Micromachined Electrophoresis Chips with Thick-Film Electrochemical

Assoc. 1998, 38, 210. (11) Rippeth, J.; Gibson, T.; Hart, J.; Hartley, I.; Nelson, G. Analyst (Cambridge,. U.K.) 1997, 122, 1425. Anal. Chem. 1999, 71...
0 downloads 0 Views 64KB Size
Download PDF
Anal. Chem. 1999, 71, 5436-5440

Micromachined Electrophoresis Chips with Thick-Film Electrochemical Detectors Joseph Wang,* Baomin Tian, and Eskil Sahlin

Department of Chemistry and Biochemistry, New Mexico State University, Las Cruces, NM 88003

A capillary electrophoresis (CE) microsystem, based on the combination of microphotolithographically fabricated separation chips and thick-film electrochemical detector strips, is described. The microsystem consists of a planar screen-printed carbon line electrode mounted perpendicular to the flow direction. Such coupling obviates the need for permanent attachment of the detector, to allow easy and fast replacement of the working electrode. Variables influencing the separation efficiency and amperometric response, including the channel-electrode spacing, separation voltage, or detection potential, are assessed and optimized. The versatility, simplicity, and low-cost advantages of the new design are coupled to an attractive performance, with submicromolar detection limits, and good precision. Applicability for assays of mixtures of nitroaromatic explosives or catecholamines is demonstrated. Such use of screen-printed detectors should also benefit conventional CE systems, particularly in applications requiring a frequent replacement of the working electrode. The development of microscale (chip-based) separation devices, particularly micromachined capillary electrophoresis (CE) systems, has witnessed an explosive growth in recent years.1,2 Such miniaturized devices represent the ability to shrink conventional “benchtop” separation systems with major advantages of speed, cost, portability, and solvent/sample consumption.3 As the field of chip-based separation microsystems continues its rapid growth, there are urgent needs for developing compatible detection modes. Much of the work on CE microchips uses laserfluorescence detection. Yet, such detection requires a large and expensive supporting optical system and is limited to analytes that fluoresce or are amenable to derivatization with a fluorophore. Electrochemistry offers a considerable promise for detection in micromachined CE chips. Such detection offers remarkable sensitivity (comparable to that of fluorescence),4 tunable selectivity, and low-volume requirements. Particularly attractive for onchip applications is the inherent miniaturization of electrochemical devices (and of the control instrumentation), their low-power requirements, extremely low cost, and high compatibility with * Corresponding author: (phone) 505-646-2140; (fax) 505-646-6033; (e-mail) [email protected]. (1) Freemantle, M. Chem. Eng. News, 1999, Feb. 22, 27. (2) Kovacs, ZG. T.; Petersen, K.; Albin, M. Anal. Chem. 1996, 68, 407A. (3) Hadd, A.; Raymond, D.; Halliwell, J.; Jacobson, S.; Ramsey, M. Anal. Chem. 1997, 69, 3407. (4) Gavin, P.; Ewing, A. G. Anal. Chem. 1997, 69, 3838.

5436 Analytical Chemistry, Vol. 71, No. 23, December 1, 1999

advanced micromachining and microfabrication technologies. Electrochemical detection has already proven to be extremely useful for conventional CE systems based on fused-silica capillaries,5,6 but rarely in connection to planar micromachined CE chips. The major challenges for such integration are similar to those of conventional CE systems, namely isolation of the working electrode from the high separation voltage and its proper alignment with the capillary. Additional consideration should be given to the different dimensions, shape, and materials of planar CE microchips. Woolley et al.7 reported recently on CE chips with lithographically fabricated electrodes located outside the exit of the channel, Gavin and Ewing4 developed a thin-film microfabricated electrochemical array detector for planar CE chips, while we8 described an on-chip detector based on sputtering the working electrode directly onto the channel outlet. This paper reports on a powerful and flexible electrochemical detection for CE chips based on thick-film detector strips. The thick-film (screen-printing) microfabrication technology is commonly used for large-scale production of extremely inexpensive and yet highly reproducible electrochemical sensors.9 For example, most commercial “one-shot” glucose biosensors rely on the thick-film fabrication process.10 Thick-film electrodes have been previously used as biosensor detectors for flow injection analyzers,11 but not in connection to CE microchips or CE systems, in general. The present microsystem couples the microphotolithographically fabricated CE glass chips with planar thick-film electrodes on ceramic wafers (Figure 1). Rather than fixing the detector permanently to the chip,6,7 the new design permits convenient and rapid replacement of the detector wafer. Such an easily exchangeable detector adds great versatility to the CE/electrochemistry operation, particularly in applications requiring a frequent electrode replacement. For example, it allows a convenient surface modification (in a separate/optimal electrochemical cell), a fast replacement of passivated electrodes (within 5-10 s), or the comparison and use of different electrode materials. Despite its remarkably low cost, the thick-film detector displays an attractive analytical performance, with lower detection limits than analogous (5) Voegel, P. D.; Baldwin, R. P. Electrophoresis 1997, 18, 2267. (6) Holland, L. A.; Lunte, S. M. Anal. Commun. 1998, 35, 1H. (7) Woolley A. T.; Lao K.; Glazer A. N.; Mathies R. A. Anal. Chem. 1998, 70, 684. (8) Wang, J.; Tian, B.; Sahlin, E. Anal. Chem.,1999, 71, 3902. (9) Hart, J.; Wring, S. Electroanalysis 1994, 6, 617. (10) Kirk, J. K.; Rheney, C. C. J. Am. Pharm. Assoc. 1998, 38, 210. (11) Rippeth, J.; Gibson, T.; Hart, J.; Hartley, I.; Nelson, G. Analyst (Cambridge, U.K.) 1997, 122, 1425. 10.1021/ac990807d CCC: $18.00

© 1999 American Chemical Society Published on Web 10/29/1999

Figure 1. Capillary electrophoretic system with electrochemical detection. (A) Glass microchip, (B) separation channel, (C) injection channel, (D) pipet tip for buffer reservoir, (E) pipet tip for sample reservoir, (F) pipet tip for reservoir not used, (G) Plexiglass body, (H) buffer reservoir, (I) sample reservoir, (J) blocked (unused) reservoir, (K) detection reservoir, (L) screen-printed working-electrode strip, (M) screen-printed working electrode, (N) silver ink contact, (O) insulator, (P) tape (spacer), (Q) channel outlet, (R) counter electrode, (S) reference electrode, (T) high-voltage power electrodes, (U) plastic screw. For clarity, the chip, its holder, and the screen-printed electrode strip are separated, and dimensions are not in scale.

thin-film electrodes.7 The microfabricated detector has no adverse effects on the CE separation and requires no decoupling mechanism or time-consuming alignment procedures, in a manner analogous to end-column detectors to a conventional CE capillary.12 The characterization, optimization, and advantages of the CE chip/thick-film electrochemical detection microsystem are reported in the following sections. EXPERIMENTAL SECTION Apparatus. The homemade high-voltage power supply had an adjustable voltage range between 0 and +4000 V. The glass microchannel separation chips were fabricated at Alberta Microelectronic Corp. (AMC model MC-BF4-001, Edmonton, Canada), using standard microphotolithographic technology, including wetchemical etching and thermal bonding techniques. The original waste reservoir had been cut off by AMC, leaving the channel outlet at the highly flat end of the chip. The glass chip, shown in Figure 1, consisted of a glass plate (120 × 87 mm), with a 77mm-long separation channel (between a deliberately blocked/ unused reservoir and the channel outlet at the detection reservoir) and a 10-mm-long injection channel (between the sample reservoir and the buffer reservoir). The two channels crossed each other halfway between the sample and the buffer reservoir and 5 mm from the blocked reservoir to yield a separation channel with an effective length of 72 mm. The channels had a half-circle cross section, with a maximum depth of 20 µm and a width of 50 µm at the top. Pipet tips were inserted into the holes of the buffer and sample reservoirs (see Figure 1). The glass chip was fixed in a laboratory-built Plexiglass holder (Figure 1), with silicone grease providing proper sealing. The holder contained reservoirs for the sample and buffer solutions, a detection reservoir, and an unused reservoir. A platinum wire (12) Ye, J.; Baldwin, R. P. Anal. Chem. 1993, 65, 3525.

was inserted into each reservoir and served as a contact for the high-voltage power supply. An additional platinum wire and an Ag/AgCl wire were also inserted into the detection reservoir, serving as the counter and reference electrodes, respectively, for the amperometric detection. The Ag/AgCl wire was prepared by electrochemical oxidization of a silver wire in 0.10 M hydrochloric acid. The detection reservoir has a special groove into which the screen-printed electrode strip fits exactly, to allow reproducible and stable positioning, perpendicular to the flow direction. The screen-printed electrode strip was further held in place by a plastic screw pressing the strip against the channel outlet. Amperometric detection was performed with an Electrochemical Analyzer 621 (CH Instruments) connected to a Pentium 166 MHz computer with 32 MB RAM. Chemicals. 1,2-Benzenediol (catechol) (99%), 3,4-dihydroxyphenethylamine hydrochloride (dopamine), 4-[1-hydroxy-2-(methylamino)ethyl]-1,2-benzenediol hydrochloride (epinephrine), and sodium hydroxide were obtained from Sigma, while 2-(4-morpholino)ethanesulfonic acid hydrate (MES hydrate) was obtained from Lancaster. Stock solutions of dopamine and catechol (10 mM) were prepared daily in deionized water. 2,4,6-Trinitrotoluene (TNT) was obtained from Chem Service (West Chester, PA). The TNT stock solution (1000 mg/L) was prepared in 70% (v/v) acetonitrile. Stock solutions (1000 mg/L in acetonitrile) of 1,3dinitrobenzene (DNB), 2,4-dinitrotoluene (2,4-DNT), 2,6-dinitrotoluene (2,6-DNT), and 4-nitrotoluene (4-NT) were obtained from Radian International (Austin, TX). Sodium dodecyl sulfate (SDS) (95%), sodium borate 10-hydrate, and hydrochloric acid (reagentgrade) were obtained from J. T. Baker. Acetonitrile was obtained from Sigma-Aldrich. Prior to use, the stock solutions were diluted in electrophoresis buffer. Screen-Printed Electrodes. The screen-printed electrodes were printed with a semiautomatic printer (model TF 100, MPM, Franklin, MA). Three different carbon inks were tested for fabricating the working electrode, including an Acheson ink Electrodag 440B (49AB90) (Acheson Colloids, Ontario, CA), Ercon ink G-448(I) (Ercon, Waltham, MA), and a ESL ink RS12113 modified to contain 30% extra carbon (Electro-Science Laboratories, Inc., PA). Printing was performed through patterned stencils (100-µm-thick, Specialty Photo-Etch, Inc., Texas) onto 100 × 100 × 0.64 mm alumina ceramic plates. Each plate consisted of 30 strips (33.3 × 10.0 × 0.64 mm) with each strip being defined by a laser pre/semi cut. The printing procedure consisted of the following steps. A carbon ink working-electrode layer (0.3 × 8.0 mm) was first printed on each of the strips of the ceramic plate and was cured at 100 °C for 30 min. Then, a silver ink (Ercon R-421(DRE-68)) contact layer (1.5 × 21.0 mm), partially overlapping the carbon layer, was printed and cured at 100 °C for 30 min. An insulating ink (Ercon R-488CI-G1 Insulator Green) layer was subsequently printed to cover the carbon-silver junction and to define the working-electrode area (0.30 × 2.5 mm), on one end, and to expose the contact area on the other side. The strips were then cured at 100 °C for 120 min. The final strip is shown schematically in Figure 1(L-P). The cured layers of carbon, silver, and insulator had thicknesses of 10, 28, and 70 µm, respectively. Prior to use, pieces of tape (Scotch, Magic Tape 810), with thicknesses of 60 µm each, were placed as shown in Figure 1(P). Analytical Chemistry, Vol. 71, No. 23, December 1, 1999

5437

These tapes served as spacers controlling the distance between the strip and the channel outlet. Electrophoresis Procedure. Prior to use, the channels were treated by rinsing them with a 1.0 M sodium hydroxide solution for 20 min, followed by deionized water for 1 min, 1.0% hydrochloric acid for 20 min, and finally with deionized water for 1 min. For the separation, the buffer and sample reservoirs (in the chip holder) and the corresponding pipet tips on the microchannel chip were filled with 250 µL of buffer and sample solutions, respectively. The chip was then placed in its holder with the pipet tips pointing downward into the reservoirs, and the detection reservoir was filled with buffer solution. Finally, the high-voltage power supply was connected to the reservoirs. To fill the injection channel between the separation channel and the sample reservoir with sample solution, +1500 V was applied for 30 s to the sample reservoir with the detection reservoir grounded and the buffer reservoir floating. The electrophoresis buffer consisted of MES buffer (25 mM, pH 6.5) for the separation of catecholamines, and the injection (sample “loading”) was performed by applying +1000 V to the sample reservoir for 2 s with the detection reservoir grounded and the buffer reservoir floating. A buffer solution, containing 15 mM borate (pH 8.7) and 25 mM SDS, was used for the separation of explosives. For this purpose, the injection was carried out by applying +1500 V to the sample reservoir for 3 s, with the detection reservoir grounded and the buffer reservoir floating. Prior to use, all buffer solutions were filtered through a 0.45-µm filter (Gelman Acrodisc) and sonicated for 20 min. Separations were usually carried out by applying +1500 V to the buffer reservoir with the detection reservoir grounded and the sample reservoir floating. The solutions were not dearated. Safety Considerations: To avoid electrical shock the high-voltage power supply should be handled with extreme care. Electrochemical Detection. Unless mentioned otherwise, the electropherograms were recorded after the background stabilization, with a time resolution of 0.1 s, using applied detection potentials (vs Ag/AgCl) of +0.70 V for catecholamines and -0.70 V for explosives. No software filtration of the signal was used. RESULTS AND DISCUSSION Characterization and Optimization. The coupling of thickfilm electrodes with micromachined CE systems requires proper attention to the screen-printing conditions and the capillarydetector spacing. The electrochemical reactivity and hence the performance of screen-printed working-electrode detectors is commonly influenced by the composition and source of the carbon ink.13 Figure 2(a) displays electropherograms for an equimolar (100 µM) mixture of dopamine (DA) and catechol (CA) using detectors based on the Ercon (A), Acheson (B), and ESL (C) carbon inks and a separation voltage of +1500 V. The Ercon and Acheson based working electrodes resulted in well-defined, sharp, and resolved peaks; a flat baseline; and favorable signal-to-noise characteristics. No response is observed using the ESL working electrode (C(a)). Anodic activation, often used for enhancing the electrochemical reactivity of thick-film detectors,14 offers a dra(13) Wang, J.; Tian, B.; Nascimento, V.; Angnes, L. Electrochim. Acta 1998, 43, 3459. (14) Wang, J.; Pedrero, M.; Sakslund, H.; Hammerich, O.; Pingarron, J. Analyst (Cambridge, U.K.) 1996, 121, 345.

5438 Analytical Chemistry, Vol. 71, No. 23, December 1, 1999

Figure 2. Influence of different carbon inks on the detection of 100 µM dopamine (DA) and catechol (CA) before (a) and after (b) pretreatment at +2.0 V for 1 min. Carbon inks: (A) Ercon, (B) Acheson, and (C) ESL. Conditions: MES buffer (25 mM, pH 6.5) as electrophoresis buffer; separation at +1500 V; sample injection at +1000 V for 2 s; detection at +0.70 V using a 60-µm spacing between the electrode surface and the channel outlet.

Figure 3. Influence of the separation voltage on the detection of 100 µM dopamine and catechol. Separation performed using (a) +1000 V, (b) +2000 V, (c) +3000 V, and (d) +4000 V. Other conditions, as in Figure 2(A(a)).

matic improvement of the response of the ESL-based detector (C(b)). Such treatment yields no further enhancement of the signals observed with the Ercon and Acheson electrodes. Erconbased thick-film detectors were selected for subsequent work. Figure 3 examines the influence of the separation voltage upon the response (using a 60-µm spacing between the channel outlet and the electrode surface). The separation efficiency, the current signals, and the baseline slope are all affected by the separation voltage. As expected, increasing the voltage from +1000 to +4000 V (in 1000 V steps, a-d) dramatically decreases the retention times for both analytes. The largest amperometric signals are observed using the +2000 V separation, while the +4000 V separation results in small peaks over a sloping baseline. The data of Figure 3, particularly the initial charging-current baseline rise, indicate incomplete isolation from the higher separation voltages. The separation voltage has a small effect upon the background noise. The peak-to-peak noise level increased from 40 to 55 pA upon changing the voltage between 1000 and 4000 V (not shown). Additional insights into this behavior, which reflects the effect of the separation voltage on the number of theoretical plates as well as upon the electrochemical detection process, can be obtained from the influence of the channel-electrode spacing.

Figure 5. Hydrodynamic voltammograms for 100 µM catechol (a) and epinephrine (b). Other conditions, as in Figure 2(A(a)).

Figure 4. Influence of the distance between the channel outlet and the screen-printed working electrode upon the response using separation voltages of +1500(A) and +3000 (B) V. The screen-printed strip is separated from the channel outlet with (a) 60, (b) 120, (c) 180, and (d) 240 µm. Other conditions as in Figure 2(A(a)).

Figure 4 displays the effect of the spacing upon the response using separation voltages of +1500 (A) and +3000 (B) V. Using the +1500 V separations, the amperometric signal decreases dramatically (∼10-fold) upon increasing the spacing between 60 and 240 µm (a-d, A). The spacing also influences the separation efficiency, as indicated from the decrease in the number of theoretical plates, from 4400 to 130 (for dopamine) and from 4100 to 490 (for catechol) between 60 and 240 µm, respectively. Such changes in the separation efficiency reflect the increased postcapillary diffusional broadening at large channel-electrode distances. Longer channels can be used for improving the separation efficiency. The peak broadening at the larger spacing is coupled to a slight increase in the retention times, from 49 to 55 s (for dopamine) and from 88 to 94 s (for catechol) between 60 and 240 µm. A surprisingly different trend is observed using the +3000 V separations, for which the 120-µm spacing yielded the largest signals, shortest retention times, and highest separation efficiency (b,B). For example, the number of theoretical plates (for dopamine) are 830, 2010, 950, and 200 for the 60-, 120-, 180-, and 240µm distances, respectively. Such a trend may be attributed to the fact that the spacing affects not only the dispersive zone broadening (outside the channel) but also the decoupling from the separation voltage (with improved isolation at larger distances), as well as the fluid mass transport at the detector surface. The net result of these, often opposite effects, is the complex behavior observed in Figures 3 and 4. The capillary-detector spacing also had a negligible effect upon the detector noise, as was indicated from another experiment employing a more sensitive scale (not shown). The present detector resembles the wall-jet design (with the channel/nozzle dimensions being much smaller than the detector wall).15 For conventional HPLC hydrodynamic wall-jet detectors, the jet issuing from the nozzle remains intact up to 10 mm.15 In view of the different velocity profile at the end-column CE detector (electrosmotic flow “pushing” the analyte), it is not clear whether the liquid breaks up before impinging the detector. Compared (15) Gunanigsham, H.; Fleet, B. Anal. Chem. 1983, 55, 1409.

with wall-jet-type detectors for conventional CE systems,12 involving a fixed disk electrode opposite to a circular capillary outlet, the present replaceable detector strip consists of a printed carbon line electrode (300-µm width) positioned and centered across from a half-circle channel (of 50-µm diameter). Figure 5 depicts hydrodynamic voltammograms (HDV) for the oxidation of catechol (a) and epinephrine (b). The curves were taken stepwise, in connection to a 1500 V CE separation, by making 100 mV changes in the detection potential. As expected for the oxidation of the catechol moiety, both compounds display no response below +0.40 V. The response rises gradually between +0.50 and +0.90 V, after which it levels off. The half-wave potentials are +0.66 V (epinephrine) and +0.69 V (catechol). Such drawn-out voltammograms reflect the resistance of the printedcarbon composite surface. Subsequent amperometric detection work employed a potential of +0.70 V that offered the best signalto-noise characteristics. A dramatic increase in the baseline current, its slope, and the corresponding noise was observed at higher potentials. Similar to end-column detection to conventional CE systems,16 the HDV profile was influenced by the separation voltage. A 100 mV anodic shift of the waves and ∼50% enhancements of the limiting current were observed upon increasing the separation voltage from 1500 to 3000 V (not shown). Analytical Performance. The thick-film electrochemical CE detector displays a well-defined concentration dependence. Electropherograms for sample mixtures containing increasing levels of dopamine and catechol in 2 × 10-5 M steps are shown in Figure 6(a-e). Defined peaks proportional to the analyte concentration are observed for both compounds. The resulting calibration plots (not shown) were linear with sensitivities of 0.160 and 0.0610 nA/ µM for dopamine and catechol, respectively (correlation coefficients, 0.998 and 0.989). Combining the high sensitivity of the thick-film detector with its low noise level results in low detection limits of 3.8 × 10-7 M dopamine and 7.8 × 10-7M catechol, respectively (based on the signal-to-noise characteristics (S/N ) 3) of the response to a mixture containing 2 × 10-6 M of these compounds; not shown; conditions as in Figure 2(Aa)). Such values are substantially lower than those (3.7 × 10-6 M dopamine and 1.2 × 10-5 M catechol) obtained with the on-chip thin-film electrochemical detector.7 Such values are similar to those common for amperometric detection with conventional (fusedsilica capillary) CE.17 The high sensitivity and speed of the CE/thick-film detector system is coupled to very good reproducibility. A series of 20 (16) Wallenburg, S. R.; Nyholm, L.; Lunte, C. E. Anal. Chem. 1999, 71, 544. (17) Zhong, M.; Lunte, S. M. Anal. Chem. 1996, 68, 2488.

Analytical Chemistry, Vol. 71, No. 23, December 1, 1999

5439

Figure 6. Electropherograms for mixtures containing 20(a), 40(b), 60 (c), 80 (d), and 100 (e) µM catechol and dopamine. Conditions, as in Figure 2(A(a)).

repetitive injections of a 50 µM dopamine solution (using the same detector strip) gave a mean value of 7.4 nA and a relative standard deviation of 4.3%. Different detector strips also displayed a good precision, as expected for the thick-film microfabrication and the reproducible detector positioning. The design of the microsystem permits rapid (5-10 s) replacement of the detector strip. This is in contrast to CE chips with integrated thin-film7 or sputtered8 working electrodes that require replacement of the entire microsystem in the case of electrode passivation. This includes fouling by accumulation of reaction products (as in the detection of phenolic compounds) or of surface-active agents (such as the SDS buffer component). Figure 7 demonstrates the utility of the CE/electrochemical system for analyzing a mixture of common nitroaromatic explosives. Owing to forensic security and environmental considerations, there are urgent needs for developing rapid, reliable, and miniaturized devices for decentralized detection of explosives. The inherent redox activity of nitroaromatic explosives makes them ideal candidates for electrochemical detection. Amperometric detection has been employed previously for measuring nitroaromatic explosives following their conventional CE 18 separations, but not in connection to microscale on-chip analysis. The microchip explosive analysis was performed with a borate buffer (15 mM, pH 8.7) containing 25 mM SDS. The utility of this buffer for separating explosive agents has been documented earlier.19 The electropherogram of Figure 7 indicates the convenient and rapid separation and detection of five such explosive compounds (DNB, 2,4-DNT, 2,6-DNT, 4-NT, and TNT) with a total time of around 3 min in connection to a separation potential of +1500 V. Despite (18) Hilmi, A.; Luong, J. H. T.; Nguyen, A.-L. Anal. Chem. 1999, 71, 873. (19) Kleibohmer, W.; Kamman, K.; Robert, J.; Musenbrock, E. J. Chromatogr. 1993, 638, 349. (20) Henry, C. S.; Zhong, M.; Lunte, S. M., Kim, M.; Bau, H.; Santiago, J. Anal. Commun. 1999, 36, 305.

5440 Analytical Chemistry, Vol. 71, No. 23, December 1, 1999

Figure 7. Electropherogram for 10 mg/L DNB; 2,4-DNT; 2,6-DNT; and TNT and 20 mg/L 4-NT. Conditions: electrophoresis medium, borate buffer (15 mM, pH 8.7) containing 25 mM SDS; separation at +1500 V; sample injection at +1500 V for 3 s; detection at -0.70 V using a 60-µm spacing between the detector strip and the channel outlet.

the negative potential (-0.70 V) essential for reducing the nitro moiety, the thick-film electrochemical detector displays a low background noise and sharp peaks for these 10-20 mg/L concentrations. Such favorable signal-to-noise characteristics result in low detection limits ranging from 0.6 mg/L (for DNB and TNT) to 2.0 mg/L (for 4-NT; S/N ) 3). Such detectability and speed are consistent with various on-site security and environmental needs. CONCLUSIONS The results clearly demonstrate that the combination of thickfilm amperometric electrodes with microchip CE systems results in a versatile analytical device. The ability to readily exchange the sensing electrode should be extremely useful for many practical applications, ranging from surface poisoning to surface modification. Such flexibility and low cost are coupled to an attractive performance, with very low detection limits and good precision. The thick-film fabrication process allows the printing of a wide range of electrode films. While the advantages of the thick-film electrochemical detector have been presented within the context of microchip CE systems, such a detector should be attractive for other micromachined flow analyzers, as well for ordinary CE systems. Coupling the detector strips with newly developed ceramic CE microchips20 should lead to disposable CE/ electrochemical systems. Work is in progress toward further miniaturization and integration of an on-chip potentiostatic circuitry and toward expanding the scope of electrochemical detectors in microanalytical systems. ACKNOWLEDGMENT This work was supported in part by Sandia NL. E.S. acknowledges financial support from the Swedish Foundation for International Cooperation Research and Higher Education (STINT). Received for review July 21, 1999. Accepted September 21, 1999. AC990807D