Anal. Chem. 2004, 76, 1175-1183
A Packaging Technique for Polymer Microfluidic Platforms Siyi Lai, Xia Cao, and L. James Lee*
Department of Chemical Engineering, The Ohio State University, Columbus, Ohio 43210
A new technique, resin-gas injection, has been developed for bonding and surface modification of polymer microfluidic devices. This method can easily bond biochips with complex flow patterns. A cascade micromixer and a multichannel DNA sequencing chip were demonstrated experimentally. By adding surface modification agents, the interfacial free energy of the substrate with water can be controlled. Local modification of the channel surface can also be achieved through sequential resin-gas injection in conjunction with a masking technique. For application, this technique is used to form a layer of dry monolithic stationary hydrogel on the walls of a microchannel, serving as a sieving material for electrophoresis separation of DNA fragments. The reagent loading and the electrophoresis separation efficiency of this technique were compared experimentally with the conventional linear polymer solution method used in the microchannel-based DNA sequencing process. It is found that our method has the advantages of more user-friendly operation, easier and faster sample loading, but slightly less separation efficiency. Polymers have a great potential to be used for BioMEMS applications because many polymers are low cost, can be processed easily, and possess a broad range of physical and chemical properties. By combining various bulk or surface machining techniques and conventional manufacturing methods such as silicone rubber casting, injection molding, hot embossing, and reaction injection molding, polymer platforms containing complex flow patterns at the microscale can be fabricated.1 After fabrication, the platforms need to be sealed in order to perform microfluidic functions. The channel surface may need to be modified in order to provide proper physical, chemical, or biological functions. Examples are hydrophilicity for easy reagent loading, hydrophobicity for capillary valving, immobilization of protein, enzyme, or immunomolecules, and electric charge on the surface (for strong electroosmotic flow, a high static charge is preferred, while a charge-free surface is preferred for pressure flow and electrophoresis). Currently, packaging (bonding, surface modification, and reagent loading) is still a challenging issue in the design and fabrication of polymer-based microfluidic devices. Bonding (i.e., sealing a device with a lid) between silicon and silicon or other materials (glass, metal, etc.) is well developed * To whom correspondence should be addressed. Phone: (614) 292-2408. Fax: (614) 292-9271. E-mail:
[email protected]. (1) Becker, H.; Gartner, C. Electrophoresis 2000, 21, 12-26. 10.1021/ac034990t CCC: $27.50 Published on Web 01/09/2004
© 2004 American Chemical Society
and can be achieved by different methods such as anodic bonding, fusion bonding, eutectic bonding, and adhesive bonding.2 These approaches involve high temperature, high pressure, or high voltages. Among them, only adhesive bonding can be applied to polymer-based microfluidic devices. When the adhesive bonding method is used, care needs to be taken in order to prevent the adhesive from flowing into the microchannels. Several techniques such as lamination (adhesive tape, thermal adhesive film),3-5 thermal (IR, hot plate, laser) bonding, ultrasonic welding, solvent bonding (i.e., partially dissolve the bonding surfaces and then evaporate the solvent),6-8 and others1,9-13 have been used on polymer microdevices in academia and industry. In general, these bonding techniques alter the surface of the microdevices by using external forces or energies. The same driving force that allows the bonding also tends to deform the microfeatures. They are mainly applicable for relatively large microchannels (several hundreds of micrometers to millimeters). For poly(dimethyl siloxane) (PDMS) microfluidic systems, the oxygen plasma treatment has been widely used. In this method, the oxygen plasma was applied to activate the PDMS surface. The PDMS can then form irreversibly sealing to itself, glass, silicon, polyethylene, and polystyrene, but not to poly(methyl methacrylate) (PMMA), polyamide, or polycarbonate.14 During the electrophoresis separation, the separation channel is usually filled with sieving materials (such as polyacrylamide or agarose gel). The sieving material is one of the most important components in DNA sequencing analysis because it determines the migration behavior and the resolution of DNA separation. Linear polymer solutions were widely used in the microchip-based electrophoresis separation. Difficulties are often experienced (2) Schmidt, M. A. Proc. IEEE 1998, 86, 1575-1585. (3) Rossier, J. S.; Schwarz, A.; Reymond. F.; Ferrigno, R.; Bianchi, F.; Girault, H. H. Electrophoresis 1999, 20, 727-731. (4) Dreuth, H.; Heiden, C. Mater. Sci. Eng. 1998, C5, 227-231. (5) McCormick R. M.; Nelson, R. J.; Alonso-Amigo, M. G.; Benvegnu, D. J.; Hooper, H. H. Anal. Chem. 1997, 69, 2626-2630. (6) Lum, P.; Greenstein, M. U.S. Patent 5,932,315, 1999. (7) Metz, S.; Holzer, R.; Renaud, P. LabChip 2001, 1, 29-34. (8) Glasgow, I. K.; Beebe, D. J.; White, V. E. Sens. Mater. 1999, 11, 269-278. (9) Becker, H.; Dietz, W. Proc. SPIE-Int. Soc. Opt. Eng. 1998, 3515, 177182. (10) Gandhi, K.; Dubrow, R. S.; Bousse, L. J. U.S. Patent 6,123,798, 2000. (11) Lee, L. J.; Madou, M. J.; Koelling, K. W.; Daunert, S.; Lai; S.; Koh, C. G.; Juang, Y.-J.; Lu, Y.; Yu, L. Biomed. Microdevices 2001, 3, 339-351. (12) Madou, M. J.; Lee, L. J.; Koelling, K. W.; Lai, S.; Koh, C. G.; Juang, Y.-J.; Yu, L.; Lu, Y. SPE ANTEC Proc. 2001, 59, 2534-2538. (13) Robert, M. A.; Rossier. J. S.; Bercier, P.; Girault, H. Anal. Chem. 1997, 69, 2035-2042. (14) Duffy, D. C.; McDonald, J. C.; Schueller, O. J. A.; Whitesides, G. M. Anal. Chem. 1998, 4974-4984.
Analytical Chemistry, Vol. 76, No. 4, February 15, 2004 1175
during loading the sieving materials because the presence of linear polymer greatly increases the viscosity of the solution. Air bubbles tend to be trapped in the microchannel, which is problematic in the electrophoresis process. External pumping (vortex or vacuum) is often necessary to assist the loading of sieving material and eliminate the air bubbles in separation channels. This paper presents a resin-gas injection technique for bonding and surface modification of polymer-based microfluidic platforms. By applying the masking technique, local modification of the channel surface can be achieved through sequential resingas injection. This approach is also used to form a layer of crosslinked polyacrylamide gel on the walls of the microchannel, serving as a sieving material for DNA separation by electrophoresis. The sieving material loading, swelling behavior of hydrogels, and electrophoresis efficiency of this new technique were investigated and compared experimentally with the linear polymer solution method on a DNA sequencing chip. EXPERIMENTAL SECTION Reagents. Bonding and Surface Modification. Hydroxyethyl methacrylate (HEMA, monomer), 2,2-dimethoxy-2-phenylacetophenone (Irg 651, photoinitiator), and sodium dodecyl sulfate (SDS, surfactant) were purchased from Aldrich (Milwaukee, WI). The probe fluids for contact angle measurements include distilled water and ethylene glycol purchased from Fisher Scientific (Fairlawn, NJ). Poly(ethylene glycol) with molecular weight of 4000 (PEG-4000) was donated by Union Carbide, now Dow Chemical (Danbury, CT). DNA Separation. Acrylamide (electrophoresis grade, monomer), a hydroxyethyl cellulose (HEC, M h V ) 720 000, sieving material), a poly(ethylene glycol) with molecular weight of 4 × 106 (PEG 4M, sieving material), and thiazole orange (TO, DNA probe) were purchased from Aldrich. The cross-link agent for acrylamide, N,N′-methylenebisacrylamide (bis, electrophoresis grade), was purchased from Fisher Scientific. A DNA standard (ΦX174 HaeIII digest, 381 µg/mL) and the 5× Tris-BorateEDTA (TBE) buffer solution were purchased from Sigma-Aldrich (St. Louis, MO). Platform Fabrication. Platforms with micrometer-sized channels were fabricated using a LIGA-like technique.11,12 For a singlechannel platform (dimensions are described in Figure 1a), the mold insert was made using a computer numerically controlled (CNC) machining on a steel block. Two microchip designs were used for DNA separation. A simple design (chip A) is shown in Figure 1b, where the separation channel is 127 µm wide, 100 µm deep, and 5.3 cm long. Four reservoirs were connected at the end of each microchannel for buffer solution, DNA sample, and wastes, respectively. They also provide access to the electrodes. A double-T structure was adopted for DNA sample injection. The detection can be carried out at any point between the sample injection point and waste reservoir 4. Therefore, the actual separation length ranges from 0 to 5.3 cm. The second design is for multiple DNA separations as shown in Figure 1c (chip B). In this design, a T structure was adopted for sample injection, and the separation channel dimension is 127 µm wide, 100 µm deep, and 8.3 cm long. This design is able to separate six DNA samples (loaded in reservoirs 2). The two DNA sequencing microchips were fabricated by the CNC machining on 3.2-mm-thick PMMA plates. 1176 Analytical Chemistry, Vol. 76, No. 4, February 15, 2004
Figure 1. Microfluidic devices used in the study: (a) a single channel, (b) DNA sequencing chip A, and (c) DNA sequencing chip B (reservoirs: 1, buffer solution; 2, sample; 3, sample injection waste, and 4, waste. Point A, sample injection point).
Bonding and Surface Modification. For a simple microfluidic platform as shown in Figure 1a, the microfluidic platform and a lid were encapsulated in a holder. The top plate of the channel holder was made of a 7.5 cm × 4.8 cm × 1.2 cm aluminum block with a view window at the center. A 0.5 cm wide × 0.6 cm deep groove was cut around the view window, so that a channel insert could be placed inside the groove. The bottom holder was made of a 7.5 cm × 4.8 cm × 2.1 cm PMMA block. After the microchannel insert was placed inside the aluminum groove, the top and bottom holders were tightened with bolts. For more complex and larger microfluidic platforms, as shown in Figure 1b and c, the lid and the platform were tightened together by C-clamps. A nitrogen gas cylinder was used as the pumping unit, and a gauge was connected to monitor the pressure. The fluid flow inside the microchannel was monitored by a video system consisting of a Cohu video camera (1.27-cm view detector area), a TV monitor, and a 6.5× zoom lens (D. O. Industries) with a 5× adapter and a 2× eyepiece. With this setup, magnifications greater than 400× can be achieved at a working distance of more than 2.54 cm. After the experiment, frames were grabbed from the motion video by commercial video software Adobe Premiere 5.0. The schematic of bonding by resin-gas injection is shown in Figure 2a. After the microfluidic platform and the lid were encapsulated in the holder, several drops of HEMA mixed with Irg 651 and a surface modification agent were injected into the platform to fill the micrometer- and millimeter-sized channels and reservoirs, as well as the gap between the platform and the lid. Nitrogen gas was then pumped in to replace most of the resin inside the channels and reservoirs. The pressure of the gas ranged from 140 to 830 kPa, depending on the channel dimensions. A vacuum pump may also be used to replace the gas cylinder, and
electrophoresis, 1× TBE buffer solution with 5 µM TO was loaded into the microchannels to swell the dry polyacrylamide gel layer left on the walls of the microchannels. Characterization of Bonding and Surface Modification. The bonded microfluidic device was filled with food dyes to detect any leakage. The cross section was examined by a scanning electron microscope (Philips XL 30, Philips Electron Optics), and the thickness of the poly(hydroxyethyl methacrylate) (PHEMA) coating was then measured. The dimension and surface roughness of the fabricated microchannels were characterized by an optical profilometer (WYKO NT3300 Profiling System, Veeco Metrology Group, Tucson, AZ). The contact angle measurement between a solid surface and water can be used to determine the hydrophilicity of the surface. The advancing and receding contact angles between the surface of the microfluidic device and probe fluids (distilled water and ethylene glycol) were measured by a dynamic contact angle analyzer (DCA-322, Cahn Instruments Inc.). For each sample, the measurement was made at least three times. The substrate surface was cleaned and dried before each measurement, and precautions were taken to prevent any contamination on the surface. The fluids were changed after each measurement to avoid any contamination. Two components of surface free energy of the substrate, polar (γsp) and dispersion (γsd) components, were calculated by using the geometric two-liquid method (DCA application software 1.0B). The interfacial free energy of the solid surface with water (γsw) was calculated by using the Good-Girifalco equation15 as follows:
Figure 2. Schematic of (a) resin-gas injection bonding (1, resin injection; 2, gas injection; 3, resin curing) and (b) local surface modification.
the resin inside the microchannel was removed by applying vacuum (12-25 cmHg). The remaining resin was cured by ultraviolet light (Mineralight lamp UVGL-58, Upland, CA). Other resins such as epoxy and unsaturated polyester can also be used for bonding. In addition to photo cure, resins can also be cured by thermal or redox initiation. Since HEMA is relatively hydrophilic, can bond PMMA well, and has been widely used in various medical applications, it was chosen as the material for boding biochips in this study. Local Surface Modification. Figure 2b shows the schematic of using the resin-gas injection method for local surface modification. The experimental setup and procedure were the same, except that prior to curing of the remaining resin mixture, a mask (e.g., a photomask used in the lithography process) was placed between the lid plate and the UV source. The mask blocked out the UV light locally, so that the resin mixture in the locations under the mask would not cure. After the curing of the unmasked parts was complete, the uncured resin mixture was removed by bleaching the surface with water or other solvent similar to the development step in the photolithography process. Gel Loading. The monomer gel solution was prepared by mixing acrylamide, cross-link agent bis, and photoinitiator Irg 651 in 1× TBE buffer solution. The gel loading process followed the same resin-gas injection technique after the platform was bonded to the lid. After UV exposure, the microchip was placed in an oven at 70 °C for 1∼2 h to achieve complete gel reaction. Prior to the
γsw ) γs + γw - 2(γds γdw)1/2 - 2(γps γpw)1/2
where γs and γw are surface free energy of the platform and water, respectively, and γwp and γwd are polar and dispersion components of surface free energy of water with values of 52 and 20.8 mJ/m2, respectively. The UV photopolymerization of HEMA and acrylamide was studied by using differential photocalorimetry (DPC) (DSC2920, TA Instrument) at 365 nm. DNA Separation. The DNA standard was diluted in a 0.2× TBE buffer solution to 10 µg/mL from the stock solution (381 µg/mL). The TO dye was prepared in a 100 µM concentration in a 1× TBE buffer solution. It was further diluted to 5 µM in the sieving (HEC or PEO) solution. Voltages were applied to the reservoirs via a high-voltage power supply system, which consists of a low-voltage, programmable power supply (72-6695, Tenma Test Equipment, Springboro, OH), two miniature dc to highvoltage converters (G10 for output up to 1000 V and G20 for output up to 2000 V, Emco High Voltage Corp., Sutter Creek, CA), and a double pole double throw relay (0700A, MCM Electronics). Detection was carried out on-chip using an inverted fluorescence microscope (Nikon Eclipse TE2000-U). A 100-W mercury light source with a 490-nm filter and a dichroic mirror was used as an excitation source. The fluorescence signal was obtained through a dichroic mirror and a 510-nm filter. Images at the detection point were recorded sequentially by a 12-bit high-resolution monochrome digital camera system (CoolSnap HQ, Roper Scientific). The fluorescence intensities were extracted directly from the (15) Girifalco, L. A.; Good, R. J. J. Phys. Chem. 1957, 61, 904-909.
Analytical Chemistry, Vol. 76, No. 4, February 15, 2004
1177
Table 1. Voltage Applied and Sequences for Electrophoresis Separation on DNA Sequencing Chips A and B voltage applied (V) process
1
injection pulling back separation
float 0 0
injection pulling back separation
float 0 0
2
3
4
time (s)
Chip A 400 0 500 500 float float
float 1100 1100
25 5 150
Chip B 600 0 600 600 float float
float 1600 1600
50 5 600
Table 2. UV Intensities under Different Lid Materials UV intensity (mW/cm2) lid
365 nm
320-500 nm
none 1-mm glass 1.5-mm PDMS 1.59-mm PMMA 1.59-mm PC
21 18.87 19.34 3.54 0.031
21 18.68 19.63 14.49 10.89
Figure 4. Fluorescence background (excitation at 490 nm) of several platform materials.
Figure 3. UV transmittance spectra of PC, PMMA, PDMS, and glass.
sequential video images by using an image analysis software (Fryer Metamorph image analysis system). The intensity was an average over a 100 µm × 30 µm detection area. As described in the literature, the electrophoresis separation of the DNA fragments follows three steps: injection of DNA samples, pulling back to form the sample plug, and the separation. The voltages applied and times for each step are summarized in Table 1. RESULTS AND DISCUSSION UV Photopolymerization. During UV curing of the resin, the UV light has to penetrate through the lid in order to cure the resin in the microchannel. The lid acts as a filter. Therefore, optical properties of lid materials play an important role during photo curing of the resin. The UV transparency of polycarbonate (PC), PMMA, PDMS, and glass was examined and is plotted in Figure 3. It can be seen that PC is almost opaque for the UV light below a wavelength of 390 nm. Therefore, it is not suitable for the lid material. The PMMA shows good transparency down to 365 nm. PDMS provides the best UV transparency in the short-wavelength range. The actual UV intensity under the lid was measured and is summarized in Table 2. The inherent autofluorescence background of the polymeric materials can be problematic for fluorescence detection in the electrophoresis separation of DNA. The fluorescence background of glass and several polymeric substrates was examined at the excitation wavelength of 490 nm (the same excitation wavelength used in the DNA separation) as shown in Figure 4. The emitted 1178 Analytical Chemistry, Vol. 76, No. 4, February 15, 2004
fluorescence intensities from 510 to 550 nm were recorded. The results show that PMMA and PDMS have a comparably low background, similar to that of glass. On the other hands, PC and polystyrene (PS) have very strong fluorescence background at 510 nm, higher than that of most DNA fragments. Because of its good combination of optical clarity, mechanical strength, and low cost, PMMA was selected as the device material in the microfluidic experiments. The fluorescence background of the bonding material may also cause problems for the detection. We found that PMMA-based microdevices bonded with PHEMA have comparable low-fluorescence background everywhere (∼170 arbitrary unit), indicating no interaction between the PHEMA layer and the dye (TO). DPC measurement of the heat flow during the UV photopolymerization indicated that the reaction of HEMA has almost finished after 120-s exposure to the UV light under a PMMA lid. The total heat of polymerization calculated from the area under the heat flow curve was ∼46 kJ/mol, while the heat of polymerization of HEMA reported in the literature was 50 kJ/mol.16 Assuming that the rate of the reaction is proportional to the reaction exotherm, the overall conversion of the photopolymerization was found to be 92%. DPC study of the UV photopolymerization of acrylamide shows that the polymerization of acrylamide has almost finished after 60 s under a PMMA lid, which is much shorter than the time for conventional redox polymerization (∼1 h).17 The measured heat of polymerization of acrylamide was ∼76 kJ/mol, while the heat (16) Ro ¨hm America, LLC product literature: http://www.rohmamerica.com/ Methacrylates/Monomers/aliphatics.html. (17) Brahmasandra, S. N.; Burke, D. T.; Mastrangelo, C. H.; Burns, M. A. Electrophoresis 2001, 22, 1046-1062.
Figure 5. (a) Top view of a bonded microchannel filled with food dye; SEM photos of cross-section view of (b) a bonded reservoir, (c) enlarged view of reservoir, (d) enlarged view of bonded reservoir, (e) a bonded microchannel (90 µm × 330 µm), and (f) top view of a bonded microchannel after delamination.
of polymerization of acrylamide from the literature was 82.8 kJ/ mol.18 The overall conversion of the UV-cured polyacrylamide was ∼91%, which is similar to that of the conventional redox polymerization.17 Resin-Gas Injection Bonding. A few drops of food dye were injected into the bonded microstructure through the inlet to detect any leakage and blockage of the microchannel. Food dye went through the inlet reservoir and the microchannel, reaching the output reservoir. A bonded channel (90 µm wide and 330 µm deep) tested by this method is shown in Figure 5a. Complete filling of the outlet reservoir indicates that there is no block inside the microchannel. The absence of food dye outside the reservoirs and microchannel indicates no leakage. Similar results were achieved in bonding of both DNA sequencing chips. A cross section (A-A) of a bonded reservoir (3 mm wide and 300 µm deep) is shown in Figure 5b. A close view of the reservoir corner before and after bonding is shown in Figure 5c and d. It can be seen that a thin layer of PHEMA resin remains in the inner surface of the reservoir and the corner is rounded after bonding. (18) U.S. Environmental Protection Agency. Toxic Substances Control Act (TSCA) Chemical Assessment Series; Washington, DC, April 15, 1981.
Figure 5e is a SEM photo of the cross section of a bonded channel (90 µm × 330 µm), showing that good bonding was achieved. The surface smoothness of the bonded microchannel was examined after the bonded lid and platform were delaminated in liquid nitrogen. A very smooth channel sidewall was achieved as shown in Figure 5f. In many applications, sharp corners in the microfluidic devices are undesirable since fluids may be trapped there.12 By using resin-gas injection bonding, the sharp corner of the reservoir can be rounded like a streamline, which may facilitate the fluid flow. Surface Modification. Table 3 compares the measured surface properties of PMMA with several PMMA surfaces treated by resin-gas injection. It can be seen that large contact angles were observed on the untreated PMMA surface and PMMA coated with PHEMA, indicating that these surfaces are hydrophobic. The interfacial free energies of PMMA surfaces with water decreased from 12.33 to 2.76 mJ/m2 after different surface treatments. The PMMA surface coated with a mixture of PHEMA and the PEG 4000 (90:10 by weight) has the lowest interfacial free energy with water. (19) Kaelble, D. H. Physical Chemistry of Adhesion; Wiley: New York, 1971.
Analytical Chemistry, Vol. 76, No. 4, February 15, 2004
1179
Table 3. Measured Water Contact Angles and Calculated Surface Free Energy of Treated and Untreated PMMAa materials
θA/water
γp (mJ/m2)
γd (mJ/m2)
γs (mJ/m2)
γsw (mJ/m2)
untreated PMMA coated with PHEMA coated with PHEMA and 0.5% SDS coated with PHEMA and 20% EG coated with PHEMA and 10% PEG 4000 glass19
70.03 ( 1.03 68.19 ( 0.71 58.15 ( 1.82 57.12 ( 1.68 53.01 ( 1.72 56
24.11 ( 1.68 17.16 ( 1.17 9.76 ( 2.52 12.68 ( 1.23 19.31 ( 1.28 19.2
12.85 ( 1.35 18.25 ( 1.10 34.57 ( 3.63 32.09 ( 1.25 29.91 ( 1.32 37.6
36.96 ( 0.47 35.41 ( 0.35 44.32 ( 2.04 44.77 ( 0.48 49.22 ( 0.91 56.8
12.33 8.31 4.10 3.44 2.76 1.03
a θ , the advancing contact angle between solid surface and distilled water; γp, polar component of surface free energy; γd, dispersion component A of surface free energy; γs, surface free energy; γsw, interfacial free energy of substrate with water ( ( the standard deviations).
Figure 6. (a) Schematic of experiment to test the effect of surface modification on reagent loading; snapshots of (b) a platform bonded by PHEMA, (c) a platform bonded by PHEMA and 10% PEG 4000, and (d) a platform with local surface modification.
To demonstrate the effect of surface hydrophilicity on reagent loading, a drop of food dye was placed on the top of the inlet reservoir of the bonded microfluidic platform and the flow of the food dye was monitored by a CCD camera from the top (Figure 6a). Figure 6b shows snapshots taken at different times of a PMMA platform bonded by PHEMA only. The food dye stayed at the inlet, and no flow was observed. Figure 6c shows snapshots for a similar platform bonded by PHEMA and 10% PEG 4000. It can be seen that, as time elapsed, the food dye flowed into the microchannel spontaneously from the inlet to the outlet reservoir. Local Surface Modification. Many researchers have applied lithography techniques to manipulate proteins, cells,20,21 and other (20) Kane, R. S.; Takayama, S.; Ostuni, E.; Ingber, D. E.; Whitesides, G. Biomaterials 1999, 20, 2363-2376.
1180
Analytical Chemistry, Vol. 76, No. 4, February 15, 2004
biological macromolecules22 on locally modified silicon surfaces. Materials other than photoresists have also been applied for micropatterning by UV polymerization techniques.23,24 Functional microfluidic components have been produced by patterning regions inside the microchannels. Using patterned hydrophobic regions inside a silicon microchannel, nanoliter-sized liquid can be accurately metered.25 The hydrophobic regions (21) Park, A.; Wu, B.; Griffith, L. G. J. Biomater. Sci. Polym. Ed. 1998, 9, 89110. (22) Li, Y.; Pfohl, T.; Kim, J. H.; Yasa, M.; Wen, Z.; Kim, M. W.; Safinya, C. R. Biomed. Microdevices 2001, 3, 239-244. (23) Ward, J. H.; Bashir, R.; Peppas, N. A. J. Biomed. Mater. Res. 2001, 56, 351-360. (24) Beebe, D. J.; Moore, J. S.; Bauer, J. M.; Yu, Q.; Liu, R. H.; Devadoss, C.; Jo, B.-H. Nature 2000, 406, 588-590. (25) Handique, K.; Burke, D. T.; Mastrangelo, C. H.; Burns, M. A. Anal. Chem. 2000, 72, 4100-4109.
Figure 7. Swell ratio of the polyacrylamide gel vs cross-linking ratio. Figure 8. Loading of sieving materials in the separation channel: (a) 0.75% HEC (Mw ) 720K) in an empty channel, (b) 1.0% PEO (Mw ) 4M) in an empty channel, and (c) TBE buffer solution in a channel prefilled with acrylamide gel.
inside the microchannel were patterned by bulk and surface machining on silicon substrate before the platform was bonded. Microvalves inside the microchannels were fabricated by liquidphase photopolymerization24 of hydrogels around prefabricated posts. The microvalves consist of a stimuli-responsive hydrogel structure that undergoes a volume change in response to changes in local pH. Swelling and deswelling of the hydrogel can regulate the fluid flow in microchannels. Patterns other than the circles for microvalves have also been demonstrated with a minimum feature size of 25 µm. Combining photolithography, photopolymerization, and resingas injection techniques, localized surface properties in the microchannels can be achieved to perform specific functions. An example is shown in Figure 6d. By local surface modification, the inlet reservoir and most of the microchannel were coated with PHEMA and 10% PEG 4000, while both the ends of the microchannel near the outlet reservoir and the outlet reservoir were coated only with PHEMA. Figure 6d shows snapshots of reagent loading on this platform. It can be seen that the food dye flowed into the microchannel spontaneously but stopped at the end of the microchannel near the outlet reservoir where the surface changed (the fourth and fifth snapshots) from hydrophilic to hydrophobic. By applying a small pressure at the inlet, the food dye could flow to the outlet reservoir (the sixth snapshot). Swelling of Acrylamide Hydrogel. The swelling ratio is an important property of hydrogels. In general, the swelling ratio, which is defined as the ratio of the weight of the swollen sample to the weigh of the dry matrix, is used to characterize the swelling behavior of hydrogels. Factors that affect the swelling ratio include the cross-linking ratio, chemical structure, and environmental conditions such as pH value and temperature.2626 The swelling ratio of the photopolymerized acrylamide gel in the 1× TBE buffer solution was measured as shown in Figure 7. It can be seen that the cross-linking ratio, the ratio of the cross-linking agent to the polymer repeating units, has a dominating effect on the swelling behavior. The higher the cross-linking ratio, the more crosslinking agent is incorporated in the hydrogel structure. Highly cross-linked hydrogels have a tighter structure and tend to swell less compared to the same hydrogels with a lower cross-linking ratio. The chemical structure of the polymer may also affect the swelling ratio. The equilibrium swelling ratio of poly(N-isopropy-
lacrylamide) cross-linked by bis in water increased from 7.61 to 16.8 when the molar ratio of N-isopropylacrylamide to bis increased from 10:1 to 100:1.27 Loading of Sieving Material. Figure 8 shows the progresses of loading three different sieving materials only by the capillary force in the single-channel DNA sequencing chip, chip A (Figure 1b). The 0.75% 720K HEC solution and 1.0% 4M PEO solution were loaded into an empty channel with dimensions of 127 µm × 100 µm. The TBE buffer solution was loaded into a channel of the same dimensions but coated with a layer of dried cross-linking polyacrylamide gel. The starting point of the separation channel shown in Figure 8 is ∼5 mm away from the sample injection point. The times for the PEO, HEC, and TBE buffer solutions to travel a 1-cm distance in the separation channels are 98, 48, and 2.43 s, respectively. It should be noted that the loading speed of each sieving material in the separation channel was not uniform through the process. The flow in the first 5 mm was much faster than that in the second 5 mm. By capillary force only, filling a linear polymer solution into a separation channel of 5-cm length took almost 10 min. Vacuum or syringe pump could assist the loading process. However, air bubbles may be introduced into the microchannel, which is problematic in the electrophoresis process. By coating a dry polyacrylamide gel on the surface of the separation channel, sample loading by capillary force becomes very simple and the loading time is very short because the TBE buffer solution has a much lower viscosity than the conventional sieving solutions. DNA Fragment Separation. When applying a resin-gas coating of polyacrylamide gel for DNA separation, it is critical that the coated polyacrylamide layer can swell and completely fill the cavity in the separation channel. For a polyacrylamide gel consisting of 10% solid content and 1% cross-linking agent, Figure 9a shows that the swollen gel after loading the buffer solution did not cover the entire cavity. There is a small channel at the center of the cavity. During DNA separation, most DNA molecules raced through this channel instead of the polyacrylamide because of low flow resistance. The swelling ratio of this gel is nearly 20 under free swelling,28 which should be more than enough to
(26) Peppas, N. A.; Bures, P.; Leobandung; W.; Ichikawa, H. Eur. J. Pharm. Biopharm. 2000, 50, 27-46.
(27) Zlatanic, A.; Petrovic, Z. S. ANTEC Proc. SPE 2001, 59, 3319-3323. (28) Furuhawa, H. J. Mol. Struct. 2000, 554, 11-19.
Analytical Chemistry, Vol. 76, No. 4, February 15, 2004
1181
Figure 10. Separation of DNA standard in chip A with cross-linked polyacrylamide gel (40% solid content, 1% cross-linking agent) as sieving material. DNA fragment sizes in number of base pairs are denoted above each peak. Detection at 3 cm.
Figure 9. Swelling of polyacrylamide gels: (a) 10% solid content, 1% cross-linking agent; (b) 40% solid content, 1% cross-linking agent; and (c) 12% solid content, 0.25% cross-linking agent with a cavity at center in 1× TBE buffer.
cover the entire cavity. However, the gel constrained in the separation channel undergoes stress buildup during swelling, which affects its ability to completely fill the cavity. As shown in Figure 9b and c, the gel with 40% solid content and 1% crosslinking agent can nearly fill the cavity and the gel with 12% solid content and 0.25% cross-linking agent can completely fill the cavity. No race tracking was observed for both gels during electrophoretic flow. The performance of this cross-linked polyacrylamide gel on DNA separation was evaluated by resin-gas coating of the gel on the wall of chip A. The gel with 40% solid content and 1% cross-link agent was chosen to prevent the race tracking of swollen gel matrix during DNA separation. The DNA standard used is ΦX174 HaeIII digest with 11 fragments ranging from 72 to 1353 base pairs. Figure 10 shows the fluorescence intensities of the separated DNA fragments in chip A. All fragments are resolved with the exception of the 271- and 281base pair doublet. To compare its separation efficiency with a linear polymer solution as sieving materials, 1.0% 4M PEO or 0.75% 720K HEC was used in chip A and chip B. Figure 11 shows the fluorescence intensities of separated fragments in the DNA standard. All fragments are resolved. The deficiency of the electrophoresis may result from the mesh structure in the gel. Recent studies showed that gel structure (solid and crosslinking agent dependent) influences mobility, diffusion, and 1182 Analytical Chemistry, Vol. 76, No. 4, February 15, 2004
Figure 11. DNA standard separated in (a) chip A with 1% 4M PEO and (b) chip B with 0.75% 720K HEC as sieving material. Detection at 3 cm.
dispersion of the DNA molecules and therefore the resolution of DNA separation.17,29-31 CONCLUSIONS A new technique, resin-gas injection, has been developed for bonding and surface modification of polymer-based microfluidic devices. This method successfully bonded microfluidic platforms from a simple single-channel structure to more complex patterns. By adding surface modification agents, the interfacial free energy of the substrate with water decreases. By applying the masking (29) Brahmasandra, S. N.; Ugaz, V. M.; Burke, D. T.; Mastrangelo, C. H.; Burns, M. A. Electrophoresis 2001, 22, 300-311. (30) Ugaz, V. M.; Brahmasandra, S. N.; Burke, D. T.; Burns, M. A. Electrophoresis 2002, 23, 1450-1459. (31) Ugaz, V. M.; Burke, D. T.; Burns, M. A. Electrophoresis 2002, 23, 27772787.
technique, local modification of the channel surface was achieved through cascade resin-gas injection. This approach was also applied to coat a layer of cross-linked polyacrylamide gel on the walls of the separation channel as the sieving material for electrophoresis. The gel coating method eliminated the tedious sieving solution preparation procedure, facilitated the sample loading process, and achieved DNA separation efficiency similar to that of the linear polymer solution except for the 271- and 281base pair doublet.
ACKNOWLEDGMENT The authors thank the National Science Foundation (DMI0102639) for financial support.
Received for review August 25, 2003. Accepted October 30, 2003. AC034990T
Analytical Chemistry, Vol. 76, No. 4, February 15, 2004
1183
