(Ga1-xZnx)(N1-xOx) - American Chemical Society

Photodeposited Rh/Cr2O3 (core/shell) nanoparticles on a (Ga1-xZnx)(N1-xOx) photocatalyst are studied as a cocatalyst for overall water splitting. Irra...
0 downloads 0 Views 190KB Size
Download PDF
7554

J. Phys. Chem. C 2007, 111, 7554-7560

Roles of Rh/Cr2O3 (Core/Shell) Nanoparticles Photodeposited on Visible-Light-Responsive (Ga1-xZnx)(N1-xOx) Solid Solutions in Photocatalytic Overall Water Splitting Kazuhiko Maeda,† Kentaro Teramura,†,⊥ Daling Lu,‡ Nobuo Saito,§ Yasunobu Inoue,§ and Kazunari Domen*,† Department of Chemical System Engineering, The UniVersity of Tokyo, 7-3-1 Hongo, Bunkyo-ku, Tokyo 113-8656, Japan, Solution Oriented Research for Science and Technology (SORST) programs of the Japan Science and Technology Agency (JST), 4-1-8 Honcho, Kawaguchi-shi, Saitama 332-0012, Japan, and Department of Chemistry, Nagaoka UniVersity of Technology, Nagaoka 940-2188, Japan ReceiVed: February 7, 2007; In Final Form: March 22, 2007

Photodeposited Rh/Cr2O3 (core/shell) nanoparticles on a (Ga1-xZnx)(N1-xOx) photocatalyst are studied as a cocatalyst for overall water splitting. Irradiation of Rh-loaded (Ga1-xZnx)(N1-xOx) with visible light (λ > 400 nm) in aqueous K2CrO4 solution results in the formation of core/shell-structured nanoparticles consisting of a Rh core and Cr2O3 shell through the site-selective photoreduction of CrO42- anions to Cr2O3 on Rh nanoparticles. The shell thickness increases with concentration of K2CrO4 in the reactant solution to a maximum of ca. 2 nm, at which a uniform Cr2O3 coating is obtained. The Cr2O3 shell suppresses water formation from H2 and O2 on Rh nanoparticles, allowing stoichiometric decomposition of pure water to be achieved. The core/shell structure also provides enhanced H2 evolution compared to that of bare Rh nanoparticles.

1. Introduction It is known that the dispersion of nanoparticles of a noble metal (e.g., Pt, Rh, Au, and Ag)1 or transition-metal oxide (e.g., NiOx, RuO2, and Rh2-yCryO3)2-4 on a semiconductor photocatalyst improves the efficiency of electron transfer. In the overall water-splitting reaction, such nanoparticulate cocatalysts havebeenappliedwithgoodsuccesstofacilitateH2 evolution.1b-d,2-4 As shown in Scheme 1A, upon irradiation at wavelengths that exceed the band gap energy of the photocatalyst, photogenerated electrons in the photocatalyst migrate to the cocatalysts and reduce adsorbed H+ to H2. However, although noble metal functions as an efficient cocatalyst for H2 evolution, it also catalyzes water formation from H2 and O2, limiting its usefulness as a cocatalyst for overall water splitting on a particulate photocatalyst.1b Several approaches have been taken in attempts to solve this problem. Sayama and Arakawa.1c and Abe et al.1d reported that water formation on a Pt-loaded TiO2 catalyst during overall water splitting is suppressed by conducting the reaction in Na2CO3 or NaI aqueous solution. Very recently, our group attempted to solve this problem by devising a core/shellstructured noble-metal/Cr2O3 nanoparticle as a new type of cocatalyst to facilitate H2 evolution (Scheme 1B).5 The Cr2O3 shell surrounding the noble-metal nanoparticles prevents water formation during photocatalytic overall water splitting while maintaining the effectiveness of the noble metal as a cocatalyst, allowing both H2 and O2 to evolve at the stoichiometric ratio. In contrast to the more conventional single-base cocatalysts displayed in Scheme 1A, the core/shell cocatalyst effectively separates the two roles of the cocatalyst: extraction of photo* To whom correspondence should be addressed. Phone: +81-3-58411148. Fax: +81-3-5841-8838. E-mail: [email protected]. † The University of Tokyo. ‡ Japan Science and Technology Agency. § Nagaoka University of Technology. ⊥ Present address: Pioneering Research Unit for Next Generation, Kyoto University, Katsura, Nishikyo-ku, Kyoto 615-8510, Japan.

SCHEME 1: Roles of Cocatalyst Loaded on a Semiconductor Photocatalyst in Overall Water Splitting: (A) Conventional Cocatalyst and (B) Core/Shell-Structured Cocatalyst

generated electrons from the photocatalyst, and providing catalytic active sites for the reduction of H+. Another core/shell-structured cocatalyst, Ni-core/NiO-shell nanoparticles, has also been proposed for photocatalytic overall water splitting.2a-c,i However, while the Ni/NiO cocatalyst has been applied successfully in many heterogeneous photocatalytic systems,2,4d the loading process involves reduction and subsequent oxidation as activation treatment, which is somewhat disadvantageous for heat-sensitive photocatalysts such as those for overall water splitting.2e,3d,6 In contrast, the noble-metal/ Cr2O3 (core/shell) cocatalyst does not require such activation treatment and is thus more suitable for the preparation of efficient water-splitting photocatalysts.5 The noble-metal/Cr2O3 cocatalyst thus represents a new and effective approach for the design of efficient photocatalytic systems. In the previous communication, visible-light-driven overall water splitting was demonstrated using a (Ga1-xZnx)(N1-xOx) photocatalyst with photodeposited Rh/Cr2O3 nanoparticles.5 Of the various noble metals examined, Rh was found to be the most effective core material.5 However, detailed studies of the structural characteristics and photocatalytic activity of this core/shell system have yet to be carried out. Since the photocatalytic activity for overall water splitting is strongly dependent on the structure of the cocatalyst,2b,3e,4c it is important to investigate the structural

10.1021/jp071056j CCC: $37.00 © 2007 American Chemical Society Published on Web 05/02/2007

Photodeposited Nanoparticles on a Photocatalyst

J. Phys. Chem. C, Vol. 111, No. 20, 2007 7555

SCHEME 2: Preparation of Rh/Cr2O3 (Core/Shell) Nanoparticles on (Ga1-xZnx)(N1-xOx)a

a

Eg is the band gap energy of (Ga1-xZnx)(N1-xOx).

characteristics of the cocatalyst in order to achieve further increases in photocatalytic activity. The present study reports the results of a more detailed investigation of the core/shellstructured Rh/Cr2O3 nanoparticulate system on (Ga1-xZnx)(N1-xOx) with respect to the structural characteristics and photocatalytic activity under visible irradiation. 2. Experimental Section 2.1. Materials and Reagents. The (Ga1-xZnx)(N1-xOx) solid solution as the base photocatalyst was prepared according to the method described in previous papers.3c,d Briefly, a mixture of Ga2O3 and ZnO powders (1.08 g of Ga2O3 and 0.94 g of ZnO) was heated at 1123 K under NH3 flow (250 mL/min). After 15 h of nitridation, the sample was cooled to room temperature, maintaining the NH3 flow throughout. The production of (Ga1-xZnx)(N1-xOx) with x ) 0.12 was confirmed by powder X-ray diffraction (XRD) and energy-dispersive X-ray (EDX) analysis. The band gap energy of the as-obtained (Ga1-xZnx)(N1-xOx) is 2.68 eV, as estimated from the onset of the diffuse reflectance spectrum.3c,d Na3RhCl6‚2H2O (Kanto Chemicals, 97% as Rh) and K2CrO4 (Kanto Chemicals, 99%) were used as precursors for preparing the core/shell cocatalyst. For measurements of X-ray absorption fine structure (XAFS) spectra and X-ray photoelectron spectroscopy (XPS), Rh2O3 (Kanto Chemicals, 99.9%), Cr2O3 (Kanto Chemicals, 98.5%), CrO3 (Kanto Chemicals, 98.0%), and K2CrO4 were used as references. A solid solution of Rh2O3 and Cr2O3 (Rh0.5Cr1.5O3) was prepared as a reference according to the method described previously.4a,b 2.2. Preparation of Rh/Cr2O3 (Core/Shell) Nanoparticles and Photocatalytic Reactions. The preparation of Rh/Cr2O3 (core/shell) nanoparticles was carried out in a Pyrex inner irradiation-type reaction vessel connected to a glass closed gas circulation system. The (Ga1-xZnx)(N1-xOx) catalyst was first loaded with Rh nanoparticles by photodeposition using Na3RhCl6‚2H2O as a precursor. The amount of loaded Rh was confirmed by inductively coupled plasma atomic emission spectroscopy (ICP-AES) to be 0.75% ( 0.02% with respect to (Ga1-xZnx)(N1-xOx). The Rh-loaded sample (0.3 g) was then dispersed in aqueous K2CrO4 solution (370 mL, 2.34 × 10-3 to 0.234 mM). After evacuation, the solution was exposed to visible irradiation (λ > 400 nm) for 4 h to reduce K2CrO4 to Cr2O3. Irradiation was conducted using a 450 W high-pressure Hg lamp (UM-452, Ushio) and a Pyrex tube filled with sodium nitrite aqueous solution as a filter to block UV light.5 The temperature of the reactant solution was maintained at room temperature by a flow of cooling water during the preparation procedure. The final product was washed well with distilled water and dried overnight at 343 K. The formation of Rh/Cr2O3 (core/shell) nanoparticles on the (Ga1-xZnx)(N1-xOx) catalyst is illustrated in Scheme 2. Photocatalytic reactions were carried out using the same experimental setup as used for cocatalyst preparation. The

Figure 1. HR-TEM images of Rh-loaded (Ga1-xZnx)(N1-xOx) (a) before and (b-f) after visible irradiation for 4 h in aqueous K2CrO4 solutions with concentrations of (b) 2.34 × 10-3, (c) 1.56 × 10-2, (d) 7.80 × 10-2, and (e) 0.234 mM. (f) Same as (e) after 8 h.

evolved gases were analyzed by a gas chromatograph connected directly to the closed gas circulation system. The quantum efficiency (Φ) was estimated by the method described previously4a using the equation Φ (%) ) (2R/I) × 100, where R and I represent the number of evolved H2 molecules and the number of incident photons, respectively. It is assumed that all incident photons are absorbed by the photocatalyst. The number of incident photons was measured using a calibrated Si photodiode. 2.3. Characterization of Catalysts. The prepared samples were studied by high-resolution transmission electron microscopy (HR-TEM; JEM-2010F, Jeol), XAFS, and XPS (ESCA3200, Shimadzu). The binding energies determined by XPS were corrected in reference to the Au4f7/2 peak (83.8 eV) for each sample. XAFS measurements were carried out at the BL01B1 beamline of the SPring-8 synchrotron facility (Hyogo, Japan) using a ring energy of 8 GeV and stored current of 100 mA in top-up mode (proposal no. 2004B0075-NXa-np) and at the BL12C beamline of the Photon Factory (High-Energy Accelerator Research Organization, Tsukuba, Japan) using a ring energy of 2.5 GeV and stored current of 450-300 mA (proposal no. 2004G317 and 2006G125). The spectra were measured in transmission or fluorescence mode at room temperature using a Si(111) two-crystal monochromator. Data reduction was performed using the REX2000 program (Rigaku Corporation). The photon energies in X-ray absorption near-edge structure (XANES) spectra were corrected in reference to Cu foil (8980.3 eV) for each sample. The Fourier transforms of k3weighted extended XAFS (EXAFS) spectra were performed typically in the 3.0-12.0 Å region. 3. Results and Discussion 3.1. Effect of Initial K2CrO4 Concentration on Shell Thickness and Photocatalytic Activity. As illustrated in Scheme 2, the formation of the Cr2O3 shell occurs through the reduction of CrO42- anions by photogenerated electrons migrating from the (Ga1-xZnx)(N1-xOx) bulk to the external surfaces of Rh nanoparticles.5 Figure 1 shows HR-TEM images of the Rh-loaded (Ga1-xZnx)(N1-xOx) catalysts (Figure 1a) before and (Figure 1b-e) after visible irradiation in aqueous K2CrO4 solutions of various concentrations for 4 h. The primary size of Rh nanoparticles photodeposited on (Ga1-xZnx)(N1-xOx) was 2-3 nm (Figure 1a), although some secondary aggregates were observed as reported previously (not shown here).5 In the sample prepared in K2CrO4 solution with a concentration of 2.34 × 10-3 mM, the Rh nanoparticles were covered with a thin shell of ca. 1 nm thickness, and the deposited shell was irregular (Figure 1b). The Cr2O3 shell coating the Rh nanoparticles

7556 J. Phys. Chem. C, Vol. 111, No. 20, 2007

Figure 2. (A) Amount of photodeposited Cr in Rh/Cr2O3-loaded (Ga1-xZnx)(N1-xOx), and (B and C) photocatalytic activity for overall water splitting (B) and H2 evolution from aqueous methanol solution (10 vol %) (C) under visible irradiation (λ > 400 nm) as a function of initial K2CrO4 concentration. Reaction conditions: 0.15 g of catalyst, 370 mL of reactant solution, high-pressure mercury lamp (450 W) light source, Pyrex inner irradiation-type reaction vessel, aqueous NaNO2 solution as cutoff filter.

became thicker and more regular with increasing Cr concentration up to 7.80 × 10-2 mM (Figure 1d), above which no further changes were observed (Figure 1e). It has been confirmed by previous HR-TEM observations that the saturated shell thickness is ca. 2 nm, regardless of the Rh particle size.5 These results are supported by the ICP-AES analyses of the proportion of Cr in the final catalyst (Figure 2A), revealing that the saturation level of Cr is 0.31 ( 0.01 wt % with respect to the Rh-loaded (Ga1-xZnx)(N1-xOx). The photocatalytic decomposition of pure water was carried out using the catalysts as prepared above. Figure 2B shows the dependence of the average rate of H2 and O2 evolution in overall water splitting on the Rh/Cr2O3-loaded (Ga1-xZnx)(N1-xOx) catalyst on the initial K2CrO4 concentration employed for preparation of the core/shell cocatalyst. As reported previously, the Rh-loaded (Ga1-xZnx)(N1-xOx) without the Cr2O3 shell displays negligible activity, primarily due to rapid water formation from H2 and O2 at Rh nanoparticle sites.5 However,

Maeda et al. stoichiometric evolution of H2 and O2 was successfully obtained on the Rh/Cr2O3 (core/shell) system, regardless of the initial K2CrO4 concentration. The activity increased markedly with increasing initial K2CrO4 concentration to a sustained maximum at 7.80 × 10-2 mM K2CrO4 and above, consistent with the trend in Cr loading and shell thickness. The quantum efficiency of overall water splitting for the catalyst prepared using a K2CrO4 solution with concentration of 0.234 mM is calculated to be approximately 0.8% at irradiation wavelengths of 420-440 nm. In general, the photocatalytic activity for overall water splitting is known to be strongly dependent on the structure of the loaded cocatalyst, pH of the reactant solution, and so on.2b,3c,e,4b,c Since the quantum efficiency of Rh/Cr2O3-loaded (Ga1-xZnx)(N1-xOx) has yet to be optimized at present, it is expected to be improved by refining the preparation method and changing the reaction condition. Nevertheless, the rates of H2 and O2 evolution on the present catalyst were apparently lower than those on the previously reported Rh2-yCryO3-loaded (Ga1-xZnx)(N1-xOx) under the similar reaction condition.4c It has been reported that aggregation of cocatalysts directly contributes to a decrease in activity for photocatalytic overall water splitting.3e,4c Therefore, this low activity appears to be primarily due to poor dispersion (i.e., aggregation) of loaded Rh/Cr2O3 nanoparticles,5 although other factors such as electronic state of the cocatalysts are expected to affect the activity as well. Photocatalytic reactions using these catalysts in the presence of methanol as a sacrificial electron donor were also examined in order to evaluate the H2 production activity. In the presence of methanol, O2 is not produced because methanol is irreversibly consumed by photogenerated holes in the valence band of the photocatalyst.2f Water formation from H2 and O2 therefore cannot take place on Rh. Figure 2C shows the dependence of the rate of H2 evolution on Rh/Cr2O3-loaded (Ga1-xZnx)(N1-xOx) from aqueous methanol solution (10 vol %) under visible irradiation (λ > 400 nm) on the initial K2CrO4 concentration. All of the present catalysts produced H2 steadily, while no H2 evolution was observed on the bare (Ga1-xZnx)(N1-xOx) catalyst, indicating that the base catalyst does not host active sites for H2 evolution and that H2 evolution takes place on the nanoparticulate cocatalyst. The rate of H2 evolution increased with initial K2CrO4 concentration to a maximum at 7.80 × 10-2 mM K2CrO4 and above, similar to the behavior seen for overall water splitting (Figure 2B). The enhancement of activity for overall water splitting (Figure 2B) is dependent on the quality of the Cr2O3 shell formed on the Rh nanoparticles. On the irregularly coated Rh nanoparticles (Figure 1b), water formation from H2 and O2 would occur during the photocatalytic reaction, because the rapid water formation is observed even in a simple physical mixture of Rh- and Rh/Cr2O3-loaded catalyst (1:1 by weight), thus leading to low activity.5 Therefore, the increase in activity for overall water splitting with increasing initial K2CrO4 concentration appears to be associated primarily with the suppression of water formation by uniformly coating the exposed Rh nanoparticles with Cr2O3. The observation of a sustained maximum activity for all samples prepared at K2CrO4 concentrations of 7.80 × 10-2 mM and above can thus be attributed to the constant shell thickness obtained over this concentration range (Figures 1 and 2A). In the present system, a solution of at least 7.80 × 10-2 mM K2CrO4 is therefore required to prepare an effective core/shell structure. It is also apparent from the results displayed in Figure 2C that Rh/Cr2O3 (core/shell) nanoparticles are more efficient cocatalysts for H2 evolution than Rh alone, indicating that the catalytic activity for H2 evolution in overall

Photodeposited Nanoparticles on a Photocatalyst

Figure 3. (A) Rh K edge XANES spectrum and (B) Fourier transforms of k3-weighted EXAFS spectrum for Rh/Cr2O3-photodeposited (Ga1-xZnx) (N1-xOx). Spectra for Rh foil, Rh2O3, and Rh0.5Cr1.5O3 are shown as references.

water splitting is in fact increased by coating the Rh nanoparticles with a Cr2O3 shell, in addition to the suppression of water formation. It has been reported that, in general, the H2 evolution activity from aqueous methanol solution using metal-oxide photocatalysts is substantially higher than that from pure water.2g,h For example, the rate of H2 evolution by Ni-loaded K4Nb6O17 in aqueous methanol solution is about 17 times higher than that in pure water.2g In contrast, the activity of Rh/Cr2O3-loaded (Ga1-xZnx)(N1-xOx) for H2 evolution from aqueous methanol solution is not so high compared to that from pure water (Figure 2). A similar result has been obtained when Rh2-yCryO3loaded (Ga1-xZnx)(N1-xOx) was used as a photocatalyst.4b Therefore, it appears that the reason for the low H2 evolution activity is not due to the kind of loaded cocatalysts but (Ga1-xZnx)(N1-xOx) itself. Our preliminary experiments revealed that the rate of H2 evolution from aqueous methanol solution by Rh/Cr2O3-loaded (Ga1-xZnx)(N1-xOx) increases by about 2 times with increasing the concentration of methanol from 10 to 75 vol %, suggesting that the methanol oxidation process corresponds to the rate-determining step for H2 evolution from an aqueous methanol solution. Taking into account the fact that the rate of O2 evolution during H2 evolution reaction from aqueous methanol solution (10-75 vol %) is negligible regardless of the methanol concentration, photogenerated holes in the valence band of (Ga1-xZnx)(N1-xOx) preferentially oxidize methanol rather than water. Thus, one possible reason for the low H2 evolution activity from aqueous methanol solution is the slow reaction rate of methanol photooxidation on the (Ga1-xZnx)(N1-xOx) surface. Further investigation regarding this matter is now under way. 3.2. XAFS and XPS Measurements. The valence states of Rh and Cr in the Rh/Cr2O3 (core/shell) nanoparticles on the (Ga1-xZnx)(N1-xOx) surface were investigated by XAFS. The catalyst analyzed was prepared using an aqueous solution of 0.234 mM K2CrO4 (corresponding to Figure 1e). Figure 3 shows the Rh K edge XANES spectrum and the Fourier transforms of the k3-weighted EXAFS spectrum for this catalyst. The data for Rh foil, Rh2O3, and Rh0.5Cr1.5O3 are shown as references. The Rh K edge XANES spectrum for the sample obviously differs from that for both Rh2O3 and Rh0.5Cr1.5O3 but is somewhat similar to that for Rh foil judging from the spectral shape around the absorption edge and at 23260 eV. The peak

J. Phys. Chem. C, Vol. 111, No. 20, 2007 7557

Figure 4. (A) Cr K edge XANES spectrum and (B) Fourier transforms of k3-weighted EXAFS spectrum for Rh/Cr2O3-photodeposited (Ga1-xZnx) (N1-xOx). Spectra for Cr foil, Cr2O3, K2CrO4, and CrO3 are shown as references.

Figure 5. XPS spectrum for Cr 2p of Rh/Cr2O3-photodeposited (Ga1-xZnx)(N1-xOx). Spectra for Cr2O3, Rh0.5Cr1.5O3, and K2CrO4 are shown as references.

assigned to the Rh-Rh shell is clearly apparent in the Rh K edge EXAFS spectrum, indicating that the Rh species photodeposited on the (Ga1-xZnx)(N1-xOx) surface is almost entirely in the metallic form. Figure 4 shows the Cr K edge XANES and corresponding EXAFS spectrum for the same sample along with data for Cr foil, Cr2O3, K2CrO4, and CrO3. The XANES spectrum for the sample is quite different from that for Cr foil, K2CrO4, and CrO3, but the spectral region around the absorption edge and at 6020 eV for the sample resembles that for Cr2O3. The EXAFS spectrum for the sample is also similar to that for Cr2O3. The relatively weak peak at ca. 2.5 Å, assignable to the Cr-(O)-Cr shell, is suggestive of its small size, as supported by TEM observations (Figure 1e). The absence of metallic Cr was also confirmed from the EXAFS spectrum. Further investigation of the valence state of the Cr shell was performed by XPS. As expected from the result of TEM observations, the Cr 2p photoelectron signal appeared upon photodeposition, while the relative intensity of the Rh 3d peak against the Ga2p3/2 (or Zn2p3/2) peak became slightly lower. Figure 5 shows the XPS spectra for the Cr 2p of the core/shell material, along with reference spectra for Cr2O3 and K2CrO4.

7558 J. Phys. Chem. C, Vol. 111, No. 20, 2007

Figure 6. H2 and O2 evolution from aqueous K2CrO4 solution (0.234 mM) on (A) Rh-loaded and (B) bare (Ga1-xZnx)(N1-xOx) under visible irradiation (λ > 400 nm). Reaction conditions: 0.3 g of catalyst, 370 mL of aqueous K2CrO4 solution, high-pressure mercury lamp (450 W) light source, Pyrex inner irradiation-type reaction vessel, aqueous NaNO2 solution as cutoff filter.

The positions of the Cr 2p3/2 photoelectron signal for the Cr2O3 (576.7 eV) and K2CrO4 (579.7 eV) references are in agreement with reported values.7 It is clear that the Cr 2p3/2 peak position for the sample appears at the same position as for Cr2O3 (576.7 eV), and differs from that for K2CrO4 (579.7 eV), consistent with the results of XAFS measurements (Figure 4). Accordingly, it can be concluded that the Cr shell surrounding the Rh nanoparticles is composed of Cr2O3. 3.3. Cr2O3 Shell Formation on Rh Nanoparticles. Irradiation of Rh-loaded (Ga1-xZnx)(N1-xOx) in aqueous K2CrO4 solution under visible light (λ > 400 nm) to form the Cr2O3 shell is accompanied by H2 and O2 evolution (Figure 6A). In the previous study, it was revealed that the Rh/Cr2O3 (core/ shell)-loaded (Ga1-xZnx)(N1-xOx) does not produce appreciable H2 or O2 evolution when uncoated Rh-loaded (Ga1-xZnx)(N1-xOx) is dispersed in the same reactant solution, which is attributable to rapid water formation from H2 and O2 on the Rh nanoparticles.5 Therefore, the fact that both H2 and O2 evolution occurred immediately upon irradiation (Figure 6A) suggests that the formation of the Cr2O3 shell on the Rh nanoparticles takes place very quickly. In the second run, after evacuation (48 h), steady and stoichiometric H2 and O2 evolution was obtained, indicating establishment of the overall water-splitting reaction (2H2O f 2H2 + O2). The total amount of evolved H2 and O2 in the two runs exceeded 2 mmol, substantially greater than the amount of initial K2CrO4 (ca. 87 µmol). No gas evolution was observed without irradiation. A low level of N2 evolution (ca. 8 µmol) was detected in the initial stage of the reaction, attributed to the oxidation of N3- species near the (Ga1-xZnx)(N1-xOx) surface to N2.4b However, the production of N2 was completely suppressed as the reaction progressed, indicating that there was no continuous degradation of the catalyst. These results clearly demonstrate that the reaction proceeded photocatalytically. The amount of evolved H2 in the initial stage of the reaction was lower than expected from the stoichiometry, primarily attributable to consumption of a fraction of the photogenerated electrons by the reduction of CrO42- anions to deposit the Cr2O3 shell. As will be mentioned below, the amount of photodeposited Cr and the remnant Cr in the reactant solution after 4 h of irradiation is well-balanced. Assuming that CrO42- functions as an electron accepter to promote O2 evolution on the (Ga1-xZnx)(N1-xOx) according to the following equation,

4CrO42- + 8H+ f 4H2O + 2Cr2O3 + 3O2

Maeda et al. the amount of consumed CrO42- should correspond to 75% of O2 evolved. In the present case, the amount of initial CrO42- is ca. 87 µmol, 20% of which was consumed by the shell formation. Thus, the amount of O2 initiated by CrO42- reduction is calculated to be ca. 13 µmol. However, the amount of excessively evolved O2 (ca. 60 µmol) was apparently larger than the expected value (ca. 13 µmol). Therefore, the excess amount of O2 evolution is not only due to the consumption of a fraction of the photogenerated electrons by the reduction of CrO42anions to deposit the Cr2O3 shell but also other factor(s). It has been reported that, at the initial stage of the photocatalytic overall water splitting, an excessive H2 or O2 evolution is sometimes observed,3c,f,g although the origin of such excessive gas evolution has yet to be systematically clarified. In this case, it is speculated that any of constituent elements of the catalyst is reduced by the photogenerated electrons during the photodeposition of Cr2O3. If the reduction of CrO42- anions to Cr2O3 (i.e., Cr2O3 shell photodeposition) continues until all of the CrO42- anions are consumed, the rate of H2 and O2 evolution should not meet the stoichiometry even after extended irradiation. However, stoichiometric H2 and O2 evolution was observed in the second run after evacuation (4-8 h), implying that photodeposition of the Cr2O3 shell on Rh nanoparticles ceases once the shell has reached a certain thickness. After 8 h of irradiation, the Cr2O3 shell thickness had not increased from that observed after 4 h (see Figure 1, parts e and f), despite the solution containing excess K2CrO4 (as indicated by the characteristic yellow color of the solution). UV-vis spectroscopy revealed that approximately 80% of the K2CrO4 (vs initial concentration) remained in the reactant solution after irradiation for 4 h.8,9 Since CrO42- anions are a strong oxidizing reagent,10 it had been expected that the anions would act as efficient electron acceptors to facilitate O2 evolution on the (Ga1-xZnx)(N1-xOx) catalyst. However, when bare (Ga1-xZnx)(N1-xOx) was used instead of the Rh-loaded sample, no appreciable O2 evolution was detected in aqueous K2CrO4 solution (Figure 6B). This result indicates that no catalytic active sites are available for the reduction of CrO42- anions on the bare (Ga1-xZnx)(N1-xOx) surface, that is, Rh nanoparticles are the active sites for reduction of CrO42anions. As reported previously, (Ga1-xZnx)(N1-xOx) exhibits high photocatalytic activity for water oxidation in the presence of Ag+ cations as an electron accepter,4a indicating that electrons photogenerated in the conduction band of (Ga1-xZnx)(N1-xOx) can efficiently reduce Ag+ cations. It is thus of interest that (Ga1-xZnx)(N1-xOx) is unable to reduce CrO42- anions even though, thermodynamically, CrO42- anions are more readily reduced than Ag+ cations (Cr6+/Cr3+, 1.33 V; Ag+/Ag, 0.8 V vs NHE).10 To clarify the origin of the inactivity for CrO42reduction by (Ga1-xZnx)(N1-xOx), the adsorption of K2CrO4 on the (Ga1-xZnx)(N1-xOx) surface under conditions similar to that of the photocatalytic reaction was examined.11 However, no adsorption was observed at K2CrO4 concentrations of less than 0.39 mM. It has been reported that the adsorption of CrO42anions on a photocatalyst is essential for achieving the reduction of CrO42- anions,12 as this process requires direct electron transfer from the photocatalyst to CrO42-. Therefore, reduction of CrO42- anions onto the (Ga1-xZnx)(N1-xOx) surface will be unable to take place due to the lack of adsorption sites, contributing to the successful preparation of Rh/Cr2O3 (core/ shell) nanoparticles on the (Ga1-xZnx)(N1-xOx). It is also apparent that Cr2O3 shell deposition only occurs in the present system up to a thickness of ca. 2 nm. This result

Photodeposited Nanoparticles on a Photocatalyst

J. Phys. Chem. C, Vol. 111, No. 20, 2007 7559

indicates that H2 evolution, or reduction of H+, takes place preferentially on the Cr2O3 shell even in the presence of CrO42anions, which are more readily reduced than H+ (Cr6+/Cr3+, 1.33 V; H+/H2, 0 V vs NHE).10 For some photocatalysts, it has been reported that nanoparticles of noble metal are selectively photodeposited at specific sites or crystal faces on the photocatalyst surface.1e,f,13 For example, Tsuji et al. reported that Pt nanoparticles are selectively photodeposited at the edges of nanosteps on the surface of (AgIn)0.22Zn1.56S2.1e This type of phenomenon indicates that the reactivity, or reductive power, of the photogenerated electrons varies among surface sites or crystal faces. When Rh nanoparticles are photodeposited on the (Ga1-xZnx)(N1-xOx) surface, some aggregate to form larger secondary particles.5 Although the present (Ga1-xZnx)(N1-xOx) system with primary particles of 200-500 nm in size does not have a characteristic surface morphology as observed in a previous study,3d the aggregation of photodeposited Rh nanoparticles suggests that the reactivity of the photogenerated electrons also varies among surface sites on (Ga1-xZnx)(N1-xOx). If the thickness of the Cr2O3 shell is dominated by the reactivity of the photogenerated electrons, the shell thickness should similarly vary among Rh particles, that is, the shell should be thicker when photodeposited on larger aggregated particles. However, the photodeposited Cr2O3 has been confirmed to form a shell with uniform maximum thickness of ca. 2 nm, regardless of particle size,5 suggesting that the thickness of the Cr2O3 shell is independent of the reactivity of photogenerated electrons in this system. The reason for the inhibition of CrO42- reduction onto Cr2O3 shells of more than 2 nm in thickness remains unclear, and the detailed mechanism by which the Cr2O3 shell is formed is complex and requires further investigation. As noted above, the reduction of CrO42- to Cr2O3 occurs very quickly in the initial stage of irradiation, accompanied by H2 evolution due to the reduction of H+. Therefore, the two reduction processes will initially be in competition. It is thus difficult to trace these processes using the present experimental setup. One possible explanation for the self-limiting nature of the Cr2O3 shell formation is the difference in reduction process between Cr6+/Cr3+ and H+/H2. It appears that the photoreduction of Cr6+ into Cr3+ involving a complex three-electron redox process14 is more difficult to proceed than kinetically simpler processes, such as H+ reduction. Moffat et al. reported that electron tunneling can take place across an electrochemically passivated oxide layer with a thickness of 0.5-2.5 nm on a Cr electrode, and the tunneling electron transfer is hindered as the thickness of the passivated oxide layer increases.15 It is therefore considered that the present electron penetration through the Cr2O3 shell is due to such a tunneling electron-transfer phenomenon, but the efficiency of electron transfer from the Rh core to the external surface of the Cr2O3 shell decreases with increasing the thickness of the Cr2O3 shell. As a result, with increasing the thickness of the Cr2O3 shell, the reduction of CrO42- into Cr2O3, involving three-electron transfer, would be difficult to occur as compared to the H+ reduction. Finally, when the thickness of the Cr2O3 shell reached about 2 nm, the reduction of CrO42- ceased and the H2 evolution as a result of H+ reduction proceeded preferentially. The most important point of this study is the fact that the Cr2O3 shell functions as an effective cocatalyst for photocatalytic H2 evolution. It is considered that photocatalytic hydrogen production on the cocatalyst surface consists of the following three elementary steps:2i

H+(a) + e- f H(a)

(1)

H(a) + H(a) f H2(a)

(2)

H2(a) f H2(g)

(3)

where (1) adsorbed protons (H+(a)) are reduced into the H atom (H(a)) by photogenerated electrons, (2) two adsorbed H atoms are combined to form an adsorbed H2 molecule (H2(a)), and finally (3) the adsorbed H2 is released from the solid surface as gaseous H2 (H2(g)). Experimental results showed that the reduction of H+ into H2 molecule certainly takes place on the Cr2O3 shell coated on the Rh core. Therefore, the Cr2O3 shell has the ability to adsorb H+ and to activate the H atom. Cr2O3 is known to be an active catalyst for hydrogen-related reactions such as hydrogenation and dehydrogenation.16 Busca has reported that H atoms can adsorb on a hydrogen-activated Cr2O3 surface to form Cr-H bonds.16b Such surface nature of Cr2O3 might contribute to the good performance as a cocatalyst for photocatalytic H2 evolution. However, as we cannot investigate the reaction mechanism for H2 evolution on the surfaces of the Cr2O3 shell using the present experimental setups, the reaction mechanism will be investigated by means of an electrochemical and spectroscopic technique in future work. 4. Conclusion Core/shell-structured nanoparticles consisting of a metallic Rh core and trivalent Cr-oxide shell were successfully prepared as a cocatalyst on (Ga1-xZnx)(N1-xOx) by photodeposition using each aqueous solution of Na3RhCl6‚2H2O and K2CrO4. The material thus obtained was demonstrated to be functional as a photocatalyst for visible-light-driven overall water splitting. The Rh core promotes charge separation in (Ga1-xZnx)(N1-xOx) and tunneling electron transfer to the Cr2O3 shell, which prevents water formation from H2 and O2 while providing catalytic active sites for H2 production. Both the Rh core and the Cr2O3 shell are essential for achieving photocatalytic overall water splitting in this (Ga1-xZnx)(N1-xOx)-based system. Acknowledgment. The authors thank the staff of Mitsubishi Chemicals Co. for elemental analyses. This work was supported by the Solution Oriented Research for Science and Technology (SORST) program of the Japan Science and Technology (JST) Agency and the 21st Century Center of Excellence (COE) and Research and Development in a New Interdisciplinary Field Based on Nanotechnology and Materials Science programs of the Ministry of Education, Science, Sports and Culture (MEXT) of Japan. References and Notes (1) (a) Kraeutler, B.; Bard, A. J. J. Am. Chem. Soc. 1978, 100, 4317. (b) Yamaguti, K.; Sato, S. J. Chem. Soc., Faraday Trans. 1 1985, 81, 1237. (c) Sayama, K.; Arakawa, H. J. Chem. Soc., Faraday Trans. 1997, 93, 1647. (d) Abe, R.; Sayama, K.; Arakawa, H. Chem. Phys. Lett. 2003, 371, 360. (e) Tsuji, I.; Kato, H.; Kobayashi, H.; Kudo, A. J. Am. Chem. Soc. 2004, 126, 13406. (f) Tsuji, I.; Kato, H.; Kobayashi, H.; Kudo, A. J. Phys. Chem. B 2005, 109, 7323. (g) Subramanian, V.; Wolf, E.; Kamat, P. V. J. Phys. Chem. B 2001, 105, 11439. (h) Tada, H.; Ishida, T.; Takao, A.; Ito, S. Langmuir 2004, 20, 7898. (2) (a) Domen, K.; Naito, S.; Soma, M.; Onishi, T.; Tamaru, K. J. Chem. Soc., Chem. Commun. 1980, 543. (b) Domen, K.; Kudo, A.; Onishi, T.; Kosugi, N.; Kuroda, H. J. Phys. Chem. 1986, 90, 292. (c) Domen, K.; Kudo, A.; Onishi, T. J. Catal. 1986, 102, 92. (d) Kato, H.; Kudo, A. Catal. Today 2003, 78, 561. (e) Kudo, A.; Nakagawa, S.; Kato, H. Chem. Lett. 1999, 28, 1197. (f) Kudo, A. Catal. SurV. Asia 2003, 7, 31. (g) Kudo, A.; Tanaka, A.; Domen, K.; Maruya, K.; Aika, K.; Onishi, T. J. Catal. 1988, 111, 67. (h) Kato, H.; Asakura, K.; Kudo, A. J. Am. Chem. Soc. 2003, 125, 3082. (i) Domen, K.; Naito, S.; Onishi, T.; Tamaru, K. J. Phys. Chem. 1982, 86, 3657.

7560 J. Phys. Chem. C, Vol. 111, No. 20, 2007 (3) (a) Inoue, Y.; Kubokawa, T.; Sato, K. J. Phys. Chem. 1991, 95, 4059. (b) Sato, J.; Saito, N.; Nishiyama, H.; Inoue, Y. J. Phys. Chem. B 2001, 105, 6061. (c) Maeda, K.; Takata, T.; Hara, M.; Saito, N.; Inoue, Y.; Kobayashi, H.; Domen, K. J. Am. Chem. Soc. 2005, 127, 8286. (d) Maeda, K.; Teramura, K.; Takata, T.; Hara, M.; Saito, N.; Toda, K.; Inoue, Y.; Kobayashi, H.; Domen, K. J. Phys. Chem. B 2005, 109, 20504. (e) Teramura, K.; Maeda, K.; Saito, T.; Takata, T.; Saito, N.; Inoue, Y.; Domen, K. J. Phys. Chem. B 2005, 109, 21915. (f) Sato, J.; Saito, N.; Nishiyama, H.; Inoue, Y. J. Photochem. Photobiol., A 2002, 148, 85. (g) Sato, J.; Kobayashi, H.; Saito, N.; Nishiyama, H.; Inoue, Y. J. Photochem. Photobiol., A 2003, 158, 139. (4) (a) Maeda, K.; Teramura, K.; Lu, D.; Takata, T.; Saito, N.; Inoue, Y.; Domen, K. Nature 2006, 440, 295. (b) Maeda, K.; Teramura, K.; Masuda, H.; Takata, T.; Saito, N.; Inoue, Y.; Domen, K. J. Phys. Chem. B 2006, 110, 13107. (c) Maeda, K.; Teramura, K.; Lu, D.; Takata, T.; Saito, N.; Inoue, Y.; Domen, K. J. Phys. Chem. B 2006, 110, 13753. (d) Maeda, K.; Teramura, K.; Saito, N.; Inoue, Y.; Domen, K. J. Catal. 2006, 243, 303. (5) Maeda, K.; Teramura, K.; Lu, D.; Saito, N.; Inoue, Y.; Domen, K. Angew. Chem., Int. Ed. 2006, 45, 7806. (6) (a) Le Gendre, L.; Marchand, R.; Laurent, Y. J. Eur. Ceram. Soc. 1997, 17, 1813. (b) Tessier, F.; Le Gendre, L.; Chevire, F.; Marchand, R.; Navrotsky, A. Chem. Mater. 2005, 17, 3570. (7) Desimoni, E.; Malitesta, C.; Zambonin, P. G.; Riviere, J. C. Surf. Interface Anal. 1988, 13, 173.

Maeda et al. (8) The concentration of K2CrO4 in the reactant suspension was measured using a UV-vis spectrometer (V-560, Jasco). The change in concentration was estimated from a characteristic absorption band near 372 nm. (9) This result also confirms the material balance, as the ratio of K2CrO4 photoreduced upon 4 h of irradiation is calculated to be approximately 21% given an initial concentration of 0.234 mM K2CrO4 (1.5 wt % Cr). (10) Dean, J. A. Lange’s Handbook of Chemistry, 13th ed.; McGrawHill: New York, 1985. (11) Adsorption experiments were performed under conditions similar to the preparation of Rh/Cr2O3 nanoparticles. The (Ga1-xZnx)(N1-xOx) powder was dispersed in aqueous K2CrO4 solution under continuous stirring in the dark. After 4 h, (Ga1-xZnx)(N1-xOx) was separated by filtration and the supernatant was analyzed by UV-vis spectroscopy to estimate the change in K2CrO4 concentration. (12) Fu, H.; Lu, G.; Li, S. J. Photochem. Photobiol., A 1998, 114, 81. (13) Ohno, T.; Sarukawa, K.; Matsumura, M. New J. Chem. 2002, 26, 1167. (14) Testa, J. J.; Grela, M. A.; Litter, M. I. Langmuir 2001, 17, 3515. (15) Moffat, T. P.; Yang, H.; Fan, F. F.; Bard, A. J. J. Electrochem. Soc. 1992, 139, 3158. (16) (a) Lugo, H. J.; Lunsford, J. H. J. Catal. 1985, 91, 155. (b) Busca, G. J. Catal. 1989, 120, 303.