As the worldwide demand for fuel continues to increase rapidly, the industrial desulfurization of diesel fuel faces increasing technological challenges to comply with environmental concerns and legal requirements [1, 2].
Hydrodesulfurization (HDS) is a mature technology and efficiently removes aliphatic and acyclic sulfur compounds (e.g. thiols, sulfides, and disulfides) [3-5] but not thiophene sulfides, such as dibenzothiophene and its derivatives [6-8]. Additionally, further improvements to the HDS process for deep desulfurization are limited to increasingly harsh operating conditions with high associated costs [9, 10]. Hence, it is critical to develop supplementary strategies for deep desulfurization, including extraction [11-14], adsorption [15, 16], oxidation [17-22], and biodesulfurization [23]. Oxidative desulfurization (ODS) can remove aromatic organosulfur compounds under mild conditions. As such, it is considered as a promising technique for deep desulfurization.
Ionic liquids (ILs) are organic salts with low melting points and vapor pressures, and have attracted wide attention owing to their unique physicochemical properties, including negligible volatility, low toxicity, excellent thermal stability, good solubility characteristics, and the variety of available structures [24]. Task-specific IL anions that are tethered to functional groups (e.g. polyoxometalates) show great potential for the modification and control of different chemical and physical functions in a material [25-27]. Polyoxometalates (POMs) are a vast class of well-defined, early transition metal oxyanion clusters that exhibit distinctive physicochemical properties, including adjustability of composition, size, shape, acid-base properties, and redox potential [28-32]. As such, POMs are extensively used as green catalysts, particularly in oxidation reactions [33, 34]. A number of studies have demonstrated that POM-based ILs exhibited efficient catalytic ODS [35-37]. However, the widespread use of these materials for homogeneous desulfurization in an industrial setting has been hindered because of their intrinsic liquid nature and high viscosity, which leads to difficulties in separation and recovery, recyclability, cost, and the large amounts of ILs that are required [38].
To overcome some of these problems, considerable efforts have been made to introduce active POM-based species into heterogeneous desulfurization systems, where materials such as zeolites, porous carbon, clays, siliceous materials, and metal-organic frameworks were employed as a support [39-44]. Among these supports, mesoporous silica possesses a number of desirable properties, including a stable mesoporous structure, a large surface area and pore volume, well-ordered pore structure, and narrow pore-size distributions. Moreover, silica has no inherent catalytic property and can be used as a support that can be functionalized by framework substitution or post-synthesis surface modification. As such, POM-based IL-supported mesoporous silica has been fabricated for ODS, but required relatively excessive oxidants. Additionally, the requirement for organic solvents (e.g. acetonitrile and dimethylformamide) may limit future applications of these materials [45, 46].
In this work, active tungsten species were introduced directly into mesoporous silica from a POM-based IL ([C16mim]3PW12O40), where the imidazole cation and polyoxometalate anion acted as template and metal source, respectively. The hybrid materials (W-SiO2) were highly effective on removing organosulfur compounds under mild conditions and were separated easily for recycling. Additionally, organic solvents were not required for the heterogeneous desulfurization. All samples were characterized using X-ray diffraction, Transmission electron microscopy, Raman spectroscopy, Fourier transform infrared spectra, Diffuse reflectance spectra, and Brunauer-Emmett-Teller analysis.
H2O2 (aqueous solution, 30 wt%), H3PW12O40·14H2O (AR grade), WO3 (AR grade), n-octane (AR grade), NH3·H2O (aqueous solution, 25 wt%), acetonitrile (AR grade), tetraethyl orthosilicate (TEOS, AR grade) and tetradecane (AR grade) were purchased from Sinopharm Chemical Reagent Co., Ltd. Benzothiophene (BT, 99%), dibenzothiophene (DBT, 98%) 1-dodecanethiol (DT, 99%), and 4, 6-dimethyldibenzothiophene (4, 6-DMDBT, 99%) were purchased from Sigma-Aldrich. [C16mim]Br was purchased from Shanghai Chengjie Chemical Co., Ltd. All the materials are used without further purification.
The POM-based IL [C16mim]3PW12O40 was prepared according to a literature procedure [47]. The mesoporous hybrid material (W-SiO2) was prepared using a one-pot method from an initial gel that consisted of TEOS:[C16mim]3PW12O40:H2O: NH3·H2O = 1:x:160:1.5 (x = 0.002, 0.004, 0.008). In a typical synthesis, [C16mim]3PW12O40 (0.14 g) was added into acetonitrile (4 mL) with constant stirring. Subsequently, TEOS (2 mL) and NH3·H2O (0.52 mL) were added into the mixture and stirred for 3 h. The resulting gel was filtered, washed with water (200 mL), dried overnight at 120 ℃, and then calcined at 550 ℃ with a heating rate of 5 ℃/min for 6 h. The samples were named based on the Si/W molar ratio used in the gel (W-SiO2-y, where y = Si/W = 10, 20 and 40).
The samples were ground into a fine powder for characterization. Fourier transform infrared (FT-IR) spectra were recorded with a Nicolet Nexus 470 FT-IR instrument using KBr pellets. Raman spectroscopy was recorded at ambient temperature using a DXR Raman microscope with an excitation wavelength of 532 nm from a laser source. Diffuse reflectance spectra (DRS) were measured between 200 and 800 nm with BaSO4 used as the reflectance standard using a UV-visible spectrometer (UV-2450, Shimadzu). The surface area of the samples was calculated using the Brunauer-Emmett-Teller (BET) method, while the pore-size distribution was obtained from the adsorption branches of the isotherms using the Barrett-Joyner- Halenda (BJH) method. Transmission electron microscopy (TEM) was recorded using a JEOL-JEM-2010 (JEOL, Japan) operated at 200 kV. The crystalline phases within the samples were analyzed using X-ray diffraction (XRD) spectroscopy with a Bruker D8 diffractometer with Cu Kα radiation (λ = 0.154 nm).
A model oil sample was prepared by dissolving DBT, BT, DT and 4, 6-DMDBT in n-octane with a corresponding sulfur content of 500, 250, 250 and 250 ppm, respectively. Desulfurization was performed in a flask equipped with a magnetic stirrer and condenser. In a typical run, a water bath was initially heated to 60 ℃ then allowed to stabilize. W-SiO2 (0.01 g) and H2O2 (20 μL) were added to the reactor and then the oil (5 mL) was injected.
The sulfur contents in the oil after the desulfurization test were analyzed using gas chromatography-flame ionization detector (GC-FID) with tetradecane as the internal standard (Agilent 7890A; HP-5, 30 m × 0.32 mm × 0.25 μm; FID: Agilent). The conversion of the sulfur compounds in the oil was used to calculate the removal of sulfur compounds.
The oxidized sulfur compounds were characterized using gas chromatography-mass spectrometer (GC-MS, Agilent 7890/ 5975C-GC/MSD; HP-5 MS column, 30 m × 250 μm × 0.25 μm; temperature program: 100 ℃ → temperature rising 15 ℃/min → 200 ℃ for 10 min).
FT-IR spectra of W-SiO2-y, WO3 and [C16mim]3PW12O40 are shown in Fig. 1. The spectrum of bulk WO3 (Fig. 1(1)) exhibited a strong and broad peak at approximately 820 cm−1 that was caused by the stretching vibrations of W-O-W. In addition to the C-H stretching vibrations of the alkyl chain (2950 to 2850 cm−1), [C16mim]3PW12O40 (Fig. 1(2)) exhibited four bands that were characteristic of Keggin units (νas(P-Oa) = 1080 cm−1, νas(W=Ot) = 978 cm−1, νas(W-Oc-W) = 898 cm−1 and νas(W-Oe-W) = 808 cm−1). The W-SiO2-y materials (Fig. 1(3)-(5)) exhibited a broad peak near 3500 cm−1 that was attributed to surface silanols and adsorbed water molecules [48]. The three peaks located at 1086, 812, and 460 cm−1 were caused by asymmetric νasym(Si-O-Si), symmetric νsym(Si-O-Si) and bending modes of the SiO4 tetrahedra vibrations within the silica, respectively [49, 50]. The band near 954 cm−1 in the W-SiO2-y materials was characteristic of Si-O-W bonds, which indicated that tungsten species were incorporated into the framework of the mesoporous silica matrix [51].
The electronic behavior of the materials was examined using UV-vis DRS, as shown in Fig. 2. The spectrum of [C16mim]3PW12O40 exhibited an absorption band at approximately 212 nm that was assigned to an O-P transition, while the absorption peak located at 260 nm was ascribed to a charge transfer transition within the Keggin units. Bulk WO3 exhibited a strong absorption near 430 nm that was assigned to a ligand-to-metal charge transfer (O2−-W6+) [52]. After the introduction of the tungsten species into the silica matrix, two ligand-to-metal charge transfer bands were observed near 202 and 267 nm (O2s-W5d-O2p). The absorption onset of the W-SiO2-y materials was shifted to a lower wavelength by approximately 100 nm when compared with bulk WO3, which indicated that the tungsten species were incorporated into the silica matrix [53].
To examine the structure of the tungsten species that were incorporated into the hybrid materials, Raman scattering spectroscopy was used (Fig. 3). Bulk WO3 (Fig. 3(1)) exhibited a series of characteristic peaks in the regions 1000-600 cm−1 and 600-150 cm−1 that were assigned to W-O stretching and bending vibrations, respectively [54]. These bands were also observed in the W-SiO2-y materials (Fig. 3(2)-(4)) and as the tungsten content increased, the bands became more intense and well defined. These results indicated that the tungsten species in the hybrid materials were tungsten oxide.
XRD pattern was performed to analyze the composition and phase structure of the W-SiO2-y materials (Fig. 4). Small angle XRD patterns of the samples exhibited a broad diffraction peak that was located at approximately 2θ = 2.2°, which is a feature of mesoporous materials (Fig. 4(a)). However, the wide-angle XRD patterns exhibited a broad peak that was consistent with amorphous silica walls (Fig. 4(b)). The diffraction peaks of crystalline WO3 were not obvious, even in the W-SiO2 sample that had a relatively high tungsten content (Si/W = 10), which indicated that the tungsten species were effectively dispersed throughout the silica matrix.
Nitrogen adsorption-desorption isotherms and the BJH pore-size distribution curves are shown in Fig. 5. The nitrogen adsorption-desorption isotherms of all samples were assigned according to the IUPAC classification as Type IV, which is characteristic of a mesostructure in W-SiO2 materials (Fig. 5(a)). Typical unimodal pore-size distribution curves with pore diameters that ranged between 2 and 5 nm indicated the uniform mesoporosity of the hybrid materials (Fig. 5(b)). The textural properties of the samples are listed in Table 1. The W-SiO2-40 sample had a surface area of 743 m2/g, a pore volume of 0.50 cm3/g, and a pore size of 3.20 nm. As the tungsten content within the samples increased, the BET surface area, pore volume and pore size decreased. This effect has been reported to be caused by the anchoring of tungsten species in the pore walls [55], which was confirmed by the results of the low-angle XRD analysis.
TEM was used to investigate the structural features of the representative sample. TEM images of W-SiO2-20 exhibited numerous interconnected worm-like mesoporous structures with a disordered arrangement (Fig. 6). The pore diameters determined from the TEM images ranged from 2 to 4 nm, which was close to the average pore size determined from the nitrogen desorption isotherm (2.9 nm). The TEM images also confirmed the low-angle XRD results and textural properties listed in Table 1.
The removal of DBT from the sample oil by the W-SiO2-y materials under identical conditions is shown in Fig. 7. Both W-SiO2-10 and W-SiO2-20achieved deep desulfurization within 40 min. Importantly, W-SiO2-20 catalyzed the complete removal of DBT in 30 min, which was higher than either W-SiO2-10 or W-SiO2-40 (97.7% and 69.4%, respectively). This may have been because W-SiO2-20 possessed a larger BET surface area and pore volume compared with W-SiO2-10, and more active sites (tungsten species) compared with W-SiO2-40. The adsorption desulfurization of DBT was also investigated without the presence of H2O2 (catalyst = 0.01 g, T = 60 ℃, t = 40 min). This experiment yielded adsorption desulfurization values of 6.9%, 5.6%, and 4.7%, which corresponded to calculated adsorptive capacities of 17.2, 14.0, and 11.7 mg/g for W-SiO2-40, W-SiO2-20, and W-SiO2-10, respectively.
The identity of a sulfur-containing compound can influence the ability of a catalyst to remove sulfur. The effectiveness of W-SiO2-20 for the removal of a series of organosulfur compounds was studied. In addition to DBT, the compounds BT, 4, 6-DMDBT and DT were examined (Fig. 8). The removal of sulfur reached equilibrium after approximately 30 min except for 4, 6-DMDBT, which took 20 min. The levels of sulfur removal from the oil after 20 min for BT, DBT, DT and 4, 6-DMDBT were 84.4%, 90.4%, 75.0% and 100.0% , respectively. Thus, the ability of W-SiO2-20 to remove sulfur decreased in the order 4, 6-DMDBT ＞ DBT ＞ BT ＞ DT. This decrease in ability to remove sulfur corresponded to the decrease in sulfur electron density of the aromatic compounds (4, 6-DMDBT (5.760) ＞ DBT (5.758) ＞ BT (5.739)) [56]. However, the sulfur removal of DT was the lowest, which was attributed to the strong steric hindrance of the long carbon chains.
The ability of the catalyst to be recycled was investigated. After the first reaction, the upper oil phase was separated by decanting from the reactor. The remaining catalyst within the reactor was dried at 60 ℃ overnight, and then fresh H2O2 and model oil were added to the reactor for the next cycle. The ability of the catalyst to remove sulfur from the reaction system was able to reach 90.3% after recycling seven times, as shown in Fig. 9.
The oxidation products of DBT were determined using GC-MS analysis after reaction (Fig. 10). The upper oil layer was decanted, and the used catalyst was extracted using carbon tetrachloride for the GC-MS analysis. A single, intense peak at 3.4 min was ascribed to n-tetradecane, which indicated that DBT was completely removed from the oil phase (Fig. 10(a)). The GC trace for the extracted catalyst only exhibited an obvious peak at 9.8 min, which corresponded to DBTO2 (Fig. 10(b), inset). The fact that a peak for DBT was not detected demonstrated that it was entirely oxidized to DBTO2 in the desulfurization process.
A potential mechanism for the desulfurization process using the W-SiO2 catalysts is shown in Fig. 11. Initially, DBT is absorbed by W-SiO2 and is oxidized by the active peroxo species, which are formed by the reaction of the tungsten species and H2O2. The sulfone species that is subsequently formed (DBTO2) can also be absorbed by W-SiO2. As the reaction proceeds, the peroxo species transformed to tungsten species, which could continually combine with H2O2 [57].
In summary, a series of mesoporous materials (W-SiO2) were synthesized using a one-pot method from a POM-based IL. These materials contained tungsten species that were effectively dispersed throughout the silica matrix. The hybrid catalysts exhibited excellent activity for the removal of a number of different sulfur-containing compounds from a model oil, with low amounts of co-oxidant (O/S molar ratio = 2.5) in approximately 40 min. This process did not require the addition of organic solvents as an extractant. The ability of the catalysts to remove sulfur compounds decreased in the order: 4, 6-DMDBT ＞ DBT ＞ BT ＞ DT. Importantly, the catalyst could be recycled up to seven times without any significant decrease in activity.
The authors wish to acknowledge the financial support provided by Advanced Talents of Jiangsu University (13JDG080), Postdoctoral Foundation of China (2014M551516), and A Project Funded by the Priority Academic Program Development of Jiangsu Higher Education Institutions (PAPD).
2016,
Vol. 37 Issue (6): 971-978 
PDF (21885 678KB)
2016,
Vol. 37 Issue (6): 971-978 PDF (21885 678KB)