Cyclic Voltammetric Detection in Capillary Electrophoresis with

Fast cyclic voltammetry (CV) was evaluated over sweep rates of 20−1000 V/s at Au disk electrodes (25 and 10 μm) for end-capillary detection in capi...
4 downloads 0 Views 116KB Size
Download PDF
Anal. Chem. 1998, 70, 2504-2509

Cyclic Voltammetric Detection in Capillary Electrophoresis with Application to Metal Ions Jenny Wen, Andrzej Baranski, and Richard Cassidy*

Chemistry Department, University of Saskatchewan, Saskatoon, SK S7N 5C9, Canada

Fast cyclic voltammetry (CV) was evaluated over sweep rates of 20-1000 V/s at Au disk electrodes (25 and 10 µm) for end-capillary detection in capillary electrophoresis with metal ions as test analytes; some studies were also done with 25-µm Pt disk electrodes. The waveform applied to the electrode consisted of a preconcentration period (55-330 ms) followed by cyclic voltammetry (2100 ms). Maximum signal-to-noise was obtained with the integrated CV current as the analytical signal, and this was linearly proportional to sweep rate; maximum response was obtained at sweep rates of >100 V/s for 10-µm electrodes and >200 V/s for 25-µm electrodes; sweep rates of >400 V/s caused peak tailing due to trapping of the analyte at the electrode. With this CV detection approach, comigrating analytes could be identified and determined. Reproducibilities for six analytes over the range 1.0 × 10-7-1.0 × 10-5 mol/L were 2%-5%, and calibration curves were linear, with response factors in the range of 2%-6%. Detection limits (2 × peak-to-peak baseline noise) were in the range of 5 × 10-9-4 × 10-8 mol/L, which are 1-2 orders of magnitude better than results obtained previously with square-wave pulsed amperometric detection of metal ions. The successful application of capillary electrophoresis (CE) relies on the ability to detect analytes at low concentrations in small sample volumes. Among the various CE detection approaches, electrochemical detection is promising for detection of inorganic,1,2 organic,3,4 and biological compounds.5,6 Of the different electrochemical detection techniques, amperometric detection is the most easily and commonly implemented method. Constant voltage detection is feasible for some electroactive analytes,7,8 but electrode fouling from sample matrixes and/or analyte reaction products limits its practical application for many * Corresponding author. Fax: 306-9664730. E-mail: [email protected]. (1) Ewing, A. G.; Mesaros, J. M.; Gavin, P. F. Anal. Chem. 1994, 66, 527A537A. (2) Matysik, F. M.; Meister, A.; Werner, G. Anal. Chim. Acta 1995, 305, 114120. (3) Lu, W.; Cassidy, R. M. Anal. Chem. 1993, 65, 2878-2881. (4) Gaitonde, C. D.; Pathak, P. V. J. Chromatogr. 1990, 514, 389-393. (5) Chen, J. G.; Weber, S. G. Anal. Chem. 1995, 67, 3596-3604. (6) (a) Deacon, M.; Oshea, T. J.; Lunte, S. M.; Smyth, M. R. J. Chromatogr. A 1993, 652, 377-383. (b) Kissinger, P. T. J. Pharm. Biomed. Anal. 1996, 14, 871-880. (7) Wallingford, R. A.; Ewing, A. G. Anal. Chem. 1987, 59, 1762. (8) Gaitonde, C. D.; Pathak, P. V. J. Chromatogr. 1990, 514, 389.

2504 Analytical Chemistry, Vol. 70, No. 13, July 1, 1998

analytes.9-11 Square-wave pulsed voltage detection can overcome electrode fouling problems and provide consistent electrode response for many species,12-16 and detection limits (2 × peakto-peak noise) are generally in the range of 10-6-10-7 mol/L;9-16 the use of special data treatments and/or different potential wave forms can give slightly better detection limits (10-7-10-8 mol/L range).17 With recent developments in fast personal computers and simplified A/D interfacing technology, it is possible that other rapid scanning techniques may provide improvements in CE electrochemical detection. Stripping voltammetry of preadsorbed analytes has always been regarded as one of the most sensitive electrochemical techniques,18-20 giving detection limits down to 10-9-10-11 mol/ L.21-23 Most early applications were limited to slow scan rates21-26 (less than 10 V/s) to avoid broadening of stripping peaks, which could result in poor discrimination of the analytes; these procedures usually involved the use of a hanging mercury drop or a mercury film electrode.27-29 The use of high-frequency potential waveforms with microelectrodes for repetitive fast-scan voltammetric analysis can eliminate most of the limitations of conventional stripping techniques for the detection of analytes in CE, but only a few studies have been reported with fast-scan voltammetry in CE. Fast cyclic voltammetry (CV) has been reported (9) Austin, D. S.; Polta, J. A. J. Electroanal. Chem. 1984, 168, 227-237. (10) Moring, S. E. Capillary Electrophoresis: Theory and Practice. In Quantitative Aspects of Capillary Electrophoresis Analysis; Grossman, P. D., Colburn, J. C., Eds.; Academic Press: San Diego, CA, 1992. (11) Cahill, P. S.; Wightman, R. M. Anal. Chem. 1995, 67, 2599-2605. (12) Lacourse, W. R.; Owens, G. S. Electrophoresis 1996, 17, 310-318. (13) Lu, W.; Cassidy, R. M. Anal. Chem. 1993, 65, 2878-2881. (14) O’Shea, T. J.; Lunte, S. M.; LaCourse, W. R. Anal. Chem. 1993, 65, 948951. (15) Weber, P. L.; Lunte, S. M. Electrophoresis 1996, 17, 302-309. (16) Wen, J.; Cassidy, R. M. Anal. Chem. 1996, 68, 1047-1053. (17) Wen, J.; Cassidy, R. M.; Baranski, A. S. J. Chromatogr. A, in press. (18) Wang, J.; Hutchins-Kumar, L. D. Anal. Chem. 1986, 58, 402-407. (19) Wang, J.; Taha, Z. Talanta 1991, 38, 489-492. (20) Wang, J.; Varughese, K. Anal. Chim. Acta 1987, 199, 185-189. (21) Baranski, A. S. Anal. Chem. 1987, 59, 662-666. (22) Wu, H. P. Anal.Chem. 1996, 68, 1639-1645. (23) Bret, C. M. A.; Brett, A. M. O.; Tugulea L. Anal. Chim. Acta 1996, 322, 151-157. (24) Deuries, W. T. J. Electroanal. Chem. 1965, 9, 448-452. (25) Wong, D. K. Y.; Ewing, A. G. Anal. Chem. 1990, 62, 2697-2702. (26) Vehmeyer, K. R.; Wightman, R. M. Anal. Chem. 1985, 57, 1989-1993. (27) Nomura, S.; Nozaki, K.; Okazaki, S. Electroanalysis 1991, 3, 617-625. (28) Barbeira, P. J. S.; Mazo, L. H.; Stradiotto, N. R. Analyst 1995, 120, 16471650. (29) Lukaszewski, Z.; Zembrzuski, W.; Piela, A. Anal. Chim. Acta 1996, 318, 159-165. S0003-2700(97)01224-9 CCC: $15.00

© 1998 American Chemical Society Published on Web 06/02/1998

for the detection of histamine and 5-hydroxytryptamine in HPLC,30 and it has been suggested that fast voltammetry should be able to resolve comigrating species in CE if their electrochemical potentials are sufficiently different.31 Repetitive voltammetric scanning has also been recently reported for CE,32,33 but scan rates were relatively low and resistance problems associated with fibers being inserted into capillaries were apparent.33 More recently, a study of high-frequency square-wave techniques used on-line CV with disk electrodes and end-capillary detection to characterize the electrochemical behavior of analytes.17,34,35 The results of these studies suggested that fast on-line CV may also offer advantages for electrochemical detection in flow injection and CE. These advantages could include peak purity evaluation, characterization of electrochemical behavior of the background electrolyte and the analyte under CE conditions, resolution of comigrating analytes, use of a preadsorption voltage in the waveform to permit analyte preconcentration via physical, chemical, or electrochemical adsorption, and selection of optimum voltage ranges for integration of the anodic and/or cathodic currents in the cyclic voltammograms. In this work, the ability of fast cyclic voltammetry to detect a series of test ions and to identify comigrating species in CE was studied over sweep rates of 25-1000 V/s. Both CV current and CV charge were used as analytical signals, and the dependence of the analytical signals and the peak shapes on sweep rate was evaluated. Other parameters, such as applied potential, size of working electrode, and preconcentration time, were also examined. EXPERIMENTAL SECTION Instrumentation and Apparatus. A separation voltage of 20 kV over a 25-µm-i.d. × 350-µm-o.d. fused silica capillary (Polymicro Technology Inc., Phoenix, AZ) was used for the separation of metal ions. Before use, new capillaries were washed with water, 0.1 mol/L HCl, and the operating electrolyte (Rhydroxyisobutyric acid and creatinine) as described previously26 to obtain reproducible electroosmotic flow. For storage, a flow of electrolyte was maintained in the capillaries via a height differential of approximately 10 cm. The input of a 30-kV power supply (Spellman, High Voltage Electronics Corp., model RHR30PN30, Plainview, NY) with reversible polarity was placed in a Plexiglas box with an interlock switch on the access door. Standard 0.8-mL polyethylene sample vials (Cole-Parmer Instrument Co., Niles, IL) were used as containers for the carrier electrolyte and for the standards and samples. Electrochemical detection was performed with a three-electrode system placed inside a faradaic cage to minimize interference from environmental noise. The working electrode and capillary were aligned to within 20 ( 5 µm with a micropositioner. Pt wire (surface area about 0.5 mm2) was used as counter electrode, a saturated KCl calomel electrode (miniature model, Fisher Scientific Co., Ottawa, ON, (30) Pihel, K.; Hsieh, S.; Jorgenson, J. W.; Wightman, R. M. Anal. Chem. 1995, 67, 4514-4521. (31) Ferris, S. S.; Lou, G.; Ewing. A. G. J. Microcolumn Sep. 1994, 6, 263-268. (32) Swanek, F. D.; Chen, G.; Ewing, A. G. Anal. Chem. 1996, 68, 3912-3916. (33) Park, S.; McGrath, M. J.; Smyth, M. R.; Diamond, D.; Lunte, C. E. Anal. Chem. 1997, 69, 2994-3001. (34) Baranski, A. S.; Norouzi, P.; Nelson, L. J. Proc. Electrochem. Soc. 1996, 96-9, 41-52. (35) Baranski, A. S.; Norouzi, P. Can. J. Chem. 1997, 75, 1736-1749.

Canada) was used as reference electrode, and the 25-µm- and 10µm-i.d. Au and Pt disk electrodes were manufactured and preconditioned as described elsewhere.16 Detection was controlled with a Pentium/16.0 MB RAM IBM personal computer equipped with a PCL-818 high-performance data acquisition card (B & C Microsystems Inc., Sunnyvale, CA). The scan rates used were lower than the 100-kHz rate of the PCL board (normally 40-70 kHz) because noise increased significantly above 75 kHz. The number of points defining the CV curve was generally about 256 points, and, depending on the sampling rate, at least two A/D conversions were averaged at each point. Details on data collection are given elsewhere.17,34,35 Sampling and Detection Limits. Samples were introduced into the capillary by 5-kV electrokinetic injection for 10 s; under these sampling conditions, the amount of sample injected for the fastest ion, Tl+, was 3.2 × 10-17 mol, and that for the slowest ion, Pb2+, was 1.3 × 10-17 mol for a sample concentration of 1.0 × 10-7 mol/L, and the sample volume injected into the capillary was evaluated by electroosmotic flow to be approximately 6.1 × 10-4 µL. Detection limits were obtained by injecting analyte at a concentration of 1.0 × 10-8-1.0 × 10-7 mol/L. Detection limits are defined as twice peak-to-peak noise, and this is equivalent to a S/N of about 10 when noise is defined as one standard deviation. Except where noted, all comparisons of detection limits are made relative to PAD procedures used previously for metal ions.16 Reagents. All chemicals were of analytical grade and were used without further purification. All solutions were prepared from doubly distilled deionized water (Corning, Mega-Pure system, MP-6A&D2, Corning, NY). The background electrolytes for separation and detection of the metal ions were 0.030 mol/L creatinine (Sigma, St. Louis, MO) and 0.008 mol/L R-hydroxyisobutyric acid (HIBA) (98% Aldrich, Milwaukee, WI), and the pH value was adjusted with acetic acid to 4.8 unless otherwise noted. The electrolyte in the separation reservoir was replaced daily to avoid chemical and pH changes. Reagent-grade metal salts used were thallium nitrate and nickel nitrate (Fisher Scientific Co., Fair Lawn, NJ, lead nitrate (Anachemia, Montreal, PQ, Canada), cadmium nitrate and zinc nitrate (Merck & Co. Inc., Rahway, NJ), and cobalt acetate (General Chemical Division, Allied Chemical & Dye Corp., New York, NY). Metal ion stock solutions (0.01 mol/L) were diluted to the desired concentration with operating electrolyte prior to use. All CE solutions were filtered through a 0.2-µm Nylon-66 membrane syringe filter (Cole-Parmer Instrument Co.). Waveform. The waveform consisted of an initial constant voltage followed by a triangular CV wave. The constant potential, E1, was normally in the range of -700- -1200 mV; this potential was held for periods of 55-330 ms, depending on the standard potentials of metal ions. CV wave portion consisted of a linear scan from the initial potential E1 to a positive vertex potential, E2, in a positive potential direction and then back to E1; the time required for a complete CV scan depended on the scan rate. Scan rates were normally in the range of 100-300V/s, and a complete CV scan took 5-20 ms. The total time for a complete waveform was the sum of the time at E1 and the time required for the CV. Oxidation reactions were designated as positive currents. Two types of signals were used for the analytical response. One was the maximum current in each voltammogram, referred to as the Analytical Chemistry, Vol. 70, No. 13, July 1, 1998

2505

Figure 1. Influence of preconcentration time on the CV charge at CE peak maximum at a 25-µm Au electrode. Experimental conditions: CV potential, -700 to 100 mV; sweep rate, 350 V/s; analyte concentration, 1.0 µmol/L; 5 kV electromigration injection for 10 s; electrolyte, 0.030 mol/L creatinine and 0.008 mol/L HIBA at pH 4.8; separation voltage, 20 kV over 25-µm × 60-cm capillary.

CV current response. The second was the integrated current over selected portions of the CV, with units of coulombs; this response will be referred to as the CV charge response. The two types of responses were plotted vs time after correction for the response from the background electrolyte, obtained during the first 7-10 scans in the CE separation. Since CV charge gave better S/N, this is the response used for most of the data discussed in this paper. RESULTS AND DISCUSSION Preconcentration. The high sensitivity obtained in stripping voltammetric techniques is mainly attributed to analyte preconcentration via either physical, chemical, or electrochemical adsorption. A series of metal ions was used for this evaluation since they offer a wide range of electrochemical behavior.17 To optimize the performance of fast CV detection in CE, the effect of preconcentration time on electrode response and peak shape was evaluated for Tl+, Co2+, Ni2+, Zn2+ Cd2+, and Pb2+ over the range of 2.0 × 10-7-1.0 × 10-4 mol/L at 25- and 10 -µm Au electrodes. A preconcentration range of 55-330 ms was examined, and the sweep rate range was 25-1000 V/s. The importance of preconcentration time is demonstrated by the results shown in Figure 1 for Tl+, Cd2+, and Pb2+, obtained at a sweep rate of 350 V/s with a 25-µm Au electrode. Results for Zn2+, Ni2+, and Co2+ were similar to that for Cd2+. Figure 1 shows that electrode response increased with preconcentration time. The rate of increase, Tl+ > Pb2+ . Cd2+, appeared to be related to the relative order of the reversibility of the overall electrode reactions, as determined previously in on-line CV studies of the analytes.17 Maximum preconcentration time was limited by the need to plot 10-15 data points over the narrow analyte peaks. In addition, peak broadening was noted when preconcentration time was g250 ms, likely due to analyte carryover from one waveform to the next because of incomplete removal of the analyte adsorbed onto the 2506 Analytical Chemistry, Vol. 70, No. 13, July 1, 1998

electrode (discussed below). Long preconcentration times also reduced the number of data points for plotting the electropherogram; thus, preconcentration times of 110-220 ms were selected as a compromise between maximum S/N and minimum peak broadening. For most of the results reported in this paper, the potential waveform normally consisted of two parts: an initial constant potential (preconcentration period) time of 220 ms, followed by a triangular potential over a time period determined by the scan rate. If analytes cannot be preconcentrated, then some loss in S/N would be expected. However, it is also possible to make use of the fact that many analytes will adsorb at electrodes at suitably applied potentials, and one of the advantages of scanning approaches for electrochemical detection is that different parts of the waveform can be selected for a variety of purposes, such as electrode cleaning, electrode conditioning, and analyte preconcentration. Recent studies in our laboratories have show that the response for analytes such as dopamine can be enhanced by preconcentration with appropriate waveform selection. Sweep Rate. In fast voltammetric analysis, the sweep rate is an important factor since analyte signal, background noise, and peak shape rely on sweep rate. Depending on experimental conditions, CV current can be proportional to ν1/2 (in diffusioncontrolled conditions)36 or directly proportional to ν (at thin-layer cells37 or Hg film electrodes).22 Little information has been reported on the dependence of CV current on sweep rate at Au microelectrodes. Theoretically, the CV charge is supposed to be independent of sweep rate, and this has been observed in slowsweep-rate voltammetric techniques.21,22 However, changes in CV charge with sweep rate may occur under fast-scanning conditions due to several factors, such as electrochemical kinetics, mass transport of analytes to and from the electrode/solution interface, the ohmic drop, and the time-dependence of the double layer capacitance caused by slow surface reconstruction processes. For some conditions, carryover of analyte from the previous CV will lead to an increase of electrode response in the next CV, and thus response may increase with sweep rate if the reaction rates of analytes are fast enough to match their mass transport rates. To find the optimal sweep rate for maximum S/N, the dependence of electrode response on sweep rate was examined under CE conditions for Tl+, Co2+, Ni2+, Zn2+, Cd2+, and Pb2+ over 2.0 × 10-7-5.0 × 10-5 mol/L at 25- and 10-µm Au electrodes over sweep rates of 25-1000 V/s. Typical results for CV current at a 25-µm Au electrode for Tl+, Zn2+, Cd2+, and Pb2+ over 25-500 V/s is shown in Figure 2. The CV current increased linearly with sweep rate, with correlation coefficients in the range of 0.9980.999; results for Ni2+ and Co2+ and results at a 10-µm Au electrode were similar. The order of the increase in slope for the different test analytes was the same as that for the reversibility of the electrochemical reaction.17 These linear relationships are similar to those reported for a thin-layer cell37 and a thin Hg film electrode.22 In contrast to the linear dependency observed between CV current and sweep rate (Figure 2), the relationship observed between CV charge and sweep rate depended on the analyte and the size of the electrode. For example, the change (36) Gosser, D. K. Cyclic Voltammetry: simulation and analysis of reaction mechanisms; VCH: New York, 1993. (37) Bard, A. J.; Faulkner, L. R. Electrochemical Methods: Bulk Electrolysis Method; John Wiley & Sons: New York, 1980; Chapter 10.

Figure 2. Dependence of CV current at the CE peak maximum on sweep rate at a 25-µm Au electrode. Experimental conditions: CV potential, -900 to 100 mV; analyte concentration, 1.0 µmol/L; preconcentration time, 220 ms; other conditions as for Figure 1.

Figure 3. Dependence of CV charge at the CE peak maximum on sweep rate at a 25-µm Au electrode. Experimental conditions: CV potential, -900 to 100 mV; analyte concentration, 1.0 µmol/L; preconcentration time, 220 ms; other conditions as for Figure 1.

in CV charge with sweep rate for Tl+, Cd2+, Zn2+, and Pb2+ shown in Figure 3 exhibited a variety of different trends; Ni2+ and Co2+ gave results similar to those for Zn2+ and Cd2+. The different patterns observed in Figure 3 appear to be associated with changes in the relationship between the rate of the charge-transfer step and mass transport rates with changes in sweep rate. For Tl+ and Pb2+, which have faster charge-transfer kinetics (small shifts in voltammograms), analyte carryover began to occur from one waveform to the next as the sweep rate increased, but for the other metal ions, the charge-transfer kinetics were not fast enough to match mass transport rates, and thus analyte carryover was not significant. Another complicating factor in the interpreta-

tion of the results in Figure 3 is the fact that they were obtained for a CE peak rather than a constant bulk concentration. Figure 3 shows that the CV charge of Tl+ and Pb2+ increased with sweep rate up to 200 V/s and then became more or less constant with the sweep rate, likely due to positive shifts observed in the cyclic voltammograms. These positive shifts, which would cause incomplete reaction, are expected from theory38 and have been shown in other experimental studies.39 With a change in the sweep rate from 20 to 400 V/s, the anodic peak potential shifted in a positive direction: 50-60 mV for Tl+, 40-50 mV for Pb2+, and 25-35 mV for Cd2+, Zn2+, Ni2+, and Co2+. Cathodic peak potentials shifted to slightly more negative values. Thus, an increase in sweep rate caused a decrease in CV charge due to the incomplete electrochemical reaction from potential shifts. Maximum S/N was achieved at sweep rates of 150-350 V/s for analytes with fast charge-transfer kinetics (2-3 times enhancement in S/N relative to low sweep rate), and for analytes with slower charge-transfer kinetics, sweep rate affected S/N only slightly. Trends similar to those shown in Figures 2 and 3 were also observed at different concentration levels, with a number of different Au microelectrodes, and also with some Pt microelectrodes. Since the ohmic drop and mass transport are influenced by electrode size, detection performance at a 10-µm Au electrode was also examined with Tl+, Co2+, Ni2+, Zn2+, Cd2+, and Pb2+ over a sweep rate of 25-900 V/s. The results showed that the value of CV charge at the 10-µm electrode was smaller (relative to that at the 25-µm electrode) by a factor of 4-5 (electrode area decreased by a factor of ∼6). The dependence of CV currents on sweep rate gave results similar to those obtained at a 25-µm electrode, and the CV charge signal for Tl+ and Pb2+ rose and dropped more quickly than that obtained at a 25-µm Au electrode. These differences in behavior for 10-µm electrodes appear to be primarily related to faster mass-transport rates, which leads to less analyte carryover from one waveform to the next. Although the background noise at the 10-µm electrode was 3-4-fold smaller than that at the 25-µm electrode, S/N was not improved. Consequently, detection at a 10-µm electrode did not provide better performance and was not examined further in these studies. The above results showed that some metals were not completely oxidized at high sweep rates. Since higher positive potentials should accelerate oxidation, the performance of CV detection was evaluated at a higher vertex potential of 400 mV. The CV results showed that, with the higher positive potential, Tl+ and Pb2+ were oxidized completely over CV sweep rates of 25-450 V/s. However, Cd2+, Zn2+, Ni2+, and Co2+ still could not be completely oxidized at 400 mV, due to their slow chargetransfer kinetics. Since the reaction of analytes at 400 mV was more complete, the CV charge of analytes should increase (relative to Figure 3), and a typical result for Tl+, Cd2+, and Pb2+ is shown in Figure 4. It is observed in Figure 4 that the CV charge of Tl+ and Pb2+ increased with sweep rate over the range of 20-450 V/s; for Cd2+, peak CV charge increased to a plateau as sweep rate was adjusted to 50 V/s and then reduced gradually at the sweep rate of >250 V/s. Zn2+, Ni2+, and Co2+ gave results similar to those for Cd2+. Unfortunately, the application of higher positive (38) Roe, D. K.; Toni, J. E. A. Anal. Chem. 1965, 37, 1503-1506. (39) Forster, R. J. Chem. Soc. Rev. 1994, 289-297.

Analytical Chemistry, Vol. 70, No. 13, July 1, 1998

2507

Figure 4. Dependence of CV charge on sweep rate at a 10-µm Au electrode. Experimental conditions: CV potential, -800 to 400 mV; analyte concentration, 1.0 µmol/L; preconcentration time, 220 ms; other conditions as for Figure 1.

potentials led to higher background currents and larger peak-topeak noise; these effects are related to larger capacitance currents and the appearance of hydrogen and oxygen reduction reactions because of stripping of the organic electrolyte film off the Au electrode at the higher potentials.17 Consequently, despite higher signals, detection limits obtained at 400 mV were almost the same as those obtained at 200 mV, and more positive CV potentials gave decreasing S/N. The above results show that optimum S/N was obtained at high sweep rates. Unfortunately, these conditions led to incomplete oxidation and carryover of analyte from one waveform to the next, and thus tailing of the peaks in the electropherogram also increased with sweep rate. To evaluate the optimum condition for both S/N and peak resolution, peak shapes for Tl+, Co2+, Ni2+, Zn2+, Cd2+, and Pb2+ were examined in the range of 5.0 × 10-7-1.0 × 10-5 mol/L over sweep rates of 25-1000 V/s. For most analytes, CE peak width at half peak height depended on sweep rate and the maximum applied positive potential. When the positive vertex potential was e200 mV, the increase in peak tailing over the sweep rate 25-400 V/s was Tl+, 100%. When the vertex potential was adjusted to 400 mV, the peak tailing of metal ions, Co2+, Ni2+, Zn2+, Cd2+, and Pb2+, was much smaller (sweep rate 25-450 V/s): Tl+,