Coulometric Detection of Components in Liquid Plugs by

Anal. Chem. 2010, 82, 8725–8732

Coulometric Detection of Components in Liquid Plugs by Microfabricated Flow Channel and Electrode Structures Fumihiro Sassa,† Hind Laghzali,‡ Junji Fukuda,† and Hiroaki Suzuki*,† Graduate School of Pure and Applied Sciences, University of Tsukuba, 1-1-1 Tennodai, Tsukuba, Ibaraki 305-8573, Japan, and Department of Information Technologies for Health Care, Institute for Science and Technology, Joseph Fourier University, BP 53-38041 Grenoble Cedex 9, France Coulometry has been demonstrated to be effective for determining the analyte in a liquid plug on the nanoliterscale confined in a flow channel. A plug prepared in a rhombus structure of an auxiliary flow channel was placed on a thin-film three-electrode system, and hydrogen peroxide was detected as a model analyte. Under a fixed potential, the current decayed rapidly, particularly in shallow flow channels, thus making reproducible amperometric detection difficult. On the other hand, the increasing charge during coulometry facilitated the measurements. A constricted flow channel structure with an array of platinum strips for the working electrode was effective at efficiently consuming the analyte to improve the sensitivity and lower the detection limit. Compared to the case of a single short working electrode with the same area, a 4-fold increase in sensitivity was observed for the best combination of flow channel height and interstrip distance. With an increase in the generated current while maintaining the background at a low level, the detection limit was lowered from 1.3 µM to 410 nM using working electrodes with the same area. Furthermore, the processing of solutions containing L-glutamate or L-glutamate oxidase and the detection of L-glutamate were demonstrated. Progress in the development of the micro total analysis system (µTAS) and lab-on-a-chip technologies has produced a rapid expansion in applications. Following this trend, the rapid and efficient processing of many different solutions has emerged as a critical issue. To achieve this goal, microfluidic devices that handle liquid plugs in a network of flow channels have been proposed as a promising processing technique.1 Plug-based devices have some compelling advantages. Unlike continuous fluids, the volume of solutions being handled can be reduced significantly, which is particularly advantageous in applications for chemical analyses that use a very small amount of sample solution and/or very expensive reagents. The capability to handle many solutions in a limited space also opens up a new way to realize large-scale * To whom correspondence should be addressed. Phone: +81-29-853-5598. Fax: +81-29-853-4490. E-mail: [email protected]. † University of Tsukuba. ‡ Joseph Fourier University. (1) Song, H.; Chen, D. L.; Ismagilov, R. F. Angew. Chem., Int. Ed. 2006, 45, 7336–7356. 10.1021/ac102289a  2010 American Chemical Society Published on Web 09/21/2010

integration.2 Furthermore, the rapid mixing of components is a promising feature1,3 that contrasts the mixing of components between continuous laminar flows using special structures.4-9 Although the devices are versatile, analyte detection is required in the final stage of processing when chemical analysis is a final goal. For the plug-based systems, however, the development and improvement of a sensing methodology is lagging. Fluorometry has been used predominantly for analyte detection in previous plug-based microsystems.10-13 However, with the optical method, instruments to carry out the procedure tend to be bulky and complicated, particularly when the number of sensing sites is increased. Thus, electrochemical approaches are more advantageous.14,15 Electrochemical sensors can achieve highly sensitive detection with relatively simple external instruments. Furthermore, they facilitate miniaturization and integration on a single chip, which provides additional advantages to microfluidic technology. Nevertheless, applications to electrochemical sensing for plug-based systems have been very rare until now.16 Along with detection by conventional instruments for flow injection analysis, chromatography, and electrophoresis,17-20 analyte detection in microfabricated electrochemical devices has (2) Thorsen, T.; Maerkl, S. J.; Quake, S. R. Science 2002, 298, 580–584. (3) Song, H.; Tice, J. D.; Ismagilov, R. F. Angew. Chem., Int. Ed. 2003, 42, 768–772. (4) Bessoth, F. G.; deMello, A. J.; Manz, A. Anal. Commun. 1999, 36, 213– 215. (5) Liu, R. H.; Stremler, M. A.; Sharp, K. V.; Olsen, M. G.; Santiago, J. G.; Adrian, R. J.; Aref, H.; Beebe, D. J. J. Microelectromech. Syst. 2000, 9, 190– 197. (6) Johnson, T. J.; Ross, D.; Locascio, L. E. Anal. Chem. 2002, 74, 45–51. (7) Stroock, A. D.; Dertinger, S. K. W.; Ajdari, A.; Mezic´, I.; Stone, H. A.; Whitesides, G. M. Science 2002, 295, 647–651. (8) Kim, D. J.; Oh, H. J.; Park, T. H.; Choo, J. B.; Lee, S. H. Analyst 2005, 130, 293–298. (9) Sudarsan, A. P.; Ugaz, V. M. Proc. Natl. Acad. Sci. U.S.A. 2006, 103, 7228– 7233. (10) Linder, V.; Sia, S. K.; Whitesides, G. M. Anal. Chem. 2005, 77, 64–71. (11) Huebner, A.; Srisa-Art, M.; Holt, D.; Abell, C.; Hollfelder, F.; deMello, A. J.; Edel, J. B. Chem. Commun. 2007, 1218–1220. (12) Sassa, F.; Fukuda, J.; Suzuki, H. Anal. Chem. 2008, 80, 6206–6213. (13) Chen, D.; Du, W.; Liu, Y.; Liu, W.; Kuznetsov, A.; Mendez, F. E.; Philipson, L. H.; Ismagilov, R. F. Proc. Natl. Acad. Sci. U.S.A. 2008, 105, 16843– 16848. (14) Suzuki, H.; Upadhyay, S.; Loughran, M. In Encyclopedia of Sensors; Grimes, C. A., Dickey, E. C., Pishko, M. V., Eds.; American Scientific Publishers: Stevenson Ranch, CA, 2006; Vol. 6, pp 161-187. (15) Sassa, F.; Morimoto, K.; Satoh, W.; Suzuki, H. Electrophoresis 2008, 29, 1787–1800. (16) Han, Z.; Li, W.; Huang, Y.; Zheng, B. Anal. Chem. 2009, 81, 5840–5845. (17) Kissinger, P. T. Anal. Chem. 1977, 49, 447A–456A.

Analytical Chemistry, Vol. 82, No. 20, October 15, 2010

8725

mostly been based on amperometry.21-26 To improve sensitivity, a direct strategy is to make the working electrode as large as possible in the restricted space.27,28 However, achieving highly sensitive reliable amperometric determination becomes increasingly difficult with decreasing sample and reagent volumes, particularly for stationary fluids, because of the rapid depletion of the analyte in the vicinity of the electrode. To avoid this problem, regeneration of the analyte by redox cycling has been used in solutions on the nanoliter scale or lower.29-35 However, this method is not applicable to general cases. Thus, coulometry can provide an effective solution.36,37 In coulometry, an output signal is provided in the form of a charge, which is the integral of the current. In other words, the information for the analyte collected during a predetermined time period is obtained not in terms of the concentration but of the amount of analyte. Therefore, in an ideal case where all of the analyte is consumed in the electrode reaction, calibration is not necessary. This is a great advantage because the method of calibration chosen often causes controversy during the inspection of batch-fabricated microdevices, particularly those designed for single use. However, despite emerging technology for the plug-based microfluidic systems, no detailed investigation on the applicability of coulometry has been carried out yet. In this study, the applicability of coulometry to the analysis of components in nanoliter-scale liquid plugs was investigated. Hydrogen peroxide was used as a test analyte because it poses problems in many fields and its combination with enzymes (oxidases) expands the application to the detection of various biomolecules ranging from small molecules to proteins. A systematic investigation revealed that rapid collection of an analyte on a working electrode is not necessarily easy, even in small liquid plugs. Structural modifications to squeeze a liquid plug of a (18) Weber, S. G.; Purdy, W. C. Ind. Eng. Chem. Prod. Res. Dev. 1981, 20, 593–598. (19) Baldwin, R. P. Electrophoresis 2000, 21, 4017–4028. (20) Lacher, N. A.; Garrison, K. E.; Martin, R. S.; Lunte, S. M. Electrophoresis 2001, 22, 2526–2536. (21) Jobst, G.; Moser, I.; Svasek, P.; Varahram, M.; Trajanoski, Z.; Wach, P.; Kotanko, P.; Skrabal, F.; Urban, G. Sens. Actuators, B 1997, 43, 121–125. (22) Woolley, A. T.; Lao, K.; Glazer, A. N.; Mathies, R. A. Anal. Chem. 1998, 70, 684–688. (23) Nagy, G.; Xu, C. X.; Buck, R. P.; Lindner, E.; Neuman, M. R. Anal. Chem. 1998, 70, 2156–2162. (24) Liu, Z.; Niwa, O.; Kurita, R.; Horiuchi, T. J. Chromatogr., A 2000, 891, 149–156. (25) Perdomo, J.; Hinkers, H.; Sundermeier, C.; Seifert, W.; Morell, O. M.; Knoll, M. Biosens. Bioelectron. 2000, 15, 515–522. (26) Pemberton, R. M.; Hart, J. P.; Mottram, T. T. Biosens. Bioelectron. 2001, 16, 715–723. (27) Lee, K.-H.; Ishikawa, T.; McNiven, S.; Nomura, Y.; Sasaki, S.; Arikawa, Y.; Karube, I. Anal. Chim. Acta 1999, 386, 211–220. (28) Lee, K.-H.; Ishikawa, T.; Sasaki, S.; Arikawa, Y.; Karube, I. Electroanalysis 1999, 11, 1172–1179. (29) Bowyer, W. J.; Clark, M. E.; Ingram, J. L. Anal. Chem. 1992, 64, 459–462. (30) Clark, R. A.; Hietpas, P. B.; Ewing, A. G. Anal. Chem. 1997, 69, 259–263. (31) Kashyap, R.; Gratzl, M. Anal. Chem. 1998, 70, 1468–1476. (32) Cai, X.; Klauke, N.; Glidle, A.; Cobbold, P.; Smith, G. L.; Cooper, J. M. Anal. Chem. 2002, 74, 909–914. (33) Troyer, K. P.; Wightman, R. M. Anal. Chem. 2002, 74, 5370–5375. (34) Sun, P.; Mirkin, M. V. J. Am. Chem. Soc. 2008, 130, 8241–8250. (35) Wolfrum, B.; Zevenbergen, M.; Lemay, S. Anal. Chem. 2008, 80, 972– 977. (36) Bard, A. J.; Faulkner, L. R. Electrochemical Methods: Fundamentals and Applications; John Wiley & Sons: New York, 1980. (37) Heineman, W. R.; Kissinger, P. T. In Laboratory Techniques in Electroanalytical Chemistry; Kissinger, P. T., Heineman, W. R., Eds.; Marcel Dekker: New York, 1996; pp 51-125.

8726


relatively large volume into a thin flow channel improved the detection sensitivity remarkably. The background charge was reduced significantly by using a working electrode consisting of an array of thin electrodes. In addition, we used a unique flow channel structure to carry out precise volume measurements. To demonstrate the applicability of our findings to biochemical analyses, L-glutamate detection was carried out by handling necessary plugs of solutions on the chip. EXPERIMENTAL SECTION Reagents and Materials. The following reagents and materials were obtained for the fabrication and characterization of the devices: glass wafers (no. 7740, 3 in., 500 µm thick) from Corning Japan (Tokyo, Japan); a thick film photoresist SU-8 from MicroChem (Newton, MA); precursor solution of polydimethylsiloxane (PDMS) KE-1300T from Shin-Etsu Chemical (Tokyo, Japan); hydrogen peroxide (30%) from Wako (Osaka, Japan); food dyes Red No. 102, Blue No. 1, and Yellow No. 4 from Kyoritsu Foods (Tokyo, Japan); precursor solution of polyimide Semicofine SP341 from Toray (Tokyo, Japan); and L-glutamate oxidase (L-GlOx, EC 1.4.3.11 from Streptomyces sp., 330 U mL-1) from Yamasa (Chiba, Japan). All solutions were prepared with distilled deionized water. The concentration of hydrogen peroxide was checked using a calibrated refractometer specifically designed for determining hydrogen peroxide (PAL-39S, Atago, Tokyo, Japan). Fabrication of Devices. Coulometric analysis of components in liquid plugs was carried out using the devices shown in Figure 1. Patterns for a thin-film three-electrode system were formed on a glass substrate (11 mm × 30 mm), and the main and auxiliary flow channels were formed with PDMS. We used two types of electrode systems. In the type I system, thin strips of working, reference, and auxiliary electrodes were formed in a flow channel, as shown in Figure 1A,C (top). The active area for the working electrode was 600 µm × 400 µm. The main flow channel was straight in all regions, including the sensing region. The width of the main flow channel was 600 µm, and the height was adjusted between 60 and 520 µm. For the type II system, we used another device, as illustrated in Figure 1B,C (bottom). In this device, part of the flow channel over the electrode area was made shallower (height, 40 or 80 µm). The width was the same as that of the type I device (600 µm). The working electrode was separated into an array of thin strips. Devices with working electrodes of different widths and interelectrode distances were prepared (Table 1). The total area of the working electrode was the same as that of the type I device (0.24 mm2). For comparison, a device with a large single working electrode was also fabricated. The reference and auxiliary electrodes were the same as those of the type I device. For both types of devices, the auxiliary flow channel consisting of a row of rhombuses was used to measure the plug volumes. The narrowest and widest portions of each rhombus were 60 and 500 µm, respectively. The length was 700 µm. The height of the channel was the same as that of the main flow channel. Silicone tubes (inner diameter, 500 µm) were connected to the inlet of the main flow channel and the auxiliary flow channel through stainless pipes (inner diameter, 400 µm) attached to the inlets. Microsyringe pumps were connected to the tubes.

Figure 1. Devices used for analysis: (A) type I device; (B) type II device; (C) magnified views of the electrode area; only part of the W.E. is shown for the type II device; and (D) cross-section of the electrode area of the type II device. (E) Electrodes used in the type II device: W.E., working electrode; R.E., reference electrode; A.E., auxiliary electrode. Table 1. Working Electrodes Used in the Type II Device height of the thin flow area total number width of interstrip channel covered electrode area of a strip distancea region by the no. (mm2) strips (µm) (µm) (µm) W.E. (µm) 1 2 3 4 5 6 7 8 9 10 a

0.24 0.24 0.24 0.24 6.72 0.24 0.24 0.24 0.24 3.36

5 10 20 40 1 5 10 20 40 1

80 40 20 10 11 200 80 40 20 10 5 600

2160 1080 540 270 1040 520 260 130

40 40 40 40 40 80 80 80 80 80

11 200 11 200 11 200 11 200 11 200 5 600 5 600 5 600 5 600 5 600

Edge to edge.

The electrode patterns were formed by a thin-film process that included metal deposition by sputtering and patterning by a liftoff process. After patterning a positive photoresist layer, a 200 nm thick platinum layer was deposited first with a 30 nm thick titanium intermediate layer. The substrate was immersed in acetone to remove the metal layers on the photoresist. This formed the patterns for the working and auxiliary electrodes. The platinum layer was also used as the base layer for the reference electrode. A 650 nm thick silver layer was then formed only on the reference electrode area by the same lift-off process. A 2.5 µm thick polyimide layer was formed to delineate the active areas for the electrodes and pad areas. To form the polyimide layer, a positive photoresist layer was patterned on the layer of polyimide. During this process, the polyimide layer was also etched by the developer solution of the photoresist. The photoresist was then

removed in ethanol. Finally, the polyimide patterns were cured at a final temperature of 280 °C. The silver pattern was also covered with the polyimide layer except for two pinholes that were 30 µm in diameter. Silver chloride was grown from there by applying a current of 5 nA for 10 min in a 0.1 M KCl solution.38 The growth of the silver chloride layer and the potential were monitored simultaneously using a three-electrode configuration with a platinum plate auxiliary electrode and commercial Ag/AgCl reference electrode (2080A-06T, Horiba, Tokyo, Japan). The flow channel structure was formed by replica molding.39 A template was formed with the thick-film photoresist (SU-8), and the precursor solution of PDMS was poured onto the template. After PDMS was cured, it was removed from the template. Flow channel patterns were thus obtained. To form the structure with the shallow region, patterning with the thick-film photoresist was repeated to make the template. On-Chip Preparation of Liquid Plugs. To handle plugs in the flow channels, positive or negative air pressure was applied using two syringe pumps connected to the ends of the main and auxiliary flow channels (Figure 1A, B). Operations such as attaching, dividing, sorting, and forming plugs of various lengths can be conducted by using the pumps and following an appropriate program. In the coulometric analysis of components in liquid plugs, the measurement precision for the volume of the sample and reagent solutions influences the precision of the obtained data. The rhombus structure of the auxiliary flow channel facilitates the measurement. First, a relatively long plug is introduced into (38) Suzuki, H.; Taura, T. J. Electrochem. Soc. 2001, 148, E468–E474. (39) Duffy, D. C.; McDonald, J. C.; Schueller, O. J. A.; Whitesides, G. M. Anal. Chem. 1998, 70, 4974–4984.


8727

Figure 2. Volume measurement in the auxiliary flow channel: (A) procedure for measuring the plug volume, (B) change in free energy accompanying the movement of the meniscus in the rhombus structure, and (C) plugs formed in the auxiliary flow channel using 2-6 rhombuses and returned to the main flow channel.

the main flow channel using the pump connected at the end. Part of the plug is withdrawn into the first rhombus by using the pump connected to the auxiliary flow channel (Figure 2A(1)). Once the meniscus passes the widest part, the solution moves spontaneously into the narrowing region and stops at the node (Figure 2A(2, 3)). The timing to switch off the pump is sensed by visual inspection or by measuring the length of time after the introduction of the solution. When plugs of larger volumes were needed, the step is repeated to fill more rhombuses (Figure 2A(4)). After the volume measurement, the solution in the main flow channel is separated (Figure 2A(5)) and the part of the solution left in the auxiliary flow channel is returned to the main flow channel (Figure 2A(6)). The volume of a single rhombus is 0.2h and has a range of 8-100 nL, where h is the height of the flow channel and has a range of 40-520 µm. In the following experiments, the volume of plugs placed in the sensing area corresponds to the volume of four rhombuses unless otherwise noted. The spontaneous movement of the solution in the rhombus structure can be explained by the change in free energy. Suppose that a plug is first placed in the main flow channel and that part of the solution is withdrawn into the auxiliary flow channel (Figure 2B). The free energy change can be calculated as the sum of contributions originating from interfacial tensions between the solid-liquid, liquid-air, and air-solid interfaces, which are directly related to the changes in the areas of the interfaces. Because the height of the main and auxiliary flow channels is the same everywhere and the total volume of the plug is constant, no changes accompany the formation and disappearance of the solid-liquid and air-solid interfaces at the top and bottom of the flow channels. In addition, there are no contributions from the two menisci (liquid-air interfaces) in the main flow channel if they remain there. On the other hand, there are changes accompanying the formation and disappearance of the liquid-air interface (meniscus) in the auxiliary flow channel and the solid-liquid and air-solid interfaces at the side walls of the flow channels. If the liquid-air interface is assumed to be flat, the dependence of the free energy on the location of the meniscus in the auxiliary flow channel is described as in Figure 2B. In the oscillating change, local maxima and minimums appear at the widest and narrowest portions of the rhombuses, respectively. 8728


Therefore, once the liquid plug passes the widest portion by applying pressure using the pump, the rest of the rhombus is filled with the solution spontaneously and the meniscus stops at the narrowest portion. Measurement Procedure. When conducting coulometric analyses on the chip, a solution containing hydrogen peroxide was first introduced into the system in the form of a relatively long liquid plug (∼500 nL). A plug of an appropriate length was then formed using the auxiliary flow channel. The plug was then transported to the sensing area. Although flow channel regions other than the sensing area were hydrophobic, the bottom of the sensing area was hydrophilic. Therefore, the plug introduced into the middle of the sensing area spread into the entire area spontaneously by capillary action and stopped at the end. Consequently, the position of the plug was precisely aligned. Electrochemical measurements were carried out with a stationary plug using an Autolab PGSTAT12 (Eco Chemie, Utrecht, The Netherlands). The measured current was integrated between 0 s and a predetermined time. After each measurement, the sensing area was rinsed with a cleansing plug, and the next measurement was carried out with a fresh plug. To analyze the enzymatic reaction of L-glutamate oxidase, plugs with different components were formed in the same manner and the components were mixed at an appropriate ratio. The mixed plug was transported to the sensing area, and hydrogen peroxide produced by the enzymatic reaction was measured. The other procedures were the same as that used for hydrogen peroxide. Prior to each measurement, the working electrode was cleaned by applying alternating voltages of +0.7 and -0.7 V (vs on-chip Ag/AgCl) several times. The experiments were carried out at room temperature. RESULTS AND DISCUSSION Preparation of Plugs Using the Auxiliary Flow Channel. The chained rhombus structure in the auxiliary flow channel facilitated precise measurement of liquid plug volumes on the nanoliter scale. Since the dimensions of the structures in the auxiliary flow channels were smaller than those of the main flow channel, a solution introduced there spread farther from the junction. We used a straight auxiliary flow channel in a previous study.12 However, because of the straightness, precise volume

Table 2. Volume of Plugs Measured in the Rhombus Structure of the Auxiliary Flow Channel no. of used rhombuses averaged volume (nL) relative standard deviation (%) no. of plugs in the main flow channel

2

3

4

5

6

200

307

408

511

618

1.86

1.15

1.20

1.15

0.98

8

7

6

5

4

measurement required careful adjustment of the position of the meniscus. In contrast, for the present device, precise digital style measurement was realized by simply measuring the number of rhombuses filled with the solution. To evaluate the precision of the volume measurement, plugs were formed using rhombuses in the auxiliary flow channel and placed in the main flow channel. Figure 2C shows rows of plugs of the same volume formed in the main flow channel using the rhombuses. Average volumes and relative standard deviations are summarized in Table 2 (plugs formed using a single rhombus were too thin to move in the main flow channel without collapsing). Although the number of measured plugs depended on their length in the main flow channel, the relative standard deviation was within 1.9% even in the worst case, demonstrating that plug volumes can be measured with reproducible and precise results using our rhombus structure and mechanism. Influence of Flow Channel Height on the Type I Device Response. The concentration profile of the analyte in the vicinity of the electrode is influenced by the height of the flow channel. Unlike conventional thin-layer cells in which only one dimension (height) is usually on a micrometer-scale, plugs have at least two dimensions that are that small. Therefore, with a constant applied potential, current decays rapidly in shallow flow channels. The influence was first examined by using the type I device. Figure 3 shows the time courses of the current recorded in flow channels of different heights. In all cases, the current decreased monotonically. With a 60 µm high flow channel, current decayed more rapidly within 10 s. The result suggests that amperometric detection becomes extremely difficult for plug analysis, particularly in shallower flow channels. The current increased with increasing height. However, as the flow channel became deeper, the curves tended to converge and a substantial current flow was observed

Figure 3. Time courses of the generated current in flow channels of various heights for the type I device. The values at the curves indicate the height. H2O2 concentration: 1 mM.

Figure 4. Coulometric detection using the type I device: (A) time courses of the generated charge; background charges are not subtracted; (B) ratio of consumption of hydrogen peroxide in a plug 10 (0), 30 (2), and 60 s (O) after applying the potential. H2O2 concentration: 1 mM. Each data point represents the average of five runs; the error bars are the standard deviations.

even after 60 s. This implies that the volume of the plug was sufficiently large with respect to the expansion of the diffusion layer extending from the working electrode and that the influence of the channel ceiling was no longer significant. To analyze this point in more detail, consumption of the analyte was examined by coulometric analysis. Figure 4A shows time courses of the charge obtained from the data in Figure 3. In contrast to the change in current, charge increased gradually, facilitating the measurement. Figure 4B shows the percentage of hydrogen peroxide collected 10, 30, and 60 s after applying the potential calculated relative to the total hydrogen peroxide that should have existed in the plug. The latter was obtained from the given concentration and volume of the plug. Furthermore, the background charge measured using a blank plug without hydrogen peroxide was subtracted from each measured value of charge. Overall, the collection efficiency decreased monotonically as the flow channel became deeper. Even for the shallowest flow channel, not all of the analyte was consumed after 60 s. A possible reason is that the plug also spread on the reference and auxiliary electrodes, and substantial time was necessary to collect the analyte from there. The obtained results suggest that hydrogen peroxide is collected more rapidly and efficiently in a shallower flow channel. Unlike ordinary thin-layer cells in which a solution is constantly flowing, excellent sensitivity and lower detection limit cannot be achieved by only decreasing the depth of the flow channel. On Analytical Chemistry, Vol. 82, No. 20, October 15, 2010

8729

the other hand, the results suggest that a substantial proportion of the hydrogen peroxide was wasted without being used for detection in deeper flow channels, even for plugs on the nanoliter scale. However, they also imply that the detection sensitivity and the lower detection limit can be improved further if the wasted hydrogen peroxide is collected more efficiently. Coulometric Detection Using the Type II Device. To examine the point raised in the previous experiments, we used the type II device with a constricted electrode region (Figure 1B). To accommodate a plug of the same volume as in the type I device, the constricted region was made longer. In addition, the working electrode was divided into an array of thin strips (Figure 1E). The total working electrode area was the same as that of the type I device. The reference and auxiliary electrodes were also the same as those of the type I device. Eight types of working electrodes with strips of different widths and interstrip distances were used (Table 1). Single large working electrodes that covered the entire working electrode region were also used for comparison (Table 1, no. 5 and no. 10). A plug of 420 nL was transported to the sensing area in the same manner as the experiment using the type I device with the 520 µm deep flow channel. First, we used devices with a 40 µm (height) × 12.5 mm (length) constricted region (Table 1, no. 1 to no. 5). When the plug was squeezed into the constricted region, the menisci of the liquid plug barely covered the entire working electrode region and the reference and auxiliary electrodes. When analyses are conducted in a stretched liquid plug, one problem that occurs is the Ohmic drop.40 In our microfabricated device, the electrodes were obliged to form as patterns on a planar surface. However, with this configuration, the current flows along the longitudinal direction of the flow channel. As a result, the potential of the part of the working electrode nearest to the reference electrode is close to the potential applied by the potentiostat, whereas the potential of the part in a more distant region may be insufficient to oxidize hydrogen peroxide. With the type I device, no substantial change was observed between solutions containing 10 mM and 0.1 M KCl as an electrolyte. However, with the type II device using a 40 µm high constricted region, the charge observed with 10 mM KCl was 51% of that with 0.1 M KCl. Therefore, the KCl concentration was increased to 0.1 M for the evaluation of the type II device. Figure 5A shows the time courses of the generated charge observed using devices no. 1 to no. 5. When a single large working electrode (no. 5) was used, a significant increase in charge was observed in comparison with the type I device. On the other hand, for type II devices no. 1 to no. 4, the current increased with the number of electrode strips with an accompanying decrease in the interstrip distance. Furthermore, the background current remained at the level observed with the type I device. Nevertheless, in type II devices no. 1 to no. 4, the smallest interstrip distance (edge to edge) was 280 µm, suggesting difficulty in collecting H2O2 efficiently. Therefore, we prepared other devices with a higher and shorter constricted region (no. 6 to no. 10). The length of the working electrode area was halved (5.6 mm), and the height of the constricted region was (40) Susan, M.; Lunte, C. E. L.; Kissinger, P. T. In Laboratory Techniques in Electroanalytical Chemistry; Kissinger, P. T., Heineman, W. R., Eds.; Marcel Dekker: New York, 1996; pp 813-853.

8730


Figure 5. Changes in the generated charge obtained with the type I and type II devices: (A) time courses of the charge obtained using the type I (flow channel height, 520 µm) and type II (no. 1 to no. 5) devices; (B) time courses of the charge obtained using the type II (no. 6 to no. 10) devices. H2O2 concentration: 1 mM. Each value includes the background charge. (C) Charge measured after 60 s using the type I and II devices (see Table 1) with solutions containing 1 mM H2O2 and no H2O2. Each data point represents the average of five runs.

doubled (80 µm). Figure 5B shows the time courses of the generated charge observed using the devices. With regard to the results obtained with devices no. 5 and no. 10 using a long single working electrode, the devices with shallower (i.e., longer) working electrode regions should generate larger charge. However, the current observed using device no. 10 with a smaller single working electrode was much larger than that of device no. 5. This result suggests that the Ohmic drop was not eliminated

Figure 6. (A) Dependence of the charge at 60 s on the concentration of H2O2 obtained using the type I (flow channel height, 520 µm) (b) and type II (no. 9) (O) devices. The inset shows the change in the lower concentration range. Each data point represents the average of five runs. Error bars represent standard deviations. (B) Time courses of the background charge in the electrolyte solution without H2O2 and their standard deviations observed using the type I (flow channel height, 520 µm) (b), type II (no. 9) (O), and type II (no. 10) (4) devices. Each data point represents the average of ten runs. Error bars represent standard deviations.

completely in device no. 5 even with 0.1 M KCl. On the other hand, charges generated using the type II devices with the shorter working electrode region were the same with 10 mM and 0.1 M KCl. The results suggest that there is a limit to the length of the working electrode area. Figure 5C compares charges recorded after 60 s with 1 mM H2O2 and without H2O2. The effect of the flow channel and electrode structures is evident. Calibration plots for hydrogen peroxide are shown in Figure 6A, where the type I device with the 520 µm high flow channel and the type II device that provided the best data (no. 9) were compared for the same plug volume (420 nL). In both cases, good reproducibility and linearity were confirmed for all examined concentrations (type I, R2 ) 0.996; type II (no. 9), R2 ) 0.998). Compared with the type I device, a 4-fold increase in sensitivity was achieved with the type II device in the examined range of concentrations. Another improvement was shown in the lower concentration region. The background current was approximately the same in the two cases. However, because of the improved sensitivity, the type II device also lowered the detection limit. To consider this point in further detail, time courses of the background charge obtained with the type I and type II devices (no. 9 and no. 10) are shown in Figure 6B. Changes observed using the type I and type II (no. 9) devices were close, and the background charge increased gradually.

Figure 6B also shows the averages of the standard deviations (n ) 10) at some points to show the distribution of the background charge. The standard deviation of the signal from the background can be related to the noise.41 If the detection limit is defined as a concentration separated by 3σ from the average of the background charge, the detectable concentration was approximately 1.3 µM and 410 nM for the types I and II (no. 9) devices, respectively. Although the application of a working electrode of a large area may promote the collection of the analyte (type II (no. 10)), it is not necessarily advantageous in improving the detection limit. With the type II (no. 10) device, the lower detection limit was 2.1 µM, which is substantially higher than that of the type II (no. 9) device. On the basis of the results, the way to improve the sensitivity and lower the detection limit is evident. The Faradaic current (or sensitivity) can be enhanced by increasing the number of strips and decreasing the interstrip distance with a given working electrode area. On the other hand, the background current depends on the total working electrode area. Therefore, since the charge is obtained by integrating the current with respect to time, a strategy to ameliorate the lower detection limit is to make the Faradaic current dominant against the background current and collect the analyte as rapidly as possible in a shorter time. Ultimately, with a minimum area for the strips and total working electrode, the interstrip distance should be minimized to the limit at which overlapping of the diffusion layers is barely avoided.42 With the shrinking diffusion layer, the height of the constricted region of the flow channel should be optimized by taking the Ohmic drop into consideration. To activate the electrode surface and enhance the sensitivity to hydrogen peroxide further, nanoporous or mesoporous platinum electrodes may have an additional effect.43,44 In our present device, the reference and auxiliary electrodes were formed in the constricted region. This electrode layout produces unwanted dead volume over the reference and auxiliary electrodes, which delays collection of the analyte. Separation of this region from the main flow channel will accelerate analyte collection. The reusability of the device was examined by repeating the experiment using the same type II device (no. 9) and a standard solution containing 100 µM H2O2. After 30 iterations, the charge measured after 60 s decreased slightly by 4.7%, which may be due to the slight damage to the thin-film reference electrode45 and contamination of the working electrode. Scattering of the measured charge was within 4.0% of the regression line, demonstrating reproducible measurement. Application to Biochemical Analysis. A device that detects hydrogen peroxide in a liquid plug can be used to analyze substrates of various oxidases. As a demonstration, hydrogen peroxide produced by the enzymatic reaction of L-glutamate oxidase was detected using our device. Here, the type II device (41) Morgan, D. M.; Weber, S. G. Anal. Chem. 1984, 56, 2560–2567. (42) Naka, K.; Hayashi, H.; Senda, M.; Shiraishi, H.; Konishi, S. Jpn. J. Appl. Phys. 2009, 48, 037001. (43) Evans, S. A. G.; Elliott, J. M.; Andrews, L. M.; Bartlett, P. N.; Doyle, P. J.; Denuault, G. Anal. Chem. 2002, 74, 1322–1326. (44) Han, J. H.; Boo, H.; Park, S.; Chung, T. D. Electrochim. Acta 2006, 52, 1788–1791. (45) Suzuki, H.; Hiratsuka, A.; Sasaki, S.; Karube, I. Sens. Actuators, B 1998, 46, 104–113.


8731

peroxide (R2 ) 0.991). The results demonstrate that our device can be used for the analysis of enzymatic reactions. L-glutamate is a representative neurotransmitter. Only a small volume of real samples needs to be drawn into the device to obtain the concentration of an analyte with a predetermined time interval. The sensing region can be used repeatedly for many different samples. To this end, a plug of a cleansing buffer solution was introduced after flushing the used plugs in the main flow channel. Additional auxiliary flow channels will facilitate this step along with the other plug-handling steps.

Figure 7. Application to the detection of L-glutamate. (A) Procedure to handle plugs and detect L-glutamate: (1-4) steps to measure the volume and form a plug of L-glutamate, (5-7) steps to measure the volume, form a plug of L-GlOx, and merge the plug with the L-glutamate plug, (8) coulometric detection on the electrodes. (B) Dependence of the charge on the concentration of L-glutamate obtained using the type II (no. 9) device. Each point represents the average of five runs. Error bars are the standard deviations. The inset shows the time courses of the generated charge. Numerals in the inset indicate the concentration of L-glutamate (micromolar).

that showed the largest response (no. 9) was used. Figure 7A shows the procedure to process liquid plugs. First, necessary solutions containing the analyte or enzyme were introduced into the main flow channel. Plugs of a predetermined volume (210 nL) were then prepared using the auxiliary flow channel and merged to mix the components following the procedure mentioned elsewhere.12 The plug was mobilized to the sensing area immediately, and the concentration of hydrogen peroxide produced by the enzymatic reaction was measured. Figure 7B shows the dependence of the charge on the concentration of L-glutamate. A good linear relationship was observed between the charge and concentration of L-glutamate just as in the case for hydrogen

8732


CONCLUSIONS For analysis of components dissolved in liquid plugs on the nanoliter scale, data acquisition by amperometry is not easy because of the rapidly and continuously decaying current. This is particularly true when the flow channel is shallow. In this respect, coulometric detection is advantageous. To improve the detection sensitivity, squeezing a plug into a constricted flow channel and detecting compounds using a working electrode consisting of an array of strips has proven to be effective. With thinner strips and appropriate interstrip distance, the collection efficiency of the analyte increases while maintaining a low background charge. The device can also be used for analyzing compounds that are oxidized enzymatically when accompanying the production of hydrogen peroxide. As mentioned earlier, the advantages of our plug-based system include a significantly reduced volume of solutions, efficient processing of many solutions in a device, rapid mixing of components, and reproducible precise measurements of plug volume. In addition, an advantage of electrochemical detectors is easy integration of multiple detectors. These features are particularly beneficial for applications in which sample solutions of very small volumes or very expensive reagents must be handled, which are often encountered in the examination of living cells, including the detection of neurotransmitters, or changes in the environment around a single cell. Furthermore, the technique is also useful for basic research, including the establishment of the number of electrons involved in an electrode reaction to elucidate the mechanism,36,37 as has been obtained by the conventional thinlayer cell technique. ACKNOWLEDGMENT This study was partially supported by a Grant-in-Aid for Scientific Research (B) (Grant No. 19360369) and a Grant-in Aid for JSPS Fellows under the Japan Society for the Promotion of Science. Received for review April 22, 2010. Accepted September 2, 2010. AC102289A

Coulometric Detection of Components in Liquid Plugs by

Recommend Documents