Ornl 4254 by The "E" Generation

HOME HELP ORNL-4254

UC-80

- Reactor Technology

Contract No. W-7405-eng-26

MOLTEN-SALT REACTOR PROGRAM SEMIANNUAL PROGRESS REPORT For Period Ending February 29,

M. W. R. B. P. R.

1968

Rosenthal, Program Director Briggs, Associate Director Kasten, Associate Director

LEGAL N q T I C E TMs r e m

was prepred 88 an account of Qovernment #ponwred work. Netther tbe Unlted states, mor the Commission, nor q pram actlng oh behU of tbe Cmmisston: A. moLes any wranty or repremntatlon.expressed or impUed,with respect to the U.Nracy completeness, or usefulness of the Iuformauon eontalmd In thls report, or that the use of a& informanon. appantus. method. or pmcess disclosed in th(s report may not

prlvately owned r@t&or 8. Aosumes any Uahilttles with respect to tbe ma of. or for dsmwes resulUw from the use of any Iuformatlon. apparatus. method. or process discload In thls report AO used the above. “prwn a&g on behalf of the CommIsslon” Includes any smployee or contractor of the Commission, or employee of much contractor. to &e e N n t that such employee or contractor of the Commission. o t employee of Nch contractor prepare.. disseminates. or provides acce8s to, anr Informatlobpuouant to hlo e m p l o m t or eontract with the Commission. br hlo employment wltb auch contractor.

AUGUST 1968

OAK RIDGE NATIONAL LABORATORY Oak Ridge, Tennessee operated by UNION CARBIDE CORPORATION for the U. 5. ATOMIC ENERGY COMMISSION

This report i s one of a series of periodic reports in which we describe briefly the progress of the program. Other reports issued in this series are listed below. ORNL-3708 i s an especially useful report, because it gives a thorough review of the design and construction-and supporting development work for the MSRE. It also describes much of the general technology for molten-salt reactor systems.

0RNL-2474 ORNL-2626 ORNL-2684. 0RNL-2723 ORNL-2799 ORNL-2890 ORNL-2973 ORNL-30 14 ORNL-3122 ORNL-3215 ORNL-3282 ORNL-3369 ORNL-3419 ORNL-3529 ORNL-3626 ORNL-3708 ORNL-38 12 ORNL-3872 ORNL-3936 0RNL-4037

Period Ending January 31, 1958 Period Ending October 31, 1958 Period Ending January 31, 1959 Period Ending April 30, 1959 Period Ending July 31, 1959 Period Ending October 31, 1959 Periods Ending January 31 and April 30, 1960 Period Ending July 31, 1960 Period Ending February 28, 1961 Period Ending August 31, 1961 Period Ending February 28, 1962 Period Ending August 31, 1962 Period Ending January 31, 1963 Period Ending July 31, 1963 Period Ending January 31, 1964 Period Ending July 31, 1964 Period Ending February 28, 1965 \

Period Ending August 31, 1965 Period Ending February 28, 1966 Period Ending August 31, 1966

ORNL-4119

Period Ending February 28, 1967

ORNL-4191

Period Ending August 31, 1967

Contents ................................. SUMMARY.............................................................................................................................................................. INTRODUCTION ................................................................................................................

;

ix xi

PART 1. MOLTEN-SALT REACTOR EXPERIMENT

.......................................... 1. MSRE OPERATIONS ......................................................................................... 1.1 Chronological Account of Operations and Maintenance.............................................................. 1.2 Operations Analysis ........................................................................................................................ 1.2.1 Reactivity Balance .......................................................................................................... 1.2.2 Variations in Reactor Access Nozzle Temperatures .................................................. 1.2.3 Radiation Heating ............................................................................................................ 1.2.4 Thermal Cycle History .................................................................................................... 1.2.5 * 36U Indication of Integrated Power .............................................................................. 1.3 Equipment Performance .................................................................................................................. 1.3.1 Salt Pumps ........................................................................................................................ 1.3.2 Heat Transfer .................................................................................................................... 1.3.3 Salt Samplers .................................................................................................................... 1.3.4 Control Rods and Drives .................................................................................................. 1.3.5 Radiator Enclosure .......................................................................................................... 1.3.6 Off-Gas Systems ................................................................................................................. 1.3.7 Main Blowers .................................................................................................................... 1.3.8 Heaters .............................................................................................................................. 1.3.9 Electrical System .............................................................................................................. 1.3.10 Salt Pump Oil Systems'..................................................................................................... 1.3.11 Cooling Water System ...................................................................................................... 1.3.12 Component Cooling Systems ............................................................................................ 1.3.13 Containment and Ventilation ..........................................................................................

2 COMPONENT DEVELOPMENT ................................................................................................................ 2.1 Off-Gas Sampler................................................................................................................................ 2.2 Fuel Sampler-Enricher .................................................................................................................... 2.3 Decontamination Studies ................................................................................................................ 2.4 Study of Pin-Hole Camera for Gamma Source Mapping................................................................

............................................................................................ 2.6 Pumps ................................................................................................................................................ 2.6.1 Mark 2 Fuel Pump ............................................................................................................ 2.6.2 MSRE Oil Pumps ..............................................................................................................

2.5

Freeze-Flange Thermal Cycle Tests

2.6.3

Oil Pump Endurance Test ................................................................................................ iii

1 1 3 3 7 9 9 10 11 11 11 12 13 13 13 14 15 15

16 16 17 17 19 19 20 20 22 23 29 29 29 29

3. INSTRUMENTS AND CONTROLS.............................................................................................................. 3.1 MSRE Operating Experience .......................................................................................................... 3.1.1 Safety System Components .............................................................................................. 3.1.2 Thermocouples .................................................................................................................. 3.1.3 Other Instruments and Controls ...................................................................................... 3.2 Control System Design .................................................................................................................... 3.3 MSRE Neutron Noise Analysis ...................................................................................................... 3.4 Test of MSRE Rod Control System Under Simulated 233U Loading Conditions ...................... 3.5 Analog Computer Studies of the MSRE System with 233U Fuel Loading ................................

30 30 30 30 31 31 32 32 35

4. MSRE REACTOR ANALYSIS...................................................................................................................... 4.1 Introduction ...................................................................................................................................... 4.2 Simulation of Nuclear Excursion Incidents .................................................................................. 4.2.1 Uncontrolled Rod Withdrawal .......................................................................................... 4.2.2 Return of Separated Uranium to the Core ......................................................................

36 36 37 37 40

4.3

Detection of Anomalous Reactivity Effects

................................................................................

PART 2 MSBR DESIGN AND DEVELOPMENT

5 DESIGN .......................................................................................................................................................... 5.1 General .............................................................................................................................................. 5.2 Flow Diagram ....................................................................................................................................

51 51 51

5.3 Plant Layout .................................................................................................................................... 5.4 Reactor Vessel and Core ................................................................................................................ 5.5 Primary Heat Exchanger ..................................................................................................................

53 55 61

5.6

Fuel Drain Tank

..............................................................................................................................

6. REACTOR PHYSICS ..................................................................................................................................... 6.1 MSBR Physics Analysis .................................................................................................................. 6.1.1 Reference Reactor ............................................................................................................ 6.1.2 Fuel-Cycle Costs .............................................................................................................. 6.1.3 Cell Calculations .............................................................................................................. 6.1.4 Reactivity Coefficients .................................................................................................... 6.1.5 Measurements of Eta for 233U and 'jSU in the MSRE ................................................

7 SYSTEMS AND COMPONENTS DEVELOPMENT .................................................................................... 7.1 Noble Gas Behavior in an MSBR .................................................................................................... 7.2 Sodium Fluoroborate Circulating Loop Test ................................................................................ 7.3 MSBR Pumps .................................................................................................................................... 7.3.1 Pump Program .................................................................................................................... 7.3.2 Fuel Salt Pump.................................................................................................................. 7.3.3 Coolant Salt Pump ............................................................................................................ 7.3.4 Molten-Salt Pump Test Facility ...................................................................................... 7.3.5 Molten-Salt Bearing Program .......................................................................................... 7.3.6 Rotor-Dynamics Feasibility Investigation ..................................................................... 7.4 Remote Maintenance ........................................................................................................................

./..

I '

66 66 66 70 70 71 72 74 74 75 75 75 76 78 78 79 82 82

4 5

8. MSBR INSTRUMENTATION AND CONTROLS ........................................................................................ 8.1

Analog Computer Studies ................................................................................................................

l -

PART 3 CHEMISTRY

9. CHEMISTRY OF THE MSRE ...................................................................................................................... 9.1 Fuel Salt Composition and Purity .................................................................................................. 9.2 9.3

MSRE Fuel Circuit Corrosion Chemistry ...................................................................................... Isotopic Composition of Uranium in MSRE Fuel Salt ..................................................................

88 89 90 93

94 10 FISSION PRODUCT BEHAVIOR................................................................................................................ 94 10.1 Fission Product Behavior in the MSRE ........................................................................................ 94 10.1.1 Fission Products in the MSRE Fuel .............................................................................. 96 10.1.2 Fission Products in the MSRE Cover Gas .................................................................... 10.1.3 Deposition of Fission Products from MSRE Cover Gas on Metal Specimens .......... 96 99 10.1.4 Examination of MSRE Surveillance Specimens After 24, 000 Mwhr ............................ 100 10.1.5 Hot-Cell Tests on Fission Product Volatilization from Molten MSRE Fuel ............ 108 10.1.6 Miscellaneous Tests ........................................................................................................ 115 10.2 Fission Product Distribution in an MSRE Graphite Surveillance Specimen ............................ 10.3 10.4

Proton Reaction Analysis for Lithium and Fluorine in MSR Graphite ...................................... Surface Phenomena in Molten Salts

..............................................................................................

11 CHEMISTRY OF FISSION PRODUCT FLUORIDES................................................................................

119 125

11.2

Mass Spectrometry of the Molybdenum Fluorides ........................................................................

129 129 129 130 131 134

11.3

Spectroscopic Studies of Fission Product Fluorides ..................................................................

136

11.4

Preparation of Niobium Pentafluoride ..........................................................................................

137

11.1 Properties of Molybdenum Fluorides ............................................................................................ 11.1.1 Synthesis of MoF, and MoF, .......................................................................................... 11.1.2 Lithium FluoromolybdatesflII) ........................................................................................ 11.1.3 Kinetics of MoF, Behavior in 2LiF.BeF ....................................................................

12. PHYSICAL CHEMISTRY O F MOLTEN SALTS ........................................................................................ 12.1 12.2 12.3

Thermodynamics of LiF-BeF, Melts by EMF Measurements .....................................................

138

Electrolysis of LiF-BeF, Mixtures with a Bismuth Cathode ....................................................

140 141

A Review of Electrical Conductivities in Molten Fluoride Systems ........................................

........................................ A Stirred Reaction Vessel for Molten F~uorides.......................................................................... 12.6 The Chemistry of Silica in Molten LiF-BeF, .............................................................................. 12.7 A Silica Cell and Furnace for Electrochemical Measurements with Fused Fluorides ........ 12.8 Status of the Molten-Salt Chemistry Information Center ............................................................

12.4 12.5

138

Measurement of Specific Conductance in LiF-BeF, (66-34 mole So)

144 146 146 149 149

13 CHEMISTRY OF MOLTEN-SALT REACTOR FUEL REPROCESSING TECHNOLOGY ....................

152

13.1 MSBR Fuel Reprocessing by Reductive Extraction into Molten Bismuth ................................ 13.2 Removal of Structural Metal Fluorides from a Simulated MSRE Fuel Solvent ........................ 13.2.1 Reduction of Structural Metal Fluorides ........................................................................ 13.2.2 Salt Filtration Studies......................................................................................................

152

Protactinium Studies in ,the High-Alpha Molten-Salt Laboratory .............................................. 13.3.1 Protactinium Reduction by Solid Thorium in the Near Absence of Iron .................... 13.3.2 Reduction of Protactinium by Bismuth-Uranium Al.loy ................................................ 13.3.3 Reduction of Protactinium by Bismuth-Thorium Alloys .............................................. 13.3.4 Two-Region Breeder Blanket Composition .................................................................... 13.3.5 Single-Region Fuel Composition ....................................................................................

159 159 159 160 160 161

14 BEHAVIOR OF BF, AND FLUOROBORATE MIXTURES .................................................................... 14.1 Phase Relations in Fluoroborate Systems .................................................................................... 14.1.1 The System NaF-NaBF,KBF,- K F ................................................................................ 14.2 Nonideality of Mixing in Potassium Fluoroborate-Sodium (or Potassium) Fluoride Systems ..........................................................................................................................................

166 166 166

13.3

155 155 157

14.4 14.5

Dissociation Pressure of BF, for the MSRE Substitute Coolant

.............................................

169

14.6

Chemical Thermodynamics of the System NaBF, N a F ................................................................

14.7 14.8

Corrosion of Hastelloy N and Its Constituents in Fluoroborate Melts

170 171

.................................... Compatibility and Immiscibility of Molten Fluorides ..................................................................

/f4 1

167

168 Heat Content of NaBF4-NaF (92.5-7.5 mole So) .......................................................................... The Solubility of Thorium Metal in Lithium Fluoride-Thorium Tetrafluoride Mixtures ........ 168

14.3

171

PART 4 . MOLTEN-SALT IRRADIATION EXPERIMENTS

15 MOLTEN-SALT CONVECTION LOOP IN THE ORR .............................................................................. 174

................................................................................................ Penetration of Fission Products into Graphite and Deposition onto Surfaces ........................ Studies of Surface Wetting of Graphite by Molten Salt................................................................ Design of a Third In-Pile Molten-Salt Loop ................................................................................

175

16 GAMMA IRRADIATION OF FLUOROBORATE ........................................................................................

180

15.1 15.2 15.3 15.4

Isotope Activity Balance (Loop 2)

175 178 178

PART 5 MATERIALS DEVELOPMENT

17 MSRE SURVEILLANCE PROGRAM .......................................................................................................... 17.1 General Comments ............................................................................................................................ 17.2

Examination of Hastelloy N Specimens from MSRE Surveillance Facility ..............................

183 183 184

18 GRAPHITE STUDIES .................................................................................................................................. 18.1 Procurement of Special Grades of Graphite ..................................................................................

188 188

...................................... X-Ray Studies on Graphite ..............................................................................................................

189

18.2 18.3

Porosity Created in Grade AXF Graphite by Oxidation Pretreatment

190

vii

18.4 18.5 18.6 18.7

Gas Impregnation of Graphite with Carbon .................................................................................. Graphite Surface Sealing with Metals ............................................................................................ Graphite Irradiation Program .......................................................................................................... Nondestructive Testing Studies .................................................................................................... 18.7.1 Graphite Ultrasonic Velocity Measurements ................................................................ 18.7.2 Low-Voltage Radiography ................................................................................................

19. HASTELLOY N ............................................................................................................................................ 19.1 Influence of Strain Rate on the Fracture Strain of Hastelloy N ................................................ 19.2 Status of Development of the Modified Alloy .............................................................................. 19.3 Effect of Carbon and Titanium on the Unirradiated Creep-Rupture Properties of 19.4 19.5

Ni-Mo-Cr Alloys ............................................................................................................................ Electrical Resistivity of Titanium-Modified Hastelloy N .......................................................... Electron Microscope Studies of Hastelloy N................................................................................ 19.5.1 Phase Identification Studies in Hastelloy N ................................................................ 19.5.2 Effect of Silicon on Precipitation in Hastelloy N........................................................ 19.5.3 Titanium-Modified Hastelloy N ...................................................................................... 19.5.4 Summary .............................................................................................................................. Diffusion of Titanium in Modified Hastelloy N ............................................................................

19.6 19.7 Measurement of Residual Stresses in Hastelloy N Welds .......................................................... 19.7.1 Experimental Results ...................................................................................................... 19.8 Application of the Narrow-Gap Welding Process to the Joining of Hastelloy N .................... 19.9 Natural Circulation Loops and Test Capsules ............................................................................ 19.9.1 Fuel 'Salts .......................................................................................................................... 19.9.2 Blanket Salts .................................................................................................................... 19.9.3 Coolant Salts .................................................................................................................... 19.10 Forced Circulation Loop ................................................................................................................ 19.11 Oxidation of Hastelloy N ................................................................................................................

20 GRAPHITE-TO-METAL JOINING .............................................................................................................. 20.1 Graphite Brazing Development ...................................................................................................... 20.2 Radiation Stability of Brazing Alloys of Interest for Brazing Graphite .................................. 20.3 Graphite-to-Hastelloy N Transition Joint .................................................................................... 20.3.1 Conceptual Design ............................................................................................................ 20.3.2 Heavy-Metal Alloy Development .................................................................................... 20.4

Nondestructive Testing Evaluation of Graphite-to-Metal Joints

..............................................

21 SUPPORT FOR COMPONENTS DEVELOPMENT PROGRAM .............................................................. 21.1 Remote Welding ................................................................................................................................

191 192 195 1% 1% 197

198 198 201 204 205 206 206 209 212 213 213 215 216 217 218 218 221 221 226 228 231 231

234 235 235 236 238 240 240

viii

PART 6. MOLTEN-SALT PROCESSING AND PREPARATION 22. MEASUREMENT OF DISTRIBUTION COEFFICIENTS IN MOLTENSALT-METAL SYSTEMS ...................................................................................................................................................... 22.1

Extraction of Uranium and Rare Earths from Fuel Salt of Two-Fluid MSBR’s ........................

242

22.2

Extraction of Uranium from Single-Fluid MSBR Fuel..................................................................

243

............................................................................ PROTACTINIUM REMOVAL FROM A SINCLE-FLUID MSBR .............................................................. CONTINUOUS FLUORINATION O F MOLTEN SALT ............................................................................ RELATIVE VOLATILITY MEASUREMENTS BY THE TRANSPIRATION METHOD ........................ DISTILLATION O F MSRE FUEL CARRIER SALT ................................................................................ PROTACTINIUM REMOVAL FROM A TWO-FLUID MSBR .................................................................... RECOVERY O F URANIUM FROM MSRE FUEL SALT BY FLUORINATION .................................... 28.1 Fluorination-Corrosion Study .......................................................................................................... 28.2 Fission Product Behavior - Analytical Assistance Program .................................................... 22.3

23. 24. 25. 26. 27. 28.

241

Experimental Procedure and Lithium “Lqss”

247 248 252 255 258 260 264 264 266

29. MSRE FUEL SALT PROCESSING ............................................................................................................

269

30. PREPARATION O F 7LiF-233UF, FUEL CONCENTRATE FOR THE MSRE .................................... 30.1

Equipment Changes ..........................................................................................................................

270 270

30.2

Equipment and Process Status

......................................................................................................

271

31. DECAY HEAT GENERATION RATE IN A SINGLE-REGION MOLTEN-SALT REACTOR .............. 275

8 L

Introduction The objective of the Molten-Salt Reactor Program

graphite is a new material having high density am small pore size. The fuel salt does not wet the graphite and therefore does not enter the pores, even at pressures well above the operating pressure. Heat produced in the reactor is transferred to a coolant salt i n the primary heat exchanger, and th coolant salt is pumped through a radiator to dissipate the heat to the atmosphere. Design of the MSRE started early in the summer of 1960, and fabrication of equipment began early in 1962. The essential installations were completed, and prenuclear testing was begun in Augu: of 1964. Following prenuclear testing and some modifications, the reactor was taken critical on June 1,1965, and zero-power experiments were completed early in July. After additional modifications, maintenance, and sealing of the containment, operation at a power of 1 Mw began in January 1966. At the 1-Mw power level, trouble was experiencc with plugging of s m a l l ports in control valves in the off-gas system by heavy liquid and varnish-lik organic materials. These materials are believed to be produced from a very small amount of oil tha leaks through a gasketed seal and into the salt i f the tank of the fuel circulating pump. The oil vaporizes and accompanies the gaseous fission'prod ucts and helium cover gas purge into the off-gas system. There the intense beta radiation from th krypton and xenon polymerizes some of the hydrocarbons, and the products plug small openings. This difficulty was overcome by installing a specially designed filter in the off-gas line. Full power, about 7.5 Mw, was reached in May 1966. The plant was operated until the middle of July for about six weeks at full power, when one of the radiator cooling blowers (which were left over from the ANP program) broke up from mechan. ical stress. While new blowers were being procured, an array of graphite and metal surveillance

is the development of nuclear reactors which use

fluid fuels that are solutions of fissile and fertile materials in suitable carrier salts. The program is an outgrowth of the effort begun over 18 years ago in the Aircraft Nuclear Propulsion (ANP) program to make a molten-salt reactor power plant for aircraft. A moltensalt reactor the Aircraft Reactor Experiment - was operated at ORNL in 1954 a s part of the ANP program. Our major goal now is to achieve a thermal breeder reactor that will produce power at low cost while simultaneously conserving and extending the nation's fuel resources. Fuel for this type of reactor would be 233UF4 or 235UF4 dissolved in a salt that is a mixture of L i F and BeF2. The fertile material would be ThF, dissolved in the same salt or i n a separate blanket salt of similar composition. The technology being developed for the breeder is also applicable to advanced converter reactors. A major program activity is the operation of the Molten-Salt Reactor Experiment (MSRE). This reactor w a s built to test t h e types of fuels and materials that would be used in thermal breeder and converter reactors and to provide experience with the operation and maintenance of a molten-salt reactor. The MSRE operates a t 1200'F and a t atmospheric pressure and produces about 7.5 Mw of heat. The initial fuel contains 0.9 mole % UF,, 5 mole % ZrF,, 29 mole % BeF,, and 65 mole % LiF, and the uranium is about 33% 235U. The melting point is 840'F. The fuel circulates through a reactor vessel and an external pump and heat exchange system. All this equipment is constructed of Hastelloy N, a nickel-molybdenum-chromium alloy with exceptional resistance to corrosion by molten fluorides and with high strength at high temperature. The reactor core contains an assembly of graphite moderator bars that are in direct contact with the fuel. The

specimens was taken from the core and examined. Power operation was resumed in October with one blower; then in November the second blower was installed, and full power was again attained. After a shutdown to remove salt that had accidentally gotten into a n off-gas line, the MSRE was operated in December and January at full power for 30 days without interruption. The next power run was begun later in January and was continued for 102 days until terminated to remove a second set of graphite and metal specimens. An additional operating period of 46 days during the summer was interrupted for maintenance work on the samplerenricher when the cable drive mechanism jammed. In September 1967, a run was begun which continued for s i x months until terminated on schedule in March 1968. Power operation during this run had to be interrupted once when the reactor was taken to zero power to repair an electrical short in the sampler-enricher. Completion of this six-month run brings to a close the first phase of MSRE operation, i n which the objective was to demonstrate on a s m a l l scale the attractive features and technical feasibility of these systems for civilian power reactors. We believe this objective h a s been achieved and that the MSRE has shown that molten fluoride reactors can be operated at temperatures above 1200'F without corrosive attack on either the metal or graphite parts of the system, that the fuel is completely stable, that reactor equipment can operate satisfactorily at these conditions, that xenon can be removed rapidly from molten salts, and that, when necessary, the radioactive equipment can b e repaired or replaced. The second phase of MSRE operation will be operation with 3U fuel i n place of 'U. A small facility in the MSRE building will be used to remove the uranium presently in the fuel salt by treatment with gaseous F 2. Highly pure 33Uwill then be added to the present carrier salt, and critical, lowpower, and full-power t e s t s will be performed.

A large part of the Molten-Salt Reactor Program is now being devoted to the requirements of future

molten-salt reactors. Conceptual design studies and evaluations are being made of large breeder reactors, and an increasing amount of work on materials, on the chemistry of fuel and coolant salts, and on processing methods is included i n the research and development program. For several years, most of our work on breeder reactors has been aimed specifically at two-fluid systems i n which graphite tubes are used to separate uranium-bearing fuel salts from thorium-bearing fertile salts. We think attractive reactors of this type can b e developed, but the core designs we have been working on are complex, and several years of experience with a prototype reactor would be required to prove that graphite can serve a s a plumbing material while exposed to high fast-neutron irradiations. As a consequence, a one-fluid breeder h a s been a long-sought goal. Two developments of the past year have established the feasibility of a one-fluid breeder. The first was establishment of the chemical s t e p s in a process which u s e s liquid bismuth to extract protactinium and uranium selectively from a salt that also contains thorium. The second was the recognition that a fertile blanket can be obtained with a salt i n which there is uranium as well as thorium by reducing the graphite-to-fuel ratio in the outer part of the core. Our studies show that a one-fluid, two-region breeder can be built that h a s fuel utilization characteristics comparable to our two-fluid designs, and probably better economics. Since the graphite serves only a s moderator, the one-fluid reactor is more nearly a scaleup of the MSRE. These features have caused us t o change the emphasis of our breeder program from the two-fluid to the one-fluid breeder. Work on both performed during the past s i x months is described in this report, but most of our design and development effort is now directed to the one-fluid system.

t' 1

Summary PART 1. MOLTEN-SALT REACTOR EXPERIMENT

Component Development

The installation and preoperational testing of the off-gas sampler were completed. Four samples of reactor off-gas were removed for xenon isotopic analysis. The isolation chamber and drive assembly that was removed from the sampler-enricher was examined and disassembled in a hot cell. After successful decontamination, the major part of the unit w a s returned for use as a spare. A proximity switch was tested that promises to help prevent cable snarls such as that which rendered the sampler inoperative. A heated carrier utilizing molten babbitt was built t o keep samples hot i n transit to the analytical laboratory. Tests of a pinhole gamma-ray camera for locating radiation sources showed promise. Thermal cycling of a prototype freeze flange was resumed to lend confidence t o predictions of fatigue life. The hot-test facility for the Mark 2 fuel pump was prepared, and the rotary element assembly was nearly finished.

1. MSRE Operations This report period was almost completely occupied with a long run that began in September and was still going at the end of February. When o p erations were first resumed after the fuel sampler mechanism was replaced, some difficulties were encountered with a radiator door and a component cooling pump. Once the long run was under way, however, the only delay was occasioned by a wiring failure in the sampler-enricher in November. The very long run, without the complicating effects of drains and dilutions or fuel additions, afforded opportunity for very close comparison of predicted and observed long-term changes in reactivity. A gradual deviation between observed and computed reactivity of -3.5 x (% W k ) / Mwhr was seen. An experiment on the effect of minor variations i n fuel temperature, system pressure, and fuel salt level on fission g a s stripping extended over a twomonth period. There were several different indications that lower temperature, higher pressure, and lower salt level increased the volume of g a s circulating with the fuel and delayed the stripping of fission gases into the cover gas.

Instruments and Controls

Only a moderate amount of maintenance on the reactor systems was required. Five of the fifteen relays in the rod scram coincidence circuit failed, leading t o a decision to replace them with a different type of relay. Water leaks in nuclear chamber cables continued to give trouble, and two fission chambers and an ionization chamber had to be replaced. A second failure occurred in a position synchro on one of the control rods. A few minor modifications were made, and some design required for salt processing was completed. Equipment and procedures for analysis of neutron noise spectra were tested and used to obtain data pertinent to reactor operation under various conditions.

Component performance was generally quite good.

Two bearings on the main blowers were replaced during a period at low power i n January, and the standby component cooling pump was not operable during most of the run. However, neither of these delayed the operation. The wiring failure in the sampler-enricher interrupted high-power operation for nine days, but it was not necessary to drain the fuel. During the six-month period, the reactor was critical 89% of the time, and the integrated power increased to 8581 equivalent full-power hours.

xii The adequacy of the rod servo control system with 3U fuel was investigated. No need for modification appeared.

4. MSRE Reactor Analysis Reactor physics studies in support of planned operation of the MSRE with 233Uwere extended to help evaluate the nuclear safety of the system. A s was the c a s e with the present 235Ufuel loading, the potentially most severe nuclear excursions were associated with two hypothetical incidents: reactivity addition by the simultaneous withdrawal of the three control rods, and the gradual separation of uranium from the mainstream of circulating salt, followed by its sudden resuspension and rapid return to the core in concentrated form. Both incidents were analyzed with the aid of a digital kinetics program supplemented by analog simulation. In the first incident, we found that rod scram initiated by the action of the safety system would limit the temperature-pressure rises in the nuclear excursion to inconsequential proportions. This same conclusion would apply in the case of the second incident, unless the abnormal reactivity loss represented by the uranium separation became larger than about 0.95% 6k/k. This abnormal reactivity loss would be easily detectable by routine computer monitoring of the reactivity balance, which should reveal any anomaly a s large a s 0.1% 6k/k.

PART 2. MSBR DESIGN AND DEVELOPMENT 5. Design Developments in fuel processing and reactor core configurations that enable a one-fluid molten-salt reactor to be an efficient breeder have led us to set aside our work on the design of a two-fluid breeder reactor and to undertake studies of a onefluid breeder. The fuel for the one-fluid breeder consists of fissile uranium and fertile thorium as tetrafluorides dissolved in a lithium fluorideberyllium fluoride carrier salt. We have been working on the design of a 2000 Mw (electrical) reactor in which the fuel circulates upward around vertical graphite bars in a reactor vessel and then completes the circuit through four pumps and four heat exchangers. A s in our designs for two-fluid reactors, the reactor vessel, heat exchangers, and

pumps are installed in a heated and shielded cell. Steam generating equipment is installed in cells adjacent to the reactor cell, and sodium fluoroborate salt circulates between the primary heat exchangers and the steam generators to transfer the fission heat to supercritical steam. Also communicating with the reactor cell are a fuel processing cell, an off-gas disposal cell, and a drain tank cell for storing fuel salt when the reactor is drained. Basic piping layouts have been made for the fuel and coolant salt systems. A s y s t e m has been conceived for cooling the drain tank by t h e r m a l convection of sodium fluoroborate salt between coils in the drain tank and air-cooled coils in a naturaldraft chimney outside the reactor building.

6. Reactor Physics Neutronic calculations of a single-fluid moltensalt breeder reactor, with fissile and fertile materials carried in the same salt stream, have shown that breeding performance comparable with that of a two-fluid MSBR can be achieved, provided the core is properly designed to minimize neutron leakage. Breeding ratios of 1.05 to 1.07, fuel specific power of 2 to 2.5 Mw (thermal)/kg, and annual fuel yields of about 5%/year appear to be attainable with fuel processing rates which probably imply fuelcycle costs less than 0.5 mill/kwhr (electrical). Such a reactor would have a small negative overall isothermal temperature coefficient of reactivity and a substantially negative prompt coefficient, that is, -3 x lo-â&#x20AC;&#x2122; 8k/OC, associated with a change in salt temperature alone.

7. Systems and Components Development The analytical model used to compute the steadystate migration of noble gases to the graphite and other sinks in an MSBR was extended to study the effects of graphite surface area and surface coatings on the xenon poison fraction. It was shown that an 8 4 1 coating of material with a diffusion ft2/hr would reduce the poison coefficient of fraction from 2 2 % to the target value of 0.5%. The alterations to and checkout of the test facility for operation with sodium fluoroborate were completed, and a flush charge of 900 lb of sodium fluoroborate was added to the sump. The flush charge is intended to remove residues of the Li-Be

U i

xiii

salt previously circulated in the loop and will be replaced after several days of operation. The loop will be operated to study the pumping characteristics of the salt and the problems associated with control and monitoring of the salt composition. A preliminary layout was made of a fuel salt pump configuration applicable to the single-fluid molten-salt breeder concept. Preliminary layouts were made of a molten-salt pump test facility and a molten-salt bearing tester suitable for the MSBE salt pumps. Work was initiated on specifications for the MSBE fuel salt pump. The rotor-dynamics feasibility investigation was completed for the long-shaft pump configuration that requires a molten-salt-lubricated bearing and was specified for our two-fluid molten-salt breeder concept. Specimens of cermet hard coatings which are plasmasprayed on Hastelloy N substrate were received, and they are being evaluated a s candidate materials for molten-salt bearings. Studies of the problems associated with the maintenance of the MSBR concepts were started. The initial effort is to evaluate the problems caused by scaleup of the general maintenance system used at the MSRE. The three specific areas under active study are (1) the application of the portable maintenance shield concept to the various cells of the MSBR, (2) remote welding for vessel entry and component replacement, and (3) replacement of the graphite moderator elements of the core on a routine basis. The early development of remotely operated vessel and pipe closures is important to the program, and the study of remote welding a s an initial approach is being expedited.

8. MSBR Instrumentation

and Controls

Analog computer studies of the dynamic behavior of the Molten-Salt Breeder Reactor were begun. A model of the two-fluid system was developed, and several transient cases were investigated to determine uncontrolled core behavior during reactivity changes, fuel salt flow reductions, and simulated load losses. The results obtained lead us to the tentative conclusion that the plant would be inherently load following at the expense of modest temperature changes. They also indicate that it should be quite easy to accommodate rather large load changes using a control system to maintain some desired temperature condition.

A model of the steam generator was also developed. The complexity of this model exhausts the capacity of the ORNL analog computer, leaving no equipment available for the simulation of the rest of the plant. Alternative methods for studying the dynamics of the entire Molten-Salt Breeder Reactor power plant are being studied, with the most promising approach being one using a simpler linearized model of the steam generator on a hybrid computer.

PART 3. CHEMISTRY 9.

Chemistry of the

MSRE

Chemical behavior in the salt, gas, oil, and water systems has been under continuous surveillance since the beginning of MSRE operations. The results continued t o show excellent materials compatibility. There was, however, an unexplained difference of about 0.02 wt % between the chemical analyses for uranium and the uranium concentration computed from operational data. The chromium concentration reached a steady-state value of 85 ppm; this represents an insignificant amount of corrosion, and the view has been advanced that much of the chromium content of the fuel came from the drain tanks. Mass spectrometric analyses of the uranium isotope distribution in the fuel promised to be useful in rating the reactor output.

10.

Fission Product Behavior

The fate of fission products in the reactor was established in considerable detail. The only unusual behavior continued to be that of the more noble metals such a s Mo, Ru, Te, and Nb. These left the fuel not only by depositing on walls but also apparently a s a smoke that was carried away in the gas phase. A new method of examining the concentration profile of fission products in graphite from the MSRE confirmed the results from the older method. Concentrations of Li and F in the same graphite were obtained by a highly sensitive method involving proton bombardment. Traces of these elements penetrated deeply, but the amounts found, though unexplained, were not chemically significant. Samples of fuel from the MSRE were studied in the hot cell; the same type of emanation of fission

XiV

products was found a s that previously encountered i n the MSRE pump bowl. This emanation was identified a s having the form of aerosols, and pictures of the particles were obtained by electron microscopy. Tests were initiated to explore the nature of colloids in molten salts.

11. Chemistry of Fission Product Fluorides Because of the peculiar behavior of fission products like molybdenum in the MSRE, an exploration of the chemical behavior of molybdenum fluorides and niobium fluorides was continued. Methods of preparing and characterizing MoF,, MoF,, and other compounds were perfected. The rate of autooxidation and -reduction of MoF, in LiF-BeF, melts was measured. A mass spectrometric investigation of molybdenum vaporization was extended to higher pressures where the vapor-phase species were measured at varying temperatures. The dimer MozF 1o was encountered. Spectrophotometric studies were attempted for MoF, and NbF, in LiF-BeF,, and NbF, was synt hesized.

A Molten-Salt Chemistry Information Center has been established and is operating successfully.

13. Chemistry of Molten-Salt Reactor Fuel Reprocessing Technology Laboratory-scale studies of the reductive extraction of protactinium, uranium, and rare earths continued to demonstrate the feasibility of reprocessing two-region molten-salt breeders. As applied to one-fluid breeders, the results were less favorable. The difficulty is that thorium tends to reduce before the rare earths. The fluorination step for removing uranium in reprocessing fuel leads t o the accumulation of a large amount of corrosion products in the form of structural metal fluorides. Modes of reducing and removing these products have been studied. Experiments on the reductive extraction of protactinium were carried out in a facility that allowed realistic amounts of alpha-active 2 3 'Pa to be followed. The reductive extractions usually employed bismuth alloys a s the reducing and extracting m e dium.

14. Behavior of BF, and Fluoroborate Mixtures 12. Physical Chemistry of Molten Salts Electrochemical measurements have revealed the solution thermodynamics of LiF-BeF m e l t s over the whole composition range. Experiments on the electrolysis of LiF-BeF mixtures with a bismuth cathode were initiated; these were of interest in connection with extractive reprocessing of breeder melts. An improved method for balancing cell impedance was employed in measuring the electrical conductivity of molten fluorides. The specific conductivity of LiF-BeF, (66-34mole So) rises with temperature and is about 1.6 mhos/cm a t 510OC. In the past the lack of vigorous agitation has seriously handicapped work on heterogeneous equilibria in molten fluorides; an improved stirred vessel was therefore designed and constructed. The inleakage of air to this vessel was scarcely detectable. Such a vessel was used in unraveling the chemistry of silica in LiF-BeF,. The use of an overpressure of SiF, permitted electrochemical cells involving LiF-BeF, to be constructed of silica.

A strong candidate a s a secondary coolant was the NaF-KF-BF, (47-5-48mole %) mixture, which had a melting point of 350 f 5OC. Experimental phase diagrams for fluoroborates were revised t o be consistent with the heats of fusion of NaF and KF. Using a copper block calorimeter, the heat of fusion of the NaBF,-NaF eutectic was found to be 31 cal/g, and the heat of transition was 14.7 cal/g. The heat capacity of the liquid was 0.36 cal g-' oc-

The pressure of BF, above fluoroborate mixtures was measured; from these measurements the chemical thermodynamics of the system NaF-NaBF, were elucidated.

PART 4. MOLTEN-SALT IRRADIATION EXPERIMENTS

15. Molten-Salt Convection Loop in the ORR During this report period we have completed additional analyses of the fuel salt, graphite, and

m e t a l in contact with fissioning fuel salt and cover g a s from the second in-pile molten-salt convection loop. These data permitted completion of the isotope activity balance. Molybdenum, tellurium, ruthenium, and niobium are almost entirely departed from the salt. These elements showed no dominant preference for graphite or metal but seemed to deposit on whatever surface was available. Short-lived noble gases appeared to have diffused appreciably into the graphite, as shown by the presence of daughter isotopes such as *'St, 14%a, and others. However, t h e majof proportions of these, and almost all of the other alkali, alkaline-earth, and rareearth isotopes (all of which form relatively stable, nonvolatile fluorides), were found in the salt. Design of a third molten-salt convection loop h a s been completed. Design features include the u s e of a modified Hastelloy N containing titanium for improved resistance to radiation-induced hightemperature embrittlement, graphite surveillance specimens in the core section to permit better postirradiation determination of the interaction of graphite and salt, and gas adsorption traps to help identify the gas-borne fission product species.

16. t

Gamma Irradiation of Fluoroborate

Gamma irradiation experiments with sodium fluoroborate salt are being conducted in the central channel of spent HFIR fuel elements to determine the effects, i f any, of such irradiations on the sodium fluoroborate and its compatibility with Hastelloy N. One gamma irradiation experiment and an unirradiated control have been completed to date. Both experiments used an NaF-NaBF4 eutectic mixture (normally 8-92mole %) in a Hastelloy N capsule containing a Hastelloy N corrosion test specimen. The unirradiated experiment was operated for 840 hr (mostly at 6OOOC). The irradiation experiment was operated for 533 hr in spent fuel element 34 (-8 x lo' r/hr at the start of the experiment, compared with an estimated 5 x lo' r/hr in the MSRE heat exchanger), and the salt accumulated an estimated absorbed dose of "1 x ev/g. The gas space i n the capsule was maintained at a nominal temperature of 600OC. We were not able to monitor salt temperatures in the lower part of the capsule after thermocouples in this region were lost early in the experiment. However, it is estimated that

at least part of the NaF-NaBF4 salt in the capsule remained frozen (melting point -38OOC) for most of the run, since water entered the container can and soaked the magnesia insulating pad under the capsule. Salt from the irradiated capsule was discolored, while that from the unirradiated capsule was entirely white except for some s m a l l green crystals, identified as Na3CrF,, found near metal surfaces. Hastelloy N coupons exposed in the irradiated and unirradiated tests exhibited negligible attack. Analysis of residual gas from the irradiated and unirradiated tests showed no trace of B F 3. A second gamma irradiation capsule assembly h a s been placed in operation, and we plan t o irradiate this experiment to a higher total dose than the first experiment.

PART 5. MATERIALS DEVELOPMENT

17. MSRE Surveillance Program We have removed two sets of surveillance specimens from the core of the MSRE and one set of metal specimens from outside the core. Metallographic studies did not reveal any evidence of corrosion of the graphite and Hastelloy N by the fluoride salt environment. The Hastelloy specimens from outside the core were exposed to the cell environment and had an oxide film of about 0.002 in. Testing showed that the mechanical properties of the Hastelloy N changed during exposure, but these changes are equivalent to those noted for m a t e r i a l s exposed to equivalent fluences i n the ORR. The reduction in the ductility seems to be about saturated for the vessel, and the properties are more than adequate for operation under design conditions.

18.

Graphite Studies

We continue to obtain samples of several grades of graphite that are potentially applicable for u s e in molten-salt breeder reactors. Some graphites have been obtained from the Y-12Plant for which we have detailed information on the raw materials and processing variables. These materials will be included in all phases of our graphite work. We are evaluating several techniques for surface sealing graphite to obtain a low surface perme-

xvi ability. Carbon coatings have been applied previously in a fluidized bed, and two new methods are under study. One involves passing the carbonaceous gas through the wall of a graphite tube, with the location of deposition being controlled by a radial temperature gradient. A second method utilizes a cyclic technique in which the sample environment is cycled between hydrocarbon mixtures and vacuum. Samples coated by both techniques are being evaluated. Sealing the graphite by the deposition of molybdenum is complicated by the need to minimize the quantity of m e t a l present in the core of the reactor. Our work shows that the graphite can be sealed adequately by about 1 mil of molybdenum if the graphite substrate has a small, uniform pore size of about 1 p. We have designed an experimental assembly for irradiating graphite in the HFIR. This facility will allow us to irradiate small graphite cylinders to a fluence of 4 x neutrons c m - 2 year-' a t 700OC. We have run two short experiments in the HFIR to check the design, have made the necessary changes, and presently have two assemblies in the HFIR. Changes in the specimen dimensions will be measured, and techniques have been developed to measure the velocity of sound in the graphite to determine changes in the elastic modulus.

19. Hastelloy N The radiation damage in Hastelloy N continues to receive considerable attention. Although the strength at elevated temperatures is reduced by irradiation, the reduction in the fracture strain is of primary concern. W e have found that the fracture strain is dependent upon the strain rate, showing a minimum fracture strain of about 0.5% at a strain rate of O.l%/hr. The titanium-modified Hastelloy shows the same behavior, but the minimum fracture strain is about 3%compared with 0.5% for the standard alloy. W e are continuing the development of the titanium-modified Hastelloy N. We have obtained from commercial vendors about twenty-five 100-lb melts and one 5000-lb m e l t of the modified composition. The yield of fabricated shapes from the 5000-lb m e l t was about 50%, a respectable value for standard m i l l practice. Studies on several s m a l l m e l t s show that both carbon and titanium contribute to the strength of the modified alloy in the unirradiated condition. The object of these studies is to

determine the titanium and carbon contents to give optimum properties. Diffusion measurements have been made to estimate how rapidly titanium can be removed from the alloy by corrosion. The results indicate that titanium diffuses less rapidly than chromium and should not cause a significant increase in the corrosion rate. Electron microscopy of standard Hastelloy N h a s shown that the large precipitates are basically carbides of the M,C type and that they contain several percent silicon. The segregation of silicon to the precipitates probably accounts for incipient melting and weld cracking in the alfoy. The silicon specification has been lowered to 0.1%maximum to minimize these adverse effects. The residual stresses that develop during welding can cause subsequent cracking of the weldment during service. We have developed a technique for measuring these stresses in welded plates that will allow us to determine welding parameters and postweld heat treatments to minimize the residual stresses. Measurements on Hastelloy N show that the maximum residual stress is reduced from about 61,000 to 5000 psi by a postweld anneal of 6 hr at 8 7 1 T . W e have several thermal convection loops in operation to investigate the compatibility of structural metals with various fluoride salts. A loop constructed of type 304L stainless steel and containing a modified fuel salt h a s operated for 40,510 hr at a peak temperature of 677OC without incident. Removable tabs of stainless steel indicate that the corrosion rate is about 2 mils/year for the first several hundred hours and then decreases, consistent with the behavior expected for a diffusion-controlled process. Analysis of the salt indicates that only the chromium level is increasing. A Hastelloy N loop containing the same salt has operated without incident at a peak temperature of 704OC for 51,810 hr. Two loops constructed of Hastelloy N and containing the potential coolant salt NaBF ,-NaF (92-8 mole %) have operated for about 3000 hr a t a peak temperature of 607OC. These loops are constructed so that samples can be removed for weighing and salt samples taken during loop operation. Chemical analysis of the salt indicates that the chromium and iron levels are both increasing at a rate consistent with a diffusion-controlled corrosion process. Weight changes of the removable specimens indicate a maximum corrosion rate of a few tenths of a mil

LJ i

xvii

u *

per year. One of these loops contains specimens of modified Hastelloy N (contains no iron), and it was found that this material corroded at a lower rate. Several other loops have been started very recently t o investigate the compatibilities of other salts and Hastelloy N. A forced-convection loop is in the final stages of construction to investigate the compatibility of Hastelloy N and NaBF,-NaF (92-8 mole %) under realistic service conditions. This loop will have a liquid velocity of 7 fps, a BF, control system, and capabilities for sampling the salt during operation. Studies of the oxidation of Hastelloy N indicate that silicon plays a critical role i n the scaling resistance. Silicon is usually present in air-melted heats at a level of about 0.6%, and we would like t o reduce t h i s further t o improve the weldability. Levels of less than 0.1% silicon can be obtained routinely by vacuum melting. However, the heats with lower silicon have higher scaling rates. The further addition of titanium t o the low-silicon materials does not cause any appreciable changes in the scaling resistance. Although we have established these trends, the oxidation resistance i n all cases is quite acceptable at 760OC.

20.

Graphite-to-Metal Joining

We have been able t o join graphite to Hastelloy N in small sizes (1 in. in diameter) by direct brazing with copper or with an Ni-Pd-Cr alloy. However, the matching of surfaces to obtain good joints in large pipe sizes requires extremely good control over dimensional tolerances, so we have not been able to make the direct joint consistently. However, joints can be made repeatedly when a molybdenum transition piece is used between the Hastelloy N and the graphite. A nondestructive testing technique has been developed for determining whether all areas of the joint are bonded. We have investigated the effects of irradiation on several brazing alloys that are potentially suitable for this joint including Cu, Ni-Pd-Cr, Cu-Ni-Cr-Be, and Cu-Ni-Ta-Be. The strength and ductility changes indicate that all the alloys have acceptable properties after irradiation to fluences of the order of lo2' neutrons/cm '. Another approach for joining Hastelloy N t o graphite is a transition joint i n which thin layers of alloys having slightly different coefficients of

thermal expansion a r e brazed together with a very ductile brazing alloy such as copper. Suitable tungsten-base alloys presently exist for the joint, and the development of molybdenum-base alloys seems imminent.

21.

Support for Components Development Program

We are setting up the equipment to evaluate several welding processes that are potentially applicable t o the remote joining of Hastelloy N. The a i m of this study is t o choose a welding process that can be further developed for making joints remotely in a high radiation field.

PART 6. MOLTEN-SALT PROCESSING AND PREPARATION

22.

Measurement of Distribution Coefficients in Molten-Salt-Metal Systems

Distribution of uranium, thorium, and rare earths between selected molten fluoride salts and lithiumbismuth solutions is being studied in support of reductive extraction processes for MSR fuels. Data obtained so far show that uranium can easily b e preferentially extracted from the fuel salt of a twofluid or a single-fluid MSBR. Separation of the uranium from rare earths is good (separation factors of at least lo3) with each type of fuel. In the processing of single-fluid MSBR fuel, uranium can b e separated from thorium by a factor of at least 10'. Rare earths can probably be separated from thorium, but more data a r e required t o establish the optimum conditions.

23.

Protactinium Removal from a Single-Fluid MSBR

The steady-state performance of a system for isolating protactinium from a single-fluid MSBR by reductive extraction was examined. Calculations based on available tentative equilibrium data indicate that adequate protactinium isolation can be obtained with a system consisting of a n extraction column equivalent t o about 12 ideal stages, a n electrolytic oxidizer-reducer, and a protactinium decay tank having a volume of about 400 ft3. Required salt and metal flow rates are about 2 gpm.

xviii The system was found to be quite sensitive to minor variations in operating conditions, and several methods for stabilizing the system have been explored. Details will be revised a s more accurate data are available, but qualitative conclusions are believed valid.

24.

Continuous Fluorination of Molten Salt

Studies are under way on protecting a continuous fluorinator from corrosion by freezing a layer of salt on the vessel wall. Operability of such a syst e m was demonstrated by countercurrently contacting molten salt and an inert gas in the presence of a frozen layer of salt in a 5-in.-diam, 8-ft-high column. An internal heat source consisting of Calrod heaters in a t-im-diam pipe along the center line of the system was used to simulate the volume heat source provided by fission product decay in the molten salt. Frozen wall thicknesses and temperature profiles in the frozen salt were in general agreement with values predicted by relations based on radial heat transfer from a volume heat source. Frozen wall thickness ranged from 0.3 to 0.8 in., depending on experimental conditions.

25.

Relative V o l a t i l i t y Measurements by the Transpiration Method

Relative volatilities for several solutes were measured in the temperature range from 900 to 105OOC using LiF-BeF2 (90-10 mole %) a s the solvent. Values obtained with respect to L i F at 1000째C were about 0.04 for UF,, 25 for RbF, 95 for CsF, and about 2 for ZrF, (when present in solution at a concentration of 0.083 mole %).

26.

Distillation of

MSRE f u e l

Carrier Salt

Study of low-pressure distillation of MSRE carrier s a l t is under way in equipment which includes a @-liter feed tank, a 12-liter still, a condenser, and a 48-liter condensate receiver. Two runs were made using a still pot temperature of about 1000째C and a condenser pressure of 0.06 to 2 mm Hg; the total salt volume distilled during the runs was about 60 liters. With the lower condenser pressures, salt distillation rates of 1.2 and 1.5 ft day-' *'tf were observed at still pot temperatures of 990 and 1005OC respectively.

27.

Protactinium Removal from a Two-Fluid

MSBR

Protactinium removal processes based on reductive extraction using liquid bismuth containing thorium were analyzed to evaluate feasibility of removing protactinium from the fertile stream of a two-fluid MSBR. Calculations indicate that adequate protactinium removal can be achieved with a system consisting of an extraction column equivalent to about three ideal stages, an electrolytic oxidizer-reducer, a protactinium decay tank having a volume of about 400 ft', and a fluorinator for removal of uranium from the decay tank. Required fertile salt and metal flow rates are about 25 and 0.1 gpm, respectively, for a 1000 Mw (electrical) reactor.

28.

Recovery of Uranium from MSRE F u e l Salt by Fluorination

Small-scale tests are being made with simulated MSRE fuel s a l t to determine the effects of temperature, fluorine concentration, and fluorine flow rate on the rate of uranium volatilization and the rate of corrosion of Hastelloy N. Preliminary results indicate that uranium can be readily removed from the salt with fluorine at 450 to 50O0C; however, corrosion rates of up to 0.5 mil/hr can be expected during the fluorination period. Fluorination is also being considered for use a s the first step in a precise method for analyzing for uranium in MSRE fuel salt. The UF, produced would be collected and decontaminated using the NaF sorption-desorption method; then the UF ,-NaF complex could be transferred to B low-level tadiation area for precision coulometric analysis of the uranium. Decontamination factors for lo3Ru, '%b, and 13'Te have been greater than 10'; however, attempts to achieve an adequate decontamination factor (greater than lo3) for iodine have been unsuccessful so far.

29. MSRE

F u e l Salt Processing

Modifications to the MSRE fuel processing facility have been completed except for the installation of the salt filter. Processing plans call for fluorination, reduction, and filtration of both flush and fuel salts.

xi x :

30.

Preparation of 7LiF-233UF, Fuel Concentrate for the MSRE

Refueling and operating the MSRE with 233U fuel in 1968 will require approximately 39.5 kg of 91.4%enriched 233U as 'LiF-233UF, (73-27 mole %) eutectic salt. This fuel concentrate will be prepared i n cell G of the TURF building because of the radiation from the high 232U (222 ppm) content of the 233U. Engineering design and installation of the process equipment i n TURF a r e essentially completed. Additional equipment has been designed and built for drilling holes in the enrichment caps u l e s and for packaging the bulk charge of salt in nine small salt cans. A shakedown run using 238u03 was started January 15, 1968. By the end of February, 90 to 95% of the uranium had been converted to UF, by treatment with H,-HF. Progress of the reaction h a s been difficult t o follow in the remote facility, and the conversion is slower than expected, probably because of scaleup probl e m s in the production equipment. Equipment changes to eliminate operating probl e m s encountered in the shakedown run are almost compl et ed .

. w

31.

Decay Heat Generation Rate in a

Single-Region Molten-Salt Reactor

Studies of heat generation by gross fission products and 233Pain a one-region 2000 Mw (electrical) MSR show that a t equilibrium with continuous processing these components are generating 289.4 and 0.74 Mw, respectively, in the fuel stream. When noble gases are sparged from the circulating fuel loop, the fission product decay heat decreases to 257 Mw; if, in addition t o sparging, the noble metals are removed by reaction with reactor surfaces, the rate is 255.8 l'vlw. The value above for 233Padecay heat corresponds t o a fuel stream concentration of 0.256 g of 233Paper liter, the equilibrium quantity when 233Pais removed on a 3-day processing cycle. The fission product processing cycle time is 38 days. This processing scheme requites decay storage of 190.5 kg of 233Pa in the processing plant. At equilibrium, this 233Pa produces 9.7 Mw of heat. Calculations of the heat-generation rate at t i m e s after reactor shutdown show that a s much as 33% less heat is generated in fuel from which both noble gases and noble metals have been mnoved. The difference between this rate and the gross rate is not always this large, but i t does average about 20% lower over the first year.

. \

II I 1

P. N. Haubenreich

The six-month period reported here was remarkable for an unprecedented run that began i n September and was still going at the end of February, after more than five months. This run, coupled with the very successful operation in the preceding report period, successfully completed the phase of the experimental program whose principal objective

was demonstration of reliability i n sustained operation. The first part of this report describes the experience with operation of the MSRE, development directly connected with the reactor, and analysis of the planned operation with 233Ufuel.

1. MSRE Operations P. N. Haubenreich

1.1 CHRONOLOGICAL ACCOUNT OF OPERATIONS AND MAINTENANCE R. J. R. P. T. A.

Blumberg L. Crowley H. Guymon H. Harley L. Hudson I. Krakoviak

R. C. M. H. R. B.

B. Lindauer K. McGlothlan Richardson C. Roller C. Steffy, Jr. H. Webster

At the beginning of this report period, the reactor was down because of trouble with the fuel samplerenricher. The sample latch had been retrieved from the pump bowl, but the drive unit and isolation chamber had not yet been replaced. This job was completed during the first week i n September while the salt systems were being preheated for the resumption of operation. The first operation i n run 13 was the circulation of flush salt for six days to permit testing of the

ccd

'MSR Program Semiann. Progr. Rept. Aug. 31, 1967, ORNL-4191, p. 15.

new sampler mechanism. Also during this t i m e the coolant salt system was filled, drained for repairs on the radiator door lifting mechanism, and again refilled. The fuel loop was filled with fuel, and full-power operation started on Friday, September 15. During the weekend, an oil leak shut down a component cooling pump. Operation continued on the second pump, but rather than start off what was expected to be a long run without a standby component cooling pump, we decided to shut down and repair the ailing unit. Repairs took only two days, and on September 20 the fuel system was refilled to start run 14. It was more than six months before the fuel was again drained. Plans were to operate the reactor at high power for several months i n connection with several studies. These included neutron irradiation of Hastelloy N specimens i n the core to higher fluences than this alloy had ever before received, measurement of uranium isotopic changes over a period of substantial burnup to provide information on cross sections, observation of long-term reac-

2 ORNL-VWG 68-3938

POWER

I240

REACTOR OUTLET TEMPERATURE

1200

I160

NOMINAL FUEL PUMP BOWL PRESSURE

-m

: : 6

N i967

Fig. 1.1.

(

1968

Outline of MSRE Operations, September 1967 to February 1968.

tivity behavior as fission products and plutonium built in, and continued studies on fission product distributions. The power history can b e seen in' Fig. 1.1. For the first two months of power operation, the experimental objectives were pursued practically without incident. On October 8 and 9, the power was lowered to 6 Mw to permit a low-temperature, low-salt-level experiment on g a s ingestion into the circulating fuel. On October 20 the power was down briefly after an area power failure. Three days at 10 kw, October 23 to 26, allowed the 13'Xe to clear out of the fuel so that a reference measurement of the reactivity balance could b e made. The only equipment problem was an oil leak that shut down a component cooling pump, but operation continued without interruption on the standby unit. Power operation was interrupted in mid-November by trouble with the wiring in the fuel sampler-enricher. Electrical leads between the outer and inner containment boxes shorted out, leaving a sample stranded i n the tube. T h e reactor power was

lowered to 10 kw while repairs were made. A containment tent was erected, and a hole was cut i n the outer box. This exposed the fault in the cable, so it was not necessary to drain the fuel and go into the inner box. Repairs were made, and tests showed all circuits were operable except one nonessential position switch. While the power was down for the sampler repair, an abnormal pressure drop developed in the fuel off-gas line near the pump bowl. The cause was not known, but to plinimize the chance of salt m i s t causing a worse blockage while the pressure was being kept low for the sampler work, the fuel pump was shut off. After two days, repairs permitted the sample to be retrieved and the isolation valves t o be closed. The restriction i n the g a s line was then blown clear, and nuclear operation was resumed. For about 15 hr after the reactor was returned to full power, indications were that the off-gas restriction was causing some flow to bypass through the overflow tank. Then the pressure drop decreased and was not detectable for the next two months.

! I;

From the beginning of run 14, the xenon effect at full power had been about 0.32% 6k/k - about what i t was i n early power operation but up measurably from the 0.27% 6k/k observed i n runs 11 and 12. The reason for this small shift was not apparent, but there had been indications that the xenon stripping was affected to a slight degree by fuel temperature, liquid level i n the pump bowl, and system pressure. Therefore an investigation was started into the effects of minor changes in these variables on the stripping of xenon and other fission gases from the fuel. In order to lower the fuel temperature without bringing the coolant salt below 1000째F at the radiator outlet, the power was first reduced t o 5 Mw. As can be seen from Fig. 1.1, the experimental operation at various temperatures and pressures extended over the two months from the middle of December t o the middle of February. Beginning on January 22, the reactor power was held at 10 kw for eight days while the temperature and salt level were varied to observe reactivity effects in the absence of xenon. Small changes were observed, reflecting variations i n the amount of bubbles circulating with the fuel salt (around 0.1 to 0.2 vol % normally). Early in the 10-kw operation the main blower bearings were inspected, and two were replaced because of damaged balls. The filter in the coolant off-gas line that had plugged a few weeks earlier was also replaced while the power was down. After 18 more days of 5-Mw operation, which completed the planned experiments on gas stripping, the reactor was.returned t o full power. Analysis of the data suggested that the system gas stripping may have changed characteristics during the ex-

Table

periments. Therefore, on February 29, the power was again reduced t o 5 Mw for a recheck of the xenon under conditions outwardly the same as those that had been tested in December. While reactor operations were proceeding, preparations were being made for the shutdown scheduled to begin late i n March. These included preparing the chemical processing system for fluorination of the salt t o remove the uranium and designing and building equipment for adding 233Uenriching salt t o the fuel carrier left after the fluorination. Operation was continuing at the end of the report period, more than five months after the startup in September. Statistics for the six-month period and totals at the end are given in Table 1.1.

1.2 OPERATIONS ANALYSIS 1.2.1

Reactivity Balance

J. R. Engel The continuous operation for over five months i n t h i s period afforded an unprecedented opportunity to study long-term changes in reactivity. The reactivity balance under the simplest conditions (low power and no xenon) was checked five times over a period of substantial fuel burnup without any complications of fuel additions or intervening fuel loop drains and flushes. In addition, there was t i m e to vary operating conditions t o observe effects on xenon removal.

1.1. Some MSRE Operating Statistics September 1967-February 1968

Total Through Feb. 29, 1968

Critical time, hr

3891 (8%)

10,909

Integrated power, Mwhr

21,823

62,130

Equivalent full-power hours

3014 (6%)

8581

Fuel loop

4054 (93%)

14,415

Coolant loop

4170 (95%)

16,229

Salt circulation, hr

Effects of Gas in Fuel Salt.

- Prior to the cur-

rent run (run 14) the reactor had operated outside of a relatively narrow range of operating conditions for only brief periods of time. The normal conditions were 1210째F at the reactor outlet, 5 psighelium overpressure at the fuel pump, and a narrow range of fuel s a l t levels i n the pump bowl. Previous deviations from these conditions showed that at least temperature and pressure affected the value of the residual term in the reactivity balance. However, the effects had not been clearly defined. During run 14, we performed a series of tests in an effort to evaluate the effects of fuel system temperature and overpressure and fuel pump level on residual reactivity. In these tests the reactor outlet temperature was varied between 1180 and 1225OF and the overpressure between 3 and 9 psig. Since the fuel pump level is restricted by the pump hydraulic performance, only the normal variation in fuel pump level was allowed. This study was performed at a reactor power of 5 Mw to provide the required latitude for the temperature changes (see Fig. 1.1). Small but significant variations in reactivity were observed. The major part of the variations was caused by changes i n the '35Xe poisoning in the reactor, apparently induced by variations i n the effectiveness of g a s stripping caused by the parameter changes. Table 1.2 presents a summary of the net observed 13'Xe poisoning terms at the various conditions. In general, the xenon poisoning increases with decreasing temperature and increasing overpressure. However, the pressure effect practically disappears at the higher temperatures. Conversely, the temperature effect is smaller at the lower pressures. The effect of fuel pump level is much less pronounced than the temperature and pressure effects,but decreasing level leads to higher xenon poisoning for the range of levels investigated. The details of the changes in '35Xe poison were obtained from the reactivity balance. In addition, several samples of the fuel off-gas were obtained under various operating conditions, so that the 13aXe/'34Xe ratio might be measured to verify that the apparent xenon poisoning was in fact due

2There is some evidence to suggest lower xenon poisoning at very low fuel-pump levels, where the helium void fraction in the circulating loop increases substantially.

to 13'Xe. Preliminary results are i n substantial agreement with the poisoning measured by the reactivity balance. Aside from the variations i n xenon poisoning, there was other independent evidence which clearly indicated changes in the effectiveness of removal of gaseous fission products t o the reactor off-gas system. Poorer removal of the gases coincided with the higher poison levels. The temperatures of the off-gas holdup volume in the reactor cell 6-522)and the particle trap are sensitive indicators of the fission product concentration i n the off-gas stream. At constant system pressure, where no changes i n trdnsit (decay) time are encountered, this concentration directly reflects the effectiveness of the fission product stripping. The observed temperatures at L-522 and near the coarse filtering medium at the entrance to the particle trap are listed in columns 5 and 6 of Table 1.2. Another effect observed during these t e s t s was significant variation in the amount of undissolved g a s i n circulation with the fuel salt. These changes were measured by the s m a l l reactivity effects at zero power with no '35Xe in the system. Additional qualitative support for variations in void fraction was obtained during power operation from spectral measurements of the inherent neutronflux noise and from temperature observatiods at the reactor access nozzle. (See also Sects. 1.2.2 and 3.3.) The reactivity balance gave a reactivity loss of 0.032% 6k/k between the condition with the fewest circulating voids (1225'F and 3 psig overpressure) and that with the most (118O0F and 9 psig). This implies a change of 0.15 to 0.2% by volume i n the void fraction between these two conditions. No change in void fraction with fuel pump level was detected, and there was no observable pressure effect at 1225'F. However, at 1180째F over half of the total change between the two extremes was due to the pressure difference. Column i' of Table 1.2 shows the approximate magnitude of the additional circulating void fraction at each condition (The void fraction at 1225OF and 5 psig is thought to b e 0.1 to 0.15% by volume.) The tests in this run indicated at least the qualitative nature of the effects of system temperature and pressure on xenon poisoning and circulating voids. However, they also demonstrated that the effects are not necessarily completely reproducible. During run 12 and earlier runs, the typical value for I3'Xe poisoning at 7.2 Mw was 0.27% 6k/k.

...

Table 1.2.

Effects of Fuel System Pressure, Temperature, and Level on 135Xe Poisoning ot 5 Mw

Fuel Pump Reactor Outlet Overpressure Temperature

Fuel Pump Level

Net Xenona

(in.)

(So akk/k)

Poisoning

Off-Gas System Temperature f F ) Holdup Volume, Line 522

Particle Trap, Yorkmesh Section

Change from Minimum Circulating Void Fraction (vol %)

(Psig)

eF)

1225

6.2

0.182

234

2 92

1225

5.6

0.187

228

272

1210

6.1

0.206

231

264

0.03

1210

5.6

0.219

229

260

0.03

1210

5.3

0.233

228

2 58

0.03

1180

5.7

0.336

221

1180

5.6

0.358

22 0

251

0.11

1180

5.3

0.374

219

246

0.11

1225

6.2

0.191

236

1225

5.6

0.201

232

231

1210

6.0

0.227

230

222

0.04

1210

5.6

0.232

228

224

0.04

1210

5.3

0.241

224

217

0.04 0.10

0.11

1195

5.9

0.263

22 1

214

1195

5.6

0.289

214

2 02

0.19

1195

5.3

0.346

204

180

0.10

1180

5.6

0.371'

193'

152'

0.18

1180

5.3

0.374'

195'

155'

0.18

1225

6.2

0.175'

187

244

1225

5.6

0.182'

193

227

1180

5.8

0.240'

185'

230'

0.10

1180

5.5

0.248'

188'

226'

0.10

1180

5.3

0.260'

189'

220'

0.10

1180

5.6

0.294'

217

2 65

0.11

1180

5.3

0.295'

216

255

0.11

'Corrected for drift in zero-power residual reactivity and for direct reactivity effect of differences i n circulating void fraction. 'These data were taken after a period of operation a t 10 kw, during which the pattern of xenon behavior changed toward less pressure-temperature-level sensitivity. 'Partial restriction i n off-gas line at fuel pump caused change in flow path and holdup time.

6 For the first four months of run 14, the characteristic value for this term was 0.33% 6k/k. After the period of low-power operation in January 1968,the xenon term reverted to the old value. Since this change occurred in the middle of the 5-Mw tests to define xenon behavior, no explicit quantitative evaluation can be made. This is illustrated by two nominally identical tests made under conditions of high xenon poisoning before and after this change. Tests on December 23 to 25, 1967, and March 3 to 6, 1968, both involved operation at 5 Mw and 1180OF with 5 psig overpressure. The respective xenon poisoning terms were 0.35 and 0.30% 6k/k. Long-Term Changes. - Because of variations i n the xenon poisoning, the best data on the long-term drift in residual reactivity are obtained at very low power with no xenon present. In earlier operations, these data were taken only near the beginning and the end of the various reactor runs, and many runs were relatively short in terms of fuel burnup and net control rod movement. Comparison of data between runs was furthet complicated by the need to compensate for the dilution effects of fuel drains and flushing operations. All these results were scattered within a relatively narrow band of reactivity values with no particular trend in evidence. During the six months of run 14, zero-power reactivity data were collected on five separate occasions. These data showed a very s m a l l but remark-

ably constant negative trend in residual reactivity as a function of integrated power. The average reactivity slope during run 14 was -3.5 x Gk/k)/Mwhr. Reexamination of earlier data showed considerable scatter around this value but no real inconsistency with it. Some of t h i s scatter occurred because significant scatter was encountered between individual reactivity balances at some conditions. Figure 1.2 shows the zero-power reactivity balance results for the entire 235Uoperation. Data points taken within a particular reactor run, that is, after a reactor fill and before the subsequent drain, are connected by straight lines. Where lines are drawn from two points at a given condition, the points represent the extremes of significant reactivity balance variations at that condition. At conditions where only trivial variations were observed, only a single point was plotted. If the negative reactivity slope observed in run 14 were applied to the entire power history, a total negative deviation of 0.22% 6k/k from the beginning of operation would b e expected. T h e fact that a total change of this magnitude has not been observed suggests the presence of errors that tend to compensate for the negative drift observed during operation. A possible source of such an error is the corrections that must be applied for the dilution effects of reactor drains and flushes between runs.

Lâ&#x20AC;?

OR N L DWG 68-3939

Fig. 1.2.

Residual Reoctivity a t Zero Power.

A reactor flush results i n a net transfer of uranium (and other fuel constituents) from the fuel salt t o the flush salt. If the amount of uranium transferred in such a flush is overestimated in the correction that is applied t o the reactivity balance, the residual reactivity is shifted in the positive direction opposite to the direction of drift during operation. An overcorrection of 2.6 kg in total uranium would be required t o compensate for a drift of -0.22% 6k/k. To date, total uranium corrections in the amount of 4.9 kg have been applied t o the reactivity balance; analyses of the flush s a l t when i t was last in the fuel loop indicated a total uranium content of 5.0 kg. In applying the calculated corrections, it is assumed that the total volumes of fuel salt and flush salt remain constant. A gain of 0.7 ft3 i n the actual flush salt inventory at the expense of the fuel salt could also conceal the expected negative reactivity drift. However, the syst e m physical inventories indicate that if there has been any shift at all, it has been to increase the fuel s a l t volume at the expense of the flush salt. Thus it does not appear that errors in corrections between reactor runs are compensating for the downward drift during the runs. The source of the compensating effect h a s not yet been identified.

There are a number of possible causes for the negative reactivity drift during any one run. However, none of these appears to be large enough to explain the drift. For example, if the drift were due t o an error in power calibration, an error of about 30%would b e required. The data on the buildup of '"U i n the fuel (see Sect. 1.2.5) indicate that the power calibration error is less than 10%. This is further supparted by the rate of 239Pu buildup in the fuel and various fission product studies-. Another possible source of e m r is nonuniform deposition of fission products in the fuel loop. Even if all the nonvolatile nonsaturating fission products were deposited on the core graphite (uniformly), the negative reactivity effect would be less than one-half the drift rate in run 14. Therefore the s m a l l negative trend in reactivity remains unexplained at the present time. It is worth noting, however, that the magnitude is well within the allowable reactivity anomaly of 0.5% 6k/k. ("his limit is based on reactor safety studies that placed conservative interpretations on reactivity losses and their possible consequences.)

1.2.2

Variations in Reactor Access N o z z l e Temperatures

J. L. Crowley

The reactor access nozzle is one of the two places in the fuel loop where there is a gas-liquid interface; the other is the fuel pump bowl. Variations in liquid level in the access nozzle, which can be inferred from thermocouple readings, afford some information on the gas circulating with the fuel. As shown in Fig. 1.3, the reactor access nozzle (RAN)is a 10-in. pipe with a cooling-air jacket on the outside. Inside is an air-cooled plug attached t o the closure flange. In the large plug is a smaller nozzle and an air-cooled plug, providing access to the sample assembly in the core. It was intended that frozen salt plugs be established id the annuli between the cooled plugs and nozzles to prevent molten salt from coming in contact with the metal ring seals at the access flanges. In actual operation, however, the indications are that it is trapped gas i n the annulus and not a frozen salt plug which prevents liquid salt from rising to the seal area. Two mechanisms transport g a s t o and from the RAN annuli. Owing to the higher static pressure at the RAN, helium is transferred from the RAN t o the pump bowl gas space as a solute. This effect was previously noted on the Engineering Test Loop i n the operation of the forerunner of the reactor access nozzle. The mechanism which delivers g a s to the RAN is the collection of circulating bubbles from the fuel passing by the lower ends of the annuli. The equilibrium condition between these two mechanisms determines the gas inventories and the liquid levels i n the RAN annuli. Changes in the liquid levels can be detected by monitoring the thermocouples attached t o the RAN. In general it h a s been noted that the RAN salt levels run higher when the system operates at lower pressure or higher temperature. An example of the above effect i s provided in Fig. 1.4. This is a plot of two RAN thermocouples (see Fig. 1.3 for location), the fuel pump helium

3MSR Program Semiann. Progr. Rept. Feb. 28, 1965,

ORNL-3812, P. 9. 4MSR Program Semiann. Progr. Rept. July 31, 1963, ORNL-3529, p. 40.

8 ORNL-DWG 68-5507

COOLING AIR

3 COOLING

I I I

w V ~in.

10-in.

I-

in.

YD R-7A R-7B

L SALT

Fig. 1.3. Reactor Acc’srr Nozzle Showing 10- and 2t-in. Annuli.

ORNL-DWG

1300

68-5508

t 200

I 100 1000

900

perature also appeared to reduce the amount of bubbles bringing gas t o the annuli. The variations in circulating gas concentrations with changes in fuel temperature and pressure, inferred from RAN temperatures, agree with the changes indicated by the reactivity (see Sect. 1.2.1).

1.2.3

800 700

Radiation Heating

C. H. Gabbard

600

500

400 a

3s .v)

za ;

. FUEL PUMP PRESSURE-

DATE, BEGINNING FEB 6. t968 Fig. 1.4.

Effect of Reactor Outlet Temperature and

Fuel Pump Pressure on Reactor-Access-Nozzle Temperatures.

pressure, and the reactor outlet salt temperature for a period of ten days in February. The t e s t in progress at that t i m e was one to determine the effects of pressure and temperature on reactivity while operating at 5 Mw power. The fuel pump helium pressure was first lowered from 9 to 5 psig with the reactor outlet at 1180'F. On February 10 and 11, the reactor outlet was increased to 1225'F, while the fuel pump pressure was maintained at 3 psig. At the beginning condition, that is, high syst e m pressure and low salt temperature, the moltensalt levels in the annuli are below the levels of the two thermocouples. When the pressure was lowered on February 6, RAN temperatures dropped initially, indicating the lowering of the salt level due t o expansion of gas trapped in the RAN annuli. Very soon, however, the molten salt rose to or above the R43B thermocouple location. Thus it would appear that bubbles were bringing gas into the annuli at a lower rate. The next changes, increases i n circulating fuel temperature on February 10 and 11, also resulted i n rises in the molten-salt level in the annuli. After the second temperature increase, the level apparently rose over a period of about two days, until it was near thermocouple R7B. Thus the increase in circulating fuel tem-

The temperature differences between the reactor inlet and the lower head and between the inlet and the core support flange continue to be the indicator for possible sedimentation buildup within the reactor vessel. These temperature differences agree satisfactorily with previous data and indicate that there has been no significant solids accumulation within the reactor vessel. Table 1.3 shows a comparison of the data collected during run 14 with those collected and reported previously. The temperature distribution on the upper surface of the fuel pump tank during run 14 was about the same as that in run 12. As reported previously, the temperatures are somewhat lower than in earlier power runs, and the reason for the difference h a s not been established. Table 1.3.

Power-Dependent Temperature Differences

Between F u e l Salt Entering, and Points on, the Reactor Vessel

Temperature Difference fF/Mw) Run No.

Dates

Core Support Flange Lower Head 6

4/66-5/66

2.11

1.54

1/67-5/67

2.14

1.50

6/67-8/67

2.20

1.55

9/67-2/68

2.25

1.42

1.2.4

Thermal C y c l e History

C. H. Gabbard The accumulated thermal cycle history of the various components sensitive to thermal cycle 'MSR Program Semiann. Progr. Rept. Aug. 31, 1967, ORNL-4191. p. 22.

10 Table

1.4.

Component

MSRE

Heat/Cool

Fill/Drain

Power

-/Off

Fuel system

Coolant system

447

Coolant pump

127

Freeze flanges 100, 101, 102

Freeze flanges 200, 201

Penetrations 200, 201

Fuel pump

Cumulotive Thermal Cycle History Through February 1968

Thaw

Thaw and Transfer

Freeze valve 103

104

105

106

107

108

109

110

111

112

204

206

damage is shown in Table 1.4. Approximately 69% of the design thermal cycle life of the fuel system freeze flanges has been used to date. This compares with a value of 63% of the design life that had been consumed at the end of the previous report period. Essentially no thermal cycle damage has been accumulated on the fuel system flanges since the fuel system was last filled on September 20, 1967, at the beginning of run 14.

1.2.5 236UIndication of R. C. Steffy, Jr.

Integrated Power

J. R. Engel

The basic measurement of the nuclear power of the MSRE is by the overall system heat balance.

The preponderant term in this balance is the heat removed from the system by the secondary (coolant) salt. I t s accuracy is therefore strongly dependent on the accuracy with which that term is evaluated. Other indications of reactor power - radiator air heat balances, fission product inventories, and nuclear instrument readings have provided only reasonable (f20%) confirmation of the heat-balance power calibration. Since all these techniques could involve substantial systematic errors, an effort has been made to check the power calibration by an independent technique. A potentially accurate standard a s a monitor for integrated power in a nuclear reactor is the change in the isotopic composition of the uranium fuel. The accuracy of this method depends on the existence of an isotope whose concentration changes

6, ?

significantly and in a manner that can be accurately related to energy production. The 236Uin the MSRE fuel is well suited to this purpose. The fraction of 236Uin the uranium mixture has increased by a-factor of 3 in the course of power operation with the 235Ufuel. Since 236Uis produced only by parasitic neutron captures in 235Uand since the rate of burnup of ”‘U is very low, the net production of this isotope per fission event depends primarily on the value of a (= Z,/Zt) for 235Uin the MSRE neutron spectrum. Extension of this relationship to total energy production depends only on the effective energy yield per fission in this reactor, The absolute rate of production of 236Uin the MSRE was inferred from several fuel samples in which the uranium isotopic composition was measured. These data were combined with “book” inventories for total uranium and corrected for various reactor reactions (drains, flushes, and fuel additions) to obtain loop inventories for 236U. Since chemical results for total uranium have consistently been within 1%of the “book” values and the precision of the isotopic assays is high, the accuracy of the 236Uinventory is probably within *5%. Figure 1.5 shows the calculated 236Uinventory plotted against the integrated power based on

the heat balance measurements. The solid line is a least-squares fit to the data points and has a slope of 0.0110 f 0.002 g of 236uper megawatthour. The theoretical slope, based on the ratios of effective 235Uneutron cross sections and fission energy yield i n the MSRE, is 0.0109 g/Mwhr, within 1%of the observed slope. The uncertainty in the theoretical slope is probably less than lo%, so the actual integrated power is probably within 10%of the value indicated by the heat balances.

1.3 EQUIPMENT PERFORMANCE 1.3.1 Salt Pumps P. N. Haubenreich The 120Ggpm pump that circulates fuel salt and the 850-gpm pump that circulates coolant s a l t continued to run uneventfully. By the end of the report period the fuel pump had run for 18,400 hr, 14,415 hr circulating salt and 3985 hr circulating helium. The coolant pump had pumped salt for 16,229 hr and helium for 3082 hr, for a total of 19,311 hr. As reported in Sect. 1.3.6 and 1.3.10, there is probably some oil leakage into the pump bowls, but the rates must be very small since they do not produce significant changes i n the oil inventories. The replacement rotary elements for both pumps are seal-welded to prevent such inleakage, but the oil problem is so minor that replacement is not contemplated.

1.3.2

H e a t Transfer

C. H. Gabbard

The heat transfer performance of the fuel-to-coolant-salt heat exchanger continues to be monitored by periodic evaluations of the heat transfer coefficient and by the heat transfer index.6 The heat transfer index was evaluated a number of t i m e s during run 14 with the reactor a t full power. However, only one additional heat transfer coefficient measurement over a range of powers was made during this report period. This measurement gave a value of the overall heat transfer coefficient of

INTEGRATED HEAT-BALANCE POWER (Mwhr)

Fig. 1.5.

236U Buildup in MSRE vs Power Production.

6MSR Program Semiann. Progr. Rept. Feb. 28, 1967, ORNL-4119, p. 21.

605 Btu hr-’ ft-’ (OF)-’, compared with 586 Btu hr-l f t - 2 (0F)- which is the average of the six

’,

previous measurements. The average heat transfer index increased during run 1 4 to 0.0376 MwPF, which is also slightly higher than the previous data. The apparent increase in heat transfer was caused entirely by a recalibration of one of the coolant salt flow elements, which increased the indicated coolant salt flow rate and the calculated heat removal rate at the radiator. (This increased the calculated heat balance power by about 100 t o 200 kw at full power.) Thus the conclusion is that the heat transfer performance of the heat exchanger has remained unchanged since the beginning of operation.

1.3.3 Salt Samplers R. B. Gallaher During this six-month report period, operation of the sampler-enricher was resumed with a new isolation chamber and cable drive assembly. Except for the wiring failure that occurred in November, only minor difficulties were encountered. A total of 70 sampling operations were made during the report period, as follows: 1O-g salt samples

50-g salt samples

Salt samples in freeze,-valve capsules

Gas sample in freeze-valve capsule

Capsule exposure in gas space

Nickel rod exposure in salt

Three of the 10-g samples were of flush salt; the others were of fuel salt. The SO-g samples were taken for uranium isotopic analyses, determination of oxide level or U 3+/U4+ ratio, and other special analyses, as reported in Part 3 of this report. All the sampling attempts were successful, except that one 50-g capsule collected only 5 g of fuel salt. The attempt appeared normal in every other respect, and the explanation for the small amount of salt is not known. Operations during the period brought the total use of the sampler-enricher to 114 uranium enrichments and 349 samples and special exposures. Since the maintenance period that ended in September, the pressure transducer in the removal

valve buffer system has been sensitive to heat from the illuminator, drifting by 25%of full scale when the light is turned on. A reflective metal shield was placed over the thermal insulation around the transducer. This helped, but it did not eliminate the effect. The wiring fault appeared i n the following manner. On November 14, as a sample was being withdrawn from the pump bowl, the 0.3-amp fuse in the drive motor circuit blew. Subsequent tests showed that the insert m o d e drew the normal 0.2 amp, but in withdrawal the current rose to 1.0 or 1.5 amp. Since the locked-rotor current is only 0.3 amp, the high current indicated leakage. During further tests, three circuits opened, disabling the motor “insert” and “withdraw” circuits and the upper limit switch.

L, I

The most likely location for the failure was b e lieved to be at or between the outer and inner containment penetrations. This area was suspected because of its s m a l l clearances and tight bends in the cable. High-frequency capacitance measure ments with a Grid-Dip meter indicated that the length of cable to the point of failure was consistent with this guess. Therefore, after a nonflammable plastic tent was erected over the samplerenricher to prevent spread of possible contamination, a 3-in.-diam hole was sawed in the cover plate directly above the cable penetration into the inner box (1-c area). The failure was found where the wires were bent back down against the side of t h e plug on the lower end of the cable between the inner and outer boxes. A temporary connection showed that everything inside the inner box was operable. The sample capsule was retrieved, and the isolation valves were closed at this time. The damaged section of cable was abandoned, and a new cable was installed having a penetration through a 4-in. pipe cap welded over the sawed hole in the top cover.

While the cap was being welded in place, a heavy current evidently went to ground through another cable that penetrated the top cover near the new cap. The receptacle and plug were destroyed. Repairs were made by cutting out and replacing this penetration. As a result of this experience, an isolation transformer and a fuse were added to each of the three drive unit cables (see Sect. 3.1.3). This allows one ground without interference with opera tion.

' W

After the repairs, all circuits were operable except the upper l i m i t switch. This switch normally stopped the drive motor when the latch reached the latch stop. Because the motor and circuit can stand blocked-rotor current, the upper l i m i t switch was bypassed, and the motor is now turned off manually when the position indicator shows the latch is fully withdrawn. Containment has been quite good. A minor release of activity to the stack occurred at least once while area 3A was being evacuated. After all disconnects i n the sampler off-gas system were checked and tightened, no further detectable release occurred. During the report period the coolant salt sampler was used t o take five 10-g samples, bringing the total to 65. No operating difficulties were encountered.

1.3.4

Control

Rods

and Drives

M. Richardson The control rods continued to operate freely throughout the report period. T e s t s i n September and late January showed no shift in position reading and no appreciable change in rod drop times. The only mechanical difficulty was with the fineposition synchro transmitter on drive No. 2, which developed an open circuit in January. This synchro was one that was installed in May 1967, so its service life was relatively short for some reason as yet unknown.

1.3.5

Radiator Enclosure

M. Richardson

During the flush salt operation i n September, at the beginning of run 13, the outlet door tended t o jam because it was hanging crooked. The trouble appeared to be i n one of the housings for the springs through which the cables lift the door. A piece of some material, apparently weld slag, had fallen in and jammed the piston i n that housing, preventing it from moving up. The piston in the other housing was free. Both coil springs were found to have b e c o m e permanently compressed, probably because of subjection to high temperatures i n the startup period before hoods were installed over the doors. Since the springs were not

essential, they were replaced with pipe sleeves that hold the pistons down. The brakes in the lifting mechanism remained adequate, limiting coastdown to about 3 in. The door seals continued to provide adequate heat containment with the doors closed. 1.3.6 Off-Gas Systems A. I. Krakoviak The off-gas systems of both the fuel and coolant system developed partial restrictions that caused minor inconveniences but did not interfere with noma1 power operation. Fuel Off-Gas System. - A restriction became evident i n the off-gas line somewhere between the pump bowl and the junction of the overflow tank vent with the 4411. holdup line. The first indication appeared after 2'/2 days of operation at 10 kw following the sampler-enricher wiring failure of November 14. During this period the operational and maintenance valves were open, and a 0.6-liter/min helium purge was maintained down the sampler tube to the pump bowl t o prevent contamination of the sampler by fission gases. The restriction was evidenced by an increase (0.5 psig) in the fuel pump bowl pressure when the overflow tank vent valve was closed during a routine return of salt from the overflow tank t o the pump bowl. After the wiring repairs had been finished, the sample had been retrieved into area lC, and the operational and maintenance valves had been closed, the restriction was relieved by pressurizing the pump bowl to 6.0 psig and suddenly venting the gas into a drain tank which was at 3 psig. The pressure drop then appeared normal (<0.1 psi), but after three more days of low-power operation] an abnormal pressure rise (0.2 psi) was again observed when salt was being returned from the overflow tank. Repetition of the mild blow-through to the drain tank had little or no effect this time. The line was not completely blocked, however, and full-power operation was resumed. At first, temperatures on the overflow tank and the off-gas line (responding to fission product heating) clearly indicated that there was e n o u a pressure drop at the pump bowl outlet to cause much of the off-gas to bubble through the overflow tank and out its vent line. Then, after 15 hr at full power, the bypass flow stopped, indicating that the pressure drop through the restriction had decreased significantly.

14 The pressure drop remained below the l i m i t s of detection throughout the next nine weeks while the reactor was operating a t 7.2 Mw or 5 Mw. Then after two days a t 10 kw, the pressure again became detectable and continued to increase over the next six days at very low power. When power operation was resumed at 5 Mw, temperatures indicated that there was again bypass flow through the overflow tank. The restriction increased during two weeks of operation at 5 Mw, but the line never became completely plugged. Then while the overflow tank was being emptied, four days after the resumption of full-power operation, the restriction partially blew out, bringing the pressure drop again below the l i m i t of detection. The pressure drop remained just at or below the l i m i t thereafter. The behavior was quite unlike the plugging that occurred earlier in the same section of line a s a result of the overfill with flush salt, mainly in that it never completely plugged and there was a suggestion of some effect of power level. Plans were made to investigate this restriction in the offgas line at the next shutdown. Tools were prepared like those used earlier to clear out the off-gas line at the pump bowl. Preparations were made to remove the short flanged section of the off-gas line for examination and replacement with a line instmmented with thermocouples. Main Charcoal Beds. The performance of the charcoal beds was very good; none of the annoying plugging problems experienced pmviously* * 9 appeared during this report period. During run 13 and the early part of run 14, with sections 1 A and 1B in service, the maximum pressure drop across the beds in parallel was 3.0 psi. The main charcoal beds were conservatively designed to delay the krypton and xenon long enough (7 days for Kr, 90 days for Xe) so that the major radioactive constituent in the effluent gas would be l c y e a r 8SKr. The beds were designed to provide these holdup times when the off-gas flow was equally divided between two sections in parallel, but there were some indications during early operation that adequate holdup might be provided with all the flow through only one section. Therefore, in November a test of the capability of a single

'MSR Program Semiann. Progr. Rept. Feb. 28, 1967, ORNL-4119. pp. 27-29. 'Zbid., p. 30. 'MSR PrOgrem Semiann. Progr. Rept. A&. 31, 1967, ORNL-4191, p. 28.

section was started, with only section 1 A on line. Only a very moderate rise in effluent activity materialized. Section 1 A remained in service alone for the remainder of the run with the exception of a ten-day reactivity test at low fuel pump pressure, when two beds were in service to permit lower pressure. Coolant Off-GasSystem. - The coolant pump offgas system has experienced chronic plugging at the s m a l l 50-p sintered-metal filter upstream of the pressure control valve. This filter has plugged regularly and h a s been replaced seven times over the three years since the start of preoperational checkout of the system. The plugging appears to be due to liquefaction of oil vapors from the coolant pump in the pores of the sintered metal. The filter plugged again during this report period, and the coolant pump pressure was controlled by venting through a line bypassing the plugged The output from the coolant pump pressure controller was switched from the normal control valve to the vent valve to give satisfactory pressure control.

1.3.7

Main Blowers

C. H. Gabbard The two main blowers, MB-1 and MB-3, continued to run when needed and by the end of the report period had accumulated 7361 and 6685 hr, respectively, since they were rebuilt in the fall of 1966. The main bearings on both blowers had to be replaced in January, but it happened that there was no interference with reactor operation. The reactor was being operated at 5 Mw on one blower, MB-3, when routine monitoring showed the bearing vibration increasing from 0.8 to 1.5 mils. When an oscilloscope trace showed abnormal high-frequency noise in the bearing, MB-1 was brought on line and MB-3 shut down. A week later, the reactor power was lowered to 10 kw a s part of the reactivity experiments; so for several days neither blower was needed. Although the vibration meter on MB-1 bearings had shown no trouble during operation, there was some periodic noise and roughness that could be detected during a coastdown. Therefore, while the power was down, the thrust bearing on each blower was replaced. In the MB-1 bearing, one ball had a relatively large surface pit, accowting for the noise and "Xbid., pp. 28-29.

MB-3

MB-4

Fig. 1.6.

Ball Bearing Failures from MSRE Main Blowers MB-1 and MB-3.

roughness. The M E 3 bearing was i n much worse condition, with one ball actually fractured into four large pieces. The damaged balls from the two bearings are shown i n Fig. 1.6. The MB-1 bearing was the one supplied with the rebuilt blower in the fall of 1966 and had seen 6633 hr of service. The MB-3 bearing was a replacement, installed in March 1967, that had accumulated 4801 hr. Lubrication appeared to have been adequate, so the relatively short service life could not be attributed t o improper lubrication (which was the suspected cause i n an earlier beari n g failure).

A n a l y s i s of t h e bearing loading and

stresses revealed that according to accepted design correlations the expected service life was only 5000 hr, about what was actually attained. The original bearings were radial-type ball bearings i n which thrust load produced high stresses. Therefore, they were replaced with angular-contacttype ball bearings having a greater thrust capacity and much longer expected life.

1.3.8

Heaters

T. L. Hudson While the reactor system was being heated up for run 13, heater FV-103 was lost by an open circuit. This 1.5-kw bent-tubular-type heater normally supplies about 200 w of heat to a 4 i n . section of the

fuel drain line within the reactor vessel furnace, between the freeze valve and the resistance-heated section of the line. The purpose of the heater was to help control the temperature profile through the freeze valve. After the heater failed, t e s t s with flush salt showed that by proper adjustment of the cooling air controls, the freeze valve could be maintained reliably with thaw t i m e s i n an acceptable range. (Thaw t i m e was around 1 2 min when the reactor vessel was at 1180째F and the center of the freeze valve was at 495OF.) Therefore this heater was not replaced. T h e last of s i x heating elements i n heater HX-1 on the primary heat exchanger failed during the previous semiannual report period. In December a lead to the adjacent heater, HX-2, opened, thereby reducing the output of this heater by 50%. A week later a second partial failure further reduced the output to only 33%of normal. Although these failures do not affect operations so long as salt is kept circulating, the lack of heat i n the two adjacent heaters will make i t necessary to repair or replace them during the next shutdown.

1.3.9

Electrical System

T. L. Hudson Power to the MSRE electrical system is supplied from the ORNL substation by either of two 13.8-kv

16 power lines, a preferred line or a n alternate. Originally the two main blowers could not be operated while the area was on the alternate feeder, but rearrangement of feeders at the ORNL substat i ~reduced i the load on the alternate supply to the MSRE. During this report period the reactor was operated for the first t i m e at full power while on the alternate feeder. The occasion was a manual switchover for 2 hr, while a defective insulator was replaced at the substation. A building power failure occurred on a foggy morning in October, when an arc developed between a 13.8-kv line to the main transformer for the building and a parallel metal activator rod to the 13.8-kv line fuse. The cause of the arc is unexplained: Although the weather was foggy, i t was not unusually wet. This fault caused the operation of the preferred feeder overcurrent ground relay located at the ORNL substation, which tripped the feeder breaker before the line overcurrent relay located at the MSRE could operate to prevent an automatic transfer to the alternate feeder. When the MSRE was transferred, the alternate feeder overcurrent ground relay operated and the alternate feeder breaker also tripped. (To prevent an automatic transfer to the alternate feeder on a similar fault, an overcurrent ground relay was later installed at the MSRE.) After an interruption of 37 min, low-power nuclear operation was resumed on emergency electrical power from the diesel generators. The damage was repaired, the spacing b e tween the line and the activator rod was increased, and normal service was restored after an intemption of 2 hr. The three diesel generators at the MSRE have proved to be quite reliable. They are started and operated unloaded for about an hour each week. Once a month they are tested under load. They have never failed to start when required during a power outage. Under emergency conditions they are usually running within 2 min after a failure of the normal power supply.

1.3.10 Salt Pump Oil Systems A. I. Krakoviak The lubricating oil systems for both salt pumps operated continuously and without incident throughout this six-month report period. The only problem was the recurring, gradual fouling of the water

side of the cooling coils on the oil reservoirs. By December the temperature of the oil supplied to the pumps was up to 150OF; so the water supply was changed from tower water to cooler process water. During about two months on process water, the heat transfer improved to the point that tower water again gave acceptable cooling. Accumulation of shaft seal oil leakage from the fuel pump was steady at about 12 cm3/day from September through November. For the next two months there was no measurable accumulation in the oil collection tank. Then about February 10, leakage began to collect again and averaged 6 cm3/day for the remainder of the month. Oil accumulation from the coolant pump shaft seal leakage also varied. In September and October it was only 7 cm3/day. The rate increased in November to 15 cm3/day, and i n the last three months was relatively steady at about 20 cm3/day. These accumulation rates are in the range observed i n earlier operation of the salt pumps. During power operation, s a m p l e s were taken at weekly intervals, end they showed no significant deterioration of the oil quality. To compensate for these samples and shaft seal leakage, 2.8 gal of oil was added to the fuel pump lubrication system and 2.0 gal to the coolant pump system. Inventories based on additions, sample removals, and accumulation of seal leakage showed, during this six-month report period, a net apparent gain in the coolant oil system of 114 c m 3 and a net apparent loss from the fuel oil system of 545 cm3. Because of inaccuracy in the inventories and a slight systematic error as unmeasured losses i n sampling the fuel pump oil, the probable error i n net change is about 1200 cm3. Thus the apparent changes are well within the probable error.

1.3.11 Cooling Water System

P. H. Harley Operation of the cooling water systems continued satisfactorily throughout the report period. The only disturbance of normal operation occurred when a 6-in. cast-iron pipe supplying water for the building fire protection system and the potable water system cracked underground just outside the building. To isolate the leak, it was necessary to shut off both this line and the line supplying the process cooling water system. Adequate cooling was main-

h-,

tained while the break was being repaired by shifting some i t e m s from process water to tower water and running a temporary hose connection from the water main to the process water system. A dry fire hose was laid from the nearby Nuclear Safety Pilot Plant for possible emergency use. Although leak tests in June 1967 detected no leakage from treated-water components in the reactor cell, a small leak apparently persists. Throughout this six-month period, condensate collection in the component cooling system averaged about 0.73 gpd. There was no accumulation of water in the reactor cell sump. Measured losses from the treatedwater system averaged 1.5 gpd. About 0.75 gpd is lost by evaporation from the degassing tank that strips radiolytic hydrogen and oxygen from the treated water. Thus the unmeasured loss agreed closely with the condensate collection rate. During a short period in December the treatedwater l o s s increased to about 4 gpd. The increase was traced t o a leak across a pressure-relief valve which discharges to the waste system to prevent excessive pressure buildup i n the event the radisltion block valves close. The leaking valve was replaced.

1.3.12 Component Cooling Systems

P. H. Harley

Oil leaks i n the lubrication systems of the main component cooling pumps were troublesome during this period. In run 13, after s i x days of flush salt circulation and two days of nuclear operation, CCP-2 was shut down by low oil pressure, leading to the decision to terminate the run (see Sect. 1.1). Repairs consisted in capping a leaking drain line and tightening several packing nuts and fittings. The other blower, CCP-1, was used for the first 873 hr of run 14. At that t i m e a serious oil leak was indicated by low oil pressure and accumulation of about 2 gal of oil i n the condensate collection tank connected t o the blower containment. The standby unit, CCP-2, was immediately started up and operated without further difficulties throughout the remainder of the period, 3036 hr. The strainer i n the discharge of the component cooling pumps developed an excessive pressure drop during run 13, and the screen was replaced at the t i m e of the repairs to CCP-2. The strainer basket that was removed had been installed before

run 12. In 1800 hr of operation its 100-mesh screen had become partially plugged with rubber dust from the drive belts. The screen was damaged i n removal, and a replacement basket was made using 16-mesh 0.023-in.-wire screen. This screen showed no pressure buildup over the five months to the end of the period. Cooling air for the coolant salt freeze valves was supplied most of the t i m e by the service air compressor, AC-3. Component cooling pump 3 was kept i n standby and used only when AC-3 was undergoing programmed miiintenance. Once while CCP-3 was in service, a bearing seized, requiring replacement of the bearing and drive belt. The blower had been operated for a total of only about 200 hr since a similar bearing failure. l 1

1.3.13 Containment and Ventilation P. H. Harley

R. C. Steffy, Jr.

Throughout all the operations i n this period the reactor and drain tank cells were held near -2 psig. The air inleakage into these cells was about 15 scfd, less than the acceptable maximum by about a factor of 5. Inleakage rates are measured by a balance that includes purge and exhaust flows and changes i n cell temperature and pressure. At the beginning of runs 13 and 14, a familiar pattern was repeated. The pattern consists of three stages:

1. In the first day after the cell is closed, while temperatures are coming up to equilibrium, the indicated inleakage rate is quite high, somet i m e s as much as 200 scfd. 2. For three days or so after the first stage, the indicated inleakage stays up around 75 t o 100 scfd. This high apparent rate is attributed t o gradual saturation of the cell atmosphere with water vapor.

3. When the cell atmosphere becomes saturated at the component cooling pump cooler, condensate begins to appear there, and the indicated cell leak rate breaks down sharply to the actual inleakage rate, usually 15 to 20 scfd.

I1MSR Program Semiann. Progr. Rept. Aug. 31, 1967, ORNL-4191, p. 29.

18 The cell atmosphere is maintained at about 3% oxygen by supplying a continuous purge of nitrogen. Oxygen analyzer readings can, in principle, b e used with other data to compute air inleakage. However, assumption of no oxygen consumption in the cell led to a calculated negative inleakage of air. Therefore the inleakage measured by the flows and pressure changes was assumed to b e correct, and the oxygen analyses were used to compute oxygen consumption in the cells. Results showed an apparent consumption of 3 to 4 scfd of oxygen, presumably by reaction with some substance in the cells. Moisture condensed in the component cooling system (which is an extension of the reactor cell) is drained once a day. Collection rates averaged 0.73 gpd over the five months of run 14 i n this report period. Daily amounts ranged from practically none to several gallons when the temperature outdoors (and in the special equipment room) was changing. Ventilation through operating areas of the reactor building (and the reactor cell during maintenance)

is maintained by one of two stack fans. Stack fan No. 1 is normally kept i n service, with No. 2 on standby. After almost continuous service since lubrication system modifications i n March 1966, fan No. 1 developed roughness in a bearing and the bearing was replaced in February 1968. During maintenance work in September 1967, the pressure drop across the roughing filters at the ventilation stack increased to the point that stack flow was significantly reduced. The filters in two of the three parallel banks were replaced, and dust filters were installed on inlets to ducts in all accessible areas to retard buildup on the roughing filters. Pressure drop across the highefficiency particulate filters downstream of t h e roughing filters remained practically unchanged. Activity released through the stack during the six-month report period amounted to less than 0.25 m c of particulate activity and only 2.3 m c of iodine. Nearly half (1.0 mc) of the iodine was released during the maintenance work at the fuel sampler early in September. The remainder came from the sampler at various times.

2. Component Development Dunlap Scott

2.1 OFF-GAS SAMPLER R. B. Gallaher

A. N. Smith

The system developed for sampling and limited on-line analysis of the fuel off-gas has been described in previous progress reports.' During this report period, the installation of the off-gas sampler was completed, and the system was given a preoperational check, including leak tests, pressure tests, and test operation of the instruments. Operational use then began with the removal of several samples for off-site analysis of xenon isotopic composition. Prior t o the removal of reactor gas samples, three samples of a standard gas containing krypton, xenon, and carbon tetrafluoride were trapped with the refrigerated molecular sieve adsorber. The adsorbed gases were swept into removable sample bombs, whose contents were then analyzed to determine the fraction of the sample passed through the bed that was recovered. The fractions recovered were calculated to be 67, 76, and 86%for the three samples. (The fraction recovered was practically the same for Kr, Xe, and CF,.) The molecular sieve and removal bombs were used to remove four samples of reactor gas that had been isolated for three to seven weeks in the permanent isolation chambers in the off-gas system. The first sample contained about 50% air, possibly because of incomplete purging of the system piping before the sample was removed. No air was detected in subsequent samples. Xenon isotopic ratios were measured, but the concentration of xenon was too low to obtain the de-

'MSR Program Semiann. Progr. Rept. Feb. 28, 1967,

ORNL-4119,pp. 4143.

sired accuracy. Therefore, after the third sample, the procedure was changed to obtain a higher concentration of gases in the bomb. One sample was trapped directly from the reactor off-gas stream while the reactor was operating at full power. When the sample was first transferred to the removal bomb, the radiation level in the containment box was over 100 r/hr, but the radioactivity decayed rapidly, and the sample was successfully removed after four days of decay. Several equipment problems were encountered, but all could be remedied. During preliminary testing, half the elements on one of the thermal conductivity cells burned out for some undetermined cause. A new cell was procured and installed, and no further trouble occurred. Just before the first off-gas sample was removed from the isolation chamber, one of the two pressure transducers failed. By modifying the procedure, i t was possible t o operate without the failed element. It was replaced, however, when the containment box was opened for another reason. The box had to be opened because of considerable difficulties with the valve extension handles. When the extension handles were first being installed, several of the operating fingers broke off easily a t welds. It turned out that a piece of Inconel had accidentally been used instead of stainless steel in making the handles, and the welds were brittle. All the handles were rewelded, using proper welding procedures. Later, while the third off-gas sample was being removed, a valve handle extension was lifted up and improperly reengaged. The extension was then damaged when an attempt was made to operate the valve. The top of the containment box had to be removed to repair the damage, and in this operation three other extensions were damaged. A slight modification was made in the top to improve

20 alignment between the valve handles and extensions, and changes in the design of the extensions were planned.

2.2 FUEL SAMPLER-ENRICHER

R. B. Gallaher The tangling of the drive cable that led to the reactor sbutdown in August2 apparently started with the capsule or latch jamming a t the lower isolation valve (the maintenance valve). Then, since the jam was undetected, the drive cable was forced off the reel into the isolation chamber above the valves, causing it to tangle and kink. The new drive unit that was installed in September has a latch of magnetic material, and during this report period, a device was developed to sense the passage of the latch past a point in the tube. The device is a commercially available unit that actuates a switch when a magnetic object moves into and out of its field. The device was tested and proved capable of sensing the latch inside the sampler tube. The available unit is not resistant to radiation damage and is therefore not suitable for mounting inside the reactor cell. A design was prepared to mount the switch about 10 in. below the maintenance valve to indicate when the latch has passed through both valves in the line. Thus if the latch hangs up before reaching this point, it should be detected in time to prevent enough excess cable being unreeled to cause a serious tangle. Some additional information on one aspect of the sampler trouble in August was uncovered when the inoperative drive assembly and isolation chamber (l-Cassembly) were disassembled in a hot cell. When the cover was removed in the hot cell, it was not possible to obtain a clear view of the drive unit. However, when the unit was removed from the box, it was evident that the Teleflex drive cable was not tangled in the gears as had been supposed. In fact, there was little or no slack in the cable between the reel cover and the hole in the latch positioner through which the cable passed to the isolation chamber below. The probable reason that the cable could not be driven up or down was found to b e a sharp kink in the drive

2MSR Program Semiann. Progr. Rept. Aug. 31, 1967, ORNL-4191, pp. 15, 32.

cable. This kink was lodged in the '$,-in.-diam hole, about in. from the bottom of the hole and 1 1 t 6 in. from the top. The kink could not be pulled easily through the hole, probably explaining why the motor stalled and then could not start the cable moving in either direction. In the removal of the drive unit from the assembly in the hot cell, most of the electrical insulation was accidentally stripped from the lead wires. It appeared that for some reason, probably radiation, the insulation had become quite brittle. After disassembly and decontamination the l-C assembly less the cable drive unit was returned to the reactor area. Replacement parts for the drive unit were ordered so that a complete assembly could be prepared as a spare. At low temperatures, radiation from fission products can produce free fluorine in frozen fuel salt. Because of concern over possible effects of fluorine in samples intended for oxide or U3' analyses, a carrier was designed and built to keep a fuel sample at about 500째F from the t i m e it is removed from the sampler until it is unloaded at the analytical laboratory. The carrier uses molten babbitt (85% Pb-10% Sb-5% Sn) as shielding and a heat reservoir. Built-in electric heaters melt the babbitt before the sample is loaded, and the heat of fusion keeps the temperature from decreasing significantly for about 9 hr. The sample is unavoidably cooled below 400'F in the sampler after it is removed from the pump bowl. The t i m e required to get the sample from the pump bowl to the carrier cannot be reduced much below 2% hr, but the hot carrier prevents this period from being extended during shipment or storage.

"/B

2.3 DECONTAMINATION STUDIES

T. H. Mauney Decontamination of the inoperative samplerenricher mechanism was used to develop further the cleaning treatments tried previously on the manipulator from the sampler-enri~her.~ The unit was grossly contaminated with fission products carried up from the pump bowl during sampling operations, so that great care was necessary in transporting and handling. The unit was

31bid., p. 40.

taken from the sampler-enricher into a specially prepared open-bottomed lead coffin, bagged, and stored in a spare cell a t the reactor s i t e until plans could b e worked out for decontamination. These plans were complicated because a facility had to be found that could contain the 3000-lb coffin, with capabilities for removing the coffin from the assembly, disassembling the mechanism, and then decontaminating it. Such a facility was found in the Fission Products Development Laboratory of the Isotopes Division. Some indication of the complexity of the disassembly and decontamination job is given by the photograph of the complete assembly (actually the replacement) shown in Fig. 2.1. Disassembly began with the very difficult task of remotely uncoupling seven tubing fittings and four bolts to remove the upper plate. Fourteen bolts were then removed from the cover plate. Removing the cover

plate permitted some inspection of the drive unit, but it was necessary to unbolt two more bolts to lift out that mechanism. It was not practical to attempt decontamination of the drive unit, especially since the drive cable was kinked and cut off and the wiring insulation was damaged. All the rest of the assembly was involved in the decontamination. Decontamination was done in two stages: the first in the Fission Products Development Laboratory and the second in the Equipment Decontamination Building. Table 2.1 lists the treatments given in each stage and radiation levels before and after. The readings shown before the first stage were made 15 weeks after the last sample was taken with the mechanism, and the readings after the final stage were taken two weeks later.

Table

2.1. Steps i n Decontamination of 1-C Assembly

Stage 1

- Fission Products Development Laboratory

Initial radiation level (11-22-67), 100-500 r/hr Initial contamination (smears), 5-1500 mr/hr Io3Ru, "Zr-Nb Treatment (8O-9O0C spray, scrub with fiber brush) 1. Water, 30min 2. 5% nitric acid, 6 0 min 3. Water, 30 min 4. 5% oxalic acid, 30 min 5. 5% Turco 4502, 30 min 6. Water, 30 min Final radiation level (12-1-67), 2-5 r/hr Final contamination (smears), 1-20 mr/hr lo3Ru, "Zr-Nb Stage 2

- Equipment Decontamination Building

Treatment (80-90'C) 1. Bab-0-oxalic acid, 30-min Scrub with wire brush 2. 0.40 M oxalic acid, 0.34 M hydrogen peroxide, 0.16 M citric acid, 30-min soak 3. Bab-0, 30-min scrub with wire brush Final radiation level (12-6-671, 200 mr/hr (1 tube a t 1 r/hr) Final contamination (smears), 1-5 mr/hr Fig. 2.1.

I-CArea

Unit from Sampler-Enricher (MSRE).

At the conclusion of the decontamination procedure, levels were low enough that the unit could be worked on directly. Thus the decontamination made available a spare component at substantially less cost than a new one. In addition, it demonstrated that the standard solutions used in the decontamination of stainless components are quite effective on the kind of contamination present in a salt sampler, giving an overall decontamination factor between 500 and 2500.

2.4

(T G)

ORNL-DWG

68-5510

'/e In.DIAM

y6 ih

\ 4h-0.016

MATERIAL-ALUMINUM APERTURE

in. DIAM

STUDY OF PINHOLE CAMERA FOR GAMMA SOURCE MAPPING T. H. Mauney

Studies on the uses of a gamma-ray pinhole camera were resumed. Although our original survey of the literature was not promising, information was later turned up on a technique which is capable of high-quality photographs. The objective in the investigations reported here was to develop the capability for making photographs that could accurately map radiation sources in the MSRE reactor cell for maintenance and experimental planning. The camera used in the studies is shown in Fig. 2.2. It consists essentially of a thick-walled lead box with an aperture that allows gamma rays and visible light to strike film a t the other end of the box. Sheet film is inserted in a slot using a standard 4- by 5-in. film cassette. The aperture is a thin section of aluminum, '/* in. in diameter, through which gamma rays can penetrate, pierced with a 0.016-in.-diam hole to admit visible light. There is no shutter exposures are long and are satisfactorily controlled by other means. In the tests described here, the lights were turned on and off, and the gamma ray source was moved into and out of the field. In a field application the film would simply be inserted and removed. Test photographs were made of a 20-curie lg21r source, located in a well-lit shielded room. Some were made with the camera only 12 in. from the source, giving a radiation level a t the camera of 110 r/hr, which is about the radiation level at the top of the MSRE reactor cell. Other photographs were made with the camera 15 ft from the source and other objects, which is typical of distances in the MSRE cells. Photographs were made with various exposure times, with both x-ray film and

FILM HOLDER SLOT

Fig.

2.2.

Pinhole Camera Used i n Tests.

visible-light film. Data and results are given in Table 2.2. T y o of the most significant photographs are Figs. 2.3 and 2.4. Figure 2.3 is a photograph on visible-light film exposed for 12 min. The source is a %-in. pellet inside a t-in.-diam capsule, held 12 in. from the camera, in front of a scarred wooden table leg. The poster is about 2% ft behind the source. Visible details were reproduced clearly, but the gamma radiation produced only a dim spot. Figure 2.4 is a photograph of the same setup, but on x-ray film exposed for 3 min. The source spot shows up very clearly, and the visible scene was reproduced well enough to identify the location of the source relative to other objects. In tests with the camera 15 ft from the source, the radiation a t the camera was only 0.5 r/hr, and the

23 Table

2.2.

Data on T e s t Photographs with Pinhole Camera

Exposure T i m e j i

Test .

Type Of Film”

Distance

Remarks Gamma Rays

Light

X ray

15 ft

6% min

Visible

15 ft

10 min

X ray

15 ft

10 min

Good spot

X ray

12 in.

30 sec

Good spot

X ray

12 in.

30 sec

45 s e c

Good picture, no spot

X ray

12 in.

3 min

Good picture, good spot

Visible

12 in.

3 min

1%min

Dim spot

Visible

12 in.

6 min

6min

Good picture; no spot

Visible

12 in.

12 min

Good picture, dim spot

Small spot, underexposed Good picture, no spot

- Fig. 2.4

- Fig. 2.3

“X-ray film was Kodak KK, “visible” film was Kodak Tri-X.

source spot was swamped by visible light when film was exposed simultaneously t o light and gamma rays. However, a very clear source spot could be produced on x-ray film exposed only to gamma rays. In this case an acceptable picture was obtained by superimposing two negatives: KK x-ray film exposed 6% min to gamma rays only and Tri-X visible-light film exposed 10 min. The source was hardly distinguishable on a composite print, but its location could be determined fairly well by comparison of the two negatives. The results indicate that the pinhole camera can be useful in surveying the MSRE reactor cell for gamma-ray sources, and plans are to photograph portions of the cell at the next shutdown.

2’5 FREEZE-FLANGE THERMAL CYCLE TESTS F. E. Lynch In the development of the freeze flanges for the MSRE, a prototype flanged joint was subjected t o 103 thermal cycles to determine the sealing and distortion characteristics in simulated reactor startups and shutdown^.^ These thermal cycles were typical of heating and salt-filling operations 4MSR Program Semiann. Progr. Rept. July 31, 1964, ORNL-3708, p. 180.

but were more severe than those encountered in the MSRE. A satisfactory leak-tight gas seal was obtained at both elevated temperature and room temperature. This number of cycles was greater than the number anticipated for the flanges in the MSRE. Thus it was concluded that the MSRE flanges should be satisfactory from the standpoint of sealing and distortion. Examination of the prototype flange during and after the 103 cycles showed no cracks such as would be produced by thermal fatigue. Predictions, based on low-cycle fatigue analyses, were that c r a c k s should be e x p e c t e d at about 300 test c y c l e s . The permissible number of cycles of various kinds on the reactor flanges was based on the same kind of analysis. but the permissible number was s e t a factor of 10 below the number a t which cracking would be e x p e ~ t e d .Now, ~ as reported in Sect. 1.2.3, the reactor flanges have reached 69% of the permissible life. Although they may not reach the permissible l i m i t , it was decided to reactivate the flange thermal cycle test. The purpose is to lend further confidence in the fatigue calculations, and plans are to continue thermal cycling the prototype

’P. N. Haubenreich et SJ., MSRE Design and Operations Report. Part V-A. Safety Analysis of Operation with 233U,ORNLTM-2111, p. 70 (February 1968).

Fig.

2.3.

Test Photograph on Visible-Light Film. Tri-X ortho film, 1921r source, 100 r/hr a t camera, 12-min ex-

posure, 5-min development i n x-ray developer.

Fig.

2.4.

T e s t Photograph on X-Ray Film,

x-ray film, 1921r source, 110 r/hr a t camera, 3-min simultaneous

exposure to light and source, 8-min development.

freeze flange until it cracks or until the MSRE is finally shut down. The Freeze-Flange Thermal Cycle Facility cons i s t s of a n upper and a lower tank with interconnecting pipe containing the test flange. In the

course of the test, molten s a l t is oscillated between the two tanks. Salt is forced to flow to the upper tank by pressurizing the lower tank with gas, and it returns to the lower tank by gravity when the pressure between the two tanks is equalized. A

ORNL-LR-DWG 52043A

SPRING,-LOADED RELIEF

SYSTEM-OVERPRESSURE REGULATING VALVE UPFLOW PRESSURE REGULATING VALVE

Fig.

2.5.

Freeze-Flqngs Thermal C y c l e Facility.

schematic of the t e s t facility is shown in Fig. 2.5. Figure 2.6 shows the cross section of the 5in. MSRE-type Hastelloy N flange. The two tanks remain heated at all times, with the molten s a l t in the lower tank when the system is not in operation. A t the start of a thermal cycle, heaters adjacent to the flanges are turned on, and the system is heated until the flange hub reaches a preset temperature. When this temperature is obtained, contacts within a temperature controller are closed, the lower sump is pres-

surized with helium, and a timer is simultaneously started. Level probes control'the flow of s a l t between the two tanks by alternately pressurizing the lower tank and equalizing the gas pressure. This oscillating flow of the salt between the two tanks continues until the preset t i m e on the timer elapses. The timer then turns off the pipe heaters, and the salt drains to the lower sump tank. A controlled cooldown is obtained by reducing the setting on the pipe heaters and reenergizing them for a period of time.

6.'

ORNL-LR-DWG 63248R2

W G E CLAMP-

BUFFER CONNECTION (SHOWN ROTATED) MODIFIED R-68 RINGGASKET

FROZEN S A L T SEAL-

w I &-in.-R

(TYP)

SLOPE 1:4

Fig. 2.6.

Cross Section of 5-in. Molten-Salt Freeze Flange.

The tan.. temperatures are maintained between 1300 and 1350OF with corresponding pipe temperatures of approximately 1300OF. The temperature a t the hub of the uninsulated flange at the t i m e the salt starts to oscillate is between 500 and 600OF. Eight to nine hours are required to preheat the pipe to obtain this hub temperature. The salt is then oscillated from one tank to the other for 5 hr. Upon completion of the oscillation time, the flanges are cooled down to approximately 130OF.

The original 103 cycles were c mpleted usin helium as the cover gas, but when operation was resumed we attempted to use argon. However, we returned t o helium when we found that the operating temperature could not be maintained at the desired level with argon. The history of the upper bore and ring temperatures during a typical thermal cycle using helium as the pressurizing gas is shown in Fig. 2.7. A typical radial thermal gradient during salt oscillation is shown in Fig. 2.8.

ORNL- DWG 68-5511

1400

1200

1000

800

a 2

w n

EI-

600

400

200

TIME (hr)

Fig. Helium.

2.7.

Uppe; Bore and Ring Temperature Characteristics of 5-in.

MSRE Freeze Flange; Salt Oscillated with

ORNL-DWG 68- 5512

I400

I200

1000

-P

800

600

I-

pump bowl, thus eliminating the need for the overflow tank used in the present MSRE pump installation. It also differs in having a longer unsupported length of shaft and a different fission gas stripper. During this six-month report period, work continued on preparation of the pump and test facility for test operation with molten salt.7 Installation of the pump tank in the test facility was completed. An electrical furnace for preheating was installed on the pump tank; various thermocouples and electrical preheaters were replaced on the test system, the pressure measuring devices on the salt flow venturi were replaced, and the necessary thermal insulation was applied to the test loop. Some delays were encountered in the assembly of the pump rotary element, and the assembly was not quite finished by the end of the report period.

2.6.2 MSRE Oil 0

4 6 8 IO DISTANCE FROM CENTER LINE OF PIPE (in.)

f i g . 2.8. Radial Temperature Distribution in !%in. MSRE Freeze Flange.

Shakedown of the second of two MSRE lubrication pumps that were refurbished* was completed, and the two pumps were returned to the MSRE to serve as spares.

2.6.3 Oil At this time, a total of 139 cycles have been completed. The outer surfaces of the flange assembly are visually inspected during and after each cycle. A complete inspection of the flange was made between cycles 103 and 104 when the facility was shut down. This inspection included dye-penetrant inspection of the internal flange face and bore surfaces as well as a dimensional check of the bore. These complete inspections will be repeated periodically through the duration of the test.

2.6 PUMPS

P. G. Smith

A. G. Grindell

2.6.1 hhrk 2

Fuel Pump

As reported previously,6 the Mark 2 fuel pump was designed to give more salt expansion in the

6MSR Program Semiann. Pro&. Rept. Feb. 28, 1967, ORNL-4119. p. 64.

Pumps

Pump Endurance Test

The oil pump endurance testg was continued. By the end of the period, the pump had run for 40,134 hr circulating oil at 160'F and 70 gpm.

7 M S R Program Semiann. Progr. Rept. Aug. 31, 1967, ORNL-4191, p. 45. 'Zbid., p. 46. 'MSR Program Semiann. Progr. Rept. Feb. 28, 1962, ORNL-3282, p. 55.

3. Instruments and Controls L. C. Oakes

curacy of the thermocouples in service on the salt systems. There have also been remarkably few failures due t o breakage, shorting, grounding, or apparent detachment from the surface being monitored. The thermocouples are Chromel-Alumel wires enclosed in a %-in. Hastelloy N or stainless steel sheath with magnesia insulation. Most junctions are grounded. The sheathed thermocouple stock was commercially manufactured t o ORNL specifications, and the junctions and disconnect seals were prepared at the MSRE by ORNL craftsmen. A total of 1071 thermocouples are installed a t the MSRE. Of these, 866 are on the salt systems: 351 on the circulating loops and t h e remainder on the drain tanks, drain lines, and freeze valves. Of the 1071 thermocouples, only 9 have failed in more than three years of service. Three others are unserviceable because of damage suffered during the final stages of construction when repair was too difficult to be worth while. A breakdown of t h e failures is given in Table 3.1.

3.1 MSRE OPERATING EXPERIENCE The instrumentation and control systems continued to function well, and the failure rate of components continued to decrease. Component failures did not compromise reactor safety or cause excessive inconvenience in reactor operation.

3.1.1 Safety System Components J. L. Redford Five more relays failed in the rod scram coincidence matrix. This made a total of seven failures in less than a year out of the 15 relays in the matrix. These relays are 115-v ac relays operating in a 32-v dc system. Relays more suited t o the service were procured and are ready to install. Numerous false trips on channels 2 and 3 were traced to intermittent false operation of core outlet temperature switches. Replacing t h e switches remedied the trouble. Other than test scrams, the control rods were scrammed only twice in the s i x months. The first scram was caused by the building power failure in October while the reactor was at full power. The other occurred while the reactor was a t 10 kw in November and resulted from circuit testing in search of a ground in the rod servo system. One of the three ionization chambers began producing a reverse current, and upon removal it was found that moisture had leaked into the cable.

'MSR Program Semiann. Progr. Rept. Aug. 31, 1967, ORNL-4191, pp. 22-23.

Table 3.1.

Failures Among 1071 MSRE Thermocouples

Through February 29, 1968

3.1.2 Thermocouples C. H. Gabbard As described in the last semiannual report' there has been very little change in the apparent ac-

Nature of Failure

Number

Damaged during construction Damaged during maintenance Lead opened during operation Abnormally low reading (detached?) Bad disconnect in reactor cell

3 2 3 3 1

Total

i LJ

3.1.3 Other Instruments and Controls J. L. Redford Two fission chamber failures occurred during the period. One was because of a leak in the cable, and the other has not yet been diagnosed. The failure in the fine position synchro transmitter on rod drive No. 2 appears to be an open rotor winding, as in the previous failure reported last period. With the transmitter in this condition, the position indicator on the console tracks normal changes but may lose 180' on rod scrams. Miscellaneous failures included the air flow switch on a radiator annulus blower. A new switch with a smaller paddle will be tried. The power supply to the single-point level switches in the drain tanks failed because of e cooling blower failure. The blower was replaced, and the power supply was repaired.

3.2 CONTROL SYSTEM DESIGN P. G. Hemdon Further additions and modifications were made to the instrumentation and controls systems as experience revealed the need or desirability of more information for the operators, improved performance, or increased protection. During the report period there were 26 design change requests directly involving instruments or controls. Seven of these required only changes in process switch operating points, sixteen resulted in changes in instruments or controls, one was canceled, and t h e remaining two were not completed. The more important changes are described below. To avoid further damage to wiring and equipment resulting from short circuits to ground, isolation transformers were installed in the power supply to the motor control circuits of the fuel sampler-enricher cable drive motor. This permits the entire circuit to float at 115 v above ground potential. Before this revision, one control power conductor was a grounded neutral. The floating circuit will operate normally when only one point is grounded, but a misoperation is likely to occur if two points become grounded a t the same time. To avoid two simultaneous grounds, grounddetecting lamps were installed, and all grounds will be. corrected as soon as possible.

The design of instrumentation and controls for a fluorinated-fuel-salt filter i n transfer line 110 between the fuel storage tank and the fuel drain tanks was started. Eleven mineral-insulated, Inconelsheathed thermocouples are attached to the salt filter and connected to the patch panel in the main control room. Temperature interlocks are provided to annunciate high temperatures in the filter gas space and to stop the transfer of salt to the fuel drain tanks by closing the fuel storage tank helium supply valve if the temperature of the top flange on the filter gets too high. A helium purge line is also provided a t the top of the filter to prevent salt from contacting the flange. Several additions to the fuel processing plant helium supply system are designed to prevent an accidental fill of the reactor during fuel transfer operations: a pressure relief valve l i m i t s the maximum pressure in the line supplying the fuel processing plant to 50 p i g , a capillary flow restrictor designed to the same specifications as FE-517 (the drain tank helium supply restrictor) limits the maximum flow rate in the salt filter purge line to a value less than 0.5 cfm, and a weld-sealed solenoid block valve is installed in the filter purge supply line. Safety-grade interlocks are designed to close the purge supply block valve and the fuel storage tank helium supply valve on an emergency drain or fill-restrict signal. A safety-grade jumper is provided around these interlocks. Plans are to use the decontamination cell to store traps loaded with UF, from the fuel fluorination. To ensure that water does not accumulate undetected in this cell, a bubbler-type liquid level measuring system identical to that in other equipment cells was designed for the decontamination cell. A level indicator is located in the transmitter room, and a high-level annunciator is provided in the auxiliary control room. The helium and standard gas supply systems on the fuel off-gas sampler proved to be inadequate during initial test operations. To obtain the precise flow control required for the purge and calibration operations and to provide better protection for delicate components, a new gas supply system was designed and installed. The volume of all lines to each piece of equipment was reduced to a minimum, special pressure regulators designed for use in chromatograph systems were provided, and additional pressure-relief valves were installed. A conductivity-type level detecting system was designed for the off-gas sampler containment box.

32 The instrument provides an alarm i f the box starts to fill with water. 3.3 MSRE NEUTRON NOISE ANALYSIS D. N. Fry R. C. Kryter J. C. Robinson The spectrum of inherent noise in the MSRE neutron flux signal has been examined in some detail. The data are collected through a special low-noise instrument channel and are digitally recorded on magnetic tape by the on-line computer. A program has also been prepared to permit direct digital analysis of the data tapes by the same computer immediately after the data are collected. A considerable amount of data was collected in run 14 with particular emphasis on the series of experiments a t 5 Mw. The most prominent feature of all the noise spectra is a peak a t about 1 hertz (see Fig. 3.1). In general, this peak increased in

amplitude under conditions where the fraction of circulating voids increased. Figure 3.1 shows two noise spectra which illustrate this effect. Other variations were also observed, and detailed correlations of the noise spectra with other reactor data are being attempted.

3.4 TEST OF MSRE ROD CONTROL SYSTEM UNDER SIMULATED 233U LOADING CONDITIONS S. J. Ball

R. J. Steffy, Jr.

Because of the smaller fraction of delayed neutrons from 233Ufission, the prompt power response to a sudden reactivity change will be about three t i m e s as large as for the present 235Ufuel loading. Conceivably this increased high-frequency

2Zbid., pp. 61-62.

ORNL-DWG 68-4017

Fig. 3.1. Typical Effect of Reactor Operating Parameters on the Spectrum of Inherent Neutron Fluctuations in MSRE. Reactor power = 5 Mw.

33 gain could cause enough overshoot following a control rod adjustment that the rod servo controller would hunt excessively. Thus it was desired to determine in advance if any changes would be necessary in the servo system to have it function properly with 233U fuel. A method was devised to test the capability of the servo system for controlling the reactor with 233U fuel. Basically it consisted in modifying the flux signal to the servo system so that the apparent response to a rod change closely resembled that from the 233U system. Appropriate resistors and capacitors would be wired around existing elements in the servo network (see Fig. 3.2) so that the flux input signal would have a high-frequency gain about three t i m e s normal but an identical steadystate value. Verification of the validity of this experiment and its implementation were assigned to an MIT

Practice School team (P. J. Wood, leader, and D. J. Roberts). The work reported below was carried out mainly by them. Before the experiment was conducted on the reactor, the method was tested by analog simulation. The simulator used essentially the same model of the MSRE that was used in the operator training simulator3 except that only three delayed neutron groups were used and that the external fuel circuit was represented by a third-order system. Three conditions were simulated: the MSRE with 235U fuel and with 233U fuel, both with an unmodified servo system, and the reactor with 235U fuel and the servo system modified by the proposed RC network. Figures 3.3 and 3.4 show the response of

3S. J. Ball, Simulators for Training Moltendaft Reactor Experiment Operators, ORNL-TM-1445 (April 1966).

ORNL-DWG 68-5513

c-201

R-249 R-2q2 3 meg A11

CALCULATION

k e g FLUX SIGNAL FROM COMP. ION CHAMBER

2 Pf

p k

200(f,

AMP.

, , O y f

>TO

la1 v

30 k R-250

SERVO AMP. LA& VV1

r: 2PfL 4 meg R- 251 R 259

ROD CONTROL RELAYS

34 ORNL-DWG 68-5514

I 233;

0.5

Fig.

3.3.

I .o TIME (mid

1.5

2 1)

Analog Simulation of 0 . 5 4 ~Lood Changes.

ORNL-DWG 68- 5515 7.5

235,,

7.0

6.5

7.0

233,, 233,,

0.25

0.50

0.?5

1.00

TIME (mid

Fig.

3.4.

Analog Simulation of 0.01% 8k/k Changes.

the three systems to a 0.5-Mw load change and a step reactivity change of 0.01% 6Wk respectively. It is apparent that the 235U+ RC system closely follows the 233Usystem response for fast changes. Table 3.2 further illustrates the success of the RC networks in making the 235Uresponse approach that of the 233Usystem. Thus the analog results showed that the modification of the 5U system response was approximately correct for highfrequency changes @ 0.3 radiaa/sec) and that the

proposed experiment on the MSRE should indeec test the servo system under conditions close to those existing with the 233Uloading. W i t h confidence established that addition of an RC network would provide a meaningful simulation, the experiment was performed on the reactor. After the proper RC elements were attached t o the circuit, the system was perturbed small amounts in three different ways: (1) a shim rod was inserted -0.5 in., (2) the radiator load was changed by

Toble

3.2.

Comporison of Flux Response to

0.01% 8k/k

Step lncreore in Reoctivity i n Three Systems on

Analog Computer

Difference Between 233u and 2 3 5 ~

Time After Reactivity Step (sec)

0. W. Burke

Difference Between 2 3 3 and ~ 2 3 5+ ~ RC

(%I

(%)

1 2 3

33 26 22

8 9 6 4

3.5 ANALOG COMPUTER STUDIES OF THE MSRE SYSTEM WITH 233UFUEL LOADING

0.5 Mw, and (3) the outlet temperature demand was changed by 3OF. After each of these tests the system was allowed to reach equilibrium and was then returned to its original state by reversal of the initial procedure. The ability of the controller to correct for large changes was tested by a 2 . 7 4 ~ power change (from 7.2 to “4.5). In only one test, the 3 O F outlet temperature demand change, did the regulating rad oscillate abnormally. When the demand was increased, the rod oscillated for about 25 sec (six withdraws and six inserts) but then steadied out without external interference. A similar occurrence was not observed when the temperature demand was decreased. The conclusion was that at least when the reactor operates a t high power, the present rad servo system will be adequate and will not introduce undue rod oscillations. Because the highest gain relative to the servo dead band occurs at lower powers, the situation may not be as favorable, but minor adjustments should be all that is required for 233Uoperation.

F. H. Clark

A s part of the analysis of the MSRE with 233Ubearing fuel, a detailed simulation of the system was s e t up on an analog computer. The simulation included the period-scram circuits that cause the control rods to scram after the period becomes shorter than 1 sec. This could not be conveniently included explicitly in the digital calculations because of the complex dependence of the scram signal on the initial neutron level, the gamma background, and the history of the flux and the period a s it approaches 1 sec. Results from these analog studies were combined with digital calculations to describe the behavior of the reactor under accident conditions. The findings of these coordinated studies are described in detail in Sect, 4.2. Some preliminary analog studies were made t o a s s e s s the performance of the servo controller with 233Ufuel based on estimated values of instrument t i m e lags and roddrive-motor inertial effects. It appears that the controller will be quite adequate when operating at the higher powers associated with the temperature mode of control. There was some indication of “hunting” at very low power levels while operating in the flux mode of control. These results support those obtained from the insitu test of the actual servo control system with the modified flux input (see Sect. 3.4). Plans were developed to measure control rod accelerations and coastdowns experimentally. These will be incorporated in the analog simulation to more accurately predict the performance of the control system with 233Ufuel.

4. MSWE Reactor Analysis 4.1 INTRODUCTION

excursions with a 233Ufuel salt were found to be associated with two postulated incidents. One incident results from the sustained withdrawal of the three reactor control rods at maximum rate with the initial neutron level very low, near source conditions. The other incident involves the gradual separation of uranium from the main stream of circulating salt during routine operation of the reactor, followed by the rapid return of the uranium in concentrated form to the core. We have performed digital simulation studies of t h e consequences of these incidents to determine both the inherent shutdown capabilities of the system (through temperature-reactivity feedback) and also the effectiveness of the reactor safety system in initiating control rod scrams to suppress the excursions. Though the latter is of principal concern in these studies, it is of interest t o obtain the approximate relations between these reactivity addition incidents and the inherent shutdown capabilities of the reactor, without action of the safety-scram system. In addition to these simulation studies, in order to further elucidate the nuclear safety consequences of changing to a 233U fuel salt, we have compared simulations of these incidents with 233U and with the present 235U fuel loading, the latter calculated using updated values of the MSRE neutronic characteristics obtained since the earlier nuclear safety studies. This comparison is described further in the results summarized in the following sections. A synopsis of the most important results of these studies is a l s o included in ref. 6 in context with general description 'of the safety of the reactor system with 233U.

B. E. Prince Reactor physics studies in support of the future operation of the MSRE with a 233U fuel loading were extended t o aid in evaluating the nuclear safety of the system. Use was made of results of previous computational studies of the neutronic properties of the system with 3U fuel, and emphasis was given to analysis of some abnormal situations which could conceivably lead t o nuclear excursions. To perform these studies, we used the reactor kinetics-digital simulation code ZORCH, developed for calculation of large transients in power, temperature, and pressure in the MSRE. The general approach taken in the safety studies was similar t o that used earlier for analysis of the nuclear safety with the present 5U-bearing fuel salt. In the evaluation of nuclear safety for the 233U loading, however, we have attempted where possible to take advantage of experience gained from both simulation studies and reactor operation with the present fuel salt. Thus, from the standpoint of magnitudes and rates of reactivity addition and initial reactor conditions, some incidents considered in the original safety analysis could either be judged unrealistic or less severe than others. As was the case with the present fuel loading, the potentially most severe nuclear

3 0 4

'MSR Program Semiann. Progr. Rept. Aug. 31, 1967, ORNL-4191, pp. 50-62. 2C. W. Nestor, Jr., ZORCH an ZBM-7090 Program for the Analysis of Simulated MSRE Power Transients with a Simplified Space-Dependent Kinetics Model, ORNL-TM-345 (September 1962). 3P. N. Haubenreich e t e l . , MSRE Design and Operation Report. Part ZZZ. Nuclear Analysis, ORNL-TM-730, pp. 122-56 (February 1964). 4S. E. Beall e t al., MSRE Design and Operation Report. Part v. Reactor Safety Analysis, ORNL-TM-732. p ~ 196-231 . (August 1964).

' B . E. Prince e t e l . , Zero-Power Experiments on the Molten-Salt Reactor Experiment, ORNL4233 (February 1968). ' P . N. Haubenreich e t al., MSRE Design and Operation Report. Part V-A. Safety Analysis o f Operation with 233U, ORNL-TM-211, pp. 40-69 (February 1968).

4.2 SIMULATION OF NUCLEAR EXCURSION INCIDENTS B. E. Prince

R. C. Steffy, Jr.

4.2.1 Uncontrolled Rod Withdrawal In this hypothetical accident, a nuclear excursion is produced by the sustained withdrawal of the

three control rods, with the reactor passing through criticality when the rods are near the position of maximum differential worth. The restrictions and control interlocks which would prevent this type of incident from occurring are described in ref. 6. In this section, we will summarize the results of digital simulations of the accident, including t h e most severe conditions which would result without consideration of the action of the safety system and the effect of this action in initiating rod scram and terminating the excursion.

CiJ

To define initial conditions in the rod-withdrawal incident, we have assumed that the reactor is loaded with excess uranium equivalent in reactivity t o the worth of one control rod (2.75% 6k/k). The reactor is initially subcritical by the insertion of all three rods, the fuel is circulating, and the core temperature is uniform at 1200OF. The initial fission rate is quite low, near a low l i m i t determined by the strength of the inherent a,n source in the 233U fuel (approximately 1 w). The three rods are now assumed to be withdrawn in unison at a speed of 0.5 in./sec. The reactor passes through criticality when the rods are approximately 28 in. above the position of maximum insertion, which is approximately 3 in. above the position of maximum differential worth. In this region the rod motion would correspond closely to a ramp addition of reactivity of 0.093%/sec. Without consideration of the action of the safety system, a short t i m e after criticality is reached, sufficient reactivity would be added to produce a prompt-critical excursion. This would be terminated by temperature feedback due to nuclear heating of the fuel. Then, if the rod withdrawal were continued, reactivity would be added at the above rate for approximately 16 sec after the t i m e of criticality. The rate would then decrease gradually until the rods reached their upper limits. After the initial prompt-critical excursion, the power and temperature would increase gradually to keep in step with the continued reactivity addition.

These characteristics are exhibited by results of digital calculations of the power-temperaturepressure excursion, shown in Fig. 4.1. For these calculations, initial steady-state conditions at a power level of 1w were assumed, and the rod withdrawal was assumed to begin a t t i m e zero, near the position of maximum differential worth. Both the temperature of the fluid at the hottest point in the reactor channel and the outlet temperature of the hottest channel are plotted in Fig. 4.1. (As described in ref. 2, the digital model includes a detailed numerical treatment of the axial convection of heat by fluid motion during the power transient.) The calculated pressure surge during the period of very rapid heating, also shown in Fig. 4.1, results primarily from the effects of acceleration of fluid in the outlet pipe. A description of the mathematical model for the pressure rise calculation is included in ref. 3. In order t o illustrate the importance of the differences in neutronics characteristics in changing the loading from the present 5U-bearing salt to a 233U salt, we also performed calculations similar to those of Fig. 4.1 for the present fuel composition. To simplify this comparison, we assumed that the reactivity addition rates were identical (0.093% Gk/k/sec) and also that the initial neutron levels were identical (1 w). In an accident simulation for the 235U fuel, the reactivity addition corresponding to rod withdrawal would actually be about 30%smaller than the above rate because of the smaller worth of the control rods. However, by making these assumptions, the nuclear excursions can be compared simply on the basis of the differences in delayed neutron fractions, temperature coefficients of reactivity, and prompt-neutron generation t i m e for the two fuels. The resulting nuclear transients calculated for the 235U fuel are shown in Fig. 4.2. It is evident that the rapid portion of the transient occurs later in t i m e with 23 ' U than with the 3U fuel, since more reactivity must be added t o reach the prompt-critical condition. As the reactivity addition continues beyond this condition, the presence of the larger amount of excess reactivity and the smaller temperature coefficient of reactivity of the 235U fuel lead to a larger temperature transient for the 235U case. In other words, the inherent shutdown mechanisms with 233Ubegin t o take effect sooner, must compensate for less reactivity, and compensate more effectively than with 235U.

4 800

4 700

4 600 c

7 5

(500

4400

t 300

1200

Fig. 4.1.

t2 TIME (sec)

Results of Uncontrolled Rod Withdrawal with No Safety Action.

The calculations shown in Fig. 4.1 for the 233U fuel indicate that the fuel temperature and core pressure rise incurred during the rapid portion of the transient would be inconsequential in terms of conservative l i m i t s specified in the nuclear safety analysis (340'F and 50 psi).6 However, it is clear that counteraction of the rod withdrawal would

233U fuel.

ultimately be necessary to prevent overheating of the core. To this aim, the safety system would actuate a rod s c r a m to limit the excursion. Scrams would be actuated by any of three conditions: (1) high flux level (over 11.25 Mw, if the fuel pump is running, or 11.25 kw if the circulation is stopped)', (2) positive periods less than 1 sec,

I800

(700

- 4600

cc 3 c

1500

W 0

I-

I400

4 300

4 200

-.-

400

20 -

- (5 u)

v, 300

200 6

8 TIME (sec)

4 00

TIME ( s e c )

Fig. 4.2.

Results of Uncontrolled Rod Withdrawal with No Safety Action.

235U fuel.

40 4.2.2

and (3) reactor outlet temperatures greater than 1300째F. In the incidents of the type considered in these studies, the first two are the active mechanisms. In all our simulation studies, we have assumed (1) that only two of the three control rods drop on request, (2) a time delay of 100 msec occurs between scram signal and the start of rod drop, and (3) the rod acceleration is'l0 ft/sec2. Figure 4.3 shows the results of actuating the rod scram due to increases in the neutron flux above the 11.25-Mw level. This mechanism would be effective in limiting the nuclear excursion to inconsequential proportions. Actually, a safety scram would occur earlier than indicated by Fig. 4.3, due to reactor periods smaller than 1 sec. The digital calculations for this transient showed that a 1-sec period is reached a t approximately 1.4 sec, a 500-msec period at 2.4 sec, the 11.25-Mw level at 5.3 sec, and the maximum power, limited by temperature feedback, at 5.8 sec. Thus, calculations based on quite conservative assumptions of the actual t i m e of actuation of the period scram showed that the power transient would be a t least two orders of magnitude below that shown in Fig. 4.3.

Return of Separated Uranium to the Core

The inclusion in the nuclear safety calculations of an incident involving the gradual separation of uranium from the circulating salt and its rapid concentration and return to the reactor core is based on the existence of two possible avenues for c h e m i c a l separation of uranium from the molten fluoride salt. One avenue requires the loss of enough fluorine from the salt to ultimately result in the production of metallic uranium. The other requires the gross contamination of the salt with enough oxygen or water vapor to result in the production of ZrO, and finally UO,. Precautions are taken to avoid contaminations, and evidence accumulated from chemical analyses during operation with the present 235U-bearing fuel salt has indicated no trends in either of these directions toward uranium separation. Although we expect these trends to continue when the reactor is operated with 233U, the existence of these remote possibilities motivates the analysis of a hypothetical nuclear incident of this type. To define the conditions for the incident, we have assumed that, during routine operation of

ORNL-DWG 68-970

1220

MAXIMUM FUEL TEMPERATURE

I-

a a W

g 4240 W

I-

4200

TIME ( s e d

Fig. 4.3.

Results of Uncontrolled Rod Withdrawal w i t h Scram a t

11.25 Mw. 233U fuel.

the reactor a t some steady power level, uranium is gradually removed from the main circulating stream and is accumulated in low-velocity regions. At the same time, the regulating rod is automati-

cally withdrawn to compensate for the reduction in uranium concentration. W e then assume that, by some mechanism, this concentrated uranium is suddenly dispersed throughout the 10 ft3 of fuel salt in the lower head of the reactor vessel and is carried upward through the channeled region of the core by the fluid motion, producing a nuclear excursion. It is clear that the assignment of a single magnitude or rate of addition of reactivity in this incident is not possible, as was the case with simulation of the uncontrolled rod withdrawal. Thus, it was necessary t o study the consequences of the uranium separation incident as a function of the amount of uranium (or reactivity) lost and also of the initial reactor operating conditions. The principal a i m of the analysis was t o compare the magnitude of t h e reactivity loss which would be required to produce severe rises in temperature

and pressure during the excursion associated with the return of all the lost uranium with the magnitude of anomalous reactivity loss which could be detected by routine computer monitoring of the reactivity balance. As in the case of t h e uncontrolled rod withdrawal incident, the nuclear excursions were simulated both with and without the safety scram of the control rods. Although the magnitude of reactivity loss was considered as a parameter in the analysis, for the conditions defined above, the timedependent shape of the reactivity addition during the uranium return would be largely determined by the fluid dynamics of the core and by the spatial distribution of nuclear importance. Figure 4.4 shows the time dependence of the reactivity calculated from the flow velocities observed in the MSRE hydraulic mockup vessel and normalized t o Ak,, the magnitude of the original reactivity loss. Equivalently, &, would be the reactivity effect of all the separated uranium if it were uniformly dispersed throughout the 70 ft3 of salt in the loop rather than con-

ORNL-DWG

Fig. 4.4.

8 TIME (sec)

68-974

Time Dependence of Reactivity Addition Due to Sudden Resuspension of Uranium in Lower Head of

Reactor Vessel.

i i

I I I

j I

/ j !

42 centrated in the salt moving through the core. From Fig. 4.4, the maximum total reactivity which could be added to the core is 4.9 Ako, and the maximum rate of addition is 1.35 Ako/sec. In choosing the initial reactor operating conditions for this incident, we have assumed that the lowest initial power level would be 1 kw and that the maximum power would be 7.2 Mw. The former value is about ten t i m e s smaller than the lowest steady power a t which the reactor would be routinely operated. The initial inlet fuel temperature was assumed to be 1200'F in all the calculations. For a typical calculated incident of this type in which there is no safety rod scram, Fig. 4.5 shows the transient behavior of the fuel salt temperature at the hottest point in the core and also the transient salt temperature a t the outlet of the hottest channel. Here, the initial power level is 1 kw, and the original reactivity loss Ako is 0.25% in multiplication factor. Associated with the rapid rise portion ( t 2 3.5 sec) of the curve for maximum fuel temperature are promptcritical excursions in power and core pressure qualitatively similar to those shown in Fig. 4.1. The maximum pressure rise for this case was about 39 psi. In Fig. 4.5, the temperatures ultimately reach a maximum and

then decrease as the concentrated uranium is again removed from the core. The dependences of these peak pressures and temperatures on the initial power level and on the magnitude of Ako are shown in Figs. 4.6 and 4.7. No safety rod scram was assumed for these calculations. Figure 4.6 shows that, for the case of Ako = 0.25%, the worst conditions for the pressure surge occur a t low initial power levels. This prompt rise in pressure results mainly from expansion of the core fluid and consequent fluid acceleration effects, which depend on the rate of increase in the power level. At higher initial power levels the nuclear heating of the core fluid tends t o slow the rate of rise in power earlier in the course of the transient. The dependence on Ako of the maximum temperature rise of the outlet of the hottest fuel channel i s shown in Fig. 4.7 for the maximum and minimum initial power levels assumed: 7.2 Mw and 1 kw. For values of Ako larger than about 0.25%, the temperature rise is nearly independent of initial power. These cases correspond to reactivity additions well above prompt critical, in which the graphite temperature change is relatively s m a l l and the fuel temperature increase depends on the integral of the power. At the opposite extreme, for

, 1700

4600

TEMPERATURE

I HOTTEST CHANNEL

I !

4300

1200

TIME (sec)

Fig.

4.5.

Temperature Excursion Caused by Sudden Resuspension of Uranium Equivdlent to

formly Distributed.

I n i t i a l power,

1 kw; no safety oction.

0.25% 8 k / k if Uni-

ORNL-DWG

68-973

0.01

0001

Fig. 4.6. to 0.25%

0.1 I N I T I A L POWER LEVEL(Mw)

Effect of I n i t i a l Power on Peak Pressure Rise Caused by Sudden Resuspension of Uranium Equivalent

8k/k i f

Uniformly Distributed.

No safety action.

ORNL-DWG 68-972

600

120

500

400

. c

-.u)

v, LL

60 w

LL 3 v)

3 LL a

200

100

Fig. 4.7.

0.4

0.2

0.3

0.4 Aka, EQUIVALENT UNIFORM REACTIVITY (70A k/k)

0 0.5-

E f f e c t of Magnitude of Reactivity Recovery on Peak Pressure and Temperature Rises During a Uranium

Resuspension Incident with

233U Fuel. No safety action.

44 small values of the reactivity (Ak, =< 0.035%), the reactor remains below prompt critical at the peak of the reactivity addition, and delayed neutron effects tend to control the response characteristics. The dependence of the magnitude of the pressure rise on Ak, is also shown in Fig. 4.7. The curve corresponds to an initial power level of 1 kw; on the same scale, the pressure rise corresponding to an initial power level of 7.2 Mw would be quite small. The preceding calculations take no account of the control rod safety scram action in suppressing the nuclear excursion. An indication of the influence of the initial reactor conditions and the magnitude of reactivity addition on the action of the safety scram system is provided by the curves in Figs. 4.8 and 4.9. For Ak, = 0.25%, Fig. 4.8 shows, as a function of the initial power level, the relative sequence in time of achieving various conditions during the transient: (1)a 1-sec period, (2) a f/,-secperiod, (3) high neutron level (11.25 Mw), and (4) peak power conditions with only inherent shutdown by temperature feedback effects. This figure indicates that a scram due to periods

ORNL- WG 68-5516

LJ c

0.01

,001

F i g , 4.8.

0.1 4 INITIAL POWER LEVEL (Mu)

Effect of I n i t i a l Power L e v e l on Times to

Reach Scram Conditions and Peak Power Conditions with

233U Fuel.

Ak0 = 0.25%.

smaller than 1 sec would considerably precede the conditions of high neutron level over all but the highest indicated initial operating power. At these high power levels, the smaller fractional increment in power required t o reach 11.25 Mw reduces the time interval relative t o that of the peak temperature conditions.

ORNL-DWG

68-5517

P rtw- 3 2

0.4

0.2 Aho

Fig. 4.9.

0.3

, EQUIVALENT

0.4

0.5

0.7

0.8

UNIFORM REACTiVlTY (% 6 k / k )

Effect of Magnitude of Reactivity Recovery on Times to Reach Scram Conditions and Peak Power

Conditions with

233U Fuel. I n i t i a l power, 1 kw.

u 8

In an analogous fashion, Fig. 4.9 shows the dependence of this t i m e sequence on the magnitude of Ako for an initial power of 1kw. Although the times to obtain each condition following the start of the incident are reduced, the t i m e increments between the curves tend to become constant as increases. A s would be expected, when the safety rod s c r a m is taken into account, the magnitude of Ak, required to produce substantial rises in temperature and pressure is increased. Figure 4.10 shows the effect of the scram of two rods, initiated by the 11.25-Mw level trip, in reducing the temperature and pressure rises associated with large reactivity additions. The curves shown correspond to an initial power level of 1kw. For the scales shown, temperature-pressure rises corresponding to 7.2 Mw initial power would be negligible. A s evidenced in Figs. 4.8 and 4.9, for the low initial power levels, scrams due to short period would actually precede the high neutron level condition.

Aka

Thus the results shown in Fig. 4.10 may be u s 4 as conservative upper l i m i t s on the temperature-

pressure rises for these reactivity magnitudes. Based only on these limits, recoveries in uranium corresponding to Ak0 5 0.78% could occur without exceeding the 340째F maximum temperature rise criterion for limiting thermal stresses to safe v a l u e s 6 In order to increase this level to a more realistic but still conservative value, a detailed simulation of the action of the short-period-scram mechanism was required. These results are described later in this section. Simulation calculations corresponding to the results shown in Figs. 4.7 and 4.9 were also performed for the present 5U fuel salt in order t o directly compare the responses of the MSRE to an incident of this type with the two fuel loadings. The normalized reactivity-addition curve shown in Fig. 4.4 can also be used for the 235Ufuel, since the core hydraulics are unchanged and the spatial distributions of nuclear importance are very nearly

ORNL-DWG 68-974

500

100

400

-.-

-a v)

300

a 3

I-

200

3 W

E I-

400a 0.4

0.5

0.8 A k o , EQUIVALENT UNIFORM REACTIVITY (% A k / k

0.6

0.7

0.9-

Fig. 4.10.

Effect of Magnitude of Reactivity Recovery an Peak Pressures and Temperatures During o Uranium

Resuspension Incident with

233U Fuel. Rod scram a t 11.25 Mw; i n i t i a l power, 1 kw.

46 equal for each fuel. The results of these calculations are summarized in Figs. 4.11 and 4.12. It can be seen from Fig. 4.11 that the temperature rise corresponding t o a given value of Ak, is higher for the 5U fuel salt, in approximately the inverse ratio of the fuel-temperature coefficients of reactivity for the two loadings.’*6 Because of the larger delayed neutron fraction for 5U,a larger addition of reactivity is required to reach the prompt-critical condition. This is evidenced in Fig. 4.11 by the larger value of Ak, required to produce substantial nuclear heating, starting from a low initial power level. In Fig. 4.12, i t is evidenced by the longer t i m e s following the start of reactivity addition which are required t o reach the specified conditions. In general, however, there are no important differences in the responses in this nuclear incident which could have a deter-

ORNL-DWG

68-5548

460

4 40

I20

400 ;.

?? 80 P 3 v) v)

Fig. 4.11.

0.I

0.4 0.5 Ak0 ,EQUIVALENT UNIFORM REACTIVITY (%86k/k 1

-0

0.2

0.3

E f f e c t of Magnitude of Reactivity Recovery

on Peak Pressures and Temperatures During a Uranium Resuspension Incident with action.

235U Fuel. No safety

mining detrimental effect on nuclear safety for either fuel salt. In order to further investigate the upper l i m i t s of reactivity addition which could be tolerated in this type of incident with 3U fuel, some analog simulation studies of the MSRE were made.7 The analog model included a representation of the dependence of the period scram signal on the initial neutron level, the gamma background, and the history of the nuclear transient prior t o scram. In general agreement with the results shown in Fig. 4.8, the analog calculations showed that, for initial power levels above about 7 Mw, the high neutron level trip would actuate t h e s c r a m signal; a t less than 7 Mw, the period trip would actuate the signal. For Ak, between 0.9 and 1.2%, the t i m e to start a rod scram varied between 300 and 800 msec, with the shorter times corresponding to higher Ak, and low initial power. Typical reactivity inputs in these cases are shown in Fig. 4.13. Before a rod scram, the input is very s m a l l and the resulting power transient is also small. This implies that the excursions of potential harm are those that could overcome the available shutdown worth of the rods. In these analyses, this was assumed t o be 4.0%, corresponding to the worth of two rods, dropping from an initial position of 44 in., in the “shadow” of the third rod over their entire length of travel. Thus, reactivity additions large enough t o cause a secondary nuclear excursion correspond to Ak, > 0.82%, as seen from Fig. 4.13. For these large additions with period scram, the power-temperature transients were calculated both with the analog model and with the digital program. Typical results are given in Figs. 4.14 and 4.15. It may be seen that the two models were in close agreement in calculating powqr transients (Fig. 4.14) but that there were some notable differences in the temperature calculations (Fig. 4.15). The analog model tends to reach a peak temperature faster than the digital calculation, but at a lower magnitude. These m o d e l s assume somewhat different mechanisms for heat transport within the core. The analog model uses a 27-lump representation of the MSRE core, within which “well stirred” conditions are

70. W. Burke and F. H. Clark, ORNL, personal communication (January 1968). ‘S. J. Ball and T. W. Kerlin, Stability Analysis of the MoZtenSaf t Reactor Experiment, ORNL-TM-1070 (December 1965).

ORNL-DWG

60-5519

0 u) Q

w 4

I-

TIME TO REACH PEAK POWEF

- 1

TIME TO REACH 44.25 MW

TIME TO REACH 4-sec PERIOC

0 0

Fig.

4.12.

04 .

0.2 0.3 0.4 0.5 0.6 0.7 Ak, ,EQUIVALENT UNIFORM REACTIVITY ("/o 6 k / k 1

Effect of Magnitude of Reactivity Recovery on l i m e s to Reach Scram Conditions and Peak Power

Conditions with 235U Fuel. I n i t i a l power,

j 1

kw.

assumed, implying immediate heat transport within each lump. In contrast, the digital model assumes no axial conduction or mixing.' Peak hot-channel outlet temperatures a s a function of both Ak, and initial power level are shown in Fig. 4.16. The analog model indicated that a Ak, of about 1.07% would be required before the tolerable temperature l i m i t s were exceeded. A more conservative value of 0.96% was obtained from the digital calculation. Both models indicated that peak temperature was not very dependent on initial power level, particularly for the higher values of The magnitudes of the peak pressure rises calculated for these large reactivity additions with period scram are shown in Fig. 4.17. These results also exhibit t h e inverse relation between

Aka.

0.8

initial power level and peak pressure discussed in the preceding sections. In order to obtain an upper limit on the maximum pressure rise, an initial power level of 100 w was chosen. Projection of the calculated data to 100 w initial power shows that a Ak, of 0.95% could be tolerated without exceeding the conservative 50-psi limit in pressure rise assigned in the safety analysis.

4.3 DETECTION OF ANOMALOUS REACTIVITY EFFECTS

B. E. Prince During routine operation of the MSRE, calculations of the complete reactivity balance are made at S-min intervals by the on-line digital computer.

48 ORNL-DWG 68-5522

ORNL-DWG 68-5520

500

-ae

400

e 300

Is W

W 0

-I

g 200

> e_ > -2 F 0 a W

100

3 t

-3

I-

-I I-

2 -100

-4

Fig. 4.13.

200

H X a

TIME (sed

Time Dependence of T o t a l Reactivity Input

100

with Combined Uranium Resuspension and Rod Scram

Due to Short Period.

-100

Ako= 0.9% I I 8 IO

TIME (sed ORNL-DWG

2000

68-552t

Fig. 4.15.

Time Dependence of Hot-Channel Outlet

Temperature, Calculated with Both Analog and D i g i t a l

IO00

Models.

I n i t i a l power = 7.5 Mw.

500 200

400

II /I

5 B

\, I

I I

4 .O

0.5

O2 0.4

2 4

TIME

Fig.

Experience with these calculations, accumulated during a substantial part of the operation with the present 5U loading, h a s been described in ref 9. At steady power, temperature, and pressure, we have found that the variation in consecutive reactivity balances is about f0.01% Sk/k. This value represents the probable limit of precision in the technique. Longer term variations in the residual reactivity can be due either to systematic errors in calculation of known reactivity effects or to anomalous changes whose detection is important, such as the hypothetical situation considered i n the preceding section. The long-term reactivity variations which have been observed with the present fuel salt have been very small. A systematic variation as small as 0.05% 6k/k would

4.14.

Time Dependence of Power L e v e l for a

Uranium Resuspension Incident with Initial Power

(sed

= 7.5

Mw.

h0= 1.2%

and

'5. R. Engel and B. E. Prince, The Reactivity Bafances in the MSRE, ORNL-TM-1796 (Mafch 1967).

4800

, , , , ,

ORNL-768-5523

400

E G 6

1600

W K

350 4400

E I-

'3 0

300

4200 4800

-.-E 250

J W

4600

E 3a

P W

4400

200

W (0

(200 1 0.7

0.8

F i g . 4.16.

Dependence of Peak Hot-Channel Outlet

Temperature on i

0.9

L\ko and

1.0 A ko (So 8 k/k)

1.4

1.2

f.3

450

100

I n i t i a l Power L e v e l for Large

Reactivity Additions with Short-Period Scram. 50

be clearly observable by the computer-calculation monitoring. A systematic long-term change larger than 0.10% &/k should be easily separable from the effects of any miscalculation of known reactivity changes. When the present fuel loading is changed to 23 'U, the fissile-uranium-concentrationreactivity effects will be larger, * by about a factor of 2. This means that the reactivity balance calculations can be expected to be more sensitive to variation in uranium content. For example, removal of 1 g of 233Ufrom the salt, by burnup or any other mechanism, may be expected to produce a reactivity reduction nearly four times that of 1 g of 2 3 5 U in the present fuel. An abnormal reactivity loss of 0.10% Sk/k would correspond approximately to a 0.26% reduction in uranium concentration (of isotopic composition to be used for the new loading), which is total l o s s of about 80 g of uranium from 70.5 ft3 of circulating salt. On the other

0.8

0.9

t .O

4.4

4.3

4.2

A k,, (%I

Fig. &to

4.17.

Dependence of Peak R e s s u r e R i s e on

and I n i t i a l Power L e v e l for Large Reactivity

Addi-

tions with Short-Period Scram.

hand, the upper limit of 0.95% 6k/k obtained in the studies described in the preceding section corresponds to approximately 2.4% in uranium concentration (750 g of total uranium). Thus, there is considerable latitude in establishing an administrative safety limit on anomalous reactivity effects to prohibit nuclear operation when the residual term in the reactivity balance becomes too large.

Part 2.

MSBR -

Design and Development I

sy-

* - _ - - x L

*-_I_

R. B. Briggs The primary objective of the engineering design and development activities of the MSR program is to design a molten-salt breeder experiment (MSBE) and develop the components and systems for that reactor. The MSBE is proposed to be a model of a large power breeder reactor. Most of the design and development effort is being spent on studies of reference designs of 1000 Mw (electrical) and larger power plants to provide the criteria and much of the basis for design of the MSBE. These studies were begun in September of 1965. The results are reported in four semiannual progress reports through August 1967 (refs. 1-4) and in ORNL-3996, Design Studies of 1000-Mw(e) Molten-Salt Breeder Reactors. Based on the studies, a two-region reactor with fissile and fertile materials in separate fuel and blanket salt streams was chosen a s the m o s t promising approach to the breeder. Factors related to fuel and blanket processing were the most important considerations in reaching this conclusion. A fluoride volatility process was available for separating uranium efficiently from fuel and blanket streams. A distillation process had been conceived for separating the 'LiF and BeF, salts from rareearth fission products in the fuel salt. Separation of 233Pa from the blanket salt was not essential to achieve breeding, although such separation appeared to be possible and offered promise of improving the breeding performance and the economics.

'MSR Program Semiann. ORNL-3936. p. 172. ,MSR Program Semiann. ORNL-4037. p. 207. 3MSR Program Semiann. ORNL4119, p. 174. 4MSR Program Semiann. ORNL-4191. p. 63.

Progr. Rept. Feb. 28, 1966, Progr. Rept. Aug. 31, 1966, Progr. Rept. Feb. 28, 1967, Progr. Rept. Aug. 31, 1967,

The design studies on the large two-fluid breeder reactors reached a reasonable stopping point in September of 1967. The design power for the MSBE had been selected as 150 Mw (thermal), the objectives and program for development of that reactor had been summarized in ORNL-TM1855, and the programs for development of components and systems had been outlined in ORNLTM-1855 and ORNL-TM-1856. Preparations were being made to begin the conceptual design and the development for the MSBE when two new developments caused us to delay that work and to continue with the design studies of large power plants. One of the new developments was evidence that 233Paand possibly the rare-earth fission products can be separated from the mixed thorium-uranium fuel salt of a one-fluid reactor by reductive extraction methods employing liquid bismuth. The second new development was the calculation that the leakage of neutrons from the core of a one5uid reactor of modest size can be reduced t o desirable levels by increasing the ratio of fuel salt to graphite to produce a greatly undermoderated region in the outer part of the core. By optimizing the distribution of salt, the reduction in leakage can be accomplished with a low specific inventory of fissile material. These developments, when combined with the greater simplicity and potential reliability, make the one-fluid breeder a more attractive concept than the two-fluid breeder, which must rely on graphite piping in the core to separate fuel and blanket streams. Our current design studies of large one-fluid breeders are aimed at describing those reactors sufficiently for us to select a s i z e and a design basis for a one-fluid breeder experiment. Development work is limited to analyses in support of the design and to experiments that are relevant to one-fluid and two-5uid reactors.

5. Design E. S. Bettis

5.1 GENERAL E. S. Bettis

The single-fluid reactor has the advantage that the graphite functions only a s moderator. In principle, i t can be present in the form of long bars with no firm connections a t top or bottom; the bars can be removed individually or in groups; the lifetime of the graphite should depend only on the bulk changes in dimensions that result from irradiation. There are also subsidiary advantages related to the amount and arrangement of the reactor equipment. These factors combined to cause us to look seriously at the design of a single-fluid reactor when there was evidence that fuel processing could be provided to make a singlefluid reactor a breeder. In the beginning we expected that the reactor would have to be a 20-ft-diam right cylinder or larger in order for the neutron leakage to be acceptably low. A power output of 2000 Mw (electrical) or greater per reactor then became necessary in order to achieve a low specific inventory. We believe that power plants of this size will be common in the 1980â&#x20AC;&#x2122;s when large molten-salt reactors could be built, so,we based our design studies on a 4444 Mw ( y r m a l ) , 2000 Mw (electrical) plant. W e have sidce found that zoning the core permits one ta obtain good breeding performance from 1000 Mw (electrical) and possibly from smaller reactors. The design studies reported here are for the 2000 Mw (electrical) reactor, but the performance of 1000 Mw (electrical) and 2000 Mw (electrical) reactors is compared in the reactor physics section.

R. C. Robertson

Our version of a 1000 Mw (electrical) power plant based on the two-fluid MSBR was described in considerable detail in our last semiannual report. W e selected the modular concept in which the plant contained four reactors, each with the capability for generating supercritical steam equivalent to 250 Mw (electrical). The core of each reactor was designed to accommodate for a reasonable time the dimensional changes produced by irradiation of the graphite a s those changes were deduced from the most recent experimental data. This led to the u s e of graphite fuel cells in the form of reentrant tubes that were brazed to m e t a l pipes. These pipes were welded into fuel supply and discharge plenums in the bottom of the reactor vessel. The fertile salt filled the interstices between fuel cells and a blanket tegion around the core. This reactor presented two major problems. The graphite in the core had to be replaceable for a variety of reasons, and the most acceptable way that had been conceived for replacing the graphite involved replacing the entire core and reactor vessel. Periodic replacement of the reactor vessel appeared to be economical i f the frequency were governed entirely by radiation-induced dimensional changes in the bulk graphite. There was, however, concern that mechanical failure of individual graphite cells or graphite-to-metal joints could cause the replacement frequency to become excessive.

5.2 FLOW DIAGRAM R. C. Robertson

H. L. Watts

The flow diagram for a one-fluid reactor plant with a capacity of 4444 Mw (thermal) and 2000 Mw

â&#x20AC;&#x2DC;MSR Program Semiam. Progr. Rept. Aug. 31, 1967, ORNL-4191, p. 63.

51 i

ORNL-DWG 68-4192

VENT OFF-GAS HOLD UP PRESSURE

COOLING COOLING COOLANT SALT PUMP

i 4

0 TO CHEMICAL PROCESSING

FREEZE VALVE PROPORTIONAL FLOW VALVE SEPARATOR -GAS FUEL

- COOLANT

----

STEAM

**STEAM GENERATING UNIT NO. 2 we+ STEAM GENERATING UNIT NO. 3 **wt STEAM GENERATING UNIT NO. 4

Fig.

5.1.

@ @ @ @ @ @) @ @

1300 t300

1150 850 It50 1150 1150 850 1150

200 50 75 75 75 0-20 16.2 16.2

Flow Diogram for 2000 Mw (Electrical) Station.

TEMP OF 850 850 600 1000 1000 700 950 1250 600 100

FLOW 75 f . . sec 10 2.56 x IO6 lb/ hr 2.56 x IO6 1.26 x lo6 1 . 2 6 ~IO6 4 ft3/,ec 4 420,000 f t Ymin 420.000

Table

5.1. Estimated Properties of F u e l and Coolant Salts for One-Fluid Breeder Reactors Fuel Salt at 1050'F

Composition, mole % Specific heat, Btu lb"

OF-'

Density, lb/ft3 Thermal conductivity, Btu ft-' hr-' Viscosity, lb ft"

OF-'

hr-'

Liquidus temperature, OF

(electrical) is shown in Fig. 5.1. This diagram is similar to those shown previously for the twofluid reactor with the obvious omission of the fertile salt circuit. The fuel s a l t for the one-fluid reactor contains both fissile and fertile atoms in 'LiF-BeF, carrier. The composition and our estimates of the properties are shown in Table 5.1. In the flow diagram this salt is circulated by four pumps through a common reactor vessel. Each pump circulates approximately 27,000 gpm of salt through the reactor and a heat exchanger. The salt enters the reactor a t 1050OF and leaves a t a mean temperature of 1300OF. The coolant salt - sodium fluoroborate - is pumped in four heat transfer circuits, one for each fuel circuit. Each pump circulates 53,000 gpm of coolant salt through a primary heat exchanger and through several superheaters and reheaters. The coolant salt enters the primary heat exchanger a t 950°F, flows to the superheater at 1150°F, and leaves the superheater a t 850OF. Since 850°F is below the liquidus temperature of the fuel salt, some 1125OF salt is bypassed around the superheater and mixed with the exit salt so that the temperature of the coolant returning to the primary heat exchanger is 950OF.

Each coolant salt loop has four superheaters, making a total of 16 units in the plant. There are two steam reheaters per coolant salt loop, a total of eight units in the plant. Steam enters the superheaters a t 700OF and exits a t 1000'F. Steam enters the reheaters a t 6W°F and is heated to l O O O O F for return to the turbine.

Coolant Salt at 8W°F

BeF2, LiF, ThF,, UF4 (20, 67.7, 12, 0.3)

B F 3 NaF (48. 52)

0.2 9

0.37

223

125

0.49

0.46

930

71 5

5.3 PLANT LAYOUT C. E. Bettis H. M. Poly c. w. Collins J. R. Rose W. K. Crowley W. Terry H. L. Watts With the new reactor and the higher power, our entire concept of the overall power breeder plant has to be reevaluated. Such concepts a s the basic idea for the thermal shield and the double containment are still valid, but we have devoted only a little t i m e to the layout of the new plant and have not examined those details. A plan view of a possible arrangement of the various cells is shown in Fig. 5.2. The reactor vessel, four heat exchangers, and four fuel pumps are located in the reactor cell. This cell is circular and is about 52 f t in diameter by 47 f t deep. It has a thermal shield to prevent overheating of the concrete. It also has double containment, with a pump-back system whereby the integrity of the containment is constantly monitored. Four steam-generating cells are located symmetrically in relation to the reactor cell. The cells are approximately 33 ft wide by 46 ft long by 43 ft deep. They contain only coolant salt and steam and therefore have no need for a thermal shield or double containment. These cells are isolated from the reactor cell and from *$e highbay area by bellows seals around pipes that com1 municate with those areas. The fuel drain tank is in a cell all its own. This cell is below the level of the reactor cell i

FUEL SALT DRAIN TANK

INSTRUMENTATION

CELL\

COOLANT SALT DRAIN CELL

COOLANT SALT

43 ft-Oin.

49ft-Oin.

43 f t - 0 in.

DRAIN CELL\

26fl-Oh.

;RADE)

-REHEATER -STEAM GENERATOR

33ft-Oin.

429ft-0in

-COOLANT SALT

-I-

PUMP

33ft-Oin.

- CHEMICAL

-I----

PROCESSING CELL

48 f t 0 in.

L-J-d

, I -

//-7-

OFF-GAS PROCESS CELL

I N S T R U M E N T A ~ CELL

Fig. 5.2.

62ft-Oh-4

\HOT

CELL

Plan View of Cell Arrangement.

in order for s a l t to drain by gravity from the reactor into the drain tank. Double containment is provided since radioactive salt is stored there. The drain tank cell is about 28 ft wide by 49 f t long by 38 ft deep. It is isolated from the reactor cell by bellows seals around the communicating salt lines. The coolant salt is stored in a separate cell about 26 ft wide by 43 ft long by45 fl deep. This cell needs little shielding and need be sealed only sufficiently to prevent excessive heat loss. The off-gas cell is approximately 18 ft wide, 63 ft long, and 43 f t deep. This cell must be doubly contained, but the ambient temperature is around 100Â°F, so it needs no electrical space heaters or thermal insulation. Cooling of the gas holdup tanks and the charcoal absorber beds is done by water which comes from the plant feedwater system. The chemical processing cell is about 18 ft wide by 63 ft long by 43 ft deep. This cell must have double containment, but it too does not require space heaters or thermal insulation. In this cell the pieces of apparatus are heated and cooled individually. On the cell plan we show two cells for instrumentation u s e s such as test points, junctions, instrument air sources, etc. These cells have approximate dimensions of 13 by 26 by 43 and 8 by 48 by 43 ft. The off-gas, chemical processing, and instrumentation cells shown in Fig. 5.2 are meant t o be only representative since no serious design effort has been put on them. The reactor, fuel drain tank, and s t e a m cells have been laid out sufficiently to show the arrangement of the components in some detail. These cell arrangements are shown in Figs. 5.3 to 5.5. Our arrangement of equipment for the one-fluid reactor is based on one-pass upward flow of fuel through the reactor vessel. This â&#x20AC;&#x153;once-throughâ&#x20AC;? flow allows the reactor and heat exchangers to be at the same elevation. The pumps are at the top of the reactor and have drive shafts that may be short enough to eliminate the need for molten-salt bearings. The heat exchangers are mounted so that they move and the thermal stresses are accommodated without the use of expansion loops or expansion joints in the fuel piping. We have used bellows in the coolant salt piping. We hope to be able to do this, for the coolant pipes are 34 in. in diameter, and expansion loops for such sizes are

very bulky. In the steam cell all components are anchored solidly. Expansion in pipelines is taken up by bellows in the pipes. The high-pressuresteam and feedwater lines have large expansion loops outside the cells to allow for dimensional changes in those lines.

5.4 REACTOR VESSEL AND CORE H. F. Bauman C. W. Collins W. C. George

H. M. Poly W. Terry H. L. Watts

W e have considered several concepts for the single-fluid reactor vessel and core. Studies have been made to try to attain the best combination of fuel inventory, flow control, temperature control, and e a s e of removal of the graphite from the reactor. A major complicating factor in the design is the radiation-induced dimensional changes in the graphite. We spent considerable effort in an attempt to employ reentrant concentric tubes of graphite for the fuel passages in the core. By confining the salt t o such tubes and filling the spaces between tubes with gas, the changes in salt volume caused by the dimensional changes in graphite are insignificant. This design, however, made necessary some kind of metal-to-graphite joint that had to be relatively leakproof, and these joints would have to be broken to replace the graphite. Also, the gas spaces in the core presented voids which could fill with salt and greatly increase the reactivity under certain conditions. Gravity flow from the reactor vessel t o the pump suction tank was believed to be desirable, but the inventory required to achieve this was excessive. The concentric reentrant tube concept was abandoned as having too many of the disadvantages of our designs for two-fluid reactors and none of the advantages. The idea of a container filled with vertical graphite pieces of some shape with fuel in the spaces between graphite pieces and flowing in a single pass upward through the vessel was finally adopted for t h e basic design. With this configuration, two possibilities for disposing the graphite within the vessel come to mind. One involves simply stacking graphite inside the vessel, each piece contacting the neighboring pieces and the entire bundle being finally restrained by the reactor vessel or by a cratelike structure within the vessel.

ORNL-OWG 68-4lSOA

INSTRUMENTATION CELL FREEZE

VALVE GROUND

FLOOR LEVEL

I HEAT STACK

Fig.

5.3.

Elevation

Reactor

and

Fuel

Drain

FUEL SALT DRAIN TANK

Cells.

ORNL-DWG 6S-4l69A

FUEL

SALT

PUMP

COOLANT

SALT PUMP

Y STEAM PIPING

Fig.

5.4.

Elevation

of Reactor

and

Steam

Cells.

ORNL-DW

66-4493

INS1

STEAM PIPING

Fig.

5.5.

Plan

View

Reactor

and

Steam

Cells.

.I.

This simple placement of the graphite might suffice if there were no dimensional changes with irradiation, but the graphite will shrink, the s a l t fraction in the core will increase, and this increase might be localized to create a rather large pool of s a l t at some indeterminate and possibly changing location within the core. Various schemes were attempted to try to cause some force (e.g., hydraulic pressure drop, buoyancy) to compact the core as irradiation caused the pieces to shrink. To date we have found no practical method of overcoming this â&#x20AC;&#x153;crevassingâ&#x20AC;? of the core; we therefore prefer another design. One way to be sure that the liquid volume will increase uniformly within the core as the graphite shrinks is to locate each piece of graphite at the top and bottom of the reactor by a m e t a l structure that is dimensionally stable. The method of confining the graphite at these locating points is important. The attachment must present a minimum of difficulty to removal of the graphite. It must accommodate dimensional changes of the graphite without undue restraint. In addition to positioning the graphite, means must be provided for supporting the considerable load of the graphite in the core. When the reactor is empty, this is a gravity load at the bottom of the vessel. When the reactor is filled with salt having approximately twice the density of the graphite, the graphite floats and exerts a buoyant force almost equal to the weight of the graphite in the empty vessel. When the pumps are running, there is an additional large upward force caused by the pressure drop across the core. W e have investigated many configurations of vessels with reinforced flat heads, dished heads, and inverted dished heads. We tried circumferential fluid entry, through transition sections from pipes to the reactor vessel. A design which seems to offer the best embodiment of the desired features is shown in elevation in Fig. 5.6. Some important parameters for the reactor are listed in Table 5.2. The vessel is about 18 ft in diameter by 24 ft high. It h a s a standard dished head at the bottom. In the center is a 4-ft-diam manifold into which the 24-in. outlet line from each heat exchanger discharges, and the four streams mix in the plenum formed by the dished head of the reactor vessel, Mounted above the dished head is a flat support plate with perforated web reinforcing. This

Table 5.2.

Design Parameters for One-Fluid Reactor

Thermal power, Mw Vessel diameter, ft Vessel height, ft Core diameter, ft Core height, ft Core volume, ft3 Average power density, kw/liter Fraction of liquid in core, % Region 1 Region 2 Region 3 Reflector thickness, in. Number of core elements Maximum velocity of salt in core, fps 3 Salt volumes, ft Core Reflector Plena Heat exchanger Pumps Piping Total F i s s i l e inventory, kg Fertile inventory, kg Specific power, Mw (thermal)/kg

4444 18.3 24.5 16 20 4020 -40 19 17 44 12 1760 13 1240 25 5 90 320 120 150 2445 1880 90,000 2.36

plate locates and supports the weight of the graphite stringers comprising the center part of the reactor core. The support plate is drilled on an even square pitch, and nipples for receiving the round ends of the moderator stringers are welded into these holes. These nipples serve as orifices for controlling flow and as sockets for locating the graphite pieces. The outer region of the core is mounted in thimbles the same as the center region. Flow i n this outer region is greatly reduced over that required in the center. This flow is obtained from the pool below the center region of graphite stringers and may be regulated by orificing the flow channels through and around the graphite pieces. Near the top of the reactor vessel a square grid is welded to the vessel. The squares in this grid are large enough to contain nine of the core pieces. This grid locates the top ends of the pieces so that each group of nine is exactly positioned within the core. The grid web is $4 in.

ORNL-DWG 68-41916

7 18 ft Oin. I D

t6ft Oin. DIAM CORE (38%-)U L E F

I-4Ift

COVER TRUSS

4in. DIAM (13% FUEL-)

-6ft 6in.DlAM(t8%FUEL)

CONTROLROD

CONTAINMENT SEAL WELDS

--I

____I

REFLECTOR

20 C

8 in.

24' Din.

Fig.

5.6.

Elevation of Reactor Vessel.

Ld"

i 61

. bd

thick. In order for the graphite to fit closely, the top 16 in. of each piece is reduced in cross section to permit its insertion past the grid. The center piece of graphite is inserted last and serves as a key to tighten the group. To remove the graphite, the center piece must be removed first. With the sockets at the bottom and the grid structure at the top of the vessel, the graphite is supported freely and is retained in one radial position. It is free to float in the salt and must be restrained by some loading bearing structure at the top. W e have made the top of the vessel a flat reinforced lid. This lid is held down by 1b e a m s clamped a t the perimeter. A seal weld is made for leak-tightness only. In each of the grid openings above the graphite stringers, we have a m e t a l spider with a center rod which transmits the buoyant and pressure forces to the reinforced lid of the vessel. This spider also has provision for orificing the flow of salt as i t leaves the core in order to help control the distribution of flow. For nuclear reasons (discussed in the physics section), the volume fraction of fuel in the core is nonuniform. As currently conceived, the core has three regions characterized by different salt fractions. The central one-sixth of the core, region 1, contains 19 vol % salt; the surrounding one-third, region 2, contains 17 vol % salt; and the outer one-half, region 3, contains 44 vol $4 salt. The power production per unit volume of salt in region 1 is nearly uniform, It decreases to about 27% of the center value at the outer edge of region 2 and t o 3%of the center value near the outer edge of region 3. W e would like for the temperature rise of the fuel that passes through the core to be nearly uniform and equal to the specified mean temperature rise. This is desirable in order to prevent overheating in the center of the core and to conserve on total flow through the reactor. Many configurations of core pieces have been examined in an attempt to achieve the desired distributions of fuel volume and flow, The one we presently like best is shown in Fig. 5.7. The graphite element is 4 in. square with the edges contoured and the center cored to obtain the specified salt fraction in each region. The design appears to eliminate places where salt could stagnate and overheat. Present plans are t o make the channels in region 1 of the core of uniform hydraulic radius and

to permit the salt to flow freely through all the channels. In the regions 2 and 3 the channels around the pieces, which are more difficult to orifice, will be reduced in thickness as the heat production decreases, and the channels through the pieces, which are more easily orificed, will be increased in size to obtain the desired fuel fraction. By distributing the s a l t in this manner, providing some orificing of the channels between pieces, and carefully orificing the channels in centers of the pieces, we hope to obtain a nearly uniform radial pressure in the interconnecting channels across the core. This should minimize the cross flow between core regions at the expense of a small increase in total flow and decrease in overall temperature rise of the fluid. W e will have no firm design for the orifices or good information about the flow distribution until a study has been made of the flow in the reactor core. This is in progress, but we have no results even of a preliminary nature.

5.5 PRIMARY HEAT EXCHANGER C. E. Bettis

H. L. Watts

W e have not designed a heat exchanger for the one-fluid reactor. What we have done is to scale up the heat transfer surface from previous calculations, base the length on the height of the reactor vessel, and calculate the diameter required t o provide the desired capacity. The heat exchangers now have one-pass flow of both fuel and coolant salt. The pumps are not integral with the heat exchanger, although they can be made so i f further analysis shows it to be desirable.

5.6 FUEL DRAIN TANK W. C. George H. M. Poly H. L. Watts

We have spent a large part of our design effort on a system for removing afterheat from fuel salt in the drain tanks. With the modular plant design of the two-fluid reactor, we proposed to cool the drain tanks by reheat steam from the high-pressure turbine. Since the likelihood of having more than two of the four modules drained at any one t i m e is small, the supply of this cooling steam should be dependable. With a single reactor, this is no longer true, and steam cooling, while still possible, looks less attractive.

ORNL-DWG 68-4196A

rz:z=l

SECTION A-A FUEL VOLUME -18%

21 '

21 f

SECTION A-A FUEL VOLUME 4370

SECTION C-C FUEL VOLUME -38%

t SECTION E-E

NOTE: DIMENSIONS ARE IN INCHES

Fig.

5.7.

Configurations of Core Elements.

Another basic change to the drain system results from our arrangement of equipment for the one-fluid reactor. A stopped pump no longer drains the reactor. W e presently see no reason for draining the reactor quickly and would propose to continue to operate at reduced power for some time with one or possibly more pumps off. W e have one drain tank capable of holding and cooling the entire salt volume (2247 ft3) of the reactor primary system. This is a cylindrical tank connected to the primary system through a 12-in. line. This line h a s a salt valve, a t present assumed to be a freeze valve, through which the system drains into the tank. An additional 6-in. line enters the tank near the top. This line is connected through a freeze valve to a drain pan in the reactor pit. In case of failure of a fuel system component, the fuel can be drained into the drain tank through this line. Salt is transferred to the reactor system by pressurizing the drain tank with gas. Figure 5.8 is an elevation drawing of this tank. Figure 5.9 is a plan of the tank showing the manifolding of the cooling coils which remove the afterheat from the fuel. The cooling is done by U-tubes which extend into the tank from headers located at the top but below the dished head cover. These are "-in. tubes, 0.42-in. wall, with a separation of 3 in. between the tubes in each U. The U-tubes are attached to headers which are arranged on two levels, one level running a t right angles t o the other level. This arrangement results in coolant tubes on 2Y8-in. square pitch throughout the volume of the drain tank. The U-tubes are assembled in modules, and these modules are connected together by bringing 1-in. lines through the dished head of the drain tank. There are many more coolant tubes in the drain tank than are required to remove the heat. We did this at least temporarily in order to reduce the ligament dimension of the fuel t o prevent hot spots within the fuel volume. The coolant surface in the drain tank can remove 300 Mw, while the maximum load amounts to only 60 Mw. In Table 5.3 we show some design parameters of the drain tank. The top layer of coolant headers are part of one autonomous coolant circuit, and the bottom layer of headers are part of another circuit. Each

Table

5.3.

Design Parameters for Drain Tank

Diameter of tank, ft Height of tank, ft Volume for fuel, ft3 Volume of coolant in tank, ft3 Number of coolant circuits Number of U-tubes per circuit Length of U-tube, ft Velocity of coolant in tubes, fps Overall U in tank, Btu hr" ft-2 Tank wall thickness, in.

16 20.5 2500 278 2 1540 30 0.75 OF-'

Design pressure, psi Number of modules of tubes per circuit Tube diameter, in. Tube wall thickness, in.

lt 70 100 3

0.042

s e t of headers is piped to a finned-tube air cooler. These coolers are located inside a chimney or chimneys outside the building. We have not yet decided whether t o provide a separate chimney for each coolant circuit. The chimneys are equipped with dampers so that the air coolers are maintained in a stagnant hot ambient when not required t o reject heat. In fact, heaters are provided to keep the air coolers above the melting point of the coolant s a l t at all times. This system of air coolers, U-tubes in the drain tank, and interconnecting piping is filled with the fluoroborate coolant salt. The s a l t circulates by natural convection, conveying the heat from the drain tank to the air cooler and dumping it into the atmosphere. We have only preliminary designs of the air coolers and chimneys. A manufacturer of such chimneys says that a stack about 150 ft high and about 30 ft in diameter will produce a pressure drop of about 1in. of water, which will be sufficient to remove 60 Mw of heat from air coolers in the base of the chimney. By using this type of heat removal system, no mechanical devices have t o run to remove the afterheat, and no reserves of water or coolants other than air, flowing also by natural convection, are required. As the afterheat load decreases with radioactive decay, dampers in the chimney can be closed to reduce the heat rejection rate.

ORNL-DWG 68-4488 A W

L 6-in. PIPE

.COOLANTHEADERS

. COOLANT HEADERS

Lts

12-in. PIPE

Fig. 5.8.

Elevation of Fuel Drain Tank.

, w c

,FUEL

SALT LEVEL

. Fig. 5.9.

Plan View of Fuel Drain Tank.

6. Reactor Physics A. M. Perry ORNL-DWG 68-5526

6.1 MSBR PHYSICS ANALYSIS 0. '

W. R. Cobb

L. Smith

H. T. Ken

Physics analysis of the MSBR for the period covered by this 'report was confined to studies of a single-fluid machine. Results of the studies demonstrated that the nuclear performance of a single-fluid reactor could be made a s attractive as that of the 40-w/cm3 two-fluid reactor discussed in previous progress reports. Some of the results presented here should be regarded a s preliminary. In the discussion below, a reactor which includes all the m o s t recent design information is identified a s the reference reactor. This designation is made for convenience of identification only. Several revisions to this design are already being studied. For example, it may be desirable to tailor the salt distribution in the axial a s well a s the radial direction. Also, the reactor fuel cell described below is intended only to show the approximate features of an MSBR single-fluid cell. Instead of having all of the salt surrounding the graphite a s depicted below, final cell designs will almost certainly have part of the salt interior to the graphite, partly to maintain suitably low graphite temperatures and partly to achieve the most advantageous self-shielding of thorium resonance cross sections.

GRID AND PLENUM

I 1

CORE 49% SALT

CORE 47% SALT

CORE

44% SALT

i f t GRAPHITE

-2 in. HASTELLOY

I I

-\SALT

AND HASTELLOY

6.1.1

-t

Reference Reoctor

8ft-4

kfft

A schematic drawing of the most recent one-fluid reactor core is shown in Fig. 6.1. The core is 16 f t in diameter by 20 ft in length. Radially the core is surrounded by 1 f t of graphite reflector. Above and below the core are several structural regions which support the core and transport salt. Since these

Fig. 6.1.

Reference Reoctor Core.

regions contain various amounts of s a l t and graphite in addition to Hastelloy N, they are also ef-

fective as blanket regions for the core. The reactor is surrounded by a pressure vessel of Hastelloy “2 in. thick. For reasons which will be described below, the core contains 19% salt in the central one-sixth of the volume, 17% in the next one-third, and 44% in the outer one-half. The salt contains 67.68 mole % LiF, 20 mole % BeF,, 12 mole % ThF,, and 0.32 mole % UF,. In addition to the salt inventory explicitly indicated in Fig. 6.1 (1745 ft3), the external system (heat exchangers, piping, etc.) contains 700 ft3. The total reactor power output is 4444 Mw (thermal) or 2000 Mw (electrical). The average power density is 40 w/cm3 of core volume. Although the details of fuel processing are incomplete, it is currently believed that the salt processing cycle time for fission product removal will be -40 days, with a processing cycle t i m e for protactinium removal of - 5 days. Figure 6.2 shows a plot of the total thermal flux below 1.86 e v and the fast flux above 50 kev vs radial position in the core midplane. Figure 6.3 shows the same information along the axis of the core. The peak-to-average thermal flux ratio is 2.7 in the radial direction and 1.5 in the axial direction, giving a total peak-to-average thermal flux ratio of 4.0. The peak-to-average fast flux (above 50 kev) ratio is 2.6 radially and 1.5 axially, giving a total peak-to-average fast flux ratio of 3.9.

The appearance of a radially nonuniform volume fraction of salt in the core deserves special discussion because this feature produces a substantial improvement in breeding performance over that of a core with a uniform volume fraction of salt. A uniformly loaded single-fluid reactor can be designed with a high breeding ratio (e. g., 1.07) simply by making the reactor so large that leakage is negligible. However, the large inventory required to fuel such a reactor greatly penalizes the yield. On the other hand, if one introduces a nonuniform loading with relatively small salt volume fractions in the center of the reactor (“17 t o 19%) and larger volume fractions (“44%) in the outer region, small leakages may be achieved with modest size cores. With this salt distribution the neutron energy spectrum is harder in the outer region than in the center, the neutron source density near the surface is reduced, and hence leakage is reduced. A t the same time the harder spectrum and the relatively large inventory of salt in the outer region enhance neutron captures in the thorium resonances. The nonuniform core may be characterized by saying that the central part of the core acts like a core and that the outer part, containing the same materials but in different proportions, acts like a blanket. Consequently a single fluid is effectively made to serve a dual purpose. (It should be pointed out that all refer-

ORNL-OWG 6 8 - 5 5 2 8

ORNL-OWG 68-5527

(~10’~)

Fig.

6.2.

120 160 RADIUS ( c m )

200

240

Fast and Thermal Flux Distributions in

Core Midplane.

280

100

200

300 400 500

600

700

000

DISTANCE FROM REACTOR TOP (cm)

Fig. 6.3.

Fast and Thermal Flux Distributions Along

Reactor Axis.

68 ences to the core include both regions. Specifically, reported power densities are averages over the two zones.) Comparison of the highest-yield one-fluid 2000 Mw (electrical) uniform core with the highestyield nonuniform core shows an increase in yield from 4.0 to 5.5%, a s well as an increase in breeding ratio from 1.048 to 1.068. The optimum salt volume fraction distribution so far as yield is concerned has a value of 14% in the central one-sixth of the core volume. However, for that core the thermal flux, and hence the specific power, is strongly peaked in the center of the core, being -80% higher than shown in Fig. 6.2. Heat removal from such a strongly nonuniform heat source distribution is difficult. By raising the salt fraction in the central one-sixth of the core to lWo, the flux in the center of the reactor can be flattened and the peak flux reduced a s shown in Fig. 6.2. The yield is reduced a t most 0.3%. The considerable simplification of heat removal problems afforded by the flattened source distribution seems to justify the s m a l l penalty on yield. Also, since the fast damage flux distribution closely follows the thermal flux distribution, reducing the peak thermal flux as much as possible is desirable from the standpoint of maximizing the useful lifetime of the graphite in the center of the core. Present calculations indicate that the graphite in the maximum flux region of the reference reactor will have a replacement lifetime of from 2 to 3 years if the graphite lifetime is limited to a fluence of 3.0 x nvt (E > 50 kev). It is instructive to compare the reference reactor with a variety of other designs. An OPTIMERC calculation determined the maximum yield attainable in each case. The comparison, in terms of relative changes in yield and breeding ratio, is shown in Table 6.1. All variations are expressed relative to the reference reactor. Comparison of cases 1 through 5 shows that the reference reactor has a higher yield than any uniform core and has the optimum diameter of the nonuniform cases. The optimum axial length a s well a s the desirability of a tailored s a l t distribution in the axial direction are under investigation. Comparison of cases 1, 6, and 7 indicates that 12 mole % ThF, is nearly the optimum percentage of this salt component. Larger mole percentages give marginally improved nuclear performance while having markedly poorer thermal properties (particularly a higher melting point).

Comparison of cases 1 and 8 shows some incentive for choosing a 2000 Mw (electrical) design over a 1000 Mw (electrical) design. However, despite the improved yield of the 2000 Mw (electrical) reactor, the performance of the 1000 Mw (electrical) reactor is good. The 1000 Mw (electrical) reactor may eventually be selected as the reference design, depending upon the outcome of studies currently under way to formulate design criteria. Case 9 shows that the reflector is of marginal value so far as neutron economy is concerned. The reflector is of small value because the outer part of the core functions as a blanket. If the reflector should prove to be a complicating or expensive part of the reactor it could be removed without great penalty to the yield. However, it may be necessary or desirable in any case to use the reflector to shield the reactor vessel against neutron irradiation. Anticipating the possibility that design problems may somehow necessitate a certain volume of salt outside the core and hence in a largely parasitic region, case 10 is intended t o show that the introduction of a substantial inventory of salt outside the core does not in fact greatly impair the performance of the system. T o a large extent this salt pays its way by serving a s blanket material and allowing a reduction in the amount of salt in the outer part of the core itself. Finally, case 11 is presented to show the kind of nuclear performance that can be achieved in a reactor much smaller in s i z e and power output than the reference reactor. In contrast with the 40-w/cm3 power density of the reference reactor, the power density in case 11 is 25 w/cm3. In c a s e 11 the volume fraction of salt in the central half of the core is *14%, and in the outer half it is -770%. The strongly nonuniform volume fraction of s a l t in the core of the small system minimizes neutron leakage and affords a good breeding ratio and a modest yield. The unflattened peak thermal flux a t the center of the small reactor is very nearly the same as the flattened thermal flux in the central one-sixth of the core of the reference reactor. Thus, it a p pears possible in a reactor the size of case 11 t o approximate rather closely the peak fluxes, flux distributions, and breeding ratio of the reference reactor, with somewhat greater departures from the core average power density and the yield of the larger machine. Having presented the principal features of the one-fluid reactor, it is worth while to compare it

. bi

69 Table 6.1.

Comparison of Other Configurations with the Reference Reactora ~ _ _ _ _ _

Reactor Description

Change in

Change in Yield

Breeding Ratio

(%I

1. Reference reactor, 20 x 16 f t nonuniform core,

2000 Mw (electrical) 2. Similar to reference reactor but with 20-ft core diameter 3. 25 X 25 ft core, uniform salt volume fraction of 0.127, 2000 Mw (electrical) 4. 2 0 x 2 0 f t core, uniform salt volume fraction of 0.14, 2000 Mw (electrical) 5. 15 x 15 f t core, uniform s a l t volume fraction of 0.18, 2000 Mw (electrical) 6. Similar to reference reactor but with 1 0 mole %

- 0.001

- 0.55

- 0.007

-1.7

0.02

-1.5

- 0.047

-2.4

- 0.006

- 0.4

- 0.01 7

-1.4

0.004

- 0.8

- 0.003

- 0.3

- 0.001

- 0.6

- 0.02

-4.0

ThF, 7. Similar to reference reactor but with 8 mole % ThF4

8. Similar to reference reactor but with 14 X 14 ft core and power output of 1000 Mw (electrical)

9. Similar to reference reactor but without radial reflector 10. Similar to case 9 but with 4 in. of pure salt around core in radial direction 11. Similar to reference reactor but with 1 0 X 1 0 f t core and power output of 250 Mw (electrical)

'All the reactors were optimized to obtain the maximum yield. bHeight and diameter.

with the two-fluid reactors considered in O R N L 3996 and i n previous reports. Table 6.2 gives pertinent information for both systems. Data are included for two versions of two-fluid breeders for 1000 Mw (electrical) power plants. One plant has a single reactor vessel with average and maximum power densities in the core of 80 and 160 kw/liter respectively. The other plant contains four 250 Mw (electrical) reactor modules, each with average and maximum power densities of 40 and 80 kw/liter respectively. This modular plant, of considerably degraded breeding performance but longer life of graphite under irradiation, was finally selected to be the reference design for the two-fluid breeder plant, primarily on the basis of cost and practicality considerations relative to replacing the graphite.

Since the two-fluid reactors are for 1000 Mw (electrical) systems, we should compare them with a 1000 Mw (electrical) one-fluid reactor. The 2000 Mw (electrical) one-fluid reference reactor is included in Table 6.2 for completeness. The numbers are not all consistent, but the reactors should generally be compared on the basis of the maximum power density (life of the graphite) in the core. It should be possible to replace the graphite in a one-fluid reactor more frequently than in a twofluid reactor for the same cost. On the basis of economic considerations, then, the one-fluid reactors shown in the table are more nearly comparable with the modular version of the two-fluid reactor. From the table it is clear that, although the two designs are rather different, the overall nuclear per-

70 Table

6.2.

Comparison of the Characteristics of Two-Fluid and Single-Fluid

Two-Fluid MSBR

250 Mw (electrical) a

MSBR's

Single -Fluid MSBR

1000 Mw (electrica 1)

2000 Mw (electrica 1)

Core height, ft

12.5

13.7

Core diameter, ft

10.0

9.7

11.32

Blanket thickness, f t

1.5

2.0

2.34

Core power. Mw (thermal)

555

2222

1812

3646

41 0

798

Blanket power, Mw (thermal) Average core power density, kw/liter

Peak-to+verage power ratio in coreb

2.0

Graphite replacement life, years'

3.4

1.7

2.1

Specific fuel inventory, kg/Mw (e lectrica 1)

1.04

0.73

1.06

0.94

1,000 0.033 0.075 0.031 0.005

1.000 0.032 0.067 0.032 0.001

1 .ooo 0.041

0.021 0.01 4

1.000.045 0.053 0.023 0.011

Breeding ratio

1.069

1.071

1.068

Annual fuel yield, %/year

5.0

7.4

4.8

5.5

Fuel doubling time, years

9.4

14.4

12.6

Neutron balance, neutron captures per neutron absorbed in fissile materia 1

Fissile materia1 ( 233U 235U) Modera tor Carrier salt Fission products Leakage

0.052

aOnequarter module of a 1000 MW (electrical) plant. bAssumed average ratio maintainable over life of graphite. CAllowable dose = 3.0 x lo2* nvt > 50 kev, plant factor = 0.8.

formance of the two systems is quite similar, The principal differences appear in the details of the parasitic losses in the neutron balances of the two systems. However, as shown in the table, the relatively modest differences essentially cancel.

have not yet been fully determined, and hence it is premature to discuss the fuel-cycle costs in any detail. However, it is currently felt'that fuel-cycle costs less than 0.5 mill/kwhr will be achievable with the liquid-metal extraction techniques which are being investigated for the one-fluid system. *

6.1.2 Fuel-Cycle Costs 6.1.3 Cell Calculations In all the calculations described above, emphasis was placed upon maximizing the yield of the various systems. Details of the salt processing scheme

A number of calculations have been performed to determine the nuclear properties of a single-fluid

MSBR fuel cell. The calculations are based on the ORNL-DWG 68-5530 cell shown schematically in cross section in Fig. 6.4. This cell contains a central graphite log -4.7 in. in diameter surrounded by a region of salt occupying 20% of the total cell volume. Conclusions FLUX ( â&#x201A;Ź < 1.86 ev) based upon calculations using this cell must be regarded as tentative for several reasons. First, the core of the reference reactor actually contains three f E L LI I I different salt volume fractions and hence three difAVERAGE FAST ferent cell designs. Second, an MSBR cell will FLUX ( E > 50 kav) -probably contain a portion of the salt in passages through the graphite, perhaps in a central cylindrical passage. Third, a cell -4 in. in diameter is probaSALT bly required i n order to maintain sufficiently low I GRAPHITE I I I graphite temperatures. Despite these shortcomings, 0 1 2 3 4 5 6 7 it is felt that the characteristics of the cell RADIUS (cm) presented here will be a useful and reasonably accurate first approximation to those of actual cells. Fig. 6.5. Fast and Thermol Flux Distributions in OneFigure 6.5 shows flux plots of the total fast flux Fluid MSBR Cell. above 50 kev and total thermal flux below 1.86 e v as a function of radial position in the cell. The curves are shown relative to the cell average fast flux and cell average thermal flux respectively. 6.1.4 Reactivity Coefficients The detailed distributions shown for the fluxes in Fig. 6.5 are in fact superimposed upon the gross A number of isothermal reactivity coefficients distributions shown in Fig. 6.2. Table 6.3 shows were calculated using the reference reactor and the the ratio of the average flux in the salt or graphite cell of Fig. 6.4. These coefficients are summarized region to the cell average flux for each of several in Table 6.4. The Doppler coefficient is primarily energy ranges above thermal. that of thorium. The graphite spectral coefficient is positive because of the competition between thermal captures in fuel, which decrease less rapidly than l/v, and thermal captures in thorium, which decrease ORNL-DWG 68-5529 nearly as l / v with increasing temperature. The total spectral component includes the Doppler term, the graphite spectral term, and the thermal spectral effects of all important moderators in the salt. The resonance density component represents the effect of decreased self-shielding of thorium with decreasing salt density. The total density component includes all effects associated with changes in the density of both s a l t and graphite, including the resonance density term. The prompt salt component includes all reactivity effects associated with an isothermal core temperature change except the graphite spectral and graphite density components, the latter of which is negligible. The last entry in Table 6.4 combines the spectral and density effects into a total isothermal reactivity coefficient. Several important conclusions can be drawn from these coefficients. First, the total power coefficient is negative. Second, both the total density Fig. 6.4. One-Fluid MSBR Fuel Cell.

I 1

72 Flux Disadvantage Factors in Fast and Epithermal Ranges

Table 6.3.

Group

Table 6.4.

Energy Range

0.821-10 MeV

0.0318-0.821

1.234-31.82

0.0479-1.234

1.8647.9 ev

Mev kev kev

Isothermal Reactivity Coefficients

Reactivity Coefficient, Component

x 10-~ -3.6

Graphite spectral

+ 1.9

Total spectral

- 0.26

Resonance density

-1.2

Total density

-0.26

Prompt salt

-2.4

Tota 1

- 0.49

and the total spectral reactivity coefficients are negative. Third, and perhaps most important, the prompt coefficient of the salt, which largely determines the fast transient response of the system, is a relatively large negative coefficient and should afford adequate reactor stability and controllability. Studies of the actual dynamic performance of the system are in progress.

6.1.5

Meorurcments of Eta for 233Uand

’45 cell

45 ..lt/4

0.967

1.132

0.985

1.060

0.999

1.004

1.006

0.97 7

1.008

0.969

cell

fuel isotope. Samples have been taken for this purpose during operation with the 235U loading, and others will be taken during operation with 233U.A precise measurement for the latter would be of interest to the MSBR program since, a s discussed by Perry,’ the uncertainty in the effective value of 7 for 233Ucontributes the major uncertainty in the breeding ratio of molten-salt breeders. The measured 7 ( 233U) will be quite comparable with that for proposed MSBR designs because of the similarity, as demonstrated by Prince,2 of the spectra of MSBR and MSRE. For both 5U and 3U loadings of MSRE, comparisons between measured and calculated values of 7 provide tests of calculational methods and cross-section data. The first experiment, with 235U, will check procedures and equipment; the result will test the validity of the method, and may, indeed, constitute a useful check on the cross sections of 2 3 5 ~ . The basic idea of the measurement, discussed in terms of Z33u for concreteness, is as follows: Absorptions in 233U are measured by observing the decrease in 233U concentration in the salt over a period of time at power; captures i n 233U are measured by observing the increase i n 234U concentration (corrected for absorptions in 234U itself). These concentrations, and the changes i n them, are measured relative to that of 238U, whose neutron absorption cross section, and hence concentration change, are far smaller than those of either 233U

of the Reference Reactor

Doppler

+graphite

235U

in the MSRE

G. L. Ragan Isotopic analysis of two MSRE fuel samples differing in depletion of the 5U or 3U fuel can be used to measure the value .of 7 for the principal

‘A. M. Perry, “Influence of Neutron Data in the Design of Other Types of Power Reactors,” t o be published in The Proceedings of the Second Conference on Neutron Cross Sections and Technology, Washington, D. C., March 4-7, 1968. ‘B. E. Prince, MSR Program Semiann. Progr. Rept. Aug. 31,1967, ORNL-4191, p. 50.

. cd

73 or 234U. These relative concentrations can be measured very precisely, yielding a precise value for the conveniently defined cross-section ratio i

Using a well-established3 value for v , one gets 9 = v-=Of

vu- y)

An expression has been developed for the crosssection ratio y in terms of the quantities actually measured by the m a s s spectrometer and the effective absorption cross sections of 4U and 238U. For the m a s s spectrometer measurements, one includes: (1) the measured current ratios (e. g., that in the 233U channel to that i n the 238U channel) ; (2) the instrument calibration factors (e. g., the relative current sensitivities of channels 233 and 238) ; (3) interchannel cross-coupling factors (e. g., the spurious current in channel 234 due to the breadth of the 233 peak); and (4) variation in these instrument factors between measurements on initial and final samples. The ratios of the effective cross sections of 234U and 238U to that of 233U are involved because of the s m a l l but signifi-

cant corrections for depletion of 234Uand 238U during the course of the experiment. An extensive error analysis4 h a s been performed for both the 233U and 235U cases. The uncertainties which appear to limit the attainable precision are those of: (1) ratio of current in another channel to that in channel 238 (assumed uncertainty +O.Ol%); (2) effective absorption cross sections of 234U, 236U, and 238U relative to those of the principal fuel isotopes 3U and 5U (assumed uncertainty +lo%). On this basis, it appears that an experiment involving 2% depletion of 233U should measure y to f3.3%, corresponding to an uncertainty of +0.008 in the effective value of q(233U) in the MSRE spectrum. The corresponding errors for 2% depletion of 235U are +1.4% in y and k0.007 in 9(235U). To these uncertainties owing to the measurements themselves, one should add an allowance for the uncertainty in v Using presently accepted values3 for these [ f0.32%'for v(233U) and +0.25% for v ( ~ ~ ' u1,) one estimates an overall uncertainty of +0.011 (fortuitously coinciding) for both 9(233U) and q(235U), a s measured in the MSRE spectrum. This value may be compared with our present estimate' of +0.015 for the uncertainty in q(233U), in a typical MSBR spectrum, attributable to uncertainties in neutron cross sections.

3See, for example, J. R. Stehn et al., Neutron Cross Sections, BNL-325, second ed., Suppt 2 (February 1965).

4A. M. Perry and G. L. Ragan, Measurement of 9(233U) in the MSRJZ (in preparation).

7. Systems and Components Development Dunlap Scott

i I 1

kw/liter. The fuel cell geometry was also changed somewhat to give a higher specific surface area. The various xenon migration parameters were chosen to obtain an 135Xe poison fraction of 2.25% (high end of the attainable range) with no coating, and then the calculations were extended to include the effects of coatings. The results of these computations appear in Fig. 7.1. In the coating the void fraction available to xenon was made variable in such a way that it changed

The programs concerned with migration of noble gases, the technology of sodium fluoroborate, and the early development of pumps for molten-salt reactors were continued during the period. A study was begun of remote maintenance for MSBR’S, including the problems involved in the remote welding of pipes and vessel closures. Studies of the effectiveness of decontamination of small components for subsequent maintenance and of the use of a gamma-ray camera for locating radioactive hot spots, both of which are pertinent to the maintenance plan for an MSBR, are reported in the MSRE section of this report.

ORNL-DWG 68-5531 c3

7.1 NOBLE GAS BEHAVIOR IN AN MSBR

I I I

I ~

I I

I-

REACTOR POWER -556 Mw(THERMAL) CORE POWER DENSITY -20 kw/liter BULK GRAPHITE DIFFUSION COEFFICIENT-

5z 2

ft2/hr

R. J. Kedl 8

SIGNIFICANT PARAMETERS:

83 0E

BULK GRAPHITE AVAILABLE VOID-IO% CIRCULATING BUBBLE SURFACE A R E A - 3 0 0 0 ft2 COMPOSED OF “ONCE THRU“ BUBBLES AND NO “RECIRCULATING” BUBBLES

The results of ‘35Xe poisoning calculations were presented for a specified two-fluid reactor concept in the last semiannual progress report (ORNL-4191). They showed that the target poison fraction of 0.5% was attainable. Since then, the proposed core has increased in size in order to lower the power density. In addition, the proposed fuel cell geometry h a s changed to significantly increase the graphite surface area per unit volume of core. These factors will increase the ‘35Xe poisoning, and it now appears that h e attainable poison fraction will be between 1 and 2%. As a result, consideration is being given to graphite with a very low-permeability coating. Calculations were made to determine the effect of coating the graphite with a thin layer of lowpermeability graphite. As a reference design, essentially the same reactor concept was used a s described in ORNL4191, but with the core size increased to yield a mean power density of 20

MASS TRANSFER COEFFICIENT TO BUBBLES2.0 fthr FUEL CELL GEOMETRY -CONCENTRIC ANNULUS

IO-^

0 5 2

Q .

10-6

IO 3

10-7

10-8

0.3 0.I

IO-^

0 5

COATING THICKNESS (mils)

Fig.

7.1.

Fraction.

Effect of Coated Graphite on 13’Xe Poison

75 by one order of magnitude when the diffusion coefficient changed by two orders of magnitude. The diffusion coefficient of lo', ft2/hr for the bulk graphite represents graphite believed to be readily available. It is important to notice that the bulk graphite must be improved to a diffusion coefficient of less than lo-' ft2/hr before there is a significant effect on the poison fraction, whereas an 8-mil coating of 10-8-ft2/hr material would reduce the poison fraction to the target value of 0.5%.

7.2 SODIUM FLUOROBORATE CIRCULATING LOOP TEST A.

N. Smith

P. G. Smith Harry Young

The alterations necessary t o convert this test facility, formerly called the Fuel Pump HighTemperature Endurance Test Facility, to operate with sodium fluoroborate were completed. The work included installation of (1) equipment for automatic control of BF, flow through a dip leg in the pump bowl and helium flow to the pump shaft annulus, (2) an access nozzle for removing salt samples from the-pump tank, (3) a thermal conductivity cell to monitor the composition of the off-gas stream, and (4) equipment for controlled ventilation of the loop enclosure and instrument cubicles with provisions for venting the discharge above the building roof. Instruments and components were checked out for proper operation, and the system was leak tested. A flush charge of about 900 lb of sodium fluoroborate was added to the sump, and this s a l t is now being circulated. The flush salt, which is intended to remove residues of the lithium fluoride-beryllium fluoride salt previously circulated in the loop, will be drained after several days of circulation, and a second batch will be added. The salt will then be circulated for several months for investigation of

Salt samples will be used for direct indication of salt composition. Devices that will be evaluated for monitoring the composition indirectly are a thermal conductivity cell, which will provide an indication of the BF, partial pressure in the offgas stream, and a cold-finger unit, which will check for changes in the liquidus temperature of the salt. The cold-finger system is currently under design and will not b e available during the initial period of circulation. Tentative calibration data on the thermal conductivity cell indicate that a change in BF, partial pressure of 3.5 mm Hg will result in a change in signal of 2.5% of the indicator scale. This sensitivity is enough to detect any significant changes i n the salt composition.

7.3 MSBR PUMPS A. G. Grindell

L. V. Wilson C. E. Bettis C. K. McGlothlan

P. G. Smith H. C. Young

7.3.1 Pump Program The present emphasis on studying the singlefluid MSBR h a s influenced the pump program. The number of hydraulic designs h a s been reduced from 3 required for the twefluid concept t o 2 (see Table 7.1). Preliminary study indicates the practicability of using a sump pump configuration similar to that of the salt pumps for the

Table 7.1.

Pump Requirements for

2000 Mw (Electrical) MSBR Fue 1

Coolant

Number required

Design temperature, OF

1300

Capacity, gpm

24,000

53,000

1. pumping characteristics of the salt, including minimum overpressure necessary to prevent cavitation, 2. problems associated with control and monitoring of the salt composition.

Head, ft

200

Speed, rpm

890

Specific speed, N e

3170

3830

Net positive suction head required, ft

'MSR Program Semiann. Progt. Rept. Aug. 31, 1967, ORNL4191. p. 95.

Impeller input power, hp

7500

7 000

76 Molten-Salt Reactor Experiment (MSRE) and of using pumps of essentially one mechanical design and one drive design. The emphasis on the single-fluid concept has not, however, altered the pump program objectives' and the desire for the strong participation of the United States pump industry in design, development, production, and operation of pumps for the Molten-Salt Breeder Experiment (MSBE). Our present plan for securing industrial participation in the pump program, in brief, consists of the following items: 1. Interested pump manufacturers will be requested to submit proposals for designing the MSBE fuel salt pump to our specifications. 2. After evaluation of these proposals, selected pump manufacturer(s) will be asked to design the fuel s a l t pump and to estimate the cost and schedule to produce a prototype. The cost estimate will include whatever development program the pump manufacturer proposes and his participation in the ORNL pump program. The pump manufacturer will also be requested to provide cost and schedule estimates for producing the required salt pumps for the Engineering Test Unit (ETU) and the MSBE. 3. After evaluation of the completed desigfls, selected pump manufacturer@) will be requested to produce the prototype fuel salt pump, complete his development program, and participate in the ORNL pump program. 4. The prototype fuel salt pump will be subjected to molten-salt testing'in the Fuel Salt Pump Test Facility with the participation of the pump manufacturer(s). 5. The salt pumps for the ETU and the MSBE will be produced by the pump manufacturer(s). 6. With the participation of the pump manufacturer(s), proof tests of the ETU and MSBE s a l t pumps 'will be performed in the Fuel Salt Pump Test Facility. 7. Adequate support, in which the pump manufacturer(s) will participate, will be provided to sustain the continued satisfactory operation of the salt pumps in the ETU and the MSBE. We will make design studies and analyses of the salt pump configuration and its relationships to the reactor system to become thoroughly

'MSR Program Semiann. Progr. Rept. Aug. 31, 1967,

ORNL4191, p. 96.

familiar with the pump problems. The studies will

be used to provide insight and direction to the pump design and development for the industrial participant(s) a s well as the Laboratory. Information is now being assembled for use in writing specifications for the fuel salt pump.

7.3.2 Fuel Salt Pump In our present design of the single-fluid breeder, the pumps are near the top of the reactor cell. The length of the shaft can be reduced considerably from that of the long-shaft pump3 proposed for our design of two-fluid reactors; this shortening presents an opportunity to eliminate the molten-salt bearing required in the long-shaft pump. Figure 7.2 shows a preliminary layout of a concept of the fuel salt pump for the single-fluid reactor. Although the capacity is much larger, the pump is similar in configuration to the MSRE s a l t pumps. The drive motor is mounted in a fixed position on top of the concrete shielding, and the bearing housing is recessed into the shielding to reduce the shaft overhang. The pump shaft is mounted on two pairs of preloaded oillubricated ball bearings. The impeller is overhung about 6% ft below the lower bearing. A preliminary study indicated a first shaft critical speed greater than 1200 rpm. Internal shielding of the bearing housing is provided by a steel or Hastelloy N shield plug cooled by an organic liquid, possibly the same as the bearing lubricant. To reduce deflections of the shaft a t the impeller caused by hydraulic radial forces on the impeller, the pump casing is of either the double-volute (as shown) or the vaned diffuser design to minimize these forces. Both casings have the added advantage of greater structural strength than a single-volute or a vaneless diffuser. At the design temperature of 1300째F, where the allowable stress for Hastelloy N is 3500 psi, this advantage is of considerable importance to the designer. The bearing housing is designed for replacement by semidirect maintenance, either in situ or in a hot cell after the entire rotary element has been removed from the reactor system.

3MSR Program Semiann. Progr. Rept. Aug. 31, 1967,

ORNL-4191, p. 98.

77 ORNL-DWG 68-5532

Fig. 7.2.

MSBR F u e l Salt Pump Concept. Preliminary layout of short-shaft pump.

78 The drive motor will be installed in a fixed position, and the pump will be permitted to move freely with the thermal expansion of the reactor system. For power transmission a floating coupling or universal joints will be used to accommodate the approximately '/*-in. horizontal and vertical displacements of the pump that are anticipated during operation at temperature. The transmission device will also have design features to accommodate the approximately 5-in. vertical and 1f/,-in. horizontal movements that occur between room temperature and reactor cell design temperature; we do not presently anticipate operating the pump at room temperature. The pump tank provides volume to accommodate the anticipated thermal expansion of the fuel salt a t off-design conditions. It is almost completely decoupled hydraulically from the flowing salt in the impeller and volute passages by (1) labyrinth seals installed in the pump casing around the pump shaft and on the periphery of the casing and (2) bridge tubes that connect the volute to the inlet and outlet nozzles attached t o the pump tank. The bridge tubes also eliminate structural redundancies between the pump tank and the volute and its supporting structure. In a noteworthy fashion the hydraulic decoupling serves to minimize the changes that may occur in the pump tank liquid level when several pumps are being operated in parallel and one of them stops. Consider, if (1) the gas volumes of the salt pumps being operated in parallel are interconnected, (2) the salt volume in each pump tank is connected directly to its pump suction, and (3) all pumps are being supplied from a common plenum in the reactor, then the following events would occur when one pump stops: The level of salt in the tank of the stopped pump would try to increase by an amount equal to the velocity head at the pump suction plus the head l o s s in the suction line from the common supply to the pump tank. This change in level would be ten or more feet, which would represent an unwanted increase in pump shaft length. Also, unless there is sufficient reserve salt volume in the other pump tanks to supply the increased salt requirement of the stopped pump, the system fluid would in-gas when the salt level in the tanks of the operating pumps is lowered to the level of the pump suction. HOWever, by connecting the liquid in the hydraulically decoupled version of the pump tank to a point i n the reactor plenum where the velocity changes

very little when one pump is stopped, and by making the pressure drop in this connecting line very low for the salt flow returning from the tank to the plenum, the level change in the pump tanks probably can be held to about 2 ft. Although not shown in Fig. 7.2, the upper portion of the pump tank and its internal structure will be cooled by a s m a l l bypass flow of the fuel salt to remove beta and gamma heat. A study using an existing heat transfer program (SIFTTOSS) is in progress to determine the temperature distribution in the pump shaft and shaft casings. The first case being solved assumes full-power gamma heating on one half of the shaft casing and no gamma heating on the other half. The results of the first case will indicate the next case to be considered. A preliminary investigation h a s been made of the capacity of the fuel salt pump and its speed on various pump parameters such as impeller diameter, net positive suction head required (NPSH), hydraulic efficiency, and shaft diameter. The results are shown in Fig. 7.3.

7.3.3 Coolant Salt Pump There is sufficient similarity in the design head, speed, temperature, and power characteristics of the fuel and coolant salt pumps (see Table 7.1) that it appears feasible to interchange the rotary elements (except for the impellers) and the drive motors. Interchangeability could lead to considerable savings in the MSBE s a l t pump program.

7.3.4 Molten-Salt Pump Test Facility The design was initiated for a molten-salt pump test facility which will provide adequate nonnuclear testing capabilities for the MSBE s a l t pumps. Basically, the facility will consist of the necessary piping system, heat sink (preferably to air), variable flow restrictor, flow measuring device, salt storage tanks, preheaters, and controls and instrumentation. The following is a partial list of the criteria for the facility: Maximum temperature Maximum cover g a s pressure Maximum flow Maximum head loss Maximum heat sink capability

1400OF 150 p s i g 10,000 gpm 200 ft '"100,OO Btu/min

79 ORNL

.-.-._._600

Fig. 7.3.

. W

800 I000 PUMP SPEED (rpm)

E f f e c t of Copacity and Speed of

The facility including test pump will be approximately 60 ft high, i f the long-shaft pump is used, and approximately 50 ft long. It is proposed to locate the facility in Building 9201-3. Preliminary design studies have been directed to the sizing of a salt-to-air heat exchanger and investigating flow characteristics of the salt piping loop.

7.3.5 Molten-Salt

Bearing

Program

Although the single-fluid reactor concept may permit the use of a short-shaft pump similar to

- DWQ 68â&#x20AC;&#x2122;5533

SHAFT DIAMETER (in.) REACTOR PDWER-2000 Mw( ELECTRICAL) TOTAL SALT FLOW 435,000 gpm PUMPHEAD -200ft

4 200

MSBR Fuel Salt

Pump on Various Pump Parameters.

that of the MSRE salt pumps, we plan to pursue the molten-salt bearing program vigorously. There is need to establish whether such bearings can be used in a pump. In addition, the bearing materials may be applicable t o the plugs and s e a t s of valves for molten salt. Bearing Materials. The Metals and Ceramics Division is preparing a program in which candidate bearing materials will be subjected to fuel and coolant salt tests. Hard-surface coatings on Hastelloy N substrate will be studied. The various methods that will be tried for applying the coating to the substrate include plasma spray,

multiple coating by chemical vapor deposition, and brazing. Sixty-four hard-coated specimens4 consisting of plasma-sprayed cermet on Hastelloy N substrate were received from Mechanical Technology Incorporated. Four different coatings are represented: (1) tungsten carbide bonded with 7 to 10%cobalt, (2) 25% tungsten carbide-7% nickel68% mixed W-Cr carbides, (3) 40% pure tungsten carbide-50% tungsten carbide with 12% cobalt binder-10% molybdenum, and (4) 75% chromium carbide-25% Ni-Cr alloy binder. These coatings will be evaluated by the Metals and Ceramics Division on the results of (1) chemical compatibility tests with molten salt, (2) thermal cycling tests in molten salt, (3) x-ray studies of coating structure made before and after moltensalt tests, (4) surface roughness measurements taken before and after molten-salt tests, (5) determination of coating adherence to Hastelloy N substrate, and (6) metallographic studies. Coatings which have promise will be evaluated further in actual molten-salt bearing operation. Bearing Tests. The materials that have sufficient promise will be fabricated into bearings and tested in molten salt. The first molten-salt tests will be conducted in an existing facility which can accommodate a 2-in.-diam by 2-in.-long bearing.4 The facility was last operated with potassium, and its refurbishment to molten salt operation is being studied. The layout of a facility for testing relatively large molten-salt bearings4 is complete. The facility utilizes such existing i t e m s from the MSRE pump program a s the Mark 1 prototype pump tank and rotary element, the structure from the water pump test stand, and a high-temperature molten-salt drain tank. The tester will accommodate a 6-in.-diam by 6-in.-long bearing and will be operable a t speeds to 1800 rpm, temperatures to 1400째F, and bearing loads to 500 lb. The tester will provide basic information and help resolve some of the problems of design, development, fabrication, and operation of moltensalt bearings. The major of these problems is proof testing the selected bearing materials and their mounting arrangement should they have a

4MSR Program Semiann. Progr. Rept. Aug. 31, 1967, ORNL-4191, p. 100.

coefficient of thermal expansion significantly different from that of Hastelloy N. A cross section of the tester is shown in Fig. 7.4. It consists of the aforementioned existing pump modified by removing the impeller, volute, and other structures inside the pump tank to make room for the test journal, bearing and mount, and bearing loading device. The journal is attached to the end of the shaft by means of a conical mount that accommodates relative thermal expansion between these two pieces by sliding along the conical surfaces and at the same t i m e maintains concentricity between the outside of the journal and the shaft center line. The bearing consists of four tilting pads mounted on a ring made of the s a m e material a s the pads. The ring, in turn, is mounted on a structure of leaf springs having radial flexibility sufficient to permit relative thermal expansion between the ring and structure without overstressing the members. The radial stiffness, however, is sufficiently high to transmit the bearing load easily. The bearing is loaded by means of a hydraulic cylinder acting on the lever that penetrates the tank through a bellows and bears against the leaf spring structure. The entire bearing assembly is supported on slender rods that permit the assembly to move with almost zero redundancy upon application of the bearing load. Lugs are provided to resist the bearing torque. The tester will be driven by a low-torque motor to reduce damage in the event of bearing seizure. We expect the bearing to operate predictably in molten salt, so no attempt will be made to m e a s ure such functional characteristics a s operating clearance and attitude angle. The tilting pad bearing configuration was selected for the tester because of its ability to better accommodate those conditions that cause bearing damage and lead subsequently to failure. Damage is most likely to result from (1) the presence of crud or particulate matter in the lubricating salt, (2) thermal distortion of the bearing surfaces or loss of the alignment between the journal and bearing, or (3) mechanical disturbance of the bearing surfaces caused by starting and stopping the journal. To a greater degree than most bearing configurations, the tilting pad is capable of self-cleaning should any particulate matter pass into the bearing clearance. It can also tolerate a relatively high degree of misalignment.

ORNL-DWO 68-1134

MOTOR

BEARING HOUSING

HYDRAULIC CYLINDER LOAD CELL

LOADING ARM

3ft 7in.

(CONICAL MOUNT)-

. 1

I W

R- .

1 Fig.

7.4.

-.R

TO DRAIN TANK

Large-Scale Molten-Salt Bearing Tester.

7.3.6 Rotor-Dynamics Feasibility Investigation An analysis of the rotor-dynamic response4 of a long-shaft pump has been completed by Mechanical Technology Incorporated. The final report is expected soon.

7.4 REMOTE MAINTENANCE Robert Blumberg

P. P. Holz

A fundamental criterion in the design of a molten-salt breeder reactor is that maintenance will be done in a safe and economical manner. Attainment of this objective requires close surveillance of the reactor design during the conceptual stage so that considerations of the basic problems of component maintenance and replacement are included. W e are now establishing the broad outlines of the maintenance system design and development program. Initially, our approach will be based on the maintenance system used a t the MSRE, which consists in disconnecting and replacing components by manipulation of expendable longhandled tools through a movable maintenance shield. The experience with this maintenance system gained a t the MSRE provides a basis for the anticipation and evaluation of problems in the much larger MSBR.

Several factors which influenced the evolution of the MSRE maintenance system and also apply to the MSBR are

1. the nature of the tasks i t had to perform (unscheduled repair and replacement of a large number of different components), 2. the physical character of the reactor plant (basically a chemical processing plant separated into several rooms, roughly according to function), 3. the economics of other approaches to the maintenance problem, generally involving the use of electrically controlled equipment operated completely remotely and limited to exactly prescribed functions. However, in transposing to the MSBR we recognize several differences that cannot be dismissed by simply scaling up from the MSRE. In essence, the MSBR maintenance system development pro-

gram will be concerned with applying MSRE techniques to a larger reactor and to overcoming problems caused by the differences. There are three specific areas under active consideration: (1) the application of the portable maintenance shield concept to the various cells of the MSBR, (2) remote welding for vessel entry and component replacement, and (3) replacement of the graphite moderator elements of the core on a regular basis. Factors involved in each of these areas will influence the reactor equipment layout and component design, so that an early resolution is thought necessary.

MSBR Maintenance Shield The application of the portable maintenance shield concept to the MSBR is important in that this basic piece of equipment will be used for almost all remote maintenance operations. A practical, workable design must be achieved before we can confidently look a t in-cell problems. The design must take into account the shieldblock layout, the method of gaining access through the containment, the layout of components in the cell, and any physical obstructions on the working floor level. Layouts of various shield block and containment membrane arrangements have been started to determine the effect on remote maintenance procedures. Also of interest in this general area are experiments to be conducted to measure the residual radiation levels in the reactor cell of the MSRE. This information will be used to estimate levels that will be encountered in the MSBR during maintenance operations and to deduce whether they will impose limitations on manipulations through the maintenance shield.

Remote Welding We propose to use remote cutting and welding a s the means for disconnecting the s a l t piping for component replacement and for gaining access into the reactor vessel for graphite moderator replacement. The early development of this capability is important in establishing the maintenance concept a s well a s in establishing the space requirements in the plant layout. Figure 7.5 shows schematically the equipment involved in the process for welding the horizontal piping.

ORNL-DWG 68-5535

.OCKS

EXCHANGER

Fig. 7.5.

Schematic Version of Remote Welding Components.

Item 1 is a vehicle which traverses the circumference of the pipe and has the capability of cutting and preparing the pipe for welding, welding the pipe, and then inspecting the joint. Item 2 consists of the long-handled tools which insert and support i t e m 1, provide some minor control or adjustment over the various processes, and provide routing for the service lines. Item 3 is the maintenance shield for the MSBR. Item 4 is the equipment which h a s the primary control of the welding, cutting, and inspection processes. Item 5 is whatever is used to support the components and position the ends of the piping for welding. The rollers shown are schematic. The

design of component supports must include consideration of the weld joint pre-positioning requirements. The development effort will be primarily concerned with an assembly consisting of items 1 and 4 and the connecting cables. Activity thus far has been limited to generating broad outlines of the overall program and to a search for information on the state of the current technology. Significant and applicable work h a s been found in the area of automatic butt welding of pipes in a horizontal position. Two firms build welding equipment that is capable of completing good welds automatically in horizontal piping, but these require manual

installation and direct monitoring. These two firms are Liquid Carbonic Corporation of Chicago, Illinois, a subsidiary of General Dynamics Corporation, and North American Aviation, Inc. Both systems u s e tungsten inert-gas welding and a control system that operates on the usual welding parameters to achieve automatic welding. There are some differences in the detailed electrical and mechanical design, in their capabilities, and possibly-in their state of perfection. While this equipment represents an advanced state of the art for automated welding, no existing system fully m e e t s our specific requirement for remote welding. W e plan to evaluate the use of such equipment in the production of the high-quality welds required in a nuclear system and then to proceed with solving the problems of weld joint design, remote alignment and positioning of the components, and remote inspection techniques which are necessary for providing confidence in the weld quality. Replacement of Graphite Moderator Elements

Because of radiation-induced damage to graphite, some of the moderator elements of the reference

design may have to be replaced every few years. The planning and preparation for this task are important parts of the remote maintenance program and involve four problem areas. These are

1. a miscellaneous area concerned with the general setup of the maintenance equipment, with the provisions for afierheat removal, and with the method for disconnecting auxiliaries, 2. the process for remote cutting and rewelding of the vessel closure, 3. the handling and storage of the very large vessel lid, 4. the handling of new and of radioactive moderator elements. Because of the large amount of radiation and contamination, item 4 presents the greatest p r o b lem. A number of methods are being evaluated that range from use of a huge charge-discharge machine to a simple long-handled grapple a p proach. In handling the contaminated graphite, we have the option of breaking the moderator bars into smaller pieces and loading them into a transport cask below the reactor shield, or of bringing them up into a container above the shield and disposing of them from there.

L, c

8. MSBR Instrumentation and Controls L. C. Oakes

8.1 ANALOG COMPUTER STUDIES 0. W. Burke F. H. Clark

method of Cohen and Skinner, and (2) the lag term of the two groups with the shortest t i m e constants was neglected; for the two groups having the longest t i m e constants, the lag terms were approximated by

S. J. Ditto R. L. Moore

Analog computer studies of the dynamics of MSBR systems were begun. These studies are expected to form the basis for the determination of a suitable control scheme as well a s the safety requirements of the plant. The basic approach is to first simulate in as great detail as possible the various individual subsystems to determine the behavior of these subsystems during postulated transient conditions. These studies will then form the basis for judging the validity of simpler models so that the entire plant can be simulated. An approximate space invariant model of the reactor kinetics yielded the following equations for precursors, C , and power, P:

and the two other delay groups were treated with the transport lag devices. Cases were run with both models, and the results showed comparable and acceptable performance. Preliminary studies of the dynamics of two-fluid MSBR dynamics were made to determine the initial reactor response to certain load and reactivity transients in the absence of any control action. The durations of some of the transients were short compared with the transit t i m e of t h e s a l t in the fuel circulation loop and therefore yielded no significant information regarding overall stability. Figure 8.1 shows the results of the simulation of the insertion of approximately 0.75% 6k/k as a s t e p while the reactor was operating at design point power level of 556 Mw [thermal power of a single module of the four-module 1000 Mw (electrical) two-fluid breeder plant]. Although the power reached a peak value of about 775 Mw in 0.375 sec, the fuel temperature at the core outlet only increased by about 30째F. After the initial peak the power settled to about 625 Mw, at which t i m e the run was terminated, since further changes would be at a rather modest rate which we considered quite amenable to routine control measures.

d P ( f )= - P(t) + xi C k f ), dt A d 1 C k f )= fi P ( f )- xi CJt) - - C(t)

The term CJt 7L)dictates the need for a transport lag generator for each delay group i f they are ta be faithfully simulated. Since only two transport lag generators are available, it is necessary to use approximate methods of computing the transient delayed neutron precursor concentration using equations involving only two transport lags. Two such methods were compared: (1) the six groups were reduced to two effective groups by use of the

'E. R. Cohen and R. E. Skinner, "Reduced Delayed Neutron Group Representations,'' Nucl. Sci. Eng. 5, 291-98 (1959).

86 ORNL-DWG 68-5536 4400

806

LL 0

I-

a a

556

I Fig. 8.1.

900

Response t o Step Reactivity Addition of 0.746%

&/k.

ORNL-DWG

68-5537 i400

-I W

3 l-

a a W a

3I900

Fig. 8.2.

Response to F l o w Coastdown.

Figure 8.2 shows the results of a run simulating the coastdown of fuel salt flow without changing coolant salt flow rate, again with initial conditions at design point. The coastdown assumed was an exponential decay from 100 to 10%flow rate with a t i m e constant of about 6 sec. In this case the power decreased to approximately 12% in about 50 sec and then began to rise slowly as the effects of the excess cooling by the heat exchanger began to reduce the mean fuel temperature in the core. A t this t i m e the temperature of the fuel leaving the core was about 1340째F and was rising at a rate of less than 1째F/sec. The curves of Fig. 8.3 show the response of the reactor and primary loop to a simulated loss of load. In this case the temperature of the coolant salt entering the primary heat exchanger was increased from its design value of 850째F to llOO'F

at a rate of 2S째F/sec. This appears to the primary heat exchanger a s a complete loss of load a t the rate of lO%/sec, since the coolant salt leaving the heat exchanger at the beginning of the transient has not had t i m e to return to the heat exchanger before termination of the study. Thermal lags in the heat exchanger and core caused the power to lag behind the load, but the response appears quite good. These results lead us to the tentative conclusion that the plant would be inherently load following a t the expense of modest temperature changes. They also indicate that it should be quite easy to accommodate rather large load changes using a control system to maintain some desired temperature condition. Although it had originally been planned to extend these studies to include investigation of the details of the required control characteristics, the

ORNL- DWG 68-5538

556

z a

2 I

HEAT EXCHANGER OUTLET I

Fig. 8.3. Response to Load Reduction of 10Wsec.

shift of emphasis from the two-fluid to the onefluid system has caused us to set aside studies of the two-fluid reactor for the present. Our next studies will be aimed at a reference design of a single-fluid MSBR. Simulation of the entire plant will require additional analog computer equipment to simulate transport lags in the piping of the secondary salt loop. An order has been placed for two lag generators, with delivery expected about the middle of the 1968 calendar year. The steam generator has been modeled for examination of its dynamic behavior on the analog. Considerable simplification of the system equations has been undertaken. Further, an attempt will be made to cause dynamic variations which are model dependent rather than real to occur in times short compared with t i m e s of physical interest. We will do this by (1) forbidding unrealistically rapid adjustment of control devices and (2) possibly rescaling time. Such procedures are often used to adapt the computer to a problem which is actually beyond its nominal capability. We cannot,

of course, be assured that the analog computer will perform satisfactorily in this mode. Accordingly, work is continuing toward programming the problem for hybrid computer solution. It should be observed that this problem is nearly identical to a problem done by the British CEGB and Electronic Associates on a hybrid computer, and their experience demonstrates clearly the need for a machine with the capabilities of a hybrid computer to handle this problem properly. These results, unfortunately, are unavailable to us because of proprietary reasons. The model itself has been linearized, a procedure we consider appropriate for control, if not for safety, purposes. The heat transfer coefficient is now permitted to be a function of the velocity of both fluids. Analog controls are provided to permit variation of salt velocity, water pump speed indirectly affecting initial pressure, initial temperature, and throttle opening. The throttle is treated at all t i m e s a s a critical aperture. This treatment causes decoupling of effects downstream of the throttle.

. - ,

d Part

3. Chemistry W. R. Grimes

The chemical research and development effort in close support of the MSBR program includes, as described in this chapter, a variety of studies. A major share of this effort continues to be devoted to the immediate and anticipated problems of the operating Molten-Salt Reactor Experiment (MSRE). Sampling of the MSRE fuel and coolant salts, interpretation of the analyses for major and minor constituents of the melt, and investigation of the distribution of fission products as revealed by specimens exposed t o the pump bowl gases continued as routine. The results of salt analyses continued to show excellent materials compatibility. Significant fractions of several fission products continued to appear in the pump bowl gas space. The fate of fission products in the reactor w a s established in considerable detail. The only unusual behavior continued to be that of the more noble metals such as Mo, Ru, Te, and Nb. These left the fuel by depositing on the walls, and possibly also as a smoke that was carried away in the gas phase. Since the chemistry of the volatile fluorides of these fission products may have relevance to future aspects of the MSBR program, in-

vestigation of the chemical behavior of their fluorides was continued. The prospects for application of fluoroborate coolants i n molten-salt reactor technology continued to appear favorable and has justified a continuing investigation of the NaF-NaBF, and NaF-KF-BF3 systems. As innovations in MSBR design increasingly favor the development of a single-fluid system, efforts to devise efficient chemical reprocessing methods have become more significant. Laboratory studies of recovery of protactinium and removal of lanthanide fission products from the core s a l t by reductive extraction into molten metals continue to show promise and were actively pursued. Solution thermodynamics and electrochemistry of LiF-BeF, melts were investigated in connection with chemical behavior and extractive reprocessing of breeder reactor m e l t s . Development studies in analytical chemistry have been directed primarily toward improvement in analysis of radioactive samples of fuel for oxide and uranium trifluoride and for impurities in the helium gas from the MSRE.

9. Chemistry of the MSRE R. E. Thoma

Chemical behavior in the salt, gas, oil, and water systems of the MSRE has been the subject of a continuous surveillance program since the beginning of MSRE operations. Surveillance is accomplished partly by on-line instrumental analysis. The condition of s a l t systems, however, is

appraised preponderantly from the results of chemical analysis. Numerous samples are obtained therefore as part of this program. They afford a means for determining the stability and compatibility of salt-metal systems, t h e condition and service life of coolants and lubricants, and the

composition, purity, and oxidation-reduction potentials of the fuel salt. In addition, analyses of the fuel salt are performed regularly to measure isotopic composition of the fissile material in order to compute fuel bumup rates. It has been the purpose of this program to apply the experience gained during the period when the MSRE contained 'U- *U fuel salt to operation of the reactor in the near future with 233U fuel salt and subsequently to the design of molten-salt breeder reactors. Results of previous MSRE materials analyses were summarized in the last report of this series. In the intervening period, the MSRE has operated continuously for its longest period of power generation uninterrupted by drain and flush operations. This period of some 25,500 Mwhr has afforded an unprecedented opportunity to determine the extent to which a program of routine analyses is relevant to reactor operations.

9.1 FUEL SALT COMPOSITION AND PURITY

It became evident at the onset of MSRE power operation that the on-line reactivity computation was about ten t i m e s as sensitive for detection of small changes in uranium concentration as individual chemical analyses of the fuel salt could be anticipated to be. Therefore, until more precise methods for the determination of uranium are developed, the chief function of individual analyses 'MSR Program Semiann. Progr. Rept. Aug. 31, 1967, ORNL-4191, pp. 102-8.

Table 9.1.

Run No. Design FP-4 FP-5-7

FP-8 FP-9 FP-10 FP-11 FP-12 FP-13-14

Average Composition of

must come from their indication of long-term trends in fuel composition and from their application to studies of chemical changes in the flush salt. Routine MSRE operations alter the composition of the fuel salt principally by the consumption or addition of "UF4 and by intermixing fuel and flush salts. In current operations the fuel salt has been circulated in the reactor for more than five months without alteration of its composition by intermixing fuel and flush salts. The average composition of the fuel salt during runs 13 and 14 is compared with previous values in Table 9.1. These results reinforce the prior inference that the circulating fuel salt would exhibit excellent chemical stability in nuclear operations and indicate that estimates of fuel Composition, as computed from operation data, are in satisfactory agreement with analytical results. Final evaluation of operating power levels and residual reactivity depend upon such concurrence. Statistical analyses of the chemically determined values for uranium concentration show that control l i m i t s have not been exceeded. However, within these l i m i t s a disparity between computed and analytical values has developed, one in which the results of chemical analyses indicate that the concentration of uranium in the fuel salt is currently lower by -0.02 wt % than that computed from operational data. This trend is suggested by comparison of the average values of uranium as determined from chemical analysis with those computed from operation analysis (Table 9.2). The results of individual chemical analyses of uranium in the MSRE fuel salt are compared with

MSRE

Fuel Circuit Salt i n Mole %

'LiF

BeF2

ZrF4

UF4

65.00 63.36 f0.57 63.29 f 0.72 65.84 f 2.49 64.17.f 0.10 64.65 &. 0.45 64.30 f 0.90 64.46 f 0 . 8 3 64.07 f 0.95

29.17 30.65 f 0.58 30.70 f 0.70 28.57 f 2.12 30.04 f 0.14 29.64 f 0.48 29.89 f 0.81 29.86 f 0.77 29.95 f 0.95

5.00 5.15 f 0.12 5.19 f 0.13 4.82 f 0.35 4.99 f 0.19 4.92 f 0.06 5.01 f 0 . 1 3 4.88 f 0.14 5.15 f 0.13

0.83 0.825 f 0.011 0.824 f 0.011 0.771 f 0 . 0 5 8 0.794 f 0.011 0.791 fO.010 0.803 fO.018 0.794 f 0.02 1 0.819 f0.015

Number of Samples

22 14 8 4

10 41 26 44

90 Table

9.2. Comparison of Analytical and Computed

Values of Uranium in the

MSRE

F u e l Salt

Percent Uranium Run No.

Book (a4

Analytical (a4

A kgU

Mwhr

x 5-7 8 9

10 1la llb 12a 12b 13-14

4.623 4.605 4.591 4.573 4.569 4.579 4.561 4.590 4.564

4.629 4.632 4.603 4.609 4.570 4.571 4.548 4.572 4.543

f0.026 50.011 f0.031 f0.020 fO.018 fO.019 50.022 fO.032 f0.032

+0.3 +1.35 +0.60 +1.80 +0.05 -1.40 -0.65 -0.90 -1.05

Number of Samples

lo3

15.6 4.4 2.5 10.5 12.5 3.0 4.5 1.5 25.5

14 8 4 10 35 6 18 8 44

the nominal book value for uranium as shown in Fig. 9.1. The actual trend in these values is more clearly evident from the control chart shown in Fig. 9.2. The slope of the points shown here was computed from a least-squares program using both weighted (where individual data points were assigned values of 4/u2 - dashed line, Fig. 9.2) and raw values (solid line, Fig. 9.2) of uranium concentration. The results show that a real disparity exists between values of uranium concentration in the fuel salt as computed from operational analysis and chemical analysis. The raw values indicate a difference in chemical and book value of 1.28 kg in total uranium. A similar disparity has also been noted in a gradual decline in the computed reactivity which has been observed in runs 13 and 14 (this report, Sect. 1.2.1) and which seems to have proceeded from the beginning of power operation, a decline which would account for an apparent difference of 0.95 kg in total uranium. We find no chemical basis for this unexplained difference between the anticipated values for uranium concentration of the fuel salt and the experimentally determined values. The reality of the difference seems to be unquestionable and is found in corresponding magnitude in reactivity computations. W e conclude therefore that it will be necessary to reevaluate the method by which the concentration of uranium in the fuel salt is computed.

The overall purity of the MSRE fuel salt, as judged from the concentrations of oxide and chromium, has been maintained during the present report period. An innovation was made in the method of handling samples of fuel salt obtained for oxide analysis or for measurement of the concentration of trivalent uranium in fuel salt (this report, Sect. 2.2). This change is the use of a heated carrier to maintain the temperature of salt specimens after isolation at 200 to 3OO0C and thereby to ensure that the fluorine recombination reaction can proceed sufficiently rapidly to prevent evolution of free fluorine from frozen salt. This technique presumably obviates the possibility that nascent fluorine might remove oxide from the salt. After this new method of handling salt specimens was employed, the oxide concentration in one salt sample (FP-14-53) was found to be 58 ppm, not significantly greater than that found in samples before the heated carrier was used. W e conclude that this result confirms the results of previous oxide analyses and indicates that the oxide concentration of the fuel salt has remained a t about 50 ppm since the beginning of MSRE operations.

c, -

9.2 MSRE FLKL CIRCUIT CORROSION CHEMISTRY In previous power operations of the MSRE, the concentration of chromium in the fuel salt h a s remained constant during periods of salt circulation. Until experiments were initiated in December 1967 to examine the relationship of residual reactivity to temperature, Falt level in the pump bowl, and cover gas pressure, the concentration of chromium in the fuel salt during run 14 was 72 8 ppm. After power was reduced from 7.2 to 5.0 Mw and reactor outlet temperature from 654 to 638OC, we found that the chromium level of the fuel salt had risen to approximately 85 ppm. Other conditions which were subsequently imposed during these tests included additional variation in operating temperatures within the range 638 to 663OC and variations from 5.0 to 9.0 psi in cover gas pressure. The results of current chemical analyses indicate that concentration of chromium in the fuel salt has reached a new steady-state value of 85 ppm. While the cause of t h e recent 13 ppm increase in chromium in the fuel salt is not known, only temperature changes during otherwise steady and

ORNL-DWG 68-5539

4.700

4.680

4.660

4.640

4.620

4.600

-L

4.580

5- 4.560 Z

a 2

4.540

4.520

4.500

4.480

4.460

4.440

4.420 __

MEGAWATT HOURS ( x 4 ~ 3 )

Fig. 9.1.

Comparison of Analytical and Computed Values of Uranium in the

MSRE F u e l Salt.

30 MEGAWATT

Fig. 9.2.

MSRE F u e l Salt

- Normalized Values for Uranium Analyses.

continuous power operation s e e m to be without precedent. The observed rates of corrosion in t h e MSRE have been significantly lower than predicted from thermodynamic data and diffusion theory and might have been expected to produce chromium in the fuel salt at rates which would have caused the salt now to contain 250 to 300 ppm of Cr". The temperature dependence of the equilibrium Cro + 2UF, = CrF, + 2UF, is too low to explain this recent increase of chromium concentration. It h a s been postulated that one of the principal reasons for the unexpectedly low values observed is that the metal surfaces of the fuel circuit have been covered with a film of the noble-metal fission products Nb, Mo, Tc, and Ru about 10 A thick. Results of electron microprobe analysis of the metal surveillance specimens removed from the MSRE in May 1967 lend support to this view in

HOURS ( x i ~ 3 ) Wt %

(analyzed)/wt %

U (book) x 4.65.

that they did not reveal any change in chromium concentration below a depth of 10 p, t h e limit of measurement. Possibly the temperature changes imposed recently caused spallation of t h e noble-metal fission products from the hotter regions of the fuel loop, behavior which could occur if the thermal coefficients of expansion of the noble-metal fission product film and base metal alloy are unequal. Chromium from the freshly exposed alloy surface would then be available for reestablishment of its steady-state activity. If this sort of mechanism is indeed operative, similar behavior should occur

'Furnished through the courtesy of C. Crouthamel and associates, Chemical Engineering Division, Argonne National Laboratory, Argonne, Ill.

u -

on resumption of power operation after periods when the reactor is drained and cooled. Actually, slight increases in chromium concentration of the fuel salt have been observed each t i m e such circumstances have occurred. The significance of the mechanism proposed here is that it requires assignment of previous changes in t h e chromium concentration of t h e fuel salt to the circuit system rather than to t h e drain tanks. If the total amount of chromium represented by this increase, +47 ppm from t h e outset of MSRE operations, were leached uniformly from the fuel circuit, it would correspond to removal of chromium from a -depth of 0.283 mil, or to a n average corrosion rate m i l per 1000 hr of operation with of 1.972 x salt circulating in the fuel circuit.

5U was depleted from or added to the fuel salt have little effect on the relative concentration of 236U. Results of the 236U concentration analyses may be used effectively therefore to compute the output of the reactor. Results of such computations are described in Sect. 1.2.5.

33.6

33.4

33.2

33.0

32.8

9.3 ISOTOPIC COMPOSITION OF URANIUM IN THE MSRE FUEL SALT We have obtained mass spectrometric analyses3 as part of a program of routine analysis of the MSRE fuel salt. Results of t h e isotopic analyses have been accumulated in order to evaluate the potential applications of such analyses to the computation of power generation in the MSRE. The results obtained from the beginning of power operations with the MSRE are summarized in Fig. 9.3. The data shown here indicate moderately good correlation of the spectrochemical determination of 5U concentration with calculated values. The analytical results show a continuous average bias of +0.6% as compared with calculated values. The results of spectrochemical analyses shown in Fig. 9.3 indicate that the concentration of 236U in the fuel salt increases as anticipated, essentially linearly with power generation. Small changes in the 5U/2 *U ratio which have occurred as 3Performed by R. E. Eby, Analytical Chemistry Division, ORNL.

-6 320..56 > 0.4 + 3

I3 W

0.3

0.2

0.1

66.5

66.0

6.5 5

.20

MEGAWATT HOURS ( x 103)

Fig.

9.3.

F u e l Salt.

Isotopic Composition of Uranium in the

MSRE

10. Fission Product Behavior 10.1 FISSION PRODUCT BEHAVIOR IN THE MSRE

S. S. Kirslis

4. an investigation of contamination problems with metal specimens exposed in the pump bowl,

F. F. Blankenship

5. hot-cell tests on the quantity and chemical form of fission products found in the gas phase above the surface of MSRE fuel salt,

The reasons for interest in the behavior of fission products in the MSRE have been discussed in previous reports. The principal practical concern continues to be with the possible deposition of the noble-metal fission products (Mo, Tc, Ru, Te, and Nb) in the graphite core of an MSBR. Neutron economy would suffer i f too large a fraction of these species possessing moderate neutron cross sections were deposited in the highflux region. Information on the concentrations of fission products in MSRE fuel salt is of immediate interest in planning the volatility processing of the current charge of MSRE fuel salt. In the current report period, t e s t s s i m i l a r to those previously carried out were continued, with emphasis on checking and improving the experimental techniques. In addition, several hot-cell tests were carried out on the nature of gas-borne activities above molten fuel salt samples from the MSRE. The following sections will report in some detail on

6. miscellaneous tests. \

10.1.1 Fission Products i n the MSRE Fuel The possibility was suggested in the previous report4 that the occasional high values and wide scatter of measured concentrations of noble-metal activities in fuel salt samples may have been due to contamination of the ladle sampler by activities present in the pump bowl gas phase. To eliminate this possibility, salt samples were taken using freeze-valve capsules of the type previously used for sampling the pump bowl g a s phase. The evacuated 20-cc capsule was sealed by a fusible plug of Li,BeF, which melted when the capsule was lowered below the fuel level in the pump bowl, allowing the fuel salt to fill the evacuated capsule. The enclosed salt sample was thus protected from contamination from outside. In the analytical hot cell, the end of the inlet capillary was plugged with wax, and the exterior of the capsule was leached with acid to remove contaminating activity. The capsule was cut into several sections with a tubing cutter, the bulk of the salt was removed from each section, and the emptied sections were leached with acid to remove adhering salt. The salt, after powdering and mixing, and the interior leach were analyzed

1. the completion and evaluation of fission product analyses from the second s e t of graphite and metal specimens from the core of the MSRE,

2. analyses of MSRE fuel salt sampled by an improved technique,

3. analysis of the pump bowl cover gas,

â&#x20AC;&#x2DC;P. R. Kasten, Graphite Behavior and Its E f f e c t on

MSBR Performance, ORNLTM-2136, p. 4.1. 2S. S. Kirslis, MSR Program Semiann. Progr. Rept. Aug. 31, 1966, ORNL4037, pp. 165-66. 3S. S . Kirslis and F. F. Blankenship, MSR Program Semiann. Progr. Rept. Aug. 31, 1967, ORNL4191, p, 116.

41bid., p. 121. The suggestion was by W. R. Grimes. â&#x20AC;&#x2122;5. S. Kirslis and F. F. Blankenship, MSR Program Semiann. Progr. Rept. Feb. 28, 1967, ORNG4119, p. 139.

95 T a b l e 10.1. ~

Sample No. Sampling date Sampling method

Isotope

Fission Products in ~

FP14-22 11-7-67 Ladle

FP14-20 11-3-67 Freeze-valve capsule

Fission Y ie Id 6.06

8.15

lolo

x io9

0.35

1.86

3.0

3.74

x 10' x io9

0.38

FP14-30 12-5-67 Freeze-valve capsule

FP14-63 2-27-68 Freeze-valve capsule

r 10'

-2.7

2.22

x lo9

8.5

x 10'

3.20

x 10'

8.19

x 10'

1.26

x io9

2.78

x 10'

<1.07 x 1.0'

<2.92

io7

2.38

x lo7

<6.4

x lo6

4.2

x io7

7.0

x io5

4.9 x

x 10'

2.54 x 10'

52.2

x lo7

x lo6

51.6

10'

53.2

<lo'

6.2

1 . 2 8 . ~10l1

1.19 x 10'

1.29 x 10l1

3.1

6.64

x 10" 9.15 x 10' 1.69 x 10"

5.51 x l o l o

3.41

x lolo

4.46 x 10"

8.15

10"

8.46

x lolo

1.41

x 10"

1.45

x 10l1

2.7

6.75

x 10"

4.34

x 1010

6.35

1.59

x 10"

1.16 x 1 0 l 1

5.7

5.96 x 10"

5.36

x 10'

10'

1.40

10'

3.22 -6.2

6.2

4.79

FP14-66 3-5-68 Freeze-valve capsule

Disintegrations per Minute per Grame

8.9

4.7

MSRE F u e l Samples

10"

3.4 x 106

x io9

1.21

x 10"

9.15 x 1 o ' O

9.07

x 10"

1.86

x 10"

9.17

x 10"

9.30

x 10"

1.68

x 10"

6.2

x 10"

4.70

x 10"

1.41

x lo1'

1.48

x 10l1

-2.5

-9.1

10"

a A l l activities calculated back t o sampling time.

kEJ

radiochemically. The inlet capillary was dissolved and analyzed as a separate sample. The results of the radiochemical analyses are reported in Table 10.1, in which each value represents the sum of the activities found in the salt, the interior leach, and the capillary dissolution, and is reported in disintegrations per minute per gram of sample calculated back to the t i m e of sampling. The concentrations of the fission products with stable fluorides (rare earths, alkalies, alkaline earths, and zirconium) were in satisfactory agreement with previous analyses of samples taken by the ladle method. However, the concentrations of the noble-metal fission products were at least an order of magnitude lower than in previous samples. In the first sample, FP14-20,the bulk of the noble metals in the total sample were found

in the dissolution of the inlet capillary. It was suspected that noble-metal activities adhering to the outside of the capillary tip were covered by the drop of wax used to seal the tip before leaching the outside of the capsule. In subsequent samples the outside of the capillary tip was sanded to remove adhering activity before applying the wax; only a s m a l l fraction of all activities was then found in the capillary. The leach of the interior of the emptied capsule usually contained an appreciable fraction of the total noble metal in the capsule, occasionally as much as half. Most or all of the 95Nb was in this leach. Since the salt sample contacted the interior of the nickel capsule in the molten state for only about 15 min, these observations indicate that noble metals deposit rapidly from molten salt on a clean nickel surface. It is therefore possible,

particularly in the case of 9sNb, that the salt sampled was previously depleted in noble metals by deposition on the exterior of the capsule. This difficulty can be avoided experimentally in several ways: (1) the capsule or the capillary can be constructed of or plated with a noble metal, (2) only a long capillary with a salt seal at its tip need be submerged in the fuel salt, or (3) the salt seal can be made of a high-melting salt like LiF, allowing the capsule exterior to saturate with noble metals before the sealing salt dissolves. Deposition on the capsule exterior probably does not deplete the salt sampled of noble metals by more than a factor of 2, except in the case of 95Nb. However, it leads to another analytical difficulty which may help to explain the considerable scatter of the noble-metal data in Table 10.1. The exterior of the capsule is leached repeatedly with 2 N HNO, at 85 to 95OC until the last leach contains less than 1%of the gamma activity of the first leach. Noble metals, however, tend to replate on nickel from the leaching solution. Normally 15 to 20 leaches are required to reach the 1% level. Therefore, some noble metal probably remains on the exterior of the capsule when leaching is halted. Since the noble-metal concentrations inside the capsule are very low, a slight remaining contamination of the exterior might account for much of the activity found in the inside leach. On this basis, the lower figures in Table 10.1 are probably more nearly correct. In any case the scatter of the noble-metal concentrations in the fuel salt samples is of little practical consequence since the values are so small, representing on the average less than 1% of the quantities which would have been found if none had left the salt phase. By contrast, the fission products with stable fluorides show concentrations within 30% of their calculated values, although scatter exceeds the claimed 20% analytical accuracy.

In summary, it is felt that the estimates of noblemetal concentrations in fuel salt are much more reliable from freeze-valve capsule samples than from ladled samples. Previously reported data should be discounted. For salt-seeking species, both sampling methods are equally satisfactory, and previously reported data should be valid. The planned fluorination of the fuel charge in the MSRE to remove UF, is simplified by the fact that only 1% or less of the noble-metal fis-

sion products remain in the fuel. Special trapping techniques are required to separate volatile noblemetal fluorides from UF,.

10.1.2 i n the

Fission Products Gas

MSRE Cover

The analyses of seven gas samples taken in the mist shield region of the MSRE pump bowl by the freeze-valve capsule technique have been thoroughly discussed in previous reports. The results indicated that a s m a l l amount of fuel salt m i s t and an appreciable share of the total noblemetal nuclides produced by fission were by some mechanism injected into the cover gas phase (see Table 10.3). Uranium concentrations in the gas samples were usually higher than those calculated from the amounts of salt mist indicated by gsZr analyses. Only a single cover gas sample, FP14-67, was taken in this report period, since experimental t i m e was preempted by other high-priority work. A freeze-valve capsule of slightly different design (Fig. 10.1) was used, and the sample was taken during 5-Mw operation under otherwise normal conditions. The analytical results, together with those of several previous runs, are shown in Table 10.2.

10.1.3 Deposition of Fission MSRE Cover Gas on Metal

from

Products Specimens

Tests in which small metal specimens were exposed to the gas phase and fuel phase in the pump bowl gave the first qualitative indications of unusual volatilization and plating behavior of the noble-metal fission products. High concentrations of noble-metal activities were found on specimens exposed to cover gas or fuel, while fission products possessing stable fluorides showed much lower activities. Results have been reported previously from a large number of tests run with several types of metal specimens (Hastelloy N, Ag,

6Zbid., pp. 13841. 'MSR Program Semiann. Progr. Rept. Aug. 31, 1967, ORNL-4191, pp. 116-19. 'S. S. Kirslis and 'F.F. Blankenship, Reactor Chem. Div. Ann. Progr. Rept. Dec. 31,1967, ORNL-4229, pp. 8-10.

ORNL-DWG

/ 3/4-in. -OD i

68-6079

NICKEL TUBE 0.020 -in. WALL

%-in.

I '46-in.

Fig. 10.1.

Ni TUBING

WELD ROD

Freeze-Valve Capsule for Sampling Pump

Bowl Cover Gas.

Ni, Au, and stainless steel) under a variety of reactor operating conditions and exposure times. The results of these t e s t s were disappointing from the standpoints of obtaining quantitative information and of showing variations in deposition under different test conditions. The reproducibility of results was poor, with variations between duplicate runs often as high as an order of magnitude. Deposition on the different metals was not significantly different. Exposure t i m e had little effect; samples exposed for 1 min collected as much activity as those exposed for 10 min. Changes in reactor operating conditions had little effect on deposition. Even samples exposed several hours after stopping the fuel pump or two days after draining the fuel out of the reactor showed as much deposition of noble m e t a l s as normal samples. The addition of beryllium to reduce the fuel caused no significant change in deposition.

The only variable which had a definite effect on gas phase deposition was a reactor shutdown of two weeks or longer. Then decreases in noblem e t a l deposition by an order of magnitude were observed. In the current report period, efforts were concentrated on trying to identify the experimental factors causing difficulty. Comparative runs were made in which the specimens (Ag, Hastelloy N, and stainless steel) were fluorinated, hydrogenated, or left in the normal slightly oxidized condition. Exposure times were 1 and 10 min. The six t e s t s gave practically undistinguishable results. Deposition of noble metals was approximately halved on the fluorinated specimens compared with the others, but this s m a l l difference is probably not significant. The last two tests identified the basic difficulty. In the first of these, the metal specimens were lowered in the sampling pipe t o a position 2 ft above the pump bowl. During the 10-min exposure, the usual helium flow of 200 cc/min was passed down the tube. The test assembly was then withdrawn and packaged for shipment to the analytical laboratory in the usual way. Approximately the same amounts of noble-metal nuclides were deposited on the test specimens a s in normal exposures in the pump bowl. The second test consisted in hanging a standard test assembly on the cable latch, leaving it there for one day as in the customary test procedure, removing it with the sampling cubicle manipulator, and packaging it for shipment and analysis in the usual way. Again the amounts of noble-metal activities found on the specimens were nearly as great (within a factor of 1to 10) as in normal pump bowl exposures, Clearly the contamination of the samples during the various manipulations involved in t h e test procedure before or after the actual pump bowl exposure was responsible for most of the observed activities. This difficulty was not identified sooner because of the astonishing reproducibility of the contamination. Variations of several orders of magnitude would be expected rather than the observed zero to one order. This uniformity suggests that most of the contamination may have occurred while the test assembly remained for a day on the latch in the highly contaminated 1C area. The mechanism by which the contamination spreads from the pump bowl to the various surfaces is un-

Table 10.2.

Experiment No. Sampling date Operating time, days. Nominal power; Mw Be addition

Features Accumulated Mwhr

Fission Products in MSRE Cover Gas as Determined from Freeze Valve Capsules

FP10-11 FP10-22 12-27-66 1-11-67 14.5 off, 12.6 on 14.5 off, 27.7 on 7.4 7.4 After 5.5 No Regular Regular 16,200 13.600

FPll42 FP1146 4-11-67. 02:49 4-18-67. M:28 65 on. 1.5 hr off 14 off, 72 on 0 7.2 After 8.40 g No Pump off 1.2 hr Regular 27.900 29,100

FPl l-53 5-2-67, l a 4 3 14 off, 86 on 7.2 No Helium bubbles 31,700

FP12-7 6-2-67, 0 6 S O 92.3 on, 42.5 off

FP12-26 7-17-67, M:03 46 off, 23 on 0 7.2 No After 37.8 g Power off 42.5 days Regular 32,650 36.500

FP14-67 36-68, 06:03 13 a t 7 Mw, 6.6 a t 5 MW 5.0 No

Regular 62,770

Disintewations per Minute in Total Sampleb 9 9 ~ 0

lo3Ru lo6Ru 132Te

6.06

2.04 x 10"

3.0

3.80 x

0.38 -4.7

"Te

0.35

95Nb

6.2

95Zr 140Ba 1311

5.73 x 10'0

1.36 X 10" 2.63 x

io9

7.74

1o7

5.08 x 10'0

6.2

lo7 <2.9 x lo6

<2.2 x 1o7 =

6.35

2.75 x 10'

3.48 x 10'

-3.1

"sr

4.79

lrlAg

0.019

14'ce

-6.0

144~e

-6.0

235째C

-6.7

io9 x io7

C3.26 x =

9.75 x

io9

2.09 x 10'

2.03 x io9

1.05 x 1011

io9 8.2 x io7

2.31 x 10"

1.57 x 10"

io9 9.49 x io7

1.12 x 1010 4.03 x 10'

3.35 x 101

1 . 8 8 10" ~

7.98 x 10'

3.51 .X io9

2.17 x 10'

1.05 x 10'0

2.26 x io9?

2.51 x

4.64 x

1.15 x 1011 6.45 x 10' <4.4 x

io7

1.3 x -2

io9 lo7

1.8 x 10'

2.74 x 10"

3.07 x 10" 1.18 x 1010

4 x IO9 1.7 x 10' 4.17 x

8.64 x

io7

5.03 x 10'

3.16 x lof0?

1.21 x 10"

6.6 x 10'

io9 2.98 x io7

2.14 x 10' 4.01 x 10'

6.16 x 10' 5.7 x

io9

2.13 x 10'

9.81 x 10' 4.08 x

io9

8.63 X 3.72 x

lo9 io9

1.67 x l o l o 8.35 x 10'

3.71 x

3.11 x 10"

io9

4.79 x 10' 4.17 x 10' 0.55

9.2

1-70 x

io9

1.83 x

lo*

1.33 x 10'

5.5 x 10'

3.86

io9

2.18 x

3.52 x

1.71 x 10' 25

.Duration of previous shutdown and of continuous operating time just before sample was taken. bDisintegraticns per minute calculated to the time of sampling or of previous shutdown. CMicrograms in sample.

certain but probably involves rubbing of the cable, latch, samplers, and manipulators against each other. These observations make it necessary to discard any quantitative interpretation of the pump bowl exposure tests. The source of the contamination is still principally the gas phase above the fuel in the pump bowl, so that the qualitative conclusion that noble-metal activities predominate in this gas remains valid. Also the conclusions from test FPll-50, in which graphite and metal specimens were at least partially protected from contamination by a perforated metal screen, probably remain sound. These observations on contamination confirm the explanation given for the high concentrations of noble-metal fission products in fuel salt sampled by the open ladle technique. A capsule with a sliding sheath h a s been designed which will protect m e t a l or graphite samples from contamination except when the test assembly is in the pump bowl.

10.1.4 Examination of MSRE Surveillance Specimens After 24,000 Mwhr

In a previous report, l o a summary was presented of the more significant results from the examinations and analyses of the graphite and Hastelloy specimens which were exposed to fissioning molten salt for 24,000 Mwhr in axial positions in the MSRE core. Since then, the analyses have been completed, and the middle graphite bar (Y-7) was also sampled by a different method and analyzed for a few isotopes with a germanium diode gamma spectrometer. The same specimen was also examined with an electron probe in an attempt to detect surface contamination and by proton bombardment to determine the concentration profiles of lithium and fluorine. The significance of these additional results will be briefly discussed. Completion of Analytical Results. - The concentration profiles of 'Mo, 2Te, 3 R ~ "Nb, , "Zr, "Sr, 140Ba, and 23sUin the exposed graphite specimens were described in the last report. The completion of analytical work on these nuclides caused no significant changes in the pro-

files given. All the samples were also analyzed for 6 R ~ .Without exception, the 6 R values ~ were a factor of about 20 lower than the '"Ru values, so that the concentration profiles have identical shape. In addition, some selected samples were analyzedfor "'Ag, 137Cs, 13'1, "Y, 14'Ce, 144Ce, and 147Nd. The ''Ag profile behaved like lo3Ru, the activity dropping steeply through four orders of magnitude in 10 mils. The 13'Cs behaved in an unusual manner, as it also did in the 7800-Mwhr exposure. Its activity dropped two orders of magnitude in 10 m i l s and then began slowly rising; 13'1 behaved similarly. It has been suggested that both these species can diffuse in graphite. The 147Nd, 144Ce, and 14'Ce exhibited fairly straight plots of the logarithm of the activity v s penetration distance, and the slopes were each quite similar to that for 14'Ba. This behavior does not agree with the theory that the slope should vary inversely with the half-life of the raregas precursor. The half-lives are "very short," less than 1sec, 1.7 sec, and 16 sec, respectively, for the precursors of 147Nd, '44Ce, l 4 'Ce, and "'Ba. Yttrium-91, with a 10-sec krypton precursor, gives a slope less steep than 'Ba. With its very short-lived xenon precursor, it is puzzling how 147Nd got into the graphite at all. From the fact that the three rare earths had similar concentration gradients, it may be guessed that they entered and diffused in the graphite as rare earths rather than a s xenons. For each of the species which gave straight plots of log activity v s distance, the slopes were steeper for the top graphite sample than for the sample in the middle of the core, although the graphite types were supposedly identical. Both of these samples showed distinctly steeper slopes than the bottom sample, which was of the more porous "lattice stock" variety. The concentration profile plots revealed an interesting indication of the difference in permeability of the same piece of graphite from different sides. For the top graphite sample, the concentration profiles of all nuclides, without exception, were higher for penetration from one of the narrow sides than for penetration from any of the other sides. Sampling by Sanding. Five graphite blanks were obtained during the course of milling the irradiated samples by milling a clean piece of the same type of graphite on the same milling machine.

'MSR Program Semiann. Pro& R e p t . Aug. 31, 1967, ORNL-4191, pp. 131-35. "Zbid., pp. 121-28.

100 It was very disturbing that analyses of the blanks sometimes showed more activity (particularly for %Io, 3 R ~ , 6 R ~ and , sNb) than the irradiated sample just previously milled. Confidence was destroyed in all analyses giving lesser values than the blanks. Nevertheless, these analyses behaved in a very consistent manner, resulting in smooth concentration profiles. To check the validity of the sampling method, a remaining stored piece of the irradiated middle graphite bar (Y-7) was handled by a sanding method that permitted sampling from the inside out (see Sect. 10.2). The concentration profiles obtained tailed off toward the center of the specimen in a manner very similar to that observed with the milled specimens. Absolute comparisons between the two sets of data usually agreed to within better than an order of magnitude. Since the area of graphite sampled by the two methods differed by a factor of about 30, local variations in graphite density and permeability might explain the difference in results. The results of the sanding method restored much of our confidence in the milling method of sampling. Certainly the reality of the tails of the concentration profiles was established. The simplest explanation of the high blanks is that the blank graphite specimen somehow became contaminated inside the hot cells, where the milling was done.

Electron Probe Examination. A s m a l l sample of the irradiated middle bar graphite 01-7) was mounted for electron probe examination for possible identification of any surface deposits. The sample was shipped to the Argonne National Laboratory, where an instrument for examining radioactive samples was available. The electron probe examination, which could be used to within 1or 2 p of the edge of the graphite, detected no impurities. Typical detection l i m i t s were 0.02 wt % for tellurium and 0.04 wt % for uranium and technetium. Lithium and Fluorine Concentration Profiles i n Irradiated Graphite. A proton bombardment method

for determining the concentration profiles of lithium and fluorine in irradiated graphite is described in Sect. 10.3. The results for the middle graphite bar (Y-7) indicated that the molar concentration of lithium exceeds that of fluorine at all points inside the graphite. This observation proves conclusively that the fission products observed in irradiated graphite do not permeate as fluorides.

They probably are present either as carbides or in the metallic form. Fission Product Distribution i n the MSRE. - Results have been reported above for the quantities of fission products in the fuel salt and in the cover gas of the MSRE. The previous report' gave values for the quantities deposited on graphite and on Hastelloy N on May 5, 1%7, after 32,000 Mwhr of power operation. In addition, the total inventory of each nuclide at this date is available from a recently developed computer program for calculating the total amount of each isotope from the operating history of the MSRE. An approximate material balance for May 8, 1967, based on these data is given in Table 10.3. It differs from the balance given in the previous report' because of the revised concentrations of noble metals in the fuel salt. The balances are good for nuclides with stable fluorides which remain in the salt phase; fair balances were obtained for 2Te, SNb, and 9M0;and 3 R i~s poorly accounted for. To enter the cover gas data into a material balance, the quantity of each species lost per day (observed concentration t i m e s volume of a v e r gas flow per day) must be compared with the production of that species by fission per day. As discussed in the previous report, l 3 it is likely that the gaseous concentrations measured in the mist shield region are higher (by a s m a l l factor) than those in the bulk of the cover gas. This will tend to yield high material balances for species like 99M0,whose concentration in the gas phase is large.

10.1.5

Hot-Cell Tests on Fission Product Volatilization from Molten MSRE Fuel

It has been established that the gas phase above the fuel in the MSRE pump bowl contains appreciable concentrations of fission products, particularly the noble metals. However, no hard information exists regarding the chemical and physical nature of the volatilized species and the mechanism of volatilization. In view of the potential

''Zbid., pp. 126-27. 12Zbid., p. 128. 13Zbid., p. 118.

101 Table 10.3.

Isotope

Approximate Fission Product Distribution in

Inventory in MSRE (dis/min)a

Percent in Fuel

Percent on Graphite

MSRE After 32,000 Mwhr

Percent on Hastelloy N

Cover Gasb (%/day)

x 10" 7.91

0.94

10.9

40.5

5.86

0.83

10.0

70.0

lo3Ru

3.36

0.13

6.6

14.9

"Nb

4.40

0.044

36.4

34.1

95Zr

6.00

96.1

"sr

5.02

77.0

0.26

1311

4.00

64.0

1.0

9 9 ~ 0

132

0.03

0.06

5.7 0.14

BThis total inventory was calculated from the power history of the MSRE. bThese values represent the percentages of daily production rate lost to the cover gas per day.

practical importance of the volatilization phenomenon as a method of removing noble-metal fission products from the fuel circulation system of an MSBR, it is desirable to investigate these more basic questions. There are serious limitations on the types of t e s t s which can be carried out in the existing MSRE pump bowl facility. However, it was noted that two of the cover gas samples taken after reactor shutdown (FPll-42 and FP12-7 in Table 10.2) contained sizable concentrations of noblemetal activities. This indicated that fission was not essential t o the volatilization process and suggested the possibility of studying volatilization from a molten sample of MSRE s a l t in a hot cell, where experimental possibilities would be much less limited. First and Second Hot-Cell Tests. - The experimental plan was originally designed to distinguish between the volatile fluoride and metallic colloid hypotheses l 4 on the nature of the volatilizing species. We planned to measure the activities in the gas phase above the surface of molten MSRE salt when (1)pure helium was slowly passed over the quiescent surface, (2) a mixture of 5% H, in helium was similarly passed, (3) pure helium was bubbled through the salt, and (4) the H,-He mix141bid., p. 119.

ture was similarly bubbled. The plan was based on the assumptions that only high-valent fluorides would volatilize in procedure 1, since gentle helium flow would provide no agitation to suspend metallic particles in the gas phase; that H, in procedure 2 would reduce volatile fluorides at the melt surface and prevent them from leaving the fuel salt; and that metallic particles should be suspended in the gas phase by the bubbling gas in procedures 3 and 4. Two hot-cell experiments of this type have been carried out: one with a sample of MSRE salt 35 days old and one on the day after sampling. The SO-g salt sample was contained in a flanged gastight stainless steel reaction vessel (Fig. 10.2) fitted with a gas inlet tube, through which a dip tube line could also be inserted, and a large tube through which the "probe" tubes were inserted to with in. of the molten-salt surface. A side tube from the large tube provided an alternate path for gas flow. The gas inlet tube and the alternate line were connected to valves on a simple gas manifold (Fig. 10.3) carrying two other valves which were connected to a vacuum pump and to the gas supply line. The tank helium was purified by two titanium sponge traps at 6OO0C, one near the tank outside the cell and the other just before entering the gas manifold. The second titanium trap was added after the first test showed signs of fuel hydrolysis.

102 ORNL-DWG 68-6080

f '/4 in.

SWAGELOK FITTING '/4-in.-OD

ALTERNATE 1.INE

1/4-in.-OD GAS INLET LINE SWAGELOK FITTING SODA-LIME 3/g-in.-OD TUBING NaF PELLETS FELT-METAL FILTER SALT SURFACE 509 MSRE FUEL SALT

REACTION VESSEL TUBE FURNACE

Fig. 10.2.

Hot-Cell Test Reaction Vessel.

The traps were not heated when the 5% H,-95% He mixture was used in order to avoid TiH, formation. W e then depended on the efficient adsorptive drying action of degassed titanium sponge, The flow of gas was measured outside the cell with a low-flow-range rotameter. To avoid exposure of the fuel salt sample to air a s much a s possible, the top of the usual 50-g sampling capsule was cut off just above the salt level, and the capsule was placed in the flanged reaction vessel without fracturing the solidified lump of salt. A cap was placed on the probe tube line, and the capsule was evacuated while the reaction vessel was heated to 200OC. A sheathed ChromebAlumel thermocouple wired to the reaction vessel was used to indicate and control the furnace temperature. During the evacuation, every joint in the system was leak tested with acetone and a Hastings gage. When no more leaks could

be found and the gage reading dropped below 10 p, the system was filled to 1 psig with purified helium, and the cap on the large tube was slightly opened to provide a helium flush of about 15 cc/min. The reactor was then heated to 6OO0C, the cap was removed, and the probe tube was quickly inserted. The Swagelok fitting on the probe tube had previously been attached a t a level such that the lower end of the probe was within 9, in. of the molten-salt surface. The only exit for the helium flow was up through the probe tube, so all in. of the moltenthe gas sampled passed within salt surface. Because of the resistance to flow of the Feltmetal filter in the probe tube (Fig. 10.2), the system pressure usually rose to 2 to 3 psig at a helium flow of 10 to 15 cc/min. In the first run, with 35-day-old salt, the flow was continued for 1 hr. In the second run the duration of flow was decreased to 30 to 40 min.

103 ORNL-OWG 68-6084

TO HASTINGS GAGE AND VACUUM PUMP PRESSURE GAGE 0-30 psi0

HOT C E L L WALL

I I

PRESSURE GAGE 0-30 psi0

/ALTERNATE

h a

LINE

h A

T i TRAP 6OOOC

ROTAME T EA

0-400cc/min NEEDLE VALVE

$I r-2I I

REACTION VESSEL

n He TANK

MSRE SALT

)PRESSURE REGULATOR FURNACE

5 % Hp. 95% He TANK

Fig. 10.3.

Complete Apparatus for Hot-Cell Tests.

The first probe tube was then removed, and the large tube was capped loosely. The titanium traps were cooled, and the 5% H,-95% He mixture was passed through the system t o flush out the pure helium. The flow of gas was adjusted at 10 t o 15 cc/min, and the second probe was inserted, run for 1hr, and removed. With the large tube capped, a Swagelok union on the gas inlet line was opened, and the dip line was placed in position so that its end was about 3.5 in. below the m e l t surface. The alternate line was open during this operation to provide flushing flow over the salt. With the gas inlet line attached to the dip line, the cap on the large line was slightly opened, the alternate line valve was closed, and a bubbling flow of 5% H,-95% He was started a t 10 to 15 cc/min. The third probe was then inserted, run for 1hr, and removed. The gas supply was converted back to pure helium, the titanium traps were heated, and a fourth probe was run for 1hr with helium bubbling at 10 t o 15 cc/min. In

the bubbling runs the pressure usually had t o be increased gradually t o about 5 psig to maintain the 10- to 15-cc/min flow rate. In the first run the pressure in the reactor was released at the end of the bubbling tests by loosening the Swagelok fitting on the probe. This required all the gas in the long line to the outside manifold t o bubble rapidly through the dip tube as the pressure decreased to atmospheric. It is estimated that the bubbling flow was about 200 cc/min for 2 min at the end of these tests. In the second run, it was realized that the pressure across the dip tube could be equalized by opening the valve on the alternate line. By this means the pressure was relieved without a sudden violent bubbling through the dip tube. Samples of the fuel salt in the reaction vessel were taken with small ladle-type samplers after each test with helium flow. The probe tubes in these t e s t s were '/-in.-diam nickel tubes containing a 1-in.-long empty section

104 at the bottom, a Y,,-in.-thick Feltmetal filter (100% retention of particles larger than 4 p), and l-in.-long sections filled with NaF and soda-lime pellets. In the analytical hot cell, the ends of each tube were plugged with s m a l l rubber stoppers, and the outside surface was leached with acid until the last leach contained less than 1%of the gamma activity of the first leach. The tubes were then cut into a bottom section, a Feltmetal section, an NaF section, and a soda-lime section. Each sample was separately dissolved, and the solution was analyzed for 12 fission products and 235U. The analytical results of these tests are given in Tables 10.4 and 10.5, with all activities corrected back to the t i m e of salt sampling. Gamma scans of the samples from the first test showed that the predominant activity was l 3'I. This behavior was undoubtedly due to fuel salt hydrolysis, indicating some water impurity in the reaction vessel atmosphere. Much less 'I was found in the samples from the second test, in which the leak testing was more thorough and the various experimental operations were performed more smoothly with the benefit of experience from the first run. However, the salt surface appeared green and scum free after both tests, so the degree of hydrolysis could not have been excessive in the first test. Since a 35-day-old sample was used in the first test, the short-lived "Mo and 13,Te activities could not be detected in many of the samples. Even where values are reported, they are not reliable. The results of these tests are very instructive in spite of the wide scatter in results within each run and between runs. The following comments apply to both tests but more specifically to the second test, whose results showed less scatter. With gas passing slowly over the quiescent molten fuel surface, appreciable quantities of all the major fission products were found inside the probes, with the noble metals predominating as in analyses of MSRE cover gas. There was no difference in volatilization behavior when H, was added to the sweep gas. If the noble metals had been present a s high-valent fluorides, they should have been reduced to nonvolatile states by H,. Thus the noble metals are probably present in the gas phase as a colloidal suspension of tiny metallic particles. The presence of activities such a s "Sr, 95Zr, 140Ba, 141Ce, and 235U(in the relative proportions found in fuel salt) inside the probes

showed that the fuel salt was also present in the gas phase a s a colloidal mist. Simple vaporization of fuel salt could not account for either the quantities or the relative proportions of the activities found in the probes. This observation has disturbing implications regarding the proposed distillation separation of fission products from the bulk components of fuel salt, since appreciable quantities of fuel salt would volatilize a s mist as well a s vapor. For each nuclide the quantities deposited in the bottom empty section of the probe and on the filter section were similar (more so for the second test). About 10% of these quantities were usually found in the NaF and soda-lime sections beyond the filter. The gaseous suspensions of noble metals and of fuel salt are thus surprisingly stable, and at least 20% is composed of particles less than 4 p in diameter. When helium or 5% H,-95% He was bubbled slowly through the melt (second test), the amounts of activities found in the various sections of the probes were very similar to the amounts when the gases slowly passed over the melt. Thus the gas phase above the melt appears to be a s easily "saturated" with the colloidal suspensions by the quiescent surface as by slow bubbling. (This observation is contraindicated by a laboratory mockup experiment in which the slow bubbling of helium through molten Li ,BeF4 produced visible droplets of salt on a horizontal metal plate 1 c m above the salt surface.) However, when the gases were bubbled rapidly through the melt for 2 min a t about 200 cc/min (first test), the amounts of all activities in all sections of the probes increased by factors of 10 to 1000 compared with the slowflow cases. Therefore the amount of fume or m i s t formation is greatly enhanced by turbulent gas-fuel contacting. The readiness with which gaseous suspensions of noble metals and of fuel salt are formed above the highly radioactive fuel melt invalidated the original presumption that only fission products with volatile fluorides could leave the melt under the gentle sweep conditions. However, the evidence from the hydrogen gentle sweep runs and particularly from the hydrogen bubbling run in the second test is nearly incontrovertible that volatile fluorides are not involved in a significant way in causing the noble-metal fission products to become gas borne. A second line of argument

Table 10.4.

F i r s t T e s t of Volatilizotion of Fission Products from MSRE F u e l

..--

Disintegrations per Minute per Total Sample' Flow Conditions

'-U

Samplea lo3Ru

10-15 cm3/min pure He over surface

B FM NaF SL

10-15 cm3/min 5% H, over surface

B FM NaF SL

10-15 cm3/min 5% H, bubbling through salt

B FM NaF SL

10-15 cm3/min pure He bubbling through salt

B FM NaF SL

io6 lo6 lo4 lo4 3.77 x lo' 4.22 x lo5 3.39 x lo4 1.47 x lo4 3.85 x lo8 2.45 X lo'

7.11 x 4.30X 8.14 x 6.70 x

7.5 1.73

x x

lo6Ru

"Nb

lo5 1.80 x 10' lo5 1 . 3 X lo6 lo4 5 2.3 x lo5 lo4 55.9 x lo4 3.01 x lo6 9.53 x lo8 3 . 5 0 ~lo4 1.6 X lo6 5.76 x lo3 '"3.7 x lo5 3.41 x lo3 -9.0 x lo4 4.7 x lo7 0

4.87 x 3.59X 1.58 x 1.00 x

1.7 X 10' lo7 7.6 X lo6 lo7 -1.5 x io6

1.5 x 2.14 x lo8 1.52 X lo' 1.65 X 9.8 x 1.28 x 108 -6.4 x lo7 -6.3 x

lo7 lo8 lo6 io6

7.15 1.67

x lo7 x io8

1.72 x 10'' 7.84 X 10' 3.53 x 10' 7.73 x lo7

9 9 ~ 0

,Te

12'Te

"2,

Low

LOW L~~

Low Low

1.99 X lo6 1.70 L~~ -1.5 L~~ <6.9 L~~ 52.2

Low

7.45

Low

LOW

Low Low

1.3 5.3

L~~

Low

1.6 x

Low Low

Low

7.8 -1.1 x 10" 7.8 -4.2 x 10' -3.4

x 10' x lo' x 10' x 10"

-1.2 x 10l1 5.9 -1.9 X lo'* 9.8 X 10" -2.7 x -1.4 x IO" -8.5 x 1olo 5.6 x 10'

x lo8 x lo5 x lo4 x lo4 x lo6 1.53 x lo8

141ce

(pg total)

lo5

lo7 lo5 lo4 lo3 1.96 x lo8 1.43 x lo6 3.73 x lo5 2.08 x lo5

-1.9 X lo7 5.86 X 10'' 6.2 X lo7 2.86 X 10" 4.2 x io6 -1.2 x lo6 1.9 x lo6 3.31 x lo5

5.18 x 10" 2.35 X lo1' 9.00 x lo5 4.36 x lo5

7096.00 3532.00 0.143 0.149

2.70 x lo' 1.17 X 10" 6.24 x lo5 7.75 x lo5

414.00 1720.00 0.297 0.367

2.2 x lo7 3.8 X lo7 6.0 x lo6 3.3 x lo6

x lo6

x lo5

3.26 x lo' 1.25 X 10' 8.8 x lo5 2.09 x lo6

6.72 x 1.1 x <1.7 x a.3 x

6.20 0.064 0.086 0.027 22.7 0.157 0.071 0.066

B ' is bottom empty 1 in. of probe, F M is section including the Feltmetal filter, and N a F and SL are the sections packed with NaF and soda l i m e respectively. 'Activities calculated back to fuel sampling time: 10-23-67, 0944 AM.

Second Test of Volatilization of Fission Products from MSRE Fuel

Table 10.5.

Disintegrations p r Minute per Total Sampleb Flow Conditions

Sample*

10 cc/min pure He over surface

B FM NaF

SL 10 cc/min pure He over surface

10 cc/min 5%Hz over

9.10 6.40 <2.8 8.7

x 106 x lo6 x 10' x 105

1.88 x 10' 1 . 0 9 ~io9 1 . 9 0 ~i o 7 2.78 lo7

7.38 lo7 3.81 x i o 7 2.44 x l o 6 3.53 x 10'

2.5 5.4 <4.1 5.5

x x x x

x lo7

2.9 4.8 -2.6 1.4

x x x x

1.94 x 10' 3.34 x 107 2.63 x io7 1.04 x 107

7.51 3.13 1.73 1.18

x 10' x io7 x io7 x lo7

3.6 5.4 2.5 1.3

x x x x

x 10' x 106

1.66 x 107 8.68 x l o 6 9.9ox10' 7.98 x 106

2.64 3.08 1.57 7.36

x 10' x 10'

6.2 3.2 2.2 1.2

x lo6 x 10' x 10' x 106

3.40 5.43 8.33 1.23

5.29 x io7 1.06 l o 9 4.44 x 10' 3.49 x l o 7

2.35 6.07 2.94 3.88

2.3 4.1 -8.1 7.9

x lo6 x lo6 x 10' x 10'

2.47 5.26 1.26 1.07

8.32 x 1.22 x 1.14x 5.05 x

-2.9 x lo6 8.8 x 106

10' 106

8.W. x l o 7 2.99 x lo7 3.50x 10' 1.47 x 10'

5.64 7.22 1.73 1.42

lo9

1.71 x 10'

FM N ~ F

1.29 x 10' 2.51 x io7 1.62~10~

1.04 x

io7

2.03 4.60 733 7.94

1.78 x i o 7 1.37 x i o 7 3 . 5 6 ~l o 6 8.39 x 106

3.23 5.00 1.48 3.05

B FM NaF

SL 10 cc/min pure H e bubbling through salt

x io7 x 106 x 106

x x

io7 lo6

"-1.7 x 10' 1.47 x 10' 5 . 2 3 ~lo6 7.55 x 106

-4.9 x 106 1 . 0 6 ~l o 7 1.07 x 106 2.16 x lo6

6.59X 10' 6.85 x l o 7 6.8Ox1O6 8.18 x 10'

2 5 4 x lo6 1 . 0 9 ~l o 7 1.59x 10' 1.71 x 107

67.5 x 10'

56.8 x

B FY NaF

SL Fuel salt, d L min-' g-'

"ke

-2.46 x lo6 6.63 x io7 1-86 x lo7 1.64 x i o 7

SL 10 cc/min 5%Habubbling through salt

lZ9Te

-7.24X lo6 3.21 x io7 4.63X10' 2.46 x 10'

lo6

9 9 ~ 0

'"Te

lo3Ru

FM NaF

surface

9sNb

'06RU

10'

106 10' 10'

1.8 x 106

'"9.6 x 10' 2.5 x 106 1.5 x 106 0

3.2 107 1.3 x lo6 1.76 x lo6 4.06 -5.40 3.12 1.82

x io7 x 10' x lo6 x 106

-lo9

1.18 x

10' 10'

io7

x lo6

x 10) x io7 x 106 x 106

106 106

10' 10' 10'

106 106

10'

<7.9 x 10'

2.07 5.51 5.47 7.69

x x x x

106 106 10'

10'

"'~e

Toto1 U 9'zr

1311

2.58 x 10' 9.56 x 10' 1.03 X 10' 1.78 x 10'

2.59 9.06 5.29 -0.5

x lo6 x 106 x 10' x 10'

2.75 x 10' 1.50 x 1010 1.77 i o 9 3.92 x 10' 9.47 6.29 1.26 7.75

1 . 3 9 ~lo6 5.16 x l o 6 2.31 x 10' 1.24 x lo6

2.77 7.84 3.20 1.98

x x x x

10' 10' 10' 10'

1.27 6.23 7.38 1.95

x lo6

x 10' x 106

4.85 7.66 5.66 1.81

x x x x

105 10' 106 10'

1.69 4.50 4.18 1.09

x 10' x 10'

x 10'

106

1.60 1.71 2.00 1.79

x lo6 x lo6 x lo6 x 10'

x 10' x lo6 x 10'

x lo7 x 10'

8.64 x 1o1O

x 10' x 106 x 106

x to7 x 10' x io9 x 10'

"~r

3.64 x 1.20 x 1.03 x 1.34 x

10' 106

10' 10'

2.99 x 10' 1.20 x 106 1.20 107 3.50 x lo6 3.03 6.96 6.70 2.04

x 10'

8.24 x 10' 6.43 x 10' 1.98 x io9 8.51 x 10' 1.71 l o 9 3.13 109 7.03 x l o 9 3.30 x 10'

3.36 2.54 1.71 2.11

x io7

x 10' x 10' x 10'

1.36 x 10' 5.29 x l o 6 1.11 x 106 7.11 x 10'

7.98 9.76 5.87 2.53

x 10' X 10' x 10s x 10'

2.46 4.64 1.44 8.29

7.30

10"

x lo6

x 10' x 106 x lo6 x 10'

1.22 x

loll

1.48 3.83 6.52 1.91

x 10'

x 1OlO x lo9 x lo9

3.11 X 10"

3.07 1.71 2.42 1.42

x x

lo6 lo6

x 106 x io7 x 10' x 106 x 106

x 10' x 10' X 10'

6.82 x

lolo

'"Bo -4.6 -1.1 <1.6 1.7

x lo' x 106 x 10' x 10'

0.06 0.18 0.14 0.04

2.6 x 9.WX -0.7 x -3.1 x

10' 105 10' 10'

0.09 0.44 0.58 0.27

6.8 x 1.1 x 1.10 x 2.8 x

10'

0.16 0.61 0.26 0.07

106

10'

3.15 x 10' 2.9s x io7 '"8.1 x 10' '"3.0X 10' -8.1 x 10' 1.81 x 10' 3.1 X 10' 3.2 X 10'

29.9 2.77 0.28 0.05 0.26 0.75 a 20 0.29

1.55 x 1011

*B is bottom empty 1 in. of probe, FM is oection including the Feltmetal filter, NoF and SL arc the section0 pocked with NaF and soda-Hme respectively. bActivitiem calculated back to fuel sampling t h e : 12447.8:45 AM.

107

bT,

to the same effect may be based on t h e demonstrated readiness of fuel salt particles t o become suspended in the gas phase. If fuel salt so easily becomes gas borne, the same should be true of noble-metal particles in the fuel. To account for the preferential volatilization of noble metals, it may be necessary t o postulate that the colloidal particles are concentrated at the salt-gas interface. Although the hot-cell t e s t s provide strong evidence as to the nature of the gas-borne activities, the question of mechanism remains unanswered. The formation of gaseous colloids from the quiescent molten surface may not be explained by normal physical processes. One possibility is that recoils from beta emissions near the surface of the molten salt may cause the ejection of tiny particles of fuel salt and of noble metals into the adjacent gas phase. Another suggestion' is that differences in thermodynamic contact potentials between metals, salt, and the gas phase would tend to eject m e t a l particles from the salt phase into the gas phase. It has also been suggested16 that the bursting of very tiny gas bubbles, perhaps formed by radioactive decay, might cause aerosol formation. Third Hot-Cell Test. - A third hot-cell test was designed to (1)confirm previous results using similar probes, (2) determine whether the suspended metal or fuel particles in the gas phase were electrically charged, (3) attempt to determine the size of the particulates in the gas phase by examining with an electron microscope a filter through which the exit gas was passed, (4) t e s t the diffusion behavior of the gaseous activities by observing the distribution of activities deposited on the inside of a t-in.-diam vertical tube whose open end was near the fuel surface and whose top end was closed, and (5) obtain an indication of particle size distribution in the gas above molten fuel by passing the gas through a long copper tube and measuring the distribution of activities down the length of the tube. It was also intended t o carry out several of these tests at intervals over the course of a month to determine the effect of activity decay on the volatilization process. HOWever, the furnace used to heat the reaction vessel

"Letter from Jere Nichols to F. L. Culler, Jan. 8. 1968. '6Personal communication from J. Braunstein, Reactor Chemistry Division.

burned out after the completion of the first sequence of these experiments, ruining the reaction vessel in the process. Samples from these five runs were delivered to the analytical laboratory, and results are not yet available. Fourth Hot-Cell Test. - The burned-out furnace in the hot cell was replaced, and a fourth hot-cell test was started with a fresh 50-g sample of MSRE fuel salt, FP14-69. Because of experimental difficulties and an intervening weekend, the first two runs of this test were carried out five days after the salt was sampled. The sample was kept at 2OOOC during this period t o avoid radiolysis. The first run contained a probe tube of the usual type and two electrodes for measurement of the charge on the particulate matter in the gas phase above the molten fuel. This test will not be described in detail since analytical results are not yet available. In the second run the test assembly was a stainless steel rod to which were attached s i x %-in.diam copper screens spaced at levels 1, 1.5, 2.5, 3.5, and 4.5 in. from the molten-salt level. The copper screens were of the type used to hold electron microscope specimens and had previously been coated with a thin (600-A) vaporized layer of carbon. The gas exiting from the reactor passed at 15 cc/min for 40 min through a '/i-in.-diam. tube which surrounded the rod holding the screens. With this experimental arrangement the exiting gas traversed the screens at a linear velocity of about 0.5 cm/sec, with good opportunity for the particulate matter to deposit on the screens. The ?,-in.-diam tube and the stainless steel rod were then removed from theireactor, and the s i x copper screens were carefully loosened from the rod and dropped into s m a l l plastic bottles. Through the plastic bottle, no sample read more than 300 mr/hr of gamma activity, so that the screens could be examined in the electron microscope in Building 3019. Gamma scans of the five lower samples showed high 'I activities (" lo8 dis/min per sample) which masked the probable presence of other gamma activities. The 'I contamination of the topmost sixth sample was lower (4.6 x lo5 dis/min), so that the presence of '%Io (3.1 x loJ dis/min), '03Ru (1.7 x IO6 dis/min), "Nb (2.8 x lo5 dis/min), and "OLa (5.9 x lo4 dis/min) could be detected. These activity readings were taken 7.0 days after the s a l t was sampled. The high

v2,

108 1 3 9 activities on most of the samples are again ascribed to fuel sample hydrolysis by traces of water in the reactor. The degree of hydrolysis was probably very slight, since special care was exercised to make the system leak tight and since the system was evacuated to a pressure below 10 p with the fuel salt at 2OOOC before the runs were started. At this writing, the three topmost copper screens (at 2.5, 3.5, and 4.5 in. above the molten salt) have been examined in the electron microscope. A sizable quantity of particulate matter was observed on the screens, often covering several percent of the areas viewed. There appeared to be three discrete particle size ranges: very fine particles 35 to 180 A in diameter, medium-size particles 1000 to 2000 A in diameter, and occasional large particles more than ten times larger than the medium-size particles (Fig. 10.4). Most of the area of the deposited material was represented by the fine-size particles. A curious circular pattern of particles was often seen, particularly on the two topmost screens (Fig. 10.5). An exact circle was outlined by medium-size particles. The interior of the circle was filled with randomly deposited medium and fine particles. The area density of the fine particles was the same inside and outside the circles, but the density of medium-size particles was at least ten t i m e s a s great inside than outside the circles. At high magnification, it can be seeh that many of the medium-size particles are transparent (Fig. 10.6). The small particles can be seen inside the outline of the medium-size spots. It is possible that the fine particles are deposited on top of the larger particles, but most of the larger spots are pale gray compared with the darker color of the s m a l l particles. Also a faint halo is visible around many of the larger spots. Most of the medium-size spots appear to be round. Several of hexagonal shape can be discerned. Figure 10.4 shows a number of cubical shapes. Some small- and medium-size triangles (usually very dark) are seen in Fig. 10.6. A dark triangle is often visible at the edge of a large pale spot. These observations suggest that the mediumsize spots, particularly those in the circular patterns, are thin f i l m s rather than thick three-dimensional particles. Microscopists who have looked at the circular patterns say they are very similar

to patterns obtained when s m a l l droplets of solution evaporate on a flat surface. It is possible that MSRE fuel salt particles on the screens may pick up atmospheric moisture over the course of several days exposure to air before examination. The most abundant phase in fuel salt, Li,BeF,, is known to be deliquescent. The solution so formed could spread over the carbon film to a thin layer. When this film is evacuated and heated by the electron beam, the water could evaporate and form the characteristic circular patterns. Close examination of Fig. 10.6 shows that many of the smaller spots are also pale in color. These may represent only partial solution. The darker spots with no pale areas may be materials which do not dissolve. An electron diffraction pattern (Fig. 10.7) is shown of the material in the area in the light square of Fig. 10.8. This type of pattern is characteristic of salts (like the fuel salt) but is not the type of pattern obtained from graphite or metals. An attempt is being made to identify positively the material in diffraction patterns of a number of different areas on the screens. The remaining three samples will be examined with the electron microscope, and care will be taken with future samples to avoid exposing them to atmospheric moisture. The observations reported are of considerable significance, since they represent the first hard direct evidence that colloidal particles are t o be found in gas sweeping slowly by the surface of MSRE fuel salt. The diffraction pattern identifies part of the volatilizing material as a salt, probably fuel salt. In addition, we now have some idea of the particle sizes we are dealing with. Most of them are extremely s m a l l of a size which could well be ejected into the gas phase by beta recoil. It will be desirable to expose some of the electron microscope screens in the MSRE pump bowl gas space and in the access tube as soon as possible. Plans have already been made to expose s o m e in the MSRE off-gas line.

10.1.6 Miscellaneous Tests Two tests which do not fit into previous categories will be reported below. Surface Salt from the MSRE Pump Bowl. - A number of observations in previous tests, particularly the finding of high concentrations of noblemetal fission products in the gas phase of the

109

Fig. 10.4.

Electron Micrograph of Particles in Gas Flowing over MSRE Salt. 88,000~.

110

Fig. 10.5.

Circular Patterns Frequently Found on Screens in Gas Above MSRE Salt. 32,OOOX.

Fig.

10.6.

Electron Micrograph Showing the Transparency of Some Particles.

96,000~.

112

Fig.

10.7.

Electron Diffraction Pattern of Particulate Matter Above MSRE Salt.

113

! I

Fig. 10.8.

Area of Deposit Giving the Diffraction Pattern of Fig. 10.7.

114 MSRE pump bowl, suggest that the noble-metal fission products may be highly concentrated at the fuel-gas interface and may even be incorporated in any surface scum or foam floating on the salt. There is basic and practical interest in finding an answer to this question. Considerable thought has been given to the experimental problem of sampling the fuel surface itself. The two major problems are contamination of the sample by the gas phase (or by deposition from the fuel phase) and dilution of the sample by salt from below the surface. Several rather elaborate devices have been designed but have not yet been tried. By accident, we may have obtained a good surface salt sample by a method which would be difficult to duplicate intentionally. Sample FP14-60 was to have been a 50-g sample for the third hotcell test. It was taken in the usual large nickel capsule with windows near the top to admit the salt. When the sample arrived at the hot cell, it was found that the capsule contained only 5 g of salt. Although the normal 50-g sampling procedure was used, it appears that the bottom of the capsule windows must have been exactly at the fuel salt level. If it had been higher, no salt would have entered the capsule; if it had been lower, the capsule would have filled completely. It is likely that the ripples in the fuel surface occasionally lapped a little salt into the capsule. Unfortunately the capsule must have been contaminated inside and out with noble metals from the gas phase, and on the outside by deposition from the fuel melt. Since the sample was therefore not an ideal one, it was decided to handle it by the procedure used for ladled 10-g samples and to compare the results with those from ladled 10-g samples. Large differences in activities would be apparent. The 50-g capsule was therefore cut off about 1 in. above the bottom, and the salt was removed by shaking in the Wig-L-Bug (as for ladled 10-g samples). Salt samples were then weighed out and analyzed for the usual radioactive nuclides and for Ni, Cr, and Fe. Radiochemical analyses in disintegrations per minute per gram, corrected back to the t i m e of sampling, were 1.49 x 10" for 99Mo, 4.31 x 1 O ' O for 132Te,9.04 x l o 9 for 1 0 3 R ~4.07 , x lo8 for lo6Ru, "4.7 x lo9 for "Nb, 1.40 x 10" for 95Zr, 8.77 x 10" for 89Sr, 1.09 x 10" for l4ICe, 9.26 x loio for 144Ce, 8.78 x 10"

for '"Ba, and 5.96 x 1 O ' O for l3II. When these figures are compared with the values in Table 9.2 of the previous report, l 7 the value for 99M0 is higher than all except one, 'j2Te is higher than the average by a factor of 3, lo3Ru is less than two and greater than four values, lo6Ru is higher than all values by an average factor of 2, "Nb is less than all values by an average factor of 0.2, and the remaining salt-seeking species have similar values. The results of analyses of Ni, Cr, and Fe were not excessively high: 152, 90, and 159 ppm respectively. These results indicate no large concentration effect for noble m e t a l s or metals of construction at the fuel surface. The data for 95Nb in fact indicate a depletion effect. These indications should be regarded as suggestive rather than conclusive, since it has not been proven that the sample analyzed was a surface salt sample and since the comparative method used assumes without proof that the different sample containers were contaminated to a similar degree by noble m e t a l s in the gas phase and in the fuel phase. Deposition of Noble Metals on Nickel from Cover Gas and from Fuel Salt. - In a number of previous pump bowl tests, the deposition of noble metals on the stainless steel cables used to suspend fuel sampling capsules was examined to determine whether deposition was greater at the fuel-gas interface than in the cover gas or in the fuel. While there was much scatter in these data, the deposition on the interface region was usually higher (by a s m a l l factor) than on the higher or lower regions. In an earlier section, it was pointed out that this type of observation was made uncertain by the contamination problem. In a special test with an unrelated objective, FP14-55, a nickel rod inside a perforated nickel basket was lowered into the pump bowl so that the middle of the rod was at the level of the fuel salt surface. This provided a sample of metal exposed to the interface region in a container which offered some protection against contamination by handling. The nickel rod was cut into a gas-phase region, an interface region, and a fuel-phase region. The position of the gas-fuel interface was determined by examination of the rod with a low-power hot17MSR Program Semiann. Progr. Rept. Aug. 31 , 1967, ORNL-4191, p. 120.

. U

115 cell microscope. The activity on each section was leached off, and the leaches were analyzed radiochemically. The gas-phase and interface regions showed very similar depositions of noble metals. The deposition of noble metals on the liquid-phase region w a s higher by factors of 30 for ”Mo, ‘03Ru, and and 5 for 13*Teand 95Nb. Since the deposition was so disparate between gas- and fuel-exposed regions, the deposition on the interface sample should have been heavily weighted by the portion of this sample which was submerged in fuel. The results suggest that the visual determination of the fuel-gas interface position on the rod was at fault. The difference in surface appearance (drop-shaped watermarks and a dark deposit) that was taken to indicate the salt level may actually have represented a region well above the salt level. In this case, the results given represent merely the difference in deposition from the gas phase and from the fuel phase. They are markedly different from the previous (questionable) results on stainless steel cables, where similar deposition of noble metals from both phases was observed.

10.2 FISSION PRODUCT DISTRIBUTION IN AN MSRE GRAPHITE SURVEILLANCE SPECIMEN D. R. Cuneo H. E. Robertson

F. Dyer L. Bate

In a previous report, * determinations of fission product distribution in MSRE surveillance specimens after 24,000 Mwhr were reported. Samples were obtained by milling off layers of the graphite surfaces i n a plane parallel to the longitudinal axis of a 0.47 by 0.66 in. cross section bar. The specimens were about 4.5 in. long. The resulting powder was dissolved and analyzed radiochemically for selected fission products. Near the surface, layers as thin a s 1 m i l were milled off a s samples; subsequent samples at greater depths were as thick as 10 mils. Generally, a total depth of about 50 m i l s was sampled from the four sides of each specimen. The following inherent uncertainties in this sampling procedure were recognized: (1) some contamination was certainly carried by the steel milling surfaces to the next

sample from the preceding (and hotter) sample; (2) with the milling apparatus used, it is possible that the amount of graphite removed per cut was not uniform over the length of the specimen; and (3) it was difficult to be sure that the amount of graphite removed was uniform over the width of the specimen. In order to determine i f the “tails” or concentrations of fission products observed at considerable depth in the specimens were real, the following sampling scheme was devised. The rectangular-shaped specimen was sawed longitudinally at midplane. A small (0.25-to 0.30-in.-diam) core was then drilled from the cold inner (new) surface to the outer (hot) surface. The hot surface of this small graphite core was glued to a cold graphite coupon, which in turn was glued to a precisionground tool steel piston. The piston f i t s into a 3.5-in.diam holder. This assembly is shown in Fig. 10.9. A dial-indicating micrometer allows reading changes in the piston vertical position to direct readings of 0.1 mil. The samples are obtained by describing a figure 8 motion on a clean surface of emery paper resting on a precision lapping plate. The resulting powder is taped in place on the emery paper. This procedure, first devised by Lonsdale and Graves, h a s been de”H. K. Lonsdale and J. N. Graves, “Diffusion of Thorium in Pyrolytic Carbon Coatings,” pp. 18-35 in Coated Particle F u e l s Research a t General Atomic, Oct. 30, 1964-April 30, 1965.

Fig. 10.9.

Apparatus Used for Sampling of Graphite

Core. V i e w on right shows location of graphite core

“S. S. Kirslis and F. F. Blankenship, MSR Program Semiann. Progr. Rept. Aug. 31, 1967, ORNL4191, p. 121.

cemented t o floating piston. assembly.

V i e w on l e f t shows entire

116 scribed in detail elsewhere. We believe that use of this apparatus certainly eliminates the uncertainties described above for the milling operation type of samples. The weak point in this procedure for these surveillance specimens is that only a small portion of the specimens is sampled when compared with milling samples. The tapedsn samples are transferred to a gammaray spectrometer for determinations of fission product concentrations. This approach to the probl e m was begun at a time when the surveillance samples had cooled beyond determination of shortlived isotopes of Mo, Te, I, and Ba. It is felt that these short-lived isotopes can also be determined satisfactorily by counting the entire powdered sample, a s we have done with the longcooled samples, for longer-lived nuclides. The first core-drill sample was taken from the portion of specimen Y-7 from which some 60 m i l s of the surface under consideration had been removed earlier by milling. l 8 Small samples (1 to 10 m i l s ) were removed from this core by grinding, starting at the cold (midplane of the specimen) end. These samples were counted using an NaI crystal that had only sufficient resolution to allow good counting data for "Zr, 13'Cs, and '44Ce. (For subsequent samples a germanium diode was available, thus allowing greater resolution and determination of more nuclides.) Part of the data obtained from these grinding samples are reported in Table 10.6. The last entry in the table, 70 mils from the original surface, was obtained from a sample taken 10 m i l s deep from the surface left following the milling. From this depth (10 mils) on to the "new" surface, the concentrations of the three nuclides dropped, a s much a s an order of magnitude. We have no explanation for this anomaly. However, it is clear that the tails observed from the milling-type samples are real since we see measurable concentrations extending to midplane in Table 10.6. The data in Table 10.6 show little change in the l 3'Cs concentration with depth. This nuclide has a 3.9-min gaseous precursor (' 7Xe), which allows considerable t i m e for diffusion in the gaseous state.

2oR. B. Evans III e t al., "Recoil of Fission Products in Pyrolytic Carbon," Ges-Cooled Reactor Program Semiann. Progr. Rept. Sept. 30, 1965, ORNL3885, pp. 13149.

Table 10.6.

Fission Product Concentrations Varying

with Depth in Specimen

Y-7 as Determined

from F i r s t Grinding of Specimen

Depth from Original Surface' (mils)

"2,

235

Atoms of Nuclide per Cubic Centimeter of Graphite '37cs

144~e

4 x 10'

5 x 10'4

2.5 x io12

200

1 x 10l2

6 x 1014

150

6 x 10l2

7 x 1014

2 x 10l2

125

1 x 1013

8 x 1014

5 x 10l2

100

2.5 x 1013

1 x 10'5

5 x io12

2 x 1013

(midplane)

2.5

1 x 10'5

k 10"

6.5 x io12

1013

1 x lo1'

8 x 10l2

10~3

1 'X 1015

2 x ios3

*Original surface is that before cuts were removed by milling operation.

To check the possibility that zirconium was produced in situ from uranium contained in the graphite before its exposure to the salt, a corresponding piece of unirradiated material was subjected to activation analysis. Analysis showed the natural uranium content of the specimen to be <0.06 ppm, whereas about 70 ppm would be required to yield the amount of 95Zr found at about 100 to 125 m i l s from the original surface. Next, a core (305 m i l s in diameter) was drilled through a portion of the Y-7 sample which had not been milled. Initially, 251 m i l s were removed from the 467-mil-thick specimen in the following increments: one 2-mil sample, nine 1 m i l each, five 5 m i l s each, five 10 m i l s each, one 15 mils, two 25 mils each, and two 50 m i l s each. At a later date, the remaining 216 m i l s were sampled by grinding ten increments 1 m i l each, five 5 m i l s each, five 10 m i l s each, three 25 m i l s each, and one 56-mil portion, Partial results of the gammaspectrometer counting data are shown in Figs. 10.10 and 10.11. We believe that the graphite core surface represented by the left ordinate in each figure is that surface which was exposed directly to free-flowing fuel salt in the reactor and that the surface represented by the right

117

0 RNL- DWG 68-6082

DISTANCE FROM SURFACE ( m i l s 1

Fig. 10.10.

60 50 40 30 20 DISTANCE FROM SURFACE (mils)

Concentrations of Nuclides v s Depth in Graphite Starting a t Surface i n Contact with Free-Flawing

F u e l Salt ( a ) and with Stagnant F u e l Salt ( b ) .

118 ORNL-OWG 68-6083

toi7

d6 5

to'

to' 0

20 30 40 50 60 DISTANCE FROM SURFACE ( m i l s )

Fig. 10.11. Concentrations of Nuclides v s Depth i n Graphite Starting o t Surface in Contact w i t h Free-Flowing F w l Solt (a) and w i t h Stognant F u e l Soh (b).

ordinates was exposed only to stagnant fuel salt which found its way through slits available where adjacent pieces of graphite were stacked together. However, this portion of Y-7 was left unmarked when stored since it was intended for permeability studies. From these curves and data not shown representing sampling to the midplane of the core, we observe the following. In Fig. 10.10, the "Nb follows a similar pattern from one surface to the other, ending at the back (or stagnant) surface at a greater concentration by a factor of about 15 to 20. The 95Zr concentration gradient is considerably steeper on the back side of the specimen, with no detectable amount between 15 and 85 mils.

Ruthenium-103 was found from the "hot" salt surface to midplane, but only to a depth of 25 m i l s from the opposite surface. Ruthenium-106 followed the shape of the 3Ru curve to about 50 m i l s and then showed a drop at 65 to 70 mils; thereafter it was below a value of 1x IO'*. Starting from the opposite surface (Fig. 10.10b) it dropped below limits of detection a t 15 mils. The lo3*lo6Ruvalues are similar, mil by mil, when we consider that the fission yield of lo3Ru is 0.38% and that of '06Ru is 2.9%. Cerium-141 has an l&min barium and a 3.7-hr lanthanum precursor, while 144Cehas only short-lived ones. These different origins of the two isotopes may help

119 explain the marked differences in penetration in graphite as shown i n Fig. 10.11; the melting point of barium is 7250C2 and operating temperature of the graphite is believed to be 650 to 75OOC. If the ordinates in Fig. 10.10b and 10. l l b represent a graphite surface exposed to stagnant fuel salt, then the higher surface concentrations and lesser depth of penetration for several of the nuclides may have occurred as follows. If this surface had contact only with a limited amount of salt for a long period of reactor operation, then penetration by any nuclide observed, except 7Cs, which probably penetrated from the opposite surface, would be limited by fission products available. However, at the end of reactor operation, barren salt circulated through the reactor for a very short time (relatively) did not have sufficient contacting time with this surface to remove fission products as thoroughly as a t the opposite free-flowing s a l t surface. In Table 10.7 are given details of the distribution of seven nuclides starting from the graphite surface in contact with free-flowing s a l t and extending to a depth where -100% of all nuclides have been found. These data present in tabular form the data represented graphically in Figs. 10.10a and 10.11a. Since these data and those reported previously by Kirslis' were obtained by quite different techniques and analytical procedures, i t is interesting to compare the results, as seen in Table 10.8. While there are apparent differences in results from the two methods, i t appears likely that the method we have described, that is, grinding of a small graphite core and dry graphite powder gamma spectrometry, yields usable results. The small core samples, of course, have the distinct possibility of including a crack, which could markedly affect the results. In Table 10.9 the data for the total atoms of individual nuclides, surface t o midplane, are given for the second sampling (representing the first half of this graphite core) and the third sampling (representing the second half). We see that with the exception of the large difference for "Nb, the total nuclide contents of the opposite halves of the core are very nearly the same. This is somewhat surprising in view of some of the large differences noted for nuclide concentrations per unit

21Handbook of Chemistry and Physics, 47th ed., by R. C . Weast, Chemical Rubber Co., 1966.

of volume that we found by comparing data in the left and right halves of Figs. 10.10 and 10.11. However, in most cases, 90 to 95% of each nuclide was found in the first 10 m i l s from the graphite surfaces. Another (the third) core was drilled from specimen Y-7. It was sampled by grinding throughout its entire length of 470 mils. Table 10.10 shows a comparison of total atoms of each nuclide per square centimeter of graphite surface from one surface through to the opposite surface for the second and third cores. Reasonable agreement is found for six of the seven nuclides; the discrepancy for l 4 'Ce between the two comes about because of low concentrations found in the first 10 m i l s of the third core. Table 10.11 gives comparisons of concentration profiles for seven nuclides through the second and third cores. By the hand grinding of small cores from graphite surveillance specimens, we can expect to obtain representative results for the distribution of fission products in the specimen. Short-cooled specimens of graphite from the MSRE may prove too radioactive for hand grinding, and methods of performing this operation remotely are being investigated. The attractive features of this method include savings by direct gamma spectrometry of the graphite powder rather than dissolution of the much larger milling samples and the necessary radiochemical separations of each nuclide. The grinding procedure has also allowed determinations of L i t and F- on each freshly exposed surface of the core.

10.3 PROTON REACTION ANALYSIS FOR LITHIUM AND FLUORINE IN MSR GRAPHITE R. L. Macklin E. Ricci T. H. Handley J. H. Gibbons D. Cuneo Consideration of the proton-induced reactions 7Li(p,n) and 1 9 F ( p , a y ) suggested that they might be used to measure the concentration of these target nuclides in graphite as a means to determine the extent to which MSR fuel salt components penetrate into the graphite under irradiation. Heretofore, none of the standard analytical techniques, including neutron activation, has shown much promise for this application at the few ppm wt %) level.

Table

10.7.

Distribution of Fission Products i n

MSRE Sample Y-7 a s Found by Second Core Grinding Starting a t Graphite Surface Soh and Extending to a Depth of 201 M i l s

in Contact with Free-Flowing

Total Distance Surface (mils)

Percent of Total "Zr i n This Cut

Cumulative $5 "2,

Percent of Total "Nb in This c u t

Cumulative % "Nb

Percent of Total lo3Ru in This c u t

Cumulative % lo3Ru

Percent of Total lo6Ru in This Cut

Cumulative % lo6Ru

Percent of Total 137Cs in This c u t

Cumulative % 137Cs

Percent of Total 141Ce in This c u t

Cumulative % 14'Ce

Percent of Total 144Ce in This Cut

0-2

73.9

80.3

82.2

85.3

12.5

38.3

75.7

8.8

82.7

10.2

90.5

9.7

91.9

4.3

89.6

1.9

14.4

15.8

54.1

8.7

84.4

2.6

85.3

1.6

92.1

1.3

93.2

1.4

91.0

0.2

14.6

3.6

57.7

0.9

85.3

1.1

86.4

1.3

93.4

1.2

94.4

1.5

92.5

0.7

15.3

5.1

62.8

1.0

86.3

1.1

87.5

1.1

94.5

0.9

95.3

1.2

93.7

0.6

15.9

3.3

66.1

0.8

87.1

0.5

88.0

0.5

95.0

0.4

95.7

0.6

94.3

0.5

16.4

2.7

68.8

0.5

87.6

0.8

88.8

0.7

95.7

0.5

96.2

0.6

94.9

0.5

16.9

1.8

70.6

0.6

88.2

0.3

89.1

0.4

96.1

0.3

96.5

0.6

95.5

0.5

17.4

1.4

72.0

0.3

88.5

2.0

91.1

0.2

96.3

0.1

96.6

0.3

95.8

0.4

17.8

1.0

73.0

0.2

88.7

0.3

91.4

0.2

96.5

0.2

96.8

0.3

96.1

0,s

18.3

1.7

74.7

0.3

89.0

from

95.7

98.2

98.4

126

98.5

99.4

99.2

151

98.7

loot

201

99.3

1ow

Cumulative % 144ce

31.0

88.2

91.9

58.7

95.4

94.5

99.4

1ow

66.9

96.5

94.9

99.6

1ow

87.6

98.7

96.9

99.1

121

Table

10.8.

Comporison

Results as

for

Nuclide

Obtained

Concentrations

are

Concentrations by Milling

atoms

of nuclide

“2,

lo3Ru

lo6Ru

13’CS

141Ce

1 x 1016

2.3 x 101’

2.7 x 10” 1.8 x 101’

6 x 10”

3 x 10’6

2.3 x 1014

2 x 10’6

1.3

1.4 x 10’5

6 x 10”

1.7 x 1016

1.3 x 10’5

1.2 x lOI6

2.2

3.8

x 10” a x 10’5

7.7

from

Y-7

of graphite

Surface

(mils)

different

but

2.5

1.4 x 1014 a.5 x 10’3

3.9 x 1o13 3.5 x 10’3

1.1 x 10’4

4.6

x 1013 4 x 10’3

1.8 x 1014

7.4 x 10’3

2.8 x 1013

5 x 1014

1.3 x 10’4

2 x 10’3

6 x 1014

x 1015

1.5 x 10’5

2.5 x 1o15

1.5 x 10’5

2.2

7.5 x 1014

x 10’5

1.5 x 10’5 1 x 1014

1.3 x 1015 1 x 10’5 4 x 10’4 2.1

10.9.

to Midplana

Atoms

of Nuclides

by Core-Drill (Second

Opposite

Sampllng)

Surface

per

Sampling

(Third

and

from Midplana

Salt @ 10”‘)

Surface

Square Surface to

Sampling)

Second, Hot

x 1014

3.5 x 1013

by 41/-in.-long specimen in May 1967; samples were dissolved Sampl& ground from surface of 0.3-in.-diam core of graphite by gamma spectrometer using a germanium diode.

Total

Found

x 10’4 6.1

to Y-7.

Nuclide

40 1.8 x 10’4 6 x 1013

x 1015

similar

1.3 x 10’5 3 x 10’4

2 x 10’5

Table

-w

Specimen

3 x 10’4

Centimeter

Surveillance

x 10’5

*Method A: samples milled from 0.66~in.wide nuclides determined radiochemically. Method B: December 1967; dry powder samples were analyzed bSpecimen

centimeter

1.3 x 101’

B ‘44Ce

cubic

Depth

Graphite

Methods

Methoda 2

“Nb

Grtnding

per

Sample Nuclide

in MSRE

and

Third, Stagnant

Salt (x

10’1’)

“Nb

3.47

72.

“2,

0.73

0.75

lo3Ru

3.35

4.43

’ 06Ru

1.0

1.17

13’cs

7.75

‘41Ce

1.6

2.13

144Ce

1.14

1.22

10.1

Surface

and in

122 Table 10.10.

T o t a l Atoms of Nuctide per Square Centimeter from One Surface to Opposite Surface for T w o Cores Removed from

MSRE Surveillance Specimen Y-7

'Nb

952,

103Ru

Second core

1.5 x 1017

7.5 x 10'8

7.8 x 1017

2.2 x 1017

1.8 x 10'8

3.9 x 1017

2.3 x 1017

Third core

9.0

6.0

7.3 x 1017

5.1 x 1017

1.1 x io18

9.5 x

io16

1.7 x 1017

1.1:l

1.:2.3

1.:6:1

106Ru

137Ru

144~e

14ke

Total atoms per square centimeter

Second: third

10'6

1.7:l

1018

1.25:l

4.1~1,

1.351

core ratio

T a b l e 10.1

1. Comparison of Fission Product Concentration Profiles for T w o Samples Core D r i l l e d from MSRE Graphite Surveillance Specimen Y-7

"Nb

Very similar concentration profile i n both cores and found in every sample t o midplane from all surfaces

952,

Profile curves similar in shape for both cores. with slightly steeper slopes for third core curves. N o detectable amount beyond 150 m i l s from first surface of third core; found all the way from first surface of second core. No detectable amount beyond 4 mils from second surface of third core; some quantity at 100 mils from second surface of second core

lo3Ru

'06Ru

Second core: found to persist to midplane (233 mils) from first surface and to depth of 25 mils from opposite surface. Third core: much steeper drop; 7 x l O I 3 atoms/cc at 5 mils from first surface v s 1.5 x lo'' for 5-mil depth in second core. Disappeared a t 10 m i l s from both surfaces of third core Similar behavior in both cores: found to about midplane from first surface of both cores and no deeper than 15 to 20 m i l s from opposite surface of both cores

l37CS

Similar concentration profiles for both cores, with slightly higher values for second core from 10 mils to midplane (233 mils)

141ce

From 1 0 mils t o midplane of both cores, concentrations are quite similar; however, for third core there is an unexplained flatness (and even lack of detection in some samples) from 10 mils t o the surfaces

1 4 4 ~ e Second core: found in every sample, first surface to midplane (233 mils), with increase i n concentration from 100 mils t o midplane; not found deeper than 15 m i l s from opposite surface. Third core; not de-

tectable below 15 to 20 mils from either surface

For 'Li the high neutron yield and relatively low threshold of the 'Li(p,n) reaction (1.881 M e V ) offer a unique identification with high sensitivity. The 9 Be(p,n) threshold occurs a t 2.059 M e V , and the 12CJ 13CJand "F thresholds still higher. Protons of 2.06 Mev penetrating the surface of graphite lose energy by Coulomb interaction with the electrons. They reach the 7Li threshold in 8 x

c m (-0.33 mil) and produce no 'Li(p,n) reaction neutrons at greater depths. The major neutron background (mostly from cosmic rays) can be conveniently determined by a companion bombardment at 1.881 MeV. The nuclear reaction gF(p,a)'60*(y)160 gives a 6.14-Mev gamma ray with a very high yield. The energy is large compared with gamma rays from

123 fission products, The reaction has recently been used to study fluorine penetration of Zircaloy by Starfelt e t al. A survey of other proton-induced reactions in graphite and in the molten salt used for reactor fuel indicated that fluorine penetration in the graphite moderator of the MSRE might also be determined to a very high degree of sensitivity. Prominent proton resonances in the * g F ( p , ~ y ) reaction occur near 0.5, 1.5, and 2 M e V . Protons of such energies would penetrate graphite to a depth of 0.001 to 0.005 c m (a few tenths to 2 mils). For smaller depths a detailed study of the gammaray yield with varying proton energy can be analyzed to give the concentration a s a function of depth, The chief background expected is from 13C(p,y) in the graphite, giving gamma rays near 9 M e V . The incomplete absorption of these in the spectrometer gives a flat background near 6 MeV. Protons at 1.88 Mev from the ORNL 3-Mv Van de Graaff were collimated to a 0.32-cmdiam spot. Steady currents from to 1pa were collected at the sample and integrated to monitor the number of protons in each run. A 7.5 x 7.5 c m NaI(T1) scintillation counter was placed 4 to 225 c m from the sample to record the gamma-ray spectrum. A long counter23 was used to measure neutron yield and was placed straight ahead from the proton beam at a distance of from 11 to 35 c m from the sample. A sketch of the apparatus is shown in Fig. 10.12. Thin standards used were 180 pg/cm2 and about 14 pg/cm2 of L i F evaporated on aluminum. A thick 7LiF crystal was used a s a standard also, after evaporating a thin conductive layer of silver on three of its faces to avoid electrostatic probl e m s from the proton charge buildup. Possible bias in our calibration for fluorine concentrations is estimated a t 10%to allow for the effect of anisotropic emission of gamma rays near 90'. For lithium concentration the uncertainty is 20%, largely because we could not use 1/r2 scaling to greatly reduce the dead-time corrections (due to high count rate) in counting the standard. In the c a s e of the MSRE sample (Y-7) mounted on an iron cylinder for lapping, an additional large correction for neutron transmission through the iron raises the possible bias for lithium to 25%.

'/2

22E. Moller, L. Nilsson, and N. Starfelt, Nucl. Znetr. Methods 50, 270 (1967). 23A. 0. Hansen and J. L. McKibben, Phys. Rev. 72, 673 (1947).

ORNL-DWG 68-2540

GRAPHITE SAMPLE

1 TO AMPLIFIER AND MULTICHANNEL ANALYZER FOR GAMMA SPECTRA

Fig. 10.12.

AND SCALER FOR . NEUTRON COUNT

P l a n V i e w Diagram of the Experiment.

Protons from the

ORNL 3.Mv Van de Graaff accelerator

were focused onto the surface o f the sample i n vacuum. F a c i l i t i e s for measuring the charge, suppressing electron emission, etc., are not s'hown. T h e sample was viewed by a gamma-ray spectrometer near 90'

to the

beam and by a neutron counter near 0'.

Figure 10.13 shows the gamma spectrum obtained using the 7LiF target, Also shown is the spectrum from a sample of beryllium metal. The gamma-ray yield from the beryllium is about 1/300 of that from fluorine and occurs mostly at a higher energy (7.48 Mev). Three samples of graphite were studied. A piece of clean graphite gave the background spectrum indicated in Fig. 10.14. A comparison with the standards indicated that a fluorine concentration of about 1 ppm could be detected in a 10-min run. The lithium detection l i m i t with the neutron counter used was about 0.1 ppm. A sample of graphite (Y-5, grade CGB, bar 635) exposed to nonradioactive molten salt for nine months was studied, a s well as one from the MSRE (Y-7, withdrawn in May 1967). Successive surface layers of these samples were ground (lapped) away to measure the fluorine concentration a t successive, deeper levels; since the back face of sample Y-7 had also been exposed to the molten salt, concentrations were studied a s this face was approached from the interior of the sample. The removed material is to be analyzed for fission products and uranium.

124

ORNL-DWG 68-2544

+O.oao

5000

2000

1000

500

z 0

200

W K

5 w

100

> W

W -I

u I

Fig. 10.13.

100

200 300 400 500 PULSE HEIGHT CHANNEL NUMBER

Puise-Height Spectra i n the

600

7.5- by 7.5-

cm N a l ( T I ) Spectrometer from Samples of 19F and 9Be T h e y i e l d from 7Li 37) was negligible i n com-

for Equal Proton Bombardment. above 0.478 Mev (channel parison.

The sample exposed to nonradioactive MSRE fuel salt (Y-5) showed surface contamination of 1.1 pg/cm2. This was removed by a 5 x cm lapping. Since the proton beam is capable of evaporating or sputtering surface layers , the sample was bombarded for more than 2 hr with a l/;c(a beam before lapping. The fluorine contamination of the surface was decreased about 5%. Accidental surface contamination was a problem in handling these samples, since the 'LiF standards used in the same apparatus had up to a million t i m e s higher concentration. Discrimination between thin surface layers and material distributed in depth was accomplished by varying the proton

energy. Figure 10.15 shows the concentrations of lithium and fluorine found a t depth in sample Y-5. The initial very high lithium value near the surface did not persist in the further measurements and is not considered typical of the bulk graphite. It may represent material originally in a small fissure about 8 x c m deep. The MSRE moderator sample (Y-7) showed about 3 r/hr (surface) fission product activity , requiring special handling. A l.&cm lead filter was used in front of the spectrometer to reduce the overall counting rate from the radioactivity. Surface fluorine measured about 3.7 pg/cm2. Concentration as a function of depth is shown in Fig. 10.16. To determine whether this fluorine concentration was due to salt penetration of a crack fortuitously crossing the sample, a microscopic examination and an acetone test were perfonned. Neither test showed any evidence of a crack. (Incipient cracks have normally led to fracture during the coring procedure used in cutting the sample from a larger piece of graphite. Sample Y-7 showed no fractures in the section we studied.) The 'Li concentrations found are summarized in Fig. 10.17. The lithium analysis was developed during the course of the experiments, so data are not available very close to the first surface studied. The sample exposed a t the center of the reactor for nine months (Y-7) shows over 100 t i m e s the lithium and fluorine content found in the control sample (Y-5) given identical treatment except for irradiation. Since graphite is expected to withstand up to 20 t i m e s heavier irradiations before replacement for structural reasons, it will be important to study penetration after longer reactor exposures. In the unirradiated sample the ratio of lithium to fluorine (-0.43) is probably consistent with that expected for the 'LiF molecule (7/19) in view of possible errors in normalization. The ratio expected from the fuel salt composition is The nonstoichiometric ratios found (particularly in the irradiated sample) seem suggestive of ionic diffusion. The fluorine concentration in the irradiated sample (Y-7, Fig. 10.16) follows the inverse of the depth from the front or back face fairly closely, although the anomaly near 0.008 c m (3 mils) from the rear face seems real. (The anomaly is also seen in the lithium data, Fig. 10.17, and may correspond to an inhomogeneity in the graphite.)

125

ORNL-DW 68-2542

ENERGY (MeV)

1 140 1

4 0 ~ ~

1 1 240 1 190

I 1

290

' 1 340

1 1 390

I l l '490 ' 440

540

590

CHANNEL NUMBER

Fig. 10.14.

Pulse-Height Spectra from Samples of Clean Graphite (Dashed L i n e s ) and Graphite Containing a

T m c e of Fluorine. spectrum near

The peaks a t 2.367, 3.150, and 3.150

The lithium concentration in the irradiated sample is not proportional to the fluorine, rising from the 7/19 ratio typical of 'LiF near the surface to 6/1 at the center. The minimum lithium concentration appears to be nearer to the first (outside) face of the sample than to the center. Only the molten-salt flow velocity was reportedly much different at the two faces.

10.4 SURFACE PHENOMENA IN MOLTEN SALTS H. W. Kohn

- 0.511

Mev are attributable to 12C(p,y) reactions, the

Mev to '3C(p,y).

F. F. Blankenship

The possibility that colloids o r metallic sols are involved in the corrosion product and fission product behavior i n the MSRE fuel h a s led to a study of colloids in molten salts. When the MSRE fuel was initially treated with hydrogen and beryllium during purification, sols as well as larger particles of reduced nickel and iron were evidently produced. Some o f these have remained suspended throughout operation of the reactor and presumably constitute the iron and nickel content regularly

reported in fuel analyses. Also, since beryllium treatments in the reactor did not decrease the chromium concentration, part of the chromium in the fuel is presumably present as a metallic sol. In general, the sols would b e expected to exist as alloys rather than as pure metals. The more noble of the fission product metals, because of the reduction potential reflected by the UF, content, are expected to exist in the MSRE as elemental metals. Most of such elements plate out on walls as metal (or possibly as carbides on graphite, in the case of molybdenum and niobium). (An appreciable portion also appears to leave with the off-gas.) However, to the extent that they do remain in the fuel, they are probably present as sols. Again these sols are undoubtedly alloys, and some may contain little or no structural metal. There is evidence (see Sect. 10.1) that sols containing fission products leave the melt, but a corresponding loss of structural metals has Cot been noted. As to the fate of these sols, we have obtained evidence from laboratory experiments that there

126 ORNL-DWG 68-25435

200

o FLUORINE

dRNL-LWG 6 5 2 5 4 4 1

1000

5M)

100

200 I-

'00

3 5 0

I-

I (3 W

& z

10 5

2 I

5.0

51.0

5 1 0 0 2

5 ~ x x )

DEPTH (mils)

IO-^

5 to-'

roo

DEPTH (em)

2.0

z W

1.0

Fig. 10.16.

Fluorine Concentrations i n Graphite

Sample Y-7, Exposed to Molten F u e l Salt in the MSRE

0.5

Reactor for Nine Months. Measurements were made as the sample was ground away in layers progressing from

0.2

the interior toward the second surface (closed triangles).

the f i r s t surface to the center (open circ1es)and then from The distances shown are as measured from the nearest surface exposed t o molten salts.

0.1

I 0

0.005

Fig. 10.15.

4 6 DEPTH (mils) I

0.04 0.045 DEPTH (c m 1

0.02

Lithium and Fluorine Concentrations in a

Sample (Y-5) of Unirradiated Graphite Exposed t o Molten Fluorides for Nine Months. The i n i t i a l high lithium value near the surface did not persist i n further measurements.

is marked tendency for flotation to occur. We studied samples from a fuel solvent m e l t to which large amounts of structural metal fluorides had been added and then reduced with hydrogen and zirconium. On melting a coarsely ground portion of this material in glass, a dark scum accumulated on the surface of the quiescent melt. When the scum was removed and discarded, the remaining salt had a decreased structural metal content. The results are shown in Table 10.12.

This suggested that the black s c u m was finely divided metal that had somehow floated to the surface. Two subsequent trials with salt from the same source were performed with helium bubbling through the m e l t in a test of flotation. The scum, shown in Fig. 10.18, formed in greater abundance, and the evidence for flotation was as shown in Table 10.13.

Table 10.12.

Structural Metal Content Before and

After Removal of Scum from Melt

Makeup Analysis

After Treatment

(PPd

bPm)

286

158

445

4680

218

127

500

ORNL-DWG 68-254517 10

MEASURING INWARD FROM FIRST SURFACE

200 i

100

+ P

> 20

! n i

;

2 I

1.0

l 0.005

Fig. 10.17.

aor a02

2 5 DEPTH 6nils) I

100 1

0.05 0.4 0.2 DEPTH (ern)

2 I

0.5

1000

4.0

Lithium Concentrations in the Graphite

Sample (Y-7) Exposed Nine Months i n the MSRE Reactor. T h e significance o f the different symbols i s as described for Fig. 10.16.

T h e distribution does not appear t o b e

symmetrical, and a nonuniformity near 0.008 cm from the second surface i s suggested.

T h e fluorine measurement

(Fig. 10.16) indicates a similar minimum.

There was a possibility that the black material included finely divided Zr02, the oxide that precipitates from fuel solvent. Since there was no significant segregation of zirconium, we concluded that flotation had removed predominantly metals. The conditions for the formation of a similar scum a t the liquid surface of the pump bowl in the MSRE are good. However, attempts to detect it directly have been inconclusively unsuccessful. Most of the material collected in the laboratory experiments on flotation seems t o be larger than colloidal in size (>0.1 p), but the sol particles are probably behaving in the same manner as the larger particles. There is evidence that s o l particles escape from radioactive MSRE fuel to form an aerosol (see Sect. 10.1). The mechanism for this implausible behavior remains unknown. One suggestion is that because the gas above radioactive fuel is conducting, a contact potential between the liquid salt and gas phases impels charged particles from

Fig. 10.18.

Flotation i n F u e l Solvent.

the interface. Another suggestion is that recoil is responsible. Bursting bubbles may have an effect, but the results of hot-cell t e s t s (see

Sect. 10.1) with quiescent fuel from the MSRE suggest that bubbles may not be necessary. The whole question of the formation and stability of colloids in molten s a l t s has been virtually untouched in the literature. Accordingly, experiments in nitrate melts have been initiated (because of the convenience of working with nitrates) to characterize such colloidal behavior.

128 Table 10.13.

Sagregotion of Reduced Structural Metals

by Flotation

Analyses in weight percent ~

0.147 0.008

0.012 <0.004

0.31 <0.006

9.79 10.2

0.074 0.013 0.010

0.030 <0.009 <0.003

0.42 <0.2 0.12

9.19 10.1 11.3

Second melt

TOP Bottom Third melt

TOP Middle Bottom

Colloidal dispersions of silver metal in molten potassium-sodium nitrate eutectic have been prepared by uv photolysis. Colloidal gold has been prepared in sodium-lithium and sodium-potassium nitrates by reduction of gold chloride and also by supersonic dispersion of particulate gold from

exploded wires. Silver iodide in potassiumlithium nitrate m e l t s has also been prepared. Transmission and scattering spectra are being measured. Another surface-related behavior that has been studied is foam formation. Foam formation in nitrate melts was induced b$ adding aluminum nitrate nonahydrate. The foam was first formed by bubbles of nitrogen oxide from the decomposition of aluminum nitrate. When helium was bubbled through the melt, the nitric oxide foam gave way to a helium bubble foam which, i f undisturbed, lasted for about 15 min. Similiar behavior was encountered some t i m e ago on an addition of thorium fluoride to a nitrate-fluoride melt.24 In both cases we could hypothesize that the foam was stabilized by finely divided oxide. In view of the rare occurrence of molten-salt foams, we have conjectured that a finely divided wetted solid may be a requisite for stable foam formation in molten salts. 24R.A. Strehlow, private communication.

11. Chemistry of Fission Product Fluorides 11.1 PROP ERTl ES OF MOLYBDENUM FLUORIDES C. F. Weaver H. A. Friedman D. N. Hess The synthesis and characterization of molybdenum fluorides and a study of their reactions in molten 2LiF.BeF2 1 , 2 were continued. The synthesis of ruthenium fluorides was initiated.

11.1.1 Synthesis of MoF, and MoF, The first s t e p in the synthesis involved forming a solution of MoF, and MoF, by reacting molybdenum metal with an excess of MoF, refluxed in a Pyrex container such a s that shown at the right in Fig.

I -

11.1. Commercial MoF, was purified with respect to HF before use by distilling it over N a F in an apparatus similar to that shown at the left in Fig. 11.1. Hydrogen fluoride is known to catalyze the reaction 2MoF, + 3Si0, + SSiF, + MOO,;, failure to remove HF resulted in several explosive ruptures of the Pyrex equipment, while, in the absence of HF, the MoF, was contained for weeks without an observable pressure increase. As much of the apparatus a s possible was flamed under vacuum before use, because these fluorides are very reactive toward moisture. After the molybdenum metal was completely reacted with the excess MoF,, the system was evacuated and t h e temperature slowly increased. During this procedure the excess MoF,

left the system, leaving a residue of MoF,. Upon further heating, part of the MoF, sublimed to the neck of the Pyrex flasks, and part disproportionated, producing MoF, and MoF, (which then left the system). Experiments in which the terminal t e m peratures were 2000 and 150° both produced MoF, as a tan residue in the bulb and MoF, as a bright yellow deposit in the neck of the flask. The ratio of MoF, to MoF, was higher at 2OOOthan at 1509 An experiment which was heated to only 1OOOproduced the MoF, in the neck of the container but left a viscous residue with a molybdenum-to-fluorine ratio slightly higher than MoF,. This residue did not solidify upon cooling to room temperature and storing for several days. The MoF, and MoF, were both identified by x-ray diffraction and by chemical analysis. Their diffraction lines are listed in Table 11.1. Those for MoF, were identical to those obtained by D. E. LaValle et in earlier synthesis and structure studies. The pattern for MoF, corresponds to that calculated6 from structural data in the literature. The analytical results’ confirming the stoichiometry a r e shown in Table 11.2. The impurity levels determined by emission spectroscopy * are also shown in Table 11.2. The optical properties of these substances and of NbF,’ were determined to provide a rapid and convenient means d . 4 9 5

D. E. LaValle e t al., J . Am. Chem. SOC.82, 243334 (1960). 5Persona1 communication, R. M. Steele, Metals and Ceramics Division. ‘Pattern calculated by G. D. Brunton, Reactor Chemistry Division. ‘Analyses performed by E. C. Lynn and Dave Canada, Analytical Chemistry Division. ‘Analyses performed by J. A. Carter et al., Analytical Chemistry Division. ’Sample prepared by L. M. Toth by vacuum distillation of impure NbF,, Reactor Chemistry Division.

‘C. F. Weaver, H. A. Friedman, and D. N. Hess, “Behavior of Molybdenum Fluorides,’’ Reactor Chem. Div. Ann. Progr. Rept. Dec. 31, 1967, ORNL-4229. 2C. F. Weaver and H. A. Friedman, “Behavior of Molybdenum Fluorides,” MSR Program Semiann. Progr. Rept. Aug. 31, 1967, ORNL-4191, pp. 142-44. ,A. J. Edwards, R. D. Peacock, and R. W. H. Small, J . Chem. SOC.1962, pp. 4486-91.

129

130 ORNL-DWG 67-11720

Fig. 11.1.

Apparatus for the Synthesis of Molybdenum Fluorides.

for future identification; they are summarized in Table 11.3. The absorption spectrum of MoF, is being investigated. Preliminary results show absorption in the region of 1260 nm and an intense ultraviolet shoulder near 340 nm.

nI'' cooperation with Jack Young, Analytical Chemistry Division.

11.1.2 Lithium Fluoromolybdates(lll) The relative stability of MoF, at temperatures of reactor operation and the fact that it can produce both molybdenum metal and volatile molybdenum products upon disproportionation suggested that behavior of Mo3 in MSRE-type solvents should be investigated. Examination of quenched melts along the join 2LiF.BeF ,-MoF, have disclosed two previously unknown compounds shown to be-

131

T a b l e 11.1.

X-Ray Diffraction L i n e s e

MoF, Powder Pattern

2e 23.3 31.8 34.4 39.3 41.9 47.5 51.7 54.9 55.3 60 61.6 66 68.2 72.6 74.3 77.3 81.3

1.1

2.5 1.4

Analyses of the Molybdenum Fluorides

MoF,

MoF5 Debye-Sherrer

looz/zo 100 25 13 5.4 8.2 8.1 12 3.2 6.5 4.7 4.4 2.1 1.9 2.7

T a b l e 11.2.

2e 11.10 16;73 19.65 20.64 22.31 24.31 25.11 26.53 27.49 28.61 31.11 32.64 33.81 37.22 39.55 42.51 45.37 47.35 48.77 51.51 53.37

Intensity W

m W

m S

vw vw ms

vw vw vw W

vw W W

Wt 70 Mo Mo (calcd) F F (calcd) Ppm AI B Be Ca cu Li

MoF5

49.6 50.2 49.9 49.8

62.7 62.7 37.9 37.3 <200 -

<loo 50 500 50 30 300 300 <500 -

Mg Mn Na Si W

500 10 <500 3000 100 <loo 500 200

vw ms

vw vw vw vw

T a b l e 11.3.

Optical Properties of the Molybdenum

Fluorides and Niobium Pentafluoride

MoF,

Uniaxial positive

N u = 1.592 N, = 1.624

%opper target.

Golden yellow MoF5

long t o the system LiF-MoF,. Their optical properties and x-ray diffraction patterns have been determined and are given in Tables 11.4 and 11.5 respectively. Attempts to determine the stoichiometry and melting behavior of these phases have been complicated by corrosion of container materials such as nickel and copper. However, the formulas are certainly close t o 5LiF.2MoF3 and LiF-MoF,, and the compounds have been so labeled in Tables 11.4 and 11.5. These efforts are being continued with the following goals in mind:

1. Definite determination of the formulas and

melting behavior of these compounds. 2. Determination of the liquidus values and primary phases along the join 2LiF.BeF2MoF 3. 3. Analogous studies of the behavior of niobium, ruthenium, and technetium.

Biaxial Optical angle = goo

N,= 1.520 N p = 1.534 N Y = 1.548 Lemon yellow NbF5

Biaxial negative Optical angle = SOo N, = 1.484 N y = 1.516

11.1.3 Kinetics of MoF, Behavior in 2LiF.BeF2 Previously reported work2 indicated that MoF may be stable for many days or disproportionated to molybdenum methl and MoF, in a few hours in the temperature range 500 to 700째 depending on the temperature and pressure. It was further shown

132 Table 11.4.

LiF .MoF

Optical Properties of the Lithium Fluoromolybdates

Uniaxial positive

Nu= 1.512 N, = 1.524

5LiF-2MoF

Biaxial negative

N,= 1.472 N y = 1.490 Optical angle = 60’

Table 11.5. X-Ray Diffraction Powder Patterns

5LiF.2MoF3 2e

20.2 21.4 25.9 32.9 35.2 35.8 40.1 40.9 41.6 43.5 44.7 48.7 51.1 52.8 55.1 57.1 61.1 63.1 68.9

LiF-MoF,

lOOZ/Z,

lOOZ/I,

2e 19.3 21.3 26.9 30.1 33.3 35.0 38.2

100 72 34 33 6 11 21 10 12 23 3 7 9 18 7 4 6 4 6

29 100 37

MSRE and that its kinetic behavior should be quantitatively investigated in the ppm range. Initial conditions were selected to provide a simple reference system which could be made progressively more complex and more analogous to the MSRE. The first experiment was made with a copper container, a 2LiF.BeF2 melt, and a helium flow rate of 8 liters/hr per kilogram of melt. The MoF, was added to the molten 2LiF. BeF, in the form of a previously fused mixture of LiF-MoF, (75-25 mole So). Filtered samples were taken over a period of several hours, a s shown in Table 11.6 under “first run.” The three for molybdenum gave estypes of sentially the same answers. The wet method provided the least scatter and indicated that, after an initial mixing period, the concentration of molybdenum in the melt did not change over a 23-hr period and that there was no difference between the filtered and unfiltered samples. The second experiment was conducted in the same way except that after 2 hr at 500° the temperature w a s

27 11 4 12 10 19 12 16

40.0

42.3 44.3 44.6 52.9

‘Analyses performed by E. I. Wyatt et al., Analytical Chemistry Division.

ORNL-OWG 68-5542

that the corrosion of the container material by MoF, was much greater if molten 2LiF.BeF2 was present and that copper was much superior to nickel a s a container material. The reaction MoF, (1 mole %)

+ 3UF,

(4 mole %)

-,Mo + 3UF, 0

was observed in molten 2LiF.BeF2 at 500OC. These observations suggest that Mo3 plays a central role in the behavior of molybdenum in the

’

80 100 TIME (hrl

120

140

(60 /--

Fig. 11.2.

Removal of Mo3+ from Molten 2LiF-BeF2.

133

Table 11.6. Kinetic Behovior of Mo3' in 2LiF*BeFp

Temperature of Sample PC) !

Type

Length of Molybdenum Time After by Wet Addition of Molybdenum Salt @Pm) (hr)

Molybdenum by Spectroscopic Analysis

Molybdenum by Activation Analysis

Copper by Wet Chemistry

Copper by Spectroscopic Analysis

(PPd

(ppd

bpm)

3 00 300 700 700 500

638,621 845,896 770,908 836,819 752,743 846,839

F i r s t Run

500

Filtered Filtered Filtered Filtered Filtered Unfiltered

0.15

620 810 820 810 810 800

2 6 23 23

500

500 30 500 100 100 100

Second Run

500 500-700 700 700 700 700 700a 700a

Filtered Filtered Filtered Filtered Filtered Unfiltered Filtered Unfiltered

690 510 440 390 190 250 <10 20

2 2.5 3.5 7 24 24 27 27

113 64 183 50 208 81 41 396

700 400 300 300 100 200 < 100

<loo

100 70 200 50 100 70 70 500

aReduced molybdenum in melt with H2.

raised to 7009 In this case there was a definite decrease in the molybdenum concentration in the filtered samples. The unfiltered samples contained m r e molybdenum than t h e filtered ones, suggesting that metallic molybdenum was present and that it was being excluded by the filter. To further confirm that the metallic molybdenum was not passing the filter, the sample was reduced with an excess of hydrogen. N o molybdenum was detected in the filtered reduced sample but was found easily in the unfiltered sample, suggesting that part of the metallic molybdenum was suspended in t h e m e l t but not passed by the filter. The copper analyses do not appear to have a trend with time. The scatter of the wet copper analyses was greater than expected from the analytical procedure and is probably an artifact of the sampling technique, which involves a copper filter. Unfiltered samples of 2LiF-BeF2 taken in a nickel tube with and without Mo3 were not +

found to contain copper. The absence of a copper concentration increase with time and the level of concentration both indicate that the decrease in molybdenum concentration was not caused by MoF, reduction by copper. The effect of the helium flow rate (MoF, removed rate) on the kinetic behavior of Mo3' in molten 2LiF.BeF2 was investigated in runs 3 and 4. The data in Table 11.7 and Fig. 11.2 indicate that the molybdenum was removed from t h e 2LiF.BeF2 by a second-order process, the most likely event being

2MoF,

+ MoF,

The rate constant k in the relation

k=(;-<)-

1 t

134 Table 11.7.

Temperature of Sample

Kinetic Behavior of Mo3+ In 2LiF*BeF2

Length of Time After Temperature Is Raised from 5O@C (hr)

PC,

Molybdenum by Wet Chemistry

effect, the back reaction was involved significantly in the kinetic behavior of Mo3+, b is the order of df or larger, and k,/k, 2 f , / f , , consistent with the experimental values

(PPd

Third Run

1.2 x lo-, hr-' ppm-' = 10 1.2 x lo-, hr-' ppm-'

500

806

690

663

700

570

700

4 t

465

700

240

700

140

11.2 are associated with a lower temperature,

700

112

700

140

since the heatup from 500 to 7OOOC was involved. However, since the rates were faster at the lower temperatures and since the rates are known to increase with temperature, the short-term effect m u s t have another explanation. Apparently this was the period necessary to establish the steadystate condition. These kinetic experiments will be continued with the following objectives:

Fourth Run

158

500

700

137

700

19i

700

\ 67 i

7 00

16 14

(c = concentration of Mo3 3 has the value 1.2 x lo-, ppm-' hr-' for a helium flow rate of liter/hr at 700Oand 1.2 x lo-, ppm-' hr-' for 12 liters/hr at 700OC. If one makes the steady-state assumptions for reaction (1) that the concentration of MoF, was constant and very low compared with the concentration of Mo3+,then R takes the form a f / ( b+ df), where a is the rate constant for the forward reaction, b is the rate constant for the back reaction, d is a constant, and f is the flow rate. If b << df, that is, if the back reaction is unimportant, then the rate expression reduces to dc a = --c2,

for

12 l i t e d h r f , 0.75 liter/hr

16.

The first two points (at 0 and 0.75 hr) in Fig.

1. To determine effects of UF,, UF,, ZrF,, and ThF, additions as well as INOR-8and graphite on the rate and extent of removal of molybdenum from the melt. 2. To identify and determine the fate of the volatile species. 3. To obtain the corresponding information for niobium, technetium, and ruthenium.

11.2 MASS SPECTROMETRY OF THE MOLYBDENUM FLUORIDES

R. A. Strehlow

J. D. Redman

The mass spectrometric study of molybdenum fluoride vaporization behavior was continued. Previous work' was conducted with a conventional Knudsen cell suitable for the study of low-vaporpressure solids. This cell was not suitable for study of gaseous species at elevated temperature. Accordingly a cell (made of nickel) was constructed which permitted addition of gases directly into the cell region. Using the new cell, the mass spectrum due to admitted MoF, was obtained at

which suggests that the flow rate is unimportant. Since, experimentally, the flow rate had a great

12MSR Program Semiann. Progr. Rept. A@. 31, 1967, ORNL4 191.

135 11.8. Mass Spectrometric Cracking Patterns for Molybdenum Fluorides and Oxyfluorides

Table ~~~

MoF, Ion

Intensity

Ion

Intensity

Ion

MoF +

100

MoF5

MoF3

+ MoF,

+ MoF4 +

+ MoF3

MoF2+

MoF, +

M ~ +F

MoF+ Mo+

MoF3

Mo+

MOOF,~

MoF,

MoF5

Intensity

Ion

100

MoOF3

MoF

M~O,F

M ~ F

+ MoOFz

lo+

+ MoOF,

Mo+

MoFz+

M~F,,+

M~OF

MoOF

M ~ F

MOO+

3.5

MOO+

+ Mo

100

MoF

+ +

Mo+

amis assignment is not certain; MoOF,

Mo02F2

Intensity

Ion

Intensity

100

Mo02F2

100

is possible.

various temperatures from 30 to 85OOC. The cracking pattern a t 3OoC differed only slightly from that obtained earlier and is reported in Table 11.8. With a steady rate of M o F , admission t o the cell containing molybdenum wire, an increase in temperature resulted first in production of M o F , and, at higher temperatures, increasing amounts of MoF,. This behavior paralleled that of MoF, (which was studied using the earlier cell design) except for the presence of dimeric impurity in the M o F , sample. The dimer, which has a principal peak family corresponding to M O , F ~ + , may have as a neutral precursor MO,F,O

or Mo,FlOo. The mass spectroscopic evidence alone is not adequate to distinguish between these possibilities. The rather striking similarity of results from the M o F , and the M o F , studies is shown in Fig. 11.3, which displays the derived percentages of the species M o F , , MoF,, and M o F , as a function of temperature for both the tri- and liexafluoride. The data for M o F , exclude the dimer. The difference in free energy of formation between M o F , and M o F , is clearly very slight, since their relative amount varies only slowly with temperature. The earlier conclusion that kinetic factors dominate the descriptive chemistry of these materials is still valid, although we have obtained data indicating

ORNL-DWG 68-5543

SOURCE MATERIAL SPECIES MoF3 MoG MoF5

Moh

MOF6

IO0 90

-L 70 0

c v)

50 0

3 30

IO 0 0

400

Fig. 11.3.

200

300 400 500 600 700 KNUDSEN CELL TEMPERATURE PC)

800

Monomeric Vapor Species over MoF6 vs

C e l l Temperature.

900

136 that some thermochemical relations can be derived for the Mo-F system. The dimer Mo2F101 wasnot the only dinuclear fluoride observed. During admission of MoF, an erosion of the tantalum heat shields in the Knudsen cell assembly occurred and was evidenced by the presence of MoTaF9+ions i n the spectrum as well as by subsequent examination of the tantalum elements Continued work on the molybdenum fluoride system will include a more complete treatment of the data obtained so far and consideration of the approach needed to obtain the desired thermochemical and kinetic data for this system. Some other fission product fluoride and oxyfluoride systems will also be examined.

11.3 SPECTROSCOPIC STUDIES OF FISSION PRODUCT FLUORIDES L. M. Toth J. P. Young G. P. Smith Because spectrophotometry offers a direct measurement of fission products such as the lowervalent niobium and molybdenum fluorides, a study of their spectra has been undertaken i n the windowless cell developed by J- P. Y 0 ~ n g . I T ~ h i s cell has the advantage that the melt need only be exposed to a container which has been selected on the basis of the melt's reactivity with it. For this reason graphite and copper windowless cells have been selected for use with MoF and NbF, solutes in LiF-BeF2 (66-34 mole %). 315 The original intent was to compare the spectra of NbF, and MoF, with the ligand-field spectra of d' and d 3 configurations, respectively, in octahedral symmetry. Initial attempts to dissolve 0.1 wt % concentrations of MoF, in L 2 B l 6 failed due to the formation of a finely divided precipitate which required 16 t o 24 hr to settle out. After it had settled, no spectrum was evident in the L2B. The cause of

I3J.

P. Young, Anal. Chem. 36, 390 (1964).

14Supplied by C. F . Weaver and H. A. Friedman "Henceforth abbreviated a s "L2B." '6This L2B was from a stancfard HF-H2 treated batch of J. H. Shaffer's, Reactor Chemistry Division. "Reactor Chemistry Division.

the precipitate has been attributed to the formation of insoluble molybdenum oxides or oxyfluorides. Successful dissolution of MoF, in L 2 B has been achieved by first treating the solvent with SiF, and filtering it through a quartz frit. The spectra of such solutions display peaks at 350 and 470 mp which are not inconsistent with those expected for a d 3 cation in octahedral symmetry. To verify that this spectrum is not due to contributions from SiF,, it was compared with one produced by a sample taken from the molybdenum fluoride kinetic studies of C. F. Weaver and H. A. Friedman." Although the latter sample yielded only a very weak spectrum, similar absorbancies at 350 and 470 mp could be identified. The poor intensity of this spectrum has been tentatively attributed to the fact that much MoF, had precipitated out due to oxide contamination of the sample. Experiments are currently under way to demonstrate this further as well as to establish a quantitative value for the absorption coefficients a t 350 and 470 mp. With these values, l i m i t s of spectroscopic detectability will be available. Further experiments will b e made in an attempt to demonstrate conclusively that the spectrum is due to dissolved MoF, in octahedral symmetry. Niobium tetrafluoride is apparently sparingly soluble in L2B. Presently, only a very weak spectrum has been obtained which displays a single absorbance at 350 mp. This spectrum is, however, similar t o that produced by NbF, in molten NbF,. Further work shall endeavor to circumvent the solvation problems by generation of NbF, in solution. Additional information available from the spectroscopic studies includes the compatibility of the solutions'with the container materials. In reactor-grade graphite, the spectrum attributed to MoF, disappears completely within 4 hr at 65OOC according to first-order kinetics. No intermediate spectra have been seen during this reaction. On the basis that the same solution shows no perceptible change under similar conditions i n a hydrogenfired copper container, reaction with the graphite is suggested. Further experiments will describe the stability of the solutions in more detail.

"L. M. Toth and G. P. Smith, Reactor Chem. Div. Ann. Progr. Rept. Dec. 31, 1967, ORNL-4229, p. 64.

137 11.4 PREPARATION OF NIOBIUM P ENTAFLUORIDE

H. A. Friedman L. M. Toth C. F. Weaver i

Because of the lack of information concerning the behavior of fission product fluorides in the MSRE, their chemistry in molten fluoride solutions is being actively pursued. One such study is concerned with niobium and its fluorides. " Niobium pentafluoride has been the most convenient starting point for the synthesis of lower niobium fluorides as well as a low-temperature It had solvent for the lower niobium fluorides. been previously prepared in a flow system at 25OOC by passing fluorine over niobium powder in a Monel reaction vessel, 2 o but this procedure would have taken several days reaction t i m e for the hundred-gram quantities of product required.

"L. M. Toth andG. P. Smith, Reacfor Chem. Div. Ann. Progr. Rept. Dec. 3 1 , 1967, ORNL-4229. 'OF. Fairbrother and W. C. Frith, J . Chem. SOC. 1951, p. 3051,

It was found in this work that decreasing the reaction temperature accelerated the reaction rate to the point that it could be completed within 8 hr. This is to b e expected for an exothermic reaction which involves a decrease in the entropy of the system: Nb + 5/2 F,

NbF,

The effect also provides a built-in safety device in which the reaction rate decreases abruptly if the temperature goes too high. The synthesis was performed in a two-chamber Monel reaction tube which was entirely sealed except for the F, inlet valve. The F, was admitted at a constant rate into the reaction chamber, containing powdered niobium metal at 70 to 100째C, and the NbF, sublimed into the other chamber, which was at room temperature. Niobium oxyfluorides present as impurities were then removed from the product by resublimation of the NbF, at 100'.

12. Physical Chemistry of Molten Salts 12.1 THERMODYNAMICS OF LiF-BeF, MELTS BY EMF MEASUREMENTS B. F. Hitch

C. F. Baes, Jr.

A potentiometric study' has been completed of the cell Beo I BeF,, L i F I H,, HF, Pt ,

This comparison of the emf data with the LiF-BeF, phase diagram also indicated that the heat of fusion for BeF, is <2.0 kcal/mole. A power series in xLiF was assumed for log yBeF,, and the coefficients were determined by a least-squares fit to the data. This gave (Fig. 12.2)

logY B e F 2 --6.8780-

-)xiiF

for which the cell reaction Beo(c)

+ 2HF(g)

BeF,(d)

+ H2(g)

was assumed. These measurements extended over the composition range 0.30 to 0.90 mole fraction BeF, and over temperatures in the range 500 to 900OC (Fig. 12.1). Since the cell potential E should be related to the activity of BeF, in the m e l t by

RT .=EO-,,.[

, H '

aBeF2

p;F

+ (-40.7375 + 94.3997

36292.8 T

)

XiiF

--84870.9 T Fi:.)

+ (-67.4178 + 52923.5 T

A Gibbs-Duhem integration of this expression gives for Y L i p

, log yLiF = 0.9384

activity coefficients could be derived for BeF, (and by a Gibbs-Duhem integration for LiF). Usefully accurate measurements could not be made with pure BeF, in the cell because of its very high viscosity; hence values of Eo were calculated using values for aBeF2derived from the phase diagram and a re-

--232T.OS +-

T T

ported heat of fusion2 of 1.13 kcal/mole. This gave, in volts,

X L F ,

113592.3 T

Eo = 2.4430 - 0.0007952T .

52923.5 'B. F. Hitch and C. F. Ebes, Jr.. An E.M.F. Study of LiF-BeF2 Solutions. ORNL-4257. ,A. R. Taylor and T. E. Gardner, Some T h e m 1 P r o p erties of Beryllium Fluoride from So to 12OO0K,U.S. Bureau of Mines Report RI-6644 (1965).

(The integration constant was determined from the L i F liquidus in the LiF-BeF, system.) Formation free energies and heats for BeF, and Be0 were also

138

. bi

139

ORNL-DWG

67-l3723R

1.95

1.90

1.85

1.80

1.75

1.70

1.65

1.60

1.55

150 0.80,0.90 BeF,

650

600 650 700

650 700 750

700 750 800

750 800 850

850 900 950

800 850 900

TEMPERATURE ("C)

Fig.

12.1.

Measured C e l l Potentials, Corrected to a Unit Value of the Pressure Quotient

of Temperature a t Various L i F - B e F 2 Melt Compositions:

Ec = E

+-23 In ( P H 2 / 4 F )

PH /PiF, 2

as a Function

140

ORNL-DWG 68- 231

1.0

0.8 Y

0.6 0

z a ,p0.4

0.2

0.I 0

0.2

0.1

0.3

0.5

0.4

0.7

0.6

0.8

0.9

1.0

XBSF,

Activity Coefficients of BeF, and LiF Derived from the EMF Measurements.

Fig. 12.2.

Table 12.1.

Compound

Temperature.

Formation Heats and Free Energies of BeFZ and B e 0 b&kcal/mole) Present Work

BeF,

298 900

Be0

bGt(kcal/rnole)

State

1000

Crystalline Liquid Liquid

-246.01

-242.54

298 900 1000

Crystalline Crystalline Crystalline

- 145.85 - 145.57(f1.5) - 145.46

-243.12(f1.1)

JANAF'

Present Work

-242.30(*2.0)

-234.39 2 1l.gO(fl.1) 208.47

JANAF'

-230.98(f2.0)

-143.10(f0.1)

-138.36 123.20(fl.S) 120.73

- 136.12(f0.1)

'Reference 3.

calculated (Table 12.1) by combining the results of the present study with available thermochemical data3 for H F and H,O and with the results of a previous study4 of the equilibrium

The Be2 I Beo and the HF,H, I F- electrodes used in this study performed acceptably for use a s reference electrodes, both being stable and reproducible.

+ BeO(c) e BeF,(d) + H,O(g) .

12.2 ELECTROLYSIS OF LiF-BeF, MIXTURES WITH A BISMUTH CATHODE

2HF(g)

3JANAF Thermochemical Tables, Clearing House for Federal Scientific and Technical Information, U.S. Department of Commerce, August 1965. 4A. L. Mathews and C. F. Baes, Jr.. J. Znorg. Chern. 7, 373 (1968).

K. A. Romberger

J. Braunstein

Experiments have been initiated on the electrolysis of LiF-BeF, mixtures with a bismuth cathode

141

in a silica cell. The purpose of these experiments is (1) t o develop lithium-bismuth alloy electrodes containing about 0.1 at. % lithium as reference electrodes for equilibrium and nonequilibrium electrochemical measurements in molten fluorides, (2) to ascertain the compatibility of lithium-bismuth alloys with silica containers, and (3) to determine whether reduced bismuth species are present in m e l t s contacted with lithium-bismuth alloys under conditions proposed for extraction processes for the single-region MSBR.5 (At very much higher lithium concentrations in bismuth, that is, with saturated alloys, there have been indications of the dissolution of reduced bismuth species in LiC1-LiF mixtures. 6 , In a preliminary experiment, 41 g (1.2 moles) of LiF-BeF, (66-34 mole %) which had previously been purified for use with the MSRE7 was electrolyzed, under a helium blanket, between a bismuth cathode (99.999%Bi, hydrogen fired a t 6OOOC to reduce any oxide) and a graphite anode isolated from the main portion of the m e l t in a silica J-tube. Forty-one grams of bismuth was used as the cathode, and the temperature was 473OC. Approximately 160 coulombs were passed (-1.7 x loF3 equivalent), enough t o produce about 0.8 at. % lithium-bismuth alloy if the current efficiency were 100%. The current density was 2 x amp/cm2. The silica cell permitted visual observation of the melt and electrodes during the electrolysis. From the inception of electrolysis, gas bubbles were formed a t the cathode (possibly hydrogen from the reduction of residual HF, which had been used in the purification of the salt,7 or traces of OH-). A black deposit was formed which partially covered the surface of the bismuth cathode. Chemical analysis of the phases has not been completed, but it is believed that the black deposit was metallic beryllium. Because of the low rate of diffusion of lithium from the surface into the bulk of the bismuth, the local activity of lithium a t the surface may have been high enough t o permit beryllium, which is insoluble in bismuth, to deposit on the surface.

J. H. Shaffer e t al., MSR Program Semiann. Progr. Rept. Aug. 31, 1967, ORNL-4191, p. 148. 6M. S. Foster e t al., J . Phys. Chem. 68, 980 (1964); M. S. Foster, Advan. Chem. Ser. 64, 136 (1967). 7J. H. Shaffer e t al., Reactor Chem. Div. Ann. Progr. Rept. Jan. 31, 1965, ORNL-3789, p. 99.

Samples of salt were withdrawn for analysis after passing about 40 coulombs and again after about 160 coulombs (equivalent to about 0.2 and 0.8 at %, respectively, lithium in bismuth). Preliminary analytical results showed that the bismuth content of the salt samples was below the limit of detection of 10 ppm. There was no visible attack of the silica by the m e t a l phase. Experiments are continuing, to evaluate the behavior of the lithiumbismuth electrodes and t o elucidate the electrochemical behavior of bismuth in molten fluorides.

12.3 REVIEW OF ELECTRICAL CONDUCTIVITIES IN MOLTEN FLUORIDE SYSTEMS G. D. Robbins

A review of electrical conductivity measurements in molten fluoride systems covering the period 1927 to 1967 has been made.E The results may be summarized as follows:

1. Because of the high specific conductance of most molten salts (1 t o 6 ohms-' cm-'),9 one of two experimental approaches' is usually employed: the use of capillary-containing cells, which results in a measured resistance of several hundred ohms, or the use of metallic cells in which the container is one electrode with a second electrode positioned in the melt. Measured resistances in the latter type of cells are less than 1 ohm. Boron nitride encased in graphite has been successfully employed in capillary construction,' while platinum, platinumrhodium (20%), Inconel, molybdenum, and graphite have been used for container and electrode materials. 2. The common practice of measuring resistance with a Wheatstone bridge having a parallel re-

'-'

*G. D. Robbins, Electrical Conductivity of Molten Fluorides, a Review, being reviewed for publication. '1. S. Yaffe and E. R. Van Artsdalen, J . Phys. Chem. 60, 1125 (1956). loG. J. Jam, C. Solomons, and H. J. Gardner, Chem. Rev. 58, 461 (1958). "E. A. Brown and B. Porter, Electrical Conductivity and Density of Molten Systems of Uranium Tetrafluoride and Thorium Fluoride with Alkali Fluorides, U.S. Dept. of Interior, Bureau of Mines, 128.23:6500 (1964). 12E.W. Yim and M. Feinleib, J - Electrochem. SOC. 104, 622 (1957). 131bid., 626 (1957).

142 sistance and capacitance in the balancing arm can result in considerable error if the relation for resistive and phase balance,

RP = Rs[l

+ R2C2 (2nf)'I , P P

is not employed in determining the solution resistance.8 In this expression RP and C are the parp. allel balancing resistance and capacitance, Rs is the solution resistance, and f is the measuring frequency. Use of this correction can b e avoided by employing a bridge with the balancing resistance and capacitance in series. 3. The practice of measuring resistance at a series of frequencies and extrapolating to infinite frequency as a function of f - ' I 2 is examined in terms of electrode process concepts. Some authors while others found no followed this appreciable variation of resistance with frequency (refs. 11-13, 17, 18). Unfortunately, the majority of investigators did not address themselves to this question. (Research on the frequency dispersion of electrical conductivity in molten salts is continuing in this laboratory.) 4. Table 12.2 lists some values of the specific conductance K and the equivalent conductance h e q of those alkali-metal and alkaline-earth fluorides which have been reported.11-13*15-17-23 These quantities are defined as K=

ORNL-MNG 68-5544

'1 (Z/a) , R

x E= 1

n 400

500

600 700 TEMPERATURE (TI

800

Fig. 12.3. Specific Conductances of Some MSRERelated Systems.

eq wt

hes = K - ,

P I4G. D. Robbins and J. Braunstein, Reactor Chem. Div. Ann. Progr. Rept. Dec. 31, 1967, ORNL-4229, p. 57. "5. D. Edwards et af.. J . Efectrochem. SOC.100, 508 (1953). 16J. D. Edwards e t a f . , ibid. 99, 527 (1952). 17J. E. Mackenzie, J. Phys. Chem. 32, 1150 (1960). "5. A. Mackenzie, Rev. Sci. Znstr. 27, 297 (1956). "1. S. Yaffe and E. R. Van Artsdalen, Chem. Div. Semiann. Progr. Rept. June 20, 1956, ORNL-2159, p. 77. 2oH. R. Bronstein and M. A. Bredig, Chem. Div. Ann. Progr. Rept. June 20, 1959, ORNL-2782, p. 59. 21H. R. Bronstein, A. S. Dworkin, andM. A. Bredig, Chem. Div. Ann. Progr. Rept. June 20, 1960, ORNL 2983, p. 65. 22T. Baak. Acta Chem. Scand. 8, 1727 (1954). 23H.Winterhager and L. Werner, Forschungsber. Wirtsch. Verkehrsministeriums Nonirhein-Westfafen, No. 438 (1957).

where R is the resistance, Z/a is the cell constant, and p is the density of the molten fluoride at the measuring temperature. Specific conductance as a function of temperature is shown in Fig. 12.3 for several LiF-ThF, mixtures and alkali fluoride-alkali fluoroborate mixtures of relevance to the Molten-Salt Reactor Program. (Insufficient experimental details are contained in ref. 24 to permit a critical evaluation of the data from this source.)

24V. G. Selivanov and V. V. Stender, Russian J . Znorg. Chem. 4, 934 (1959).

143

Table 12.2.

System LiF

Reference 19

847-1027

12,13 23

900 875 958 1037 870-1010

NaF

1000 1040 1080 1020 1003 1086 1138 1030-1090

869-1040

12.13 23

20 19

725-921

BeF2

CaF2

737 784 852

17,18

700 800 950

1.2

cm-'. an-'.

Aeq(cm2 ohms-'

cm-'1

128

160.8 118

1.05

1.2

5.74 (&several %) 5.95 5.15 (&l%) 5*52> 4.960 5.179 5.335 5.29 + 5.64 X 10-3(T 103OOC) (*l%)

113

156.6

-3.493 + 1.480 x 6.608 x 1 0 - 6 9 3.80 ( h % ) 3.573 3.793 4.021 3.77 (f2%)

lo-'

-4.511 + 1.642 7.632 x 2.51 2.73 3.03

10-2T(oC)

1.S7

T measured (OK)

(ohms-'

3.805 + 1.004 X 10-2noC) 3.516 X 8.43 (&I%) 8.663 9.058 9.306 9.06 + 5.83 X 10-3(T 87OoC) (*l%)

236 x

1020

T melting (OK) bStandard deviation = 0.008 ohm-' 'Standard deviation = 0.009 ohm-'

1.05

0.71 x

1.05

900 859 938 1012 905

12,13 23

CSF

T ("C>

T(OC)

124

eq-')

144

12.4 MEASUREMENT OF SPECIFIC CONDUCTANCE IN LiF-BcF, (66-34 MOLE

G. D. Robbins

J. Braunstein

The silica cell shown in Fig. 12.4 was designed for use in the determination of electrical conductance in molten LiF-BeF, (66-34 mole %). The 68-5545

ORNL-DWG

R O D , 4 x 30 cm

\ P t W I R E , 4 mm /GLASS

/ I

TAPE

TUBING, 0.3

2 0 cm

small-diameter tubing increased the conduction path and resulted in measured resistances of the order of 100 ohms. A Wheatstone bridge incorporating a variable resistance and capacitance connected in series was constructed, and its use eliminated errors resulting from approximations often employed in conductivity measurements.2 The platinized platinum electrodes (d 2 1mm) were held in fixed position relative to the cell assembly by glass tape. The cell constant, Z/a, of the silica cell was determined a t 340 and 38OOC in molten potassium nitrate by measuring the resistance in the frequency range 600 to 4000 hertz and extrapolating to infinite frequency vs P 1 I 2 . The observed frequency dependence is shown in Fig. 12.5. From the infinite-frequency resistance R m and literature values of the specific conductance K vs temperature for KN0,,26 the cell constant was determined to be 129 cm-' a t both temperatures. These quantities are related by the expression

f THERMOCOUPLE WELL

The frequency dependence of measured resistance of LiF-BeF, (66-34 mole 74) is shown in Fig. 12.6 at several temperatures. The specific conductivity, calculated from the values of resistance at infinite frequency and Eq. (l),is presented in Fig. 12.7 a s a function of temperature. From the slope of this line, the temperature coefficient at 50OOC is calculated to be 3.9 x ("C)-'. Employing the density data of Cantor2 and the relation mol wt

Am =-K

density

the molar conductance Am a t 5OOOC is 25.5 cm2 ohms-' moles-'. The measurements were completed during the first hour the cell was in contact with the molten fluoride, and we believe the accuracy to be better than *lo%. It was observed that in the molten fluoride system the measured resistance increased slowly as a function of time. In a subsequent experiment

Fig. .12.4. Silico Conductivity Cell.

25G. D. Robbins, Electrical Conductivity of Molten Fluorides, a review, being reviewed for publication. 26A. Klemm, p. 567 in Molten Salt Chemistry (ea. by M. Blander), Interscience, New York, 1964. 27S. Cantor, internal memorandum, 1967.

145

ORNL-DWG 68-5546

ORNL-DWG 68-5548

f (Hz)

4000

2000

1200

800

600

220

1.6 240

-E

4E 1.4

200

490 1.2

180

1.0

450 170

0.a

0.02

'h.7

0.03

0.04

510

530

(Hz-"')

12.7. Specific Conductance vs Temperature for

L i F - B e F Z (66-34 Mole %).

12.5. Meosured Resistance vs Frequency i n KN03.

Fig.

490 TEMPERATURE ("C)

Fig.

Molten

470

0.05

ORNL-DWG 68-5549

480 ORNL- DWG 68-5547

f (Hz) 160 1

100

-0-0--

./.--*--

.-.--.-.- .

oQ g

47OoC

485'

./-.-A.

.-e--

r 0

-E -

.*5000

1 0

0.09

0.02

0.03

0.04

0.05 "V

100

120

TIME (hr)

Fig.

12.6.

Measured Resistance vs Frequency i n

Molten L i F - B e F 2

(66-34 Male %).

employing a different cell [(Z/a) = 164 (fl) cm-' 1 and platinized platinum electrodes which had been previously exposed t o molten LiF-BeF, (66-34 mole %), the slopes of the measured resistance vs f-'I2 plots were investigated as a function of time. The results a r e shown in Fig. 12.8. The frequency range over which R vs f-'I2 was linear increased at longer exposure times; however, the

Fig. 12.8. Slope of Measured Resistance vs f-1'2 485OC i n L i F - B e F 2 (66-34 Mole %).

value of resistance extrapolated to infinite frequency also varied (Fig. 12.9). This cannot be explained in terms of changes in cell constant, since (Z/a) was redetermined at the completion of the experiment to be 171 cm-'. Investigation of the frequency dependence of measured resistance is proceeding.

146

140

ORNL-DWG 68-5550

110

60 TIME (hr)

400

120

Fig. 12.9. Values of Resistance Extrapolated to Infinite Frequency vs Time for'LiF-BeF2 (66-34 Mole at

4asOc.

12.5 A STIRRED REACTION VESSEL FOR MOLTEN FLUORIDES C. E. Bamberger T. J. Golson

C. F. Baes, Jr.

J. Nicholson

In the past, gas sparging, stirring, and rocking have been used to provide agitation in studies of heterogeneous equilibria involving molten fluorides. The stirred vessels, while providing the most vigorous mixing, tended to leak a t the seal of the stirrer shaft. A rocking furnace or gas sparging did not provide very vigorous agitation. More recently, the need for an improved reaction vessel became urgent in connection with the planned study of the distribution of U4 +,Th4 +, and P a 4 + between fluoride and oxide phases.28 The attractiveness of such an extraction system a s a protactinium recovery process obviously depends as much on favorable equilibration rates as on favorable equilibrium behavisr. But to achieve good extraction rates, one needs good suspension and mixing of the heavy oxide solids with the fluoride salt; this requires a suitably designed mixer. An improved stirred vessel has therefore been designed and constructed. It includes the following features (Fig. 12.10):

1. The shaft is sealed by means of Teflon chevron rings which are stacked around the shaft and compressed by a nut. This seal is water cooled.

28B. F. Hitch. C. S. Bambereer. and C. F. Baes. Tr...

MSR Program S A i a n n . Progr. kept. Aug. 31, 1967, " ORNL-4191, p. 136.

2. The upper (cool) shaft bearing is a sleeve of oil-impregnated metal. The lower (hot) bearing is a sleeve of graphite.

3. The stirrer is driven by a flexible cable connected to a variable-speed dc motor. It allows controlled stirrer speeds in the range 60 to 1750 rpm. 4. The shaft and stirrer blade (of INOR-8)can easily be removed and replaced. The vessel, flanged to the stirrer assembly, is cheap to fabricate and is easily replaced.

The apparatus currently is being used to complete a study of the reactions of SiO, with LiFBeF, m e l t s (see Sect. 12.6). T h e helium cover gas, maintained at -0.4 atm (gage), is circulated continuously through the melt and then around a closed circuit which includes an infrared gas absorption cell. Measurement of the infrared absorption bands of SiF, is used to monitor the course of the reaction. The growth of the CO, absorption peak also provides a sensitive measure of the rate of oxygen inleakage, since CO, is produced by reaction of oxygen with the traces of graphite present in the melt. With the stirrer-shaft seal lubricated by a small amount of oil, the stirrer has been operated continuously at >300 rpm for several days a t a time. While wear of the Teflon rings requires an occasional tightening of the packing nut, the outleakage of helium is easily maintained below "1 c m Hg of pressure per day, while the rate of air inleakage, a s indicated by the CO, content of the cover gps, has been nearly undetectable. This reaction vessel should prove a versatile apparatus for a variety of future studies of heterogeneous equilibria involving molten salts.

12.6 THE CHEMISTRY OF SILICA . IN MOLTEN LiF-BcF C. E. Bamberger

C. F. Baes, Jr.

As reported p r e v i o u ~ l y silica , ~ ~ ~has ~ ~ been found to be a satisfactory container for melts in

"C. E. Bamberger, C. F. Baes, Jr., and J. P. Young, J . Znorg. Nucl. Chem., to be published. 30C.E. Bambereer. R. B. Allen, and C. F. Baes. Jr.. Reactor Chem. Di;. Ann. Pro&. Rept. Dec. 31, 1967, ORNL-4229, p. 60.

147

ORNL-DWG

68-3432

FLEXIBLE DRIVE THRUST BEARING

OlLlTE BEARING

THERMOCOUPLE

WELL

SPARGE GAS TEFLON CHEVRON PACKING SEAL SAMPLING

TUBE

H20 COOLING COILS

“O‘RING

SEAL

GRAPHITE BEARING

NI VESSEL

Fig. 12.10.

Improved Stirred Reaction Vessel for Molten Fluorides.

PACKING N U T

148 ORNL-DWG 68-4647R

which Li,BeF4 is the solvent salt. The partial pressures of SiF, generated by the reaction

t ("C) 800 200

+ 4F- e 20,- + SiF4(g)

SiO,(c)

(1)

generally are low, and the oxide ion produced by this reaction can easily be held to low levels if SiF, is introduced into the cover gas. In visual experiments and in spectrophotometric studies, silica containers were found to remain intact and did not contaminate the m e l t provided the SiF, generated was retained in the cover gas or if SiF, was already present in the cover gas. The chemistry of SiO, in the presence of MSR fluoride mixtures is being studied further with the objectives of (1)establishing the equilibrium solids which can coexist with such melts and (2) examining their ion exchange properties. If the above reaction is allowed to proceed to the right by sweeping away the SiF, produced, the oxide ion concentration should increase until a new oxide-containing solid phase precipitates. Beryllium oxide is normally the oxide phase which first precipitates in molten Li, BeF,, but according to the Si0,-Be0 phase diagram, SiO, should react with B e 0 to produce Be2Si04 (phenacite), 2BeO(c)

+ SiO,(c)

Be2Si04

I '

500

600

400

E E

k' $ 1

0.I

0.04 0.9

1.O

4.4

4.2

4.3

4.4

4.5

'ooyT(oK)

(2)

Hence, unless some more stable oxyfluoride phase can form in the presence of the molten fluoride, it is expected that as reaction (1) proceeds far enough to the right, phenacite should appear as a stable solid phase, establishing the equilibrium

700

12.11. Partial Pressures of SiF, i n Equilibrium Be2Si04, and Molten 2LiF-BeF T h e point (0) a t upper l e f t i s from Novoselova et a 1 . 6 T h e solid line i s based on Eq. (4); the dashed line i s based on avai lab le thermochemi ca I data. Fig.

with SiOq,

30031

2SiO,(c)

+ 2BeF,(d) e Be,SiO,(c) + SiF,(g) , (3)

K = Psi,

4I a B e F 2 *

This equilibrium has been investigated directly by the recirculation of helium carrier gas through a n Li,BeF, melt to which both SiO, and Be2Si0, had been added and then through an infrared absorption cell, where the intensity of the SiF, peak at 1033 cm-' was determined. The gas was pumped around this circuit by means of a finger pump. The following expression reproduces the partial pressure measurements within their estimated uncertainty (*lo%,Fig. 12.11): log Psi,

(mm) = 8.745 4

- 7576/T.

(4)

The observed partial pressures are in general slightly higher than may be predicted from available thermochemical data30-32 (cf. Fig. 12.11). These partial pressures are seen to be quite low, however, confirming that quite modest overpressures of SiF, should prevent the precipitation of an oxide-containing phase in the presence of SiO,. At present, a stirred vessel (see Sect. 12.5) is being used in continuing these measurements. The

31JANAF Thermochemical Tables, Clearing House for Federal Scientific and Technical Information, U.S. Department of Commerce, August 1965. 32A. V. Novoselova and Yu. V. Ashikina, Znorg. Materials USSR (English Transl.) 2(9), 1375 (1966).

149 ratio of B e 0 to SiO, in the added solids is being varied to ascertain whether or not any stable oxidecontaining phases other than Be,SiO,, SiO, , and B e 0 are formed in the presence of the molten fluorides.

12.7 A SILICA CELL AND FURNACE FOR ELECTROCHEMICAL MEASUREMENTS WITH FUSED FLUORIDES

K. A. Romberger Two distinct advantages are gained if an electrochemical cell can be constructed from fused silica rather than from graphite or corrosion-resistant metals. First, fused silica provides excellent electrical insulation for the electrodes, and second, the cell components are visible so that formation of gas bubbles, solids, etc., may be detected rapidly. Even so, fused silica has seldom been used with fused fluorides in the past because of the corrosive attack of the fluoride salt on the silica. Recently, however, Bamberger et al. suggested that the presence of an SiF4-containing atmosphere would reduce the rate of attack of some molten fluorides on fused silica. This protection of the silica should be the greatest with those fluoride s a l t s which have a relatively low basicity or alkali fluoride activity, for example, LiF-BeF, mixtures that contain 0.33 or higher mole fraction of BeF,. In order t o utilize the advantages of silica, a silica cell and furnace has been constructed for electrochemical studies in LiF-BeF, mixtures. The design is such that it will permit, in a single system with a controlled atmosphere, both equilibrium and nonequilibrium electrochemical measurements including emf, chronopotentiometry, linear sweep voltammetry at fixed solid or liquid electrodes, rotating disk voltammetry, and polarography at a dropping (bismuth) electrode. The purpose of these measurements is to yield information about the diffusion coefficients, decomposition potentials, and oxidation s t a t e s of the species dissolved in the molten salt. The heating element of the furance is a 20-gage asbestos-covered Nichrome wire helix supported by a vertical 4-in.-OD silica tube. A 2'$-in.-ID, 6-in.high silica cup, centered within the heated zone,

33C.E. L. Bamberger e t al., MSR Program Semiann. Progr. R e p t . Aug. 31, 1967, ORNL-4191, p. 137.

holds the fused salt. Atmospheric containment (and secondary salt containment) is provided by an 11in.-long, S-in.-ID closed-end silica tube. This latter tube extends upward to the top of the furnace, where i t is joined, via a silicone rubber seal, to a water-cooled nickel collar. A nickel head having eight ports which terminate in Swagelok fittings provides entry to the cell. Five ports are vertically aligned risers. Three of these risers are used for entry by electrodes. A fourth riser is used either for solid sample additions or for a fourth electrode. The fifth riser admits either a stirrer or a rotating disk electrode through a Swagelok tee fitted with Teflon seals. Helium pressure is maintained through the tee to provide an inert-gas seal. The remaining three ports are for a thermocouple well and inlet and outlet gas lines. The usable working depth in the cell of approximately 4 in. is centered on a line between two 6in.-square windows. The heating element is surrounded by a platinum-covered nickel reflector. Insulation is of Lavite. The furnace requires a 208-v ac source and can be safely operated a t 8OO0C, a t which temperature the required power is 2700 w. A detailed drawing of this cell and furnace is shown in Fig. 12.12.

12.8 STATUS OF THE MOLTEN-SALT CHEMISTRY INFORMATION CENTER H. F. McDuffie In the summer of 1965, we decided to establish a Molten-Salt Chemistry InfOrmation Center. The primary purpose was to provide a detailed subject index t o all the chemical material representing ORNL work in the field since 1950. A secondary purpose was to provide a framework into which other technical reports and open-literature publications could b e inserted. The center has never been formally funded but has required approximately half the time of one technical man and half the t i m e of one secretary. Because of the low level of funding available and the purposes of the center, we designed i t so that it could b e used by technical staff members without assistance and so that new entries could be created with the minimum investment of technical time. An optical coordination indexing system (Termatrex) was chosen as being the most appropriate

150

ORNL-OWG

0 -STIRRER

68-4381

SHAFT

ELECTRODES

SWAGELOK FITTING TEFLON LINERS

THERMOCOUPLE

HEAD (NICKEL)

TEFLON O-RING GASKET

COOLING COLLAR (NICKEL)

TOP PLATE (STAINLESS STEEL) !

FRAME (STAINLESS STEEL)

REFLECTOR (PLATINUM)

SUPPORT AND SALT OVERFLOW CUP (STAINLESS STEEL)

BOTTOM PLATE (STAINLESS STEEL)

Fig. 12.12.

Silica Furnace and Cell for Electrochemical Measurements.

151

! i

(simplest and cheapest) for the purpose. This is the same system used by the ORNL Isotopes Information Center; it was possible to u s e some of their equipment in setting up the Molten-Salt Chemistry Information Center. The two most important functions, from the technical point of view, in establishing an information center for the retrieval of the information contained within a r e the selection of a n adequate list of key words for indexing purposes and the assignment of these key words to particular documents or portions of documents. W e have tried to use care in the selection of our key-word list and have, s o far, limited it to 480 terms. The list is broken into two parts, general terms and materials. General terms include words and phrases such as ABSORPTION SPECTRA, BIBLIOGRAPHY, BLANKET REPROCESSING, CALORIMETRY, DENSITY OF LIQUID, HYDROLYSIS, KINETICS, etc. Materials words include more specific items such as ACTINIDES, ALUMINUM, ARGON, GLASS, GRAPHITE, HF, KF, LiF-BeF, (the system), MERCURY, MOLYBDENUM, MSRE FUEL, etc. Technical people selected these key words, and technical people have made the assignments of key words to articles. Clerical and secretarial people have handled all the other aspects of entering the information into the system. A 5 by 8 card is typed containing the assigned serial number, the identification of the document, the abstract (if available), and the assigned key words. A single cross-reference card is typed, showing the serial number and the identification of the document; it is filed by journal or report number. Then the Termatrex cards for the several selected key words are assembled and punched with t h e serial number of the document. (The arrangement of the serial number punches is on a 100 by 100 grid, s o that any number up to 10,000 can be located by a separate hole.) In addition to the indexing of relevant reports which are brought to our attention, we maintain a subscription to the ASCA (automatic subject citation alert) function of the Institute for Scientific

Information. Each week, their current scientific literature is scanned for articles written by any of 106 source authors whom we have chosen as publishing wholly or significantly in the molten-salt field. The titles of all the articles thus brought to our attention are screened, and the relevant articles are indexed, thus providing some coverage of current scientific literature as well as of our own reports. At the present t i m e the system has accumulated a total of 6600 entries; the rate of accumulation has been somewhat more rapid than we had expected, particularly in view of the low level of effort which we have assigned to the work. The system has been most useful to new members of our staff in permitting theii ready access to the earlier project work in the Molten-Salt Reactor Program. Additional support, which would be expected to materialize with a n increase in the overall program, would permit u s to expand our coverage of both the report literature and the open literature through systematic surveying of Chemical Abstracts and Nuclear Science Abstracts for relevant documents. When the storage capacity, 10,000, of the present system is reached, we can set up a second deck of Termatrex cards and continue with another 10,000 entries, or we can consider whether we have arrived at the stage of justifying the staff and complexity required to put the information into a digital computer data processing system. This operation would be essentially nontechnical, since the system was designed with this ultimate conversion in mind. The flexibility of the system permits continuous improvement of the indexing and e a s y correction of any errors which are found. For the present, it s e e m s preferable to maintain the center as a device for retrieval of requested information rather than as a place for continuing analytical reviews with a regular output to customers. W e can answer some telephone queries quickly and c a n handle some requests from outside writers for specific information, but we do not have the staff to do elaborate state-of-the-art studies at this time.

13. Chemistry of Molten-Salt Reactor Fuel Reprocessing Technology 13.1 MSBR FUEL REPROCESSING BY REDUCTIVE EXTRACTION INTO MOLTEN BISMUTH D. M. Moulton W. R. Grimes J. H. Shaffer Studies of the reductive extraction of protactinium, uranium, and rare earths from molten fluoride mixtures into molten bismuth have continued to yield results which favor i t s application for reprocessing fissile and fertile mixtures of the two-region moltensalt breeder reactor concept. As previously reported in this series, the experimental data relate the equilibrium distribution of extracted metals to concentrations of reducing agent dissolved in the m e t a l phase' and permit the construction of an electromotive series where the relative positions of metals indicate their order of extraction., These values have demonstrated the chemical feasibility of reductive extraction for reprocessing a two-region breeder reactor within a reasonable degree of confidence and have provided a basis for initiating an engineering evaluation of the processing concept. More recent efforts have been primarily oriented toward reprocessing requirements of the conceptual single-region breeder reactor. This concept combines both fissile and fertile material in a single fluoride mixture which will nominally contain 12 mole % ThF, and 0.3 mole % UF, dissolved in a n LiF-BeF, solvent mixture. The reprocessing requirements remain the same, but the chemistry becomes more difficult since there are now two com-

'MSR Program Semiann. Progr. Rept. Feb. 28, 1966,

OWL-3936, p. 141. 'MSR Program Semiann. Pro&. Rept. Aug. 3 1 , 1 9 6 7 , ORNL-4191, p. 148.

binations of elements with similar reduction potentials. Earlier data for a somewhat different solvent, LiF-BeF,-ThF, (73-2-25 mole %) ,indicate that uranium will have to be removed as the first step, with protactinium following under slightly stronger reduction conditions. If the separation factor (-10) obtained from these data is valid for the single-region system, the process of separating uranium from protactinium should be feasible but will require more careful process control. Further studies of the extractability of these elements in the single-region fuel mixture are in progress. Fission product removal will logically be performed on the salt stream effluent from the uraniumprotactinium removal unit ahead of the fuel reconstitution section of the reprocessing plant. Consequently, laboratory studies are concerned with the extraction of rare earths from a mixture of LiF, BeF, , and ThF, where the concentration of thorium will be about 12 mole %, lithium may be varied from 58 to 72 mole %, and beryllium will make up the balance. In a preliminary experiment a small quantity of ThF, was added to LiF-ReF, (66-34 mole %) containing about 100 ppm europium spiked with 52-1 5 4 Eactivity. ~ Extraction from this salt mixture (salt A) into bismuth was carried out as in previous experiments by making metallic lithium additions to the metal phase and by monitoring the progress with a beryllium electrode. Reduction potentials calculated from the measured distributions are shown as part of Table 13.1, which contains current values for several elements obtained from the same salt mixture. If the thorium activity remains equal to its concentration, then at 12 mole % ThF, in the single-region fuel mixture, thorium bismuthide will precipitate at 1.49 v. The distributions of various important elements between salt A and bismuth at 6OOOC are illustrated in Fig. 13.1 as

152

153

f a b l e 13.1.

Element

Extroction Potentials of Fission Products from L i F - B e F 2 (66-34 Mole %) into Bismuth

& 0'

Valence 5oooc

6OO0C

7OO0C

8OO0C

2.00

1.93

1.88

1.83

Bea

1.93

1.85

1.78

1.71

1.61 (1.8)

1.53 (1.8)

1.45 (1.8)

1.56

2.5

1.52

1.6

2.7

1.48

1.39

1.80 1.69(2.0)

1.58 (3.0)

1.48 (2.6)

1.41 (2.4)

1.33 (2.2)

aeo(as referred to pure metal); Be is insoluble in Bi. bValence in parentheses at each temperature.

functions of reduction potentials. The vertical lines mean that a solid phase of constant composition is being formed and will limit the reduction potential of the system until its concentration in the salt phase is exhausted. For thorium, the virtual position of Eo is taken from dilute measurements in salt A, but the onset of solid phase formation is calculated for ThF, = 12 mole % in the salt. The important considerations for the single-region MSBR reprocessing scheme are the voltage difference between uranium and protactinium and the rare-earth positions relative to the vertical thorium line. If these values are valid, the extraction process should be feasible but not as favorable as i t s application with the two-region reactor. Because of the uncertainty of thorium activity a t higher thorium concentrations in the salt mixture and i t s effect on the extraction process, a systematic investigation of rare-earth extraction from salt mixtures of practical interest as a fuel solvent h a s been initiated. Since increasing the L i F concentration of the mixture (hence increasing its free

fluoride content) might increase the stability of Th4' by fluoride complex formation,' initial studies have held the thorium concentratio,, of the salt at 12 mole % and varied the ratio of lithium to beryllium. Three experiments with europium, samarium, and neodymium, each dissolved in LiFBeF,-ThF, (58-30-12 mole %) , were conducted by adding thorium metal to the bismuth pool. Thorium saturation in bismuth was inferred when the potential of a beryllium electrode inserted i n the s a l t mixture stopped falling upon further addition of thorium metal. The amount of extracted rare earth in each case was too low for quantitative evaluation under analytical conditions imposed on the system. Radiochemical analyses were further complicated bv the presence of gamma-emitting thorium daughters, some of which extracted readily into bismuth. Neodymium was seen in the bismuth a t 700 and

3S. Cantor and C. F. Baes, private communication, December 1967.

154

ORNL-DWG 68-3119

I I I I NOTE: FORMATION OF ThBiz CALCULATED FOR ThF, 112 mole % IN SALT

1.9

E. REDUCTION POTENTIAL

(volts)

Fig. 13.1. Distribution of Uranium, Protactinium, and Rare Earths Between LiF-BeF2 (66-34 Mole %) and

values for a appreciably above 1 will denote feasible separations, and. the magnitude of a will indicate extraction efficiency. Thus, if the concentration of rare earth in the salt mixture is 100 ppm by weight, the thorium concentration in the salt is 12 mole %, and the limiting potential corresponds to 3500 ppm thorium (i. e., XTF(Bi) = 3.14 x in the bismuth, then the experimental procedure should detect rare-earth concentrations as low as 0.8 ppm in the bismuth phase. Experiments are now in progress to examine the effects of salt composition on the extractability of rare earths. These experiments will be conducted with cerium because of the convenient radiochemical properties of the 44Ce isotope for analytical' application. When the salt composition has been optimized, extractions of the various important rare earths will be studied. The results of one experiment (still in progress) with the salt mixture LiF-BeF2-ThF4 (72-16-12 mole %) have shown the favorable extraction of cerium into bismuth at 60OOC. Figure 13.2 illustrates the dependence of rare-earth extraction on the thorium concentration in bismuth up to its saturation value. The separation coefficient a is also related to the thorium concentration in bismuth, as shown in Fig. 13.3. The decrease in a with increasing thorium results from the higher valence of thorium than cerium.

Bismuth at 60OoC.

800째C, but samarium, though it should extract more easily, was not definitely detected, probably because it has only a weak and short-lived radioisotope. Potentiometric readings in this salt mixture did indicate changes in the activities of the major salt constituents from values reported for salt A. Values for beryllium and thorium were found about 0.1 and 0.E v, respectively, more noble than in salt A if lithium is assumed unchanged. Since the feasibility of the extraction process depends on the relative extractability of rare earths as compared with thorium, the experimental procedures are being refined to yield significant values for very low extraction efficiencies. If we define this separation coefficient a as

-I f

INITIAL 'CONC OF be IN SAL+- 407ppm' WEIGHT OF SALT = 3.26 kg WEIGHT OF Bit3.00kg *==POINTS AT 7OOOC o ==POINTS AT 600.C Ce CONTENT OF SALT REMAINED AT 107 2 2 ppm DURING ENTIRE EXPERIMENT I I I I

0.5

Fig. 13.2.

'RE(Bi)

/'RE(salt)

'Th(Bi)

/*Th(salt)

4.0

4.5

2.0

2.5

3.0

THORIUM FOUND IN BISMUTH (mole fraction ~403)

Extraction of Cerium from LiF-BeF2-ThF4

(72-16-12 Mole %) into Molten Bismuth by Reduction with Thorium at 60@C.

155

ORNL-DWG

68-4690

CERIUM CONCENTRATION IN SALT = 407 ppm BY WEIGHT D= NMETAL "SALT I W E WEIGHT l j l i T OF OF BSALT=3.26 i T kg kg ~

The experimental program was carried out with a 104-kg batch of the fluoride mixture LiF-BeF2-ZrF4 (65-30-5 mole %), prepared from the component fluoride salts by routine production procedures. In order to represent conditions expected after fluorination of the MSRE fuel salt, approximately 80 g of CrF,, 50 g of FeF, , and 800 g of NiF, were added to the salt mixture, which was hydrofluorinated for 6 hr with 10 to 20 mole % HF in helium to ensure complete dissolution of the corrosion product fluorides. The analysis of a filtered salt sample taken after the hydrofluorination period is compared with expected concentrations in Table 13.2.

b a 0 0.5

4.0

4.5

2.0

2.5

THORIUM FOUND IN BISMUTH (mole fraction x

3.0

Table 13.2. Concentrations of Corrosion Product Fluorides after Hydrofluorinotion

to3) Estimated Concentrationa

Analysisa

(PPI

CrF

445

470

FeF2

2 86

62 0

NiF,

4680

3700

Material Fig. 13.3.

Relative Extraction of Cerium and Thorium

from LiF-BeFZ-ThF4 (72-16-12 Mole 600%.

X) into Bismuth a t

13.2 REMOVAL OF STRUCTURAL METAL FLUORIDES FROM A SIMULATED MSRE FUEL SOLVENT

aConcentrations reported on a metal basis.

L. E. McNeese F. A. Doss J. H. Shaffer 13.2.1 Reduction of Structural Metal Fluorides Current plans for the MSRE call for substituting 233U for 235u in the reactor fuel mixture. The fuel solvent mixture, LiF-BeF2-ZrF4 (65-30-5 mole %), together with its fission product inventory, will be retained; this will be accomplished by fluorinating the fuel salt to remove its present uranium content and by adding 233U as the concentrate mixture LiFUF, (73-27 mole %) During fluorination the salt will become contaminated with fluorides of nickel, iron, and chromium as a result of corrosion of the Hastelloy N fluorinator. Although the expected concentrations of these fluorides will not exceed their solubilities in the fuel salt, their return to the reactor is undesirable. Therefore, the reduction of structural metals from solution in the fluorinated fuel solvent and their removal by filtration from the solvent during i t s transfer to the reactor system are requirements of the refueling operation. An experimental simulation of these steps has been conducted to define better the feasibility of the refueling procedure.

Equilibrium data4 for reduction of the corrosion product fluorides indicate that if equilibrium is achieved, a hydrogen sparge will reduce NiF, easily, FeF, with difficulty, and CrF, to a negligible extent in a practical t i m e period. The a i m of the first part of the experiment was to demonstrate NiF, reduction with a hydrogen sparge and to compare hydrogen utilization with that predicted from equilibrium data. After the initial hydrofluorination, free HF was stripped from the salt mixture by helium sparging, and the molten s a l t was sparged for 9 hr with hydrogen at a rate of 10 standard liters per minute a t 65OoC. The 48.2-liter salt mixture was contained in a 12-in.-diam vessel, and the sparge line terminated 2 6 in. below the melt surface. The gas effluent from the treatment vessel was analyzed 4C. M. Blood, Solubility and Stability of Structural Metal Difluorides in Molten Fluoride Mixtures, ORNLCF-61-54 (Sept. 21, 1961).

156 ORNL-DWG 68-5551

f4 ._ 3 -

I-3

SALT w t : 404.4 kg [Nil,,,T: 3700 ppm

\4 1

SALT COMP: LiF 65mole 70 BeFz 30mole 70

H p FLOW: 40 liters/min NTP SALT VOL: 48.2 l i t e r s

0-e

ORNL-DWG 68-5552

0.6

z W

3 LL -I

LL W

9 2 W

e .'-

,,-e-

. 0

Fig. 13.4.

Reduction of N i F Z from Simulated MSRE ZIRCONIUM METAL ADDED (9)

Fuel Solvent by Hydrogen Sparging at 6SOoC. Fig.

13.5.

Pseudoequilibrium of Hydrogen with

Structural Metal Fluorides During Their Reduction from

periodically for H F (Fig. 13.4), and sparging was discontinued when the HF concentration decreased to a low but near constant value. The quantity of HF produced during this period was equivalent t o about 93% of the NiF, initially present in the s a l t mixture. Hydrogen utilization was considerably lower than would have resulted if equilibrium had been achieved; under equilibrium conditions a sparge time of 20.4 min would have resulted in reduction of 9% of the NiF, present initially. Since equilibrium data predict very low hydrogen utilization during reduction of FeF, and CrF, , data were obtained on reduction of these fluorides with zirconium metal. The zirconium was first machined in order t o increase the surface area and thereby reduce passivation of the surface by deposition of reduced iron and chromium. The turnings were pressed into 3~-in.-diam cylinders ( p -4.6 g/cm3 ) having lengths of 0.3 to 1.0 in. to facilitate addition to the treatment vessel. The pressed zirconium slugs were added t o the salt mixture after reduction of NiF, with hydrogen sparging. After each addition the approximate extent of reduction was noted by sparging the m e l t with hydrogen and analyzing the effluent gas for HF. As shown in Fig. 13.5, the

Simulated MSRE Fuel Solvent by Zirconium Metal a t

650OC.

HF concentration decreased steadily, and an arbitrary end point for the reduction reaction was established at a concentration of 0.02 meq of H F per liter of H as a result of experience during preparation ol the initial MSRE fuel carrier salt. The calculated quantity of zirconium required to complete reduction of the corrosion product fluorides was 115 g; this was based on the analysis of the s a l t mixture shown in Table 13.2 and assumed reduction of 12.2 equivalents of these metal fluorides during the hydrogen reduction period. Analyses of filtered samples taken near the conclusion of the zirconium addition period are shown below. Zirconium Added

(e)

Metals Found (ppm by weight) Cr

253

<30

327

6.1

157

These values are below limits established for the initial MSRE fuel carrier salt. On the basis of these data, zirconium utilization was at least 46 to 65%.

The experimental assembly used for the filtration tests was essentially that used for the routine production of fluoride mixtures. The salt preparation vessel was connected by a 2-in.-diam Inconel tube 13.2.2 Salt Filtration Studies to a heated s a l t receiver vessel. This transfer tube extended near the bottom of the salt preparation The reduction Of structural metal corrosion products vessel and only penetrated the top of the receiver. from solution in the MSRE fuel solvent is anticiThe porous metal filter was mounted vertically in pated to produce approximately 330 g (0.73 lb) of the transfer line above the receiver vessel. A solid particles per cubic foot of salt mixture. The second s a l t return line extended near the bottom of suspension or entrainment of a major fraction of the receiver vessel and connected to the salt prepathese solids during the transfer of the fuel solvent ration vessel. These lines were used alternately back to the reactor drain tank assembly could have for forcing the 1.7 ft3 of salt mixture from one vesadverse effects on filter performance. Therefore, sel to the other and were capped when not in use. the primary purpose of this program was an examiThis arrangement permitted repeated filtration t e s t s nation of salt filtration characteristics under conto be conducted on the s a l t mixture. Eight filtraditions which would reasonably simulate those tions were conducted during this investigation. anticipated in the reactor application. Filtration Although conditions which would promote the tests were separately conducted on a sintered nickel decantation process are desired for the reactor product of the Micro Metallics Corporation, having application, attempts were made also to observe pores about 40 p in diameter, and two grades of filter loading capacities. Tests under static conFeltmetal, a product of Huyck Metals Company. ditions prevailed when the melt had remained Each filter was fabricated as a 25/-in.-diam plate quiescent for a minimum time of 4 hr prior to transso that geometric surface areas of all filters would fer. Agitated melt conditions were established by be identical. One grade was of Monel with pores rapidly sparging the melt with helium just prior to 20 p in diameter and a second of 347 stainless steel The temperature of the m e l t was consalt transfer. having 41-p-diam pores. The latter, designated trolled a t 650째C (1200OF) for all tests. The indiFM250 by the manufacturer, had a porosity of 78% cated temperatures of the s a l t transfer line fell to and a rated capability for removing 98% of all about 565OC (1050OF) during the extended transfer particles over 10 p in diameter. This material was operations. Lower temperatures were erroneously chosen for application in the reactor system because recorded from thermocouples that became separated of its adaptability to fabrication requirements and from the transfer line. In no instances did the s a l t its satisfactory performance during these filtration freeze in the transfer line or filter during the tests. tests. The credibility of the multiple filtration tests that In each test, s a l t transfer was induced by inwere performed on a single salt preparation is based creasing to 11 psig the pressure of helium above on the anticipation that the primary process for rethe salt mixture in the treatment vessel and evacumoving solids from the MSRE fuel solvent will be ating all gases from the receiver vessel. These that of decantation. This conclusion is supported conditions were maintained throughout each test. by considerable experience in filtering fluoride mixThe pressure drop across the filter varied as the tures through sintered nickel during the production level of s a l t in the treatment vessel decreased. A of fluoride mixtures for the MSRP. Examinations of summary of the filter performance tests is shown in filters used in this process have consistently shown Table 13.3. The 20-p Feltmetal was tested first, the collection Qf very small quantities of solids on and i t w a s disqualified after 2 hr operation. Esthe filter plates. This condition has persisted sentially no s a l t was passed through the filter despite the accumulation of considerable quantities during this period. Inspection of the filter also of solids in the salt treatment vessels during their showed that the predominant part of the 22-g holdup was salt, and only a thin layer of metallic material repeated use i n the production process. Thus, the novel feature of these current filtration tests is, was on the filter plate. The second test, with sintered nickel, showed remarkable performance but perhaps, the indirect estimate of particle size distribution of solids formed by the reduction process collected only 7 g of material on the filter plate. proposed for the reactor application. Since no visible failure of the filter plate was ap-

158 Table 13.3.

Summary of Filtration Tests

Salt composition: LiF-BeFz-ZrF, (65-30-5 mole %) Weight of salt mixture: 104.1 kg Volume of salt a t 65OoC: 1.7 ft3 Indicated pressure differential: 11 psig forepressure vs vacuum

Test

Filter Material

No.

Pore Diameter

Salt Conditions

Monel Feltmetal

Static

Sintered nickel

Static

Transfer Time

Weight Gain

(hr)

(E)

0.5

Remarks

Test terminated after 2 hr; essentially no salt transfer

No visible material on filter or evidence of failure

Sintered nickel

Agitated

1.75

347 SS Feltmetal

Static

2.0

Test stopped after 90-kg transfer

Sintered nickel

Static

2.17

Filter plugged efter 40-kg transfer; material on filter predominantly s a l t

Sintered nickel

Static

1.84

Balance of s a l t transferred; filter ruptured

347 SS Feltmetal

Agitated

2.0

Sintered nickel

Static

1.75

parent, we presume that either the decantation process was extremely efficient or the suspended particles were too small to impinge on the filter plate. The balance of the filtration tests showed comparable results which were essentially independent of the static or agitated condition of the melt. Under the test conditions, filtration times for FM250 Feltmetal were about 1.19 and 1.36 hr per cubic foot of s a l t mixture. The longer transfer time probably reflected a lower salt temperature a t the filter plate. The occasional plugging of the filter in tests 4 and 5 suggests that the loading capacity of the 4Ckp filters was about 50 to 75 g of metal particles. Samples of the s a l t mixture were withdrawn by filter stick prior to the first filtration experiment and by a composite sampler downstream from the

80-kg back transfer of salt from receiver to treatment vessel; filter plugged.

filter plate after tests 4, 6, and 7. The results of chemical analyses of these samples are shown in Table 13.4. Although test 6 resulted in a filter rupture, corresponding chemical analyses do not reflect excessively large concentrations of structural metals in the filtered salt. Variations in the other results are probably within the combined reproducibility of the analytical methods and the test procedure. If these concentrations are indicative of solids passing through the filter plates rather than unreduced metals, then the data reflect the particle size distribution of the reduced material. We conclude that only 1.8 to 3.4% of metals reduced from solution in the MSFS fuel solvent will pass through the filter material proposed for use in the reactor application.

159 Table

13.4.

Summary

of Analytical Results During Filtration T e s t s Impurity Concentration (ppm)

Sample Interval Cr

Total

Before test la

1 06

After test 4

165

After test 6

256

132

404

After test 7

aFilter stick sample.

13.3 PROTACTINIUM STUDIES IN T H E HIGH-ALPHA MOLTEN-SALT LABORATORY C. J. Barton H. H. Stone

J. C. Mailen W. R. Grimes

Technical feasibility of the “Brillo process,” a method for removal of protactinium from 7LiF-BeF,ThF, melts by reduction from the salt phase and adsorption on s t e e l wool, was demonstrated previously and has been described. s Because application of this method involves separations of liquids and solids in highly radioactive environments, i t s adaptation at the engineering scale does not seem to be attractive. For this reason, we have begun studies in which protactinium is reduced by liquid bismuth-thorium alloys. The results of similar studies6 conducted at the tracer level indicate that removal of protactinium from molten fluorides by liquid m e t a l extraction is a potentially successful fuel reprocessing method. This investigation is designed to confirm the chemical validity of the process and to provide basic information needed to establish optimal processing conditions. Before beginning liquid-metal extraction studies, we performed one solid thorium reduction experiment that was planned to elucidate the role of iron in protactinium precipitation. The results of this experiment are summarized below.

13.3.1 Protactinium Reduction by Solid Thorium i n the Near Absence of Iron Previously, we found that in LiF-ThF4 melts, iron is reduced rapidly at 6OO0C by thorium and at a much slower rate by hydrogen. We chose to examine the reduction of iron a t 7000C in an LiF-ThF4 m e l t which also contained protactinium. Treatment of the melt with hydrogen reduced the iron concentration from an original concentration of 500 ppm to 3 ppm (as indicated by 59Fe counts) in about 18 hr, in contrast to a 40-hr period required at 6000C.7 The quantitative significance of the results obtained was somewhat impaired by partial loss of protactinium during hydrogen treatment and during transfer from the nickel pot to a graphite-lined pot. However, it appears qualitatively that there is little difference i n the behavior of protactinium metal during reduction with solid thorium except that in the presence of iron a slightly greater fraction of protactinium remains in the unfiltered s a l t sample after reduction than is found i n using the bismuththorium alloy.

13.3.2 Reduction of Protactinium by Bismuth-Uranium Alloy Two equilibration experiments were performed with bismuth-uranium alloys which initially contained 1 at. % uranium in order to simulate condi-

’C. J. Barton and H. H. Stone, MSR Program Semiann. Progr. Rept. Aug. 31. 1967, ORNL4191, p. 153.

‘C. J . Barton and H. H. Stone, Reduction of Iron Dis6J. S. Watson and M. E. Whatley, Protactinium Removal from Molten-Salt Reactor Fertile Salt, internal memorandum, solved in Molten LiF-ThF,, ORNL-TM-2036 (Nov. 2, Jan. 24, 1968. 1967).

160 tions under single-fluid reactor conditions. The first of these experiments is described elsewhere. a In summary, analyses of both filtered and unfiltered samples of bismuth and salt phases removed after 16 and 20 hr contact accounted for only about 42% of the uranium added initially to the bismuth. This probably indicated that part of the uranium was precipitated a s UO, by oxygen or water vapor inadvertently introduced into the system. The presence of UO, means that uranium might have been introduced into the salt by the reaction UO, + ThF, -,UF, + Tho,, as well a s by the reaction Uo + ThF, + UF, + Tho. Both reactions have positive free energy for the direction indicated, but since ThF, is present in large excess over the amount of Uo or UO, and since complexing of the UF, by LiF occurs in the molten-salt phase, either reaction might be expected to take place. Since the relative contribution of the two possible reactions cannot be evaluated, the data obtained have only qualitative significance. The uranium content of the frozen unfiltered salt was much higher than that of the unfiltered samples removed from the molten salt. This indicates that the UO, either dissolved or dispersed in the salt as it froze, and since the salt was quite green, it s e e m s most likely that it dissolved. Repetition of the experiment resulted i n a significant improvement. Uranium losses were found to be negligible, and the protactinium removal rate was encouragingly rapid. Protactinium distribution data displayed in Fig. 13.6 showed a marked drop in concentration in the bismuth during the first 5 hr with relatively little change during the l a s t 11 hr. Although the concentrations of protactinium in the salt phase appear to vary somewhat, about 90% of the initial amount of the element was removed from the salt after 5 hr of contact. The initial low value in the salt phase coincides with the initial high uranium concentration in the filtered bismuth (Fig. 13.6).

13.3.3 Reduction of Protactinium by Bismuth-Thorium Alloys We have examined the transfer of protactinium from molten fluoride mixtures to bismuth-thorium alloys

' C . J. Barton and H. H. Stone, Reactor Chem. Div. Ann. Progr. Rept. Dec. 31, 1967, ORNL-4229, p. 47.

ORNL-DWG 68-5553

CONTACT TIME (hr)

Fig. 13.6.

Protactinium Distribution Between LiF-

BeFZ-ThF4 (73-2-25 Mole Alloy. Run 1-16.

W) and

Bismuth-Uroniurn

using the two-region breeder blanket composition LiF-BeF,-ThF, (73-2-25 mole %) and the singleregion fuel composition LiF-BeF,-ThF, (72-16-12 mole %). They will be discussed in that order.

13.3.4 Two-Region Breeder Blanket Composition Two experiments were performed to measure the extent of transfer of protactinium to bismuth-thorium alloy and the distribution of uranium. In both experiments the amount of thorium added initially to the bismuth was sufficient to give a concentration of 2500 ppm if all the thorium dissolved. Analysis of a sample of bismuth-thorium alloy removed 2 hr after addition of the thorium showed 2220 ppm, but five days later, after extended treatment of the s a l t phase occasioned by difficulty in keeping the protactinium in solution during hydrogen treatment: the thorium concentration had dropped to 220 ppm. W e assume that water vapor or oxygen in the system was causing the precipitation of protactinium be-

161

I 1

cause 90% of the thorium in the bismuth-thorium alloys was converted to Tho,. Despite the presence of Tho,, about 1.0 to 1.5% of the protactinium was transferred to the bismuth with low concentration of thorium metal in the bismuth. The protactinium transfer rate was increased sharply to about 20% of the total within 1 hr when an additional amount of thorium (totaling 2500 ppm) was added. The protactinium content of the bismuth dropped to about 12% of the total in the system after standing overnight under flowing helium without agitation, and a further drop, to about 3%, occurred when the phases were mixed by sparging with helium for 1'/4hr. The abovementioned existence of solid Tho, in the syst e m probably accounts for the removal of the protactinium from bismuth. A large increase in the protactinium content of unfiltered salt was noted after the gas sparging. An additional experiment resulted in the most complete transfer of protactinium from salt to bismuth that has been experienced to date. The results, displayed in Fig. 13.7, show about 80% of the protactinium in the filtered bismuth after only 30 min contact of the phases. Since the first set of samples accounted for 125% of the protactinium, it seems probable that the protactinium was not homogenously dispersed in the bismuth at that time.

If we assume that the calculated protactinium content of the bismuth in the first sample was too high, then it appears that there was relatively little change in the protactinium content of the bismuth over an 18-hr period. The minimum protactinium balance observed in this experiment was 80%(18-hr samples). The uranium distribution data are displayed in Fig. 13.8. The uranium was present in the salt at a n initial concentration of about 50 ppm. About 80% was found in the filtered bismuth, and there was very little left in the filtered salt after 18 hr contact time. Furthermore, the data indicate that the transfer of uranium to the bismuth occurred rapidly, as did the transfer of protactinium. 13.3.5 Single-Region F u e l Composition A series of four experiments was performed with the single-region reactor fuel composition LiF-BeF,ThF4 (72-16-12 mole %). The principal variables were the concentration of uranium fluoride added to the salt phase and the thorium concentration in the bismuth. Lanthanum was added to the salt in s o m e

I00

90 ORNL-DWG 68-2747

4 00

e c

z 60 0

'0 7 0

E 50

W V

5 40

k W

2 30

10 40

0 0

Fig. 13.7.

6 8 +O 42 44 46 TIME OF CONTACT ( hr 1

Distribution of Protactinium Between

Fig. 13.8.

6 8 $0 12 CONTACT TIME (hr)

Distribution of Uranium Between LiF-

Bismuth-Thorium Alloy (2500 ppm Th) and LiF-BeFZ-

BeF2-ThF4 (73-2-25 Mole %) and Bismuth-Thorium

ThF4 (73-2-25 Mole %) at 625OC.

Alloy.

Run 1-8.

162 Table 13.5.

Run No.

Experimental Conditions for Extraction Experiments with Single-Region Fuel, LiF-BeFZ-ThF4 (72-16-12 Mole %)

Uranium Concentration

Lanthanum Concentration

231pa Concentration (PP4

Weight of Thorium Added to Bismuth (g)

(mole %)

(PPm)

1-30

0.00

Tracer

3.0

2 -5

0.25

100

3.0

2-15

0.125

100

1.93

2-26

0.01 0

None

140

of the experiments in an effort to compare the transfer of a rare earth with that of protactinium under the same conditions. Experimental conditions in the four experiments are included in Table 13.5. Protactinium distribution data in run 1-30, shown in Fig. 13.9, indicate that protactinium reduction was nearly complete after 7% hr contact of the salt and bismuth phases, with gas (helium) mixing. However, protactinium transfer to the bismuth was much less complete than i n the case of the tworegion blanket composition (Fig. 13.7), and the protactinium balance after the first (3Gmin) set of samples was poor. This shows that a large fraction of the protactinium was associated with some solid phase and was inaccessible to sampling. The amount of thorium metal present was more than enough to saturate the bismuth at the temperature of the experiment (62SoC), so that insoluble intermetallic complexes may have been present in addition to solid thorium. Lanthanum reduction was quite incomplete, and the l r o L a balance was not very good except for the 30-min samples, where 98%was found in the filtered salt and 2% in the filtered bismuth. Subsequent filtered bismuth samples showed only 0.2 to 0.5% of the initial amount of IrOLa (not shown in Fig. 13.9). The amount of thorium added in run 2-5 (3.0 g of thorium to 300 g of bismuth) was calculated to provide an excess of 2500 ppm Tho. The protactinium distribution data are shown in Fig. 13.10. Protactinium in filtered salt samples went through a minimum at about 8 hr contact t i m e and then increased, while that in the filtered bismuth decreased after reaching a maximum at the same contact time. The uranium distribution (Fig. 13.11) shows a minimum in the uranium concentration of filtered salt samples, but

0.50

+ 0.50

the uranium concentration in the bismuth did not drop off during the same period. The thorium concentration in bismuth, also shown in Fig. 13.11, dropped off steadily during the course of the experiment, from an initial concentration of 2500 ppm to 160 ppm after 19 hr contact time. The final low concentration of thorium in the bismuth provides a possible explanation for the protactinium distribution data observed in this experiment (Fig. 13.lo), but there is no obvious explanation for thorium loss. Except for short exposure of stainless steel samplers during removal of bismuth samples, graphite was the only structural material in contact with the two phases. The copper s a l t samplers only contact the salt phase during sampling. Further investigation will be required to provide an explanation for thorium loss and the poor protactinium balance observed in the later part of the experiment. In this experiment we demonstrated that the reduced protactinium could be returned to solution in the molten s a l t in the presence of bismuth by treating the mix with anhydrous H,-HF (H, :HF volume ratio varied from 10:1 to 5:l). Approximately 28 hr were required to redissolve all the protactinium because it was necessary to hydrofluorinate all the metallic thorium and uranium in the system as well as the reduced protactinium. W e had previously demonstrated reversibility of the protactinium precipitation in some experiments with solid thorium, but this was the first t i m e that we showed that liquid bismuth does not interfere with the hydrofluorination process. (previously reported9 tracer-level experiments indicated that this result would be expected.) '5. H. Shaffer e t al., MSR Program Semiann. Progr. Rept. Aug. 31, 1967, ORNL-4191, p. 148.

163

L*â&#x20AC;&#x2122;

ORNL-DWG A

68-5555 I

LANTHANUM CONTENT OF FILTERED SALT

. 0

LANTHANUM CONTENT OF FILTERED SALT LANTHANUM CONTENT OF UNFILTERED SALT PROTACTINIUM CONTENT OF FILTERED SALT

PROTACTINIUM CONTENT OF UNFILTERED SALT PROTACTINIUM CONTENT OF FILTERED BISMUTH PROTACTINIUM CONTENT OF FILTERED BISMUTH PLUS FILTER

8 to 12 CONTACT TIME (hr)

Fig. 13.9. Reduction of Protoctinium and Lanthanum in LiF-BeFg-ThF4 (72-16-12 Mole %) by Bismuth-Thorium Alloy at 625OC.

Uranium distribution data in run 2-15 (Fig. 13.12) show that uranium reduction was slow and incomplete with a poor uranium balance, especially during the later part of the experiment, The thorium concentration in the metal phase (corrected for salt contamination by assuming that all beryllium present was due to salt) is also shown in the same figure. The decrease in thorium concentration proceeded at a linear rate (177 ppm/hr) during the first 9 hr and much more slowly during the last 13*4 hr. This compares with the uniform rate of decrease of 125 ppm/hr over a n 18-hr period in run 2-5. T h e protactinium distribution data (Fig. 13.13), like the uranium data, show slow and incomplete reduction, probably because of the thorium loss from the metal phase. The protactinium concentration in filtered bismuth reached a maximum (3%of total) at 9 hr contact time and then decreased to 0.6%. The pro-

tactinium balance was poor, and the amount found on the stainless steel filter through which the filtered bismuth passed was relatively large, showing that there was a great deal of insoluble protactinium in the metal phase. The lanthanum data were much like those i n run 2-5. The 40Laconcentration in filtered salt samples decreased to 82% of the initial value after 22%hr contact, but the maximum concentration observed in filtered bismuth was 0.5%(3 hr contact time). The last experiment in the series (run 2-26) represented an effort to avoid insoluble phases in the bismuth, The initial addition of thorium (504 mg) was calculated to leave 1000 ppm excess thorium in the bismuth after all uranium and protactinium were completely reduced. After 17%hr, a second addition of 506 mg of thorium was made. Only 233Padata are available at present. On the basis of these data,

164 ORNL-DWG 68-5556

ORNL-DWG 68-2740 4 00

4020

2200

4010

2000

4000

4800

4600

6 c 80

Pa CONTENT OF FILTEREDBi PLUS FILTER m Pa CONTENT OF UNFILTERED-

I-

c 0. +

E60 t

8 I

z a

E 8

I-

O K

5 K

6 8 40 42 44 CONTACT TIME ( h r )

4200

4000

800

--.

-."-.- ..-.-

I..

Alloy at 6 2 P C .

600

400

200

8 5 E 9 I-

Fig. 13.10. Distribution of Protoctinium Between LIFR m F 2-. _The 1\,-7 9 - 1 A - 1 9 Unlpr 9%) ,-,nnd -..- Bismuth.Tharium -----.-

f a

a I

4400

Q-

-5

Fig. 13.12.

9 42 45 CONTACT TIME (hr)

Distribution of Uranium Between LiF-

BeF2-ThF4 (72-16-12 Mole

W ) and

Bismuth-Thorium

Alloy. Run 2-15.

--0 0

6 400

-ez?

70 0

z 2 z

W v)

9 n

! i20

40 0

CONTACT TIME ( h r )

Fig. 13.11.

Uranium Distribution Between LiF-BeF2-

ThF4 (72-16-12 Mole

625'C.

W )and

Bismuth-Thorium Alloy at

the protactinium distribution does not appear to be drastically different from that observed in the three previous experiments. The protactihium content of the filtered salt appeared to be leveling off at about 46% of initial Concentration after 16 hr contact, and the second thorium addition produced only a slight further decrease (to 34%). Only about 4% of the protactinium w a s found in the filtered bismuth prior to the second addition of the thorium, but a sharp increase, to a maximum of 11%,was noted after the addition. However, after another l&hr period, this value had dropped to 3.1%. This experiment will be discussed in more detail when the analytical data are complete. Transfer of protactinium from two-region breeder blanket compositions to bismuth-thorium alloys occurred rapidly and very nearly completely. Reduction of both uranium and protactinium in the single-region reactor fuel composition LiF-BeF,-ThF4 (72-16-12 mole %) occurred much more slowly in contact with

165 ORNL-DWG

68-5557

CONTACT T I M E ( h r )

Fig. 13.13.

Distribution of Protactinium Between LiF-BeFg-ThF4 (72-16-12 Mole %) and Bismuth-Thorium Alloy.

Run 2-15.

periments with single-region fuel may be attributed bismuth-thorium alloys, and the fraction of proto reaction between thorium and BeF,. The reaction tactinium found in filtered bismuth samples was product could be Beo, Be2C, or ThBe, 3 , and the much smaller than for the breeder blanket composiavailable thermodynamic data seem to favor the tions tested. Since the principal difference is in compounds rather than pure metallic beryllium. the BeF, content (16 mole 7% for singleregion vs 0 or 2% for two-region compositions), it seems likely Further investigations are planned in an effort to elucidate the thorium loss phenomenon. that the disappearance of thorium noted in two ex-

14. Behavior of BF, and Fluoroborate Mixtures 14.1 PHASE RELATIONS IN FLUOROBORATE SYSTEMS C. J. Barton

L. 0. Gilpatrick

14.1.1 The System NaF-NaBF4-KBF4-KF Investigation of phase relations in the system NaF-KF-BF a t atmospheric pressure is limited to the portion of the system bounded by the compounds NaF-NaBF,-KBF,-KF, because fluoroborate preparations containing more than 50 mole % of BF, exist only in equilibrium with high pressures of BF,. This system is of interest because it would be desirable t o have a coolant with an even lower liquidus temperature than is provided by the NaF-NaBF, system (38OOC). Phase diagrams for two pseudobinary systems that form part of the ternary system have been published in another report. The NaF-KBF, system has a eutectic containing about 96 mole % of KBF, melting at 540 f 5OC. The hightemperature forms of the compounds NaBF, and KBF, make a continuous series of solid solutions with a minimum-melting composition containing approximately 85 mole % NaBF,. A tentative diagram for the ternary system is shown in Fig. 14.1. There is a ternary eutectic composition in the triangle NaF-KBF,-KF close to the KF-KBF, binary eutectic. The NaF primary phase field covers a large area of the diagram. There appears to be a valley forming the boundary between the NaF and NaBF,-KBF, solid solution primary phase fields, approximately parallel t o the NaBF,KBF, join and quite close to it. The minimummelting composition along this valley, containing approximately 47 mole % NaF, 5 mole % KF, and 48 mole % BF,, is the minimum-melting composition in the ternary system, with a liquidus temperature of 386 f 5OC.

H. Insley' T. N. McVay'

We mentioned in our previous report2 that there were indications that NaBF, recrystallized from dilute hydrofluoric acid solutions was less pure than KBF, prepared by the same procedure and that alternative methods of purifying NaBF, were being considered. One process that has worked well with some materials, slow recrystallization from the melt by passing a tube filled with the material through a temperature gradient, failed to produce any significant improvement in the sharpness of the differential thermal analysis (DTA) melting curve, the most sensitive measure of compound purity that we have available. We found that improved purity of NaBF was achieved by treating molten NaBF, at 4258C with a mixture of anhydrous HF, BF,, and helium (2 volumes of BF, per volume of HF). The sharpness of the DTA melting curve for NaBF, resulting from this treatment approached that produced by our best KBF,. We believe that this treatment removes oxygen, which is probably present as NaBF30H formed by hydrolysis of NaBF,. The presence of a small quantity of this impurity could account for at least a portion of the corrosive action of earlier fluoroborate preparations. A sample of material from a large batch of the NaBF, purchased by a special order from .a commercial producer3 had a melting point only about l0C lower than that of our best laboratory preparation.

'Harshaw Chemical Co. Division of Kewanee Oil Co., Cleveland, Ohio. Specifications supplied by the manufacturer were: NaBF,, 99.08%; oxygen, 0.025%; Pb, 0.004%; Si, 0.01%; Ca, 0.01%; Fe, 0.023%; water insoluble <0.01%; H20, 0.01%. These specifications were confirmed by ORNL analyses, and a total of 2400 lb of this material was received. Major portions are available for study.

'Consultant. 2C. J. Barton et el., MSR Program Semiann. Progr. Rept. Aug. 31, 1967, ORNL4191, pp. 158-59.

166

167 ORNL- DWG 6 8 - 5 5 6 0

NaF (9950)

Fig. 14.1.

The System NaF-KF-NaBF4-KBFr

14.2 NONIDEALITY OF MIXING IN POTASSIUM FLUOROBORATE-SODIUM (OR POTASSIUM) FLUORIDE SYSTEMS

GE = RT In y N e F = AH, [(T/T,) - 11

+ AC, (T, - T) - T x AC, x In (T,/T) - RT In N i a F ,

M. A. Bredig

KF (8580)

720

Experimental phase diagram data available from the preceding semiannual report were examined for deviations from ideality of mixing. Slightly revised liquidus curves for KF and NaF in the binary salt system KF-KBF, and the reciprocal salt system NaF-KBF,, respectively, are shown in Fig. 14.2. The revision is based upon an appropriate application of the known enthalpies of fusion of NaF and KF (AH, = 7.90 and 6.75 kcal/mole) to the liquidus near 100%NaF and KF respectively. It is apparent that the KF-KBF, system deviates little from ideality, whereas the reciprocal salt mixtures NaF-KBF,, containing cations and anionswf widely differing sizes, show large positive deviations from ideality. The latter were estimated semiquantitatively in terms of the partial molar excess free energy of NaF, G E , by means of the equation

Fig. 14.2.

Revised Liquidus Interpolations for the

Systems KF-KBF4 and NaF-KBF4. Experimental points by Barton et at. 2

168

ORNL-WG

sn-sss

ORNL-DWG 68-5164

250 I

400

200

300

400

500

600

‘200

\“

450

0 1

Fa+

400

n 0

WC)

Fig. 14.3.

Excess Chemical Potential of NaF in NaF-

Fig. 14.4.

Heat Content of NaBF4-NaF (92.5-7.5

XI.

KBF4 from Phase Diagram. Experimental points by Barton et aI. 2

Mole

with A C p = 1.0 cal deg” mole-’. Figure 14.3 gives Fe a s a function of, and shows it to be not simply proportional to, the ideal “Temkin” (It must be activity of KBF,, (1 noted that this is not an isothermal plot.) The shape of the NaF liquidus in Fig. 14.2 and the high value - (12,000 cal/mole) of the (interpolated) slope of GiaFbelow (1 N N a J 2= 0.1 in Fig. 14.3 suggest strongly that replacement of NaF by LiF, with the much greater difference in s i z e of L i t from Kt, would lead to separation into two liquids.

mixture was contained in a sealed platinum capsule, which in turn was sealed into an Inconel capsule especially designed for our heat content apparatus. The results are shown in Fig. 14.4. The heat of fusion is 31 cal/g, and the heat of transition is 14.7 cal/g. The heat capacity of the liquid from 380 to 6OO0C is 0.36 cal g-’ oCl’ and the heat capacity of the high-temperature solid from 243 to 38OOC is 0.34 cal g” O C ” l .

14.3 HEAT CONTENT OF NaBF4-NaF (92.5-7.5 MOLE %) A. S. Dworkin The heat content of the eutectic NaBF,-NaF has been measured using a copper block drop calorimeter., The salt mixture w a s supplied by L. 0. Gilpatrick and was from the same batch a s that used for the phase diagram s t u d i e s 5 The ,A. S. Dworkin and M. A. Bredig, J . Phys. Chem. 64,

269 (1960). ’MSR Program Semiann. Progr. Rept. Aug. 31, 1967,

ORNL-4191, pp. 158-59.

14.4 SOLUBILITY OF THORIUM METAL IN LITHIUM FLUORIDE-THORIUM TETRAFLUORIDE MIXTURES

M. A. Bredig

H. R. Bronstein

A knowledge of the free energy of formation of ThF, in various molten mixtures of LiF-ThF, would be of value t o the MSR fuel reprocessing study. An emf study of a cell of the type

Tho I ThF,(x), L i F (1 x)lH,, HFlPt

is the most obvious method for determining this quantity. The cell container must of necessity be nickel because of the u s e of the H,, HF I Pt electrode.

169

However, a study by C. J. Barton and H. H. Stone6 h a s cast s o m e doubt about the feasibility of such measurements in a nickel container. A thorium rod suspended in a molten mixture of the LiF-ThF, (73-27 mole %) contained in a nickel crucible disappeared after 16 hr at a temperature of 600OC. Black magnetic material removed from the cooled m e l t analyzed 45% N i and 30% Th, while nonmagnetic material contained 22% N i and 49.5% Th. An explanation was based on the assumption that the reaction 3ThF,

+ Tho -+ 4ThF,

occurs and that the ThF,, when it diffuses to the nickel wall, disproportionates because of the formation of the Th-Ni intermetallic compound. However, the speculative nature of this explanation was emphasized, as very little is known about the existence of a ThF,. If such a reaction with the nickel container actually occurs, severe complications would arise in a n attempt to measure the free energy of formation of ThF, in the above cell. Before abandoning this much-preferred cell arrangement, a reinvestigation of the reaction of Tho with the LiF-ThF, m e l t was deemed desirable. A carefully dehydrated mixture of LiF-ThF, (73 and 27 m o l e % respectively) was melted in a thorium crucible and held in the molten state under an atmosphere of dry argon at a temperature of 620OC. The m e l t was frequently stirred with a thorium stirrer. After a heating period of 16 hr, a sample of the molten liquid was withdrawn in a tantalum cup. The apparatus utilized for this experiment has been previously described' and illustrated. The analysis of the sample indicated that a p proximately 2.0 mole % thorium metal had dissolved in the melt. Next, a i-in. nickel rod was suspended i n the melt for a period of 5 hr at 620OC. At the termination of the experiment the solidified m e l t was removed from the thorium crucible. A blackish

6C. J. Barton and H. H. Stone, ORNLTM-2036 (November 1967). 7 J. C. Warf, J . Am. Chem. SOC.74. 1864 (1952). *H.R. Bronstein. A. S. Dworkin, and M. A. Bredig, J . Phys. Chem. 66, 44 (1962). 'A. S. Dworkin, H. R. Bronstein. and M. A. Bredig, Discussions Faraday SOC.32, 188 (1962).

layer of material was found at the bottom of the melt. Apparently segregation had occurred upon slow cooling. This black material reacted vigorously with hydrochloric acid. A hydrogen evolution analysis yielded a value of 12.5% by weight calculated as free thorium metal. Spectrographic analysis showed the presence of only very minor quantities of other metals a s impurities. Of some significance is the absence of detectable quantities of nickel, possibly attributable to the very s m a l l surface ratio of nickel rod to thorium crucible. X-ray examination of this blackish material showed lines only of the compounds existing in this m e l t composition, l o that is, mainly Li,ThF, and a small amount of LiThF,. In agreement with this, microscopic examination showed the blackish material to be comprised mainly of the salt mixture. Further experimental work will be performed to corroborate the above findings. 14.5 DISSOCIATION PRESSURE OF BF, FOR THE MSRE SUBSTITUTE COOLANT Stanley Cantor Melts composed mainly of NaBF,, because they possess attractive thermal properties, are under consideration a s coolants in molten-salt reactors. An unattractive property of fluoroborates is the relatively high vapor pressure of BF, caused by dissociation, for example, NaBF,(l)

NaF(2)

+ BF,(g).

To determine the equilibrium constant of the above reaction and to derive other thermodynamic data (see next section), BF, pressures in equilibrium with melts of the system NaBF,-NaF have been measured. Some of the results are shown in Fig. 14.5. (Preliminaj measurements reported in ORNL-4191, pp. 159-61, are superseded by more accurate results shown in Fig. 14.5.) For the mixture proposed as the MSRE substitute coolant, 92-8 mole % NaBF,-NaF, vapor pressures of BF, in equilibrium with the melt of this fixed composition can be represented by the equation 'OR. E. Thoma (ed.), Phase Diagrams of Nuclear Reactor Materials, ORNL-2548, p. 72 (1959).

170 ORNL-DWG 68-5562

10.000

5000

SUPERCOOLED LIQUID

2000

From the observables of the experiment (pressure and volume of BF, vapor, temperature, number of moles of NaBF, and N a F initially charged into the apparatus), an equation was derived for the thermodynamic activity of NaBF,. The equation is:

IO00 500

200

I00

2 5 0 ln W

20 IO 5 2 -

Fig. 14.5.

Vapor Pressure of BFg i n Equilibrium with

NaBF4-NaF Melts.

log P (mm)

9.024

5920

- -. T (OK)

At normal power operating temperature (100O'F) in the MSRE coolant pump bowl, the pressure of BF, will be 53 mm; at zero-power temperature (1200'F) the pressure will be 401 mm.

where P , , is the pressure of BF, in equilibrium 3 with a melt whose composition is denoted by f(N,), defined as the number of moles of NaF initially loaded minus the number of moles of BF, in the vapor divided by the number of moles of NaBF, and NaF initially loaded. The composition variable, f ( N , ) , is very close to, but not exactly equal to, the equilibrium mole fraction of NaF in the melt. Equation (1) was derived from the Gibbs-Duhem equation for the molten phase at constant temperatute. The integral on the right-hand side of Eq. (1) was solved graphically, using Simpson's rule. By integration of the Gibbs-Duhem equation, activities of NaF were derived from the activities of NaBF,. For the most part the activities exhibited slightly positive deviations from ideality. In other words, activity coefficients (defined in terms of the pure liquid components) were greater than unity. Figure 14.6 shows the activity coefficients of NaBF, and NaF a t 1000'K. By combining BF, pressures and activities the equilibrium constant K, was obtained for the reaction ORNL- DWG 68-2994

3.2 2.8

2.4

14.6 CHEMICAL THERMODYNAMICS OF THE SYSTEM NaBF4-NaF

7 2.0

Stanley Cantor

1.6

In a closed apparatus of known volume, BF, pressures in equilibrium with the m e l t were measured manometrically in the composition range 2 to "100 mole % NaBF,. For each composition, pressures were measured over at least a 150' temperature interval within overall limits of 425 to 1200OC.

1.2 1.0

0.8 0

0.f

0.2

0.3

0.4 Q5 0.6 0.7 mde fraction NaBG

0.8

0.9

1.0

Fig. 14.6. A c t i v i t y Coefficients of N a B F 4 and N a F at 1000' K.

171 NaBF,(O

+ BF,(g) ,

NaF(1)

where a

KP = P B F a

At 1000째K, Kp equals 0.182 atm, while at llW째K, Kp is 0.670 atm.

14.7 CORROSION OF HASTELLOY N AND ITS CONSTITUENTS IN FLUOROBORATE MELTS Stanley Cantor A preliminary corrosion test' showed that metallic chromium (the component of Hastelloy N most readily oxidized in molten fluoride media) reacted with the molten mixture proposed as the MSRE substitute coolant, NaF-NaBF, (92-8 mole So). Two products of the reaction were crystalline Na,CrF, and gaseous BF,. Most likely, the removal of NaF from the m e l t i n forming Na,CrF6 caused a shift in the equilibrium NaBF,(I)

NaF(1)

+ BF,(g) ,

thereby establishing a higher vapor pressure of BF,. A program of tests is in progress to

1. determine i f Hastelloy N itself reacts unfavorably with NaBF, m e l t s , 2. verify the initial results with pure chromium, 3. observe whether nickel and iron also react. In each test a salt sample of about 100 g partially decomposes (according to the above reaction) in a n evacuated container fabricated from or containing samples of the test metal; the gas pressure and temperature are continuously monitored. On the basis of the preliminary study, we anticipated that for a constant temperature, the gas pressure would increase at a rate proportional to the chromium corrosion rate. After runs are completed, the containers are cut open and the contents examined. A summary of s i x tests is given in Table 14.1.

'S. Cantor, MSR Program Semiann. Progr. Rept. Aug. 3 1 , 1967, ORNL-4191. p. 161.

The condition of the Hastelloy N containers and specimens following tests 1 and 2 (see Table 14.1) indicated rather sluggish reaction with NaBF,. On the other hand, increases of nickel and chromium in the salt indicated that moderate corrosion occurred; taking into account the surface area of metal exposed to the melt, the concentrations of nickel and chromium in the salt are equivalent to a corrosion rate of 1.5 to 3 mils/year. Test 3, however, provided more encouraging results with regard to both nickel and chromium corrosion. The lack of nickel corrosion in test 3 was essentially verified in test 5. It is possible that the vessels in tests 1 and 2 had undergone some corrosion prior to being loaded with fluoroborate and that the corrosion product was subsequently dissolved by the salt. Test 4 indicates that chromium reacts with either fluoroborate or some impurity contained i n NaBF,. As yet we have not found reduced boron, either in an elemental form or as a metallic boride. Hence, it cannot be stated with any certainty that NaBF, oxidizes chromium. The blackish material noted after completion of test 4 is being carefully examined for a possible boron content.

14.8 COMPATIBILITY AND IMMISCIBILITY OF MOLTEN FLVORIDES C. E. Bamberger C. F. Baes, Jr.

J. P. Young

C. S. Sherer"

In connection with the investigation of fluoroborates as a coolant for molten-salt reactors, the following tests of compatibility and solubility of KBF, and NaBF, with various fluoride mixtures and with B,O, were carried out. It was found that silica and, at lower temperatures, even Pyrex containers were suitable for these tests, thus providing the great advantage of direct visual examination. A s a result, the occurrence of liquid-liquid immiscibility was quickly detected, a phenomenon which was not detected by lessdirect DTA and quenching techniques. In all tests the samples were evacuated several t i m e s and flushed with helium prior to melting. 'Research participant, Alabama College, Montevallo, Alabama.

172 Table 14.1. Test No.

Exposed

Summary of Corrosion Tests with

Time a t Temperature

92-8 Mole % NaBF4-NaF

Vapor Pressure

Observations ,at Termination of Test

Hastelloy N

284 hr a t 56OoC

Rose from 100 to 210 mm in 96 hr; remained a t 210 mm thereafter

Metal exposed t o salt and BF, vapor had blackened appearance; Hastelloy N specimens immersed in melt had not changed weight. Salt cake was white with black scum a t interfaces. Concentrations of chromium and nickel in salt increased by 100 and 650 ppm respectively.

Hastelloy N. s a l t loaded was pure NaBF4

449 hr a t 58OoC

585 mm continuously

Slight blackening of exposed metal surfaces; no change in weight of Hastelloy N specimens. Chromium and nickel in s a l t phase increased by 230 and 1590 ppm respectively.

Hastelloy N

603 hr a t 605OC, then 253 hr at 657OC

150 mm continuously, then 370 mm continuously

Slight blackening of exposed metal surfaces; specimens unchanged in weight. Salt had slight greenish cast, but Na3CrF6 was not detected. Chromium had increased in s a l t phase by 100 ppm but nickel remained < 2 0 ppm.

5 0 g chromium chips; nickel container

837 hr a t 600째C

Slowly rose from 170 to 280 mm

Chromium chips lost 0.5 g. Much green Na3CrFs formed and located on or near chromium chips within the salt cake. Water-insoluble blackish material mixed with Na3CrF6.

Nickel coupons; nickel container

1428 hr at 606OC

170 mm unchanged for 384 hr; slow rise to 460 mm thereafter

Pressure remained a t 310 mm when sample cooled to room temperature. It may therefore be inferred that the rise in pressure during the run was caused by a very slow inleakage of air. Nickel in s a l t phase increased by by about 25 ppm. Nickel metal coupons were unchanged in weight and showed no evidence of attack.

Iron coupons; nickel container

936 hr a t 600째C

220 mm continuously

Test still continuing.

In the following solubility tests, except one, the temperature was 48OOC and the containers were Pyrex. Samples remained molten for periods up to 1 2 hr.

1. NaBF, + CrF, (0.3 wt So): no solubility detected; after l to 2 hr the insoluble CrF, changed to a bright green color, possibly 3NaFXrF 3.

173 2. NaBF, + CrF, ("1 wt %): no solubility detected; change of color indicating possible 3NaF.CrF3. 3. NaBF, + UF,: neither solubility of UF, nor formation of UO, was detected, even after adding B,O,.

5. NaBF, + 2LiF.BeF2, phase ratio -10:1, temperature 48OoC, Pyrex container. 6. NaBF, + MSRE salt (0.64 LiF, 0.29 BeF,, 0.015 ZrF,, 0.009 UF,), phase ratio "1:lO and 4:1, temperature 48OoC, Pyrex container.

4. NaBF, + HoF,: this rare earth was readily available, and its spectrum in molten fluorides is known; no measurable solubility was detected. 5. KBF, + B (amorphous) a t 650OC in fused silica: no significant solubility w a s detected; the finely divided powder coagulated with time. The following list summarizes the tests in which liquid immiscibility w a s observed; unless otherwise stated the container material was fused silica. Since UF, did not show spectrophotometrically any solubility in NaBF, or in KBF,, it was used for spiking other fluorides as a visual aid in the detection of phase separations. In some instances KBF, was used rather than NaBF, because its higher melting point w a s closer to the melting point of some fluoride mixtures.

1. NaBF, + blanket salt (0.71 LiF, 0.02 BeF,, 0.27 ThF,), phase ratio "l:lO, temperature 585OC. 2. KBF, + blanket salt (0.71 LiF, 0.02 BeF,, 0.27 ThF,), temperature 625OC. 3. NaBF, + BULT-4 (0.65 LiF, 0.30 BeF,, 0.01 UF,, 0.04 ThF,), phase ratio * l : l O , temperature 585OC. 4. NaBF, + 3LiF-UF,, phase ratio 1:30, temperature 48OoC, Pyrex container.

i j

13W. Hellriegel, Ber. 708, 689-90 (1937).

NaBF,

+ 2B,O,+BF,

+ NaF.B,06

Mole Fraction Phase NaF

LiF

BF3

=FZ

Bottom

0.3 91 0.165

0.139 0.43 5

0.419 0.129

0.052 0.271

TOP Bottom

0.365 0.1 56

0.174 0.448

0.406 0.138

0.055 0.257

TOP

0.373 0.110

0.164 0.545

0.432 0.044

0.031 0.300

(g NaBF,/g LiZBeF4) 3

TOP

Bottom

(1)

Indeed this is a commonly used method to generate BF,. Nonetheless, we decided to explore the effect of B,O, additions to immiscible liquids of NaBF, plus LiF-BeF,-ThF,-UF,. G a s evolution was detected in all tests where B,O, was added directly, and no extraction of UF, or precipitation of UO, w a s noted. The resulting borate phases were very viscous. Finally, in another series of tests 2LiF.BeF, was equilibrated with NaBF, in sealed nickel containers at 600OC. The capsules were quenched and the two phases sampled and analyzed. The results (Table 14.2) show appreciable distribution of all the components (NaF, LiF, BF,, and BeF,) between the two phases. In each phase

Liquid-Liquid Distribution Behavior for Li2BeF4-NaBF4 Mixtures at 6OO0C

Initial Phase Ratio

suggesting that the phases may be represented approximately as reciprocal mixtures of the ions Li', Na', BF,-, BeF,2-. In terms of these ions there is a tendency for the smaller ions, Li' and BeF4*-, to favor one phase and the larger ions, Na' and BF,', to favor the other phase.

Table 14.2.

It has been reported' that mixtures of NaBF, and B,O, react vigorously on heating to produce BF

Part 4.

Molten-Salt- Irradiation Experiments E. G. Bohlmann

Molten-salt breeder reactors will operate at fuel salt power densities of 300 to 700 w/cc in comparison with the peak level of 24 w/cc in the MSRE. Knowledge of the effects of such exposures on fuel salt stability and materials compatibility for both design and foreseeable offdesign conditions is essential to the success of the breeder program. The fates of the fission products are also of interest in terms of separations processing and possible poisoning due to accumulation on or in the graphite moderator. Molten-salt convection loop experiments are being operated in beam hole "-1 of the ORR to develop data pertinent to the breeder program. Examination of the second loop experiment was completed during this period. No unexpected materials problems were encountered except for the fact that the salt had wetted the graphite, presumably because of small amounts of moisture present in the gas used in loading, sampling, and draining manipulations. Fission product behavior

was well defined (good material balances for most isotopes of interest) and generally consistent with results of analyses on MSRE materials. A prototype of a third loop, modified to provide for better graphite examinations both pre and post irradiation, has been constructed, and components for the in-pile version have been fabricated of Hastelloy N modified to improve high-temperature ductility after irradiation. Work on the experiment has been discontinued because of budget limitations and investigations of effects of moisture levels in the cover and manipulatory gas. In the interim a program of capsule irradiations of fluoroborate coolant s a l t in spent HFIR fuel elements is in progress. Gamma fluxes comparable with those present in the MSRE heat exchanger are available in the center of such elements. Examination of the first capsule experiment showed no gross effects of accumulation of a dose of 2.5 x 10" r.

15. Molten-Salt Convection loop in the ORR E. L. Compere

H. C. Savage

J. M. Baker

the fuel salt, graphite, and metal in contact with fissioning fuel salt and cover gas and thereby determined the distribution of fission product activities in the loop components. A prototype of a third loop assembly has been designed, and component parts have been fabricated. Results of these additional postirradiation analyses and present design features of the third loop assembly are described below.

We have previously described the operation and postirradiation examination of the second in-pile molten-salt convection loop experiment operated in beam hole "-1 of the ORR.' During this report period, we completed additional analyses of

'E. L. Compere et at., MSR Program Semiam. Progr. Rept. Aug. 31, 1967, ORNL4191, pp. 176-93.

174

175 15.1 ISOTOPE ACTIVITY BALANCE (LOOP 2)

As reported previously,' the activity of a given isotope to be expected in the system at a particular t i m e was estimated by a detailed application of standard equations2 to the individual irradiation and inventory periods with adjustment for decay to final ORR shutdown for the experiment. Data received during the current period permit completion of the isotope activity balance. The activities of the fission products 'Cs, '44Ce, and 95Zr were used as internal standards to estimate the average flux received by the salt under the assumption that they were not appreciably lost from the salt. A mean flux to the salt of 0.88 x 10l3 was thus estimated. From this value, total activities of the various isotopes produced in the ORR loop experiment were calculated. The calculations described above provided a n estimate of the amount of each isotope to be accounted for. W e determined the total amount of isotope actually found in the system by dividing the loop into regions, analyzing a specimen from each region, and allocating a proportionate amount of activity to the region. For the various samples obtained from the loop, activity determinations for the 13 isotopes shown in Table 15.1 were made, as well as sensitive determinations of 235U. The activities have been totaled for each isotope under the categories of graphite, loop metal, salt sample lines, gas sample and inlet lines, and salt. These values, plus the estimated total activities calculated from inventory and irradiation history, are also shown in Table 15.1. It may be seen that over half (but generally less than all) the expected activity was accounted for in the cases of 99Mo, 132Te,95Zr, *'Sr, '37Cs, 14'Ce, '44Ce, 91Y, and 147Nd. A substantial proportion, although less than half, was accounted for in the cases of "Nb, 40Ba, and I. Inasmuch as iodine readily volatilizes from all solid samples (without doubt from powdered graphite in particular), it is to be expected that iodine determinations would be low.

'J. M. West, pp. 7-14 in Nuclear Engineering Handbook, ed. by H. Etherington, McGraw-Hill, New York, 1953.

Only about 11%of the '03Ru was found, although proper traps were used to recover any ruthenium compounds volatilized during t h e preparation of radiochemical samples. Molybdenum, tellurium, ruthenium, and niobium are a l m o s t entirely departed from the salt. These elements showed no dominant preference for graphite or metal but seemed to deposit on whatever surface was available. Short-lived noble gases appeared to have diffused appreciably into the graphite, as shown by the presence of daughter isotopes such as 89Sr, l4'Ba, and others. However, the major proportions of these, and almost all of other alkali, alkaline-earth, and rare-earth isotopes (all of which form relatively stable, nonvolatile fluorides), were found in the salt.

15.2 PENETRATION OF FISSION PRODUCTS INTO GRAPHITE AND DEPOSITION ONTO SURFACES The amounts of the respective fission product isotopes which penetrated the graphite to given depths were obtained from the samples shaved from the fuel channels. Data for each isotope and each common cut depth were summed over all the fuel channels. The value was corrected for the amount of the isotope in salt in the sample based on the amount of 235U which was found and on the concentration of the isotope in regular salt samples. This correction was appreciable only for those isotopes found principally in salt and was never dominant. The amount of an isotope in the graphite up to a given depth was then divided by the total amount of the isotope found in the loop, yielding the percentage of the particular fission product that penetrated to that depth in the graphite. Thus, about 43%of the fission product '32Te that was found was deposited within 1.3 m i l s of the graphite surface, and about 58%within the first 35 mils. Values so obtained are shown in Fig. 15.1, where the percentage of each fission product which penetrated to given depths is shown as a function of depth. The fission products 95Zr, 141Ce, 144Ce, 91Y, and 147Nd are found in low amounts in the graphite (between 0.4 and 3%)with nearly all that observed being found close to the surface.

176 Toble 15.1.

Comparison of F i s s i o n Product A c t i v i t y Found in Various Loop Regions w i t h Activity Produced in Loop

Percentage of loop inventory (calculated from power history) found in given region

Fission

Inventory,'

Yield (So)

Calculated (10" dis/min)

Isotope

Salt Samplesb

Gas Sample

Loop Graphite'

Metal

Sample Line

Line (Heated)

Gas Inlet Line ('=Old)

Foundd

<0.01 0.0001 <0.01 <0.01

74 11 54 35

<0.01 0.02 <0.01 0.01 <0.00001 0.02 <0.01 <0.0003 <0.01

71;106 33 34 76;92 125;65 91;99 72;80 112;126 77;69

Total

Isotopes That Leave the Solt

66-hr 99M0 39.7-day 03Ru 78-hr 132Te 35-day "Nbb

6.1 3.0 4.3 6.2

940 8,100 1,140 7,400

0.2 0.07;O. 09 0.8; 1.8 0.07

41.0 6.0 31.6 13.4

23.4 4.5 14.0 17.6

9.9 0.64 7.2 3.3

0.02 0.0005

0.04 <0.01

Isotopes That Remain i n the S o h

65-day "Zr 8.05-day 1311 12.8-day l 4 O B a ' 50.5-day " ~ r ' 58.3-day 'lye 30-year 1 3 7 ~ s e 32.8-day 141Ce 284-day 144Ce 11.1-day 147Nd

6.2 2.93 6.35 4.79 5.8 6.0 6.0 5.6 2.6

12,800 5,000

16,500 11,500 12,900 108

17,300 3,500 6,100

69; 107 30;31, 29;30 66;82 121;61 87;95 70;78 109; 124 72;64

0.4 0.3 2.3 8.7 3.0 1.8 1.3 1.3 2.0

0.6 1.6 1.3 1.1 0.6 2.2 0.4 0.5 0.7

0.8

1.1 0.9 0.7 0.6 0.02 0.7 0.7 1.9

<0.01 0.1 <0.01

0.1 0.003 0.15 <0.01 0.0003 <0.01

'Assuming a mean flux to salt of 0.88 X 1013 based on average of values from "Zr, 137Cs, and 144Ce in final salt samples. bEstimated for total salt based on each of two final samples. Niobium-95 based on earlier sample because production from '%, in salt during final period tended to remain in the salt, which was frozen most of the final two weeks. (Seventeen and twenty-one percent were found in the two final samples.) 'Corrected for salt content. dBoth totals are shown if the two salt samples differed appreciably. 'Isotopes with noble-gas precursors.

On the other extreme in quantity are 132Te,

"Mo, and lo3Ru. These isotopes showed 35 to 43% in the first 1.3 m i l s , 52 to 56% within 3.1 mils, and not much more at greater depths. These isotopes are thus indicated to deposit strongly from the s a l t onto the graphite surface but migrate only weakly if at all after deposition. Noble-gas isotopes would be expected to diffuse into g r a ~ h i t e ,so ~ that a more gradual concentration gradient of daughter isotopes resulting from their decay should be encountered. Thus,

3R. J. Kedl, A Model for Computing the Migration of Very Short-Lived Noble Gases into MSRE Graphite, ORNGTM-1810 (July 1967).

the total amount of daughter isotope within a given depth would continue to increase to appreciable depths. This pattern is evident for 89Sr, 14'Ba, 137Cs, and possibly 'lY. Iodine131 also exhibited such a pattern, although it was present in low quantity and amounts varied from sample to sample. Since iodine is not expected to be volatile from molten salt, either a volatile precursor is implied or migration could have occurred after the salt was frozen or during subsequent handling. Niobium-95 is present a t concentrations far in excess of its parent 95Zr, and its action does not appear coupled to that of 95Zr. Accumulated amounts increase rapidly for the first 5 m i l s from

177

100

5 20 a

z r

e W

$ I-

a W

I j

0 (0 v, LL -I

w 0.5 W

a FIGURES IN PARENTHESES SHOW PERCENTAGE ACCOUNTED FOR IN ISOTOPE BALANCE

10.2

0.4 0

DEPTH (mils)

Fig. 15.1.

Penetration of Various Fission Product Isotopes into Graphite a s a Percentage of Toto1 Isotope i n Loop.

178

11 to 32% and then increase regularly but more slowly up to about 3% a t 35 mils. Apparently this isotope was deposited relatively strongly on the gaphite b q also tended to penetrate inward, in this way exhibiting the pattern attributed above to gases. 15.3 STUDIES OF SURFACE WETTING OFGRAPHITE BYMOLTENSALT During postirradiation examination of the second in-pile molten-salt loop, evidence of surface wetting of the core graphite by salt was noted. W e wish to establish procedures to prevent this in the next loop experiment. W e have recently studied in a vacuum box the wetting characteristics of droplets of MSRE solvent salt on a platelet of MSRE graphite. We confirm generally the observation of Kreyger, Kirslis, and Blankenship4 that wetting is due to three-phase contact of gas, graphite, and molten salt at moisture levels as low as 10 ppm or lower. In an atmosphere of tank helium (below 4 ppm water), the droplets on melting did not wet the graphite. However, in a few minutes they developed a transluceht crust and in l to 2 hr after melting had slumped and spread spasmodically over the graphite surface. Another droplet was melted on graphite under good vacuum (less than torr, the present gage limit) and remained clear and round for about 20 hr. However, when tank helium was admitted to a pressure of 0.2 torr, the molten droplet promptly slumped and wetted the graphite surface. The surface wetting behavior of the s a l t thus appears to be very sensitive to very slight impurities, presumably water, in the helium. This could have been accentuated in these studies because of the small amount of salt and the large amount of gas. Further, the lack of wetting of the graphite by salt in the experiment under vacuum indicates that the prior condition of graphite or s a l t did not control the wetting behavior. Consequently, gas cleanup procedures are being tested a s the next step in studying the wetting phenomenon.

4 ~ J.. Kreyger et al., MSR Program Semiann. Progr. Rept. July 31, 1963, ORNL-3539, p. 125.

15d DESIGN OF A THIRD IN-PILE MOLTEN-SALT LOOP In-pile operation of the second molten-salt loop experiment was terminated because of a crack in the core outlet pipe5 with resultant fission product leakage. From evidence of postirradiation examination and evaluation of the effects of neutron irradiation on properties of Hastelloy N, it was concluded that this failure was caused by deterioration of stress rupture properties due to neutron irradiation.6 Therefore, the third loop assembly will be constructed of a modified Hastelloy N containing titanium as an additive for improving resistance to radiation-induced high-temperature embrittlement. Required shapes for construction of the loop components have been obtained from an ingot of such (-0.5% Ti, 2% W) Hastelloy N. We have also redesigned the graphite core section of the next loop to provide additional graphite surveillance specimens. The redesigned core has four holes of f/,-in.-square cross section instead of the eight '/,-in.-diam round holes for salt flow used for the two loops previously operated inpile. Graphite specimens with a t-in.-square cross section are inserted coaxially in each hole, resulting in a ,-in. annular salt flow channel, a s shown in Fig. 15.2. The flat surfaces of these graphite specimens will permit improved preirradiation evaluation of the graphite quality and postirradiation examination to determine interaction of graphite and salt. Metal locating brackets will be used to position the graphite specimens and will also serve a s surveillance specimens for metals of interest. In addition, we plan to install an adsorption trap immediately adjacent to the vapor space in the loop gas separation tank in an effort to trap and identify the gas-borne fission product species.

'/1

'E. L. Compere et el., op. cit., p. 182. 6H.C. Savage e t al., Operation of Moltenbalt Convection Loop in the ORR, ORNL-TM-1960 (December 1967). 7H.E. McCoy, Jr., e t al., MSR Program Semiann. Progr. Rept. Aug. 31, 1967, ORNL-4191, p. 18. 8E.L. Compere e t el., op. cit., p. 190.

179 ORNL-DWG 68-5563

'1, x

, , -HASTELLOY CAN

'12

GRAPHITE CORE

j .

OCOUPLE

It__ I I Fig. 15.2.

Graphite Core and Specimens, Molten-Salt

\ 1%6

In-Pile Loop.

n / '

x 3/n

- -

GRAPHITE SPEC IMEN

SALT FLOW ANNULUS

---_

DIMENSIONS IN INCHES

. 16. Gamma Irradiation of Fluoroborate E. L. Compere H. C. Savage J. M. Baker The use of sodium fluoroborate as a coolant in the Molten-Salt Breeder Reactor will involve the possibility of irradiation damage due to gamma rays and delayed neutrons from the fuel salt in the heat exchanger. Gamma rays might ionize and decompose sodium fluoroborate to produce boron, fluorine, and sodium fluoride. Neutrons in a "B(n,a) reaction would be expected to yield an alpha particle, lithium fluoride, fluorine, and sodium fluoride. The highenergy alpha particle could also ionize and decompose the sodium fluoroborate. The fluorine generated by either process would probably react with structural metals of the system to form their fluorides. Thus, a major effect of radiation damage to sodium fluoroborate by either mechanism would be a corresponding amount of corrosion. In the projected test of the 8%NaF-92% NaBF, eutectic mixture in MSRE, the neutron flux is negligible ("6 x 10' nv); therefore, the present studies are emphasizing the effects of gamma radiation. The gamma irradiation experiments are conducted in the central channel of spent HFIR fuel elements in the storage pool of the HFIR, where very high gamma fluxes are available (-8 x lo' r h r for elements four days old); this is comparable with the gamma flux in the MSRE heat exchanger. For the first gamma irradiation experiment 34 g of an NaF-NaBF, eutectic mixture with a melting point of *380째C was placed in a Hastelloy N capsule (0.93 in. OD x 0.78 in. ID x 3.5 in. long) containing a Hastelloy N corrosion test specimen. The salt was melted in the capsule in an argon atmosphere box, and then the capsule was welded shut. A capillary tube connected the gas space above the salt to a pressure transducer. A heater and three thermocouples were used to control and

monitor temperature. The capsule assembly was then placed in an aluminum container to isolate it from the pool water. Prior to the start of irradiation, the capsule containing +e sodium fluoroborate w a s vacuum pumped (5 x torr) for 16 hr at 150째C to remove any moisture and residual gas. The capsule was then sealed by closing the valve on the gas line and held a t 45OOC for 3 hr. On cooling to 2OoC a residual pressure of *5 torrs was observed. Analysis of the residual gas by mass spectrograph showed i t to contain N, (73%),Ar (13%),CO, (8%), H, (4%), 0,(2%), and H,O (0.2%). The capsule was then vacuum pumped a t 15OOC for 1hr, closed off, and operated for 68 hr at 600OC. The observed pressure at 6OOOC was 145 mm Hg close to the value anticipated for BF, vapor pressure according to Cantor's data. When cooled to 2OoC, residual gas pressure was below the limit of detection (* 5 torrs). The capsule assembly was then placed in the center of HFIR fuel element 34 on February 2, 1968. This element had been removed from the reactor on January 29, 1968, after operating for 23 days at 100 Mw. The sodium fluoroborate salt temperature reached 465OC from gamma heat alone (estimated to b e 0.25 w/g). Electrical heat was then added to bring the temperature to 600째C. On February 4, 1968, i t was observed that the electrical power required to maintain the capsule at 600OC was higher than anticipated and that the indicated wall temperature a t the center of the capsule (below the salt-gas interface) was below

'H. C . Savage e t e l . , NISRProgram Semiann. Progr. Rept. Aug. 31, 1967, ORNL-4191, pp. 194-95. ,S. Cantor, MSR Program Semiann. Progr. Rept. Aug. 31, 1967, ORNL-4191, pp. 159-61.

180

181 that of the upper section (gas space). These observations indicated the possibility of water in the container surrounding the capsule, and the experiment was removed from the fuel element. No evidence of water in the container was found by vacuum pumping while heating the capsule to 100OC. However, only the thermocouple located on the wall by the gas space of the capsule was s t i l l operating. The experiment was again placed in the center of the fuel element, and operation at 6OO0C, a s indicated by the above thermocouple, was continued. The experiment was removed from the fuel element on February 26, 1968, after 533 hr of irradiation. Observed pressure during most of the irradiation period ranged between 100 and 150 mm. For the 533 hr of irradiation the NaF-NaBF, s a l t accumulated an estimated absorbed dose of 1 x ev/g. Residual gas pressure in the capsule was * 10 torrs a t 5OoC and was found to contain a large percentage of hydrogen by m a s s spectrographic analysis (H2, 85%;N,, 12%; H,O, 2%; CO,, 1%; Ar, 0.3%; 0,, O.l%), which was confirmed by gas chromatography. The source of the hydrogen in the residual cover gas is not known; incompletely fluorinated fluoroborate (NaBF,OH) and in-diffusion of radiolytic or corrosion hydrogen from the water in the containment are possibilities. Upon opening t h e aluminum container can, it was found t o contain sufficient water to soak the magnesia insulating pad under the capsule. Since the capsule bottom was in contact with a watersoaked insulation pad, a severe temperature gradient undoubtedly existed from the top t o bottom of the capsule for most of the irradiation period. The two thermocouples at the center and near the bottom of the capsule were lost early in the run; thus the temperature at the bottom of the capsule is unknown. However, it is estimated that at least part of the NaF-NaBF, salt in the capsule remained frozen after t h e first two days of operation. Since no leaks were found in the aluminum can by pressure testing, it is assumed that water was inadvertently spilled into the can through the container access tube. The capsule was cut open, and the corrosion test specimen and s a l t were removed. No visual evidence of corrosion was seen. The salt, however, was discolored. Several sharply defined layers of salt of varying shades of pink from the bottom to about the center were observed with a

thin dark-brown layer on the bottom. The upper half of the salt was a mottled gray white. The layers are believed to be associated with solidification progressing upward from the bottom as the internal gamma heating diminished, with the moist insulation permitting significant heat removal. An unirradiated capsule assembly identical to the irradiated capsule was operated molten for 840 hr (mostly at 600OC) as a control for the irradiation experiment. Salt removed from this experiment was entirely white and rather frangible with none of the pink-brown discoloration observed in the irradiated salt. However, s m a l l brilliant-green crystals identified microscopically a s Na,CrF, were found near metal surfaces particularly at the bottom. For the unirradiated capsule, there was very little temperature gradient from top to bottom as was initially the case for the irradiated capsule. Samples of salt removed at various levels from the respective irradiated and unirradiated experiments were examined by mass spectrographic analyses. In each case, the bottom layer contained 10,000 ppm Cr or more and about 2000 ppm Ni, Fe, andCa, and 500 ppm of Mo, Ti, Mg, and Al, as well as smaller but appreciable amounts of a number of other impurities. The Na/B atom ratio ranged between 0.96 and 1.12 for most regions as might be expected for the sodium fluoroboratesodium fluoride mixture (nominally, Na/B 1.09). The region of the unirradiated material with green crystals showed an Na/B ratio of 1.26, whereas the corresponding dark-brown bottom region of the irradiated s a l t had a ratio of 1.09. The upper surfaces of both salts had a dark scum, and samples from this region in each case showed an Na/B ratio of -0.9. It is believed that a mild corrosion of Hastelloy N occurred in both cases, with a dark boron scum being produced. The corrosion products collected near the bottom. Hastelloy N coupons exposed in the irradiated and unirradiated t e s t s exhibited negligible attack other than a few surface stains. Microscopic examination revealed a general slight dulling - more evident at the liquid surface and above and somewhat more pronounced on the unirradiated coupon. The weight change of the unirradiated specimen corresponded t o less than 0.01 mil/year; for the irradiated coupon, the weight change was zero. The above findings are not indicative of any noteworthy effect of gamma irradiation on the

182 sodium fluoroborate salt or its compatibility with Hastelloy N. A second gamma irradiation capsule assembly has been fabricated, and it is planned t o irradiate this experiment to a higher total dose than the

i ii 1

f first experiment. Additional insulation at the top and bottom of the capsule has been added to ensure that the entire capsule is a t a nearly uniform temperature.

J 5.

Part

Materials Development H. E. McCoy, Jr.

i 1

The primary structural materials in molten-salt reactors are Hastelloy N and graphite. Our experience with the MSRE has demonstrated the excellent compatibility of these materials with fluoride salts in a nuclear environment. W e have a surveillance facility in the MSRE that allows us to periodically remove graphite and Hastelloy N specimens for examination. Both materials have been used in our fission product studies, and we have run mechanical pmperty tests on the Hastelloy N to follow the changes in the properties of the reactor vessel. Future MSBR's will probably utilize similar structural materials, but s o m e advances in technology are desirable. The requirements placed on the graphite are somewhat more severe in an MSBR

J. R. Weir

than in the MSRE. The graphite will be exposed to higher fluences and must have low permeability to gaseous fission products. We presently have a facility for irradiating graphite to a fluence of about 3 x lo2* nvt (>SO kev) per year, and we are investigating the dimensional changes that occur in various grades of graphite. We plan to reduce the permeability of graphite by surface sealing with metals or pyrocarbon. We have developed a modified Hastelloy N with superior resistance to irradiation damage and are studying its properties in detail. Some new salt compositions are of interest for MSBR's, and we are expanding our corrosion program to investigate their compatibility with Hastelloy N.

17. MSRE Surveillance Program 17.1 GENERAL COMMENTS H. E. McCoy, Jr.

W. H. Cook

We maintain a graphite and Hastelloy N surveillance program to follow the changes in properties with irradiation of the graphite and Hastelloy N used in the MSRE. A special fixture has been designed' so that these materials can b e exposed to the MSRE environment and removed periodically for examination. The graphite has been used principally for fission product deposition studies, since we presently do not have hot-cell techniques

'MSR Program Semiann. Pmgr. Rept. Aug. 31, 1965, ORNL-3872, pp. 87-92.

for measuring the very small changes i n physical properties that have likely taken place. The Hastelloy N is i n a form that can be converted to mechanical test specimens quite easily, and we have evaluated its postirradiation properties by tensile and creep-rupture tests. The s p a c e in the surveillance fixture is somewhat larger than required to follow the property changes in the same materials used to construct the MSRE. Hence, we have inserted some special types of graphite and Hastelloy N (1)to gain a better understanding of fission product behavior, (2) to obtain compatibility information, and (3) to determine irradiation damage information. We plan to remove another group of surveillance specimens from the MSRE about April 1, 1968.

183-

184 --.

17.2 EXAMINATION OF HASTELLOY N SPECIMENS FROM MSRE SURVEILLANCE FACILITY H. E. McCoy, Jr. We removed a second set of Hastelloy N specimens in June 1967. The specimens removed from the core had been at temperature for 5500 hr and had accumulated a peak thermal dose of approximately 4 x 1020 neutrons/cm2. The core specimens were modified alloys containing approximately 0.5% Ti (heat 21545) and 0.5% Zr (heat 21554). A stringer of standard Hastelloy N specimens from outside the reactor vessel was also removed. These specimens had been exposed t o the MSRE cell environment for about 11,000 hr and had accumulated a peak thermal dose of about 3 x lo’’ neutrons/cm2. The vessel specimens were made of the same heats used in constructing the MSRE.

-.-E

The creeprupture properties of the heats of standard Hastelloy N are shown in Fig. 17.1. Data are shown for (1) the first set of specimens removed from the core after 4800 hr with a thermal fluence of 1.3 x lozo neutrons/cm2 and (2) the second set of specimens located outside the vessel and removed after 11,000 hr with a thermal fluence of 3 x lo’’ neutrons/cm2. The specimens irradiated to the lower fluence generally had a larger rupture life at a given stress, although the differences become quite small a t low stresses (compare points for heat 5085). We have no explanation for the superior rupture life of heat 5081. The minimum creep rates are compared in Fig. 17.2 for several heats of Hastelloy N in theirradiated and unirradiated conditions. As observed previously,2 this parameter is not affected. 2H.E. McCoy, Jr., and J. 96 (1968).

R. Weir, Nucl. Appl. 4(2), ORNL-DWG 68-6576

0 0

E cn

I11111

to -

-to-’

I1111

MSRE SURVEILLANCE VESSEL CORE HEAT

Fig. 17.1.

5065

m 3XtO” 4t:OO

io0

A t.3Xt020 THERMAL FLUENCE 4800 TIME AT 650% (hr)

to‘

-l+l

I II

2 5 to* RUPTURE TIME (hr)

to3

Postirradiation Creep-Rupture Properties of MSRE Surveillance Specimens at

65PC

185

‘

from 3 x 1019 to 1.3 x lo2’ neutrons/cm2. However, the data at higher strain rates indicate a large reduction in the fracture strain as a result of the increased fluence. The tensile properties of t h e air-melted surveillance materials are shown in Table 17.1 for test conditions of 25 and 650OC. The fracture strains at 65OOC are considerably lower at the higher fluence. Thereduction in ductility at 25OC is thought to b e associated with carbide precipitation in the alloy and is probably a function of both the thermal history and the fluence. The property changes that have been noted in Hastelloy N as a result of service in the MSRE are similar to those noted for Hastelloy N irradiated in an inert-gas environment, thus reflecting the excellent compatibility of the alloy with

The fracture strains of the vessel heats are shown in Fig. 17.3. The scatter band was determined from a large number of data points, with heat 5065 being on the lower s i d e and heat 5085 being on the upper side. The minimum fracture strain at a strain rate of about O.l%/hr clearly exists for materials exposed to irradiation for about 1000 hr at 65OoC, whereas t h e data for the surveillance specimens do not suggest the existence of this minimum. The fracture strains of most of the surveillance specimens fall between 1.5 and 2.5%; however, heat 5081 is somewhat anomalous and exhibits higher fracture strains with the trend of increasing fracture strain with decreasing creep rate. The results on heat 5085 indicate that the fracture strain may be decreased only slightly by increasing the fluence

ORNL-DWG

68-6577

’

= 30 ki

I I Ill1

MSRE SURVEILLANCE VESSEL

CORE

20 3x40’’ 41,000

A 1.3XtOe0 4800

HEAT 5065 5085 5084 THERMAL FLUENCE TIME AT 650°C (hr)

0 2

(0-2 2

fo-‘

4oo

40’

MINIMUM CREEP RATE (%/hr)

Fig, 17.2.

Variation of Minimum Creep Rate of Hastelloy N

Tests a t 65OoC.

MSRE Surveillance Samples i n Postirradiation Creep

186 ORNL-DWG 68-6578

15 44 13

4I ll1lI1l

I I lllllll I

MSRE SURMILLANCE

t”%’

-s

HEAT

CORE

5065 5085 A 5081 4 . 3 ~ 1 0 ~THERMAL FLUENCE TIME AT 650°C (hrl 48 I m

- 9

!iJ w

9 7

E 6 5

3 2

0 10-3

Fig. 17.3.

10-2

16‘

100 STRAIN RATE (%/hr)

Variation of Fracture Strain with Strain Rate for MSRE Surveillance Specimens at 650OC.

fluoride salts. The fracture strain and the rupture life at a given stress are both reduced, but not to levels that should prevent the continued successful operation of the MSRE. The property changes noted in increasing the fluence from 3 x 10’’ to 1.3 x lo2’ neutrons/cm2 w e e quite small. The vessel should not reach a thermal fluence of 1.3 x lo2’ neutrons/cm2 until after 150,000 Mwhr of operation. Thus, if the effects of thermal aging are not adverse, the properties of the vessel should not deteriorate much below those observed

for the material irradiated to 1.3 x IO2’ neutrons/ cm2. The titanium- and zirconium-modified alloys that were removed from the core were heat treated to obtain a very fine grain size. At the time these were inserted, it was thought that the fracture strain increased with decreasing g a i n size. However, subsequent t e s t s showed that the reverse was true. The postirradiation properties of these materials were quite comparable with those noted for the standard Hastelloy N.

,--

187 Table 17.1.

0.05 d n - l e

650

0.002 dn-'

34.6 25.8

11.4 Core

0.05 mb-'

(3 x

Heat 5081

0.002 min-'

0.05 min-'

0.002 min-'

neutrons/cm 2 )

42.5 22.4

12.0

(1.3 x lo2' neutrons/cm 2 )

27.3

650

13.7

astrain rate.

Surveillance Specimens

Heat 5085

Vessel

(W)for MSRE

Heat 5065

Test Temperature (OC)

T e n s i l e Elongotion

9.4

38.7 42.6 14.3

9.0

18. Graphite Studies 18.1 PROCUREMENT OF SPECIAL GRADES OF GRAPHITE

and molybdenum, (3) graphite-to-metal joining studies, and (4) fabrication of test assemblies. Graphite for the initial phases of i t e m s 1 through 3 has been obtained in reasonable quantities. Some graphite for item 4 is on order. The graphite manufactured by Poco Graphite? Inc., continues to be our reference graphite â&#x201A;Źor the sealing and joining work because it (1) has reasonable uniformity? (2) has pore entrance diameters

W. H. Cook The current studies on graphite for molten-salt breeder reactors include (1)the determination of the physical and mechanical properties before and after irradiation, (2) sealing research with pyrolytic carbon and chemical-vapor-deposited niobium

Table 18.1.

Grade

Receipt and U t i l i z a t i o n of Special Grades of Graphite Received Since January

Source

mPe

Density

Nominal Dimensions

Pieces

1, 1968

Utilization

(in. 1

(g/cm3) 9950

Speer"

Near isotropic

1.71

2 i x 4 x 8

9949

Speer"

Near isotropic

1.71

2gX4X8

9948

Speer"

Near isotropic

1.92

2 i x 4 X 8

AXF

pocob

Isotropic

AXF-SQ

Poco

Isotropic

AXF-SQBG

Poco

Isotropic

ATJ-S

CPDC

Near isotropic

ATJ-SG

CPD

Near isotropic

Grade 9948 will b e evaluated in HFIR irradiation studies

l6diam x 6

200

1.81

4 diamx18

Graphite-to-metal joining studies

1.89

1 i x 4 x 6

Graphite-to-metal joining studies

8 diam x 12

Made by a special process which has potential for fabrication of MSBR shapes; evaluation to include irradiation in the HFIR

8 diam x 12

Made by the ATJ-S process using a more isotropic filler material, Gilsocarbon flour; evaluation to include irradiation in the HFIR

'"1.82

For sealing studies \

-1.83

1.81

"Gratis material from Speer Carbon Company. 'Poco Graphite, Inc. 'Carbon Products Division of Union Carbide Corporation.

188

189

that are in the desired range (approx 1 p), and (3) is readily obtainable in s m a l l but useful s i z e s for these studies. The grades and disposition of graphite obtained recently are summarized in Table 18.1.

18.2 POROSITY CREATED I N GRADE AXF GRAPHITE BY OXIDATION PRETREATMENT W. H. Cook

ment was used to remove fingerprints or other contaminants from graphite test pieces before they were sealed by the vapor deposition of metals. Specimens given this pretreatment were unusually permeable and were not sealed by reasonable metal deposits. Microstructural examination showed that the normal, relatively uniform pore structure with entrance diameters approximately 1 p now had a secondary set of pores with some more than a hundred times as large as the basic pores. This was a new lot of grade AXF graphite in stock sizes of ,-in.-diam, 6-in.-long rods; so

The porosity of grade A X F graphite' used in coating and sealing studies was increased to an unacceptable extent when the specimens were treated for hr in air at 60OOC. This pretreat-

'/1

'/2

'Manufactured by Poco Graphite, Inc., Garland, Tex.

'The original pore entrance diameters were measured with a mercury intrusion porosimeter, and the secondary pores were measured with a microscope.

PHOTO 91877

1 ._ c I

Fig. 18.1. I

Microstructures of Grade AXF Isotropic Graphite Specimens from t6-in.-diam

Rods. (a) As received,

( b ) after exposure for hr i n air at 6OO0C, and (c) machined to 0.400 in. OD x 0.100 in. I D and then given the same treatment given the specimen shown in b. The normal pores are not resolved i n these microstructures; they are usud

ally associated with the dark-gray phase of the structure, but assuming that they are greater than 1 /L (the approximate diameter of their entrances), they would be only approximately 0.004 in. on the photomicrographs. However, the oxidized voids (pores) are clearly visible as the large black circular spots. 1 0 0 ~ . No etchant.

190 a series of tests were made to determine if the secondary set of pores was the result of the pretreatment or was characteristic of the new lot of material. Control and test specimens were taken from three random rods. Test specimens from each rod were given the standard pretreatment, and duplicate specimens were given the same anneal in an argon environment. The normal grade AXF graphite microstructure, a s shown in Fig. 18.1~1,was observed for all three controls and the test specimens treated in argon. All those given the standard oxidation pretreatment developed the large secondary s e t of pores shown in Fig. 186 and c. The voids were a t irregular depths ranging from 20 mils to completely through the ,-in.-diam specimens. The oxidized void (pore) sizes graded from large to small from the external surfaces to inside the specimens. The samples used in the sealing studies were 0.4 in. OD x 0.1 in. ID x 1.45 in. long, and the large voids introduced by oxidation generally penetrated the entire wall. The pretreatment in air will not be used in future work. Surface contamination will be reduced by more careful handling, followed 'by anneals in argon or vacuum environments.

'/1

18.3 X-RAY STUDIES ON GRAPHITE

are closely related to i t s resistance to radiation damage. W e are concerned with the determination of lattice parameters (in both the c and a directions), preferred orientation, and crystallite size (Lc). Quantitative data are difficult to obtain and are strongly affected by equipment and analytical procedures. We have begun measurements of the crystallite sizes and lattice parameters of nearly isotropic polycrystalline graphite to develop an accurate approach for data gathering and analysis and to characterize these particular graphites. Preliminary results are given in Table 18.2. The lattice parameter measurements are reasonably consistent and agree well with those obtained by other investigators. The values for the a parameter are reasonably constant, but the values for c vary. These variations probably reflect differences in crystal perfection. The crystallite s i z e measurements were determined independently from the broadening of the (002) and (004) diffraction peaks. These determinations should give identical values for Lc, but the data in Table 18.2 show that this was not the case. The basic Scherrer3 equation used in analyzing these results involves an instrumental correction factor. Again different results were obtained using the correction factor suggested by Schemer3 and the one later proposed by Warren.3

0. B. Cavin X-ray diffraction is a useful technique for examining many of the properties of graphite that Table 18.2.

3H.P. King and L. E. Alexander, X-Ray Diffraction Procedures, pp. 491, 500, Wiley, New York, 1954.

X-Ray Measurements on Several Graphites

Sample

AXF

2.462

6.769

189

116

250

156

AXF-SQBG

2.464

6.763

196

130

262

183

AXF-5QBG3

2.463

6.749

224

142

313

205

BY-12

2.463

6.737

272

204

406

342

RY-12-00029-32

2.463

6.732

250

198

361

330

RY-12-00031-34

2.464

6.736

272

191

406

312

CGB base stock

2.464

6.731

277

235

417

424

*usingWarren's instrumental correction. Schemer's instrumental correction.

191

These apparently contradictory results are due to the fact that the Scherrer equation assumes that the only source of line broadening is that due to the crystallite size. In addition t o crystallite size, there are a t least six other factors which may contribute to line broadening: (1)compositional variations, (2) small crystallite size, (3) inhomogeneous strains, (4) stacking faults, (5) range of wavelengths in incident beams, (6) instrumental factors, and (7) low absorption in the sample. Thus the Scherrer equation leaves us with a rather meaningless summation of these factors. Therefore, in the future we plan t o use a more accurate Fourier approach for the analysis of peak shapes rather than using only peak breadths. In this manner we can separate more of the factors and obtain a more realistic measurement of the crystallite size Lc.

18.4 GAS IMPREGNATION OF GRAPHITE WITH CARBON R. L. Beatty

D. V. Kiplinger

One of the requirements for graphite t o be used in a molten-salt breeder reactor is a surface with low permeability to prevent xenon absorption. Calculations suggest that a helium permeability of less than lo-' cm2/sec a t the graphite surface will be required to keep the xenon concentration in the core t o the desired level. Since commercially available graphites usually have helium permeabilities some five or s i x orders of magnitude higher than this required level, it is necessary t o consider coating or sealing the graphite surface by some means. Present considerations are to seal the graphite surface either with pyrolytic carbon or with a m e t a l such as molybdenum or niobium. If a m e t a l is employed, the amount must be strictly limited to avoid excessive neutron absorption penalties. Though no such volume limitation per se exists in the case of pyrolytic carbon, there is a limitation on the means of employing it. Since the crystalline character of the pyrolytic carbon deposit is likely to differ markedly from that of

the base graphite, the neutron-induced dimensional changes of the two materials will probably be different also. Therefore, if the pyrolytic carbon is applied simply as a coating, i t may be subject to spalling during irradiation. This problem may be circumvented, however, if the carbon is deposited in the pores near the surface rather than on the surface We are exploring methods of gas impregnating graphite with carbon. A technique for providing a very low-permeability surface seal by immersing a graphite specimen in a fluidized bed has been d e m ~ n s t r a t e d but , ~ this method may be impractical for large bodies such as are required in a moltens a l t reactor. Another method we a r e studying involves forcing the carbon-bearing gas radially through a heated specimen, up a thermal gradient. The thermal gradient is achieved radially in the specimen by heating i t with a 1.2-kw, 450-kc induction generator while cooling the inside by means of an axial ,-in. water line. W e selected the specimen geometry t o meet the requirements of the HFIR irradiation facility (i.e., a hollow right circular cylinder of nominal dimensions 0.400 in. OD, 0.125 in. ID, and 0.500 in. land. The base stock is Poco graphite grade AXF. During impregnation the specimen is clamped between spring-loaded mullite tubes which seal against the ends with Grafoil gaskets. The carbon-bearing gas and a diluent are forced into one of the mullite rams and can exit only by passing radially through the graphite. Sources of carbon being considered are methane, propylene, and butadiene. We checked 30 as-machined AXF specimens for leak tightness and found great variation. We obtained vacuum levels ranging from 4 to 120 p Hg when the specimens were pumped with the roughing pump of a Veeco leak detector. Preliminary impregnation experiments indicate that this range of as-received permeabilities may not be important for carbon impregnation, but we will examine this point in detail. With the above techniques we have sealed two specimens so that the helium leak rates measured by the Veeco detector are less than half that of our 4.5 x lo-' s t d cm3/sec standard leak. Though sealing at the ends of the specimen is a problem with this

'/1

4C. S. Barrett and T. B. Massalski, Structure of Metals, pp. 155, 454, McGraw-Hill, New York, 1966.

5MSR Program Semiann. Progr. Rept. Aug. 31, 1967, ORNL-4191, p. 212.

192 technique, we believe it is promising and can be used either alone or in conjunction with surface diffusion impregnation. We are also studying a system which cycles between vacuum and hydrocarbon atmosphere while the graphite substrate is heated to temperatures of 800 to 1300OC. The variables being considered in this system are temperature, cycle time, concentration of hydrocarbon gas, system pressure, and original substrate porosity. The experimental apparatus includes a silica tube with the test specimen suspended inside by a graphite thread. The silica tube is closed a t each end by a solenoid valve, the bottom one being connected to a vacuum pump and the top one to a source of hydrocarbon gas. The two solenoids are controlled by a pulse timer that automatically switches the system between vacuum and hydrocarbon atmosphere a t preset intervals. The first carbon source we used with this system was a mixture of 1.3-butadiene in argon (10% C,H, + 90% At). We found that by using argon a s the diluent, we could operate from 150 to 2OOOC hotter with the same power setting than with helium as the diluent. We later found that the t i m e required to s e a l a sample can be reduced by using undiluted butadiene. W e selected butadiene as the carbon source because it decomposes a t a lower temperature than hydrocarbons such a s methane and propylene and has a high carbon to volume of gas ratio. The impregnation process appears to be quite sensitive to temperature. A t temperatures above 1000°C the processes studied tend to coat the samples rather than impregnate them. A t temperatures in the range of 800 to 950°C, undiluted butadiene appears to penetrate the pores, but the deposit is probably one or more of a family of highly viscous hydrocarbons. These tars, when baked out, release hydrogen, which tends to reopen the pores to some degree. Several samples were prepared by pulsing between 5 sec of vacuum (<25 in. Hg) and a pressure of 10 psig of undiluted butadiene in the temperature range 800 to 900OC. We found that samples could be made leak-tight after a total hydrocarbon contact t i m e of 1 to 2 hr. The expression “leak-tight’’ here means a leak rate less than the standard used, which was 4.5 x std c m 3 / s e c of helium. These samples were then heated to 13OO0C, cooled, and rechecked for

leakage. In a l l samples studied, except one sample impregnated a t 900°C, the heat treatment was found to increase the leak rate. The sample impregnated a t 900°C that remained impermeable to helium after heating at 130OOC was heated to 200O0C for 5 min. The dimensions remained the same with a very slight loss in weight, and the helium leak rate was 200 x lo-* std cm3/ sec after 20 min of helium purge. This specimen was then reimpregnated a t 900°C and heated a t 200O0C. This t i m e no leak was detected by the Veeco instrument, which means that the helium permeability was less than cm2/ sec. There was no measurable thickness of coating on the outside of the specimen as a result of the two impregnations. Thus we feel that the pulsing technique for gas impregnation of graphite with carbon is the m o s t promising technique we have tried. Further, this technique should be relatively easy to scale up and is equally applicable to either hollow tubes or solid rods.

c, a

18.5 GRAPHITE SURFACE SEALING WITH METALS

W. C. Robinson Detailed analysis of the molybdenum-coated R-0025 graphite samples reported previously6 indicated that there was no direct connection between the thickness of m e t a l coating and the helium permeability of the coated sample. The first experiment, M-Mo-1, which had a coating of approximately 0.05 mil, had a permeability too small to be detected by a Veeco leak detector. The other eight samples had coatings up to 0.275 mil and would not pump down low enough to measure the permeability. We decided that this lack of correlation must be due to the inhomogeneity of the R-0025 starting material. Twenty samples of AXF graphite were obtained. Prior to coating, each sample was tested by attempting to pump a vacuum on it with the pumps of the Veeco leak detector. Pressures down to 0,001 torr can be measured. After coating, the sample was again pumped and the pressure was recorded. If the pressure was 50.001 ton, an attempt was made to measure the permeability. c

‘MSR Program Semiann. Progr. Rept. Aug. 32, 1967, ORNL-4191, p. 211.

193 T a b l e 18.3.

Experiment

No.

Gas Flow Rate (cm3/min) MoF6

M-Mo-10 50 M-Mo-11 50 M - M o - ~ ~ ~ 50 M-MO-13 50 M-Mo-14 50 M-Mo-15 50 M-Mo-16 50 M-Mo-17 50 M-Mo-18 50 M-Mo-20 M-Mo-21 M-Mo-22 50 M-Mo-23 50 M-Mo-24 50 M-Mo-25 50 M-Mo-26 50 M-MO-27 50 M-Mo-2 8 50

Surface Sealing of Graphite by Metal Coatings

Temperature

Pressure

(OC)

(torrs)

Time

Coating Thickness

(

Precoat Vacuum

Final Vacuum

Permeabilitya

~ (mils)

(toms)

(torrs)

(cm3/sec)

<1.5 x 10-10

H, 800 800 800 800 800 800 800 800 800 Oxidized Oxidized

700 700 700 700 700 700 700 700 800

800

800 800 800 800

800 800

800

5 5 10 5 5 10 10 10 5

5 10 5 5 10 5 30 10 5

0.033 0.073 0.049 0.046 0.091 0.102 0.520 0.066 0.080

0.120 0.115 0.010 0.075 0.120 0.135 0.155 0.082 0.001

0.110 0.082 0.017 0.065 0.020 0.022 0.01 0.05 <0.001

5 5 10 5 5 10 10

5 10 5 10 20 10 20

0.081 0.252 0.136 0.161 0.374 0.055 0.485

0.050 0.075 0.100 0.090 0.095 0.090 0.100

0.010 0.001 0.010 0.001 0.001 0.075 0.001

aA helium leak rate of <lo-* cm3/sec corresponds to a surface diffusion coefficient of about bFingerprint. 'Would pump down sufficiently to leak check, but leak was too large to measure.

The experimental conditions and results are shown in Table 18.3. The coatings reduced the permeability of all samples except the third, M-Mo-12, which was more permeable after coating. On this sample the coating outlined a fingerprint very clearly, and this contaminant probably contributed t o the poor coating characteristics. All subsequent samples were annealed for $ t o 1 hr a t 6OOOC in air before coating to remove surface contaminants. For all the remaining samples (M-Mo-13 through M-Mo-28) the vacuum achieved on the coated samples seems to be a function of both the precoat vacuum and deposit thickness. Only five samples had permeabilities low enough to be pumped to a pressure less than 0.001 torr. Permeability within to 1.5 x lo-'' cm3/sec can the range of 7 x be measured with the leak detector. In M-Mo-23 the measured helium permeability was 3.15 x

lo-'

3.15 x >7 x > 7 x 10-6 > 7 x 10-6c cm2/sec.

cm3/sec. This required a deposit of 0.252 mil on

a sample that pumped to 0.075 torr originally. In M-Mo-18 the permeability of the control sample was too low to measure on the Veeco leak detector (i.e., less than 1.5 x cm3/sec). This sample had only 0.080 m i l of molybdenum, but the sample pumped to a vacuum of 0.001 torr before coating. For M-Mo-25, M-Mo-26, and MMo-28, the coated samples could be pumped to 0.001 torr, which is sufficiently low to leak test, but they leaked too much to be measured (i.e., the permeability was greater than 7 x c m 3/se~). The test results are plotted in Fig. 18.2 to depict the pressure drop obtained before and after coating as a function of deposition thickness. This correlation indicates that thinner deposits are required at 700 than at 8OO0C, particularly on samples that will not originally pump below 0.080

194 ORNL-DWG 68-3986

0.02

0.04

0.06

(PORG-PF,,,&)

Fig. 18.2.

0.08 0.40 REDUCTION PRESSURE (torr)

0.42

a44

Change in Pressure Drop Across a Graphite Specimen with Coating Thickness.

to 0.090 torr. On samples that will pump below 0.075 torr originally, both coatings are equally sound, and less than 0.1 m i l Mo is sufficient to reduce the final vacuum to the limit of the Veeco pressure gage. On samples that will not pump below 0.095 torr, over 0.3 m i l of molybdenum is necessary for an 8OOOC deposit. A 700OC deposit may .be capable of sealing the samples with 0.1 m i l of molybdenum if the sample will originally pump below 0.110 torr. If the original samples will not pump below this range, then our data indicate that they cannot be sealed with a reasonable amount of molybdenum. The validity of the plot is being checked with 20 more samples of the air-fired gaphite, which are on hand. When one compares M-Mo-18 with the other samples, it is apparent that the original permeability is very important. A recent study by Cook (Sect. 18.2) has revealed that the 600OC air fire can open large pores in the graphite samples. These large pores may be more difficult to s e a l than the pores in the as-received sample. The 6OOOC air fire will therefore be discontinued in the hope that the argon-fired samples will have better permeability prior to coating and will thus require thinner coatings. New coating experiments on AXF graphite with an argon heat treatment will be compared with the air-treated samples to test this hypothesis. Examination of a l l the coatings revealed that the metal was principally bridging the external pores of the graphite, with only the most open

and shallow voids being filled. An attempt was made to exert a vacuum on the inside of the graphite cylinder during deposition in an effort to pull the reacting gases into the graphite pores and possibly achieve the desired reduction in gas permeability with less molybdenum. This technique is called the differential pressure CVD coating process. The first attempts to do this were not successful because we were not able to obtain a vacuum seal around the graphite during heating and therefore pulled only a partial vacuum of 1 to 2 torrs inside the graphite sleeve. However, as shown in Fig. 18.3, some penetration was achieved. The Welding and Brazing Laboratory will make a braze joint between the graphite cylinder and a molybdenum tube, which should solve this problem. Thus we will obtain a lower vacuum inside the graphite cylinder and perhaps obtain deeper penetration of coatings within the pores in subsequent coating experiments. A series of 7OOOC argon anneals were performed on selected samples of molybdenum-coated graphite to evaluate the tendency of spalling during thermal cycling and to see if the coating totally converted to a carbide. The samples were heated to 7OOOC for 17 hr, furnace cooled, and then given three more 48-hr anneals a t 7OOOC with cooling to room temperature between anneals. No tendency to spa11 was observed on any sample. After the final anneal the samples examined still had 10%molybdenum carbide or less, a s estimated using x-ray diffraction techniques.

â&#x20AC;?

195

Fig. 18.3.

Penetration of Molybdenum Deposit into Graphite, M-Mo-30.

18.6 GRAPHITE IRRADIATION PROGRAM

C. R. Kennedy

W e have initiated graphite irradiation experiments in the target rod positions in the core of the HFIR. This facility is being used to obtain high fluences in a relatively short time. The first two experiments, described previously,' have completed a single cycle in the HFIR and have been examined. The purpose of these first two experiments was primarily t o confirm the heating rates and the design of the experiment. The design utilized tungsten susceptors to compensate for the lower nuclear heating near the ends and thus obtain an axial temperature distribution of 690 t o 730OC. The reactor itself is quite stable, and temperature variations over a cycle or from cycle to cycle should be less than 10째C. The results of the first series show that the peak nuclear heating was 33.6 w/g, compared with a n estimated 35 w/g. The axial 'MSR Program Semiann. Progr. Rept. Aug. 31, 1967, ORNL-4191, p. 215.

500~.

gradient did not follow the estimated cosine function but was slightly distorted, with a ratio of end heating t o peak heating of 0.56. Calculations based on flux monitors of stainless steel doped with cobalt indicate a flux of 1.3 x 10' neutrons cm-2 sec-' (E > 50 kev). The results of the first two experiments suggested a few minor design changes. These changes have been made, and two long-term experiments have been constructed and inserted in the HFIR. These experiments will be removed from the reactor after five cycles with a n exposure of about 1 x loz2 neutrons/cm2 (E > 50 kev). The specimens will b e evaluated by measurements of physical dimensions , permeability, elastic modulus, and other physical parameters. Selected specimens will be subjected to low-angle x-ray scattering and electron microscope studies. The materials included in the two experiments are listed in Table 18.4. The program of evaluating candidate materials is still in its beginning stages, and many of the more promising materials have not been delivered. This affords a chance t o include grades AGOT,

196 Table

18.4.

Source

Materials Used in Graphite Irradiation Experiments

Remarks

Grade ~~~~~

Poco

AXF

Base stock

Poco

AXF-3000

Base stock heated t o 30OO0C

Poco

AXF-5QB

Impregnated base stock

Poco

AXF-5QB-3000

Impregnated b a s e stock heated t o 300O0C

Y-12

BY-12

85% graphite flour-15% fines-pitch,

Y-12

RY-12-31

85% graphite flour-15% fines-Varcum,

Y-12

RY-12-29

85% graphite flour-15% Thermax-Varcum,

Y-12

Natural flake

Natural flake-pitch, 30OO0C

GLCC

H315-A

Pipe material similar to GA materials, isotropic

GLCC

H364

Fine grained, high fracture elongation, isotropic

ucc ucc ucc

1425

Pipe material similar t o grade CGB

ATJ-S

An improved ATJ grade

AGOT

Material used in PNL irradiations

ucc

Hot-pressed pyrographite

T o be used for x-ray scattering experiments

UKAEA

UK isotropic

Material used in UK irradiations

OWL

Pyrocarbon-coated Poco

Surface sealed with pyrocarbon

UK isotropic, and H315-A in the HFIR irmdiations for direct comparison with PNL, UK, and GA irradiation results. Those grades demonstrating poor irradiation resistance will be removed and replaced during the recycle periods. The Y-12 grades are a controlled fabrication series to isolate material variables affecting the growth rates and ultimate irradiation resistance. These four graphites are unique in that although several different starting materials were used, the physical properties and their isotropics are about the same. This will allow us to a s s e s s the effects of starting materials with a minimal change in other parameters. The Poco series is included to gain a better understanding of the exceptional dimensional stability demonstrated by this material.* The coated materials were included to obtain some indication of the ability of these materials to adhere to a graphite body. The coatings were not applied by an optimized process, since our coating work is in its early stages.

'Yosh Kawa, BNWL, private communication.

30OO0C 30OO0C 30OO0C

18.7 NONDESTRUCTIVE TESTING STUDIES

H. L. Whaley

R. W. McClung

18.7.1 Graphite Ultrasonic Velocity Measurements

H. L. Whaley Progress has been made toward the measurement of ultrasonic longitudinal (V,) and shear (V,) velocities for determining the elastic properties of proposed MSBR graphites. Longitudinal measurements have been performed with both commercial transducers with oil coupling and with thin ceramic disks bonded to the samples with Salol (phenyl salicylate). Shear crystals require a solid couplant and have been bonded with Salol. One major problem arises from the configuration of the specimens (0.400-in.-diam cylinders, 0.500 in. long, with a 0.120-in.-diam axial hole). The concentric smooth surfaces result in mode conversion of the ultrasound and spurious reflections that make identification of the desired signals difficult. To help overcome this situation, bulk pieces of

197

u *

one kind of the graphite (Poco AXF) have been obtained t o establish approximate time intervals required for the ultrasonic pulses to travel known distances. Unfortunately, bulk pieces are not available for all types of the graphite to be measured. To avoid activation in the reactor due t o prior contamination by a couplant, ultrasonic measurements will be performed initially on pieces taken adjacent to the irradiation samples and then, after exposure, on the irradiation samples themselves. About 25 of the cylinders (machined from Poco AXF and Great Lakes H315-A graphite) have been obtained for studying the uniformity of the materials and the reproducibility of bonding and measuring techniques. The anisotropy of grade H315-A was detected and may pose some problems.

18.7.2 Low-Voltage Radiography R. W. McClung Fig. 18.4. Low-Voltage Radiograph of a Graphite

Low-voltage radiographs were made for the Reactor Chemistry Division on several sections of graphite which had undergone s a l t impregnation testing in an irradiation field (Fig. 18.4). The radiographs showed the presence of high-density material (compared with graphite) in obvious cracks and in other areas which may have been cracks oriented so that a characteristic shape could not be observed. The inspection was a further demonstration of the low-voltage radiographic technique:

1. to achieve high-sensitivity radiographs on thin sections of low-density material,

Slice Taken from ORR Natural Circulation Loop No. 2. Uranium-bearing fluoride salt passed through the eight holes.

The dark areas indicate the presence of the

high-density salt.

The diameter of the piece is

2 in.

2. t o detect minute amounts of higher-density material in the low-density matrix, 3. to perform these evaluations despite the specimen being radioactive due to irradiation testing.

. 19. Hastelloy N diated condition. In the unirradiated condition the fracture strain at temperatures <500째C is only weakly influenced by the strain rate. At SOOoC the fracture strain decreases at very low strain rates, due to the transition from transgranular to intergranular fracture. At 65OOC the fracture strain is very dependent upon strain rate, with the lower ductilities being characterized by intergranular fracture. A t 760 and 85OoC the fracture strains are again quite high, with only a slight dependence on the strain rate.

19.1 INFLUENCE OF STRAIN RATE ON THE FRACTURE STRAIN OF HASTELLOY N H. E. McCoy, Jr. Previous results have indicated that the fracture strain of irradiated Hastelloy N is a function of strain rate.' n 2 We have now obtained sufficient test results to see the influence of this variable more clearly. The variation of fracture strain at the strain rates normally encountered in tensile tests, 2 x to 2 min-' , is shown in Fig. 19.1 for a typical standard air-melted alloy in the unirra-

The results from specimens that were irradiated in the MSRE are shown in Fig. 19.2. A t 4OOOC the fracture strains are almost independent of strain rate, although the strains are slightly lower than those in Fig. 19.1 for unirradiated materials. At SOOOC a distinct effect of strain rate is obvious, the decreased ductility again being a result of the

'MSR Program Semiann. Progr. Rept. Feb. 28, 1967, ORNL-4119, p. 103. 2MSR Program Semiann. Progr. Rept. Aug. 31, 1967, ORNL-4191, p. 200.

ORNL-DWG 67-245%

0.001

0.01

0.4

STRAIN RATE ( m i d )

Fig.

19.1.

Influence of Strain Rate on the D u c t i l i t y of

MSRE Surveillance Control Specimens. Heat 5081.

198

199

ORNL-DWG 67-3961R

2 40

Ef 30 -1

."0.oot n

0.Ot

0.1 STRAIN RATE (min-')

Fig. 19.2.

Influence of Strain Rate on the Ductility of Hastelloy

Irradiated to thermal fluence of

i I

! !

ig,

1.3 x lo2' neutrons/cm 2

transition from transgranular to intergranular fracture. However, the transition occurs at a higher strain rate for the irradiated than for the unirradiated material. At 65OOC and higher temperatures the ductility decreases progressively a s the test temperature increases. The variation in strain over the temperature range from 650 to 85OOC becomes smaller as the strain rate decreases, thus suggesting that the fracture strains a t very low strain rates (creep tests) may be independent of test temperature (above 65OOC). We shall later present data that indicate that this may indeed be the case. Most nuclear applications expose the materials to creep conditions rather than the high strain rates encountered in tensile tests. The fracture strains for several heats of irradiated air-melted material are given in Fig. 19.3. The data are represented by a scatter band, but there is an obvious trend for the data from heat 5065 to fall on the lower side and that for heat 5085 on the upper side. However, the spread is muoh larger a t high strain rates. The fracture strain is a minimum at a strain rate of about O.l%/hr. The irradiation temperature does not seem to be a significant factor in the fracture strain at lower strain rates, but can cause a factor of 2 variation at high strain rates, the higher values being obtained for the lower irradiation temperature.

N (Heat 5081) MSRE Surveillance Specimens.

The fracture strains are shown in Fig. 19.4 for several vacuum m e l t s of standard Hastelloy N. These materials seem to show a larger influence of irradiation temperature, although the scatter in the data prevents this from being an unequivocal conclusion. Generally, the material irradiated at < 150OC has the higher ductility. Specimens irradiated a t 65OOC have fracture strains quite close to those shown in Fig. 19.3 for the air-melted heats. This observation is quite interesting since the vacuum-melted alloys have boron levels of 0.5 to 1 0 ppm, whereas the boron levels of the air-melted materials are as high as 50 ppm. Although there are many other differences between air- and vacuum-melted materials, our tests do hot indicate that reducing the boron level alone is an effective solution to the problem of irradiation damage in this material. We have found that modifications in the chemical composition of Hastelloy N are very effective in reducing the radiation damage. The microstructure is generally characterized by large stringers of carbide. These carbides are of the M,C type and are very high in molybdenum and contain from 2 to 5% Si. We found that the stringers were eliminated i f the molybdenum content were reduced to 12 to 13%rather than the 16% normally used. The strength penalty for the reduction in molybdenum content seems rather modest. However, this

200

ORNL-DWG 68-4+9%

PRETEST ANNE1 SOLID POINTS- RADIATED 150째 C IRI 10

ISILE TESTS

. 1

0 0.001

aoc

0.1

MINIMUM CREEP RATE (%/hr) Fig. 19.3. Influence of Strain Rote on the Fracture Strain o f Several Air-Melted Heats of Hastelloy N. Irradiated to a thermal fluence of 2 to 8 x 10'' neutrons/cm 2 ond tested a t 65OoC. Solid points indicate specimens irradiated at less than 150OC; open points indicate an irradiation temperature of 65OoC.

change alone does not seem to improve the resistance t o irradiation damage. The further modification of adding about 0.5 wt % Ti to the basic composition of Ni-7% Cr0.2% Mn-0.05% C was found to give additional improvement. The creeprupture properties of two small commercial melts are shown in Fig. 19.5. The fracture strains are shown in Fig. 19.6. Again, the ductility minimum a s a function of strain rate is apparent. However, the minimum

ductility is not a s low for the titanium-modified material and rises very sharply with strain rate above or below the value of about O.l%/hr associated with the minimum ductility. The variation in the properties of the two titanium-modified heats is presently unexplained, although it is quite interesting that heat 21545, which has the higher boron level, has the better ductility. The influence of test and irradiation temperatures is also shown in Fig. 19.6. The standard

. c

201

ORNL-DWG 68-4198A 15

14 z

PRETEST ANNEAL-+ hr AT .SOLID POINTS- IRRADIATED OPEN POINTS- IRRADIATED 500-65O’C

- 9

-s

$ 8

bi w

9 7 I-

YL L 6 5

4 3

2 *

0 0.001

0.04

0.4

10 f MINIMUM CREEP RATE (% /hr)

400

1000

N. 2 to 8 x 10” neutrondcm’ and tested a t 65OoC. Solid points indicate specimens irradiated a t less than 150°C; open points indicate a n irradiation temperature of 65OoC. Fig.

19.4.

Influence of Strain Rate an the Fracture Strain of Several Vacuum-Melted Heats of Hastelloy

Irradiated to o thermal fluence of

heats included are only the vacuum-melted materials. They were irradiated and tested at the same temperature either 650 or 760OC. Although the iiacture strains observed in a tensile test were quite different, the strains under creep conditions were similar for 650 and 760OC. The titaniummodified specimens (21545 and 66-548) were all irradiated at 65OoC and then tested a t 650 or’ 76OoC. Again, the fracture strain in a tensile test varies with test temperature, but the strains under creep conditions do not vary significantly with test temperature.

19.2 STATUS OF DEVELOPMENT OF THE MODIFIED ALLOY

H. E. McCoy, Jr. Alloying additions to Hastelloy N of titanium, zirconium, and hafnium have been found to improve the resistance of the alloy to irradiation damage.3 Several small laboratory heats have been produced 3MSR Program Semiann. Progr. Rept. Aug. 31, 1967, ORNL-4191, p. 217.

202 tNL-DWG

0 0

I 1

i t

0.4

PRE;

2477 7304 6252 65-55

1 Ill1 ST ANNEAL-lhr AT 4l77OC I 'I1'----'OINTS- IRRADIATED 45O'Cm 'OINTS- IRRADIATED 500-650째C 8

IIII 1

I I I I11111 I I I I 1 1 1 1

4 00 RUPTURE TIME (hr)

1000

F i g . 19.5. Stress-Rupture Properties of Irradiated Titanium-Modified Hastellay 2 65OoC. Thermal fluence was 2 to 5 x IO2' neutrondcm

that contain up to 1%of each alloying addition. Two 100-lb commercial m e l t s containing 0.05 and 0.5% Hf have been procured and found to have excellent properties. The zirconium addition reduced the fabricability of the alloy, so that the first s m a l l commercial m e l t s containing 0.5 wt % Zr broke during fabrication at 1177OC. A subsequent m e l t was fabricated successfully by lowering the working temperature to about 1038OC. Thus, two 100-lb commercial m e l t s containing 0.05 and 0.5% Zr were obtained. Both of these materials exhibited extensive hot cracking during el ding.^ Twenty-five 1 W l b commercial melts have been procured with titanium levels up to 1.2% and carbon levels ranging from 0.003 to 0.1%. These alloys have fabricated well, have exhibited good welding characteristics, and have shown good pre- and postirradiation properties. We have decided to continue further development of only the titanium-modified alloy. This decision was based on: (1)the poor weldability of the zirconium-bearing alloys, (2) the excellent properties of the titanium-modified alloy, and (3) the comparatively large amount of experience that we 4hlSR Program Semiann. Progr. Rept. Aug. 31, 1967,

ORNL-4191,p. 221.

40,000

Irradiated and tested at

have with titanium-bearing materials compared with the hafnium-modified alloys. The f i r s t large-scale melt has been delivered. This 5000-lb melt was produced as a doublevacuum-melted ingot and subjected to standard mill practice. The yield will be about 2000 lb, which is quite reasonable for the various shapes of products that were made. The composition of this melt is given below. This material will be included in all phases of our testing program. Chromium Molybdenum Tungsten Iron Carbon Silicon Cobalt Manganese Vanadium Phosphorus

Sulfur Aluminum Titanium Copper Nitrogen Boron

7.84% 12.36% 0.12% 0.25% 0.066% 0.02% 0.07% 0.14% 0.04% 0.002% 0.002% 0.12% 0.56% 0.01% 0.004% 0.2 ppm

203

ORNL-DWG

HEAT 0

e 0 4 A

4 o

IRRADIATED TEMPERATURE PC1

TEST TEMPERATURE CC1

650 760 650 760 500-650 500-650 500-650

650 760 650

6252 6252

5944 5944 24545 21545 66-548

68-4203A

760 650 760 650 760

ANNEAL 4 hr AT 4477O C BEFORE TESTING

J Z 8

ti W

5 7

I -

I-

kl 0.4

(

0.04

MINIMUM CREEP RATE (%/hr)

Fig. 19.6.

Variation of Fracture Strain w i t h Strain Rote for Standard and Titanium-Modified Hastelloy N.

Thermal fluencs w a s

8 x lo2'

neutrons/cm

204 Further large m e l t s are necessary before we or the vendors will consider this alloy a standard production item. The next large m e l t to be procured will have approximately the same composition as that shown previously, with relaxed specifications on iron and manganese to allow more liberal use of scrap by the vendor.

193. EFFECT OF CARBON AND TITANIUM ON THE UNIRRADIATED CREEP-RUPTURE PROPERTIES OF Ni-Mo-Cr ALLOYS

C. E. Sessions As part of our efforts to optimize the mechanical properties of titanium-modified Hastelloy N, we are studying the effects of carbon concentration on the alloy. The interaction of carbon with strong carbide formers such as molybdenum, chromium, and, more significantly, titanium is expected to strengthen the alloy through precipitation reactions. In addition, however, we suspect that the beneficial effects of s m a l l titanium additions on the postirradiation ductility of these alloys is in part due to titanium-(boron, carbon) interactions which reduce the concentration of helium at the grain boundaries following neutron irradiation. Since boron and carbon should act similarly in this alloy, we have attempted t o define the titanium-carbon interactions in order to better deduce the role of titanium in reducing the irradiation damage of nickel-base alloys.

Five small laboratory heats of modified Hastelloy N (Ni-12% Mo-7% Cr-0.5% Ti) were made with carbon levels of 0.003, 0.007, 0.037, 0.053, and 0.27%. A sixth heat containing 0.04%C and no titanium was a l s o included. After fabricating into '/,-in.-rod stock, miniature mechanical property specimens were machined and given one of three different heat treatments in argon. Samples were then creep tested in air at 65OoC at stresses of 47,000, 40,000, and 32,350 psi. An evaluation of the creep rupture results indicated the following:

1. The rupture life at a given stress increases with increasing carbon content, as shown in Fig. 19.7. 2. Heat treating for 1 hr at 1177OC prior to testing resulted in longer rupture lives than the other two heat treatments (Le., 1 hr at 1260째C, or 1 hr at 1177OC followed by aging 100 hr at 87OOC). 3. The creep ductility at 40,000 psi showed little variation (14-20%) for carbon levels up to 0.05%,but at the highest carbon level, 0.27%, the creep elongation increased significantly to 30-40% (Fig. 19.8). 4. The minimum creep rate at all stress levels de-

creased continuously with increasing carbon content.

5. The addition of 0.5% Ti to the basic Ni-12% Mo-7% Cr-0.04% C alloy increased the rupture

ORNL-DWG 68-4993

.-n 0 0

-ge

Eb 20 W

a 0 40

494

195

A (96

- 0 (97 A 498

0.007% 0.037 2 0.05% 0.27 %

0.042 (WITHOUT TITANIUM)

0.4

400

4000

RUPTURE LIFE (hr)

Fig. 19.7.

Effect of Carbon on the Rupture Life of Ni-12% Mo-7% Cr-0.5% Ti Alloy at 650OC.

205

ORNL-MNG 68-6084

-3 I

NO Ti

&4SIC ALLOY: Ni- 42 % Mo- 7% Cr 0.5%Ti

ANNEALED Ihr AT 4260OC

0.004 0.002

0.02 0.05 CARBON CONTENT ( w t %1

0.005 0.04

0.4

0.2

0.5

Fig. 19.8. Effect of Carbon and Titanium Content on the Total Creep Strain a t 40,000 psi and 650OC.

life by a factor of from 3 to 10, depending on the stress level.

6. The 0.5% Ti addition a t the 0.04% C level did not change the creep ductility of samples tested in the solution-annealed condition; however, the ductility of samples that were aged 100 hr a t 87OOC before testing was increased by a factor of 2 a t the lower stress level.

i I i k . , I

These observations indicate that there is a pronounced strengthening effect due to titanium-carbon interactions in this alloy; however, no significant increase in creep ductility occurs a t the 0.5% T i level when compared with samples containing no titanium addition. Changes in the microstructure with increases in carbon concentration are shown in Fig. 19.9 for samples solution annealed a t 1177OC and aged 100 hr a t 87OOC. The lower carbon levels are virtually single-phase alloys, but the amount of second phase increases significantly with increasing carbon. Heat 198, which contains 0.04% C and no titanium, shows a large amount of finely divided precipitates that are probably M,C type. Each sample, except the 0.27% C level, was single phase after a 126OOC heat treatment. To summarize, the beneficial effect of titanium additions on the out-of-reactor properties of this alloy is primarily a strengthening advantage. There appears to be no effect of titanium on the

creep ductility for samples tested in the sohtionannealed condition; however, a small increase in ductility was found a t lower stresses for samples heat treated to precipitate the carbon prior to testing. Carbon increases the creep strength significantly, especially over the range from 0.007 to 0.037 wt %.

19.4 ELECTRICAL RESISTIVITY OF TITANIUMMODIFIED HASTELLOY N H. E. McCoy, Jr. The resistivity of Hastelloy N varies with temperature in a somewhat anomalous fashion5 Upon heating, the resistivity increases with temperature up to about 6OO0C, decreases with increasing temperature up to 1000째C, and then increases with further increases in temperature. The resistivity -.\ is decreased by cold working rather than increased, as normally noted for metals. This behavior is generally attributed to short-range order, with the ordering reaction causing an increase in resistivity. We have investigated the resistivity of the titanium-modified alloy and found that the same 5H.E. McCoy, Jr., Resistivity Anomaly in NickelBase Alloys (in preparation).

206

L, PHOTO 94879

0.003 YoC

493 0.007%C

4 94

037Yo C

0105 Ye C

495

190

Fig. 19.9. Effect of Carbon on the Microstructure of Titanium-Modified Hastelloy N (Ni-12% Mo-7% Cr-0.5% Ti). Samples solution annealed 1 hr at 1177°C and aged 100 hr at 87OoC.

general behavior prevails. However, the resistivity curve is lower by about 12%for the modified alloy. The chemical changes involved for the modified alloy include a decrease in the molybdenum content from 16 to 1296, the addition of 0.5%Ti, a decrease in the iron content from 4 to O%, and a general decrease in residual elements such as silicon and manganese. These changes lead to a decrease in the total alloy content and hence would be expected to yield an alloy with lower resistivity. A composite set of curves is shown in Fig. 19.10 for the titanium-modified alloy. Cold working decreases the resistivity by about 15%. After a complete heating and cooling cycle, an “equilibrium” curve is established that is followed during subsequent cycles. Annealing the material for 1 hr at 1177OC reduces the resistivity slightly, but the equilibrium curve is followed after the first heating. The resistivity changes over a normal operating range of 25 to 7OOOC are quite small, only 4%.

19.5. ELECTRON MICROSCOPE STUDIES OF HASTELLOY N

R. E. Gehlbach 19.5.1 Phase Identification Studies in Hastelloy N We are continuing our investigation into the characterization of precipitated phases occurring in Hastelloy N. As we reported previously6 the microstructure of standard Hastelloy N is characterized by stringers of blocky M,C carbides after annealing in the temperature range 1177 to 1260%. After aging and/or testing at temperatures between 427 and 87loC, a fine dendritic grain-boundary precipitate is generated, which coarsens with increasing aging temperature and with increasing t i m e s at the lower temperatures. --.

,R. E. Gehlbach, MSR Program Semiann. Progr. Rept. Aug. 31, 1967, ORNL-4191,pp. 219-21.

. I W i

102

4 00

Fig. 19.10.

400

200

300

207

500

600

\ \

800

\ \

700 TEMPERATURE ("C 1

400

ORNL-DWG

1100

68-6085

1000

900

Variation of Resistivity with Temperature for Titanium-Modified Hastelloy

208 Two tensile specimens were examined that had been annealed a t 1177 and 126OOC and tested at 982OC. They exhibited elongations of 58 and 4.6% respectively. Transmission electron microscopy of electrochemically thinned slices from the gage lengths showed grain boundaries nearly completely free of precipitates and completely recrystallized for the specimens annealed a t 1177OC, whereas the grain boundaries of samples taken from the specimen with the higher temperature anneal contained a very extensive amount of precipitate (Fig. 19.11). Extraction replication revealed a cellular precipitate morphology (Fig. 19.12) in the grain boundaries of the latter specimen and confirmed the relative absence of precipitation in the grain boundaries of the former. The cellular morphology has the same lattice parameter (11A) a s the blocky and dendritic forms of the M6C carbides. Chemical analysis of the various M,C-type carbides in standard Hastelloy N has been completed, employing both conventional electron microprobe techniques and also the electron microprobe analyzer accessory on one of our electron microscopes. With the latter instrument, we are able to analyze

'Xi. E. McCoy, Influence o f Several Metallurgical Variables on the Tensile Properties o f Haetelloy N , ORNL-3661, p. 8 (August 1964).

Fig. 19.12. Morphology of M,C-Type Precipitate i n Fig. 19.11b.

Grain-Boundary

precipitates on extraction replicas, which permits the study of fine grain-boundary precipitates that cannot be analyzed in the bulk sample. The results are summarized in Table 19.1. Several PHOTO 94880

Fig. 19.11.

Effect of Annealing a t (e) 1177% and ( b ) 126OoC on Microstructure After Tensile Fracture a t 982OC.

Fracture elongations were 58 and 4.6% respectively.

209

T a b l e 19.1.

Analysis of Precipitates in Standard Hastelloy N in %

Large Stringer Major Element

Nominale

Matrix

Precipitate' b d, e

Grain Boundaryd" Precipitate

High-Temperature Phas ebne

74.5

33.8

37.4

15-18

64.6

12.1

55.8

49.2

70-75

22.2

7.1

4.8

5.7

2-4

7.4

4.2

0.8

0.6

0.3

2.7

0.06

2.7 3.7

1.1

2.5-4.5

2-1 . 7

eNominal alloy composition from bulk chemical analysis. bAnalysis employing conventional microprobe techniques on bulk specimen; normalized to 98%. 'M6C-type carbides. dAnalysis employing microprobe analyzer accessory on electron microscope. eCorrected for absorption, atomic number, fluorescence; normalized to 96%.

'Thin precipitates, no corrections required; normalized to 96%. *Noncarbide.

errors are inherent in the techniques involved, and the results are only semiquantitative. The blocky, or stringer, morphology approximately fits the stoichiometric Ni,Mo,C carbide, whereas the grain-boundary morphology corresponds closely to the Ni,Mo,C composition. We have carried out additional work on Hastelloy N which contained 14C introduced into the molten alloy. Autoradiography performed on these specimens indicates that the M,C-type precipitates actually are carbides, as can b e seen in Fig. 19.13. The high-temperature phase present after annealing the 1 4 C at 2316OC and cooling quickly was reported previously6 as being a noncarbide phase, probably resulting from a transformation from the carbide precipitates. Electron microscope microprobe analyses performed on extractions of this phase indicate that it is primarily molybdenum and chromium in the ratio of 9:l by weight. No other elements of atomic number greater than 11were detected. The carbides that are also present in this specimen have been identified by electron diffraction as

*H. E. McCoy, Studies of the Carbon Distribution in Hastelloy N, ORNL-TM-1353,jp. 2 (February 1966).

the Mo,C type, and analysis of individual particles and f i l m s indicates that the actual composition is close to (Mo,.~C~,~,),C. It is possible that the high-temperature phase in the 1 4 C material is not the same phase a s that formed in the commercial heats, because conventional microprobe analysis of the high-temperature phase in standard Hastelloy N differs markedly from the above results (see Table 19.1) and because Mo,C-type carbides have not been observed in specimens from standard alloys which were annealed at high temperature. We are now attempting to identify the high-temperature phase after extraction from standard Hastelloy N.

19.5.2

Effect of Silicon on Precipitation in Hastelloy N

R. E. Gehlbach

H. E. McCoy, Jr.

In air-melted heats of Hastelloy N, the large carbides do not go into solid solution, even at high annealing temperatures, but appear to melt and transform to a lamellar phase between 1260 and 1316OC. However, in vacuum-melted heats the precipitates do go into solution, and the hightemperature phase is not formed (Fig. 19.14). The

210 PHOTO 91881

Fig. 19.13. Carbon-14 Specimen Annealed a t 1177OC and Aged 100 hr a t 649OC. Showing Activity a t Precipitates and in Grain Boundaries. ( a ) Lightly etched. ( b ) Specimen with NTB-2 autoradiogrophie emulsion, exposed

100 hr. SOOX.

3. The precipitates go into solution after annealing the low-silicon heats at temperatures of 1260 to 137loC,whereas those in the highsilicon heats melt and transform to the hightemperature phase seen in Fig. 19.14. 4. The temperature for the melting and transformation was lower for the 1% than the 0.48% alloy. 5. The concentration of silicon in the precipi-

major compositional difference between the airand vacuum-melted alloys is the silicon concentration. To evaluate the effect of silicon on precipitation, several alloys of nominal composition (16 Mo, 7 Cr, 4 Fe, etc.) with silicon concentrations of 0.02, 0.12, 0.48, and 1.05 wt % were prepared. The following observations were made regarding the effect of increasing silicon concentration on precipitation:

tates, determined by conventional microprobe analysis, increases with increasing silicon in the specimens and is generally two to five times the matrix silicon concentration. The matrix composition is approximately half the nominal silicon concentration in the alloy. These data are shown in Table 19.2.

1. The amount of M,C-type precipitates increases noticeably; however, many are present in the low-silicon heats after annealing at 1177OC.

2. The as-cast structure is retained somewhat in the 0.48% Si heat, greatly in the 1% alloy. 'H. E. McCoy, Influence o f Several Metallurgical Variables on the Tensile Properties o f Hastelloy N,

ORNL-3661 (August 1964).

We extracted precipitates from a specimen of 1% Si which had been annealed at 1260째C, using a direct carbon extraction replica (Fig. 19.15). The

211

PHOTO 94882

AIR MELTED (0.6% ' Si )

VACUUM MELTED (0.045' 7- Si)

245OOF

Fig. 19.14.

25OOOF

High-Temperature Phase Transformation Occurring in Air-Melted Hastelloy N. Note that the M6C-

type carbides go into solution in the low-silicon vacuum-melted heat and do not melt or form the high-temperature phase.

grain boundaries contain much precipitate in the form of both particles and thin films associated with these particles. In contrast, very little precipitate is found in grain boundaries of standard Table

19.2. Semiquantitative Microprobe Analysis of Silicon i n Hastelloy

* /

Heat

Nominal (wt%)

Matrix (wt %I

Precipitatesa (wt %)

0.02

0.03

0.12

0.1

0.6

0.48

0.3

1.5

1.05

0.5

1.7

aPrecipitates are very small; silicon concentration is actually greater due to matrix dilution resulting from bulk metal surrounding particles.

Hastelloy N. Selected area electron diffraction indicates that the film precipitate has essentially the same lattice parameter a s the M,C-type carbides found in the standard alloy. X-ray diffraction powder patterns confirm this observation, and long exposures fail t o show any diffraction lines other than those associated with the M,C-type carbide. Microprobe analysis of these extracted particles using the electron microscope microprobe accessory resulted in a composition similar to the blocky carbides present in the commercial material with the exception of silicon, which exists in concentrations up to about 5%. We know that the M6C-type precipitates in standard Hastelloy N are rich in silicon (Sect. 19.5.1). It is quite obvious that silicon is playing an important tole both in the high-temperature noncarbide phase and in the behavior of the M6C-type precipitates. We have not been able to correlate the M6C-type particles with carbon, but their be-

212

Fig.

19.15. Direct Carbon Extraction Replica Showing Blocky Precipitates and Grain-Boundary F i l m s Extracted Specimen was annealed a t 126OoC before extraction.

from Hastelloy N with 1.05% Si.

havior is sensitive to silicon concentration in the alloy. Both the blocky and grain boundary precipitates are rich in silicon and increasingly so with increases in the silicon concentration of the bulk material. The available information suggests that the silicon may occupy carbon positions with the lattice parameter not being changed significantly. If silicon were to occupy all carbon sites, about 5.2% Si could be accommodated. Microprobe analysis of individual particles and precipitate films has not suggested that the silicon concentration exceeds this value.

19.5.3

Titanium-Modified Hastelloy

in -Fig. 19.16, a s well as some larger precipitates, is actually of the Mo,C type, the major phase present. The lattice constants are slightly con-

We have identified fine grain-boundary precipitates which occur in titanium-modified Hastelloy N after annealing at 1177 and 871OC. It was previously reported' that the thin-film morphology was probably face-centered cubic Tic and the particles a noncubic precipitate. However, we have found by electron and x-ray diffraction that the thin-film morphology shown 'OR. E. Gehlbach. MSR Program Semiann. Progr. Rept. A @ . 31, 1967, ORNL-4191, pp. 219-21.

Fig.

19.16.

Grain-Boundary Precipitate in Titanium-

Modified Hastelloy

N. See text

for discussion.

213

tracted from those for pure Mo2C, and we have found that the metallic portion of the carbides contains approximately 10 wt % chromium. Electron diffraction also shows that the small particles associated with this film, a s well a s the needle-like morphology, have a face-centered cubic lattice parameter of 4.24 A, similar to the values reported for TiN. X-ray diffraction shows a slightly expanded (4.26-4.27 A) face-centered cubic parameter, suggesting a complex Tic-type phase which might contain nitrogen, oxygen, and possibly another metallic atom besides titanium. The amount of this phase is quite small compared with the amount of Mo2C present.

19.5.4 Summary These studies have shown that standard Hastelloy N annealed a t 1177OC contains large M,C carbides that are rich in molybdenum. Aging causes the precipitation of fine M,C particles along the boundaries. All the M,C particles studied are enriched in silicon. A series of special alloys revealed that the concentration of silicon i n the precipitate increased a s the silicon content of the alloy increased. This segregation of silicon to these precipitates is probably responsible for the localized melting that occurs when this alloy is heated t o about 1300OC. The low-temperature M,C transforms to a lamellar product when the alloy is heated t o about 1260OC. Carbon-14 studies show that this product is not a carbide, but an identification has not been made. The titanium-modified Hastelloy N w a s found to have small amounts of grain-boundary precipitates of the Mo,C type and also a complex face-centered cubic phase.

concerned with whether this element would increase the corrosion rate. The corrosion of Hastelloy N in fluoride salt systems is controlled by the diffusion of chromium, and w e have carried out diffusion measurements to determine whether titanium diffuses rapidly enough to contribute significantly to the corrosion rate. W e have completed measurements of the diffusion coefficient for the radioactive tracer 4 4 T i in modified Hastelloy N over the temperature range 800 to 1250OC. The data obtained can be described by an Arrhenius expression, D

(15

* 2) exp 73,000RTk 3000

cm2/sec, (I)

a s illustrated in Fig. 19.17. Technique limitations prevented meaningful experiments at lower temperatures. ORNL-DWG 68- 3734

r (oc) 1200

1000

800

700

500

600

IO+

1b'O

lo-" 0

%s E u 10-12

19.6 DIFFUSION OF TITANIUM IN MODIFIED HASTELLOY N C. E. Sessions

T. S. Lundy

Experiments and service in the MSRE have demonstrated the excellent corrosion resistance of Hastelloy N in fluoride salt systems. Our radiation damage studies have shown that the resistance of this alloy to radiation damage can be improved markedly by the addition of about 0.5% Ti. Titanium forms a very stable fluoride, and w e are

Io-'6

~4000/~ (OK)

Fig. 19.1 7. Temperature Dependence of the Diffusion Coefficient of Titanium in Modified Hastelloy N.

214 The data were used to predict the maximum rate of titanium diffusion out of Hastelloy N under steady-state corrosion conditions. This occurs for the case where the titanium concentration at the surface in contact with the s a l t is maintained constant at zero. The appropriate solution of Fick's second law of diffusion is then

Table 19.3.

Depth at Which Titanium Concentration

lo Within Approximately 95% of I t s Initial Concentration in the Alloy

~~~

5 x 10-16 X

c(x, t) = C, erf -

2f11

c(x, t) = concentration at a distance x from the surface,

t = t i m e at temperature, D = diffusion coefficient,

"J. H. DeVan, Effect of Alloying Additions on Corrosion Behavior of Nickel-Molybdenum Alloys In Fused Fluoride Mixtures, M.S. thesis, The University of Tennessee, August 1960.

0.00025

8,000

0.00037

0.0005 0.002

4,000

0.0007

8.000

0.0010 0.0015 0.0060

16,000 250,000 2 x 10-l~

4,000 8,000

16.000

C, = initial uniform concentration throughout the specimen. Using Eq. (2) and experimental values of the diffusion coefficient D, we have calculated titanium concentration distributions that would result after various t i m e s at 700 and 800OC. At 7OOOC the extrapolated value of D is approximately 5 x lo-' cm2/sec. However, since this value is for volume diffusion, one might argue that the effective diffusivity, which includes the effect of short-circuit diffusion paths, might be as much as a n order of magnitude larger than the extrapolated value. Thus a value for D of 5 x cm2/sec has been used in our calculations for 700OC. The measured value at 80OOC was 2 x cm2/sec. The latter value corresponds roughly to the effective chromium diffusion rate in Hastelloy N at 65OOC." The depth of the diffusion zone from which titanium is depleted from the alloy increases with both increasing time and increasing temperature. Table 19.3 lists the depth at which the titanium concentration is about 95%of its original value. For the maximum probable coefficient at 70O0C (Le., 5 x 10" '), the depleted zone after 30 years (-250,000 hr) extends only over a distance approximately 60 p from the surface of t h e alloy. The total amount of material Mt which diffuses from the alloy held under isothermal conditions with a zero surface concentration is given by12

4,000

16,000 250,000 5 x 10-l~

where

250,000

Mt = 2c,

0,0015 0.002 0.003

0.0120

(3)

Using this equation and various values of D and t, we calculated the expected buildup in solute concentration in the salt of a typical system such as the MSRE. (The total surface area was assumed to be 852 ft2, and the total amount of molten salt 4.92 x lo6 g.) The results are shown in Table 19.4. A t 8OOOC the amount of titanium removed would be approximately 169 g after 30 years, which is the equivalent of only 30 ppm increase in the salt concentration; at 7OOOC the comparable values would be 84 g of titanium and 17 ppm. We conclude from these studies that the addition of titanium to Hastelloy N has not affected the corrosion resistance of the alloy. Because of the lower rate of titanium diffusion and the lower concentration of titanium present, the selective removal of chromium will continue to be the primary corrosion process in this alloy in malten fluoride salts.

12J. Crank, The Mathematics of Diffusion, pp. 11-13, Clarendon, 1956.

215 T a b l e 19.4.

Calculated Titanium and Chromium Loss from Modified Hastelloy

Under Z e r o Surface Concentration Steady-State Corrosion Conditions

Hastelloy N surface area used was 852 ft2. Amount of fluoride s a l t considered a t 4.92

Alloy

Temperature

Diffusion

Exposure

Flux Out

Total Amount

Constituent

(OC)

Coefficient (cm2/sec>

Time (hr)

(g/cm2)

Removed (g)

5 x 10-16

1,000 4,000 9,000 16,000 250,000

2.13 4.26 6.39 8.52 3.37

x x

1,000 4,000 9,000 16,000 250,000

6.74 1.35 2.02 2.69 1.06

x x 10-~ x

2 x 10-l~

1,000 4,000 9,000 16,000 250,000

2 x 10-l~

1,000 4,000 9,000 16,000 250,000

Titanium

700

800

Chromium

-650

lo6 g. Salt Concentration Increase @pm>

1.7 3.4 5.0 6.7 26.5

0.3 0.7 1.0 1.3 5.1

x 10-~ x 10-~

5.3 10.7 15.8 21.2 84.0

1.1 2.2 3.2 4.3 17.0

1.35 2.69 4.04 5.39 2.13

x 10-~ x 10-~ x 10-~ x 10-~ x 10-~

10.7 15.8 31.9 42.6 169

2.2 3.2 6.5 8.7 30.5

1.89 3.77 5.65 7.55 2.99

x x x x x

x x 10-~

10-~ 10-~ 10-~ 10-~

149 298 446 596 2360

30.0 60.6 90.8 121.2 480

chromium and titanium concentrations were 7 and 0.5 wt % respectively.

19.7 MEASUREMENT OF RESIDUAL STRESSES IN HASTELLOY N WELDS A. G. Cepolina

D. A. Canonico

We are continuing to study the effects of welding conditions and postweld heat treatment on the djstribution and level of the residual stresses in Hastelloy N. The method of measuring the residual stresses is a modification of the Boring-Sachs technique,' which permits the continuous investigation of the distribution of the planar stresses in and near the weld.

13MSR Program Semiann. Progr. Rept. Au& 31, 1967, ORNL-4191, pp. 223-26.

The technique used for machining the specimens has been modified slightly from that previously r e ported.I3 A lathe was substituted for the milling machine previously used. The cutting tool is fixed more rigidly, resulting in an improved cut with more uniform quality and reduced work hardening. Large volumes of cutting fluid a r e now used for both lubricating and cooling. The temperature of the welded specimen is monitored, and a thermocouple placed near the strain gages showed a maximum temperature rise during machining of about 3OC. The normal lathe chucking technique has been revised. The 12-in.-diam, i-in.-thick welded specimen is bolted to a Micarta bed. The specimen is then held by clamping to the plate. This system eliminates the stresses in-

216 wide were made simultaneously on each face of a '/,-in.-thick Hastelloy N disk. Because of this welding procedure, the stress distribution induced in the plate is assumed to be planar. Welds involving variations in shielding gas and postweld heat treatment have been investigated. Prior to welding, all plates were annealed for 1 hr a t 1177OC in hydrogen. Table 19.5 lists the variations studied to date. The computer printout curves showing the distribution of radial and tangential stresses across the welds of specimens 1, 5, and 6 are shown in Fig. 19.18. From an analysis of the welds made without postweld heat treatment, it appears that the maximum tangential stress occurs about in. from the weld axis toward the disk center." Using helium a s the arc shielding gas, the peak stress was found to be approximately 61,000 psi; for argon shielding gas, the value was about 56,000 psi. It appears that the stress gradient across the weld is steeper â&#x201A;Źor argon than for helium. This probably results from the hotter and wider weld puddle which characterizes welds made under helium. Two postweld heat treatments were also considered (specimens 5 and 6). Sample 5 was given a full anneal a t 1177OC for 1 hr, while sample 6 was heated 6 hr a t 871Â°C, a standard stress relief for Hastelloy N. Both treatments reduced the residual welding stress to extremely low values. Whereas the full anneal eliminated any peaking of

duced in the plane of the specimen by the conventional holding system. The machining operation on the lathe requires that the strain gages and their associated wires be handled in a way that will prevent them from being damaged. To accomplish this, the wires are fed through the hollow lathe shaft and connected t o silvercoated plugs. During machining, these plugs are held in a manner that permits their rotating with the shaft. After a cut is completed, the lathe is stopped and the plugs are connected to the strainmeasuring unit. After the reading is taken, the plugs are disconnected, and the next cut is made. The error due to contact resistance in the plugs is negligible. The tangential residual stress (with respect to the weld axis) is a function of the change of the tangential strain with the radius, that is, dc,/dR, where tzT is the gage reading in microinches per inch and where R is the value of the radius of the cut in inches. In order t o evaluate dcT/dR, it is necessary to fit the experimental readings of strain and radius with a continuous function. The curve fitting and evaluation of the slope at various values of R are done by a computer. The tangential stresses are evaluated as a function of location across the weldment. The radial stresses are determined mathematically from the longitudinal strain data. These calculations have also been computerized.

19.7.1

Experimental Results

"This asymmetry might be explained by considering the differential heat sink and restraint conditions that exist between portions of the sample. This is attributable to the difference in the amount of material present (ratio of approx 1/3).

Bead-on-plate welds (fusion pass with no addition of filler metal) have been examined. Two 6-in.-diam circular welds with a weld bead in.

Table 19.5.

Conditions Used for Moking Weld Test Specimens

Heat Input Specimen No.

for Each of T w o Simultaneous Welds

Shielding Gas

Postweld Heat Treatment

ci/in.) 15,000

Argon

None

15,000

Argon

6 hr at 871OC in H,

15,000

Argon

1 hr at llf7OC in H,

15.000

Helium

None

3 '

217

ORNL-DWG 68-6086

3 SPECIMEN N0.4 (AS WELDED)

-.-

(I In

0 0

e & O W J

9 v)

-4

'/4

SPECIMEN N0.6 STRESS RELIEVED 6 hr AT 16OO0F

SIGMA R SIGMA T

-4 4

SPECIMEN N0.5 ANNEALED 1 hr AT 245OOF

-4

2.5

-_ 3.5

3.0

SIGMA R SIGMAT

4.0

R (in.)

Fig. 19.18. Weldments.

Residual Stresses i n Various Hastelloy N

Sigma

of thick Hastelloy N plate has been received from Battelle Memorial Institute (BMI). Battelle was supplied with a 1-in.-thick plate of MSRE grade Hastelloy N and a 25-lb coil of 0.125-in.-diam Hastelloy N weld wire. Because of the needs of the NGW process, it was necessary to reduce the wire diameter to 0.045 in. Problems were encountered by BMI during processing, and only 7 lb of usable wire was finally obtained. The NGW process is a proprietary process developed by BMI. It is basically a metal-arc inertgas process wherein welding is done a t the bottom of a narrow gap formed by the root faces of the two materials being joined. The process has the advantages that it requires no edge preparation and the amount of filler metal needed to fill the joint is minimal. Operationally, the two plates t o be joined are placed about in. apart. Two torches slightly out of alignment (within the allowable in. dimension) a r e employed. The second torch trails the first by about 2 in. The lay of the weld wires is oriented so that each tends t o deposit its bead toward the wall nearest it. The process permits low-heat-input welds, which, owing t o the narrow joint, results in the fabrication being completed with a minimum number of passes. Three welds were made during the course of the investigation. Only the first weldment was made under the normal twin-wire-narrow-gap technique. This weldment was made using an argon-rich atmosphere (60 Ar-40 He). Center-line cracking occurred in the trailing weld bead after about half the plate thickness had been welded. The second weld was made using a single-wire technique and a helium-rich atmosphere (70 He-30 Ar). Cracking again occurred after slightly less than half the plate thickness had been welded. The third weld was preceded by a bead-on-plate study to further optimize the weld bead contour. It was found that a n 80 He-20 Ar atmosphere produced the best weld bead, one with a reduced capillary and no long columnar grains. Special precautions were taken with the filler wire prior to welding, with the interpass temperature, and with the condition of the surface. This weld was also unsatisfactory, since it contained large blowholes and a number of cracks and fissures.

R = radial stress, and sigma T =

tangential stress.

the stress, the 871OC treatment left a maximum of about 5000 p s i i n the weld deposit. From these results, it appears that the heat treatment a t 871OC for 6 hr is indeed beneficial. The difference in s t r e s s distribution between argon and helium welds is also noteworthy.

19.8 APPLICATION OF THE NARROW-GAP WELDING PROCESS TO THE JOINING OF HASTELLOY N

D. A. Canonico The final report on the application of the NarrowGap Welding Process (NGW)' s to the joining

"R. P. Meister and D. C. Martin, Brit. Welding J . 13, 252 (May 1966).

218 This study concerning the applicability of the NGW process was inconclusive. In all instances the welds were unsatisfactory. However, a strong possibility exists that the filler wire was not of ,the best quality.

19.9 NATURAL CIRCULATION LOOPS AND TEST CAPSULES

J. W. Koger

A. P. Litman

Six loops and four capsules are presently in operation. Tables 19.6 and 19.7 detail the service parameters of these test units. Since last year we have utilized a newly designed natural circulation loop. The device is essentially a test bed system wherein metal specimens suspended in the s a l t and salt samples can be removed and/or replaced a t operating temperature without disturbing flow or introducing air contamination. This design makes it possible to obtain kinetic data on corrosion and to study corrosion mechanisms prior to dismantling and subjecting an entire loop to metallurgical analysis. Figure 19.19 shows a schematic of the new loop configuration, while Fig. 19.20 displays two loops, NCL-15 and -16, in a protective hood prior to operation. We are continuing to concentrate on the compatibility of Hastelloy N with fuel, blanket, and coolant salts. The compositions of standard Hastelloy N and the titanium-modified Hastelloy N are shown in Table 19.8.

19.9.1 Fuel Salts Loop 1255, constructed of Hastelloy N and containing a simulated MSRE fuel salt plus 1 mole % ThF, , continues to operate without difficulty after almost six years. Loop 1258, constructed of type 304L stainless steel and containing the same salt a s loop 1255, has operated about 4.6 years with only minor changes in flow characteristics. In January 1967, ten new stainless steel specimens were placed in the hot leg of loop 1258. The specimen in the hottest position was replaced after 3700 hr. A plot of the weight change of all the specimens a s a function of time and temperature is given in Fig. 19.21. The curves are rather typical of corrosion processes that are diffusion controlled. A s ex-

pected, the weight loss is highest at the hottest point in the loop and decreases with decreasing temperature. Salt analyses show that only chromium is leached from the metal, and a diffusion coefficient can be calculated from Eq. (1) by assuming that the rate-controlling process is the diffusion of chromium to the solid surface:

2 =-

(C,

- CJcr Wl

(1)

where

= weight loss, g/cm2,

C, = bulk chromium concentration in material, d c m 3, C6 = surface chromium concentration in material, d c m 3, t = time, sec,

= diffusion coefficient, cm2/sec.

Assuming a surface chromium concentration of zero, the calculated values for D at 649 and and 1.4 x lo-' cm2/sec 677OC were 4.0 x (ref. 16). The diffusion coefficient for chromium in austenitic stainless steel at 677OC was extrapolated from existing data as 4 x lo-' cm2/sec. Thus our corrosion data indicate an apparent diffusion coefficient two orders of magnitude higher than that extrapolated from diffusion measurements. There are several possible reasons why these values do not agree. (1) The measured values were obtained over the temperature range from 849 to 1398OC and extrapolated to 677OC. Shortcircuit diffusion mechanisms have been shown to be effective at lower temperatures and lead to a deviation in the standard Arrhenius plot used for extrapolation. (2) A corrosion mechanism more complicated than leaching of chromium from stainless steel is involved. (3) The solid-state diffusion of chromium in stainless steel is not the rate-controlling step in the process. W e are continuing to investigate these factors in an effort to explain the differences above. Loop NCL 16, the first two-fluid MSBR fuel salt natural circulation loop incorporating the new loop configuration, is now operating. The weight change obtained after 250 hr on the titanium-modified Hastelloy N specimen at 7MoC was 0.35 mg/cm2.

16R. A. Wolfe and H. W. Paxton, Trans. Met. AIME 230, 1426 (1964).

SOC.

-c .

Table 19.6. MSRP Natural Circulation Loop Operation Through February 2 9 , 1968 Maximum Loop No.

Loop Material

Salt Composition (mole 70)

Specimens

TYpe

+ 2%

T i m e (hr)

Temperature (OC)

(OC)

Scheduled

Operated

1255

Hastelloy N

Nb'"

Fuel

LiF-BeF,-ZrF,-UF,-ThF, (70-23-5-1-1)

704

Indefinite

51,810

1258

Type 304L stainless steel

Type 304L stainless steelb#'

Fuel

LiF-BeFZ-ZrF4-UF4-ThF4 (70-23-5-1-1)

677

100

Indefinite

40,510

NCL 13

Hastelloy N

Hastelloy Ncnd

Coolant

NaBF4-NaF (92-8)

607

150

5000

2,930

NCL 14

Hastelloy N

Titanium-modified Hastelloy Nc*d

Coolant

NaBF4-NaF (92-8)

607

150

Indefinite

2,840

NCL 15

Hastelloy N

Titanium-mo dif ie d Hastelloy N, Hastelloy N

Blanket

LiF-BeFz-ThF, (73-2-25)

677

Indefinite

530

NCL 16

Hastelloy N

Titanium-modified Hastelloy N; Hastelloy N

Fuel

LiF-BeF2-UF, (65.5-34.0-0.5)

704

170

Indefinite

340

NCL 17

Hastelloy N

Titanium-modified Hastelloy N, Hastelloy N controlsc*d

Coolant

NaBF4-NaF (92-8)plus water vapor additions

607

150

Indefinite

Estimated startup 5-1-68

NCL 18

Hastelloy N

Titanium-modified Hastelloy N; Hastelloy N

Fuel

LiF-BeF2-UF, (65.534.0-0.5) plus FeF, additions

704

170

Indefinite

Estimated startup 5-1-68

NCL 19

Hastelloy N

Titanium-modified Hastelloy N; Hastclloy N controlscod

Fuel

LiF-BeF,-UF, (65.534.0-0.5)plus bismuth plus lithium in molybdenum hot finger

704

170

5000

Estimated startup 5-1-68

aPennanent specimens. bHot leg only. 'Removable specimens. dHot and cold leg.

Table 19.7. MSRP Capsule Program

Container Material

Specimens

Test Fluid (mole %)

Hastelloy N (four containers)

Hastelloy N in vapor, liquid, and interface

NaBF4-NaF (92.8) plus BF3 a t 120 mm Hg

T i m e (hr)

Temperature (OC)

607

Scheduled 2000

Support coolant s a l t project and 840 (No. 1) determine effect of BF, pres822 (No. 2) sure on compatibility of so815 (No. 3) dium fluoroborate salts with 805 (No. 4) Hastelloy N

500

Estimated startup 4-1.68

(No. l), 50 psig (No. 2), 100 psig (No.3), and 400 psig (No.4)

TZM~

TZM

LiF-BeF2 (88-12)

1093

Operated'

Support MSRP fuel processing program; potential vacuum still material

'Through February. %0-0.5% Ti-O.08% Zr-0.02% C.

c:'

221 ORNL-DWG

68-3987

There is little doubt that an unsteady-state condition still'prevails in the system, and any data obtained at this time must b e considered tentative. In support of the MSRP fuel processing scheme using vacuum distillation, we have fabricated a TZM alloy capsule that will contain LiF-BeF, (88-12 mole %). This composition is typical of the salt distilland expected during processing. The test will be run a t 1093OC, the approximate distillation temperature. This temperature precludes use of nickel-base alloys, and thus the molybdenum-base alloy TZM has been selected as a candidate retort material. 19.9.2 Blanket Salts

Loop NCL 15 is now operating with a typical two-fluid MSBR blanket salt. Considerable difficulty was experienced in filling this loop. The salt seemed to have an exaggerated tendency toward incongruent melting and segregation during solidification. The weight change obtained after 400 hr on the titanium-modified Hastelloy N specimen at 677OC was 0.05 mg/cm2. This figure is lower than that for any other loop we have studied recently. It is again emphasized that these data are preliminary, since the loop has not reached a steady state.

CLAMSHELL HEATERS

30 in.

INSULATION

I/ I I I

CORROSION SPECIMENS

19.9.3 Coolant Salts 1

I I

SAMPLER

u INCHES

I'I

Loops NCL 13 and NCL 14, containing NaBF,NaF (92-8 mole %), continue to operate under identical conditions. Loop NCL 13 contains standard Hastelloy N specimens, while loop NCL 1 4 has titanium-modified Hastelloy N specimens suspended in the salt stream. These differences make possible an excellent comparison of the effect of fluoroborate s a l t on these alloys. Utilizing weight changes on the specimens, changes in fluoroborate salt chemistry, and assuming loop behavior and specimen behavior are the same at a given temperature, a mass balance of the system was obtained within 10%: A'system

gain

''salt

(2)

where

loss = ''system

Fig. 19.19. Loop.

Schematic of New Natural Circulation

''system

loss

= weight

loss for specimens and loop components,

222

Fig.

19.20.

Natural Circulation Loops

NCL-15 and -16 Prior to operation.

223 Table 19.8.

Composition of Hastelloy N Alloys

Chemical Content (wt %) Alloy Ni

Standard Hastelloy N

17.2

7.4

4.5

0.6

Titanium-modified Hastelloy N

13.6

7.3

<o. 1

<0.01

ORNL-DWG

0.54

0.02

0.14

0.5

68-6087

F i g . 19.21.

4 5 6 7 SPECIMEN TIME IN SYSTEM (hr)

Specimen Weight Loss a s Functions of Time and Temperature for Loop 1258 Heated Section.

and specimens fabricated from type

AWsystem

gain = weight

ACselt

gain for specimens and components,

Loop

304L stainless steel.

chemistry change in salt.

A profile of t h e weight change of loop NCL 13 at various times is given in Fig. 19.22. It can be seen that there a r e two positions (balance points) at 520 f. 15OC which have not changed weight as a function of time. One of these points is located in the upper crossover section and the other in the lower portion of the hot leg. Considering these

balance points, we estimate that about 55% of the loop is gaining weight and the remainder is losing weight. The change in chromium and iron content in the salt with t i m e is plotted in Fig. 19.23. At the end of this reporting period, 2200 hr exposure time, about 1500 mg of material has been lost from the metal. Approximately half of this stayed in the salt, and the other half deposited in the cold portions of the loop. We believe the container material, standard Hastelloy N, in loop NCL 14 is behaving like

224

h, ORNL-DWG 68-3985

N -

f G O CJ z a

650

t I

-I

600

c3 W

3-2 550

-3

500

-4 0

(00

DISTANCE (in.)

UPPER CROSS OVER

COLD LEG (VERTICAL)

BEND LOWER BEND CROSS OVER

HOT LEG (VERTICAL)

Fig. 19.22. Weight Change a s Function of Position (Temperature) for Standard Hastelloy N Specimens in Loop NCL 13.

loop NCL 13 tubing despite the results on specimens from loop NCL 14, which are titaniummodified Hastelloy N. A weight change vs distance (temperature) profile for the specimens in loop NCL 14, Fig. 19.24, illustrates the lower weight changes for the modified alloy compared with the standard Hastelloy. The value of ACsalt calculated using Eq. (2) was 150 ppm Cr, in excellent agreement with experiment. This suggests that a loop fabricated from modified Hastelloy N should behave like our test specimens of the same material. Proof of this thesis will be studied as modified material becomes available for construction of a loop. Weight change-diffusion (rate constant) calculations similar to those done on other loops were performed on loops NCL 13 and NCL 14 using Q. (3): 2 AWloss = - C @'F, T > 520 f 15OC , (3)

D = diffusion coefficient, cm2/sec,

t = rime, sec. It was found that by using the weight changes and the combined iron and chromium concentration for the loop specimens, a l m o s t identical D values were obtained for the standard and modified Hastelloy N. This is quite significant, since the iron content of titanium-modified Hastelloy N is negligible. Thus D, or in reality a rate constant K , , is independent of the concentration of chromium and iron and other diffusing species. The weight change, however, is quite sensitive to the total concentration of the migrating elements. Confirmation of this hypothesis is evident when comparing K l from our loops with a value obtained for Dc, in Hastelloy N;17 K , was found to be 2 1 O 2 exp (-57,50O/RT), and specifically K , for loops NCL 13 and 14 at 607OC is

6 = c, l/E*

where

C = concentration of Fe + Cr in Hastelloy N, g/cm3 ,

17J. H. DeVan, E f f e c t o f Alloying Additions on Cor rasion Behavior of Nickel-Molybdenmn Alloys in Fused Fluoride Mixtures, ORNL-TM-2021 (to be published).

225

ORNL-DWG 68-6088

cm2/sec, while t h e literature value for 5x Dcr is 5 x cm2/sec. We believe the sodium fluoroborate salts may be less selective in their dissolution tendency of the major elements in nickel- and iron-base alloys. However, it is emphasized that the corrosion rates of Hastelloy N in fluoroborate salts are relatively s m a l l and now appear to be leveling off. This behavior increases our confidence in the compatibility of Hastelloy N in candidate salts for molten-salt reactors. Capsule tests to evaluate the compatibility of Hastelloy N with fluoroborate salts at 607OC as a function of BF, pressure have operated 800 hr through February. There are four capsules with three titanium-modified Hastelloy N specimens in each; one specimen is in the salt, one in the BF, vapor, and one in the liquid-vapor interface. The BF, pressures used are approximately 130 torrs (the normal BF, vapor pressure at 607OC), 50, 100, and 400 psig. The time, originally scheduled for 1000 hr, has been extended to 2000 hr to permit gathering of more conclusive results. W e have observed that the specimens in the vapor phase of loops NCL 13 and NCL 14 continue to lose weight and retain a blackened appearance. These capsule tests should shed light on this phenomenon.

400

350

300

250

-z

g 200 E

k z w z u

450

100

4 000

3000

2000 TIME (hr)

Fig. 19.23.

Iron and Chromium Levels i n the NaBF4-

NaF (92-8 Mole %) Salt in Loop N C L 13.

ORNL-DWG 68-6089

2 E

- 0 W

650

z a

-1

I-

600 L

c3 W

-2

550

-3

500

-4

DISTANCE UPPER CROSS OVER

Fig. 19.24.

COLD LEG (VERTICAL)

BEN^

400

(in.) LOWER

BEN^

CROSS OVER

HOT LEG (VERTICAL)

Weight Change a s Function of Position (Temperature) for Titanium-Modified Hostelloy N in LOOPN C L 14.

226 19.10 FORCED CIRCULATION LOOP

P. A. Gnadt

W. R. Huntley

Work is in progress on a loop facility that will enable us to study the compatibility of alloys and salts under forced-circulation conditions of interest to the MSRP. A t present, we are constructing a test facility for studying the compatibility of Hastelloy N with NaBF,-NaF (92-8 mole %). The loop, MSR-FCL-1, will operate at a bulk maximum temperature of 607OC with a temperature difference of 150OC. This s a l t is very attractive as a coolant because of its low cost and low melting point. If found to be compatible and satisfactory in other respects, it will be used in the MSRE in place of the present coolant, 'LiF-BeF2 (66-34 mole %). The loop and facility designs from earlier forced

circulation loops were used, with minor modifications, in a n attempt to reduce costs. The loop MSR-FCL-1 has been designed to approximate conditions in the MSRE coolant circuit. Liquid velocities will be limited to 7 fps because a n available model L F B pump is being used. Other salient features of the loop include a BF, purging and addition system over the free liquid surface in the pump tank, means for sampling molten salt during operation, and permanent metallurgical specimens installed in the bulk stream. The salt in the loop will be heated by electrical resistance heating of the tubing wall and will be cooled by a finned air-cooled heat exchanger. Special controls are provided to reduce the possibilities of freezing the salt in the loop piping during electrical power interruptions or other equipment failures. The loop is shown in simplified scheORNL-DWG 68-2797

SALT SAMPLE LINE

AJUSTO SPEDE

OIL LINES TO PUMP

LFB PUMP 3 gpm METALLURGICAL

'METALLURGICAL SPECIMEN

HEATER LUG

RESISTANCE HEATED SECTION FREEZE VALVE

850 OF

HEATER LUG

18 p s i 0 REYNOLD'S NUMBER

-3400 VELOCITY -7

Fig. 19.25. Schematic of Molten-Soh Forced-Convection Corrosion Loop.

ft/sec

W "

1 I --I

TO PUMP

@--

LIQUID TEMPERATURE WALL TEMPERATURE

1400 1300

- 4200 -k W

1400

4000

B!-

900 800

700 600

LENGTH (ft)

Fig. 19.26.

Temperature Profile of Molten-Salt Forced-Convection Corrosion Loop.

matic form in Fig. 19.25. Figure 19.26 shows the expected temperature profile of the liquid and the metal wall. Selected engineering data for the loop design are given in Table 19.9. The data given are the design conditions, but uncertainties in the physical properties of the s a l t may cause the actual operating conditions to be different. T a b l e 19.9.

Engineering Data for Loop MSR-FCL-1

Materials, temperature, velocities Tubing and specimens Standard Hastelloy N Nominal tubing size in. OD by 0.045-in. wall Total tubing length 54 ,ft Bulk fluid temperature 607OC

'/2

(max) Bulk fluid AT Flow rate Liquid velocity

150Oc 3e m Approximately 7% fps

Table 19.9 (Coni.)

Heat transfer Heat load a t finned cooler Liquid Reynolds number Liquid film heat transfer coefficient Length of finned i - i n . - O D cooler coil Coolant air flow Coolant air AT

322,000 Btu/hr (approx

94 kw) Approximately 3450 Approximately 775 Btu hr-' ft-' (OF)-' 26 ft

3000 cfm 5ooc

Pumping requirements System AP a t 3 gpm Required pump speed

53 psi (60 ft) 4700 rpm

Circulating salt volume Tubing Pump bowl Total volume

85 in.3 47 in.3 132 in.3

228 Approximately 98% of the design modifications have been completed during the reporting period. Ninety-five percent of the loop components have been procured and fabricated. Assembly of components and installation of electrical equipment are in progress.

19.11 OXIDATION OF HASTELLOY N

B. McNabb, Jr.

tion resistance of this alloy is quite good,zo and we have been concerned with whether these chemical modifications would reduce the oxidation resistance. The rate of oxidation at a given peak temperature is usually higher when the material is being cycled periodically to lower temperatures. Although a very protective and tenacious oxide may "W.

R. Martin and J. R. Weir, Nucf. Appf. 1(2), 160-67

(1965).

Although Hastelloy N has generally performed quite well, its susceptibility to radiation dama g e ' 8 * f 9has caused us t o consider small modifications of its chemical composition. The oxida-

"W. R. Martin and J. R. Weir, Nucf. Appf. 3, 167 (1967). 2oH. E. McCoy, Jr., and J. R. Weir, Jr.. Materials Devefopment for MoftenSaft Breeder Reactors, ORNLTM-

1854 (June 1967), pp. 31-34.

Toble 19.1'0. Compositions of Test Alloys Chemical Content (wt %)

Heat No.

2477 X2497 5065 5067 5320 5911 7304 7305 Y8487 6252-2 65552

16.32 17.1 16.48 17.30 16.85 17.01 16.1 16.3 16.78 16.53 15.85

7.05 7.15 7.'25 7.40 6.92 6.14 7.16 7.27 7.32 7.26 7.43

61 71 72 73 74 75 76 102 104 105

12.14 11.69 11.74 11.87 11.81 11.89 11.93 11.2 11.74 12.28

7.77 7.93 7.90 7.90 7.94 7.89 7.84 6.92 7.14 6.72

0.089 0.10 0.089

66535 66536 66541 66548 67502 67503 67504

12.66 12.36 13.17 12.44 12.74 12.54 12.68

7.2 6.9 6.81 7.66 7.24 7.75 7.49

0.07 0.07 0.03 0.03 0.08 0.15 0.12

4.25 c0.02 3.90 4.0 0.05 0.03 2.36 2.40 4.11 0.12 4.45

0.04 0.55 0.48 0.016 0.21 0.92 0.90 0.30 0.20 0.43

0.057 0.048 0.065 0.06 0.056 0.056 0.104 0.059 0.05 0.051 0.045

0.015 0.016 0.595 0.43 c0.006 0.05 0.037 0.21 0.17

0.21 0.23 0.23 0.27 0.25 0.24 0.28 0.20 0.21 0.20

0.048 0.059 0.057 0.058 0.062 0.062 0.127 0.045

0.12 0.10

0.072 0.028 0.066 0.039 0.04 0.06 0.07

0.15 0.08 0.12

0.05

0.049

0.05

0.15

Other

0.055 0.01 0.01 0.15 0.26 0.41 0.16 0.20 0.25

0.01 0.05 0.08 0.15 0.43 0.95 1.0 0.39 0.55 0.66

<0.005

0.005 c0.005 co.005

0.01 c0.005 c0.005 <0.005 0.03 0.025 0.05

0.05 co.01 <0.01 0.02

0.06 0.08 c0.03 c0.03 0.12 0.12 0.03

0.15 0.32 0.27 0.45 0.53 0.06 <0.01

2.15 W 0.08 Hf 0.43 Hf

LI , L

229

, b

be formed at t h e peak temperature, this oxide may spa11 during cooling due t o a difference between the thermal expansion of the oxide and t h e metal substrate. When reheated t o the peak temperature, a period of rapid oxidation occurs until another protective layer of oxide is formed. We used a cycle period of 25 hr in our tests (i.e., 25 hr at temperature), after which the specimens were removed from the furnace, allowed to cool t o room temperature, weighed, and reinserted in the furnace. The t e s t times were generally 1000 hr, with the samples being weighed periodically. Test temperatures of 760 and 982OC were studied. The test specimens were '4 in. in diameter by '4 in. long; the materials studied are given in Table 19.10. The results from several m e l t s of standard Hastelloy N are shown in Fig. 19.27. These

ORNL- DWG 68- 3990

600

alloys were made by air and vacuum melting and contain various amounts of Fe, Al, Mn, and Si. However, the variation in silicon appears to be most important, with the weight loss at 982OC varying by a factor of about 6 over the range of 0 to 0.6 wt %. At 76OOC a similar trend is noted, although about 0.05% Si seems sufficient to improve the oxidation resistance appreciably. Several 2-lb laboratory melts of the modified base composition (12% Mo-7% Cr-0.2% Mn-0.05% C) were prepared with various amounts of titanium. The results of t e s t s on this material are shown in Fig. 19.28. These m e l t s are all quite low in silicon, and t h e weight losses are quite similar to those shown in Fig. 19.27 for the low-silicon alloys. The titanium additions seem to decrease the weight loss slightly, the effect being greater at 76OoC. Several commercial heats of titanium-modified materials have been tested, and the results are shown in Fig. 19.29. With the exception of the one point a t 76OoC, the oxidation rates are reasonably independent of titanium level. The rates

700 1

500

ORNL-DWG

68-3991

0.2

0.4

0.6

0.8

i.o

600

$ 400 E

0 0

2 500

0 0

300

$ 60

0 J

+ I

200

I-

c W

100

0 I I

0 0 0

0.2

0.4 0.6 PERCENT SILICON

0.8

PERCENT Ti

1.0

Fig. 19.28. Fig. 19.27. Effects of Silicon on the Scaling Resistance of Hastelloy N.

Effects of Titanium Additions on the

Scaling Resistance of Laboratory Melts of Ni-12% Mo-7% Cr-0.2% Mn-0.05%C Alloys.

230

are slightly lower than those shown in Fig. 19.28, probably because of the higher concentration of beneficial impurities in the commercial alloys. These results indicate that the removal of silicon and other residuals from the alloy is more detrimental to the oxidation resistance than the addition of small amounts of titanium. The weight losses experienced at 76OoC are still relatively low for such stringent conditions (e.g., a weight loss of 100 mg/in.2 in 1000 hr corresponds to the loss of about 50 m i l s of metal in 10,000 hr). Isothermal oxidation tests are being run on these materials, and the rates are much lower; however, the trends noted here under cyclic conditions seem to apply.

500

\ W

8 400 2

z v)

300 -I

200.

40f! 00

Fig. 19.29.

760%

(

Variation of Scaling Resistance with Ti-

tanium Content for Severol Commercial Melts of Ni-12%

Mo-7% Cr-0,2% Mn-0.05% C Alloys.

. ,

20. Graphite-to-Metal Joining 20.1 GRAPHITE BRAZING DEVELOPMENT W. J. Werner

In addition to developing brazing filler metals which directly wet and flow on graphite, we are investigating the u s e of thin (0.0002 to 0.0005 in.) pyrolytically deposited metallic coatings for en'MSR Program Semiann. Progr. Rept. Aug. 31, 1967,

ORNL4191. p. 236.

Fig. 20.1.

hancing brazability Thus, more conventional ductile, corrosion-resistant brazing alloys such as pure copper and 60 Pd-40 N i (wt X) could be used. Both molybdenum and tungsten are being deposited on high-density , low-permeability graphite, and we have found that brazing filler metals readily wet and flow on the coated graphite. Figure 20.1 shows the morphology of a thin coating of pyrolytic molybdenum on MSRE-grade graphite. X-ray studies showed that immediately

Microstructure and Morphology of 0.0002-in. Coating of Pyrolytic Molybdenum on CGB Graphite.

pol i s hed.

231

232 adjacent to the graphite the molybdenum had reacted with the graphite to form molybdenum carbide. Also, the morphology of the coating suggests a high degree of mechanical bonding between the materials. The test joint shown schematically in Fig. 20.2 was devised to check the strength of graphiteHastelloy N joints made by the above method. The graphite used in these tests was high-density, low-permeability isotropic material made by Poco. The joints were brazed with copper and pulled at room temperature. Figure 20.3 shows one of the broken test specimens. The position of fracture varied from specimen to specimen. This was taken a s an indication of the extent of cracking due to differential thermal expansion in the plane of the taper. In addition, one specimen fractured a t the metal-graphite junction during the brazing cycle. Thus more severe stresses than anticipated were encountered with this particular test specimen design. Currently, a series of coated graphite-to-molybdenum brazed specimens is being prepared. Experimental results on full-size joints (3% in. OD, $-in. wall) are preliminary. Most of the tests were direct joints of Hastelloy N to pyrolyticcoated graphite using copper as the brazing filler metal. Both induction and furnace brazing techniques were used. The induction-brazing technique is quite attractive in view of the large number of joints needed and the length and uniform configuration of the reactor components. Inductionbrazing experiments have resulted in fracture of the graphite at the graphite-Hastelloy N junction

'Manufactured by Poco Graphite, Inc.. Garland, Tex.

ORNL-DWG 68-609q

GRAPHITE

HASTELLOY N OR MOLYBDENUM

Fig. 20.2. Schematic Illustration of Graphite-Hastelloy N or Graphite-Molybdenum Test Joint.

Fig. 20.3. Graphite-Hostelloy N Brazed Specimen Tested at Room Temperature. The location of the fracture was approximately i n the middle of the taper. Joint design shown in Fig. 20.2.

during a relatively fast cooldown from the brazing temperature. Uniform heating has been difficult to obtain, and, as a result, we found portions of these joints containing unmelted brazing filler metal. We are presently upgrading the inductionbrazing unit to minimize these problems. Several direct joints of various designs have also been furnace brazed. Again the direct joints between graphite and Hastelloy N cracked in all cases, although not to the extent that the induction-brazed joints did. This experience indicates the potential advantages of the low-expansion transition metal concept. For example, a brazed graphiteto-molybdenum joint showed no visual or radiographic evidence of cracking. Radiographically, good flow of the brazing filler metal was noted throughout the joint area. Figure 20.4 shows the microstructure of such a joint (note the excellent flow of the copper filler metal). Next, a piece of Hastelloy N will be brazed to the molybdenum to form the complete transition joint, and then the entire joint will be thermally cycled.

233

Fig.

20.4.

with Copper.

Microstructure of Pyrolytic-Molybdenum-Coated Etchant:

H202 and NH40H.

ATJ Graphite Brazed t o Molybdenum Transition Piece

234 20.2 RADIATION STABILITY OF BRAZING ALLOYS OF INTEREST FOR BRAZING GRAPHITE W. J. Werner

H. E. McCoy, Jr.

W e have irradiated a series of brazing filler metals which find application for joining Hastelloy N and/or molybdenum to high-density- graphite. Hastelloy N test specimens of the Millerpeaslee design3 were used in the study. The specimens were 0.125 in. thick x 0.375 in. x 1.875 in. The joint spacing before brazing was maintained at 0.002 in. The specimens were irradiated in a single-test capsule in the poolside position in the ORR. The peak thermal flux was 6 x 1013neutrons cm'2 sec", and the peak fast (>2.9 M e V ) flux was 5 x 1OI2 neutrons cm'2 sec". The duration of the experiment was 1128 hr (time at temperature and full power), so the thermal and fast doses were 2.43 x 1020 and 2.03 x 1019 neutrons/cm* respectively. The compositions of the brazing filler metals used in the study are shown in Table 20.1. The pure copper and nickel-palladium-chromium brazes were made in a hydrogen atmosphere at 1125 and 125OOC respectively. The copper-base filler alloys were brazed in vacuum at 1200'C. After irradiation the specimens were tested on an Instron tensile machine. As-brazed control specimens were also tested. In addition, a set of specimens having the same thermal and en-

'F. M. Miller and R. L. Peaslee, Welding .I. Res. S ~ p p l 34(4), . 1 4 4 ~ - 1 5 0 (1958). ~

Table

20.1.

vironmental history as the reactor specimens is being prepared. Strength data for the reactor and control specimens are shown in Table 20.2. It can be seen that all the brazing filler metals seem to have adequate mechanical strength for nuclear system joints. We cannot obtain quantitative values for the shear strains until the width of the sheared region is determined. Complete analysis of the data must await the test results from the specimens having the same thermal history as the irradiated specimens.

Table

20.2.

Properties of Several. Brazing Alloys

Material

Maximum Shear Strength &si) 29C

7OO0C

Copper Unirradiated Irradiated

29 15

9 7

35 Ni-60 Pd-5 Cr Unkradiated Irradiated

66 66

28 12.5

68 Cu-17 Ni-10 Cr-5 Be Unirradiate d Irradiated

64 54

32 17

60 Cu-15 Ni-20 Ta-5 Be' Unirradia te d Irradiated

78 49

31 18

'Specimens failed in Hastelloy N due to the high strength of the brazing alloy.

Composition of Brazing F i l l e r Metals Used in Irradiation T e s t

Alloy Composition (wt

so)

Brazing Application

Pure copper

Pyrolytic-molybdenum-coetedgraphite to molybdenum or Hastelloy N and molybdenum to Hastelloy N

3 5 N i - 6 0 Pd-5 Cr

Graphite to molybdenum or Hastelloy N

68 Cu-17 Ni-10 Cr-5 Be

Graphite to molybdenum or Hastelloy N

60 Cu-15 Ni-20 Ta-5 Be

Graphite to molybdenum or Hastelloy N

235 20.3 GRAPHITE-TO-HASTELLOY TRANS IT ION J 01NT

Table 20.3. Expected Coefficients of Thermal Expansion for Thermal Joint Materials

J. P. Hammond

Coefficient of Thermal Expansiona

Material i

Work was initiated to develop a transition joint for joining graphite to Hastelloy N. The transition joint concept is an attractive solution to the problem of joining materials of widely differing thermal coefficients of expansion and consists in inserting a series of connected segments whose properties change gradually from one component of the juncture t o the other.4 The segments are usually fabricated by powder metallurgy and are joined by diffusion bonding or brazing. The constituents of the joint should be metallurgically compatible with one another under the fabrication and service conditions. There are two widely differing properties of graphite and Hastelloy N which give concern: the coefficient of thermal expansion and the dimensional stability under irradiation. Both can produce damaging strain at the joint interface, the former being of concern while cooling from the brazing temperature and during thermal cycling and the latter being of concern during prolonged irradiation of the graphite member.

20.3.1 Conceptual Design Two families of materials exist which offer unique characteristics for designing the type of link we require. One is the familiar class of heavy-metal alloys5 of tungsten or molybdenum

Structure

(cLin.PC)

Isotropic graphite

Hex

4.3

Hastelloy N

Fcc

12.3

Tungsten

Bcc

4.7

Molybdenum

Bcc

5.3

Graphite-carbide composites

Hex-fcc

4.3 to -5.3

Heavy-metal alloys

Bcc-fcc

‘“5 to -11

aMean value between 2 0 and 600OC.

alterable over a wide range by varying the concentration of heavy-metal phase in the structure. Since tungsten and molybdenum are near graphite in expansion coefficient (Table 20.3), the segmen t of highest hea vy-met a1 concentration wou Id be located adjacent to the graphite member, while the segment of highest nickel matrix would be next to Hastelloy N. To mollify undesirable effects stemming from placing dimensionally unstable graphite against stable heavy-metal alloy, a transition with respect to irradiation-induced dimensional instability is also introduced. This is done by incorporating one or more segments of a graphite-carbide composite between the graphite

with a nickel alloy as the binder constituent,

and the heavy-metal members. Since the graphite-

which is fabricated by liquid-phase sintering. The structure consists of round heavy-metal grains enveloped by a nickel alloy matrix. The other class of materials is the graphite-transitional metal carbide composites recently developed for rocket nozzles.6 These are fabricated by hot isostatic compaction. As noted in Table 20.3 the coefficient of thermal expansion of the heavy-metal alloy should be

carbide composite would be comprised of a highly dense isotropic-type graphite without any graphitizable binder, i t should be fairly resistant to dimensional change during irradiation. Observe in Table 20.3 that the combination of graphitecarbide composites and heavy-metal alloys spans the wide difference in coefficient of thermal expansion between the graphite and the Hastelloy N. Schematic drawings illustrating the principles of this design are shown in Fig. 20.5. Fabrication of suitable graphite-carbide composites for this joint appears straightforward using

4F. Zimmer. Metal Pro& 83, 101 (January 1963). 5F. C. Holtz. Developnent and Evaluation of Hightemperature Tungsten Alloys, Final Report, ARF-22097, LAR 59 (September 1961). 6Y.Harada, Graphite-Metal Carbide Composites, NASACr-507 (June 1966).

’

‘D. C. Carmichael, W. C. Chard. and M. C. Brockway, Dense Isotropic Graphite Fabricated by Hot Isostatic Compaction, BMI-1796 (March 1967).

236

ORNL-DWG 67-10338

GRAPHITE-CARBIDE COMPOSITES

GRAPHITE

HEAVY METAL ALLOYS

@ I

sf&ttfi!L

@ HASTELLOY N

TRAN~ITION IN COEFFICIENT OF THERMAL EXPANSION

EXPOSURE

O O O @ > O @ @ COMPOSITION OF SEGMENTS

Fig. Tap:

20.5. Transition Joint, Graphite to Hastelloy N. array o f segments comprising the joint, showing

transitions i n properties.

Middle:

transition i n irra-

diation-induced dimensional instabi I i t y o f graphitecarbide composite segments.

Bottom:

transition i n co-

efficient o f thermal expansion o f graphite-carbide composite and heavy-metal segments.

the high-temperature gas-pressure bonding technique. The joining of the components of the joint is believed to be feasible with current methods and knowledge relative to brazing and diffusion bonding.

20.3.2

Heavy-Metal A l l o y Development

Preliminary fabrication studies were conducted on tungsten heavy-metal alloys to confirm the good sinterability and ductility experienced with this class of materials in other quartems By achieving microstructures wherein a ductile nickel phase surrounds the otherwise fragile tungsten grains, exceptionally ductile composites have been obtained. A s indicated in Table 20.4, good fabri-

cation results were achieved with a nickel-iron mixture (4/1 nickel-to-iron ratio) as the binder constituent. The tungsten concentrations studied were 90, 75, and 60 wt %; these levels appear appropriate for the individual segments of the transition joint. The high toughness of these particular alloys is attested to by the high degree to which they are cold rollable without edge cracking (Fig. 2 0 . 6 ~and ~ b). We extended our studies to include alloys based on molybdenum because they are expected to show superior resistance to embrittlement by irradiation. The important consideration here is that a t the service temperature (around 7OO0C), fast displacement damage would anneal out of the molybdenum but possibly not out of tungsten. Early attempts to fabricate alloys based on molybdenum either gave rather brittle alloys or they were difficult to fabricate (Table 20.4, specimens 4 to 11). The observation that these molybdenum heavymetal alloys were prone to be brittle whereas the tungsten alloys were not is explained by comparing the nickel-molybdenum and nickel-tungsten phase diagrams. On cooling mixtures of molybdenum and nickel which have been heated at the liquid-phase sintering temperature (just above 1453OC), a eutectic containing the brittle intermetallic compound NiMo forms. This happened for the 90% Mo-7% Ni-3% Fe alloy (specimen 4 of Table 20.4) and gave a very brittle alloy. Nickel-tungsten alloys do not form an intermetallic compound during cooling from the liquid phase, and tungsten which was liquid-phase sintered with nickel-iron as the binder was free of the nickelmolybdenum intermetallic and was quite ductile. The problem of developing suitable molybdenum heavy-metal alloys thus resolves into one of finding additives to molybdenum-nickel that suppress the intermetallic-forming reaction sufficiently to preclude brittle alloys while at the same t i m e ensuring good fabricability. Palladium, platinum, copper, and tungsten were selected for examination palladium, platinum, and copper as soluble matrix phase additions and tungsten as a molybdenum displacement additive. The tungsten additive greatly reduced the amount of intermetallic in the alloys, but some of this phase still remained (specimens 5 to 8 of Table 20.4). Also, these compacts, when containing the larger amounts of binder constituent, had very narrow sintering temperature ranges,

237

Table

20.4.

Preliminary Fabrication Results on Tungsten and Molybdenum Heavy-Metal A I loys

Specimen

Composition

N 0.

(wt %)

Optimum Sintering Temperaturea PC)

Remarks

90 W-10

(7 Ni/3 F e )

1450

Good structure with one-phase matrix; ductile

75 W-25 (7 Ni/3 F e )

1435

Good structure with one-phase matrix; very ductile

60 W-40 (7 Ni/3 Fe)

1415

Good structure with one-phase matrix; very ductile

90 Mo-10 (7 Ni/3 Fe)

1400

Good structure, but matrix was a eutectic mixture; brittle

48 M o a 5 W-15 Ni-2 Fe

1410

Fair structure, but matrix contained some eutectic; not ductile

85 (7 Mo/lO W)-15 (4 Ni/l Fe)

1475

Fair structure, but some intermetallic in matrix; not doctile

70 (7 Mo/lO W ) 3 0 (4 N i / l F e )

55 (7 Mo/lO W)-45

75 Mo-20 Ni-5 P t

87 Mo-6.5

Ni-6.5

90 Mo-7 Ni-3

‘“1310

Similar to 6 and very difficult to sinter without forming ‘‘globl’c

(4 Ni/l Fe)

Similar to 6 and difficult t o sinter without forming llglob**c

1370

Poor structure, with some eutectic and porosity in the matrix; not ductile

1475

Fair structure, with one-phase matrix; not ductile

1375

Excellent structure with one-phase matrix but poor surface character and difficult-to-control sintering; ductile

aFirings were conducted in a dry hydrogen atmosphere. bComments relative t o quality of microstructure refer t o degree of success in achieving the desired dispersion of round heavy-metal grains in a nickel alloy matrix; ductility was assessed by determining amenability to cold rolling. ‘Glob refers t o a slumping rounded mass.

showing a marked tendency t o “glob.” Platinum also gave unsatisfactory results (specimen 9), but palladium and copper (specimens 10 and 11) gave the desired continuous-matrix-type structures free of any discernible intermetallic. However, the palladiumcontaining alloy, 87 Mo-6.5 Ni-6.5 Pd (wt %), was unexpectedly brittle. Since palladium h a s a rather high affinity for hydrogen, this may have been caused by having sintered in hydrogen a s an atmosphere. The copper-containing material, 90% Mo-7% Ni-3% Cu, was ductile but proved difficult to sinter without severe surface

flaws. The difficulties experienced with the latter alloy were partially attributed t o a temperature-lowering endothermic reaction which occurred just as sintering began. These preliminary results on heavy-metal alloys are encouraging, and we feel that the transitionjoint approach to joining graphite to Hastelloy N will prove practicable. A joint involving tungstenbase alloys is presently possible, and efforts to develop suitable molybdenum heavy-metal alloys will continue. Also, cursory evaluations will be made of graphite-carbide composites.

238

Fig. 20.6. Microstructures of a Number of Experimental Heavy-Metal Alloys. (a)90% W-7% Ni-3% Fa, cold rolled 60% in reduction o f area without cracking. (b) 60% W-28% Ni-12% Fe, cold rolled 95% without cracking. (c) 87% Mo-6.5% Ni-6.5% Pd, cracked w i t h first rolling pass. (d)90% Mo-7% Ni-3% Cu, as sintered, ductile. Etchant: equal parts of hydrogen peroxide

(30%) and ammonium hydroxide.

20.4 NONDESTRUCTIVE TESTING EVALUATION OF GRAPHITE-TO-METAL JOINTS

K. V. Cook We are developing nondestructive testing techniques to detect nonbond in various designs of graphite-to-metal joints. Our preliminaly studies indicate that an acceptable pulse-echo ultrasonic method can be developed for certain joint configurations. Mechanical scanning is a ,problem for the joint designs with which we have been working. We assembled a turntable device to u s e

in conjunction with our large ultrasonic tank for X-Y scanning; however, modifications will be necessary before we C M economically scan a number of joints of different configurations. W e were able to demonstrate that the ultrasonic method (using the turntable scanning device) could distinguish bond and nonbond on one joint design. The ultrasonic data indicated that bonding was accomplished on only one area of the total joint. The destructive test of this joint resulted in the separation of the graphite material in the bonded area. Figure 20.7 shows that result. As is evi-

239

Fig.

20.7.

Results of o Destructive T e s t on o

Graphite-to-Metal Brazed Joint.

T h e dark areas, where

the graphite material seporoted due to the presence of good bond, coincide w i t h the area which the ultrasonic test had predicted t o b e bonded.

dent, only one portion of the joint, the dark area, was bonded, coinciding with the prediction based on the ultrasonic examination. These results demonstrate the feasibility of the ultrasonic pulse-echo method for this particular joint configuration. No attempt h a s been made to determine the optimum sensitivity to nonbonding. A second joint design we have been working with is a tensile specimen with the graphite-tometal joint in the gage length. The maximum diameter of the specimen is 1.00 in., and the joint is a 30" cone, so that the joint diameter varies from 1 in. to a point. The graphite comprises the male part of the joint and the Hastelloy N the female part. Since both the radius of curvature of the joint and the area that reflects the ultrasonic energy change with the diameter, calibration is a problem because we encounter variations in reflected signal as the diameter changes even for complete nonbond. A further complication is the changing metal thickness from the outer surface of the specimen to the joint interface as the joint diameter changes. Attempts to evaluate this joint with the pulse-echo ultrasonic method have been unsuccessful thus far; however, we feel that it is possible to tolerate some variation in our test, and we plan to pursue this technique when modifications on our mechanical system are completed. Other joint designs to be fabricated will also require investigation by NDT evaluation techniques. These needs are being coordinated closely with the Welding and Brazing Group.

21. Support for Components Development Program 21.1 REMOTE WELDING

gas-tungsten arc welding processes have been demonstrated adequately for this alloy. The primary problem lies i n adapting the techniques to remote operation. This is not true for the gas-metal arc process, which h a s been used very little in the welding of Hastelloy N. Hence, we have in progress a modest effort to adapt this potentially desirable process to Hastelloy N. Test plates have been machined, and filler wire is being fabricated and spooled for u s e in these studies. Approximate parameters will be obtained for depositing sound weld metal. Since the maintenance may involve the welding of highly irradiated (with consequent low ductility) Hastelloy N, we plan to conduct welding studies on some irradiated creep specimens. If weld cracking appears to be serious, various approaches will be investigated to obtain a suitable joint.

R. W. Gunkel The Metals and Ceramics Division and the Reactor Division are engaged in a joint program for investigating and developing a welding process for automated remote maintenance of the MSBR. Most work so far has been in the form of literature surveys and contacts with automatic equipment manufacturers and users. Automated remote cutting and welding have been done at several facilities in the United States, and information from these programs is beneficial to us. Most of the applicable work h a s been conducted using the gas-tungsten arc (cold-wire feed) and gas-metal arc processes. Hastelloy N will be the metallic structural material to be joined, and both automatic and manual

U 240

Part 6. Molten-Salt Processing und Preparation M. E. Whatley

The development work on processes for moltensalt breeder reactors h a s undergone a significant change i n direction during this reporting period. The change resulted from the recognition and acceptance of reductive extraction as a feasible and attractive basis for processes to isolate protactinium from the reactor and, possibly, to remove fission products. A protactinium isolation flowsheet based on reductive extraction now appears applicable to a one-fluid breeder reactor. This processing development, along with the recognition that good breeding performance can be achieved with a single-fluid system, has changed the direction of the reactor development and hence has redirected the development of chemical processing methods away from a two-fluid concept. While fluorination and distillation for a two-fluid reactor will probably find application as steps in the processing of a single-fluid machine, they will be secondary to reductive extraction, which will

22.

comprise the heart of the processing flowsheet. Our work on protecting exposed surfaces from corrosion by a dynamically cast layer of frozen salt, originally developed for the fluorination step, will find application i n many steps of the new process. This section includes the presentation of some of the data on reductive extraction which lend credit t o i t s feasibility, and the calculational analy s i s of some proposed flowsheets with alternatives. Although hardware for engineering experiments with reductive extraction is being assembled, this work is not yet ready to report. Our participation i n the activities of the MoltenSalt Reactor Experiment during i t s shutdown in March 1968 will include recovery of the 235U from the present fuel salt, processing this salt to remove chromium and corrosion products, delivery of a prepared charge of 233UF,-LiF eutectic for refueling the reactor, and distillation of 48 liters of MSRE carrier salt in an experimental vacuum still.

Measurement of Distribution Coefficients in Molten-Salt-Metal L. M. Ferris

J. F. Land

Evaluation of reductive extraction methods for the processing of molten-salt breeder reactor fuels is being continued. The process now under study involves the selective extraction of uranium, protactinium, and rare earths from the molten salt using liquid lithium-bismuth solutions. Originally the main effort was devoted to the processing of

Systems

J. J. Lawrence

C. E. Schilling

two-fluid MSBR salts: specifically, the removal of protactinium from the blanket salt and the removal of the rare earths from the fuel salt. More recent

'M. W. Rosenthal, MSR Program Semiann. Progr. Rept. Aug. 31, 1967, ORNL-4191, pp. 148 and 248.

241

242 work2l3 has indicated that uranium and protactinium can probably be separated and recovered from a single-fluid MSBR salt; furthermore, a prelimidary evaluation4 showed that it should b e possible to engineer such a process. Consequently, most of the current effort is directed toward development of processing methods for single-fluid MSBR fuels. In order to obtain a more detailed evaluation of these methods, accurate distribution coefficients for uranium, thorium, and the rare earths must be determined. Our program is, therefore, designed to provide these data.

22.1 EXTRACTION OF URANIUM AND RARE EARTHS FROM FUEL SALT OF TWO-FLUID MSBR'S The equilibrium distribution of a component M between a molten salt and a liquid metal phase can be expressed as a distribution coefficient D, that is defined as

mole fraction of component M in metal phase mole fraction of component M in salt phase

Thermodynamic treatments (supported by experimental data) of the equilibria involved with salts containing LiF indicates that, at a fixed temperature, each distribution coefficient should vary with the lithiun concentration in the metal phase according to the equation log D PA

log XLi(,)

+ log I

if the concentration of M in each phase is low. In Eq. (l), n is the valence of the component (as i t s fluoride, MF,) i n the salt, XLi(,) is the equilibrium lithium concentration in the metal phase (atom fraction), and I is a constant. Presentation of equilibrium distribution data in the manner indicated by Eq. (1) is desirable; however, because of the preliminary nature of our initial experiments, this is not possible. In our ex'J.

periments, usually only a few data points were obtained at a given temperature. Our data can be compared with those obtained by others either i n terms of the quantity Q, defined as

4M. E. Whatley, L. E. McNeese, and J. S. Watson, personal communication, Feb. 6, 1968. 'W. R. Grimes, Reactor Chem. Div. Ann. Progr. Rept. Dec. 31, 1966, ORNL-4076, p. 34.

or as the difference in standard reduction potentials (as defined by Moulton '):

E;,,

A E i = (RT/nF) In D,

Values for Q and AEi can be calculated from a single distribution coefficient and the corresponding equilibrium lithium concentration. From Eqs. (2) and (3), the relationship between Q and AE; is found t o be log Q =

nF AE; - n 1% XLi, 2.303RT

(4)

in which XLi, is the mole fraction of LiF in the salt. Data obtained on the distribution of several solutes between two-fluid MmR fuel salt, LiF-BeF, (6634 mole %), and lithium-bismuth solutions are given i n Tables 22.1 and 22.2. In calculating values of Q and mi, n was assumed to be unity for lithium and sodium, 2 for europium, 3 for lanthanum, and 4 for uranium. Average AEi values (Table 22.1) are given at those temperatures where more than one distribution coefficient was obtained. The values obtained for lanthanum and europium are in reasonable agreement with those reported by Moulton and Shaffer. However, our value for uranium at 6OO0C (based on three data points, Table 22.2) is 0.14 v higher than the value reported by Moulton and Shaffer. The reason for this difference is inexplicable at present. It is interesting to note that the AE,,' values appear to be practically constant over the temperature range from 500 to 70OoC. If is constant, plots of log Q vs 1/T should be linear [see Eq. (4)I. Plots

H. Shaffer, memo to W. R. Grimes, Nov. 27, 1967.

3C.J. Barton, personal communication, December 1967.

'M. W. Rosenthal, MSR Program Serniann. Progr. Rept. Feb. 28, 1967, QRNL-4119, p. 150. 'D. M. Moulton and J. H. Shaffer, unpublished results, Dec. 14, 1967. *M. W. Rosenthal, op. cit., pp. 155-56.

243 T a b l e 22.1.

hEi

Valuesa for the Distribution of Various Solutes Between Different Fluoride Salts and

Salt Composition (mole 5%)

L i th i um- Bi smuth Solut ions

AEi (v>

Temperature C째C>

LiF

B F 2

ThF4

500

0.27

600b

0.32

600'

600b

0.54

600

0.68

600

0.66

602

605

675

700

0.30

700b

0.34

0.40

0.30

0.45

0.32

0.36

0.33

0.41

0.31 0.44 0.60

0.27

0.20

0.43

where M is the other metal involved; see text.

bD.M. Moulton and J. H. Shaffer, ref 7. =D.M. Moulton and J. H. Shaffer, ORNL-4191, p. 155; ref 8.

of the data for europium and lanthanum are shown in Fig. 22.1; the lines were placed using average values of AE; of 0.43 and 0.30 for lanthanum and europium respectively (Table 22.1). (Moulton's data for cerium give a very kood straight line over the temperature range from 500 to 70OOC.) Although the data points are highly scattered, it does appear that Q decreases with increasing temperature. This dependence of Q on temperature shows that the rare earths are less easily extracted at the higher temperatures.

The data given in Tables 22.1 and 22.2 show that uranium can easily b e removed from the fuel s a l t of a two-fluid MSBR by reductive extraction. Furthermore, separation from the rare earths is very good; at 6OO0C the separation factors (Du/DLn) are at least 1000. The data also show that, once the uranium has been extracted, the rare earths can b e removed from the salt by increasing the lithium concentration in the metal phase.

22.2 EXTRACTION OF URANIUM FROM SINGLE-FLUID MSBR FUEL Two experiments involving equilibration of simulated single-fluid MSBR fuel salt with lithiumbismuth solutions at 6OO0C have been completed. In each experiment the salt, initially LiF-BeF2ThF,-UF4 (about 70-19-11-0.3mole So), and about an equal volume of bismuth were heated in a mild steel crucible under argon to 60OoC; then lithium ( a s Li-Be alloy) was added to the bismuth i n s m a l l increments, and the equilibrium distribution of the various components between the two phases was determined. The salt in one experiment contained 250 ppm LaF,, while EuF, was present in the other system. The distribution coefficients obtained are given in Table 22.3. Those for uranium can be represented by the relationship log D,

log CLi + 7.874 ,

244

Table

22.2. Effect of Temperature on the Distribution of Europium and Lanthanum Between L i F - B e F 2 (66-36 Mole %) and Li-Bi Solutions

Temperature

ec>

Li Concentration in Metal Phase

10'1~~

Q" DEU

DLe

(at. %>

Eu x

io3

La x

lo8

500

12.94

4.02

14.3

8.83

583

11.68

4.95

16.4

6.67

583

11.68

3.46

17.1

600

11.45

0.0133

1.74

600

11.45

0.0232

5.41

600

11.45

0.0313

600'

11.45

600'

11.45

11.4

600'

11.45

11.4

602

11.43

2.56

605

11.39

0.119

0.092

0.55

605

11.39

0.119

0.90

5.34

605

11.39

0.208

0.705

0.78

605

11.39

0.298

5.12

1.93

605

11.39

4.84

6.36

2.72

6 08

11.35

0.85

0.82

11.3

675

10.55

0.238

0.487

1021

0.36

675

10.55

0.298

0.672

1788

0.25

675

10.55

0.298

0.92 6

0.35

675

10.55

0.416

1.09

0.15

675

10.55

0.506

2.01

0.16

700

10.28

5.11

700'

10.28

7.64

800'

9.32

5.53

14.3

53.6 6.68

5.41

8.66

0.29

2.16

8.26

3.32

aQ = DLnXL:cSI,; Ln = Eu or L a ; the values of n used for Eu and La were 2 and 3 respectively. 'Calculated from data given by D. M. Moulton and J. H. Shaffer, ref. 8. 'D. M. Moulton and J. H. Shaffer, ref. 7.

245

Table 22.3. Distribution of Lithium, Uranium, Thorium, and Europium Between LiF-BeF2-ThF,

(70-19-1 1 Mole %) and Bismuth Solutions

6OO0C

Experiment

No.

Lithium Concentration in Metal Phase (at. So)

0.000672

0.00021

0.00119

0.00013

0.0032 1

0.0638

0.00579

0.0267

0.00584

0.311

0.00726

0.132

0.00747

0.238

0.00873

0.134

0.00988

2.47

0.00997

0.827

0.01 12

1.12

0.0128

1.47

0.0133

2.28

0.0136

4.08

0.0176

5.07

0.0182

15.1

0.0285

15.2

0.0376

26.1

0.0570

102

0.00131

0.0623

129

0.00229

0.0786

168

0.0032 7

0.0816

155

0.00573

0.0855

0.00802

0.0939

0.00998

0.011

0.0997

0.00712

0.014

0.114

0.175

333

DTh

0.0131 759

0.0152

0.0041

246 ORNL-DWG 68-6095

TEMPERATURE ("C) 500

600

/ I

I I

0.1 A DATA OF D.M. MOULTON, ORNL-4191, p. 155. 20 7 D.M. MOULTON. PERSONAL COMMUNICATION.-

0 x

2 9

'O.OOO/rCK)

Fig. 22.1.

Dependence of log Q on Tempemture.

i n which CLi is the lithium concentration in the metal phase (at. %). This equation was obtained by visually fitting what appeared to be the best line of slope 4 through the points in the lithium concentration range of 0.002 to 0.03 at. %, where the analyses were the most accurate. These data are shown in Fig. 22.2; the distribution coefficients obtained using LiF-BeF2 (6634 mole %) as the salt phase are included for comparison. The distribution coefficients for thorium (Table 22.3) can be represented by the equation log DTh = 4 log CLi + 2.00, which was also obtained by a visual fit of the data (Fig. 22.2). The line through the three points obtained for europium (Fig. 22.2) was drawn with a slope of 2. In these experiments, a s lithium-bismuth alloy was added to thesystem, the thorium concentration in the metal phase increased to what appeared to be a limiting (steady-state) value of 1700 f 100 ppm. The corresponding uranium and lithium con-

Fig. 22.2. Effect of Lithium Concentration i n Metal Phase on the Distribution of Uronium, Thorium, and ) Europium Between LiF-BeF2-ThF4 (70-19-1 1 Mole I

and Bismuth a t 60OoC.

centrations were 2500 f 300 and 50 f 10 ppm respectively. The distribution coefficients a t steady state were about 0.014 for both thorium and europium and less than 0.07 for lanthanum (the lanthanum concentration in the metal phase was always less than 4 ppm). The data indicate that the solubility of thorium in bismuth is markedly depressed by the presence of uranium and/or lithium, since the solubility of thorium in pure bismuth at 6OOOC is 3000 to 3900 ppm. '* l o 'M. Hansen and K. Anderko, Constitution of Binary Alloys, 2d ed., p. 341, McGraw-Hill. New York, 1958. 'OR. P. Elliott, Constitution of Binary Alloys, First Supplement, p. 202, McGraw-Hill, New York, 1965.

247

Th data re wted ab ve s h w that uranium c n be easily extracted from single-fluid MSBR salt, leaving thorium and rare earths in the salt; the separation factors (D,/DTh and D,/DEU) are at least lo4. These data also show by extrapolation t o low lithium concentrations that the rare earths can probably b e separated from thorium; however, i n order to achieve reasonable separation factors in this system, the thorium concentration in the metal phase will probably have to b e somewhat lower than its maximum solubility. 22.3 EXPERIMENTAL PROCEDURE AND LITHIUM “LOSS”

’ ’

It has been noted previously * that during reductive extraction experiments up t o 75%of the lithium added t o the system was consumed in an inexplicable manner. In m o s t of these experiments, the lithium was added as the pure metal. The results of our latest experiments show that this probl e m can be largely eliminated when the lithium is added to the system as pieces of frozen lithiumbismuth alloy. An example of the equivalent balances that can be achieved by this method (Fig. 22.3) was derived from data obtained i n an experiment in which LiF-BeF ,-UF, (initially 66-34-0.3 mole %) was equilibrated with liquid bismuth to which lithium-bismuth alloy (7 at. % Li) was added periodically. The figure is a plot of the milliequivalents of Li + U found i n the metal phase [4(mgatoms u) + mg-atoms Lil v s the milliequivalents of lithium added to the system as lithium-bismuth alloy. The first 10 meq of lithium added was probably consumed i n reactions with FeF2, OH-, and other

easily reduced species that were present in the salt. Then, a s more lithium-bismuth alloy as added, the lithium was consumed by reduction of uranium fluoride from the salts. The consumption of lithium was nearly stoichiometric for the reaction 4Li

4LiF

+U .

This behavior indicates that the uranium in the salt phase was still in the tetravalent state (undoubtedly as UF,). After the uranium was extracted into the metal phase, the increase i n milliequivalents of Li + U was due entirely t o the addition of more lithium. As may be seen i n Fig. 22.3, the equivalent balance was quite acceptable (Le., the slope of the line was unity) throughout most of the experiment.

ORNL-DWG 68-6097

50 U

$40

-I

,5 30 W

8 20 3 B 3 IO

Li ADDED TO SYSTEM AS L i -Bi

Fig.

“R. B. Briggs, MSR Program Semiann. Progr. Rept. Aug. 31, 1966, ORNL-4037, p. 142.

+ UF,

22.3.

Reductant Balance i n Experiment Involving

of Lithium-Bismuth Solutions with LiFBeF2-UF4 ( I n i t i a l l y 66-34-0.3 Mole 56) at 600°C.

Equilibration

ALLOY (mea)

23.

Protactinium Removal from a Single-Fluid MSBR L. E. McNeese

Laboratory experiments have shown that protactinium anduranium can be extracted into bismuth from molten salt which contains fluorides of lithium, beryllium, and thorium by using metallic thorium as the reductant. In the proposed flowsheet (Fig. 23.1), a salt stream from the reactor enters the bottom of the extraction column and flows countercurrently to a stream of bismuth containing reduced metals. Ideally, the metal stream entering the top of the column contains sufficient thorium and lithium to extract only the uranium entering the column. The system exploits the fact that protactinium is less noble than uranium but more noble than thorium. Hence,

M. E. Whatley in the lower part of the column, uranium is preferentially extracted from the incoming salt, while the protactinium progresses farther up the column to where it is reduced by thorium. In this manner, protactinium refluxes in the center of the column, and relatively high protactinium concentrations result. A retention tank is provided at the center of the column, where the maximum protactinium concentration occurs in the salt, to retain the protactinium until it decays to uranium. An essential part of the flowsheet is an electrolytic oxidizer-reducer which serves the dual purpose of recovering the extracted uranium from the metal stream leaving the extraction column

F2 OXIDIZER (ANODE)

REDUCER (CATHODE)

Fig.

23.1. Single-Fluid MSBR Processing by Reductive Extraction.

249

and of preparing the lithium-thorium-bismuth stream which is fed to the extraction column, The metal phase containing the uranium extracted in the column can serve as the cell anode, where uranium and lithium will be oxidized to UF, and LiF. Salt from the top of the extraction column serves as the cell electrolyte and first passes over a pool of liquid bismuth which serves as the cathode into which thorium and lithium are reduced. Typical performance of the protactinium isolation system is shown in Fig. 23.2 for a case which will be taken as the reference case. The minimum reactor protactinium concentration is obtained when the bismuth flow rate is just sufficient to extract the uranium entering the system. At slightly higher bismuth rates, protactinium will also be extracted, since it is the next component in order of decreasing nobility. At bis-

muth rates slightly lower than the optimum rate, some uranium will not b e extracted; this uranium and most of the protactinium will flow out the top of the column. In either case, some protactinium is allowed to return to the reactor, and the effectiveness of the system is diminished. The protactinium isolation system becomes ineffective almost immediately for bismuth flow rates lower than the optimum rate; for bismuth flow rates higher than the optimum, the reactor protactinium concentration increases from the minimum value of 22 ppm at the rate of 28 ppm for each percent increase in metal flow rate for the conditions of this particular case. Similar effects would be produced by variations in salt flow rate or in total concentration of reduced metals (equivalents of reduced m e t a l s per mole of bismuth) fed to the column. Calculated concentration profiles in the extraction column are shown in Fig. 23.3 for

ORNL-DWG

68-42907A

Fig.

23.2.

Variation

of Protactinium Concentration in Reactor and Protactinium Decay Tank with Bismuth F l o w Rate,

250

OANL-DWG 67-l2893A

10-0

90-2

10-6

to-8

2 ::

,o-'O

Po = 0.00132

to-=

6 ' 6

Fig.

23.3.

6 8 STAGE NUMBER

cludes loss of protactinium, 233U, or other components of the reactor fuel salt. However, the protactinium removal efficiency is undesirably sensitive to minor variations in operating conditions such a s s a l t or bismuth flow rate and concentration of reduced metals i n the bismuth stream fed to the extraction column. Methods for making system performance less sensitive to minor variations in operating conditions have been considered. Removal of uranium from the center of the column by fluorination of molten salt was found to make steady-state system performance insensitive to s m a l l changes in operating conditions; the concentration of UF, in the fluorinator off-gas was found to be extremely sensitive to changes in operating conditions and can be used as the primary process control point. The'effectiveness of a system stabilized by uranium removal was compared with that of a nonstabilized system by assuming the error in control of bismuth flow rate to be distributed normally around a best mean flow rate FMB opt with standard deviation 0,so that an average reactor protactinium concentration F could be calculated by

Calculated Concentration Profiles i n Re-

ductive Extraction l o w e r .

where

x = FMB - FMBopt/u,

FMB = bismuth flow rate, steady-state operation under optimum conditions. The concentration of uranium in the salt decreases from the reactor concentration of 3 x mole fraction to 6.3 x mole fraction at the inlet to the protactinium decay tank, whereas the protactinium concentration increases from the reactor concentration of 2.2 x mole fraction to 1.32 x mole fraction at the inlet to the decay tank. The concentrations of protactinium and uranium in the decay tank are 1.3 x and 8.3 x lo-' mole fraction respectively. Above the decay tank the uranium and protactinium concentrations decrease steadily to negligible values. The flowsheet h a s several very desirable characteristics, which include a neglibible holdup of fissile 233uin the processing plant; an almost immediate return of newly produced 233U to the reactor system; and a closed system which pre-

c(x)

= steady-state

reactor protactinium concentration at bismuth flow rate corresponding to x.

The average reactor protactinium concentration is shown in Fig. 23.4 as a function of the standard deviation in bismuth flow rate for cases with stabilization by uranium removal as well as for two cases with no uranium removal which have protactinium processing cycle t i m e s of three days {reference case) and one day. For a standard deviation of 0.35% of FMBopt (optimum bismuth flow rate), the minimum reactor protactinium concentration (22 ppm) is obtained if the fraction of uranium removed from the inlet to the decay tank is 2% or greater; with no uranium removal the average concentration is 38.5 ppm. With a protactinium processing cycle t i m e of one day and a standard deviation of 0.35%, an average con-

251

ORNL-DWG 68-2090A

280 f = FRACTION OF URANIUM REMOVED FROM STREAM ENTERING DECAY

-1

240

TANK Po PROCESSING CYCLE TIME = 3 days Po PROCESSING CYCLE TIME= 4 doy

460

i20

NO REMWAL

0.0707

0.4

0.2

1.O

0.5

( u /FMBOp,)

Fig.

2.0

5.0

io0 (%I

23.4. Variation of Average Reactor Protactinium Concentration with Standard Deviation of Bismuth F l o w

Rate for Cases w i t h Stabilization by Fluorination and Without Stabilization.

centration of 24.5 ppm is obtained. A protactinium isolation system having a one-day processing cycle t i m e with no uranium removal produces roughly the same average reactor protactinium concentration as a system having a three-day cycle t i m e with 1%removal of uranium from the s a l t stream entering the decay tank for standard deviations larger than about 0.35%.

In choosing the optimum protactinium isolation system, the relative costs associated with higher processing rates, with closer process control, and with increased uranium removal must be considered as well as the value of obtaining a given average reactor protactinium concentration.

24. Continuous Fluorination of Molten Salt B. A. Hannaford

Uranium present in the fuel stream of an MSBR must be removed prior to the distillation step since UF, would not be completely volatilized during this operation. Equipment is being developed for the continuous removal of UF, from a salt stream by countercurrently contacting the salt with F, in a salt-phase-continuous system. The equipment will be protected from corrosion by freezing a layer of salt on the vessel wall; the heat necessary for maintaining molten salt adjacent to frozen salt will be provided by the decay of fission products in the salt stream. Recent development work has been directed toward demonstrating operability of a frozen-wall fluorinator using countercurrent flow of molten s a l t and an inert gas. The experimental equipment consisted of a 5-in.-diam, 8-fthigh column fabricated from sched 40 nickel pipe (Fig. 24.1). An internal heat source consisting of three Calrod heaters contained in a t-in.-diam sched 40 Inconel pipe is used to simulate a volume heat source in the molten salt. Two sets of internal thermocouples located near the center of each of two test sections indicated the radial temperature gradient, from which one could infer the location of the interface between molten and frozen salt. Each test section was independently cooled by air flowing through spirally wound "/-in.- diam nickel tubing. Additional Calrod heaters were wound on the external surface of the fluorinator to provide auxiliary heat during heatup and to provide control of temperatures a t the ends of the column. A 66-34 mole % LiF-ZrF, mixture, which has a liquidus of 595OC and a phase diagram similar to the LiF-BeF, system, was metered from the feed tank for periods a s long a s 5 hr, which allowed collection of data for a 1- to 2-hr period of steady-state operation.

L. E. McNeese

The principal objective of the experiments, demonstrating that a layer of frozen salt could be formed and maintained under approximate operating conditions, was achieved. Experimental conditions are compared in Table 24.1 with reference conditions for a fluorinator having a salt throughput of 1 5 ft3/day and an inlet uranium concentration of 0.8 kg per cubic foot of s a l t with a 50% F, utilization. In general the frozen wall thickness and temperature profiles in the frozen s a l t were in good agreement with the values predicted by relations obtained by assuming purely radial heat flow from a volume heat source. Frozen wall thickness ranged from 0.3 to 0.8 in. The effect of heat generation in the layer of frozen salt was not simulated in these experiments. The thermal conductivity of the frozen salt was calculated for each run from the experimentally determined temperature gradient and the measured heat flux; the relative agreement of calculated values was taken to b e indicative of consistency of the experimental data. Thermal conductivities calculated from the upper test section data were closely grouped around 0.75 Btu hr" ft" (OF)"; however, values from the lower section were more widely scattered and were generally about 100% higher. The heating-cooling system used on the column produced some variation in external wall temperature (and hence frozen wall thickness); in a typical run the difference between s a l t liquidus and wall temperature ranged from 85 to 140OC. Protection of the fluorine inlet nozzle from corrosion is an anticipated problem associated with operation of a frozen-wall fluorinator. A possible solution incorporated in the present system consists in introducing the fluorine through a short

252

253 ORNL- DWG 68- 3340

-VENT

SALT

ARGON

Ic - - - -

I -

S A L T RECEIVER T 625OC

Fig.

24.1.

FROZEN

6OO0C

FEED TANK T 625OC

Experimental Equipment for Studying Formation of Frozen Salt Layer for Corrosion Protection.

Height of column, 8 ft; height of each test section,

1.75 ft.

254

Table

24.1.

Comparison of Experimental ond

Reference Conditions for Fluorination of

IS ft3

of Molten Salt

pertDay'

Experimental Salt flow rate, liters/hr

-3.3

Gas flow rate, standard lite rs/min

0.5-2.0

Heat flux, w per foot of column height

600-1600

Reference 17.7

(At)

2.0 (F2) -2000

'15 ft3/day corresponds to processing fuel stream of a 1000 Mw (electrical) two-fluid MSBR.

section of 3-in.4iam pipe which intersects the fluorinator at a 4 5 O angle, as shown in Fig. 24.1.

The inlet section would be protected from c o r m sion by a layer of frozen salt as i n the fluorinator. Tests with the present system indicate satisfactory operation when the surfaces of the inlet section are covered by a layer of frozen salt produced by maintaining wall temperatures below the salt liquidus. Heat is supplied to this section in the present case as a result of turbulence in the molten salt caused by bubbles; in an actual case, heat would be generated in the salt as a result of fission product decay. This method of g a s introduction appears to be feasible, although it will not produce smalldiameter gas bubbles.

In future experiments, the heat flux will be increased to the reference value, and salt flow rate will be increased to approximately 50% of the reference rate.

25. Relative Volatility Measurements by the Transpiration Method F. J. Smith

C. T. Thompson

Data for three LiF-BeF, solutions are given in Table 25.1. The composition LiF-BeF, (90-10 mole %) is approximately the one expected in the still pot during vacuum distillation and gives LiFBeF, (66-34mole %) as the condensate.' The relative volatilities of BeF, with respect to LiF obtained in experiments with LiF-BeF, binary systems are in reasonable agreement with those reported by Cantor,, who a l s o used the transpiration method. For example, Cantor obtained values of 4.28 and 3.75 a t 1000째C for LiF-BeF, (85-15 mole %) and LiF-BeF, (90-10mole %) respectively; the corresponding values from our work were about 3.8 and 3.77 (Table 25.1). The value obtained with LiF-BeF, (90-10mole %) is somewhat lower than the average value of 4.71 obtained by Hightower and McNeese, who used an equilibrium still method, and is higher than the values obtained when the salt contained small amounts of RbF, CsF, ZrF4, or UF, (Table 25.1). This scatter in values is not surprising when one considers that small variations in the composition of the liquid and/or vapor lead to large changes in the relative volatility. For example, Hightower and McNeese' reported that the vapor in equilibrium with LiF-BeF2 (90-10mole %) a t 1000째C was LiF-BeF, (66-34 mole So). This gives a value for

Liquid-vapor equilibrium data necessary for the evaluation and development of the possible distillation step in the processing of two-fluid MSBR fuel salt, LiF-BeF, (66-34mole %), are being determined by the transpiration method. In the absence of any information regarding complex molecules in the vapor phase, the partial pressures of LiF, BeF,, and solute fluorides were calculated on the assumption that only monomers existed in the vapor. In each experiment, the apparent partial pressures P could be described adequately by linear expressions, log P (mm Hg) = A

- B/T

L. M. Ferris

(OK),

in which A and B were constants over the temperature range investigated, 900 to 1050OC. The data are summarized in Table 25.1. Relative volatilities are also included, as well as effective activity coefficients. For a multicomponent system, the activity coefficients of component A a t each temperature are given by D

where N , is the mole fraction of A in the melt and P i is the vapor pressure of pure A. Relative volatility is defined by

34/66 4.64. 10/90

a=-= *

'J. R. Hightower, Jr., and L. E. McNeese, Measurement of the Relative Volatilities of Fluorides of Ce, La, Pr, Nd, Sm. Eu, Ba, Sr, Y, and Zr in Mixtures of LiF and BeF,, ORNL-TM-2058 (January 1968). ,R. B. Briggs. MSR Program Semiann. Progr. Rept. Feb. 28, 1966, OR-3936, p. 128.

where u A Bis the relative volatility of A with respect to B, y is the mole fraction of the designated component in the wpor phase, and x is the mole fraction in the liquid phase.

255

256 Table 25.1.

Apparent Partial Pressures, Relative Volatilities, and Effective Activity Coefficients in LiF-BeFZ-Metal Fluoride Systems

Salt Composition (mole So) LiF

BeF,

90 95 90

89.6

86.4

89.9

Species

3d Component

10 5 0.02 UF,

9.9

0.5 UF,

4.0 UF,

9.6

0.09 RbF

0.03 C s F

0.083 ZrF,

Effective Activity Coefficient at

Apparent Pressure'

lOOo0C

Relative Volatility with Respect to LiF a t 1000째C

LiF BeF,

8.497 7.983

11,055 10,665

1.60 4.42 X lo-,

3.82

LiF BeF,

7.604 8.707

10,070 11,884

1.30 3.55 x 10-2

3.77

LiF BeF,

8.804 11.510

11,505 15,303

4.33 x

LiF BeF,

9.481 9.339 4.361

12,386 12,411 12,481

1.33 5.96 x lo', 7.36 x 1 0 - ~

8.384 7.421 6.686

10,987 10,112 13,443

4.65 X 1.09 x

lo-, lo-,

4.78 4.2 x lo',

10.790 10.177 10.272

13,992 13,726 16,786

1.55 3.84 x 10-2 1.25 x lo-,

3.42 4.2 X lo',

8.286 6.596 5.187

10,811 10,552 8,907

1.47 3.11 X 2.19

2.93 24.7

LiF BeF, CsF

9.654 8.310 0.819

13,459 11,313 3,375

1.99 4.07 x loa2 1.17

2.82 95.1

LiF BeF, ZrF,

7.915 7.167 13.095

10,358 10,070 20.382

1.41 2.83 X loa2 5.39 x lo',

2.77 2.19

UF4 LiF BeF, UF4 LiF BeF, UF4 LiF BeF, RbF

1.30

lo-,

4.40 6.19 2.9 x lo-,

1.34

'Log p (mm) = A B/T (OK). Temperature range: 900 to 105OOC. It was assumed that LiF, BeF,, and the solute fluorides existed only as monomers i n the vapor.

On the other hand, Cantor, reported that the vapor in equilibrium with LiF-BeF, (88-12 mole %) has the composition LiF-BeF, (67-33 mole %), which leads to a value for

33/67 a = -- 3.6. 12/88 The partial pressure data obtained in our work are incompatible with the total pressure data of C a n t ~ r . He ~ reports the total pressure of LiFBeF, (90-10 mole %) to be 1.8 mm Hg a t 1000째C.

For the same system a t 1000째C, we obtained PLiF= 0.55 and PBeF = 0.23 mm Hg, cor2

responding to a total pressure of 0.78 mm (assuming that no association or dissociation occurred in the vapor phase). Association (known to occur4 in the vapor phase above pure LiF) or

3W. R. Grimes, Reactor Chem. Div. Ann. Progr. Rept. Dec. 31, 2965, ORNL-3913, p. 24. ,R. S. Scheffeeand J. L. Margrave, 1. Chem. Phys. 31, 1682 (1959).

257 complexation (observed m a s s spectrometrically in the vapors above LiF-BeF, solutions3) would make the total pressure even lower than that predicted by our transpiration experiments. Our results a r e also in disagreement with those of Cantor at several other LiF-BeF, compositions. The reason for this disparity is not obvious. The activity coefficients for BeF, are in good agreement with the values obtained by Kelly3 using distillation data and assuming unit activity for LiF. Experiments performed with a n LiF-BeF, (9010 mole %) solvent containing UF4 a t concentra-

tions of 0.02, 0.5, and 4.0 mole % (by analysis), respectively, gave relative volatilities of UF, with respect to LiF (Table 25.1) which confirmed the earlier assumption that adequate uranium recovery cannot be achieved in the distillation step. The experimental relative volatilities of RbF and C s F with respect to LiF are within a factor of 2 of the theoretically predicted values (assuming ideal behavior of the system). Hence, as previously assumed, these two fission products will probably s e t the discard rate. The relative volatility of ZrF,, present a t a concentration of 0.083 mole %, was 2.19.

Distillation of MSRE Fuel Carrier Salt L. E. McNeese

J. R. Hightower

We have described previously' equipment for study of low-pressure distillation of MSRE fuel carrier salt which includes a 48-liter feed tank, a 12-liter still pot, a condenser, and a 48-liter condensate receiver. The equipment has been installed and checked out in a test facility to perform nonradioactive experiments prior to operation with irradiated MSRE fuel carrier salt. Two experiments have been made to date. During nonradioactive operation, four 48-liter batches of MSRE fuel carrier s a l t (65-30-5 mole % LiF-BeF,ZrF,) will b e distilled at 1000°C and at a pressure of about 1 mm Hg. The thermal insulation and heaters will then be removed from the still, and a thorough examination of the equipment will be made. After examination the still will be moved to a cell a t the MSRE s i t e for distillation of a 48-liter batch of fluorinated fuel s a l t from the reactor. The first distillation run was completed in 83 hr using approximately 48 liters of MSRE fuel carrier salt. At the beginning of the run, 9.4 liters of salt was transferred to the still pot, and the condenser pressure was lowered to 2.0 mm Hg. Startup of the equipment was hampered by unsatisfactoy operation of the liquid level probes in the still pot. It is believed that argon dissolved in the molten salt formed bubbles on the liquid level probes initially as the pressure was reduced. The s y s t e m was held a t 900°C and 2 mm Hg for approximately 4 hr, after which time the probes performed adequately. The system was then operated with a still pot temperature of 900 to 950OC and a condenser pressure of 0.65 to 2.0 mm Hg for approximately 40 hr. Salt was trans-

ferred to the still pot periodically in order to maintain a volume of about 9 liters. This operating period reduced the concentration of BeF, in the still pot from the initial concentration of 30 mole %. The still pot temperature was raised to 990°C, and distillation rates were measured during a 40-hr period at condenser pressures of 0.5, 0.3, and 0.055 mm Hg. Results at these conditions are summarized in Table 26.1. Distillation rates were calculated from the rate of change of salt level in the feed tank and condensate receiver. During the last 40 hr of operation the still liquid level and salt feed rate were controlled automatically. Equipment operation was-smooth. Four condensate samples were taken and have been submitted for analysis. A t the end of the run, approximately 8 liters of the initial salt mixture was used t o flush the high-melting salt from the still pot in order to produce salt having a liquidus of less than 700°C. For the second run, approximately 45 liters of MSRE fuel carrier salt (65-30-5 mole % LiF-BeF,ZrF,) was transferred to the feed tank. Salt was transferred to the still pot to yield a volume of about 9 liters. A still pot temperature of 900°C

Table 26.1.

Distillation Rates and Operating

Conditions for Distillation of MSRE Carrier Salt

'Ondenser

Pressure (mm Hg)

0.5 0.3 0.055 0.07

'MSR Program Semiann. Progr. Rept. Feb. 28, 1967, OKNL-4119, p. 211.

258

Distillation Rate k g h r f t 3 day-'

1.54 1.60 1.67 2.02

1.15 1.20 1.20 1.51

ft-,

Fuel

Still Pot Run Temperature Time

CC>

990 990 990 1005

(hr)

7.2 9.5 16.5 29.8

259

and a condenser pressure of 2 mm Hg were held for 1.5 hr, after which (over a 13-hr period) the still pot temperature was slowly increased to 1005OC and the condenser pressure was decreased to 0.07 mm Hg. These conditions were maintained for about 30 hr. The distillation rate during this period is given in Table 26.1. Still pot liquid level and salt feed rate were controlled automatically during the entire run. Minor difficulty was experienced in taking condensate samples when s m a l l amounts of salt vapor condensed in the 1.5-in. line through which samples are taken. Two condensate samples were obtained. Under the operating conditions in the still, the distillation rate is controlled by the condition that pressure drop in the passage connecting the vapor-

ization and condensation surfaces equals the difference between t h e vapor pressure of salt in the still pot and the pressure at the lower end of the condenser. For this reason, distillation rate should b e essentially independent of condenser pressure for condenser pressures much lower than the salt vapor pressure (1.0 to 1.5 mm Hg), as was o b served. With condenser pressures much lower than the salt vapor pressure, distillation rate should be proportional to salt vapor pressure and hence quite dependent on still pot temperature. A 21% increase in distillation rate was observed as the still pot temperature was increased from 990 to 1005OC; the corresponding increase in salt vapor pressure is 28%.

27. Protactinium Removal from a Two-Fluid MSBR J. S. Watson

M. E. Whatley

A higher breeding ratio can be achieved in molten-salt reactors i f a low protactinium concentration is maintained in regions of high neutron flux so a s t o avoid parasitic capture by the protactinium before it decays to 233U.One conceptual MSBR design is based on a two-fluid concept, with the fertile stream circulating in separate channels through the high-flux core region a s well as through the blanket surrounding the core. The protactinium concentration in the fertile stream can be kept low by processing the stream rapidly for removal of protactinium soon after it is formed, leaving little t i m e for neutron capture. Promising protactinium removal processes based on reductive extraction using liquid bismuth have been analyzed to evaluate their feasibility.

The flowsheet chosen for study is shown in Fig. 27.1. The modification shown in Fig. 27.2 could be used if development of a reliable reducer proves more difficult than expected. In both flowsheets, fertile salt from the reactor (stream 1)is contacted with liquid bismuth (stream 3) containing thorium (0.003 mole fraction). The bismuth contains a lithium concentration such that no thorium or lithium transfers between the metal and the fertile salt. The metal stream from the extractor containing protactinium flows to an oxidizer which converts the thorium, protactinium, and lithium to fluorides. The bismuth is not oxidized and is recycled to the extractor after the proper amounts of thorium and lithium reductant are added by electrolytically reducing all or part of a

ORNL-DWG

F2 (SALT)

60-707A

F3 (METAL)

I uF6

DECAY TANK

F4 (METAL)

e a

REDUCER

= I

F4-2 (METAL)

Fig. 27.1.

Reference Flowsheat for Removing Protactinium from a Two-Fluid MSBR.

260

261 ORNL-DWG 68-788A

F2 (SALT)

Fig.

27.2.

"Throwoway"

F7 (METAL)

NATURAL LiF OR NoF ADDITION

Flowsheet for Removing Protactinium from a Two-Fluid

MSBR Without Use of on

Electrolytic Ox id izer-Reducer.

recycle salt stream from the decay tank (stream 7). The salt mixture formed by oxidation of stream 4 would have a n undesirably high liquidus; the liquidus is lowered in the preferred flowsheet (Fig. 27.1) by recycling salt from the decay tank through the reducer and into the oxidizer. T o minimize the protactinium return to the extractor, salt recycled from the decay tank is contacted with the metal stream leaving the first, or main, extractor. This transfers most of the protactinium to the metal stream, which then flows to the oxidizer; there the protactinium is oxidized and returns to the decay tank without passing through the main extractor. In the modification shown in Fig. 27.2, fresh 'Li and thorium m e t a l are added to bismuth to form stream 3; no reducer is used. The me tal stream from the first extractor is contacted with a recycle from the decay tank, as in Fig. 27.1, before entering the oxidizer, but the recycle salt is sent to waste. Lithium fluoride is added to theoxidizer to reduce the liquidus of the resulting salt, and since this material will not return to the reactor blanket, natural lithium c a n b e used. Another alkali metal fluoride (e.g., sodium fluoride) could be used if it does not interfere with protactinium transfer in the second extractor.

T o evaluate the relative merits of these flowsheets, flow rates and compositions of all streams were calculated over a wide range of conditions. The lithium and thorium concentrations in the blanket were fixed at 0.72 and 0.28 mole fraction, respectively, and the protactinium p e r a t i o n rate was fixed at 10.6 g-moles/day. [This corresponds to a 2200 M w (thermal) reactor.] A l l calculations were based on a p w e s s i n g rate of 5.81 x lo6 g-moles/day of blanket salt, which is 4750 ft3/day (25 gpm, approximately two blanket volumes per day). Results are shown in Figs. 27.3 and 27.4. Figure 27.3 shows the required m e t a l flow rate in gram-moles per day (multiply by 3.9 x to get gallons per minute) as a function of the number of stages i n the extractor for three protactinium concentrations in the blanket (5 x 10 x lov6, and 20 x mole fraction). These blanket concentrations may be compared with the value of approximately 80 x mole fraction which resulted i n the blanket of a "no-Pa-removal" reactor system which h a s been considered.' The 'R. B. Briggs, Summary of the Objectives, the Design, and a Program of Development of MoItenSalt Breeder Reactors, ORNL-TM-1851(June 12, 1967).

262

ORNL-DWG 68-6098

5 IO N. NUMBER OF STAGES

Fig. Fig.

27.3.

Metal Rate Required as a Function of

27.4.

Ratio of Salt Rate from Oxidizer to Metal

Rate a s a Function of

LiF

Concentration i n Decay Tank.

Number of Stages in Extractor.

metal rate is particularly important for two reasons: it affects the extractor design and bismuth inventory and it sets the reducer capacity (or ’Li and thorium consumption in the “no-reducer” flowsheet). A low metal rate is desirable because the reducer anode surface may be expensive to maintain. Raising the protactinium concentration in the blanket raises the equilibrium (or maximum) metal loading and lowers the required bismuth rate. Increasing the number of stages i n the contactor also improves the performance of the extractor and lowers the required metal rate; however, Fig. 27.3 shows that the gain will b e relatively s m a l l for more than two or three stages. Required m e t a l flow rates will be about 0.3 x lo5 g-moles/day (0.1 Qm). In the reference flowsheet the salt recycle rate to the oxidizer is determined by the desired liquidus temperature. Figure 27.4 shows the decay tank composition a s a function of the ratio of thesalt recycle rate to the metal rate F,. With an infinite

recycle rate, thelithium composition only approaches the blanket composition, so that it is necessary to operate with a slightly lower lithium Composition in the decay salt than in the blanket. The no-reducer flowsheet h a s theoption of a lower reductant composition, which increases the lithium-to-thorium ratio in the metal. This allows a lower salt recycle but requires higher m e t a l circulation rates. The volume of the decay tank is set by the concentration of protactinium desired in the tank but is subject to limitations. The tank must hold sufficient protactinium that 10.6 g-moles decay in the tank per day (neglecting decay outside the tank). Under the conditions of interest, the tank volume will be limited by heat-removal considerations rather than by the maximum concentration which could be achieved. There will be 6.7 Mw of heat generated in the decay tank from protactinium decay, which suggests a minimum tank volume of a few hundred cubic feet.

263 Important considerations unique to the no-reducer flowsheet are 'Li and thorium consumption and protactinium losses. The 'Li and thorium consumptions reflect both material transferred to fertile salt i n extractor I(0.5 to 5% of materid fed) and the material remaining in the effluent metal stream from this extractor, which is ultimately sent to waste. The protactinium loss results from incomplete removal of protactinium from the salt stream flowing through extractor 11. The 'Li consumptions are 5.7 and 1.4 lb/day, respectively, for blanket protactinium concentrations of 5 x 10- and 20 x 10- mole f taction. The corresponding thorium consumptions would b e 150

and 38 lb/day. Consumption of these quantities of 'Li and thorium would be tolerated, but the costs are significant. With a 400-ft3 decay tank, theprotactinium loss would b e less than 0.02% of that produced in the reactor. The reference flowsheet shown in Fig. 27.1 is the preferred flowsheet, and further studies and development work for two-region processing should be aimed at this system. This flowsheet provides a minimum-size reducer and bismuth rate while placing little restriction on the decay tank volume; the decay tank volume is governed largely by heatremoval considerations.

28. Recovery of Uranium from M S R E Fuel Salt by Fluorination M. R. Bennett G. I. Cathers C. J. Shipman

The future schedule of MSRE operation includes removal of the uranium (about 30%enriched in 235U) from the carrier salt, LiF-BeF2-ZrF, (6530-5 mole %), by fluorination. The purified carrier salt will be reused for the subsequent operation of the MSRE with 233UF4.Fluorination of UF, dissolved in a molten fluoride salt, with the attendant evolution of volatile UF,, is quite corrosive to all known metallic containers. It is anticipated that significant amounts of chromium, iron, and nickel will b e put into the salt by corrosion, making necessary its cleanup by hydrogen treatment and other means. A series of small-scale fluorination tests is being made with simulated MSRE salt to study the effects of temperature, fluorine concentration, and fluorine flow rate on the rate of uranium volatilization and the rate of corrosion of Hastelloy N. In addition, some experimental results are being obtained on the volatilization of chromium and molybdenum fluorides (which are corrusion products) and on the behavior of volatile fission products (Ru, Nb, I, Te) in the fluorination of short-decay fuel. Knowledge of fission product behavior is essential to the development of a special fluorination procedure to be used by the Analytical Chemistry Division in making precise uranium analyses of fuel salt.

28.1 FLUOR1NATION-CORROSlON STUDY The fluorination-corrosion tests are being made with MSRE-type salt in a 1.87-in.-ID Hastelloy N reactor under conditions roughly approximating those being considered for u s e in the 49-in.-ID fluorination tank at the MSRE site. The uranium

264

is being volatilized out of the salt with an intermediate inert-gas sparging period to simulate changing of the NaF sorbent beds used for r e covery of the UF,. A gas flow rate of 146 ml/min (STP) is being used in the 1.87-in.-ID reactor to simulate the flow rate of 100 liters/min in the 49-in.-ID reactor. In the larger reactor, more uranium is involved, and the surface-to-volume ratio is lower. Hence the change in oxidation state of the salt during the course of UF, volatilization may be different in the two systems. Two tests have been made, one at 450OC and one at 500째C, each with 340 g of LiF-BeF2-ZrF,-UF, (63-32-5-0.8 mole %). Salt samples were taken during the runs at intervals (after He sparging) determined by the necessity of changing a 12-g NaF trap used for recovery of the UF,. In each test, 50% F,-50% He (146 ml/min) was used initially, followed by pure fluorine (146 ml/min) at the end of the run to ensure complete uranium recovery. In another test at 500째C, pure fluorine was used throughout. A trap containing 2 M KOH was employed to sorb unused fluorine and volatile chromium and molybdenum compounds that passed through the NaF trap. As shown in Table 28.1, the UF, volatilization rate and the degree of fluorine utilization were higher at 500OC than at 450OC. The amount of chromium leached from the Hastelfoy N reactor increased with temperature. The effect of temperature on the solubilization of iron and nickel from t h e reactor is not as clear (the dip tube, thermocouple well, and reactor bottom were made of Ni-200). Figure 28.1 shows the relation be-

265

T a b l e 28.1.

Uranium Volatilization and Corrosion Results in Fluorination T e s t s with

MSRE Salt i n Hastalloy N at Two Temperatures

Temperature

Fluorine Concentration

Total Time

(so)

(OC)

(hr)

Fluorine Utilization Efficiency

(%I

(mg)

(%I

50 50 50 50 100 100 100 100

0.5 2.0 5.0 6.0 8.0 12.0 16.0

13 42 93 118 103 143 128 117

4 35 60 73 82 152 93 96

5.4 14.2 44.7 68.0 92.2 100

50 50 50 100 100 100

1.0 1.75 2.25 2.75 6.75 10.75

75 153 194 224 150 140

49 95 122 126 139 180

18.5 38.5 55.1 83.4 100

100 100

0.75 1.25 1.58 2.58 6.58 10.58

74 112 199 136 125 102

17 48 48 71 71 82

28.7 59.4 90.7 98.9 100

450

500

100 100

3.5

Corrosion (mg in salt8)

Volatile Material

2.4 5.5 14.7

4.0 2.1 7.5 5.6 2.2 0.7

7.6 11.0 13.8 11.7 1.8

9 12

7.1 11.4 17.6 1.5 0.1

%orrected for initial values at zero t h e .

. W

tween chromium and iron leaching and uranium volatilization for the third run (with pure fluorine); corrosion appears to occur predominantly in the uranium volatilization period. Various estimates of the corrosion rate can be calculated from the chromium data using different t i m e periods (Table 28.2). There is good agreement between the rate calculated from the period in which most of the chromium was leached and that over the period where most of the uranium was volatilized. The estimates for the complete runs are somewhat arbitrary because fluorination was continued after the uranium had been removed to observe chromium behavior. Volatility of chromium and molybdenum was also observed in these runs. A chromium compound (probably CrF5) appeared to volatilize only

after most of the uranium had been removed from the salt (Table 28.1). This chromium compound was effectively trapped by NaF at 2SoC, forming an orange complex compound. Molybdenum (probably as MoF,) volatilized to a large extent throughout each run and, a s expected from the dissociation pressure of MoF6.2NaF, was not completely recovered in the 25OC NaF trap. Most of it passed through to the KOH scrub solution. The fluorine utilization efficiency figures (calculated from UF, output and fluorine input) suggest that the utilization was highest during the first 2 hr of fluorination; this is not strictly true, since the method of calculation does not take into account the fluorine consumed in changing the average oxidation number of the uranium. Also, the data do not imply that a definite consumption

266 Table 28.2.

Temperature

CC>

Estimates of Rates of Corrosion of Hastellay

Fluorine Concentrstion (%)

in Fluorination Tests

Corrosion Rate (milshr) in Period of: 80-95% U

Maximum Rate of Cr Change

Volatilization

0.19 0.58 0.50

0.12 0.45 0.49

50 and 100

450 500 500

50 and 100 100

Complete Run 0.05 0.12 0.07

28.2 FISSION PRODUCT BEHAVIOR ANALYTICAL ASSISTANCE PROGRAM

3'6

FLUORINATION TIME (hr)

Fig. 28.1.

Relationship Between Corrosion of Hostel-

and U F 6 Volatilization in Fluorination of MSREType Salt a t 50OoC. loy

of fluorine has t o occur before any UF, is evolved. In recent fluorination tests, some volatilization of UF, occurred when the uranium oxidation number in the salt was a s low a s 4.3. This was calculated on the basis of complete consumption of fluorine introduced i n the first 1.5 min of fluorine flow.

In the currently proposed analytical procedure for the determination of uranium in MSRE salt, which involves fluorination a s the first step, a precision of 3~0.1% is needed. Since it is anticipated that the use of remotely controlled hotcell coulometric instrumentation will introduce an inherent error of fl%,decontamination of the UF,NaF mmplex product from the fluorination step to a level appropriate for analysis i n a low-level radiation area will be essential. In addition, uranium losses in the waste salt and fluorination system (traps, etc) must be less than 0.1 wt %. Results of preliminary tests in which uranium has been fluorinated from MSRE-type salt containing volatile fission product activity have been encouraging. Four fluorination tests have been completed using specially prepared MSRE-type salt samples supplied by the Analytical Chemistry Division. The samples consisted of copper capsules each containing about 46 g of LiF-BeF,-ZrF,-UF, (65.4-29.4-5.0-0.2 mole %). Each salt sample was melted out under an inert atmosphere into a suitable nickel fluorinating vessel and then was spiked with about 0.1 m c of lo3Ru, "Nb, I3l1, or I3,Te. Solutions of these tracers were first evaporated to dlyness with the corresponding carrier (about 40 to 50 mg) onto 0.5-g quantities of LiF. Fluorination was carried out by heating the salt to 55OOC in an inert atmosphere and then passing a mixture of 25% F,-75% N, through the melt for 30 min followed by pure fluorine for an additional 30 min at 100 ml/min (STP). The product gas stream (containing F,, UF,, and volatile fission product activity) was passed through a 20-g NaF

267

during fluorination were, respectively, about 10, and greater than 99%,with corresponding material balances of about 95, 10, and 0.1%. Essentially all the ruthenium and niobium that was volatilized was trapped on the 4OOOC NaF trap; however, no appreciable sorption of tellurium was detected (Table 28.4). Attempts to a d i e v e an adequate decontamination factor (greater than lo3) for l 3 I I have been unsuccessful so far. In the first two tests, where the UF, sorption trap consisted of 10 g of NaF at 25OC, decontamination factors of 14 and 8 were obtained. In the third test, decreasing the amount of NaF to 5 g and increasing the sorption temperature to 100째C resulted in a higher, but still inadequate, decontamination factor of 35 (Table 28.5). In this test the measurement of the temperature of the sorption bed may have been highly in error; therefore, a duplicate test will be made.

trap at 400째C to remove the fission products and then into a 10-g N a F trap at 25OC, where the UF, product was sorbed. An additional 5-g NaF trap was inserted downstream for determination of any uranium breakthrough. Based on a makeup value of 1.185 wt % uranium in e a c h of the initial salt samples, the total uranium found in three tests corresponded to recoveries of 99.27, 100.22, and 97.65%. However, corresponding uranium losses i n the tests, based on analyses of the traps and waste salt, were, respectively, 0.034, 0.059, and 0.076 wt %, indicating actual recoveries greater than 99.9%(see Table 28.3). Decontamination factors obtained for lo3Ru, 95Nb, and '32Te during the fluorination step have been greater than lo5, and thus these nuclides apparently do not make a significant contribution to the overall activity level of the UF, product. The relative amounts volatilized from t h e salt

T a b l e 28.3.

Uranium Recoveries and Losses During Fluorination from MSRE-Type Salt and Sorption on N a F

Uranium Makeup Value

Run No.

Uranium L o s s e s (% of total found)

Uranium

(wt %)

Recoverya (wt %)

Trap No. l b

99.27 100.22 97.65

0.012 0.051 0.026

Trap No. ' 2

Waste Salt

Tota 1

0.013 0.002 0.022

0.034

1.185 1.185 1.185

1 2 3

0.009 0.006 0.022

0.059 0.076

BBased on analysis of all traps and scrubber solutions. b4000C NaF trap. '25OC NaF backup trap.

Table

28.4.

Behavior of '03Ru,

95Nb,

and 1 3 2 T e During Fluorination of Uranium from MSRE-Type Salt a t 5MoC

Amount

Total Activity (dis/min) Nuclide

4OO0C NaF Trap

NaF Backup Trap

lo3Ru

-2X 10'

<lo3

1.9

95Nb

- 2 x 10'

3 x lo'

<lo3

9 x 10'

<lo3

132Te

"'2~ 10'

<lo3

c103

4.3 x i o 5

<lo3

> 99

x 10'

UF, Product Trap

Nuclides Volatilized

MSRE Salt (Initial)

Waste Salt

(%I

<lo3

UF, Product Decontamination Factor

> lo5 > lo5 > lo5

268 J-_

Table

Total Activity (dis/min) RunNo.

1 2 3

28.5. Behavior of l3’I During Fluorinotion of Uranium from MSRE-Type Salt at 550°C 13 lI

Backup

KOH

Wa ste

(Initial)

4OO0C NaF Trap

Trap

Scrub Trap

Salt

-2 lo8 ‘“2X lo8 -1 x lo9

-lo3 1 . 8 lo4 ~ 5.6 x 10,

1.3 10, 5.3 X106 7.2 x lo7

2x107

2x104 2.7 x 10,

MSRESalt

<lo3

UF, Trap

Volatilized

1 . 4 lo7 ~ 2 . 5 ~ 1 0 ~ 2.9 x lo7

Decontamination Factor

(7.1

>99 >99 >99

14 8 35

= I

chi

29. M S R E Fuel Salt Processing R. B. Lindauer The modifications to the MSRE fuel processing facility t o permit processing with a 30-day decay t i m e have been completed. In November, it was decided to remove the structural-metal fluorides formed during fluorination from the salt before returning it to the reactor system. Design of a salt filter for removing the metals after reduction h a s been completed, and fabrication is in progress. The filter will have 9 f t 2 of Inconel porous metal as a filter medium. The filter element is designed for remote replacement in case of plugging or rupture. The filter h a s a capacity of about 40 l b of corrosion products, the amount formed during 50 hr of fluorination with a corrosion rate of 0.2 mil/hr. Another change in processing plans is the elimination of the H,-HF treatment of the flush and fuel

269

salts prior to fluorination. A heated salt sample carrier was constructed as a simpler and more economical method for verifying the oxide analyses of fuel salt samples. Handling of the sodium fluoride absorbers loaded with UF, will b e simplified considerably, since it has been learned that most of the noble-metal fission products leave the salt during reactor operation. Molybdenum-99 would have been the principal radiation source with short-decayed fuel. The problem of overheating of the high-temperature sodium fluoride trap from "Nb is also eliminated. A third problem which has been eliminated is the discharge of tellurium during processing. Training of part of the reactor operations group to operate the fuel processing plant is in progress.

30. Preparation of 7UF-

233

UF, Fuel Concentrate for the MSRE

J. W. Anderson S. Mann S. E. Bolt

E. L. Nicholson J. M. Chandler W. F. Schaffer, Jr.

operated drilling, weighing, and packaging box has been built and tested for this operation. The lifting cables and keys will be attached to the capsules before they are filled with salt, thus avoiding a remote welding operation. After drilling, weighing, and testing, the capsules will be placed in the special holders that are used as part of the samplerenricher s y s t e m a t the MSRE, packaged in groups of six, and shipped to the MSRE in a shielded carrier. The method used for bulk additions of 235U fuel salt from the 4 i n . storage containers to the reactor via a heated transfer line was discarded i n favor of a simpler method for the highly active 233U salt. The bulk additions will be made from 2’/,-in.-diam salt cans that are vented and have an opening in the bottom. They are essentially “giant” enrichment capsules that are lowered into the heated dump tank, so that the s a l t melts and drains into the tank. Four 7.0-kg uranium cans, one 3.0-kg uranium can, one 2.0-kg uranium can, one l.CLkg uranium can, and two 0.5-kg uranium cans will be delivered to the MSRE. The cans are assembled i n two clusters for filling with salt. One cluster consists of the four 7.0-kg cans, and the other is the remaining five s m a l l cans plus one extra 2.0-kg can which is arranged to hold recoverable salt heels from the s a l t processing system. The capsule filling furnace in cell G was replaced with a larger furnace that will accommodate both the capsules and 2t-in.diam s a l t can clusters. Other equipment changes included a new furnace liner, a new furnace stand, can cluster dismantling tools, and a larger work table to accommodate both the capsule drilling and packaging box and the dismantling of the can

The MSRE will be refueled with 233U in 1968. Approximately 39.5 kg of 91.4%enriched 233U a s 7LiF-233UF, (73-27 mole %) eutectic s a l t will be required. This material will be prepared in a shielded facility because of the high 232U content (222 ppm) of the 233U. The process and equipment have been described in the previous report. The following sections describe the changes in process and equipment since the original report and the status of the program.

30.1 EQUIPMENT CHANGES The MSRE personnel completed a detailed evaluation of their fuel requirements and loading operations for the 233U program. The predicted operational fuel load for the reactor is 35.9 kg of uranium (-91.4% 233U), and the contingency is 3.6 kg, giving a total of 39.5 kg of 233U to be delivered to the MSRE a s the eutectic salt. This is a decrease of about 3 kg from earlier planning. At the same time, the number of enriching capsules was reduced from 60 to 45; so the surplus capsules were removed from the existing array. The unshielded glove box used at the MSRE for drilling the uranium enriching capsule vent and drain holes and for weighing the capsules before loading into the sampler-enricher is inadequate for the highly active 233U. These operations will now be done in the TURF before the capsules are shipped to the MSRE. A remotely

’

‘MSR Program Semiann. Progr. Rept. Aug. 31. 1967, ORNL-4191, pp. 252-53.

270

271

clusters for shipment. The c a n s will be weighed in the cell before shipment by suspending them from a wire attached to a scale located on top of the cell. W e have determined that the density of the stored 233U03 is higher than the original estimate. This, plus the decrease in the total amount of uranium required, as noted above, will allow u s to make three batches of approximately 13.5 kg U each instead of the four batches planned earlier. A significant saving in operating t i m e will result from the elimination of one of the production batches. The 45 enriching capsules will be fiIled from the first batch of salt. The remainder of the first batch, the second batch, and part of the third batch will be used to fill the cluster of four 7.0-kg uranium s a l t cans. The small c a n s will be filled last, from the remainder of the third batch. The 233U03 oxide can opener and dumping box were designed using can drawings and sample cans from Savannah River. The cans in storage in the pilot plant were not checked closely because of the high radiation levels of the oxide. After the box was completed, it was found that the cans were 1.75 in. shorter than shown by the drawings and can samples. Since modification of the charging box to accommodate the shorter cans was impractical, we will cement a 1.5-111. . extension on each can. The fixtures for tapering the can end and removing the weld bead, for pressing the extension on the cans, and for curing the cemented joint have been built and tested. The fabrication costs and operating problems caused by this development are minor.

30.2 EQUIPMENT AND PROCESS STATUS The TURF building was turned over to ORNL on September 15, 1967, and the C P F F contractor started preparing cell G for installation of the process equipment and completed this by midOctober. The C P F F work consisted in installing the instrument panel, the gas supply and electrical systems to tenninals inside the cell, the cell platform, the shielding plugs for unused pipe s l o t s and windows, the PARmmanipulator, and the can charging chute. The process equipment was completed in the ORNL shops by mid-October, and O W L forces finished the installation of the process equipment in cell G in January. Part of

the processing equipment installed on and in cell G is shown in Figs. 30.1 to 30.3. The equipment for adding the extensions to the oxide cans, for dismantling the capsule and can assemblies, and for drilling and weighing the capsules will be located in the northwest and northeast comers of the cell. About 130 formal drawings plus 150 field sketches have been issued to date for the job. W e estimate that the design work is more than 98% complete as of February 29, 1968. Pressure vessel, criticality, building, and radiochemical plant safety reviews were completed, and the process and buildings were approved for MSRE 233U fuel production. Equipment checkout and minor alterations and changes were completed, and the cold test run with 238U03(11.6 kg 238U) was started January 15, 1968. Some problems were encountered with the can opener and dumping box. After a few cans were opened, a groove rolling device accidentally jammed the can chuck, stripping the motor drive gears. The gears were replaced, and the remaining cans were opened and dumped, though with difficulty due to binding of the cans in the box. The box was removed from the cell, decontaminated, and sent to the shop for repairs and for alterations to improve operability. The chemical part of the run started on January 24, 1968, with the WOOC sintering treatment followed by hydrogen reduction of the UO, to UO,. The offgas filter plugged during the reduction step but was opened by blowback and h a s not been a problem since the reduction step. Treatment with H2-HF has been under way since January 27. It was planned to follow the progress of the conversion of UO, to UF, by titrating the off-gas to determine HF utilization and to periodically determine the freezing point of the charge as it approached the eutectic point. The off-gas titration was hampered by HF-water condensation i n the plastic lines to the caustic scrubber bottle and by problems with remote manipulator operation of the conventional burets i n the cell. This rig h a s since been replaced by a reliable titration system, but in the early stages of the conversion it w a s impossible to get reliable results with the existing titration system. Attempts to run freezing point curves on the s a l t i n the reaction vessel were unsuccessful because of the high thermal capacity of the reaction vessel and furnace as compared with the s a l t charge. Samples of the

272

Fig.

30.1.

Control Panel and Gas Supply for

233U F u e l Salt Preparation

melt taken after about 290 hr of HF-H, treatment showed that only 5 to 10% conversion of UO, to UF, had occurred. The H,-HF sparger tube was removed from the reaction vessel on February 14 and was found to be corroded off at the liquid-gas interface. It was replaced, and HF utilization rose from zero to 70%, indicating good conversion of UO, to UF,. After about 200 hr of additional operation, about 71% of the UO, had been reacted. The H F utilization, as determined by titration with the new apparatus, has decreased in the expected manner since then. Progress of the reaction is being followed mainly by sampling the salt at frequent intervals. We had problems with the remote salt sampling box because of excessive friction between the

- Cell G,

Building

7930.

s a l t sampler probe and the sampler box. The box was returned to the shop for repairs and for alterations to improve operability. At the end of the month, the H,-HF treatment was still under way, and 90 to 95% of the UO, had been converted to UF,. The progress of the reaction h a s been slow owing to poorer contacting of gas and salt i n the 8-in.diam pot as compared with the rl-in.-diam pots used during flowsheet development, and interruptions of the process due to equipment probl e m s and for salt sampling operations. We now estimate that the chemical part of e a c h 233U run will take about 28 days instead of the 17 estimated earlier. Tentative plans are to complete the cold run in March and prepare the three batches of 233U salt in April, May, and June.

Lid

i 273

i1 w i i I

Fig. 30.2. Building 7930.

Main Reaction Vessel Furnace and Auxiliary Equipment for 233U Fuel Salt Preparation

- Cell G,

1 274

. Fig. 30.3.

Auxiliary Furnaces and Equipment for

233U Fuel Salt

Preparation

Cell G, Building 7930.

31.

Decay Heat Generation Rate in a Single-Region Molten-Salt Reactor J.

W. L. Carter

Decay heat from fission products and 233Pa has been calculated for a 2000 Mw (electrical) singleregion MSR operating on LiF-BeF2-'ThF,-2 33UF, fuel. Heat-generation rates are shown graphically in Figs. 31.1 and 31.2 a s a function of time after reactor shutdown. The data extend from reactor equilibrium to a decay t i m e of about 11years. It was assumed that the MSR had been operating sufficiently long for fission products and 233Pa to be present in equilibrium concentrations. For 233Pa the characteristic equilibrium was for a 3 d a y processing cycle. For fission products, equilibrium was established for three different situations: (1) all fission products removed on a 38-day cycle through the processing plant; (2) noble gases removed on a 50-sec cycle by sparging with an inert gas in the circulating loop; and (3) in addition to noble-gas sparging, noble metals removed on a 50hr cycle by reaction with piping, heat exchanger, and reactor surfaces. In cases 2 and 3, remaining fission products were removed by the 3 8 d a y cycle through the processing plant. The vertical separation of the three appropriately labeled curves of Fig. 31.1 is a measure of the difference in decay heat from these three assumed

275

S. Watson

conditions. The largest difference for t i m e s soon after shutdown occurs at about 1 hr, when the noble-gas- and noble-metal-free fuel produces about 33%less heat than fuel containing gross fission products. At equilibrium for the three above conditions, fission products produce, respectively, 144.7, 128.5, and 127.9 kw per cubic foot of fuel. This reactor has a breeding ratio of 1.076, and there is 205 kg of 233Pa in the fuel and processing plant. The 233Pa concentration in the fuel is 7.25 g/ft3 (14.5 kg total), which a t equilibrium generates 0.367 kw of heat per cubic foot; there is 195.5 kg in the processing plant, producing 9.68 Mw of heat. An interesting result of this study is the importance of a few fission products in the total heat generation. At equilibrium, Rb, Cs, and Sb, respectively, account for 22.8, 21, and 17.3% of fission product decay heat. At 1hr decay the amounts are I (18.6%), Kr (13%), La (lO.S%), and Y (9.6%); at 10 hr decay, the values are I (23.8%), La (16.5%), and Y (11.4%). For much longer decay t i m e s (e.g., 125 days), about 80%of the energy is from Nb, Zr, Pr, and Y.

276

8 6 -

4 -

REACTORPOWER FUEL VOLUM& EPUlLlBRlUM Po FUEL PROCESSING CYCLE *33Pa PROCESSINGCYCLE

Fig.

31.1.

.. .

4444 MW IThwrnal) 2000 ft9

= 0.256 *

~llilw 36 days

3dOys

F i s s i o n Product and Protactinium Decay Heat in One-Region

MSR Fuel.

277 0RNL- DWG 68 -2 407A

2FUEL

FUEL VOLUME FUEL PROCESSING CYCLE Po PROCESSING CYCLE

4444 MW (thrrmol) 2000 ft' 38 DAYS 3 DAYS

~ E A Z T O RPOWER

IO8 8 6 -

2IO'

I I I I11111

Fig. 31.2.

6810

I I7 I I IIIII

IOd

30 d

I1 I I IIIIII

3~10

6mo

4 6810' 2 4 6810' 2 4 TIME AFTER DISCHARGE FROM REACTOR (hours)

5 Yr

6810'

I 1 IIIIIII

Total Heat-Generation Rate in One-Region MSR from Gross Fission Products and 233Pa.

68106

278 OAK RIDGE NATIONAL LABORATORY

MOLTEN-SALT REACTOR PROGRAM FEBRUARY 29, IW

I EvIMrYYHT

c E. h i . . 1. LCm.

r. H. c i d *

n ruur DLVEI

c 1.Cmllint

1. K. &In' L 1. Urn* 1.P. E . k r l y A. 0. GdnLII

AGB

J. 1. kw*m R L. l k d H. A Nh.' L c O.L... R C &bm* 1. R. &n* L R T.llrkm* w. 1H. L. W1. L. v. Wilm. 1.c. Grrc H. Y. Pdy

A YPm*

J. R. Wdr H. E. U d h

YLC

HAITELMY Y STUDIES YLC YLC YLC YLC YLC YK YLC YLC

UIl

P. G Llh L. v. W i l v n H. C luna D. D. hr. Dql. h n

G H LI.rllm' H. A. Y d d n

UTERIALS

PHYYU

R R R

YLC WC YLC YLC YLC

SYSTU

YLC YLC YLC YLC

R R RC R R UTE R R

R UT V. G Lm. E. 1. L..nnc. E. H. Ln' L. hlq. 1.1. Y.m'

B. W

1. A

?.*.lt*

N. a PI-^^ R. 5. Pullla. 1. H. hi*,1.'

L. D. +.&* G. D. M ! d E. Y h u s

n. RTinh. 1. 1. Wwhovv B. c. w1111ams

CT CT CT CT CT CT

R 1.Willims

YK YLC YLC YLC YLC YLC WC YLC

1. R. En$ C. H. G&d'* 1. C. Wr/Jr. .

n CT n CT CT CT CT CT CT CT CT CT

CT CT CT

AW m

C. D. M n . 1. L. R**rd R. 1.TU&" S E. Ellis C.Chhnm C. E. K l r l r d

Rc Rc

D. Y h l h L. G 0rbl.r 1. H. IMh* R A. %dab* E. L. N i d o l a 1. Y. h d l 5 . h AYRoI W. F. kh&"

L CnC.

YLC YK YLC

AC W L V T I U L QIWISTRY C CiWlSTRY D I ~ l S l m _ CT QIWlCU T E W L O O Y I D DlRECmR5 DIVISION GE GENERAL ENGINEERING0 I U INSTRWENTATION W D C YLC Y E T U L C E R * Y I C S D I

J, 0. Bl&* 1.L. Gh. L Y Frrh B. A -H 1.1. H~CCY S Li" 1. C. kiln L. E. Ydlru 1. P. Nihd. C. E. khilllq F. 1. hlh 1. S W s t m E. L. Younshtd 1. h e V. L. FadJ. F. L d 11.0.Pqn. c. 1. Thwm

R E . h** L. Y l a h C F. W - 4 C E. L. f. A. D n s H. A Fd&. L. 0. G d w i & * D. K Ha',

B. F. H i i h H. 1.Kdan J. 0. R h * G. D. R*L*I. K. A. R & p * 1. G & s i H. H. Itnd W. K. R. Fhnll 1.kndws. h. C. E. Ro*n. 1.P. T d h n

M u WC

YLC YLC YLC YLC YLC YK

YLC

IC IK RC IC RC RC RC RC

RC RC Rc IC Rc C Rc IC

S 1.w.

IK Rc RC Rc IC

Rc Rc

YM YLC YLC YLC YLC YLC E. L. C.rr H. C. Smwa 1. Y Bd1.1. H n

* PART TWE ON M5RP

J. C. Ihim*

wucbP*an

S.y.*r R. F. &I.

c. Y eoyd J. Y Ikl. 1.H. MI.*' D. L. h l n s * 1.1.h*k* E. nirri* F. L. Ihitin) 1. P. run'

Ac Ac Ac Ac AS AC Ac AC UT Ac

SPECIAL * U I I T A K E

C C C P GE P GE YK

ILC

279

ORNL-4254

UC-80

- Reactor Technology

INTERNAL DlSTRlSUTlON 1. 2. 3. 4. 5. 6. 7. 8. 9. 10. 11. 12. 13. 14. 15. 16. 17. 18. 19. 20. 21. 22. 23. 24. 25.

R. K. Adams

G. M. Adamson R. G. Affel L. G. Alexander R. F. Apple C. F. Baes J. M. Baker S. J. Ball C. E. Bamberger C. J. Barton H. F. Bauman S. E. Beall R. L. Beatty M. J. Bell M. Bender C. E. Bettis E. S. Bettis D. S. Billington R. E. Blanco F. F. Blankenship J. 0. Blomeke R. Blumberg A. L. Boch E. G. Bohlmann C. J. Barkowski 26. G. E. Boyd 27. J. Braunstein 28. M. A. Bredig 29. E. J. Breeding 30-44. R. B. Briggs 45. H. R. Bronstein 46. W. E. Browning 47. F. R. Bruce 48. G. D. Brunton 49. G. H. Burger 50. D. A. Canonico 51. S. Cantor 52. D. W. Cardwell 53. W. L. Carter 54. G. 1. Cathers 55. 0. B. Cavin 56. A. Cepolino

57. J. M. Chandler 58. W. R. Cobb 59. C. W. Collins 60. E. L. Compere 61. J. A. Conlin 62. W. H. Cook 63. L, T. G r b i n 64. W. B. Cottrell 65. B. Cox 66. G. A. Cristy 67. S. J. Cromer (K-25) 68. J. L. Crowley 69. F. L. Culler 70. D. R. Cuneo 71. J. M. Dale 72. D. G. Davis 73. W. W. Davis 74. R. J. DeBakker 75. J. H. DeVan 76. S. J. Ditto 77. R. G. Donnelly 78. 1. T. Dudley 79. N. E. Dunwoody 80. A. S. Dworkin 81. D. A. Dyslin 82. W. P. Eatherly 83. M. C. Edlund (K-25) 84. J. R. Engel 85. E. P. Epler 86. W. K. Ergen 87. D. E. Ferguson 88. L. M. Ferris 89. A. P. Fraas 90. H. A. Friedman 91. J. H. Frye, Jr. 92. C. H. Gabbard 93. W. R. Gall 94. R. B. Gallaher 95. R. E. Gelbach 96. J. H. Gibbons 97. R. G. Gilliland 98. L. 0. Gilpatrick

280 h

99. H. E. Goeller 100. W. R. Grimes 101. A. G. Grindell 102. R. W. Gunkel 103. R. H. Guymon 104. J. P. Hammond 105. R. P. Hammond 106. B. A. Hannaford 107. P. H. Harley 108. D. G. Harman 109. W. 0. Harms 110. C. S. Harrill 111. P. N. Haubenreich 112. F. A. Heddleson 113. R. E. Helms 114. P. G. Herndon 115. D. N. Hess 116. R. F. Hibbs (Y-12) 117. J. R. Hightower 118. M. R. H i l l 119. E. C. Hise 120. H. W. Hoffman 121. D. K. Holmes 122. V. D. Holt 123. P. P. Holz 124. R. W. Horton 125. A. S. Householder 126. T. L. Hudson 127. H. lnouye 128. W. H. Jordan 129-143. P. R. Kasten 144. R. J. Ked1 145. M. T. Kelley 146. M. J. Kelly 147. C. R. Kennedy 148. T. W. Kerlin 149. H. T. Kerr 150. R. F. Kimball 151. S. S. Kirslis 152. D. J. Knowles 153. J. W. Koger 154. A. 1. Krakoviak 155. J. S. Kress 156. J. W. Krewson 157. C. E. Lamb 158. J. A. Lane 159. C. E. Larson 160. E. J. Lawrence 161. J. J. Lawrence 162. M. S. Lin 163. T. A. Lincoln

164. R. B. Lindauer 165. A. P. Litman 166. J. L. Liverman 167. R. S. Livingston 168. G. H. Llewellyn 169. E. L. Long 170. M. 1. Lundin 171. R. N. Lyon 172. R. L. Macklin 173. H. G. MacPherson 174. R. E. MacPherson 175. F. C. Maienschein 176. J. C. Mailen 177. D. L. Manning 178. C. D. Martin 179. W. R. Martin 180. C. E. Mathews 181. T. H. Mauney 182. H. McClain 183. R. W. McClung 184. H. E. McCoy 185. H. F. McDuffie 186. C. K. McGlothlan 187. C. J. McHargue 188. L. E. McNeese 189. J. R. McWherter 190. H. J. Metz 191. A. S. Meyer 192. E. C. Miller 193. C. A. Mills 194. R. L. Minue 195. W. R. Mixon 196. R. L. Moore 197. K. Z. Morgan 198. D. M. Moulten 199. J. C. Moyers 200. T. R. Mueller 201. H. A. Nelms 202. J. P. Nichols 203. E. L. Nicholson 204. E. D. Nogueira 205. L. C. Oakes 206. W. R. Osborn 207-208. R. B. Parker 209. L. F. Parsly 210. P. Patriarca 211. H. R. Payne 212. A. M. Perry 213. T. W. Pickel 214. H. B. Piper 215. B. E. Prince

w t

bsl

281

81,

G. L. Ragan J. L. Redford M. Richardson G. D. Robbins R. C. Robertson W. C. Robinson H. C. Roller K. A. Romberger M. W. Rosenthal R. G. Ross H. C. Savage A. W. Sawlainen W. F. Schaffer C. E. Schilling Dunlap Scott J. L. Scott H. E. Seagren C. E. Sessions J. H. Shaffer E. D. Shipley W. H. Sides M. J. Sk'inner G. M. Slaughter A. N. Smith F. J. Smith G. P. Smith 0. L. Smith P. G. Smith A. H. Snell W. F. Spencer 1. Spiewak R. C. Steffy C. E. Stevenson W. C. Stoddart 424. H. H. Stone 425. R. A. Strehlbw

216. 217. 218. 219. 220. 221. 222. 223. 224-398. 399. 400. 401. 402. 403. 404. 405. 406. 407. 408. 409. 410. 411. 412. 413. 414. 415. 416. 417. 418. 419. 420. 421. 422. 423.

426. 427. 428. 429. 430. 431. 432. 433. 434. 435. 436. 437. 438. 439. 440. 441. 442. 443. 444. 445. 446. 447. 448. 449. 450. 451. 452. 453. 454. 455. 456-457.

D. A. Sundberg J. R. Tallackson E. H. Taylor W. Terry R. E. Thoma P. F. Thomason L. M. Toth D. B. Trauger R. W. Tucker W. C. Ulrich D. C. Watkin G. M. Watson J. S. Watson H. L. Watts C. F. Weaver B. H. Webster A. M. Weinberg J. R. Weir W. J. Werner K. W. West M. E. Whatley J. C. White G. C. Williams L. V. Wilson

G. J.

Young

H. C. Young J. P. Young E. L. Youngblood F. C. Zapp Biology Library ORNL - Y-12 Technical Library Document Reference Section 458-460. Central Research Library 461-615. Laboratory Records Department 616. Laboratory Records, ORNL R.C.

EXTERNAL DlSTRlBUTlON

617. 618. 619. 620. 621. 622. 623. 624. 625. 626. 627.

W. 0. Allen, Atomics International, P.O. Box 309, Canoga Park, California 91304 J. G. Asquith, Atomics International, P.O. Box 309, Canoga Park, California 91304 J. C. Bowman, Union Carbide Technical Center, 12900 Snow Road, Parma, Ohio 44130 G. D. Brady, Materials Systems Division, UCC, Kokomo, Indiana 46901 J. H. Brannan, Carbon Products Division, 270 Park Avenue, New York, New York 10017 W. S. Butler, Dow Chemical Company, Freeport, Texas 77541 Paul Cohen, Westinghouse Electric Corp., P.O. Box 158, Madison, Pennsylvania 15663 D. F. Cope, Atomic Energy Commission, RDT Site Office (ORNL) J W. Crawford, Atomic Energy Commission, Washington 20545 M. W. Croft, Babcock and Wilcox Company, P.O. Box 1260, Lynchburg, Virginia 24505 C. B. Deering, Atomic Energy Commission, RDT Site Office (ORNL)

282

628. D. A. Douglas, Materials Systems Division, UCC, Kokomo, Indiana 46901 629. H. L. Falkenberry, Tennessee Valley Authority, 303 Power Building, Chattanooga, Tenn. 37401 630. C. W. Fay, Wisconsin Michigan Power Company, 231 W. Michigan Street, Milwaukee, Wisconsin 53201 631. Gregory Flynn, General Motors, 12 Mile and Mound Roads, Warren, Michigan 48089 632. A. Giambusso, Atomic Energy Commission, Washington 20545 633. Gerald Golden, Argonne National Laboratory, 9700 S. Cass Avenue, Argonne, Illinois 60439 634. W. W. Grigorieff, Assistant to the Executive Director, Oak Ridge Associated Universities 635. J. T. Kehpe, Burns and Roe, Inc., 700 Kinderkamach, Oradell, New Jersey 07649 636. L. W. Lang, Douglas United Nuclear, 703 Bldg., Richland, Washington 99352 637. W. J. Larkin, Atomic Energy -. Commission, OR0 638. J. A. Lieberman, Atomic Energy Commission, Washington 20545 639. R. A. Lorenzini, Foster Wheeler, 110 S. Orange, Livingston, N. J. 07039 640. W. D. Manly, Material Systems Division, UCC, 270 Park Avenue, New York, New York 10017 641. J. P. Mays, Great Lakes Carbon Co., 299 Park Avenue, New York, New York 10017 642. W. B. McDonald, Battelle-Pacific Northwest Laboratory, Hanford, Washington 99352 643-644. T. W. McIntosh, Atomic Energy Commission, Washington 20542 645. W. J. Mordarski, Nuclear Development, Combustion Engineering, Windsor, Connecticut 646. Sidney Parry, Great Lakes Carbon, P.O. Box 667, Niagara Falls, New York 14302 647. Worth Percival, General Motors, 12 Mile and Mound Roads, Warren, Michigan 48089 648. G. J. Petretic, Atomic Energy Commission, Washington 20545 649. M. A. Rosen, Atomic Energy Commission, Washington 20545 650. H. M. Roth, Atomic Energy Commission, OR0 651. R. W. Schmitt, General Electric Co., Schenectady, New York 12301 652. M. Shaw, Atomic Energy Commission, Washington 20545 653. Rem0 Silvestrini, United Nuclear Corporation, Grasslands Road, Elmsford, New York 10523 654. E. E. Sinclair, Atomic Energy Commission, Washington 20545 655. W. L. Smalley, Atomic Energy Commission, OR0 656. T. M. Snyder, General Electric Co., 175 Curtner Ave., San Jose, California 95103 657. L. D. Stoughton, UCC, P.O. Box 500, Lawrenceburg, Tennessee 38464 658. Philip T. Stroup, Alcoa, P.O. Box 772, New Kensington, Pennsylvania 659. J. A. Swartout, UCC, 270 Park Avenue, New York, New York 10017 660. R. F. Sweek, Atomic Energy Commission, Washington 20545 661. Richard Tait, Poco Graphite, P.O. Box 1524, Garland, Texas 75040 662. D. R. Thomas, Commonwealth Associates, Inc., 209 E. Washington Ave., Jackson, Michigan 49201 663. M. TSOU,General Motors, 12 Mile and Mound Roads, Warren, Michigan 48089 664. J. W. UIImann, UCC, P.O. Box 278, Tarrytown, New York 10591 665. C. H. Waugaman, Tennessee Valley Authority, 303 Power Building, Chattanooga, Tenn. 37401 666. D. B. Weaver, Tennessee Valley Authority, New Sprankle Building, Knoxville, Tennessee 667. G. 0. Wessenauer, Tennessee Valley Authority, Chattanooga, Tennessee 37401 668. M. J. Whitman, Atomic Energy Commission, Washington 20545 669. H. A. Wilber, Power Reactor Development Company, 1911 First Street, Detroit, Michigan 670. James H. Wright, Westinghouse Electric, P.O. Box 355, Pittsburgh, Pennsylvania 15230 67 1-672. Laboratory and University Division (ORO) 673-93 1. Given distribution as shown in TID-4500 under Reactor Technology category (25 copies-CFSTI)

I I *