Facile Construction of Magnetic Ionic Liquid Supported Silica for Aerobic Oxidative Desulfurization in Fuel

: With the rapid growth in fuel demand, deep desulfurization of fuel oil is vitally necessary for the sake of health and environmental protection. In this work, a kind of magnetic ionic liquid supported silica is prepared by a facile ball milling method, and applied in the aerobic oxidative desulfurization of organosulfurs in fuel. The experimental results indicated that ball milling proce-dure can increase the speciﬁc surface area of samples, which is beneﬁcial to oxidative desulfurization process. Under the optimal reaction conditions, the prepared materials can have an entire removal of aromatic sulfur compounds as well as a good recycling ability. Moreover, the introduction of Fe 3 O 4 did not decline the desulfurization performance, but help the catalyst to be easily separated after reaction.


Introduction
With the rapid development of the automobile industry, the demand for fuel oil was greatly increased in the past decades. However, sulfur compounds in fuel oil are converted to SO X after combustion, which is a major source of acid rain, haze, and air pollution [1][2][3][4]. Hence, the legal threshold values for sulfur content in fuel oil is limited to less than 10 ppm in many countries [5][6][7]. Currently, the traditional technology to remove sulfur compounds in the oil refining industry is hydrodesulfurization (HDS), which is extremely efficient in removing aliphatic and acrylic sulfur compounding such as thiols, sulfides, and disulfides. However, it is difficult to apply in the removal of aromatic sulfur compounds such as dibenzothiophene (DBT) and its derivatives [8,9]. Hence, some supplementary technologies have been developed, including extraction desulfurization (EDS), adsorption desulfurization (ADS), biological desulfurization (BDS), and oxidative desulfurization (ODS) [10][11][12][13][14]. In particular, ODS is considered to be a promising technology, which can efficiently remove aromatic sulfur compounds under mild conditions. Therefore, deep oxidative desulfurization is a very important research topic, especially using green oxidants such as molecular oxygen.
The choice of catalyst is also critical for the ODS process. Various kinds of materials have been applied in the ODS process, including organic acids, inorganic acids, and peroxide materials [15]. Although these homogeneous catalysts exhibit high activity, it is difficult for them to be separated from the ODS system after reaction. In order to solve this problem, a series of heterogeneous catalysts have gradually been developed [16][17][18], such as metal oxides [19][20][21], activated carbon [22], molecular sieves (e.g., SBA-15 and MCM-41) [23,24], and polyoxometalates (POMs) [25][26][27]. Among them, polyoxometalate (POMs) catalysts have attracted wide attention for their advantages of strong redox properties, environmental friendliness, and high designability. However, due to the high lattice energy, most POMs materials have poor dispersibility in organic solvents, resulting in low activity in ODS process [26,27]. To solve this problem, Li's group [25,28,29] designed a series of ionic liquids (ILs) with organic cation and POMs-based anion to improve the interface between the catalyst and oil phase, promoting the desulfurization performance. However, the low specific surface area of POMs materials often leads to a long reaction time and poor recycling performance. Hence, the immobilization of POMs onto suitable carriers would solve the problem of improving the catalytic activity in oxidative desulfurization [30][31][32][33][34][35][36]. Hence, various types of POMs supported materials such as W-SiO 2 -type [12,13,[37][38][39], Mo-SiO 2 -type [40][41][42], V-SiO 2 -type [43], and Au-SiO 2 -type [44,45] have been studied, exhibiting good activity in the ODS process. However, the used catalysts could not be easily separated from the recycling system, and other separation methods such as centrifugation are required, increasing the energy consumption and reaction cost. Magnetic nanocomposites such as Fe 3 O 4 are a type of material that can be moved by an external magnetic field [46]. Therefore, Fe 3 O 4 has the application potential for rapid separation of the catalyst after the desulfurization reaction. On the other hand, mechanochemical methods have been widely explored in the preparation of catalysts due to their advantages such as being solvent-free, environmentally friendly, and their rapid preparation [47][48][49].
In this work, a kind of magnetic ionic liquid supported silica (C 16 PW/SiO 2 -Fe-BM) is prepared by a facile ball milling method, and applied in the aerobic oxidative desulfurization of organosulfurs in fuel. The physicochemical properties of the catalyst were characterized by various technical methods such as FT-IR, Raman, BET, and XPS. The catalytic performance of the catalyst was evaluated by oxidizing the sulfides in the presence of oxygen. The effect of Fe 3 O 4 on the magnetic properties was also investigated as well as the desulfurization performance. The experimental results demonstrated that the sample C 16 PW/SiO 2 -Fe-BM exhibited good desulfurization activity, which can be quickly separated after reaction.

Characterization of Samples
FT-IR spectra of various samples is depicted in Figure 1. For the ionic liquid C 16 PW, four characteristic peaks of Keggin structure could be observed in the region of 800-1100 nm −1 (Figure 1a), which was assigned to the vibration band of P-O (1080 cm −1 ), W=O (982 cm −1 ), W-O c -W (corner-sharing, 897 cm −1 ) and W-O e -W (edge-sharing, 806 cm −1 ), respectively [50]. For IL supported silica catalysts (Figure 1b-d), the main vibration peaks for Keggin units could also be found as well as the symmetrical stretching vibration of Si-O-Si (800 cm −1 ) and tetrahedron bending vibration of [SiO 4 ] (469 cm −1 ) [51]. In addition, C-H stretching vibration peaks of ionic liquid cations appeared in the range of 2800 to 2900 cm −1 . These results indicated that the ionic liquid [C 16 mim] 3 PW 12 O 40 was successfully introduced to the SiO 2 . Figure 2 shows the Raman spectra of various samples. For C 16 PW (Figure 2a), two distinct peaks at 1006 and 991 cm −1 could be observed, which was respectively ascribed to ν s (W=O) and ν as (W=O) in the Keggin units [52]. After introduction to silica (Figure 2b-d), the two peaks were red-shifted in the Raman spectra, which may be due to the reduced interaction between the bonds of W=O. No obvious bands for Fe 3 O 4 was observed in the Raman spectra, which could be attributed to the low Raman intensity of Fe 3 O 4 with a low content. These results also indicated that ionic liquid was successfully supported, and retained its structure.  Figure 2 shows the Raman spectra of various samples. For C16PW (Figure 2a), two distinct peaks at 1006 and 991 cm −1 could be observed, which was respectively ascribed to νs(W=O) and νas(W=O) in the Keggin units [52]. After introduction to silica (Figure 2b-d), the two peaks were red-shifted in the Raman spectra, which may be due to the reduced interaction between the bonds of W=O. No obvious bands for Fe3O4 was observed in the Raman spectra, which could be attributed to the low Raman intensity of Fe3O4 with a low content. These results also indicated that ionic liquid was successfully supported, and retained its structure.    Figure 2 shows the Raman spectra of various samples. For C16PW (Figure 2a), two distinct peaks at 1006 and 991 cm −1 could be observed, which was respectively ascribed to νs(W=O) and νas(W=O) in the Keggin units [52]. After introduction to silica (Figure 2b-d), the two peaks were red-shifted in the Raman spectra, which may be due to the reduced interaction between the bonds of W=O. No obvious bands for Fe3O4 was observed in the Raman spectra, which could be attributed to the low Raman intensity of Fe3O4 with a low content. These results also indicated that ionic liquid was successfully supported, and retained its structure.  Ultraviolet-visible diffuse reflectance spectroscopy is performed to investigate the electron transfer in different samples ( Figure 3). For the ionic liquid C 16 PW (Figure 3a), the absorption peaks could be observed around 268 and 208 nm assigned to the charge transfer of O 2− →W 6+ and O→P [53]. For the ionic liquid supported materials, the broad and strong absorption peaks could be found in the range 200-400 nm, which may be due to the charge transfer induced by the excitation electron from the valence band to the conduction band [54]. The results also indicated the successful synthesis of the ionic liquids supported materials. fer of O →W and O→P [53]. For the ionic liquid supported materials, the broad and strong absorption peaks could be found in the range 200-400 nm, which may be due to the charge transfer induced by the excitation electron from the valence band to the conduction band [54]. The results also indicated the successful synthesis of the ionic liquids supported materials. XPS spectra are employed to investigate the effect of ball milling method and Fe3O4 incorporation on the chemical properties on the elements of various samples ( Figure 4). For the W 4f core-level spectrum of C16PW/SiO2 (Figure 4a), two peaks at the binding energy of 37.38 eV and 35.23 eV could be found, which was ascribed to W 4f5/2 and W 4f7/2 respectively. For the sample C16PW/SiO2-BM, the peak for the W 4f5/2 and W 4f7/2 could be observed at the same position (Figure 4b), indicating no change on the valence state of tungsten after ball milling. However, for the sample C16PW/SiO2-Fe-BM (Figure 4c), the binding energy for W 4f5/2 and W 4f7/2 was increased to 38.18 eV and 36.03 eV, indicating the electron transfer between Fe3O4 and supported ionic liquid.  (Table 1). Moreover, the incorporation of Fe 3 O 4 during the grinding process could further increase the specific surface area, which is beneficial for the ODS process.  N2 adsorption-desorption isotherms of various samples are presented to study the porous property ( Figure 5). According to the classification of the IUPAC, all the samples exhibited a typical type-Ⅳ curve [55]. For the sample C16PW/SiO2 (Figure 5a), a H3 hysteresis loop appeared in 0.55 < p/po < 0.95, indicating the presence of a stacking hole in the catalyst. For C16PW/SiO2-BM and C16PW/SiO2-Fe-BM (Figure 5b,c), a H2 hysteresis loop could be found in the relative pressure of 0.45-1.0, indicating the slit pores existing in the samples. These results demonstrated that the ball milling method could reduce the particle size and increase the specific surface area (Table 1). Moreover, the incorporation of

ODS Performance
The desulfurization performance of different samples is presented in Figure 6. Under the same reaction conditions, the sulfur removal for C16PW/SiO2, C16PW/SiO2-BM, and C16PW/SiO2-Fe-BM could reach 24.9%, 98.5%, and 99.1%, respectively. It could be found that the ODS performance of the samples was improved after ball milling. Moreover, the desulfurization efficiency was also increased after the incorporation of Fe3O4, which was ascribed to the higher surface area (Table 1)

ODS Performance
The desulfurization performance of different samples is presented in Figure 6. Under the same reaction conditions, the sulfur removal for C 16 PW/SiO 2 , C 16 PW/SiO 2 -BM, and C 16 PW/SiO 2 -Fe-BM could reach 24.9%, 98.5%, and 99.1%, respectively. It could be found that the ODS performance of the samples was improved after ball milling. Moreover, the desulfurization efficiency was also increased after the incorporation of Fe 3 O 4 , which was ascribed to the higher surface area (Table 1).

ODS Performance
The desulfurization performance of different samples is presented in Figure 6. Under the same reaction conditions, the sulfur removal for C16PW/SiO2, C16PW/SiO2-BM, and C16PW/SiO2-Fe-BM could reach 24.9%, 98.5%, and 99.1%, respectively. It could be found that the ODS performance of the samples was improved after ball milling. Moreover, the desulfurization efficiency was also increased after the incorporation of Fe3O4, which was ascribed to the higher surface area (Table 1) In order to study the desulfurization activity of the sample C 16 PW/SiO 2 -Fe-BM toward different sulfur-containing substrates, the typical sulfide DBT and its derivatives (4-MDBT and 4,6-DMDBT) are carried out under the same conditions (Figure 7). For DBT, the desulfurization rate can reach 100% within 7 h. On the other hand, the desulfurization rate toward the aromatic sulfide 4-MDBT and 4,6-DMDBT can reach 90.2% and 80.5% in 8 h, respectively. The difference on ODS performance may be ascribed to the steric hindrance in the sulfur compounds [56]. The steric hindrance of sulfur compounds followed the order: 4,6-DMDBT > 4-MDBT > DBT [57]. Compared to DBT, 4-MDBT and 4,6-DMDBT have one or two more methyl groups on the benzene ring, respectively. Their steric hindrance is greater than that of DBT, making it difficult for sulfur atoms to be oxidized during the reaction process. In order to study the desulfurization activity of the sample C16PW/SiO2-Fe-BM toward different sulfur-containing substrates, the typical sulfide DBT and its derivatives (4-MDBT and 4,6-DMDBT) are carried out under the same conditions (Figure 7). For DBT the desulfurization rate can reach 100% within 7 h. On the other hand, the desulfurization rate toward the aromatic sulfide 4-MDBT and 4,6-DMDBT can reach 90.2% and 80.5% in 8 h, respectively. The difference on ODS performance may be ascribed to the steric hindrance in the sulfur compounds [56]. The steric hindrance of sulfur compounds followed the order: 4,6-DMDBT > 4-MDBT > DBT [57]. Compared to DBT, 4-MDBT and 4,6-DMDBT have one or two more methyl groups on the benzene ring, respectively. Their steric hindrance is greater than that of DBT, making it difficult for sulfur atoms to be oxidized during the reaction process. The recycling performance of the catalyst is another important factor in the desulfurization process ( Figure 8). The recycling process of the typical sample C16PW/SiO2-Fe-BM was performed as follows: The upper oil phase was decanted directly after the reaction, and the lower catalyst was dried at 50 °C overnight. Afterwards, the next run was evaluated after adding the fresh model oil and air flow. It can be observed that the sulfur removal could still reach 100% after being recycled four times. The recycling performance of the catalyst is another important factor in the desulfurization process (Figure 8). The recycling process of the typical sample C 16 PW/SiO 2 -Fe-BM was performed as follows: The upper oil phase was decanted directly after the reaction, and the lower catalyst was dried at 50 • C overnight. Afterwards, the next run was evaluated after adding the fresh model oil and air flow. It can be observed that the sulfur removal could still reach 100% after being recycled four times.

Magnetic Hysteresis Test
In order to investigate the magnetism of the catalyst, the magnetization of Fe 3 O 4 and C 16 PW/SiO 2 -Fe-BM is measured through an external magnetic field at room temperature ( Figure 9). As shown in Figure 9A,B, the saturation magnetization of C 16 PW/SiO 2 -Fe-BM (0.8 emg/g) is significantly lower than that of Fe 3 O 4 (82 emg/g). This result was attributed to the low content of Fe 3 O 4 and the addition of non-magnetic material (C 16 PW/SiO 2 ) in the hybrid materials. Moreover, no hysteresis phenomenon was found in Figure 9A, indicating the superparamagnetism of the prepared samples. To further confirm the above results, the separation expriment of the sample in oil is carried out in this study ( Figure 9C). It was obviously found that the catalyst could be easily separated from the oil phase through the action of an external magnetic field.

Magnetic Hysteresis Test
In order to investigate the magnetism of the catalyst, the magnetization of Fe3O4 and C16PW/SiO2-Fe-BM is measured through an external magnetic field at room temperature ( Figure 9). As shown in Figure 9A,B, the saturation magnetization of C16PW/SiO2-Fe-BM (0.8 emg/g) is significantly lower than that of Fe3O4 (82 emg/g). This result was attributed to the low content of Fe3O4 and the addition of non-magnetic material (C16PW/SiO2) in the hybrid materials. Moreover, no hysteresis phenomenon was found in Figure 9A, indicating the superparamagnetism of the prepared samples. To further confirm the above results, the separation expriment of the sample in oil is carried out in this study ( Figure 9C). It was obviously found that the catalyst could be easily separated from the oil phase through the action of an external magnetic field.

Magnetic Hysteresis Test
In order to investigate the magnetism of the catalyst, the magnetization of Fe3O4 and C16PW/SiO2-Fe-BM is measured through an external magnetic field at room temperature ( Figure 9). As shown in Figure 9A,B, the saturation magnetization of C16PW/SiO2-Fe-BM (0.8 emg/g) is significantly lower than that of Fe3O4 (82 emg/g). This result was attributed to the low content of Fe3O4 and the addition of non-magnetic material (C16PW/SiO2) in the hybrid materials. Moreover, no hysteresis phenomenon was found in Figure 9A, indicating the superparamagnetism of the prepared samples. To further confirm the above results, the separation expriment of the sample in oil is carried out in this study ( Figure 9C). It was obviously found that the catalyst could be easily separated from the oil phase through the action of an external magnetic field.

Synthesis of Magnetic Fe3O4
A total of 3.0 g FeCl3·6H2O was dissolved in 60 mL of ethylene glycol with continuous stirring for 30 min. Afterwards, 6.0 g CH3COONa·3H2O was added to the above solution with continuous stirring for 30 min. The mixed solution was transferred to a reaction kettle and kept at 200 °C for 12 h. Finally, the precipitate was collected by a magnet, washed with water and alcohol for several times, and dried at 50 °C in a vacuum.

Synthesis of Magnetic Fe 3 O 4
A total of 3.0 g FeCl 3 ·6H 2 O was dissolved in 60 mL of ethylene glycol with continuous stirring for 30 min. Afterwards, 6.0 g CH 3 COONa·3H 2 O was added to the above solution with continuous stirring for 30 min. The mixed solution was transferred to a reaction kettle and kept at 200 • C for 12 h. Finally, the precipitate was collected by a magnet, washed with water and alcohol for several times, and dried at 50 • C in a vacuum.

Synthesis of Catalysts
Ionic liquid [C 16 mim] 3 PW 12 O 40 (C 16 PW) was synthesized according to the previous report [58]. In a typical process, 0.3 g of C 16 PW was dissolved in 4 mL of acetonitrile with continuous stirring. The above solution was dropwise added to 26 mL of deionized water at 30 • C under stirring for 10 min. Subsequently, 2 mL of TEOS and 0.5 mL of aqueous ammonia was dropwise added to the above mixture. After stirring for 3 h, the resulting solution was dried at 50 • C overnight to obtain the supported ionic liquid C 16

Synthesis of Catalysts
Ionic liquid [C16mim]3PW12O40 (C16PW) was synthesized according to the previous report [58]. In a typical process, 0.3 g of C16PW was dissolved in 4 mL of acetonitrile with continuous stirring. The above solution was dropwise added to 26 mL of deionized water at 30°C under stirring for 10 min. Subsequently, 2 mL of TEOS and 0.5 mL of aqueous ammonia was dropwise added to the above mixture. After stirring for 3 h, the resulting solution was dried at 50 °C overnight to obtain the supported ionic liquid C16PW/SiO2. Finally, C16PW/SiO2 was mechanically milled with magnetic Fe3O4 (1 wt.%) with a rotation speed of 200 rpm for 3 h to obtain C16PW/SiO2-Fe-BM (Scheme 1). For comparison, supported ionic liquid C16PW/SiO2 without Fe3O4 is obtained with the same ball milling method and is denoted as C16PW/SiO2-BM. Scheme 1. The preparation process of samples.

Oxidative Desulfurization Process
The model oil was prepared by dissolving DBT, 4-MDBT, and 4,6-DMDBT in dodecane with a corresponding S-concentration of 200 ppm, respectively. In a typical reaction process, 0.01 g of the prepared samples and 20 mL of model oil was added in a threenecked flask equipped with a magnetic stirrer and a heater set at a certain temperature. Then, the air was injected to the reactor at a flow rate of 100 mL/min. The residual sulfur content in the model oil was determined by gas chromatography on Shimadzu GC 2010 Plus (SH-Rtx-5, 30 m × 0.25 mm × 0.25 μm). The injector temperature was 250 °C, and the detector temperature was 300 °C. The temperature of the GC process started at 100 °C and rose to 250 °C at 25 °C/min.

Characterization
Fourier transform infrared spectra (FT-IR) of the samples (KBr pellets) were recorded on a Nicolet FT-IR spectrophotometer (Nexus 470, Thermo Electron Corporation, Waltham, MA, USA) with a range of 400-4000 cm −1 at room temperature. Crystalline structures of hybrid materials were analyzed by the XRD technique using an X-ray diffractometer (Bruker D8 ADVANCE, Billerica, MA, USA) using Cu Kα radiation at 30 kV (λ = 1.54 Å). The Raman spectrum was obtained on a DXR Raman microscope (Waltham, MA, USA) with a 532 nm laser source under ambient conditions at room temperature. The Scheme 1. The preparation process of samples.

Oxidative Desulfurization Process
The model oil was prepared by dissolving DBT, 4-MDBT, and 4,6-DMDBT in dodecane with a corresponding S-concentration of 200 ppm, respectively. In a typical reaction process, 0.01 g of the prepared samples and 20 mL of model oil was added in a three-necked flask equipped with a magnetic stirrer and a heater set at a certain temperature. Then, the air was injected to the reactor at a flow rate of 100 mL/min. The residual sulfur content in the model oil was determined by gas chromatography on Shimadzu GC 2010 Plus (SH-Rtx-5, 30 m × 0.25 mm × 0.25 µm). The injector temperature was 250 • C, and the detector temperature was 300 • C. The temperature of the GC process started at 100 • C and rose to 250 • C at 25 • C/min.

Characterization
Fourier transform infrared spectra (FT-IR) of the samples (KBr pellets) were recorded on a Nicolet FT-IR spectrophotometer (Nexus 470, Thermo Electron Corporation, Waltham, MA, USA) with a range of 400-4000 cm −1 at room temperature. Crystalline structures of hybrid materials were analyzed by the XRD technique using an X-ray diffractometer (Bruker D8 ADVANCE, Billerica, MA, USA) using Cu Kα radiation at 30 kV (λ = 1.54 Å). The Raman spectrum was obtained on a DXR Raman microscope (Waltham, MA, USA) with a 532 nm laser source under ambient conditions at room temperature. The wavenumbers range from 400 to 4000 cm −1 . X-ray photoelectron spectroscopy (XPS) was recorded on a spectrophotometer (Thermo VG, Waltham, MA, United States) with 30 eV energy to explore the scanning of the narrow spectra of the elements on the surface. Textural properties were measured by Nitrogen adsorption-desorption isotherms. It was carried out using Micromeritics ASAP 2000 equipment (Norcross, GA, USA) at liquid N 2 temperature. The content of sulfur-containing left in oil phase was characterized by Gas chromatographymass spectrometry (GC-MS7890/5975C-GC/MSD, Agilent, Santa Clara, CA, USA).

Conclusions
To sum up, magnetic ionic liquid supported silica (C 16 PW/SiO 2 -Fe-BM) was synthesized by ball milling and applied in the removal of organic sulfides with oxygen as the oxidant. The experimental results indicated that polyoxometalate based ionic liquids was successfully introduced to the silica as well as the Fe 3 O 4 particle to give the prepared sample superparamagnetism. In the aerobic oxidative desulfurization process, the hybrid catalyst exhibited good catalytic performance, which could achieve complete removal of dibenzothiophene in 7 h. Without any further regeneration procedure, the sulfur removal could still reach 100% after being recycled four times. The desulfurization performance of various aromatic sulfur compounds decreased in the order of DBT > 4-MDBT > 4,6-DMDBT.

Data Availability Statement:
The data presented in this study are available in article.