An SiO₂-PVA nanocomposite, which is the precursor of a porous silica monolith, was prepared by using fumed silica nanoparticles (Aerosil 300, BET surface area of 300 m²・g⁻¹ according to the supplier, Nippon Aerosil, Japan) and PVA (polymerization degree of 1500 - 1800, Wako pure chemical Industries, Japan) as starting materials by a following procedure [21] . Fumed silica was added distilled water adjusted at pH 3 followed by dispersing by means of an ultra-sonication for 1 h to obtain 8 wt% of a silica suspension. Meanwhile, PVA was dissolved into distilled water and stirred using a magnetic stirrer at a speed of 800 rpm to prepare 8 wt% of a PVA solution. Then, the obtained silica suspension and PVA solution were mixed for 12 h to obtain 80SiO₂-20PVA (wt%) suspension. The resulting SiO₂-PVA suspension was cast into a Teflon container and dried at 30˚C under atmospheric pressure in an oven. A gelation of the suspension was taken place gradually and then a monolithic SiO₂-PVA nanocomposite was formed.

A monolithic porous silica monolith was fabricated by calcining the SiO₂-PVA nanocomposite at 600˚C for 3 h in air to combustion of the PVA, followed by sintered at 600˚C - 1100˚C for 5, 10, 20, 30 min, or 1 h, respectively, in air.

BET surface area and pore size distribution of the porous silica monolith were determined by means of a porosimeter (BELSORP-mini, Bel Japan Inc., Japan) via Brunauer-Emmett-Teller (BET) method and Barrett-Joyner-Halenda (BJH) method. Structural change in the porous silica monolith by the sintering was examined by means of a Raman spectrometer (STR-HTUV, Seki Technotron Co.) equipped with a He-Cd laser. Optical transmittances of the porous silica monolith were measured by means of a spectrophotometer (U-3500, Hitachi High-Technologies Co., Japan).

Figure 1 shows pore size distribution of the SiO₂-PVA nanocomposite. It is clearly seen that the SiO₂-PVA nanocomposite had mesopores with radius below 20 nm and BET surface area of 141 m²・g⁻¹. This SiO₂-PVA nanocomposite was used as a precursor for a porous silica monolith. The SiO₂-PVA nanocomposite precursor was sintered at given temperature in the range of 600˚C - 1100˚C to fabricate a porous silica monolith. Figure 2 denotes dependence of BET surface area on sintering temperature for the sintered samples. BET surface area for the sample sintered at 600˚C was 291 m²・g⁻¹. This value is larger than that of the SiO₂-PVA nanocomposite precursor and comparable to that of the as-received fumed silica particles. This result indicated that the PVA in the SiO₂-PVA nanocomposite was just combusted at the temperature and sintering of the silica nanoparticles did not occur. As increasing sintering temperature up to 900˚C, BET surface area of the sample decreased with increasing treatment temperature. Over 900˚C, BET surface area of the sample abruptly decreased with increasing sintering temperature. This fact suggested that the sintering of silica

nanoparticles would occur above 900˚C. At 1100˚C, BET surface area reached to zero. This result indicates that sintering of the silica nanoparticles was accomplished at 1100˚C and dense (non-porous) body was formed. Figure 3 shows dependence of the sintering temperature on pore size distributions of the sintered silica samples. It is clearly seen that the sintered silica samples had broad peaks with the pore radius below 25 nm. For the sample sintered at 600˚C, the peak center of pore size distribution is located at ~15 nm. As increasing sintering temperature, the peak was shifted to smaller side and its peak intensity became small. For the sample obtained at 1100˚C, no pores exist in the examined range.

Sintering behavior of the silica nanoparticles of the SiO₂-PVA nanocomposite was elucidated in terms of change in amorphous structure by using the Raman spectroscopy. Figure 4 depicts Raman spectra of the sintered silica samples. Raman spectrum for a commercial silica glass was also given in the figure to

compare the spectra for the sintered samples. For the spectrum of the commercial silica glass, some characteristic peaks attributed to an amorphous silica were observed: The bands at around 500 and 620 cm⁻¹ are assigned respectively to symmetric oxygen breathing vibrations of regular four-membered (D1) and three-membered (D2) silica rings [23] [24] [25] . The broad band at around 430 cm⁻¹ is attributed to bending motions of oxygen atoms in the random network of silica glass [26] . In the present study, we focused the bands for D1 and D2 to elucidate the sintering behavior of silica nanoparticles because D1 and D2 bands are sensitive to change in the amorphous structure. Figure 5 depicts relative intensities for D1 and D2 bonds of each sample. For the spectrum of SiO₂-PVA nanocomposite, the intensities for D1 and D2 bands relatively larger than those of the commercial silica glass. As increasing sintering temperature, intensities for D1 and D2 bands decreased. As a result, the intensities for the sample sintered at 1100˚C became to be comparable to those of the commercial silica glass. These changes in the intensities would involve structural change in silica membered rings. Amorphous silica contains silica membered rings such as four-membered (D1) and three-membered (D2) silica rings, and larger membered silica rings. In an industry, the fumed silica nanoparticle, used in this study as the starting material, is produced by hydrothermal reaction of SiCl₄ in oxyhydrogen flame at high temperature. The production process for the fumed silica nanoparticles gives characteristic surface properties such as large surface area and structural defects such as four-membered (D1) and three-membered (D2) silica rings. These four-membered and three-membered silica rings are considered to be less stable than the other membered rings such as six-membered rings [27] . Thus,

such small membered rings would be transformed to the other stable structure during sintering. Figure 4 and Figure 5 suggested that the structural change of the amorphous silica was observed in the sample sintered above 900˚C. Consequently, the amorphous structure for the sample sintered at 1100˚C is comparable to that of the commercial silica glass. From these results, it is considered that sintering of silica nanoparticles occurs above 900˚C, associated with change in the amorphous structure between transformed from three- and four-membered silica rings to the other membered rings.

Porous structure of sintered silica monolith is also controlled by sintering time, as well as sintering temperature. Figure 6 shows variation of BET surface areas by change in the treatment time at 1100˚C in air. The BET surface area of the sample decreased with increasing time. As shown in Figure 7, the pore size distribution for the sample shifted to lower side with increasing sintering time. These results indicated that the porous structure was tailored with BET surface

area in the range of ca. 0 - 300 m² g⁻¹ and pore size distribution in ca. 5 - 30 nm, respectively.

The sintering time affects not only on the porous structure but also an optical transmittance in UV-Vis and near-infrared regions. Figure 8 shows optical transmittance in UV-Vis and near-infrared regions for the sintered samples. For the sample (a) obtained by sintering at 1100˚C for 5 min, whose BET surface area was 146 m²・g⁻¹, the spectrum indicated opaque in the examined range. In contrast, for the sample; (b) obtained by sintering at 1100˚C for 30 min, whose BET surface area was 20 m²・g⁻¹, the sample was translucent. In order to clarify the relation between the morphology and the optical transmittance, we observed the microstructure of the sintered sample by using FE-SEM. Figure 9 shows FE-SEM images for these samples. For the opaque sample, the silica grains with several hundred nanometer and inter-particle pores were clearly observed. For

the translucent sample, meanwhile, the grain size became large and inter-particle pore size was reduced. It is considered that decreasing of inter-particle pore size reduced a light scattering at the particle interface. Therefore, transmittance of the porous silica increases with decreasing BET surface areas.

Figure 10 shows the porous silica monolith fabricated by using the present method. We obtained the disc shape with 63 mm diameter and 7 mm thickness. No crack was observed by the naked eye. The BET surface area for the porous silica monolith (e.g. 146 m²・g⁻¹) is much larger than that of silica monolith obtained by a conventional colloidal gel process (e.g. less than 50 m²・g⁻¹) [2] . Therefore, the present method is attractive alternative to fabricate porous silica monolith with over centimeter size possessing mesopores.