Langmuir 2008, 24, 617-620
617
Using Breath Figure Patterns on Structured Substrates for the Preparation of Hierarchically Structured Microsieves Claudia Greiser, Susann Ebert, and Werner A. Goedel* Physical Chemistry, Chemnitz UniVersity of Technology, Strasse der Nationen 62, 09111 Chemnitz, Germany ReceiVed September 22, 2007. In Final Form: NoVember 16, 2007 Microsieves are advanced filtration membranes characterized by a uniform pore size, a high pore density, and a thickness smaller than the pore diameter. The uniform pore size provides a high selectivity; the small thickness gives rise to a high flux and allows efficient removal of any filter cake by backflushing. However, microsieves are sensitive to mechanical stress. Thus, they need either an external macroporous support or a hierarchical structure that provides an integrated supportive structure. We prepare microsieves with a hierarchical pore structure by creating breath figure patterns within layers of solutions of polymers in a volatile solvent that are spread out on top of structured supports. For the formation of breath figure patterns, the volatile solvent is evaporated in a moist atmosphere. This cools the surface to such an extent that dew droplets form on the thin film, partially penetrate into the layer, and create a concave imprint in the final solid polymer layer. This procedure is usually done on flat surfaces; in our case the spreading of the polymer solution is done on a support decorated with protrusions. In this procedure, the dew droplets touch the protrusions of the structured support before the polymer solution vitrifies. At the same time, the trenches of the structured substrate are filled with polymer much deeper than the penetration depth of the dew droplets. After the separation of the vitrified layer from the substrate, we obtain thin polymer membranes with a hierarchical structure consisting of an ultrathin active separation layer with submicrometer pores and a supporting layer with larger pores.
Membranes are naturally occurring or artificially produced two-dimensional barriers that separate two fluid phases from each other, show selective permeability, and thus may be utilized to enrich or deplete components in and out of one of the fluid phases, respectively. Porous membranes are predominantly utilized as filtration media to remove particles, macromolecules, or microorganisms from liquids or gases. In conventional porous membranes, the pores are usually much smaller than the membrane thickness, are nonuniform in diameter, and form tortuous paths instead of straight channels. The actual filtration process is often performed by a filter cake that is formed on top of the membrane. These structural details decrease the performance of the membranes and increase their flow resistance. They can be circumvented, however, by using recently developed microsieve technology.1-3 Microsieves are filtration membranes showing a thickness smaller than the pore diameter and thus providing optimum performance at minimum flow resistance; the formation of a filter cake can be efficiently prevented by applying cross-flow filtration and frequent backflushing.4 Until recently,5 microsieves have been prepared exclusively using microlithography. The virtue of microsieves, the small thickness, renders them simultaneously fragile. Thus, they have to be stabilized by a hierarchically structured support that has larger openings, a feature that can easily be integrated into the lithographic process. However, microlithography is a comparatively expensive and elaborate technique and inherently has limitations with respect to the lateral dimensions of the prepared membranes. On the other hand, porous materials can be made in much simpler ways by embedding small objects such as block * Corresponding author. E-mail:
[email protected]. Tel: +49 371 531-31713. Fax: +49 371 531-21249. (1) van Rijn, C. J. M.; van der Wekken, M.; Nijdam, W.; Elwenspoek, M. J. Microelectromech. Syst. 1997, 6, 48-54. (2) van Rijn, C. J. M.; Veldhuis, G. J.; Kuiper, V. Nanotechnology 1998, 9, 343-345. (3) Desai, T. A.; Hansford, D.; Ferrari, M. J. Membr. Sci. 1999, 159, 221. (4) Kuiper, S.; van Rijn, C. J. M.; Nijdam, W.; Krijnen, G. J. M.; Elwenspoek, M. C. J. Membr. Sci. 2000, 180, 15-28. (5) Vogelaar, L.; Barsema, J. N.; van Rijn, C. J. M.; Nijdam, W.; Wessling, M. AdV. Mater. 2003, 15, 1385.
copolymers,6,7 surfactant arrays,8-10 emulsion droplets, and colloidal crystals11-16 into an initially liquid matrix, followed by solidification of the matrix and removal of the objects. Of special interest is the preparation of microsieve-like structures by embedding just a monolayer of templating objects. Most elegant among these techniques is the formation of so-called breath figure patterns.17-19 A dilute solution of a polymer is spread out on a planar surface and the solvent is vaporized in the presence of a moist atmosphere. Upon evaporation of the solvent, the surface of the polymer solution cools down, and dew droplets form, merge, grow, sink into the polymer solution, create an imprint in the solidifying polymer, and finally evaporate at the end of the process when the surface warms up to the original temperature. Usually, this technique gives rise to a layer bearing pores that have openings at the top but are closed at the bottom;20 there are a few reports on utilizing breath figure patterns to prepare microsieves bearing pores that penetrate from the top to the bottom of the polymer layer.21 However, until now the preparation of sieve-like structures via breath figure patterns has yielded (6) Thurn-Albrecht, T.; Schotter, J.; Ka¨stle, G. A.; Emley, N.; Shibauchi, T.; Krusin-Elbaum, L.; Guarini, K.; Black, C. T.; Tuominen, M. T.; Russell, T. P. Science 2000, 290, 2126-2129. (7) Chan, V. Z.-H.; Hoffman, J.; Lee, V. Y.; Iatrou, H.; Avgeropoulos, A.; Hadjichristidis, N.; Miller, R. D.; Tomas, E. L. Science 1999, 286, 1716-1719. (8) Tavolaro, A.; Drioli, E. AdV. Mater. 1999, 11, 975-993. (9) Kresge, C. T.; Leonowicz, M. E.; Roth, W. J.; Vartuli, J. C.; Beck, J. S. Nature 1992, 359, 710-712. (10) Raman, N. K.; Anderson, M. T.; Brinker, C. J. Chem. Mater. 1996, 8, 1682-1701. (11) Velev, O. D.; Lenhoff, A. M. Curr. Opin. Colloid Interface Sci. 2000, 5, 56-63. (12) Velev, O. D.; Jede, T. A.; Lobo, R. F.; Lenhoff, A. M. Nature 1997, 389, 447. (13) Holland, B. T.; Blanford, C. F.; Stein, A. Science 1998, 281, 538-540. (14) Wijnhoven, J. E. G. J.; Vos, W. L. Science 1998, 281, 802-804. (15) Jiang, P.; Hwang, K. S.; Mittleman, D. M.; Bertone, J. F.; Colvin, V. L. J. Am. Chem. Soc. 1999, 121, 11630-11637. (16) Johnson, S. A.; Ollivier, P. J.; Mallouk, T. E. Science 1999, 283, 963965. (17) Widawski, G.; Rawiso, M.; Francois, B. Nature 1994, 369, 387. (18) Pitois, O.; Francois, B. Eur. Phys. J. B 1999, 8, 225-231. (19) For a recent review see: Bunz, U. H .F. AdV. Mater. 2006, 18, 973-989. (20) Connal, L. A.; Gurr, P. A.; Qiao, G. G.; Solomon D. H. J. Mater. Chem. 2005, 15, 1286-1292. (21) Nishikawa, T.; Ookura, R.; Nishida, J.; Arai, K.; Hayashi, J.; Kurono, N.; Sawadaishi, T.; Hara, M.; Shimomura, M. Langmuir 2002, 18, 5734.
10.1021/la7029449 CCC: $40.75 © 2008 American Chemical Society Published on Web 12/29/2007
618 Langmuir, Vol. 24, No. 3, 2008
Figure 1. Scheme of the preparation of a hierarchical microsieve: (a) A polymer solution is applied to a three-dimensionally structured surface (here, spheres embedded in a water layer); (b-d) the solvent is evaporated in a moist atmosphere, the cooling of the polymer solution due to evaporation leads to the condensation of dew droplets, and the polymer solution finally solidifies; (e) after solidification of the polymer, the water is removed/evaporated; and (f) removal of the structured substrate gives rise to the desired porous membrane.
only laterally smooth membranes that do not have any hierarchical structure that may act as a support. Therefore, the hierarchical structure needs to be added subsequent to the preparation (e.g., by transferring the sieve-like structures to an external support with larger openings). However, this extra step poses the danger of damaging the structure and demands additionally the optimization of adhesion and the matching of thermal expansion coefficients of the support and the sieve. We are aware of one report on the preparation of breath figure patterns on a laterally structured substrate (electron microscopy grid) not giving rise to microsieve-like structures with pores penetrating the polymer layer from the top to the bottom.22 Here we show that one can utilize breath figure patterns to prepare microsieves that are made in one stroke out of a single material and have the desired hierarchical structure. The special feature in our process is to spread the polymer solution not onto a flat surface but onto a surface that has a three-dimensionally structure (Figure 1). The trenches in this structure are filled with the polymer solution and finally give rise to thicker, nonporous regions that may act as a mechanical support; the elevations are covered with a comparatively thin layer of the polymer solution acting as substrates for breath figure formation and finally give rise to sieve-like structures. In principle, this process might work on any three-dimensionallystructured surface. It will be advantageous if the elevations make up most of the area of the structured surface, and the trenches are comparatively deep, and assume only a small fraction of the total area. To facilitate spreading and ease the separation between the surface and the final polymer layer undercuts, vertical walls and sharp edges should be avoided. It is also desirable that the surface be prepared without sophisticated techniques such as photolithography. Our most favored surface that fulfills all the above-mentioned criteria is a monolayer of micrometer-sized glass beads that are partially embedded in a thin layer of water (as depicted in Figure 1). Onto this layer we spread a solution of a non-water-soluble glassy polymer (poly(methyl methacrylate)) in a non-water-soluble volatile solvent (chloroform), evaporate the solvent at 23 °C in an atmosphere of 98% humidity, lift the resulting structure off of the water layer, dry it, and obtain the intermediate stage depicted in Figure 2: the glass beads are glued together; the bottom halves of the glass beads are not covered by the polymer (Figure 2d), but the top parts are embedded and covered by the polymer (Figure 2a), and the polymer surface shows a microscopic structure that reflects the structure of the underlying layer of glass beads. A closer inspection of the top of the structure reveals that the (22) Connal, L. A.; Qiao, G. G. AdV. Mater. 2006, 18, 3024-3028.
Letters
polymer layer also bears pores on a submicrometer length scale, indicating the formation of breath figure patterns (Figure 2b,c). At this stage, it is not clear whether these pores actually reach the bottom of the thin polymer layer that covers the glass spheres, or whether they are just dimples that are open only at the top. The glass beads can easily be removed by dipping the structure into a mixture of dilute hydrofluoric acid and sulfuric acid. This procedure gives rise to the final hierarchically structured membrane shown in Figure 3. The bottom view reveals thicker, beam-like structures due to the filled trenches in the threedimensionally structured surface. Top and bottom views reveal submicroscopic pores in the thin polymeric layers that were formed on top of the elevations and thus confirm the desired microsieve-like structure of these parts of the membrane. The membrane was broken into pieces, and the exposed cross section is shown at the bottom of Figure 3. The beam-like structures have a feature size of 25 to 30 µm (Figure 3g,h), and the microsieve-like parts have a thickness in the range of 80 to 200 nm (Figure 3i, the thickness varies with the position, and the given range was estimated from several images). Image analysis of a series of images such as those shown in Figure 3c,f) reveals for the top view as well as for the bottom view a monomodal pore width distribution. The mean pore diameter obtained from the top view images is 0.264 µm, and the standard deviation of the pore width is 0.146 µm. The corresponding values obtained from the bottom view images are 0.09 ( 0.07 µm. (The analysis was based on 6 images comprising 250 analyzed pores.) The smaller pore size obtained from the analysis of the bottom view images reflects the slightly conical shape of the pores that is also visible in images of the top view and the cross section (Figure 3c,i). The images shown in Figure 3 reveal that the above-described strategy for the preparation of hierarchical structures using breath figure patterns was successful: we obtain large hemispherical pores due to the structured substrate and submicroscopic conical pores due to breath figure patterns that to a large extent penetrate the thin parts of the membrane. In most papers that investigate the details of pore penetration, it is stated that the bottom of the pores is covered by polymer. In accordance with these observations, published mechanisms of pattern formation often comprise an intermediate polymeric “skin” that completely covers the interface between the water droplets and the polymer solution.18 However, systematic variations of the preparation conditions revealed that one could induce a transition from nonpenetrating pores to penetrating pores by reducing the concentration of the polymer solution.21 In our case, the conical shape of the submicroscopic pores indicates that before solidification the contact angle between the interface of the water droplet/polymer solution and the glass surface (measured in the water phase) is higher than 90°. We are unaware of the composition of the polymer solution at the moment of solidification, and thus we do not know the exact contact angles. We may assume that the contact angle at the moment of solidification is in the range of contact angles of a water droplet covered by either chloroform (advancing, 108.4 ( 10.0°; receding, 98.8 ( 5.0°) or ethylacetate, which is a liquid similar to the polymer, but with low molecular mass (advancing, 140.6 ( 5.1°; receding, 140.8 ( 10.0°). The contact angles of these model liquids are higher than 90°, in agreement with the observation of the conical shape of the final pores, and in agreement with the assumption that the water droplets touch the solid substrate and establish a finite contact angle before any polymer skin may form. We must admit that the submicroscopic pores are nonuniform in diameter. The formation of submicro-
Letters
Langmuir, Vol. 24, No. 3, 2008 619
Figure 2. Porous polymer layer still on top of the solid substrate: (a-c) porous polymer layer on top of the glass beads (stage e in Figure 1) imaged from the top at various magnifications and (d-f) the same layer imaged from the bottom, revealing the embedded glass beads.
Figure 3. Hierarchically structured microsieve-like membrane, separated from the substrate: (a-c) images obtained from the top at various magnifications revealing the curved microsieve-like parts that originally covered the protruding parts of the substrate; (d-f) images obtained from the bottom revealing the beam-like supportive structures that were created by filling the trenches of the substrate; (g-i) images of a cross section that reveal the dimensions of the beam-like structures (25-30 µm) as well as the thickness of the microsieve-like parts (80-200 nm).
scopic breath figure patterns is often claimed in the literature. Unfortunately, the regular structures actually shown reveal in most cases length scales well above a micrometer.
To make the pores penetrate through the bottom of the layer, we chose a comparatively dilute polymer solution and spread an amount that was barely enough to cover the glass surface
620 Langmuir, Vol. 24, No. 3, 2008
completely. We believe that this gives rise to only a short time span between cooling below the dew point and solidification and that the early contact of the droplets with the solid surface reduces the lateral mobility required for regular order. 23,24 Preliminary permeation and filtration tests were conducted by guiding 2-propanol, water, and dispersions of yeast cells through the membranes. If subjected to a permeating flow of water and 2-propanol at a pressure of approximately 103 Pa, then the membranes have a permeability of 5.60 ( 0.75 L h-1 m-2 Pa-1 for water and 0.99 ( 0.15 L h-1 m-2 Pa-1 for 2-propanol. Dispersions of yeast cells that are passed through the membranes show a significant reduction in the number of cells per volume (reduction from 40 × 105 cells/µL ( 1.3 × 105 cells/µL down to 14 × 105 cells/µL ( 3.9 × 105 cells/µL 25), but not the complete removal of the cells. The latter fact is in agreement with the polydispersity of the pore diameters and the pores visible in the electron microscopy images that exceed the typical size of yeast cells of 5-10 µm. In summary, we conclude that the formation of breath figures on top of polymer solutions that are spread onto threedimensionally structured surfaces is a versatile tool for the preparation of hierarchically structured microsieve-like porous membranes in one stroke, and made out of one material. In the example shown, we chose a substrate as simple as possible (the top parts of a layer of spherical glass beads protruding out of a planar surface) that can be easily prepared. As a result, we obtain curved microsieve-like parts and supportive beam-like structures with an aspect ratio given by the geometry of the beads. However, the method may be extended to other surfaces (e.g., plates structured via embossing, engraving, or etching). Such substrates may be more elaborate to prepare but may offer independent control over the size, shape, and curvature of the microsieve-like parts, and the width and depth of the beam-like supportive parts of the hierarchical structure. In principle, this method offers the possibility for scale up to areas considerably larger than the limits of microlithography. The technique thus might be suitable for the preparation of large-area sturdy microsieves. Experimental Section Glass beads of 85 µm diameter and a standard deviation of the diameter of 0.5 µm (Supelco glass beads, acid washed, 75 µm) were hydrophobized by immersion in a mixture of 3:1 v/v sulfuric acid (96 wt %)/hydrogen peroxide (35%) at 80 °C for 10 min, washed with copious amounts of purified water, and dried in air at 120 °C overnight, subsequently immersed overnight into a solution of 10 mmol/L octadecyl triethoxysilane (Alfa Aesar, 94%) in toluene (technical grade, dried prior to use with 3 Å molecular sieves), washed three times with copious amounts of pure toluene and twice (23) Steyer, A.; Guenoun, P.; Beysens, D.; Knobler, C. M. Phys. ReV. B 1990, 42, 1086. (24) Fritter, D.; Knobler, C. M.; Beysens, D. Phys. ReV. A 1991, 43, 28582869. Knobler, C. M.; Beysens, D. Europhys. Lett. 1988, 6, 707. (25) Mean value of three independently prepared membranes, showing individual reductions of 16 × 105 cells/µL ( 0.3 × 105 cells/µL, 9.8 × 105 cells/µL ( 1.3 × 105 cells/µL, and 17 × 105 cells/µL ( 0.9 × 105 cells/µL; errors given here are standard deviations of two independent permeation runs per membrane, with each run analyzed twice with the cell counter.
Letters with acetone, and then dried at room temperature in air. A glass Petri dish of 50 mm diameter and 10 mm height was filled to a height of 6 mm with water. The above-described glass beads (0.1227 g) were applied dry onto the water surface. (The glass beads distributed themselves evenly across the water surface as an almost densely packed layer covering an almost circular area of approximately 40 mm diameter.) The Petri dish was transferred to a glove box with an atmosphere of 98% relative humidity at 23 °C. Onto the layer of the glass beads we applied by the aid of a disposable polypropylene syringe 0.313 g of a solution of 2 wt % poly(methyl methacrylate) in chloroform and evaporated the chloroform in a gentle stream of argon at the same humidity, with the whole evaporation process being completed within 15 min. The resulting membranes were lifted off of the water surface, cut in pieces and used to obtain the electron microscopy images shown in Figure 2. To remove the glass beads, parts of the membrane were immersed in a solution of 0.75 g (approximately 18 drops) of sulfuric acid (48 wt %) and 0.75 g of hydrofluoric acid (40 wt %) (Caution! Hydrofluoric acid is toxic and corrosiVe and should be handled only by trained personnel aware of the dangers associated with the use of this chemical !) in 15 mL of water and were finally used to acquire the electron microscopy images shown in Figure 3. Electron microscopy was performed using a NanoNovaSEM (FEI); samples were coated with an 18 nm thick platinum layer prior to imaging. Contact angles were determined by imaging either water droplets covered by ethylacetate or chloroform droplets covered by water resting on the surface of a glass slide coated with octadecyltriethoxysilane as described above using a contact angle measuring system (G2, Kru¨ss, Stephanskirchen/Rosenheim). Each value represents a mean value of five independent measurements and the corresponding standard deviation. A homebuilt filtration device was made according to ref 26 consisting of two glass tubes with a 90° bend and a planar glass plate with a central circular opening attached to one end of each tube. Membranes prepared using the above-described procedures were sandwiched between two rings of paraffin wax and mounted between the two glass plates in such a way that the final device had an overall zigzag shape consisting of a vertical section of the first tube (running downward), followed by a horizontal section of the same tube, with the membrane mounted vertically between the glass plates, and then the horizontal and vertical sections of the second glass tube. The area of the membrane that was exposed to the liquid to be filtered was a circle of 1.5 mm diameter. After mounting the device, 2-propanol (technical grade), water, or dispersions of yeast cells were placed into the first vertical glass tube and passed through the membrane by the aid of the natural hydrostatic pressure exerted by the 14 cm height of the liquid within the tube (approximately 1100 Pa in the case of 2-propanol and 1400 Pa in the case of water). The number concentrations of yeast cells before and after filtration were determined via optical microscopy using a cell counter chamber (Neubauer improved, Marienfeld, Lauda-Ko¨nigshofen, Germany).
Acknowledgment. We thank C. Werner, Max Bergmann Center of Biomaterials/Institute of Polymer Research, Dresden, Germany, for lending us the cell counter chamber. LA7029449 (26) Liu, G.; Ding, J.; Hashimoto, T.; Kimishima, K.; Winnik, F. M.; Nigam, S. Chem. Mater. 1999, 11, 2233-2240.
