J. Phys. Chem. B 2001, 105, 3289-3294
3289
Photocatalytic Property and Electronic Structure of Lanthanide Tantalates, LnTaO4 (Ln ) La, Ce, Pr, Nd, and Sm) Masato Machida,* Shunsuke Murakami, and Tsuyoshi Kijima Department of Applied Chemistry, Faculty of Engineering, Miyazaki UniVersity, Gakuenkibanadai-nishi, Miyazaki 889-2192, Japan
Shigenori Matsushima Kitakyushu National College of Technology, Kitakyushu 802-0985, Japan
Masao Arai National Institute for Research in Inorganic Materials, Tsukuba 305-0044, Japan ReceiVed: NoVember 28, 2000; In Final Form: February 5, 2001
Photocatalytic activity of lanthanide tantalates, LnTaO4 (Ln ) La, Ce, Pr, Nd, and Sm), for water splitting was studied in connection to the effect of Ln 4f levels on the electronic structure. Valence band XPS, UVvis, and first-principle electronic calculations suggested that the position of Ln 4f levels becomes lowered monotonically across the series of Ln. The empty La 4f level is supposed to be higher than the conduction band edge, whereas the filled 4f levels of the Nd and Sm compounds lie in the valence band. For Ln ) Ce and Pr, occupied 4f levels lying within the forbidden gap lead to the lower band gaps (Eg) compared to others. The highest photocatalytic activity was attained by LaTaO4, which solely can decompose water into stoichiometric H2/O2 mixtures. This is largely due to the overlap between Ta 5d and O 2p orbitals, which produces a wide density-of-states distribution in the conduction band. On the other hand, the tantalates containing from one (Ce3+) to five (Sm3+) 4f electrons were much less active, because unoccupied 4f levels lying on or below the conduction band edge would play as a trapping center for photoexcited electrons. The interactions between the empty and filled Ln 4f levels and carriers generated under UV irradiation are possible reasons for the Ln-dependent photocatalytic activity.
Introduction Despite a number of studies on perovskite-related oxide (ABO3) photocatalysts,1-11 there were no reports so far on the effect of A-site lanthanide (Ln) on their activity. Generally, it was recognized that the band gap transition responsible for photocatalytic processes is mainly related to B-d and O-p states, whereas the localized character of Ln 4f levels would make little contribution. In our proceeding papers;12,13 however, we have first demonstrated the Ln-dependent photocatalytic activity of RbLnTa2O7 (Ln ) La, Pr, Nd, and Sm) for overall water splitting. All of these materials that crystallized in a layered perovskite-type structure exhibited the activity increasing in the sequence of Pr < La , Sm < Nd. To elucidate such an unexpected effect of Ln, we have next applied valence band XPS measurement and electronic structure calculation based on all-electron full potential linear augmented plane wave (FLAPW) method.14 The result revealed that replacing Ln lowers the energy level of Ln 4f bands continuously with increasing the number of 4f electrons, i.e., the unoccupied La 4f level is located on the bottom of the conduction band, whereas the partially occupied 4f levels become lowered with an increase of 4f electrons down to the position above the valence band edge (Ln ) Pr) and finally fall into the valence band mainly comprised of O 2p orbitals (Ln ) Nd and Sm). Moreover, it * To whom correspondence should be addressed. Fax: +81-985-587312. Tel: +81-985-58-7323. E-mail:
[email protected].
was also suggested that unoccupied as well as occupied Ln 4f orbitals are not completely localized but partly contribute to the hybridization with O 2p and Ta 5d orbitals. Therefore, the photocatalytic activity seems to be closely related to the degree of the Ln-O-Ta hybridization, which affect not only the edges of conduction band and valence band but also the density of states (DOS) distributions in both bands. Although RbNdTa2O7 is a first example of active photocatalysts containing partially occupied Ln 4f levels, for more general understanding of their role, further studies are requested for other tantalate photocatalysts containing various different lanthanides. This paper reports the synthesis, spectroscopic characterization, electronic structure calculation, and photocatalytic property of simple binary oxides with the composition of LnTaO4. The electronic structure was discussed by using XPS and UV-vis data as well as first-principle calculation. The oxides both as prepared and partially oxidized nickel (NiOx) loaded were applied to the photocatalytic water splitting, and the result was correlated to the electronic structure. Finally, comparisons were made with the results in our previous work on RbLnTa2O7 to elucidate the role of Ln 4f levels on the photocatalytic property. Experimental Section Sample Preparation and Characterization. LnTaO4 (Ln ) La, Ce, Pr, Nd, and Sm) was synthesized by calcining powder mixtures of oxides, Ln2O3 (Ln ) La, Nd, and Sm), CeO2 or Pr6O11, and Ta2O5 (99.99%, Rare Metallic Co., Ltd., Japan) at
10.1021/jp004297z CCC: $20.00 © 2001 American Chemical Society Published on Web 03/29/2001
3290 J. Phys. Chem. B, Vol. 105, No. 16, 2001 1200 °C for 10 h. Calcination of Ce compounds was carried out in a stream of H2, whereas calcination of all other compounds was carried out in air. The Ni-loaded catalyst was prepared by impregnation of as-prepared powders with an aqueous solution of Ni(NO3)2 (Wako Pure Chemicals Ind., Ltd., Japan). The impregnated sample was then submitted to reduction in a stream of H2 at 500 °C and subsequent reoxidation in a stream of O2 at 200 °C to prepare partially the oxidized nickel (NiOx) catalyst.15 The loading amount of Ni metal was from 0.1 to 2.0 wt %. Characterization. The crystal structure of samples was identified by the use of a powder X-ray diffractometer (XRD; Shimadzu XD-D1) with monochromated Cu KR radiation (30 kV, 20 mA). Diffuse reflectance spectra were recorded with a Jasco V-550 UV-vis spectrometer. The optical band gap energy was calculated from the onset of absorption edges by using commercial software (Jasco VWGA589). The XPS measurement was performed on a Shimadzu-Kratos AXIS-HS spectrometer using monochromated Al KR radiation (10 kV, 15 mA). For the measurement, each sample was pressed into ca. 0.1 mm thick pellets and preevacuated under a pressure less than 10-2 Pa for 24 h at ambient temperature. The normal operating pressure in the analysis chamber was controlled less than 10-7 Pa during the measurement. The binding energy (EB) calibration was checked by line position of C 1s as an internal reference. Because a charging effect was observed for every sample, a flood gun with low-energy (ca. 1 eV) electrons from a neutralizer was used to stabilize the spectra. Electronic Structure Calculation. The first-principle calculation approach in the present study is based on the all-electron FLAPW method within the local density approximation (LDA)16 to the density function theory. The calculation of the electronic structure of LaTaO4 was conducted by use of the WIEN97 package,17 the results of which include a fully optimized groundstate structure obtained with total energy and atomic forces, band structure, and densities of states. The crystallographic parameters for the calculation, including lattice parameters and atomic positions, were those reported by Kurova and Akessandrov.18 The parameters of individual atoms were optimized prior to the calculation using the package program. Photocatalytic Reactions. The photocatalytic H2 evolution from water was conducted in an inner irradiation quartz cell, which was connected to a closed gas-circulating system (dead volume: 250 cm3) consisting of a circulation pump, a pressure sensor, gas sampling valves, and stainless steel tubing. A powder sample of the tantalate (0.2 g) was suspended in distilled water (200 cm3) in the cell. Prior to the reaction, the mixture was deaerated by evacuation and then flushed with Ar gas (20 kPa) repeatedly to remove O2 and CO2 dissolved in water. The photocatalytic gas evolution was also measured from aqueous solutions of 4M CH3OH or 0.1M AgNO3 (200 cm3) in an inner irradiation Pyrex cell. The reaction was carried out by irradiating the mixture with light from a 400 W high-pressure Hg lamp. Gas evolution was observed only under photoirradiation, being analyzed by an on-line gas chromatograph (Shimadzu GC8AIT, TCD, Ar carrier, MS-5A and Porapak-Q columns). Any contamination from air was confirmed negligible during at least 50 h of photoreactions. Results and Discussion Crystal Structure. Figure 1 shows the powder X-ray diffraction patterns of the Ln2O3-Ta2O5 system after calcination at 1200 °C for 10 h. Except for a few very weak peaks ascribable to unreacted Ta2O5, all of the diffraction peaks could be
Machida et al.
Figure 1. Powder X-ray diffraction patterns of LnTaO4 (Ln ) La, Ce, Pr, Nd, and Sm).
Figure 2. Crystal structure of LnTaO4 (Ln ) La, Ce, and Pr).
TABLE 1: Interatomic Distance of LaTaO4 La-O (1) La-O (2) La-O (3) La-O (4)
0.249 nm 0.274 nm 0.252 nm 0.247 nm
Ta-O (1) Ta-O (2) Ta-O (3) Ta-O (4)
0.205 nm 0.207 nm 0.201 nm 0.191 nm
attributed to either a monoclinic LaTaO4-type for Ln ) La, Ce, and Pr or a monoclinic LaNbO4-type (fergusonite) for Ln ) Nd and Sm. As is shown in Figure 2, the monoclinic structure with La, Ce, and Pr compounds is constructed by corner-shared TaO6 octahedra forming zigzag strings extending in the direction of the b axis.17 Two adjacent sheets of TaO6 thus formed are held together by their bonding with Ln ions, which are coordinated to six oxygen ions in one sheet and two in the adjacent sheet. Table 1 compares the interatomic distance relevant to the four different oxygen sites in the structure of LaTaO4.18 Note that there are no significant differences for the distance of Ta-O pairs. The fergusonite-type lattice obtained for the Nd and Sm compounds is isostructural with a distorted version of the sheelite-type (CaWO4) structure.19 Because two oxygens of each TaO6 octahedron possess larger Ta-O distances, the structure can also be regarded as arrays of TaO4 tetrahedrons that are not linked together. As can be judged from
Property and Structure of Lanthanide Tantalates the shape of XRD peaks (Figure 1), all of these tantalates possess similar crystallinity. Indeed, their BET surface areas calculated from N2 isotherm at -196 °C were in the same range (ca. 5 m2/g). The Ce valence of as-prepared CeTaO4 was analyzed by measuring Ce 3d XPS spectra. It is reported that the multiple states in this energy region arise from different Ce 4f level occupancies in the final state.20 Six peaks corresponding to three pairs of spin-orbit doublets can be identified in the spectrum of Ce(IV) oxides.21 On the contrary, the Ce 3d spectrum from CeTaO4 in the present study showed only two pairs of doublets, being compatible with Ce ions in the trivalent state. Calculated Electronic Structure. On the basis of the crystal structure described above, the electronic structure of LaTaO4 was studied by the first principle calculation using the FLAPW method, which has been most widely applied for the electronic structure calculation of bulk as well as surface of solids.22-26 Figure 3 shows calculated partial and total DOS of LaTaO4. 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 -5 to 0 eV are mainly composed of the O 2p orbital, which is hybridized with Ta 5d. The partial DOS of four types of oxygen sites distribute similarly in accordance with their similar Ta-O interatomic distances (Table 1). The valence band also contains very small contributions from La d and f orbitals. The conduction band in the energy range from 3.5 to 12 eV mainly consists of La d, La f, and Ta d orbitals. The Ta partial DOS is largely ascribable to 5d orbitals, which are divided into two distribution peaks at 3.5∼7 eV and 8∼12 eV as a result of the crystal field splitting in the octahedral TaO6 environment. The similar double-peak distributions in the four O partial DOS in this energy region support the large hybridization between O 2p and Ta 5d. Above the main weight of the La 5d band, a strong and narrow peak of La 4f and broad distributions of La 5d are observed. The localized character of La 4f levels is appreciable from the comparison with Ta and O partial DOS, distributions of which are quite different from the La partial DOS. We have also tried to apply the FLAPW calculations to the system containing partially occupied Ln 4f shells (Ln ) Ce, Pr, Nd, and Sm). According to the preliminary results, the Ln 4f level shifts to lower energy with increasing 4f electrons as was observed in the valence band XPS spectra as shown below. However, the Ln 4f levels do not split into occupied and unoccupied ones in the local-density approximation. As a consequence, the Fermi level is pinned at the 4f bands and the calculated electronic structures fail to reproduce the insulating behavior. This failure makes it impossible to extract the quantitative information about partially occupied 4f shells. More advanced approximations would be necessary to treat the partially occupied 4f shells properly. Absorption Spectra and Band Gap Energy. Figure 4 shows the UV-vis diffuse reflectance spectra of as-prepared LnTaO4. The absorption spectra of Ln ) Pr, Nd, and Sm are characterized by a set of sharp peaks at >350 nm ascribable to internal transitions in a partly filled 4f shell. Because not only the shape but also the position of these peaks are in agreement with those for Ln(III) complexes27 and simple sesquioxide (Ln2O3), the Ln 4f electron in LnTaO4 has a localized character and the corresponding photoexcitation process should not be responsible for photocatalytic reactions. Beside these absorption peaks, the LnTaO4 compounds, with the exception of Ce, exhibited clearcut absorption edges at around 300 nm. Their positions were slightly different, depending on Ln, i.e., red shift was observed
J. Phys. Chem. B, Vol. 105, No. 16, 2001 3291
Figure 3. Total and partial DOS for LaTaO4 calculated by FLAPW method.
for Pr and much more for Ce. The optical band gap energy (Eg) of LnTaO4 calculated from the position of the absorption onset is compared in Figure 5. As is evident from Figure 3, the Eg value of LaTaO4 corresponds to the gap between the valence band formed by O 2p and the conduction band formed by Ta 5d. In contrast to the La, Nd, and Sm compounds with similar Eg’s, 3.8-3.9 eV, the other two compounds showed decreased Eg’s with the deep minimum on Ce. Such a Ln dependence of Eg was also confirmed in other lanthanide tantalates, Ln3TaO7 (Figure 5) as well as RbLnTa2O7 in our previous study.12,13 Moreover, all of these variations of Eg agreed with that in simple sesquioxides of Ln, being closely associated with the energy levels of 4f shell in the electronic structure of these lanthanide tantalates. Valence Band XPS Spectra. To evaluate the energy levels of partly filled 4f electrons, further studies employed XPS
3292 J. Phys. Chem. B, Vol. 105, No. 16, 2001
Machida et al.
Figure 6. Valence band XPS spectra of LnTaO4 (Ln ) La, Ce, Pr, Nd, and Sm). Figure 4. UV-vis diffuse reflectance spectra of LnTaO4 (Ln ) La, Ce, Pr, Nd, and Sm). Symbols for each absorptions show the excited states of isolated Ln 4f. The ground states are La, 1S0, Pr, 3H4, Nd, 4I , Sm, 6H . 9/2 5/2
Figure 7. Band structure models of LnTaO4 (Ln ) La, Pr, and Nd).
Figure 5. Variation of band gap energy (Eg) of LnTaO4 and Ln3TaO7 (Ln ) La, Ce, Pr, Nd, and Sm).
measurement in the valence band photoemission region as shown in Figure 6. The spectra of LnTaO4 in this region are characterized by a broad and strong peak stretching from 3 to 11 eV. As can be judged from a comparison with Figure 3, the peak corresponds to the highest filled band mainly composed of O 2p orbitals that are hybridized with Ta 5d. CeTaO4 presented another weak peak at EB ) ca. 2 eV, which can be assigned to the emission from the occupied Ce 4f levels lying within the band gap. For the Pr compound, this shoulder peak was shifted toward the valence band edge with increased intensity. Further shift toward higher EB (lower energy level) was continued in the Nd as well as Sm compounds, suggesting the filled 4f levels lying within the O 2p valence band. In Figure 6, the position of highest filled band of each tantalates can be determined from the intercept of a linear fit to the right shoulder of the observed peak profiles with the zero line. The difference of the position thus obtained is found to be consistent with the different Eg in Figure 5, i.e., the variation of Eg can be explained
by the energy of filled Ln 4f levels that become lowered with an increase of 4f electrons. Band Structure Model. With all of these results taken into consideration, the band structure model of LnTaO4 (Ln ) La, Ce, and Pr) can be illustrated schematically as shown in Figure 7. In accordance with the FLAPW calculation of LaTaO4, both the valence band and the conduction band are largely consisted of hybridization between Ta 5d (t2g) and O 2p orbitals. Similar distributions of the conduction band as well as the valence band DOS can be expected for CeTaO4 and PrTaO4 with basically same crystal structure. However, the most noticeable difference should be observed for the position of Ln 4f levels. The highly localized single La 4f band is positioned at ca. 2 eV above the bottom of the conduction band in Figure 3. For Ln with partly occupied 4f orbitals, however, Coulomb correlation should cause not only splitting of the 4f band into a filled band and an empty band but also lowering of their energy level across the series of Ln. For the Ce as well as Pr compounds, therefore, the filled 4f band is located within the band gap. Although the band structure is not known for Nd and Sm with fergusonite-type structure, the filled 4f band becomes further lowered to overlap with O 2p valence band as was observed in the XPS spectra (Figure 6). As soon as the filled 4f band enters into the O 2p valence band, the Eg values in Figure 5 become almost constant.
Property and Structure of Lanthanide Tantalates
J. Phys. Chem. B, Vol. 105, No. 16, 2001 3293
TABLE 2: Photocatalytic Activity of LnTaO4 for Water Splitting ratea/µmol h-1 unloaded
0.7 wt % Ni-loaded
rateb/ µmol h-1
ratec/ µmol h-1
Ln
Eg/eV
H2
O2
H2
O2
H2
O2
La Ce Pr Nd Sm
3.9 2.4 3.6 3.9 3.8
6.9 0.1 0.4 0.5 0.3
2.5 0.0 0.0 0.0 0.0
115.6 0.6 4.7 3.7 9.2
51.5 0.0 0.7 0.0 0.0
5.3 0.0 0.5 0.9 0.1
0.3 0.8 0.5
a
Measured in an inner irradiation quartz cell under irradiation from a 400 W high-pressure Hg lamp: H2O, 200 cm3; catalyst, 0.2 g. b Measured in an inner irradiation Pyrex cell under irradiation from a 400 W high-pressure Hg lamp: 4M CH3OH, 200 cm3; catalyst, 0.2 g. c Measured in an inner irradiation Pyrex cell under irradiation from a 400 W high-pressure Hg lamp: 0.1M AgNO3, 200 cm3; catalyst, 0.2 g.
Figure 8. Effect of Ni loading on the photocatalytic evolution of H2 and O2 over LaTaO4 under irradiation of UV light from a 400 W highpressure Hg lamp. Catalyst, 0.2 g; water, 200 cm3.
Thus, the highest occupied level is formed by O 2p for La, Nd, and Sm or by filled Ln 4f lying above the O 2p bands for Ce and Pr. Photocatalytic Activity for Water Splitting. Table 2 summarizes the activity of as-prepared LnTaO4 for photocatalytic water splitting under UV irradiation. Among as-prepared LnTaO4, the La compound showed the highest rate for the evolution of H2/O2 mixtures with a nearly stoichiometric ratio from pure water. Possible impurity phasessLa2O3, Ta2O5, or Ln3TaO7swith larger band gap energies were also prepared and submitted to the reaction, but all of these compounds were less active. The other lanthanide tantalates from Ce to Sm also showed much less activity and no detectable O2 during 10 h of photoirradiation. We have already reported the Ln-dependent photocatalytic activity of RbLnTa2O7, which increases in the sequence of Pr < La , Sm < Nd.13 Thus, the photocatalytic activity of LnTaO4 depends on Ln in a totally different way. We have also conducted the reaction over LaTaO4 loaded with NiOx that is known to be effective for H2 evolution by photocatalytic water splitting.15 Figure 8 represents the rate of H2 and O2 evolved over the LaTaO4 catalyst as a function of Ni loading. The activity increased with increasing the loading up to 0.7 wt % Ni, at which the rates of H2 and O2 evolutions attained 115.6 and 51.5 µmol/h, respectively, compared to 6.9 and 2.5 µmol/h for unloaded LaTaO4. However, a further increase of Ni loading decreased the activity because considerable UV absorption by NiOx inhibits the generation of carriers in the tantalate. The amount of gas evolved increased linearly with the time progress, and no sign of deactivation was observed during 50 h of the reaction. The cumulative amount of gas evolution during 12 h of photoirradiation exceeded that of
Figure 9. Possible mechanisms of photocatalytic reaction in LaTaO4 and RbLaTa2O7.
LaTaO4 used, proving that the gas evolution proceeded catalytically. Although the increased rate of H2 evolved by loading NiOx was also observed for other tantalates (Table 2), the effect was very small compared to the La compound, and O2 was still not evolved. The different effect of NiOx loading appears to be closely related to the band structure of LnTaO4 as discussed in the following section. Correlation between Photocatalysis and Electronic Structure of LnTaO4. The most important point for the purpose of the present study is to elucidate the reason the observed photocatalytic activity of LnTaO4 is strongly dependent on Ln. For this purpose, a comparison of the observed rate of photocatalytic gas evolution from pure water and aqueous solutions of CH3OH is of interest. The rate of H2 evolved from aqueous solutions of 4 mol/L CH3OH was measured under irradiation of UV-visible light through the Pyrex cell. As shown in Table 2, the rate of H2 evolved was low even in the presence of the sacrificial electron donor (CH3OH). The Pyrex cell decreases the number of photons with energies greater than Eg of LaTaO4, and this is the main reason for the low rate of H2 evolution. In this case, however, it should be noted that LaTaO4 exhibited much higher activity again. The anomalous photocatalytic activity of LaTaO4 in the present system is associated with so high an empty 4f band in the conduction band as shown in Figure 7. When the valence band electrons are photoexcited, carriers generated in the conduction band can migrate through largely hybridized Ta-O bond to the surface to evolve H2 thereon (Figure 9). On the contrary, the extremely low activity of the Ce as well as Pr compounds with lower band gap energies suggests that photoexcited electrons should not be utilized for the reduction of H+ to H2. This is supported by the effect of loading NiOx, which was not effective in improving the photocatalytic gas evolution over these two tantalates. These results can be understood by assuming that electrons are readily trapped by the empty Ln 4f levels, which locate near the bottom of the conduction band for Ln ) Ce and Pr as shown in Figure 7. Owing to the strongly localized character of the Ln 4f levels, the trapped carriers cannot reach the catalyst surface. The same consideration cannot be made for the Nd and Sm compounds, which are not isostructural with LaTaO4. The fergusonite-type structure of these two compounds should form the electronic structure deformed from that of the LaTaO4 (Figure 3). Irrespective of the structural difference, the optical band gap energy (Figure 5) and XPS spectra (Figure 6) may suggest similar behavior of filled 4f level; the position of which becomes lowered with an incremental number of 4f electron and thus overlaps with the O 2p valence band. It also should be pointed out that the fergusonite-type structure appears to be unsuitable for a photocatalyst, because a TaO4 tetrahedron is apart from each other, i.e., neither corners, edges, nor faces are shared, not to allow the photoexcited carriers to migrate.
3294 J. Phys. Chem. B, Vol. 105, No. 16, 2001 Another interaction is suspected for the filled Ln 4f level, which situates very close to the valence band edge for Ln ) Ce and Pr (Figure 7). In these systems, if photoexcited holes generated in the top of the valence band are also trapped by the filled Ln 4f levels, the efficiency of O2 evolution should be drastically decreased. This may imply that Ln plays a role of recombination center for photoexcited electrons and holes. As can be judged from Table 2, such interactions would be a possible reason the La compound can solely produce stoichiometric O2 from pure water under full-arc irradiation. To ascertain the effect of Ln 4f on the O2 evolution, we have evaluated the rate of O2 evolution in the presence of an electron acceptor (Ag+). Irradiation of the light through the Pyrex cell caused photochemical deposition of Ag on the catalyst, whereas the rate of O2 evolved in this case was very small and similar for the three compounds (Ln ) La, Ce, and Pr). The deposition of Ag suggests the generation of photoexcited carriers, and the holes thus produced in the valence band should possess potential energy enough to oxidize water to evolve O2. Thus, the low rate of O2 evolution is associated with the property of catalyst surface. It is known that several oxide photocatalysts cannot evolve O2 from aqueous AgNO3 solution under band-gap irradiation because of the lack of oxygen evolution site on the surface.28,29 Further study is under way to examine the effect of the interaction between holes and filled Ln 4f bands on the photocatalytic activity of LnTaO4. A comparison of the present result with our previous study on RnLnTa2O7 is useful in understanding the Ln-dependent photocatalytic activity of various lanthanide tantalates. The RbLnTa2O7 photocatalysts exhibit Ln-dependent activity, which is completely different from LnTaO4, i.e., RbLaTa2O7 was less active as compared with the Nd and Sm compounds.12,13 In accordance with the electronic structure calculation based on the FLAPW method, the lower activity for La is closely related to calculated band structure schematically shown in Figure 9.14 Unlike LaTaO4, the unoccupied La 4f levels of RbLaTa2O7 are located very close to the bottom of the conduction band and thus may trap photoexcited carriers. With an increase of 4f electrons, however, the 4f levels become lowered more deeply relative to the conduction band edge, being enough not to play as accepting centers. Consequently, the catalytic activities of the Nd as well as Sm compounds are not affected by the empty Ln 4f levels. In addition to this effect, it is also pointed out that the hybridization between Ta 5d, O 2p, and Ln 4f observed in the RbLnTa2O7 system may bring about a positive effect on the photocatalytic activity.14 The result of the present study implies that the Ln dependence of photocatalytic activity is governed by the energy level of Ln 4f levels in the band structure, which is dependent on the chemical composition and crystal structure. The empty 4f levels close to the conduction band edge would trap carriers and thus
Machida et al. decrease the photocatalytic activity for the H2 evolution. The hybridization of Ln 4f with other atomic orbitals also appears to affect the generation and migration of carriers, but a further study is necessary to understand the correlation with the photocatalytic property. Acknowledgment. The present study was financially supported by the Sumitomo Foundation and Grant-in aid for Scientific Research from the Ministry of Education, Science, Sports, and Culture. References and Notes (1) Domen, K.; Naito, S.; Ohnishi, T.; Tamaru, K. Chem. Phys. Lett. 1982, 92, 433. (2) Inoue, Y.; Kubokawa, T.; Sato, K. J. Chem. Soc., Chem. Commun. 1990, 1298. (3) Inoue, Y.; Niiyama, T.; Asai, Y.; Sato, Y. J. Chem. Soc., Chem. Commun. 1992, 579. (4) Uchida, S.; Yamamoto, Y.; Fujishiro, Y.; Watanabe, A.; Ito, O.; Sato, T. J. Chem. Soc., Dalton Trans. 1997, 93, 3229. (5) Kudo, A.; Tanaka, A.; Domen, K.; Maruya, K.; Aika, K.; Ohnishi, T. J. Catal. 1988, 111, 67. (6) Takata, T.; Shinohara, K.; Tanaka, A.; Hara, M.; Kondo, J. N.; Domen, K. J. Photochem. Photobiol. 1997, 106, 45. (7) Kudo, A.; Kato, H. Chem. Lett. 1997, 867. (8) Kudo, A.; Hijii, A. Chem. Lett. 1999, 1103. (9) Ihihara, T.; Takita, T. J. Phys. Chem. B 1999, 103, 1-3. (10) Kim, H. G.; Hwang, D. W.; Kim, J.; Kim, Y. G.; Lee, J. S. Chem. Commun. 1999, 1077. (11) Fujishiro, Y.; Uchida, S.; Sato, T. Int. J. Inorg. Mater. 1999, 1, 67. (12) Machida, M.; Yabunaka, J.; Kijima, T. Chem. Commun. 1999, 1939. (13) Machida, M.; Yabunaka, J.; Kijima, T. Chem. Mater. 2000, 12, 812. (14) Machida, M.; Yabunaka, J.; Kijima, T.; Matsushima, S.; Arai, M. J. Mater. Chem. 2001, in press. (15) Domen, K.; Naito, S.; Soma, M.; Ohnishi, T.; Tamaru, K. J. Chem. Soc. Chem. Commun. 1980, 543. (16) Jones, R. O.; Gunnarsson, O. ReV. Mod. Phys. 1989, 61, 689. (17) Blaha, P.; Schwarz, K.; Luitz, J. WIEN97; Vienna University of Technology: Vienna, Austria, 1997. [Improved and updated UNIX version of the original copyrighted WIEN code, which was published Blaha, P.; Schwarz, K.; Sorantin, P.; Tricky. S. B. Comput. Phys. Commun. 1990, 59, 399. (18) Kurova, T. A.; Akessandrov, V. B. Dokl. Akad. Nauk. SSSR 1971, 201, 1095. (19) Komlov, A. I. Dokl. Akad. Nauk. SSSR 1959, 126, 853. (20) Kotani, A.; Jo, T.; Parlebas, J. C. AdV. Phys. 1988, 37, 37. (21) Mullins, D. R.; Overbury, S. H.; Huntley, D. R. Surf. Sci. 1998, 409, 307. (22) Krakauer; H.; Posternak, MA.; Freeman, A. J. Phys. ReV. 1979, B19, 1706. (23) Wimmer, E.; Krakauer, H.; Weinert, M.; Freeman, A. J. Phys. ReV. 1981, B24, 864. (24) Weinert, M.; Wimmer, E.; Freeman, A. J. Phys. ReV. 1982, B26, 4571. (25) Mattheiss, L. F.; Hamann, D. R. Phys. ReV. 1983, B28, 4227. (26) Asahi, R.; Taga, Y.; Mannstadt, W.; Freeman, A. J. Phys. ReV. B 2000, 61, 7459. (27) Ryan, J. L.; Jorgensen, C. K. J. Phys. Chem. 1966, 70, 2845. (28) Kudo, A.; Hijii, A. Chem. Lett. 1997, 421. (29) Bamwenda, G. R.; Arakawa, H. J. Mol. Catal. A 2000, 161, 105.
