Theoretical Investigation of the Metal-Doped ... - ACS Publications

Mar 26, 2012 - The effects of a series metal dopants on the photocatalytic activity of SrTiO3-based photocatalyst are investigated using first princip...
0 downloads 0 Views 4MB Size
Article pubs.acs.org/JPCC

Theoretical Investigation of the Metal-Doped SrTiO3 Photocatalysts for Water Splitting Hsin-Chieh Chen, Chao-Wei Huang, Jeffrey C. S. Wu, and Shiang-Tai Lin* Department of Chemical Engineering, National Taiwan University, Taipei 10617, Taiwan ABSTRACT: The effects of a series metal dopants on the photocatalytic activity of SrTiO3-based photocatalyst are investigated using first principle DFT calculations. The SrTiO3:Rh(1%) loaded with Pt has been found to give the best efficiency in water splitting. However, the same host doped with Ru leads to very low H2 evolution rate even it has a better visible light response. The analysis of the density of states and the calculated absorption spectra were used to illustrate the mechanisms that influence the photocatalytic efficiency. Our calculation results suggested that the two competing factors, the free electron generation (via light harvesting) and the charge recombination (due to the presence of recombination centers) process resulted in the existence of the optimal dopant concentration for the transition-metal-doped SrTiO3 lattice. The energy states introduced by dopant Rh in the bandgap of SrTiO3 were found to be very close to the valence band maximum. These new states thus reduce the bandgap of catalyst and enhance its light absorption capability. Furthermore, the proximity of these states to the valence band allows for efficient electron replenishment and thus reduces the probability of trapping electrons from the conduction band. In contrast, the energy states introduced by dopant Ru are significantly higher than the valence band making them an isolated recombination center. These Ru associated states also reduces the driving force for oxidation reaction. As a result, the Rh-doped SrTiO3 catalysts are found to provide a high H2 evolution rate.



INTRODUCTION Hydrogen production via water splitting with solar energy provides an attractive way that produces hydrogen without reliance on fossil fuels and the emission of carbon dioxide.1−6 Photolysis of water requires radiation with a wavelength less than 190 nm.7 Fujishima and Honda8 showed that water can be decomposed under visible light with the use of a photocatalyst whose conduction band minimum (CBM) is higher (more negative) than the H+/H2 reduction potential (0 V vs NHE), while the valence band maximum (VBM) is lower (more positive) than the O2/H2O oxidation potential (1.23 V vs NHE). The ABO3-type perovskite strontium titanate (SrTiO3) is one of the promising photocatalysts that meets such requirements (see Figure 1) and has been explored extensively.2,9−11 However, the 10% commercial catalytic efficiency is far from achieved due to the lack of suitable photocatalytic material with appropriate electronic bandsturcture.2,5,12 The wide bandgap (3.3 eV)13 of this catalyst results in a less than 5% of sunlight absorption and low evolution efficiency. Introducing a small amount of foreign elements (usually transition metals) into the conventional metal oxide is a strategy frequently used to manipulate the bandgap.9,10,14,15 The dopants may generate new energy levels (Figure 1) between the VBM and CBM of the metal oxide resulting in a reduced minimum light absorption energy gap of the host lattice. Attempts have been made to dope particulate SrTiO3 with noble metals, such as Mn, Ru, Rh, Pd, Ir, and Pt to © 2012 American Chemical Society

enhance its photocatalytic activity and systematically examined the reaction mechanism.9,11 However, it was found that the lowered energy gap (and thus enhanced photon harvesting ability) does not guarantee an improvement in the H2 evolution efficiency. For example, the Rh-doped SrTiO3 gave the best H2 evolution rate while the Ru-doped one that possessed the best visible light harvesting ability had a negligible evolution rate.9,11 Other factors such as the driving potential for the redox reaction, the recombination of the active electrons, the structural defect of the crystal, and the energy barrier at the surface reaction site, also play an important role throughout the reaction process.16,17 Unfortunately, these factors are complicated and mutually correlated. It is difficult to evaluate each individual step experimentally, thereby making the design of high efficiency water-splitting photocatalysts a challenging task. We used first principle calculations to determine the electronic and optical properties of metal-doped SrTiO3 materials and analyze the factors that may affect their photocatalytic activity. Our results show that the position of the newly formed energy levels introduced by the metal dopants is of primary importance in determining the outcome catalytic activities. Furthermore, the two competing factors, light harvesting ability, and the amount of recombination Received: January 27, 2012 Revised: March 20, 2012 Published: March 26, 2012 7897

dx.doi.org/10.1021/jp300910e | J. Phys. Chem. C 2012, 116, 7897−7903

The Journal of Physical Chemistry C

Article

where ε1 and ε2 are the real and imaginary parts of frequency dependent complex dielectric constant, λ0 is the light wavelength of the corresponding frequency.



RESULTS AND DISCUSSION Pristine SrTiO3. Strontium titanate is a centrosymmetric material with a perovskite structure (Figure 2). The TiO3 atoms

Figure 2. Schematic plot of the 2 × 2 × 2 perovskite SrTiO3 supercell. Red and green spheres are oxygen and strontium atoms, respectively. The titanium atoms are located in the center of the octahedron.

Figure 1. Schematic plot of the band energy alignment of the photocatalyst and the redox potential of the water splitting reaction. The introduction of dopants creates new energy levels that effectively reduce the band gap of the photocatalyst.

form a framework of interconnected octahedrons, whereas the Sr atoms distribute over the tunnel space. The relaxed bond length of Ti−O is 1.959 Å, and the distance between Sr and its

centers result in the existence of the optimal dopant concentration for the transition-metal-doped SrTiO3 lattice.



COMPUTATIONAL DETAILS All the calculations were implemented by the vienna ab-initio simulation package (VASP).18−22 The Kohn−Sham equations were solved with the generalized gradient approximation (GGA) proposed by Perdew, Burke and Ernzerhof.23 Besides, the electron wave function was expanded in plane waves up to a cutoff energy of 400 eV. The electron−ion interaction was described by the pseudopotentials generated within the projector-augmented wave (PAW) scheme. The gammacentered k-point sets that ensured the fulfillment of the convergence criteria for both the geometry optimization and the static calculation for electronic properties are listed in Table 1. In the structure optimization (minimization of energy) Table 1. k-Point Sets Used in This Study system size unit 2×2×2 3×3×3 4×4×4

optimization 12 × 12 10 × 10 6×6× 1×1×

optical and electronic properties

× 12 × 10 6 1

15 × 15 12 × 12 8×8× 3×3×

× 15 × 12 8 3

procedure, the conjugate-gradient algorithm24 was used to relax the ions into their ground state structures. Both atomic positions and cell parameters were optimized until the residual forces were below 10−4 eV/Å. The optical absorption spectra for various designed configurations were calculated by converting the complex dielectric function obtained directly from the VASP program to the absorption coefficient αabs according to the following relation: αabs =

2 2π ( ε12 + ε22 − ε22)1/2 λ0

Figure 3. (a) Calculated and (b) the experimental13 band structure of perovskite SrTiO3. The Fermi levels are set to zero.

(1) 7898

dx.doi.org/10.1021/jp300910e | J. Phys. Chem. C 2012, 116, 7897−7903

The Journal of Physical Chemistry C

Article

resembles closely to the measured one in their shapes. The projected density of states (PDOS) shows that the valence band of pristine SrTiO3 is mainly composed of the O2p−Ti3d hybridized orbitals, whereas the conduction band is merely the contribution of the Ti3d orbitals. The overlap of O2p and Ti3d orbitals in the VB represents the component of covalent in Ti− O bond. On the other hand, the Sr3d orbitals contribute very little to both the valence band and the lower part of conduction band. In other words, the Sr atoms only construct the skeleton of the perovskite structure but do not affect the electronic structure around the Fermi level. Figure 4 shows the isosurface of the valence band maximum (VBM) and the conduction band minimum (CBM) of the pristine SrTiO3. It is clear that the VBM and CBM are composed of the nonbonding O2p orbitals and the Ti3d orbitals, respectively. This electronic characteristic is similar to that of TiO2 because the electronic states of Sr do not interfere with both band edges.27,28 The above calculation results for the pristine SrTiO3 demonstrate that the use of the present theoretical calculation is suitable for the study of SrTiO3. Doped SrTiO3. To investigate the effect of dopant on the light absorption capability, we calculated the optical and electronic properties of doped SrTiO3 by replacing one Ti atom in the 4 × 4 × 4 supercell (i.e., 1.6% dopant concentration) with one of the six different elements (Mn, Ru, Rh, Pd, Ir, and Pt). The calculated dopant−oxygen bond lengths, the lattice parameters, and the corresponding formation energies for the doped system are given in Table 2. The formation energy was calculated by Ef = (Edoped + μ Ti) − (Epristine + μ M)

where E denotes the total potential energy of the studied system; μTi and μM are the chemical potential of Ti and dopant M, respectively. Because the ionic radii of these metal dopants are similar to that of Ti, the calculated cell lengths and dopant− oxygen bond length remained essentially unchanged after doping. Furthermore, the formation energies upon doping are on the same order of magnitude. The degree of the local structural distortion is measured by the displacement in the distance between the dopant and its neighboring Ti and O. As shown in Table 3, the displacement of the dopant-Ti distances is generally small. The largest difference is found for the distance between the dopant and its nearest neighboring Ti atoms. The dopant is mainly accommodated by the bond length alternation between Ti and O, which also decays rapidly away from the defect center. These facts imply that the structural distortion arisen from the dopant is localized. Our calculation results are consistent with the experimental observations, which showed no significant structural change after doping the SrTiO3 lattice.11,29 The calculated absorption profiles of metal-doped SrTiO3 are shown in Figure 5. The result shows that the Ru-doped SrTiO3 has the broadest absorption region among the six types of dopants, which agrees with the experiment.9 Both theoretical and experimental results show that the Ru-doped SrTiO3 has the most prominent absorption in the visible region and thus has the best optical absorption efficiency. Theoretical calculations may also provide the position of energy states attributed from the dopants. Figure 6 shows the calculated densities of state (DOS) for the doped SrTiO3. The introduced dopants interfere very little with the semicore level of the host lattice. The reference energy was set by matching the position of their O2s semicore level to the pristine one.30

Figure 4. Orbital distribution of (a) the valence band maximum and (b) the conduction band minimum for the pristine SrTiO3.

Table 2. Structural Parameters, Ionic Radii,32 and the Corresponding Formation Energies of the M-Doped STO dopant type Ti (pristine) Mn Ru Rh Pd Ir Pt

ionic radius (Å)

M−O bond length (Å)

lattice parameter (Å)

0.61

1.959

3.917

0.53 0.62 0.60 0.62 0.63 0.63

1.903 1.968 1.976 1.981 1.973 1.990

3.915 3.917 3.917 3.917 3.917 3.918

(2)

formation energy (kcal/mol dopant)

126.8 159.8 175.6 194.2 170.6 180.9

ambient oxygen atoms is 2.770 Å. The optimized lattice parameters for pure SrTiO3 is 3.917 Å, which agrees well with the experimental value (3.91 Å).25 The calculated electronic structure of pristine SrTiO3 is depicted in Figure 3. All of the band energies are shifted in reference to the Fermi level, which is set to be zero. The bandstructure profile shows an indirect bandgap nature in agreement with the experimental observation. The lower calculated bandgap (1.73 eV) compared to the experimental value (3.3 eV)13 is a deficiency in the current state-of-the-art DFT method due to the insufficient cancellation of the selfinteraction correction inherent in the local exchange functional.26 Nonetheless, the calculated bandstructure profile 7899

dx.doi.org/10.1021/jp300910e | J. Phys. Chem. C 2012, 116, 7897−7903

The Journal of Physical Chemistry C

Article

Table 3. Average Distance Variation between the Dopant (M) and Its Neighboring Atoms M−(nth nearest Ti) (Å)

dopant type pristinea Mn Ru Rh Pd Ir Pt

1st 0.000 −0.023 0.005 0.003 0.012 0.006 0.010

2nd 0.000 −0.006 −0.002 −0.002 −0.002 −0.002 −0.001

3rd 0.000 −0.007 0.003 0.002 0.002 0.006 0.004

M−(nth nearest O) (Å) 4th 0.000 −0.005 0.000 0.000 0.000 0.000 0.000

1st 0.000 −0.056 0.009 0.017 0.022 0.014 0.031

2nd 0.000 −0.017 0.002 0.007 0.011 0.007 0.014

3rd 0.000 −0.010 −0.002 −0.002 −0.003 −0.002 −0.002

4th 0.000 −0.008 0.002 0.003 0.006 0.006 0.008

The distances between Ti and its first, second, third, and fourth nearest neighboring Ti and oxygen atoms in the pristine structure are used as reference. a

Figure 6. Calculated density of states for the (a) pristine STO, and the (b) Pd-, (c) Pt-, (d) Mn-, (e) Ir-, (f) Ru-, (g) Rh-doped STO. The Fermi level of the pristine STO is set to be zero, whereas the others are shifted to match the energy of O2s semicore with the pristine one.

These observations are in accordance with the calculated absorption spectra, where the absorption profiles of the Pd- and Pt-doped systems remain unchanged, whereas the others obtain additional absorption in the visible light region, especially for

Figure 5. (a) Calculated spectra and (b) the experimental absorption spectra for various kind of doped SrTiO3 (dopant concentration: expt. 0.5%, calaculation 1.6%). The absorption energy in the calculated spectra is shifted by 1.5 eV to compensate the underestimated band gap.

Table 4. Cell Sizes, Dopant Concentrations, Optimized Geometry Parameters, and the Formation Energies for the Rh-Doped Systems in This Study

The projected DOS of the doped Pd appears in the bottom of the conduction band and thus does not alter the bandgap. However, the projected DOS of the doped Pt lies in the upper part of the valence band, and thus only results in negligible disturbance to its band edge. However, the dopant levels of other dopants (Mn, Ru, Rh, and Ir) lie in the midgap, which may significantly narrow the minimum energy gap of the SrTiO3 lattice and are responsible for the photon absorption in the visible light region. The Ru-doped system shows the narrowest energy gap between dopant states and the VBM. 7900

dopant type

type of supercell

dopant concentration

M-O bond length (Å)

Rh

4×4× 4 + Rh 4×4× 4+ 2Rh 3×3× 3 + Rh 2×2× 2 + Rh

1.6%

1.976

3.917

175.56

3.2%

1.976

3.918

177.63

3.7%

1.977

3.915

173.76

12.5%

1.979

3.916

173.60

Lattice parameter (Å)

formation energy (kcal/ mol dopant)

dx.doi.org/10.1021/jp300910e | J. Phys. Chem. C 2012, 116, 7897−7903

The Journal of Physical Chemistry C

Article

Figure 7. Calculated absorption spectra for Rh-doped systems for various dopant concentrations: 0% (i), 1.6% (ii), 3.2% (iii), 3.7% (iv), and 12.5% (v), respectively.

Figure 9. Compiled PDOS of the Rh-doped and systems for various dopant contrations: 0% (i), 1.6% (ii), 3.2% (iii), 3.7% (iv), and 12.5% (v), respectively. The Fermi levels of each systems are set zero.

however, the newly formed dopant states appear in the middle of the bandgap; therefore, the electrons in these levels have an excitation energy considerably lower than that of the O2p electrons. However, the larger energy gap between the VB and the Ru states prohibits the replenishment of electrons via lattice thermal vibration (on the order of 10 meV). The vacancy states here can be refilled only by absorbing additional photons indicating that the lifetime of the Ru recombination center would be longer. In addition, the higher Ru-states compared to Rh-states implies that they are much closer to the O2/H2O oxidation potential (Figure 1) and thus reduces the driving force for the oxidation reaction. The strong charge trapping nature inherent in the Ru dopant and the reduced driving force for oxidation thus may result in its low evolution efficiency. Effect of the Dopant Concentration. Konta’s study showed the existence of an optimal dopant concentration for the Rh-doped SrTiO3 (i.e., 1%),9 indicating that there must be at least two competing factors related to the change of dopant concentrations. To identify these factors, we constructed systems with varying dopant concentrations using several supercells with different sizes and replaced one or two Ti atoms with Rh dopants. Table 4 summarizes the cell sizes, dopant concentrations, optimized geometry parameters, and the formation energies for the Rh-doped systems in this study. The calculated lattice parameters and the equilibrium dopantoxygen bond lengths show little change with dopant concentration and are almost identical to those of the pristine. Furthermore, the formation energy of the Rh-doped SrTiO3 lattice has no significant dependency on the dopant concentration. Figure 7 depicts the calculated absorption spectrum for the Rh-doped SrTiO3 lattices. The absorption of visible light is found to increase with increasing dopant concentration. More importantly, the introduction of Rh leads to direct band transition from the dopant levels to the CBM, whereas the SrTiO3 host has an indirect band transition

Figure 8. Calculated PDOS and bandstructure for the 3 × 3 × 3 SrTiO3 supercells doped with Rh.

the Ru-doped lattice. It is also important to note that these dopants might not affect the CBM of SrTiO3. This allows the reduction of the energy gap without reducing the driving forces for the H2 reduction process.31 Experimentally it was found that Rh-doped SrTiO3 exhibits the highest H2 evolution efficiency among these doped SrTiO3 lattice.9,11 The outstanding photocatalytic activity of the Rhdoped SrTiO3 may be explained by the suitable position of the dopant energy levels. The newly formed dopant states of the Rh-doped SrTiO3 lie slightly above the VBM. Therefore, the minimum energy gap is slightly reduced because the excitation energy of the electrons in these states is slightly lower than that of the O2p electrons. After the excitation process, the vacancies of electron have to be replenished, or the vacant dopant states would become a strong recombination center that traps the hot electrons traveling in the CB. The proximity of the Rh dopant states to the VB edge allows for rapid electron replenishment by absorbing either low energy photons or phonons (thermal vibration energy of the lattice). For the Ru-doped SrTiO3, 7901

dx.doi.org/10.1021/jp300910e | J. Phys. Chem. C 2012, 116, 7897−7903

The Journal of Physical Chemistry C

Article

Figure 10. Spatial distribution of the dopant energy levels for the (a) 1.6%, (b) 3.2%, (c) 3.7%, and (d) 12.5% Rh-doped SrTiO3. The dopant energy levels remain localized arounnd Rh regardless of its concentration in the photocatalyst.

energy levels remains the same so that the absorption efficiency (number of photons that can be absorbed) is not much improved. However, the localized nature of states introduced by Rh does not change with the increase of Rh concentration. The Rh atoms that just lose electrons after the photoexcitation process would present as active electron accepting sites. Because the charge separation process of the photogenerated electrons occurs at the catalytic surface, any doping site along or near the traveling path would possibly become a trap. Therefore, while the light absorption efficiency increases with the increase of concentration of Rh in SrTiO3 (Figures 7 and 9), the density of electron traps (spatially localized dopant states, Figure 10) increases as well and therefore lowers its internal quantum yield. The competition of these two effects can lead to an optimal Rh concentration for H2 evolution efficiency.

between the VBM and CBM (Figure 8). As a result, the absorption peak spreading over the visible light region becomes eminent even at low dopant concentrations. This visible region absorption becomes even more evident as the dopant concentration increases. Furthermore, there appears to be some additional absorption around the absorption peak corresponding to the O2p-Ti3d transition when the dopant concentration exceeds 4%. The enhanced absorption at 310 nm for the 12.5% Rh-doped SrTiO3 is ascribed to the dopant levels that stretch into the VB edge and enhance the transition probability from the top of the VB to the CBM. The calculated DOS for Rh-doped systems with different dopant concentrations are compared in Figure 9. The Fermi level of each doped system lies within the dopant states, meaning that every dopant inherently has empty states that can accept the active electrons. The dopant states become broader and more eminent as the concentration increases. Despite of the broadening of the dopant levels, the electronic structures of both band edges are not affected and the energy gap does not become significantly narrower. Moreover, the distribution of the dopant states remains localized (Figure 10) so that the replenishment of electron does not become easier. From the above analysis, the absorption intensity in the visible light region rises as the dopant concentration increases for the Rh-doped systems and thereby beneficial to the photocatalytic activity. However, the position of the dopant



CONCLUSIONS

Three key factors that affect H2 evolution efficiency of SrTiO3 are analyzed via first principle DFT calculations: (1) the light absorption efficiency, (2) the recombination of excited electrons, and (3) the driving forces for the oxidation and reduction reactions. Doping transition metals into the SrTiO3 lattice has been proven effective in creating visible light response by introducing new energy states into the bandgap, which effectively narrows the energy gap. However, these new 7902

dx.doi.org/10.1021/jp300910e | J. Phys. Chem. C 2012, 116, 7897−7903

The Journal of Physical Chemistry C

Article

(21) Kresse, G.; Hafner, J. Phys. Rev. B 1994, 49, 14251−14269. (22) Kresse, G.; Hafner, J. Phys. Rev. B 1993, 48, 13115−13118. (23) Perdew, J. P.; Burke, K.; Ernzerhof, M. Phys. Rev. Lett. 1996, 77, 3865−3868. (24) White, S. D. M. Scientist 1986, 1, 23−23. (25) Ohtomo, A.; Muller, D. A.; Grazul, J. L.; Hwang, H. Y. Nature 2002, 419, 378−380. (26) Cramer, C. J.; Truhlar, D. G. Phys. Chem. Chem. Phys. 2009, 11, 10757−10816. (27) Wei, W.; Dai, Y.; Guo, M.; Yu, L.; Jin, H.; Han, S.; Huang, B. Phys. Chem. Chem. Phys. 2010, 12, 7612−7619. (28) Scaife, D. E. Sol. Energy 1980, 25, 41−54. (29) Subramanian, V.; Roeder, R. K.; Wolf, E. E. Ind. Eng. Chem. Res. 2006, 45, 2187−2193. (30) Mrovec, M.; Albina, J. M.; Meyer, B.; Elsasser, C. Phys. Rev. B 2009, 79. (31) Mills, A.; LeHunte, S. J. Photoch. Photobiol. A 1997, 108, 1−35. (32) Shannon, R. D. Acta Crystallogr., A 1976, 32, 751−767.

dopant associated states may reduce the driving forces of the oxidation reaction, and may serve as charge recombination centers. The Rh-doped SrTiO3 exhibits best H2 evolution efficiency among the six doped lattice studied. The energy states introduced by dopant Rh in the bandgap of SrTiO3 was found to be close to the valence band maximum. The proximity of these states to the valence band allows for efficient electron replenishment and thus may reduce the probability of trapping electrons from the conduction band. More importantly, these states might not interfere with the edge of the conduction band, which plays an essential role for determining the driving force of H2 evolution. The increase of Rh concentration has a positive effect on the light absorption efficiency. However, due to the localized nature of the Rh energy states, the number of recombination centers increases with the increasing of Rh in SrTiO3. The existence of the optimum dopant concentration is attributed to two competing factors, the light harvesting ability and the amount of recombination centers.



AUTHOR INFORMATION

Corresponding Author

*E-mail [email protected]. Notes

The authors declare no competing financial interest.



ACKNOWLEDGMENTS The authors would like to thank the financial support from NSC 98-2221-E-002-087-MY3 and NSC 100-2221-E-002-137 by the National Science Council of Taiwan and computation resources from the National Center for High-Performance Computing of Taiwan and the Information Networking Center of National Taiwan University.



REFERENCES

(1) Chen, X. B.; Shen, S. H.; Guo, L. J.; Mao, S. S. Chem. Rev. 2010, 110, 6503−6570. (2) Kudo, A.; Miseki, Y. Chem. Soc. Rev. 2009, 38, 253−278. (3) Osterloh, F. E. Chem. Mater. 2008, 20, 35−54. (4) Lee, J. S. Catal. Surv. Asia 2005, 9, 217−227. (5) Maeda, K.; Domen, K. J. Phys. Chem. C 2007, 111, 7851−7861. (6) Youngblood, W. J.; Lee, S. H. A.; Maeda, K.; Mallouk, T. E. Acc. Chem. Res. 2009, 42, 1966−1973. (7) Coehn, A. Berichte Der Deutschen Chemischen Gesellschaft 1910, 43, 880−884. (8) Fujishima, A.; Honda, K. Nature 1972, 238, 37−38. (9) Konta, R.; Ishii, T.; Kato, H.; Kudo, A. J Phys Chem B 2004, 108, 8992−8995. (10) Miyauchi, M.; Takashio, M.; Tobimatsu, H. Langmuir 2004, 20, 232−236. (11) Bae, S. W.; Borse, P. H.; Lee, J. S. Appl. Phys. Lett. 2008, 92. (12) Maeda, K.; Domen, K. J. Phys. Chem. Lett. 2010, 1, 2655−2661. (13) Takizawa, M.; Maekawa, K.; Wadati, H.; Yoshida, T.; Fujimori, A.; Kumigashira, H.; Oshima, M. Phys. Rev. B 2009, 79, 113103. (14) Hwang, D. W.; Kirn, H. G.; Lee, J. S.; Kim, J.; Li, W.; Oh, S. H. J. Phys. Chem. B 2005, 109, 2093−2102. (15) Zou, Z. G.; Ye, J. H.; Sayama, K.; Arakawa, H. J. Photoch. Photobiol. A 2002, 148, 65−69. (16) Yerga, R. M. N.; Galvan, M. C. A.; del Valle, F.; de la Mano, J. A. V.; Fierro, J. L. G. ChemSusChem 2009, 2, 471−485. (17) Hisatomi, T.; Maeda, K.; Takanabe, K.; Kubota, J.; Domen, K. J. Phys. Chem. C 2009, 113, 21458−21466. (18) Kresse, G.; Furthmuller, J. Phys. Rev. B 1996, 54, 11169−11186. (19) Kresse, G.; Furthmuller, J. Comput. Mater. Sci. 1996, 6, 15−50. (20) Kresse, G.; Hafner, J. J. Phys.-Condes. Matter 1994, 6, 8245− 8257. 7903

dx.doi.org/10.1021/jp300910e | J. Phys. Chem. C 2012, 116, 7897−7903