbacks. The following
describes each system, and how it is typically used.
Multi-Collector (MC) ICPMS
A MC-ICPMS consists of an ICP source, a series of focusing lenses, an electrostatic
analyzer (energy filter), a magnet (mass analyzer), and a detector array. The ion source of
all ICP instruments is at atmospheric pressure, which therefore requires several step-down
vacuum chambers, through which the ion beam travels on its way to the detector. Multi-
collector instruments tend to be the largest, most precise instruments, and thus much effort
is placed to reduce the pressure in the analyzer region to ≤ 10−8 mbar (running pressure). This
has the effect of decreasing the possible number of collisions, thereby increasing sensitivity
and improving abundance sensitivity. An MC-ICPMS uses multiple collectors to measure all
isotopes of interest simultaneously. The advantage over a single-collector (SC) instrument
(described below) is two-fold: first, sensitivity is increased linearly by measuring each isotope
in a different collector. For example, for U–Pb measurements, 6 different isotopes—238U, 232Th,
208
Pb, 207Pb, 206Pb, and 204Pb—are measured coincidentally on an MC-ICPMS, whereas each
has to be measured sequentially on an SC-ICPMS. Given equal dwell time, 6 times as many
counts are measured on the MC-ICPMS, increasing precision by ~2.5 ×. Second, because all
isotopes are measured at the same time, any change in signal intensity affects all isotopic
measurements at the same time, such that measured ratios remain constant. Multi-collector
instruments are designed to be extremely stable, and as a result, it takes too long to switch
the magnet (i.e., mass) during the course of a laser ablation analysis. Given their stability
and sensitivity, MC-ICPMS is used exclusively for ratio measurements, both for U–Pb age
determinations and single-element isotope ratio measurements such as for Sr (titanite, apatite,
calcite), Nd (titanite, monazite), Hf (zircon), and Pb (apatite, titanite, calcite).
Single-collector (SC) ICPMS
SC-ICPMS instruments are inherently less sensitive than MC-ICPMS instruments. They
are designed for rapid mass switching and therefore the instrument of choice in which a
large number of masses are analyzed and precision is less important. Element concentration
analysis fits into this category, but SC instruments can also be used for isotopic analyses,
such as U–Pb, and in recent cases Rb–Sr (Zack and Hogmalm 2016). Two fundamental types
of SC instruments exist, though both generally serve the same purpose: a quadrupole MS is
less sensitive, and has less resolving power than a magnetic-sector MS, but is considerably
less expensive and can be more flexible, especially if analysis across a large mass range is
desirable. A magnetic-sector instrument requires relatively long settling times for magnet
switching, so small mass jumps are obtained by steering the beam with a voltage potential.
This minimizes the number of magnet positions during a single cycle.
Petrochronologic analyses can be performed solely by SC-ICPMS, whereby the TE
composition and U–Pb age are determined entirely by one instrument, wherein each sweep of
the mass range includes all necessary isotopes (e.g., Iizuka and Hirata 2004; Yuan et al. 2008;
Kohn and Corrie 2011; Kylander-Clark et al. 2013). This technique can be sufficient in many
cases, however, compared to LASS analyses in which a MC-ICPMS is used for U–Pb age
determination, this yields relatively poor precision on the age, and requires larger pits, more
analyses, or both, to yield the ultimate precision on an age of 1–2%.
186 Kylander-Clark
Laser
The laser-ablation system may be the most critical component of any LA-ICPMS setup.
Two main components of the LA system affect the quality of the data: 1) laser wavelength and
pulse width, and 2) geometry of the sample cell. Absorbance of light by minerals increases as
the wavelength of the light decreases. Lasers with very short (femtosecond) pulse widths have
been shown to minimize melting of the sample (Russo et al. 2002), which produces volatility-
controlled inter-element fractionation as the laser pit is progressively deepened—an effect
referred to as “down-hole laser-induced elemental fractionation (LIEF)”. Thus, instrumentation
has moved towards employing lasers with shorter wavelength and shorter pulse widths. Both
yield more consistent and cleaner ablation craters, producing particles with smaller sizes and
more uniform size distributions (Guillong et al. 2003), which are in turn ionized more efficiently
in the ICP. This increase in sensitivity allows smaller pit depths, reducing LIEF, which is a major
factor limiting the application of analyses of standards to analyses of unknowns. Because LIEF
varies least within a single mineral type, it is best to use a matrix-matched standard, i.e., a standard
of the same mineral as the unknown. This applies to both geochronology and geochemistry;
however, suitable homogenous reference materials (RMs) are not yet available for every type of
analysis. Many matrix-matched reference materials exist for U–Pb geochronologic analyses for
the most commonly employed geochronometers (zircon, monazite, titanite, apatite, and rutile),
but fewer have been found / calibrated for TE analyses. Fortunately, internal standards—such as
Ca, Si, and Ti—can be employed with non-matrix-matched reference materials, such that signal
intensity can be partially corrected for variations in ablation rate between the reference material
and the unknown. Because accuracy and precision of an age tend to be more important than the
accuracy and precision on elemental abundance, the use of a short wavelength laser and matrix-
matched standard is more critical in the acquisition of geochronologic data.
Some of the most important advances in improving accuracy, precision, and reproducibility
of LA-ICPMS data have come from improving flow dynamics in the sample chamber. This is
justly so, as the inter-element mass bias is strongly affected by the flow of the carrier gas across
the sample surface (Bleiner and Günther 2001). Modern MC-ICPMS instruments can measure
U–Pb ratios with a precision of much better than 1% for a single spot analysis, however,
relatively large deviations (> 2%) in inter-element fractionation can occur from spot to spot
due to changes in flow dynamics in different parts of the cell (e.g., Bleiner and Bogaerts 2006;
Muller et al. 2009), and thus cell design in recent years has been aimed at creating identical
flow characteristics in every part of the cell. That way, samples and standards are subject to the
same flow-induced fractionation—that is, the conditions under which the unknown is ablated
are the same as the standard, and the true vs. measured ratio of each is the same.
The first sample chambers used for LA-ICPMS were simple single, small-volume cavities
in blocks or discs. The newest cells involve a dual-volume design, employing a small-volume
cup that resides inside a large-volume cell that contains several samples. The small-volume
portion allows for rapid washout times (e.g., Bleiner and Günther 2001), and similar flow
dynamics independent of position within the large-volume cell. The first conceptions of
this dual-volume design—such as the HelEx cell (Eggins et al. 2005) found on the latest
Photon Machines lasers—require an arm from inside the cell to the outside, and a moderately
complicated system to keep the cell sealed as the arm position changes. A similar configuration
exists for ESI lasers, however, other configurations, such as the RESOlution S155 cell, have
a larger cell that includes the entire sample stage, allowing the small-volume cup to remain
stationary in the large-volume cell minimizing the potential for changes in flow dynamics
between sample and standard. If sealed and aligned properly, both designs work with the same
efficiency and accuracy and are vast improvements over earlier cell designs.
TREATMENT OF UNCERTAINTIES
The treatment of uncertainties for all LA-ICPMS petrochronologic data invariably
differs from lab to lab and is likely poorly understood by many users, such that uncertainty
propagation of data in submitted manuscripts varies considerably. This variability is
problematic, because it makes comparison of datasets from different sources quite difficult.
Some geochronologic datasets from a single laboratory can be compared at a level better than
2% (2σ). However, it should be assumed that any two datasets from different LA-ICPMS
laboratories cannot be distinguished, should their ages lie within 2% of one another (Košler
et al. 2013). Furthermore, unless a study produces some confirmation of reproducibility,
there can be little confidence in the data presented. Therefore, it is necessary to report the
reproducibility of primary and secondary reference materials within each session.
Some attempts have been made to standardize the propagation of uncertainty in U–Pb
data (Horstwood et al. 2016; McLean et al. 2016) and recommendations are published
online (www.plasmage.org). Single element concentration and isotope ratio data should
be treated in a similar manner with propagation of uncertainty using stated methods,
including results for primary and secondary RMs. As noted below, uncertainties in TE
data typically receive less scrutiny than isotopic data, possibly because composition is
commonly compared in log space or as internally-derived ratios. Nevertheless, TE data
can be collected and reduced with a wide variety of methods, and as such, those methods
should be reported along with estimates on uncertainties for compositional data.
LASER ABLATION SPLIT STREAM (LASS)

Most recently, LA-ICPMS petrochronology has begun to take advantage of the virtues
of each instrument: the MC-ICPMS, for its superior isotope-ratio measurements, and the SC-
ICPMS, for its rapid mass switching. In such a configuration, the laser aerosol + He ± Ar carrier
gas is split into two streams, each of which are sent to the two different mass spectrometers.
Several configurations are possible, and will be discussed in sequence.
The first demonstrated LASS technique used a Nu Plasma MC-ICPMS coupled
with an Elan 6100DRC quadrupole instrument (Yuan et al. 2008). This setup enabled the
petrochronologic analysis of zircon, whereby Hf isotopic data were measured on the MC-
ICPMS, and U–Pb age ± TE data were measured on the quadrupole. This type of analysis
saves time and money in terms of laboratory operations, compared to separate sessions
of analysis by each instrument. More importantly, it guarantees that the same volume of
material is used for the 176Hf /177Hf ratio, the U–Pb age, and TE concentration. The downside
is that Hf isotopic data requires a large signal intensity—for both acceptable signal:noise
ratio and measurement precision—and thus, diverting part of the laser aerosol from the
MC-ICPMS to the quadrupole instrument, requires the spot size to virtually double. As an
example, Yuan et al. (2008) required a 44 µm diameter spot in the LASS configuration to
get the same precision that they achieved using a 32 µm spot with only the MC-ICPMS; this
is a nearly a 2-fold increase in spot area and volume. Furthermore, precision of the U–Pb
age data is compromised using this technique: a SC-ICPMS has a much lower effective
sensitivity than a MC-ICPMS, and is further reduced when TE data are also collected.
Another example is given in Figure 2. In this case, U–Pb zircon ages from a LASS analysis
on the quadrupole are shown with U–Pb data from a MC-ICPMS; the quadrupole analyses
included TE data (Si, Ti, Y, Zr, Nb, REE, Hf), but the signal intensity was increased by ~40 × over
the MC-ICPMS analyses by using a larger spot (65 µm vs 15 µm) and higher laser repetition rate
(8 Hz vs. 4 Hz). Even with the increase in ablated volume for the SC-ICPMS, the precision of the
U–Pb date using the SC-ICPMS is far inferior, and some samples with small, few, and complexly
zoned petrochronometer minerals may not afford an abundance of large-spot analyses.
188 Kylander-Clark
0.082
quadupole LASS U-Pb+TE (65 μm, 8 Hz) uncertainties are 2σ
Concordia Age = 1063.8 ± 6.0 Ma

MSWD = 0.24
1160
0.078 MC-ICPMS U-Pb (15 μm, 4 Hz)
1120 Concordia Age = 1062.7 ± 2.1 Ma
MSWD = 0.58
1080
0.074 1040
Pb/206Pb
1000
0.070 1100
207
LASS MC-ICPMS U-Pb (25 μm, 4 Hz)

Concordia Age = 1063.1 ± 1.8 Ma 1040
U/206Pb
238 MSWD = 0.93
0.066
5.0 5.2 5.4 5.6 5.8 6.0 6.2
Figure 2. U–Pb2.zircon
Figure Data age
fromdata
twofrom two sessions,
sessions, highlighting
highlighting differences
difference in analytical
in analytical uncertainty between
uncertainty
quadrupole and MC-ICPMS
between quadrupole data. Data
data andcollected
MC-ICPMS withdata.
the quadrupole
Data takenwas during
with a LASS session
the quadrupole was(approxi-
mately half the alaser
during LASSaerosol was(approximately
session diverted to a MC-ICPMS for aerosol
half the laser Hf isotopic
wasmeasurement)
directed to aand
MC-also includes
TE concentration
ICPMS fordata, employing
Hf isotopic a 65 μm spot and
measurement) andtwice
also the laser repetition
includes rate of U–Pb
trace-element age analyses on the
concentrations,
MC-ICPMS. Much better precision on U–Pb zircon ages can be obtained using
employing a 65 μm spot and twice the rep rate of U-Pb analyses on the MC-ICPMS.the MC-ICPMS; a 25 μm spot
using LASS (inset) yields approximately the same precision as a 15 μm spot using solely the MC-ICPMS.
Much better precision on U-Pb ratios can be obtained using the MC-ICPMS; a 25 μm
spot using LASS (inset) yields approximately the same precision as a 15 μm spot using
Onesolely
alternative to measuring all data of interest with just a single spot is to measure Hf
the MC-ICPMS.
isotopes in one session by MC-ICPMS, and use LASS in a second session to measure U–Pb
age by MC-ICPMS and TE data by SC-ICPMS. If the second session is performed using
LASS, a relatively small spot can be used for each session because one has the advantage
of increased Hf isotopic precision in the first session and U–Pb precision in the second
session; Figure 2 exemplifies this. This gives the user some freedom to place 2 smaller spots
in roughly the same area as required by a single spot, which can be more advantageous in
grains with complex zoning. This alternative method takes longer and is more costly; for
simple, large grains, a single session of LASS, with isotopic composition measured on a
MC-ICPMS and U–Pb age ± TE data measured on a SC-ICPMS is the most effective. At the
time of this writing, LASS Hf-isotopic composition + U–Pb age ± TE has been employed
only for metamorphic and igneous zircon (e.g., Chen et al. 2010; Wu et al. 2010).
If isotopic data are not required, LASS can be employed with MC-ICPMS for U–Pb
age and SC-ICPMS for TE analysis (Kylander-Clark et al. 2013). As noted in the previous
paragraph and shown in Figure 2, this technique was developed to take advantage of the
superior stability and sensitivity of the MC-ICPMS to retrieve precise and accurate age data
from single spots, while measuring elemental concentrations on a SC-ICPMS. As with the
LASS Hf + U–Pb age ± TE technique described above, a larger spot for a single analysis takes
the place of two single smaller spots. In this case, however, because measurements on a MC-
ICPMS are so precise, sufficiently small spots can still be employed with excellent precision
on the age. The precision and accuracy of element concentrations have not received the same
scrutiny as U–Pb dates, and depend on many factors—including the matrix characteristics
of the reference materials and unknowns—but LASS provides the same accuracy, albeit
with a drop in precision commensurate with reduced signal intensity (Kylander-Clark et
al. 2013). One important feature of LASS is the ability to match TE data with U–Pb data
throughout the analysis to identify inclusions or changes in growth zoning within a mineral,
down-hole during the analysis. This can be done without LASS (solely on a SC-ICPMS
as described above), but the U–Pb analytical precision is generally prohibitively poor to
recognize changes in U–Pb age with changes in chemistry. Two examples are given in
Figures 3 and 4. The downside of LASS may be the larger spot size which increases the
likelihood of intersecting a growth zone or inclusion; however, growth zoning or inclusions
are difficult to identify without corresponding U–Pb age / TE information.
A
206
Pb/238U
800
Ma
Lu 20
ppm
600
15
400
10
Figure 3. A) LASS
200 5 collected with ot
laser off
laser on
U-Pb age data w

ts began the analys
0 2 4 6 8 10 12 14 16 400 Ma metamo
inherited, ca. 140
B Intercepts: 20
points from A, pl
ca. 400, 1400 Ma time-resolved da
0.09 Lu 10% uncertainty
ppm phic age and inh
0.08 2
Lu concentration
analysis (solid, 2S
theless, the entir
0.07
1040 Average the age/TE data
880 pits.
720
Pb/206Pb
0.06
560
400
0.05
U/ Pb
207
238 206
4 8 12 16
Figure 3. A) LASS trace of a single laser pit in a zircon; Lu data collected with other TE, using a SC-ICPMS
and U–Pb age data was collected using a MC-ICPMS. The laser began the analysis sampling a garnet-stable
(i.e., low Lu) ca. 400 Ma metamorphic rim, but eventually intersected the inherited, ca. 1400 Ma igneous,
high-Lu core. B) Individual data points from A, plotted on a Tera-Wasserburg diagram. The time-resolved
data (transparent ellipses shown with a uniform, 10% uncertainty) yield a discordia array between the meta-
morphic age and inherited age, with a corresponding increase in Lu concentration with apparent age. The
average of the analysis (solid, 2SE, ellipse) is not nearly as informative, nevertheless, the entire example
illustrates the benefits of collecting the age/TE data from one laser pit, rather than two adjacent pits.
190 Kylander-Clark
15000
Hf
ellipses are 2σ
Pb/206Pb
70
0 1400
P ppm
2000
207
60
50
Inclusion
1500
40 112 110 108 106 104 102 100

238
U/206Pb 1000
30
Hf ,Y, P ppm
Nd ppm
20
Y 500
10
P
0 Nd 0
U/206Pb 207Pb/206Pb date
400
Pb cps
200
500
204
0
0
238
Figure 4. The U–Pb/TE trace of 5 laser pits in igneous zircon; TE data was obtained simultaneously with
Figure
U–Pb data4. The U-Pb/TE
(quadrupole trace respectively).
and MC-ICPMS, of 5 laserDuringpits the
in igneous zircon;
second analysis, the laser intersected
TE data was
an inclusion, whichobtained simultaneously
shows up in LREEs, such as Nd, and P with U-Pb
(apatite?); data
elements that are elevated in zircon,
such as Hf and Y, are unaffected. Though there is no measurable increase in 204Pb, the age is significantly
(quadrupole and
affected; most affected is theMC-ICPMS,
207 respectively).
Pb/206Pb age. The coincident TE dataDuring
allows the the
user tosecond
confidently discard
analysis, the laser intersected an inclusion, which shows up in
this data point when calculating an age for the sample.
LREEs,
LASSsuch as Nd,science
has advanced and Pby(apatite?); elements
rapidly providing a wealththat are
of information about both age
elevated in zircon, such as Hf and Y, are unaffected. Though
and geochemistry (see Applications below for examples), however, LASS analyses should be
used to complement, rather than replace the information gained by imaging and major elemental
thereprior
analysis is no measurable
to ablation. increase
It is prudent to make inimagesPb,
204
the
prior age istosignifi-
to analysis guide spot placement—
cathodoluminescence (CL) of zircon, or Y concentration maps of monazite,TE
cantly affected, the Pb/ Pb more so. The coincident
207 206
for example (e.g.,
data
Engi allows
2017; Lanaritheand user to confidently
Engi 2017; Williams et al. discard
2017). Thesethisimages,
dataand pointmajor- and minor-
element analyses, used for thermobarometry (e.g., monazite–garnet; Pyle et al. 2001) for example,
when
are easier tocalculating
gather prior toan age for
ablation, whichthecansample.
remove substantial areas of grains of interest.
Once prior, non-destructive mapping is done, maps can also be made using LA(SS)-
ICPMS, which can add significant insight to processes which formed / modified the
geochronometers and / or coeval phases (Woodhead et al. 2007; Paul et al. 2014; Ubide et al.
2015). An example is shown in Figure 5; original igneous zircon is replaced and overgrown
by rims and embayments of metamorphic zircon. Low HREE concentrations coincide with
(neo / re)crystallized zircon, implying that metamorphism occurred in the garnet-stability
field, whereas apparent age data is partly decoupled, implying that the U–Pb system may
have been modified by a later, different process.
High Low age/TE
Lu
Ce
Dy
Sm
Figure 5. LASS m
locations) of ign
Age underwent part
quent ca. 400 M
and embaymen
moderately less
that of REE, sug
inherted and (n
lized radiogenic
maps greatly in
of data.
analyzed zircon
200 μm
Figure 5. LASS maps (circles on Sm map show spot locations) of igneous zircon that grew at ca. 900 Ma,
and underwent partial (re/neo)crystallization during a subsequent ca. 400 Ma garnet-stable metamorphic
event. Rims and embayments contain considerably less HREE and moderately less MREE. Age map does
not perfectly match that of REE, suggesting that late fissures (cutting both inherted and (neo/re)crystallized
zircon) may have mobilized radiogenic Pb. The combination of age, TE and CL maps greatly increase the
ability to interpret any single set of data.
Plumbing for LASS

There are a few different possible combinations to feed the He, Ar and ablated material
into two different ICPMS instruments (Fig. 6). The simplest arrangement is a T- or Y-connector
placed after the Ar + He mixing plumbing (be it a mixing bulb, ‘T-connector,’ etc.), sending
192 Kylander-Clark
sample
He in1 cup
He in2
sample Laser
cell
homogenizer
Ar in1
Ar in2 Ar in3
restricting
valves
SC-ICP MC-ICP
Figure 6. Configuration of instrumental setup for petrochronology as described in the text.
approximately halfFigure
of the6.gas and aerosol toinstrumental
A configurable each ICPMS setup
instrument.
for The advantage of this
setup is its simplicity,
LASS but it leaves little room
petrochronology to tune each
as described ininstrument
the text. separately by changing
the Ar flow rate. Because each instrument will require a specific gas flow for maximum
sensitivity, the backpressure of each line must be matched well enough so that tuning can take
place by either changing the z-position of the torch, or by increasing or decreasing the auxiliary
Ar flow to the torch. If the inner-torch inside diameter (ID) of each instrument is similar (and
smaller than the tubing to the torch), the backpressure is likely to be the same in each line, and
minor adjustments for maximum sensitivity of both instruments should be possible. If not, the
backpressure in each line can be equalized by using similar length, identical ID tubing to both
instruments, or by placing a restricting valve to the instrument with lower initial backpressure.
In theory, if an uneven distribution of signal intensity is desired, one can force more of the
aerosol to one instrument by adding Ar to each ICPMS after the split (see the setup described
in Ibanez-Mejia et al. 2015). An uneven split in Ar will force the aerosol to the line with the
lower additional flow rate. This may be desirable for analyses in which one half of the split
needs greater precision (i.e., count rates). An example of this would be a LASS setup in which
Hf isotopes in zircon are measured on a MC-ICPMS and U–Pb age on a SC-ICPMS. As
mentioned above, because the required precision of Hf isotopes is so much greater than that
for U–Pb age, it would be advantageous to send a larger proportion to the MC-ICPMS.
APPLICATIONS
LA-ICPMS petrochronology, with and without LASS, has been applied to numerous
minerals—both igneous and metamorphic—including, but not limited to zircon, monazite, titanite,
rutile, and apatite. Sedimentary rocks have also received recent attention, as TE or Hf isotopic
data can complement U–Pb detrital zircon ages to aid in provenance determination, or to better
understand the processes leading to source terrane formation. To give the reader an idea of what is
possible, a few case studies are described in detail, with an emphasis on the latest LASS techniques.
LASS of metamorphic zircon

Young and Kylander-Clark (2015) explored the transformation or lack thereof of felsic
gneiss during subduction to (ultra)high-pressure (UHP) depths. In the Western Gneiss Region
of western Norway, the bulk of the Caledonian (U)HP terrane consists of amphibolite-
facies, plagioclase-bearing (i.e., low-pressure) assemblages. The extent of transformation
is of great interest to those who model the density and strength of continental crust during
subduction and exhumation. Transformation was assessed by comparing the observed
mineral assemblage to that expected at high pressure and by using LASS petrochronology
on zircon; a MC-ICPMS was employed for U–Pb geochronology, and a SC-ICPMS was
used for TE compositions. It was deduced that most, but not all, of the continental crust did
not transform at (U)HP conditions; the LASS data were essential in reaching this conclusion
because the bulk of the data yielded inherited and mixed inherited-Caledonian dates with
typical igneous zircon rare-earth-element (REE) patterns: a steep increase through the
heavy REE (HREE), and a negative Eu anomaly. Mixed analyses could be recognized
through matched U–Pb–TE traces. Caledonian rims yielded different patterns: lower REE
concentrations and reduced HREE / MREE, indicate that the bulk rock underwent minor
prograde, garnet-stable metamorphism early in the subduction / exhumation cycle. The
disappearance of a negative Eu anomaly concomitant with the youngest dates indicate that
some, if minor (re)crystallization, also occurred at near-peak metamorphic conditions.
Single-shot LASS (SS-LASS) analysis of thin metamorphic zircon rims
In an attempt to decipher the timing and conditions of minor metamorphic zircon growth
in HP metamorphic rocks from the Cordillera de la Costa, Venezuela, Viete et al. (2015) used
LASS—MC-ICPMS was used for U–Pb age and a quadrupole MS for TE analysis—to measure
ages and TE on thin (<1 µm) rims of zircon. The authors mounted the zircons without polishing
them, and ablated the rims in single-shot, 100 nm pulses (Cottle et al. 2012) to retrieve pit-depth
changes in age vs. TE concentration; Y was analyzed as a proxy for garnet stability, Ti as a proxy
for temperature, and Eu as a proxy for plagioclase-stability. Because of the transient nature of
the signal, only two TEs were measured in any given session; this resulted in ~20% long-term
precision and accuracy on the analyses. Four discrete pulses of metamorphism were discovered,
between 33–18 Ma, with a negative correlation between age and Y, Th, U, Eu concentration. In
many grains, the first several pulses give the metamorphic crystallization age, and are followed
by pulse-dates indicating an increasing mixture of inherited zircon.
Depth profiling of rutile
Thermally activated Pb diffusion in rutile occurs at sufficiently low temperatures such that
rutile can be used to recover cooling paths (see Zack and Kooijman 2017). Smye and Stockli
(2014) used a LASS configuration with two SC-ICPMS instruments to obtain a µm-scale
depth profile of both U–Pb date (first instrument) and Zr concentration (second instrument).
The authors found a monotonic decrease in age toward the rim; an inversion of that profile
indicated rapid initial cooling in the Early Jurassic, followed by a period of slower cooling.
The Zr profile showed a decrease in concentration from the core, but a sharp increase in the last
few µm at the rim. These profiles indicated that the rutile, which originally crystallized during
a Permian granulite-facies event, was reheated in the Early Jurassic. This conclusion was only
possible because of the combined age + composition information.
Petrochronology of detrital zircon
Age combined with Th / U ratio may be the earliest petrochronologic tool used for
evaluating detrital zircon data (e.g., Amelin 1998; Hartmann and Santos 2004), as Th / U
ratio has long been known to indicate changes in petrogenetic conditions during zircon (re)
crystallization (Ahrens 1965; Hoskin and Schaltegger 2003 and references therein). The
194 Kylander-Clark
Th / U ratio in detrital zircon via standard LA-ICPMS is easily obtained and has therefore
been used frequently (e.g., Bingen et al. 2005; Iizuka et al. 2005; Phillips et al. 2006). Recent
advances have led to the ability to combine Hf isotopic data with age; for example a recent
study by Ibanez-Mejia et al. (2015) used LASS data (MC-ICPMS for Hf and a SC-ICPMS
for U–Pb age) from (meta)sediments in the northern Andes to better understand how the
Amazon craton was involved in the assembly of Rodina. Using the combined age–isotopic
data, the authors found an early interval of steeply increasing 176Hf / 177Hf with decreasing
age, followed by a later interval in which the trend shallows; these data were interpreted to
indicate two contrasting phases of orogenesis. Other studies have combined TE composition
with the age of detrital zircon data; Johnston and Kylander-Clark (2016) tracked the evolution
of the Mesozoic California margin using detrital zircon age and chemistry (MC-ICPMS for
U–Pb age and quadrupole for TE concentrations). The authors showed evidence for the rapid
exhumation and denudation of a Jurassic ophiolite in the forearc from a distinct set of zircons
found only in 102–97 Ma sediments, as well as present supporting evidence for forearc and
intra-arc shortening during the Late Cretaceous via abrupt changes in HREE and Eu / Eu*.
Monazite petrochronology
Monazite has had an increasingly important role as a petrochronometer because of its
affinity for both U and Th (and thus age determination), and a broad range of TE that can be
linked to (re)crystallization conditions (Foster et al. 2002; Kohn and Malloy 2004; Kohn et al.
2005; Finger and Krenn 2007; Cottle et al. 2009b,c; Janots et al. 2009). Because of its large
concentrations of Th and U (commonly 1000s of ppm) monazite grains can be analyzed with
small laser spots, allowing for rapid acquisition of both U–Th / Pb and elemental composition
data. For example, Stearns et al. (2013) used ages and HREE chemistry (MC-ICPMS + SC-
ICPMS) of Himalayan and Pamir monazite to document synchronous prograde–retrograde
garnet growth, in which 28–20 Ma monazite revealed decreasing HREE abundance, and
20–15 Ma monazite showed increasing HREE abundances. Štípská et al. (2015) documented
monazite growth in several metamorphic grades along a P–T–d path; staurolite-grade
monazite yielded older ages with high Y and HREE (garnet unstable), kyanite-grade monazite
contained the same early zones in their cores, but with low Y, HREE mantles (garnet stable),
and sillimanite-grade monazite contained both staurolite- and kyanite-grade monazite in their
cores and mantles, respectively, but with high Y, HREE rims (retrograde garnet resorption).
Holder et al. (2015) used LASS (MC-ICPMS + quadrupole) to examine the change of
monazite chemistry during high-grade metamorphism, concluding that coincident increases
in Sr, Pb, and Eu could be attributed to decrease in feldspar stability at high pressure.
Hacker et al. (2015) applied monazite LASS (MC-ICPMS + quadrupole or SC-ICPMS)
petrochronology to nearly 70 samples from the Scandinavian Caledonides to determine the
response of monazite to widespread, high-grade metamorphism. In this case, the majority of
samples revealed (re)crystallization of monazite during late-stage retrogression; both the age
and chemistry were consistent. Both pre-Caledonian, and peak-Caledonian monazite was
preserved in several cases, again evident in both age and composition.
Monazite also contains considerable concentrations of Nd, and favorable Sm / Nd ratios,
affording the opportunity for in-situ isotopic Nd measurements. Poletti et al. (2017) used LA-
ICPMS U–Th / Pb, elemental composition, and Nd isotopic composition of monazite from
the Mountain Pass Intrusive suite, and were able to genetically link the monazite-bearing
carbonatites to a nearby ultrapotassic suite of plutonic rocks.
Titanite Petrochronology
Titanite is a common trace phase in many quartzofeldspathic rocks that can persist
through multiple metamorphic, igneous, and sedimentary transport events that span a wide
range of pressures, temperatures, and deformation and fluid conditions. It often (re)crystallizes
in response to changes in these parameters and can record these changes—in both age and
chemistry—in distinct compositional domains. Though it can be more problematic to date,

given its lower relative U and Th and higher relative common Pb contents than the other
minerals mentioned in this section, it is widespread, has an affinity for a large range of TE, the
latter of which can aid in the assignment of conditions during (re)crystallization (e.g., Franz
and Spear 1985; Green and Pearson 1986; Frost et al. 2001; Prowatke and Klemme 2005;
Hayden et al. 2008). Given the variable common-Pb content of titanite, the rapid analytical
capability of LA-ICPMS is ideally suited for the mineral, because an isochronous discordia
array between an upper intercept, common-Pb value and concordia (on a Tera-Wasserburg
diagram) is easily achieved with a large number of spots (e.g., Storey et al. 2006). Perhaps
the most common use of titanite in petrochronology by LA-ICPMS is that in which the age
and temperature (Zr-in-titanite; Hayden et al. 2008) are linked: Kohn and Corrie (2011) used
combined age–Zr profiles from titanite from the Himalaya to argue for long-lived thick and
weak crustal channels; Gao et al. (2011) used LA-SC-ICPMS titanite data to distinguish
between early titanite that crystallized at relatively cold temperatures, and late retrograde,
high-temperature titanite; Schwartz et al. (2016) employed LASS (MC-ICPMS + quadrupole)
for a persistent high-temperature lower crust from temperature and age data of titanites from
Zealandia; Stearns et al. (2016) was the first to provide depth-profile Zr ppm + U–Pb date via
single-shot LASS (MC-ICPMS + quadrupole), and argued for (re)crystallization rather than
diffusional modification of Pamir titanites. Other LA-ICPMS studies have used other TE or
isotopic analyses in titanite in concert with their age in order to solve geologic problems:
Gao et al. (2011) also used REE composition of titanites to distinguish whether titanite grew
in the presence of garnet and / or plagioclase; Stearns et al. (2015) presented TE + U–Pb age
data (LASS; MC-ICPMS + quadrupole) from titanites in which age, Zr, Y, Sr covaried; B. R.
Hacker (personal communication; Fig. 7) used in-situ Nd isotopic data to reveal changes in
melt isotopic composition through time throughout the evolution of the Pamir plateau.
0
Miocene Oligocene Eocene
-2
-4
e Nd (t)
-6
-8
-10
-12
-14
10 15 20 25 30 35 40 45
age (Ma)
Figure 7. Petrochronologic (U–Pb age + Nd isotopes) data of Pamir titanites obtained by LA-ICPMS show
that intrusions become more evolved over time.
time.
196 Kylander-Clark
CONCLUDING REMARKS
Petrochronologic data via LA-ICPMS, particularly by LASS, is more readily obtained
and widely used than ever before, and new applications are continuing to be developed at
an increasing rate. Improvements in instrument sensitivity, stability, laser coupling and gas-
flow dynamics have broadened the applicability of LA-ICPMS petrochronology to a larger set
of geologic problems, but limitations in necessary precision, sample size, composition still
exist, and, in most cases LA(SS)-ICPMS petrochronology is most effective when combined
with analytical data from other sources. A range of options exist, from single-instrument
age–composition analysis to multiple-instrument, multiple session age–isotopic composition
analysis, and choosing the right type of analysis requires careful considerations of all the
requirements involved in producing interpretable results.
ACKNOWLEDGMENTS
This manuscript was significantly improved by inputs from reviewers D. Chew and P.
Sylvester, editor P. Lanari, and B. Hacker.
REFERENCES
Ahrens LH (1965) Some observations on the uranium and thorium distributions in accessory zircon from granitic
rocks. Geochim Cosmochim Acta 29:711–716, doi:10.1016 / 0016–7037(65)90064–5
Amelin YV (1998) Geochronology of the Jack Hills detrital zircons by precise U–Pb isotope dilution analysis of
crystal fragments. Chem Geol 146:25–38, doi:10.1016 / S0009-2541(97)00162–9
Bingen B, Griffin WL, Torsvik TH, Saeed A (2005) Timing of Late Neoproterozoic glaciation on Baltica
constrained by detrital zircon geochronology in the Hedmark Group, south-east Norway. Terra Nova 17:250–
258, doi:10.1111 / j.1365–3121.2005.00609.x
Bleiner D, Günther D (2001) Theoretical description and experimental observation of aerosol transport processes
in laser ablation inductively coupled plasma mass spectrometry. J Anal Atom Spectrom 16:449–456,
doi:10.1039 / b009729m
Bleiner D, Bogaerts A (2006) Computer simulations of laser ablation sample introduction for plasma-source
elemental microanalysis. J Anal Atom Spectrom 21:1161–1174, doi:10.1039 / b607627k
Chen R-X, Zheng Y-F, Xie L (2010) Metamorphic growth and recrystallization of zircon: Distinction by
simultaneous in-situ analyses of trace elements, U–Th–Pb and Lu–Hf isotopes in zircons from eclogite-facies
rocks in the Sulu orogen. Lithos 114:132–154, doi:10.1016 / j.lithos.2009.08.006
Cottle JM, Horstwood MSA, Parrish RR (2009a) A new approach to single shot laser ablation analysis and its
application to in situ Pb / U geochronology. J Anal Atom Spectom 24:1355–1363, doi:10.1039 / b821899d
Cottle JM, Kylander-Clark ARC, Vrijmoed JC (2012) U–Th / Pb geochronology of detrital zircon and monazite
by single shot laser ablation inductively coupled plasma mass spectrometry (SS-LA-ICPMS). Chem Geol
332–333:136–147, doi:10.1016 / j.chemgeo.2012.09.035
Cottle JM, Searle MP, Horstwood MSA, Waters DJ (2009b) Timing of midcrustal metamorphism, melting, and
deformation in the Mount Everest region of Southern Tibet revealed by U(–Th)–Pb geochronology. J Geol
117:643–664, doi:10.1086 / 605994
Cottle JM, Jessup MJ, Newell DL, Horstwood MSA, Noble SR, Parrish RR, Waters DJ, Searle MPCTC (2009c)
Geochronology of granulitized eclogite from the Ama Drime Massif: Implications for the tectonic evolution
of the South Tibetan Himalaya. Tectonics 28, doi:10.1029 / 2008tc002256
Eggins SM, Grün R, McCulloch MT, Pike AW, Chappell J, Kinsley L, Mortimer G, Shelley M, Murray-Wallace
CV, Spötl C, Taylor L (2005) In situ U-series dating by laser-ablation multi-collector ICPMS: new prospects
for Quaternary geochronology. Quat Sci Rev 24:2523–2538, doi:10.1016 / j.quascirev.2005.07.006
Geochem 83:365–418
loop. Lithos 95:103–115, doi:10.1016 / j.lithos.2006.06.003
Foster G, Gibson HD, Parrish R, Horstwood M, Fraser J, Tindle A (2002) Textural, chemical and isotopic insights
into the nature and behaviour of metamorphic monazite. Chem Geol 191:183–207
Franz G, Spear FS (1985) Aluminous titanite (sphene) from the Eclogite Zone, south-central Tauern Window,
Austria. Chem Geol 50:33–46, doi:10.1016 / 0009–2541(85)90110-X
Frost BR, Chamberlain KR, Schumacher JC (2001) Sphene (titanite); phase relations and role as a geochronometer.
Chem Geol 172:131–148
Gao XY, Zheng YF, Chen YX (2011) U–Pb ages and trace elements in metamorphic zircon and titanite from UHP
eclogite in the Dabie orogen: constraints on P–T–t path. J Metamorph Geol 29:721–740, doi:10.1111 / j.1525–
1314.2011.00938.x
Green TH, Pearson NJ (1986) Rare-earth element partitioning between sphene and coexisting silicate liquid at high
pressure and temperature. Chem Geol 55:105–119, doi:10.1016 / 0009–2541(86)90131–2
Guillong M, Horn I, Gunther D (2003) A comparison of 266 nm, 213 nm and 193 nm produced from a single solid
state Nd:YAG laser for laser ablation ICP-MS. J Anal Atom Spectom 18:1224–1230, doi:10.1039 / b305434a
Geol 409:28–41, doi:10.1016 / j.chemgeo.2015.05.008
Hartmann LA, Santos JOS (2004) Predominance of high Th / U, magmatic zircon in Brazilian Shield sandstones.
Geology 32:73–76, doi:10.1130 / g20007.1
Hayden LA, Watson EB, Wark DA (2008) A thermobarometer for sphene (titanite). Contib Mineral Petr 155:529–
540, doi:10.1007 / s00410-007-0256-y
of (ultra)high-pressure metamorphism: Examples from the Western Gneiss Region, Norway. Chem Geol
409:99–111, doi:10.1016 / j.chemgeo.2015.04.021
Horstwood MS, Košler J, Gehrels G, Jackson SE, McLean NM, Paton C, Pearson NJ, Sircombe K, Sylvester
P, Vermeesch P, Bowring JF (2016) Community-derived standards for LA-ICP-MS U–(Th–)Pb
geochronology—Uncertainty propagation, age interpretation and data reporting. Geostandard Geoanal Res,
doi:10.1111 / j.1751-908X.2016.00379.x
Iizuka T, Hirata T (2004) Simultaneous determinations of U–Pb age and REE abundances for zircons using ArF
excimer laser ablation-ICPMS. GEOCHEMICAL JOURNAL 38:229–241, doi:10.2343 / geochemj.38.229
Iizuka T, Hirata T, Komiya T, Rino S, Katayama I, Motoki A, Maruyama S (2005) U–Pb and Lu–Hf isotope
systematics of zircons from the Mississippi River sand: Implications for reworking and growth of continental
crust. Geology 33:485–488, doi:10.1130 / g21427.1
Jackson SE, Longerich HP, Dunning GR, Fryer BJ (1992) The application of laser ablation-microprobe-inductively
coupled plasma-mass spectrometery (LAM-ICP-MS) to in situ trace-element analysis in minerals. Can
Mineral 30:1049–1064
Jackson SE, Pearson NJ, Griffin WL, Belousova EA (2004) The application of laser ablation-inductively coupled
plasma-mass spectrometry to in situ U–Pb zircon geochronology. Chem Geol 211:47–69, doi:10.1016 / j.
chemgeo.2004.06.017
Janots E, Engi M, Rubatto D, Berger A, Gregory C, Rahn M (2009) Metamorphic rates in collisional orogeny from
in situ allanite and monazite dating. Geology 37:11–14, doi:10.1130 / g25192a.1
Johnston SM, Kylander-Clark ARC (2016) Rapid ophiolite exhumation and arc thickening in the southern california
late cretaceous convergent margin as defined by nacimiento block foreack detrital zircon geochronology and
geochemistry. Geol Soc Am Abstracts with Programs 48:18–5
silicates: Implications for age determinations. Geochim Cosmochim Ac 68:101–113
Evidence for a static hot channel in the Himalayan orogen. Earth Planet Sc Lett 311:136–143, doi:10.1016 / j.
epsl.2011.09.008
doi:10.1111 / j.1525–1314.2005.00584.x
Košler J, Sláma J, Belousova E, Corfu F, Gehrels GE, Gerdes A, Horstwood MS, Sircombe KN, Sylvester
PJ, Tiepolo M, Whitehouse MJ (2013) U–Pb Detrital Zircon Analysis—Results of an inter-laboratory
comparison. Geostandard Geoanal Res 37:243–259, doi:10.1111 / j.1751-908X.2013.00245.x
345:99–112, doi:10.1016 / j.chemgeo.2013.02.019
Geochem 83:55–102
Longerich H (2008) Laser Ablation-Inductively Coupled Plasma-Mass Spectrometry (LA-ICP-MS): An
Introduction. Laser Ablation ICP-MS in the Earth Sciences: Current Practices and Outstanding Issues (P
Sylvester, ed) Mineral Assoc Canada Short Course Ser 40:1–18
McLean N, Bowring J, Gehrels G (2016) Algorithms and software for U–Pb geochronology by LA-ICPMS.
Geochem Geophy Geosy, doi:10.1002 / 2015gc006097
Muller W, Shelley M, Miller P, Broude S (2009) Initial performance metrics of a new custom-designed ArF
excimer LA-ICPMS system coupled to a two-volume laser-ablation cell. J Anal Atom Spectom 24:209–214,
doi:10.1039 / b805995k
198 Kylander-Clark
Paul B, Woodhead JD, Paton C, Hergt JM, Hellstrom J, Norris CA (2014) Towards a method for quantitative
LA-ICP-MS imaging of multi-phase assemblages: mineral identification and analysis correction procedures.
Geostandard Geoanal Res 38:253–263, doi:10.1111 / j.1751-908X.2014.00270.x
Phillips G, Wilson CJL, Campbell IH, Allen CM (2006) U–Th–Pb detrital zircon geochronology from the southern
Prince Charles Mountains, East Antarctica—Defining the Archaean to Neoproterozoic Ruker Province.
Precambrian Res 148:292–306, doi:10.1016 / j.precamres.2006.05.001
Poletti JE, Cottle JM, Hagen-Peter GA, Lackey JS (2017) Petrochronological constraints on the origin of
the Mountain Pass ultrapotassic and carbonatite intrusive suite, California. J Petrol 57:1555–1598,
doi:10.1093 / petrology / egw050
Prowatke S, Klemme S (2005) Effect of melt composition on the partitioning of trace elements between titanite and
silicate melt. Geochim Cosmochim Acta 69:695–709, doi:10.1016 / j.gca.2004.06.037
Pyle JM, Spear FS, Rudnick RL, McDonough WF (2001) Monazite–xenotime and monazite–garnet equilibrium in
a prograde pelite sequence. J Petrol 42:2083–2117
Russo RE, Mao X, Gonzalez JJ, Mao SS (2002) Femtosecond laser ablation ICP-MS. J Anal Atom Spectom
17:1072–1075, doi:10.1039 / b202044k
Schaltegger U, Schmitt AK, Horstwood MSA (2015) U–Th–Pb zircon geochronology by ID-TIMS, SIMS, and
laser ablation ICP-MS: Recipes, interpretations, and opportunities. Chem Geol 402:89–110, doi:10.1016 / j.
chemgeo.2015.02.028
Schwartz JJ, Stowell HH, Klepeis KA, Tulloch AJ, Kylander-Clark ARC, Hacker BR, Coble MA (2016)
Thermochronology of extensional orogenic collapse in the deep crust of Zealandia. Geosphere 12:647–677,
doi:10.1130 / ges01232.1
evolution. Earth Planet Sci Lett 408:171–182, doi:10.1016 / j.epsl.2014.10.013
Stearns MA, Cottle JM, Hacker BR, Kylander-Clark ARC (2016) Extracting thermal histories from the near-rim
zoning in titanite using coupled U–Pb and trace-element depth profiles by single-shot laser-ablation split
stream (SS-LASS) ICP-MS. Chem Geol 422:13–24, doi:10.1016 / j.chemgeo.2015.12.011
Stearns MA, Hacker BR, Ratschbacher L, Rutte D, Kylander-Clark ARC (2015) Titanite petrochronology of the
Pamir gneiss domes: Implications for middle to deep crust exhumation and titanite closure to Pb and Zr
diffusion. Tectonics 34:784–802, doi:10.1002 / 2014tc003774
Stearns MA, Hacker BR, Ratschbacher L, Lee J, Cottle JM, Kylander-Clark ARC (2013) Synchronous Oligocene–
Miocene metamorphism of the Pamir and the north Himalaya driven by plate-scale dynamics. Geology
41:1071–1074, doi:10.1130 / g34451.1
Štípská P, Hacker BR, Racek M, Holder R, Kylander-Clark ARC, Schulmann K, Hasalová P (2015) Monazite
dating of prograde and retrograde P–T–d paths in the Barrovian terrane of the Thaya window, Bohemian
Massif. J Petrol 56:1007–1035, doi:10.1093 / petrology / egv026
Storey CD, Jeffries TE, Smith M (2006) Common lead-corrected laser ablation ICP-MS U–Pb systematics and
geochronology of titanite. Chem Geol 227:37–52, doi:10.1016 / j.chemgeo.2005.09.003
Ubide T, McKenna CA, Chew DM, Kamber BS (2015) High-resolution LA-ICP-MS trace element
mapping of igneous minerals: In search of magma histories. Chem Geol 409:157–168, doi:10.1016 / j.
chemgeo.2015.05.020
Viete DR, Kylander-Clark ARC, Hacker BR (2015) Single-shot laser ablation split stream (SS-LASS)
petrochronology deciphers multiple, short-duration metamorphic events. Chem Geol 415:70–86,
doi:10.1016 / j.chemgeo.2015.09.013
Geochem 83:153–182
Woodhead JD, Hellstrom J, Hergt JM, Greig A, Maas R (2007) Isotopic and elemental imaging of geological
materials by laser ablation inductively coupled plasma-mass spectrometry. Geostandard Geoanal Res
31:331–343, doi:10.1111 / j.1751-908X.2007.00104.x
Wu F-Y, Ji W-Q, Liu C-Z, Chung S-L (2010) Detrital zircon U–Pb and Hf isotopic data from the Xigaze fore-
arc basin: Constraints on Transhimalayan magmatic evolution in southern Tibet. Chem Geol 271:13–25,
doi:10.1016 / j.chemgeo.2009.12.007
Young DJ, Kylander-Clark ARC (2015) Does continental crust transform during eclogite facies metamorphism? J
Metamorph Geol 33:331–357, doi:10.1111 / jmg.12123
Yuan H-L, Gao S, Dai M-N, Zong C-L, Günther D, Fontaine GH, Liu X-M, Diwu C (2008) Simultaneous
determinations of U–Pb age, Hf isotopes and trace element compositions of zircon by excimer laser-ablation
quadrupole and multiple-collector ICP-MS. Chem Geol 247:100–118, doi:10.1016 / j.chemgeo.2007.10.003
Zack T, Hogmalm KJ (2016) Laser ablation Rb / Sr dating by online chemical separation of Rb and Sr in an oxygen-
filled reaction cell. Chem Geol 437:120–133, doi:10.1016 / j.chemgeo.2016.05.027
Zheng JP, Griffin WL, O’Reilly SY, Zhao JH, Wu YB, Liu GL, Pearson N, Zhang M, Ma CQ, Zhang ZH, Yu CM
(2009) Neoarchean (2.7 Ga–2.8 Ga) accretion beneath the North China Craton: U–Pb age, trace elements and
Hf isotopes of zircons in diamondiferous kimberlites. Lithos 112:188–202, doi:10.1016 / j.lithos.2009.02.003
Vol. 83 pp. 199–230, 2017 7
Secondary Ionization Mass Spectrometry Analysis

in Petrochronology
Axel K. Schmitt
Institute of Earth Sciences
Heidelberg University
69120 Heidelberg
Germany
axel.schmitt@geow.uni-heidelberg.de
1,2
Jorge A. Vazquez
1
Stanford-USGS Ion Microprobe Laboratory
U.S. Geological Survey
Menlo Park, CA 94025
2
Stanford University
Stanford, CA 94305
USA
jvazquez@usgs.gov

The goal of petrochronology is to extract information about the rates and conditions at
which rocks and magmas are transported through the Earth’s crust. Garnering this information
from the rock record greatly benefits from integrating textural and compositional data with
radiometric dating of accessory minerals. Length scales of crystal growth and diffusive
transport in accessory minerals under realistic geologic conditions are typically in the range
of 1–10’s of μm, and in some cases even substantially smaller, with zircon having among the
lowest diffusion coefficients at a given temperature (e.g., Cherniak and Watson 2003). Intrinsic
to the compartmentalization of geochemical and geochronologic information from intra-crystal
domains is the requirement to determine accessory mineral compositions using techniques
that sample at commensurate spatial scales so as to not convolute the geologic signals that are
recorded within crystals, as may be the case with single grain or large grain fragment analysis
by isotope dilution thermal ionization mass spectrometry (ID-TIMS; e.g., Schaltegger and
Davies 2017, this volume; Schoene and Baxter 2017, this volume). Small crystals can also be
difficult to extract by mineral separation techniques traditionally used in geochronology, which
also lead to a loss of petrographic context. Secondary Ionization Mass Spectrometry, that is
SIMS performed with an ion microprobe, is an analytical technique ideally suited to meet the
high spatial resolution analysis requirements that are critical for petrochronology (Table 1).
In SIMS, bombardment of solid targets with an energetic ion beam removes atoms from
the sample where primary ions are implanted into the target material to a depth of < 5–10 nm.
Lateral resolution is controlled by primary ion beam dimensions (sub-μm to few 10’s of μm) with
an upper limit set by the acceptance angle of the ion optical system which collects the secondary
ions emitted from the source into the mass spectrometer. This upper limit is typically a few
100’s of μm, provided sample charging can be mitigated, but only rarely are such large areas
analyzed in geologic samples. Depth resolution is typically one order-of-magnitude smaller
than lateral resolution, depending on conditions, and total crater depths for petrochronologic
analysis are few μm at most. Analysis durations are typically a few to 10’s of minutes.
1943-2666/17/0083-0007$05.00 (online)
200 Schmitt & Vazquez
Table 1. Advantages and limitations of in-situ SIMS analysis for petrochronology in comparison
with other isotope selective methods
In-situ analysis capabilities Limitations
Versatile for radiogenic and stable isotopes (−ID-TIMS, Quantitation where ultimate precision and accuracy is
−LA-ICP-MS) limited by sample volume and/or sputtering physics (−
ID-TIMS, ±LA-ICP-MS)
Isotope ratios (e.g., complete U–Th–Pb system with Dependence on reference materials (−ID-TIMS,
common Pb correction) and trace elements (±ID-TIMS, +LA-ICP-MS)
±LA-ICP-MS)
Generates diverse data from spots (including depth Trade-off between small sample volume and precision
profiling) or areas by ion imaging (−ID-TIMS, so that reducing sampling volumes by using smaller
+LA-ICP-MS) beams limits precision, also through higher background
to signal ratios (+ID-TIMS, +LA-ICP-MS)
Spatial selectivity (−ID-TIMS, +LA-ICP-MS) Complex mass spectrum (−ID-TIMS, −LA-ICP-MS)
Isotope analysis in petrographic context (−ID-TIMS, Complex and costly instrumentation required
+LA-ICP-MS) (−ID-TIMS, −LA-ICP-MS)
Reveals mixed populations (±ID-TIMS, +LA-ICP-MS)
Can reveal diffusion in natural and experimental
samples (−ID-TIMS, ±LA-ICP-MS)
High sensitivity and low backgrounds (−ID-TIMS,
−LA-ICP-MS)
No sample chemistry (−ID-TIMS, +LA-ICP-MS)
Individual particle analysis (−ID-TIMS, +LA-ICP-MS)
Minimal sample consumption enabling correlated
analysis (−ID-TIMS, −LA-ICP-MS)
Note: + indicates property shared with other methods; - indicates property unique, and not shared with other methods; other methods: isotope
dilution thermal ionization mass spectrometry ID-TIMS, laser ablation-inductively coupled plasma-mass spectrometry LA-ICP-MS.
In contrast to laser ablation-inductively coupled plasma-mass spectrometry (LA-ICP-MS;

e.g., Kylander-Clark 2017, this volume) the principle of SIMS requires that sample removal
(“sputtering”) and ionization occur in the same location. Hence the sample surface is part of an
ion optical system that controls the trajectories of the primary ion beam, and the secondary ion
emission. While this results in high ion collection efficiencies, it also places strict requirements
on the sample surface which must be flat and oriented parallel to the secondary ion extraction
optical components. Moreover, the sample must be conductive, or coated with a conductive
material, and sustain high vacuum. Sputtering generates complex mass spectra of atomic,
molecular, and multiply charged ion species, which requires a careful spectral evaluation and
the adoption of measures that can mitigate mass spectral interferences. These measures may
result in a loss of transmission in the mass spectrometer.
This article summarizes the strengths of SIMS for petrochronology which include isotope
specific analysis, superb spatial resolution, and high sensitivity. We also address some limitations
of SIMS analysis which comprise potential pitfalls resulting from complex mass spectra and
matrix-dependent variability in secondary ion formation that can compromise data accuracy if
not properly accounted for (Table 1). There are many excellent summary publications of SIMS
methodologies that have focused on key applications in the Earth Sciences: geochronology,
trace element, and stable isotope analysis (e.g., Shimizu and Hart 1982; Ireland 1995; Williams
and McKibben 1998; Ireland and Williams 2003; Stern 2009; Valley and Kita 2009; Ireland
2015). Although this literature is as old as 35 years, the principles of geochemical SIMS
analysis and even some of the fundamental instrumental designs have changed only marginally,
although reliability, routinely achieved precision (especially in multi-collection stable isotope
analysis), and automation have substantially progressed since then. Our goal for this review is
SIMS Analysis in Petrochronology 201
to focus on those aspects of SIMS analysis and data treatment that are of interest to igneous or
metamorphic petrologists and that are unique to the requirements of optimizing the acquisition
of spatially correlated geochronological and geochemical information from crystal domains
and analyzing accessory minerals in a textural context (in-situ analysis sensu stricto). We only
review basic aspects of SIMS geochronology, trace element, and stable isotopic analysis where
it is necessary to provide a context for the integration of these methods; we refer to the excellent
reviews listed above for more detail on individual methodologies.
INSTRUMENTATION AND SAMPLE PREPARATION

Large-magnet radius ion microprobes for petrochronology, and complementary
instrumentation
The chronology aspect in petrochronology is almost exclusively the domain of large
magnet radius SIMS instruments that can routinely resolve the complex mass spectra that
result from sputtering chemically complex accessory minerals. We therefore focus below on
currently used instruments of the CAMECA ims 1270/1280 and extended SHRIMP families
(e.g., SHRIMP-II, and -RG). Smaller SIMS instruments can yield valuable information on
the chemical composition and stable isotopes in certain minerals, but they are limited in
resolving geochronologically relevant isotopes (e.g., 204Pb, 206Pb, 207Pb, 208Pb) for accessory
minerals where isobaric interferences abound. The CAMECA NanoSIMS instruments that
are specifically designed to utilize small primary ion beams (1 μm) are widely applied in
cosmochemistry and biogeochemistry (e.g., Hoppe et al. 2013). Because the ultra-high spatial
resolution that results from a highly focused primary beam intrinsically comes at a loss of
secondary ion intensity, the role of NanoSIMS in the analysis of trace species important
for petrochronology is limited (Table 1). Where this instrument has been used successfully
for dating purposes (e.g., Stern et al. 2005; Yang et al. 2012), primary beam dimensions are
typically > 1 μm and thus equivalent to those achieved in large magnet radius instruments
(e.g., Harrison and Schmitt 2007). NanoSIMS has occasionally been used to investigate trace
element compositions of accessory minerals (e.g., Hofmann et al. 2009, 2014), and for high-
resolution geospeedometry to derive the durations, rather than absolute dates, over which
diffusion occurred between the compositional zones in major minerals that record rapid
petrologic evolution. When combined with petrologic information from the compositions of
the mineral zoning, these diffusion studies have been used for “petrospeedometry” to resolve
the durations of processes such as silicic magma generation (e.g., Till et al. 2015) and quartz
growth and residence in rhyolitic magma that was eventually erupted (e.g., Seitz et al. 2016).
Large magnet radius instruments for geochronology and petrochronology
CAMECA-family. CAMECA has developed ion microprobes since the late 1960s, with
the first commercially successful small radius magnet instrument, the IMS 3f, essentially
earning the company monopoly status for magnetic sector SIMS in the semi-conductor
industry (de Chambost 2011). Following the initial development of the SHRIMP instrument
(see historical review by Ireland et al. 2008) in the 1970s at Australian National University
(ANU), CAMECA decided to also branch out into geological applications (de Chambost
2011). The idea was to scale up the concurrently marketed IMS 4f to meet the transmission
and mass resolution requirements for Pb isotope analysis of zircon, while at the same time
maintaining the ion imaging capabilities of the IMS instruments that had proven extremely
helpful for tuning and target location purposes. This design is also beneficial in enhancing
lateral resolution, and in turn the resolution of depth profiling by means of inserting a field
aperture into the secondary ion path (Fig. 1). The instrument’s name IMS 1270 was in fact a
misnomer which was based on a 10× scaling of the IMS 4f magnet radius assumed to be 127 mm
(in fact it was only 117 mm), with the realized magnet radius being only 5 × 117 mm = 585 mm
A Electrostatic
analyzer
Cs-ion source
Field Energy slit
aperture Magnet
Duoplasmatron Entrance slit and
Primary beam contrast aperture Multi-
mass filter Normal incidence collection
electron gun
Airlock Axial
Sample detectors
holder
Microchannel Plate
CAMECA IMS 1280-HR and phosphor screen
Energy Electrostatic
Slit Analyzer
B
Magnet
Matsuda Duoplasmatron
quadrupole lens
Divergence
slits
Multiple SHRIMP II Source slit

Collector
Source chamber
100 cm
Primary
apertures
C
Q3
QQH chamber
Electrostatic
Analyzer Magnet
Primary
Q4 column
Q2
QQH monitor
Q1
Collector
Source chamber
SHRIMP RG
Figure 1. A) Schematic diagrams of large magnet radius SIMS instruments. A) CAMECA IMS 1280
with multi-collector. Double-arrows and black boxes indicate ion optical devices (electrostatic lenses and
deflecting plates, respectively). B) SHRIMP II and C) SHRIMP-RG. Major differences include the magnet-
electrostatic analyzer geometries, and the presence of four quadrupoles (Q1-Q4) in SHRIMP-RG.
(de Chambost 2011). The first instrument was installed at the University of California Los
Angeles (UCLA) in December 1992. A multi-collector detection system became available
in 1997. Following an upgrade of the instrument electronics, the IMS 1270 became the IMS
1280, the first of which was installed at University of Wisconsin in 2005, but the fundamental
instrumental layout (Fig. 1) has experienced only incremental changes and remains largely
constant to the present. There are currently about 30 instruments operating worldwide.
The IMS1270/1280 instruments feature a split primary column where both duoplasmatron
gas and Cs+ surface ionization primary ion sources can be held simultaneously under
vacuum (Fig. 1). The primary beam mass filter (PBMF; Fig. 1) generates an isotopically
pure primary ion beam. This is critical for certain uses, such as for avoiding the implantation
of OH− together with O− into the target which would generate difficult-to-resolve hydride
interferences (see below). The PBMF also acts as a shunt magnet which allows for
sequential use of both sources without breaking vacuum. Latest generation instruments can
continuously monitor primary beam intensity through rapid beam switching.
Samples are introduced via an airlock into an antechamber which can hold multiple
samples for rapid exchange. They are pumped to vacuum (~10−5 to 10−7 Pa) and then transferred
into the sample chamber where the surface can be observed in reflected light through an
off-axis microscope objective with illumination through an oblique vacuum-external light
source. A unique design is the presence of a gas inlet through the extraction plate opposite
the sample which can locally enhance the concentration of gas species in the sample chamber
(“oxygen flooding”). Oxygen flooding enhances ionization of Pb+ in some minerals (e.g.,
zircon, baddeleyite, rutile), but not others (e.g., monazite; Schuhmacher et al. 1994; Schmitt
et al. 2010; Schmitt and Zack 2012). Mount holders can accommodate samples up to 25.4 mm
in diameter (note: the useful area is smaller; see below) and a height of preferably < 7 mm;
smaller samples can be mounted in holders with smaller (e.g., 5 mm) cutouts in the top plate,
but the limited useful surface area renders this type of mounting less preferable. To compensate
for electron loss from the sputter area when the sample is at negative potential (analyzing
negative ions, e.g., 18O−/16O−), a normal incidence electron gun is used to prevent charging of
insulators (Fig. 1). Positive or negative secondary ions are extracted in an electrostatic field
between the biased sample surface and the grounded extraction plate at a potential gradient of
up to 2 V/μm. The total secondary beam path is ~5 m resulting in a time-of-flight of 10’s of μs.
The secondary (‘Transfer’) ion optical system permits adjustment of the secondary ion
trajectory relative to the entrance slit and the co-located, and generally wide open contrast
aperture, as well as the pre-electrostatic analyzer (ESA) field aperture. Tuning of the transfer
system achieves the required transmission and limits spherical (angular) aberrations that
could compromise the focusing capabilities of the mass spectrometer (Fig. 1). Moreover, the
field aperture plane contains a magnified image of the ions emitted from the sample surface.
The field aperture width can be continuously modified to achieve selection of a spatial subset
of secondary ions. For analyses where surface contamination is of concern (as is the case
for trace Pb), it can be adjusted for transmission of ions from the center of the analysis
crater, thus blocking most ions emitted from the more slowly sputtered edges of the crater.
While clipping of the beam in the field aperture reduces all secondary ion intensities, surface
contaminants from the edges are preferentially excluded, thus enhancing depth resolution.
The mass spectrometer section of the instrument has a double-focusing geometry with an
ESA followed by a magnetic prism (Fig. 1). Mass resolution can be enhanced by electrostatically
modifying the secondary ion beam in a direction perpendicular to the mass focal plane in the
so-called XY mode (Z being the direction of secondary ion travel in CAMECA coordinates);
this is commonly done for U–Pb analysis. The energy slit between the ESA and the magnet
(Fig. 1) provides a band-pass for secondary ion energies. The low-energy slit is set to a
position that permits sampling of all ions below the nominal accelerating potential, whereas
the high-energy slit is independently moved to typically admit ions within a few 10’s of eV of
the maximum. Analysis types where interferences are difficult to resolve benefit from sampling
only the high-energy population of secondary ions. This “energy filtering” mode of operation
takes advantage of the narrow energy distribution of most molecular ions (hydrides being a
notable exception), whose intensities are significantly reduced at high (several 10’s to 100’s
of eV) energy offsets (e.g., Shimizu 1978). Energy filtering reduces the relative contribution
of molecular interferences on atomic ion species, for example that of light rare earth element
(REE) oxides on heavy REE atomic ions (Zinner and Crozaz 1986), and mitigates matrix
effects (e.g., Shimizu and Hart 1982).
Secondary ions are detected either by Faraday cup (FC) or electron multiplier detectors
(EM). Single collection is achieved by electrostatically bending the secondary ion beam
coming through the exit slit in a small ESA and steering it into the so-called axial EM, or one
of two bracketing FCs (Fig. 1). Multi-collection instruments additionally have an array of five
moveable detector carriers (“trolleys”), which are equipped with miniaturized EMs or FCs. On
the current model of the IMS 1280-HR multi-collection is possible at 1 u dispersion from mass
7 to 240. Pre-amplifier circuits for FCs contain resistors between 109 and 1011 Ω; maximum
currents (at 1010 Ω) are 6 × 109 counts per second (cps). Thermal drift at ~4 × 103 cps/°C is
mitigated by housing of the resistor circuit in a thermally stable environment, but count rates
<106 cps are difficult to measure by these FCs because of Johnson-Nyquist noise. At low
secondary ion intensities (<106 cps) electron multipliers in pulse-counting mode are currently
used because they offer extremely low noise levels (< 0.01 cps). One drawback of the EM
is dead time which requires significant corrections at high count rates required for accurate
analysis (i.e., 1‰ correction per nsec dead time at 106 cps). Moreover, aging of the EM
detectors requires regular monitoring of the pulse voltage distribution to maintain sensitivity,
and because the EM yield deteriorates with total accumulated counts, frequent cross-
calibration between different EMs or EM and FC detectors is required if multiple detectors
are used. Lastly, a terminal microchannel plate detector on the axis of the spectrometer allows
for direct ion imaging of a mass-filtered secondary ion beam (Fig. 1). Direct ion imaging
is not as sensitive as scanning ion imaging with EMs and at best semi-quantitative due to
detector non-linearity and heterogeneous aging. Nevertheless, many practitioners value it
as an indispensable aid for tuning, rapid beam location, and characterization of analytical
targets. Direct ion imaging is a unique feature of all IMS (= ion microscope) instruments.
Nevertheless, scanning ion imaging (analogous to scanning electron microscopy) where the
primary beam is rastered over the analyte area and secondary ion signals are detected in fast-
response EM detectors is the preferred method to create ion images and derivative spatially-
correlated geochemical and isotope data on these instruments.
SHRIMP-family. The Sensitive High Resolution Ion Microprobes (SHRIMP)
manufactured by Australian Scientific Instruments (ASI) are employed for geochronology,
isotopic, and trace element analyses of geologic materials. These large-format, double-
focusing mass spectrometers are descendants of the original SHRIMP (subsequently
named SHRIMP I) instrument developed by William Compston and Steve Clement in the
late 1970’s at the Research School of Earth Sciences, ANU. SHRIMP I was developed
to perform isotopic analyses with high sensitivity and at high mass resolution sufficient
to resolve isobaric interferences when analyzing geologic samples. Many of the design
features of SHRIMP I are employed in newer models of the SHRIMP family. SHRIMP I
was designed based on the Matsuda (1974) ion optic configuration for a double-focusing
mass spectrometer. To optimize transmission at high (m/Δm > 5000) mass resolution,
SHRIMP I was designed with an astigmatic secondary ion extraction system, a 1270 mm
radius electrostatic analyzer, and a 1000 mm turning radius magnet, all of which led to an
instrument with an ~7 m beam path. These basic elements have been used in subsequent
SHRIMP models. Consequences of the astigmatic nature of the secondary ion optics are
that SHRIMP instruments cannot be used as an ion microscopes like the CAMECA IMS
family of ion probes, and that the depth profiling capability is limited because secondary
ions from the bottom and the walls of the sputter pit are convoluted. After about three
decades of service, SHRIMP I was decommissioned in late 2010.
In the early 1990’s, SHRIMP II was introduced based on the successful design elements of
SHRIMP I, and became commercially available through ASI. SHRIMP II featured a redesigned
source (= sample) chamber, primary column and secondary ion optics system that reduced
aberrations and doubled sensitivity. The redesigned primary column incorporated a lens and
aperture system for Köhler illumination to produce flat-bottomed and steep-sided analysis spots,
thereby reducing spherical aberrations and contamination from the non-sputtered surface of
the target. The incorporation of a Schwarzschild microscope system, which allows coincident
reflected light imaging of samples and alignment of the primary-secondary ion optics, has
allowed user-friendly operation and precise sampling of target domains. The sample stage
in the source chamber can hold two mounts for analysis; one position is typically reserved
for a setup mount that contains a suite of concentration and secondary reference materials. A
separate sample lock houses a motorized rack that holds up to four mounts under high vacuum
in preparation for exchanges into the sample stage. Mount exchange is automated using a
motorized hoist and pneumatic clamp system. Originally, SHRIMP II instruments were fitted
with a single-collector system, but newer versions have featured a multi-collection system with
ion counters and FCs (Holden et al. 2009). For most geochronology (e.g., U–Pb, U–Th) and
trace element (e.g., REE, Ti) applications, a primary beam of negative oxygen ions that are
generated in a cold-cathode duoplasmatron is used to bombard a sample and generate positive
secondary ions. A Wien mass filter is used to select preferred primary beam species (e.g., O− vs
O2−) and exclude species, such as OH− and NO2− that may generate interferences. For trace
elements that more readily yield negative secondary ions during sputtering, such as halogens,
the duoplasmatron is physically replaced by a cesium source to yield Cs+ ions, with charge
compensation on the sample surface provided by an electron gun. Helmholtz coils surround
the source chamber and are used to moderate light isotope fractionation by canceling out the
vertical component of the geomagnetic field.
After the introduction of SHRIMP II, a variant (SHRIMP-RG) was designed based on the
reversely configured mass spectrometer of Matsuda (1990), in order to reach ultra-high mass
resolution for analyses of intermediate-mass elements, such as the REE, that are plagued by
interferences during SIMS analysis. Only two SHRIMP-RG instruments have been manufactured;
one is located at ANU, and the other is at Stanford University where it is cooperatively operated
with the U.S. Geological Survey. The reverse geometry of SHRIMP-RG, which employs the
magnet before the ESA (Fig. 1), allows a mass resolution of up to four times higher than for other
SHRIMPs (Compston and Clement 2006; Ireland et al. 2008). Consequently, the REE and other
trace elements in accessory minerals can be resolved without the need to employ aggressive
energy filtering that leads to a reduction of secondary ions transmitted to the collector, or where
substantial reduction in molecular intensities cannot be achieved at higher energies, as is the
case for hydrides of intermediate to high atomic number elements (Fig. 2). Energy filtering
can be employed as needed using an energy selection slit located immediately before the ESA,
and abundance sensitivity can be enhanced with a retardation lens located before the collector.
An off-axis FC detector is available and can be employed in conjunction with an on-axis EM
for measuring trace elements that yield count rates varying by several orders of magnitude, for
example 139La+ and 155Gd+ in allanite or monazite. This approach has been employed by the
Stanford-USGS SHRIMP-RG to perform simultaneous U–Th–Pb geochronology and analysis
of REE in light REE-rich minerals. However, a key limitation of the SHRIMP-RG design is that
multicollection of isotopes is not possible.
104 93
Nb+
92
103 ZrH+
Intensity (cps)
102
101
100 SHRIMP-RG
MAD zircon
m/m = 14000
10-1
92.900 92.905 92.910 92.915 92.920
Mass/charge (u)
Figure 2. High mass resolution scan of MAD zircon on mass/charge 93 u (93Nb+) at a mass resolving power
m/Δm = 14000. Scan shows the 92ZrH+ interference resolved on the the high-mass side of the Nb-peak. Data
generated on the Stanford-USGS SHRIMP-RG, an instrument designed to achieve high mass resolution
with minimal loss in transmission.
Recently, a variant (SHRIMP-SI, aka SHRIMP V) has been specifically designed for
precise multi-collector measurement of stable isotopes including oxygen (Ireland et al. 2008;
Ireland et al. 2014). SHRIMP-SI has been fitted with a new electrometer system for FCs
with 1010–12 Ω resistors and a capacitor charge mode capability optimized for count rates
(105–106 cps) over the upper and lower effective performance limits of EM and FC detectors,
respectively (Ireland et al. 2014).
Sample preparation for petrochronology
Because the sample surface is part of the ion extraction geometry in all SIMS instruments,
stringent requirements exist for samples to be vacuum-proof, flat, and conductive. Sample
preparation is thus a critical step for successful SIMS analysis. Furthermore, extensive sample
characterization prior to SIMS analysis (e.g., by optical and/or electron beam imaging) and
documentation is essential for efficient targeting. Key aspects of common preparation and
imaging procedures used for petrochronological SIMS analysis are summarized here.
Grain mounts. SIMS analysis of geologic materials traditionally employs grain mounts
with isolated crystals embedded into commercially available epoxy (Fig. 3). Grain mounts are
typically composed of one mineral type to maximize the number of analyzed grains per sample
mount, minimize the time spent changing mounts, and facilitate automated analysis, all of which
are often important factors for research productivity. Mineral grains, either as whole crystals or
fragments, are typically isolated via conventional mineral separation involving rock crushing
and pulverizing, heavy liquid density separation, and magnetic separation. Individual mineral
grains are then segregated and/or are hand-picked in preparation for mounting. As is performed
before most ID-TIMS analyses, individual zircon crystals can be treated to “chemical abrasion”
prior to mounting in order to remove domains affected by cryptic Pb-loss (Watts et al. 2016).
Grain mounts are typically prepared by placing individual or aggregations of grains

onto double-sided tape. Individual grains or groups are commonly arranged into rows or
other regular patterns in order to provide organization and later assist navigation using the
ion microprobe’s reflected light microscope (Fig. 3). It is important that mounts also include
reference materials needed for the intended geochronology, trace element or stable isotope
calibration. The CAMECA and SHRIMP ion microprobes have steel mount holders that accept
25.4 mm diameter round mounts. Because it has been demonstrated that instrumental mass
fractionation can be exacerbated for analyses at the margins of the mounts near the steel-epoxy
interface (e.g., Treble et al. 2007; Ickert et al. 2008), grains should be placed within the inner
15 mm diameter of the mount, away from the mount edge. After organization, the tape-mounted
grains are surrounded with a 25.4 mm inner diameter plastic or teflon ring which is filled with a
well-stirred mixture of resin and hardener. Choice of the epoxy is critical: the epoxy must meet
quality criteria for categories such as minimal outgassing under vacuum, hardness and good
bonding properties, resilience under electron beam bombardment, and preferably having low
cathodoluminescence backgrounds. The University of Edinburgh Ion Microprobe facility has
provided an excellent survey of the characteristics of numerous varieties of epoxy, available
online (http://www.ed.ac.uk/geosciences/research/facilities/ionprobe/technical/epoxyresins).
It is also important that the epoxy is free of bubbles, and that bubbles do not form on the sides
of crystals, flaws that may result when the epoxy is overly viscous or poured too quickly.
Sectioned bubbles can result in relief near or adjacent to grains, which can deform the local
electrostatic field and affect secondary ion trajectories, possibly resulting in significant isotope
fractionation and reduced precision (Kita et al. 2009). Curing the epoxy under pressure can
mitigate bubble formation. After hardening, the epoxy mount may be milled to an appropriate
overall thickness (~5 mm), and the side exposing the sample is ground with abrasives (e.g.,
diamond) to expose grain interiors at the required depth. Finally, the mount is polished to
generate a flat surface across the entire mount with minimal relief.
Pre-analysis microscopic (transmitted and reflected light) and scanning electron
microscope (SEM, using cathodoluminescence CL and/or back-scattered electrons BSE)
imaging is essential because it provides a guide for navigation and targeting during analytical
sessions and later provides a spatial context for organization, evaluation, and interpretation of
combined geochronologic and trace element data. Electronic mount images may be uploaded
and registered to stage coordinates for point-and-click recording of spot locations that are later
referenced for automated analyses. For electron beam imaging, a conductive carbon (C) coating
is usually applied. The C coating is then typically removed by gentle re-polishing and replaced
prior to SIMS analysis by a gold (Au) coating. The Au coating has the advantages of being
more conductive, monoisotopic, and rapidly removed due to its faster sputtering compared to C.
A Zircon
B C Indium
Epoxy Unknowns
10 µm
Zircon
Reference
100 µm 1000 µm 100 µm

Figure 3. Examples for different mounting techniques. A) grain mount with zircon (imaged in cathodolu-
minescence CL) embedded in epoxy. B) optical image of a SIMS mount with diamond core petrographic
thin-section disks mounted together with zircon reference grains prepared as a sectioned epoxy mount. In-
set shows CL image of a small zircon from within the rock. C) zircon crystals pressed into indium showing
unsectioned grains with their pristine crystal surfaces (back-scattered electron (BSE) image).
In-situ mounts. In-situ analysis using polished rock sections is a viable alternative
to conventional procedures that aim at liberating accessory minerals from the rock and pre-
concentrating them in a heavy mineral separate. The advantages of in-situ dating relative to
mineral separation are partially offset by the requirement of preparing a larger number of mounts
because of the typically low abundance of accessory grains exposed on the surface of a rock
section. Consequently, the need for frequent sample changes and the time required to locate
targets, especially if they are small, reduces analytical throughput when in-situ mounts are used.
Standard rectangle petrographic thin-sections where datable accessory minerals have
been located and identified must be trimmed to dimensions that can fit ion microprobe holders
(e.g., Rayner and Stern 2002). Even 25.4 mm round petrographic sections are best reformatted
to permit co-mounting of reference materials. This is advantageous for throughput and data
quality because swapping between mounts during the analysis is time-consuming and exact
analytical conditions may not always be replicated after sample exchange. Trimming can be
performed by cutting sections into smaller pieces containing the region of interest using a
low-speed diamond saw, or by coring rounds using a diamond drill bit. The cut-outs are placed
face-down on adhesive tape together with blocks of pre-sectioned and pre-polished reference
material grains (Fig. 3). Subsequent preparation steps are identical to those described in the
previous section, with or without the grinding and polishing steps.
Surface analysis and depth profiling mounts. Secondary ion emission in SIMS occurs
from the top few nm of the target, making it ideally suited for the analysis of thin surface layers
(e.g., Hunter 2009). SIMS is widely used to detect dopants in synthetic electronics materials
via depth profiling where continuous sputtering produces a time-resolved signal that can be
related to depth if the sputtering rate is known (Hunter 2009). Instrumental parameters that
determine SIMS depth resolution include impact energy, beam density distribution, primary
ion species, sample flatness, and secondary ion tuning. Especially the field aperture setting
in CAMECA ion microprobes is critical as it serves to preferentially transmit secondary ions
from the center of the crater and mitigate contribution of ions emitted from the more slowly
sputtered crater edges (Hunter 2009). SIMS depth profiling of synthetic or experimentally
treated natural samples has been successfully applied to quantify diffusion properties of
geologic materials (e.g., Cherniak et al. 2004). In such samples, it is generally easier to control
the geometric parameters (i.e., orientation of chemical/isotopic layers) that enhance depth
resolution, whereas accessory minerals that are targets for petrochronology often lack these
ideal properties. Nonetheless, SIMS surface analysis and depth profiling has potential to
identify age or chemical variations at a scale that is inaccessible by other techniques. Crystals
that have well developed flat crystal faces such as the (100) or (110) prisms in zircon lend
themselves to depth profiling, provided that internal growth layers are oriented parallel to the
surface (e.g., Breeding et al. 2004; Vorhies et al. 2013).
The shallow depth resolution characteristics of SIMS analyses thus can be leveraged to
provide ultra-high spatial resolution sampling of these crystal faces, i.e., last crystallization layers
of crystals that grew in a concentric fashion. The last interval of crystal growth by accessory
minerals, often recording the final magmatic or metamorphic history, may archive important
time-compositional information for petrochronology. In the case of accessory minerals like
zircon, the outermost compositional zones may be only nano- to several micrometers in thickness,
and thus beyond the spatial resolution of most other approaches for micro-geochronology and
trace element quantification. When sampling crystal faces, lateral resolution is less critical
because growth zones are often parallel to the plane of sputtering. Special techniques for depth
profiling sample preparation where crystal faces are arranged parallel to the mount surface
involve pressing crystals into soft and malleable metal such as indium (In; Fig. 3). Although the
bonding of In is weaker than with epoxy, sectioning and polishing In-mounted crystals after
depth profiling is possible and permits subsequent analysis of the crystal interiors. Alternatively,
crystals can be extracted from the In mount, and recast into epoxy.
QUANTITATIVE SIMS ANALYSIS

Relative sensitivity and instrumental mass fractionation factors
The two major aspects of quantitative SIMS analysis are the interpretation of mass spectra
that exist in the mass range of interest, and the quantification of secondary ion intensities of
the species of concern with the goal to obtain accurate abundance information. Spectrum
interpretation is thus a prerequisite to reliably extract the SIMS signal of interest, and isolate
it from potentially overlapping signals such as interferences or backgrounds (e.g., abundance
sensitivity). This is often not a trivial task because mass spectra in SIMS are intrinsically
complicated, and comprise isobaric interferences composed of atoms from the target material,
primary ion beam species, conductive coating, vacuum residua, and other potential contaminants
(polishing materials, embedding medium, implanted atoms from previous SIMS analysis, etc.).
Hydrides are particularly critical because they require high mass resolution (e.g., m/Δm = 14000
for 92ZrH+ on 93Nb+; Fig. 2), and are difficult to eliminate by energy-filtering. Achieving
low H backgrounds, where critical, requires organics-free mounting media and the use of a
liquid-N2 cold trap or helium cryopump to pump hydrogen gas efficiently. Singly-charged
ions dominate the mass spectrum, but dual positively charged ions (e.g., 90Zr++ on 45Sc+) can
cause interferences with an analysis if unmitigated, e.g., through high mass resolution (in the
case of Sc at m/Δm = 13000). EM backgrounds are typically negligibly low, but corrections
for abundance sensitivity (the tailing of adjacent high-intensity peaks at a distance of 1 u; it is
~250 ppb for large magnet radius SIMS instruments) can be essential for ultra-trace analysis
such as 230Th or 231Pa. Abundance sensitivity is often monitored adjacent to the peak of interest
using an interference-free mass station, sometimes at the opposite mass side relative to the
interfering peak, or a nearby mass with interference patterns similar to the peak of interest.
Secondary ion signals must then be accurately quantified as isotope ratios or concentrations.
The fundamental equation that underlies quantitative SIMS analysis (Eqn. 1):
I x= I p ⋅ Ys ⋅ α ±x ⋅ fx ⋅ cx (1)
relates Ix = current of ion species x in terms of Ip = primary ion beam current, Ys = the sputter
yield (atoms removed per incoming primary ion), αx = ionization probability of species x (as
positive or secondary ions), fx = fraction of x ions transmitted and detected, and cx = concentration
of x, with concentration being the quantity of interest. Sputter yield, ionization probability,
transmission and detection efficiency can be combined into an element-specific useful yield
(which is the ratio between ions of x detected divided by atoms x removed). For positive atomic
secondary ions at low energies useful yields range between 10−3 and 10−1 for most elements
(Hinton 1990; Hervig et al. 2006). For similar matrices (e.g., silicates) and analytical conditions
(e.g., primary ion species, secondary ion energy), useful yields or absolute sensitivities follow
basic chemical trends, and tabulated data can be used to make first-order predictions about the
expected intensities of an ion species for a given concentration (Hinton 1990; Hervig et al. 2006).
Quantification of element abundances or relative abundances for analytical purposes
require empirical corrections because the sputtering and instrument related unknowns
in Equation (1) cannot be deduced theoretically, and they are variable with the material
analyzed. These corrections are based on a comparison of unknowns with natural or synthetic
measurement standards analyzed under reproducible conditions. Secondary ion intensities for
the species of interest x are corrected for detector gain, dead time (in the case of EM detectors),
and background (in particular for FC detectors which have much higher noise levels compared
to EMs), and divided by the intensity of a normalizing species m corrected in the same manner.
For a reference material, concentrations of x and m (or their ratio cx / cm) are known, and from
the ratio “measured over true” a relative sensitivity factor (RSF) can be determined. This RSF
combines all unknown parameters in Equation (1). The primary beam intensity cancels, provided
it remains constant, or is corrected for drift if masses are analyzed sequentially (Ludwig 2009):
Ix I p ⋅ Ys ⋅ α ±x ⋅ fx ⋅ cx c
= = RSF ⋅ x (2)
I m I p ⋅ Ys ⋅ α m± ⋅ fm ⋅ cm cm
The RSF represents the correction factor that needs to be applied to the measured quantity
on the sample to obtain the “true” abundance (e.g., U ppm for trace element analysis) or
ratio (e.g., 206Pb/238U for geochronology). Provided the RSF is identical for unknowns and
references, the following relation holds (Eqn. 3 as an example for the quantification of U
abundance in zircon):
I Uunknown cZr
unknown
I reference c reference (3)
RSF = unknown
⋅ unknown = Ureference ⋅ Zr
I Zr cU I Zr cUreference
where IZr is the intensity of a zirconium (Zr) species of interest (e.g., 94Zr2O+ at mass/charge =
203.8076). To estimate concentrations (e.g., U ppm), cZr needs to be quantified for the reference
zircon and the unknown, unless it can be assumed that they are identical and cancel each other.
This is—to a first order—the case for stoichiometric Zr in zircon. The U concentration (in the
chosen units, e.g., in ppm) then simplifies to (Eqn. 4):
I Uunknown 1 (4)
c=
unknown
U ⋅
I Zrunknown RSF
with the RSF determined on a reference zircon (e.g., 91500 with ~80 ppm U; Wiedenbeck et
al. 2004). If a glass (e.g., NIST SRM 610) is used to quantify U in zircon and a Si ion species
is the normalizing species, then different SiO2 abundances in unknown zircon and reference
glass need to be carried through for quantification (e.g., SiO2 = 70 wt% for NIST SRM 610
and 32 wt% for stoichiometric zircon). Multiple measurements of reference materials with
different abundances of the trace element of interest can be used to establish a working
curve, which helps to validate the linearity of the method and account for variations in the
reliability of reference materials (e.g., for Ti-in-quartz; Fig. 4).
Before moving on to more complex calibrations, it should be mentioned that externally
calibrated stable isotope ratios are quantified in an analogous way, whereby an instrumental
mass fractionation factor (IMF) is determined on a reference, and then applied to the
unknown, as expressed in Equation (5) and (6) for the example of oxygen isotopic ratios
(where Rtrue is the “true” abundance ratio of 18O/16O and Rmeas the measured ratio of the
secondary ion intensities for 18O− and 16O−):
=IMF R=
reference
meas
reference
/ Rtrue unknown
Rmeas unknown
/ Rtrue (5)
1 (6)
R=
unknown
true
unknown
Rmeas ⋅
IMF
Two- and three-dimensional relative sensitivity calibrations
It has long been recognized that Pb/U and Pb/Th RSF values in SIMS can vary strongly,
even when analytical conditions are kept as constant as possible. The reasons for this behavior
remain poorly understood, but covariations in Pb/U and Pb/Th RSF have been observed with U
and Th oxide formation from which a RSF tailored to a given XOm / XOn of the unknown can be
obtained (XOm / XOn, X = U or Th; m = 1–2; n = 0–1; m ≠ n; Hinthorne et al. 1979, and reviews in
0.010
0.008 d
te
ola
p
tra
Ti+/30SiO2+
0.006 ex
QTIP39
linear regression
-5
RSF = 1.11×10
0.004
48
(QTIP 39 excluded
QTIP38
QTIP14
QTIP7 from regression)
0.002
Session 1
Session 2
0.000
0 200 400 600 800 1000
Ti-in-quartz (ppm)
Figure 4. Example for a working curve for SIMS trace element analysis with the useful range shown as
solid line and extrapolation (not used for quantification) as dashed line. Synthetic Ti-in-quartz reference
materials (“QTIP”) from Thomas et al. 2010). Secondary ion intensities for 48Ti+ and 30SiO2+ were counted
simultaneously using dual electron multiplier detection with a detector slit width of 150 µm (correspond-
ing to m/Δm = 8000). The relative sensitivity factors (RSF) for two analytical sessions (May and August
2013) are indistinguishable. Note that the QTIP 39 reference with the highest Ti is heterogeneous, and
therefore omitted from the calibration. Calibration data from Shulaker et al. (2015) and collected with
multi-collection on a CAMECA IMS 1270.
Williams and McKibben 1998; Ireland and Williams 2003; Jeon and Whitehouse 2015). These
empirically determined functions (Fig. 5) can have the form of a power law:
b
Pb  XO m  (7)
= a 
XO n  XO n 
or, if spread is minor, reasonably be approximated by a linear fit:
Pb  XO m 
= a +b (8)
XO n  XO n 
Such two-dimensional (2-D, because they depend on two measured indicators) RSF
calibrations have been developed for different accessory minerals (e.g., allanite, apatite,
baddeleyite, monazite, rutile, titanite, xenotime, zircon), and also for calculating U
abundances (e.g., Williams and McKibben 1998). The optimal choice of XOm / XOn will
depend on the matrix and instrument parameters (e.g., O− and O2− primary ion beam, or
use of O2-flooding), but differences in the resulting calibration uncertainties are generally
minor. A long-term intercomparison of different calibration schemes for the same reference
material (91500 zircon) shows that RSF values scattered within 0.79 and 1.1% relative
uncertainty (Jeon and Whitehouse 2015), and monitoring this variability is a fundamental
requirement for precise SIMS geochronology.
In some instances, three-dimensional calibrations (3-D) have been proposed, e.g., for Th–Pb
geochronology of allanite where matrix effects could be monitored by also including FeO+ / SiO+ in
the calibration (Catlos et al. 2000). Other studies have subsequently established that conventional
2-D calibrations for allanite yielded sufficiently accurate results (Gregory et al. 2007).
10.0
A
206 * 238
Pb / U age (Ma) ZIPS 3.0.4.
9.5 10
Pb/U RSF
9.0 5
8.5 540 560 580

0
8.0
7.5 UO/U
9.50 9.75 10.00 10.25 10.50
2.4 206
Pb*/238U age (Ma)
B SQUID 2
ln Pb/U RSF
2.3 10
5
2.2
0
540 560 580
2.1
ln UO/U
2.0
2.250 2.275 2.300 2.325 2.350 2.375
Figure 5. Example for U–Pb zircon calibration plots for zircon reference z6266 with an age of 559 Ma (Stern
and Amelin 2003). Panels A and B show the same data, with A using a linear calibration and B a power law
calibration. Data reduction was performed using ZIPS 3.0.4. (created by C. Coath), and SQUID 2 (K. Lud-
wig). Significant differences in the age and uncertainty calculations are absent, demonstrating the compara-
bility of data reduced with different calibrations and software. Data collected using a CAMECA IMS 1270.
Another variation of a relative sensitivity calibration is used to determine accurate Th/U

in accessory minerals. This is important as Th/U is often used as a first-order indicator for
magmatic vs. metamorphic origins (where magmatic Th/U in zircon is ~0.3–0.7, whereas
many metamorphic zircons have Th/U < 0.1; Rubatto 2002; Hoskin and Schaltegger 2003), and
to implement accurate disequilibrium corrections (see below). Although Th/U RSF values are
much closer to unity than those for Pb/U, quantification is hampered by Th and U heterogeneities
in many reference zircons (with the notable exception of 91500 zircon). An elegant solution is
a calibration, where “true” Th/U is proxied by 208Pb*/206Pb* which shows insignificant isotopic
fractionation in SIMS. Assuming concordancy between Th–Pb and U–Pb decay systems, Th/U
can be derived from measured 208Pb*/206Pb* (Hinthorne et al. 1979; Reid et al. 1997).
In all these cases, potential bias can arise if reference materials used for calibration
behave differently during primary ion bombardment than the unknowns. Matrix effects are
biases that can result from differences in chemical composition, which for zircon can be high
(> 2500 ppm) U, although White and Ireland (2012) have proposed that bias in the analysis
of high-U zircons may reflect incipient metamictization rather than a chemical matrix effect.
Crystallographic orientation can also influence relative sensitivities (e.g., baddeleyite, Wingate
and Compston 2000) and instrumental mass fractionation (e.g., oxygen isotopes in magnetite
and rutile; Huberty et al. 2010; Shulaker et al. 2015). Adapting analytical conditions can
mitigate both effects. Dispersion of U–Pb was reduced for randomly oriented baddeleyite and
rutile crystals when O2-flooding was applied (Li et al. 2010; Schmitt et al. 2010; Schmitt and
Zack 2012), and oxygen isotope data showed less spread if low energy primary ions were used
for magnetite and rutile (Huberty et al. 2010; Shulaker et al. 2015).
Specific analytical consideration and strategies for petrochronology
Geochronology. Coupling geochronology with textural and petrologic information requires
adhering to an analytical strategy to optimize data quality and efficiency. Preparation for
geochronology considers the availability of datable minerals in a rock, mineral compositions,

and the range of expected dates. This also includes the availability of reference minerals for
calibration. Zircon is probably the most-straightforward mineral for geochronology using
SIMS because it is widespread in crustal rocks and has been employed as a geochronometer for
decades (e.g., Rubatto 2017, this volume). Multiple reference materials exist for calibration of
Pb/U ratios and trace elements in zircon. Most silicic plutonic rocks and many of their volcanic
equivalents contain zircon, either as an early or late microphenocryst or as inclusions in major
phases. Mafic intrusions often reach zircon saturation as they cool to their solidi. Metamorphic
rocks typically contain zircon grains as crystals found in the matrix as well as in porphyroblasts.
Other accessory minerals tend to be more restricted in their occurrences, for example baddeleyite
is found mostly in mafic rocks, chevkinite in alkalic rocks, and monazite found predominantly in
metamorphic rocks. Some accessory minerals may be preferable to others for petrochronology
depending on compositions and project goals. For example, the complex composition of allanite
facilitates the combination of geochronology with elemental concentrations that are reflective of
the major-minor element evolution of the parent from which it grew, such as changes in Mn and
Mg (e.g., Vazquez and Reid 2004), whereas zircon generally provides information about trace
element evolution (e.g., REE, Hf) of its parent. In addition, accessory minerals differ in their
closure behavior, and thus can be harnessed as thermochronometers.
The range of expected dates and mineral compositions, e.g., Tertiary monazite versus
Proterozoic zircon, will dictate if dates are best derived by 207Pb/206Pb, 206Pb/238U, 207Pb/235U
208
Pb/232Th, etc., methods, or 238U–230Th disequilibrium in the case of young (< 350 ka) accessory
minerals. Because dates using any of these methods are derived from parent/daughter or daughter
isotope ratios, a sufficient concentration of parent isotope is needed, especially when dating young
minerals. High U concentrations have allowed U–Th dating of zircons as young as ca. 2.5 ka
(Schmitt et al. 2013; Wright et al. 2015). However, there can be too much of a good thing: high
U concentrations may be associated with metamictization, which compromises the assumption
of matrix-matching between unknown and the reference material (see above), and which may
be associated with Pb-loss and result in spurious crystallization ages. High U intensities can
also compromise the EM detector and trigger safeties to protect it. High U concentrations and
metamict domains are often recognizable in cathodoluminescence and backscattered electron
images, and thus can be avoided via pre-analysis imaging. In general, instrumental conditions
for U–Pb and U–Th dating are similar, with mass resolutions of m/Δm > 5000 needed to resolve
206
Pb+ from nearby 178Hf28Si+ and 174Hf16O2+ in zircon. In CAMECA instruments, oxygen
flooding is routinely used to enhance secondary ion yields of Pb+ in zircon.
Analogous to the 40Ar/39Ar method, SIMS U–Pb dates are calculated relative to a reference
that yields concordant U–Pb dates, typically independently constrained by conventional isotope
dilution mass spectrometry (ID-TIMS). Several reference zircons are commonly used for U–Pb
dating by SIMS, including Temora (418 Ma), R33 (419 Ma), SL13 (572 Ma), z6266 (559 Ma),
91500 (1065 Ma), AS3, and FC1 (both 1099 Ma). For some of these reference materials
an approximately 1 million year range in the 206Pb/238U dates has been reported by different
laboratories, e.g., for zircon R33 (Black et al. 2004; Mattinson 2010; Schaltegger et al. 2015).
Through the use of commonly distributed spikes (e.g., EARTHTIME) for isotope dilution, the
range of dates from different laboratories is likely to be reduced in the future. Minerals with high
Th concentrations, such as monazite and allanite, are attractive for their potential to generate
precise 208Pb/232Th dates. High Th minerals as reference materials for U–Pb and/or Th–Pb dating
include: monazite 44069 (Aleinikoff et al. 2006), monazite 554 (Harrison et al. 1995), titanite
BLR-1 (Aleinikoff et al. 2007), allanite Siss and allanite Bona (von Blanckenburg 1992), allanite
Tara (Gregory et al. 2007), and xenotime MG-1 (Fletcher et al. 2004), although not all of these
minerals have 208Pb/232Th dates that are confirmed by ID-TIMS (see below).
SIMS U–Pb dating also requires corrections for common Pb; the resulting radiogenic
Pb* is used for the age calculation. For this, stable 204Pb is analyzed and used for correction in
combination with appropriate isotopic compositions of common Pb. For zircon, 204Pb intensities
are low, and often decreasing during the analysis, indicating that common Pb is surface
derived. Consequently, anthropogenic common Pb compositions are adequate. Other minerals
(e.g., allanite, apatite, monazite, titanite) have significant intrinsic common Pb, and common
Pb compositions used for correction need to be carefully chosen. Accurate determination of
204
Pb is hampered by unresolvable interferences (e.g., 232Th144Nd16O2++ in monazite). Peak-
stripping (using 232Th143Nd16O2++ at mass 203.5) or some amount of energy filtering and/or use
of a pre-collector retardation lens is needed to reduce isobars that interfere with the Pb isotopes
in monazite (Fletcher et al. 2010). Alternative corrections for common Pb are therefore often
advantageous: for unradiogenic materials, or minerals with low Th (e.g., baddeleyite, rutile), a
208
Pb-based correction (Compston et al. 1984) is a viable alternative to the 204Pb-based common
Pb correction. If 204Pb is difficult to measure and/or the dated material is sufficiently young that
concordancy can be reasonably assumed, the common-Pb uncorrected data can be regressed to
obtain a concordia intercept age (e.g., Baldwin and Ireland 1995). The effects of variable initial
disequilibrium, however, can hamper the accuracy of this approach (e.g., Janots et al. 2012).
Accessory minerals in Quaternary (< 300 ka) rocks are amenable to U–Th dating. Due to
the greater secondary ion yields for actinide oxides, U–Th dating by SIMS typically employs
measurements of UO+ and ThO+ ions at m/Δm = 5000–6000 in order to resolve 230Th16O+
from any REE molecules and other species of the same unit mass (Schmitt 2011). Analyses
for U–Th dating of zircon, allanite, or chevkinite typically require primary beam intensities
that are 5–10 times greater than used for U–Pb dating. Count rates for 230Th during analyses
of zircons are typically 0.1–10 cps for U concentrations of 10–1000 ppm and a 50 nA O−
primary beam (Schmitt 2011). Zircons with low U (< 200 ppm) can be challenging to date by
U–Th or U–Pb methods, and may necessitate relatively large datasets to resolve statistically
meaningful dates (e.g., Coombs and Vazquez 2014). The duration of analyses for 238U–230Th
dating is typically 30–45 minutes, but this can be shortened by ~30% using multi-collection
(e.g., Storm et al. 2011; Vazquez et al. 2014). Quantification of the relative ionization between
U and Th is needed for U–Th dating, and can be accomplished using the observed versus
true 208Pb/206Pb–232Th/238U relation for a reference mineral of known age or by precise
measurement of a mineral in 238U–230Th secular equilibrium (Schmitt 2011).
Trace elements in accessory minerals. The combination of trace element analyses and
geochronology using SIMS has been extensively used to understand the petrologic evolution
of magmas and compositional evolution of metamorphic rocks. The use of zircon for
petrochronology has been popular over the last decade, in particular due to the development
and refinement of the Ti-in-zircon geothermometer (Watson and Harrison 2005; Ferry and
Watson 2007) and advancements in the methodology for high-spatial resolution and high-
precision dating of old and young zircons (e.g., Schaltegger et al. 2015). The combination
of trace element geochemistry, Ti thermometry, stable isotope analysis (see below) and high-
precision U–Th geochronology has been used to resolve the chronology of changing magma
composition and temperature at young volcanoes, including the identification of crystals
recycled by the thermal rejuvenation of stalled intrusions. This has contributed significantly
to a better understanding of realistic crystallization conditions of zircon in magmas and their
longevity (e.g., Claiborne et al. 2010; Storm et al. 2011).
In general, a mass resolution m/Δm > 8000 is required for interference-free analysis of REE
in zircon, including the resolution of 48Ti+ from 96Zr2+. This level of resolution normally requires
some energy filtering and/or closing of slits. Alternatively, high energy filtering (at ~100 eV
offsets) and peak-stripping of independently determined atomic/oxide ion ratios can be applied
for REE analysis at moderate mass resolution m/Δm = 2000. Significantly higher mass resolution
(>10,000) is needed to adequately resolve some other trace elements, such as 93Nb+ from 92ZrH+
(Fig. 2) that have been used for tracing the tectono-magmatic environments of zircon (e.g.,
Grimes et al. 2015). Allowing epoxy mounts to outgas at least overnight under high vacuum
can minimize hydride interferences. Guide peaks are typically needed to autocenter the peaks
of REE species that tend to be found at relatively low concentrations, such as Eu in zircon. In
zircon, the ubiquitous presence of Zr molecules within the LREE spectrum provides a reliable set
of guides (e.g., 91Zr30Si16O2+ for 153Eu+) that can be employed for automated analysis.
Simultaneous measurement of selected trace elements and the isotopes used for
geochronology is feasible (e.g., Abbott et al. 2012). For example, the Stanford-USGS
SHRIMP–RG routinely performs U–Th–Pb dating at mass resolution of m/Δm > 8000, with
simultaneous measurement of REE, Y, and Hf. Most studies, however, have used a two-step
approach to combine dates and trace element concentrations, with trace element measurements
occurring before or after those analyses utilized for dating, usually by placing an analysis
spot within the same compositional domain sampled for geochronology with a smaller
primary beam spot size and at lower intensity as for geochronology. Pre- and post-analysis
imaging is of critical importance to confirm the link between the dating and compositional
information. The correspondence between the domains sampled for trace elements and dating
can be checked by evaluating the concordance of a single trace element, like uranium, that is
measured during both analyses (e.g., Mattinson et al. 2006).
Trace element concentrations determined by SIMS are typically calculated by comparing
the count rates measured in an unknown to those measured in a standard with known
concentrations, typically normalized to a mutually stoichiometric element such as Si or Zr
in the case of zircon. Matrix-dependent ionization of elemental and isotopic compositions
is characteristic of SIMS analyses (e.g., Ireland 2015), and thus unknowns should ideally
be matched to standards in terms of composition and crystal structure. Trace element
concentrations for accessory minerals are generally calibrated to a single standard during an
analytical session, but calibration curves (Fig. 3) that are derived from the relation between
secondary ion yield and known concentrations for a suite of standards may be employed
(e.g., Hiess et al. 2008). Various trace element reference zircons are employed by different
SIMS laboratories, with 91500 (Wiedenbeck et al. 2004), z6266 (Stern and Amelin 2003),
and MAD (Barth and Wooden 2010) as the most commonly used for calibration of REE and
other elements in zircons. These are homogenous gem-quality materials and are derived from
megacrysts. SL13 zircon is noteworthy because its Ti concentration is well characterized
and homogenous (~6.3 ppm; Hiess et al. 2008), which is important for application of the
Ti-in-zircon geothermometer. In contrast, the array of references that are commonly used
for geochronology (e.g., Temora, R33, and AS3 for zircon, and 44069 for monazite) are
mostly heterogeneous with respect to trace elements. References for calibrating SIMS
measurements of trace elements in the other accessory minerals used for geochronology are
less abundant than for zircon, but include xenotime BS-1 (Fletcher et al. 2004; Aleinikoff
et al. 2012), monazite NAM (Aleinikoff et al. 2012), and titanite BLR (Mazdab 2009). For
those accessory minerals with high REE concentrations, domains that have been dated via
SIMS may be characterized using electron microprobe analyses, and their results combined
to evaluate time-compositional evolution (e.g., Vazquez and Reid 2004; Vazquez et al. 2014).
The presence of micro-inclusions may hinder efforts to quantify the trace element
concentrations of accessory minerals; applying fine beams (e.g., on instruments dedicated to
high spatial resolution analysis such as NanoSIMS) may be required in such cases. Accessory
minerals, in particular zircon, are often riddled with micro-inclusions trapped during crystal
growth including other accessory minerals such as apatite, allanite, or oxides, as well as
melt (glass) inclusions. Inadvertent sampling of these inclusions by the primary beam can
dramatically skew trace element concentrations. For example, incorporation of apatite and
glass inclusions, which are especially common in igneous zircons, may result in an REE
pattern that is reversed from the high HREE/LREE pattern that typifies zircon. Cracks and
other imperfections in zircons may contain elevated concentrations of trace elements such
as Ti, which if sampled might skew calculated crystallization temperatures (Harrison and
Schmitt 2007). To identify problematic analysis, it is recommended to always monitor
elements that are indicative of inclusions (e.g., Mg, P, Fe) along with the trace elements of
interest. Reproducibility for Ti in SL13 is < 5% (2σ; Hiess et al. 2008), but contamination
from beam overlap with crystal imperfections can be a major source of bias for the typically
low abundances of Ti in zircon: at ~5 ppm Ti (corresponding to a crystallization temperature of
~680 °C) a 1 ppm excess would increase apparent temperatures by ~15 °C. RSF reproducibility
is thus a minor source of uncertainty for the Ti-in-zircon thermometer compared to surface
contamination and inclusion of non-zircon phases in the analysis volume. Because of the large
sample volume of LA-ICP-MS compared to SIMS, the presence of subsurface inclusions can
be problematic in LA-ICP-MS even when they are not visible at the surface.
Stable and long-lived radiogenic isotopes. Stable isotopes (specifically, O and Li) and
long-lived radiogenic isotopes (Hf- and Nd-isotopes) have revealed important constraints on
the origins of Hadean detrital zircons where rock context is lacking. These isotope systems
are also relevant for the study of igneous and metamorphic zircon, and, in the case of oxygen
isotopes, for accessory monazite and rutile. Because of the comparatively slow sputter rate
of SIMS, secondary ion intensities are typically too low to be measured in FC detectors, and
consequently data lack the precision that is required for a meaningful characterization of
the Hf and Nd isotopic composition (e.g., Kinny et al. 1991), where LA-ICP-MS is utilized
for zircon and monazite, respectively (e.g., McFarlane and McCulloch 2007; Fisher et al.
2014). In developing the optimal analytical strategy, it is, however, important to consider that
SIMS is the most conservative analytical technique with regard to sample consumption, and
that it is generally advantageous to start a sequence of petrochronological analyses with the
least destructive methods, whereby SIMS can be considered as practically non-destructive
when comparing the small amounts of materials sputtered to those removed during
sample preparation using grinding and polishing abrasives. Because of the comparatively
large amounts of material consumed in LA-ICP-MS Hf- and Nd-isotopic analysis (crater
dimensions are several 10’s of μm in width and depth), such data are best collected after all
other required measurements have been completed on the target crystals.
Oxygen isotopes in zircon analysis protocols, including reference materials, have been
established in different labs and using different instrumentation. Of these, oxygen isotopes in
zircon have experienced the widest range of applications, and SIMS has practically become
the method of choice for oxygen isotope (δ18OVSMOW) analysis of zircon (Valley and Kita
2009). If oxygen isotopes cannot be analyzed first, then previously analyzed grains (e.g., for
geochronology) require removal of existing craters or placing spots away from them because
of ion implantation with the commonly used mass-filtered 16O− beam. Multicollection of
16 −
O and 18O− is performed whereby an intermediate detector is sometimes dedicated to 17O−
(ca. 4-times lower in abundance than 18O) when extraterrestrial materials are investigated for
mass-independent fractionation effects which were inherited in the rocky planets from early
solar system oxygen isotopic heterogeneities (e.g., Nemchin et al. 2006). For terrestrial zircon,
simultaneous analysis of mass 17 along with 16 and 18 can also target 16OH− (requiring a mass
resolution m/Δm = 6000 to be resolved from 17O−) which can reveal the presence of water in
nominally anhydrous zircon as an indication of metamictization (Pidgeon et al. 2013). For
high precision analysis, shaped beam (Köhler) illumination or rastering of a focused Cs+ beam
over ~5 to 10 µm is applied to minimize downhole fractionation of isotope ratios. Charge build-
up in the crater is prevented by re-supplying electrons via a normal incidence electron gun
held at equipotential to the sample surface (CAMECA) or an oblique electron gun delivering
a constant current (SHRIMP). Raw counts are corrected for FC backgrounds and detector
yield. Bracketing and intermittent analyses of zircon references are performed to calibrate
instrumental mass fractionation, and to correct isotopic ratios of the unknowns. External
reproducibility of zircon δ18O is ~0.2–0.3‰ (2σ), provided that the sample is flat (Kita et
al. 2009), and targets are located at some distance from the edges of the crater (Treble et al.
2007). New mount designs (see below) have improved reproducibility of isotope analyses over
an area up to ~15 mm in diameter. High spatial resolution analysis of 18O− and 16O− negative
ions employs Cs+ sources at crater dimensions (width and depth) of down to ~1 μm (Page et al.
2007). An ~0.5 μm Ga+ beam emitted from a liquid metal ion source has also been successfully
used in reconnaissance studies (Bindeman et al. 2014). For high spatial resolution analysis,
secondary intensities are too low to measure 18O− in an FC detector. Instead, an EM is used for
detection of 18O− which limits precision to ~1–2‰ (2σ; Page et al. 2007).
Accurate oxygen isotope analysis of monazite is more problematic than of zircon because
of monazite’s strong compositional variability, whereas zircon is essentially stoichiometric with
only minor variability in Hf content causing generally negligible matrix effects. Several studies
have demonstrated a matrix dependent variation in IMF for monazite where compositional
variations of Th (via the huttonite substitution: Th4+ + Si4+ = REE3+ + P5+) and Ca (via the
brabantite substitution Ca2+ + Th4+ = 2REE3+) cause IMF values to vary by several permil,
depending on the instrumentation used, and the overall chemical variation delineated by the
reference monazites analyzed (Breecker and Sharp 2007; Rubatto et al. 2014). Uncertainty in
the bulk oxygen isotope analysis of reference monazite (Breecker and Sharp 2007; Rubatto et
al. 2014) is another source of potential bias. Because of these uncertainties, reliable oxygen
isotope analysis of monazite (e.g., for the purpose of monazite–quartz isotope exchange
thermometry) requires careful consideration of all external and internal sources of error.
Lithium isotope (7Li/6Li expressed as δ7Li LSVEC) analysis is emerging as a potentially
useful indicator of magma sources and the thermal history of zircon (Ushikubo et al. 2008; Trail
et al. 2016). Zircon incorporates Li as an interstitial cation through the coupled substitution
2(REE, Y)3+ = Li+ + P5+ (Bouvier et al. 2012). Zircon in magma-derived continental crust has
higher Li abundances compared to zircon from oceanic crustal plagiogranites (Grimes et
al. 2011). Zircon from Archean felsic plutonic rocks mostly have near-mantle δ7Li isotopic
compositions of +3.8 ± 1.5‰ (from the compilation in Bouvier et al. 2012), whereas Hadean
Jack Hills (Australia) zircon has a significant population with negative δ7Li characteristic
of continentally-derived soils and sediments (Ushikubo et al. 2008; Bouvier et al. 2012).
Li-in-zircon is a potentially useful diffusion thermochronometer, where the preservation of
diffusion profiles can constrain the maximum temperature a crystal has experienced when
timescales of high-temperature crystal storage are known or can be estimated (Rubin et al.
2014; Trail et al. 2016). Conversely, incomplete diffusive equilibration and uncertainties
resulting from difficulties in identifying homogeneous (with regard to Li abundances and
δ7Li) reference zircon crystals hamper the interpretation of Li isotopic data for zircon.
Data reporting for U–Th–Pb geochronology
Shared attributes of SIMS with other U–Th–Pb dating techniques. SIMS is a highly
versatile tool, and can produce diverse data such as trace element and isotope compositions
from spot analyses, depth profiles, as well as 2-D and 3-D images. Here, we concentrate
on general aspects of data reporting in SIMS geochronology. SIMS quantification includes
assumptions and corrections that result from interelement fractionation in different matrices
where true values rely on measurements of reference materials by other techniques (typically
ID-TIMS), and as such, it shares many attributes with LA-ICP-MS (e.g., Horstwood et al. 2016).
As with LA-ICP-MS, uncertainties resulting from the determination of RSFs on reference

materials are typically larger than internal analytical precision. Accurate dating also requires
corrections for common Pb and initial disequilibrium. In combination, these corrections limit
SIMS age uncertainties to 1–2% (relative error). Systematic errors from uranium isotopic
abundances and decay constants (e.g., Boehnke and Harrison 2014) are common to all U–Th–
Pb dating methods, but are frequently negligible for SIMS and LA-ICP-MS ages because of
their comparatively large uncertainties, especially for unpooled data.
Proper calibration. The ratio of interest should always be calibrated on the same ratio
of the reference material. This means that matching ID-TIMS values should be used: e.g., for
206
Pb/238U RSF calibrations on a reference zircon, its reported 206Pb/238U is relevant, and not
a ratio calculated from a 207Pb/206Pb age. Discordance, or corrections for disequilibrium in
individual decay chains (e.g., 230Th in the 238U decay chain) applied to ID-TIMS dates need to be
accounted for if published ages for reference materials used to calculate isotopic ratios. Monazite
is particularly critical because of the scarcity of ID-TIMS 208Pb/232Th ages, and it is potentially
problematic to derive 208Pb/232Th from U–Pb or Pb–Pb ages without independent verification.
Common Pb correction. Common Pb corrections are an additional source of uncertainty
because 204Pb counts are typically low and fraught with high uncertainties. Measurement
uncertainties for the Pb species used for the common Pb correction should be propagated
into the final age uncertainty, but uncertainties stemming from the choice of common Pb
compositions should only be included in the data point uncertainty if they are likely to be
variable, or are derived from a measurement pertinent to individual samples (Schoene 2014).
Disequilibrium corrections. Uranium series disequilibrium is a major source of age
uncertainty for Quaternary zircons (Fig. 6), and together with pre-eruptive zircon residence
can significantly bias cross-calibration of decay constants for 40K- and U-based chronometers
(Simon et al. 2008). This mainly affects the 238U and 235U decay chains whereas disequilibrium
effects in the 232Th-decay chain can be reasonably neglected due to the absence of long-lived
intermediate daughter isotopes. Using 230Th (the longest-lived non-uranium intermediate
daughter isotope in the U-decay chains) as an example, the ingrowth of 206Pb under conditions
of initial disequilibrium is expressed as (Eqn. 9; modified from Wendt and Carl 1985):
206
Pb λ  (230
Th ) 
= (eλ238 t
)
−1 + e λ238 t
⋅ 238 ⋅ 0
(
− 1 ⋅ 1 − eλ230 t ) (9)
238
U λ230 
 ( 238
U)
0


with (230Th)0 / (238U)0 representing the initial (at the time of crystallization) activity ratio.
The deviation from equilibrium can be predicted from the ratio of Th and U partitioning
coefficients, assuming that the melt is in equilibrium (Eqn. 10; Schärer 1984):
(= Th ) ( )
melt
230 230
mineral-melt
DTh Th
=f 0
⋅ 0 (10)
( U)
238
0
DUmineral-melt ( 238
U)
melt
D-values can be determined from 232Th and 238U concentrations measured in the mineral and
glass, or if unavailable whole-rock. Alternatively, model D values (e.g., Blundy and Wood
2003) can be applied. Because 208Pb/232Th ages are not measurably affected by disequilibrium,
they are preferred when DU/DTh is poorly constrained. Janots et al. (2012) have inverted
the problem and used the difference between monazite 206Pb/238U and 208Pb/232Th ages to
reconstruct the U/Th of hydrothermal fluids.
1.5x10-4
DTh/DU = 0.2
DRa/DU = 0
Wendt & Carl (1985)
230
Th only Wendt & Carl (1985)
230 226
-4 Th and Ra
1.0x10
Pb*/238U
206
equilibrium Δt = 2 ka
5.0x10-5
Δt = 87 ka
Schärer (1984) 770 ka

0.0
0 200 400 600 800 1000
Age (ka)
Figure 6. Effect of initial disequilibrium on 206Pb*/238U zircon ages. Four different curves are shown be-
tween 0 and 1000 ka: secular equilibrium (solid line), disequilibrium for 230Th according to Equation (9)
(DTh / DU = 0.2; dash-dot), a simplified estimate from Schärer (1984) using f = 0.2 (short dash), and the un-
abridged calculation from Wendt and Carl (1985) which also accounts for 226Ra disequilibrium (DRa / DU = 0).
Inset illustrates the resulting age differences (Δt) for disequilibrium uncorrected and corrected ages for ca.
770 ka zircon (equivalent to the disequilibrium-corrected ID-TIMS zircon age of Bishop Tuff; Crowley et
al. 2007). Ickert et al. (2015) emphasize that disequilibrium corrections introduce systematic uncertainty,
and should not be propagated prior to calculation of weighted average ages (cf. Crowley et al. 2007).
By analogy to the recommended treatment of uncertainty of common Pb corrections,

disequilibrium-correction uncertainties should be treated as systematic uncertainties if f
(Eqn. 10) depends on model assumptions which is mostly the case because of the difficulties
in reconstructing melt Th/U and the impossibility of defining (230Th) / (238U) once equilibrium
has been reached (Ickert et al. 2015; Boehnke et al. 2016). These considerations also hold for
disequilibrium corrections in the 235U-decay chain, where DPa can only be estimated from model
(e.g., Blundy and Wood 2003) or empirical DPa / DU partitioning coefficient ratios (Schmitt
2007, 2011; Rioux et al. 2015) because no long-lived or stable Pa isotope exists, unlike 230Th
where long-lived 232Th can be used to estimate partitioning. Additional bias may result from
initial disequilibrium in 234U and 226Ra (Fig. 6). Unity activity ratios for (234U) / (238U) can be
reasonably expected for melts, but geothermal fluids often have elevated values which are
reflected by high initial (234U) / (238U) in hydrogenic minerals such as opal (e.g., Paces 2015).
Little is known on 226Ra partitioning in zircon and other accessory minerals, but D values are
expected to be very low (10−2–10−3) for zircon when using Ba as a proxy (Thomas et al. 2002).
Validation. Analyzing a secondary reference intermittently during an analytical session
and processing the data as if it was an unknown is an essential check to detect instrumental
malfunction and/or human error. This is important for both 206Pb/238U and 207Pb/206Pb ages.
The secondary reference should be as similar to the unknown as possible (e.g., in age and/or
U abundance) to facilitate the detection of potential analytical bias that could go undetected in
other reference materials. Potential bias could occur from unmitigated surficial contamination
(e.g., 204Pb or 204Hg), matrix effects (e.g., high-U zircon; White and Ireland 2012), or crystal
orientation effects (e.g., baddeleyite, or rutile; Wingate and Compston 2000; Schmitt and Zack
2012; Taylor et al. 2012). Validation data should be included in the publications, either as a
summary, or preferably together with data tabulations. To ensure comparability and possibly
recalculation of data, it is essential that data sources for reference materials are adequately cited.
When comparing data point analyses from the same analytical session (e.g., for calculating a
population average, uncertainty, and values for the mean square of weighted deviates MSWD),
systematic uncertainties should be excluded. Conversely, when comparing data from different
sessions, or labs, systematic uncertainties for different reference materials should be included.
COMPARING SIMS TO OTHER TECHNIQUES

SIMS stands out amongst other techniques for micro-isotopic analysis for its small
sampling volume and depth, and the preservation of samples for later analyses that may
consume large domains or entire crystals. Other techniques, such as micro-milling or laser
ablation, typically sample much larger volumes and often penetrate or consume entire
crystals. The high sensitivity and relatively slow sputter rates (e.g., ~0.1 µm3/s/nA for O−
sputtering of zircon) characteristic of SIMS ensure that sampling depths are shallow and
that experiments will be non-destructive relative to the typical size of accessory minerals.
Analysts can have confidence that “what you see is what you get” when analyses are guided
by cathodoluminescence or backscattered electron images. However, SIMS analysis times are
significantly longer than those for comparable analyses by LA-ICP-MS, which is impractical
for studies that require large datasets, such as the datasets that are standard for detrital mineral
geochronology. Atom probe tomography (APT) has been used to obtain isotope-specific 3D
reconstructions of individual atoms in samples shaped to tips < 10 nm wide and few 10’s of
nm long (e.g., Valley et al. 2015; Peterman et al. 2016). While this technique permits unique
insights into the chemical structure of accessory minerals at the nanoscale, the expensive
and time-consuming preparation of APT tips, and the small sample volume resulting in large
uncertainties for trace components limit the petrochronologic applicability of APT.
Preservation of the textural context of datable minerals is one of the key advantages of
microbeam methods, and SIMS excels for analysis of accessory mineral inclusions which are
too small to be separated for ID-TIMS, or to yield sufficient signal in LA-ICP-MS analysis. In-
situ dating of monazite, for example, has revealed age differences between crystals included in
garnet vs. those hosted in the matrix, which was attributed to shielding of monazite by its garnet
host (e.g., Foster et al. 2000; Catlos et al. 2002). By contrast, it is known that rock disintegration
and conventional mineral separation using gravity separation techniques (e.g., through water-
shaking, or heavy liquids) may fail in recovering the full spectrum of accessory minerals
present (e.g., Sláma and Košler 2012), or even any at all if grains are small (especially < 10 μm)
or included in mechanically strong host phases. Lab equipment used in mineral separation (e.g.,
crushers, shaking tables, magnetic separators, etc.) is also a potential source for contamination,
and especially for samples with low accessory mineral abundances it is imperative to use new
equipment and/or adhere to strict cleaning procedures to minimize the risk of crystal carry-over
from previously processed rocks (e.g., Torsvik et al. 2013). Even when thorough measures are
implemented in the mineral separation laboratory, there remains potential for unexpected, and
difficult to mitigate, contamination as demonstrated by the recent discovery of abundant zircon
in sepiolite drill mud used during ocean drilling (Andrews et al. 2016).
Lastly, SIMS is preferable when intact samples need to be preserved for later analyses.
Several studies have demonstrated that a combination of geochronology, trace elements,
and oxygen isotopes (measured by SIMS) and Hf isotopes (measured by LA-ICP-MS), all
from the same zircons, can provide essential insights ranging from the secular evolution
of Earth’s crust and mantle (e.g., Harrison et al. 2008) to the evolution of young magma
chambers beneath restive volcanoes (e.g., Stelten et al. 2015; Rubin et al. 2016). In these
studies, SIMS analyses were used first to generate dates and measure trace elements, then
the more penetrating LA-ICP-MS analyses were used to measure Hf isotope compositions.
The relatively non-destructive nature of SIMS allows single zircon crystals to be “double-
dated” (e.g., in volcanic rocks analyzed for their crystallization and eruption ages). For
example, Danišík et al. (2012) used SIMS to generate 238U–230Th dates for zircons from a
stratigraphically important tephra in New Zealand. The dated zircons were plucked from their
mounts and individually heated to release and analyze their radiogenic helium. This “double-
dating” approach was essential for generating a robust eruption age for the tephra, which was
confirmed by 14C dating of logs that were charred by the hot volcanic ash.
CASE STUDIES FOR SIMS PETROCHRONOLOGY

Other chapters in this volume cover case studies for SIMS in petrochronology in depth
(e.g., Engi 2017, this volume; Lanari and Engi 2017, this volume; Zack and Kooijman 2017,
this volume), and we therefore only briefly showcase some recent studies from the authors’
respective laboratories where the strengths of SIMS (i.e., in-situ analysis capability at high
spatial resolution, high-sensitivity, and versatility) proved essential for the success of the
research project. One such study (Shulaker et al. 2015) was the first to utilize rutilated quartz
as a thermogeochronometer. In rutilated quartz, low-temperature quartz (as evidenced from
the presence of fluid inclusions) hosts rutile, often in the form of fine needles with diameters
of few 10’s of μm. Although the exact formation mechanisms remain unclear, the close
spatial relationship between quartz and rutile offers the possibility to apply independent
geothermometers: Ti-in-quartz (TITANIQ; Wark and Watson 2006; Thomas et al. 2010), Zr-
in-rutile (Zack et al. 2004; Tomkins et al. 2007), and oxygen isotope exchange between quartz
and rutile (Matthews 1994). Trace element and isotopic equilibrium between rutile and quartz
is more likely for contact pairs, and analyzing rutile and quartz in petrographic context is
therefore a prerequisite for testing consistency between these independent geothermometers.
Shulaker et al. (2015) investigated a suite of six rutilated quartz samples from Alpine fissures
from host rocks which experienced peak metamorphic temperatures between  350 and
600 °C. Two of six samples yielded significant radiogenic 206Pb and a concordia intercept age
of ca. 15 Ma, consistent with radiometric ages for scarce monazite and titanite in Alpine clefts
(e.g., Janots et al. 2012). Reproducibility of SIMS oxygen isotope data in reference quartz was
~0.3‰ (2σ) whereas randomly oriented rutile crystals (R10B; Luvizotto et al. 2009) were only
reproducible to ~1‰ under the same analytical conditions. Shulaker et al. (2015) attributed
this to a crystal orientation effect. Because of strong isotopic differences over the pertaining
temperature interval (Δ18Oquartz–rutile = δ18Oquartz – δ18Orutile = +7 to +15‰), uncertainties related
to crystal orientation translate to only minor (20–60 °C) temperature uncertainties for the
quartz–rutile oxygen isotope exchange thermometer. Interestingly, oxygen isotope exchange
temperatures show no correlation with TITANIQ temperatures (using calibrations of Thomas
et al. 2010; Huang and Audétat 2012) at low temperatures (Fig. 7), whereas such a correlation
exists between oxygen isotope and Zr-in-rutile thermometers (Zack et al. 2004; Tomkins
et al. 2007), albeit with a significant offset for the Tomkins et al. (2007) calibration. These
results demonstrate how the strong oxygen isotopic fractionation between rutile and quartz
can be harnessed as a geothermometer, but they also urge caution for the application of trace
element geothermometers at low temperatures. The common occurrence of rutile and quartz in
metamorphic and hydrothermal vein rocks offers the possibility to further exploit this mineral
pair as a geothermometer with the bonus that rutile can be radiometrically dated.
Another recent study where the versatility and superb spatial resolution of SIMS came
to bear is Matthews et al. (2015), who combined trace elements and U–Pb geochronology of
Pleistocene zircons to resolve the eruption age and petrologic evolution of the magma chamber
responsible for the Lava Creek Tuff super-eruption and the formation of Yellowstone caldera
(Fig. 8). The Lava Creek eruption deposited a voluminous ignimbrite around Yellowstone, and
generated an ash bed that blanketed much of the western and central United States. The Lava
Creek ash has been a key marker bed for correlating glacial and pluvial deposits in North
America (Sarna-Wojcicki et al. 1984), delimiting rates of uplift (e.g., Darling et al. 2012) and
establishing faunal sequences (Bell et al. 2004). To constrain the eruption age of the Lava Creek
magma and resolve the final stages of its thermal and compositional evolution, Matthews et
al. (2015) sampled the unpolished faces of zircons from different members of the eruptive
stratigraphy, and integrated the results with a comparable dataset for the interiors of sectioned
crystals from the same populations of zircon. Matthews et al. (2015) derived a 206Pb/238U
date of ca. 627 ± 6 ka (2σ error) from the zircon crystal faces, which included a ca. +95 ka
T (°C) from Matthews (1994) O-isotope thermometer

600 500 400 300
10
Ti-in-quartz (ppm)
100
Zr-in-rutile
Zr-in-rutile (ppm)
?
1
Ti-in-quartz
10
2.5 3.0 3.5 4.0
√Δ
18
Oquartz-rutile
Figure 7. Comparison between oxygen isotope temperatures (top horizontal axis) derived from the rutile–
quartz oxygen isotope exchange thermometer (Matthews 1994; bottom axis) and temperature-dependent
Ti-in-quartz and Zr-in-rutile thermometers for rutilated quartz from Alpine fissures. Ti-in-quartz and Zr-
in-rutile temperatures are not quantified, but shown as abundances. Zr-in-rutile and oxygen isotope tem-
peratures correlate as expected, whereas no correlation exists between Ti-in-quarz and oxygen isotope
temperatures. Data from Shulaker et al. (2015) collected using a CAMECA IMS 1270.
20,000
17,500
15,000
12,500
Y (ppm)
Crystal faces
Dark CL cores
10,000
Med-Light CL cores
Intermediate interiors
7,500
5,000
2,500
0
550 600 650 700 750
206 238
Pb/ U date (ka)
Figure 8. Petrochronology of zircons from Member B of the Lava Creek Tuff, erupted from Yellowstone
caldera at ca. 630 ka. The 206Pb/238U dates for the cores and intermediate zones of sectioned crystals yield a
mean crystallization age that is ca. 35 kyr older than the mean age derived from corresponding crystal faces
(Matthews et al. 2015). Cores that are dark gray in cathodoluminescence images have elevated U, Y, and
large negative europium anomalies, suggesting growth from low temperature and evolved, i.e., near solidus,
rhyolitic magma prior to incorporation in less evolved rhyolite (Matthews et al. 2015). Data generated on the
Stanford-USGS SHRIMP-RG. Scale bar is 100 µm.
correction for initial 230Th/238U disequilibrium. This zircon rim crystallization age effectively
matched the eruption age apparent from the authors’ 40Ar/39Ar dating of coexisting sanidines,
as well as the stratigraphic position of the Lava Creek ash bed in marine sediments marking
the Marine Isotope Stage 15–16 boundary (Dean et al. 2015). In addition, the petrochronologic
approach used by Matthews et al. (2015) revealed that the Lava Creek magma evolved over a
ca. 35 kyr interval, with reversely zoned rims (i.e., lower U, Th, REE, Hf, and higher Eu/Eu*
compared to interior domains) indicating a change to a hotter and less evolved composition
within about 10 kyr of eruption (Fig. 8). Independent U–Pb dating of Lava Creek zircons by ID-
TIMS (Wotzlaw et al. 2015) confirmed the range of crystallization dates derived by Matthews
et al. (2015) via SIMS, and has illustrated the ability of high-spatial resolution SIMS dating to
provide reliable constraints on eruption ages with application to tephrochronology.
OUTLOOK
Improved detection
The fundamental design of large magnet radius SIMS instruments has experienced little
modification since the prototypes were commissioned at ANU (SHRIMP I) and UCLA (IMS
1270) in 1981 and 1992, respectively. In the late 1990’s, the reverse-geometry SIMS instrument
(SHRIMP-RG) deviated significantly from earlier designs, but it was not commercially
successful. Multi-detection capabilities were introduced for CAMECA and SHRIMP II
instruments in the late 1990’s and early 2000’s, respectively. Since then, SIMS platforms have
developed high-precision stable isotope analysis capabilities, which became only possible
with the advent of multi-collection using FC detectors. Further improvements of the detection
system remain highly desirable from a SIMS analyst’s perspective. Although multi-collection
of low intensity secondary ion signals is possible with discrete-dynode (CAMECA) and
continuous (SHRIMP) EM detectors, dead-time uncertainties and ageing are intrinsic draw-
backs of EM detectors. Multi-collection at low counts rates typical for the analysis of Pb and
U-series intermediate daughter isotopes is hardly affected by this, as EM gain drift correlates
with cumulative counts detected. For these applications, multi-collection can dramatically
increase through-put, if gain changes can be monitored and corrected for (e.g., Holden et al.
2009). Much more problematic are count rates around the “Megahertz gap”, the intensity range
of 106 counts per second, where EMs age rapidly. Although software permits gain drift and dead
time of EMs to be monitored in automated fashion, these parameters still result in significant
(at the 1‰ level) uncertainty into isotope ratio measurements. In this regard, designing FC pre-
amplifier circuits with high-impedance resistors is promising, as noise is expected to decrease
by a factor of ~3 per order-of-magnitude increase in resistivity. For SIMS instruments, this
has been successfully demonstrated for 1012 Ω resistors (the iFlex Faraday counting system for
SHRIMP), and even for 1013 Ω resistors in TIMS instruments where FC detection yielded more
precise measurements than EMs when currents were > 2 × 104 cps (Koornneef et al. 2014).
Ion source developments
Another field of current development is to improve the brightness, stability, and longevity
of ion sources. For positive secondary ion analysis, duoplasmatron sources are traditionally
applied. Maximum primary ion intensities of these sources are ~1 μA for O−, ~250 nA for O2−,
and 10 µA for O2+. Poor stability and frequent replacement of metal parts exposed to the oxygen
plasma (typically every 200–300 hours of use) are significant draw-backs of the duoplasmatron.
Recently, plasma ion sources that can operate in dual polarity (producing O− or O2+) have
been commercialized (Hyperion II by Oregon Physics) and been installed on CAMECA
instruments, including the large-magnet radius sources at UCLA. Initial tests showed a factor
of 8 gain in beam density for O− (Ming-Chang Liu, pers. comm). Higher brightness of the
oxygen ion source will lead to improved spatial resolution, and/or higher sputtering rates at the
same spot size compared to duoplasmatron sources, thus enhancing secondary ion signals. The
higher secondary ion currents resulting from the use of the plasma ion source will often be in
the range where they transcend the optimal intensity range for EM detection, which also calls
for the implementation of low-noise FC detectors. In addition to increasing source brightness,
there are developments in improving secondary ion yields through post-sputtering ionization.
Resonance ionization mass spectrometry (RIMS) at much higher (~30–40%) useful yields than
possible in normal SIMS is currently performed with experimental instrumentation (e.g., the
Chicago Instrument for Laser Ionization CHILI; Stephan et al. 2016).
Sample holder design and automation
Precision of stable isotope analysis has benefited from the introduction of sample
holders by CAMECA (Peres et al. 2013) with a larger annulus to hold the edge of the sample
than previous holders. This has the advantage that a more uniform sample potential can
be maintained across the central part of the holder than is possible with traditional sample
holders, thus eliminating strong fractionation when approaching the edges of the sample
(Treble et al. 2007). SHRIMP megamounts with 35 mm diameter were designed to achieve the
same uniform IMF over the central 19 mm diameter bullseye (Ickert et al. 2008). Compared
to commercial laser ablation cells, however, the useful surface area in SIMS is much smaller.
Automated sample exchange improves throughput, and while this is available for SHRIMP
instruments, it is currently under development for large magnet radius CAMECA instruments.
Final considerations
The past decade has seen a dramatic improvement of stable isotope analytical capabilities
with SIMS, with 0.1–0.2‰ precision for replicate individual spot analyses for δ18O in
chemically simple materials such as zircon, quartz, or calcite. By contrast, analytical limitations
for geochronologic SIMS have remained at the ~1–2% (2σ) level which was originally achieved
about three decades ago. These limitations are imposed by the reproducibility of the 2-D
calibration curve for U–Pb and Th–Pb, and are largely independent of the instrumental design.
This does not imply complacency by SIMS geochronologists: significant instrument time was
dedicated by all laboratories invested in petrochronology to gain a better understanding of the
sources of uncertainty and bias in SIMS. However, no silver bullet has been identified to break
through the 1% reproducibility barrier for U–Pb and Th–Pb despite many valuable insights
gained into the understanding of how instrumental parameters (e.g., Magee et al. 2014) as
well as matrix and crystal orientation effects (e.g., Taylor et al. 2012; White and Ireland 2012)
can affect SIMS relative sensitivities for geochronologically relevant elements. In this regard,
the high-spatial resolution methods, SIMS and LA-ICP-MS, remain limited in precision to
ID-TIMS. For petrochronological analysis of complexly zoned accessory minerals the key
advantage is accuracy albeit at lower precision compared to high-precision ID-TIMS ages where
bulk analysis of crystal aliquots, and even single crystals and crystal fragments, would yield
geologically meaningless averages (Table 1). This level of resolution is particularly valuable
where crystals are analyzed in situ. Reconnaissance studies indicate a lack of bias if accessory
minerals are analyzed by SIMS as grain separates conventionally embedded in epoxy vs.
in-situ in a petrographic thin-section, even when the primary beam partially overlaps adjacent
minerals in the thin section (Schmitt et al. 2010), but few such systematic comparisons have
been conducted. Because in-situ analysis is such an important field of SIMS petrochronology,
revisiting targeting, mounting, instrumental tuning, and calibration strategies for small accessory
minerals hosted in a rocky matrix should become a major goal for improving throughput and
bolstering confidence in the data. Another highly promising field is the combination of chemical
(via SIMS) and structural information (via electron backscatter diffraction, or micro-Raman
spectroscopy). These techniques are powerful in identifying phase changes, metamictization or
overgrowths at the micrometer scales which can aid in the interpretation of petrochronologic
data (e.g., for impact metamorphosed meteorites; Darling et al. 2016). At present, users of
SIMS petrochronological data should stay aware of the non-routine nature of many of these
analyses which can still provide some formidable challenges for the SIMS specialist.
ACKNOWLEDGMENTS
AKS expresses special thanks to all present and former members of the UCLA ion
microprobe lab for their long-term support, in particular Chris Coath, Marty Grove, T. Mark
Harrison, George Jarzebinski, Ming-Chang Liu, Oscar Lovera, Kevin McKeegan, and Lvcian
Vltava. Insightful discussions with Thomas Ludwig (Heidelberg University) are acknowledged.
JAV thanks Charlie Bacon, Matt Coble, Marty Grove, Brad Ito, Marsha Lidzbarski, and Joe
Wooden for their help and support in the Stanford-USGS ion microprobe laboratory, and is
grateful to the UCLA ion probe lab for his graduate training. We also thank John Aleinikoff,
Brian Monteleone, and Richard Stern for their helpful reviews, and Pierre Lanari for careful
editorial handling. The ion microprobe facility at UCLA is partly supported by a grant from
the Instrumentation and Facilities Program, Division of Earth Sciences, National Science
Foundation. Any use of trade, firm, or product names is for descriptive purposes only and does
not imply endorsement by the U.S. Government.
REFERENCES
Abbott SS, Harrison TM, Schmitt AK, Mojzsis SJ (2012) A search for thermal excursions from ancient
extraterrestrial impacts using Hadean zircon Ti–U–Th–Pb depth profiles. PNAS 109:13486–13492
Aleinikoff JN, Schenck WS, Plank MO, Srogi L, Fanning CM, Kamo SL, Bosbyshell H (2006) Deciphering
igneous and metamorphic events in high-grade rocks of the Wilmington Complex, Delaware: Morphology,
cathodoluminescence and backscattered electron zoning, and SHRIMP U–Pb geochronology of zircon and
monazite. Geol Soc Am Bull 118:39–64
Aleinikoff JN, Wintsch RP, Tollo RP, Unruh DM, Fanning CM, Schmitz MD (2007) Ages and origins of rocks of
the Killingworth dome, south-central Connecticut: Implications for the tectonic evolution of southern New
England. Am J Sci 307:63–118
Aleinikoff JN, Grauch RI, Mazdab FK, Kwak L, Fanning CM, Kamo SL (2012) Origin of an unusual monazite–
xenotime gneiss, Hudson Highlands, New York: SHRIMP U–Pb geochronology and trace element
geochemistry. Am J Sci 312:723–765
Andrews GD, Schmitt AK, Busby CJ, Brown SR, Blum P, Harvey J (2016) Age and compositional data of
zircon from sepiolite drilling mud to identify contamination of ocean drilling samples. Geochem Geophys
Geosyst:3512–3526
Baldwin SL, Ireland TR (1995) A tale of two eras: Pliocene-Pleistocene unroofing of Cenozoic and late Archean
zircons from active metamorphic core complexes, Solomon Sea, Papua New Guinea. Geology 23:1023–1026.
Barth AP, Wooden JL (2010) Coupled elemental and isotopic analyses of polygenetic zircons from granitic rocks
by ion microprobe, with implications for melt evolution and the sources of granitic magmas. Chem Geol
277:149–159
Bell CJ, Lundelius Jr EL, Barnosky A, Graham R, Lindsay E, Ruez Jr D, Semken Jr H, Webb S, Zakrzewski R,
Woodburne M (2004) The Blancan, Irvingtonian, and Rancholabrean mammal ages. Late Cretaceous and
Cenozoic mammals of North America: Biostratigraphy and Geochronology:232–314
Bindeman I, Serebryakov N, Schmitt A, Vazquez J, Guan Y, Azimov PY, Astafiev BY, Palandri J, Dobrzhinetskaya
L (2014) Field and microanalytical isotopic investigation of ultradepleted in 18O Paleoproterozoic “Slushball
Earth” rocks from Karelia, Russia. Geosphere 10:308–339
Black LP, Kamo SL, Allen CM, Davis DW, Aleinikoff JN, Valley JW, Mundil R, Campbell IH, Korsch RJ,
Williams IS (2004) Improved 206Pb/238U microprobe geochronology by the monitoring of a trace-element-
related matrix effect; SHRIMP, ID–TIMS, ELA–ICP–MS and oxygen isotope documentation for a series of
zircon standards. Chem Geol 205:115–140
Blundy J, Wood B (2003) Mineral–melt partitioning of uranium, thorium and their daughters. Rev Mineral
Geochem 52:59–123
Boehnke P, Harrison TM (2014) A meta-analysis of geochronologically relevant half-lives: what’s the best decay
constant? Int Geol Rev 56:905–914
Boehnke P, Barboni M, Bell E (2016) Zircon U/Th model ages in the presence of melt heterogeneity. Quat
Geochronol 34:69–74
Bouvier A-S, Ushikubo T, Kita NT, Cavosie AJ, Kozdon R, Valley JW (2012) Li isotopes and trace elements as a
petrogenetic tracer in zircon: insights from Archean TTGs and sanukitoids. Contrib Mineral Petrol 163:745–768
Breecker DO, Sharp ZD (2007) A monazite oxygen isotope thermometer. Am Mineral 92:1561–1572
Breeding CM, Ague JJ, Grove M, Rupke AL (2004) Isotopic and chemical alteration of zircon by metamorphic
fluids: U–Pb age depth-profiling of zircon crystals from Barrow’s garnet zone, northeast Scotland. Am
Mineral 89:1067–1077
Catlos E, Sorensen SS, Harrison TM (2000) Th–Pb ion-microprobe dating of allanite. Am Mineral 85:633–648
Catlos E, Gilley L, Harrison TM (2002) Interpretation of monazite ages obtained via in situ analysis. Chem Geol
188:193–215
Cherniak DJ, Watson EB (2003) Diffusion in zircon. Rev Mineral Geochem 53:113–143
Cherniak D, Watson EB, Grove M, Harrison TM (2004) Pb diffusion in monazite: a combined RBS/SIMS study.
Claiborne LL, Miller CF, Flanagan DM, Clynne MA, Wooden JL (2010) Zircon reveals protracted magma storage
and recycling beneath Mount St. Helens. Geology 38:1011–1014
Compston W, Clement SWJ (2006) The geological microprobe: The first 25 years of dating zircons. Appl Surf Sci
252:7089–7095
Compston W, Williams I, Meyer C (1984) U‐Pb geochronology of zircons from lunar breccia 73217 using a
sensitive high mass‐resolution ion microprobe. J Geophys Res Solid Earth 89(S02): B525-B534
Coombs ML, Vazquez JA (2014) Cogenetic late Pleistocene rhyolite and cumulate diorites from Augustine Volcano
revealed by SIMS 238U‐230Th dating of zircon, and implications for silicic magma generation by extraction
from mush. Geochem Geophys Geosyst 15:4846–4865
Crowley J, Schoene B, Bowring S (2007) U–Pb dating of zircon in the Bishop Tuff at the millennial scale. Geology
35:1123–1126
Danišík M, Shane P, Schmitt AK, Hogg A, Santos GM, Storm S, Evans NJ, Fifield LK, Lindsay JM (2012) Re-
anchoring the late Pleistocene tephrochronology of New Zealand based on concordant radiocarbon ages and
combined 238U/230Th disequilibrium and (U–Th)/He zircon ages. Earth Planet Sci Lett 349:240–250
Darling AL, Karlstrom KE, Granger DE, Aslan A, Kriby E, Ouimet WB, Lazear GD, Coblentz DD, Cole RD
(2012) New incision rates along the Colorado River system based on cosmogenic burial dating of terraces:
Implications for regional controls on Quaternary incision. Geosphere:1020–1041
Darling JR, Moser DE, Barker IR, Tait KT, Chamberlain KR, Schmitt AK, Hyde BC (2016) Variable microstructural
response of baddeleyite to shock metamorphism in young basaltic shergottite NWA 5298 and improved U–Pb
dating of Solar System events. Earth Planet Sci Lett 444:1–12
de Chambost E (2011) A history of CAMECA (1954–2009). Adv Imaging Electron Phys 167:1–119
Dean WE, Kennett JP, Behl RJ, Nicholson C, Sorlien CC (2015) Abrupt termination of Marine Isotope Stage 16
(Termination VII) at 631.5 ka in Santa Barbara Basin, California. Paleoceanogr 30:1373–1390
Geochem 83:365–418
Fisher CM, Vervoort JD, Hanchar JM (2014) Guidelines for reporting zircon Hf isotopic data by LA-MC-ICPMS
and potential pitfalls in the interpretation of these data. Chem Geol 363:125–133
Fletcher IR, McNaughton NJ, Aleinikoff JA, Rasmussen B, Kamo SL (2004) Improved calibration procedures
and new standards for U–Pb and Th–Pb dating of Phanerozoic xenotime by ion microprobe. Chem Geol
209:295–314
Fletcher IR, McNaughton NJ, Davis WJ, Rasmussen B. (2010) Matrix effects and calibration limitations in ion
probe U–Pb and Th–Pb dating of monazite. Chem Geol 270:31–44.
181:327–340
Gregory CJ, Rubatto D, Allen CM, Williams IS, Hermann J, Ireland T (2007) Allanite micro-geochronology: a
LA-ICP-MS and SHRIMP U–Th–Pb study. Chem Geol 245:162–182
Grimes CB, Ushikubo T, John BE, Valley JW (2011) Uniformly mantle-like δ18O in zircons from oceanic
plagiogranites and gabbros. Contrib Mineral Petrol 161:13–33
Grimes C, Wooden J, Cheadle M, John B (2015) “Fingerprinting” tectono-magmatic provenance using trace
elements in igneous zircon. Contrib Mineral Petrol 170:1–26
Harrison TM, Schmitt AK (2007) High sensitivity mapping of Ti distributions in Hadean zircons. Earth Planet Sci
Lett 261:9–19
Harrison TM, McKeegan K, LeFort P (1995) Detection of inherited monazite in the Manaslu leucogranite by 208Pb–
232
Th ion microprobe dating: crystallization age and tectonic implications. Earth Planet Sci Lett 133:271–282
Harrison TM, Schmitt AK, McCulloch MT, Lovera OM (2008) Early (≥ 4.5 Ga) formation of terrestrial crust: Lu–
Hf, δ18O, and Ti thermometry results for Hadean zircons. Earth Planet Sci Lett 268:476–486
Hervig RL, Mazdab FK, Williams P, Guan Y, Huss GR, Leshin LA (2006) Useful ion yields for Cameca IMS 3f
and 6f SIMS: Limits on quantitative analysis. Chem Geol 227:83–99
Hinthorne J, Andersen C, Conrad R, Lovering J (1979) Single-grain 207Pb/206Pb and U/Pb age determinations with
a 10-μm spatial resolution using the ion microprobe mass analyzer (IMMA). Chem Geol 25:271–303
Hinton R (1990) Ion microprobe trace-element analysis of silicates: Measurement of multi-element glasses. Chem
Geol 83:11–25
Hofmann AE, Valley JW, Watson EB, Cavosie AJ, Eiler JM (2009) Sub-micron scale distributions of trace elements
in zircon. Contrib Mineral Petrol 158:317–335
Hofmann AE, Baker MB, Eiler JM (2014) Sub-micron-scale trace-element distributions in natural zircons of
known provenance: implications for Ti-in-zircon thermometry. Contrib Mineral Petrol 168:1–21
Holden P, Lanc P, Ireland TR, Harrison TM, Foster JJ, Bruce Z (2009) Mass-spectrometric mining of Hadean
zircons by automated SHRIMP multi-collector and single-collector U/Pb zircon age dating: the first 100,000
grains. Int J Mass Spectrom 286:53–63
Hoppe P, Cohen S, Meibom A (2013) NanoSIMS: technical aspects and applications in cosmochemistry and
biological geochemistry. Geostand Geoanal Res 37:111–154
Horstwood MS, Košler J, Gehrels G, Jackson SE, McLean NM, Paton C, Pearson NJ, Sircombe K, Sylvester P,
Vermeesch P (2016) Community‐derived standards for LA‐ICP‐MS U‐(Th‐) Pb geochronology–uncertainty
propagation, age interpretation and data reporting. Geostand Geoanal Res 40:311–332
Hoskin PW, Schaltegger U (2003) The composition of zircon and igneous and metamorphic petrogenesis. Rev
Huang R, Audétat A (2012) The titanium-in-quartz (TitaniQ) thermobarometer: a critical examination and re-
calibration. Geochim Cosmochim Acta 84:75–89
Huberty JM, Kita NT, Kozdon R, Heck PR, Fournelle JH, Spicuzza MJ, Xu H, Valley JW (2010) Crystal orientation
effects in δ18O for magnetite and hematite by SIMS. Chem Geol 276:269–283
Hunter J (2009) Improving depth profile measurements of natural materials: Lessons learned from electronic
materials depth-profiling. Secondary Ion Mass Spectrometry in the Earth Sciences: Gleaning the Big Picture
from a Small Spot. Mineral Assoc Can Short Course 41:133–148
Ickert R, Hiess J, Williams I, Holden P, Ireland T, Lanc P, Schram N, Foster J, Clement S (2008) Determining high
precision, in situ, oxygen isotope ratios with a SHRIMP II: analyses of MPI-DING silicate-glass reference
materials and zircon from contrasting granites. Chem Geol 257:114–128
Ickert RB, Mundil R, Magee CW, Mulcahy SR (2015) The U–Th–Pb systematics of zircon from the Bishop Tuff: A
case study in challenges to high-precision Pb/U geochronology at the millennial scale. Geochim Cosmochim
Acta 168:88–110
Ireland TR (1995) Ion microprobe mass spectrometry: techniques and applications in cosmochemistry,
geochemistry, and geochronology. Adv Anal Geochem 2:1–118
Ireland TR (2015) Secondary ion mass spectrometry (SIMS). In: Encyclopedia of Scientific Dating Methods. Rink
WJ, Thompson JW (eds) Springer, Berlin, p 739–740
Ireland TR, Williams IS (2003) Considerations in zircon geochronology by SIMS. Rev Mineral Geochem 53:215–241
Ireland T, Clement S, Compston W, Foster J, Holden P, Jenkins B, Lanc P, Schram N, Williams I (2008)
Development of SHRIMP. Australas J Earth Sci 55:937–954
Ireland T, Schram N, Holden P, Lanc P, Avila J, Armstrong R, Amelin Y, Latimore A, Corrigan D, Clement S (2014)
Charge-mode electrometer measurements of S-isotopic compositions on SHRIMP-SI. Int J Mass Spectrom
359:26–37
Janots E, Berger A, Gnos E, Whitehouse M, Lewin E, Pettke T (2012) Constraints on fluid evolution during
metamorphism from U–Th–Pb systematics in Alpine hydrothermal monazite. Chem Geol 326:61–71
Jeon H, Whitehouse MJ (2015) A critical evaluation of U–Pb calibration schemes used in SIMS zircon
geochronology. Geostand Geoanal Res 39:443–452
Kinny PD, Compston W, Williams IS (1991) A reconnaissance ion-probe study of hafnium isotopes in zircons.
Kita NT, Ushikubo T, Fu B, Valley JW (2009) High precision SIMS oxygen isotope analysis and the effect of
sample topography. Chem Geol 264:43–57
Koornneef J, Bouman C, Schwieters J, Davies G (2014) Measurement of small ion beams by thermal ionisation
mass spectrometry using new 1013 Ohm resistors. Analytica Chim Acta 819:49–55
Lanari P, Engi M (2017) Local bulk composition effects on metamorphic mineral assemblages. Rev Mineral Geochem
83:55–102
Li Q-L, Li X-H, Liu Y, Tang G-Q, Yang J-H, Zhu W-G (2010) Precise U–Pb and Pb–Pb dating of Phanerozoic
baddeleyite by SIMS with oxygen flooding technique. J Anal At Spectrom 25:1107–1113
Ludwig K (2009) Errors of isotope ratios acquired by double interpolation. Chem Geol 268:24–26
Luvizotto G, Zack T, Meyer H, Ludwig T, Triebold S, Kronz A, Münker C, Stockli D, Prowatke S, Klemme S
(2009) Rutile crystals as potential trace element and isotope mineral standards for microanalysis. Chem Geol
261:346–369
Matsuda H (1974) Double focusing mass spectrometers of second order. Int J Mass Spectrom Ion Phys 14:219–233
Matsuda H (1990) High performance mass spectrometers of magnetic sector type. Int J Mass Spectrom Ion
Processes 100:31–39
Matthews A (1994) Oxygen isotope geothermometers for metamorphic rocks. J Metamorph Geol 12:211–219
Matthews NE, Vazquez JA, Calvert AT (2015) Age of the Lava Creek supereruption and magma chamber assembly
at Yellowstone based on 40Ar/39Ar and U‐Pb dating of sanidine and zircon crystals. Geochem Geophys
Geosyst 16:2508–2528
Mattinson JM (2010) Analysis of the relative decay constants of 235U and 238U by multi-step CA-TIMS
measurements of closed-system natural zircon samples. Chem Geol 275:186–198
Mattinson CG, Wooden JL, Liou JG, Bird D, Wu C (2006) Age and duration of eclogite-facies metamorphism,
North Qaidam HP/UHP terrane, Western China. Am J Sci 306:683–711
Magee C Jr, Ferris J, Magee C Sr (2014) Effect of impact energy on SIMS U–Pb zircon geochronology. Surf
Interface Anal 46(S1):322–5.
Mazdab FK (2009) Characterization of flux-grown trace-element-doped titanite using the high-mass-resolution ion
microprobe (SHRIMP–RG). Can Mineral 47:813–831
McFarlane C, McCulloch M (2007) Coupling of in-situ Sm–Nd systematics and U–Pb dating of monazite and
allanite with applications to crustal evolution studies. Chem Geol 245:45–60
Nemchin A, Whitehouse M, Pidgeon R, Meyer C (2006) Oxygen isotopic signature of 4.4–3.9 Ga zircons as a
monitor of differentiation processes on the Moon. Geochim Cosmochim Acta 70:1864–1872
Paces JB (2015) Uranium Series, Opal. Encyclopedia of Scientific Dating Methods:837–843
Page F, Ushikubo T, Kita NT, Riciputi L, Valley JW (2007) High-precision oxygen isotope analysis of picogram
samples reveals 2 μm gradients and slow diffusion in zircon. Am Mineral 92:1772–1775
Peres P, Kita N, Valley J, Fernandes F, Schuhmacher M (2013) New sample holder geometry for high precision
isotope analyses. Surf Interface Anal 45:553–556
Peterman EM, Reddy SM, Saxey DW, Snoeyenbos DR, Rickard WD, Fougerouse D, Kylander-Clark AR (2016)
Nanogeochronology of discordant zircon measured by atom probe microscopy of Pb-enriched dislocation
loops. Sci Adv 2:e1601318
Pidgeon R, Nemchin A, Cliff J (2013) Interaction of weathering solutions with oxygen and U–Pb isotopic systems
of radiation-damaged zircon from an Archean granite, Darling Range Batholith, Western Australia. Contrib
Rayner NM, Stern RA (2002) Improved sample preparation method for SHRIMP analysis of delicate mineral
grains exposed in thin sections. Natural Resources Canada, Geological Survey of Canada, http://www.
publications.gc.ca/collections/collection_2011/rncan-nrcan/M44–2002-F10-eng.pdf
Reid MR, Coath CD, Harrison TM, McKeegan KD (1997) Prolonged residence times for the youngest rhyolites
associated with Long Valley Caldera: 230Th–238U ion microprobe dating of young zircons. Earth Planet Sci
Lett 150:27–39
Rioux M, Bowring S, Cheadle M, John B (2015) Evidence for initial excess 231Pa in mid-ocean ridge zircons.
Chem Geol 397:143–156
Rubatto D, Putlitz B, Gauthiez-Putallaz L, Crépisson C, Buick IS, Zheng Y-F (2014) Measurement of in-situ
oxygen isotope ratios in monazite by SHRIMP ion microprobe: Standards, protocols and implications. Chem
Geol 380:84–96
Rubin A, Cooper K, Kent A, Costa Rodriguez F, Till C (2014) Using Li diffusion to track thermal histories within
single zircon crystals. AGU Abstract Fall 2014.
Rubin A, Cooper KM, Leever M, Wimpenny J, Deering C, Rooney T, Gravley D, Yin Q-Z (2016) Changes in
magma storage conditions following caldera collapse at Okataina Volcanic Center, New Zealand. Contrib
Sarna-Wojcicki AM, Bowman H, Meyer CE, Russell P, Woodward M, McCoy G, Rowe Jr J, Baedecker P, Asaro
F, Michael H (1984) Chemical analyses, correlations, and ages of upper Pliocene and Pleistocene ash layers
of east-central and southern California. In: Chemical Analyses, Correlations, and Ages of Ipper Pliocene and
Pleistocene Ash Layers of East-Central and Southern California. USGS Prof P 1293:1–40.
Schaltegger U, Schmitt A, Horstwood M (2015) U–Th–Pb zircon geochronology by ID-TIMS, SIMS, and laser
ablation ICP-MS: recipes, interpretations, and opportunities. Chem Geol 402:89–110
Schärer U (1984) The effect of initial 230Th disequilibrium on young U–Pb ages: the Makalu case, Himalaya. Earth
Schmitt AK (2007) Letter: Ion microprobe analysis of (231Pa)/(235U) and an appraisal of protactinium partitioning
in igneous zircon. Am Mineral 92:691–694
Schmitt AK (2011) Uranium series accessory crystal dating of magmatic processes. Annu Rev Earth Planet Sci
39:321–349
Schmitt AK, Zack T (2012) High-sensitivity U–Pb rutile dating by secondary ion mass spectrometry (SIMS) with
an O2+ primary beam. Chem Geol 332:65–73
Schmitt AK, Chamberlain KR, Swapp SM, Harrison TM (2010) In situ U–Pb dating of micro-baddeleyite by
secondary ion mass spectrometry. Chem Geol 269:386–395
Schmitt AK, Martín A, Stockli DF, Farley KA, Lovera OM (2013) (U–Th)/He zircon and archaeological ages for a
late prehistoric eruption in the Salton Trough (California, USA). Geology 41:7–10
Schoene B (2014) 4.10-U–Th–Pb Geochronology. Treatise on Geochemistry, Second Edition, Elsevier,
Oxford:341–378
Schuhmacher M, De Chambost E, McKeegan K, Harrison TM, Migeon H (1994) In situ dating of zircon with the
CAMECA ims 1270. Secondary Ion Mass Spectrometry SIMS IX:919–922
Seitz S, Putlitz B, Baumgartner LP, Escrig S, Meibom A, Bouvier AS (2016) Short magmatic residence times of
quartz phenocrysts in Patagonian rhyolites associated with Gondwana breakup. Geology 44:67–70
Shimizu N (1978) Analysis of the zoned Plagioclase of different magmatic environments: A Preliminary ion
microprobe study, Earth Planet Sci Lett 39:398–406.
Shimizu N, Hart S (1982) Isotope fractionation in secondary ion mass spectrometry. J Appl Phys 53:1303–1311
Shulaker DZ, Schmitt AK, Zack T, Bindeman I (2015) In-situ oxygen isotope and trace element geothermometry
of rutilated quartz from Alpine fissures. Am Mineral 100:915–925
Simon JI, Renne PR, Mundil R (2008) Implications of pre-eruptive magmatic histories of zircons for U–Pb
geochronology of silicic extrusions. Earth Planet Sci Lett 266:182–194
Sláma J, Košler J (2012) Effects of sampling and mineral separation on accuracy of detrital zircon studies. Geochem
Geophys Geosyst 13:Q05007, doi:10.1029/2012GC004106
Stelten ME, Cooper KM, Vazquez JA, Calvert AT, Glessner JJ (2015) Mechanisms and timescales of generating
eruptible rhyolitic magmas at Yellowstone caldera from zircon and sanidine geochronology and geochemistry.
J Petrol 56:1607–1642
Stephan T, Trappitsch R, Davis AM, Pellin MJ, Rost D, Savina MR, Yokochi R, Liu N (2016) CHILI–the Chicago
Instrument for Laser Ionization–a new tool for isotope measurements in cosmochemistry. Int J Mass
Spectrom 407:1–15
Stern R (2009) An introduction to secondary ion mass spectrometry (SIMS) In: Geology. Secondary Ion Mass
Spectrometry in the Earth Sciences: Gleaning the big picture from a small spot. Mineral Assoc Can Short
Course 41:1–18
Stern RA, Amelin Y (2003) Assessment of errors in SIMS zircon U–Pb geochronology using a natural zircon
standard and NIST SRM 610 glass. Chem Geol 197:111–142
Stern RA, Fletcher IR, Rasmussen B, McNaughton NJ, Griffin BJ (2005) Ion microprobe (NanoSIMS 50) Pb-
isotope geochronology at < 5μm scale. Int J Mass Spectrom 244:125–134
Storm S, Shane P, Schmitt AK, Lindsay JM (2011) Contrasting punctuated zircon growth in two syn-erupted
rhyolite magmas from Tarawera volcano: Insights to crystal diversity in magmatic systems. Earth Planet Sci
Lett 301:511–520
Taylor R, Clark C, Reddy SM (2012) The effect of grain orientation on secondary ion mass spectrometry (SIMS)
analysis of rutile. Chem Geol 300:81–87
Thomas J, Bodnar R, Shimizu N, Sinha A (2002) Determination of zircon/melt trace element partition coefficients
from SIMS analysis of melt inclusions in zircon. Geochim Cosmochim Acta 66:2887–2901
Thomas JB, Watson EB, Spear FS, Shemella PT, Nayak SK, Lanzirotti A (2010) TitaniQ under pressure: the effect
of pressure and temperature on the solubility of Ti in quartz. Contrib Mineral Petrol 160:743–759
Till CB, Vazquez, JA, Boyce JW (2015) Months between rejuvenation and volcanic eruption at Yellowstone
caldera, Wyoming. Geology 43:695–698
Tomkins H, Powell R, Ellis D (2007) The pressure dependence of the zirconium‐in‐rutile thermometer. J
Torsvik TH, Amundsen H, Hartz EH, Corfu F, Kusznir N, Gaina C, Doubrovine PV, Steinberger B, Ashwal LD,
Jamtveit B (2013) A Precambrian microcontinent in the Indian Ocean. Nat Geosci 6:223–227
Trail D, Cherniak DJ, Watson EB, Harrison TM, Weiss BP, Szumila I (2016) Li zoning in zircon as a potential
geospeedometer and peak temperature indicator. Contrib Mineral Petrol 171:1–15
Treble P, Schmitt AK, Edwards R, McKeegan KD, Harrison T, Grove M, Cheng H, Wang Y (2007) High resolution
Secondary Ionisation Mass Spectrometry (SIMS) δ18O analyses of Hulu Cave speleothem at the time of
Heinrich Event 1. Chem Geol 238:197–212
Ushikubo T, Kita NT, Cavosie AJ, Wilde SA, Rudnick RL, Valley JW (2008) Lithium in Jack Hills zircons:
Evidence for extensive weathering of Earth’s earliest crust. Earth Planet Sci Lett 272:666–676
Valley JW, Kita NT (2009) In situ oxygen isotope geochemistry by ion microprobe. Secondary Ion Mass
Spectrometry in the Earth Sciences: Gleaning the big picture from a small spot. Mineral Assoc Can Short
Course 41:19–63
Valley JW., Reinhard DA, Cavosie AJ, Ushikubo T, Lawrence DF, Larson DJ, Kelly TF, Snoeyenbos DR, Strickland
A (2015) Presidential Address. Nano-and micro-geochronology in Hadean and Archean zircons by atom-
probe tomography and SIMS: New tools for old minerals. Am Mineral 100:1355–1377
Vazquez JA, Reid MR (2004) Probing the accumulation history of the voluminous Toba magma. Sci 305:991–994
Vazquez JA, Velasco NO, Schmitt AK, Bleick HA, Stelten ME (2014) 238U–230Th dating of chevkinite in high-silica
rhyolites from La Primavera and Yellowstone calderas. Chem Geol 390:109–118
von Blanckenburg F (1992) Combined high-precision chronometry and geochemical tracing using accessory
minerals: applied to the Central-Alpine Bergell intrusion (central Europe). Chem Geol 100:19–40
Vorhies SH, Ague JJ, Schmitt AK (2013) Zircon growth and recrystallization during progressive metamorphism,
Barrovian zones, Scotland. Am Mineral 98:219–230
Wark DA, Watson EB (2006) TitaniQ: a titanium-in-quartz geothermometer. Contrib Mineral Petrol 152:743–754
Watson E, Harrison T (2005) Zircon thermometer reveals minimum melting conditions on earliest Earth. Sci
308:841–844
Watts KE, Coble MA, Vazquez JA, Henry CD, Colgan JP, John DA (2016) Chemical abrasion-SIMS (CA-SIMS)
U–Pb dating of zircon from the late Eocene Caetano caldera, Nevada. Chem Geol 439:139–151
Wendt I, Carl C (1985) U/Pb dating of discordant 0.1 Ma old secondary U minerals. Earth Planet Sci Lett 73:278–284
White L, Ireland T (2012) High-uranium matrix effect in zircon and its implications for SHRIMP U–Pb age
determinations. Chem Geol 306:78–91
Wiedenbeck M, Hanchar JM, Peck WH, Sylvester P, Valley J, Whitehouse M, Kronz A, Morishita Y, Nasdala L,
Fiebig J (2004) Further characterisation of the 91500 zircon crystal. Geostand Geoanal Res 28:9–39
Williams I, McKibben M (1998) Applications of microanalytical techniques to understanding mineralizing
processes. Rev Econ Geol 7:1–35
Wingate M, Compston W (2000) Crystal orientation effects during ion microprobe U–Pb analysis of baddeleyite.
Wotzlaw J-F, Bindeman IN, Stern RA, D’Abzac F-X, Schaltegger U (2015) Rapid heterogeneous assembly of
multiple magma reservoirs prior to Yellowstone supereruptions. Sci Rep 5, doi:10.1038/srep14026
Wright HM, Vazquez JA, Champion DE, Calvert AT, Mangan MT, Stelten M, Cooper KM, Herzig C, Schriener A
(2015) Episodic Holocene eruption of the Salton Buttes rhyolites, California, from paleomagnetic, U‐Th, and
Ar/Ar dating. Geochem Geophys Geosyst 16:1198–1210
Yang W, Lin Y-T, Zhang J-C, Hao J-L, Shen W-J, Hu S (2012) Precise micrometre-sized Pb–Pb and U–Pb dating
with NanoSIMS. J Anal At Spectrom 27:479–487
Zinner E, Crozaz G (1986) A method for the quantitative measurement of rare earth elements in the ion microprobe.
Int J Mass Spectrom Ion Processes 69:17–38
Vol. 83 pp. 231–260, 2017 8
Petrochronology and TIMS

Blair Schoene
Princeton University
Princeton, NJ 08544
USA
bschoene@princeton.edu
Ethan F. Baxter
Boston College
Chestnut Hill, MA 02467
USA
baxteret@bc.edu
INTRODUCTION
Thermal ionization mass spectrometers, or TIMS, were developed by the pioneers of mass
spectrometry in the mid-20th century, and have since been workhorses for generating isotopic
data for a wide range of elements. Later-developed mass spectrometric techniques have many
advantages over TIMS, including higher spatial resolution with in situ techniques, such as
secondary ion mass spectrometry (SIMS) and laser ablation inductively coupled plasma mass
spectrometry (LA-ICPMS), and greater versatility in terms the elements that can be easily-
and well-measured. The reason TIMS persists as an important method for geochronology
is that for some key parent-daughter systems (e.g., U–Pb, Sm–Nd), it can produce isotopic
data and resultant dates with 10−100 times higher precision and more quantifiable accuracy
than in situ techniques, even when sample sizes are very small (such as those that might
result from single crystals, or even small portions of zoned crystals). For many questions in
the geosciences, the highest achievable precision and accuracy are required to resolve the
timescales of processes and/or correlate events globally. As an example, modern TIMS U–Pb
geochronology is capable of producing dates with precision and accuracy better than 0.1%
of the age for single crystals with only a few picograms (pg) of Pb. Therefore, it is possible
to constrain the durations of single zircon crystal growth in magmatic systems over tens to
hundreds of kyr in Mesozoic and younger rocks. If these dates and rates can be connected
with other igneous processes such as magma transfer, emplacement and crystallization, then
it becomes possible to calibrate thermal and mass budgets in magmatic systems and evaluate
competing models for pluton assembly and subvolcanic magma storage. As another example,
Sm–Nd geochronology of garnet permits dates with precision better than ±1 million years for
garnets of any age, including multiple concentric growth zones in single crystals. Such zoned
mineral chronology of this common porphyroblastic mineral in metamorphic rocks can be
linked with thermodynamic modeling, permitting constraints on rates, durations, and pulses of
heating, burial, and devolatilization during tectonometamorphic evolution.
As with other techniques used in geochronology, it is important by TIMS to combine the parent
and daughter isotope ratios with complementary geochemical, isotopic, and textural information
in order to interpret those dates within the context of the geologic processes being investigated.
In this volume, focus is directed to the science of understanding the rates of mineral- and rock-
1943-2666/17/0083-0008$05.00 (online)
232 Schoene & Baxter
forming processes, or petrochronology. Generally the goal of petrochronology is to capitalize

on a thermodynamic, geochemical, or mechanical framework to understand metamorphic or
igneous processes. This goal is no different in TIMS compared to other techniques, but TIMS
poses a different set of challenges that make harmonizing textural and geochemical information
with dates more difficult than with in situ techniques. This chapter outlines some of the aspects
of TIMS that are important in doing good petrochronology, and then highlights some examples
from the recent literature that exemplify both conceptual advances and creative workflows. We
focus primarily on the U–Pb and Sm–Nd systems, but acknowledge that other systems, for
example Rb–Sr and Re–Os, have also been exploited to obtain petrochronologic data with the
same goal of understanding the rates of rock-forming and geodynamic processes.
A BRIEF REVIEW OF TIMS GEOCHRONOLOGY

While the details of thermal ionization mass spectrometry will not be covered in this
chapter (see Carlson 2014, for a recent review), we mention the mechanism of sample
ionization because it is the major difference between TIMS and other mass spectrometers
used in geochronology: in short, it is the primary control on both the positive and negative
aspects of TIMS in petrochronology. Sample ionization in TIMS, as the name suggests, is
carried out by heating a sample on a thin metal filament until the target element in question
is either volatilized or ionized (preferably the latter, but ionization efficiencies are typically
< 5%; although recent advances approach or exceed 25% ion yield for NdO+ analysis, e.g.,
Harvey and Baxter 2009), and these ions are accelerated through a focusing lens and into
a mass separator such as a magnet. In contrast, ICPMS provides 100% ionization for all
elements. However, ICPMS provides very low ion transmission from the plasma source to
the analyzer, typically < 1%, whereas TIMS ion transmission is much higher, probably > 50%.
Either method can result in ion beams that are stable for several hours for typical sample sizes,
and through simple counting statistics this can result in high precision isotopic ratios. Once
the ion beam is established, run duration is dictated by the desired analytical precision and is
identical via TIMS or MC-ICPMS. Generally, TIMS outperforms MC-ICPMS for isotopic
analysis when ionization efficiency for a given element exceeds ~2%, especially for smaller
sample sizes. As will be highlighted below, for small sample sizes (defined below) targeted in
petrochronology, minimizing laboratory contamination, or blank, is also important and often
easier by TIMS given complex sample introduction systems associated with solution ICPMS.
There are, however, several drawbacks to creating ions by thermal ionization. One is that
the sample must be transformed into a substance that can be adhered onto, and easily ionized
on, a filament that is a few millimeters wide. The other is that TIMS are generally low mass
resolution instruments, meaning they cannot resolve isobaric or polyatomic interferences
of the same nominal mass (e.g., cannot differentiate between masses 205.9 and 206.0). The
first issue is addressed by dissolving the sample in acid and concentrating it into a small
volume of liquid substrate, selected based on its ability to enhance ionization, then drying
that volume on the filament before loading the sample into the TIMS source. Sample size is
thus limited in this way to that which the geochronologist can physically manipulate with
tweezers or a pipette to transfer the sample into beakers for dissolution.
The issue of low mass resolution is remediated by putting the dissolved sample through
various types of ion separation chemistry that use organic resins capable of separating
different elements based on their chemical properties and the type/concentration of reagent
used (Schönbächler and Fehr 2014). The exact type and volume of resin, in addition to
the reagent recipe, for the procedure depends on the elements requiring separation. For
example, high-precision Sm–Nd geochronology requires separation and isolation of the
rare earth elements (REEs) prior to mass spectrometry due to isobaric interferences that
would obfuscate the measured isotopic ratios (Wasserburg et al. 1981). TIMS also carries
Petrochronology and TIMS 233
the benefit of differentiating certain interfering elements from the desired element due to
the different temperatures at which each element ionizes. This can help to recognize and
mitigate the effects of interfering elements during an analysis though in general the goal of
column chemistry is to remove interfering elements as much as possible.
Because the whole purpose of ion chromatography is to separate elements in the sample
(including the parent from the daughter), the measurement of parent/daughter ratio required
for geochronology is not possible during a single analysis. Instead, in order to still calculate
accurate ages, a technique called isotope dilution (ID) is used. Isotope dilution is the process
by which a tracer (or “spike”) solution is added to the sample, usually prior to dissolution,
that contains concentrated isotopes of the parent and daughter elements (Stracke et al.
2014). Because the ratio of those isotopes in the tracer solution are very well known through
a series of calibration experiments, the concentration of parent and daughter isotopes can be
backed out by measuring the sample isotopes relative to spike isotopes (Wasserburg et al.
1981; Condon et al. 2015; McLean et al. 2015). For some applications, synthetic isotopes
such as 205Pb and 233U and 236U are used, whereas for others natural isotopes are used.
The mathematics for calculating parent/daughter ratios using the differing approaches are
well-defined (Compston and Oversby 1969; Galer and Abouchami 1998; Thirlwall 2000;
Schmitz and Schoene 2007; McLean et al. 2011; Stracke et al. 2014a). An added benefit of
isotope dilution as a means of calculating parent/daughter ratios is that the tracer solution
is homogeneous and analyzed at the exact same time as the sample, thereby eliminating
concerns about variable parent/daughter fractionation between sample and standard. This
point is an important source of uncertainty using in situ techniques that correct both isotopic
and elemental fractionation through sample/standard bracketing (Williams 1998; Horstwood
et al. 2016). In other words, the accuracy of a sample’s date can be tracked directly to the
accuracy to which one knows the composition of the tracer solution.
The process of sample preparation is briefly described above, but in practice is quite
time-intensive. When compounded with long analysis times required for high precision data,
TIMS geochronology is an extremely time consuming process. Furthermore, because the steps
listed above require extensive sample handling and use of reagents, blank is an issue and
for some methods can be a limiting factor in the sample size analyzed. Therefore, sample
dissolution and ion separation chemistry are all done in clean labs, further adding to the cost
and complexity of TIMS geochronology.
There are obvious and not-so-obvious difficulties in doing TIMS petrochronology. By way
of comparison, with in situ techniques sample preparation can be as simple as making polished
thin sections or grain mounts as a way of obtaining petrographic and textural information of
dated minerals. Following thorough optical and geochemical sample characterization, the
target minerals can be directly dated with spatial resolution of < 30 µm (see Kylander-Clark
2017; Schmitt and Vazquez 2017, both this volume). Developments of both conventional
and split-stream LA-ICPMS techniques permit analysis of both U–Pb date and geochemistry
simultaneously, further simplifying the work-flow and removing uncertainties about
comparing dates and geochemistry from different volumes of material (Yuan et al. 2008;
Kohn and Corrie 2011; Kylander-Clark et al. 2013). In TIMS geochronology, the sample
preparation follows some variation of that used for in situ analysis, with the caveat that the
sample to be dated must be physically removed from grain mount or thin section or bulk rock
and dissolved, which requires steady hands and/or precise tools (such as a MicroMill). The
specifics of the workflow needed to achieve petrochronologic characterization vary depending
on the isotopic system and minerals being targeted. As such, the rest of this chapter is divided
into the two isotopic dating methods most commonly used for petrochronology by TIMS,
and within each section the narrative focuses mostly on the workflow considerations while
presenting some results from the literature. The reader is referred to those publications for
more detail on the interpretations and geologic significance of the data.
U–Pb ID-TIMS PETROCHRONOLOGY

Workflows in petrochronology
As broadly outlined above, there are many important steps prior to analysis by isotope dilution
(ID) TIMS geochronology (see Davis et al. 2003; Corfu 2013; Schoene 2014, for additional details
and references). Because most high-U minerals targeted for U–Pb geochronology are accessory
phases (i.e., < 1% volume of rock), typical sample processing begins with crushing followed by
density+magnetic separations to concentrate the mineral(s) of interest, e.g., zircon, apatite, titanite,
monazite, etc. This was an essential step several decades ago when a pure aliquot of several
milligrams of high-U mineral was required for a single analysis. Comparatively, modern ID-
TIMS U–Pb geochronology focuses on single accessory crystals (mostly < 300 µm in diameter) or
fragments of single crystals. Bulk crushing of rocks to obtain large mineral separates is therefore
not necessarily required unless the concentration of accessory minerals is very low, as may be the
case with crystal-poor volcanic rocks. This emphasis on smaller sample size is driven largely by
increased awareness of the complexity of mineral growth in igneous and metamorphic systems.
In theory, the ideal TIMS workflow would not differ much from that commonly pursued by in situ
techniques, namely: 1) identify high-U minerals in thin section, 2) characterize their petrographic
context and geochemistry in relation to rock-forming minerals that contain complementary
information about the system’s petrogenetic history, and 3) remove those same minerals from thin
section for ID-TIMS analysis. However, such an approach is rare in reality due to the challenge of
physically manipulating micrometer-scale mineral domains. Despite these difficulties, the clear
advantage of TIMS temporal precision relative to in situ dating methods underscores its utility
in petrologic studies (e.g., Fig. 1), and researchers have increasingly developed workflows to
complement TIMS geochronology with additional textural and geochemical information (Fig. 2).
Nonetheless, bulk crushing remains an important method of mineral separation for three reasons:
1) it is not uncommon to identify high-U minerals through bulk crushing that are either absent
or rare in thin section, 2) to ensure more accurate representation of the total crystal population
of a sample, from which to then characterize and select crystals for analysis, and 3) it is difficult
to extract small minerals from thin section. Examples from the recent literature of creative
workflows in ID-TIMS petrochronology are given below in the section Linking geochemistry
with U–Pb ID-TIMS geochronology, but first a discussion of limitations imposed by sample size
is presented, as this factor dictates what workflows are possible.
Limits on sample size and precision
ID-TIMS U–Pb geochronology of high-U minerals with low amounts of initial Pb (e.g.,
zircon) is the most precise available geochronological method. Reported age precision for
ID-TIMS has increased dramatically over the past few decades, from the percent level on
large milligram size fractions of minerals to better than 0.1% (2s) on fragments of single
crystals (Schoene 2014). Advances in mass spectrometry have played a pivotal role in this
development, particularly in relation to ion generation, transmission sensitivity, and ion beam
measurement. However, other sources of uncertainty can be far greater than associated with
analytical precision, depending on the age of the sample, the ratio of initial U to Pb, and
laboratory blank. Several recent contributions have discussed these sources in detail (Schmitz
and Schoene 2007; McLean et al. 2011; Schoene 2014); examples are depicted in Fig. 3 for
zircons with different ages and U contents. Because one goal of ID-TIMS petrochronology
is to explore potentially complex and protracted mineral growth histories through dating
increasingly smaller samples, those uncertainties that pose the greatest limitations to sample
size are discussed below. Other sources of uncertainty, such as mass-dependent isotope
fractionation during mass spectrometric measurements (Amelin and Davis 2006; McLean
et al. 2015), corrections for initial daughter product disequilibrium (Parrish 1990; Schärer
1984; Crowley et al. 2007; Schmitz and Schoene 2007), and oxygen isotopic composition of
UO2 (Condon et al. 2015) are independent of sample size but can be the dominant source of
uncertainty for cases where sample size is not a limit on precision (Fig. 3).
Hornblende porphyry Hornblende-biotite porphyry

10CC22 10CC51
39
LA-ICPMS
LA-ICPMS
38
37
Pb/238U Date (Ma)
36
CA-ID-TIMS CA-ID-TIMS
206
35
34
33
Zircon grains plucked out of epoxy mount

and subsequently dated by CA-ID-TIMS
32
Figure 1. Rank order plot illustrating the precision of Chemical Abrasion (CA-)ID-TIMS geochronology
relative to LA-ICPMS U–Pb geochronology. Shown are two examples of 206Pb/238U dates on zircon from
the same rocks, and in some cases for the exact same zircon grains. These data illustrate both the ability of
chemical abrasion to remove zones of Pb-loss and also that higher precision reveals complexity in zircon
growth that is unresolvable with lower precision data. The height of symbols represents 2s uncertainties.
Data from Chelle-Michou et al. (2014).
U+Pb±Th Tracer
2) in situ trace elements:
Zr, Hf, Y, Sc, Ti
REE, Nb, Ta, ±Th
laser or ion beam spots

column with
ACI D ion exchange
3) remove single grain
1) textural imaging or grain fragment resin
by CL or BSE from mount or 4) Dissolution
thin section
5) ion exchange separation
7a) trace elements:
single-collector
solution
trace element aliquot U+Pb±Th aliquot
ICP-MS
with Zr, Hf, Y, Sc,
REE, Nb, Ta, ±Th
7b) Hf isotopes:
solution 6) U-Pb isotopes: TIMS
MC-ICP-MS
Figure 2. Possible workflow for ID-TIMS U–Pb petrochronology.
Pb-blank. The greatest limitation on sample size is the amount of radiogenic Pb in a

sample (Pb*) compared to the non-radiogenic, or common, Pb (Pbc). Pbc integrates Pb taken up
by the grain during crystallization and the laboratory blank. For zircon, there is no quantifiable
Pb included in the crystal structure during crystallization, so laboratory blank is the only
source of Pbc (although some Pbc may be incorporated into zircon during metamictization and
measured measured
U mass measured 238
U/235U measured
Pb/205Pb
206
fractionation 206
Pb/205Pb
206
Pb/204Pb
measured
U/235U
238
measured
206
Pb/204Pb Th diseq.;
230
(Th/U)magma
=4±2
αPb
mass fractionation Pb
αPb = 0.11±0.04 %/a.m.u. (206Pb/204Pb)blank (206Pb/204Pb)blank
=18.30±0.33 =18.30±0.33
North Mt. Basalt zircon ash bed zircon rhyolitic tuff zircon
central atlantic magmatic Triassic-Jurassic boundary Kos volcanic complex
province
200.13 ± 0.31 Ma 0.357 ± 0.008 Ma
200.38 ± 0.05 Ma
Pb*/Pbc = 3.8 Pb*/Pbc = 0.5
Pb*/Pbc = 219
Pbc = 0.5 pg Pbc = 0.4 pg
Pbc = 0.8 pg
Figure 3. Pie charts detailing the variance contributions to the total uncertainty budget of some 206Pb/238U
zircon dates. Modified from Schoene (2014). See text for discussion.
recrystallization (Geisler et al. 2003)). For other minerals, such as monazite, small amounts
of Pbc (typically < 1 ppm) may be incorporated into the crystal lattice. For titanite, apatite and
allanite, among others, 10s to 100s ppm Pbc may be incorporated during crystallization.
Figure 3 shows that for young and/or low U zircons, the correction for Pbc is by far the
largest source of uncertainty for a given U–Pb date (manifested in the correction for the
206
Pb/204Pb ratio of the Pb blank). Figure 4 further quantifies this by showing a compilation
of zircon data from a single study where the Pb*/Pbc varies from ~1 to > 150 as a function of
the sample size (all analyses are < 1 zircon grain), U and Pb* content, and amount of blank
Pb. This figure clearly demonstrates that the precision of an analysis scales with Pb*/Pbc,
with dates becoming exponentially less precise when Pb*/Pbc <~15−20. Laboratories that
perform high precision geochronology of small and/or young zircons typically have Pb
blanks < 1 pg, but strive to remain < 0.5 pg.
Comparatively, in the case of Archean zircons with tens, if not hundreds, of picograms
Pb*, sample size is essentially limited by one’s skill with a set of tweezers. For most
applications, however, there is a trade-off between the volume of sample selected, age, U
content, and routinely-achievable blanks for a given laboratory. The average uncertainty
in the Oligocene zircon dataset in Figure 3 of 0.07% on a 206Pb/238U date translates into
about ±20 ka for a single zircon fragment, which is far smaller than the average range of
zircon crystallization ages for a granitoid sample from that study (i.e., ~500 kyr spread in
dates for a given hand sample). While this implies that even smaller sample sizes could
elucidate further details of zircon growth histories, it also requires lower Pb blanks. The
difference between a 0.6 and 0.3 pg Pb blank in such studies can determine whether one
can successfully resolve temporal variability between grains or detect intra-crystal growth
histories by subsampling and dating multiple fragments from single grains.
A large part of the uncertainty associated with Pbc in dated minerals comes from the
difficulty in constraining the Pbc isotopic composition. Because 204Pb is non-radiogenic and
stable, in ID-TIMS U–Pb geochronology, the moles of 204Pb can be determined through
isotope dilution and the amount of 206Pb, 207Pb and 208Pb from blank can be subtracted from the
radiogenic Pb isotopes by calculating their abundance relative to 204Pb. In the case of minerals
that incorporate relatively high Pbc into their mineral structure (e.g., titanite), the composition
of this initial Pb has been estimated in several ways, including 1) assuming a composition
0.40
0.35
uncertainty of 206Pb/238U date (%)

0.30 Pbc [pg]
n=135
0.25
0 1 2 3 4
0.20
0.15
variability in precision
from other sources
0.10
0.05
0.00
0 50 100 150
Pb*/Pbc
Figure 4. Precision of ID-TIMS U–Pb date as a function of the ratio of radiogenic Pb (Pb*)
to unradiogenic Pb (laboratory blank in this case; Pbc). Data are for single zircon fragments
from (Samperton et al., 2015), and given 2s uncertainties, as a percent of the age. Inset
shows histogram of Pbc values from the same dataset.
based on whole-earth Pb evolution models (Stacey and Kramers 1975), 2) measuring the Pb
isotopic composition of a coexisting low-U mineral, e.g., feldspar (Chamberlain and Bowring
2001), and 3) application of standard isochron methods. The benefits and drawbacks of each of
these approaches are reviewed in detail by Schoene (2014) and the references therein.
For minerals such as zircon where blank Pb is the dominant source of Pbc, knowledge of
that composition can be a limiting factor in the precision of a date, even if blank amounts are
very low. The blank isotopic composition depends on mixtures between the different sources
of blank Pb in the lab. The obvious sources are the reagents added to the sample during grain
dissolution, ion separation chemistry, and filament loading, in addition to Pbc ionized from the
filament, diffused from Teflon during dissolution, or not removed from the surface of the sample
during pre-cleaning steps. Because the amounts of each source will all vary slightly between
analyses, a typical approach to estimating the composition of these mixtures is to analyze “total
procedural blanks”, where a dissolution capsule is left empty, spiked with tracer solution, and
treated identically to normal samples. The standard deviation of these measurements should
characterize the range of possible blank compositions for samples, and this uncertainty can be
propagated into dates (Schmitz and Schoene 2007; McLean et al. 2011). Figure 5 shows the
measured isotopic compositions of ~600 Pb blanks in the Princeton ID-TIMS laboratory over
a 3 year period. Most of these analyses were measured with the intention of calculating an
accurate mass of Pb, not the Pb isotopic composition, and thus emphasized speed over quality.
For example, it is observed in many TIMS labs that isobaric interferences are present under
all Pb isotopes at low temperature, and that as the filament is ramped to higher temperatures
these interferences “burn off”. This effect takes time, and while ensuring that interferences are
burned off is not crucial for determining the amount of Pb blank in an analysis, it is crucial
for determining the Pb blank isotopic composition. Therefore the accuracy of the isotopic
composition is questionable in the majority of measurements shown in Figure 5. Two datasets
16.5
16.3
16.1
15.9
207Pb/204Pb
15.7
15.5
15.3
15.1
all data (n= 586)

14.9
subset 1 (n=10)
14.7 subset 2 (n=20)
14.5
16 16.5 17 17.5 18 18.5 19 19.5 20 20.5
206Pb/204Pb
Figure 5. Compilation of measurements of the isotopic composition of Pb from 2013–2016 in the Princeton
ID-TIMS laboratory. Gray circles represent all data, including measurements not meant to target the blank
isotopic composition. These analyses were done quickly with less care taken to allow low temperature
isobaric interference “burn off”, noting these interferences are observed to decrease through an analysis
until high quality data is obtained. Black squares (Samperton et al. 2015) and white triangles (Barboni et al.
2017) are two examples of total procedural blanks measured by a single user over the course of one or two
studies. Mean values and 1-standard deviation are: 206Pb/204Pb = 18.42 ± 0.18, 207Pb/204Pb = 15.48 ± 0.17 for
(Samperton et al. 2015) and 206Pb/204Pb = 18.44 ± 0.18, 207Pb/204Pb = 15.53 ± 0.18 for (Barboni et al. 2017).
Uncertainties (1RSE) for individual data points are approximately the symbol size.
are also shown in Figure 5 wherein analyses were carefully performed over individual studies
and run identically to typical samples; these analyses were used to estimate the Pb blank
composition. Typically the uncertainty of a given blank analysis is far less than the variance
observed in blank isotopic composition of multiple measurements; thus, while measuring more
blanks results in a more accurate description of that variance, it does not reduce the uncertainty
in the correction. In other words, measuring more blanks over the course of a study increases the
accuracy, but not necessarily the precision of the resulting U–Pb dates. Reducing the variance
in Pb blank composition, and therefore reducing the uncertainty in the blank subtraction, is an
important way to increase precision on dates for samples with low Pb*/Pbc.
Analytical precision. Another limitation on sample size relates to the ionization of Pb
and U in the mass spectrometer. While analytical precision would appear to be of secondary
importance based on Figure 3, this is primarily because those examples were chosen to
highlight the importance of other sources of uncertainty. For the zircon to the left in Figure 3,
using a double-Pb, double-U isotopic tracer (e.g., the EARTHTIME 202Pb–205Pb–233U–235U
tracer; Condon et al. 2015; McLean et al. 2015) would eliminate the uncertainty of Pb mass
fractionation during mass spectrometry, and in that case the measured 206Pb/205Pb would be
the largest contributor to the uncertainty in the date. Figure 6 compiles data from the Princeton
ID-TIMS lab between 2012–2016 for zircons with a range of intensities of 205Pb and variable
206
Pb/205Pb ratios. The data were measured on an IsotopX PhoeniX62 TIMS on a Daly
photomultiplier ion counter, all with ~10 pg 205Pb. While these data encompass a range of
N-values (i.e., number of ratios measured, or total integration time) and total ion yield (average
intensity is plotted), they form a roughly linear cloud in log–log space, where precision
increases as a function of increasing intensity and increasing sample/spike Pb (represented as
the 206Pb/205Pb, where 205Pb is the tracer isotope). The plot also shows the precision of uranium
analyses for the same grains. Uranium was measured as UO2+ on three static Faraday cups
3
Pb measured peak hopping on ion counter U measured static on faraday cups
2.5
10 -1
UO2/233UO2 or 206Pb/205Pb ratio

10 -1
Standard error 238UO2/233UO2 (%)

Standard error 206Pb/205Pb (%)
1.5
-2
10
1
-2
10
238
0.5
10 -3 0
10 4 10 5 10 2 10 3
Beam intensity 205
Pb (cps) Beam intensity 233UO2 (mV)
Figure 6. Demonstration of analytical precision of Pb and U isotopic ratios as a function of sample size. Data
are from routine measurements with the EARTHTIME ET535 or ET2535 tracer in the Princeton ID-TIMS
laboratory between 2012–2016, and were chosen to include a wide range in sample/spike ratio and beam
intensities. All measurements performed on an IsotopX PhoeniX TIMS by peak-hopping on a Daly photomul-
tiplier ion counter for Pb and as static measurements with 1012 ohm amplifier boards on Faraday cups for U.
Each sample has ~1–1.5 ng of 233U and 10–15 pg of 205Pb. Datasets include ~50–100 ratios (250–500 second
integration on each isotope) for Pb and 100–300 ratios (500–1500 second integration) for U. Grayscale on
datapoints represent the sample/spike ratio, with key at right. Beam intensities are reported as the average
intensity over the entire analysis. Uncertainties (vertical axes) are reported as 1RSE. See text for discussion.
on 1012 ohm resistors. Although each measurement contains ~1 ng of 233U, larger scatter in
the measured precision as a function of intensity and 238UO2/233UO2 compared to Pb reflects
less predictable ionization for U (i.e., more variable N or total integration time). For very low
238
UO2/233UO2 or particularly poorly-ionizing samples, UO2 can also be measured on an ion
counter with similar results to Pb. A couple take-home points can be made from Figure 6. One,
not surprisingly, is that precision is improved as more ions are extracted from the filament and
measured. Most labs mix their sample with a slurry of colloidal silicic acid and phosphoric
acid, which has been shown to increase ionization (Gerstenberger and Haase 1997) compared
to loading the sample on the filament with other materials. Variability in ionization using this
technique is large from sample-to-sample and user-to-user, resulting in much of the dispersion
in Figure 6. Regardless, ionization for both Pb and U (typically calculated by summing the
total number of ions measured relative to the calculated sample size) is typically ≤ 5% of the
total sample on the filament. There is, therefore, an opportunity for drastic improvements
in ionization, which would decrease the amount of sample needed for ID-TIMS U–Pb
geochronology and permit examination of increasingly smaller mineral domains.
Chemical abrasion. As with any geochronological technique, assessing open-system
behavior is of primary importance. In the U–Pb system, for minerals older than a couple hundred
millions of years (Ma), discordance between 206Pb/238U and 207Pb/235U dates can be used as
a means of identifying whether initial Pb, or loss/gain of Pb or U, have affected a mineral’s
isotopic systematics (Wetherill 1956; Corfu 2013). This information can be used to assess the
accuracy of a given date. Of all the possible sources of discordance in the U–Pb system, Pb-
loss has been the prime suspect, and great measures have been undertaken to both understand
and remediate this problem (Corfu 2013; Schoene 2014). In U–Pb thermochronometers such
as titanite, apatite, and rutile, Pb-loss can be used advantageously to constrain the thermal
history of rocks (Blackburn et al. 2011; Smye and Stockli 2014; note however that for titanite
and rutile there is ongoing discussion about diffusion kinetics; e.g., Kohn 2016; Kohn and
Penniston-Dorland 2017, this volume). In minerals with negligible rates of Pb diffusion under
most crustal conditions, however, little has been learned in attempts to interpret Pb-loss;
instead, the overwhelming goal has been to exclude from analysis crystal domains affected
by Pb-loss. Following decades of attempts, Mattinson (2005) presented a technique called
chemical abrasion TIMS (CA-TIMS), in which zircon grains are annealed at > 900 °C for
48−60 hours and then leached with HF acid in order to remove domains compromised by
Pb-loss. In short, this technique has revolutionized ID-TIMS zircon U–Pb geochronology
such that almost every lab has adopted some form of this method and adapted it for single- or
sub-grain zircon analysis (see Mundil et al. 2004, Schoene et al. 2006) for early examples,
but note that all recent U–Pb ID-TIMS papers employ CA). CA-TIMS has taken huge steps
in eliminating discordance in Archean zircon and improved accuracy in dating Phanerozoic
zircons as well. Notably, attempts to extend CA-TIMS to minerals other than zircon have
not yet been successful (Rioux et al. 2010; Peterman et al. 2012). In CA-TIMS, an analyst
will typically leach single pre-selected grains, perform several cleaning steps, spike with an
isotopic tracer, and then dissolve the whole grain. An important point about this workflow for
petrochronology is that, due to the leaching step, no information is available about what parts
of the grains are dissolved. Polishing and imaging of zircon that has experienced chemical
abrasion shows that the leaching procedure is complicated and does not simply attack the
outermost zircon layers (Mundil et al. 2004; Mattinson 2011). Hydrofluoric acid apparently
penetrates zircon along fractures and attacks discrete domains, often burrowing into grains
in unpredictable, asymmetric ways. It is therefore not straightforward to connect zircon
textural or geochemical information obtained prior to chemical abrasion with the age that
is eventually measured. This “information decoupling” is similar to the practice of many in
situ workflows that, e.g., involve measuring age from one laser spot and geochemistry ± Hf
isotopes from a nearby spot in the grain. While it is not unreasonable to associate such data,
there is potentially nontrivial uncertainty in such a correlation. For in situ methods, the LASS
method helps remediate this problem (Kylander-Clark 2017, this volume); for TIMS, the
TIMS-TEA method, described below, aims to achieve similar goals.
Linking textures with dates in ID-TIMS U–Pb geochronology
In situ attempts. In situ geochronological methods have revealed that linking dates to
internal mineral textures and petrographic context can be highly useful in interpreting such
dates in terms of P–T paths in metamorphic rocks and mineral growth histories in both
metamorphic and igneous rocks (see Rubatto 2017; Schaltegger and Davies 2017, both this
volume). ID-TIMS has benefited from these insights but is faced with the obvious challenge
of physically isolating grains targeted for geochronology prior to dissolution. There have
been some notable examples where grains have been removed from thin sections and dated
by ID-TIMS following characterization in thin section.
One such example is that of (Lanzirotti and Hanson 1996), who characterized two
populations of monazite in metapelites from the Appalachian-aged Wepawaug Schist in
Connecticut, USA. They were able to characterize multiple populations of monazite on the
basis of morphology and geochemistry, and also link their growth to peak versus retrograde
metamorphic conditions. By simply plucking the monazites out of thin section and dating
them by TIMS, they were able to show that the populations differed in age by ~30 Ma, thereby
placing robust time constraints on the retrograde P–T path.
More recently, Corrie and Kohn (2007) used a microdrill to remove cores of monazite
grains from thin section and then date them by TIMS. The monazite forms part of a Barrovian-
type assemblage in the Great Smoky Mountains, USA, and was demonstrated to have
grown near the staurolite-in isograd during the Taconian orogeny. Because the monazites
were relatively large (> 200 µm diameter) and high-U, Corrie and Kohn (2007) were able
to drill out and analyze < 50 µm diameter pieces of monazite that had been characterized
petrographically. One of the challenges of this is the possibility of contamination during
drilling, since monazite has a high acid solubility and is difficult to clean. Regardless, this
approach to linking TIMS dates to metamorphic reactions remains underexploited.
Although not carried out in thin section, Barboni and Schoene (2014) were able to
link zircon dates to their setting in the rock by extracting zircon inclusions from K-feldspar
megacrysts in the 7 Ma Mt. Capanne pluton, Elba, Italy. By cutting megacrysts into core
and rim pieces and then crushing them, they were able to build a stratigraphy through
the K-feldspar with uncertainties on U–Pb dates of only a few thousand years (Fig. 7).
Using logic similar to that of a detrital zircon study in sedimentary rocks, they showed
that youngest zircons from the megacryst cores are tens of thousands of years older than
the rims, and this duration is therefore the maximum growth interval of the K-feldspar. By
combining U–Pb data with observations of K-feldspar inclusion suites and thermodynamic
phase modeling, they argued that megacryst growth occurred in an upper crustal magmatic
mush, thus placing time constraints on upper crustal cooling and pluton crystallization.
7.6
K-feldspar megacryst
2 cm
Pb/238U date (Ma)
Core
7.5
core to rim
growth =
Rim
32 ± 14 ka
7.4
zircon inclusions
206
from K-spar core
youngest zircon
zircon inclusions
in host rock from K-spar rim
7.3
Figure 7. Determining crystal growth rates with ID-TIMS U–Pb zircon geochronology. In this example, K-
feldspar megacrysts were removed from the host granotoid, the Sant Andrea phase of the Mt. Capanne pluton,
Elba, Italy. Zircon inclusions from within K-feldspar megacrysts were removed by bulk crushing after slicing
up K-feldspar into core and rim domains. 206Pb/238U dates from individual zircons are plotted in a rank order
plot. The youngest zircon date from the host granitoid is shown to be identical to the youngest K-feldpsar
rim zircon date; core to rim growth duration is thus a maximum. Data are from Barboni and Schoene (2014).
Ex situ attempts. It is now common to characterize internal textures of minerals

following bulk mineral separation. This can give insight into whether minerals have igneous
or metamorphic origins, or whether igneous or metamorphic cores of grains are overgrown
by later episodes of mineral growth. Interpreting these textures has been important for
attaching significance to dates but remains a qualitative and imperfect tool. Nonetheless,
knowledge regarding whether a date isolates or mixes visible growth domains within a
mineral is an important first-order petrochronologic observation. Textures within these
domains can give information about the igneous or metamorphic origin of mineral growth;
the nature of the contacts between crystal domains also provides information regarding
periods of dissolution, reprecipitation, or recrystallization.
In U–Pb ID-TIMS geochronology, linking internal mineral textures with dates is

most commonly done by imaging minerals in grain mount and then removing and dating
them as whole grains or fragments (Crowley et al. 2006; Matzel et al. 2006; Gordon et al.
2010; Rivera et al. 2013). Although admittedly crude, it is possible to break grains near
or along boundaries between different growth domains and date them in order to isolate
the timing of growth of different sub-domains. There are, of course, large uncertainties in
how observed domains project into the third dimension of the grain and therefore into the
sampled volume. Microsampling previously imaged minerals is not yet routine in most
labs, but is becoming increasingly common and in every case having textural information
that can be linked to dates is better than the alternative.
Linking geochemistry with U–Pb ID-TIMS geochronology
In situ geochemical analysis prior to TIMS. An extension of the petrographic and textural
characterization techniques described above involves carrying out geochemical analyses in
thin section or grain mount and then microsampling grains for geochronology. Depending
on the concentration of the elements or isotopes measured, the analyses can be performed by
electron probe microanalysis (EPMA), LA-ICPMS or SIMS. In the case of LA-ICPMS care
must be taken not to ablate so much sample such that the remainder is too small to analyze by
TIMS. The goal is to combine geochemical information with age information in a way that
can be related back to the composition of the liquid from which the minerals crystallized or
identifying equilibrium mineral assemblages through application of partition coefficients.
There are not many examples of this approach, but the increasing accessibility of in situ
techniques and the number of TIMS labs capable of measuring small amounts of Pb will
surely lead to more common implementation. Crowley et al. (Crowley et al. 2007), working
on zircon from the Bishop Tuff, California, measured U and Th concentrations and produced
CL images by EPMA prior to removing grains from grain mount for U–Pb analysis. U
and Th increased by up to a factor of 5 across grain transects. In addition to providing
information about magma evolution during zircon growth, this approach provided a means
to compare bulk U and Th concentrations obtained from the TIMS analysis with finer scale
zonation that is obscured by sampling whole grains or fragments.
Rivera et al. (2014) used LA-ICPMS followed by TIMS U–Pb zircon geochronology
on the same grains to investigate the magmatic history of the precursor magma that fed the
eruption of the Huckleberry Ridge Tuff at ca. 2.1 Ma. By measuring the europium anomaly in
zircons (a proxy for fractional crystallization of plagioclase and sanidine within the magmatic
system), and combining this with Ti-in-zircon measurements to investigate temperature,
they were able to model the crystallinity in the magma as a function of temperature (Fig. 8).
These same zircons were then removed from grain mount and dated by TIMS. The results
were used to texturally and geochemically identify cores of zircons that predated the main
phase of emplacement and residence of the magma in the upper crust by up to 120 ka. The
combination of Ti-in-zircon, geochemistry, and geochronology showed that magma residence
and differentiation in the upper crust predated eruption by only tens of ka, arguing for high
magma flux and short upper crustal residence leading to the Huckleberry Ridge supereruption.
Similarly, Samperton et al. (2015) conducted trace element transects across zircon grains
to look at the relative geochemical evolution during zircon growth in a suite of samples from
the Bergell Intrusion, Italy/Switzerland. These data were used to argue that relatively gradual
changes in zircon trace elements reflected evolving magma composition over several hundred
kyr due to fractionation of both major and accessory phases in the magma. Microsampling of
the grains provided up to 5 analyses from single zircons with individual uncertainties as low as
±20 ka, permitting evaluation of growth timescales of individual zircons within a geochemical
framework. A remaining uncertainty in quantitatively relating trace elements in zircon to
HRT TIMS ages (all)

860
A HRT TIMS ages (young)
HRT 40Ar/39Ar age (young)
840
SRB TIMS ages (all)
relative probability
820
Ti-in zircon TºC
800
780
760
HRT SRB
740
CD1
720 CD2
700 CD3
Figure 8. LA-ICPMS + ID-TIMS
680
2.04 2.08 2.12 2.16 2.20 2.24 2.28
U–Pb petrochronology in the
Age (Ma) Huckleberry Ridge Tuff (HRT) and
900 Snake River Butte rhyolite (SRB)
880 B (Rivera et al. 2014). (A) Tempera-
860
ture versus 206Pb/238U age from
zircon fragments that were first
Ti-in zircon TºC
840
measured for trace elements by LA-
820 ICPMS in situ, then removed from
800 Grain interiors grain mount and dated by ID-TIMS.
780 CD1, etc., are manual groupings
760 based on zircon geochemistry. (B)
740
Geochemistry of zircon core–rim
pairs that were subsequently dated.
720 Core-rim tie line
Eu/Eu* is the europium anomaly.
700 (C) Examples of zircon CL images,
680
0 0.05 0.10 0.15 0.20 0.25 0.30 0.35 0.40 textural interpretations, and LA
Eu/Eu* spots, color coded by geochemical
Truncations domain to those in A.
C Dark
bands
43
1 }
42 26
53
2 41 }
27 54
Overgrowths Melt
Inclusion
that of the liquid, while not unique to TIMS petrochronology, derives from limited partition
coefficient data under a range of pressures, temperatures, and magma compositions. More
experiments are needed to help address the paucity of partitioning data with respect to such
intensive variables for zircon and other high-U minerals used for petrochronology.
Another important uncertainty remains in relating trace element geochemistry determined
in situ on the surface of a polished grain mount to dates derived by ID-TIMS on grains removed
from that mount: in other words, one is left comparing two-dimensional geochemical data to
three-dimensional chronological data. In every type of analysis, be it in situ or whole-grain, the
trace element and age information obtained are integrations of the analyzed volume. If those
volumes are not identical, then it is difficult to quantitatively compare the results.
Indirect in situ analysis. One of the parameters that falls out of U–Pb geochronology is
the Th/U of the grain. While in U–Th–Pb analysis this can be directly determined by 232Th/238U
measurement, a model Th/U can be calculated without measuring Th through measurement of
the 208Pb/206Pb and assuming concordance between 206Pb/238U and 208Pb/232Th dates. Therefore, it
is common to plot the Th/U of a suite of minerals versus the measured dates, and this has been
used to distinguish between metamorphic versus igneous grains and/or evaluate the petrologic
implications of evolving Th/U during crystallization. Given this information, it is also possible
to measure Th/U of datable minerals in thin section and correlate those data indirectly with Th/U
derived from the isotopic measurements. As an example of this approach, Oberli et al. (2004)
measured major and trace element geochemistry of magmatic allanite by EPMA. They were
able to relate the Th/U measured by EPMA with that determined through U–Th–Pb isotopic
analysis, and by inference could then relate the major element geochemistry to the calculated
U–Th–Pb dates. They conclude that allanite grew in the presence of an evolving silicate liquid
composition over a period of 5 Ma following intrusion of the studied magma pulse.
Geochemistry and dates from the same dissolved minerals. As described in the section
A brief review of TIMS geochronology, an essential part of TIMS geochronology is that target
elements in the sample must be separated from other elements that may interfere with the
masses measured. In the cases discussed in this section, U, Th and Pb can be separated
quantitatively from the other elements in the dissolved minerals. Though often overlooked, it
is possible to retain the portions of the sample that are not analyzed for geochronology—the
so-called “wash” solution—and these can then be measured for geochemistry or isotopes.
The benefit of this approach over in situ analysis prior to geochronology is that it compares
the dates and geochemistry and/or isotopes from the exact same volume of dated material.
Thus, the average geochemical composition should in most cases correspond to the moment
in time measured for average growth age over that volume of mineral (Samperton et al. 2015).
A relatively common example is that of Hf in zircon. Given that Hf is present in zircon
at the weight percent level, it is possible to measure the Hf isotopic composition of the wash
solution by multicollector ICPMS. The first published example was in Amelin et al. (1999),
carried out on a set of Hadean Jack Hills zircons, in order to combine high-precision dates with
their corresponding Hf isotopic values to gain insight into the presence or absence of depleted
early-Earth reservoirs. Subsequent investigations have used measured Hf isotopic values to
look at the importance of magma mixing and assimilation during the crystallization history of
igneous zircons (Schaltegger et al. 2002, 2009; Schoene et al. 2012; Broderick et al. 2015),
as well as the sources of silicate fluids in metamorphic zircon growth (Crowley et al. 2006).
Because recovery of Hf in the wash solution is very close to 100%, isotopic fractionation
during ion separation chemistry is not a concern. In some cases, the wash solution has been
put through a second stage separation to remove REEs from the Hf fraction, given isobaric
interferences such as 176Yb on 176Hf. Alternatively, this interference can be corrected for by
measuring 175Yb, as is routinely done during LA-ICPMS Hf analysis (Woodhead et al. 2004;
Fisher et al. 2014; D’Abzac et al. 2016).
Similarly, given the sensitivity of modern ICPMS instruments, it is possible to measure
the trace element geochemistry of the wash solutions as well for single minerals or fragments
of minerals (Schoene et al. 2010). This technique, dubbed TIMS-TEA (for Trace Element
Analysis), is simply a downsizing of previous work on larger aliquots of dated minerals (Heaman
et al. 1990; Root et al. 2004), and thus usable for modern ID-TIMS U–Pb geochronology where
an emphasis is placed on shrinking sample size to isolate heterogeneities. Schoene et al. (2010,
2012) applied this method to zircon and titanite from igneous rocks ranging in composition
from gabbroic to granitic and also to a granulite-derived zircon. Those studies showed that
elemental fractionation during ion separation chemistry was less than a few percent and
that Hf isotopes could be measured from the same aliquot as the other trace elements. Trace
element data from Schoene et al. (2012) served several purposes. The first was to identify
zircons that are similar in age but inherited from other batches of magma (sometimes called
antecrysts; Miller et al. 2007)). The second was to show that zircon and titanite geochemistry
reflects the geochemistry of the liquid from which these minerals crystallized. The third was to
show that changes in trace element signatures of the accessory minerals as a function of time
record the timescales of magma differentiation processes such as assimilation, magma mixing
and fractional crystallization. Those data, from the Adamello Batholith, Italy, show that the
timescales associated with these processes are from 104−105 years, and therefore require the
precision of TIMS U–Pb geochronology to resolve them.
Samperton et al. (2015) used TIMS-TEA of imaged and microsampled zircons to
investigate the timescales of magma differentiation in the Bergell Intrusion, Italy/Switzerland.
They combined this approach with in situ determination of trace elements by LA-ICPMS to
compare analytical approaches and to learn more about both high spatial resolution relative
changes in trace element geochemistry and lower spatial resolution (i.e., zircon crystal- and
fragment-scale) trace element evolution in absolute time (Fig. 9). One of the purposes of
obtaining in situ trace elements is to answer the question of whether the TEA data record
mixing of core versus rim signatures of zircon, given the crude nature of microsampling zircons
prior to TIMS analysis, or whether trends in time exhibited by TEA data record protracted
evolution of magma geochemistry. In this particular case, it was shown that tonalites from the
Bergell recorded zircon crystallization in a differentiating liquid during post-emplacement
cooling; granodioritic zircon preserved inheritance from slightly older tonalitic magma pulses,
which was subsequently overgrown by zircon that tracked liquid compositions. The main point
to make here is that it seems necessary to combine CL imaging with both in situ and TEA
geochemistry to fully understand of zircon growth and magma evolution in these systems.
However, the agreement observed in studies with both TEA data and in situ geochemical
analyses in both igneous and metamorphic rocks (DesOrmeau et al. 2015; Samperton et al.
2015) show that either approach may be appropriate depending on the questions that are asked
(Schoene et al. 2010; DesOrmeau et al. 2014; Broderick et al. 2015; Deering et al. 2016).
A Δtz = 178 ± 34 kyr B LA-ICPMS

geochemistry 20
30.89 ± 0.03
Hf ×103 (ppm)
31.03 ± 0.02
8 30.85 ± 0.02 15
7 6 5 4 3 2 1
#1 #3
#2 10
2σ
50 μm core rim 5
18
C TIMS-TEA geochronology+geochemistry
Hf ×103 (ppm)
16 tonalites granodiorites
hybrids
14
12
2σ
10
8
32.0 31.8 31.6 31.4 31.2 31.0 30.8 30.6 30.4
CA-ID-TIMS 206Pb/238U zircon date (Ma)
Figure 9. U–Pb TIMS-TEA, illustrating combined LA-ICPMS/TIMS workflow. (A) CL image of zircon.
Circles show where geochemical data was obtained by LA-ICPMS. Grain fragments #1–#3 were then
extracted by breaking grain along dotted lines. Fragments were then dated by CA-ID-TIMS, and dates
are indicated with 2s uncertainties. Dtz is the minimum growth duration of the crystal. (B) LA-ICPMS
geochemical transects across multiple zircons, including that shown in A. Plotted as relative core to rim
transects. (C) Geochemical and age data determined by TIMS-TEA for seven hand samples ranging in
composition from tonalite to granodiorite. Datapoints represent dates from single zircon fragments as in A,
and geochemistry was measured on the same volume of dissolved zircon by solution ICPMS. Data come
from the Bergell Intrusion, Italy/Switzerland, and are from Samperton et al. (2015). See text for discussion.
SAMARIUM–NEODYMIUM ID-TIMS PETROCHRONOLOGY

While the first Nd isotopic measurements were conducted in the mid-1970’s (e.g., DePaolo
and Wasserburg 1976; Lugmair et al. 1976; Richard et al. 1976), the earliest mineral-specific
geochronology using Sm–Nd can probably be traced to 1980 focusing on the mineral garnet (van
Bremen and Hawkesworth 1980; Griffin and Brueckner 1980). Like U–Pb in zircon, garnet has an
unusually high Sm–Nd (parent to daughter) ratio making it amenable to Sm–Nd geochronology.
Since then, the Sm–Nd system has also been used to constrain the mineralization of other
relatively high Sm/Nd minerals such as scheelite (e.g., Bell et al. 1989), fluorite (e.g., Chesley
et al. 1991), apatite (e.g., Rakovan et al. 1997), carbonate (e.g., Peng et al. 2003; Henjes-Kunst
et al. 2014), and rare earth rich xenotime (e.g., Thoni et al. 2008) and eudialyte (e.g., Sjöqvist
et al. 2016). Unlike U–Pb in zircon (where U/Pb ratio often exceeds 100), the Sm–Nd ratio in
garnet (or any high Sm/Nd mineral) is too low (clean garnet has 147Sm/144Nd generally between
1.0 and 5.0; other minerals rarely exceed 1.0) for us to ignore “common neodymium” or initial
143
Nd/144Nd. Rather, all Sm–Nd petrochronology is isochron geochronology, requiring precise
analysis not just of the mineral of interest but also of a second point (or points) with low Sm/Nd
to pin the isochron and calculate an age. The reader is referred to other papers describing the
choice of second point(s) on the isochron and associated age derivation (e.g., Baxter and Scherer
2013; Baxter et al. 2017, this volume), though most often it is appropriate to use the whole rock
or matrix surrounding the garnet (or other mineral) as the anchor point for an isochron. Here,
we review the use of TIMS to extract precise and accurate Sm–Nd isotopic data needed for
precise geochronology, with potential application not just to garnet, but to any relatively high
Sm/Nd mineral. Then, we turn our focus to garnet-specific workflows and analytical advances,
including the advent of chemically contoured micromill-TIMS of zoned crystals. Most garnet
petrochronology combines an Sm–Nd age (and/or Lu–Hf age, which would be gained through
MC-ICPMS analysis) with any number of complementary petrologic, geochemical, or textural
constraints (see Baxter et al. 2017, this volume). On the other hand, whether for garnet or for
other dateable minerals such as zircon or monazite, Nd-isotopic measurements (i.e., eNd(init))
can themselves provide valuable petrologic or tectonic context for “petrochronology” (e.g.,
zircon, monazite, allanite; Amelin 2004; McFarlane and McCulloch 2007; Iizuka et al. 2011).
Such work has utilized LA-ICPMS when Nd concentrations are high enough, though TIMS can
provide data on materials yielding much smaller amounts of Nd (especially zircon; e.g., Amelin
2004). This review of TIMS Sm–Nd methodologies and capabilities is thus motivated by both
the petrologic and chronologic aspects of petrochronology.
Why ID-TIMS Sm–Nd for Petrochronology?
Sm–Nd can be analyzed with high precision by TIMS or by MC-ICPMS. However, the
main advantage of TIMS is the ability to maintain such high precision (and accuracy) even
as sample size decreases. Part of the reason involves the option to ionize Nd as the oxide,
NdO+, first recognized in the earliest days of Nd-isotope geochemistry (e.g., Lugmair et al.
1976; DePaolo and Wasserburg 1976). Numerous methods have been developed to maximize
ionization including the use of an oxygen bleed valve (e.g., Sharma et al. 1995), and various
loading activator solutions (e.g., Thirlwall 1991; Griselin et al. 2001; Amelin 2004; Chu et
al. 2009; Harvey and Baxter 2009). For example, Harvey and Baxter (2009) employed a
Ta2O5 powder slurry in dilute phosphoric acid as an activator solution. Without the need for
an oxygen bleed valve, source pressures remain low during analysis (about two orders of
magnitude lower than when using the bleed valve). Calculated total ionization efficiency using
the Ta2O5 method are as high as ~30%, representing a major boost over more common TIMS
(or MC-ICPMS) analysis of Nd metal where ion yields are < 10%. Given the unavoidable
sample attrition during focusing, this total ionization efficiency is very close to the theoretical
maximum possible value. 143Nd/144Nd precision via TIMS NdO+ analysis using the Harvey
and Baxter (2009) loading solution yields 10 ppm 2RSD for 4 ng loads, and 28−37 ppm 2RSD
for 400 pg loads of a standard solution (Fig. 10). In practice, most natural samples that are
dissolved and run through columns yield precision (internal 2RSE) 2 to 5 times worse than
pure standards (external 2RSD; Fig. 10) for loads of at least 100 pg. For loads smaller than
100 pg, performance of natural samples worsens further in most cases. This familiar and
frustrating difference between standard solution and natural sample performance in any lab
likely results from loading technique, and additional non-Nd material coming from the column
chemistry that may inhibit or destabilize ionization of Nd. Reducing the performance gap
between standard solutions and dissolved samples remains a frontier for TIMS analysis.
NdO+ analysis also requires correction for interferences from the various oxide species
formed when a particular Nd isotope combines with 16O, 17O, or 18O. In practice, this
interference correction is quite straightforward. Oxygen isotope corrections should be made
first, beginning with lowest mass Nd isotope (142Nd) up to the highest mass (150Nd). This
therefore requires measurement of 140Ce and 141Pr during analysis (even though good column
chemistry should have eliminated these interfering elements, see below). Harvey and Baxter
(2009) ignored the 140Ce correction as it has negligible effect on the crucial measurement of
143
Nd/144Nd needed for traditional 147Sm–143Nd geochronology. Note, however, that failure
to measure and correct for 140Ce oxide interference can lead to spurious 142Nd abundances;
as petrochronologic applications have now arisen that may require ultrahigh precision on
142
Nd/144Nd as a tracer of fractionated reservoirs in the early Earth (e.g., O’Neil et al. 2008),
the analytical method would need to be modified to include measurement of 140Ce. The oxide
correction also requires an accurate and precise knowledge of the oxygen isotopic composition
during analysis. Published “natural” oxygen isotope compositions are not sufficient; rather this
must be carefully calibrated on the TIMS using monoisotopic REE solutions such as pure 141Pr
or 150Nd spike (e.g., Baxter and Harvey 2009). Baxter and Harvey (2009) also showed that the
oxygen isotope reservoir was large enough not to be fractionated during typical runs.
10000
NdO standard 10^11
Nd/144Nd precision (ppm)
NdO standard 10^12
Nd standard 10^13
1000
natural samples
100
143
10
0.010 0.100 1.000
Nd load size (ng)
Figure 10. Analytical precision of 143Nd/144Nd measured by TIMS vs. Nd load size. Large symbols rep-
resent external precision (2RSD) of repeat loads of a pure Nd standard solution. Filled symbols are for
samples run as NdO+ following Baxter and Harvey (2009). Filled circles and triangle were run in dynamic
analysis on an Isotopx Phoenix TIMS with 1011 and 1012 ohm amplifiers, respectively. Open symbols are
for samples run as Nd-metal in static analysis with 1013 ohm amplifiers on a Triton TIMS (from Koornneef
et al. 2014). Small diamond symbols represent internal analytical precision (2RSE) of natural samples
(mostly from tiny garnets) from which Nd was extracted via column chemistry and run as NdO+ with
1011 ohm amplifiers on a Triton TIMS following Baxter and Harvey (2009) by Kathryn Maneiro (Maneiro
2016). Solid line is a simple statistical projection of expected precision based on the 10 ppm precision of
the 4 ng Nd standard solution. Dashed line is five times higher than the statistical projection of the standard.
Final Nd isotope ratios must be normalized to account for TIMS mass-dependent fractionation.
The accepted normalizing ratio is 146Nd/144Nd = 0.7219. Normalization should be done after
oxygen isotope and interference corrections have been performed. As discussed above, ID-TIMS
involves the addition of an enriched isotopic tracer or “spike” for precise measurements of Sm
and Nd. Typically, 150Nd is the Nd spike, and 147Sm or 149Sm is used for Sm spike. By using a well-
calibrated mixed spike (e.g., 147Sm and 150Nd) added to and fully equilibrated with a dissolved
sample before column chemistry, any inaccuracies in absolute concentration of Sm and Nd will
perfectly correlate thus resulting in accurate Sm/Nd ratios. With spiked samples, normalization
and spike subtraction (to achieve both accurate concentration and accurate 143Nd/144Nd ratios)
must be done iteratively. Sm ID-TIMS analysis is done separately on its own filament; either a
double/triple Re filament, or on a single Re or Ta filament run as the metal.
Recent years have witnessed attempts by TIMS manufacturers to utilize high resistance
amplifiers which should theoretically boost signal to noise ratio for smaller beams, further
improving precision for the smallest sample sizes (e.g., Koornneef et al. 2013, 2014; Tappa
et al. 2016). However, these higher resistance amplifiers (1012 and 1013 ohm) may require a
longer decay time during analysis leading ultimately to loss of much of the analyzable signal
as compared to typical 1011 ohm resistors. While still few published data from higher ohm
amplifiers exists, it remains unclear how much improvement will ultimately manifest from
such amplifiers (Fig. 10). This is particularly true when considering the other limitations and
sources of error (such as the blank, see below) that may dwarf TIMS analytical errors at small
sample sizes when the potential benefits of 1012 or 1013 ohm amplifiers might be maximized.
Sample Preparation for Sm–Nd TIMS Petrochronology
Recall that all Sm–Nd geochronology is isochron geochronology. Thus, a whole rock or
matrix sample (prepared using traditional methods of powdering and dissolution) is required
to pair with the mineral of interest. In the next section we will address the challenges of
extracting and preparing the specific mineral of interest for dissolution. Once any sample (a
whole rock, matrix, or mineral powder) has been acquired, the next step is full dissolution.
This is followed by addition of the mixed calibrated Sm–Nd spike solution. The sample
is then ready for column chromatography. There are several methods for clean separation
of Nd and Sm for analysis, each generally involving a pre-concentration step (or steps) to
extract the REEs. For traditional Nd-metal analysis, most labs use LN-Spec resin for the final
separation of Sm from Nd (e.g., Pin and Santos Zalduegui 1997). For NdO analysis, column
chemistry must do a good job of also eliminating Pr from the Nd; LN-Spec resin generally
does a poor job at that. So, methyl-lactic acid (also called alpha-hydroxy-isobutyric acid)
is frequently used for final separation of all the REEs (e.g., DePaolo and Wasserburg 1976;
Harvey and Baxter 2009). Care must be taken to calibrate these columns as they are pH and
also temperature sensitive. Even with good column chemistry, interferences from Pr and Sm
(on Nd) and from Gd (on Sm) should be monitored during TIMS analysis.
Garnet Petrochronology. Sm–Nd petrochronology is most widely applied to garnet
because it is a common porphyroblastic mineral in a wide array of metamorphic rocks, rather
than an accessory phase. The use of garnet as a monitor of tectonometamorphic evolution—
via its well preserved concentric chemical, textural, and isotopic zonation—is well established
after many decades of fundamental thermodynamic, geochemical, and analytical investments
(see Baxter et al. 2017, this volume, for a review). Relatively speaking, the chronology of
garnet is still being refined. Garnet petrochronology can be done either via bulk mineral
analysis of whole grains, or via analysis of multiple growth zones within a single crystal via
chemically contoured microsampling (see Pollington and Baxter 2011, Baxter and Scherer
2013, or Baxter et al. 2017, this volume for further review). We note that one of the very first
pioneering studies of zoned garnet geochronology used the Rb–Sr system (Christensen et al.
1989) though subsequent work on garnets has moved much more towards Sm–Nd (via TIMS)
and Lu–Hf (via MC-ICPMS). Lu–Hf geochronology is not discussed in this TIMS-focused
chapter; for a full discussion of Lu–Hf garnet geochronology see Baxter et al. 2017, this
volume and references therein. Sample preparation for bulk garnet analysis requires standard
techniques which may include heavy liquids, magnetic separation, and handpicking to extract
an optically pure garnet separate. Lapen et al. (2003) notes the possibility of unintentionally
fractionating different growth zones in garnet during magnetic separation if magnetite or
ilmenite inclusions occur heterogeneously. But, garnet petrochronology via TIMS is at its
most powerful when pairing high precision, chemically contoured, concentric age zonation
in single garnet crystals to the commonly zoned major element, trace element, isotopic,
and textural information that can also be extracted from the same growth zones prior to
their consumption during Sm–Nd analysis. Zoned garnet geochronology first requires a thin
(1−2 mm) disc-shaped slice through the geometric center of a garnet from which a map of the
zoned garnet can be produced to guide the sampling. The adjacent garnet surface may also be
preserved for a thin section and to add petrologic and structural context. The map could be
based on simple textures, colors, or concentric geometries visible with naked eye, or it may be
a chemical map of the disc’s surface acquired via electron microprobe or LA-ICPMS. In any
case, once the map has been created, the petrochronologist’s next task is to physically extract
different portions of the garnet. Different methods have been developed over the years (stick
it on tape and whack it with a hammer–Cohen et al. 1988; rectangular saw cuts–Christensen
et al. 1989; cylindrical drill cores–Stowell et al. 2001) culminating in the use of a high spatial
resolution MicroMill device either to extract microdrilled garnet powders (Ducea et al. 2003)
or intact solid garnet annuli (Pollington and Baxter 2011). Once garnet separates have been
prepared, they must then be cleaned of their mineral inclusions. Avoiding visible inclusions
during micromilling and hand picking can go a long way but most of the time an additional
“partial dissolution” step is required to eliminate microscopic inclusions (see Baxter et al.
2017, this volume for further discussion). Once the garnet has been properly cleaned, it is
ready for full dissolution and column chemistry described above. Figure 11 shows the typical
workflow for chemically contoured Sm–Nd garnet geochronology via TIMS.
Numerous studies have now paired zoned garnet geochronology with related petrologic,
geochemical, or textural observations in the same garnets; the possibilities for future garnet
petrochronology are vast. Figure 12 illustrates four examples of chemically contoured garnet
microsampling conducted following the workflow of Figure 11. 5−6 cm diameter garnet
crystals (rare as they may be) have produced concentric age information from up to 13
sampled annuli revealing not just when the garnet grew, and not just how long it grew, but
how the rate of growth changed during that timespan. Figure 13 highlights examples of the
4) partial dissolution
to dissolve away Sm-Nd Tracer 6) ion exchange
inclusions
separation
2) in situ chemically
coutoured MicroMilling
ACI D
columns with
ACI D ion exchange
resins
5) recover & fully
dissolve pure garnet
1) chemical map of
centrally sectioned
garnet disc by EMP
Nd and Sm aliquots
or LA-ICPMS
3) remove, crush,
seive, Frantz,
handpick, each 7) Nd and Sm isotopes: TIMS
solid garnet annulus
Figure 11. Workflow involved in chemically contoured Sm–Nd ID-TIMS garnet geochronology. Example
garnet shown in #1,2,3 is from Gatewood et al. (2015). Micromill image is from Pollington and Baxter (2010).
A B
10 zones in a
5cm garnet -
13 zones in 6cm garnet - 7.5 Myr span 8 Myr span
C D
6 zones in a 2.2cm garnet - 4 Myr span 3 zones in a 1.2cm garnet - <1 Myr span
Figure 12. Four examples of chemically contoured microsampling for garnet geochronology via TIMS. Pan-
els are shown at the same scale; 1 cm scale bars indicated. A. garnet from a chlorite schist in a local shear
zone, Tauern Window, Austria (from Pollington and Baxter 2010); B. garnet from a felsic blueschist-facies
rock from Sifnos, Greece (from Dragovic et al. 2015); C. garnet from a regional metamorphic schist at Town-
shend Dam, Vermont, USA (from Gatewood et al. 2015); D. garnet from a mafic blueschist from Sifnos,
Greece (from Dragovic et al. 2012). Dark or colored concentric lines on each figure represent the drill trench-
es from which garnet powder was discarded. Instead, solid annuli between each trench was kept for analysis.
th
row
Age (Myr; relative to core)
e tg s
4 r n se
Ga pul
growth pulses
-2 resolved
-4
0.0
0.0 0.5
0.5 1.0
1.0 1.5
1.5 2.0
2.0 2.5
2.5 3.0
3.0
radial distance from core (cm)
Figure 13. Resolving power of zoned garnet geochronology. Circles are data from Pollington and Baxter
(2010) as pictured in Figure 12a. Diamonds are data from Dragovic et al. (2015) as pictured in Figure 12b.
Note individual age precision is generally < ±1 Myr. Two gray diamonds are age data not included in
interpretations due to a likely artifact from sample preparation (see Dragovic et al. 2015 for discussion).
Shaded bars highlight growth pulses resolved by the data. Note the 2-order of magnitude acceleration in
volumetric growth rate of the garnet from Dragovic et al. (2015). The Pollington and Baxter (2010) garnet
is consistent with a constant volumetric growth rate with the two rapid pulses superimposed.
geochronologic resolution that can be gained from this method for the two largest garnets.
Pollington and Baxter (2010) showed geochronologic evidence for two rapid pulses of garnet
growth that correlate with garnet chemical zonation; these pulses reflect evolving kinetic and
thermodynamic drivers during the waning stages of the Alpine Orogeny 20−28 million years
ago. Dragovic et al. (2015) showed that while this 5 cm garnet crystal from Sifnos, Greece
grew over 8 million years, the majority of its growth happened in a final rapid pulse lasting
just a few hundred thousand years. Using thermodynamic analysis of the garnet forming
reaction, this pulse was linked to rapid dehydration from this lithology during subduction
45 million years ago. Smaller garnets 1−2 cm in diameter are also suitable for this approach.
Gatewood et al. (2015) conducted zoned garnet chronology on ten different crystals from
the same rock volume (one of these is picture in Fig. 12c) of a regional metamorphic schist
from Townshend Dam Vermont. Comparison of Mn chemical zonation showed a correlation
between Mn and age indicating first-order rock-wide chemical equilibrium for major elements
in the rock volume. Dragovic et al. (2012) conducted zoned chronology on two crystals in a
mafic blueschist from Sifnos Greece providing evidence again of a very rapid pulse of garnet
growth and related dehydration. The reader is directed to these papers for a full accounting
of the petrochronologic results, but we present them here as recent examples of what is
already possible with chemically contoured garnet Sm–Nd petrochronology. Given available
methodologies, chemically contoured garnet petrochronology on a single crystal requires a
grain diameter of at least ~5 mm (Pollington and Baxter 2011), though smaller crystals may be
drilled and lumped together in some cases. On the one hand, the scale of this “microsampling”
seems enormous compared to the tiny zircons of U–Pb geochronology; however the sample
size of Nd extracted from the garnet zones itself is still quite small due to the low concentration
of Nd (< < 1 ppm) in most garnets. Fortunately, unlike accessory minerals, garnet crystals often
grow large enough to permit zoned chronology and high resolution petrochronology.
Internal Sm–Nd Isochron Petrochronology. It is also theoretically possible to construct
an isochron solely from the mineral to be dated, called an “internal mineral” isochron.
This only works if the mineral exhibits some natural zonation in Sm/Nd, which allows a
reasonable spread along the isochron and worthwhile age precision (e.g., Rakovan et al.
1997), and a physical means of separating and analyzing the different zones (e.g., Sjoqvist
et al. 2016). Sector zoned minerals are good candidates for this approach as they can have
easily recognizable zones with strongly variable chemistry, but also grew together at the same
time (thus conforming to the fundamental isochron assumption). The elegance of an internal
isochron is that it no longer requires the use of points on the isochron besides the mineral of
interest, nor all of the (sometimes fraught) assumptions that must be made about the suitability
of any second (or third, fourth, etc) point added to an isochron besides the mineral of interest.
Rakovan et al. (1997) used sector zoned apatite and cut small chunks from different crystal
faces to construct a 2 point isochron of 43.8 ± 4.7 Ma based on a spread in 147Sm/144Nd
from 0.16 to 0.52. Somewhat analogously, Sjövqist et al. (2016) dated the REE ore mineral
eudialyte from the Norra Kärr deposit, Sweden, via an internal isochron (Fig. 14). In this case
sector zonation revealed much more subtle variations in 147Sm/144Nd ratio. Given the high Nd
concentrations in the mineral (several thousand ppm) tiny 150 um diameter microdrilled pits
provided enough material for precise analysis, and permitted careful spatial control to sample
pits from different sector zones as guided by in situ LA-ICPMS mapping. Figure 14 shows
the resulting five point internal isochron of 1040 ± 44 Ma based. The relatively poor precision
is reflective of the small spread in 147Sm/144Nd from 0.141 to 0.186. The petrologic value of
this age is that it directly constrains the primary ore mineral’s formation even as evidence
for complex multistage open system processes abounds. The age provides important context
for the specific processes that led to ore formation within this complex system. While few
minerals will be amenable to such internal isochron petrochronology, it is an elegant approach
that can yield valuable petrochronologic constraints when possible.
0.5126
Eudialyte Internal Isochron

1040 ± 44 Ma
0.5125
(n=5; MSWD=1.7)
143Nd/144Nd
0.5124
0.5123
0.5122
0.13 0.14 0.15 0.16 0.17 0.18 0.19
147Sm/144Nd
Figure 14. Eudialyte internal isochron. Each datapoint represents a different ~75 mm radius microdrilled pit
within a single sector zoned eudialyte crystal. Inset shows in situ micromilling of the sample in progress.
Data from Sjövqist et al. (2016).
Sm–Nd Age Precision

The precision of an Sm–Nd isochron age depends on the following factors: (1) Maximum
spread in 147Sm/144Nd between points on the isochron; (2) Analytical precision of the
143
Nd/144Nd data; (3) Analytical precision of the 147Sm/144Nd data; (4) Number of points on the
isochron and the MSWD.
If we consider a simple two-point isochron between whole rock and garnet, in general,
the garnet (or other high Sm/Nd phase) controls the precision and the age. The higher the
garnet’s 147Sm/144Nd the more precise the isochron age can potentially be. After that, the
analytical precision of the 143Nd/144Nd and 147Sm/144Nd determine two-point isochron age
precision, though their relative importance varies with the age of the garnet and the absolute
147
Sm/144Nd ratio (Fig. 14). The 147Sm/144Nd precision matters most to age precision when
the garnet is very old (i.e., Proterozoic to Archean) and has a higher 147Sm/144Nd ratio. For
young garnet and with lower absolute 147Sm/144Nd ratio, it is the 143Nd/144Nd precision that
matter most to age precision. Baxter and Scherer (2013) and Baxter et al. (2017, this volume)
show how age precision varies as a function of 147Sm/144Nd ratio in garnet, and for different
143
Nd/144Nd analytical precisions. Here, Figure 15 shows how age precision varies as a
function of 147Sm/144Nd ratio in garnet, both for different 147Sm/144Nd analytical precision
and for differently aged garnet. An important nuance to appreciate here, which differs in
the way we typically think about U–Pb ages, is that the absolute precision of a garnet age
generally does not increase proportionally with increasing age of the garnet, except when Sm/
Nd ratio is highest and/or 147Sm/144Nd analytical precision is poorest. That is, whereas U–Pb
geochronologists typically speak about % age uncertainty, a Sm–Nd geochronologist would
speak about absolute uncertainty. For example, as shown in Figure 15B, all things being equal,
if 147Sm/144Nd analytical precision is very good (e.g., 0.02%) one could date a 10 million year
old garnet at ±1 Ma absolute precision (±10%) and also a 1000 million year old garnet at
the same ±1 Ma absolute precision (±0.1%). The reason is the slow decay of 147Sm, and the
relatively low parent/daughter ratio, as compared to most minerals dated via U–Pb.
The major limitation in analytical precision, as already discussed, is sample size. The
concentration of Nd in truly clean garnet is very low, usually between 0.01 and 1.0 ppm
(Baxter et al. 2017, this volume). Thus, as the petrochronologist seeks to microsample
smaller and smaller zones, smaller amounts of Nd (and Sm) will need to be analyzed. Or,
we may be working with a different mineral rich in Nd (e.g., monazite, xenotime, eudialyte)
but we may be limited in sample size simply by natural grain size or by the complexity and
resolution of internal zonation we wish to microsample, for example via tiny MicroMill
pits just 50−150 mm wide (e.g., Sjövqist et al. 2016; Fig. 14). Figure 10 provides a sense of
the analytical precision one can expect from different load sizes of Nd. 147Sm/144Nd can be
measured with external reproducibility no worse than 0.1% to 0.5% with advances towards
0.01% precision underway (e.g., Dragovic et al. 2015 reports 0.023% external precision
on 147Sm/144Nd on repeat analyses of a mixed gravimetric Sm–Nd standard, spiked and
run through columns). Only at sub-nanogram sample sizes does internal run precision of
147
Sm/144Nd exceed the external precision and thus come into play. In practice, one should
always use the greater (poorer) of the internal run precision or the long term external
precision (based on repeat runs of standards of similar sample size) in age error propagation.
The final aspect of isochron age precision comes from additional points on the isochron
and the degree to which they fit on the same isochron. The choice of additional points on an
isochron is a delicate matter (see Baxter et al. 2017, this volume) so one must resist the urge to
populate an “isochron” with as many points as possible just to meet some statistical criterion.
Indeed the business of high-resolution petrochronology seeks to resolve real geological
differences and subtleties in age, not smear them out and obfuscate them with the brush of
statistics. Still, if multiple preparations of the same generation of garnet (or mineral) growth
(as determined by microdrilling of similar age zones, or by chemical correlation, or textural
arguments) can be collected, a multipoint isochron is always preferred. The MSWD is a
statistical measure of the goodness of fit accounting for the scatter that would be expected given
the reported analytical uncertainties (Wendt and Carl 1991). When the fit is good, an MSWD
is close to 1.0 and age precision will improve as compared to a simple two-point isochron age
precision. When fit is poor, an MSWD is  1 and age precision can worsen as compared to a
simple two-point isochron. As discussed in Baxter et al. (2017, this volume) and Kohn (2009) a
poor MSWD should not be viewed as a failure; rather a high MSWD can be an indication that
the samples on the isochron did not in fact grow at a single time, but rather that a more complex
geochronologic story is there, waiting for a clever petrochronologist to dissect.
Another crucial variable that can affect the isochron age precision—and accuracy—is
contamination from blank. As discussed for U–Pb geochronology, characterization and correction
for the blank can often be the primary limitation to age precision. While blank corrections have
always been a major concern for U–Pb dating, Sm–Nd geochronology has only recently pushed
into small enough sample sizes where blank now becomes a major limitation. Blanks affect
both the 143Nd/144Nd and the 147Sm/144Nd measurements. Thus, a blank correction should be
made (and may become significant) whenever sample/blank ratio falls below ~1000:1. To
accomplish this, a good constraint on the 147Sm/144Nd and 143Nd/144Nd of the blank is required.
This can pose a major challenge given that good lab blanks are < 10 pg of Nd. The good news
is that the 147Sm/144Nd and 143Nd/144Nd blank corrections are correlated, and generally pull
each datapoint down along the isochron. This is because most blanks should approximate most
common rocks, thus plotting near the whole rock point on the isochron for most samples. This
problem is exacerbated for the oldest samples that deviate the most from modern crustal average
values. Just as is common practice in U–Pb geochronology, rigorous statistical propagation of
correlated errors in blank corrections are needed especially when sample/blank is  1000:1. For
Sm–Nd, especially given the somewhat more reagent intensive NdO column chemistry, the best
100
147Sm/144Nd
@ 0.5% (2RSD)
Age precision (Ma)
10
A
0.1
100
147
Sm/144Nd=0.5
147
Sm/144Nd=1.0
147Sm/144Nd
147
Sm/144Nd=2.0
@ 0.02% (2RSD) 147
Sm/144Nd=4.0
Age precision (Ma)
10
B
0.1
1 10 100 1000
age (Ma)
Figure 15. Sm–Nd age precision vs. age of sample. All calculations are for a two-point garnet–matrix Sm–Nd isochron where matrix
has a 147Sm/144Nd = 0.15, and the analytical precision for 143Nd/144Nd is 10 ppm (2RSD) for both datapoints. Each curve is for a dif-
ferent absolute 147Sm/144Nd of the garnet ranging from 0.5 (solid) to 4.0 (long dashed). Panel A is for a 147Sm/144Nd analytical preci-
sion of 0.5% (2RSD) whereas Panel B is for a 147Sm/144Nd analytical precision of 0.02%. Note that the importance of 147Sm/144Nd
analytical precision is greatest at older ages.
Nd blanks are in the 1−10 pg range. Thus, as we dive into the sub-nanogram range of analysis,
blank begins to represent the largest source of uncertainty, dwarfing the improvements that could
be made by higher ohm resistors or new-and-improved loading solutions to further enhance ion
yield (e.g., Baxter et al. 2014). The authors are not aware of any reports of labs having achieved
total chemistry blanks below 1 pg of Nd; that will be an important frontier as we seek to push the
barriers of petrochronologic age resolution further. In this regard, Sm–Nd geochronologists have
much to learn from U–Pb geochronologists as the two fields have advanced at different paces.
THE FUTURE
As geochronologists continue to calibrate the rates of geologic processes at a finer and
finer scale, there is an inevitable result that geologists and geochronologists are driven to
ask questions that then require higher precision dates. This cyclic process is why TIMS
geochronology continues to thrive despite its time-consuming and relatively cumbersome
nature. While three decades ago it was sufficient to ask what the age of a pluton was, igneous
petrologists now want to know over what timescales a pluton is constructed. While it was
once common to classify a metamorphic “event”, it is now more interesting, and also possible,
to look at the rates of metamorphic reactions within an event that occurred over millions of
years. As we learn more about these processes and about how they are recorded by minerals
we can actually date, it becomes even more important to place geochronologic data within the
context of petrogenesis. The challenges and opportunities that face ID-TIMS petrochronology
outlined above must be met by sustained innovation, starting with sample collection in the
field and ending with mass spectrometry and data reduction and analysis.
While this review focuses on U–Pb and Sm–Nd petrochronology, other systems including
Rb–Sr and Re–Os are also amenable linking dates with rock forming processes, and numerous
examples from these systems exist. Because the Rb–Sr system is useful in dating the growth or
evolution of numerous rock forming minerals, this tool has been used to link such minerals to
growth or deformation textures some of which may relate to larger scale tectonic or geochemical
forcing (e.g., Christensen et al. 1989; Muller et al. 2000; Cliff and Meffan-Main 2003; Charlier
et al. 2006; Glodny et al. 2008; de Meyer et al. 2014; Walker et al. 2016). Generally, in situ
measurements of 87Sr/86Sr are more robust with MicroMill-TIMS (followed by column chemistry
to remove Rb interference) than with LA-ICPMS due to the challenges of Rb and Kr interferences
that can exist in LA-ICPMS analysis (though reaction-cell methods now allow direct in-situ LA-
ICPMS Rb–Sr geochronology; e.g. Zack and Hogmalm 2016). Petrochronologic applications
employing high resolution MicroMill sampling of individual growth zones in feldspars (e.g.,
Charlier et al. 2006) or polymineralic porphyroblast strain fringes (e.g., Muller et al. 2000)
represent examples of what is possible with in situ MicroMill-TIMS methods and Rb–Sr. Re–Os
geochronology has proven useful in calibrating the timescales of ore-forming processes through
molybdenite geochronology (Selby and Creaser 2001; Stein et al. 2001; Bingen and Stein 2003),
and continued interest in understanding these deposits and their links to igneous and hydrothermal
systems will continue to motivate work in Re–Os petrochronology (Zimmerman et al. 2014).
It is unlikely that ID-TIMS geochronology will meet the spatial resolution of in situ
techniques (though in situ MicroMill pits may be as small as ~50−100 µm diameter using
fine tipped tungsten-carbide bits; Charlier et al. 2006), but it has yet to be seen whether in situ
geochronology will match the precision and accuracy currently afforded by ID-TIMS. The
goal in ID-TIMS geochronology, therefore, is to bridge the gap between spatial resolution
and petrographic and geochemical context that is more easily attained by in situ techniques.
Many of the tools to do this are in place but underutilized. Sample characterization via in
situ techniques such as petrography, SIMS, LA-ICPMS and EPMA should be a rule rather
than an exception for those wishing to understand high-precision geochronologic data and
attach petrologic significance to it. Although ID-TIMS is rightly touted as the most precise
and accurate technique for some geochronologic systems, increasing accuracy of reported
ages remains dependent on correct data interpretation. Developing workflows that benefit most
from multiple analytical and theoretical tools for interpreting geochronologic data in terms of
petrogenesis will therefore become even more important in the future.
ACKNOWLEDGMENTS
EFB gratefully acknowledges support from NSF grants EAR-1250497/1561882 and
PIRE-1545903. EFB thanks Kathryn Maneiro for sharing otherwise unpublished data from
her PhD thesis at Boston University, and for her pioneering efforts in sub-ng Nd analysis.
BS would like to thank Kyle Samperton and Michael Eddy for feedback on the manuscript,
and Kyle Samperton for help compiling data from the Princeton lab. The authors are grateful
for the careful reviews from Urs Schaltegger and Randy Parrish, in addition to constructive
feedback, patience, and editorial handling by Matt Kohn.
REFERENCES
Amelin Y (2004) Sm–Nd systematics of zircon. Chem Geol 211:375–387
Amelin Y, Davis WJ (2006) Isotopic analysis of lead in sub-nanogram quantities by TIMS using a 202Pb–205Pb
spike. J Anal At Spectrom 21:1053–1061
Amelin Y, Lee D-C, Halliday AN, Pidgeon RT (1999) Nature of the Earth’s earliest crust from hafnium isotopes in
single detrital zircons. Nature 399:252–255
Barboni M, Schoene B (2014) Short eruption window revealed by absolute crystal growth rates in a granitic
magma. Nat Geosci 7:524–528
Barboni M, Boehnke P, Keller CB, Kohl I, Schoene B, Young ED, McKeegan KD (2017) Early formation of the
moon 4.51 billion years ago. SciAdv 3.1:e1602365
Baxter EF, Scherer EE (2013) Garnet geochronology: timekeeper of tectonometamorphic processes. Elements
9:433–438
Baxter EF, Honn DK, Sullivan NS, Eccles KA (2014) Sub-nanogram Nd isotope analysis via TIMS: Magic potions,
fancy resistors, but don’t forget the blank. Goldschmidt Meeting, Sacramento CA
Geochem 83:469–533
Bell K, Anglin CD, Franklin JM (1989) Sm–Nd and Rb–Sr isotope systematics of scheelites: Possible implications
for the age and genesis of vein-hosted gold deposits. Geology 17:500–504
Bingen B, Stein H (2003) Molybdenite Re–Os dating of biotite dehydration melting in the Rogaland high-
temperature granulites, S. Norway. Earth Planet Sci Lett 208:181–195
Blackburn T, Bowring S, Schoene B, Mahan K, Dudas F (2011) U–Pb thermochronology: creating a temporal
record of lithosphere thermal evolution. Contrib Mineral Petrol 162:479–500
Broderick C, Wotzlaw JF, Frick DA, Gerdes A, Ulianov A, Günther D, Schaltegger U (2015) Linking the thermal
evolution and emplacement history of an upper-crustal pluton to its lower-crustal roots using zircon
geochronology and geochemistry (southern Adamello batholith N. Italy). Contrib Mineral Petrol 170:1–17
Carlson RW (2014) 15.18 - Thermal Ionization Mass Spectrometry A2. In:Treatise on Geochemistry (Second
Edition) Holland, Heinrich D, Turekian KK (ed.) Oxford, Elsevier, p. 337–354
Chamberlain KR, Bowring SA (2001) Apatite–feldspar U–Pb thermochronometer: a reliable mid-range (~450 °C),
diffusion controlled system. Chem Geol 172:173–200
Charlier BLA, Ginibre C,Morgan D,Nowell GM, Pearson,DG, Davidson JP, Ottley CJ (2006) Methods for the
microsampling and high-precision analysis of strontium and rubidium isotopes at single crystal scale for
petrological and geochronological applications. Chem Geol 232:114–133
Chelle-Michou C, Chiaradia M, Ovtcharova M, Ulianov A, Wotzlaw J-F (2014) Zircon petrochronology reveals
the temporal link between porphyry systems and the magmatic evolution of their hidden plutonic roots (the
Eocene Coroccohuayco deposit, Peru). Lithos 198:129–140
Chesley JT, Halliday AN, Scrivener RC (1991) Samarium–neodymium direct dating of fluorite mineralization.
Science 252:949–951
Christensen JN, Rosenfeld JL, DePaolo DJ (1989) Rates of tectonometamorphic processes from rubidium and
strontium isotopes in garnet. Science 244:1465–1469
Chu ZY, Chen FK, Yang YH, Guo JH (2009) Precise determination of Sm, Nd concentrations and Nd isotopic
compositions at the nanogram level in geological samples by thermal ionization mass spectrometry. J Anal
At Spectrom 24:1534–1544
Cliff RA, Meffan-Main S (2003) Evidence from Rb–Sr microsampling geochronology for the timing of Alpine
deformation in the Sonnblick Dome, SETauern Window, Austria. Geol Soc Spec Publ 220:159–172
Cohen AS, O’Nions RK, Siegenthaler R, Griffin WL (1988) Chronology of the pressure–temperature history
recorded by a granulite terrain. Contrib Mineral Petrol 98:303–311
Compston W, Oversby VM (1969) Lead isotopic analysis using a double spike. J Geophys Res 74:4338–4348
Condon D, Schoene B, McLean N, Bowring S, Parrish R (2015) Metrology and traceability of U–Pb isotope
dilution geochronology (EARTHTIMETracer Calibration Part I). Geochim Cosmochim Acta 164:464–480
Corfu F (2013) A century of U–Pb geochronology: The long quest towards concordance. Geol Soc Am Bull
125:33–47
Corrie SL, Kohn MJ (2007) Resolving the timing of orogenesis in the Western Blue Ridge, southern Appalachians,
via in situ ID-TIMS monazite geochronology. Geology 35:627–630, DOI: 610.1130/G23601A23601
Crowley JL, Schmitz MD, Bowring SA, Williams ML, Karlstrom KE (2006) U–Pb and Hf isotopic analysis of
zircon in lower crustal xenoliths from the Navajo volcanic field: 1.4 Ga mafic magmatism and metamorphism
beneath the Colorado Plateau. Contrib Mineral Petrol 151:313–330, doi: 310.1007/s00410-00006-00061-z
Crowley JL, Schoene B, Bowring SA (2007) U–Pb dating of zircon in the Bishop Tuff at the millennial scale.
Geology 35:1123–1126; doi: 1110.1130/G24017A
D’Abzac F-X, Davies JH, Wotzlaw J-F, Schaltegger U (2016) Hf isotope analysis of small zircon and baddeleyite
grains by conventional multi collector-inductively coupled plasma-mass spectrometry. Chem Geol 433:12–23
Davis DW, Williams IS, Krogh TE (2003) Historical development of zircon geochronology. Rev Mineral 53:145–181
Deering CD, Keller B, Schoene B, Bachmann O, Beane R, Ovtcharova M (2016) Zircon record of the plutonic-
volcanic connection and protracted rhyolite melt evolution. Geology 44:267–270
de Meyer CMC, Baumgartner LP, Beard BL, Johnson CM (2014) Rb–Sr ages from phengite inclusions in garnets
from high pressure rocks of the Swiss Western Alps. Earth Planet Sci Lett 395:205–216
DePaolo DJ, Wasserburg GJ (1976) Nd isotopic variations and petrogenetic models. Geophys Res Lett 3:249–252
DesOrmeau JW, Gordon SM, Little TA, Bowring SA (2014) Tracking the exhumation of a Pliocene (U) HP terrane:
U–Pb and trace-element constraints from zircon, D’Entrecasteaux Islands, Papua New Guinea. Geochem
Geophys Geosystems 15:3945–3964
DesOrmeau JW, Gordon SM, Kylander-Clark AR C, Hacker BR, Bowring SA, Schoene B, Samperton KM (2015)
Insights into (U)HP metamorphism of the Western Gneiss Region, Norway: A high-spatial resolution and
high-precision zircon study. Chem Geol 414:138–155
314–317:9–22
Dragovic B, Baxter EF and Caddick MJ (2015) Pulsed dehydration and garnet growth during subduction revealed
by zoned garnet geochronology and thermodynamic modeling, Sifnos, Greece. Earth Planet Sci Lett
413:111–122
Ducea MN, Ganguly J, Rosenberg EJ, Patchett PJ, Cheng WJ and Isachsen C (2003) Sm–Nd dating of spatially
Fisher CM, Vervoort JD, DuFrane SA (2014) Accurate Hf isotope determinations of complex zircons using the
“laser ablation split stream” method. Geochem Geophys Geosystems 15:121–139
Galer S, Abouchami W (1998) Practical application of lead triple spiking for correction of instrumental mass
discrimination. Mineral Mag A 62:491–492
Vermont, USA. Chem Geol 401:151–168
Geisler T, Pidgeon RT, Kurtz R, van Bronswijk W, Schleicher H (2003) Experimental hydrothermal alteration of
partially metamict zircon. Am Mineral 88:1496–1513
Gerstenberger H, Haase G (1997) A highly effective emitter substance for mass spectrometric Pb isotope ratio
determinations. Chem Geol 136:309–312
Glodny J, Kühn A, Austrheim H (2008) Geochronology of fluid-induced eclogite and amphibolite facies
metamorphic reactions in a subduction—collision system, Bergen Arcs, Norway. Contrib Mineral Petrol,
156:27–48
Gordon SM, Bowring SA, Whitney DL, Miller RB, McLean N (2010) Time scales of metamorphism, deformation,
crustal melting in a continental arc, North Cascades, USA. Geol Soc Am Bull: B30060-1
Griffin WL and Brueckner HK (1980) Caledonian Sm–Nd ages and a crustal origin for Norwegian eclogites.
Nature 285:319–321
Griselin M, van Belle JC, Pomies C, Vroon PZ, van Soest MC, Davies GR (2001) An improved chromatographic
separation of Nd with application to NdO+ isotope analysis. Chem Geol 172:347–359
small aliquots (1–10 ng). Chem Geol 258:251–257
Heaman LM, Bowins R, Crocket J (1990) The chemical composition of igneous zircon suites: implications for
geochemical tracer studies. Geochim Cosmochim Acta 54:1597–1607
Henjes-Kunst F, Prochaska,W, Niedermayr A, Sullivan N, Baxter E (2014) Sm–Nd dating of hydrothermal
carbonate formation: the case of the Breitenau magnesite deposit (Styria, Austria). Chem Geol 387:184–201
Horstwood MSA, Košler J, Gehrels G, Jackson SE, McLean NM, Paton C, Pearson NJ, Sircombe K, Sylvester P,
Vermeesch P, Bowring JF, Condon DJ, Schoene B (2016) Community-derived standards for LA-ICP-MSU-
Th–Pb geochronology—uncertainty propagation, age interpretation and data reporting. Geostand Geoanal
Res, doi: 10.1111/j.1751-908X2016.00379.x
Iizuka T, Nebel O, McCulloch MT (2011) Tracing the provenance and recrystallization processes of the Earth’s
oldest detritus at Mt. Narryer and Jack Hills, Western Australia: An in situ Sm–Nd isotopic study of monazite.
Mineral 101:25–42
Evidence for a static hot channel in the Himalayan orogen. Earth Planet Sci Lett 311:136–143
Geochem 83:103–152
Koornneef JM, Bouman C, Schwieters JB, Davies GR (2013) Use of 1012 ohm current amplifiers in Sr and Nd
isotope analyses by TIMS for application to sub-nanogram samples. J Anal At Spectrom 28:749–754
Koornneef JM, Bouman C, Schwieters JB, Davies GR (2014) Measurement of small ion beams by thermal
ionisation mass spectrometry using new 1013 ohm resistors. Anal Chim Acta 819:49–55
345:99–112
Lanzirotti A, Hanson GN (1996) Geochronology and geochemistry of multiple generations of monazite from the
Wepawaug Schist, Connecticut, USA: implications for monazite stability in metamorphic rocks. Contrib
Lugmair GW, Marti K, Kurtz JP, Scheinin NB (1976) History and genesis of lunar troctolite 76535 or: how old is
old? Proc 7th Lunar Sci Conf:2009–2033
Maneiro, KA (2016) Development of a Detrital Garnet Geochronometer and the Search for Earth’s Oldest Garnet.
PhD Thesis Boston University
Mattinson JM (2005) Zircon U–Pb chemical-abrasion (“CA-TIMS”) method: combined annealing and multi-step
dissolution analysis for improved precision and accuracy of zircon ages. Chem Geol 220:47–56
Mattinson JM (2011) Extending the Krogh legacy: development of the CATIMS method for zircon U–Pb
geochronology. Can J Earth Sci 48:95–105
Matzel JP, Bowring SA, Miller RB (2006) Timescales of pluton construction at differing crustal levels: examples
from the Mount Stuart and Tenpeak intrusions, North Cascades, WA: Geol Soc Am Bull 118:1412–1430 doi:
1410.1130/B25923.25921
McFarlane CRM, McCulloch MT (2007) Coupling of in-situ Sm–Nd systematics and U–Pb dating of monazite
and allanite with applications to crustal evolution. Chem Geol 245:45–60
McLean NM, Bowring JF, Bowring SA (2011) An algorithm for U–Pb isotope dilution data reduction and
uncertainty propagation. Geochem Geophys Geosystem 12:Q0AA18
McLean NM, Condon DJ, Schoene B, Bowring SA (2015) Evaluating uncertainties in the calibration of isotopic
reference materials and multi-element isotopic tracers (EARTHTIME Tracer Calibration Part II). Geochim
Miller JS, Matzel JP, Miller CF, Burgess SD, Miller RB (2007) Zircon growth and recycling during the assembly
of large, composite arc plutons. J Volcanol Geotherm Res 167:282–299
Muller W, Aerden D, Halliday AN (2000) Isotopic dating of strain fringe increments: duration and rates of
deformation in shear zones. Science 288:2195–2198
Mundil R, Ludwig KR, Metcalfe I, Renne PR (2004) Age and timing of the Permian mass extinctions: U/Pb dating
of closed-system zircons. Science 305:1760–1763
Oberli F, Meier M, Berger A, Rosenberg CL, Giere R (2004) U–Th–Pb and 230Th/238U disequilibrium isotope
systematics: Precise accessory mineral chronology and melt evolution tracing in the Alpine Bergell intrusion.
O’Neil J, Carlson RW, Francis D, Stevenson RK (2008) Neodymium-142 evidence for Hadean mafic crust. Science
321:1828–1831
Peng J-T, Hu R-Z, Burnard PG (2003) Samarium–neodymium isotope systematics of hydrothermal calcites from
the Xikuangshan antimony deposit (Hunan, China): the potential of calcite as a geochronometer. Chem Geol
200:129–136
Peterman EM, Mattinson JM, Hacker BR (2012) Multi-step TIMS and CA-TIMS monazite U–Pb geochronology.
Chem Geol 312–313:58–73
Pin C, Santos Zalduegui JF (1997) Sequential separation of light-rare-earth elements, thorium and uranium by
miniaturization extraction chromatography: application to isotopic analyses of silicate rocks. Anal Chim Acta
339:79–89
tectonometamorphic processes. Earth Planet Sci Lett 293:63–71
Pollington AD, Baxter EF (2011) High precision microsampling and preparation of zoned garnet porphyroblasts
for Sm–Nd geochronology. Chem Geol 281:270–282
Rakovan,J., McDaiel, D.K., Reeder R.J., 1997, Use of surface-controlled REE sectoral zoning in apatite from
Llallagua, Bolivia, to determine a single-crystal Sm–Nd age. Earth Planet Sci Lett 146:329–336
Richard P, Shimizu N, Allegre CJ (1976) 143Nd/146Nd, a natural tracer: an application to oceanic basalts. Earth
Rioux M, Bowring S, Dudás F, Hanson R (2010) Characterizing the U–Pb systematics of baddeleyite through
chemical abrasion: application of multi-step digestion methods to baddeleyite geochronology. Contrib
Rivera TA, Storey M, Schmitz MD, Crowley JL (2013) Age intercalibration of 40Ar/39Ar sanidine and chemically
distinct U/Pb zircon populations from the Alder Creek Rhyolite quaternary geochronology standard. Chem
Geol 345:87–98
Rivera TA, Schmitz MD, Crowley JL, Storey M (2014) Rapid magma evolution constrained by zircon
petrochronology and 40Ar/39Ar sanidine ages for the Huckleberry Ridge Tuff, Yellowstone, USA. Geology
42:643–646
Root DB, Hacker BR, Mattinson JM, Wooden JL (2004) Zircon geochronology and ca. 400 Ma exhumation of
Norwegian ultrahigh-pressure rocks: an ion microprobe and chemical abrasion study. Earth Planet Sci Lett
228:325–341
Samperton KM, Schoene B, Cottle JM, Brenhin Keller C, Crowley JL, Schmitz MD (2015) Magma emplacement,
differentiation and cooling in the middle crust: Integrated zircon geochronological–geochemical constraints
from the Bergell Intrusion, Central Alps. Chem Geol 417:322–340
Schaltegger U, Zeilinger G, Frank M, Burg JP (2002) Multiple mantle sources during island arc magmatism: U–Pb
and Hf isotopic evidence from the Kohistan arc complex, Pakistan. Terra Nova 14:461–468
Schaltegger U, Brack PB, Ovtcharova M, Peytcheva I, Schoene B, Stracke A, Bargossi GM (2009) Zircon U,
Pb, Th, Hf isotopes record up to 700 kyrs of magma fractionation and crystallization in a composite pluton
(Adamello batholith, N. Italy). Earth Planet Sci Lett 286:208–218
Schärer U (1984) The effect of initial 230Th disequilibrium on young U–Pb ages: the Makalu case, Himalaya. Earth
Geochem 83:199–230
Schmitz MD, Schoene B (2007) Derivation of isotope ratios, errors, error correlations for U–Pb geochronology
using 205Pb-235U-(233U)-spiked isotope dilution thermal ionization mass spectrometric data. Geochem
Geophys Geosystem 8:Q08006
Schoene B (2014) U–Th–Pb geochronology. In:Treatise on Geochemistry, Volume 4.10, Rudnick R (ed.) Oxford
UK, Elsevier, p. 341–378
Schoene B, Crowley JL, Condon DC, Schmitz MD, Bowring SA (2006) Reassessing the uranium decay constants
for geochronology using ID-TIMS U–Pb data. Geochim Cosmochim Acta 70:426–445
Schoene B, Latkoczy C, Schaltegger U, Gunther D (2010) A new method integrating high-precision U–Pb geochronology
with zircon trace element analysis (U–Pb TIMS-TEA). Geochim Cosmochim Acta 74:7144–7159
Schoene B, Schaltegger U, Brack P, Latkoczy C, Stracke A, Günther D (2012) Rates of magma differentiation and
emplacement in a ballooning pluton recorded by U–Pb TIMS-TEA, Adamello batholith, Italy. Earth Planet
Sci Lett 355–356:162–173
Schönbächler M, Fehr MA (2014) 15.7—Basics of ion exchange chromatography for selected geological
applications A2. In:Treatise on Geochemistry (Second Edition), Holland, Heinrich D, Turekian KK (eds.)
Oxford, Elsevier, p. 123–146
Selby D, Creaser RA (2001) Re–Os geochronology and systematics in molybdenite from the Endako porphyry
molybdenum deposit, British Columbia, Canada. Econ Geol 96:197–204
Sharma M, Wasserburg GJ, Papanastassiou DA, Quick JE, Sharkov EV, Laz’ko EE (1995) High 143Nd/ 144Nd in
extremely depleted mantle rocks. Earth Planet Sci Lett 135:101–114
Sjöqvist ASL, Zack T, Baxter EF, Honn DK (2016) Post-magmatic implications for rare-earth element
mineralisation from a microgeochemical in situ ID-TIMS Sm–Nd isochron from a single magmatic eudialyte
crystal from the Norra Kärr alkaline complex. IGC Conference, Cape Town, South Africa
Stacey JC, Kramers JD (1975) Approximation of terrestrial lead isotope evolution by a two-stage model. Earth
Stein H, Markey R, Morgan J, Hannah J, Scherstén A (2001) The remarkable Re–Os chronometer in molybdenite:
how and why it works. Terra Nova 13:479–486
Stowell HH, Taylor DL, Tinkham DL, Goldberg SA and Ouderkirk KA (2001) Contact metamorphic P–T–t paths
from Sm–Nd garnet ages, phase equilibria modelling and thermobarometry: Garnet Ledge, south-eastern
Alaska, USA. J Metamorph Geol 19:645–660
Stracke A, Scherer EE, Reynolds BC (2014) 15.4 –Application of Isotope Dilution in Geochemistry A2 - Holland,
In: Heinrich D, Turekian KK (eds) Treatise on Geochemistry (Second Edition): Oxford, Elsevier:71–86
Tappa MJ, Baxter EF, Maneiro KA, Guest RE (2016) Sub-nanogram neodymium isotope measurements on a New
Isotopx Phoenix TIMS using 1011 and 1012 ohm resistors: GSA Fall Conference, Denver CO, USA
Thirlwall MF (1991) High-precision multicollector isotopic analysis of low levels of Nd as oxide. Chem Geol
94:13–22
Thirlwall MF (2000) Inter-laboratory and other errors in Pb isotope analyses investigated using a 207Pb–204Pb
double spike. Chem Geol 163:299–322
Thoni M, Miller C, Blichert-Toft J, Whitehouse MJ, Konzett J, Zanetta A (2008) Timing of high-pressure
metamorphism and exhumation of the eclogite type-locality (Kupplerbrunn–Prickler Halt, Saualpe, south-
eastern Austria): constraints from correlations of the Sm–Nd, Lu–Hf, U–Pb and Rb–Sr isotopic systems. J
Metamorph Geol 26:561−81, doi:10.1111/j.1525–1314.2008.00778.x
van Breemen O and Hawkesworth CJ (1980) Sm–Nd isotopic study of garnets and their metamorphic host rocks.
Trans R Soc Edinburgh: Earth Sci 71:97–102
Walker S, Thirlwall MF, Strachan RA, Bird AF (2016) Evidence from Rb–Sr mineral ages for multiple orogenic
events in the Caledonides of Shetland, Scotland. J Geol Soc London 173:489–503
Wasserburg GJ, Jacousen SB, DePaolo DJ, McCulloch MT, Wen T (1981) Precise determinations of Sm/Nd ratios,
Sm and Nd isotopic abundances in standard solutions. Geochim Cosmochim Acta 45:2311–2323
Wendt I, Carl C (1991) The statistical distribution of the mean squared weighted deviation. Chem Geol 86:275–285
Wetherill GW (1956) Discordant uranium-lead ages. Trans Am Geophys Union 37:320–326
Williams IS (1998) U–Th–Pb geochronology by ion microprobe. In:Applications of Microanalytical Techniques to
Understanding Mineralizing Processes. McKibben MA, Shanks WC III, Ridley WI (eds.) Volume 7, p. 1–35
Woodhead J, Hergt J, Shelley M, Eggins S, Kemp R (2004) Zircon Hf-isotope analysis with an excimer laser, depth
profiling, ablation of complex geometries, concomitant age estimation. Chem Geol 209:121–135
Yuan H-L, Gao S, Dai M-N, Zong C-L, Günther D, Fontaine GH, Liu X-M, Diwu C (2008) Simultaneous
quadrupole and multiple-collector ICP-MS. Chem Geol 247:100–118
Zack T, Hogmalm KJ. Laser ablation Rb/Sr dating by online chemical separation of Rb and Sr in an oxygen-filled
reaction cell. Chem Geol 437:120–133
Zimmerman A, Stein HJ, Morgan JW, Markey RJ, Watanabe Y (2014) Re–Os geochronology of the El Salvador
porphyry Cu–Mo deposit, Chile: tracking analytical improvements in accuracy and precision over the past
decade. Geochim Cosmochim Acta 131:13–32
Vol. 83 pp. 261–295, 2017 9
Zircon: The Metamorphic Mineral

Daniela Rubatto
University of Bern
Baltzerstrasse 1–3
CH-3012, Switzerland
and
Research School of Earth Sciences
Australian National University
Acton 2601, ACT
Australia
daniela.rubatto@geo.unibe.ch
INTRODUCTION
A mineral that forms under conditions as variable as diagenesis to deep subduction, melt
crystallization to low temperature alteration, and that retains information on time, temperature,
trace element and isotopic signatures is bound to be a useful petrogenetic tool. The variety
of conditions under which zircon forms and reacts during metamorphism is a great asset,
but also a challenge as interpretation of any geochemical data obtained from zircon must
be placed in pressure–temperature–deformation–fluid context. Under which condition and by
which process zircon forms in metamorphic rocks remains a crucial question to answer for the
correct interpretation of its precious geochemical information.
In the last 20 years there has been a dramatic evolution in the use of zircon in metamorphic
petrology. With the advent of in situ dating techniques zircon became relevant as a mineral for
age determinations in high-grade metamorphic rocks. Since then, there has been incredible
progress in our understanding of metamorphic zircon with the documentation of growth
and alteration textures, its capacity to protect mineral inclusions, zircon thermometry, trace
element patterns and their relation to main mineral assemblages, solubility of zircon in melt
and fluids, and isotopic systematics in single domains that go beyond U–Pb age determinations.
Metamorphic zircon is no longer an impediment to precise geochronology of protolith rocks,
but has become a truly indispensable mineral in reconstructing pressure–temperature–time–
fluid-paths over a wide range of settings. An obvious consequence of its wide use, is the rapid
increase of literature on metamorphic zircon and any attempt to summarize it can only be partial:
in this chapter, reference to published works are intended as examples and not as a compilation.
This chapter approaches zircon as a metamorphic mineral reporting on its petrography
and texture, deformation structure and mineral chemistry, including trace element and
isotopic systematics. Linking this information together highlights the potential of zircon as
a key mineral in petrochronology.
PREAMBLE: THE MANY FACES AND NAMES OF METAMORPHIC ZIRCON

Various terms have been used, more or less loosely, to describe features and processes
that form zircon in metamorphic conditions. Different authors may have used the same
term differently, causing additional confusion. Any classification has to be based on texture,
1943-2666/17/0083-0009$05.00 (online)
262 Rubatto
crystal structure and zircon composition and requires understanding of the formation
process, which is not always the case in published studies. Within this contribution a generic
terminology is adopted that uses three principal terms for metamorphic zircon: alteration,
replacement/recrystallization and new growth. Examples of texture of these three categories
are shown in Figure 1 and an attempt to summarize zircon features distinctive of particular
metamorphic environments and processes is presented in Table 1.
Alteration is a process that partly overprints and disturbs a relict mineral and where textural
and chemical vestiges of the relict are preserved. Alteration is often, but not necessarily, aided
by prior metamictization as the damaged zircon domains are particularly prone to alteration by
Table 1. Summary of metamorphic zircon characteristics.
Characteristic Metamorphic conditions Process and/or cause

Regular polygonal zoning, Anatexis, granulite facies Crystallization from a melt or
oscillatory or sector, generally and hydrothermal conditions precipitation from a fluid
weak, and mostly euhedral external
shape
Patchy, mosaic zoning Subsolidus Metamictization, fluid
alteratioin, initial stages of
replacement
Zoning
Unzoned or weak convolute zoning Subsolidus above greenschist Replacement including

facies dissolution-precipitation
Sawtooth overgrowths Diagenesis to low Dissolution-precipitation
greenschist facies < 400°C
Intragrain crystallographic Amphibolite to UHT High strain rates and
missorientation and formation of milonitization
subgrains
Microzircon around major minerals Cooling from high Expulsion of Zr during
(rutile, ilmenite, garnet) temperatures mineral breakdown or
Microstructure
recrystallization
Porosity and inclusions of Th and Subsolidus Dissolution–precipitation
U phases
Presence of non-formula elements From diagenesis to extreme Metamictization and fluid
(Ca, Al...) conditions alteratioin
Pb nuggets Ultra high temperature Pb mobilization
>900°C
composition
Low Th/U Subsolidus to migmatites, Coexistence with Th-rich

Anomalous
less common in UHT and phase such as monazite or

mafic compositions allanite
Flat HREE pattern Amphibolite, eclogite and Coexistence with garnet
granulite facies to extreme
P–T
Strong LREE depletion and steep Amphibolite to granulite Coexistence with abundant
REE pattern facies LREE-rich phases such as
titanite, allanite, monazite
Chemistry
Absence of negative Eu-anomaly Eclogite facies (or Lack of significant amount of

assemblages lacking feldspars in the assemblage
feldspar)
Decoupling of U–Pb and Hf Subsolidus to granulite Alteration and incomplete
systematics replacement
Zircon: The Metamorphic Mineral 263
growth from fluids growth from melt

E 20µm F G
ion
tat
Inherited
ipi
n- nt
D
ec
tio me
pr
olu ce
iss epla
Inherited
H
r
4
dd
Eclogite UHP
an
3
ite r tz
C coes ua
quart
z +q
eite lbite I
Eclogite HP jad a
Pressure (GPa)
te wet solidus
alteration
t Inherited
B his Granulite
e sc
Blu
growth from melt

1
oli
ist
Inherited L
ch
hib
dissolution-precipitation
ns
UHT Inherited
p
ee
Am
Low grade
0
Gr
200 400 600 800 1000

Temperature (°C)
A leucosome (Qtz + Fsp) M
Zircon
Detrital
Zircon
Rutile
N
Zr expulsion from rutile
Figure 1. Typical internal zoning and textures of zircons from different metamorphic grades. Images in A,
B and N are BSE, others are CL images. Horizontal scale bar in all images is 20 microns. Labels on the out-
side of images indicate the main process responsible for zircon growth or disturbance. The central diagram
summarizes the main metamorphic facies. A) Zircon overgrowth on detrital core in greenschist facies shale
[used by permission of Springer, license 3930810283470, from Rasmussen et al. (2005), Contributions to
Mineralogy and Petrology, Vol. 150, Fig. 3k, p. 149]. B) Altered inherited zircon in digenetic sandstone
[used by permission of John Wiley and Sons, license 3930080281612, from Hay and Dempster (2009),
Sedimentology, Vol. 56, Fig. 4A, p. 2181]. C) Zircon with core altered during sea-floor alteration and rim
formed during high-pressure metamorphism, same sample as described in Spandler et al. (2004). D) Zircon
with inherited core and two metamorphic rims from eclogitic micaschist [used by permission of Nature Pub-
lishing Group, from Rubatto et al. (2011), Nature Geoscience, Vol. 4, Fig. 2a, p. 339]. E) Zircon formed in a
fluid vein within eclogite [used by permission of Elsevier, from Rubatto and Hermann (2003), Geochimica
et Cosmochimica Acta, Vol. 67, Fig. 4b, p. 2179]. F) Zircon with inherited core and two metamorphic rims
from UHP whiteschist [used by permission of Springer, license 3930090407326, from Gauthiez-Putallaz
et al. (2016), Contributions to Mineralogy and Petrology, Vol. 171, Fig. 3a, p. 15]. G) Zircon grown under
UHP to granulite facies metamorphic conditions in a Kokchetav gneiss, courtesy of A. Stepanov. H) Fir-tree
sector zoning in metamorphic zircon from eclogite [used by permission of Elsevier, license 3930810027220,
Root et al. (2004), Earth and Planetary Science Letters, Vol. 228, Fig. 3a, p. 330]. In this case the process for
zircon formation is unclear. I) Zircon from granulite with two metamorphic overgrowths around inherited
core [used by permission of John Wiley and Sons, license 3930090711788, from Hermann and Rubatto
(2003), Journal of Metamorphic Geology, Vol. 171, Fig. 3a, p. 15]. L) Zircon from low temperature mig-
matite with two metamorphic overgrowths around inherited core [used by permission of Springer, license
3930090864124, from Rubatto et al. (2009), Contributions to Mineralogy and Petrology, Vol. 158, Fig. 3l,
p. 708]. M) Zircon from a leucocratic vein that records the age of UHT metamorphism [used by permis-
sion of Oxford University Press, license 930800811618, from Harley and Nandakumar (2014), Journal of
Petrology, Vol. 55, Fig. 8b, p. 1978]. Note the feathered texture and a multiphase inclusion in the core. N)
Microzircons around rutile grain formed by expulsion of Zr during recrystallization of rutile upon cooling
from UHT metamorphism [used by permission of Springer, license 3930091093780, from Ewing et al.
(2013), Contributions to Mineralogy and Petrology, Vol. 165, Fig. 5d, p. 766].
264 Rubatto
interaction with fluids. Alteration is the term used, for example, for old relict cores that have
only partly lost their U–Pb and other isotopic signatures, and may have ghost or fuzzy zoning.
Replacement is an in situ process that changes the chemical composition of an existing
domain, occurs at sub-solidus conditions and is commonly aided by fluids. Recrystallization
is another term that is often used to describe this process. Replaced/recrystallized zircon
domains show evidence of complete resetting of the chemical/isotopic system, a sharp
boundary with inherited domains, and lack regular growth textures (shape or internal
zoning), suggesting the domain formed in replacement of pre-existing zircon. One of the
best investigated and widely recognized replacement processes that forms metamorphic
zircon in sub-solidus conditions is in situ dissolution–precipitation (Geisler et al. 2003a,
2007; Tomaschek et al. 2003; Rubatto et al. 2008; Putnis 2009).
Overgrowth or new growth indicates a new crystal domain that shows sharp boundaries
with any existing relict core and regular growth textures (shape or internal zoning). These
domains have distinct chemical and isotopic compositions that are generally homogenous
within a single domain. This type of zircon is commonly found in melt or fluid-rich systems
and is caused by crystallization from a melt or precipitation from a fluid. Overgrowths on
detrital zircon grains at very low grade would also be included in this category.
This terminology has to be taken as zircon specific, and may not apply to other minerals.
While many zircons in metamorphic rocks can be described with these three categories, it
is evident that individual cases exist that do not clearly fit a single category as multiple
processes may affect the same zircon population, and the distinction between one and
another may be blurred (Spandler et al. 2004; Tichomirowa et al. 2005; Zheng et al. 2005;
Chen et al. 2011; Gao et al. 2015).
PETROGRAPHY OF ZIRCON
Textural relationships and inclusions
Metamorphic petrology is grounded on careful textural observations to establish mutual
relationships between minerals. Defining which mineral is coexistent with others (paragenesis)
based on textural equilibrium, inclusion relationships and composition is the backbone of
metamorphic petrology. Textural criteria are often questionable when small accessory and
refractive minerals like zircon are involved (Fig. 2). Small zircons are commonly included
in larger grains, but this does not mean that the host and the inclusion are in equilibrium.
A common observation under the optical microscope is the presence of inherited zircons
included in key metamorphic minerals, that despite the apparent equilibrium texture with
straight grain boundaries, have no petrological relationship with the host mineral (Fig. 2B).
Zircon in gneisses, eclogitic metagabbros, and amphibolites that are included in metamorphic
garnet, pyroxene or amphibole may be inherited and unrelated to metamorphism. Textural
relationships between major minerals and zircon (and refractory accessory minerals in
general) alone are not a robust criterion for age interpretation, particularly at low to medium
metamorphic grade. A more robust link between the stability of major minerals and zircon
can be established based on mineral inclusions in zircon (see below).
Petrography has proven a powerful and necessary tool to identify low-grade metamorphic
zircon that overgrows detrital grains (Rasmussen 2005; Hay and Dempster 2009b). The small
size and sawtooth-shape of these overgrowths is so characteristic that the texture alone is
a strong evidence of metamorphic growth (Fig. 1A). Similarly, petrography is crucial in
identifying micro-zircon that may form by exsolution or metamorphic reactions during cooling
and breakdown of Zr-rich minerals. Examples are micro zircons in cordierite coronas around
garnet (Fig. 2C, Degeling et al. 2001) or micro zircon around ilmenite and rutile (Bingen et
500 µm Garnet
Zircon
Amphibole
Omphacite
Allanite
A 100 µm B
C D
Cordierite
Inherited
Garnet zircon
Garnet
50 µm
Figure 2. Textural relationships between zircon and other metamorphic minerals. A) Zircon in apparent
textural equilibrium with allanite (brown) in an eclogite facies rock; zircon and monazite are both meta-
morphic but differ in age by 10 Ma [used by permission of Mineralogical Society of America, from Rubatto
et al. (2007), American Mineralogist, Vol. 94, Fig. 1, p. 1521]. B) Inherited magmatic zircon included in
amphibole in an eclogite. C) Micro-zircons (indicated by arrows) in the cordierite corona around garnet,
which formed during decompression from granulite facies [used by permission of Mineralogical Society of
Great Britain and Ireland, license 3930121477614, from Degeling et al. (2001), Mineralogical Magazine,
Vol. 65, Fig. 2b, p. 752]. D) Zircon included in garnet in amphibolite; most of the zircon is inherited and
only the thin rim (indicated by arrows) is metamorphic, same sample as described in Buick et al. (2006).
al. 2001; Ewing et al. 2013). Unfortunately the size of these zircons is commonly below the
common spatial resolution of microbeam dating techniques (10–30 µm, Fig. 1N and 2C), but
their presence still provides important information on the petrogenesis of the rocks.
Particularly important is the use of petrography to recognize and characterize mineral
inclusions in zircon. These inclusions are not only valuable for relating zircon ages to
metamorphic assemblages, but may also provide unique petrological information on the P–T
evolution of the host sample. The most striking example is from high and ultra-high pressure
rocks (UHP), were prograde to peak mineralogy is easily replaced during decompression.
Coesite is a key indicator mineral for ultra-high pressure metamorphism, but a robust container,
such as garnet or zircon, is often needed to preserve coesite in natural rocks. Additionally, during
decompression of felsic UHP rocks at T > 700 °C, phengite melting occurs leading to a pervasive
recrystallization of the main rock-forming minerals (Hermann et al. 2006c). For this reason,
relicts of UHP metamorphism in felsic rocks are most commonly found in refractory minerals
such as garnet or zircon. In the subducted continental rocks of the Kokchetav Metamorphic
Complex, diamonds in gneisses and marbles are mainly found as inclusions in zircon and garnet
(Shatsky and Sobolev 2003). The spectacular record of peak to retrograde inclusions in zircon
has proven a key tool for age interpretation (Hermann et al. 2001; Katayama and Maruyama
2009). In the vast Dabie-Sulu orogen, the extent of the crust subducted to UHP conditions could
only be demonstrated through the widespread occurrence of coesite inclusions in zircon from
gneisses (Ye et al. 2000) that otherwise preserve little or no relict of UHP assemblages.
266 Rubatto
Petrography of mineral inclusions in zircon must however deal with the possibility of
secondary inclusions (Fig. 3). Altered and metamict zones in inherited zircon cores have
been proven to contain secondary inclusions, as for example high pressure minerals in
“magmatic” zircon (Gebauer et al. 1997; Zhang et al. 2009a; Gauthiez-Putallaz et al. 2016).
These secondary inclusions may be identified because of disturbed cathodoluminescence
zoning, fractures, porosity and evidence of metamictization in the zircon around the
inclusions. More controversial is the finding of white mica inclusions (Hopkins et al.
2008; Rasmussen et al. 2011) and carbon-phase inclusions (Menneken et al. 2007) in Early
Archean zircons from the Jack Hills quartzite, where the debate is ongoing to what extent
these inclusions are primary or secondary (see discussion in Harrison et al. 2017).
Figure 3. Secondary metamorphic inclusions in

inherited zircon cores that underwent high pres-
sure metamorphism. The cores also show dis-
CL image BSE image turbance of the original zoning likely due to al-
teration by metamorphic fluids. A) Zircon from
hic
talc micaschist recovered from the Chinese Conti-

morp
nental Scientific Drilling Main Hole in the Sulu

phengite orogeny, China [used by permission of John
meta
Wiley and Sons, license 3930800210763, from

Zhang et al. (2009a), Journal of Metamorphic
fluid Geology, Vol. 27, Fig. 3c, p. 321]. B) Zircon
50 µm
50 µm from a whiteschist of the Dora Maira unit, West-
ern Alps, Italy [used by permission of Springer,
A B phengite license 3930130984661, from Gauthiez-Putal-
laz et al. (2016), Contributions to Mineralogy
and Petrology, Vol. 171, Fig. 3b, p. 15].
Internal zoning
It is impossible to overstate the importance of characterizing internal zoning in metamorphic
zircon: recognizing the presence of detrital cores, relict protolith magmatic zircon, and multiple
domains formed during metamorphism is a fundamental step in the correct interpretation of
any zircon geochemical information and eventually age. This is most commonly achieved with
panchromatic cathodoluminescence (CL) or back-scattered-electron (BSE) imaging, which have
become ubiquitous tools in zircon studies. Since the early applications to metamorphic zircon
it became clear that CL and BSE images reveal internal structure otherwise invisible in light
microscopy or etching (e.g., Hanchar and Miller 1993; Hanchar and Rudnick 1995; Vavra et
al. 1996). CL and BSE emission are both proxies for chemical signals and are broadly anti-
correlated because intrinsic CL attributed primarily to Dy is suppressed by the heavy element U,
which increases the BSE signal (Rubatto and Gebauer 2000; Poller et al. 2001). Composition is
however not the only player in CL emission, which is also controlled by structural parameters
such as crystallinity or the presence of defect centers (Nasdala et al. 2002). The application of
CL to metamorphic zircon has revolutionized its use in metamorphic petrology, making it easy
to identify distinct growth zones and, to some degree, deformation features.
There have been a few attempts to categorize zoning (particularly CL zoning) of
metamorphic zircon (Rubatto and Gebauer 2000; Corfu et al. 2003; Rubatto and Hermann
2007b), but more commonly every case has been presented in separate studies. The variety
of textures in zircon is extremely wide, but some general systematics exist (Fig. 1, Table 1).
It is commonly recognized that metamorphic zircon domains (including zircons forming in
anatectic melts, which are considered metamorphic in this paper) have weak zoning when
compared with the marked oscillatory and sector zoning of magmatic zircon (see a review
in Rubatto and Gebauer 2000; Corfu et al. 2003). Zircon formed in sub-solidus conditions
most commonly shows no regular zoning, having either a homogeneous CL/BSE emission or
a cloudy and irregular zonation. However exceptions exist particularly for zircons attributed
to fluid-related processes like in metamorphic veins or jadeitites. Weak internal zoning also
characterizes metamorphic zircon that crystallized in high-grade rocks, likely from anatectic
melts. In this case, weak oscillatory, sector, or fir-tree zoning is more common (Vavra et al.
1996; Schaltegger et al. 1999; Corfu et al. 2003; Claesson et al. 2016). High grade metamorphic
zircon often displays a relatively low CL emission, possibly related to a high U content (see
also section on Th–U composition), but the opposite has also been observed (Vavra et al. 1996;
Corfu et al. 2003; Fu et al. 2008). Notably low-U zircon domains that are irregular in shape and
form embayments into magmatic zircon, and thus look quite similar to metamorphic domains,
are also formed during late-magmatic processes (Corfu et al. 2003). While CL, and to a lesser
extent BSE imaging remains a powerful tool to recognize different growth domains, additional
data are often required to determine the growth environment of metamorphic zircon.
Zircon imaging by more advanced techniques such as Electron Back Scattered Diffraction
(EBSD), Transmission Electron Microscopy (TEM), and element mapping is more time
consuming than CL and BSE imaging. These advanced techniques are applied to cases where
deformation or particular chemical information is targeted (Reddy et al. 2008, 2010; Austrheim
and Corfu 2009; Timms et al. 2011; Piazolo et al. 2012; Vonlanthen et al. 2012). It has been
proposed that magmatic and metamorphic zircon can be distinguished based on Raman spectra
(Xian et al. 2004), but this application remains limited and widely untested. The analytical
approach for Raman identification has indeed been contested (Nasdala and Hanchar 2005)
because it is based on a laser-induced photoluminescence peak. Given the possible difference
in trace element composition between magmatic and metamorphic (i.e., lower in rare earth
elements—REE—and Th) zircon, it is plausible that in some cases a distinction based on
spectroscopy may work. The greatest use of Raman spectroscopy is in the documentation of
completely to partly metamict zircon, where the amorphisation process leads to changes in the
wavenumber and half-width of the Raman bands (Nasdala et al. 2003).
The superior spatial resolution of TEM analyses (McLaren et al. 1994; Hay and
Dempster 2009b; Hay et al. 2010; Vonlanthen et al. 2012) and, more recently, atom probe
(Valley et al. 2014) are promising investigative tools, but they require advanced sample
preparation and are partly destructive. Their future application to recrystallization fronts
and domain boundaries within metamorphic zircon may be particularly interesting for the
understanding of processes of metamorphic zircon formation.
Deformation
There is no regional metamorphism without deformation and thus the effect of deformation
on metamorphic zircon systematics must be taken into account. Evidence of crystal-plastic
deformation of zircon crystals affecting composition and most importantly U–Pb systematics
mainly come from high temperature shear zones (Reddy et al. 2006; Timms et al. 2006; Austrheim
and Corfu 2009; Piazolo et al. 2012), although reports from unfoliated rocks also exist (Timms
et al. 2011). Large zircon crystals (mm in size) in rocks that deformed under amphibolite- to
granulite-facies conditions (> 700 °C) display intragrain crystallographic misorientation of
2–20° at the crystal tips (Reddy et al. 2006, 2007; Timms et al. 2006; Piazolo et al. 2012). This
miss-orientation correlates with panchromatic CL emission (reduced CL at the loci of low angle
boundaries), REE composition (increase in total REE and of middle-REE—MREE—with
respect to heavy-REE—HREE) or increase in Th/U (Timms et al. 2006; Piazolo et al. 2012). In
some cases, deformation results in microfractures that define small subgrains that are misoriented
by up to 10° and contain less Ti than the original crystal (Timms et al. 2011; Piazolo et al. 2012).
Planar deformation features have been observed in zircon from pseudotachylytes (Austrheim
and Corfu 2009). While these features are common in impact-related minerals, in this case they
have been attributed to extreme strain rate during seismic deformation.
268 Rubatto
Studies agree that the dislocations and deformation features act as fast diffusion pathways
for trace elements, U, Th and Pb. The creation of subgrains may also enhance chemical
exchanges with any alteration fluid due to high surface area (Piazolo et al. 2012). How much
this deformation disturbs the age is not always clear because its detection depends on the relative
timing of crystallization versus deformation. Partial to complete resetting of U–Pb ages in the
deformed domain is observed in some cases (Timms et al. 2011; Piazolo et al. 2012).
While full characterization of deformation features (best done by EBSD) and degree of
chemical and isotopic resetting may not always be possible, panchromatic CL images can give
a first hint on the presence of deformation. In all cases reported, a general correspondence
between low-angle boundaries and low CL emission exists and such features should be a
warning for any isotopic analysis. This correspondence is in agreement with observed recovery
of CL emission by annealing of crystal defects (Nasdala et al. 2002).
MINERAL CHEMISTRY
A few chemical indicators have been commonly used in identifying metamorphic
zircon and to create links between measured ages and metamorphic conditions (see also
Table 1). Chemical criteria that can relate zircon composition to metamorphic assemblages
are particularly useful for age interpretation. This however requires that the chemical (trace
elements including Th and U) and isotopic (Pb) systems are equally robust. Experimental studies
indicate that diffusion of the large divalent Pb2+ ion is comparable to that of trivalent REE and
orders of magnitude faster than tetravalent ions like Th and U, which are essentially immobile
under most geological conditions (Cherniak and Watson 2003). Additionally, radiogenic Pb
is internally produced and may not be bonded in the crystal lattice in a structural site, and
this might enhance its capacity to escape the crystal. Decoupling of U–Pb ages and element
abundances has been reported for samples that have experienced high temperature (Kusiak
et al. 2013a), metamictization or intense deformation (Reddy et al. 2006; Timms et al. 2006,
2011). Kaczmarek et al. (2008) reported zircon from deformed metagabbros that preserved
magmatic REE patterns, but whose apparent ages varied between that of the protolith and
of later metamorphism. Zircons from the Dabie-Sulu high pressure rocks commonly contain
relicts of magmatic zircon, that may still have their original high Th/U or steep REE patterns,
but whose U–Pb system has been reset to the age of metamorphism (Zheng et al. 2005; Xia et
al. 2009). Studies of natural samples that suggest diffusional re-equilibration of REE, but not
of Th and U, as predicted by diffusion experiments, are lacking.
Th/U systematics
The Th–U composition of zircon is routinely measured during dating and thus has become
an easy-to-acquire and widely used criteria for zircon classification. The Th/U of metamorphic
zircons is generally < 0.1, but exceptions do exist (see below). This criteria was proposed
based on the study of low temperature, high pressure zircons (Rubatto and Gebauer 2000), and
it has been proven valid in countless cases. Exemplary are numerous studies of eclogite-facies
zircon, and zircon in migmatites and granulites (Fig. 4). In general terms, the robustness of this
simple chemical criteria appears to be independent of the process that led to zircon formation,
from solid-sate replacement to crystallization from anatectic melts (Zhao et al. 2015). It is also
important to note that the opposite is also true: most magmatic zircons have Th/U > 0.1 unless
altered (e.g., Belousova et al. 2002; Grimes et al. 2015).
It has been demonstrated that metamorphic zircon does not always have low Th/U.
The most occurrences of metamorphic zircon with Th/U > 0.1 are from high and ultra high
temperature (> 900 °C) samples [(Vavra et al. 1996; Schaltegger et al. 1999; Möller et al.
2003; Kelly and Harley 2005) see also a discussion in (Harley et al. 2007)]. The incorporation
10000
UHT Estern Gats
10000
Eclogites, WGR
D. Granulites, UHT rocks and ultramafics

Eclogites, Sulu
1000
Ultramafic
1000
UHT Regaland
B. UHP crust
Gneisses, Bohemian
U (ppm)
Whiteschists, DM
Malenco
100
100
UHT Antarctica
Gneisses, Kokchetav
Hojazo
10
10
Gneisses, Sulu
HT-UHP Ivrea
1
1
10
0.1
0.01
0.001
1
0.1
0.01
0.001
A. Various HP rocks (eclogites, veins, metasediments)
10000
10000
Metasediments, Caledonia
Metasediments, Rhodope
C. Migmatites: metapelite and metatonalites

Metasediments, Alps
Australia Stafford
1000
1000
Alaska
U (ppm)
Eclogites, West Africa
Alps
Eclogites, Australia
100
100
Australia Reynolds Range
Eclogites, PNG
Himalaya
10
10
Leucogabbro, Alps
Eclogites, Alps
Veins, Alps
1
1
0.1
0.01
0.001
0.001
0.01
0.1
1
Th/U Th/U
Figure 4. Th/U versus U content (in ppm) of metamorphic zircon from different tectonic settings. The
compilation is based on circa 1400 published analyses (Vavra et al. 1996; Gebauer et al. 1997; Rubatto
and Gebauer 2000; Hermann et al. 2001, 2006b; López Sánchez-Vizcaíno et al. 2001; Rubatto et al.
2001, 2006, 2008, 2009, 2013; Möller et al. 2002; Cesare et al. 2003; Hermann and Rubatto 2003; Rub-
atto and Hermann 2003; Root et al. 2004; Kelly and Harley 2005; Spandler et al. 2005; Bauer et al. 2007;
Zhang et al. 2009b; Gasser et al. 2012; Gordon et al. 2012; Ewing et al. 2013; Korhonen et al. 2013;
Kylander-Clark et al. 2013; Ganade de Araujo et al. 2014; Phillips et al. 2015; Rubatto and Angiboust
2015; Gauthiez-Putallaz et al. 2016; Stepanov et al. 2016b). See text for discussion.
270 Rubatto
of Th in zircon is primarily controlled by the availability of Th and U in the system and

partitioning with other phases. The common presence in crustal metamorphic rocks of Th-
rich phases such as monazite and allanite is an obvious reason for low Th/U in coexisting
metamorphic zircon in eclogite, amphibolite and granulite facies rocks. The absence of
these phases in some crustal rocks (either by melting under ultra-high temperature—UHT,
or because of composition) should produce metamorphic zircon with high Th/U.
A compilation of Th/U versus U content of metamorphic zircon in different tectonic
settings shows some interesting systematics (Fig. 4, ca. 1400 analyses). Data are grouped in 4
categories representing different metamorphic conditions.
(A) Relatively low temperature, high-pressure rocks of various compositions from mafic
eclogites and leucogabbros, to micaschists and metamorphic veins (Rubatto and Gebauer
2000; Rubatto and Hermann 2003; Spandler et al. 2005; Bauer et al. 2007; Rubatto et al. 2008;
Gordon et al. 2012; Ganade de Araujo et al. 2014; Phillips et al. 2015; Rubatto and Angiboust
2015). In all cases metamorphic temperatures are below the solidus and there is no evidence
of melting in the rocks. According to the studies, most metamorphic zircon in these samples
formed under HP conditions. In such “cold” eclogites Th/U of metamorphic zircon is mostly,
but not restricted to < 0.1, with values between 0.001 and 0.6. U contents are also variable
from a few to 1000s of ppm, but mostly below 1000 ppm. The variability in zircon Th–U
composition of HP rocks overall is large compared to any other category. The large range
reflects the variety of rock types but also the lack of a dominant buffering phase. Monazite
is not a common mineral in these rocks, but allanite is present in many samples. Another
secondary effect that influences Th/U in these samples may be the temperature dependence
of Th incorporation in zircon: the relatively larger Th+4 ion may fit proportionally less in a
lower-T crystal structure than the smaller U+4 ion (Rubatto and Gebauer 2000).
(B) In mafic and felsic crustal rocks that underwent UHP conditions and thus higher
temperatures of re-equilibration, metamorphic zircon Th–U composition is more restricted.
Uranium ranges between 10–2500 ppm (at least in the selected samples) and Th/U is mainly
below 0.2, with less than 10% of data (total analyses 524) higher than this value. As documented
by Stepanov et al. (2016b) for some UHP–T gneisses of the Kokchetav metamorphic complex,
the relatively high solubility of monazite in ultrahigh temperature melts (Stepanov et al.
2012; Stepanov et al. 2014) will allow zircon crystallization with high Th/U in some rocks.
Indeed, in one Kokchetav sample it has been documented that prograde metamorphic zircon
cores with low Ti-contents have Th/U < 0.1, consistent with coexisting monazite, whereas
peak metamorphic zircon domains with high Ti contents have Th/U of 0.4–0.6 and formed at
1000 °C, 5 GPa when monazite was completely dissolved in the partial melt. Zircon rims with
low Ti that formed during retrograde crystallization of melts, when monazite is again present
in the assemblage, show a low Th/U < 0.1 (Stepanov et al. 2016a). Zircons in gneisses from the
Bohemian Massif UHP unit have a restricted composition (Th/U 0.02–0.1, Kylander-Clark et
al. 2013). UHP mafic eclogites from the Western Gneiss Region (Root et al. 2004; Kylander-
Clark et al. 2013) and from the Dora Maira whiteschists (Gebauer et al. 1997; Gauthiez-
Putallaz et al. 2016) also have zircon Th/U consistently below 0.1. Zircon rims from the Sulu
UHP mafic and felsic rocks show higher values up to 0.4 (Zhang et al. 2009b).
(C) Zircon in migmatites, where a significant amount of leucosome is preserved,
consistently has Th/U at 0.1 or below. This is independent of metamorphic temperature or
pressure, from the 800 °C and 9 kbar of the Himalayan Higher Crystalline (Rubatto et al. 2013),
to the water assisted melting at 650–700 °C, 5–13 kbar in the Central Alps (Rubatto et al. 2009)
and Alaska Chugach complex (Gasser et al. 2012), including low pressure migmatites of
central Australia. Most of these samples are metapelitic migmatites, where monazite is always
an abundant accessory mineral in both paleosome and leucosome. In metatonalites from the
central Alps, allanite is a nearly ubiquitous accessory (Rubatto et al. 2009). Notably, in the
metapelites the U content of metamorphic zircon is also quite restricted, never below ~100 ppm,
whereas it can be as low as 10 ppm in the migmatitic tonalites. In migmatites of intermediate
composition where neither allanite nor monazite are stable then higher Th/U are expected (see
an example in the Lewisian granulites of Norhern Scotland Whitehouse and Kemp 2010). The
remarkably consistent Th–U composition of zircons in many migmatites may be also related to
the presence of partial melts, which are a Th- and U-bearing phase. Experimental studies show
that the relative partitioning of Th and U between zircon and granitic melt does not significantly
change with T in the range 800–1050 °C (Rubatto and Hermann 2007a). In very oxidized
environments, some of the U might occur as 5+ or even 6+ cation, which are significantly
more incompatible than U4+ (Burnham and Berry 2012). As Th remains as 4+ cation this might
potentially contribute to high Th/U in highly oxidizing environments.
(D) UHT rocks (T > 900 °C) are a distinct case in Th–U metamorphic zircon composition,
as the majority of data plot above Th/U of 0.1 with values as high as 3. Samples include the
pigeonite-bearing granulites from the Rogaland anorthosite complex (Möller et al. 2002),
the saffirine-bearing orthogneiss and charnokite of the Napier Complex in Antarctica (Kelly
and Harley 2005), enderbite and migmatitic gneisses of the Eastern Ghats belt in India
(Korhonen et al. 2013), and metapelitic rocks of the lower crustal section of the Ivrea Zone
(Vavra et al. 1996; Ewing et al. 2013). Note that some of the UHP samples from the Kokchetav
massif plotted in category B also recorded T of 900–1000 °C and Th/U can be > 0.1. Lower
crustal metapelites that did not reach temperatures > 850 °C and where monazite is stable
are plotted in Figure 4D for comparison (Hojazo and Malenco, Cesare et al. 2003; Hermann
and Rubatto 2003); these relatively lower T granulites have very low Th/U (0.001–0.01)
and U content is above 100 ppm. Zircon from ultramafic rocks where there is no stable
Th-phase are also included in this plot and indeed they show Th/U of 0.1–1. Examples are
metamorphic zircons from a HP metapyroxenite (López Sánchez-Vizcaíno et al. 2001) and
zircon from the Duria garnet peridotite (Hermann et al. 2006b).
Rare earth elements
Rare earth elements (REE) and particularly mid to heavy REE (M-HREE, Sm–Lu) can
also be used for recognizing metamorphic zircons. The principle is based on partitioning with
co-existing minerals that sequester M-HREE in the metamorphic assemblage (Fig. 5). Garnet is
commonly a main host of HREE in medium to high-grade mafic to pelitic metamorphic rocks.
Zircon that grows in a garnet-rich assemblage, where HREE are sequestered in garnet, will
show a relatively flat HREE pattern compared to the HREE enrichment in magmatic zircon.
This low HREE signature has been widely reported for metamorphic zircon in garnet-bearing
eclogitic and granulitic rocks (Schaltegger et al. 1999; Rubatto 2002; Hermann and Rubatto
2003; Rubatto and Hermann 2003; Whitehouse and Platt 2003; Bingen et al. 2004; Gilotti
et al. 2004; Hokada and Harley 2004; Root et al. 2004; Kelly and Harley 2005; Hermann et
al. 2006a; Rubatto et al. 2006, 2013; Wu et al. 2008a; Wu et al. 2008b; Fornelli et al. 2014;
Whitehouse et al. 2014; Gauthiez-Putallaz et al. 2016).
Similarly, Eu deficiency relative to other REE in zircon (i.e., negative Eu anomaly) has been
attributed to the presence of feldspars that sequester Eu (Schaltegger et al. 1999; Rubatto 2002).
The fact that zircon mainly incorporates Eu3+, whereas feldspars take up mainly Eu2+ is a further
complication (see also a discussion in Kohn 2016). Two main metamorphic conditions have been
related to changing Eu anomaly in metamorphic zircon. (i) In eclogite facies assemblages, where
albite breaks down to jadeite and quartz, zircon commonly has a weak or no Eu anomaly, at least in
rocks that have no strong bulk Eu anomaly. As eclogitic assemblages also commonly yield garnet,
the lack of a negative Eu-anomaly is coupled to a relatively flat HREE pattern (Fig. 5A) (Rubatto
2002; Rubatto and Hermann 2003; Baldwin et al. 2004; Gilotti et al. 2004; Wu et al. 2008a,b;
Gauthiez-Putallaz et al. 2016) (ii) In migmatites, melting reactions involving micas produce
272 Rubatto
10000 peritectic K-feldspar that incorporates all

on the available Eu2+. Metamorphic zircon
zirc (and monazite) growing from anatectic
1000 atic
a gm melts acquires a stronger negative Eu
m
100 anomaly relative to their protolith or sub-
eclogitic zircon
solidus counterpart (Fig. 5B, Schaltegger
mineral / chondrite
10
et al. 1999; Rubatto et al. 2006, 2013).
garnet Similarly, metamorphic zircon growing
in an assemblage rich in L-MREE phases
1
such as titanite, allanite or monazite can
develop a particularly light-REE (La–
0.1
Nd) depleted pattern (Fig. 5C). This has
been observed for example in zircon from
0.01 the amphibolite-facies migmatites of the
A Central Alps, which are rich in accessory
0.001 allanite and titanite (Rubatto et al. 2009).
La Ce Pr Nd Sm Eu Gd Tb Dy Ho Er Tm Yb Lu
It is important to bear in mind that,
10000 n as already stated for Th/U, these REE
o
z irc
tic signatures are not absolute and depend
1000 ma on a number of factors: (i) bulk rock
g
ma granulitic zircon
composition, e.g., in rocks strongly
100 enriched in HREE, both garnet and
metamorphic zircon will have relatively
mineral / chondrite
10 garnet
high HREE; zircon growing in a feldspar-
free, HP assemblage may still have a
1 negative Eu anomaly if the bulk rock is Eu-
feldspar
depleted; e.g., HP zircon in some Kokchetav
0.1 gneisses (Hermann et al. 2001) and Dora
Maira whiteschists (Gauthiez-Putallaz et
0.01 al. 2016). (ii) The volume percent of the
B HREE or Eu controlling phase: rocks in
0.001 which garnet is only a minor phase may
La Ce Pr Nd Sm Eu Gd Tb Dy Ho Er Tm Yb Lu still have metamorphic zircon with high
1000000
HREE contents. (iii) The presence of other
phases controlling the HREE or Eu budget,
monazite
100000 as for example abundant orthopyroxene
allanite
ir con that can accommodate significant HREE
10000 atic z
a gm (Fornelli et al. 2014) and will increase the
m
1000 titanite HREE depletion in granulite facies zircon.
mineral / chondrite
100 Figure 5. Representative REE patterns of zircon

con types and other relevant minerals, normalized to
10 zir Chondrite values. Magmatic zircon is plotted for
s
cie reference in all diagrams. Relevant REE-bearing
1 fa minerals coexisting with metamorphic zircon in
it e- eclogite (A), granulite (B) and amphibolite fa-
0.1 bol cies (C) assemblages are represented in the re-
phi
spective diagrams to illustrate the competition
0.01 am for REE and the consequent REE signature of
C metamorphic zircon in each assemblage. See text
0.001 for discussion.
La Ce Pr Nd Sm Eu Gd Tb Dy Ho Er Tm Yb Lu
The use of HREE in linking metamorphic zircon to garnet in the co-existing assemblage
can be exploited further if the equilibrium partitioning between zircon and garnet is known
for different temperatures and garnet compositions. In samples with zoned garnet and
multiple metamorphic zircon growth this could lead to identifying which specific garnet and
zircon growth are in equilibrium (Hermann and Rubatto 2003). Experimental and empirical
trace element zircon/garnet equilibrium partitioning vary over an order of magnitude,
particularly for the HREE (Table 2) (Rubatto 2002; Hermann and Rubatto 2003; Rubatto
and Hermann 2003, 2007a; Hokada and Harley 2004; Kelly and Harley 2005; Buick et al.
2006; Rubatto et al. 2006; Taylor et al. 2014). Element partitioning between two phases
is firstly controlled by temperature, and secondarily by composition, whereas pressure is
likely to have a negligible effect (Rubatto and Hermann 2007a). The published values cover
metamorphic temperatures from 550–1000 °C for natural samples and 800–1000 °C for
experiments, and variable garnet compositions from 0 to 8 wt% CaO (subset of studies
for which the garnet composition is provided). In this wide range of conditions, variations
are expected. Taking as an example Yb, the most abundant HREE in both minerals and
thus relatively easy to measure, correlations between these parameters can be seen (Fig. 6).
The zircon/garnet partition coefficient for Yb shows a negative correlation with T when
the experimental studies are considered. This T-dependence is confirmed by some natural
samples at 1000 °C that have a low zircon/garnet partition coefficient for Yb of 0.4–1.2,
whereas granulites at ~800 °C have a Yb partition coefficient of 5–17. Both experimental
studies and natural samples at these T are equilibrated with melts. Notably, the three samples
that recorded lower metamorphic temperatures, where melt was not present, fall off the trend
defined by melt-present samples/experiments. For the subset of studies that report garnet
composition, most data also show a correlation between zircon/garnet Yb partitioning and
grossular component in garnet. This highlights an additional complexity in the application of
equilibrium partitioning as garnet major element composition varies widely, whereas zircon
major composition remains constant. More systematic studies are needed to fully map out
the effects of temperature and garnet composition on HREE partitioning.
In samples where garnet is zoned and zircon has multiple growth zones partition
coefficients could be used to recognize equilibrium versus disequilibrium growth. In general,
it is easier to detect when garnet and zircon are clearly not in equilibrium than when they
potentially are. The partitioning can be strongly affected by lack of preservation of growth
zones and original REE composition. Resorption of garnet or zircon during decompression
or melting would impede correct partitioning determination. In rocks that reach relatively
high temperatures, garnet commonly grows during the prograde evolution, whereas zircon
is more likely to form during melt crystallization upon cooling, and thus the two phases are
100.0 100.0
natural samples experiments
17.3 17.3
zircon/garnet Yb
10.0 8.6 10.0 8.6 8.4

5.3 5.8 4.6 5.3 5.8
4.6
2.9 2.9
1.8 1.8
1.0 1.2 1.0 1.0 1.2
subsolidus 0.7 0.6 0.7 0.6
0.7 0.4 0.4
0.1 0.1
400 500 600 700 800 900 1000 1100 0.00 0.05 0.10 0.15 0.20 0.25
T (°C) X grossular in garnet
Figure 6. Zircon/garnet Yb partition coefficient from experimental studies and natural samples plotted
against T (A) and garnet composition represented by molar proportion of grossular (B). The dotted ellipse
in A groups natural samples that did not reach melting conditions. The number next to each symbol is the
partition coefficient. Data are from Table 2 that also contains the references. See text for discussion.
Table 2. Zircon/garnet partition coefficient for middle to heavy REE obtained from natural rocks and experiments. The major element garnet 274
composition for each sample is reported when available.
Rock type Reference Gd Tb Dy Ho Er Tm Yb Lu Y T (°C) XGrs XAlm XPrp XSpess
HP vein Rubatto and Hermann 2003 1.6 1.6 2.1 2.9 3.6 4.5 5.3 6.0 3.4 550 0.15 0.55 0.30 0.001
HP schist Rubatto 2002 0.3 0.4 0.7 1.0 1.2 1.5 1.8 2.1 1.3 600 0.14 0.68 0.16 0.018
Gneiss Buick et al. 2006 2.3 2.5 3.8 5.6 8.2 10.9 15.5 800 0.016 0.63 0.35 0.006
Granulite Hermann and Rubatto 2003 0.8 1.0 1.3 1.7 2.6 3.9 5.8 8.2 2.1 750 0.18 0.63 0.18 0.008
Granulite Rubatto 2002 1.6 1.8 2.6 4.3 7.1 11.1 17.3 23.9 4.8 800 0.017 0.81 0.16 0.015
Granulite Rubatto 2002 0.9 0.9 1.3 2.0 3.2 5.2 8.6 12.1 2.3 800 0.023 0.82 0.13 0.026
Granulite Rubatto et al. 2006 0.7 0.8 1.1 1.6 2.4 3.4 4.6 6.3 1.9 800 0.031 0.84 0.1 0.027
Granulite Whitehouse and Platt 2003 0.5 0.7 0.8 0.8 1.0 0.7 0.7 0.8 700
UHT gneiss Hokada and Harley 2004 1.3 0.9 0.7 0.7 0.7 0.7 0.6 0.7 1000 0.024 0.52 0.46 0.003
Rubatto
UHT gneiss Kelly and Harley 2005 0.7 0.7 0.7 0.8 0.9 0.9 1.0 1.2 1000
Experiment Rubatto and Hermann 2007 1.0 2.8 5.5 8.4 11.6 3.9 800 0.22 0.5 0.3 0.0
Experiment Taylor et al. 2014 0.6 0.5 0.5 0.5 0.5 0.6 0.7 0.8 900 0.0 0.53 0.47 0.0
Experiment Taylor et al. 2014 0.7 0.6 0.6 0.6 0.7 0.8 0.9 1.1 950
not necessarily in trace element equilibrium. An additional complication is diffusional re-

equilibration, which can affect particularly the garnet composition: it has been shown that
above 700–900 °C, diffusional equilibration of trace elements in garnet occurs over geological
timescales, whereas zircon still preserves prograde growth stages (Stepanov et al. 2016b). As
it is not trivial to detect resorption or diffusion in garnet and zircon, and textural coexistence
is not a valid argument to prove chemical equilibrium, the use of HREE zircon/garnet
partitioning for a direct link between metamorphic zircon and garnet growth zones must be
carefully evaluated case by case. However, the general rule that metamorphic zircon has flat
HREE in garnet-rich assemblages, and variable Eu depending on the presence of feldspar has
been observed in numerous studies and remains a useful tool for relating metamorphic zircon
to assemblages and ultimately metamorphic conditions.
Ti-in-zircon thermometry
The Ti-in-zircon thermometer is based on the principle that, in a buffered assemblage,
the incorporation of Ti in zircon depends on T (Watson and Harrison 2005; Ferry and Watson
2007). One of the main attractions of this single mineral thermometer is that it can yield
temperatures from isolated zircon crystals because in most crustal rocks the activity of SiO2
is high and a Ti-phase is present. First applied to Early Archean detrital zircons of magmatic
origin (Watson and Harrison 2005), the Ti-in-zircon thermometer has since proved a seemingly
simple tool for petrology and its application has quickly spread from detrital to magmatic and
metamorphic zircon. For temperature-dominated metamorphism the capacity to relate an age
to a temperature in the P–T path is certainly appealing. Countless studies have applied this
thermometer obtaining reasonable metamorphic temperatures, especially in upper amphibolite
facies to lower granulite facies conditions, where zircon starts to be reactive and full buffering
by quartz and a Ti-phase is achieved. Under lower and higher metamorphic grade, caution is
needed in the application of the Ti-in-zircon thermometer as discussed below.
Metamorphic zircons that have formed under low temperature in sub-solidus conditions
(< 600 °C) have low Ti as predicted by the thermometer (< 2 ppm, unpublished data, Tailby et
al. 2011). Tailby et al. (2011) have shown that different sectors in a fir-tree zoned zircon from
a vein containing rutile and quartz have variation in Ti contents that would correspond to
variations of ~40 °C. This cannot be reconciled with T oscillations during growth and indicates
that crystallographic sectors have some influence on Ti content rather than solely crystallization
temperature. This effect may be particularly relevant at very low Ti contents, i.e., low temperature,
but zircon is hard to react under such conditions and thus very few cases exist to test this issue.
For example Ti-thermometry of diagenetic or low-grade zircon has never been achieved. At low Ti
concentrations, any contamination during analysis from Ti-bearing micro-inclusions, neighboring
phases or fractures could be serious (Harrison and Schmitt 2007). This demands particular
cautions during analysis of Ti in potentially low temperature zircons. Additionally, it is much
more difficult to prove that a buffer assemblage was reactive at low metamorphic temperatures.
On the other end, under extreme metamorphic temperatures, Ti content in zircon is higher and
easier to measure (18–90 ppm at 800–1000 °C). The Ti-in-zircon thermometer can indeed record
T of 900–1000 °C, as shown in some Kaapvaal xenoliths containing two stages of zircon growth
(Baldwin et al. 2007). Extreme T is recorded by zircons from the Kokchetav diamondiferous
rocks where Ti-in-zircon thermometry returns T of 910–1080 °C (Stepanov et al. 2016b),
corroborating peak conditions of UHP metamorphism. In high grade rocks, which commonly
undergo partial melting, a major issue can instead be that zircon crystallizes upon crystallization
of melt during decompression and cooling and not at the T peak. For example, in the Anápolis-
Itauçu Complex in central Brazil, samples with UHT assemblages that recorded metamorphic
temperature of ~1000 °C, contain metamorphic zircon that record Ti-in-zircon temperatures
276 Rubatto
mainly in the range 800–950 °C (Baldwin et al. 2007). Obtained T and textural relationships
between zircon and other minerals were used to conclude that the zircon formed mainly during
prograde and retrograde reactions and not at the UHT peak. The lower crustal section of the
Ivrea Zone, Italy, contains metapelitic septa within gabbros that recorded metamorphic T of 900–
1050 °C according to Zr-in-rutile thermometry (Ewing et al. 2013). In these rocks Ti-in-zircon
thermometry records lower temperatures around 750–800 °C, likely because zircon crystallized
only upon cooling. Ti-in-zircon temperatures below the peak T (700–850 °C versus ~900 °C)
are also reported for metamorphic zircon in leucosomes of the Bohemian Massif (Kotkova and
Harley 2010). Therefore, in the case of zircon crystallization from a partial melt, the T of Zr
saturation and thus zircon crystallization can be significantly lower than the peak T experienced
by the rocks, but also well below the T of melting (see details in thermodynamic modeling by
Kelsey et al. 2008; Yakymchuk and Brown 2014; Kohn et al. 2015).
As any successful petrological tool, Ti-in-zircon thermometry has its limitations. Hofmann
et al. (2009) pointed out possible problems of this thermometer due to non-equilibrium effects in
Ti incorporation in zircon, the effect of other substituting elements, and the contamination of Ti
analyses from edges and fractures (see also Harrison and Schmitt 2007; Hiess et al. 2008). Others
have highlighted the issue of underestimation of crystallization temperatures in magmatic zircon
(Fu et al. 2008). In applying Ti-in-zircon thermometry to metamorphic rocks there are other
potential limitations to be considered. (1) Equilibrium with a Ti-phase is a prerequisite for the
correct application of the thermometer (Ferry and Watson 2007). In metamorphic rocks where
more than one metamorphic assemblage is preserved establishing the presence of a Ti-phase at
the stage when zircon formed may not be trivial. Most commonly this leads to an underestimation
of the real Ti-activity and thus the temperature of zircon crystallization. (2) The effect of pressure
on Ti incorporation in zircon remains unconstrained. The Ti-in-zircon thermometer (Ferry and
Watson 2007) was calibrated for pressures close to 10 kbar (Ferry and Watson 2007). Tailby
et al. (2011) showed that Ti substitutes for Si at these pressures and proposed that at high
pressure the solubility of Ti should decrease. As a result, the Ti-in-zircon thermometer may
underestimate temperature. On the other hand, it is expected that under low pressure conditions
(< 5 kbar), the thermometer likely overestimates temperatures. Ferriss et al. (2008) suggested that
with increasing pressure, Ti might additionally substitute into the Zr site, in turn increasing Ti
solubility in zircon. Testing on natural samples is limited, but Ti-in-zircon temperature estimates
for at least the UHP Kokchetav rocks are close to the peak T estimated by other thermometers
(Stepanov et al. 2016b), indicating that, with pressure, decreasing amounts of Ti on the Si site
might be compensated by increasing Ti concentration on the Zr site. Until a better understanding
of the effect of pressure on this thermometer is gained, it is safer to use Ti-in-zircon temperatures
in (U)HP rocks as a relative T indicator, as suggested by Stepanov et al. (2016b). (3) Deformation
under high temperature can modify zircon chemical composition, generally inducing loss of Pb,
REE and Ti or redistribution of elements (Reddy et al. 2009); application of the thermometer to
highly deformed rocks has to be done with caution or supported by EBSD analysis and chemical
mapping of the zircon crystals (see section on Deformation).
ISOTOPE SYSTEMATICS
Since the very early days of geochronology, zircon has been a prime target for U–Pb
isotopic investigations in order to obtain crystallization ages. In the last decades Lu–Hf and
oxygen isotope investigations in zircon domain have been developed and primarily applied to
magmatic or detrital crystals. In metamorphic settings, the use of isotopic tracers to understand
zircon petrogenesis and assist in U–Pb age interpretation is less widespread, but increasing.
Correlations between different isotopic system (U–Pb, Lu–Hf and oxygen) and chemical
signatures acquired from the same growth domain remain underexplored, but have already
demonstrated some valid concepts that are summarized below. For any isotopic systematic it
is important to consider diffusivity (see also Kohn and Penniston-Dorland 2017), robustness to
metamorphic resetting/alteration and thus possible decoupling of different systems.
U–Pb isotopes
Zircon is the most widely used mineral for U–Pb age determination also in metamorphic
rocks. U–Pb geochronology (Th is not sufficiently abundant in zircon to be a useful
chronometer) in metamorphic zircon has increased dramatically since the development of
micro-beam techniques that can measure U–Pb isotopic ratios at the 10–50 µm scale, namely
high resolution ion microprobes and Laser Ablation ICP-MS (Kylander-Clark 2017; Schmitt
and Vazquez 2017). Whereas these methods may not reach the sub-1% age precision of thermal
ionization mass spectrometry (TIMS, Schoene and Baxter 2017) they can spatially resolve the
internal growth zones typical of metamorphic zircon.
Zircon alteration and replacement under sub-solidus conditions are difficult to date
accurately. In altered zircon domains the U–Pb system is disturbed and measured dates are
commonly a mix between the age of the inherited grain and that of metamorphic disturbance.
Several examples of these systematics exist in eclogite of the Dabie-Sulu orogenic belt,
where altered zircon cores yield a range of spurious dates (Zheng et al. 2005; Xia et al. 2009;
Chen et al. 2011; Sheng et al. 2012). Textures also suggests that alteration affects zircon at
a micron scale below the spatial resolution of micro-beam geochronology (10–50 µm) and
thus mixing of domains is a common problem in analyzing altered zircon. Alteration that
produces spurious dates is not limited to sub-solidus conditions, but can also occur in higher
grade samples (Tichomirowa et al. 2005). Metamictization of zircon domains commonly
results in dark, mottled CL emission (Table 1) and zircons with these features are preserved
in all metamorphic environment from diagenesis (Hay and Dempster 2009a) to high grade
(Tichomirowa et al. 2005). Reversely discordant dates are commonly measured in metamict,
altered domains where there has been Pb mobility (Tichomirowa et al. 2005; Kusiak et al.
2013a). Additional indications of zircon alteration are (i) high initial Pb contents and (ii)
homogeneous Lu–Hf and oxygen isotopic composition in domains whose ages scatter (see
sections on Lu–Hf and oxygen isotopes). Complete metamorphic replacement of zircon
domains does achieve resetting of the U–Pb system: such domains (commonly zircon rims)
have poor zoning, homogeneous chemistry and yield accurate and reproducible ages. In sub-
solidus rocks, metamorphic zircon commonly has low U content and relatively high initial Pb
content, which can limit age precision. Eclogite-facies rocks that were subducted along a cold-
geotherm yield this type of zircon (Rubatto et al. 1999, 2011; Baldwin et al. 2004; Spandler
et al. 2005). Replacement/recrystallization may proceed along a well defined front (Hoskin
and Black 2000), but it can also be more localized and randomly distributed to generate
sub-domains where CL-zoning is chaotic (Rubatto et al. 2008). Where pristine, altered and
replaced domains are combined at a micron scale, age determination results in mixed ages.
Most other chemical signals are analyses at the same scale as ages and thus will also be mixed.
Metamorphic zircon forms more readily in high-grade rocks where it commonly
constitutes overgrowths on inherited magmatic or detrital cores. Chronology of migmatites and
granulites that crystallized zircon from melts is relatively straightforward: metamorphic zircon
domains are texturally and chemically distinct from the inherited cores and record faithfully
the age of crystallization, at least below UHT conditions. Common Pb is generally not an
issue in such zircon as Pb strongly partitions in the melt during crystallization. Reproducible
and concordant dates are commonly achieved for zircon overgrowths and their chemistry can
assist in age interpretation. However, even under conditions where Pb diffusion/mobility,
278 Rubatto
strong deformation or later fluid alteration do not disturb the zircon formation age, it is
possible to obtain a spread in zircon ages. This is not necessarily poor geochronology, as this
spread can have a geological significance indicating a long-lasting process. Indeed, in several
terranes it has been demonstrated that melting and high-grade metamorphism can persist
for 10s of million years and thus zircon can form over a period of time. However, during
protracted metamorphism the environment of zircon growth changes because of metamorphic
reactions or new melt injections and thus zircon grows metamorphic domains that are distinct
in chemistry and internal texture (Vavra et al. 1996; Hacker et al. 1998; Schaltegger et al. 1999;
Rubatto et al. 2001, 2009, 2013; Möller et al. 2002; Hermann and Rubatto 2003; Montero et al.
2004; Root et al. 2004; Tichomirowa et al. 2005; Hermann et al. 2006a; Gerdes and Zeh 2009;
Kotkova and Harley 2010; Gordon et al. 2012, 2013; Korhonen et al. 2013; Kylander-Clark et
al. 2013; Harley and Nandakumar 2014; Young and Kylander-Clark 2015). In such samples,
the presence of distinct metamorphic domains that are internally homogeneous, suggests that
zircon grows in discrete, relatively brief episodes.
Extreme conditions with T > 1000 °C and P up to 50 kbar are documented in metamorphic
rocks and such conditions might persist for long geological times (10s Ma). In order
to understand ages obtained from zircon in UHT rocks it is thus important to discuss the
robustness of the U–Pb system in zircon under these conditions. Experimental studies indicate
that Pb diffusion in zircon is not significant at T < 900 °C even for geological times of several
million years (see a review in Cherniak 2010). Metamorphism, however, can reach extreme
temperatures up to 1050 °C, where even a perfectly crystalline zircon could diffuse Pb over
geological time scales (Cherniak and Watson 2003). Inherited zircon (detrital or magmatic
cores/grains) that still retain pre-metamorphic ages are preserved in various terranes that
underwent extreme T for long periods of time. Examples are the Rogaland sapphirine-
granulites (peak and decompression at 950–1000 °C between 7.5 and 5.5 kbar, Möller et
al. 2003; Drüppel et al. 2013), and the lower crustal rocks of the Ivrea Zone (900–1000 °C,
6–10 kbar, Vavra et al. 1996; Ewing et al. 2013). Rare inherited cores with pre-metamorphic
ages are also preserved in crustal rocks from the Kokchetav metamorphic belt in Kazakhstan,
where peak T was around 1000 °C at elevated P of 40–50 kbar. The existence of inherited
zircons in some UHP–T samples imply that zircon was a stable phase even under extreme
conditions and that U–Pb ages can at least partially survive such high T, even when associated
with deformation and melting. In the Kokchetav belt, the duration of UHP–T metamorphism
was only a few million years (Claoué-Long et al. 1991; Hermann et al. 2001; Katayama et al.
2001; Stepanov et al. 2016b). Zircons in some Kokchetav samples have dates scattering over 20
Ma that do not correlate with the zircon internal zoning texture or composition (Stepanov et al.
2016b). This observation was interpreted as partially reset during peak temperature of 1000 °C
in zircon that do not show any evident sign of metamictization. What it is hard to establish is if
any age disturbance under these extreme metamorphic conditions is due to diffusional Pb loss
or dissolution–precipitation of zircon in several stages during metamorphism.
A particular case of Pb mobility is documented in the Archean zircons of the Napier
Complex, Antarctica, which underwent UHT metamorphism at > 900 °C. The discovery of
reversely discordant ages (206Pb/238U age older than the 207Pb/206Pb age) led to postulate
the presence of unsupported radiogenic Pb in some zircon domains (Williams et al. 1984).
More recent studies imaged in detail the distribution of Pb isotopes in these zircons and
proved the patchy distribution of radiogenic Pb at the micron scale. Pb forms nuggets of Pb
metal and its distribution is mostly unrelated to U content (Kusiak et al. 2013a; Kusiak et al.
2013b; Kusiak et al. 2015) and has no correlation with Th/U, REE or oxygen isotopes either.
The mobility of radiogenic Pb caused apparent 207Pb/206Pb dates within a modified domain
to vary over 500 Ma, with the oldest dates being spurious and older than the formation of
the crystal. Kusiak et al. (2013a) proposed that extreme metamorphic temperatures in a
dry environment caused Pb mobility within the crystal, without net Pb loss. This process
of within-crystal Pb mobility has been also identified in the UHT rocks of southern India
(Whitehouse et al. 2014). Such reports will likely increase with the development of atom
probe analysis and ion microprobe mapping. Deformation under relatively high temperature
has also been found to cause Pb mobility (see section on deformation).
Pb diffusion profiles in pristine zircon (not metamict, nor deformed) have so far never
been measured, but increased spatial resolution in modern analytical techniques (atom probe,
NanoSIMS analysis and SIMS depth profiling) may eventually resolve diffusion at a submicron
scale. The bulk of available evidence indicates that loss of Pb by volume diffusion in non-
metamict zircon is not an important factor even under extreme metamorphic conditions.
Lu–Hf isotopes
Hafnium isotopes in zircon have become a widely used petrogenetic tool in magmatic
and sedimentary rocks for crustal evolution studies. For this isotopic system, the role of
metamorphism is primarily a negative one, as metamorphism can cause Pb loss and thus
compromise the veracity of calculated epsilon Hf values and model ages (see a review in
Vervoort and Kemp 2016). The application of Hf isotopes to metamorphic zircon has been
directed to understand zircon petrogenesis and resetting of the U–Pb system. Decoupling
of the U–Pb and Lu–Hf systems in zircon during metamorphism and alteration has been
documented in natural samples (Zheng et al. 2005; Gerdes and Zeh 2009; Xia et al. 2009;
Zhao et al. 2015) and by experimental work (Lenting et al. 2010). While variable U–Pb dates
can be the product of alteration/resetting and be unrelated to geological events (see above),
distinct 176Hf/177Hf signature are expected if zircon domains formed during different episodes
of the rock evolution (Gerdes and Zeh 2009). The plot in Figure 7A illustrates the different
evolution in 176Hf/177Hf over time of components (bulk rock, magmatic zircon, metamorphic
zircon, garnet) of a metagabbro over a 160 Ma period. Because of the high Hf contents, there
is insignificant ingrowth of radiogenic Hf into the magmatic zircon. For other minerals and the
bulk rock, radiogenic ingrowth depends on the Lu/Hf ratio. As most of the Hf is locked away
in zircon, the “reactive” bulk rock Hf isotopes will evolve much faster than the bulk Lu/Hf.
Let’s consider a prograde metamorphic event producing garnet 100 Ma after the protolith
crystallization, when magmatic zircon was not reactive. Given the high Lu/Hf of garnet,
even within a few million years of a metamorphic cycle, garnet can evolve highly radiogenic
176
Hf/177Hf. Metamorphic zircon that forms at peak metamorphic conditions a few million
year after prograde garnet can potentially acquire any 176Hf/177Hf in between the value of the
protolith zircon and the highly radiogenic value of the garnet depending on which reservoir
(protolith zircon, bulk rock, bulk without zircon, or garnet) the new zircon equilibrated with.
Zircon that forms after dissolution of some protolith zircon and some garnet will acquire an
intermediate 176Hf/177Hf, higher than the protolith zircon. The same effect can be produced by
dissolution of other high Lu/Hf phases such as apatite (Valley et al. 2010).
In metamorphic gneisses and mafic eclogites of the Dabie-Sulu HP terrane, inherited
magmatic cores, whose ages were partly reset during metamorphism, still preserve high
176
Lu/177Hf and Th/U relative to metamorphic zircon (Zheng et al. 2005; Xia et al. 2009; Gao
et al. 2015). Even when metamorphic disturbance of the U–Pb age and Th/U system is nearly
complete (spongy textures, modified CL, apparent ages close to lower intercept and low Th/U)
the Lu/Hf system of the protolith zircon remains undisturbed (Xia et al. 2009). On the contrary,
zircon grown during metamorphism has not only new U–Pb age and Th–U compositions but also
consistently lower 176 Lu/177Hf (2–10 times lower) than the protolith zircon value (Fig. 7B). Some
of the zircon domains that crystallized during metamorphism also have higher 176 Lu/177Hf, which
points to release of radiogenic Hf from other sources, most likely garnet. The low 176 Lu/177Hf and
high 176Hf/177Hf in some metamorphic zircon implies that, at the time of zircon growth, garnet
was present as a Lu sink (see section on REE), but recrystallized releasing radiogenic 176Hf/177Hf
280 Rubatto
0.2838 0.2828
/Hf 4
A B
0.2836
garnet, Lu
0.2824 magmatic
Hf/177Hf
0.2834
Hf/177Hf
on metamorphic
zirc 0.2820
out zircon
with
176
0.2832 bulk
176
f 0.026
bulk rock, Lu/H
0.2816
0.2830 protolith zircon, Lu/Hf 0.004
magmatic
0.2828 0.2812
0 20 40 60 80 100 120 140 160 0.0000 0.0006 0.0012 0.0018
Time since protolith crystallization (Ma) 176
Lu/177Hf
Figure 7. Lu–Hf systematics in metamorphic zircon. A) 176Hf/177Hf isotopic evolution over time for differ-
ent components of a gabbroic rock that crystallized magmatic zircon at time 0 and metamorphic zircon 100
Ma later, in a garnet bearing assemblage. Because zircon is the main host of Hf in the rock, the evolution of
the bulk rock where protolith zircon remains isotopically isolated (bulk without zircon) is significantly more
radiogenic than that of the protolith zircon. Garnet that forms during prograde metamorphism has a much
higher Lu/Hf than the bulk (4 versus 0.026) or the magmatic zircon (Lu/Hf 0.004) and rapidly increases its
176
Hf/177Hf with time. The 176Hf/177Hf ratio of metamorphic zircon that forms a few Ma after garnet (peak
to retrograde path) will depend on which component(s) of the system the zircon equilibrates with: only the
magmatic zircon, the bulk rock fully equilibrated, the bulk rock without participation of the magmatic zircon,
or even only the garnet. The grey arrows indicate the trajectory of new metamorphic zircon depending on its
acquired 176Hf/177Hf, and their variable length represents the likelihood of that composition occurring. Model
based on data from Monviso eclogite described in Spandler et al. (2011). B) Difference in 176Hf/177Hf and
176
Lu/177Hf systematics between magmatic protolith zircons (large fields) and new metamorphic zircon (small
circles) in two HP samples from the Dabie orogeny. Data are from Zheng et al. (2005), white symbols, and
Gao et al. (2015), grey symbols. Note the tendency of the metamorphic zircon to higher 176Hf/177Hf and lower
176
Lu/177Hf due to the likely effect of garnet in the metamorphic assemblage.
(Zheng et al. 2005; Xia et al. 2009). Even more complicated 176 Lu/177Hf versus 176Hf/177Hf trends
can be created in metamorphic zircon if multiple Hf sources are identified, such as dissolution
in anatectic melts of much older detrital zircon of various age and multiple metamorphic events
overprinting each other (Gerdes and Zeh 2009; Zhao et al. 2015). In the process of U–Pb and
Lu–Hf decoupling, there is a fundamental role for aqueous fluids and melts in redistributing and
transporting these elements. The decoupling of the U–Pb and Lu–Hf systems can also be used
to identify analyses mixing inherited cores and metamorphic rims, which produces a correlation
between the two systems (Xia et al. 2009; Zhao et al. 2015).
Oxygen isotopes
Oxygen isotopes can be readily measured in single zircon growth zones by ion microprobe
to a precision of 0.2‰ (Valley 2003). A common application of this isotopic system is tracing the
source of zircon grains, particularly in magmas and sediments, to reconstruct crustal reworking.
This strategy applies equally to inherited and detrital zircons in metamorphic rocks (Rumble
et al. 2002; Zhao et al. 2008; Rubatto and Angiboust 2015), and is particularly powerful in
recrystallized rocks where relic zircon cores may be the only remnant of the protolith (Zheng
et al. 2008; Fu et al. 2010; Chen et al. 2011; Sheng et al. 2012; Gauthiez-Putallaz et al. 2016).
Similarly to the Lu–Hf system, oxygen isotopes can give insight into metamorphic
replacement and modification of zircons, as well as partial resetting of the U–Pb system and
mixed ages. Natural samples indicate that, at least in some environments, the U–Pb system is
easier to reset than the oxygen isotopes and thus partly reset zircon domains may still preserve
protolith oxygen isotopic composition (Wu et al. 2006; Martin et al. 2008). This may seem
at odds with experimental data that report diffusion of oxygen isotopes faster than that of
Pb (Cherniak and Watson 2003), but age resetting during metamorphism is generally not
dominated by simple solid state diffusion. In other cases, metamorphic resetting of zircon age
and Th/U weakly correlates with changes in δ18O suggesting that metamorphism eventually
affects oxygen isotopes (Petersson et al. 2015; Rubatto and Angiboust 2015). This difference
may reflect different processes and degree of resetting. Therefore the use of oxygen isotopes to
detect metamorphic disturbance of the age may have to be considered case by case.
While replacement/recrystallization can be an effective way to reset the oxygen isotopic
composition of zircon, natural and experimental studies indicate that diffusion of oxygen does
not play a significant role (Watson and Cherniak 1997; Page et al. 2007). Oxygen isotope
diffusion in zircon is still debated (see also Kohn and Penniston-Dorland 2017) because of
disagreement whether “dry” or “wet” diffusion applies in natural rocks and the general lack
of data (i.e., measurable profiles in natural samples). The point has been made that even in
apparently “dry” metamorphic rocks where free fluids may be lacking, the activity of H2O
is still buffered by mineral phases and significant, and thus “wet” diffusion generally applies
(Kohn 1999). The potential retentivity of 18O/16O isotopic ratio in zircon has been modeled
by Cherniak and Watson (2003) using the only experimentally determined wet diffusion data.
Compared to Pb, U or other trace elements, oxygen diffusion in zircon is much faster so that
oxygen isotope signature of a ~100 µm domain should survive less than 0.1 Ma at temperature
of 700 °C, and only 5700 years at 900 °C. Natural cases, however, suggest higher retentivity of
oxygen isotopes in zircon. Inherited, relict zircon cores with very low δ18O that survived UHP
metamorphism at ~750 °C are common in the Dabie-Sulu gneiss (Chen et al. 2011; Sheng et
al. 2012). Bowman et al. (2011) have shown that even zircon that underwent relatively high-
grade metamorphism for several 10s of million years preserve sharp δ18O changes (within the
5–10 µm resolution of the analyses) between inherited cores (~6 ‰) and metamorphic rims (8–
10‰). Claesson et al. (2016) suggest that the oxygen isotopic composition of non-metamict
zircons that still preserve magmatic-type zoning were modified under high temperature
(~700 °C) in hydrothermal conditions. They base their conclusion on the finding of unusually
high δ18O in Archean zircons and on previous experimental determination of relatively fast
oxygen diffusion under wet conditions (Cherniak and Watson 2003). However, also in this
case, no oxygen isotope diffusion profile could be measured in zircon. These examples show
that oxygen isotope diffusion in zircon is still far from being resolved and requires future work.
Oxygen isotopic composition of zircon can assist in the challenging task to link zircon
ages to metamorphic assemblages. Oxygen fractionation factors between zircon and some major
minerals are reasonably well known (Valley 2003). For example, given known metamorphic
temperatures, oxygen isotopic equilibrium between zircon and garnet can support a case for
metamorphic zircon formation (Martin et al. 2006). Conversely, isotopic disequilibrium
between zircon and major phases would suggest that inherited zircons are preserved within
metamorphic assemblages, where major minerals are isotopically equilibrated (Zheng et al.
2003). Fu et al. (2010) demonstrated that the mantle-like δ18O (4.8–5.2 ‰) of zircon and Ti-
in-zircon thermometry (600–800 °C) in numerous jadeitites are not supportive of hydrothermal
zircon formation in a vein, but rather of their oceanic crust protolith. The systematics of oxygen
isotopes is more complicated during sub-solidus alteration and partial to complete replacement
of zircon. Does a zircon formed by in situ dissolution–precipitation equilibrate with the bulk rock
oxygen isotopic signature or inherit the δ18O of the protolith zircon it replaces? Two examples
of metamorphic zircon that did not re-equilibrate with the bulk oxygen signature are presented
in Figure 8. In a metabasite from Naxos that experienced metamorphism at 500 °C, a number
of metamorphic zircon rims yield variable δ18O that in each grain is identical to the values of
the magmatic core (Martin et al. 2006), suggesting that the rims inherited the core isotopic
signature. In an eclogitic metasediment from the Sesia Zone in the Italian Alps (Rubatto et al.
1999), metamorphic zircon rims yield different δ18O than the detrital cores, but still differing one
from another. In this case it is postulated that the isotopic composition of the metamorphic zircon
equilibrated only locally and not rock-wide. On the other hand, zircon overgrowths that formed
in a metamorphic vein, hosted by the eclogitic metasediment, displays a homogenous oxygen
isotope composition, as expected for newly formed zircon that equilibrated with the bulk rock.
282 Rubatto
Particularly insightful is the use of oxygen isotopes in zircon and co-existing minerals to
trace metasomatism and open versus closed system metamorphism. In orthogneisses, significant
shifts in δ18O between relict magmatic cores and metamorphic rims have been interpreted as
open system behavior and crystallization of zircons from metamorphic fluids (Chen et al. 2011;
Sheng et al. 2012). Zircon within metamorphic veins in the Monviso eclogite, have significantly
lower δ18O than inherited magmatic zircon in the country rock. The zircon oxygen composition
and age, together with the variable δ18O in garnet, reveal deep sea floor alteration as well as
metasomatism during high pressure metamorphism (Rubatto and Angiboust 2015). As this
example shows, a more convincing argument for metasomatism can be made when the isotopic
variation in metamorphic zircon is correlated to the composition of major minerals, such as
garnet (Martin et al. 2006; Page et al. 2014; Rubatto and Angiboust 2015).
Changes in the oxygen isotopic composition of minerals can change according to
temperature, assemblages and external fluids. Therefore the interpretation of oxygen isotope
signatures in metamorphic systems in terms of external fluids requires a control on the change in
assemblages and temperatures between the protolith and metamorphic stages. Large variations
in mineral crystallization and volume of quartz and plagioclase can, for example, shift the
δ18O of metamorphic minerals of a few ‰ within a constant bulk δ18O. This was modeled in
metamorphic rocks from Alpine Corsica (France) where Permian granulites (T 650–800 °C)
composed mainly of garnet, quartz and feldspars were transformed by Eocene high-pressure
metamorphism (T 400–500 °C) into garnet, amphibole, lawsonite, quartz and phengite
(Martin et al. 2014). The shift in δ18O between the granulitic garnet core (9.9‰) and the high-
pressure garnet domains (7.2‰) is significant, but it can be largely reconciled with changes in
assemblage and T alone. In that sample, a further stage of metamorphism under high pressure
inherited core metamorphic rim

10.0
Metabasite, Metasediment
Naxos, Greece [1] Alps, Italy [2] Vein within metasediment [2]
9.0
8.0
δ18O‰ zircon
7.0
6.0
5.0
4.0
50µm 50µm
3.0
50µm
Figure 8. Oxygen isotopic composition of zircon core and rim from two samples that underwent meta-
morphism at 500–600 °C and where zircon rims record the age of metamorphism. Analyses of rim and
core from the same crystal are linked by thin vertical lines. In both samples, the zircon rim oxygen isotope
composition did not equilibrate with the bulk rock. In contrast, zircon in the metamorphic vein within the
micaschist (right) have a uniform oxygen isotope composition. Example of the zircon texture is shown in
the CL image below the data. Errors on δ18O values are not plotted. Data and images are from [1] Martin
et al. (2006) and [2] unpublished data from sample MST2 and MST1 described in Rubatto et al. (1999)
[Images are used by permission of Elsevier, license 3930790667589, in Martin et al. (2006), Vol. 87, Fig. 7,
p.183; license 3930131166344, from Rubatto et al. (1999), Vol. 167, Fig. 2, p. 146]. See text for discussion.
was related to metasomatism by external fluids. While the Corsica rocks may represent a
rather extreme case (large variation in T and assemblages between metamorphic cycles), the
case highlights that the interpretation of δ18O in zircon or any metamorphic mineral must be
supported by accurate P–T constraints, and that changes in δ18O between growth zones of up
to a couple of ‰ do not necessarily imply external fluids.
Particularly low δ18O have been reported for metamorphic zircons (Fig. 9) that grew in
high-pressure rocks such as mafic eclogites and veins (Fu et al. 2010, 2012; Chen et al. 2011;
Sheng et al. 2012; Rubatto and Angiboust 2015). The δ18O value of these zircons is well below
the value of the relict magmatic zircons present in the same or nearby rocks and below the mantle
value of 5.3 ± 0.3 ‰, 1SD (Valley 2003). In some cases the low zircon values are also found
in coexisting garnet (Rubatto and Angiboust 2015). Two explanations have been proposed for
these low δ18O zircons. In the case of values between 1–5 ‰ in mafic rocks and included veins
the low δ18O has been attributed to ocean floor alteration of the protolith and/or presence of low
δ18O metasomatic fluids generated from nearby altered crust (ultramafic or mafic) during zircon
growth. Indeed, bulk rock analyses have proven that altered oceanic crust, even when subducted
to high pressure conditions, generally preserves the δ18O typical of ocean floor alteration
(Putlitz et al. 2000; Miller et al. 2001). However profiles through oceanic crust in Oman report
minimum δ18O values of 3‰ for serpentinization under relatively high temperature (Gregory
and Taylor 1981); these values are still higher of what is measured in some metamorphic zircon
in eclogites and veins (Fig. 9). Internal fractionation of oxygen isotopes during metamorphism
by crystallization of high δ18O phases could additionally shift the zircon oxygen composition
to lighter values. Alternatively, the protolith of subducted oceanic crust could have experienced
more extreme high T alteration than found in the Oman sequence.
A particularly intriguing case is that of negative δ18O values (as low as -10 ‰) in
metamorphic zircons of the Dabie–Sulu eclogites, veins and gneisses, where inherited zircon
are -4 to 3.5‰ in the gneiss and 2–10‰ in the eclogite (Chen et al. 2011; Sheng et al. 2012).
These negative values have been interpreted as metamorphic zircon growth from externally
derived negative δ18O fluids produced by dehydration of gneissic protoliths that were glacial-
hydrothermally altered, in a particularly cold clima (the so called “Snowball Earth”). Such
extremely light oxygen signature has been extensively documented in the inherited magmatic
zircon over a wide section of the South China Craton (Zheng et al. 2008).
metamorphic to -10‰
Sulu gneiss (5)
to +10‰
Sulu eclogite (5)
Dabie vein (4)

Dabie eclogite (4)
Greece eclogite (3)
Japan jadeitite (2)

metamorphic
magmatic Alps vein (1)
Alps eclogite (1)
-4 -3 -2 -1 0 1 2 3 4 5 6 7
δ18O‰ zircon
Figure 9. Compilation of low δ18O values in metamorphic zircon (black bars) compared to protolith zircon
of the same sample (empty bars). Data are from (1) Rubatto and Angiboust (2015), (2) Fu et al. (2010), (3)
Fu et al. (2012), (4) Sheng et al. (2012) and (5) Chen et al. (2011). See text for discussion.
284 Rubatto
PETROGENESIS
Compared to the wealth of petrological knowledge on major metamorphic minerals,
understanding the behavior of accessory phases is still in its infancy. Relatively few studies
have dealt with the stability of accessory minerals. Zircon is a robust accessory mineral and
case studies show that its stability is wide and its reactivity can vary case by case. Only a few
reactions that form new metamorphic zircon from major or accessory minerals have been
proposed. Processes that produce metamorphic zircon from precursor zircon by replacement/
recrystallization at the solid state, dissolution (re)precipitation in situ aided by fluids, or
melting and crystallization in leucosomes have been recognized as significant.
In this section the complex topic of the petrogenesis of zircon in metamorphic rocks
is discussed for four main processes/environments roughly corresponding to increasing
metamorphic grade: (i) diagenesis and low-T metamorphism, (ii) zircon replacement or
dissolution–precipitation in sub-solidus conditions and in the presence of aqueous fluids, (iii)
zircon in melt-dominated metamorphic systems and (iv) zircon-forming reactions involving
major/minor minerals under amphibolite to granulite facies conditions. An overview of typical
zircon types under variable metamorphic conditions is shown in Figure 1.
Diagenesis and low-T metamorphism
What may happen to zircon in the early stages of metamorphism is best shown by the
study of lightly metamorphosed sandstones (Hay and Dempster 2009a). During burial at
< 100 °C, detrital zircons in these sediments show signs of alteration and new metamorphic
zircon is formed. Delicate textures of zircon showing alteration have been documented under
such conditions (Fig. 1A): they are unlikely to have survived the depositional process, and thus
represent diagenetic/metamorphic modifications. The modification is driven by fluids and affects
particularly radiation-damaged zones (Fig. 1B), as predicted by experimental studies (Geisler et
al. 2003a). The altered zones are enriched in elements such as Ca, Al, Fe, Mn and LREE that do
not enter the zircon structure (Geisler et al. 2003b; Hay and Dempster 2009a), and show porosity
and fractures. These zones are similar to what observed in zircons in hydrothermal altered
granites (Geisler et al. 2003b). The altered zones are progressively replaced by new crystalline
zircon that also forms jagged, sawtooth-shaped zircon overgrowths of a few microns in thickness
(Hay and Dempster 2009a), which become relatively more abundant in lower greenschist facies
samples (200–400 °C) (Rasmussen 2005; Hay and Dempster 2009b). The formation of low-
grade metamorphic xenotime, even finely intergrown with zircon, is commonly associated to
these first metamorphic zircon overgrowths (Rasmussen 2005; Hay and Dempster 2009a; Hay
et al. 2010). The mechanisms that have been proposed to be dominant at this low grade are
dissolution–precipitation and solid-state reaction (Hay and Dempster 2009a).
The sawtooth-shaped zircon overgrowths are generally absent (see an exception in Franz et
al. 2015) from higher grade rocks (upper greenschist to lower amphibolite facies, Rubatto et al.
2001; Hay and Dempster 2009b) and thus it is likely that later dissolution or Ostwald ripening
is erasing this early record. Zr released by dissolution of small zircon overgrowths during
prograde metamorphism could be accommodated in other growing metamorphic minerals
(Kohn et al. 2015) explaining why metamorphic zircon is generally absent in greenschist to
lower amphibolite facies rocks. Partly altered, porous and metamict zircon domains, which
are the likely source of Zr for low-grade zircon growth, are partly dissolved during early
metamorphism, but can also be preserved at higher grade (see below).
Metamorphic zircon and fluids at sub-solidus conditions
Zircon replacement and modification by fluids has been identified in a variety of metamorphic
settings from diagenetic environments (see above) to granulite facies metamorphism, and
particularly in rocks that record high pressure metamorphism (see also a review in Rubatto and
Hermann 2007b). A variety of textures have been attributed to fluid-driven alteration/replacement

of zircon (Fig. 1, Table 1, see also Corfu et al. 2003). One of the lowest T reported for zircon
alteration in metamorphic rocks is from Spandler et al. (2004), who proposed that relict zircon
cores preserved in schist from New Caledonia underwent alteration by fluids during sea floor
alteration of the mafic protolith at T < 100 °C, based on their inclusion assemblage and texture
(Fig. 1C). Such process expelled trace elements from the original zircon, increased porosity, and
produced mottled and patchy zoning. These features have many similarities to the dissolution–
precipitation process described in zircon from rocks at 400–600 °C (Tomaschek et al. 2003;
Rubatto et al. 2008). Similar alteration and replacement processes have been described for other
accessory minerals such as apatite and monazite and reproduced in controlled experiments
(Hetherington et al. 2010; Harlov et al. 2011; Seydoux-Guillaume et al. 2012). For a more
general description of mineral replacement see also Putnis (2009).
More extensive alteration leads to full replacement/recrystallization of domains that are
texturally discordant with the original zoning and often marked by a recrystallization front
(Hoskin and Black 2000; Vonlanthen et al. 2012). These replaced domains are chemically
and microstructurally different from the cores: they commonly have a lower trace element
content, they lack any elements typical for alteration zones such as LREE, Ca, Al and
Fe, and they are inclusion poor. TEM investigation of these domains shows that they are
relatively free of defects (Vonlanthen et al. 2012). Most studies of natural samples report
that during this process the chemical composition and isotopic systematics of the replaced
zircon is totally reset. However, remnants of the chemical and isotopic composition of
protolith zircon have been reported in such domains when analyses were performed with a
spatial resolution of ~20 mm (Hoskin and Black 2000). This indicates that the replacement
process is not always complete and that “islands” of altered zircon are preserved at the
micro to nanoscale. Such micro-relicts have been identified in zircon where the replacement/
recrystallization process led to micro-zircon formation (Rubatto et al. 2008).
It has been proposed that dissolution–precipitation can proceed efficiently even with
minimum free fluid, and with low Zr solubility in that fluid (Tomaschek et al. 2003; Geisler
et al. 2007). The process does not require transport of Zr and other zircon-forming elements
outside the zircon itself, and can proceed in a virtually closed system (Hoskin and Black
2000). This implies that this process, that cannibalizes inherited zircon and resets its isotopic
and chemical composition to form metamorphic zircon, is not necessarily communicating
with the reactive bulk rock and can occur independently of Zr solubility or metamorphic
reactions. It requires aggressive fluids and it is enhanced by temperature and the presence of
metamict zircon (Geisler et al. 2003a).
Zircon with textures suggestive of a metamict state is preserved up to high metamorphic
conditions. For example, porous zircon domains that contain very low-grade (<100 °C) mineral
inclusions have been recovered in blueschist- to eclogite-facies rocks (Spandler et al. 2004). A
whiteschists that recorded ultra high pressure metamorphism (35 kbar and 750 °C) preserves
altered zircon cores from the precursor granite that contain prograde to peak metamorphic
mineral inclusions in healed alteration zones (Gauthiez-Putallaz et al. 2016). These reports
suggest that dissolution of metamict zircon domains by metamorphic fluids that provides Zr and
other essential elements for metamorphic zircon formation can occur under diverse conditions.
Preservation of metamict zircon up to high grade conditions is supported by experimental
work and calculations. Zircon is predicted to stop accumulating radiation damage at ~250 °C,
the T above which amorphisation will no longer occur, because annealing becomes faster
than damage accumulation (Meldrum et al. 1999; Ewing et al. 2003). The process of recovery
and structural re-organisation of zircon, as described in Ewing et al. (2003, and references
therein) occurs in stages over a T range and is particularly low in zircon compared to any
286 Rubatto
related phase. In experiments, the first stage of recovery occurs below ~700 °C (recovery of
point defects and short length damage). The second stage of re-organisation occurs above
~700 °C and, full structural re-organisation has been documented at 900 °C or above (Geisler
et al. 2001b). During this process, islands of recovered material are surrounded by amorphous
domains until damaged material is no longer detectable (by Raman or TEM investigation).
Diffusion dominates the recovery process which is thus T- and time-dependent (Ewing et al.
2003). Calculations indicate that it would take up to 370 Ma to fully recover the structure of
metamict zircon at 700 °C (Geisler et al. 2001a), but the process is significantly more efficient
under hydrothermal or wet geological conditions (Geisler et al. 2001b).
Experimental investigations have shown very low solubility of Zr in aqueous fluids, that
increases with increasing Si contents and alkalinity of the fluids (Ayers et al. 2012). Both,
Si contents and alkalinity in aqueous fluids increase significantly with increasing pressure
(Hermann and Rubatto 2014). This might explain why abundant metamorphic zircon occurs in
high-pressure crustal rocks that record metamorphic temperatures below the solidus < 700 °C,
as for example in Alpine eclogites. Examples are large metamorphic zircon in the omphacite–
garnet–rutile veins within the Monviso eclogites (Rubatto and Hermann 2003) and the intense
zircon dissolution–precipitation process in the jadeite–leucogabbro of the Lanzo Massif (Rubatto
et al. 2008). Metamorphic zircon in jedeitites could also form by precipitation from alkaline-rich
fluids (e.g., Mori et al. 2011). The abundance of metamorphic zircon in the HP rocks of the
Dabie-Sulu region has also been largely attributed to the activity of aggressive fluids (Zheng et
al. 2005; Zhang et al. 2009b; Zhao et al. 2015). In contrast, metamorphic zircon is very rare in
greenschist to amphibolite facies crustal rocks that record Barrovian metamorphism at similar T.
If fluids are the driver of zircon dissolution–precipitation, zircon reactivity would be much
less dependent on T, but rather enhanced by the release of fluids in the rock system. Thus,
metamorphic zircon formation can occur even while T is increasing, unlike what indicated
by thermodynamic models that require equilibrium among all phases (Kohn et al. 2015).
Prograde metamorphic zircon have been for example documented in the Dora Maira schists
(Gauthiez-Putallaz et al. 2016), where zircon domains have been related to episode of fluid
release by dehydration reactions (see texture in Fig. 1F). Such cases of prograde zircon must
be considered when interpreting zircon ages.
Metamorphic zircon and melts
The most common form of metamorphic zircon is overgrowth on detrital or inherited
magmatic grains during partial melting. Field studies that looked at prograde metamorphic
sequences, mainly of pelitic/arcosic compositions, show that new zircon formation under
sub-solidus conditions is virtually absent, but becomes abundant as soon as partial melting
is observed (Rubatto et al. 2001; Williams 2001). Such metamorphic overgrowths on zircon
cores are more abundant in leucosomes than melanosomes (Rubatto et al. 2001).
Zr solubility in anatectic melts ranges from 10s to 100s of ppm according to temperature
and melt composition (Boehnke et al. 2013). A few studies have attempted to model the behavior
of zircon and monazite in migmatites using thermodynamic databases (Kelsey et al. 2008;
Kelsey and Powell 2011; Yakymchuk and Brown 2014; Kohn et al. 2015). These models agree
that in migmatites, most of the bulk Zr will be either stored in the melt or locked in undissolved
inherited zircon. They also predict that zircon crystallization will occur significantly during
cooling when Zr saturation is reached in leucosomes or in interstitial melt. A common conclusion
is that most of the new zircon will crystallize in leucosomes compared to melanosomes or restitic
portions. The solubility models also predict that, in felsic compositions, zircon dissolution and
thus crystallization from a melt is less effective than that of monazite, and thus relict zircon
will survive to higher metamorphic grade than monazite. The models consistently predict that
anatectic zircon should grow during cooling when Zr solubility decreases in the melt or the
solidus is reached (Kelsey et al. 2008; Kelsey and Powell 2011; Kohn et al. 2015).
The thermodynamic models all make significant assumptions that may differ from natural
cases. A common assumption to all models is that all zircon and monazite crystals are in contact
with the melt over the entire P–T evolution (Kelsey et al. 2008; Kelsey and Powell 2011;
Yakymchuk and Brown 2014; Kohn et al. 2015). In natural rocks, shielding of inherited grains
from contact with the melt (or fluids) is often the case. The observation that metamorphic zircon
rims even in migmatites are variable in size from grain to grain, with the common presence of
grains that are lacking overgrowths, support this scenario. Therefore any attempt to model the
Zr budget in metamorphic rocks has the limitation that a variable but significant proportion of
Zr may remain shielded from the melt. A second assumption in some models is that no melt loss
from the system occurs. The pioneering modeling of Kelsey et al. (2008) mainly considered
closed system behavior, with one episode of melt loss. Melt loss is undoubtedly a complication
in natural rocks and successive models demonstrated that such behavior will reduce the amount
of melt generated, and thus the solubility of Zr and the production of new zircon (Yakymchuk
and Brown 2014). Ignoring the effect of Zr-release and uptake from other Zr-bearing phases
in thermodynamic models (Kelsey et al. 2008; Yakymchuk and Brown 2014) does not seems
to affect the general conclusions compared to studies that budgeted for Zr in minerals such as
garnet, rutile and amphibole (Kelsey and Powell 2011; Kohn et al. 2015).
Crystallization from a Zr saturated melt upon cooling, as predicted by the models, is
not the only process to form zircon overgrowth at high-grade metamorphic conditions. Some
studies concluded that zircon in migmatites or granulites can form during prograde or peak
metamorphism (Hermann and Rubatto 2003; Baldwin et al. 2007; Gordon et al. 2013; Rubatto
et al. 2013). Arguments are based on zircon inclusions, trace element composition, Ti-in-zircon-
temperatures and different zircon ages from continuous sequences. In the most compelling
cases, a change in REE composition has been observed between relatively older and younger
metamorphic zircon domains (Hermann and Rubatto 2003; Gordon et al. 2013; Rubatto et
al. 2013). When the change in zircon REE composition has been related to the abundance of
coexisting phases (particularly garnet and feldspars, see discussion above) and in turn to the
P–T path, prograde to peak zircon growth was proposed. Such examples span different tectonic
setting and P–T evolutions from collisional Barrovian metamorphism (Rubatto et al. 2013),
UHP metamorphism (Gordon et al. 2013) and lower crustal melting in extensional settings
(Hermann and Rubatto 2003). In the case of fluid-induced melting in the central Alps, Rubatto
et al. (2009) observed that zircon growth from melt occurred at different times (million of
years apart) in segregated leucosomes sampled only meters from each other. Some migmatites
contain multiple growth zones (Fig. 1) that have distinct age and composition, within single
samples or even zircon grains. The intermittent availability of water for fluid-induced melting
has been proposed by Rubatto et al. (2009) as a mechanism to explain multiple zircon growth
zones in the same crystal and with age differences of several million years. These observation
provide strong evidence that dissolution–precipitation of zircon in the presence of a melt can
occur at any stage—prograde or retrograde—as long as melt is present.
Zircon forming reactions
Textural observations in natural samples that support metamorphic zircon growth from
the breakdown of another phase have been reported for ilmenite (Bingen et al. 2004), rutile
(Ewing et al. 2013; Pape et al. 2016) and garnet (Fraser et al. 1997; Degeling et al. 2001);
see textures in Fig. 1N and Fig. 2C. In some cases the textural observations are supported by
trace element mass balance considerations (Degeling et al. 2001; Ewing et al. 2013). The main
idea is that, in a metamorphic reaction, there is a decrease of Zr solubility in source minerals
(rutile, garnet etc...) and the expelled Zr results in the precipitation of metamorphic zircon. It
has been suggested that garnet plays a minor role in the Zr budget of crustal rocks (Kelsey
and Powell 2011), and thus garnet breakdown reactions might result in the formation of new
zircon. A recent compilation of Zr content in major minerals (Kohn et al. 2015) identified
288 Rubatto
rutile, ilmenite, titanite, garnet and hornblende as carriers for Zr: such minerals contain a
few to 100s of ppm of Zr, and even 1000s of ppm in the case of rutile. A review of own
data on Zr content in garnet (320 analyses, 20 samples) in rocks metamorphosed from 500
to 1000 °C where metamorphic zircon is found, show Zr contents of only 5–20 ppm Zr in
garnet. Zr concentrations in minerals increase with increasing temperature. Thus, the reaction
of magmatic minerals such as pyroxene, amphibole and ilmenite, but also volcanic glass to
lower temperature metamorphic minerals provides a mechanism to form metamorphic zircon.
The storage capacity of Zr in rock forming minerals can be considered in a simple
example: A garnet-amphibolite with 20% garnet, 20% hornblende with contents of 3, 30 or
100 ppm Zr, that also contains 2 % of rutile with 40, 1500 or 3500 ppm Zr (T of 500, 700
and 900 °C, respectively) would provide a maximum of 2, 42 or 110 ppm Zr for the bulk,
respectively. Thus, Zr release at low temperature of 500 °C from these minerals is irrelevant,
whereas is more significant when very high temperature minerals are affected. As modeled
by Kohn et al. (2015) for a basaltic and metapelitic composition, Zr released from major and
accessory minerals during prograde metamorphism is expected to be entirely taken up by
other minerals, at least as long as the T is increasing. Particularly in rutile-bearing rocks, any
Zr release in the reactive bulk can be taken up by growing rutile (Kohn et al. 2015). Therefore
zircon-forming reactions will be mainly related to decompression when rutile transforms to
ilmenite or titanite, or related to retrograde replacement of garnet by minerals that have a lower
capacity to store Zr (cordierite, biotite or chlorite).
CONCLUSIONS AND OUTLOOK

The investigation of metamorphic zircon has dramatically increased since the development
of in situ analytical methods that allow measuring diverse chemical and isotopic signals in
distinct zircon domains. Combined with essential imaging of internal textures, geochemical
information has provided the necessary base for metamorphic petrology of zircon.
Metamorphic zircon forms by a series of processes from the lowest grade to extreme
metamorphism. At low grade, fluid alteration and solid-state replacement are dominant
mechanisms affecting zircon. During partial melting zircon is particularly reactive with
high solubility in the melts and crystallization of overgrowths. Under extreme conditions
metamorphic zircon remains stable, but different elements and isotope systems are likely to
be affected in different ways. The most prominent process is the loss of incompatible Pb that
occurs in altered, deformed or metamict zircon, whereas the REE and HFSE trace element
chemistry and Hf systematics are generally preserved. The behavior of oxygen isotopes in
zircon under extreme conditions remains uncertain.
Linking U–Pb ages to metamorphic conditions for correct age interpretation requires the
combination of multiple information, including internal zoning, deformation features, inclusions,
Ti-thermometry, trace element patterns, Lu–Hf and/or oxygen isotopes. These different systems
may have different retentivity and thus are not always coupled, and their comparison provides
additional information. Zircon is not only a robust geochronometer, but also a mineral relevant
for the petrogenesis of metamorphic rocks that can provide details on protolith, temperature
evolution, deformation, fluids and melts and assists the reconstruction of crustal processes.
Current and future analytical developments will increase our capability to collect
geochemical information with a greater spatial resolution, that in turn will allow resolving
element and isotopic diffusion, fine scale alteration and replacement/recrystallization. The
systematics of numerous trace elements hosted in zircons remains underexplored (e.g., H, Li,
Nb, Ta, Sc) and requires systemic studies. Trace element partitioning with other phases and melts
of different compositions have to be further investigated over a range of P–T to fully exploit the
capacity of zircon to monitor geochemical differentiation. Ti-in-zircon thermometry is lacking
studies on the effect of pressure and Ti activity. Diffusion of crucial elements, for example
oxygen, requires further investigation both experimentally and of natural samples. Petrogenesis
of metamorphic zircon will gain from additional knowledge of Zr and Hf distribution in
metamorphic minerals and fluids, and of zircon forming reactions. The complex processes of
zircon alteration, replacement, dissolution, precipitation and modification in general should be
approached through simulations, experiments and systematic studies in natural samples.
ACKNOWLEDGMENTS
A wide thanks goes to all the students and collaborators that I have worked with while
investigating the many faces and lives of metamorphic zircon. Their constant supply of
interesting samples has much benefited my research.
I thank Matt Kohn, Martin Engi and Pierre Lanari for inviting me to contribute to this
volume and keeping up the information flow and discussion during manuscript preparation.
This contribution has benefited from the constructive reviews of Martin Whitehouse, Bradley
Hacker and Matt Kohn. A special thanks goes to Jörg Hermann for discussion and comments
on this chapter.
REFERENCES
Austrheim H, Corfu F (2009) Formation of planar deformation features (PDFs) in zircon during coseismic faulting
and an evaluation of potential effects on U–Pb systematics. Chem Geol 261:24–30
Ayers JC, Zhang L, Luo Y, Peters TJ (2012) Zircon solubility in alkaline aqueous fluids at upper crustal conditions.
Baldwin SL, Monteleone B, Webb LE, Fitzgerald PG, Grove M, Hill EJ (2004) Pliocene eclogite exhumation at
plate tectonic rates in eastern Papua New Guinea. Nature 431:263–267
Baldwin JA, Brown M, Schmitz MD (2007) First application of titanium-in-zircon thermometry to ultrahigh-
temperature metamorphism. Geology 35:295–298
Bauer C, Rubatto D, Krenn K, Proyer A, Hoinkes G (2007) A zircon study from the Rhodope Metamorphic
Complex, N-Greece: Time record of a multistage evolution. Lithos 99:207–228
Belousova E, Griffin W, O’Reilly SY, Fisher N (2002) Igneous zircon: trace element composition as an indicator of
source rock type. Contrib Mineral Petrol 143:602–622
Bingen B, Austrheim H, Whitehouse M (2001) Ilmenite as a source of zirconium during high-grade metamorphism?
Textural evidence from the Caledonides of Western Norway and implications for zircon geochronology. J
Petrol 42:355–375
Bingen B, Austrheim H, Whitehouse MJ, Davis WJ (2004) Trace element signature and U–Pb geochronology of
eclogite-facies zircon, Bergen Arcs, Caledonides of W Norway. Contrib Mineral Petrol 147:671–683
351:324–334
Bowman JR, Moser DE, Valley JW, Wooden JL, Kita NT, Mazdab FK (2011) Zircon U–Pb isotope, δ18O and trace
element response to 80 m.y. of high temperature metamorphism in the lower crust: Sluggish diffusion and
new records of Archean craton formation. Am J Sci 311:719–772
Buick IS, Hermann J, Williams IS, Gibson R, Rubatto D (2006) A SHRIMP U–Pb and LA-ICP-MS trace element
study of the petrogenesis of garnet–cordierite–orthoamphibole gneisses from the Central Zone of the
Limpopo Belt, South Africa. Lithos 88:150–172
Burnham AD, Berry AJ (2012) An experimental study of trace element partitioning between zircon and melt as a
function of oxygen fugacity. Geochim Cosmochim Acta 95:196–212
Cesare B, Gómez-Pugnaire MT, Rubatto D (2003) Residence time of S-type anatectic magmas beneath the Neogene
Volcanic Province of SE Spain: a zircon and monazite SHRIMP study. Contrib Mineral Petrol 146:28–43
Chen Y-X, Zheng Y-F, Chen R-X, Zhang S-B, Li Q, Dai M, Chen L (2011) Metamorphic growth and recrystallization
of zircons in extremely 18O-depleted rocks during eclogite-facies metamorphism: Evidence from U–Pb ages,
trace elements, and O–Hf isotopes. Geochim Cosmochim Acta 75:4877–4898
Cherniak DJ (2010) Diffusion in accessory minerals: Zircon, titanite, apatite, monazite and xenotime. Rev Mineral
Geochem 72:827–869
Cherniak DJ, Watson BE (2003) Diffusion in zircon. Rev Mineral Geochem 53:113–143
290 Rubatto
Claesson S, Bibikova EV, Shumlyanskyy L, Whitehouse MJ, Billström K (2016) Can oxygen isotopes in magmatic
zircon be modified by metamorphism? A case study from the Eoarchean Dniester-Bug Series, Ukrainian
Shield. Precambr Res 273:1–11
Claoué-Long JC, Sobolev NV, Shatsky VS, Sobolev AV (1991) Zircon response to diamond-pressure metamorphism
in the Kokchetav massif, USSR. Geology 19:710–713
Corfu F, Hanchar JM, Hoskin PWO, Kinny P (2003) Atlas of zircon textures. Rev Mineral Geochem 53:469–500
Degeling H, Eggins S, Ellis DJ (2001) Zr budget for metamorphic reactions, and the formation of zircon from
garnet breakdown. Mineral Mag 65:749–758
Drüppel K, Elsäßer L, Brandt S, Gerdes A (2013) Sveconorwegian mid-crustal ultrahigh-temperature
metamorphism in Rogaland, Norway: U–Pb LA-ICP-MS geochronology and pseudosections of sapphirine
granulites and associated paragneisses. J Petrol 54:305–350
Ewing RC, Meldrum A, Wang L, Weber WJ, Corrales LR (2003) Radiation effects in zircon. Rev Mineral Geochem
53:387–425
Ewing T, Hermann J, Rubatto D (2013) The robustness of the Zr-in-rutile and Ti-in-zircon thermometers during
Ferriss EDA, Essene EJ, Becker U (2008) Computational study of the effect of pressure on the Ti-in-zircon
geothermometer. Eur J Mineral 20:745–755
rutile thermometers Contrib Mineral Petrol 154:429–437
Fornelli A, Langone A, Micheletti F, Pascazio A, Piccarreta G (2014) The role of trace element partitioning between
garnet, zircon and orthopyroxene on the interpretation of zircon U–Pb ages: An example from high-grade
basement in Calabria (Southern Italy). Int J Earth Sci 103:487–507
Franz G, Morteani G, Rhede D (2015) Xenotime-(Y) formation from zircon dissolution–precipitation and HREE
fractionation: an example from a metamorphosed phosphatic sandstone, Espinhaço fold belt (Brazil). Contrib
Mineral Petrol 170
Fu B, Page FZ, Cavosie AJ, Fournelle J, Kita NT, Star Lackey J, Wilde SA, Valley JW (2008) Ti-in-zircon
thermometry: applications and limitations. Contrib Mineral Petrol 156:197–215
Fu B, Valley JW, Kita NT, Spicuzza MJ, Paton C, Tsujimori T, Bröcker M, Harlow GE (2010) Multiple origins of
zircons in jadeitite. Contrib Mineral Petrol 159:769–780
Fu B, Paul B, Cliff J, Bröcker M, Bulle F (2012) O–Hf isotope constraints on the origin of zircon in high-pressure
mélange blocks and associated matrix rocks from Tinos and Syros, Greece. Eur J Mineral 24:277–287
Ganade de Araujo CE, Rubatto D, Hermann J, Cordani UG, Caby R, Basei MAS (2014) Ediacaran 2,500-km-long
synchronous deep continental subduction in the West Gondwana Orogen. Nature Commun 5:5198
Gao X-Y, Zheng Y-F, Chen Y-X, Tang H-L, Li W-C (2015) Zircon geochemistry records the action of metamorphic
fluid on the formation of ultrahigh-pressure jadeite quartzite in the Dabie orogen. Chem Geol 419:158–175
Gasser D, Rubatto D, Bruand E, Stüwe K (2012) Large-scale, short-lived metamorphism, deformation, and
magmatism in the Chugach metamorphic complex, southern Alaska: A SHRIMP U–Pb study of zircons.
Geol Soc Am Bull 124:886–905
and oxygen isotope analysis. Contrib Mineral Petrol 171:15
Gebauer D, Schertl H-P, Brix M, Schreyer W (1997) 35 Ma old ultrahigh-pressure metamorphism and evidence for
very rapid exhumation in the Dora Maira Massif, Western Alps. Lithos 41:5–24
Geisler T, Pidgeon RT, van Bronswijk W, Pleysier R (2001a) Kinetics of thermal recovery and recrystallization of
partially metamict zircon: a Raman spectroscopic study. Eur J Mineral 13:1163–1176
Geisler T, Ulonska M, Schleicher H, Pidgeon RT, van Bronswijk W (2001b) Leaching and differential
recrystallization of metamict zircon under experimental hydrothermal conditions. Contrib Mineral Petrol
141:53–65
Geisler T, Pidgeon RT, Kurtz R, van Bronswijk W, Schleicher H (2003a) Experimental hydrothermal alteration of
partially metamict zircon. Am Mineral 88:1496–1543
Geisler T, Rashwan AA, Rahn MKW, Poller U, Zwingmann H, Pidgeon RT, Schleicher H, Tomaschek F (2003b)
Low-temperature hydrothermal alteration of natural metamict zircons from the Eastern Desert, Egypt.
Mineral Mag 67:485–508
Geisler T, Schaltegger U, Tomaschek F (2007) Re-equilibration of zircon in acqueous fluids and melts. Elements
3:43–50
Gerdes A, Zeh A (2009) Zircon formation versus zircon alteration—New insights from combined U–Pb and Lu–Hf
in-situ LA-ICP-MS analyses, and consequences for the interpretation of Archean zircon from the Central
Zone of the Limpopo Belt. Chem Geol 261:230–243
Gilotti JA, Nutman AP, Brueckner HK (2004) Devonian to Carboniferous collision in the Greenland Caledonides:
U–Pb zircon and Sm–Nd ages of high-pressure and ultrahigh-pressure metamorphism. Contrib Mineral
Petrol 148:216–235
Gordon SM, Little TA, Hacker BR, Bowring SA, Korchinski M, Baldwin SL, Kylander-Clark ARC (2012) Multi-
stage exhumation of young UHP–HP rocks: Timescales of melt crystallization in the D’Entrecasteaux
Islands, southeastern Papua New Guinea. Earth Planet Sci Lett 351–352:237–246
Gordon SM, Whitney DL, Teyssier C, Fossen H (2013) U–Pb dates and trace-element geochemistry of zircon
from migmatite, Western Gneiss Region, Norway: Significance for history of partial melting in continental
subduction. Lithos 170–171:35–53
Gregory RT, Taylor HP (1981) An oxygen isotope profile in a section of Cretaceous oceanic crust, Samail Ophiolite,
Oman: Evidence for δ18O buffering of the oceans by deep (>5 km) seawater-hydrothermal circulation at mid-
ocean ridges. J Geophys Res 86:2737–2755
Grimes CB, Wooden JL, Cheadle MJ, John BE (2015) “Fingerprinting” tectono-magmatic provenance using trace
elements in igneous zircon. Contrib Mineral Petrol 170:1–26
Hacker RB, Ratschbacher L, Webb L, Ireland T, Walker D, Dong S (1998) U/Pb zircon ages constrain the
architecture of the ultrahigh-pressure Qinling–Dabie Orogen. Earth Planet Sci Lett 161:215–230
Hanchar JM, Miller CF (1993) Zircon zonation patterns as revealed by cathodoluminescence and backscattered
electron images: Implications for interpretation of complex crustal histories. Chem Geol 110:1–13
Hanchar JM, Rudnick RL (1995) Revealing hidden structures: The application of cathodoluminescence and back-
scattered electron imaging to dating zircons from lower crust xenoliths. Lithos 36:289–303
Harley SL, Nandakumar V (2014) Accessory mineral behaviour in granulite migmatites: a case study from the
Kerala Khondalite Belt, India. J Petrol 55:1965–2002
Harley SL, Kelly NM, Möller A (2007) Zircon behaviour and the thermal histories of mountain chains. Elements
3:25–30
Harrison TM, Bell EA, Boehnke P (2017) Hadean zircon petrochronology. Rev Mineral Geochem 83:329–363
Lett 261:9–19
Hay DC, Dempster TJ (2009a) Zircon alteration, formation and preservation in sandstones. Sedimentology
56:2175–2191
Hay DC, Dempster TJ (2009b) Zircon behaviour during low-temperature metamorphism J Petrol 50:571–589
Hay DC, Dempster TJ, Lee MR, Brown DJ (2010) Anatomy of a low temperature zircon outgrowth. Contrib
Hermann J, Rubatto D (2014) Subduction of continental crust to mantle depth: Geochemistry of ultrahigh-pressure
rocks. In: Rudnick R (ed) The Crust vol 4. Elsevier Amsterdam pp 309–340
diamondiferous, deeply subducted continental crust (Kokchetav massif, Kazakhstan). Contrib Mineral Petrol
141:66–82
Hermann J, Rubatto D, Korsakov A, Shatsky VS (2006a) The age of metamorphism of diamondiferous rocks
determined with SHRIMP dating of zircon. Russ Geol Geophys 47:513–520
Hermann J, Rubatto D, Trommsdorff V (2006b) Sub-solidus Oligocene zircon formation in garnet peridotite during
fast decompression and fluid infiltration (Duria, Central Alps). Mineral Petrol 88:181–206
Hermann J, Spandler C, Hack A, Korsakov AV (2006c) Aqueous fluids and hydrous melts in high-pressure and
ultra-high pressure rocks: Implications for element transfer in subduction zones. Lithos 92:399–417
Hetherington CJ, Harlov DE, Budzyn B (2010) Experimental metasomatism of monazite and xenotime: Mineral
stability, REE mobility and fluid composition. Mineral Petrol 99:165–184
Hokada T, Harley SL (2004) Zircon growth in UHT leucosome: constraints from zircon–garnet rare earth elements
(REE) relations in Napier Complex, East Antarctica. J Mineral Petrol Sci 99:180–190
Hopkins M, Harrison TM, Manning CE (2008) Low heat flow inferred from > 4 Gyr zircons suggests Hadean plate
boundary interactions. Nature 456:493–496
Hoskin PWO, Black LP (2000) Metamorphic zircon formation by solid-state recrystallization of protolith igneous
zircon. J Metamorph Geol 18:423–439
Kaczmarek M-A, Müntener O, Rubatto D (2008) Trace element chemistry and U–Pb dating of zircons from
oceanic gabbros and their relationship with whole rock composition (Lanzo, Italian Alps). Contrib Mineral
Petrol 155:295–312
Katayama I, Maruyama S (2009) Inclusion study in zircon from ultrahigh-pressure metamorphic rocks in the
Kokchetav massif: an excellent tracer of metamorphic history. J Geol Soc 166:783–796
292 Rubatto
Katayama I, Maruyama S, Parkinson CD, Terada K, Sano Y (2001) Ion micro-probe U–Pb zircon geochronology
of peak and retrograde stages of ultrahigh-pressure metamorphic rocks from the Kokchetav massif, northern
Kazakhstan. Earth Planet Sci Lett 188:185–198
Kelly N, Harley S (2005) An integrated microtextural and chemical approach to zircon geochronology: refining the
Archean history of the Napier Complex, east Antarctica. Contrib Mineral Petrol 149:57–84
SiO2–H2O–TiO2–ZrO2 system. J Metamorph Geol 29:151–166
Kohn MJ (1999) Why most “dry” rocks should cool “wet”. Am Mineral 84:570–580
Mineral 101:25–42
Geochem 83:103–152
Korhonen FJ, Clark C, Brown M, Bhattacharya S, Taylor R (2013) How long-lived is ultrahigh temperature (UHT)
metamorphism? Constraints from zircon and monazite geochronology in the Eastern Ghats orogenic belt,
India. Precambr Res 234:322–350
Kotkova J, Harley SL (2010) Anatexis during high-pressure crustal metamorphism: evidence from garnet–whole-
rock REE relationships and zircon–rutile Ti–Zr thermometry in leucogranulites from the Bohemian Massif.
J Petrol 51:1967–2001
Kusiak MA, Whitehouse MJ, Wilde SA, Dunkley DJ, Menneken M, Nemchin AA, Clark C (2013a) Changes in zircon
chemistry during archean UHT metamorphism in the Napier Complex, Antarctica. Am J Sci 313:933–967
Kusiak MA, Whitehouse MJ, Wilde SA, Nemchin AA, Clark C (2013b) Mobilization of radiogenic Pb in zircon
revealed by ion imaging: Implications for early Earth geochronology. Geology 41:291–294
Kusiak MA, Dunkley DJ, Wirth R, Whitehouse MJ, Wilde SA, Marquardt K (2015) Metallic lead nanospheres
discovered in ancient zircons. PNAS 112:4958–4963
Kylander-Clark ARC, Hacker BR, Cottle JM (2013) Laser-ablation split-stream ICP petrochronology Chem Geol
345:99–112
Lenting C, Geisler T, Gerdes A, Kooijman E, Scherer EE, Zeh A (2010) The behavior of the Hf isotope system in
radiation-damaged zircon during experimental hydrothermal alteration. Am Mineral 95:1343–1348
López Sánchez-Vizcaíno V, Rubatto D, Gómez-Pugnaire MT, Trommsdorff V, Müntener O (2001) Middle Miocene
HP metamorphism and fast exhumation of the Nevado Filabride Complex, SE Spain. Terra Nova 13:327–332
Martin L, Duchêne S, Deloule E, Vanderhaeghe O (2006) The isotopic composition of zircon and garnet: a record
of the metamorphic history of Naxos (Greece). Lithos 87:174–192
Martin LAJ, Duchêne S, Deloule E, Vanderhaeghe O (2008) Mobility of trace elements and oxygen in zircon
during metamorphism: Consequences for geochemical tracing. Earth Planet Sci Lett 267:161–174
Martin L, Rubatto D, Crepisson C, Hermann J, Putlitz B, Vitale-Brovarone A (2014) Garnet oxygen analysis by
SHRIMP-SI: matrix corrections and application to high pressure metasomatic rocks from Alpine Corsica.
Chem Geol 374–375:25–36
McLaren AC, Fitz Gerald JD, Williams IS (1994) The microstructure of zircon and its influence on the age
determiantion from Pb/U isotopic ratios measured by ion microprobe. Geochim Cosmochim Acta 58:993–1005
Meldrum A, Boatner LA, Zinkle SJ, Wang S-X, Wang L-M, Ewing RC (1999) Effects of dose rate and temperature
on the crystalline-to-metamict transformation in the ABO4 orthosilicates. Can Mineral 37:207–221
Menneken M, Nemchin AA, Geisler T, Pidgeon RT, Wilde SA (2007) Hadean diamonds in zircon from Jack Hills,
Western Australia. Nature 448:917-U915
Miller JA, Cartwright I, Buick I, Barnicoat A (2001) An O-isotope profile through the HP–LT Corsican ophiolite,
France and its implications for fluid flow during subduction. Chem Geol 178:43–69
Möller A, O’Brien PJ, Kennedy A, Kröner A (2002) Polyphase zircon in ultrahigh-temperature granulites
(Rogaland, SW Norway): Constraints for Pb diffusion in zircon. J Metamorph Geol 20:727–740
Möller A, O’Brien PJ, Kennedy A, Kröner A (2003) Linking growth episodes of zircon and metamorphic textures
to zircon chemistry: an example from the ultrahigh-temperature granulites of Rogaland (SW Norway). Geol
Soc, London, Spec Publ 220:65–81
Montero P, Bea F, Zinger TF, Scarrow JH, Molina JF, Whitehouse M (2004) 55 million years of continuous anatexis
in Central Iberia: Single-zircon dating of the Peña Negra Complex. J Geol Soc 161:255–263
Mori Y, Orihashi Y, Miyamoto T, Shimada K, Shigeno M, Nishiyama T (2011) Origin of zircon in jadeitite from the
Nishisonogi metamorphic rocks, Kyushu, Japan. J Metamorph Geol 29:673–684
Nasdala L, Hanchar JM (2005) Comment on: Application of Raman spectroscopy to distinguish metamorphic and
igneous zircon (Xian et al., Anal Lett 2004, v. 37, p. 119). Anal Lett 38:727–734
Nasdala L, Lengauer CL, Hanchar JM, Kronz A, Wirth R, Blanc P, Kennedy AK, Seydoux-Guillaume AM (2002)
Annealing radiation damage and the recovery of cathodoluminescence. Chem Geol 191:121–140
Nasdala L, Zhang M, Kempe U, Panczer G, Gaft M, Andrut M, Plötze M (2003) Spectroscopic methods applied to
zircon. Rev Mineral Geochem 53:427–467
Page FZ, Essene EJ, Mukasa SB, Valley JW (2014) A garnet-zircon oxygen isotope record of subduction and
exhumation fluids from the Franciscan complex, California. J Petrol 55:103–131
Page FZ, Ushikubo T, Kita NT, Riciputi LR, Valley JW (2007) High-precision oxygen isotope analysis of picogram
samples reveals 2 μm gradients and slow diffusion in zircon. Am Mineral 92:1772–1775
temperature (UHT) rocks. Contrib Mineral Petrol 171:1–20
Petersson A, Scherstén A, Andersson J, Whitehouse MJ, Baranoski MT (2015) Zircon U–Pb, Hf and O isotope
constraints on growth versus reworking of continental crust in the subsurface Grenville orogen, Ohio, USA.
Phillips G, Rubatto D, Phillips D, Offler R (2015) High-pressure metamorphism in the southern New England
Orogen: implications for long-lived accretionary orogenesis in eastern Australia. Tectonics 34:1979–2010
Piazolo S, Austrheim H, Whitehouse M (2012) Brittle–ductile microfabrics in naturally deformed zircon:
Deformation mechanisms and consequences for U–Pb dating. Am Mineral 97:1544–1563
Poller U, Huth J, Hoppe P, Williams IS (2001) REE, U, TH, and HF distribution in zircon from Western Carpathian
Variscan granitoids: A combined cathodoluminescence and ion microprobe study. Am J Sci 301:858–876
Putlitz B, Matthews A, Valley JW (2000) Oxygen and hydrogen isotope study of high-pressure metagabbros and
metabasalts (Cyclades, Greece): implications for the subduction of oceanic crust. Contrib Mineral Petrol
138:114–126
Putnis A (2009) Mineral replacement reactions. Rev Mineral Geochem 70:87–124
Rasmussen B (2005) Zircon growth in very low grade metasedimentary rocks: evidence for zirconium mobility at
~250 °C. Contrib Mineral Petrol 150:146–155
Rasmussen B, Fletcher IR, Muhling JR, Gregory CJ, Wilde SA (2011) Metamorphic replacement of mineral inclusions
in detrital zircon from Jack Hills, Australia: Implications for the Hadean Earth. Geology 39:1143–1146
Reddy SM, Timms NE, Trimby P, Kinny PD, Buchan C, Blake K (2006) Crystal-plastic deformation of zircon: A
defect in the assumption of chemical robustness. Geology 34:257–260
Reddy SM, Timms NE, Pantleon W, Trimby P (2007) Quantitative characterization of plastic deformation of zircon
and geological implications. Contrib Mineral Petrol 153:625–645
Reddy SM, Timms NE, Eglington BM (2008) Electron backscatter diffraction analysis of zircon: A systematic
assessment of match unit characteristics and pattern indexing optimization. Am Mineral 93:187–197
Reddy SM, Timms NE, Hamilton PJ, Smyth HR (2009) Deformation-related microstructures in magmatic zircon
and implications for diffusion. Contrib Mineral Petrol 157:231–244
Reddy SM, Clark C, Timms NE, Eglington BM (2010) Electron backscatter diffraction analysis and orientation
mapping of monazite. Mineral Mag 74:493–506
Root DB, Hacker BR, Mattinson JM, Wooden JL (2004) Zircon geochronology and ca. 400 Ma exhumation of
Norwegian ultrahigh-pressure rocks: an ion microprobe and chemical abrasion study. Earth Planet Sci Lett
228:325–341
Rubatto D (2002) Zircon trace element geochemistry: distribution coefficients and the link between U–Pb ages and
Rubatto D, Angiboust S (2015) Oxygen isotope record of oceanic and high-pressure metasomatism: a P–T–time–
fluid path for the Monviso eclogites (Italy). Contrib Mineral Petrol 170:44
Rubatto D, Chakraborty S, Dasgupta S (2013) Timescales of crustal melting in the Higher Himalayan Crystallines
Rubatto D, Gebauer D (2000) Use of cathodoluminescence for U–Pb zircon dating by ion microprobe: some
examples from the Western Alps. In: Pagel M, Barbin V, Blanc P, Ohnenstetter D (eds) Cathodoluminescence
in geosciences, vol. Springer, Berlin Heidelberg New York, pp 373–400
Rubatto D, Gebauer D, Compagnoni R (1999) Dating of eclogite-facies zircons: the age of Alpine metamorphism
in the Sesia-Lanzo Zone (Western Alps). Earth Planet Sci Lett 167:141–158
Rubatto D, Hermann J (2003) Zircon formation during fluid circulation in eclogites (Monviso, Western Alps):
implications for Zr and Hf budget in subduction zones. Geochim Cosmochim Acta 67:2173–2187
Rubatto D, Hermann J (2007a) Experimental zircon/melt and zircon/garnet trace element partitioning and
implications for the geochronology of crustal rocks. Chem Geol 241:62–87
Rubatto D, Hermann J (2007b) Zircon behaviour in deeply subducted rocks. Elements 3:31–35
Rubatto D, Hermann J, Buick IS (2006) Temperature and bulk composition control on the growth of monazite and
zircon during low-pressure anatexis (Mount Stafford, central Australia). J Petrol 47:1973–1996
294 Rubatto
Rubatto D, Müntener O, Barnhorn A, Gregory C (2008) Dissolution–reprecipitation of zircon at low-temperature,

high-pressure conditions (Lanzo Massif, Italy). Am Mineral 93:1519–1529
Rubatto D, Hermann J, Berger A, Engi M (2009) Protracted fluid-induced melting during Barrovian metamorphism
in the Central Alps. Contrib Mineral Petrol 158:703–722
Rubatto D, Regis D, Hermann J, Boston K, Engi M, Beltrando M, McAlpine SRB (2011) Yo-Yo subduction
recorded by accessory minerals in the Sesia Zone, Western Alps. Nature Geosci 4:338–342
Rumble D, Giorgis D, Ireland T, Zhang Z, Xu H, Yui TF, Yang J, Xu Z, Liou JG (2002) Low δ18O zircons, U–Pb
dating, and the age of the Qinglongshan oxygen and hydrogen isotope anomaly near Donghai in Jiangsu
Province, China. Geochim Cosmochim Acta 66:2299–2306
Schaltegger U, Fanning M, Günther D, Maurin JC, Schulmann K, Gebauer D (1999) Growth, annealing and
recrystallization of zircon and preservation of monazite in high-grade metamorphism: conventional and in-
situ U–Pb isotope, cathodoluminescence and microchemical evidence. Contrib Mineral Petrol 134:186–201
Geochem 83:199–230
damage, and disturbance of the U–Pb and Th–Pb chronometers. Chem Geol 330–331:140–158
Shatsky VS, Sobolev AV (2003) The Kokchetav massif, Kazakhstan. In: EMU Notes in Mineralogy Vol. 5:
Ultrahigh Pressure Metamorphism, p. 75–103
Sheng Y-M, Zheng Y-F, Chen R-X, Li Q, Dai M (2012) Fluid action on zircon growth and recrystallization during
quartz veining within UHP eclogite: Insights from U–Pb ages, O–Hf isotopes and trace elements. Lithos
136–139:126–144
Spandler C, Hermann J, Rubatto D (2004) Exsolution of thortveitite, yttrialite and xenotime during low temperature
recrystallization of zircon from New Caledonia, and their significance for trace element incorporation in
zircon. Am Mineral 89:1795–1806
Spandler C, Rubatto D, Hermann J (2005) Late Cretaceous–Tertiary tectonics of the southern Pacific; insight
from U–Pb SHRIMP dating of eclogite-facies rocks from New Caledonia. Tectonics 24:TC3003,
doi:3010.1029/2004TC001709
Spandler C, Pettke T, Rubatto D (2011) Internal and external fluid sources for eclogite-facies veins in the Monviso
meta-ophiolite, Western Alps: Implications for fluid flow in subduction zones. J Petrol 52:1207–1236
implications for the REE, Th and U geochemistry of crustal rocks. Chem Geol 300–301:200–220
Stepanov A, Hermann J, Korsakov AV, Rubatto D (2014) Geochemistry of ultrahigh-pressure anatexis: fractionation
of elements in the Kokchetav gneisses during melting at diamond-facies conditions. Contrib Mineral Petrol
167:1002
Stepanov A, Hermann J, Rubatto D, Korsakov AV, Danyushevsky (2016a) Melting history of an ultrahigh-pressure
paragneiss revealed by multiphase solid inclusions in garnet, Kokchetav massif, Kazakhstan. J Petrol 57:
1531–1554
Stepanov A, Rubatto D, Hermann J, Korsakov AV (2016b) Constrasting P–T paths within the Barchi-Kol UHP terrain
(Kokchetav Complex): Implications for subduction and exhumation of continental crust. Am Mineral 101:788
Tailby ND, Walker AM, Berry AJ, Hermann J, Evans KA, Mavrogenes JA, O’Neill HSC, Rodina IS, Soldatov AV,
Rubatto D, Newville M, Sutton SR (2011) Ti site occupancy in zircon. Geochim Cosmochim Acta 75:905–921
Taylor RJM, Harley SL, Hinton RW, Elphick S, Clark C, Kelly NM (2014) Experimental determination of REE
partition coefficients between zircon, garnet and melt: A key to understanding high-T crustal processes. J
Tichomirowa M, Whitehouse MJ, Nasdala L (2005) Resorption, growth, solid state recrystallisation, and annealing
of granulite facies zircon—a case study from the Central Erzgebirge, Bohemian Massif. Lithos 82:25–50
Timms NE, Kinny PD, Reddy SM (2006) Deformation-related modification of U and Th in zircon. Geochim
Cosmochim Acta 70:A651
Timms NE, Kinny PD, Reddy SM, Evans K, Clark C, Healy D (2011) Relationship among titanium, rare earth
elements, U–Pb ages and deformation microstructures in zircon: Implications for Ti-in-zircon thermometry.
Tomaschek F, Kennedy AK, Villa IM, Lagos M, Ballhaus C (2003) Zircons from Syros, Cyclades, Greece—
recrystallization and mobilization of zircon during high-pressure metamorphism. J Petrol 44:1977–2002
Valley JW (2003) Oxygen isotopes in zircon. Rev Mineral Geochem 53:343–385
Valley PM, Fisher CM, Hanchar JM, Lam R, Tubrett M (2010) Hafnium isotopes in zircon: A tracer of fluid–rock
interaction during magnetite–apatite (“Kiruna-type”) mineralization. Chem Geol 275:208–220
Valley JW, Cavosie AJ, Ushikubo T, Reinhard DA, Lawrence DF, Larson DJ, Clifton PH, Kelly TF, Wilde SA,
Moser DE, Spicuzza MJ (2014) Hadean age for a post-magma-ocean zircon confirmed by atom-probe
tomography. Nature Geosci 7:219–223
Vavra G, Gebauer D, Schmidt R, Compston W (1996) Multiple zircon growth and recrystallization during
polyphase Late Carboniferous to Triassic metamorphism in granulites of the Ivrea Zone (Southern Alps): an
ion microprobe (SHRIMP) study. Contrib Mineral Petrol 122:337–358
Vervoort JD, Kemp AIS (2016) Clarifying the zircon Hf isotope record of crust–mantle evolution. Chem Geol
425:65–75
Vonlanthen P, Fitz Gerald JD, Rubatto D, Hermann J (2012) Recrystallization rims in zircon (Valle d’Arbedo.
Switzerland): An integrated cathodoluminescence, LA-ICP-MS, SHRIMP, and TEM study. Am Mineral
97:369–377
Watson BE, Cherniak DJ (1997) Oxygen diffusion in zircon. Earth Planet Sci Lett 148:527–544
Watson EB, Harrison TM (2005) Zircon thermometer reveals minimum melting conditions on earliest Earth.
Science 308:841–844
Whitehouse M, Kemp AIS (2010) On the difficulty of assigning crustal residence, magmatic protolith and
metamorphic ages to Lewisian granulites: constraints from combined in situ U–Pb and Lu–Hf isotopes.
Geological Society, London, Special Pubblications 335:81–101
Whitehouse MJ, Platt JP (2003) Dating high-grade metamorphism: constraints from rare-earth elements in zircon
and garnet. Contrib Mineral Petrol 145:61–74
Whitehouse MJ, Ravindra Kumar GR, Rimša A (2014) Behaviour of radiogenic Pb in zircon during ultrahigh-
temperature metamorphism: An ion imaging and ion tomography case study from the Kerala Khondalite Belt,
southern India. Contrib Mineral Petrol 168:1–18
Williams I, Compston W, Black L, Ireland T, Foster J (1984) Unsupported radiogenic Pb in zircon: a cause of
anomalously high Pb–Pb, U–Pb and Th–Pb ages. Contrib Mineral Petrol 88:322–327
Williams IS (2001) Response of detrital zircon and monazite, and their U–Pb isotopic systems, to regional
metamorphism and host-rock partial melting, Cooma Complex, southeastern Australia. Aust J Earth Sci
48:557–580
Wu Y-B, Zheng Y-F, Zhao Z-F, Gong B, Liu XM, Wu F-Y (2006) U–Pb, Hf and O isotope evidence for two episodes
of fluid-assisted zircon growth in marble-hosted eclogites from the Dabie orogen. Geochim Cosmochim Acta
70:3743–3761
Wu YB, Gao S, Zhang HF, Yang SH, Jiao WF, Liu YS, Yuan HL (2008a) Timing of UHP metamorphism in the
Hong’an area, western Dabie Mountains, China: Evidence from zircon U–Pb age, trace element and Hf
isotope composition. Contrib Mineral Petrol 155:123–133
Wu YB, Zheng YF, Gao S, Jiao WF, Liu YS (2008b) Zircon U–Pb age and trace element evidence for Paleoproterozoic
granulite-facies metamorphism and Archean crustal rocks in the Dabie Orogen. Lithos 101:308–322
Xia QX, Zheng YF, Yuan H, Wu FY (2009) Contrasting Lu–Hf and U–Th–Pb isotope systematics between
metamorphic growth and recrystallization of zircon from eclogite-facies metagranites in the Dabie orogen,
China. Lithos 112:477–496
Xian WS, Sun M, Malpas J, Zhao GC, Zhou MF, Ye K, Liu JB, Phillips DL (2004) Application of Raman
spectroscopy to distinguish metamorphic and igneous zircons. Anal Lett 37:119–130
Yakymchuk C, Brown M (2014) Behaviour of zircon and monazite during crustal melting. J Geol Soc 171:465–479
Ye K, Yao Y, Katayama I, Cong B, Wang Q, Maruyama S (2000) Large areal extent of ultrahigh-pressure
metamorphism in the Sulu ultrahigh-pressure terrane of East China: new implications from coesite and
omphacite inclusions in zircon of granitic gneiss. Lithos 52:157–164
Young DJ, Kylander-Clark ARC (2015) Does continental crust transform during eclogite facies metamorphism? J
Zhang ZM, Schertl HP, Wang JL, Shen K, Liou JG (2009a) Source of coesite inclusions within inherited magmatic
zircon from Sulu UHP rocks, eastern China, and their bearing for fluid–rock interaction and SHRIMP dating.
Zhang ZM, Shen K, Wang JL, Dong HL (2009b) Petrological and geochronological constraints on the formation,
subduction and exhumation of the continental crust in the southern Sulu orogen, eastern-central China.
Tectonophysics 475:291–307
Zhao ZF, Zheng YF, Wei CS, Chen FK, Liu X, Wu FY (2008) Zircon U–Pb ages, Hf and O isotopes constrain the
crustal architecture of the ultrahigh-pressure Dabie orogen in China. Chem Geol 253:222–242
Zhao L, Li T, Peng P, Guo J, Wang W, Wang H, Santosh M, Zhai M (2015) Anatomy of zircon growth in high
pressure granulites: SIMS U–Pb geochronology and Lu–Hf isotopes from the Jiaobei Terrane, eastern North
China Craton. Gondwana Res 28:1373–1390
Zheng YF, Fu B, Gong B, Li L (2003) Stable isotope geochemistry of ultra-high pressure metamorphic rocks from
the Dabie-Sulu orogen in China; implications for geodynamics and fluid regime. Earth-Sci Rev 62:105–161
Zheng YF, Wu Y-B, Zhao Z-F, Zhang S-B, Xu P, Wu F-Y (2005) Metamorphic effect on zircon Lu–Hf and
U–Pb isotope systems in ultrahigh-pressure eclogite-facies metagranite and metabasalt. Earth Planet Sci Lett
240:378–400
Zheng YF, Gong B, Zhao ZF, Wu YB, Chen FK (2008) Zircon U–Pb age and O isotope evidence for Neoproterozoic
low-18O magnetism during supercontinental rifting in South China: Implications for the snowball earth event.
Am J Sci 308:484–516
Vol. 83 pp. 297–328, 2017 10
Petrochronology of Zircon and Baddeleyite

in Igneous Rocks: Reconstructing Magmatic Processes
at High Temporal Resolution
Urs Schaltegger
Department of Earth Sciences
University of Geneva
1205 Geneva,
Switzerland
urs.schaltegger@unige.ch
Joshua H.F.L. Davies

University of Geneva
1205 Geneva
Switzerland
joshua.davies@unige.ch
INTRODUCTION
Zircon (ZrSiO4) and baddeleyite (ZrO2) are common accessory minerals in igneous rocks
of felsic to mafic composition. Both minerals host trace elements substituting for Zr, among
them Hf, Th, U, Y, REEs and many more. The excellent chemical and physical resistivity of
zircon makes this mineral a perfect archive of chemical and temporal information to trace
geological processes in the past, utilizing the outstanding power and temporal resolution of the
U–Pb decay schemes. Baddeleyite is a chemically and physically much more fragile mineral.
It preserves similar information only where it is shielded from dissolution and physical
fragmentation as an inclusion in other minerals or in a fine-grained or non-reactive rock matrix.
It offers the potential for dating the solidification of mafic rocks with high-precision through
its crystallization in small pockets of Zr-enriched melt, after extensive olivine and pyroxene
fractionation. Zircon and baddelelyite U–Pb dates are, for an overwhelming majority of cases
and where we can assume a closed system, considered to reflect the time of crystallization.
The development of the U–Pb dating tool CA-ID-TIMS (chemical abrasion-isotope
dilution-thermal ionization mass spectrometry) since 2005 has led to unprecedented precision
of better than 0.1% in 206Pb / 238U dates (Bowring et al. 2005). Increased sensitivity of mass
spectrometers and low laboratory blanks due to reduction of acid volumes allow routine U–Pb
age determinations of micrograms of material at sufficiently high radiogenic / common lead
ratios (see Schoene and Baxter 2017, this volume).
In situ U–Pb age analysis using laser ablation or primary ion beam sputtering allows
analysis of sub-microgram quantities of zircon material from polished internal sections or
zircon surfaces with spot diameters ranging from ~30 µm for laser-ablation, inductively
coupled plasma mass spectrometry (LA-ICP-MS) to 10 µm for secondary ion mass
spectrometry (SIMS), lateral resolutions of 2–5 µm for NanoSIMS (Yang et al. 2012), or
from high-voltage pulse induced evaporation of needle-tips of 200 nm diameter through
1943-2666/17/0083-0010$05.00 (online)
298 Schaltegger & Davies
atom probe tomography (APT; Valley et al. 2015). Increased spatial resolution is logically
linked to a drastic decrease of analyzed volume and to increased analytical uncertainty.
In contrast, ID-TIMS analysis integrates the chemical and isotopic information over the
analyzed volume of an entire grain or parts of a fragmented grain, which yields integrated
compositional and temporal information at high precision. The dilemma of low-precision at
high spatial resolution and high-precision at low spatial resolution is in many cases asking
for an intelligent combination of the two approaches, both for elemental and isotopic analysis
(Schaltegger et al. 2015). However, to resolve the timescale of processes that are active in
magmatic systems, we need temporal information in the 103–105 years range, precluding
the use of in situ dating techniques for rocks that are older than ~10 Ma; ID-TIMS analysis
can provide the necessary temporal resolution up to ages of ~500 Ma. In rare cases, precise
207
Pb / 206Pb dates from Proterozoic and Archean zircon may provide sufficient resolution to
trace magmatic processes (e.g., Schoene and Bowring 2010; Zeh et al. 2015).
In this chapter we concentrate mainly on the approach of combining high-precision
CA-ID-TIMS dating of zircon and baddeleyite with compositional imaging and analysis to
obtain the maximum amount of information on the petrogenesis of the host rock as well as the
physio-chemical conditions during crystallization at maximum temporal resolution. We also
give credit to low-precision U–Pb petrochronology utilizing LA-ICP-MS, as well as SIMS, for
the latter with a focus on disequilibrium dating of young volcanic zircon.
“Petrochronology” we define as understanding the petrological evolution of a given rock
and a given mineral assemblage in an absolute time frame that has been established by radio-
isotopic dating.
WORKFLOW FOR ZIRCON PETROCHRONOLOGY

The aim of petrochronology is to relate an age determination as closely as possible with
elemental and isotopic analysis of the dated minerals. Looking beyond the horizon of zircon
geochronology, we may want to analyze trace element concentrations and / or Hf, Nd, Pb or Sr
isotopes in a variety of U-bearing minerals amenable to U–Pb dating (titanite, monazite, xenotime,
rutile and potentially others). Concentrating on the mineral zircon, the analytical workflows are
significantly different depending on the choice of the dating method – high temporal resolution
U–Pb dating using CA-ID-TIMS, or high spatial resolution LA-ICP-MS or SIMS techniques.
The concept of petrochronology was initially developed for studying the growth of minerals
in metamorphic systems, where a combination of composition and age helped to determine the
timing of growth as well as the controlling metamorphic reactions (e.g., Fraser et al. 1997). For
igneous systems, petrochronology developed later, partially in combination with laser-ablation,
split-stream analysis (LASS; Yuan et al. 2008; Kylander-Clark et al. 2013). LASS refers to the
simultaneous analysis of chemical or isotopic composition and age on the exact same volume of
a previously imaged grain (backscatterd electrons, BSE, or cathodo-luminescence, CL) mounted
in epoxy resin and polished, or directly from a polished thin section. Initial chemical analysis
could be done by electron microprobe (EMP), a required step for minerals other than zircon
to quantify one element as an internal standard in LA-analysis. The LASS approach involves
splitting the ablated particle stream, suspended in a Ar–He–N2 gas mixture, and deviating,
e.g., one half into a sector-field ICP-MS (SF-ICP-MS) for U–Pb dating, the other half into a
multicollector ICP-MS (MC-ICP-MS) for Hf isotope analysis. Such setups have proven suitable
for precise and accurate age determination and Hf isotopic analysis of zircon (e.g., Fisher et
al. 2014; Reimink et al. 2016). Another setup may be designed using a quadrupole ICP-MS
(Q-ICP-MS) to determine elemental concentrations, and a SF or MC-ICP-MS for dating, when
relating, e.g., heavy rare earth element (HREE) fractionation during magmatic or metamorphic
Petrochronology of Zircon and Baddeleyite in Igneous Rocks 299
crystallization to zircon ages. If LASS capabilities are not present in a given laboratory, the
different analytical steps can be done sequentially; one may first analyse the U–Pb age in a
30 µm diameter UV-laser pit, which is later “overdrilled” by a 70 µm UV-laser pit for Hf isotope
analysis. However, age and Hf isotope information cannot always be considered identical since
the volumes analysed differ. This can lead to artificial Hf isotope variation in Archean and
Hadean zircon due to inaccurate correction for the proportion of radiogenic Hf produced in situ
from the radioactive decay of the parent isotope 176Lu (e.g., Kemp et al. 2010).
LASS was successfully used to address the question of zircon growth in the stability
field of garnet through HREE analysis in high-pressure metamorphic terranes (e.g., Yuan
et al. 2008; Kylander-Clark et al. 2013, and references therein). An example for sequential
use of SIMS for U–Pb age determination, REE and Y elemental concentration analysis and
O isotope analysis, followed by LA-MC-ICP-MS analysis of Hf isotope composition in
diamond bearing zircon may be found in Kotková et al. (2016). Metamorphic reactions
or short periods of zircon saturation in a magma may lead to thin overgrowths of zircon
on pre-existing grains; such overgrowths may be analysed for age and elemental / isotopic
composition using depth profiling on a SIMS (e.g., Reid and Coath 2000; Schmitt 2011), or
by single shot analysis with a LA-ICP-MS equipment (e.g., Cottle et al. 2012).
SIMS analysis is the method of choice when applying 238U–232Th–230Th dating
techniqes to zircons from young volcanic rocks, utilizing initial 230Th disequilibrium during
the crystallization of zircon. However, the ages produced are model ages that rely on an
estimated initial 232Th / 230Th ratio, usually obtained from the whole rock, and therefore may
not be acurate (Reid et al. 1997). The major benefit that can be obtained from disequilibrium
zircon dating by SIMS is depth profiling of individual zircon grains for age, trace elements
and isotopic composition (e.g., Storm et al. 2014; Rubin et al. 2016). Using these tools,
the geochemical evolution of the zircon can be linked to ages in the 10’s of ka to reveal
magmatic processes, protracted zircon growth or recycling of old grains. In a general sense,
U-series petrochronology has uncovered identical magmatic processes at young time scales
to conventional U–Pb petrochronology. However, this technique is limited to ages younger
than ~500 ka, and the ages produced have to be considered as model ages. In the following
we focus our discussion mostly to examples of conventional U–Pb petrochronology.
All of the approaches descrived above are inadequate for resolving rapid geological
processes in rocks that are older than ~100 Ma. For these we need to utilize high-precision U–Pb
dating techniques involving CA-ID-TIMS. A complete workflow for zircon petrochronology
involving CA-ID-TIMS high-precision U–Pb age determination is complex and consists of a
combination of in situ and bulk-dissolution techniques (Fig. 1).
A) Imaging of the textural relationships from a thin section, using BSE images from a
scanning electron microscope (SEM) or BSE and X-ray imaging on a QEMScan (Quantitative
Evaluation of Minerals by Scanning electron microscopy). This analysis provides detailed
information on the textural position of zircon and baddeleyite in the rock, thus allowing
selective sampling of an accessory phase according to its paragenetic position. In addition, the
QEMScan data set yield grain size distribution statistics for zircon if the imaging resolution is
appropriately selected. This textural characterization prior to sampling of zircon directly from
the imaged section will become increasingly important, since it allows us to directly link a
paragenetic sequence and information with chemical composition and date.
B) Selected zircon grains may be extracted from thin section using a microdrill and
individually annealed in quartz crucibles for later chemical abrasion treatment (e.g., Broderick
2013). It is also possible to mount grains after annealing in epoxy resin, thus preserving the
initially polished surface from the thin section.
Figure 1. Schematic workflow for CA-ID-TIMS U–Pb petrochronology in igneous systems.
C) After a very light repolishing, imaging of the growth structures of zircon using optical,
or panchromatic or wavelength-resolved cathodo-luminescence (CL) on an SEM. This
information is essential for relating zircon textures with U–Pb dates in order to establish a
relative crystallization sequence and produce high quality images to guide in situ work.
D) Carrying out chemical or isotopic in situ analyses, such as trace element, oxygen
or hafnium isotopic analysis by SIMS, or trace element and hafnium isotopic analysis
through LA-ICP-MS. These in situ procedures allow separate and selective analysis of
single growth zones of zircon for their chemical and isotopic composition. We note that
ID-TIMS U–Pb age determinations performed at the University of Geneva laboratory of
zircon grains that were previously sampled for laser ablation analysis did not reveal any
influence from prior laser analysis on the U–Pb system of zircon (e.g., Chelle-Michou et
al. 2014). Effectively, melt rims around laser craters are removed entirely during the partial
dissolution step of the chemical abrasion procedure.
E) After careful extraction of selected, imaged and chemically and isotopically characterized
zircon crystals, partial dissolution is undertaken to remove effects of radiation-related lattice
damage (Mattinson 2005). The grains are then subjected to dissolution after adding a tracer
solution, such as, the (202Pb–) 205Pb–233U–235U EARTHTIME tracer (Condon et al. 2015), see
details of analytical techniques in Schoene and Baxter (2017, this volume) or in Schoene (2014).
Lead and uranium are isolated using anion-exchange chromatographic separation techniques and
their isotopic compositions are measured on a thermal ionization mass spectrometer (TIMS).
The resulting 206Pb / 238U dates are very precise (at ± 0.05% or better) and since many labs use
the same EARTHTIME tracer solutions, the individual ages can be compared between these labs
using only the internal (analytical, random) uncertainties, rather than including tracer calibration
and decay constant uncertainties. These later uncertainties should be included when comparing
ages between non EARTHTIME labs, or when comparing different geochronometers.
F) Chromatography also isolates the trace element and REE fraction that can be analyzed
for elemental concentrations and ratios using a SF-ICP-MS in dry plasma mode with a
desolvating nebulizer, a technique described as TIMS-TEA (Schoene et al. 2010). This allows
determination of chemical information from the same volume of the grain that is dated.
G) From the same trace element solution Hf isotope analyses can be carried out (if
the Hf concentration is sufficiently high) using a MC-ICP-MS in dry plasma mode. At low
Hf concentrations, standard bracketing with matrix-matched natural reference materials is
crucial to obtain accurate results (D’Abzac et al. 2016).
WORKFLOW FOR BADDELEYITE PETROCHRONOLOGY

Baddeleyite petrochronology has been applied very rarely. The traditional chemical
abrasion method applied to zircon has not proven to be effective for baddeleyite (Rioux et al.
2010). Hence the problem of secondary lead loss in baddeleyite, which significantly biases the
accuracy of U–Pb dates, remains an issue. Furthermore, the small size of baddeleyite, typically
< 70 µm, and its monoclinic platy habit typically prohibit a complex workflow involving
extraction from thin section, remounting for trace elements / isotopic analysis and imaging,
and then extraction from the mount again for U–Pb dating. However, a typical workflow for
baddeleyite petrochronology proceeds similarly to zircon:
A) Electron imaging of thin sections to identify baddeleyite and determine its host
or co-crystallizing minerals. This is the best way to determine the baddeleyite content of
the rock and its potential for U–Pb geochronology. From the authors’ personal experience,
samples without baddeleyite in thin section usually do not have this mineral in their mineral
separates. Since extraction from whole rock powder is trickier and more time consuming
for baddeleyite than zircon, checking thin sections for the presence of baddeleyite and
determining the average size of the grains before separation is advised.
B) Coring of the thin section to date the identified baddeleyite crystals ‘in situ’ by SIMS can
be conducted at this point (e.g., Schmitt et al. 2010). Alternatively, traditional zircon separation
techniques, or modifications thereof (Soderlund and Johansson 2002) are employed on whole
rock powder crushed to < 250 µm to extract the baddeleyite crystals. To our knowledge, no
studies have employed extraction of baddeleyite directly from thin section for TIMS dating.
C) Large grains (> 100 µm) can be mounted for further electron imaging by SEM or in situ
analysis by electron microprobe or LA-ICP-MS for Hf isotopes (Ibanez-Mejia et al. 2014).
D) Separated grains are either removed from the grain mount and dissolved for ID-TIMS
U–Pb analysis or dissolved without mounting and imaging if the grains are small. Dissolution
procedures are the same as for zircon, but without chemical abrasion pre-treatment. Uranium
and Pb purification column chemistry washes can also be collected and analyzed for Hf
isotopic composition on an MC-ICP-MS if the grains are large enough to provide a sufficiently
high Hf signal. Special analytical procedures may be required to measure the Hf isotopic
composition of small grains (e.g., D’Abzac et al. 2016).
PETROCHRONOLOGY OF ZIRCON
IN INTERMEDIATE TO FELSIC SYSTEMS
In the following section we discuss the information that can be obtained from careful
chemical and isotopic analysis of zircon. We show that zircon can record information on the
temporal evolution of the chemical composition of a magmatic system and how it changes
as a function of the physical processes such as crystallisation, and magma extraction and
recharge. We first discuss how zircon in intermediate to felsic rocks reflects the evolution of
a calc-alkaline system, then, in the following chapter, we discuss how baddeleyite and zircon
crystallize in evolved residual liquid portions of mafic tholeiitic magmas.
Crystallization of zircon in intermediate–felsic, calc-alkaline melts
First consider crystallization in a simplified magmatic system along a monotonic cooling
path without any thermal disturbance from magma recharge. Temperature decreases over a
period of time, before arriving at the solidus of a given chemical system.
Saturation of a given magmatic liquid with respect to zircon can be expressed as the zircon
saturation temperature (Tsat; Watson and Harrison 1983). The zircon saturation temperature
Tsat is a function of the magma composition, depending on the concentration of Zr and the
concentration ratio of network-forming elements Si + Al over network-modifiers K + Na + Ca
in the melt, expressed as M = [(Na + K + 2Ca)/(Si + Al)], and defined for compositions of
M = 1.0 to 2.1. This range covers granitic to granodioritic composition, but does not include
more mafic melts. The dependence of the zircon saturation temperature on parameters other
than the ones listed above (such as the variation in H2O or changing melt composition through
crystallization of concurrent silicate phases) leads to an uncertainty on Tsat of tens of degrees
at least, see discussion in Boehnke et al. (2013). We also define here the “saturation interval”,
which extends between the saturation temperature (Tsat) and the solidus temperature, i.e., the
temperature range during which we expect zircon to grow.
Alternatively, the crystallization temperature can be approximated by measuring the
concentration of Ti in zircon, since the incorporation of Ti into the lattice of cyrstallizing zircon
has been shown to be dependent on temperature (Watson and Harrison 1983; Ferry and Watson
2007). However, further discussion of this approach by Fu et al. (2008) and Hofmann et al.
(2009) suggested that Ti partitioning into zircon is strongly influenced by factors other than just
aTiO2 and aSiO2. Those include pressure, trace element composition of the parent melt, relative
diffusion rates of Ti, Si and Zr in the melt surrounding the growing zircon, and Zr 4+ + 1 Hf 4+−1
and (Si4++Zr4+)1 (P5++Y3+)−1 exchange vectors (hafnon and xenotime solid solution vectors).
At the saturation temperature (Tsat), zircon nucleates either on pre-existing older zircon
crystals through so-called heterogeneous nucleation, or spontaneously through homogeneous
nucleation from a liquid that is (over)saturated with respect to zircon, which requires a lower
activation energy. Zircon then continues growing through an Ostwald-ripening process.
Heterogeneous nucleation takes place at the surface of older, resorbed zircon grains as well as on
grains that have crystallized earlier in the magmatic system, in crystal mushes that were present
at lower levels of the system (“antecrystic” zircon; Miller et al. 2007). Any magma may contain
recycled zircon crystals from earlier phases of crystallization within the same magmatic system.
In some cases (e.g., in Charlier et al. 2004), what looked like an “inherited core” in a zircon grain
was shown to be only a few 10’s to 100 kas older than the mostly oscillatory zone magmatic rim.
Obviously, zircon crystallizes over a certain period of time, during which the magma is
cooling and the percentage of the total volume of crystals (zircon and others) is increasing.
In intermediate magma, crystallization of large volumes of amphibole and plagioclase occur
together with or prior to zircon precipitation. Zircon may thus occur as inclusions in the major
minerals as well as in matrix minerals that crystallized subsequently from the interstitial liquid
(or it may be fortouitously included in secondary minerals). In particular, the crystallization
of plagioclase will fractionate Zr as an incompatible element into the residual liquid, where
zircon will crystallize. The change of melt composition through fractional crystallization
of major and accessory phases will also alter Tsat. Pressure is not known to significantly
influence zircon saturation in the liquid, but will affect the crystallization sequence of
plagioclase and amphibole, and these in turn influence the trace element partitioning into
zircon. Trace element partitioning into zircon is therefore a petrogenetic tool that allows
reconstruction of the fractional crystallization path of a magmatic system.
In case of monotonic cooling and crystallization of a very small model pluton (1000 °C
melt temperature, 2 km thick cylindrical pluton with a volume of 12 km3, 10 km intrusion
depth; 30 °C / km geothermal gradient; solidus temperature 680 °C) zircon will crystallize over
7000 years, between 5 and 16 ka after intrusion, in a portion of the pluton that resides at an
average temperature, along the solid thin line in Fig. 2A. Note that near the top and sides of the
pluton zircon will crystallize over a shorter interval as cooling is faster than in interior and deep
portions. Therefore, at the margin of the intrusion zircon crystallization is instant (the stippled
curve “minimum temperature” remains below the solidus temperature at all times), whereas
Figure 2. Modelled thermal decay curves and melt fractions for a cooling and crystallizing pluton (1000 °C
melt temperature, 2 km thick cylindrical pluton = 12 km3, 10 km intrusion depth; 30 °C/km geothermal
gradient; solidus temperature 680 °C); the grey horizontal band indicates the range of zircon saturation
temperatures (Tsat) in intermediate to felsic magmatic systems; A: monotonous cooling; B: with one mafic
recharge after 13 ka (melt temperature 1100 °C). Stippled lines account for minimum temperature at the
outermost border of the pluton, and maximum temperature in the core, the solid blue line is the melt
fraction measured on the right axis. The model pluton is entirely solid at 20 and 27 ka, respectively. C)
Theoretical dimensionless distribution of newly formed zircon crystals in a crystallizing magma, assuming
intrusion over 50 ka at a flux of 10−2 km3/a and a final volume of 500 km3 (calculated and redrawn after
Caricchi et al. 2014); D) theoretical, sigmoidal age distribution of zircon growth bands crystallizing from
a monotonously cooling magma. Stages 1–4 discussed in text.
in the hottest part the liquid remains within the zircon saturation interval until ~25 ka after
emplacement. This time span corresponds to the “magmatic residence time” and to the period of
“autocrystic” zircon growth (Miller et al. 2007). The expected distribution of 206Pb / 238U dates
recorded by zircon crystals in a sample of this pluton will describe a unimodal, skewed distribution
pattern, assuming linear nucleation and crystallization rates, as anticipated by the crystallization
model of Caricchi et al. (2014; see Fig. 2C;). A zircon crystallized in such a simplified system
would exhibit undisturbed oscillatory zoning visible in cathodo-luminescence (CL; Fig. 3A). If
we could analyze each growth segment of an oscillatory zoned zircon separately, zones 1 to 4
(as an example) would reflect a systematic decrease of the date of crystallization (Fig. 2D). If
a linear crystallization rate is assumed, the temporal distribution of individual growth zones
would produce a curved, sigmoidal pattern with a majority of dates concentrated between dates
equivalent to zone 2 and zone 3 (Fig. 2D); the curvature of the distribution of individual dates is
a function of the changing proportion of magma within the zircon saturation interval, relative to
magma at temperatures below or above during progressive crystallization.
For whole-grain ID-TIMS dates it is necessary to evaluate how a single zircon grain
reflects these processes because it potentially integrates the entire, or at least part of the growth
history including growth segments 1 to 4. Zircon single-grain dates from plutons or volcanic
units show dispersions that can vary between near-zero (Wotzlaw et al. 2014) and up to several
10 to 100 ka (Leuthold et al. 2012; Schoene et al. 2012; Broderick et al. 2015). Despite the fact
Figure 3. Cathodo-luminescence images of magmatic zircon crystals. A: oscillatory zoning from undis-
turbed growth in a magmatic liquid (leucotonalite, S. Adamello, N. Italy); B: magmatic zircon showing three
growth zones—1) sector zoning, with truncated boundary versus 2) sector zoning, 3) oscillatory zoning
(granite porphyry, Mongolia); C: zircon crystal from same rock sample as B, but devoid of growth zone 2,
evidencing that not all zircon crystals of the same rock sample share the same growth history; D: complexly
zoned zircon crystal with three growth periods—1) combined oscillatory and sector zoning, 2) sector zon-
ing with somewhat subordinate oscillatory zoning/banding, 3) outer zone with faint oscillatory and sector
zoning –growth zones are separated by intermittent resorption events (tonalite, S. Adamello, N. Italy); E:
zircon from the same sample as D, showing very similar characteristics; note the non-planar zoning in zone
2, interpreted by diffusion-reaction processes (dissolution/reprecipitation; see Geisler et al. 2007).
that zircon in a real magmatic system grows at different rates, and it nucleates homogeneously
as well as heterogeneously, the age distribution of single-grain ID-TIMS dates is always
expected to be sigmoidal in shape when a sufficient number (N > 15) of grains is analyzed (e.g.,
Samperton et al. 2015). Other processes, such as magma recharge and inheritance from re-
melting of partially or entirely crystallized precursor systems, will blur the initial distribution
pattern. It has been demonstrated that a large part of the history of a magmatic plumbing system
can be reconstructed from the zircon population of a single hand-sample (e.g., Broderick et al.
2015; Samperton et al. 2015). This fact invokes homogenization processes that transports zircon
between different portions of a crystallizing magma reservoir and / or pass melts of different
composition through the porous framework of a crystal mush. If thermally driven convection is
strong enough, initial small-scale, local and ephemeral equilibrium states get averaged out; by
contrast, in low-volume and rapidly cooling systems such states will be preserved (Broderick et
al. 2015). As a result, a U–Pb date of a zircon is to be considered as a chemical signal and cannot
be readily related to a physical process other than its crystallization.
The more realistic case of a complex magmatic system, with periods of recharge from
mafic and hot magma batches will lead to significant temperature oscillations within the
crystallizing magma, and to repeated intermittent periods of zircon resorption and dissolution
during the growth history.
Consider a crystal-free, water-saturated basaltic melt (50% SiO2) that experiences
monotonic crystallization from 1300 °C (Caricchi and Blundy 2015) and significant olivine,
clinopyroxene, plagioclase and amphibole fractionation at a deep level of the magmatic
plumbing system (the “hot zone”, following Annen 2011). A small volume of zircon forms
during this period of rapid crystallization (rapid decrease of the melt fraction Mf ; interval I
on Fig. 4) in small, marginal volumes of the magmatic reservoir. In most magmatic systems,
zircon would preferentially crystallize during a long period of slow crystallization (interval
II), when the biggest part of the magma reservoir is within the Tsat window. Zircon growth
during this time interval has been termed “equilibrium crystallization” (Watson and Liang
1995; Wang et al. 2011) and is commonly associated with sector-zoned growth (see, e.g.,
Corfu et al. 2003; Kotková et al. 2010; Tapster et al. 2016). Such early growth zones with
sector zoning are displayed in Figs. 3B, C (zone 1) and Figs. 3 D, E (zone 2). Crystallization
is terminated by rapid precipitation of oscillatory zoned zircon when approaching the
solidus (interval III in Fig. 4). However, several additional processes may occur within the
system (stippled lines in Fig. 4): i) andesitic magma may be extracted after interval I from
the deep part of the plumbing system and rises into the middle crust; ii) subsequently, more
batches of increasingly evolved melts are extracted from the crystallizing magma (A, B,
C); iii) crystallizing magma batches (“crystal mushes”) in the middle crust are injected and
thermally rejuvenated by incoming hot (1100 °C) andesitic magma, or, iv) volatile loss shifts
the crystallization curve at a given melt fraction to higher temperature.
Considering a single injection of 1100 °C hot andesite magma 10 ka after the initial
emplacement (Fig. 2B), we observe several effects, compared to the simple case shown in
Fig. 2A: (i) the time the magma takes to arrive at the solidus is prolonged to 20 ka in the part of the
intrusion at average temperature (Fig. 2B); (ii) if mafic melt is mixing with re-juvenated mush,
the change in chemical composition will lead to a change in Tsat; (iii) withdrawal of Zr through
zircon crystallization results in a decrease of Tsat; (iv) the zircon population will record a bimodal
distribution of 206Pb / 238U dates centered around ~8 and ~15 ka, due to crystallization after the
recharge events; (v) the peak temperature during this thermal rejuvenation may exceed Tsat for
the given magma composition and thus lead to a short period of zircon resorption (Fig. 2B).
The models show that recharge and rejuvenation of a magmatic system through hotter,
incoming magma leads to thermal rejuvenation (decrease of the crystal proportion) and to repeated
periods of enhanced zircon crystallization. And indeed, multi-peak zircon age distributions over
200 to 300 ka from high-precision U–Pb dates are a common feature of most magmatic systems,
Figure 4. Fractional crystallization of major minerals and zircon from a tholeiitic basalt as a function of
crystal content and temperature (after Caricchi et al. 2014, 2015). A, B and C denote extraction events of
interstitial melt from the crystallizing mush. Dashed lines trace the melt fraction during extraction and
subsequent crystallization of melt batches. The grey band delineates the temperature interval between
arrival at Tsat (zircon) at 900 °C and the solidus, i.e., the interval during which zircon forms. Decreasing
water content of melts due to volatile loss would shift the melt fraction curve towards higher temperatures
(Caricchi and Blundy 2015).
suggesting that temperature oscillations are a common result of magma recharge events. Figures
5 A to C show three examples of complex magmatic systems (taken from Broderick et al. 2015;
Samperton et al. 2015, and Wotzlaw et al. 2013, respectively). Other examples with high-
precision geochronology can be found in Barboni et al. (2015; see distribution curves in Caricchi
et al. 2016), Schoene et al. (2012), or Wotzlaw et al. (2012). Temperature cycling often involves
periods of resorption, visible in CL or BSE images, during which temperature exceeded Tsat
(zir). Sector and / or oscillating growth zones are truncated along rounded internal surfaces that
are epitaxially overgrown by subsequent euhedral growth zones (Figs. 3B, C truncated zone
1; Figs. 3 D, E, truncated zones 1 and 2). The sequence of zircon growth zones represents a
temporal “stratigraphy” of a thermally decaying magmatic system, where the more internal
zones of a zircon crystal grew during earlier stages of the magmatic system. In case of mushes
that were > 50% crystalline, these were subsequently rejuvenated due to temperature oscillations
during magma recharge. Internal zones that formed during an early stage, at deeper levels of a
magmatic plumbing systems, often show sector zoning or a combination of sector and oscillatory
zoning (Fig. 3) and may be termed “antecrystic” in the sense of Miller et al. (2007). The onset of
magmatic zircon crystallization (t = 0) in the example of the Val Fredda Complex (Fig. 5A) was
set so as to ignore two ca. 150 ka older zircon analyses from the same hand sample. Such zircons
(several 100 ka older) may be derived from cannibalized “proto-plutons” (e.g., Chelle-Michou
et al. 2014; Reubi and Blundy 2008; Wotzlaw et al. 2013) that were reactivated and rejuvenated
during injection of new magma. A distinction between these different types of “antecrysts” may
be made through chemical and isotopic (O, Hf) analysis, see below. Not all zircon crystals of one
hand sample need to show the same “growth stratigraphy”. For example, the crystals shown in
Figure 5. Examples of multi-peak zircon date distributions from magmatic systems of increasing volume (ker-
nel density plots; Vermeesch et al. 2012); A: Val Fredda Complex, S. Adamello intrusion, N. Italy (Broderick
et al. 2015); B: Bergell intrusives, N. Italy (Samperton et al. 2015); C: Fish Canyon Tuff (Wotzlaw et al. 2013).
Figs. 3 B and C are separated from the same granite porphyry sample, but C did not experience
the growth event 2, intermediate in age between 1 and 3, or, alternatively did not preserve it
during the resorption event between 2 and 3.
Age distribution curves as shown in Figure 5 also carry information about the physical
state of a magma reservoir. A small reservoir of magma (tens of km3) that has been periodically
replenished will reflect ephemeral, short episodes during which melt crystallized within 105
year timescales (Fig. 5A). Increasing the volume of a magmatic system to hundreds of km3,
longer periods at higher (>50% vol.) melt proportion can be produced, creating reservoir-wide
melt convection and exchange of crystals (Fig. 5B). These features are shown in the zircon age
distributions as numerous single peaks that get smoothed out, then the principal and secondary
maxima become wider. Zircon that is crystallizing in equilibrium in a larger volume of melt
over longer periods of time has to be considered “autocrystic”, as proposed in Samperton et
al. (2015). In a giant magmatic system (thousands of km3), these homogenizing effects are
even more important and lead to a large age dispersion with a single mode and a large standard
deviation (Fig. 5C). Using the model approach of Caricchi et al. (2014, 2016), mode, median and
standard deviation of populations of zircon U–Pb dates can be tentatively employed to estimate
the integrated magma flux and the volume of a magmatic system, even where it is not known a
priori whether the resulting data derive from a lower or upper crustal magma reservoir.
Distributions of zircon dates over several 100 ka often are not in agreement with the short
solidification times required for small magma volumes. To explain this apparent contradiction,
crystallization of zircon at deeper crustal levels, followed by transport in suspension, in small
magma batches, into the upper crust is typically invoked. Sequential accumulation of small
magma batches into larger plutons at low magma flux rates is a characteristic of many mid to
upper crustal plutons (e.g., Tappa et al. 2011; Schoene et al. 2012; Rosera et al. 2013; Chelle-
Michou et al. 2014; Barboni et al. 2015; Broderick et al. 2015). Incremental accumulation of
small magma batches became popular as the major paradigm for the buildup of large plutons
(after Glazner et al. 2004, and Coleman et al. 2004), as an alternative concept to “big tank”
models that involve magma emplacement through diapirism.
Finally, a word of caution is due when referring to a number of dispersed (antecrystic)
U–Pb single zircon dates grains as “magmatic residence time”, or “period of protracted zircon
growth”: The age dispersion defined by the antecrysts (Miller et al. 2007) is the result of different
processes, involving interaction with previously partly crystallized magmas during melt ascent
and recycling of zircon from rejuvenated mushes. The oldest antecrysts may be many 100 ka
older than the latest crystallization. It is therefore implied that “magmatic residence” includes
in many cases long periods of storage of previously crystallized zircon in cold mushes close to
the solidus and not as suspended crystals in a melt (Claiborne et al. 2010; Klemetti et al. 2014).
What does zircon chemistry tell us about magmatic processes?
Co-precipitating mineral assemblages. Zircon crystallizes together with a series of major
and accessory minerals, which all compete for some critical trace elements. Analyses of these
elements in dated zircon allow the calculation of time-resolved crystallization paths and the
evolution of equilibrium melt compositions, via known partition coefficients. In practical terms,
only a small series of elements in zircon is accessible to quantification through electron microprobe
and laser ablation ICP-MS analyses, such as Zr, Hf, Ti, REE (mostly heavy REE, HREE), Y, Th,
and U. The Th / U ratio may also be obtained from the radiogenic 208Pb / 206Pb ratio from ID-TIMS
age determination. In the following, we inspect the chemical variation of selected elements in
zircon crystallizing in calc-alkaline magmas in continental arc settings, specifically excluding
those melts in which monazite and xenotime are also stable, because no data are available yet for
the chemical and isotopic composition of precisely (ID-TIMS) dated zircon from these.
Partitioning of rare earth elements. Detailed chemical investigation of the major and
accessory minerals in a quartz–plagioclase diorite (PQD) and a tonalite (VFT) of the Val Fredda
Complex (S. Adamello, N. Italy; Broderick 2013) reveals the relative importance of amphibole,
plagioclase, apatite, titanite, allanite and zircon for the partitioning of REE’s between the melt and
the mineral phases (Fig. 6). The influence of zircon is restricted to the HREE Dy to Lu; the main
competitors for this element group are amphibole, apatite and titanite, in decreasing importance
and modal abundance. The Dy concentration as well as the heavy / middle rare earth element
ratio (HREE / MREE) Yb / Dy can be used as a chemical proxy for melt crystallinity, and also
allow a qualitative assessment of the importance of titanite, allanite and apatite crystallization
on the melt via the D(Yb / Dy) of these minerals. Allanite can be ubiquitous in dioritic to tonalitic
melts although usually in low abundance. It preferentially scavenges light rare earth elements
(LREE) and thus is of limited importance for the HREE / MREE ratio. For the examples shown
in Fig. 6, Gd to Lu abundances in plagioclase were below the limit of detection, so this mineral
is of subordinate importance for HREE / MREE fractionation. The REE patterns of zircon in
Fig. 6 reveal the outstanding capacity of zircon to scavenge Ce4+ from the melt, leading to very
high Ce4+/Ce3+ ratios (Hoskin and Schaltegger 2003) that have been shown experimentally to be
related to the oxidation state of the melt (Trail et al. 2012). The oxidation state may reflect the
composition of source rocks and / or the oxidation state of assimilated material at higher levels
of the crust. The Ce anomaly (Ce / Ce*) is commonly defined by linear interpolation between La
and Pr. But the concentration of these two elements in zircon are often below or close to the limit
of detection, so the Ce / Nd ratio is used instead (Chelle-Michou et al. 2014).
The trace element composition of zircon (especially its MREE and HREE concentrations)
may thus be used to model the chemical composition of the melt in equilibrium with zircon,
as follows: (1) select partition coefficients (KD values) for REE in zircon from Rubatto and
Hermann (2007), (2) use the whole-rock composition as the chemical starting composition
in a closed system, (3) use published trace element partition coefficients for the major and
accessory minerals and the modal composition from thin section observations. Subsequently,
the trace element composition of the fractionating bulk assemblage and the evolving melt
may be reconstructed. As a result, the chemical composition of the zircon can be used, in
combination with the other minerals, to reconstruct the magmatic evolution, for example,
yielding a percentage of crystals as a function of time (e.g., Wotzlaw et al. 2013).
Figure 6. Rare earth element mineral budgets for a plagioclase–quartz–diorite (PQD) and a tonalite (VFT;
Val Fredda Complex, Southern Adamello batholith, N. Italy), after Broderick (2013).
However, this model relies on single, published sets of REE partition coefficients
for different accessory minerals and does not allow any quantification of accuracy and
uncertainty. Theoretically, Monte Carlo simulations of the bulk partition coefficients from
the fractionating assemblage (from the published range of experimentally determined
partition coefficients of major and accessory minerals) could help to narrow down the range
of possible KD values by inverting them from the melt composition, back to measured mineral
compositions. In addition, to validate the result, the calculated modal compositions can be
compared to the modal abundance of mineral assemblages determined experimentally from
Piwinskii and Wyllie (1968) and Nandedkar et al. (2014).
The large scatter of Dy concentration plotted versus the Yb / Dy ratio in zircon from
different continental arc plutonic and volcanic rocks (Fig. 7A; all data from dated zircon
grains or populations) suggests that zircon records the degree of fractionation of the LREE
and MREE scavenging minerals (apatite, titanite, zircon) during crystallization. The trajectory
defined by the Fish Canyon Tuff zircon can be recreated using a model with ~0.6 vol%
titanite crystallization from a granodioritic initial magma (Wotzlaw et al. 2013). Zircon
derived from Adamello and Bergell intrusions (N. Italy and S. Switzerland, respectively)
align grossly along the same trend. Two well-dated intrusive units of the Southern Adamello,
the Lago della Vacca Complex (Schoene et al. 2012) and the Val Fredda Complex (Broderick
et al. 2015) show hardly any overlap, pointing to different starting compositions and / or
different crystallization pathways. REE concentrations in zircon from gabbros and dolerite
dykes of the Central Atlantic Magmatic province (Davies et al., in press; Schoene et al.
2010) show significant scatter in Dy (and probably HREE) concentration up to 1800 ppm Dy
(Fig. 7A). These zircon grains are interpreted as growing in small pockets of trace element-
enriched, residual melt that arrived at zircon saturation just above the solidus, after abundant
fractionation of olivine, pyroxene and plagioclase, but none of the REE-scavenging accessory
minerals, thus representing local equilibria at small (possibly cm-to-dm) scale.
Partitioning of Th, U, Zr and Hf. Very similar systematics are displayed by Th / U vs. Zr / Hf
ratios of the same zircon populations (Fig. 7B; for D values see overview in Table 1). Zircon
incorporates U4+ in preference to Th4+ because the smaller ionic radius of U4+ fits better into the
zircon lattice (Shannon 1976), leading to estimates of DTh / DU (zir) of between 0.15 and 0.39
for a temperature range from 700 to 900 °C (Rubatto and Hermann 2007). Th / U (zircon / rock)
ratios show an even larger spread of 0.19–0.65 (Kirkland et al. 2015), but these values are
in conflict with values for zircon DTh and DU derived from the lattice strain model (Blundy
and Wood 2003). Knowing the Th / U of the melt is of paramount importance for obtaining
accurate 206Pb / 238U zircon dates because of the correction for initial 230Th disequilibrium that
leads to a deficit of radiogenic 206Pb in the zircon lattice relative to the abundance of the parent
238
U (Schärer 1984). The DZr / DHf ratio of zircon has been estimated to be around 2, leading
to a higher compatibility of Zr in the zircon lattice than Hf in agreement with the lattice strain
model (Blundy and Wood 2003; no data provided). This value is in disagreement with the
experimental data of Rubatto and Hermann (2007), who propose DZr / DHf values of 0.58–1.37
(Table 1), neither of which are fully supported by the lattice strain model (or the observed
trends in real data, see below). These systematics imply that for Th, U, Zr and Hf, fractionation
is solely due to the respective zircon / melt partition coefficients. Crystallization of zircon
itself will leave a residual liquid with higher Th / U and lower Zr / Hf ratios, thus subsequently
forming zircon will show increasing Th / U and decreasing Zr / Hf (Table 1).
To explain the co-variation of Th / U and Zr / Hf ratios in zircon (Fig. 7B) we need
to infer co-precipitation of a Th-scavenging phase such as apatite, titanite or allanite.
Contemporaneous crystallization of these mineral phases can indeed explain the correlated
Figure 7. Trace element concentrations and ratios measured in radio-isotopically dated zircon. (A) Dy vs. Yb/
Dy; (B) Th/U vs. Zr/Hf. CAMP = Central Atlantic Magmatic Province, dotted lines show blown up version of
plots without CAMP data for clarity. Sources of data: [1] Davies et al., in press.; [2] Broderick et al. 2015; [3]
Wotzlaw et al. 2013; [4] Schoene et al. 2012; [5] Skopelitis 2014; [6] Samperton et al. 2015; [7] Schoene et al.
2010. Schematic fractionation vectors of zircon, titanite and apatite are indicated, based on values in Table 1.
Table 1. Selected D-values for zircon, titanite and apatite.

Mineral Yb/Dy Th/U Zr/Hf
Zircon 1.81 [1] 0.15 [3] ~2 [3]
3.45–4.33 [2] 0.29–0.25 [2] 0.58–1.37 [2]
Titanite 0.12* [4] 2 [4] 5 [4]
Apatite 0.19–0.14* [5] 0.94–12.8 [5] 3.2–4.6 [5]
*
Lu/Gd ratio; [1] Thomas et al. (2002); [2] Rubatto and Hermann (2007), temperature
dependent; [3] Blundy and Wood (2003); [4] Prowatke and Klemme (2005), dacite; [5]
Prowatke and Klemme (2006).
decrease of Th / U and Zr / Hf in Val Fredda and Lago della Vacca units of Southern
Adamello, as well as for the Fish Canyon Tuff, over time (as seen in Fig. 7B). This implies
that the crystallization of accessory minerals is controlling the Th / U in the melt. Titanite
and apatite are ubiquitous minerals in calc-alkaline rocks, and allanite has been reported as
an importantly fractionating phase from the Bergell intrusion (Oberli et al. 2004; Samperton
et al. 2015), for example. Titanite from the Fish Canyon Tuff has a high Th / U ratio of 6.7
(Bachmann et al. 2005), titanite of the Val Fredda Complex between 2.1 and 0.5 (Broderick
2014), whereas the Th / U ratio in titanite from the Lago della Vacca unit varies over several
orders of magnitude between 0.02 and 11.8 as a function of host rock composition and in
part due to subsolidus reactions (Schoene et al. 2012). Experimental data of Prowatke and
Klemme (2005) point to a DTh / DU of ~2 (Table 1) for titanite. The fractionation of apatite
with a DTh / DU of 0.94–12.78 (Prowatke and Klemme 2006) can have an even bigger impact
on the Th / U budget of a crystallizing melt due to its higher modal abundance.
It would, however, be strange if titanite, apatite and allanite rather than zircon were
controlling the Zr / Hf of a melt. Adopting a DZr / DHf ratio in zircon below one (e.g., a value of
0.58 from the 800 °C run of Rubatto and Hermann 2007, instead of the value of ~2 from Blundy
and Wood 2003) would lead to increasing Zr / Hf over time through zircon crystallization
alone, contrary to the trend in Figure 7B. This may be taken as an indication that the DTh / DU
value in Rubatto and Hermann (2007) is not representative of these melt compositions.
Zircon crystallizing in small pockets of highly evolved residual melts in gabbros and dolerite
dykes of the Central Atlantic Magmatic province show highly elevated Th / U up to 4, which may
be explained by the lack of Th-scavenging minerals in the crystallizing assemblage (Fig. 7B).
Simple versus complex magmatic evolution. Temporal trends in the chemical composition of
zircon are monotonic and continuous for the case of closed-system crystallization and cooling. In a
magmatic system with a more complex history, involving periodic magma recharge / rejuvenation
events that result in temperature oscillation and lead to magma mingling / mixing are unlikely to
yield smooth and monotonous chemical trends. The significant scatter in Dy and Yb data from
the central and northern part of the Adamello intrusive suite (Skopelitis 2014), from the Val
Fredda complex of S. Adamello (Broderick et al. 2013) or from the Bergell pluton (Samperton et
al. 2015) are likely to indicate some of these processes (Fig. 7A).
The temporally resolved changes in zircon chemistry in the complex evolution of a large
and homogeneous magma reservoir, such as represented by the Fish Canyon Tuff (Wotzlaw et
al. 2013), provide good insight into the power of zircon petrochronology. In a monotonously
cooling granodioritic magma, chemical fractionation related to crystallization of titanite as
the sole major carrier of HREE beside zircon would lead to an apparent monotonous increase
in Yb / Dy ratios of consecutively crystallizing zircon grains 1 to 4 (Fig. 8). The case of the
Fish Canyon Tuff shows, however, that the Yb / Dy ratios from sequentially crystallizing
zircon grains 1’ to 4’ follow a significantly more complex fractionation trend over 450 ka.
The geochemical model can be inverted to calculate hypothetical melt / crystal proportions,
reflecting a sequence of increasing crystal proportion, up to ~80%, followed by thermal
rejuvenation, re-melting, a decrease of the crystal content, to ~45%, prior to violent eruption
at 28.19 Ma (Fig. 8; see Wotzlaw et al. 2013 for details of the numerical model).
This example impressively shows how high-precision zircon petrochronology can yield
a temporally resolved chemical record of magma evolution, from which parameters such as
equilibrium melt composition, percentage of crystals, modal composition of the crystallizing
mineral assemblage or even magma rheology may be computed, such as in Wotzlaw et al.
(2013), or in Barboni and Schoene (2014).
Figure 8. Temporal evolution of

Yb/Dy as a function of crystal-
lizing titanite in the granodioritic
magma of the Fish Canyon Tuff
reservoir; titanite is the only HREE
scavenging mineral beside zircon.
Sequentially crystallized zircons 1
to 4 would delineate the evolution
of an undisturbed, closed-system
magmatic system; zircons 1’ to 4’
crystallize during complex system
evolution involving thermal reju-
venation between 28.4 and 28.2 Ma
of a nearly solidified mush followed
by eruption at 28.19 Ma (redrawn
after Wotzlaw et al. 2013).
Mingling magma batches of different origin. When dating an igneous rock through
zircon U–Pb geochronology, one underlying assumption is that zircon crystallized in the last
melt batch at or at least close to the level of emplacement. This a priori assumption is wrong
in some cases, as shown by the disagreement between apparent zircon residence times of
several 100 ka and thermal models (as e.g., in Barboni and Schoene 2014, or Chelle-Michou
et al. 2014). As an alternative it has been proposed that zircon crystallizes in compositionally
differing magma batches at different temperatures and at different depths of the magmatic
plumbing system that have been sequentially and rapidly assembled in the upper crust after
most of the zircon formed. Very short timescales of 102–103 years for merging smaller
compositionally different magma reservoirs into large ones, and to subsequent supereruption
have been suggested for Yellowstone-type magmatic systems (e.g., Wotzlaw et al. 2014,
2015). Cores of zircon reflect compositional differences of the smaller-scale reservoirs shown
by O and Hf isotope heterogeneity, whereas rim compositions may be homogeneous and in
equilibrium with the last melt. However, for the Kilgore Tuff magma it has been shown that
these processes were too rapid for the outermost rims of zircon to reach isotopic equilibrium
with the interstitial glass (e.g., Wotzlaw et al. 2014). Mingling of distinct magma batches -
from different sources or variably contaminated by different crustal materials - will be reflected
by heterogeneous oxygen and hafnium isotope composition of zircon, beside excess scatter in
trace element abundances in zircon and other accessories.
The oxygen isotopic composition of zircon has been shown to be an excellent indicator for
the assimilation of hydrothermally altered material, because remelting of upper crustal material
that has been in contact with meteoric water (δ18O = −14‰ ; Taylor and Forester 1979) will
produce zircon with δ18O values significantly below the canonical value of 5.3 ± 0.3 ‰ for mantle
zircon (Valley 2003). Low-δ18O zircon has been discovered in volcanic rocks related to the Snake
River Plain – Yellowstone plume (e.g., Bindeman and Valley 2001; Bindeman 2008; Wotzlaw
et al. 2015), indicating assimilation of hydrothermally altered juvenile caldera material shortly
before eruption. Similarly, late-stage differentiates of the Skaergaard complex apparently digested
previously emplaced, meteorically-altered and 18O-depleted material (Wotzlaw et al. 2012).
Incorporation of old crustal components into a melt may also lead to changes in the δ18O
values, but a far more dramatic effect can be seen in the Hf isotopic composition of zircon from
such hybrid melts. For example, Wotzlaw et al. (2015) demonstrated that 45 vol% assimilation
of Archean crust (εHf = −40 to −60; 4–8 ppm Hf; δ18O = +6 to +9) into juvenile mantle melt
(εHf = +5 to +15; 2–6 ppm Hf; δ18O = +5 to +15) would lead to an increase of 1‰ for the
δ18O value, but a marked decrease of 23 εHf units in analyzed zircon. Oxygen and Hf isotope
compositions of precisely-dated magmatic zircon are therefore perfectly suitable to trace the
timing of the assembly of magma batches from different parts of the magmatic system, with
different crystallization / assimilation histories and / or different sources. Young magmatic
systems provide sufficient temporal resolution to demonstrate the ephemeral character of melt
batches and the high speed at which the mingling between them must be acting, which cannot
be resolved even by high-precision dating techniques. Therefore, zircon in a given portion of
melt can crystallize over short time periods in equilibrium with a small-scale chemical system
and may be different to zircons crystallized in other parts of the same magmatic system (e.g,
Farina et al. 2014, Wotzlaw et al. 2015). During the assembly of melt portions with different
history and / or source, suspended zircon can become physically mingled, which is recorded as
non-systematic trace element and isotopic scatter.
Broderick et al. (2015) documented a case of small and ephemeral magma batches in the
Val Fredda Complex in the Southern Adamello that were emplaced in the upper crust over 104
years’ timescales and represent variably contaminated hybrid melts. Individual zircon crystals
within a sample document variations of 4–6 εHf units, unrelated to the actual high-precision
U–Pb date (Fig. 9B) and pointing to mechanical mixing of zircon from different chemical
environments into the final melt. The co-variation of εHf with Th / U suggests, however, that
contamination by SiO2-rich crustal melts and fractional crystallization of possibly apatite
and titanite were contemporaneous (Fig. 9A). The lacking correlation with time in Fig. 9B
suggests that these chemical characteristics were, however, neither acquired in a coherent
magma batch, nor in the last melt at the moment of emplacement, but in different small melt
portions representing small and ephemeral micro-equilibria prior to final emplacement.
Figure 9. Zircon Hf, Th, U chemical and isotopic characteristics from the Val Fredda complex, southern
Adamello. The lacking temporal correlation between crustal contamination and crystal fractionation pro-
cesses points to zircon growth in distinct and small magma portions.
Incremental assembly of magma batches in the upper crust—wrapping up what we

have learned
One of the essential messages emerging from the field of petrochronology is that magma
is a suspension of crystals in a liquid and zircon is a part of this suspended crystal cargo. The
suspended zircon crystals have formed over several 10 to 100 ka, as directly demonstrated by
SIMS dating of volcanic zircon (e.g., Cooper and Kent 2014; Klemetti and Clynne 2014) or
through zircon inclusions in K-feldspar phenocrysts (e.g., Barboni and Schoene 2014). As shown
by the Fish Canyon Tuff example of Wotzlaw et al. (2013), zircon may even record crystallization
in a precursor mush or “proto-pluton” (Annen 2011) up to 500 ka prior to final eruption,
despite rejuvenation and subsequent eruption of the mush. This is considered a typical case for
“cold” arc type magmatic settings, where a short magma recharge event leads to the eruption
of crystal mushes of granodioritic composition, the so-called “monotonous intermediates” (e.g.,
Bachmann et al. 2007), but the temperature does not significantly exceed Tsat over a long period
of time, i.e., pre-existing zircons are not significantly dissolved. Quantitative thermal models
demonstrate that zircon dissolution is rather inefficient in calc-alkaline magmatic systems with
magma temperatures below 800 °C (Watson 1996; Frazer et al. 2014). Zircon in such a case is
thus recording a significant part of the entire lifespan of a magmatic system, from the initiation
of crystallization in the lower part of the crust, through phases of intermediate storage and
rejuvenation, until final emplacement in the upper crust or eruption.
At the other extreme, almost solidified crystal mushes may be almost entirely rejuvenated
by incoming, very hot basaltic magma in plume-type settings such as shown by the Kilgore
Tuff (Wotzlaw et al. 2014), which leads to substantial re-melting before eruption. The pre-
eruptive zircon memory may be entirely erased because the majority of the magma volume
became thermally rejuvenated at temperatures above Tsat (zir), and additionally by magma that
is undersaturated in Zr. Zircon geochronology will thus solely reveal the very short timescales
of crystallization during rapid cooling from this rejuvenation event until eruption, i.e., the
lifetime of the very last magmatic liquid. Nearly complete extraction of this high-temperature
rhyolitic liquid containing very limited crystal cargo may be seen as the process leading to
high-silica, phenocryst-free rhyolite, which apparently formed over very short lifetimes in hot
systems at high magma fluxes (Tappa et al. 2011; Wotzlaw et al. 2014).
Tilton et al. (1955) and Silver and Deutsch (1963) were among the first geochronologists
to apply U–Pb zircon dating to date the intrusion of granitoid plutons – what has remained
from this initial motivation? With the advances of high-precision CA-ID-TIMS dating, the
uncertainties of individual analyses of single grains or parts of grains have reached 0.05% in
their 206Pb / 238U date. These high precision ages inform us on the timescales of some of the
above discussed processes, but with our improved understanding we find that the ages and
compositions do not necessarily relate to the physical emplacement of the magma in the crust.
Zircon U–Pb dates provide ample information about the evolution of a magmatic plumbing
system from its roots in the deep crust to its eventual emplacement in the upper crust or its
eruption. The information we can obtain from high-precision U–Pb dating includes:
Zircon U–Pb dates very commonly display multi-peak distributions, describing a history of
thermal oscillation over periods of 104 to 105 years. Zircon grains from one hand-sample do not
necessarily share the same history but were assembled into the same melt batch. They represent
snapshots from different parts of a thermally, physically and chemically evolving magma
reservoir. Zircon crystals are moving around in interstitial liquids percolating through crystal
mushes, and / or record the passage of different liquids over time via their growth history. Epitaxial
growth bands form, and these are in equilibrium with liquids of different chemical composition.
The trace element characteristics of zircon from small (tens of km3) magma volumes
may not show any coherent chemical trends, since every grain and every growth zone may
represent a local and ephemeral equilibrium. Any diagram plotting two trace element ratios
against each other, or plotting them against crystallization date, will yield non-systematic or
chaotic trends (e.g., Broderick et al. 2015). If the magma volume is large enough and the heat
content sufficiently high, such a magma reservoir may start to convect and homogenize, which
is evidenced by chemical trends from zircon that exhibit systematic variations (e.g., Schoene
et al. 2012). In case of large-volume, high-flux magma systems, the trace element chemistry of
zircon may represent pluton-wide chemical trends that can be modelled in terms of fractional
crystallization, (e.g., shown for the Bergell intrusion by Samperton et al. 2015, for Zr and Hf,
or for the Fish Canyon Tuff magma chamber by Wotzlaw et al. 2013, for HREE and Th / U).
The mode in the distribution of zircon dates (such as in Figs. 5 A to C) reflects the fact
that most of the zircon crystals form when the major part of the magma volume was within
the zircon saturation interval. However, a small number of crystals will be preserved that
formed earlier in cooler portions, already at lower levels of the crust, another population
of zircon will form in small residual melt pockets when the magma approaches the solidus,
after emplacement at shallow levels in the crust. Temperature oscillations due to multiple
recharge events of hot mafic magma would not alter these systematics, but prolong the
duration of the main peak (or peaks) of zircon crystallization. High-precision U–Pb zircon
dates thus reflect the integrated crystallization history from zircon growth in precursor
mushes to final solidification. In most cases it is therefore impossible to make a direct link
to the physical movement of magma, i.e., to emplacement. Development of time-resolved
thermal-physical models may help to quantify the moment when a magma has reached ca.
50% crystallinity, which would correspond to an approximate minimum age of emplacement.
After emplacement, zircon may continue to crystallize in a stagnant interstitial liquid.
These considerations imply that zircon may not at all record the age relationships we can
deduce in the field, because it has crystallized mostly at deeper crustal levels than the present-
day outcrop level. This may be suspected based on excessive dispersion of zircon dates up to
200-300 ka, in contradiction with thermal models. The U–Pb ages and the incoherent trace
element systematics of zircon suggest injection of small melt batches into the upper crust,
emplacing and crystallizing over no more than 10 ka (e.g., Chelle-Michou et al. 2014; Barboni
et al. 2015), eventually building large plutons by sequential accretion over millions of years.
PETROCHRONOLOGY OF BADDELEYITE AND ZIRCON

IN MAFIC SYSTEMS
Crystallization of zircon in mafic (tholeiitic) melts
In this section, we concentrate our efforts on characterizing zircon and baddeleyite in
tholeiitic rocks since recent geochemical and geochronological work on zircon from mid
ocean ridge (MOR) gabbro and large igneous provinces (LIPs) suggests subtle differences
between zircon in these environments and those found in calc-alkaline magmas. Zircon and
baddeleyite are not uncommon in tholeiitic rocks, especially in the more evolved portions
of coarse grained intrusions or in coarse mesostasis. Zircon is commonly documented from
tonalites and trondhjemites associated with oxide-rich MOR gabbros (Coogan and Hinton
2006; Grimes et al. 2007, 2009; Lissenberg et al. 2009; Rioux et al. 2015a,b, 2016) and less
commonly from thick sills, dykes and flows in LIPs (Svensen et al. 2009; Schoene et al. 2010;
Blackburn et al. 2013; Sell et al. 2014; Burgess and Bowring 2015; Davies et al. in press).
The common occurrence of zircon and baddeleyite in tholeiitic rocks may lead to the
speculation that zircon can crystallize directly from tholeiitic melts at high temperature.
However, application of the zircon saturation thermometry equations, with an extrapolation
to higher M values for mafic melts (e.g., M > 2.5, DeLong and Chatelain 1990) suggests that
zircon will not crystalize from basaltic magmas unless they have Zr concentrations of >~7400
ppm at > 1000 °C (Boehnke et al. 2013), whereas average mid ocean ridge gabbro only has
~20 ppm Zr (Niu and O’Hara 2003). This simplistic application of the zircon saturation
equations is consistent with Ti-in-zircon thermometry in MOR gabbro, although the absence
of quartz or rutile from many of these magmas limits the application of titanium thermometry
in mafic zircon. However, the observation that most silicic rocks have aTiO2 between 0.6–0.9
and aSiO2 > 0.5 (Ferry and Watson 2007) allows us to make estimates for the Ti and Si activities
that result in temperature estimates with uncertainties of ~30 °C. Compilations of mafic Ti-
in-zircon temperatures indicate zircon crystallization at temperatures of ~950–700 °C (Fu et
al. 2008; Grimes et al. 2009; Jöns et al. 2009; Rioux et al. 2015a, 2016). These temperatures
can be compared to models of the liquid lines of descent for tholeiitic magmas which
show that concentrations of Ti and Fe increase in the liquid as the basaltic liquid cools to
~1080–1120 °C and crystallizes until ~85% fractional crystallization, where the saturation
point for Fe-Ti oxides is reached (depending on ƒO2 and H2O). From this point on in the
crystallization sequence, SiO2 increases in the melt along with other incompatible elements,
causing zircon and other accessory minerals (apatite, quartz) to saturate (Niu et al. 2002).
The point at which zircon crystallizes in these late-stage, evolved melts determines to what
extent zircon chemistry can be used to trace petrogenetic processes. Zircon crystallization
can also be modeled by partial re-melting of hydrothermally altered gabbro rather than by a
fractional crystallization process directly from a tholeiitic melt (Koepke et al. 2007), but the
geochemistry of zircon produced either way should be similar.
Chemical characteristics of zircon in mafic magmas
In a general sense, the trace element geochemistry of zircon crystallized from tholeiitic
magmas is quite distinctive compared with zircon in calc-alkaline rocks. Grimes et al. (2007)
showed that, because U and Yb have similar partition coefficients between zircon and melt
(226 ± 64 for Yb and 157 ± 51 for U, experiments at 850 °C, Rubatto and Hermann 2007), their
ratio in zircon should reflect the ratio in the melt they crystallize from. Arc magmas have high
U / Yb ratios > 1 due to high concentrations of U, whereas MORB has low U and is typically
more enriched in Yb producing U / Yb < 1. When these ratios are plotted vs. HREE or Hf
concentrations, tholeiitic zircon can be effectively distinguished from zircon grown in other
magmas. However, classification of zircon (especially tholeiitic zircon) on the basis of trace
elements is hampered by the fact that variation within a single sample can match the variation
shown by all samples in a group. Trace element variations within MOR gabbroic zircon can
reveal fractional crystallization processes, similar to zircon from calc-alkaline settings, with
negative correlations between Ti and Hf concentrations, which likely reflect incompatible
behavior of Hf in the melt, and that zircon is the main Hf host (Fig. 10A; Grimes et al. 2009;
Rioux et al. 2015a,b). At lower temperatures (<~750 °C), and more elevated Hf concentrations
(~1.7 wt% Hf), the relationship between Hf and Ti appears to break, with increases in Hf
occurring at constant Ti, which can be interpreted as reflecting eutectic crystallization (Grimes
et al. 2009). Within the same sample set, most REE display more scattered patterns compared
with Hf when plotted against Ti (or temperature), suggesting that saturation of other accessory
minerals (apatite, titanite, etc.) plays a role in controlling the budget of these elements in
fractionated tholeiitic melts. Other inferences may be made based on Yb / Dy ratios, which
in zircon from acidic rocks has been shown to record titanite + zircon crystallization (see
above, Fig. 7A). In tholeiitic zircon, Yb / Dy appears to be extremely consistent over a range
of temperatures, suggesting that the partition coefficients of these elements in tholeiitic melts
may be close to unity, which is quite different from what is expected (D(Yb / Dy) ≈ 1.81 from
Thomas et al. 2002; or 3.45–4.3 from Rubatto and Hermann, 2007; see Table 1). After Ti
concentrations drop below ~10 ppm (~725 °C), the consistent behavior of Yb / Dy stops,
possibly indicating that a LREE-MREE scavenging phase starts to sequester Dy, increasing the
Yb / Dy ratios of zircon (Fig. 10B). This transition to non-zircon controlled HREE partitioning
occurs simultaneously between Yb / Dy and the possibly eutectic growth observed with Hf.
One interesting difference in tholeiitic zircon geochemistry between MOR and LIP magmas
is shown in Fig. 10C where the Th / U of LIP zircon is consistently elevated relative to MOR
zircon. This effect is easily modeled in LIP samples since zircon and baddeleyite are the only
minerals that strongly partition U and Th into their crystal structure, whereas evolved melts
that crystallize zircon from MOR magmas may contain apatite, titanite, and other U and Th
scavaging trace phases. Both zircon and baddeleyite have DTh / U mineral–melt below 1, ~0.2 for
zircon (Rubatto and Hermann, 2007), and < 0.18 for baddeleyite (Klemme and Meyer 2003), but
since almost all analyses of baddeleyite have Th / U ~0.01, DTh / U is likely to be very low. Th / U in
a melt should increase with fractional crystallization of zircon or baddeleyite, and consequently,
zircon or baddeleyite grown at later stages should have elevated Th / U, assuming constant
Figure 10. Zircon trace element ratios vs Ti-in-zircon temperatures for zircon from tholeiitic melts. A)
Titanium temperature vs Hf (wt %), B) Titanium Temperature vs Yb/Dy, C) Th/U vs Yb/Dy. MOR =
Mid Ocean Ridge, data from Grimes et al. 2009; Rioux et al. 2015a,b, 2016, CAMP = Central Atlantic
Magmatic Province, data from Davies et al. in press and Schoene et al. (2010). The arrows denote
general trends in the data.
partitioning (Barboni and Schoene 2014; Wotzlaw et al. 2014). However, the high Th / U ratios
in zircon shown in the LIP dataset from the CAMP, suggest Th / U in the magma of up to 30. Such
high Th / U values for whole rocks have been found for some granites (see Kirkland et al. 2015)
but not for mafic rocks, which are typically <~6. To create these extreme Th / U magmatic values,
zircon and / or baddeleyite need to make up a large proportion of the fractionally crystallizing
assemblage, up to ~20%, however simple mass balance calculations indicate that this can not be
the case. For example, the Zr concentrations of CAMP whole rocks are < 160 ppm (Marzoli et
al. 2014), and crystallizing zircon and baddeleyite in large proportions (20% of the crystallizing
phases) would reduce the Zr concentration of the melt to ~60 ppb after only 4.5% fractional
crystallization while only increasing the Th / U of the melt by ~2, nowhere near the required
Th / U of 30 even if the starting Th / U of the melt is 6 (assuming partition coefficients of Rubatto
and Herman 2007; and Klemme and Meyer 2003). Such extreme and apparently contradictory
compositions required for these melts indicate an alternative explanation for the high Th / U in
the LIP zircon such as the partition coefficient for Th / U may be closer to 1 than 0.2. Highly
fractionated mafic melts have not specifically been investigated in zircon–melt, or baddeleyite–
melt trace element partitioning experiments, so this possibility remains currently speculative.
Tholeiitic zircon petrochronology has been applied much less than petrochronology in
calc-alkaline rocks, and since zircon in tholeiitic rocks are typically crystallizing in late-
stage evolved, to extremely fractionated melts, the recordable temporal history is likely to
be short. Few geochronological studies from tholeiitic zircon have had the resolution to
record protracted crystallization histories (Grimes et al. 2008; Rioux et al. 2012, 2015a,b),
and until very recently (Rioux et al. 2016) high-resolution geochronological studies
have not also collected geochemical information on the dated grains to directly compare
chemistry to age. This comparison would be especially interesting in LIP magmas due
to the apparently more frequent occurrence of baddeleyite than in MOR gabbros. Since
baddeleyite petrochronology so far is underdeveloped, it is not clear if the difference in
baddeleyite abundance for MOR and LIP reflects merely a sampling effect, as more studies
have targeted baddeleyite in LIP samples than in MOR, or if it is petrologically controlled.
Baddeleyite geochronology
Part of the problem with high precision U–Pb geochronology in mafic rocks is the limited
abundance of zircon. But when it is extracted, chemical abrasion techniques can be applied
to ensure that the effects of Pb loss on the grain are removed (Mattinson 2005), resulting
in reliable and accurate ages. For baddeleyite, there are no currently accepted techniques
to remove the effect of Pb loss (see Rioux et al. 2010), and therefore geochronology with
this mineral results in scattered ages and difficult interpretations. Another datable mineral in
mafic rocks, zirconolite CaZrTiO2O7 presents similar analytical drawbacks (Wu et al. 2010).
There have been numerous studies highlighting the power of baddeleyite for dating different
rock types. Recent examples and topics include meteorite impact events (Moser et al. 2013,
Darling et al. 2016), alkaline magmas (Heaman and LeCheminant 2000; Heaman et al.
2009; Ibáñez-Mejía et al. 2014), diabase dykes and gabbros (Olsson et al. 2011; Davies and
Heaman 2014; Ernst et al. 2016), (anorthosites, Wall and Scoates 2016; Wall et al. 2016), and
vesicle-rich segregations in silica undersaturated lavas (Wu et al. 2015). Here we concentrate
on some of the unresolved issues with U–Pb geochronology in baddeleyite.
U–Pb discordance in baddeleyite. Theoretically, in a cooling mafic melt with high Zr
concentrations, baddeleyite saturation may be reached before silica saturation and zircon
crystallization. Such relationships have been identified qualitatively in thin section (Fig. 11A) and
quantitatively though geochronology, where baddeleyite is shown to crystallize 0.16 ± 0.07 Ma
before zircon in a gabbroic intrusion in the Canary Islands, followed by a period of baddeleyite
and zircon co-precipitation (Allibon et al. 2011). In the Allibon study, zircon was chemically
abraded to remove Pb loss, whereas if baddeleyite and zircon are dated from the same sample
without any chemical abrasion applied to the zircon, baddeleyite ages are usually more
concordant than zircon (Davies and Heaman 2014), suggesting that baddeleyite and zircon lose
Pb in different ways. In detailed, high-resolution studies involving zircon chemical abrasion
ages, baddeleyite frequently records ages younger than zircon from the same rock. In a mafic
dyke from the Moroccan CAMP, baddeleyite ages are up to 1.5 ± 0.4 Ma younger than the zircon,
despite petrographic evidence that they crystallized before (Fig. 11B; Davies et al. in press).
Discordance in baddeleyite is often attributed to mixtures between igneous (and presumably
concordant) baddeleyite, with later hydrothermal or metamorphic zircon, producing a linear
array between the igneous baddeleyite age, and the later zircon age (Heaman and LeCheminant
1993). In the CAMP example (Fig. 11), none of the dated baddeleyite grains contained zircon
overgrowths or alteration suggesting that the discordance is attributable to baddeleyite itself.
Attempts to air-abrade baddeleyite before TIMS U–Pb analysis to remove zircon rims have
resulted in reduced discordance, but it has not been completely (e.g., Heaman and LeCheminant
2000; Wall et al. 2016). Part of the reason for only partial removal of discordance through air
abrasion is that alteration of baddeleyite to produce zircon occurs both at the rim of the crystals
and internally, where it will remain unaffected by air abrasion (Rioux et al. 2010; Wall et al.
Figure 11. Relationship between baddeleyite and zircon in a CAMP basaltic dyke. A) BSE image from a
thin section showing baddeleyite with zircon overgrowth, indicating that baddeleyite crystallized before
zircon, B) Concordia diagram for the same sample, chemically abraded zircon analyses are shaded red,
baddeleyite data without chemical abrasion are shaded grey. Note that the zircon analyses overlap within
uncertainty and the baddeleyite ages are all younger, but still concordant. Data from Davies et al. in press.
2016). Also the small size of typical crystals < 100 μm, and their fragility means that air abrasion
is not suitable in cases where only a few grains are extracted from a sample. Multi-step digestion
has been demonstrated to isolate baddeleyite from secondary zircon where each component,
i.e., baddeleyite and zircon, can be dated using high precision TIMS techniques independently
(Rioux et al. 2010). This technique results in more concordant results than air abrasion alone,
however does not deal with discordance issues affecting baddeleyite itself.
On top of the Pb loss problems, baddeleyite is known to record reverse discordance, i.e.,
the U–Pb uncertainty ellipse plots to the left of the Concordia curve (Schoene and Bowring
2006; Nilsson et al. 2010; Söderlund et al. 2010). The cause of reverse discordance is enigmatic,
although reverse discordance data commonly display linear arrays with normally discordant
analyses, suggesting a common origin both physically and in time, for example U loss, Pb gain
or isotopic fractionation of Pb during Pb loss. Baddeleyite can be forced to produce reverse
discordance in a lab setting through partial HCl dissolution between 125 and 210 °C, similar to
zircon (Mattinson 2005; Rioux et al. 2010), but again it is not clear what the exact mechanism
is that leads to reverse discordance. With increased dissolution, the reverse Pb loss disappears,
suggesting that it is associated with more disturbed areas of the crystal; however, ‘normal’ Pb loss
then becomes dominant (Rioux et al. 2010), i.e., analyses fall to the right of the Concordia line.
Part of the problem in understanding discordance in baddeleyite is that there are no
available experimental diffusion data for Pb in baddeleyite. This area is receiving some
attention (Bloch et al. 2014) but the observation that it preserves concordant, or slightly
discordant ages though granulite grade metamorphism suggests that the closure temperature
for Pb diffusion is high (Soderlund et al. 2008; Beckman et al. 2014), hence thermally
activated volume diffusion may not be a viable mechanism for Pb loss. Lead disturbance
was identified during laser ablation analysis of baddeleyite from Duluth gabbro, where
increased Pb counts were identified next to high U zones (Ibáñez-Mejía et al. 2014). These
anomalous Pb enrichments were interpreted to reflect intracrystalline Pb* migration from
the surrounding U rich zones, possibly due to effects of radiation damage. Baddeleyite
responds to radiation damage differently than zircon and maintains its crystallinity through
high levels of radiation (see review by Trachenko 2004). However, ion bombardment
experiments on monoclinic zirconia (a.k.a. baddeleyite) have shown that baddeleyite
can undergo phase transformations due to the radiation damage: baddeleyite becomes
tetragonal at high ion radiation fluxes, and this phase survives at atmospheric pressures
and temperatures, unlike tetragonal zirconia produced thermally (Phillippi and Mazdiyasni
1971; Sickafus et al. 1999; Simeone et al. 2006). The phase change is a recrystallization of
polymerized domains created during the collision cascade of the incoming ion (in terms of
geochronological processes this would correspond to damage associated with alpha recoil),
and in a natural baddeleyite crystal, recrystallization would likely cause Pb migration and
fast pathway diffusion. These phase changes have not been identified in natural baddeleyite
crystals, but Raman spectra from the Phalaborwa baddeleyite do suggest the presence of
a tetragonal phase (Fig. 12). Phalaborwa baddeleyite (2060 Ma) is known to record small
amounts of Pb loss (Heaman 2009; Rioux et al. 2010), and the identification of a tetragonal
band in the Raman spectra leads to the speculation that Pb loss may reflect Pb mobility
and diffusion induced by a phase transformation (Davies 2014). Further study is required
to show whether the specific Raman spectra are due to the presence of a tetragonal phase,
since Raman spectral bands can be created in a number of ways (Lenz et al. 2015). Once the
mechanism for Pb loss in baddeleyite is identified, targeted methods can be developed to
reduce its effect, which would increase the potential for baddeleyite petrochronology.
Figure 12. Baddeleyite Raman spectra. Spectrum for synthetic monoclinic baddeleyite (black line) com-
pared with spectra for Phalaborwa natural baddeleyite (grey lines) which has an age of 2060 Ma (Heaman
2009). The location of the dominant tetragonal band at 257 cm−1 is highlighted. Note the appearance of a
signal in this location in the Phalaborwa baddeleyite, but not the synthetic baddeleyite.
OUTLOOK
Zircon (and to a modest extent baddeleyite) petrochronology has revealed a great deal
about magmatic processes through combining techniques such as geochronology, petrology
and geochemistry. The field is, however, still rapidly moving ahead, through refining of
currently used techniques, more extensive use of chemical and structural mapping of
mineral grains prior to isotopic analysis, or through the development of novel tools. For
example, Li diffuses much faster in zircon than most other detectable cations (Cherniak and
Watson 2010). Recent experimental results suggest that relaxation of Li profiles from sector
or oscillatory zones may be used as a geospeedometer and potentially a peak temperature
indicator (Trail et al. 2016). The combination of Li profiles with Ti temperatures may provide
interesting new information either on the thermal history of plutons after zircon saturation,
or on zircon recycling events, depending on the temperatures involved. Another avenue for
future work, so far under-utilized, is petrographically controlled sampling—context matters:
whereas zircon crystallizing from the mesostasis between minerals in late-crystallizing melts
should have a chemistry and age that reflects this, zircon inclusions in pyroxene, or in cores
of plagioclase may have different age / chemical information. Even in a single pluton hand
sample, it is clear that many zircons do not record the age of pluton emplacement but some
protracted pre-emplacement history, but it is not so clear to what extent the other minerals in
plutonic rocks record corresponding information. Targeting zircon inclusions with different
minerals could be used to add constraints on the temporal history of other phases.
A problem in urgent need of development relates to element partition coefficients between
melt of different compositions and zircon and baddeleyite. The usefulness of chemical
information in zircon and baddeleyite partially depends on such data, and as highlighted
throughout this chapter, inconsistencies between different studies reduce the ability to extract
robust petrochronological histories.
Reducing sample size has been a target of the TIMS U–Pb geochronology community
for a long time and, with further reductions in laboratory blank below the 0.2 pg level,
analyte sizes will continue to shrink allowing access to finer and finer spatial resolution while
maintaining temporal precision. Obvious targets are zircons with multiple overgrowths that
are too close in age for ion probe analysis to resolve (e.g., Reimink et al. 2016). However,
recent high-resolution ion imaging studies have shown that Pb* distribution is not correlated
with U concentrations, instead it is concentrated in areas of the crystal (Kusiak et al. 2013,
2015). This may cause problems as sample sizes decrease, although, on the other hand it
may open new avenues for future developments.
ACKNOWLEDGMENTS
Members of the Earth Science department at University of Geneva have contributed
ideas and data to this review, especially C.A. Broderick, F. Martenot, G. Simpson, and
L. Caricchi. The support from the Swiss National Science Foundation for the research of
both authors and the isotope laboratory at University of Geneva is highly appreciated. An
early version of the manuscript benefitted from comments of M. Ovtcharova and F. Farina
(Geneva), which are both acknowledged. The reviews of Drew Coleman, Matt Rioux and
Blair Schoene helped to improve the text further, as well as the comments and careful
editorial handling by Martin Engi. No research is possible without skilled and competent
technical help at all stages of work—a special “Thank you!” therefore goes to the technical
personnel, who work in the background and make the work possible.
REFERENCES
Allibon J, Ovtcharova M, Bussy F, Cosca M, Schaltegger U, Bussien D, Lewin E (2011) Lifetime of an ocean
island volcano feeder zone: constraints from U–Pb dating on coexisting zircon and baddeleyite, and 40Ar/39Ar
age determinations, Fuerteventura, Canary Islands. Can J Earth Sci 48:567–592, doi.org/10.1139/E10-032
Annen C (2011) Implications of incremental emplacement of magma bodies for magma differentiation, thermal
aureole dimensions and plutonism–volcanism relationships. Tectonophysics 500:3–10, doi.org/10.1016/j.
tecto.2009.04.010
Bachmann O, Dungan MA, Bussy F (2005) Insights into shallow magmatic processes in large silicic magma
bodies: the trace element record in the Fish Canyon magma body, Colorado. Contrib Mineral Petrol 149:338–
349, doi.org/10.1007/s00410-005-0653-z
Bachmann O, Miller C, de Silva S (2007) The volcanic–plutonic connection as a stage for understanding crustal
magmatism. J Volcanol Geotherm Res 167:1–23
Barboni M, Schoene B (2014) Short eruption window revealed by absolute crystal growth rates in a granitic
magma. Nat Geosci 7:524–528, doi.org/10.1038/ngeo2185
Barboni M, Annen C, Schoene B (2015) Evaluating the construction and evolution of upper crustal magma
reservoirs with coupled U/Pb zircon geochronology and thermal modeling: A case study from the Mt.
Capanne pluton (Elba, Italy). Earth Planet Sci Lett 432:436–448, doi.org/10.1016/j.epsl.2015.09.043
Beckman V, Möller C, Söderlund U, Corfu F, Pallon J, Chamberlain K (2014) Metamorphic zircon formation at the
transition from gabbro to eclogite in Trollheimen–Surnadalen, Norwegian Caledonides. In: New Perspectives
on the Caledonides of Scandinavia and Related Areas. Corfu F, Gasser D, Chew D (eds) Geol Soc Lon Spec
Publ 390
Bindeman I (2008) Oxygen isotopes in mantle and crustal magmas as revealed by single crystal analysis. Rev
Mineral Geochem 69:445–478, doi.org/10.1016/j.epsl.2015.09.043
Bindeman IN, Valley JW (2001) Low-delta O-18 rhyolites from Yellowstone: Magmatic evolution based on
analyses of zircons and individual phenocrysts. J Petrol 42:1491–1517
Blackburn TJ, Olsen PE, Bowring SA, Mclean NM, Kent DV, Puffer J, McHone G, Rasbury TE, Et-Touhami M
(2013) Zircon U–Pb geochronology links the End-Triassic extinction with the Central Atlantic magmatic
province. Science 340:941–945, doi.org/10.1126/science.1234204
Bloch M, Watkins J, Van Orman J (2014) Diffusion kinetics of geochronologically relevant species in baddeleyite.
EOS Trans, Am Geophys Union #V43B-4827
Blundy J, Wood B (2003) Mineral–melt partitioning of uranium, thorium and their daughters. Rev Mineral
Geochem 52:59–123
351:324–334, doi.org/10.1016/j.chemgeo.2013.05.028
Broderick C (2013) Timescales and petrologic processes during incremental pluton assembly: a case study from
the Val Fredda Complex, Adamello Batholith N Italy. PhD thesis nr. 4612, University of Geneva, Terre &
Environnement 125, 169 pp
Bowring SA, Erwin D, Parrish RR, Renne P (2005) EARTHTIME: A community-based effort towards high-
precision calibration of earth history. Geochim Cosmochim Acta 69:A316
Broderick C, Wotzlaw JF, Frick DA, Gerdes A, Ulianov A, Günther D, Schaltegger U (2015) Linking the thermal
evolution and emplacement history of an upper-crustal pluton to its lower-crustal roots using zircon
geochronology and geochemistry (southern Adamello batholith, N. Italy). Contrib Mineral Petrol 170:28,
doi.org/10.1007/s00410-015-1184-x
Burgess SD, Bowring SA (2015) High-precision geochronology confirms voluminous magmatism before, during,
and after Earth’s most severe extinction. Sci Adv 1:e1500470–e1500470, doi.org/10.1126/sciadv.1500470
Caricchi L, Blundy J (2015) The temporal evolution of chemical and physical properties of magmatic systems.
Geol Soc London, Spec Publ 422:1–15, doi.org/10.1144/SP422.11
Caricchi L, Simpson G, Schaltegger U (2014) Zircons reveal magma fluxes in the Earth’s crust. Nature 511:457–
461, doi.org/10.1038/nature13532
Caricchi L, Simpson G, Schaltegger U (2016) Estimates of volume and magma input in crustal magmatic systems
from zircon geochronology: the effect of modeling assumptions and system variables. Frontiers Earth Sci
4:409, doi.org/10.1016/j.epsl.2015.02.035
Charlier BLA, Wilson C, Lowenstern J, Blake S, van Calsteren P, Davidson, J (2004) Magma generation at a large,
hyperactive silicic volcano (Taupo, New Zealand) revealed by U–Th and U–Pb systematics in zircons. J
Petrol 46:3–32, http://doi.org/10.1093/petrology/egh060
Chelle-Michou C, Chiaradia M, Ovtcharova M, Ulianov A, Wotzlaw JF (2014) Zircon petrochronology reveals
the temporal link between porphyry systems and the magmatic evolution of their hidden plutonic roots (the
Eocene Coroccohuayco deposit, Peru). Lithos 198–199:129–140, doi.org/10.1016/j.lithos.2014.03.017
Cherniak DJ, Watson EB (2010) Li diffusion in zircon. Contrib Mineral Petrol 160:383–390
Claiborne LL, Miller CF, Flanagan DM, Clynne MA, Wooden JL (2010) Zircon reveals protracted magma storage
and recycling beneath Mount St. Helens. Geology 38:1011–1014, doi.org/10.1130/G31285.1
Coleman D, Gray W, Glazner A (2004) Rethinking the emplacement and evolution of zoned plutons: Geochronologic
evidence for incremental assembly of the Tuolumne Intrusive Suite, California. Geology 32:433–436
Condon D, Schoene B, Mclean NM, Bowring SA, Parrish RR (2015) Metrology and traceability of U–Pb isotope
dilution geochronology (EARTHTIME Tracer Calibration Part I). Geochim Cosmochim Acta 164:464–480,
doi.org/10.1016/j.gca.2015.05.026
Coogan LA, Hinton RW (2006) Do the trace element compositions of detrital zircons require Hadean continental
crust? Geology 34:633–636
Cooper K, Kent A (2014) Rapid remobilization of magmatic crystals kept in cold storage. Nature 506:480–483.
http://doi.org/10.1038/nature12991
Cottle JM, Kylander-Clark AR, Vrijmoed JC (2012) U–Th/Pb geochronology of detrital zircon and monazite by
single shot laser ablation inductively coupled plasma mass spectrometry (SS-LA-ICPMS). Chem Geol 332–
333: 136–147, doi.org/10.1016/j.chemgeo.2012.09.035
D’Abzac F-X, Davies JHFL, Wotzlaw JF, Schaltegger U (2016) Hf isotope analysis of small zircon and baddeleyite
grains by conventional Multi Collector-Inductively Coupled Plasma-Mass Spectrometry. Chem Geol 433:12–
23, doi.org/10.1016/j.chemgeo.2016.03.025
Darling JR, Moser DE, Barker IR, Tait KT, Chamberlain KR, Schmitt AK, Hyde BC (2016) Variable microstructural
response of baddeleyite to shock metamorphism in young basaltic shergottite NWA 5298 and improved U–Pb
dating of Solar Sytem events. Earth Planet Sci Lett 444:1–12
Davies JHFL (2014) Insights into the origin of the Scourie Dykes from geochemistry and geochronology. PhD
Dissertation. University of Alberta, Alberta, Canada
Davies JHFL, Heaman LM (2014) New U–Pb baddeleyite and zircon ages for the Scourie dyke swarm: A long-
lived large igneous province with implications for the Paleoproterozoic evolution of NW Scotland. Precamb
Res 249:180–198
Davies JHFL, Marzoli A, Bertrand H, Youbi N, Schaltegger U (2017) End-Triassic mass extinction started by
intrusive CAMP activity. Nat Comm (in press)
DeLong SE, Chatelain C (1990) Trace element constraints on accessory-phase saturation in evolved MORB
magma. Earth Planet Sci Lett 101:206–215
Ernst RA, Hamilton MA, Söderlund U, Hanes JA, Gladkochub DP, Okrugin AV, Kolotilina T, Mekhonoskin AS,
Bleeker W, LeCheminant AN, Buchan KL, Chamberlain KR, Didenko AN (2016) Long-lived connection
between southern Siberia and northern Laurentia in the Proterozoica. Nat Geosci 9:464–469, doi.org/10.1038/
NGEO2700
Farina F, Stevens G, Gerdes A, Frei D (2014) Small-scale Hf isotopic variability in the Peninsula pluton (South
Africa): the processes that control inheritance of source 176Hf/177Hf diversity in S-type granites. Contrib
Mineral Petrol 168:1065, doi:10.1007/s00410-014-1065-8
Ferry J, Watson E (2007) New thermodynamic models and revised calibrations for the Ti-in-zircon and Zr-in-rutile
thermometers. Contrib Mineral Petrol 154:429–437
Fisher CM, Vervoort JD, DuFrane SA (2014) Accurate Hf isotope determinations of complex zircons using the “laser
ablation split stream” method. Geochem Geophys Geosystem 15:121–139, doi.org/10.1002/2013GC004962
Frazer RE, Coleman DS, Mills RD (2014) Zircon U–Pb geochronology of the Mount Givens Granodiorite:
Implications for the genesis of large volumes of eruptible magma. J Geophys Res-Solid Earth 119:2907–
2924, doi.org/10.1002/2013 JB010716
Fu B, Page FZ, Cavosie AJ, Fournelle J, Kita NT, Lackey JS, Wilde SA, Valley JW (2008) Ti-in-zircon thermometry:
applications and limitations. Contrib Mineral Petrol 156:197–215
Geisler T, Schaltegger U, Tomaschek F (2007) Re-equilibration of zircon in aqueous fluids and melts. Elements
3:43–50
Glazner A, Bartley J, Coleman D, Gray W, Taylor R (2004) Are plutons assembled over millions of
years by amalgamation from small magma chambers? GSA Today 14:4–11, doi: 10.1130/1052–
5173(2004)014 < 0004:APAOMO > 2.0.CO;2
Grimes CB, John BE, Kelemen PB, Mazdab FK, Wooden JL, Cheadle MJ, Hanghøj K, Schwartz JJ (2007) Trace
element chemistry of zircons from oceanic crust: A method for distinguishing detrital zircon provenance.
Geology 35:643–646
Grimes CB, John BE, Cheadle MJ, Wooden JL (2008) Protracted construction of gabbroic crust at a slow spreading
ridge: Constraints from 206Pb/238U zircon ages from Atlantis Massif and IODP Hole U1309D (30°N, MAR).
Geochem Geophys Geosyst 9, doi:10.1029/2008GC002063
Grimes CB, John BE, Cheadle MJ, Mazdab FK, Wooden JL, Swapp S, Schwartz JJ (2009) On the occurrence, trace
element geochemistry, and crystallization history of zircon from in situ ocean lithosphere. Contrib Mineral
Petrol 158:757–783, doi.org/10.1007/s00410-009-0409-2
Heaman LM, LeCheminant AN (1993) Paragenesis and U–Pb systematics of baddeleyite (ZrO2). Chem Geol
110:95–126
Heaman LM, LeCheminant AN (2000) Anomalous U–Pb systematics in mantle-derived baddeleyite xenocrysts
from Ile Bizard: evidence for high temperature radon diffusion? Chem Geol 172:77–93
Heaman LM (2009) The application of U–Pb geochronology to mafic, ultramafic and alkaline rocks: An evaluation
of three mineral standards. Chem Geol 261:43–52
Ibanez-Meija M, Gehrels, GE, Ruiz J, Vervoort JD, Eddy MP, Li C (2014) Small-volume baddeleyite (ZrO2) U–Pb
geochronology and Lu–Hf isotope geochemistry by LA-ICP-MS Techniques and applications. Chem Geol
348:149–167
Jöns N, Bach W, Schroeder T (2009) Formation and alteration of plagiogranites in an ultramafic-hosted detachment
fault at the Mid-Atlantic Ridge (ODP Leg 209). Contrib Mineral Petrol. 157:625–639 doi: 10.1007/s00410-
008-0357-2
Kemp AIS, Wilde SA, Hawkesworth CJ, Coath CD, Nemchin A, Pidgeon RT, Vervoort JD, DuFrane AS (2010)
Hadean crustal evolution revisited: New constraints from Pb–Hf isotope systematics of the Jack Hills zircons.
Earth Planet Sci Lett 296:45–56, doi.org/10.1016/j.epsl.2010.04.043
Kirkland C L, Smithies RH, Taylor RJM, Evans N, McDonald B (2015) Zircon Th/U ratios in magmatic environs.
Lithos 212–215:397–414, doi.org/10.1016/j.lithos.2014.11.021
Klemetti EW, Clynne MA (2014) Localized rejuvenation of a crystal mush recorded in zircon temporal and
compositional variation at the Lassen volcanic center, Northern California. PLoS ONE 9:e113157, doi.
org/10.1371/journal.pone.0113157.s002
Klemme S, Meyer H-P (2003) Trace element partitioning between baddeleyite and carbonatite melt at high
pressures and high temperatures. Chem Geol 199:233–242
Koepke J, Berndt J, Feig ST, Holtz F (2007) The formation of SiO2 rich melts within the deep oceanic crust by
hydrous partial melting of gabbros. Contrib Mineral Petrol 153:67–84
Kotková J, Schaltegger U, Leichmann J (2010) Two types of ultrapotassic plutonic rocks in the Bohemian Massif—
Coeval intrusions at different crustal levels. Lithos 115:163–176, doi.org/10.1016/j.lithos.2009.11.016
Kotková J, Whitehouse M, Schaltegger U, D’Abzac FX (2016) The fate of zircon during UHT-UHP metamorphism:
isotopic (U/Pb, δ18O, Hf) and trace element constraints. J Metamorph Geol 34:719–739, doi.org/10.1111/
jmg.12206
Kusiak MA, Whitehouse MJ, Wilde SA, Nemchin AA, Clark C (2013) Mobilization of radiogenic Pb in zircon
revealed by ion imaging: Implications for early Earth geochronology. Geology 41:291–294
Kusiak MA, Dunkley DJ, Wirth R, Whitehouse MJ, Wilde SA, Marquardt K (2015) Metallic lead nanospheres
discovered in ancient zircons. PNAS 112:4958–4963
Kylander-Clark AR, Hacker BR, Cottle JM (2013) Laser-ablation split-stream ICP petrochronology. Chem Geol
345:99–112, doi.org/10.1016/j.chemgeo.2013.02.019
Lenz C, Nasdala L, Talla D, Hauzenberger C, Seitz R, Kolitsch U (2015) Laser-induced REE3+ photoluminescence
of selected accessory minerals—An “advantageous artefact” in Raman spectroscopy. Chem Geol 415:1–16
Leuthold J, Müntener O, Baumgartner L P, Putlitz B, Ovtcharova M, Schaltegger U (2012) Time resolved
construction of a bimodal laccolith (Torres del Paine, Patagonia). Earth Plan Sci Lett 325–326:1–8, doi.
org/10.1016/j.epsl.2012.01.032
Lissenberg CJ, Rioux M, Shimizu N, Bowrin, SA, Mével C (2009) Zircon dating of oceanic crustal accretion.
Science 323:1048–1050
Marzoli A, Jourdan F, Bussy F, Chiaradia M, Costa F (2014) Petrogenesis of tholeiitic basalts from the Central
Atlantic magmatic province as revealed by mineral major and trace elements and Sr isotopes. Lithos 188:44–59
Mattinson J (2005) Zircon U–Pb chemical abrasion (“CA-TIMS”) method: combined annealing and multi-step
partial dissolution analysis for improved precision and accuracy of zircon ages. Chem Geol 220:47–66
Miller J, Matzel J, Miller C, Burgess S, Miller R (2007) Zircon growth and recycling during the assembly of large,
composite arc plutons. J Volcanol Geotherm Res 167:282–299
Moser DE, Chamberlain KR, Tait KT, Schmitt AK, Darling JR, Barker IR, Hyde BC (2013) Solving the Martian
meteorite age conundrum using micro-baddeleyite and launch-generated zircon. Nature 499:454–457
Nandedkar RH, Ulmer P, Müntener O (2014) Fractional crystallization of primitive, hydrous arc magmas: an
experimental study at 0.7 GPa. Contrib Mineral Petrol 167:1015, doi.org/10.1007/s00410-014-1015-5
Nilsson MKM, Söderlund U, Ernst RE, Hamilton MA, Scherstén A, Armitage PEB (2010) Precise U–Pb
baddeleyite ages of mafic dykes and intrusions in southern West Greenland and implications for a possible
reconstruction with the Superior craton. Precamb Res 183:399–415
Niu Y, O’Hara MJ (2003) Origin of ocean island basalts: a new perspective from petrology, geochemistry, and
mineral physics considerations. J Geophys Res 108:1–19
Niu Y, Gilmore T, Mackie S, Greig A, Bach W (2002) Mineral chemistry, whole-rock compositions, and
petrogenesis of Leg 176 gabbros: data and discussion. Proc Ocean Drill Prog Sci Results 176:1–60
Oberli F, Meier M, Berger A, Rosenberg CL, Gieré R (2004) U–Th–Pb and 230Th/238U disequilibrium isotope
systematics: Precise accessory mineral chronology and melt evolution tracing in the Alpine Bergell intrusion.
Geochim Cosmochim Acta 68:2543–2560, doi.org/10.1016/j.gca.2003.10.017
Olsson JR, Söderlund U, Hamilton MA, Klausen MB, Helffrich GR (2011) A late Archaean radiating dyke swarm
as possible clue to the origin of the Bushveld Complex. Nat Geosci 4:865–869
Phillippi CM, Mazdiyasni KS (1971) Infrared and Raman spectra of zirconia polymorphs. J Am Ceram Soc
54:254–258
Piwinskii AJ, Wyllie PJ (1968) Experimental studies of igneous rock series: A zoned pluton in the Wallowa
batholith, Oregon. J Geol 76: 205–234
silicate melt. Geochim Cosmochim Acta 69:695–709, doi.org/10.1016/j.gca.2004.06.037
Prowatke S, Klemme S (2006) Trace element partitioning between apatite and silicate melts. Geochim Cosmochim
Acta 70:4513–4527, doi.org/10.1016/j.gca.2006.06.162
Reid MR, Coath CD (2000) In situ U–Pb ages of zircons from the Bishop Tuff: No evidence for long crystal
residence times. Geology 28:443–446
Reid MR, Coath, Ca.D, Harrison TM, McKeegan KD (1997) Prolonged residence times for the youngest rhyolites
associated with Long Valley Caldera: 230Th–238U ion microprobe dating of young zircons. Earth Planet. Sci.
Lett., 150:27–39
Reimink JR, Davies JHFL, Chacko T, Stern RA, Heaman LM, Sarkar C, Schaltegger U, Creaser RA, Pearson DG
(2016) No evidence for Hadean continental crust within Earth’s oldest evolved rock unit. Nat Geosci, doi.
org/10.1038/ngeo2786
Reubi O, Blundy J (2008) Assimilation of Plutonic Roots, Formation of High-K “Exotic” Melt Inclusions and
Genesis of Andesitic Magmas at Volcan De Colima, Mexico. J Petrol 49:2221–2243, doi.org/10.1093/
petrology/egn066
Rioux M, Bowring S, Dudás F, Hanson R (2010) Characterizing the U–Pb systematics of baddeleyite through
chemical abrasion: application of multi-step digestion methods to baddeleyite geochronology. Contrib
Mineral Petrol 160:777–801, doi.org/10.1007/s00410-010-0507-1
Rioux M, Lissenberg CJ, McLean NM, Bowring SA, MacLeod CJ, Hellebrand E, Shimizu N (2012) Protracted
timescales of lower crustal growth at the fast-spreading East Pacific Rise. Nat Geosci 5:275–278, doi.
org/10.1038/ngeo1378
Rioux M, Bowring S, Cheadle M, John B (2015a) Evidence for initial excess 321 Pa in mid-ocean ridge zircons.
Chem Geol 397:134–156
Rioux M, Jöns N, Bowring S, Lissenberg CJ, Bach W, Kylander-Clark A, Hacker B, Dudás F (2015b) U–Pb dating
of interspersed gabbroic magmatism and hydrothermal metamorphism during lower crustal accretion, Vema
lithospheric section, Mid-Atlantic Ridge. J Geophys Res: Solid Earth 120:2093–2118
Rioux M, Cheadle M, John B, Bowring S (2016) The temporal and spatial distribution of magmatism during lower
crustal accretion at an ultraslow-spreading ridge: High-precision U–Pb zircon dating of ODP Holes 735B
and 1105A, Atlantis Bank, Southwest Indian Ridge. Earth Planet Sci Let doi: 10.1016/j.epsl.2016.05.047
Rosera JM, Coleman DS, Stein HJ (2013) Re-evaluating genetic models for porphyry Mo mineralization at Questa,
New Mexico: Implications for ore deposition following silicic ignimbrite eruption. Geochem Geophys
Geosystem 14:787–805, doi.org/10.1002/ggge.20048
Rubatto D, Hermann J (2007) Experimental zircon/melt and zircon/garnet trace element partitioning and implications
for the geochronology of crustal rocks. Chem Geol 241:38–61, doi.org/10.1016/j.chemgeo.2007.01.027
Rubin A, Cooper KM, Leever M, Wimpenny J, Deering C, Rooney T, Gravley D, Yin QZ (2016) Changes in
magma storage conditions following caldera collapse at Okataina Volcanic Center, New Zealand. Contrib
Samperton KM, Schoene B, Cottle JM, Keller CB, Crowley JL, Schmitz MD (2015) Magma emplacement,
differentiation and cooling in the middle crust: Integrated zircon geochronological–geochemical constraints
from the Bergell Intrusion, Central Alps. Chem Geol 417:322–340, doi.org/10.1016/j.chemgeo.2015.10.024
Simakin A, Bindeman I (2008) Evolution of crystal sizes in the series of dissolution and precipitation events in
open magma systems. J Volcanol Geothermal Res 177:997–1010, doi.org/10.1016/j.jvolgeores.2008.07.012
Schaltegger U, Schmitt AK, Horstwood MSA (2015) U–Th–Pb zircon geochronology by ID-TIMS, SIMS,
and laser ablation ICP-MS: Recipes, interpretations, and opportunities. Chem Geol 402:89–110, doi.
org/10.1016/j.chemgeo.2015.02.028
Schärer U (1984) The effect of initial 230Th disequilibrium on young U–Pb ages; the Makalu case, Himalaya. Earth
Schmitt AK (2011) Uranium series accessory crystal dating of magmatic processes. Ann Rev Earth Planet Sci
39:321–349, doi.org/10.1146/annurev-earth-040610-133330
Geochem 83:199–230
Schmitt AK, Chamberlain KR, Swapp SM, Harrison TM (2010) In situ U–Pb dating of micro-baddeleyite by
secondary ion mass spectrometry. Chem Geol 269:386–395
Schoene B (2014) U–Th–Pb Geochronology. Treatise of Geochemistry, The Crust (2nd ed., Vol. 4, pp. 341–378).
Elsevier Ltd., doi.org/10.1016/B978-0-08-095975-7.00310–7
Schoene, B, Bowring S (2006) U–Pb systematics of the McClure Mountain syenite: thermochronological
constraints on the age of the 40Ar/39Ar standard MMhb. Contrib Mineral Petrol 151:615–630
Schoene B, Bowring SA (2010) Rates and mechanisms of Mesoarchean magmatic arc construction, eastern kaapvaal
craton, Swaziland. Geol Soc Amer Bull 122:408–429, doi.org/10.1016/0012-821X(96)00049–0
Schoene B, Latkoczy C, Schaltegger U, Günther D (2010) A new method integrating high-precision U–Pb
geochronology with zircon trace element analysis (U–Pb TIMS-TEA). Geochim Cosmochim Acta 74:7144–
7159, doi.org/10.1016/j.gca.2010.09.016
Schoene B, Schaltegger U, Brack P, Latkoczy C, Stracke A, Günther D (2012) Rates of magma differentiation and
emplacement in a ballooning pluton recorded by U–Pb TIMS-TEA, Adamello batholith, Italy. Earth Planet
Sci Lett 355–356:162–173, doi.org/10.1016/j.epsl.2012.08.019
Sell B, Ovtcharova M, Guex J, Bartolini A, Jourdan F, Spangenberg JE, Vicente JC, Schaltegger U (2014)
Evaluating the temporal link between the karoo LIP and climatic–biologic events of the Toarcian Stage with
high-precision U–Pb geochronology. Earth Plan Sci Lett 408:48–56 doi.org/10.1016/j.epsl.2014.10.008
Shannon RD (1976) Revised effective ionic radii and systematic studies of interatomic distances in halides and
chalcogenides. Acta Crystallogr A32:751–767
Sickafus KE, Matzke H, Hartmann T, Yasuda K, Valdez JA, Chodak P, Nastasi M, Verrall RA (1999) Radiation
damage effects in zirconia. J Nucl Mat 274:66–77
Silver LT, Deutsch S (1963) Uranium-lead isotopic variations in zircons: a case study. J Geol 71:721–758
Simeone D, Baldinozzi G, Gosset D, Caër SLe (2006) Phase transition of pure zirconia under irradiation: A
textbook example. Nucl Inst Methods Phys Res B 250:95–100
Skopelitis A (2014) Formation of a tonalitic batholith through sequential accretion of magma batches: a study of
chemical composition, age and emplacement mechanisms of the Adamello Batholith N Italy. Unpubl PhD
thesis nr. 4660, University of Geneva
Söderlund U, Johanson L (2002) A simple way to extract baddeleyite (ZrO2). Geochem Geophys Geosystem 3,
doi.org/10.1029/2001GC000212
Söderlund U, Hellström FA, Kamo SL (2008) Geochronology of high-pressure mafic granulite dykes in SW
Sweden: tracking the P–T–t path of metamorphism using Hf isotopes in zircon and baddeleyite. J Metamorph
Geol 26:539–560
Söderlund U, Hofmann A, Klausen MB, Olsson JR, Ernst RE, Persson P-O (2010) Towards a complete magmatic
barcode for the Zimbabwe craton: Baddeleyite U–Pb dating of regional dolerite dyke swarms and sill
complexes. Precam Res 183:388–398
Storm S, Schmitt AK, Shane P, Lindsay JM (2014) Zircon trace element chemistry at sub-micrometer resolution
for Tarawera volcano, New Zealand, and implications for rhyolite magma evolution. Contrib Mineral Petrol
167:1–19
Svensen H, Planke S, Polozov AG, Schmidbauer N, Corfu F, Podladchikov YY, Jamtviet B (2009) Siberian gas
venting and the end-Permian environmental crisis. Earth Planet Sci Lett 277:490–500
Tappa MJ, Coleman DS, Mills RD, Samperton KM (2011) The plutonic record of a silicic ignimbrite from the Latir
volcanic field, New Mexico. Geochem Geophys Geosystem 12, doi.org/10.1029/2011GC003700
Tapster S, Condon DJ, Naden J, Noble SR, Petterson MG, Roberts NMW, Saunders AD, Smith JD (2016) Rapid
thermal rejuvenation of high-crystallinity magma linked to porphyry copper deposit formation; evidence
from the Koloula Porphyry Prospect, Solomon Islands. Earth Planet Sci Lett 442:206–217, doi.org/10.1016/j.
epsl.2016.02.046
Taylor HP, Forester RW (1979) An oxygen and hydrogen isotope study of the Skaergaard intrusion and its country
rocks: A description of a 55 My-old fossil hydrothermal system. J Petrol 20:355–419
Tilton GR, Patterson C, Brown H, Inghram M, Hayden R, Hess D, Larsen E (1955) Isotopic composition and
distribution of lead, uranium and thorium in a Precambrian granite. Geol Soc Am Bull 66:1131–1148
Trachenko K (2004) Understanding resistance to amorphiziation by radiation damage. J Phys Condens Matter
16:R1491–R1515
Trail D, Watson EB, Tailby ND (2012) Ce and Eu anomalies in zircon as proxies for the oxidation state of magmas.
Geochim Cosmochim Acta 97:70–87, doi.org/10.1016/j.gca.2012.08.032
geospeedometer and peak temperature indicator. Contrib Mineral Petrol 171, doi 10.1007/s00410-016-1238-8
Valley JW (2003) Oxygen isotopes in zircon. Rev Mineral Geochem 53:343–385
Valley JW, Reinhard DA, Cavosie AJ, Ushikubo T, Lawrence DF, Larson DJ, Kelly TF, Snoeyenbos DR, Strickland
A (2015) Nano- and micro-geochronology in Hadean and Archean zircons by atom-probe tomography and
SIMS: New tools for old minerals. Am Mineral 100:1355–1377, doi.org/10.2138/am-2015-5134
Vermeesch P (2012) On the visualisation of detrital age distributions. Chem Geol 312–313:190–194, doi.
org/10.1016/j.chemgeo.2012.04.021
Wall CJ, Scoates JS (2016) High precision U–Pb zircon-baddeleyite dating of the J-M Reef platinum group element
deposit in the Still Water complex, Montana (USA). Econ Geol 111:771–782
Wall CJ, Scoates JS, Weis D (2016) Zircon from the Anorthosite zone II of the Stillwater Complex as a U–Pb
geochronological reference material for Archean rocks. Chem Geol 436:54–71
Wang X, Griffin WL, Chen J, Huang P, Li X (2011) U and Th contents and Th/U ratios of zircon in felsic and mafic
magmatic rocks: improved zircon–melt distribution coefficients. Acta Geol Sinica 85:11164–17411
Watson EB (1996) Dissolution, growth and survival of zircons during crustal fusion: kinetic principles, geological
models and implications for isotopic inheritance. Trans R Soc Edinburgh 87:43–56, doi.org/10.1017/
S0263593300006465
Watson EB, Liang Y (1995) A simple model for sector zoning in slowly grown crystals: Implications for growth
rate and lattice diffusion, with emphasis on accessory minerals in crustal rocks. Am Mineral 80:1179–1187
Wotzlaw J-F, Bindeman IN, Schaltegger U, Brooks, Ca.K, Naslund HR (2012) High-resolution insights into
episodes of crystallization, hydrothermal alteration and remelting in the Skaergaard intrusive complex. Earth
Planet Sci Lett, 355–356:199–212
Wotzlaw JF, Schaltegger U, Frick DA, Dungan MA, Gerdes A, Günther D (2013) Tracking the evolution of large-
volume silicic magma reservoirs from assembly to supereruption. Geology 41:867–870, doi.org/10.1130/
G34366.1
Wotzlaw JF, Bindeman IN, Watts KE, Schmitt AK, Caricchi L, Schaltegger U (2014) Linking rapid magma
reservoir assembly and eruption trigger mechanisms at evolved Yellowstone-type supervolcanoes. Geology
42:807–810, doi.org/10.1130/G35979.1
Wotzlaw JF, Bindeman IN, Stern RA, D’Abzac FX, Schaltegger U (2015) Rapid heterogeneous assembly of
multiple magma reservoirs prior toYellowstone supereruptions. Sci Rep 1–10, doi.org/10.1038/srep14026
Wu F-Y, Yang Y-H, Mitchell RH, Bellatreccia F, Li Q-L, Zhao Z-F (2010) In situ U–Pb and Nd–Hf–(Sr) isotopic
investigations of zirconolite and calzirtite. Chem Geol 277:178–195
Wu WN, Schmitt AK, Pappalardo L (2015) U–Th baddeleyite geochronology and its significance to date the
emplacement of silica undersaturated magmas. Am Mineral 100:2082–2090
Yang W, Lin YT, Zhang JC, Hao JL, Shen WJ, Hu S (2012) Precise micrometre-sized Pb–Pb and U–Pb dating with
NanoSIMS J Anal Atom Spectrom 27:479, doi.org/10.1039/c2ja10303f
Yuan HL, Gao S, Dai MN, Zong Ca.L, Gunther D, Fontaine GH, Liu XM, Diwu Ca.R,. (2008) Simultaneous
quadrupole and multiple-collector ICP-MS Chem Geol 247:100–118
Zeh A, Ovtcharova M, Wilson AH, Schaltegger U (2015) The Bushveld Complex was emplaced and cooled in less
than one million years—results of zirconology, and geotectonic implications. Earth Planet Sci Lett 418:103–
114, doi.org/10.1016/j.epsl.2015.02.035
Vol. 83 pp. 329–363, 2017 11
Hadean Zircon Petrochronology

T. Mark Harrison, Elizabeth A. Bell, Patrick Boehnke
Department of Earth, Planetary and Space Sciences
University of California, Los Angeles
Los Angeles, CA 90095
USA
tmark.harrison@gmail.com
ebell21@ucla.edu
pboehnke@gmail.com
INTRODUCTION
The inspiration for this volume arose in part from a shift in perception among
U–Pb geochronologists that began to develop in the late 1980s. Prior to then, analytical
geochronology emphasized progressively lower blank analysis of separated accessory mineral
aggregates (e.g., Krogh 1982; Parrish 1987), with results generally interpreted to reflect a singular
moment in time. For example, a widespread measure of confidence in intra-analytical reliability
was conformity to an MSWD (a form of χ2 test; Wendt and Carl 1991) of unity. This approach
implicitly assumed that geological processes act on timescales that are short with respect to
analytical errors (e.g., Schoene et al. 2015). As in situ methodologies (e.g., Compston and
Pidgeon 1986; Harrison et al. 1997; Griffin et al. 2000) and increasingly well-calibrated double
spikes (e.g., Amelin and Davis 2006; McLean et al. 2015) emerged, geochronologists began to
move away from interpreting geological processes as a series of instantaneous episodes (e.g.,
Rubatto 2002). At about the same time, petrologists developed techniques that permitted in situ
chemical analyses to be interpreted in terms of continuously changing pressure–temperature–
time histories (e.g., Spear 1988). The recognition followed that specific mineral reactions
yielded products that could be directly dated or interpreted in terms of protracted petrogenetic
processes. Part of this shift was due to an appreciation that trace elements in accessory phases
could identify the changing nature of modal mineralogy during crystal growth (e.g., Pyle et al.
2001; Kohn and Malloy 2004) and thus potentially relate petrogenesis to absolute time. The
transition to petrochronology was complete upon recognition that high MSWDs were in fact
the expected case for most metamorphic minerals (Kohn 2009).
One of the great frontiers for fundamental discovery in the geosciences is earliest Earth
(DePaolo et al. 2008). However, investigations of the first five hundred million years of Earth
history—known as the Hadean eon (Cloud 1972, 1976)—are limited by the lack of a rock
record older than 4.02 Ga (Bowring and Williams 1999; Mojzsis et al. 2014; Reimink et al.
2016; cf. O’Neil et al. 2008). This potentially leaves only a single strategy—the examination
of Hadean detrital or inherited minerals—to directly assess the geophysical conditions, and
therefore habitability, of early Earth. Not having access to the rock context in which a mineral
geochronometer formed does reduce opportunities to understand its growth medium and the
forces acting on it during crystallization. However, all is not lost. Virtually every accessory
phase contains both trace element signatures and inclusions of coexisting minerals that were
incorporated during its formation. Thus we can stand what seems to be a limitation on its head
by viewing, for example, a detrital zircon as both a micro-rock encapsulation system and an
elemental partition mirror of the magma from which it grew.

1943-2666/17/0083-0011$05.00 (online)
330 Harrison, Bell & Boehnke
In this chapter, we review the petrochemical and -chronological systems in zircon and
their application to detrital grains from the Jack Hills regions, enumerate the thirteen presently
known localities from which Hadean zircons have been documented, review their age and
geochemical properties, and discuss the results in context of possible and unlikely sources. For
our purposes, we arbitrarily define the Hadean as the period of Earth history prior to formation
of the oldest documented rock (i.e., older than the age of the Acasta metatonalite at 4.02 Ga). We
conclude that, in contrast to the longstanding paradigm of a hellish early Earth devoid of oceans,
continents and life, the Hadean zircon record, and the micro-rocks that they encapsulate, largely
grew under a range of conditions far more similar to the present than once imagined.
WHY STUDY HADEAN ZIRCONS?

Due to zircon’s inherent resistance to alteration by weathering, dissolution, shock, and
diffusive exchange, and its enrichment in U and Th relative to daughter product Pb (Hanchar
and Hoskin 2003), the U–Pb zircon system has long been regarded as the premier crustal
geochronometer. While highly valued in that role, the trace element and isotopic compositions
of zircon have become recognized as valuable probes of environmental conditions experienced
during crystallization. Even in cases where zircon has been removed from its original rock context,
such as detrital grains in clastic rocks, inclusions, trace element patterns and isotopic signatures
can yield important information regarding source conditions if the record is undisturbed. Having
emphasized the remarkably refractory nature and resistance to diffusive exchange of zircon
(Cherniak and Watson 2003), it is important to note its Achilles heel. Zircon is sensitive to
radiation damage and can degrade into heterogeneous microcrystalline zones encompassed by
amorphous material (Ewing et al. 2003). Nature has, to some degree, already weeded out those
grains most susceptible to metamictization from detrital zircon populations as high U and Th
grains are unlikely to survive sediment transport (e.g., Hadean Jack Hills zircons with original U
concentrations > 600 ppm are exceedingly rare). Thus care must be taken to ensure that effects of
post-crystallization alteration are not mistaken as primary features.
The importance of Hadean zircons is then in the coupling of their great antiquity with
their amenability to U–Pb dating and capacity to retain geochemical information. Although
first documented at nearby Mt. Narryer (Froude et al. 1983), the vast majority of investigations
of > 4 Ga zircons sampled heavy-mineral-rich quartz–pebble conglomerates from a locality
in the Erawondoo region of the Jack Hills (Fig. 1) (Compston and Pidgeon 1986; Maas et al.
1992; Spaggiari et al. 2007). Zircons are typically extracted from these rocks using standard
separatory methods based on their high density and low magnetic susceptibility, handpicked
and secured in an epoxy mount which is then polished and analyzed using the 207Pb/206Pb ion
microprobe dating approach (see Holden et al. 2009).
Over 200,000 of these grains have been 207Pb/206Pb dated in this fashion with over 6,000
yielding ages older than 4 Ga. Most of the ~3% of the analyzed grains that are > 4 Ga were then
U–Pb dated using an ion microprobe, with several found to be as old as 4.38 Ga (Compston
and Pidgeon 1986; Holden et al. 2009; Valley et al. 2014).
Although zircon is dominantly a mineral of the continental crust, its formation is not restricted
to that environment nor, for that matter, to Earth. However, zircons of continental affinity can
be readily distinguished from those derived from the mantle or oceanic crust by trace element
characteristics (e.g., U/Y vs. Y) and significantly lower crystallization temperatures (Grimes et al.
2007; Hellebrand et al. 2007). Lunar and meteoritic zircons can be distinguished from terrestrial
counterparts by their REE signature (e.g., lack of a Ce anomaly; Hoskin and Schaltegger 2003).
Furthermore, apparent crystallization temperatures for lunar zircons range from 900 to 1100 °C
(Taylor et al. 2009) in contrast to terrestrial Hadean zircons which are restricted to 600–780 °C
(Harrison et al. 2007; Fu et al. 2008). Thus it is amply clear that the vast majority of Hadean
Hadean Zircon Petrochronology 331
Figure 1. Quartz–pebble conglomerates on Erawondoo Hill, Jack Hills region of Western Australia.
Samples collected within ~100 m of this site have produced > 95% of all Hadean zircons yet documented
(Photo credit: Bruce Watson).
zircons are derived from terrestrial continental lithologies. Furthermore, textural characteristics
of Hadean zircons from Jack Hills (e.g., growth zoning, inclusion mineralogy) indicate that most
are derived from igneous sources (e.g., Cavosie et al. 2004; Hopkins et al. 2008).
Geochemical studies using upwards of half of the total Hadean grains thus far documented
have inspired a variety of interpretations. However, there is a broad consensus that evidence
derived from these ancient zircons implies abundant water at or near Earth’s surface during
that era (e.g., Wilde et al. 2001; Mojzsis et al. 2001; Rollinson 2008; Shirey et al. 2008;
Harrison 2009). This represents a dramatic reversal from the conception of an uninhabitable,
hellish world from which this time period gets its name (Solomon 1980; Smith 1981; Maher
and Stevenson 1988; Abe 1993; Ward and Brownlee 2000).
We note at the outset that the intrinsic limitations (including preservation bias) of the Hadean
zircon record could prevent ‘smoking gun’ conclusions about earliest Earth from ever being
drawn. For example, the 70% of the Earth’s surface that is today covered by MORB contributes
essentially nothing to the archive of detrital or xenocrystic zircons. This concern is sure to diminish
as Hadean zircons are documented from a growing number of globally diverse locations.
Modes of investigation
Most investigations of Hadean zircons to date have emphasized ion microprobe
analysis to minimize the volume of mineral excavated during age surveys, thus maximizing
the signals in our subsequent analyses (i.e., δ18O, 176Hf/177Hf, Ti, etc.). This made sense for
early studies of Jack Hills zircons (e.g., Compston and Pidgeon 1986; Mojzsis et al. 2001)
as there were then no serious alternatives to the ion microprobe. Typical practice was to
handpick individual zircons and mount them on double-sided adhesive tape in systematic
grids together with zircon standards. This enabled the most ancient grains identified by
the 207Pb/206Pb age survey to be easily located for subsequent analysis. When we began
the program to date over 100,000 Jack Hills zircons in 2001, this laborious mounting
process was not the rate limiting step in creating a large archive of Hadean zircons as then
no ion microprobe yet had automated analysis capability. Indeed, that project led to the
development of the automated stage on the SHRIMP instruments, followed shortly thereafter
by CAMECA’s ‘chain analysis’ tool. Arguably the most remarkable analytical development
over the subsequent 15 years has been the development and refinement of laser ablation,
inductively-coupled mass spectrometry (LA-ICP-MS). Effective yields have increased by
over an order of magnitude dropping both analytical time and the mass of material needed
to attain a specified precision. This in turn has decreased costs dramatically. While the
ion microprobe remains the ultimate tool for in situ U–Pb dating, it is perhaps no longer
cost effective in undertaking large (i.e., ≥ 5,000 U–Pb zircon) age surveys. Where once
zircons were largely evaporated to attain an LA-ICP-MS U–Pb age, modern multicollector
instruments can obtain a U–Pb with ±2% precision from a ~1000 µm3 volume (Ibanez-
Mejia et al. 2014) which begins to compare favorably with the ~150 µm3 volume long
attainable using the ion microprobe. Even lower cost, quadrupole LA-ICP systems can
attain similar precision from ~6,000 µm3 ablation craters, which still represents only ~4%
of a typical Hadean zircon mass (i.e., ~1 µg). However, a common Pb correction based on
204
Pb is problematic for LA-ICP instruments owing to near ubiquitous background at 204Hg
leading to seemingly precise but potentially inaccurate U–Pb ages.
Age distributions. Numerous age studies of Jack Hills detrital zircons all show a
characteristic bimodal distribution with peaks close to 3.4 and 4.1 Ga with some grains as old
as nearly 4.4 Ga (Compston and Pidgeon 1986; Maas et al. 1992; Amelin 1998; Amelin et al.
1999; Mojzsis et al. 2001; Cavoise et al. 2004; Trail et al. 2007; Holden et al. 2009; Bell et al.
2011, 2014; Bell and Harrison 2013).
How abundant are Hadean zircons on Earth? Most of the dozen or so localities for which
at least one > 4 Ga zircon has been documented were not targeted for that purpose but rather
discovered serendipitously. Ancient metasediments and orthogneisses for which 10s to 100s of
zircons have been U–Pb dated without identifying at least one Hadean zircon must fall into one
of two categories: 1) those in which > 4 Ga zircons are present at a level of less than ~1% but
have not yet been detected, and 2) those in which they are simply absent. How many zircons
should be dated to ascertain to which category a sample belongs? The probability of detecting
at least a single > 4 Ga zircon as a function of abundance is shown in Fig. 2. A reasonable
assumption is that the ca. 3% > 4 Ga zircon abundance in the Jack Hills (Holden et al. 2009) is
anomalously high for most Archean quartzites and thus we examine the detection probabilities
where abundances are between one and two orders of magnitude lower (i.e., 0.2–0.02%). From
Fig. 2, we can see that at the 95% confidence level (bright red band), diminishing returns are
achieved following analysis of ~5000 zircons, corresponding to an effective abundance limit
of ~0.05%. To our knowledge, only the Erawondoo locality in the Jack Hills has had more than
5000 zircons dated and thus intercomparisons are as yet of limited value.
0.0020
1.000
0.0018 0.9702
0.0016 0.8806
Fraction of >4 Ga zircons
0.7612 Figure 2. Plot showing the

0.0014 probability of identifying at
0.5225
least one Hadean zircon from
0.0012 0.2838 populations with a range of
assumed occurrence rates of
0.0010 0.04500 > 4 Ga grains as a function of
number of dated grains. Note
0.0008 that achieving 89–93% confi-
dence in detecting a Hadean zir-
con from a population in which
0.0006 1-in-2000 are > 4 Ga requires
U–Pb dating ~5000 grains.
0.0004
0.0002
5000 10000 15000 20000
Number of zircons sampled
Indeed, the significance of Hadean zircons is sometimes dismissed by their seeming rarity—
with less than a handful of these grains (the total mass acquired is less than 6 g), how can one
begin to articulate the nature of early Earth? Our view is that this type of question is akin to
asking how the Big Bang could possibly be characterized by capture of a vanishingly small
fraction (<10−70) of the photons in the observable universe. That is, observations on rare materials
can lead to profound insights. There are two principal factors that inform this view. The first is,
we do not expect to preserve a significant proportion of early formed crust on a dynamic planet.
Assuming that Earth has recycled crust since formation at least as efficiently as we recognize
it has throughout the Phanerozoic, only a few percent at most would likely remain (Armstrong
1981). The second is that we simply have not tried hard enough to find these remnants; we have
sampled considerably less than 10−17 of the continental crust for geochronology, and support for
reconnaissance dating surveys (i.e., “fishing trips”) is notoriously difficult to obtain.
Isotope geochemistry. Several elements abundant in zircon comprise isotopic systems
relevant to petrogenesis and have a significant role in defining conditions not only during
the Hadean but throughout Earth history. The 18O/16O of magmas contains information
regarding their sources with primary variations often reflecting incorporation of aqueously
altered materials. Mantle-derived magmas display a narrow range of 18O/16O, corresponding
to zircons with an average δ18OSMOW of 5.3 ± 0.3 (1σ) (Valley et al. 1998). Aqueous alteration
at low temperatures results in clay-rich sediments with higher δ18O, whereas hydrothermal
alteration generally imparts lower δ18O values. Incorporation of these altered materials into
later magmas results in significant deviation from the mantle average value (e.g., O’Neil and
Chappell 1977) which is reflected in the compositions of all silicate and oxide phases present,
including zircon. That some Hadean Jack Hills zircons are significantly above the mantle
value suggests abundant liquid water in the surface or near-surface environment as early as
ca. 4.3 Ga (e.g., Mojzsis et al. 2001; Peck et al. 2001).
The Lu–Hf system is based on the decay of 176 Lu to 176Hf (t½ = 37 Ga; Söderlund et al.
2004). Lu and Hf are fractionated during partial melting, such that higher Lu/Hf ratios form
in depleted mantle and lower Lu/Hf ratios form in continental crust. Over Earth’s history,
this leads to differences in 176Hf relative to the stable, primordial isotope 177Hf (represented
by εHf) in these two reservoirs. The spread in εHf between the continental crust and depleted
mantle allows for the calculation of a model age of mantle extraction for igneous rocks. This
isotopic system is useful on the level of individual zircons due to the incorporation of abundant
(up to several weight percent) Hf in zircon and the lesser, ca. 100 ppm-level incorporation of Lu
(e.g., Hoskin and Schaltegger 2003), which allows zircon to preserve the original magma εHf
with very little age-correction for 176 Lu ingrowth. Hadean Jack Hills zircons show dominantly
negative (i.e., old crustal) εHf with some grains requiring separation of very low-Lu/Hf (i.e.,
felsic) reservoirs by ca. 4.5 Ga (Harrison et al. 2008; Bell et al. 2014).
Mineral inclusions. As zircon crystallizes in the solid state or from magmas, it almost
invariably traps exotic phases such as crystals and melt or other fluids (e.g., Maas et al. 1992;
Chopin and Sobolev 1995; Tabata et al. 1998; Liu et al. 2001; Fig. 3). Coupled with the
capacity for accurate U–Pb dating of the zircon host, these inclusions are a potentially rich
source of information about petrogenetic conditions of formation and/or provenance.
Although mineral inclusions in magmatic zircon can record diagnostic information about the
environment in which they formed, the extent to which the mineralogy and chemistry of zircons
and their inclusion can be used to reconstruct petrogenesis and provenance is only now becoming
clear. Darling et al. (2009) concluded that mineral inclusions in zircons grown in intermediate to
felsic melts within the Sudbury impact melt sheet imply somewhat more felsic melt conditions
than the associated whole rock in terms of the modal proportions of quartz, alkali feldspar, and
plagioclase. However, Jennings et al. (2011) showed that the chemistry of igneous apatite and
mafic phases is typically similar between crystals included in zircon and those in the whole rock.
Establishing the primary nature of inclusions and contamination introduced during
sample preparation are potentially serious concerns that need to be explicitly addressed. For
example, reports of abundant diamonds and graphite in Hadean Jack Hills zircons (Menneken
et al. 2007; Nemchin et al. 2008) were later determined to be contaminants introduced during
sample preparation (Dobrzhinetskaya et al. 2014).
Figure 3. Cathodoluminescence image of Hadean Jack Hills zircon RSES77-5.7. This 4.06 ± 0.1 Ga con-
cordant grain contains likely primary inclusions of quartz, rutile and muscovite permitting reliable Ti ther-
mometry and phengite barometry which can then be used to infer near surface thermal structure.
Mineral inclusions coupled with the chemistry of their host zircon are an underexploited
resource for establishing internally consistent evidence for host rock character. The advent
of the Ti-in-zircon thermometer, for instance, underscored the potential for thermodynamic
relationships between included phases and elements partitioned into the zircon structure.
Similarly, the incorporation of aluminous and carbonaceous inclusions into zircon (Hopkins et al.
2008; Rasmussen et al. 2011; Bell et al. 2015b; Harrison and Wielicki 2015) raises the possibility
of calibrating trace elements in zircon as an indicator of host melt chemistry or volatile content.
Zircon geochemistry. Zircon incorporates many elements at the trace or minor level
during crystallization, some of which are useful petrologic indicators. For example, the
content of Ti in zircon serves as a crystallization thermometer given knowledge of the melt
aSiO2 and aTiO2 (Watson and Harrison 2005; Ferry and Watson 2007). Ce/Ce*, or the excess in
Ce over the other light rare earth elements (LREE) La and Pr, is a proxy for magma fO2 (Trail
et al. 2011a). Th/U can generally be used to distinguish magmatic from metamorphic zircons,
with metamorphic zircons typically < 0.07 and igneous zircon at higher values (Rubatto 2002,
2017). Other trace element concentrations have less quantitative ties to petrogenesis but may
have the potential to yield important information.
Rare earth elements (REE) occur in terrestrial zircon with a characteristic chondrite-
normalized abundance pattern characterized by relatively low LREE and increasingly
abundant REE with increasing Z (e.g., Hoskin and Schaltegger 2003). Two exceptions to
this rule include the aforementioned excess in Ce (Ce/Ce*) which is seen among virtually all
unaltered terrestrial zircons and a deficit in Eu (Eu/Eu*). The steady increase in compatibility
with increasing atomic mass for most REE in the zircon lattice results from the steady decrease
in ionic radius coupled with the trivalent oxidation state in which most REE are found in
the crustal and surficial environment. Significant amounts of tetravalent Ce (which is more
compatible in zircon) and divalent Eu (largely taken up by plagioclase) lead to their respective
anomalous contents. However, interpreting REE patterns in terms of zircon petrogenesis
requires distinguishing pristine from altered zircon chemistry, and hydrothermal alteration
of zircon is usually accompanied by an increase in LREE relative to the other REE and a
flattening of the LREE pattern, obscuring the Ce/Ce* and potentially the Eu/Eu*.
JACK HILLS ZIRCONS

Isotopic results
U–Pb age. Various age surveys of detrital zircons from the Erawondoo Hill discovery
site conglomerate (e.g., Crowley et al. 2005; Holden et al. 2009) generally show the zircons
to have a bimodal age distribution with major peaks at ca. 3.4 and 4.1 Ga. Concordant zircons
older than ca. 3.8 Ga make up approximately 5% of the population, and zircons become much
less abundant with age older than ca. 4.2 Ga (Holden et al. 2009; Fig. 4). The remaining 95%
of the population is mostly concentrated between 3.3 and 3.6 Ga, with a deficit of zircon ages
between 3.6 and 3.8 Ga (Bell and Harrison 2013).
The confidence with which one can interpret the meaning of a U–Pb date of a > 4 Ga
zircon is challenged by the potential for later fluid alteration and thermal disturbances. While
the concordance of U–Pb analyses (or lack thereof) can be used to assess the robustness of an
interpreted age, this is generally insensitive to early Pb loss. Therefore, assessing the general
reliability of ion microprobe U–Pb ages is an open challenge, especially given that most Hadean
Jack Hills zircons contain multiple age domains. Valley et al. (2014) examined the possibility of
Pb redistribution in a 4.38 Ga zircon core, imaged using atom probe tomography, encompassed
by a ca. 3.4 Ga, 10-20 μm rim. In this analytical technique, a zircon sliver extracted using a
focused ion beam was field evaporated and the emergent ions mass analyzed with a spatial
Figure 4. Histograms and probability-density curves for concordant Jack Hills zircons. (a) Histogram of rapid
initial survey of individual 207Pb/206Pb ages undertaken to identify the > 3.9 Ga population. Inset shows the
whole population of 4500 rapidly scanned 207Pb/206Pb ages. (b) Histogram and probability density for the con-
cordant > 4.2 Ga zircons. The small peak at 4.35 Ga may be the oldest surviving crustal remnant. Reprinted
fromPeter Holden et al. Mass spectrometric mining of Hadean zircons by automated SHRIMPmulticollector
and singlecollector U/Pb zircon age dating: The first 100,000 grains, International Journal of Mass Spectrometry
286(2–3):53–63 (2009), with permission from Elsevier.
resolution of <1 nm. Results from their analysis show Pb redistribution into “nanoclusters” with
~10 nm diameter and spacing of ~10-50 nm. The 207Pb/206Pb age of the zircon outside of the
“nanoclusters” is ~3.4 Ga and ~4.4 Ga for the entire analyzed volume. These data are consistent
with an event at 3.4 Ga that mobilized radiogenic Pb that had accumulated since the zircon’s
crystallization at 4.38 Ga. The mobilized Pb migrated into “nano clusters” on a length scale of
<50 nm, which is below the lateral spatial resolution of an ion microprobe. This finding supports
the view that due to the generally slow diffusion of Pb in zircon and zircon’s resistance to
alteration, concordant U–Pb analyses likely record actual zircon crystallization ages.
Oxygen isotopes. Elevated values of δ18OSMOW observed in Hadean Jack Hills zircons led
two independent groups to simultaneously propose (Mojzsis et al. 2001, Wilde et al. 2001) that
the protolith of these grains contained 18O-enriched clay minerals, in turn implying that liquid
water was present at or near the Earth’s surface by ~4.3 Ga. Numerous follow-up measurements
(e.g., Cavosie et al. 2005, Trail et al. 2007, Harrison et al. 2008; Bell et al. 2016) confirmed that
a significant fraction of Hadean Jack Hills zircons contain 18O-enrichments 2 to 3‰ above the
mantle zircon value of 5.3‰ (Valley et al. 1998). As the oxygen isotope fractionation between
zircon and granitoid melt is approximately −2‰ (Valley et al. 1994, Trail et al. 2009), δ18O
values of the melt from which the zircons crystallized are inferred to have been up to +9 ‰.
Phanerozoic granitoids derived largely from orthogneiss protoliths (I types) tend to
have δ18O between 8-9‰, whereas those derived by melting of clay-rich (i.e., 18O enriched)
metasedimentary rocks (S types) have higher δ18O (O’Neil and Chappell 1977). Granitoids with
δ18O values significantly less than 5‰ likely reflect hydrothermal interaction with meteoric
water (Taylor and Sheppard 1986) rather than weathering. In general, S-type granitoids form by
anatexis of metasediments enriched in 18O, compared with I-type granitoids that form directly
or indirectly from arc processes (Chappell and White 1974). Jack Hills zircons enriched in 18O
thus provide evidence indicating the presence in the protolith of recycled crustal material that had
interacted with liquid water under surface, or near surface, conditions (i.e., at low temperature).
A limitation to this interpretation is the possibility of oxygen isotope exchange under
hydrous conditions, even at post-depositional temperatures experienced by Jack Hills zircons
(i.e., ~450 °C). For example, the characteristic diffusion distance for oxygen in zircon at
500 °C for 1 Ma is ~1 µm, assuming a high water activity (Watson and Cherniak 1997). Thus
it is conceivable that oxygen isotope exchange during protracted thermal events could have
introduced the heavy oxygen signature. This concern is somewhat mitigated by the relative
improbability that hydrothermal fluids were highly δ18O enriched. However, it does not
preclude isotopic equilibration from having occurred prior to deposition at ca. 3 Ga.
Lutetium–Hafnium. Studies of initial 176Hf/177Hf in > 4 Ga Jack Hills zircons show large
deviations in εHf(T) from bulk silicate Earth (Kinny et al. 1991; Amelin et al. 1999; Harrison
et al. 2005, 2008; Blichert-Toft and Albarède 2008; Bell et al. 2011, 2014; Kemp et al. 2010)
that have been generally interpreted to reflect an early major differentiation of the silicate Earth
(Fig. 5). Modeling these data by associating εHf(T) with the range of 176 Lu/177Hf observed in large
datasets of analyzed crustal rocks are consistent with the formation of crust occurring essentially
continuously since 4.5 Ga. Several data (Harrison et al. 2008; Bell et al. 2014) yield εHf(T) within
uncertainty of the solar system initial ratio (Iizuka et al. 2015) requiring that the zircon protoliths
had been removed from a chondritic uniform reservoir (CHUR) by 4.5 Ga. Harrison et al. (2005)
initially reported several Hadean Jack Hills zircons with positive εHf(T), but subsequent in situ
studies have not confirmed significantly positive values. This likely reflects complications arising
from the lack of simultaneous age and Hf isotope analysis, as described by Harrison et al. (2005).
The most robust aspect of this now large dataset is the cluster of results along a line
corresponding to a Lu/Hf ≈ 0.01, a value characteristic of continental crust. Such a low-Lu/Hf
reservoir at ~4 Ga is consistent with either early extraction of this very felsic crust or its
Figure 5. eHf(T) vs. 207Pb/206Pb

age of Jack Hills zircons.
Reference 176Lu/177Hf ratios
for continental crust (0.01) is
shown along with value for
Bulk Earth (0.034) and primor-
dial 176Hf/177Hf (i.e., Lu/Hf = 0).
These data are consistent with
the formation of continental
crust occurring essentially con-
tinuously since 4.5 Ga (modi-
fied from Bell et al. 2014).
generation by remelting of a primordial more basaltic reservoir, but in either case extrapolation
of this trend yields a present-day εHf(T) of approximately -100. This is substantially lower than
the most negative value yet measured (εHf(T) = −35; Guitreau et al. 2012). The lack of such a
signal suggests substantial recycling of crust into the mantle during the early Archean (Bell
et al. 2011, 2014). More specifically, zircons with εHf(T) consistent with continuing evolution
of this reservoir appear absent from the Jack Hills record after 3.7 Ga (Bell et al. 2014).
Combined with Hf isotopic evidence for juvenile mantle melts at ca. 3.9–3.7 Ga at both Jack
Hills (Bell et al. 2014) and the nearby Mt. Narryer site with similarly aged zircon (Nebel-
Jacobsen et al. 2010), these observations likely point to a recycling event ca. 3.9–3.7 Ga which
resembles the Hf isotopic evolution of modern subduction-related orogens (e.g., Collins et al.
2011) and so may have additional tectonic significance.
Plutonium–Xenon. The meteorite record reveals that 244Pu was present in the early solar
system with an initial Pu/U abundance of ~0.007 (Ozima and Podosek 2002). However, its
use as a geochemical tracer is restricted by its relatively short half-life (t½ = 82 Ma). As the
only known relics of the Earth’s earliest crust, analysis of Xe in Hadean zircons offers a way
to determine terrestrial Pu/U ratios and potentially investigate Pu geochemistry during early
crust forming events. Because these ancient zircons are detrital and of unknown provenance, it
is essential that individual grains be analyzed. Turner et al. (2004) discovered the first evidence
of extinct terrestrial 244Pu in individual 4.15–4.22 Ga Jack Hills zircons. These measurements
yielded initial Pu/U ratios ranging from chondritic (~0.007) to essentially zero. The latter
results were first interpreted to be due to Xe loss during later metamorphism. This assumption
was tested by irradiating 3.98–4.16 Ga zircons with thermal neutrons to generate Xe from 235U
neutron fission to determine Pu/U simultaneously with U–Xe apparent ages. Comparison of
U–Pb and U–Xe ages showed varying degrees of Xe loss, but about a third of the zircons yield
207
Pb/206Pb and U-Xe ages that are concordant within uncertainty (Turner et al. 2007).
Given that U becomes oxidized to the soluble uranyl ion (UO22+) under even mildly
oxidized aqueous conditions while the solubilities of essentially all Pu species are generally
much lower, variations in Pu/U has been suggested as a potential indicator of aqueous
alteration in the Jack Hills zircon protoliths (Harrison 2009). To test this hypothesis, Bell
(2013) collected a multivariate dataset on eleven zircons, including analysis of Xe isotopic
ratios, U–Pb age, trace element contents, and δ18O, to look for correlations (e.g., δ18O vs. Pu/U)
expected from aqueous processes. With the exception of Nd/U, none were found. High-Nd/U
zircons display only low Pu/U, while low Nd/U zircons show more heterogeneous Pu/U. The
high-Nd/U group appears less magmatically evolved than other Hadean zircons, has REE
patterns suggestive of some degree of alteration, either by hydrothermal fluid interaction
or phosphate replacement, and consists of solely low-Pu/U zircons with a range of Hadean
to Proterozoic U–Xe ages. The higher diversity of Pu/U among the rest of the population
may reflect more heterogeneous processes, including possible primary Pu/U variations from
a variety of processes that were not well-constrained. Thus the early promise that Pu/U
variations might record aqueous fractionation events in the Hadean may not be realized.
Lithium. δ7 Li analyses of Hadean Jack Hills zircons range from −19 to +13‰ (Ushikubo
et al. 2008). These authors interpreted highly negative values to reflect zircon crystallization
from a source that experienced intense weathering, thus placing the protolith at one time
at Earth’s surface. A limitation of this interpretation is that Li diffuses readily in zircons at
relatively low temperatures (Cherniak and Watson 2010) and thus could have exchanged with
hydrogen species during metamorphism (Trail et al. 2011b). Ushikubo et al. (2008) speculated
that Li migration might be limited by coupling with the very slow REE diffusion in zircons
thus limiting its geological transport rate. Recently Trail et al. (2016) examined just this
relationship and found no detectable link between Li and REE diffusion.
Inclusions in zircon
The plentiful mineral inclusions preserved in detrital zircons from the Jack Hills, western
Australia, have been the subject of several studies beginning with Maas et al. (1992) who
recognized their dominantly granitic character.
Muscovite. Hopkins et al. (2008, 2010) followed up the Maas et al. (1992) study by
examining > 1700 inclusion bearing zircons from Jack Hills. Their examination revealed that
quartz and muscovite are the principal inclusion phases, potentially pointing to aluminous granitic
sources; see example in Fig. 3). Hopkins et al. (2010) used a thermodynamic solution model for
celadonite substitution in muscovite (White et al. 2001) to estimate pressures for muscovite
inclusions in magmatic zircons. In all cases, pressures greater than 5 kbar (unsurprising given
the presence of magmatic muscovite) were obtained which, coupled with the relatively low host
zircons crystallization temperature (ca. 700 °C), implies remarkably low near surface heat flows
(≤ 80 mW/m2). This stands in stark contrast to previous model estimates of 160–400 mW/m2
(Smith 1981; Sleep 2000). By analogy to modern Earth, this led them to suggest formation in an
underthrust, or subduction-like, environment (Hopkins et al. 2008, 2010). However, the primary
nature of these inclusions was brought into question by Rasmussen et al. (2011), who surveyed
1000 Jack Hills zircons from 4.2 to 3.0 Ga and showed that some inclusions fall on cracks in
their host zircons and that phosphate inclusions generally record post-depositional U–Pb ages.
They suggested that much of the mineral inclusion record was due to secondary mineralization.
However, a closer look at the Jack Hills mineral inclusion record reveals complexities
not well explained by a largely secondary origin and argues for the preservation of many
primary inclusions. Inclusions that intersect cracks in their host zircons display a different
modal mineralogy than those isolated from cracks (Bell et al. 2015a). Muscovite inclusions
record a wide range of silica substitution with Si-per-formula-unit (12 oxygen basis) ranging
from 2.9 to 3.4, unlikely to all form from the same metamorphic fluid. The assemblage
that intersects cracks is roughly intermediate between the isolated assemblage and the
assemblage of secondary phases seen filling void space along cracks, probably showing partial
replacement (Bell et al. 2015a). The isolated and likely primary assemblage is muscovite-
dominated with abundant quartz, still suggestive of aluminous granitic protoliths, and minor
phases such as biotite, apatite, and feldspars vary in abundance with zircon age (Bell et al.
2015a). Certain phases present in the isolated assemblage and absent in the crack-intersecting
assemblage probably point to selective destruction of the minerals apatite and feldspar.
Because of the relatively low numbers of identified rare phases (e.g., aluminosilicates), it is
difficult at present to determine their significance for zircon provenance or for identifying the
nature of the altering fluids that invaded the zircons along cracks over geologic time.
In many instances where a sound case for a preserved primary inclusion assemblage can
be made, analyses cannot currently be effected due to size limitations. Indeed, only 6 of the
31 muscovites documented in the Hopkins et al. (2008) study could be reliably analyzed using
EMPA due to their small (< 2 mm on shortest dimension) size and the effects of secondary
fluorescence. Typically, the oldest zircons (> 4.2) contain the smallest white mica inclusions.
For example, we have identified a zircon as old as 4.34 Ga containing white mica that, except
for its size, is a candidate for thermobarometric analysis.
Fe oxides. The development of textural criteria for identifying primary inclusions (Bell
et al. 2015a) opens up possibilities for recognizing zircons’ changing provenance with time
and investigating their post-depositional alteration history. One intriguing aspect of zircon
provenance that could be further understood through the inclusion record is that of protolith
magma fO2 and its evolution. As described in more detail in the next section, Trail et al. (2011a)
demonstrated that the Ce anomaly of a zircon (Ce/Ce*) is a quantitative estimate for host magma
fO2, and furthermore that Hadean Jack Hills zircons show a range in fO2 with an average near the
fayalite–magnetite–quartz (FMQ) buffer, i.e., similar to the modern upper mantle. Granitoids
form at a range of fO2, controlled both by source region and assimilation of wall rock material
during ascent. Characteristic series of granites with contrasting fO2 in accretionary environments
are identified by their accessory Fe-Ti oxides, with more oxidized granites dominated by
magnetite and more reduced granites dominated by ilmenite (Ishihara 1977).
Fe-Ti oxides occur commonly as inclusions in zircon (Rasmussen et al. 2011) and appear to
be a robust if minor component of the Hadean primary assemblage (Bell et al. 2015a). Primary
Fe-oxide inclusions may preserve geomagnetic information and multiple groups are currently
investigating whether such signals are the oldest known records of a Hadean dynamo. Knowing
when the geodynamo arose potentially constrains the Earth’s early thermal structure and potential
for atmospheric loss, as well as when compositionally-driven core convection began. At present,
the oldest reliable determination of the terrestrial magnetic field is 3.45 Ga (Biggin et al. 2011).
Tarduno et al. (2015) interpreted Jack Hills zircons as containing magnetite inclusions
that retained primary remanent magnetization as old as 4.2 Ga (cf. Weiss et al. 2015) but
failed to demonstrate whether they had been remagnetized by thermal processes subsequent to
formation. Tarduno et al. (2015) argued that their zircons had not experienced high-temperature
metamorphism, as Pb would be redistributed in an inhomogeneous fashion at the nm-scale
(Valley et al. 2014). This process would result in non-systematic Pb/U variations during SIMS
depth profiling, which they did not observe. That view misrepresents their ion microprobe
capability in three ways: 1) the sputtering process mixes near surface atoms at the ~10 nm-scale,
2) the SHRIMP instrument they used cannot truly depth profile as sputtered atoms from both
crater bottom and surface are simultaneously accelerated into the mass spectrometer, and 3) the
10–20 µm diameter spot they used is three orders of magnitude larger than would be needed to
reveal such heterogeneities, even if they existed. As described below, Trail et al. (2016) suggested
that zircons exhibiting Li concentration heterogeneities, including oscillatory zoning, could be
calibrated in this role as a peak temperature geothermometer for paleomagnetic studies.
Biotite. Biotite inclusions in magmatic zircon are relatively common (Rasmussen et al.
2011) and appear to reflect the composition of biotite in the host (Jennings et al. 2011), which
varies considerably among granitoids (e.g., Buda et al. 2004; Abdel-Rahman 1994). Biotite
shows characteristic variations in FeO, MgO, and Al2O3 contents that can discriminate among
calc-alkaline, peraluminous, and alkaline anorogenic granitoids (Abdel-Rahman 1994). Thus,

identifying and analyzing primary Hadean biotite inclusions could better constrain the nature
of Hadean melt compositions that may have tectonic implications. In addition, rare sulfide
(Mojzsis 2007) and carbonaceous (see next section) phases have also been identified in Hadean
zircons. A systematic survey for these and other rare phases will further illuminate the volatile
contents of Hadean magmas and their source materials.
Graphite. A key challenge in pondering the existence of life elsewhere is that we know
of only one occurrence. While Earth is the only planet on which life is known to have
emerged, we remain largely ignorant of the conditions, timing and mechanisms by which
this occurred. A broad array of morphological and isotopic evidence supports the view that
by 3.8 to 3.5 billion years (Ga) ago our planet hosted microbiota, including some with
relatively sophisticated metabolisms (e.g., Mojzsis et al. 1996; Rosing 1999; McKeegan et
al. 2007; Schopf 2014; Brasier et al. 2015).
As noted earlier, geochemical studies of Hadean Jack Hills zircons have led several
authors to suggest relatively clement conditions on earliest Earth (e.g., Mozjsis et al. 2001;
Wilde et al. 2001; Harrison 2009). This leaves open the possibility that our planet became
habitable, and life emerged, during the first 500 million years of Earth history. Knowing when
and under what conditions life emerged could tell us a great deal about the likelihood of life
elsewhere. Were conditions clement or hellacious? Did life emerge virtually immediately or
only after a half billion years of planetary preparation?
Thus reports of abundant diamond and graphite inclusions in the Jack Hills zircons (4%
of each in the zircons investigated) and the spectrum of light carbon isotopic compositions
they contained (Menneken et al. 2007; Nemchin et al. 2008) was met with both excitement and
skepticism; the latter reflecting the seeming inconsistency of the presence of diamonds with the
many inferences drawn from other zircon inclusions (e.g., their derivation from crustal melts;
Mojzsis et al. 2001; Peck et al. 2001; Watson and Harrison 2005; Hopkins et al. 2010; Bell et al.
2015). Recently, the diamonds were shown definitively to be contamination from the polishing
compound that was used during sample preparation (Dobrzhinetskaya et al. 2014). The origin
of the graphite was less certain but deemed also likely due to contamination. This left the true
occurrence rate and nature of carbonaceous materials in the Jack Hills zircons uncertain.
Bell et al. (2015b) optically examined a large number of > 3.8 Ga Jack Hills zircons and
found ~25% contain opaque inclusions. Imaging these selected grains by Raman spectroscopy
revealed two isolated carbonaceous inclusions in a concordant, 4.10 Ga zircon (RSES 61-18.8;
Fig. 6). To ensure that these inclusions were never in contact with the laboratory environment
prior to structural and isotopic analyses, Bell et al. (2015b) extracted a ~160 ng sliver of the
zircon containing the two carbonaceous phases via focused ion beam milling and examined
it using X-ray nanotomography (Fig. 6). The 40 nm spatial resolution of this imaging method
revealed no cracks associated with the graphite inclusions. Their isolation within the zircon
crystal and from cracks indicated a primary origin. Carbon isotopic measurements using SIMS
yielded an average δ13CPDB of −24 ± 5‰. As carbon isotopic fractionation between gaseous
and condensed species in magmas is expected to be relatively small (e.g., ≤ 4‰; Javoy et al.
1978), such a low δ13C value is consistent with a biogenic origin. While there are possible
inorganic mechanisms that could also produce such a signal, they require what we see as
an unlikely chain of geologic events (Bell et al. 2015b). Alternatively, House (2015) offered
the “wild” suggestion that a high carbon content in Earth’s core could have resulted in an
initially highly 13C-depleted mantle. If the Bell et al. (2015b) result does indeed represent an
isotopic signal of biologic activity, it would extend our knowledge of the timing of terrestrial
life back to at least 4.1 Ga, or ≥ 300 Ma earlier than the previously suggested and coincident
with estimates derived from molecular divergence among prokaryotes (Battistuzzi et al. 2004).
Figure 6. High resolution transmission X-ray im-

age of RSES 61-18.8 with arrows pointing to the two
graphite inclusions analyzed for carbon isotopes (Bell
et al. 2015b).
Reports of graphite in S-type granites are relatively rare but cases have been documented
in which it was inherited from the source (Seifert at al. 2010; Zeng et al. 2001), incorporated via
wallrock assimilation (Duke and Rumble 1986), or precipitated during subsolidus interactions
with CO2 (Frezzotti et al. 1994; also see Carroll and Wyllie 1989). Graphite inclusions have
been reported in metamorphic zircon (Song et al. 2005) but, to our knowledge, ours is the
first documented case of primary graphite in magmatic zircon. Given the relative paucity of
investigations of zircon inclusion populations, it is difficult to know whether this reflects their
low abundance or simply the lack of a concerted search.
The oxygen fugacity over which graphite can be stable in a granitic magma depends on
H2O and H2 activities (Ohmoto and Kerrick 1977), but relatively reducing redox conditions
(i.e., below FMQ) are implied.
Zircon geochemistry
Titanium. Because the abundance of a trace element partitioned between mineral and
melt is temperature dependent, crystallization temperatures can in principle be estimated
from knowledge of the concentration of that element in the solid phase if the magma is
appropriately buffered. The advent of the Ti-in-zircon thermometer permitted zircon
crystallization temperatures to be assessed provided the activities of quartz and rutile can
be estimated (Watson and Harrison 2005; Watson et al. 2006; Ferry and Watson 2007). The
diffusion of Ti in zircon is vanishingly slow under crustal conditions (Cherniak and Watson
2007) and thus the potential for re-equilibration of the thermometer is very low. In the case
in which zircon co-exists with both quartz and rutile (i.e., aSiO2 ≈ aTiO2 ≈ 1), an accurate and
precise temperature (i.e., ±15 °C) can routinely be determined.
The first application of the Ti-in-zircon thermometer was to Hadean zircons from Jack
Hills. Watson and Harrison (2005) measured Ti in zircons ranging from 3.91 to 4.35 Ga,
and the vast majority of these plotting in a normal distribution. Excluding high temperature
outliers yielded an average temperature of 680 ± 25 °C (data shown in Fig. 7). However, a
limitation in applying this thermometer to detrital zircons is the unknown aTiO2 of the parent
magma. In the case of the zircon shown in Fig. 3, which contains both primary quartz and
rutile, an accurate crystallization temperature is expected. However, unless co-crystallization
with rutile is known, the calculated temperature it is a minimum estimate. In the absence of
rutile inclusions, Watson and Harrison (2005) argued that aTiO2 is largely restricted to between
Jack Hills (Harrison, 2009)

Model (Wielicki et al. (2012)
Impacts (Wielicki et al. (2012)
Iceland (Carley et al., 2011)
Relative probability
SIMS impacts (Kenny et al., 2016)
600 650 700 750 800 850 900

o
Temperature ( C)
Figure 7. Probability plot of apparent zircon crystallization temperature comparing Hadean data (blue)
with data for Icelandic (magenta) and impact formed (black) zircons. The dashed curve shows the
distribution predicted by a model incorporating impact thermal effects, continental rock chemistry, and
zircon saturation behavior (Modified from Wielicki et al. 2012).
~0.5 and 1 in continental igneous rocks as the general nature of evolving magmas leads to
high aTiO2 prior to zircon saturation. Thus for Hadean zircons of magmatic origin, it would
be a rare case in which zircon formed in the absence of a Ti-rich phase (e.g., rutile, ilmenite,
titanite), thus generally restricting aTiO2 to ≥ 0.5. In case of aTiO2≈ 0.5, calculated temperatures

in the range 650–700 °C would be underestimated by 40–50 °C, although this is entirely

compensated for if aSiO2 = aTiO2 (Ferry and Watson 2007). Hofmann et al. (2009) inferred that
enhanced Ti contents could be incorporated during non-equilibrium crystallization resulting in
higher than actual calculated temperatures. If this effect were significant in the generation of
granitic Jack Hills zircons this would further support their low-temperature origin.
While it is widely acknowledged that water saturation in intracrustal magmas is rare and
that the vast majority of intermediate to siliceous magmas form by dehydration melting under
vapor absent conditions (Clemens 1984), Watson and Harrison (2005) concluded that the tight
cluster of Hadean zircon crystallization temperatures at 680 ± 25 °C (Fig. 7) reflects prograde
melting under conditions at or near water saturation. They arrived at this interpretation because
prograde, vapor-absent melting of metapelites and orthogneisses containing typical crustal Zr
concentrations (i.e., 150–200 ppm; Harrison et al. 2007) at 5 to 10 kbar would be expected
to record zircon crystallization temperature peaks corresponding to the relevant dehydration
melting equilibria (e.g., muscovite at ca. 740 °C, biotite at ca. 770–800 °C, amphibole at
≥ 800 °C; Spear 1993). Thus, for example, Hamilton’s (2007) assertion that Hadean zircons
were derived solely through melting resulting from hornblende breakdown is fundamentally
inconsistent with all thermometric results to date.
Rock porosities in the middle and deep crust are typically < 0.1% (Ingebritsen and Manning
2002) and thus < 0.03 wt.% free H2O is available to flux melting. During metamorphism, water
is progressively lost from rocks via discontinuous, subsolidus dehydration reactions through
the greenschist and amphibolite facies (Spear 1993). Structural water is stored in hydrous
minerals (e.g., ~4% in muscovite, ~3% in biotite, ~2% in hornblende, ~2–4% in altered basalt at
greenschist facies; Clemens and Vielzeuf 1987; Franzson et al. 2010). The correspondingly low
water contents of pelitic (~1.2%) and quartzofeldspathic rocks (~0.6%) are expected to produce
only small amounts of melt at temperatures close to 700 °C (Clemens and Vielzeuf 1987). Figure 8
(White et al. 2001) underscores the limited melting potential of a metapelite (represented by the
Na + CaO–K2O–FeO–MgO–Al2O3–SiO2–H2O system) for the temperature range (655–705 °C)
Figure 8. P–T pseudosection for a model pelite in the NCKFMASH system containing added 20 mol%
added H2O (modified from White et al. 2001). Note that even in the presence of this free water, the P–T
region populated by Hadean zircons would result in essentially melt-free conditions indicating that very
high water contents (> 9 wt%; Burnham 1975) would be required to create significant, mobile magmas.
and pressures (> 6 kbars) inferred for Hadean Jack Hills zircons which fall below the “effective
solidus” melt fraction of 0.03 (even in the presence of 20 mol% added H2O). That is, under vapor
absent conditions at the pressure–temperature range documented for Hadean zircons (Hopkins et
al. 2010), both pelitic and quartzofeldspathic source rocks would be effectively melt free.
Concluding that Hadean Jack Hills zircons largely formed under water-saturated
conditions sharply limits the possible tectonic settings in which they formed. The key issue is
that silicate magmas at pressures above 6 kbar dissolve much more H2O than is available in
rocks (up to 70 mol% at 10 kbar; Burnham 1975; Clemens 1984) and thus requires an external
(e.g., dehydrating underthrust sediments) source of water for saturation to be achieved.
Rare earths. As previously noted, the abundance ratio of Ce in zircon relative to that
interpolated from the light rare earth pattern (Ce/Ce*) has been developed as a quantitative
estimate for host magma fO2 (Trail et al. 2011a). Most Hadean Jack Hills zircons are within
error of FMQ (similar present-day upper mantle) but range as low as IW suggesting a
diversity of source materials (Trail et al. 2011a). Another, qualitative estimate for magma fO2
involves the mineralogy of Fe-Ti oxide phases. Ishihara (1977) observed that both oxidized
and reduced series of granitoids occur in accretionary settings, with the oxide mineralogy
of high fO2 granites dominated by magnetite and that of the reduced granites dominated by
ilmenite. Since oxide inclusions are often a minor constituent of igneous zircon inclusion
suites (e.g., Rasmussen et al. 2011), the coupled investigation of oxide inclusion mineralogy
with Ce/Ce* in the host zircon provides the ability to check for internal consistency between
these two estimates of Hadean magma redox conditions.
It will be helpful to establish both the characteristic ranges of zircon Ce/Ce* for magnetite
vs. ilmenite series granitoids and whether the oxide mineralogy of the whole rock is accurately
reflected by the mineralogy included in zircons. Reconnaissance EDS analysis (Hopkins et al.
2010; reported by Bell et al. 2015a) suggests that the Hadean opaque inclusions are dominated
by Fe oxides, potentially magnetite. Further investigation of the mineralogy of Hadean oxide
inclusions, coupled with their host zircon Ce/Ce*, may help to better classify the granitoids
they derive from or potentially to diagnose alteration affecting the inclusions or host zircon.
This coupled approach will better illuminate the redox conditions in the Hadean crust and any
potential complexities that igneous zircon may record.
However, petrologically important trace element characteristics such as Ce/Ce* and
Ti content can be obscured by alteration or contamination (e.g., cracks, inclusions, etc.).
Hydrothermal alteration of zircon is often diagnosed by a high, flat light rare earth element
(LREE) pattern. Among Jack Hills zircons, such alteration is dominantly characterized by
anomalously high Ti, Fe, P, U, and LREE contents (Bell et al. 2016). To remediate this issue,
Bell et al. (2016) developed a trace element indicator (i.e., the LREE-Index; Fig. 9) which
permits altered and hydrothermal zircons to be clearly identified.
Lithium. As noted earlier, Trail et al. (2016) proposed the use of Li zoning in zircon
as a peak temperature indicator, particularly for use in ascertaining the retention of primary
remanent magnetic signals. Figure 9 shows a direct ion image of 7Li+ of the surface of a
sectioned 4.02 Ga Jack Hills zircon containing a ~5-μm-wide Li concentration band (Fig. 10).
The general preservation of this band requires that peak heating temperature(s) for this
detrital zircon did not exceed ~500 °C for million-year timescales. Thus this grain did not
exceed the Curie temperature for magnetite of 585 °C and would be a viable candidate for
study of primary magnetism. Of course each detrital zircon in a population may have a
different pre-depositional thermal history, but this result indicates that, post deposition, the
metaconglomerates at Erawondoo Hill have not experienced temperatures greater than 500 °C,
consistent with other thermometric determinations (Rasmussen et al. 2010).
Figure 9. A) Typical zircon trace element pattern vs. that for metamorphic xenotime in Jack Hills discovery
site quartzite and average upper continental crust. Most potential contaminating materials would have higher
LREE and/or a higher LREE/HREE slope than primary magmatic zircon. B) Definition of the Light Rare
Earth Element Index (LREE-I), based on the ratio of the LREE (represented by Nd and Sm) to the MREE
(represented by Dy). Reprinted from Bell et al. (2016) Recovering the primary geochemistry of Jack Hills zir-
cons through quantitative estimates of chemical alteration Geochimica et Cosmochimica Acta 191:287–292.,
with permission from Elsevier.
Aluminum. Trail et al. (2017) found that zircons from peraluminous granitoids contain
average Al concentrations of ~10 ppm (with a range from 0 to 23 ppm), in contrast to I-
and A-type zircons, which average ~1.3 ppm. Although alumina activity appears not to be
Figure 10. Li-in-zircon geospeedometry ap-

plied to a 4.03 Ga Jack Hills zircon. (Left)
CL image showing growth zoning. (Right)
Li+ image in boxed region at left. The 5 µm
scale of Li banding requires peak heating
temperatures to have never exceed 500 °C
for Ma-timescales. This constrains the Jack
Hills Hadean zircons to have never been
heated above the magnetite Curie point since
4.03 Ga. Reprinted from Trail (2016) Li zon-
ing in zircon as a potential geospeedometer
and peak temperature indicator. Contribu-
tions to Mineralogy and Petrology 171:1–15,
with permission from Springer.
a simple function of the degree of the peraluminosity, zircon Al concentration could be

calibrated as a proxy for melt Al2O3 / (CaO + Na2O + K2O) where molar values > 1 reflect
an origin from recycled pelitic material (Trail et al. 2017). They applied this approach to
Hadean Jack Hills zircons and found both metalunimous and peraluminous sources, albeit
with the former apparently dominating the population. Although the scarcity of high Al
contents from Hadean zircons suggests that metaluminous crustal rocks may have been
more common than peraluminous rocks in the Hadean, the ~20% overlap of low Al (i.e.,
< 5 ppm) in S-type zircons somewhat obscures this inference.
Carbon. As carbon is long known to dissolve in silicates at trace levels (e.g., Freund et
al. 1980; Oberheuser et al. 1983; Mathez et al. 1984; Tingle et al. 1988; Keppler et al. 2003;
Rosenthal et al. 2015), the coexistence of zircon and graphite raises the possibility that C could
be present at measurable levels in zircon. As SIMS has the potential for detection levels of C
as low as ~1 ppb, it is ideally suited for such a search.
The lack of dependency of carbon solubility in silicates on oxygen fugacity suggested
to Shcheka et al. (2006) that C4+ substitutes for Si4+, with increased levels as the volume
of the SiO4 tetrahedron decreases. In this regard, they emphasized that the relatively small
volume of the SiO4 tetrahedron in zircon should enhance carbon solubility. Alternatively, Sen
et al. (2013) found evidence that C sbustitutes for nonbridging oxygen in synthesized silicate
nanodomains. As such they hypothesized that trace carbon could be incorporated into silicates
across a broader range of fO2 than previously thought and speculated that this incorporation
mechanism might have been preferentially important during the Hadean eon.
In the same way that a zircon co-crystallizing with rutile contains a predictable temperature-
dependent Ti concentration, zircons growing in the presence of a carbonaceous species appear to
partition C in a fashion that could be calibrated as a magma volatile probe. Having an approach
with which to detect Hadean crustal C could potentially reconcile the disparate views regarding
the magnitude of carbon in the crust during that eon. Some authors argued for a net increase
in crustal carbon from essentially zero at 4 Ga (e.g., Hayes and Waldbauer 2006; Kelemen and
Manning 2015) to its present day inventory in a broadly linear fashion. Marty et al. (2013)
envisioned an essentially continuous transfer of carbon from undegassed mantle reservoirs
implying a net increase to the crust over time. In contrast, Dasgupta (2013) advocated for higher
than present day concentration on early Earth. The development of a proxy to detect the presence
of carbon in Hadean (and younger) melts may eventually permit selection among these models.
OTHER WESTERN AUSTRALIAN HADEAN ZIRCON OCCURRENCES

Mt. Narryer
Ion microprobe dating of detrital zircons from several quartzites at Mt. Narryer in Western
Australia (Fig. 11) have revealed a minor Hadean component ranging from 2% (Froude et al. 1983)
to 12% grains > 4.0 Ga (Pidgeon and Nemchin 2006), with younger zircons ranging to ca. 3 Ga. A
LA-ICP-MS study of Mt. Narryer zircons of all ages suggested that they generally display higher
U contents and lower Ce/Ce* than Jack Hills zircons (Crowley et al. 2005). Our preliminary ion
microprobe data for 80 zircons between ca. 3 and 3.75 Ga in age from two Mt. Narryer quartzites
suggests that zircons with unaltered magmatic chemistry (i.e., via the LREE-Index) do indeed
show slightly higher U contents and lower Ce/Ce* than Jack Hills zircons of similar age, although
there is significant overlap between the populations. However, crystallization temperatures for
these zircons is higher than at Jack Hills, averaging ~750 ± 50ºC. All but one of our studied
zircons has Th/U > 0.2, indicative of magmatic origins. Calculated fO2 for these zircons suggests
values on average several log units below the FMQ buffer, which overlaps the range of many
less-oxidized Hadean Jack Hills zircons. Although we have not yet identified > 4 Ga zircons from
Mt. Narryer in our preliminary survey, these differences in chemistry likely point to a diversity of
Eoarchean-Hadean source rocks represented in Western Australia, as also suggested by Crowley
et al. (2005) and by Hf isotopic compositions (Nebel-Jacobsen et al. 2010).
Churla Wells
Ion microprobe dating of a zircon from an orthogneiss from near Churla Well,~25 km
west of the Mt. Narryer site (Fig. 11), yielded grains with 207Pb/206Pb ages of 4.14 to 4.18 Ga
(Nelson et al. 2000). Electron microprobe traverses show that the core containing the oldest
ages has much lower Hf, REE, U and Th than the outer regions. Nonetheless, several
observations—U contents in the core ranging up to 666 ppm, Th/U as high as 0.6, and trace
element concentrations and ratios—strongly suggest its origin in a granitic magma.
Maynard Hills
In the Southern Cross Granite–Greenstone Terrane, Western Australia (Fig. 11), ion
microprobe dating of a single zircon from a quartzite within the Maynard Hills greenstone belt
(Wyche 2007) yielded a mean 207Pb/206Pb age of 4.35 ± 0.01 Ga.
Figure 11. Location map of the 13 sites from which > 4 Ga zircons have been documented.
Mt Alfred
At the Mt. Alfred locality of the Illaara Greenstone Belt further along strike, Nelson (2005)
documented a concordant zircon with an age of 4.17 ± 0.01 Ga. Thern and Nelson (2012)
reported three additional Hadean zircon ages from this sample ranging from 4.23 to 4.34 Ga.
To our knowledge, no geochemistry for Hadean zircons from this sample have been published.
NORTH AMERICAN HADEAN ZIRCON OCCURRENCES

Northwest Territory, Canada
The Acasta tonalite orthogneiss from the Western Slave craton (Fig. 11) yields a range
of U–Pb zircon ages interpreted to date protolith crystallization at 3.96 Ga (Bowring and
Williams 1999; Stern and Bleeker 1998; Mojzsis et al. 2014). However, Iizuka et al. (2006)
documented a 4.20 ± 0.06 Ga zircon grain using LA-ICP-MS. This apparent xenocryst has a
LREE pattern (Bell et al. 2016) within the field associated with unaltered zircon. Its Th/U
suggests a magmatic origin and, along with other trace element concentrations and ratios,
derivation from a felsic melt by a process other than differentiation of a mafic magma.
Pronounced Ce and Eu anomalies correspond, respectively, to an fO2 close to the FMQ buffer
(assuming a crystallization temperature of 750 °C) and a crustal, as opposed to mantle, origin.
Greenland
Detailed ion microprobe dating of a tonalitic orthogneiss from Akilia Island, West
Greenland (Fig. 11), that crosscuts the oldest known marine sediment (Manning et al. 2006)
established a crystallization age of 3.83 ± 0.01 Ga (Mojzsis and Harrison 2002). A U–Pb survey
of zircons identified in thin section documented a single zircon, concordant within uncertainty,
with an age of 4.08 ± 0.02 Ga.
ASIAN HADEAN ZIRCON OCCURRENCES

Tibet
Duo et al. (2007) report a 4.1 Ga ion microprobe age for a zircon from a quartzite in
Buring County, western Tibet (Fig. 11). Th/U ratios greater than 0.7 suggest a magmatic origin
for this detrital grain.
North Qinling
Wang et al. (2007) reported a LA-ICP-MS age of 4.08 ± 0.01 Ga for a xenocrystic zircon
from Ordovician volcanics of the Caotangou Group, North Qinling Orogenic Belt (Fig. 11).
Subsequent ion microprobe and LA-ICP-MS analyses identified additional Hadean grains with
ages ranging from 4.03 to 4.08 Ga (Diwu et al. 2010, 2013). Hafnium isotope analyses of these
grains are consistent with origin in crust extracted between 4.0 to 4.4 Ga (Diwu et al. 2013).
North China Craton
Cui et al. (2013) reported a LA-ICP-MS U–Pb date of 4.17 ± 0.05 Ga, concordant
within uncertainly, for a xenocrystic zircon from the Anshan–Benxi Archaean supracrustal
greenstone belt (Fig. 11). Correction of common Pb was made using 208Pb and assumed
concordancy between the U–Pb and Th–Pb systems. The zircon was separated from fine-
grained amphibolites intruded into banded iron formation and bedded coarse-grained
amphibolites. Its Th/U of 0.46 suggests a magmatic origin.
Southern China
Using ion microprobe U–Pb dating, two Hadean detrital zircons were documented from
a quartzite within Neoproterozoic metasediments from the Cathaysia Block in southwestern
Zhejiang (Fig. 11; Xing et al. 2014). One zircon core yielded a 207Pb/206Pb age of 4.13 ± 0.01 Ga
with a δ18O = 5.9 ± 0.1‰. The other zircon grain has a 4.12 ± 0.01 Ga magmatic core, a δ18O
of 7.2 ± 0.2‰, a positive Ce anomaly indicative of highly oxidizing conditions, and a high
apparent Ti-in-zircon crystallization temperature of 910 °C. While the authors interpreted these
results to suggest the zircons originated via dry melting of oxidized and hydrothermally altered
supracrustal rocks, closer examination of the high apparent Ti content may be warranted due to
Ti contamination effects (Harrison and Schmitt 2007). Trace element discrimination diagrams
and REE patterns place the zircon core within the continentally-derived field.
SOUTH AMERICAN HADEAN ZIRCON OCCURRENCES

Southern Guyana
A xenocrystic zircon from a felsic volcanic unit of the Iwokrama Formation, Guyana
Shield (Fig. 11), yielded a concordant LA-ICP-MS U–Pb age of 4.22 ± 0.02 (Nadeau et al.
2013). No other geochemical analyses of this zircon have been reported.
Eastern Brazil
The Archean core of the São Francisco Craton, northeastern Brazil (Fig. 11), contains
meta-volcanosedimentary supracrustal rocks including the Ibitira–Ubirac greenstone belt.
Paquette et al. (2015) analyzed a zircon from an amphibolite facies pelite from this belt by LA-
ICP-MS yielding a distribution of U–Pb ages that intersected concordia at 4.22 ± 0.02 Ga (four
207
Pb/206Pb ages > 4.01 Ga). The core Th/U ratios of 0.8 and high U contents (up to 1400 ppm)
suggest a felsic magmatic origin of this probably detrital grain.
The above discussion is complicated by limitations comparing data generated by different
analytical methodologies. For example, the LA-ICP-MS approach lacks the capacity to
measure common 204Pb and thus the possibility exists that single zircon occurrences with
apparent ages of > 4 Ga could be due to inclusion of non-radiogenic Pb.
OTHER PROPOSED MECHANISMS FOR FORMING

HADEAN JACK HILLS ZIRCONS
The collective data obtained over the last 15 years broadly support an origin of many
Hadean zircons as crystallizing from relatively cool, relatively wet felsic melts sourced at least
in part from sedimentary protoliths at a plate boundary. Nonetheless, numerous other models
have been proposed, and we evaluate the consistency of the data with these hypotheses.
Icelandic rhyolites
Iceland’s unusual geochemical character and thick basaltic crust couple to produce an
unusually high proportion (~10%) of silicic magmatism. These rocks in turn host abundant
zircon making Iceland a seemingly attractive model to explain the production of Hadean
zircons (Taylor and McLennan 1985; Galer and Goldstein 1991; Valley et al. 2002).
However, comprehensive investigations of the trace element and oxygen isotope composition
of Icelandic zircons show that the two populations to be distinctively different (Carley et
al. 2011, 2014; cf. Reimink et al. 2014). In contrast to the elevated 18O signature in some
Hadean zircons, Icelandic zircons are characterized by 18O-depleted values, likely the result of
recycling of altered basaltic crust rather than direct melting of sedimentary rocks (Bindeman
et al. 2012). Zircon crystallization temperatures are similarly different, with Icelandic zircons
yielding an average of 780 °C compared to the 680 °C average of Jack Hills Hadean zircons.
As noted in the previous section, higher temperature data from the Mt. Narryer and
southwestern Zhejiang sites may indicate their formation in different environments relative to
the Jack Hills population, possibly similar to an Iceland-like source.
Intermediate igneous rocks
Several authors have argued that the low temperature Hadean peak could reflect zircon
saturation at low temperatures in rocks of the tonalite-trondhjemite-granodiorite (TTG) suite
(Glikson 2006; Nutman 2006). Glikson (2006) proposed that Hadean zircons could have
originated in TTG’s that formed at high-temperatures but did not crystallize zircon until near
eutectic temperatures were reached. Similarly, Nutman (2006) argued on the basis of calculated
saturation temperatures (Watson and Harrison 1983) for TTG’s that high-temperature melts
do not crystallize zircon until they cool to temperatures near that of minimum melting
(i.e., zircons from both wet tonalite and minimum melts yield similarly low crystallization
temperatures). However, Harrison et al. (2007) showed bulk rock saturation thermometry
to be inapplicable to zircons crystallizing from TTGs. Rather, a cooling magma system first
crystallizes modally abundant phases, increasing the Zr concentration in the residual melt
while moving the melt towards compositions with much lower capacities to dissolve zircon.
Thus temperatures calculated from bulk rock chemistry significantly underestimate the onset
temperature of zircon crystallization in TTGs. The expected and observed result is that zircons
crystallized from TTG melts (as documented by electron imaging studies) yield significantly
higher temperatures than seen in the Hadean Jack Hills population (e.g., Harrison et al. 2007).
Mafic igneous rocks
A variety of authors have suggested that the > 4 Ga zircon temperature distribution could
be derived from zircons originating in mafic magmas (Coogan and Hinton 2006; Valley et
al. 2006; Rollinson 2008). However, zircon formation temperatures in these environments
are significantly higher (> 750 °C) than the Hadean peak (e.g., Harrison et al. 2007;
Hellebrand et al. 2007). As a case in point, Rollinson (2008) argued that the δ18O and trace
element signatures in Hadean Jack Hills zircons were consistent with an origin in ophiolitic
trondhjemites rather than continental crust. The author pointed to water-saturated, low
pressure melting experiments on oceanic gabbros at > 900 °C that yielded trondhjemitic melts.
While the origin of the excess water is potentially explicable in this scenario, the origin of
muscovite, a mineral uncharacteristic of trondhjemite but the most common inclusion in Jack
Hills Hadean zircons (Hopkins et al. 2008), was not addressed. Although we noted earlier the
clear separation between Hadean and MORB zircons on a plot of U/Yb vs. Y plot (Grimes et
al. 2007), Rollinson (2008) argued that data on such discrimination diagrams showed a ~20%
overlap and were thus permissive of such an origin. While this is true when plotting present
U concentrations, the separation becomes essentially complete once an appropriate correction
for U decay has been made (e.g., a 4.3 Ga zircon presently containing 100 ppm U originally
crystallized with 244 ppm U).
As noted earlier, zircons derived from a wide range of mafic rocks yield much higher
average temperatures (~770 °C; Valley et al. 2006; Fu et al. 2008) than the Hadean population
(Harrison et al. 2007). In the absence of a natural selection mechanism that preferentially
excludes zircons formed at high temperature (the opposite of what is expected from
preservation effects on high radioactivity zircons), intermediate to mafic sources are unlikely
to have contributed significantly to the Hadean Jack Hills population. However, sparse, higher
temperature zircons from the Mt. Narryer and southwestern Zhejiang locations may well be
consistent with such an origin and could be tested by studies of their inclusion populations.
Sagduction
A feature of modern plate tectonics is that oceanic lithosphere older than about 20 Ma is
negatively buoyant and thus can be underthrust beneath adjacent plates. At 80–120 km depths,
the basaltic crust undergoes a transformation to much denser eclogite and the resulting pull
on the downgoing slab provides a first-order contribution to the global plate tectonic energy
budget. In the hotter mantle of early Earth, it is assumed that oceanic lithosphere would have
been thicker and thus may not have been able to achieve the neutral buoyancy required for
subduction making plate tectonic-type behavior uncertain (Davies 1992; cf., Korenaga 2013).
Under such conditions, a low apparent geotherm could be achieved locally where thermally and/
or compositionally dense crust sinks into the mantle as downward moving drips (sagduction;
Macgregor 1951). While this can insulate the descending mass from reaching melting temperature
until high pressures are attained (e.g., Davies 1992), more nuanced scenarios are also possible
(e.g., François et al. 2014). Such a mechanism was invoked by several authors (Williams 2007;
Nemchin et al. 2008) to explain the anomalously low (< 10 °C/km) geotherms required by the
apparent occurrence of diamond in Hadean Jack Hills zircons, although recognition that the
diamonds were contamination (Dobrzhinetskaya et al. 2014) obviated the need for such models.
The sagduction model shares similar limitations to those discussed above, the source of the
needed water and, in the case of blocks delaminated into the mantle, the lack of a mechanism
to return zircons formed by this mechanism to the surface. Consider the case of a sagducting
block of mafic eclogite. As noted earlier, below the brittle-ductile transition rock porosities are
typically < 0.1% (Ingebritsen and Manning 2002). Structural water stored in hydrous minerals is
limited to ≤ 2% of virtually all rock types and is lost progressively via discontinuous, subsolidus
dehydration reactions through the greenschist and amphibolite facies (Spear 1993).
Any water liberated by dehydration is likely to ascend from the sagducting drip into colder,
overlying rocks. Thus fusion is likely to be forestalled until temperatures greatly exceeding
that of minimum melting are reached. In the case of complete devolatilization, temperatures of
> 900 °C would be required for melting of dry rock. As noted earlier, the absence of peaks in the
Hadean zircon crystallization spectrum corresponding to dehydration melting does not support
such a mechanism and such melts are unlikely to be characterized by quartz and muscovite
inclusions. Even the most appealing scenario involving eclogitized pelite containing a 50:50
mixture of muscovite and quartz contains only ~2 wt.% water and vapour absent melting of such
a protolith produces highly water-undersaturated melts (e.g., Patiño Douce and Harris 1998).
Sagduction models lack a mechanism to introduce water-rich fluids into fertile source
rocks capable of yielding both peraluminous and metaluminous magmas at temperatures
close to minimum melting (as required by Ti thermometry) and then sustain the supply
of water until the rock’s melt fertility is essentially exhausted (thus resulting in the single
Hadean zircon peak at ca. 680 °C). The twofold appeal of a plate boundary environment is
the continuous source of water available in the hangingwall of a submarine underthrust and
the potential for long-term (i.e., > 4 Ga) preservation of any zircons created by water-fluxed
melting. In contrast, how zircons formed during dehydration melting in a block sagducting
into the mantle will reappear to be preserved at the Earth’s surface is unclear.
Impact melts
Given the likelihood of high bolide fluxes to early Earth, the potential for impact melts
to be a source of Hadean zircons requires investigation. Studies of neo-formed zircon in
preserved terrestrial basins large enough to have created melt sheets (e.g., Sudbury, Morokweng,
Manicouagan, Vredefort) show that their crystallization temperatures average more than 100 °C
greater than that of > 4 Ga Jack Hills zircons and thus impacts do not represent a dominant
source for that Hadean population (Darling et al. 2009; Wielicki et al. 2012). This observation
is supported by models that relate expected impact thermal anomalies with early crustal rock
chemistry (Wielicki et al. 2012; Fig. 7), which confirm that observations from the handful of
known impact melt sheets is indeed globally representative. Recently, Kenny et al. (2016) argued
that zircon crystallization temperatures for the granophyre layer at the Sudbury impact crater had
been underemphasized and proposed this rock type as a source of at least a portion of Hadean
Jack Hills zircons (Fig. 7). They raised the prospect of an unspecified selection process that had
preferentially destroyed high temperature Hadean zircons and thus biased the detrital record
to low temperatures. Nature does tend to bias the detrital zircon record but that mechanism
operates in exactly the opposite sense. Late crystallizing, thus low temperature, granitoid zircons
are known to contain elevated U and Th concentrations which lead to metamictization (Claiborne
et al. 2010) and thus they are more likely lost from the detrital record, resulting in preferential
preservation of higher temperature zircons (Harrison and Schmitt 2007). Wielicki et al. (2016)
tested the Kenny et al. (2016) hypothesis statistically and showed that the probability of extracting
the Hadean Jack Hills Ti-in-zircon temperature distribution from their data, or any permutation
of the published dataset of impact-produced zircons, is vanishingly small.
Marchi et al. (2014) proposed that an intense bombardment event at ~4.1 Ga covered the
planet with ca. 20 km of flood basalt in which the Hadean Jack Hills zircons were formed.
In brief, this model is fundamentally incompatible with virtually every geochemical record
obtained from that population (i.e., hydrous melting conditions, low geotherm, peraluminous
compositions, etc.).
The above statements are specifically relevant to the Hadean Jack Hills zircons and
conclusions should be tempered by the reconnaissance-scale data obtained for samples from Mt.
Narryer and southwestern Zhejiang, which yield apparent zircon crystallization temperatures
of 750 and 910 °C, respectively. While both results are subject to potential Ti contamination
effects, developing a strong geochemical database from the dozen or so locations for which
> 4 Ga zircons have been documented should be a research priority.
Heat pipe tectonics
Moore and Webb (2013) investigated the thermal effects of “heat-pipe” magmatism in
which volcanism dominates the near surface thermal structure early in planetary evolution.
Their simulations showed that low geotherms could develop in response to frequent volcanic
eruptions that advect surface material downwards. They argued that Hadean zircons arose
in ascending TTG plutons within the diamond stability field produced at the intersection of
the wet basalt solidus and their exceedingly low calculated geotherms. Unfortunately, this
constraint was predicated on a report that diamonds had been included in these grains during
formation (Menneken et al. 2007). This was subsequently shown to be due to contamination
during sample preparation (Dobrzhinetskaya et al. 2014). Even putting this issue aside, left
unaddressed is the source of sufficient water to saturate an intermediate melt (> 25 wt.%;
Mysen and Wheeler 2000) at depths of > 100 km. As noted earlier, the geochemistry of Hadean
Jack Hills zircons is inconsistent with low water activity melting and the inclusion assemblage
is unlikely to arise from a basaltic source.
Terrestrial KREEP
KREEP is an acronym reflecting K-, REE- and P-enriched materials on Earth’s moon,
which are thought to reflect progressive crystallization of a magma ocean (Warren and Wasson
1979). As noted earlier, initial 176Hf/177Hf of Jack Hills zircons show large deviations in εHf(T)
from bulk silicate Earth (see summary in Bell et al. 2014). The initial report of Harrison et al.
(2005) of positive εHf results utilizing ion microprobe U–Pb age spots with LA-ICP-MS Lu–Hf
results on differing portions of the analyzed zircon were not reproduced in a follow-up study
(Harrison et al. 2008) in which age and Hf isotopes were measured on the same volume. This
was ascribed to non-linear mixing effects between zircon rims and cores (see Harrison et al.
2005) that almost certainly also affected the bulk results of Blichert-Toft and Albarède (2008).
Kemp et al. (2010) chose a small subset of the Jack Hills eHf data that aligned along
a subchondritic array extrapolating back to 4.4–4.5 Ga. They interpreted the coherence
of this limited dataset to reflect formation of an incompatible element enriched reservoir
during solidification of a magma ocean—in effect, a terrestrial KREEP layer. In their
model, ~400 Ma of subsequent intra-crustal melting of basalt hydrated by interaction with
an early atmosphere/hydrosphere produced the Hadean Jack Hills zircons including those
with high 18O. They interpreted the results of an experimental study of the simple system
CaO+MgO+Al2O3 + SiO2 + H2O (Ellis and Thompson 1986), which produced peraluminous
melts at ≥ 800 °C under water-saturation, as explaining the presence of muscovite inclusions
in Jack Hills zircons and thus obviating the requirement for a metasedimentary source. While
true that corundum-normative melts are produced under these conditions, muscovite was of
course not present in the K-free experimental system and would have been an unlikely
modal phase to form from a basaltic protolith. As with most of the above hypotheses, the
principal problem with the Kemp et al. (2010) model is the lack of a source for the copious
amounts of water required to saturate melts at high pressure and the implied high zircon
formation temperatures. As described in the Sagduction and Heat pipe sections, carrying
water from surface through a continuous series of dehydration reactions during burial would
result in production of inextractable amounts of melt below 750 °C.
Multi-stage scenarios
Shirey et al. (2008) interpreted the origin of Hadean zircons through a multi-part model
that included: 1) global separation of an early (> 4.4 Ga) enriched reservoir, 2) deep mantle
fractionation of Ca-silicate and Mg-silicate perovskite from a terrestrial magma ocean following
lunar formation, 3) formation of a mafic to ultramafic crust, and 4) repeated cycles of remelting
of that crust under “wet” conditions to produce progressively more silica-oversaturated
TTGs. Their arguments against Hadean Jack Hills zircons forming in a dominantly granitic
crust are the occurrence of zircons in MORB and Icelandic settings (see Icelandic rhyolites
and Mafic igneous rocks sections as well as the requirement of water-saturated melting for
counter arguments). Shirey et al. (2008) note that an Iceland-like environment would permit
hydrothermally altered basalt to be buried to the depths of wet melting to produce zircons of
similar character to the Jack Hills zircons. However, no such population has been documented
in extensive studies of Icelandic zircons (Carley et al. 2011, 2014).
Summary
Although considered separately above, most of the alternative explanations for the
geochemical characteristics of the Jack Hills zircons share the assumption that melting
occurred intracrustally in the absence of an external source of water (i.e., sagduction, Davies
1992; burial beneath impact melts, Marchi et al. 2014; heat pipe burial, Moore and Webb 2013;
terrestrial KREEP, Kemp et al. 2010). As noted previously (see Fig. 8), the P–T conditions
indicated by these zircons puts them outside conditions that would produce extractable
melts fractions thus requiring addition of significant water from an exotic source, such as
dehydration of a downgoing slab. Despite the attractions of a plate boundary-type model,
as noted at the outset, it is not possible to ascribe unique conditions to Hadean Earth from
geochemical records preserved in > 4 Ga zircons. Rather, our goal should be to identify a
parsimonious, internally-consistent model of Earth evolution that explains robust aspects of
the zircon record. We acknowledge that our perspective is strongly biased towards Jack Hills
zircons because the vast majority of information about Hadean conditions has been derived
from them. Increasingly comprehensive studies of other zircon populations, e.g., from Narryer
Hills, might reveal other aspects of Hadean geology. Nonetheless, given data available to date,
plate boundary interactions not only provide a setting that explains the nature of Hadean zircon
geochemistry and the inclusions they host, but also invoke the simplest dynamical mode that
is clearly plausible for this planet. Granted, internal consistency is not smoking gun proof and
most accumulated evidence is indirect and open to alternate interpretations. Overly elaborate
scenarios, especially those that stand in contradiction to important aspects of the geochemical
evidence, do not help advance our understanding of the first five hundred million years of Earth
history. Rather, they muddle a discussion that is generally poorly understood by those outside
this somewhat narrow field. Furthermore, although it should go without saying, mantle-derived
rocks do not possess a record that rocks of continental affinity do not exist elsewhere on the
planet (e.g., Kamber et al. 2005; Kemp et al. 2015; Reimink et al. 2016). This is the equivalent
of concluding that no cratons exist on the planet today from analysis of a rock from Samoa.
Even if we ultimately conclude that plate boundary interactions were the dominant source
of the Hadean zircon geochemical signals, numerous questions remain unanswered. Was the
process continuous throughout the Hadean or did it repeatedly start and stop? Is the inferred
convergent boundary an island arc, a continent-continent collision, a mixture of the two, or an
entirely different kind of setting unique to early Earth?
A LINK TO THE LATE HEAVY BOMBARDMENT?

As noted previously, Jack Hills detrital zircons show a characteristic bimodal age
distribution with peaks at about 3.4 and 4.1 Ga, and ages as old as 4.38 Ga (Holden et al.
2009). A curious feature of this distribution is the relative rarity of zircons between 3.9 and
3.6 Ga (Bell and Harrison 2013), a period which includes a hypothesized spike in impacts to
the Earth-Moon system (termed the Late Heavy Bombardment; LHB). Evidence of such an
event was first seen in ca. 3.9 Ga isotopic disturbances of lunar samples (Tera et al. 1974),
although others (e.g., Hartmann 1975) interpreted this as the tail of a decreasing bolide
flux. The lack of an identifiable signature in the fragmentary terrestrial rock record from
the LHB era (ca. 3.9 Ga) has limited the study of this period of solar system history almost
entirely to extraterrestrial samples. Given its scaling to the Moon in terms of gravitational
cross section and surface area, the Earth likely experienced ~20 times the impact flux to the
Moon causing a widespread crustal thermal disturbance. Thus it is somewhat surprising that
the Jack Hills zircon population does not contain a significant proportion grown in impact
melt sheets (Wielicki et al. 2012, 2016). Because of their crustal origin, all Hadean Jack
Hills zircons share one feature in common—they all must have resided within 10s of km of
the Earth surface during the LHB era. Thermal perturbations in the crust during this time,
perhaps due to impacts, could mobilize Zr to form epitaxial growths on Hadean-age zircons.
Trail et al. (2007) U–Pb depth profiled four Hadean zircons and found that they preserved
3.94–3.97 Ga rims. While they could not rule out endogenic processes as the precipitating
event, they speculated that this common trait might be the terrestrial evidence of the LHB.
Abbott et al. (2012) followed up this study by simultaneously depth profiling U–Pb age and
crystallization temperature of overgrowths on Hadean zircons. Of the eight grains examined,
four had 3.85–3.95 Ga rims that yield significantly higher formation temperatures (> 840 °C)
than either younger rims or older cores. This was again seen as suggestive of an LHB link.
Bell and Harrison (2013) undertook an intensive age survey to archive a large number
(> 100) of Jack Hills zircons formed in the age range 3.9 to 3.6 Ga. Geochemical analyses
on this population showed surprising differences. Specifically, zircons between ca. 3.91 and
3.84 Ga were found to be unique in the > 3.6 Ga Jack Hills zircon record in having two distinct
trace element groupings. The existence of a distinct high-U (and Hf), low-Ti (and Ce, P, Th/U)
zircon provenance (they termed “Group II”) is specific to this ca. 70 million year period. The
remaining 3.91–3.84 Ga zircons (termed “Group I”) resemble the majority of Hadean zircons
both in apparent crystallization temperature and numerous other trace elements. These patterns
in trace element depletion and enrichment, the seemingly paradoxical coincidence of the highest
U contents with high degrees of concordance, and the homogeneous nature or very faint zoning
found in many Group II grains, were interpreted to result from thermally-driven, transgressive
recrystallization (Hoskin and Schaltegger 2003) at 3.91–3.84 Ga. The persistent coincidence of
an apparent thermal event within the period postulated for the LHB suggests that the terrestrial
archive of Hadean material may potentially be a superior to lunar samples for establishing the
timing, and even existence (Boehnke and Harrison 2016), of a Late Heavy Bombardment.
BROADER IMPACTS OF HADEAN ZIRCONS

The role of Hadean zircons in geochemical innovation
Geochemical studies of Hadean zircons have collectively led to a paradigm shift in our
concept of early Earth—there is general agreement that evidence derived from these zircons
implies abundant water at or near Earth’s surface during the Hadean. Almost as interesting as
the paradigm shift itself is the manner by which it occurred and what that says about the role
of early Earth research in driving geochemical innovation. As previously noted, it was the need
to date large (> 105) numbers of zircons to create an archive of > 103 Hadean grains that drove
development of the first automated ion microprobe stage.
Zircon has long been appreciated as the leading crustal geochronometer for its robust U–Pb
system and resistance to physical and chemical alteration in most geologic environments. Thus
knowing the temperature at which magmatic zircon crystallizes sharpens both interpretations of
U–Pb dates and permits a range of new petrochronologic investigations. Although this appeal has
existed since at least the 1970s, it was through the challenge of the unknown provenance of Hadean
zircons that the Ti-in-zircon thermometer was realized (Watson and Harrison 2005; Watson et al.
2006; Ferry and Watson 2007). This application subsequently eruptedacross geochemistry and
petrology (as attested by the over 1,700 citations these three papers have attracted since 2005;
Google Scholar). Subsequent developments, including terrestrial Pu/U tracing (Turner et al.
2007), zircon fO2 barometry (Trail et al. 2011a), and a magma aluminosity proxy (Trail et al.
2016), underscore the degree to which a vanishingly small lithic record has inspired innovation.
The role of Hadean zircons in scientific thought
Perhaps the most remarkable feature of inferences drawn from > 4 Ga zircons is that none
were gleaned from theory. Rather, generations of models innocent of observational constraints
fed a paradigm of a hellish, desiccated, uninhabitable Earth (e.g., Cloud 1976; Smith 1981;
Sleep et al. 1989; Collerson and Kamber 1999; Ward and Brownlee 2000; O’Neill and Debaille
2014) for which there is no empirical evidence. What compelled the scientific community
to create an origin myth in the absence of direct evidence? While science is distinguished
from mythology by its emphasis on verification, its practitioners may be as subject to the
same existential need for explanations as any primitive society. In context with high expected
Hadean heat production and impact flux, it proved irresistible to explain the lack of ancient
continental crust by its non-existence rather than the equally or more plausible notion that it
has been largely consumed by the same processes we see operating on the planet today.
Whether or not this episode represents a scientific anomaly is, to us, a matter of debate.
It is at least arguable that such behavior has been a feature of geophysical modelers approach
to Earth evolution since Kelvin’s “certain truth” that the planet was less than 2% of its actual
age (Thompson 1897) or Jeffreys’ denial of the fit between South America and Africa (Jeffreys
1924). For reasons that remain obscure, calculations carried out in the absence of observational
constraints can take on an edifice-like character in the geo- and planetary sciences that slows
progress and distracts the community from fresh, and possibly better, ideas. Perhaps David
Stevenson (1983) said it best when referring to speculations of Hadean dynamics: “Basic
physical principles need to be understood but detailed scenarios or predictions based upon
them are best regarded as ‘convenient fictions’ worthy of discussion but not enshrinement”.
SUMMARY
Advances in geochemical microanalysis and innovative applications of zircon geochemistry
have made this mineral perhaps the only currently known probe of the time predating Earth’s
known rock record. Zircon’s resistance to mechanical breakup and chemical weathering allow
it to preserve chemical, isotopic, and mineral inclusion information with an associated U–Pb
timestamp through later metamorphism and sedimentary cycling. Although the Jack Hills
locality in Western Australia is the best known and most highly studied source of Hadean
zircons, worldwide twelve additional localities are known to have yielded at least one zircon
older than 4 Ga. Geochemical investigations carried out on Jack Hills zircons have yielded a
view of Hadean Earth fundamentally at odds with traditional notions of a dry, impact-disrupted,
certainly uninhabitable environment, and these lines of investigation can serve as a useful guide
as attention turns to these additional Hadean zircon-bearing localities. Isotopic investigations
reveal the likelihood of a surface or near-surface hydrosphere as early as 4.3 Ga (δ18O; e.g.,
Peck et al. 2001; Mojzsis et al. 2001) and of felsic crust as early as 4.5 Ga (Lu–Hf system; e.g.,
Harrison et al. 2008). Magmatic Th/U ratios (e.g., Cavosie et al. 2006; Bell and Harrison 2013;
Bell et al. 2014), dominantly granitic mineral inclusions (Hopkins et al. 2008, 2010; Bell et al.
2015a), and innovations such as the Ti-in-zircon thermometer (Watson and Harrison 2005) and
Ce/Ce* as fO2 barometer (Trail et al. 2011a) further suggest minimum melt conditions and redox
conditions near the present-day upper mantle for the Hadean magmas that produced the Jack
Hills zircons. One zircon containing primary graphite with a light isotopic value reminiscent
of biologic carbon fixation may point to a terrestrial biosphere as early as 4.1 Ga (Bell et al.
2015b). Preliminary geochemical investigation of zircons from the other localities suggest
higher-temperature origins, and further study of these materials will doubtless shed light on the
diversity of preserved Hadean magmatic environments. Epitaxial rims on Hadean zircon cores
tend toward ca. 3.9 Ga ages and their anomalously high formation temperatures have led some
workers to speculate that this could be a terrestrial signal of the Late Heavy Bombardment.
Although a complete picture of Hadean Earth will likely not be possible from detrital zircon
alone, they have thus far provided the only known empirical data on pre-4 Ga Earth. Models for
this period must consider all of the data from the various isotopic, trace chemical, and mineral
inclusion lines of evidence in the zircons. Increasing the geographic (and possibly petrogenetic)
diversity of Hadean samples available for study will undoubtedly continue to drive geochemical
innovation and new insights into this obscure yet important period of our planet’s history.
ACKNOWLEDGMENTS
We thank Dustin Trail, an anonymous reviewer, and editor Matt Kohn for helpful reviews,
and we thank Matthew Wielicki, Ellen Alexander, Bruce Watson, Stephen Mojzsis, and Ben
Weiss for discussions that helped refine many of the arguments made in this paper. This work
was supported by NSF grant EAR-1551437. The ion microprobe facility at the University of
California, Los Angeles, is partly supported by a grant from the Instrumentation and Facilities
Program, Division of Earth Sciences, National Science Foundation.
REFERENCES
Abdel-Rahman AFM (1994) Nature of biotites from alkaline calc-alkaline and peraluminous magmas. J Petrol
35:525–541
Abe Y (1993) Physical state of the very early Earth. Lithos 30:223–235
Abbott SS, Harrison TM, Schmitt AK, Mojzsis SJ (2012) A search for terrestrial evidence of the Late Heavy
Bombardment in Ti–U–Th–Pb depth profiles of ancient zircons. PNAS 109:13,486–13,492
Amelin YV (1998) Geochronology of the Jack Hills detrital zircons by precise U–Pb isotope dilution analysis of
crystal fragments. Chem Geol 146:25–38
Amelin Y, Davis WJ (2006) Isotopic analysis of lead in sub-nanogram quantities by TIMS using a 202Pb–205Pb
spike. J Anal Atomic Spectrom 21:1053–1061
Amelin YV, Lee DC, Halliday AN, Pidgeon RT (1999) Nature of the Earth’s earliest crust from hafnium isotopes
in single detrital zircons. Nature 399:252–55
Armstrong RL (1981) Radiogenic isotopes: the case for crustal recycling on a near-steady-state no-continental
growth Earth. Phil Trans R Soc London Ser A 301:443–71
Battistuzzi FU, Feijao A, Hedges SB (2004) A genomic timescale of prokaryote evolution: insights into the origin
of methanogenesis phototrophy and the colonization of land. BMC Evol Biol 4:44–51
Bell EA (2013) Hadean–Archean transitions: Constraints from the Jack Hills detrital zircon record. PhD thesis,
University of California, Los Angeles, 285 pp
Bell EA, Harrison TM (2013) Post-Hadean transitions in Jack Hills zircon provenance: A signal of the Late Heavy
Bombardment? Earth Planet Sci Lett 364:1–11
Bell EA, Harrison TM, McCulloch MT, Young ED (2011) Early Archean crustal evolution of the Jack Hills
Zircon source terrane inferred from Lu–Hf, 207Pb/206Pb, and δ18O systematics of Jack Hills zircons. Geochim
Bell EA, Harrison TM, Kohl IE, Young ED (2014) Eoarchean evolution of the Jack Hills zircon source and loss of
Hadean crust. Geochim Cosmochim Acta 146:27–42
Bell EA, Boehnke P, Hopkins-Wielicki MD, Harrison TM (2015a) Distinguishing primary and secondary inclusion
assemblages in Jack Hills zircons. Lithos 234:15–26
Bell EA, Boehnke P, Harrison TM, Mao W (2015b) Potentially biogenic carbon preserved in a 4.1 Ga zircon.
PNAS 112:14518–14521
Bell EA, Boehnke P, Harrison TM (2016) Recovering the primary geochemistry of Jack Hills zircons through
quantitative estimates of chemical alteration. Geochim Cosmochim Acta 191:187–202
Biggin AJ, de Wit MJ, Langereis CG, Zegers TE, Voûte S, Dekkers MJ, Drost K (2011) Palaeomagnetism of
Archaean rocks of the Onverwacht Group Barberton Greenstone Belt (southern Africa): Evidence for a stable
and potentially reversing geomagnetic field at ca. 3.5 Ga. Earth Planet Sci Lett 302:314–328
Bindeman I, Gurenko A, Carley T, Miller C, Martin E, Sigmarsson O (2012) Silicic magma petrogenesis in Iceland
by remelting of hydrothermally altered crust based on oxygen isotope diversity and disequilibria between
zircon and magma with implications for MORB. Terra Nova 24:227–232
Blichert-Toft J, Albarède F (2008) Hafnium isotopes in Jack Hills zircons and the formation of the Hadean crust.
Earth Planet Sci Lett 265:686702
Boehnke P, Harrison TM (2016) Illusory Late Heavy Bombardments. PNAS 201611535
Bowring SA, Williams IS (1999) Priscoan (4.00–4.03 Ga) orthogneisses from northwestern Canada. Contrib
Brasier MD, Antcliffe J, Saunders M, Wacey, D (2015) Changing the picture of Earth’s earliest fossils (3.5–1.9 Ga)
with new approaches and new discoveries. PNAS 112:4859–4864
Buda G, Koller F, Kovács J, Ulrych J (2004) Compositional variation of biotite from Variscan granitoids in Central
Europe: a statistical evaluation. Acta Mineral Petrograph Szeged 45:21–37
Burnham, CW (1975) Water and magmas; a mixing model. Geochim Cosmochim Acta 39:1077–1084
Carley TL, Miller CF, Wooden JL, Padilla AJ, Schmitt AK, Economos RC, Bindeman IN, Jordan BT (2014) Iceland
is not a magmatic analog for the Hadean: Evidence from the zircon record. Earth Planet Sci Lett 405:85–97
Carley TL, Miller CF, Wooden JL, Bindeman IN, Barth AP (2011) Zircon from historic eruptions in Iceland:
reconstructing storage and evolution of silicic magmas. Mineral Petrol 102:135–161
Cavosie AJ, Wilde SA, Liu D, Weiblen PW, Valley JW (2004) Internal zoning and U–Th–Pb chemistry of Jack
Hills detrital zircons: a mineral record of early Archean to Mesoproterozoic (4348–1576 Ma) magmatism.
Cavosie AJ, Valley JW, Wilde SA, EIMF (2005) Magmatic δ18O in 4400–3900 Ma detrital zircons: A record of the
alteration and recycling of crust in the Early Archean. Earth Planet Sci Lett 235:663–681
Cavosie AJ, Valley JW, Wilde SA, EIMF (2006) Correlated microanalysis of zircon: Trace element, δ18O, and U–
Th–Pb isotopic constraints on the igneous origin of complex > 3900 Ma detrital grains. Geochim Cosmochim
Acta 70:5601–5616
Chappell BW, White AJR (1974) Two contrasting granite types. Pac Geol 8:173–174
Cherniak DJ, Watson EB (2003) Diffusion in zircon. Rev Mineral Geochem 53:113–143
Cherniak DJ, Watson EB (2007) Ti diffusion in zircon. Chem Geol 242:470–483
Cherniak DJ, Watson EB (2010) Li diffusion in zircon. Contrib Mineral Petrol 160:383–390
Chopin C, Sobolev NV (1995) Principal mineralogic indicators of UHP in crustal rocks. In: Ultrahigh-Pressure
Metamorphism. Coleman RG, Wang XM (Eds.) Cambridge University Press. England, p. 96–133
Claiborne LL, Miller CF, Wooden JL (2010) Trace element composition of igneous zircon: a thermal and
compositional record of the accumulation and evolution of a large silicic batholith, Spirit Mountain, Nevada.
Clemens JD (1984) Water contents of silicic to intermediate magmas. Lithos 17:273–287
Clemens JD, Vielzeuf D (1987) Constraints on melting and magma production in the crust. Earth Planet Sci Lett
86:287–306
Cloud P (1972) A working model of the primitive Earth. Am J Sci 272:537–548
Cloud P (1976) Major features of crustal evolution. De Toit Lecture, Geol Soc S Afr:1–33
Collerson KD, Kamber BS (1999) Evolution of the continents and the atmosphere inferred from Th–U–Nb
systematics of the depleted mantle. Science 283:1519–1522
Collins WJ, Belousova EA, Kemp AIS, Murphy JB (2011) Two contrasting Phanerozoic orogenic systems revealed
by Hf isotopic data. Nat Geosci. 4:333–337
Compston W, Pidgeon RT (1986) Jack Hills Evidence of more very old detrital zircons in Western Australia. Nature
321:766–769
Coogan LA, Hinton RW (2006) Do the trace element compositions of detrital zircons require Hadean continental
crust? Geology 34:633–636
Crowley JL, Myers JS, Sylvester PJ, Cox RA (2005) Detrital zircon from the Jack Hills and Mount Narryer,
Western Australia: Evidence for diverse > 4.0 Ga source rocks. J Geol 113:239–263
Cui PL, Sun JG, Sha DM, Wang XJ, Zhang P, Gu AL, Wang ZY (2013) Oldest zircon xenocryst (4.17 Ga) from the
North China Craton. Int Geol Rev 55:1902–1908
Darling J, Storey C, Hawkesworth C (2009) Impact melt sheet zircons and their implications for the Hadean crust.
Geology 37:927–930
Dasgupta R (2013) Ingassing storage and outgassing of terrestrial carbon through geologic time. Rev Mineral
Geochem 75:183–229
Davies GF (1992) On the emergence of plate tectonics. Geology 20:963–966
DePaolo DJ, Cerling TE, Hemming SR, Knoll AH, Richter FM, Royden LH, Rudnick RL, Stixrude L, Trefil JS
(2008) Origin and evolution of Earth: Research questions for a changing planet. Nat Acad Press 152 pp
Diwu CR, Sun Y, Dong ZC, Wang HL, Chen DL, Chen L, Zhang H (2010) In situ U–Pb geochronology of Hadean
zircon xenocryst (4.1–3.9 Ga) from the western of the Northern Qinling Orogenic Belt. Acta Petrol Sin
26:1171–1174
Diwu C, Sun Y, Wilde SA, Wang H, Dong Z, Zhang H, Wang Q (2013) New evidence for~ 4.45 Ga terrestrial crust
from zircon xenocrysts in Ordovician ignimbrite in the North Qinling Orogenic Belt China. Gondwana Res
23:1484–1490
Dobrzhinetskaya L, Wirth R, Green H (2014) Diamonds in Earth’s oldest zircons from Jack Hills conglomerate
Australia are contamination. Earth Planet Sci Lett 387:212–218
Duke EF, Rumble D (1986) Textural and isotopic variations in graphite from plutonic rocks South-Central New
Hampshire. Contrib Mineral Petrol 93:409–419
Duo J, Wen CQ, Guo JC, Fan XP, Li XW (2007) 4.1 Ga old detrital zircon in western Tibet of China. China Sci
Bull 52:23–26
Ellis DJ, Thompson AB (1986) Subsolidus and partial melting reactions in the quartz-excess CaO
+ MgO + Al2O3 + SiO2 + H2O system under water-excess and water-deficient conditions to 10 kb: some
implications for the origin of peraluminous melts from mafic rocks. J Petrol 27:91–121
Ewing RC, Meldrum A, Wang L, Weber WJ, Corrales LR (2003) Radiation effects in zircon. Rev Mineral Geochem
53:387–425
François C, Philippot P, Rey P, Rubatto D (2014) Burial and exhumation during Archean sagduction in the east
Pilbara granite–greenstone terrane. Earth Planet Sci Lett 396:235–251
Franzson H, Guðfinnsson GH, Helgadóttir HM, Frolova J (2010) Porosity, density and chemical composition
relationships in altered Icelandic hyaloclastites. In:Water–Rock Interaction. Birkle P, Torres-Alvarado IS
(eds.) London: Taylor & Francis Group, p. 199–202
Freund F, Kathrein H, Wengeler H, Knobel R, Heinen HJ (1980) Carbon in solid solution in forsterite—A key
to the intractable nature of reduced carbon in terrestrial and cosmogenic rocks. Geochim Cosmochim Acta
44:1319–1333
Frezzotti ML, Di Vincenzo G, Ghezzo C, Burke EA (1994) Evidence of magmatic CO2-rich fluids in peraluminous
graphite-bearing leucogranites from Deep Freeze Range (northern Victoria Land Antarctica). Contrib Mineral
Petrol 117:111–123
Froude DO, Ireland TR, Kinny PD, Williams IS, Compston W (1983) Ion microprobe identification of 4,100–4,200
Myr-old terrestrial zircons. Nature 304:616–618
Fu B, Page FZ, Cavosie AJ, Fournelle J, Kita NT, Lackey JS, Wilde SA, Valley JW (2008) Ti-in-Zircon thermometry:
applications and limitations. Contrib Mineral Petrol 156:197–215
Galer SJG, Goldstein SL (1991) Early mantle differentiation and its thermal consequences. Geochim Cosmochim
Acta 55:227–239
Glikson A (2006) Comment on “Zircon thermometer reveals minimum melting conditions on earliest Earth”.
Science 311:779a
Griffin WL, Pearson NJ, Belousova E, Jackson SE, Van Achterbergh E, O’Reilly SY, Shee SR (2000) The Hf
isotope composition of cratonic mantle: LAM-MC-ICPMS analysis of zircon megacrysts in kimberlites.
Grimes CB, John BE, Kelemen PB, Mazdab FK, Wooden JL, Cheadle MJ, Hanghøj K, Schwartz JJ (2007) Trace
element chemistry of zircons from oceanic crust: A method for distinguishing detrital zircon provenance.
Geology 35:643–646
Guitreau M, Blichert-Toft J, Martin H, Mojzsis SJ, Albarède F (2012) Hafnium isotope evidence from Archean
granitic rocks for deep-mantle origin of continental crust. Earth Planet Sci Lett 337:211–223
Hamilton WB (2007) Earth’s first two billion years—The era of internally mobile crust. Geol Soc Am Memr
200:233–296
Hanchar JM, Hoskin PWO (eds.) (2003) Zircon. Rev Mineral Geochem 53, Mineral Soc Am.
Harrison TM (2009) The Hadean crust: Evidence from > 4 Ga zircons. Ann Rev Earth Planet Sci 37:479–505
Lett 261:9–19
Harrison TM, Wielicki MM (2015) From the Himalaya to the Hadean. Am Mineral 101:1348–1359
Harrison TM, Ryerson FJ, Le Fort P, Yin A, Lovera OM, Catlos EJ (1997) A Late Miocene–Pliocene origin for the
Central Himalayan inverted metamorphism. Earth Planet Sci Lett 146:E1–E8
Harrison TM, Blichert-Toft J, Müller W, Albarede F, Holden P, Mojzsis SJ (2005) Heterogeneous Hadean hafnium:
Evidence of continental crust by 4.4–4.5 Ga. Science 310:1947–1950
Harrison TM, Watson EB, Aikman AK (2007) Temperature spectra of zircon crystallization in plutonic rocks.
Geology 35:635–638
Harrison TM, Schmitt AK, McCulloch MT, Lovera OM (2008) Early (≥4.5 Ga) formation of terrestrial crust: Lu–
Hf, δ18O, and Ti thermometry results for Hadean zircons. Earth Planet Sci Lett 268:476–486
Hartmann WK (1975) Lunar “cataclysm”: a misconception? Icarus 24:181–187
Hayes JM, Waldbauer JR (2006) The carbon cycle and associated redox processes through time. Phil Trans R Soc
B: Biol Sci 361:931–950
Hellebrand E, Möller A, Whitehouse M, Cannat M (2007) Formation of oceanic zircons. Geochim Cosmochim
Acta 71:A391–A391
in zircon. Contrib Mineral Petrology 158:317–335
Holden P, Lanc P, Ireland TR, Harrison TM, Foster JJ, Bruce ZP (2009) Mass-spectrometric mining of Hadean
zircons by automated SHRIMP multi-collector and single-collector U/Pb zircon age dating: The first 100 000
grains. Int J Mass Spectrom 286:53–63
Hopkins M, Harrison TM, Manning CE (2008) Low heat flow inferred from > 4 Gyr zircons suggests Hadean plate
boundary interactions. Nature 456:493–496
Hopkins M, Harrison TM, Manning CE (2010) Constraints on Hadean geodynamics from mineral inclusions in
> 4 Ga zircons. Earth Planet Sci Lett 298:367–376
House CH (2015) Penciling in details of the Hadean. PNAS 112:14410–14411
Ibanez-Mejia M, Gehrels GE, Ruiz J, Vervoort JD, Eddy MP, Li C (2014) Small-volume baddeleyite (ZrO2) U–Pb
geochronology and Lu–Hf isotope geochemistry by LA-ICP-MS. Techniques and applications. Chem Geol
384:149–167
Ishihara S (1977) The magnetite-series and ilmenite-series granitic rocks. Min Geol 27:293–305
Iizuka T, Horie K, Komiya T, Maruyama S, Hirata T, Hidaka H, Windley BF (2006) 4.2 Ga zircon xenocryst in an
Acasta gneiss from northwestern Canada: Evidence for early continental crust. Geology 34:245–248
Iizuka T, Yamaguchi T, Hibiya Y, Amelin Y (2015) Meteorite zircon constraints on the bulk Lu−Hf isotope
composition and early differentiation of the Earth. PNAS 112:5331–5336
Ingebritsen SE, Manning CE (2002) Diffuse fluid flux through orogenic belts: Implications for the world ocean.
PNAS 99:9113–9116
Javoy M, Pineau F, Iiyama I (1978) Experimental determination of the isotopic fractionation between gaseous CO2
and carbon dissolved in tholeiitic magma. Contrib Mineral Petrol 67:35–39
Jeffreys H (1924) The earth. Its origin, history and physical constitution. Cambridge University Press
Jennings ES, Marschall HR, Hawkesworth CJ, Storey CD (2011) Characterization of magma from inclusions in
zircon: Apatite and biotite work well feldspar less so. Geology 39:863–866
Kamber BS, Whitehouse MJ, Bolhar R, Moorbath S (2005) Volcanic resurfacing and the early terrestrial crust:
zircon U–Pb and REE constraints from the Isua Greenstone Belt, southern West Greenland. Earth Planet Sci
Lett 240:276–290
Kelemen PB, Manning CE (2015) Reevaluating carbon fluxes in subduction zones what goes down mostly comes
up. PNAS 112:E3997–E4006
Kemp AIS, Wilde SA, Hawkesworth CJ, Coath CD, Nemchin A, Pidgeon RT, Vervoort JD, DuFrane SA (2010)
Hadean crustal evolution revisited: new constraints from Pb–Hf isotope systematics of the Jack Hills zircons.
Kemp AIS, Hickman AH, Kirkland CL, Vervoort JD (2015) Hf isotopes in detrital and inherited zircons of the
Pilbara Craton provide no evidence for Hadean continents. Precambrian Res 261:112–126
Kenny GG, Whitehouse MJ, Kamber BS (2016) Differentiated impact melt sheets may be a potential source of
Hadean detrital zircon. Geology 44:435–438
Keppler H, Wiedenbeck M, Shcheka SS (2003) Carbon solubility in olivine and the mode of carbon storage in the
Earth’s mantle. Nature 424:414–416
Kinny PD, Compston W, Williams IS (1991) A reconnaissance ion-probe study of hafnium isotopes in zircons.
silicates: implications for age determinations. Geochim Cosmochim Acta 68:101–113
Korenaga J (2013) Initiation and evolution of plate tectonics on Earth: theories and observations. Ann Rev Earth
Planet Sci 41:117–151
Krogh TE (1982) Improved accuracy of U–Pb zircon ages by the creation of more concordant systems using an air
abrasion technique. Geochim Cosmochim Acta 46:637–649
Liu J, Ye K, Maruyama S, Cong B, Fan H (2001) Mineral inclusions in zircon from gneisses in the ultrahigh-
pressure zone of the Dabie Mountains China. J Geol 109:523–535
Maas R, Kinny PD, Williams IS, Froude DO, Compston W (1992) The Earth’s oldest known crust: A
geochronological and geochemical study of 3900–4200 Ma old detrital zircons from Mt. Narryer and Jack
Hills Western Australia. Geochim Cosmochim Acta 56:1281–1300
Macgregor AM (1951) Some milestones in the Precambrian of Southern Rhodesia. Proc Geol Soc S Afr 54:27–71
McLean NM, Condon DJ, Schoene B, Bowring SA (2015) Evaluating uncertainties in the calibration of isotopic
reference materials and multi-element isotopic tracers (EARTHTIME Tracer Calibration Part II). Geochim
Maher KA, Stevenson DJ (1988) Impact frustration of the origin of life. Nature 331:612–614
Manning CE, Mojzsis SJ, Harrison TM (2006) Geology, age and origin of supracrustal rocks at Akilia, West
Greenland. Am J Sci 306:303–366
Marchi S, Bottke WF, Elkins-Tanton LT, Bierhaus M, Wuennemann K, Morbidelli A, Kring DA (2014) Widespread
mixing and burial of Earth’s Hadean crust by asteroid . Nature 511:578–82
McKeegan KD, Kudryavtsev AB, Schopf JW (2007) Raman and ion microscopic imagery of graphitic inclusions
in apatite from older than 3830 Ma Akiliasupracrustal rocks west Greenland. Geology 35:591–594
Marty B, Alexander CMD, Raymond SN (2013) Primordial origins of Earth’s carbon. Rev Mineral Geochem
75:149–181
Mathez EA, Blacic JD, Beery J, Maggiore C, Hollander M (1984) Carbon abundances in mantle minerals
determined by nuclear reaction analysis. Geophys Res Lett 11:947–950
Menneken M, Nemchin AA, Geisler T, Pidgeon RT, Wilde SA (2007) Hadean diamonds in zircon from Jack Hills
Western Australia. Nature 448:917–920
Moore WB, Webb AAG (2013) Heat-pipe earth. Nature 501:501–505
Mojzsis SJ, Harrison TM (2002) Establishment of a 3.83-Ga magmatic age for the Akiliatonalite (southern West
Greenland). Earth Planet Sci Lett 202:563–576
Mojzsis SJ, Arrhenius G, McKeegan KD, Harrison TM, Nutman AP, Friend CRL (1996) Evidence for life on Earth
by 3800 Myr. Nature 384:55–59
Mojzsis SJ, Harrison TM, Pidgeon RT (2001) Oxygen-isotope evidence from ancient zircons for liquid water at the
Earth’s surface 4300 Myr ago. Nature 409:178–181
Mojzsis SJ, Cates NL, Caro G, Trail D, Abramov O, Guitreau M, Blichert-Toft J, Hopkins MD, Bleeker W (2014)
Component geochronology in the polyphase ca. 3920 Ma Acasta Gneiss. Geochim Cosmochim Acta 133:68–96
Moore WB, Webb AAG (2013) Heat-pipe earth. Nature 501:501–505
Mysen BO, Wheeler K (2000) Solubility behavior of water in haploandesitic melts at high pressure and high
temperature. Am Mineral 85:1128–1142
Nadeau S, Chen W, Reece J, Lachhman D, Ault R, Faraco MTL, Fraga LM, Reis NJ, Betiollo LM (2013) Guyana:
the Lost Hadean crust of South America? Braz J Geol 43:601–606
Nebel-Jacobsen Y, Munker C, Nebel O, Gerdes A, Mezger K, Nelson DR (2010) Reworking of Earth’s first crust:
constraints from Hf isotopes in Archean zircons from Mt. Narryer Australia. Precambrian Res 182:175–186
Nelson DR (2002) Compilation of Geochronology Data 2001.No. 2002/2 Western Australia Geological Survey
Nelson DR (2005) Compilation of Geochronology Data 2003. No. 2005/2 Western Australia Geological Survey
Nelson DR, Robinson BW, Myers JS (2000) Complex geological histories extending for ≥ 4.0 Ga deciphered from
xenocryst zircon microstructures Earth Planet Sci Lett 181:89–102
Nemchin AA, Whitehouse MJ, Menneken M, Geisler T, Pidgeon RT, Wilde SA (2008) A light carbon reservoir
recorded in zircon-hosted diamond from the Jack Hills. Nature 454:92–95
Nutman AP (2006) Comments on “Zircon thermometer reveals minimum melting conditions on earliest Earth”:
Science 311:779b
Oberheuser G, Kathrein H, Demortier G, Gonska H, Freund F (1983) Carbon in olivine single crystals analyzed
by the 12C(d, p)13C method and by photoelectron spectroscopy. Geochim Cosmochim Acta 47:1117–1129
Ohmoto H, Kerrick DM (1977) Devolatilization equilibria in graphitic systems. Am J Sci 277:1013–1044
O’Neil JR, Chappell BW (1977) Oxygen and hydrogen isotope relations in the Berridale Batholith, Southeastern
Australia. J Geol Soc London 133:559–571
O’Neill C, Debaille V (2014) The evolution of Hadean–Eoarchaean geodynamics. Earth Planet Sci Lett 406:49–58
O’Neil J, Carlson RW, Francis D, Stevenson RK (2008) Neodymium-142 evidence for Hadean mafic crust. Science
321:1828–1831
Ozima M, Podosek FA (2002) Noble Gas Geochemistry. Cambridge Univ. Press. 291 pp
Patiño Douce A, Harris N (1998) Experimental constraints on Himalayan anatexis. J Petrol 39:689–710
Paquette JL, Barbosa JSF, Rohais S, Cruz SC, Goncalves P, Peucat JJ, Leal ABM, Santos-Pinto M, Martin H (2015)
The geological roots of South America: 4.1 Ga and 3.7 Ga zircon crystals discovered in NE Brazil and NW
Argentina. Precambrian Res 271:49–55
Parrish RR (1987) An improved micro-capsule for zircon dissolution in U–Pb geochronology; Chem Geol (Isotope
Geosci Sec) 66:99–102
Peck WH, Valley JW, Wilde SA, Graham CM (2001) Oxygen isotope ratios and rare earth elements in 3.3 to 4.4 Ga
zircons: Ion microprobe evidence for high δ18O continental crust and oceans in the Early Archean. Geochim
Pidgeon RT, Nemchin AA (2006) High abundance of early Archaean grains and the age distribution of detrital
zircons in a sillimanite-bearing quartzite from Mt Narryer, Western Australia. Precambrian Res 150:201–220
Rasmussen B, Fletcher IR, Muhling JR, Wilde SA (2010) In situ U–Th–Pb geochronology of monazite and
xenotime from the Jack Hills belt: Implications for the age of deposition and metamorphism of Hadean
zircons. Precambrian Res 180:26–46
Rasmussen B, Fletcher IR, Muhling JR, Gregory CJ, Wilde SA (2011) Metamorphic replacement of mineral inclusions
in detrital zircon from Jack Hills Australia: Implications for the Hadean Earth. Geology 39:1143–1146
Reimink JR, Chacko T, Stern RA, Heaman LM (2014) Earth’s earliest evolved crust generated in an Iceland-like
setting. Nat Geosci 7:529–533
Reimink JR, Davies JHFL, Chacko T, Stern RA, Heaman LM, Sarkar C, Schaltegger U, Creaser RA, Pearson
DG (2016) No evidence for Hadean continental crust within Earth’s oldest evolved rock unit. Nat Geosci:
doi:10.1038/ngeo2786
Rollinson H (2008) Ophiolitic trondhjemites: a possible analogue for Hadean felsic ‘crust’. Terra Nova 20:364–369
Rosenthal A, Hauri EH, Hirschmann MM (2015) Experimental determination of C, F and H partitioning between
mantle minerals and carbonated basalt CO2/Ba and CO2/Nb systematics of partial melting and the CO2
contents of basaltic source regions. Earth Planet Sci Lett 412:77–87
Rosing MT (1999) 13C-depleted carbon microparticles in > 3700-Ma sea-floor sedimentary rocks from West
Greenland. Science 283:674–676
Schoene B, Samperton KM, Eddy MP, Keller G, Adatte T, Bowring SA, Khadri SF, Gertsch B (2015) U–Pb
geochronology of the Deccan Traps and relation to the end-Cretaceous mass extinction. Science 347:182–184
Schopf JW (2014) Geological evidence of oxygenic photosynthesis and the biotic response to the 2400–2200 Ma
“Great Oxidation Event”. Biochem (Moscow)79:165–177
Sen S, Widgeon SJ, Navrotsky A, Mera G, Tavakoli A, Ionescu E, Riedel R (2013) Carbon substitution for oxygen
in silicates in planetary interiors. PNAS 110:15904–15907
Seifert W, Thomas R, Rhede D, Förster HJ (2010) Origin of coexisting wüstite Mg-Fe and REE phosphate minerals
in graphite-bearing fluorapatite from the Rumburk granite. Eur J Mineral 22:495–507
Shcheka SS, Wiedenbeck M, Frost DJ, Keppler H (2006) Carbon solubility in mantle minerals. Earth Planet Sci
Lett 245:730–742
Shirey SB, Kamber BS, Whitehouse MJ, Mueller PA, Basu AR (2008) A review of the isotopic and trace element
evidence for mantle and crustal processes in the Hadean and Archean: Implications for the onset of plate
tectonic subduction. Geol Soc Am Spec Pap 440:1–29
Sleep NH (2000) Evolution of the mode of convection within terrestrial planets. J Geophys Res 105:17563–17578
Sleep NH, Zahnle KJ, Kasting JF, Morowitz HJ (1989) Annihilation of ecosystems by large asteroid impacts on
the early Earth. Nature 342:139–142
Smith JV (1981) The 1st 800 million years of earth’s history. Philos Trans R Soc London Ser A 301:401–422
Söderlund U, Patchett PJ, Vervoort JD, Isachsen CE (2004) The 176Lu decay constant determined by Lu–Hf and
U–Pb isotope systematics of Precambrian mafic intrusions. Earth Planet Sci Lett 219:311–324
Solomon SC (1980) Differentiation of crusts and cores of the terrestrial planets: Lessons for the early Earth?
Precambr Res 10:177–194
Song S, Zhang L, Niu Y, Su L, Jian P, Liu D (2005) Geochronology of diamond-bearing zircons from garnet
peridotite in the North Qaidam UHPM belt Northern Tibetan Plateau: a record of complex histories from
oceanic lithosphere subduction to continental collision. Earth Planet Sci Lett 234:99–118
Spaggiari CV, Pidgeon RT, Wilde SA (2007) The Jack Hills greenstone belt, Western Australia: part 2: lithological
relationships and implications for the deposition of ≥ 4.0 Ga detrital zircons. Precambrian Res 155:261–286
Spear FS (1988) The Gibbs method and Duhem’s theorem: The quantitative relationships among P, T, chemical
potential, phase composition and reaction progress in igneous and metamorphic systems. Contrib Mineral
Petrol 99:249–256
Spear FS (1993) Metamorphic phase equilibria and pressure–temperature–time-paths. Mineral Soc Am 799 pp
Stern RA, Bleeker W (1998) Age of the world’s oldest rocks refined using Canada’s SHRIMP the Acasta gneiss
complex Northwest Territories Canada. Geosci Canada 25:27–31
Stevenson DJ (1983) The nature of the earth prior to the oldest known rock record—The Hadean earth. In: Earth’s
Earliest Biosphere: Its Origin and Evolution. Princeton University Press, p. 32–40
Tabata H, Yamauchi K, Maruyama S, Liou JG (1998) Tracing the extent of a UHP metamorphic terrane: Mineral-
inclusion study of zircons in gneisses from the Dabie Shan. In: When Continents Collide: Geodynamics and
Geochemistry of Ultrahigh-Pressure Rocks. Springer Netherlands, p. 261–273
Tarduno JA, Cottrell RD, Davis WJ, Nimmo F, Bono RK (2015) A Hadean to Paleoarchean geodynamo recorded
by single zircon crystals. Science 349:521–524
Taylor SR, McLennan SM (1985) The Continental Crust: Its Composition and Evolution. Oxford: Blackwell. 312 pp
Taylor HP, Sheppard SMF (1986) Igneous rocks. I. Processes of isotopic fractionation and isotope systematics.
Rev Mineral 16:227–71
Tera F, Papanastassiou DA, Wasserburg G (1974) Isotopic evidence for a terminal lunar cataclysm. Earth Planet
Sci Lett 22:1–21
Thern ER, Nelson DR (2012) Detrital zircon age structure within ca. 3 Ga metasedimentary rocks Yilgarn Craton:
Elucidation of Hadean source terranes by principal component analysis. Precambrian Res 214:28–43
Thompson W (1863) On the secular colling of the Earth. Phil Mag 25:1–14
Tingle TN, Green HW, Finnerty AA (1988) Experiments and observations bearing on the solubility and diffusivity
of carbon in olivine. J Geophys Res 93:15289–15304
Trail D, Mojzsis SJ, Harrison TM, Schmitt AK, Watson EB, Young ED (2007) Constraints on Hadean zircon
protoliths from oxygen isotopes, REEs and Ti-thermometry. Geochem Geophys Geosyst 8:Q06014
Trail D, Bindeman IN, Watson EB, Schmitt AK (2009) Experimental calibration of oxygen isotope fractionation
between quartz and zircon. Geochim Cosmochim Acta 73:7110–7126
Trail D, Watson EB, Tailby ND (2011a) The oxidation state of Hadean magmas and implications for early Earth’s
atmosphere. Nature 480:79–82
Trail D, Thomas JB, Watson EB (2011b) The incorporation of hydroxyl into zircon. Am Mineral 96:60–67
geospeedometer and peak temperature indicator. Contrib Mineral Petrol 171:1–15
Trail D, Tailby, N, Wang Y, Harrison TM, Boehnke P (2017) Al in zircon as evidence for peraluminous melts and
recycling of pelites from the Hadean to modern times. Geochem Geophys Geosystem. In Press
Turner G, Harrison TM, Holland G, Mojzsis SJ, Gilmour J (2004) Xenon from extinct 244Pu in ancient terrestrial
zircons. Science 306:89–91
Turner G, Busfield A, Crowther SA, Harrison TM, Mojzsis SJ, Gilmour J (2007) Pu–Xe, U–Xe, U–Pb chronology
and isotope systematics of ancient zircons from Western Australia. Earth Planet Sci Lett 261:491–499
Ushikubo T, Kita NT, Cavosie AJ, Wilde SA, Rudnick RL, Valley JW (2008) Lithium in Jack Hills zircons:
Evidence for extensive weathering of Earth’s earliest crust. Earth Planet Sci Lett 272:666–676
Valley JW, Chiarenzelli JR, McLelland JM (1994) Oxygen isotope geochemistry of zircon. Earth Planet Sci Lett
126:187–206
Valley JW, Kinny PD, Schulze DJ, Spicuzza MJ (1998) Zircon megacrysts from kimberlite: oxygen isotope
variability among mantle melts. Contrib Mineral Petrol 133:1–11
Valley JW, Peck WH, King EM, Wilde SA (2002) A cool early Earth. Geology 30:351–354
Valley JW, Cavosie AJ, Fu B, Peck WH, Wilde SA (2006) Comment on “Heterogeneous Hadean Hafnium:
Evidence of continental crust at 4.4 to 4.5 Ga”. Science 312:1139a
Valley JW, Cavosie AJ, Ushikubo T, Reinhard DA, Lawrence DF, Larson DJ, Spicuzza MJ (2014) Hadean age for
a post-magma-ocean zircon confirmed by atom-probe tomography. Nat Geosci 7:219–223
Wang H, Chen L, Sun Y, Liu X, Xu X, Chen J, Zhang H, Diwu C (2007) ∼4.1 Ga xenocrystal zircon from Ordovician
volcanic rocks in western part of North Qinling Orogenic Belt. Chin Sci Bull 52:3002–3010
Ward PD, Brownlee D (2000) Rare Earth: Why Complex Life is Uncommon in the Universe. Copernicus Books
New York
Warren PH, Wasson JT (1979) The origin of KREEP. Rev Geophys 17:73–88
Watson EB, Cherniak DJ (1997) Oxygen diffusion in zircon. Earth Planet Sci Lett 148:527–544
Watson EB, Harrison TM (2005) Zircon thermometer reveals minimum melting conditions on earliest Earth.
Science 308:841–844
Petrol 151:413–433
Weiss BP, Maloof AC, Tailby N, Ramezani J, Fu RR, Hanus V, Trail D, Watson EB, Harrison TM, Bowring SA,
Kirschvink JL, Swanson-Hysell NL, Coe RS (2015) Pervasive remagnetization of detrital zircon host rocks
in the Jack Hills Western Australia and implications for records of the early geodynamo. Earth Planet Sci
Lett 430:115–128
Wendt I, Carl C (1991) The statistical distribution of the mean squared weighted deviation. Chem Geol (Isotope
Geosci Sect) 86:275–285
White RW, Powell RW, Holland TJB (2001) Calculation of partial melting equilibria in the system Na2O–CaO–
K2O–FeO–MgO–Al2O3–SiO2–H2O (NCKFMASH). J Metamorph Geol 19:139–153
Wielicki MM, Harrison TM, Schmitt AK (2012) Geochemical signatures and magmatic stability of terrestrial
impact produced zircon. Earth Planet Sci Lett 321:20–31
Wielicki MM, Harrison TM, Schmitt AK (2016) Reply to Kenny et al. “Differentiated impact melt sheets may be
a potential source of Hadean detrital zircon”. Geology 44:e398
Wilde SA, Valley JW, Peck WH, Graham CM (2001) Evidence form detrital zircons for the existence of continental
crust and oceans 4.4 Ga ago. Nature 409:175–178
Williams IS (2007) Old diamonds and the upper crust. Nature 448:880–881
Wyche S (2007) Evidence of Pre-3100 Ma Crust in the Youanmi and South West Terranes, and Eastern Goldfields
Superterrane, of the Yilgarn Craton. Dev Precambrian Geol 15:113–123
Wyche S, Nelson DR, Riganti A (2004) 4350–3130 Ma detrital zircons in the Southern Cross Granite Greenstone
Terrane Western Australia: implications for the early evolution of the Yilgarn Craton. Aust J Earth Sci 51:31–45
Xing G, Wang X, Wan Y, Chen Z, Yang J, Jitajima K, Ushikubo T, Gopon P (2014) Diversity in early crustal
evolution: 4100 Ma zircons in the Cathaysia Block of southern China. China Sci Rep 4:51–43
Zeng Y, Zhu Y, Liu J (2001) Carbonaceous material in S-type Xihuashan granite. Geochem J 35:145–153
Vol. 83 pp. 365–418, 2017 12
Petrochronology Based on REE-Minerals:

Monazite, Allanite, Xenotime, Apatite
Martin Engi
Department of Geological Sciences
University of Bern
Baltzerstrasse 3
3012 Bern
Switzerland
engi@geo.unibe.ch

REE-minerals
Monazite, xenotime, and allanite are REE1-minerals sensu stricto because lanthanides
(La…Lu) and yttrium are critical constituents in them. Apatite does not require REE, but
because it contains substantial REE in many rocks, it is included in this review. All four
minerals also host unusually high radionuclide concentrations, notably Th and U, forming the
basis of their utility as geochronometers.
This quartet of accessory minerals is playing an increasingly important role in
petrochronology because they provide ways to link robust spot ages to petrogenetic (P–T)
conditions so can lend petrogenetic context to chronology based on other minerals. Part I of this
review assembles the basic requisites prior to integrative petrochronologic analysis. Individual
characteristics of the four REE-minerals are addressed first, i.e., their crystal chemistry and
stability relations. Thermobarometers and trace element geochemistry used for tracing
petrogenesis are discussed next, and finally their chronology is summarized. Part II presents
case studies to highlight the specific strengths of REE-minerals used to resolve the dynamics
of a broad range of processes, from diagenetic to magmatic conditions. Finally, a brief section
at the end outlines a few of the current challenges and promising perspectives for future work.
To introduce the four REE-minerals in style, let us recall the origins of their names. The
three phosphates have well respected Greek grandparents, and allanite has solid Scottish roots,
yet of all four of them show idiosyncracies in etymology or type material.
Apatite had long puzzled naturalists, as it shows great chemical and physical variability
and can resemble other minerals. Once properly identified, Abraham Gottlieb Werner named
it apatite. His reasoning referred to the Greek root ἀπατὰω and giving the precise Latin
translation: decipio. Taken literally, both mean “I deceive” or “I mislead”, which sounds like
an apt confession from this mineral for having fooled humans so long. It is perhaps appropriate
that the original material identified probably was fluorapatite.
Monazite derives from μουάζειυ = to be solitary. This name probably refers to monazite’s
sparse occurrence in early discovered locations as well as its tendency to form isolated crystals
rather than aggregates. I have been unable to trace this fully, but the now abandoned synonyms
eremite (from ερημία = solitude) and kryptolite (= hidden rock) would affirm this interpretation.
1
The rare earth elements (REE ) sensu stricto comprise the lanthanides (La… Lu), plus Sc and Y

1943-2666/17/0083-0012$10.00 (online)
366 Engi
Xenotime again has Hellenic roots, but with a twist, as its original name was kenotime.
An early copy-paste error led to xenotime. Both names contain the stem τιμή = honor.
But κευός means “vain”, and the vain honor refers to the false initial claim by Jöns Jacob
Berzelius that xenotime was a new element, which he then correctly recognized to be a
mineral (e.g., Berzelius 1824) but under a different name. The name xenotime, introduced
later (Beudant 1832), must be considered a misnomer, for ξένος means foreign or odd,
making little sense, except that it is odd to honor a mere spelling mistake. The first reported
kenotime was from a pegmatite in Hidra, SW Norway.
Allanite is named for Thomas Allen (Thomson 1810), an Edinburgh mineralogist who
first singled out the mineral from a load of materials he had acquired, but which had previously
been pilfered by English warships (Secher and Johnsen 2008). The original collector was Karl
Ludwig Giesecke, and the type locality is Aluk, SE Greenland (Ibler 2010; p.86–87).
CRYSTAL CHEMISTRY AND CONSEQUENCES

Mineral structures and compositions relevant to petrochronology are outlined here.
Emphasis is placed on clarifying substitution mechanisms, since this is required to appreciate
how these minerals incorporate and fractionate REE, U, Th, and other elements used in
chronology and petrology. Which exchange vectors operate is essential also to formulate
chemical equilibria that constrain conditions of growth or interaction with other minerals, melt
or hydrous fluid. Among the REE three groups are often distinguished, i.e., light, middle, and
heavy REE, where LREE: La…Nd, MREE: Sm, Eu, Gd, and HREE: Tb…Lu. Y has an ionic
radius within the range of HREE and is usually included with these.
More complete and comprehensive mineralogical accounts include the following:
Huminicki and Hawthorne (2002) should be consulted for phosphate structure in general; Ni
et al. (1995) for monazite and xenotime, Clavier et al. (2011) for monazite, and Hetherington
et al. (2008) for xenotime. For apatite refer to White et al. (2005), Hughes and Rakovan
(2002), Pan and Fleet (2002) or, for an elegant overview, Hughes and Rakovan (2015). The
crystal chemistry of allanite is briefly summarized in Armbruster et al. (2006) and with more
background in Gieré and Sorensen (2004).
Monazite and xenotime
Crystal structure, chemical substitutions. Monazite and xenotime are large-ion
orthophosphates (REE)[PO4] with related crystal structures; subtle differences between these
help to explain their substantially different REE-fractionation (Ni et al. 1995): The large
trivalent ion in monazite is coordinated to nine O-ions that form an irregular polyhedron
(REE)O9; in xenotime, Y (plus other HREE3+) is coordinated to eight O-ions that form a
regular dodecahedron (Y)O8 (Fig.1). By sharing edges, these polyhedra form chains in the
b-direction in both crystal structures. Along c these chains are linked by [PO4] tetrahedra
to form a (100) sheet, and these layers are then stacked along a by sharing edges with the
REE-polyhedra. While monazite is monoclinic and favors LREE, its dimorph xenotime-
(REE) is tetragonal (part of the zircon group) and prefers HREE.
Large-ion phosphates show extensive solubility with related Th-phosphates and -silicates,
such as cheralite CaTh(PO4)2 and huttonite ThSiO4. Y (or some HREE or Gd) is a required
constituent for xenotime; for monazite it is Ce (or La), which can be substituted by other
LREE and Sm. Apart from exchanges involving only REE3+ and Y3+, two main substitutions
involve coupled charge balance:
2 REE3+ ↔ Th4+ + Ca2+ (1)
REE3+ + P5+ ↔ Th4+ + Si4+ (2)

Petrochronology Based on REE-Minerals 367
A B
Figure 1. Local structure of REE-sites (A) in monazite and (B) in xenotime. Phosphate (PO4) tetrahedra
shown in green. The large trivalent ion (red sphere) in monazite occupies an irregular (REE)O9 polyhedron
that preferentially accomodates the larger, light to middle REE and Th4+; in xenotime the regular (Y,REE)O8
dodecahedron provides a suitable site for Y and the smaller, heavy REE.
In both cases, U4+ may stand for Th4+; most monazites have Th/U > 1, but xenotime usually
shows Th/U < 1. Data compiled by Spear and Pyle (2002) indicate that the cheralite exchange
(Eqn. 1) dominates for monazite, but the huttonite exchange (Eqn. 2) is important as well
(Table 1). Experimental data (Stepanov et al. 2012) indicate that huttonite dominates over
cheralite at high pressure. In xenotime the chemical correlation plots are less clear; Zr4+ may
substitute for Th4+, so Zr should be included in xenotime analyses. Other minor substitutions
are heterovalent as well, e.g., those involving halogens (mostly F), sulfur (in monazite), As
(Janots et al. 2011), and Sr for REE (Krenn and Finger 2004).
As reviewed by Harrison et al. (2002) and Williams et al. (2007), experiments in binary
systems of LaPO4 with cheralite, Ca½U½PO4, and huttonite show complete or extensive solid
solutions (at 780 °C), but natural compositions cover only parts of the range (e.g., Förster
and Harlov 1999). Where the two phosphates are cogenetic, different exchange vectors may
dominate, e.g., cheralite in monazite, but huttonite in xenotime (Fig. 7 in Regis et al. 2012).
Compared to xenotime, natural monazite shows considerably more variability in the
heterovalent substitutions; this may be due to the flexibility of the larger and less regular
polyhedron. The difference in coordination may also account for much of the observed REE-
partitioning: Larger LREE ions (and Th4+) find a “happy home” in the larger 9-O polyhedra
in monazite, while Y3+ and the smaller HREE clearly prefer the smaller 8-O polyhedra in
xenotime. In addition, some LREE show a tetrad effect, which seems due to orbital preference
(of their unpaired 4 f-electrons) for the irregular REE-site in monazite (Duc-Tin and Keppler
2015). The magnitude of these REE-preferences is visible in the distribution coefficients
(Fig. 2), and these data also indicate that U4+ fractionates but weakly between xenotime and
monazite. In general, the actinide ratio (Th/U) is higher for monazite than for xenotime.
Table 1. Main compositional endmembers of REE-phosphates.

Endmember Name in monazite Name in xenotime Notes
(REE)PO4 monazite-(LREE) xenotime-(LREE) LREE: Ce > La ≈ Nd > Sm
Ca½Th½PO4 cheralite formerly brabantite
ThSiO4 huttonite thorite
USiO4 coffinite
368 Engi
Figure 2. Distribution coefficient (con-

centration ratio KD = Cmon / Cxen) be-
tween monazite and xenotime for REE,
U, and Th. Trends for LREE to MREE
highlighted in orange, for MREE to
HREE in green, for actinides in blue;
based on data from Franz et al. (1996).
In the REE-budget of many rocks, monazite plays a major role for the LREE, as does
xenotime for the HREE. Monazite is the most common of the radioactive minerals (Overstreet
1967) and the main host of Th and U in many rocks.
Fortunately for geochronology, both monazite and xenotime do not favor incorporation
of lead in their structure, although Pb2+ readily fits into monazite. Typically, common lead
contents are low (often at ppm-level: Parrish 1990), although for precise isotopic data a
common lead correction can be significant. In chemical U–Th–Pb dating (Williams and
Jercinovic 2017; this volume), such correction is not possible, and its effect often would be
within the analytical uncertainty due to electron probe counting statistics. It should be noted,
however, that elevated common lead contents in monazite are known, e.g., in high pressure
environments (Holder et al. 2015), in rocks altered by interaction with F-rich hydrous fluid
(Didier et al. 2013), and occasionally in pegmatites (Kohn and Vervoort 2008); chemical
dating of monazite is not advisable for such samples.
Sector zoning. Crystal-chemical controls of monazite can influence its uptake of
REE. Cressey et al. (1999) documented differential incorporation of ions due to structural
differences between specific growth surfaces. In that study, monazite from a carbonatite
was sectorially enriched in La in {011} but depleted in {101} and {100} zones, whereas
Nd showed an inverse preference; the effect is insignificant for Ce. Fractionation factors are
modest in this case (~1.5), but caution is needed when concentration differences are observed,
and such chemical heterogeneity in monazite is used to assess reaction relations or local REE-
mobility. Stepanov et al. (2012) experimentally confirmed sector zoning in monazite from
granitic melt (at 1000 °C, 1 GPa, 16 wt% H2O); fractionation for the LREE was weak, but
much more pronounced for Sm, Gd, Y, and notably also for Th. The data indicate that different
growth sectors show substitutions of LREE for huttonite and for MREE (or HREE).
However, sector enrichment is an equilibrium feature of mineral surfaces, and close
spatial proximity in a natural sample does not guarantee coeval growth. For instance, in a
suite of lower greenschist to middle amphibolite facies metapelites, Janots et al. (2008) found
substantially different La/Nd ratios in two populations of monazite (and similar fractionation
factors as reported by Cressey et al. 1999), but local textures (Fig.10 in Janots et al. 2008)
indicate sequential growth rims, not simultaneous sector fractionation. Sector zoning appears
to be particularly common at very high metamorphic grade, notable in the granulite facies
(DeWolf et al. 1993; Bingen and van Breemen 1998; Zhu and O’Nions 1999), while magmatic
monazite can show oscillatory zoning (Montel 1993).
Apatite
Apatite, Earth’s most abundant phosphate, has an extraordinarily flexible crystal structure.
It incorporates a wide range of minor and trace elements, including some most interesting
ones for geochemistry and chronology, e.g., S, Sr, U, Th, and REE. Furthermore, anion
substitution in apatite is common; in most rocks apatite shows F > OH >> Cl, but F/OH ratios
vary. Apatite usually shows hexagonal symmetry, but anion ordering causes structural changes
in Ca5[PO4]3(F,OH,Cl) (Hughes and Rakovan 2002).
In terms of total REE-content, natural apatite is not comparable to monazite and xenotime,
but it can be significant in the LREE-budget of rocks, especially in carbonatites and mafic
rocks (e.g., Y + La + Ce to 22,000 ppm: Cruft 1966; LREE > 15,000 ppm: Finger et al. 1998),
in migmatites (LREE ~ 4,750 ppm: Bea 1996). REE concentrations appear to increase from
amphibolite to granulite grade (LREE + MREE > 10,000 ppm: Bingen et al. 1996).
Crystal-chemically, REE do find a “happy home” in the M2 site of apatite, which is
coordinated to six O-ions plus one monovalent anion (Pan and Fleet 2002). U4+ also favors the
M2 site in most apatites (Luo et al. 2009), but Th contents are very low. The site occupancy
ratio (M2REE/M1REE) changes with ionic radius: MREE such as Nd show no site preference,
but HREE prefer the M1 site, which is coordinated to nine oxygen ions; M1 is smaller than
M2, so ionic size is the probable cause of the site preference (Wood and Blundy 1997).
Substitution mechanisms involving REE in natural apatite involve charge compensation
(Pan and Fleet 2002); the main exchanges proposed are:
REE3+ + Na+ ↔ 2 Ca2+ (3)
REE3+ + SiO44− ↔ Ca2+ + PO43− (4)

Such heterovalent exchange mechanisms typically lead to much slower diffusion than homovalent
exchanges; this was indeed found in experiments for REE in fluorapatite (Cherniak 2000).
Because coupled exchanges depend on the concentration of other mono- to pentavalent cations
on different sites in the crystal lattice, REE-diffusivity in apatite must be sensitive to composition.
Sector zoning. Apatite shows sector preference for certain REE (Rakovan and Reeder
1996); in this case the MREE and HREE are most affected, and fractionation is much more
pronounced than in monazite: One study (Rakovan et al. 1997) observed nearly an order of
magnitude difference for Nd between {0001} and {1010} zones. On petrographic grounds,
the sectors analyzed were judged to represent a single growth phase. Paired with substantially
different Sm/Nd ratios in the two sectors, the compositional spread from analyses of different
sectors of one crystal was sufficient to define an isochron age of 43.8 ± 4.7 Ma (2σ).
Allanite
Crystal structure, chemical substitutions. Gieré and Sorensen (2004) gave an excellent
entry into the world of allanite; Armbruster et al. (2006) clarified the nomenclature for the
epidote-group and the non-trivial task of site-assignment, and they nominally ended the
debate of what should be called allanite. Following Ercit (2002), allanite-(Ce), allanite-(La)
etc. are now recognised as distinct mineral species. Here, the term allanite refers to what is
officially the allanite subgroup.
Minor amounts of REE are often present in epidote group minerals, but in allanite LREE and
Fe2+ are structurally essential constituents. The structural formula A2M3[SiO4][Si2O7](O)(OH)
describes the entire epidote group and allows the main substitutions to be understood. Two
groups of sites are identified, labeled A and M: A involves A(1) and A(2), which are 9- and
11-fold coordinated, respectively; M comprises three octahedra, of which M(1) and M(2) are
essentially occupied by Al. With the main constituent of A(1) being Ca (plus minor Mn2+), most
370 Engi
substitutions occur on A(2) and M(3) (Fig. 3): In natural allanite, A(2) is mostly occupied by
REE3+, with M(3) containing Fe2+ and minor Mg. By contrast, most natural epidotes (formally
members of the clinozoisite subgroup) have A(2) principally occupied by Ca and M(3) by
trivalent ions (Fe3+, Al, Mn3+). Allanite frequently shows chemical zoning or overgrowths,
demonstrating solid solution with epidotes. The main exchange is heterovalent and can be
written in site-specific notation as
A2
(REE)3+ + M3M2+ → A2Ca2+ + M3M3+ (5)
Data compiled by Gieré and Sorensen (2004) show chemical analyses covering the complete
range possible due to this exchange. In many cases the contents in REE3+ and M2+ are higher
than will fit into the A2 and M3 sites, respectively, indicating that other sites can be involved
as well. In addition to 2–3 wt% REE2O3, allanite shows tetravalent substitutions, notably to
accommodate Th (typically 1,000–10,000 ppm) and U (mostly < 2,000 ppm):
A2
(Th4+,U4+) + 2 M3M2+ → A2Ca2+ + 2 M3M3+ (6)
Unless Fe2+/Fe3+ is analyzed by Mössbauer spectrometry (Fehr and Heuss-Assbichler 1997),
the ferrous fraction must be estimated when normalizing chemical analyses for epidote group
minerals from electron probe data (and LA-ICP-MS2 data for trace elements). Analyses are
normalized to 12.5 oxygen atoms on an anhydrous basis, as explained by Gieré and Sorensen
(2004; their Figs. 5 and 6), based on Petrík et al. (1995).
As already noted by Goldschmidt and Thomassen (1924), allanite prefers LREE
(La–Nd), similar to monazite, but in marked contrast to zircon and garnet, two commonly
coexisting phases that tend to be HREE-enriched. Partitioning patterns—often shown in spider
diagrams, normalized to chondrite or some suitable reference mineral—can thus be used for
paragenetic indications (Hermann 2002). For allanite, the absolute amounts of the REE, the
slope and curvature in the diagram, the presence or absence of anomalies (Eu, Ce?) all reflect
competition between allanite and other REE-bearing minerals as the allanite grew.
Figure 3. Structural elements of allanite shown as combined coordination polyhedra with specific ions to
emphasize characteristic allanite substitutions relative to basic epidote structure. Large red spheres with
sticks mark the 11-fold A2 site for the REE3+ ⇔ Ca2+ exchange; charge balance is via the Fe2+ ⇔ Fe3+ on
the M3 (distorted) octahedral site, shown in bright green. Isolated SiO4 tetrahedra in yellow, Si2O7 groups
in turquoise. Pink spheres show 9-fold coordinated Ca in large cavities.
2
Laser ablation inductively coupled plasma mass spectrometry
Sector zoning
Sharply discrepant compositions are often found in epidote group crystals, and their
significance is not always clear. Apart from magmatic and metamorphic overgrowths,
a possible miscibility gap has been debated, but this option was dismissed by Banno and
Yoshizawa (1993) in favor of sector growth.
Radiation damage
Emission of α-particles in the decay chains of 238U, 235U, and 232Th cause ionization and
inflict radiation damage on any mineral with appreciable U or Th contents. Spontaneous 238U
fission is several million times rarer than α-decay, but causes much more severe structural
damage to a host crystal. To indicate the extent of damage, optical effects in thin section
(color, reduced anisotropy) and Raman spectra (McFarlane 2016) are useful. Minerals respond
differently to the effects of both types of damage: In zircon and allanite with high actinide
contents, structural damage accumulates over time, disrupting their crystalline integrity and
eventually leading to a metamict (largely amorphous) state. By contrast, REE-phosphates do
not become metamict, even though they can accommodate up to 30 wt% ThO2 + UO2 (van
Emden et al. 1997); the damage inflicted to their structure appears to heal. Thermal annealing
occurs at low temperatures, i.e., at ca. 100 °C in apatite and ca. 300 °C in monazite (Gleadow
et al. 2002). While the mechanisms of such restoration are not well understood, they
certainly have positive effects: Monazite, xenotime, and apatite remain structurally intact,
and are less likely to lose their radiogenic Pb (Seydoux-Guillaume et al. 2004) and interact
with fluid than heavily damaged allanite or zircon. Annealing can occur in allanite as well
(Karioris et al. 1981), but it is more protracted. Structurally damaged or metamict grains are
quite common in actinide-rich varieties, especially in Paleozoic and older rocks. But caution
is indicated—even where old grains rich in U + Th now appear structurally intact, radiation-
damaged grains may have healed after a period of open system behaviour. Such a scenario
may explain why ages from non-metamict allanite occasionally appear implausibly young,
such as for the Pacoima Canyon pegmatite (California), for which reportedly non-metamict
allanite was dated to 1006 ± 37 Ma (zircon: Weber 1990; allanite: Gieré and Sorensen 2004,
p. 466–472), whereas zircon is 1191 ± 4 Ma (SIMS Th–Pb, Catlos et al. 2000).
Disturbed Th–U–Pb dates in radiation-damaged allanite may result from Pb-loss or
actinide remobilization (TIMS U–Pb, Barth et al. 1994). In this and other studies, discordances
in ages (from TIMS and LA-ICP-MS) are larger between 208Pb/232Th and Pb/U than between
206
Pb/238U and 207Pb/235U. This implies unequal chemical mobility of Th and U in allanite.
Barth et al. (1994) discovered minute grains of thorite and uraninite along fluid pathways
(cracks) in two metamict samples. This indicates an indirect effect of radiation damage,
and such secondary phases would account for the decoupling between U- and Th-systems.
Beyond a critical cumulative irradiation dose, hydrothermal fluid attack appears to be much
more effective. According to Smye et al. (2014), a critical dose sufficient to effect isotopic
disturbance is an order of magnitude lower in allanite than in zircon. This susceptibility to
damage and alteration, and the generally much higher concentration of Th + U, help explain
why allanite ages > 300 Ma often show more effects of radiation damage than zircon.
Diffusion and closure temperature
Several aspects of petrochronology must consider diffusive mobility, notably
• when constraining petrogenetic conditions, applying thermobarometers etc., based on
local equilibria in sample domains (Lanari and Engi 2017);
• when considering the effects of chemical re-equilibration upon subsequent changes in
P–T (pro- and/or retrograde), whether by pure exchange- or combined net–transfer-
and exchange-reactions (Kohn and Penniston-Dorland 2017);
372 Engi
• when analyzing geochronological data (e.g., normal / inverse discordance) to assess

full/partial isotopic equilibration during mineral growth, and subsequent loss of parent
or daughter isotopes (Wasserburg 1963).
The essential theoretical aspects of diffusion, experimental constraints, select analytical
data from natural examples, and insight gained from petrochronological studies are presented
and evaluated in this volume (Kohn and Penniston-Dorland 2017). Here, a brief account is
given only to assess the current state of knowledge about isotopic closure temperatures (Tc)
relevant to U–Pb, Th–Pb, and Pb–Pb chronometry. While the link between Tc and diffusivity
(D) is well defined (Dodson 1973, 1986; Ganguly and Tirone 1999; Kohn and Penniston-
Dorland 2017), minerals differ in their capacity to self-heal damage due to fission and
α-decay, e.g., allanite can become metamict with time. Insight gained from work on natural
samples occasionally is in conflict with results from experimental diffusion studies, and
structural damage certainly increases ionic mobility. While this topic has been addressed
in numerous studies, we are far from understanding how closure temperatures change in
minerals undergoing (partial) metamictization and thus affect what ages are recorded.
For monazite, Parrish (1990) estimated Tc = 725 ± 25 °C, a value still commonly quoted,
but Tc values of 800–850 °C have subsequently been proposed, based on preserved age relations
in high-temperature samples (Spear and Parrish 1996; Bingen and van Breemen 1998; Kamber
et al. 1998). Experimental diffusion data initially indicated Tc ~ 500 °C for 10 µm monazite
(Smith and Giletti 1997), but subsequent work (Cherniak et al. 2004) found Tc > 900 °C. Pb
diffusion in xenotime was found to be even slower than in monazite, and Tc must be similarly
high (Hawkins and Bowring 1997; Cherniak 2006, 2010).
No experimental diffusion data are available for Pb in allanite; closure temperatures
have been estimated based on dated high-temperature samples. A minimum Tc of 750 °C is
required to account for the retentivity of allanite (Oberli et al. 2004) crystallized in Bergell
tonalite during magmatic differentiation (see Case Studies below, Fig. 21). Heaman and
Parrish (1991) proposed 650 ± 25 °C, but also noted that allanite U–Pb closes at higher
temperature than titanite. Tc for titanite has recently been revised upward to near 800 °C
(Kohn 2017), and I am not aware of any natural datasets that would indicate a lower Tc for
allanite. Note, however, that allanite may seem young because of retrograde growth (Finger
et al. 1998; Wing et al. 2003), well below its Tc.
At current understanding, the conclusion is that lead loss by diffusion from monazite,
xenotime, and allanite is probably negligible up to temperatures of > 750–800 °C over 10 Ma
for grains >20 µm. Such conditions certainly encompass “normal” granulite samples, i.e.,
U–Th–Pb chronometry applied to high-temperature growth zones should yield formation
ages. Recent studies have interpreted monazite data even from UHT terrains (> 900 °C) as
formation ages (e.g., Suzuki et al. 2006; Korhonen et al. 2013), but rigorous corroboration
using diffusion profiles from such samples would be desirable. The shape of concentration
profiles across different growth zones should allow diffusion models to be tested (e.g.,
reasonable D·∆t3) and may potentially constrain the T–t history (e.g., Watson and Harrison
1984). However, it is important to remember that fluid- or deformation-induced retrogression
can be far more important than purely diffusional processes (Teufel and Heinrich 1997;
Seydoux-Guillaume et al. 2002b; Harlov et al. 2011; Shazia et al. 2015).
3
since the diffusion length is 2(D·t)½
GEOTHERMOBAROMETRY
Monazite and xenotime thermometry
A welcome effect of the pronounced Y-HREE fractionation is the miscibility gap between
xenotime and monazite, which is quite asymmetric in YPO4–(REE)PO4. The Y-content
of monazite serves as a geothermometer, with only modest pressure-sensitivity (Gratz and
Heinrich 1997, 1998; Heinrich et al. 1997; Andrehs and Heinrich 1998). Calibration data
are summarized and discussed in Spear and Pyle (2002, Fig.24); they emphasize effects of
additional components (Th, U, Ca, Si), but so far only the effect of Th on the miscibility gap
has been calibrated (Seydoux-Guillaume et al. 2002a).
Figure 4 compares calibrations based on experiments. From these data, Mogilevsky
(2007) developed a generalized regular solution model that can handle other components as
well. For thermometry, this formalism (as presented in Mogilevsky’s Equations 4–7) demands
a numerical solution, and so far it seems that no applications have appeared in geology. Spear
and Pyle (2010) used the same experimental data to construct a thermodynamic model based on
estimated endmember properties, and approximate temperatures can be obtained from these.
This approach is promising, but so far restricted to a very small compositional space; a wider
basis of phase equilibrium data is sorely needed to allow a more generalized thermodynamic
calibration and then applications thereof.
Apart from such technical issues, applying monazite–xenotime thermometry to rocks has
its limits. Notably, xenotime should have been present when monazite grew, else the method
yields minimum temperature values only. Since garnet (YAG component) is a major competitor
for Y, xenotime is consumed during prograde garnet growth and commonly disappears from the
assemblage; it may reappear where garnet is consumed, at higher grade or on decompression or
retrogression (e.g., Hallett and Spear 2015). For these reasons, it can be tricky to ascertain that a
specific monazite zone grew while saturated in xenotime, and in fact they rarely occur together
in equilibrium (e.g., Pyle and Spear 1999). In zoned monazite, the Y-content is a good guide:
annuli with high Y are good candidates for thermometry, whereas an abrupt decrease in Y may
indicate growth in a xenotime-absent assemblage. Geochemical zoning and textural relations
can convincingly discriminate such populations (e.g., Foster et al. 2004; Daniel and Pyle 2006,
p. 108–113). In rocks where monazite and xenotime are both present in local assemblages,
they often show zoning, and it can be unclear how to assess which compositional pairs were
in equilibrium at any stage of the evolution; checking if partition coefficients (notably Gd/Y,
Dy/Y: Pyle et al. 2001; perhaps also Th/U) show equilibrium values is preferable to blind trust.
Figure 4. Monazite limb of the misci-

bility gap to xenotime, comparing cal-
ibrations based on experiments. Two
isobars (0.2 and 1.5 GPa) from Gratz
and Heinrich (1997) show sensitivity
to pressure; three isopleths from Sey-
doux-Guillaume et al. (2002a) display
effect of the Th-content of monazite;
note that ~20% Th-endmember shows
an effect comparable to ~1 GPa pres-
sure increase. Curves labeled Ce, La,
Ce, Nd show the dependence on the
respective REE from models cali-
brated by Mogilevsky (2007) using
experimental data.
374 Engi
Two further thermometers make use of the Y-contents of garnet in equilibrium with either
xenotime or monazite (Pyle and Spear 2000; Pyle et al. 2001), but do not depend on both
phosphates being present. The empirical garnet–xenotime thermometer is useful but over a small
temperature range (~450–560 °C). As long as xenotime buffers the Y-activity of an assemblage,
only the Y-content of garnet (grt) is needed for thermometry (Pyle and Spear 2000):
16031( ±862 )
ln ( Y/ppm
= )grt − 13.25 ( ±1.12 ) (7)
(T / K )
The garnet–monazite thermometer extends this limited temperature range; it is
most sensitive up to ~620 °C. At higher temperature the Y-concentration in garnet is low,
approaching the detection limit of EMPA4 data, hence LA-ICP-MS analyses are preferred.
In addition, quartz, plagioclase and apatite must coexist, and the thermometer depends on
H2O, albeit weakly (Pyle et al. 2001):
 −1.45 × ( P / bar ) + 103 × 447.8 ( ±32.1) 

T /K =  (8)
567 ( ±40 ) − R ln ( K eq ) 
 
requiring analyses and solution models for garnet (grt, Grs: Ca3Al2Si3O12, YAG: Y3Al5O12),
plagioclase (pl, An: CaAl2Si2O8), monazite (mnz), and apatite (ap, OH-ap: Ca5[PO4]3(OH))
to obtain
K eq =
(a5/ 4
grs,grt
5/ 4
aan, 3 1/ 2
pl aYPO 4 ,mnz fH 2 O ) (9)
(a YAG,grt aOH-ap,ap )
Provided the mineral analyses are accurate5, the calibration usually yields temperature
estimates to within ±30 °C of independent thermometers even if a pure hydrous fluid is
assumed. However, to be petrochronologically useful, the thermometer should be restricted to
monazite–garnet pairs that grew simultaneously, and to ascertain this, petrographic scrutiny is
essential (Pyle et al. 2001).
Monazite–melt thermobarometry
Solubility data for monazite in granitic melt were used to calibrate a thermometer for
magmas (Montel 1993; Plank et al. 2009); it relies on low-pressure experiments (0.2 GPa,
Montel 1986; Rapp and Watson 1986). A similar thermometric equation, but with a pressure
correction, was proposed by Stepanov et al. (2012), incorporating their new data to 5 GPa.
All these thermometers assume that monazite dissolves to dominantly associated LREE(PO4)
species. However, recent data (Duc-Tin and Keppler 2015) indicate that at least for some
magma compositions partial dissociation in the melt occurs at low pressure, and at pressures of
3 GPa (Skora and Blundy 2012) dissociated LREE3+ and (PO4)3− ions are dominant in hydrous
silicic melts. This behavior parallels other hydrous fluids at high pressure (Manning 2004).
As pointed out by Skora and Blundy (2012), thermometry such as advocated by Plank et
al. (2009) would be seriously (>100°) in error at pressures typical of subduction systems.
Stepanov et al. (2012) developed the following equation describing monazite solubility in
peraluminous and metaluminous granitic melts:
ln ( CLREE /ppm ) = 16.16 ( ±0.30 ) + 0.23 ( ±0.07 ) × ( H 2O /wt.% ) + ln ( X LREE )

1/ 2 mnz
11494 ( ±410 ) 1.94 ( ±0.40 ) × ( P /GPa ) (10)

− –
(T /K ) (T /K )
4
EPMA: Electron probe microanalyzer
5
As OH-apatite is obtained by difference, F in apatite requires analytical attention
(CLREE := Σ La…Sm in ppm; XLREE := LREE / [LREE + Y + Th + U]). This linear fit

reproduces a large dataset (750–1400 °C, 0.2–5 GPa, 1–20 wt% H2O and 0.82–1.36 ASI 6),
but includes neither the experimental data for melts with (CaO + FeO + MgO) > 3 wt% nor for
peraluminous and phosphate-rich melts (Duc-Tin and Keppler 2015). Figure 5 shows that the
strong temperature-dependence vanishes for highly aluminous melts (ASI > 1.0).
TRACE ELEMENT GEOCHEMISTRY AND PETROGENESIS

Magmatic and partial melting range
Phosphates. Monazite grains crystallized from melt typically show chemical zoning,
and this can be very useful to single out specific growth zones for spot dating. So far, most
emphasis has been placed on Th- and Y-zoning (Pyle and Spear 2003; Kohn et al. 2005;
Rubatto et al. 2013). This is addressed below, but recent experimental work has clarified the
interplay of these elements with the REE, and thus zoning in the LREE and MREE, even
in individual REE, should be deciphered in detail. Such information is likely to add useful
control on the dissolution and growth history and elucidate the control phosphate solubility
plays in fractionation, e.g., of Sm over Nd (Zeng et al. 2005) or LREE showing tetrad effects
for monazite (Duc-Tin and Keppler 2015).
Figure 5. Experimental solubility data for REE-phosphates in granitic melt by Duc-Tin and Keppler (2015)
and earlier studies. (A) Monazite solubility is more strongly temperature-dependent for the LREE than for
MREE (and HREE); size of symbols shows uncertainty in data. (B) The effect of the aluminum saturation
index (ASI) on REE solubility. Select trends (based on data for 800 °C, 0.2 GPa, ~0.1 wt% P2O5) show
solubility generally increasing from LREE (La–Nd) to MREE (Sm–Gd) to HREE (Tb–Lu); differences are
not significant for peraluminous melts (ASI > ~1.0), but for peralkaline melts solubility strongly increases.
Steep slopes for Nd and Gd indicate tetrad effect.
6
ASI := Al/(Na + K + 2Ca)
376 Engi
REE. Data on monazite solubility (Montel 1986; Rapp and Watson 1986; Rapp et al.
1987; Montel 1993; Duc-Tin 2007; Stepanov et al. 2012; Duc-Tin and Keppler 2015) show
a dependence on temperature, pressure and magma composition, notably the H2O- and
phosphate-content, and the aluminum saturation index (ASI := molar Al/(Na + K +2 Ca)). As
explained above, the pressure dependence of the solubility has been attributed to changes in
complexation in the melt (Skora and Blundy 2012), but recent data for monazite and xenotime
(Duc-Tin and Keppler 2015) indicate that partial dissociation also occurs at low pressures—
more notably in peralkaline, less strongly in peraluminous melt (Fig. 5)—a fact that had
gone unnoticed in previous studies. In the pressure-composition range of partial dissociation
(0 < y < 1), orthophosphate (oph: monazite and xenotime) dissolution should be written as:
REE ( PO 4 )oph → (1 − y )  REE ( PO 4 )  + y  REE 3 + + ( PO 4 ) 

0 3−
(11)
  melt   melt
The P–T–X conditions of partial dissociation are not yet fully explored, but solubility is known
to increase with H2O-content and alkalinity (for ASI < 1.0), whilst it decreases with actinide
(Th + U) concentration and pressure. Solubility of monazite is most temperature-dependent for
LREE, less so for MREE (Stepanov et al. 2012; Duc-Tin and Keppler 2015). With increasing
dissociation (higher y-values), orthophosphate solubility increasingly depends on the
phosphate contents of the melt, which can vary widely—by at least two orders of magnitude
for mafic to felsic magmas, and from 0.1 to 1.5 wt% in granitic magmas alone (Pichavant et al.
1992; Breiter et al. 2008). Because REE and P2O5 are both incompatible, their concentration
depends on the degree of melting: At high melt fractions P2O5 concentrations are low, whereas
LREE are high; during crystallization they move in the opposite sense. Apatite solubility is
low in metaluminous melts ((ASI = 1), but increases with the ASI, owing to AlPO4 complexes
(e.g., Wolf and London 1994), and it is also elevated at low ASI, owing to alkali phosphate
complexing (Ellison and Hess 1988). In melts derived from apatite-saturated metasediments,
phosphate contents may initially be buffered by apatite solubility, but since monazite is
considerably less soluble in melt (and hydrous fluid) than apatite, monazite (and xenotime)
largely control the REE- (and Y-) content of melts and vice versa. LREE-rich apatite dissolves
incongruently (Antignano and Manning 2008; Tropper et al. 2011), so leaves behind a LREE-
enriched monazite. Leucosomes in migmatites may well preserve pre-anatectic monazite and
xenotime (Rapp and Watson 1986), or apatite in very hydrous melts (Zeng et al. 2005).
Recent work (Duc-Tin and Keppler 2015) shows substantial differences in solubility
among pure LREE-phosphates; even neighboring LREE can differ by a factor of two (Fig. 5).
This opens the possibility to refine our understanding and interpretation of zoned monazite
by analyzing the patterns for individual REE or groups thereof, particularly as solubility is
strongly temperature-dependent for LREE but less so for heavier REE.
Current understanding of the solubility of orthophosphates (Fig. 5), especially in granitic
systems and their REE evolution, allows some specific conclusions relevant to monazite
petrochronology. Attempts to model the behavior of monazite (and zircon) in the partial
melting range (Kelsey et al. 2008; Yakymchuk and Brown 2014) have combined computed
phase diagrams that show the stable phase relations, computed from thermodynamic models,
with the solubility equations. Yakymchuk and Brown used the updated solubility equation
(Eqn. 10) and also included open-system scenarios (with intermittent melt loss). The results
of both studies predict that partial melting of metapelitic and metapsammitic protoliths will,
depending on the exact P–T path, lead to dissolution of (most or all) monazite. Noting
that melt-isopleths typically have positive P–T slope, decompression is expected to lead to
dissolution, and monazite growth is expected only upon cooling. In ascending magmas, the
oldest monazite age most likely dates cooling after decompression, magma emplacement
or injection. This is certainly the most likely moment on a clockwise P–T–t path, say in a
collisional setting, where initial decompression may be attended by heating: Most monazite
is expected to crystallize sometime after cooling from Tmax, and it seems most unlikely that
one could use monazite to date the stage after Pmax but before Tmax is reached (Kohn et al.
2005). Things are more complicated if magmatic allanite is involved, i.e., in more calcic
compositions (e.g., granodiorite and tonalite). At late-magmatic stages, as partial melts
become enriched in volatiles, monazite resorption is likely7.
In addition, dissolution of phosphates can be kinetically limited, especially in dry melts,
where diffusion of LREE and phosphate species away from the dissolving solids is very slow
(Harrison and Watson 1984).
Th, Y, and U. These elements partition into monazite, much like the LREE, but to decreasing
degrees, i.e., Th most, U least. During crystallization, thorium is dominantly fractionated into
monazite, and Rayleigh-like fractionation occurs, except in more calcic magmas, where allanite
may interfere. Th/U ratios in monazite of course depend on Th/U in the magma, but also on
the modal abundance of coexisting phases competing for actinides, notably allanite—the other
main Th-sink—and zircon, xenotime, and titanite, all of which favor U over Th. In migmatites
and generally in granitoid magmas, fragments or corroded relics of pre-melting monazite (from
one or more prograde generations, discussed below) may survive even long periods of high
temperature and extensive melting (Copeland et al. 1988; Harrison et al. 1995; Kohn et al. 2005;
Dumond et al. 2015). Their eventual dissolution may explain Th-rich rims in magmatic monazite
or allanite. In the absence of such complications, the earliest magmatic monazites typically show
the highest Th-contents; subsequent growth zones are Rayleigh-distilled. Such patterns, reported
primarily from anatectic, often haplogranitic magmas, have been put to use in deciphering and
dating pre- and post-anatectic stages (Pyle and Spear 2003; Kohn et al. 2005; Corrie and Kohn
2011). Xenotime saturation, i.e., growth from melt or resorption in it, is critically dependent on
and inversely correlated with garnet and monazite, the main competitors for Y. Competition for
actinides during magmatic crystallization is reflected in apatite, which shows Th/U ratios that
vary inversely with the modal presence of monazite (and possibly allanite?) during crystallization
(Cochrane et al. 2014). Where apatite occurs as magmatic inclusions in zircon or titanite, its
composition carries useful petrogenetic information (Bruand et al. 2016).
Allanite
REE-zoning in igneous allanite has long been recognized to reflect fractional crystallization
(Levinson 1966), most commonly from a melt of high water contents (Johnston and Wyllie
1988; Beard et al. 2006). Experimental data on allanite solubility (Kessel et al. 2005; Klimm et
al. 2008; Hermann and Rubatto 2009) cover a range of melt compositions and pressures. Where
allanite forms in magma, it largely controls the LREE contents of the melt simply because it
contains LREE at weight-% level. For Th, U, and Y, as noted above, competition from coexisting
phases (zircon, monazite, xenotime, titanite or rutile) may intervene. As summarized by Smye
et al. (2014) REE-fractionation in allanite depends on magma composition: The most LREE-
enriched allanites are found in Ca-rich types (e.g., carbonatite); La/Nd and Ce/Nd are high for
diorite and granodiorite, intermediate for granite, and lowest for syenite. Allanites in rhyolitic
obsidian often show La/Yb 250–2300 and beyond, but only 3 to ~100 for the glass (Mahood
and Hildreth 1983). Typically, chondrite-normalized REE-patterns are steep for early-formed
allanite, but depletion of LREE then leads to nearly flat patterns, and the overall core-to-rim
zoning in allanite may end with REE-enriched (or even REE-poor) peripheral epidote (e.g.,
Oberli et al. 2004). REE-contents (normalized to chondrite) in > 1700 analyses compiled by
Gieré and Sorensen (2004) show typical ranges for LREE vs. MREE (e.g.La: 104–2·105, but
Sm: 4·103–6·104 only). Allanite preference for LREE is exceeded (in absolute concentration)
only by its appetite for Th, i.e., allanite fractionates Th/La relative to the melt (Mahood and
Hildreth 1983; Klimm et al. 2008; Hermann and Rubatto 2009).
7
However, contrary to previous belief, fluorine in melt was shown to have no effect on solubility of monazite and xenotime,
indicating no fluoride complexing of REE (Duc-Tin and Keppler 2015).
378 Engi
The actinide ratio Th/U is often used as a diagnostic means to distinguish magmatic
from metamorphic allanite growth zones: Typical magmatic values are Th/U > 100(Gregory
et al. 2007), metamorphic ones are < 50 (Gregory et al. 2009), except under very oxidizing
conditions, where U4+ may lead to Th/U >> 100 (Janots and Rubatto 2014). Trends during
magmatic differentiation depend on competing minerals; for example, Th/U increases in
allanite, whereas it decreases in coexisting titanite. Fractional crystallization of allanite can
produce strong zoning in major and minor elements. Figure 6 shows typical compositional
data, crystal morphology, and zoning in igneous allanite from a granodiorite. In a tonalite
studied by Oberli et al. (2004) allanite shows a striking magmatic evolution, from dark brown
core to faintly colored rim, in wt%: REE2O3 17.6 → 1.9, CaO 13.7 → 23.6, Fe2O3 5.3 → 11.2,
ThO2 1.4 → < 0.1. In mafic rocks at high pressure allanite plays a leading role in sequestering
REE and actinides. Data for eclogites show that allanite stores 90–99% of a rock’s budget in
LREE, Th, and U, > 60% of its Sm and still about a third of its Eu (Hermann 2002). Allanite
finds its main competitor for LREE + Th in the composition range where monazite coexists.
This is discussed below jointly with the evolution of the phosphates, since REE-fractionation
between these and allanite is petrochronologically significant.
Subsolidus petrogenesis
Apatite, monazite, and xenotime are known to occur from diagenetic conditions (e.g.,
Kingsbury et al. 1993; Evans and Zalasiewicz 1996; Knudsen and Gunter 2002; Rasmussen et
al. 2002; Vallini et al. 2002; Rasmussen 2005) all the way up to upper amphibolite, granulite,
and eclogite grades (e.g., Black et al. 1984; Bingen et al. 1996; Zhu and O’Nions 1999;
Rubatto et al. 2001; Erickson et al. 2015; Gasser et al. 2015; Tucker et al. 2015).
Diagenetic apatite, monazite and xenotime have proven datable (Evans et al. 2002;
Rasmussen 2005; Davis et al. 2008), and ages can be very helpful in understanding diagenetic
processes involving phosphates. Remarkably, diagenetic xenotime has proven to preserve its
formation age (Vallini et al. 2002), resisting later penetrative deformation and a greenschist
facies overprint. This is most welcome because textural relations are complex (Fig. 7), so
present a challenge to infer specific irreversible reactions and link them to P–T conditions. In
metaclastic sequences, accessory minerals at anchizonal conditions typically include mostly
authigenic, diagenetic and/or detrital monazite, apatite, and xenotime, in addition to detrital
zircon, but detrital cores show monazite or xenotime overgrowths (Fig. 7 A, F) that form in the
lower greenschist (Wing et al. 2003; Rasmussen and Muhlig 2007; Janots et al. 2008, 2011;
Allaz et al. 2013) and low-temperature blueschist facies (Janots et al. 2006). Low-grade apatite
tends to be high in Th and U, and it is the main REE reservoir in deep sea mud (Elderfield and
Pagett 1986; Kato et al. 2011; Kon et al. 2014). Relative to their (mostly clastic) precursors,
low-grade metamorphic monazite is typically low in Th and U, depleted in HREE, but
enriched in LREE. In many instances, apatite, thorite, and xenotime are visible as satellites
or intergrown small grains, indicating local reactions and spatially limited redistribution of
REE and actinides over some 50–100 µm. Reaction volumes reach much larger dimensions
(mm to cm) in high-strain zones or where ample fluid interacted with the rocks (Janots et
al. 2012; Seydoux-Guillaume et al. 2012), even at low temperatures. Most growth zones are
small (< 10 µm), however, and in situ chronometry in very low-grade metamorphic monazite
(< 350 °C) is probably best done by EMPA (Fig. 7 G, H).
Above the lower greenschist facies, grain coarsening and replacement reactions typically
enhance petrochronological options for REE minerals. Grain to grain heterogeneity in trace
element contents and isotopes of detrital phosphates, which can hinder correlations, tends to
disappear due to such processes above ~500 °C (Hammerli 2014; Hammerli et al. 2014, 2016).
Figure 6. Magmatic allanite from Cima d’Asta pluton (CAP), a commonly used standard reference mate-
rial (Burn 2016). (A) BSE image of allanite fragment shows weak oscillatory zoning in core; inclusions are
apatite, zircon, and quartz; circles mark spots of LA-analysis. (B) X-ray maps (La, Th) of central part (white
frame in A) show chemical zoning due to fractionation from melt during growth; rhythmic banding reflects
inverse correlation of La and Th (a.p.f.u.: atoms per formula unit). From core to rim, elements with Th-
affinity increase (1.1 wt% ThO2 in core, 1.6% in rim) , those with La-affinity decrease in concentrations. (C)
Based on the La–Th diagram five magmatic growth zones were identified. (D and E) U–Th diagrams show
core to rim evolution and identify two core zones (I, II) based on their Th/U ratio and Th-contents. Their
Th/Pb ages are within error, based on LA-ICP-MS data, whilst ages can be resolved between core I and rim
(D). All data from Burn (2016). Ages overlap with previous TIMS and LA-ICP-MS analyses (Barth et al.
1994; Gregory et al. 2007), though cores and rim were not separately treated in these (compare Figure 12).
380 Engi
Monazite and allanite

Precursors to metamorphic monazite and xenotime include, in addition to their own
ancestors, various low-temperature Al-phosphates, carbonates, and phyllosilicates; textures
commonly indicate dissolution-reprecipitation processes (Fig. 7G,H), but few studies have
worked out phase relations (e.g., Janots et al. 2006; Spear 2010) or attempted local mass
balance for REE and actinides—a topic in need of study. In the greenschist or low-temperature
blueschist facies, monazite in clastic metasediments breaks down to form allanite + apatite.
Various reactions (Table 2) have been suggested on petrographic grounds, and P–T conditions
for the first appearance of allanite clearly depend on rock composition. The limiting reactions
of REE-phases have not been studied by experimental reversals, so remain inadequately
known. For a small part of the relevant composition space experimental stability relations are
available (Janots et al. 2007; Hermann and Spandler 2008) and can be compared (Fig. 8) to
those computed from thermodynamic data, which were calibrated using natural assemblage
data as well (Spear 2010). Effects of the bulk rock composition are evident when comparing
Figure 8A and 8B: allanite is stabilized by an increase in CaO content, and substitutions
such as Fe–Mg and La–Nd have a major effect on its stability relative to monazite. These
thermodynamic models are a promising start, but at this point none of them adequately
reproduce the reaction sequences documented in metamorphic rocks.
A next generation of monazite is known to grow at conditions around the garnet or staurolite
isograd in rocks that previously contained allanite (e.g., Smith and Barreiro 1990; Catlos et
al. 2001; Spear and Pyle 2002; Wing et al. 2003; Corrie and Kohn 2008; Janots et al. 2008),
but it can also grow at lower grades in true (very low-Ca) metapelites and -psammites that did
not contain allanite (e.g., Kingsbury et al. 1993). Allanite can persist to higher temperatures in
Ca-rich rock types (Ca/Na > 1) and at high-pressure conditions (Janots et al. 2008; Kim et al.
2009; Radulescu et al. 2009), but phase relations remain poorly understood. Allanite grown in
the eclogite facies can be Sr-rich (Rubatto et al. 2008; Cenki-Tok et al. 2011) due to feldspar
reacting to sodic pyroxene. Typical prograde reaction textures are shown in Figure 9 from
amphibolite facies metapelites and in Figure 10 from eclogite facies micaschists.
Monazite formation in the amphibolite facies has been documented from many rock types
and pressure regimes, in some cases reflecting a single growth stage (e.g., Kohn and Malloy
2004), in others showing polyphase growth (Rubatto et al. 2001; Pyle and Spear 2003; Foster
et al. 2004; Kohn et al. 2005; Finger and Krenn 2007). Grains that formed early in a growth
sequence tend to sequester Th and Y, zones added later thus are more depleted, but since
precursors and reaction mechanisms can change, patterns are not always so simple. Unless
other accessories intervene, REE- and Y-patterns of monazite often reflect strong partioning
with coexisting major assemblage minerals, notably garnet and feldspar, and monazite REE-
patterns essentially reflect the modally abundant REE-competitors. In granulites, for instance,
monazite that formed after garnet can become very depleted in HREE (Foster et al. 2000;
Holder et al. 2015), a trend noted in apatite as well (Bea and Montero 1999).
Upon partial melting, monazite tends to dissolve and may or may not completely disappear;
it typically grows during crystallization of leucosome in migmatites (Williams 2001; Pyle and
Spear 2003; Kohn et al. 2005; Yakymchuk and Brown 2014).
Since allanite and monazite can form at several stages, and thus samples commonly
contain several (over)growth zones, chronometry is of course tempting wherever these
are of sufficient size for in situ analysis. Constraining the physical conditions of growth
stages is thus a central issue, and in the absence of thermodynamic models for the REE-
minerals involved, textural relations (e.g., core–rim) and local assemblages of coexisting
minerals currently are our best guides. Detailed observation of inclusions and zoned relics
of characteristic phases (e.g., garnet, mica, corroded accessories) may indicate a relative
Figure 7. BSE images of low-temperature xenotime and monazite. (A, B) Authigenic/diagenetic xeno-
time overgrowths on detrital zircon grains and rounded detrital zircon completely engulfed in xenotime;
SHRIMP dates of two growth rims yielded 1693 ± 4 and 1645 ± 3 Ma (Vallini et al. 2002, 2005). Samples
are from the Mount Barren Group, Western Australia. (C, D) Greenschist-grade monazite (light) in a black
shale with abundant pyrite euhedra in matrix; SHRIMP analysis (oval pits) yielded ages of 2416 ± 12 Ma
for monazite, and 2430 ± 19 Ma for xenotime. Samples are from the Neoarchaean Roy Hill Shale Member,
Jeerinah Group, Pilbara Craton, Western Australia (Rasmussen et al. 2005). (E) Detrital zircon with euhe-
dral xenotime overgrowths, quartz muscovite schist, Palaeoproterozoic Mount Barren Group (Rasmussen
et al. 2011). (F) Xenotime nucleating on zircon fragment in a metapelitic blueschist (350 °C, 1 GPa) from
Rif, Morrocco (Janots et al. 2006). Texturally similar outgrowths were reported by Rasmussen (2005,
Fig. 8) from Carboniferous Lower Coal Measures (UK). (G,H) Variscan monazite core (~320 Ma) with thin
Alpine rim (< 30 Ma) grown in metamarl at chloritoid grade (440 °C, 0.6 GPa). EPMA data across profile
A–B show rim is low in Th, Y, U, and radiogenic Pb. Sample from Garvera, Central Swiss Alps. Data from
(Janots et al. 2008). Photo credits: Birger Rasmussen (A–E) and Emilie Janots (F,G).
382 Engi
Table 2A. Proposed prograde reactions involving monazite, allanite, apatite, and xenotime.
Label Reactants Products Reference
M0 flo ± xen mnz + … Janots et al. (2006)
A1 flo + mnz + carb aln + syn + … Janots et al. (2006)
A2 mnz + mus + carb + qz aln + bio + ap Wing et al. (2003)
A3 mnz + chl + pl + cc + qz aln + ap + … Wing et al. (2003)
A4 mnz + chl + cc ± hem aln + xen + ap + ctd Janots et al. (2008)
A5 mnz + chl + an ± hem aln + xen + ap + ctd Janots et al. (2008)
A6 mnz + … aln + xen? Smith and Barreiro (1990)
A7 mnz + pl + … aln + ep + tho + ap Finger et al. (1998)
A8 mnz + chl + … aln + ctd Radulescu et al. (2009)
A9 mnz + bt + … aln + F-ap + qtz + … (Yi and Cho 2009)
M1 aln + apa + mu + AS + qz mnz + bio + pl + H2O Wing et al. (2003); Ferry (2000)
M2 aln + apa + mnz + … Yang and Pattison (2006)
M3 aln + apa mnz + pl + mag? Tomkins and Pattison (2007)
M4 aln + apa + AFM1 mnz + an + AFM2 Janots et al. (2008)
M5 gar + apa + mu mnz + sill + bio Pyle and Spear (2003)
M6 gar + chl + mu mnz + st + bio + pl Kohn and Malloy (2004)
Table 2B. Pressure–temperature conditions estimated for allanite- and monazite-forming reactions
Label Temperature Pressure Thermobarometry Isograd Rock type
M0 350–420 °C 1.0–1.2 GPa ctd–chl thermometry metapelite
A1 300–350 °C 0.3–1.0 GPa ctd–chl thermometry metapelite
A2/3 ~400 °C 0.35 GPa Ferry (1994) bio–in metapelite
A4/5 420–450 °C TWQ ctd–in metapelite/marl
A6 500–550 °C 0.28–0.34 GPa gar–bio bio–in metapelite
A7 550–600 °C 0.4–0.5 GPa regional? metagranitoid
A8 540–600 °C 2.0 GPa phase diagrams micaschist
A9 500–600 °C phase relations (vein) leucogneiss
M1 ~500 °C 0.35 GPa Ferry (1994) and/ky–in metapelite
M2 480–520 °C 0.20–0.44 GPa Helms and Labotka (1991) ~gar–in metapelite
M3 530–580 °C 0.30 GPa Pattison and Vogl (2005) ~cord–in metapelite
M4 530–580 °C 0.8–0.9 GPa TWQ metapelite
M5 580–605 °C 0.3–0.4 GPa gar–mnz thermometry metapelite
M6 ~600 °C 0.6–0.8 GPa stau–in metapelite
Minerals. AFM: Al-Fe-Mg-silicates (bio, mus, gar), aln: allanite, an: anorthite, and: andalusite, ap: apatite,
AS: aluminosilicate, bio: biotite, carb: carbonate, cc: calcite, chl: chlorite, cord: cordierite, ctd: chloritoid, ep:
epidote, flo: florencite, gar: garnet, hem: hematite, ky: kyanite, mag: magnetite, mnz: monazite, mu: musco-
vite, pl: plagioclase, qz: quartz, st: staurolite, syn: synchisite, tho: thorite.
Figure 8. Models for the stability fields of allanite and monazite in P–T space based on thermodynamic data
(A,B) and experimental phase equilibria (C). Mineral abbreviations: Aln = allanite, Am = amphibole, And
= andalusite, Ap = apatite, Bt = biotite, Cpx = clinopyroxene, Chl = chlorite, Cs = coesite, Ctd = chloritoid,
Dis: dissakisite, F: hydrous fluid, Gt = garnet, Kfs = K-feldspar, Ky = kyanite, L: melt (64–67 wt% SiO2),
Lmt = laumontite, Lw = lawsonite, Mnz: monazite, Phe = phengite, Qtz = quartz, Opx = orthopyroxene, Sil
= sillimanite, Tc = talc, WM = white K-mica, Zo = zoisite. (A): Phase relations calculated for two compo-
sitions typical of clastic metasediments (both 16.57 wt% Al2O3; left: average pelite, 2.17 wt% CaO; right:
slightly marly, 4.34 wt% CaO) using a model for Ce-allanite and Ce-monazite (Spear 2010). The predicted
reaction sequence is consistent with those observed in several terrains above 500 °C, but no low-temperature
monazite stability field (below 400–450 °C, Janots et al. 2008) is predicted. (B) Calculated medium-pressure
stability field of dissakisite (La–Mg allanite—cliozoisite solution) (+ apatite) against various assemblages
comprising La-monazite in calcic pelite model composition based on calorimetric data (Janots et al. (2007).
(B) High-pressure field of Mg-allanite in phengite-eclogite model system (KNCMASH + trace elements),
based on experimentally determined phase relations. Allanite first appears in experiments at temperatures
just below the zoisite-out reaction. Disappearance of allanite at high temperatures is by gradual dissolution
into melt, not linked to the major phase reactions; after Hermann (2002). Blue dot–dashed boundary is for
a pelitic model system; this limit of allanite versus monazite was inferred by Hermann and Rubatto (2009)
from experiments by Hermann and Spandler (2008).
384 Engi
Figure 9. X-ray maps of REE-minerals at middle amphibolite facies, northern Lepontine belt (Central
Alps, Switzerland). Textures indicate partial breakdown of allanite (+apatite) to monazite + xenotime.
No apatite appears in the assemblage, and its consumption probably terminated the reaction (at ~560 °C,
0.8 GPa; Janots et al. 2008). Color bars (top right) indicate concentrations of elements in maps, with cold
colors indicating low contents, red being maximum; typical ranges are shown for main REE-minerals. (A)
La-rich allanite cores are preserved, and splay monazite aggregates interwoven with tiny xenotime are
probably pseudomorphic after allanite. (B) Similar mineral textures, but preserved allanite cores are rich
in Nd and Th (note heterogeneity). Both textures in same sample attest to very limited mobility of REE
and actinides even during mineral reactions involving fluid. Heterogeneity in and among reaction domains
poses problems in U–Th–Pb chronology (e.g., common Pb). BSE images courtesy of Daniele Regis.
Figure 10. Allanite in eclogite facies phengite quartzite (Sesia Zone, Western Alps). Textures in BSE images
show 3 HP-stages (I-III). (A) Corroded core of allanite-I (75 Ma) includes Phe-I (Si: 3.45 a.p.f.u.1, 540 °C,
1.8 GPa), aln-II overgrowth (68 Ma) contains Phe-II (Si: 3.20 apfu, < 520 °C, < 1.5 GPa), and thin Y-epidote
rim (60–65 Ma) is in equilibrium with matrix assemblage (550 °C, ~1.8 GPa); Ages are from SHRIMP Th/
Pb isotopic analyses. Data indicate pressure-cycling (yoyo tectonics) in a subduction channel (Rubatto et al.
2011). (B, C) Samples from the same unit display similar complex resorption and replacement textures in al-
lanite core, with Y-epidote rim. Relic monazite can be preserved inside allanite aggregates. (Regis et al. 2012).
(D, E) Compositions of growth zones visible in allanite from (C). Successive generations show decrease in
REE contents and increase in ferric fraction (Fe3+/Fetotal: labels along radial lines in diagram) (after Petrík et
al. 1995). Clz = clinozoisite, Ep = epidote, Aln = allanite. (E, F) Minor and trace element compositions in
overgrowth sequence shown in (C). REE patterns and the sequestration of the critical minor elements Sr, Y, Th
and U indicate that monazite was the main precursor of allanite. While allanite grew at eclogite facies condi-
tions, in the absence of feldspar, REE patterns of aln-I and aln-II reflect LREE enriched monazite composition
and preserve its Eu-anomaly. Note strong Th/U fractionation. All data from Regis et al. (2014). Chondrite
normalization based on Sun and McDonough (1989). BSE images courtesy of Daniele Regis.
1
atoms per formula unit KAl2−xMgx[Al1−xSi3+xO10](OH)2
386 Engi
paragenetic sequence, and their geochemical characterization often allows estimates of P–T
conditions (Regis et al. 2014; Mottram et al. 2015). However, textures within high-grade
monazite (and less commonly allanite) tend to be complex, often patchy, and age relations
are not always clear. Microstructural relations of allanite and monazite (and their inclusions)
to metamorphic fabrics can be invaluable to relate P–T–t relations to deformation (Manzotti
et al. 2012; Regis et al. 2012; Goswami-Banerjee and Robyr 2015; Wawrzenitz et al. 2015);
considerable progress is being made relating monazite and allanite growth to deformation
events (e.g., Ayers et al. 2002; Williams and Jercinovic 2002; Berman et al. 2005; Dahl et
al. 2005; Simmat and Raith 2008; Kirkland et al. 2009; Cenki-Tok et al. 2011, 2014; Kelly
et al. 2012; Wawrzenitz et al. 2012; Didier et al. 2013, 2014). Subsolidus retrograde growth
of monazite (± xenotime) at the expense of allanite or apatite + garnet has been documented
in several instances; it is usually related to localized or pervasive fluid-influx, and such
monazite has proven datable (Bollinger and Janots 2006; Tobgay et al. 2012). From similar
environments, but also in retrograde corona formation from high-grade rocks, monazite
breakdown to allanite + apatite has commonly been reported (Broska et al. 2005; Krenn et al.
2012; Ondrejka et al. 2016) and occasionally dated (Yi and Cho 2009).
Xenotime. Rasmussen et al. (2011) studied a metaclastic sequence (Mount Barren Group)
in SW Australia and found xenotime closely associated with detrital zircon, commonly as
syntaxial outgrowths. Up to ~450 °C, detrital and diagenetic xenotime is partly preserved,
but wholly metamorphic grains largely replace these at higher temperatures. Dissolution–
reprecipitation reactions produced compositionally distinct rims forming at greenschist and
amphibolite facies conditions. As in situ U–Pb chronology of xenotime yielded at least four
discrete age groups (2.0 to 1.2 Ga), a discontinuous sequence of reactions is likely to be
responsible for the growth of metamorphic xenotime, at the expense of detrital and diagenetic
precursors. A few specific reactions have been formulated so far (Table 2). Metamorphic
xenotime shows distinctly lower Th and U contents, in some instances Th-rich annuli, and tiny
U- and Th-rich product grains (e.g., Figs. 3 and 8 in Rasmussen et al. 2011). Similar phase
relations were reported by Allaz et al. (2013) and, in eclogite facies metaclastics, by Regis et
al. (2012). The latter study showed zircon to be one of the reactant precursors and identified
thorite and aeschynite among the satellite products. In more calcic domains of the same
samples, monazite and xenotime broke down to produce allanite. Phase relations (Fig. 11)
calculated for metapelites from a thermodynamic basis (Spear and Pyle 2010) that includes
(preliminary) data for the REE-phosphates, show approximate stability limits of xenotime in
the subsolidus range. To distinguish different xenotime generations, REE are useful: compared
to magmatic xenotime, which is typically HREE-enriched, at subsolidus conditions enrichment
increases from LREE to MREE, and patterns from MREE to HREE are either flat or decrease.
Monazite–apatite–xenotime textural relations. Mutual inclusions and overgrowth relations
are often observed among REE-phosphates (and allanite) and can be of great utility in relating
ages to (partial) replacement reactions and metasomatic effects, e.g., related to ore formation. For
instance, inclusions of monazite and/or xenotime commonly occur in apatite, and using chemical
fingerprinting (halogens, Sr, Pb), it is possible to track fluid–rock interaction (Harlov and Förster
2003; Harlov et al. 2005; Broska et al. 2014; Shazia et al. 2015; Jonsson et al. 2016).
Effects of hydrous fluids
Apatite, monazite, and xenotime are commonly found in fluid-rich environments, both at
low and high temperature (Harlov et al. 2008; Hetherington and Harlov 2008). The solubility
of phosphate minerals in fluids is strongly composition-dependent, and substantial evidence
from experimental studies (Harlov et al. 2005, 2007, 2011; Budzyń et al. 2011; Seydoux-
Guillaume et al. 2012; Grand’Homme et al. 2016b) indicates that dissolution-reprecipitation
processes are particularly effective when alkaline hydrous fluids are involved; in many cases
F-apatite is a characteristic product phase, and xenotime as well as monazite are involved.
Figure 11. Phase relations calculated for a model pelite composition in K2O–Na2O–CaO–MnO–MgO–
FeO–Al2O3–SiO2–H2O–Y2O3–Ce2O3–P2O5–Fluid from Spear and Pyle (2010); the model is substantially
founded on natural assemblages from a specific locality in Maine, includes data for pure xenotime and
binary Ce-Y-monazite, but does not consider allanite. Stability limits for xenotime depend on bulk Y-
contents; a realistic field is shown. Mineral abbreviations as in Figure 8, plus AlSi = kyanite or sillimanite,
Crd = cordierite, St = staurolite.
For chronologic efforts, partial replacement by such processes can be a challenge (Williams
et al. 2011), but interaction with pervasive fluid may be a most effective way to completely
reset clocks in phosphate minerals (Harlov and Hetherington 2010; Harlov et al. 2011). These
studies have spawned efforts to date hydrothermal systems (Williams et al. 2011; Shazia et
al. 2015), and indeed several studies have thus investigated Alpine clefts (Janots et al. 2012;
Grand’Homme et al. 2016a) and altered shear zones (Berger et al. 2013).
Particular geochemical characteristics of REE-minerals, e.g., the tetrad effect (Bau
1996) have long been attributed to effects of fluids, but Duc-Tin and Keppler (2015) showed
convincingly that this effect can be explained by the orbital structure of particular LREE being
particularly favorable for the irregular REE-coordination polyhedra in monazite and xenotime
(Fig. 1). Their REE partitioning behavior—relative to melt, hydrous fluids or minerals with
regular REE polyhedra—is thus different than expected solely from ionic radius considerations
(Fig. 5). As a consequence, and contrary to previous work (Bau 1996), tetrad patterns may not
directly indicate the effect of fluids but of crystal-structure dependent fractionation.
CHRONOLOGIC SYSTEMS
Owing to the elevated actinide contents of the REE-minerals, chronometry has been
based on the 208Pb/232Th, 206Pb/238U and 207Pb/235U (or 207Pb/206Pb) ratios far more often than
on alternative isotopic systems. As discussed above, the closure temperatures for these systems
are not precisely known, but are certainly high enough for chronometry in many applications
at subsolidus environments and, with caution, into the partial melting range. For apatite, U–Pb
closure is estimated at lower temperature (350–550 °C; Cochrane et al. 2014; Chew and Spikings
388 Engi
2015), suitable only for dating crystallization conditions of upper crustal and blueschist facies
samples. Lu–Hf (Tc 675–750 °C) is an alternative for higher temperatures, but has been rarely
applied so far (Barfod et al. 2003; Larsson and Söderlund 2005). And Sm–Nd dating has been
applied to apatite based on intersectoral differences in partitioning (Rakovan et al. 1997).
Spatial resolution versus age resolution
Compared to ID-TIMS analysis, all in situ techniques have inferior age resolution and
precision, but many studies focus on answering questions that demand high spatial resolution.
For many samples and problems, finding grains of sufficient homogeneity and size for TIMS
dating ranks somewhere between challenging and impossible. In other cases, the structural
context can be preserved at least in part, with effort, by microdrilling (Kohn et al. 2005;
Smye et al. 2011) or directly cutting monazite or allanite from a polished section. In this
way, monazite has been successfully isolated and then analyzed with optimal precision using
ID-TIMS (Hawkins and Bowring 1997; Baldwin et al. 2006). Allanite can also be separated
into several age groups or magmatic generations based on the color and density of mineral
fragments; precise ID-TIMS analysis of these then give a range of ages (Oberli et al. 2004).
These indicate the timescale of plutonic evolution, though the specific significance of each
age is not clear because no textural link to geochemical tracers is possible. So, precision
is excellent in ID-TIMS studies, but context is sacrificed, and thus petrogenetically critical
information for each fraction analyzed is missing.
Alternatives include a range of micro-analytical methods, some of which allow in situ dating
of growth zones in polished sections. EPMA age mapping of monazite is a powerful way to resolve
first-order geological problems, especially in polymetamorphic rocks with a complex evolution
(Williams et al. 1999; Hermann 2002; Buick et al. 2010; Dumond et al. 2015; Johnson et al. 2015;
Tucker et al. 2015) and in rocks that experienced extensive interaction with fluid (Shazia et al.
2015; Grand’Homme et al. 2016a; Kirkland et al. 2016). Allanite maps have similar potential,
especially in high-pressure rocks and migmatites (Gregory et al. 2012; Regis et al. 2014), but
have been much less utilized so far, since EPMA dating is not possible, and SIMS or laser-based
analysis has lower spatial resolution and demands considerably more analytical effort.
U–Th–Pb
Monazite and xenotime. Monazite often contains several wt% Th, several thousand ppm U,
and usually negligible initial lead (Montel et al. 1996). These high actinide concentrations result in
high levels of radiogenic lead, except in very young samples. In xenotime Th and U contents are
lower, but both monazite and xenotime are proven U–Th–Pb chronometers. Monazite and allanite
normally contain both Th and U in concentrations sufficient to combine their decay systems for
chronometry; for xenotime only U–Pb is possible. In both phosphates, REE contents are also high
(e.g., 104–105 ppm Sm and Nd), and as tracer isotopes they add important geochemical constraints,
especially combined with chronometry (Nemchin et al. 2013; Whitehouse et al. 2014).
The commonly high Th- and U-contents allow chemical U–Th–Pb dating of monazite and
xenotime by EPMA (Suzuki and Adachi 1991; Montel et al. 1996; Williams and Jercinovic
2002; Pyle et al. 2005b; Williams et al. 2007; Hetherington et al. 2008; Spear et al. 2009). The
power of the approach, its limits, technical aspects, and some applications are not reviewed
here, as they are specifically addressed in this volume (Williams and Jercinovic 2017); similarly,
isotopic dating by LA-ICP-MS and SIMS are covered separately (Kylander-Clark 2017; Schmitt
and Vazquez 2017). Two developments are particularly suited to link ages with their geochemical
and textural context: LASS ICP-MS (laser ablation split stream: Kylander-Clark et al. 2013;
Goudie et al. 2014) and Scanning Ion Imaging using SIMS (Harrison et al. 2002).
In situ isotopic analysis by LA-ICP-MS or ion probe / SHRIMP involves particular concerns
for each instrument, notably regarding standardization, fractionation, matrix correction, and
realistic error propagation. The magnitude of the effects varies for each type of mineral; in
a general way, monazite and xenotime pose fewer problems in isotopic analysis than does
allanite, but successful analytical protocols for all of them have been required. In particular,
the question of matrix-matched8 standardization has been repeatedly raised (Kohn and Vervoort
2008), especially for allanite (Catlos et al. 2000), because of its large chemical variability. Since
matrix-matching is more critical in ion probe analysis than in laser ablation analysis (Gregory
et al. 2007; Darling et al. 2012), LA-ICP-MS may be the preferred method, given the paucity
of suitable allanite standards (Smye et al. 2014). Fortunately, it now appears that the concerns
over matrix-matching and standards in allanite analysis can be largely avoidable with carefully
optimized LA-ICP-MS conditions (McFarlane 2016). However, matrix-matching remains
essential in ion probe analysis (e.g., Fletcher et al. 2004).
Apatite
Apatite contains several thousand ppm REE and has typically high Lu/Hf ratios, making
Lu–Hf dating possible (e.g., Barfod et al. 2005), with high closure temperatures similar to
U–Pb in monazite (e.g., McFarlane and McCulloch 2007). Apatite also has relatively high
U-contents, most useful in U–Pb thermochronometry (Chamberlain and Bowring 2001; Chew
et al. 2011; Chew and Donelick 2012), often in conjunction with (U–Th)/He and/or fission track
dating. These applications are beyond the scope of this review; for a short introduction consult
Cochrane et al. (2014). Since metamorphic samples often contain more than one generation of
apatite, U–Pb or Th–Pb isotopic analysis should be done by in situ methods allowing sufficient
spatial resolution. Ion probe methods (Sano et al. 1999) or, more commonly, LA-ICP-MS
(Willigers et al. 2002; Chew et al. 2011) has been used and is sufficiently precise to date low-U
apatites, such as from mafic dykes (Pochon et al. 2016). Sr and Nd isotopes can be analyzed in
situ, offering new possibilities of petrogenetic utility (Hammerli et al. 2014)
Allanite
Allanite usually contains 1,000–16,000 ppm Th and typically about a tenth that amount
in U. Th/U > 2 is common, but shows substantial variation and may give genetic indications.
Allanite has shown its potential as a robust Th/U–Pb chronometer, but several challenges
remain, especially its high initial lead contents. Unlike monazite and xenotime, which
commonly contain nearly 100% radiogenic lead, allanite (and apatite) have an affinity for Pb,
and isotopic data used for dating must be corrected for this component. This is straightforward
for ID-TIMS data using 204Pb where mass interferences are eliminated. In SHRIMP analysis,
energy filtering effectively reduces isobaric interference (Rubatto et al. 2001) on mass 204,
but spot analyses obtained by LA-ICP-MS combine 204Pb and 204Hg. Several methods are
in use for estimating the composition of the initial lead component (Pbini), which includes
inherited radiogenic and non-radiogenic lead; some are based on models, others on analysis
of the sample being dated. As petrochronology typically aims to resolve local growth stages,
and relics are the rule, not the exception, overall isotopic equilibration cannot be assumed, and
methods relying on a single type of common lead (Wendt 1984) are not applicable in general9.
If the fraction of non-radiogenic Pb is high—and in allanite (Gabudianu Radulescu et al.
2009a; Gregory et al. 2012) and apatite (Cochrane et al. 2014) it often is in the range 20–
70%, but may be > 90%—the common lead correction affects ages and their uncertainty very
substantially. In addition, minerals with high Th/U contain excess 206Pb produced from excess
230
Th (an intermediate daughter nuclide in the 238U chain) incorporated during growth, for
which a correction is needed (Schärer 1984; Barth et al. 1994). These complications enlarge
age uncertainties and make it difficult to check for possible Pb-loss, which may occur at
temperatures above ~750–800 °C (Kamber et al. 1998).
8
Both structurally and chemically
9
This warning does not apply where the analytical uncertainty in the common Pb measurement exceeds the effect of potential
variations in its composition.
390 Engi
Initial lead. A complete discussion of the issues and controversies surrounding common
Pb is beyond the scope of this review, but data presented in the literature are commonly based
on premises that have since been questioned. As these can affect ages substantially, one should
understand the basic problem involved in estimating the composition (isotopic ratios) of the
initial lead, know the approaches used to solve them, and consider at least some of the perils
and implications. This is true for all chronometers containing high levels of non-radiogenic
lead, notably Ca-rich minerals, where Pb can replace Ca. The specific situation for allanite is
presented in Gregory et al. (2007, 2012). In essence, for both the Th–Pb and the U–Pb system,
two approaches are used: A single-spot correction or an isochron method. The single-spot
correction assumes a model Pb composition that does not change during allanite growth, and
whose isotopic ratio is predicted by a global lead evolution model (Cumming and Richards
1975; Stacey and Kramers 1975). Even though local dissolution of accessory precursors are a
likely staple on which allanite fed (Romer and Rötzler 2011), single spot correction has been
widely used to treat allanite dates. An alternative practice is to determine Pbini composition
from internal heterogeneity within an allanite crystal. This approach assumes a single growth
stage with a common age but variable chemistry. Allanite commonly shows variable U/Th and
common lead, so the isochron correction usually uses a Tera-Wasserburg diagram approach
(plotting 207Pb/206Pb vs. 238U/206Pb). Data showing variably non-radiogenic Pb (Fig. 12)
are used to obtain the age (2007; Gregory et al. 2012). The common practice to determine
common lead composition by analyzing both U-rich and coexisting U-poor minerals (e.g.,
feldspar) assumes isotopic equilibration in the same compositional domain as the mineral
chronometer; within the context of an assumed single age, the isochron method allows this
assumption to be tested. As pointed out by Burn (2016), it is advisable to combine these
correction methods and use them to check the quality of data by first processing them in an
uncorrected Tera-Wasserburg and a 206Pbc normalized isochron diagram (Fig. 12). Provided a
dataset from different parts of a sample define a linear array in a Tera-Wasserburg diagram, the
intercept with the y-axis defines the 207Pb/206Pb initial lead composition. Each diagram returns
an estimate of the age and the initial lead composition of allanite (or any other mineral with
high common lead contents, such as apatite; Chew et al. 2011). The concordance of these two
age estimates is used to evaluate the quality of the data and estimate the age uncertainty.
During growth, both allanite and apatite may incorporate extraneous Pb; if growth is from a
melt, isotopic equilibrium is often assumed, perhaps unjustly, as indicated by pre-melting relics
in leucosomes from migmatites (Berger et al. 2009). At sub-solidus condition, metamorphic
growth may occur in local domains or from isotopically inhomogeneous precursors (Lanari and
Engi 2017), and local lead isotope ratios may be quite different from those of the bulk rock,
let alone of a terrestrial evolution model (Stacey and Kramers 1975). Inheritance is of major
concern in polymetamorphic basement rocks (Romer and Siegesmund 2003; Romer and Xiao
2005; Romer and Rötzler 2011), but since many metamorphic rocks contain relic accessory
minerals, local variability should always be tested because local digestion of such relics may
cause major deviations from bulk rock U/Pb, Th/Pb and 207Pb/206Pb (e.g., Wawrzenitz et al.
2015). Inheritance of this kind is evident in datasets as excessive scatter in the U/Pb or Th/Pb
ages, i.e., scatter beyond the purely analytical uncertainty (~0.1% for ID-TIMS, 2–3% for LA-
ICP-MS or ion probe data). Major discrepancies between measured and model lead compositions
have repeatedly been found in metamorphic allanite; e.g., Gabudianu Radulescu et al. (2009b)
obtained a concordant U–Pb age 37.4 ± 0.9 Ma (208Pb/232Th: 38.3 ± 1.0 Ma) for allanite from
micaschist (eclogite facies), in which the regressed 207Pb/206Pb was 0.498 ± 0.005. According to
Pb evolution models 207Pb/206Pb would be 0.838 and imply apparent ages > 200Ma!
As emphasized by Janots and Rubatto (2014), for young samples (with low amounts
of radiogenic lead) containing allanite with high Th/U and high initial Pb fraction 206Pbc-
normalized Th–Pb isochron dating will yield the most reliable age and also provides an
independent control on the isotopic composition of initial Pb.
Figure 12. Allanite ages for CAP, a frequently used standard reference material from Permian granodiorite
(Cima d’Asta Pluton, NE Italy). 6 datasets obtained by various techniques (ID-TIMS, SHRIMP, SIMS, LA-
ICP-MS) are compared here. (A) As all datasets show, Th–Pb ages (weighted mean: 275.5 ± 1.4 Ma) are lower
than U–Pb ages, indicating that 206Pb is partly thorogenic (from 230Th); LA + SH for LA-ICP-MS + SHRIMP.
(B, C) Concordia and Tera-Wasserburg (inverse Concordia) plots. TIMS and LA-ICP-MS data show agreement
within error. Symbols of ID-TIMS data are larger than analytical uncertainty; the reverse is true for LA-ICP-
MS data, and gives rise to apparent excess scatter. Common lead contents are 5–46%. Inset in (B) shows BSE
image of CAP allanite illustrating oscillatory zoning typical of magmatic growth. Compare with Figure 6.
392 Engi
CASE STUDIES
Selection, purpose
With the wealth of studies—hundreds on monazite alone—it is quite impossible to do justice
to so many and diverse results. Any selection of exemplary studies is necessarily subjective—for
me, looking for balance seemed futile. Rather my goal is to focus on the understanding gained
by combining petrogenetic analysis and chronology based on REE-minerals. Examples were
chosen, in addition to those mentioned in the first part of this review, because they either helped
resolve a major scientific question—some at mineral grain scale, some at tectonic or geodynamic
scales—or they shed new light into one of the classic terrains. Some studies used particularly
novel thinking, others high-tech approaches or smart combinations of methods.
Single REE mineral species
Isolated inclusions. Dating isolated inclusions in porphyroblasts can be attractive for
several reasons: The host may have protected inclusions from high-strain or high-temperature
stages, thus preserving a trustworthy upper age limit to the host; If the host is petrogenetically
useful, such as garnet, staurolite, or cordierite, inclusion ages may also be linked to P–T
conditions. If the same phase can be dated in other structural contexts, not just from the same
geological unit, but in the same sample, the age difference may give an indication of duration
or rates, or it may be used to test currently held beliefs about closure. This approach has
been applied to xenotime by Kamber et al. (1998) and to monazite by Montel et al. (2000).
In addition, chemical gradients between inclusions and host mineral may lend themselves to
diffusion chronology (Spear et al. 2012; Kohn and Penniston-Dorland 2017), especially if
diffusion profiles for more than one element of unequal diffusion rates are analyzed.
Kamber et al. (1998) dated zircon, monazite, and xenotime in late Archean granulites
from the Limpopo Belt (Zimbabwe) by SIMS (Fig.13). Xenotime was found in two
structural contexts—as inclusions (together with zircon and monazite) in fresh cordierite,
and interstitially in the matrix. Interstitial xenotime yielded 207/206Pb ages ~60 Ma younger
(2551 ± 5 Ma) than xenotime inclusions (2612 ± 8 Ma), which were close to ages of monazite
inclusions (2600 ± 5 Ma) and interstitial matrix monazite (2602 ± 17 Ma). Cordierite had formed
by biotite dehydration-melting at ~850 °C, 0.7 GPa and included the assemblage of accessory
minerals: Tiny zircon and large (150–350 µm), partly corroded monazite and xenotime were
unzoned but strongly enriched in LREE and HREE, respectively. During subsequent cooling,
trapped inclusions were armored from hydrothermal alteration that produced interstitial
xenotime (2551 ± 5 Ma), skarns (2544 ± 39 Ma), as well as tiny zircon (2554 ± 27 Ma). Since
no zoning could be detected in the large xenotime inclusion, Kamber et al. proposed that the
closure temperature for xenotime had to be > 800 °C, which is 100–150° higher than previously
believed. However, it would seem that the survival of older xenotime and monazite inside
cordierite mere shows that retention of Pb in the phosphates was surprisingly high (probably
due to low solubility or low diffusivity of Pb in cordierite).
Montel et al. (2000) used an analogous approach to constrain possible closure temperature
for monazite based on age systematics in kinzigites from Beni Bousera (Morocco). Only
monazite grains (< 20 µm size) included in garnet could be dated (284 ± 27 Ma) by EMPA,
whereas larger interstitial grains (20–70 µm across) had Pb contents below the EMPA detection
limit. Incomplete chemical and mechanical “shielding” was proposed for monazite inclusions
near cracks in garnet, which show ages some 150 Ma less than those fully enclosed in garnet.
Because well shielded monazite inclusions retained the age, and temperatures >850 °C were
determined for the hottest samples investigated, Montel et al. (2000) proposed that the closure
temperature for monazite might be higher than previously accepted. However, as in the previous
study by Kamber et al. (1998), the observation that monazite shows good Pb-retention when
the phosphate is included in a refractory host such as garnet, does not indicate by itself a high
closure temperature for monazite. Such observations can be used, however, to interpret the age
of the inclusion as a minimum formation age.
Figure 13. Mineral chronometry in the Limpopo (Zimbabwe) granulite belt based on SIMS analysis of
inclusions and matrix minerals (Kamber et al. 1998). The robustness to resetting of several mineral chro-
nometers can be assessed in response to thermal, deformational, and chemical triggers. Such HT–terrains
cool slowly, but constraining their T–t paths is challenging, more because of uncertainty in thermometry
(e.g., Blenkinsop and Frei 1997) than in age. Direct mineral thermometry (e.g., Y-in-monazite) would help,
provided the same intragrain domains are dated.
Diversity of monazite
Following some initial studies (e.g., Suzuki et al. 1991), provenance analysis in clastic
sedimentary basins has started to focus more on monazite, in addition to zircon (Hietpas et
al. 2010; Moecher et al. 2011). Both minerals are resilient to transport and weathering, thus
common in heavy mineral fractions; both commonly contain identifiable growth zones that are
likely to retain formation ages. But zircon is extremely refractory, whereas monazite is more
responsive to low- and medium-grade metamorphism and fluid stimulus, hence can preserve a
more complete record of tectonic activity in hinterland source regions.
Hietpas et al. (2011) compared zircon and monazite age spectra from sandstones in the
Appalachian foreland basin and found monazite to retain 232Th–208Pb age spectra that reflect
all of the major Paleozoic orogenic phases known from the Appalachian orogen. By constrast,
90% of the zircon recorded none of the three major tectonic pulses, but instead retained Meso-
Proterozoic or older ages (Fig. 14). The youngest detrital monazites analyzed by Hietpas et
al. indicate a maximum sedimentation age some 550 Ma younger than detrital zircons.
Clearly, the two mineral systems can provide complementary information in provenance studies,
and arguably monazite should take on a larger role than zircon in constraining depositional ages.
Inheritance is common for zircon and monazite, and sometimes for xenotime, so textural
criteria must be used to identify overgrowths (Reimink et al. 2016). To avoid meaningless
mixed dates and convolution in age spectra, spatially separated ages should be acquired.
EPMA analysis may be suitable for analysis of growth zones in monazite and xenotime, but
SIMS or LA-ICP-MS is required for zircon.
Apatite: testing old ideas in new light
Many petrochronological studies assume that recorded growth zoning is essentially
preserved, i.e., that recrystallisation did not occur and that diffusive alteration of chemistry
or ages during the subsequent T–t evolution was negligible. But at high metamorphic grades
and in the magmatic temperature range, initial concentration gradients can be diffusionally
broadened, especially in apatite. While diffusion may modify or destroy the prograde record,
it opens the possibility to investigate thermal timescales by studying diffusion profiles for
suitable elements and minerals (e.g., Spear et al. 2012; Kohn and Penniston-Dorland 2017).
394 Engi
Figure 14. Age distribution of detrital heavy minerals in Middle Carboniferous to Permian sandstones
from two basins (Lee, Pocahontas) in the Appalachian foreland, central eastern USA. BSE images were
used to avoid overgrowth zones visible in some monazites from Pacahontas samples. Six basins were ana-
lyzed, and illustrative data for only two of them are shown here. Given the different preservation character-
istics of monazite and zircon, provenance analysis is more representative if these minerals are used in com-
bination: Whereas zircon (U/Pb, LA-ICP-MS) shows only Grenvillian (~1250–950 Ma) and older ages,
monazite (Th/Pb, SIMS) retains predominantly Mesoproterozoic and Paleozoic ages, including Taconian
(470–440 Ma) and Acadian (420–380 Ma) relics. Data from (Hietpas et al. 2011),
This idea was developed for Sr in apatite by Ague and Baxter (2007), who used forward
modeling to constrain the duration of metamorphism in the Dalradian of Scotland, a classic
metamorphic sequence (Barrow 1912). Ague and Baxter analyzed Sr-concentrations across
chemically inhomogeneous apatite grains and compared these data to profiles modeled by
solving the 1-D diffusion equation
REE ( PO 4 )oph → (1 − y )  REE ( PO 4 )  + y  REE 3 + + ( PO 4 ) 

0 3−
(11)
  melt   melt
where c is concentration, t is time, D is diffusivity, and x is distance. A step function was
assumed for the initial CSr profile. Then, using the temperature-dependent diffusion coefficient
DSr (Cherniak and Ryerson 1993) and the maximum temperature documented (±30°) for each
sample, the time required to produce the observed profile was determined. This turned out to be
surprisingly short, only ~250 ka (Fig. 15). Analysis of the Sr-profiles in apatite was corroborated
by analogous data for Fe, Mg, Ca, and Mn profiles in garnet, and time scales again were similarly
short. The cycle of regional Barrovian metamorphism is thought to have lasted at least two orders
of magnitude longer, and indeed previous studies in the Dalradian had proposed considerably
longer time scales. Ague and Baxter (2007) instead explain this brief thermal pulse by advective
heating due to magma or aqueous fluids, immediately followed by rapid exhumation.
Figure 15. Data and models of Sr diffusion in apa-

tite (after Ague and Baxter 2007). (A) EPMA data
(blue ovals, size reflects ±2σ uncertainty) of zoned
apatite from the staurolite zone of the Dalradian
metamorphic terrane, Scotland. Results of best-fit
diffusion model are shown by the red curve, and
inferred initial concentration (step function) by
the stippled red line. (B) Various thermal scenarios
(e.g., Models A, B, C shown) are compatible with
the data shown above, as these constrain the prod-
uct D·∆t only. (C) The average timescale (∆t) for
peak thermal conditions is ~250 ka and indepen-
dent of grade (temperature) within error.
Modeling of Pb-diffusion in 3.15 Ga old apatite (and titanite) was used by Schoene
and Bowring (2007) to obtain T–t paths for the Kapvaal craton (South Africa). ID-TIMS
207
Pb/206Pb ages of single grains were lower for smaller grains; numerical modeling indicates
a compatible T–t evolution for all of them. In the absence of mineral chemical data, e.g.,
on the U-distribution within grains, it is impossible to test some of the assumptions made.
For this reason, these data are not well suited to inform current debates regarding realistic
diffusion coefficients for apatite and titanite (Kohn 2017; Kohn and Penniston-Dorland 2017).
For example, larger grains might have begun growing earlier than smaller grains, or contain
earlier-formed cores, leading to age dispersion independent of diffusion effects.
Monadic monazite
Polyphase growth (of wisdom) in New England. A metamorphic suite in the Chesham
Pond Nappe (SW New Hampshire, USA) yields monazite–xenotime temperatures between
470–740 °C (Pyle and Spear 2003). As is typical in sample suites with multiple monazite
generations, a perfect separation of stages is not possible; scatter and some overlap in the
Y-contents (XYPO4)mnz occur within a single sample and certainly among samples, even if
closely spaced (Fig. 16). Combined uncertainties (from calibration and analytical scatter in
populations) are given as ±50° (2σ). Pyle et al. (2005a) added chemical and SIMS isotopic
ages for four distinct monazite generations: an early detrital one plus three produced by
metamorphic reactions. Prograde evolution covers an age range of ~20 Ma.
396 Engi
Figure 16. P–T–t evolution in the Chesham Pond nappe (New Hampshire, northeastern USA) based on
monazite petrochronology. Pyle et al. (2005a) distinguished four monazite domains based on textural and
chemical criteria. Overgrowth geometry, partial resorption, and Th/U vs. Y in monazite shown in insets.
SIMS 207Pb/206Pb ages (2 samples) and EPMA chemical U–Th–Pb ages (3 samples) for each domain were
linked to P–T conditions using monazite–garnet thermometry (Pyle and Spear 2003) and metapelite phase
equilibria. An early sharp thermal spike reflects contact metamorphism. Subsequent tectonically driven
P–T evolution involved a pressure increase and fast heating (14.4 ± 4.0 °C/Ma) towards the thermal peak
(domains 2–3), growth of monazite from partial melt upon initial cooling from the thermal peak (domain
4), and rapid cooling to ~350 °C to reach Ar closure in muscovite and biotite (at -8.5 ± 1.6 °C/Ma). All
uncertainties and closure temperatures for 40Ar/39Ar ages from Pyle et al. (2005a).
Such monazite ages are difficult to use for direct correlations amongst tectonic units, let
alone an entire orogenic belt. Detailed analysis of individual samples with tight links to the
tectono-metamorphic evolution for each of them is more promising.
Canadian Cordillera. Foster et al. (2004) documented several P–T–t points from a single
sample containing sufficiently large monazite growth zones. For each such zone, in situ LA-
ICP-MS ages are tied to P–T data (from Y in monazite thermometry and/or phase equilibrium
modeling of the associated assemblage; Foster et al. 2002). X-ray maps are useful to guide
trace element analysis, and relating the Y-contents in garnet and monazite is of particular
importance to understand the metamorphic evolution (Pyle and Spear 1999). In two samples
from the Monashee complex (Canadian Cordillera), 3–4 monazite growth zones were identified
and, using textural criteria, their temporal relations to poikiloblastic kyanite and garnet were
carefully established (Foster et al. 2002). U–Pb ages and thermobarometry for each monazite
zone (Fig. 17) could thus be tightly linked to the overall petrogenetic evolution deduced from
detailed petrography and phase diagram modeling. The resulting P–T–t paths show 2–3 phases
of prograde monazite growth over a period of 20 Ma; the data indicate an average heating rate
of 2.4 ± 1.2 °/Ma and constrain the onset of rapid exhumation.
This study emphasized the integrated approach used. Though based on only two
strategically chosen samples from the Omineca Belt, the results quantified a temporal
framework of the tectonic evolution in this collisional orogen.
Accretion. Based on detailed and systematic mapping aimed to decipher the tectonic
accretion of the Yukon-Tanana Terrane to western North America, Berman et al. (2007)
used petrochronology on six samples in the Stewart River geology. Structurally well
defined metamorphic samples from this complex, polydeformed terrane were studied to link
Figure 17. Petrochronological data for two samples studied by Foster et al. (2004) from the Canadian
Cordillera. Samples were collected from 600 m (DG167) and 2.1 km (DG167) below the Monashee
décollement, which separates two amphibolite facies continental units. Up to four generations of monazite
were distinguished, 2–3 could be dated by LA-ICP-MS U–Pb analysis in each rock. Garnet–monazite,
xenotime–matrix monazite, and garnet–biotite thermometry was combined with pseudosection analysis to
infer P–T paths. Although temperature is well constrained, pressure is not. (A, B) T–t and P–T data and
trends; vertical bands indicate independent constraints for the minimum age of extension and the end of
thrusting, respectively. (C) P–T paths are schematic, and uncertainties are larger than shown. Asymmetric
uncertainties in pressure indicated by position of red age labels.
in situ SHRIMP U–Pb monazite ages and a few TIMS U–Pb titanite ages to changing P–T
conditions. Robust metamorphic constraints permit a temporal framework to be assigned
to the multiple tectonic episodes (Fig. 18): Late Devonian arc formation (365–357 Ma),
Mississippian magmatism and deformation—first above the subducting east-vergent
proto-Pacific slab (355–340 Ma), then above a W-vergent subduction system during the
Mid-Permian (260–253 Ma) involving eclogite formation and arc magmatism—followed
by contact metamorphism from supra-subduction plutons (200–190 Ma), and accretion
by thrusting onto North America (185 Ma). Metamorphic constraints were derived by
combining forward and inverse thermodynamic models, and 206Pb/238U chronometry was
used for ages. This study again demonstrates that detailed analysis of a select few samples
can greatly enhance the understanding of the tectonic implications, provided the necessary
groundwork was previously laid to select appropriate samples for petrochronology.
398 Engi
Figure 18. Schematic cross sections (from Devonian to Cretaceous) depicting the evolution of the Yukon-
Tanana Terrane (YTT) analyzed by Berman et al. (2007) in the Stewart River area, Yukon, Canada. The
polydeformational and polymetamorphic evolution of the YTT spans 250 Ma, as unraveled by structural
work, TWQ-thermobarometry, U–Pb dating of titanite by TIMS, and U–Pb dating of chemically dis-
tinct monazite domains by SHRIMP. Five metamorphic stages (M1–M5) found by Berman et al. in YTT
samples are listed with P–T conditions and ages. This study demonstrated that petrochronology with de-
tailed analysis of structurally well controlled samples can uniquely illuminate continental margin tectonics.
Himalayan tectonics. The capacity of monazite to record several sequential stages of

growth and retain formation ages up to high temperature has been put to good use in many
studies. Some of these have documented quite complex P–T–D–t-paths and helped elucidate
large scale tectonics. This is notably the case in the Himalaya, where petrochronology has
been combined with detailed structural studies to allow current models of collisional orogeny
to be critically tested (Kohn 2014a). Specific temperature–time points—commonly from
monazite—and their corresponding paths are essential, but their value is greatly enhanced if it
is possible to link them to the polyphase structural evolution.
In central Nepal, four main thrust sheets were studied in detail in two main sections (Kohn
et al. 2004; Corrie and Kohn 2011). The approach developed was first to identify monazite grains
in thin section and produce EPMA maps of chemical zoning; select grains were then drilled out,
mounted, polished, and analyzed by SIMS. X-ray maps (Th, U, Y, Si) were used to preselect
spots for dating. In zoned monazite, low-Y zones (at higher grade also low-Th) were identified
as late prograde monazite; high-Y rims grew around these (in the higher tectonic units), and these
rims indicate growth from melt upon cooling. Xenotime was found in few samples and appears
to be mostly retrograde; allanite was essentially limited to samples of the Lesser Himalayan
sequence. The 232Th–208Pb ages indicate similar patterns within each tectonic unit and correlate
with equally systematic P–T data (mostly garnet–biotite thermometry plus barometry based on
Ca-contents of garnet): Within each unit, the growth age of late prograde (pre-anatectic, low-Y)
monazite decreases structurally downward. Age-discontinuities were discovered between thrust
sheets, and cooling in the hangingwall temporally was found to coincide with heating in the
footwall (Fig. 19). These patterns indicate in-sequence thrusting of structurally higher over lower
units, i.e., the unit being thrust on top started transferring heat to the one below.
In the Annapurna region, peak P–T conditions increase structurally upwards, from the lowest
to the highest parts of the Greater Himalayan Sequence, with ~525 °C, 0.8 GPa in the lowest unit;
750 °C, 1.2 GPa in the middle units, and up to 775 °C, 1.3 GPa in the topmost tectonic unit. A
similar sequence was also documented further east, in the Langtang section (Kohn 2008), starting
in the lower Lesser Himalayan sequence and going up to the middle Greater Himalayan sequence.
Here, migmatites in the top unit were crystallizing while the unit in their footwall was still heating.
P–T–t paths for each unit benefit from the tight links among mineral chemistry, textures, and
ages, again showing the need to identify the petrogenetic context of polyphase monazite growth;
avoiding mixed dates (or at least weeding them out based on well separated ages) is crucial.
These data (and similar ones from western Nepal, Montomoli et al. 2013) appear to have
resolved the long-standing question of what caused the inverted metamorphic sequence in the
Himalaya, at least in central Nepal: Protracted thrusting produced sigmoidal isotherms10; a
series of such thrusts was activated in sequence, each of them a gently inclined splay of the Main
Himalayan Thrust, thus distributing the lithosphere scale strain (see also Harrison et al. 1995;
Catlos et al. 2001). Stacking of these thrust sheets involved several relatively short-distance
thrust periods, producing a duplex structure, which then may have (partially) collapsed by
orogenic flattening. Discussions on tectonic mechanisms remain controversial, but the
monazite datasets from Nepal samples certainly favor critical wedge models, even if more
recent work concludes that convergence was accommodated through both wedge taper and
lateral mid-crustal flow processes (Larson et al. 2013).
Combining strengths: petrochronology from monazite plus allanite
In several geological settings prograde reaction sequences involving monazite and allanite
have been analyzed, providing detailed insight into reaction mechanisms (e.g., Rubatto et
al. 2001; Wing et al. 2003; Janots et al. 2006). In favorable cases the age and conditions of
formation of both REE-phases could be determined, with implications on the duration and
rates of metamorphic processes. These studies also helped to correct long-held views of how
mineral ages relate to P–T–t paths, i.e what monazite ages do (or do not) mean.
10
the term “folded isotherms” is misleading
400 Engi
Figure 19. Monazite-based petrochronology (Corrie and Kohn 2011) in four thrust sheets of the Greater
Himalayan Sequence (GHS) along Modi Khola valley, Annapurna region (Nepal). (A) Monazite data cover
three segments of Formation I (Fm I). Two stages of prograde monazite growth are observed in all units,
with a stage of post-anatectic growth in the central two units. Ages of each generation decrease structurally
downward. Cooling was probably due to thrust emplacement (Kohn et al. 2004). This study documents that
good structural control combined with detailed and petrologically solid monazite chronometry can yield
tectonic (as well as thermal) rates and thus allows tectonic models to be quantitatively evaluated.
Central Alps. In metaclastic sequences of the Lepontine Alps (Switzerland) low-grade

pelitic and marly rocks show evidence of first metamorphic monazite (high in La, low in
Th and Y) grown at the expense of detrital monazite (dated at ~300–320 Ma) below 400 °C,
but these could not be dated (diameter usually < 5 µm and too young for EPMA). Along the
classic metamorphic field gradient examined near Passo Lucomagno (Janots et al. 2008,
2009), at 440–450 °C (~0.4 GPa) monazite was a reactant in one of the first reactions that
produced chloritoid, apatite, and allanite (zoned in REE, Y and Th). In the biotite zone (and
up to 610 °C), only allanite but neither monazite nor xenotime remained in metapelites. In
marly rocks between 530–500 °C (0.5–0.9 GPa), allanite retains its zoning, but is rimmed
by epidote in the kyanite zone. Where staurolite formed, allanite started to break down to
monazite + xenotime clusters (Fig. 20), but persisted up to 580 °C. The compositions of this
second generation of metamorphic monazite reflect those of their local REE-precursors (e.g.,
MREE-rich allanite cores produced MREE-rich monazite, LREE-rich allanite rims produced
corresponding LREE-monazite; this likely reflects a sequential breakdown); Th-rich monazite
rims and xenotime satellites always ended up at the periphery of clusters. In samples of low
Ca/Na-ratio only monazite + xenotime remained above 585 °C, but at high Ca/Na Y-rich
epidote armored allanite, and breakdown to monazite did not occur.
In situ SHRIMP (208Pb/232Th) ages of coexisting allanite and monazite gave 31.5 ± 1.3 and
18.0 ± 0.3 Ma in one sample (and in a second one 29.2 ± 1.0 and 19.1 ± 0.3 Ma), constraining
the time interval of prograde heating from 430–450 °C to 560–580 °C, implying a heating
rate of 8–15 °C/m.y. (Fig. 20). It is uncertain whether monazite ages in the higher-grade
parts of the Lepontine metamorphic belt (up to sillimanite grade, ~700 °C) grew by similar
mechanisms and P–T conditions. Given the diversity of rock types and local deformation
styles, other growth modes are fairly likely, though further in situ dating has confirmed
the observed pattern (Boston et al. 2014). Many precise monazite TIMS ages (Köppel
and Grünenfelder 1975, 1978; Köppel et al. 1981) indicate a gradual increase in age from
~19 to ~23 Ma across the belt, with abruptly more diverse ages in the Southern Steep belt,
where monazite is ~27 Ma, and migmatites contain allanite and zircon that range between
32 and 20 Ma (Rubatto et al. 2009; Gregory et al. 2012). Yet, as in other classic orogens,
many highly precise ages remain of uncertain petrogenetic and tectonic significance. The
previously propagated notion that monazite ages simply document Tmax conditions is almost
as unlikely as their interpretation as cooling ages (to 450 °C, e.g., Steck et al. 2013).
Monazite at extreme conditions
Partial melting range. Prograde zoning has been found well preserved in at least some
of the monazites from various migmatites (Pyle and Spear 2003; Kohn et al. 2005), and
thermodynamic models (Kelsey et al. 2008; Spear and Pyle 2010) have proven very useful to
analyze and interpret complexly zoned monazites, especially in settings involving partial melting
(Hallett and Spear 2015). However, dissolution of monazite in melt or extended periods at high
temperature reduce the record retained. Some datasets show that chemical and isotopic age
zoning respond differently to temperature. Monazite at UHT has not yet been widely studied.
In migmatites (600–775 °C, 0.63–1.0 GPa) from polycyclic basement in northern
Norway (Kalak Nappe Complex, Gasser et al. 2015), chemically homogeneous monazite
gave concordant SIMS U–Pb ages that span > 200Ma. Despite careful structural control plus
additional zircon, titanite, and rutile ages, this study could not resolve whether the 200 Ma time
span reflects protracted monazite growth or partial U–Pb resetting. Ages obtained (Fig. 21)
from a majority of monazite grains (698 ± 11 Ma) are within error of zircon (702 ± 5 Ma),
and both reflect crystallization from partial melt, otherwise similar monazite grains cluster
along Concordia between ~600 and ~800 Ma. As rutile is texturally coeval, its Zr-in-rutile
temperatures (550–630 °C) clearly must be reset, and indeed its age is Caledonian, mostly
440 Ma, within error of the titanite that overgrew rutile at 440–430 Ma. In this case monazite
402 Engi
Figure 20. Ages and P–T data for medium-pressure amphibolite facies metapelites from the northern
Lepontine belt, Central Alps, Switzerland. (A) BSE image shows metamorphic allanite and monazite.
Dark pits in allanite are from SHRIMP analysis. Th/Pb ages (31.5 ± 1.3 Ma) date prograde allanite growth
at chloritoid-in isograd (440 °C). (B) BSE photo of shows monazite and xenotime formed by allanite break-
down at staurolite–kyanite grade (560 °C, 0.8 GPa). 10 (out of 13) spots were used to obtain a SHRIMP
U/Pb age of 18.0 ± 0.3 Ma; Tera-Wasserburg plot shows 5–12% common lead in monazite (inherited from
allanite). (C) These ages, combined with silicate-based thermobarometry (Todd and Engi 1997; Janots et
al. 2008) and fission track data (Janots et al. 2009), establish a P–T–t path for this classic orogen as well as
a moderate heating rate. The high cooling rate mostly reflects tectonic unroofing.
evidently retained no chemical evidence of prograde growth. Gasser et al. were unsure whether
post-migmatic cooling involved substantial decompression (from ~0.9 GPa) prior to the onset
of the Caledonian cycle. But the study concluded that the samples they studied spent some
200 Ma at temperatures above 650 °C, so whatever chemical zoning monazite (in melanosome)
initially had, must have been erased. These migmatites were not overprinted by the otherwise
pervasive Caledonian S2 fabrics (peaking at 600 ± 30 °C, 1.13 ± 0.13 GPa), but rutile ages are
partly concordant (552—432 Ma), and Zr-in-rutile yields temperatures consistent with S2,
and titanites yield concordant TIMS ages of 440—429 Ma. The P–T evolution between the
two stages is not constrained, and schematic paths such as shown (Fig. 21a, b) are possible.
However, it is clear that monazite in migmatites formed over some 200 Ma at temperatures
> 650 °C (Gasser et al. 2015), following anatexis in the sillimanite- and kyanite-fields.
Figure 21. Ages and P–T path for metapelitic migmatites from the Caledonian Kalak Nappes, (northern
Norway), that show pre-Caledonian (S1) fabrics. (A) Monazite in leucosome is chemically homogeneous,
and a Tera-Wasserburg plot of SIMS data shows 10 concordant U–Pb ages between 710 ± 9 and 609 ± 12 Ma.
The oldest 7 grains cluster at 698 ± 11 Ma (within error of zircon: 702 ± 5 Ma),whereas 3 younger monazites
(stippled outline) scatter along Tera-Wasserburg. (B) Monazites from melanosome are mostly homogeneous
(rare cores differ in Th and REE). Excepting two core analyses, SIMS data are concordant within 2σ, and
range from 786 ± 12 to 594 ± 18 Ma. (C) Possible P–T–t path. Pseudosection modeling for S1-assemblages
indicates peak conditions of 760 ± 15 °C and 0.93 ± .05 GPa at ~700 Ma for the S1 stage, and 600 ± 30 °C and
1.13 ± 0.13 GPa at ≥ 440Ma for the S2 (Caledonian) stage. Monazite growth in leucosomes entirely postdates
the S1 stage, whereas in melanosomes, a few older cores remain. In melanosomes, chemical heterogeneity
in monazite was largely erased along this protracted high-T path, but the large spread of concordant U–Pb
dates probably represents reliable formation ages. Data from Gasser et al. (2015).
404 Engi
UHT conditions. In the Eastern Ghats (India) UHT terrain, slow cooling from peak
conditions >960 °C to ~900 °C and 0.97 to 0.75 GPa (determined from phase equilibrium
modeling, Korhonen et al. 2011) is reflected in monazite mean 207Pb/206Pb SHRIMP ages11
(Korhonen et al. 2013) between 970 to 930 Ma. In this context, a spread of ages (1130 and
970 Ma) from monazites with very diverse internal textures and Y-distributions are interpreted
to recall prograde stages. Not surprisingly, the data indicate partial Pb-loss (Fig. 22), though
no simple core-to-rim age zoning is evident, and most of the analyses of zircon and monazite
are reported as concordant (defined as <10% discordance). The P–T–t data indicate a counter-
clockwise path, with certainly > 50 Ma (and possibly 200 Ma) at UHT conditions. Discrete,
concordant populations of monazite are common (more so than zircon overgrowths on zircon
cores) and could be analyzed in situ and in grain fractions; they are clearly offset against
Figure 22. Petrochronology of UHT rocks (Mg-Al-rich granulites) from Sunki locality, Eastern Ghats Prov-
ince, India. These rocks record cooling from > 960 °C, ~1 GPa to the solidus at ~900 °C, 0.75 GPa during
slow exumation (Korhonen et al. 2011). (A) P–T paths inferred for two samples of different composition.
Initial petrographic and pseudosection analysis indicated rather steep (rapid) initial decompression (thick
black paths), but later analysis inferred lower early pressures and nearly isobaric paths (thin red paths).
Inclusion relations and coronal successions involving major silicates (opx, cord) and accessory minerals al-
lowed monazite ages to be tied to mineral reactions. (B) Temperature–time paths for same samples. EPMA
X-ray mapping and SHRIMP dating show slow cooling (1–2 °C/Ma) from ~1000 °C at 1000–1050 Ma to
850–900 °C at 950 Ma, when zircon in a third sample grew. Many data are discordant data in Sample 1, few
in Sample 2. Weighted mean ages are shown as 2σ (grey bands), but envelopes of individual data are 1σ.
11
except where excess 206Pb was noted, in which case 207Pb/235U ages were preferred
late Neoproterozoic overprinting. Given monazite grain sizes of typically 80–200 µm, it is

remarkable that only a small proportion of the data appear disturbed and had to be considered
as outliers. The P–T conditions and duration of the UHT-evolution appear reliably defined
on the basis of coherent age groups; monazite and zircon grown from (trapped) melt show
overlapping ages, though some monazite ages may exceed those of zircon. In Sample 2, two
stages of decompression were separately datable: monazite in opx is ~60 Ma older than in
cord. The relatively large proportion of concordant dates suggests protracted monazite growth
(Sample 1: 1014–959 Ma, Sample 2: 1043–922 Ma) due to multiple reactions during initial
cooling. Younger discordant ages in Sample 1 were interpreted to reflect Pb loss, but monazites
from Sample 2 do not show this.These data suggest that monazite chemistry and ages are
resistant to diffusional resetting at temperatures of ca. 950 ˚C (Korhonen et al. 2013).
UHP conditions. Hacker et al. (2015) investigated the response of monazite and its host rocks
to metamorphism during subduction to UHP conditions and subsequent return to the surface. A
large sample suite from the Western Gneiss Region (Norway) was analyzed by LASS to obtain
simultaneous spot analyses of trace-elements, U–Pb ages, and Th–U chemistry. This instrument
makes an inherent spatial link between spot ages and their petrological context, so to interpret
the significance of individual spots, a few well established principles were used: Spot ages
from monazite showing low HREE concentrations were taken to reflect the presence of garnet
(Rubatto 2002); elevated Eu and Sr indicated growth in the eclogite facies, where plagioclase
had broken down to pyroxene and garnet (Finger and Krenn 2007; Holder et al. 2015); Th and
U concentrations indicated where/when fluid was available (Hoskin and Schaltegger 2003) that
may have facilitated monazite recrystallization (Seydoux-Guillaume et al. 2002a). Hacker et
al. (2015) found that relatively few monazite ages reflect growth at UHP-conditions; in coarse-
grained rocks some monazites first survived amphibolite- to granulite-facies metamorphism and
partial melting, followed by UHP-subduction—with their original ages intact.
Allanite: a hot finale
Allanites from migmatites and intrusive or volcanic rocks have sparked several studies in
which the control available from strong zoning in major and trace elements and isotopes was
combined with retentive U–Th–Pb chronometry on allanite, zircon, and titanite.
Melt evolution and disequilibrium isotope systematics. Oberli et al. (2004) studied
allanite from a single sample taken in a massive feeder dyke of the syn-orogenic Bergell tonalite
(Swiss Alps) intrusion. Conventional ID-TIMS dating on mineral separates was used. Detailed
documentation by EPMA and SEM are perfectly compatible with magmatic fractionation;
hornblende thermobarometry indicates 940 °C and ~0.8 GPa, allanite crystallization may
have started below the zircon saturation temperature (>730 °C). Fractionation is recorded in
strongly zoned allanite (Fig. 23): Core compositions show REE2O3 ~17.6 wt%, Fe2+/Fetotal
~0.6; both are near zero at the rim; Th/U rapidly decreases from ~170 to ~1. Apparent ages
based on 206Pb/238U in allanite were 1.8 to 6.3 Ma older than 208Pb/232Th ages due to excess
206
Pb, reflecting initial isotopic 230Th/238U disequilibrium, and allanite’s ravenous appetite for
Th. The sequence of ages obtained from (in part air-abraded) fractions of growth-zoned zircon
(33.0–32.0 Ma), density-separated allanite (32.0–28.0 Ma), and titanite (29.85 Ma) document
growth in a differentiating melt and are related to finite intervals of melt fractionation during
emplacement and in situ crystallization. One core-rich zircon fraction (dated at 44.7 Ma)
showed evidence of inheritance, whereas six others indicate a differentiation trend over ~1 m.y.
When corrected for initial isotopic disequilibrium in 230Th/238U, these six data (and titanite)
were concordant, whereas the upper intercept of the zircon core was at 1070 Ma.
The benefit of very precise TIMS-data was paired with painstakingly careful characterization
of mineral chemical and isotopic composition. While in situ dating might have tightened the
links between microchemical and age information, it is doubtful whether the more uncertain
isotopic ratios in LA-ICP-MS or SIMS data could discern such fine age detail in a system
406 Engi
Figure 23. Systematic chemistry of allanite vs. time for 9 density-separated allanite fragments from the Bergell
tonalite, Iorio Pass, Swiss/Italian Central Alps) that crystallized during magmatic differentiation and that show
complex chemical zoning (Oberli et al. (2004). Painstaking U–Pb and Th–Pb dating by TIMS allowed chemi-
cal differentiation to be linked to a crystallization interval spanning 4 Ma. (A) Th is sequestered early, markedly
depleting the melt in Th. (B) U increases in the melt, but its high concentration implies that crystallization of
major silicates (which generally exclude U) enrich the melt faster than allanite can deplete it. (C–D) Simulta-
neous fractionation of Th and U, and changes to allanite and melt compositions, rapidly diminish the allanite
Th–U ratio and partitioning of Th and U between allanite and melt. (E) Common lead content increases as
allanite grows, demanding correction in dating. (F) 206Pb/238U are systematically higher than 208Pb/232Th ages
ages because of disequilibrium in 230Th/238U. A decrease in Th/U as allanite crystallizes causes the disparity
between ages [Dage(U–Th)] to be most pronounced (> 6 Ma) for the oldest allanite, least in youngest (< 2 Ma).
Preservation of this initial disequilibrium signature shows that the Th–Pb data reflect growth ages during
magma differentiation (between 750 and 650 °C), and that diffusive Pb-resetting was not effective.
evolving so rapidly. It is significant that the observed age differences among grains relate to
magmatic differentiation, not to Pb-diffusion. Allanite thus has a sufficiently high closure
temperature (> 750 °C) to warrant chronometry in acid to intermediate magmatic systems.
Late Quaternary volcanoes. In very young magmatic systems, isotopic disequilibrium
in 238U/230Th (i.e., the U-series age) can be used to estimate residence times (e.g., Vazquez
and Reid 2004); a beautiful example utilized allanite crystals in Mono Lake magma. Cox et
al. (2012) determined the expected secular equilibrium value of 230Th (from the Th/U ratio
analyzed in host glass) and calculated the time needed to reach the 230Th measured in allanite
(Fig. 24). From the difference between the U-series age of 66.0 ± 3.3 ka and the eruption age
of 38.7 ± 1.2 ka (determined independently by (U–Th)/He dating), they calculated a mean
residence time for these allanites of 27 ± 4 ka. This precision rivals determinations using zircon
or baddelyite (Schaltegger and Davies 2017).
Figure 24. U-series ages from allanite and (U–Th)/He ages of eruption from ash layer 15, Mono Lake,
eastern Sierra Nevada, California, USA. The U-series date (66.0 ± 3.3 ka, 2σ) was determined using ICP-
MS (Cox et al. 2012) and reflects the average crystallization age in the magma. The (U–Th)/He date
(38.7 ± 1.2 ka, 2σ) is the (helium) eruption age. These data yield a 27 ± 4 ka average residence time of
allanite in the rhyolitic magma. Shown for comparison are 40Ar/39Ar data for Ash 15 and ages for two
prominent magnetic excursions (Mono Lake and Laschamp).
FUTURE DIRECTIONS
• Further analytical developments will undoubtedly allow dating ever smaller volumes at
higher precision. Of equal importance is sharpening and wielding tools that provide textural
and microchemical context to these ages: X-ray maps from EPMA need to be combined with
trace element maps from laser ablation methods, prior to spot dating.
• Thermodynamic models are needed for the main REE-minerals to make reliable
prediction of the reaction relations involving major and accessory minerals and to enhance
thermobarometry based on their composition. The lead taken by Janots et al. (2007), Spear
(2010), and Spear and Pyle (2010) should be expanded, and for this purpose experimentally
determined phase equilibria—ideally reaction reversals—are required that involve REE-
phases and fully documented natural assemblages from well known P–T conditions.
408 Engi
• Pressure estimates need improving to link chronologic data, especially based on monazite
and allanite. Since allanite frequently contains quartz inclusions, Raman micropiezometry
(Kohn 2014b) on these might be attempted, though shear modulus data for epidote (Hacker
et al. 2003) are but approximately valid for allanite. Radiation softening may also prove
problematica. Elastic properties for the REE-phosphates were compiled by Mogilevsky (2007).
ACKNOWLEDGMENTS
I thank Thomas Armbruster for producing images of crystal structures. Emilie Janots, Rob
Berman, Marco Burn, Pierre Lanari, Birger Rasmussen, and Daniele Regis donated original
images. I am grateful for this and to Emilie Janots and Andrew Smye for their very helpful
reviews. Thanks go to Matt Kohn for his thoughtful advice, editorial rigor and his patience,
and to Christine for doing everything else while I was preparing this. The Swiss National
Science foundation has provided funding for work presented here (project 200020_126946).
REFERENCES
Allaz J, Selleck B, Williams ML, Jercinovic MJ (2013) Microprobe analysis and dating of monazite from the
Potsdam Formation, New York: A progressive record of chemical reaction and fluid interaction. Am Mineral
98:1106–1119
Andrehs G, Heinrich W (1998) Experimental determination of REE distributions between monazite and xenotime:
potential for temperature-calibrated geochronology. Chem Geol 149:83–96
Antignano A, Manning CE (2008) Fluorapatite solubility in H2O and H2O–NaCl at 700 to 900 °C and 0.7 to
2.0 GPa. Chem Geol 251:112–119
Armbruster T, Bonazzi P, Akasaka M, Bermanec V, Chopin C, Gieré R, Heuss-Assbichler S, Liebscher A,
Menchetti S, Yuanming PA, Pasero M (2006) Recommended nomenclature of epidote-group minerals. Eur
J Mineral 18:551–567
Ayers JC, Dunkle S, Gao S, Miller CF (2002) Constraints on timing of peak and retrograde metamorphism in the
Dabie Shan Ultrahigh-Pressure Metamorphic Belt, east-central China, using U–Th–Pb dating of zircon and
monazite. Chem Geol 186:315–331, doi:10.1016/S0009-2541(02)00008–6
Baldwin JA, Bowring SA, Williams ML, Mahan KH (2006) Geochronological constraints on the evolution of high-
pressure felsic granulites from an integrated electron microprobe and ID-TIMS geochemical study. Lithos
88:173–200
Banno S, Yoshizawa H (1993) Sector-zoning of epidote in the Sanbagawa schists and the question of an epidote
miscibility gap. Mineral Mag 57:739–743
Barfod GH, Otero O, Albarède F (2003) Phosphate Lu–Hf geochronology. Chem Geol 200:241–253, doi:10.1016/
S0009-2541(03)00202-X
Barfod GH, Krogstad EJ, Frei R, Albarède F (2005) Lu–Hf and PbSL geochronology of apatites from Proterozoic
terranes: A first look at Lu-Hf isotopic closure in metamorphic apatite. Geochim Cosmochim Acta 69:1847–1859
Barrow G (1912) On the geology of Lower Dee-side and the Southern Highland border. Proc Geol Assoc 1912:1–17
Barth S, Oberli F, Meier M (1994) Th–Pb versus U–Pb isotope systematics in allanite from co-genetic rhyolite and
granodiorite: implications for geochronology. Earth Planet Sci Lett 124:149–159
Bau M (1996) Controls on the fractionation of isovalent trace elements in magmatic and aqueous systems: Evidence
from Y/Ho, Zr/Hf, and lanthanide tetrad effect. Contrib Mineral Petrol 123:323–333
Bea F (1996) Residence of REE, Y, Th, and U in granites and crustal protoliths; implications for the chemistry of
crustal melts. J Petrol 37:521–552
of Ivrea-Verbano, NW Italy. Geochim Cosmochim Acta 63:1133–1153
Beard JS, Sorensen SS, Gieré R (2006) REE zoning in allanite related to changing partition coefficients during
crystallisation; implications for REE behaviour in an epidote-bearing tonalite. Mineral Mag 70:419–435
Berger A, Rosenberg C, Schaltegger U (2009) Stability and isotopic dating of monazite and allanite in partially
molten rocks: examples from the Central Alps. Swiss J Geosci 102:15–29, doi:10.1007/s00015-009-1310-8
Berger A, Gnos E, Janots E, Whitehouse M, Soom M, Frei R, Waight TE (2013) Dating brittle tectonic movements
with cleft monazite: Fluid–rock interaction and formation of REE minerals. Tectonics 32:1176–1189,
doi:10.1002/tect.20071
Berman RG, Sanborn-Barrie M, Stern RA, Carson CJ (2005) Tectonometamorphism at ca. 2.35 and 1.85 Ga in the
Rae Domain, western Churchill Province, Nunavut, Canada: Insights from structural, metamorphic and in
situ geochronological analysis of the southwestern Committee Bay Belt. Can Mineral 43:937–971
Berman RG, Ryan JJ, Gordey SP, Villeneuve M (2007) Permian to Cretaceous polymetamorphic evolution of the
Stewart River region, Yukon-Tanana terrane, Yukon, Canada: P–T evolution linked with in situ SHRIMP
monazite geochronology. J Metamorph Geol 25:803–827, doi:10.1111/j.1525–1314.2007.00729.x
Berzelius JJ (1824) Undersökning af några Mineralier; 1. Phosphorsyrad Ytterjord. Kungliga Svenska Vetenskaps-
Akademiens handlingar, Stockholm:334–358
Beudant F-S (1832) Traité élémentaire de Minéralogie. Verdière
Bingen B, van Breemen O (1998) U–Pb monazite ages in amphibolite-to granulite-facies orthogneiss reflect
hydrous mineral breakdown reactions: Sveconorwegian Province of SW Norway. Contrib Mineral Petrol
132:336–353
Bingen B, Demaiffe D, Hertogen J (1996) Redistribution of rare earth elements, thorium, and uranium over
accessory minerals in the course of amphibolite to granulite facies metamorphism: the role of apatite and
monazite in orthogneisses from southwestern Norway. Geochim Cosmochim Acta 60:1341–1354
Black LP, Fitzgerald JD, Harley SL (1984) Pb isotopic composition, colour, and microstructure of monazites from
a polymetamorphic rock in Antarctica. Contrib Mineral Petrol 85:141–148
Blenkinsop TG, Frei R (1997) Archean and Proterozoic mineralization and tectonics at the Renco Mine (northern
marginal zone, Limpopo Belt, Zimbabwe); reply. Econ Geol 92:747–748
Bollinger L, Janots E (2006) Evidence for Mio-Pliocene retrograde monazite in the Lesser Himalaya, far western
Nepal. Eur J Mineral 18:289–297
Boston K (2014) Investigation of accessory allanite, monazite and rutile of the Barrovian sequence of the Central
Alps (Switzerland). PhD, Australian National University, Canberra
Breiter K, Förster H-J, Skoda R (2008) Extreme P-, Bi-, Nb-, Sc-, U and F-rich zircon from fractionated
perphosphorous granites: the peraluminous Podlesi granite system, Czech Republica. Lithos 88:15–34
Broska I, Williams CT, Janák M, Nagy G (2005) Alteration and breakdown of xenotime-(Y) and monazite-(Ce) in
granitic rocks of the Western Carpathians, Slovakia. Lithos 82:71–83
Broska I, Krogh Ravna EJ, Vojtko P, Janák M, Konečný P, Pentrák M, Bačík P, Luptáková J, Kullerud K (2014)
Oriented inclusions in apatite in a post-UHP fluid-mediated regime (Tromsø Nappe, Norway). Eur J Mineral
26:623–634, doi:10.1127/0935–1221/2014/0026–2396
Bruand E, Storey C, Fowler M (2016) An apatite for progress: Inclusions in zircon and titanite constrain petrogenesis
and provenance. Geology 44:91–94, doi:10.1130/G37301.1
Budzyń B, Harlov DE, Williams ML, Jercinovic MJ (2011) Experimental determination of stability relations
between monazite, fluorapatite, allanite, and REE-epidote as a function of pressure, temperature, and fluid
composition. Am Mineral 96:1547–1567
Buick IS, Clark C, Rubatto D, Hermann J, Pandit M, Hand M (2010) Constraints on the Proterozoic evolution
of the Aravalli–Delhi Orogenic belt (NW India) from monazite geochronology and mineral trace element
geochemistry. Lithos 120:511–528, doi:10.1016/j.lithos.2010.09.011
Burn M (2016) LA-ICP-QMS Th–U/Pb allanite dating: methods and applications. PhD, University of Bern
Catlos EJ, Sorensen SS, Harrison TM (2000) Th–Pb ion microprobe dating of allanite. Am Mineral 85:633–648
Catlos EJ, Harrison TM, Kohn MJ, Grove M, Ryerson FJ, Manning C, Upreti BN (2001) Geochronologic and
thermobarometric constraints on the evolution of the Main Central Thrust, central Nepal Himalaya. J
Geophys Res B, Solid Earth Planets 106:16177–16204
Cenki-Tok B, Oliot E, Rubatto D, Berger A, Engi M, Janots E, Thomsen TB, Manzotti P, Regis D, Spandler C,
Robyr M (2011) Preservation of Permian allanite within an Alpine eclogite facies shear zone at Mt Mucrone,
Italy: Mechanical and chemical behaviour of allanite during mylonitization. Lithos 125:40–50, doi:10.1016/j.
lithos.2011.01.005
Cenki-Tok B, Darling JR, Rolland Y, Dhuime B, Storey CD (2014) Direct dating of mid-crustal shear zones with
synkinematic allanite: new in situ U–Th–Pb geochronological approaches applied to the Mont Blanc massif.
Terra Nova 26:29–37
Chamberlain KR, Bowring SA (2001) Apatite–feldspar U–Pb thermochronometer: a reliable, mid-range (∼ 450°
C), diffusion-controlled system. Chem Geol 172:173–200
Cherniak D (2000) Rare earth element diffusion in apatite. Geochim Cosmochim Acta 64:3871–3885
Cherniak DJ (2006) Pb and rare earth element diffusion in xenotime. Lithos 88:1–14, doi:10.1016/j.
lithos.2005.08.002
Cherniak DJ (2010) Diffusion in accessory minerals: zircon, titanite, apatite, monazite and xenotime. Rev Mineral
Geochem 72:827–869, doi:10.2138/rmg.2010.72.18
Cherniak DJ, Ryerson FJ (1993) A study of strontium diffusion in apatite using Rutherford backscattering
spectroscopy and ion implantation. Geochim Cosmochim Acta 57:4653–4662
Cherniak DJ, Watson EB, Grove M, Harrison TM (2004) Pb diffusion in monazite: a combined RBS/SIMS study
1. Geochim Cosmochim Acta 68:829–840, doi:10.1016/j.gca.2003.07.012
410 Engi
Chew DM, Donelick RA (2012) Combined apatite fission-track and U–Pb dating by LA-ICP-MS and its application
in apatite provenance analysis. Quantitative Mineralogy and Microanalysis of Sediments and Sedimentary
Rocks: Mineral Assoc Canada, Short Course 42:219–247
Chew DM, Spikings RA (2015) Geochronology and thermochronology using apatite: time and temperature, lower
crust to surface. Elements 11:189–194, doi:10.2113/gselements.11.3.189
Chew DM, Sylvester PJ, Tubrett MN (2011) U–Pb and Th–Pb dating of apatite by LA-ICPMS. Chem Geol
280:200–216
Clavier N, Podor R, Dacheux N (2011) Crystal chemistry of the monazite structure. J Eur Ceram Soc 31:941–976
Cochrane R, Spikings RA, Chew D, Wotzlaw J-F, Chiaradia M, Tyrrell S, Schaltegger U, Van der Lelij R (2014)
High temperature (> 350 ºC) thermochronology and mechanisms of Pb loss in apatite. Geochim Cosmochim
Acta 127:39–56
Copeland P, Parrish RR, Harrison TM (1988) Identification of inherited radiogenic Pb in monazite and its
implications for U–Pb systematics. Nature 333:760–763
implications for monazite formation. J Metamorph Geol 26:451–464
Corrie SL, Kohn MJ (2011) Metamorphic history of the central Himalaya, Annapurna region, Nepal, and
implications for tectonic models. Geol Soc Am Bull 123:1863–1879, doi:10.1130/B30376.1
Cox SE, Farley KA, Hemming SR (2012) Insights into the age of the Mono Lake Excursion and magmatic crystal
residence time from (U–Th)/He and 230Th dating of volcanic allanite. Earth Planet Sci Lett 319–320:178–
184, doi:10.1016/j.epsl.2011.12.025
Cressey G, Wall F, Cressey BA (1999) Differential REE uptake by sector growth of monazite. Mineral Mag
63:813–813, doi:10.1180/002646199548952
Cruft EF (1966) Minor elements in igneous and metamorphic apatite. Geochim Cosmochim Acta 30:375–398
Cumming G, Richards J (1975) Ore lead isotope ratios in a continuously changing Earth. Earth Planet Sci Lett
28:155–171
Dahl PS, Terry MP, Jercinovic MJ, Williams ML, Hamilton MA, Foland KA, Clement SM, Friberg LM (2005)
Electron probe (Ultrachron) microchronometry of metamorphic monazite: Unraveling the timing of polyphase
thermotectonism in the easternmost Wyoming Craton (Black Hills, South Dakota). Am Mineral 90:1712–1728
Daniel CG, Pyle JM (2006) Monazite–xenotime thermochronometry and Al2SiO5 reaction textures in the Picuris
Range, Northern New Mexico, USA: New evidence for a 1450–1400 Ma orogenic event. J Petrol 47:97–118,
doi:10.1093/petrology/egi069
Darling J, Storey C, Engi M (2012) Allanite U–Th–Pb geochronology by laser ablation ICP-MS. Chem Geol
292–293:103–115
Davis W, Rainbird R, Gall Q, Jefferson C (2008) In situ U–Pb dating of diagenetic apatite and xenotime: Paleofluid
flow history within the Thelon, Athabasca and Hornby Bay basins. Geochim Cosmochim Acta 72:A203
DeWolf CP, Belshaw N, O’Nions RK (1993) A metamorphic history from micron-scale 207Pb/206Pb chronometry of
Archean monazite. Earth Planet Sci Lett 120:207–220, doi:10.1016/0012-821X(93)90240-A
Didier A, Bosse V, Boulvais P, Bouloton J, Paquette J-L, Montel J-M, Devidal J-L (2013) Disturbance versus
preservation of U–Th–Pb ages in monazite during fluid–rock interaction: textural, chemical and isotopic
in situ study in microgranites (Velay Dome, France). Contrib Mineral Petrol 165:1051–1072, doi:10.1007/
s00410-012-0847-0
Didier A, Bosse V, Cherneva Z, Gautier P, Georgieva M, Paquette J-L, Gerdjikov I (2014) Syn-deformation fluid-
assisted growth of monazite during renewed high-grade metamorphism in metapelites of the Central Rhodope
(Bulgaria, Greece). Chem Geol 381:206–222, doi:10.1016/j.chemgeo.2014.05.020
Petrol 40:259–274
Dodson M (1986) Closure profiles in cooling systems. Mater Sci Forum 7:145–154
Duc-Tin Q (2007) Experimental studies on the behaviour of rare earth elements and tin in granitic systems. PhD,
University of Tuebingen
Duc-Tin Q, Keppler H (2015) Monazite and xenotime solubility in granitic melts and the origin of the lanthanide
tetrad effect. Contrib Mineral Petrol 169:1–26, doi:10.1007/s00410-014-1100-9
Dumond G, Goncalves P, Williams ML, Jercinovic MJ (2015) Monazite as a monitor of melting, garnet growth and
feldspar recrystallization in continental lower crust. J Metamorph Geol 33:735–762, doi:10.1111/jmg.12150
Elderfield H, Pagett R (1986) Rare earth elements in ichthyoliths: variations with redox conditions and depositional
environment. Sci Total Environ 49:175–197
Ellison A, Hess P (1988) Peraluminous and peralkaline effects upon “monazite” solubility in high-silica liquids.
Eos, Trans Am Geophys Union 69:498
Ercit TS (2002) The mess that is allanite. Can Mineral 40:1411–1419
Erickson TM, Pearce MA, Taylor RJM, Timms NE, Clark C, Reddy SM, Buick IS (2015) Deformed monazite
yields high-temperature tectonic ages. Geology 43:383–386
Evans J, Zalasiewicz J (1996) U–Pb, Pb–Pb and Sm–Nd dating of authigenic monazite: implications for the
diagenetic evolution of the Welsh Basin. Earth Planet Sci Lett 144:421–433
Evans JA, Zalasiewicz JA, Fletcher I, Rasmussen B, Pearce NJG (2002) Dating diagenetic monazite in mudrocks:
constraining the oil window? J Geol Soc 159:619–622
Fehr KT, Heuss-Assbichler S (1997) Intracrystalline equilibria and immiscibility along the join clinozoisite–
epidote: An experimental and 57Fe Mössbauer study. Neues Jahrb Mineral Abh 172:43–67
Ferry JM (1994) Overview of the petrologic record of fluid flow during regional metamorphism in northern New
England. Am J Sci 294:905–988
Ferry JM (2000) Patterns of mineral occurrence in metamorphic rocks. Am Mineral 85:1573–1588
loop. Lithos 95:103–115
Finger F, Broska I, Roberts MP, Schermaier A (1998) Replacement of primary monazite by apatite–allanite–
epidote coronas in an amphibolite facies granite gneiss from the eastern Alps. Amer Mineral 83:248–258
Fletcher IR, McNaughton NJ, Aleinikoff JA, Rasmussen B, Kamo SL (2004) Improved calibration procedures
and new standards for U–Pb and Th–Pb dating of Phanerozoic xenotime by ion microprobe. Chem Geol
209:295–314, doi:10.1016/j.chemgeo.2004.06.015
Förster HJ, Harlov DE (1999) Monazite-(Ce)–huttonite solid solutions in granulite-facies metabasites from the
Ivrea-Verbano Zone, Italy. Min Mag 63:587–594
181:327–340
Foster G, Gibson HD, Parrish R, Horstwood M, Fraser J, Tindle A (2002) Textural, chemical and isotopic insights
into the nature and behaviour of metamorphic monazite. Chem Geol 191:183–207, doi:10.1016/S0009-
2541(02)00156–0
Sci Lett 228:125–142, doi:10.1016/j.epsl.2004.09.024
Franz G, Andrehs G, Rhede D (1996) Crystal chemistry of monazite and xenotime from Saxothuringian-
Moldanubian metapelites, NE Bavaria, Germany. Eur J Mineral:1097–1118
Gabudianu Radulescu I, Rubatto D, Gregory C, Compagnoni R (2009a) The age of HP metamorphism in the Gran
Paradiso Massif, Western Alps: A petrological and geochronological study of “silvery micaschists”. Lithos
110:95–108, doi:10.1016/j.lithos.2008.12.008
Gabudianu Radulescu I, Rubatto D, Gregory C, Compagnoni R (2009b) The age of metamorphism in the Gran
Paradiso Massif, Western Alps: A petrological and geochronological study of “silvery micaschists”. Lithos
110:95–108, doi:10.1016/j.lithos.2008.12.008
Ganguly J, Tirone M (1999) Diffusion closure temperature and age of a mineral with arbitrary extent of diffusion:
Gasser D, Jeřábek P, Faber C, Stünitz H, Menegon L, Corfu F, Erambert M, Whitehouse M (2015) Behaviour
of geochronometers and timing of metamorphic reactions during deformation at lower crustal conditions:
phase equilibrium modelling and U–Pb dating of zircon, monazite, rutile and titanite from the Kalak Nappe
Complex, northern Norway. J Metamorph Geol 33:513–534
Gieré R, Sorensen SS (2004) Allanite and other REE-rich epidote-group minerals. Rev Mineral Geochem 56:431–493
Gleadow AJ, Belton DX, Kohn BP, Brown RW (2002) Fission track dating of phosphate minerals and the
thermochronology of apatite. Rev Mineral Geochem 48:579–630
Goldschmidt VM, Thomassen L (1924) Geochemische Verteilungsgesetze der Elemente. III.
Röntgenspektrographische Untersuchungen über die Verteilung der Seltenen Erdmetalle in Mineralen.
Videnskapsselskapets Skrifter I Mathematisk-Naturvidenskabelig Klasse 5:1–58
Goswami-Banerjee S, Robyr M (2015) Pressure and temperature conditions for crystallization of metamorphic
allanite and monazite in metapelites: a case study from the Miyar Valley (high Himalayan Crystalline of
Zanskar, NW India). J Metamorph Geol 33:535–556
Goudie DJ, Fisher CM, Hanchar JM, Crowley JL, Ayers JC (2014) Simultaneous in situ determination of U–Pb
and Sm–Nd isotopes in monazite by laser ablation ICP-MS. Geochem Geophys Geosystem 15:2575–2600
Grand’Homme A, Janots E, Bosse V, Seydoux-Guillaume A, Guedes RDA (2016a) Interpretation of U–Th–Pb
in-situ ages of hydrothermal monazite-(Ce) and xenotime-(Y): evidence from a large-scale regional study in
clefts from the western Alps. Mineral Petrol:1–21
Grand’Homme A, Janots E, Seydoux-Guillaume A-M, Guillaume D, Bosse V, Magnin V (2016b) Partial resetting
Geology 44:431–434
Gratz R, Heinrich W (1997) Monazite–xenotime thermobarometry: experimental calibration of the miscibility gap
in the system CePO4–YPO4. Am Mineral 82:772–780
Gratz R, Heinrich W (1998) Monazite–xenotime thermometry. III. Experimental calibration of the partitioning of
gadolinium between monazite and xenotime. Eur J Mineral 10:579–588
412 Engi
Gregory CJ, McFarlane CRM, Hermann J, Rubatto D (2009) Tracing the evolution of calc-alkaline magmas: In-
situ Sm–Nd isotope studies of accessory minerals in the Bergell Igneous Complex, Italy. Chem Geol 260:73–
86, doi:10.1016/j.chemgeo.2008.12.003
Gregory CJ, Rubatto D, Allen CM, Williams IS, Hermann J, Ireland T (2007) Allanite micro-geochronology: A
LA-ICP-MS and SHRIMP U–Th–Pb study. Chem Geol 15:162–182
Gregory C, Rubatto D, Hermann J, Berger A, Engi M (2012) Allanite behaviour during incipient melting in the
southern Central Alps. Geochim Cosmochim Acta 84:433–458, doi:10.1016/j.gca.2012.01.020
Hacker BR, Abers GA, Peacock SM (2003) Subduction factory 1. Theoretical mineralogy, densities, seismic wave
speeds, and H2O contents. J Geophys Res 108(B1):2029, doi:10.1029/2001JB001127
Hallett BW, Spear FS (2015) Monazite, zircon, and garnet growth in migmatitic pelites as a record of metamorphism
and partial melting in the East Humboldt Range, Nevada. Am Mineral 100:951–972
Hammerli J (2014) Using microanalysis of minerals to track geochemical processes during metamorphism:
examples from the Mary Kathleen fold belt, Queensland, and the Eastern Mt. Lofty Ranges, South Australia.
PhD, James Cook University
Hammerli J, Kemp A, Spandler C (2014) Neodymium isotope equilibration during crustal metamorphism revealed
by in situ microanalysis of REE-rich accessory minerals. Earth Planet Sci Lett 392:133–142
Hammerli J, Spandler C, Oliver NHS (2016) Element redistribution and mobility during upper crustal metamorphism
of metasedimentary rocks: an example from the eastern Mount Lofty Ranges, South Australia. Contrib
Mineral Petrol 171:36, doi:10.1007/s00410-016-1239-7
Harlov DE, Förster HJ (2003) Fluid-induced nucleation of (Y + REE)-phosphate minerals within apatite: Nature
and experiment. Part II. Fluorapatite. Am Mineral 88:1209–1229
Harlov DE, Wirth R, Förster H-J (2005) An experimental study of dissolution–reprecipitation in fluorapatite: fluid
infiltration and the formation of monazite. Contrib Mineral Petrol 150:268–286
Harlov DE, Wirth R, Hetherington CJ (2007) The relative stability of monazite and huttonite at 300–900 °C and
200–1000 MPa: metasomatism and the propagation of metastable mineral phases. Am Mineral 92:1652–1664
Harlov DE, Procházka V, Förster H-J, Matejka D (2008) Origin of monazite–xenotime–zircon–fluorapatite
assemblages in the peraluminous Melechov granite massif, Czech Republic. Mineral Petrol 94:9–26
Harrison TM, Watson EB (1984) The behavior of apatite during crustal anatexis: equilibrium and kinetic
considerations. Geochim Cosmochim Acta 48:1467–1477
Harrison TM, McKeegan KD, LeFort P (1995) Detection of inherited monazite in the Manaslu leucogranite
by 208Pb/232Th ion microprobe dating: Crystallization age and tectonic implications. Earth Planet Sci Lett
133:271–282
Harrison TM, Catlos EJ, Montel J-M (2002) U–Th–Pb dating of phosphate minerals. Rev Mineral Geochem
48:524–558, doi:10.2138/rmg.2002.48.14
Hawkins DP, Bowring SA (1997) U–Pb systematics of monazite and xenotime: case studies from the
Paleoproterozoic of the Grand Canyon, Arizona. Contrib Mineral Petrol 127:87–103
Heaman LM, Parrish RR (eds) (1991) U–Pb geochronology of accessory minerals. Mineral Assoc Canada, Toronto
Heinrich W, Andrehs G, Franz G (1997) Monazite–xenotime miscibility gap thermometry. I. An empirical
calibration. J Metamorph Geol 15:3–16
Helms TS, Labotka TC (1991) Petrogenesis of Early Proterozoic pelitic schists of the southern Black Hills, South
Dakota: constraints on regional low-pressure metamorphism. Geol Soc Am Bull 103:1324–1334
Hermann J (2002) Allanite: thorium and light rare earth element carrier in subducted crust. Chem Geol 192:289–306
Hermann J, Spandler CJ (2008) Sediment melts at sub-arc depths: an experimental study. J Petrol 49:717–740
Hermann J, Rubatto D (2009) Accessory phase control on the trace element signature of sediment melts in
subduction zones. Chem Geol 265
Hetherington CJ, Harlov DE (2008) Partial metasomatic alteration of xenotime and monazite from a granitic
pegmatite, Hidra anorthosite massif, southwestern Norway: Dissolution–reprecipitation and the subsequent
formation of thorite and uraninite inclusions. Am Mineral 93:806–820
Hetherington CJ, Jercinovic MJ, Williams ML, Mahan K (2008) Understanding geologic processes with xenotime:
Composition, chronology, and a protocol for electron probe microanalysis. Chem Geol 254:133–147
Hietpas J, Samson S, Moecher D, Schmitt AK (2010) Recovering tectonic events from the sedimentary record:
Detrital monazite plays in high fidelity. Geology 38:167–170
Hietpas J, Samson S, Moecher D (2011) A direct comparison of the ages of detrital monazite versus detrital zircon
in Appalachian foreland basin sandstones: Searching for the record of Phanerozoic orogenic events. Earth
of (ultra)high-pressure metamorphism: Examples from the Western Gneiss Region, Norway. Chem Geol
409:99–111, doi:10.1016/j.chemgeo.2015.04.021
Hoskin PW, Schaltegger U (2003) The composition of zircon and igneous and metamorphic petrogenesis. Rev
Hughes JM, Rakovan J (2002) The crystal structure of apatite, Ca5(PO4)3(F,OH,Cl). Rev Mineral Geochem 48:1–12
Hughes JM, Rakovan JF (2015) Structurally robust, chemically diverse: apatite and apatite supergroup minerals.
Elements 11:165–170
Huminicki DM, Hawthorne FC (2002) The crystal chemistry of the phosphate minerals. Rev Mineral Geochem
48:123–253
Ibler G (2010) Karl Ludwig Giesecke (1761–1833): das Leben und Wirken eines frühen europäischen Gelehrten;
Protokoll eines merkwürdigen Lebensweges. Mitt Österr Miner Ges 156:37–114
Janots E, Rubatto D (2014) U–Th–Pb dating of collision in the external Alpine domains (Urseren zone, Switzerland)
using low temperature allanite and monazite. Lithos 184:155–166
Janots E, Negro F, Brunet F, Goffé B, Engi M, Bouybaouène ML (2006) Evolution of the REE mineralogy in
HP–LT metapelites of the Sebtide complex, Rif, Morocco: monazite stability and geochronology. Lithos
87:214–234
Janots E, Brunet F, Goffé B, Poinssot C, Murchard M, Cemic L (2007) Thermochemistry of monazite-(La) and
dissakisite-(La): implications for monazite and allanite stability in metapelites. Contrib Mineral Petrol 154:1–
14, doi:10.1007/s00410-006-0176-2
Janots E, Engi M, Berger A, Allaz J, Schwarz J-O, Spandler C (2008) Prograde metamorphic sequence of REE-
from 250 to 610 °C. J Metamorph Geol 26:509–526, doi:10.1111/j.1525–1314.2008.00774.x
Janots E, Engi M, Rubatto D, Berger A, Gregory C (2009) Metamorphic rates in collisional orogeny from in situ
allanite and monazite dating. Geology 37:11–14, doi:10.1130/G25192A.1
Janots E, Berger A, Engi M (2011) Physico-chemical control on the REE-mineralogy in chloritoid-grade
metasediments from a single outcrop (Central Alps, Switzerland). Lithos 121:1–11, doi:10.1016/j.
lithos.2010.08.023
Janots E, Berger A, Gnos E, Whitehouse M, Lewin E, Pettke T (2012) Constraints on fluid evolution during
metamorphism from U–Th–Pb systematics in Alpine hydrothermal monazite. Chem Geol 326–327:61–71,
doi:10.1016/j.chemgeo.2012.07.014
Johnston AD, Wyllie PJ (1988) Constraints on the origin of Archean tronhjemites based on phase relationships of
Nuk gneiss with H2O at 15 kbar. Contrib Mineral Petrol 100:35–46
Johnson TE, Clark C, Taylor RJM, Santosh M, Collins AS (2015) Prograde and retrograde growth of monazite
in migmatites: An example from the Nagercoil Block, southern India. Geosci Frontiers 6:373–387,
doi:10.1016/j.gsf.2014.12.003
Jonsson E, Harlov DE, Majka J, Högdahl K, Persson-Nilsson K (2016) Fluorapatite–monazite–allanite relations in
the Grängesberg apatite–iron oxide ore district, Bergslagen, Sweden. Am Mineral 101:1769–1782
Kamber B, Frei R, Gibb A (1998) Pitfalls and new approaches in granulite chronometry: an example from the
Limpopo Belt, Zimbabwe. Precambrian Res 91:269–285
Karioris FG, Gowda K, Cartz L (1981) Heavy ion bombardment on monoclinic ThSiO4, ThO2, and monazite.
Radiation Eff Lett 58:1–3
Kato Y, Fujinaga K, Nakamura K, Takaya Y, Kitamura K, Ohta J, Toda R, Nakashima T, Iwamori H (2011) Deep-
sea mud in the Pacific Ocean as a potential resource for rare-earth elements. Nat Geosci 4:535–539
Kelly NM, Harley SL, Möller A (2012) Complexity in the behavior and recrystallization of monazite during high-T
metamorphism and fluid infiltration. Chem Geol 322–323:192–208, doi:10.1016/j.chemgeo.2012.07.001
Kelsey DE, Clark C, Hand M (2008) Thermobarometric modeling of zircon and monazite growth in melt-bearing
Kessel R, Schmidt MW, Ulmer P, Pettke T (2005) Trace element signature of subduction-zone fluids, melts and
supercritical liquids at 120–180 km depth. Nature 437:724–727
Kim Y, Yi K, Cho M (2009) Parageneses and Th–U distributions among allanite, monazite, and xenotime in
Barroviantype metapelites, Imjingang belt, central Korea. Am Mineral 94:430–438
Kingsbury JA, Miller CF, Wooden JL, Harrison TM (1993) Monazite paragenesis and U–Pb systematics in rocks of
the eastern Mojave Desert, California, U.S.A.: implications for thermochronometry. Chem Geol 110:147–167
Kirkland CL, Whitehouse MJ, Slagstad T (2009) Fluid-assisted zircon and monazite growth within a shear zone: a case
study from Finnmark, Arctic Norway. Contrib Mineral Petrol 158:637–657, doi:10.1007/s00410-009-0401-x
Kirkland CL, Erickson TM, Johnson TE, Danišík M, Evans NJ, Bourdet J, McDonald BJ (2016) Discriminating
prolonged, episodic or disturbed monazite age spectra: An example from the Kalak Nappe Complex, Arctic
Norway. Chem Geol 424:96–110, doi:10.1016/j.chemgeo.2016.01.009
Klimm K, Blundy JD, Green TH (2008) Trace element partitioning and accessory phase saturation during H2O-
saturated melting of basalt with implications for subduction zone chemical fluxes. J Petrol 49:523–553
Knudsen AC, Gunter ME (2002) Sedimentary phosphorites—an example: Phosphorite formation, southeastern
Idaho, U.S.A. Rev Mineral Geochem 48:363–390
414 Engi
Kohn MJ (2008) PTt data from central Nepal support critical taper and repudiate large-scale channel flow of the
Greater Himalayan Sequence. Geol Soc Am Bull 120:259–273
Kohn MJ (2014a) Himalayan metamorphism and its tectonic implications. Ann Rev Earth Planet Sci 42:381–419
Kohn MJ (2014b) “Thermoba-Raman-try”: Calibration of spectroscopic barometers and thermometers for mineral
inclusions. Earth Planet Sci Lett 388:187–196
silicates: Implications for age determinations. Geochim Cosmochim Acta 68:101–113
Geochem 83:103–152
Himalaya of central Nepal. Earth Planet Sci Lett 228:299–310
doi:10.1111/j.1525–1314.2005.00584.x
Kohn MJ, Vervoort JD (2008) U–Th–Pb dating of monazite by single-collector ICP-MS: pitfalls and potential.
Geochem Geophys Geosystem 9, doi:10.1029/2007GC001899
Kon Y, Hoshino M, Sanematsu K, Morita S, Tsunematsu M, Okamoto N, Yano N, Tanaka M, Takagi T (2014)
geochemical characteristics of apatite in heavy REE-rich deep-sea mud from Minami–Torishima area,
Southeastern Japan. Resour Geol 64:47–57
Köppel V, Grünenfelder M (1975) Concordant U–Pb ages of monazite and xenotime from the Central Alps and the
timing of high temperature metamorphism, a preliminary report. Schweiz Mineral Petrog Mitt 55:129–132
Köppel V, Grünenfelder M (1978) The significance of monazite U–Pb ages; examples from the Lepontine area of
the Swiss Alps. US Geol Survey, Open File Report 78–701:226–227
Köppel V, Günthert A, Grünenfelder M (1981) Patterns of U–Pb zircon and monazite ages in polymetamorphic
units of the Swiss Central Alps. Schweiz Mineral Petrog Mitt 61:97–120
Korhonen F, Saw A, Clark C, Brown M, Bhattacharya S (2011) New constraints on UHT metamorphism in the
Eastern Ghats Province through the application of phase equilibria modelling and in situ geochronology.
Gondwana Res 20:764–781
Korhonen F, Clark C, Brown M, Bhattacharya S, Taylor R (2013) How long-lived is ultrahigh temperature (UHT)
metamorphism? Constraints from zircon and monazite geochronology in the Eastern Ghats orogenic belt,
India. Precambrian Res 234:322–350
Krenn E, Finger F (2004) Metamorphic formation of Sr-apatite and Sr-bearing monazite in a high-pressure rock
from the Bohemian Massif. Am Mineral 89:1323–1329
Krenn E, Harlov DE, Finger F, Wunder B (2012) LREE-redistribution among fluorapatite, monazite, and allanite at
high pressures and temperatures. Am Mineral 97:1881–1890, doi:10.2138/am.2012.4005
345:99–112, doi:10.1016/j.chemgeo.2013.02.019
Geochem 83:55–102
Larson KP, Gervais F, Kellett DA (2013) A P–T–t–D discontinuity in east-central Nepal: Implications for the
evolution of the Himalayan mid-crust. Lithos 179:275–292, doi:10.1016/j.lithos.2013.08.012
Larsson D, Söderlund U (2005) Lu–Hf apatite geochronology of mafic cumulates: An example from a Fe–Ti
mineralization at Smålands Taberg, southern Sweden. Chem Geol 224:201–211
Levinson AA (1966) A system of nomenclature for rare-earth minerals. Am Mineral 51:152–158
Luo Y, Hughes JM, Rakovan J, Pan Y (2009) Site preference of U and Th in Cl, F, and Sr apatites. Am Mineral
94:345–351
Mahood G, Hildreth W (1983) Large partition coefficients for trace elements in high-silica rhyolites. Geochim
Manning CE (2004) The chemistry of subduction zone fluids. Earth Planet Sci Lett 223:1–16
Manzotti P, Rubatto D, Darling J, Zucali M, Cenki-Tok B, Engi M (2012) From Permo-Triassic lithospheric
thinning to Jurassic rifting at the Adriatic margin: Petrological and geochronological record in Valtournenche
(Western Italian Alps). Lithos 146–147:276–292, doi:0.1016/j.lithos.2012.05.007
McFarlane CR (2016) Allanite U–Pb geochronology by 193 nm LA ICP-MS using NIST610 glass for external
calibration. Chem Geol 438:91–102
McFarlane CRM, McCulloch MT (2007) Coupling of in-situ Sm–Nd systematics and U–Pb dating of
monazite and allanite with applications to crustal evolution studies. Chem Geol 245:45–60, doi:10.1016/j.
chemgeo.2007.07.020
Moecher D, Hietpas J, Samson S, Chakraborty S (2011) Insights into southern Appalachian tectonics from ages of
detrital monazite and zircon in modern alluvium. Geosphere 7:494–512
Mogilevsky P (2007) On the miscibility gap in monazite–xenotime systems. Phys Chem Minerals 34:201–214,
doi:10.1007/s00269-006-0139-1
Montel J-M (1986) Experimental determination of the solubility of Ce-monazite in SiO2–Al2O3–K2O–Na2O melts
at 800 °C, 2 kbar, under H2O-saturated conditions. Geology 14:659–662
Montel J-M (1993) A model for monazite/melt equilibrium and application to the generation of granitic magmas.
Chem Geol 110:127–146
Montel J-M, Foret S, Veschambre M, Nicollet C, Provost A (1996) Electron microprobe dating of monazite. Chem
Geol 131:37–53
Montel J-M, Kornprobst J, Vielzeuf D (2000) Preservation of old U–Th–Pb ages in shielded monazite: example
from the Beni Bousera Hercynian kinzigites (Marocco). J Metamorph Geol 18:335–342
Montomoli C, Iaccarino S, Carosi R, Langone A, Visonà D (2013) Tectonometamorphic discontinuities within the
Greater Himalayan Sequence in Western Nepal (Central Himalaya): Insights on the exhumation of crystalline
rocks. Tectonophysics 608:1349–1370
Mottram CM, Parrish RR, Regis D, Warren CJ, Argles TW, Harris NBW, Roberts NMW (2015) Using U–Th–Pb
Himalaya. Tectonics 34:1355–1374, doi:10.1002/2014TC003743
Nemchin AA, Horstwood MS, Whitehouse MJ (2013) High-spatial-resolution geochronology. Elements 9:31–37
Ni Y, Hughes JM, Mariano AN (1995) Crystal chemistry of monazite and xenotime structures. Am Mineral 80:21–26
Oberli F, Meier M, Berger A, Rosenberg CL, Gieré R (2004) U–Th–Pb and 230Th/238U disequilibrium isotope
systematics: precise accessory mineral chronology and melt evolution tracing in the Alpine Bergell intrusion.
Ondrejka M, Putiš M, Uher P, Schmiedt I, Pukančík L, Konečný P (2016) Fluid-driven destabilization of REE-
bearing accessory minerals in the granitic orthogneisses of North Veporic basement (Western Carpathians,
Slovakia). Mineral Petrol, doi:10.1007/s00710-016-0432-8
Overstreet WC (1967) The Geologic Occurrence of Monazite. USGS Professional Paper 530, p. 327
Pan Y, Fleet ME (2002) Composition of the apatite group minerals: Substitution mechanisms and controlling
factors. Rev Mineral Geochem 48:13–49
Pattison DR, Vogl JJ (2005) Contrasting sequences of metapelitic mineral-assemblages in the aureole of the
tilted Nelson Batholith, British Columbia: Implications for phase equilibria and pressure determination in
andalusite–sillimanite-type settings. Can Mineral 43:51–88
Petrík I, Broska I, Lipka J, Siman P (1995) Granitoid allanite-(Ce): substitution relations, redox conditions and REE
distributions (on an example of I-type granitoids, Western Carpathians, Slovakia). Geologica Carpathica 46:79–94
Pichavant M, Montel J-M, Richard LR (1992) Apatite solubility in peraluminous liquids: experimental data and an
extension of the Harrison–Watson model. Geochim Cosmochim Acta 56:3855–3861
Plank T, Cooper LB, Manning CE (2009) Emerging geothermometers for estimating slab surface temperatures.
Nat Geosci 2:611–615
Pochon A, Poujol M, Gloaguen E, Branquet Y, Cagnard F, Gumiaux C, Gapais D (2016) U–Pb LA-ICP-MS dating
of apatite in mafic rocks: Evidence for a major magmatic event at the Devonian-Carboniferous boundary in
the Armorican Massif (France). Am Mineral 101:2430–2442
Pyle JM, Spear FS, Cheney JT, Layne G (2005a) Monazite ages in the Chesham Pond Nappe, SW New Hampshire,
U.S.A.: Implications for assembly of central New England thrust sheets. Am Mineral 90:592–606
Pyle JM, Spear FS, Wark DA, Daniel CG, Storm LC (2005b) Contributions to precision and accuracy of monazite
microprobe ages. Am Mineral 90:547–577
Radulescu IG, Rubatto D, Gregory C, Compagnoni R (2009) The age of HP metamorphism in the Gran Paradiso
Massif, Western Alps: a petrological and geochronological study of “silvery micaschists”. Lithos 110:95–108
Rakovan J, Reeder RJ (1996) Intracrystalline rare earth element distributions in apatite: Surface structural influences
on incorporation during growth. Geochim Cosmochim Acta 60:4435–4445
Rakovan J, McDaniel DK, Reeder RJ (1997) Use of surface-controlled REE sectoral zoning in apatite from
Llallagua, Bolivia, to determine a single-crystal Sm–Nd age. Earth Planet Sci Lett 146:329–336
Rapp RP, Watson EB (1986) Monazite solubility and dissolution kinetics: implications for the thorium and light
rare earth chemistry of felsic magmas. Contrib Mineral Petrol 94:304–316, doi:10.1007/bf00371439
Rapp RP, Ryerson F, Miller CF (1987) Experimental evidence bearing on the stability of monazite during crustal
anaatexis. Geophys Res Lett 14:307–310
416 Engi
Rasmussen B (2005) Radiometric dating of sedimentary rocks: the application of diagenetic xenotime
geochronology. Earth-Sci Rev 68:197–243
Rasmussen B, Muhlig JR (2007) Monazite begets monazite: evidence for dissolution of detrital monazite and
reprecipitation of syntectonic monazite during low-grade regional metamorphism. Contrib Mineral Petrol
154:675–689
Rasmussen B, Bengtson S, Fletcher IR, McNaughton NJ (2002) Discoidal impressions and trace-like fossils more
than 1200 million years old. Science 296:1112–1115
Rasmussen B, Fletcher IR, Sheppard S (2005) Isotopic dating of the migration of a low-grade metamorphic front
during orogenesis. Geology 33:773-776 doi:10.1130/G21666.1
Rasmussen B, Fletcher IR, Muhling JR (2011) Response of xenotime to prograde metamorphism. Contrib Mineral
Petrol 162:1259–1277
Regis D, Cenki-Tok B, Darling J, Engi M (2012) Redistribution of REE, Y, Th, and U at high pressure: Allanite-
forming reactions in impure meta-quartzites (Sesia Zone, Western Italian Alps). Am Mineral 97:315–328
Regis D, Rubatto D, Darling J, Cenki-Tok B, Zucali M, Engi M (2014) Multiple metamorphic stages within an
eclogite-facies terrane (Sesia Zone, Western Alps) revealed by Th–U–Pb petrochronology. J Petrol 55:1429–
1456, doi:10.1093/petrology/egu029
Reimink JR, Davies JHFL, Waldron JWF, Rojas X (2016) Dealing with discordance: a novel approach for analysing
U–Pb detrital zircon datasets. J Geol Soc 173:577–585, doi:10.1144/jgs2015-114
Romer RL, Siegesmund S (2003) Why allanite may swindle about its true age. Contrib Mineral Petrol 146:297–
307, doi:10.1007/s00410-003-0494-6
Romer RL, Xiao Y (2005) Initial Pb-Sr(-Nd) isotopic heterogeneity in a single allanite-epidote crystal: implications
of reaction history for the dating of minerals with low parent-to-daughter ratios. Contrib Mineral Petrol
148:662–674
Romer RL, Rötzler J (2011) The role of element distribution for the isotopic dating of metamorphic minerals. Eur
metamorphism. Geology 184:123–138
Rubatto D, Müntener O, Barnhoorn A, Gregory C (2008) Dissolution–reprecipitation of zircon at low-temperature,
high-pressure conditions (Lanzo Massif, Italy). Am Mineral 93:1519–1529
Rubatto D, Hermann J, Berger A, Engi M (2009) Protracted fluid-present melting during Barrovian metamorphism
in the Central Alps. Contr Mineral Petrol 158:703–722, doi:10.1007/s00410-009-0406-5
Rubatto D, Regis D, Hermann J, Boston K, Engi M, Beltrando M, McAlpine S (2011) Yo-Yo subduction recorded
by accessory minerals (Sesia Zone, Western Alps) Nat Geosci 4:338–342, doi:10.1038/ngeo1124
Rubatto D, Chakraborty S, Dasgupta S (2013) Timescales of crustal melting in the Higher Himalayan Crystallines
Contrib Mineral Petrol 165:349–372, doi:10.1007/s00410-012-0812-y
Sano Y, Oyama T, Terada K, Hidaka H (1999) Ion microprobe dating of apatite. Chem Geol 153:249–258
Schärer U (1984) The effect of initial 230Th disequilibrium on young U Pb ages: The Makalu case, Himalaya. Earth
Geochem 83:199–230
Schoene B, Bowring SA (2007) Determining accurate temperature–time paths from U–Pb thermochronology: An
example from the Kaapvaal craton, southern Africa. Geochim Cosmochim Acta 71:165–185, doi:10.1016/j.
gca.2006.08.029
Secher K, Johnsen O (2008) Minerals in Greenland. Geology and Ore, GEUS 12:1–12
Seydoux-Guillaume A-M, Wirth R, Heinrich W, Montel J-M (2002a) Experimental determination of Thorium
partitioning between monazite and xenotime using analytical electron microscopy and X-ray diffraction
Rietveld analysis. Eur J Mineral 14:869–878
Seydoux-Guillaume AM, Paquette JL, Wiedenbeck M, Montel J-M, Heinrich W (2002b) Experimental resetting of
the U–Th–Pb systems in monazite. Chem Geol 191:165–181, doi:10.1016/S0009-2541(02)00155–9
Seydoux-Guillaume AM, Wirth R, Deutsch A, Scharer U (2004) Microstructure of 24–1928 Ma concordant
monazites; implications for geochronology and nuclear waste deposits. Geochim Cosmochim Acta 68:2517–
2527, doi:10.1016/j.gca.2003.10.042
damage, and disturbance of the U–Pb and Th–Pb chronometers. Chem Geol 330–331:140–158, doi:10.1016/j.
chemgeo.2012.07.031
Shazia J, Harlov D, Suzuki K, Kim S, Girish-Kumar M, Hayasaka Y, Ishwar-Kumar C, Windley B, Sajeev K (2015)
Linking monazite geochronology with fluid infiltration and metamorphic histories: Nature and experiment.
Lithos 236:1–15
Simmat R, Raith MM (2008) U–Th–Pb monazite geochronometry of the Eastern Ghats Belt, India: Timing and
spatial disposition of poly-metamorphism. Precambrian Res 162:16–39, doi:10.1016/j.precamres.2007.07.016
Skora S, Blundy J (2012) Monazite solubility in hydrous silicic melts at high pressure conditions relevant to
subduction zone metamorphism. Earth Planet Sci Lett 321–322:104–114, doi:10.1016/j.epsl.2012.01.002
Smith HA, Barreiro B (1990) Monazite U–Pb dating of staurolite grade metamorphism in pelitic schists. Contrib
Mineral Petrol 105:602–615, doi:10.1007/bf00302498
Smith HA, Giletti BJ (1997) Lead diffusion in monazite. Geochim Cosmochim Acta 61:1047–1055, doi:10.1016/
S0016-7037(96)00396–1
Smye AJ, Bickle MJ, Holland TJ, Parrish RR, Condon DJ (2011) Rapid formation and exhumation of the youngest
Alpine eclogites: a thermal conundrum to Barrovian metamorphism. Earth Planet Sci Lett 306:193–204
Smye AJ, Roberts NMW, Condon DJ, Horstwood MSA, Parrish RR (2014) Characterising the U–Th–Pb systematics
of allanite by ID and LA-ICPMS: Implications for geochronology. Geochim Cosmochim Acta 135:1–28
Spear FS (2010) Monazite–allanite phase relations in metapelites. Chem Geol, doi:10.1016/j.chemgeo.2010.10.004
Petrol 37:733–765
Spear FS, Pyle JM, Cherniak D (2009) Limitations of chemical dating of monazite. Chem Geol 266:218–230,
doi:10.1016/j.chemgeo.2009.06.007
Spear FS, Ashley KT, Webb LE, Thomas JB (2012) Ti diffusion in quartz inclusions: implications for metamorphic
time scales. Contrib Mineral Petrol 164:977–986
Stacey JS, Kramers JD (1975) Approximation of terrestrial lead isotope evolution by a 2-stage model. Earth Planet
Sci Lett 26:207–221
Steck A, Della Torre F, Keller F, Pfeifer H-R, Hunziker J, Masson H (2013) Tectonics of the Lepontine Alps:
ductile thrusting and folding in the deepest tectonic levels of the Central Alps. Swiss J Geosci 106:427–450,
doi:10.1007/s00015-013-0135-7
implications for the REE, Th and U geochemistry of crustal rocks. Chem Geol 300–301:200–220,
doi:10.1016/j.chemgeo.2012.01.007
Sun S-s, McDonough WF (1989) Chemical and isotopic systematics of oceanic basalts: implications for mantle
composition and processes. Geol Soc, London, Spec Publ 42:313–345
Suzuki K, Adachi M (1991) Precambrian provenance and Silurian metamorphism of the Tsubonosawa paragneiss
in the South Kitakami terrane, Northeast Japan, revealed by the chemical Th–U-total Pb isochron ages of
monazite, zircon and xenotime. Geochem J 25:357–376
Suzuki K, Adachi M, Tanaka T (1991) Middle Precambrian provenance of Jurassic sandstone in the Mino Terrane,
central Japan: Th–U-total Pb evidence from an electron microprobe monazite study. Sediment Geol 75:141–147
Suzuki S, Arima M, Williams IS, Shiraishi K, Kagami H (2006) Thermal history of UHT metamorphism in
the Napier Complex, East Antarctica: Insights from zircon, monazite, and garnet ages. J Geol 114:65–84,
doi:10.1086/498100
Teufel S, Heinrich W (1997) Partial resetting of the U–Pb isotope system in monazite through hydrothermal
experiments: An SEM and U–Pb isotope study. Chem Geol 137:273–281
Thomson T (1810) Experiments on allanite, a new mineral from Greenland. Trans R Soc Edinburgh: Earth Sci
8:371–386
Tobgay T, McQuarrie N, Long S, Kohn MJ, Corrie SL (2012) The age and rate of displacement along the Main
Central Thrust in the western Bhutan Himalaya. Earth Planet Sci Lett 319–320:146–158, doi:10.1016/j.
epsl.2011.12.005
Todd CS, Engi M (1997) Metamorphic field gradients in the Central Alps. J Metamorph Geol 15:513–530
Tomkins HS, Pattison DRM (2007) Accessory phase petrogenesis in relation to major phase assemblages in pelites
from the Nelson contact aureole, southern British Columbia. J Metamorph Geol 25:401–421
Tropper P, Manning CE, Harlov DE (2011) Solubility of CePO4 monazite and YPO4 xenotime in H2O and H2O–
NaCl at 800 °C and 1 GPa: Implications for REE and Y transport during high-grade metamorphism. Chem
Tucker NM, Hand M, Kelsey DE, Dutch RA (2015) A duality of timescales: Short-lived ultrahigh temperature
metamorphism preserving a long-lived monazite growth history in the Grenvillian Musgrave–Albany–Fraser
Orogen. Precambrian Research 264:204–234, doi:10.1016/j.precamres.2015.04.015
Vallini DA, Rasmussen B, Krapež B, Fletcher IR, McNaughton NJ (2002) Obtaining diagenetic ages from
metamorphosed sedimentary rocks: U–Pb dating of unusually coarse xenotime cement in phosphatic
sandstone. Geology 30:1083–1086
418 Engi
Vallini DA, Rasmussen B, Krapež B, Fletcher IR, Mcnaughton NJ (2005) Microtextures, geochemistry and
geochronology of authigenic xenotime: constraining the cementation history of a Palaeoproterozoic
metasedimentary sequence. Sedimentology 52:101–122
van Emden B, Thornber MR, Graham J, Lincoln FJ (1997) The incorporation of actinides in monazite and xenotime
from placer deposits in Western Australia. Can Mineral 35:95–104
Vazquez JA, Reid MR (2004) Probing the accumulation history of the voluminous Toba magma. Science 305:991–
994, doi:10.1126/science.1096994
Wasserburg G (1963) Diffusion processes in lead–uranium systems. J Geophys Res 68:4823–4846
Watson E, Harrison T (1984) Accessory minerals and the geochemical evolution of crustal magmatic systems: a
summary and prospectus of experimental approaches. Phys Earth Planet Inter 35:19–30
Wawrzenitz N, Krohe A, Rhede D, Romer RL (2012) Dating rock deformation with monazite: The impact of
dissolution precipitation creep. Lithos 134:52–74
Wawrzenitz N, Krohe A, Baziotis I, Mposkos E, Kylander-Clark AR, Romer RL (2015) LASS U–Th–Pb monazite
and rutile geochronology of felsic high-pressure granulites (Rhodope, N Greece): Effects of fluid, deformation
and metamorphic reactions in local subsystems. Lithos 232:266–285
Weber WJ (1990) Radiation-induced defects and amorphization in zircon. J Mater Res 5:2687–2697
Wendt I (1984) A three-dimensional U–Pb discordia plane to evaluate samples with common lead of unknown
isotopic composition. Chem Geol 46:1–12
White T, Ferraris C, Kim J, Madhavi S (2005) Apatite–an adaptive framework structure. Rev Mineral Geochem
57:307–401
Whitehouse MJ, Kumar GR, Rimša A (2014) Behaviour of radiogenic Pb in zircon during ultrahigh-temperature
metamorphism: an ion imaging and ion tomography case study from the Kerala Khondalite Belt, southern
India. Contrib Mineral Petrol 168:1–18
Williams I (2001) Response of detrital zircon and monazite, and their U–Pb isotopic systems, to regional
metamorphism and host–rock partial melting, Cooma Complex, southeastern Australia. Aust J Earth Sci
48:557–580
Williams ML, Jercinovic MJ (2002) Microprobe monazite geochronology: putting absolute time into microstructural
analysis. J Struct Geol 24:1013–1028
Geochem 83:153–182
Williams ML, Jercinovic MJ (2012) Tectonic interpretation of metamorphic tectonites: integrating compositional
mapping, microstructural analysis and in situ monazite dating. J Metamorph Geol 30:739–752, doi:10.1111/
j.1525–1314.2012.00995.x
Williams ML, Jercinovic MJ, Terry MP (1999) Age mapping and dating of monazite on the electron microprobe:
Deconvoluting multistage tectonic histories. Geology 27:1023–1026
Williams ML, Jercinovic MJ, Hetherington CJ (2007) Microprobe monazite geochronology: Understanding
geologic processes by integrating composition and chronology. Ann Rev Earth Planet Sci 35:137–175,
doi:10.1146/annurev.earth.35.031306.140228
Williams ML, Jercinovic MJ, Harlov DE, Budzyń B, Hetherington CJ (2011) Resetting monazite ages during fluid-
related alteration. Chem Geol 283:218–225, doi:10.1016/j.chemgeo.2011.01.019
Willigers BJA, Baker JA, Krogstad EJ, Peate DW (2002) Precise and accurate in situ Pb–Pb dating of apatite,
monazite, and sphene by laser ablation multiple-collector ICP-MS. Geochim Cosmochim Acta 66:1051–1066
Wing BA, Ferry JM, Harrison TM (2003) Prograde destruction and formation of monazite and allanite during contact
and regional metamorphism of pelites: petrology and geochronology. Contrib Mineral Petrol 145:228–250
Wolf MB, London D (1994) Apatite dissolution into peraluminous haplogranitic melts: an experimental study of
solubilities and mechanisms. Geochim Cosmochim Acta 58:4127–4145
Wood BJ, Blundy J (1997) A predictive model for rare earth element partitioning between clinopyroxene and
anhydrous silicate melt. Contrib Mineral Petrol 129:161–181
Yakymchuk C, Brown M (2014) Behaviour of zircon and monazite during crustal melting. J Geol Soc 171:465–479
Yang P, Pattison D (2006) Genesis of monazite and Y zoning in garnet from the Black Hills, South Dakota. Lithos
88:233–253, doi:10.1016/j.lithos.2005.08.012
Yi K, Cho M (2009) SHRIMP geochronology and reaction texture of monazite from a retrogressive transitional
layer, Hwacheon Granulite Complex, Korea. Geosci J 13:293, doi:10.1007/s12303-009-0028-y
Zeng L, Asimow PD, Saleeby JB (2005) Coupling of anatectic reactions and dissolution of accessory phases and
the Sr and Nd isotope systematics of anatectic melts from a metasedimentary source. Geochim Cosmochim
Acta 69:3671–3682
Zhu XK, O’Nions RK (1999) Zonation of monazite in metamorphic rocks and its implications for high temperature
thermochronology: a case study from the Lewisian terrain. Earth Planet Sci Lett 171:209–220
Vol. 83 pp. 419–441, 2017 13
Titanite Petrochronology
Matthew J. Kohn
Boise, ID 83725
USA

Titanite (CaTiSiO5) is a common mineral in calc-silicates, metamorphosed igneous rocks,
and calc-alkaline plutons. The mineral was first named by Martin Klaproth in 1795 for its high
content of the element titanium, which had been discovered only a few years prior, and named by
Klaproth for the Titans of Greek mythology. The alternate name sphene was proposed by Rene
Haüy in 1801 for the mineral’s characteristic wedge-shape (sphenos in Greek means “wedge”),
but in 1982 the IMA recommended that the name titanite be used in technical writing. The name
sphene is still used in the gem industry, and retains a loyal following among some mineralogists.
Titanite’s unusual crystal structure—including a 7-fold decahedral site—preferentially
takes up numerous geochemically interesting elements, especially U, which enhances its
geochronologic utility, but also other high field-strength elements like Zr, and the rare-earth
elements (REE). It is one of a handful of major Ti-bearing phases that occur in almost every
rock either as a silicate (titanite), as a pure Ti-oxide (rutile or anatase) or as a Fe–Ti oxide
(ilmenite or magnetite). Although usually present as a minor or accessory mineral, titanite
differs from many other accessory minerals in that its main chemical constituents participate
in reactions with other major minerals. Significant substitution of Al and OH enhances this
reactivity. Thus, although the stability and reactivity of accessory minerals such as monazite,
zircon, etc. are also tied to major mineral reactions (Pyle and Spear 1999; Ferry 2000; Wing et
al. 2003; Kohn and Malloy 2004; Tomkins and Pattison 2007; Spear 2010; Kohn et al. 2015,
etc.), titanite’s connection is much more direct and forms the basis of quantitative thermometry
and barometry. More generally, titanite has served a key role in understanding igneous,
metamorphic and ore-forming processes (Kerrich and Cassidy 1994; Frost et al. 2000) and
is even used to constrain the depositional ages of sedimentary rocks through chronologic
analysis of bacterial pseudomorphs (Banerjee et al. 2007; Calderon et al. 2013).
This review updates the outstanding comprehensive work of Frost et al. (2000), who reviewed
the crystal chemistry, phase relations and chronologic utility of titanite. I specifically address the
fundamental crystal structure, including crystal chemical idiosyncrasies and thermobarometric
equilibria; the stability fields and reactions responsible for forming igneous and metamorphic
titanite; the U–Pb and Sm–Nd chronologic systems, including potential diffusional biases;
petrochronologic case studies in metamorphic and igneous systems; and recommendations for
future refinements. While titanite may not be the most ubiquitous of minerals, it stands as a
singularly useful petrochronometer in certain common rock types (especially calc-silicates)
where other minerals such as zircon or monazite are either absent or less reactive.

1943-2666/17/0083-0013$05.00 (online)
420 Kohn
CRYSTAL CHEMISTRY OF TITANITE

Crystal structure and chemical substitutions
Titanite contains three structural sites—a tetrahedral site with Si, an octahedral site with
Ti, and an unusual 7-fold decahedral site with Ca (Fig. 1). The 7 oxygens of the decahedral
site are arranged in a distorted ring of 5 oxygens, with one oxygen above and one oxygen
below. The octahedral sites are arranged in chains parallel to the a-axis and are crosslinked
through the tetrahedral and decahedral sites (Fig. 1). In most titanite crystals, Si essentially
fills the tetrahedral site, but significant chemical substitution can occur in the octahedral and
decahedral sites. The most important of these substitutions include Sr, REE, Pb and U (and
possibly Na) for Ca, and Al, Fe2+, Fe3+, Zr, Nb and Ta for Ti. Charge balance among unlike
cations is achieved in two different ways. Many proposed substitutions follow typical crystal-
chemical principles of coupled cation substitutions on one or more sites (Fig. 1; Ribbe 1980;
Paterson and Stephens 1992; Smith et al. 2009), for example:
Nb5+ + Al3+ = 2Ti4+ (1)
REE3+ + Al3+ = Ca2+ + Ti4+ (2)
One oxygen in the titanite structure, however, is slightly underbonded (Ribbe 1980), and
consequently an unusual substitution involves replacement of this O1 oxygen with either OH
or F (Fig. 1), with charge balance normally achieved through substitution of Al or Fe3+ for Ti:
(Al, Fe)3+ + (OH, F)− = Ti4+ + O2− Fig. 1 Kohn (3a)
Partial operation of Reaction 3a induces high F content in highly aluminous titanite (Fig. 2;
Franz and Spear 1985). Alternatively, the hydroxyl substitution can be written as:
(Al, Fe)3+ + H+ = Ti4+ (3b)
Ca, Na, Sr, REE, U
O, F, Ti, Al, Zr, Fe2+,3+, Nb, Ta

OH
Si, (Al)
Proposed Chemical Substitutions
a O (Al, Fe)3+ + (OH, F)- = Ti4+ + O2-

Nb5+ + (Al, Fe)3+ = 2Ti4+
b 2REE3+ + Fe2+ = 2Ca2+ + Ti4+
REE3+ + (Al, Fe)3+ = Ca2+ + Ti4+
REE3+ + Na+ = 2Ca2+
Figure 1. Crystal structure and chemistry of titanite projected along c-axis showing tetrahedral (Si), octa-
hedral (Ti), and decahedral (Ca) sites. Geochemically important cation substitutions are listed. Oxygen at
the O1 site is shared between octahedra and can exhibit major F and OH substitutions. Other O sites, e.g.,
between octahedra and tetrahedra, are occupied solely by O. Crystal structure image based on http://www.
uwgb.edu/dutchs/petrology/Titanite Structure.HTM
Titanite Petrochronology 421
0.6 1.0
79-60 79-102
Ttn
Ttn
Py
Ti (atoms pfu)
F (atoms pfu)
0.4 75 0.8
0.
.0
1
)=
H
O
0
+ 0.5
(F
79-102
F/
0.2 0.6
0.25 Ttn
0.0
A 0.4
B Fl
0.0 0.2 0.4 0.6 0.0 0.2 0.4 0.6
Al+Fe3+ (atoms pfu) Al+Fe3+ (atoms pfu)
Figure 2. Compositional correlations and substitution of Al + Fe3+ in titanite (from Franz and Spear 1985).
Backscattered electron images of titanite from two marbles studied by Franz and Spear (1985) show zon-
ing between low Al-concentration (bright areas) and high Al-concentration (dark areas). Py = pyrite; Fl =
fluorite. Scale bars are 100 µm. (A) Fluorine broadly correlates with Al + Fe3+, but additional substitution of
OH is likely. (B) Substitution of Al + Fe3+ (and F + OH) occurs primarily for Ti4+ (and O2−).
where H+ protonates the underbonded O. Comparable protonation–deprotonation substitutions

are also observed in hydrous minerals such as tourmaline and amphibole. Complete operation
of Reaction 3b leads to the endmember vuagnatite [CaAl(OH|SiO4)], a mineral with a different
structure from titanite. Difficulty in measuring OH contents in minerals means that the
magnitude of the hydroxyl substitution is rarely determined in natural crystals, and because
the OH content of titanite is variable, a unique anion charge cannot be assigned. Thus, most
titanite compositions are normalized on a three-cation basis.
Thermometry and barometry
Thermometry and barometry may seem distinct from crystal chemistry, but require a clear
understanding of chemical substitutions, so are discussed here. Key advances over the last decade
now allow direct calculation of the temperature (T) and pressure (P) of crystallization using titanite
chemistry in equilibrium with other common minerals. The Zr-in-titanite thermometer (Hayden
et al. 2008; Fig. 3) relies on direct substitution of Zr4+ for Ti4+ according to the equilibrium:
CaTiSiO5 (titanite) + ZrSiO4 (zircon) =
(4)
CaZrSiO5 (Zr-titanite) + TiO2 (rutile) + SiO2 (quartz)
One major advantage to this equilibrium is that constituents are either sufficiently pure (titanite,
zircon, rutile and quartz) that Raoult’s law applies, or sufficiently dilute (Zr-titanite) that Henry’s
law applies, mitigating concerns about component activities except for unusually aluminous
titanite (Tropper and Manning 2008). The Zr-in-titanite calibration covers a wide range of
temperatures and pressures (Fig. 3A), and demonstrates a moderate P dependence (Fig. 3B).
Indeed, analytical errors propagate to only ~±2 °C for laser-ablation ICP-MS analyses, so the
largest sources of uncertainty commonly lie in the uncertainties in P and in the activity of
rutile [a(Rt)] in rutile-absent rocks. Pressures can be determined either in coexisting rocks or
through titanite-specific equilibria (Kapp et al. 2009). However, titanite and rutile rarely coexist
or exhibit equilibrium textures, so usually a(Rt) ≤ 1.0. A nominal a(Rt) of 0.75–0.85 may be
assumed based on mineral equilibria applied across a wide range of metamorphic rocks (Kapp
et al. 2009, Chambers and Kohn 2012), or the activity of rutile can be directly determined
in quartz-bearing rocks from the chemistry of coexisting amphibole, muscovite, or biotite
(Chambers and Kohn 2012), as long as compositions are within the range of calibration data.
422 Kohn
T (°C)
1000 900 800 700 600
3.0
[Zr] (ppm) 10 50 200
Experimental data a(Rt) = 0.85 5 20 100
10000 0.2
Natural data a(Ttn) = 0.95
0.5
500
Pressure (GPa)
±2σ
1000 2.0
Zr (ppm)
0 0 0 1000
-3. -2. -1.
100 0.0 2000
2.4 1.8 1.0 o
1.0 2Cz 1.0
t + H 2O
+R +
10 Qtz + Ttn 5000
3An 2.0
Pressure
Isopleths
1
A 0.0
B
0.0008 0.0009 0.0010 0.0011 0.0012 500 600 700 800 900 1000
1/T(K) Temperature (°C)
Figure 3. Basis for titanite thermobarometers. (A) Experimental and natural data used to calibrate the
Zr-in-titanite thermometer (Hayden et al. 2008). Isopleths show calibration at different pressures (in GPa).
(B) Pressure–temperature diagram, contoured for Zr concentrations [Zr, ppm] for the Zr-in-titanite ther-
mometer, and log(K) for the TZARS barometer (Kapp et al. 2009). Contours for Zr-in-titanite assume satu-
ration with quartz and zircon, but typical reduced activities of rutile [a(Rt)] and titanite [a(Ttn)] of 0.85 and
0.95 respectively. Error bar for T represents propagated analytical errors for LA-ICP-MS analysis (≤ ±2 °C),
and ±0.1 uncertainty in the activity of rutile (c. ±7 °C). Error bar for P represents propagated uncertainties
in activities. Qtz = quartz, Rt = rutile, Czo = clinozoisite, An = anorthite, Ttn = titanite.
Pressures of titanite crystallization, particularly in calc-silicates, may be estimated from

the TZARS equilibrium (Fig. 3B; Kapp et al. 2009):
CaTiSiO5 (titanite) + 3 CaAl2Si2O8 (anorthite) + H2O =
(5)
2 Ca2Al3Si3O12(OH) ([clino]zoisite) + TiO2 (rutile) + SiO2 (quartz)
A specific calibration has not been published, so instead automated calculations are made
using an internally consistent thermodynamic database such as Thermocalc (Holland and
Powell 2011). Although petrologists commonly avoid estimating P or T using equilibria that
involve volatile species, this reaction is relatively insensitive to a(H2O), and many calc-silicate
assemblages are restricted to XH2O ≥ 0.8 (Kapp et al. 2009; see later discussion). Although the
reaction has a moderate T-dependence, the largest source of uncertainty lies in estimating
activities. These include not only the activity for rutile, as discussed above, but also the activity
of zoisite or clinozoisite (most rocks contain epidote), and anorthite (plagioclase can exhibit a
wide range of compositions). Still, errors less than ~±0.1 GPa are possible.
Sector zoning
Titanite exhibits a strong propensity to develop sector zoning, which manifests as
crystallographically controlled differences in compositions (Paterson et al. 1989). Sector
zoning is ascribed to differences in chemical partitioning among distinct crystal faces
(Hollister 1970), which is preserved where crystal growth rates were rapid compared to
intracrystalline diffusion (Watson and Liang 1995). Cation diffusivities in titanite are
generally quite low (see later discussion), so even relatively slow crystal growth can lead
to sector zoning. Among all possible crystal faces, the {111} crystal form in titanite is most
commonly developed, but other crystal faces [{102}, {001}, etc.] can also form. These
different faces partition trace elements quite differently, for example non-{111} sectors have
higher Zr and U contents by as much as a factor of ~3, while Pb and REE contents can be
higher by a factor of ~2 (Fig. 4A, B; Hayden et al. 2008; Bauer 2015; Walters 2016). Exactly
which sectors are exposed in a thin section or grain mount depends on the angle at which the
plane of the section intersects the crystal (Fig. 4C; Paterson and Stephens 1992). Because the
Fig. 4 Kohn
Zr U Age Zr U Age Zr U Age Zr U Age Zr U Age

1 145 205 7 125 155 23±2 10 165 375 24±2 1 145 105 20±8 4 120 130 26±4
2 175 285 8 75 205 19±2 11 105 165 18±4 2 90 135 20±3 5 110 105 25±5
3
3 250 150 25±2 9 80 205 21±5 3 95 135 18±4 2
4 125 165 22±2
5 95 205 17±2
8 9 11 4
6 130 165 24±2
8 1
3
4 4 5 7 10 4
1 2 6
2 Zr U Age 3
1 3 3 390 555 23±1
7 4 290 325 24±2
1 Zr U Age
5 6 5 240 250 18±2
2 1 255 665 24±1
6 290 425 23±1
Zr U Age 2 140 325
7 205 500 24±2
100 µm 1 220 400 100 µm 5 3 235 550 26±2
8 230 285 23±2
A 2 255 445 20±3 B 4 240 275 24±2
850
(001)
(102) (100) High-Zr, U
Temperature (°C)
(001)
800
(102) (111)
(111)
750
(111)
(100)
(001) Low-Zr, U
(100) 700
(110) 0.0 10.0 20.0 30.0
C D Age (Ma)
Figure 4. Titanite textures and chemistry illustrating compositional impact of sector zoning. Darkest
symbols indicate highest-Zr sectors, intermediate symbols indicate moderate-Zr sectors or mixed analy-
ses, and white symbols indicate low-Zr sectors. (A–B) Backscattered electron images of metamorphic
titanite grains from a Himalayan calc-silicate gneiss (sample KN14–51a), showing dark gray, {111}
zones with low Zr and U contents flanking light gray core zones with high Zr and U contents. Data from
Walters (2016). (C) Typical crystal form of titanite showing dominant {111} faces with minor faces
along other directions. Expected patterns of sector zoning (darker vs. lighter shading) are illustrated for
transverse vs. longitudinal sections. Modified from Paterson and Stephens (1992). (D) Titanite tempera-
ture–time data for KN14–51a define parallel trends with a ~50 °C temperature offset, reflecting high- vs.
low-Zr data collected in different sectors. Data from Walters (2016).
{111} faces normally dominate, however, the other sectors are most typically found in titanite
cores or along an elongate medial domain that is flanked by {111} domains (Fig. 4A–C).
Not every titanite crystal exhibits sector zoning, but the intersectoral compositional
differences do affect petrogenetic applications of titanite geochemistry. The Zr-in-titanite
thermometer was calibrated for {111} sector chemistry (Hayden et al. 2008), so use of the
unusually high Zr content of other sectors will result in calculated temperatures that are too
high (e.g., by 40–80 °C; Fig. 4D; Hayden et al. 2008). Ironically, the non-{111} sectors also
have the highest U / Pb, so provide superior age resolution. But until a systematic calibration of
Zr partitioning among different sectors is established (beyond a nominal “factor of 3”), these
higher-quality ages cannot be linked unequivocally to precise Zr-in-titanite temperatures.
IGNEOUS TITANITE
Petrogenesis
In igneous rocks, titanite occurs most commonly as a late-stage mineral in felsic calc-
alkaline plutons. Its stability is strongly influenced by oxygen fugacity (fO2) and water fugacity
(fH2O). In fO2–T space, two key reactions limit titanite stability (Fig. 5A; Wones 1989; Frost and
Lindsley 1992; Xirouchakis and Lindsley 1998; Frost et al. 2000):
Fig. 5 Kohn
424 Kohn
1.0
Ttn-common 1 bar Fe*
Mt Fay
ΔlogfO2 (-FMQ)
0.0 Ttn
+M
Cpx Ilm Cpx
+I t
-1.0
+
lm
Qt
Cpx + Ilm + Qtz
z
+
Ti Ttn Ca
O2
Ttn + Fay
-2.0 Ttn-rare
+Qtz
-3.0
A
-4.0
400 600 800 1000 1200
T (°C)
Mt90 Hd25
3.0 Mt85
Ttn + Mt
Mt80
ΔlogfO2 (-FMQ)
Typical calc- Mt75

2.0 alkaline magmas Hd50
1.0 0.3 GPa

Hd80
0.0 Ilm90 Hd100
Cpx + Ilm Ilm94 Ilm92
B
-1.0
600 700 800 900 1000
T (°C)
Ttn-stable with H2O

Ttn z + H +
Qt mph
Ca lm +
+ M 2O
-A O 2
I
t+
ΔlogfO2 (-FMQ)
Ttn + Mt + Qtz
Cpx + Ilm + O2
Cp
Cpx +
ph
x
Am
Ca-A
Ca-
+
t + + H 2O
Ttn + Qtz
M
Ilm + Qtz + H2O
+ m
Ca-Amph + Ttn
Ttn px + Il
mph + Ilm
C Ttn-absent
+ H2O
C
T (°C) a(H2O)
Figure 5. Phase equilibrium constraints on igneous titanite stability. (A) fO2 relative to fayalite–
magnetite–quartz equilibrium vs. T. Cpx = clinopyroxene, Fay = fayalite, Ilm = ilmenite, Mt =
magnetite, Qtz = quartz, Ttn = titanite, Usp = ulvöspinel. The composition diagrams are pro-
jected from quartz and combine Fe2+ and Fe3+. Ilmenite and magnetite are assumed to contain
small amounts of hematite and ulvöspinel. Small black bar spans typical igneous rock composi-
tions. Modified from Xirouchakis et al. (2001a). (B) fO2 relative to fayalite–magnetite–quartz
equilibrium vs. temperature, with detail of Cpx + Ilm = Ttn + Mt equilibrium contoured for the
mole fraction of hedenbergite (XHd). Thin lines illustrate contours of ilmenite (XIlm) and magne-
tite (XMt) for coexisting oxides. Modified from Frost et al. (2000) and Xirouchakis et al. (2001b)
(C) fO2 relative to fayalite–magnetite–quartz equilibrium vs. temperature and water fugacity in
H2O-bearing systems. Modified from Xirouchakis et al. (2001b). See also Harlov et al. (2006).
3 CaFeSi2O6 (clinopyroxene) + 3 FeTiO3 (ilmenite) + O2 =

(6)
3 CaTiSiO5 (titanite) + 2 Fe3O4 (magnetite) + 3 SiO2 (quartz)
CaFeSi2O6 (clinopyroxene) + FeTiO3 (ilmenite) =
(7)
CaTiSiO5 (titanite) + Fe2SiO4 (olivine)
High-temperatures and low fO2 stabilize ilmenite rather than titanite, so with decreasing T or
increasing fO2, titanite may form via Reactions like (6) and (7). Projecting from quartz into
CaO–TiO2–(FeO + Fe2O3) space shows that these reactions reflect crossing tie-line relations
(tie-line flip) that should discontinuously produce titanite as a late-stage major titanian phase
after earlier-formed ilmenite. Felsic calc-alkaline magmas crystallize at relatively high fO2 and
at moderate- to low-temperatures, in the stability field of titanite + magnetite (Fig. 5B). In
particular, evolution along ilmenite-buffered equilibria towards higher DfO2 during cooling is
expected drive formation of titanite (Frost and Lindsley 1992; Fig. 5B). These relationships
explain both the occurrence of titanite + magnetite + quartz assemblages in many felsic rocks
(Wones 1989) and the occurrence of titanite within latest-stage distillates of igneous bodies
like the Skaergaard layered mafic intrusion (Xirouchakis et al. 2001b)
High water activity can also stabilize titanite (Fig. 5C), as exemplified by the reaction:
7 CaFeSi2O6 (clinopyroxene) + 3 FeTiO3 (ilmenite) + 5 SiO2 (quartz) + 2 H2O =
(8)
2 Ca2Fe5Si8O22(OH)2 (Ca-amphibole) + 3 CaTiSiO5 (titanite)
which transforms clinopyroxene + ilmenite-bearing assemblages into amphibole + titanite-
bearing assemblages (Fig. 5C; Frost et al. 2000). Crystallization of anhydrous minerals during
cooling increases the activity of water, ultimately stabilizing titanite at the expense of ilmenite.
Especially in the context of high fO2 calc-alkaline magmas, titanite is expected to form with
amphibole, late in the crystallization sequence (Xirouchakis et al. 2001b). Relatively high Ti
solubility in amphibole tends to increase the stability of titanite-bearing assemblages towards
higher temperatures and lower fH2O, whereas moderate Ti solubility in clinopyroxene increases
the stability of ilmenite-bearing assemblages to slightly higher fO2.
Trace element geochemistry
Trace element geochemistry of igneous titanite has focused primarily on REE. Several
experimental studies confirm titanite’s preference for the middle REE (Fig. 6A; Green and
Pearson 1986; Tiepolo et al. 2002; Prowatke and Klemme 2005). Fitted to lattice strain quasi-
parabolic models (Blundy and Wood 1994), partition coefficients reach maxima at about the
ionic radius of Dy through Sm, notably lower than the ionic radius of 7-fold coordinated Ca (c.
1.06 Å, comparable to Pr). Thus Ca may not be perfectly sized for the decahedral site, rather it
is simply the most abundant divalent cation with approximately the correct ionic radius. Some
dependence of partitioning on melt composition is observed, and the discrimination of middle
REE from light and heavy REE becomes more pronounced with greater melt polymerization,
i.e., the parabolas become tighter (Prowatke and Klemme 2005). Natural titanite–whole rock
composition ratios invariably exceed experimental partition coefficients, and do not always
conform to simple parabolic models (Fig. 6B). This discrepancy arises because titanite generally
forms late. Many incompatible elements like REE are more concentrated in the residual melt,
whereas crystallization of other minerals with different REE preferences (e.g., light REE in
apatite, heavy REE in zircon, etc.) affects availability of REE and consequently the REE patterns.
Thus, titanite geochemistry is expected to reflect the later stage processes of melt crystallization.
Igneous titanite crystals can exhibit sector zoning, and non-{111} sectors consistently
show higher REE and Zr concentrations than {111} sectors (Fig. 6C). Extensive data from
the Half Dome granodiorite (Bauer 2015) suggest differences of a factor of ~2 for REE, and a
426 Kohn Fig. 6 Kohn
100 1000
Andesite 900°C A Core B
D(titanite/melt)
{111}
10 100
Natural
Lamproite 850°C volcanic
Dacite 1150°C titanite Half Dome
Rhyolite 1150°C Granodiorite
Lu Yb Er Y DyTb Gd Eu Sm Nd Pr Ce La Lu Yb Er Y Dy Gd Eu Sm Nd Pr Ce La
1.0 10
0.90 0.95 1.0 1.05 1.10 0.90 0.95 1.0 1.05 1.10
Ionic Radius (Å) Ionic Radius (Å)
105 105
Half Dome granodiorite C Kiruna iron district D
Sample/Chondrite
Sample/Chondrite
104 104 1 mm
Core sectors
Whole-rock
x100 Rims: c. 1870 Ma
Rims
103 103
{111} sectors Cores: c. 2050 Ma

1 mm
102 102
La Pr Sm Gd Dy Ho Tm Lu La Pr Sm Gd Dy Ho Tm Lu
Zr Ce Nd Eu Tb Y Er Yb Ce Nd Eu Tb Y Er Yb
Figure 6. Trace element patterns in titanite. (A) Experimental titanite–melt partitioning data (D = mineral/
melt partition coefficient) vs. ionic radius, fit with lattice strain models. Minimum strain (maximum D) typi-
cally occurs for MREE (Gd, Eu and Sm). Data and models from Green and Pearson (1986), Tiepolo et al.
(2002) and Prowatke and Klemme (2005). (B) Partitioning data for igneous titanite vs. whole-rock or glass,
showing generally flatter trends and larger D-values than expected from experiments (note change in vertical
scale). Low-temperatures and fractional crystallization likely impact composition of melt from which titanite
crystallized. Low Eu likely reflects prior plagioclase fractionation. Core vs. {111} sectors discriminate more
differently for MREE than for LREE and HREE. Summarized from Green and Pearson (1986), Solgadi and
Sawyer (2008), and Bauer (2015). (C) Zirconium and REE data for igneous titanite and whole rocks from the
Half Dome granodiorite. Core sectors (blue) have higher MREE and Zr contents and larger negative Eu anom-
alies than {111} sectors (orange). Different zones are distinguishable through back-scattered electron images
(inset schematic). Data from Bauer (2015). (D) REE data for titanite from the Kiruna iron district, Sweden.
Patchy zoning in back-scattered electron images (inset schematic) distinguishes cores vs. rims, which have
different ages, REE concentrations, and REE patterns. Data from Smith et al. (2009).
~25% enrichment in Zr for non-{111} sectors compared to {111}. Interestingly, the depth of
the Eu anomaly for non-{111} sectors exceeds that of {111} sectors by ~1 / 3, suggesting that
{111} sectors may take up more Eu2+. Differences in Eu2+ uptake could be tested by measuring
inter-sectoral differences in other divalent cations, such as Sr, Pb, and Ba.
In ore deposits, titanite commonly exhibits complex chemical zoning, and REE patterns
may be used to identify fluid sources, whether magmatic or hydrothermal (Fig. 6D; e.g., Smith
et al. 2009). In general, light REE enrichment indicates a magmatic origin for ore-forming
fluids. For example, the REE patterns for titanite rims at Kiruna compare favorably with trends
for titanite from the Half Dome granodiorite, suggesting that ore-forming fluids at Kiruna
were derived from associated magmas. Other REE patterns, such as flat profiles in titanite
cores, are ascribed to other processes, such as initial hydrothermal alteration, either during or
soon after deposition of the original volcanic sequence (Smith et al. 2009). In some deposits,
however, host-rock chemistry appears to buffer REE patterns (Chelle-Michou et al. 2015).
Titanite Petrochronology Fig. 427
7 Kohn
2.5
Felsic rocks
Rutile
Metarhyolite
2.0
Titanite
Pressure (GPa)
Stable
1.5
Metadacite
Titanite
1.0 Stable
Ilmenite
0.5
A
2.0
Mafic rocks,
Calc-silicates Rutile 0
Pressure (GPa)
1.5 1.
)=
CO
2
X(
1
1.0 0.
Metabasite
)2 =
O Titanite
C tz
X( Stable
+Q
0.5 C al O 2
+ C
Calc- Rt tn + Ilmenite
T
Silicates B
300 400 500 600 700 800
Temperature (°C)
650
Titanite
Stable t Ilmenite
600
Cpx Cpx+B
Ep Bt (-Ilm)
T (°C)
550 Ep+Amp
Dol+Ms Rutile
Amp
500
P=0.7 GPa +Qtz +Fsp +Cal

450
C
0.95 0.80 0.65 0.50 0.35 0.20 0.05
X(H2O)
Figure 7. Phase equilibrium constraints on metamorphic titanite stability. (A, B) Composition-specific
mineral assemblage diagrams for felsic through mafic bulk compositions (metarhyolite: Menold et al.
2009; metadacite: Spencer et al. 2013; metabasite: Carty et al. 2012; Kohn et al. 2015). Curves (thick red
lines) from Frost et al. (2000) show a limiting reaction in calc-silicates that stabilizes titanite at the expense
of rutile for two different X(CO2) values. (C) T–X diagram at P = 0.7 GPa for a typical calc-silicate composi-
tion (whole-rock data from Cottle et al. 2011).
428 Kohn
METAMORPHIC TITANITE
Petrogenesis
The pressure–temperature (P–T) stability field of titanite has now been calculated for
specific rocks spanning a wide range of bulk compositions (Fig. 7). For metaigneous rocks
ranging in composition from metabasite to metarhyolite, the titanite stability field exhibits
a wedge-like (arguably “sphenoidal”) geometry (Fig. 7A, B). High Ca and Ti in metamafic
rocks tend to stabilize titanite at lower P–T conditions than for compositionally intermediate
and felsic rocks. Among titanian phases, rutile is stable at the highest pressures and lowest
temperatures, titanite at intermediate P–T conditions, and ilmenite at the lowest pressures
and highest temperatures. The common occurrence of titanite rims on rutile (e.g., Lucassen et
al. 2010a) attests to P–T paths that evolve with increasing T or decreasing P. In contrast, the
transition from titanite to rutile recorded by inclusions in garnet or, rarely, rutile overgrowths
on matrix titanite, reflects prograde paths that involve increasing P or decreasing T. Examples
include the Cordillera Darwin metamorphic complex, southern Chile (Kohn et al. 1993), and
Franciscan complex blueschists and eclogites exposed at Jenner, California (Krogh et al.
1994). Thus, titanite in metaigneous rocks is expected to record aspects of the earlier and later
stages of metamorphism, but not peak conditions.
In calc-silicate rocks, titanite becomes stable only at the highest temperatures and X(H2O)
conditions (Fig. 7C). Thus, unlike in mineral assemblage diagrams for metaigneous rocks, rutile
and titanite flank an intermediate stability region occupied by ilmenite. Such models further
demonstrate that titanite and epidote coexist only at high X(H2O), which reduces uncertainties in
P calculated using the TZARS barometer (Kapp et al. 2009; Fig. 3B). In contrast to its behavior
in metaigneous rocks, titanite in calc-silicates may be hoped to record processes occurring near
the peak of metamorphism. Note, however, that the F–OH substitution (see Fig. 2) can strongly
influence the P–T-conditions of formation, and the irregular zoning patterns in these samples
point to variable compositions of coexisting fluid on small spatial or temporal scales.
CHRONOLOGIC SYSTEMS
U–Pb
Tilton and Grünenfelder (1968) first recognized the chronologic utility of titanite, owing to
a high U content that can reach hundreds of ppm (e.g., Fig. 4). Its affinity for U notwithstanding,
not all rocks contain much U, and unlike other high-U minerals such as zircon and monazite,
titanite also has a strong affinity for Pb. Consequently, most titanite U–Pb measurements require
correction for substantial common Pb before an age can be calculated. Although historically
titanite data were collected using ID-TIMS with a minimum of interferences on isotopic masses,
many data are now collected in situ with LA-ICP-MS (Kylander-Clark 2017, this volume),
where high 204Hg backgrounds compromise measurement of 204Pb. Because of this interference,
corrections are now made in reference to inverse isochrons (Tera-Wasserburg diagrams),
regressing less radiogenic and more radiogenic measurements to derive an age (Fig. 8A).
An unusual concern in dating titanite is that different grains or domains may take up
different common Pb compositions (Essex and Gromet 2000). Rather than sampling a
homogeneous whole-rock Pb value, titanite and other high-Pb minerals have been proposed to
inherit a Pb composition that reflects local, isotopically heterogeneous, reactant phases (Romer
and Rötzler 2003, 2011; Romer and Xiao 2005). Variations in common lead 207Pb / 206Pb may
reach ~2% (data from Essex and Gromet 2000, Romer and Xiao 2005), which is at least 20
times larger than analytical precisions for ID-TIMS analysis (Schoene and Baxter 2017, this
volume). Low U / Pb matrix minerals include feldspars, epidote, mica, pyroxene, amphibole
and tourmaline (e.g., see Kohn and Corrie 2011), and in principle permit variations in common
Pb composition to be characterized. Large common Pb corrections mean that highly precise
0.8 0.520
Brown
Common Pb range
ra n
cy
High-U
c u s io
0.518 00
Aceci
Clear 36 0
Pr
0.6 206 203 0
200 197 194 30
0.516 50
TIMS 26
Nd/144Nd
Pb/206Pb
LA-ICP-MS
0.4 0.514
LA-ICP-MS
207
143
TIMS
00
0.512
32
12 00 00
0.2
16 24
0.510
00
0
0
0
A B
60
40
30
20
0.0 0.508
0 10 20 30 40 50 0.0 0.1 0.2 0.3 0.4 0.5
238
U/206Pb 147
Sm/144Nd
Figure 8. (A) Hypothetical example of U–Pb data for titanite, illustrating form of inverse isochron for
a 200 Ma titanite analysis with moderate common Pb. Insets show relative sizes of error ellipses for
ID-TIMS (0.1%) vs. LA-ICP-MS (2–3%) and importance of common Pb correction in assigning age
uncertainties. Inner vs. outer brackets show age uncertainty when common Pb composition is uncertain
by only 0.1% vs. 2% respectively. (B) Example of Sm–Nd dating of titanite illustrating apparent age
variation. Data scatter reflects overprinting events at c. 3600 and c. 2650 Ma, as well as variation in initial
143
Nd/144Nd. Modified from Amelin (2009).
(e.g., c. 0.1%) titanite ages are not always possible and that each titanite domain must be
evaluated independently for U / Pb and potential impact of common Pb variability. For example,
consider a hypothetical 200 Ma titanite grain with 207Pb / 206Pb = 0.2 and 238U / 206Pb = 24.4, with
common Pb 207Pb / 206Pb = 0.700. Assuming ID-TIMS measurement errors of ±0.1%, an inverse
isochron would yield an age of 200.13 ± 0.22 Ma (0.11% error; Fig. 8A). If the 207Pb / 206Pb
value is subject to 2% uncertainty, however, the error increases to 1.3 Ma (0.65%; Fig. 8A).
Polygenetic or polymetamorphic rocks may be more susceptible to common Pb
compositional variation, if the source of this variability indeed lies in reactant phases (Romer
and Rötzler 2011). For example, a granite that is subsequently metamorphosed could pose
serious chronologic problems if common Pb in two different titanite crystals is derived
from low-U feldspar (virtually no radiogenic Pb) vs. high U accessory minerals (mostly
radiogenic Pb). In contrast, simple crystallization processes in a magma operating on sub-
Ma time scales should not generate large variations in Pb isotope ratios, as long as recharge
or assimilation does not alter 207Pb / 206Pb spatially or temporally. Clearly, the details of
the geologic setting define the level of concern. A small uncertainty in the common Pb
composition may be warranted in many rocks, but that assumption must be independently
rationalized or demonstrated with measurements on low-U minerals.
For two reasons, use of single collector LA-ICP-MS or ion microprobe is less sensitive
to assumptions about common Pb composition. First, common Pb composition and its
variation can be measured directly on low-U minerals. Second, analytical uncertainties already
encompass likely variation in 207Pb / 206P. For example, a nominal ~2% variation in 207Pb / 206Pb
falls within the 2–4% analytical uncertainty of single-collector LA-ICP-MS and ion probe,
so is essentially already accounted for. Moreover, the isotopic measurements on titanite, not
common Pb, typically control the age uncertainty. For example, consider again a 200 Ma
titanite grain with 207Pb / 206Pb = 0.2 and 238U / 206Pb = 24.4, but with measurement errors of 2%
and 3%, respectively. Assuming common Pb 207Pb / 206Pb = 0.700 ± 0.014 (2% error) implies an
age of 200.1 ± 6.2 Ma. Reducing the uncertainty in common Pb composition from 2% to 0.1%
decreases the age uncertainty by only 0.1 Ma, to ±6.1 Ma (Fig. 8A). Even doubling the error
on common Pb composition to 4% increases the age uncertainty by only 0.4 Ma to ± 6.6 Ma.
430 Kohn
Sm–Nd
In principle, high REE contents, and a general preference for middle REE (Fig. 6) permits
Sm–Nd dating of titanite as an alternative to U–Pb dating. For accurate dating, a 147Sm / 144Nd
ratio ≥ 0.2 is desirable, and although many igneous grains have ratios < 0.2, some do reach
values of c. 0.5. In his study of the Archean Itsaq Gneiss Complex, western Greenland
(near Isua), Amelin (2009) compared U–Pb and Sm–Nd ages for the same separated titanite
grains and multi-grain splits. Data for Sm–Nd have 147Sm / 144Nd between ~0.1 and ~0.5, and
scatter broadly (Fig. 8B). Much of the scatter represents mixing of ages between initial and
overprinting metamorphic events at ~3600 and ~2650 Ma (Fig. 8B). Some data, however,
suggest that initial 143Nd / 144Nd was not constant for each grain. Because Sm–Nd ages are
strongly dependent on initial 143Nd / 144Nd, future dating will require special efforts to ensure
the initial ratio is determined accurately.
Diffusional biases
Most chronologic studies of titanite find younger U–Pb ages than for coexisting zircon,
and this age difference is typically ascribed to continuous diffusional loss of Pb from titanite
during cooling or to diffusional resetting during later events. A classic titanite dataset from
the Western Gneiss Region, Norway illustrates discordant titanite data distributed between
an orthogneiss protolith age of 1657 Ma and an overprinting metamorphic age of 395 Ma
(Fig. 9A; Tucker et al. 2004). Tucker et al. (2004) clearly favored diffusive resetting to explain
Fig. 9 Kohn
this discordancy, but they also note that simple mixing of protolith and metamorphic grains
and domains would yield an identical distribution of data. The ubiquity of complex zoning
in titanite grains (e.g., Figs. 2, 4, 6) recommends characterization of zoning patterns using
back-scattered electron imaging, X-ray mapping, or trace element analysis prior to any dating
attempts. Because the dated titanite separates were not characterized chemically prior to
analysis, the source of discordancy (diffusive Pb loss vs. mixing) remains ambiguous.
T (°C)
1200 1000 800 600
0.30 -17
1600
Titanite U-Pb ages from the Diffusion Data
Western Gneiss Region, Norway 1657 for Titanite
Devonian sedimentary rocks ±3 Ma Ta
0.25 1400 -19
Upper/uppermost allochthons
Gula/Seve/Støren nappes Nb
log10D(m2/s)
Middle allochthon 1200 Sr

Basement units
Pb/238U
0.20 -21
Sample locality
1000
Zr
206
Nd
0.15 -23 Pb
800
Trondheim Pb,U Dabie Pb,U WGR

600 Kristian- Pb, Nepal
0.10 sund -25 Zr,Pb,U,
Th Pamir
400 50 km
395±3 Ma
0.05
15 samples A -27
B
1.0 2.0 3.0 4.0 7 8 9 10 11
207
Pb/235U 10000/T(K)
Figure 9. (A) Chronologic data from titanite separates from rocks in the Western Gneiss Region, Norway,
showing a well-defined chord between 1657 ± 3 and 395 ± 3 Ma. Inset shows regional geology and location
of samples. From Tucker et al. (2004). (B) Diffusivities (D = diffusion coefficient) of various elements in
titanite, and constraints on D’s from natural samples. Boxes with downward-pointing arrows indicate maxi-
mum limits on D for Zr, Pb, U and Th derived from natural samples. Thick gray lines show experimental
data for Zr and Pb, which differ from natural constraints. Thick black lines show experimental data for
other cations. Modified from Cherniak (2006, 2015); natural data from Kohn and Corrie (2011), Gao et al.
(2012), Spencer et al. (2013), and Stearns et al. (2016).
The longstanding debate about Pb diffusion rates in titanite has led to a wide range of
proposed closure temperatures (Tc) based on various natural settings: 450–500 °C (Mattinson
1978), 600 ± 25 °C (Heaman and Parrish 1991; Spear and Parrish 1996), 500–670 °C (Mezger
et al. 1991), > 680 ± 20 °C (Scott and St-Onge 1995), > 650 °C (Pidgeon et al. 1996), > 712 °C
(Zhang and Schärer 1996), > 775 °C (Kohn and Corrie 2011), ≥ 825 °C (Gao et al. 2012). Arguably,
experimentally-determined Pb diffusion rates (Cherniak 1993) have influenced discussion most
strongly. These relatively high diffusivities (thick gray line, Fig. 9B) imply Tc of c. 600 °C for
100 µm diameter grains cooling at 10 °C / Ma. In her subsequent study of Sr diffusion, however,
Cherniak (1995) commented on the surprising inconsistency between Sr and Pb diffusion rates;
their experimental diffusivities differ by ~4 orders of magnitude even though both cations are
divalent, substitute at the same crystallographic site, and have nearly indistinguishable ionic
radii. The difficulty in evaluating diffusive resetting or estimating Tc using natural whole- or
multi-grain data as a field-check to experiments is that titanite can grow over a large range
of temperatures (Kerrich and Cassidy 1994; Frost et al. 2000), and should not be expected to
preserve the same age as, say, a zircon or monazite (e.g., Corfu 1996; Verts et al. 1996). So the
fact that a titanite crystal records a younger age than zircon or monazite provides little direct
information regarding Pb diffusivities and Tc—titanite might be quite retentive of Pb but simply
grow later. For example, Lucassen and Becchio (2003) found that titanite crystallized over a time
span of c. 50 Ma in different rocks from the exhumed roots of an arc.
Recent petrochronologic studies (Kohn and Corrie 2011; Gao et al. 2012; Spencer et
al. 2013; Stearns et al. 2016) place relatively precise limits on D (uncertainties of a factor of
2–4) and suggest slow diffusion for Pb and other cations. Estimates of D require solving the
diffusion equation, and all such solutions involve the combined parameter D·Dt, where Dt is the
duration at maximum T. Thus, to constrain D in titanite, the duration of peak metamorphism
must be known well. The four highlighted studies are unusual in providing relatively tight
constraints for Dt, as well as for T. For the studies of Gao et al. (2012; Dabie-Sulu) and Spencer
et al. (2013; Western Gneiss Region), maximum D is estimated from the preservation of relict
pre-metamorphic titanite cores, c. 50–100 µm in diameter, inside metamorphic titanite grains.
Given protolith ages, relict cores could have lost, at most, ~50% of their Pb during subsequent
metamorphism, which implies D·Dt / a2 = 0.14, where a is grain radius (Crank 1975, his
Eqn. 6.19); for 90% loss, D·Dt / a2 = 0.30, so even a large uncertainty in the maximum percent
loss translates into a small error in the limit for D. For Dabie-Sulu rocks, the duration at
maximum temperature for the titanite core-overgrowth couple is estimated at c. 5–15 Ma, based
on zircon petrochronology that demonstrates protracted heating (e.g., Liou et al. 2012). For
Western Gneiss Region rocks, titanite petrochronology requires a minimum duration of 15 Ma
at a temperature of c. 780 °C (Spencer et al. 2013; Kohn 2015). This latter limit is a very robust
minimum (D is a robust maximum) because titanite rims post-date maximum pressures, and
titanite cores could have lost Pb during prograde and maximum pressure metamorphism prior
to titanite rim overgrowth. Temperatures are determined both from Zr-in-titanite and regional
thermometry. Uranium shows consistent differences in core vs. overgrowth concentrations,
implying broadly similar resistance to resetting.
For the studies of Kohn and Corrie (2011, Nepal) and Stearns et al. (2016, Pamir), D is
estimated by solving the diffusion equation for a fixed boundary condition and identifying the
maximum magnitude of diffusive loss with depth in a crystal:
 C ( x ) − Cr   x 
erf −1  =  (9)
 C 0 − Cr   4 D∆t 
where C(x), Cr and Co represent concentrations at position x, at the rim, and initially in
the titanite interior respectively. Nepal titanite shows no resolvable age resetting at depths
432 Kohn
within 5 µm of the crystal surface for durations of 15 Ma at peak temperature, as determined
from other titanite grains from the same rock. Some grains show c. 10 Ma age differences over
distances of 15 µm, and in general there is close correspondence between U–Pb age and Zr-
in-titanite temperature, irrespective of position within each grain. Pamir titanite shows ~4 µm
age gradients produced over a maximum duration of 3–6 Ma. Similarly steep concentration
gradients are observed in U, Th, and Zr. Preserved sector zoning in titanite, despite protracted
high temperatures, further supports slow diffusivities for Zr and U (Fig. 4, Walters 2016).
Solving the corresponding diffusion equations for maximum D indicates Pb diffusivities
at least 4 orders of magnitude lower than experiments, but similar to experimental constraints
on Sr. Thus, multiple natural datasets reconcile the curious experimental mismatch between Sr
and Pb diffusivities. Lead actually appears to diffuse at about the same rate as Sr, as would be
expected from ionic charge and radius considerations. These data imply that Pb diffusion and
chronologic resetting are ineffective at temperatures below c. 800 °C. In support of this view,
at least one study that originally interpreted younger titanite ages as reflecting diffusive Pb loss
(Spear and Parrish 1996) has been reinterpreted using independent data. In light of monazite
petrochronology, the “young” titanite ages in the Valhalla Complex (Spear and Parrish 1996)
overlap the peak of metamorphism (Spear 2004; c. 820 °C), implying that titanite can preserve
peak metamorphic ages even when temperatures exceed 800 °C. Lucassen and Becchio (2003)
also argued that the c. 50 Ma age differences observed among titanite separates from different
rocks reflected deformation-enhanced recrystallization or growth associated with thermal
pulses, not with differential diffusional resetting. Natural intracrystalline gradients for Zr, U
and Th that are as steep or steeper than Pb (Stearns et al. 2016) imply that these elements
diffuse at least as slowly as Pb, if not more slowly. Although diffusivities for U and Th have not
been measured experimentally, results for Zr differ from experimental data (Cherniak 2006)
by about 2 orders of magnitude. Natural data imply that Zr-in-titanite temperatures can be
preserved on spatial scales of c. 10 µm, even at temperatures of c. 800 °C.
EXAMPLES
Temperature–time histories from single rocks
Data from high-grade calc-silicate gneisses in the central Nepal Himalaya illustrate the
potential of combined titanite geochronology and thermometry to illuminate temperature–time
(T–t) histories (Kohn and Corrie 2011). These rocks consist of relatively simple high-grade
assemblages of hornblende + clinopyroxene + quartz + plagioclase + biotite + calcite + titanite +
tourmaline + apatite + zircon. Analyses of low-U hornblende, clinopyroxene, biotite and
tourmaline provide measures of 207Pb/206Pb to anchor inverse isochrons, whereas in situ spot
analyses within thin sections (Fig. 10A) and depth profiles of separated titanite grains (inset, Fig.
10A,B) provide constraints on T and age, via Zr-in-titanite thermometry and U–Pb geochronology.
The depth profile data (Fig. 10B) are novel in providing T and age constraints with µm spatial
resolution, and demonstrate preservation of systematic changes in T and age over length scales
≤15 µm. Indeed, it was the failure of these data to conform to the predictions of experimentally
estimated diffusivities that led to the recognition that the diffusivity of Pb in titanite must be
quite slow (Fig. 9; Kohn and Corrie 2011). Relatively large age uncertainties (typically ~5%,
but up to ~10%) reflect low Pb contents for such young grains and high proportions of common
Pb, but are still sufficient to resolve T–t histories with acceptable precision. It is only the high U
content of these calc-silicates that permits a T–t history to be resolved, and similar attempts with
metabasites elsewhere in the Himalaya have failed (unpubl. data).
The T–t history reveals slow heating from ~700 to ~775 °C between ~38 Ma and ~23 Ma,
with slight cooling to c. 760 °C by ~21 Ma (Fig. 10C). A second, structurally lower rock
yields a similar, albeit higher-T history. No titanite ages younger than ~21 Ma are observed,
Fig. 10 Kohn
Spot Analyses Bt 850 ~25 Ma ~30 Ma 35.1 37.0

31.4 ±2.0
, 25,100 µm Cpx 800 ±1.8 31.0
~35 Ma
24.8 25.0
750
Ttn
Modified
±3.0
Cpx 700 19.0
Ttn 781 °C 753 °C 734 °C
Temperature (°C)
850 ~20 Ma ~25 Ma ~25 Ma 37.0
Cpx
Age (Ma)
800 31.0
25.9 26.6
750 25.0
Ttn ±3.4
Modified
21.1 ±4.9
±3.2
700 19.0
775 °C 779 °C 778 °C
850 ~30 Ma ~30 Ma ~30 Ma 37.0
Modified
Hbl Depth 800 29.9 28.2 29.4 31.0
Profile ±2.3 ±2.5 ±2.6
750 25.0
700 19.0
737 °C 738 °C 740 °C
Cpx 0.0 4.0 8.0 12.0 16.0
A B Distance (µm)
800 800
Pelite
Solidus Mnz
Temperature (°C)
600
750
Thrust
Initiation Ms
400
Mnz Titanite crystal cross-section S STDS N
700
6 km
Depth MCT, STDS
profile ( )
Zrn Initiation
200 MCT
Spot analyses ( ) Ttn

650 0
15.0 20.0 25.0 30.0 35.0 40.0 0.0 10.0 20.0 30.0 40.0
C Age (Ma) Age (Ma)
Figure 10. Titanite petrochronologic data from a calc-silicate gneiss, central Nepal Himalaya, illustrating
how temperature–time curves are developed for a single rock. All data from Kohn and Corrie (2011). Ti-
tanite grains imaged in Fig. 4 reflect similar structural levels and metamorphic conditions. (A) Photomicro-
graph showing locations of laser ablation analyses. Bt = biotite; Cpx = clinopyroxene; Hbl = hornblende;
Ttn = titanite. Other matrix minerals include zircon, quartz, calcite and feldspar. Inset shows separated
titanite that was analyzed in depth-profile mode. (B) Depth profile data for 3 titanite grains, showing
different zoning in Zr (temperature) and U–Pb ages. Similar background shading indicates comparable
temperature and age domains. (C) Temperature time history derived from titanite and regional data show-
ing slow heating followed by rapid cooling as bounding structures initiated above and below the samples.
Mnz = monazite Th–Pb ages; Ms = muscovite 40Ar/39Ar ages; MCT = Main Central Thrust; STDS = South
Tibetan Detachment System. Insets illustrate types of analysis, typical zoning observed in backscattered
electron images (scale bar is 100 µm; Zrn = zircon), and structural position of sample relative to MCT and
STDS. A second sample, c. 3 km structurally below, shows a similar T–t history.
and all temperatures are consistently above muscovite dehydration-melting. Such protracted
slow heating at such high temperatures poses an interesting problem for understanding
orogenesis because models of lower crustal flow (e.g., Beaumont et al. 2001) predict either
that extensional shearing should nucleate at the top of a weakened crustal channel or that
thrust-sense shearing should preferentially seek out weakened layers. The South Tibetan
Detachment System did not form until much later (c. 21–22 Ma; see summary of Sachan et al.
2010), challenging applicability of the channel flow model. The growing recognition of cryptic
metamorphic discontinuities associated with intra-GHS thrusts (first identified using monazite
petrochronology by Kohn et al. 2004; cf. Montomoli et al. 2013), however, may explain the
titanite data if other thrusts were active above the level of these samples.
434 Kohn
Late-stage monazite growth during melt crystallization constrains cooling below

~700 °C to 17–21 Ma (Corrie and Kohn 2011; Fig. 10C). In combination with regional
muscovite ages of c. 15 Ma (Martin et al. 2015), these data show a profound change in T–t
history coinciding with regional initiation of the structurally lower Main Central Thrust
and the structurally higher South Tibetan Detachment System. Cooling likely resulted from
two contributing processes. Transfer of rocks to the hanging wall of underlying thrusts
(e.g., the Main Central Thrust) and transport through a lateral thermal gradient would have
terminated slow heating and dramatically increased cooling rates (e.g., Kohn et al. 2004;
Kohn 2008). Simultaneously, initiation of extensional denudation along the South Tibetan
Detachment System may have cooled these rocks if the shear zone propagated proximally
to these samples. In conjunction with regional petrochronologic observations (Corrie and
Kohn 2011), titanite T–t data generally support in-sequence thrust models of orogenesis
modified by extensional processes, but are harder to explain by lower crustal flow.
Temperature–time histories from multiple rocks
Chronologic resolution of ±3 to 10% provides age uncertainties of only a few million years
for young orogens (Alps, Himalaya, etc.), but these uncertainties expand to tens of Ma for older
orogens. In these instances, a different approach is needed, specifically pooling all analyses of
titanite grains and domains from a rock to deduce its average T–t point. Because different rocks
experience different degrees of titanite growth and recrystallization during metamorphism (e.g.,
Lucassen and Becchio 2003; Spencer et al. 2013), a robust regional T–t history may be constructed.
For the ultrahigh-pressure rocks of the Scandinavian Caledonides, Spencer et al. (2013)
developed T–t points on a rock-by-rock basis, typically collecting Zr and U–Pb data on
c. 30 spots distributed across one or more titanite grains in each sample (inset, Fig. 11A).
Titanite grains analyzed included grains within host gneisses as well as grains from leucocratic
segregations. Excluding a few instances of scattered or bimodal data, each rock defines a
precise inverse isochron age and T (Fig. 11A). Combining T–t points across multiple rocks
defines a regional T–t path (Fig. 11B; Kohn et al. 2015; Kohn 2016). Titanite is not commonly
stable at UHP conditions (Fig. 7), so must form during exhumation below ~1.8 GPa (inset,
Fig. 11B), via reaction of rutile plus a calcic phase. Thus, T–t points reflect a late-stage
exhumation and cooling history. These data show slow cooling between ~405 and ~390 Ma,
followed by rapid cooling to muscovite closure at ~385 Ma.
The composite titanite T–t data yield several important insights. First, they demonstrate
hot exhumation, markedly unlike T–t paths from typical subduction zones that should exhibit
significant cooling during exhumation (e.g., Gerya et al. 2002). Clearly, mechanisms of post-
UHP exhumation can differ from wedge dynamics expected in subduction zones (e.g., Cloos
1982; Gerya et al. 2002). Second, protolith U–Pb ages are at least partially preserved in the cores
of some titanite grains. Preservation of Proterozoic ages and disparate U contents through the
entire UHP cycle, including residence for at least 10–15 Ma at temperatures of at least 775 °C,
provides strict constraints on Pb and U diffusivities (Fig. 9B). Third, temperatures above typical
partial melting reactions were sustained for c. 15 Ma, and ages overlap the timing of post-tectonic
pegmatite intrusion (Fig. 11B). Thus the rocks experienced protracted residence at mid/deep
crustal levels, where titanite crystallized from partial melts, then reacted with other minerals
or recrystallized in response to deformation (Spencer et al. 2013). Last, numerous estimates of
peak UHP conditions have been proposed based on Sm–Nd ages of garnet and U–Pb ages of
zircon (see summary of Kohn et al. 2015). Yet, temperatures were generally too high to preserve
peak metamorphic Sm–Nd ages (e.g., see Burton et al. 1995, Ganguly et al. 1998), and zircon
is expected to record retrograde, not prograde or peak ages (Kohn et al. 2015). Thus, these data
unequivocally constrain exhumation to have initiated prior to ~405 Ma. In conjunction with
estimates of peak UHP metamorphic ages of c. 415 to 407 Ma (e.g., Terry et al. 2000, Kylander-
Clark et al. 2009), the titanite data imply exhumation rates on the order of 1–3 cm / yr.
1.0
T (°C)
Western Gneiss Region 740 760 780
A07- A07- G97-
8 24H 22G3 05Y1
6
±2σ
n
4
A0
72
Pb/206Pb
2
4H
0
0.5 50 100 150 200
Zr (ppm)
207
405±2 Ma
392±2 Ma
385±7 Ma
100 µm
G9
7 05
Ttn Y1 A07
22
G 3
0.0
A
0.0 6.0 12.0 18.0
238
U/206Pb
800
Temperature (°C)
G9705Y1 4.0 ?
A0722G3
775
3.0 UHP
P (GPa)
A0724H
750 Less than
2.0
~1.5 GPa
Post-tectonic
Rutile
Pegmatites
e
Titanit
725 1.0
To Ms
500 600 700 800

700
B T (°C)
375 385 395 405 415 425
Age (Ma)
Figure 11. Titanite data from the Western Gneiss Region, Norway, illustrating how T–t curves are
developed from multiple rocks. All data from Spencer et al. (2013). (A) U–Pb data generally sug-
gest indistinguishable ages but variable amounts of common Pb among spots; Zr contents (upper
inset) commonly exhibit heterogeneity. Lower inset illustrates analytical style, with 10 rim and
5 core analyses. Modified from Spencer et al. (2013). (B) Composite T–t history, showing slow
cooling between ~405 and ~395 Ma, and rapid cooling afterwards. Error bars include a typical
1–2% calibration error. Inset shows general P–T stability of titanite. Ages must reflect post-UHP
conditions. Modified from Kohn et al. (2015).
Pressure–time histories
If the pressures and ages of formation for different titanite crystals or domains can be
determined, a pressure-time history can be derived, permitting burial or exhumation rates to
be inferred. In their study of metamorphic titanite from two different calc-silicate nodules
from the Dora Maira massif, western Italian Alps, Rubatto and Hermann (2001) identified
compositionally distinctive domains in titanite whose ages could be linked to an overall
P–T history (Fig. 12). One sample was minimally overprinted, and titanite grains contained
cores inherited from a previous event and well-developed rims with highly sodic omphacite
inclusions (Jd38). Solving the thermodynamic expression of the reaction:
436 Kohn
6 TiO2 (rutile) + Ca3Al2Si3O12 (grossular) + 3 CaMgSi2O6 (diopside)
(10)
= Mg3Al2Si3O12 (pyrope) + 6 CaTiSiO5 (titanite)
for observed mineral compositions ties the titanite rims and their inclusions to UHP conditions
of c. 3.5 GPa at ~35 Ma. In the second sample, relict high-pressure cores with moderately sodic
omphacite inclusions (Jd15) were overgrown by rims that were texturally equilibrated with late-
stage amphibolite-facies mineral assemblages, including low-Na pyroxene (Jd06). Solving the
thermodynamic expression for the albite–jadeite–quartz reaction, in conjunction with regional
thermobarometry, Rubatto and Hermann (2001) calculated much lower pressures at c. 33 Ma
(c. 1 GPa) and c. 32 Ma (c. 0.5 GPa; Fig. 12). Altogether the petrologic and chronologic data
indicate initial exhumation from UHP conditions at rates ≥ 3 cm / yr, slowing progressively to
~1.5 and 0.5 cm / yr as rocks reached typical crustal levels. Fig. 12 Kohn
As in other terranes, these titanite grains preserve inherited core ages, despite peak
metamorphic temperatures of c. 750 °C, perhaps suggesting slower Pb diffusion than expected
from experiments (Cherniak 1993). We do not know, however, how long the Dora Maira rocks
resided at maximum temperatures, so calculations of Pb diffusivity are not possible.
4.0
34.5±2.4 Ma
35.1
Jd38 ±0.9
UHP Ma
3.0 253±7 Ma
Coesite
Pressure (GPa)
HP
m/y
Jd15 Quartz
c
LP
3.4
31.3
2.0
±1.8 Ma 33.1
±1.6 Ma Qtz
Jd 20 +
Ab 85
32.9±0.9 Ma
1.0
Jd 06 + Qtz
1.6 cm/yr Ab 92
31.8±0.5 Ma
0.5 cm/yr
0.0
200 400 600 800
Temperature (°C)
Figure 12. Titanite data and P–T–time history from Dora Maira massif calc-silicates, western Italian Alps.
Insets show titanite textures evident in back-scattered electron images and locations of several analytical
spots. Several ages were averaged to derive the pooled age listed for each stage on the P–T path. Titanite
contains inherited cores and omphacite inclusions with compositions distinctive of ultrahigh-pressure
(Jd38) and later high-pressure (Jd15) conditions. UHP = ultrahigh-pressure, HP = high pressure, LP = low
pressure; Jd = jadeite (component in clinopyroxene), Qtz = quartz, Ab = albite. Scale bars are 50 µm. Con-
straint at lowest P–T conditions is derived from fission track age of 29.9 ± 1.4 Ma on zircon. All errors are
±2σ. From Rubatto and Hermann (2001).
Igneous processes
Despite numerous ages collected for igneous titanite, relatively little direct
petrochronologic work has attempted to link ages to chemistry, temperature, etc. One
major hurdle is that petrochronologic investigations inherently require more than one age.
The typically short timescales of igneous processes dramatically complicate this endeavor,
because rapid, spatially resolute analytical techniques like LA-ICP-MS and SIMS generally
lack the precision to resolve small age differences (Kylander-Clark 2017; Schmitt and Vazquez
2017, both this volume). One ongoing study (Schmitz and Crowley 2014) examines magma
chamber dynamics for the Fish Canyon Tuff through the analysis of relatively large (c.
0.5 mm-diameter) titanite grains. Crystals were extracted, doubly polished to thicknesses of
~300 µm, imaged using back-scattered electrons, and analyzed for trace elements using LA-
ICP-MS. This approach discriminated two generations of titanite growth. Cores are REE-rich
and U + Sr-poor with large Eu anomalies, whereas rims are REE-poor and U + Sr-rich with
small Eu anomalies. Reaction zones with intermediate compositions separate cores and rims.
The trace element data are interpreted to reflect rejuvenation of the magma chamber driven
by underplated mafic magmas. A mafic flux reacted with original titanite grains, which were
then overgrown by titanite rims with distinct trace element compositions. Crucially, ID-TIMS
dating of different titanite grains suggests older ages for cores (28.4 to 29.0 Ma) than for rims
(c. 28.2 Ma). The latter age is chemically linked to magma rejuvenation and is indistinguishable
from the age of eruption (Kuiper et al. 2008; Wotzlaw et al. 2013).
These data support previous zircon petrochronologic interpretations of the Fish Canyon Tuff
magma chamber (Wotzlaw et al. 2013) that inferred c. 200 ka of initial crystallization to produce
a 75–80% crystalline mush, followed by c. 200 ka of remelting associated with underplating
of andesitic magmas and infiltration of hot, hydrous fluids, culminating in eruption at 28.2 Ma.
Titanite cores must have grown during the initial, nearly complete crystallization of the magma
chamber, partially dissolved during fluid influx, then regrown soon prior to eruption.
FUTURE DIRECTIONS
Progress on the following topics would help promote more accurate petrochronologic
interpretations of titanite:
Diffusivities
Profound differences distinguish diffusion rates determined experimentally (Cherniak
1993, 2006) vs. estimates from natural composition and age gradients (Fig. 9). A more
concerted program of analysis is needed, both experimentally and based on natural samples,
to explain these discrepancies. Even if natural constraints on cation diffusion rates in titanite
are broadly correct, higher precision observations are needed to refine estimates of D.
Chemical domain structure
Sector-zoning aside, many metamorphic titanite crystals show patchy zoning (e.g.,
Figs. 2, 4, 6, 12). What causes this behavior? Does titanite dissolve and reprecipitate continuously
and randomly, or do textural and local bulk compositional factors, especially heterogeneities
in fluid composition and availability, influence how titanite grows or dissolves (e.g., Lucassen
et al. 2010b)? Answers to these questions would help target grains more efficiently for
petrochronologic analysis, and support thermodynamic and trace element modeling efforts.
Models
Major element solution models for titanite are still lacking (although see Tropper and
Manning 2008), and Al- F- and OH-components are not incorporated into comprehensive
thermodynamic datasets. While the basic reactions stabilizing titanite are known (Figs. 5, 7),
a refined understanding of titanite stability fields and expected compositional variability
of titanite across P–T space would enhance the petrologic utility of titanite compositions.
Simultaneously, trace element mass balance is difficult to model, particularly where different
sectors partition trace elements differently, and titanite apparently undergoes chaotic
dissolution-reprecipitation. Both processes can potentially bias trace element mass balance
and trace element patterns (e.g., Amelin 2009), and should be characterized and quantified.
438 Kohn
Rutile activity
Many equilibria, including Zr-in-titanite thermometry (Fig. 3), assume a(Rt). Although
estimates of a(Rt) are possible in many quartz-bearing metamorphic rocks (Chambers and Kohn
2012), application to igneous rocks, which contain dissimilar compositions to the calibration
dataset, is lacking. A more comprehensive assessment of a(Rt) across rock types and P–T
conditions is needed to fully realize the potential of equilibria involving titanian minerals.
Common Pb
Ultimately, as the precision of isotope measurements improves, the variability in
common Pb 207Pb/206Pb taken up by different titanite crystals and domains will limit age
resolution (Fig. 8A). Techniques and case studies are needed not only to identify the range
of 207Pb/206Pb variation but also to determine the cause of variations in common Pb among
titanite grains or within domains of single grains. If local mineral reactions control common
Pb (e.g., Romer and Rötzler 2011), methods will need to be developed to identify how other
minerals release Pb and how titanite acquires it.
ACKNOWLEDGMENTS
This research was funded by NSF grants EAR-1321897, -1419865, and -1545903. Thanks
to Gerhard Franz and Jesse Walters for providing images and data used in Figs. 2 and 4, Gerhard
Franz and John Schumacher for detailed reviews, and Martin Engi for editorial handling.
REFERENCES
Amelin Y (2009) Sm–Nd and U–Pb systematics of single titanite grains. Chem Geol 261:52–60, doi:10.1016/j.
chemgeo.2009.01.014
Banerjee NR, Simonetti A, Furnes H, Muehlenbachs K, Staudigel H, Heaman L, Van Kranendonk MJ (2007)
Direct dating of Archean microbial ichnofossils. Geology 35:487–490, doi:10.1130/g23534a.1
Bauer JE (2015) Complex zoning patterns and rare earth element varitions across titanite crystals from the Half
Dome Granodiorite, central Sierra Nevada, California. M. S. University of North Carolina at Chapel Hill
Beaumont C, Jamieson RA, Nguyen MH, Lee B (2001) Himalayan tectonics explained by extrusion of a low-
Blundy J, Wood B (1994) Prediction of crystal–melt partition coefficients from elastic moduli. Nature 372:452–454
Burton KW, Kohn MJ, Cohen AS, O’Nions RK (1995) The relative diffusion of Pb, Nd, Sr and O in garnet. Earth
Calderon M, Prades CF, Herve F, Avendano V, Fanning CM, Massonne HJ, Theye T, Simonetti A (2013) Petrological
vestiges of the Late Jurassic-Early Cretaceous transition from rift to back-arc basin in southernmost Chile:
New age and geochemical data from the Capitan Aracena, Carlos III, and Tortuga ophiolitic complexes.
Geochem J 47:201–217
Carty JP, Connelly JN, Hudson NFC, Gale JFW (2012) Constraints on the timing of deformation, magmatism
and metamorphism in the Dalradian of NE Scotland. Scottish J Geol 48:103–117, doi:10.1144/sjg2012–407
Chambers JA, Kohn MJ (2012) Titanium in muscovite, biotite, and hornblende: Modeling, thermometry and rutile
activities in metapelites and amphibolites. Am Mineral 97:543–555
Chelle-Michou C, Chiaradia M, Selby D, Ovtcharova M, Spikings RA (2015) High-resolution geochronology
of the Coroccohuayco porphyry-skarn deposit, Peru: A rapid product of the Incaic Orogeny. Econ Geol
110:423–443
Cherniak DJ (1995) Sr and Nd diffusion in titanite. Chem Geol 125:219–232
Cherniak DJ (2006) Zr diffusion in titanite. Contrib Mineral Petrol 152:639–647
Cherniak DJ (2015) Nb and Ta diffusion in titanite. Chem Geol 413:44–50, doi:10.1016/j.chemgeo.2015.08.010
Cloos M (1982) Flow melanges – numerical modeling and geologic constraints on their origin in the Franciscan
subduction complex, California. Geol Soc Am Bull 93:330–345, doi:10.1130/0016–7606(1982)93<330:fm
nmag>2.0.co;2
Corfu F (1996) Multistage zircon and titanite growth and inheritance in an Archean gneiss complex, Winnipeg
River Subprovince, Ontario. Earth Planet Sci Lett 141:175–186
Corrie SL, Kohn MJ (2011) Metamorphic history of the central Himalaya, Annapurna region, Nepal, and
implications for tectonic models. Geol Soc Am Bull 123:1863–1879, doi:10.1130/b30376.1
Cottle JM, Waters DJ, Riley D, Beyssac O, Jessup MJ (2011) Metamorphic history of the South Tibetan Detachment
System, Mt. Everest region, revealed by RSCM thermometry and phase equilibria modelling. J Metamorph
Geol 29:561–582, doi:10.1111/j.1525–1314.2011.00930.x
Crank J (1975) The Mathematics of Diffusion. Oxford University Press, London
Essex RM, Gromet LP (2000) U–Pb dating of prograde and retrograde titanite growth during the Scandian orogeny.
Geology 28:419–422
Franz G, Spear FS (1985) Aluminous titanite from the Eclogite Zone, south central Tauern Window, Austria. Chem
Geol 50:33–46
Frost BR, Lindsley DH (1992) Equilibria among Fe-Ti oxides, pyroxenes, olivine, and quartz: Part II. Application.
Am Mineral 77:1004–1020
Chem Geol 172:131–148
Ganguly J, Tirone M, Hervig RL (1998) Diffusion kinetics of samarium and neodymium in garnet, and a method
for determining cooling rates of rocks. Science 281:805–807
Gao X-Y, Zheng Y-F, Chen Y-X, Guo J (2012) Geochemical and U–Pb age constraints on the occurrence
of polygenetic titanites in UHP metagranite in the Dabie orogen. Lithos 136:93–108, doi:10.1016/j.
lithos.2011.03.020
channel: A numerical simulation. Tectonics 21:doi:10.1029/2002TC001406
Green TH, Pearson NJ (1986) Rare-earth element partitioning between sphene and coexisting silicate liquid at high
pressure and temperature. Chem Geol 55:105–119, doi:10.1016/0009–2541(86)90131–2
Harlov D, Tropper P, Seifert W, Nijland T, Forster H-J (2006) Formation of Al-rich titanite (CaTiSiO4O–
CaAlSiO4OH) reaction rims on ilmenite in metamorphic rocks as a function of fH2O and fO2. Lithos 88:72–84,
doi:10.1016/j.lithos.2005.08.005
155:529–540
Heaman L, Parrish R (1991) U–Pb geochronology of accessory minerals. In: Applications of radiogenic isotope
systems to problems in geology. Short Course Handbook. Vol 19. Heaman L, Ludden JN, (eds). Mineral
Assoc Canada, Toronto, Canada, p 59–102
Hollister LS (1970) Origin, mechanism, and consequences of compositional sector-zoning in staurolite. Am
Mineral 55:742–766
Kapp P, Manning CE, Tropper P (2009) Phase-equilibrium constraints on titanite and rutile activities in mafic
epidote amphibolites and geobarometry using titanite–rutile equilibria. J Metamorph Geol 27:509–521,
doi:10.1111/j.1525–1314.2009.00836.x
Kerrich R, Cassidy KF (1994) Temporal relationships of lode gold mineralization to accretion, magmatism,
metamorphism and deformation—Archean to present: A review. Ore Geol Rev 9:263–310, doi:10.1016/0169–
1368(94)90001–9
Kohn MJ (2008) P–T–t data from central Nepal support critical taper and repudiate large-scale channel flow of the
Greater Himalayan Sequence. Geol Soc Am Bull 120:259–273, doi:10.1130/b26252.1
Mineral 101:25–42
evidence for a static hot channel in the Himalayan orogen. Earth Planet Sci Lett 311:136–143
silicates: Implications for age determinations. Geochim Cosmochim Acta 68:101–113, doi:10.1016/s0016–
7037(03)00258–8
Kohn MJ, Spear FS, Dalziel IWD (1993) Metamorphic P–T paths from Cordillera Darwin, a core complex in
Tierra del Fuego, Chile. J Petrol 34:519–542
Himalaya of central Nepal. Earth Planet Sci Lett 228:299–310, doi:10.1016/j.epsl.2004.10.007
Krogh EJ, Oh CW, Liou JG (1994) Polyphase and anticlockwise P–T evolution for Franciscan eclogites and
blueschists from Jenner, California, USA. J Metamorph Geol 12:121–134
Kuiper KF, Deino A, Hilgen FJ, Krijgsman W, Renne PR, Wijbrans JR (2008) Synchronizing rock clocks of Earth
history. Science 320:500–504, doi:10.1126/science.1154339
440 Kohn
Kylander-Clark ARC, Hacker BR, Johnson CM, Beard BL, Mahlen NJ (2009) Slow subduction of a thick ultrahigh-
pressure terrane. Tectonics 28:doi:10.1029/2007TC002251
Lucassen F, Becchio R (2003) Timing of high-grade metamorphism: Early Palaeozoic U–Pb formation ages of
titanite indicate long-standing high-T conditions at the western margin of Gondwana (Argentina, 26–29°S).
Lucassen F, Dulski P, Abart R, Franz G, Rhede D, Romer RL (2010a) Redistribution of HFSE elements during
rutile replacement by titanite. Contrib Mineral Petrol 160:279–295, doi:10.1007/s00410–009-0477–3
Lucassen F, Franz G, Rhede D, Wirth R (2010b) Ti–Al zoning of experimentally grown titanite in the system CaO–
Al2O3–TiO2–SiO2–NaCl–H2O–(F): Evidence for small-scale fluid heterogeneity. Am Mineral 95:1365–1378,
doi:10.2138/am.2010.3518
Martin AJ, Copeland P, Benowitz JA (2015) Muscovite 40Ar/39Ar ages help reveal the Neogene tectonic evolution of
the southern Annapurna Range, central Nepal. Geol Soc, London, Spec Publ 412:199–220
Mattinson JM (1978) Age, origin, and thermal histories of some plutonic rocks from the Salinian Block of
California. Contrib Mineral Petrol 67:233–245, doi:10.1007/bf00381451
Menold CA, Manning CE, Yin A, Tropper P, Chen XH, Wang XF (2009) Metamorphic evolution, mineral chemistry
and thermobarometry of orthogneiss hosting ultrahigh-pressure eclogites in the North Qaidam metamorphic
belt, Western China. J Asian Earth Sci 35:273–284, doi:10.1016/j.jseaes.2008.12.008
Mezger K, Rawnsley C, Bohlen S, Hanson G (1991) U–Pb garnet, sphene, monazite and rutile ages: Implications
for the duration of high grade metamorphism and cooling histories, Adirondack Mts., New York. J Geol
99:415–428
Montomoli C, Iaccarino S, Carosi R, Langone A, Visona D (2013) Tectonometamorphic discontinuities within the
Greater Himalayan Sequence in Western Nepal (Central Himalaya): Insights on the exhumation of crystalline
rocks. Tectonophys 608:1349–1370
Montomoli C, Carosi R, Iaccarino S (2015) Tectonometamorphic discontinuities in the Greater Himalayan
Sequence: a local or a regional feature? Geol Soc, London, Spec Publ 412:25–41
Paterson BA, Stephens WE (1992) Kinetically induced compositional zoning in titanite: implications for accessory-
phase/melt partitioning of trace elements. Contrib Mineral Petrol 109:373–385, doi:10.1007/bf00283325
Paterson BA, Stephens WE, Herd DA (1989) Zoning in granitoid accessory minerals as revealed by backscattered
electron imagery. Mineral Mag 53:55–61
Pidgeon RT, Bosch D, Bruguier O (1996) Inherited zircon and titanite U–Pb systems in an archaean syenite
from southwestern Australia: Implications for U–Pb stability of titanite. Earth Planet Sci Lett 141:187–198,
doi:10.1016/0012–821x(96)00068–4
silicate melt. Geochim Cosmochim Acta 69:695–709, doi:10.1016/j.gca.2004.06.037
reactions. Geol Mat Res 1:1–49
Ribbe PH (1980) Titanite. Rev Mineral 5:137–154
Romer RL, Rötzler J (2003) Effect of metamorphic reaction history on the U–Pb dating of titanite. Geol Soc
Special Publ 220:147–158
Romer RL, Rötzler J (2011) The role of element distribution for the isotopic dating of metamorphic minerals. Eur
Romer RL, Xiao Y (2005) Initial Pb–Sr(–Nd) isotopic heterogeneity in a single allanite–epidote crystal:
implications of reaction history for the dating of minerals with low parent-to-daughter ratios. Contrib Mineral
Petrol 148:662–674
Sachan HK, Kohn MJ, Saxena A, Corrie SL (2010) The Malari leucogranite, Garhwal Himalaya, northern India:
Chemistry, age, and tectonic implications. Geol Soc Am Bull 122:1865–1876, doi:10.1130/b30153.1
Schmitz MD, Crowley JL (2014) Titanite petrochronology in the Fish Canyon Tuff. EOS–Trans Am Geophys
Union 94:V34A-08
Scott DJ, St-Onge MR (1995) Constraints on Pb closure temperature in titanite based on rocks from the Ungava
orogen, Canada: Implications for U–Pb geochronology and P–T–t path determinations. Geology 23:1123–1126
Geochem 83:199–230
Smith MP, Storey CD, Jeffries TE, Ryan C (2009) In Situ U–Pb and trace element analysis of accessory minerals in
the Kiruna District, Norrbotten, Sweden: New Constraints on the timing and origin of mineralization. J Petrol
50:2063–2094, doi:10.1093/petrology/egp069
Solgadi F, Sawyer EW (2008) Formation of igneous layering in granodiorite by gravity flow: a field, microstructure
and geochemical study of the Tuolumne Intrusive Suite at Sawmill Canyon, California. J Petrol 49:2009–
2042, doi:10.1093/petrology/egn056
Spear FS (2010) Monazite–allanite phase relations in metapelites. Chem Geol 279:55–62
Petrol 37:733–765
Spencer KJ, Hacker BR, Kylander-Clark ARC, Andersen TB, Cottle JM, Stearns MA, Poletti JE, Seward GGE
(2013) Campaign-style titanite U–Pb dating by laser-ablation ICP: Implications for crustal flow, phase
transformations and titanite closure. Chem Geol 341:84–101
Stearns MA, Cottle JM, Hacker BR, Kylander-Clark ARC (2016) Extracting thermal histories from the near-rim
zoning in titanite using coupled U–Pb and trace-element depth profiles by single-shot laser-ablation split
stream (SS-LASS) ICP-MS. Chem Geol 422:13–24, doi:10.1016/j.chemgeo.2015.12.011
Terry MP, Robinson P, Hamilton MA, Jercinovic MJ (2000) Monazite geochronology of UHP and HP
metamorphism, deformation, and exhumation, Nordøyane, Western Gneiss Region, Norway. Am Mineral
85:1651–1664
Tiepolo M, Oberti R, Vannucci R (2002) Trace-element incorporation in titanite: constraints from experimentally
determined solid/liquid partition coefficients. Chem Geol 191:105–119, doi:10.1016/s0009–2541(02)00151–1
Tilton GR, Grünenfelder MH (1968) Sphene: uranium–lead ages. Science 159:1458–1460, doi:10.1126/
science.159.3822.1458
Tomkins HS, Pattison DRM (2007) Accessory phase petrogenesis in relation to major phase assemblages in pelites
from the Nelson contact aureole, southern British Columbia. J Metamorph Geol 25:401–421
Tropper P, Manning CE (2008) The current status of titanite–rutile thermobarometry in ultrahigh-pressure
metamorphic rocks: The influence of titanite activity models on phase equilibrium calculations. Chem Geol
254:123–132, doi:10.1016/j.chemgeo.2008.03.010
Tucker RD, Robinson P, Solli A, Gee DG, Thorsnes T, Krogh TE, Nordgulen O, Bickford ME (2004) Thrusting
and extension in the Scandian hinterland, Norway: New U–Pb ages and tectonostratigraphic evidence. Am J
Sci 304:477–532, doi:10.2475/ajs.304.6.477
Verts LA, Chamberlain KR, Frost CD (1996) U–Pb sphene dating of metamorphism: The importance of sphene
growth in the contact aureole of the Red Mountain pluton, Laramie Mountains, Wyoming. Contrib Mineral
Petrol 125:186–199, doi:10.1007/s004100050215
Walters JB (2016) Protracted thrusting followed by late rapid cooling of the Greater Himalayan Sequence,
Annapurna Himalaya, central Nepal: Insights from titanite petrochronology. MS Boise State University
Watson EB, Liang Y (1995) A simple model for sector zoning in slowly grown crystals; implications for growth
Wing BN, Ferry JM, Harrison TM (2003) Prograde destruction and formation of monazite and allanite during
145:228–250
Wones DR (1989) Significance of the assemblage titanite + magnetite + quartz in granitic rocks. Am Mineral
74:744–749
Wotzlaw J-F, Schaltegger U, Frick DA, Dungan MA, Gerdes A, Guenther D (2013) Tracking the evolution of
large-volume silicic magma reservoirs from assembly to supereruption. Geology 41:867–870, doi:10.1130/
g34366.1
Xirouchakis D, Lindsley DH (1998) Equilibria among titanite, hedenbergite, fayalite, quartz, ilmenite, and
magnetite: Experiments and internally consistent thermodynamic data for titanite. Am Mineral 83:712–725
Xirouchakis D, Lindsley DH, Andersen DJ (2001a) Assemblages with titanite (CaTiOSiO4), Ca-Mg-Fe olivine and
pyroxenes, Fe-Mg-Ti oxides, and quartz: Part I. Theory. Am Mineral 86:247–253
Xirouchakis D, Lindsley DH, Frost BR (2001b) Assemblages with titanite (CaTiOSiO4), Ca-Mg-Fe olivine and
pyroxenes, Fe-Mg-Ti oxides, and quartz: Part II. Application. Am Mineral 86:254–264
Zhang LS, Schärer U (1996) Inherited Pb components in magmatic titanite and their consequence for the
interpretation of U–Pb ages. Earth Planet Sci Lett 138:57-65, doi:10.1016/0012-821x(95)00237-7
Vol. 83 pp. 443–467, 2017 14
Petrology and Geochronology of Rutile

Thomas Zack
University of Gothenburg
PO Box 460
41430 Gothenburg
Sweden
zack@gvc.gu.se
Ellen Kooijman
Swedish Museum of Natural History
Box 50007
104 05 Stockholm
Sweden
ellen.kooijman@nrm.se

Rutile (TiO2) is an important accessory mineral that, when present, offers a rich source of
information about the rock units in which it is incorporated. It occurs in a variety of specific
microstructural settings, contains significant amounts of several trace elements and is one
of the classical minerals used for U–Pb age determination. Here, we focus on information
obtainable from rutile in its original textural context. We do not present an exhaustive review
on detrital rutile in clastic sediments, but note that an understanding of the petrochronology
of rutile in its source rocks will aid interpretation of data obtained from detrital rutile. For
further information on the important role of rutile in provenance studies, the reader is referred
to previous reviews (e.g., Zack et al. 2004b; Meinhold 2010; Triebold et al. 2012). Coarse
rutile is the only stable TiO2 polymorph under all crustal and upper mantle conditions, with
the exception of certain hydrothermal environments (Smith et al. 2009). As such, we will focus
on rutile rather than the polymorphs brookite, anatase and ultrahigh-pressure modifications.
In this chapter, we first review rutile occurrences, trace element geochemistry, and U–Pb
geochronology individually to illustrate the insights that can be gained from microstructures,
chemistry and ages. Then, in the spirit of petrochronology, we show the interpretational power
of combining these approaches, using the Ivrea Zone (Italy) as a case study. Finally, we suggest
some areas of future research that would improve petrochronologic research using rutile.
RUTILE OCCURRENCE
Rutile is a characteristic mineral in moderate- to high pressure metapelitic rocks, in high
pressure metamorphosed mafic rocks, and in sedimentary rocks (e.g., Force 1980; Frost 1991;
Zack et al. 2004b; Triebold et al. 2012). Rutile also occurs rarely in magmatic rocks, e.g.,
anorthosites, as well as in some hydrothermal systems. Coarse-grained rutile is notably absent
in many fresh quartzofeldspathic gneisses, peridotites, carbonates (with the exception of some
calc-silicates, see Ferry 2000), low-grade metamorphic rocks, volcanic rocks, and fresh granites

1943-2666/17/0083-0014$05.00 (online)
444 Zack & Kooijman
(with the exception of porphyry copper deposits and certain highly differentiated granites that
straddle the line to pegmatites or greisens) (Force 1980; Zack et al. 2004b; Carruzzo et al. 2006).
The presence of rutile depends on its stability in relation to other major Ti phases such
as titanite, ilmenite, biotite and to a lesser degree amphibole. Bulk rock compositions with
high Ca and Fe2+ contents stabilize titanite and ilmenite, respectively, over rutile, and in many
metapelitic rocks, rutile occurs only if biotite is absent, in low abundance or starting to break
down. Chemical reactions can be written to describe the stability of rutile in more quantitative
ways. In mafic systems, Frost (1991) highlights the reaction:
Anorthite + Quartz + 2 Ilmenite = Garnet (Gr1Alm2) + 2 Rutile (GRIPS) (1)

as a major reaction bounding rutile stability. This reaction was calibrated experimentally
by Bohlen and Liotta (1986). It is strongly pressure dependent, with rutile being the high
pressure phase. Several other reactions with ilmenite or titanite (Manning and Bohlen
1991) are feasible, all more or less pressure dependent. In an experimental study on typical
N-MORB, rutile is found only above 1.2 GPa (Ernst and Liu 1998; see Fig. 1). Equilibrium
phase diagrams of oxidized N-MORB in the system NCKFMASHTO show rutile stable
above 1.0 GPa (Diener and Powell 2012; see Fig. 1).
In high-Al metapelites, Frost (1991) emphasizes the reaction:
3 Ilmenite + Al2SiO5 + Quartz = Almandine + 3 Rutile (GRAIL) (2)

which was experimentally calibrated by Bohlen et al. (1983). Bohlen et al. (1983) also list
examples of presumably coexisting rutile and ilmenite occurring between 0.5 and 0.8 GPa in
metapelitic assemblages (see Fig. 1). In several medium-pressure metamorphic areas, e.g., the
Ivrea Zone, Italy, (Zingg 1980; see below), rutile is reported in metapelitic assemblages but is
largely absent in metamorphosed mafic rocks. This is consistent with the relationship shown
in Figure 1, where rutile appears to be stabilized at lower pressure in metapelitic systems
compared to metabasites, which can be partly explained by higher Fe contents of the latter.
In metamorphic terranes that were metamorphosed at P ≤ 0.5 GPa, rutile is typically very
rare and occurs only in restricted lithologies. For example, in amphibolite- to granulite-facies
rocks of Namaqualand (South Africa), a classic low-pressure granulite-facies terrane, ilmenite
is the dominant Ti-phase in felsic gneisses and garnet–cordierite metapelitic assemblages over
an area of more than 10,000 km2 (e.g., Waters 1986a; Willner 1995). Rutile has been reported
as the dominant Ti-phase only in rare MgAl-rich rocks (Waters 1986a; Moore and Waters
1990) and sillimanite-rich rocks (Willner 1995). The protoliths of both rock types are assumed
to be deeply weathered Mg-enriched soil horizons where Ca, Na and/or Fe have been leached
out, whereas Al and by inference Ti have been passively enriched (Moore and Waters 1990),
leaving a source composition suitable for abundant rutile over a wide P–T range. Examples
of similar rutile distribution are found in MgAl-rich rocks from several low-pressure terranes
such as the Reynolds Range, Australia (Vry and Baker 2006). Pure quartzites are equally
suitable for rutile preservation/formation in low pressure metamorphic areas as they can be
almost devoid of Fe and Ca, whereas Ti together with Zr can be relatively enriched as part of
the original heavy mineral suite in sandstones (Hubert 1962). A detailed description of rutile in
quartzites from the Reynolds Range is found in Rösel et al. (2014; see also Fig. 2I).
A locally important occurrence of rutile is certain metasomatic rock types, most
notably formed by Na-metasomatism. Engvik et al. (2014) describe widespread albitisation
in several areas of Scandinavia associated with large-scale fluid fluxes during high-grade
metamorphism. Here, most elements were leached out of the rock, leaving almost pure
albite rocks with a sometimes substantial proportion of refractory minerals, such as rutile
and apatite, of economic value. The cm-sized rutile mineral standard R10 from the Bamble
Sector, Norway (Luvizotto et al. 2009a) formed in such an environment.
Figure 1
Petrology and Geochronology of Rutile 445
30
WGC
25
20 rutile
P (in kbar)
15
(1)
Ivrea
10 (2)
PGD
5 Namaqua
ilmenite/titanite
0
600 650 700 750 800 850 900
T (in oC)
Figure 1. Pressure–temperature diagram showing stability fields for rutile and ilmenite/titanite depending
on host rock composition. Transition field in metapelitic rock compositions is marked in light grey. Dia-
monds are P–T conditions for metapelites with coexisting rutile and ilmenite, from Bohlen et al. (1983).
Transition field in metabasic rock compositions is marked in dark grey. Solid line (1) is the transition from
titanite/ilmenite to rutile in experiments from Ernst and Liu (1998). Solid line (2) is the transition from
titanite to rutile in pseudosection models from Diener and Powell (2012). Stippled lines mark the approxi-
mate fields for amphibolite-facies (lower left), granulite-facies (lower right) and eclogite facies (upper field)
for reference. Solid black outlines mark the P–T conditions for metamorphic areas mentioned in the text:
Namaqua, after Waters (1986b); Pikwitonei Granulite Domain (PGD), after Mezger et al. (1990); Ivrea, only
rutile-bearing units after Redler et al. (2012); Western Gneiss Complex (WGC), after Carswell et al. (2003).
MICROSTRUCTURES OF RUTILE IN METAMORPHIC ROCKS

Microstructural relationships between accessory minerals and surrounding phases in
thin section have been shown to be of fundamental importance in several studies (e.g., Engi
2017; Lanari and Engi 2017; Williams et al. 2017). Also, careful investigation of internal
chemical textures, like compositional zonation, is indispensable for accessory phases, as
illustrated particularly for zircon (e.g., Corfu et al. 2003). Unfortunately, only very limited
use has been made so far of these layers of information to better understand processes
related to rutile. The benefits of carefully describing microstructural relationships of rutile
are best demonstrated when combined with Zr-in-rutile thermometry (e.g., Zack et al. 2004a;
Luvizotto and Zack 2009; Kooijman et al. 2012; Ewing et al. 2013; Pape et al. 2016; Pauly
et al. 2016). In this context, it is unfortunate that many studies do not report occurrences
of rutile, but rather the presences of “oxides” or “opaque phases” (although rutile is not
opaque) or simply ignore the occurrence of mineral phases beyond the interest of the study.
The microstructural relationships of rutile and neighboring phases can be seen best
either in reflected light microscopy or in backscattered electron (BSE) imaging (see Fig. 2).
If rutile is light-colored (often corresponding to low Fe-contents), thin section images in
transmitted light microscopy may also be useful, although it is noted that using infrared
light microscopy dramatically enhances visibility of internal features within rutile (Cabral
et al. 2015). This section focuses on metamorphic processes, because rutile is found most
commonly in metamorphic rocks, and because textural information is particularly important
to unravel complex histories recorded by these rocks.
446 Zack & Kooijman
A B
Chl
Ilm
Rt
C D
Grt Rt Grt
Rt Rt
Bt
E F
Grt
Rt
Ilm
Rt
G H
Rt
Rt Ilm
Tnt
70 µm
Figure 2. A–C: prograde formation of rutile. A) Rutile and chlorite replacing ilmenite (Erzgebirge, Ger-
many; from Luvizotto et al. 2009b); B) Minute rutile (and ilmenite) aggregates growing around biotite
during initial stages of biotite breakdown (Ivrea Zone, Italy). [Used by permission of Elsevier Limited,
from Luvizotto and Zack (2009) Chemical Geology, Vol 261, Fig. 5b, p. 307]; C) Coarse rutile in garnet
rims and in matrix from advanced to terminal biotite breakdown (Ivrea Zone, Italy). [Used by permission
of Elsevier Limited, from Luvizotto and Zack (2009) Chemical Geology, Vol 261, Fig. 6, p. 307]. D–E:
retrograde formation of rutile. D) Rutile crystallizing out of melt together with euhedral garnet and biotite
(Itaucu Complex, Brazil; unpublished); E) Rutile exsolution needles in garnet (xenolith of mafic granulite
from Sisimiut aillikite; described in Smit et al. 2016, photo courtesy of M. Smit). F–G: replacement of
rutile. F) Detrital rutile replaced by ilmenite during prograde metamorphism (Erzgebirge, Germany; from
Luvizotto et al. 2009b); G) Rutile replaced by titanite during retrograde metamorphism (garnet amphibolite
from Catalina Island; photo courtesy of A. Cruz-Uribe). H–I: Exsolution lamellae. H) Ilmenite exsolution
lamellae in rutile, with beginning corona texture (crustal xenolith from the Pamir, Tajikistan, unpublished).
Mineral abbreviations: Bt–biotite, Chl–chlorite, Coe–coesite, Grt–garnet, Ilm–ilmenite, rt–rutile, Sil–sil-
limanite, Tnt–titanite, Tur–tourmaline, Zrc–zircon.
I Zrc J
Rt
K L
M N
Figure 2 (cont’d). H–I: Exsolution lamellae I) Zircon exsolution lamellae in rutile (from UHT septa in Ivrea
Zone; from Ewing et al. 2013). J–N) other features. J) Irregularly recrystallized rutile in an amphibolite-facies
heavy mineral layer in quartzite next to unaffected round detrital zircon and tourmaline crystals (Reynolds
Range, Australia; From PhD study of D. Rösel 2014); K) beginning granoblastic–polygonal recrystallization
at the edge of a cm-sized rutile crystal with abundant fluid inclusions and subgrain boundaries (in chlorite-
rich blackwall zone; Tiburon Peninsula, California; unpublished); L) pristine coesite inclusion in rutile (Dora
Maira, Italy; from Hart et al. 2016); M) Relatively rare visible chemical zonation, here oscillatory zonation of
Nb and Ta-rich (light) and Nb and Ta-poor (dark) layers (Adelaide Orogen, Australia; described in Crowhurst
et al. 2002; unpublished picture); N) Zr WDS mapping of a rutile crystal illustrating asymmetric zoning rang-
ing from ca 700 to 1350 ppm, here higher Zr content (brighter) on the right (Pikwitonei Granulite Domain,
Canada). [Used by permission of John Wiley and Sons, from Kooijman, Smit, Mezger and Berndt (2012)
Journal of Metamorphic Geology, Vol 30, Fig. 5a, p. 405]. All figures are BSE images, except E (transmitted
light microphotograph), J (reflected light microphotograph) and N (Zr WDS map).
Mineral abbreviations: Bt–biotite, Chl–chlorite, Coe–coesite, Grt–garnet, Ilm–ilmenite, rt–rutile, Sil–sil-
limanite, Tnt–titanite, Tur–tourmaline, Zrc–zircon.
Different generations of rutile growth can be tied to different stages of a metamorphic

P–T path. During prograde metamorphism, rutile growth has been attributed to breakdown
of ilmenite, titanite and biotite. For ilmenite breakdown, pseudomorphs of rutile and chlorite
intergrowth are reported at the former site of ilmenite (Fig. 2A; Luvizotto et al. 2009b).
For the beginning of biotite breakdown, aggregates of small rutile are found around biotite
flakes (Fig. 2B; Luvizotto and Zack 2009). At more advanced stages of biotite breakdown,
large rutile is found in garnet rims and in the matrix, and biotite relicts are preserved only
in garnet cores (Fig. 2C; Luvizotto and Zack 2009). Rutile growth on the retrograde path
can be observed in UHT granulites from the Anapolis-Itaucu Complex, Brazil. Here, rutile
448 Zack & Kooijman
is found together with and included in small euhedral garnets in a biotite-rich melanosome
(Fig. 2D) that can clearly be related to peritectic garnet growth out of a crystallizing melt
during early stages of exhumation (see reaction 16 in Moraes et al. 2002). Rutile commonly
exsolves from other Ti-bearing phases during cooling, including quartz, biotite (often as
oriented needles called sagenite), pyroxenes and garnet (see Fig. 2E).
Microstructures tied to rutile replacement are widespread and well-documented. During
prograde metamorphism, pre-metamorphic detrital rutile in impure Fe-rich quartzites can be
(partially) replaced by ilmenite (Fig. 2F). Similar coronas can also form during retrograde
metamorphism via the GRIPS and GRAIL reactions in the down-pressure direction. As shown
in Figure 2G, titanite coronas can also form around rutile, in places exhibiting spectacular
resorption. The latter reaction facilitates reaction rate determination from clearly visible Nb
back-diffusion into remaining rutile (Lucassen et al. 2010; Cruz-Uribe et al. 2014) although
the applicability of this method has been questioned (Kohn and Penniston-Dorland 2017).
Chemical and textural modifications of rutile both during prograde and retrograde
metamorphism are critical for understanding the temporal and thermal evolution of rutile-
bearing rocks. Most common are sub-micron thin oriented exsolution lamellae of Fe-oxides
(ilmenite, hematite, magnetite) within rutile, sometimes forming larger aggregates around rutile
(Fig. 2H). Although many probably formed during cooling, changes in oxygen fugacity may
also play a role in some cases (e.g., Dill et al. 2007). For Zr-in-rutile thermometry the exsolution
of baddelyite and zircon (Fig. 2I) is of great interest, and has been documented in several cases
(see below). Poorly documented is the potential of rutile to recrystallize. A spectacular example
has been described by Rösel (2014) in a preserved heavy mineral layer within amphibolite-facies
quartzites from the Reynolds Range. Here, the only abundant detrital accessory phases are zircon,
tourmaline and rutile, typical for a highly mature sediment source (ZTR index of Hubert 1962).
Pristine rounded detrital grains of tourmaline and zircon are preserved, whereas former detrital
rutile is found only as recrystallized rutile aggregates with irregular grain boundaries (Fig. 2J).
These textures show at least qualitatively that rutile has a greater tendency to recrystallize than
tourmaline and zircon in this type of environment. It appears that the influence of deformation
is not the dominant factor. For example, a partly recrystallized cm-sized rutile crystal is found
embedded in weak chlorite-dominated matrix in a blackwall surrounding a blueschist body
within the serpentinite mélange of Tiburon, California. The partly preserved early rutile,
showing signs of subgrain boundaries, has been recrystallized to a perfect example of a fine-
grained granoblastic-polygonal aggregate (Fig. 2K; unpublished). These internal structures are
only vaguely visible in high-contrast SEM images, and more refined techniques such as recently
calibrated EBSD imaging of rutile (Taylor et al. 2012) are recommended.
Finally, we emphasize the vast amount of information that may be stored within single rutile
crystals with respect to microstructures. In the past, inclusions in rutile have been completely
ignored. Only recently a large number of phases recording HP and UHP metamorphic conditions
have been documented using careful BSE imaging and EDS and Raman identification (Hart et
al. 2016), including coesite inclusions (Fig. 2L). In infrared light microscopy, rutile is largely
transparent enabling detailed investigation of rutile inclusions in 3D. This has been utilized by
Cabral et al. (2015) to investigate fluid inclusions in rutile related to a hydrothermal gold deposit.
Internal chemical zonation is of fundamental importance in understanding accessory minerals
(see e.g., Corfu et al. 2003; Engi 2017; Kohn 2017; Lanari and Engi 2017; Rubatto 2017).
Unfortunately, in contrast to zircon, rutile is not sensitive to cathodoluminescence (CL) imaging.
Also, BSE imaging is generally not able to resolve chemical contrasts, as rutile is always close to
end-member composition. Only in extreme examples where wt% differences of heavy elements
such as Nb, Ta, W and Sb occur is BSE imaging able to resolve these (e.g., Smith and Perseil
1997; Rice et al. 1998; Fig. 2M). In the current absence of a simple imaging technique, chemical
heterogeneities within rutile crystals mostly have been made visible by time-consuming WDS
mapping, as shown for a Zr mapping in Figure 2N.
ZIRCONIUM-IN-RUTILE THERMOMETRY
AND RUTILE–QUARTZ OXYGEN ISOTOPE THERMOMETRY
Rutile provides the opportunity to obtain precise temperature information for a large
variety of rocks. There are several reasons for making such an optimistic statement, in
particular: (1) two independent geothermometers have been experimentally calibrated that
are applicable for moderate to high temperatures (Zr-in-rutile) as well as for moderate to low
temperatures (rutile–quartz oxygen isotope exchange) and (2) rutile occurs in many mineral
assemblages that are otherwise inappropriate for temperature estimations (e.g., hydrothermal
quartz veins, quartzites). The Zr incorporation in rutile is based on the simple relationship:
ZrO2 (in rutile) + SiO2 (quartz/coesite) = ZrSiO4 (zircon) (3)
where the two other phases (quartz/coesite and zircon) represent saturating near-endmember
phases when present. It is important that rutile coexists with quartz/coesite and zircon in
order to apply Zr-in-rutile thermometry as outlined below. Alternatively, the activities of
SiO2 and ZrSiO4 need to be constrained (Zack et al. 2004a). Rutile Zr concentrations are
remarkably temperature-sensitive over the range of naturally occurring values from a few
10s to 10000 ppm (Zack et al. 2004a). This relationship has been experimentally calibrated
by Watson et al. (2006) at 1 GPa and a significant pressure effect has been calibrated by
Tomkins et al. (2007). Both experimental data sets are incorporated in the equation:
83.9 + 4.10 P
=T ( °C ) − 273 (4)
0.1428 − R ln ( Zr in ppm )
(Tomkins et al. 2007; see Fig. 3), which currently defines the best calibration in the a-quartz
stability field. Other expressions apply to the b-quartz and coesite stability fields. Here, P is
pressure given in GPa, and R is the gas constant (0.0083144 kJ K-1). This equation technically
is calibrated for α-quartz, but is less than 5 °C divergent from the calibration for b-quartz
(Tomkins et al. 2007). Temperature determinations are remarkably consistent with other
geothermometers at medium temperature (ca. 500–750 °C), especially in eclogites (e.g., Zack
and Luvizotto 2006; Miller et al. 2007). However, both at higher and lower temperatures,
results are more difficult to interpret. In HT and UHT granulites, rutile may show a large
variation in Zr concentration, even within one thin section (e.g., Kooijman et al. 2012).
Zirconium concentrations are typically homogeneous within the core of a given grain, but
strongly different between grains. In some cases, highest Zr concentrations were observed
for rutile grains included within garnet and orthopyroxene (Zack et al. 2004a); in other cases
such systematics do not occur (e.g., Luvizotto and Zack 2009; Ewing et al. 2013). In one case,
increasing Zr contents of rutile inclusions in garnet cores to garnet rims are consistent with a
prograde heating path in UHT granulites from Antarctica (Pauly et al. 2016). In some cases,
pronounced intra-grain zoning, including asymmetric zoning, has been observed (Kooijman et
al. 2012). Additionally, decreasing Zr concentrations occur towards grain boundary contacts to
zircon (Kooijman et al. 2012; Pape et al. 2016). Different mechanisms have been proposed to
explain these features. Most of these models involve a combination of Zr diffusion within rutile
as experimentally calibrated by Cherniak et al. (2007; see Fig. 4), and grain boundary diffusion
of Zr. The latter is difficult to quantify and is possibly strongly dependent on the presence and
nature of a grain-boundary fluid (Luvizotto and Zack 2009) or on the attachment/detachment
of ions on buffering phases (Kohn et al. 2016). In consequence, the intra-grain differences in
Zr concentration, typically up to values of 1000 ppm Zr at the contact to zircon, have been
interpreted to correspond to a closure temperature for Zr diffusion in rutile, corresponding to
ca. 700 °C for the given Zr concentration (Luvizotto and Zack 2009; Kooijman et al. 2012). The
large inter-grain variation can be explained through different scenarios that are not mutually
exclusive: (1) rutile formed at different times during P–T histories (Kooijman et al. 2012;
Pape et al. 2016; Pauly et al. 2016); (2) differences in the rates of grain-boundary diffusion
450 Zack & Kooijman Figure 3
T (in oC) Figure 3. Zr content in rutile
750 500 300 as a function of temperature.
5 Solid lines are calculated from
experiments of Tomkins et al.
(2007; Eqn. 4). The influence
of pressure is shown by dif-
4 ferent lines with pressure in
GPa. Open symbols represent
log (Zr, in ppm)
results for 24 metamorphic

3 areas from studies of Zack
et al. (2004a), Watson et al.
(2006) and Zack and Luvi-
zotto (2006), with squares
2
from ca 1 GPa, diamonds from
ca 2 GPa and triangles from
ca 3 GPa. Black solid circles
1 are results for rutilated quartz
veins from low pressures (Shu-
3 2 1 0.0001 laker et al. 2015) that seem to
0 indicate a different tempera-
7 9 11 13 15 17 19 ture dependence at low tem-
peratures (stippled line).
104/T (in K)
(Luvizotto and Zack 2009; drawing on the concept of a fluid diffusion-controlled regime of
Dohmen and Chakraborty 2003; see also Kohn et al. 2016), (3) distance of a given rutile
crystal to the nearest zircon (Luvizotto and Zack 2009) and (4) potential recrystallization of
some rutile crystals (Ewing et al. 2013). A robust distinction between the exact mechanisms
in different environments has not yet been laid out. Nevertheless, there is general consensus
that temperatures derived from the maximum Zr content in a given rutile population is a good
indication for the minimum peak metamorphic temperature of a sample (e.g., Luvizotto and
Zack 2009; Kooijman et al. 2012; Ewing et al. 2013; Pape et al. 2016).
A common property of rutile from HT and UHT granulites is the occurrence of baddelyite
and/or zircon exsolution lamellae (Fig. 2I). These have been interpreted to be produced during
cooling caused by the decreasing Zr solubility of rutile at lower temperature (Kooijman et al.
2012; Ewing et al. 2013). Formation of zircon lamellae within rutile is possible by transporting
Si from the surrounding matrix, for example at sufficient Si diffusion rates within rutile,
speculated to be a rate-limiting step (Taylor-Jones and Powell 2015). However, the presence
of fluids along fractures in rutile may enable sufficient Si supply (Pape et al., 2016). Rutile
may contain sufficient Si to form zircon without requiring Si in-diffusion (Kohn et al. 2016),
although it is currently unconstrained how much Si typically occurs in natural rutile. In the
absence of sufficient Si supply, baddelyite exsolution can occur (Kooijman et al. 2012). The
formation of zircon and baddeleyite may depend on factors other than chemical transport such
as nucleation (Kohn et al. 2016; Kohn and Penniston-Dorland 2017). Reintegrating the Zr
content in rutile before exsolution has been shown to give good estimates of minimum peak
metamorphic temperature of a sample (Ewing et al. 2013; Pape et al. 2016; Smit et al. 2016)
although no consensus exists for the ideal quantification method. For UHT samples from the
Ivrea Zone (see below), significantly lower temperatures were calculated when reintegration
was performed using a 60-µm defocused beam EPMA measurement (940–1060 °C; Pape et
al. 2016) and a defocused spot LA-ICP-MS measurement (1000–1020 °C; Ewing et al. 2013)
compared to reintegration based on grayscale BSE imaging (1085–1194 °C; Pape et al. 2016).
In any case, Zr-in-rutile temperatures often provide robust estimates on peak metamorphic
conditions for UHT granulites (e.g., Zack et al. 2004a; Harley 2008; Ewing et al. 2013) where
traditional thermobarometers (e.g., Fe–Mg exchange) are re-equilibrated.
Different challenges performing Zr-in-rutile thermometry exist at low temperature

(< 500 °C): (1) natural conditions are well outside the calibrated range (650–1300 °C; Watson et
al. 2006; Tomkins et al. 2007); (2) intra-grain Zr diffusion in rutile becomes very slow (Cherniak
et al., 2007); (3) expected Zr concentrations become increasingly challenging to measure
precisely by some techniques such as electron probe micro analysis (< 30 ppm below 500 °C);
and (4) Zr solubility in any grain boundary fluid will also be substantially reduced, making it
increasingly difficult for equilibrium to be attained among rutile, quartz and zircon grains that are
not in direct contact. Unfortunately, the latter parameter has not been experimentally quantified.
An alternative that avoids these problems is provided by the rutile–quartz oxygen isotope
thermometer. Experimentally calibrated down to 300 °C (Matthews et al. 1979), it has the
advantage that oxygen diffusion in rutile is ca four orders of magnitude faster than Zr in
rutile (see Fig. 4) and touching grain boundaries are easy to find. An attractive bonus is that
the rutile–quartz oxygen isotope thermometer is, next to magnetite–quartz, the second most
temperature-sensitive oxygen isotope-exchange thermometer (e.g., Matthews and Schliestedt
1984), potentially distinguishing temperature differences to better than ±50 °C (Shulaker et
al. 2015). In quartz-rich and rutile-poor lithologies (in particular quartzites) effective closure
temperatures are defined by O diffusion in rutile, which is rarely below 600 °C even for
slow cooling and small grain sizes (see Valley 2001). Traditionally, the rutile–quartz oxygen
isotope thermometer has been used successfully for more than 30 years (see examples in
Matthews et al. 1979 and Agrinier 1991). Surprisingly, it is only recently that it has received
renewed attention. Shulaker et al. (2015) calibrated oxygen isotope measurements of rutile by
HR-SIMS, and applied the new protocol to ‘rutilated’ quartz from alpine clefts from a range
of different metamorphic settings. The oxygen isotope exchange temperatures derived from
rutile–quartz pairs span from 310 to 540 °C, consistent with the range of regional metamorphic
conditions in that area. Interestingly, Zr-in-rutile temperatures correlate strongly with oxygen
isotope temperature, but do not follow the temperatures expected from the Tomkins et al.
(2007) calibration (see Fig. 3). An explanation cannot be given so far, but a misfit of the
Zr-in-rutile thermometer with other temperature estimates is apparent in several other low-T
metamorphic and hydrothermal examples (Cabral et al. 2015). Shulaker et al. (2015) caution
against the use of the Zr-in-rutile thermometry at low-T environments and recommend using
the rutile–quartz oxygen isotope thermometer instead. Figure 4
T (in ºC)
1200 1000 800 700
-15
Nb He
-17 O Figure 4. Arrhenius diagram illustrating

seven orders of magnitude different diffu-
Log D (m2sec-1)
sion systematics in rutile. Solid lines mark

the temperature range of experiments for
each system; stippled line for He is ex-
-19 trapolated from lower temperature range.
Data source: Zr-Cherniak et al. (2007);
Pb Pb-Cherniak (2000); Nb-Marschall et al.
(2013); O-Moore et al. (1998); He-Cher-
-21 niak and Watson (2011).
Zr
-23
6 7 8 9 10 11
104/T (in K)
452 Zack & Kooijman
NIOBIUM AND Cr DISTRIBUTION AS SOURCE ROCK INDICATORS

Niobium and Cr in detrital rutile have been used extensively to distinguish different
source rocks in provenance studies. The concept is based on the observation that the bulk Nb
and Ti contents in many rocks are predominantly concentrated in rutile when rutile is stable;
consequently, Nb concentrations in rutile reflect Nb/Ti ratios in host rock, which vary widely
(see Zack et al. 2002, 2004b). Several attempts have been made to distinguish, geochemically,
rutile that originates from metapelites and metabasites, which are the main source lithologies
for detrital rutile (e.g., Triebold et al. 2012). All such attempts are entirely empirical, have a
pragmatic approach and are not meant to be based on first principles. Rather these two rock
types reflect endmembers on a compositional continuum from mafic to felsic and aluminous
rocks. Furthermore, Nb and Ti are not always controlled only by rutile, and Nb/Ti fractionation
can occur between rutile and biotite (Luvizotto and Zack 2009) and rutile and titanite (Lucassen
et al. 2010; Cruz-Uribe et al. 2014).
Despite all those shortcomings, discrimination diagrams based on Nb and Cr for a large
suite of detrital rutile grains provide a statistical perspective on the distribution of other trace
elements derived from different source rocks. We have selected four studies on trace element
distribution of detrital rutile that report Nb and Cr (in addition to several other trace elements,
including U): Jurassic to recent sediments from the North Sea (Morton and Chenery 2009),
Paleozoic sediments from Germany (Rösel 2014) and Turkey (Okay et al. 2011) as well as
Mesoproterozoic greenschist facies quartzites from Australia (Rösel et al. 2014). Altogether
1170 analyses are considered (excluding only 55 rutiles with Ce and/or Y with > 10 ppm from
the North Sea data set, potentially mixed analyses including phases like zircon, monazite, etc).
Figure 5
About 19% of detrital rutile grains fall in the metabasite field and 81% in the metapelite field
based on the latest discrimination diagram (Fig. 5; Triebold et al. 2012).
10000
Metabasite
field
1000
Cr (in ppm)
100
Metapelite
field
CC (R&G14)
10
10 100 1000 10000 100000
Nb (in ppm)
Figure 5. Log Cr versus log Nb concentrations for 1170 detrital rutile analyses. Symbols correspond to
four different studies covering a wide range of depositional age and geographic position: open triangles-
North Sea sediments, Jurassic to Recent (Morton and Chenery 2009, n = 463); solid circles—Carboniferous
sediments from Turkey (Okay et al. 2011, n = 105); open circles—Ordovician sediments from Germany
(Rösel 2014, n = 426); solid triangles—Mesoproterozoic sediments from Reynolds Range, Australia (Rösel
et al. 2014, n = 186). Metabasite versus metapelite fields from Triebold et al. (2012). Arrow represents Nb
content of a hypothetical rutile from average continental crust (CC) from Rudnick and Gao (2014).
This compilation of more than 1100 detrital rutile grains illustrates how a clear distinction
between those two rock types is not possible (see for example Meyer et al. 2011, Kooijman
et al. 2012). Overall, there are not two separate fields: only one main cluster exists within a
diffuse cloud of possible Nb and Cr concentrations. Still, it is interesting that the median of
all Nb values of 1790 ppm closely follows the expected hypothetical Nb content in rutile of
1875 ppm calculated from average continental crust (0.64 wt% TiO2 and 12 ppm Nb; Rudnick
and Gao 2014), assuming all Ti and Nb are concentrated in rutile (see Fig. 5). On the one hand
it supports our assumption that detrital rutile chosen for this compilation is representative of
normal crustal sections. On the other hand it hints to the exciting prospect that large numbers
of detrital rutile of different ages may provide direct insight into the Nb/Ti fractionation in
the crust throughout Earth’s history. The oldest rutile grains analysed for these trace elements
(from the Mesoproterozoic Reynolds Range) have the highest median Nb/Ti ratio, which
nevertheless is based only on a small population and may not be representative.
URANIUM–LEAD GEOCHRONOLOGY
Uranium, Th and common Pb distribution in rutile
To retrieve U–Pb age information from rutile, three chemical components need to be
considered: U, Th and Pb. Unlike zircon, the U contents of some rutile crystals are too low
(< 0.1 ppm, see Zack et al. 2011) to be datable. Single spot 206Pb/238U 1σ age errors of < 6% for
the lowest-U rutile are reported down to 1.0 ppm (Li et al. 2011; Topuz et al. 2013). On the other
extreme, U contents of > 100 ppm have been reported from several locations (e.g., Mezger et al.
1989). A predictive model for targeting high-U rutile has not been presented yet, and several
factors (e.g., whole rock U content, coexisting accessory phases, temperature, oxygen fugacity)
seem to be important parameters. In the absence of an encompassing predictive model, we exploit
the detrital rutile data set presented above to survey U–Th–Pb systematics of rutile in general.
Figure 6A shows that for detrital rutile classified as “metapelitic”, only 27% contains
< 4 ppm U. In contrast, for “metabasic” rutile 60% has U contents of < 4 ppm (Fig. 6B). This
confirms the common knowledge that metapelitic rocks are more likely to contain datable rutile
than coexisting metabasic rocks. In the absence of choice, it is nevertheless possible to date rutile
in some metabasic rocks (e.g., tectonic sliver of blueschist or eclogite). However, it requires a
thorough search for high-U rutile (e.g., Li et al. 2011). Additionally, rutile with > 100 ppm U is
rare, but it can be discovered through a systematic search of a large pool of lithologies.
An advantageous property of most rutile is the absence of significant amounts of Th. This
property, and its corresponding lack of 208Pb ingrowth, facilitates common Pb correction via 208Pb
measurement (Zack et al. 2011). Measurement of 208Pb instead of 204Pb is favorable, because it is
ca. 40 times more abundant and is not compromised by 204Hg interference—a common problem
in laser ablation ICP-MS (Kylander-Clark 2017). As can be seen in Figure 6C, the majority of
analyzed detrital rutile (52%) has Th contents < 0.5 ppm. This can be explained by the large size
of Th4+ compared to the much smaller Ti4+ (Zack et al. 2002). However, not all rutile is low in
Th, and exceptions with Th contents of up to 35 ppm are reported from diagenetically grown
rutile from Oman (Dunkl and von Eynatten 2009). This is also visible in the detrital data set
in Figure 6C, where 3% of all analyzed grains had > 10 ppm Th. Such cases compromise the
common 208Pb correction, although we are not able to fully evaluate the data set used, and part of
the 3% may be based on mixed analyses. Nevertheless, we strongly recommend monitoring Th
contents when using the common 208Pb correction method during U–Pb rutile dating.
It is a widespread misconception that rutile can be rich in common Pb, as most rutile has
common Pb contents < 0.2 ppm (Zack et al 2011). In many cases it is the low U content that
leads to an unfavorably high ratio of common to radiogenic Pb. To a certain degree this can and
should be corrected by monitoring 206Pb/208Pb (Zack et al. 2011). Down to 206Pb/208Pb values
of ca 10, the common Pb correction does not lead to a severe loss in age precision.
454 Zack & Kooijman
Figure 6
27% A
”metapelitic” rutile
n=952
Number
3%
U (in ppm)
60% B
”metabasite” rutile
n=218
Number
0%
U (in ppm) Figure 6

51%
C
all rutile
n=649
Number
3%
Th (in ppm)
Figure 6. Histograms for U and Th concentrations of same detrital rutile grains as listed in Figure 5A)
Uranium concentrations for “metapelitic” rutile; B) Uranium concentrations for “metabasic” rutile; C)
Thorium concentrations for all rutile where data are available.
Analytics
Uranium-Lead geochronology of rutile has been applied for decades to constrain
metamorphic histories (e.g., Ludwig and Cooper 1984; Schärer et al. 1986; Corfu and Muir
1989; Mezger et al. 1989), to date hydrothermal alteration processes (e.g., Richards et al. 1988),
and eruption of xenoliths (Davis et al. 1997). These pioneering studies applied isotope-dilution
thermal ionization mass spectrometry (ID-TIMS) analysis of grain separates. This method has
the advantage of high precision, but has the potential complication of providing inaccurate
results for grains having more than one age component or grains having U/Pb-bearing mineral
inclusions such as zircon. More recently, high spatial resolution techniques have been developed
for rutile geochronology including laser ablation inductively coupled plasma mass spectrometry
(LA-ICP-MS; e.g., Vry and Baker 2006; Kooijman et al. 2010; Zack et al. 2011; Kylander-Clark
2017) and secondary ion mass spectrometry (SIMS; e.g., Clark et al. 2000; Li et al. 2011; Taylor
et al. 2012; Schmitt and Zack 2012; Ewing et al. 2015; Schmitt and Vasquez 2017). These in
situ techniques enable accurate dating of heterogeneous rutile and are largely complementary;
LA-ICP-MS allows rapid analysis of large populations required for detrital provenance studies,
whereas SIMS has higher spatial resolution and is less destructive. Both techniques are not
without challenges; depth profiling by LA-ICP-MS is at a stage where only integrating several
profiles yields a level of precision where age zoning becomes resolvable (Smye and Stockli
2011), whereas U–Pb SIMS analysis may be compromised by strong crystal orientation effects
(Li et al. 2011; Taylor et al. 2012; Schmitt and Zack 2012). Nevertheless, in situ rutile dating has
been successfully applied in an increasing number of studies covering most of the geological
time-scale, from as old as 3.3 Ga (Upadhyay et al. 2014) to as young as only 9 Ma (Smit et al.
2014). Further improvement of in situ techniques will be key to advancing our ability to resolve
zoning in heterogeneous grains and enable dating of samples with low radiogenic Pb contents.
The data quality of in situ techniques such as LA-ICP-MS and SIMS largely depends on
the availability of suitable matrix-matched calibration standards. Such standards should: 1) be
homogeneous in age; 2) have a high U concentration, 3) contain little or no common Pb, and
4) be commonly available. The number of natural rutile standards has significantly increased in
the last decade with various groups developing material for large-scale distribution (Luvizotto
et al. 2009a; Zack et al. 2011; Bracciali et al. 2013). Currently available standards cover a range
of ages from 489.5 Ma (R19; Zack et al. 2011) to 2625 Ma (WHQ; Taylor et al. 2012) with
the most commonly used standards being R10 and R10b (1090 Ma; Luvizotto et al. 2009a;
Schmitt and Zack 2012). Bracciali et al. (2013) show that these standards can sometimes be
heterogeneous or high in common Pb, which may be a problem if this is not corrected for. They
present two new potential standards of ~1.8 Ga (Sugluk-4 and PCA-S207), which show good
reproducibility (1–2% RSD). Although such precision is only slightly larger than the long-term
reproducibility of most zircon standards, there is an ongoing community effort to characterize
more abundant and more homogeneous reference materials (e.g., E. Axelsson, pers. comm.).
Uranium–lead systematics in rutile
A reliable interpretation of U–Pb ages of any mineral requires knowledge of the closure
temperature (Tc) for Pb diffusion. The mathematical expression to determine Tc is defined by
Dodson (1973) as:
Ea / R
Tc =
 
( 2
 ARTc D0 / a
ln 
2
) (5)
dT 
 Ea 
 dt 
in which Ea is the activation energy, R is the gas constant, A is a geometric factor, D0 is the
frequency factor, i.e., the diffusion coefficient at infinite temperature, a is the effective diffusion
radius, and dT/dt is the cooling rate. This expression gives a weighted average Tc for a whole
grain. Important assumptions of the model are: (1) the matrix around the crystal acts as a
homogeneous infinite reservoir for the diffusing species, (2) the concentration of the species
at the onset of cooling was homogeneous, (3) no overgrowth, resorption or recrystallization
456 Zack & Kooijman
occurred after the onset of cooling, (4) there was sufficient diffusive loss during cooling to
modify the concentration of the species in the core of the grain (i.e., cooling was slow), and
(5) temperature is inversely proportional to time during cooling. The first estimate of Tc for
Pb in rutile was made by Mezger et al. (1989), who applied ID-TIMS analysis to rutile grains
of 90–210 µm of the Proterozoic Adirondacks and compared the resulting ages by assuming
closure temperatures of other dated minerals in the same terrane. Their estimates were between
380–420 °C, one of the lowest closure temperatures for Pb in accessory minerals. Vry and Baker
(2006) recalculated this value to 500–540 °C using higher Tc estimates for Pb in monazite,
Pb in titanite, Ar in hornblende, and Ar in biotite in the same geological setting. The revised
Tc estimate is more consistent with the diffusion coefficients determined in experiments by
Cherniak (2000; Ea = 250 ± 12 kJ mol-1 and D0 = 3.92 × 10–10 for synthetic rutile, Ea = 242 ± 10 kJ
mol-1 and D0 = 1.55 × 10–10 for natural rutile; Fig. 4). These coefficients predict a whole-grain Tc
of ~620 °C for a spherical rutile of 200 µm at an average cooling rate of 1 °C/Ma.
The diffusion rates for Pb in rutile imply that diffusive age gradients will develop in
rutile during cooling from lower crustal conditions. The description of Tc as a function of
the position within the grain [Tc(x)] was mathematically defined before it was viable to
demonstrate analytically (Dodson 1986). When age gradients are detectable, in principle it is
possible to obtain a continuous thermal history by forward modelling T–t histories that can
match the observed age gradient (Grove and Harrison 1999; Harrison et al. 2005). With the
advancements of analytical techniques it has become possible to resolve U/Pb zoning or age
profiles in rutile (Kooijman et al. 2010, 2015; Smye and Stockli 2014). These studies present
Figure 7. Age profiles in rutile grains. A) 207Pb/206Pb age profiles for three rutile grains from sample 462d
from the Pikwitonei Granulite Domain, Canada. Dashed lines represent error function best fit to the data
and R2 values indicate goodness of fit (Kooijman et al. 2010). B) error-weighted 206Pb/238U age depth-
profiles calculated for sample IVZR19 (n = 10) from the Ivrea Zone, Southern Alps, Italy. Dashed line
represents power law best fit to the data (Smye and Stockli 2014).
resolvable diffusion profiles in very different geological settings (Fig. 7). As such, each of
these profiles is of a different length scale and amplitude, but all adhere closely to Fickian
diffusion behavior allowing modeling of cooling rates.
Extremely slow cooling of the Archaean lower crust in the Pikwitonei Granulite Domain,
Canada resulted in age profiles in rutile that show a core-to-rim age range of up to 200 Ma
in grains of up to 270 µm (Fig. 7; Kooijman et al. 2010). Their study involved error-function
based diffusion modelling of Pb in rutile using the diffusion coefficients of Cherniak (2000).
They show that the Tc as a function of position in the grain ranged from 640 °C in the core
to 490 °C in the rim of grains. The latter temperature is currently the lowest Tc estimate from
an empirical study. Although the spatial resolution of the LA-ICP-MS analyses was slightly
limiting, the cooling rate modeling results fitted well with previous estimates of average
cooling rate based on comparison with other thermochronometers (Mezger et al. 1989). The
time-resolved model provided additional constraints by enabling construction of a continuous
cooling history from 2.2 °C/Ma at 640 °C to 0.4 °C/Ma at 490 °C (Kooijman et al. 2010).
The study by Smye and Stockli (2014) demonstrates age closure profiles in rutile at a higher
resolution and over a much smaller scale of ~20 µm using LA-ICP-MS depth profiling. Their
study targeted rutile grains from the Ivrea Zone where modeling of the age profiles constrained
rapid cooling from 180 Ma (Fig. 7; see below). High-resolution depth profiling permitted the
identification of a transient thermal event that so far remained undetected with traditional
spot analysis (Smye and Stockli 2014). In the well-studied UHP zone of the Western Gneiss
Complex, Norway, Kooijman et al. (2015) resolved Pb diffusion profiles in mm-size rutile
grains. The diffusion length in these grains is 400 µm, which could be resolved well by LA-
ICP-MS using a typical 30-µm spot size. Such diffusion profiles allow for reconstructing a
significant part of the cooling history of a metamorphic terrane (Dodson 1986; Kooijman
et al 2010). These studies show that U–Pb age profiles in rutile can occur, and be resolved,
on different scales, depending on the geological setting. This is in contrast to Villa (2016)
who stated that “diffusion profiles of isotope ratios in minerals used for geochronology are
absent”. Current analytical techniques allow for precise determination of such profiles in rutile
and, theoretically, permit forward modelling of the profiles as a function of T–t to obtain
a continuous thermal history for temperature intervals that are inaccessible to conventional
thermochronology (Smye and Stockli 2014; Kooijman et al. 2015).
Finally, it is important to note that the effective diffusion radius needs to be accurately defined
before invoking volume-diffusion controlled cooling ages. One of the main issues that could pose a
problem for interpreting U–Pb rutile ages is the common presence of exsolution lamellae (Fig. 2I)
and/or twinning planes. These features may reduce the effective diffusion radius and hence the
effective Tc of Pb at any spot in the grain if 1) they act as effective pathways for Pb movement,
or 2) if Pb partitions into the lamellae. If either one of these occurs, the effective Tc of rutile will
be controlled by the distance between the lamellae, and be significantly different from the whole-
grain Tc. If the lamellae form at temperatures above Tc, the U–Pb systematics will be affected. To
date, there are no studies that have quantified these potential effects. A further complication is that
exsolution lamellae a priori do not have to form above Tc, but can be a late feature.
Cooling vs formation ages
The intermediate closure temperature of Pb in rutile implies that it is possible to find
crystallization ages if rutile grains are large enough or if cooling is very fast. Diffusion-
induced resetting is as yet unresolvable in low-grade metamorphic rocks such as blueschists
and low-temperature eclogites (diffusion length scale < 1 µm) and formation ages can be
found in hydrothermal deposits, where, e.g., U–Pb ages for rutile and monazite are identical
(Shulaker et al. 2015). In higher-grade metamorphic rocks, the preservation of a formation age
in the core depends on a number of factors including peak-metamorphic temperature, cooling
rate, and grain size. Figure 8 shows the relation between these variables based on a diffusion
458 Zack & Kooijman
Figure 8. Relationship between effective diffusion length (a), average cooling rate (dT / dt) and peak meta-
morphic temperature for Pb in rutile based on diffusion modeling (Smit et al. 2013) using the diffusion
coefficients for natural rutile (Cherniak 2000). Shaded areas show peak metamorphic temperatures and
average cooling rates for the Pikwitonei Granulite Domain (PGD) in Canada (Mezger et al. 1989; Kooij-
man et al. 2010), the Ivrea Zone (IVZ) in the Southern Alps, Italy, and the Western Gneiss Complex (WGC)
in Norway (Carswell et al. 2003).
modeling approach by Smit et al. (2013) using the diffusion coefficients for natural rutile
(Cherniak 2000) for a range of geologically relevant conditions. Assumptions for this model
are the same as described earlier for the definitions of Dodson (1973, 1986).
The isotherms in Figure 8 indicate that higher temperatures and slower cooling will require
larger grains to preserve crystallization ages in grain cores. To illustrate the potential applicability
of this model, metamorphic terranes discussed in the text have been marked (Fig. 8). In the case
of the Pikwitonei Granulite Domain, which has peak T estimates of 750–800 °C and cooling
rates of 1–2 °C/Ma the minimum diffusion radius would be 0.5–1.5 mm. This means that a grain
size larger than 1–3 mm is required to find a crystallization age in the core. Although diffusion
profiles have been demonstrated in rutile from the Pikwitonei Granulite Domain (Kooijman et
al. 2010), the largest grains were 270 µm implying that the core ages must be far removed from
the original crystallization age. The Western Gneiss Complex in Norway underwent slightly
higher peak T conditions at 750–850 °C, which was followed by much faster cooling at rates
of 10–20 °C/Ma (e.g., Carswell et al. 2003; Kylander-Clark et al. 2008). This implies that a
diffusion length on the order of 0.2–1.0 mm could be expected in rutile grains from this terrane
(Fig. 8). Kooijman et al. (2015) resolved age profiles in rutile from this area with a diffusion
length of 400 μm in mm-sized grains, demonstrating that cooling in this terrane was fast enough
to allow for crystallization ages to be preserved in the cores of rutile grains.
In many cases in metamorphic systems rutile will record only cooling ages, but this still
yields a wealth of useful information about a different part of the rock’s history, for example
cooling and exhumation. This makes rutile ages very much complementary to zircon and
monazite ages in the same rock. Even in metamorphic rock samples where crystallization ages
still can be found, parts of rutile grains would be expected to record cooling ages, which can
also be exploited. In this regard, rutile is not likely to record pre-metamorphic information,
i.e., multiple generations, because rutile is not stable under most low- to medium-grade
metamorphic conditions and will either break down completely, or fully recrystallise (Zack et
al. 2004b). It therefore exhibits strikingly different behavior compared to zircon or monazite,
which commonly preserve multiple growth zones.
Comparison with U–Pb titanite ages

There is a significant body of literature where cooling paths of high grade metamorphic
terranes are derived using ages from several metamorphic phases and isotope systems. In
most cases, rutile ages are younger than U–Pb ages from titanite (e.g., Mezger et al. 1998;
Möller et al. 2000; Flowers et al. 2006), and similar to Ar–Ar ages from hornblende and
Rb–Sr ages from muscovite (e.g., Vry and Baker 2006). We will not discuss complexities
that may arise with Ar–Ar and Rb–Sr systems or with recrystallization behavior of micas
(see e.g., Villa 1998), but rather focus on a comparison of U–Pb rutile ages with U–Pb
titanite ages. Rutile and titanite are among a handful of phases where experimental diffusion
rates are available that can be directly related to thermochronology (Cherniak 1993, 2000).
One of the challenges in thermochronology is that experimental diffusion rates for Pb in
both phases are broadly similar, which contrasts with natural data.
One possibility is that the experimentally derived diffusion parameters for Pb in rutile
may be wrong, as advocated in several studies. For example, Schmitz and Bowring (2003)
suggest that experiments for rutile may be compromised by not considering the effect of water
and defect distribution in the crystal lattice and therefore prefer to adhere to the original Tc
estimate of ~400 °C by Mezger et al (1989). It has to be stated that this may be applicable to
titanite experiments as well, which Schmitz and Bowring do not question. To this end, several
studies show that titanite can record pre-metamorphic/magmatic U–Pb ages in areas heated up
beyond the closure temperature predicted (e.g., Pidgeon et al. 1996; Frost et al. 2000; Rubatto
and Hermann 2001; Spencer et al. 2013). Kohn and Penniston-Dorland (2017) therefore argue
that Pb diffusion in titanite must be much slower in nature than indicated experimentally.
Another possibility for the conflict between experiment and some studies on natural
samples can be fundamentally addressed by questioning the validity of volume diffusion
for the U–Pb system. Villa (1998, 2016) argues that in the presence of fluids, processes like
recrystallization, resorption and overgrowth can be much faster than volume diffusion and
should dominate. While we have shown above that clear examples for volume diffusion-
controlled age zonation profiles exist for rutile, the same is currently missing in the literature
for titanite, with Kohn and Corrie (2011) giving examples of U–Pb age and Zr-in-titanite
temperature relationships attributed to growth zoning and Stearns et al. (2016) documenting
U–Pb age and trace element variations due to recrystallization.
CASE STUDY: THE IVREA ZONE

The Ivrea Zone (also Ivrea-Verbano Zone) in the Southern Alps, Italy is arguably the
area best-studied in terms of rutile occurrence, geochemistry and U–Pb age behavior. It is a
classical area that exposes a near-complete section through all levels of the lower crust (Zingg
1980, 1983; Handy et al. 1999; Zingg et al. 1990) with clear connections to contemporaneous
upper crustal sections, like the metamorphic Serie de Laghi (also called Strona-Ceneri zone;
Zingg 1983) and sedimentary and volcanic cover rocks, e.g., in the Monte Nudo Basin,
the Lombardian Basin and the Canavese Zone (e.g., Beltrando et al. 2015). The geologic
importance of this area in combination with a wide geographic occurrence of assemblages
containing large rutile grains (up to 1 mm in diameter) makes the Ivrea Zone a testbed for new
methodologies applicable to rutile. In the following we focus on the unique advantages that
rutile petrochronology affords in unraveling the geologic history of the Ivrea Zone.
The geologic history of the Ivrea Zone spans more than 300 Ma, including regional
metamorphism up to granulite-facies during the Carboniferous, massive Permian magma
intrusions, Jurassic rifting and finally exhumation during the Alpine orogeny (Handy et al.
1999). The first clearly resolvable episode is the regional metamorphism at around 316 ± 3 Ma
460 Zack & Kooijman
Figure 9. Geologic map of the Ivrea Zone, modified after Ewing et al. (2015). Dark stippled unit- mostly am-
phibolite- and granulite-facies metapelites (called Kinzigite Formation); light stippled unit- Mafic Complex;
light unstippled unit- Paragneiss-bearing belt with so-called metapelite septa (dark unstippled); black unit-
mantle peridotite; white stippled lines are isograds (Mu–muscovite out; Kfs–K-feldspar in; Opx–orthopyrox-
ene in) from Zingg (1980). Large circles are coarse-grained rutile occurrences in metapelites described in Luvi-
zotto and Zack (2009), and Ewing et al. (2013). Small circles are additional rutile occurrences mapped in Zingg
(1980). The rutile-in isograd is inferred after Zingg (1980). [Used by permission of Springer, from Ewing,
Rubatto, Beltrando and Hermann (2015) Contributions to Mineralogy and Petrology, Vol 169, Fig. 1, p. 44]
(Ewing et al. 2013), which affected the entire Ivrea Zone, but to various peak P–T conditions.
It is best recorded in Val Strona di Omegna and Val d’Ossola (see Fig. 9), where a general
increase in metamorphic grade is visible from ca 650 oC / 0.5 GPa in the eastern, mid-crustal
section to > 900 oC / 1.1 GPa in the western, lowermost crustal section (Redler et al. 2012).
P–T estimates are based on pseudosections that include a Ti-bearing biotite model, which is
crucial as Ti stabilizes biotite to higher temperatures compared to models without Ti (see also
Tajčmanová et al. 2009). It is this area where one of the few published cases where a regional
rutile-in isograd has been mapped (Zingg 1980), observable in metapelitic assemblages. The
dominance of ilmenite as the major Ti-phase in metabasites (e.g., Kunz et al. 2014) broadly
corresponds to different minimum pressures required to stabilize rutile in those different
lithologies (Fig. 1). The rutile-in isograd is attributed to biotite breakdown, because the
products (garnet, K-feldspar and sillimanite) and pressures exceed the GRAIL equilibrium for
these rocks (Zingg 1980; Luvizotto and Zack 2009). The increasing metamorphic gradient in
the rutile-bearing assemblages in Val Strona di Omegna and Val d’Ossola is also recorded in
rutile in the form of an increase in maximum Zr contents in a given rutile population. Here,
the maximum Zr contents of rutile range from 2500 ppm Zr close to the rutile-in isograd, to
5000 ppm Zr in the highest grade rocks (Luvizotto and Zack 2009). The calculated temperatures
of 850 °C (at 0.8 GPa) and 950 °C (at 1.1 GPa), respectively, are in good agreement with Redler
et al. (2012), and are strong evidence that previous temperature estimates of this part of the
Ivrea Zone based on Fe–Mg exchange thermometers (e.g., a maximum of 860 °C by Henk et
al. 1997 for the highest grade section) are prone to diffusional resetting.
In Permian times (288 ± 4 Ma; Peressini et al. 2007), a massive layered intrusion, the
Mafic Complex (Fig. 9), ca. 10 km in thickness intruded the lower crust (e.g., Quick et al.
2003). Although no rutile is recorded in a wide range of intrusive rock types from norites up
to granodiorite (gabbro is the main rock type), spectacular rutile occurrences are nevertheless
found in metapelitic slivers (so-called septa; Sinigoi et al. 1995) ingested by mafic intrusive
units (Quick et al. 2003; Ewing et al. 2013). In these ≤ 100 m thick septa, highly restitic
lithologies contain biotite-free, garnet- and sillimanite-rich gneisses with abundant rutile
(Ewing et al. 2013), sometimes with additional corundum (Quick et al. 2003; Pape et al. 2016).
Having obviously been exposed to contact metamorphism by a gabbroic magma, those rutile
grains record some of the highest published Zr contents (up to 10000 ppm Zr, corresponding
to temperatures of ca. 1000 °C; Ewing et al. 2013). Other rutile grains show textbook-example
exsolution lamellae of zircon and in one case baddeleyite formed during cooling (Ewing et al.
2013; Pape et al. 2016). Reintegrating the expelled Zr in rutile, Ewing et al. (2013) and Pape
et al. (2016) concluded that peak metamorphic temperatures were even higher than 1000 °C.
Interestingly, Ti-in-zircon does not record such high temperatures (only up to 870 °C), likely
because it crystallized from in situ anatectic melts during cooling (Ewing et al. 2013).
Between intrusion of the Mafic Complex and the Jurassic rifting, it is uncertain what
magmatic and/or tectonic events affected the entire Ivrea Zone. In the northern end, gabbros
around the prominent mafic-ultramafic Finero body formed at 232 ± 3 Ma (Zanetti et al. 2013).
Whether sporadic zircon ages between 260 and 200 Ma throughout the entire Ivrea Zone (Vavra
et al. 1999; Ewing et al. 2013) are attributed to accompanying magmatic activities or due to local
fluid influx, are a matter of debate (see discussion in Ewing et al. 2013 and references therein).
In any case, a geochemical or age imprint on rutile has not yet been ascribed to these events.
Jurassic rifting is recorded in many parts of the Alps and beyond (e.g., Handy et al.
1999), leading to the opening of the Alpine Tethys Ocean. In the Ivrea Zone, the most
prominent feature recording this event is the Pogallo line (Fig. 9), an amphibolite-facies
shear zone partly juxtaposing the lower crustal Ivrea Zone against upper crustal Serie dei
Laghi units. Movement along the Pogallo line has been dated at 182.0 ± 1.6 Ma (Mulch et al.
2002). It appears that rutile may be the phase that provides the most coherent picture of these
thermal events during this time span. Several studies record U–Pb rutile ages between 160
and 200 Ma throughout the Ivrea Zone (Zack et al. 2011; Smye and Stockli 2014; Ewing et al.
2015). However, what makes rutile such a key phase here is the view that rutile crystallized
before the Jurassic (Ewing et al. 2015). In particular, Zr contents of > 4000 ppm in rutile
can originate only at temperatures acquired during Variscan granulite-facies conditions
462 Zack & Kooijman
and especially during the intrusion of the Permian Mafic Complex for rutile in the septae.
Therefore, U/Pb rutile ages have to be interpreted in the framework of Pb loss. Complications
from recrystallization and/or overgrowth can be excluded (Ewing et al. 2015).
As can be expected and as is shown in Figure 8, slow cooling granulite-facies areas like the
rutile-bearing parts of the Ivrea Zone do not preserve formation ages in rutile even in the largest
recorded grains of 1 mm. Therefore, widespread rutile core ages of 180 Ma (Zack et al. 2011;
Smye and Stockli 2014; Ewing et al. 2015) correspond to temperature conditions of ca. 600 °C.
However, Smye and Stockli (2014) were able to detect age zoning profiles in the outermost
15 microns of euhedral rutile crystals (Fig. 7), which they cannot interpret with a monotonous
cooling path, but rather with rapid cooling at ca. 180 Ma. In combination with Zr diffusion
profiles, the ages and cooling paths can be interpreted to originate after a short-lived transient
temperature spike up to ≥ 700 °C. Those results show the advantage of diffusion-controlled age
zoning profiles, where instead of T–t points from bulk mineral results, continuous sections of
a T–t evolution can be determined. In comparison with extensive previous age determinations
of multiple dating systems in the same section of the Ivrea Zone (Siegesmund et al. 2008), a
rift-related reheating event is apparently visible only from the new U–Pb rutile data set. We
stress that such a Jurassic reheating event has also been proposed previously (see e.g., Fig. 13 in
Siegesmund et al. 2008), but it became evident only using rutile petrochronology.
The Jurassic U/Pb rutile ages of ca 180 Ma are observable in all rutile-bearing sections
in the Ivrea Zone (Zack et al. 2011; Smye and Stockli 2014; Ewing et al. 2015). Ewing et al.
(2015) also describe a second age population of ca 160 Ma using SIMS, not found in nearby
locations studied by Zack et al. (2011) and Smye and Stockli (2014) who used LA-ICP-MS,
and ascribe this to a second reheating event at this time. Future studies should examine whether
one or several reheating events affected the Ivrea Zone and how this is recorded in the entire
rutile-bearing section, spanning a paleodepth of almost 4 km.
CONCLUDING REMARKS AND RECOMMENDATIONS

Rutile is an ideal recorder of Earth’s (in particular metamorphic) processes due to the
fact that it provides several layers of information regarding temperature and time. Although
petrologic information such as microstructures and trace element composition can be
interpreted independently from geochronologic information, it is the combination of those
two fields in the spirit of petrochronology, where progress will be made. This is particularly
relevant when considering U–Pb thermochronology of rutile. Knowledge of the Tc of Pb in
rutile is crucially important to accurately interpret the age. Caution is warranted when the
following textural features occur, because these may result in a higher or lower apparent Tc
than would be predicted by grain size:
• Crystals with resorbed grain shapes (e.g., Fig. 2G),
• Multicrystal aggregates (e.g., Figs. 2J and 2K) and
• Grains with twinning planes and/or exsolution lamellae (Figs. 22H and 2I).
In high-grade metamorphic rocks, specifically of granulite-facies grade, rutile is potentially
the most suitable mineral for studying the influence of intra- and intergrain diffusion during
cooling on thermochronological interpretation. If maximum temperatures and cooling rates
are known, it can be estimated what minimum grain sizes of rutile crystals are required to
resolve age zoning profiles consistent with Fickian diffusion behavior (Fig. 8). Following
Ewing et al. (2015), we recommend an active search for microstructural and chemical criteria
that demonstrate the absence of modification of crystal cores during cooling such as:
• Evidence for older textural relationships, e.g., contact with older minerals like zircon or
garnet (e.g., Fig. 2C) or original rounded grain shape of detritus in quartzites
• Preservation of inclusions (Fig. 2L)

• Chemical composition pointing towards an older history (e.g., Zr contents of > 1000 ppm)
The same criteria also apply for the search of U–Pb formation ages in rutile from lower
grade metamorphic areas, specifically in blueschists and eclogites, except that zoning profiles
may not be resolvable.
The element Zr, used for Zr-in-rutile thermometry, diffuses much more slowly than Pb
in rutile (Fig. 4) and, hence, can be linked to crystallization or recrystallization of rutile.
Depending on rock type and metamorphic history Zr may record prograde, peak or retrograde
conditions, and thus should be commonly decoupled from U–Pb chronology. Only when
formation ages are found in rutile can they be directly tied with temperature information from
Zr contents or, at temperatures below ca 500 °C, to O isotopes.
ACKNOWLEDGMENTS
We would like to express our gratitude towards Tanya Ewing and Julie Vry for their
very extensive and constructive reviews. Additional comments and suggestions by Matt
Kohn are gratefully acknowledged. Pierre Lanari is thanked for his editorial guidance and
patience. Tanya Ewing is also thanked for providing an editable version of Figure 9. We are
grateful to Alicia Cruz-Uribe, George Luvizotto, Delia Rösel and Matthijs Smit for providing
(partly unpublished) microphotographs for this article. Matthijs Smit is also thanked for
discussions about this review and help with some figures. They are also thanked, along
with Jasper Berndt, John Cottle, Helen Degeling, Istvan Dunkl, Tanya Ewing, Andreas
Kronz, Andrew Kylander-Clark, Craig Manning, Horst Marschall, Klaus Mezger, Andreas
Möller, Renato Moraes, Patty O’Brien, Roger Powell, Axel Schmitt, Andrew Smye, Danny
Stockli, Silke Triebold and Hilmar von Eynatten for stimulating discussions regarding
Rutilology throughout the years. TZ acknowledges funding from Svenska Vetenskapsradet
(VR E0582701) and previous rutile-related grants from the German Science Foundation.
REFERENCES
Agrinier P (1991) The natural calibration of O/16O geothermometers—application to the quartz–rutile mineral
18
pair. Chem Geol 91:49–64

Beltrando M, Stockli DF, Decarlis A, Manatschal G (2015) A crustal-scale view at rift localization along the fossil
Adriatic margin of the Alpine Tethys preserved in NW Italy. Tectonics 34:1927–1951
Bohlen SR, Liotta JJ (1986) A barometer for garnet amphibolites and garnet granulites. J Petrol 27:1025–1034
Bohlen SR, Wall VJ, Boettcher AL (1983) Experimental investigations and geological applications of equilibria in
the system FeO–TiO2–Al2O3–SiO2–H2O. Am Mineral 68:1049–1058
Bracciali L, Parrish RR, Horstwood MSA, Condon DJ, Najman Y (2013) U–Pb LA-(MC)-ICP-MS dating of rutile:
New reference materials and applications to sedimentary provenance. Chem Geol 347:82–101
Cabral AR, Rios FJ, de Oliveira LAR, de Abreu FR, Lehmann B, Zack T, Laufek F (2015) Fluid-inclusion
microthermometry and the Zr-in-rutile thermometer for hydrothermal rutile. Int J Earth Sci 104:513–519
Carruzzo S, Clarke DB, Pelrine KM, MacDonald MA (2006) Texture, composition, and origin of rutile in the South
Mountain Batholith, Nova Scotia. Can Mineral 44:715–730
Carswell DA, Brueckner HK, Cuthbert SJ, Mehta K, O’Brien PJ (2003) The timing of stabilisation and the exhumation
rate for ultra-high pressure rocks in the Western Gneiss Region of Norway. J Met Geol 21:601–612
Cherniak DJ (2000) Pb diffusion in rutile. Contrib Mineral Petrol 139:198–207
Cherniak DJ, Watson EB (2011) Helium diffusion in rutile and titanite, and consideration of the origin and
implications of diffusional anisotropy. Chem Geol 288:149–161
Cherniak DJ, Manchester J, Watson EB (2007) Zr and Hf diffusion in rutile. Earth Planet Sci Lett 261:267–279
Clark DJ, Hensen BJ, Kinny PD (2000) Geochronological constraints for a two-stage history of the Albany-Fraser
Orogen, Western Australia. Precam Res 102:155–183
464 Zack & Kooijman
Corfu F, Muir TL (1989) The Hemlo-Heron Bay greenstone belt and Hemlo Au-Mo deposit, Superior Province,
Ontario, Canada; 2. Timing of metamorphism, alteration and Au mineralization from titanite, rutile, and
monazite U–Pb geochronology. Chem Geol 79:201–223
Cruz-Uribe AM, Feineman MD, Zack T, Barth M (2014) Metamorphic reaction rates at similar to 650–800 oC from
diffusion of niobium in rutile. Geochim Cosmochim Acta 130:63–77
Davis WJ (1997) U–Pb zircon and rutile ages from granulite xenoliths in the Slave province: evidence for mafic
magmatism in the lower crust coincident with Proterozoic dike swarms. Geology 25:343–346
Diener JFA, Powell R (2012) Revised activity-composition models for clinopyroxene and amphibole. J Met Geol
30:131–142
Dill HG, Melcher F, Füẞl, Weber B (2007) The origin of rutile–ilmenite aggregates (“nigrine”) in alluvial–fluvial
placers of the Hagendorf pegmatite province, NE Bavaria, Germany. Mineral Petrol 89:133–158
Petrol 40:343–346
Dodson MH (1986) Closure profiles in cooling systems. Mat Sci Forum 7:145–154
Dohmen R, Chakraborty S (2003) Mechanism and kinetics of element and isotopic exchange mediated by a fluid
phase. Am Mineral 88:1251–1270
Dunkl I, von Eynatten H (2009) Anchizonal-hydrothermal growth and (U–Th)/He dating of rutile crystals in the
sediments of Hawasina window, Oman. Geochim Cosmochim Acta 73:A314.
Geochem 83:365–418
Engvik A, Ihlen PM, Austrheim H (2014) Characterisation of Na-metasomatism in the Sveconorwegian Bamble
Sector of South Norway. Geosci Front 5:659–672
Ernst WG, Liu J (1998) Experimental phase-equilibrium study of Al- and Ti-contents of calcic amphibole in
MORB - A semiquantitative thermobarometer. Am Mineral 83:952–969
Ewing TA, Hermann J, Rubatto D (2013) The robustness of the Zr-in-rutile and Ti-in-zircon thermometers during
Ewing TA, Rubatto D, Beltrando M, Hermann J (2015) Constraints on the thermal evolution of the Adriatic margin
during Jurassic continental break-up: U–Pb dating of rutile from the Ivrea-Verbano Zone, Italy. Contrib
Mineral Petrol 169:44. Doi:10.1007/s00410–015-1135–6
Flowers RM, Mahan KH, Bowring SA, Williams ML, Pringle MS, Hodges KV (2006) Multistage exhumation
and juxtaposition of lower continental crust in the western Canadian Shield: Linking high-resolution U–Pb
and 40Ar/39Ar thermochronometry with pressure–temperature–deformation paths. Tectonics 25:TC4003.
Doi:10.1029/2005TC001912
Force ER (1980) The provenance of rutile. J Sediment Petrol 50:485–488
Frost BR (1991) Stability of oxide minerals in metamorphic rocks. Rev Mineral 25:469–488
Chem Geol 172:131–148
Grove M, Harrison TM (1999) Monazite Th–Pb age depth profiling. Geology 27:487–490
Handy MR, Franz L, Heller F, Janott B, Zurbriggen R (1999) Multistage accretion and exhumation of the
continental crust (Ivrea crustal section, Italy and Switzerland). Tectonics 18:1154–1177
Harley SL (2008) Refining the P–T records of UHT crustal metamorphism. J Metamorph Geol 26:125–154
Harrison TM, Grove M, Lovera OM, Zeitler PK (2005) Continuous thermal histories from inversion of closure
profiles. Rev Mineral Geochem 58:389–409
Hart E, Storey C, Bruand E, Schertl H-P, Alexander BD (2016) Mineral inclusions in rutile: A novel recorder of
HP-UHP metamorphism. Earth Planet Sci Lett 446:137–148
Henk A, Franz L, Teufel S, Oncken O (1997) Magmatic underplating, extension and crustal equilibration: insights
from a cross-section through the Ivrea Zone and Strona–Ceneri Zone, Northern Italy. J Geol 105:367–377
Hubert JF (1962) A zircon–tourmaline–rutile maturity index and the interdependence of the composition of heavy
mineral assemblages with the gross composition and texture of sandstone. J Sediment Petrol 32:440–450
Geochem 83:103–152
Evidence for a static hot channel in the Himalayan orogen. Earth Planet Sci Lett 311:136–143
Kohn MJ, Penniston-Dorland SC, Ferreira JCS (2016) Implications of near-rim compositional zoning in rutile
for geothermometry, geospeedometry, and trace element equilibration. Contrib Mineral Petrol 171:78, doi
10.1007/s00410–016-1285–1
Kooijman E, Mezger K, Berndt J (2010) Constraints on the U–Pb systematics of metamorphic rutile from in situ
LA-ICP-MS analysis. Earth Planet Sci Lett 293:321–330
Kooijman E, Smit MA, Mezger K, Berndt J (2012) Trace element systematics in granulite facies rutile: implications
for Zr geothermometry and provenance studies. J Metamorph Geol 30:397–412
Kooijman E, Hacker BR, Smit MA, Kylander-Clark ARC (2015) Rutile thermochronology constrains time-
resolved cooling histories in orogenic belts. Goldschmidt Abstr:1657
Kunz BE, Johnson TE, White RW, Redler C (2014) Partial melting of metabasic rocks in Val Strona di Omegna,
Ivrea Zone, northern Italy. Lithos 190:1–12
Kylander-Clark ARC, Hacker BR, Mattinson JM (2008) Slow exhumation of UHP terranes: Titanite and rutile ages
of the Western Gneiss Region, Norway. Earth Planet Sci Lett 272:531–540
Geochem 83:55–102
Li Q-L, Lin W, Su W, Li X-H, Shi Y-H, Liu Y, Tang G-Q (2011) SIMS U–Pb rutile age of low-temperature eclogites
from southwestern Chinese Tianshan, NW China. Lithos 122:76–86
Lucassen F, Dulski P, Abart R, Franz G, Rhede D, Romer RL (2010) Redistribution of HFSE elements during rutile
replacement by titanite. Contrib Mineral Petrol 160:279–295
Ludwig KR, Cooper JA (1984) Geochronology of Precambrian granites and associated U–Ti–Th mineralization,
northern Olary province, South Australia. Contrib Mineral Petrol 86:298–308
Luvizotto GL, Zack T (2009) Nb and Zr behavior in rutile during high-grade metamorphism and retrogression: An
example from the Ivrea–Verbano Zone. Chem Geol 261:303–317
Luvizotto GL, Zack T, Meyer HP, Ludwig T, Triebold S, Kronz A, Muenker C, Stockli DF, Prowatke S, Klemme
S, Jacob DE, von Eynatten H (2009a) Rutile crystals as potential trace element and isotope mineral standards
for microanalysis. Chem Geol 261:346–369
Luvizotto GL, Zack T, Triebold S, von Eynatten H (2009b) Rutile occurrence and trace element behavior in
medium-grade metasedimentary rocks: example from the Erzgebirge, Germany. Mineral Petrol 97:233–249
Manning CE, Bohlen SR (1991) The reaction titanite + kyanite = anorthite + rutile and titanite–rutile barometry in
eclogites. Contrib Mineral Petrol 109:1–9
Marschall HR, Dohmen R, Ludwig T (2013) Diffusion-induced fractionation of niobium and tantalum during
continental crust formation. Earth Planet Sci Lett 375:361–371
Matthews A (1994) Oxygen-isotope geothermometers for metamorphic rocks. J Met Geol 12:211–219
Matthews A, Schliestedt M (1984) Evolution of the blueschist and greenschist facies rocks of Sifnos, Cyclades,
Greece- a stable isotope study of subduction-related metamorphism. Contrib Mineral Petrol 88:150–163
Matthews A, Beckinsale RD, Durham JJ (1979) Oxygen isotope fractionation between rutile and water and
geothermometry of metamorphic eclogites. Mineral Mag 43:405–413
Meinhold G (2010) Rutile and its applications in earth sciences. Earth-Sci Rev 102:1–28
Meyer M, John T, Brandt S, Klemd R (2011) Trace element composition of rutile and the application of Zr-in-rutile
thermometry to UHT metamorphism (Epupa Complex, NW Namibia). Lithos 126:388–401
Mezger K, Hanson GN, Bohlen SR (1989) High-precision U–Pb ages of metamorphic rutile- application to the
cooling history of high-grade terranes. Earth Planet Sci Lett 96:106–118
Mezger K, Bohlen SR, Hanson GN (1990) Metamorphic history of the Archean Pikwitonei granulite domain and
the Cross Lake Subprovince, Superior Province, Manitoba, Canada. J Petrol 31:483–517
Miller C, Zanetti A, Thoeni M, Konzett J (2007) Eclogitisation of gabbroic rocks: Redistribution of trace elements
and Zr in rutile thermometry in an Eo-Alpine subduction zone (Eastern Alps). Chem Geol 239:96–123
Möller A, Mezger K, Schenk V (2000) U–Pb dating of metamorphic minerals: Pan-African metamorphism and
prolonged slow cooling of high pressure granulites in Tanzania, East Africa. Precambrian Res 104:123–146
Moore JM, Waters DJ (1990) Geochemistry and origin of cordierite–orthoamphibole/orthopyroxene–phlogopite
rocks from Namaqualand, South Africa. Chem Geol 85:77–100
Moore DK, Cherniak DJ, Watson EB (1998) Oxygen diffusion in rutile from 750 to 1000 °C and 0.1 to 1000 MPa.
Am Mineral 83:700–711
Moraes R, Brown M, Fuck RA, Camargo MA, Lima TM (2002) Characterization and P–T evolution of melt-
bearing ultrahigh-temperature granulites: An example from the Anapolis-Itaucu Complex of the Brasilia Fold
Belt, Brazil. J Petrol 43:1673–1705
Morton AC, Chenery SP (2009) Detrital rutile geochemistry and thermometry as guides to provenance of Jurassic–
Paleocene sandstones of the Norwegian Sea. J Sediment Res 79:540–553
Mulch A, Cosca MA, Handy MR (2002) In-situ UV-laser 40Ar/39Ar geochronology of a micaceous mylonite: an
example of defect-enhanced argon loss. Contrib Mineral Petrol 142:738–752
Okay N, Zack T, Okay AI, Barth M (2011) Sinistral transport along the Trans-European Suture Zone: detrital
zircon–rutile geochronology and sandstone petrography from the Carboniferous flysch of the Pontides. Geol
Mag 148:380–403
temperature (UHT) rocks. Contrib Mineral Petrol 171:44, doi:10.1007/s00410–016-1254–8
Pauly J, Marschall HR, Meyer HP, Chatterjee N, Monteleone B (2016) Prolonged Ediacaran-Cambrian metamorphic
history and short-lived high-pressure granulite-facies metamorphism in the H.U. Sverdrupfjella, Dronning Maud
Land (East Antarctica): evidence for continental collision during Gondwana assembly. J Petrol 57:185–228
466 Zack & Kooijman
Peressini G, Quick JE, Sinigoi S, Hofmann AW, Fanning M (2007) Duration of a large mafic intrusion and heat
transfer in the lower crust: A SHRIMP U–Pb zircon study in the Ivrea-Verbano Zone (Western Alps, Italy).
J Petrol 48:1185–1218
Pidgeon RT, Bosch D, Bruguier O (1996) Inherited zircon and titanite U–Pb systems in an Archaen sysenite from
southwestern Australia: implications for U–Pb stability of titanite. Earth Planet Sci Lett 141:187–198
Quick JE, Sinigoi S, Snoke AW, Kalakay TJ, Mayer A, Peressini G (2003) Geological Map of the Southern
Ivrea-Verbano Zone, Northwestern Italy. Geologic Investigations Series Map I-2776 and booklet (22 p) US
Geological Survey
Redler C, Johnson TE, White RW, Kunz BE (2012) Phase equilibrium constraints on a deep crustal metamorphic
field gradient: metapelitic rocks from the Ivrea Zone (NW Italy). J Metamorph Geol 30:235–254
Rice CM, Darke KE, Still JW (1998) Tungsten-bearing from the Kori Kollo gold mine, Bolivia. Mineral Mag
62:421–429
Richards JP, Krogh TE, Spooner ETC (1988) Fluid inclusions characteristics and U–Pb rutile age of late
hydrothermal alteration veining at the Suoshi stratiform copper deposit, central African copper belt. Econ
Geol 83:118–139
Rösel D (2014) U–Pb dating of detrital rutile: implications for sedimentary provenance and Pb diffusion in rutile.
PhD study, Universität Mainz, Germany, 196 p
Rösel D, Zack T, Boger SD (2014) LA-ICP-MS U–Pb dating of detrital rutile and zircon from the Reynolds
Range: A window into the Palaeoproterozoic tectonosedimentary evolution of the North Australian Craton.
Rudnick RL, Gao S (2014) Composition of the continental crust. Treatise in Geochemistry (2nd Edition) In:
Holland H, Turekian K (ed) Elsevier, Vol. 4, p. 1–51
Schärer U, Krogh TE, Gower CF (1986) Age and evolution of the Grenville Province in eastern Labrador from
U–Pb systematics in accessory minerals. Contrib Mineral Petrol 94:438–451
Geochem 83:199–230
Schmitt AK, Zack T (2012) High-sensitivity U–Pb rutile dating by secondary ion mass spectrometry (SIMS) with
an O2+ primary beam. Chem Geol 332:65–73
Schmitz MD, Bowring SA (2003) Constraints on the thermal evolution of continental lithosphere from U–Pb
accessory mineral thermochronometry of lower crustal xenoliths, southern Africa. Contrib Mineral Petrol
144:592–618
Shulaker DZ, Schmitt AK, Zack T, Bindeman I (2015) In-situ oxygen isotope and trace element geothermometry
of rutilated quartz from Alpine fissures. Am Mineral 100:915–925
Siegesmund S, Layer P, Dunkl I, Vollbrecht A, Steenken A, Wemmer K, Ahrendt H (2008) Exhumation and
deformation history of the lower crustal section of the Valstrona di Omegna in the Ivrea Zone, southern Alps.
In: Tectonic Aspects of the Alpine–Dinaride–Carpathian System. Vol 298. Siegesmund S, Fugenschuh B,
Froitzheim N (eds). Geol Soc Spec Publ 298:45–68
Sinigoi S, Quick JE, Mayer A, Demarchi G (1995) Density-controlled assimilation of underplated crust, Ivrea-
Verbano zone, Italy. Earth Planet Sci Lett 129:183–191
Smit MA, Scherer EE, Mezger K (2013) Peak metamorphic temperatures from cation diffusion zoning in garnet.
Smit MA, Ratschbacher L, Kooijman E, Stearns MA (2014) Early evolution of the Pamir deep crust from Lu–Hf
and U–Pb geochronology and garnet thermometry. Geology 42:1047–1050
Smit MA, Waight TE, Nielsen TFD (2016) Millennia of magmatism recorded in crustal xenoliths from alkaline
provinces in Southwest Greenland. Earth Planet Sci Lett 451: 241–250
Smith DC, Perseil EA (1997) Sb-rich rutile in the manganese concentrations at St. Marcel-Praborna, Aosta valley,
Italy: Petrology and crystal-chemistry. Mineral Mag 61:655–669
Smith SJ, Stevens R, Liu S, Li G, Navrotsky A, Boerio-Goates J, Woodfield BF (2009) Heat capacities and
thermodynamic functions of TiO2 anatase and rutile: Analysis of phase stability. Am Mineral 94:236–243
Spencer KJ, Hacker BR, Kylander-Clark ARC, Andersen TB, Cottle JM, Stearn MA, Poletti JE, Seward GGE
(2013) Campaign-style titanite U–Pb dating by laser ablation ICP: Implications for crustal flow, phase
transformation and titanite closure. Chem Geol 341:84–101
Stearns MA, Cottle JM, Kylander-Clark AR, Hacker BR (2016) Extracting thermal histories from the near-rim
zoning in titanite using coupled U–Pb and trace element depth profiles by single shot-split stream laser
ablation (SS-LASS) ICP-MS: Chem Geol 422:13–24
Tajčmanová L, Connolly JAD, Cesare B (2009) A thermodynamic model for titanium and ferric iron solution in
biotite. J metamorphic Geol 27:153–165
Taylor R, Clark C, Reddy SM (2012) The effect of grain orientation on secondary ion mass spectrometry (SIMS)
analysis of rutile. Chem Geol 300:81–87
Topuz G, Göcmengil G, Rolland Y, Celik ÖF, Zack T, Schmitt AK (2013) Jurassic accretionary complex and
ophiolite from northeast Turkey: No evidence for Cimmerian continental ribbon. Geology 41:255–258
Triebold S, von Eynatten H, Zack T (2012) A recipe for the use of rutile in sedimentary provenance analysis.
Sediment Geol 282:268–275
Upadhyay D, Chattopadhyay S, Kooijman E, Mezger K, Berndt J (2014) Magmatic and metamorphic history
of Paleoarchean tonalite–trondhjemite–granodiorite (TTG) suite from the Singhbhum craton, eastern India.
Valley JW (2001) Stable isotope thermometry at high temperature. Rev Mineral Geochem 43:365–413
Vavra G, Schmid R, Gebauer D (1999) Internal morphology, habit and U–Th–Pb microanalysis of amphibolite-to-
granulite facies zircons: geochronology of the Ivrea Zone (Southern Alps). Contrib Mineral Petrol 134:380–404
Villa IM (1998) Isotopic closure. Terra Nova 10:42–47
Villa IM (2016) Diffusion in mineral geochronometers: present and absent. Chem Geol 420:1–10
Vry JK, Baker JA (2006) LA-MC-ICPMS Pb–Pb dating of rutile from slowly cooled granulites: Confirmation of
the high closure temperature for Pb diffusion in rutile. Geochim Cosmochim Acta 70:1807–1820
Waters DJ (1986a) Metamorphic history of sapphirine-bearing and related magnesian gneisses from Namaqualand,
South Africa. J Petrol 27:541–565
Waters DJ (1986b) Metamorphic zonation and thermal history of pelitic gneisses from western Namaqualand,
South Africa. Trans Geol Soc S Afr 89:97–102
Petrol 151:413–433
Geochem 83:153–182
Willner AP (1995) Pressure–temperature evolution of a low-pressure amphibolite-facies terrane in central
Bushmanland (Namaqua Mobile Belt, South Africa). Commun Geol Surv Namibia 10:5–20
Zack T, Luvizotto GL (2006) Application of rutile thermometry to eclogites. Mineral Petrol 88:69–85
Zack T, Kronz A, Foley SF, Rivers T (2002) Trace element abundances in rutiles from eclogites and associated
garnet mica schists. Chem Geol 184:97–122
Zack T, Moraes R, Kronz A (2004a) Temperature dependence of Zr in rutile: empirical calibration of a rutile
Zack T, von Eynatten H, Kronz A (2004b) Rutile geochemistry and its potential use in quantitative provenance
studies. Sed Geol 171:37–58
Zack T, Stockli DF, Luvizotto GL, Barth MG, Belousova E, Wolfe MR, Hinton RW (2011) In situ U–Pb rutile
dating by LA-ICP-MS: 208Pb correction and prospects for geological applications. Contrib Mineral Petrol
162:515–530
Zanetti A, Mazzucchelli M, Sinigoi S, Giovanardi T, Peressini G, Fanning M (2013) SHRIMP U–Pb Zircon
Triassic Intrusion Age of the Finero Mafic Complex (Ivrea-Verbano Zone, Western Alps) and its Geodynamic
Implications. J Petrol 54:2235–2265
Zingg A (1980) Regional metamorphism in the Ivrea zone (Southern Alps, N-Italy): field and microscopic
investigations. Schweiz Mineral Petrograph Mitteil 60:153–179
Zingg A (1983) The Ivrea and Strona-Ceneri zones (Southern Alps, Ticino and N-Italy—a review. Schweiz Mineral
Petrograph Mitteil 63:361–392
Zingg A, Handy MR, Hunziker JC, Schmid SM (1990) Tectonometamorphic history of the Ivrea Zone and its
relationship to the crustal evolution of the Southern Alps. Tectonophysics 182:169–192
Vol. 83 pp. 469–533, 2017 15
Garnet: A Rock-Forming Mineral Petrochronometer

E.F. Baxter
Boston College
Chestnut Hill, Massachusetts 02467
USA
baxteret@bc.edu
M.J. Caddick
Virginia Tech
Blacksburg, Virginia 24060
USA
caddick@vt.edu
B. Dragovic
Virginia Tech
Blacksburg, Virginia 24060
USA
dragovic@vt.edu
INTRODUCTION
Garnet could be the ultimate petrochronometer. Not only can you date it directly (with
an accuracy and precision that may surprise some), but it is also a common rock-forming and
porphyroblast-forming mineral, with wide ranging—yet thermodynamically well understood—
solid solution that provides direct and quantitative petrologic context. While accessory phase
petrochronology is based largely upon establishing links to the growth or breakdown of key rock-
forming pressure–temperature–composition (P–T–X) indicators (e.g., Rubatto 2002; Williams
et al. 2007), garnet is one of those key indicator minerals. Garnet occurs in a great variety of
rock types (see Baxter et al. 2013) and is frequently zoned (texturally, chemically) meaning that
it contains more than just a snapshot of metamorphic conditions, but rather a semi-continuous
history of evolving tectonometamorphic conditions during its often prolonged growth. In this
way, garnet and its growth zonation have been likened to dendrochronology: garnet as the tree
rings of evolving tectonometamorphic conditions (e.g., Pollington and Baxter 2010).
In some ways, the dream of ‘petrochronology’ all started with garnet (Fig. 1). When
Atherton and Edmunds (1965) or Hollister (1966) recognized the chemical zonation in
garnet, when Rosenfeld (1968) noted the spiral ‘snowball’ of inclusions in rotated garnet,
or when Tracy et al. (1976) drew the first 2-D map of garnet chemical zonation, illuminating
those ‘tree-rings’ for the first time, they could only imagine what is now a reality decades
later—direct zoned garnet geochronology of those concentric rings of growth. Geoscientists
soon thereafter attempted the first garnet geochronology (van Breemen and Hawkesworth
1980), though several factors severely limited the development and wider-spread use of

1943-2666/17/0083-0015$5.00 (online)
470 Baxter, Caddick & Dragovic
A FeO,
wt%
Mn
B C
Fe 1
6.0 31.3
2
3
54
5.0 MgO, 6
7
30.0 Mg wt% 8
MnO, wt%
Fe 4.75
4.0
9
4.1
28.8 Mg
3.0 3.45
8
2.0 7
6
5
4
1.0 3
2
Mn 2 mm
1
0.0
300 200 100 0 100 200 300
Radius, µm
Figure 1. Important observations in the development of garnet petrochronology. A. Chemical zoning in

a 1-D traverse across garnet from the Kwoiek area of British Columbia (redrawn from Hollister 1966).
The ‘bell-shaped’ Mn zoning profile led to the now classic interpretation of garnet growth adhering to a
Rayleigh fractionation model. B. 2-D zoning of Mn in garnet from central Massachusetts (redrawn from
Tracy et al 1976, numbers equate to atomic percent Mn and dots indicate position of microprobe analyses).
C. Spiral ‘snowball’ inclusions in rotated garnet described by Rosenfeld (1968); image shown is from thin
section with first-order red filter provided by John Rosenfeld.
garnet geochronology from that point. These factors included 1) contamination of garnet by
micro-mineral inclusions, 2) analytical limitations of small sample size, 3) the requirement
of anchoring a garnet age analysis with another point on an isochron, and 4) the significant
time and effort required for age determination. Even today, whether via MC-ICPMS or TIMS
(e.g., Schoene and Baxter 2017, this volume), garnet geochronology requires weeks of time-
consuming sample preparation. So, while petrologists boldly forged ahead in the use and
development of garnet as probably the premier mineral recorder of evolving metamorphic and
tectonic processes during the 1970’s, 80’s and 90’s, garnet geochronology was mostly (though
not completely) supplanted by the excitement, relative ease and undoubted utility of accessory
phase geochronology (as reviewed by Engi 2017, this volume; Rubatto 2017, this volume).
Then, because the petrology and thermodynamics of accessory phases had not been as well
studied, the task of ‘petrochronology’ was to bring petrologic context to the age information
accessible in phases such as zircon and monazite. Often, this endeavor came back to linking
the growth of accessory phases with a key rock-forming mineral—garnet. The last 20 years
have now seen the advancement of direct garnet geochronology driven by several factors,
including 1) the addition of Lu–Hf to Sm–Nd as viable systems to date garnet (e.g., Duchene
et al. 1997; Scherer et al. 2000), 2) robust methods to eliminate the effects of contaminating
inclusions (e.g., Amato et al. 1999; Baxter et al. 2002; Anczkiewicz and Thirlwall 2003), 3)
improved analytical techniques to reduce sample size limitations (e.g., Harvey and Baxter
2009; Bast et al. 2015), and 4) microsampling methods whereby those individual tree-rings
can be sampled at higher and higher spatial resolution (e.g., Stowell et al. 2001; Ducea et al.
2003; Pollington and Baxter 2010, 2011). What this makes possible today is the introduction—
or re-introduction—of garnet into the cadre of modern ‘petrochronometers’.
Our purpose in this chapter is threefold. We begin with a review of the ‘petro-’ of garnet,
followed by the ‘-chrono-’ of garnet, setting us up for the modern possibilities—many yet
to be fully explored—of garnet ‘petrochronology’. First, we re-acquaint the reader with the
myriad ways in which garnet has been used directly to reconstruct past tectonometamorphic
conditions. Ranging from foundational thermodynamic P–T modeling and textural analysis to
recent advances in inclusion barometry and trace element zonation, our aim is to illuminate for
all readers the remarkable scope and potential for garnet as a recorder of tectonometamorphic
context. It is not garnet per se that we are interested in, but rather the conditions and processes
Garnet: A Rock-Forming Mineral Petrochronometer 471
that it records. Second, we bring the reader up to date on recent advances in direct chronology
of garnet via Sm–Nd and Lu–Hf systems that have made it the robust and precise chronometer
that it is. Our hope is to dispel the misconceptions that continue to limit the use and credence
of garnet geochronology, and clearly lay out the wide-ranging opportunities for direct garnet
geochronology, appropriately framed within the challenges and limitations which still exist
for this (as with any!) geochronometer. This section on the ‘chrono’ of garnet also discusses
diffusion geospeedometry of garnet, which provides complementary information on heating/
cooling durations and rates. Finally, we bring the ‘petro’ and ‘chrono’ together and present
several examples from the past decade of true garnet petrochronology wherein these methods
are integrated in increasingly innovative ways. Figure 2 serves as both an outline of our
chapter, and as our definition of garnet petrochronology: the effort to integrate any aspect(s) of
garnet-based petrology (on the left of Fig. 2) with any aspect(s) of garnet geochronology, and
in so doing, to gain insights about evolving tectonometamorphic conditions that complement
what may be extracted from other methodologies. Perhaps the most important thing to
emphasize is that garnet petrochronology is really still in its infancy. Many challenges have
been overcome, and it is time now for creative petrologists to come back to garnet and try new
ways of integrating the ‘petro’ and the ‘chrono’ (by choosing from the left and the right of
Fig. 2) to make true advances in our understanding of petrologic and tectonic evolution of the
crust-mantle system. Never before has the time been so ripe, and we hope the reader is inspired
to take these next steps in creatively deploying garnet petrochronology in the future.
PETRO- CHRONO-
Textures Sm-Nd
Geothermobarometry Lu-Hf
Pseudosections GARNET Coupled Sm-Nd & Lu-Hf

PETROCHRONOLOGY
Geo-ba-raman-try Zoned chronology
Stable isotope tracers Geospeedometry
Trace element tracers [accessory phase chronology]
Figure 2. Outline of the chapter and definition of garnet petrochronology. Any “petro-“ of garnet integrated
with any “chrono” of garnet makes for garnet petrochronology. Accessory phase geochronology is included
with a dashed line to note its integrative value, as valuable efforts have been made to link it to garnet.
PETRO- OF GARNET
Said the professor to their student: “please try to collect some samples with garnet in them;
there’s not nearly as much we can do without it”. Although this statement is apocryphal, the
sentiment will be familiar to many readers. Metamorphic petrology has been tied to the search
for and study of garnet-bearing samples since soon after Barrow defined the ‘garnet zone’ in the
eastern Scottish Highlands (Barrow 1893, 1912). Advancements in our ability to characterize
the compositions and textures of metamorphic minerals, and in thermochemical models to
describe the conditions of their evolution have commonly focused on garnet-bearing samples.
In this section we review some of the reasons why garnet has been so central to the endeavor of
reconstructing tectonometamorphic processes and conditions. We give an overview of some of
the mechanisms that have been developed to infer pressure, temperature, deformation, mineral
reaction or fluid–rock interaction, highlighting some of the recent and current developments
that will aid this. Much has already been written about most of this, and we have thus made
no attempt to be absolutely comprehensive, instead directing the reader to several of the other
excellent reviews where each provide more detail about each technique. But by bringing all of
this content together, our goal is to provide a comprehensive taste of what garnet has previously
been used to quantify and constrain, with the expectation that these methods will increasingly
be coupled with each other and with direct dating, as described later in the ‘Chrono- of garnet’
section of this chapter. We will begin with the most easily observable (often at the hand sample
scale) textural information locked in garnet, before reviewing its role in geothermobarometry
and in tracing evolving petrologic reactions and open system fluid flow processes.
Textures of garnet—provider of tectonic context
Garnet’s rigid porphyroblastic growth endows it with a propensity to retain and record its
growth history in chemical as well as textural manifestations. This section provides several
brief examples of how textural observations can provide valuable petrologic and tectonic
context for garnet’s growth span.
Garnet as a deformation and strain monitor. Garnet crystals are frequently riddled with
cracks and inclusions of other minerals that can reveal a reaction history and the co-genetic
evolution of tectonic stresses. For example Angiboust et al. (2011) interpreted cracks in UHP
eclogite garnets as reflecting seismic brecciation during subduction to ~80 km depth in the
Western Alps. Shear stress can also deform the matrix and cause garnet to rotate (Spry 1963;
i.e., ‘snowball garnets’; Rosenfeld 1968) or record changing metamorphic fabrics during
porphyroblast growth (i.e., Ramsay 1962; Bell 1985). Controversy still exists between these
two end-member interpretations (i.e., Johnson 1993; Ikeda et al. 2002; Stallard et al. 2002)
but in any case, these common and often vivid patterns of inclusions inside garnets (e.g.,
Fig. 1c) clearly record rock deformation that may reflect localized structural—(e.g., Robyr
et al. 2009) or regional tectonic-scale (e.g., Aerden et al. 2013; Sayab et al. 2016) processes.
Beyond documentation of inclusion patterns in thin section, valuable textural information
can be gleaned from additional methods of observation. Robyr et al. (2009) employed EBSD
(electron backscattered diffraction) analysis to show that crystallographic orientations within
single garnet porphyroblasts remained constant while the garnet was rotating. Then, those
crystallographic orientations shifted to reflect surrounding phyllosilicate foliation when
garnet grew during subsequent non-rotational regimes. Aerden et al. (2013) relate different
textural generations of inclusions within garnets, each separated by a FIA (fold intersection
axis) to major shifts in tectonic convergence vectors during the assembly and deformation
of the Iberian Peninsula. These FIA are carefully mapped through analysis of radial sets
of vertical thin sections and would not otherwise be readily apparent in a simple 2D thin
section. Sayab et al. (2016) use similar methods to document tectonic convergence vectors
in the Himalayan Orogen. Among the very first applications of true garnet petrochronology,
Christensen et al. (1989) and Vance and Onions (1992) dated the cores and rims of such
rotated snowball garnets to place constraints on the tectonic shear strain rate. Cases in which
the identity or composition of the included minerals change from core to rim of the host
garnet reveal additional, powerful information, as detailed below (e.g., St-Onge 1987).
Garnet resorption textures. As systems evolve and P–T–X changes, it is common for
garnet to become unstable and start to break down by consumption from its rim inward (see
Lanari et al. 2017 for additional examples). This resorption is driven by changes in P–T–X
occurring during prograde metamorphism (e.g., Florence and Spear 1993), during retrograde
metamorphism, due to chemical change related to influx of an external fluid, and sometimes
due to polymetamorphism (when old garnet is introduced into a new orogenic cycle). Such
resorbed surfaces often exhibit irregular morphologies, such as embayments, which often
contrast with sharp original crystal growth faces. On the one hand, once a crystal (or its
outermost portion) has been resorbed, there is little direct ‘garnet petrochronology’ that can
be done to study that resorption event. However, if a resorption surface can be identified
(texturally or chemically), the relative timing and conditions surrounding the resorption event
may still be constrained. For example, if an element liberated during garnet resorption cannot
be readily incorporated into product phases, a small amount of garnet might re-precipitate,
preferentially incorporating that element, or the element itself might be reincorporated into
any residual garnet (e.g., see discussion below relating to Y and HREE incorporation into
resorbed crystal rims). In general, such elements would be those that tend to be strongly
partitioned into garnet over other matrix phases including Mn, Y, Lu and other HREE (e.g.,
Kohn and Spear 2000; Kelly et al. 2011; Gatewood et al. 2015). In some cases, secondary
minerals may form at a garnet resorption surface driven by the sudden and proximal influx of
those particular elements. Xenotime, for example, has been observed decorating the resorbed
surfaces of garnets where localized flux of Y (from the resorbed garnet) promotes growth of
xenotime (e.g., Gatewood et al. 2015). In this case, there is a useful textural link between
resorbed garnet surface and neocrystallized accessory phase xenotime.
Textural evidence for polymetamorphism. We use the term ‘polymetamorphic garnet’
to describe crystals that grew during more than one tectonic ‘event’ separated by a significant
hiatus that can be resolved texturally, chemically, or temporally with existing methods.
Several examples of polymetamorphic garnet have been recognized in the Alpine-Himalayan
system that grew during multiple orogenic cycles, separated by millions to hundreds of
millions of years. Argles et al. (1999) dated fractions of garnet crystals, identifying that
only crystal rims record Tertiary metamorphism, with crystal cores hundreds of millions of
years older. Gaidies et al. (2006), Herwartz et al. (2011), Robyr et al. (2014) and (Lanari
et al. 2017) recognized, garnet zonation textures in which Alpine overgrowths are vividly
separated by prominent microstructural and chemical discontinuities from older Variscan
cores. Manzotti and Ballevre (2013) recognized chemically heterogeneous detrital garnet
cores—derived from multiple sources—that were overgrown by a homogeneous Alpine
garnet rim. All of these examples serve to illustrate some of the textural and chemical
means by which polymetamorphic garnet may be recognized. Of course, only by adding
direct geochronology to each of these growth generations may we establish the absolute
chronology and length of the hiatus between phases of garnet growth.
Porphyroblast nucleation and growth models. Metamorphic porphyroblastic minerals
comprise vivid records of progressive nucleation and growth kinetics. Treated individually,
each porphyroblast may record aspects of the rock’s overall history colored by local chemical-
textural features. Taken as a whole, porphyroblast crystal size distributions, their relative
spatial disposition, and chemical zonation reveal information about the nucleation and growth
process rock wide (e.g., Kretz 1973; Carlson et al. 1995; Chernoff and Carlson 1997; Meth
and Carlson 2005). Several forward models have been developed to predict and interpret
garnet CSD’s and spatial dispositions within porphyroblastic rocks (Galwey and Jones 1966;
Kretz 1966, 1974; Cashman and Ferry 1988; Carlson 1989, 2011; Spear and Daniel 2001;
Gaidies et al. 2008b; Gaidies et al. 2011; Schwarz et al. 2011; Ketcham and Carlson 2012;
Kelly et al. 2013a,b). These models offer testable predictions about the rate limiting processes
for crystal growth and progressive metamorphism, including important implications for the
attainment of local and rock-wide chemical equilibrium. Existing models make justified
assumptions and interpretations about whether the rate-limiting step is breakdown of parent
minerals (e.g., Schwarz et al. 2011), surface energetics of garnet (e.g., Gaidies et al. 2011),
or transport of the least mobile nutrient (often aluminum) through the intergranular medium
(e.g., Ketcham and Carlson 2012). The importance of Ostwald ripening (favoring growth
of larger porphyroblasts at the expense of smaller porphyroblasts) has been debated (e.g.,
Carlson 1999), and some studies have cited heating rate as the key limitation. Each of these
published cases shows persuasively that numerical simulations of one or several processes
can successfully model natural crystal occurrences, suggesting that different rate-limiting
processes may dominate under different circumstances. Furthermore, EBSD analysis has
revealed that garnets porphyroblasts may often (perhaps > 20% of the time) include more than
one primary nucleation site and, thus, more than one fundamental crystal lattice orientation.
These early forming crystal clusters may ultimately coalesce, separated by high-angle grain
boundaries, to form a macroscopically visible porphyroblast (e.g., Daniel and Spear 1998;
Spiess et al. 2001; Whitney et al. 2008; Whitney and Seaton 2010).
Garnet as a pressure and temperature sensor

One of the most important, and familiar, aspects of the ‘petro-’ in petrochronology involves
deciphering metamorphic pressure and temperature to infer rocks’ journeys through the crust
and mantle. Garnet has arguably been the single most useful mineral for the estimation of
evolving metamorphic P–T conditions for over forty years1. It is especially useful because of
1) its occurrence in diverse lithologies and P–T conditions, and 2) its relatively simple and
well understood chemistry, which is dominated by divalent Fe, Mg, Ca, Mn, and trivalent
Al, Fe cations (e.g., Grew et al. 2013) but is also sensitive to certain trace elements like Y.
Because garnet is a major rock-forming mineral, its growth history is directly linked to broader
rock wide petrologic evolution; so we must also develop the use of broader descriptions of
the entire mineral assemblage and provide context for the growth of garnet and its changing
chemistry. Here we review some of the techniques that have been developed to utilize garnet’s
chemical and mechanical properties to constrain P–T.
Compositional thermometry—Major elements. The advent of the electron microprobe
in the 1950s sparked a revolution in metamorphic petrology as systematic characterization
revealed that the compositions of co-existing mineral phases are often strongly correlated with
metamorphic grade (e.g., Ramberg 1952; Kretz 1959). Garnet composition almost immediately
became a focus, leading to the discovery that crystals commonly exhibit zoning from core to
rim (Atherton and Edmunds 1965; Evans 1966; Evans and Guidotti 1966; Harte and Henley
1966; Hollister 1966; Atherton 1968) and in complex patterns around inclusions and upon
contact with specific matrix phases (Tracy et al. 1976; Thompson et al. 1977; Tracy 1982;
Figures 1 and 3). Relationships between mineral composition and metamorphic temperature
suggested a geothermometer based upon the partitioning of Fe and Mg between garnet and
biotite (e.g., Frost 1962; Kretz 1964; Saxena 1968, 1969; Thompson 1976b; Goldman and
Albee 1977). This results primarily from garnet’s large distorted cubic site favoring Fe rather
than Mg at low temperature, under which conditions biotite’s octahedral sites prefer Mg.
Indeed, garnet typically contains far lower Mg content than most coexisting phases, with the
ratio Mg / Mg + Fe(XMg) decreasing in the order cordierite > chlorite > biotite > chloritoid >
staurolite > garnet (Albee 1965, 1972; Hensen 1971). The preference of each phase for Mg or
Fe is reduced upon heating, thus imparting a temperature dependence of Fe–Mg partitioning
between most co-existing phases (the volumetric effect of Mg–Fe exchange is comparatively
minor). Garnet plays a particularly important role here because its low Mg content at low
temperature establishes a much larger ∆XMg between it and a second equilibrated mineral (e.g.,
orthopyroxene or biotite) than a pairing between most other phase combinations would permit,
explaining why garnet often occurs as a key mineral in exchange thermometer calibrations.
The garnet side of these equilibria is shown in Figure 4, which demonstrates the relatively
extensive Mg and Fe variation expected under possible crustal conditions, assuming a fixed
bulk rock composition. After the classic experimental calibration by Ferry and Spear (1978),
numerous refinements to the garnet–biotite exchange thermometer (Perchuk and Lavrent’eva
1983; Ganguly and Saxena 1984; Indares and Martignole 1985; Kleeman and Reinhardt 1994;
Berman and Aranovich 1996; Gessmann et al. 1997; Holdaway et al. 1997; Mukhopadhyay et
al. 1997; Holdaway 2000) have made this one of the most often used geothermometers.
Temperature dependent Fe–Mg equilibria between garnet and many other phases have
also been calibrated as thermometers, all based on similar principles to the garnet–biotite
example outlined above. It is beyond the scope of this review to discuss all of these in detail,
but the reader is directed to Essene (1982, 1989) and Spear (1993) for more information.
1
For example, Spear (1993) lists 37 geothermobarometers on page 517−519 of his book; 25 of these include
garnet, far more than any other mineral.
A) XPy E) XPy B) XGr F) X Gr
Inclusions Inclusions
Staurolite Muscovite
Chloritoid 0 2 mm Margarite–paragonite
Chlorite
C) XSp G) XSp D) Fe/Fe+Mg H) Fe/Fe+Mg
Inclusions Inclusions
Ilm-Rut Staurolite
Ilm-Hem Chloritoid
Mag Chlorite
Figure 3. Garnet zoning, then and now. Panels A-D are some of the first 2-D garnet element zoning maps,
hand-contoured from point analyses by R.J. Tracy as part of his graduate studies and published as Thomp-
son et al (1977). These images of a crystal from the Gassetts Schist, Vermont, first hinted at the complexity
of zoning in natural garnet crystals and were used to advance a model of progressive crystal growth during
evolving P and T. The very first 2-D images had appeared a year earlier (Tracy et al 1976). Panels E-H are
element maps of the same crystal, collected ~ 35 years later by the same R.J. Tracy in a ~12 hour run with
an automated electron microprobe (Tracy et al 2012). The general zoning patterns were extremely well
documented by the previous work, but modern techniques yield a far more detailed picture, particularly
highlighting fine-scale zoning in the crystal core of the Ca map that is still difficult to construct a satisfac-
tory growth model for. Computerized automation has revolutionized many aspects of metamorphic petrol-
ogy, paving the way for the petrochronology discussed in this volume.
Compositional barometry. While low ∆V exchange equilibria were developed into

mineral thermometers, high ∆V equilibria were explored as potential barometers, again with
an early emphasis on garnet-bearing assemblages. Observation of and experimentation on the
garnet–plagioclase system (Kretz 1959; Ghent 1976; Goldsmith 1980; Newton and Haselton
1981) led to calibration of a mineral barometer that uses the end member ‘GASP’ (garnet–
aluminosilicate–silicate–plagioclase) reference reaction:
Ca3Al2Si3O12 + 2 Al2SiO5 + SiO2 = 3 CaAl2Si2O8
This is pressure sensitive because of the large molar volume difference of reactants and
products, and has been calibrated and recalibrated multiple times as new experimental data
and activity–composition (a–X) models for garnet and plagioclase have become available
(e.g., Ghent 1976, 1977; Newton and Haselton 1981; Koziol and Newton 1988; McKenna
and Hodges 1988; Berman 1990; Holdaway 2001; Caddick and Thompson 2008). The relative
performance of most of these calibrations was compared by Wu and Cheng (2006), but taken
as a group the GASP calibrations have become probably the most widely used chemical
geobarometer for metamorphic rocks, due in part to the large range of P–T conditions over which
garnet and plagioclase can both co-exist with a suitable buffering assemblage. Uncertainties
on temperature of equilibration, end-member thermodynamic properties (particularly
enthalpy and entropy), mineral solution behavior, and measurements of garnet and plagioclase
composition, likely combine to yield GASP uncertainties reaching or exceeding ±2 kbars in
many cases (see discussions by, Hodges and McKenna (1987), McKenna and Hodges (1988),
Kohn and Spear (1991) and Holdaway (2001) for more information).
2.0
Garnet abundance (%) Biotite ( FeMg
+ Mg )
1.8
1.6
1.4
Pressure, GPa
1.2 14 0.75
12 0.70
1.0
10 0.65
0.8 8 0.60
0.55
0.6 6
(garnet 0.50
4
0.4 not stable) 0.45
2
0.2 0.40
400 450 500 550 600 650 700 750 800 450 500 550 600 650 700 750 800
Temperature, ˚C Temperature, ˚C
2.0
X almandine (Fe + MgFe
+ Ca + Mn ) X pyrope (Fe + MgMg
+ Ca + Mn )
1.8
1.6
1.4
Pressure, GPa
0.70
1.2 0.35
0.65
0.30
1.0 0.60
0.25
0.8 0.55
0.20
0.6 0.50
0.15
(garnet 0.45
0.4 not stable) 0.10
0.40
0.2 0.05 (garnet not stable)
0.35
400 450 500 550 600 650 700 750 800 450 500 550 600 650 700 750 800
2.0
X spessartine (Fe + MgMn
+ Ca + Mn) X grossular (Fe + MgCa
+ Ca + Mn)
1.8
1.6
1.4
Pressure, GPa
0.45 0.30
1.2
0.40
0.25
1.0 0.35
0.30 0.20
0.8
0.25
0.6 0.20 0.15
(garnet 0.15 (garnet 0.10
0.4 not stable) not stable)
0.10
0.2 0.05 0.05
400 450 500 550 600 650 700 750 800 450 500 550 600 650 700 750 800
Figure 4. Mineral abundances and compositions in a pelitic rock at crustal conditions as a function of P and
T for the ‘average pelitic bulk-rock’ composition from Caddick and Thompson (2008). Calculated using
Perple_X (Connolly 2005) with the thermocalc ds5.5 dataset (updated from Holland and Powell 1998),
and activity models described in Caddick and Thompson (2008), except for garnet (White et al 2005),
amphibole and omphacite (both Diener and Powell 2012). The underlying calculation from which contours
were extracted considers many additional phases that are not shown, with biotite and garnet in most regions
of the diagrams co-existing with 5 or 6 additional phases (see annotated pseudosection for this rock, which
appeared as Figure 1 in Caddick and Thompson (2008), noting that that diagram was calculated with earlier
thermodynamic data and is thus not identical).
Many additional mineral systems have been suggested and calibrated for inferring pressure
and temperature, with a significant proportion of these relying partly on the properties of garnet,
e.g., garnet–plagioclase–biotite–muscovite, garnet–muscovite–plagioclase–quartz, garnet–
biotite–plagioclase–quartz, garnet–ilmenite–rutile–kyanite–quartz, garnet–ilmenite–rutile–
plagioclase–quartz, garnet–clinopyroxene–plagioclase–quartz, garnet–orthopyroxene, garnet–

clinopyroxene–phengite (Bohlen and Essene 1980; Ghent and Stout 1981; Bohlen et al. 1983;
Hodges and Crowley 1985; Bohlen and Liotta 1986; Moecher et al. 1988; Hoisch 1990, 1991;
Waters and Martin 1993; Pattison et al. 2003; Wu et al. 2004; Wu and Zhao 2006). Again, more
details about many of these were provided by Essene (1989) and by Spear (1993).
Compositional thermometry—Trace elements. While the distribution of various
trace elements in garnet can be used to determine reaction sequence, the applicability of
geochronometers, and fluid/melt history (see discussions below), trace element (particularly
yttrium) concentrations of coexisting accessory phases and garnet have also been utilized as
thermometers in pelitic lithologies.
Pyle and Spear (2000) developed an empirical trace element thermometer for the
temperature range of 450–550 ºC, by study of regional metamorphic rocks from New England.
This method utilizes the yttrium content of garnet in xenotime-bearing pelites, assuming that the
presence of xenotime buffers the rock in Y. Temperature and pressure estimates for calibration
purposes, along with P–T paths, were determined by traditional thermobarometry and differential
thermodynamics, respectively. Pyle et al. (2001) expanded this approach to the assessment
of equilibrium partitioning of coexisting monazite–xenotime–garnet, developing a garnet–
monazite thermometer. Critical in employing this thermometer is 1) accurate compositional
information about all phases involved in the reaction (Foster et al. 2004; Hallett and Spear 2015)
and 2) that monazite was a stable phase during garnet growth, i.e., that garnet and monazite
were in equilibrium. Inclusion of homogeneous monazite in homogeneous or continuously
zoned garnet may be considered a good candidate for such equilibration, though zoning in that
monazite and the chances of its isolation from the reactive part of the rock warrant particular
care, as further discussed by Lanari and Engi (2017, this volume). Uncertainties in the other
parameters for calculation likely amount to a combined uncertainty of approximately ±25 ºC
(Pyle et al. 2001). Pyle et al. (2001) note that garnet and monazite maintain compositional
equilibrium even as garnet fractionation dramatically changes the Y budget in the reactive bulk
composition, allowing for a consistent element partitioning during the rock’s prograde history.
Thermobarometry of zoned garnet. If mineral composition is to be used to infer pressure
and temperature but natural phases such as garnet are commonly chemically zoned (e.g.,
Figs. 1 and 3), where should one make a microprobe analysis, where should analyses of
coexisting phases be made, and how can results be put into the context of an evolving P–T
history? These questions have been addressed in detail (e.g., Essene 1989; Kohn and Spear
2000), and extreme caution should be taken before averaging data whose variability may be
geologically meaningful. For example, the apparent repeatability of garnet zoning patterns
suggests that this zoning records changes in pressure and temperature during crystal growth,
leading to early models of garnet zoning due to sequential mineral reactions (e.g., Hollister
1966; Loomis 1975; Thompson 1976a; Tracy et al. 1976; Tracy 1982), which in turn led to
direct use of garnet zoning to infer P–T paths during prograde metamorphism. Development
of a differential thermodynamic approach, commonly referred to as the Gibbs method,
allowed derivation of P–T paths for rocks, based upon garnet crystal zoning that is assumed
to have been established during growth and an estimate of the P–T point at which garnet first
began to grow (Spear and Selverstone 1983; Spear et al. 1984). Zoned garnet porphyroblasts
containing complex suites of mineral inclusions present an additional opportunity, with
thermobarometry on the hosts and their inclusions yielding complex information that may be
interpreted as a simple P–T progression within a single orogenic event (e.g St-Onge 1987) or
as overprinted conditions associated with multiple events or tectonic processes (e.g., Dorfler
et al. 2014). As such, the most appropriate course for thermobarometry generally begins with
careful consideration of mineral texture and mineral zoning (e.g., Essene 1989; Kohn and
Spear 2000) with garnet often at the heart of this endeavor.
Forward models of phase equilibria. A powerful use of equilibrium thermodynamics

in the understanding of metamorphic processes involves the calculation and interpretation of
phase equilibria, and much modern petrochronology involves the coupling of geochronological
methods with phase equilibria constraints. Prediction of the proportion and composition of
each major phase in a rock, and how these would change as functions of pressure, temperature
and system composition (in so-called ‘pseudosections’, ‘isochemical phase diagrams’ or
‘mineral assemblage diagrams’ such as Fig. 4) has had enormous influence on the petrologic
community. This has been reviewed recently by Spear et al. (2016) and in this volume by
Yakymchuk et al. (2017, this volume), and here we focus on the primary value from a garnet
petrochronology standpoint: revealing how garnet growth or dissolution would be expected
along any P–T path, how this might modify the residual rock composition (discussed also
in detail by Lanari and Engi 2017, this volume), how garnet chemistry would evolve during
growth, and how this is associated with changes in the co-existing assemblage, including the
production or consumption of other minerals, fluids, or melt.
Most studies of garnet in crustal metamorphism are concerned with pelitic, greywacke,
basaltic or carbonate lithologies, in which case reasonable approximations can be made by
considering a thermodynamic system comprised of Na2O, K2O, CaO, FeO, MgO, Al2O3,
SiO2, H2O, ± MnO, ± TiO2, ± Fe2O3, ± CO2. Incorporation of all or most of these components
permits description of garnet as (Fe2+,Mg,Ca,Mn)3(Al,Fe3+)2Si3O12 and calculation of
phase equilibria involving most major phases that this garnet is likely to co-exist with. The
thermodynamics of MnO-bearing mineral end-members are generally less well-constrained
than others, but MnO plays such a significant role in stabilizing garnet at low P–T conditions
and on the modal proportion of garnet at higher P–T conditions (Symmes and Ferry 1991;
Mahar et al. 1997; Tinkham et al. 2001; Caddick and Thompson 2008; White et al. 2014b)
that its inclusion is often warranted. A plethora of relevant mineral, fluid and melt models
have been described, with most recent contributions associated with various updates of the
‘thermocalc’ end-member dataset (Holland and Powell 1985, 1990, 1998, 2011). Even
models for compositionally well-constrained minerals such as garnet have been reassessed
numerous times (inlcuding but not limited to Wood and Banno 1973; Engi and Wersin 1987;
Powell and Holland 1988; Hackler and Wood 1989; Koziol and Newton 1989; Berman 1990;
Ganguly et al. 1996; Mukhopadhyay et al. 1997; White et al. 2000, 2005, 2014a,b; Stixrude
and Lithgow-Bertelloni 2005; Malaspina et al. 2009). Seemingly minor updates to solid
solution models can lead to significant changes in calculated phase equilibria unless end-
member thermodynamic properties are modified accordingly (see discussion in White et al.
2014a). Furthermore, the relative stabilities and coexisting compositions of complex mineral
solutions may respond inappropriately if applied beyond the limits (in composition, pressure
or temperature) of their original calibration, though it can often be difficult to ascertain
Gar – Bio
whether this is the case. It is thus informative to make direct comparison between K Mg – Fe
Gar – Bio
derived from the mineral compositions predicted by phase equilibria calculations and K Mg – Fe
derived from experimental calibrations such as Ferry and Spear (1978) across a range of P–T
conditions and for a range of bulk-rock compositions (and thus additional buffering minerals
in the calculated equilibria). Such comparisons, (e.g., Figure 6 of Caddick and Thompson
2008) highlight that, although experimentally determined partitioning is well recovered at
many P–T conditions, it is poorly fit elsewhere, and reveal some unexpected consequences of
utilizing complex activity-compositions models for numerous phases.
Comparison between calculated and natural garnet composition is one of the primary
constraints on segments of P–T paths inferred from pseudosection calculations (Konrad-
Schmolke et al. 2006; Gaidies et al. 2008a; Chambers et al. 2009; Cutts et al. 2010; Vorhies and
Ague 2011; Hallett and Spear 2014a; Mottram et al. 2015). Many factors can lead to calculated
compositions that fail to intersect at P–T conditions consistent with other observations,
including inappropriate assumptions of bulk composition or short length-scale compositional

heterogeneity (e.g., Tinkham and Ghent 2005; Kelsey and Hand 2015; Guevara and Caddick
2016; Palin et al. 2016; Lanari and Engi 2017, this volume), the potential for crystal nucleation
and growth at conditions removed from sample-wide thermodynamic equilibrium (e.g., Kretz
1973; Waters and Lovegrove 2002; Gaidies et al. 2011; Pattison et al. 2011; Kelly et al. 2013a,b,
b; Carlson et al. 2015b) and the possibility of diffusive modification of crystal composition
following growth (discussed in more detail below). Several recent approaches have focused more
explicitly on garnet, developing computer codes that compare measured natural crystal zoning
with thermodynamic predictions to automatically search for the most appropriate P–T path
experienced, generally also considering modification of the bulk-rock composition upon garnet
growth (e.g., Moynihan and Pattison 2013; Vrijmoed and Hacker 2014; Lanari et al. 2017). To
some extent these are natural successors to the pioneering Gibbs method (Spear and Selverstone
1983; Spear et al. 1991; Menard and Spear 1993) and to calculations that forward model the
chemical zoning established along prescribed P–T paths (e.g., Florence and Spear 1991; Gaidies
et al. 2008a,b; Konrad-Schmolke et al. 2008; Caddick et al. 2010).
Given the relative ease of calculating phase equilibria, what is their value? As
emphasized recently (e.g., Pattison 2015; Spear et al. 2015, 2016), the ability to calculate
diagrams may have exceeded our capability to assess their predictions. For example, if a
pseudosection predicts a 5 ºC, 0.5 kbar window within which a garnetiferous assemblage is
stable in a relatively dry lithology, it may be unlikely that this assemblage will actually be
found at these P–T conditions in nature due to (i) uncertainties in the thermodynamic data,
and (ii) possible kinetic effects that may hinder sample-scale equilibration on the timescale
that the rock passes through the P–T conditions of the field (Carlson et al. 2015b). This can
be closely associated with the concept that the volume of any rock in mutual thermodynamic
equilibrium (i.e., the equilibrium length-scale) probably changes throughout metamorphism
as a function of P, T and fluid availability (e.g., Stüwe 1997; Guiraud et al. 2001; Carlson
2002). In fact, the equilibrium length-scale may be different for each element depending on
the relative diffusive transport rate (in turn a function of elemental partitioning/solubility
into the intergranular transporting medium; Baxter and DePaolo 2002b; Carlson 2002). The
choice of appropriate bulk composition for a pseudosection calculation is thus often not
trivial and this composition may have to change as subsequently refractory phases such as
garnet (e.g., Spear et al. 1991; Marmo et al. 2002; Tinkham and Ghent 2005; Caddick et
al. 2007; Konrad-Schmolke et al. 2008), hydrous fluid (e.g., Guiraud et al. 2001) or melt
(e.g., Tajčmanová et al. 2007; Yakymchuk and Brown 2014; Guevara and Caddick 2016) are
produced and fractionated from the reactive system. Lanari and Engi (2017, this volume)
review bulk composition in more detail, demonstrating that use of an XRF-based rock
composition to retrieve P–T conditions of zoned garnet growth, without modification of that
composition to account for components sequestered by garnet, can produce significantly
erroneous estimates if more than ~ 2 vol% garnet has been produced. Eventually the length-
scales of equilibration may become small enough to establish domainal textures such as
coronae, in which substantial chemical potential (µ) gradients are preserved over geological
timescales and can best be interpreted through µ–µ equilibria (e.g., White et al. 2008; Štípská
et al. 2010; Baldwin et al. 2015). The retention of chemical zoning in garnet is one obvious
indicator that µ gradients are commonly maintained in metamorphic rocks.
Despite these various complications and limitations, equilibrium thermodynamic models
are of immense importance to garnet petrochronology and to metamorphic petrology more
broadly, as testified by the fact that the combined citations of the main papers describing
the thermocalc, theriak-domino and Perple_X programs (e.g., de Capitani and Brown
1987; Powell and Holland 1988; Connolly 1990; Powell et al. 1998; Connolly 2005, 2009; de
Capitani and Petrakakis 2010) and the thermocalc dataset (e.g., Holland and Powell 1990,
1998, 2011) significantly exceed 10,000 at the time of writing. Recent successes with respect
to application within garnet petrochronology are given at the end of this chapter, but we would
like here to highlight that the potential power of pseudosections in garnet petrochronology
(albeit without the title) was elegantly demonstrated ~25 years ago when Vance and Holland
(1993) coupled garnet compositions from the Gassetts Schist rocks that had previously been
element mapped by Thompson et al. (1977; e.g., Fig. 3) with pseudosection constraints on the
P–T path of garnet growth and isotopic dating of that garnet. This early adoption of garnet
petrochronology led to estimations of heating and decompression rate, and to inferences about
tectonic and thermal mechanisms for this metamorphism.
The loss of information. Compositional thermobarometry and comparison of
pseudosections with natural mineral assemblages rely on the fundamental assumption that
rocks faithfully record information associated with equilibration at a set of P–T conditions,
often assumed to be peak T. However, a substantial body of empirical (e.g., Anderson and
Olimpio 1977; Woodsworth 1977; Yardley 1977), experimental (e.g., Elphick et al. 1981;
Loomis et al. 1985; Chakraborty and Ganguly 1991; Ganguly et al. 1998a; Vielzeuf and
Saúl 2011) and modeled (e.g., Carlson 2006; Chu and Ague 2015) evidence suggests that
major element zoning in garnet will relax at high temperature through volume diffusion.
Furthermore, detailed observations indicate that exchange and net-transfer reactions can
operate on the rims of garnet crystals after peak metamorphic conditions and that these can
substantially modify the apparent P–T conditions recovered from mineral thermometry (e.g.,
Tracy 1982; Spear and Florence 1992; Kohn and Spear 2000; Pattison et al. 2003; Spear
2004). An illustrative example comes from a transect through the western Himalaya, where
oxygen isotope thermometry and phase equilibria constraints (Vannay et al. 1999) reveal far
higher temperatures than previously inferred for the same samples through major element
garnet-based thermobarometry (Vannay and Grasemann 1998). The implication is that garnet–
biotite compositions were reset by Mg–Fe exchange during cooling from peak temperatures
of ~750 ºC to ~600 ºC, where exchange and diffusion became sufficiently sluggish to inhibit
further substantial modification (effectively representing a closure temperature).
Diffusional resetting of metamorphic systems is covered in depth in this volume (Kohn
and Penniston-Dorland 2017, this volume) and later in this chapter we discuss how diffusional
partial resetting can be used to constrain heating and cooling timescales (i.e., geospeedometry).
But here, we must consider the conditions at which compositional information in garnet is altered
due to volume diffusion, thus compromising geothermobarometry. This is particularly pertinent
to petrochronology because garnet is such an important constituent of many thermobarometers
and major element diffusion within the garnet lattice is typically considered to be slower than in
many other major silicate minerals in metamorphic rocks (Brady and Cherniak 2010). Numerous
studies have thus sought to quantify the extent to which garnet zoning established during
prograde growth will be modified at subsequent stages of metamorphism, with several trying
to generalize results for a range of conditions (Florence and Spear 1991; Gaidies et al. 2008a;
Caddick et al. 2010). Figure 5 is an example that shows the extent to which garnet prograde
zoning will be preserved as a pelitic rock heats and is buried along the indicated P–T path,
exploring a wide range Tmax, heating rate and eventual crystal sizes. It indicates, for example,
that if a crystal nucleates at thermodynamically-constrained ‘garnet-in’ and grows spherically to
1 mm diameter during ca. 6 Ma of heating to TMax of 580 ºC (e.g., by following path ‘b’ in Fig. 5,
Mg Ca Fe
equating to ~25 ºC/Ma), its core will effectively preserve initial X Gar and X Gar contents but X Gar
Mn
(not shown) and X Gar (not shown) will have been modified by more than 1 mole fraction unit
due to volume diffusion subsequent to growth. Crystals achieving a 5 mm diameter along the
same prograde path will retain their initial compositions at Tmax. If heating continues at the same
rate to 750 ºC, the core composition of crystals exceeding 1 cm diameter will have been partially
modified and crystals smaller than 500 µm may preserve little or no major element zoning.
Slower heating rates increase the crystal sizes for which crystal core compositions are lost (e.g.,
paths ‘c’ and ‘f’ in Fig. 5), while fast heating tend to preserve growth information (e.g., path ‘a’).
a) 18
5 Gt-Pl-Bt- b) 850 Fe 2000 µm
17
Ms-Q-Ru 1500 µm
16 3 Gt-Pl-Bt-
(Melt-in) 800 1000 µm
15 Ms-Pa-Q-Ru-
14 (Sph Zo) 6
13 1 Gt-Bt-Ms- 750
Temperature (°C)
12 Pa-Zo-Sph-Q- 6 Gt-Pl-Bt-
Pressure (kbar)
11 (Ab Pl) Ms-Q-Ru-Melt path 'a' path 'c'

700
10 (Ky-in) path 'b' 500 µm
9 5
8 650
7 4 Gt-Pl-Bt- 10000 µm
6 Ms-Pa-Q-Ru 'd' 'e'
600 5000 µm 'Path f'
5 (Zo-out)
3000 µm
4 2 Gt-Pl-Bt- 2000 µm
1500 µm
3 Ms-Pa-Sph-Q- 550 4 1000 µm
(Zo-out)
2 3 500 µm
1 500
400 450 500 550 600 650 700 750 800 850 0 10 20 30 40 50
Temperature (°C) Time (Ma)
c) 850 Ca 1500 µm d) 850 Mg 1500 µm

1000 µm
800 1000 µm 800
6 6
750 750 500 µm

Temperature (°C)
Temperature (°C)
path 'a' path 'b' 500 µm path 'a' path 'b' path 'c'
700 path 'c' 700
5 5
10000 µm 10000 µm
650 650
5000 µm 5000 µm
'd' 'e' 'd' 'e' 500 µm
600 3000 µm 'Path f' 600 3000 µm 'Path f'
2000 µm 2000 µm
1500 µm
1000 µm 1500 µm
550 550 1000 µm
4 4
3 3 500 µm
500 µm 1000 µm
500 500
0 10 20 30 40 50 0 10 20 30 40 50
Time (Ma) Time (Ma)
Figure 5. Garnet opening to resetting upon heating, modified from Caddick et al (2010). Phase stabilities
in a pelitic rock were calculated along the burial and heating path shown in panel A, which also highlights
significant equilibria encountered. Equilibrium garnet compositions at successive points along that path
were used to establish growth zoning, and diffusional relaxation of this zoning was calculated between
each successive point. Along this P–T path, a rock encounters reactions in the same order regardless of
heating rate, which only controls the time available for diffusion at each temperature step. This crystal
growth and diffusion model was run for ~ 13,000 variations of heating rate, maximum temperature and
crystal size at that TMax. Panels B-D summarize results of these calculations, contoured to show the TMax–
heating-duration–crystal-diameter relations for which garnet crystals first ‘open’ sufficiently for diffusion
to noticeably modify the crystal core composition. Note that results concern prograde metamorphism only,
simply showing which combination of parameters will retain initial garnet crystal core growth composi-
tions upon reaching TMax and which will be modified by diffusion (solid curves, labeled for crystal diam-
eter). Horizontal gray lines numbered 3-6 indicate the assemblage phase boundaries traversed in the model
P–T path shown in A. Dashed curves show the conditions for which diffusion becomes sufficiently effec-
tive to completely eradicate zoning in garnet. See Caddick et al (2010) for additional details.
A consequence is that crystals may appear to preserve prograde zoning patterns, such
as bell-shaped Mn profiles, up to granulite-facies conditions but the absolute composition at
all interior points within the crystal may have been substantially modified from their initial
growth composition, rendering thermobarometry misleading. These resetting estimates are
somewhat faster than previous estimates (e.g., Florence and Spear 1991), partly because of the
use of different diffusivity data. Discretization of garnet porphyroblasts into sub-grains that
may not be obvious optically will further reduce timescales of diffusional homogenization,
sometimes substantially so (e.g., Konrad-Schmolke et al. 2007). This loss of compositional

information remains both a challenge and an opportunity for garnet petrochronology, reducing
the sensitivity of P–T estimates but providing additional methods for inferring duration, as
summarized in the geospeedometry section, below.
The strength of garnet: Geo-ba-Raman-try2. Previously discussed methods of
thermobarometry rely on phase compositions that are generally assumed to have equilibrated
near metamorphic peak temperature and/or pressure and to have experienced limited subsequent
modification. These assumptions can be difficult to assess in natural samples, so alternative
methods that make fewer (or at least different) assumptions are valuable. One of the most
promising of these relies on a very different property of garnet: its resistance to elastic deformation.
Minerals trapped as inclusions during metamorphic growth of porphyroblasts generally
experience approximately lithostatic pressure at the time of entrapment. Rocks expand
(decompress) upon exhumation and undergo phase transformations accordingly, but mineral
inclusions can retain a substantial proportion of the pressure at which they were trapped if
the host phase can resist the deformation required to permit expansion of the inclusion, and if
pressure is not ‘lost’ through development of cracks in the host. Identification of deformation
around inclusions in high pressure diamond (Sorby and Butler 1869; Sutton 1921) first led to
the concept of thermobarometry by study of stress in minerals hosting inclusions (Rosenfeld
and Chase 1961). Focus turned to garnet porphyroblasts (Rosenfeld and Chase 1961; Rosenfeld
1969) because phases with low thermal expansivity and high bulk and shear moduli make good
natural pressure vessels: their inclusions, which would naturally expand upon exhumation, are
likely to retain a substantial fraction of their entrapment pressure (Zhang 1998; Izraeli et al.
1999; Guiraud and Powell 2006). A classic example comes from preservation of coesite and
palisade quartz inclusions in garnet porphyroblasts, which led to the first identification of
continental rocks that had been subducted to > 90 km (Chopin 1984) based on the inference
that coesite once existed as the stable SiO2 polymorph but was pervasively inverted to quartz
upon exhumation unless trapped as inclusions within garnet.
Substantial literature suggests that residual pressure should also be maintained in rocks
lacking the obvious palisades inclusion textures, leading to studies that recovered ultra-high
pressure conditions by measuring pressure-sensitive laser Raman peak positions of coesite
and olivine inclusions in diamond and garnet (Izraeli et al. 1999; Parkinson and Katayama
1999; Sobolev et al. 2000). Correlations between pressure and the laser Raman spectra of
several phases have been developed, due partly to the need for calibrants in high-pressure
experiments (e.g., Hemley 1987; Schmidt and Ziemann 2000; Schmidt et al. 2013). Residual
pressure (i.e., current pressure felt by a mineral inclusion, which may be substantially different
to the pressure felt by crystals in other textural settings) can thus now be determined for these
phases in situ with Raman techniques. However, a mineral inclusion with pressure sensitive
Raman characteristics only makes an effective thermo-barometer in natural samples if its
elastic properties contrast sufficiently with its hosting phase (otherwise the initial pressure
felt by the inclusion may be lost as it deforms its host). A detailed survey of candidate mineral
pairs (Kohn 2014) revealed several promising combinations, with quartz-in-garnet among the
most sensitive barometers because of the relative compressibility of quartz and rigidity of
garnet. Additionally, zircon inclusions in garnet might make a very sensitive thermometer
(Kohn 2014), with uncertainties of quartz-in-garnet and zircon-in-garnet thermobarometry
potentially as small as several hundreds of bars and tens of degrees, respectively.
The quartz-in-garnet laser Raman barometer was applied to quartz–eclogite, epidote–
amphibolite and amphibolite facies metamorphic samples by Enami et al. (2007), who used
a simple numerical model (Van der Molen 1981) to infer original metamorphic pressure from
2
Kudos to editor Kohn for coining the term Thermoba-Raman-try (Kohn 2014).
measured residual pressure and reveal the high pressure history of samples that otherwise
preserve little evidence of eclogite-facies equilibration. The need for more sophisticated modeling
approaches has since been identified (Angel et al. 2014; Ashley et al. 2014a; Kohn 2014; Kouketsu
et al. 2014) and available software now offers automated calculation with several choices of
elastic model and simple correction of elastic properties for garnet composition (Ashley et al.
2014b). In an important validation of this methodology, application to experimentally derived
quartz inclusions in garnet yielded “an entrapment pressure at 800 ºC of 19.880 kbar—essentially
identical to the experimental pressure of 20 kbar” (Spear et al. 2014).
Application of Raman barometry (or ba-Raman-try) to high pressure lithologies has
retrieved pressures of quartz or coesite trapping (e.g., Korsakov et al. 2010; Zhukov and
Korsakov 2015), analysis of more complex suites of inclusions has revealed P–T paths of
subduction zone garnet growth (Ashley et al. 2014a), and comparison with thermodynamic
modeling has indicated the likely extent of overstepping the garnet isograd in subduction
zone and Barrovian-sequence samples (Spear et al. 2014; Castro and Spear 2016). However,
application to high temperature metamorphic rocks remains challenging, with Sato et al.
(2009) describing Raman shifts which imply that the interface between quartz inclusion
and garnet host can be subjected to tensile stress upon cooling and that the quartz inclusions
thus preserve negative residual pressure, as discussed further by Kouketsu et al. (2014).
Ashley et al. (2015) detailed an extreme case in which this effective under-pressuring led to
polymorphic transformation of quartz inclusions to cristobalite. A further challenge for high
temperature metamorphic rocks is that modeling currently relies on an assumption that only
elastic deformation acted on the host and inclusion, raising substantial problems for rocks that
may have experienced significant plastic deformation. Indeed, given that chemical diffusion
in garnet could either establish pressure gradients after crystal growth (Baumgartner et al.
2010) or could act as a mechanism to reduce pressure variations, garnet crystals experiencing
substantial intra-crystalline diffusion after inclusion of quartz probably make unreliable
pressure vessels without additional consideration and modeling. However, for the many
cases in which diffusive re-equilibration is minimal, thermoba-Raman-try is an exciting, and
complementary, method of inferring pressure–temperature evolution during garnet growth.
An exciting future development involves the combined application of multiple systems.
In particular, given that quartz-in-garnet and zircon-in-garnet respectively act a barometer
and a thermometer (Kohn 2014), we note that their combined use would be compelling. This
would raise the possibility of extremely detailed petrochronology in samples containing
garnet (whose P–T and duration of growth can be determined chemically and with zoned
isotope geochronology) with inclusions of both zircon (providing additional P–T information
with Raman and trace element thermometry and additional time constraints with U–Pb
geochronology) and quartz (providing sensitive constraint on evolution of P during garnet
porphyroblast growth). This represents a very powerful suite of techniques for deciphering
rates of change of pressure and temperature in the mid crust, and is clearly an important
potential avenue for further development.
Garnet as a tracer of reaction pathways and fluid–rock interaction
Thus far, we have focused on the use of garnet as a monitor of the P–T evolution of the
systems wherein it grows. In the following section we move from the ‘P–T’ to the ‘X’—
or composition—of the system during garnet growth. As system chemistry changes, often
reflecting the role of fluids or the local production or consumption of key phases, garnet crystals
can record aspects of that change especially in their trace element and isotopic zonation.
Stable Isotopes in Garnet. Oxygen is the most abundant element in the Earth’s crust and a
main component of the fluids that transport heat and mass during crustal metamorphism. Oxygen
isotopes in whole rocks and minerals have been used to characterize the composition, relative
timing, and sources of these metamorphic fluids (e.g., Bickle and Baker 1990; Baumgartner
and Valley 2001). In cases where the fluid history is complex, distinct garnet growth zones may
be associated with evolving fluid compositions and sources. Additionally, the oxygen isotopic
values of co-existing mineral pairs have been utilized as a thermometer due to the measurable
and temperature-dependent fractionation between different minerals (e.g., Chacko et al. 2001;
Valley 2001; Valley et al. 2003). In this section, we review the use of oxygen isotopes in garnet as
a marker of open system fluid flow, local reaction history, and as a crustal thermometer.
Historically, a principal objective of stable oxygen isotope studies was to determine peak
temperatures by utilizing the temperature-dependent fractionation of oxygen isotopes between
coexisting mineral pairs, determined either experimentally or empirically (Taylor and Sheppard
1986; Richter et al. 1988; Clayton et al. 1989; Zheng 1991, 1993; Kohn and Valley 1998; Chacko
et al. 2001; Valley et al. 2003). The principles of equilibrium fractionation factors (Chacko et
al., 2001) and stable isotope thermometry (Valley 2001) have been reviewed in great detail in
a previous volume of this series, so only more recent contributions, specifically in relation to
its application with garnet (∆18O mineral–garnet thermometry), will be discussed briefly here.
Quartz–garnet oxygen isotope pairs have been used most often to constrain metamorphic
temperatures (Rumble and Yui 1998; Peck and Valley 2004; Moscati and Johnson 2014). Peck
and Valley (2004) utilize garnet–quartz pairs for thermometry of quartzites from the southern
Adirondacks, NY. Conceptually, if most of the oxygen in the quartzites is in the quartz, any effect
of oxygen isotope exchange will be recorded in the δ18O of the minor phase, in this case garnet.
Refractory accessory minerals like garnet are closed to oxygen diffusion during growth (Valley
2001), and as quartz in a quartzite acts as an infinite reservoir for oxygen diffusion, this mineral
pair can be retentive of the peak temperature isotope fractionation, even if the rock is slowly
cooled (Kohn and Valley 1998). Metamorphic temperatures of ~700−800 ºC were calculated from
the quartzites from the southern Adirondacks, with no grain size dependence, suggestive of slow
oxygen diffusion in garnet and closure temperatures of at least 730 °C (Peck and Valley 2004).
The analysis of oxygen isotopes in garnet continues to benefit from increased accuracy
and precision related to the improvement of in-situ analytical techniques (Vielzeuf et al.
2005; Kita et al. 2009; Page et al. 2010), and also from texturally constrained samples (see
Electronic Appendix for a discussion on the historical challenges related to analysis of δ18O
in garnet). When quartz is fully armored by a mineral with slow diffusion of oxygen, such
as garnet, then any diffusive exchange of oxygen between quartz and host during cooling is
minimal. Hence, garnet–quartz mineral pairs as determined by in-situ techniques can offer
a detailed temperature and fluid (δ18O) history (Strickland et al. 2011). As Russell et al.
(2013) note, if combined with constraints on the timing (via garnet geochronology) and P–T
conditions (via phase equilibria modeling, or even Raman barometry of quartz in garnet) of
garnet growth, then a robust P–T–t–f metamorphic history is attainable.
Without metasomatism at elevated pressures and temperatures, metamorphic rocks will
tend to inherit the bulk rock oxygen isotopic composition of the protolith (Fig. 6). This can
be recorded in the δ18O of the individual minerals, though an accurate knowledge of the
equilibrium isotope fractionation factors among coexisting phases is required in order to
properly interpret these measured isotope compositions. Garnet has one of the lowest δ18O of
all minerals, lower than that of bulk rock δ18O, and thus increases with increasing temperature
(Kohn 1993) . In some lithologies (e.g., metabasites), garnet δ18O is close to the whole rock
and can encode the δ18O of the whole rock (Putlitz et al. 2000). Any deviation in garnet δ18O
from initial values can be produced by processes including a) change in temperature inducing
changing equilibration fractionation factors (see above), b) changing mineral assemblages
and modal proportions during progressive metamorphism (Kohn 1993; Young and Rumble
1993), and c) open system change in the reactive δ18O composition of the bulk rock from the
seawater
siliceous ooze
WHOLE ROCKS
carbonate ooze
pelagic clays
altered basalt
gabbros
mantle peridotites
limestones
shales
I-type granites
S-type granites
eclogites (1)-(11)
GARNETS
pelites (12)-(24)
skarns (25)-(30)
amphibolites (31)-(33)
calcsilicates (34)-(36)
felsic/intermediate gneisses (37)-(38)
meteoric water other (39)-(42)
as low as -65‰
-20 -10 0 10 20 30 40
δ18O
Figure 6. Compilation of δ18O data of whole rock and garnet analyses. Whole rock analyses and average
mantle value (δ18O = 5.80‰) from Eiler (2001). References to garnet data include: (1)(31)Errico et al (2013),
(2)
Martin et al (2014), (3)(33)Page et al (2014), (4)Rubatto and Angiboust (2015), (5)Rumble and Yui (1998),
(6)
Russell et al (2013), (7)Zheng et al (1998), (8)Burton et al (1995), (9)Chen et al (2014), (10)(12)Gordon et al
(2012), (11)(24)Masago et al (2003), (13)Kohn and Valley (1994), (14)Kohn et al (1993), (15)Kohn et al (1997),
(16)
Lancaster et al (2009), (17)(32)(34)Martin et al (2006), (18)Martin et al (2011), (19)Raimondo et al (2012), (20)
Skelton et al (2002), (21)van Haren et al (1996), (22)Young and Rumble (1993), (23)(30)(36)(39)Ferry et al (2014),
(25)
Crowe et al (2001), (26)Clechenko and Valley (2003), (27)D’Errico et al (2012), (28)Jamtveit and Hervig
(1994), (29)Page et al (2010), (35)Sobolev et al (2011), (37)Gauthiez-Putallaz et al (2016), (38)Vielzeuf et al
(2005), (39)Peck and Valley (2004), (41)Moscati and Johnson (2014), (42)Abart (1995).
interaction of externally-derived fluids (Kohn and Valley 1994). Because oxygen diffusion in
garnet at crustal conditions is sluggish (Coghlan 1990; Burton et al. 1995; Vielzeuf et al. 2005),
any change in garnet δ18O during its growth can be preserved in its concentric zonation (Kohn
et al. 1993; Skelton et al. 2002), in one case even after a regional granulite facies overprint
(Clechenko and Valley 2003). These effects on garnet δ18O are further discussed below.
Kohn (1993) developed a model showing the effect of net transfer reactions during prograde
heating for systems that are closed to externally-derived fluids. It was predicted that the fractional
crystallization of (low δ18O) garnet during metamorphism would result in an increase in the δ18O
of garnet from core to rim. However, this effect was calculated to be minor (~1‰ or less over
150 ºC of heating) and independent of bulk composition (Kohn 1993). This was further tested in
natural samples from regionally metamorphosed rocks from Chile (Kohn et al. 1993), where the
difference between garnet cores and rims from metapelites and amphibolites was 0.5‰ and <
0.1‰, respectively. These observations suggest that garnet growth occurred in a system closed
to infiltration of fluids out of δ18O equilibrium with the bulk rock, and are similar to observations
from regionally metamorphosed amphibolites and schists from Vermont (Kohn and Valley 1994)
and New Hampshire, U.S.A (Kohn et al. 1997). Comparably, the fractionation of δ18O due to
(internally-derived) devolatilization of a meta-basalt was shown to be small, with garnet rims
having a δ18O less than 1‰ compared to garnet cores (Valley 1986; Kohn et al. 1993).
The mechanism most likely to result in large changes in garnet δ18O from core to rim is
open system fluid infiltration. A large amount of work on δ18O in garnet has focused on its
ability to elucidate the source of externally derived fluids in various metamorphic settings. The
infiltration of these fluids, often over substantial thermal and chemical gradients, can cause
significant zoning of δ18O in garnet, offering a relative time marker for fluid flow (Skelton
et al. 2002), especially when the protolith composition differs significantly from mantle-like
compositions (~5−6‰; Fig. 6), or when the fluids are derived from devolatilization of high
δ18O sediments (Errico et al. 2013), interaction with low δ18O meteoric water (Russell et al.
2013; Fig. 6), or rehydration in a mid-crustal shear zone (Raimondo et al. 2012).
Russell et al. (2013) examined the δ18O record of garnets from a series of orogenic
eclogites and found that these crystals can display a range of δ18O that falls well outside that
expected for mantle-derived protoliths. While the intracrystalline variation of δ18O in these
garnets is small (increase of ~1−2‰ between core and rim), the absolute δ18O values record
the origin of these eclogites, distinguishing between those derived from altered, oceanic upper
crust (Trescolmen, Alps; 7.7 to 8.3‰) and high-temperature infiltration of meteoric water
into mafic intrusions buried in-situ with subaerial, continental crust (Western Gneiss Region,
Norway; −1.2 to −0.2‰). Indeed, extremely negative δ18O values in eclogitic garnet from the
Dabie (−10‰; Zheng et al. 2006) and Kokchetav (−3.9‰; Masago et al. 2003) terranes have
been interpreted as evidence of extreme infiltration of meteoric water, possibly derived from
a rift environment involving low δ18O glacial melt water. Rumble and Yui (1998) analyzed
eclogitic garnets from the Qinglongshan ridge in the Sulu terrane, and found bulk garnet δ18O
values as low as −11.1‰ (Fig. 6), also attributing such negative values as reflecting alteration
of the protolith by meteoric waters at high paleo-latitudes and/or –altitudes prior to subduction.
Other low garnet δ18O values are observed in several studies from hydrothermal skarns where
variable fluid sources in open systems are recorded in zoned garnet (Jamtveit and Hervig 1994;
Crowe et al. 2001; Clechenko and Valley 2003; D’Errico et al. 2012). D’Errico et al. (2012)
observed a large range in intragrain δ18O values (−4 to 4‰) from a hydrothermal system in the
Sierra Nevada batholith, attributing early, low δ18O values to meteoric water input, with higher
values resulting from variable mixing of magmatic and metamorphic fluids.
Some examples offer instances where garnet δ18O records changing fluid compositions
(e.g., Kohn and Valley 1994; Martin et al. 2011; Errico et al. 2013; Page et al. 2014; Rubatto and
Angiboust 2015). Many of these cases of strongly zoned garnet δ18O are recorded in eclogites.
Rubatto and Angiboust (2015) observed eclogitic garnet cores from the Monviso ophiolite
in the Western Alps with δ18O values of 0.2−2.0‰, recording ocean floor metasomatism of
the basaltic protolith. High-pressure metasomatism is recorded in the garnet rims, with δ18O
values of 3.5−6.0‰. While these rocks are found in a shear zone adjacent to down-going
lithospheric mantle, Monviso serpentinites (δ18O of 3.0−3.6‰) were not found to be the
source of the metasomatizing fluid; high δ18O fluids from dehydrating metasediments were
instead suggested to be the primary source for fluid influx at depth.
A final example of garnet δ18O recording multiple metasomatic events comes from two
separate findings from the well-studied Franciscan Complex (Errico et al. 2013; Page et al.
2014). Garnet crystals from eclogite and amphibolite blocks record contrasting fluid flow
histories (Errico et al. 2013). Amphibolites, which are inferred to record an early subduction
initiation history (Anczkiewicz et al. 2004), are metasomatized by an early influx of
sediment-derived fluids, recorded in their garnet rims (~8‰; Errico et al. 2013; see Fig. 6).
Eclogitic blocks have high δ18O garnet cores (~6−11‰), interpreted to record extreme ocean
floor alteration of the protolith, with fluid-mediated exchange with the overlying mantle
wedge at depth recorded in low δ18O garnet rims (3−5‰; Errico et al. 2013). Page et al.
(2014) suggest that their observed shifts in garnet δ18O (2−3‰) are evidence of limited fluid
flow during burial where garnet rims are formed during metasomatism at high fluid/rock
ratios during exhumation/re-burial, with possible interaction with highly oxidized fluids.
In all of the cases shown above, analysis of garnet δ18O has elucidated the source
and relative timing of fluid–rock interaction in a variety of tectonic settings. Garnet δ18O,
combined with a detailed P–T and temporal record, has great potential to provide an important
marker for metamorphic fluid flow, especially in hydrothermal and subduction zone systems
where the role of exotic external fluids is most observable. Combining δ18O with studies of
fluid inclusions in garnet (where preserved) can provide an even more detailed fluid history
of burial/prograde (primary fluid inclusions) and exhumation/retrogression (secondary fluid
inclusions), coupling these techniques to elucidate the source and composition of fluids.
Increasingly, stable isotopes other than oxygen are utilized in metamorphic systems to
study the nature and scale of fluid flow, particularly in subduction zone systems, and relying
(mainly) on the interpretation of whole rock isotopic analyses. Stable isotopes of lithium,
calcium, and magnesium are ideal for studying the effects of diffusion due to their large
relative mass differences, and are likely to be particularly useful in natural systems where
strong chemical potential gradients exist. If garnet growth occurs in an environment with high
fluid flux or large chemical gradients, then large diffusion-driven mass fractionation may be
recorded in growing porphyroblasts, and in-situ or microsampled stable isotope analyses of
garnet can produce a detailed record of such fluid and mass transfer. The subduction interface,
where the overlying mantle wedge comes into contact with subducting metasediments and
metabasaltic rocks, is an ideal locale for such studies, given the strong chemical contrasts
in several chemical components between these lithologies. Studies that have focused on the
length-scales and timescales of subduction zone fluid flow have utilized isotope systems
including calcium (John et al. 2012), lithium (Zack et al. 2003; Penniston-Dorland et al.
2010) and magnesium (Pogge von Strandmann et al. 2015). Bulk isotopic analyses on garnet
separates have been limited to just a few studies (e.g., on magnesium; Pogge von Strandmann
et al. 2015). Notably also, Bebout et al. (2015) developed a technique to measure the variation
in δ7Li in garnet using SIMS, in order to determine whether the lithium isotopic record could
help elucidate the nature of fluid flow during subduction zone metamorphism of metasediments
from Lago Di Cignana, Western Alps. Finally, Cr (Wang et al. 2016) and Fe (Williams et al.
2009) isotopes have been utilized for studying evolving redox conditions (using Cr and Fe)
and the effects of partial melting processes (using Fe).
As these isotopes are primarily limited to measurement by solution ICP-MS or TIMS, bulk
analyses of garnet have been the primary focus of earlier studies. A promising new direction may
involve the use of SIMS for in-situ analysis, where applicable. However, detailed microdrilling
from texturally or compositionally constrained garnet growth zones, in tandem with various
stable isotope analyses (oxygen and otherwise), has the potential to link these geochemical
tracers to P–T-t constraints for various metamorphic (and igneous) processes, thus expanding the
incorporation of various stable isotope systems to the garnet petrochronologist’s toolbox.
Trace elements in garnet. Garnet is often zoned in trace elements, with cores that are
enriched in Y and HREE (Hickmott et al. 1987; Chernoff and Carlson 1999; Pyle and Spear
1999; Otamendi et al. 2002). These highly compatible elements tend to resist re-equilibration
at elevated temperatures owing to their low inter- and intragranular diffusivities (Lanzirotti
1995; Spear and Kohn 1996; Otamendi et al. 2002; Carlson 2012; Bloch et al. 2015). Thus,
trace element zoning can be preserved at upper amphibolite and even granulite facies, in which
major element zoning is often erased (Fig. 5), providing a useful tool for deciphering high-
temperature metamorphic processes (Spear and Kohn 1996; Hermann and Rubatto 2003). Early
garnet trace element studies required secondary ion mass spectrometry (Hickmott et al. 1987)
or synchrotron (Lanzirotti 1995) techniques. However, LA-ICPMS is increasingly now used
for trace element analyses at adequate spatial resolution and sensitivity for many applications.
The analyzed distribution of trace elements, specifically REEs, can be utilized for
interpretation of garnet geochronology (e.g., zonation of radioactive parent elements; see
later section), changes in garnet growth rate, and changes in reaction history, specifically
with relation to accessory phases. Often, trace element zonation is sensitive to different
specific mineral reactions or matrix processes than the major elements, thus offering
complementary information. Insofar as trace element zonation relates to dateable accessory
phases, trace elements in garnet represent an often crucial correlative part of accessory phase
petrochronology. Garnet can be vividly zoned in REE’s (see expanded discussion below), as
well as elements such as Zr (Anczkiewicz et al. 2007), Cr (Spear and Kohn 1996); Yang and
Rivers (2001); (Martin 2009; Angiboust et al. 2014), P (Spear and Kohn 1996; Chernoff and
Carlson 1999; Kawakami and Hokada 2010; Hallett and Spear 2015; Jedlicka et al. 2015;
Ague and Axler 2016), and As (Jamtveit et al. 1993). Here, we present a summary of the
controlling mechanisms for partitioning of trace elements in garnet.
Hickmott et al. (1987), Hickmott and Spear (1992), and Hickmott and Shimizu (1990)
were among the earliest contributions to interpret remnant trace element zoning in garnet (in
metapelites from the Tauern Window and Massachusetts), finding controls on trace element
concentrations related to elemental fractionation during progressive metamorphism, fluid–rock
interaction, and the breakdown of trace element-rich minerals. Numerous contributions have
built upon these earlier studies in attempting to decipher metamorphic processes and rock
reaction histories from garnet trace element zoning, with additional examples cited below.
Carlson (2012), Moore et al. (2013), Cahalan et al. (2014) and Carlson et al. (2014) have
sought to determine how various trace elements are structurally incorporated into natural
garnet, and to calibrate the rates and mechanism of diffusion of these elements in garnet,
utilizing samples collected from the aureole of the Makhavinekh Lake Pluton, Labrador.
Using lattice dynamic calculations, Carlson et al. (2014) found that the incorporation of trace
elements into garnet is likely dominated by menzerite- (for REE) and alkali-type (for Li
and Na) coupled substitutions at a low energetic cost. This energetic cost will (i) decrease
as the host garnet unit-cell increases, (ii) decrease as temperature increases or pressure
decreases, and (iii) decrease substantially with contraction of the ionic radius across the
lanthanide series. These observations, as well as diffusivities calculated through the use of
stranded diffusion profiles in the Labrador garnets (Carlson 2012) have been shown to have
profound implications for the interpretation of Lu–Hf garnet ages (Kelly et al. 2011; see also
later sections), the preservation of matrix trace element distributions during porphyroblast
crystallization (or overprint zoning; see examples below), element mobility, and intergranular
solubility (Carlson et al. 2015a). Finally, Cahalan et al. (2014) found that the diffusion of Li
in garnet is strongly governed by coupled substitution with slowly diffusing REE, and thus Li
zoning may be retained during metamorphism (even at elevated temperatures) and utilized as
a monitor of fluid– (and/or melt–) rock interaction.
Simple Rayleigh fractionation between a growing garnet crystal and the rock matrix
would result in the incorporation of Y + HREE into garnet cores, with smoothly decreasing
‘bell-shaped’ profiles of these elements towards garnet crystal rims due to their relatively high
garnet/matrix partition coefficients (e.g., Hollister 1966; Otamendi et al. 2002; Lapen et al.
2003; Anczkiewicz et al. 2007; Kohn 2009; Fig. 7). In contrast, MREE and LREE generally
display a slight increase in abundance towards garnet crystal rims, or show no zonation at
all. Garnet REE profiles can, however, deviate from these simple profiles, as observed in a
number of settings. This may be due to a number of factors including 1) diffusion-limited
REE uptake during prograde metamorphism, 2) resorption (or recrystallization), 3) syn-
growth breakdown of a REE-bearing accessory phase, 4) breakdown of REE-bearing major
phases, 5) change in the kinetics of garnet growth, 6) overprint zoning, and 7) infiltration of
trace element-rich fluid. Let us treat these each in turn.
Dragovic et al. (2016) Otamendi et al. (2002)

4 6000
A) Lu B) Y
ppm
ppm
0 0
rim core rim rim core rim
Dragovic et al. (2016) Skora et al. (2006)
24 160
C) Lu D) Lu
ppm
ppm
0 0
Cruz-Uribe et al. (2015) Moore et al. (2013)

25 140
E) Lu F) Lu
ppm
ppm
0 0
core rim rim core rim
Figure 7. Garnet trace element zonation from various studies. A. Example of Rayleigh fractionation;
modified from Dragovic et al (2016). B. Rayleigh fractionation example; modified from Otamendi et
al (2002). C. Lu enriched in garnet rim, possibly from breakdown of accessory phases; modified from
Dragovic et al (2016). D. Diffusion-limited REE uptake example; modified from Skora et al (2006). E.
Core to rim profile showing Lu annulus from breakdown of accessory phase; modified from Cruz-Uribe
et al (2015). F. Example showing intermediate peaks in Lu, possibly from change in local garnet-forming
reaction; modified from Moore et al (2013).
Diffusion-limited REE uptake. Diffusion of trace elements to a growing garnet crystal

can be relatively slow compared to the crystal’s growth rate (Skora et al. 2006; Moore et al.
2013). An early nucleating garnet crystal will sequester highly compatible elements (KD > 1),
forming a high concentration central peak in zoning profiles (Fig. 7). As a result of the slow
intergranular diffusion, a steep chemical potential ‘depletion halo’ will form around the early-
growing crystal (supply is slower than uptake). Increased temperature during progressive
metamorphism will increase intergranular mobility, relaxing the chemical potential gradient
of REE around the growing garnet crystal and possibly permitting a secondary REE peak in
the growing garnet. REE abundance in the crystal core is dependent on the KD (relative to
garnet precursor phases) and matrix concentration of each element, with HREE exhibiting
the strongest fractionation into garnet. The location and magnitude of the secondary peak will
depend on the intergranular diffusivity of the element and its dependence on temperature, with
MREE exhibiting a rimward secondary peak compared to the HREE, owing to their lower
intergranular diffusion rate (and higher ionic radii). For LREE, this will manifest itself as a low
abundance in crystal cores, potentially increasing at the crystal rim.
Resorption (/recrystallization). Annular maxima in trace element abundances may result

from the partial consumption of garnet during an orogenic cycle (i.e., during prograde heating/
burial or retrograde dissolution). At elevated temperatures, Y + HREE would be liberated from
the resorbed garnet rim. Elements that have a strong affinity for garnet will preferentially
re-partition into the remnant crystal rim, and subsequently diffuse inward at a rate controlled
partly by temperature (Lanzirotti 1995; Pyle and Spear 1999). Kelly et al. (2011), utilized Lu–
Hf garnet data from Makhavinekh Lake Pluton, Labrador, to suggest that the strong preferential
reincorporation of Lu relative to Hf results in false apparent Lu–Hf garnet ages. Using their
model, higher degrees of garnet resorption and Lu retention would result in younger apparent
ages and stranded diffusion profiles of Lu at remnant crystal rims. Gatewood et al. (2015)
make a similar observation based on Sm-enriched rims likely formed during resorption and
leading to young apparent Sm–Nd rim ages.
Interface-coupled dissolution–reprecipitation (ICDR) can also result in unique trace
element zoning. Ague and Axler (2016) present garnet Na, P, and Ti distribution maps from a
high-pressure felsic granulite from the Saxon Granulite Massif, showing sharply defined cross-
cutting chemical domains that likely record ICDR during retrograde fluid–rock interaction.
While major element zonation is absent at such temperatures (> 900 °C) due to intracrystalline
diffusion, retention of zonation of these trace elements highlights their far lower diffusivities
at these conditions (Ague and Axler 2016).
Breakdown of an accessory phase. The presence of accessory phases can buffer trace
element activities, thereby exerting strong controls on their partitioning into garnet. Several
studies have linked REE zoning in garnet with the growth and breakdown of phases such as
monazite, allanite, epidote, apatite, or xenotime (Hickmott and Shimizu 1990; Hickmott and
Spear 1992; Chernoff and Carlson 1999; Pyle and Spear 1999; Yang and Rivers 2002; Corrie
and Kohn 2008; Stowell et al. 2010; Gieré et al. 2011; Cruz-Uribe et al. 2015; Dragovic et
al. 2016; Engi 2017, this volume; Fig. 7). Pyle and Spear (1999) showed that in xenotime-
bearing, garnet-zone assemblages, the early presence of xenotime would buffer Y and result
in high-Y garnet cores. Continued growth of garnet would deplete the matrix in Y, eventually
resulting in loss of xenotime and a subsequent decrease in Y towards garnet rims. Yang and
Rivers (2002) also suggested that enrichment of certain REE in garnet can be associated with
the breakdown of particular minerals: enrichment in LREE associated with the breakdown of
allanite, in MREE with epidote, and in HREE with xenotime or zircon. Using trace element
zoning patterns in both garnet and tourmaline, Gieré et al. (2011) modeled the metamorphic
evolution of rocks from the Central Alps, relating this evolution to the growth and breakdown
of accessory and major phases. The presence of subhedral annuli enriched in Ca, Sr, Y, and
HREE was associated with the breakdown of allanite, while internal zones of these annuli
were associated with garnet growth coincident with the breakdown of detrital monazite.
Hallett and Spear (2014a) suggest that P zoning in garnets from the Humboldt Range, NV,
results from breakdown of a phosphate phase such as apatite, or melting reactions involving
the breakdown of plagioclase. Finally, the discontinuous breakdown of Y-rich mineral phases
such as epidote and allanite was invoked to explain Y annuli in garnets from metapelitic rocks
of the Black Hills, South Dakota (Yang and Pattison 2006).
Breakdown of major rock-forming phases. Konrad-Schmolke et al. (2008) used a path
dependent thermodynamic forward model and published trace element partition coefficients
to model major and trace element distribution in garnet, comparing modeled elemental
distributions to those observed in UHP garnets from the Western Gneiss Region, Norway.
The modeled trace element distributions result from a sequence of garnet-producing mineral
breakdown reactions (specifically the breakdown of chlorite, epidote, and amphibole) during
subduction, combined with progressive fractional modification of the effective bulk composition
upon garnet growth. Along an inferred subduction P–T path, changes in the calculated
abundance of major phases and in trace element partitioning between those phases and garnet
generated synthetic distributions of major and trace elements in garnet. HREE zoning patterns
are largely predicted by fractionation into the garnet core by a Rayleigh-type process, with
possible perturbations towards the garnet rim for some effective bulk compositions resulting
from breakdown of epidote and amphibole. The liberation of MREE and LREE into garnet
results from epidote- and amphibole-consuming reactions, and is expressed as intermediate
peaks in a growing garnet crystal. These peaks are shown to form progressively further from
the crystal core for decreasing atomic numbers (or garnet/matrix partition coefficient), as is
also described in Zermatt-Saas Fee garnets (Skora et al. 2006).
Konrad-Schmolke et al. (2008) also relate trace element distributions in garnet to possible
interpretations of garnet geochronological methods. Strong partitioning of radioactive parent
elements such as Lu and Sm into a growing garnet crystal will have significant effects on the
interpretation of a garnet age (see below). The models from Konrad-Schmolke et al. (2008) show
that most of the Lu partitioned into garnet during early growth comes from the breakdown of
chlorite. Therefore, Lu–Hf in garnet from a subducting lithology could date chlorite breakdown
reactions, with the Sm–Nd system possibly tracking the growth of garnet via the breakdown of
epidote or amphibole and thus the amphibolite/blueschist to eclogite facies transition.
Changes in the kinetics of garnet growth. For elements that strongly partition into
a growing garnet crystal (i.e., Y + HREE), a decrease in the garnet growth rate can lead to
an increase in the concentration of that particular element in the garnet’s outermost surface
(Hickmott and Spear 1992). The combined presence of these elemental annuli in a zone of
relatively inclusion-free garnet has been partly attributed to such decreases in garnet growth
rate (Lanzirotti 1995; Yang and Rivers 2001; Yang and Pattison 2006).
Overprint zoning. Pre-existing matrix heterogeneities or former accessory phases can be
retained as overprint zoning in the chemistry of garnet that grows in these locations (Menard
and Spear 1996; Spear and Kohn 1996; Yang and Rivers 2001, 2002; Kohn 2004; Vielzeuf et
al. 2005; Martin 2009; Carlson et al. 2015a). Martin (2009) shows sub-millimeter scale Y and
Cr zoning in garnet from central Nepal that defines an internal foliation, spiraling from the rim
to the core of crystals, similar to patterns often portrayed by inclusion trails. These garnets lack
well-defined inclusion patterns, so only the trace element zoning in garnet offers a diagnostic
tool for determining the relative timing of deformation, possibly even helping to determine the
sense of rotation of a porphyroblast relative to the matrix foliation.
Infiltration of trace element-rich fluid. Infiltration of an externally-derived fluid whose
trace element concentrations are out of equilibrium with the mineral assemblage may result
in distinct trace element zoning patterns (Jamtveit and Hervig 1994; Stowell et al. 1996;
Moore et al. 2013; Angiboust et al. 2014; Hallett and Spear 2014b). Invoking the ‘exotic
fluid’ interpretation has often proved challenging because examples of patchy or oscillatory
zoning can often be equally well explained by pre-existing heterogeneities (see above). This
is especially true with respect to REE zoning in garnet, as corroborating evidence from fluid
inclusions or coincident zoning in major elements is usually missing. Moore et al. (2013)
explain REE zoning in rims of Franciscan blueschist garnets as related to fluid infiltration,
because zoning in these elements is also coincident with annuli in Ca and Mn, presumably also
enriched in the metasomatizing fluid. Moore et al. (2013) note that such zonation can act as a
time marker because rock-wide infiltration would result in similar trace element zoning for all
garnet crystals growing during the interval influenced by the fluid flow event.
A vivid example of garnet trace element zonation involving the infiltration of exotic fluids
comes from the fossil subduction zone of the Western Alps, where Angiboust et al. (2014)
observe patchy and oscillatory Cr zoning in garnets from shear zone eclogites (Fig. 8). The
complex Cr zoning, coupled with both enrichments of Mg, Ni, Cr, and LILEs and with boron
isotopic signatures in metasomatic minerals, was interpreted to reflect the large-scale episodic
infiltration of serpentinite-derived fluids (via antigorite breakdown) at ca. 80 km depth at the
slab–mantle interface (Angiboust et al. 2014). While this appears to contrast with garnet
oxygen isotope interpretations presented earlier (sediment-derived fluid source), Monviso
eclogites experienced multiple phases of metasomatism, and as the Tethyan seafloor (slow-
spreading ridge) dominantly comprised serpentinized mantle and pelites, mafic blocks were
likely to have interacted with multiple sources of fluid (Angiboust, pers. comm.).
Ultimately, linking the concentrations and zoning of trace elements in garnet to the
evolution of the mineral assemblage and the timing of garnet growth can be a powerful
petrochronologic tool, especially when comparing the competing roles of fluid infiltration,
intergranular element mobility, and crystal growth rate.
Concluding this review of the ways in which garnet can record petrologic or tectonic
processes and conditions, we turn now to the topic of direct garnet chronology. As we shift gears,
we encourage the reader to keep in mind the “petro-“ of garnet, and realize that direct garnet
chronology allows the petrochronologist to date any of these conditions and processes, and in
the best case, constrain their rate or duration (e.g., Fig. 2). The “petro-“ of garnet provides the
motivation for garnet petrochronology; otherwise we would just be dating a mineral.
Figure 8. Cr, Mn, and Ti zoning from a garnet from the Monviso ophiolite of the Western Alps, modified from
Angiboust et al (2014). Note patchy zoning in the core and oscillatory-zoned rims. Trace element annuli attrib-
uted to influx of externally derived fluid from antigorite breakdown of associated mantle wedge serpentinites.
CHRONO- OF GARNET
Even today, many in the geoscience community are unaware, or unconvinced, that
the growth of garnet can be dated directly, precisely, and accurately with the Sm–Nd or
Lu–Hf isotope systems. Much of that skepticism stems from the early history of garnet
geochronology in the 1980’s and 1990’s before certain modern methods were developed.
Since then, direct garnet geochronology has evolved to approach the hopes of would-be
petrochronologists of decades past. At the same time, garnets have also become one of most
often utilized minerals for ‘geospeedometry’ (Lasaga 1983; e.g., Chakraborty and Ganguly
1991) whereby the durations of tectonic and metamorphic events (generally events related
to heating and cooling) can be accurately reconstructed by modeling stranded diffusion
profiles within garnet. Valuable advances have also been made in linking the growth (and
chronology) of accessory phases such as monazite and zircon to the growth or breakdown
of garnet via textural and chemical means (e.g., Engi 2017, this volume; Lanari and Engi
2017, this volume; Rubatto 2017, this volume; Williams et al. 2017, this volume); we view
the integration of accessory phase petrochronology with true garnet petrochronology as one
of the more exciting future applications which may be inspired by this full RiMG volume.
Indeed, the decision to approach time-consuming garnet geochronology might be established
and motivated by preliminary accessory phase petrochronology, thermodynamic modeling,
and thin section petrography. While we leave it to other authors to describe advances in
linking accessory phase dating to garnet, in this section, we will review the past, present,
and future of direct garnet geochronology and garnet geospeedometry. Together, these two
methods provide the ‘chrono’ in garnet petrochronology including absolute ages, absolute
durations, and relative rates of thermal processes at conditions wherein garnets exist.
Garnet geochronology
van Breemen and Hawkesworth (1980) and Griffin and Brueckner (1980) first recognized
that garnet’s relatively high Sm/Nd ratio (as compared to most other common minerals)
could be exploited for geochronology. Garnet growth is dated with the isochron method by
pairing it with the hosting ‘whole rock’ and/or other mineral separates with which it grew (and
subsequently evolved isotopically) in an isochron diagram (for a review of garnet isochron
basics, see Baxter and Scherer 2013). Further exploration of Sm–Nd garnet geochronology
continued in a few labs through the 1980’s until 1989 when the first attempts to date garnet via
U–Pb (Mezger et al. 1989) and Rb–Sr (Christensen et al. 1989) were published. Despite the
novelty of these studies, concerns surrounding the problem of contaminating mineral inclusions
in garnet, matrix heterogeneity, and open system mobility of daughter isotopes in the rock
system pushed garnet geochronology via U–Pb and Rb–Sr out of favor (for example, see Zhou
and Hensen 1995; DeWolf et al. 1996; Romer and Xiao 2005; Sousa et al. 2013). In a positive
twist, it was in part the recognition that such U–Pb analyses of ‘garnet’ were dominated by
accessory mineral inclusions that led to the advancement of accessory phase petrochronology
with textural and chemical links to garnet growth (e.g., Hermann and Rubatto 2003; Pyle and
Spear 2003; Wing et al. 2003; Foster et al. 2004). Sm–Nd garnet geochronology survived this
period, but not without lingering skepticism stunting its further development, especially given
the time and analytical effort required to overcome the challenges of contaminating inclusions
and low Nd concentrations. In 1997, the first Lu–Hf garnet geochronology was published by
Duchene et al. (1997). Since that time, significant advances have been made for both Lu–Hf
and Sm–Nd garnet geochronology, both in terms of sample preparation, isotopic analysis, and
data interpretation. With modern methods, garnet geochronology via Lu–Hf and Sm–Nd can
yield accurate and precise ages of garnet directly from the garnet itself. In this section of the
chapter, we will review some of the key advances, and remaining caveats, of which any user
or interpreter of garnet geochronology should be aware.
Garnet isochron age precision. The precision of an isochron age (Fig. 9) depends on
how well the slope of the isochron is constrained. Age precision (i.e., slope precision) depends
on three main factors: 1) the analytical precision of the isotopic datapoints, 2) the spread
in parent/daughter between lowest and highest points in the isochron, and 3) the scatter in
the isochron. In general, the slope of the isochron is constrained most precisely when the
analytical precision of each point is best, the spread in parent/daughter ratio is largest, and
the scatter is smallest. Let us first address the importance of analytical precision and isochron
spread in theoretical two-point isochrons. Then in the next section, we will explore the effect
of adding additional points to an isochron and their scatter therefrom.
Modern mass spectrometry (TIMS or MC-ICP-MS for Sm–Nd, MC-ICP-MS for Lu–Hf)
permits <10 ppm 2RSD analytical precision for 143Nd/144Nd and 176Hf/177Hf when sample
size is unlimited (~100 ng of Hf or Nd). TIMS is the optimal tool for Nd analysis given
higher net ion efficiency (especially as sample size decreases), whereas MC-ICP-MS is the
optimal tool for Hf for the converse reason. Oftentimes, especially in applications requiring
smaller amounts of garnet to be extracted and analyzed (such as zoned garnet chronology,
Sm-Nd Lu-Hf
10 10
30 Ma 30 Ma
1 1
0.1 0.1
0 1 2 3 4 5 6 7 8 9 10 0 1 2 3 4 5 6 7 8 9 10
10 10
Age precision (Ma)
380 Ma 380 Ma
1 1
0.1 0.1
0 1 2 3 4 5 6 7 8 9 10 0 1 2 3 4 5 6 7 8 9 10
10 10
1 1
2600 Ma 2600 Ma
0.1 0.1
0 1 2 3 4 5 6 7 8 9 0 1 2 3 4 5 6 7 8 9 10
147Sm/144Nd 176Lu/177Hf
Figure 9. Garnet–matrix two-point isochron age precision as a function of parent/daughter isotope ratio
for garnets of three different ages: Alpine (30 Ma), Acadian (380 Ma), Archean (2.6 Ga). In all plots, the
bold lower line is for 10ppm (2σ) analytical precision on the garnet daughter isotope analysis, the dashed
middle line is for 30ppm, and the thin upper line is for 100ppm. 10ppm analytical precision is achievable
for 176Hf/177Hf when load size is unlimited (10s to 100s of ng Hf run on MC-ICPMS) and for 143Nd/144Nd
(run as NdO+ on TIMS) when load size is > 4ng. As sample size decreases, as is usually the case for small
amounts of clean garnet with low Hf and Nd concentration, analytical precision worsens (see text for
discussion). 147Sm/144Nd analytical precision is 0.03% and 176Lu/177Hf is 0.2% indicative of optimal perfor-
mance on TIMS and MC-ICPMS, respectively. Matrix 147Sm/144Nd is 0.15 whereas matrix 176Lu/177Hf is
0.02, each typical values for crustal rocks. If additional points are included in the isochron, the age preci-
sion can change depending on the MSWD (see text for discussion).
see below), the amount of Nd and Hf is much less than desired for optimal analysis. This
sample size limitation has been one of the major factors limiting garnet geochronology.
Figure 10 shows a compilation of garnet data, cleaned as well as possible with modern
methods (see discussion below). Daughter element concentrations in clean garnet are
generally less than 0.5 ppm and often below 0.1 ppm for Nd and Hf, meaning that tens of
>
0 1 2 3 4 5 6 7 8 9 10 0 1 2 3 4 5 6 7 8 9 10
147 144 176 177
Sm/ Nd Lu/ Hf
0.0 0.2 0.4 0.6 0.8 1.0 1.2 1.4 1.6 1.8 2.0 2.2 0.0 0.2 0.4 0.6 0.8 1.0 1.2 1.4 1.6 1.8 2.0 2.2
[Nd] ppm [Hf] ppm

Figure 10. Histograms of reported garnet Sm–Nd and Lu–Hf data after cleansing efforts to remove inclu-
sions using the methods of Pollington and Baxter (2011) for Sm–Nd and Lagos et al (2007) for Lu–Hf.
147
Sm/144Nd and 176Lu/177Hf data are taken from Baxter and Scherer (2013), [Nd ppm] and [Lu ppm] con-
centration data correspond to the parent/daughter data and are compiled from the Baxter Lab and Scherer
Lab (Erik Scherer, pers. comm.). Some 176Lu/177Hf data are off scale reaching as high as 50. Most well
cleansed garnet has 147Sm/144Nd and 176Lu/177Hf > 1.0, and [Nd ppm] and [Hf ppm] < 0.4.
milligrams of clean garnet may still only provide a few nanograms of Nd or Hf for analysis.
Note that in estimating required sample size, one must consider the sample loss that occurs
during mineral separation and garnet cleansing (see below) which can often exceed 80%
of the raw garnet mass (e.g., Pollington and Baxter 2011). As analytical methods continue
to improve, we can push to smaller and smaller garnet sample sizes and higher resolution
petrochronology. Fortunately, recent and continuing advances in mass spectrometry permit
improved analytical precision even as sample size decreases (Harvey and Baxter 2009;
Schoene and Baxter 2017, this volume; Baxter and Scherer 2013; see Bast et al. 2015). For
example, sub-nanogram loads of pure Nd standard solution analyzed as NdO+ via TIMS can
still yield external precision < 30 ppm 2RSD, though experience shows that sub-nanogram
samples run through column chemistry typically return internal precisions 2−5 times worse
(Schoene and Baxter 2017, this volume). Small sample analysis of Hf via MC-ICP-MS
remains a limitation in most labs, though use of a desolvating nebulizer (Aridus), refined
cone geometry, and 1012 ohm resistors, permits sub-nanogram loads of Hf from dissolved
samples run through column chemistry and analyzed via MC-ICP-MS to yield 50−180 ppm
2RSE internal precision (Bast et al. 2015). Daughter isotope precision plays a larger role
in controlling age precision for Paleozoic and younger samples and when parent/daughter
ratio is lower, whereas parent/daughter isotope precision plays a larger role for Proterozoic
and older samples and when parent/daughter ratio is higher (see Baxter and Scherer 2013).
In general, 147Sm/144Nd ratios measured by isotope dilution (ID)TIMS can yield external
precision (i.e., by assessing reproducibility of mixed standard solutions, or homogeneous

dissolved natural rock samples) better than 0.1% with modern methods. 176Lu/177Hf analysis
via ID-MC-ICPMS is less precise—more typically 0.1 to 1.0%—because making precise
mass bias corrections and spike subtraction is more challenging given that there are only
two isotopes of Lu (though MC-ICP-MS permits addition of a secondary element, such as
Erbium, for purposes of mass bias correction; e.g., Bast et al. 2015).
Most clean garnet yields parent/daughter greater than 1.0 for both isotopic systems
(Fig. 10), perhaps slightly more often for 147Sm/144Nd than for 176Lu/177Hf. It is also
interesting to note that the very highest parent/daughter ratios (as high as 50; Lagos et al.
2007) are from 176Lu/177Hf. These observations are an indication that 176Lu/177Hf tends to be
much more strongly zoned from core (highest Lu/Hf) to rim (lowest Lu/Hf) owing to strong
Rayleigh fractionation of parent Lu (over daughter Hf) into garnet, than 147Sm/144Nd which
tends to have more uniform zonation (for example, see Fig. 7, or modeling of Kohn (2009), or
LA-ICP-MS data from Anczkiewicz et al. (2007) and Anczkiewicz et al. (2012). Overall, as
shown in Figure 9, theoretical two-point isochrons between clean garnet with parent/daughter
> 1.0 and a whole rock or matrix as the lower second point on the isochron can generally
yield age precision between ±1.0 and ±0.1 million years (Baxter and Scherer 2013) assuming
optimal analytical precision. If the garnet has parent/daughter << 1.0, age precision (and
sometimes accuracy, see below) degrades quickly. As sample size decreases and analytical
precision predictably worsens, poorer age precision also results, though ±1−5 million years
age precision is often still achievable in most cases.
Multi-point isochrons and the MSWD. If more than two points populate an isochron
(which is always desirable), a statistical opportunity exists to evaluate that scatter and include
it in the uncertainty of the isochron’s slope, and the age. The most geochronologically
relevant statistical measure of this scatter is the ‘mean square of weighted deviates’ or MSWD
(e.g., Wendt and Carl 1991). It compares the scatter expected given the reported analytical
uncertainty to the actual scatter of data from an isochron. When the observed scatter matches
the statistically predicted scatter given analytical uncertainty, the MSWD is near 1.0; in this
case additional points to the isochron can lead to better precision than that shown in Figure 9.
Much higher MSWD will in turn lead to poorer (higher) age uncertainty and means that
there is real geologic scatter from a single isochron; that is, some of the points populating the
isochron fail one of the fundamental isochron assumptions. High MSWD isn’t necessarily
bad news though. If the high MSWD is due to multiple garnet analyses that scatter off the
isochron, this may be an indication of resolvable age variation in the garnets being analyzed.
Since it has been shown that garnet from the same rock (e.g., Skora et al. 2008; Herwartz
et al. 2011) or the same crystal (e.g., Pollington and Baxter 2010; Dragovic et al. 2015) can
span growth ages of many millions of years or even tens to hundreds of millions of years,
such scatter and high MSWD is a useful and expected indicator of resolvable age zonation
(also see Kohn 2009) that might encourage further textural or zoned garnet geochronologic
analysis. In this regard, a range of individual two-point garnet–matrix ages is of greater
meaning than a lumped average growth age produced by a multi-point isochron with high
MSWD (e.g., Pollington and Baxter 2010; and see Fig. 16 discussed in the section on Zoned
Garnet Geochronlogy below). If high MSWD is due to scatter from multiple whole rock,
matrix, or mineral analyses (including poorly cleansed garnet) on the low side of the isochron,
that could mean that one or more of those points doesn’t belong and should be removed
(this is discussed in a later section). Or, it may reflect real heterogeneity in the local rock
matrix that, unfortunately, must be included in the age and its uncertainty (also discussed
in a later section). MSWD less than 1.0 generally means that analytical uncertainties have
been overestimated. Overestimation—sometimes referred to as ‘conservative’ estimation—
of analytical uncertainty can also change a dataset that would yield a very high MSWD into
a dataset that yields a lower MSWD, unintentionally masking what may in fact be important
geological scatter. The savvy geochronologist, or interpreter of geochronologic data, will be
careful to look for such high ‘conservative’ estimates of analytical uncertainty in reported
isochron data. It is therefore crucial that estimates of analytical uncertainty are accurate,
not ‘conservative’, as both over and underestimates can have deleterious effects on age
interpretation masking what could be important information about the system.
The garnet point on the isochron and the problem of inclusion contamination. Garnet
geochronology with Lu–Hf or Sm–Nd works because garnet uniquely fractionates parent from
daughter creating an unusually high parent/daughter ratio. The garnet always represents the
high point (or points, if one makes multiple measurements of the garnet) on the isochron.
However, essentially all garnets contain micro-inclusions of other minerals that may be older
(if inherited) or younger (if accessed by cracks) than the crystallization age of the garnet
itself. The problem of inclusions continues to be the greatest challenge for successful garnet
geochronology; if inclusions in garnet are not sufficiently removed, resulting ‘garnet’ ages
can be imprecise, and worse, grossly inaccurate. Unfortunately, the published literature is
full of such examples and the reader is invited to critically evaluate the literature themselves
in light of the perspective and data we offer in this section. At the same time, the notion
that we cannot overcome the challenge of inclusions remains the greatest misconception that
continues to limit the credibility and broader use of garnet geochronology. In fact, numerous
methodologies have been developed that provide solutions to the inclusion problem in almost
all cases, leading to clean garnet and robust, precise, and accurate garnet ages.
Why inclusions are a problem. The problem of contaminating inclusions was recognized
in the very first paper on garnet geochronology in 1980 (van Breemen and Hawkesworth 1980).
Micro-inclusions can be inherited from significantly older episodes, and some inclusions
can be reset or precipitated after garnet growth during retrograde cracking and fluid influx.
Therefore, in the worst case, a ‘garnet’ analysis that still contains abundant inclusions can lead
to grossly inaccurate apparent ages that can be younger or older than the true garnet age (e.g.,
Fig. 11). In general, inherited (i.e., older) inclusions that have a parent/daughter isotope ratio
lower than the host rock (or younger inclusions that have a parent/daughter ratio greater than
the host rock) will pull the contaminated garnet down off the true isochron to create falsely
young ages. The converse is also true. Of course, the more contaminated the ‘garnet’ analysis
is, the lower its apparent parent/daughter, the greater the potential inaccuracy (Fig. 12). Even
if included phases are in age equilibrium with the garnet (i.e., they lie perfectly on the garnet–
matrix isochron) their presence in the ‘garnet’ analysis will pull the ‘garnet’ point down along
the isochron, reducing the spread along the isochron, thus worsening the age precision (Fig.
12). The reader is referred to a great number of published studies that define and describe the
problem of inclusion contamination for garnet geochronology (e.g., Zhou and Hensen 1995;
Scherer et al. 2000; Prince et al. 2001; Thoni 2002; Baxter and Scherer 2013).
How to solve the problem of inclusions. In the past 20 years numerous methods have
been developed to eliminate most inclusions from most garnets thereby solving the inclusion
problem. Any attempt to clean a garnet separate begins and ends with careful handpicking
of finely crushed separates. However, good handpicking alone is usually inadequate to fully
alleviate the inclusion problem (e.g., Thoni 2002). Micro-inclusions that may not be visible
to the naked eye can escape even the most diligent handpicking. So, the most successful
methods to cleanse garnets of their micro-inclusions all involve a ‘leaching’ or ‘partial
dissolution’ procedure in various strong acids either to dissolve away problem inclusions in
discarded solution leaving pure garnet for analysis, or, to dissolve away pure garnet in the
analyzed solution leaving problem inclusions behind in a solid residue. The former has been
successfully employed for Sm–Nd geochronology (e.g., Zhou and Hensen 1995; DeWolf et
Successfully
0.5140
cleansed garnet
a
0.5135
. 3 6 M8)
0 0.7
9 ±D =
Nd/144Nd
. 4
46 SW
(M 0.51269 ch
ro
n
0.5130 so
0.51267 xi
a tri
0.51265 et-m
rn
143
ga
0.51263
0.51261
0.5125 ed
0.51259 ns
clea net
0.51257 un gar
0.51255
0.10 0.20 0.30 0.40 0.50
0.5120
0.0 1.0 2.0 3.0 4.0 5.0
147 144
Sm/ Nd
Figure 11. The problem of inclusions, and the success of partial dissolution to solve it. Sm–Nd isochron
data are from a mafic blueschist (modified from Dragovic et al 2012). The lower inset shows uncleansed
garnet (open symbols) dominated by inclusions prior to partial dissolution. Any attempt to create an age
by pairing an uncleansed garnet with the matrix (filled circles) would lead to falsely young—even nega-
tive—ages. Main figure shows the same garnets after proper partial dissolution cleansing (filled diamonds
and squares) and resultant accurate and precise isochron age.
al. 1996; Amato et al. 1999; Scherer et al. 2000; Baxter et al. 2002; Thoni 2002; Anczkiewicz
and Thirlwall 2003; Pollington and Baxter 2011) where the most insidious inclusions are
REE-rich minerals like monazite or clinozoisite which will dissolve in acid more readily than
their garnet host. The latter has been successfully employed for Lu–Hf geochronology (e.g.,
Scherer et al. 2000; Lagos et al. 2007) where the most insidious inclusion is Hf-rich zircon,
which is extremely resistant even to hydrofluoric acid at typical hotplate temperatures. This
dichotomy does mean it is difficult to design a single cleansing method optimized for both
Lu–Hf and Sm–Nd geochronology on the same sample aliquot. This, along with the analytical
differences described earlier, in turn explains the dearth of studies where both Lu–Hf and
Sm–Nd geochronology on the same samples yield optimal quality results from both systems.
Some of the variables to consider when testing a partial dissolution method on a given garnet
sample include: 1) acids used (e.g., Anczkiewicz et al. (2004) use sulfuric acid which attacks
phosphates well, but does little to silicate inclusions like epidote if they are a factor, whereas
HF partial dissolution of Amato et al. (1999) removes silicates well, but dissolves much of
the garnet too requiring a delicate balance), 2) duration (anywhere from 15 to 180 min in HF
have proven useful in different cases; e.g., Baxter and Scherer (2013)), 3) grain size (anywhere
between 250 and 100 mesh size is recommended; finer grain size may lead to problems with
garnet reactive surface area being too high; Pollington and Baxter 2011).
The term ‘partial dissolution’ cleansing is preferred over ‘leaching’ as the latter syntax
could be interpreted to mean that something is actually leaching out of the garnet lattice itself.
This leads to questions about whether the ‘leaching’ procedure is preferentially ‘leaching’ out
0.5170 0.5124
A B
0.5160 0.5123
0.5150 0.5122
Nd/144Nd
0.5140 0.5121
143
0.5130 0.5120
Inclusion contaminated
0.5120 0.5119 “garnet”-matrix isochron
0.5110 0.5118
0 0.5 1 1.5 2 0.09 0.14 0.19 0.24 0.29
385 1000
C D
380
800
375
Apparent Age (Ma)
600
370
365 400
360 200
355
0
350
-200
345
340 -400
0 0.5 1 1.5 2 0.09 0.14 0.19 0.24 0.29
147
Sm/144Nd 147
Sm/144Nd
Figure 12. Effects of contamination by inclusions. These theoretical diagrams shown are for the Sm–Nd
system but also qualitatively apply to the Lu–Hf system. Clean garnet (circle) has high 147Sm/144Nd ratio.
Diamond is the matrix. Square is a low 147Sm/144Nd inherited inclusion (e.g., monazite for Sm–Nd or zircon
for Lu–Hf) yielding an age of 450 Ma when paired with the host matrix. A and B show the true garnet–ma-
trix isochron (solid line, with an age of 380 Ma) and a garnet–inclusion mixing array (bold dashed line).
C and D show the apparent “garnet”-matrix age for “garnet” data points (circles) with varying amounts of
inclusion contamination along the garnet–inclusion mixing line. Note that heavily contaminated “garnet”
plotting near the matrix 147Sm/144Nd can yield imprecise and grossly inaccurate ages that are older or
younger than the true garnet–matrix age, including even negative ages. B and D show a single example of a
heavily contaminated “garnet” data point paired with the matrix to create a falsely young apparent isochron
age with shallower slope (fine dashed) than the true garnet–matrix age. Two-point isochron age error bars
shown are determined assuming the analytical parameters outlined in Figure 9.
parent vs. daughter (or vice versa). Here, it is important to note that none of the aforementioned
studies have shown any evidence that the partial dissolution methods being employed create
any fractionation or change of the garnet chemistry itself. When garnets are well cleaned
they give remarkably consistent age results that would be impossible if some fractionation
via ‘leaching’ was occurring (see Fig. 11). Differential ‘leaching’ of parent and daughter is
also extremely unlikely because the mechanism for ‘leaching’ would be solid state diffusion
through the garnet lattice; diffusivities at hot-plate temperatures (~100−200 ºC) are too slow
to allow for any significant diffusion of REE and even slower Hf. Alpha-decay induced lattice
damage that may enhance diffusive loss in U/Th-rich minerals like metamict zircon is not
an issue in garnet given the orders of magnitude smaller alpha flux in garnet due to lower
concentrations of the alpha producing elements (e.g., Sm), the much slower decay rate of Sm,
and single alpha decay for 147Sm to stable 143Nd. Should any fractionation effects be found
(and no evidence yet exists to suggest there should be) it would be more likely to affect the Lu–
Hf rather than the Sm–Nd system given the stronger chemical difference between the REE Lu
and the HFSE Hf (as opposed to the REE’s Sm and Nd). In summary, there is no evidence to
suggest that partial dissolution cleansing of inclusions is affecting the remaining clean garnet’s
Sm–Nd and Lu–Hf chemistry at all.
How to know the garnet is clean (or clean enough). Because contaminating inclusions
have much lower parent/daughter ratios, and much higher daughter element concentrations
than their garnet hosts, well-cleaned garnet generally will exhibit high parent/daughter ratio
and low daughter element concentration. Figures 10 and 13 provide two different compilations
of ‘garnet’ parent/daughter ratio and daughter element concentration data to illustrate this.
Figure 13 depicts the full range of published garnet data from a partial compilation of the
literature. No screening was used in this compilation except to include only data reported
as ‘garnet’ by the authors. Note the remarkable range in the data spanning many orders of
magnitude, and the conspicuous negative relationship between parent/daughter ratio and
daughter element concentration. In general, higher 147Sm/144Nd and lower Nd concentration
indicates a cleaner garnet. But where do we draw the line between clean, clean enough, and
dirty garnet? Rather than single out specific studies as good or bad, we present in Figure 10
similar compilations of published and unpublished Sm–Nd and Lu–Hf garnet data from over
100 different garnet-bearing rocks analyzed in the Boston University Lab (data from Ethan
Baxter) and the Muenster Lab (personal communication from Erik Scherer) as they represent
a subset of data prepared in a consistent manner, reflecting modern practices for Sm–Nd (by
Baxter and colleagues using partial dissolution methods described by Pollington and Baxter
2011) and Lu–Hf (by Scherer and colleagues using partial dissolution methods described
by Lagos et al. 2007). By no means does this mean that every sample in these plots is a
perfectly cleansed garnet (that is surely not the case). But it does give an indication of the
range of measured garnet compositions that typically results from best practices in these two
particular labs that is instructive for sake of comparison. For Sm–Nd, 90% of the data indicate
147
Sm/144Nd > 1.0 and [Nd] ppm < 0.4. For Lu–Hf, 90% of the data indicate 176Lu/177Hf > 0.5
and [Hf] ppm < 0.4. If the garnet doesn’t exceed these cutoffs after the first attempt to cleanse
it, we suggest additional attempts. Baxter and Scherer (2013) recommend that garnet is clean
enough to avoid significant effects of inclusion contamination when it has parent/daughter
ratio > 1.0, generally coinciding with daughter element concentration < 1.0. These of course
are arbitrary cutoffs but are meant to provide some guidance in establishing the confidence
(i.e., accuracy) of a given garnet age. Figure 12 also shows an example of the progressively
diminished effect of inclusion contamination as a garnet is cleansed.
A good way to evaluate whether a garnet separate is clean is by comparison to in situ
LA-ICP-MS or SIMS analysis (e.g., Prince et al. 2001; Anczkiewicz et al. 2007; Stowell
et al. 2010; Stowell et al. 2014; Gatewood et al. 2015; Dragovic et al. 2016). Ideally, any
garnet would first be lasered in situ to establish a baseline for clean parent/daughter and
daughter element concentration in the garnet. Careful screening of laser data (considering
multiple elements) is crucial to eliminate even slightly contaminated spots from the data (i.e.,
contamination by non-garnet mineral inclusions), because garnet Nd and Hf concentrations
in clean garnet are so low (< 1 ppm, sometimes < 0.1 ppm). Still, LA-ICP-MS analysis can
provide a valuable comparison to full garnet analysis of cleansed samples to see if partial
dissolution was successful. Figure 14 shows an example from large garnets from Townshend
Dam, Vermont from Gatewood et al. (2015). Note first the raw dataset for [Nd] ppm showing
abundant spikes of high Nd (over 100 ppm!) within the garnet vividly depicting the significant
presence of inclusions. After screening away most of the inclusion-contaminated points,
Figure 13. 147Sm/144Nd vs. [Nd] ppm for “garnet”

data. The oval outlines a compilation of hundreds
of published “garnet” data in the literature, a great
10 many of which had not been properly cleansed
Cleaner garnet: of inclusions. Compare to the data from cleansed
Higher
and uncleansed garnet shown in Figures 10, 11.
147Sm/144Nd
confidence
Garnets with 147Sm/144Nd >1 and [Nd] <1 ppm is
more likely to be clean and will return an accurate
0.01 0.1 10 100 garnet age (darker shading). Garnets with very low
Dirtier garnet: [Nd] 147
Sm/144Nd and/or very high [Nd] ppm (lighter
Lower ppm
confidence
shading) constitute a large amount of the published
0.1 data, these are almost certainly contaminated by in-
clusions; use of such analyses for a garnet isochron
age will lead to less precise and possibly inaccurate
results (see Figure 12).
Sm Nd
1.5 ppm 1.0 ppm
1.2 ppm 0.8 ppm
0.9 ppm 0.6 ppm

0.75
0.6 ppm 0.4 ppm
TIMS
0.34
TIMS
0.3 ppm 0.2 ppm
0 ppm 0.04 0 ppm 0.03

1 cm 1 cm
Lu Unfiltered Nd
30 ppm >100 ppm
24 ppm 80 ppm
18 ppm 60 ppm
12 ppm 40 ppm
6 ppm 20 ppm
0 ppm 0 ppm
1 cm 1 cm
Figure 14. LA-ICPMS maps of Sm, Nd, Lu [ppm] in garnet from Townshend Dam, Vermont (modified
from Gatewood et al 2015). Note Sm is slightly zoned with enrichments at the rim, whereas Lu is more
strongly zoned with enrichments at the core. Nd is unzoned, except for numerous high spots reflecting
Nd-rich inclusions. Bottom right panel shows raw unfiltered Nd data where the highest analyses show
[Nd] in excess of 100ppm. Top right panel shows [Nd] data after carefully filtering out the most egregious
contamination, revealing the underlying mostly clean garnet with concentrations <1ppm. White bars and
values show the range of ID-TIMS analyses of the same Vermont garnets after proper cleansing reported
by Gatewood et al (2015), showing good agreement with the laser data.
the refined LA-ICP-MS data show low [Nd] from 0.03 ppm to 0.5 ppm. In this study, these
values are a close match for the cleaned bulk garnet data analyzed via TIMS, indicating
success in cleansing away these inclusions and recovering true clean garnet.
Another vivid example of the success of partial dissolution cleansing, and the great
importance of doing so, is shown in Figure 11 from Dragovic et al. (2012). Here, note where the
‘garnet’ data plot when they were intentionally analyzed without partial dissolution cleansing.
These data yield low 147Sm/144Nd < 0.45 and high [Nd] ppm between 0.9 and 2.8 ppm along a
crude mixing trend with an inherited inclusion population with low 147Sm/144Nd (akin to the
theoretical example shown in Fig. 12b). Any of these ‘garnets’, when paired with the matrix,
would give falsely young ages; a few would even give negative ages. But consider these very
same garnet samples after experiencing a proper partial dissolution. Now, these data give high
147
Sm/144Nd between 1 and 5 with [Nd] between 0.03 ppm and 0.07 ppm and plot together on a
remarkably tight isochron indicating the time of garnet growth in this mafic blueschist.
The second point on the isochron. It takes a slope to calculate an age, a line to calculate a
slope, and two points to make a line. Three, four, or five points lend further statistical credence
to a linear isochron relationship but without at least a second point, isochron geochronology is
a non-starter. In this regard, the second point on the isochron is just as important and powerful
as the garnet point on the isochron in determining age precision and accuracy. The second point
on the isochron must be an accurate and precise measurement of the rock matrix reservoir with
which the garnet initially grew in isotopic equilibrium (e.g., 143Nd/144Nd or 176Hf/177Hf). In the
very simplest scenario (rarely achieved in real systems), an isotopically homogeneous parent
rock or melt simultaneously crystallizes all its minerals, including garnet, such that the entire
rock is in initial isotopic equilibrium and all minerals remain closed systems from that point
until the present day. In this case, one could measure anything else in the rock—the entire
whole rock itself, just the matrix, or any subset of the various other minerals in the rock—as a
representation of that initial rock reservoir and pair it with the pure garnet to form a two point
isochron. Additional garnets (with high Sm/Nd or Lu/Hf) or additional whole rock, matrix,
or mineral analyses (with low Sm/Nd of Lu/Hf) would further populate the isochron and a
happy multi-point isochron with a perfect MSWD of 1.0 would result. In reality, many rocks
(or protoliths) are not perfectly isotopically homogeneous at the onset of garnet growth, the
whole rock and matrix may differ due to parent/daughter Rayleigh fractionation during garnet
growth, the matrix may experience open system change of parent/daughter or daughter isotope
composition after/during garnet growth, and different mineral phases may grow or close to
isotopic exchange with the rock reservoir at different times than the garnet. Any of these
factors can conspire to create an inaccurate or imprecise garnet isochron age if such data that
don’t belong are included on an isochron. Stated differently, more points on an isochron are
only a good idea if those points belong on the isochron. The good news is that most of these
effects are very minor to negligible most of the time. Let us explore these each in turn.
Initial matrix heterogeneity. If the rock reservoir is initially heterogeneous for whatever
reason, how can we know what the garnet actually grew in equilibrium with? Igneous protoliths
are probably least susceptible to this issue given that they are fairly homogenous when they first
form. However, ancient layered sedimentary protoliths where each layer may include inherited
minerals of varying provenance can present problems in this regard. For isotopic systems like
U–Pb or Rb/Sr where the range of parent/daughter ratio among phases can be several orders of
magnitude, this can create enormous heterogeneities (via differential radiogenic ingrowth) the
effects of which can be severely problematic for garnet (or any) isochron geochronology in those
systems (e.g., Romer and Xiao 2005). Fortunately, with the exception of garnet, most common
minerals (and rocks) have a limited range of Lu/Hf and Sm/Nd ratios such that the magnitude of
matrix heterogeneities is much smaller. Still, matrix heterogeneity can be significant given the
high precision and accuracy often desired (and achievable) with modern methods.
Let us consider a layered sedimentary protolith. The layers have varying 147Sm/144Nd (or
176
Lu/177Hf) which has led to varying 143Nd/144Nd (or 176Hf/177Hf). Let us now permit garnet
to grow instantaneously and with uniform distribution in that layered rock. [In reality, the
different layers may have different enough major element chemistry so as to preferentially
crystallize garnet more on certain layers than in others, but let us ignore that for the present
discussion]. At any given location in the layered rock system, the new garnet crystallizes
with an average composition reflective of the equilibrium lengthscale (Le) for that element
in the system (see Baxter and DePaolo (2002a,b), and DePaolo and Getty (1996) for further
discussion). The equilibrium lengthscale for a given element, Le, is dependent on the effective
diffusivity (D*) of the element within the intergranular transporting medium (ITM; Baxter
and DePaolo 2002b), and the local reaction/exchange rate (R) for that element between the
matrix minerals and the ITM: Le = (D* / R)1/2. The effective diffusivity (D*) includes both the
diffusivity in the ITM, and the partitioning of that element between the solid minerals and
the ITM. Elements (like Sr) that are strongly partitioned (i.e., soluble) into the fluid filled
ITM have high D* and smear out and average 87Sr/86Sr over a large equilibrium lengthscale.
Elements like Nd and Hf are very weakly partitioned (i.e., insoluble) into the fluid filled ITM
of most crustal fluids and thus have relatively low D*; thus 143Nd/144Nd and 176Hf/177Hf of the
ITM (and any garnet crystallizing from it) will closely match the local bulk matrix solid. In
practice, this means you wouldn’t want to pair a garnet from one layer with bulk matrix from
another layer. But, if you crushed up the entire rock and sampled a representative average
garnet separate and matrix from the entire volume, the heterogeneities average out and the
problem goes away. However, now consider the growth of a very large garnet porphyroblast
large enough to grow over several compositional layers. A single crystal like this would be
attractive for high-resolution microsampling and zoned geochronology. In this scenario,
each concentric growth ring, or even different portions of the same growth ring, of the garnet
may inherit a different starting 143Nd/144Nd (or 176Hf/177Hf) depending on the layer(s) it is in.
When focusing on a single crystal, or portions thereof, it is no longer advisable to use a large
averaged whole rock to anchor the isochron in a heterogeneous protolith. Instead, one must
carefully analyze and evaluate the extent of porphyroblast scale matrix heterogeneity via
multiple measurements and include them all on an isochron. This is an important statistical
acknowledgement of the uncertainty about the second point on the isochron that will result
in larger age errors and larger MSWD. Gatewood et al. (2015) shows a vivid example of
the effects of such cm scale matrix heterogeneity on zoned garnet geochronology from a
layered meta-sedimentary rock. On the contrary, Pollington and Baxter (2010) saw no such
effect when conducting zoned garnet chronology in a homogeneous rock matrix, most likely
inherited from an igneous protolith. Overall, it is always advisable to evaluate the extent of
matrix heterogeneity of any rock, try to avoid major heterogeneities and lithologic contacts
if possible, and match multiple garnets (or bulk garnet separates) and average matrix from
the same rock volume to average out heterogeneities.
Whole rock vs. matrix. We define a ‘whole rock’ as the entire rock volume including
the garnet itself. We define the ‘matrix’ as the entire rock volume excepting the garnet. As a
garnet with much higher Sm/Nd and Lu/Hf than the original whole rock grows, it will alter
the Sm/Nd and Lu/Hf of the remaining matrix via simple Raleigh fractionation. Strictly
speaking, the core of the garnet (the initial fraction of garnet to grow) should be paired with
the ‘whole rock’, whereas the outermost rim of the garnet (the final bit to grow) should be
paired with the matrix. In general it is advisable to measure both matrix and whole rock
from the same rock averaged volume to establish the significance of that difference for
your particular rock. For Sm–Nd, the general finding is that matrix and whole rock are
very rarely different enough to make a significant difference to the age. This is largely
due to the fact that the concentration of Nd and Sm in clean garnet is usually one or more
orders of magnitude smaller than the concentration in the matrix. It would require a lot
of garnet to create a significant difference in parent/daughter. The Lu–Hf system is more
susceptible to this process as Lu concentrations in garnet can be much higher (up to an order
of magnitude or more) than the whole rock especially at the onset of garnet growth, whereas
Hf concentrations are very low. This could lead to more significant Rayleigh fractionation
effects leaving the residual matrix with even lower Lu/Hf after garnet growth. Fortunately,
for either system, if the time duration between growth of core and growth of rim is short,
negligible radiogenic daughter in-growth (and rotation of the isochron beyond horizontal)
will have occurred. In this case, shifting Sm/Nd or Lu/Hf of the matrix to different values
along a near horizontal line near the time of garnet growth will have negligible effect on the
final isochron age determination. Finally, because most garnets also include (and effectively
sequester) other minerals the net change to the host rock composition is offset and minimized.
Rocks with low Nd or Hf concentrations (< 10 ppm) and a large timespan between growth of
garnet core and rim are therefore more susceptible to this easily accounted for effect.
Open system change of matrix. If the matrix experienced any kind of open system
exchange, or loss, or gain of Sm, Nd or Lu, Hf there is potential to skew the garnet–matrix
isochron age relationship. Mechanisms may include partial melting where a melt is lost from
the system, the injection of a melt into the system, or the passage of fluids through the system.
Most common crustal fluids have low solubilities of REE and Hf so only in extreme cases of
focused fluid flow should we observe major Sm/Nd or Lu/Hf loss or gains. Important examples
of such open system mobility of REE have been found in crustal fluids (Zack and John 2007;
Ague 2011). Loss of an internally derived partial melt would leave the solid residue with a
higher Sm/Nd or Lu/Hf ratio. But as long as the internally derived melt would be in Nd and
Hf isotopic equilibrium with the bulk solid there would be no isotopic fractionation in the melt
depleted residue. This should generally be the case, unless incongruent melting of a very old
protolith (where significant mineral scale Nd or Hf isotopic differences had already evolved)
created a melt with different isotopic composition (e.g., Zeng et al. 2005) and that melt was
extracted quickly, before isotopic equilibrium could be restored via diffusion; in this case the
resulting effect on age should be evaluated. Fortunately for the Sm–Nd and Lu–Hf systems (as
opposed to U–Pb or Rb–Sr), such radiogenic differences are relatively small given the narrow
range of Sm/Nd and Lu/Hf that exist among most common minerals (e.g., Romer and Xiao
2005). Introduction of an external melt or fluid into the system has the potential to be much
more problematic, regardless of when the metasomatic event occurred with respect to garnet
growth. External melts or fluids could bring completely different Nd or Hf isotopic chemistry
that could mix with the original rock matrix pulling it vertically up or down well off the
garnet–matrix isochron. Changes in Sm/Nd or Lu/Hf may also occur which could add to the
effect if the metasomatic event happened a long time after garnet growth (when the isochron
had already rotated well past horizontal). In general, the garnet geochronologist should avoid
samples bearing evidence of open system metasomatism: in a vein, next to pegmatite, near
a lithologic contact. Or, great care should be taken to evaluate the isotopic extent of that
metasomatism to try and recreate the original rock matrix as well as possible. In some cases,
the garnet inclusions may serve as a guide towards reconstructing the original rock matrix, but
as we have seen, these minerals inclusions (if they are inherited from earlier events) do not
always accurately reflect the garnet’s original isotopic growth environment.
Non-garnet mineral growth and/or closure. The very first garnet geochronology papers
chose minerals (like clinopyroxene in an eclogite, for example; Griffin and Brueckner 1980)
rather than a whole rock to pair with the garnet in two-point isochrons. Some papers continue to
add other minerals along with a whole rock or matrix to anchor the garnet isochron. Most of the
time, these mineral data plot very near the whole rock or matrix (i.e., with similarly low parent/
daughter) and have little effect on the age. But sometimes these individual minerals plot slightly
off the isochron, leading to higher MSWD if included, or different absolute ages if employed
instead of the whole rock. Which point is a better representation of the rock reservoir with
which the garnet grew in isotopic equilibrium? On the one hand, the clinopyroxene and garnet
in an eclogite both reflect growth at eclogite facies conditions (whereas the whole rock may
have other phases that grew before or since eclogite facies) and thus one could argue they are a
good match if the goal is to date eclogite facies conditions. However, it is almost certainly the
case that clinopyroxene has a different effective ‘closure’ time (due to processes like diffusion
or matrix recrystallization) than the garnet, so if the clinopyroxene has a different 147Sm/144Nd
than the matrix it will evolve off the garnet–matrix isochron and should not be included. The
magnitude of this effect depends on the difference in closure/growth age of clinopyroxene vs.
garnet, and on the difference in 147Sm/144Nd between clinopyroxene and the matrix. But, if the
matrix itself is made up of all these minerals, how then can we argue that the matrix is any better
a choice? The reason is that the matrix itself has remained a closed system and represents an
appropriate rock-averaged Sm–Nd and Lu–Hf composition from which the garnet first grew.
Resistant accessory phases (like monazite or zircon) are the exception in that they can retain
their inherited isotopic signatures and generally do not participate as reactants in garnet growth.
The second point(s) on the isochron cannot be overlooked. At a minimum, it is worth
measuring a representative whole rock and matrix. Pure non-garnet mineral separates may not
be good choices for the isochron, especially when their parent/daughter ratio differs greatly
from the matrix and/or their growth or closure time differs greatly from that of garnet. Avoid
rocks with evidence for open system exchange, of significant layering or heterogeneity. If
heterogeneity exists, it must be evaluated and included in the isochron to acknowledge that
uncertainty (e.g., Gatewood et al. 2015). While many of these issues have the potential to
completely ruin U–Pb or Rb–Sr garnet geochronology, they are rarely a major problem for
Sm–Nd and Lu–Hf geochronology if reasonable care is taken to evaluate them.
Why Do Lu–Hf and Sm–Nd Ages Differ? The Complementarity of Lu–Hf and
Sm–Nd Garnet Geochronology. One of the powerful opportunities of garnet geochronology
is the theoretical ability to date the same garnet (or garnet-bearing rock) with both Lu–Hf
and Sm–Nd. As discussed above, this hasn’t been done as often as we would like given the
different instruments required for optimal analysis (i.e., TIMS for Sm–Nd and MC-ICPMS
for Lu–Hf) and because the most popular methods employed for removing inclusions are
opposite for Lu–Hf and Sm–Nd. Thus, most labs are optimally equipped for one or the other,
but rarely for both. Still, a growing number of labs have attempted to date the same garnet-
bearing rocks with both methods (e.g., Lapen et al. 2003; Kylander-Clark et al. 2007; Cheng
et al. 2008, 2016; Skora et al. 2009; Anczkiewicz et al. 2012; Smit et al. 2013) leading to
interesting observations that, when fully understood can lead to valuable insights about the
rock’s history. Both systems are well suited for garnet geochronology, though their different
pros and cons may make one more suitable for a given sample suite or application, and thus
powerfully complementary when used in concert.
A notable observation made in many (though not all) of such combined studies is that
Lu–Hf garnet ages tend to be older than Sm–Nd ages from the same rock. Why is this the case?
The possible answers can be grouped into three categories, one (or more) of which may apply
in any given situation:
1. Different parent isotope zonation for Lu–Hf and Sm–Nd (e.g., Skora et al. 2009)
2. Different ‘closure temperature’ for Lu–Hf and Sm–Nd (e.g., Yakymchuk et al. 2015)
3. Something is wrong with the Sm–Nd age or the Lu–Hf age
Let us address each scenario in turn as each provides an opportunity to further explore
some of the important differences and complementarity of Lu–Hf and Sm–Nd garnet
geochronology.
Different parent isotope zonation for Lu–Hf and Sm–Nd. As discussed in the section on
trace elements above, Lu is very strongly partitioned into garnet during its growth as compared
to other common matrix minerals. On the contrary, Sm, Nd, and Hf are all weakly partitioned
into garnet as compared to other matrix minerals (though Sm not as weakly as Nd, hence the
reason that garnet can still have high Sm/Nd ratios despite relatively low Sm and very low Nd
concentrations). Thus, garnets tend to be strongly zoned in Lu (with highest Lu concentrations
in the early grown core as much as ~100 times higher than rims with low Lu; Figs. 7 and 14)
whereas Sm, Nd, and Hf tend to be unzoned or with slight enrichments toward the rim (e.g.,
Lapen et al. 2003; Cheng et al. 2008; Skora et al. 2008; Peterman et al. 2009; Anczkiewicz
et al. 2012; Smit et al. 2013; Gatewood et al. 2015; Fig. 15). Because the age information
contained within a garnet crystal is based on the concentration of the parent element (Lu or
Sm) this chemical zonation means that Lu–Hf ages will generally be skewed towards the age
of core growth (where more of the Lu resides) whereas Sm–Nd ages will be skewed towards
the age of rim growth (where more of the Sm resides). It is also important to note that while
these general trends for Lu and Sm zonation are common, more complex zonation in Lu and/
or Sm can exist in certain garnets ((see trace element section, above). This underscores the
value of analyzing the core-to-rim Lu and Sm zonation of the actual samples in question as they
have implications for interpreting the different ‘bulk’ Lu–Hf and Sm–Nd garnet ages. Lapen
et al. (2003), Skora et al. (2009), and Kohn (2009) show examples of how measured Lu and
Sm zonation can be used to model bulk ages by identifying where/when during garnet growth
the parent element and mass averaged age would plot (e.g., Fig. 15). Such modeling requires
certain assumptions including growth symmetry (e.g., perfect spherical shells), growth rate
(constant vs. episodic), but are generally very useful in explaining age differences and, in the
best case, placing constraints on total growth duration (e.g., Skora et al. 2009). For garnets that
never experience temperatures above ~650 ºC (above which diffusive mobilization may become
significant; see below) the difference in Lu vs. Sm zonation is the dominant reason why Lu–Hf
ages are different (generally older) than Sm–Nd ages. Both systems date primary growth, but
they reflect different stages of that prolonged growth duration.
Lu/Hf bulk age Sm/Nd bulk age

skewed towards skewed towards
CORE... Older! RIM... Younger!
100 5
Sm (ppm),
Lu/177Hf
80 4
50%
176
60 3
Lu (ppm),
147
Sm/144Nd
40 2
50%
20 1
core 0.2 0.4 0.6 0.8 rim

grain radius (r/a)
Figure 15. Conceptual zonation diagram of Lu, Hf, Sm, Nd resulting from simple Rayleigh fractionation
during growth (modified from Kohn 2009). “r/a” on the x-axis indicates relative radial location from garnet
core to rim of radius “a”. “50%” shows the radial location where exactly half of the parent isotope is in core-
ward and rim-ward garnet portions. Lu is strongly zoned, varying by two orders of magnitude from core to
rim, whereas Sm is only slightly zoned. Note spherical symmetry involved in this model. This zonation is the
primary reason that Lu–Hf bulk grain ages tend to be older than Sm–Nd bulk grain ages. Because 147Sm/144Nd
is nearly unzoned, the Sm–Nd system offers a better chance at precise zoned garnet chronology from core-to-
rim, whereas Lu–Hf offers a better chance to precisely date the earliest growth of garnet core.
Different ‘closure temperature’ for Lu–Hf and Sm–Nd. Because most garnets grow below
their ‘closure temperature’ for Sm–Nd and Lu–Hf, and never experience higher temperatures,
diffusive ‘closure’ is generally not at play when interpreting garnet geochronologic data,
including the difference between Sm–Nd and Lu–Hf ages. However, for garnets that did
experience temperatures in excess of ~650 ºC during or after their growth, the issue of
diffusive ‘closure’ (i.e., Dodson 1973) or perhaps more appropriately diffusive ‘re-opening’
(i.e., Caddick et al. 2010; Watson and Cherniak 2013) may come into play. Numerous papers
have attempted to place constraints on the ‘closure temperature’ of garnet for both systems
but, as pointed out and modeled by Ganguly and Tirone (1999) the classic Dodson formalism
for closure is rarely appropriate for garnet because it is rarely completely reset before dropping
through a closure temperature interval. As Watson and Cherniak (2013) describe, the matter
is more often one of diffusive ‘opening’ whereby a garnet is heated for some period of time
to temperatures high enough that diffusive resetting may begin. For example, Baxter et al.
(2002) used the diffusion data of Ganguly et al. (1998b) to show that only the outermost few
10s of microns of a garnet crystal would be affected by age resetting during heating up to
~660 ºC (over a span of 12 million years), resulting in a negligible degree of bulk age resetting
(depending on the grain size of course). Since then, numerous studies—both experimental
and empirical—have sought to constrain the diffusivity of the REE (i.e., Sm, Nd, Lu: e.g.,
Carlson (2012), Tirone et al. (2005), Van Orman et al. (2002) and Hf: Bloch et al. (2015)).
While all REE (including Sm, Nd, and Lu) appear to diffuse at the same rate within existing
uncertainties, some overall discrepancy exists amongst existing studies of REE diffusion.
Ganguly et al. (1998b) and Tirone et al. (2005) return diffusivities for the REE about an order
of magnitude higher than those of Van Orman et al. (2002), Carlson (2012), and Bloch et al.
(2015). Bloch et al. (2015) suggest that the difference may reflect a concentration dependence
on REE diffusivity (i.e., lower concentrations diffuse via a faster diffusion mechanism) such
that the faster Tirone et al. (2005) data are more appropriate for geochronologic applications.
In any case, the data of Tirone et al. (2005) represent the fastest and thus provide a lower
constraint on the temperatures at which REE in garnet become mobile for garnets, which
of course itself is also dependent on grain size and cooling rate. Rather than quote blanket
‘closure temperatures’, interpreters of garnet geochronologic data should explore the partial
diffusive resetting of ages by means of simple analytical or numerical modeling for the grain
size and specific heating/cooling history that may have existed in the given system. Ganguly
and Tirone (2001) and Watson and Cherniak (2013) provide useful analytical formalisms that
may be used in general cases. Baxter et al. (2002), Korhonen et al. (2012), and Dragovic et
al. (2016) are examples where simple case specific modeling was employed to constrain the
likely extent of resetting. Baxter and Scherer (2013) model the range of temperatures required
to reset a bulk garnet age just 5% (i.e., just beginning to ‘re-open’) and 95% (i.e., essentially
completely reset) for given grain size, temperature, and dwell time at that temperature. The
reader is encouraged to consult these papers and model the diffusive resetting of Nd or Hf age
information in their specific case.
The recent experimental data of Bloch et al. (2015) confirmed previous empirical inferences
(e.g., Scherer et al. 2000; Kohn 2009; Anczkiewicz et al. 2012; Smit et al. 2013) that the
diffusivity of Hf is considerably lower (by an order of magnitude, or more) than that of all the
REE, including Lu. Without question, Hf diffusivity is much slower than Nd diffusivity; thus the
Sm–Nd system is more susceptible to thermal diffusive resetting than the Lu–Hf system. That is,
the ‘closure temperature’ of Hf is significantly higher than that of Nd. Bloch and Ganguly (2015)
present diffusion modeling to show that the time required to fully reset Hf isotopic composition
of garnet is 10−1000 times longer (at a given temperature) than the time required to fully reset
Nd isotopes. Thus, for garnets experiencing temperatures above ~650 °C, the difference in Hf vs.
Nd diffusivity should result in generally younger Sm–Nd ages (which have been partially reset)
vs. Lu–Hf ages. Yakymchuk et al. (2015) shows a good example of this in a high temperature
(~850 ºC) migmatite where other possibilities (Lu vs. Sm zonation) can be ruled out.
Something wrong with the Lu–Hf or the Sm–Nd Age? As discussed above, there are
numerous other reasons why a given Sm–Nd or Lu–Hf age might simply be flawed, in which
case comparisons to good Lu–Hf or Sm–Nd data are misguided. Perhaps most notorious is the
effect of inclusions, especially as most samples are not cleaned with methods optimized for
both Sm–Nd and Lu–Hf (see above). While this chapter prefers not to enter the business of
identifying ‘bad’ published data, we will point out that both good and bad certainly do exist in
the literature. As previously discussed, Sm–Nd or Lu–Hf garnet data where the parent/daughter
ratio is < 1.0 and/or the daughter element concentration is > 0.5 ppm should be scrutinized
carefully as these are the hallmarks of inclusion contamination that can create significant age
inaccuracies (see Figs. 12, 13, 14). While high Nd or Hf concentration and low 147Sm/144Nd or
176
Lu/177Hf do not necessarily prove age-distorting contamination, experience shows that such
data usually indicate an inclusion problem that should be addressed or evaluated in some way
(Smit et al. 2013) before the resulting age is accepted.
An additional factor unique to the Lu–Hf system is the possibility of diffusive decoupling
of parent Lu from daughter Hf. Because Lu (and all REEs) diffusive at a much faster rate
than Hf (see above discussion and Bloch et al. 2015), it is theoretically possible that Lu
zonation may smooth out within a garnet, or that Lu may diffuse into or out of a garnet as
P–T changes (only if garnet–matrix Lu partitioning changes) subsequent to garnet growth,
all while the Hf isotopic composition remains locked in. While the effects of such Lu
mobility on resulting age can vary depending on the situation, most often this would result
in a counterclockwise rotation of the garnet–matrix isochron and falsely old ages. As with
all closure-related arguments for garnet geochronology, it is important to note again that
this process is only possible when garnet is heated above the ‘closure temperature’ for Lu
(> ~650 ºC) for a significant period of time, and most enhanced for smaller garnets, and
the highest temperatures. This argument rarely applies to the majority of greenschist and
amphibolite facies garnets that never experience such high temperatures. When this process
is at play, Lu–Hf ages can be compromised, thus rendering meaningful comparisons to good
Sm–Nd data impossible. Papers by Kohn (2009), Anczkiewicz et al. (2012), Bloch et al.
(2015), Bloch and Ganguly (2015), and (Kohn and Penniston-Dorland 2017, this volume)
discuss and develop this potentially confounding problem for Lu–Hf geochronology while
Anczkiewicz et al. (2012) shows compelling evidence for its occurrence in a natural setting.
Zoned garnet geochronology. The vast majority of published garnet geochronology,
and the entire discussion in this chapter up to this point, involves what we call ‘bulk garnet’
geochronology. As long as no unintended fractionation of the garnet occurs (for example,
due to magnetic separation which may select against garnet portions with differing amount
of magnetic inclusions; e.g., Lapen et al. 2003), a ‘bulk age’ simply represents the parent
isotope weighted (based on parent element zonation) average age of the garnet (Fig. 15).
For a garnet that grew very rapidly (i.e., whose growth duration is within the age precision),
such a bulk age is a valuable and precise measure of the growth event. However, if the garnet
growth duration exceeds the attainable age precision (as may often be the case) the ‘bulk age’
is of limited petrochronologic utility, representing only an average age within an unresolved
timespan of garnet growth. Insofar as multi-point bulk garnet isochrons might display scatter
amongst clean garnet data with high parent/daughter, the bulk age precision (poorer) and
MSWD (higher) represent a valuable indication that the garnet may have resolvable age
zoning. Consider for example, the dataset in Figure 16 from Pollington and Baxter (2010).
All of these garnet data are from the same rock. All of these garnet data are clean, passing any
test of inclusion contamination. Matched with the matrix data, we can calculate a multi-point
age of 25.5 Ma with a disappointing precision of ±5.3 Ma and eyebrow-raising MSWD of
270. Here is a reminder that poor MSWD, especially when caused by scatter of clean garnet
at high parent/daughter, is not necessarily (nor even often) an indication that something has
gone wrong. Rather it is a statistical invitation to explore the hypothesis of resolvable age
zonation through means of zoned garnet geochronology.
In fact, the data shown in Figure 16 are from different zones of the same single garnet
crystal, sampled as concentric rings from core to rim. With this context, the dataset suddenly
makes sense. Instead of a single ill-advised multi-point isochron, what we really have are
twelve individual two-point isochrons that reveal a 7.5 million year growth duration (Pollington
and Baxter 2010). With the textural context of each garnet growth zone, we find that multiple
two-point isochrons are better, and more appropriate, than a single multi-point isochron.
Zoned garnet geochronology is not new. In fact, the first attempts to extract and date different
garnet growth zones spanned work between 1988−1999 (Cohen et al. 1988; Christensen et al.
1989, 1994; Vance and Onions 1990, 1992; Burton and Onions 1991; Mezger et al. 1992; Getty
et al. 1993; Vance and Harris 1999). These papers used Rb–Sr and/or Sm–Nd to date two or
three concentric growth zones in single 1–5 cm garnet crystals, separated via some combination
of sawing, crushing, and plucking. Stowell et al. (2001) pioneered a core-drilling procedure to
extract different garnet zones for Sm–Nd geochronology in a number of settings including contact
metamorphism in Alaska, granulite facies migmatites in Fiordland, New Zealand, and granulites
from the Cascades of Washington USA. Solva et al. (2003) separately analyzed core and rim of
magmatic pegmatite garnet constraining Permian magmatism in the Alps. Ducea et al. (2003)
were the first to employ a microdrilling apparatus (the MicroMill™) to extract garnet from three
concentric growth zones from high temperature garnets in two Western North American sites.
They modeled their data to constrain cooling rates from peak magmatic arc temperatures. In their
study, the MicroMill™ was used to directly extract garnet in powdered form from individual
drilled pits in core and rim. Pollington and Baxter (2010) used the MicroMill™ for chemically
contoured sampling of concentric zones in a large 6 cm garnet from the Austrian Alps (Fig.
16). The spatial and age resolution of the Pollington and Baxter (2010) study permitted not
just a constraint on total growth duration, but also the recognition of two significant pulses in
the rate of garnet growth, correlated with chemical and textural features indicative of evolving
thermodynamic and tectonic conditions, that would have otherwise have been missed. Pollington
and Baxter (2010, 2011) used the Micromill™ to drill out concentric trenches, between which
solid garnet annuli were left for collection, crushing, partial dissolution cleansing, and Sm–Nd
TIMS analysis. They found that garnet powders derived from the drill trenches could not be
cleaned of inclusions due to the extremely fine grain size; thus, the material from drilled trenches
was discarded. Since then, three other studies (Dragovic et al. 2012, 2015; Gatewood et al.
2015) have employed the microdrilling methodology outlined in Pollington and Baxter (2011)
extracting between 3 and 10 annuli per garnet crystal in blueschist facies rocks from Sifnos,
Greece (Dragovic et al. 2012, 2015) and regional metamorphic schists from Townshend Dam,
Vermont (Gatewood et al. 2015; the same rocks first dated by Christensen et al. 1989). Most of
the studies cited in this paragraph show that < 1.0 Myr Sm–Nd age precision is achievable even
as sample size decreases with more highly resolved sampling and smaller sample sizes.
Until recently, Lu–Hf zoned geochronology had not been attempted, in large part due
to sample size limitations given that most labs still require 10’s–100’s ng Hf for MC-ICPMS
analysis, but this is changing. Herwartz et al. (2011) separated cores and rims of strongly color
zoned garnet from the Alps by crushing and handpicking, which were subsequently dated via
Lu–Hf. This first study of garnet age zonation with Lu–Hf revealed an old inherited Variscan
(~333 Ma) core surrounded by much younger Alpine (~38 Ma) rims, which these authors
interpreted as evidence of two distinct cycles of subduction to mantle depths separated by three
hundred million years. Nesheim et al. (2012) separated and grouped cores and rims of multiple
garnet crystals in a single Mesoproterozoic sample from Idaho, USA, revealing mixed garnet
ages bracketed by two major garnet growth events at ~1330 and ~1080 Ma. Anczkiewicz et al.
(2014) extracted and dated three growth zones, based on textural and chemical patterns, from
a large 3 cm garnet crystal from the Himalaya (see further discussion below). Schmidt et al.
(2015) used saw cuts to sample up to 4−6 solid growth zones in four different 3–5 cm garnets
for Lu–Hf geochronology. Rather than the surrounding matrix (which could not be sampled in
this study), Schmidt et al. (2015) anchored each interior garnet analysis with the Lu-depleted
0.5132
0.5131 Zoned Chronology:

25.5 ± 5.3 Ma
12 concentric 2-pt (MSWD=270)
0.513
garnet-matrix isochrons multi-point
0.5129 “errorchron”
0.5128
Nd/144Nd
Zone 1
Zone 2
0.5127 Zone 4
Zone 5
31 Zone 6
0.5126
143
29 Zone 7
Age (Ma)
27 Zone 8
0.5125 25 Zone 9
23 Zone 10
0.5124 21 Zone 11
19 Zone 12
0.5123 0.0 1.0 2.0 3.0 Zone 13
radius (cm) Measured garnets
0.5122 Measured matrix
0 0.5 1 1.5 2 2.5 3 3.5 4 4.5 5
147
Sm/144Nd
Figure 16. Zoned garnet geochronology for a single porphyroblast from Stillup Tal, Austria (modified from
Pollington and Baxter 2010). Each garnet “zone” represents a concentric growth zone from core (zone 1)
to rim (zone 13). When each is paired with the matrix, a pattern of systematically younging two-point iso-
chron ages is seen from core to rim, spanning about 7.5 million years as shown (inset). Bold dashed line is
the ill-advised multi-point ‘errorchron’ resulting from combining all garnet data and the matrix.
outermost rim in two-point ‘garnet only’ isochrons. The resulting age data suggest a total span
of garnet growth of at least 12 million years. Cheng et al. (2016) presented Lu–Hf analysis of
12 micro-sawed garnet zones (not concentric) within two single crystals from the Huwan shear
zone, China. Resulting two-point isochrons display ages ranging from 400 to 264 Ma which are
interpreted to reflect mixing of at least two distinct episodes of garnet growth spanning that time.
Zoned garnet geochronology, despite its first applications almost 30 years ago, remains
in its infancy, but its potential is great. Methodologies have now been developed to extract
well-defined chemically contoured growth zones of garnet from single crystals of ~5 mm of
greater diameter, clean those garnet portions of inclusions, analyze their isotopic composition
even as sample size diminishes, and extract accurate and precise Sm–Nd, or Lu–Hf ages on
each zone. Pollington and Baxter (2011) review the potential and current limitations for zoned
garnet geochronology by Sm–Nd. As analytical approaches continue to improve for Sm–Nd
and Lu–Hf smaller, more highly resolved records will become increasingly feasible. This is
the exciting future that awaits garnet petrochronology.
Geospeedometry with garnet
Given the time and expense required to obtain high precision zoned garnet geochronology
data, additional methods of inferring approximate metamorphic duration are desirable,
particularly for understanding events on timescales too short to resolve isotopically. Although
diffusion was discussed previously as an obstacle to thermobarometry, chemical profiles that
retain stranded and incompletely flattened chemical (or more precisely, chemical potential)
gradients provide an opportunity to infer maximum metamorphic duration. This obstacle/
opportunity duality often referred to as ‘geospeedometry’ (c.f. Lasaga 1983) is discussed in
detail by Kohn and Penniston-Dorland (2017, this volume) and reviewed briefly here because
garnet has proven to be a particularly useful phase for geospeedometric study of tectono-
metamorphism (e.g., Chakraborty and Ganguly 1991; Perchuk et al. 1999; Dachs and Proyer
2002; Fernando et al. 2003; Faryad and Chakraborty 2005; Storm and Spear 2005; Ague and
Baxter 2007; Galli et al. 2011; Hallett and Spear 2011; Viete et al. 2011; Spear 2014).
Fick’s first and second laws relate flux, J, and chemical concentration, C, of component i
to distance, x, and time, t:
∂Ci
J i = − Di
∂x
∂Ci ∂ 2C
= Di 2 i
∂t ∂x
where Di is the diffusivity. Numerous experimental and modeling studies have ascertained
Arrhenius relationships between D and T for various major (e.g., Elphick et al. 1981; Loomis
et al. 1985; Chakraborty and Ganguly 1991; Ganguly et al. 1998a; Carlson 2006; Ganguly
2010; Vielzeuf and Saúl 2011; Chu and Ague 2015) and trace elements (e.g., Tirone et al.
2005; Carlson 2012; Bloch et al. 2015) in garnet. The primary fitting parameters in most cases
are a pre-exponential constant and an activation energy term, although many formulations
consider activation volume and oxygen-fugacity terms, and some (e.g., Carlson 2006; Chu and
Ague 2015) explicitly account for changes in diffusion coefficient across the range of possible
garnet compositional space (e.g., stating that Fe would diffuse at a different rate through a
grossular-dominated crystal than through an almandine-dominated crystal).
Geospeedometry (Lasaga 1983) typically uses Ficks’ second law cast into an appropriate
geometry (e.g., Crank 1975) to model how an imposed initial chemical zoning profile would
relax with time, with the aim of finding the duration that best-fits an analyzed profile (e.g., in a
microprobe traverse through a garnet crystal). For diffusion of trace elements, D is essentially
independent of chemical composition (though see Bloch et al. 2015 who posit a concentration
dependence on REE diffusivity) and diffuses only in response to its own chemical potential
gradients, so that calculation is straightforward with analytical (e.g., Crank 1975) or numerical
techniques. Major element diffusion through the garnet lattice is less straightforward, with mass
and charge balance constraints requiring neutrality at each point throughout the crystal and
mandating coupling of diffusion. In essence one can think of this as follows: Fe cannot freely
flow left through a 1-D model of a garnet crystal unless one or several other components are
flowing right to compensate and maintain charge and volume. The four main divalent cations
in garnet are thus often considered simultaneously, by generation of a 3 × 3 element diffusion
matrix that couples tracer diffusivities of each component with their relative compositions.
The fourth element is generally considered a dependent component (Lasaga 1979):
 D* z z C 
Dij = Di*δij −  n i i j i   D*j − Dn* 
 ∑ zk2Ck Dk* 
 k =1 
where D* is a tracer diffusivity, z is the charge of the component of interest, n is the dependent
component, and δij is the Kronecker delta. For the example of Ca as the dependent component,
Ficks’ second law thus expands to:
 ∂cFe   ∂DFeFe ∂DFeMg ∂DFeMn   ∂cFe   ∂ 2 cFe 

 ∂x  
 ∂t  ∂x   ∂x 
2
∂x  ∂x 
      DFeFe DFeMg DFeMn  2
 ∂cMg   ∂DMgFe ∂DMgMg ∂DMgMn   ∂cMg    ∂ c 
 ∂x + DMgFe DMgMg DMgMn   Mg 
 ∂t  ∂x ∂x   ∂x  
2
      DMnFe   ∂x 
DMnMg DMnMn   2
 ∂cMn   ∂DMnFe ∂DMnMg ∂DMnMn   ∂cMn   ∂c 
 ∂x  Mn 
 ∂t 
 ∂x ∂x   ∂x  2
 ∂x 
which can be solved simply with a finite difference approach, as described in more detail in
Spear’s MSA monograph (1993) and in Ganguly’s EMU Notes in Mineralogy (2002) and
Reviews in Mineralogy and Geochemistry (2010) chapters.
Although the numerical solution is straightforward, constraining the initial, pre-diffusion
profile is often difficult. Typically, initial profiles are either drawn subjectively, based upon the
remnant preserved zoning and a petrologically and geometrically reasonable estimate of what
the ‘undiffused’ profile must have looked like (e.g., Dachs and Proyer 2002; Ague and Baxter
2007), or are assumed to be insignificantly zoned at peak temperature and develop zoning on
rims due to changing equilibrium conditions during exhumation and cooling (e.g., O’Brien and
Vrána 1995; Ganguly et al. 2000; Storm and Spear 2005). Kohn and Penniston-Dorland (2017,
this volume) review such retrograde resetting in detail. A final constraint on initial profile
comes from incorporation of information from equilibrium thermodynamic calculations to
generate model profiles that would form during growth along a given P–T path (e.g., Florence
and Spear 1991, 1993; Gaidies et al. 2008b; Caddick et al. 2010; Galli et al. 2011).
In all cases, the timescales retrieved by diffusion speedometry are only as good as (i) the
initial profile modeled, (ii) the diffusivity data applied, (iii) the appropriate boundary conditions
being set on the model, (iv) knowledge of the P–T and fO2 conditions that the sample experienced.
Garnet-based speedometry is probably most useful for defining maximum timescales of
metamorphic heating that are consistent with retention of observed chemical zoning, thus
yielding minimum permissible rates. Most diffusion occurs at or near to Tmax and a characteristic
T of diffusion can be quantified for a complex P–T history accordingly (e.g., Chakraborty and
Ganguly 1991). This means that speedometry is generally most sensitive to the duration spent near
peak temperature, with more subtle information required to infer lower temperature metamorphic
rates. Methodologies are also generally incapable of distinguishing single long ‘events’ from
multiple, briefer pulses, typically only revealing the total integrated D(T)·t (e.g., Ague and Baxter
2007; Kohn and Penniston-Dorland 2017, this volume). Thus without additional textural or
compositional (thermodynamic) constraints there can be significant non-uniqueness in results,
with strong, negative correlation between apparent peak temperature and apparent metamorphic
duration. An example of this (from Galli et al. 2011) is described in the following section, which
draws together aspects of both the ‘petro’ and ‘chrono’ of garnet into true petrochronology.
EXAMPLES OF PETRO-CHRONOLOGY OF GARNET

Figure 2 of this chapter provided a list of techniques to extract the ‘petro-’ of garnet and a list
of techniques to extract the ‘chrono’ of garnet. The first two sections of the chapter have detailed
those methodologies. Now, the stage is set for their integration into true garnet petrochronology.
Below, we have chosen five examples from the literature of the past decade that show the great
potential of garnet petrochronology. Other good examples exist (most of them have already
been cited elsewhere in this chapter) and many more are in progress. Our hope is that, through
these examples and our template in Figure 2, the reader will appreciate how to conceive of a
garnet petrologic application, appreciate the unique and complementary value of the information
extracted therefrom, and independently decide that it is worthy of the time and effort required.
Petrochronology of garnet: High pressure crustal metamorphism
Cheng et al. (2013; 2016) and Cheng and Cao (2015) present an impressive garnet
petrochronologic dataset from eclogite samples of the Huwan shear zone in the northwestern
boundary of the Hong-an orogen (Western Dabie). This suite of papers represents an
excellent example of the power of garnet petrochronology via: 1) combined Lu–Hf and
Sm–Nd geochronology, 2) zoned garnet geochronology, 3) P–T constraints on garnet growth
via thermodynamic modeling, and 4) comparison to zircon ages from the same sample
suites. Cheng et al. (2013) presented combined Lu–Hf and Sm–Nd geochronology from the
same handpicked eclogite garnet aliquots. Multi-point isochrons (Fig. 17) show remarkable
consistency yielding 260.4 ± 2.0 Ma (n = 10; MSWD = 1.4) for Sm–Nd and 260.0 ± 1.0 (n = 9;
MSWD = 1.0) for Lu–Hf. The tight agreement of each isochron data point, and between
Sm–Nd and Lu–Hf ages are indicative of successful geochronology. The agreement between
Lu–Hf and Sm–Nd ages itself is noteworthy, contradicting the oft-cited generalization that
Lu–Hf ages are older than Sm–Nd ages. Here, according to the authors, handpicking efforts
appear to have preferentially removed the Lu-rich garnet cores (that would normally skew
a bulk crystal Lu–Hf age toward the early stages of garnet growth); thus both Lu–Hf and
Sm–Nd ages date the same rim portion of the garnet. Eclogite facies mineral inclusions
and P–T estimates from garnet rim chemistry pin the 260 Ma age to peak eclogite facies
conditions. However the story, and the innovative garnet petrochronology, does not end
there. The 260 Ma age of eclogite facies metamorphism is not reflected in zircon age data
from the same area, which instead cluster mostly around 310 Ma. Thus, the garnet records
an entire eclogite facies event that is apparently missed by the zircons. Subsequent papers
by Cheng and Cao (2015) and Cheng et al. (2016) presented zoned garnet chronology on
other eclogites from the field area to test whether garnet cores might in fact record an age
more consistent with the zircon data. In the 2015 paper, garnet cores yielded Lu–Hf ages
of 296.7 ± 3.8 Ma and rims gave both Lu–Hf and Sm–Nd ages of ~255 Ma. The Lu–Hf core
age likely represents a mix between ~310 Ma eclogitic garnet and ~255 Ma eclogitic garnet,
reflecting 40 million years of prolonged subduction zone metamorphic conditions between
1.9 and 2.4 GPa at 500 °C to 575 °C, as constrained by intersecting core and rim isopleths
in pseudosections. Finally, the 2016 paper revealed an even broader scatter of Lu–Hf data
from multiple microsawed zones in a large garnet megacryst spanning 400 to 250 Ma; these
ages matched age spectra of newly acquired zircon data. The data presented in this series
of papers shows a range of ages because each garnet bearing sample, and each zone of
individual garnet crystals, reflects and records a different stage in the complex evolution of
this terrane. Only through direct garnet petrochronology were these authors able to recognize
at least two distinct (ca. 400 Ma and ~310−255 Ma), and prolonged, episodes of subduction
zone metamorphism, thus motivating further study on convergent margin geodynamics.
0.291 0.517
A) B)
Garnets
0.516 Garnets
0.288
Nd/144Nd
Hf/177Hf
0.515
260.0 ± 1.0 Ma
176
0.514
260.4 ± 2.0 Ma
143
0.285 MSWD = 1.0

bomb WR bomb WR MSWD = 1.4
0.513
Omp
Lu-Hf Sm-Nd
saw WR
Omp
0.282 0.512
0 0.4 0.8 1.2 1.6 0 0.4 0.8 1.2 1.6 2.0 2.4
147
176
Lu/177Hf Sm/143Nd
Figure 17. Lu–Hf and Sm–Nd multi-point garnet isochrons from the same garnet-bearing eclogite (modi-
fied from Cheng et al 2013). Note the perfect agreement of the two ages. When Lu–Hf and Sm–Nd are used
to date the exact same age generation of garnet (in this case the rim) the two chronometers agree. Later
work by Cheng et al (2016) found that garnet cores from other eclogites in this field area indeed contain an
older generation of garnet. All garnet growth generations were linked to P–T conditions via thermodynamic
modeling of garnet chemical isopleths, revealing multiple subduction episodes. “Bomb-WR” indicates
sample in which the whole rock was fully dissolved in a Parr bomb.
Petrochronology of garnet: Geospeedometry and the timescales of granulite facies

metamorphism
A primary motivation for deciphering metamorphic P–T paths is that knowledge of P–T
evolution can help to reveal processes that control the evolution of plate margins and influence
heating and cooling of Earth’s crust. Garnet petrochronology helped to address this in the
case of granulite facies rocks from the Gruf Complex, eastern Central Alps. Assemblages in
these rocks have long been thought to record substantially higher temperatures than in many
surrounding lithologies (e.g., Droop and Bucher-Nurminen 1984) and they do not correlate
easily with simple tectonic models of Alpine evolution. In a study of Gruf granulites and
charnockites, Galli et al. (2011) constrained reaction sequences through careful textural
analysis before estimating P–T conditions by (i) combining experimental constraints to
determine the position of key reactions, (ii) mineral thermobarometry, (iii) construction of
appropriate pseudosections for peak metamorphic conditions, and (iv) multiphase mineral
thermobarometry to constrain the equilibration P–T of late-stage coronae and symplectites.
Results suggest peak temperatures in excess of 900 ºC at ~9 kbar, with subsequent
symplectite formation at closer to 720–740 ºC and ~7 kbar (Fig. 18a). Intriguingly, garnet
crystals retain remnant ‘prograde’ chemical zoning (dashed curves in Fig. 18d), raising
the question of how such high temperatures were achieved, how briefly they must have
been maintained for, and what tectonic mechanisms were responsible. Zircons from Gruf
charnockites contain Permian (~290–260 Ma) cores overgrown with Tertiary (~34–29 Ma)
rims, with orthopyroxene inclusions implying that core ages record or post-date granulite
facies metamorphism (Galli et al. 2012). Galli et al. (2011) thus described a two stage model,
with ~900 ºC Permian metamorphism overprinted by symplectite formation during Alpine
orogenesis (Fig. 18b-c). A coupled garnet growth and diffusion model was constructed for an
assumed P–T path that experiences both events, using an appropriate bulk-rock composition
and thermodynamic constraints to establish growth zoning during prograde metamorphism.
Results were inconsistent with observed garnet zoning unless the initial event was rapid,
experiencing ultra-high temperature conditions for less than ~1 Myrs (Fig. 18d). Typical
Alpine metamorphic timescales of several tens of millions of years result in complete loss
of prograde zoning if peak T is ~900 ºC, though a rapid early 900 ºC event overprinted by
tens of millions of years of Alpine metamorphism reaching more typical peak temperatures
for the region (~ 720 ºC) yielded relatively good fits to measured garnet compositions. In this
petrochronology study, Galli et al. (2011) thus concluded that granulite facies temperatures
were only sustained for a very brief period and were probably unrelated to Alpine collisional
tectonics. Rocks then resided in the mid crust for ~250 Myrs, before experiencing exhumation
and a second phase of metamorphism at conditions and rates compatible with interpretations
from elsewhere in the region, possibly due to mechanical interaction with melts that now
form an adjacent pluton (Galli et al. 2013).
Petrochronology of garnet: Timescales of lower crustal melting
Stowell et al. (2010, 2014) provide excellent examples of the use of petrochronology
in understanding the timescales and conditions of high-temperature metamorphism, partial
melting, and the generation of high Sr/Y magmas in the lower arc crust of Fiordland, NZ.
Stowell et al. (2010) combined Sm–Nd garnet geochronology on drilled cores and rims,
trace element zoning in garnet, phase equilibria modeling and U–Pb zircon geochronology to
provide an integrated history of a short duration, high-pressure melting event recorded by the
Pembroke Granulite. Garnet core and rim ages were indistinguishable within each sample (e.g.,
core age of 123.5 ± 2.1 Ma; rim age of 122.5 ± 2.1 Ma for a dioritic gneiss), attesting to the short
duration of garnet growth. The duration of garnet growth and subsequent partial melting event
12 12
a) b)
UHT Perian
metamorphism
(~ 272 Ma)
10 10 Granulite PT conditions
from Droop and
Bucher-Nurminen (1984)
~ 32 Ma Alpine
Pressure, kbar
Pressure, kbar
8 8 metamorphism
Granulite PT
Thermocalc, average PT conditions from Galli
et al. (2011)
Best estimate post-peak Alpine
6 conditions consistent with 6 symplectite PT
the central Alpine conditions from
Lepontine metamorphism Galli et al. (2011)
(from literature sources)
4 4 Two metamorphic cycles (Galli et al.,
2011)
Equilibria associated with Symplectite PT conditions
symplectite-forming from Droop and Bucher- Single Alpine metamorphic cycle (e.g.
reactions Nurminen( 1984) Droop and Bucher-Nurminen, 1984)
2 2
600 700 800 900 1000 700 800 900 1000
12 Growth profile based on equilibrium Path type 1: Rapid heating to b,
c) d) thermodynamic prediction for path a-b 250 Myr cooling to c, 20 Myr cooling to d
(without diffusive relaxation)
molefraction component
10 0.8 0.8
b
Pressure, kbars
8 0.6 0.6
Mg
Fe Fe
a Mg
6 e c 0.4 0.4
natural Mg
d note that temperatures in the range 0.2 0.2 natural Ca
4 450 ˚C – 650 ˚C were tested for point e Ca
0 Mn 0
400 500 600 700 800 900 1000 core 500 1000 1500 core 500 1000 1500
Temperature, ˚C
Path type 2: 1 Myr ‘granulite event’ (a–e), Path type 2: 1 Myr ‘granulite event’ (a–e),
P-T paths investigated in diffusion study 240 Myr at e = 450 ˚C, 240 Myr at e = 650 ˚C,
Initial heating to ‘granulite conditions’ (curve a–b) 30 Myr ‘Alpine event’ (e–d) 30 Myr ‘Alpine event’ (e–d)
0.8
0.8
followed by:
Path type 1: Slow (250 Myrs) cooling to ‘Alpine 0.6 0.6
conditions’ (black dashed curve b–c) then 20 Myrs Fe Fe
cooling to 400˚C (dashed curve c–d) 0.4 0.4
Mg Mg
Path type 2: Rapid cooling to low temperature (gray
0.2
curve b–e), 240 Myrs residence at ‘e’, then reheating Ca
0.2
(e–c), exhumation and cooling (c–d) at rates and P-T Ca
0 Mn Mn
conditions consistent with other Alpine studies. Note 0
that the temperature of point ‘e’ is unkown, so various core 500 1000 1500 core 500 1000 1500
possibilities were tested). Radial distance from crystal core, µm Radial distance from crystal core, µm
Figure 18. Evolution of the Gruf complex modified from Galli et al (2011). A. Calculated locations of observed
reactions in coronae and symplectites. B. P–T estimate for corona formation in the context of the peak P–T
experienced, with age constraints where available. C. P–T path types for which model results are shown in
subsequent panels. E. Garnet growth and diffusion model results showing: core-to-rim zoning in the absence
of intra-crystalline diffusion in garnet; ‘type 1’ path experiencing slow cooling from peak temperature; ‘type 2’
path experiencing rapid UHT conditions followed by a long duration at 450 ˚C, then a 30 Ma ‘Alpine metamor-
phic event’; ‘type 2 path’ but with intermediate residence at 650 ˚C. In each case, the solid curves are model
results and the dashed curves show natural crystal zoning. See Galli et al (2011) for more details.
is constrained by comparing bulk garnet and zircon ages from a gabbroic gneiss showing no
evidence of melting (garnet age of 126.1 ± 2.0 Ma) to bulk garnet and zircon ages from a dioritic
gneiss that shows melting evidence (a garnet reaction zone and leucosome; peritectic garnet
age of 122.6 ± 2.0 Ma). Well-preserved HREE zoning in the dioritic gneiss was attributed to the
continued consumption of accessory phases during garnet growth. Phase equilibria modeling
of the gabbroic gneiss indicates peak metamorphic conditions at 1.1−1.4 GPa and 680−815 ºC,
with garnet growth occurring during a > 0.5 GPa increase. Intrusion of the Western Fiordland
Orthogneiss (125−115 Ma) is synchronous with the granulite-facies event, indicating that
voluminous arc magmatism may have triggered garnet growth and partial melting.
The combined geochronologic and thermodynamic approach allowed for the interpretation
that thickening of arc crust resulted in high-pressure metamorphism and that granulite-facies
garnet growth lasted 3−7 Ma, with subsequent partial melting occurring shortly thereafter,
perhaps as little as 3 Ma later.
Stowell et al. (2014) dated an additional nine granulite-facies rocks from around the
Malaspina pluton using Sm–Nd garnet geochronology, and obtained ages ranging from
115.6 ± 2.6 Ma to 110.6 ± 2.0 Ma, roughly 10 Ma later than that from the Pembroke Granulite
(at conditions of ~920 ºC and 1.4−1.5 GPa), suggesting that high-pressure metamorphism and
partial melting was diachronous over a > 3000 km2 area of the mid- to lower crust. Stowell et al.
(2014) suggest this may result from pulsed underplating of magma in the lower crust or ‘drip-
style’ delamination of the lowermost crust. High-precision garnet geochronology allowed
resolution of the diachronous nature of high-pressure metamorphism and partial melting, and
when combined with constraints on the P–T conditions (and P–T path) during garnet growth,
allows for interpretation of the regional tectonic framework upon loading, partial melting, and
extensional collapse of the Fiordland arc crust.
Petrochronology of garnet: Subduction zone dehydration
Garnet growth typically results from the breakdown of hydrous phases during subduction
zone metamorphism, and can thus be a powerful proxy for subduction zone devolatilization
(Baxter and Caddick 2013). Dragovic et al. (2012, 2015) utilized zoned Sm–Nd garnet
geochronology coupled with phase equilibria modeling in order to constrain the duration
and rate of devolatilization during subduction of mafic and felsic lithologies, respectively,
from the Cycladic Blueschist Unit of Sifnos, Greece. These papers combined some of the
techniques described in previous sections (i.e., zoned geochronology and pseudosections), but
also highlighted and resolved some of the challenges associated with Sm–Nd isotopic analyses
of very small sample sizes. Dragovic et al. (2012) microdrilled three chemically distinct,
concentric zones from each of two separate garnet crystals and dated them using Sm–Nd. All
six garnet fractions could be placed onto a single isochron (see Fig. 11), indicating to a first
order that garnet growth was very brief. When separated into three isochrons, the duration of
garnet growth was calculated to be 0.04 Ma, with a 2σ maximum duration of ~1 Ma. Dragovic
et al. (2015) took this much further, microdrilling and dating ten growth zones from each of
two garnet crystal which revealed three distinct pulses of crystal growth, with the final pulse
growing a majority of the garnet and occurring in < 0.8 Ma (Fig. 19a).
a) b) Pressure (GPa)
1st 2nd 3rd 0.5 1.0 1.5 2.0 2.1 2.15 2.19
4.5 1.2 5.0
garnet
garnet
water
Weight % water stored in rock
Garnet abundance observed
water
Garnet abundance predicted
1.0
4.0
Observed Predicted
earliest possible growth
3.0 0.8
3.0
0.6
2.0
1.5 Grt1 0.4
pulse!
Grt2 1.0
0.2
0.0
250
300
350
400
450
500
550
58 54 50 46 42
Age (Ma) Temperature (˚C)
Figure 19. Devolatilization and garnet growth history recorded in zoned garnet porphyroblasts from Sif-
nos, Greece (modified from Dragovic et al. 2015). A. Volume abundance of garnet over time showing
pulses of garnet growth in two garnet crystals from nineteen microdrilled growth zones. The final pulse
lasted < 0.8 Ma. B. Associated fluid release as modeled using a path dependent thermodynamic forward
model. The aforementioned final pulse released > 0.5 wt.% fluid from the bulk rock.
In Dragovic et al. (2012), the P–T conditions for the interval of garnet growth were calculated
with separate pseudosections for whole rock (for initiation of growth) and ‘garnet depleted
rock’ (for the termination of growth) bulk compositions. Results imply initiation of crystal
core growth at ~2.0 GPa and ~460 ºC and termination of rim growth ~2.2 GPa and ~560 ºC.
In Dragovic et al. (2015), a path dependent thermodynamic forward model was employed in
which phase equilibria were calculated at 0.5 ºC intervals along a prescribed P–T path, with
stable garnet and fluid fractionated from the effective bulk composition at every interval. In both
contributions, the change in bulk rock water content during the < 1 Ma garnet growth interval
was calculated at ~ 0.3–0.5 wt%. The path dependent thermodynamic model was especially
useful here, simultaneously 1) discerning the likely subduction P–T path, 2) modeling observed
growth pulses of garnet, and 3) predicting the evolution of fluid release (Fig. 19b).
Both studies thus suggested a pulse of fluid release and heating (~100 ºC/Ma) spanning
just hundreds of thousands of years. The high spatial resolution attainable by zoned garnet
geochronology, the ability to measure smaller samples sizes (thinner growth zones), and
the coupling with thermodynamic constraints gave the ability to distinguish whether garnet
growth, and thus metamorphic devolatilization, was temporally continuous or focused into
pulses of mineral growth and fluid production.
Petrochronology of garnet: Collisional tectonics and inverted metamorphic gradients
An improved understanding of the evolution of the well-known Himalayan inverted
metamorphic gradient has been a goal of tectonic studies for several decades. The Lesser
Himalayan sequence in Sikkim exposes an unusually complete section through the inverted
Barrovian sequence, and a remarkable group of studies have repeatedly analyzed a suite
of samples with various techniques to build a detailed picture of the evolution of these
rocks. For example, sample 24/99 is an aluminous garnet schist from the Lesser Himalaya,
first described by Dasgupta et al. (2009) and representing the low temperature base of the
inverted metamorphic sequence. Dasgupta et al. (2009) subjected 24/99 to element mapping
and quantitative microprobe analysis, using garnet–biotite thermometry and two methods
of multiphase thermobarometry to infer that it reached peak conditions of ~ 525 ºC, 5 kbar.
Pseudosection constraints suggested a somewhat higher peak temperature but confirmed
texturally-derived interpretations that garnet grew at the expense of chlorite and chloritoid.
Garnets from all of the Barrovian zones were dated with Lu–Hf at very high precision
in order to constrain the temporal evolution of the sequence. For example, garnet in sample
24/99 was dated using Lu–Hf at 10.6 ± 0.2 Ma, suggesting that it experienced the youngest
garnet growth in the sampled transect (Anczkiewicz et al. 2014). In the same study, zoned
Lu–Hf geochronology on a kyanite grade sample implied garnet core growth at 13.7 ± 0.2 Ma
and rim growth at 9.9 ± 3.8 Ma (Anczkiewicz et al. 2014). Taken together, data constrain the
duration of prograde metamorphism (using first garnet growth as a marker) across the Lesser
Himalayan sequence, showing that ages decrease towards structurally lower levels (and lower
metamorphic grade) and suggesting that the garnet isograd swept through the entire sequence
from garnet-grade to sillimanite-grade in ~6 Myrs.
The size, shape and distribution of every garnet crystal in a several cubic centimeter block
of sample 24/99 was then mapped in 3-D (Gaidies et al. 2015). Analysis of results suggested
clustering and no spatial ordering of garnet, providing little evidence for diffusion-controlled
crystallization and suggesting that garnet nucleation and growth could have been controlled by
crystal interface processes (Gaidies et al. 2015). Thin sections were cut through the geometric
cores of the largest garnet crystals and were carefully characterized to define inclusion
mineralogy and fabrics, and to quantify major element zoning profiles. Thermodynamic
models helped to refine P–T paths, and garnet diffusion speedometry gave constraints on both
the heating and cooling rates, though the extent of diffusion was limited in 24/99 because it did
not reach a particularly high maximum temperature.
This entire remarkable dataset is shown to be consistent with models of a coherent Lesser
Himalayan block being buried during collision (Anczkiewicz et al. 2014; Chakraborty et al.
2016). All rocks reached their respective peak metamorphic conditions simultaneously but
with those samples that eventually reach highest grade passing through garnet-in reactions first
and thus recording earliest garnet growth (Anczkiewicz et al. 2014). This truly represents an
integrated garnet petrochronology application.
OUTLOOK
Both the ‘petro’ and the ‘chrono’ of garnet have evolved significantly over the past
several decades, and that evolution will continue as analytical methodologies improve and as
the creativity of the scientific applications grows. We have touched on many aspects of that
evolution here, showcasing a few examples of the current state of the art garnet petrochronology.
Once, we were barely able to see the ‘tree rings’ of garnet growth. If the concentric growth rings
in a garnet crystal can be likened to the pages of a history book (as suggested in Baxter and
Scherer 2013), while we first only got to read the title, and then progressed to reading the title
of each chapter, we are now able to read the record with the resolution of several pages. New
petrologic techniques (thermodynamic, isotopic, textural, mechanical) are like learning new
vocabulary that allows us to extract greater meaning and context from every chapter and page
of that garnet chronology. What will the future hold? How soon and with what new analytical
breakthroughs will we be able to read every single word of that garnet tree ring history book?
Garnet petrochronology is not a panacea; its limitations and challenges must also be understood
to fully harness its potential and it is quite possible that we will never be fully satisfied with
the resolution of the story preserved in metamorphic rocks, or our ability to read that story.
But as Lu–Hf and Sm–Nd geochronology become increasingly precise and accessible, and
our understanding of the processes that embed chemical and isotopic signatures in crystals
improve, garnet will surely become an ever more important member of the cadre of modern
petrochronometers discussed in this RiMG volume. So although we have dwelled mostly here
on previous examples and current methods, above all we encourage the reader to consider the
potential of garnet petrochronology in deciphering the evolution of crustal processes.
ACKNOWLEDGMENTS
EFB gratefully acknowledges NSF award 1250497/1561882. MJC gratefully acknowledges
NSF awards EAR−1250470 and EAR−1447568 (to MJC and BD). We thank E.E. Scherer for
providing data in Figure 10 as well as numerous discussions on garnet geochronology. We thank
Robert Anczkiewicz, David Pattison, and Pierre Lanari for valuable reviews, Matthew Kohn
for his feedback, and Martin Engi for his patient, detailed and constructive editorial handling.
We thank R.J. Tracy for providing the materials for Figure 3, for patiently emphasizing the
importance of ‘chemo-petrography’, and for guiding our collection and interpretation of
countless ‘chemo-petrographic’ images. We thank John Rosenfeld for providing his ‘favorite
rotated garnet photo’ from an original print for Figure 1c.
REFERENCES
Abart R (1995) Phase-equilibrium and stable-isotope constraints on the formation of metasomatic garnet–
vesuvianite veins (SW Adamello, N Italy). Contrib Mineral Petrol 122:116–133
Aerden D, Bell TH, Puga E, Sayab M, Lozano JA, de Federico AD (2013) Multi-stage mountain building vs.
relative plate motions in the Betic Cordillera deduced from integrated microstructural and petrological
analysis of porphyroblast inclusion trails. Tectonophysics 587:188–206
Ague JJ (2011) Extreme channelization of fluid and the problem of element mobility during Barrovian
metamorphism. Am Mineral 96:333–352
Ague JJ, Axler JA (2016) Interface coupled dissolution–reprecipitation in garnet from subducted granulites and
ultrahigh-pressure rocks revealed by phosphorous, sodium, and titanium zonation. Am Mineral 101:1696–1699
Albee AL (1965) Phase equilibria in three assemblages of kyanite-zone pelitic schists, Lincoln Mountain
Quadrangle, central Vermont. J Petrol 6:246–301
Albee AL (1972) Metamorphism of pelitic schists: reaction relations of chloritoid and staurolite. Geol Soc Am
Bull 83:3249–3268
Amato JM, Johnson CM, Baumgartner LP, Beard BL (1999) Rapid exhumation of the Zermatt–Saas ophiolite
deduced from high-precision Sm–Nd and Rb–Sr geochronology. Earth Planet Sci Lett 171:425–438
Anczkiewicz R, Thirlwall MF (2003) Improving precision of Sm–Nd garnet dating by H2SO4 leaching: a simple
solution to the phosphate inclusion problem. Geol Soc London Spec Pub 220:83–91
Anczkiewicz R, Platt JP, Thirlwall MF, Wakabayashi J (2004) Franciscan subduction off to a slow start: evidence
from high-precision Lu–Hf garnet ages on high grade-blocks. Earth Planet Sci Lett 225:147–161
Anczkiewicz R, Szczepanski J, Mazur S, Storey C, Crowley Q, Villa IM, Thirlwall ME, Jeffries TE (2007) Lu–Hf
geochronology and trace element distribution in garnet: Implications for uplift and exhumation of ultra-high
pressure granulites in the Sudetes, SW Poland. Lithos 95:363–380
Anczkiewicz R, Thirlwall M, Alard O, Rogers NW, Clark C (2012) Diffusional homogenization of light REE in
garnet from the Day Nui Con Voi Massif in N-Vietnam: Implications for Sm–Nd geochronology and timing
of metamorphism in the Red River shear zone. Chem Geol 318:16–30
Anczkiewicz R, Chakraborty S, Dasgupta S, Mukhopadhyay D, Kołtonik K (2014) Timing, duration and inversion
of prograde Barrovian metamorphism constrained by high resolution Lu–Hf garnet dating: A case study from
the Sikkim Himalaya, NE India. Earth Planet Sci Lett 407:70–81
Anderson DE, Olimpio JC (1977) Progressive homogenization of metamorphic garnets, south Molnar, Scotland:
Evidence for volume diffusion. Can Mineral 15:205–216
Angel RJ, Mazzucchelli ML, Alvaro M, Nimis P, Nestola F (2014) Geobarometry from host–inclusion systems:
The role of elastic relaxation. Am Mineral 99:2146–2149
Angiboust S, Agard P, Raimbourg H, Yamato P, Huet B (2011) Subduction interface processes recorded by
eclogite-facies shear zones (Monviso, W. Alps). Lithos 127:222–238
Angiboust S, Pettke T, De Hoog JCM, Caron B, Oncken O (2014) Channelized fluid flow and eclogite-facies
metasomatism along the subduction shear zone. J Petrol 55:883–916
Argles TW, Prince CI, Foster GL, Vance D (1999) New garnets for old? Cautionary tales from young mountain
belts. Earth Planet Sci Lett 172:301–309
Ashley KT, Caddick MJ, Steele-MacInnis M, Bodnar RJ, Dragovic B (2014a) Geothermobarometric history of
subduction recorded by quartz inclusions in garnet. Geochem Geophys Geosystem 15:350–360
Ashley KT, Steele-MacInnis M, Caddick MJ (2014b) QuIB Calc: A MATLAB® script for geobarometry based on
elastic modeling of quartz inclusions in garnet. Comp and Geosci 60:155–157
Ashley KT, Darling RS, Bodnar RJ, Law RD (2015) Significance of “stretched” mineral inclusions for
reconstructing P–T exhumation history. Contrib Mineral Petrol 169:1–9
Atherton MP (1968) The variation of garnet, biotite and chlorite composition in medium grade pelitic rocks from
the Dalradian, Scotland, with particular reference to the zonation in garnet. Contrib Mineral Petrol 18:347–371
Atherton MP, Edmunds WM (1965) An electron microprobe study of some zoned garnets from metamorphic
rocks. Earth Planet Sci Lett 1:185–193
Baldwin JA, Powell R, White RW, Štípská P (2015) Using calculated chemical potential relationships to account
for replacement of kyanite by symplectite in high pressure granulites. J Metamorph Geol 33:311–330
Barrow G (1893) On an intrusion of muscovite–biotite gneiss in the east Highlands of Scotland, and its
accompanying metamorphism. Quart J Geol Soc London 49:330–358
Barrow G (1912) On the geology of lower Deeside and the southern Highland border. Proc Geol Assoc 23:268–284
Bast R, Scherer EE, Sprung P, Fischer-Godde M, Stracke A, Mezger K (2015) A rapid and efficient ion-exchange
chromatography for Lu–Hf, Sm–Nd, and Rb–Sr geochronology and the routine isotope analysis of sub-ng
amounts of Hf by MC-ICP-MS. J Anal At Spectrom 30:2323–2333
Baumgartner LP, Valley JW (2001) Stable isotope transport and contact metamorphic fluid flow. Rev Mineral
Geochem 43:415–467
Baumgartner LP, Floess D, Podladchikov Y, Foster CT (2010) Pressure gradients in garnets induced by diffusion
relaxation of major element zoning. Geol Soc Denver Ann Meeting
Baxter EF, DePaolo D (2002a) Field measurement of high temperature bulk reaction rates II: Interpretation of
results from a field site near Simplon Pass, Switzerland. Am J Sci 302:465–516
Baxter EF, DePaolo DJ (2002b) Field measurement of high temperature bulk reaction rates I: Theory and technique.
Am J Sci 302:442–464
Baxter EF, Caddick MJ (2013) Garnet growth as a proxy for progressive subduction zone dehydration. Geology
41:643–646
Baxter EF, Scherer EE (2013) Garnet Geochronology: Timekeeper of Tectonometamorphic Processes. Elements
9:433–438
Baxter EF, Ague JJ, Depaolo DJ (2002) Prograde temperature–time evolution in the Barrovian type-locality
constrained by Sm/Nd garnet ages from Glen Clova, Scotland. J Geol Soc London 159:71–82
Baxter EF, Caddick MJ, Ague JJ (2013) Garnet: common mineral, uncommonly useful. Elements 9:415–419
Bebout GE, Tsujimori T, Ota T, Shimaki Y, Kunihiro T, Carlson WD, Nakamura E (2015) Lithium in UHP
metasedimentary garnets from Lago Di Cignana, Italy: Coupled substitutions and the role of accessory
phases. Geol Soc Am Ann Meeting
Bell TH (1985) Deformation partitioning and porphyroblast rotation in metamorphic rocks: a radical reinterpretation.
Berman RG (1990) Mixing properties of Ca–Mg–Fe–Mn garnets. Am Mineral 75:328–344
Berman RG, Aranovich LY (1996) Optimized standard state and solution properties of minerals. I. Model
calibration for olivine, orthopyroxene, cordierite, garnet, and ilmenite in the system FeO–MgO–CaO–Al2O3–
TiO2–SiO2. Contrib Mineral Petrol 126:1–24
Bickle MJ, Baker J (1990) Advective diffusive transport of isotopic fronts—an example from Naxos, Greece. Earth
Bloch E, Ganguly J (2015) 176Lu–176Hf geochronology of garnet II: numerical simulations of the development
of garnet–whole-rock 176Lu–176Hf isochrons and a new method for constraining the thermal history of
metamorphic rocks. Contrib Mineral Petrol 169:1–16
Bloch E, Ganguly J, Hervig R, Cheng W (2015) 176Lu–176Hf geochronology of garnet I: experimental determination
of the diffusion kinetics of Lu3+ and Hf4+ in garnet, closure temperatures and geochronological implications.
Bohlen SR, Essene EJ (1980) Evaluation of coexisting garnet–biotite, garnet–clinopyroxene and other Mg–Fe
exchange thermometers in Adirondack granulites. Geol Soc Am Bull 91:685–719
Bohlen SR, Liotta JJ (1986) A barometer for garnet amphibolites and garnet granulites. J Petrol 27:1025–1034
Bohlen SR, Wall VJ, Boettcher AL (1983) Experimental investigations and gelogical applications of equilibria in
the system FeO–TiO2–Al2O3–SiO2–H2O. Am Mineral 68:1049–1058
Brady JB, Cherniak DJ (2010) Diffusion in minerals: an overview of published experimental diffusion data. Rev
Burton KW, Onions RK (1991) High-resolution garnet chronometry and the rates of metamorphic processes. Earth
Burton KW, Kohn MJ, Cohen AS, Onions RK (1995) The relative diffusion of Pb, Nd, Sr and O in garnet. Earth
Caddick MJ, Thompson AB (2008) Quantifying the tectono-metamorphic evolution of pelitic rocks from a wide
range of tectonic settings: Mineral compositions in equilibrium. Contrib Mineral Petrol 156:177–195
Caddick MJ, Bickle MJ, Harris NBW, Holland TJB, Horstwood MSA, Ahmad T (2007) Burial and exhumation
history of a Lesser Himalayan schist: Recording the formation of an inverted metamorphic sequence in NW
India. Earth Planet Sci Lett 264:375–390
Caddick MJ, Konopásek J, Thompson AB (2010) Preservation of garnet growth zoning and the duration of
prograde metamorphism. J Petrol 51:2327–2347
Cahalan RC, Kelly ED, Carlson WD (2014) Rates of Li diffusion in garnet: Coupled transport of Li and Y + REEs.
Am Mineral 99:1676–1682
Carlson WD (1989) The significance of intergranular diffusion to the mechanisms and kinetics of porphyroblast
crystallization. Contrib Mineral Petrol 103:1–24
Carlson WD (1999) The case against Ostwald ripening of porphyroblasts. Can Mineral 37:403–414
Carlson WD (2002) Scales of disequilibrium and rates of equilibration during metamorphism. Am Mineral 87:185–204
Carlson WD (2006) Rates of Fe, Mg, Mn and Ca diffusion in garnet. Am Mineral 91:1–11
Carlson WD (2011) Porphyroblast crystallization: linking processes, kinetics, and microstructures. Int Geol Rev
53:406–445
Carlson WD (2012) Rates and mechanism of Y REE, and Cr diffusion in garnet. Am Mineral 97:1598–1618
Carlson WD, Denison C, Ketcham RA (1995) Controls on the nucleation and growth of porphyroblasts: Kinetics
from natural textures and numerical models. Geol J 30:207–225
Carlson WD, Gale JD, Wright K (2014) Incorporation of Y and REEs in aluminosilicate garnet: Energetics from
atomistic simulation. Am Mineral 99:1022–1034
Carlson WD, Hixon JD, Garber JM, Bodnar RJ (2015a) Controls on metamorphic equilibration: the importance of
intergranular solubilities mediated by fluid composition. J Metamorph Geol 33:123–146
Carlson WD, Pattison DR, Caddick MJ (2015b) Beyond the equilibrium paradigm: How consideration of kinetics
enhances metamorphic interpretation. Am Mineral 100:1659–1667
Cashman KV, Ferry JM (1988) Crystal size distribution (CSD) in rocks and the kinetics and dynamics of
crystallization. 3. Metamorphic crystallization. Contrib Mineral Petrol 99:401–415
Castro AE, Spear FS (2016) Reaction overstepping and re-evaluation of peak P–T conditions of the blueschist unit
Sifnos, Greece: implications for the Cyclades subduction zone. Int Geol Rev:1–15
Chacko T, Cole DR, Horita J (2001) Equilibrium oxygen, hydrogen and carbon isotope fractionation factors
applicable to geologic systems. Rev Mineral Geochem 43:1–81
Chakraborty S, Ganguly J (1991) Compositional zoning and cation diffusion in garnets. In: Diffusion, Atomic
Ordering, and Mass Transport. Ganguly J (ed) Springer-Verlag, New York, p 120–175
Chakraborty S, Anczkiewicz R, Gaidies F, Rubatto D, Sorcar N, Faak K, Mukhopadhyay D, Dasgupta S (2016) A
review of thermal history and timescales of tectonometamorphic processes in Sikkim Himalaya (NE India)
and implications for rates of metamorphic processes. J Metamorph Geol 34:785–803
Chambers J, Caddick M, Argles T, Horstwood M, Sherlock S, Harris N, Parrish R, Ahmad T (2009) Empirical
constraints on extrusion mechanisms from the upper margin of an exhumed high-grade orogenic core, Sutlej
valley, NW India. Tectonophysics 477:77–92
Cheng H, Cao D (2015) Protracted garnet growth in high-Peclogite: constraints from multiple geochronology
and P–T pseudosection. J Metamorph Geol 33:613–632
Cheng H, King RL, Nakamura E, Vervoort JD, Zhou Z (2008) Coupled Lu–Hf and Sm–Nd geochronology constrains
garnet growth in ultra-high-pressure eclogites from the Dabie orogen. J Metamorph Geol 26:741–758
Cheng H, Zhang C, Vervoort JD, Zhou Z (2013) New Lu–Hf and Sm–Nd geochronology constrains the subduction
of oceanic crust during the Carboniferous–Permian in the Dabie orogen. J Asian Earth Sci 63:139–150
Chen YX, Zheng YF, Li L, Chen RX (2014) Fluid–rock interaction and geochemical transport during protolith
emplacement and continental collision: A tale from Qinglongshan ultrahigh-pressure metamorphic rocks in
the Sulu orogen. Am J Sci 314:357–399
Cheng H, Liu XC, Vervoort JD, Wilford D, Cao DD (2016) Micro-sampling Lu–Hf geochronology reveals episodic
garnet growth and multiple high-P metamorphic events. J Metamorph Geol 34:363–377
Chernoff CB, Carlson WD (1997) Disequilibrium for Ca during growth of pelitic garnet. J Metamorph Geol
15:421–438
growth. Geology 27:555–558
Chopin C (1984) Coesite and pure pyrope in high-grade blueschists of the Western Alps: a first record and some
consequences. Contrib Mineral Petrol 86:107–118
Christensen JN, Rosenfeld JL, Depaolo DJ (1989) Rates of tectonometamorphic processes from rubidium and
strontium isotopes in garnet. Science 244:1465–1469
Christensen JN, Selverstone J, Rosenfeld JL, DePaolo DJ (1994) Correlation by Rb–Sr geochronology of garnet
growth histories from different structural levels within the Tauern Window, Eastern Alps. Contrib Mineral
Petrol 118:1–12
Chu X, Ague JJ (2015) Analysis of experimental data on divalent cation diffusion kinetics in aluminosilicate garnets
with application to timescales of peak Barrovian metamorphism, Scotland. Contrib Mineral Petrol 170
Clayton RN, Goldsmith JR, Mayeda TK (1989) Oxygen isotope fractionation in quartz, albite, anorthite and
calcite. Geochim Cosmochim Acta 53:725–733
Clechenko CC, Valley JW (2003) Oscillatory zoning in garnet from the Willsboro Wollastonite Skarn, Adirondack
Mts, New York: a record of shallow hydrothermal processes preserved in a granulite facies terrane. J
Coghlan RAN (1990) Studies in diffusional transport: grain boundary transport of oxygen in feldspars, diffusion of
oxygen, REE’s in garnet. Doctoral dissertation, Brown Univ.
Cohen AS, O’Nions RK, Siegenthaler R, Griffin WL (1988) Chronology of the pressure–temperature history
recorded by a granulite terrain. Contrib Mineral Petrol 98:303–311
Connolly JAD (1990) Multivariable phase diagrams: an algorithm based on generalized thermodynamics. Am J
Sci 290:666–718
Connolly JAD (2005) Computation of phase equilibria by linear programming: A tool for geodynamic modeling
and its application to subduction zone decarbonation. Earth Planet Sci Lett 236:524–541
Connolly JAD (2009) The geodynamic equation of state: What and how. Geochem Geophys Geosystem 10
implications for monazite formation. J Metamorph Geol 26:451–464
Crank J (1975) The Mathematics of Diffusion, 2nd Edition. Clarendon Press, Oxford
Crowe DE, Riciputi LR, Bezenek S, Ignatiev A (2001) Oxygen isotope and trace element zoning in hydrothermal
garnets: Windows into large-scale fluid-flow behavior. Geology 29:479–482
Cruz-Uribe AM, Hoisch TD, Wells ML, Vervoort JD, Mazdab FK (2015) Linking thermodynamic modelling,
Lu–Hf geochronology and trace elements in garnet: new P–T–t paths from the Sevier hinterland. J
Cutts K, Kinny P, Strachan R, Hand M, Kelsey D, Emery M, Friend C, Leslie A (2010) Three metamorphic events
recorded in a single garnet: Integrated phase modelling, in situ LA-ICPMS and SIMS geochronology from
the Moine Supergroup, NW Scotland. J Metamorph Geol 28:249–267
D’Errico ME, Lackey JS, Surpless BE, Loewy SL, Wooden JL, Barnes JD, Strickland A, Valley JW (2012) A
detailed record of shallow hydrothermal fluid flow in the Sierra Nevada magmatic arc from low-18O skarn
garnets. Geology 40:763–766
Dachs E, Proyer A (2002) Constraints on the duration of high-pressure metamorphism in the Tauern Window from
diffusion modelling of discontinuous growth zones in eclogite garnet. J Metamorph Geol 20:769–780
Daniel CG, Spear FS (1998) Three-dimensional patterns of garnet nucleation and growth. Geology 26:503–506
Dasgupta S, Chakraborty S, Neogi S (2009) Petrology of an inverted Barrovian sequence of metapelites in Sikkim
Himalaya, India: Constraints on the tectonics of inversion. Am J Sci 309:43–84
de Capitani C, Petrakakis K (2010) The computation of equilibrium assemblage diagrams with Theriak/Domino
software. Am Mineral 95:1006–1016
DePaolo DJ, Getty SR (1996) Models of isotopic exchange in reactive fluid–rock systems: Implications for
geochronology in metamorphic rock. Geochim Cosmochim Acta 60:3933–3947
DeWolf CP, Zeissler CJ, Halliday AN, Mezger K, Essene EJ (1996) The role of inclusions in U–Pb and Sm–Nd
garnet geochronology: Stepwise dissolution experiments and trace uranium mapping by fission track analysis.
Diener J, Powell R (2012) Revised activity–composition models for clinopyroxene and amphibole. J Metamorph
Geol 30:131–142
Petrol 40:259–274
Dorfler KM, Tracy RJ, Caddick MJ (2014) Late-stage orogenic loading revealed by contact metamorphism in the
northern Appalachians, New York. J Metamorph Geol 32:113–132
314–317:9–22
Dragovic B, Baxter EF, Caddick MJ (2015) Pulsed dehydration and garnet growth during subduction revealed by
zoned garnet geochronology and thermodynamic modeling, Sifnos, Greece. Earth Planet Sci Lett 413:111–122
Dragovic B, Guevara VE, Caddick MJ, Baxter EF, Kylander-Clark AR (2016) A pulse of cryptic granulite-
facies metamorphism in the Archean Wyoming Craton revealed by Sm–Nd garnet and U–Pb monazite
geochronology. Precambrian Res 283:24–49
Droop GTR, Bucher-Nurminen K (1984) Reaction textures and metamorphic evolution of sapphirine-bearing
granulites from the Gruf Complex, Italian Central Alps. J Petrol 25:766–8031
Ducea MN, Ganguly J, Rosenberg EJ, Patchett PJ, Cheng WJ, Isachsen C (2003) Sm–Nd dating of spatially
Duchene S, BlichertToft J, Luais B, Telouk P, Lardeaux JM, Albarede F (1997) The Lu–Hf dating of garnets and
the ages of the Alpine high-pressure metamorphism. Nature 387:586–589
Eiler JM (2001) Oxygen isotope variations of basaltic lavas and upper mantle rocks. Rev Mineral Geochem
43:319–364
Elphick SC, Ganguly J, Loomis TP (1981) Experimental study of Fe–Mg interdiffusion in aluminosilicate garnet.
Trans Am Geophys Union 62:411
Enami M, Nishiyama T, Mouri T (2007) Laser Raman microspectrometry of metamorphic quartz: A simple
method for comparison of metamorphic pressures. Am Mineral 92:1303–1315
Geochem 83:365–418
Engi M, Wersin P (1987) Derivation and application of a solution model for calcic garnet. Schweiz Mineral
Petrograph Mitteil 67:53–73
Errico JC, Barnes JD, Strickland A, Valley JW (2013) Oxygen isotope zoning in garnets from Franciscan eclogite
blocks: evidence for rock-buffered fluid interaction in the mantle wedge. Contrib Mineral Petrol 166:1161–1176
Essene EJ (1982) Geologic thermometry and barometry. Rev Mineral Geochem 10:153–206
Essene EJ (1989) The current status of thermobarometry in metamorphic rocks. In: Evolution of Metamorphic
Belts. Daly JS, Cliff, RA and Yardley BWD (eds) Geological Society of London, p 1–44
Evans BW (1966) Microprobe study of zoning in eclogite garnet. Geol Soc Am Spec Pap 87:54
Evans BW, Guidotti CV (1966) The sillimanite–potash feldspar isograd in Western Maine, USA. Contrib Mineral
Petrol 12:25–62
Faryad SW, Chakraborty S (2005) Duration of Eo-Alpine metamorphic events obtained from multicomponent
diffusion modeling of garnet: a case study from the Eastern Alps. Contrib Mineral Petrol 150:306–318
Fernando G, Hauzenberger CA, Baumgartner LP, Hofmeister W (2003) Modeling of retrograde diffusion zoning
in garnet: evidence for slow cooling of granulites from the Highland Complex of Sri Lanka. Mineral Petrol
78:53–71
Ferry JM, Spear FS (1978) Experimental calibration of the partitioning of Fe and Mg between biotite and garnet.
Ferry JM, Kitajima K, Strickland A, Valley JW (2014) Ion microprobe survey of the grain-scale oxygen isotope
geochemistry of minerals in metamorphic rocks. Geochim Cosmochim Acta 144:403–433
Florence FP, Spear FS (1991) Effects of diffusional modification of garnet growth zoning on P–T path
calculations. Contrib Mineral Petrol 107:487–500
Florence FP, Spear FS (1993) Influences of reaction history and chemical diffusion on P–T calculations for
staurolite schists from the Littleton Formation, northwestern New Hampshire. Am Mineral 78:345–359
Foster GL, Parrish RR, Horstwood MSA, Chenery S, Pyle JM, Gibson HD (2004) The generation of prograde
P–T–t points and paths; a textural, compositional, and chronological study of metamorphic monazite.
Frost MJ (1962) Metamorphic grade and iron–magnesium distribution between coexisting garnet–biotite and
garnet–hornblende. Mineral Mag 99:427–438
Gaidies F, Abart R, De Capitani C, Schuster R, Connolly JAD, Reusser E (2006) Characterization of
thermobarometry. J Metamorph Geol 24:451–475
Gaidies F, De Capitani C, Abart R (2008a) THERIA_G: a software program to numerically model prograde garnet
Gaidies F, De Capitani C, Abart R, Schuster R (2008b) Prograde garnet growth along complex P–T–t
paths: results from numerical experiments on polyphase garnet from the Wölz Complex (Austroalpine
basement). Contrib Mineral Petrol 155:673–688
Gaidies F, Pattison DRM, de Capitani C (2011) Toward a quantitative model of metamorphic nucleation and
Gaidies F, Petley-Ragan A, Chakraborty S, Dasgupta S, Jones P (2015) Constraining the conditions of
Barrovian metamorphism in Sikkim, India: P–T–t paths of garnet crystallization in the Lesser Himalayan
Belt. J Metamorph Geol 33:23–44
Galli A, Le Bayon B, Schmidt MW, Burg, J-P, Caddick MJ, Reusser E (2011) Granulites and charnockites of
the Gruf Complex: evidence for Permian ultra-high temperature metamorphism in the Central Alps. Lithos
124:17–45
Galli A, Le Bayon B, Schmidt MW, Burg JP, Reusser E, Sergeev SA, Larionov A (2012) U–Pb zircon dating of
the Gruf Complex: disclosing the late Variscan granulitic lower crust of Europe stranded in the Central Alps.
Galli A, Le Bayon B, Schmidt MW, Burg, J-P, Reusser E (2013) Tectonometamorphic history of the Gruf complex
(Central Alps): exhumation of a granulite–migmatite complex with the Bergell pluton. Swiss J Geosci
106:33–62
Galwey AK, Jones KA (1966) Crystal size frequency distribution of garnets in some analysed metamorphic rocks
from Mallaig Inverness Scotland. Geol Mag 103:143-and
Ganguly J (2002) Diffusion kinetics in minerals: Principles and applications to tectonometamorphic processes.
In: EMU Notes in Mineralogy: Energy Modelling in Minerals. Gramaccioli CMO (ed) Eur Mineral Union,
p 271–309
Geochem 72:559–601
Ganguly J, Saxena SK (1984) Mixing properties of aluminosilicate garnets: constraints from natural and
experimental data, and applications to geothermo-barometry. Am Mineral 69:88–97
Ganguly J, Tirone M (1999) Diffusion closure temperature and age of a mineral with arbitrary extent of diffusion:
Ganguly J, Tirone M (2001) Relationship between cooling rate and cooling age of a mineral: Theory and
applications to meteorites. Meteorit Planet Sci 36:167–175
Ganguly J, Cheng WJ, Tirone M (1996) Thermodynamics of aluminosilicate garnet solid solution: New
experimental data, an optimized model, and thermometric applications. Contrib Mineral Petrol 126:137–151
Ganguly J, Cheng W, Chakraborty S (1998a) Cation diffusion in aluminosilicate garnets: experimental determination
in pyrope-almandine diffusion couples. Contrib Mineral Petrol 131:171–180
Ganguly J, Tirone M, Hervig RL (1998b) Diffusion kinetics of samarium and neodymium in garnet, and a method
for determining cooling rates of rocks. Science 281:805–807
Ganguly J, Dasgupta S, Cheng W, Neogi S (2000) Exhumation history of a section of the Sikkim Himalayas,
India: records in the metamorphic mineral equilibria and compositional zoning of garnet. Earth Planet Sci
Lett 183:471–486
Vermont, USA. Chem Geol 401:151–168
and oxygen isotope analysis. Contrib Mineral Petrol 171:1–20
Gessmann CK, Spiering B, Raith M (1997) Experimental study of the Fe–Mg exchange between garnet and biotite:
Constraints on the mixing behavior and analysis of the cation-exchange mechanisms. Am Mineral 82:1225–1240
Getty SR, Selverstone J, Wernicke BP, Jacobsen SB, Aliberti E, Lux DR (1993) Sm–Nd dating of multiple garnet
growth events in an arc-continent collision zone, northwestern United-States Cordillera. Contrib Mineral
Petrol 115:45–57
Ghent ED (1976) Plagioclase–garnet–Al2SiO5–quartz: a potential geothermometer–geobarometer. Am Mineral
61:710–714
Ghent ED (1977) Applications of activity-composition relations to displacement of a solid-solid equilibrium
anorthite = grossular + kyanite + quartz. In: Short Course in Application of Thermodynamics to Petrology
and Ore Deposits. Greenwood HJ (ed) Mineral Assoc Canada, p 99–108
Ghent ED, Stout MZ (1981) Geobarometry and geothermometry of plagioclase–biotite–garnet–muscovite
assemblages. Contrib Mineral Petrol 76:92–97
Gieré R, Rumble D, Gunther D, Connolly J, Caddick MJ (2011) Correlation of growth and breakdown of major
and accessory minerals in metapelites from Campolungo, Central Alps. J Petrol 52:2293–2334
Goldman DS, Albee AL (1977) Correlation of Mg/Fe partitioning between garnet and biotite with O18/O16
partitioning between quartz and magnetite. Am J Sci 277:750–761
Goldsmith JR (1980) Melting and breakdown reactions of anorthite at high pressures and temperatures. Am
Mineral 65:272–284
Gordon SM, Luffi P, Hacker B, Valley J, Spicuzza M, Kozdon R, Kelemen P, Ratshbacher L, Minaev V (2012) The
thermal structure of continental crust in active orogens: insight from Miocene eclogite and granulite xenoliths
of the Pamir Mountains. J Metamorph Geol 30:413–434
Grew ES, Locock AJ, Mills SJ, Galuskina IO, Galuskin EV, Halenius U (2013) Nomenclature of the garnet
supergroup. Am Mineral 98:785–810
Griffin WL, Brueckner HK (1980) Caledonian Sm–Nd ages and a crustal origin for Norwegian eclogites. Nature
285:319–321
Guevara VE, Caddick MJ (2016) Shooting at a moving target: phase equilibria modeling of high-temperature
metamorphism. J Metamorph Geol 34:209–235
Guiraud M, Powell R (2006) P–V–T relationships and mineral equilibria in inclusions in minerals. Earth Planet
Sci Lett 244:683–694
Guiraud M, Powell R, Rebay G (2001) H2O in metamorphism and unexpected behaviour in the preservation of
metamorphic mineral assemblages. J Metamorph Geol 19:445–454
Hallett BW, Spear FS (2011) Insight into the cooling history of the Valhalla complex, British Columbia. Lithos
125:809–824
Hackler RT, Wood BJ (1989) Experimental determination of Fe and Mg exchange between garnet and olivine and
estimation of Fe–Mg mixing properties in garnet. Am Mineral 74:994–999
Hallett BW, Spear FS (2014a) The P–T history of anatectic pelites of the Northern East Humboldt Range,
Nevada: Evidence for tectonic loading, decompression, and anatexis. J Petrol 55:3–36
Hallett BW, Spear FS (2015) Monazite, zircon, and garnet growth in migmatitic pelites as a record of metamorphism
and partial melting in the East Humboldt Range, Nevada. Am Mineral 100:951–972
Harte B, Henley KJ (1966) Occurence of compositionally zoned almanditic garnets in regionally metamorphosed
rocks. Nature 210:689–692
small aliquots (1–10 ng). Chem Geol 258:251–257
Hemley RJ (1987) Pressure dependence of Raman spectra of SiO2 polymorphs; α-quartz, coesite and stishovite. In:
High-Pressure Research in Mineral Physics. Murli, HM and Syono Y (eds) Am Geophys Union, Washington
DC, p 347–359
Hensen BJ (1971) Theoretical phase relations involving cordierite and garnet in the system MgO–FeO–Al2O3–
SiO2. Contrib Mineral Petrol 33:191–214
Herwartz D, Nagel TJ, Munker C, Scherer EE, Froitzheim N (2011) Tracing two orogenic cycles in one eclogite
sample by Lu–Hf garnet chronometry. Nat Geosci 4:178–183
Hickmott DD, Shimizu N (1990) Trace-element zoning in garnet from the Kwoiek area, British-Columbia—
disequilibrium partitioning during garnet growth. Contrib Mineral Petrol 104:619–630
Hickmott D, Spear FS (1992) Major-element and trace-element zoning in garnets from calcareous pelites in the
NW Shelburne Falls quadrangle, Massachusetts—garnet growth histories in retrograded rocks. J Petrol
33:965–1005
Hickmott DD, Shimizu N, Spear FS, Selverstone J (1987) Trace-element zoning in a metamorphic garnet. Geology
15:573–576
Hodges KV, Crowley PD (1985) Error estimation and empirical geothermobarometry for pelitic systems. Am
Mineral 70:702–709
Hodges KV, McKenna LW (1987) Realistic propagation of uncertainties in geologic thermobarometry. Am
Mineral 72:671–680
Hoisch TD (1990) Empirical calibration of six geobarometers for the mineral assemblage quartz + muscovite +
biotite + plagioclase + garnet. Contrib Mineral Petrol 104:225–234
Hoisch TD (1991) Equilibria within the mineral assemblage quartz + muscovite + biotite + garnet + plagioclase,
and implications for the mixing properties of octahedrally-coordinated cations in muscovite and biotite.
Holdaway MJ (2000) Application of new experimental and garnet Margules data to the garnet–biotite
geothermometer. Am Mineral 85:881–892
Holdaway MJ (2001) Recalibration of the GASP geobarometer in light of recent garnet and plagioclase activity
models and versions of the garnet–biotite geothermometer. Am Mineral 86:1117–1129
Holdaway MJ, Mukhopadhyay B, Dyar MD, Guidotti CV, Dutrow BL (1997) Garnet–biotite geothermobarometry
revised: New Margules parameters and a natural specimen data set from Maine. Am Mineral 82:582–595
Holland TJB, Powell R (1985) An internally consistent thermodynamic dataset with uncertainties and correlations:
2. Data and results. J Metamorph Geol 3:343–370
Holland TJB, Powell R (1990) An enlarged and updated internally consistent thermodynamic dataset with
uncertainties and correlations—the system K2O–Na2O–CaO–MgO–MnO–FeO–Fe2O3–Al2O3–TiO2–
SiO2–C–H2–O2. J Metamorph Geol 8:89–124
Hollister LS (1966) Garnet zoning: an interpretation based on the Rayleigh fractionation model. Science 154:1647–1651
Ikeda T, Shimobayashi N, Wallis SR, Tsuchiyama A (2002) Crystallographic orientation, chemical composition
and three-dimensional geometry of sigmoidal garnet: evidence for rotation. J Struct Geol 24:1633–1646
Indares AD, Martignole J (1985) Biotite–garnet geothermometry in the granulite facies: the influence of Ti and Al
in biotite. Am Mineral 70:272–278
Izraeli ES, Harris JW, Navon O (1999) Raman barometry of diamond formation. Earth Planet Sci Lett 173:351–360
Jamtveit B, Hervig RL (1994) Constraints on transport and kinetics in hydrothermal systems from zoned garnet
crystals. Science 263:505–508
Jamtveit B, Wogelius RA, Fraser DG (1993) Zonation patterns of skarn garnets—records of hydrothermal system
evolution. Geology 21:113–116
Jedlicka R, Faryad SW, Hauzenberger C (2015) Prograde Metamorphic History of UHP Granulites from the
Moldanubian Zone (Bohemian Massif) Revealed by Major Element and Y + REE Zoning in Garnets. J Petrol
56:2069–2088
John T, Gussone N, Podladchikov YY, Bebout GE, Dohmen R, Halama R, Klemd R, Magna T, Seitz HM (2012)
Volcanic arcs fed by rapid pulsed fluid flow through subducting slabs. Nat Geosci 5:489–492
Johnson SE (1993) Testing models for the development of spiral-shaped inclusion trails in garnet porphyroblasts—
to rotate or not to rotate, that is the question. J Metamorph Geol 11:635–659
Kawakami T, Hokada T (2010) Linking P–T path with development of discontinuous phosphorus zoning
in garnet during high-temperature metamorphism—an example from Lützow-Holm Complex, East
Antarctica. J Mineral Petrol Sci 105:175–186
Kelly ED, Carlson WD, Connelly JN (2011) Implications of garnet resorption for the Lu–Hf garnet geochronometer:
an example from the contact aureole of the Makhavinekh Lake Pluton, Labrador. J Metamorph Geol 29:901–916
Kelly ED, Carlson WD, Ketcham RA (2013a) Crystallization kinetics during regional metamorphism of
porphyroblastic rocks. J Metamorph Geol 31:963–979
Kelly ED, Carlson WD, Ketcham RA (2013b) Magnitudes of departures from equilibrium during regional
metamorphism of porphyroblastic rocks. J Metamorph Geol 31:981–1002
Kelsey DE, Hand M (2015) On ultrahigh temperature crustal metamorphism: phase equilibria, trace element
thermometry, bulk composition, heat sources, timescales and tectonic settings. Geoscience Frontiers 6:311–356
Ketcham RA, Carlson WD (2012) Numerical simulation of diffusion-controlled nucleation and growth of
porphyroblasts. J Metamorph Geol 30:489–512
Kita NT, Ushikubo T, Fu B, Valley JW (2009) High precision SIMS oxygen isotope analysis and the effect of
sample topography. Chem Geol 264:43–57
Kleeman U, Reinhardt J (1994) Garnet–biotite thermometry revised: the effect of AlVI and Ti in biotite. Eur J
Mineral 6:925–941
Kohn MJ (1993) Modeling of prograde mineral d18O changes in metamorphic systems. Contrib Mineral Petrol
113:249–261
Kohn MJ (2004) Oscillatory- and sector-zoned garnets record cyclic (?) rapid thrusting in central Nepal. Geochem
Geophys Geosystem 5:9
Kohn MJ (2014) “Thermoba-Raman-try”: Calibration of spectroscopic barometers and thermometers for mineral
inclusions. Earth Planet Sci Lett 388:187–196
Geochem 83:103–152
Kohn MJ, Spear FS (1991) Error propagation in barometers: 2. Application to rocks. Am Mineral 76:138–147
Kohn MJ, Spear FS (2000) Retrograde net transfer reaction insurance for pressure–temperature estimates. Geology
28:1127–1130
Kohn MJ, Valley JW (1994) Oxygen-isotope constraints on metamorphic fluid-flow, Townshend Dam, Vermont,
USA. Geochim Cosmochim Acta 58:5551–5566
Kohn MJ, Valley JW (1998) Obtaining equilibrium oxygen isotope fractionations from rocks: theory and examples.
Kohn MJ, Valley JW, Elsenheimer D, Spicuzza MJ (1993) O isotope zoning in garnet and staurolite—evidence for
closed-system mineral growth during regional metamorphism. Am Mineral 78:988–1001
Kohn MJ, Spear FS, Valley JW (1997) Dehydration-melting and fluid recycling during metamorphism: Rangeley
Formation, New Hampshire, USA. J Petrol 38:1255–1277
Konrad-Schmolke M, Babist J, Handy MR, O’Brien PJ (2006) The physico-chemical properties of a subducted
slab from garnet zonation patterns (Sesia Zone, Western Alps). J Petrol 47:2123–2148
Konrad-Schmolke M, O’Brien PJ, Heidelbach F (2007) Compositional re-equilibration of garnet: the importance
of sub-grain boundaries. Eur J Mineral 19:431–438
Konrad-Schmolke M, O’Brien PJ, De Capitani C, Carswell DA (2008) Garnet growth at high- and ultra-high
pressure conditions and the effect of element fractionation on mineral modes and composition. Lithos
103:309–332
Korsakov AV, Zhukov VP, Vandenabeele P (2010) Raman-based geobarometry of ultrahigh-pressure metamorphic
rocks: applications, problems, and perspectives. Anal Bioanal Chem 397:2739–2752
Korhonen FJ, Brown M, Grove M, Siddoway CS, Baxter EF, Inglis JD (2012) Separating metamorphic events in
the Fosdick migmatite–granite complex, West Antarctica. J Metamorph Geol 30:165–191
Kouketsu Y, Nishiyama T, Ikeda T, Enami M (2014) Evaluation of residual pressure in an inclusion-host system
using negative frequency shift of quartz Raman spectra. Am Mineral 99:433–442
Koziol AM, Newton RC (1988) Redetermination of the anorthite breakdown reaction and improvement of the
plagioclase– garnet–Al2SiO5–quartz geobarometer. Am Mineral 73:216–223
Koziol AM, Newton RC (1989) Grossular activity-composition relationships in ternary garnets determined by
reversed displaced-equilibrium experiments. Contrib Mineral Petrol 103:423–433
Kretz R (1959) Chemical study of garnet, biotite and hornblende from gneisses of southwestern Quebec, with
emphasis on distribution of elements in coexisting minerals. J Geol 67:371–403
Kretz R (1964) Analysis of equilibrium in garnet–biotite–sillimanite gneisses from Quebec. J Petrol 5:1–20
Kretz R (1966) Interpretation of the shape of mineral grains in metamorphic rocks. J Petrol 7:68–94
Kretz R (1973) Kinetics of the crystallization of garnet at two localities near Yellowknife. Can Mineral 12:1–20
Kretz R (1974) Some models for the rate of crystallization of garnet in metamorphic rocks. Lithos 7:123–131
Kylander-Clark ARC, Hacker BR, Johnson CM, Beard BL, Mahlen NJ, Lapen TJ (2007) Coupled Lu–Hf and Sm–
Nd geochronology constrains prograde and exhumation histories of high- and ultrahigh-pressure eclogites
from western Norway. Chem Geol 242:137–154
Lagos M, Scherer EE, Tomaschek F, Munker C, Keiter M, Berndt J, Ballhaus C (2007) High precision Lu–Hf
geochronology of eocene eclogite-facies rocks from syros, cyclades, Greece. Chem Geol 243:16–35
Lanari P, Engi M (2017) Local bulk composition effects on mineral assemblages. Rev Mineral Geochem v. 83
Lanari P, Giuntoli F, Loury C, Burn M, Engi M (2017) An inverse modeling approach to obtain P–T conditions
of metamorphic stages involving garnet growth and resorption. Eur J Mineral in press:doi:10.1127/
ejm/2017/0029–2597
Lancaster PJ, Fu B, Page FZ, Kita NT, Bickford ME, Hill BM, McLelland JM, Valley JW (2009) Genesis of
metapelitic migmatites in the Adirondack Mountains, New York. J Metamorph Geol 27:41–54
Lanzirotti A (1995) Yttrium zoning in metamorphic garnets. Geochim Cosmochim Acta 59:4105–4110
Lasaga AC (1983) Geospeedometry: An extension of geothermometry. In: Kinetics and Equilibrium in Mineral
Reactions. Saxena SK (ed) Springer-Verlag, New York, p 81–114
Loomis TP (1975) Reaction zoning of garnet. Contrib Mineral Petrol 52:285–305
Loomis TP, Ganguly J, Elphick SC (1985) Experimental determination of cation diffusivities in aluminosilicate
garnets II. Multicomponent simulation and tracer diffusion coefficients. Contrib Mineral Petrol 90:45–51
Mahar E, Powell R, Holland TJB, Howell N (1997) The effect of Mn on mineral stability in metapelites. J
Malaspina N, Poli S, Fumagalli P (2009) The oxidation state of metasomatized mantle wedge: insights from
C–O–H-bearing garnet peridotite. J Petrol 50:1533–1552
Manzotti P, Ballevre M (2013) Multistage garnet in high-pressure metasediments: Alpine overgrowths on Variscan
detrital grains. Geology 41:1151–1154
Marmo BA, Clarke GL, Powell R (2002) Fractionation of bulk rock composition due to porphyroblast growth:
effects on eclogite facies mineral equilibria, Pam Peninsula, New Caledonia. J Metamorph Geol 20:151–165
Martin AJ (2009) Sub-millimeter heterogeneity of yttrium and chromium during growth of semi-pelitic garnet. J
Petrol 50:1713–1727
Martin L, Duchene S, Deloule E, Vanderhaeghe O (2006) The isotopic composition of zircon and garnet: A record
of the metamorphic history of Naxos, Greece. Lithos 87:174–192
Martin LAJ, Ballevre M, Boulvais P, Halfpenny A, Vanderhaeghe O, Duchene S, Deloule E (2011) Garnet re-
equilibration by coupled dissolution-reprecipitation: evidence from textural, major element and oxygen
isotope zoning of ‘cloudy’ garnet. J Metamorph Geol 29:213–231
Martin LAJ, Rubatto D, Crépisson C, Hermann J, Putlitz B, Vitale-Brovarone A (2014) Garnet oxygen analysis by
SHRIMP-SI: Matrix corrections and application to high-pressure metasomatic rocks from Alpine Corsica.
Chem Geol 374–375:25–36
Masago H, Rumble D, Ernst WG, Parkinson CD, Maruyama S (2003) Low d18O eclogites from the Kokchetav
massif, northern Kazakhstan. J Metamorph Geol 21:579–587
McKenna LW, Hodges KV (1988) Accuracy versus precision in locating reaction boundaries: Implications for the
garnet–plagioclase–aluminum silicate–quartz geobarometer. Am Mineral 73:1205–1205
Menard T, Spear FS (1993) Metamorphism of calcic pelitic schists, Strafford Dome, Vermont: Compositional
zoning and reaction history. J Petrol 34:977–1005
Menard T, Spear FS (1996) Interpretation of plagioclase zonation in calcic pelitic schist, south Strafford, Vermont,
and the effects on thermobarometry. Can Mineral 34:133–146
Meth CE, Carlson WD (2005) Diffusion-controlled synkinematic growth of garnet from a heterogeneous precursor
at Passo del Sole, Switzerland. Can Mineral 43:157–182
Mezger K, Hanson GN, Bohlen SR (1989) U–Pb systematics of garnet—dating the growth of garnet in the late
Archean Pikwitonei Granulite Domain at Cauchon and Natawahunan lakes, Manitoba, Canada. Contrib
Mezger K, Essene EJ, Halliday AN (1992) Closure temperatures of the Sm–Nd system in metamorphic garnets.
Moecher DP, Essene EJ, Anovitz LM (1988) Calculation and application of clinopyroxene–garnet–plagioclase–
quartz geobarometers. Contrib Mineral Petrol 100:92–106
Moore SJ, Carlson WD, Hesse, Ma (2013) Origins of yttrium and rare earth element distributions in metamorphic
garnet. J Metamorph Geol 31:663–689
Moscati RJ, Johnson CA (2014) Major element and oxygen isotope geochemistry of vapour-phase garnet from the
Topopah Spring Tuff at Yucca Mountain, Nevada, USA. Mineral Mag 78:1029–1041
Mottram CM, Parrish RR, Regis D, Warren CJ, Argles TW, Harris NB, Roberts NM (2015) Using U–Th–Pb
Himalaya. Tectonics 34:1355–1374
with application to metapelitic schist from the Kootenay Arc, British Columbia, Canada. J Metamorph
Geol 31:525–548
Mukhopadhyay B, Holdaway MJ, Koziol AM (1997) A statistical model of thermodynamic mixing properties of
Ca–Mg–Fe2+ garnets. Am Mineral 82:165–181
Nesheim TO, Vervoort JD, McClelland WC, Gilotti JA, Lang HM (2012) Mesoproterozoic syntectonic garnet
within Belt Supergroup metamorphic tectonites: Evidence of Grenville-age metamorphism and deformation
along northwest Laurentia. Lithos 134:91–107
Newton RC, Haselton HT (1981) Thermodynamics of the garnet–plagioclase–Al2SiO5–quartz geobarometer. In:
Thermodynamics of Minerals and Melts. Newton RC, Navrotsky, A and Wood BJ (eds) Springer-Verlag,
New York, p 129–145
O’Brien PJ, Vrána S (1995) Eclogites with a short-lived granulite facies overprint in the Moldanubian Zone, Czech
Republic: Petrology, geochemistry and diffusion modelling of garnet zoning. Geol Rundsch 84:473–488
Otamendi JE, de la Rosa JD, Douce AEP, Castro A (2002) Rayleigh fractionation of heavy rare earths and yttrium
during metamorphic garnet growth. Geology 30:159–162
Page FZ, Kita NT, Valley JW (2010) Ion microprobe analysis of oxygen isotopes in garnets of complex chemistry.
Page FZ, Essene EJ, Mukasa SB, Valley JW (2014) A garnet–zircon oxygen isotope record of subduction and
exhumation fluids from the Franciscan Complex, California. J Petrol 55:103–131
Palin RM, Weller OM, Waters DJ, Dyck B (2016) Quantifying geological uncertainty in metamorphic phase
equilibria modelling; a Monte Carlo assessment and implications for tectonic interpretations. Geosci
Frontiers 7:591–607
Parkinson C, Katayama I (1999) Present-day ultrahigh-pressure conditions of coesite inclusions in zircon and
garnet: Evidence from laser Raman microspectroscopy. Geology 27:979–982
Pattison D (2015) Challenges in phase diagram modelling, focusing on metapelites. Geological Society of America
Pattison DRM, Chacko T, Farquhar J, McFarlane CRM (2003) Temperatures of granulite-facies metamorphism:
constraints from experimental phase equilibria and thermobarometry corrected for retrograde exchange. J
Petrol 44:867–900
Pattison DRM, De Capitani C, Gaidies F (2011) Petrological consequences of variations in metamorphic reaction
affinity. J Metamorph Geol 29:953–977
Peck W, Valley J (2004) Quartz–garnet isotope thermometry in the southern Adirondack Highlands (Grenville
Province, New York). J Metamorph Geol 22:763–773
Penniston-Dorland SC, Sorensen SS, Ash RD, Khadke SV (2010) Lithium isotopes as a tracer of fluids in a
subduction zone melange: Franciscan Complex, CA. Earth Planet Sci Lett 292:181–190
Perchuk LL, Lavrent’eva IV (1983) Experimental investigation of exchange equilibria in the system cordierite–
garnet–biotite. In: Kinetics and Equilibrium in Mineral Reactions. Saxena SK (ed) Springer-Verlag, New
York, p 199–239
Perchuk A, Philippot P, Erdmer P, Fialin M (1999) Rates of thermal equilibration at the onset of subduction
deduced from diffusion modeling of eclogitic garnets, Yukon–Tanana terrane, Canada. Geology 27:531–534
Peterman EM, Hacker BR, Baxter EF (2009) Phase transformations of continental crust during subduction and
exhumation: Western Gneiss Region, Norway. Eur J Mineral 21:1097–1118
Pogge von Strandmann PAE, Dohmen R, Marschall HR, Schumacher JC, Elliott T (2015) Extreme magnesium
isotope fractionation at outcrop scale records the mechanism and rate at which reaction fronts advance. J
Petrol 56:33–58
tectonometamorphic processes. Earth Planet Sci Lett 293:63–71
Pollington AD, Baxter EF (2011) High precision microsampling and preparation of zoned garnet porphyroblasts
for Sm–Nd geochronology. Chem Geol 281:270–282
Powell R, Holland TJB (1988) An internally consistent dataset with uncertainties and correlations. 3. Applications
to geobarometry, worked examples and a computer-program. J Metamorph Geol 6:173–204
Powell R, Holland TJB, Worley B (1998) Calculating phase diagrams involving solid solutions via non-linear
equations, with examples using THERMOCALC. J Metamorph Geol 16:577–588
Prince C, Harris N, Vance D (2001) Fluid-enhanced melting during prograde metamorphism. J Geol Soc London
158:233–241
Putlitz B, Matthews A, Valley JW (2000) Oxygen and hydrogen isotope study of high-pressure metagabbros and
metabasalts (Cyclades, Greece): implications for the subduction of oceanic crust. Contrib Mineral Petrol
138:114–126
Pyle J, Spear F (1999) Yttrium zoning in garnet: Coupling of major and accessory phases during metamorphic
and a new monazite- garnet thermometer. J Petrol 42:2083–2107
Raimondo T, Clark C, Hand M, Cliff J, Harris C (2012) High-resolution geochemical record of fluid–rock
interaction in a mid-crustal shear zone: a comparative study of major element and oxygen isotope transport
in garnet. J Metamorph Geol 30:255–280
Ramberg H (1952) Chemical bonds and distribution of cations in silicates. J Geol 60:331–355
Ramsay JG (1962) The geometry and mechanics of formation of “similar” type folds. J Geol 70:309–327
Richter R, Spiering B, Hoernes S (1988) Petrology and isotope-geochemistry of granulite-facial marbles and
calcium-silicate rocks of Sri-Lanka. Fortschr Mineral 66:134–134
Robyr M, Carlson WD, Passchier C, Vonlanthen P (2009) Microstructural, chemical and textural records during
growth of snowball garnet. J Metamorph Geol 27:423–437
Robyr M, Darbellay B, Baumgartner LP (2014) Matrix-dependent garnet growth in polymetamorphic rocks of the
Sesia zone, Italian Alps. J Metamorph Geol 32:3–24
Romer RL, Xiao YL (2005) Initial Pb–Sr(Nd) isotopic heterogeneity in a single allanite–epidote crystal:
implications of reaction history for the dating of minerals with low parent-to-daughter ratios. Contrib Mineral
Petrol 148:662–674
Rosenfeld JL (1968) Garnet rotations due to the major Paleozoic deformations in southeast Vermont. In: Studies of
Appalachian Geology: Northern and Maritime. Zen E (ed) John Wiley and Sons, Inc., New York, p 185–202
Rosenfeld JL (1969) Stress effects around quartz inclusions in almandine and the piezothermometry of coexisting
aluminum silicates. Am J Sci 267:317–351
Rosenfeld JL, Chase AB (1961) Pressure and temperature of crystallization from elastic effects around solid
inclusions in minerals? Am J Sci 259:519–541
Rubatto D (2017) Zircon: the metamorphic mineral. Rev Mineral Geochem 83
Rubatto D, Angiboust S (2015) Oxygen isotope record of oceanic and high-pressure metasomatism: a P–T–
time–fluid path for the Monviso eclogites (Italy). Contrib Mineral Petrol 170
Rumble D, Yui TF (1998) The Qinglongshan oxygen and hydrogen isotope anomaly near Donghai in Jiangsu
Province, China. Geochim Cosmochim Acta 62:3307–3321
Russell AK, Kitajima K, Strickland A, Medaris LG, Schulze DJ, Valley JW (2013) Eclogite-facies fluid infiltration:
constraints from d18O zoning in garnet. Contrib Mineral Petrol 165:103–116
Sato K, Santosh M, Tsunogae T (2009) A petrologic and laser Raman spectroscopic study of sapphirine–spinel–
quartz–Mg-staurolite inclusions in garnet from Kumiloothu, southern India: Implications for extreme
metamorphism in a collisional orogen. J Geodynam 47:107–118
Saxena SK (1968) Distribution of elements between coexisting minerals and the nature of solid solution in garnet.
Saxena SK (1969) Silicate solid solutions and geothermometry: 3. Distribution of Fe and Mg between coexisting
garnet and biotite. Contrib Mineral Petrol 22:259–267
Sayab M, Shah SZ, Aerden D (2016) Metamorphic record of the NW Himalayan orogeny between the Indian
plate-Kohistan Ladakh Arc and Asia: Revelations from foliation intersection axis (FIA) controlled P–T–
t–d paths. Tectonophysics 671:110–126
Scherer EE, Cameron KL, Blichert-Toft J (2000) Lu–Hf garnet geochronology: Closure temperature relative to the
Sm–Nd system and the effects of trace mineral inclusions. Geochim Cosmochim Acta 64:3413–3432
Schmidt C, Ziemann MA (2000) In-situ Raman spectroscopy of quartz: A pressure sensor for hydrothermal
diamond-anvil cell experiments at elevated temperatures. Am Mineral 85:1725–1734
Schmidt C, Steele-MacInnis M, Watenphul A, Wilke M (2013) Calibration of zircon as a Raman spectroscopic
pressure sensor to high temperatures and application to water-silicate melt systems. Am Mineral 98:643–650
Schmidt A, Pourteau A, Candan O, Oberhansli R (2015) Lu–Hf geochronology on cm-sized garnets using
microsampling: New constraints on garnet growth rates and duration of metamorphism during continental
collision (Menderes Massif, Turkey). Earth Planet Sci Lett 432:24–35
Schwarz JO, Engi M, Berger A (2011) Porphyroblast crystallization kinetics: the role of the nutrient production
rate. J Metamorph Geol 29:497–512
Skelton A, Annersten H, Valley J (2002) d18O and yttrium zoning in garnet: time markers for fluid flow? J
Skora S, Baumgartner LP, Mahlen NJ, Johnson CM, Pilet S, Hellebrand E (2006) Diffusion-limited REE uptake
by eclogite garnets and its consequences for Lu–Hf and Sm–Nd geochronology. Contrib Mineral Petrol
152:703–720
Skora S, Baumgartner LP, Mahlen NJ, Lapen TJ, Johnson CM, Bussy F (2008) Estimation of a maximum Lu
diffusion rate in a natural eclogite garnet. Swiss J Geosci 101:637–650
Skora S, Lapen TJ, Baumgartner LP, Johnson CM, Hellebrand E, Mahlen NJ (2009) The duration of prograde
garnet crystallization in the UHP eclogites at Lago di Cignana, Italy. Earth Planet Sci Lett 287:402–411
Smit MA, Scherer EE, Mezger K (2013) Lu–Hf and Sm–Nd garnet geochronology: Chronometric closure and
implications for dating petrological processes. Earth Planet Sci Lett 381:222–233
Sobolev NV, Fursenko BA, Goryainov SV, Shu J, Hemley RJ, Mao A, Boyd FR (2000) Fossilized high pressure
from the Earth’s deep interior: the coesite-in-diamond barometer. PNAS 97:11875–11879
Sobolev NV, Schertl HP, Valley JW, Page FZ, Kita NT, Spicuzza MJ, Neuser RD, Logvinova AM (2011) Oxygen
isotope variations of garnets and clinopyroxenes in a layered diamondiferous calcsilicate rock from
Kokchetav Massif, Kazakhstan: a window into the geochemical nature of deeply subducted UHPM rocks.
Solva H, Thoni M, Habler G (2003) Dating a single garnet crystal with very high Sm/Nd ratios (Campo basement
unit, Eastern Alps). Eur J Mineral 15:35–42
Sorby HC, Butler PJ (1869) On the structure of rubies, sapphires, diamonds, and some other minerals. Proc R Soc
London 17:291–302
Sousa J, Kohn MJ, Schmitz MD, Northrup CJ, Spear FS (2013) Strontium isotope zoning in garnet: implications
for metamorphic matrix equilibration, geochronology and phase equilibrium modelling. J Metamorph Geol
31:437–452
Spear FS (1993) Metamorphic phase equilibria and pressure–temperature–time paths. Mineral Soc Am,
Washington, DC
Spear FS (2014) The duration of near-peak metamorphism from diffusion modelling of garnet zoning. J Metamorph
Geol 32:903–914
Spear FS, Daniel CG (2001) Diffusion control of garnet growth, Harpswell Neck, Maine, USA. J Metamorph Geol
19:179–195
Spear FS, Florence FP (1992) Thermobarometry in granulites: pitfalls and new approaches. Precambr Res 55:209–241
Spear FS, Selverstone J (1983) Quantitative P–T paths from zoned minerals: theory and tectonic applications.
Spear FS, Selverstone J, Hickmott D, Crowley PD, Hodges KV (1984) P–T paths from garnet zoning—a new
technique for deciphering tectonic processes in crystalline terranes. Geology 12:87–90
Spear FS, Kohn MJ, Florence FP, Menard T (1991) A model for garnet and plagioclase growth in pelitic
schists: Implications for thermobarometry and P–T path determinations. J Metamorph Geol 8:683–696
thermodynamic modeling. Contrib Mineral Petrol 168
Spear FS, Pattison DRM, Cheney JT (2015). It’s a mad, mad, mmad world. GSA Annual Meeting. Baltimore
Pap 523:31–73
Spiess R, Peruzzo L, Prior DJ, Wheeler J (2001) Development of garnet porphyroblasts by multiple nucleation,
coalescence and boundary misorientation-driven rotations. J Metamorph Geol 19:269–290
Spry A (1963) The origin and significance of snowball structure in garnet. J Petrol 4:211–222
St-Onge MR (1987) Zoned poikiloblasic garnets: P–T paths and syn-metamorphic uplift through 30 km of
structural depth, Wopmay Orogen, Canada. J Petrol 28:1–21
Stallard A, Ikei H, Masuda T (2002) Numerical simulations of spiral-shaped inclusion trails: can 3D geometry
distinguish between end-member models of spiral formation? J Metamorph Geol 20:801–812
Štípská P, Powell R, White RW, Baldwin JA (2010) Using calculated chemical potential relationships to account for
coronas around kyanite: an example from the Bohemian Massif. J Metamorph Geol 28:97–116
Stixrude L, Lithgow-Bertelloni C (2005) Thermodynamics of mantle minerals—I. Physical properties. Geophys
J Inter 162:610–632
Storm LC, Spear FS (2005) Pressure, temperature and cooling rates of granulite facies migmatitic pelites from the
southern Adirondack Highlands, New York. J Metamorph Geol 23:107–130
Stowell HH, Menard T, Ridgway CK (1996) Ca-metasomatism and chemical zonation of garnet in contact-
metamorphic aureoles, Juneau Gold Belt, southeastern Alaska. Can Mineral 34:1195–1209
Stowell HH, Taylor DL, Tinkham DL, Goldberg SA, Ouderkirk KA (2001) Contact metamorphic P–T–t paths
from Sm–Nd garnet ages, phase equilibria modelling and thermobarometry: Garnet Ledge, south-eastern
Alaska, USA. J Metamorph Geol 19:645–660
Stowell H, Tulloch A, Zuluaga C, Koenig A (2010) Timing and duration of garnet granulite metamorphism in
magmatic arc crust, Fiordland, New Zealand. Chem Geol 273:91–110
Stowell H, Parker KO, Gatewood M, Tulloch A, Koenig A (2014) Temporal links between pluton emplacement,
garnet granulite metamorphism, partial melting and extensional collapse in the lower crust of a Cretaceous
magmatic arc, Fiordland, New Zealand. J Metamorph Geol 32:151–175
Strickland A, Russell AK, Quintero R, Spicuzza MJ, Valley JW (2011) Oxygen isotope ratios of quartz inclusions
in garnet and implications for mineral pair thermometry. Geol Soc Am Abstr with Programs 43:93
reaction textures. Contrib Mineral Petrol 129:43–52
Sutton JR (1921) Inclusions in diamond from South Africa. Mineral Mag 19:208–210
Symmes GH, Ferry JM (1991) Evidence from mineral assemblages for infiltration of pelitic schists by aqueous
fluids during metamorphism. Contrib Mineral Petrol 108:419–438
Tajčmanová L, Konopásek J, Connolly JAD (2007) Diffusion-controlled development of silica-undersaturated
domains in felsic granulites of the Bohemian Massif (Variscan belt of Central Europe). Contrib Mineral
Petrol 153:237–250
Taylor HP, Sheppard SMF (1986) Igneous rocks. 1. Processes of isotopic fractionation and isotope systematics.
Rev Miner 16:227–271
Thompson AB (1976a) Mineral reactions in pelitic rocks: I. Prediction of P–T–X (Fe–Mg) phase relations.
Am J Sci 276:401–424
Thompson AB (1976b) Mineral reactions in pelitic rocks: II. Calculation of some P–T–X (Fe–Mg) phase
relations. Am J Sci 276:425–454
Thompson AB, Tracy RJ, Lyttle PT, Thompson JB (1977) Prograde reaction histories deduced from compositional
zonation and mineral inclusions in garnet from the Gassetts schist, Vermont. Am J Sci 277:1152–1167
Thoni M (2002) Sm–Nd isotope systematics in garnet from different lithologies (Eastern Alps): age results, and an
evaluation of potential problems for garnet Sm–Nd chronometry. Chem Geol 185:255–281
Tinkham DK, Ghent ED (2005) Estimating P–T conditions of garnet growth with isochemical phase-diagram
sections and the problem of effective bulk-composition. Can Mineral 43:33–50
Tinkham DK, Zuluaga CA, Stowell HH (2001) Metapelite phase equilibria modeling in MnNCKFMASH: The
effect of variable Al2O3 and MgO / (MgO + FeO) on mineral stability. Geol Mater Res 3:1–42
Tirone M, Ganguly J, Dohmen R, Langenhorst F, Hervig R, Becker, H-W (2005) Rare earth diffusion kinetics in
garnet: Experimental studies and applications. Geochim Cosmochim Acta 69:2385–2398
Tracy RJ (1982) Compositional zoning and inclusions in metamorphic minerals. Rev Mineral 10:355–397
Tracy RJ, Robinson P, Thompson AB (1976) Garnet composition and zoning in the determination of temperature
and pressure of metamorphism, central Massachusetts. Am Mineral 61:762–775
Tracy RJ, Caddick MJ, Thompson AB (2012) Garnet growth zoning in metapelites: not so simple after all, and
revealing more than we once thought? GSA Annual Meeting
Valley JW (1986) Stable isotope geochemistry of metamorphic rocks. Rev Mineral 16:445–489
Valley JW (2001) Stable isotope thermometry at high temperatures. Rev Mineral Geochem 43:365–413
Valley JW, Bindeman IN, Peck WH (2003) Empirical calibration of oxygen isotope fractionation in zircon.
van Breemen O, Hawkesworth CJ (1980) Sm–Nd isotopic study of garnets and their metamorphic host rocks.
Transactions of the Royal Society of Edinburgh: Earth Sciences 71:97–102
Van der Molen I (1981) The shift of the α–β transition temperature of quartz associated with the thermal expansion
of granite at high pressure. Tectonophysics 73:323–342
van Haren JLM, Ague JJ, Rye DM (1996) Oxygen isotope record of fluid infiltration and mass transfer during
regional metamorphism of pelitic schist, Connecticut, USA. Geochim Cosmochim Acta 60:3487–3504
Van Orman JA, Grove TL, Shimizu N, Layne GD (2002) Rare earth element diffusion in a natural pyrope single
crystal at 2.8 GPa. Contrib Mineral Petrol 142:416–424
Vance D, Harris N (1999) Timing of prograde metamorphism in the Zanskar Himalaya. Geology 27:395–398
Vance D, Holland T (1993) A detailed isotopic and petrological study of a single garnet from the Gassetts Schist,
Vermont. Contrib Mineral Petrol 114:101–118
Vance D, Onions RK (1990) Isotopic chronometry of zoned garnets—growth-kinetics and metamorphic histories.
Vance D, Onions RK (1992) Prograde and retrograde thermal histories from the central Swiss Alps. Earth Planet
Sci Lett 114:113–129
Vannay JC, Grasemann B (1998) Inverted metamorphism in the High Himalaya of Himachal Pradesh (NW India):
Phase equilibria versus thermobarometry. Schweiz Mineral Petrograph Mitteil 78:107–132
Vannay JC, Sharp ZD, Grasemann B (1999) Himalayan inverted metamorphism constrained by oxygen isotope
thermometry. Contrib Mineral Petrol 137:90–101
Vielzeuf D, Saúl A (2011) Uphill diffusion, zero-flux planes and transient chemical solitary waves in garnet.
Vielzeuf D, Champenois M, Valley JW, Brunet F, Devidal JL (2005) SIMS analyses of oxygen isotopes: Matrix
effects in Fe–Mg–Ca garnets. Chem Geol 223:208–226
Viete DR, Hermann J, Lister GS, Stenhouse IR (2011) The nature and origin of the Barrovian metamorphism,
Scotland: diffusion length scales in garnet and inferred thermal time scales. J Geol Soc London 168:115–131
Vorhies SH, Ague JJ (2011) Pressure–temperature evolution and thermal regimes in the Barrovian zones, Scotland.
J Geol Soc London 168:1147–1166
Vrijmoed JC, Hacker BR (2014) Determining P–T paths from garnet zoning using a brute-force computational
method. Contrib Mineral Petrol 167:1–13
Wang X, Planavsky NJ, Reinhard CT, Zou H, Ague JJ, Wu Y, Gill BC, Schwarzenbach EM, Peucker-Ehrenbrink
B (2016) Chromium isotope fractionation during subduction-related metamorphism, black shale weathering,
and hydrothermal alteration. Chem Geol 423:19–33
Waters D, Martin H (1993) The garnet–clinopyroxene–phengite barometer. Terra Abst 5:410–411
Waters DJ, Lovegrove DP (2002) Assessing the extent of disequilibrium and overstepping of prograde metamorphic
reactions in metapelites from the Bushveld Complex aureole, South Africa. J Metamorph Geol 20:135–149
Watson EB, Cherniak DJ (2013) Simple equations for diffusion in response to heating. Chem Geol 335:93–104
Wendt I, Carl C (1991) The statistical distribution of the mean squared weighted deviation Chem Geol 86:275–285
White RW, Powell R, Holland TJB, Worley B (2000) The effect of TiO2 and Fe2O3 on metapelitic assemblages at
greenschist and amphibolite facies conditions: mineral equilibria calculations in the system K2O–FeO–MgO–
Al2O3–SiO2–H2O–TiO2–Fe2O3. J Metamorph Geol 18:497–511
White RW, Pomroy NE, Powell R (2005) An in situ metatexite–diatexite transition in upper amphibolite facies
rocks from Broken Hill, Australia. J Metamorph Geol 23:579–602
the textural evolution of metamorphic rocks. J Metamorph Geol 26:181–198
White RW, Powell R, Holland TJB, Johnson TE, Green ECR (2014a) New mineral activity–composition relations
for thermodynamic calculations in metapelitic systems. J Metamorph Geol 32:261–286
White RW, Powell R, Johnson TE (2014b) The effect of Mn on mineral stability in metapelites revisited: new a–X
relations for manganese-bearing minerals. J Metamorph Geol 32:809–828
Whitney DL, Seaton NCA (2010) Garnet polycrystals and the significance of clustered crystallization. Contrib
Whitney DL, Goergen ET, Ketcham RA, Kunze K (2008) Formation of garnet polycrystals during metamorphic
crystallization. J Metamorph Geol 26:365–383
Williams ML, Jercinovic MJ, Hetherington CJ (2007) Microprobe monazite geochronology: understanding
geologic processes by integrating composition and chronology. Ann Rev Earth Planet Sci 35:137
Williams HM, Nielsen SG, Renac C, Griffin WL, O’Reilly SY, McCammon CA, Pearson N, Viljoen F, Alt JC,
Halliday AN (2009) Fractionation of oxygen and iron isotopes by partial melting processes: Implications for
the interpretation of stable isotope signatures in mafic rocks. Earth Planet Sci Lett 283:156–166
Geochem 83:153–182
145:228–250
Wood BJ, Banno S (1973) Garnet–orthopyroxene and orthopyroxene–clinopyroxene relationships in simple and
complex systems. Contrib Mineral Petrol 42:109–124
Woodsworth GJ (1977) Homogenization of zoned garnets from pelitic schists. Can Mineral 15:230–242
Wu C-M, Zhao G (2006) Recalibration of the garnet–muscovite (GM) geothermometer and the garnet–muscovite–
plagioclase–quartz (GMPQ) geobarometer for metapelitic assemblages. J Petrol 47:2357–2368
Wu C-M, Zhang J, Ren L-D (2004) Empirical garnet–biotite–plagioclase–quartz (GBPQ) geobarometry in
medium- to high-grade metapelites. J Petrol 45:1907–1921
Wu C-M, Cheng B-H (2006) Valid garnet–biotite (GB) geothermometry and garnet–aluminum silicate–
plagioclase–quartz (GASP) geobarometry in metapelitic rocks. Lithos 89:1–23
Yakymchuk C, Brown M (2014) Consequences of open-system melting in tectonics. J Geol Soc London
Yakymchuk C, Brown M, Clark C, Korhonen FJ, Piccoli PM, Siddoway CS, Taylor RJM, Vervoort JD (2015)
Decoding polyphase migmatites using geochronology and phase equilibria modelling. J Metamorph Geol
33:203–230
Yakymchuk C, Clark C, White RW (2017) Phase relations, reaction sequences and petrochronology. Rev Mineral
Geochem 83:13–53
Yang PS, Pattison D (2006) Genesis of monazite and Y zoning in garnet from the Black Hills, South Dakota. Lithos
88:233–253
Yang P, Rivers T (2001) Chromium and manganese zoning in pelitic garnet and kyanite: Spiral, overprint, and
oscillatory (?) zoning patterns and the role of growth rate. J Metamorph Geol 19:455–474
Yang P, Rivers T (2002) The origin of Mn and Y annuli in garnet and the thermal dependence of P in garnet and Y
in apatite in calc- pelite and pelite. Gagnon terrane, western Labrador. Geol Mater Res 4:1–35
Yardley BWD (1977) An empirical study of diffusion in garnet. Am Mineral 62:793–800
Young ED, Rumble D (1993) The origin of correlated variations in insitu 18O/16O and elemental concentrations in
metamorphic garnet from southeastern Vermont, USA. Geochim Cosmochim Acta 57:2585–2597
Zack T, John T (2007) An evaluation of reactive fluid flow and trace element mobility in subducting slabs. Chem
Geol 239:199–216
Zack T, Tomascak PB, Rudnick RL, Dalpe C, McDonough WF (2003) Extremely light Li in orogenic eclogites:
The role of isotope fractionation during dehydration in subducted oceanic crust. Earth Planet Sci Lett
208:279–290
Zeng L, Asimow PD, Saleeby JB (2005) Coupling of anatectic reactions and dissolution of accessory phases and
the Sr and Nd isotope systematics of anatectic melts from a metasedimentary source. Geochim Cosmochim
Acta 69:3671–3682
Zhang Y (1998) Mechanical and phase equilibria in inclusion–host systems. Earth Planet Sci Lett 157:209–222
Zheng YF (1991) Calculation of oxygen isotope fractionation in metal-oxides. Geochim Cosmochim Acta
55:2299–2307
Zheng YF (1993) Calculation of oxygen-isotope fractionation in hydroxyl-bearing silicates. Earth Planet Sci Lett
120:247–263
Zheng Y-F, Fu B, Li Y, Xiao Y, Li S (1998) Oxygen and hydrogen isotope geochemistry of ultrahigh-pressure
eclogites from the Dabie Mountains and the Sulu terrane. Earth Planet Sci Lett 155:113–129
Zheng YF, Zhao ZF, Wu YB, Zhang SB, Liu XM, Wu FY (2006) Zircon U–Pb age, Hf and O isotope constraints
on protolith origin of ultrahigh-pressure eclogite and gneiss in the Dable orogen. Chem Geol 231:135–158
Zhou B, Hensen BJ (1995) Inherited Sm/Nd isotope components preserved in monazite inclusions within garnets in
leucogneiss from East Antarctica and implications for closure temperature studies. Chem Geol 121:317–326
Zhukov VP, Korsakov AV (2015) Evolution of host-inclusion systems: a visco-elastic model. J Metamorph Geol
33:815–828.
Vol. 83 pp. 535–575, 2017 16
Chronometry and Speedometry of Magmatic Processes

using Chemical Diffusion in
Olivine, Plagioclase and Pyroxenes
Ralf Dohmen, Kathrin Faak
Institut für Geologie, Mineralogie und Geophysik
Ruhr-Universität Bochum
Universitätsstraße 150
44801 Bochum
Germany
ralf.dohmen@rub.de
kathrin.faak@rub.de
Jon D. Blundy
School of Earth Sciences
University of Bristol
Wills Memorial Building
Bristol BS8 1RJ
United Kingdom
jon.blundy@bristol.ac.uk
INTRODUCTION
The magmatic processes that fuel volcanism, crustal growth, ore formation and discharge of
volcanic gases and aerosols to the atmosphere occur across a range of timescales, from millions
of years to just a few seconds. For example, the production of new oceanic crust at mid-ocean
ridges is a near-continuous process that can operate in any one ocean basin on timescales of more
than 100 m.y. However, the driving force for such processes is the spreading of the ocean plates
that happens on a cm/yr timescale. At the other end of the spectrum, explosive volcanic eruptions
involve the ascent and fragmentation of magma at velocities of the order of 100 m/s such that the
journey from a magma chamber to an ash cloud may take place in a matter of minutes. In this
case the driving force is the rapid expansion of magmatic gas in response to changes in pressure.
At intermediate timescales magmatic processes may give rise to hydrothermal ore deposits on
timescales of less than a million years for an individual deposit, while growth of giant granite
batholiths may require piecemeal assembly of magma batches on timescales of a few million
years. Although each of these processes has a characteristic, time-averaged timescale on which
it operates, this is typically the end result of one or more natural processes that operate on much
shorter timescales. For example, mid-ocean ridges do not extrude magma continuously onto the
ocean floor, mineralising fluids do not discharge continuously through the shallow crust, and
granitic magmas do not dribble continuously into evolving batholithic chambers. In some cases
it is the long-term timescales that are important, for example the spreading rate of ocean basins,
in others it is the short-term timescales that are important, for example the episodic growth of
lava domes at active volcanoes. Although the long-term timescales are reasonably well known,
accessing the shorter timescales is notoriously difficult, yet is vital if we are to successfully model
magmatic systems. The importance of gauging the appropriate timescale for a given magmatic

1943-2666/17/0083-0016$05.00 (online)
536 Dohmen, Faak & Blundy
process is that, coupled to an extensive parameter such as volume, mass or length, it enables
calculations of rates or fluxes of matter and heat. As the tempo of magmatism in the widest
sense is controlled ultimately by the interplay of several different processes each with their own
characteristic rate, e.g., magma input, convection, degassing, cooling etc, so the determination
of timescales is fundamental to the understanding of magmatism. In this chapter we review first
the available petrochronometers and then focus specifically on the burgeoning field of diffusion
chronometry as a means of accessing timescales for a variety of magmatic processes.
Types of chronometers
Magmatic timescales can be constrained primarily by three methods: observational,
radiometric and diffusive. Observations are limited to active magmatic systems, such
as volcanoes that are erupting or degassing passively. For example the growth of lava
domes at Mount St. Helens (Washington) and Soufrière Hills (Montserrat) volcanoes has
been documented with unprecedented detail over the course of more than two decades.
Visual observations of dome characteristics (volume, height, temperature, etc.) have been
supplemented by geodesy, tiltmeters, seismology and measurements of gas chemistry. The
latter can also be achieved remotely, using satellites, for volcanoes that degas syn- or inter-
eruptively, such as Etna (Italy). All of these observations provide precise timescales that can be
used to refine physical understanding of sub-volcanic processes. For example, at Soufrière Hills
observations of lava dome extrusions have been used to develop physical models of eruption
periodicity that embrace conduit flow and gas exsolution (Melnik and Sparks 2005). Likewise,
coupled observations of lava dome extrusion and inter-eruption gas discharge have been used
to invoke mushy, sub-volcanic magmatic systems that are subject to periodic instability that
leads to a decoupling of gas and magma (Christopher et al. 2015). Kilauea volcano (Hawaii) is
a classic example of a well-monitored volcano where frequent, long-lasting observations have
been used to formulate a detailed image of the magma plumbing system and dynamics (Tilling
and Dvorak 1993; Dvorak and Dzurisin 1997).
The drawback of quantitative observational timescales is that they are limited to the modern
era, notably since the advent of routine volcanic monitoring of the type that is now widespread
at many active and restless volcanoes. To extend the observational timescale further back in
time requires the availability of archival records, such as those found in eyewitness accounts.
For example, colonial records of volcanism in Ecuador have provided insights into eruptions
of Cotopaxi volcano (Pistolesi et al. 2011); Albert et al. (2016) used historical accounts of
seismicity to explore the build-up to eruptions at monogenetic volcanoes. Other temporal
indicators of volcanism, such as tree rings or ice-cores, can also be exploited, notably because
they contain robust chronologies. Linking historic eyewitness accounts to dendrochronology
and ice-core records is, perhaps, a relatively unexplored approach to volcanic timescales. Yet
still, this approach is limited, at the very most, to the last several hundred years.
For longer timescales the methodology of choice has been radiometric dating, whereby
the decay of naturally occurring radioisotopes to daughter isotopes has long been exploited
as a geochronometer. Originally this approach was limited to radioisotopes with multi-
million year half-lives, restricting applicability to long timescales. The lower limit on these
timescales is constrained by the analytical precision with which the ratio of daughter to
parent isotopes can be measured and the half-life of the parent. In the case of the well-
established Ar–Ar dating technique, in the most favourable of circumstances (i.e., high
initial potassium contents) ages down to as little as 2000 yr can be recovered (Lanphere et
al. 2007). For most other long-lived radioisotopes, such as 87Sr, 238U, 235U, 230Th, 147Sm, etc.,
timescales of less than about 1 m.y. are inaccessible. Carbon-14 dating (t1/2 = 5730 yr) affords
glimpses of shorter timescales, although its application relies on knowledge of the global
14
C flux and the availability of carbonised material that can be directly related to volcanism,
such as charred tree trunks in pyroclastic flows.
Chronometry and Speedometry in Olivine, Plagioclase & Pyroxenes 537
With the advent of improved mass spectrometric methods a new class of daughter isotopes
became accessible; those of the short-lived uranium and thorium decay series that form en
route to the stable daughter isotopes of lead. The so-called U-series dating method (e.g., Turner
and Costa 2007; Cooper and Reid 2008) has proven especially valuable in constraining many
magmatic timescales, not least because the daughter isotopes comprise elements with widely
differing chemical affinity. For example, radon is a gas whose loss from the decay chain can
provide valuable insights into degassing processes (e.g., Gauthier and Condomines 1999;
Berlo et al. 2006; Kayzar et al. 2009). Other daughter isotopes are preferentially sequestered
by particular minerals, thereby constraining the onset of crystallisation of those minerals
in an evolving magmatic system. The uptake of radium by plagioclase is a case in point
(e.g., Cooper and Reid 2003; Turner et al. 2003). The wide variety of half-lives of U-series
radionuclides means that, in principle, a wealth of timescales from seconds to millennia can
be recovered. However, once secular equilibrium between parent decay and daughter ingrowth
is established, typically after about 5 daughter half-lives, all chronometric information is lost.
Thus, the central problem of radiometric dating persists; the parent and daughter isotopes can
become decoupled, such that what is dated is a time of decoupling (or cessation of secular
equilibrium). A further limitation of the U-series approach is that the analytical techniques
used do not lend themselves to high spatial resolution. Consequently it may be difficult to
disentangle multiple timescales from zoned crystals, especially where crystal cores and rims
differ greatly in age (e.g., Cooper and Reid 2003). The application of U-series methods to
constrain timescales of magmatic processes is discussed in some detail by Cooper and Reid
(2008). U-series dating of zircons is covered by Schaltegger and Davies (2017, this volume).
The focus of this chapter is an alternative family of geochronometers that rely on the time-
dependent flux (diffusion) of elements in response to chemical potential gradients established
during magmatic processes. These gradients are most readily manifest in zoned crystals of
common minerals in magmatic systems, e.g., Costa et al. (2008), Kahl et al. (2011). As a
crystal grows, it takes up major and trace elements in response to changes in the melt with
which it is bathed. Perturbations to the melt chemistry or the physical parameters that control
uptake, such as pressure, temperature and redox, result in chemical changes between one layer
(zone) and the next in the growing crystal. At the instant of growth, each layer has some simple
relationship to the melt from which it grew, but no such relationship to the previous layer.
The relationship between each layer and the melt may be one of chemical equilibrium or, if
growth is relatively rapid, kinetic factors may dominate (e.g., Watson and Liang 1995; Watson
1996). Foremost amongst these is the rate at which chemical components can be supplied
diffusively to the boundary layer around the growing crystal (Albarède and Bottinga 1972)
and the relative preference of trace species for the surface of growing crystals relative to their
interior (Pinilla et al. 2012). There is a rich literature on the uptake of trace elements into
growing crystals that is beyond the scope of this chapter. What is important is that the growth
of zoned magmatic crystals sets up chemical potential gradients between successive zones.
Diffusion strives to eliminate such gradients by moving chemical components from one layer
to another. The rate at which this happens, the chemical diffusivity, is a complex function of
the diffusion mechanism, the point defect chemistry of the crystal, the temperature at which
diffusion occurs and the geometry of the interface across which diffusion takes place. The
significance of these various parameters is relatively well established through experiment and
theory for a range of geological substrates, both crystals and melts, and the diffusivity of
many trace species has been determined experimentally. Diffusion is, ultimately, a thermally
mediated process, such that it occurs most rapidly at high (magmatic) temperatures and is
effectively quenched once a system cools below a characteristic closure temperature (e.g.,
Dodson 1986; Kohn and Penniston-Dorland 2017, this volume). Consequently, diffusion
chronometry is a very effective means to constrain the timescale between the event that caused
the chemical perturbation responsible for a zone to form in a crystal and the moment at which
this system cooled below some closure temperature when diffusion was effectively arrested.
For explosive volcanic systems closure occurs upon eruption such that diffusion chronometry
is a valuable tool for recovering the timescales of pre-eruptive processes that are responsible
for modifying the chemistry of volcanic crystals. The sensitivity of crystal chemistry to both
intensive parameters (P, T, fO2) and melt composition means that diffusion chronometry can
be used to establish the time-lapse between a change in the configuration of the sub-volcanic
system and its eventual eruption. The challenge is to establish the temperature at which the
original perturbation occurred, the nature of the diffusive interface, the original chemical
potential driving diffusion and the type of diffusion that is occurring. Conversely, for effusive
volcanic systems diffusion of some species may continue during cooling of the lava flow. In
that case, diffusion chronometry may be used to constrain cooling times or flow rates of lava
flows. For example, Newcombe et al. (2014) have shown how diffusion of major and volatile
(H2O, F) elements across the walls of olivine-hosted melt inclusions may be used as a cooling
rate speedometer for basaltic lava flows, while Sio et al. (2013) used Fe–Mg interdiffusion in
olivine phenocrysts to determine the cooling rate of the Kilauea Iki lava lake.
In this review we will explore the use of diffusion-based geochronometers in deriving
magmatic timescales by highlighting a number of recent applications. The considerable potential
of diffusion chronometry in understanding magmatic systems is evidenced by the significant
number of recent papers in high profile journals such as Nature and Science (Martin et al. 2008;
Druitt et al. 2012; John et al. 2012; Saunders et al. 2012a; Ruprecht and Plank 2013; Cooper and
Kent 2014). Here we will discuss the assumptions implicit in these and other applications, their
limitations and the insights that can (and cannot) be gained from this approach. This review can
be read as an update of that given by Costa et al. (2008). Therefore, we will not give a detailed
introduction into the theoretical background of diffusion modelling, which is given elsewhere
(e.g., Ganguly 2002; Watson and Baxter 2007; Costa et al. 2008; Costa and Morgan 2010; Zhang
2010). However, to make it easier for the reader to follow the text and the logic within this
chapter we will illustrate some basics of the approach, wherever necessary, without going into
the mathematical details. In addition the review of Kohn and Penniston-Dorland (2017, this
volume) introduces into some of the basics of diffusion and its role for petrochronology.
Basic approach of diffusion chronometry
Diffusion chronometry (sometimes called “geospeedometry”, e.g., Lasaga 1983; Kohn and
Penniston-Dorland 2017, this volume) is based on the extent to which ion exchange reactions
that are sensitive to a change in some environmental condition (e.g., temperature, pressure,
or chemical potential) proceed during this change. In general, the approach used is forward
modelling of concentration distributions (typically, element or isotope profiles measured along
traverses across sections of crystals, glasses or bulk rocks). Firstly, a diffusion model needs to
be defined by describing the element fluxes and geometry for the real system. The model has to
be translated to partial differential equations that must be solved for the given initial conditions
(the concentration distributions at some defined point in time) and boundary conditions (the
element fluxes at the geometric boundaries of the system). The boundary conditions define the
interaction of the system (typically a chemically zoned crystal) with the exterior, thermodynamic
environment, e.g., a silicate melt. The measured distributions of concentration are then fitted
by this solution, where time (or the temperature–time path) is the only unknown, provided the
relevant diffusion coefficients are known for the conditions of interest. For the fundamentals
of diffusion chronometry, different strategies to define appropriately the diffusion problem,
and methods to solve the respective diffusion equation, we refer the reader to the detailed
reviews of Ganguly (2002), Chakraborty (2008), Costa et al. (2008) and Costa and Morgan
(2010), including the various textbooks on diffusion cited therein. In addition, some discussion
on diffusion and diffusion coefficients can be found in Kohn and Penniston-Dorland (2017,
this volume). For the purposes of this chapter we first summarize briefly the pre-requisites,
advantages, and shortcomings of diffusion chronometry (see also Chakraborty 2006, 2008),
before we address specific examples for olivine, orthopyroxene, and plagioclase.
Prerequisites. (1) A measurable chemical or isotope zoning that is conserved in a mineral

and that has been at least partly produced by diffusion. A homogeneous crystal would allow
for an estimate of the minimum time scales if it can be assumed that the crystal was originally
zoned. (2) Accurate diffusion coefficients: ideally, for the experimental determination of the
diffusion coefficients, their dependence on various intensive thermodynamic parameters
should be quantified, which in addition to P, T, and major element composition could be
oxygen fugacity, water fugacity or, for example, activity of silica in the case of silicates
(e.g., Dohmen and Chakraborty 2007; Zhukova et al. 2014). The relevance of these different
parameters is discussed in detail in Chakraborty (2008). Diffusion of major as well as trace
elements in ionic solids involves the coupling of diffusion to other ions, which might require
the application of a multicomponent diffusion equation (see also discussion on effect of
crystal chemistry and substitution mechanisms in Kohn and Penniston-Dorland 2017, this
volume). (3) An appropriate diffusion model, which includes geometry (e.g., simplified
geometry vs. real crystal shape; 1D vs. 2D vs. 3D models), initial condition (e.g., homogenous
concentration vs. step-like zoning vs. complex zoning), and boundary condition (e.g., fixed
rim composition vs. variable rim composition as controlled by a temperature-dependent
exchange reaction, a geothermometer, or by variable element fluxes from the environment;
open system vs. closed system, Chakraborty and Ganguly 1991).
General Strengths and Advantages. The main strength of diffusion chronometry is that
we can determine a time scale of the respective process (duration) that is independent of the
absolute age of the crystal. Thus the same methodology can be applied to crystals in volcanic
rocks from any age, i.e., Archean to the present day. Moreover, with the right choice of the
element-mineral pair there is no limit on the time scale that can be investigated, from seconds
up to billions of years. Short-lived processes can be determined in old crystals. For example,
with diffusion modelling the cooling rate of refractory high temperature condensates (CAIs)
in the solar nebula or the peak temperature of metamorphism were determined for chondrite
parent bodies at the onset of our solar system (Simon et al. 2011; Schwinger et al. 2016).
Further examples are the determinations of short residence times or ascent rates of crystals
from historical volcanic eruptions (e.g., Druitt et al. 2012; Ruprecht and Plank 2013). In
addition, petrology can be used to relate the chemical gradient in the system to a specific
process (e.g., magma ascent, magma mixing, cooling), and hence the duration of this process
can be determined by diffusion modelling—that is the essence of diffusion chronometry.
Moreover, it is possible to access timescales that are commensurate with observational
timescales of magmatic processes, as discussed above. Consequently there is potential to link
timescales derived from young volcanic crystals with geophysical or geodetic observations of
the volcanoes that erupted them (e.g., Kahl et al. 2011; Saunders et al. 2012a).
Shortcomings and Uncertainties. (1) The accuracy of diffusion coefficients, D, is typically
within 0.2–0.5 log units for the experimental parameter range (T, P, fO2, etc.). This uncertainty
becomes significantly larger when the data have to be extrapolated beyond the experimental
parameter range. Furthermore, considering an isothermal process, the temperature needs to be
known from thermometry, and any uncertainties in T further affect the uncertainty in D. Since the
total extent of diffusion that occurred at this temperature T (generally equivalent to the loosely
defined diffusion profile length) is proportional to Dt , the uncertainty in the respective time,
t, as determined by fitting the measured concentration distribution, comprises both the quality
of the fit and the uncertainty in D. Assuming an almost perfect fit, an uncertainty in D of 0.5
log units translates to an uncertainty in the duration of the isothermal event by the same 0.5 log
units. Even considering these relative large uncertainties in D, the order of magnitude of the
time scale can be determined by diffusion modelling for a specific geologic process, whether
it lasts for seconds, days, months or millions of years. It should be also noted that for trace
elements (as experimentally demonstrated for H and Li diffusion in olivine and clinopyroxene,
see sections below) different diffusion mechanisms and rates may be involved. In this case the
operating diffusion mechanism needs to be identified for the given crystal, since the rates may
differ by orders of magnitudes (see also Dohmen et al. 2016b). (2) The difference between
the initial and the final (observed) concentration distribution (or, more correctly, the chemical
potential distribution) of the element/isotope of interest is the measure of the total extent of
diffusion, which basically contains the time information. The reconstruction of the initial
profile can be ambiguous, but various possible strategies are discussed in the specific case
studies below. In general, the observed concentration profile/distribution to be fitted limits
the overall range of possible initial profiles (see also discussion in Chakraborty and Ganguly
1991; Costa et al. 2008). (3) As for the initial condition, the modeller has to make a decision
regarding the boundary condition based on the petrologic understanding of the system. When
minerals are in contact with a melt it is usually assumed that transport within the melt is
fast enough (at least much faster than in the crystal) and therefore (i) the melt is treated as a
homogeneous phase and (ii) local equilibrium at the crystal–melt interface is attained. These
two assumptions may be incorrect for very compatible or very slow-diffusing elements (see
for example the model examples of Watson and Müller (2009) or Dohmen et al. (2003) or in
the case of very rapid crystal growth. It should be noted that an incorrect boundary condition
can have a stronger affect on the inferred time scale than a wrong initial condition (Costa et
al. 2008). Further discussions on uncertainties of diffusion chronometry are given in Costa et
al. (2008) and Faak et al. (2013). (4) The obtainable timescale and its accuracy are not limited
only by the relevant diffusion rate but also by the spatial resolution of the available analytical
method. If the concentration gradient is very steep, the measured diffusion profile could be
convolved (i.e., not adequately resolved spatially), thus a larger apparent timescale would be
obtained (Ganguly et al. 1988). Considering a Gaussian distribution function for the intensity
of the signal centred at the measured spot, characterized by the standard deviation σ with unit
length, only for a diffusion profile length Dt  σ we can exclude significant contributions
of the convolution effect to the obtained time scale (Ganguly et al. 1988). However, if a steep
concentration gradient can be well resolved spatially it provides a better constraint for the
timescale compared to a shallower gradient due to the fitting procedure, which needs to account
for the precision of the individual data points along a diffusion profile (see for example Figs. 3
and 4 in Ruprecht and Plank 2013). If the gradient is too steep to be resolved by a given
analytical method, it yields a constraint for a maximum permissible timescale of the process.
OLIVINE
Olivine (Ol) has been used to constrain time scales of processes related to mafic rocks (e.g.,
Gerlach and Grove 1982; Coogan et al. 2002, 2007; Pan and Batiza 2002; Costa and Dungan
2005; Kahl et al. 2011, 2013, 2015; Ruprecht and Plank 2013; Newcombe et al. 2014; Albert
et al. 2015; Oeser et al. 2015; Faak and Gillis 2016; Hartley et al. 2016; Rae et al. 2016), where
it is an ubiquitous phase. Application to more silicic volcanic rocks (andesite, dacite) can be
found in Nakamura (1995), Coombs et al. (2000), Costa and Chakraborty (2004), and Martin
et al. (2008), where olivine is often present in the form of xenocrysts or as part of more mafic
inclusions/enclaves; and to basanite in Martí et al. (2013) and Longpré et al. (2014). We will
use olivine as a “reference case” to illustrate the general approach of diffusion chronometry
where, in addition, many recent developments have been made to improve the accuracy of the
method, for example to distinguish diffusion zoning from growth zoning (e.g., Oeser et al.
2015; Shea et al. 2015a) and user-friendly diffusion modeling algorithms are now available
for olivine (e.g., DIPRA, Girona and Costa 2013). This and the following sections will be
organized in a similar way whereby the available diffusion coefficients are first discussed
briefly (more details can be found for olivine in Chakraborty 2010).
Diffusion coefficients
The diffusion rates of major and minor cations occupying the metal site in olivine are
relatively fast when compared to pyroxenes and garnet (e.g., Müller et al. 2013). However, the
published diffusion rates of trace cations in olivine are controversial (e.g., Cherniak 2010 vs.
Spandler and O’Neill 2010, see also Chakraborty 2010). This controversy might be related to
the different diffusion mechanisms that are possible in olivine as demonstrated for Li and H
diffusion (Dohmen et al. 2010 and Mackwell and Kohlstedt 1990, respectively). In any case,
for most practical applications where the electron microprobe was used to map concentration
distributions or measure element profiles, the most relevant elements are Fe, Mg, Ca, Ni and Mn
for which diffusion datasets exist that have been reproduced in various studies using different
experimental approaches over a large temperature range (700–1250 °C), and therefore data
extrapolation is, for the most part, unnecessary (Fe–Mg interdiffusion: Chakraborty 1997;
Dohmen and Chakraborty 2007; Holzapfel et al. 2007; Ca “tracer” diffusion: Coogan et al.
2005; Ni and Mn “tracer” diffusion: Petry et al. 2004). Diffusion of these cations (and other
minor or trace elements occupying the M-site) is in principle a coupled diffusion process that
would require a multi-component diffusion model (e.g., Lasaga 1979), but it can be shown
numerically that diffusion of minor or trace elements having the same charge as the major
cations (Fe2+ and Mg2+) can be treated independently by using the so-called “tracer diffusion
coefficient” as measured experimentally (see Costa et al. 2008 or Vogt et al. 2015 for numerical
examples). The diffusion coefficients of Fe–Mg interdiffusion, and Mn, and Ni tracer diffusion
are rather similar and respective diffusion profiles are of similar length, but the Ca tracer
diffusion coefficient is about an order of magnitude smaller. Other trace elements in olivine
have also been used as a diffusion chronometer in cases where their concentration zoning is
large enough to be resolved by other micro-analytical methods such as laser ablation inductively
coupled mass spectrometry (LA-ICP-MS, e.g., Qian et al. 2010), Fourier transform infrared
spectroscopy (FT-IR) for H (e.g., Demouchy et al. 2006), or secondary ion mass spectroscopy
(SIMS) for Li (e.g., Jeffcoate et al. 2007). To measure short chemical zoning profiles in olivine
nm-scale analytical methods can be also applied, like Nano-SIMS, or field emission electron
microprobe (FEG-probe), or time-of-flight SIMS (TOF-SIMS). Alternatively, back-scattered
electron (BSE) imaging can be used to infer the major component (fayalite–forsterite) zoning
in olivine (Martin et al. 2008; Hartley et al. 2016; Rae et al. 2016), for details of this approach
see the section on orthopyroxene (Opx) and Morgan et al. (2004).
Determination of magma residence times using Fe–Mg, Ca, Ni, and Mn diffusion
Magma residence times by diffusion modelling of chemical zoning in olivine
phenocrysts or xenocrysts have been determined for a large variety of volcanic rocks, from
basanite to dacite (e.g., Costa and Chakraborty 2004; Costa and Dungan 2005; Marti et al.
2013; Albert et al. 2015). In all of these studies a straightforward diffusion chronometry
approach was applied whereby the system is treated as isothermal, such that a constant
diffusion coefficient can be used. The simplistic scenario commonly used is that the olivine
crystal became suddenly exposed to a different chemical/magmatic environment (which
could be related to a sudden change in temperature, magma mixing, decompression, magma
recharge etc…), the olivine rim re-equilibrates with this new chemical environment by
element exchange or an overgrowth in equilibrium with the melt is formed, then progressive
internal re-equilibration of olivine with time occurs by solid-state diffusion. The system and
hence the chemical zoning is eventually quenched by the volcanic eruption and the chemical
zoning is therefore a measure of the time span between the change of the environment and
the eruption. Thus, we do not measure the total residence time (or life time) of a given
crystal, but infer instead timescales related to the dynamics of the magmatic system.
Figure 1 shows a chemical zoning map from Kahl et al. (2011), which evidently cannot be
fitted by a model wherein the olivine was initially homogeneous and only the rim composition
changed. Compositional plateaus that could be identified in olivine phenocrysts in basalts of Mt.
Etna indicate various growth stages. Equally, a compositional profile such as shown in Figure 1
cannot be produced by fractional crystallization in a continuously cooled system. Thus, Kahl
et al. (2011) assumed that chemical composition of each of these plateaus reflects a different
magmatic environment. Based on a system analysis of the diverse compositions and types of
zonings, three different magma environments were identified. Diffusional relaxation between
these compositional plateaus contains information on the time spans between the different growth
steps and the final eruption. Here two sequential diffusion-modelling steps were performed
assuming that specific regions in the profile reflect different growth stages of the crystal in two
respective chemical environments (concept illustrated in Figure 2). Finally, from simulation of
chemical zoning in 28 crystals from the 1991–1993 SE-flank eruptions, the dynamic evolution of
the plumbing system was reconstructed by incorporating information from volcano monitoring.
The same approach was applied to eruptions between 2001 and 2008 of Mt. Etna (Kahl et al.
2013, 2015) where, in addition, thermodynamic phase equilibrium calculations with MELTS
(Ghiorso and Sack 1995) were performed. It was confirmed that different types of olivine core
compositions (plateau regions) “reflect compositionally distinct batches of magma that have
resided in different sections of the plumbing system”. The MELTS calculations allowed Kahl
et al. (2013, 2015) to identify the intensive parameters that control the different types of olivine
compositions in these magmatic environments. Water content in the melt and temperature of
the magma appear to exercise the dominant control. In addition the connectivity between these
different magmatic environments was established and, finally, diffusion modelling was used
to obtain the time scales (ranging from days to 2 yr) of mixing between these different magma
batches and the time of final eruption.
Pronounced compositional steps in olivine were also found in basaltic andesites from the
1963–1965 eruption of Irazu volcano, Costa Rica, indicating again magma mixing (Ruprecht
and Plank 2013). In particular Ni shows strong variations in primitive olivine crystals with
high and relatively constant forsterite contents (> 90%). This variation was ascribed primarily
to an olivine crystallization trend in primary melts of the mantle, as olivine controls the Ni
budget in the melt. It further implies that primary melts coexist and mix in the source region.
Corresponding zoning of P was used to infer the growth zoning and estimate the effect of
diffusion for Ni. Diffusion modelling of the Ni zonations indicates that, between the magma
a Mg b c 0° a = 3°
80 b = 20°
c = 70°
78 Traverse
Fo [mol %]
a
c
76 270° 90°
74 b
100 µm rim core rim
lower
180° hemisphere
50 100 150 200 250 300
Distance [µm]
Figure 1. (a) Mg distribution map, (b) Fo concentration profile with model fit, and (c) respective profile
orientation data of an olivine crystal from the 1991–1993 SE-flank eruptions products of Mt. Etna (Sicily).
The dotted line in (b) is the assumed initial step profile formed during three different growth stages (core,
1st overgrowth, 2nd overgrowth) reflecting different magmatic environments. The model fit (solid line) re-
produces the measured concentration profile by two sequential diffusion-modelling steps, which indicates
that the 1st overgrowth was formed about 250 days before the 2nd overgrowth, which was formed 113 days
before eruption. Figure modified after Kahl et al. (2011). Note the asymmetric shape of the zoning and the
crystal habit in 2D, which is most likely a sectioning effect based on the simulations of Shea et al. (2015b).
g
a t0
C0
C
x
b C1 C1
C0 Intrusion I
C tA
x
c C1 C1
tA
C0
C
d
C1
C0
C1
Intrusion II
C tB
C2 C2
e
C1 C1
C0
C
C2 C2
tB
x
f
C1 C1
Eruption
C C0 tC
C2 C2
x
Figure 2. Illustration of the concept to reconstruct the growth history and residence times in different
magmatic environments of a phenocryst by sequential diffusion modelling. (a)–(f) illustrate concentration
profiles (x denotes distance) induced sequentially by magmatic events during time line (g): (a) At some
time t0 a homogeneous crystal with composition C0 was formed; (b) After the crystal came into contact with
a different magmatic environment at time tA (e.g., a new batch of melt arrived) a new rim with composition
C1 was formed; (c) After the formation of the rim, a diffusive flux starts to smooth the initial step profile;
(d) At time tB a second magmatic event (intrusion, magma mixing) changes the local thermodynamic
conditions and a second rim is formed with composition C2; (e) Concentration gradients between C0, C1
and C2 tend to homogenize the entire concentration profile by chemical diffusion. (f) The eruption cools
the magma quickly and freezes the concentration profile, which is eventually a product of growth and dif-
fusion zoning (black solid line). The black dashed line represents the concentration profile produced only
by growth without diffusion (the initial profile used for diffusion modelling). The duration of diffusion at
different stages before eruption, here ΔtA and ΔtB, are the quantities that can be obtained by simulating the
observed concentration profiles. (Figure modified after Kahl et al. 2011).
mixing processes deep within the mantle and final eruption, a time between months and years
elapsed. That implies the erupted crystals record recharge of the magma during the course of the
eruption. Because the mixing process occurred in a source region at about 35 km depth, from
this timescale an integrated ascent rate of the order 10–50 m per day was estimated. We discuss
below how an ascent rate can be inferred more directly from H diffusion profiles in olivine.
Growth vs. diffusive zoning: stable isotopes as diffusive fingerprints
A fundamental problem of the approach described above is to identify how much of
the zoning was formed simply by growth (the initial profile) and how much was modified
subsequently by solid-state diffusion. Several strategies can be applied to exclude that the
inferred time scale was biased due to pre-diffusional crystal growth:
First, diffusion in olivine is anisotropic (e.g., for Fe–Mg interdiffusion, D//c = 6 × D//a
= 6 × D//b in the T range 700–1250 °C, diffusion anisotropy changes at higher temperatures,
Tachibana et al. 2013); this allows to check whether the relative length of diffusion profiles in
the same crystal along different directions is in accordance with the predicted relative profile
lengths (e.g., Costa and Chakraborty 2004; Costa et al. 2008); crystal growth would not be
expected to be similarly anisotropic. Crystallographic orientations of measured profiles are
usually measured using electron back-scatter diffraction (EBSD).
Secondly, the zoning of major or trace elements that apparently have slow diffusivities could
be a tracer of the original zoning type and different growth stages. For example P in olivine
conserves typically complex zoning (e.g., Milman-Barris et al. 2008; Welsch et al. 2013) that in
some cases indicates dendritic growth of olivine in the initial phase (Welsch et al. 2014; Shea et
al. 2015a), but zoning of major elements and other divalent cations is typically smooth. Bouvet
de Maisonneuve et al. (2016) compared trace element zoning (P, Ti, Sc, V, Al) vs. major and
minor element zoning (Fe–Mg, Ni, Mn, Ca) in olivine and plagioclase. They concluded that trace
elements like P, Ti, Sc, and Al in olivine record more and earlier events in the magmatic system
than Fe–Mg, Ni, Mn, and Ca. However, it was also concluded that incorporation of these trace
elements is decoupled from those of the major and minor elements, and their zoning is controlled
by kinetic factors related to rapid growth rather than equilibrium uptake into olivine.
Thirdly, zoning of Fe–Mg, Ca, Ni, and Mn due simply to crystal growth is distinct from
that due to diffusion (Costa et al. 2008). Consequently, if diffusion, rather than growth, is
the cause of the observed zoning patterns, then simultaneous diffusion modelling of different
elements should ideally give the same or similar time scale (e.g., Costa and Dungan 2005).
Fourthly, isotopic zoning provides detailed insights into the origin of chemical zoning (Sio et
al. 2013; Oeser et al. 2015). Light isotopes diffuse slightly faster than heavy isotopes (the “isotope
effect”, e.g., Schoen 1958). Thus, in contrast to equilibrium partitioning, isotopic diffusion
produces strong isotope anomalies even at magmatic temperatures (Richter et al. 1999, 2003).
In the case of Fe–Mg interdiffusion in olivine, such diffusive fractionation results in negatively
correlated isotope anomalies for Mg and Fe due to their opposing diffusion flux directions. Strong
negative correlations of δ56Fe with δ26Mg isotopes were first identified by Teng et al. (2011) (see
also Dauphas et al. 2010 for first indications of this effect) in analyses of olivine fragments from
Hawaiian basalts (Fig. 3), a feature that can be explained by diffusive fractionation. Diffusion
modelling and in situ measurements of Fe and Mg isotopes by microdrilling and of Fe isotopes
by in situ measurements using LA-ICP-MS and multi collector SIMS (MC-SIMS) found
a similar correlation as in Figure 3 (Sio et al. 2013), confirming this explanation. In addition
Sio et al. (2013) were able to reproduce the known cooling rate for these samples by diffusion
modelling of the fayalite zoning (which also confirms the experimental diffusion data). New
analytical developments using a femtosecond laser ablation system allow in situ analysis of
both Fe and Mg isotopes (Oeser et al. 2014). Chemically zoned olivine crystals from different
types of basaltic rocks have been analysed using this technique by Oeser et al. (2015, Fig. 4).
0.6
2
0.4
0.2
0.0
Fe (‰)
-0.2
56
-0.4
-0.6 KOO-17A
T4D4-01
-0.8 KI75-1-139.3
T4D2#1
KI81-5-254.5
-1.0
Oceanic basalt
-0.40 -0.30 -0.20 -0.10 0.00 0.10

26
Mg (‰)
Figure 3. Mg and Fe isotope fractionations in olivine fragments from Hawaiian lavas. The values for
δ56Fe = ((56Fe / 54Fe)Sample / (56Fe / 54Fe)Standard − 1) ×1000 and δ26Mg = ((26Mg / 24Mg)Sample / (26Mg / 24Mg)Standard
− 1) ×1000 define a linear relationship with a slope of −3.3 ± 0.3 (solid line). Olivine data are reported in
Table 1 of Teng et al. (2011). Oceanic basalt data are reported in Teng et al. (2008, 2010, 2011). Figure
modified after Teng et al. (2011).
As in Sio et al. (2013) the clear fingerprint of diffusion on Fe and Mg isotopes was observed. In
addition Oeser et al. (2015) discussed and modelled different scenarios, including growth and
dissolution, to produce chemical zoning and associated isotope zoning. Based on these models it
can be concluded that solid state diffusion leaves a unique fingerprint on the Fe and Mg isotopes
and, for example, it can be distinguished whether the olivine zoning was formed by element
exchange at the surface or whether an overgrowth was formed (Fig. 5). In most of the olivine
crystals investigated by Oeser et al. (2015) the isotope variations could be reproduced by the
diffusion model shown in Figure 4, indicating element exchange at the rim, potentially driven by
cooling of the magma. However, the parameter β, which for two isotopes with masses Mi and Mj
is used to empirically describe the isotope effect according to the relation Di / D j = ( M j / Mi )β,
was used as a free parameter for the fitting procedure and varied for Mg and Fe isotopes between
0.06–0.12 and 0.08–0.2, respectively, with the value for Mg systematically about a factor of
two smaller. Whether this variation is related to additional processes responsible for the isotope
anomalies or related to anisotropy and compositional effects should be clarified by experiments
to calibrate β. As, for example, discussed by Van Orman and Krawczynski (2015) this parameter
is controlled by the diffusion mechanism and should be somehow coupled for elements that
interdiffuse with each other such as Fe and Mg in olivine.
Determination of cooling rates from Fe–Mg zoning in olivine
Based on the assumption that the Fe–Mg zoning was produced during cooling, Oeser et
al. (2015) determined the respective cooling rates of basaltic rocks by applying an approach
suggested by Ganguly (2002) (for another simplified approach to determine cooling rates from
diffusion models, see also Watson and Cherniak 2015). In this model the diffusion equation
is transformed and the diffusion coefficient is integrated over time (see also Lai et al. 2015 or
Schwinger et al. 2016), which is a measure of the total extent of diffusion (“diffusion profile
length”), and is used as a fitting parameter. The reciprocal cooling rate for an assumed initial
Fe
olivine
Mg
melt melt' melt'
16 b c
28
14
Mg [wt%]
Fe [wt%]
26
12
10 24
rim core rim core
0.0
d e
Fe [‰, IRMM-14]
26Mg [‰, [DSM-3]
0.0
-0.4
-0.2
-0.8
-0.4
56
-1.2
-0.6
-1.6
rim core rim core
200 400 600 800 200 400 600 800
Distance [µm] Distance [µm]
Figure 4. Simplified modeling of Fe–Mg chemical and isotopic diffusion profiles for the Massif Central ol-
ivine assuming element exchange at the crystal rim after a homogenous olivine was formed. (a) Illustration
of the modelled scenario: pure Fe–Mg interdiffusion due to a compositional contrast between olivine and
melt. (b)–(e) Measured data and modelled profiles for (b) Fe, (c) Mg, (d) d56Fe, and (e) d26Mg. Modelling
results are shown as solid lines, which indicate either an effective residence time of 3.26 yr at 97% of the
peak temperature of 1297 °C, or alternatively a cooling rate of 57 °C/ yr−1 from this peak temperature. Note
the negative and positive bumps for d56Fe and d26Mg, respectively, whose minimum/maximum coincides
with the position of the strongest curvature in the chemical profiles. The negative bump for d56Fe forms
because the diffusive flux of Fe into the olivine is faster for the lighter Fe isotope (56Fe) than for the heavier
one (58Fe). In the case of Mg, diffusion is in the opposite direction, i.e., out of the crystal, hence a positive
bump forms for d26Mg. Figure modified after Oeser et al. (2015).
temperature can be determined from this fitting parameter. For the different crystals and intra-
plate volcanoes studied, the inferred cooling rates varied typically between 30–300 °C/yr.
Fe–Mg exchange between olivine and melt is sensitive to temperature but apparently in
the basaltic rocks studied by Sio et al. (2013) and Oeser et al. (2015) this is not reflected in
the profile shape (e.g., Fig. 4). Here a fixed rim composition is assumed, which during cooling
did not change significantly. The interplay between the exchange reaction at the rim controlled
by temperature and the resulting diffusion profile is discussed in more detail below for Ca in
olivine and Mg in plagioclase.
a
Fe
Mg
olivine
melt melt' melt'
b 28
15
27
14
Mg [wt%]
Fe measured 27
Fe [wt%]
Fe modeled
13 26
Mg measured
Mg modeled
26
12
25
11
25
rim core rim
0.20
Fe and Mg [‰]
Fe modeled
0.10 Mg modeled
26
0.00
56
-0.10
-0.20
rim core rim

50 100 150 200 250 300
Distance [µm]
Figure 5. Simulation of Fe and Mg diffusion for a scenario (a) where an initial step profile was formed
due to quick overgrowth onto a homogeneous olivine core. (b) Fe and Mg concentration profiles across
an olivine phenocryst (EH_31-10-11_OL9) investigated by Longpré et al. (2014) and the respective
simulation by diffusion modelling. (c) The corresponding Fe–Mg isotopic profiles across this olivine as
expected from modelling the diffusion-generated chemical zoning. The Fe and Mg concentration data
are from Table DR4 in Longpré et al. (2014). Note that the bulk crystal does not show any isotope frac-
tionation due to internal diffusive redistribution of elements and their respective isotopes. Internally, the
d56Fe and d26Mg values are negatively correlated because of the interdiffusion process, but each show
both a negative and a positive bump in the profile. This isotopic signature is clearly different to that
produced in the scenario for Figure 4. Figure modified after Oeser et al. (2015).
Determination of cooling rates from Ca zoning in olivine

The cooling rate of rocks can alternatively be determined by using the closure profiles
concept of Dodson (1986) or the extension of this formulation by Ganguly and Tirone (1999)
(see also Kohn and Penniston-Dorland 2017, this volume). This approach requires that the
composition at the crystal rim changes significantly during cooling (approximately linear with
time) and that the external environment behaves as an infinite reservoir. An example, applied
by Coogan et al. (2002), is the temperature-sensitive exchange of Ca between olivine and
clinopyroxene (Cpx) (Köhler and Brey 1990). The net exchange reaction may be written as:
Mg2SiO4 (Ol) + CaMgSi2O6 (Cpx) = CaMgSiO4 (Ol) + Mg2Si2O6 (Cpx) (1)
For a minor variation in the major element composition of Ol and Cpx (the enstatite
component in Cpx and forsterite component in Ol) according to this reaction an effective partition
coefficient of Ca between Ol and Cpx, ln K pOl-Cpx
(Ca ) , can be defined. The temperature (and pressure)
dependence of ln K pOl-Cpx
(Ca ) was calibrated by Köhler and Brey (1990) (a more recent calibration
considering also the effect of the forsterite content is given in Shejwalkar and Coogan 2013):
−0.425P [ MPa ] − 5792 − 1.25 ( T )

(Ca ) =
ln K pOl-Cpx (2)
(T )
Equation (2) predicts that ln K pOl-Cpx
(Ca )
decreases with decreasing temperature, leading to a
diffusive flux of Ca out of olivine and into clinopyroxene during cooling. According to the
exchange reaction in Equation (1) this directly implies an opposing flux of Mg. Assuming that
this exchange reaction is controlled by Ca diffusion in olivine (i.e., clinopyroxene is considered
to be an infinite reservoir for Ca in this case), a closure temperature, Tc, can be defined,
which depends on the distance, x, from the interface (Dodson 1986; Onorato et al. 1981).
At Tc(x) the diffusive flux becomes too small to effectively change the composition (in this
case the Ca content). Thus, the rims of an olivine crystal will be able to maintain equilibrium
Ca-concentrations down to lower temperatures than the core, leading to the development of a
closure profile that is convex upwards for a continuous cooling history (Fig. 6). According to
Dodson (1986) this closure profile (Tc-profile) can be calculated as:
E
Tc ( x ) =
  RT D  2
 (3)
R ln   c 2 0  + G ( x ) 
 Esa  
where R = ideal gas constant, s = cooling rate around the closure temperature Tc, and a = radius
of the grain. The parameters E and D0 refer to the activation energy and pre-exponential factor
in the diffusion equation. Coogan et al. (2005) provide experimentally determined diffusion
coefficients for Ca in olivine (e.g., E = 207 kJ/mol and D0 = 1 × 10−10 m2/s for diffusion along
the c-axis and fO2 = 10−7 Pa). The closure function G(x) depends on the geometry of the cooling
object and the position x within it. For example, using the geometry of a sphere (Dodson 1986):
( −1) sin(nπx )ln ( nπ )

n +1
∞
G( x ) = γ + 4∑ (4)
n =1 nπx
where γ ≈ 0.57721 is Euler’s constant. As Tc appears on both sides of Equation (3), the equation
has to be solved iteratively.
Coogan et al. (2002, 2007) applied the above set of equations to determine cooling rates of
gabbros from the Oman ophiolite using the measured Ca-concentration profiles in olivine. These
profiles can be translated into closure temperatures using the thermometer given in Equation (2):
d T /d t d T /d t
0.1°Cyr-1 0.001°Cyr-1
a b
0.004 0.004
Ca [atoms p.f.u.]
Ca [atoms p.f.u.]
0.003 0.003
0.002 0.002
0.001 0.001

0 500 1000 1500 0 500 1000 1500
T0=1200°C T1=1100°C T2=1000°C T3=800°C T4=600°C
closure temperature [°C]

0.0014 c 900 d
Ca [atoms p.f.u.]
0.0012 850
0.0010 800 closure profile Tc(x)
0.0008 for dT/dt = 0.0013°Cyr-1
750
0.0006
700
0.0004
0.0002 650
200 400 600 800 200 400 600 800
Figure 6. (a) and (b) Simulated Ca concentration profiles in olivine for two different cooling rates based on dif-
fusive exchange between Cpx and Ol with the temperature-sensitive partition coefficient given in Equation (2).
(c) A measured Ca concentration profile and corresponding closure profile fit (d) according to Equation (5).
−0.425P [MPa] − 5792

=Tc ( x ) − 273
 CCa
Ol
( x) 
ln  Cpx  + 1.25
(5)
 CCa 
Ol
where P = pressure, CCa (x) = measured concentration of Ca in olivine at position x,
Cpx
CCa = measured concentration of Ca in clinopyroxene (adjacent to the olivine grain).
Cpx
The assumption that Cpx behaves as an infinite reservoir directly implies that CCa is
effectively homogeneous. The cooling rate s in Equation (3) is iteratively changed until
the best fit between the closure profile calculated from Equation (3) and the closure profile
calculated from Equation (5) is attained (Fig. 6).
Coogan et al. (2007) performed numerical diffusion simulations to test whether the
assumptions of the Dodson model are justified. For example, whether the environment of
olivine can be treated as an infinite reservoir for Ca depends on the relative diffusion rates
in the exchange couple (Ca diffusion in clinopyroxene and olivine) and their relative modal
abundances. Although Coogan et al.(2007) have demonstrated that in this specific case this
assumption is not exactly fulfilled, it could be shown that the final error for the derived
cooling rate with the Dodson model is on the order of 0.1–0.2 log units. In addition, further
errors were introduced because of a simplified geometry for the olivine (sphere) and the fact
that diffusion anisotropy was not considered.
VanTongeren et al. (2008) used the same approach as Coogan et al. (2002, 2007) to
obtain cooling rates from the lower oceanic crust exposed in the Wadi Khafifah section of
the Oman ophiolite. However, instead of measuring and fitting full concentration profiles,
these authors measured between 2 and 7 (on average 3) analyses per olivine grain and then
used the highest measured Ca values from each crystal core for the calculation of a single Tc
and bulk cooling rate. Following this approach, they report uncertainties up to 1.5 log units
on the obtained cooling rates. Faak and Gillis (2016) show that chemical alteration as well as
secondary fluorescence during electron microprobe analysis from adjacent Ca-rich phases can
lead to anomalously high Ca contents, which can only be filtered out meaningfully by looking
at complete zoning profiles. They demonstrate how the cooling rate extracted from averaged
analyses in the core can overestimate the cooling rate by up to 2 orders of magnitude compared
to that extracted from diffusion profiles over the whole grain.
Determination of cooling rates to constrain thermal structure of a crustal segment
Quantifying the rate of temperature changes of rocks during magmatic processes additionally
provides insights into energy transfer processes. Different processes of heat exchange (e.g.,
conductive vs. convective heat transport via hydrothermal circulation) leave behind different
patterns of cooling rates as a function of lateral and vertical position within a cooling body.
Mapping cooling rates recorded in natural rock samples provides the possibility to determine
the thermal structure and thermal evolution of a crustal segment formed by igneous processes.
Coogan et al. (2002, 2007) used the Ca-in-olivine geospeedometer to determine cooling
rates as a function of stratigraphic depth (i.e., cooling rate profiles demonstrated) within five
different segments of the plutonic section of the oceanic crust. They demonstrated that the Ca
content in olivine was systematically reset to lower temperatures for deeper gabbros, indicating
systematically lower cooling rates with increasing depth (e.g., for fast-spreading mid-ocean
ridges cooling rates decreased from 0.1 °Cyr−1 at the top of the plutonic section to 0.0001 °Cyr−1
at the Moho, Fig. 7). An issue here is that the Ca concentrations towards the olivine rim could
not be reproduced by the model, which was explained by fluorescence effects of the electron
microprobe measurements. A comparison between cooling rate profiles predicted from different
thermal models and their associated cooling rate profiles from diffusion modelling allowed them
to extract information about the dominant mechanism of heat loss in the lower oceanic crust.
Faak and Gillis (2016) derived cooling rates from olivine gabbros and troctolites formed
at the fast-spreading East Pacific Rise and drilled from the Hess Deep Rift. Here, the Ca
exchange between olivine and clinopyroxene as well as Mg exchange between plagioclase
and clinopyroxene were used to obtain cooling rates from diffusion modelling (the details
on the Mg-in-plagioclase thermometer are given below). The obtained cooling rates for the
deeper level plutonics (> 2km below the dike/gabbro boundary) span a range from 0.005 to
0.0001 °Cyr−1, with a mean value of 0.0011 °Cyr−1. In combination with cooling rates obtained
from shallow level gabbros from the Hess Deep Rift (Faak et al. 2014), they were able to
construct a cooling rate profile over ~2 km, starting from the top of the plutonic sequence
(Fig. 7). According to Faak and Gillis (2016), both the shape of the cooling rate profile and the
actual values are best explained by thermal models with near-conductive cooling of the lower
crust and hydrothermal heat extraction at the top.
Determination of magma ascent rates using H diffusion
By analogy with the determination of cooling rates where diffusion modelling is coupled
to a thermometer, for the determination of decompression or ascent rates a barometer is
needed. Olivine, a nominally anhydrous mineral (NAM), contains H in trace amounts up
to 1000 ppm, and experimental calibrations demonstrate that the solubility of H in olivine
increases strongly with pressure in water-saturated settings (e.g., Kohlstedt et al. 1996).
Thus, during isothermal or adiabatic decompression in fluid-saturated magmas a strong
-1
Cooling rate dT/dt [°Cyr ]
0.0001 0.001 0.01 0.1
0
Data sets
200 shallow Hess Deep Rift
(Faak et al., 2014;
400 gabbros Coogan et al., 2007;
Faak & Gillis, 2016)
600 Oman ophiolite

(Coogan et al., 2007)
800
below DGB [m]
Diffusion
chromometer
Depth
Mg-in-Pl
Ca-in-Ol
Thermal models
Conductive
(cooled from
>2000 the top)
deeper
plutonics Hydrothermal
(extensive deep
hydrothermal
circulation)
-4 -3 -2 -1
log dT/dt [log °Cyr-1]
Figure 7. Cooling rates obtained from the Ca in olivine and Mg in plagioclase diffusion modelling plotted
against the stratigraphic depth of gabbroic rocks from the Hess Deep Rift and the Oman ophiolite. The
cooling rate profile is compared to predictions of cooling rates with depth from different thermal models
(for details see Faak and Gillis 2016). Figure modified after Faak and Gillis (2016).
driving force for H diffusion out of olivine exists. Unlike most other elements, since H is the
fastest diffusing element in olivine (Kohlstedt and Mackwell 1998; Demouchy and Mackwell
2006) its H content is able to respond rapidly to magma ascent (e.g., Fig. 8), as demonstrated
by Demouchy et al. (2006) and Peslier and Luhr (2006) who measured H profiles in olivine
from mantle xenoliths using Fourier transform infrared spectroscopy. FT-IR spectroscopy
is very sensitive to O–H bonds in different materials and is routinely used to analyse the H
content in NAMs as well as to infer its chemical environment. From diffusion modelling of
H, a total travel time to the Earth’s surface can be determined and after estimating an initial
depth, an ascent rate can be calculated. With this approach, the following olivine ascent rates
were inferred: ~3–9 m/s for garnet–peridotite xenoliths in alkali basalts (Fig. 8; Demouchy et
al. 2006); 0.2–0.5 m/s for spinel–peridotite xenoliths in alkali basalts (Peslier and Luhr 2006);
5–37 m/s for garnet and spinel–peridotite xenoliths in kimberlites, Peslier et al. (2008);
3–12 m/s for spinel–peridotite xenoliths in (Eifel volcanoes, Denis et al. 2013); and 0.2–
25.3 m/s (Hawaian volcanoes, Peslier et al. 2015). These ascent rates are broadly consistent
with independent methods (see for example Fig. 7 in Peslier et al. 2015).
However, there are a number of additional complexities that need to be considered in this case:
(i) Estimation of the initial profile: Comparison of the water contents between pyroxenes and
olivine indicate that the olivine cores have lost their original H content, once in equilibrium with
the thermodynamic environment at greater depth (e.g., Demouchy et al. 2006; Denis et al. 2013).
The H contents of pyroxene are in general not zoned, which indicates that they still preserve their
initial H content. This observation appears to be consistent with the known relative diffusion
rates of H in orthopyroxene and olivine. The initial H content in olivine can be then reconstructed
by using the H partition coefficient between orthopyroxene and olivine, K pOl/Opx (H) . Either the
experimentally measured K pOl/Opx(H) were used (e.g., Peslier and Luhr 2006; Denis et al. 2013) or
estimated indirectly using experimental calibrations of the water content in orthopyroxene as a
function of water fugacity, which was used to estimate the water fugacity in the magma chamber
at depth. In thermodynamic equilibrium this water fugacity reflects also the initial water content
Diff // [100] Diff // [010] Diff // [001]
50
40
H2O [µg/g]
30
20
10
0.51 1.5 2 0.5 1 1.5 0.5 1 1.5

x [mm] y [mm] z [mm]
Figure 8. Water content of olivine (in ppm H2O by weight) as a function of position parallel to each axis
using calibration of Bell et al. (2003). Polarized infrared profiles are across an olivine crystal from Pali-
Aike alkali basalt, Patagonia, Chile. Solid lines are not a fit but represent calculated diffusion profiles
assuming the slower diffusion mechanism for H in olivine. As a function of temperature, different dura-
tions for dehydrogenation are extracted and yield a global ascent rate of 6 ± 3 m/s Dotted lines give ±5%
on ascent time. Note that the measured data indicate a stronger H loss parallel to the a-axis compared to
the c-axis, which would be more consistent with the faster diffusion mechanism according to Kohlstedt
and Mackwell (1998). See also the more recent model calculations of Thoraval and Demouchy (2014)
for these profiles. Figure modified from Demouchy et al. (2006).
in olivine and which can be calculated in an analogous way by using experimental solubility data
(Demouchy et al. 2006). In the case that the pyroxenes may have also lost their initial H content,
an alternative way of estimating the initial H content in olivine could be used for fluid-saturated
systems: the water fugacity can be calculated from available thermodynamic models (for
example using the software VolatileCalc, Newman and Lowenstern 2002) if realistic estimates
of the initial volatile contents can be made, e.g., from melt inclusions. Assuming thermodynamic
equilibrium at depth, the experimental calibration of the water content in olivine as a function of
water fugacity, P, T, and XFe (e.g., Zhao et al. 2004) could be used to calculate the equilibrium
content of H2O in olivine (Fig. 9). In any of these cases independent estimates of temperature and
pressure are necessary to perform these calculations. If, for example, the estimate of the initial H
content were too high, a slower ascent rate would be predicted.
(ii) The boundary condition: In all published models so far a fixed boundary condition
was used, where it has been assumed that from the outset of magma ascent the entire H
content at the rim is lost. This assumption is clearly erroneous as the H content at the rim
tends to equilibrate with the local environment, which defines at depth a water fugacity that
is significantly greater than 0 Pa (see for example the discussion in Peslier et al. 2008). With
decreasing depth the water fugacity decreases (more strongly in fluid-saturated systems)
imposing an increased driving force for H diffusion out of olivine (or any other NAMs), as
illustrated in Figure 9. As a net result, by assuming a fixed boundary condition the ascent rate
would be overestimated (e.g., Costa et al. 2010b).
(iii) Another consequence of the loss of H2O in the core is that a 3D model is required
since diffusion fluxes from all directions out of the core need to be taken into account. Due to
the diffusion anisotropy of H, it is also necessary to measure the H concentration distribution
in a cross section along two main crystallographic axes, ideally including the fastest diffusion
direction to compare with the model calculations (Demouchy et al. 2006). Measurement along
one just profile direction can be very misleading (as illustrated in Thoraval and Demouchy 2014).
In a recent paper Ferriss et al. (2015) illustrated an approach by performing transmission FT-IR
on whole crystals without cutting them to compare the measurements with 3D simulations.
(iv) Diffusion and incorporation mechanisms of atoms in crystals are strongly coupled. How
H in olivine is incorporated is still debated with little agreement between research groups in the
interpretation of the IR spectrum and hence the structural site of water (e.g., see the review article
of Beran and Libowitzky 2006). This is chiefly related to the presence of different peaks in the IR
spectrum of olivine, which vary markedly from one olivine to another, in contrast to other NAMs.
The following sites have been suggested for olivine: hydroxyl groups associated with vacancies
either on the metal site or the Si site; or hydroxyls coupled to trace elements like Al or Ti. Certain
peaks in the spectrum have been clearly identified to be characteristic of association of a hydroxyl
group with trivalent cations (Fe3+, Cr3+, Al3+) and Ti4+ (Berry et al. 2005, 2007). Interestingly,
the latter peaks dominate in most IR spectra of mantle olivines, whereas in most experimental
high-pressure studies where large amounts of water are incorporated (> 200 ppm) these peaks are
absent or very minor. Consequently, it has been established that at higher pressures (Mosenfelder
et al., 2006; Withers and Hirschmann 2008) there is no variation of solubility with different trace
element concentrations e.g., Al, Cr, Ti, and Fe3+. These observations show that various sites for
hydroxyl in olivine have to be considered and in particular for hydroxyls that are associated with
other trace elements this must have also a consequence for the dehydrogenation mechanism and
the corresponding rate. For example, for the “Titanium-clinohumite” defect (Walker et al. 2007),
H2O in olivine [µg/g]

Depth [km]
P [MPa]
log (fH2O [Pa])

Figure 9. Model calculation to illustrate the driving force for water diffusion out of olivine during
ascent. Water fugacity (dotted line) for a basaltic melt was calculated as a function of pressure using
VolatileCalc (Newman and Lowenstern 2002) assuming a closed system, T = 1200 °C, and initial H2O
and CO2 contents of 5 wt% and 3000 ppm, respectively. The corresponding H2O content in olivine in
equilibrium with the melt (solid line) was calculated using the experimental calibration of Zhao et al.
(2004). Note that only in the second stage during ascent, after H2O degasses from the melt, can a strong
decrease of H2O in the olivine be expected.
dehydrogenation could form Ti-bearing precipitates and/or involve the coupled flux of Ti and
H. However, diffusion of H in olivine has only been investigated previously by hydrogenation
experiments (Mackwell and Kohlstedt 1990; Kohlstedt and Mackwell 1998; Demouchy and
Mackwell 2006) (with one unpublished study on H–D exchange in forsterite cited by Ingrin
and Blanchard 2006). Two different rates of hydrogenation have been observed, where the faster
one only allows a partial equilibration of olivine (Kohlstedt and Mackwell 1998). The different
rates are related to different charge-balancing mechanisms and it was suggested that the faster
mechanism involves interdiffusion of electron holes and protons and the slower mechanisms
a coupled diffusion of metal vacancies and two protons. Metal vacancies would then be the
favoured site for the protons as suggested by Kohlstedt et al. (1996). In any case, for both
proposed diffusion mechanisms, it is unclear which one controls the rate of dehydrogenation, or
even if olivine will be dehydrogenated by the slow mechanism and whether once a certain level
is reached the dehydrogenation accelerates. In Demouchy et al. (2006), Peslier and Luhr (2006),
Denis et al. (2013), and Peslier et al. (2008, 2015) only the slow diffusion mechanism has been
used to infer ascent rates since using the faster diffusion rate would indicate unrealistically fast
rates. However, the two diffusion mechanisms have a different anisotropy and the measured
profiles of Demouchy et al. (2006) appear more consistent with the anisotropy of the faster
mechanism, whereby D// a > D// c, opposite to the slow mechanism (Fig. 8). Different mobilities of
different hydrogen defects were also documented by Peslier et al. (2015) in olivine crystals from
Hawaiian basalts. Here H associated with metal vacancies is apparently more mobile compared
to H associated with Ti (see also Padron-Navarta and O’Neill 2014). Therefore, the main issue
of this approach is the uncertain dehydrogenation rate in olivine, which strongly depends on the
various possible incorporation mechanisms and structural H sites. Clearly more experiments are
needed for better constraints on the derived ascent rates.
Ferriss et al. (2016) concluded that the common observation of hydrogen zonation in
mantle xenolith olivines but not in clinopyroxenes implies that hydrogen diffusion is much
faster in olivine than in pyroxene, which then requires the operation of the fastest diffusion
mechanism quantified in olivine and diffusivities in clinopyroxene at the lower end of the
measured range (see also section on Cpx below). Such high diffusivities in olivine strongly
suggest that water in mantle xenoliths has at least partially equilibrated with the host magma,
and that the diffusion profiles observed in mantle xenolith olivine reflect only the final stage
of ascent after water begins to exsolve.
PLAGIOCLASE
Since the fundamental experiments of Bowen (1913), plagioclase (Pl) feldspar has a long
tradition of being used as an archive of magmatic processes. Various zoning types for major
element (albite–anorthite) zoning have been classified (normal, reversed, oscillatory, patchy,
e.g., Streck 2008), which are related to different types of growth conditions and related
magmatic settings. The ability of plagioclase to conserve even nm-scaled major element
zoning (e.g., Ginibre et al. 2002a; Saunders et al. 2014) is due to the very sluggish coupled
inter-diffusion rates of NaSi and CaAl (e.g., Grove et al. 1984). Minor and trace element
zoning have been studied to infer open versus closed system processes as these elements
are thought to be less dependent on the intensive thermodynamic parameters of the system
(e.g., Blundy and Shimizu 1991; Singer et al. 1995; Davidson and Tepley 1997; Ginibre
et al. 2002b; Berlo et al. 2007; Ruprecht and Wörner 2007; Smith et al. 2009). However,
diffusivities of trace and minor elements like Li, K, Mg, and potentially Sr and Ba, are fast
enough that their respective growth zoning patterns could be disturbed, depending mainly on
the magma temperature and crystal residence times. An extreme case is Li diffusion, which
is extremely fast in plagioclase (Giletti and Shanahan 1997) and has been used to determine
the final ascent rate of an explosive eruption (Charlier et al. 2012).
The first approach to reproduce chemical zoning in plagioclase by diffusion modelling

was applied by Zellmer et al. (1999) for Sr. Based on the strong dependence of Sr
plagioclase–melt partition coefficients on the anorthite content (Blundy and Wood 1991) a
crucial point was already realized—if the albite (Ab)-anorthite (An) content of plagioclase
is zoned, and can be considered to be effectively immobile, the re-equilibration of more
mobile trace elements results in a non-uniform element distribution within the plagioclase.
This is related to the dependence of the chemical potential of trace element components in
plagioclase with the anorthite content, which can be described by non-ideal interactions
of individual trace elements with albite and anorthite (Blundy and Wood 1991; Dohmen
and Blundy 2014). However, as shown by Costa et al. (2003) this variation of the chemical
potential with anorthite has also consequences for the respective diffusion equation of the
trace element since the diffusive flux is proportional to the chemical potential gradient, not
the concentration gradient, in the more general case (e.g., Ganguly 2002). Consequently,
diffusion modelling for chemically zoned plagioclase with non-ideal interactions is more
complex and requires knowledge of how the chemical potential of the trace or minor element
(or the respective chemical component) varies with the anorthite content.
Diffusion coefficients and specifics for trace element diffusion in plagioclase
Costa et al. (2003) derived an equation for diffusion modeling of Mg in plagioclase, where
the variation of the chemical potential of the Mg-component with anorthite is considered:
∂CMg ∂  ∂CMg DMgCMg ∂X 

=  DMg ⋅ − ⋅ AMg ⋅ An  (6)
∂t ∂x  ∂x RT ∂x 
where DMg stands for the diffusion coefficient of Mg in plagioclase, which may be also a function
of the fraction of anorthite, XAn, and/or the activity of SiO2, aSiO2 (Faak et al. 2014 vs. Van Orman
et al. 2014, see discussion below). The parameter AMg is a thermodynamic factor reflecting the
non-ideal interaction of the Mg-component with Ab and An and was inferred from the empirical
fit of Bindeman et al.(1998) for plagioclase–melt partition coefficients, as follows:
−l
RT ⋅ ln K ppl(Mg 2+
)
= AMg ⋅ X An + BMg (7)
Equation (6) can be also generalized for other elements showing similar linear compositional
dependence of the partition coefficient as in Equation (7). In the derivation of Equation (6)
by Costa et al. (2003) it was assumed that Equation (7) reflects the explicit dependence of the
chemical potential of Mg on An. However, as discussed in Bindeman et al. (1998) the parameters
in Equation (7) were obtained from Drake and Weill’s (1975) experiments on basaltic melts
where the variation of the An content was related to a simultaneous change in T and liquid
composition. Thus, as demonstrated by thermodynamic modelling of Dohmen and Blundy
(2014), Equation (7) also includes the implicit dependence of An on T and liquid composition.
In the thermodynamic model of Dohmen and Blundy (2014) the chemical potential variation
with anorthite of various trace components were inferred from the application of the lattice
strain model of Blundy and Wood (1994) to experimental plagioclase–melt partitioning data
from isothermal series in the diopside–albite–anorthite system. The derivation of the diffusion
equation applying this thermodynamic model and the respective quasi steady state condition,
which describes the partial equilibrium of trace elements in plagioclase with the surrounding
melt (open system condition) or internal equilibration (closed system condition) is presented in
the Electronic Appendix, including a stable numerical scheme to solve this diffusion equation
(a Mathematica or C++ source code can be requested from the authors). Finally, the derived
diffusion equation (Eqn. A8) is shown to be effectively equivalent to Equation (6) since the
variation of RT ln (gi) with XAn is approximately linear.
The main implication of the thermodynamic model of Dohmen and Blundy (2014)
is that the thermodynamic parameter Ai differs from the values inferred from the relations
presented in Bindeman et al. (1998), as shown in Table 1. The latter values were used in
various diffusion modelling studies with Sr and Mg (e.g., Costa et al. 2003; Druitt et al. 2012;
Ruprecht and Cooper 2012; Cooper and Kent 2014) and are potentially incorrect according
to Dohmen and Blundy (2014) and Faak et al. (2013). Different values for the parameter Ai
have consequences for the diffusion profiles and the inferred time scales. In particular for Mg
the sign of Ai of Dohmen and Blundy (2014) and Faak et al.(2013) compared to those inferred
from Bindeman et al. (1998) is the opposite, which has implications both for the diffusion flux
and the “equilibrated profile”. In Figure 10 model results are shown for Mg and Sr diffusion in
a synthetic, normally zoned plagioclase crystal (albitic overgrowth on anorthite-rich core and
corresponding positive correlations of Mg and Sr). A particular advantage of using Sr is that it
is compatible in plagioclase. Consequently the variation in Sr with XAn resulting from growth
will have a positive slope, whereas diffusive re-equilibration will impart a negative slope (see
also Cooper and Kent 2014). According to the diffusion data of Cherniak and Watson (1994)
Sr diffusion should be detectable with ion probe measurements (5–10 mm spatial resolution)
after about 10 yr at 890 °C or a month at 1200 °C.
For Mg the distinction between growth zoning and zoning affected by diffusion is
more difficult. According to Dohmen and Blundy (2014) Mg has an A-parameter of positive
sign, opposite to those of other relevant divalent cations like Sr or Ba, and opposite to those
inferred by Bindeman et al. (1998). According to the lattice strain model adopted by Dohmen
and Blundy (2014) the reason for the positive sign is that, unlike Sr2+ and Ba2+, the ionic
radius of Mg2+ is smaller than the optimum ionic radius for divalent cations, r02 + , which
decreases from Ab to An. Therefore with increasing XAn, cations like Sr2+, with a larger ionic
radius than r02 + , become energetically less favourable for this site, whereas Mg2+ becomes
more favourable. These energetics quantified by the lattice strain model are reflected in
the A-parameters reported in Table 1. According to the positive AMg, at partial equilibrium
(or quasi steady state) of Mg it would show a positive, roughly linear, variation with XAn.
Unlike for Sr, the growth zoning of Mg in plagioclase depends on the co-precipitating mafic
minerals that control the Mg budget in the melt. In case of the cotectic growth with Pl of
mafic minerals like Ol, Opx, Cpx, the Mg in the melt tends to decrease and for normal zoning
of XAn, Mg should decrease as well. Thus, during growth a positive correlation of Mg with
XAn is obtained, potentially difficult to distinguish from diffusive equilibration, depending on
the exact Mg vs. XAn slope. However, in a number of studies of plagioclase phenocrysts from
a variety of volcanic rocks (Costa et al. 2003, 2010a; Moore et al. 2014) a negative correlation
of Mg with XAn was observed, which were reproduced assuming partial equilibrium and using
the negative AMg parameter of Bindeman et al. (1998) (e.g., see also Fig. 10). Alternatively, if
the positive AMg of Dohmen and Blundy (2014) and Faak et al. (2013) is correct, the observed
Table 1 Values for the thermodynamic parameter Ai (kJ/mol) representing the non-ideal interaction
of the trace element i with the major components Ab and An, as inferred from plagioclase–melt and
plagioclase–clinopyroxene partitioning data.
Reference AMg ASr ABa ALi AK ARb
Dohmen and Blundy (2014) 1200 °C 15.8 −17.4 −35.1 −1.7 −8.0 −15.9
Dohmen and Blundy (2014) 900 °C 13.7 −15.1 −30.5 −2.5 −8.8 −16.7
Bindeman et al. (1998) −26.1 −30.4 −55.0 −6.9 −25.5 −40.0
Faak et al. (2013) 16.9
a b initial
0.8 0.01
0.1
0.7 1
XAn 10
0.6 100
200 200 300
rim Distance [µm] core equilibrium
Sr [µg/g]
Sr [µg/g]
eq Bindeman
100 100
rim core
c d initial
0.001
2500 2500 0.01
0.1
1
10
Mg [µg/g]
Mg [µg/g]
2000 2000 equilibrium
eq Bindeman
1500 1500
1000
1000
rim core
50 100 150 200 250 300 0.60 0.65 0.70 0.75 0.80
Distance [µm] XAn
Figure 10. Results of model calculations for Sr (a) and Mg (c) diffusion profiles and respective correla-
tions of their concentrations with XAn in plagioclase, (b) and (d), respectively. For the initial step growth
zoning of XAn (see inset in panel (a)) and trace elements (synthetic crystal) a positive correlation of both,
Sr and Mg, with XAn was assumed. Temperature was assumed to be 1200 °C. The numbers in the legend
denote the time in years. For DSr and DMg we used the experimental data of Cherniak and Watson (1994)
and Van Orman et al. (2014), respectively. Note the uphill diffusion flux for Sr at the interface between
low and high anorthite, which is the result of the second term in the diffusion equation (Eqn. 6). It should
be emphasized that such concentration gradients have been not observed so far in any natural sample,
which would be an independent test of the diffusion model and would also allow to estimate the param-
eter ASr or AMg. Therefore, the diffusion model should be experimentally tested.
Mg concentration patterns would lie far from equilibrium and hence should reflect growth
zoning, whereby Mg in plagioclase increases as An decreases. Growth zoning with such a
negative Mg vs. XAn slope would require plagioclase-dominated crystallization, such that Mg
behaves as an incompatible element in the crystallising magma. This need not mean that no
mafic minerals were present or growing in the system but simply that the larger plagioclase
phenocrysts that are typically analysed may reflect a stage of magma differentiation where
mostly plagioclase crystallizes (or mafic minerals are resorbed). An alternative resolution to
this apparent discrepancy between experimental and natural observations is if Mg occupies
the T-site, rather than the M-site, in natural plagioclases. The lattice strain model described
above pertains to substitution of Mg2+ for Ca2+ or Na1+ on the large, approximately 10-fold
co-ordinated M-site. It has been proposed (e.g., Longhi 1976; Peters et al.1995; Miller et al.
2006) that in Ca-rich plagioclases Mg may also be incorporated onto the tetrahedral site, via
the substituent molecule CaMgSi3O8. In that case it is the size mismatch between Mg2+ ions
and the T-site that will control partitioning and the relationship with XAn may differ from that
described above for M-site incorporation. Further work on the Mg environment in natural and
synthetic plagioclases is required to resolve this matter.
For other incompatible trace elements, such as Ba, Li, K, and Rb, both growth and
diffusion will impart negative correlations with XAn, making the effects of growth and diffusion
difficult to deconvolve. Thus, Sr and potentially Mg (in case of reversed growth zoning) are the
elements for which, within reasonable timescales, an effect of diffusive re-equilibration should
be clearly observed, without knowing in detail the initial growth zoning.
According to the experimental diffusion data of Cherniak and Watson (1994) for DSrPl and
Pl
Van Orman et al. (2014) for DMg , the concentration profiles of Sr and Mg change at different
time scales (e.g., Fig. 10), whereby Mg re-equilibrates about a factor 30 faster than Sr for
XAn between 0.6 and 0.8. Both diffusion coefficients depend strongly on the anorthite content
(Fig. 11). In contrast, Faak et al. (2013) did not observe a dependence of the diffusion rate of
Mg in plagioclase on XAn (for an investigated range of XAn between 0.5 and 0.8) but found a
Pl
strong influence of DMg on the chemical potential of silica in the system. Both studies, Faak
et al. (2013) and Van Orman et al. (2014), are consistent with earlier data of LaTourrette and
Wasserburg (1998) measured for natural single crystals of anorthite (XAn = 0.95) at 0.1 MPa
Pl
between temperatures of 1200 and 1400 °C. However, based on Faak et al. (2013), DMg is
Pl
rather similar to DSr for plagioclase with XAn around 0.4. Whether Mg diffuses on a similar
timescale to Sr may be inferred from investigation of natural plagioclases.
Determination of magma residence times using Mg or Sr diffusion
Magma residence times or, to be more precise, timescales between mixing (or mingling)
events using Mg and Sr diffusion modelling in plagioclase were determined by Zellmer et
al. (1999), Tepley et al. (2000), Costa et al. (2003, 2010a), Druitt et al. (2012), Ruprecht and
Cooper (2012), Cooper and Kent (2014), and Singer et al. (2016). In these studies typically
either the Mg or the Sr zoning was modelled to extract timescales. A general strategy to
reconstruct the growth zoning of trace elements was to use the anorthite growth zoning and
find a correlation of the Sr or Mg contents with XAn during growth. For example in Cooper
and Kent (2014) for plagioclases of Mt. Hood the initial Sr profiles were estimated by two
separate methods: (1) using the observed general correlation between Sr and anorthite and
(2) using the results of Rhyolite-MELTS modelling of the liquid line of descent to calculate
Sr contents of plagioclase from the plagioclase–melt partition coefficient. From both methods
only a minor difference of the slopes for Sr vs. XAn was obtained. This similarity suggests
here that very little equilibration of Sr by diffusion occurred. As illustrated in Figure 10, the
residence time was then inferred from the deviation of the measured Sr vs. XAn correlation
to the “partial” chemical equilibrium correlation, which is negative. Cooper and Kent (2014)
proposed strong temperature fluctuations in the magmatic system between 750–900 °C,
making Sr effectively immobile in plagioclase throughout most of the residence time.
-14
LT F₆₀ a(SiO₂)=1
W
-15 ₉₅ (
LT W 001)
log (DMg [m s ])
2 -1
₉₅ (
0
10
-16 )
VO
Pl
VO ₂₃
-17 ₄₃
F₆₀ a(SiO₂)=0.6 VO
₆₇
-18 F₆₀ a(SiO₂)=Ol+Opx
-19
VO
₉₃
6 7 8 9
4 4
10 /T [10 /K]
Figure 11. Arrhenius plot of measured Mg diffusion coefficients from the literature (LTW: LaTourrette
and Wasserburg 1998; F: Faak et al. 2013; VO: Van Orman et al. 2014) and their effect on XAn and aSiO2
(XAn and aSiO2 are given as subscripts).
Thus only diffusive effects close to the peak temperature were visible. By this model, the
apparent discrepancies between the residence times inferred from the short-lived U-series
isotopes and those of diffusion modelling was explained. Clearly the effect of a fluctuating
temperature field on diffusion should be borne in mind when interpreting timescales.
Determination of cooling rates using Mg diffusion
A widespread mineral association in mafic and intermediate igneous rocks is plagioclase–
clinopyroxene. Faak et al. (2014) developed an Mg-in-plagioclase geospeedometer, based on
the diffusive exchange of Mg between plagioclase and clinopyroxene during cooling. The
Mg-in-plagioclase geospeedometer builds upon the diffusion model proposed by Costa et al.
(2003) to account for the effect of zoned XAn (Eqn. 6).
According to Faak et al. (2013) the partition coefficient of Mg between plagioclase and
clinopyroxene can be parameterised, as follows:
−9219 16913  Jmol 

-1
ln K Pl - Cpx
=+
1.6 + XAn + ln aSiO2 (8)
p (Mg)
T RT
where T is in Kelvin. The partition coefficient K pPl(Mg)
- Cpx
decreases with temperature, i.e., during
cooling an exchange of Mg between plagioclase and clinopyroxene rims is required to maintain
equilibrium, which can be described by one of the following net transfer reactions (dependent
on the assumed site occupancy of Mg in plagioclase):
Mg on T-site: CaSiO3 (Cpx) + MgSiO3 (Cpx) + SiO2 =
CaMgSi3O8 (Pl) (9)
Mg on M-site: MgSiO3 (Cpx) + SiO2 + Al 2O3 =

MgAl 2Si 2O8 (Pl) (10)
Consequently, this thermometer is not based on a simple exchange reaction and, in addition
to temperature, also depends on the silica activity (or more precisely the chemical potential of
silica). The relevance of the chemical environment is common for trace element thermometry
(e.g., effect of silica or TiO2 activity for the Ti-in-zircon or Zr-in-rutile thermometer, e.g., Ferry
and Watson 2007). Since Mg is a major element in clinopyroxene, the Mg-content (the MgSiO3
component) is effectively buffered and does not change due to the Mg flux from plagioclase
to clinopyroxene during cooling (cf. Ca exchange between Ol and Cpx discussed above).
Thus, for a known concentration of Mg in clinopyroxene, an apparent temperature can be
calculated at each point along a measured Mg-profile in plagioclase. Based on Reaction (9),
the CaMgSi3O8 component dissolves into clinopyroxene and produces some extra silica that
in many cases could in turn be used to produce minor amounts of MgSiO3 (En) from Mg2SiO4
(Fo) according to reaction Fo + SiO2 = En. Therefore, local equilibrium between plagioclase
and clinopyroxene interfaces is not controlled only by lattice diffusion within these minerals
but also involves processes such as grain boundary diffusion (e.g., transport of silica), which
might be the rate limiting process (e.g., Dohmen and Chakraborty 2003).
The diffusion model of Faak et al. (2014) is based on the assumption that local equilibrium
is achieved instantaneously at the interface between both phases, and hence the concentration
of Mg at the interface of the plagioclase crystal can be calculated from K pPl(Mg) - Cpx
at any T.
According to Equations 6 and 8, the evolution of the resulting concentration profile depends
on the sub-solidus cooling history, as well as on the XAn-profile (see Fig. 12). For that reason,
the standard Dodson model (or any recent extension, e.g, Ganguly and Tirone 1999) cannot
be used to obtain the closure profile and hence cooling rates (e.g., as for Ca in olivine). Hence
the Mg profiles were calculated numerically by solving the appropriate diffusion equation
(Eqn. 6). The temperature interval over which cooling rates can be recorded ranges from Tc
at the crystal core (Tccore) to Tc at the crystal rim (Tcrim). The exact values for Tccore and Tcrim
depend on the cooling rate and on the size of the crystals (for plutonic rocks from the lower
oceanic crust, this interval is between 1100–750 °C; Faak et al. 2015). However, modelling
the redistribution of Mg between plagioclase and clinopyroxene during cooling requires an
estimate of the initial conditions (Tstart) for the system. Faak et al. (2014) show that for a
system with peak temperatures above 1100 °C (i.e., most mafic igneous systems), a grain size
of 1 mm and a cooling rate of 0.1 °Cyr−1, the memory of the initial conditions would be erased
by diffusion (thus one assumption of the Dodson model would be fulfilled). In these cases, a
reasonable approach is to calculate an equilibrium concentration profile of Mg in plagioclase
based on equilibrium partitioning with clinopyroxene as starting profile.
Two crucial parameters in the Mg-in-plagioclase geospeedometer, the diffusion
Pl
coefficient DMg and the partition coefficient K pPl(Mg)
- Cpx
, depend on the silica activity aSiO2 of the
system (Faak et al. 2013). Therefore, the application of this approach requires knowledge of
the silica activity (as a function of temperature). In many systems of interest, aSiO2is either
buffered by a given mineral assemblage or can at least be constrained to lie between two
buffer reactions (e.g., Carmichael et al. 1970). For example, Faak et al. (2013, 2014) provide
a method to constrain silica activity for the case of a system buffered either by pure SiO2
(e.g., quartz) or by coexisting olivine–orthopyroxene.
The Mg-in-plagioclase geospeedometer has been tested with regard to the robustness
of the calculated cooling rates and the precision is found to be better than half an order of
magnitude (Faak et al. 2014, 2015). Faak et al. (2014) investigated the effect of different
a b dT/dt c dT/dt
0.8 0.20 0.1°Cyr-1 0.001°Cyr-1
MgO [wt%]
0.7
0.15
XAn
0.6
0.10
0.5
0.05
0.4
rim core rim
d e f dT/dt
0.8
dT/dt
0.20 0.1°Cyr-1 0.001°Cyr-1
MgO [wt%]
0.7
0.15
XAn
0.6
0.5 0.10
0.4 0.05
rim core rim
g h dT/dt i dT/dt
0.8
0.20 0.1°Cyr-1 0.001°Cyr-1
MgO [wt%]
0.7
0.15
XAn
0.6
0.10
0.5
0.4 0.05
rim core rim
0 500 1000 1500 0 500 1000 1500 0 500 1000 1500
Distance [µm] Distance [µm] Distance [µm]
T0=1200°C T1=1100°C T2=1000°C T3=800°C T4=600°C diffusion profile partitioning profile
Figure 12. Schematic evolution of Mg-profiles in plagioclase during cooling (in contact with clinopyroxene)
for three different assumed plagioclase crystals (length = 1500 mm, plane sheet geometry). P1 (a, b, c) has a flat
XAn-profile, P2 (d, e, f) and P3 (g, h, i) have stepped XAn-profiles that are reversely zoned. Panels (b), (e) and
(h) show the calculated Mg-profiles at different temperatures for a linear cooling rate of 0.001 °Cyr−1. Panels
(c), (f) and (i) shows the calculated Mg-profiles for the same temperatures as above, but for faster linear cool-
ing of 0.1 °Cyr−1. Dashed lines show the calculated theoretical partitioning profiles for Mg in plagioclase in
equilibrium with clinopyroxene at different temperatures. (Figure modified after Faak et al. 2014).
diffusion coefficients (Faak et al. 2013 vs. Van Orman et al. 2014) on the cooling rate that
would be obtained for plagioclase crystals with normally and reversely zoned XAn profiles.
For their specific example, they show that the cooling rate obtained for a normally zoned
Pl
plagioclase crystal will be by a factor of 3 faster when using an XAn-dependent form of DMg
Pl
compared to a DMg that does not depend on XAn. For a reversely zoned plagioclase crystal, the
Pl
obtained cooling rate will be a factor of 4 slower when an XAn-dependent DMg is used.
Additionally, Faak and Gillis (2016), in a combined approach, applied the Mg-in-
plagioclase geospeedometer and the Ca-in-olivine geospeedometer (Coogan et al. 2002,
2007) on the same samples from the lower oceanic crust (Fig. 7). They demonstrate that these
independent geospeedometers yield similar cooling rates, implying that the absolute values
obtained from the two methods are robust.
PYROXENES
Pyroxenes have been used to constrain time scales of processes in silicic systems that lack
olivine. The observed diffusion profile lengths of Fe-Mg in natural orthopyroxene are often
much shorter (a few mm or less) than in olivine. This observation appears to be consistent
with the measured diffusion rates in Opx and Cpx that are generally slower than in olivine
(e.g, about 1–2 log units for Fe–Mg: Müller et al. 2013; Dohmen et al. 2016a), considering
similar timescales for the respective magmatic processes. Therefore BSE imaging (for Fe–
Mg, e.g., Morgan et al. 2004, Saunders et al. 2012a) or other higher resolution methods
have been used (TOF-SIMS: Saunders et al. 2012b; NanoSIMS: Saunders et al. 2014) to
resolve in detail the chemical zoning. Fast processes can be determined using Li diffusion in
Cpx. For example, concentration profiles of Li in Cpx from martian meteorites (nakhlites)
were modelled to determine cooling rates of the respective magmatic system (Beck et al.
2006). Here, in situ zoning of the Li isotopes was measured, and found to be consistent with
diffusive fractionation, as was also demonstrated in other cases (e.g., Gallagher and Elliott
2009). For a review of diffusion data in pyroxenes we refer to Cherniak and Dimanov (2010)
and for more recent data to Müller et al. (2013) for Cpx and Dohmen et al. (2016a) for Opx.
Here only a short overview and an update of diffusion data are given.
Diffusion coefficients
Orthopyroxene. Until very recently (Dohmen et al. 2016a) no directly measured diffusion
Opx
data for Fe–Mg interdiffusion in Opx were available. Therefore in the past, DFe-Mg for diffusion
chronometry was inferred from Mg tracer diffusion measured by Schwandt et al. (1998) in Opx
with 88% of enstatite component (En88) at fO2 of the iron–wuestite buffer, IW, (e.g., Saunders et
al. 2012a) or from theoretical correlations between Fe–Mg diffusion and Fe–Mg order–disorder
rates on the M1 and M2 sites (Ganguly and Tazzoli 1994; Allan et al. 2013). Based on the
theoretical work of Ganguly and Tazzoli (1994) and the experimental data for order–disorder
Opx
data of Stimpfl et al. (2005), Allan et al. (2013) parameterised the dependence of DFe-Mg on
T, fO2, and XFe = Fe / (Fe + Mg). However, the first direct experimental measurements of Fe–Mg
interdiffusion in Opx by Dohmen et al. (2016a) demonstrate that the dependence on fO2 is much
smaller than estimated by Allan et al. (2013). In addition, the available data on order–disorder
rates of Fe–Mg in Opx were discussed in Dohmen et al. (2016a) and it was concluded that at
Opx
constant P, T, and fO2, DFe-Mg varies less than 0.5 log units for 0.1 < XFe < 0.5. As a consequence,
the Mg tracer diffusion coefficients of Schwandt et al. (1998) measured for Opx close to enstatite
Opx
composition and at fO2 = IW are very similar to DFe-Mg with XFe < 0.5 at fO2 close to the Ni–NiO
Opx
buffer, NNO, (Fig. 13). According to Dohmen et al. (2016a), DFe-Mg parallel to the c-axis is a
factor of 3.5 faster than parallel to the a-axis. Therefore for accurate determination of time scales
the crystallographic orientation of the profile should be established.
With the exception of H (Li has so far not been measured for Opx), other elements
(Cr, REE, Pb, and Ti) diffuse significantly more slowly than Fe–Mg, by at least an order
of magnitude. Recent data measured after the review of Cherniak and Dimanov (2010) are
those of Sano et al. (2011) for Nd and of Cherniak and Liang (2012) for Ti. As for olivine the
behaviour of H diffusion in Opx is complex and depends on the relevant incorporation and
diffusion mechanism (e.g., Stalder and Skogby 2003, 2007; Stalder and Behrens 2006).
Clinopyroxene. DFe-Mg
Cpx
was measured by Müller et al. (2013) between 800–1200 °C and is
slower than in Opx with a similar XFe (Müller et al. 2013), but has the same activation energy
Cpx
within experimental error, 320 kJ/mol and 308 kJ/mol, respectively. The DFe-Mg of Müller et al.
Cpx
(2013) are consistent with DFe-Mg-Mn measured by Dimanov and Wiedenbeck (2006), although
in the latter work a stronger effect of fO2 could be identified (for details, see discussion in
Müller et al. 2013). Ca tracer diffusion rates or inter-diffusion processes involving Ca are at
least one log unit slower than Fe–Mg interdiffusion in Cpx.
Faster diffusing elements are H and Li. As in olivine and orthopyroxene, their respective
diffusion rates are dependent on more than just temperature. H diffusion experiments by, for
example, Sundvall et al. (2009) and Ferriss et al. (2016) have shown that diffusion rates of
H vary by up to five orders of magnitude at a given temperature where the total Fe content
(and respective effects on the Fe3+ content) and tetrahedral Al content in Cpx appear to have a
dominant control (Ferriss et al. 2016).
T [°C]
[2
]/
/[0 //[100]
01
]- [1] //[010]
7
//[001]
[2
]/ [4]
/[1
log (DFe-Mg [m /s])
00 NN
]- O
2
7 [2
[2 ]/
]/ /[0
/[0 01 [3]
1100 1000 900 01 ]N IW
]I NO
-18 [2 W
]/
/[0
01
-19 ]-
[4] 7
-11
[2
]/
-20 /[0
01
]-
11
-21
4 4
10 /T [10 /K]
Figure 13. Measured and estimated DFe–Mg from the literature in an Arrhenius plot: [1] Schwandt et al.
(1998); [2] Dohmen et al. (2016); [3] Ganguly and Tazzoli (1994); [4] Allan et al. (2013). Different symbols
are used for Mg tracer diffusion data from different crystallographic directions of Schwandt et al. (1998),
as indicated in the figure. The diffusional anisotropy of DMg is not well resolved with this dataset measured
at fO2 = IW, but can be well described by the Arrhenius relation of Dohmen et al. (2016), which has only a
weak dependence on fO2. Also shown are DFe–Mg for En91 as calculated from the expressions of Ganguly and
Tazzoli (1994) at fO2 = IW [2] and from Allan et al. (2013) at fO2 = NNO [3]. The calculated values of DFe–Mg
at a constant oxygen fugacity are shown in the inset of this figure (log fO2 [Pa] = –11 according to Allan et
al. (2013) and according to Dohmen et al. (2016) at log fO2 [Pa] = –11 and log fO2 [Pa] = –7). The stronger fO2
dependence of diffusivity postulated by Allan et al. (2013) produces a low activation energy and significant
deviation from measured values at various temperatures. Figure modified from Dohmen et al. (2016).
The first measurements of Li diffusion in clinopyroxene found diffusion profiles with

a simple Arrhenius relationship for fO2 close to the wüstite–magnetite buffer, WM, (Coogan
et al. 2005). In a more recent study by Richter et al. (2014), where fO2 was varied between
IW and NNO, a more complex behaviour of Li diffusion was found, analogous to that in
olivine (Dohmen et al. 2010), where a model with two Li species (Li on interstitial and on
metal sites) was required to simulate the unusual diffusion profiles. The two mechanisms
may operate simultaneously to produce variable diffusion rates (by more than two log units)
at a given temperature depending on the Li gradient as well as fO2 (Richter et al. 2014, see
also Tomascak et al. 2016). These conditions also impact on the observed Li isotope profiles
and total extent of diffusive fractionation between 6 Li and 7 Li (Richter et al. 2014). Richter
et al. (2014) modelled two Li isotope profiles from the literature and showed that only
the single species (slow) mechanism operated. Furthermore they could distinguish between
different boundary conditions—a sudden change (fixed boundary condition) or a continuous
change at the rim of the crystal. Depending on the boundary condition, the diffusive flux into
the crystal changes and hence the total extent of diffusive isotope fractionation is affected,
which is an effect of mass balance. The stable isotopes therefore also leave a fingerprint of
the operating diffusion mechanisms and, if measured in natural samples, help to identify the
correct boundary condition and diffusion rate.
Determination of magma residence times using Fe–Mg diffusion in Opx and Cpx
In BSE images, pyroxenes in silicic volcanic rocks typically show a diversity of zoning
styles, with normal, reversed, multiple, or patchy zoning. However, it is quite common that a
final, outermost rim with a thickness of 10–100 µm is formed at some point before the magma
erupts (e.g., Fig. 14). The compositional jump at the interface between the outermost rim
and the interior of the crystal is typically very sharp and cannot be resolved using electron
microprobe analysis without convolution effects (Ganguly et al. 1988; Morgan et al. 2004).
Therefore, Morgan et al. (2004) inferred variations in the Fe content using the grey scale of BSE
images of clinopyroxene, since the contribution to the changes in the grey scale by variations
of other elements, Al or Si, are minor. For Cpx it must also be assumed that the variation
of Ca was small enough to not affect the grey scale. The linear correlation of the grey scale
with the Mg# = Mg / (Fe + Mg) in Opx was demonstrated by Allan et al. (2013). BSE imaging
allows for a spatial resolution down to 100 nm or less (e.g., Saunders et al. 2012b) and is a very
time efficient way to map Fe in pyroxene (or olivine) crystals. Using image analysis software,
Morgan et al. (2004) transferred the grey scale of a BSE image into pixelated image data and
residence times at magmatic temperatures were obtained from the following equation
n   x 
C( x ,t ) = = 1 − 0.5erfc   (11)
2(n + 1)   2 Dt  
at which C represents the compositional contrast on a scale of 0–0.5 between the junction
of the observed maximum extent of diffusion at the half width x (see Fig. 14 and Appendix
A in Morgan et al. 2004 for details), t is the residence time, n is the number of different grey
shades in the BSE image in the zone effected by diffusion, and D is the diffusion coefficient
of Fe-Mg in clinopyroxene. For the application of Equation (11), D needs to be approximately
constant, an initial step profile is assumed, and the diffusion profiles must be short enough to
assume an infinite one-dimensional medium as geometry (see Crank 1975 for explicit initial
and boundary conditions; e.g., Fig. 14). A main advantage of the short profiles measured along
a traverse at the half width of the crystal is that sectioning effects are less likely and a one-
dimensional geometry for the diffusion model is often justified (Krimer and Costa 2016). A risk
is that the time scale is still overestimated if the Fe–Mg zoning was not measurably modified
following growth, meaning the diffusion distance is smaller than the spatial resolution of the
BSE imaging. The measurement of more slowly diffusing elements, e.g., Al and Ca, provides
(a) (b)
C1
0.5 (C1-C0)
C0
x
C1
y
0.75
50 μm 0.5 (C1-C0)
C0
Figure 14. (a) BSE image of a chemically zoned orthopyroxene from Mt St. Helens (crystal SH127_
KS4, Saunders et al. 2012a). Note the geometrically well-defined rim with a relatively sharp interface (b)
Illustration of the method of Morgan et al. (2004) to invert the grey scale zoning of the BSE image to a
timescale (see Eqn. 11).
clues to the original growth zoning (Morgan et al. 2004). This has been demonstrated by
Saunders et al. (2012a,b, 2014) for orthopyroxene (and Pl) by using TOF-SIMS and Nano-
SIMS measurements that allowed for sub-mm resolution of slow-diffusing elements.
The appropriate diffusion coefficient depends on the magmatic temperature T, which
Morgan et al. (2004) determined based on the composition of melt inclusions in clinopyroxene
using the geothermometer of Cioni et al. (1999). Using the diffusion data of Dimanov and
Cpx
Sautter (2000) for DFe-Mg , Morgan et al. (2004) determined residence times in multiple
clinopyroxene crystals (following growth of the final rim) from the 1944 eruption of Vesuvius
and obtained a range from 0.4 to 9 yr with a distribution skewed to young ages. They used the
shape of the obtained cumulative residence curve for comparison with modelled populations
with different magma chamber volumes. Their analysis reveals that a constant magma chamber
volume is inadequate for Vesuvius and suggest instead a volume of 8.0 × 107 m3 that expanded
to 1.15–1.5 × 108 m3 prior to the 1944 eruption.
Following the basic approach of Morgan et al. (2004) five recent studies have explored the
potential of Fe–Mg zoning in Opx to quantify magmatic time scales (Saunders et al. 2012a, Allan
et al. 2013; Chamberlain et al. 2014; Kilgour et al. 2014). Here we focus on the methodological
developments for diffusion chronometry and the time scales inferred in these studies. As
illustrated already for olivine by Kahl et al. (2011) a main strength of diffusion chronometry is
the direct comparison of timescales of magmatic processes at depth with monitoring signals at
Earth’s surface. This potentially allows for the interpretation of certain surface signals (seismic,
degassing, or ground deformation) in terms of a specific causative magmatic process at depth
(see also Albert et al. 2016). For example from the zoning at the outer rims of Opx crystals from
Mount St. Helens (e.g., Fig. 14a) Saunders et al. (2012a) established the time elapsed between
the final rim growth and the eruption, which varied between days to months. The value they
Opx
used for DFe-Mg was 4.72 × 10−21 m2/s, as calculated at 884 °C (estimated from Fe-Ti oxides,
Blundy et al. 2008) from Mg tracer diffusion of Schwandt et al. (1998). The recent experimental
Opx
data for DFe-Mg parallel to the c-axis measured for Opx with En91 of Dohmen et al. (2016a)
predict at the same temperature for fO2 = NNO + 2 a value of DFe-Mg Opx
= 7.39 × 10−21 m2/s, which
justifies the use of Schwandt et al. (1998). The Opx in Saunders et al. (2012) were more Fe-
rich, with a range of XFe = 0.24 – 0.46. The higher Fe content might enhance the Fe–Mg diffusion
rates by a factor 2–3 compared to Opx with XFe = 0.09 (see discussion in Dohmen et al. 2016a
Opx
on the compositional dependence of DFe-Mg ), but for profile orientations parallel to the a-axis,
the diffusion rate is smaller by a factor 3.5, compensating the effect of composition. From the
types of zoning and the trace element systematics, Saunders et al. (2012a) concluded that the
final rim growth was formed by injection of hotter magma in the shallow magma chamber, and
hence the date of magma intrusion could be determined from diffusion modelling. It was shown
that this timing correlated with recorded episodes of seismicity, suggesting that swarms of small
earthquakes heralded the ascent of new magma into the sub-volcanic reservoir.
In Allan et al. (2013) effects of growth zoning were identified in Opx crystals from
the 25.4 ka, 530 km3 Oruanui eruption (Taupo volcano, New Zealand) for profile directions
parallel to the c-axis, and hence only diffusion profiles roughly parallel to the a- or b-axis
were considered. The overgrown rim was typically thicker for directions parallel to the c-axis
compared to other directions and hence the growth rate was larger. This might have been
responsible for a diffusion boundary layer in the melt and a respective evolution in the rim
composition during growth. The measured profile shapes for Fe could not be fitted by a simple
diffusion model that does not consider simultaneous growth. Four different thermometers
were applied, which gave a consistent temperature of 770 ± 30 °C. To relate the compositional
change and textures of Opx crystals to physical conditions, the trace element composition
of amphibole (Amp) crystals (mainly Zn and Mn contents) were correlated to pressure
calculations using the thermobarometric formulations of Ridolfi et al. (2010). Allan et al.
Opx
(2013) formulated their own parameterized DFe-Mg based on order–disorder rates and the model
of Ganguly and Tazzoli (1994). For the relevant conditions, T = 770 °C and fO2 = NNO − 0.2,
Opx
DFe-Mg = 1.28 × 10−22 m2/s according to Dohmen et al. (2016a), which is more than two log
Opx
units smaller than the DFe-Mg = 3.97 × 10−20 m2/s predicted from the relation in Allan et al. (2013)
for Opx with Mg# = 0.5. Consequently, the magma residence times of Allan et al. (2013),
Opx
varying between tens of years up to 2000 yr, would be longer (1–200 ka) if DFe-Mg of Dohmen
et al. (2016a) had been used. Time scales of 200 ka appear to be very long and inconsistent
with other estimates using U–Th model ages with zircon. One explanation could be that for
Opx
T  900 °C extrapolation of the expression of Dohmen et al. (2016a) underestimates DFe-Mg
because of a change in the diffusion mechanism, as discussed in Dohmen et al. (2016a). From
the kinetic data of the Fe–Mg order–disorder for T < 900 °C a smaller activation energy of Fe–
Mg interdiffusion could be inferred as well as a lack of any effect of fO2, which would imply an
about one log unit faster diffusion rate at the relevant temperature of 770 °C.
Opx
The strong discrepancy between estimated DFe-Mg using Allan et al. (2013) and measured
Opx
DFe-Mg is most likely the reason for some of the inconsistencies obtained by Chamberlain
et al. (2014) who compared timescales obtained from four different diffusion chronometers
for multiple samples of Bishop Tuff, California: Fe–Mg in Opx, Ti in Qtz, and Sr and Ba
Opx
in sanidine. Here again the relation of Allan et al. (2013) was used to estimate DFe-Mg but
compared to the Ti in Qtz chronometer, the residence times were a factor of 10 shorter.
Temperatures were derived form two-feldspar thermometry and span a range of 753–813 °C
for the different samples. Based on the previous discussion and Dohmen et al. (2016a), it is
Opx
likely that Chamberlain et al. (2014) used an estimate for DFe-Mg that is too large by at least a
Opx
factor of 10. Therefore, considering the new DFe-Mg data, timescales inferred from Fe–Mg in
Opx and Ti in Qtz might be reconciled. Timescales inferred from Sr and Ba zoning in sanidine
are much longer, which according to Chamberlain et al. (2014) are probably related to only a
minor diffusive modification of the Sr and Ba growth zoning.
Singer et al. (2016) determined time scales of magma recharge for the 1912 Novarupta-
Katmai eruption from diffusion modelling of Mg in Opx and Pl. In this case for Pl only rough
estimates for the timescale could be made, but these are in general on the order of hundreds
of years. From Fe–Mg zoning in Opx significantly shorter time scales (weeks) were inferred.
SYNOPSIS/PERSPECTIVES
The great potential of reconstructing chemical and isotope zoning by diffusion modelling
is that it not only allows for the determination of timescales (duration or rate) of a specific
magmatic process, but also can be used in combination with petrological and observational
(e.g., volcano monitoring) constraints to provide an integrated picture of a magmatic system.
However, the methodology relies on robust estimates of magmatic temperatures, measurement
of the concentration profiles without convolution effects, appropriate diffusion coefficients, a
knowledge of chemical potential (in order to establish the equilibrium profile), and appropriate
choice of initial and boundary conditions. Without careful appraisal of these factors, a
timescale can always be obtained, but its accuracy is poor and its significance dubious. As
illustrated above for a number of case studies, different strategies have been developed to
improve the accuracy of the various input parameters of the forward modelling procedure.
In addition to the quality of the individual diffusion profile fits (standard deviations), which
is a first indicator of a diffusion controlled zoning, for non-cubic minerals the consistency
of the timescales obtained from different elements measured along different crystallographic
directions in the same crystal is one way to gauge the quality of the model and the inferred
time scale. This of course requires a good experimental calibration of the diffusion anisotropy.
In addition, measurement and fitting of chemical zoning in a larger number of crystals allows
to identify sectioning effects (e.g., Pearce 1984), and whether 2D or 3D modelling is required
to improve the accuracy of the timescale (Costa and Chakraborty 2004). For example, for
Fe–Mg diffusion modelling with olivine it was shown numerically by Shea et al. (2015b)
that with the appropriate choice of crystals and fitting of 20 concentration profiles with a
1D model an accuracy of 5% for the time scale can be achieved (ignoring the uncertainty of
the experimentally determined diffusion coefficient and input parameters such as T, fO2, etc.).
However, if the core composition remained unchanged by diffusion, and with the appropriate
choice of the profile direction, sectioning effects are in general less likely.
In-situ analysis of stable isotopes for the relevant elements has been shown to provide
strong constraints on the origin of the chemical zoning (growth/dissolution vs. diffusion), the
initial zoning, as well as the boundary condition (e.g., Oeser et al. 2015). The simultaneous
fitting of the chemical and isotope zoning reduces greatly the degrees of freedom for the
boundary condition, which is often coupled to the detailed thermal history. Consequently,
the thermal history can be reconstructed in greater detail compared to the case when only
chemical zoning is fitted (Sio and Dauphas 2016). This technique is particularly useful for
interdiffusion processes and related correlations of isotope anomalies for the interdiffusing
elements. Experimental calibrations of the respective isotope effects in the minerals of interest
are needed to obtain even stronger constraints. Currently, based on the spatial resolution of the
available techniques and constraints related to the counting statistics, this approach requires
long diffusion profiles on the order of 100 mm or more, which is rarely the case in natural
samples and hard to achieve in experiments. But even bulk analysis of olivine fragments
provides some hints to the nature of the chemical zoning as shown by Teng et al. (2011). In
the near future it is likely that the spatial resolution of these isotope profiling techniques will
improve and even shorter diffusion profiles or those of minor elements could be analysed.
Diffusion coefficients are the most fundamental input parameter for diffusion chronometry
and, as discussed for Fe–Mg interdiffusion in orthopyroxene or Mg diffusion in plagioclase,
controversies persist for important diffusion data. Only detailed experimental studies with
the focus of identifying the diffusion mechanism and factors controlling the relevant point
defect concentrations can provide robust and appropriate diffusion data sets for natural
systems. Conflicting diffusion data sets that still exist, for example REE diffusion in olivine,
clinopyroxene and garnet, may be explained by different diffusion mechanisms operating in the
various types of experiments (e.g., Dohmen et al. 2016b). Two types of diffusion mechanisms
with drastically different rates (3–4 orders of magnitude) have been identified for Li and H
diffusion in olivine or pyroxene. The relevant diffusion rate can depend on the detailed trace
element chemistry of the mineral and the chemical environment (boundary condition) of the
crystal. Thus, in certain cases a simple Arrhenius type of equation to describe the respective
D is not appropriate, and a more complex parameterization is required that considers in detail
the effect of oxygen fugacity or trace elements present as heterovalent substitutions on the
point defect chemistry of the mineral and hence its diffusion properties (e.g., Dohmen and
Chakraborty 2007). In the case of an element occupying two lattice sites, the interaction of the
different diffusing species and point defects (e.g., Dohmen et al. 2010; Richter et al. 2014) need
to be considered in the governing diffusion equations. Again, measurement of stable isotopes
provides additional constraints on the operating diffusion mechanisms (Richter et al. 2014).
In addition to Ol, Opx, Cpx, and Pl, chemical zoning in quartz (Qtz), alkali feldspar (Afs),
spinel (Sp) and melts/glasses have been used to estimate timescales of magmatic processes. The
general approaches used in these studies do not differ fundamentally from those discussed above
and therefore we do not discuss them in detail here. For example Ba zoning in sanidine crystals
from the A.D. 79 eruption of Vesuvius was modelled to determine the time elapsed between
magma recharge and eruption (Morgan et al. 2006). As described for Fe–Mg in pyroxenes, the
zoning of Ba was directly inferred from BSE images. In a similar study diffusion of Sr and Ba
in sanidine have been used to introduce the concept of binary element diffusion modelling and
determine magma residence times for sanidine crystals in the Bishop Tuff (Morgan and Blake
2006). Here again a slower (Ba) diffusing element was used to reconstruct the initial profile of
a faster diffusing element (Sr) (see also Till et al. 2015). A more comprehensive reconstruction
of the crystal history can be obtained, as demonstrated by Zellmer and Clavero (2006), when
trace element zoning in different zones of a sanidine crystal are modelled simultaneously
(analogous to the example shown in Figs. 1 and 2). In their study Sr diffusion was modelled in a
sanidine crystal with multiple growth zones from Taapaca volcano, Central Andes. A simplified
procedure was applied to model the oscillatory zoned Sr concentration profiles, which showed
that the core was older by about 750 yr compared to the outer growth zone, indicating an
average growth rate of the crystal on the order of 10−10 cm s−1. Ti diffusion in Qtz can be used
to reconstruct the thermal history of a magma as its content is strongly sensitive to temperature
(Wark et al. 2007). However, complex zoning patterns of Ti in Qtz, typically inferred indirectly
by cathodoluminescence (CL) intensity, could equally be the result of fast growth and related
kinetic effects and therefore no equilibrium of Ti between quartz and co-existing melt can be
assumed as a means to estimate temperatures (Seitz et al. 2016).
In a number of recent papers the potential of H diffusion in melt inclusions (e.g., Liu et
al. 2007; Chen et al. 2013; Newcombe at al. 2014) and melt embayments (e.g., Humphreys
et al. 2008; Lloyd et al. 2014; Ferguson et al. 2016) to determine magma ascent rates was
explored. Compared to olivine, the diffusion rate of H appears to be better constrained but
involves additional complexities related to the presence of two H species in glasses and melts,
molecular water and hydroxyl ions (e.g., Wasserburg 1988; Zhang et al. 1991). As for olivine,
the initial water content and the depth needed to be estimated, and a fixed boundary condition
was assumed, whereby the H content is completely lost from the boundary, which is not
realistic since the H content in melt is a function of pressure and related water fugacity in the
system. Clearly more diffusion chronometers will be developed in the future (for example Nb
and Zr diffusion in rutile as, for example, applied for metamorphic systems, e.g., see Zack and
Kooijman 2017, this volume) but this is obviously dependent on the availability and accuracy
of the respective diffusion data set. For example diffusion of Fe–Mg or other elements in
amphibole could be useful for magmatic systems, but diffusion data are currently lacking.
Finally, it can be concluded that for diffusion chronometry we cannot apply a simple recipe
to obtain reliable timescales for a generic magmatic scenario. But scientific progress in this
field is fast and, in conjunction with experimental studies, modelling strategies are developed
for the diffusion chronometer that is appropriate for the magmatic process of interest. Diffusion
chronometry of magmatic processes has yet to reach its full potential, although rapid and
concerted developments in the field suggest that this may not be the case for much longer.
ACKNOWLEDGMENTS
The authors would like to thank Martin Engi, Fidel Costa, and Philip Rupprecht for their
constructive and helpful reviews that improved the manuscript. In addition we thank Sumit
Chakraborty for helpful comments. JB acknowledges support from ERC Advanced Grant
“CRITMAG”
REFERENCES
Albarède F Bottinga Y (1972) Kinetic disequilibrium in trace element partitioning between phenocrysts and host
lava. Geochim Cosmochim Acta 36:141–156
Albert H, Costa F, Marti J (2015) Timing of magmatic processes and unrest associated with mafic historical
monogenetic eruptions in Tenerife Island. J Petrol 56:1945–1965
Albert H, Costa F, Marti J (2016) Years to weeks of seismic unrest and magmatic intrusions precede monogenetic
eruptions. Geology 44:211–214
Allan ASR, Morgan DJ, Wilson CJN, Millet M-A (2013) From mush to eruption in centuries: assembly of the
super-sized Oruanui magma body. Contrib Mineral Petrol 166:143–164
Beck P, Chaussidon M, Barrat JA, Gillet Ph, Bohn M (2006) Diffusion induced Li isotopic fractionation during the
cooling of magmatic rocks: the case of pyroxene phenocrysts from nakhlite meteorites. Geochim Cosmochim
Acta 70:4813–4825
Bell DR, Rossman GR, Maldener J, Endisch D, Rauch F (2003) Hydroxide in olivine: a quantitative determination of
the absolute amount and calibration of the IR spectrum. J Geophys Res 108:2105, doi:10.1029/2001 JB000679
Beran A, Libowitzky E (2006) Water in natural mantle minerals II: Olivine, garnet and accessory minerals. Rev
Berlo K, Turner S, Blundy J, Black S, Hawkesworth C (2006) Tracing pre-eruptive magma degassing using
(210Pb/226Ra) disequilibria in the volcanic deposits of the 1980–1986 eruption of Mount St. Helens. Earth
Berlo K, Blundy J, Turner S, Hawkesworth C (2007) Textural and chemical variation in plagioclase phenocrysts
from the 1980 eruptions of Mount St. Helens, USA Contrib Mineral Petrol 154:291–308
Berry A, Hermann J, O’Neill HSC, Foran GJ (2005) Fingerprinting the water site in mantle olivine. Geology
33:869–872
Berry AJ, O’Neill HStC, Hermann J, Scott DR (2007) The infrared signature of water associated with trivalent
cations in olivine. Earth Planet Sci Lett 261:134–142
Bindeman IN, Davies AM, Drake MJ (1998) Ion microprobe study of plagioclase–basalt partition experiments at
natural concentration levels of trace elements. Geochim Cosmochim Acta 62:1175–1193.
Blundy JD, Shimizu N (1991) Trace element evidence for plagioclase recycling in calc-alkaline magmas. Earth
Blundy JD, Wood BJ (1991) Crystal-chemical controls on the partitioning of Sr and Ba between plagioclase
feldspar, silicate melts and hydrothermal solutions. Geochim Cosmochim Acta 55:193–209
Blundy JD, Wood BJ (1994) Prediction of crystal–melt partition coefficients from elastic moduli. Nature 372:452–454
Blundy J, Cashman KV, Berlo K (2008) Evolving magma storage conditions beneath Mount St. Helens inferred
from chemical variations in melt inclusions from the 1980–1986 and current (2004–2006) eruptions. In:
Sherrod DR, Scott WE, Stauffer PH (Eds) A volcano rekindled: the renewed eruption of Mount St. Helens
2004–2006. US Geol Surv Prof Pap 1750:755–790
Bouvet de Maisonneuve C, Costa F, Huber C, Vonlanthen P, Bachmann O, Dungan M (2016) Do olivines faithfully
record magmatic events? Contrib Mineral Petrol 171: doi 10.1007/s00410-016-1264-6
Bowen NL (1913) The melting phenomena of the plagioclase feldspars. Am J Sci 35:577–599
Carmichael ISE, Nicholls J, Smith AL (1970) Silica activity in igneous rocks. Am Mineral 55:246–263
Chakraborty S (1997) Rates and mechanisms of Fe–Mg interdiffusion in olivine at 980–1300 °C J Geophys Res
102:12317–12331
Chakraborty S (2006) Diffusion modeling as a tool for constraining timescales of evolution of metamorphic rocks.
Chakraborty S (2008) Diffusion in solid silicates: a tool to track timescales of processes comes of age. Ann Rev
Earth Planet Sci 36:153–190
Chakraborty S (2010) Diffusion coefficients in olivine, wadsleyite and ringwoodite. Rev Mineral Geochem
72:603–639
Chakraborty S, Ganguly J (1991) Compositional zoning and cation diffusion in garnets. In: Ganguly J (ed)
Diffusion, atomic ordering and mass transport. Advances in physical geochemistry, vol 8. Springer, New
York, pp 120–175
Chamberlain, KJ, Morgan, DJ, Wilson, CJN (2014) Timescales of mixing and mobilisation in the Bishop Tuff
magma body: perspectives from diffusion chronometry. Contrib Mineral Petrol 168:1034 DOI 10.1007/
s00410-014-1034-2
Charlier BLA, Morgan DJ, Wilson CJN, Wooden JL, Allan ASR, Baker JA (2012) Lithium concentration gradients
in feldspar and quartz record the final minutes of magma ascent in an explosive supereruption. Earth Planet
Sci Lett 319–320:218–227
Chen Y, Provost A, Schiano P, Cluzel N (2013) Magma ascent rate and initial water concentration inferred from
diffusive water loss from olivine-hosted melt inclusions. Contrib Mineral Petrol 165:525–541
Cherniak DJ (2010) Cation diffusion in feldspars. Rev Mineral Geochem 72:691–733
Cherniak DJ, Watson EB (1994) A study of strontium diffusion in plagioclase using Rutherford backscattering
spectroscopy. Geochim Cosmochim Acta 58:5179–5190
Cherniak DJ, Dimanov A (2010) Diffusion in pyroxene, mica and amphibole. Rev Mineral Geochem 72:641–690
Cherniak DJ, Liang Y (2012) Ti diffusion in natural pyroxene. Geochim Cosmochim Acta 98:31–47. doi:10.1016/j.
gca.2012.09.021
Christopher TE, Blundy J, Cashman K, Cole P, Edmonds M, Smith PJ, Sparks RSJ, Stinton A (2015) Crustal-
scale degassing due to magma system destabilization and magma–gas decoupling at Soufriere Hills Volcano,
Montserrat. Geochem Geophys Geosystem 16:2797–2811
Cioni R, Marianelli P, Santacroce R (1999) Temperature of Vesuvius magma. Geology 27:443–446
Coogan LA, Jenkin GRT, Wilson RN (2002) Constraining the cooling rate of the lower oceanic crust: a new
approach applied to the Oman ophiolite. Earth Planet Sci Lett 199:127–146
Coogan LA, Hain A, Stahl S, Chakraborty S (2005) Experimental determination of the diffusion coefficient for
calcium in olivine between 900 and 1500 °C Geochim Cosmochim Acta 69:3683–3694
Coogan LA, Jenkin GRT, Wilson RN (2007) Contrasting cooling rates in the lower oceanic crust at fast- and slow-
spreading ridges revealed by geospeedometry. J Petrol 48:2211–2231
Coombs ML, Eichelberger JC, Rutherford MJ (2000) Magma storage and mixing conditions for the 1953–1974
eruptions of Southwest Trident Volcano, Katmai National Park, Alaska. Contrib Mineral Petrol 140:99–118
Cooper KM, Kent AJR (2014) Rapid remobilization of magmatic crystals kept in cold storage. Nature 506:480–483
Cooper M, Reid MR (2003) Re-examination of crystal ages in recent Mount St. Helens lavas: implications for
magma reservoir processes. Earth Planet Sci Lett 213:149–167
Cooper KM, Reid MR (2008) Uranium-series crystal ages. Rev Mineral Geochem 69:479–544
Costa F, Chakraborty S (2004) Decadal time gaps between mafic intrusion and silicic eruption obtained from
chemical zoning patterns in olivine. Earth Planet Sci Lett 227:517–530
Costa F, Dungan M (2005) Short timescales of magmatic assimilation from diffusion modelling of multiple
elements in olivine. Geology 33:837–840
Costa F, Morgan D (2010) Time constraints from chemical equilibration in magmatic crystals. In: Dosseto
A, Turner SP, Van Orman JA (eds) Timescales of magmatic processes: from core to atmosphere. Wiley,
Chichester, p. 125–159
Costa F, Chakraborty S, Dohmen R (2003) Diffusion coupling between trace and major elements and a model for
calculation of magma residence times using plagioclase. Geochim Cosmochim Acta 67:2189–2200
Costa F, Dohmen R, Chakraborty S (2008) Timescales of magmatic processes from modeling the zoning patterns
of crystals. Rev Mineral Geochem 69:545–594
Costa F, Coogan LA, Chakraborty S (2010a) The timescales of magma mixing and mingling involving primitive
melts and melt–mush interaction at mid-ocean ridges. Contrib Mineral Petrol 159:173–194
Costa F, Dohmen R, Demouchy S (2010b) Modelling the dehydrogenation of mantle olivine with implications for
the water content of the Earth’s upper mantle, and ascent rates of kimberlite and alkali basaltic magmas. AGU
Fall Meeting 2010, V24C-06.
Crank J (1975) The Mathematics of Diffusion, 2nd edition, 414 p. Oxford Science Publication, Oxford.
Dauphas N, Teng F-Z, Arndt NT (2010) Magnesium and iron isotopes in 2.7 Ga Alexo komatiites: Mantle
signatures, no evidence for Soret diffusion, and identification of diffusive transport in zoned olivine. Geochim
Davidson JP, Tepley FJ (1997) Recharge in volcanic systems; evidence from isotope profiles of phenocrysts.
Science 275:826–829
Demouchy S, Mackwell S (2006) Mechanisms of hydrogen incorporation and diffusion in iron-bearing olivine.
Phys Chem Miner 33:347–355
Demouchy S, Jacobsen SD, Gaillard F, and Stern CR (2006) Rapid magma ascent recorded by water diffusion
profiles in mantle olivine. Geology 34:429–432
Denis CMM, Demouchy S, Shaw CSJ (2013) Evidence of dehydration in peridotites from Eifel volcanic
field and estimates of the rate of magma ascent. J Volcanol Geotherm Res 258:85–99, doi:10.1016/j.
jvolgeores.2013.04.010
Dimanov A, Sautter V (2000) “Average” interdiffusion of (Fe,Mn)–Mg in natural diopside. Eur J Mineral 12:749–760
Dimanov A, Wiedenbeck M (2006) (Fe,Mn)–Mg interdiffusion in natural diopside: effect of pO2. Eur J Mineral
18:705–718
Dodson MH (1986) Closure profiles in cooling systems. Mater Sci Forum 7:145–154
Dohmen R, Chakraborty S, Palme H, Rammensee W (2003) The role of element solubility on the kinetics of
element partitioning: in situ observations and a thermodynamic kinetic model. J Geophys. Res 108: B3 2157,
doi:10.1029/2001 JB000587
Dohmen R, Blundy J (2014) A predictive thermodynamic model for element partitioning between plagioclase and
melt as a function of pressure, temperature and composition. Am J Sci 314:1319–1372
Dohmen R, Chakraborty S (2007) Fe–Mg diffusion in olivine II: Point defect chemistry, change of diffusion
mechanisms and a model for calculation of diffusion coefficients in natural olivine. Phys Chem Mineral
34:409–430
Dohmen R, Kasemann SA, Coogan L, Chakraborty S (2010) Diffusion of Li in olivine. Part I: Experimental obser-
vations and a multi species diffusion model. Geochim Cosmochim Acta 74:274–292
Dohmen R, Ter Heege JH, Becker H-W, Chakraborty S (2016a) Fe–Mg interdiffusion in orthopyroxene. Am
Mineral 101:2210–2221
Dohmen R, Marschall H, Wiedenbeck M, Polednia J, Chakraborty S (2016b) Trace element diffusion in minerals:
the role of multiple diffusion mechanisms operating simultaneously. AGU, Fall meeting 2016, MR51A-2682.
Drake MJ, Weill DF (1971) Petrology of Apollo-11 sample-10071—Differentiated mini-igneous complex. Earth
Druitt TH, Costa F, Deloule E, Dungan M, Scaillet B (2012) Decadal to monthly timescales of magma transfer and
reservoir growth at a caldera volcano. Nature 482:77–80
Dvorak JJ, Dzurisin D (1997) Volcano geodesy: The search for magma reservoirs and the formation of eruptive
vents. Rev Geophys 35: doi: 10.1029/97RG00070
Faak K, Gillis KM (2016) Slow cooling of the lowermost oceanic crust at the fast-spreading East Pacific Rise.
Geology 44:115–118
Faak K, Chakraborty S, Coogan LA (2013) Mg in plagioclase: experimental calibration of a new geothermometer
and diffusion coefficients. Geochim Cosmochim Acta 123:195–217
Faak K, Chakraborty S, Coogan LA (2014) A new Mg-in-plagioclase geospeedometer for the determination of
cooling rates of mafic rocks. Geochim Cosmochim Acta 140:691–707
Faak K, Coogan LA, Chakraborty S (2015) Near conductive cooling rates in the upper-plutonic section of crust
formed at the East Pacific Rise. Earth Planet Sci Lett 423:36–47, doi:10.1016/j.epsl.2015.04.025
Ferguson DJ, Gonnermann HM, Ruprecht P, Plank T, Hauri EH, Houghton BF, Swanson DA (2016) Magma
decompression rates during explosive eruptions of Kilauea volcano, Hawaii, recorded by melt embayments.
Bull Volcanol 78:71, doi: 10.1007/s00445-016- 1064-x
Ferriss E, Plank T, Walker D, Nettles M (2015) The whole-block approach to measuring hydrogen diffusivity in
nominally anhydrous minerals. Am Mineral 100:837–851
Ferris E, Plank T, Walker D (2016) Site-specific hydrogen diffusion rates during clinopyroxene dehydration.
Contrib Mineral Petrol 171:55 DOI 10.1007/s00410-016-1262-8
Gallagher K, Elliott T (2009) Fractionation of lithium isotopes in magmatic systems as a natural consequence of
cooling. Earth Planet Sci Lett 278:286–296
Ganguly J (2002) Diffusion kinetics in minerals: Principles and applications to tectono-metamorphic processes.
Eur Mineral Union Notes 4:271–309
Ganguly J, Tazzoli V (1994) Fe2+–Mg interdiffusion in orthopyroxene: Retrieval from the data on intracrystalline
exchange reaction. Am Mineral 79:930–937
Ganguly J, Tirone M (1999) Diffusion closure temperature and age of a mineral with arbitrary extent of diffusion;
Ganguly J, Bhattacharya RN, Chakraborty S (1988) Convolution effect in the determination of compositional
profiles and diffusion coefficients by microprobe step scans. Am Mineral 73:901–909
Gauthier PJ, Condomines M (1999) 210Pb–226Ra radioactive disequilibria in recent lavas and radon degassing:
inferences on the magma chamber dynamics at Stromboli and Merapi volcanoes. Earth Planet Sci Lett
172:111–126
Gerlach D, Grove T (1982) Petrology of Medicine Lake Highland volcanics: Characterization of endmembers of
magma mixing. Contrib Mineral Petrol 80:147–159
Ghiorso MS, Sack RO (1995) Chemical mass transfer in magmatic processes. IV A revised and internally consistent
thermodynamic model for the interpolation and extrapolation of liquid–solid equilibria in magmatic systems
at elevated temperatures and pressures. Contrib Mineral Petrol 119:197–212
Giletti BJ, Shanahan TM (1997) Alkali diffusion in plagioclase feldspar. Chem Geol 139:3–20
Ginibre C, Kronz A, Wörner G (2002a) High-resolution quantitative imaging of plagioclase composition using
accumulated backscattered electron images: New constraints on oscillatory zoning. Contrib Mineral Petrol
142:436–448
Ginibre C, Wörner G, Kronz A (2002b) Minor- and trace- element zoning in plagioclase: implications for magma
chamber processes at Parinacota volcano, northern Chile. Contrib Mineral Petrol 143:300–315
Girona T, Costa F (2013) DIPRA: a user-friendly program to model multi-element diffusion in olivine
with applications to timescales of magmatic processes. Geochem Geophys Geosyst 14:422–431,
doi:10.1029/2012 gC004427
Grove TL, Baker MB, Kinzler RJ (1984) Coupled CaAl–NaSi diffusion in plagioclase feldspar: experiments and
applications to cooling rate speedometry. Geochim Cosmochim Acta 48:2113–2121
Hartley ME, Morgan DJ, Maclennan J, Edmonds M, Thordarson T (2016) Tracking timescales of short-term
precursors to large basaltic fissure eruptions through Fe–Mg diffusion in olivine. Earth Planet Sci Lett
439:58–70
Holzapfel C, Chakraborty S, Rubie DC, Frost DJ (2007) Effect of pressure on Fe–Mg, Ni and Mn diffusion in
(FexMg1−x)2SiO4 olivine: Phys Earth Planet Inter 162:186–198
Humphreys MCS, Menand T, Blundy JD, Klimm K (2008) Magma ascent rates in explosive eruptions: Constraints
from H2O diffusion in melt inclusions. Earth Planet Sci Lett 270:25–40
Ingrin J, Blanchard M (2006) Diffusion of hydrogen in minerals. Rev Mineral Geochem 63:291–320
Jeffcoate AB, Elliott T, Kasemann SA, Ionov D, Cooper K, Brooker R (2007) Li isotope fractionation 4144 in
peridotites and mafic melts. Geochim Cosmochim Acta 71:202–218
John T, Gussone N, Podladchikov YY, Bebout GE, Dohmen R, Halama R, Klemd R, Magna T, Seitz HM (2012)
Volcanic arcs fed by rapid pulsed fluid flow through subducting slabs. Nat Geosci 5:489–492
Kahl M, Chakraborty S, Costa F, Pompilio M (2011) Dynamic plumbing system beneath volcanoes revealed by
kinetic modeling and the connection to monitoring data: An example from Mt. Etna. Earth Planet Sci Lett
308:11–22
Kahl M, Chakraborty S, Costa F, Pompilio M, Liuzzo M, Viccaro M (2013) Compositionally zoned crystals
and real-time degassing data reveal changes in magma transfer dynamics during the 2006 summit eruptive
episodes of Mt. Etna. Bull Volcanol 75:1–14
Kahl M, Chakraborty S, Pompilio M, Costa F (2015) Constraints on the nature and evolution of the magma
plumbing system of Mt. Etna Volcano (1991–2008) from a combined thermodynamic and kinetic modelling
of the compositional record of minerals. J Petrol 56:2015–2068
Kayzar TM, Cooper KM, Reagan MK, Kent AJR (2009) Gas transport model for the magmatic system at Mount
Pinatubo, Philippines: Insights from (210Pb)/(226Ra). J Volcanol Geotherm Res 181:124–140
Kilgour GN, Saunders KE, Blundy JD, Cashman KV, Scott BJ, Miller CA (2014) Timescales of magmatic
processes at Ruapehu volcano from diffusion chronometry and their comparison to monitoring data. J
Volcanol Geotherm Res 288:62–75
Köhler TP, Brey GP (1990) Calcium exchange between olivine and clinopyroxene calibrated as a geothermometer
for natural peridotites from 2 to 60 kb with applications. Geochim Cosmochim Acta 54:2375–2388
Kohlstedt DL, Mackwell SJ (1998) Diffusion of hydrogen and intrinsic point defects in olivine. Z Phys Chem
207:147–162
Kohlstedt DL, Keppler H, Rubie DC (1996) Solubility of water in the α, β and γ phases of (Mg,Fe)2SiO4. Contrib
Geochem 83:103–152
Krimer D, Costa F (2016) Evaluation of the effects of 3d diffusion, crystal geometry, and initial conditions on
retrieved time-scales from Fe–Mg zoning in natural oriented orthopyroxene crystals. Geochim Cosmochim
Acta 196:271–288
Lanphere M, Champion D, Melluso L, Morra V, Perrotta A, Scarpati C, Tedesco D, Calvert A (2007) 40Ar/39Ar ages
of the AD 79 eruption of Vesuvius, Italy. Bull Volcanol 69:259–263
Lasaga AC (1983) Geospeedometry: an extension of geothermometry. In: Kinetics and equilibrium in mineral
reactions (ed. Saxena SK). Springer-Verlag, New York. pp. 82–114
Lai Y-J, Pogge von Strandmann PAE, Dohmen R, Takazawa E, Elliott T (2015) The influence of melt infiltration
onthe Li and Mg isotopic composition of the Horoman Peridotite Massif. Geochim Cosmochim Acta
164:318–332
LaTourrette T, Wasserburg GJ (1998) Mg diffusion in anorthite: implications for the formation of early solar system
planetesimals. Earth Planet Sci Lett 158:91–108
Liu Y, Anderson AT, Wilson CJN (2007) Melt pockets in phenocrysts and decompression rates of silicic magmas
before fragmentation. J Geophys Res, Solid Earth 112:B06204, doi: 10.1029/2006 JB004500
Lloyd AS, Ruprecht P, Hauri EH, Rose W, Gonnerman HM, Plank T (2014) NanoSIMS results from olivine-hosted
melt embayments: magma ascent rate during explosive basaltic eruptions. J Volc Geotherm Res 283:1–18
Longhi, J, Walker D, Hays J (1976) Fe and Mg in plagioclase. Proc Lunar Sci Conf 7:1281–1300
Longpré MA, Klügel A, Diehl A, Stix J (2014) Mixing in mantle magma reservoirs prior to and during the 2011–
2012 eruption at El Hierro, Canary Islands. Geology 42:315–318
Mackwell SJ, Kohlstedt DL (1990) Diffusion of hydrogen in olivine: implications for water in the mantle. J
Geophys Res 95(B4):5079, doi:10.1029/JB095iB04p05079
Marti J, Castro A, Rodriguez C, Costa F, Carrasquilla S, Pedreira R, Bolos × (2013) Correlation of magma evolution
and geophysical monitoring during the 2011–2012 El Hierro (Canary Islands) Submarine Eruption. J Petrol
54:1349–1373
Martin VM, Morgan DJ, Jerram DA, Caddick MJ, Prior DJ, Davidson JP (2008) Bang! Month-scale eruption
triggering at Santorini volcano. Science 321:1178
Melnik O, Sparks RSJ (2005) Controls on conduit magma flow dynamics during lava dome building eruptions. J
Geophys Res 110: B02209, doi:10.1029/2004 JB003183
Mierdel K, Keppler H, Smyth JR, Langenhorst F (2007) Water solubility in aluminous orthopyroxene and the
origin of Earth’s asthenosphere. Science 315:364–368
Milman-Barris MS, Beckett JR, Michael MB, Hofmann A, Morgan Z, Crowley MR, Vielzeuf D, Stolper E (2008)
Zoning of phosphorus in igneous olivine. Contrib Mineral Petrol 155:739–765
Miller SA, Asimow PD, Burnett DS (2006) Determination of melt influence on divalent element partitioning
between anorthite and CMAS melts. Geochim Cosmochim Acta 70:4258–4274
Moore A, Coogan L, Costa F, Perfit MR (2014) Primitive melt replenishment and crystal-mush disaggregation in
the weeks preceding the 2005–2006 eruption 9° 50’N, EPR Earth Planet Sci Lett 403:15–26
Morgan DJ, Blake S (2006) Magmatic residence times of zoned phenocrysts: Introduction and application of the
binary element diffusion modelling (BEDM) technique: Contrib Mineral Petrol 151:58–70
Morgan DJ, Blake S, Rogers NW, DeVivo B, Rolandi G, Macdonald R, Hawkesworth J (2004) Timescales of
crystal residence and magma chamber volume from modeling of diffusion profiles in phenocrysts: Vesuvius
1944. Earth Planet Sci Lett 222:933–946
Morgan, DJ, Blake S, Rogers NW, DeVivo B, Rolandi G, Davidson J (2006) Magma recharge at Vesuvius in the
century prior to the eruption AD 79. Geology 34:845–848
Mosenfelder JL, Deligne NI, Asimow PD, Rossman GR (2006) Hydrogen incorporation in olivine from 2–12 GPa.
Müller T, Dohmen R, Becker HW, ter Heege J, Chakraborty S (2013) Fe–Mg interdiffusion rates in clinopyroxene:
experimental data and implications for Fe–Mg exchange geothermometers. Contrib Mineral Petrol 166:1563–
1576
Nakamura M (1995) Residence time and crystallization history of nickeliferous olivine phenocrysts from the
northern Yatsugatake volcanoes, Central Japan: Application of a growth and diffusion model in the system
Mg–Fe–Ni. J Volcanol Geotherm Res 66:81–100
Newcombe ME, Fabbrizio A, Zhang Y, Ma C, Le Voyer M, Guan Y, Eiler JM, Saal AE, Stolper EM (2014)
Chemical zonation in olivine-hosted melt inclusions. Contrib Mineral Petrol 168:1–26
Newman S, Lowenstern JB (2002) VolatileCalc: a silicate melt–H2O–CO2 solution model written in Visual Basic
for Excel. Comp Geosci 28:597–604
Oeser M, Weyer S, Horn I, Schuth S (2014) High-precision Fe and Mg isotope ratios of silicate reference glasses
determined in situ by femtosecond LA-MC-ICP-MS and by solution nebulisation MC-ICP-MS Geostand
Geoanal Res 38:311–328
Oeser M, Dohmen R, Horn I, Schuth S, Weyer S (2015) Processes and time scales of magmatic evolution as
revealed by Fe–Mg chemical and isotopic zoning in natural olivines. Geochim Cosmochim Acta 154:130–150
Onorato PIK, Hopper RW, Yinnon H, Uhlmann DR, Taylor LA, Garrison JR, Hunter R (1981) Solute partitioning
under continuous cooling conditions as a cooling rate indicator. J Geophys Res 86:9511–9518
Padron-Navarta JA, Hermann J, O’Neill HStC (2014) Site-specific hydrogen diffusion rates in forsterite. Earth
Pan Y, Batiza R (2002) Mid-ocean ridge magma chamber processes: Constraints from olivine zonation in lavas
from the East Pacific Rise at 9°30’N and 10°30’N J Geophys Res 107:2022
Pearce TH (1984) The analysis of zoning in magmatic crystals with emphasis on olivine. Contrib Mineral Petrol
86:149–154
Peslier AH, Luhr JF (2006) Hydrogen loss from olivines in mantle xenoliths from Simcoe (USA) and Mexico:
Mafic alkalic magma ascent rates and water budget of the sub-continental lithosphere. Earth Planet Sci Lett
242:302–319
Peslier AH, Woodland AB, Wolff JA (2008) Fast kimberlite ascent rates estimated from hydrogen diffusion profiles
in xenolithic olivines from Southern Africa. Geochim Cosmochim Acta 72:2711–2722
Peslier AH, Bizimis M, Matney M (2015) Water disequilibrium in olivines from Hawaiian peridotites: Recent
metasomatism, H diffusion and magma ascent rates. Geochim Cosmochim Acta 154:98–117
Peters MT, Shaffer EE, Burnett DS, Kim SS (1995) Magnesium and titanium partitioning between anorthite and
Type B CAI liquid: Dependence on oxygen fugacity and liquid composition Geochim Cosmochim Acta
59:2785–2796
Petry C, Chakraborty S, Palme H (2004) Experimental determination of Ni diffusion coefficients in olivine and
their dependence on temperature, composition, oxygen fugacity, and crystallographic orientation. Geochim
Pinilla C, Davis SA, Scott TB, Allan, NL, Blundy JD (2012) Interfacial storage of noble gases and other trace
elements in magmatic systems. Earth Planet Sci Lett 319:287–294
Pistolesi M, Rosi M, Cioni R, Cashman KV, Rossotti A, Aguilera E (2011) Physical volcanology of the post-XII
century activity at Cotopaxi Volcano, Ecuador: behavior of an andesitic central volcano. Geol Soc Am Bull
123:1193–1215
Qian Q, O’Neill HSC, Hermann J (2010) Comparative diffusion coefficients of major and trace elements in olivine
at 950 °C from a xenocryst included in dioritic magma. Geology 38:331–334
Rae ASP, Edmonds M, Maclennan J, Morgan D, Houghton B, Hartley ME, Sides I (2016) Time scales of magma
transport and mixing at Kilauea Volcano, Hawai’i. Geology 44:463–466
Richter FM, Liang Y, Davis AM (1999) Isotope fractionation by diffusion in molten oxides. Geochim Cosmochim
Acta 63:2853–2861
Richter FM, Davis AM, DePaolo DJ, Watson EB (2003) Isotope fractionation by chemical diffusion between
molten basalt and rhyolite. Geochim Cosmochim Acta 67:3905–3923
Richter F, Watson B, Chaussidon M, Mendybaev R, Ruscitto D (2014) Lithium isotope fractionation by diffusion
in minerals. Part 1: Pyroxenes. Geochim Cosmochim Acta 126:352–370
Ridolfi F, Renzulli A, Puerini M (2010) Stability and chemical equilibrium of amphibole in calc-alkaline magmas:
an overview, new thermobarometric formulations and application to subduction-related volcanoes. Contrib
Ruprecht P, Cooper KM (2012) Integrating uranium-series and elemental diffusion geochronometers in mixed
magmas from Volcan Quizapu, Central Chile. J Petrol 53:841–871
Ruprecht P, Plank T (2013) Feeding andesitic eruptions with a high-speed connection from the mantle. Nature
500:68–72
Ruprecht P, Wörner G (2007) Variable regimes in magma systems documented in plagioclase zoning patterns: El
Misti stratovolcano and Andahua monogenetic cones. J Volcanol Geotherm Res 165:142–162
Sano J, Ganguly J, Hervig R, Dohmen R, Zhang X(2011) Neodymium diffusion in orthopyroxene: Experimental
studies and applications to geological and planetary problems. Geochim Cosmochim Acta 75:4684–4698
Saunders K, Blundy J, Dohmen R, Cashman K (2012a) Linking petrology and seismology at an active volcano.
Science 336:1023–1027
Saunders K, Rinnen S, Blundy J, Dohmen R, Klemme S, Arlinghaus HF (2012b) TOF-SIMS and electron
microprobe investigation of zoned magmatic orthopyroxenes: first results of trace and minor element analysis
with implications for diffusion modelling. Am Mineral 97:532–542
Saunders K, Buse B, Kilburn MR, Kearns S, Blundy J (2014) Nanoscale characterisation of crystal zoning. Chem
Geol 364:20–32
Schoen AH (1958) Correlation and the isotope effect for diffusion in crystalline solids. Phys Rev Lett 1:138–140
Schwandt CS, Cygan RT, Westrich HR (1998) Magnesium self-diffusion in orthoenstatite. Contrib Mineral Petrol
130:390–396
Schwinger S, Dohmen R, Schertl HP (2016) A combined diffusion and thermal modeling approach to determine
peak temperatures of thermal metamorphism experienced by meteorites. Geochim Cosmochim Acta
191:255–276
Seitz S, Putlitz B, Baumgartner LP, Escrig S, Meibom A, Bouvier AS (2016) Short magmatic residence times of
quartz phenocrysts in Patagonian rhyolites associated with Gondwana breakup. Geology 44:67–70
Shea T, Lynn KJ, Garcia MO (2015a) Cracking the olivine zoning code: Distinguishing between crystal growth
and diffusion. Geology 43:935–938
Shea T, Costa F, Krimer D, Hammer JE (2015b) Accuracy of timescales retrieved from diffusion modeling in
olivine: A 3D perspective. Am Mineral 100:2026–2042
Simon JI, Hutcheon ID, Simon SB, Matzel JEP, Ramon EC, Weber PK, Grossman L, DePaolo DJ (2011) Oxygen
isotope variations at the margin of a CAI records circulation within the solar nebula. Science 331:1175–1178,
doi: 10.1126/science.1197970
Singer BS, Dungan MA, Layne GD (1995) Textures and Sr, Ba, Mg, Fe, K, and Ti compositional profiles in
volcanic plagioclase: Clues to the dynamics of calc-alkaline magma chambers. Am Mineral 80:776–798
Singer B, Costa F, Herrin JS, Hildreth W, Fierstein J (2016) The timing of compositionally zoned magma reservoirs
and mafic ‘priming’weeks before the 1912 Novarupta-Katmai rhyolite eruption. Earth Planet Sci Lett
451:125–137
Sio CK, Dauphas N (2016) Thermal and crystallization histories of magmatic bodies by Monte Carlo inversion of
Mg–Fe isotopic profiles in olivine. Geology DOI: 10.1130/G38056.1
Sio CK, Dauphas N, Teng F-Z, Chaussidon M, Helz RT, Roskosz M (2013) Discerning crystal growth from diffusion
profiles in zoned olivine by in-situ Mg-Fe isotopic analyses. Geochim Cosmochim Acta 123:302–321
Smith VC, Blundy JD, Arce JL (2009) A Temporal Record of Magma Accumulation and Evolution beneath
Nevado de Toluca, Mexico, Preserved in Plagioclase Phenocrysts. J Petrol 50:405–426
Spandler C, O’Neill HStC (2010) Diffusion and partition coefficients of minor and trace elements in San Carlos
olivine at 1300 °C with some geochemical implications. Contrib Mineral Petrol 159:791–818
Sundvall R, Skogby H, Stalder R (2009) Dehydration–hydration mechanisms in synthetic Fe-poor diopside. Eur J
Mineral 21:17–26. doi:10.1127/0935–1221/2009/0021–1880
Stalder R, Behrens H (2006) D/H exchange in pure and Cr-doped enstatite: implications for hydrogen diffusivity.
Phy Chem Mineral 33:601–611
Stalder R, Skogby H (2003) Hydrogen diffusion in natural and synthetis orthopyroxene. Phy Chem Mineral 30:12–19
Stalder R, Skogby H (2007) Dehydration mechanisms in synthetic Fe-bearing enstatite. Eur J Mineral 19:201–216
Streck MJ (2008) Mineral textures and zoning as evidence for open system processes. Rev Mineral Geochem
69:595–622
Tachibana S, Tamada S, Kawasaki H, Ozawa K, Nagahara H (2013) Interdiffusion of Mg–Fe in olivine at 1,400–
1,600 °C and 1 atm total pressure. Phys Chem Minerals 40:511–519
Teng F-Z, Dauphas N, Helz RT (2008) Iron isotope fractionation during magmatic differentiation in Kilauea Iki
lava lake. Science 320:1620–1622
Teng F-Z, Li W-Y, Ke S, Marty B, Dauphas N, Huang S, Wu F-Y, Pourmand A, (2010) Magnesium isotopic
composition of the Earth and chondrites. Geochim Cosmochim Acta 74:4150–4166
Teng F-Z, Dauphas N, Helz R T, Gao S, Huang S (2011) Diffusion-driven magnesium and iron isotope fractionation
in Hawaiian olivine. Earth Planet Sci Lett 308:317–324
Tepley FJ III, Davidson JP, Tilling RI, Arth JG (2000) Magma mixing, recharge and eruption histories recorded in
plagioclase phenocrysts from El ChichÓn volcano, Mexico. J Petrol 41:1397–1411
Thoraval C, Demouchy S (2014) Numerical models of ionic diffusion in one and three dimensions: Application to
dehydration of mantle olivine. Phys Chem Minerals 41:709–723. doi:10.1007/s00269-014-0685-x
Till CB, Vazquez JA, Boyce JW (2015) Months between rejuvenation and volcanic eruption at Yellowstone caldera,
Wyoming. Geology 43:695–698
Tilling RI, Dvorak JJ (1993) Anatomy of a basaltic volcano. Nature 363:125–132
Tomascak PB, Magna T, Dohmen R (2016) Advances in Lithium Isotope Geochemistry. Springer Cham Heidelberg
Turner S, George R, Jerram DA, Carpenter N, Hawkesworth C (2003) Case studies of plagioclase growth and
residence times in island arc lavas from Tonga and the Lesser Antilles, and a model to reconcile discordant
age information. Earth Planet Sci Lett 214:279–294
Turner S, Costa F (2007) Measuring timescales of magmatic evolution. Elements 3:267–272
Van Orman JA, Cherniak DJ, Kita NT (2014) Magnesium diffusion in plagioclase: dependence on composition,
and implications for thermal resetting of the 26Al–26Mg early solar system chronometer. Earth Planet Sci Lett
385:79–88
Van Orman JA, Krawczynski MA (2015) Theoretical constraints on the isotope effect for diffusion in minerals.
VanTongeren JA, Kelemen PB, Hanghoj K (2008) Cooling rates in the lower crust of the Oman ophiolite: Ca in
olivine, revisited. Earth Planet Sci Lett. 267:69–82, doi:10.1016/j.epsl.2007.11.034
Vogt K, Dohmen R, Chakraborty S (2015) Fe–Mg diffusion in spinel—New experimental data and a point defect
model. Am Mineral 100:2112–2122
Walker AM, Hermann J, Berry AJ, O’Neill HSC (2007) Three water sites in upper mantle olivine and the role of
titanium in the water weakening mechanism. J Geophys Res 112:B05211, doi:10.1029/2006 JB004620
Wark DA, Hildreth W, Spear FS, Cherniak DJ, Watson EB (2007) Pre-eruption recharge of the Bishop magma
system. Geology 35:235–238
Wasserburg GJ (1988) Diffusion of water in silicate melts. J Geol 96:363–367
Watson EB (1996) Surface enrichment and trace-element uptake during crystal growth. Geochim Cosmochim Acta
60:5013–5020
Watson EB, Baxter EF (2007) Diffusion in solid-Earth systems. Earth Planet Sci Lett 253:307–327
Watson EB, Cherniak DJ (2015) Quantitative cooling histories from stranded diffusion profiles. Contrib Mineral
Petrol 169:1–14
Watson EB, Liang Y (1995) A simple model for sector zoning in slowly grown crystals: Implications for growth
Watson EB, Müller T (2009) Non-equilibrium isotopic and elemental fractionation during diffusion-controlled
crystal growth under static and dynamic conditions. Chem Geol 267:111–124
Welsch B, Faure F, Famin V, Baronnet A, Bachèlery P (2013) Dendritic crystallization: a single process for all the
textures of olivine in basalts? J Petrol 54:539–574
Welsch B, Hammer JE, Hellebrand E (2014) Phosphorus zoning reveals dendritic architecture of olivine. Geology
42:867–870
Withers AC, Hirschmann MM (2008) Influence of temperature, composition, silica activity and oxygen fugacity on
the H2O storage capacity of olivine at 8 GPa. Contrib Miner Petrol 156:595–605
Zellmer GF, Blake S, Vance D, Hawkesworth C, Turner S (1999) Plagioclase residence times at two island arc
volcanoes (Kameni Islands, Santorini, and Soufriere, St. Vincent) determined by Sr diffusion systematics.
Zellmer GF, Clavero JE (2006) Using trace element correlation patterns to decipher a sanidine crystal growth
chronology: An example from Taapaca volcano, Central Andes. J Volcanol Geotherm Res 156:291–301
Zhang Y (2010) Diffusion in minerals and melts: Theoretical background. Rev Mineral Geochem 72:5–59
Zhang YX, Stolper EM, Wasserburg GJ (1991) Diffusion of water in rhyolitic glasses. Geochim Cosmochim Acta
55:441–456
Zhao Y-H, Ginsberg SB, Kohlstedt DL (2004) Solubility of hydrogen in olivine: dependence on temperature and
iron content. Contrib Miner Petrol 147:155–161, doi:10.1007/s00410-003-0524-4
Zhukova I, O’Neill H StC, Cambell IH, Kilburn MR (2014) The effect of silica activity on the diffusion of Ni and
Co in olivine. Contrib Mineral Petrol 168:1029, DOI 10.1007/s00410-014-1029-z

Petrochronology Methods and Applications by Matthew J. Kohn

Uploaded by

Document Information

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

Petrochronology Methods and Applications by Matthew J. Kohn

Uploaded by

Copyright:

Available Formats

REVIEWS in MINERALOGY

Martin Engi & Pierre Lanari

Series Editor: Ian Swainson

1 Significant Ages—An Introduction to Petrochronology

2 Phase Relations, Reaction Sequences and Petrochronology

SUPRASOLIDUS PHASE RELATIONS AND REACTION SEQUENCES.........................30

3 Local Bulk Composition Effects on Metamorphic Mineral Assemblages

APPENDIX 1—LOCAL BULK COMPOSITIONS

4 Diffusion: Obstacles and Opportunities in Petrochronology

INTRODUCTION AND SCOPE...........................................................................................103

5 Electron Microprobe Petrochronology

6 Petrochronology by Laser-Ablation Inductively Coupled

7 Secondary Ionization Mass Spectrometry Analysis in Petrochronology

INTRODUCTION AND SCOPE...........................................................................................199

Large magnet radius instruments for geochronology and petrochronology...............201

8 Petrochronology and TIMS

9 Zircon: The Metamorphic Mineral

Rare earth elements....................................................................................................271

10 Petrochronology of Zircon and Baddeleyite in Igneous Rocks:

11 Hadean Zircon Petrochronology

OTHER WESTERN AUSTRALIAN HADEAN ZIRCON OCCURRENCES.....................347

12 Petrochronology Based on REE-Minerals:

INTRODUCTION AND SCOPE...........................................................................................365

Monazite and xenotime thermometry.........................................................................373

INTRODUCTION AND SCOPE...........................................................................................419

14 Petrology and Geochronology of Rutile

INTRODUCTION AND SCOPE...........................................................................................443

15 Garnet: A Rock-Forming Mineral Petrochronometer

Petrochronology of garnet: Collisional tectonics

16 Chronometry and Speedometry of Magmatic Processes

Significant Ages—An Introduction

INTRODUCTION AND SCOPE

1529-6466/17/0083-0001$05.00 (print) http://dx.doi.org/10.2138/rmg.2017.83.1

SIGNIFICANCE OF AGE DATA

PETRO-CHRONO-LOGICAL APPROACH AND AMBITION

Phase Relations, Reaction Sequences

(a) Ti-in-quartz thermometer

-40 aTi = 0.7

400 500 600 700 800 900 1000

(b) Ti-in-zircon thermometer

350 450 550 650 750 850

PHASE EQUILIBRIA MODELLING

Table 1. Bulk compositions used for phase equilibria modelling (mol%)

Phase equilibria modelling of subsolidus and suprasolidus systems requires different

SUBSOLIDUS PHASE RELATIONS AND REACTION SEQUENCES

17 l ilm pl ilm pl ilm

(a) Metapelite: clockwise P–T max T

ttn ru max P ilm prograde retrograde mt

Average passive margin greywacke + H2O + q

1. g ep mu chl pa ab ru ttn 11. g pl bi mu ilm hem pa ab ru 21. g pl ep bi mu chl ilm pa ab ru

(a) Greywacke: clockwise P–T

(b) Greywacke: hairpin P–T max P max T

SUPRASOLIDUS PHASE RELATIONS AND REACTION SEQUENCES

Average Pelite + liq + ilm

1: g bi mu pl ru q 8: g cd bi ksp pl q 15: g sill ksp pl

Metapelite: clockwise P–T

(b) 100 1.25

Metapelite: counterclockwise P–T

After muscovite is completely consumed, partial melting continues through the

Average Passive Margin Greywacke + q + ilm + liq

Greywacke: clockwise P–T

(b) 100 1.25

Greywacke: counterclockwise P–T

Average mid ocean ridge basalt