-X-Ray Photoelectron Diﬀraction Study on the Surface and Interface Structure of VO 2 /TiO 2 (110) Model Catalyst

We have studied the atomic structure of the surface and interface of VO 2 /TiO 2 (110) model catalyst by X-ray photoelectron diﬀraction (XPED). In this study, vanadium was deposited on the TiO 2 (110) surface in the oxygen atmosphere. We characterized the VO 2 ultra-thin ﬁlm grown on the TiO 2 (110) surface by low energy electron diﬀraction (LEED), X-ray photoelectron spectroscopy (XPS), and XPED. As a result, VO 2 ultra-thin ﬁlm grown epitaxially on TiO 2 (110) was determined by using the multiple scattering cluster model with spherical wave (MSC-SW). Finally, we carried out the quantitative analysis by using the reliability factor R (R-factor) method and clariﬁed the surface and interface structures of VO 2 /TiO 2 (110). [DOI: 10.1380/ejssnt.2008.272]


I. INTRODUCTION
Vanadium oxides supported on TiO 2 surface are highly active catalysts for the selective catalytic reduction (SCR) of NO x with NH 3 to N 2 [1,2]. An important and interesting feature of supported vanadium oxide catalysts is that few-layer structures of vanadium oxides play a property of catalyst. It is reported that the monolayer structure is the most active. However, its catalytic mechanisms are still unresolved on atomic scale. It is extremely important to clarify the correlation between the atomic structure and the catalytic activity. Therefore, we measured VO 2 /TiO 2 (110) model catalyst by X-ray photoelectron diffraction (XPED) excited by the Al-Kα line (hα = 1486.6 eV). XPED is a powerful tool used in thin-film materials research because forward-scattering peaks along interatomic axes between photoelectron emitters and their neighboring atoms can be used effectively for analyzing * This paper was presented at International Symposium on Surface Science and Nanotechnology (ISSS-5), Waseda University, Japan, 9-13 November, 2008. † Corresponding author: mnojima@rs.noda.tus.ac.jp simultaneously each structure of a thin film: an interface and a substrate. Finally, the surface and interface structures of VO 2 /TiO 2 (110) model catalyst were clarified by using the multiple scattering cluster model with spherical wave (MSC-SW) method and the R-factor.

II. EXPERIMENTAL
All measurements were performed in ultra-high-vacuum (the base pressure was 10 −8 Pa). The TiO 2 (110) crystal surface was cleaned by cycles of Ar + sputtering at 600 V, followed by annealing at 850 K in O 2 at 1.5 × 10 −4 Pa. 99.9% pure vanadium was deposited by molecular beam epitaxy (MBE) in low pressure O 2 atmosphere. Vanadium depositions were performed in O 2 at 2.0 × 10 −4 Pa at a temperature of 373 K. We characterized the 3.0 ML VO 2 ultra-thin film grown on the TiO 2 (110) surface by low energy electron diffraction (LEED), X-ray photoelectron spectroscopy (XPS), and XPED. In the XPED measurements, we used the strong Al-Kα line at 4 kW (20 kV, 200 mA). We selected an angular resolution of ±2.0 • as the measurement conditions for V 2p photoelectrons to obtain more photoelectron intensities. In the two-dimensional XPED measurements, we chose the pho- toelectrons of Ti 2p 3/2 (E b = 454.1 eV), O 1s (E b = 531.0 eV), and V 2p 3/2 (E b = 512.2 eV) that were emitted from the 3.0 ML VO 2 /TiO 2 (110) surface. The measurement range of the polar angle (θ) was selected as θ = 0 • -76 • to the normal in steps of 2 • . The measurement range of the azimuthal angle (φ) was selected as φ = 0 • -90 • in steps of 2 • . The theoretical XPED patterns were calculated by using the MSC-SW method. The algorithm of the MSC-SW method is based on separable Green's function approach described by Rehr and Albers [3]. Finally, we carried out the quantitative analysis by using the R-factor method and clarified the surface and interface structures of VO 2 /TiO 2 (110). Figure 1 shows the XPS spectra of the clean TiO 2 (110) surface and two vanadium oxides ultra-thin films (1.5 and 3.0 ML thick, respectively) deposited on the TiO 2 (110) surface. In our previous study, we reported about vanadium oxides ultra-thin film for 1.5 ML [4]. The thickness of the vanadium oxides ultra-thin film was calculated from V 2p and Ti 2p photoelectron intensities of the XPS spectra. In order to reduce the forward-scattering effect, both V 2p and Ti 2p photoelectron intensities were averaged at several azimuthal angles. The thickness of the film can be represented as follows.

III. RESULTS AND DISCUSSIONS
where K is the instrumental function; n, the atomic percentage; σ, the total photoionization cross section; λ, the inelastic mean free path (IMFP); and I, the intensity of X-ray. The measurement conditions selected were an energy resolution of 0.05 eV. As a result, the O 1s satellite peaks were centered at 520.6 and 518.6 eV, which result from the presence of the Kα 3 and Kα 4 X-rays in the non-monochromatic X-ray source. In the XPS spectra of 3.0 ML VO 2 /TiO 2 (110), the O 1s satellite peaks were completely overlapped with the V 2p photoelectron peaks because the photoelectron intensities of O 1s satellite peaks were very low. In XPS spectra of 1.5 and 3.0 ML VO 2 /TiO 2 (110), the V 2p 3/2 peak was centered at 515.6 and 515.8 eV, which indicates that the vanadium cations were in the +4 oxidation state (515.7 ± 0.3 eV). Figure 2 shows the LEED patterns of the clean TiO 2 (110) surface and two vanadium oxides ultra-thin films (1.5 and 3.0 ML thick, respectively) deposited on the TiO 2 (110) surface. A clean (1 × 1) LEED pattern was observed on the TiO 2 (110) surface. The (1×1) LEED pattern typical of the substrate is retained by the overlayer up to 3.0 ML. Consequently, the vanadium oxides were considered to have the VO 2 structure. The more the vanadium oxides grew, the more the intensity of background increased. The signal-to-background ratio in Figs. 2(b)-2(c) indicates the quality of surface structure. Figure 3 shows the two-dimensional XPED patterns of O 1s and Ti 2p photoelectrons from clean and ordered TiO 2 (110), V 2p photoelectron from 1.5 ML VO 2 /TiO 2 (110), and O 1s, Ti 2p, and V 2p photoelectrons from 3.0 ML VO 2 /TiO 2 (110) excited by Al-Kα line. In the XPED method, the photoelectron intensities of a specific peak of the element are detected by varying the polar and azimuthal angles of the sample, and the brightness in the XPED patterns shows the photoelectron intensities at each measurement angle [5,6]. However, the V 2p photoelectron peak overlaps with the O 1s peak excited by the Al-Kα 3,4 line in the XPED and XPS measurement of VO 2 /TiO 2 (110). Therefore, the structural information of vanadium oxides was not obtained with the usual measurement conditions. In order to resolve this problem, we obtained pure XPS spectra and XPED patterns of V 2p photoelectrons by subtracting the O 1s satellite peak [4]. The center of the XPED pattern corresponds to θ = 0 • and the periphery corresponds to θ = 76 • . In the XPED measurement of cleaned TiO 2 (110), the forwardscattering peaks of Ti 2p XPED pattern at θ = 27 • , 45 • , 60 • , 65 • were not shifted after deposition of 3.0 ML VO 2 . Further, Kikuchi-like patterns on O 1s XPED pattern also remained the same after depositing 3.0 ML VO 2 . The forward-scattering peaks and Kikuchi-like patterns are characteristics of the bulk structure [7][8][9][10]. In the XPED measurement of 1.5 ML VO 2 /TiO 2 (110), the clear forward scattering peaks were acquired on the twodimensional XPED pattern of V 2p. And, the Kikuchilike patterns on the two-dimensional XPED pattern of O 1s were clearly acquired. Moreover, clean (1 × 1) LEED pattern was observed in Fig. 2(b). Therefore, it is considered that the 1.5 ML VO 2 structure is highly ordered. On the other hand, in the XPED measurement of 3.0 ML VO 2 /TiO 2 (110), two-dimensional XPED pattern of V 2p was not clear at high polar angle. For example, forward scattering peak at θ = 65 • in Fig. 3(f) is not clear. In contrast, forward scattering peak at low polar angle is very clear. In fact, it is considered that the surface structure of 3.0 ML VO 2 is not highly ordered compared with the well ordered bulk structure. Also the disordered surface of 3.0 ML VO 2 makes the spots more diffuse and the intensity of background increase in LEED pattern in Fig. 2(c). Figure 4 shows the theoretical XPED patterns of O 1s and Ti 2p from clean and ordered TiO 2 (110), V 2p photoelectron from 1.5 ML VO 2 /TiO 2 (110), and O 1s, Ti 2p, and V 2p photoelectrons from 3.0 ML VO 2 /TiO 2 (110) obtained by using the structural model by the MSC-SW algorithm. The 1.0 ML VO 2 consists of six atomic layers and the vanadium atoms lie in two of the six layers. Therefore, 1.5 and 3.0 ML VO 2 contains three and six layers in which the vanadium atoms exist. On comparing the experimental patterns with the theoretical patterns, it was observed that the positions of the forward-scattering peaks and the Kikuchi-like patterns were agreed well with each other. Figure 5 shows the structural model of VO 2 /TiO 2 (110) obtained from these discussions. A quantitative analysis was carried out by using the R-factor method [4]. Each of Parameter b is the longer lattice parameter. Parameter c is the length between V atoms of the lowest layer of VO2 ultrathin film and Ti atoms in the highest layer of TiO2(110). In short, parameter c is the interfacial distance. a, b, and c in Fig. 5 is dominant parameter to determine the model structure. Each parameter was optimized in steps of 0.01Å. In literature, parameters on TiO 2 bulk and VO 2 bulk are described as Table I [ [11][12][13]. As a result, R-factor suggested its minimum to a model (a = 2.93Å and b = 6.27Å) in the 3.0 ML VO 2 ultra-thin film. There are mismatches between the lattice parameters of 3.0 ML VO 2 surface and those of TiO 2 bulk in Table I: ∆a/a = −1.0%, ∆b/b = −3.5%. Also, there are mismatches between the lattice parameters of 3.0 ML VO 2 surface and those of VO 2 bulk: ∆a/a = +2.1% and ∆b/b = −2.2%. It is difficult to determine that the true structure of 3.0 ML VO 2 is VO 2 or TiO 2 structure. In contrast, there are few mismatches between the lattice parameters of 1.5 ML VO 2 surface and those of TiO 2 bulk: ∆a/a = −1.0%, ∆b/b = +0.6%. Also, there are mismatches between the lattice parameters of 1.5 ML VO 2 surface and those of VO 2 bulk: ∆a/a = +2.1% and ∆b/b = +2.0% [4]. Accordingly, the lattice parameters of 1.5 ML VO 2 are more corresponding to those of TiO 2 bulk than those of VO 2 bulk. On the system of 3.0 ML VO 2 /TiO 2 , R-factor was converged to minimum value as parameter c was 2.91Å. About the parameter c, there are large mismatch between the cases on 3.0 ML VO 2 /TiO 2 (110) and TiO 2 bulk: ∆c/c = −10.2%.
And, there are also large mismatch between the 3.0 ML VO 2 /TiO 2 (110) and VO 2 bulk: ∆c/c = −9.6%. These value indicate that the interface between the 3.0 ML VO 2 ultra-thin film and TiO 2 (110) surface should be quite shorter than interlayer distance of VO 2 bulk and TiO 2 bulk. In contrast, R-factor was converged to minimum value as parameter c of 1.5 ML VO 2 /TiO 2 was 3.50Å [4]. The interfacial distance of 3.0 ML VO 2 /TiO 2 is relatively smaller than those of 1.5 ML VO 2 /TiO 2 (110). For understanding these results, the steps on the surface of TiO 2 (110) must be considered. In XPED measurement, we acquire average information on the atomic structure. A lot of steps on the TiO 2 (110) first surface are error factors of interfacial analysis for VO 2 /TiO 2 . It is considered that TiO 2 (110) surface was roughened up by intensive Ar + sputtering. In our previous study of 1.5 ML VO 2 /TiO 2 (110), the TiO 2 (110) crystal surface was cleaned by Ar+ sputtering at 1.5 kV [4]. In contrast, the accelerating voltage was 600 V in this study. It is considered that TiO 2 (110) surface was roughened up by Ar + sputtering at high accelerating voltage in our previous study. It is reported that a height of single step on TiO 2 (110) surface is approximately 3.25Å [14]. Therefore, the interfacial distance might be estimated longer than a true distance. Consequently, it is considered that the interfacial distance between 3.0 ML VO 2 ultra-thin film and TiO 2 (110) surface is more reliable. In addition, the interfacial distance of VO 2 /TiO 2 (110) is smaller than the interlayer distance of VO 2 bulk and TiO 2 bulk.

IV. CONCLUSIONS
As a result of LEED and two dimensional XPED measurements, it is considered that 3.0 ML VO 2 surface structure is not highly ordered compared with the well ordered bulk structure. In XPS spectra, the V 2p 3/2 peak was centered at 515.6 and 515.8 eV, which indicates that the vanadium cations were in the +4 oxidation state (515.7 ± 0.3 eV). In R-factor analysis, a lot of steps on the TiO 2 (110) first surface might act as error factors of interfacial analysis. Finally, it is considered that the interfacial distance between VO 2 ultra-thin film and TiO 2 (110) surface is shorter than interlayer distance of VO 2 bulk and TiO 2 bulk.