Autonomous Microfluidic Control by Chemically Actuated Micropumps

Jul 29, 2010 - Elliot J. Smith , Wang Xi , Denys Makarov , Ingolf Mönch , Stefan Harazim , Vladimir A. Bolaños Quiñones , Christine K. Schmidt , Yongf...
0 downloads 0 Views 2MB Size
Download PDF
Anal. Chem. 2010, 82, 6870–6876

Autonomous Microfluidic Control by Chemically Actuated Micropumps and Its Application to Chemical Analyses Atsushi Takashima, Kenichi Kojima, and Hiroaki Suzuki* Graduate School of Pure and Applied Sciences, University of Tsukuba, 1-1-1 Tennodai, Tsukuba, Ibaraki 305-8573, Japan Autonomous control of microfluidic transport was realized through the use of chemically actuated diaphragm micropumps connected to a network of controlling flow channels. A hydrogen peroxide (H2O2) solution was transported in the controlling flow channel by capillary action. Upon the solution’s arrival at the lower compartment of a micropump filled with manganese dioxide (MnO2) powder, a volume change that accompanied the production of oxygen caused by the catalytic decomposition of H2O2 induced inflation of the diaphragm. This in turn caused the movement of a solution in another network of flow channels formed in the upper layer. Micropumps that only exert pressure were also fabricated. By positioning the micropumps at appropriate locations in conjunction with additional flow-delaying components, the ejection of solutions from the reservoir of each micropump could be initiated at coordinated times. Furthermore, the solutions could be transported by the application of pressure from other micropumps. In other words, the information for switching from one micropump to another could be described on the chip in the form of a network of flow channels. This autonomous processing of solutions was demonstrated for enzymatic analyses of H2O2, glucose, and lactate. With the progress now being made in microfluidic technologies, innovative devices that make possible the complicated manipulation of solutions have been proposed for various applications.1-5 However, as far as microfluidic transport is concerned, many of these previous devices have relied on external instruments such as microsyringe pumps or power sources to produce pressure-driven flows or to generate electroosmotic flows. Such bulky instruments, however, are obstacles to the increased integration of components and miniaturization of the entire system. * To whom correspondence should be addressed. Phone: +81-29-853-5598. Fax: +81-29-853-4490. E-mail: [email protected]. (1) Thorsen, T.; Maerkl, S. J.; Quake, S. R. Science 2002, 298, 580–584. (2) Balagadde´, F. K.; You, L.; Hansen, C. L.; Arnold, F. H.; Quake, S. R. Science 2005, 309, 137–140. (3) Shiu, J.-Y.; Chen, P. Adv. Mater. 2005, 17, 1866–1869. (4) Wang, C.-H.; Lee, G.-B. Biosens. Bioelectron. 2005, 21, 419–425. (5) Satoh, W.; Hosono, H.; Yokomaku, H.; Morimoto, K.; Upadhyay, S.; Suzuki, H. Sensors 2008, 8, 1111–1127.

6870

Analytical Chemistry, Vol. 82, No. 16, August 15, 2010

To address this problem, trials have been performed to integrate active microfluidic components on a single chip.6-10 The long history of the development of micropumps and microvalves has produced a variety of devices that are based on various principles.11-13 In the reported devices, actuation of the components has usually been based on switching by electrical signals that are programmed in a number of ways. In this approach, however, specially designed electronic circuits and software are needed to realize cooperative operation of the components. In addition, for disposable devices, it would be preferable for the microfluidic system to function autonomously. To resolve this problem, chemical actuators that are based upon the spontaneous volume change of a hydrogel have been reported.14-16 Capillary action has also been used to realize a variety of devices for the autonomous transport of solutions and various other applications.17-23 These previous approaches, particularly the latter one, suggest a direction for the realization of more sophisticated devices in the next generation. The manipulation of solutions in devices has been based on programmed instructions described on the chip as a structural (6) Choi, J.-W.; Oh, K. W.; Han, A.; Okulan, N.; Wijayawardhana, C. A.; Lannes, C.; Bhansali, S.; Schlueter, K. T.; Heineman, W. R.; Halsall, H. B.; Nevin, J. H.; Helmicki, A. J.; Henderson, H. T.; Ahn, C. H. Biomed. Microdevices 2001, 3, 191–200. (7) Srinivasan, V.; Pamula, V. K.; Fair, R. B. Lab Chip 2004, 4, 310–315. (8) Satoh, W.; Hosono, H.; Suzuki, H. Anal. Chem. 2005, 77, 6857–6863. (9) Nashida, N.; Satoh, W.; Fukuda, J.; Suzuki, H. Biosens. Bioelectron. 2007, 22, 3167–3173. (10) Abdelgawad, M.; Wheeler, A. Adv. Mater. 2009, 21, 920–925. (11) Gravesen, P.; Branebjerg, J.; Jensen, O. S. J. Micromech. Microeng. 1993, 3, 168–182. (12) Shoji, S.; Esashi, M. J. Micromech. Microeng. 1994, 4, 157–171. (13) Laser, D. J.; Santiago, J. G. J. Micromech. Microeng. 2004, 14, R35–R64. (14) Beebe, D. J.; Moore, J. S.; Yu, Q.; Liu, R. H.; Kraft, M. L.; Jo, B.-H.; Devadoss, C. Proc. Natl. Acad. Sci. U.S.A. 2000, 97, 13488–13493. (15) Suzuki, H.; Kumagai, A.; Ogawa, K.; Kokufuta, E. Biomacromolecules 2004, 5, 486–491. (16) Suzuki, H.; Tokuda, T.; Kobayashi, K. Sens. Actuators, B 2002, 83, 53–59. (17) Zhao, B.; Moore, J. S.; Beebe, D. J. Science 2001, 291, 1023–1026. (18) Ahn, C. H.; Choi, J.-W.; Beaucage, G.; Nevin, J. H.; Lee, J.-B.; Puntambekar, A.; Lee, J. Y. Proc. IEEE 2004, 92, 154–173. (19) Bouaidat, S.; Hansen, O.; Bruus, H.; Berendsen, C.; Bau-Madsen, N. K.; Thomsen, P.; Wolff, A.; Jonsmann, J. Lab Chip 2005, 5, 827–836. (20) Delamarche, E.; Juncker, D.; Schmid, H. Adv. Mater. 2005, 17, 2911– 2933. (21) Chung, K. H.; Hong, J. W.; Lee, D.-S.; Yoon, H. C. Anal. Chim. Acta 2007, 585, 1–10. (22) Zimmermann, M.; Hunziker, P.; Delamarche, E. Microfluid. Nanofluid. 2008, 5, 395–402. (23) Swickrath, M. J.; Burns, S. D.; Wnek, G. E. Sens. Actuators, B 2009, 140, 656–662. 10.1021/ac1009657  2010 American Chemical Society Published on Web 07/29/2010

Figure 1. Chemically actuated micropumps with flow channels. (A) Exploded view of the micropumps and the flow channels. (B) Operation of the micropump. Cross-sections are shown that include the flow channel for transport, the diaphragm, and the lower compartment for the H2O2 solution. First, the reservoir of the micropump is filled with a solution to be transported (top). When a H2O2 solution is transported in the controlling flow channel and reaches the lower compartment of the micropump, bubbles are produced, the diaphragm inflates, and the solution in the upper reservoir is injected into the upper flow channel (bottom).

arrangement of components, including the flow channel network.18 Although there have been limitations in manipulation that is performed only through capillary action, even the complicated manipulation of solutions may be realized by coupling a programmed microfluidic network with chemically actuated microfluidic components. In a number of previous studies, gas bubbles produced by the electrolysis of water were used to produce a volume change that would mobilize a solution in a microflow channel.24-29 This principle of operation is attractive for the realization of a chemically actuated micropump, because gas production is accompanied by many chemical reactions. In creating our device, we used the volume change of oxygen bubbles produced by the catalytic decomposition of H2O2.26 To trigger the pumping action, a H2O2 solution was transported and supplied to the micropumps by capillary action in a controlling flow channel. A network of controlling flow channels described on a chip could be used as a program to operate many micropumps cooperatively. In other words, the timing of the switching among pumps could be adjusted by changing the relative positions of the micropumps and the length or other dimensional parameters of the flow channels. In this paper we present the basic concept for a chemically actuated micropump and its programming and characterize the performance of the device. (24) Bo ¨hm, S.; Timmer, B.; Olthuis, W.; Bergveld, P. J. Micromech. Microeng. 2000, 10, 498–504. (25) Suzuki, H.; Yoneyama, R. Sens. Actuators, B 2003, 96, 38–45. (26) Choi, Y. H.; Son, S. U.; Lee, S. S. Sens. Actuators, A 2004, 111, 8–13. (27) Satoh, W.; Shimizu, Y.; Kaneto, T.; Suzuki, H. Sens. Actuators, B 2007, 123, 1153–1160. (28) Shimizu, Y.; Takashima, A.; Satoh, W.; Sassa, F.; Fukuda, J.; Suzuki, H. Sens. Actuators, B 2009, 140, 649–655. (29) Blanco-Gomez, G.; Glidle, A.; Flendrig, L. M.; Cooper, J. M. Anal. Chem. 2009, 81, 1365–1370.

EXPERIMENTAL SECTION Materials and Reagents. A thick-film photoresist (SU-8) was purchased from MicroChem, Newton, MA. A precursor solution of poly(dimethylsiloxane) (PDMS) (KE-1300T) was purchased from Shin-Etsu Chemical, Tokyo, Japan. A precursor solution of PVA-SbQ, SPP-H-13, was purchased from Toyo Gosei Kogyo, Chiba, Japan. H2O2, manganese dioxide, and poly(oxyethylene) sorbitan monolaurate (Tween 20) were purchased from Wako Pure Chemical Industries, Osaka, Japan. The enzymes and related reagents were obtained from the following commercial sources: horseradish peroxidase (HRP; 100 U/mg), lactate oxidase (LOD; 38 U/mg), and bovine serum albumin (BSA) from Wako Pure Chemical Industries, Osaka, Japan; glucose oxidase (GOD; 151 U/mg) and 25% glutaraldehyde (GA) solution from Sigma-Aldrich, St. Louis, MO; N-acetyl-3,7dihydroxyphenoxazine (Amplex Red) from AnaSpec, San Jose, CA. Basic Structure and Fabrication of the Microfluidic Devices. The devices were constructed by stacking two PDMS substrates on a glass substrate (Figure 1). Flow channels were formed with PDMS using a template formed with a thick-film photoresist (SU-8). The compartments for the pumps and solutions to be transported were formed in the lower and upper PDMS layers by punching. A critical part of each micropump was a circular compartment (diameter 2.5 mm) with a diaphragm. The diaphragm was formed by intercalating a 50 µm thick PDMS sheet between the two PDMS substrates. The lower part of the compartment was connected to a controlling flow channel for the transport of a H2O2 solution. To form a MnO2 layer in the vicinity of the diaphragm, a droplet of water containing a suspension of MnO2 powder was put into the compartment that was then placed upside Analytical Chemistry, Vol. 82, No. 16, August 15, 2010

6871

down until the water evaporated. The amount of the powder that remained was between 0.75 and 1.35 mg, depending on the size of the micropump. To prevent leakage of the powder and more effectively produce oxygen bubbles and exert pressure upon the diaphragm, the compartment beneath the diaphragm was stuffed with a plug of a PVA-SbQ gel, leaving a space below it that could be filled with the H2O2 solution introduced from the controlling flow channel. In forming the plug, that space was filled with a precursor solution of PVASbQ, which was then cured under a UV light. Air vents were formed at appropriate locations to release pressure and facilitate the transport and filling of the H2O2 solution. A circular reservoir (diameter 1.0 mm) for a solution to be transported was also formed with PDMS on the diaphragm layer and was connected to a flow channel. The flow channels extending from the reservoirs of several micropumps formed an appropriate network that reflected their different purposes. An inlet was formed on the reservoir to be filled with a transported solution. Micropumps that were used only to apply pressure were formed in a similar manner. In this case, an inlet for the transported solution was not formed. For all devices, the heights of the controlling flow channels and of the flow channels in which solutions would be transported were 75 and 150 µm, respectively. The width of the flow channel for transportation was 500 µm. After the reservoirs in the upper PDMS substrate were filled with the necessary solutions, the entire structure was inserted between two poly(methyl methacrylate) (PMMA) plates and then fixed in place with bolts and nuts. Procedure and Principle of Operation. A change in the volume in the upper solution reservoir is caused by the deformation of the diaphragm that follows the production of oxygen bubbles produced by the catalytic decomposition of H2O2. First, a H2O2 solution is transported in the controlling flow channel by capillary action and is injected into the lower compartment of the micropump (Figure 1B, top). When the solution reaches the MnO2 powder below the diaphragm, oxygen bubbles are produced by the catalytic decomposition of H2O2: MnO2

2H2O2 98 2H2O + O2

The diaphragm then inflates and exerts pressure upon the solution in the reservoir. As a result, the solution is pushed forward in the flow channel and is transported to the lower stream (Figure 1B, bottom). The structure can also be modified so that the reservoir is filled with only air, instead of a solution to be transported. In this case, only a change in pressure is generated to move a liquid column that may be present in the lower stream of the extending flow channel. Several micropumps can be connected to the controlling flow channel and the flow channel for the transport of necessary solutions. Each micropump can be switched on individually and sequentially according to a predetermined schedule that is programmed in the shape of the network of controlling flow channels. Construction and Operation of Analysis Systems. To demonstrate the sequential manipulation of solutions for chemical 6872

Analytical Chemistry, Vol. 82, No. 16, August 15, 2010

analyses, two devices were fabricated. In one of them (Figure 5), a solution containing 50 U/mL HRP and another solution containing H2O2 as an analyte and Amplex Red (5 mM) were filled in the reservoir of two micropumps. The solutions were prepared with a 50 mM Tris-HCl buffer solution (pH 7.4). After the solutions were ejected from the pumps and merged at the T-junction according to the programmed pumping, their intensity of fluorescence was measured using a fluorescence microscope (VB-G25, Keyence, Tokyo, Japan) equipped with a CCD detection system (Keyence VB-7000/7010). In a more complicated device (Figure 6), enzymes (GOD and LOD) were immobilized in two of three injection ports. HRP was immobilized in a reaction chamber formed in the lower stream. In the immobilization of the enzymes, an enzyme solution, a 0.1 wt % BSA solution, and a 0.1 wt % GA solution were mixed in a 1:1:1 ratio. The mixed solution was then dropped into the corresponding reservoirs or the reaction chamber (5 µL for the GOD and LOD solutions and 1 µL for the HRP solution), and a cross-linking reaction was allowed to proceed. Following this, the enzymeimmobilized layers were immersed in a 0.1 M glycine solution for 60 min. The activity of the immobilized enzymes was 1.3 U for GOD, 0.17 U for LOD, and 3.3 × 10-2 U for HRP. After the injection ports were filled with solutions containing either glucose or lactate or both of them, along with the immobilized enzymes, H2O2 was produced by the enzymatic reactions. The solutions were then transported to the reaction chamber. The enzymatic reactions of HRP were accompanied by the generation of fluorescence, whose intensity was measured. Values for time, flow velocity, and fluorescence intensity were obtained in five measurements, whose averages are used in the following discussion. RESULTS AND DISCUSSION Movement of a Solution in the Controlling Flow Channel. The controlling flow channel consisted of three walls of PDMS and a bottom of glass. Although PDMS is hydrophobic (contact angle 110°), the hydrophilic glass bottom alone (contact angle 15°) was able to generate a sufficient driving force to produce capillary action. The velocity of the column of H2O2 changed depending on device parameters such as the width, height, and wettability of the flow channel. The velocity of a liquid column, x˙, in a straight flow channel is expressed as follows:30-33 x˙ )

γLV hw 8ηx h + w

(

)[ 2

cos θPDMS + cos θglass 2 cos θPDMS + w h

]

Here, γLV is the interfacial tension between the solution and the air, η is the viscosity of the solution, x is the distance from the inlet of the capillary to the meniscus of the moving liquid column, h and w are the height and width of the flow channel, and θPDMS and θglass are the contact angles on PDMS and glass, respectively. As could be anticipated from the equation, the movement of the column slowed as it moved forward in the fabricated flow (30) Satoh, W.; Yokomaku, H.; Hosono, H.; Ohnishi, N.; Suzuki, H. J. Appl. Phys. 2008, 103, 034903. (31) Janshoff, A.; Ku ¨ nneke, S. Eur. Biophys. J. 2000, 29, 549–554. (32) Delamarche, E.; Bernard, A.; Schmid, H.; Bietsch, A.; Michel, B.; Biebuyck, H. J. Am. Chem. Soc. 1998, 120, 500–508. (33) Kim, E.; Xia, Y.; Whitesides, G. M. Science 1995, 376, 581–584.

Figure 2. Influence of a surfactant (Tween 20) on microfluidic transport. The distance indicates that of the meniscus of the liquid column from the reservoir for the H2O2 solution in a straight flow channel of 500 µm × 75 µm in cross-section. Concentration of the surfactant: [, 0 wt %; ×, 0.005 wt %; 2, 0.01 wt %; b, 0.05 wt %; 9, 0.1 wt %.

channels. In addition, the movement occasionally became irregular, possibly due to the morphological or chemical nonuniformity of the channel. In the worst case, the column stopped midway in its journey and did not reach the lower compartment of the micropump. This problem was solved by adding a surfactant to the solution, which facilitated smooth movement. Figure 2 shows the dependence of the movement of the solution on the concentration of the surfactant (Tween 20) that was added. The influence of the surfactant was dramatic, and the solution’s movement became smoother and faster with increasing concentration of surfactant. At concentrations higher than 0.01 wt %, the flow velocity almost leveled off. In the following experiments, the concentration was therefore fixed at 0.01 wt %. In biochemical analyses in microsystems, the length of time required for a reaction is often on the order of seconds or minutes. In view of this, an additional requirement in such cases is the presence of structures that can slow the flow velocity of solutions. For this reason, we then examined how the velocity of the column changed with changes in the width of the flow channel. In Figure 3A, flow channels 2-4 are straight and have widths of 250 µm, 500 µm, and 1.0 mm, respectively. With a change made only in

the width, a marked difference in flow velocity was observed that demonstrated accelerated movement of liquid plugs in wider flow channels. The presence of compartments positioned along the flow channel exerts an additional similar influence.34,35 Therefore, rectangular compartments with dimensions of 1.5 mm × 880 µm and 6.0 mm × 3.5 mm were attached to the 250 µm wide and 1.0 mm wide flow channels (flow channels 1 and 5, respectively). A portion of the solution, however, also penetrated into the extending controlling flow channel while the solution filled each compartment. As a result, the movement of the column was not significantly different from the case in which there were no compartments. This result indicated that branched compartments are not effective for this purpose. We then tried a sequential arrangement. Figure 3B shows 500 µm wide flow channels. Flow channel 1 is straight, and flow channels 2 and 3 have compartments of different sizes (2.0 mm × 2.0 mm and 3.5 mm × 3.5 mm, respectively). By locating the exit at an appropriate position in the compartment, the transport in the flow channel was resumed after the compartment was filled completely, and the effect of the structures was more significant than that in Figure 3A. There was a tendency for bubbles to remain in the corners of the square compartments. Although these bubbles were small and had no adverse effect on the transport of solutions, circular or elliptic compartments might be better, both to avoid this potential problem and to realize a more accurate adjustment of timing. The movement of the column could also be delayed by the introduction of constrictions. The width of the flow channels in Figure 3C is 500 µm. For flow channels 2 and 3, the constrictions were positioned near the inlet and had widths of 300 and 200 µm, respectively. The constrictions also had an effect, and the movement of the column was slowed with narrower constrictions. Although we used only simple delaying structures because of the limited space, microfluidic transport can be delayed further through the use of a more complicated network of flow channels.22 The movement of a solution in a flow channel with compartments and/or constrictions can be understood using numerical

Figure 3. Movement of solutions in the flow channels with various delaying structures. (A) Effect of changing the width of the flow channel and attaching rectangular compartments on the sides. (B) Effect of adding rectangular compartments in series. (C) Effect of creating constrictions. The images were taken 1, 3, and 6 s (from left to right) after the introduction of the solution from the left. Scale bars correspond to 2 mm. Analytical Chemistry, Vol. 82, No. 16, August 15, 2010

6873

simulations,34,35 which can be helpful in the design of controlling flow channels that punctually provide H2O2 solution to micropumps according to a predetermined schedule. Autonomous Sequential Switching of the Micropumps. We next studied the function of the micropump. After the lower compartment of the pump was filled with H2O2 solution, the diaphragm inflated and the solution in the upper reservoir was ejected and then mobilized in the extending flow channel. During this step, the H2O2 solution moved forward in the controlling flow channel and filled the lower compartment of the next pump. From that point forward, the same step was repeated. Needless to say, the concentration of H2O2 injected into the controlling flow channel affects the flow velocity of the solution in the upper flow channel. Although the flow velocity definitely depends on the H2O2 concentration, significant leakage and destruction of the gel layer was observed over a certain threshold, which also depended on the size of the pump. Considering these dynamics, 1.6 or 3.2 M H2O2 was used in the following experiments, depending on the size of the pump. For the handling of many solutions, several micropumps can be connected to the controlling flow channel and to the upper flow channels for solutions to be transported. To adjust the timing and order of the injection of solutions, the distance between pumps can be adjusted in a network of flow channels. Compartments and constrictions can also be introduced, as was discussed earlier. In the device shown in parts A and B of Figure 4, micropumps are connected with one another by rectangular compartments at their sides and centers, respectively. In the device shown in Figure 4B, the compartments are filled with solution after the compartment of a neighboring micropump in the upper stream is filled with solution. With the small compartments in the crowded layout, however, the extension of the liquid columns from the upper reservoirs was not so significant compared with that of the device shown in Figure 4A. The device shown in Figure 4C has additional small compartments to delay the movement of the solution. Note the difference in the length of the liquid columns, which shows that the actuation of the micropump can be distinctly delayed in the lower stream of the controlling flow channel, unlike the device shown in Figure 4A and 4B. This result demonstrates that additional delaying structures can in fact be used to adjust the timing to trigger the actuation of the micropumps. Coordinated Operation of Micropumps for Chemical Analyses. By properly designing a network of flow channels with micropumps located in appropriate positions, microfluidic devices can be constructed for various purposes, including chemical analyses that require the specific processing of solutions. In the device shown in Figure 5, there are two micropumps to eject solutions and two pumps to apply pressure. Here, H2O2 was also used as an analyte. A solution containing H2O2 and Amplex Red (5 mM) and another solution containing 50 U/mL HRP, both prepared with a 50 mM Tris-HCl buffer solution (pH 7.4), were used to fill the reservoirs. After another H2O2 solution was injected into the controlling flow channel, it first filled the compartments of the micropumps. Following this, the analyte H2O2 solution in the reservoir was injected into the upper flow (34) Erickson, D.; Li, D.; Park, C. B. J. Colloid Interface Sci. 2002, 250, 422– 430. (35) Young, W.-B. Colloids Surf., A 2004, 234, 123–128.

6874

Analytical Chemistry, Vol. 82, No. 16, August 15, 2010

Figure 4. Arrays of micropumps with controlling flow channels with different structures. Rectangular compartments are connected with the controlling flow channels at the edges (A), at the center (B), and via smaller delaying compartments (C).

channels for transport after 9 s (Figure 5B, panels 1 and 2). After a time delay (117 s), the larger pumps that apply pressure were switched on and the solutions in the upper flow channel were pinched off from the rest of the solutions, transported to the center, and merged in the mixing channel (Figure 5B, panels 3 and 4). The average flow velocity in the mixing channel was 94 µm/s. Solutions containing the enzyme and the substrates were transported and mixed in the flow channel. The enzymatic reaction by HRP produced highly fluorescent resorufin, which generated red fluorescence under a fluorescence microscope. Figure 5C shows the dependence of the fluorescence intensity on the concentration. The fluorescence intensity was measured 3 min after mixing. In the graph plotted on the semilog scale, the dependence of the fluorescence intensity on the concentration could be clearly observed.

Figure 5. Device to eject and mix solution plugs. (A) Layout of the controlling flow channels (dashed line, shaded) and the flow channels for transport (solid line). (B) Movement of a H2O2 solution in the controlling flow channels (dashed arrows) and an analyte H2O2 solution in flow channels for the transport of solutions (solid arrows). (1) The reservoirs of the micropumps were filled with solutions. (2) The solutions were injected into the main channel. (3, 4) The solutions were transported by the exertion of pressure from the leftmost and rightmost pumps and were merged at the T-junction. (C) Dependence of the fluorescence intensity on the concentration of H2O2 ejected from a micropump. Five runs were performed, and the averages and standard deviations are shown. The dashed lines show the average +3σ of the background fluorescence. The concentration of the H2O2 solution injected into the controlling flow channel was 1.6 M. The dimensions of the chip were 25 mm × 16 mm.

More complicated manipulation of solutions could also be carried out. The device shown in Figure 6 has three injection pumps and three micropumps to exert pressure upon the ejected plugs. Three lines of controlling flow channels were used. In each channel, a pump to eject a necessary solution and a pump to apply pressure were connected. The lower compartment of each injection pump was used to delay the arrival of the H2O2 solution to the pumps to apply pressure located in the lower stream. GOD and LOD were immobilized at the bottom of the reservoirs of pumps A and C. First, we used solutions containing either glucose or lactate. The reservoir of pump A was filled with a phosphate buffer solution containing glucose and Amplex Red (5 mM), and the reservoir of pump C was filled with another phosphate buffer solution containing lactate and Amplex Red (5 mM). The reservoir of pump B was filled with a phosphate buffer solution to wash the reaction chamber. The enzymatic reactions by the oxidases produced H2O2. In view of the volume of the solutions and the activity of the immobilized enzymes, it could be assumed that the enzymatic conversion was virtually completed in the examined ranges of concentration of glucose and lactate during this preparatory period. Three minutes after the filling with solutions, the H2O2 solution was introduced into the controlling flow channel. Following this, a row of plugs was injected into the main flow channel from the reservoirs of the pumps located in the lower stream. In accordance with the flow channel design, pumps A-F were switched on 28, 85, 174, 187, 259, and 336 s after the injection of the H2O2 solution into the controlling flow channel. The velocity of a plug passing through the reaction chamber was 35 µm/s, the same velocity at which the three pumps would apply pressure. When the first plug containing Amplex Red and H2O2 produced by GOD arrived at the reaction chamber, fluorescence was generated in the chamber, as it was in the previously mentioned device, and its intensity was measured 90 s after the solution reached the chamber. After the reaction chamber was rinsed with a rinsing plug that was ejected from pump B, the last plug containing Amplex Red and H2O2 produced by LOD was introduced into the reaction chamber, and the fluorescence intensity was measured as before. The same experiment was carried out using the same solution containing glucose, lactate, and Amplex Red to fill pumps A and C. Figure 7 shows the calibration plots obtained using solutions containing only glucose or lactate and the data points obtained for solutions containing both of them. For the limited ranges of concentration that were employed, the plots were apparently linear. The values obtained for solutions containing both glucose and lactate in different combinations of concentration came close to the calibration plots, which demonstrated that the device can be used for the analysis of different analytes that coexist in the same solution. We note again that no wires or tubes were deployed around the chip to apply electrical signals or pressure. The necessary reagent solutions can be stored in the injection ports beforehand. The only thing that must be done is to introduce a sample solution and initiate the flow of the H2O2 solution in the controlling flow channel. The H2O2 solution can also be stored in a separate reservoir and can then be injected by applying a small degree Analytical Chemistry, Vol. 82, No. 16, August 15, 2010

6875

Figure 7. On-chip detection of glucose and lactate using the device shown in Figure 6. Dependence of the fluorescence intensity on the concentration of glucose (O, 0) and lactate (b, 9). O and b indicate data obtained in experiments using solutions that contained only glucose or lactate. Five runs were performed, and the averages and standard deviations are shown. 0 and 9 indicate data obtained by filling the injection ports of pumps A and C with solutions containing both glucose and lactate. The inset shows the plot at a lower concentration range near the detection limits. The dashed lines show the average +3σ of the background fluorescence.

of pressure that enables it to pass through a hydrophobic valve set at the entrance.

Figure 6. Enzyme analysis that accompanies the manipulation of three plugs. (A) Layout of the controlling flow channels (dashed line, shaded) and flow channels for transport (solid line). For purposes of clarity, the two flow channel networks are drawn separately in the lower figure. (B) Fluorescence images showing the movement of the H2O2 solution in the network of controlling flow channels (dashed arrows) and solution plugs in the network of upper flow channels (solid arrows). (1) The first solution was transported to the reaction chamber. (2) After flushing of the first solution, the reaction chamber was washed with the second plug. (3) After flushing of the solution, the third solution was transported to the reaction chamber. Although the flows in the controlling flow channels are described separately, they began to flow in the flow channels simultaneously. The concentration of the H2O2 solution injected into the controlling flow channel was 3.2 M. The dimensions of the chip were 33 mm × 27 mm. 6876

Analytical Chemistry, Vol. 82, No. 16, August 15, 2010

CONCLUSIONS Chemically actuated micropumps can be realized by making use of the volume change produced by the catalytic decomposition of H2O2. The pumps are located along a controlling flow channel that transports a H2O2 solution via capillary action. The row of pumps can be switched on sequentially following the introduction of the H2O2 solution into the controlling flow channel. The structure of the pump itself can also be used to exert pressure upon a solution in a flow channel. The timing of the switching among pumps can be adjusted by locating them at appropriate positions in a network of flow channels or by employing additional structures such as compartments and/or constrictions. In other words, the information for switching among pumps is directly described on the chip as a program. As demonstrated, one potential application for our autonomous devices is that of portable analysis systems. Although on-chip biochemical analyses for molecules such as proteins have already been carried out using integrated microfluidic components,6,27,28 this technique will simplify the construction of the entire system. Moreover, such an autonomous device can be useful for a variety of purposes, such as micromixing,23 tissue culturing,19 and gas-liquid reactions,17 since it can minimize the burden involved in the handling of solutions.

Received for review April 12, 2010. Accepted July 8, 2010. AC1009657