Forensic Science Forensic ELSEVIER Science 77 (1996) International 211 ~22’) Capillary electrophoresis: principles and applications in illicit drug analysis’ F. Tagliaro”“.‘, Received 12 June 1995: revision S. Turrina”, received F.P. Smith” 8 September 1995: accepted 4 October 1995 Abstract Capillary electrophoresis, which appeared in the early 198Os, is now rapidly expanding into many scientific disciplines. including analytical chemistry. biotechnology and biomedical and pharmaceutical sciences. In capillary electrophoresis. electrokinetic separations are carried out in tiny capillaries at high voltages (lo--30 kV). thus obtaining high efficiencies (,li > 105) and excellent mass sensitivities (down to 10 Ix- 10 “I moles). The main features of capillary electrophoresis are: versatility of application (from inorganic ions to large DNA fragments), use of different separation modes with different selectivity. extremely low demands on sample volume, negligible running costs, possibility of interfacing with different detection systems, ruggedness and simplicity of instrumentation. Capillary electrophoresis applications in forensic sciences have appeared only recently. but are now rapidly growing, particularly in forensic toxicology. The present paper briefly describes the basic principles of capillary electrophoresis, from both the instrumental and analytical points of view. Furthermore. the main applications in the analysis of illicit,controlled drugs in both illicit preparations and biological samples are presented and discussed (43 references). It is concluded that the particular separation mechanism and the high complementarity of this technique to chro matography makes capillary electrophoresis a new pow-erful tool of investigation in the hands of forensic toxicologists. * Corresponding author. ’ A first version was presented National Drug Testing Laboratories as plenary from lecture at the UNDCP the countries participating Laboratory Project’, Thessaloniki. 3 ‘At present. visiting professor in Alabama at Birmingham, Birmingham. 7 October, 1994. the Graduate Program AL. USA. 037Y-0738’96~Sl5.00 Science SSDl 0379.0738(95)01863-E 12 1996 Elsekier Ireland Ltd. in All consultative in the Forensic rights reswed Meeting ‘Regional Science. The of Heads 01 Balkan Route University of 212 F. Tagliaro et al. / Forensic Science International 77 (1996) 21 I-229 Keywords: Capillary electrophoresis;Forensic toxicology; Illicit drugs; Analysis; Seizures; Biological samples 1. Introduction Capillary Electrophoresis (CE) (or high-performance capillary electrophoresis (HPCE)), probably, is the most rapidly expanding analytical technique appeared in the recent years. In fact, as it is depicted in Fig. 1, the papers appeared in the leading international journals have rapidly increased, as well as books, volumes [l - lo] and symposia dedicated to this emerging technique. In CE, electrokinetic separations are carried out in tiny capillaries at high voltages (lo-30 kV), thus obtaining high efficiencies (N > 105) and excellent mass sensitivities (down to 10 - 18- 10 - *’ moles). The main features of CE could be summarized as follows: - versatility of application (from inorganic ions to large DNA fragments); - modes of separation with different selectivity can be applied with the same hardware; - extremely low demands upon sample volume (few nl are injected); - minimum need of solvents (running buffer) and low cost of other consumables (capillaries); - possibility of interfacing with different detection systems; - ruggedness and simplicity of the basic instrumentation. 100 Cl 1982 1983 1984 1985 1986 1987 1988 1989 1990 1991 1992 1993 years Fig. 1. Papers on CE quoted yearly in Chemical Abstracts in the years 1883-1993. Fig. 2. Scheme of a capillary electropherograph 2. Instrumentution The basic equipment (Fig. 2), consists of an injection system, a capillary (in most instances made of fused silica, generally, 20&100 pm internal diameter (i.d.), 20-100 cm long), a high voltage source (operating within lo-30 kV), electrodes, electrode jars and detector(s). 2. I. Samplr injection Because of the miniaturization of the separation compartment (the capillary), only minute, but well controlled volumes of sample should be injected, if high efficiency in separations is to be attained. Modern, automated instruments do assure in real conditions injection reproducibility better than 2”/0. Manually operated CE systemsmay be less reproducible. but still acceptably precise with the use of appropriate internal standards. In general, two injection techniques can be used: hydrodynamic injection and electrokinetic injection. The two differ essentially becausethe former is non-specific (what is injected represents the composition of the sample exactly), while the latter is selective, depending on ion mobility, charge and concentration of the ions in the sample. The simplest way to introduce nanoliter amounts of sample solution into the capillary is by gravity, raising the injection end of the capillary for a fixed time. Also, hydrodynamic injection can be accomplished by application of pressure for a fixed time onto the sample vial in which the injection end of the capillary is dipped, or by application of vacuum at the opposite end. In order to allow injection by means of standard high performance liquid chromatography (HPLC) syringes, a split-flow injector for CE has also been developed, in which the liquid injected is split between the capillary and a split-vent. 214 F. Tagliaro et al. / Forensic Science International 77 (1996) 21 l-229 The percent volume of the sample introduced in the capillary (usually - l/1000) depends on the difference in hydrodynamic resistance between the ‘capillary port’ and the ‘vent port’, which depends on the relative differences in length and diameter of the connected tubings (i.e. the capillary and the waste). In electrokinetic injection, the sample is introduced into the capillary by applying a voltage (in general, lower than that used in the separation), while the injection end is dipped into the sample. Thus, the molecules contained in the sample are injected by electromigration as well as transportation with the electroosmotic flow (see later). Consequently, both ion mobility and electroosmotic flow influence the amount of the analytes injected. In conclusion, although theoretically superior in terms of selectivity, electrokinetic injection is dependent on factors often difficult to keep under control. 2.2. Capillary Typical dimensions of capillaries for CE are id. 20-100 pm and lengths ranging from 20-100 cm. The capillary is the compartment where separation and, in general, detection occur. Consequently, the capillary should be chemically and physically resistant, inexpensive, UV transparent, precisely produced with narrow internal diameters, not prone to adsorb solutes and, in order to allow dissipation of the Joule heating, should have a high thermal conductivity. Fused silica capillaries, resembling those currently used in capillary gas chromatography, can be commercially found with i.d. values ranging from less than 10 pm to more than 200 pm and meet almost all the above mentioned requirements, except that, in some instances, they tend to adsorb the analytes. CE capillaries can be uncoated, internally coated or filled with gels. In order to give them good flexibility, they are externally protected with polyimide. Where the capillary needs to be optically transparent, in order to allow photometric detection, the external coating is burned off, leaving a highly UV transparent, but fragile, wall. The conditions of the internal surface of the capillary (in the case of fused silica, the ionization of the acidic silanol groups) are extremely important to achieve the best performances from CE. The simplest way to obtain a ‘reproducibly clean’ inner capillary surface is washing the capillary with a strong base solution (e.g. 1.0-0.1 M NaOH), in order to dissolve a thin film of silica from the wall. Washing must be followed by a thorough rinse with the run buffer: the choice of pH, ionic strength and composition of buffer is, of course, crucial. Alternatively, the capillary wall can be coated with ‘modifiers’, which can be physically adsorbed, chemically bound or simply added to the buffer. The most common coatings are polyacrylamide, PEG, cellulose, PVA, amino acids, proteins, amines, surfactants, etc. 2.3. Power supply High voltage power supplies, in commercial CE instrumentation, generally deliver up to 30 kV and currents up to 200-300 PA. Because of the direction of the F. Tagliaro er al. I Forensic Science Internafional 77 (1996) 31 l-2-79 215 electroosmotic flow, the common polarity is with the anode at the injection end of the capillary and the cathode close to the detector; however, under particular experimental conditions the polarity is reversed. CE separations are usually carried out at constant potential, but constant current can be used too and, sometimes, this is preferable, because the system tends to compensate for temperature changes. Constant power operation is typical of isoelectric focusing. 2.4. Detection Detection in CE has to face two challenging problems, namely small amounts of analytes injected and tiny peak volumes (in the order of nanolitres). Up to the present, I-IV-visible detectors are by far the most widely adopted. In order to maintain the high efficiency of separation and to avoid loss of sensitivity due to peak dilution, detection is carried out directly ‘in-column’, through a window obtained by burning (or scraping) off the polyimide external coating of the capillary. However, the short i.d. of the capillary (i.e. the detection cell path-length) limits the concentration sensitivity to lo-“- 10e6 M and the linear range. A promising way to increase the path length (up to few mm) is to use a Z-shaped capillary, in which detection is by axial illumination. This, reportedly, improves the sensitivity by a factor of ten or more, with minimal decrease of resolution (see Ref. [l] pp. 61-69). Alternatively, the path length can be increased by, a ‘bubble cell’ design of the capillary [ll]. Improved selectivity in absorbance detection is achieved by photodiode array or multiwavelength detectors, analogous to HPLC, which, with only a moderate loss of sensitivity, allow on-line recording of UV(-visible) spectra of the separated zones. This is important for peak purity evaluation and for investigation of peak identity, on the basis of the spectral analysis of the peaks. Lamp-based fluorescence in-column detection encounters difficulties in focusing enough excitation energy into the capillary and in efficiently collecting the emitted radiation. This limits the theoretical sensitivity that fluorescence detection could provide; on the other hand, in terms of selectivity, the advantage of fluorescence over UV absorbance is maintained. Laser-induced fluorescence, allowing high energy excitation, is more sensitive (reportedly 10 - ’ up to 10 - I2 M) but instrumentally more complex and limited in the excitation wavelengths available from the current types of laser sources (325 nm for He-Cd and 488 nm for Ar ion lasers). Since a minority of analytes are naturally fluorescent, pre-column derivatization is often applied, primarily according to the procedures well established for HPLC. Post-column derivatization, although feasible, is basically more complex, due to the addition and mixing of reagents after the separation, which tend to cause band spreading incompatible with the tiny volumes of the separated zones. Electrochemical detection, in particular conductivity and amperometric detection, have been applied in CE with encouraging results (see Ref. [l] pp. 108-121). However, a problem still not fully resolved is the need to separate the high potential 216 F. Tagliaro et al. / Forensic Science International 77 (1996) 21 I-229 compartment of the capillary from the detection compartment. This insulation has been obtained by means of a porous conductive joint inserted before the detection end of the capillary, but no commercial devices are at present available. Conductimetric detection is also possible, e.g. by applying two electrodes into the capillary through a perpendicularly laser drilled hole. A CE instrument fitted with a conductimetric detector has recently become commercially available. A peculiar detection mode, suitable for measurement of ionic compounds which neither absorb UV radiation nor are fluorescent or electrochemically active, is ‘indirect detection’ (see Ref. [l] pp. 121-1260). The analytes are detected as ‘holes’ of UV absorbance (or fluorescence or electrochemical activity) of the background electrolyte, which contains a UV absorbing (or fluorescent or electrochemically active) ionic additive. This additive is displaced, to maintain the electroneutrality, from the zones occupied by the analytes of interest, thus producing ‘negative’ peaks at the detector. Indirect detection is ‘universal’ (but often too non-specific) and does not require instrumental hardware changes. Drawbacks are sensitivity, which is l-2 orders of magnitude lower than with the corresponding ‘direct’ detection mode, and the narrower range of linearity. A variety of other detection techniques have been applied in CE including: laser based thermo-optical detection, refractive index detection, radioisotope detection and notably, mass spectrometric (MS) detection, also in the configuration MS-MS. Because of the importance of information provided by MS techniques in almost all analytical environments, many efforts have been spent in the development of interfaces between CE and MS (see Ref. [l] pp. 131- 145). The electro-spray interface seems the most suitable for CE. The extremely low flow rate of CE is easily handled by this interface, in which also a sheath flow can be added in order to allow the composition of the electrospray liquid to be independent of CZE buffer (often with high ionic strength). Also, ion-spray/atmospheric pressure ionization and continuous flow fast-atom bombardment have been coupled to CE with success. 3. Separation techniques Highly efficient separations can be performed in CE using different modes, such as capillary zone electrophoresis (CZE), micellar electrokinetic chromatography (MEKC or MECC), capillary electrochromatography (CEC), capillary isoelectric focusing (CIEF), capillary gel electrophoresis (CGE) and capillary isotachophoresis (CITP). It is noticeable that the separation modes can be switched with no or minimal changes in the instrumental hardware. 3.1. Capillary zone electrophoresis (CZE) The theory of electrophoretic separations finds its most direct application in CZE, but most of its fundamental principles operate also in the other CE separation modes. In CZE the ionic analytes are separated according to a plain electrophoretic principle, on the basis of the differences in their electrophoretic mobilities, resulting in different migration velocities. The migration velocity (c) of a charged particle in an electric field depends on its electrophoretic mobility (1~~) and on the applied electric field (E) in which ,LL,is described by the following equation: p, = 4 67rtp where y = ion charge, r = ion radius, ~1 = solution viscosity. From the above equation, it is clear that molecular dimensions and charge are the main factors influencing ion mobility. For weak acids and bases, it is obvious that the charge is dependent on the individual pK values and on the pH of the running buffer. An additional fundamental factor operating in CZE (as well as in other CE modes) is electroosmosis. This phenomenon, causing an electrically driven flow of liquid in the capillary, occurs in any electrophoretic separation, but, while in slab gel electrophoresis it is a disturbing factor, in CE it is responsible for important aspects of the separation process. Electroosmosis originates from the charges present on the inner surface of the capillary resulting, in the case of fused silica capillaries, from the dissociation of silanols (SO ~ ) occurring at pH values > 2. These negative charges at the wall attract cations of the buffer, which create a double-layer and a potential difference (zeta potential, i). The application of a voltage difference along the capillary, causes the migration of these cations toward the cathode, which osmotically drag water in the same direction. The linear velocity of the electroosmotic flow is described by the following equation: i: 1' ,.<, = ~ 4Jlt1 i “E where: ;: = dielectric constant; i = zeta potential, q = solution viscosity. In short, electroosmotic flow, in most cases, is directed from the anodic to the cathodic end of the capillary; it can be empirically measured by injecting a neutral marker (e.g. acetone) and determining its migration time. A specific feature of electroosmotic flow is that, being generated at the walls of the capillary. its flow profile is almost flat, without the parabolic shape typical of the pressure generated laminar flow, as it occurs in capillary liquid chromatography (Fig. 3). It is clear that this characteristic is important in limiting the broadening of the zones during their migration. Considering that a ‘normal’ set up of the CE instrument necessitates that injection is at the anodic end and detection is close to the cathodic end, and that the electroosmotic flow velocity is greater than the electrophoretic velocity of most analytes, it is clear that in CE cations, neutral species and anions can be detected in the same run. 218 F. Tagliaro et al. 1 Forensic Science International 77 (1996) 21 l-229 The actual migration velocity of cations and anions will result from the algebraic summation of the individual electrophoretic mobilities (in the same direction of the electroosmotic flow or in the opposite, respectively) and the electroosmotic flow. All neutral species will migrate together at the same velocity as the electroosmotic flow: this means that non-ionic compounds, could not be separated, reaching the detector in a single zone. In the presence of the electroosmotic flow, the migration velocity follows the equation: where: tie0 and ,u~= mobilities of EOF and of the analyte, respectively; V = potential; L = capillary length. A careful control of electroosmotic flow is necessary for obtaining the best performances from CE: in fact it is fundamental in MEKC and CEC, often useful in CZE, but not desirable (or acceptable only if slow) in CGE, CIEF and CITP. To this aim, the capillary surface is passivated with modifiers physically adsorbed or chemically bound or added to the buffer (dynamic coating). Most common coatings are polyacrylamide, PEG, cellulose, surfactants, organic amines, etc. Since CE is a highly efficient separation technique, any cause of zone broadening is deleterious. Zone broadening, with resulting loss of efficiency, can be caused by different factors, among which are worth mentioning: Joule heating, adsorption of analytes onto the capillary wall, siphoning between the two electrode jars, injection volume, high conductivity of injected samples (in comparison to that of the running buffer) and extra column void volumes. electroosmotic pressure Fig. 3. Schematic capillary. representation flow profile driven flow profile of the electroosmotic and the pressure driven liquid flow profiles in a Joule heating can be controlled by using narrow bore capillaries, low conductivity buffers and column thermostating. Adsorption of analytes onto the capillary wall can be decreased by increasing the buffer ionic strength (by competing with the charges of the wall), working at low pH ( < 3) (in order to suppress silanol ionization) or at high pH ( > 9) (in order to suppress the positive charges of analytes), or using buffer additives or capillary surface coatings. In general, high buffer concentrations have been shown to reduce the adsorption of analytes to the capillary walls. In CZE, the main factors affecting the separation are buffer pH and composition. Important characteristics of the background buffer are to have a good buffering capacity at the chosen pH, low conductivity (in order to allow the use of high potentials without generating high currents) and low interferences with the detection system (basically, low UV absorbance). In order to avoid peak distortion. mobility of buffer ions should be matched with that of analytes. For most purposes. phosphate, borate, citrate and phosphate/borate buffers are suitable. Buffer additives can markedly affect selectivity and resolution. Zwitterionic buffers are sometimes necessary in order to work at high concentrations without excessive current. From the principles mentioned above, it is clear that in CZE neutral substances migrate at the same velocity of the electroosmotic flow and can not be separated. A different analytical approach, which applies also to non-ionic compounds, is represented by MEKC, which was introduced around the middle of 1980s by Terabe [12-141. In this system, which is based on chromatography-like separation mechanisms, a micellar pseudo-stationary phase is introduced in the buffer. which retains the solutes according to the partitioning between the ‘stationary phase’ itself and the mobile phase (i.e. the running bulfer). In MEKC, the flow of mobile phase is generated by electroosmosis, which produces a liquid flow, with a ‘plug-like shape’, which maintains the high efficiency of separation. The so-called ‘pseudo-stationary’ phase is obtained by addition to the buffer of a surfactant at supramicellar concentration, in most cases, sodium dodecyl sulfate (SDS). The anionic SDS micelles migrate towards the anode, thus reducing the net mobility of other solutes, depending on the degree of mutual interactions (mainly partitioning) they establish. A selectivity criterion based on partitioning of solutes in the lipophilic core of the micelles is. thus, introduced. This operates with non-ionic substances in a ‘quasi chromatographic’ mode, the micelles acting as the stationary phase and the electroosmotic flow of buffer (often containing additives) the mobile phase. In short, polar molecules will elute faster than hydrophobic’ compounds, within a ‘time window’ determined by the mobility of the electroosmotic flow (to) and that of the micelles (t,,,) (Fig. 4). For neutral solutes, the only difference from chromatography is that the ‘stationary phase’ in MEKC has its own migration velocity (t,,), which is measured by using a water-insoluble dye (e.g. Sudan III or Sudan IV) completely included in the micelles. 220 F. Tagliaro et al. 1 Forensic Science [ International 77 (1996) [<hydrophobic . 21 I-229 pj . electroosmotic flow micelle migration Fig. 4. Schematic representation of a MEKC electropherogram: t, = time of migration of a compound excluded from the micelles, simply driven by the electroosmotic flow; t, = time of migration of a compound fully included in the micelles. For charged solutes, the migration behaviour is dependent not only on the degree of micellar solubilization, but also on the electrophoretic mobility of the ionized form of the compound in the aqueous phase. Several surfactants are used in MEKC, but the most popular are: sodium dodecyl sulphate (SDS), bile salts and hydrophobic-chain quaternary ammonium salts. Factors important in the modulation of selectivity are additives of the aqueous buffer phase, in particular, organic solvents (methanol, isopropanol, acetonitrile) have been used up to a concentration of 50% in order to reduce both the hydrophobic interactions between analytes and micelles, similarly to reversed-phase chromatography, and to improve efficiency. 3.3. Cupillury isotachophoresis(CITP) CITP is another important separation mode in CE, which resembles classical isotachophoresis. Although based on the electrophoretic mobilities of analytes, the separation principle of this technique is absolutely peculiar. The ionic compounds migrate in discrete zones at the same velocity between two ionic solutions, one with the highest mobility (leading electrolyte) the other with the lowest mobility (terminating electrolyte), among all the ions. The different ionic substances migrate in discrete zones after the leading electrolyte, according to the individual mobilities, similar to carriages in a train (for a comprehensive presentation, see Ref. [15]). 3.4. Inclusion conlpkues ~~ Cliird scpr-rrtions A field in which CE is already proving to be superior to liquid and gas chromatography is the separation of enantiomers (seeRef. [l] pp. 257 -269). Chiral separation can be achieved either in the presence of chirally selective solutes or by means of chirally selective micelles. The chirally active solutes include cyclodextrins (CDs), modified CDs, bile salts, crown ethers, Cu(Il)-L-histidine. Cu(II)-aspartame. etc. Chiral surfdctants forming chirally functionalized micelles have also been used. The separation is based on the reversible formation of sterically selective inclusion complexes between the chiral selector and the analyte molecules during electrophoresis. Obviously, in order to allow chiral separation, the migration velocity of the complexed molecules must be different from that of the ‘free’ drugs. Many variables can be adjusted to optimize CE chiral separation, among which chiral selector type and concentration, organic modifier concentration. ionic strength of the buffer. temperature, applied field strength, etc. Special mention should be made to CDs [40-421. These oligosaccharides have an external hydrophilic surface and a hydrophobic cavity. in which they can include other compounds by hydrophobic interaction. This inclusion mechanism is steritally selective, but also massselective; analytes must fit the size of the cavity, which changes in dependence of the number of glucose units in their structure (6, 7 and 8, respectively for x-, p- and ;t-CDs). Native CDs are neutral and highly hydrophilic: therefore they migrate at the velocity of the electroosmotic flow. Newly introduced charged CDs can display a counter migration in relation to the electroosmotic flow. thus showing a retardation effect on the compounds they interact with. Neutral CDs, when added to micellar solutions, selectively decrease the interactions of hydrophobic compounds with the micelles. Another. alternative possibility for achieving chiral selectivity is the dericatization of the analytes with chiral reagents followed by separation of the resulting diastereomers. Among CE separation modes. CEC is that which most resembleschromatography (see Ref. [l] pp. 194- 195). In fact, a chromatographic stationary phase is contained (mostly packed) in the capillary and interacts with the solutes according to the usual chromatographic separation mechanisms.The main peculiarity of CEC is that the flow of the mobile phase (i.e. run buffer) is generated by electroosmosis. The plug shape of the liquid flow and the absenceof pressurepulsations (usual with reciprocating pumps) represent clear advantages over the pressure driven flow. Thus, higher separation efficiency can. in general. be obtained. With chirally selective packings, efficient chiral separations are accomplished. However, the current stage of development of CEC is lower than that of the other CE separation modes and its practical application is still limited. In the analysis of drugs, CE has mainly been applied as MEKC. CZE and, sometimes, as CITP and CEC. Other separation modes popular in CE. such as 222 F. Tagliaro et al. / Forensic Science International 77 (1996) 21 l-229 capillary gel electrophoresis and capillary isoelectric focusing are suitable for the separation of biopolymers, but have not yet found application in forensic toxicology. 4. Applications of CE in forensic toxicology The analysis of drugs, in both pharmaceutical preparations and body fluids, represents one of the most rapidly growing application areas of CE. Specific applications in the field of illicit or controlled drugs have appeared only a few years ago and are still limited, but tend to increase rapidly. For this reason, the present review can not be fully updated, but simply gives a glimpse of the potential of this novel analytical technology, until now fairly esoteric in analytical toxicological disciplines. An excellent review on the principles of CE and its use for the analysis of seized drugs, by IS. Lurie [16], has appeared in a recent book dedicated to the analysis of illicit and misused drugs, edited by J.A. Adamovics. To the best of our knowledge, Weinberger and Lurie [17], in 1991, first applied CE and particularly MEKC, to the analysis of illicit drug substances. The authors used 50 pm i.d. bare silica capillaries, 25-100 cm in length, and a background buffer consisting of 85 mM SDS, 8.5 mM phosphate, 8.5 mM borate, at a pH of 8.5, which was added with 15% acetonitrile. The applied voltages were 25-30 kV; detection was by UV absorption at 210 nm. Under the described conditions, it was possible to separate with high efficiency, acidic and neutral impurities in illicit samples of heroin. In general, relative standard deviation (RSD) for migration times was w 0.5% and for areas and peak heights 448%. However, with the late peaks (migration time > 40 min) analytical precision was worse, a phenomenon ascribed to the inconsistent evaporation of the organic modifier from the buffer reservoirs. As could be expected, better resolution was achieved using a smaller diameter capillary (25 pm instead of 50 pm i.d.), but the sensitivity was inferior, because of the reduced optical path length. Also, fluorescence detection (ex. wavelength 257 nm, em. 400 nm) was tested, achieving higher sensitivity for the fuorescent analytes (more recently, to the same aim, krypton-fluoride laser excitation at 248 nm has also been used; see Ref. [43]). In the same paper, the separation of bulk heroin, heroin impurities, degradation products and adulterants was demonstrated. The substances migrated in the following order: morphine, phenobarbital, 6-monoacetylmorphine (MAM), 3MAM, methaqualone, heroin, acetylcodeine, papaverine, noscapine. The CE separation pattern was different from that of reversed-phase HPLC. A sample of illicit cocaine was also analyzed, with excellent peak shapes and resolution of benzoylecgonine, cocaine, cis- and trans-cinnamoylcocaine. In comparison with HPLC, MEKC allowed the resolution of about twice as many peaks; on the other hand, HPLC was more sensitive. However, using MEKC, it was still possible to detect heroin impurities down to 0.2%. To test the performances of MEKC in the separation of a larger panel of compounds of potential forensic interest, a mixture of as many as 18 drugs. including psilocybin, amphetamines, benzodiazepines, PCP, cannabinoids, etc. was separated with baseline resolution [ 171. The application of MEKC for the analysis of illicit heroin and cocaine was also successfully demonstrated by Staub and Plaut [18], who used 50 mM SDS in phosphate/borate buffer (10 and 15 mM, respectively, pH 7.8) containing 15% of acetonitrile. In 25 min, paracetamol, caffeine. 6-MAM, acetylcodeine, procaine, papaverine, heroin and noscapine were separated in a heroin seizure and paracetamol, benzoylecgonine and cocaine in an illicit cocaine sample. On-line recorded UV spectra of the peaks helped the identification of the sample components. A more extensive study of MEKC for the analysis of heroin and amphetamine seizures has recently been published by Krogh et al. [19]. A 50 cm x 50 pm i.d. bare silica capillary was used; the running buffer was composed of 25 mM SDS in phospate/borate buffer (10 mM for each salt) pH 9, containing 5% acetonitrile. Detection was by UV absorption at 214 nm. A test mixture of the main alkaloids found in illicit heroin and heroin adulterants (on the whole, 14 compounds), as well as a mixture of seven drugs structurally related to amphetamine were completely resolved in 15 min. RSD values of 0.5-l .9X and of 0.8992.23’% were obtained for migration times (relative to crystal violet, the internal standard) in within-day and between-day reproducibility tests, respectively. In quantitative studies, typical standard curves were linear in the range from 0.02 to 0.5 mgjml, with correlation coefficients from 0.997 to 0.999. Typical RSD values in the analysis of illicit heroin and amphetamine were in the range 2.0 -4.3%. Samples could be injected every 13 min; the fused silica capillary was replaced only after 500 injections to assure good reproducibility. The authors concluded that CE is a valuable complement to HPLC and GC in for the analysis of illicit drugs. Trenerry et al. [20] reported the use of 50 mM cetyltrimethylammonium bromide (CTAB), as micellar agent, as an alternative to SDS. The use of this cationic micellar agent (50 mM in phosphate/borate buffer 10 mM, pH 8.6 with 10% acetonitrile, at - 15 kV (reversed polarity), UV detection at 280 nm) was reported to give faster separations ( - 15 min) of heroin and related substances, in comparison to SDS. This method was successfully tested with different illicit heroin preparations in comparison to HPLC. A good correlation was found between the two techniques. Precision of MEKC was slightly worse than that of HPLC. but the resolving power of CE was much higher. allowing the complete resolution even of very complex heroin seizures. The same authors [21] reported an application of the above mentioned MEKC method (only variations in acetonitrile from 10 to 7.5% and detection wavelength from 280 to 230 nm) for the analysis of cocaine and related substances. Benzoylecgonine, cocaine, cis- and trn/r.s-cinnamoylcocaine were separated in real cocaine seizures and the results compared with HPLC and GC. MEKC and GC correlated pretty well to each other, with similar RSD values, while HPLC gave slightly higher results (no explanation was gi\:en). 224 F. Tagliaro et al. / Forensic Science International 77 (1996) 21 l-229 According to the authors, these MEKC methods passed inter-laboratory proficiency tests. The separation of enantiomers of amphetamine, methamphetamine, ephedrine, pseudoephedrine, norephedrine and norpseudoephedrine with application to forensic samples was reported by Lurie [22], as a tool for investigations on the synthetic methodologies. The author used derivatization with a chiral derivatizing reagent, 2,3,4,6-tetra-0-acetyl-/?-D-glucopyranosyl isothiocyanate, followed by MEKC of the resulting diastereomers. The 12 enantiomers of all the above mentioned phenethylamines were resolved in a single run, using a bare silica capillary (48 cm x 50 pm i.d.) and a buffer consisting of 20% methanol and 80% SDS solution (100 mM SDS, 10 mM phosphate/borate buffer, pH 9.0). More recently, Lurie et al. [23] reported the chiral resolution of a number of basic drugs of forensic interest, namely amphetamine, methamphetamine, cathinone, methcathinone, cathine, cocaine, propoxyphene and various a-hydroxyphenethylamines with the use of neutral and anionic CDs. In this separation scheme, resolution was optimized by varying the ratio of the neutral/anionic CDs. In fact, both types of CDs have chiral selectivity, but anionic CDs, due to their negative charge, display an electrophoretic counter migration and consequently higher retarding effect on analytes. Chiral resolution, could be achieved also in CEC, using an enantioselective stationary phase, thus mimicking HPLC separations, but, reportedly, with higher efficiency. This approach was adopted by Li and Lloyd [24], who used @,-acid glycoprotein as stationary phase packed in fused silica capillaries of 50 pm id. for the separation of the enantiomers of hexobarbital, pentobarbital, P-blockers and other drugs. As an alternative to MEKC, Chee and Wan [25], used CZE with 50 mM phosphate buffer, pH 2.35 in a 75 pm id. 60 cm long bare silica capillary. Detection was by UV absorption at 214 nm. The authors achieved the separation of 17 basic drugs in only 11 min, including amphetamine, methamphetamine, medazepam, lidocaine, diazepam, methaqualone, etc. Under these conditions, according to a plain electrophoretic mechanism of separation, drugs having lower pK, values and consequently less positive charge, showed higher migration times. However, the influence of other factors (according to the authors: molecular size, tendency to interact with the column and ability to form doubly charged species) hampered a clear correlation between pK, values and migration times. The migration time RSD values were, in general, less than 1% and peak-area RSDs were between 1.5 and 4.3%. Poorer reproducibility was found for analytes with very slow migration. The described method was applied to test drug mixtures, but not on real forensic exhibits. However, it proved suitable for the analysis of biological fluids, after simple chloroform/isopropanol (9: 1) extraction. The reported sensitivity was N 0.45 pug/ml, either in blood or urine. According to the authors, CZE offered advantages over MEKC for drug screening and, particularly, simple background electrolyte preparation and shorter analysis times. The main limitation was the inahility to analyze acidic, neutral and basic drugs all together. Other research groups active in the field of forensic and clinical toxicology have focused their attention on the use of CE for the toxicological investigation of biological fluids. Wernly and Thormann [26], using MEKC with an aqueous borate/phosphate buffer, pH 9.1, containing 75 mM SDS, reported the qualitative analysis of many drugs of abuse and metabolites in urine, including benzoylecgonine, morphine, heroin, 6-MAM, methamphetamine. codeine, amphetamine, cocaine, methadone, methaqualone and benzodiazepines. Urine purification (5 ml) and concentration was by ‘double mechanism’ (cation exchange and reversed phase) solid-phase extraction cartridges; the dried extract was then reconstituted with 100 1~1of buffer, and directly injected, allowing a sensitivity of 100 ng:‘ml. Peak identification was achieved on the basis of the retention times and of the on-line recorded UV spectra of the peaks. CE showed a sensitivity comparable to usual non-isotopic immunoassays and was proposed for confirmation testing, following usual screenings by enzyme-immunoassays. Also, MEKC (50 mM SDS in phosphate-borate buffer, pH 7.8) was used for rapid and high-resolution separations of barbiturates, namely barbital, allobarbital, phenobarbital, butalbital, thiopental, amobarbital. pentobarbital, with on-column multi-wavelength detection [27]; achieving sensitivities in the range of the low /Lgg/ml. Urine samples required extraction of barbiturates prior to analysis, but with human serum barbiturates eluted in an interference-free window of the electropherogram, allowing these substances to be determined by direct sample injection also. MEKC with 75 mM SDS in phosphate-borate buffer, pH 9.1 allowed the determination of l l-nor-delta-9-tetrahydrocannabinol-9-carboxylic acid, the major metabolite of delta-9-tetrahydrocannabinol in urine [28]. Using basic hydrolysis of 5 ml of urine, followed by solid phase extraction, it was possible to achieve the sensitivity of lo-30 rig/ml. Again, the on-line recording of the peak spectrum. by means of a fast scanning UV detector. gave an additional opportunity for confirmation. The use of MEKC for determining morphine-3-glucuronide. the major metabolite of heroin and other opiates, omitting any hydrolysis, was reported by Wernly for confirmation of et al. (1993), but the sensitivity. - 1 /lg;rnl, was unsatisfactory the results of the enzyme-immunoassays around the NIDA cut-off. Benzodiazepines are also suitable for CE analysis. as reported by Schafroth et al. [29]. who determined the major urinary compounds of eight common benzodiazepines (flunitrazepam. diazepam. midazolam. clonazepam. bromazepam, temazepam, oxazepam and lorazepam) by MEKC. using 75 mM SDS in phosphate/borate buffer, pH 9.3. After enzymatic hydrolysis and solid phase extraction with commercial ‘double mechanism’ cartridges, they achieved a sensitivity better than with the EMIT”’ immunoassay. Due to the limited sensitivity of CE with UV detection, sample preparation and pre-concentration is crucial for applications in biological samples. A stepwise solid-phase extraction for human urine preliminary to MEKC analysis, using 226 F. Tagliaro et al. / Forensic Science International 77 (1996) 21 i-229 commercial disposable cartridges with hydrophobic and ion-exchange interactions, was described by Wernly and Thormann [30]. This clean-up procedure allowed up to 50 times concentration of urine samples. An overview of the strategies for monitoring drugs in body fluids (serum, urine and saliva) by MEKC, focused on buffer selection and sample preparation procedures, including direct injection, ultrafiltration and solid-phase extraction, has appeared recently [3 11. The comparative use of MEKC, CZE and CITP for the determination of drugs in body fluids was reported by Caslavska et al. [32]. For MEKC a buffer composed of 75 mM SDS, in a borate-phosphate buffer, pH 9.1 was employed. CZE analyses were executed with a 33 mM phosphate buffer, pH 8.3. CITP was performed with a leader of 10 mM HCl and histidine, pH 6 and a terminator composed of 10 mM 2-(N-morpholino)ethanesulphonic acid and histidine, pH 6.0. Salicylate, paracetamol and antiepilectics were analyzed in serum and urine. In case of high drug concentrations, body fluids could be injected directly or with simple pre-treatments, such as dilution (urine) or ultrafiltration (serum). However, in cases of drugs at the low pg/ml level, extraction and concentration were required. The authors concluded that MEKC and CZE were easier to be applied, whereas CITP required careful selection of buffers and proved generally less sensitive. Only with the ITP spike technique, in which baseline-resolved UV absorption peaks are produced by bracketing the solute with discrete non-absorbing spacers, CITP could approach the sensitivity of MEKC and CZE. Hair analysis is gaining increasing popularity in the forensic toxicology environment, as a tool for investigating past, chronic exposure to illicit drugs, as is shown by an increasing body of literature [33,34]. To this end, CE, in principle, could offer several advantages over current techniques (GC, GC-MS, HPLC). mainly due to the requirement of minimal sample. To the best of our knowledge, the few reports published until now come only from our group. Initally we adopted CZE separations, because this mode of operation appeared simpler to be applied and replicated [35,36]. In order to allow rapid reconditioning of the capillary after the washing cycles with alkali (0.1 M NaOH), a must, after injection of complex biological extracts, we chose a basic CZE background buffer, i.e. 50 mM borate, pH 9.2. For the quantitative determination of cocaine and morphine, detection was either at 200 nm wavelength (for both substances) or at the absorbance maxima of each analyte (for cocaine 238 nm; for morphine 214 nm). Tetracaine and nalorphine were chosen as I.S. for cocaine and morphine, respectively. Excellent resolution and peak shape was obtained for both the analytes and the respective I.S. The separations were highly efficient (up to 350 000 theoretical plates) and repeatable (migration time RSDs: < 1% intra-day, < 3% inter-days), the determinations accurate and precise (intra-day RSDs in the range 335%). Due to the tiny volumes of sample injected (few nl) and the moderate concentration sensitivity of CE with UV detection (in the order of the low pg/ml), the limit of detection in hair was acceptable ( < 0.2 ng/mg) only if the hair extracts were reconstituted with lo--20 1’1, which is almost impracticable in real routine work situations. More recently [37], the use of sample stacking techniques has allowed the injection of about ten times larger volumes (i.e. - 60 nl), thus allowing reconstitution of the extracts with as much as 100 ~11of water, without sacrificing efficiency. In the same paper, MEKC has been applied to hair samples using 0.1 mol/l SDS in 0.025 mol/l borate containing 20% methanol. The sensitivity achieved was only slightly worse than with CZE, but the selectivity was much higher, due to the additional ‘reversed-phase like’ separation mechanism exploited by the SDS micelles. 5. Conclusion In conclusion, CE has proved to be a new analytical tool suitable for the investigation of both seized preparations of illicit drugs. as well as as biological samples. This technique displays several characteristics, such as electrophoretic and/or chromatography-like separation mechanisms, negligible consumption of samples and reagents, instrumental simplicity and multiple detection modes, which make it unique on the scene of modern analytical technology. In particular, the possibility of carrying out separations based on different physical-chemical mechanisms (CZE. MEKC and CITP) with the same instrumentation, provides an interesting possibility of ‘internal’ confirmation of the results. The possibility of interfacing with mass spectrometry is an additional feature of CE crucial in the forensic environment, which is ready to become commercially available, mainly with electrospray [38] or atmospheric pressure ionization [39] interfaces. Therefore. CE seems to have an important future in forensic toxicology as well as in other fields of forensic science (e.g. explosive and gunshot residue analysis and DNA fingerprinting) and is already ready for use as a complement to the usual. more consolidated analytical techniques. Acknowledgements The authors gratefully acknowledge the assistance of the personnel of the Library Services of the Hospitals of Verona (Biblioteca Marani) and of the Glaxo Research Center, Verona, Italy, in obtaining the bibliographic material needed for this review. References [I] S.F..Y, LI. Capillary Electrophorws. J U~wm/rtq,-. L/hwv~.. Vol. 52. Elsevier. Amsterdam. 1991. 228 F. Tagliaro et al. / Forensic Science International 77 (1996) 21 l-229 [2] R.A. Mosher, D.A. Saville and W. Thormann, The Dynamics of Electrophoresis, VCH, Weinheim, Germany, 1992. [3] P.D. Grossman and J.C. Colburn, Capillary Electrophoresis Theory and Practice, Academic Press Inc., San Diego, 1992. [4] J. Vindevogel and P. Sandra, Introduction to Micellar Electrokinetic Chromatography, Huthig Verlag, Heidelberg, Germany, 1992. [5] R. Weinberger, Practical Capillary Electrophoresis, Academic Press, Boston, 1993. [6] F. Foret, L. Krivankova and P. Bocek, Capillary Zone Electrophoresis, VCH, Weinheim, Germany, 1993. [7] P. Camilleri, Capillary Electrophoresis: Theory and Practice, New Directions in Organic and Biological Chemistry, CRC Press, Boca Raton, FL, USA, 1993. [S] R. Khun and S. Hoffstetter-Kuhn, Capillary Electrophoresis: Principles and Practice, Springer-Verlag, New York, 1993. [9] P. Jandik and G. Bonn, Capillary Electrophoresis of Small Molecules and Ions, VCH, New York, 1993. [IO] N.A. Guzman, Capillary Electrophoresis Technology Chromatographic Science Series, M. Dekker, New York, 1993. [l l] D. N. Heiger, High Performance Capillary Electrophoresis, Hewlett-Packard, Waldbronn, Germany, 1992, pp. 100~101. [12] S. Terabe, K. Otsuka, K. Ichikawa, A. Tsuchiya and T. Ando, Electrokinetic separations with micellar solutions and open-tubular capillaries. Anal. Chem., 56 (1984) 111-I 13. [I31 S. Terabe, K. Otsuka and T. Ando, Electrokinetic chromatography with micellar solution and open-tubular capillaries. Anal. Chem., 57 (1985) 8344841. [14] S. Terabe, Electrokinetc chromatography: an interface between electrophoresis and chromatography. Trends Anal. Chem., 8 (1989) l29- 134. [l5] F.M. Everaerts, J.L. Beckers and T.P. Verheggen, Isotachophoresis: Theory, Instrumentation and Applications, J. Chromatogr. Library, Vol. 6, Elsevier, Amsterdam, 1976. [16] I.S. Lurie, Analysis of seized drugs by capillary electrophoresis. In J.A. Adamovics (ed.), Analysis of Addictioe and Misused Drugs, Marcel Dekker, New York, 1994, pp. 151-219. [17] R. Weinberger and I.S. Lurie, Micellar electrokinetic capillary chromatography of illicit drug substances. Anal. Chem., 63 (1991) 823-827. [18] C. Staub and 0. Plaut, High performance capillary electrophoresis. A new tool in forensic toxicology? In R.K. Mueller (ed.), Proceedings of the 31st International Meeting o/’ TIAFT, Leipzig ‘93, MOLINApress, Leipzig, 1994, pp. 4522458. [19] M. Krogh, S. Brekke, F. Tonnesen and K.E. Rasmussen, Analysis of drug seizures of heroin and amphetamine by capillary electrophoresis. J. Chromatogr., 674 (1994) 235-240. [20] V.C. Trenerry, R.J. Wells and J. Robertson, The analysis of illicit heroin seizures by capillary zone electrophoresis. J. Chromatogr. Sci., 32 (1994) l-6. [21] V.C. Trenerry, J. Robertson and R.J. Wells, The determination of cocaine and related substances by micellar electrokinetic capillary chromatography. Electrophoresis, 15 (1994) 103- 108. [22] I.S. Lurie, Micellar electrokinetic capillary chromatography of the enantiomers of amphetamine, methamphetamine and their hydroxyphenethylamine precursors. I. Chromatogr., 605 (1992) 269215. [23] I.S. Lurie, R.F.X. Klein, T.A. Dal Cason M.J. LeBelle, R. Brenneisen and R.E. Weinberger, Chiral resolution of cationic drugs of forensic interest by capillary electrophoresis with mixtures of neutral and anionic cyclodextrins. Anal. Chem., 66 (1994) 4019-4026. [24] S. Li and D.K. Lloyd, Direct chiral separations by capillary electrophoresis using capillaries packed with an %,-acid glycoprotein chiral stationary phase. Anal. Chem., 65 (1993) 368443690. [25] G.L. Chee and T.S.M. Wan, Reproducible and high-speed separation of basic drugs by capillary zone electrophoresis. I. Chromatogr., 612 (I 993) 172- 177. 1261 P. Wernly and W. Thormann, Analysis of illicit drugs in human urine by micellar electrokinetic capillary chromatography with on-column fast scanning polychrome absorption detection. Anal. Chem., 63 (1991) 2878-2882. 1271 W. urine Thormann, by P. Meier. C. Marcolli high-performance and capillary F. Binder. Analysis of barbiturates electrophoresis-micellar in human electrokinetic capillary serum [28] phy with on-column multi-wavelength detection. J. Clrromnro~r., 545 (1991) 445.-460. P. Wernly and W. Thormann, Confirmation testing of I I-nor-delta9-tetrahydrocannabinol-Y-carboxylic acid in urine with micellar electrokinetic capillary chromatography. .1. Cl~omuror-.. [29] (1992) X-256. M. Schafroth. benzodiazepines [30] [?I] W. P. Wernly solid-phase Thormann in human and W. extraction 2155 2159. W. Thormann, and urine. Thormann, and micellar S. Lienhard and [32] micellar electrokinetic capillary J. Caslabska. S. Lienhard and methods for the determination [33] F. Tagliaro. [34] investigation. P. Kinz (ed.), [35] [36] D. Allcman. Micellar electrokinetic ~/‘lcc,~~o/,/~o~r,.\i.v. IS ( 1994) 72 78. M. Chiarotti and Special Issue. Hair analysis Drug of electrokinetic P. Wernlq. abuse Z. Deyl F0lerr,trc, in forensic (eds.). for the monitoring Hair analysis Sci. fti/.. 63 (1993) touicologb Hair 70 (1995) I 222. F. Tagliaro, C. Poiesi, R. Aiello. R. DorIG. S. Ghielmi and for the investigation of drugs in hair: determination 01‘ cocaine (lY93) 303 F. Tagliaro, 309. C. Antonioli. S. Moretto. F. Tagliaro. in forensic W.F. Smyth. S. Turrina, toxicology. Applications S. Archetti, 70 (1995) 93 104. R.D. Smith, H.R. Usdeth. [39] capillar) electrophoresis-mass I.M. Johansson, R. Pavelka spectrometry. and J.D. Henion. pressure [40] electrophoresis-atmospheric 515 528. S. Fanali, Separation [4l] complexation M. Heuermann [42] pseudo-stationary M.W.F. Nielen, [43] clectrophoresis. I.S. Lurie. and neutral C.J. of Barinaga optical K.C. A?rul. Chan. impurities heroin Edmonds. mass capillary fuids bq capillary 337. drugs of abuse Marigo. Capillary electrophoresis and morphine. J. C’l~on~crrc~~~.. 638 and M. I;oren.sic Marigo, High liquid illicit sensitivity chromatography electrophoresis: a new tool drugs. Fo~rrrvic, SC,/. /r/r.. Instrumentation for high-performance 559 (1991) of small spectrometry. capillary electrophoresis. of basic drugs tin in body Irff.. M. Marigo. Capillary in hair analysis for Clle/)~., 65 (1993) XX5 893. T.K. Spratley. J.F. Casale and in illicit stepwise 64 ( 1092) .%r. zone I97 208, drug molecules J. H.J. Issaq. Separation 559 based on and J. C/rro,n~r/og~.. ( 1991 ) host-guest cyclodextrms J. C’/~ro,nrrto~~r.. 648 (1993) using cycloderctrin-modified clectrophoresib. by capillary C‘l~~rncrrofi,.. electrophoresis ./. C./~rorr~trrogr.. 474 (I 989) 441 -446. Chiral separation of basic drugs using in capillary separation of 63 (I 993) 237--2X3. C.G. by using Clrern.. for Issue. high-performance J. Ci~ro~~clro~r.. Determination ioniration isomers with cyclodextrins. and G. Blaschke. phases Chiral and M. tool II. Special S. Ghielmi in hair: 2. Deyl and and prospects [38] urine Am//. of drugs as a diagnostic I 314. analysis 608 chromatography J. Cl~uotntrro,qr.. 636 (1993) 137- 148. Comparatne use of three electrokinetic fluids. J. (%rormrtogr.. 638 (1993) 335 low-cost methods for determination of cocaine and capillary electrophoresis. Forc~trt~ Si. Iut.. [37] confirmation in human capillar! chromatography. Strategies chromatography. W. Thormann. of drugs in body capillary and chromatogra- as chiral 267-~274. capillary detection 669 lone of acidic (1995) 1 13.