Polymer 45 (2004) 8469–8474 www.elsevier.com/locate/polymer
Preparation of porous polymer membranes using nano- or micro-pillar arrays as templates Xiaohu Yana, Guojun Liua,*, Michael Dickeyb, C. Grant Willsonb a
Department of Chemistry, Queen’s University, 90 Queen’s Cresent, Kingston, Ont., Canada K7L 3N6 b Department of Chemical Engineering, University of Texas at Austin, Austin, TX 78712-1167, USA Received 12 August 2004; accepted 30 September 2004 Available online 18 October 2004
Abstract ZnO or polystyrene (PS) pillar arrays were formed on solid substrates and used as templates for the formation of porous polymer membranes. The membranes were formed by casting a polymer solution in the inter-pillar space of these templates and evaporating the solvent. Dissolution of the pillars in a selective solvent resulted in thin films containing monodisperse micrometer-sized channels. Membranes produced using the pillar template technique showed high water permeability and high size selectivity. q 2004 Elsevier Ltd. All rights reserved. Keywords: Membrane; Pillars; Templated synthesis
1. Introduction Microporous membranes can be formed in numerous ways. Few methods yield membranes with cylindrical pores of uniform size that span the membrane like those found in the ‘track-etched’ polymer membranes [1] or in the ‘electrochemically-etched’ alumina membranes [2]. The track-etched membranes suffer, however, from relatively low porosity and thus low effluence. Recently methods have emerged that are capable of forming large arrays of small, cylindrical pores in polymer films through the use of block copolymers [3,4] or with imprint lithography [5,6]. The ability to create ever-smaller, monodisperse pores is enabling more sophisticated applications of porous materials in areas as diverse as drug delivery [7], enantiomer separation [8], DNA and biomolecule separation [9], lithography [10], chemical sensoring and catalysis [11] etc. The cylindrical pores of such membranes have been used recently also as ‘templates’ to grow various micro- and nano-structures including metal multiblocks [12]. In this paper, we report a simple method for creating mono-disperse cylindrical pores in polymer films based on * Corresponding author. Tel.: C1 613 533 6996; fax: C1 613 533 6669. E-mail address:
[email protected] (G. Liu). 0032-3861/$ - see front matter q 2004 Elsevier Ltd. All rights reserved. doi:10.1016/j.polymer.2004.09.073
‘reverse templating’. A polymer film is cast about a template of a sacrificial pillar array. The pillars are removed after film formation, leaving behind a porous membrane with cylindrical pores that span the complete thickness of the membrane (w5 mm). The pore size may be tuned by adjusting the diameter of the templating pillars from 100 nm to 10 mm or by shrinking the pore diameter by plating the walls with metal [13]. Although the preparation of only polymer membranes will be discussed, the process may be adaptable also to the preparation of sol–gel inorganic or other films. The concept of replica molding against a template for the formation of porous materials is not without precedent [14]. The most notable example is the use of micelles of lowmolar-mass [15] or block copolymer [16] surfactants as templates to prepare mesoporous bulk silica with threedimensional porous structures. The technique is, however, not readily amenable to generate thin films with macroscopically-aligned permeating channels [17]. Other popular templating techniques include the use of colloidal spheres [14,18] or water droplets [19] as templates to yield membranes with concave pores. There are numerous methods available for creating arrays of pillars for the template, such as photolithography or e-beam lithography [20]. We chose to form both organic
8470
X. Yan et al. / Polymer 45 (2004) 8469–8474
(polystyrene, PS) and inorganic (ZnO) pillar arrays by patterning methods that are feasible using common, inexpensive equipment. Arrays of PS pillars were formed by the amplification of electrohydrodyamic instabilities, a method developed by Chou et al. [21] and Russell et al. [22] that is adaptable to other polymeric materials. The ZnO micropillars were prepared on SnO2-coated glass plates following a method reported recently by Vayssierres and coworkers [23].
2. Experimental section 2.1. Materials Zinc nitrate hydrate, Zn(NO3)2$6H2O (98%), Hexamethylenetetramine, C6H12N4 (99%), 0.90 and 1.1 mm polystyrene latex spheres (10% in water) were purchased from Aldrich. Glass plates coated with polycrystalline F– SnO2 were purchased from Hartford Glass Inc. (USA). Polysulfone (Udelw P3500) was a generous gift from Solvay Advanced Polymers and was purified by precipitation fractionation using THF as solvent and methanol as precipitant. The weight-average molar mass Mw of the fractionated sample was determined by light scattering using the specific refractive index increment of 0.201 ml gK1 that we measured in THF and was 9.0!104 g molK1. The polydispersity index Mw/Mn of the sample was 1.42 as determined by size exclusion chromatography based on poly(methyl methacrylate) standards. The PS sample used had a bimodal distribution with Mw of the components at 200,000 and 4000 g molK1, respectively. Both PS (catalogue no. 33,164-1) and nylon 6/6 (catalogue no. 42,917-1, melt viscosity at 230–280) were purchased from Aldrich. 2.2. ZnO pillar preparation The SnO2-coated glass plates were cut into 5!5 cm2 pieces. The pieces were placed vertically against the inner wall of a beaker containing a 1 wt% aqueous detergent solution and boiled with stirring for 1 h to clean the surface. Before use, the plates were rinsed thoroughly by distilled water and methanol passed through a 0.1 mm filter and dried under nitrogen flow. For pillar formation, Zn(NO3)2$6H2O and an equal molar amount of C6H12N4 were dissolved in 200 ml of distilled water in a cylindrical bottle with a screw-on cap. The cleaned glass plate was inserted into the bottle and placed approximately vertically against the bottle wall. After capping, the bottle was placed in a regular laboratory oven with temperature set at 90 8C for a pre-determined period of time. The plate after removal was rinsed by distilled water. The plate was dried under ambient conditions overnight after wrapping with aluminum foil.
2.3. PS pillar preparation PS films were spun-cast from a 12.5 wt% PS solution in toluene on aluminum-backed polished silicon wafers. After baking at 90 8C for 60 s to remove residual solvent, a top electrode consisting of glass plate coated with 4-nm thick chromium was placed w5 mm above the polymer film. The capacitor device was assembled with the help of etched spacers, clamps and electric leads. A voltage of 200 V was applied across the gap while heating the film to 120 8C. The film was held at 120 8C for several minutes until the pillars appeared. The pillar structure was frozen by lowering the temperature and the two electrodes were pried apart with a razor. Most of the pillars adhered to the Cr-coated glass plate after the separation of the electrodes and the polymer film was thus cast on the glass plate to prepare eventually a membrane. 2.4. Membrane preparation A dam that was 2–3 mm tall was built from a roomtemperature-curing epoxy resin (Varian) along the periphery of a glass plate bearing ZnO pillars. After the dam was cured for 4–5 h, a calculated amount of PSf in THF at 3.3 mg mlK1 was injected through a 0.1-mm filter onto the plate. The plate was loosely covered with aluminum foil and the solvent was evaporated under ambient conditions for 1 h. To etch off the PSf spherical end caps on the pillars, the glass plate bearing the PSf film was soaked in THF/methanol (v/vZ1/1) for 20 s. This was followed by THF/methanol evaporation and the immersion of the plate into 0.5-M HCl for w5 h to dissolve the ZnO pillars. After ZnO dissolution, the membrane spontaneously detached from the glass plate. The membrane was rinsed by distilled water for several times and stored in water until use. Membranes were prepared from the PS pillar template similarly. Formic acid was used as the selective solvent for nylon 6/6 and THF was used to selectively dissolve the PS pillars. 2.5. Water permeability test The permeation cell consisted of two cylindrical tubes with a cross-sectional inner diameter of 0.481 cm. One of the tubes assumed the L-shape and the other was straight. The end of the horizontal portion of the L-shaped tube was flattened and ground, leaving a small opening with a diameter of 4.5 mm in the center. So was one end of the straight tube. The PSf membranes used in permeability tests were 4.3-mm thick as confirmed by SEM. To mount a membrane, the membrane was first sandwiched between two paraffin films that were 10 mm thick and had an opening with a diameter of 3.89 mm or an area of 0.119 cm2 in the center, which was taken as the permeation area. The composite film was lightly pressed and then mounted between the two halves of the cell. The two halves were held
X. Yan et al. / Polymer 45 (2004) 8469–8474
together by a clamp. The distilled water used was first ultrasonicated for 30 min using a Branson Model 1200 R-C instrument to remove air bubbles and then passed through a 0.2 mm filter to remove dust particles. Once the membrane channels were wetted by forcing water through them, water drained through the membranes spontaneously due to gravity or a height h difference between the two sides of the cell. In a given run, the flow rate defined as dh/dt decreased exponentially with h. At a given h, dh/dt decreased somewhat after each refilling of the L-tube with water and reached steady readings after some 12 cycles of draining and refilling. The permeation rate 2 in units l mK2 hK1 barK1 was calculated from the permeability data using
2Z
pr2 1 d ln h ! ! rg dt A
(1)
which was independent of h. In Eq. (1), A is the permeating area of the membrane, pr2 is the cross-sectional area of the L tube, r denotes water density, and g denotes the gravitational acceleration. 2.6. Separation of latex spheres A PSf membrane was sandwiched by two paraffin films with an opening of 0.397 cm2 (diameter 7.1 mm) in the center. The sandwiched membrane was used to compartment two half-cells. To one of the half-cell was added 20 ml of a latex solution containing 0.90-mm and 1.1-mm spheres at 0.23 and 0.16 mg mlK1. The solution was dialyzed against 20 ml of distilled water placed in the other half-cell. The liquid in the half-cells was constantly stirred. After 7 days, the solution from the permeate side was taken for SEM analysis.
8471
2.7. Scanning electron microscopy SEM images were obtained using a FEI XL310 ESEM scanning electron microscope. To obtain a side-view of the ZnO or PS pillars, the glass plate was cut with a diamond knife to narrow strips. The strips were mounted side-wise on a double-sided adhesive carbon tape (Electron Microscopy Science), which had been glued on to a stainless-steel sample support. To obtain a top-down view of the pillars, a small piece of the sample was mounted laterally on a sample support. To obtain images for the latex spheres, the latex solution was dropped on a glass slide cover and the solvent water was left evaporating under ambient conditions. The glass slide cover was then glued on a sample support. To image porous membranes, a membrane was first floated on water surface and picked up with a paraffin film. The supported film was then mounted on a double-sided adhesive tape and lightly pressed. Before viewing, all of the samples were coated with a 6 nm-thick platinum-gold layer using a Hummer I Sputter coater.
3. Results and discussion Scheme 1 re-iterates the steps involved in preparing the membranes. After pillar array preparation (A/B), a dam was built along the periphery of the substrate using an epoxy resin (B/C). A polymeric solution was cast into the resulting well and the solvent was evaporated, yielding a polymer thin film. If the membrane thickness was comparable with the height of the pillars, the pillars were end-capped by a hemispherical layer (C/D). The end caps were removed by briefly soaking the glass plate in a marginal solvent for the polymer. This also thinned out the rest of the membrane (D/E). Thin films with permeating channels were obtained by dissolving the pillars with a selective solvent (E/F). This section describes our success in achieving each of these steps and presents some preliminary membrane characterization results. 3.1. ZnO micropillar preparation The ZnO micro-pillar arrays were grown on SnO2-coated glass plates from the thermal decomposition of Zn(NO3)2 and hexamethylenetetramine (C6H12N4) at 90 8C following [24]: 90 8C
2ZnðNO3 Þ2 C C6 H12 N4 C 8H2 O # 2ZnO C 4NH4 NO3 C 6HCOH
Scheme 1. Membrane preparation procedure.
The nucleation of ZnO crystals occurred more readily on the glass surface, because the interfacial energy between the ZnO crystals and the substrate was lower than that between the crystals and the solution. Following nucleation, the
8472
X. Yan et al. / Polymer 45 (2004) 8469–8474
crystals grew along the preferred direction of crystallization, normal to the substrate. The top left image of Fig. 1 shows a top-down scanning electron microscopy (SEM) image of a ZnO micropillar array. This particular micropillar array was prepared by soaking a SnO2-coated glass plate for 18 h in a heated mixture of Zn(NO3)2 and C6H12N4, each at 0.050 M. The pillars assume the shape of a hexagonal prism with a side and prism length of w1.0 and w6.5 mm, respectively. Reducing reaction time and precursor concentration decreased these dimensions. For example, decreasing the reaction time to 6 h reduced the side and prism length to 0.4 and 2.8 mm, respectively. Increasing the concentration of Zn(NO3)2 and C6H12N4 to 0.100 M yielded pillars with side and prism length of 3.0 and 10.2 mm, respectively, after 18 h of immersion. The ZnO pillar arrays used as templates in this study had an average pillar side length of (0.90G 0.23) mm and a height of 6.5 mm. 3.2. Polysulfone membranes Polysulfone (PSf) films were cast about the ZnO pillars using tetrahydrofuran (THF) as solvent. When the thickness of the resulting PSf film was comparable to the pillar height, the ends of the ZnO pillars appeared capped by a layer of PSf (D in Scheme 1). Simply using thinner PSf films eliminated this problem. The end caps could also be removed by soaking
briefly (w20 s) soaking the PSf film in THF/methanol (v/vZ50/50), which slowly dissolved the PSf. The ZnO pillars were removed by soaking the film in 0.5 M HCl for 2 h to yield a microporous membrane as shown in the top right image of Fig. 1. The polymer evidently deformed somewhat after ZnO pillar removal since the cross-section of the channels looks circular rather than hexagonal. The diameter averaged over those measured manually for some 300 channels is (0.95G0.18) mm. 3.3. PS pillar arrays Although the use of ZnO pillars yields cylindrical channels, the channels are neither perfectly vertical nor regularly spaced. The PS pillar formation process is capable of forming ordered arrays of pillars normal to the substrate. As a proof of principle, a crude array of PS pillars was formed to show the adaptability of the membrane process to other templates. The preparation of the PS pillars involved spin-coating a 1 mm-thick PS film on a bottom planar electrode and placing the PS film at distance w5 mm from a top planar electrode. Upon heating the PS film to 120 8C and applying an electric field (w40 V mmK1), the electrostatic pressure generated at the PS melt surface induced polymer
Fig. 1. Top-down SEM view of membranes prepared using ZnO pillars (top right) and PS pillars (bottom right) as the template and a ZnO pillar array (top left). Shown on bottom left is an oblique angle view of a PS pillar array.
X. Yan et al. / Polymer 45 (2004) 8469–8474
flow, producing an array of approximately hexagonallyordered pillars across the capacitor type device. Fig. 1 (bottom left) shows an oblique angle SEM image of a PS pillar array on one electrode after the prying apart of the two electrodes. Defect spots or regions are seen in the pillar array as some pillars got stuck on the other or opposite electrode for the fact that we did not surface-treat the electrodes for specific adhesion. The particular sample of pillars that was used as template had an average diameter of (6.5G2.6) mm at the pillar top and a height of 4.5 mm. 3.4. Nylon membranes After casting a nylon 6/6 film out of formic acid about the template, the PS pillars were selectively dissolved by THF, yielding a nylon 6/6 membrane as shown in the top down SEM image in Fig. 1 (bottom right). The average diameter of the pores is (5.0G1.4) mm. The slight decrease in pore size is likely a result of the cast nylon film being thinner than the height of the pillars. The diameter at the top of the pillar is slightly larger than the average pillar diameter since the PS must conform to its natural contact angle with the upper electrode. In principle, the average pillar diameter can be tuned by changing pillar formation conditions including the PS film thickness and the electric field. More importantly, the size and spacing of the pillars can be patterned down to w100 nm using a relief patterned top electrode [22]. The electric field is strongest where the template pattern is closest to the polymer interface, forcing the polymer to selectively grow first under the protruding features. Thus, this method can, in principle, yield membranes with regularly-packed and uniformly-sized pores with diameter tunable between w100 nm and several microns. 3.5. Water permeability of a PSf membrane A PSf membrane was characterized to demonstrate the high permeability and size selectivity of the membranes formed using this templating process. For the water
8473
permeability test, the PSf membrane previously described was mounted at the bottom end of a L-shaped cylindrical cell [14]. The permeation rate of ultrasonicated water was measured as 2.2!105 l mK2 hK1 barK1 for a 4.3-mm thick membrane after the flow had been stabilized by filling and draining the cell 12 times. This is w4000 times higher than w60 l mK2 hK1 barK1 determined for porous PSf membranes prepared by the precipitation immersion method [25], suggesting the extraordinarily high permeability of the membrane. 3.6. Size selectivity of a PSf membrane To test size-selectivity, we used a PSf membrane with an average diameter of (0.95G0.18) mm to divide two half cells with one containing a dilute solution of 1.1 and 0.90 mm diameter polystyrene latex spheres and the other containing water. Fig. 2 compares SEM images of the latex spheres taken from the feed and the permeate side after seven days of dialysis. The membrane showed excellent size selectivity for the 0.9-mm spheres. Only two 1.1-mm beads permeated the sample, as noted by the arrows in Fig. 2. A quantitative analysis showed that the pre- and afterseparation samples had the average diameters of 0.97G 0.12 and 0.89G0.04 mm, respectively, with the latter essentially the same as that of the pure 0.9-mm spheres.
4. Conclusions We have demonstrated the feasibility to prepare porous membranes using pillars grown on a solid substrate as templates. The membranes thus prepared have high liquid permeability and high size selectivity. The channel size in the membrane can be tuned by changing the pillar preparation conditions. This technique is simple and is readily amenable to membrane preparation from other polymers, monomers, and sol–gel precursors.
Fig. 2. SEM images of 1.1- and 0.9-mm latex spheres before (left) and after (right) sample permeation across a porous PSf membrane.
8474
X. Yan et al. / Polymer 45 (2004) 8469–8474
Acknowledgements NSERC of Canada and NSF of USA are gratefully acknowledged for sponsoring this research.
References [1] Fleischer RL, Price PB, Walker RM. Nuclear tracks in solids. Berkley: University of California Press; 1975. [2] Furneaux RC, Rigby WR, Davidson AP. Nature 1989;337:147. [3] (a) Liu GJ, Ding JF, Hashimoto T, Saijo K, Winnik FM, Nigam S. Chem Mater 1999;11:2233. (b) Liu GJ, Ding JF, Guo A, Herfort M, Bazett-Jones D. Macromolecules 1997;30:1851. [4] See, for example, (a) Lee J-S, Hirao A, Nakahama S. Macromolecules 1989;22:2602. (b) Zalusky AS, Olayo-Valles R, Taylor CJ, Hillmyer MA. J Am Chem Soc 2001;123:1519. (c) Thurn-Albrecht T, Schotter J, Kastle GA, Emley N, Shibauchi T, Krusin-Elbaum L, Guarini K, Black CT, Tuominen MT, Russell TP. Science 2000;290:2126. [5] Chou SY. In: MRS Bull, 2001. p. 512. July issue. [6] Johnson SC, Bailey TC, Dickey MD, Smith BJ, Kim EK, Jamieson AT, Stacey NA, Ekerdt JG, Willson CG, Mancini DP, Dauksher WJ, Nordquist KJ, Resnick DJ. Proc SPIE 2003;5037:197. [7] Grayson RCR, Choi IS, Tyler BM, Wang PP, Brem H, Cima MJ, Langer R. Nat Mater 2003;2:767. [8] Lee SB, Mitchell DT, Trofin L, Nevanen TK, Hans So¨derlund H, Martin CR. Science 2002;296:2198. [9] See, for example, Cabodi M, Turner SWP, Craighead HG. Anal Chem 2002;74:5169.
[10] See, for example, Park M, Harrison C, Chalkin PM, Register RA, Adamson DH. Science 1997;276:1401. [11] Lin VS-Y, Motesharei K, Dancil KPS, Sailor MJ, Ghadiri MR. Science 1997;278:840. [12] (a) Martin CR. Science 1994;266:1961. (b) Al-Mawlawi D, Liu CZ, Moskovits M. J Mater Res 1994;9:1014. (c) Kovtyukhova NI, Mallouk TE. Chem Eur J 2002;8:4355. [13] Kshama BJ, Hulteen JC, Martin CR. Science 1997;278:655. [14] See, for example, Gates B, Yin YD, Xia YN. Chem Mater 1999;11: 2827 for a long list of examples. [15] Kresge CT, Leonowicz ME, Roth WJ, Vartuli JC, Beck JS. Nature 1992;359:710. [16] (a) Attard GS, Glyde JC, Goltner CG. Nature 1995;378:366. (b) Bagshaw SA, Prouzet E, Pinnavaia TJ. Science 1995;269:1242. [17] See, for example, Melosh NA, Davidson P, Feng P, Pine DJ, Chmelka BF. J Am Chem Soc 2001;123:1240. [18] Yan F, Goedel WA. Chem Mater 2004;16:1622. [19] (a) Widawski G, Rawiso M, Franc¸o¨is B. Nature 1994;369:397. (b) Jenekhe SA, Chen XL. Science 1999;283:372. [20] See, for example, (a) Kuo CW, Shiu JY, Chen P, Somorjai GA. J Phys Chem B 2003;107:9950. (b) Geim AK, Dubonos SV, Grigorieva IV, Novoselov KS, Zhukov AA, Shapoval SY. Nat Mater 2003;2:461. [21] Chou SY, Zhuang L. J Vac Sci Technol, B 1999;17:3197. [22] Schaffer E, Thurn-Albrecht T, Russell TP, Steiner U. Nature 2000; 403:874. [23] Vayssieres L, Keis K, Lindquist S-E, Hagfeldt A. J Phys Chem B 2001;105:3350. [24] Fujita K, Matsuda K, Mitsuzawa S. Bull Chem Soc Jpn 1992;65:2270. [25] See, for example, Lu ZH, Liu GJ, Duncan S. Macromolecules 2004; 37:174–80.
