Fully Packed Capillary Electrochromatographic Microchip with Self

So gradient-free design of the chip mount and proper leveling of the solution are ... The aqueous colloidal silica solution (800 nm, 0.1 wt %) was hea...
3 downloads 0 Views 502KB Size
Download PDF
Anal. Chem. 2007, 79, 3214-3219

Fully Packed Capillary Electrochromatographic Microchip with Self-Assembly Colloidal Silica Beads Jongman Park,* Dami Lee, and Won Kim

Analytical Laboratory, Department of Chemistry, Konkuk University, 1 Hwayangdong, Gwangjingu, Seoul 143-701, Korea Shigeyoshi Horiike and Takahiro Nishimoto

Technology Research Lab, Shimadzu Corporation, 3-9-4 Hikaridai, Seika-cho, Soraku-gun, Kyoto 619-0237, Japan Se Hwan Lee and Chong H. Ahn*

Microsystems and BioMEMS Laboratory, Department of Electrical and Computer Engineering and Computer Science, University of Cincinnati, Cincinnati, Ohio 45221-0030

A fully packed capillary electrochromatographic (CEC) microchip showing improved solution and chip handling was developed. Microchannels for the CEC microchip were patterned on a cyclic olefin copolymer substrate by injection molding and packed fully with 0.8-µm monodisperse colloidal silica beads utilizing a self-assembly packing technique. The silica packed chip substrate was covered and thermally press-bonded. After fabrication, the chip was filled with buffer solution by self-priming capillary action. The self-assembly packing at each channel served as a built-in nanofilter allowing quick loading of samples and running buffer solution without filtration. Because of a large surface area-to-volume ratio of the silica packing, reproducible control of electroosmotic flow was possible without leveling of the solutions in the reservoirs resulting 1.3% rsd in migration rate. The capillary electrophoretic separation characteristics of the chip were studied using fluorescein isothiocyanate (FITC)-derivatized amino acids as probe molecules. A mixture of FITC and four FITC-derivatized amino acids was successfully separated with 2-mm separation channel length. Recently, extensive research have been focused on microfluidic chip-based analytical devices because of their promising advantages in instrumentation, minimized sample/reagent consumption, speed of analysis, high throughput, and efficiency. As a result of great efforts, various microfluidic analytical devices, DNA chips, clinical diagnostic chips, and capillary electrophoretic (CE) microchips have been developed and applied in practical analyses. Especially the utilization of CE separation techniques in microchip platforms greatly promises the efficiency and versatility of microfluidic analytical devices. In fact, numerous reports on microchipbased CE separations have been published including theoretical * To whom correspondence should be addressed. Phone: +82-2-450-3438. Fax: +82-3436-5382. E-mail: [email protected]; Phone: 513-556-4767. Fax: 513-556-7326. E-mail: [email protected].

3214 Analytical Chemistry, Vol. 79, No. 8, April 15, 2007

treatments, chip design, fabrications, sample pretreatments, sample injections, detection, and their practical applications.1-12 Nevertheless, the reliability of the analytical results from the chip-based CE separations suffers from poor reproducibility of sample handling, instability of electroosmotic flow (EOF), and difficulties of detection in miniaturized systems. Various sample injection techniques including a “pinching technique” and designs of injection channels have been proposed to minimize band broadening during sample injections for higher efficiency.13 Another major problem in microchip separations arises from the structure of the open tubular capillary channels. Any substantial vibrations or shocks on the chip mount and gradients in fluid heights between the reservoirs of the solution are fatal for reproducible flow control of fluids. In addition, gradients of fluid heights between the reservoirs can be generated easily as a result of the electroosmotic drive of the fluids during repetitive sample injections and separations. Once the extent of flows caused by other than electroosmotic drive of the solution becomes comparable to electroosmotic flow itself, it is hard to get reproducible (1) Manz, A.; Graber, N.; Widmer, M. H. Sens. Actuators, B 1990, 1, 244. (2) Harrison, J. D.; Manz, A.; Fan, Z.; Luedi, H.; Widmer, M. H. Anal. Chem. 1992, 64, 1926. (3) Woolley, A. T.; Lao, K.; Glazer, A. N.; Mathies, R. A. Anal. Chem. 1998, 70, 684. (4) Reyes, D. R.; Iossifidis, D.; Auroux, P.-A.; Manz, A. Anal. Chem. 2002, 74, 2623. (5) Auroux, P.-A.; Iossifidis, D.; Reyes, D. R.; Manz, A. Anal. Chem. 2002, 74, 2637. (6) Dolnik, V.; Liu, S.; Jovanovich, S. Electrophoresis 2000, 21, 41. (7) Jacobson, S. C.; Hergenroder, R.; Koutny, L. B.; Ramsey, M. J. Anal. Chem. 1994, 66, 1114. (8) Bruin, G. J. M. Electrophoresis Electrophoresis 2000, 21, 3931. (9) Becker, H.; Gartner, C. Electrophoresis 2000, 21, 12. (10) Wang, J.; Ibanez, A.; Chatrathi, M. P.; Escarpa, A. Anal. Chem. 2001, 73, 5323. (11) Figeys, D.; Pinto, D. Electrophoresis 2001, 22, 208. (12) Lion, N.; Rohner, T. C.; Dayon, L.; Arnaud, I. L.; Damoc, E.; Youhnovski, N.; Wu, Z.; Roussel, C.; Josserand, J.; Jensen, H.; Rossier, J. S.; Przybylski, M.; Girault, H. H. Electrophoresis 2003, 24, 3533. (13) Roddy, E. S.; Xu, H.; Ewing, A. G. Electrophoresis 2004, 25, 229. 10.1021/ac061714g CCC: $37.00

© 2007 American Chemical Society Published on Web 03/15/2007

analytical results. So gradient-free design of the chip mount and proper leveling of the solution are essential.14 These requirements reduce the utility of the CE microchip for on-site analysis at this moment, although a reliable hand-held microchip-based CE instrument is available.15 To allow the CE microchip to leave the laboratory, a new approach in microchip design is desired. Recently, a microchip with built-in polymeric monolithic structure having microchannels through the whole channel was reported. Because of the high surface area-to-volume ratio, this fully packed microchip with the monolithic polymeric structure showed improved uniformity of electroosmotic flow through the channels. Improved uniformity of electroosmotic flow resulted in lowered band broadening in electrophoretic separation.16 Meanwhile, during our recent work on self-assembly submicrometer colloidal silica packing in a microchannel, it was realized that the flow of aqueous buffer solution through the colloidal silicapacked microchannel was possible only by either capillary action or electroosmotic drive.17 Any pressure-driven flow was unnoticeable and negligible because of the extra high surface area-tovolume ratio of the colloidal silica-packed microchannels. On the basis of this observation, we expected the possibility of further improvement in sample and chip handling in the microchip CE experiment by packing the microchannels of the microchip fully. Indeed, the fully packed microchip did not require any sophisticated chip-mounting apparatus. Because of the built-in filter function of the packed beads, neither a chip cleaning process nor filtration of the solution was required. The close crystalline packing of the colloidal silica in the microchannels served as a stationary phase for good separation even with a very short separation column length. The details of chip preparation and their characteristics are described below. EXPERIMENTAL SECTION Reagents and Materials. Plain colloidal silica having a uniform size (800 nm in diameter) was purchased from Bang’s Laboratories, Inc. (Fisher IN). Fluorescein isothiocyanate (FITC) and amino acids (L-arginine, DL,L-phenylalanine, L-glutamic acid, glycine) were purchased from Sigma-Aldrich. FITC derivatization of amino acids for fluorescence detection was done a day before the CE experiment and stored in a refrigerator as described elsewhere.18 Solutions used in CE experiments were neither filtered nor degassed. Cyclic olefinic copolymer (COC) resins (Topas 8007, 5013) were obtained from Ticona (Summit, NJ). Fabrications. COC plastic microchip substrate patterned with a simple cross microchannel was prepared by an injection molding technique using an electroplated nickel mold.19 The separation channel was 25 mm long and other channels were 5 mm long from the channel cross. All channels were 100 µm in width and 80 µm in depth. The patterned COC substrate was made of a resin (14) Wang, J; Tian, B.; Sahlin, E. Anal. Chem. 1999, 71, 3901. (15) Jackson, D. J.; Naber, J. F.; Roussel, T. J., Jr.; Crain, M. M.; Walsh, K. M.; Keynton, R. S.; Baldwin, R. P. Anal. Chem. 2003, 75, 3643. (16) Lazar, L. M.; Li, L.; Yang, Y.; Karger, B. L. Electrophoresis 2003, 24, 3655. (17) Horiike, S.; Lee, S. H.; Nishimoto, T.; Ahn, C. H. Proceedings of µ-TAS 2003, 7th International Conference on Micro Total Analysis Systems; October 5-9, 2003l Vol. 1, p 417. (18) Fang, Q; Xu, G.-M.; Fang Z.-L. Anal. Chem. 2002, 74, 1223. (19) Trichur, R.; Kim, S.; Lee, S. H.; Abdelaziez, Y. A.; Starkey, D. E.; Halsall, H. B.; Heineman, W. R.; Ahn, C. H. Proceedings of µ-TAS 2002, 6th International Conference on Micro Total Analysis Systems, November 3-7, 2002; p 560.

Figure 1. Structure of the fully packed microchip and its actual shape.

having a high glass transition temperature (Tg ) 134 °C). After packing with submicrometer silica beads, the packed substrate was covered with a plain COC plate having low Tg (78 °C) and sealed using a homemade hot embossing machine. To avoid the destruction of the silica packing by the pressure applied during press-bonding, the temperature of the chip in the hot embossing machine was maintained at 115 °C for 1 h without pressing and the cover plate was allowed to soften. The softened cover plate was squeezed from the top with the weight of the hot plate of the embossing machine for 30 min in order to bond it to the packed substrate. Then it was cooled down to room temperature without removing the weight. The bonded chip was further fabricated for CEC test. Three reservoirs for sample, sample waste, and buffer solutions were formed by drilling with a milling bit (2.4 mm in diameter) at positions of 6.2 mm from the cross center. The waste reservoir at the end of the separation channel was formed at the position desired. Figure 1 shows the structure and the real appearance of the packed chip. Silica Bead Packing. The microchannels were packed with silica colloidal beads by a self-assembly packing process as reported earlier.17 The patterned COC chip substrate was pretreated with O2 plasma for 5 min to give hydrophilicity to the microchannel surface. The aqueous colloidal silica solution (800 nm, 0.1 wt %) was heated to 40 °C in a beaker with gentle stirring to prevent slow precipitation of the aggregated silica particles. The end of the separation channel was dipped cautiously in the solution by holding with a jig. The beaker was covered with a Petri dish to reduce evaporation of water. Pretreated open microchannels showed enough high hydrophilicity to drive the water by capillary action to the ends of the channels along with silica colloidal particles suspended in it as illustrated in Figure 2. Once the colloidal silica particles reached the end of the capillary channels, spontaneous three-dimensional packing of the silica beads started from the end of the microchannels due to the slow evaporation of water. Such a self-assembly packing technique of colloidal beads for the preparation of multilayer crystalline films was already demonstrated in the area of photonic devices.20 The self-assembly packing process continued toward the end of the empty microchannel at the bottom. The packing speed was controlled by the (20) Jiang, P.; Bertone, J. F.; Hwang, K. S.; Colvin, V. L. Chem. Mater. 1999, 11, 2132.

Analytical Chemistry, Vol. 79, No. 8, April 15, 2007

3215

Figure 2. Illustration of self-assembly colloidal silica packing of microchannels.

opening between the beaker and the Petri dish. The packing process was stopped when the desired length of packing was achieved. The packed chip was washed very gently and cautiously with plenty of deionized water to remove extra silica particles at the dipped area and was dried completely at room temperature. Then it was subjected to cover bonding and further fabrications as described earlier Capillary Electrochromatographic Experiments. A multichannel high-voltage power supply for the osmotic flow drive of solutions was from Lab Smith (HVS 448-3000V). A fluorescence microscope (Olympus BMX 51) was interfaced to a CCD multichannel spectrometer (S2000 PCI, Ocean Optics) using optical fiber for the detection of the analytes. A 75-W Xe lamp light source for excitation was used. A filter block with band-pass of 400-440 nm and barrier filter of >475 nm was used in experiments to select excitation and detection wavelengths. Detection bandwidth of 20 nm was set at 520 nm, and the signal was integrated for every 100 ms. The detection zone was defined by controlling the focusing slit size on the microscope. An objective lens of 20× magnification was used for the fluorescence detection. A small USB interfaced CCD camera was mounted on the eye piece of the microscope to monitor the electroosmotic flow. FITC and four FITC-derived amino acids were used as probe molecules. The drilled reservoirs were filled with 5-10 µL of solutions including sample solution using a micropipet without filtering. The reservoirs for running buffer, sample, sample waste, and waste solution are denoted as A, B, C, and D, respectively, for convenience (Figure 1). A pinching technique in sample loading was applied as well as a back-pushing technique during separation using the programmable high-voltage power supply. The voltage program was varied from chip to chip depending on the packing quality of the chips. Safety Considerations. The high-voltage power supply should be handled with extreme care to avoid possible electrical shock. 3216

Analytical Chemistry, Vol. 79, No. 8, April 15, 2007

RESULTS AND DISCUSSION The microchannels were filled with silica beads in perfect crystalline packing except for some irregular packing at the edge as shown in Figure 3a. This imperfect packing near the sidewall is mainly due to the roughness of the channel sidewall. Recently, De Pra et al. reported the effect of the geometry of channel sidewall. Stagnant zones near the sidewall of their micropillar structured chip caused severe band broadening due to the interrupted regularity of flow path.21 Indeed, the band broadening was quite large in the separations with the fully packed microchip. The microchannel patterning process on COC substrate should be improved to minimize the imperfect packing near the side wall for better separation efficiency. Occasionally some fractures or crevices were encountered after drying of the packing or bonding of the chip cover as seen in Figure 3b. It is likely due to the residual stress that occurred during the drying process or mechanical strain that occurred during an improper bonding process. Enough allowance for softening of the cover plate at elevated temperature before pressing minimized such fractures of packing. Uniform close crystalline packing of the submicrometer silica beads offers great advantages in various aspects of CEC microchip studies such as simplicity of solution/chip handling, enhanced controllability of EOF, and analytical characteristics. After preparation of the packed chip, the microchannels were filled with 20 mM sodium borate solution (pH 9.2) and conditioned overnight as follows. In order to fill the microchannels with buffer solution, a drop of the buffer solution was added to one of the reservoirs using a micropipet. Upon the addition, the buffer solution was self-primed toward the other ends of the channels by capillary action. It took about 5-10 min until the whole channels filled. Once it is filled with buffer solution, it should be kept wet all the times without drying. If it is dried by mistake, the packing structure is damaged by the formation of a crystalline solid of buffer chemicals. The chip was kept in some buffer solution or between moisturized tissues with buffer in a small zipper bag. Storage between moisturized tissues was preferred. Some erosion of the packing at the ends of microchannels was observed when stored in buffer solution. Once it is conditioned with buffer solution, it is ready to use any time. The chip was washed with distilled water, and then the residual water at the surface was wiped out with paper tissue. The solution inside of the channels can be replaced with any new buffer solution desired by flushing it out electroosmotically. It is worthwhile to mention that the chip has built-in submicrometer filters formed by submicrometer silica packing. The cumbersome filtration process of sample and buffer solution needed in an ordinary CE microchip can be omitted, even the chip cleaning process. As mentioned in the Experimental Section, the chips and solutions were used without further cleaning or filtration. Handling of the microchip also became much easier. The silica packing has a very high surface-to-volume ratio so that the solutions were held safely with minimized gravitational flow. The buffer solution was retained without any apparent flow by height difference when it was tilted. So the chip can be handled easily during CEC analysis without leveling problems of the solution, which is one of the major problems in ordinary open tubular CE (21) De Pra, M.; Kok, W. Th.; Gardeniers, J. G. E.; Desmet, G.; Eeltink, S.; van Nieuwkasteele, J. W.; Schoenmakers, P. J. Anal. Chem. 2006, 78, 6519.

Figure 3. Scanning electron microscopic images of silica packing. (a) Normal silica packing in microchannel. (b) Fractured silica packing showing three-dimensional closest packing and crevices formed by stress.

microchip work. Unleveled height between the buffer reservoirs disturbs the controllability of EOF by forming a water head backward to EOF in such a CE microchip. As electroosmotic pumping proceeds, the height differences between the reservoirs become larger and larger resulting in poor reproducibility of the migration time. In the fully packed microchip, however, the effect of the water head developed by improper leveling of the solution was suppressed. Reproducible flow control was possible by electroosmotic driving as can be seen later. A similar result in the enhanced EOF controllability was demonstrated recently by employing monolithic polymer packing in microchannels on a glass microchip as mentioned earlier.16 However, the microchannels formed in monolithic polymer packing were large and random in shapes and sizes. These characteristics of the fully packed microchips mentioned above can greatly enhance the utility of CE microchips for practical point-of-care analysis.

Figure 4 shows the injection and separation process of FITCderivatized amino acids. The chip had a 10-mm-long separation channel from the channel cross to the waste reservoir. The other channels from the channel cross to the reservoirs were 5 mm long. A pinching technique was used to minimize band broadening during sample loading by applying +300 V at the sample reservoir (A) and +280 V at the other two reservoirs (C, D) against ground at waste reservoir D. It took a long time to get a steady-state flow of highly negative FITC derivative of glutamic acid from the sample reservoir to the waste reservoir. Enough time for sample loading (30 s) was allowed to get a reproducible signal for FITCglutamic acid (Figure 4a).18 After 300 ms of rest time with potential floating for all reservoirs, the sample was injected to the separation channel by applying +600 V at the running buffer reservoir (C) against ground at the waste reservoir (D). Direct switching of the voltage from sample loading to separation caused turbulence Analytical Chemistry, Vol. 79, No. 8, April 15, 2007

3217

Figure 4. Photographs captured during capillary electrochromatographic separation of FITC and FITC-derivatized amino acids. (a) Loading of sample mixture, (b) the moment of sample injection, and (c, d) 0.5 and 2 s after injection.

Figure 5. Separation of FITC-derivatized amino acid mixture, 0.25 mM each. Key: A, arginine; F, FITC; P, phenylalanine; G, glycine; G*, glutamic acid. Detected at 2.5 mm from the channel cross. 20 mM Na2B4O7 running buffer (pH 9.2). Sample loading (A, +200 V) with pinching (C and D, +170 V; B, ground), and separation (C, +200 V) with back pushing(A and B, +70 V). The arrow denotes “injection”.

of the sample at the channel cross because of a large and sudden charging current flow at the moment of voltage switching. The voltage switching mode with rest time was helpful to reduce the band broadening. The flow of the sample channels was pushed back effectively by applying +380 V at both reservoirs A and B (Figure 4b). The analytes in the sample mixture started to be separated as soon as injected (Figure 4c). Four fluorescing bands were captured within a frame of the picture only after 2 s from the injection (Figure 4d). The length of the separation channel captured here was only 1.4 mm from the channel cross. It is thought that the close crystalline packing of the colloidal silica 3218 Analytical Chemistry, Vol. 79, No. 8, April 15, 2007

Figure 6. Effect of detection position. (a) 1, (b) 2, and (c) 3 mm from the channel cross. Other conditions are same as Figure 5.

serves as a good stationary phase to give enough interactions for the separation even with this short column length. Figure 5 is an electrochromatogram for the separation of FITC and four FTIC-amino acid derivatives. In this chip, all microchannels were 5 mm long from the channel cross. The fluorescing

signals were detected only at 2 mm from the channel cross. Five components were successfully separated as can be seen. Negatively charged FITC-derivatized glutamic acid at a given pH showed slow migration rate with large band broadening. Large band broadening lowered the separation efficiency of the microchip. However, it should be mentioned that the microchip has the width of 100 µm resulting in an ∼200-µm bandwidth at the injection point (Figure 4b). Only a 10 times longer column length was enough for effective separation of the mixture. The channel size should be optimized to reduce the sample injection bandwidth. The quality of the microchannels on the chip also should be improved to minimize the band broadening owing to the irregularity of the packing near the sidewall as commented on earlier. Figure 6 shows the effect of detection position in the separation. One millimeter of column length was not sufficient, but 2 mm was enough for successful separation of this sample. When the signal was sampled at 3 mm from the channel cross, the separation was not further improved. Relatively severe band broadening suffered the resolution of the separation. Unlikely with an ordinary CE microchip, the channel is packed with submicrometer silica beads forming three-dimensional submicrometer multichannels. It is likely various kinetic parameters in the multichannels including the stagnant zones formed by irregular packing contribute to the band broadening and compete with electrophoretic migration for separation in electroosmotic flow. Details in parameter optimization for improved separation are under study in our laboratory. The reproducibility of the migration time and peak height was evaluated using the arginine derivative and FITC mixture. The

sample was injected repeatedly for 10 times and detected at 2 mm from the channel cross. The relative standard deviation (rsd) of the migration time for FITC was only 1.3%, which means reproducible EOF control was possible in the fully packed microchip. The rsd’s of 3.4% for the arginine derivative (first peak) and 1.5% for FITC (second peak) in peak height were achieved. In conclusion, although the fully packed microchip illustrated above is not completely optimized in colloidal silica packing and following fabrications as well as operating parameters yet, it offers great advantages in solution and chip handling. It can be another step toward microchip CEC analysis at the point of care with handheld instruments. Moreover effective microchip CEC separation with a several millimeters-long separation column was possible so that further miniaturization of the CE microchip would be possible. Full characterization of the analytical characteristics of the chip and further improvement in chip fabrication for better separation performance are being studied in our laboratory. ACKNOWLEDGMENT This work was supported partly by Konkuk University through a visiting scholar program at the University of Cincinnati and by the Korea Research Foundation Grant funded by the Korean Government(MOEHRD) (KRF-2005- 041-C00297)

Received for review September 12, 2006. Accepted February 8, 2007. AC061714G

Analytical Chemistry, Vol. 79, No. 8, April 15, 2007

3219