Facile Construction of Supported Polyoxometalate Ionic Liquids for Deep Oxidative Desulfurization of Fuel

Abstract

A series of commercial silica-supported polyoxometalate ionic liquids ([C_nmim]₃PMo₁₂O₄₀) with different substitutes in the imidazole ring were prepared via a facile ball milling method and employed as catalysts in the oxidative desulfurization of organosulfur compounds. The experimental results demonstrated that the active polyoxometalate center was successfully immobilized with a highly uniform dispersion on the silica. Without any organic solvent as the extractant, sulfur removal with the sample C₁₆PMo/SiO₂-BM reached 99.5% in 30 min under the optimal condition, which was ascribed to the high specific surface area and lipophilicity. In addition, the active site was verified by a free radical trapping experiment and an ESR spin capture experiment. In addition, the oxidative product was confirmed by GC-MS analysis.

1. Introduction

As the world’s most widely used fossil fuel, oil drives global development. However, sulfur oxides are released into the environment after the combustion of oil, leading to significant pollution [1,2]. This situation has prompted the implementation of stricter environmental regulations in many countries [3]. Currently, deep desulfurization to reduce sulfur content to an ultra-low level in oil (<10 ppm) is a critical research topic [4]. The primary industrial process for desulfurization is applied as hydrodesulfurization (HDS) in fuel [5]. However, this conventional method requiring a high operation temperature and pressure is effective in removing aliphatic sulfide but less efficient in removing aromatic sulfur compounds [6], including dibenzothiophene (DBT) and its derivatives (4-MDBT, 4,6-DMDBT) [7,8]. Therefore, various alternative processes have been explored, including extractive desulfurization (EDS) [9], adsorptive desulfurization (ADS) [10], biological desulfurization (BDS) [11], and oxidative desulfurization (ODS) [12]. Among them, ODS process is effective in removing organic sulfides from oils under mild conditions (ambient temperature and pressure) [13].

The selection of an oxidant and a catalyst in the ODS process is crucial for determining desulfurization performance [14]. Hydrogen peroxide (H₂O₂) as a green oxidant is widely used for its strong oxidizing properties with the product of water, making it environmentally benign and non-damaging to ecosystems [15]. Julião et al. reported that the sulfide in the model oil could be completely removed in 2 h at 70 °C with H₂O₂ as the oxidant, while organic solvent (acetonitrile) was used as the extractant [16]. On the other hand, ionic liquids (ILs) as a kind of green solvent can be used as substitutes for organic solvents in many catalytic reactions due to their unique properties [17]. However, the limitations of ionic liquids impede their application in fuel desulfurization, including low specific surface area, high cost and difficulty in separation after reaction [18]. According to previous studies, the immobilization of ionic liquids on a solid carrier to obtain supported IL materials can not only decrease the amount of ILs in the reaction, but also enhance the separation of the catalyst from the reaction system [19]. DeCastro et al. successfully prepared the hybrid materials SiO₂/AL-IL via the impregnation method for an alkylation reaction [20]. Under optimal conditions, the alkylation conversion for dodecane and benzene can reach 99% and 98% respectively. Kohler et al. successfully synthesized a kind of IL-based catalyst [C₁₂mim]Cl/SnC_l2 [21]. The experimental results demonstrated that dimer and trimer stannate ions [Sn₂C_l5]⁻ and [Sn₃C_l8]⁻ were formed during the loading process, which could promote the chemical adsorption of n-butyl mercaptan onto the sample. In the desulfurization process of n-butyl mercaptan (500 ppm), the sulfur content could be reduced to below 5 ppm [22]. Based on the studies above, a kind of commercial silica-supported polyoxometalate ionic liquid material was constructed via a facile ball milling method.

In this work, a kind of commercial silica-supported polyoxometalate ionic liquid material was prepared by using a facile ball milling method and was employed as a catalyst in the ODS process. The composition and structure of the samples were investigated in detail by using various methods, as well as the structure–activity relationship in the ODS process. In addition, the reaction factors including the temperature, reaction duration, and amount of catalyst and oxidant were optimized in the ODS system. Moreover, the catalytic activity on different sulfur compounds (DBT, 4-MDBT, and 4,6-DMDBT) was also studied under the same conditions.

2. Results

The polyoxometalate ionic liquid ([C_nmim]₃PMo₁₂O₄₀, n = 4, 8, 12, 16) was prepared through the ion exchange of phosphomolybdic acid (H₃PMo₁₂O₄₀) and imidazole chloride ([C_nmim]Cl, n = 4, 8, 12, 16). Then, silica-supported polyoxometalate ionic liquids (C_nPMo/SiO₂-BM, n = 4, 8, 12, 16) were obtained by mixing commercial silica and polyoxometalate ionic liquids via a facile ball milling method. XRD measurement is an effective method to characterize crystallinity and structural information in solid materials [23]. Figure 1 shows the XRD spectra of supported polyoxometalate ionic liquids after ball milling. A diffraction hump of SiO₂ can be observed in the range of 15°–40° for all catalysts, indicating an amorphous structure in the samples. However, no obvious diffraction shift for C_nPMo/SiO₂-BM (n = 4, 8, 12, 16) was observed compared to that of pure SiO₂, indicating the successful introduction of POM ionic liquids on the surface of SiO₂ [24].

FT-IR spectra further indicate the presence of the Keggin structure in the samples (Figure 2a,b). Four distinct characteristic peaks of Keggin structures can still be observed for all samples in the range of 1100–750 cm⁻¹: the peak around 1080 cm⁻¹ is assigned to P–O stretching; the bands in the range of 980–960 cm⁻¹ are ascribed to the Mo=O_t anti-telescopic vibrational absorption peak between the ligand atoms and the terminal oxygen bond; and the peaks in the area of 870–890 cm⁻¹ and 790–890 cm⁻¹ are attributed to the absorption of the Mo=O_t anti-telescopic vibrational absorption between the ligand atoms and the terminal oxygen bond. The peaks around 795 cm⁻¹ represent the Mo-O_c-Mo and Mo-O_e-Mo telescopic vibrational absorption peaks, respectively [25]. Moreover, C–H stretching vibration peaks for IL cations can be found ranging from 2800 to 2900 cm⁻¹ and the tetrahedron bending vibration of [SiO₄] is observed around 470 cm⁻¹, indicating the successful introduction of POM-IL in SiO₂ with intact Keggin structure during the balling process.

Raman scattering spectrogram was used to explore the presence of molybdenum incorporated into the hybrid materials (Figure 3). The main peak positions of the catalyst are provided as follows: 950–1000 cm⁻¹, 885–900 cm⁻¹, 590–605 cm⁻¹, and around 248 cm⁻¹. The bands in the range of 950–1000 cm⁻¹ and around 248 cm⁻¹ are attributed to the stretching vibration of the Mo-O bond. The peaks in the range of 885–900 cm⁻¹ are ascribed to the tensile vibration absorption peak of Mo-O_b-Mo, while the characteristic absorption peak of Mo-O_b-Mo is represented by 590–605 cm⁻¹. No significant change in the Raman peaks of the samples was observed, indicating the uniform dispersion of the molybdenum sites on the carriers.

To obtain further information about the Mo chemical state of the catalysts, the surface Mo elemental morphology of C₁₆PMo and C₁₆PMo/SiO₂-BM was investigated using X-ray photoelectron spectroscopy (Figure 4). For the Mo 3d energy spectrum of C₁₆PMo, the peak positions of 232.4 eV and 235.6 eV represent Mo 3d3/2 and Mo 3d5/2, respectively, which are ascribed to Mo (VI). Most of the Mo elements are present in the form of Mo (VI), as well as a small portion of Mo (V), with binding energies of 234.5 eV and 231.3 eV. The formation of Mo (V) may be attributed to the fact that part of the Mo (VI) is reduced to Mo (V) by imidazole cations in the preparation process [26,27]. For the sample C₁₆PMo/SiO₂-BM, the ratio of Mo (VI) to Mo (V) valence peak areas in the Mo 3d peaks decreased, suggesting the successful introduction of the POM ionic liquids in the silica via the ball milling method [28].

To investigate the effect of ball milling on the morphology of the catalyst, SEM images of the catalyst before and after ball milling were captured, as shown in Figure 5. For the sample C₁₆PMo/SiO₂ (Figure 5a), an amorphous and bulk state with a fractured surface and aggregated particles can be observed before ball milling. For the sample C₁₆PMo/SiO₂-BM (Figure 5b), the particle size of the sample is significantly reduced after the ball milling process, facilitating uniform dispersion of the active center to improve catalytic activity. This change in texture is ascribed to the shear force applied by the ball milling method can plastically deform the powder particles and reassemble the small particles [29].

Figure 6 shows the N₂ adsorption–desorption curves for catalysts with different carbon chain lengths to further investigate the structural properties. Using the IUPAC classification, C₁₆PMo/SiO₂-BM with an H4 desorption hysteresis loop shows a typical type IV nitrogen adsorption isotherm. In this curve, macroporous and mesoporous structure can be found in the catalyst, which is formed by the superposition of the materials during the ball milling process, agreeing well with the results of the SEM images. As shown in

Table 1. Textural properties of various samples.

	Sample	SBET (m2/g)	Pore Diameter (nm)	Pore Volume (cm3/g)
1	C4PMo/SiO2-BM	41	7.985	0.1860
2	C8PMo/SiO2-BM	81	8.258	0.1861
3	C12PMo/SiO2-BM	89	10.53	0.2144
4	C16PMo/SiO2-BM	96	25.024	0.6028

, the specific surface area increases as the carbon chain becomes longer in the sample, which is good for the oxidation reaction of the sulfide on the surface of the catalyst.

The water and oil contact angles (CA: θ) of various samples were measured to study the affinity of the catalysts to the interfaces of the aqueous and oil phases during the ODS process (Figure 7). As the carbon chains in the sample increase, the hydrophilicity of the catalysts decreases and the lipophilicity increases. The contact angle of the sample C₁₆PMo/SiO₂-BM is the smallest (17°), indicating the best lipophilicity among all of the samples, while the slight decrease in hydrophilicity has little effect on the ODS process of the catalyst. The increase in lipophilicity helps the catalyst to fully contact the oil phase to improve the ODS efficiency [30].

UV-vis spectra of the catalysts are depicted in Figure 8. It can be found that all of the samples have similar absorption peaks in the ultraviolet spectra, and these absorption peaks appear in the range of 200–400 nm. The maximum absorption wavelength is around 315 nm, representing the charge transfer process of Mo-O ligands and metals in the Keggin structure. The absorption peak near 210 nm is related to the charge transfer of P-O bonds [23]. According to the ultraviolet absorption principle of SiO₂, no UV absorption for SiO₂ would be found beyond 300 nm [31]. For the supported IL materials, the UV absorption beyond 300 nm was related to the polyoxometalate ionic liquid in the catalyst, also indicating the successful introduction of POM sites in the sample.

To evaluate the desulfurization efficiency of the synthetic catalysts, ODS experiments were carried out to obtain the optimal conditions. As shown in Figure 9a, C₁₆PMo/SiO₂-BM has the best catalytic activity for DBT, and the catalytic efficiency can reach 99.5% within 30 min. The sulfur removal rate for C₁₂PMo/SiO₂-BM and C₈PMo/SiO₂-BM is 95.7% and 89.5% in 1 h, respectively. C₄PMo/SiO₂-BM has the worst desulfurization performance. Additionally, the reaction rate constant of C₁₆PMo/SiO₂-BM is as high as 0.183 min⁻¹, which is the highest among the samples (Figure 9b). These results demonstrate that the longer carbon chain in the catalyst is beneficial for the improvement in ODS performance, and the sample C₁₆PMo/SiO₂-BM is selected as the typical catalyst. As illustrated in Figure 9c, the sulfur removal rate of 4-MDBT and 4,6-DMDBT is 91.7% and 86.3% within 60 min, respectively, while the sulfur removal rate of DBT can reach 99.5% in 30 min. These results are generally affected by the spatial steric hindrance of organic sulfur compounds. For the structure of the sulfides, 4-MDBT has one more side-chain methyl group compared to DBT, leading to a larger steric hindrance, while 4,6-DMDBT contains two substitutional methyl groups and has the largest steric hindrance. In the ODS reaction, the larger steric hindrance makes it difficult for sulfur atoms to be oxidized. As shown in Figure 9d, the sulfur removal rate is 66.5% and 77.5% at 40 °C and 50 °C in 60 min, respectively, which could not achieve the aim of deep desulfurization. At 60 °C, the sulfur removal rate is increased rapidly, reaching 99.5% in 30 min. Hence, 60 °C is selected as the optimal reaction temperature. Figure 9e demonstrates the effect of different O/S ratios on ODS activity. According to the stoichiometry in the ODS reaction, 1 mol of S (DBT) needs 2 mol of O (H₂O₂) to be oxidized to its corresponding sulfone (DBTO₂). When the O/S ratio was 1 and 2, the desulfurization rate could only reach 36.7% and 80.5%, respectively, which was due to the lack of oxidants. When the O/S ratio was 3, the desulfurization rate could reach 99.5% in 30 min. The desulfurization rate with O/S = 3 and O/S = 4 was 99.5% and 100% in 30 min, respectively. Considering economic and energy efficiency, an O/S ratio of 3 was selected for the subsequent reaction [32]. Therefore, the O/S ratio of 3 is selected as the optimal dosage of the oxidant. As shown in Figure 9f, when the catalyst mass is 15 mg, the sulfur removal rate can reach 98.8% in 60 min. When the catalyst mass is increased to 20 mg, the desulfurization efficiency is significantly improved, reaching 99.5% in 30 min. Continuing to increase the quantity of the catalyst to 25 mg, the improvement in the desulfurization rate at the same time is not significant, so the mass of the catalyst was selected as 20 mg in the ODS process. Additionally, the recyclability of the typical sample C₁₆PMo/SiO₂-BM is shown in Figure 10a. After the reaction, the upper oil phase was directly separated by decantation, while the remaining catalyst was washed with carbon tetrachloride and dried at 60 °C for 8 h. After that, fresh oil and the oxidant were added for the next run. After recycling six times, the sulfur removal rate could reach 86.7%. To verify the structural stability of the catalyst, FT-IR spectra of the catalyst were recorded before and after the ODS reaction (Figure 10b). The characteristic peaks of the catalyst after the reaction can almost match those of the fresh catalyst, indicating the stability of the sample.

To explore the main active sites in the ODS process, free radical trapping experiments were carried out using p-benzoquinone (BQ) and tert-butanol (TBA) to trap the superoxide radical (•O₂⁻) and the hydroxyl radical (•OH), respectively (Figure 11). No significant change in ODS activity was observed when BQ was added. However, after the addition of TBA, sulfur removal was significantly reduced, suggesting the generation of hydroxyl radicals as the main active sites. To further verify the active species during the reaction, ESR spin capture experiments were carried out with DMPO as the capture agent (Figure 12). No characteristic peak could be found in the reaction system without catalysts, while a set of quadruple-peak peaks ascribed to the characteristic of the typical DMPO-•OH was observed after the addition of the catalysts, further indicating that the generation of the hydroxyl radical (•OH) formed in the ODS process.

In order to understand the products produced by DBT oxidation, GC-MS analysis was used to detect the oil and catalyst in the reaction process (Figure 13). During the reaction, the oil phase was directly separated by decantation, while the catalyst was extracted by carbon tetrachloride. The presence of peaks of DBT could be observed, as well as peaks for DBTO₂, in the catalyst phase, indicating that DBT was first adsorbed by the catalyst from the oil phase and then oxidized to its corresponding sulfone (DBTO₂) by the oxidant (H₂O₂).

3. Materials and Methods

3.1. Materials

DBT (98%) and 4,6-DMDBT (98%) were obtained from Sigma-Aldrich (St. Louis, MO, USA). 4-MDBT (96%), cetane (98%), and fumed silica were purchased from Aladdin Industrial Corporation (Los Angels, CA, USA). Dodecane was netted from Shanghai Maclin Biochemical Technology Co., Ltd. (Shanghai, China) H₂O₂ (30 wt%), H₃PMo₁₂O₄₀·14H₂O (A.R. grade), and acetonitrile (A.R. grade) were acquired from Sinopharm Chemical Reagent Co., Ltd. (Shanghai, China). [C₄mim]Cl (99%), [C₈mim]Cl (99%) and [C₁₂mim]Cl (99%) [C₁₆mim]Cl (99%) were obtained from Shanghai Chenjie Chemical Co., Ltd. (Shanghai, China). All reagents and solvents were of analytical grade and were used directly without further purification.

3.2. Catalyst Preparation

The polyoxometalate ionic liquid [C_nmim]₃PMo₁₂O₄₀ (n = 4, 8, 12, and 16) was prepared according to a previous study [33]. The typical synthesis procedure was as follows: 0.139 mmol of [C₄mim]₃PMo₁₂O₄₀ and 0.5 g SiO₂ were added in 30 mL of acetonitrile and stirred at 50 °C for 30 min. Then, the mixture was transferred to the reactor at 120 °C for 24 h. After the removal of acetonitrile in the mixture, a green solid was obtained and named as C_nPMo/SiO₂-BM (n = 4, 8, 12, and 16). Finally, the resulting product was mechanically ball-milled with a rotation speed of 200 rpm to obtain C_nPMo/SiO₂-BM (n = 4, 8, 12, 16).

3.3. Characterization

The X-ray diffraction (XRD) patterns of the samples were recorded on D8 diffractometeron XRD-6100 (Shimadzu, Kyoto, Japan) in the 10°–80° region with a scanning rate of 7°/min. The morphology of various samples was observed by using a scanning electron microscope (SEM) with JSM-7800F (JEOL, Tokyo, Japan). Solid UV-vis diffuse reflection spectra (UV-vis DRS) were collected on UV-2450 UV-Vis Spectrophotometer (Japan Shimadzu, BaSO₄ discs) in the 200–800 nm region. The Fourier transform infrared (FT-IR) spectra were measured with a Nicolet Nexus 470 infrared spectrometer from Thermo Electron Corporation (Waltham, MA, USA) in the 400–4000 cm⁻¹ region. KBr was used as a dispersion medium for the samples. The X-band electron spin resonance (ESR) spectrum was recorded by JES-FA200 (Japan JEOL).

3.4. Catalytic Activity Test

The model oil was obtained by dissolving the desired amount of DBT, 4-MDBT, and 4,6-DMDBT in dodecane with a corresponding S content of 200 ppm. The ODS process was carried out in a double-neck bottle connected to a reflux condenser and heated in a constant-temperature water bath. In a typical run, the model oil (5 mL) and the catalyst (20 mg) were added to the reactor. The reaction was started by adding H₂O₂ with continuous stirring. The residual S content was analyzed using a gas chromatograph, Agilent7890A/5975C (Santa Clara, CA, USA), with cetane as the inner standard.

4. Conclusions

In summary, a series of commercial silica-supported polyoxometalate ionic liquids ([C_nmim]₃PMo₁₂O₄₀, n = 4, 8, 12, and 16) with different substitutes in the imidazole ring were successfully prepared by using a facile ball milling method and employed as catalysts in the oxidative desulfurization of organosulfur compounds. The samples with different carbon chains were characterized in detail to study their composition and structure. The experimental results demonstrate that the active sites were successfully immobilized with a highly uniform dispersion on the silica. Owing to the larger specific surface area, and better lipophilicity, the sample C₁₆PMo/SiO₂-BM had the optimal desulfurization performance without the addition of any organic solvents as extractants, and could reach a sulfur removal rate of 99.5% on DBT in 30 min. In addition, the prepared sample also exhibited good desulfurization performance on other sulfides with the following order of DBT > 4-MDBT > 4,6-DMDBT. Moreover, the hydroxyl radical (•OH) was verified as the active species through free radical trapping experiments and ESR spin capture experiments in the ODS process.