Preparation of Visible-light Responsive Rutile-TiO2 (110) Wafer with Well-defined Surface by Chromium and Antimony Codoping

Transition-metal doping for titanium dioxide (TiO 2 ) is attract-ing attention for the study of visible-light responsive photocatalyst. Its photocatalytic properties were investigated via various spectroscopic approaches, though surface studies had not yet progressed owing to the difficulty in obtaining its well-defined surface. In this report, we propose that a well-defined crystalline TiO 2 (110) surface may be obtained by the codoping of chromium (Cr) and antimony (Sb) with commercially available wafers. Cr and Sb are codoped by a solid-state reaction of TiO 2 (110) wafer and dopant powder. The prepared wafer exhibited visible-light responsivity on absorption below wavelengths of 600 nm. The surface morphology characterization, performed by atomic force microscopy (AFM) revealed that the Cr and Sb codoped TiO 2 (110) surface has a well crystallized step-terrace structure that is atomically flat, while monodoped TiO 2 (110) surface does not. The codoping of Cr and Sb with TiO 2 (110) wafer should contributes towards retaining the stable rutile-TiO 2 lattice structure and produces a well-defined TiO 2 (110) surface structure with visible-light responsive characteristics.


I. INTRODUCTION
Titanium dioxide doped with certain transition metals exhibit attractive properties and potentialities for making functional materials such as gas sensors [1], catalysts [2−4], electrodes [5], and fine ceramics [6]. Especially, it has been discovered that doping transition metals into titanium dioxides increases the visible-light absorption, and this is promising in manufacturing practical photocatalysts that operate under sun-light irradiation. Therefore, titanium dioxides doped with transition metals have been extensively investigated via various viewpoints such as spectroscopic study [7] and calculating approach [8].
Conventionally, sol-gel, wet impregnation, hydrothermal synthesis, etc. were widely used for introducing transition metals into TiO2 and the doped materials prepared by these methods were investigated as visible-light responsive photocatalysts. However, as the materials that are prepared using conventional doping methods are mostly powders or have polycrystalline forms, it became difficult to perform fine surface studies. As the photocatalytic reaction is triggered in the bulk phase as well as on the surface, the surface observation of the transition metal doped TiO2 is important for photocatalytic studies. It will provide us with considerable information for elucidating photocatalyst properties, such as the surface structure [9], desorption and adsorption mechanism of reactants [10,11], etc. Therefore, it is highly desirable to employ doped TiO2 wafer surfaces with well-defined structures to perform scientific surface research.
To address this issue, we proposed the codoping of chromium (Cr) and antimony (Sb) with TiO2(110) wafer as a promising alternative for acquiring a well-defined TiO2(110) surface with visible-light response. We used commercial rutile-TiO2(110) wafer and prepared monodoped TiO2(110) wafer with Cr or Sb, and further codoped TiO2(110) wafer samples via a solid-state reaction, which was conventionally used for making ceramic powder and several doping materials that exhibit high photocatalytic activities are produced via this method [12,13]. Note that rutile TiO2 codoped with Cr and Sb functions as a visible-light responsive photocatalyst [14] and its photocatalytic properties have been widely investigated [14−16]. The prepared doped samples were assessed using optical absorption and contact-mode atomic force microscopy (AFM) imaging to reveal the dependency of the preparing conditions on their surface structure.

II. EXPERIMENTAL SECTION
Rutile TiO2(110) wafer, purchased from SHINKOHSHA, Co. Ltd., cut into 10 mm × 2 mm × 0.5 mm sized samples were codoped with Cr and Sb via conventional solid-state reactions. The dopant materials, Sb2O3 and Cr2O3 powders, (99.9%, purchased from Wako-chemical, co. Ltd.) were mixed in a molar ratio of 1 : 1, and placed on the polished surface of the TiO2(110) wafer. Next, calcination at 1270 K in air for 22 h using alumina crucibles (99.7% Al2O3) was performed. The Al2O3 crucible was sealed to prevent escape of evaporated Sb oxide. During the calcination process, the dopant diffuses with the wafer surface. The detailed procedure of the sample preparation is shown in Figure S1 (Supplementary Material). After calcination, TiO2(110) wafer turns from white to yellow in color, indicating the acquired visible-light absorption property, which was further confirmed via UV-VIS diffuse reflection spectroscopy. To measure the spectra, we used integrating-sphere measurement, and the doped wafers were attached to the sample holder with their polished surfaces upwards. Surface observation was performed by scanning probe microscopy (JSPM-4200, JEOL. Inc.) along with contact-mode AFM in air at room temperature. The sample before calcination (upper column) exhibited white color as pristine TiO2(110) wafer (a non-doped sample), while the sample prepared with pure Cr2O3 powder (middle column) exhibited dark color. In the powder characterization, the monodoping of Cr2O3 (Cr 3+ ) with TiO2 generate oxygen vacancies or the higher oxide state of chromium cations (Cr 6+ ), owing to the compensation of charge neutralities of Cr 3+ and Ti 4+ substitution. Therefore, monodoped TiO2 (Cr-TiO2) has various impurity states in the bandgap of TiO2, leading to the absorption of photons with various energies, and the color turns from white to black [14]. The features of the prepared Cr-TiO2(110) wafer sample corresponded well corresponding with these findings. On the other hand, TiO2(110) wafer codoped with Cr/Sb (Cr2O3 : Sb2O3 = 1 : 1 mixed) powder, as shown in the lower column showed clear yellow color. The codoping of Cr and Sb into the TiO2 lattice, where Cr 3+ and Sb 5+ were substituted with 2Ti 4+ maintained the charge neutrality. No oxygen vacancies were generated, and doped chromium had a fixed oxide state as Cr 3+ . Therefore, the codoped TiO2 has yellow color as its photon absorption is exactly derived from the fixed Cr 3+ → Ti 4+ charge transfer [14]. Figure 1(b) shows the UV-VIS diffuse reflection spectra of the doped TiO2(110) wafers. Codoped TiO2 sample displayed absorption bands due to the Cr 3+ → Ti 4+ charge transfer transition in the visible-light region (λ < 600 nm) in contrast to pristine TiO2. Monodoped TiO2 with only Cr (Cr-TiO2) displayed wide absorption in the visible-light region as reported in the previous study pertaining to powders [14]. From these results, it is apparent that we would prepare visible-light responsive TiO2(110) wafers with transition metal doping.

III. RESULTS AND DISCUSSION
In the doped TiO2(110) wafers, the polished specular surface was retained except in the lower region of the powder, and it allowed us to acquire meaningful AFM images on their surfaces. Figure 2 presents the contact-mode AFM images of doped TiO2(110) wafers. Figure 2(a) shows the surface structure of pristine TiO2(110) treated with the same calcination conditions of doped wafer (heated 1270 K in air for 22 h) without the dopant powder. One can observe some terraces, and the step structure is dominated by step edges running parallel to the ❬001❭ direction [9], similar to the typical well-defined structure of the non-doped TiO2(110) surface. In contrast, the surface of the Cr doped wafer as shown in Figure 2(b) is not similar to the well-defined TiO2(110) surface structure. As explained above, when Cr 3+ ions are partly substituted for Ti 4+ ions in TiO2, oxygen vacancies and/or Cr 6+ ions should be formed to keep the charge balance, and they may cause the disarrangement of the rutile structure [14]. These structural deteriorations of the Cr-TiO2(110) surface were observed at any part and on various imaging scales, as shown in Figure S2 (Supplementary Material). Figure 2(c) shows the surface structure of wafer that is doped with only Sb. Several streaky structures that extend in the ❬001❭ direction were observed. When only Sb is doped with TiO2, Sb 3+ and Sb 5+ were produced to maintain the charge neutrality, and they formed double oxides (Sb 3+ Sb 5+ O4) [17]. We assume that the streaky structures are made by these double oxides extending along the ❬001❭ direction of the rutile TiO2(110) surface. On the other hand, as shown in Figure 2(d), the result of the TiO2(110) wafer codoped with Cr and Sb is different from any case of the monodoped TiO2(110) wafer surface. The surface resembles the pristine TiO2(110) well-defined structure. Figure 2(e) shows the surface profile along with the A−B line of Figure  2(d). The step height is approximately 0.3 nm and this is well in agreement with the height of the single step of the (110) surface of the rutile TiO2. These results indicate that the TiO2(110) wafer codoped with Cr and Sb resembles the rutile type structure, and the reason behind this may be explained in terms of crystallography. When Cr and Sb are codoped with TiO2, it is known that Cr 3+ and Sb 5+ produce a double oxide (CrSbO4). Furthermore, CrSbO4 resembles the rutile type structure (a = b = 4.59 Å, c = 3.05 Å) similar to the TiO2 crystal (a = b = 4.59 Å, c = 2.953 Å) [18]. Eventually, Cr and Sb were codoped without disturbing the TiO2 rutile structure.
It is known that the photocatalytic activities of TiO2 codoped with Cr and Sb are remarkably increased in comparison with only Cr doped TiO2 [14] and the reason for this is explained in terms of photo-induced dynamics. However, photocatalytic activities of doped TiO2 are controlled by not only the photo-dynamic processes in bulk, but by the condi-tion of their surface structures as well, as the catalyst reaction occurs on the crystal surface, and our result reflects this point. The surface of the codoped TiO2(110) wafer is considerably more crystallized than that of the only Cr doped TiO2, and we suggest that the differences in their surface structures is influenced by the differences in their catalytic activities.
We prepared TiO2(110) wafer samples codoped with Cr and Sb under various calcination times (for 12 h, 22 h, and 30 h). AFM images of prepared sample surfaces were summarized in Figure 3. In these samples, the surface observation was performed on three parts of the wafer surface as shown in A−C of Figure 1(a). In the prepared sample calcinations for 12 h, we can observe some steps and terraces on the surface at position A. This structure resembles the surface structure of codoped TiO2(110) wafer as in Figure 2(d). Although the same sample, several streaky structures were observed at position C and this surface structure is similar to that observed on the wafer only doped with Sb as shown in Figure 2(c). These characteristic surface structures were further observed at various imaging scales, as shown in Figure S3 (Supplementary Material). From this, we estimate that Sb was excessively doped compared to Cr at the far position from dopant powder source, and it is indicated that Sb diffuses more quickly than Cr in spite of its atomic weight. This can be explained by the volatilization of Sb oxide. During the calcination process, a part of the Sb oxide will be volatilizing, and spreading on the wafer surface as vapor phase. Note that the vaporized Sb-oxide should fill the sealed Al2O3 crucible. Therefore, the Sb dopant is contacted with the TiO2 wafer surface to a greater extent than the Cr dopant and will be quickly doped on any part of the TiO2 wafer surface via vapor-solid interface reaction. Once the Sb was doped into TiO2 lattice, formed Sb-TiO2 should be stable and Sb would not evaporate from the rutile lattice.
At the same observation position, the surface structures vary with respect to the calcination times.
Step and terrace structures of the rutile (110) are more reproduced by longer calcinations times. When the sample is calcined for 30 h, its surface structure is uniform at any observation position as the Cr dopant is diffused adequately to reconstruct the rutile structure. Eventually, the surface structure of the codoped TiO2(110) wafer depends on the diffusing process of Cr, and a longer calcinations time allows us to prepare the well codoped TiO2(110) surface with a clear step and terrace structure. It should be suitable for basic surface studies, will aid in reaching fine atomic scale investigations [19].

IV. CONCLUSIONS
We prepared a visible-light responsive TiO2(110) wafer with a well-defined surface structure along with an atomically flat step and terrace, via the codoping of Cr and Sb through a conventional solid-state reaction. The prepared TiO2(110) wafer codoped with Cr and Sb exhibited visible-light absorption that arises from the Cr 3+ → Ti 4+ charge transfer transition. The prepared doped TiO2(110) wafer retains its polished specular surface except for the region beneath the dopant powder and was appropriate for AFM imaging. The surface structure of the monodoped Cr-TiO2 (110) wafer was clearly varied from the pristine TiO2(110) surface. However, the surface structure of rutile TiO2(110) is maintained by codoping with Cr and Sb. We estimated that the diffusion of the Cr dopant is rate-determining during the codoping process owing to the volatility of the Sb oxides. We prepared CrSb-TiO2(110) wafer with a well-defined step and terrace surface structure via adequate codoping. The wafer will be suitable for the basic surface study of visible-light responsive photocatalysts.