Photocatalytic Hydrogen Production from Water over M-Doped

J. Phys. Chem. B 2005, 109, 2093-2102

2093

Photocatalytic Hydrogen Production from Water over M-Doped La2Ti2O7 (M ) Cr, Fe) under Visible Light Irradiation (λ > 420 nm)† Dong Won Hwang,‡ Hyun Gyu Kim,‡ Jae Sung Lee,*,‡ Jindo Kim,§ Wei Li,§ and Se Hyuk Oh§ Department of Chemical Engineering, Pohang UniVersity of Science and Technology (POSTECH), San 31 Hyoja-dong, Pohang 790-784, Republic of Korea, and General Motors R&D Center, Warren, Michigan 48090 ReceiVed: February 14, 2004; In Final Form: May 19, 2004

In the search for efficient photocatalysts working under visible light, we have investigated the effect of cation substitution on a layered perovskite, La2Ti2O7. Among various metal dopants, only Cr and Fe induced intense absorption of visible light (λ > 400 nm), and only these catalysts produced H2 photocatalytically from water in the presence of methanol under visible light irradiation (λ > 420 nm). The polymerized complex method was found to be more efficient for fabrication of the present catalysts producing a more homogeneous structure than the solid-state reaction. The characterization by XRD, UV-vis DRS, XPS, and XANES revealed that doped Cr and Fe were present in the Cr3+ and Fe3+ states substituting for Ti sites in the La2Ti2O7 lattice. The theoretical calculation indicated that the most significant feature in the electronic band structure of the metaldoped La2Ti2O7 was the formation of a partially filled 3d band in the band gap of La2Ti2O7, while the contribution of these dopants on the valence band was negligible. Excitation of electrons from this localized interband to the conduction band of La2Ti2O7 was responsible for visible light absorption and the H2 evolution from water under visible light.

Introduction Much attention has been paid to photocatalytic water splitting, since H2 with its high-energy capacity and environmental friendliness could be obtained from renewable water and solar light from the process. Although there has been remarkable progress in recent decades for photocatalysts working under ultraviolet (UV) light,1-4 this progress has rarely extended to the visible light region. Chalcogenides such as CdS and CdSe have been studied extensively, since they have ideal edge positions of the valence and conduction bands for the oxidation and reduction of water molecules, respectively.5,6 Unfortunately, they also have a fatal disadvantage of photocorrosion, i.e., CdS itself is oxidized into Cd2+ and S2- by the hole produced in its valence band. In a recent review, several new photocatalyst materials for overall water splitting have been described.7 A layered perovskite, La2Ti2O7, has been studied extensively for photocatalytic water splitting and decomposition of environmentally toxic compounds such as CH3Cl. Its high photocatalytic activity was ascribed to the peculiar electronic structure, i.e., the hypervalency of the La atom constructing the layered perovskite structure.8-10 However, the band gap energy of La2Ti2O7 was ca. 3.8 eV, which made it impossible to absorb any visible light (> 400 nm). Although transition metal oxides such as Cr2O3, Fe2O3, CuO, and PbO alone could not produce hydrogen from the photocatalytic reduction of water because their conduction band edge positions are less negative than the reduction potential of water, these compounds have the advantage of absorbing visible light, since their band gap energy is smaller than 3.0 eV (> 400 nm). Therefore, if we could †

Part of the special issue “Michel Boudart Festschrift”. * Corresponding author. E-mail: [email protected]. Phone: 82-542792266. Fax: 82-54-2795528. ‡ Pohang University of Science and Technology. § General Motors R&D Center.

incorporate these transition metals that can absorb visible light into a semiconductor with its conduction band level more negative than the reduction potential of H2O, H2 might be obtained under visible light irradiation by the electron excited to the conduction band of the host semiconductor. This concept has been demonstrated, for example, for Cr-doped TiO2.11 The absorption in the visible light range for Cr-doped TiO2 could be attributed to Cr3+ to Ti4+ charge-transfer excitation, the excitation of an electron of Cr3+ into the conduction band of TiO2, and the band gap absorption of TiO2 is not affected by Cr doping. In this case, H2 could be produced only in the presence of a hole scavenger such as methanol because the energy level of Cr is less positive than the water oxidation potential, and thus the photogenerated hole in the Cr interband could not oxidize OH-. Recently, Kudo et al. reported that SrTiO3 codoped with antimony and chromium could be used for H2 production from an aqueous methanol solution.12 For Cr-SrTiO3, the H2 evolution rate was higher compared with that of Cr-TiO2, since the conduction band level of SrTiO3 is slightly more negative than that of TiO2. Ru-doped TiO2 was also used as a photocatalyst for oxygen evolution, which occurred by irradiation of visible light at wavelengths longer than 440 nm using iron(III) ions as the electron acceptor.13 To develop photocatalytic materials that could split water into H2 and O2 under visible light irradiation, we have studied cationsubstituted photocatalysts based on La2Ti2O7. This report describes the details of the synthesis of Fe- or Cr-doped La2Ti2O7 prepared by both the polymerized complex method and the solid-state reaction method. Despite numerous previous studies on metal-doped photocatalysts, the structure and role of dopants have not been well understood. Hence, our catalysts will be studied using various characterization tools such as X-ray diffraction (XRD), UV-vis diffuse reflectance spectroscopy (UV-vis DRS), X-ray photoelectron spectroscopy (XPS), and

10.1021/jp0493226 CCC: $30.25 © 2005 American Chemical Society Published on Web 06/17/2004

2094 J. Phys. Chem. B, Vol. 109, No. 6, 2005

Figure 1. Flowchart for synthesis of M-doped La2Ti2O7 (M ) Cr, Fe).

X-ray absorption near-edge structure (XANES) together with a theoretical calculation of the electronic band structure based on the FLAPW method. Experimental Section Material Synthesis. Cr- or Fe-doped La2Ti2O7 was synthesized by the citrate complex method as outlined in Figure 1. An amount of 0.01 mol of titanium tetraisopropoxide (Ti[OCH(CH3)2]4) was first dissolved into 0.4 mol of ethylene glycol [EG: (OHCH2CH2OH)], and subsequently 0.1 mol of citric acid [CA: HOOCCH2C(OH)(COOH)CH2COOH] was added to this solution. After achieving complete dissolution, 0.01 mol of La(NO3)3‚6H2O and the required amount (Cr/La ) 0.01 to ∼0.2) of Cr(NO3)3‚6H2O or Fe(NO3)3‚6H2O was added, and the mixture was stirred magnetically for 2 h at 50 °C until it became a transparent greenish or reddish clear solution for Cr- or Fedoped La2Ti2O7, respectively. The solution obtained was heated at 130 ( 5 °C with stirring to promote esterification between EG and CA and remove excess solvents. The continued heating at ca. 130 °C over several hours made the solution become highly viscous with a change in color from greenish or reddish to deep yellow while generating white fumes, and finally it gelled into a transparent brown glassy resin without any precipitation. Charring the glassy resin at 350 °C for 2 h in an

Hwang et al. electric furnace resulted in a black powder, which was heat treated at 650-1250 °C for 2 h in static air on an Al2O3 boat followed by cooling to room temperature. After this heat treatment, the obtained light yellowish and light pink powders were Cr- and Fe-doped La2Ti2O7, respectively. For comparison, Cr- and Fe-doped La2Ti2O7 were also prepared by the conventional solid-state reaction; a stoichiometric mixture of Cr2O3 or Fe2O3 together with La2O3 and TiO2 was ground mechanically in the presence of ethanol for 1 h followed by calcination at 1150-1250 °C for 10 h. The Pt deposition on the prepared catalysts was performed by the photoplatinization method; 0.05 g of H2PtCl6 was introduced into the reaction system (EtOH 50 mL + distilled water 300 mL) containing 1 g of catalyst, and then UV light from a high-pressure Hg lamp (Ace Glass) was illuminated through a Pyrex filter in an Ar atmosphere for 2 h. Characterization. The crystal structure of the sintered powder was determined by X-ray diffraction (XRD) on a Mac Science Co. M18XHF with monochromated Cu KR radiation at 40 kV and 200 mA. The optical properties were analyzed by a UV-vis diffuse reflectance spectrometer (Shimadzu, UV 2401). The BET surface area was evaluated by N2 adsorption on a constant volume adsorption apparatus (Micrometrics, ASAP 2012). The ionic state of the doped Fe and Cr as well as the valence band spectra were obtained from X-ray photon spectroscopy measurements (VG Scientific, ESCALAB 220iXL) using Mg KR radiation (1253.6 eV). The binding energy was calibrated using the C1s peak as a reference energy. Thermogravimetry (TG) was carried out to follow the decomposition of precursors with a heating rate of 10 °C/min from 30 to 800 °C. Electronic Structure Calculation. The electronic structure calculation was based on the FLAPW (full potential linearized augmented plane wave) method which used the generalized gradient approximation (GGA), an improvement of the local spin density approximation (LSDA) within the density functional theory, known to be an efficient and accurate scheme for solving many-electron problems of a crystal. The Wien97 package was used in this study.14 The crystallographic parameters including lattice parameters and atomic positions were adopted from the literature for the calculation.15 XANES Analysis. X-ray absorption spectra of the Fe K-edge and Ti K-edge were recorded at the Pohang Accelerator Laboratory (PAL) on beamline BL3C1 operating at 2.5 GeV with ca. 100-160 mA of stored current. The radiation was monochromatized using a Si(111) double-crystal monochromator to collect high-resolution XAS spectra. Data were collected at room temperature in transmission mode. N2 was used as a detector gas for both the incident and transmitted beam. The absorption energy was calibrated by measuring Fe and Ti metal foil standards, assigning the first inflection point to 7112 eV and 4966 eV, respectively. The XANES analysis was processed according to the standard procedure using the program WinXAS 97 (version 2.3).16-18 Thus, the preedge background was removed by fitting a preedge region of a spectrum with a straight line and by subtracting the extrapolated values from the entire spectrum. The resulting elemental absorption was then normalized in absorbance by using the edge jump defined by the cubic spline fit in the postedge region. Photocatalytic Reaction. For the photocatalytic reaction under UV irradiation, the reaction was carried out at room temperature in a closed gas circulation system using a highpressure Hg lamp (Ace Glass Inc., 450 W) placed in an inner irradiation-type quartz reaction cell. The catalyst (1.0 g) was

Photocatalytic H2 Production over M-Doped La2Ti2O7

J. Phys. Chem. B, Vol. 109, No. 6, 2005 2095

Figure 3. Dependence of the X-ray diffraction pattern of Cr-doped La2Ti2O7 (Cr/La ) 0.01) on the preparation method.

Figure 2. XRD patterns of Cr-doped La2Ti2O7 (Cr/La ) 0.01) with different calcination temperatures. All catalysts were prepared from the polymerized complex method followed by calcination at the indicated temperature for 2 h.

suspended in distilled water (500 mL) by magnetic stirring. For irradiation of visible light with wavelengths larger than 420 nm, the reaction was performed in an outer-irradiation-type Pyrex reactor equipped with a cutoff filter (λ > 420 nm) using a highpressure Hg lamp (Oriel, 500 W). The catalyst (0.5 g) was suspended in distilled water (100 mL) together with methanol (50 mL) by magnetic stirring. The amount of H2 evolved was analyzed by gas chromatography (TCD, molecular sieve 5 Å column, and Ar carrier). Results and Discussion Crystal Structures of Metal-Doped La2Ti2O7. According to the TGA analysis of the Cr-doped La-Ti precursor in air, most of the organics introduced during the catalyst preparation decomposed at temperatures below 550 °C. Thus, calcination of all catalysts was performed above 650 °C. The XRD patterns in Figure 2 represent the crystal growth of Cr-doped La2Ti2O7 (Cr/La ) 0.01) with increasing calcination temperatures. All catalysts were prepared by the citrate complex method followed by calcination at the indicated temperatures for 2 h. The precursor prepared by charring the gel sustained an amorphous phase below 650 °C, as shown by a broad peak covering the 2θ range between 24° and 34°. Crystallization began to occur at 700 °C, and all diffraction peaks of the precursor heat treated above 950 °C could be assigned to the pure La2Ti2O7 phase with a monoclinic structure (JCPDS: 28-0517). No other peaks related to a Cr-containing structure were observed. The width of all reflection lines was decreased with increasing calcination temperatures without a change in the overall patterns of the diffraction lines. The crystal structure of Cr- and Fe-doped La2Ti2O7 also depended highly on the preparation method. The crystal phase present in Cr-doped La2Ti2O7 prepared by the polymerized complex method (PC) grew without any impurity phase, as shown in Figure 3. However, the same sample prepared by the solid-state reaction method (SSR) calcined at 1050 °C for 10 h yielded some unreacted precursor phases. As the calcination temperature was increased to 1150 °C, the intensity of the unreacted phases was reduced, which resulted in almost the same

pattern as that made by the PC method. Therefore, it could be concluded that the polymerized complex method using citrate and ethylene glycol as the complexation agent made it possible to fabricate a more homogeneous structure than the solid-state reaction (SSR). This advantage of the PC method appears to have resulted from mixing of each component (Cr or Fe, La, and Ti) on the molecular level. The crystal structure was also highly dependent on the doping content of Cr and Fe. As shown in Figure 4A, Cr-doped La2Ti2O7 showed a phase assigned to orthorhombic LaCrO3 at dopant mol concentrations higher than Cr/La ) 0.05 (c), and the concentration of this phase increased with Cr content (d, e). For a La-deficient sample (f), the intensity of the LaCrO3 phase was comparable to that of La2Ti2O7. From this result, it could be concluded that the direct substitution of Cr ion for a La site in La2Ti2O7 was impossible, and instead, three separate phases of LaCrO3, La2Ti2O7, and TiO2 were obtained, as shown in following equation.

0.2Cr + 1.8La + 2.0Ti + 7[O] f 0.2LaCrO3 + 0.8La2Ti2O7 + 0.4TiO2 (1) However, for a Ti-deficient sample (g), the diffraction peak corresponding to orthorhombic LaCrO3 was not observed at all. To maintain the total charge balance, the structure equation would be La2[Cr0.2Ti1.8]O6.9, which means that oxygen defects exist inevitably in the lattice of La2Ti2O7. The stoichiometric mixture of Cr/Ti ) 1 without La (h) yielded a pyroclore-like Cr2Ti2O7 structure together with some unidentified peaks. The stoichiometric mixture of Cr/La ) 1 without Ti (i) yielded the pure orthorhombic LaCrO3 structure. Another feature to be mentioned here is that as the concentration of Cr dopant increased, the intensity of the diffraction peak corresponding to La2Ti2O7 decreased, which inferred that the Cr dopant inhibited the growth of La2Ti2O7 crystals. Fe-doped La2Ti2O7 showed a similar pattern of structural growth as that of the Cr-doped one, as shown in Figure 4B. The monoclinic structure of La2Ti2O7 was sustained up to an Fe dopant concentration of Fe/La ) 0.04 (c), and above this loading (d-f), the Fe3Ti3O10 phase, not FeLaO3, began to grow. For the La-deficient sample (g), the intensity of the Fe3Ti3O10 phase was comparable to that of the La2Ti2O7 phase. From this result, it could be concluded that the direct substitution of Fe ion for a La site in La2Ti2O7 seemed to be very difficult, and instead, two separate phases of Fe3Ti3O10 and La2Ti2O7 were synthesized, as shown in following equation.

0.2Fe + 1.8La + 2.0Ti + 6.97[O] f 0.067Fe3Ti3O10 + 0.9La2Ti2O7 (2)


Hwang et al.

Figure 4. Dependence of the XRD pattern of La2Ti2O7 on Cr dopant concentration (A) and on Fe dopant concentration (B); both in atom ratios.

For the Ti-deficient sample (h), the diffraction peak corresponding to the orthorhombic Fe3Ti3O10 phase disappeared completely. In this case, the structure equation would be La2[Fe0.2Ti1.8]O6.9 to maintain the total charge valence. The substitution of Fe and Cr for Ti instead of La is consistent with expectation considering cationic sizes. The ionic radii of La3+ and Ti4+ are 1.04 and 0.60 Å, respectively, while those of Fe3+ and Cr3+ are both 0.53 Å.19 Thus these doped cations are more likely to substitute for Ti4+ sites of similar size. UV-Vis Diffuse Reflectance Spectra. The absorption pattern of La2Ti2O7 was highly dependent on the dopant concentration as shown in Figure 5. When Cr was doped on La2T2O7, two types of absorption were generated in the visible light region and the band gap absorption of La2Ti2O7 with an absorption edge of 325 nm (3.82 eV) was not affected. The absorption edge of 560 nm was ascribed to the charge transfer from Cr3+ to Ti4+, while broad absorption ranging from 600 to 800 nm was ascribed to a d-d transition of 4A2 f 4T2 in Cr3+ ions in octahedral systems.20 The energy gap of Cr-doped La2Ti2O7 was 2.2 eV, which suggested that the energy level of Cr3+ was 2.2 eV more negative than the conduction band of La2Ti2O7. It was noticeable that the Ti-deficient sample (h) showed a smaller absorption than both the stoichiometric (i) and La-deficient samples (j). As proved by the X-ray diffraction result, both Cr2Ti2O7 (k) and LaCrO3 (l) phases with large absorption coefficients were not observed in the Ti-deficient sample, which resulted in smaller absorption.

The absorption pattern of Cr-doped La2Ti2O7 was also dependent on both sintering temperatures and preparation methods, as shown in Figure 6. Samples prepared by the polymerized complex method followed by calcination at 850 °C (a, d, g) showed an intermediate absorption peak near 360 nm. As the Cr content was increased (a f d f g), the absorption peak became flat due to the larger absorption of Cr-containing species such as CrLaO3 and Cr2Ti2O7. The calcination at 1150 °C (c, h, i) made this absorption peak disappear even for the lowest loading (a f c). Therefore, it could be concluded that the Cr ion was inserted into the lattice of the monoclinic La2Ti2O7 after heat treatment at temperatures higher than 1150 °C. UV-vis DRS patterns of the samples prepared by the solidstate reaction method (b, e, f) showed an additional absorption near 400 nm, which was ascribed to unreacted precursor oxides, as demonstrated by the XRD results. In addition, the absorption intensity starting at 560 nm resulting from the Cr3+ f Ti4+ transition for samples prepared by the solid-state reaction was smaller than for samples made by the polymerized complex (PC) method (c, h, i). Therefore, it was confirmed from the results of the UV-vis DRS that the PC method was superior to the solid-state reaction (SSR) in terms of the homogeneity of the product. Fe-doped La2Ti2O7 prepared by the PC method showed a similar absorption pattern as the Cr-doped one; as the Fe contents and calcination temperatures were increased, the visible light absorption was increased. However, the absorption near



Figure 5. UV-vis diffuse reflectance spectra of Cr-doped La2Ti2O7 prepared by the polymerized complex method followed by calcination at 1150 °C for 2 h.

Figure 6. Dependence of the UV-vis diffuse reflectance spectra of Cr-doped La2Ti2O7 on the preparation method. PC: polymerized complex method. SSR: solid-state reaction method.

325 nm resulting from the band gap excitation of La2Ti2O7 was red-shifted even by calcination at 850 °C for the Fe-doped one, which was not observed when Cr was used as a dopant. The band gap energy of Fe-doped La2Ti2O7 was ca. 2.6 eV. The UV-vis absorption property of Cr- and Fe-doped La2Ti2O7 is compared in Figure 7. For the same dopant concentration, the Cr-doped sample with the absorption edge of 560 nm absorbed a larger amount of visible light than the Fe-doped one with the absorption edge of 480 nm, since the absorption edge wavelength of the Cr-doped sample was longer. XPS Study on the Cr- and Fe-Doped La2Ti2O7. The oxidation state of doped Cr and Fe was investigated by XPS. As shown in Figure 8A, Cr doped on La2Ti2O7 showed a large peak assigned to trivalent chromium as that in Cr2O3. But, their binding energies were shifted to higher energies from that of

Cr2O3 at 577.0 eV, with increasing Cr concentrations up to 0.6 eV at a Cr/La of 0.05. Similarly, the ionic state of doped Fe was also found to be trivalent and no other ionic states were found (not shown). The Ti2p spectra in Figure 8B demonstrated that the ionic state of Ti4+ did not change upon doping up to a Cr/La ) 0.05. The La3d XPS spectra showed doublets at 835.0 and 852.0 eV typical of trivalent La ion21 and were not affected by doping with Cr or Fe (not shown). The valence band spectra were also taken to find out the contribution of individual atoms to the valence band. As shown in Figure 9, the physical mixture of La2O3, TiO2, and Cr2O3 showed two broad peaks at 2.8 and 6.7 eV. The peaks at 2.8 and 6.7 eV were ascribed to the valence bands of Cr2O3 and TiO2 combined with La2O3, respectively. The Cr-doped La2Ti2O7 of both concentrations showed only one peak at 6.7 eV, which


Figure 7. UV-vis diffuse reflectance spectra of Cr- and Fe-doped La2Ti2O7.

was the same for undoped La2Ti2O7. Therefore, it could be concluded that doped Cr or Fe was substituted into the lattice of the La2Ti2O7 structure by calcination at 1150 °C, and thus the contribution of the doped Cr or Fe ion on the valence band of La2Ti2O7 was negligible. The energy level of these dopants would be located between the conduction and the valence band. X-ray Absorption Near-Edge Spectroscopy (XANES). The K-edge XANES spectra of Fe and Ti were measured in order to find out the electronic and local coordination structure of iron oxide stabilized in La2Ti2O7. Figure 10 shows the Fe K-edge spectra of reference compounds and Fe-doped La2Ti2O7 calcined at 1150 °C for 2 h. Iron oxides have three forms, FeO, Fe2O3, and Fe3O4. Fe3+ ions in Fe2O3 and Fe2+ ions in FeO are both in six-coordinate octahedral sites (Oh). In Fe3O4, all of the Fe2+ and one-half of the Fe3+ ions are in Oh symmetry and the other half of the Fe3+ ions are in four-coordinate tetrahedral sites (Td). By parity consideration of X-ray absorption, the 1s f 3d transition is dipole-allowed for Td, while it is forbidden for Oh. The features of Fe K-edge XANES originate from the transitions of 1s electrons to empty 3d and 4p orbitals. At the preedge near an energy of 7112 eV, only very weak features

Figure 8. X-ray photoelectron spectra of Cr2p (A) and Ti2p (B).

Hwang et al. are observed for Fe2O3 and FeO since the 1s f 3d transition is forbidden by their Oh symmetry. As stated above, one-quarter of the Fe centers in Fe3O4 are in Td sites; therefore, a more intense 1s f 3d peak is observed generally for Fe3O4.22-24 The main absorption centered at about 7130 eV represents the 1s f 4p transition. The features on the absorption edge ranging from 7115 to 7130 eV result from the transition of 1s electrons to all of the possible energy levels above the lowest empty d orbitals. The absorption edge energies (defined as the energy with the maximum value of the first derivative function in the edge-step of the absorbance-versus-energy plot) for Fe foil, FeO, and Fe2O3 were 7112.1, 7117.8, and 7122.5 eV, respectively. XANES of Fe-doped La2Ti2O7 calcined at 1150 °C for 2 h showed an edge energy of 7123.4 eV, indicating that iron in the sample is trivalent. However, its fine structures did not match exactly with any of the reference compounds. Most obvious was the near absence of the preedge peak, especially for the sample with an Fe/La of 0.02. The fine structures of the main absorption peak were also different from those of Fe2O3. Among the prepared samples, only Fe2Ti2O7 showed the preedge peak. The absence of the preedge peak should not be due to particle size, because nanosized Fe2O3 showed an intensified preedge peak relative to bulk Fe2O3.22 Thus, Fe-doped La2Ti2O7 does not contain isolated Fe2O3 particles. Instead, Fe3+ is present probably in the lattice of La2Ti2O7 with a higher degree of Oh symmetry compared to that of Fe2O3. The preedge features of Ti K-edge XANES for various Ticontaining oxides are shown in Figure 11. The preedge spectrum of crystalline TiO2-anatase which has a somewhat distorted TiO6 octahedron (4 × Ti-O ) 1.939 Å, 2 × Ti-O ) 1.980 Å) contains the characteristic triplet structure in this region (49674977 eV) due to the 1s f 3d core level excitations. However, in addition to the three obvious peaks labeled A1, A3, and B, a fourth peak, A2, is indicated on the low-energy side of the central A3 peak. The A1 peak has been attributed to the core hole potential or 3d-4p hybridized states. The A2 transition was ascribed to an increasing contribution from five-coordinate Ti species located on the crystallite surface in nanoparticles.25-28 As particle size is reduced, more Ti atoms on the surface are in an anisotropic environment, causing distortions around the Ti atoms from an octahedral TiO6 unit. For all samples, the A1 and B peak positions were constant. However, the A2 and A3


Figure 9. Valence band spectra of Cr-doped La2Ti2O7.

Figure 10. Fe K-edge XANES of reference samples and Fe-doped La2Ti2O7 calcined at 1150 °C for 2 h.

peak positions changed with the type of oxide. TiO2 with the rutile structure showed the A1 transition with a lower intensity and the A3 transition at a lower energy compared with TiO2 with the anatase structure. The overall shape of Fe2Ti2O7 and Cr2Ti2O7 was similar to that of TiO2 with the rutile structure, which implied that Cr2Ti2O7 and Fe2Ti2O7 have the similar local structure as Ti with rutile TiO2. But, the A3 peak of Cr2Ti2O7 and Fe2Ti2O7 was located at a lower energy than that of rutile TiO2. For La2Ti2O7, the A3 peak (4972 eV) almost disappeared and the A2 peak became dominant. The peak B (4974.5 eV) position showed no significant change, although the intensity was increased a little. The existence of the A2 peak was ascribed to the distorted TiO6 in La2Ti2O7 due to the asymmetrical distance between Ti and neighboring O atoms. XANES spectra of Fe- and Cr-doped La2Ti2O7 calcined at 1150 °C did not change significantly in both peak position and intensity. This might be ascribed to a concentration of dopant that was too low to affect the environment of Ti in La2Ti2O7. However, Cr-doped La2Ti2O7 calcined at 850 °C showed a little

J. Phys. Chem. B, Vol. 109, No. 6, 2005 2099 different feature; the octahedral symmetry was broken at dopant concentrations higher than Cr/La ) 0.2, as indicated by the loss of the A1 peak. This inferred that the crystal growth of La2Ti2O7 was highly affected by dopant concentration, as was also indicated by the results of XRD and UV-vis DRS. In any case, it could be concluded from the XANES of the Fe and Ti K-edge of Fe-doped La2Ti2O7 that the doped Fe3+ occupies Ti4+ sites in the La2Ti2O7 lattice that have a similar environment (distorted octahedral symmetry). Calculation of Electronic Band Structure of Cr-Doped La2Ti2O7. Considering that doped Fe or Cr substituted for Ti in La2Ti2O7, the electronic band structures of Cr-substituted La2Ti2O7 were calculated for a fictitious composition of La2Cr0.5Ti1.5O7. As shown in Figure 12A, the edge position and shape of both the valence band and conduction band as well as the band gap energy of 3.2 eV were not affected by substitution of Cr for Ti. However, the most significant feature in the band structure of Cr-doped La2Ti2O7 was the formation of a new band near 1.8-2.3 eV. This localized band formed in the original band gap of La2Ti2O7 came from the Cr3d orbital as shown in Figure 12C. Figure 12C also indicated small contributions of oxygen and Ti and essentially no contribution of La to the formation of this interband. The contribution of the unoccupied Cr3d orbital to the conduction band was comparable to that of Ti3d (Figure 12D), while the occupied Cr3d orbital contributed mainly to the lower energy part of the valence band (Figure 12B). The conduction band of La2Ti2O7 consisted mainly of the unoccupied Ti3d orbital, especially in the lower energy region (3-5 eV), and thus the contribution of the unoccupied Cr3d orbital to the conduction band was substantial for La2Cr0.5Ti1.5O7. The localized Cr3d interband was also positioned closer to the conduction band than to the valence band. This relative position in the band gap, together with the relative contribution of Cr and La to the formation of this interband, could be explained by the fact that overlap of Cr with Ti was more substantial than that with La in La2Cr0.5Ti1.5O7. Photocatalytic Activity of Cr- and Fe-Doped La2Ti2O7. Photocatalytic activity was measured for pure water under UV light (>200 nm) and in the presence of methanol under visible light irradiation with a cutoff filter (>420 nm). To promote hydrogen generation, Cr-doped La2Ti2O7 (Cr/La ) 0.01) was loaded with 1.0 wt % Ni for UV experiments and 1.0 wt % Pt for visible light experiments.8,29 In both cases, H2 was evolved steadily with irradiation time and the H2 evolution stopped when light was turned off. Therefore, it could be concluded that H2 evolution over Cr-doped La2Ti2O7 occurred photocatalytically. Under UV irradiation, the photocatalytic activity of Cr-and Fedoped La2Ti2O7 was lower than that of the undoped one for both nickel- and platinum-loaded catalysts as shown in Table 1. As the dopant concentration was increased, the activity was lowered further. For undoped La2Ti2O7, electrons in the conduction band excited from the valence band under UV irradiation could be utilized to produce H2 from the oVerall splitting of pure water. According to the band structure calculation of Crdoped La2Ti2O7, the major effect of Cr doping was the formation of an interband in the band gap. Under UV irradiation, the electrons in the interband as well as the electrons in the valence band could be excited to the conduction band, resulting in holes in both the interband and valence band. However, water molecule could not be oxidized by the holes in the interband due to the lower oxidation potential than that required for water oxidation, and thus the new interband could not contribute to the water-splitting activity. Instead, it could serve as a recom-


Hwang et al.

Figure 11. Ti K-edge XANES of Cr- and Fe-doped La2Ti2O7 with reference compounds.

Figure 12. Electronic band structure of La2Cr0.5Ti1.5O7.

bination site for photogenerated electrons and holes, which resulted in the lower activity of Cr- and Fe-doped La2Ti2O7.

Under visible light irradiation, nickel oxide loading on these catalysts did not promote photocatalytic activity, although nickel



TABLE 1: Photocatalytic Hydrogen Production from Water over Cr- and Fe-Doped La2Ti2O7 activity (µmol H2/h)

material

UV irradiation (λ > 200 nm)a

visible light irradiation (λ > 420 nm)b

Pt/La2Ti2O7c Pt/Cr-La2Ti2O7 (Cr/La ) 0.01)c Pt/Cr,Sb-La2Ti2O7 (Cr/La ) 0.01)c Pt/Cr-La2Ti2O7 (Cr/La ) 0.05)c Pt/Fe-La2Ti2O7 (Fe/La ) 0.01)c Pt/Fe-La2Ti2O7 (Fe/La ) 0.05)c NiOx/La2Ti2O7d NiOx/Cr-La2Ti2O7 (Cr/La ) 0.01)d NiOx/Fe-La2Ti2O7 (Fe/La ) 0.01)d

120 90 70 50 45 32 400 275 269

0 15 4 8 10 5 0 0 0

a Measured in an inner irradiation quartz reaction cell under irradiation from a 450 W high-pressure Hg lamp; H2O 500 mL, catalyst 1.0 g. b Measured in an outer irradiation cell from a 500 W highpressure Hg lamp; H2O 100 mL + MeOH 50 mL, catalyst 0.5 g. A 420 nm cutoff filter was used. c 1.0 wt % of Pt was loaded using H2PtCl6 by photoplatinization. d Materials were loaded with Ni (1.0 wt %) and then pretreated by reduction at 773 K for 2 h followed by oxidation at 473 K for 1 h.

loading yielded higher activities than Pt deposition under UV irradiation for all catalysts. In addition, no activity was observed when only pure water was used as the reactant. Continuous H2 evolution was observed when CH3OH was added as a hole scavenger. For the same dopant concentration, the H2 evolution rate was a little higher for Cr-doped La2Ti2O7 than for the Fedoped one, which resulted from the difference in the amount of visible light absorption. As shown in Figure 7, the band gap energy of Cr- and Fe-doped La2Ti2O7 was 2.2 eV and 2.6 eV, respectively, which implied that a larger number of photons could be absorbed by Cr-doped La2Ti2O7. As the dopant concentration was increased, the activity was decreased for both catalysts, although the amounts of visible light absorption were larger for catalysts with higher dopant concentrations. This was again ascribed to the reduced crystallinity of La2Ti2O7, which was highly dependent on the dopant concentration, as demonstrated by the XRD results. The photocatalytic activity of Sband Cr-codoped La2Ti2O7 was also measured. According to Kudo’s report for Cr-doped SrTiO3, the activity was enhanced by codoping of Sb5+, since the charge balance was kept by codoping of Sb5+ with Cr3+, resulting in the suppression of the formation of Cr6+ ions and oxygen defects in the lattice which could work as nonradiative recombination centers between photogenerated electrons and holes.12 However, additional doping of Sb5+ with Cr3+ on La2Ti2O7 did not induce the enhancement of photocatalytic activity. This implies that at very low dopant concentrations (Cr/La ) 0.01 and 0.05), the formation of Cr6+ ion would not be substantial, as proved by the Cr2p spectra of Cr-doped La2Ti2O7, and thus the role of Sb5+ as a charge-balancing agent was not so important. Mechanism of H2 Evolution over M-Doped La2Ti2O7. From the electronic structure calculation and characterization results of XRD, UV-vis, XANES, and XPS, the band structure of Cr- or Fe-doped La2Ti2O7 can be schematically described as in Figure 13. The conduction band of La2Ti2O7 consists mainly of broad Ti3d and sharp La4f orbitals, while the valence band consists mainly of O2p.30 The band gap energy between the valence and conduction bands of M-La2Ti2O7 is 3.8 eV, while the partially filled Cr3d and Fe3d bands are located 2.2 eV and 2.6 eV below the conduction band, respectively. When light with wavelengths longer than 420 nm is used for illumination, the electrons in the Cr3d or Fe3d band are excited to the conduction band, while Cr3+ or Fe3+ loses one electron and

Figure 13. Schematic band structure of M-La2Ti2O7 (M ) Cr, Fe) and its mechanism of H2 evolution from water containing CH3OH.

becomes Cr4+ or Fe4+. There is no such photoexcitation of electrons in the valence band of La2Ti2O7 because the energy of the incident light is much less than the band gap energy. The electron excited to the conduction band has a sufficient reduction potential to reduce H+ ion. Yet, H2 was not generated over bare (Cr, Fe)-La2Ti2O7 due to the substantial recombination of the excited electron and the oxidized metal species (Cr4+ or Fe4+). H2 was generated only when Pt metal was present on the external surface of this catalyst, where the formation of an ohmic junction between Pt and doped La2Ti2O7 might enhance the charge carrier separation.31 Cr4+ or Fe4+, the oxidized form of Cr3+ or Fe3+, is utilized to oxidize the CH3OH, and then it returns to its original state, Cr3+ or Fe3+. Conclusion Among various metal dopants for La2Ti2O7, only Cr and Fe showed intense absorption in the visible light region (>400 nm), and only these catalysts produced H2 photocatalytically in the presence of methanol under visible light irradiation (>420 nm). The polymerized complex method was found to be more efficient for fabrication of the present catalysts producing a more homogeneous phase than the conventional solid-state reaction. The characterization by XRD, UV-vis DRS, XPS, and XANES revealed that doped Cr and Fe were present in the Cr3+ and Fe3+ states substituting for Ti sites in the La2Ti2O7 lattice. The theoretical calculation indicated that the most significant feature in the electronic band structure of the metal-doped La2Ti2O7 was the formation of a partially filled 3d band in the band gap of La2Ti2O7, while the contribution of these dopants on the valence band was negligible. Excitation of electrons from this localized interband to the conduction band of La2Ti2O7 was responsible for visible light absorption and the H2 evolution from water under visible light. Acknowledgment. This work was supported by the General Motors R&D Center, the R&D Center for Hydrogen Energy, the Brain Korea 21 Project, the National R&D Project for Nano Science and Technology, and from the Research Center for Energy Conversion and Storage. References and Notes (1) Domen, K.; Kudo, A.; Onishi, T. J. Catal. 1986, 102, 92. (2) Inoue, Y.; Asai, Y.; Sato, K. J. Chem. Soc., Faraday Trans. 1994, 90, 797. (3) Kudo, A.; Kato, H. Chem. Lett. 1997, 867. (4) Kim, H. G.; Hwang, D. W.; Kim, J.; Kim, Y. G.; Lee, J. S. Chem. Commun. 1999, 1077. (5) Koca, M.; Sahin, M. Int. J. Hydrogen Energy 2002, 27, 4, 363. (6) Wu, J.; Lin, J. M.; Shu, Y. B.; Sato, T. J. Mater. Chem. 2001, 11, 12, 3343.

2102 J. Phys. Chem. B, Vol. 109, No. 6, 2005 (7) Domen, K.; Hara, M.; Kondo, J. N.; Takata, T.; Kudo, A.; Kobayashi, H.; Inoue, Y. Korean J. Chem. Eng. 2001, 18, 862. (8) Hwang, D. W.; Kim, H. G.; Kim, J.; Cha, K. Y.; Kim, Y. G.; Lee, J. S. J. Catal. 2000, 193, 40. (9) Hwang, D. W.; Cha, K. Y.; Kim, J.; Kim, H. G.; Bae, S. W. Lee, J. S. Ind. Eng. Chem. Res. 2003, 42, 1184. (10) Kim, J.; Hwang, D. W.; Bae, S. W.; Kim, Y. G.; Lee, J. S. Korean J. Chem. Eng. 2001, 18, 941. (11) Borgarello, E.; Kiwi, J.; Gra¨tzel, M.; Pelizzetti, E.; Visca, M. J. Am. Chem. Soc. 1982, 104, 2996. (12) Kato, H.; Kudo, A. J. Phys. Chem. B 2002, 106, 5029. (13) Ohno, T.; Tanigawa, F.; Fujihara, K.; Izumi, S.; Matsumura, M. J. Photochem. Photobiol. A 1999, 127, 107. (14) Blaha, P.; Schwarz, K.; Luitz, J. WIEN97; A Full Potential Linearized Augmented Plane WaVe Package for Calculating Crystal Properties; Karlheinz Schwarz Technical Universita¨t: Wien, Austria, 1999 ISBN 3-9501031-0-4. (15) Schmalle, H. W.; Williams, T.; Reller, A.; Linden, A.; Bednorz, J. G. Acta Crystallogr. 1993, B49, 235. (16) Kim, W. B.; Choi, S. H.; Lee, J. S. J. Phys. Chem. B 2000, 104, 8670. (17) Park, E. D.; Choi, S. H.; Lee, J. S. J. Catal. 2000, 194, 33. (18) Kim, M. G.; Lee, M. K.; Shin, H. J.; Lee, J. M.; Lee, J. S.; Yo, C. H.; Yoon, W. S.; Kim, K. B. J. Phys. Chem. B 2002, 106, 2526.

Hwang et al. (19) Cotton, F. A.; Wilkinson, G. AdVanced Inorganic Chemistry, 4th ed.; Wiley: New York, 1980. (20) Serpone, N.; Lawless, D. Langmuir 1994, 10, 643. (21) Yong, W. Y.; Thorn, R. J. J. Chem. Phys. Solids 1980, 41, 75. (22) Chen, L. X.; Liu, T.; Thurnauer, C.; Csencsits, R.; Rajh, T. J. Phys. Chem. B 2002, 106, 8539. (23) Lin, L. W.; Kou, Y.; Zou, M.; Yan, Z. Phys. Chem. Chem. Phys. 2001, 3, 1789. (24) Han, Y.; Choi, J.; Jang, J.; Kim, D. J. Solid State Chem. 2001, 160, 435. (25) Luca, V.; Djajanti, S.; Howe, R. F. J. Phys. Chem. B 1998, 102, 10650. (26) Hanley, T.; Luca, V.; Pickering, I.; Howe, R. F. J. Phys. Chem. B 2002, 106, 1153. (27) Wu, Z. Y.; Ouvrard, G.; Gressier, P.; Natoli, C. R. Phys. ReV. 1997, B55, 10382. (28) Parlebas, J. C.; Khan, M. A.; Uozumi, T.; Okada, K.; Kotani, A. J. Electron Spectrosc. Relat. Phenom. 1995, 71, 117. (29) Hwang, D. W.; Kim, J.; Lee, J. S. Catal. Lett. 2002, 80, 53. (30) Hwang, D. W.; Li, W.; Oh, S. H.; Lee, J. S. J. Phys. Chem. B 2003, 107, 4963. (31) Gregory, A.; Allen, J. B. J. Phys. Chem. 1983, 87, 1979.

Photocatalytic Hydrogen Production from Water over M-Doped

Recommend Documents