Introduction As a semiconductor material with superior optical properties, amorphous TiO2 with the valence band, conduction band, and bandgap structure of crystalline TiO2 is prone to forming a loose porous structure with more surface defects. During the photocatalytic reaction process, it can generate a large number of electron and hole capture traps, which is conducive to improving a photocatalytic efficiency. However, its photocatalytic activity still needs to be improved. Widening its absorption range, narrowing its band gap and promoting the migration of the internal electron-hole to the surface to react with the target product are all still the current important research aspects.Doping or loading is a normal modification method of TiO2. The doping with rare-earth element La is simple and effective, narrowing the band gap of TiO2 and reducing the energy consumption of photocatalytic reaction. To further improve the photocatalytic activity, loading La-TiO2 on SBA-15 can be an effective option. The high specific surface area and the mesopores of SBA-15 can be beneficial to the dispersion of the active component and the diffusion of reactants, thereby improving the photocatalytic performance. In this paper, TiO2 was firstly doped with La, and La-TiO2 was then supported on mesoporous molecular sieves SBA-15. The obtained catalyst La-TiO2/SBA-15 was characterized and applied in the photocatalytic oxidation desulfurization for the model and real diesel oil. In addition, the photocatalytic reaction mechanism was also proposed.Methods 4 mmol tetrabutyl titanate was dropped slowly into a solution of 0.9 mL of deionized water with 10 mL of absolute ethanol under magnetically stirring until a white gel was generated. After stirring for 0.5 h, the white gel was aged at 25 ℃ for 14 h, and then stirred at 40 ℃ until the absolute ethanol was completely evaporated. This white powder becomes an amorphous TiO2 after drying at 100 ℃ and grinding. The synthesis process of La-TiO2 was similar to that of amorphous TiO2. However, deionized water was replaced with a solution of lanthanum nitrate (Its concentration was calculated according to the doping amount of La). To support La-TiO2 on SBA-15, lanthanum nitrate and tetrabutyl titanate were put into 10 mL absolute ethanol. The ratio of the feed stocks in this mixture is same as that of the synthesis of La-TiO2. SBA-15 mesoporous molecular sieve was also put into this mixture, and the loading of La-TiO2 on the supported catalyst was 10% (in mass content). 1 mL deionized water was dropped slowly into the mixture under stirring.The X-ray diffraction (XRD) patterns of the samples were determined on a model D/max-RB X-ray diffractometer with Cu-Kα radiation operating at 40 kV and 150 mA. The composition of the catalyst was identified by a model D/max-R X-ray fluorescence spectrometer (XRF) with a tungsten target at 40 kV and 50 mA. The elemental quantitative or semi-quantitative analysis was carried out based on the characteristic peak intensity of each element. The pore structure was characterized by a model ASAP 2010 physical adsorption instrument (BET). After 16-h pretreatment in vacuum at 110 ℃, N2 adsorption-desorption was operated at -196 ℃. The transmission electron microscope (TEM) images were determined on a model JEM-2010CX electron microscope at 200 kV. The Fourier-transform infrared spectra (FT-IR) in the range of 4 000-400 cm-1 were recorded on a model WQF 510 spectrometer at a resolution of 4 cm-1. The UV-Vis diffuse reflection spectra (UV-Vis) were recorded on a model Cary 2450 UV-Vis spectrometer with BaSO4 as a reference sample in the wavelength range of 200-800 nm. The photoluminescence (PL) spectra were recorded on a model FluoroMax-4 spectrofluorometer, which was excited at 240 nm in the scanning range of 260-460 nm. The valence-band X-ray photoemission spectroscopic data (VB-XPS) were obtained on a model ESCALAB 250 X-ray photoelectron spectrometer equipped with an Al Kα X-ray source and the spot size of 500 μm. At ambient temperature, PODS reaction was operated in a beaker under magnetically stirring. 10 mL model or real fuel was added firstly into a beaker, and then some amount of CH3OH as an extractant (calculated in volume ratio of the extractant to the fuel), a catalyst (calculated in the mass percent in the model fuel) and H2O2 (calculated in n(O)/n(S)) were added. The beaker was placed in the dark under magnetically stirring for 0.5 h to establish adsorption-desorption equilibrium of DBT on catalysts. PODS proceed under the visible light irradiation and the samples were withdrawn periodically every 0.5 h from the upper phase, and the sulfur content of the clear liquid sample was detected on a model TSN-5000 sulfur nitrogen detector after high-speed centrifuge.Catalyst was separated from the reaction mixture via filtration and washing with deionized ethanol, dried at 80 ℃ and then directly used for the next run.Four PODS processes were repeated, and in each process 1.0 mmol/L a kind of active intermediate capture agent was added. Isopropanol (IPA), p-benzoquinone (p-BQ) and disodium ethylenediaminetetraacetate (EDTA-2Na) were used for detecting hydroxyl radicals, superoxide radicals and holes, respectively.Results and discussion The XRD patterns show that La-TiO2/SBA-15 retains a two-dimensional hexagonal highly ordered mesoporous structure of SBA-15. TiO2, La-TiO2 and La-TiO2/SBA-15 do not show any characteristic diffraction peaks, indicating that they all are amorphous. Compared with the sample SBA-15, the specific surface area, pore volume and pore size of sample La-TiO2/SBA-15 decrease due to the loading of La-TiO2. However, its specific surface area is still higher than that of amorphous TiO2. The TEM images present an ordered mesoporous structure of La-TiO2/SBA-15, and La-TiO2 is also evenly dispersive in the pore channels. The UV-Vis analysis shows that the absorption edge of La-TiO2 is higher than that of TiO2. A reason is that lanthanum doping reduces a band gap energy of TiO2 and improves a light absorption ability of the catalyst. Although a band gap width is almost unchanged for loading. The PL spectra indicate that the intensity of fluorescence emission peak decreases after doping and loading, proving that the doping and loading both can reduce an electron-hole recombination rate, thereby prolonging the photogenerated carrier lifetime of amorphous titania material.The effect of reaction condition on the results of PODS was investigated and optimized. The results are obtained under optimum conditions (i.e., catalyst dosage of 1% (in mass fraction), n(O):n(S) of 15:1, and the of agent:oil ratio of 1:1). Under such conditions, the desulfurization rate is 95.12%, and it can still reach more than 85% after four runs of reusing. For La-TiO2/SBA-15 as a catalyst, the active intermediate species in PODS were ?O2- and h+ based on active intermediate species capture experiments.Conclusions La-TiO2/SBA-15 prepared had a two-dimensional hexagonal pore structure, a high specific surface area, and well-dispersion. Lanthanum doping broadened a light response range, increased the absorbance of the light, and reduced the photogenerated electron-hole recombination rate of TiO2. La-TiO2/SBA-15 had a higher catalytic activity rather than TiO2 and La-TiO2, and a good reusability. The main active intermediate species were ?O2- and h+.