Anal. Chem. 2005, 77, 7478-7482
Patterned Solvent Delivery and Etching for the Fabrication of Plastic Microfluidic Devices Paul C. Brister and Kenneth D. Weston*
Department of Chemistry and Biochemistry, Florida State University, Tallahassee, Florida 32306
A very simple method for micropatterning flat plastic substrates that can be used to build microfluidic devices is demonstrated. Patterned poly(dimethylsiloxane) elastomer is used as a template to control the flow path of an etching solvent through a channel design to be reproduced on the plastic substrate. The etching solvent was a acetone/ethanol mixture for poly(methyl methacrylate) substrates or a dimethylformamide/acetone mixture for polystyrene. The method is extremely fast in that duplicate plastic substrates can be patterned in just a few minutes each. We identified conditions that lead to smooth channel surfaces and characterized the rate of etching under these conditions. We determined that, for sufficiently short etching times (shallow channel depths), the etch rate is independent of the linear flow rate. This is very important since it means that the etch depth is approximately constant even in complex channel geometries where there will be a wide range of etchant flow rates at different positions in the pattern to be reproduced. We also demonstrate that the method can be used to produce channels with different depths on the same substrate as well as channels that intersect to form a continuous fluid junction. The method provides a nice alternative to existing methods to rapidly fabricate microfluidic devices in rigid plastics without the need for specialized equipment. The number and diversity of applications of microfluidic devices are rapidly increasing across a broad range of scientific and technological disciplines.1 Successful chromatographic2 and electrophoretic3 separations have been demonstrated using microfluidic systems. Applications of biotechnology that use microchips are gaining popularity as well, such as cell culture,4 polymerase chain reaction,5 and DNA sequencing.6 Progress in developing useful microfluidic systems will be benefited by simple, fast, and inexpensive approaches to fabrication. A new technique for fabricating microfluidic devices is reported in this note. * To whom correspondence should be addressed. E-mail:kweston@ chem.fsu.edu. (1) Vilkner, T.; Janasek, D.; Manz, A. Anal. Chem. 2004, 76, 3373-3385. (2) Clicq, D.; Vervoort, N.; Vounckx, R.; Ottevaere, H.; Buijs, J.; Gooijer, C.; Ariese, F.; Baron, G. V.; Desmet, G. J. Chromatogr., A A 2002, 979, 3342. (3) Backhouse, C. J.; Gajdal, A.; Pilarski, L. M.; Crabtree, H. J. Electrophoresis 2003, 24, 1777-1786. (4) Martin, K.; Henkel, T.; Baier, V.; Grodrian, A.; Schon, T.; Roth, M.; Kohler, J. M.; Metze, J. Lab Chip 2003, 3, 202-207. (5) Sun, K.; Yamaguchi, A.; Ishida, Y.; Matsuo, S.; Misawa, H. Sens. Actuators, B 2002, 84, 283-289.
7478 Analytical Chemistry, Vol. 77, No. 22, November 15, 2005
Silica-based substrates have traditionally been used in microfluidics because of their excellent optical properties and welldocumented (and predictible) surface properties.7 However, the fabrication of glass microchips is time-consuming, somewhat dangerous due to the use of hydrofluoric acid for etching, and relatively expensive, due in part to the required safety precautions. Interest has shifted to polymeric materials for microdevices due to the versatility, reduced cost, and potential for large-scale manufacturing.8,9 An array of fabrication techniques for polymeric substrates has appeared in the literature; these include hot embossing,10 soft lithography,11 injection molding,12 X-ray lithography,13 and laser ablation techniques.14 Excluding soft lithography, these fabrication methods require specialized instrumentation that is not always readily available. For production of many duplicate devices, none of the methods are as rapid or simple as the method we report here. Soft lithography, in which a liquid prepolymer, usually poly(dimethylsiloxane) (PDMS), is cast against a relief structure, usually photoresist on a silicon wafer, has become one of the most widespread techniques for prototyping microfluidic devices. The resulting patterned elastomer can be peeled away from the template and sealed to a flat substrate to form a complete device. Cured PDMS is optically clear, reproduces the template with high fidelity, and can be reversibly sealed to any flat, smooth substrate or, with some surface treatment, irreversibly sealed to glass or another sheet of PDMS. Unfortunately, PDMS has some disadvantages. For loading PDMS devices with aqueous solutions, PDMS surfaces are generally oxidized before use, making them hydrophilic and negatively charged. The surface charge density yields devices with electroosmotic flow (EOF) characteristics that are comparable to glass devices. 15 However, the PDMS surface (6) Liu, S. R. Electrophoresis 2003, 24, 3755-3761. (7) Harrison, D. J.; Fluri, K.; Seiler, K.; Fan, Z. H.; Effenhauser, C. S.; Manz, A. Science 1993, 261, 895-897. (8) Becker, H.; Locascio, L. E. Talanta 2002, 56, 267-287. (9) Soper, S. A.; Ford, S. M.; Qi, S.; McCarley, R. L.; Kelly, K.; Murphy, M. C. Anal. Chem. 2000, 72, 642A-651A. (10) Martynova, L.; Locascio, L. E.; Gaitan, M.; Kramer, G. W.; Christensen, R. G.; MacCrehan, W. A. Anal. Chem. 1997, 69, 4783-4789. (11) McDonald, J. C.; Duffy, D. C.; Anderson, J. R.; Chiu, D. T.; Wu, H. K.; Schueller, O. J. A.; Whitesides, G. M. Electrophoresis 2000, 21, 27-40. (12) McCormick, R. M.; Nelson, R. J.; AlonsoAmigo, M. G.; Benvegnu, J.; Hooper, H. H. Anal. Chem. 1997, 69, 2626-2630. (13) Ford, S. M.; Davies, J.; Kar, B.; Qi, S. D.; McWhorter, S.; Soper, S. A.; Malek, C. K. J. Biomech. Eng.-Trans. ASME 1999, 121, 13-21. (14) Roberts, M. A.; Rossier, J. S.; Bercier, P.; Girault, H. Anal. Chem. 1997, 69, 2035-2042. (15) Hu, S. W.; Ren, X. Q.; Bachman, M.; Sims, C. E.; Li, G. P.; Allbritton, N. L. Langmuir 2004, 20, 5569-5574. 10.1021/ac050847j CCC: $30.25
© 2005 American Chemical Society Published on Web 10/01/2005
is not stable over time and tends to adsorb solutes onto the surface, thus degrading resolution in PDMS devices used for separations. Since hydrophobic solutes tend to penetrate into the PDMS, there can be significant sample loss. When used with fluorescence detection methods, the retention of fluorescent sample in PDMS can cause significant background and consequent loss of detection sensitivity. Rigid plastics such as poly(methyl methacrylate) (PMMA) and polystyrene (PS) tend to exhibit more robust physical and chemical properties, such as stable EOF over time and over a broad range of pH’s and decreased surface adsorption. In this report, we present a simple method for micropatterning flat PMMA and PS substrates that can be used to build microfluidic devices. Our approach is based on a method reported by Rodriguez et al. that used patterned PDMS as a template to etch a channel design into a glass substrate.16 The etching solvent is passed over a substrate in a spatially selective manner that is defined by the pattern of the PDMS template. In our case, the process utilizes the solubility of PMMA and PS in organic solvents. The method could also be applied to pattern a wide range of plastics that are commercially available as polished sheets. This method alleviates difficult fabrication steps usually associated with plastic microfabrication and should be beneficial to a variety of researchers. EXPERIMENTAL SECTION Reagents and Equipment. The silicon wafers were obtained from Silicon Inc. (Boise, ID); the SU-8 2100 photoresist was purchased from Micro Chem Corp. The PDMS elastomer kit (Sylgard 184) was from Dow Chemical. An oil-free vacuum pump was used to draw solvents through the PDMS/PMMA or PDMS/ PS channels (model 2511B-01, Welch Vacuum). The Acrylite PMMA sheets used were from CYRO Industries (Rockaway, NJ), and clear PS sheets were from Plaskolite Inc. (Columbus, OH). A KLA-Tencor P-15 profilometer was used to measure the etching depth of the PMMA channels. A JEOL 5900 scanning electron microscope (SEM) was used to image the surface contours. To obtain a cross-section perspective of the PMMA microdevice, we first used a band saw to make a partial cut into the PMMA slab perpendicular to the etched channel from the opposite side of the etched surface and then applied pressure to fracture the PMMA along the cut. This allowed us to image a cross section of the channel. Prior to the SEM, all samples were sputter coated with gold. PDMS Template Fabrication. The PDMS fabrication process we use has been described in detail.17,18 Briefly, a photoreduction method yields a 35-mm film with the desired pattern created in drafting software. A thin layer of photoresist, SU-8 2100 (Micro Chem Corp.), is spin cast onto an etched silicon wafer and the 35-mm photomask is placed directly on top of the photoresist during exposure to UV light. The positive surface relief of the original photomask is reproduced in PDMS by curing the prepolymer and curing agent (10:1 ratio) for 2 h at 60 °C against (16) Rodriguez, I.; Spicar-Mihalic, P.; Kuyper, C. L.; Fiorini, G. S.; Chiu, D. T. Anal. Chim. Acta 2003, 496, 205-215. (17) Duffy, D. C.; McDonald, J. C.; Schueller, O. J. A.; Whitesides, G. M. Anal. Chem. 1998, 70, 4974-4984. (18) Deng, T.; Wu, H.; Brittain, S. T.; Whitesides, G. M. Anal. Chem. 2000, 72, 3176-3180.
the patterned SU-8 master. The cross-linked elastomer is then gently peeled away leaving open channels with high aspect ratios. The PDMS template can then be used to pattern many PMMA sheets without being consumed or degraded by the process. Plastic Etching with a PDMS Template. Prior to etching, all surfaces were cleaned by rinsing with ethanol and DI water for ∼2 min and dried in a stream of N2 gas. The PDMS template is gently placed onto a flat sheet of either PMMA or PS. The contact yields a conformal seal capable of confining the etching solution. We apply a constant vacuum of 77 kPa (11 psiv) to flow solvents through the PDMS channel. Figure 1A shows an illustration of the setup used. Experiments were done in which the etchant was pumped through the PDMS-plastic construct continuously for defined periods of time, or the etchant was loaded into the channels, allowed to etch without flow, and then emptied after a period of time. After etching, the channels were rinsed by flowing ethanol through the channels for 3 min. The PDMS template could be used repeatedly. One of our templates was used >50 times without an observable change in the quality of the microchannels fabricated. Sealing Plastic Sheets to Glass. Several methods for enclosing channels patterned in plastics have been reported.10,12-14,16 We chose a method that uses a thin glass coverslip because it enables us to use a high numerical aperture microscope objective for maximum light collection efficiency from the completed devices. First, fluid reservoirs were created by drilling a ∼3-mm-diameter hole through the patterned PMMA or PS sheets at the end of the etched channels. The open channels are then sealed to a glass coverslip using an ultraviolet curing adhesive, SK-9 (Sommers Optical). Glass coverslips (43 × 50, no. 1) were cleaned by rinsing in acetone, then ethanol, and then filtered water and dried in a stream of N2 gas. SK-9 is diluted 1:1 in methanol to reduce the viscocity, and 1 mL is placed on the center of a coverslip. The coverslip is spun at 1500 rpm for 40 s to create a thin layer of adhesive. An etched PMMA or PS sheet is then pressed lightly against the thin adhesive. Although it is optional, we rinsed away most of the SK-9 adhesive within channels by applying a vacuum to one reservoir and adding 10 µL of methanol to the other. Finally, the adhesive was cured under UV illumination using the bulb/ power supply from a Stinger UV40 electronic insect killer for 1 h. RESULTS AND DISCUSSION We found that the smoothness of the etched plastic walls depends strongly on the solvent used. For PMMA, the best combination of wall smoothness and etch rate was obtained when using mixtures of acetone and ethanol. In one set of experiments, we varied the percentage of acetone in the etching solution from 0 (v/v) to 100% in 5% steps and inspected the wall smoothness and etch depth of channels. The smoothness of the channel surface was assessed by SEM, profilometry, and optical microscopy. Representative examples of these are shown in Figure 1B, Figure 1C, and Figure 1D, respectively. Below 20% acetone, we observed no etching. Above ∼65% acetone, the rate of etching was high but yielded rough and irregular channel surfaces. Rough channel walls cause light scattering and band broadening in separations and are generally undesirable. Between ∼25 and 65% acetone, uniform channels with smooth surfaces could be obtained. A 50% acetone solution was determined to be optimal and reliably gave a fast etch rate and smooth channel surfaces. Analytical Chemistry, Vol. 77, No. 22, November 15, 2005
7479
Figure 1. Schematic of the etching process and representative analysis of an etched channel by SEM, profilometry, and optical microscopy. Panel A illustrates the setup used to etch the PMMA channels. Each channel in panels B-D was made by pumping a 50% acetone/ethanol solution through a 100 µm wide by 60 µm tall channel in a PDMS template for 15 min. Panel B shows a SEM image of the etched PMMA channel, panel C shows the channel profile, and panel D shows an fluorescence image of the channel when a solution containing 200-nm fluorescent polymer beads was passed through a sealed PMMA channel.
Figure 2. Rate of PMMA etching in the vertical and horizontal directions characterized by profilometry. Panel A shows profiles taken along the length of a 12-mm straight channel. The depth is constant, and the width is nearly constant along the full length of the etched channel. Panel B shows profiles of seven different channels etched for 0.5, 1, 2.5, 5, 10, 15, and 20 min. Panel C shows the channel width (at the PMMA surface and half the depth) and depth as a function of etching time. The vertical etch rate is 1.61 µm/min, and the horizontal etch rate is 6.8 µm/min. Panel D shows the aspect ratio of etched channels as a function of the etch time calculated using the width at half the depth and the vertical depth. The aspect ratio decreases as the channel become rounded at the bottom.
Using the acetone/ethanol (1:1) solution to etch channels in PMMA, we investigated the uniformity of etched channels along the length of the channels. We also characterized the etch rate and aspect ratio of channels as a function of etch time. For both sets of measurements, we used a PDMS template of a single straight channel with a width of 100 µm and height of 60 µm. Figure 2A shows five profiles of the same channel (etched time 10 min) at 2.5-mm intervals along the 12-mm length of the channel. The first and last profiles are black, and the middle three are gray. 7480
Analytical Chemistry, Vol. 77, No. 22, November 15, 2005
The slightly wider black profile was taken 1 mm from the reservoir where the etchant enters, and the slightly narrower black profile was taken 1 mm from the reservoir where the etchant exits. No measurable variation in the etch depth along the length of the channel is observed, and there is only a very slight decrease in the width from one end to the other. The slight decrease in channel width might be due to a reduced etch rate as solvent becomes loaded with dissolved PMMA. Another plausible explanation is that the vacuum applied at the exit reservoir helps to
hold the PDMS more tightly against the PMMA substrate at that end, thus resulting in slightly narrower channels. In the second series of measurements, we characterized the depth and width of channels in PMMA as a function of etch time by collecting profiles of seven different channels that formed by continuously flowing the solvent mixture for 0.5, 1.0, 2.5, 5, 10, 15, and 20 min. These profiles are shown in Figure 2B. The depth and width of each channel versus etch time is plotted in Figure 2C. The filled circles in the plot show the width of channels at the PMMA surface (the widest part), and the filled squares show the width of channels at half their depth. The open squares correspond to the vertical etch depth (note the scale for the depth is along the right axis). All three of these parameters scale linearly with etch time. Under the constant flow rate used (∼0.8 mL/min), the vertical etching rate was 1.6 µm/min. Note that, even for the very short etch time of 30 s, the width at the PMMA surface is 160 µm, so 60 µm larger than the channel in PDMS. This implies that, even for very short etch times, the solvent is able to penetrate under or through the PDMS by 30 µm on either side of the channel. This would also help to explain the surprisingly fast horizontal rate of etching. Based on the rate of increase in the width of the channel at the top surface of the PMMA, the rate of etching in the horizontal direction is 6.8 µm/min. Unfortunately, the fast rate of etching under the PDMS will severely limit the aspect ratio one can obtain using this method in its current form. We plot the effective aspect ratio, the ratio of the width at halfdepth to the depth, as a function of etch time in Figure 2D. As the channel becomes rounded at the bottom with longer etch times, the effective aspect ratio increases to a value of 10. One concern we had in using this method to fabricate microfluidic devices was that complex structures may not be reproduced with fidelity due to differences in the rate of flow of the etchant though the PDMS template. To examine this possibility, we etched a channel geometry with irregular flow paths; we designed a channel with a wide region that abruptly converges to a narrow region and then expands again. The linear flow rate in the narrow region will be much higher than in the wide regions, and one might expect the narrow region to etch faster and thus become deeper and wider relative to the wide sections. The geometry of the channel is shown by the SEM image in Figure 3A. The SEM image illustrates the consistency in the smoothness of etched surfaces in both the wide and narrow channel regions. Profiles across the narrow (gray traces) and wide regions (black traces) of the channel in this device are shown in Figure 3B; the etch time was 5 min with constant applied vacuum. The profiles from the channel clearly show that the etch depth is the same in both the narrow and wide sections despite the fact that the linear flow velocities in the two regions are not. This indicates that the etch rate is independent of the linear flow rate and depends only of the etch time, at least for the relatively high rates of flow we are using. A similar result was reported by Yokokawa et al. using a PDMS template for patterned etching of glass with hydrofluoric acid.19 They observed that, above a threshold flow rate, the rate of etching was independent of flow rate. We confirmed this conclusion with another experiment in which we filled the channel with 50% acetone, allowed the PMMA to be exposed to the solvent without flow for 5 min, and then emptied the channel by pumping (19) Yokokawa, R.; Takeuchi, S.; Fujita, H. Analyst 2004, 129, 850-854.
Figure 3. Acetone etching method applied to a channel geometry for which the linear flow rate varies with the position along the length of the channel. Panel B shows a SEM of the channel geometry and illustrates the constant depth and smoothness of the surfaces. Panel A shows profiles along the length of the channel. The black traces are profiles in the wide region while the gray traces are profiles of the narrow central region. The etching depth remains constant regardless of the channel geometry (flow rate).
N2 gas through it. In this case, the etched PMMA profile was nearly identical (
