Yu Zhan; Shi Li; Meng Xuan

(The State Key Laboratory of Chemical Engineering, East China University of Science and Technology, Shanghai 200237)

Preparation of Low Cost Activated Carbon from Deactivated Resin Catalyst for Methyl Tert-Butyl Ether Synthesis and Its Application in Dimethyl Sulfide Adsorption with Transition Metal Impregnation

Yu Zhan; Shi Li; Meng Xuan

(The State Key Laboratory of Chemical Engineering, East China University of Science and Technology, Shanghai 200237)

In this paper, a low-cost activated carbon (AC) was prepared from deactivated resin catalyst (DRC) for methyl tert-butyl ether (MTBE) synthesis through carbonization and subsequent steam activation treatment. The activated carbon was characterized in detail. After loading various transition metals, including Cu2+, Ag+, Co2+, Ni2+, Zn2+, and Fe3+via the ultrasonic-assisted impregnation method, a series of metal-loaded adsorbents (xM-AC) were obtained and their dimethyl sulfde (DMS) adsorption performance was investigated in a batch system. Among these adsorbents, 15Cu-AC presented a superior DMS adsorption capacity equating to 58.986 mg/g due to the formation of S-M (σ) bonds between Cu2+and sulfur atoms of DMS as confrmed by the Raman spectra and kinetic study.

activated carbon, deactivated resin catalyst, dimethyl sulfde, transition metal

1 Introduction

Methyl tert-butyl ether (MTBE) is a primary gasoline addictive, which is almost exclusively used as the oxygenate to raise the octane number of gasoline. MTBE is manufactured through the catalytic reaction of methanol and isobutylene. Catalysts used in the production of MTBE include the zeolites, inorganic acids, ion exchange resins, polyoxometalate, etc. Among them, the ion exchange resins have dominated the MTBE catalyst market due to its high activity, good stability and low cost. However, when the ion exchange resin loses its catalytic activity, the resin catalyst would turn into a solid waste which could hardly be degraded in the environment. Traditional treatments of deactivated resin catalyst include such approaches as landfll and incineration, but with an increasing social concern for environmental protection, disposal of this solid waste in an environmentally friendly and economical way has become a serious issue.

Thanks to its high specific surface area, complex porous structure, various surface functional groups and good electrical conductivity, activated carbon (AC) has been used in a wide range of industrial sectors, such as removing unwanted substances[1-4], catalyst and catalyst support[5-7], and electric energy storage[8-9]. Activated carbon can be obtained from various kinds of carbonaceous materials through carbonization and subsequent activation under different conditions. In recent years, agricultural and industrial wastes, such as waste rubber tires[2], waste potato peels[3], waste printed circuit boards[10], arundo donax linn[11], cotton linter fbers[4]and black liquor lignin[12]have absorbed much interest in producing activated carbon. The resin catalyst for MTBE synthesis is made up of cross-linked sulfonated styrene and divinylbenzene copolymer, and it contains a large amount of carbon. Thus, the deactivated resin catalyst (DRC) can be utilized as a precursor for the preparation of activated carbon, which not only can reduce the production cost of activated carbon but also solve the environmental pollution problems. To our knowledge, producing AC from DRC has not been reported yet.

Dimethyl sulfide (DMS), a volatile organosulfurcompound with characteristic unpleasant odor, is regarded as one of the most recalcitrant sulfur compounds that could hardly be removed from light hydrocarbons because of its non-polarity[13-15]. Recently since the quality of crude oil becomes increasingly inferior, the DMS content in the liquid hydrocarbon streams especially in light hydrocarbons increases rapidly. DMS existing in liquid hydrocarbon streams would severely poison catalysts used in the subsequent processes and further affect the purity of petrochemical products, so it is necessary to develop a suitable method for removing DMS as a pretreatment process[14]. Different technologies can be applied to remove DMS, such as oxidation[16-17], bio-filtration[17-18], and adsorption[14-15]. Among them, the adsorption process has been widely accepted thanks to its effectiveness, easy operation and low cost. And various types of adsorbents, such as activated carbon[13], bentonite[14-15], Y-zeolite[19]and metal organic framework (MOF)[20], have been used for the removal of DMS so far. However, most of the adsorbents are modified with certain metals before use to improve the adsorption capacity and selectivity of DMS. Cui and co-workers[13]have reported that commercial coconut shell based activated carbon modifed with FeCl3solution showed an improved adsorption capacity for DMS, because the modification with FeCl3on the carbon surface created new active sites and these new active sites had stronger affinity with DMS molecules. Huang and co-workers[14-15]have investigated the adsorption behavior and mechanism of DMS adsorption on the silver and copper modifed bentonite. They confrmed the formation of S–M (σ) bonds between Ag+, Cu2+and sulfur atoms of DMS.

Therefore, the objective of this study is: (1) to prepare activated carbon from deactivated resin catalyst and characterize activated carbon by elemental analyzer, N2adsorption-desorption isotherms, scanning electron microscopy (SEM) and powder X-ray diffraction (PXRD); (2) to load different transition metals, including Cu2+, Ag+, Co2+, Ni2+, Zn2+, Fe3+on activated carbon by the ultrasonic-assisted impregnation method; and (3) to investigate the DMS adsorption performance of these adsorbents in a batch system.

2 Experimental

2.1 Materials

The deactivated resin catalyst (DRC) was provided by the Cary Environmental Technology Co., Ltd., located in Cangzhou, Hebei province. HCl, CuCl2, FeCl3, ZnCl2, CoCl2, NiCl2andn-octane, were purchased from the Shanghai Lingfeng Chemical Reagent Co., Ltd. AgNO3and DMS were purchased from the Sinopharm Chemical Reagent Co., Ltd. and Aldrich, respectively. All the chemicals were used as received without further treatment.

2.2 Preparation of AC

Before carbonization and activation, the residual quartz sand was firstly removed from DRC through a mesh screen. Then, DRC was soaked in the HCl (0.1 mol/L) solution for 12 h to dissolve the metal residue and was subsequently washed with deionized water until the pH of effuent wash water reached around 7. Finally, the wet DRC was dried in a drying oven at 120oC for 24 h.

The carbonization procedure was carried out in a horizontal tubular furnace in a nitrogen flow (at a rate of 80 mL/min). About 2 g of pretreated DRC placed on a quartz boat were heated to 500oC at a temperature increase rate of 10oC /min and were kept at this temperature for 30 min. Then the carbonization samples were continuously heated to 700oC at a temperature increase rate of 5oC /min during activation and were kept at this temperature also for 30 min. During the activation process, the nitrogen fow was interrupted and the deionized water (at a flow rate of 1 mL/min) was added to the horizontal tubular furnace to be vaporized instantaneously at the high temperature. After the activation process, the horizontal tubular furnace was cooled down in nitrogen fow to room temperature. The prepared AC was washed with deionized water to remove some of the ash and decomposed fragments and was then dried at 120oC for 24 h.

2.3 Loading transition metals on AC

Several transition metals were loaded on AC by the ultrasonic-assisted impregnation method[21]. Briefly, a certain amount of metal salts including CuCl2, FeCl3, ZnCl2, CoCl2, NiCl2and AgNO3was dissolved in 2 mL ofdeionized water, and 1 g of activated carbon was added to the aqueous solution. After that, the mixtures were treated by ultrasonic wave for 30 min to overcome mass transfer resistance and ensure the uniform dispersion of transition metals. Finally, the samples were dried at 40oC until the complete evaporation of solvent and were subsequently baked at 150oC for 2 h. These metalloaded AC was designated as xM-AC, where M was referred to as the type of transition metal and x meant the weight percentage of transition metal. The procedure for preparation of AC and metal loading process is shown in Figure 1.

Figure 1 Preparation of AC and metal loading process

2.4 Characterization methods

The morphology and shape of the samples were observed using a field-emission scanning electron microscope (NOVA Nano SEM450). The element (C, H, O, N, and S) content of the AC and DRC was estimated by an elemental analyzer (Vario Macro Cube), with the oxygen content being calculated by difference. The surface area and pore size distribution of the samples were obtained from the N2adsorption?desorption isotherms at -196oC (JW-BK112). Before testing, the samples were degassed under vacuum for 120 min at 200oC. The specifc surface area (SBET) was calculated by the Brunauer–Emmet–Teller (BET) equation. The total pore volume (Vt) was estimated from the volume of adsorbed liquid nitrogen at a relative pressure of 0.99. The micropore area (Smicro) and the micropore volume (Vmicro) were obtained from the t-plot method; the mesopore area (Smeso) and mesopore volume (Vmeso) were calculated from the difference betweenSBETandSmicro,VtandVmicro, respectively. The pore size distribution was determined by the nonlocal density functional theory (NLDFT) with a slit pore model. The powder X-ray diffraction (PXRD) was used to characterize the crystal structure on a Siemens D-500 X-ray diffractometer equipped with the Ni-fltered Cu Kα radiation (operating at 40 kV and 100 mA). The Raman spectra were recorded by a Renishaw System 100 Raman spectrometer with a 514 nm red excitation from an Ar laser. The thermo-gravimetric (TG) curves were obtained using a TA Instruments thermal analyzer. The samples were preheated at a temperature increase rate of 10oC /min up to 1 000oC under a constant nitrogen fow (100 mL/min). The pyrolysis gases were introduced simultaneously to a mass spectrometer (MS) for further analysis.

2.5 DMS adsorption experiment

The DMS adsorption experiment was carried out at room temperature (25oC) in a batch system. The model oil was obtained by dissolving DMS inn-octane, and the concentration of DMS ranged from 125 mg/L to 1 500 mg/L within the experimental error. In a typical adsorption experiment, 50 mg of adsorbent were added into a sealed glass flask containing 5 mL of model oil under stirring over an appropriate time. Then, the concentration of DMS in model oil was determined by an ultraviolet fuorescence sulfur nitrogen analyzer (TS-3000). The amount of adsorbed DMS can be calculated by Equation 1:

whereqt(mg/g) is the adsorption amount of DMS,C0andCt(mg/L) are the concentration of DMS in model oil at a time of 0 andt, respectively,V(mL) andm(mg) are the volume of the model oil and the mass of the adsorbent, respectively. All the adsorption experiments were conducted twice to obtain an average value.

3 Results and Discussion

3.1 Thermal analysis of DRC

The TG-DTG curves of DRC and mass-spectrometric patterns of pyrolysis gases are shown in Figure 2. There were three sections of weight loss process. The first section ranging from room temperature to 200oCwas attributed to the water evaporation stage. The second section covering 200oC to 500oC, with the weight loss increasing signifcantly, was caused by the decomposition of sulfonic acid groups and polystyrene chains, accompanied with dehydro-condensation and aromatization. When the temperature exceeded 500oC, the weight loss became slower gradually, implying that the pyrolysis procedure was almost complete. The fxed carbon left was about 20% after decomposition of DRC.

Figure 2 (a) TG-DTG curves of DRC; (b) MS patterns of pyrolysis gases

3.2 Characterization of AC and DRC

Through carbonization and subsequent steam activation, DRC was successfully transformed into AC. The textural properties and element content of DRC and AC are summarized in Table 1. Obviously, the carbon content of AC was much higher than DRC, while the hydrogen, oxygen, sulfur content of AC decreased sharply as compared with DRC. This might be resulted from the release of volatile compounds (CO, CO2, CH4, SO2, H2S, etc.) during the carbonization and activation process[22]. The existence of nitrogen could be attributed to poisoning of the resin catalyst caused by basic nitrogen compounds during its use. Figure 3 shows the N2adsorption?desorption isotherms and the pore size distribution. DRC possessed limited pore volume (0.299 cm3/g) and very low specific surface area (23.772 m2/g). After steam activation, the porosity was greatly developed, while the BET surface area and total pore volume of AC reached 789.965 m2/g and 0.615 cm3/g, respectively. The adsorption isotherm of AC showed a combination of type I and type IV according to the IUPAC classifcation, implying the hierarchical pore structure of AC, which was also confirmed by the pore size distribution curve. The adsorption amount at a relative pressure of close to 0 increased sharply due to the micropore flling, and the hysteresis loop at a high relative pressure corresponded to the capillary condensation in mesopores. The SEM micrographs of AC (Figure 4) showed that AC still maintained the spherical morphology with cavities on its surface, suggesting that the wall of DRC became open and a wider porosity was formed through steam activation. Therefore, this DRC-based AC could be a favorable adsorbent for DMS adsorption and a good support for metal loading due to its hierarchical pore structure.

Table 1 Textural properties and element content of DRC and AC

Figure 3 N2adsorption-desorption isotherms and pore size distributions of DRC and AC

Figure 4 SEM images of AC

3.3 Effect of metal loading amount on DMS adsorption

To improve the DMS adsorption performance of AC, CuCl2was firstly selected as the copper compound to be immobilized on AC by the ultrasonic-assisted impregnation method. After the impregnation of AC with CuCl2, the N2adsorption isotherms of xCu-AC (Figure 5a) signifcantly moved down with the increase of loading amount, but the form of these isotherms did not change, indicating that the introduction of Cu2+would not change the pore type of AC. The textural properties of xCu-AC are presented in Table 2. Their specifc surface area and total pore volume decreased dramatically, denoting that a part of the pores have been blocked by Cu2+particles. The X-ray diffraction analysis (Figure 5b) was carried out to identify the mineralogical structure of AC and xCu-AC. Two diffraction peaks of AC were linked to the feature of graphite which showed characteristic refections at 2θ= 26.546° and 44.517° (slightly formed), corresponding to the (002) and (101) planes of hexagonal graphite lattice, respectively. The XRD patterns of 5Cu-AC and 10Cu-AC did not show any other obvious characteristic reflections in comparison with that of AC, whereas the XRD pattern of 15Cu-AC and 20Cu-AC showed slight characteristic refections of Cu(OH) Cl at 2θ= 15.947°, 32.436° and 40.208° that were consistent with the (001), (110), (301) planes of cubic Cu(OH)Cl, respectively, which were formed from the partial hydrolysis of CuCl2. These facts suggested that Cu2+existed in an amorphous state and was highly dispersed on AC at low content; and with the increase of loading amount, some particles might aggregateas the crystalline form in the pore channel. Thus, an appropriate loading amount of Cu2+would be a good assurance of DMS adsorption.

Figure 5 (a) N2adsorption-desorption isotherms and pore size distributions; (b) XRD patterns of xCu-AC

Table 2 Textural properties and DMS adsorption capacity of AC and xCu-AC

The performance of xCu-AC on adsorption of DMS is shown in Figure 6. The capacity of xCu-AC on adsorption of DMS rose gradually with an increasing amount of Cu2+loaded. When the loading amount of Cu2+increased to 15%, its DMS adsorption capacity reached a maximum, which was equal to 58.986 mg/g. Further increasing the loading amount of Cu2+resulted in a slight decrease of DMS adsorption capacity, since the over-loading of Cu2+might aggregate as crystalline form in the pore channels and then the active adsorption centers would be shielded by the excess Cu2+crystallites.

Figure 6 Effect of copper loading amount on DMS adsorption from the model oil

3.4 Kinetic study

The effect of contact time on DMS adsorption on AC, 5Cu-AC and 15Cu-AC is presented in Figure 7a. (To guarantee the adsorption equilibrium, other adsorption experiments were performed in 24 h). It can be found that the adsorption amount increased rapidly at the initial stage, then slowed down and finally reached an equilibrium after about 360 min. In comparison with AC (with aqeof 15.765 mg/g), 15Cu-AC (with aqeof 59.563 mg/g) showed much higher saturation adsorption capacity at equilibrium, almost 3.8 times as large as that of AC; whereas the specific surface area and total pore volume of 15Cu-AC (161.314 m2/g, and 0.227 cm3/g) were much smaller than AC (789.965 m2/g, and 0.615 cm3/g). This contradiction conveyed an important message that DMS was adsorbed by chemical interaction with Cu2+, since the sulfur atom of DMS had a low electronegativity and could easily lose its lone electron pairs, while Cu2+just had empty orbitals to accept lone electron pairs. To elucidate the adsorption mechanism, the Raman spectra (Figure 8) of AC and 15Cu-AC were obtained after the adsorption ofDMS. Two characteristic peaks at 293 cm-1and 305 cm-1were observed on 15Cu-AC, which were attributed to the stretching vibrations of Cu-Cl and Cu-S[17], respectively, providing an evidence for the formation of S-M (σ) bonds between Cu2+and the sulfur atom of DMS.

Figure 7 (a) Kinetic curves for the adsorption of DMS (1500 mg/L) on AC, 5Cu-AC and 15Cu-AC; (b)Adsorption isotherms for the adsorption of DMS on AC, 5Cu-AC and 15Cu-AC

Figure 8 Raman spectra of AC and 15Cu-AC after adsorption of DMS

In order to gain a better understanding of the adsorption process, the pseudo-first-order kinetic model (Equation 2) and the pseudo-second-order kinetic model (Equation 3)[4,11]have been used to fit the experimental data, which are expressed by the following equations:

wherek1(min-1) andk2(g/(mg·min)) are the adsorption rate constants of pseudo-frst-order equation and pseudosecond-order equation, respectively;qe(mg/g) is the amount adsorbed at the equilibrium andqt(mg/g) is the amount adsorbed at a contact time oft(min). The kinetic parameters of the two models are listed in Table 3. For all adsorbents, the pseudo-second-order model had higher correlation coeffcients (＞0.99) and the adsorption capacity values (qe,cal) calculated through pseudo-secondorder model were much closer to the experimental data (qe,exp), indicating that the pseudo-second-order model was more appropriate for describing the DMS adsorption process than the pseudo-first-order model. This fact revealed that the rate-controlling step in the adsorption process of copper loaded AC was the chemical sorption involving valence forces through sharing or exchange of electrons between the adsorbent and the adsorbate[23], as justifed by the Raman spectra.

3.5 Adsorption isotherms

The isotherms (Figure 7b) for DMS adsorption on AC, 5Cu-AC and 15Cu-AC were determined at different concentrations under equilibrium condition. To fgure out the relationship between the adsorbate and the adsorbent at equilibrium, the adsorption isotherms were ftted with the Langmuir and the Freundlich isotherm models[4,11-12]. The Langmuir model (Equation 4) assumes monolayer adsorption on the adsorbent surface and the Freundlich model (Equation 5) is an empirical equation employed to describe the multilayer adsorption. The equations of these two models are given as follows:

whereKL(L/mg) represents the Langmuir constant, andq0(mg/g) is the maximum adsorption capacity;KF(mg(L/ mg)1/n/g) andnare the Freundlich constants indicative of adsorption capacity and adsorption intensity, respectively. The constants of these two isotherm models are listed in Table 4. The correlation coeffcients of the Langmuir model were greater than those of the Freundlich model, suggesting that the adsorption of DMS on all adsorbents showed a nature of monolayer adsorption, in accordance with the chemical or micropores adsorption of the adsorbents. The maximum adsorption capacities calculated by the Langmuir model are 20.916 mg/g, 31.124 mg/g, and 70.671 mg/g for AC, 5Cu-AC, and15Cu-AC, respectively, indicating that these adsorbents have a good potential to remove DMS. Furthermore, judging from the fitting results with the Freundlich model, the values ofnare greater than 1, meaning that the adsorption is favorable.

Table 3 Kinetic model parameters for DMS adsorption

Table 4 Isotherm model parameters for DMS adsorption

3.6 DMS adsorption performance of other transition metals

Inspired by the good DMS adsorption performance of xCu-AC, more transition metal salts, including AgNO3, CoCl2, NiCl2, ZnCl2, and FeCl3, had been loaded on AC. As presented in Figure 9a, the optimum loading amount covered: 12% of Ag+, 8% of Co2+, 12% of Ni2+, 15% of Zn2+, and 9% of Fe3+. And their maximum adsorption capacity decreased in the following order: 12Ni-AC > 12Ag-AC > 15Zn-AC > 8Co-AC > 9Fe-AC. Compared with AC, the specific surface area and total pore volume of these adsorbents (Table 5) decreased sharply, indicating to the successful loading of transition metals. Among these metals, Ag+did not show an excellent DMS adsorption capacity, which was contradictory to the literature report[14]. It has been reported that Ag+can be reduced to metallic silver on AC[24-25]. To verify this hypothesis, 12Ag-AC was analyzed by XRD technique. The XRD pattern (Figure 9b) showed the sharp characteristic refections of Ag at 2θ= 38.115°, 44.299°,64.443° and 77.397°, corresponding to the (111), (200), (220), and (311) planes of face-centered cubic metallic silver crystals, respectively, while no diffraction peaks of AgNO3or Ag2O were detected . The above results have provided direct evidences for the reduction of Ag+on AC, thus AC is not a desired support for Ag+loading.

Table 5 Textural properties and DMS adsorption capacity ofxM-AC with an optimum loading amount

Figure 9 (a) DMS adsorption capacity of different transition metal ions; (b) XRD pattern of 12Ag-AC

3.7 Regeneration performance

Cycling ability of adsorbents is a key component for potential industrial application. The regeneration experiments of 15Cu-AC were carried out at different regeneration temperatures for several times at anair flow rate of 100 mL/min for 2 h. The effect of regeneration temperature and time on DMS adsorption ability of 15Cu-AC is shown in Figure 10. It can be seen from Figure 10a that when the regeneration temperature increased to 150oC from 100oC, the DMS adsorption ability was restored signifcantly. A further increase of regeneration temperature had little effect on DMS adsorption ability, suggesting that almost all the reversibly adsorbed DMS had been desorbed at 150oC. After three cycles of regeneration, the DMS adsorption ability of 15Cu-AC could still be equal to 69.86% of the fresh adsorbent. The results indicated that the regenerated adsorbent still showed a good DMS adsorption ability.

Figure 10 (a) Effect of regeneration temperature (a) and number of cycles (b) on DMS adsorption ability of 15Cu-AC.

4 Conclusions

The activated carbon derived from the deactivated resin catalyst has been successfully prepared through carbonization and subsequent steam activation. The hierarchical pore structure has enabled the activated carbon to be a favorable adsorbent for DMS adsorption and a good support for metal loading. After loading various transition metals including Cu2+, Ag+, Co2+, Ni2+, Zn2+, and Fe3+through the ultrasonic-assisted impregnation method, a series of metal-loaded adsorbents (xM-AC) were obtained. The optimum DMS adsorption capacity for the desired amount of metal decreased in the following order: Cu2+> Ni2+> Ag+> Zn2+> Co2+> Fe3+. The Raman spectra confrmed the formation of S-M (σ) bonds between Cu2+and sulfur atoms of DMS. On the basis of the kinetic studies, it was revealed that the ratecontrolling step in the adsorption process of the copper loaded AC was the chemical adsorption. The adsorption isotherms were well fitted to the Langmuir model, and the calculated maximum adsorption capacity of 15Cu-AC was equal to 70.671 mg/g. These results have shown that the production of activated carbon is a feasible and effective way to dispose of the deactivated resin catalyst, and this activated carbon impregnated with transition metals could be a promising candidate for the application in the area of DMS adsorption.

Acknowledgement: This work is fnancially supported by the National Natural Science Foundation of China (No. 21276086).

[1] Romero-Anaya A J, Lillo-Ródenas M A, Linares-Solano A.Factors governing the adsorption of ethanol on spherical activated carbons[J]. Carbon, 2015, 83: 240-249

[2] Saleh T A, Danmaliki G I. Influence of acidic and basic treatments of activated carbon derived from waste rubber tires on adsorptive desulfurization of thiophenes[J]. Journal of the Taiwan Institute of Chemical Engineers, 2016, 60: 460-468

[3] Kyzas G Z, Deliyanni E A, Matis K A. Activated carbons produced by pyrolysis of waste potato peels: Cobalt ions removal by adsorption[J]. Colloids and Surfaces A: Physicochemical and Engineering Aspects, 2016, 490: 74-83

[4] Sun Y, Yue Q, Gao B, et al. Preparation of activated carbon derived from cotton linter fbers by fused NaOH activation and its application for oxytetracycline (OTC) adsorption[J]. Journal of Colloid and Interface Science, 2012, 368: 521-527

[5] Tseng H H, Wey M Y, Fu C H. Carbon materials as catalystsupports for SO2oxidation: catalytic activity of CuOAC[J]. Carbon, 2003, 43: 139-149

[6] Lee J J, Suh J K, Hong J S, et al.The synthesis of spherical activated carbons containing zinc and their photochemical activity[J]. Carbon, 2008, 46: 1648-1655

[7] Kudo S, Maki T, Miura K, et al.High porous carbon with Cu/ZnO nanoparticles made by the pyrolysis of carbon material as a catalyst for steam reforming of methanol and dimethyl ether[J]. Carbon, 2010, 48: 1186-1195

[8] Han S J, Kim Y H, Kim K S, et al.A study on high electrochemical capacitance of ion exchange resin-based activated carbons for supercapacitor[J]. Current Applied Physics, 2012, 12: 1039-1044

[9] Jiang L, Yan J, Hao L, et al. High rate performance activated carbons prepared from ginkgo shells for electrochemical supercapacitors[J]. Carbon, 2013, 56: 146-154

[10] Kan Y, Yue Q, Gao B, et al.Preparation of epoxy resinbased activated carbons from waste printed circuit boards by steam activation[J]. Materials Letters, 2015, 159: 443-446

[11] Sun Y, Yue Q, Gao B, et al. Comparison of activated carbons from Arundo donax Linn with H4P2O7activation by conventional and microwave heating methods[J]. Chemical Engineering Journal, 2012, 192: 308-314

[12] Fu K, Yue Q, Gao B, et al. Preparation, characterization and application of lignin-based activated carbon from black liquor lignin by steam activation[J]. Chemical Engineering Journal, 2013, 228: 1074-1082

[13] Cui H, Turn S Q. Adsorption/desorption of dimethylsulfde on activated carbon modifed with iron chloride[J]. Applied Catalysis B: Environmental, 2009, 88: 25-31

[14] Huang H, Yi D, Lu Y, et al.Study on the adsorption behavior and mechanism of dimethyl sulfide on silver modified bentonite by in situ FTIR and temperatureprogrammed desorption[J]. Chemical Engineering Journal, 2013, 225: 447-455

[15] Cui Y, Lu Y, Yi D, et al. Adsorptive Desulfurization of Propylmercaptan and Dimethyl Sulfide by CuBr2Modified Bentonite[J]. China Petroleum Processing and Petrochemical Technology, 2015, 17(2): 55-63

[16] He J, Xu A, Hu L, et al.Layered KTiNbO5photocatalyst modified with transitional metal ions (Mn2+, Ni2+): Investigation of microstructure and photocatalytic reaction pathways for the oxidation of dimethyl sulfide and ethyl mercaptan[J]. Powder Technology, 2015, 270: 154-162

[17] Caceres M, Silva J, Morales M, et al.Kinetics of the biooxidation of volatile reduced sulphur compounds in a biotrickling flter[J]. Bioresource Technology, 2012, 118: 243-248

[18] Ho K L, Chung Y C, Lin Y H, et al. Microbial populations analysis and field application of biofilter for the removal of volatile-sulfur compounds from swine wastewater treatment system[J]. Journal of Hazardous Materials, 2008, 152: 580-588

[19] Yu X, Lu Y, Huang H, et al. Adsorption desulphurisation of dimethyl sulphide using nickel-based Y zeolites pretreated by hydrogen reduction[J]. Chemical Papers, 2016, 70: 294-304

[20] Li Y, Wang L J, Fan H L, et al. Removal of Sulfur Compounds by a Copper-Based Metal Organic Framework under Ambient Conditions[J]. Energy & Fuels, 2015, 29: 298-304

[21] Tang X L, Qian W, Hu A, et al. Adsorption of thiophene on Pt/Ag-supported activated carbons prepared by ultrasonicassisted impregnation[J]. Industrial & Engineering Chemistry Research, 2011, 50 (15): 9363-9367

[22] Shi Q, Li A, Zhu Z, et al.Adsorption of naphthalene onto a high-surface-area carbon from waste ion exchange resin[J]. Journal of Environmental Sciences, 2013, 25: 188-194

[23] Ho Y S, McKay G. Pseudo-second order model for sorption processes[J]. Process Biochemistry, 1999, 34: 451-465

[24] Song X, Gunawan P, Jiang R, et al.Surface activated carbon nanospheres for fast adsorption of silver ions from aqueous solutions[J]. Journal of Hazardous Materials, 2011, 194: 162-168

[25] Tran Q T, Nguyen V S, Hoang T K, et al.Preparation and properties of silver nanoparticles loaded in activated carbon for biological and environmental applications[J]. Journal of Hazardous Materials, 2011, 192: 1321-1329

date: 2017-01-17; Accepted date: 2017-02-13.

Dr. Meng Xuan, Telephone: +86- 21-64252383; E-mail: mengxuan@ecust.edu.cn.