Globally, crude oil continues to be a widely used and important resource, particularly, for transportation fuels ^{[1, 2, 3]}. However, crude oil-derived fuels contain a complex range of components, which themselves include various sulfur compounds. As well as deactivating catalysts that are fitted to automotive exhausts to decrease pollutant emissions, combustion of these sulfur compounds releases pollutants that contribute to acid rain ^{[4, 5]}. Increasing environmental protection awareness has therefore led to an urgent need for fuel desulfurization ^{[6, 7]}. Many countries have issued environmental regulations to limit fuel-sulfur concentrations to less than 10 ppm ^[8]. However, achieving such low concentrations with the range of complicated sulfur compounds found in petroleum makes refining more challenging ^{[2, 9]}.

At present, the removal of sulfur compounds from liquid fuels is achieved by hydrodesulfurization that can remove thiols, sulfides, and disulfides efficiently ^{[10, 11]}. However, hydrodesulfurization is a nonselective hydrogenation process that requires high temperature and high pressure. The process also lowers the fuel’s octane level and requires substantial quantities of H₂, which is expensive ^[12]. Furthermore, hydrodesulfurization is a less successful method for achieving desulfurization of thiophene and benzothiophene derivatives ^{[13, 14, 15]}.Therefore, for the production of cleaner transportation fuels, supplementary approaches to achieve this deep desulfurization are required.

A number of deep desulfurization approaches have been investigated, including biodesulfurization, extractive desulfurization by ionic liquid, adsorptive desulfurization, and oxidative desulfurization ^{[16, 17, 18, 19, 20, 21]}. Among them, the oxidative desulfurization is considered one of the most effective methods ^{[22, 23, 24, 25, 26, 27]}. This is because in the oxidative desulfurization process dibenzothiophene (DBT) and its derivatives can be selectively oxidized to their corresponding sulfoxides and sulfones (Scheme 1) under mild reaction conditions (< 373 K and ambient pressure) using common and inexpensive oxidants ^{[28, 29]}. The sulfoxides and sulfones can then be easily removed using water-soluble polar solvents as the extractant ^{[8, 24]}.

Scheme 1. Example of the oxidative desulfurization process.

The oxidant plays a very important role in the oxidative desulfurization process. Many types of oxidants have been studied, such as O₂, H₂O₂, and organic peroxides. Among them, H₂O₂ has been widely used as it is cheap, commercially available, and relatively environmentally friendly ^{[30, 31, 32, 33]}.

Catalysts are responsible for activating the oxidants and are therefore essential in the oxidative desulfurization process ^[34]. Different catalytic systems have been reported for the process; these include those based on an organic acid ^[35], heteropolyoxometalates ^[36], an ionic liquid ^[37], a molecular sieve ^[38], and photocatalysts ^[39]. Within this group, polyoxometalates have a number of unique features that allow their processes to exhibit higher sulfur removal efficiencies ^{[40, 41]}.

In recent years, supporting catalysts on solid mesoporous materials have been reported as a potential solution to deep desulfurization. Li et al. ^[42] reported a mesoporous silicate incorporated with phosphotungstic acid for DBT oxidation that obtained 98%-99% DBT conversion in experiments with model oil. In another experiment, Li et al. ^[43] prepared molybdovanadophosphoric acid on functionalized SBA-15 for oxidative desulfurization and reported a 97.8% conversion of DBT in a model oil while the supported catalyst was shown to be easily recovered; an important finding given that catalyst recovery and recycling is now a significant focus of the catalytic industry. The same work also noted that immobilization of the catalyst on mesoporous materials with weak bonding led to partial leaching when reacted in a polar solvent ^[43]. However, this and a number of other works have shown that when the catalyst is supported on modified silica materials with electrostatic interactions it exhibits better stability ^{[43, 44, 45]}.

In 2011, the catalyst used in this work (Q₄-H₂Se^IV₃W₆) was reported to exhibit highly efficient and selective properties in the oxidative desulfurization process with DBT conversion of more than 99% in a desulfurization test using model oil ^[46]. However, catalyst recovery in this case remains an unresolved topic and is the focus of this work.

Here, an optimized approach is reported whereby the peroxotungstate catalyst was impregnated in MCM-41 and MCM-41-NH₂ (functionalized by 3-aminopropyltriethoxysilane (APTES)) materials. The ability of these materials to facilitate the oxidative desulfurization process was tested using a model oil. The optimum reaction conditions were investigated, including reaction time, reaction temperature, molar ratio of [O]/[S], and dosage of catalyst. In addition, recycling tests were carried out.

DBT (99%), APTES, sodium tungstate dihydrate (AR), cetyltrimethylammonium bromide (CTABr), selenium dioxide (AR), typical tetraethyl orthosilicate (TEOS), and stearyl trimethyl ammonium chloride (STAC, 99%) were purchased from Aladdin Reagent Inc. H₂O₂ (30%, AR) and n-octane (CP) were obtained from Sinopharm Chemical Reagent Co. Ltd. NaOH (AR), chloroform (AR), acetonitrile (AR), toluene (AR), and HCl (37%, AR) were supplied by Nanjing Chemical Reagent Co. Ltd. All of the reagents except toluene were used as received without any further purification. The toluene was distilled to remove water and sealed in bottles to keep it dry.

The model fuels were prepared by dissolving DBT in n-octane to obtain three solutions with a sulfur content of 200, 500, and 800 ppm, respectively.

MCM-41 was synthesized following the procedures described in the literature ^{[47, 48, 49]}. In brief, this involved TEOS slowly being added to a vigorously stirred solution of NaOH and CTABr at room temperature. The solution was then stirred for 2 h at 353 K. Heating was stopped though stirring continued until the temperature had cooled to room temperature. The obtained gel was crystallized at 383 K for 48 h in stainless steel reactors with Teflon liners. After filtration, the crystallized gel was calcined in a muffle furnace at 823 K for 5 h and allowed to cool naturally yielding MCM-41 as a white powder.

MCM-41-NH₂ was prepared using a grafting technique ^[50]. Here, MCM-41 was first activated in a muffle furnace at 773 K for 4 h and cooled in a vacuum oven. One gram of activated MCM-41 was then dispersed in 50 mL of distilled toluene, and 1.5 g of APTES was added to the mixture. The mixture was magnetically stirred and refluxed at room temperature for 8 h. The resultant white solid was filtered and washed with distilled toluene. MCM-41-NH₂ powder was obtained after the white solid had been dried under vacuum at 303 K for 8 h.

Catalyst A, [C₁₈H₃₇N(CH₃)₃]₄[H₂Se^IV₃W₆O₃₄] (abbreviated to Q₄-H₂Se^IV₃W₆ ), was synthesized according to the method described in the literature ^[46]. A yellow flaky solid was obtained after the crude product had been dried at room temperature under vacuum overnight.

Catalyst B, MCM-41/Q₄-H₂Se^IV₃W₆, was prepared by the wet impregnation method ^[42]. A desired amount of Q₄-H₂Se^IV₃W₆ was weighed and dissolved into an aqueous solution, and a calculated amount of MCM-41 was added to the solution. As previously reported ^[51], to ensure that the peroxotungstate retains its structure and is well dispersed on MCM-41 materials, the peroxotungstate content should be lower than the amount required for monolayer coverage. Therefore, a loading of 5 wt% was chosen. To prepare 5 wt% MCM-41/Q₄-H₂Se^IV₃W₆, 0.020 g Q₄-H₂Se^IV₃W₆, 1.550 g of deionized water, and 0.389 g of MCM-41 were mixed in a beaker with the mixture then continuously stirred overnight at room temperature. To activate MCM-41/Q₄-H₂Se^IV₃W₆,the impregnated materials were placed into a vacuum oven for 12 h at 383 K.

Catalyst C, MCM-41-NH₂/Q₄-H₂Se^IV₃W₆,was prepared following a similar method as that used for catalyst B. To prepare 5 wt% MCM-41-NH₂/Q₄-H₂Se^IV₃W_6, 0.020 g of Q₄-H₂Se^IV₃W₆, 2.680 g of deionized water, and 0.420 g of MCM-41-NH₂ were mixed. Prior to use, the sample was heated overnight in a vacuum oven at 383 K to remove gas or water trapped in the sample.

Fourier transform infrared spectra (FTIR) in the range of 500−4000 cm⁻¹ were obtained using a NEXUS 870 spectrometer (Nicolet, America) using a dried KBr pellet.

X-ray diffraction (XRD) measurements were carried out from 0.6° to 10° with a scan rate of 0.5°/min using a D8 ADVANCE system using a Cu K_α radiation source at 40 kV and 40 mA.

Images of the samples were captured by transmission electron microscopy (TEM), which was carried out using a JEOL 2100 electron microscope operating at 200 kV.

N₂ adsorption-desorption isotherms were measured by an ASAP 2020 (Micromeritics). The samples were pretreated at 473 K under a high vacuum, and the pore structure was calculated from the adsorption isotherm. Specific surface areas were calculated in accordance with the BET method and pore size distributions were calculated using the desorption branch of the isotherm and employing the BJH method.

In a 50 mL round-bottom flask equipped with a magnetic stirrer and a heated circulating bath, 0.5 mL of acetonitrile was mixed with 10 mL of DBT, the model fuel, to be tested. Acetonitrile was used as it is one of the most suitable solvents for the oxidation desulfurization system ^[52,53,54]. The amount of catalyst was maintained at a molar ratio of catalyst/DBT = 1:60 ^[46]. Following the reaction detailed in Scheme 1, the amount of H₂O₂ (30 wt%) required was determined by maintaining [O]/[S] = 2:1. After the catalyst and H₂O₂ were added, the reaction was started at atmospheric pressure and a given set temperature that depended on the test with the reaction time then recorded.

The progress of oxidation was promptly analyzed by a sulfur-specific gas chromatography system that was equipped with a flame photometric detector (GC-FPD). The specimens were sampled from the supernatant liquid at different time intervals. The GC-FPD was set as follows: capillary column SE-30 (50 mm x 0.53 mm x 0.5 μm); injector temperature 513 K; column temperature 453 K; detector temperature 493 K; ultra-pure N₂ as carrier gas (1.0 mL/min); reagent gases air (60 mL/min) and ultra-pure H₂ (60 mL/min); injection volume 1 μl. Confidence in measurements repeatability was ensured by repeating each measurement three times.

The DBT conversion (D) was calculated as D = (S₀−S_F)/S₀ x 100%, where S₀ and S_F are the concentration of sulfur (DBT) in the oil phase initially and after the reaction had completed, respectively.

The FTIR spectrum of Q₄-H₂Se^IV₃W₆ (catalyst A) is shown in Fig. 1(1) and exhibits characteristic peaks at 969 cm⁻¹ (W=O stretching vibration), 911 cm⁻¹ (O-O stretching vibration), and 721 cm⁻¹ (Se-O stretching vibration). Fig. 1(2)-(5) shows the FTIR spectra of MCM-41, MCM-41/Q₄-H₂Se^IV₃W₆ (catalyst B), MCM-41-NH₂, and MCM-41-NH₂/Q₄-H₂Se^IV₃W₆ (catalyst C). The peaks at 780 and 1090 cm⁻¹ are associated with the Si-O-Si symmetric and asymmetric stretching vibrations of mesoporous silica, respectively. The band at 780 cm⁻¹ is assigned to the stretching vibration of Si-O-H. The peak at 721 cm⁻¹ shifted to 696 cm⁻¹ after Q₄-H₂Se^IV₃W₆ was impregnated on MCM-41 and MCM-41-NH₂. Significantly, compared with MCM-41 and MCM-41-NH₂, a new band appeared at approximately 1470 cm⁻¹ over catalyst B and catalyst C. These results indicate that Q₄-H₂Se^IV₃W₆ was highly dispersed on the surface of MCM-41 and MCM-41-NH_2.

Fig. 1. FTIR spectra of various samples. (1) Catalyst A; (2) MCM-41; (3) MCM-41-NH2; (4) Catalyst B; (5) Catalyst C.

Small-angle XRD patterns of four samples are depicted in Fig. 2. Both MCM-41 and MCM-41-NH₂ exhibited a strong reflection at 2θ = 2.1° because of the (100) plane and a few weak peaks because of (110) and (200) plane reflections. These peaks show highly ordered, two-dimensional, hexagonal and mesoporous channels of MCM-41 and MCM-41-NH₂. For catalysts B and C, all of these peaks were still detected but with a decreased intensity. This indicates that the ordered pore structure of mesoporous materials was retained well and that the Q₄-H₂Se^IV₃W₆ was well-dispersed across the mesoporous channels. The intensity of the peaks is associated with the degree of crystallinity ^[55]. The lower intensity demonstrates a decrease in the arrangement of ordered, two-dimensional, hexagonal and mesoporous channels of silica. It reveals that Q₄-H₂Se^IV₃W₆ was loaded onto the mesoporous channels successfully. This could be caused by interaction between mesoporous silica materials and Q₄-H₂Se^IV₃W_6. This result agrees with other reported studies ^{[55, 56]}.

Fig. 2. Small-angle XRD patterns of MCM-41 (1), MCM-41-NH2 (2), catalyst B (3), and catalyst C (4).

TEM was used to image the porous structure of the synthesized materials, as shown in Fig. 3. It clearly shows the well-ordered, hexagonal arrays of the mesopores typical of MCM-41 materials. The modified samples (MCM-41-NH₂) also provide well-ordered, hexagonal array structures. The pores of catalyst B were slightly smaller than those of the MCM-41 samples and the channels were less clearly defined. This implies that the Q₄-H₂Se^IV₃W₆ inserted into the pores, but that the mesoporous structure of the support was retained after being impregnated by Q₄-H₂Se^IV₃W₆. The pore size has clearly changed following the impregnation of Q₄-H₂Se^IV₃W₆ in the channels of MCM-41-NH₂. The channels of catalyst C were thinner and less sharply defined than those of MCM-41-NH₂, but the mesoporous structure of the support was retained. The TEM images provide strong evidence that Q₄-H₂Se^IV₃W₆ was well dispersed in the channels of the mesoporous materials, agreeing well with the XRD results.

Fig. 3. TEM images of MCM-41 (a), MCM-41-NH2 (b), catalyst B (c), and catalyst C (d).

The measured values of surface area and pore volume for all catalysts are presented in Table 1. Pure Q₄-H₂Se^IV₃W₆ showed a limited surface area (< 10 m²/g) and a small pore volume (0.03 cm³/g). The average pore volume of MCM-41 was larger than that of MCM-41-NH₂, indicating that the MCM-41 was successfully modified by APTES. As shown in Fig. 4, the internal channels became more complex and smaller after modification. The values for both surface area and pore volume decreased after the two supports (MCM-41 and MCM-41-NH₂) were impregnated with Q₄-H₂Se^IV₃W₆. Considering the low surface area of Q₄-H₂Se^IV₃W₆ (< 10 m²/g), the decrease of surface area and pore volume was likely caused by small clusters of Q₄-H₂Se^IV₃W₆ forming in the channels of the mesoporous materials. The data shown in Table 1 confirmed that Q₄-H₂Se^IV₃W₆ was dispersed in the channels of the mesoporous materials.

Fig. 4. MCM-41 functionalized by APTES.

Table 1 Surface area and pore volume of mesoporous materials and catalysts. Sample ABET (m2/g) Pore volume (cm3/g) Catalyst A < 10 0.03 MCM-41 647 0.76 MCM-41-NH2 324 0.31 Catalyst B 551 0.52 Catalyst C 229 0.21

Table 1 Surface area and pore volume of mesoporous materials and catalysts.

As mentioned above, simple impregnation of catalyst on mesoporous materials could lead to partial leaching. Therefore, favoring the stability of catalyst, catalyst C was chosen to conduct the following tests.

DBT conversion tests were carried out for fuel samples with different sulfur contents. The results are shown in Fig. 5. It is obvious that the conversion of DBT increases with time, though the majority of the conversion completed within 30 min for each of the tests. Moreover, DBT was almost completely converted (> 99%) after 180 min. When the sulfur content was raised to 800 ppm, the conversion of DBT decreased. As seen from Fig. 5, for model fuels with 500 ppm sulfur content, the conversion of DBT was 98.7% (approximately the same as the > 99% required for deep desulfurization) after a reaction time of 120 min. These conditions (500 ppm sulfur content and 120 min reaction time) were chosen for comparative experiments for the desulfurization process using the different catalysts. A similar time was required to achieve deep desulfurization as that reported in the literature ^[46]. This suggested that the high efficiency of DBT conversion by Q₄-H₂Se^IV₃W₆ was retained following being impregnated on MCM-41-NH_2.

Fig. 5. Influence of reaction time on DBT conversion over catalyst C. Reaction conditions: model fuels 10 mL, 318 K, [O]/[S] = 2:1, catalyst/DBT = 1:60.

Fig. 6 shows that at a given reaction time the conversion of DBT was improved by increasing the reaction temperature. When the reaction temperature was elevated from 298 to 318 K, the conversion of DBT reached 90% within 60 min (compared to 78% at 298 K). The variation between reaction rates at different temperatures suggests a suitable temperature for accelerating reaction needs to be found. According to the Arrhenius equation (K = Aexp^−Ea/RT) the reaction rate increases exponentially as temperature rises. Although increasing the reaction temperature increases the reaction rate, a tradeoff is needed to decrease energy requirements and to account for other reactions that could occur. For example, elevating the temperature from 318 to 338 K had little impact on the DBT conversion at 100 min, with this attributed to the fact that the increased temperature could also cause decomposition of H₂O₂. This decomposition of H₂O₂ decreases the amount of oxidant available and also dilutes the oxidant as water is one of the decomposition products. In this study, 318 K was therefore chosen as the most suitable temperature for the desulfurization process, somewhat lower than that used in previously reported work ^[43].

Fig. 6. Effect of temperature on DBT oxidation reaction. Reaction conditions: 500 ppm model fuels 10 mL, [O]/[S] = 2:1, catalyst/DBT = 1:60.

In the oxidative system, one of the main factors impacting the conversion is the dosage of oxidant used. The effects of oxidation dosage were studied under different [O]/[S] molar ratios. As seen in Fig. 7, the conversion of DBT quickly increased when [O]/[S] increased to 2:1. However, increasing [O]/[S] further only slightly increased DBT conversion.

Fig. 7. Impact of [O]/[S] molar ratio on DBT conversion. Reaction conditions: 500 ppm model fuels 10 mL, 318 K, catalyst/DBT = 1:60.

According to the stoichiometry of the reaction, 1 mol DBT would consume 2 mol H₂O₂ and produce 1 mol sulfone (DBTO₂). As reported elsewhere ^{[57,58,59,60]}, the H₂O₂/sulfur molar ratio is usually > 2 in the actual reaction process because both oxidation of DBT by H₂O₂ and decomposition of H₂O₂ simultaneously occur. When [O]/[S] > 2, the diminishing marginal increase in reactivity is attributed to an increase in the thermal decomposition of H₂O₂ and dilution effects from water produced from the oxidative reaction. Hence, an excess of H₂O₂ was deemed unnecessary and [O]/[S] = 2 was chosen for this experiment.

To investigate the effect of catalyst amount on the conversion of DBT, a series of experiments with varying amounts of catalyst C were carried out under the same reaction conditions. The results shown in Fig. 8 indicate that the conversion of DBT sharply increased from 8.3% to 99.7% as the molar ratio of catalyst/DBT ranged from 0 to 0.020. It was found that 98.7% DBT conversion was obtained when catalyst/DBT = 0.015. However, further increasing catalyst/DBT up to 0.025 little improved the conversion of DBT. The catalyst/DBT = 0.015 was the point at which sufficient catalytic sites were available to drive the reaction to completion, and was chosen for further experiments. This catalyst dosage was much lower than that reported elsewhere ^{[43, 46]}, perhaps because the channel structure in the current work increased the number reactive sites.

Fig. 8. Effect of catalyst C dosage on DBT conversion. Reaction conditions: 500 ppm model fuels 10 mL, 318 K, [O]/[S] = 2:1.

The activity of various catalysts applied to the oxidative desulfurization reaction is compared in Fig. 9. Results for MCM-41 and MCM-41-NH₂ were similar to the control, demonstrating that the mesoporous materials make no or very limited contributions to catalytic activity. Similar results were found in the literature ^{[61, 62]}. Catalysts A, B, and C all exhibited high and similar catalytic activity with conversion in the range 98.3%-99.1%.

Fig. 9. Performance of three kinds of catalyst on DBT conversion. Reaction conditions: 500 ppm model fuels 10 mL, 318 K, [O]/[S] = 2:1, catalyst/DBT = 1:60.

Catalysts B and C can be recovered and reused in subsequent catalytic oxidative desulfurization runs. Following the end of a reaction, the used catalysts were separated from the reaction solution, mixed with acetonitrile and stirred for 10 min before being filtered out. This washing was carried out three times, and the catalysts were then pretreated in vacuum oven at 383 K for 12 h before being used again.

As seen in Fig. 10, the catalytic efficiency of catalyst C was higher than that of catalyst B through four reaction cycles. This could be due to increased leaching of unstable immobilized Q₄-H₂Se^IV₃W₆, causing a reduction in active sites for catalyst B. The activity loss in subsequent reaction cycles could also be attributed to the inactivation of some active sites during the oxidation or recovery processes, which was also more pronounced for catalyst B. The more durable nature of catalyst C in the oxidative desulfurization process can be explained as follows. The silanol groups, which are present at the surface and inside the channels of the mesoporous MCM-41, reacted with aminosilane molecules forming ≡Si(CH₂)₃NH₃^{+ [63]}. However, after Q₄-H₂Se^IV₃W₆ was impregnated on MCM-41-NH₂, the salt ≡Si(CH₂)₃NH₃·H₂Se^IV₃W₆-Q₄ was formed instead ^[64], with the chemical interaction between ≡Si(CH₂)₃NH₃⁺and Q₄-H₂Se^IV₃W₆ enhancing the stability of catalyst C.

Fig. 10. Impact of recycling supported catalysts on DBT conversion. Reaction conditions: 500 ppm model fuels 10 mL, 318 K, [O]/[S] = 2:1, catalyst/DBT = 1:60.

After the reaction, GC-MS and FTIR analysis were carried out on solid products that were formed on the wall of the vessel. GC-MS analysis (Fig. 11) registered M⁺ at m/z = 216 confirming the presence of the sulfone of DBT. The intense m/z peaks appeared at 187.1 and 168.1 and were attributed to the C₁₁H₇SO and C₁₂H₈O moieties of DBT S,S-dioxide, respectively ^[65].

Fig. 11. GC-MS spectrogram for oxidation products present in n-octane.

As the FTIR results in Fig. 12 show, peaks appeared at 1164 and 1288 cm⁻¹ and were associated with the O-S-O symmetric and asymmetric stretching vibrations of DBTO₂, respectively.

Fig. 12. FTIR spectrum of the solid products formed by the desulfurization reaction.

The acetonitrile phase and the solid product (first dissolved in acetonitrile) were analyzed by GC-FPD. The results shown in Fig. 13 demonstrate that the final products in this reaction were mainly DBTO₂.

Fig. 13. The megascopic GC-FPD traces of post-reaction acetonitrile phase (a) and the solution of the solid products formed by the desulfurization reaction dissolved in acetonitrile (b).

The results shown in Fig. 7 demonstrate that DBT could be oxidized by the catalyst alone (i.e. without H₂O₂) with DBT oxidized into sulfoxide or sulfone. In this study, no sulfoxide was detected by GC analysis under reaction conditions. The oxidation of sulfide to sulfoxide is considered fast. So, the mechanism of DBT oxidation in the presence of catalyst C can be described as the process shown in Fig. 14 and described below.

Fig. 14. Mechanism of DBT oxidation.

It could be suggested that the catalyst on the surface of mesoporous silica samples acted as the catalytic active sites in the oxidative desulfurization. However, the oxidation of DBT actually contains two steps. First, the sulfur atom of dibenzothiophene stages a nucleophilic attack on the specific oxygen atom in the catalyst, two protons shift, and the oxygen atom transfers from the catalyst to the dibenzothiophene molecule, forming DBTO or DBTO₂ (Fig. 14(i)). Second, the catalyst that lost the oxygen atom contacts with H₂O₂, and the peroxide bond regenerates (Fig. 14(ii)). Together these two half reactions complete the oxidization of DBT ^[66,67,68].

Mesoporous materials impregnated with Q₄-H₂Se^IV₃W₆ show effective performance in the oxidative desulfurization at mild reaction conditions. The catalyst MCM-41-NH₂/Q₄- H₂Se^IV₃W₆ showed excellent properties when MCM-41 functionalized by APTES was used as the support structure. The desulfurization reaction achieved a high conversion of DBT at mild conditions (318 K) with almost all hydrogen peroxide used in the oxidative reaction ([O]/[S] = 2:1). Furthermore, only a small amount of catalyst (catalyst/DBT = 0.015) was required to achieve a high conversion of DBT (98.7%). The catalyst activity slightly decreased following subsequent reaction cycles though the conversion of DBT remained around 90% after four rounds of recycling.

Taken together these results demonstrate that the mesoporous material of MCM-41-NH₂ impregnated with Q₄-H₂Se^IV₃W₆ is an excellent catalyst for the oxidative desulfurization system. The present work on the most economical and efficient conditions for the use of Q₄-H₂Se^IV₃W₆ not only solves the problem of recovery and reuse of Q₄-H₂Se^IV₃W₆, but also provides a reference for future work in oxidative desulfurization.

Acknowledgments

The authors wish to acknowledge the financial support provided by the School of Chemistry and Chemical Engineering, Nanjing University.