Parallel-Opposed Dual-Electrode Detector with Recycling

the on-capillary electrode is parallel-opposed to the disk electrode in a thin-layer geometry. In this region, the redox cycling established between t...
0 downloads 0 Views 84KB Size
Download PDF
Anal. Chem. 1999, 71, 3200-3205

Parallel-Opposed Dual-Electrode Detector with Recycling Amperometric Enhancement for Capillary Electrophoresis Der-chang Chen, Shou-Shu Chang, and Chun-hsien Chen*

Department of Chemistry, National Sun Yat-Sen University, Kaohsiung, Taiwan 80424, R.O.C.

The assembly and characterization of dual-electrode amperometric detection for capillary electrophoresis are described. The detector consists of a disk electrode and an integrated on-capillary electrode fabricated by depositing a gold film onto the end of the separation capillary. The two electrodes are brought together, aligned, and fixed in position using a pair of acrylic plates with a straight groove on one of the plates, the same design as that of a conventional end-column detector. A portion of the on-capillary electrode is parallel-opposed to the disk electrode in a thin-layer geometry. In this region, the redox cycling established between these two electrodes significantly enhances the amperometric signals of electrochemically reversible analytes. For measurements of dopamine in pH 6.9 phosphate electrolyte with a 12.5-µm-i.d. capillary, such a configuration is 10-fold more sensitive than conventional end-column detection. The linear range exceeds 4 orders of magnitude (1.2 mM-50 nM) and the detection limit is 12 nM (4.2 amol, S/N ) 3). Various modes of potential settings for the dual-electrode detection are also discussed. In liquid chromatography (LC) and flow injection analysis, electrochemical (EC) detection with two working electrodes has gained general acceptance as a routine analytical technique.1-3 The two electrodes have been configured in parallel, in series, and facing each other.1,2 In the parallel mode, analytes flow past the two electrodes simultaneously. There are two types of potential settings. One of the electrodes can be set to measure the oxidizable species and the other the reducible species. The other type is to quantify analytes at two different oxidation potentials (or two reduction potentials) by measuring some analytes at one of the electrodes and all analytes at the other. The latter procedure is used to verify peak identity and purity. In the series mode, analytes pass across one electrode and then the other. By properly choosing the potentials, the upstream electrode can either generate electroactive species or screen out the electrochemically irreversible interferants for the downstream electrode. For the * Corresponding author: (e-mail) [email protected]; (fax) +886-75253909. (1) Kissinger, P. T., Heineman, W. R., Eds. Laboratory Techniques in Electroanalytical Chemistry; 2nd ed.; Marcel Dekker: New York, 1996. (2) Roston, D. A.; Shoup, R. E.; Kissinger, P. T. Anal. Chem. 1982, 54, 1417A1434A. (3) Fenn, R. J.; Siggia, S.; Curran, D. J. Anal. Chem. 1978, 50, 1067-1073.

3200 Analytical Chemistry, Vol. 71, No. 15, August 1, 1999

third type of dual-electrode configuration, the electrodes are facing each other in a thin-layer geometry. One electrode is termed the generator and the other the collector because the reaction product of an electrochemically reversible species at the generator electrode becomes the reactant at the collector. Such continuous redox cycling enhances the detection signal. Therefore, dualelectrode detection mode is often superior to single-electrode mode in selectivity and sensitivity in LC-EC. Capillary electrophoresis (CE) is a highly efficient separation technique in which the migration of analytes is under the influence of an extremely high potential field across the separation capillary.4 Although capillaries with small inner diameters have the advantage of small volume requirements for the analysis of volume-limited samples, their use concomitantly demands highly sensitive detection systems. Among the detection systems adapted in CE, amperometric electrochemical detection has the advantages of high sensitivity, good selectivity, and relatively low cost. However, to apply EC detection in CE, in practice it is necessary to optimize two experimental conditions. First, interference of electrophoretic current with the measurements of EC current should be minimized. Second, the alignment of the working electrode relative to the outlet of the separation capillary significantly affects the sensitivity and reproducibility of the EC detection. A design for routine analysis is necessary to facilitate the applications of EC detection in CE. Since the first CE-EC study developed by Wallingford and Ewing,5 methods to address the two issues have been documented in numerous reports5-18 and review articles.19,20 (4) Li, S. F. Y. Capillary Electrophoresis; Elsevier: New York, 1992. (5) Wallingford, R. A.; Ewing, A. G. Anal. Chem. 1987, 59, 1762-1766. (6) Huang, X.; Zare, R. N.; Sloss, S.; Ewing, A. G. Anal. Chem. 1991, 63, 189192. (7) Yik, Y. F.; Lee, H. K.; Li, S. F. Y.; Khoo, S. B. J. Chromatogr. 1991, 585, 139-144. (8) 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. (9) Park, S.; Lunte, C. E. Anal. Chem. 1995, 67, 4366-4370. (10) Chen, I.-C.; Whang, C.-W. J. Chromatogr. 1993, 644, 208-212. (11) Whang, C.-W.; Chen, I.-C. Anal. Chem. 1992, 64, 2461-2464. (12) Kok, W. T.; Sahin, Y. Anal. Chem. 1993, 65, 2497-2501. (13) Hu, S.; Wang, Z.-L.; Li, P.-B.; Cheng, J.-K. Anal. Chem. 1997, 69, 264-267. (14) Chen, M.-C.; Huang, H.-J. Anal. Chem. 1995, 67, 4010-4014. (15) Fermier, A. M.; Gostkowski, M. L.; Colon, L. A. Anal. Chem. 1996, 68, 1661-1664. (16) Zhong, M.; Lunte, S. M. Anal. Chem. 1996, 68, 2488-2493. (17) Voegel, P. D.; Zhou, W. H.; Baldwin, R. P. Anal. Chem. 1997, 69, 951957. (18) Chen, D.-c.; Hsieh, R. M. R.; Chen, C.-h. J. Chin. Chem. Soc. 1998, 45, 257-262. (19) Voegel, P. D.; Baldwin, R. P. Electrophoresis 1997, 18, 2267-2278. 10.1021/ac990069t CCC: $18.00

© 1999 American Chemical Society Published on Web 06/30/1999

Figure 1. Schematic of the dual-electrode assembly. The inset illustrates redox cycling of EC reversible substances flowing past the electrodes.

In the case of coupling CE with dual-electrode detectors, however, thus far there are only a few studies.21-26 The small inner diameters of capillaries make it difficult to position the dualelectrode configurations within a very limited space and to arrange the electrodes along the electrolyte flow pathways. The dualelectrode design developed by Zare et al.21 was constructed with a laser beam to drill a hole on the separation capillary. A gold wire was then inserted into the hole and served as the upstream electrode. The second electrode was a conventional disk electrode and was employed with end-column detection mode6 in which the electrode was placed at the outlet of the separation capillary. The dual-electrode configurations of Lunte22 and Matysik and Backofen23 were similar. An end-column scheme was used to assemble the separation capillary and the dual-electrode detector. Such detectors were prepared by shrouding an insulating layer around the two working electrodes which were configured either in ringdisk or in parallel arrangement. Voegel and Baldwin24 paired two identical capillaries in parallel, each with a metal-film working electrode on its tip and with independent EC detection, such that samples could be injected and detected simultaneously. Recently, Lunte et al.25,26 reported new dual-electrode detectors composed of a tubular electrode and a wire electrode. The former was a hollow gold tube (200 µm i.d.) glued at the outlet of the separation capillary. The wire electrode was then inserted into the tubular electrode and served as the collector electrode. Lunte et al. further extended their design of on-capillary electrodes16 where the working electrode was a metal wire epoxy-glued across the outlet of a separation capillary to a dual-electrode detector by inserting another metal wire, the generating electrode, into the separation capillary.26 We introduce here is a dual-electrode CE-EC detector with high collection and Coulomb efficiencies. A simplified scheme is depicted in Figure 1. One of the electrodes is a conventional diskshaped electrode15,27,28 fabricated by threading a gold wire (90 µm in diameter) through a 1.5-cm-long capillary (100 µm i.d. × 375 (20) Beale, S. C. Anal. Chem. 1998, 70, 279R-300R. (21) Lin, B. L.; Colon, L. A.; Zare, R. N. J. Chromatogr., A 1994, 680, 263-270. (22) Zhong, M.; Zhou, J.; Lunte, S. M.; Zhao, G.; Giolando, D. M.; Kirchhoff, J. R. Anal. Chem. 1996, 68, 203-207. (23) Matysik, F.-M.; Backofen, U. Fresenius J. Anal. Chem. 1996, 356, 169172. (24) Voegel, P. D.; Baldwin, R. P. Electrophoresis 1998, 19, 2226-2232. (25) Zhong, M.; Lunte, S. M. Anal. Chem. 1999, 71, 251-255. (26) Holland, L. A.; Lunte, S. M. Anal. Chem. 1999, 71, 407-412. (27) Ye, J. N.; Baldwin, R. P. Anal. Chem. 1993, 65, 3525-3527. (28) Ye, J. N.; Baldwin, R. P. Anal. Chem. 1994, 66, 2669-2674.

µm o.d.). The other one is an on-capillary electrode prepared by thermal evaporation of gold onto the outlet of the separation capillary (12.5 µm i.d. × 375 µm o.d.).17,18 Because the inner diameter of the separation capillary is as small as 12.5 µm, the resistance of the electrolyte in the capillary is so large that the resulting small electrophoretic current does not interfere with the electrochemical current from analytes.6 The disk electrode is placed at the end of the on-capillary electrode between a pair of acrylic plates. One of the plates has a straight groove used to guide the electrodes in place. The two electrodes can be aligned satisfactorily because their outer diameters are nearly identical. The assembly is thereupon fixed by acrylic screws.15 Figure 1 illustrates that this setup makes electrodes parallelopposed in a thin-layer geometry along the electrolyte flow pathway. When electroactive analytes migrate through the separation capillary, EC reactions of the analytes take place first at the gold film on the inner capillary wall. The film is formed due to diffusion of gold vapor into the capillary during thermal evaporation. The on-capillary electrode can thus have two working modes. Similar to a generator electrode, the on-capillary electrode can be used to initiate the redox cycling (Figure 1). Because the redox cycling significantly enhances the signal, the disk electrode has Coulomb efficiency of more than 150%, about 10 times larger than that of single-electrode end-column detection (vide infra). Alternatively, the on-capillary electrode can be used to screen the interference in the sample. Therefore, this configuration has the characteristics of two electrodes facing each other and two electrodes in series. RESULTS AND DISCUSSION Separation of Catecholamines. Catecholamines are selected for characterization of the dual CEEC operation and performance because they are the most frequently studied substances in previous CE-EC literature. Figure 2 is an example of a separation of dopamine (DA), norepinephrine (NE), epinephrine (Epi), and isoproterenol (IP). The analytes are oxidized at the on-capillary electrode and subsequently reduced at the disk electrode. The disk electrode is butted against the on-capillary electrode. The interelectrode gap is nominally 2-4 µm estimated by an optical microscope. The electrolyte is a 0.1 M phosphate solution (pH 5). The phosphate concentration and the pH of the electrolyte are chosen because under these conditions the amperometric Analytical Chemistry, Vol. 71, No. 15, August 1, 1999

3201

Figure 2. Electropherogram obtained with the dual-electrode amperometric detection. Analytes: 1.6 µM DA, 0.7 µM NE, 3.4 µM Epi, and 2.5 µM IP. Experimental conditions: fused-silica capillary, 15 cm long (12.5 µm i.d. × 375 µm o.d.); electrolyte, 0.1 M phosphate (pH 5.0); injection, 1 kV for 2 s; separation voltage, 3 kV; detection potential, 0.4 V (vs EAg/AgCl) at on-capillary electrode, -0.1 V at disk electrode; interelectrode distance, ∼3 µm.

response is the largest for catecholamines in LC.29,30 The 15-cmlong on-capillary electrodes are chosen to evaluate the performance of the dual-electrode detection because it is less convenient to place long capillaries in the space-limited evaporator. The collection efficiencies (defined as the ratio of the charge detected at the disk electrode to the charge at the on-capillary electrode, i.e., the ratio of peak areas) are 85, 82, 42, 84, and 90% for DA, NE, Epi, IP, and catechol (CA), respectively. In terms of ratios of maximal peak currents, corresponding values are 84, 74, 40, 77, and 86%. The retention time of CA is 14.31 min, far behind the other catecholamines, and thus peaks of CA are not shown in Figure 2. These collection efficiencies are three22 to ten21,25 times larger than CE-EC literature values reported by far. Comparison in detail could be meaningless because collection efficiencies are affected by many factors. For example, the relatively poor EC reversibility of Epi revealed in cyclic voltammograms suggests that its turnover percentage in each redox cycle is smaller than the other four catecholamines. Accordingly, the effect accumulated from multiple redox cycles gives Epi the lowest collection efficiency among the substances in Figure 2. Also, in solutions (29) Moyer, T. P.; Jiang, N.-S. J. Chromatogr. 1978, 153, 365. (30) Allenmark, S. J. Liq. Chromatogr. 1982, 5(S1), 1-41.

3202 Analytical Chemistry, Vol. 71, No. 15, August 1, 1999

Figure 3. Effect of the distance between the two electrodes on the performance of (A) collection efficiency (0, DA; 2, CA), (B) Coulomb efficiency for DA ([, on-capillary electrode; 0, disk electrode), and (C) Coulomb efficiency for CA ([, on-capillary electrode; 3, disk electrode). Analytes: 1.2 mM DA; 1.0 mM CA in 0.1 M phosphate (pH 6.9). Other conditions were the same as Figure 2.

with less basicity catecholamines are, in general, more reversible and consequently their collection efficiencies are relatively large. Other factors such as interelectrode distance, electrolyte flow velocity, and the size of the disk electrodes are discussed in the next section. Overall, results shown in Figure 2 clearly demonstrate the feasibility and advantages of applying the dual-electrode detector in CE-EC. Collection and Coulomb Efficiencies. As illustrated in Figure 1, this dual-electrode configuration has the advantage of signal enhancement because of the characteristics of thin-layer cells and the generator/collector mode. The turnover percentage of each redox cycle is dependent on the cell thickness, which is equivalent to the distance between the electrodes. Figure 3 shows the detector performance as a function of the interelectrode distance adjusted with a micropositioner onto which the disk electrode is firmly fixed. The starting point (0 µm distance in Figure 3) is set at where the disk electrode is butted against the on-capillary electrode. The redox substances are DA and CA. Because in low-pH solutions the migration time for CA is very long, instead of the pH 5 solution employed in Figure 2, a pH 6.8

phosphate electrolyte is used. Figure 3A shows that the collection efficiency begins to rise at an interelectrode distance of 100 µm and increases rapidly at 40 µm. Maximum values for DA and CA are 60 and 65%, respectively. The variation in charge is only 8% RSD for removal and reassembly of the disk electrode. Therefore, the collection efficiency of DA is utilized to ensure experimental reproducibility for a newly assembled cell. Because the procedures of cell assembly are the same for this setup and end-column detection,15 the reproducibility of both detectors is affected by the same factors, such as the capillary-electrode distance, alignment, and surface flatness on both the capillary and electrodes. The effect of interelectrode distance on Coulomb efficiency for DA and CA is shown in Figure 3B and C, respectively. Coulomb efficiency is defined as the charge detected relative to the charge equivalent to the amount of analyte injected.31 When the disk electrode is removed from the configuration, the Coulomb efficiencies of the on-capillary electrode for DA and CA are 91 and 96%, respectively, in a good agreement with literature values.17 When the interelectrode distance is shorter than 100 µm, not only the on-capillary electrode but also the disk electrode have Coulomb efficiencies greater than 100%. The maximum values at the on-capillary electrode are typically around 300%. The corresponding values at the disk electrode are around 165 and 200% for DA and CA, respectively. When the detection is changed to conventional single-electrode end-column detection configured simply by disconnecting the on-capillary electrode from potentiostat, the Coulomb efficiencies at the disk electrode decrease to 12% for DA and 16% for CA. Therefore, we conclude that the mechanism of redox cycling enhances the Coulomb efficiency more than 10-fold at the disk electrode. We should point out that such a parallel-opposed dual-electrode detector is believed to enhance only slightly the signal of catecholamines in LC where the interelectrode distance is 25 µm.1,3 In this present assembly, the gap is only 2-4 µm, which allows more redox cycles to occur. Therefore, the amount of redox charge is significantly increased. Figure 4 shows the effect of flow velocity on the performance of the dual-electrode detector. As the flow velocity becomes slower, both collection (Figure 4A) and Coulomb efficiencies (Figure 4B) increase because the time period for analytes staying between the electrodes becomes longer and the number of turnover cycles increases. Therefore, the signal is enhanced for analytes that can be regenerated by a reversible EC reaction. The curves of collection efficiency for both DA and CA are essentially overlapped and do not level off at slow flow velocities. The number of theoretical plates decreases significantly at slow flow velocities. The behavior of CA is the same as that of DA, but is not shown in Figure 4B and C for clarity. Efficiencies are significantly affected by the sizes of electrodes. Under the same experimental conditions as in Figure 3 (interelectrode distance, ∼3 µm), when the diameters of disk electrodes are 115 and 200 µm, the Coulomb efficiencies at the disk electrode increase to 230 and 310% and 400 and 650% for DA and CA, respectively. The improvement in efficiencies is a result of longer reaction zones for EC detection. Consequently, the numbers of theoretical plates decreased to ∼20 000 for both DA and CA. (31) Sloss, S.; Ewing, A. G. Anal. Chem. 1993, 65, 577-581.

Figure 4. Effect of flow velocity on the electrode performance of (A) collection efficiency, (B) Coulomb efficiency ([, on-capillary electrode; 0, disk electrode), and (C) number of theoretical plates at the disk electrode for 0.56 mM DA. Electrolyte, 0.1 M phosphate (pH 6.9). The flow velocity was adjusted by varying separation voltage from 1 to 6 kV in 1-kV steps. Other conditions were the same as Figure 2. The error bars indicate standard deviations (n ) 4).

Various Potential Settings. Summarized in Table 1 is Coulomb efficiency measured from five different modes of potential settings for the dual-electrode detector. Listed in Table 2 are the results of linear range, detection limit, sensitivity, and number of theoretical plates at the disk electrode for Modes I, III, and at the on-capillary electrode for mode II. The RSD values are within 8% (n ) 5). Modes I and II are equivalent to the configurations of end-column detection27 and integrated oncapillary detection,17,18 respectively. The potential settings of modes III and IV are reversed. Because of redox cycling, the efficiencies at both modes III and IV are significantly greater than those of modes I and II. Table 2 shows that the performance of mode III is far superior to the commonly used end-column configuration (mode I). The result of mode IV shows that, in addition to being the collector, the disk electrode can be the generator and gain significant enhancement in signal as well. Therefore, for samples containing EC-reversible substances in oxidized and reduced states, signals at both electrodes can be enhanced. In mode V, the potentials at both electrodes are set to oxidize the electroactive species in solution. By setting the onAnalytical Chemistry, Vol. 71, No. 15, August 1, 1999

3203

Table 1. Coulomb Efficiency at Various Detection Modesa Coulomb efficiency (%) potential setting detection mode I II III IV V

DA

on-capillary electrode

disk electrode

on-capillary electrode

turned off oxidation oxidation reduction oxidation

oxidation removal reduction oxidation oxidation

91 280 152 87

CA disk electrode

on-capillary electrode

disk electrode

12 168 159 1.9

16 96 312 107 93

203 128 1.0

a The concentrations for DA and CA were 0.56 and 0.60 mM, respectively. The electrolyte was 0.1 M phosphate (pH 6.9). Other conditions were the same as in Figure 2.

Table 2. Performance Characteristics of End-Column (I), On-Capillary (II), and Dual-Electrode (III) Detection Modesa for DA and CA mode

potential setting on-capillary disk electrode electrode

I II III

turned off oxidation oxidation

linear range

DA

CA

detection limit, µM (amol)

detection limit, µM (amol)

oxidation 1.0 µM-1.2 mM 0.21 (74) removal 50 nM-1.2 mM 0.026 (9.2) reduction 50 nM-1.2 mM 0.012 (4.2)

sensitivity, theoretical (pA/µM) plate no.b 6.9 54.5 105.6

59 000 56 000 52 000

linear range

1.0 µM-1.0 mM 0.34 (70) 0.5 µM-1.0 mM 0.082 (16.8) 0.1 µM-1.0 mM 0.029 (5.9)

sensitivity, theoretical pA/µM plate no.b 5.0 20.2 57.4

38 000 36 000 34 000

a Conditions were the same as in Table 1. The listed values are measured at the disk electrode for modes I, III, and at the on-capillary electrode for mode II. b Number of theoretical plates was calculated when both DA and CA were 6.0 µM.

capillary electrode at potentials that can oxidize interferences but not the analytes, the interference at the disk electrode can be suppressed. To illustrate this mode, DA and CA are taking the role of interferences. In mode V, interfering signals decrease to only 1.9 (DA) and 1.0% (CA), a 80-90% drop from that of endcolumn detection (mode I). This comparison suggests that a significant percentage of interferences can be removed by the oncapillary electrode. The number of theoretical plates (N) shown in Figure 4C and Table 2 appears smaller than most CE reports. Even the endcolumn configuration (mode I) results in N smaller than literature values (70 00015-74 00027) obtained using disk electrodes with similar diameters. Therefore, the peak broadness should not be attributed to the large disk electrode employed in the dualelectrode detector. A plausible explanation of the peak broadening is suggested in the following. In this study, a hole was drilled on the side of 1-mL vials and these were used as the sample and electrolyte reservoirs. The solution was confined in the vial because of surface tension. Injection and separation were performed by inserting the separation capillary into the vials through the hole. This operation is convenient for a 15-cm-long capillary. Because the capillary is short, it is difficult to bend and insert the capillary into the vial from the top as with a conventional 50-cmlong capillary. Between sampling and separation, it is operated by maneuvering the separation capillary and the detector as a whole rather than by switching vials because the anode of the high-voltage power supply is inserted into the vial from the top, vertical to the separation capillary. Therefore, deflection of the capillary during operation might cause broadening of sample zones and thus the relatively small plate numbers. This problem can be resolved by utilizing a long separation capillary and the conventional method of sample injection. For example, with a 60-cm3204 Analytical Chemistry, Vol. 71, No. 15, August 1, 1999

long capillary, plate numbers of 145 000 and 147 000 for DA and CA, respectively, can be obtained at the disk electrode under the same conditions as those used in Table 2. In summary, a dual-electrode detector is developed for capillary electrophoresis. Various detection modes shown in this study manifest the versatility of the dual-EC detector. The configuration inherits the properties of a thin-layer cell and generator/collector mode. Consequently the redox cycling between the electrodes greatly improves the sensitivity and detection limits. EXPERIMENTAL SECTION Electrode Fabrication and Cell Design. Fabrication of Au film on-capillary electrodes,17,18 Au disk electrodes,27 and the acrylic holder15 was similar to others reported in the literature. Fusedsilica capillaries (15 cm long; Polymicro Technologies Inc., Phoenix, AZ) were used as the separation capillaries and the oncapillary electrodes. The inner and outer diameters of the capillary were examined by an optical microscope and were found 12.5 and 375 µm, respectively. The capillaries were wrapped in Al foil, except for the region 0.5 cm from the end. They were then placed in the bell jar evaporator (KV-301, KEY High Vacuum Co., Nesconset, NY). The exposed region was placed ∼60° off from normal of the gold source. The base pressure, deposition rate, and film thickness were 1 × 10-8 Torr, 0.5 nm/s, and 100 nm, respectively. Problems of capillaries clogged by gold chunks never occurred. Electrical connection to the on-capillary electrode was made by cementing a thin varnished Cu thread on the Au film on the side of the capillary with conductive silver epoxy (Epoxy Technologies, Billerica, MA). After the silver epoxy was ovendried, the contact was covered with a layer of nonconductive epoxy (Epoxi-Patch, Dexter, Seabrook, NH) and the area of exposed Au film was minimized.

The disk electrode was constructed by inserting an Au wire (99.9%, Leesan, Tainan, Taiwan, ROC) through a capillary (375µm o.d.). The inner diameters of the capillaries were 100, 185, and 250 µm for Au wires of 90, 115, and 200 µm in diameter, respectively. The wire was sealed in the capillary with Epoxi-Patch. The alignment of the dual electrode was achieved with a straight groove machined on one of the acrylic plates as sketched in Figure 1. The electrodes and acrylic plates were secured in place by four acrylic screws. Polypropylene vials (1 mL) with a tiny hole drilled on the side were employed as reservoirs for the sample and running electrolyte. Because of surface tension, the solution was sustained in the vial without leaking out. Sampling and separation were conducted by insertion of the end of the separation capillary into the hole. Between sampling and separation, the tip of the separation capillary, along with the detector, was moved from one vial to another. To do so, the electrode holder was firmly attached onto a Teflon block so that the capillary was at the same height level as the hole. Both the sample vial and the electrolyte vial were equipped with an anode for the high-voltage power supply. Reagents and Solutions. All reagents were analytical reagent grade and were purchased from Sigma or Aldrich. Solutions were prepared with purified water (18 MΩ cm, Millipore-Q, Millipore Inc.). The electrolyte solutions were prepared from phosphates of monosodium and disodium salts. Solutions of catecholamines with desired concentrations were made fresh daily by serial

dilution in the separation electrolyte. Prior to injection, all solutions were filtered through a 0.45-µm PVDF filter (Lida Manufacturing Corp.). The injection voltage and injection time are 1 kV and 2 s, respectively. Apparatus. A regulated high-voltage dc power supply (model CZE 1000R, Spellman High-Voltage Electronics Corp., Plainview, NY) was used to apply high potential field for injection and separation. CE-EC detection was carried out on a potentiostat (model LC-4C, Bioanalytical Systems, West Lafayette, IN) with a four-electrode configuration including an on-capillary working electrode, a disk working electrode, an Ag-wire coated with chloride as a quasi-reference electrode, and a Pt counter electrode. Cyclic voltammetry was performed with a CHI model 630 potentiostat (CH Instruments, Cordova, TN). All CE data were collected by a PC with a data acquisition software ChemLab (Scientific Information Service Corp., Taipei, Taiwan, ROC). ACKNOWLEDGMENT This work was supported by National Science Council, R.O.C. C.h.C. gratefully acknowledges the Chemistry Department of National Sun Yat-Sen University for generous research support.

Received for review January 27, 1999. Accepted May 3, 1999. AC990069T

Analytical Chemistry, Vol. 71, No. 15, August 1, 1999

3205