DFT study on microstructures and electronic structures of Pt mono-/bi-doped anatase TiO₂ (101) surface

Abstract

The microstructures and electronic structures of Pt mono- and bi-doped anatase TiO₂ (101) surfaces were investigated by density functional theory calculations to elucidate the surface doping effects and the initial stages of Pt cocatalyst formation on a TiO₂ photocatalyst surface. Several substitutional and interstitial configurations for the Pt impurity on the surface were studied, and the relative stability of different doping configurations was compared by the impurity formation energy. Under reducing conditions, surface interstitial (in other words, surface adsorbed) doping for Pt is more stable than any competitors (substitutional, bulk or subsurface doping) on an anatase TiO₂ (101) surface. Compared with bulk doping, the metal induced gap states were very localized and outstanding in the case of Pt surface interstitial doping onto a TiO₂ (101) surface. In the case of Pt substitutional doping, the surface doping effects were harmful for photocatalysis due to the metal induced gap states in the middle of the band gap. However, in the case of Pt interstitial doping, the surface doping effects were very favorable for photocatalysis due to the overlapping of metal induced gap states with the top of the valance band. Systematic calculations revealed that Pt doping is prone to occupy the interstitial sites on the (101) surface. In particular, in the case of Pt bi-doping, two Pt impurity atoms can be co-adsorbed on the surface to form a stable configuration due to the strong Pt–Pt atomic interaction. Therefore, Pt mono-/bi-doping on an anatase TiO₂ (101) surface can be considered as the initial stage (nucleation process) of Pt cocatalyst loading onto a TiO₂ photocatalyst surface. Therefore, the calculated results can form the basis for further investigations about Pt_n cluster loadings or Pt/TiO₂ interface formation as well as water molecule adsorption or decomposition on a Pt/TiO₂ composite photocatalyst.

---

1. Introduction

Over the past few decades, the technology of hydrogen generation from photocatalytic water splitting, as one of the most promising and potential renewable energy technologies, has attracted increasing interest. However, the solar quantum conversion efficiency is very low, which hinders the further development and industrial applications of this technology.¹ To promote the photocatalytic activity, researchers have attempted to load cocatalysts onto the photocatalyst surface to enhance the photocatalytic activity.^2–10 For example, Sato et al. reported that the reactivity of hydrogen and oxygen production is enhanced using a Pt/TiO₂ composite photocatalyst;⁹ Sclafani et al. loaded Ag particles on TiO₂ surface to improve the hydrogen evolution from the decomposition of alcohol, and attributed this improvement to Ag particles capturing the photogenerated electrons.¹⁰

When a cocatalyst is loaded onto a photocatalyst surface, the photogenerated electron–hole pairs can be spatially separated because electrons and holes can be respectively localized onto the surfaces of the semiconductor and cocatalyst. Consequently, recombination of the photogenerated electron–hole pairs can be inhibited, and the oxidation and reduction reactions can occur at different surface positions. In this way, the photocatalytic activity and selectivity can be improved greatly. Recently, the noble metal cocatalyst loading has been reported to be a suitable cocatalyst, because it can improve the solar quantum conversion efficiency and enhance the light-absorption efficiency. The common noble metals used as a cocatalyst in photocatalysis include Ag, Pt, Pd, Au, and Ru.^11–16 The Pt/TiO₂ composite is the most common system in this field. In addition, most studies have confirmed that the photogenerated electrons could transfer to Pt nanoparticles.^17–19 Compared to the pure TiO₂ semiconductor, Pt/TiO₂ composites show stronger photo-reduction ability. At the same time, the photogenerated holes can diffuse freely at the TiO₂ surfaces, and further photo-oxidize the absorbed organic compounds on the TiO₂ surfaces.²⁰ Facchin et al. loaded Pt particles on the TiO₂ surface using a sol–gel method, and observed a type of oxidation of functional groups on the sample surfaces, which enhances the photocatalytic activity of TiO₂.²¹

Most studies demonstrated that the tunable photocatalytic activity of a Pt cocatalyst loading on a TiO₂ surface can be achieved by controlling the Pt loading and the microstructure of the Pt/TiO₂ interface.^22,23 In addition to a Pt nanoparticle loading onto the surfaces, Pt surface doping is also another way for doping a TiO₂ photocatalyst. The incorporation of Pt atoms onto a TiO₂ substrate can be interstitial or substitutional doping as well as adsorption. Indeed, experiments have shown that Pt atoms can diffuse thermally into the TiO₂ lattice under an oxidizing atmosphere, and be oxidized to Pt²⁺ to substitute for Ti⁴⁺ or form interstitial ions.²⁴ Some reports have suggested that Pt acts as an electron–hole separation center and therefore inhibits electron–hole recombination at lower loadings, whereas Pt acts as an electron–hole recombination center at higher loadings.^12,25,26 Importantly, Han et al. used density functional theory (DFT) calculations to investigated the adsorption of Pt clusters on an anatase TiO₂ (101) surface, and demonstrated that Pt_n favors the coordinately unsaturated step edge sites, which may serve as nucleation sites for the growth of metal clusters on the oxide surface.²⁷ Gong et al. observed Pt clusters, as small as monomers, on the terraces of an anatase TiO₂ (101) surface with a large binding energy using DFT calculations and room-temperature scanning tunneling microscopy (STM) measurements;¹⁹ Zhou et al. also considered that Pt prefers to form 3-D rather than 2-D structures or monolayers on an anatase TiO₂ (101) surface.²⁸ Similar phenomena and characteristics were observed for Pd clusters on the anatase TiO₂ (101) surface.²⁹

However, these studies mainly considered the growth mechanism of Pt_n clusters on the anatase TiO₂ (101) surface. To the best of our knowledge, there are few detailed reports of the electronic properties of Pt-doped TiO₂ surfaces, revealing the physical and chemical origin of the enhanced photocatalytic performance of the Pt/TiO₂ composite photocatalyst. Furthermore, surface doping could be considered the initial stage for the Pt_n cluster loading or Pt/TiO₂ interface formation, which means the microstructure and properties will determine the subsequent stages. In a previous study, we systematically investigated the overall evolutionary process of the Pt cocatalyst loading on an anatase TiO₂ (101) surface and the related microstructure and properties. We found that the growth of a Pt cocatalyst on an anatase TiO₂ (101) surface sequentially experiences the following stages: surface doping, cluster nucleating, cluster loading, one-dimensional nanowire loading, two-dimensional nanowire grid loading, ultrathin film ripening, and film formation via the layer-by-layer mode, with increasing loading. In this growth process, Pt surface doping is crucial for the subsequent stages.³⁰ To further improve the fundamental understanding of the Pt/TiO₂ composite photocatalyst, and provide helpful information on the relationship between surface doping and the photocatalytic performance, Pt mono- and bi-doped anatase TiO₂ (101) surface were chosen as the research object, and DFT calculations were adopted as the research method to investigate their microstructure and electronic structure. The findings are expected to be crucial for further research on these noble metal–TiO₂ composite photocatalyst for practical applications.

2. Computational methods and models

The TiO₂ (101) surface was simulated by a (2 × 3) periodic slab of 12 O–Ti–O trilayers (72 TiO₂ units, 216 atoms). The slab model was separated by a 20 Å-thick vacuum layer. The lengths of the model along the [101̄], [010] and [101] directions were 11.0092, 11.3235 and 65 Å, respectively. Thus, all the lengths of the model were larger than 10 Å, which were sufficient to avoid the self interaction effects of the periodic boundary conditions. The bottom half of the slab was fixed to mimic the bulk effects. For the substitutional models, one or two Ti atoms were replaced with Pt atoms, whereas one or two Pt atoms were placed at the interstice sites of the surface for the interstitial models. To compare the similarity and difference between surface doping and bulk doping, a 3 × 3 × 1 supercell model was constructed for Pt bulk doping.

In the present study, all the DFT calculations were carried out using Cambridge Serial Total Energy Package (CASTEP) codes, employing the ultrasoft pseudopotential.³¹ The exchange and correlation effects were described by the revised Perdew–Burke–Ernzerhof for solid (PBEsol) of the generalized gradient approximation (GGA).³² An energy cutoff of 340 eV was used to expand the Kohn–Sham wave functions. The minimization algorithm used was the Broyden–Fletcher–Goldfarb–Shanno (BFGS) scheme.³³ The K-points grid sampling of Monkhorst–Pack scheme was set to 1 × 1 × 1 in the irreducible Brillouin zone, and the fast Fourier transform grid was set to 60 × 60 × 360. To obtain accurate results, we optimized the atomic coordinates that were obtained by minimizing the total energy and atomic forces. This was done by performing an iterative process in which the coordinates of the atoms were adjusted so that the total energy of the structure is minimized. The relaxation run was considered converged when the force on the atomic nuclei was less than 0.03 eV Å⁻¹, the stress on the atomic nuclei was less than 0.05 GPa, the displacement of the nuclei was less than 1 × 10⁻³ Å, and the energy change per atom was less than 1 × 10⁻⁵ eV. To improve the accuracy of the calculated adsorption energies for Pt atoms on the TiO₂ (101) surface, the dipole corrections were used for all models, which can be essential in eliminating the nonphysical electrostatic interactions between periodic images.³⁴ To obtain an accurate electronic structure, the GGA + U method was further adopted to overcome the well-known shortcomings of GGA. A U value of 4.2 eV was applied to the Ti-d states, Pt-d states and the O-p states. Using these values, the accurate band gap of anatase TiO₂ (3.209 eV) could be obtained and compared with the experimental measurements, and the main features of the electronic structure obtained by standard DFT calculations could be maintained.

Using the above calculation method, the bulk crystal structure of anatase TiO₂ was first optimized, and the following lattice constants were obtained: a = b = 3.7747 Å, c = 9.6289 Å, d_ap = 1.9898 Å, d_eq = 1.9329 Å, 2θ = 155.058°. These calculation results are consistent with the experimental measurements:³⁵ a = b = 3.7848 Å, c = 9.5124 Å, d_ap = 1.9799 Å, d_eq = 1.9338 Å, 2θ = 156.230°. The structure of a perfect TiO₂ (101) surface was then optimized using the above calculation method to obtain the surface configuration that is consistent with previous reports.^36,37 On the other hand, the lattice constants of fcc Pt metal was 4.0043 Å. These calculated results indicate that the calculation models and method in the present work are reasonable.

3. Results and discussions

3.1 Single Pt atom surface doping

When Pt atoms adsorb onto a TiO₂ surface, they are not only likely occupy the uppermost adsorption sites, but may also diffuse to the regime below the surface. In other words, for Pt surface mono-doping, not only is there a possibility of substitutional or interstitial doping for the topmost surface atoms, but also a certain possibility of substitutional or interstitial doping for the subsurface atoms. Thus, in this section, we first investigated the doping effects of a single Pt atom on an anatase TiO₂ (101) surface. For this purpose, various possible nonequivalent surface and subsurface doping positions were considered. To examine the relative difficulty for different doping configurations, the impurity formation energy (E_f) for every doping system was calculated according to the following equation defined by Van de Walle et al., which is a widely accepted method to compare the relative degree of difficulty or solubility of impurity incorporated into the host lattice:³⁸

E_f = E_pt/TiO2 − E_TiO2 − nμ_pt + mμ_Ti (1)

where E_pt/TiO2 is the total energy of Pt-doped TiO₂ system (bulk or surface), and E_TiO2 is the total energy of pure TiO₂ system (bulk or surface), which has the same corresponding atom numbers with the doping system. μ_pt and μ_Ti are the chemical potentials of Pt and Ti, respectively. n is the number of Pt doping atoms and m is the number of replaced Ti atoms. The chemical potential depends on the experimental growth conditions (oxidizing conditions or reducing conditions). For pure TiO₂, μ_Ti and μ_O satisfy the relationship, μ_Ti + 2μ_O = E_TiO2. Under extreme oxidizing conditions, 2μ_O is the total energy of an O₂ molecule in the gas phase; while under extreme reducing conditions, μ_Ti is the total energy of the bulk metal of Ti. In the present work, μ_pt is the total energy of bulk metal of Pt. In practice, the preparation of Pt doping is normally carried out between these two extremes, suggesting that the final doping configuration of Pt will depend on the synthesis conditions as well as after-preparation treatments. The impurity formation energies under reducing conditions are plotted in Fig. 1.

Fig. 1 The impurity formation energy of single Pt mono-doping on anatase TiO₂ (101) surface on different doping positions under reducing conditions.

Under reducing conditions, Pt interstitial doping on an anatase TiO₂ (101) surface is favored over substitutional doping. For Pt interstitial doping, the impurity formation energy first increases, and then decreases with increasing distance of the Pt impurity from the surface. The most stable interstitial position is the hollow site between two bridging O_2c atoms, as shown in Fig. 2(a). On the other hand, the impurity formation energy of Pt substitutional doping decreases with increasing distance from the surface. The most stable substitutional position is to replace a Ti_6c atom in the second trilayer below the surface at about 5.053 Å. Owing to the relaxation constraint along the directions parallel to the surface, the value of E_f for the surface system is larger than that of the corresponding bulk doping systems. It is particularly noteworthy that E_f of Pt doping at the hollow site (in other words, Pt adsorbs onto this position) is much lower than the E_f of all other doping positions, even lower than the E_f of bulk doping system. This suggests that Pt atoms are enriched more easily on the TiO₂ (101) surface when the sample is prepared under reducing conditions. The most stable doping configurations of Pt in the present work are identical to previous theoretical works.^19,27,28 To explore the modification effect of Pt on the TiO₂ (101) surface for the water adsorption and decomposition (which is ongoing by our group at present), we only considered the doping effect of Pt on the uppermost surface (as shown in Fig. 2) in the following sections.

Fig. 2 Top view (upper plane) and side view (lower plane) of stable models for: (a) Pt mono-doping; (b) Pt bi-doping. The red balls represent oxygen atoms; the grey balls represent titanium atoms; the navy blue balls represent platinum atoms. The top view only shows the top two O–Ti–O trilayers.

The electronic structure of Pt-doped anatase TiO₂ (bulk and (101) surface) in different forms is shown in Fig. 3. For both bulk doping, the impurity energy bands consist mainly of hybridization between the Pt-5d states and O-2p states, and they are located at the top and bottom of the valence band (VB). In addition, these impurity energy bands overlap with the top or bottom of VB. The main difference between them is shown by the effect of Fermi energy level (E_F) shifting: E_F more obviously shifted upward, even moves into the bottom of conduction band (CB), in the case of interstitial doping. For Pt substitutional doping of the TiO₂ surface, there is an isolated impurity energy band in the middle of the band gap, which consists predominantly of Pt-5d states. These impurity energy bands can be considered metal induced gap states (MIGS). This type of impurity energy band can become the recombination center of photogenerated electron–hole pairs. The impurity energy bands at the top or bottom of the VB are more prominent compared to Pt substitutional doping in bulk TiO₂. For Pt interstitial doping on the TiO₂ surface, there are three differences compared to Pt interstitial doping in bulk TiO₂ as follows: the impurity energy band located at the top of the VB is also more prominent, while the impurity energy band is located at the bottom of the VB disappeared, and the E_F is located at the top of the impurity energy band. The major advantage of this type doping is the realization of visible-light absorption, in which two lower-energy photons successfully excite one electron from the VB to the impurity energy band, and then to the CB.

Fig. 3 The local and partial density of states of Pt mono-doped anatase TiO₂: (a) pure bulk anatase TiO₂; (b) Pt substitutional doped into bulk anatase TiO₂; (c) Pt interstitial doped into bulk anatase TiO₂; (d) pure anatase TiO₂ (101) surface; (e) Pt substitutional doped into anatase TiO₂ (101) surface; (c) Pt interstitial doped into anatase TiO₂ (101) surface. The Fermi energy level (E_F) is located at 0 eV.

As shown in Fig. 4, the direction of electron transfer can be well described by maps of the electron density difference. Compared to pure TiO₂ (bulk or surface), the oxygen atoms neighboring the Pt atom obtain more electrons. In other words, the Pt atom loses more electrons than Ti atoms. However, there is an exception that the bridging O_2c atoms bonded to a Pt atom do not obtain more electrons and the Pt atom only loses a few electrons in the case of Pt interstitial doping on the surface. Either bulk doping or surface doping, the neighboring O atoms bonded to Pt collect more electrons in the case of Pt substitutional doping than that of Pt interstitial doping. Thus, in this type of doping case, the interaction of Pt atom with neighboring O atoms is relatively stronger. Furthermore, in these figures, the doping effect of Pt is rather localized and gradually disappears with increasing distance from the Pt atom. This implies that the Pt dopant only influences the electronic states of the neighboring atoms and the local crystal or surface structure.

Fig. 4 The electron density difference of Pt mono-doped anatase TiO₂: (a) pure bulk anatase TiO₂; (b) Pt substitutional doped into bulk anatase TiO₂; (c) Pt interstitial doped into bulk anatase TiO₂; (d) pure anatase TiO₂ (101) surface; (e) Pt substitutional doped into anatase TiO₂ (101) surface; (c) Pt interstitial doped into anatase TiO₂ (101) surface.

3.2 Double Pt atoms surface doping

Owing to the larger ionic radius of Pt, double Pt atom doping is very difficult to incorporate into the bulk TiO₂ host, neither substitutional nor interstitial doping. Furthermore, in practical photocatalysis, Pt is usually deposited onto the TiO₂ surface as a cocatalyst. Thus, in this subsection, we only considered Pt surface doping. For double Pt atoms surface doping, we considered three types of model: both substitutional doping, one substitutional doping plus one interstitial doping, and both interstitial doping. The corresponding E_f under reducing conditions is shown in Fig. 5. On the (101) surface, the variations of E_f as function of the dopant distance are very small over the entire surface. When the distance between two Pt atoms is increasing, the situation of these three types of doping is similar to single Pt atom doping. For these models, the values of E_f decreased with decreasing dopant distance, implying the existence of a strong Pt–Pt atomic interaction. As well as Pt mono-doping, the values of E_f of both interstitial bi-doping is much lower than that of other two bi-doping ways under reducing conditions. Moreover, it is negative, implying that the preparation process is an exothermic reaction. Because Pt atoms as the added atoms are deposited onto the surface in a practical sample preparation, in most cases, the most probable results are surface adsorbing (or surface interstitial doping), nanoparticle loading and interface forming. Furthermore, the E_f value decreases significantly when the distance between Pt atoms is less than 2.7 Å. At this distance, Pt atoms could be bonded together, resulting from the Pt–Pt atomic interaction. Therefore, Pt interstitial bi-doping can be seen as the process of nucleation of the Pt cocatalyst on the anatase TiO₂ (101) surface.

Fig. 5 The impurity formation energy of Pt bi-doping on anatase TiO₂ (101) surface as a function of the distance between the dopants under reducing conditions: (a) both substitutional doping; (b) one substitutional doping, and other interstitial doping; (c) both interstitial doping.

Fig. 6 shows the electronic structure of each stable configuration of the three doping ways. In the case of both substitutional doping, as shown in Fig. 6(a), the most stable configuration is where one Pt atom replaces the Ti_5c atom and the other Pt atom replaces the neighboring Ti_6c atom in the second trilayer. The distance between Pt atoms is 3.05 Å. There is an isolated impurity energy band in the middle of band gap, which is formed by the hybridization of Ti_5c-5d states with O_2c-2p states. At the top of the VB, there is an impurity energy band overlapped with the VB, which consists mainly of Ti_5c-5d states, Ti_6c-5d states, and O_2c-2p states. However, at the bottom of the CB, there is an impurity energy band overlapped with the CB, which is formed by the hybridization of Ti_6c-5d states with O_2c-2p states. In addition, at the bottom of the VB, there is also another impurity energy band that consists mainly of Ti_5c-5d states, Ti_6c-5d states, and O_2c-2p states. In the case of one substitutional doping plus one interstitial doping, as shown in Fig. 6(b), the most stable configuration is one Pt atom replaces the Ti_5c atom and other Pt atom occupies the neighboring hollow site between the two bridging O_2c atoms. The distance between Pt atoms is 2.779 Å. At the top of the CB, there are two impurity energy bands. The lower band is mainly formed by the hybridization of interstitial Pt-5d states with O_2c-2p states, and the higher band is mainly formed by the hybridization of substitutional Pt-5d states with O_2c-2p states. At the bottom of the CB, there is another impurity energy band that is almost isolated. It consists mainly of interstitial Pt-5d states, substitutional Pt-5d states and O_2c-2p states. The impurity energy band at the bottom of VB is relatively weakened. The most stable configuration for the case of both interstitial doping were described above, as shown in Fig. 2(b), and the corresponding electronic structure is shown in Fig. 6(c). In this case, there is only one impurity energy band at the top of the VB, which consist mainly of Pt_H-5d states, and the hybridization of Pt_L-5d states with O_2c-2p states. In the band gap and the bottom of the VB, there are no impurity energy bands. Similarly, the E_F positions are shifted relatively upward in the latter two cases, because of interstitial Pt atom doping. Compared to Pt interstitial mono- and bi-doping, a similar impurity energy band could be found on the top of the VB. Therefore, the Pt–Pt atomic interaction enhances the advantage of Pt interstitial doping for the photocatalytic activity, as mentioned above.

Fig. 6 The local and partial density of states of Pt bi-doped onto anatase TiO₂ (101) surface: (a) both substitutional doping; (b) one substitutional doping, and other interstitial doping; (c) both interstitial doping.

The results of the electron density difference shown in Fig. 7 confirmed the conclusion that electron transfer is more prominent for the local atoms around the Pt dopant. In the first case of both substitutional doping, Pt atoms lose more electrons than that of Ti_5c or Ti_6c atoms, and their neighboring O_3c atoms obtain more electrons, as shown in Fig. 7(a). However, the O_2c atoms that bonded with Pt atom do not seem to obtain more electrons. Instead, the number of electrons transferred to them is slightly lower than that of the other bridging O_2c atoms on the surface. On the contrary, the number of transferred electron to them is obviously more than that of the other bridging O_2c atoms on the surface in the latter two cases, as shown in Fig. 7(b) and (c). In the second case of one substitutional doping plus one interstitial doping, electron transfer of the substitutional Pt atom is more obvious than that of the interstitial Pt atom. In the case of both interstitial doping, the electron transfer of the lower Pt atom is also more obvious than that of the higher Pt atom. With different types of impurity energy bands and different modes of electron transfer, the transfer process of the photogenerated carriers from the TiO₂ substrate to the Pt cocatalyst will be influenced differently.

Fig. 7 The electron density difference of Pt bi-doped onto anatase TiO₂: (a) both substitutional doping; (b) one substitutional doping, and other interstitial doping; (c) both interstitial doping.

4. Conclusions

For a Pt cocatalyst loading onto the anatase TiO₂ (101) surface, its cocatalyst effects present more of the characteristics of surface doping at a very low coverage. First, it was found that the surface interstitial (in other words, surface adsorbed) doping for Pt are more stable than any competitors (substitutional, bulk or subsurface doping) on anatase TiO₂ (101) surface, under reducing conditions. In the case of Pt bi-doping, two Pt impurity atoms can be co-adsorbed on the surface to form a stable configuration, as a result of the strong Pt–Pt atomic interaction. Thus, Pt mono- and bi-doping of a anatase TiO₂ (101) surface could be considered as the initial stage (nucleation process) of Pt cocatalyst loading onto a TiO₂ photocatalyst surface. In the case of Pt interstitial surface doping, the metal induced gap states of Pt are very localized and outstanding. In the case of Pt substitutional doping, the surface doping effects are harmful to the photocatalysis due to the metal induced gap states in the middle of the band gap. However, with Pt interstitial doping, the surface doping effects are considerably favorable for photocatalysis due to the metal induced gap states overlapping with the top of the valence band. The major advantage of Pt interstitial surface doping is the realization of visible-light absorption, in which two lower-energy photons successfully excite one electron from the valence band to the impurity energy band, and then to the conduction band. Based on these calculated results, further studies on the Pt_n cluster loading or the formation of a Pt/TiO₂ interface, as well as the surface doping effects on water adsorption and decomposition are currently underway.

Acknowledgements

The authors would like to acknowledge financial support from the National Natural Science Foundation of China (Grant no. 21263006).

References

1. A. Kudo and Y. Miseki, Chem. Soc. Rev., 2009, 38, 253.
2. J. Yang, D. Wang, H. Han and C. Li, Acc. Chem. Res., 2013, 46, 1900.
3. G. N. Schrauzer and T. D. Guth, J. Am. Chem. Soc., 1977, 99, 7189.
4. S. Sato and J. M. White, J. Phys. Chem., 1981, 85, 592.
5. A. V. Bulatov and M. L. Khidekel, Russ. Chem. Bull., 1976, 25, 1794.
6. G. R. Bamwenda, S. Tsubota, T. Nakamura and M. Haruta, J. Photochem. Photobiol., A, 1995, 89, 177.
7. V. Subramanian, E. Wolf and P. V. Kamat, J. Phys. Chem. B, 2001, 105, 11439.
8. M. Jakob, H. Levanon and P. V. Kamat, Nano Lett., 2003, 3, 353.
9. S. Sato and J. M. White, Chem. Phys. Lett., 1980, 72, 83.
10. A. Sclafani, M.-N. Mozzanega and P. Pichat, J. Photochem. Photobiol., A, 1991, 59, 181.
11. N. Sobana, M. Muruganadham and M. Swaminathan, J. Mol. Catal. A: Chem., 2006, 258, 124.
12. G. N. Kryukova, G. A. Zenkovets, A. A. Shutilov, M. Wilde, K. Günther, D. Fassler and K. Richter, Appl. Catal., B, 2007, 71, 169.
13. L. Ge, M. Xu and H. Fang, Appl. Surf. Sci., 2006, 253, 2257.
14. Z. Bian, T. Tachikawa, P. Zhang, M. Fujitsuka and T. Majima, J. Am. Chem. Soc., 2014, 136, 458.
15. K. T. Ranjit, T. K. Varadarajan and B. Viswanathan, J. Photochem. Photobiol., A, 1995, 89, 67.
16. K. Drew, G. Girishkumar, K. Vinodgopal and P. V. Kamat, J. Phys. Chem. B, 2005, 109, 11851.
17. Z. Zhang and J. T. Yates, Chem. Rev., 2012, 112, 5520.
18. H. Tada, T. Kiyonaga and S.-i. Naya, Chem. Soc. Rev., 2009, 38, 1849.
19. X.-Q. Gong, A. Selloni, O. Dulub, P. Jacobson and U. Diebold, J. Am. Chem. Soc., 2008, 130, 370.
20. K. Hiehata, A. Sasahara and H. Onishi, Nanotechnology, 2007, 18, 084007.
21. G. Facchin, G. Carturan, R. Campostrini, S. Gialanella, L. Lutterotti, L. Armelao, G. Marcì, L. Palmisano and A. Sclafani, J. Sol–Gel Sci. Technol., 2000, 18, 29.
22. D. R. Jennison, O. Dulub, W. Hebenstreit and U. Diebold, Surf. Sci., 2001, 492, L677.
23. T. Tamura, S. Ishibashi, K. Terakura and H. Weng, Phys. Rev. B: Condens. Matter Mater. Phys., 2009, 80, 195302.
24. M. Zhang, Z. Jin, Z. Zhang and H. Dang, Appl. Surf. Sci., 2005, 250, 29.
25. B. K. Vijayan, N. M. Dimitrijevic, J. Wu and K. A. Gray, J. Phys. Chem. C, 2010, 114, 21262.
26. F. Han, V. S. R. Kambala, M. Srinivasan, D. Rajarathnam and R. Naidu, Appl. Catal., A, 2009, 359, 25.
27. Y. Han, C.-j. Liu and Q. Ge, J. Phys. Chem. B, 2006, 110, 7463.
28. Y. Zhou, C. L. Muhich, B. T. Neltner, A. W. Weimer and C. B. Musgrave, J. Phys. Chem. C, 2012, 116, 12114.
29. J. Zhang, M. Zhang, Y. Han, W. Li, X. Meng and B. Zong, J. Phys. Chem. C, 2008, 112, 19506.
30. Z.-Y. Zhao, J. Phys. Chem. C, 2014, 118, 24591.
31. S. J. Clark, M. D. Segall, C. J. Pickard, P. J. Hasnip, M. J. Probert, K. Refson and M. C. Payne, Z. Kristallogr., 2005, 220, 567.
32. J. P. Perdew, A. Ruzsinszky, G. I. Csonka, O. A. Vydrov, G. E. Scuseria, L. A. Constantin, X. Zhou and K. Burke, Phys. Rev. Lett., 2008, 100, 136406.
33. B. G. Pfrommer, M. Câté, S. G. Louie and M. L. Cohen, J. Comput. Phys., 1997, 131, 233.
34. J. Neugebauer and M. Scheffler, Phys. Rev. B: Condens. Matter Mater. Phys., 1992, 46, 16067.
35. J. K. Burdett, T. Hughbanks, G. J. Miller, J. W. Richardson and J. V. Smith, J. Am. Chem. Soc., 1987, 109, 3639.
36. M. Lazzeri, A. Vittadini and A. Selloni, Phys. Rev. B: Condens. Matter Mater. Phys., 2001, 63, 155409.
37. Z. Zhao, Z. Li and Z. Zou, J. Phys.: Condens. Matter, 2010, 22, 175008.
38. C. G. Van de Walle and J. Neugebauer, J. Appl. Phys., 2004, 95, 3851.

---