Materials Research Bulletin 44 (2009) 741–746
Contents lists available at ScienceDirect
Materials Research Bulletin journal homepage: www.elsevier.com/locate/matresbu
Electronic structure and photocatalytic water splitting of lanthanum-doped Bi2AlNbO7 Yingxuan Li, Gang Chen *, Hongjie Zhang, Zhonghua Li Department of Applied Chemistry, Harbin Institute of Technology, Harbin 150001, People’s Republic of China
A R T I C L E I N F O
A B S T R A C T
Article history: Received 13 June 2007 Received in revised form 17 September 2008 Accepted 24 September 2008 Available online 2 October 2008
Bi2xLaxAlNbO7 (0 x 0.5) photocatalysts were synthesized by the solid-state reaction method and characterized by powder X-ray diffraction (XRD), infrared (IR) spectra and ultraviolet–visible (UV–vis) spectrophotometer. The band gaps of the photocatalysts were estimated from absorption edge of diffuse reﬂectance spectra, which were increased by the doping of lanthanum. It was found from the electronic band structure study that orbitals of La 5d, Bi 6p and Nb 4d formed a conduction band at a more positive level than Bi 6p and Nb 4d orbitals, which results in increasing the band gap. Photocatalytic activity for water splitting of Bi1.8La0.2AlNbO7 was about 2 times higher than that of nondoped Bi2AlNbO7. The increased photocatalytic activity of La-doped Bi2AlNbO7 was discussed in relation to the band structure and the strong absorption of OH groups at the surface of the catalyst. ß 2008 Elsevier Ltd. All rights reserved.
Keywords: A. Semiconductors C. Infrared spectroscopy D. Catalytic properties
1. Introduction Hydrogen (H2) is an important chemical feedstock and is an ultimately clean form of energy. Photocatalytic splitting of water into H2 and O2 using oxide semiconductor powders has received much attention because of its potential for direct conversion of solar energy into chemical energy. Since the report of Fujishima and Honda on water splitting using a TiO2 photoelectrode , numerous attempts have been made to develop new semiconductor photocatalysts for efﬁcient water splitting [2–7]. For photocatalytic water splitting, the conduction band level should be more negative than the reduction potential of H2O to form H2 and the valence band should be more positive than the oxidation potential of H2O to form O2. Many photocatalysts with a pyrochlore-type structure with a generic composition of A2B2O7 were reported recently, such as Bi2MNbO7 (M = Al, Ga, In and Fe), Bi2RNbO7 (R = Y, rare earth elements) and Bi2MTaO7 (M = In, Ga, Fe, La and Y) [8–10]. Among these photocatalysts, Bi2AlNbO7 showed a relatively high activity. The rate of H2 evolution in a Pt/CH3OH/H2O cell with Bi2AlNbO7 is much larger than that of a TiO2 photocatalyst (TiO2–P25) . We consider that substitution of Bi3+ by La3+ in the Bi2AlNbO7
* Corresponding author. Permanent address: Department of Applied Chemistry, Harbin, Institute of Technology, Harbin 150001, People’s Republic of China. Tel.: +86 451 86413753; fax: +86 451 86413753. E-mail address: [email protected]
(G. Chen). 0025-5408/$ – see front matter ß 2008 Elsevier Ltd. All rights reserved. doi:10.1016/j.materresbull.2008.09.020
compound might yield a slight modiﬁcation of crystal structure and offer a change in photophysical properties. It is generally agreed that a slight modiﬁcation of the structure of a semiconductor has a dramatic effect on the concentration and mobility of charge , which directly affect the photocatalytic and photophysical properties of the semiconductor. Thus, it is important to clarify the effect of the doping on the photocatalytic properties in order to obtain information for the design of photocatalysts. In this paper, the effect of La substitution for Bi in Bi2AlNbO7 on the photocatalytic activity for water splitting will be discussed on the basis of the results by X-ray diffraction, UV–vis diffuse reﬂection spectra, infrared spectrum, BET surface area, and electronic band structure calculation. 2. Experimental 2.1. Preparation of samples Polycrystalline Bi2xLaxAlNbO7 (0 x 0.5) was synthesized by the conventional ceramic method from the starting materials of La2O3, Bi2O3 Al2O3, and Nb2O5 with a purity of 99.9%. These starting materials were heated to assure stoichiometry before use as follows: La2O3 was heated at 1273 K in air for 24 h. Bi2O3 and Al2O3 were heated at 1073 K while Nb2O5 was dried at 873 K in air for 12 h. A mixture of these starting materials with an appropriate molar ratio were ground well in an agate mortar and then pressed into pellets. The pellets were calcined at 800 8C for 24 h and 950 8C for 24 h in an alumina crucible with intermediate regrinding to
Y. Li et al. / Materials Research Bulletin 44 (2009) 741–746
complete reaction. At the ﬁnal process, the column samples were calcined and reacted for 1 day at 1273–1373 K depending on the composition. NiOx co-catalyst was loaded on the pyrochlore powders to promote H2 production. The photocatalyst powder was immersed in an aqueous solution containing the required amount of Ni(NO3)2. The solution was then evaporated to a dry solid using a water bath. The dry solid was then reduced by ﬂowing H2 at 773 K for 2 h and subsequent oxidation at 473 K for 1 h in order to get NiOx supported photocatalyst. 2.2. Characterization X-ray diffraction using a Rigaku D/max-2000 diffractometer equipped with Cu Ka radiation (l = 0.15406 nm) was carried out for the sintered samples at room temperature. The 2u range was 208 2u 808 with increment of 0.028. The infrared spectrum was recorded by a PerkinElmer Spectrum and a KBr-disc technique. Photocatalysts were measured using a UV–vis spectrophotometer (TU-1900). The surface area was measured by the BET method (ST2000). The plane-wave-based density functional theory (DFT) [12,13] calculation with the generalized gradient approximation (GGA) PW91 was carried out for the pyrochlore-type Bi2xLaxAlNbO7 to obtain information about the energy structure of the solid solutions. The core electrons were replaced with ultrasoft normconserving pseudopotential, and the O 2s22p4, Al 3s23p1, Bi 6s26p3, La 5s25p65d16s2 and Nb 4s24p64d45s1 electrons were treated explicitly. The kinetic energy cut-off was set at 300 eV, and the unit cell includes (Bi2AlNbO7)2 structure. The system structures were optimized by ﬁrst-principle calculations. The photocatalytic gas evolution from aqueous solutions of methanol was conducted in an outer irradiation quartz cell, which was connected to a closed gas-circulating system. A powder sample of Bi2xLaxAlNbO7 (0 x 0.5) was suspended in an aqueous CH3OH/H2O solution (0.1 g powder catalyst, 20 ml CH3OH, 400 ml deionized H2O) in the cell by use of a magnetic stirrer. The reaction was carried out by irradiating the mixture with light from a 350 W Hg lamp. Gas evolution was analyzed by a gas chromatograph (Agilent 6820, TCD, Ar carrier). 3. Results and discussion 3.1. X-ray diffraction Bi2AlNbO7 has the pyrochlore crystal structure of consisting of the network of MO6 (M = Al3+; Nb5+) as shown in Fig. 1. The structure described as an ordered cubic close-packed array of cations with the Bi and M (M = Al3+; Nb5+) cations located, respectively at the 16c and 16d sites in the space group Fd3m. Oxide ions occupy 7/8 of the tetrahedral sites between the cations. The oxide subarray is divided into two sets of tetrahedral sites. Six of the seven oxide ions are located at the 48f position with the remaining oxide ion residing on an 8a site. Two interstitial sites are available for oxygen. One is a tetrahedral site designated 8b, whereas the other is an octahedral site designated 32e. The pyrochlore structure is related to that of ﬂuorite, AO2, but with ordered cations and ordered vacancies in 1/8 of the tetrahedral anion sites. Fig. 2 shows the powder X-ray diffraction patterns of Bi2xLaxAlNbO7 (0 x 0.5). All the diffraction peaks were assigned to the Bi2AlNbO7 single phase with the pyrochlore structure with space group Fd3m . The similarity between the crystal structures of pure and La3+ doped samples suggests that the doped La3+ ion does not change obviously the crystalline structure of the pure Bi2AlNbO7; which may mainly result from the similar
Fig. 1. View of the (Bi2AlNbO7)4 structure, emphasizing three-dimensional arrangement of the (Al3+, Nb5+)O6 diamond-like octahedral framework. Small and large circles represent oxygen atoms, and Bi cations, respectively.
ionic size of La3+ and Bi3+. Meanwhile, no impurity phases are observed in all the XRD patterns, which also suggests that the doped La3+ ions substitute mainly for the Bi3+ ions in Bi2AlNbO7. That is, the doped La3+ ion is located at the A site of A2B2O7 pyrochlore lattice and no second phases are formed. 3.2. Infrared spectroscopy The infrared (IR) absorption spectroscopy can be used to characterize the compound. The IR absorption bands of solids in the range 100–1000 cm1 are usually assigned to vibrations of ions in the crystal lattice. There are seven IR-active optic modes originating from vibration and bending metal–oxygen bonds in the IR spectra of the pyrochlore oxides . The band (y1) at about 600 cm1 corresponds to the B–O stretching vibration in the BO6 octahedron and the band (y2) at about 500 cm1 is from the A–O0 in the AO6O20 polyhedron of A2B2O6O0 . The IR spectra of the investigated compositions recorded in the range 400–770 cm1 are shown in Fig. 3. It is reasonable to assign the band at 619 cm1 to the Nb–O stretching vibration and the band with a higher frequency (y1 = 763 cm1) to the Al–O stretching vibration. This is because the stretching vibration
Fig. 2. XRD patterns of Bi2xLaxAlNbO7 (0 x 0.5).
Y. Li et al. / Materials Research Bulletin 44 (2009) 741–746
lower wavelength upon increasing the La loading. Consequently, the band gap of La-doped Bi2AlNbO7 also increases gradually with increasing the La doping and is much higher as compared to that of pure Bi2AlNbO7. These observations strongly suggest that the La doping signiﬁcantly affects the absorbance properties. Moreover, these results are in good agreement with conclusions derived from the DFT electronic structure calculation results. 3.4. DFT electronic structure calculation
Fig. 3. IR spectra of Bi2xLaxAlNbO7 (0 x 0.5).
frequency (y) of the bond is related to the mass of the bonding atoms. The atomic weight of aluminum is about 3/10 of the atomic weight of niobium. The stretching vibration frequency of the bond decrease with increasing mass of the bonding atoms . Secondly, The characteristic frequency of the Nb–O stretching vibration in any NbO6 octahedron may be at about 650 cm1 . Similar results can be found in Eu2Cu2/3Nb4/3O7 and RE2Co2/3Nb4/ 1 3O7 (RE = Nd, Sm and Eu) [15,16]. The weak band at y2 = 563 cm is in the region expected for A–O0 stretching vibrations. It is obvious that the stretching vibration frequency at about y2 = 563 cm1 is caused by the Bi–O0 stretching vibration in the BiO6O20 polyhedron. The major lines associated with Bi2AlNbO7, as shown in our results in Fig. 3, is easily detected by the intense IR lines at 619, 563 and 763 cm1, which is in good agreement with previous report . The only Bi2AlNbO7 bands found in the IR spectra also suggested the La doped Bi2AlNbO7 maintains a cubic structure similar to the Bi2AlNbO7 even under extensive modiﬁcation by La ions.
In order to understand the doping effects on the photocatalytic activity of Bi2xLaxAlNbO7 (0 x 0.5) compounds, the electronic structures of Bi2AlNbO7 and La-doped Bi2AlNbO7 were studied by the ﬁrst principle calculations. The calculated band structure of Bi2AlNbO7 is shown in Fig. 5(a). Fig. 5(b) shows the total density of states (TDOS) and partial density of states (PDOS), corresponding to the energy region in Fig. 5(a). Both the valence band maximum (VBM) and the conduction band minimum (CBM) are located at the F point, indicating that Bi2AlNbO7 is a direct bandgap semiconductor. The calculated band gap (1.8 eV, see Fig. 5) is smaller than that obtained experimentally, which should be pointed out as a common feature of DFT calculations. This underestimation is an artifact of the GGA method used for this calculation . The top of the valence band is set at zero on the abscissa and is referred to as the valence band edge. The states in the valence band region from
3.3. UV–vis diffuse reﬂectance spectra UV–vis diffuse reﬂectance spectra of the catalysts are shown in Fig. 4. The reﬂectivity spectrum was transformed to absorbance intensity through Kubelka–Munk method. It can be clearly seen from Fig. 4 that the maximum of the absorbance band shifts toward
Fig. 4. UV–vis diffuse reﬂectance spectra of Bi2xLaxAlNbO7 (0 x 0.5).
Fig. 5. The calculated energy band and density of states (DOS) of Bi2AlNbO7: (a) the energy band and (b) the density of states.
Y. Li et al. / Materials Research Bulletin 44 (2009) 741–746
6 to 0 eV are mainly composed of the O 2p, Nb 4d and Bi 6s orbitals, but the contributions of Al 3s3p orbitals are not negligible. The conduction band of Bi2AlNbO7 is made up of the Nb 4d and Bi 6p orbitals, still little degree of Bi 6s and the contribution of O 2p and Al 3s3p orbitals is negligibly small. As shown in Fig. 5(b) the partial DOS of two types of oxygen sites, O(1) and O(2), distribute very differently, reﬂecting their different coordination environment and interatomic distances. The model of Bi1.5La0.5AlNbO7 contains fewer atoms than that of other compositions, which means that the electronic structure calculation of it can be done more easily. Moreover, the calculated results of Bi1.5La0.5AlNbO7 can illuminate completely the effect of La doping on the electronic structure. Therefore, the electronic structure of one solid solution Bi1.5La0.5AlNbO7 was calculated as a typical result in order to evaluate the atomic-orbital contributions to the valence and conduction bands and to analyze the factors that inﬂuence the sizes of their optical band gaps. Considering that doped La substituted for Bi in Bi2AlNbO7, the electronic band structure was calculated for a composition of Bi1.5La0.5AlNbO7. The model consisted of two (Bi2AlNbO7) units is shown in Fig. 6. As shown in Fig. 6, the model consisted of two distinct sites for Bi atoms. To test the validity of the calculation results, geometry optimization of these models was carried out to minimize the system energy. Model (b) gave the lower total energy of 10713.5644 eV, and relative energy measured from model (a) was 3.7754 eV higher than that from model (b). Therefore, the optimized model (b) was selected as the stable structure of the solid solution for the further band structure calculations. Fig. 7 shows calculated band structure, TDOS and PDOS of Bi1.5La0.5AlNbO7. The VBM is lies at Q point, while the CBM is at Z point. This means that Bi1.5La0.5AlNbO7 is an indirect-gap semiconductor material, which is different from Bi2AlNbO7. An indirect transition is favorable for carrier migration , which might be one of the factors for the improvement in the photocatalytic activities by doping of lanthanum. It can be seen from Fig. 6 that the width of the energy gap of the Bi1.5La0.5AlNbO7 crystal is 2.5 eV. The obvious result obtained is that the band gap of Bi1.5La0.5AlNbO7 is bigger than that of the Bi2AlNbO7 crystal calculated by DFT, which is in good agreement with the band gap estimated from the absorption edge of the diffuse reﬂectance spectra. From the PDOS of La, we can see that the contribution of La
Fig. 7. The calculated energy band and density of states (DOS) of Bi1.5La0.5AlNbO7: (a) the energy band and (b) the density of states.
5p5d6s orbitals to the valence band is negligibly small. Thus the valence band is not affected by substitution of La for Bi. However, the La 5d orbital makes a larger conduction to the conduction band. The similar distributions of the La 5d and Bi 6p partial DOS curves in the conduction band region from 0 to 6 eV suggest a considerable hybridization between La 5d and Bi 6p orbitals. Affected by this La 5d, the states of Bi 6p and Nb 4d would move to the high energy region resulting in broaden of the energy gap. This may be the origin of the violet shift of the absorption edge of the crystal doped as predicted by experiment. 3.5. Band structure model
Fig. 6. Structural models for Bi1.5La0.5AlNbO7 solid solution. In model (a), La atoms locate at the lattice plane of the supercell. In model (b), La atoms located at the center of the supercell. Models (a) and (b) preserve P1 symmetry.
With all of these results taken into consideration, the band structure model of Bi2xLaxAlNbO7 (0 x 0.5) can be illustrated schematically as shown in Fig. 8. In accordance with DFT electronic structure calculation of Bi2xLaxAlNbO7, both the valence band and the conduction band consist of Nb 4d and Bi 6s orbitals. However, the most noticeable difference between the La-doped Bi2AlNbO7 and undoped Bi2AlNbO7 should be observed for the position of La 5d levels. Band gaps of La-doped Bi2AlNbO7 were larger than that of nondoped Bi2AlNbO7, which is caused by the hybridization between La 5d and Bi 6p orbitals as suggested by the DFT calculated results (Fig. 7). The increase in the band gaps would be
Y. Li et al. / Materials Research Bulletin 44 (2009) 741–746
Fig. 9. Inﬂuence of the amounts of the NiOx co-catalyst on the activity of Bi1.8La0.2AlNbO7. Fig. 8. Band structures of Bi2AlNbO7 and Bi2xLaxAlNbO7 solid solution.
due to changes in the potential of conduction band. The contribution of La 5d orbital to the conduction band may increase with the increasing content of La ions in the solid solution in accordance with the continuous shifts in the UV–vis absorption spectra (as shown in Fig. 4). 3.6. Photocatalytic water splitting over Bi2xLaxAlNbO7 (0 x 0.5) For water splitting which is an uphill reaction, it is especially important to avoid the back reaction between H2 and O2 and their intermediates to form H2O. Loading of co-catalysts such as Pt, RuO2 and NiO is necessary for many photocatalyst materials to introduce the active sites for H2 formation. The efﬁciency of electron–hole separation and the dynamics of interfacial electron transfer can be dramatically inﬂuenced. Therefore, surface modiﬁcation will be useful for the design of the separated active sites to develop highly active photocatalysts for water splitting. Of several metals loaded on photocatalyst materials, nickel was found to be one of the most effective metals for water decomposition and this might also be due to the role of nickel in charge separation. As reported by Domen et al. , this variation in photocatalytic activity for water splitting with loaded nickel might be attributed to the role of nickel in p–n junction between nickel and the photocatalyst material. The NiOx co-catalyst was loaded onto the surface of the Bi2xLaxAlNbO7 (0 x 0.5) photocatalysts by the method in Section 2. The p–n junction was made possible by reduction at 773 K for 2 h followed by oxidation at 473 K for 1 h. The amount of the NiOx co-catalyst was varied between 0 and 0.50 wt% as Ni metal particles. Fig. 9 shows the rates of the H2 evolution under UV light Table 1 Photocatalytic activities of NiOx/Bi2xLaxAlNbO7 (0 x 0.5) for water splittinga. Amount of La doped (mol)
0 0.1 0.2 0.3 0.4 0.5
Band gap (eV)
2.86 2.89 2.89 2.94 2.94 3.20
Surface area (m2 g1)
0.43 0.53 0.52 0.48 0.42 o.42
H2 evolution (mmol h1 g1) Without NiOx
With NiOx (0.2 wt%)
45.4 63.8 105.3 77.4 66.2 65.6
73.7 108.8 141.4 123.2 119.3 120.8
a Catalyst: 0.1 g, reactant solution: 20 ml CH3OH, 400 ml deionized H2O, outer irradiation cell made of quartz, 350 W high-pressure mercury lamp.
irradiation of the Bi1.8La0.2AlNbO7 photocatalyst loaded by different amounts of the NiOx co-catalyst. The photocatalytic activity increased with the amount of nickel loading up to 0.2 wt% but the further increase of nickel loading was detrimental to photocatalytic water splitting. Both nickel oxide, a p-type semiconductor, and the photocatalyst material, an n-type semiconductor, should absorb the sufﬁcient photons needed for its band gap excitation so that the p–n junction can be operated properly. When this ratio was not optimized (nickel loading <0.2 wt% or >0.2 wt%), the photocatalytic activity of catalyst was reduced. This indicates that the optimum loading amount of the NiOx co-catalyst is around 0.2 wt% on the Bi1.8La0.2AlNbO7 photocatalyst. Table 1 shows the band gaps, BET surface areas and the photocatalytic activities of Bi2xLaxAlNbO7 (0 x 0.5) with and without a NiOx co-catalyst. The lanthanum doping improved the photocatalytic activity. In the naked Bi2xLaxAlNbO7 (0 x 0.5), the highest activity was obtained when 0.2 mol of lanthanum was doped. As the amount of doped lanthanum was increased, the activity was decreased. The photocatalytic activities of Bi2xLaxAlNbO7 (0 x 0.5) were increased when the NiOx co-catalyst was loaded. Among NiOx/Bi2xLaxAlNbO7 (0 x 0.5) photocatalysts, NiOx/Bi1.8La0.2AlNbO7 showed the highest activity. The rate of H2 evolution over this catalyst was 141.4 mmol h1 g1. The experimental results show that doping with La3+ can signiﬁcantly increase the photocatalytic activity of Bi2AlNbO7. The activity ﬁrst increased, reached a maximum, and then decreased with increasing doping concentration. In addition to the possible reason for the improvement in the photocatalytic activities by doping of lanthanum mentioned above, the change from directgap to indirect-gap semiconductor material, the La3+ at the surface of the photocatalysts is also considered to be an important factor. As reported, La substituted for Bi in Bi4Ti3O12 introduces a large quantity of oxygen vacancies at the surfaces of crystal . In addition, the lanthanum sites preferably populated close to the surface rather than in the bulk of particles . According to the above reports, the oxygen vacancies at the surface of Bi2xLaxAlNbO7 should be also introduced in the same way, which results to the strong absorption of OH groups at the surface of the photocatalyst . More OH groups at the surface of the photocatalyst can trap the photogenerated holes and form the OH radicals, which is assumed to be the reason for the improved photocatalytic property . A photocatalysis reaction to split water utilizing an oxide semiconductor includes: (i) the direct absorption of photons by the band gap of the material that
Y. Li et al. / Materials Research Bulletin 44 (2009) 741–746
generates electron–hole pairs in the semiconductor particles, the excitation of an electron from the valence band to the conduction band being initiated by light absorption with energy equal to or greater than the band gap of the semiconductor. (ii) Migration of the electrons to the surface of the particles, where reduction and oxidation of surface-adsorbed water by them takes place to produce H2 and O2, respectively, followed by desorption of H2 and O2 from the surface of the photocatalyst. The former is controlled by the energy band structure of a semiconductor, while the latter is strongly correlated with the bulk and surface properties of the semiconductor. Generally, the more electron–hole pairs migrate to the surface of the semiconductor, and the more active sites the surface possesses, the higher photocatalytic activity will be expected. How to suppress the energy-wasteful recombination of the formed electron–hole pairs is a principal challenge. Because recombination of the photogenerated electron and hole is so rapid, interfacial electron transfer is kinetically competitive only when the relevant donor or acceptor is preadsorbed before photolysis. In aqueous metal oxide suspensions, dangling hydroxy groups or water molecules were thought to serve as the surface-bound traps for the photogenerated holes and to form surface-adsorbed hydroxy radicals . The high concentration of the OH groups adsorbed at the surface of crystals will trap more photogenerated holes and thus prevent electron–hole recombination . The trapping sites can prolong the lifetime of charge carriers and improve their separations. Thus the photocatalytic properties of bismuth aluminum niobium oxide should be improved by La doping at a suitable concentration. The reason for the existence of an optimum doping concentration can be explained as follows. When the doping concentration is relatively low, the number of trapping sites increased with dopant concentration, resulting in an increase in photocatalytic activity. However, after the optimum concentration is reached, the OH groups may act as electron–hole recombination centers, thereby exerting a negative effect on the migration of current carriers and decreasing the photocatalytic activity. 4. Conclusion Bi2xLaxAlNbO7 (0 x 0.5) phases were prepared by solidstate reaction. The electronic structure, optical absorption and photocatalytic activity for water splitting have been investigated. Absorption bands in the infrared spectra of Bi2xLaxAlNbO7 (0 x 0.5) have been assigned. The DOS calculations conﬁrmed
that the VB of Bi2AlNbO7 is deﬁned by the O 2p, Nb 4d and Bi 6s orbitals and CB by the Nb 4d and Bi 6p orbitals. Compared with the band structure of Bi2AlNbO7, the bottom of the Nb 4d and Bi 6p bands (CB) of La-doped samples seem to have shifted upwards, due to the inﬂuence of the orbitals’ hybridization. On the basis of these electronic structures, their photocatalytic properties were discussed. It is revealed that photocatalytic activity is also dependent on the La3+ at the surface of the photocatalysts. Among NiOx/ Bi2xLaxAlNbO7 (0 x 0.5) photocatalysts, NiOx/Bi1.8La0.2AlNbO7 showed the highest activity. The present research provides useful information for designing new photocatalysts for water splitting. Acknowledgements This work was supported by the National Nature Science Foundation of China (Project No. 20571019), The Project-sponsored by SRF for ROCS, SEM, The Project-sponsored by SRF for ROCS, HIT, Development Program for Outstanding Young Teachers in Harbin Institute of Technology (HITQNJS.2006.028) and the Scientiﬁc Research Foundation of Harbin Institute of Technology (HIT.2002.54). References  K. Honda, A. Fujishima, Nature 238 (1972) 37.  K. Domen, S. Naito, T. Onishi, K. Tamaru, M. Soma, J. Phys. Chem. 86 (1982) 3657.  K. Sayama, H. Arakawa, J. Photochem. Photobiol. A: Chem. 77 (1994) 243.  A. Kudo, H. Kato, Chem. Phys. Lett. 331 (2000) 373.  J.W. Liu, G. Chen, Z.H. Li, Z.G. Zhang, J. Solid State Chem. 179 (2006) 3704.  H. Mizoguchi, K. Ueda, M. Orita, S. Moon, K. Kajihara, M. Hirano, H. Hosono, Mater. Res. Bull. 37 (2002) 2401.  Z. Zou, J. Ye, H. Arakawa, Mater. Res. Bull. 36 (2002) 1185.  Z. Zou, J. Ye, H. Arakawa, Chem. Mater. 13 (2001) 1765.  Z. Zou, J. Ye, H. Arakawa, J. Phys. Chem. B 106 (2002) 517.  J.F. Luan, X.P. Hao, S.R. Zheng, G.Y. Luan, X.S. Wu, J. Mater. Sci. 41 (2006) 8001.  J. Tang, Z. Zou, J. Yin, J. Ye, Chem. Phys. Lett. 382 (2003) 175.  P. Hohenberg, W. Kohn, Phys. Rev. B 136 (1964) 864.  M. Levy, Proc. Natl. Acad. Sci. U.S.A. 76 (1979) 6062.  G.V. Bazuev, O.V. Makarova, G.P. Shveikin, Russ. J. Inorg. Chem. 30 (1985) 1253.  Y. Xuan, R. Liu, Y.Q. Jia, Mater. Lett. 36 (1998) 198.  Y. Xuan, R. Liu, Y.Q. Jia, Mater. Chem. Phys. 53 (1998) 256.  Y. Maruyama, H. Irie, K. Hashimoto, J. Phys. Chem. B 110 (2006) 23274.  K. Domen, A. Kudo, T. Onishi, J. Catal. 102 (1986) 92.  A.Q. Jiang, Z.X. Hu, L.D. Zhang, Appl. Phys. Lett. 74 (1999) 114.  A. Yamakata, T. Ishibashi, H. Kato, A. Kudo, H. Onishi, J. Phys. Chem. B 107 (2003) 14383.  W.F. Yao, H. Wang, X.H. Xu, X.N. Yang, Y. Zhang, S.X. Shang, M. Wang, Appl. Catal. A 251 (2003) 235.  M.A. Fox, M.T. Dulay, Chem. Rev. 93 (1993) 341.