Dispersion of ruthenium oxide on barium titanates (Ba6Ti17O40 , and and photocatalytic activity Ba4Ti13O30, BaTi4O9 Ba2Ti9O20) for water decomposition Mitsuru Kohno, Takatoshi Kaneko, Shuji Ogura, Kazunori Sato and Yasunobu Inoue* Department of Chemistry, Nagaoka University of T echnology, Nagaoka, Niigata 940-21, Japan Ruthenium oxide supported on barium titanates and was employed as a photo- (Ba6Ti17O40, Ba4Ti13O30 , BaTi4O9 Ba2Ti9O20) catalyst for water decomposition. The titanates were subjected to either reduction or reduction»oxidation. RuCl3-impregnated High-resolution electron microscopic images demonstrated that ruthenium metal and ruthenium oxides were uniformly dispersed on with an average particle size of 2.6 nm.Similar uniform ruthenium oxide dispersions were observed for the other BaTi4O9 barium titanates ; the average particle sizes were 4.7 nm for 2.3 nm for and 4.4 nm for Ba6Ti17O40, Ba4 Ti13O30, Ba2 Ti9O20 . Particle size distributions were narrower for and and slightly larger for and BaTi4O9 Ba4Ti13O30, Ba6 Ti17O40 Ba2Ti9O20 .Stoichiometric production of oxygen and hydrogen occurred for a photocatalyst. A small amount of hydrogen RuO2/BaTi4O9 and no oxygen were produced from the other barium titanates and combined with ruthe- (Ba6Ti17O40, Ba4Ti13O30 Ba2Ti9O20) nium oxides. EPR spectra at 77 K in He or with UV irradiation demonstrated that a strong signal, assigned to a surface O~ O2 radical, appeared for but not for the other barium titanates. These produced small complicated signals, indicating that BaTi4O9 only has a high efficiency for photoexcited charge formation.Raman spectra showed that a strong single peak at a high BaTi4O9 wavenumber of 860 cm~1, characteristic of was absent in the rest of the barium titanates. The diÜerent photocatalytic BaTi4O9 , properties among these titanates are discussed on the basis of structure diÜerences of the barium titanates, and the presence of internal –elds ; a long TiwO bond in the distorted octahedra is proposed to be important in photocatalysis.TiO6 Introduction We have recently shown that the combination of ruthenium oxide and makes a good photocatalyst which is BaTi4O9 able to decompose water to oxygen and hydrogen in stoichiometric ratio.1,2 Barium titanates are represented by where n\2»4.3h6 In view of the corre- Ba2(n~1)Ti4n`1O10n , lation between the photocatalytic activity and oxide structure, it is desirable to extend research to photocatalysts using a series of barium titanates combined with ruthenium oxide which works as a promoter.In particular, it is of importance to see whether is the only barium titanate that BaTi4O9 becomes a good photocatalyst by combination with ruthenium oxide and, if so, to clarify why RuO2-deposited BaTi4O9 is speci–c as a photocatalytic titanate. In this work, in addition to and BaTi4O9, Ba6 Ti17O40, Ba4 Ti13O30 Ba2Ti9O20 were chosen as representative barium titanates.has BaTi4O9 orthorombic symmetry with a unit cell of a\14.53, b\3.79, c\6.29 has monoclinic symmetry with a Aé 3.7 Ba6Ti17O40 unit cell of a\9.88, b\17.08, c\18.92 b\98.7°.3 Aé 3, has orthorombic symmetry (a\17.06, b\9.86, Ba4Ti13O30 c\14.05 and triclinic symmetry (a\14.36, Aé 3),4 Ba2Ti9O20 b\14.10, c\7.48 a\95.53, b\98.7, c\89.95°).5 Aé 3, has a pentagonal prism tunnel structure, and BaTi4O9 a hollandite-like tunnel structure, whereas Ba2Ti9O20 and have close-packed arrays of Ba6Ti17O40 Ba4Ti13O30 oxygen and barium atoms in which some of the octahedral voids are –lled by titanium atoms.3h8 In this study, the photocatalytic properties for water decomposition of these barium titanates with supported ruthenium oxides were examined. The photocatalytic activity is controlled by two factors regarding the efficiency of photoexcited charge formation and of charge transfer to the surface reactants.In the development of efficient photocatalysts, both factors have to be taken into account.In a photocatalytic system of ruthenium oxidesupporting barium titanates, the titanates absorb light and produce photoexcited electrons and holes, thus it is essential to compare the ability of photoexcited charge formation among the titanates. In a previous study on the photocatalyst, we demonstrated that the EPR RuO2/BaTi4O9 signals of under UV-irradiation in diÜerent gas BaTi4O9 atmospheres were a good measure of the photoexcitation ability.9h12 Thus, this method was employed here.For the establishment of correlations between the photoexcitation ability and titanate structures, the structural features of the titanates were investigated by laser Raman spectroscopy. As for the charge transfer, no photocatalytic activity was observed in the absence of ruthenium oxide in the photocatalytic system using Ruthenium oxide thus plays BaTi4O9 .1 an important role in the transfer of the photoexcited charges to the surface reactants.Thus, in comparison of photocatalytic activities among the titanates, it is desirable to prepare photocatalysts with similar ruthenium oxide distributions. In the previous study, direct oxidation after impregnation was RuCl3 employed to prepare ruthenium oxide.1,2 This study involves reduction and reduction»oxidation of RuCl3-impregnated barium titanates in an attempt to obtain well dispersed states of ruthenium. The dispersion was investigated by highresolution transmission electron microscopy (HRTEM) combined with elemental analysis by EDS.From comparison with the photocatalytic activity for water decomposition, the ability to produce photoexcited charges, the extent of ruthenium oxide dispersion, and the structural features, a model for photocatalysis by RuO2-deposited barium titanates is proposed. Experimental The barium titanates were prepared by calcining a stoichiometric mixture of (high-purity grade, Soekawa Chemical TiO2 Co.) and (high-purity grade, Soekawa Chemical Co.) BaCO3 in air at 1323 K and 20 h for 1523 K and 10 h for BaTi4O9 , 1423 K and 10 h for and 1523 K Ba6Ti17O40, Ba4 Ti13O30 , J.Chem. Soc., Faraday T rans., 1998, 94(1), 89»94 89and 10 h for The structures of the titanates were Ba2Ti9O20 . investigated by powder X-ray diÜraction using a Rigaku Denki RAD III diÜractometer. Their morphology was examined by observations with a JEOL JXA-733 scanning electron microscope.The surface area of titanates was measured by the BET method using nitrogen at 77 K. Photocatalysts were prepared by impregnation of the barium titanates with aqueous solutions at 353 K to RuCl3 incipient wetness and then dried in air at 353 K. The loading of ruthenium was 0.5 wt.% metal content. The RuCl3- barium titanates were subjected to either impregnated reduction in an —ow at 723 K for 2 h or reduction fol- H2 lowed by oxidation in air. The reduction temperature was maintained at 723 K, whereas the oxidation temperature was changed from 373 to 773 K.For simplicity, reduction and oxidation are denoted by R and O and only the oxidation temperature (in K) is given : e.g. a photocatalyst prepared by reduction and subsequent 573 K oxidation of impregnated is represented by The dis- BaTi4O9 RuO2(R,O573)/BaTi4O9 . persion of ruthenium oxides and the photocatalytic activity were compared to those prepared by direct oxidation of impregnated at 773 K; this is denoted as BaTi4O9 RuO2(O773)/BaTi4O9 .The photocatalytic water decomposition was carried out in a gas-circulation system described elsewhere.1,7 Brie—y, 0.2 g of powder photocatalysts were dispersed in 20 cm3 of pure water in a quartz cell, stirred by bubbling with Ar gas circulation, and irradiated with a 400 W Xe lamp through a water –lter. The UV —ux (330»390 nm) was ca. 12 mW cm~2. Hydrogen and oxygen produced were analyzed by a gas chromatograph connected to the reaction system.High-resolution electron microscopic images of barium titanate-supported ruthenium and ruthenium oxides were obtained with a JEOL 2010 transmission electron microscope (TEM) operated at an accelerating voltage of 200 kV. The sample preparation for TEM observation was made by dispersing the powder sample ultrasonically in methanol and putting the resultant suspension onto a holey carbon coated –lm supported by a Cu grid. The energy-dispersive X-ray (EDS) spectra were collected for nanometre-size areas of the samples with a Voyager energy dispersive analyzer (Noran Instruments) installed on the microscope.The beam diameters for the microanalysis of the samples were varied in the range 2»5 nm. EPR spectra were recorded on a JEOL JES-RE2X spectrometer at a microwave power of 0.1 mW, a microwave frequency of ca. 9.1 GHz and a modulation width of 1 G. In a typical run, 350^1 mg of sample was placed in an X-band quartz cell and degassed at 573 K.For adsorption experiments, 30 Torr of He or were introduced at room tem- O2 perature. EPR measurements were performed at 77 K, unless otherwise speci–ed, without or with UV irradiation of the samples with a 500 W high-pressure mercury lamp. The g values were calibrated by Mn2` in MgO, the error of which was within ^0.001. Results All of the XRD patterns of the barium titanates were in good agreement with those reported in the XRD data –les.13 The particle sizes of these titanates ranged from 1 to 5 lm, and there were no signi–cant diÜerences.The speci–c surface area of the barium titanates was 1 m2 g~1 for both Ba6Ti17O40 and 2.3 for and 1.2 for BaTi4O9, Ba4 Ti13O30 Ba2Ti9O20 . Fig. 1 shows the laser Raman spectra of the barium titanates. As reported previously, had clearly distin- BaTi4O9 guishable three main bands at 430»450, 590»650 and 860 cm~1. The characteristic peak was a strong single peak appearing at a higher wavenumber than 860 cm~1 since it re—ects a short TiwO bond.9,10 In contrast, Ba6Ti17O40 , Fig. 1 Raman spectra of (a) (b) (c) Ba6Ti17O40, Ba4 Ti13O30 , and (d) BaTi4O9 Ba2Ti9O20 and exhibited complicated spectra Ba4Ti13O30 Ba2Ti9O20 consisting of many small peaks. Although they also showed peaks at 875, 858 and 846 cm~1 , respectively, the relative intensities of these peaks were considerably lower. Fig. 2 shows the UV diÜuse re—ectance spectra of these titanates.The absorption had a threshold wavelength at around 420 nm for and and shifted to around BaTi4O9 Ba2Ti9O20 , 400 nm for and The maximum Ba6Ti17O40 Ba4Ti13O30 . absorption shifted to shorter wavelength in the order of (340 (335 Ba6Ti17O40 nm)[Ba2Ti9O20 nm)[Ba4Ti13O30 (330 (320 nm). A shoulder was observed at nm)[BaTi4O9 around 380 nm for and BaTi4O9 Ba2Ti9O20 . Fig. 2 UV diÜuse re—ectance spectra of (-------), Ba6Ti17O40 (» » » »), (»»») and Ba4Ti13O30 BaTi4O9 Ba2Ti9O20(»-»-) 90 J.Chem. Soc., Faraday T rans., 1998, V ol. 94Fig. 3 shows an HRTEM image of an photocatalyst. Spherical dark spots, RuO2(R,O573)/BaTi4O9 2»5 nm in diameter, were uniformly dispersed on the regular lattice image. These dark spheres were observed over BaTi4O9 the entire surface. EDS analysis was performed for BaTi4O9 the two regions represented by circles (a) and (b) on the TEM image (Fig. 4). For region (a), two peaks observed at 4.5 and 4.9 keV were assigned to Ba(Ka and Lb) and Ti(Ka and Kb), respectively.For region (b), in which a dark spherical spot was present, an additional peak appeared at 2.6 keV which was due to Ru La. Thus it was obvious that the dark spherical spots were composed of ruthenium. Fig. 5 compares the HRTEM images of Ru/BaTi4O9 , RuO2(R,O773)/BaTi4O9 and For black spots RuO2(O773)/BaTi4O9 . Ru/BaTi4O9 , ranging from 0.8 to 3.0 nm were observed, which were due to metallic Ru. For and RuO2(R,O773)/BaTi4O9 spots with similar particle sizes and RuO2(O773)/BaTi4O9 , distributions were observed. Note that these particle sizes and distributions are analogous to those of (cf.Fig. 3). RuO2(R,O573)/BaTi4O9 Fig. 6»8 show the HRTEM images of sup- RuO2(R,O573) ported on and respec- Ba6Ti17O40, Ba4Ti13O30 Ba2Ti9O20 , tively. In each image, spherical dark spots due to ruthenium were clearly seen on the regular lattice images of the barium titanates. Fig. 9 shows particle size distributions. Ba6Ti17O40 and had symmetrical distributions having centred Ba2Ti9O20 around 4»5 nm in which the distribution was broader for than for For and Ba2Ti9O20 Ba6Ti17O40.Ba4 Ti13O30 smaller particles were present with a higher density, BaTi4O9 , and the maximum of distributions shifted to particle sizes as small as 2»3 nm. The average diameter of the particles increased in the order of (2.3 (2.6 Ba4Ti13O30 nm)\BaTi4O9 (4.4 (4.7 nm). nm)\Ba2Ti9O20 nm)\Ba6Ti17O40 Fig. 10 compares the photocatalytic activities for water decomposition of as-impregnated BaTi4O9 , Ru/BaTi4O9 , Fig. 3 HRTEM image of Circles (a) and RuO2(R,O573)/BaTi4O9 . (b) correspond to areas used for EDS analysis. Fig. 4 EDS spectra of regions (a) and (b) on BaTi4O9 Fig. 5 HRTEM images of (a) (b) Ru/BaTi4O9 , and (c) RuO2(R,O773)/BaTi4O9 RuO2(O773)/BaTi4O9 and in which RuO2(O773)/BaTi4O9 RuO2(R,O)/BaTi4O9 , oxidation was performed at 373, 573 or 773 K. The production of was extremely small for as-impregnated H2 and slightly increased with For BaTi4O9 , Ru/BaTi4O9 .a considerable increase in the activ- RuO2(R,O373)/BaTi4O9 , ity of hydrogen production was observed, but oxygen was poorly produced. A remarkable enhancement of photocatalytic activity occurred for and RuO2(R,O573)/BaTi4O9 the activities of both hydrogen and oxygen production were Fig. 6 HRTEM image of RuO2(R,O573)/Ba6Ti17O40 J. Chem. Soc., Faraday T rans., 1998, V ol. 94 91Fig. 7 HRTEM image of RuO2(R,O573)/Ba4Ti13O30 Fig. 8 HRTEM image of RuO2(R,O573)/Ba2Ti9O20 Fig. 9 Distributions of particles deposited on (a) RuO2 Ba6Ti17O40 , (b) (c) and (d) Ba4Ti13O30 , BaTi4O9 Ba2Ti9O20 Fig. 10 Photocatalytic activity for water decomposition of BaTi4O9- (a), Ru(R723) (b), (c), supported RuCl3 RuO2(R,O373) RuO2(R,O573) (d), (e) and ( f ). 4, RuO2(R,O773) RuO2(O773) 34, H2, O2 . Fig. 11 Photocatalytic activity for water decomposition on titanates of (a) (b) RuO2(R,O573)/barium Ba6Ti17O40, Ba4 Ti13O30 , (c) and (d) 4, BaTi4O9 Ba2Ti9O20 . 34: H2; O2 . 167 and 78 lmol g~1 h~1, respectively, in which the stoichiometric hydrogen-to-oxygen ratio was nearly achieved. For the activity was lowered by a factor RuO2(R,O773)/BaTi4O9 , of ca. 30%, but was ca. twice as large as that of Note that the activity of RuO2(O773)/BaTi4O9 . was similar to that of 0.5 wt.% RuO2(O773)/BaTi4O9 which was obtained in the previous RuO2(O848)/BaTi4O9 study.1 These results indicate that reduction»oxidation is useful for catalyst activation, compared to direct oxidation.Fig. 11 shows the activities of a photocatalyst system of and RuO2(R,O573)/ (Ba6Ti17O40, Ba4 Ti13O30 Ba2Ti9O20), together with the activity of for a RuO2(R,O573)/BaTi4O9 comparison. There was no evolution of oxygen, and the rate of production was extremely small : 1.0 lmol g~1 h~1 for H2 7.0 lmol g~1 h~1 for RuO2(R,O573)/Ba6Ti17O40 , and 1.5 lmol g~1 h~1 for RuO2(R,O573)/Ba4Ti13O30 Note that there were quite large RuO2(R,O573)/Ba2Ti9O20 .diÜerences in activity between photocatalysts using BaTi4O9 and the other titanates. Fig. 12 shows the EPR spectra of the barium titanates at 77 K in 30 Torr of He with UV irradiation. Two signals (g\2.034 and 1.981) observed at both ends of each spectrum are due to Mn2` in MgO which is used for the calibration of g values. provided a strong signal with g\2.018 BaTi4O9 and 2.004, in agreement with previously reported values.9h12 On the other hand, and Ba6Ti17O40, Ba4Ti13O30 Ba2Ti9O20 showed no such characteristic strong signals with irradiation.Instead, produced small signals at g\2.021 and Ba6Ti17O40 2.002. Complicated small signals appeared at g\2.022 , 2.018 and 2.003 for and broad signals for Ba4Ti13O30 Ba2Ti9O20 . In order to observe the reactivity of the EPR-active species, signal changes with heat treatment in an oxygen atmosphere were examined. The barium titanates were irradiated at 77 K in 30 Torr of instead of He, and the sample temperature O2 was raised to room temperature in the presence of gaseous oxygen.EPR spectra were measured at 77 K without UV irradiation. showed characteristic large signals at BaTi4O9 g\2.010 and 2.003 and a very small signal at g\2.018, consistent with those reported previously.11 However, the small signals observed in disappeared completely, and Ba6Ti17O40 the signals in and diminished, as Ba4Ti13O30 Ba2Ti9O20 shown in Fig. 13. Discussion The HRTEM observations demonstrated that ruthenium metal and ruthenium oxides were uniformly dispersed as small particles on the surface.The interesting feature is BaTi4O9 that Ru metal and Ru oxides had similar particle sizes and distributions, as shown in Fig. 5; furthermore, nearly the same particle sizes of were obtained in the oxidation of RuO2 at temperatures ranging from 373 to 773 K, and Ru/BaTi4O9 92 J. Chem. Soc., Faraday T rans., 1998, V ol. 94Fig. 12 EPR spectra of (a) (b) (c) Ba6Ti17O40, Ba4 Ti13O30 , and (d) BaTi4O9 Ba2Ti9O20 Fig. 13 EPR spectra of (a) (b) (c) Ba6Ti17O40, Ba4 Ti13O30 , and (d) These titanates were UV irradiated at 77 BaTi4O9 Ba2Ti9O20 . K in heated to room temperature without evacuation, and then O2 , cooled to 77 K. All the spectra were measured at 77 K without UV irradiation. even in the direct oxidation at 773 K. A comparison with the results on and RuO2(O773)/BaTi4O9 RuO2(O848)/BaTi4O9 which was obtained by direct oxidation of the impregnated at 848 K,1 also showed that there were no signi–cant BaTi4O9 diÜerences in particle sizes and distributions of between RuO2 reduction»oxidation and direct oxidation.On the other hand, the photocatalytic activity for water decomposition was strongly dependent on the oxidation temperature of ruthenium and on the presence or absence of reduction prior to oxidation. Similar dispersion but diÜerent photo- RuO2 catalytic activity indicates that the density of RuO2-based active sites is nearly the same, but their efficiency is higher for produced by reduction»oxidation than by direct oxida- RuO2 tion without reduction.One of the reasons for this is possibly that reduction»oxidation permits strong interfacial contact of small particles with surfaces, which facilitates RuO2 BaTi4O9 the transfer of the photoexcited charges to particles. It RuO2 is plausible that the transformation of metallic to oxide phases brings about strong interfacial interactions between RuO2 particles and the oxide surfaces.The appearance of a maximum in the activity dependence on oxidation temperatures (Fig. 10) means that there are optimum states in for charge transfer. The possibility is not ruled out that RuO2 reduction produces ruthenium metal particles and then, the following oxidation leads to the formation of a ruthenium oxide shell covering the metal particles. Domen et al. proposed a structure of NiO-covered Ni as a promoter in a photocatalytic system for water decomposition of NiO deposited on SrTiO3 .14 The photocatalysts except for had very RuO2/BaTi4O9 poor performance with respect to photocatalytic activity for water decomposition, i.e.no oxygen evolution and a very small amount of hydrogen production. As for the lack of oxygen in the gas phase, it is likely that the oxygen is retained on the surface or interior of the barium titanates, as observed in Since the SEM observation and the measure- Pt/TiO2 .15 ments of surface area showed that there were no signi–cant diÜerences in macromorphology and in speci–c surface area of the titanates, it follows that large diÜerences in photocatalytic activity between and the rest of the titanates are BaTi4O9 explained in terms of diÜerences in either particle sizes and distributions of or microstructure-related photo- RuO2 chemical properties of the titanates.TEM observations showed that the particles were RuO2 composed of analogous spherical shapes, irrespective of the type of barium titanates, although showed nar- Ba4Ti13O30 rower distributions, similar to that of whereas BaTi4O9 , and exhibited broader distributions : Ba6Ti17O40 Ba2Ti9O20 the average particle size varied in the order (2.3 Ba4Ti13O30 (2.6 (4.4 nm)\BaTi4O9 nm)\Ba2Ti9O20 nm)\Ba6Ti17O40 (4.7 nm).The surface area of was calculated from the RuO2 size by assuming round shapes: the ratio was 1.1 : 1 : 0.6 : 0.5, respectively. This ratio was too small to explain the diÜerence of photocatalytic activity shown in Fig. 11. Moreover, the abovementioned order of the particle sizes was not in agreement with the order of photocatalytic activity. Thus, these results indicate that the particle sizes and dispersions of RuO2 are not a key factor in distinguishing the photocatalytic activity between and the three other barium titanate- BaTi4O9 based photocatalysts. As shown in Fig. 12 and 13, a strong EPR signal with g\2.018 and 2.004 appeared only for whereas the BaTi4O9 , signal was not observed for the other barium titanates and except for very Ba6Ti17O40, Ba4 Ti13O30 Ba2Ti9O20 small complicated signals. Furthermore, the characteristic signal was transformed to a new signal with g\2.018, 2.010 and 2.003 upon exposure to an atmosphere at room tem- O2 perature. On the other hand, the very small complicated peaks for the other barium titanates disappeared under similar conditions.These EPR experiments clearly showed that the radical production with UV irradiation was quite diÜerent between and the other titanates.As shown in the BaTi4O9 previous study, the signal with g\2.018 and 2.004 was assigned to a surface radical, O~,11,12,16,17 derived from lattice oxygen O2~,18 and the new signal with g\2.018, 2.010 and 2.003 was associated with the radical.11,19 Note that O3~ the formation of the O~ radical was not due to bulk phenomena, but occurred only at the surface region of as BaTi4O9 , is evident from the fact that the radical completely changed to J.Chem. Soc., Faraday T rans., 1998, V ol. 94 93Table 1 Percentage of TiwO bonds in the unit cell bond length \1.8 ” P2.3 ” barium titanate (%) (%) Ba6Ti17O40 2.0 2.0 Ba4Ti13O30 5.1 0.0 BaTi4O9 8.3 8.3 Ba2Ti9920 4.6 0.9 a new radical species by the reaction with gaseous These O2 . radicals undoubtedly appeared as a consequence of the photoexcited electron and hole formation, which is indicative of the high ability of to produce photoexcited charges.BaTi4O9 Thus the high photocatalytic activity of is RuO2/BaTi4O9 strongly associated with the higher efficiency of the surface radical production of which is intrinsically diÜerent BaTi4O9 , from the other barium titanates employed here. possesses a pentagonal prism tunnel structure BaTi4O9 consisting of two kinds of strongly distorted octahedra. TiO6 One octahedron has displacement of a Ti ion by 0.030 TiO6 nm from the center of gravity of six surrounding oxygen ions and the other by 0.021 nm.7 The displacement leads to the presence of large dipole moments (5.7 and 4.1 D§).It has been previously proposed that the internal –elds due to the dipole moments were responsible for the higher efficiency of separation of the photoexcited charges.12 Furthermore, from the fact that the surface O~ radical was very stable in the presence of gas molecules, it has been suggested that TiwO bond breaking occurred by charge transfer with UV irradiation.11 This breaking is considered to be strongly related to the presence of the long TiwO bond in the distorted octahedra.TiO6 Crystallographic data are available for the barium titanates, 3h5,7 and Table 1 compares the proportion of TiwO bonds with short (O0.18 nm) and long (P0.23 nm) bonds in the unit cell of the titanates. In the Raman spectra of barium titanates, the sharp peak observed at 860 cm~1 re—ects the presence of short TiwO bonds.The most interesting point is that has percentages of as large as 8.3% for the pres- BaTi4O9 ence of both short (O0.18 nm) and long (P0.23 nm) TiwO bonds, whereas the other barium titanates have low percentages, especially for long TiwO bonds. Since the O~ radical is con–ned to the surface region,11 its production requires higher percentages of long TiwO bonds. This accounts for the absence of the surface O~ radical in barium titanates apart from BaTi4O9 .From these considerations, it is con–rmed that the internal –elds due to the dipole moments of promote the BaTi4O9 charge transfer, and TiwO bond breaking stabilizes the § 1DB3.335 64]10~30 C m. surface O~ radical produced. Therefore, in comparison with the other barium titanates, the unique photocatalytic properties of are that it is composed of distorted with BaTi4O9 TiO6 short and long TiwO bonds. In relation to photocatalysis, the stable surface O~ radical works as a hole center to oxidize OH~, whereas the transferred electrons, which could be delocalized at the surface, move to ruthenium oxides and then reduce H` to hydrogen.In the present work, it has been revealed that in a photocatalytic system of ruthenium oxide-deposited barium titanates and (Ba6Ti17O40, Ba4 Ti13O30 , BaTi4O9 Ba2Ti9O20), only is active for water decomposition. The RuO2/BaTi4O9 particle sizes and distributions of are similar among the RuO2 titanates, and the photocatalytic diÜerences are attributed to the diÜerent structures of the titanates, in which important factors are the presence of internal –elds due to dipole moments for promoting photoexcited charge transfer, and long TiwO bonds leading to the formation of stable surface O~ radicals.work was supported by a Grant-in-Aid for Scienti–c This Research (B)(07555249) from The Ministry of Education, Science, Sports, and Culture of Japan. References 1 Y. Inoue, Y. Asai and K. Sato, J. Chem. Soc., Faraday T rans., 1994, 90, 797. 2 Y. Inoue, T. Niiyama, Y. Asai and K. Sato, J. Chem. Soc., Chem. Commun., 1992, 579. 3 E. Tillmanns and W. H. Baur, Acta Crystallogr., Sect. B, 1970, 26, 1645. 4 E. Tillmanns, Inorg. Nucl. Chem. L ett., 1971, 7, 1169. 5 G. D. Fallon and B. M. Gatehouse, J. Solid State Chem., 1983, 49, 59. 6 E. Tillmanns, W. Hofmeister and W. H. Baur, J. Solid State Chem., 1985, 58, 14. 7 D. H. Templeton and C. H. Dauben, J. Chem. Phys., 1960, 32, 1515. 8 W. Hofmeister, E. Tillmanns and W. H. Baur, Acta Crystallogr., Sect. C, 1984, 40, 1510. 9 M. Kohno, S. Ogura, K. Sato and Y. Inoue, Stud. Surf. Sci. Catal. A, 1996, 101, 143. 10 M. Kohno, S. Ogura and Y. Inoue, J. Mater. Chem., 1996, 6, 1921. 11 M. Kohno, S. Ogura, K. Sato and Y. Inoue, Chem. Phys. L ett., 1997, 267, 72. 12 M. Kohno, S. Ogura, K. Sato and Y. Inoue, J. Chem. Soc., Faraday T rans., 1997, 93, 2433. 13 JCPDS, No. 34-70, No. 35-51, No. 35-750 and No. 35-817. 14 K. Domen, A. Kudo, T. Onishi, N. Kosugi and H. Kuroda, J. Phys. Chem., 1986, 90, 292. 15 K. Yamaguti and S. Sato, J. Phys. Chem., 1985, 89, 5510. 16 N. B. Wong, T. B. Taarit and J. H. Lunsford, J. Chem. Phys., 1974, 60, 2148. 17 V. A. Shvets and V. B. Kazansky, J. Catal., 1972, 25, 123. 18 M. Kohno, S. Ogura, K. Sato and Y. Inoue, in preparation. 19 N. B. Wong and J. H. Lunsford, J. Chem. Phys., 1972, 56, 2664. Paper 7/04947A; Received 10th July, 1997 94 J. Chem. Soc., Faraday T rans., 1998, V ol. 94