A Novel One-Step Hydrothermal Preparation of Ru/SnxTi1-xO2 Diesel Oxidation Catalysts and its Low-Temperature Performance.

The rutile SnxTi1−xO2 (x = 0, 0.33, 0.5, 0.67, 1) solid solution was synthesized by a one-step hydrothermal method, in which tetrabutyl titanate and Tin (IV) chloride pentahydrate were used as raw materials. A series of Ru/SnxTi1−xO2 were then prepared by the impregnation process in RuCl3 to investigate the performance and stability of CO and C3H8 oxidation. These catalysts were characterized through XRD, N2 adsorption-desorption, FT-IR, TEM, XPS, H2-TPR, and O2-TPD techniques. The effect of Sn/Ti molar ratio and hydrothermal condition on the low-temperature catalytic oxidized performance and stability of Ru/SnxTi1−xO2 were investigated. The results indicated that Ru/Sn0.67Ti0.33O2 catalyst showed an excellent activity and stability at low temperatures. The CO conversion reached 50% at 180 °C and 90% at 240 °C. Besides, the C3H8 conversion reached 50% at 320 °C, the complete conversion of C3H8 realized at 500 °C, and no deactivation occurs after 12 h of catalytic reaction. The excellent low-temperature activity and stability of the Ru/Sn0.67Ti0.33O2 were attributed to the following factors. Firstly, XRD results showed that Sn4+ was successfully introduced into the lattice of TiO2 to replace Ti4+ forming a homogeneous solid solution (containing –Sn4+–O–Ti4+– species), which was consistent with TEM and N2 adsorption-desorption results. The introduction of Sn could suppress the growth of anatase crystal and promote the formation of rutile phase, and this phase transition was helpful to improve the low-temperature activity of the catalysts. Secondly, TEM images showed that ultrafine Ru nanoparticles (~ 5 nm) were dispersed on Sn0.67Ti0.33O2 support, suggesting that the formation of SnxTi1−xO2 solid solution was beneficial to the dispersion of Ru particles.


Background
Diesel engines are widely applied in the field of transportation, mining, and engineering machinery due to these advantages of low fuel consumption, high thermal efficiency, and good stability [1]. However, carbon monoxide (CO), unburned hydrocarbons (HCs), various oxides of nitrogen (NO x ), and the particulate matter (PM) in diesel vehicle exhaust have caused a serious threat to ecological environment and human health [2,3]. Furthermore, stringent environmental laws and regulations have driven recent advances in diesel emission control technologies. An integrated exhaust after-treatment system consisting of diesel oxidation catalyst (DOC), selective catalytic reduction (SCR), and catalyzed diesel particulate filter (DPF) has been widely used to purify diesel exhausts. The functions of the DOC in the after-treatment system are converting CO, HCs, and NO into CO 2 , H 2 O, and NO 2 , the NO 2 being used as raw material for subsequent de-NO x reaction to promote SCR reaction. In addition, it could also oxidize the soluble organic fraction (SOF) to decrease PM emissions. HCs excessive emission will be caused owing to the incomplete combustion of HCs during the cold start of diesel vehicles. Therefore, the catalysts need to ignite rapidly at low temperatures [4]. Presently, noble metal catalysts (such as Pt, Pd, and Rh) supported on carbon materials or oxides (such as TiO 2 , Al 2 O 3 , CeO 2 , and ZrO 2 ) are commercialized diesel oxidation catalysts with good performance for CO, NO, and HCs catalytic oxidation. However, there are drawbacks to commercialized catalysts, such as poor thermal stability, strong self-inhibition by CO, and high cost [5].
Ru and RuO x catalysts are widely applied in oxidizing CO [6], methane [7], and chlorobenzene [8]. Importantly, Ru catalysts have excellent low-temperature activity and poison resistance [8][9][10][11]. But Ru and RuO x are easily sintered, resulting in active sites' exposure decreases [12]. Therefore, Ru catalysts should be supported on a carrier to prevent their sintering and improve catalytic activity. TiO 2 has been widely used to purify diesel exhausts. RuO x and rutile phase TiO 2 have a similar lattice constant, and the rutile TiO 2 in Ru/TiO 2 catalysts plays an important role in stabilizing RuO x particles during calcination process in comparison with anatase-supported RuO x catalysts. Therefore, RuO x can be highly dispersed on the surface of TiO 2 . Furthermore, there is a synergistic effect between RuO x and TiO 2 , which is beneficial to improve the redox ability of Ru/TiO 2 [13][14][15][16][17][18]. In order to further improve the thermal stability, dispersion of active components, and transformation of anatase to rutile phase, many studies have introduced Sn 4+ into TiO 2 to form Sn x Ti 1−x O 2 solid solution. Huang et al. [16] found that the introduction of Sn 4+ into TiO 2 lattice could improve the stability of the CuO/Ti x Sn 1−x O 2 catalysts and dispersion of CuO. Bai et al . [17] indicated Sn 4+ significantly improved the thermal stability of TiO 2 . Mehraz et al. [18] found doping Sn 4+ promoted the phase transition of TiO 2 from anatase to rutile.
Previous researches have focused on the preparation of diesel oxidation catalysts by co-precipitation method, solgel method, and solid-phase reaction [5,6,15,19,20]. Yang et al. [19] prepared the Pt/TiO 2 catalysts via the coprecipitation method and found that the conversion of CO and C 3 H 6 only reaches 50% at 232°C. Li et al. [15] synthesized TiO 2 -SnO 2 nano-composite by the sol-gel method and suggested that the conversion of TiO 2 -SnO 2 to CO was 90% at 260°C. Sharif et al. [6] 2 to CO was only 82% at 240°C due to lower dispersion of Ru. Therefore, there are critical challenges that remain in the lowtemperature activity of diesel oxidation catalysts and a lot of efforts are still needed to remove CO and HCs caused in the diesel cold start. Furthermore, the current research [8,16,19,21,22] is mainly focused on the preparation of DOC catalysts by co-precipitation and sol-gel methods, which has a small grain size, but the samples have poor crystallinity and multiple crystal phases; furthermore, the subsequent heat treatment process of mixture by coprecipitation method is required. Hydrothermal treatment is adopted in the preparation process to avoid the traditionally followed calcination processes and the formation of hard aggregation of the catalysts, which could improve lowtemperature catalytic activity [23]. However, there is a lack of systematic and comprehensive studies on the one-step hydrothermal method [24,25].
Therefore, we reported that RuO x particles supported on the Sn 4+ -modified TiO 2 by the one-step hydrothermal method were excellent CO and HC oxidation catalysts with promising low-temperature activity and stability. A series of Sn x Ti 1−x O 2 (x = 0, 0.33, 0.5, 0.67, 1) solid solution were prepared by the one-step hydrothermal method. Ru/Sn x Ti 1−x O 2 were then prepared by impregnation of the Sn x Ti 1−x O 2 with RuCl 3 to oxidize CO and C 3 H 8 . The effect of hydrothermal temperatures, hydrothermal time, calcination temperatures, and the molar ratios of Sn/Ti of Ru/Sn x Ti 1−x O 2 catalysts were investigated in order to improve low-temperature activity and stability.

Catalytical Performance
The activities of the catalysts were evaluated on a fixed bed quartz reactor with an electric heater. The simulative reactant gases contained a mixture of 3000 ppm CO, 600 ppm C 3 H 8 , 600 ppm NO, 50 ppm SO 2 , 7% O 2 , and N 2 balance at a gas space velocity of 60,000 mL g −1 h −1 . The gas flow rate was regulated by mass flow controllers. The temperature of fixed bed was tested by a 0.5-mm Kthermocouple which was placed in the middle of the center channels. The outlet CO and C 3 H 8 were measured by a KM9106 flue gas analyzer (Kane International Limited, Britain). The conversion (X) of CO and C 3 H 8 was calculated using the following equation: where c in is the initial concentration of CO or C 3 H 8 and c out is the instantaneous of CO or C 3 H 8 at the reaction temperature; T 50 is denoted as the low-temperature catalytic activity index.
Catalyst Characterization X-ray diffraction (XRD) patterns of the samples were performed by power X-ray diffraction on a BRUKER D8 ADVANCE diffractometer equipped with a hightemperature chamber using Cu Kα radiation (0.15418 nm). The X-ray tube was operated at a source power of 40 kV × 40 mA.
The Brunauer-Emmett-Teller (BET) surface areas were tested by nitrogen adsorption at 77 K on a Micromeritics ASAP2020 adsorption apparatus; the specific surface area and pore distribution were calculated by the BET and BJH methods, respectively. These samples were degassed under vacuum at 300°C for 4 h before each analysis.
Fourier transform infrared (FT-IR) spectroscopy was examined using a Nicolet is5 spectrometer at a spectral resolution of 4.0 cm −1 . The powders were pressed into a self-supporting wafer (about 15 mg, 12 mm diameter). The wafer was pretreated with N 2 at 300°C for 1 h. After cooling to ambient temperature, the spectrum of samples was recorded.
Transmission electron microscopy (TEM) images of these samples were obtained by a Tecnai G2 F20 instrument at an acceleration voltage of 200 kV. The samples were ground, dispersed in ethanol, and deposited on carbon-coated copper grids prior to observation.
X-ray photoelectron spectroscopy (XPS) analysis was performed on a ESCALAB250Xi spectrometer, using monochromatic Al Kα radiation (1486.6 eV) at an accelerating power of 15 kW. The obtained sample spectra were corrected using C1s (284.6 eV) as the internal reference standard.
H 2 -temperature-programmed reduction (H 2 -TPR) experiments were performed in a quartz reactor connected to a thermal conductivity detector (TCD) with H 2 (6.9% vol. %)-Ar mixture (30 mL/min) as reductant. Prior to the reaction, the sample (50 mg) was pretreated in N 2 at 300°C for 1 h and then cooled to room temperature. TPR started from room temperature to target temperature at a rate of 10°C/min. Temperature-programmed oxygen desorption (O 2 -TPD) experiments were carried out using the same device as H 2 -TPR. The spent catalyst (50 mg) was pretreated at 300°C for 1 h under flowing Ar at 30 mL/min. Then, oxygen adsorption was conducted under an O 2 -Ar mixture (20% O 2 vol. %) at 500°C for 0.5 h. After cooling to room temperature, the system was purged in Ar (30 mL/min) for 1 h. After the treatment, the temperature was raised to target temperature (10°C/min).
In situ infrared spectroscopy (IR) of CO adsorption was collected on a Nicolet 5700 FT-IR spectrometer at a spectral resolution of 4.0 cm −1 . CO adsorption was performed by exposing a self-supporting wafer of catalyst (about 15 mg) and mounted in a commercial controlled environment chamber (HTC-3). The samples were exposed to a controlled stream of CO-Ar (10% of CO by volume) at a rate of 5.0 mL/min for 40 min. The spectra were recorded at various target temperatures at a rate of 10°C/min from room temperature to 300°C.

Results and Discussion
Catalytic Activity and Stability Figure 1 shows the catalytic activities of CO and C 3 H 8 oxidation on the Ru/Sn x Ti 1−x O 2 catalysts under the optimal preparation conditions (Fig. S1, S2 and S3) of hydrothermal temperature at 180°C, hydrothermal time at 24 h, and calcination temperature at 400°C. It can be seen that catalytic performances of Ru/Sn x Ti 1−x O 2 catalysts increased firstly and then tended to be stabilized with the increase of reaction temperature. When the molar ratio of Sn/Ti is 2/1, the T 50 of Ru/Sn 0.67 Ti 0.33 O 2 to oxidize CO and C 3 H 8 is 180°C and 320°C, respectively, which is lower reaction temperature than other Sn/Ti molar ratios. The conversion of CO reached 90% at 240°C , and the complete conversion of C 3 H 8 could be achieved at 500°C on the Ru/Sn 0.67 Ti 0.33 O 2 catalyst. The catalytic performance for each sample was normalized with respect to the Ru atoms on the surface and expressed in terms of turnover frequency (TOF), as shown in Fig. 2 [27] reported that the conversion of Au/Nb 2 O 5 and Au/SiO 2 to CO was 55% and 38%, respectively, at 250°C. Compared with other literatures [27,28], when the molar ratio of Sn/Ti is 2/1, higher CO conversion can be achieved at lower temperature in this study. Okal et al. [29] reported that the T 50 of CH 4 oxidized by Ru/ZnAl 2 O 4 catalysts was 480, 500, and 540°C, when the loading of Ru was 0.5 wt.%, 1.0 wt.%, and 4.5 wt.%, respectively. Wilburn et al. [30] reported that the T 50 of CH 4 oxidation over 0.3Pd-0.7Pt/γ-Al 2 O 3 catalyst was 360°C. The catalytic activities of different catalysts for CO and C 3 H 8 oxidation are shown in Table S1 and Table S2. Complete transformation of C 3 H 8 can be achieved at a lower temperature in this work. The optimum molar ratio of Sn/Ti is 2/1, which is consistent with the activity of CO. From the above analysis, it can be concluded that the conversion of CO and C 3 H 8 is greatly affected by the molar ratio of Sn/Ti. When the molar ratio of Sn/Ti is 2/1, the T 50 of Ru/Sn 0.67 Ti 0.33 O 2 to CO and C 3 H 8 is 180°C and 320°C, respectively. When the reaction temperature is 240°C, the conversion of CO can reach 90% and the complete conversion of C 3 H 8 can be achieved when the reaction temperature is 500°C.
The stability of CO and C 3 H 8 was investigated in Fig. 3, under hydrothermal temperature at 180°C, hydrothermal time at 24 h, and calcination temperature at 400°C (Fig. S1, S2 and S3). The conversion of CO reached 90% at 240°C, and the complete conversion of C 3 H 8 could be achieved at 500°C. Interestingly, Ru/Sn 0.67 Ti 0.33 O 2 catalyst is basically inactivated after a 12 h catalytic reaction; however, the activity of Ru/TiO 2 and Ru/SnO 2 catalysts    (Table 1), respectively. A phase transition from anatase to rutile appeared with the introduction of Sn. The Ru diffraction peaks are not observed, indicating that Ru is highly dispersed on Sn x Ti 1 −x O 2 surface or beyond the XRD detection limitation [31]. Furthermore, the diffraction peaks of Sn x Ti 1−x O 2 and Ru/Sn x Ti 1−x O 2 move gradually to lower angles with increasing Sn content, suggesting that the interplanar spacing d increases according to the Bragg equation, 2d sinθ = nλ. This is consistent with the increase in tetragonal lattice parameters (a and c) in Table 1, which is attributed to the substitution of larger ionic radius Sn 4+ (0.071 nm) for Ti 4+ (0.068 nm). The results suggest the Sn 4+ has been successfully doped into the TiO 2 lattice to form a uniform (-Sn 4+ -O-Ti 4+ -) solid solution while maintaining the rutile phase structure, which is in agreement with some previous studies [5,18].
To determine the texture properties of samples, the N 2 adsorption-desorption technique was used. The N 2 adsorption-desorption isotherms and corresponding pore diameter distribution curves of these samples are shown in Fig. 5. The N 2 adsorption-desorption isotherms of SnO 2 distinctly belong to type II; others are classical type IV according to IUPAC classification and present a H2 complex hysteresis loop in a p/p 0 range of 0.4-0.95, which is a common feature of mesoporous material (Fig. 5a, c) [17,32]. The existence of these mesopores is an important reason for the large specific surface area of catalysts [33]. All of Sn x Ti 1−x O 2 supports and Ru/Sn x Ti 1−x O 2 catalysts exhibited a narrow distribution of small-sized pores (3-8 nm), especially the Sn 0.67 Ti 0.33 O 2 support and Ru/Sn 0.67 Ti 0.33 O 2 catalysts, with the pore diameter mainly uniformly distributed around 5 nm (Fig. 5b, d). This phenomenon suggested that an appropriate amount of Sn can weaken the diffusion coefficient of the catalytic surface and indirectly hinder the agglomeration of the crystallites [17].
The texture properties of Sn x Ti 1−x O 2 supports and Ru/Sn x Ti 1−x O 2 catalysts are listed in Table 1 Figure 6 shows the FT-IR spectra of Sn x Ti 1−x O 2 supports and Ru/Sn x Ti 1−x O 2 catalysts. All the samples present similar vibration peaks at analogous wavenumber positions. The adsorption at around 3223.68 cm −1 is due to surface hydroxyl groups neighboring oxygen vacancy sites [34,35]. The bands of 1501.86-1618.18 cm −1 belong to the angular vibration peak of water. The symmetrical stretching vibration peak of lattice oxygen appears at 1028.17 cm −1 . The band of 527.27-681.2 cm −1 may be attributed to the stretching vibration peak of TiO 2 or SnO 2 [34]. Compared with Sn x Ti 1−x O 2 supports, Ru/Sn x Ti 1−x O 2 spectrum broadens, indicating that the active component Ru and support Sn x Ti 1−x O 2 have some interaction, resulting in the surface defects of catalysts [36,37].

Morphology of Catalysts
Low-and high-resolution TEM, HRTEM images, and the particle size distribution of Ru/Sn x Ti 1−x O 2 are exhibited in Fig. 7. Based on the observation of the TEM images presented in Fig. 7a, d, g, j, and m, we find that all samples are composed of well-defined particles with irregular shapes and disordered mesoporous structure, which is formed by the agglomeration of the nanoparticles [38]. Furthermore, it can be seen that the Ru/ Sn 0.67 Ti 0.33 O 2 sample has the highest degree of agglomeration because of the smallest grain size among these samples. From the HRTEM images (Fig. 7b, e, h, k, n), there is only one kind of lattice fringes with 0.327 nm, which is compatible with (110) plane of these samples. Besides, we find that the lattice fringes of TiO 2 and SnO 2 are not observed, which is attributed to Sn 4+ having been successfully doped into the lattice of TiO 2 to form a homogeneous Sn x Ti 1−x O 2 solid solution [39].
The results are consistent with XRD. The Ru particle size distribution (Fig. 7c, f, i, l, o) shows that the approximate sizes of Ru particles ranged from 3 to 20 nm. The introduction of Sn 4+ could effectively decrease the sizes of Ru particles and achieve a higher dispersion on the Sn x Ti 1−x O 2 surface. Comparing with other samples, the Ru particle size distribution of Ru/Sn 0.5 Ti 0.5 O 2 sample was wider (< 13 nm), which may be caused by the interaction between (-Sn 4+ -O-Ti 4+ -) species and Ru [26]. The Ru/Sn 0.67 Ti 0.33 O 2 catalyst has better Ru dispersion and smaller particle size (5.49 nm) among all samples.

Surface Properties of Catalysts
To further determine the elementary states and surface composition, XPS analysis was carried out. Figure 8 shows the XPS spectra of Sn 3d, Ti 2p, O 1s, and Ru 3d   Table 3 that the Sn/Ti molar ratio by XPS is observed to be slightly higher than theoretical calculation, indicating that Sn is enriched on the surface of catalysts, which leads to more oxygen vacancies. Because the electronegativity of Sn (1.96) is larger than that of Ti (1.62), in other words, the electroncapturing ability of Sn is stronger than that of Ti, which causes the redox equilibrium (Sn 4+ +Ti 3+ → Sn δ+ +Ti 4+ ) shifting to right [32]. The high-resolution spectra of the O 1s ionization feature are numerically consistent with the Gaussian feature and deconvoluted into two peaks [5] Ru 3d spectra present Ru 4+ and lower value Ru δ+ . The signal of Ru 3d 5/2 is often used to analyze the charge state of the Ru species, since another Ru 3d 3/2 overlaps with C 1s at around 284.0 eV [40]. The binding energy of 282.0-283.5 eV is assigned to Ru 3d 5/2 , which corresponded to Ru 4+ . The lower binding energy at around 280.2-281.7 eV is attributed to lower state Ru δ+ , and the Ru δ+ relative ratio in Ru/Sn 0.67 Ti 0.33 O 2 reaches 53.9%, which is higher than other catalysts. It could be explained that the strong interaction between Sn 0.67 Ti 0.33 O 2 and Ru caused a larger amount of surface reactive oxygen species [26].
XPS and EDS analyses are performed to determine the surface and bulk composition of the samples as shown in Table 2. Surface and bulk Ru analysis shows that Ru/ Sn 0.67 Ti 0.33 O 2 has the highest surface Ru (0.69 wt.%) and bulk Ru (0.40 wt.%) among all the catalysts, indicating that the active component Ru is more evenly distributed on the Sn 0.67 Ti 0.33 O 2 support, and more Ru species enters the internal of Sn 0.67 Ti 0.33 O 2 to form a strong interaction.
In order to further investigate the reduction performance of the Ru/Sn x Ti 1−x O 2 catalysts, temperatureprogrammed reduction studies are performed (Fig. 9). The shapes of these H 2 -TPR profiles are almost identical. The reduction peaks of Ru/Sn x Ti 1−x O 2 are divided into two parts: the low-temperature reduction peaks 80-270°C are associated to the lower state Ru δ+ reduced from RuO 2 and a significant amount of Sn 4+ which could be reduced to lower valent Sn δ+ or can be attributed to the reduction of surface oxygen [41], while the high-temperature reduction peaks 600-640°C are associated to Sn 0 reduced from Sn δ+ or the reduction of bulk oxygen of catalysts [26,42], which is consistent with XPS results. The reduction temperature of Ru/Sn x Ti 1−x O 2 moves towards lower temperature, peaks broaden and H 2 consumption increase with the addition of Sn, and hydrogen consumption from the H 2 -TPR measurements are shown in Table 3. The dispersion of active components on the surface of the samples has a significant effect on the reduction of surface oxygen, and hydrogen could be more easily activated with higher dispersion of Pd, resulting in the increase of H 2 consumption [43]. Therefore, we can infer that the introduction of Sn significantly increased the dispersion of Ru on the carrier, which may have resulted from the formation of Sn ; the main peak centered at 280°C or 500°C which is attributed to the desorption of the structure oxygen species, and the peaks above 600°C are assignable to the desorption of the lattice oxygen (O 2− ) species [44]. The incorporation of Sn increased the adsorbed oxygen species and shifted to a lower temperature [45]. The results indicate that the incorporation of Sn improved the oxygen activation ability of the Ru/Sn x Ti 1−x O 2 samples and the interaction between the carriers Sn x Ti 1−x O 2 and active component Ru [46,47].

CO and/or O 2 Interaction with these Samples
The in situ FI-IR spectra of CO adsorption are recorded to further investigate the effect of the ruthenium oxide species, as shown in Fig. 11. The band located at 2052 cm −1 is attributed to linear CO adsorbed on reduced Ru crystallites (Ru δ+ -CO), the band at 2140 cm −1 and 2075 cm −1 can be assigned to two different types of multicarbonyl species on partially oxidized Ru sites (Ru n+ (CO) x ), and the band at 1765 cm −1 is attributed to (Sn x Ti 1 −x O 2 )Ru-CO species [48,49]. The Ru δ+ -CO adsorption peaks at room temperature indicate the presence of some lower state Ru δ+ species. This is in agreement with the XPS results. However, the desorption temperature of the Ru δ+ -CO peak is related to the Sn/Ti ratio and temperature. As the temperature increases, the peak intensity enhances firstly and then decreases gradually. Simultaneously, the CO adsorption peak moves to a higher wave number (2052 cm −1 at 25°C and 2060 cm −1 at higher temperatures). This red-shift indicates that Sn 4+ has stronger electron-donating capability [50]. According to the characterizations mentioned above, a possible reaction mechanism of CO and C 3 H 8 oxidation is proposed and schematized in Fig. 12

Conclusions
A series of Ru/Sn x Ti 1−x O 2 catalysts were prepared by a one-step hydrothermal method for the catalytic oxidation of CO and C 3 H 8 . The preparation conditions of Ru/ Sn x Ti 1−x O 2 catalysts were optimized for CO oxidation reaction. Ru/Sn 0.67 Ti 0.33 O 2 catalyst shows best CO catalytic activity and stability at low temperature under the condition of hydrothermal temperature at 180°C, hydrothermal time at 24 h, and calcination temperature at 400°C . The effects of different molar ratios of Sn/Ti on the catalytic properties of Ru/Sn x Ti 1−x O 2 catalysts for CO and C 3 H 8 were investigated under the optimum preparation conditions. The results show that the Ru/ Sn 0.67 Ti 0.33 O 2 catalyst exhibits better low-temperature activity and stability. The conversion of CO reached 90% at 240°C, and T 50 of which keeps at 180°C. The complete conversion of C 3 H 8 could be achieved at 500°C , and its T 50 remains at 320°C. The excellent catalytic activity of Ru/Sn 0.67 Ti 0.33 O 2 catalyst is attributed to the factors listed as follows.
Additional file 1: Figure S1 Effect of different hydrothermal temperature on Ru/Sn 0.67 Ti 0.33 O 2 catalytic CO. Figure S2 Effect of different hydrothermal time on Ru/Sn 0.67 Ti 0.33 O 2 catalytic CO. Figure S3 Effect of different calcination temperature on Ru/Sn 0.67 Ti 0.33 O 2 catalytic CO.