The Sol-Gel Preparation and Characterization of Nanoporous Silica Membrane with Controlled Pore Size T. Fujii, T. Izumi, Dept. of Food Sci., Niigata Univ. of Pharm. & Appl. Life Sci., Niitsu, Niigata 956-8603, Japan Phone&Fax: 0250-25-5145 e-mail: atomic@mbk.nifty.com and T. Yano, K. Nakamura, O. Miyawaki Dept. of Appl. Biol. Chem., Univ. of Tokyo, Bunkyo-ku, Tokyo 113-8657, Japan Inorganic membranes have been increasingly used in food and biochemical industry where sterilization is required, owing to their excellent chemical and thermal stability. The separation mechanism of by a porous membrane is closely related to the membrane pore size relative to the permeate molecule size. It is thus important to control the pore size uniformly to nanometer scale without pinholes or cracks. A wet procedure of self-assembly of films from colloidal particles, which are held together by non-specific bonding forces at a solid/solution interface, has shown its worth as a cheap and technologically very simple method for the construction of ultrathin nanostructured filmes. In the sol-gel method, the gelation usually happens when aggregation or branching of the solid matter takes place in a liquid (caused by van der Waals forces). An aggregate having the higher fractal dimension has the less uniform void space than the aggregate having the lower fractal dimension. We prevent the formation of higher fractal aggregates in order to prepare the membrane with uniform pore size by the sol-gel method. Preparation of the nanoporous silica membrane A porous a-alumina membrane tube (10 mm in external diameter, 8 mm in internal diameter, 1.6 mm in mean pore size) was purchased from Mitsui Kensaku Toishi (Utsunomiya, Japan). Two colloidal silicas, Snowtex-UP, a fibrous silica colloid (5-20 nm thick and 40-300 nm long), and Snowtex-XS, spherical silica nanoparticles (4-6 nm in diameter), were presented by Nissan Chemical Industries (Tokyo, Japan). First, the alumina membrane was coated with the fibrous silica colloid to fill macropores on the surface. The silica concentration of coating solution for the first step modification by fibrous silica colloid was 20 wt.%. The operation of brush-coating was repeated several times. The membrane sintered at 570 C for 5 min with heating rate of 5 C/min. This procedure was repeated several times. Next, the membrane, secondly, was coated with the silica nanoparticles to prepare a thin microporous surface layer in the same procedure as the first step modification. The silica concentration of coating solution for the second step modification by silica nanoparticles was 4.0 wt.%. Characterization of the porous membranes Table 1 shows that the fractal dimension for membrane surfaces, BET surface area and the peak diameter of the pore size distribution of the porous membrane. The value of fractal dimension for the surface of the nanoporous silica membrane was close to the flat surface dimension d=2. These results indicate that we could prepare the silica membrane to prevent the formation of the aggregates with the higher fractal dimension in the coating solution. BET surface area values were approximately 3-4 m 2 /g in this work. As coating progressed, the peak diameter became smaller. The peak at 3.9 nm for the membrane after the second step modification by silica nanoparticles is close to the size of the nanoparticles of diameter 4-6 nm showing that this corresponds to the vacant space among nanoparticles. Table 1 Physical properties of porous membranes Membranes D f S BET [m 2 /g] d peak [nm] a-alumina porous membrane 2.010 0.739 - Membrane after the first step modification 2.152 3.87 11.5 Membrane after the second step modification 2.014 2.97 3.9
The Sol-Gel Preparation and Characterization of Nanoporous Silica Membrane with Controlled Pore Size T. Fujii, T. Izumi, Dept. of Food Sci., Niigata Univ. of Pharm. & Appl. Life Sci., Niitsu, Niigata 956-8603, Japan Phone&Fax: 0250-25-5145 e-mail: atomic@mbk.nifty.com and T. Yano, K. Nakamura, O. Miyawaki Dept. of Appl. Biol. Chem., Univ. of Tokyo, Bunkyo-ku, Tokyo 113-8657, Japan Introduction Inorganic membranes have been increasingly used in food and biochemical industry where sterilization is required, owing to their excellent chemical and thermal stability. 1 The separation mechanism of by a porous membrane is closely related to the membrane pore size relative to the permeate molecule size. 2,3 It is thus important to control the pore size uniformly to nanometer scale without pinholes or cracks. A wet procedure of self-assembly of films from colloidal particles, which are held together by non-specific bonding forces at a solid/solution interface, has shown its worth as a cheap and technologically very simple method for the construction of ultrathin nanostructured films. 4,5 In the sol-gel method, the gelation usually happens when aggregation or branching of the solid matter takes place in a liquid (caused by van der Waals forces). An aggregate having the higher fractal dimension has the less uniform void space than the aggregate having the lower fractal dimension. We prevent the formation of higher fractal aggregates in order to prepare the membrane with uniform pore size by the sol-gel method. Experimental Preparation of the nanoporous silica membrane (Fig. 1) A porous -alumina membrane tube (10 mm in external diameter, 8 mm in internal diameter, 1.6 µm in mean pore size) was purchased from Mitsui Kensaku Toishi (Utsunomiya, Japan). Two colloidal silicas, Snowtex-UP, a fibrous silica colloid (5-20 nm thick and 40-300 nm long), and Snowtex-XS, spherical silica nanoparticles (4-6 nm in diameter), were presented by Nissan Chemical Industries (Tokyo, Japan). First, the alumina membrane was coated with the fibrous silica colloid to fill macropores on the surface. The silica concentration of coating solution for the first step modification by fibrous silica colloid was 20 wt.%. The operation of brush-coating was repeated several times. The membrane sintered at 570 C for 5 min with heating rate of 5 C/min. This procedure was repeated several times. Next, the membrane, secondly, was coated with the silica nanoparticles to prepare a thin microporous surface layer in the same procedure as the first step modification. The silica concentration of coating solution for the second step modification by silica nanoparticles was 4.0 wt.%. The pore size distribution support nanospace : interparticle void spaces Fig. 1 Design of nanoporous silica membrane.
of the prepared asymmetric inorganic membrane was evaluated from the N 2 desorption isotherm measured using a Shimadzu accelerated surface area and porosimeter ASAP 2000. Results and discussion Observation of the surface microstructure of the membrane after the first step modification with AFM AFM observations can be available for monitoring the changes in the surface microstructure of the membranes. Figure 2 shows that AFM topography for the surface of the porous -alumina support membrane (A) and that of the membrane after the first step modification by the fibrous silica colloid (B). It is obvious that the coarseness of the surface decreased. This result indicates the effectiveness of the first step modification by the fibrous silica colloid. The roughness of the surfaces is quantitatively described by the root-mean-squared roughness (rms), R rms, which is the standard deviation in th e height of the surface measured by the AFM image as follows; R rms = N ( z n Š z) 2 n=1 N Š 1 z substrate were 242 and 112 nm, respectively. Cluster radius distribution in the coating solution for the second step modification The coating solution for the second step Fig. 2 AFM topography for -alumina support (A) and for the membrane after the first step modification by fibrous silica colloid (B). higher than 8.32 wt. %. The average cluster sizes of each peak are peak increased divergently with an increase in Fig. 3 Cluster radius distribution in sol of silica nanoparticles.
the silica concentration to the sol-gel transition point (Cg = 31.8 wt. %). The sol-gel transition concentration was estimated from the density of silica particles, assuming that the critical volume fraction was 0.16. The aggregate formation takes place as a result of connecting particles to clusters of various sizes. The clusters combine to form aggregates, and voids between them are filled with other clusters or single particles, or disappear due to reorganization of aggregates, accompanied by capturing the small clusters and particles. The correlation length of silica sol was described by the power law. The critical exponent was determined to be 0.78, similar to the numerical value. 7 Pore size distribution of the thin porous silica membranes after the first step and second step modification The plots show two peaks in Fig. 5. As coating progressed, the smaller peak became smaller, while the larger one didn t shift. The peak at 3.3 nm for the completely coated membrane was thought to be related to the porosity of a skin layer in the asymmetric membrane, because it was close to the size of the vacant space in a bed packed with spheres with diameters of 4-6 nm. The results indicated that the pore size of this thin porous silica membrane could be controlled by the repetition of coating. The pore size of 3.3 nm was a few times JJ BBB J J B J J J B B B Fig. 4 Effect of the concentration of silica nanoparticles On average cluster radius. Fig. 5 Pore size distribution of the tubular ceramic membranes as large as the molecular sizes of the solutes selected in this study. We applied a box-counting method to evaluate the fractal dimension for the determined surface profiles. A part of plane including a profile as shown in AFM topography is covered with the boxes with side length d. If the profile is completely covered with N boxes, the fractal geometry shows the following relationship: B J N(d) d Š D f where D f is the fractal dimension. By changing size of the boxes, the number of the boxes covering the profile is counted. The fractal dimension for the surface is obtained from the data plotted on the log(n)-log(d) plane for each profile. The value of fractal dimension for the surface of the nanoporous silica membrane was close to the flat surface dimension d=2 (See Table 1). These results indicate that we could prepare the silica membrane to prevent the formation of the aggregates with the higher fractal dimension in the coating solution.
Characterization of the porous membranes Table 1 shows that the fractal dimension for membrane surfaces, BET surface area and the peak diameter of the pore size distribution of the porous membrane. The value of fractal dimension for the surface of the nanoporous silica membrane was close to the flat surface dimension d=2. These results indicate that we could prepare the silica membrane to prevent the formation of the aggregates with the higher fractal dimension in the coating solution. BET surface area values were approximately 3-4 m 2 /g in this work. As coating progressed, the peak diameter became smaller. The peak at 3.9 nm for the membrane after the second step modification by silica nanoparticles is close to the size of the nanoparticles of diameter 4-6 nm showing that this corresponds to the vacant space among nanoparticles. Table 1 Physical properties of porous membranes Membranes D f S BET [m 2 /g] d peak [nm] -Alumina porous membrane 2.010 0.739 - Membrane after the first step modification 2.152 3.87 11.5 Membrane after the second step modification 2.014 2.97 3.9 Conclusion A thin nanoporous silica membrane was fabricated by the two-step sol-gel method. The silica concentration of the second-step coating solution was critical to prepare high quality membranes; the coating solution with high concentration of colloidal silica was not able to adhere onto the substrate after sintering because of the formation of the aggregates with the higher fractal dimension Therefore, the silica concentration in the coating solution should be low enough to avoid the formation of the aggregates. The prepared membrane was proved to have a dense skin layer with a nanospace as interparticle void space from the AFM analysis of the surface image and the nitrogen adsorption porosimetry. References: (1) A.F.M. Leenarrs, K. Keizer and A.J. Burggraaf, J. Mater. Sci., 19, 1077-1088 (1984). (2) R.J.R. Uhlhorm, K. Keizer and A.J. Burrggaaf, J. Membr. Sci., 46, 225-241 (1989). (3) O. Li and S.T. wang, J. Membr. Sci., 59, 331-352 (1991). (4) J.. Fendler and F.C. Meldrum, Adv. Mater., 7, 607-632 (1995). (5) N.A. Kotov, I. Dekany and J.. Fendler, J. Phys. Chem., 99, 13065-13069 (1995). (6) B. Chu, Dynamic Light Scattering, Academic Press, London, 1975. (7) D. Stauffer, A. Coniglio and M. Adam, Adv. Polym. Sci., 44, 103-158 (1982).