Single crystaln-Si(100) and p-Si(100) wafers with a thickness of 525 ± 25 µm were purchased from Kyodo International Inc. The n-Si(100) and p-Si(100) wafer surfaces were cleaned by a RCA cleaning method [26] . That is, the successive immersions of the wafers in a boiling mixture of 95% sulfuric acid (H₂SO₄) and 30% hydrogen peroxide (H₂O₂) at a volume ratio of 3:1, in a 5% hydrofluoric acid (HF) solution for 5 min, in a boiling mixture of 25% aqueous ammonium (NH₃), 30% H₂O₂ and distilled water at a volume ratio of 1:1:3 for 15 min, again in the 5% HF solution for 5 min, and in a 40% ammonium fluoride (NH₄F) solution for 5 min [26] . On the cleaned n-Si(100) or p-Si(100) surface, a WO₃ film was deposited by sputtering a W metal target under oxygen (O₂, 40 SCCM)/argon (Ar, 60 SCCM) gas mixture and 1.5 Pa for 16 min at substrate temperature of 400˚C, using a radio frequency (RF) magnetron sputtering apparatus (Tokuda, Model CFS-8EP). The thickness was controlled to be ~200 nm.

To obtain n-Si and p-Si powders, the purchased n-Si(100) and p-Si(100) wafers, respectively, were roughly pulverized by a mortar and then finely pulverized using a planetary ball-milling apparatus at 500 rpm for 5 min before use. Then WO₃ was loaded by a liquid phase deposition (LPD) on the surface of either pulverized n-Si or p-Si powder as follows [27] . Briefly, 5.01 g of tungsten acid (H₂WO₄, Kanto Chemical) was dissolved in 50 mL of an aqueous solution of 2% HF. 7.45 g of boric acid (H₃BO₃) was dissolved in 50 mL of distilled water and was used as the reagent which acts as F⁻ scavenger. These two solutions were mixed to use as the reaction solution for WO₃ deposition. 7.72 × 10⁻² g of either pulverized n-Si or p-Si powder (Si/WO₃ = 1/60 wt% or 1/7.3 mol%) was stirred with the mixed solution using a magnetic stirrer for 6 h at room temperature. The reaction product was obtained by a filtration, followed by washing with sufficient distilled water and drying at 50˚C. Then the samples were heated at 500˚C for 1 h in air.

The crystal structures of the prepared n-Si/WO₃ and p-Si/WO₃ electrodes and powders were examined by X-ray diffraction (XRD) using a PW-1700 system (Panalytical). A scanning transmission electron microscope (SEM, Hitachi, S-4500) was used to observe the morphology of the prepared samples. UV-visible absorption spectra for the n-Si/WO₃ and p-Si/WO₃ powders were obtained by the diffuse reflection method using a V-650 (JASCO) spectrometer.

The current-voltage (I-V) analysis in the presence or absence of light from a Xe lamp (LA-251Xe, Hayashi Tokei)for the n-Si/WO₃ and p-Si/WO₃ electrodes were performed in a conventional two electrode system using a potentiostat at (Hokuto Denko, HSV-10). To serve an ohmic electrode, platinum was deposited on WO₃ using a quick coater (Sanyu Electron Co., Ltd., SC-708) and indium (Kanto Chemical) was attached on either n-Si or p-Si.

The photocatalytic activity of the n-Si/WO₃ and p-Si/WO₃powdered photocatalysts were evaluated by the oxidative decomposition of gaseous 2-propanol irradiated with visible light (>420 nm, 1 mW/cm²) from the Xe lamp (the same above) equipped with a glass filter (Y-44, HOYA). For the analysis, 300 mg of the photocatalyst was uniformly spread over a 5.5-cm² irradiation area in a 500-ml quartz vessel. Prior to the injection of 6 µmol (~300 ppm) gaseous 2-propanol, the organic pollutants (originating from the air) absorbed on the surface of the photocatalysts were first photo-oxidized into CO₂ and the gas in the quartz vessel was then replaced with pure synthetic air (in the absence of CO₂ and organic pollutants). Following the injection of 2-propanol, the reaction vessel was kept in the dark overnight and was then subjected to visible light irradiation to start the photocatalytic reactions. The concentrations of acetone and CO₂ produced were monitored using a gas chromatograph (model GC-8A, Shimadzu Co., Ltd.).

Figure 1 shows XRD patterns of n-Si/WO₃ and p-Si/WO₃ electrodes. The faces of n-Si(100) and p-Si(100) were utilized to deposit WO₃, so the peak at ~69˚ corresponding to (400) should be large, and in fact the extremely large (400) peak of n-Si was observed. However, that of p-Si was not so large, which would be attributable to the deviation from the right angle in setting p-Si/WO₃ to the sample holder of the XRD apparatus. The peaks originated from WO₃ on both n-Si and p-Si wafers were quite similar, and WO₃ on both wafers was confirmed to have a single phase, probably the triclinic phase. WO₃ in both n-Si/WO₃ and p-Si/WO₃ powders was confirmed to have a single phase of triclinic WO₃ in the obtained XRD patterns (Figure 2). As for Si, being different from Figure 1, all the peaks originating from cubic Si were observed although some of the peaks overlapped with those from WO₃. The peak intensity of Si was not as high as that

of WO₃ because WO₃ powders completely covered the surface of Si as discussed below.

Figure 3 shows the cross sectional SEM image of the n-Si/WO₃ electrode. The dense WO₃ film with a thickness of ~200 nm was observed, similar to p-Si/WO₃ (not shown here). In Figure 4, the SEM image of the n-Si/WO₃ powder is shown. The entire surface of each Si powder was covered (Figure 4(a)) by the needle-like WO₃ powders (Figure 4(b)), which coincided well with the results of Deki et al. [27] . Figure 5 shows the UV-visible absorption spectra for commercially available WO₃, prepared n-Si/WO₃ and p-Si/WO₃ powders. The absorption over a wider wavelength region (>500 nm) clearly increased for n-Si/WO₃ and p-Si/WO₃, indicating the successful connection of WO₃ and n- or p-Si. In addition, the absorptions over 500 nm of n-Si/WO₃ and p-Si/WO₃ were similar within a several percent, so the amounts of WO₃ connected to n-Si and p-Si were presumed to be similar.

We examined the I-V analysis in the dark and under light irradiation as shown in Figure 6. The typical rectifying I-V (typical p-n junction) behavior in p-Si/ WO₃ was observed, particularly, under irradiation with light. It is plausible to consider the contact of p-Si and WO₃ (n-type semiconductor). That is, p-Si has the more negative energy of Fermi level (E_f) than that of WO₃ when we consider

the vacuum level as zero energy (Scheme 2(a)). In contrast, the ohmic I-V characteristic in n-Si/WO₃ was observed. As shown in Scheme 2(b), WO₃ has the more negative energy of E_f than that of n-Si, so it is probable to form the ohmic contact between n-Si and WO₃. In addition, irradiated with light, the current density of n-Si/WO₃ was demonstrated to be much larger than that of p-Si/WO₃. This means that the interparticle charge transfer, that is, charge transfer between photo generated holes in the valence band (VB) of n-Si and photoexcited electrons in the conduction band (CB) of WO₃ proceeded (Scheme 2(b)). Such an ohmic I-V characteristic was also observed in In₂O₃-Cu₂O system with poorphotovoltaic properties [28] . However, we can anticipate that the ohmic contact will function positively in terms of photocatalytic activity as discussed below.

We next examined the 2-propanol decomposition in the presence of the p-Si/ WO₃ and n-Si/WO₃ photocatalysts under visible-light irradiation (Figure 7). In the presence of n-Si/WO₃, the evolved acetone initially increased and then decreased. This decrease was accompanied by the increase in the CO₂ production. This behavior is plausible as it is known that 2-propanol decomposes into CO₂, which is the final product, via acetone, the intermediate product [29] . In contrast, in the presence of p-Si/WO₃, both acetone and CO₂ increased monotonically up to irradiation time of ~330 h. We cannot exclude the possibility that the acetone concentration would decrease after further irradiation of visible light, accompanied by the increase in the CO₂ evolution in the presence of p-Si/WO₃. Even in such a case, it is readily apparent that the CO₂ generation rate was smaller compared to that of n-Si/WO₃ during the acetone-increasing period. In addition, the longer acetone-increasing period indicates that acetone is reluctantly decomposed to CO₂ in the presence of p-Si/WO₃. It is generally accepted that 2-propanol is easily decomposed to acetone; however acetone is hardly decom

Scheme 2. Band alignments of (a) before and after connection of p-Si and WO₃, and (b) that of n-Si and WO₃. The charge transfer processes are also shown in the alignments after connection.

posed to CO₂. Thus, in any case, we can confidently conclude that the photocatalyticoxidative activity of n-Si/WO₃ is much higher than that of p-Si/WO₃.

It is well-known that the photo-produced holes play an important role in the generation of photocatalytic oxidative activity. In this sense, WO₃ is a candidate for having holes with strong oxidative power because the VB top potential of WO₃ is 3.1 - 3.2 V (vs. SHE, pH = 0 [30] ). The potential is even more positive than that of anatase TiO₂ (3.04 V vs. SHE, pH = 0 [31] ), which has already been widely utilized as practical applications. The photo-generated electrons also play a crucial role in the generation of the photocatalytic oxidative activity. That is, to generate the photocatalytic oxidative activity, the photo-generated electrons need to be consumed in the O₂ reduction because photocatalysts are usually utilized in air. If the photo-generated electrons are not consumed, the photo-produced holes will be recombined with them and eliminated. The CB bottom of TiO₂ lies at −0.16 V (vs. SHE, pH = 0 [31] ), which is slightly more negative than that of one-electron O₂ reduction (O₂ + H⁺ + e⁻ → HO₂, −0.046 V vs. SHE, pH = 0 [31] ). Thus, O₂ reduction is expected to proceed in the TiO₂ photocatalyst. In contrast, the CB bottom of WO₃ lies at 0.3 - 0.5 V vs. SHE [30] , so the photo-generated electrons cannot react with O₂ through one-electron reaction. This is the reason why WO₃ exhibits very low photocatalytic oxidative activity although the photo-produced holes in its VB have the strong oxidative power. To realize WO₃ for the highly active photocatalyst, either Pt or Cu(II) is loaded on WO₃ (Pt/WO₃, Cu(II)/WO₃) [32] [33] . Again, WO₃ exhibits very low photocatalytic oxidative activity because the photogenerated electrons cannot reduce O₂ through the one-electron reaction reduction. Pt or electron injected Cu(II) (i.e., Cu(I)) acts as a catalyst for multi-electron oxygen reduction (two electron reduction: O₂ + 2H⁺ + 2e⁻ → H₂O₂, 0.68 V; or four-electron reduction: O₂ + 2H₂O + 4H⁺ + 4 e⁻ → 4H₂O, 1.23 V [32] ).Thus the photogenerated electrons in either Pt/WO₃ or Cu(II)/WO₃ are consumed in the multi-electron reduction of O₂.

In the n-Si/WO₃ system (Scheme 2(b)), we consider that the 2-propanol decomposition performance was derived from the photo-produced holes with the strong oxidative power that were generated in the VB of WO₃ contributing to 2-propanol oxidation, and the photo-excited electrons that were generated in the CB of n-Si contributing to O₂ reduction through one-electron reaction. Importantly, the ohmic contact between n-Si and WO₃ acts as electron-and-hole mediator for the transfer of electrons and holes in the CB of WO₃ and in the VB of n-Si, respectively. Note that, to the best of our knowledge, Si (both n-Si and p-Si) does not function as the multi-electron O₂ reduction catalyst. In the p-Si/ WO₃ system (Scheme 2(a)), the photo-produced holes in p-Si do not have the potential to oxidize 2-propanol, considering its VB top potential. Contrastly, a portion of the photo-produced holes in WO₃ that exist on its surface across the Si particle can react with 2-propanol, and the photo-excited electrons reduce WO₃ itself to produce protonated WO₃ (H_xWO_3−y) and are eliminated. Thus, p-Si/WO₃ exhibited the activity for the 2-propanol decomposition, however the activity was low.