Microchip Capillary Electrophoresis Coupled with a Boron-Doped

Anal. Chem. 2003, 75, 935-939

Microchip Capillary Electrophoresis Coupled with a Boron-Doped Diamond Electrode-Based Electrochemical Detector Joseph Wang,* Gang Chen, and Madhu Prakash Chatrathi

Department of Chemistry and Biochemistry, New Mexico State University, Las Cruces, New Mexico 88003 Akira Fujishima,* Donald A. Tryk, and Dongchan Shin

Department of Applied Chemistry, School of Engineering, University of Tokyo, 7-3-1 Hongo, Bunkyo-ku, Tokyo 113-8656, Japan

The attractive behavior and advantages of a diamond electrode detector for a micromachined capillary electrophoresis (CE) system are discussed. A chemically vapordeposited boron-doped diamond (BDD) film band (0.3 × 6.0 mm) electrode is used for end-column amperometric detection. The favorable performance of the diamond electrode microchip detector is indicated from comparison to a commonly used thick-film carbon detector. The diamond electrode offers enhanced sensitivity, lower noise levels, and sharper peaks for several groups of important analytes (nitroaromatic explosives, organophosphate nerve agents, phenols). The favorable signal-to-background characteristics of the BDD-based CE detector are coupled with a greatly improved resistance to surface fouling and greater isolation from high separation voltages. The enhanced stability is indicated from a RSD of 0.8% for 60 repetitive measurements of 5 ppm 2,4,6-trinitrotoluene (vs RSD of 10.8% at the thick-film carbon electrode). A highly linear response is obtained for the explosives 1,3dinitrobenzene and 2,4-dinitrotoluene over the 2001400 ppb range, with detection limits of 70 and 110 ppb, respectively. Factors influencing the performance of the BDD detector are assessed and optimized. The attractive properties of BDD make it very promising material for electrochemical detection in CE microchip systems and other micromachined flow analyzers. Microfluidic devices, particularly micromachined capillary electrophoresis (CE) systems, can dramatically change the scale and speed at which chemical analysis is performed.1-3 Electrochemical (EC) detection offers great promise for CE microchips, with features that include high sensitivity, inherent miniaturization of both the detector and control instrumentation, low cost, low power demands, and high compatibility with micromachining * Corresponding authors: (e-mail) [email protected]; akira-fu@ fchem.chem.t.u-tokyo.ac.jp. (1) Figeys, D.; Pinto, D. Anal. Chem. 2000, 71, 330A. (2) Reyes, D. R.; Iossifidis, D.; Auroux, P. A.; Manz, A. Anal. Chem. 2002, 74, 2623. (3) Kutter, J. P. Trends Anal. Chem. 2000, 19, 352. 10.1021/ac0262053 CCC: $25.00 Published on Web 01/14/2003

© 2003 American Chemical Society

technologies.4,5 The performance of CE-EC microchips is strongly influenced by the working electrode material. The working electrode should provide favorable signal-to-background characteristics, as well as a reproducible response. A range of materials, including platinum, gold, and various forms of carbon, have thus found useful for chip-based EC detection, with each material offering certain advantages and disadvantages for a given application.6-11 This paper reports on the attractive performance and advantages of a diamond electrode detector for CE microchips. Borondoped diamond (BDD) electrodes have been studied intensely over recent years owing to their attractive properties.12-22 These properties include a wide potential window, low and stable background currents, negligible adsorption of organic compounds, and low sensitivity to oxygen. Diamond electrodes have thus (4) Wang, J. Talanta 2002, 56, 223. (5) Lacher, N. A.; Garrison, K. E.; Martin, R. S.; Lunte, S. M. Electrophoresis 2001, 22, 2526. (6) Woolley, A.; Lao, K.; Glazer, A.; Mathies, R. A. Anal. Chem. 1998, 70, 684. (7) Wang, J.; Tian, B.; Sahlin, E. Anal. Chem. 1999, 71, 5436. (8) Schwarz, M. A.; Galliker, B.; Fluri, K.; Kappes, T.; Hauser, P. C. Analyst 2001, 126, 147. (9) Martin, R. S.; Sawron, A. J.; Fogarty, B. A.; Regan, F. B.; Dempsey, E.; Lunte, S. M. Analyst 2001, 126, 277. (10) Wang, J.; Tian, B.; Sahlin, E. Anal. Chem. 1999, 71, 3901. (11) Zeng, Y.; Chen, H.; Pang, D. W.; Wang, Z. L.; Cheng, J. K. Anal. Chem. 2002, 74, 2441. (12) Xu, J. S.; Granger, M. C.; Chen, Q. Y.; Strojek, J. W.; Lister, T. E.; Swain, G. M. Anal. Chem. 1997, 69, A591. (13) Tatsuma, T.; Mori, H.; Fujishima, A. Anal. Chem. 2000, 72, 2919. (14) Prado, C.; Flechsig, G. U.; Grundler, P.; Foord, J. S.; Marken, F.; Compton, R. Analyst 2002, 127, 329. (15) Prado, C.; Wilkins, S. J.; Marken, F.; Compton, R. G. Electroanalysis 2002, 14, 262. (16) Spataru, N.; Sarada, B. V.; Popa, E.; Tryk, D. A.; Fujishima, A. Anal. Chem. 2001, 73, 514. (17) Xu, J.; Swain, G. A. Anal. Chem. 1998, 70, 1502. (18) Rao, T. N.; Yagi, I.; Miwa, T.; Tryk, D. A.; Fujishima, A. Anal. Chem. 1999, 71, 2506. (19) Terashima, C.; Rao, Tata N.; Sarada, B. V.; Tryk, D. A.; Fujishima, A. Anal. Chem. 2002, 74, 895. (20) Marken, F.; Paddon, C. A.; Asogan, D. Electrochem, Commun, 2002, 4, 62. (21) Rao, T. N.; Sarada, B. V.; Tryk, D. A., Fujishima, A. J. Electroanal. Chem. 2000, 491, 175. (22) Shin, D.; Sarada, B. V.; Tryk, D. A., Fujishima, A.; Wang, J. Anal. Chem. 2003, 75, 530-534.

Analytical Chemistry, Vol. 75, No. 4, February 15, 2003 935

proved extremely useful for a wide range of electroanalytical applications, including oxidase-based biosensors,13 electrical detection of nucleic acids,14 stripping analysis of heavy metals,15 voltammetric measurements of cysteine,16 azide anion,17 NADH,18 or chlorophenols,19 direct electrochemistry of cytochrome c,20 HPLC monitoring of sulfa drugs,21 or amperometric detection of neurotransmitters in conventional CE systems.22 The attractive properties of BDD make it a very promising material for EC detection in micromachined CE systems. The new BDD-based CE microchip detector offers very favorable signal-to-background characteristics, good resistance to surface fouling, isolation from high separation voltages, sharp peaks, and other attractive properties. Such advantages, along with detailed characterization and optimization of the diamond electrode microchip CE detector, are reported in the following sections. EXPERIMENTAL SECTION Reagents. Stock solutions (1000 ppm in acetonitrile) of 2,4,6trinitrotoluene (TNT), 1,3-dinitrobenzene (DNB), and 2,4-dinitrotoluene (DNT) were obtained from Radian International (Austin, TX). Paraoxon and methyl parathion were purchased from Supelco (Bellefonte, PA), and stock solutions (1000 ppm) were prepared in acetonitrile. Phenol was purchased form Sigma. 2-Chlorophenol (2-CP), 2,3-dichlorophenol (2,3-DCP), and 2,4-dichlorophenol (2,4-DCP) were supplied by Aldrich. Stock solutions (10 mM) of phenols were prepared daily in deionized water. Sodium dodecyl sulfate (SDS), Borax (sodium tetraborate), potassium phosphates, 2-(4-morpholinio)ethanesulfonic acid hydrate (MES), and sodium hydroxide were all obtained from Sigma. All sample solutions were prepared in the corresponding running buffer solutions. Apparatus. Details of the integrated CE-EC glass chip microsystem were described previously.7 Briefly, the homemade highvoltage power supply had an adjustable voltage range between 0 and +4000 V. The simple-cross single-separation channel glass microchip was obtained from Micralyne (model MC-BF4-001, Edmonton, Canada). The detection reservoir was cut off to facilitate the end-column electrochemical detection. The microchip had a four-way injection cross that was connected to the three reservoirs and the separation channel. The 86 × 20 × 2 mm chip had a 75-mm-long separation channel and a 5-mm-long injection channel. The channels had a maximum depth of 20 µm and a width of 50 µm at the top. Short pipet tips were inserted into the holes of the various reservoirs. A Plexiglas holder was fabricated for housing the separation chip and the detector and allowing their convenient replacement. Platinum wires, inserted into the individual reservoirs on the holder, served as contacts to the highvoltage power supply. Electrode Fabrication. The screen-printed carbon electrodes (used for comparison) were printed with a semiautomatic printer (Model TF 100, MPN, Franklin, MA) on a 33 × 10 mm ceramic substrate. A carbon ink (Acheson Colloids, Electrodag 440B, Catalog No. 49AB90, Ontario, Canada) was used for printing the electrode strips. Details of the printing process were described elsewhere.7 The dimensions of the bare screen-printed carbon and diamond band electrodes were identical (0.30 × 6.0 mm). Chemically vapor-deposited boron-doped diamond films were prepared at the Applied Chemistry Department of Tokyo University (Tokyo, Japan).16 A piece of diamond film (7 mm × 0.3 mm 936

Analytical Chemistry, Vol. 75, No. 4, February 15, 2003

× 40 µm) was glued onto a 33 × 10 mm ceramic plate by quicksetting nonconductive epoxy (Radio Shack Inc., Forth Worth, TX). The contact copper wire (with the ends scraped to remove the insulating coating) was placed on one end of the BDD band electrode. A conductive epoxy (Chemtronics, Kennesaw, GA) was subsequently applied to the contact point between the diamond and the copper wire and was cured at 100 °C for 30 min. Subsequently, a layer of the quick-setting nonconductive epoxy was applied to cover the surface of the electrode, leaving a 0.3 × 6.0 mm area available. This ensures geometrical dimensions of the diamond electrode similar to that of the screen-printed electrode. Prior to use, the diamond band electrode was sonicated in 2-propanol for 10 min. End-Column Amperometric Detection. The diamond band working electrode was placed at the channel outlet in a manner analogous to the placement of the screen-printed detector.7 The distance between the screen-printed as well as the diamond electrode surface and the channel outlet was controlled by a plastic screw and a thin-layer spacer (50 µm). Amperometric detection was performed with an Electrochemical Analyzer 620 (CH Instruments, Austin, TX) using the “amperometric i-t curve” mode. The electropherograms were recorded with a time resolution of 0.1 s (without any software filtration) while applying the detection potential. Sample injections were performed after stabilization of the baseline. All experiments were performed at room temperature. Electrophoretic Procedure. The channels of the glass chip were treated before use by rinsing with 0.1 M NaOH and deionized water for 10 min each. The running buffers for the separation of explosives, nerve agents, and phenols were 15 mM borate buffer (pH 9.2, containing 15 mM SDS), 20 mM MES (pH 5.0, containing 10 mM SDS), and a mixed solution of 10 mM borate/20 mM phosphate buffer (pH 8.0), respectively. The “buffer” reservoirs were filled with the CE running buffer solution, and the “sample” reservoir was filled with the sample. The injections were affected by applying the desired potential between the sample reservoir and the grounded-detection reservoir for 2 s, with all other reservoirs floating. Subsequently, separations were performed by switching the high-voltage contacts and applying the corresponding separation voltages to the running buffer reservoir with the detection reservoir grounded and all other reservoirs floating. Flow injection measurements of TNT were performed in borate buffer(15 mM; pH 9.2) containing no SDS by applying +1000 V to the “running buffer” reservoir. Safety Considerations. The high-voltage power supply and associated electrical connections should be handled with extreme care to avoid electrical shock. Nitroaromatic explosives, organophosphate nerve agents, and phenols are highly toxic and dangerous and should be handled with extra care and in a fume hood. Skin and eye contact and accidental inhalation or ingestion should be avoided. RESULTS AND DISCUSSION In the present work, the BDD working electrode was coupled with the micromachined CE system as an end-column electrochemical detector. The attractive performance of the diamond electrode microchip detector is indicated from its response to several groups of important analytes. Such favorable performance was compared critically, in the following sections, to a commonly used screen-printed electrode. To ensure that the improved

Figure 1. Electropherograms for phenols (A), nerve agents (B), and explosives (C) with screen-printed carbon (a) and diamond band (b) electrodes. Sample mixture A: 100 µM phenol (1), 100 µM 2-chlorophenol (2), 200 µM 2,4-dichlorophenol (3), and 200 µM 2,3-dichlorophenol (3). Sample B: 10 ppm paraoxon (1), 10 ppm methyl parathion (2). Sample C: 15 ppm DNB (1), 15 ppm TNT (2), 15 ppm DNT (3). Operation conditions: separation medium, 10 mM borate/ 20 mM phosphate buffer (pH 8.0) (A), 20 mM MES (pH 5.0, containing 10 mM SDS) (B); 15 mM borate buffer (pH 9.2, containing 15 mM SDS) (C); separation voltage, +1500 (A), + 2000 (B), and +1000 V (C); injection voltage, +1500 V (A), + 2000 (B), and +1000 V (C); injection time, 2 s; detection potential, +0.95 V (A), -0.55 V (B), and -0.70 V (C) (vs Ag/AgCl wire).

behavior is attributed solely to the new electrode material itself rather than to other factors, the following precautions were taken: (1) use of similar geometrical dimensions of both the electrodes (0.3 × 6.0 mm), (2) use of same detector configuration/ geometry, and (3) use of same electrode-capillary distance, controlled by a 50-µm thin spacer. Shown in Figure 1 are representative electropherograms for phenols (A), organophosphate nerve agents (B), and nitroaromatic explosives (C) recorded with the screen-printed carbon (a) and diamond band (b) working electrodes. In all cases, the BDD detector displays higher sensitivity and a lower noise level under the same operational conditions. Notice also the flatter baseline current of the BDD detector, particularly during the initial 20 s. The enhanced signalto-background characteristics of the BDD electrode are coupled to sharper peaks and, hence, to enhanced resolution. The halfpeak widths of phenol, 2-chlorophenol, 2,3-dichlorophenol, 2,4dichlorophenol, paraoxon, methyl parathion, DNB, TNT, and DNT at the BDD electrode are 4.0, 4.0, 5.2, 5.5, 5.5, 7.8, 4.5, 4.8, and 4.8 s, respectively (vs 6.3, 6.8, 7.9, 9.5, 5.7, 9.1, 6.4, 6.6, and 6.7 s at the screen-printed detector). The resolution of the phenol/ chlorophenol (A, 1/2) and DNB/TNT (C, 1/2) peaks are 1.01 and 1.08, respectively (vs 0.72 and 0.76 at the carbon detector). Overall, the data of Figure 1 suggest that the BDD is a very attractive electrode material for microchip CE detection. Additional advantages and a detailed characterization of the diamond electrode CE microchip detector are reported below in connection to the monitoring of nitroaromatic explosives. One such advantage is the reduced surface fouling offered by the diamond electrode detector. Figure 2 compares the stability of the response for repetitive flow injection microchip measurements of 5 ppm TNT at the diamond (a) and screen-printed carbon (b) detectors. The thick-film carbon detector displays a gradual decrease of the TNT response (with a 30% decrease and a RSD of 10.8%; n ) 60). In contrast, a highly stable signal is observed over the entire operation upon using the diamond electrode (RSD ) 0.8%). Such resistance to surface fouling reflects the negligible adsorption of organic compounds at the BDD surface. High

Figure 2. Stability of the response for repetitive flow injection measurements of 5 ppm TNT at the diamond (a) and screen-printed carbon (b) electrodes. Separation voltage, +2000 V; injection voltage +2000 V; running buffer, 15 mM borate buffer (pH 9.2) without SDS. Other conditions, as in Figure 1C(b).

Figure 3. Hydrodynamic voltammograms for 15 ppm TNT at screenprinted (a) and diamond (b) electrodes. Other conditions, as in Figure 1C(b).

resistance to passivation of BDD electrodes was demonstrated earlier in connection with conventional amperometric measurements of NADH18 and chlorophenols.19 Figure 3 depicts typical hydrodynamic voltammograms (HDV) for the reductive detection of 15 ppm TNT at the thick-film carbon (a) and diamond (b) electrode detectors. The curves were recorded pointwise by changing the potential over the -0.2 to -1.0 V range. Both detectors display similar voltammetric profiles. The cathodic response starts at -0.4 V, increasing gradually to reach maximum values at a potential of -0.7 V. The higher limiting current of the BDD detector is coupled with a slightly lower half-wave potential (-0.54 vs -0.59 V for the screen-printed one). All subsequent work employed a potential of -0.70 V that offered the most favorable signal-to-noise characteristics. A dramatic increase in the baseline current, its slope (initial rise of the electropherogram), and the corresponding noise level was observed at higher potentials. Analogous HDV for DNT displayed similar profiles and trends (not shown). The effect of the separation voltage upon the separation efficiency and amperometric response of DNB, TNT, and DNT is shown in Figure 4. As expected, increasing the separation voltage from 1000 to 4000 V (in 500-V increments) dramatically decreases the migration time for DNB, TNT, and DNT from 150.7 to 33.5, 164.7 to 36.6, and 184.0 to 41.0 s, respectively (a-f). The peak width (at half-height) of DNB, TNT, and DNT decreases from 5.0, 5.2, and 5.1 s to 1, 1.1, and 1.1 s, respectively. The plate number of DNB, TNT, and DNT increases from 5108, 5474, and 7106 (1000 V) to 6218, 6134, and 7697 (4000 V), respectively. The current response of the three explosives increases in a nearly linear fashion with the separation voltage (e.g., see inset for TNT). Notice also the flat baseline current even at high separation Analytical Chemistry, Vol. 75, No. 4, February 15, 2003

937

Figure 6. Influence of separation voltage on the peak-to-peak baseline noise for diamond (a) and screen-printed carbon (b) electrode detectors. Other conditions, as in Figure 1C(b).

Figure 4. Influence of the separation voltage on the response of the BDD detector for a mixture containing 15 ppm DNB (1), TNT (2), and DNT (3). Separation voltage, (a) +1000, (b) +1500, (c) +2000, (d) +2500, (e) +3000, (f) +3500, and (g) +4000 V. Also shown (inset) is the dependence of the TNT peak current upon the separation voltage. Other conditions, as in Figure 1C(b).

Figure 7. Calibration data for mixtures containing increasing levels of DNB (a) and DNT (b) in increments of 200 ppb (A-E). Also shown (insets) are the resulting calibration plots and electropherograms for a mixture containing 400 ppb (a) DNB and (b) DNT at the screenprinted carbon (A) and diamond (B) detector electrodes. Separation voltage, +1500 V. Other conditions, as in Figure 1C(b).

Figure 5. Influence of the separation voltage on the initial baseline of diamond (A) and screen-printed carbon (B) electrodes. Separation voltage: (a) +1000, (b) +2000, (c) +3000, and (d) +4000 V. Other conditions, as in Figure 1C(b).

voltages. A significantly larger initial baseline current slope is reported below for the screen-printed electrode. The baseline current and the noise level of the diamond electrode detector are substantially less affected by the separation voltage. Figure 5 shows the influence of separation voltage on the initial baseline current (first 25 s) of the diamond (A) and screen-printed carbon (B) electrodes. The thick-film carbon detector displays a rising initial baseline slope upon increasing the separation voltage during the first 5, 10, and 20 s for 2000, 3000, and 4000 V, respectively (B, b-d), indicating incomplete isolation from high separation voltages. The BDD detector, in contrast, exhibits a rapid stabilization of the background, with a completely flat baseline up to 3000 V (A, a-c), and only a small initial baseline slope (for the first 5 s) using 4000 V (A, d). Figure 6 displays the dependence of the peak-to-peak baseline noise for the diamond (a) and screen-printed carbon (b) electrodes upon the separation voltages. At both electrodes, the noise level increases with increasing separation voltage. However, the back938 Analytical Chemistry, Vol. 75, No. 4, February 15, 2003

ground noise at the diamond electrode is substantially less influenced by the separation voltage (with only 75% increase over the +1000 to +4000 V range, compared to 180% change for the screen-printed carbon; slopes, 19 and 100 pA/kV, respectively). Note also that the initial peak-to-peak noise levels (at +1000 V) are 97 and 170 pA for the BDD and screen-printed carbon electrodes, respectively (and that these electrodes have the same surface area). An even larger improvement in the noise level is obtained for anodic measurements in the positive potential range (e.g., Figure 1C). The diamond electrode detector displays a well-defined concentration dependence. Electropherograms for mixtures containing increasing levels of DNB (a) and DNT (b) in 200 ppb steps are shown in Figure 7 (A-E). Defined peaks, proportional to the concentration of both explosives are observed. The resulting calibration plots (shown as inset) are highly linear with sensitivities of 5.8 and 3.7 nA/ppm for TNT and DNT, respectively (correlation coefficients, 0.998 and 0.999). Also shown (as an inset) are the electropherograms of a mixture containing 400 ppb DNB and DNT at the screen-printed carbon (A) and diamond (B) electrodes. Unlike the poorly defined response of the screen-printed carbon detector at this low level, the diamond electrode responds favorably to this concentration. Detection limits of 70 ppb DNB and 110 ppb DNT can thus be estimated (based on S/N of 3). In conclusion, we have demonstrated the utility and the advantages of diamond electrode amperometric detectors for CE

microchips. The attractive properties of diamond electrodes, including their favorable signal-to-background characteristics, negligible adsorption of organic compounds, and wide potential window, make the BDD a very promising material for detection in CE microchip systems. Similar improvements are expected for additional groups of analytes, particularly those possessing high redox potentials or displaying surface fouling effects. Priliminary data on diamond electrode-based detection of aromatic amines and purines are very encouraging. The detector design allows easy and fast replacement of the diamond electrode. Other forms of diamond could expand the scope of CE microchip electrochemical detection (e.g., use of nickel-implanted BDD23 for the detection (23) Ohnishi, K.; Einagan, Y.; Notsu, H.; Terashima, C.; Rao, T. N.; Park, S. G.; Fujishima, A. Electrochem. Solid-State Lett. 2002, 5, D1.

of amino acids). While the advantages of the BDD-based electrochemical detector have been presented within the context of microchip CE systems, such a detector should be attractive for other micromachined flow analyzers. ACKNOWLEDGMENT This research was supported by grants from Oklahoma City National Memorial for Prevention of Terrorism (MIPT; Project 2002-J-A-139) and U.S. ONR (Award N00014-01-1-0213). D.S. thanks the Rotary Yoneyama Memorial Foundation Inc. of Japan for an YD Scholarship. Received for review October 7, 2002. Accepted December 13, 2002. AC0262053

Analytical Chemistry, Vol. 75, No. 4, February 15, 2003

939