Anal. Chem. 2000, 72, 4821-4825
Capillary Electrophoresis with an Integrated On-Capillary Tubular Detector Based on a Carbon Sol-Gel-Derived Platform Lin Hua and Swee Ngin Tan*
Division of Chemistry, School of Science, Nanyang Technological University, 469 Bukit Timah Road, Singapore 259756
An integrated on-capillary tubular electrochemical detector for capillary electrophoresis systems has been fabricated based on sol-gel technique. It consists of a solgel carbon composite tubular electrode attached permanently onto the outlet of the separation capillary. The device greatly eases the setting up of capillary electrophoresis with electrochemical detection (CEEC) as it makes possible electrode/capillary alignment without the aid of a micromanipulator since this integrated unit can be simply immersed in the CE separation buffer in an ordinary three-electrode stationary cell. To improve analytical performance of the integrated unit, the external wall of the exit capillary was etched with HF after the polyimide coating of the capillary had been removed. Influences of the working electrode length and the wall thickness at the outlet of capillary on the separation efficiency and amperometric sensitivity were assessed and optimized. The practical applicability of this configuration is demonstrated with the detection of both catecholamines and carbohydrates. The advantages, namely, versatility, convenience, ease of operation, and low-cost, of the new design combined with an excellent performance lead to high stability and low detection limits. Since its introduction, capillary electrophoresis with electrochemical detection (CEEC) is of considerable interest for biological and chemical analysis. In many cases, EC methods provide a remarkably sensitive detection approach with low-volume requirements, often rivaling optical detection methods which are currently used for most CE applications. However, CEEC still possesses some inherent problems that hamper its wide practical applicability and utility. Due to the use of the micrometer-size diameter capillaries in CE and consequently the small dimension of the electrodes employed for CEEC, one of the major challenges for such an analytical technique is the difficulty encountered in aligning the working electrode with the end of the separation capillary. In addition, a relatively few kinds of electroactive species have been reported that allow direct monitoring at typical conventional electrodes, e.g., carbon electordes. Thus, these factors are likely to restrict the wide general acceptance of CEEC as a common routine analytical method. * Corresponding author: (e-mail):
[email protected]. 10.1021/ac0005363 CCC: $19.00 Published on Web 09/08/2000
© 2000 American Chemical Society
Fortunately, several attempts have been reported in the literature on the development of new EC detector configuration for the precise positioning of the detection electrode against the outlet of the separation capillary. In their initial report, Wallingford and Ewing1 introduced a now commonly used design, termed incapillary detection, in which a 10-µm-diameter carbon fiber electrode was physically inserted into the exit end of a 75-µm-i.d. capillary. Another most commonly used configuration for capillary/electrode alignment is the end-capillary detection, where the working electrode is placed against the end of the capillary.2 In both in-capillary and end-capillary configurations, the capillary outlet and electrode tip are difficult to be seen by the naked eye. The wall-jet configuration represents one of the very successful solutions for solving the electrode/capillary alignment problem with a working electrode larger than the inner diameter of the capillary. In the original wall-jet CEEC system, suggested by Ye and Baldwin,3 a flat disk Cu electrode was placed very close to the outlet of a relatively small capillary. Despite these developments, most procedures, in practice, are involved with the use of a micromanipulator to align the electrode with the capillary. This approach is time-consuming, requires the specialist to operate the system, and is difficult to maintain reproducibly over an extended operating period. These problems related to electrode/capillary alignment do somehow limit the wide CEEC applications. It is obvious that the incorporation of both the capillary and the electrode into a single integrated unit/platform can overcome the alignment problem. Surprisingly, not much research effort has been focused on the integration of CE with on-capillary EC detection, despite its remarkable sensitivity, compatibility with simple fabrication, and complete elimination of the requirements of micromanipulators for electrode alignment devices. Recently, Baldwin et al.4 described an on-capillary detector based on sputtering a thin conductive coating of Au or Pt onto the exit tip of the capillary, while Zhong and Lunte fabricated on-capillary electrodes (OCEs) with a gold wire mounted across the outlet of a capillary.5 Subsequently, they developed an on-capillary electrochemical detector which consisted of a gold tube (200-µm i.d., 300-µm o.d.) fixed onto the end of a capillary.6 Since the gold tube is a commercial product with fixed dimensions, the investigators (1) Wallingford, R. A.; Ewing, A. G. Anal. Chem. 1987, 59, 1762. (2) Huang, X.; Zare, N. R.; Sloss, S.; Ewing, A. G. Anal. Chem. 1991, 63, 189. (3) Ye, J.; Baldwin, R. P. Anal. Chem. 1993, 65, 3525. (4) Voegel, P. D.; Zhou, W.; Baldwin, R. P. Anal. Chem. 1997, 69, 951. (5) Zhong, M.; Lunte S. M. Anal. Chem. 1996, 68, 2488. (6) Zhong, M.; Lunte, S. M. Anal. Commun. 1998, 35, 209.
Analytical Chemistry, Vol. 72, No. 20, October 15, 2000 4821
did not optimize the configuration of this integrated on-capillary detection setup. A significant challenge in the on-capillary detection scheme is the selection of suitable rugged and flexible electrode fabrication materials. To date, the electrode materials employed for OCEs fabrication have been typically based on conventional electrode materials, such as Au4-6 and Pt.4 The major disadvantage of these materials is the lack of flexibility for electrode design due to their inherent hardness, high background currents, and the problem of electrode fouling. Moreover, such metal-based electrodes are less easy to be chemically modified. These drawbacks severely limit the OCE concept to be widely used in CE applications. Recently, a new class of carbon materials based on the flexible and versatile sol-gel technique were developed for electrode fabrication and modification.7-9 In this method, the electrodes are fabricated from graphite powder dispersed homogeneously in solgel-derived ceramic materials. Such sol-gel carbon composite electrodes (CCEs) have been reported to have numerous desirable properties, including high chemical stability, amenability to chemical modification, low background current, high abrasive resistance, and mechanical stability or rigidity. In addition, the sol-gel molding technique makes it possible to fabricate electrodes in virtually any geometrical configuration, e.g., flat plate, monolithic rods or disks, and even microelectrodes to suit any specific experimental requirement.10 To our knowledge, the application of this sol-gel carbon composite material involving on-capillary configuration has not yet been investigated in CE systems. Thus, the main objective of this work is to demonstrate the capabilities of a carbon-based matrix coupled with the solgel technique for an on-capillary setup. In recent years, chemically modified electrodes (CMEs), which introduce a catalyst or redox mediator, have been demonstrated to possess distinct advantages over conventional electrode materials in various CE application areas. One of the important properties of CMEs has been their ability to extend the applicability of CEEC to species that could not be detected directly via electrochemical means or that exhibit high overpotentials at unmodified electrodes. In principle, such CMEs should also be able to provide enhanced performance for on-capillary configurations as well. One major difficulty of fabricating CME-based OCEs is the lack of suitable electrode materials for such a purpose. The modified CCEs demonstrate a new class of electrochemical detectors, which was shown to permit bulk modification by organic, inorganic, and biochemical species. This simple technique can be easily performed by mixing the matrix material with the desired amount of catalyst to achieve a homogeneous bulk material.11-13 In our previous work, we investigated the feasibility of coupling sol-gel CCEs with CE systems14,15 based on wall-jet configuration. An important extension of this work would be to examine the capabilities of CE with a tubular on-capillary working electrode (7) Tsionsky, M.; Gun, G.; Glezer V.; Lev, O. Anal. Chem. 1994, 66, 1747. (8) Sampath, S.; Lev, O. J. Electroanal. Chem. 1997, 426, 131. (9) Li, J.; Chia, L. S.; Goh, N. K.; Tan, S. N. J. Electroanal. Chem. 1998, 460, 234. (10) Lin, J.; Brown, C. W. Tr. Anal. Chem. 1997, 16, 200. (11) Wang, J.; Taha, Z. Anal. Chem. 1990, 62, 1413. (12) Xie, Y.; Huber, C. O. Anal. Chem. 1991, 63, 1714. (13) Huang, X.; Pot, J. J.; Kok, W. Th. Anal. Chim. Acta 1995, 300, 5. (14) Hua, L.; Chia, L. S.; Goh, N. K.; Tan, S. N. Electroanalysis 2000, 12, 287. (15) Hua, L.; Tan, S. N. Anal. Chim. Acta 2000, 403, 179.
4822
Analytical Chemistry, Vol. 72, No. 20, October 15, 2000
for electrochemical detection based on the sol-gel technique. The design and application of both blank and modified integrated oncapillary CCEs are reported. This particular novel configuration, where the electrode is affixed to the exit end of the separation capillary, provides a very easy and reliable method to construct and use CEEC devices. We selected two application examples (neurotransmitters and carbohydrates) from a wide range of possibilities to demonstrate the applicability of such novel oncapillary electrode configuration. EXPERIMENTAL SECTION Apparatus. Electrophoresis in capillary was driven by a highvoltage supply (model CZE1000R, Spellman, New York) capable of delivering 0 ( 30 kV. For operation safety, the high-voltage input was placed in a Plexiglas box with an interlock switch on the access door. The outlet end of the capillary was always maintained grounded through a platinum wire in the amperometric detection cell. Experiments were carried out using fused-silica capillary of 25-µm i.d. and 360-µm o.d. (Polymicro Technologies, Phoenix, AZ). Before use, new capillaries were washed with distilled water, followed by 0.1 M NaOH and the separation buffer. Sample introduction was accomplished electrokinetically. Amperometric detection at a constant potential with CE was performed with a BAS LC-4CE amperometric detector (West Lafayette, IN). A conventional three-electrode system was used with the sol-gel carbon composite tubular working electrode fabricated on the capillary. A saturated calomel reference electrode (SCE) as a reference electrode and a platinum auxiliary electrode were used. The electropherograms were monitored with a stripchart recorder (model 8376-20, Cole-Parmer, Chicago, IL). Reagents. Methyltrimethoxysilane (MTMOS) was obtained from Merck-Schuchardt. Graphite powder (extra pure) was obtained from Merck. Cuprous oxide (Cu2O) was purchased from Aldrich. Catechol, dopamine, (()-epinephrine, (()-norepinephrine hydrogentartarate, and all the carbohydrate compounds were obtained from Sigma. All the reagents were used as received without further purification. Stock solutions of analytes and the separation electrolyte were prepared in distilled water. Prior to use, the stock solutions were diluted in electrophoresis buffer. Fabrication of Tubular On-Capillary Electrodes. A solgel solution was prepared by mixing 350 µL of MTMOS, 10 µL of HCl (2 M), and 300 µL of distilled water. The above mixture was sonicated for 2 min to ensure uniform mixing. Alcohol was not used as the solvent for homogenization since sonication was sufficient to mix all the reagents. Subsequently, 0.5 g of graphite powder was mixed thoroughly with the above sol-gel solution in a mortar to form a homogeneous carbon-sol-gel paste. A short section (∼2 cm) of polymer coating was removed from the detection end of the separation capillary by burning. This end of the exposed fused-silica capillary was then chemically etched by soaking in ∼50% HF solution for a specific time to achieve the desired thickness of the capillary (δ). This thickness was measured under a calibrated microscope (model SE, Nikon, Japan). The etched capillary was first inserted through a plastic tubing of internal diameter ∼0.5 mm. This plastic tubing was then inserted through another glass tubing (o.d. 5 mm, i.d. 3.5 mm, 8 cm long) which had been previously packed with the carbon solgel paste. At the same time, a copper wire was also passed through the same glass tubing parallel to the plastic tubing for electrical
Figure 1. Schematic configuration of the on-capillary tubular solgel carbon composite electrode (not drawn to scale): (A) separation capillary, (B) plastic tubing, (C) sol-gel carbon composite, (D) nail polish coating, (E) glass tubing support, and (F) copper wire.
contact. The etched end of the fused-silica capillary was pushed through the plastic tubing until it just emerged from the carbon sol-gel paste surface. Under a microscope, the plastic tubing was clamped firmly and the separation capillary was carefully pulled back to a desired length (l) so that a tubular electrode was formed surrounding the exit end of the separation capillary. Subsequently, the on-capillary carbon sol-gel composite electrode was left to polymerize for 4 days under laboratory ambient conditions. Finally, the surface of the sol-gel carbon composite was very carefully covered with a thin coating of nail polish excluding the little exit needed for capillary to emerge. Figure 1 illustrates the schematic diagram of the novel on-capillary sol-gel carbon composite electrode. Cu2O-modified tubular on-capillary electrodes were prepared following the same procedure above except that a thoroughly blended mixture consisting of 0.04 g of Cu2O powder and 0.5 g of graphite powder replaced the pure 0.5 g of graphite powder. RESULTS AND DISCUSSION Characterization and Optimization. The integrated oncapillary tubular detector in this CEEC system consisted of a solgel carbon composite tubular electrode affixed permanently at the end of the separation capillary. The critical parameters affecting the performance of this integrated unit shown in Figure 1 are the wall thickness (δ) at the separation capillary outlet and length (l) of the tubular electrode. Etching the outside wall of the exit end of the separation capillary provides a means for reducing the capillary thickness. Thus, this effectively sets up a steep diffusion profile at the tubular electrode as the solution flows through the exit end of the capillary. The sensitivity of the detector was improved significantly when δ was thin. Figure 2 shows the effect of δ on both separation efficiency and background noise for the detection of dopamine (200 µM) in a typical CE separation buffer consisting of 20 mM phosphate buffer at pH 5.8. It is clear that the optimum efficiency for dopamine can be obtained when the capillary wall at the exit is relatively thin. The cause of the loss in separation efficiency is mainly from the dead volume introduced
Figure 2. Dependence of separation efficiency (b) and noise level (9) on the wall thickness (δ) at capillary outlet using on-capillary tubular CCE for the detection of dopamine (200 µM). CEEC conditions: run buffer, 20 mM phosphate (pH 5.8); separation voltage, 16 kV; sample injection, electrokinetic at 16 kV for 3 s; capillary, 64 cm long; electrode length (l), 300 µm; detection potential, +0.80 V vs SCE.
by the difference of the internal diameter of the separation capillary and the tubular electrode. The internal diameter of the tubular electrode is a function of δ. When δ is reduced, the dead volume decreases, which will improve the separation efficiency, and this trend is depicted in Figure 2. During the investigation of the effect of δ on efficiency, it was observed that the background noise increased dramatically as the capillary wall was etched below 50 µm. This is attributed to the dramatic drop of the separation potential across the tubular electrode when δ is relatively small. On the basis of the results of Figure 2, as a compromise between separation efficiency and noise level, a wall thickness of ∼50 µm was chosen for all the subsequent experiments. In fact, a thinner capillary wall below 50 µm could be experimentally challenging as the capillary wall will break easily. The length (l) of the tubular electrode was found to have a profound effect on the separation efficiency and detection sensitivity obtained. This observation is illustrated in Figure 3, which shows the effect of the electrode length on the peak current response and separation efficiency for the detection of dopamine under conditions similar to those in Figure 2. It can be observed that when the electrode length is greater than 300 µm, broadening of the peak width becomes significant, thus lowering the separation efficiency for dopamine. Figure 3 also shows the effect of l on the sensitivity of the tubular electrode for the detection of 50 µM dopamine and epinephrine. It is expected that as the electrode length increases, the sensitivity increases. This increase in sensitivity with electrode length is caused by the increase of the area of the tubular working electrode. As the electrode area increases, however, the background current of this sol-gel carbon composite-based electrode does not increase a lot as such electrode material has inherent low background current. The current response for dopamine and epinephrine did not significantly decrease until the electrode length was less than 300 µm. Analytical Chemistry, Vol. 72, No. 20, October 15, 2000
4823
Figure 4. Electropherogram of 50 µM (1) dopamine, (2) norepinephrine, (3) epinephrine, and (4) catechol at an on-capillary tubular CCE. Other CE conditions are as in Figure 2; wall thickness (δ) at capillary outlet 50 µm, and electrode length (l) 300 µm.
If the length is set too short below 200 µm, the response decreases very sharply or even total disappearance of the peak current has been observed. The experimental results became very irreproducible. One possible explanation for the very low sensitivity obtained when l is below 200 µm is that the analyte does not have enough time to diffuse to the very short tubular working electrode surface as it is continuously being forced through the exit capillary by the electroosmotic force in the capillary. The dependence of coulometric efficiency on the wall thickness and the electrode length was investigated when electrode parameters (l and δ) were altered. It was found that coulometric efficiency of the on-capillary CCE decreased with decreasing wall thickness δ. There was a 15% decrease for dopamine when δ was changed from 168 to 25 µm. The relationship between the electrode length and coulometric efficiency was also investigated. When the length of CCE was altered from 600 to 200 µm, the coulometric efficiency decreased by 68% for dopamine. Under CE condition (δ ) 50 µm, l ) 300 µm), the coulometric efficiency for dopamine is 53% with separation voltage of 16 kV. In this on-capillary configuration, it is not necessary to use a specific decoupling system to isolate the sensing electrode from high CE fields due to the relatively small CE voltage drop across the length (l) of CCE. Thus, it is possible to accurately maintain the potential of the working electrode. This on-capillary detection scheme can be modeled in a similar way as optimized end-column detection16 because most of the applied CE voltage is dropped across the separation capillary. Since the relatively small resistance of the buffer solution (typically ∼100 Ω) can be ignored, the resistance of whole electric circuit can be modeled as the combination of the internal resistance of the separation capillary (Rcap) and the resistance resulting from the solution inside tubular electrode (Rcce). Under optimized fabrication conditions (l ) 300
µm, δ ) 50 µm), Rcap is estimated to be 1.6 × 1010 Ω, while Rcce is ∼105 Ω, assuming electrophoretic current is 1 µA. Therefore, the applied separation potential results in ∼100 mV potential drop across the 300-µm-long electrode. This potential drop is suitable for the amperometric detection. In comparison, the potential drop inside the separation capillary is 250 V/cm (7.5 V/300 µm). Analytical Performance of Unmodified Sol-Gel Carbon Composite Tubular Electrode. The dependence of the anodic current response on the applied potential was assessed by means of hydrodynamic voltammograms (HDVs). The peak current of four neurotransmitters was recorded after separation by varying the applied potential. The HDVs behavior obtained under CE conditions for these compounds showed almost a similar trend. Typically, the anodic wave started from ∼+0.30 V, followed by a current plateau in the vicinity of +0.80 V. Thus, the detection for these four compounds was performed at an applied potential of +0.80 V. The analytical performance of the on-capillary EC detection with CE system was characterized by running the electrophoresis with an equimolar (50 µM) mixture of dopamine, epinephrine, norepinephrine, and catechol in typical CE experimental conditions. A representative electropherogram is illustrated in Figure 4. Nearly completed resolved peaks are observed for these neurotransmitters, but peaks with minute tailing for some cationic analytes are also found in electropherograms, which may be ascribed to electrostatic interactions observed in the similar separations.17 The linear range of the calibration curve was evaluated from 2 to 500 µM for all compounds examined with a correlation coefficient of g0.998 (n ) 7). The on-capillary tubular CCE detector with its low noise level results in low limits of detection (LOD) of 0.7 µM (0.76 fmol) for dopamine and 1.2 µM (0.63 fmol) for catechol, based on S/N ) 3. These LOD values were slightly higher than those reported for optimized end-column amperometric detection at a carbon fiber microelectrode (0.4 µM (18 amol) for dopamine and 0.7 µM (19 amol) for catechol with a 2-µm-i.d. capillary),16 while comparable with those obtained for metal film electrodes directly deposited onto the capillary tip (0.6 µM (2.6 fmol) for dopamine and 1.5 µM (3.2 fmol) for catechol with a 25-µm-i.d. capillary).4 The better performance of this work may be attributed to the lower background current of CCE and faster electrode kinetics of the reaction at CCE. These results
(16) Sloss, S.; Ewing, A. G. Anal. Chem. 1993, 65, 577.
(17) Wallingford, R. A.; Ewing, A. G. Anal. Chem. 1987, 59, 1762.
Figure 3. Dependence of separation efficiency (O) and peak current on the electrode length (l) using on-capillary CCE for the detection of 50 µM dopamine (9) and 50 µM epinephrine (b) with the wall thickness (δ) at capillary outlet 50 µm. Other electrophoretic conditions are as in Figure 2.
4824 Analytical Chemistry, Vol. 72, No. 20, October 15, 2000
Figure 5. Electropherogram of carbohydrate compounds at a Cu2O modified on-capillary tubular electrode: (1) sucrose, (2) lactose, (3) glucose, (4) fructose, and (5) ribose. CEEC conditions: electrophoretic medium, 50 mM NaOH solution; separation voltage at 11 kV; sample injection at 11 kV for 3 s; detection at +0.65 V vs SCE using a 300-µm-long tubular CCE; capillary, 25-µm i.d., 360-µm o.d., 75 cm long with wall thickness 50 µm at capillary outlet. Concentrations of all analytes were 100 µM.
indicate that the integrated on-capillary detector is useful for highsensitivity assays. The excellent stability of this integrated on-capillary detector is confirmed by successive injections of samples through the detector. Quantitative reproducibility of peak currents for 15 injections of 100 µM dopamine and epinephrine injected over a period of 4 h, for example, was only 2.3 and 3.5%, respectively. These modest variations are likely due to the slight differences in injection volume rather than due to the performance of the oncapillary sol-gel tubular detector. The results clearly showed that high reproducibility for on-capillary configuration can be achieved by simply leaving the integrated on-capillary tubular detectors in the buffer reservoir without the aid of a micropositioner. Due to the integration of the capillary and electrode as a single unit, the typical alignment issue which causes irreproducibility in other configurations, e.g., end-capillary or wall-jet configuration, does not arise in this configuration. As far as long-term CEEC application was concerned, individual tubular OCEs can be used for 2 months or longer before the fused-silica capillary needed to be replaced. Thus, this integrated on-capillary tubular sol-gel detector could be useful for potential routine laboratory application operated by a nonspecialist. Analytical Performance of Chemically Modified Sol-gel Carbon Composite Tubular Electrode. One of the future directions with CEEC development work is the application of new electrode materials. The applicability of chemically modified solgel carbon composite in CEEC work was first reported by our group used in the conventional wall-jet configuration.14 The chemical modification of the sol-gel carbon composite tubular electrode in the integrated on-capillary configuration will extend the applications of CEEC. Figure 5 demonstrates the utility of the CE/Cu2O-modified on-capillary detectors for analyzing a mixture (18) Colon, L. A.; Dadoo, R.; Zare, R. N. Anal. Chem. 1993, 65, 476. (19) Huang, X.; Kok, W. T. J. Chromatogr., A 1995, 707, 335. (20) Fermier, A. M.; Gostkowski, M. L.; Colon, L. A. Anal. Chem. 1996, 68, 1661. (21) O’Shea, T. J.; Lunte, S. M.; LaCourse, W. R. Anal. Chem. 1993, 65, 948. (22) Roberts, R. E.; Johnson, D. C. Electroanalysis 1995, 7, 1015. (23) Lu, W.; Cassidy, R. M. Anal. Chem. 1993, 65, 2878.
of five common carbohydrates. Due to the lack of a strong chromophore in carbohydrates for direct UV detection, the high catalytic activity for the oxidation of these compounds12 make them ideal candidates for electrochemical detection. Carbohydrates has been previously determined following conventional CEEC configuration3,14,18-20 or pulse amperometric detection mode on a gold electrode,21-23 which requires more complicated instrumentation compared to direct amperometric detection. In this work, we demonstrated that a Cu2O-modified OCE, prepared by bulk modification with sol-gel carbon composite, could be employed for the direct amperometric detection of carbohydrates. With a capillary of 75-cm length, almost complete resolution of the mixture was achieved in 22 min with a separation electrolyte consisting of 50 mM NaOH. The utility of this buffer for separation and detection of carbohydrates has been documented earlier.14,18,19 The electropherogram of Figure 5, coupled with the very low noise, indicates the convenient quantification of micromolar levels of such compounds with a separation voltage of 11 kV. Lower separation efficiencies were found for the carbohydrates than for the catecholamines, which could have resulted from the Joule heating generated by the hydroxide electrolyte. Quantitative reproducibility was also excellent. The relative standard deviation obtained for 10 injections of glucose was only 3.8%. The limit of detection, based on S/N ) 3, was estimated to be below 1.5 µm (1.3 fmol) for the five sugars examined. Such values are similar to those obtained based on amperometric detection with a conventional CE system.14,19 CONCLUSIONS The results clearly demonstrate that the coupling of the integrated on-capillary tubular CCEs provides a very convenient and versatile analytical device for the CEEC system. This integrated design requires no alignment procedures when setting up CEEC experiments because the tubular electrode is permanently attached onto the capillary outlet and thus can be operated by a nonspecialist. Most important, many kinds of different OCEs can be fabricated based on the homogeneous dispersion of chemical modifiers in the sol-gel carbon composite. In comparison, CME-based carbon paste electrodes, which have been widely used for other electroanalysis, are not suitable in the CEEC work. The main drawback is due to the very poor mechanical stability of the soft carbon paste. The loss in separation efficiency resulting from the dead volume is minimized by means of the etching of the exit end of the separation capillary wall. The integrated system exhibits very attractive performance characteristics such as very low detection limits, good precision, and long-term stability. Due to their inherent versatility, convenience, low cost, and design flexibility, further research efforts are expected to expand the scope of the sol-gel-based integrated electrochemical detectors in other analytical microsystems. ACKNOWLEDGMENT The authors are very grateful to Nanyang Technological University (Singapore) for financial support, Grant RP13/98. L.H. acknowledges the award of NTU Postgraduate Research Scholarship. Received for review May 11, 2000. Accepted July 20, 2000. AC0005363 Analytical Chemistry, Vol. 72, No. 20, October 15, 2000
4825
