A realization of low-temperature NO x reduction from an exhaust gas at stationary emission sources, such as factories, boilers, power stations and incinerators, is strongly desired. In addition, some ammonia slips from factories and pig and poultry farms are serious problems from an environmental viewpoint. We found that a TiO 2 -based photocatalyst is effective for the abatement of NO (photo-SCR) and the removal of NH 3 (photo-SCO), and that the loading of tungsten (W) oxide species on TiO 2 promotes photocatalyses [1,2]. Not only the loading amount but also the local structure of W oxide species affects the activities. A conventional XAFS analytical method could give us only a vague result for the structure of the loaded W oxide species. W L 1 - and L 3 -edge X-ray absorptions have often been used to provide information on the local symmetry, coordination and valence of W oxide species loaded on supports [3]. The valence of the W oxide species is determined by the position of the W L 1 -edge. The local symmetry is determined by the area of the pre-edge peak in W L 1 -edge XANES. The coordination number and bond distance of the W oxide species are estimated by the analysis of W L 3 -edge EXAFS. However, the study of W L 3 -edge XANES is limited. The white line appearing in the L 3 -edge XANES of transition metals is mainly attributed to electronic transitions from 2 p 3/2 orbitals to vacant d orbitals. In the case of Mo L 3 -edge X-ray absorption, two white lines are observed in the XANES. The splitting and areas of the two white lines depend on the symmetry of the MoO x ( x = 4, 5, 6) unit because these correspond to the ligand field splitting of the d orbitals [4]. Therefore, it is expected that W L 3 -edge XANES will reveal the characteristics of the 5 d orbitals in a manner similar to that of Mo L 3 -edge XANES. However, no such relationship between W L 3 - edge XANES and the structure has been described. This is because the white line is not distinctly split as is the case for Mo [5]. In the present study, we recorded the W L 1 - and L 3 -edge X-ray absorption spectra of reference samples, whose structures are already known, at beamline BL01B1 in a transmission mode. We analyzed the W L 3 - edge XANES carefully to clarify the relationship between the splitting of the W L 3 -edge white line and the structure of the WO x unit. Additionally, we found that the structure of the WO x unit can be estimated from the splitting of the W L 3 -edge white line and the pre-edge peak area of the W L 1 -edge XANES. This new structural analysis method can also easily estimate the structure of the W 6+ oxide loaded on TiO 2 . Figure 1(A) shows the W L 3 -edge XANES spectra of reference samples with the W 6+ ion. Cr 2 WO 6 and Ba 2 NiWO 6 , which have D 2 and Oh symmetries, give two peaks in the white line. On the other hand, WO 3 , H 3 PW 12 O 40 , and (NH 4 ) 10 W 12 O 41 • 5H 2 O samples, which have a distorted WO 6 unit, and Sc 2 W 3 O 12 and Na 2 WO 4 , which respectively have nearly Td and Td symmetries, exhibit only one peak. However, the second derivatives of the W L 3 -edge white line clarify that the white line consists of two peaks (Sc 2 W 3 O 12 and Na 2 WO 4 give a large peak with a shoulder peak at a low energy), as shown in Fig. 1(B). In general, 5 d orbitals of the octahedral (tetrahedral) WO 6 (WO 4 ) unit are split into e g and t 2g ( e and t g ) orbitals by a ligand field. The poorer the symmetry of WO 6 (WO 4 ) from Oh ( Td ), the smaller (larger) the energy gap of the split 5 d orbitals. Additionally, the energy gap for the WO 4 unit is smaller than that for the WO 6 unit. Therefore, the energy gap of the two peaks in Fig. 1(B) reflects the splitting of 5 d orbitals by the ligand field. Additionally, we carried out the deconvolution of the W L 3 -edge XANES spectra. All spectra were fitted with two Lorentz functions, whose top positions were determined on the basis of the values of the two peaks in Fig. 1(B), and an arctangent function. Figure 2 shows the results of the deconvolution of WO 3 and Na 2 WO 4 , as examples. The ratio of the areas under the two peaks (peak 1: peak 2; peak 1 appears at a lower energy than peak 2) for the samples with the WO 6 unit was about 3:2, whereas that for the sample with the WO 4 unit was about 2:3. For the WO 6 (WO 4 ) unit, the t 2g ( e) orbital appears at a 138 X XA AF FS S S St tu ud dy y o of f T Tu un ng gs st te en n L L 1 1 - - , , L L 3 3 - - E Ed dg ge es s: : S St tr ru uc ct tu ur ra al l A An na al ly ys si is s o of f L Lo oa ad de ed d T Tu un ng gs st te en n O Ox xi id de e S Sp pe ec ci ie es s Environmental Science A B C 12160 12120 12080 Photon Energy (eV) (a) (b) (d) (c) (f) (e) (g) 10220 10200 Photon Energy (eV) (g) (f) (e) (d) (c) (b) (a) 10220 10210 10200 Photon Energy (eV) (a) (b) (c) (d) (f) (g) (e) Normalized Absorption (arb. units) Normalized Absorption (arb. units) Second Derivatives (smoothed) Fig. 1. (A) W L 3 -edge XANES spectra, (B) second derivatives of (A) , and (C) W L 1 -edge XANES spectra of (a) Ba 2 NiWO 6 , (b) Cr 2 WO 6 , (c) (NH 4 ) 10 W 12 O 41 • 5H 2 O, (d) WO 3 , (e) H 3 PW 12 O 40 • 13H 2 O, (f) Sc 2 W 3 O 12 and (g) Na 2 WO 4 . 2.5 2.0 1.5 1.0 0.5 0.0 3.0 2.5 2.0 1.5 1.0 0.5 0.0 10240 10220 10200 10240 10220 10200 (a) (b) Photon Energy (eV) Photon Energy (eV) Normalized Absorption (arb. units) Normalized Absorption (arb. units) Spectrum Arctangent function Peak 1 Peak 2 Fitting function Spectrum Arctangent function Peak 1 Peak 2 Fitting function lower energy than the e g ( t g ) orbital and the X-ray absorption intensity is t 2 g : e g = 3 : 2 ( e : t g = 2 : 3). Thus, the ratios of peak 1 : peak 2 reflect the absorption intensities of the split 5 d states. From the above results, we conclude that the white line in the W L 3 - edge XANES indicates the electron transition from the 2 p 3 / 2 orbital to the split 5 d orbitals. Figure 1(C) shows the W L 1 -edge XANES spectra of the reference samples. The pre-edge peak is mainly attributed to the forbidden electron transition from a 2 s orbital to 5 d orbitals. The less symmetric structure of W oxide species (distortion from the Oh symmetry) provides a larger pre-edge peak because this forbidden electron transition is induced by mixing p orbitals of tungsten and ligand into empty d orbitals. Therefore, Na 2 WO 4 and Sc 2 W 3 O 12 exhibit a large pre- edge peak. We found that the combination of the area of the pre-edge peak of the W L 1 -edge XANES and the splitting of the W L 3 -edge white line provides more reliable information on the structure of the W oxide species. Figure 3 shows the area of the pre-edge peak of the W L 1 -edge XANES plotted against the energy gap of the split W L 3 -edge white line for the reference samples (represented by red solid circle). Figure 3 reveals an interesting relationship, in that the area of the pre-edge peak has a linear correlation with the energy gap of the split W L 3 -edge white line. This linear relationship is supported by density functional theory calculations of the octahedral and tetrahedral W models [2]. Therefore, we conclude that the structure of the W oxide species can be easily estimated by a combination of the pre-edge peak area of the W L 1 -edge XANES and the energy gap of the split W L 3 -edge white line. Using this new analytical method, we estimated the structure of the W oxide species loaded on TiO 2 samples from the above linear correlation. The samples used were x (2, 4, 12, 20, 40, and 80) mol g -1 H 3 PW 12 O 40 · 13H 2 O loaded on TiO 2 ( x HPA / TiO 2 ). The area of the pre-edge peak of W L 1 -edge XANES is plotted against the energy gap of the split W L 3 -edge white line for x HPA / TiO 2 in Fig. 3 (represented by blue solid square). As shown in Fig. 3, 2HPA/TiO 2 has a WO 6 structure and the amount of WO 4 species increases with an increase in the amount of loaded HPA. These estimated structures of W oxide species were in agreement with the curve fitting analysis results for the W L 3 -edge EXAFS [2]. Seiji Yamazoe a, *, Tetsuya Shishido b and Tsunehiro Tanaka b a Dept. of Materials Chemistry, Ryukoku University b Dept. of Molecular Engineering, Kyoto University *E-mail: yamazoe@rins.ryukoku.ac.jp References [1] S. Yamazoe et al. : Appl. Catal. B 83 (2008) 123. [2] S. Yamazoe, Y. Hitomi, T. Shishido and T. Tanaka: J. Phys. Chem. C 112 (2008) 6869. [3] M. Fernández-García et al. : J. Phys. Chem B 109 (2005) 6075. [4] G.N. George et al. : J. Am. Chem. Soc. 112 (1990) 2541. [5] F. Hilbrig et al. : J. Phys. Chem. 95 (1991) 6973. Fig. 2. Deconvolution of W L 3 -edge XANES spectra of (a) WO 3 and (b) Na 2 WO 4 . Fig. 3. Dependence of pre-edge peak area of W L 1 -edge XANES on energy gap of split W L 3 -edge white line of reference samples; (red solid circle): reference samples ( (a) Ba 2 NiWO 6 , (b) Cr 2 WO 6 , (c) (NH 4 ) 10 W 12 O 41 • 5H 2 O, (d) WO 3 , (e) H 3 PW 12 O 40 • 13H 2 O, (f) Sc 2 W 3 O 12 and (g) Na 2 WO 4 ), (blue solid square): x HPA/TiO 2 samples ( x =2, 4, 12, 20, 40, 60). 139 7 6 5 4 3 2 1 6 5 4 3 2 1 (a) (b) (d) (c) (e) (f) (g) 4 2 12 20 40 60 Energy Gap of W L 3 -edge White Line (eV) Pre-edge Peak Area of W L 1 -edge XANES (eV) W L 1 -edge XANES W L 3 -edge white line pre-edge peak Second derivative Energy gap
home
- JP
- EN