1 Introduction

In the past years, new energy sources such as solar power and hydrogen energy in the auto industry have been springing up. However, fuel oil still occupies a dominant position. Sulfur compounds in fuel oil generate sulfur oxides (SOx) after combustion, and this is a main source of acid rain, haze and other environmental issues (Zhang et al. 2017a; Ibrahim et al. 2017). Thus, most countries have implemented strict standards to limit the sulfur content in fuel oil to below 10 ppm. Up to now, conventional hydrodesulfurization (HDS) is efficient in removing thiols, sulfides and disulfides, but not efficient for the removal of heterocyclic sulfur compounds such as dibenzothiophene and its derivatives, especially 4,6-dimethyldibenzothiophene (Zhang et al. 2017b; Li et al. 2016). Accordingly, the development of alternative desulfurization progress is in the ascendant, including biodesulfurization (Soleimani et al. 2007), adsorption desulfurization (Ren et al. 2018; Yang et al. 2018), extraction desulfurization (Rafiee et al. 2016) and oxidative desulfurization (Akopyan et al. 2015; Zhang et al. 2014). Among these, oxidative desulfurization is considered to be the most promising technology due to its mild operating conditions and high removal efficiency for the heterocyclic sulfur compounds mentioned above (Zhang et al. 2013).

Ionic liquids (ILs) with unique physicochemical properties have been widely employed as solvents, extractants, templates and precursors (Zhou and Qu 2017; Dai et al. 2017). In the desulfurization process, ionic liquids are often used as extractants (Paduszyński et al. 2017; Moghadam et al. 2017), but the sulfur removal is not so satisfactory. Thus, polyoxometalates (POMs), a family of transition metal–oxide clusters with specific physical properties and controllable redox and properties, are often introduced as anions to the family of specific ILs (Zhao et al. 2012; Zhu et al. 2013). In the oxidative desulfurization, POM-based ILs could not only capture the sulfur compounds from oil but also activate the oxidant (e.g., hydrogen peroxide) to oxidize the sulfur compounds to sulfones. The desulfurization performance of POM-based ILs was desirable, but the catalytic systems often suffered from some problems such as high dosage and low specific area of ILs, difficulty in the separation of catalyst (Xun et al. 2016; Li et al. 2015). Thus, immobilization of POM-based ILs on a suitable carrier (e.g., silica) is a good method to solve the above problems.

Hence, in this work, a series of molybdenum-containing silica-supported ionic liquids [C4mim]3PMo12O40/SiO2 were designed to achieve the advantages of both the POM materials and heterogeneous catalysis. The hybrid materials [C4mim]3PMo12O40/SiO2 from a one-pot hydrothermal method were highly efficient in the oxidative desulfurization of 4,6-dimethyldibenzothiophene and other sulfur compounds, where no organic solvents were added as extractants. Moreover, the samples were also systematically characterized by XRD, XPS, FT-Raman, FT-IR, UV–Vis and N2 adsorption–desorption analysis.

2 Experimental section

2.1 Chemicals and materials

n-octane (CP grade), H3PMo12O40·26H2O (AR grade), acetonitrile (CH3CN, AR grade), tetraethylorthosilicate (TEOS, AR grade), hydrogen peroxide (H2O2, 30 wt%) and ammonia (NH3·H2O, 25 wt%) were acquired from Sinopharm Chemical Reagent Co. Ltd., China. Benzothiophene (BT, 99%), dibenzothiophene (DBT, 98%), 4-methyldibenzothiophene (4-MDBT, 98%) and 4,6-dimethyldibenzothiophene(4,6-DMDBT, 99%) were purchased from Sigma-Aldrich. [C4mim]Cl (99%) was obtained from Shanghai Chenjie Chemical Co. Ltd., China. All the reagents were used directly as received.

2.2 Synthesis of the catalysts

The ionic liquid [C4mim]3PMo12O40 was synthesized according to a previous study (Rajkumar and Rao 2008). Then, silica-supported ionic liquid was synthesized by a one-pot hydrothermal method as follows: Firstly, 0.17 g of [C4mim]3PMo12O40 was dissolved in 4 mL of acetonitrile at 50 °C under continuous stirring. After that, the solution above was added dropwise to 26 mL of deionized water under continuous stirring for 10 min and followed by addition of 2 mL of TEOS. Afterward, 0.5 mL of 25% aqueous ammonia was added to the mixture. After stirring for 3 h at room temperature, the mixture was transferred into the reactor and kept at 120 °C for 24 h. Then, the resultant was filtered and washed with deionized water three times and treated at 150 °C for 3 h. For comparison, other materials with different Mo:Si molar ratios and different treating temperatures were prepared by a similar method and denoted as x-[C4mim]3PMo/SiO2-y (x = 0.05, 0.1 and 0.2, and y = 150, 200 and 250 °C).

2.3 Characterization

FT-IR spectra of the catalysts were recorded on a Nicolet Nexus 470 spectrometer using KBr pellets (Thermo Electron Corporation, USA). SEM imaging was performed on a JEOL JEM-7001F field-emission microscope (JEOL Corporation, Japan). XPS was collected on a PHI 530 with a monochromatic Mg Kα source (Ulvac-Phi Corporation, Japan). Raman spectroscopy was studied on a DXR Raman microscope (Thermo Fisher Scientific, USA) with a 532-nm excitation laser power. UV–visible spectra were recorded on a UV–Vis spectrometer (UV-2450, Shimadzu, Japan). X-ray diffraction patterns were recorded with a Bruker D8 diffractometer using Cu Kα radiation (\(\lambda\) = 1.5418 Å). The N2 absorption–desorption isotherms were collected on a TriStar II 3020 surface area and porosity analyzer (Micromeritics Corporation, USA). The oxidation products of 4,6-DMDBT were studied by GC–MS on an Agilent 7890A (Agilent, USA).

2.4 Catalytic activity test

Model oil was obtained by dissolving the desired amount of BT, DBT, 4-MDBT and 4,6-DMDBT in n-octane to achieve a corresponding S-content of 250 ppm. A mixture of 0.01 g catalyst and 5 mL model oil was added into a two-necked reactor equipped with a magnetic stirrer and a condenser. Afterward, oxidation of sulfur compounds was carried out by adding a desired amount of H2O2 to the mixture above. After the reaction, the residual sulfur content was measured on GC-FID HP-5 MS column, 30 m×320 µm i.d.×0.25 µm; FID: Agilent, rising from 100 to 160 °C at a heating rate of 20 °C/min and then rising to 230 °C at a heating rate of 25 °C/min. The sulfur removal (%) was calculated as follows:

$${\text{Sulfur removal }}\left( \% \right) = \frac{{{\text{Initial sulfur content }} - {\text{Residual sulfur content}} }}{{{\text{Initial sulfur content}} }} \times 100\%$$

3 Results and discussion

3.1 Characterization of samples

Figure 1 shows the FT-IR spectra of supported [C4mim]3PMo and IL. For [C4mim]3PMo (Fig. 1a), the peaks around 3145 and 3107 cm−1 belong to the stretching vibration of C–H of the alkyl chain. The peaks around 1564 and 1465 cm−1 are ascribed to the stretching vibration of C=N and scissoring vibration of C–H, respectively (Xun et al. 2014). In addition, the four characteristic absorption peaks of the Keggin structure of ionic liquid can be clearly observed in the range of 1100 to 750 cm−1, belonging to νas (P–Oa), vas (Mo–Ot), vas (Mo–Ob–Mo) and vas (Mo–Oc–Mo), respectively (Dong et al. 2014). For IL supported on silica, the characteristic absorption peaks of the silica support were found around 3430 and 963 cm−1 caused by the stretching vibration of Si–OH on the silica surface, and the characteristic peaks at 1085, 807 and 470 cm−1 were attributed to the bending vibration of νas (Si–O–Si), νas (Si–O–Si) and silicate ions, respectively. The absorption bands for the POM anion could not be detected, as they were overlapped by the characteristic absorption of silica.

Fig. 1
figure 1

FT-IR spectra of various samples: a [C4mim]3PMo, b 0.2[C4mim]3PMo/SiO2-250, c 0.1[C4mim]3PMo/SiO2-250, d 0.05[C4mim]3PMo/SiO2-250, e 0.1[C4mim]3PMo/SiO2-200 and f 0.1[C4mim]3PMo/SiO2-150

Full size image

Further structural information about the various samples was obtained from Raman scattering spectroscopy (Fig. 2). For [C4mim]3PMo, the characteristic absorption peaks located at 1005 and 997 cm−1 could be clearly detected. These were ascribed to the vibration of vs (Mo–Ot) and vas (Mo–Ot), respectively. The bands ranging from 920 to 860 cm−1 and from 620 to 580 cm−1 could be ascribed to the vibration of vs (Mo–Ob–Mo) and vs (Mo–Oc–Mo), respectively. In addition, an strong absorption peak around 247 cm−1 was caused by the vibration of vs (Mo–Oa) (Sainero et al. 2001). For the IL-supported materials (Fig. 2b–f), as the IL contents increased from 0.05 to 0.2, the intensity of the characteristic absorption peaks around 1000 cm−1 also increased gradually. Other characteristic absorption peaks of the Keggin structure could also be found, indicating that the IL was successfully introduced. On the other hand, for silica-supported IL with different treatment temperatures, no peak shift was observed, indicating that the original structure of the ionic liquid did not change.

Fig. 2
figure 2

Raman spectra of various samples: a [C4mim]3PMo, b 0.2[C4mim]3PMo/SiO2-250, c 0.1[C4mim]3PMo/SiO2-250, d 0.05[C4mim]3PMo/SiO2-250, e 0.1[C4mim]3PMo/SiO2-200 and f 0.1[C4mim]3PMo/SiO2-150

Full size image

Figure 3 shows the electron transfer properties of various samples by UV–visible spectroscopy. For the supported silica, no obvious absorption was found. For [C4mim]3PMo, two absorption bands could be observed in the region of 220–320 nm, which were ascribed to the electron transfer from O–P and a ligand–metal charge transfer (O2–→Mo6+), respectively (Liu et al. 2013). For the supported IL materials, it is observed that the intensity of the two characteristic absorption peaks mentioned changes clearly. As IL content increased, the intensity of characteristic absorption peaks in the UV–Vis range increased gradually.

Fig. 3
figure 3

UV-vis spectra of various materials

Full size image

Figure 4 reveals the wide-angle XRD patterns of various materials. For [C4mim]3PMo, a series of sharp peaks can be detected in the 2θ range of 5°–35°. For supported IL materials with different IL contents, only a broad peak of amorphous silica could be observed instead of the diffraction of [C4mim]3PMo, which indicates a high dispersion of IL in the silica. However, for the sample 0.2[C4mim]3PMo/SiO2-250, except for the amorphous peaks of amorphous silica, the diffraction peaks of [C4mim]3PMo can be found around 26.5°, which could be ascribed to the aggregated IL with a high loading content.

Fig. 4
figure 4

XRD patterns of various materials: a [C4mim]3PMo, b 0.2[C4mim]3PMo/SiO2-250, c 0.1[C4mim]3PMo/SiO2-250, d 0.05[C4mim]3PMo/SiO2-250, e 0.1[C4mim]3PMo/SiO2-200 and f 0.1[C4mim]3PMo/SiO2-150

Full size image

To investigate the surrounding chemical environment and volatile valence of the active Mo atoms in the sample, XPS spectra were obtained (Fig. 5). As shown in Fig. 5a, the peaks located at 284.7, 398.8, 104.2, 134.5 and 233 eV are assigned to C 1s, N 1s, Si 2p, P 2p and Mo 3d respectively, indicating that [C4mim]3PMo was successfully introduced. In addition, two distinct peaks belonging to Mo 3d5/2 and Mo 3d3/2 could be detected with binding energies of 232.7 and 236.2 eV, respectively, indicating the oxidation states of Mo6+ (Zhang et al. 2012). These results also demonstrated the ionic liquid maintained its original structure after introduction to silica.

Fig. 5
figure 5

XPS analysis survey (a) and Mo 3d core-level spectrum (b)

Full size image

Figure 6 shows N2 absorption–desorption isotherms (Fig. 6a) and BJH pore size distribution curves (Fig. 6b) of various samples. In Fig. 6a, all the samples exhibited type IV with a clear H1-type hysteresis loop according to the IUPAC classification, indicating the presence of mesopores. Moreover, the hysteresis loop of the N2 adsorption–desorption isotherm appears in the high relative pressure (P0/P > 0.8), which was attributed to the accumulation of particles during the formation of the samples (Liu et al. 2012). The pore size distribution of the catalysts was measured by the BJH model. As shown in Fig. 6b, the pore size distributions of all the samples are centered in the range of 10–20 nm. The specific surface areas and pore volumes of the samples are shown in Table 1. For the samples with the same treatment temperature (Entry 1–3, Table 1), the specific surface area and pore volume parameters decreased as the IL content increased. On the other hand, for the samples with different treatment temperatures (Entry 3–5, Table 1), the specific surface area increased from 217.9 to 277.3 m2 g−1, and the pore volume increased from 0.87 to 1.03 cm3 g−1, which may be caused by the compactness of the catalyst structure at higher temperature, leading to an increase of specific surface area and pore volume.

Fig. 6
figure 6

N2 adsorption–desorption isothermals (a) and BJH pore size distribution (b) of various materials: a 0.2[C4mim]3PMo/SiO2-250, b 0.1[C4mim]3PMo/SiO2-250, c 0.05[C4mim]3PMo/SiO2-250, d 0.1[C4mim]3PMo/SiO2-200 and e 0.1[C4mim]3PMo/SiO2-150

Full size image
Table 1 Structural properties of various samples
Full size table

3.2 Catalytic performance

Figure 7 displays the performance of various samples on the removal of 4,6-DMDBT under the same conditions. For 0.1[C4mim]3PMo/SiO2 with different treatment temperatures, it can be found that the catalytic performance increased with higher treatment temperature, which is ascribed to the increase of the specific surface area and pore volume, leading to the sufficient contact of the catalyst active centers and the sulfur compounds. For [C4mim]3PMo/SiO2-250 with different IL contents, it can be observed that the catalytic performance was improved as the IL content increased. Although the increase of the IL content resulted in a decrease of the specific surface area and pore volume, the increase of active centers would also accelerate the reaction. The sulfur removal with the sample 0.1[C4mim]3PMo/SiO2-250 could reach 100% in 40 min with an oxidant/sulfur molar ratio (O/S) =3.

Fig. 7
figure 7

Influence of various samples on the catalytic performance. Reaction conditions: mass of catalyst = 0.01 g, T = 60 °C, t = 60 min, molar ratio O/S = 3

Full size image

Since real oil contains a variety of thiophene sulfides, the catalytic oxidation performance of various sulfur compounds including BT, DBT, 4-MDBT and 4,6-DMDBT is investigated using 0.1[C4mim]3PMo/SiO2-250 as the typical catalyst in this study (Fig. 8). It is clearly observed that the desulfurization of various sulfur compounds decreased in the order of 4,6-DMDBT > 4-MDBT > DBT > BT. These results can be mainly accounted for by the different electron densities on the sulfur atom of 4,6-DMDBT, 4-MDBT, DBT and BT which were 5.760, 5.759, 5.758 and 5.739, respectively (Xu et al. 2009). 4,6-DMDBT has the largest electron density among the substrates, making it easy to be oxidized.

Fig. 8
figure 8

Influence of various substrates on the catalytic performance. Reaction conditions: mass of catalyst = 0.01 g, T = 60 °C, t = 60 min, molar ratio O/S = 3

Full size image

3.3 Analysis of oxidation product

To investigate the oxidation products of 4,6-DMDBT, GC-MS analysis was performed as shown in Fig. 9. After the reaction, the upper oil phase was directly decanted, and the lower catalyst was extracted by CCl4 for GC-MS analysis. In Fig. 9a, it is observed that there is no peak for 4,6-DMDBT, and only an intense peak for n-tetradecane at 4.3 min could be detected. On the other hand (Fig. 9b), the peaks for 4,6-DMDBTO2 (m/z = 244.1) and 4,6-DMDBT (m/z = 208.1) could be observed at 16.4 min and 9.9 min, respectively. These results demonstrated that 4,6-DMDBT was firstly adsorbed by the catalyst and subsequently oxidized to 4,6-DMDBTO2.

Fig. 9
figure 9

GC-MS analysis of (a) upper oil phase after reaction and (b) extracted phase from the catalyst after reaction

Full size image

3.4 Reusability of the catalyst

It is also important to investigate the recycling capability of the reaction system (Fig. 10). After the first reaction, the upper clear oil phase was directly decanted for separation, and the used catalyst (0.1[C4mim]3PMo/SiO2-250) was dried at 50 °C overnight. Under the same conditions, fresh H2O2 and model oil were added for the next run. It is worth noting that the sulfur removal could still reach 93% after cycling for seven cycles. The good recycling ability was ascribed to the stability of silica-supported IL.

Fig. 10
figure 10

Recycling performance of hybrid materials. Reaction conditions: mass of catalyst = 0.01 g, T = 60 °C, t = 40 min, molar ratio O/S = 3

Full size image

4 Conclusion

Silica-supported ionic liquid was successfully prepared by a one-pot hydrothermal method and employed in the heterogeneous oxidative desulfurization of various sulfur compounds. The characterization results indicated that the POM-based IL was uniformly dispersed on the silica matrix and maintained its structural integrity. The sample 0.1[C4mim]3PMo/SiO2-250 had excellent catalytic activity for the removal of various sulfur compounds without any solvents and could still reach sulfur removal of 93% after seven cycles. The oxidative efficiency of different substrates decreased in the order of 4,6-DMDBT > 4-MDBT > DBT > BT.