Anal. Chem. 2001, 73, 3193-3197
A Method for Filling Complex Polymeric Microfluidic Devices and Arrays Jennifer Monahan, Andrew A. Gewirth,* and Ralph G. Nuzzo*
Department of Chemistry, University of Illinois, 600 S. Mathews, Urbana, Illinois 61801
This paper describes an improved method for filling microfluidic structures with aqueous solutions. The method, channel outgas technique (COT), is based on a filling procedure carried out at reduced pressures. This procedure is compared with previously reported methods in which microfluidic channels are filled either by using capillary forces or by applying a pressure gradient at one or more empty reservoirs. The technique has proven to be >90% effective in eliminating the formation of bubbles within microfluidic networks. It can be applied to many devices, including those containing PDMS-terminated channel features, a single channel inlet, and threedimensional arrays. For several years, there has been considerable interest centered on the fabrication of integrated microanalytical devices. These include systems for chemical separations (so-called labon-a-chip)1-5 as well as highly functional architectures for bioanalysis (BioMEMS).6,7 An issue of considerable importance highlighted in the recent literature involves the materials used to fabricate these structures. Although the original devices in this field were fabricated in glass and silicon using conventional photolithography, current work has expanded fabrication into polymeric materials.8-18 Much of this latter interest has focused on making microfluidic devices in polydimethyl siloxane (PDMS).8,9,15-17,19 * To whom correspondences should be addressed. Fax: 217-333-2685. E-mail:
[email protected],
[email protected]. (1) Dolnik, V.; Liu, S.; Jovanovich, S. Electrophoresis 2000, 21, 41-54. (2) Harrison, D. J.; Glavina, P. G.; Manz, A. Sens. Actuators B 1993, B10, 107116. (3) Graber, N.; Ludi, H.; Widmer, H. M. Sens. Actuators, B 1990, B1, 239243. (4) Manz, A.; Harrison, D. J.; Verpoorte, E. M. J.; Fettinger, J. C.; Paulus, A.; Ludi, H.; Widmer, H. M. J. Chromatogr. 1992, 593, 253-258. (5) Knapp, M. R.; Sundberg, S.; Kopf-Sil, A.; Nagle, R.; Gallagher, S.; Chow, C.; Wada, G.; Nikiforov, T.; Cohen, C.; Parce, J. W. Am. Lab. 1998, 30 (22), 22-26. (6) Santini, J. T.; Richards, A. C.; Scheidt, R.; Cima, M. J.; Langer, R. Angew. Chem., Int. Ed. 2000, 39, 2396-2407. (7) Micro Total Analysis Systems 2000: Proceedings of the µTAS 2000 Symposium; Kluwer Academic Publishers: Dordrecht, The Netherlands, 2000. (8) Folch, A.; Ayron, A.; Hurtado, O.; Schmidt, M. A.; Toner, M. J. Biomech. Eng. 1999, 121, 28-34. (9) Jackman, R. J.; Duffy, D. C.; Ostuni, E.; Wilmore, N. D.; Whitesides, G. W. Anal. Chem. 1998, 70, 2280-2287. (10) Martynova, L.; Locascio, L. E.; Gaitan, M.; Kramer, G. W.; Christensen, R. G.; MacCrehan, W. A. Anal. Chem. 1997, 69, 4783-4789. (11) Chen, Y.-H.; Chen, S.-H. Electrophoresis 2000, 21, 165-170. (12) Becker, H.; Gartner, C. Electrophoresis 2000, 21, 12-26. (13) McCormick, R. M.; Nelson, R. J.; Alonso-Amigo, M. G.; Benvegnu, D. J.; Hopper, H. H. Anal. Chem. 1997, 69, 2626-2630. 10.1021/ac001426z CCC: $20.00 Published on Web 05/24/2001
© 2001 American Chemical Society
Molded PDMS is used in many applications, including surface patterning methods such as micromolding in capillaries (MIMIC)20-26 and analytical devices for electrophoretic separations.15-18,27 Unfortunately, several properties of PDMS make its use as a medium for microfluidics difficult. Unless treated,16,17 PDMS is hydrophobic, a feature that stems from its low surface free energy.28,29 As a result, capillary action alone is not always sufficient to completely fill large or elaborate devices from a reservoir of aqueous media. For separation devices, this problem is frequently overcome by applying a pressure gradient to flush buffer through the column. In most methods described in the literature, a vacuum hose is applied directly to one of the device reservoirs in order to draw fluid into the channel network from the remaining solution-filled reservoirs.15 Unfortunately, multiple reservoirs and channel branches can result in trapped air bubbles within a microfluidic network. Additionally, PDMS is not mechanically rigid. Thus, strong mechanical forces resulting from a vacuum applied to one end of a elastomeric PDMS channel can result in its collapse if the aspect ratio is not properly controlled.19,30 We have found that as the number of intersecting channels increases or internal angles within the microfluidic network decrease, the filling process becomes more difficult. This difficulty stems from the unfavorable wetting properties associated with the confinement of high surface tension liquids within hydrophobic (14) Wang, S.-C.; Perso, C. E.; Morris, M. D. Anal. Chem. 2000, 72, 17041706. (15) Effenhauser, C. S.; Bruin, G. J. M.; Paulus, A.; Ehrat, M. Anal. Chem. 1997, 69, 3451-3457. (16) McDonald, J. C.; Duffy, D. C.; Anderson, J. R.; Chiu, D. T.; Wu, H.; Schueller, O. J. A.; Whitesides, G. M. Electrophoresis 2000, 21, 27-40. (17) Duffy, D. C.; McDonald, J. C.; Schueller, O. J. A.; Whitesides, G. M. Anal. Chem. 1998, 70, 4974-4984. (18) Delamarche, E.; Bernard, A.; Schmid, H.; Bietsch, A.; Michel, B.; Biebuyck, H. J. Am. Chem. Soc. 1998, 120, 500-508. (19) Delamarche, E.; Schmid, H.; Michel, B.; Biebuyck, H. Adv. Mater. 1997, 9, 741-746. (20) Kim, E.; Xia, Y.; Whitesides, G. M. Nature 1995, 376, 581-585. (21) Kim, E.; Xia, Y.; Whitesides, G. M. J. Am. Chem. Soc. 1996, 118, 57225731. (22) Xia, Y.; Kim, E.; Whitesides, G. M. Chem. Mater. 1996, 8, 1558-1567. (23) Jeon, N. L.; Choi, I. S.; Xu, B.; Whitesides, G. M. Adv. Mater. 1999, 11, 946-950. (24) Zhao, X.-M.; Xia, Y.; Whitesides, G. M. J. Mater. Chem. 1997, 7, 10691074. (25) Kim, E.; Whitesides, G. W. J. Phys. Chem. B 1997, 101, 855-863. (26) Xia, Y.; Whitesides, G. W. Angew. Chem., Int. Ed. Engl. 1998, 37, 550575. (27) Martin, R. S.; Gawron, A. J.; Lunte, S. M.; Henry, C. S. Anal. Chem. 2000, 72, 3196-3202. (28) Owen, M. J. J. Coat. Technol. 1981, 53, 49-53. (29) Chaudhury, M. K.; Whitesides, G. M. Langmuir 1991, 7, 1013-1025. (30) Biebuyck, H. A.; Larsen, N. B.; Delmarche, E.; Michel, B. IBM J. Res. Dev. 1997, 41, 159-170.
Analytical Chemistry, Vol. 73, No. 13, July 1, 2001 3193
channels. Frequently, the filling procedures based on pressure gradients are severely impacted by bubble formation.21 As a result, far more time can be spent in trying to remove bubbles from a PDMS electrophoretic device than in using it. To simplify filling processes for PDMS-based microfluidic devices, we have developed a variation of capillary-force or pressure-gradient-based filling methods. Our procedure, channel outgas technique (COT), takes advantage of a decreased pressure over the entire PDMS device and a large excess of fill solution. This procedure effectively fills complex channel designs, including those possessing discontinuous branches that are terminated internally and 3-D arrays. The fill procedure for most devices has been consistently reduced to a 10-min process. EXPERIMENTAL: Chemicals and Reagents. Polydimethyl siloxane was purchased from Dow Corning (Midland, MI) under the product name Sylgard 184. Fluorescein and glucose were purchased from Aldrich (Milwaukee, WI). Aqueous fluorescent buffers were used in this study to facilitate easy visual analysis of filling success. Instrumentation. PDMS device fabrication has been outlined elsewhere21,26,31 and consists of a molded PDMS pattern sealed on a second PDMS flat to generate channel networks. The surface of the PDMS was not oxidized16,17 but, instead, was sealed reversibly by hydrophobic, conformal contact after washing.15 Reservoir holes were punched through the patterned PDMS mold using a steel leather punch. The success of each fill method was documented using a 35mm camera mounted on an epifluorescent upright microscope (Olympus AX-70; Melville, NY). Fill solutions were spiked with fluorescein at ∼5 mM concentration for easy monitoring using an appropriate dichroic mirror assembly (Olympus U-MIB - 470505-nm band-pass filter, >515-nm high-pass filter). The vacuum chamber (National Appliance Company 5831-6; Portland, OR) used for these experiments was equipped with a mechanical pump. The mechanical pump was equipped with a trap for aqueous vapors. Fill Procedures. The three filling methods compared in this paper are illustrated in Figure 1. Methods 1a and 1b are variations of methods found in the literature. Experiments using capillary action, Figure 1a, were conducted on a flat to slightly canted surface. The lower reservoir was filled with buffer and care was used to ensure that no air bubbles blocked solution access to the channels. With this method, buffer was drawn into the PDMS channels via capillary action. Figure 1b outlines the second filling method, which required filling one or more reservoirs with buffer. A tube was then used to attach the empty reservoir to a vacuum source or syringe. The pressure gradient generated in this way was used to pull the solution from the filled reservoir down the channel. For our comparison studies, a vacuum of ∼50 Torr was applied to the device using a mechanical pump. For the channel outgas technique shown in Figure 1c, the sealed PDMS device was adhered to a glass slide and placed, reservoir-side-up, on the bottom of a beaker containing the fill solution of interest. The total volume of solution depends on the overall dimensions of the device, but the method works well if (31) Kumar, A.; Biebuyck, H. A.; Whitesides, G. W. Langmuir 1994, 10, 14981511.
3194 Analytical Chemistry, Vol. 73, No. 13, July 1, 2001
Figure 1. Schematic representation of three methods for filling microfluidic channels in a PDMS device: (a) capillary forces alone, (b) direct application of a vacuum at a reservoir outlet, (c) atmospheric pressure reduction above a submerged device.
the device is covered by several millimeters of solution. (This volume can be reduced to as little as 1 mL using a reservoirextension, as described below.) The glass slide serves as a weight to keep the device submerged and care should be taken to ensure that no bubbles are trapped under the glass slide. (It is not necessary to remove any bubbles that form within the PDMS reservoirs once the device is immersed in the solution.) The beaker containing the fill solution and submerged device was then placed in a vacuum chamber and the atmospheric pressure was decreased slowly over a 1-min period. As the surrounding pressure is decreased, gas trapped within the PDMS device escapes via the channel access reservoirs. We have found that evacuating to e 125 Torr was optimal for filling these channel systems. After 10 min, the pressure was returned to atmospheric pressure and the device was left submerged until the network appeared to be completely filled. For most devices this takes less than 1 min. With a transparent solution it is possible to monitor the fill progress without removing the device from solution. However, if the solution is opaque, the device can be removed from solution and rinsed to determine the extent of filling. In some cases, trapped air bubbles must be dislodged from any nonvertical reservoir walls for filling to commence. If necessary, the reservoirs can be emptied and the entire device resubmerged for a second treatment. Once successfully filled, the device is removed, rinsed, and gently dried using a lab tissue or a gentle gas flow.
Figure 2. Comparison of the three methods for filling a simple microfluidic channel in a PDMS device. The fill solution was a 2 mM fluorescein buffer at pH ≈ 7 (a pH where PDMS is not wet by the liquid): (a) capillary forces alone, (b) direct application of a vacuum at a reservoir outlet, (c) atmospheric pressure reduction above a submerged device. In all three images, the channel width was 90 µm; channel depth was 15 µm; scale bar indicates 100 µm.
DISCUSSION AND RESULTS: The performance of the three fill methods are qualitatively compared for the hydrophobic PDMS devices using a simple t-junction, a device that is commonly used for electrophoretic separations.32-36 For this demonstration, the column cross-section was 90 µm × 15 µm with short arm lengths of 8 mm and the longest arm of the cross pattern extending for 50 mm. As described above, the first method simply uses capillary action to draw solution from a reservoir into channels within the PDMS. This method has often been used for MIMIC20-26 processes to fabricate complicated polymeric patterns on surfaces. Figure 2a illustrates the limited fill volume achieved for an aqueous solution in a hybrid PDMS-glass device after 10 minutes. An identical device with a four-sided PDMS channel could not be filled using capillary action alone. This limited degree of fill appears to follow the boundary value constraints predicted by a model of the acting capillary forces.20,21 The second method uses a vacuum gradient to draw solution from a filled reservoir into the PDMS channels. This method is routinely used to fill simple electrophoretic devices having a relatively small number of channel intersections. Figure 2b demonstrates some of the problems that typically occur with this method. Although much of the channel is filled successfully, other areas routinely contain trapped bubbles. COT does not generate such bubbles. For the channel shown in Figure 2c, the (33) Harrison, D. J.; Fluri, K.; Seiler, K.; Fan, Z.; Effenhauser, C. S.; Manz, A. Science 1993, 261, 895-897. (34) Effenhauser, C. S.; Manz, A.; Widmer, H. M. Anal. Chem. 1993, 65, 26372642. (35) Jacobson, S. C.; Hergenroder, R.; Koutny, L. B.; Ramsey, J. M. Anal. Chem. 1994, 66, 1114-1118. (36) Jacobson, S. C.; Hergenroder, R.; Koutny, L. B.; Warmack, R. J.; Ramsey, J. M. Anal. Chem. 1994, 66, 1107-1113. (37) CRC Handbook of Chemistry and Physics, 53rd ed.; CRC Press: Cleveland, 1972.
Figure 3. Comparison of fill methods for complicated devices: (a), (c), (e) channels filled using capillary forces alone; (b), (d), (f) channels using a reduced-pressure chamber. (a)-(d) Channel depth, 9 µm, scale bar indicates 120 µm. (e)-(f) Channel depth, 16.5 µm; scale bar indicates 85 µm.
device was submerged in a beaker containing a 2 mM fluorescein buffer at pH ≈ 7. The chamber pressure was reduced to 75 Torr and held at that level for five minutes before venting. After holding the device submerged in the buffer for a further minute at ambient pressure, it was then removed. Upon detailed examination we found that the entire electrophoretic-channel t-junction was uniformly filled. This test experiment was repeated more than 25 times (using deionized water or 20 cm long can be filled quickly and easily from a single supply reservoir entrance. It is important to note, however, that as with other methods, only channel regions that are part of a continuous fluidic pattern will be filled. This is clearly demonstrated by the image shown in Figure 4b, which shows that the isolated center chambers (100 × 100 µm in size) of a complex array were not filled by the fluorescein buffer. Such observations confirm that the PDMS channel system stays in firm, compliant contact with the flat substrate during the filling process. With some applications, it may not be desirable to submerge the entire device in the fill solution of interest. For example, in electrophoretic applications, there may be concern about a residual film of conducting buffer on the surface of the device. For these applications we have found it convenient to first fill the network with water or a low ionic strength solution. The appropriate conducting buffer can then be flushed through the system electrophoretically. For applications in which the fill solution is in short supply, we have applied a further modification. A PDMS extension reservoir is added to the top of the device’s access hole, as shown in Figure 5a. Because PDMS will hermetically self-seal, it is possible to temporarily expand the access hole volume to a milliliter or more for the filling process. Once this smaller (1 mL)
extension reservoir is filled, it replaces the large excess of solution needed to submerge the device, and the device is placed in the vacuum chamber for COT. This mini-reservoir allows the device to fill in the same manner as a completely submerged device. The reservoir extension can be removed after the channel network is filled (using care not to separate other layers of the PDMS device). This leaves the bulk of the external device uncontaminated by fill solution and eliminates the need for large amounts (>50 mL) of expensive reagents. There is one design limitation with this modification: the device itself must be large enough to support the extension reservoir and smooth enough to make a good seal. Figure 5 also shows a final application we explored for COT. These results demonstrate that this method can be used to easily fill three-dimensional devices with aqueous solutions. As illustrated in Figure 5a, the channels in two patterns were crossed to make an interconnected stacked array. The access hole in this device supplied fluid directly to only three parallel lines of the top array. Using COT, fluid filled both levels of the array through the joined continuous network, as shown in Figure 5b. As measured by surface profilometry, the limiting channel dimensions in this structure were roughly 20 µm wide and 50 µm high. Further multidimensional tests demonstrated that 5 cm × 5 cm arrays of intersecting PDMS lines with a single access point filled with a 100% success rate for 20 successive attempts. The dimensions of these lines ranged from 20 to 150 µm wide by 15 µm tall. It should be possible to expand this example to many levels by incorporating connecting channels between layers. Although a range of viable viscosities was not determined for this method, COT was successfully tested in a 90 µm × 15 µm channel using a 2 M glucose, 2 mM fluorescein solution having a nominal relative viscosity of 10 cP.36 Higher viscosity solutions, such as mineral oil, can cause very low aspect ratio channels (15
µm high × 1200 µm wide) to collapse. This is probably due to the inability of elastomeric PDMS to withstand the large pressure drop as the viscous solution is drawn into the device. It should be noted that an important restriction on this system is one that holds for all PDMS devices: this method works well with aqueous solutions, but care must be used with organic solvents that can cause PDMS to swell or distort (thus causing it to detach from the substrate and ruin the device pattern). Additionally, applying COT to resins or polymeric solutions containing volatile solvents is difficult because high vapor pressure solvents will evaporate in the vacuum chamber. CONCLUSION: We have demonstrated that filling PDMS devices with aqueous solutions can be simplified using a channel outgas technique. The use of a vacuum chamber and enough solution to submerge a microfluidic device allows bubble-free filling of complex patterns with very little effort. These channel networks can be complex, with varying channel aspect ratios. It has been demonstrated that this method easily fills single entry, PDMS terminated channels as well as 3-D arrays. In our lab, COT has been used to fill over 100 devices with a success rate of greater than 90%. ACKNOWLEDGMENT This work was supported by National Science Foundation Grants CHE9626871 and CHE9820828 and used central facilities of the Frederick Seitz Materials Research Laboratory supported by the Department of Energy (DEFG02-96ER45439). Received for review December 5, 2000. Accepted April 2, 2001. AC001426Z
Analytical Chemistry, Vol. 73, No. 13, July 1, 2001
3197
