Postcolumn Reaction Detection with Dual-Electrode Capillary

This approach is a modification of a previously reported design in which gold rather ..... The authors thank Nancy Harmony for assistance in preparing...
3 downloads 0 Views 59KB Size
Download PDF
Anal. Chem. 1999, 71, 407-412

Postcolumn Reaction Detection with Dual-Electrode Capillary Electrophoresis-Electrochemistry and Electrogenerated Bromine Lisa A. Holland and Susan M. Lunte*

Department of Pharmaceutical Chemistry and Center for Bioanalytical Research, The University of Kansas, 2095 Constant Avenue, Lawrence, Kansas 66047

This is the first report of postcolumn amperometric reaction detection for capillary electrophoresis and dualelectrode detection. Bromide present in the run buffer is oxidized to bromine at the first electrode and subsequently detected at a second electrode downstream. Analytes that react with bromine cause a decrease in signal at the downstream electrode that is proportional to analyte concentration. Bromine is known to react with a variety of compounds, including thiols, thioethers, disulfides, amines, and unsaturated organic compounds. In this paper, the development of a new wire-wire on-capillary dual electrode that is well suited to bromine-based postcolumn reaction detection is described. System performance was evaluated using glutathione, cysteine, and methionine as test analytes. The final optimized system could be operated continuously for 24 h and was stable for day-to-day use for at least two weeks. The response for cysteine was linear from 0.5 to 20 µM with a limit of detection of ∼80 nM. The use of electrochemical (EC) detection with capillary electrophoresis (CE) has become increasingly widespread since the first report of such a system by Wallingford and Ewing.1 Several reviews on the instrumentation and applications of CEEC are available.2-4 Electrochemical detection has a number of advantages for use with CE. It is extremely selective because there are very few compounds that can be oxidized or reduced in biological samples, and this selectivity is tunable through control of the applied potential. Moreover, in contrast to most optical detection methods, the detector is easily miniaturized without a loss in sensitivity. Limits of detection (LOD) are in the nanomolar range for many compounds. Most applications of CEEC employ direct detection at carbon electrodes.1-9 Recently, the applicability of CEEC has been further (1) Wallingford, R. A.; Ewing, A. G. Anal. Chem. 1987, 59, 1762-1766. (2) Ewing, A. G.; Mesaros, J. M.; Gavin, P. F. Anal. Chem. 1994, 66, 527A537A. (3) Voegel, P. D.; Baldwin, R. P. Electrophoresis 1997, 18, 2267-2278. (4) Holland, L. A.; Lunte, S. M. Anal. Commun. 1998, 35, 1H-4H. (5) O’Shea, T. J.; Greenhagen, R. D.; Lunte, S. M.; Lunte, C. E.; Smyth, M. R.; Radzik, D. M.; Watanabe, N. J. Chromatogr. 1992, 593, 305-312. (6) Bergquist, J.; Tarkowski, A.; Ekman, R.; Ewing, A. Proc. Natl. Acad. Sci. U.S.A. 1994, 91, 12912-12916. 10.1021/ac980926d CCC: $18.00 Published on Web 12/12/1998

© 1999 American Chemical Society

Figure 1. Scheme of amperometric reaction detection employing electrogenerated bromine.

expanded through the use of additional electrode materials and new potential waveforms.10-14 Another way of increasing the applicability of CEEC is through dual-electrode detection15-17 and by indirect detection methods.18 King and Kissinger first reported liquid chromatography with postcolumn reaction detection employing electrogenerated bromine in 1980.19 In this approach, two working electrodes are employed (Figure 1). The first electrode is used to generate bromine by oxidation of the bromide present in the mobile phase. The bromine is then detected at the downstream electrode. If an analyte reacts with the bromine as it passes through the detector cell, there will be a decrease in signal at the second electrode that is proportional to the analyte concentration. (7) Malone, M. A.; Zou, H.; Lunte, S. M.; Smyth, M. R. J. Chromatogr., A 1995, 700, 73-80. (8) Hadwiger, M. E.; Park, S.; Torchia, S. R.; Lunte, C. E. J. Pharm. Biomed. Anal. 1997, 15, 621-629. (9) Xu, D. K.; Hua, L.; Li, Z. M.; Chen, H. Y. J. Chromatogr., B 1997, 694, 461-466. (10) Zhou, J.; O’Shea, T. J.; Lunte, S. M. J. Chromatogr., A 1994, 680, 271-277. (11) Huang, X.; Kok, W. Th. J. Chromatogr., A 1995, 716, 347-353. (12) O’Shea, T. J.; Lunte, S. M. Anal. Chem. 1993, 65, 247-250. (13) Zhou, W.; Baldwin, R. P. Electrophoresis 1996, 17, 319-324. (14) Owens, G. S.; LaCourse, W. R., J. Chromatogr., B 1997, 695, 15-25. (15) Lin, B. L.; Colon, L. A.; Zare, R. N. J. Chromatogr., A 1994, 680, 263-270. (16) Zhong, M.; Zhou, J.; Lunte, S. M.; Zhao, G.; Giolando, D. M.; Kirchhoff, J. R. Anal. Chem. 1996, 68, 203-207. (17) Zhong, M.; Lunte, S. M. Anal. Chem., in press. (18) Olefirowicz, T. M.; Ewing, A. G. J. Chromatogr. 1990, 499, 713-719. (19) King, W. P.; Kissinger, P. T. Clin. Chem. 1980, 26, 1484-1491.

Analytical Chemistry, Vol. 71, No. 2, January 15, 1999 407

One of the major advantages of this approach is the large number of compound types that react with bromine: these include thiols, thioethers, and unsaturated aliphatic compounds. Many of these compounds are difficult to detect by any other means. LC with postcolumn amperometric reaction detection has been used for the determination of fatty acids,19 prostaglandins,19 phenols,19,20 thiols,21,22 disulfides,21,22 proteins,22 and drugs.21,23,24 To date there has been no report of postcolumn amperometric reaction detection with capillary electrophoresis. To implement amperometric reaction detection in capillary electrophoresis, a dual-electrode configuration must be used. The first dual electrode reported for use with CE consisted of two amalgamated gold wire electrodes (25-µm o.d.) for the detection of thiols and disulfides.15 The first electrode was inserted through the capillary perpendicular to the flow, while the second electrode was inserted into the end of the capillary. More recently, both a micro-ring-disk carbon electrode and a tubular wire electrode have been described for CEEC with dual-electrode detection.16,17 In the current work, three different dual-electrode configurations for CE were evaluated for postcolumn reaction detection with electrogenerated bromine. The applicability of the final system for the detection of thiols and thioethers was demonstrated. EXPERIMENTAL SECTION Reagents. Cysteine, reduced glutathione, methionine, potassium bromide, and sodium phosphate were purchased from Sigma (St. Louis, MO). Deionized water was obtained from a Labconco water purification system (Labconco, Kansas City, MO). All standard solutions were prepared in background electrolyte. General Instrumentation. The four electrodes used for this detection scheme (platinum generating electrode, platinum detection electrode, platinum auxiliary electrode, Ag/AgCl reference electrode) were connected to a bipotentiostat (LC-4CE, Bioanalytical Systems, West Lafayette, IN). In all experiments, the generating electrode was maintained at +1.00 V. This potential was selected since it consistently led to the generation of a sufficient amount of bromine for the sensitive detection of thiols and thioethers with postcolumn amperometric reaction detection. Incomplete coupling of the detection voltage and the separation voltage in off-column CEEC can lead to a shift in the potential of the working electrode. Therefore, to ensure that the potential applied to the detection electrode was on the current-limiting plateau for the reduction of bromine, a potential of 0.00 V was used. Previous LC EC methods incorporating bromine-based amperometric reaction detection have employed detection potentials equal to or lower than 0.400 V vs Ag/AgCl.19,20,23 Data were collected and analyzed using a DA-5 analog-to-digital converter and Chromgraph software (Bioanalytical Systems). Injection. All injections were pressure-driven at 0.34 bar (5 psi) for 1 s. The injection volumes for the tube-wire, wire-ring, and on-capillary wire-wire configurations were 10, 33, and 59 nL, (20) Kok, W. Th.; Brinkman, U. A. Th.; Frei, R. W. Anal. Chim. Acta 1984, 162, 19-32. (21) Debets, A. J. J.; van de Straat, R.; Voogt, W. H.; Vos, H.; Vermeulen, N. P. E.; Frei, R. W. J. Pharm. Biomed. Anal. 1988, 6, 329-336. (22) Isaksson, K.; Lindquist, J.; Lundstrom, K. J. Chromatogr. 1985, 324, 333342. (23) Kok, W. Th.; Halvax, J. J.; Voogt, W. H.; Brinkman, U. A. Th.; Frei, R. W. Anal. Chem. 1985, 57, 2580-2583. (24) Isaksson, K. J. Chromatogr. 1987, 411, 229-236.

408

Analytical Chemistry, Vol. 71, No. 2, January 15, 1999

respectively. A timer built in-house controlled the duration of the injection using a commercially available programmable multifunction time delay relay/counter (Potter and Brumfield, Princeton, IN). The in-house timer triggered an electrically actuated pneumatic valve (MAC Valves, Wixom, MI) that allowed pressure pulses of brief duration to be delivered to the anodic reservoir. Decoupler. The same type of decoupler was used in all three configurations; it consisted of three bare fractures located between 0.5 and 1.5 cm from the detection end of the capillary. A singlefracture version of this decoupler has been described previously.25 Prior to fracturing, the end of the fused-silica capillary was immobilized onto a piece of glass with UV glue (UVEXS, Sunnyvale, CA). After the glue was cured, the capillary was scored at three locations (0.5, 1.0, and 1.5 cm from the end of the capillary) and then fractured by gently applying pressure at these positions. The decoupler was placed in a 3-mL vial, which had two small holes bored in it to accommodate the section of the capillary. Approximately 3 mm of the detection capillary protruded from the outer surface of the vial. The vial was filled with the background electrolyte/potassium bromide solution and served as the cathodic reservoir. Because the location of the decoupler determines the capillary length available for the CE separation, the effective capillary length for each configuration was 1.5 cm less than the total length. The exact separation and detection conditions for each individual CEEC configuration are described below. Tube-Wire Configuration. This approach is a modification of a previously reported design in which gold rather than platinum electrodes were employed.17 A diagram of the tube-wire configuration is shown in Figure 2A. The platinum wire and tube (Goodfellow Corp., Berwyn, PA) were pretreated using a protocol described previously.26 Briefly, they were placed in a solution of 50:50 deionized water/microcleaning solution (International Products Corp., Burlington, NJ) and sonicated for 15 min, after which they were rinsed several times with deionized water. They were then placed in a 1 N solution of sodium hydroxide and sonicated for an additional 10 min. Last, they were again rinsed several times with deionized water and allowed to air-dry. The generating electrode was fabricated from a 2-mm-long, 100 µm i.d. × 400 µm o.d. platinum tube. A 50-cm-long fusedsilica capillary (40 µm i.d. × 100 µm o.d.) was employed as the separation capillary (Polymicro Technologies, Phoenix, AZ). A 1-mm section of polyimide was removed from the end of the capillary, and the exposed section of capillary was placed inside the tube and sealed with epoxy (Miller-Stephenson Chemical Co., Danbury, CT). The section of the capillary that was still coated with polyimide would not fit inside of the tube; this defined the length of tube that was accessible to the run buffer as 1 mm. Electrical contact was made to the generating electrode by attaching a wire to the outer surface of the tube using silver epoxy (Ted Pella, Redding, CA). After the epoxy cured, the outer surface was coated with insulating epoxy (Miller-Stephenson). The 25µm-o.d. platinum wire detection electrode was fabricated in a manner identical to that used for the gold electrodes previously described.12 (25) Linhares, M. C.; Kissinger, P. T. Anal. Chem. 1991, 63, 2076-2078. (26) Chetwyn, N. P. Rapid Screening of Enzyme Inhibitors Using Capillary Electrophoresis with Enzyme-based Biosensors. Ph.D. Dissertation, University of Kansas, 1998; p 38.

Figure 2. Dual-electrode configurations investigated for postcolumn amperometric reaction detection: (A) tube-wire; (B) wire-ring; (C) wire-wire on-capillary.

The 25-µm-o.d. platinum wire was attached with silver epoxy to a larger copper wire (24 gauge). The silver epoxy was allowed to cure for 1 h. The copper/platinum wires were then placed inside a glass capillary that had been pulled to an outer diameter of ∼100 µm, such that 1.0 mm of the platinum wire protruded from the end of the glass. The wire was fixed in place with UV glue (UVEXS) by filling the tip of the glass capillary with glue using a syringe. During this step it was essential that the platinum did not become contaminated with the UV glue. The glue was allowed to cure under a long-pass UV lamp for 15 min; the opposite end of the capillary was then sealed with UV glue, and the entire electrode was placed under the UV lamp for 1 h. The electrochemical cell was housed in a rubber septum, and the generating electrode attached to the end of the detection capillary was sealed in this septum with hot glue. The detection electrode was placed in a capillary holder built in-house and positioned inside the tubular generating electrode with the aid of micromanipulators and a light microscope. Prior to each run, the generation and detection electrodes were treated by application of a bipolar square wave with an amplitude of (1.0 V, a frequency of 50 Hz, and a duration of 40 s while the capillary was flushed with 0.1 N sulfuric acid. The capillary was then rinsed with background electrolyte consisting of 25 mM sodium phosphate buffer, pH 7.5, and the cathodic reservoir was filled with 10 mM potassium bromide in background electrolyte. Separations were performed at 13.5 kV. Wire-Ring Configuration. This configuration is shown in Figure 2B. The platinum used for the wire-ring configuration

was pretreated as described for the tube-wire system. The generating electrode was fabricated in a manner similar to that for the detection electrode described above with the exception that the glass capillary was specifically drawn to a diameter between 30 and 45 µm at one end. Approximately 1.0 mm of the platinum wire protruded from the end of the glass, and a very small portion of the electrode close to the glass was insulated with epoxy. The 50-µm-i.d. platinum ring detection electrode was produced by wrapping the middle section of a ∼2.1-cm piece of 25-µm platinum wire around a 50-µm-i.d. wire with the aid of a light microscope and fine-tipped forceps. The portions of wire on either side that were not used to make the ring were then twisted together and bent at a 90° angle. The 50-µm-i.d. wire was then removed and the wire-ring electrode was produced by threading the smaller diameter platinum wire detection electrode through the ring with the aid of a light microscope and fine-tipped forceps. The ring was pulled over the insulated section of the detection electrode and was held in place by fixing the nonring portion of the electrode to the outer glass surface with epoxy. Electrical connection was accomplished via an aluminum wire attached to this section using silver epoxy. Finally, all electrical connections were insulated so that only the platinum ring and platinum wire were exposed. The electrochemical cell and pretreatment protocols were identical to those used for the tube-wire configuration. The separation capillary consisted of a 50-cm-long fused-silica capillary with an inner diameter of 75 µm and an outer diameter of 360 µm. The background electrolyte was composed of 20 mM sodium phosphate, pH 6.5. The cathodic reservoir was filled with 10 mM potassium bromide dissolved in background electrolyte. Separations were performed at 13.5 kV. On-Capillary Wire-Wire Configuration. The platinum used for the wire-wire system was pretreated as described for the wire-tube system. The generating electrode was fabricated from a 2-cm-long section of 25-µm platinum wire. The wire was bent with forceps using a light microscope so that 1 mm of the wire could be inserted inside the separation capillary. The other end of the platinum wire was fixed to the outer surface of the capillary with epoxy, as shown in Figure 2C. A larger aluminum wire was connected to the other end of the platinum wire with silver epoxy. This was used for electrical connection to the bipotentiostat. The detection electrode was fabricated from a 2-cm-long section of 10-µm-diameter wire that was bent to fit snugly over the end of the fused-silica capillary. The wire was placed on the capillary perpendicular to the generating electrode so that it did not make electrical contact and was affixed to the outer edges of the fusedsilica capillary with epoxy. After the epoxy cured, one of the ends of the detection electrode was attached to a larger aluminum wire with silver epoxy and connected to the bipotentiostat. All electrical connections outside the capillary were insulated with glue (ITW Brands, Wood Dale, IL). The active electrode length was defined by the internal diameter of the capillary (75 µm). The electrochemical cell and pretreatment protocol used were identical to those used for the tube-wire configuration. The separation capillary consisted of a 35-cm-long fused-silica capillary with an inner diameter of 75 µm and an outer diameter of 360 µm. The background electrolyte was 50 mM phosphate, pH 6.0. Analytical Chemistry, Vol. 71, No. 2, January 15, 1999

409

The cathodic reservoir was filled with 25 mM potassium bromide in background electrolyte. Separations were performed at 10 kV. RESULTS AND DISCUSSION With postcolumn reaction detection, the analyte of interest reacts with an electroactive reagent that is produced at the generating electrode and the concentration of analyte is determined by monitoring loss of this reagent at the detection electrode. The general detection strategy was discussed at the start of this paper and is illustrated in Figure 1. To demonstrate the use of this detection mode for CE, bromine was used as the electroactive reagent and its applicability for the determination of thiols and thioethers was explored. Several elements must be taken into account when one develops a reaction detection method for capillary electrophoresis. First, the inherently small volume of the separation capillaries used in CE requires that the detector be placed in-line to minimize band broadening. Second, to achieve good sensitivity, a dual-electrode configuration with good generating capacity and collection efficiency must be employed. Third, the components of the detection cell must exhibit good chemical stability to the bromine that is generated. Along with the detector requirements listed above, another important factor that must be considered is the way in which bromide ion is introduced into the system. High concentrations of bromide salts in the background electrolyte can lead to large separation currents, which can be detrimental to the CE separation efficiency. Bromide could not be introduced into the system via the anodic reservoir because it was converted at the anode to bromine, which reacted with the analyte prior to detection and produced increased noise at the detector. In addition, the noxious properties of bromine make any unnecessary production of this compound unappealing. Fortunately, with this CEEC system, bromide could be successfully introduced at the cathodic reservoir through the fracture joint used to decouple the detector from the separation capillary. Using this approach, the bromide was introduced into the capillary and transported to the generating electrode by the pressure introduced by the electroosmotic flow generated in the separation capillary, where it was oxidized to bromine. Under these circumstances, the separation current was unaffected by the addition of the bromide salt as long as the capillary was rinsed with fresh buffer between runs. This led to a significant reduction in baseline noise as well as elimination of any superfluous bromine generation. All of the postcolumn reaction systems were evaluated for the detection of thiols. Thiols have been demonstrated to be excellent candidates for this detection method.19,21,22 Many biologically important thiols, such as cysteine and glutathione, are charged and, therefore, amenable to separation by CE. Cysteine is an essential amino acid involved in protein synthesis, and glutathione plays an important role in the detoxification of active oxygen species and electrophilic xenobiotics.27-29 One of the attractive features of postcolumn amperometric reaction detection for drug metabolism studies is that it is possible to monitor both glutathione (27) Meister, A.; Anderson, M. E. Annu. Rev. Biochem. 1983, 52, 711-760. (28) Kaplowitz, N.; Aw, T. Y.; Ookhtens, M. Annu. Rev. Pharmacol. Toxicol. 1985, 25, 715-744. (29) Lawler, J. M.; Powers, S. K. Can. J. Appl. Physiol. 1998, 23, 23-55.

410

Analytical Chemistry, Vol. 71, No. 2, January 15, 1999

and the thioether conjugates of a drug in the same sample. This has been previously demonstrated for acetaminophen using LCEC.21 The unique demands of postcolumn reaction detection with capillary electrophoresis led to the development and evaluation of several different dual-electrode configurations. Initially, the tube-wire configuration was considered. This configuration has been successfully employed for the detection of thiols and disulfides by CEEC.17 Unfortunately, it was found that when the tube-wire electrode was constructed as previously reported, the bromine generated at the tube reacted with the insulating epoxy, slowly destroying the detection cell. To improve the ruggedness of the cell with regard to the electrogenerated bromine, a platinum tube with almost the same outer diameter as the separation capillary was employed. This made it possible to place the capillary deep inside the platinum tube and to apply epoxy externally. Due to the smaller inner diameter of the platinum tube, a 25-µm platinum wire was used as the detection electrode. Although a detectable signal was obtained for cysteine and glutathione using the modified tube-wire configuration (Figure 2A), the detection limits were extremely high. This is primarily due to the combination of a very small inner diameter separation capillary, a relatively large inner diameter generating electrode, and a small detection electrode. These constraints were placed on the system due to the limited availability of platinum tubes of larger diameter that could be fitted tightly onto the CE capillary without a gap that required epoxy. It was demonstrated previously that a large mismatch between the tube and wire in this configuration leads to a large dead volume in the detector cell, resulting in peak broadening and loss of analyte signal.17 Larger detection electrodes were tried but were difficult to insert without making electrical contact with the tube. To improve the LOD for thiols, a second dual-electrode design was investigated (Figure 2B). In this design, the generating electrode consisted of a 25-µm platinum wire placed inside the 75-µm capillary. It has already been shown that this configuration shows good coulometric efficiency under CE separation conditions in the single-electrode mode.1,5 The 1-mm-long wire also acts as an in-capillary mixer because it interrupts the flow path of the run buffer out of the capillary. Therefore, the analyte is mixed with the bromine as it is generated along the length of the electrode in a volume that is smaller than that of the capillary itself. The detection electrode (ring) was positioned directly outside of the entrance to the capillary to detect the unreacted bromine before it diffused away. The detection limit for cysteine was estimated to be 600 nM (S/N ) 3) using this configuration. Unfortunately, this design also suffered from the reactivity of bromine with the epoxy used to hold the ring electrode in place. The electrode lifetime was ∼1 day with continuous use. The lack of ruggedness of the two previous designs led to the investigation of a third and final dual-electrode design shown in Figure 2C. This design takes advantage of the positive attributes of placing the generating electrode inside the capillary described above. The detection electrode is similar to the integrated oncapillary electrode previously described by our group.30 This arrangement is considerably more robust than the two previous designs because none of the insulating epoxy is located directly (30) Zhong, M.; Lunte, S. M. Anal. Chem. 1996, 68, 2488-2493.

Figure 3. Electropherogram of cysteine, reduced glutathione, and N-acetylcysteine, each at a concentration of 5 µM, obtained with the wire-wire on-capillary dual-electrode reaction detector. The upper line was obtained with the generating electrode on, while the lower line was obtained with the generating electrode off. See text for separation conditions.

in line with the orifice of the fused-silica capillary. It has been utilized successfully for several weeks, including several 24-h periods, without any signs of degradation. The linearity for cysteine was determined over the range of 0.5-20 µΜ (n ) 4). The method was found to be linear with a correlation coefficient of 0.994 and a slope of 0.1 nA/µΜ. The relative standard deviation of the migration time within day was 2.2% (n ) 7) and day to day was 6.1% (3 days). The reproducibility for three injections of cysteine varied from 8% for 20 µM to 19% for 2 µM (n ) 3). The LOD for cysteine at a S/N ) 3 was estimated to be 80 nM. In postcolumn amperometric reaction detection, the concentration of the analyte is determined based on the amount of bromine depletion by the analyte. In the wire-wire configuration, this reaction occurs in the last millimeter of the detection capillary. Assuming a linear velocity of ∼0.08 cm/s, this yields a reaction time of ∼1.25 s. This was previously shown to be a sufficient amount of time for the reaction of thiols with bromine.21 To confirm that the signal obtained at the detection electrode is due solely to the depletion of bromine and not the reduction of some other species, an electropherogram was obtained with the generating electrode off. The result can be seen in Figure 3. The solid line shows the response obtained for a sample containing cysteine, reduced glutathione, and N-acetylcysteine under these conditions. It can be seen that there was no detectable signal for these thiols, confirming that the signal obtained with the generating electrode on (also seen in Figure 3) is due only to the reaction of the electrogenerated bromine with the thiols. Postcolumn amperometric reaction detection is an indirect detection method. Therefore, both the linear range and limit of detection are dependent on the amount of bromine generated in the system. This is contingent on the concentration of bromide in the cathodic buffer reservoir and the conversion efficiency of the generating electrode. In cases where a low LOD is required, (31) Albery, W. J.; Svanberg, L. R.; Wood, P. J. Electroanal. Chem. 1984, 162, 45-53. (32) Mank, A. J. G.; Lingeman. H.; Gooijer, C. J. High Resolut. Chromatogr. 1994, 17, 797-798. (33) Hogan, B. L.; Yeung, E. S. Anal. Chem. 1992, 64, 2841-2845. (34) Ling, B. L.; Baeyens, W. R. G. Anal. Chim. Acta 1991, 255, 283-288.

Figure 4. Electropherogram of 10 µΜ methionine obtained with the wire-wire on-capillary dual-electrode reaction detector.

generation of only a small amount of bromine is necessary. However, this will also lead to a small dynamic range. Conversely, the production of a large quantity of bromine will lead to a larger dynamic range, but the LOD will be higher. In LCEC applications of this technique, the optimal concentration of bromine that should be generated has been determined both empirically20 and by using a feedback method to continuously adjust the amount of bromine generated according to the signal response.31 These strategies could also be implemented for the CE system. It is clear from these results that the wire-wire on-capillary dual-electrode system is best suited to bromine-based postcolumn reaction detection. The concentration detection limits obtained for cysteine are comparable to those reported for LCEC using postcolumn amperometric reaction detection21-23 and competitive with those of other CE-based methods for the determination of thiols.8-12,14,32-34 The other electrode configurations described here were found not to be applicable to postcolumn reaction detection, but may be useful for other dual-electrode applications that do not involve generation of such a highly reactive reagent. One important attribute of postcolumn amperometric reaction detection is its versatility. Any compound that reacts with bromine will show a response. This makes it possible to detect compounds that are difficult to determine by other means. One class of compounds that fits this description is aliphatic thioethers. To demonstrate the approach described here for thioethers, it was evaluated for the detection of methionine. An electropherogram obtained for a 10 µM sample of methionine is shown in Figure 4. CONCLUSIONS Postcolumn amperometric reaction detection has been successfully employed with capillary electrophoresis for the detection of thiols and thioethers. Three different electrode configurations were evaluated. Two of these suffered from a lack of ruggedness due to the high reactivity of the bromine generated in the reaction system. The wire-wire on-capillary dual-electrode configuration, which was the most stable and provided the best limits of detection, appears to be the best cell design for this application. The fact that the signal is due to bromine depletion was confirmed by the lack of a response when the generating electrode was disconnected from the system. Future directions will focus on the detection of additional classes of compounds using this technique and evaluation of the two remaining detector designs for other dual-electrode applications. Analytical Chemistry, Vol. 71, No. 2, January 15, 1999

411

ACKNOWLEDGMENT Funding for this project and financial support for L.A.H. from the National Cancer Institute (Grant 2-T32-CA09242-20) and National Science Foundation (Grant CHE-9702631) are gratefully acknowledged. The donation of the bipotentiostat and DA-5 analogto-digital converter from Bioanalytical Systems is also acknowledged. The authors thank Nancy Harmony for assistance in

412

Analytical Chemistry, Vol. 71, No. 2, January 15, 1999

preparing the manuscript and Dr. Min Zhong for helpful discussions.

Received for review August 18, 1998. Accepted November 2, 1998. AC980926D