Effect of Water on Extractive Desulfurization of Simulated FCC Gasoline Using Io

Fang Liuya; Shen Zhi; Shen Xizhou; Kang Shunji; Huang Wei; Song Hao; Liang Tian

(1. Key Laboratory for Green Chemical Process, Ministry of Education, School of Chemical Engineering and Pharmacy, Wuhan Institute of Technology, Wuhan 430074;2. The College of Post and Telecommunication, Wuhan Institute of Technology, Wuhan 430074;3. SINOPEC Research Institute of Petroleum Processing, Beijing, 100083)

Abstract: A series of novel aqueous ionic liquids (NMP-FeCl3-nH2O) were prepared and the effects of water in the aqueous ionic liquids on desulfurization rate and selectivity of simulated FCC gasoline were investigated. The results showed that adding a small amount of water into the ionic liquid NMP-FeCl3 could effectively improve the desulfurization rate and selectivity, and the optimal amount of water was equal to 5%―10% of NMP. Finally, the possible desulfurization mechanism activated by a small amount of water was proposed.

Key words: ferric chloride; N-methylpyrrolidone; water; activation mechanism

1 Introduction

S-compounds are converted into SOxthat causes acid rain and air pollution when fuel oil is burned in automobile engine[1]. For environmental protection purpose, many countries have been demanding the reduction of S-level in fuels down to 10 μg/g, and with more stringent regulatory constraints being proposed, zero-emission and zero-level of sulfur content are expected to be previewed in the near future[2]. In order to reduce the negative impact of SOxemission, it is very important to eliminate the sulfurcompounds in the fuel oil.

Presently, hydrodesulfurization (HDS) as a mature and conventional desulfurization technology, which involves the reaction of sulfur compounds with hydrogen gas to form H2S gas, is highly effective in the reduction of sulfur level in liquid fuels[3]. However, a large amount of hydrogen consumption, expensive catalysts, high temperature (＞300 ℃), and high pressure (3―10 MPa)are generally needed in HDS process[4-5]. Therefore,alternative desulfurization technologies for fuel oils have been developed, such as oxidative desulfurization(ODS)[6], extractive desulfurization (EDS)[7], adsorptive desulfurization (ADS)[8], bio-desulfurization (BDS)[9],and others[10]. Among them, the EDS technology deserves special attention as a result of mild operating conditions(temperature and pressure) and low cost for commercial applications[11-12]. However, it is quite challenging to find or synthesize a perfect extractant with high selectivity and chemical stability[13]. At present, most extractants used in EDS process are mainly conventional organic solvents,deep eutectic solvents (DESs), and ionic liquids (ILs)[7].Among them, ILs, which possess many desirable properties, such as being nonvolatile and nonflammable with high thermal stability, have been widely discussed as extractants for the desulfurization of fuel[5,13]. Initially,imidazolium-, pyridinium-, and ammonium salt-based ILs have been employed in EDS[14-16]. However, the high price of imidazolium- or pyrrolidinium-based ILs makes it difficult to realize the large scale application in industry.To reduce the cost of ILs, some relatively inexpensive metal halide-based ILs have been developed, such as trimethylamine hydrochloride-AlCl3[17], Me3NCH2C6H5Cl-ZnCl2[18], [(C8H17)3CH3N]Cl-SnCl2[19], Et3NHCl-metal halides[20], and a series of FeCl3- and ZnCl2-based solvents[21]. Nevertheless, these ILs are prepared from two solid materials, which would lead to a high dynamic viscosity of the obtained ILs and thus their extraction ability needs to be further enhanced, because the viscosity of liquids is a key parameter responsible for the extraction ability[22]. Conventional organic solvents such as DMF, DMSO, and NMP have appropriate dynamic viscosity and strong polarity[23-25]. The S-compounds in FCC gasoline can be effectively removed using these strongly polar solvents. However, because of the high content of aromatics and olefins in FCC gasoline, these solvents are still not favored for industrial applications due to the high mutual solubility of the solvents with the fuel[26-27]. Adding a certain amount of metal halide into polar solvent can effectively reduce the solubility of fuel in the solvents and improve the selectivity. Li, et al.prepared a novel coordination ionic liquid C5H9NO-SnCl2with good selectivity under nitrogen flow[28]. Because anhydrous SnCl2is too expensive, Li, et al. further used inexpensive anhydrous FeCl3instead of anhydrous SnCl2and obtained an approximately equivalent desulfurization effect[25]. However, the 8B transition metals including iron are very hygroscopic because of their unoccupied orbitals[29]. As a matter of fact, the presence of water is to some extent inevitable in EDS systems with IL solvent. Firstly, most of ILs initially contain more or less water (ranging from 102 μg/g to 104 μg/g due to their different hydrophilicity and hydrophobicity), which is very hard and costly to remove completely[30]. Moreover,water usually exists as atmospheric moisture or at a low concentration in the fuel feed, which can be absorbed by IL solvent and accumulated during the extraction process[31]. Therefore, it is necessary to consider the effect of water on the desulfurization performance of these ionic liquids if they are going to be applied in commercial scale. The effect of water on desulfurization performance of ILs has been studied by some researchers. Gao, et al.demonstrated the dramatic decrease of the desulfurization performance of [C6PY] [BF4] with a low water content in the IL[32]. Zheng Song, et al. demonstrated that when the water concentration in IL [C4mim] [H2PO4] was below 10%, the desulfurization ratio increased and the solubility of fuel in the IL decreased[31]. However, the effect of water on desulfurization performance of aqueous NMP-FeCl3has not been reported yet.

In order to study the effect of water on the desulfurization performance of aqueous NMP-FeCl3, it is necessary to systematically study the coordination mode of anhydrous NMP-FeCl3and aqueous NMP-FeCl3ILs. The coordination mode of anhydrous NMP-FeCl3IL has been studied[25]. It is necessary to further study the coordination mode of aqueous NMP-FeCl3IL, find out the difference of coordination modes between these two types of ILs, and investigate the effect of different coordination modes on the desulfurization performance.

In this paper, a series of new aqueous NMP-FeCl3ILs were prepared. The anhydrous NMP-FeCl3IL reported in literature[25]and the new aqueous NMP-FeCl3IL were used to extract thiophene (Th), 2-methylthiophene (2-MT)and benzothiophene (BT) from simulated FCC gasoline.The effects of water on the desulfurization rate and the selectivity of aqueous NMP-FeCl3IL were investigated.Finally, Fourier-transform infrared spectroscopy (FTIR)and X-ray photoelectron spectroscopy (XPS) were used to investigate the coordination mode of H2O, NMP, and FeCl3in NMP-FeCl3-H2O IL. On the basis of the above results, the desulfurization mechanism activated by water in aqueous NMP-FeCl3IL is proposed.

2 Experimental

2.1 Materials

1-Methyl-2-pyrrolidone (NMP, AR grade), anhydrous FeCl3(AR grade), FeCl3·6H2O (AR grade),n-octane(AR grade), 1-octene (AR grade), toluene (AR grade),thiophene (Th, AR grade), 2-methylthiophene (2-MT, AR grade), and benzothiophene (BT, AR grade) were obtained from the Macklin Reagent Co., Ltd. All chemicals in this work were used as received without further purification.

2.2 Simulated FCC gasoline preparation

The FCC gasoline reported in the research work contained 29.61% of olefins, 21.54% of aromatics, and 48.26% of alkanes[33]. The simulated FCC gasoline base oil was prepared according to the above chemical composition in whichn-octane represented alkanes,1-octene represented olefins and toluene represented aromatics. Thiophene-containing, 2-methylthiophenecontaning, and benzothiophene-contaning simulated FCC gasoline samples were prepared by dissolving thiophene, 2-methylthiophene, and benzothiophene,respectively, in simulated FCC gasoline base oil. The sulfur contents of simulated FCC gasoline samples were determined by using an ultraviolet fluorescence sulfur analyzer. The sulfur content of thiophene-containing,2-methylthiophene-contaning, and benzothiophenecontaning simulated FCC gasoline samples was 127.6 mg/L, 151.0 mg/L, and 120.7 mg/L, respectively.

2.3 Extractants preparation

(1) NMP and anhydrous FeCl3were mixed in a 250-mL two-necked flask according to a mass ratio of 1:0.1. The mixture was stirred for 30 min under nitrogen blanketing.Finally, a clear anhydrous IL was obtained and denoted as NMP+10% of FeCl3.

(2) In order to conveniently compare with NMP+10%of FeCl3, NMP and FeCl3·6H2O were mixed in a 250-mL two-necked flask according to a mass ratio of 1:0.166. A clear aqueous IL (with water originating from FeCl3·6H2O) was obtained quickly after stirring for 2 min without nitrogen blanketing and was denoted as NMP+16.6% of FeCl3·6H2O. Thus, the mass of FeCl3in NMP+16.6% of FeCl3·6H2O and the mass of FeCl3in NMP+10% of FeCl3were both equal to10% of NMP.

(3) In order to further investigate the similarities and differences of desulfurization effect of water originating from FeCl3·6H2O and free water on NMP+10% of FeCl3IL, the free water with a dose equating to 6.6% of NMP mass was added into NMP+10% of FeCl3in step (1).Afterward, a clear aqueous IL (free water) was obtained after stirring for 2 min and was denoted as NMP+10% of FeCl3+6.6% of H2O.

(4) In order to investigate the effect of different amounts of free water on the desulfurization effect of NMP+10%of FeCl3, the free water with a dose equating to 5%, 6.6%,10%, 20% and 40% of the mass of NMP, respectively,were added into NMP+10% of FeCl3in step (1). After stirring for 2 min, five kinds of aqueous ILs with different amounts of free water were obtained and were denoted as NMP+10% of FeCl3+5% of H2O, NMP+10% of FeCl3+6.6% of H2O, NMP+10% of FeCl3+10% of H2O,NMP+10% of FeCl3+20% of H2O, and NMP+10% of FeCl3+40% of H2O, respectively.

(5) In order to investigate the difference of desulfurization effect between anhydrous NMP-FeCl3IL in step (1) and aqueous NMP-FeCl3IL under the condition of using the same amount of two ILs, the NMP and FeCl3·6H2O were mixed in a 250-mL two-necked flask according to a mass ratio of 1:0.1. The mixture was stirred for 2 min without nitrogen blanketing. Afterward, a clear aqueous IL was quickly obtained and was denoted as NMP+10% of FeCl3·6H2O. In NMP+10% of FeCl3·6H2O IL, the mass of FeCl3·6H2O is equal to 10% of NMP.

Because XPS analysis can only detect solid powder,NMP-FeCl3-H2O solid powder was synthesized according to a molar ratio of 1:1:1. One mol of anhydrous iron (III)chloride (97%, Mclean, China) was dissolved in 1 mol of deionized water to prepare ferric chloride solution. Then,the ferric chloride solution was mixed with 1 mol of NMP(with a purity of 99%, Mclean, China) and the obtained mixture was quickly stirred to become a thick yellow liquid. After that, the thick yellow liquid was vacuumevaporated at 55 °C for 5 days to remove excess NMP and a pale yellow crystalline powder (recrystallization)was formed. The yellowish crystalline compound was finally obtained after washing with anhydrous dimethyl ether (with a purity of 99%, Mclean, China) followed by drying under vacuum at 55 °C.

2.4 Desulfurization experiments and structural characterization

In all experiments, the desulfurization experiments were conducted in a 250-mL two-necked flask where model oils and extractants (with the model oil to extractant mass ratio equating to 1:1) were mixed under vigorous stirring for 30 min at 30°C in a water bath, with the same conditions replicated for all samples. After 30 min of extraction, the mixture was laid aside for a couple of minutes for phase splitting and settling. In order to quantify the sulfur content, the upper layer (oil phase)was then collected and analyzed by an UV fluorescence sulfur analyzer, using previously established calibration curves. The desulfurization rate (DR) were obtained by relating sulfur content in the model oil phase before(Ci) and after extraction (Cf), as shown in Eq. (1). The Nernst partition coefficientsKwere also determined by relating the mass fraction of sulfur and model oil in the two phases after the extraction, as presented in Eq. (2)―Eq. (12).

whereCiis the initial sulfur content, andCfis the sulfur content after extraction.

The distribution coefficients of sulfur (Ksulfur), model oil (Km-oil) and selectivity (S) are introduced to evaluate the desulfurization performance of extractant as three important parameters.

The distribution coefficient (Km-oil) is the ratio of oil concentration in the extract phase to oil concentration in the raffinate phase. The model oil concentration can be expressed using mass fraction (g/g)[14]or concentration(g/L)[34]. The distribution coefficient of the model oil in this paper was achieved through the calculation of mass fraction. The method of calculating the distribution coefficient (Km-oil) was as follows: The extraction was performed by mixing extractant (MIL) with the model oil(Moil) in a 250-mL two-necked flask. The mixture was stirred for 30 min at room temperature and then was laid aside for phase splitting and settling for 2 h. The mass of extract phase (Mextract) was weighed. The mass of raきnate phase (Mraきnate) was calculated by using Eq. (2):

The extractant in the raffinate phase was extracted with water (m0) for 3 min. The mixture was laid aside for phase splitting and settling for 30 min. The lower water phase(m1) was weighed. The mass of extractant in the raきnate phase (MILinraきnate) was calculated by using Eq. (3):

The mass of model oil in extract phase (Moilinextract) was calculated by using Eq. (4):

The mass of model oil in raきnate phase (Moilinraきnatephase)was calculated by using Eq. (5):

AsMThinraきnatewas far lower thanMoilinraきnate,MThinraきnatewas ignored during calculation.

The mass fraction of model oil in extract phase (W1) was calculated by using Eq. (6):

The mass fraction of model oil in raきnate phase (W2) was calculated by using Eq. (7):

The distribution coefficient (Km-oil) of model oil was calculated by using Eq. (8):

The mass fraction of sulfur oil in the extract phase (W3)was calculated by using Eq. (9):

The mass fraction of sulfur oil in the raきnate phase (W4)was calculated by using Eq. (10):

The distribution coefficient (Ksulfur) of sulfur was calculated by using Eq. (11):

The selectivity (S) was calculated as shown in Eq. (12):

Irrespective of ternary matrices or binary matrices, all sulfur contents were determined using a UV fluorescence sulfur analyzer ST0689-2A (Wuhan Yanrun Technology Development Co., Ltd.).

Interaction among H2O, FeCl3, and NMP molecules was monitored with a KBr-pelleted sample on an FT-IR spectrometer (Perkin-Elmer Spectrum 2000 spectrometer,USA) with a scan number of 32 and a resolution of 4 cm-1.XPS spectra were recorded on a Surface Science Instruments spectrometer (SSI, 2803-S) equipped with a monochromatic Al KR X-ray source of 200 W.

3 Results and Discussion

A lot of researches on EDS use single chain alkanes as model oils, includingn-heptane[35],n-octane[8,11,34,36],and isooctane[36]. In fact, besides the less polar alkanes,FCC gasoline also contains the more polar alkenes and aromatics[33]. In order to better reflect the desulfurization performance of extractants, the simulated FCC gasoline containing 48.26% ofn-octane, 29.61% of 1-octene,and 21.54% of toluene was adopted as the oil to be investigated. Moreover, NMP and simulated FCC gasoline were mutually soluble (oil was completely dissolved in extractant), while adding anhydrous FeCl3in NMP could decrease the solubility of oil in the extractant. It is difficult for excessive anhydrous FeCl3to dissolve into NMP completely, so 10% of anhydrous FeCl3could be considered as the optimal amount and then NMP+10% of FeCl3IL could act as the control group in this paper. Since the cost of FeCl3·6H2O is almost half of that of anhydrous FeCl3, an equal mass of FeCl3·6H2O was used instead of anhydrous FeCl3. Then the effects of different sulfur compounds on the desulfurization performance of IL were investigated, using new NMP+10% of FeCl3·6H2O IL as the experimental group and using NMP+10% of FeCl3IL as the control group. In order to study the effect of water (with the water originating from FeCl3·6H2O)on desulfurization performance more strictly, the desulfurization effect of NMP+16.6% of FeCl3·6H2O on simulated FCC gasoline was investigated. Herein, the mass of FeCl3in NMP+16.6% of FeCl3·6H2O and the mass of FeCl3in anhydrous NMP+10% of FeCl3are both 10% of NMP. To study the similarities and differences of water originating from FeCl3·6H2O and free water on the desulfurization effect, the desulfurization performance of added 6.6% of free water into anhydrous NMP+10%of FeCl3IL was investigated. Furthermore, the effects of different amounts of free water on the desulfurization performance of ILs were also investigated.

3.1 Desulfurization performance of aqueous and anhydrous ILs on different sulfur compounds in model oils

The extraction experiments of simulated FCC gasoline samples containing different sulfur compounds (Th,BT, and 2-MT) were carried out using NMP+10% of FeCl3and NMP+10% of FeCl3·6H2O, respectively. The results are shown in Figure 1, Figure 2, and Figure 3,respectively.

Figure 1 Effect of Th-containing simulated FCC gasoline on desulfurization rate and selectivity

Figure 2 Effect of BT-containing simulated FCC gasoline on desulfurization rate and selectivity

Figure 1 shows that, for Th-containing simulated FCC gasoline, the novel ionic liquid NMP+10% of FeCl3·6H2O has better desulfurization rate and selectivity than NMPFeCl3. The desulfurization rate of NMP+10% FeCl3·6H2O(46.2%) is higher than that of NMP-FeCl3(38.4%),increasing by 20.3%. The selectivity of NMP+10% of FeCl3·6H2O (5.46) is also higher than that of NMP-FeCl3(3.85), increasing by 41.8%. Therefore, for Th-containing simulated FCC gasoline, water in NMP+10% of FeCl3·6H2O can improve the desulfurization rate and selectivity.It can be seen from Figure 2 that, for benzothiophenecontaining simulated FCC gasoline, the desulfurization rate and selectivity of NMP+10% of FeCl3·6H2O are also better than those of NMP+10% of FeCl3. The desulfurization rate of NMP+10% of FeCl3·6H2O(51.5%) is higher than that of NMP+10% FeCl3(41.8%),increasing by 23.2%. The selectivity of NMP+10% of FeCl3·6H2O (7.61) is higher than that of NMP+10%of FeCl3(4.09), increasing by 86.3%. Therefore, for benzothiophene-containing simulated FCC gasoline,water in NMP+10% of FeCl3·6H2O can also improve the desulfurization rate and selectivity.

It can be seen from Figure 3 that, for 2-methylthiophenecontaining simulated FCC gasoline, the desulfurization rate and selectivity of NMP+10% of FeCl3·6H2O are still better than those of NMP+10% of FeCl3. The desulfurization rate of NMP+10% of FeCl3·6H2O(38.5%) is by 9.1% higher than that of NMP+10% of FeCl3(29.4%), increasing by 30.9%. The selectivity of NMP+10% of FeCl3·6H2O (4.46) is by 1.15 percentage points higher than that of NMP+10% of FeCl3(3.31),increasing by 34.7%. Therefore, for 2-methylthiophenecontaining simulated FCC gasoline, water in NMP+10%of FeCl3·6H2O can also improve the desulfurization rate and selectivity.

Figure 3 Effect of 2-MT-containing simulated FCC gasoline on desulfurization rate and selectivity

A comparison of Figure 1, Figure 2, and Figure 3 shows that NMP+10% of FeCl3·6H2O has better desulfurization performance than NMP+10% of FeCl3for simulated FCC gasoline samples containing thiophene,benzothiophene, and 2-methylthiophene, respectively,indicating that water in NMP+10% of FeCl3·6H2O can improve the desulfurization rate and selectivity. The order of desulfurization rate and selectivity are as follows:benzothiophene ＞ thiophene ＞ 2-methylthiophene. Because BT is the easiest component to remove from simulated FCC gasoline among the three sulfur compounds, the benzothiophene-containing simulated FCC gasoline was used in the further study on the effect of water on desulfurization performance of IL, as shown below.

3.2 Effect of water originating from FeCl3·6H2O on the performance of ILs for desulfurization of simulated FCC gasoline

Section 3.1 indicates that the desulfurization performance of NMP+10% of FeCl3·6H2O is better than that of NMP+10% of FeCl3. In fact, when the mass of FeCl3·6H2O is equal to 10% of NMP, the mass of FeCl3in NMP+10% of FeCl3·6H2O is equal to only 6% of NMP, while the mass of FeCl3in NMP+10% of FeCl3is equal to 10% of NMP. In order to study the effect of water originating from FeCl3·6H2O on the desulfurization performance of IL more strictly, the desulfurization effect of NMP+16.6% of FeCl3·6H2O on the simulated FCC gasoline was investigated. Herein, the mass of FeCl3in NMP+16.6% of FeCl3·6H2O and the mass of FeCl3in NMP+10% of FeCl3are both equal to 10% of NMP. The results for application of NMP+16.6% of FeCl3·6H2O and the control extractant NMP+10% of FeCl3are shown in Figure 4.

Figure 4 A comparison of desulfurization rates and selectivity between NMP+16.6% of FeCl3·6H2O and NMP+10% of FeCl3

It can be seen from Figure 4 that the desulfurization rate and selectivity of NMP+16.6% of FeCl3·6H2O are higher than those of NMP+10% of FeCl3. The desulfurization rate rose from 41.8% to 47.1%, increasing by 12.7%, and the selectivity was raised from 4.44 to 8.30, increasing by 86.9%, which demonstrates that water in NMP+16.6% of FeCl3·6H2O can effectively improve the desulfurization rate and selectivity of IL. The results indicate that the desulfurization performance is also activated by water in the NMP+16.6% of FeCl3·6H2O based IL.

3.3 Effect of free water on the performance of IL for desulfurization of simulated FCC gasoline

In order to investigate whether free water can also enhance the desulfurization performance of NMPFeCl3IL, the extraction experiment was carried out using NMP+10% of FeCl3+6.6% of H2O and NMP+16.6% of FeCl3·6H2O. The masses of FeCl3and H2O in these two ILs are nearly equal, with the results shown in Figure 5.

It can be seen from Figure 5 that the desulfurization rate and selectivity of NMP+10% of FeCl3+6.6% of H2O are basically the same as those of NMP+16.6% of FeCl3·6H2O.

Figure 5 A comparison of desulfurization rates and selectivity of NMP+16.6% of FeCl3·6H2O and NMP+10% of FeCl3+6.6% of H2O

Therefore, the desulfurization effect activated by the free water in NMP+10% of FeCl3+6.6% of H2O based IL or activated by the water originating from FeCl3·6H2O in NMP+16.6% of FeCl3·6H2O are basically the same.

3.4 Effect of the amount of free water on the performance of IL for desulfurization of simulated FCC gasoline

Section 3.2 and Section 3.3 initially indicated that the desulfurization performance of NMP-FeCl3-H2O could be activated by water in aqueous ILs. However, the effect of different amounts of free water added into NMPFeCl3on the performance of IL for desulfurization of simulated gasoline has not been studied. Therefore,the effects of added 5%, 10%, 20%, and 40% of water,respectively, into NMP+10% of FeCl3on the performance of IL for desulfurization of simulated FCC gasoline were investigated. The results are shown in Figure 6.It can be seen from Figure 6 that, with the gradual increase of the added amount of free water, the desulfurization rate first increases, and then decreases.When the amount of added free water increases from 0 to 5%, the desulfurization rate increases from 41.8%to 47.5%, which is an increase of 13.6%. When the amount of added free water increases from 5% to 10%, the desulfurization rate decreases slightly from 47.5% to 44.9%, which is a decrease of 5.47%. When the amount of added free water increases from 10% to 20%, the desulfurization rate decreases from 44.9% to 35.7%, which is a decrease of 20.5%. When the amount of added free water increases from 20% to 40%, the desulfurization rate decreased significantly from 35.7%to 19.7%, which is a decrease of 44.8%. With the gradual increase of added amount of free water, the selectivity increases gradually. It should be noted that when the amount of added free water is between 5% and 10%, the desulfurization rate and selectivity of aqueous ionic liquid are higher than those of anhydrous ionic liquid. These results show that the addition of a small amount of water not only can improve the selectivity, but also can increase the desulfurization rate, while the addition of excess water can only significantly improve the selectivity and makes the desulfurization rate reduce greatly. Considering the desulfurization rate and selectivity, the optimal amount of water is equal to 5%―10% of NMP. The result further indicated that a small amount of water can activate the desulfurization performance of aqueous NMP-FeCl3IL.

Figure 6 Effect of the amount of free water on desulfurization rate and selectivity

3.5 FTIR spectra of NMP, NMP-FeCl3 and NMPFeCl3·6H2O

In order to explore the desulfurization mechanism activated by water in aqueous NMP-FeCl3IL, the FTIR spectra of NMP, NMP-FeCl3and NMP-FeCl3·6H2O were determined firstly. The results are shown in Figure 7 and Figure 8.

Figure 7 FTIR spectra of NMP

Figure 8 FTIR spectra of NMP-FeCl3 (a) and NMPFeCl3·6H2O (b)

It can be seen from Figure 7 that NMP has a strong absorption peak of C=O at 1 668 cm-1. It can be seen from Figure 8(a) that when FeCl3is added into NMP, the band of C=O at 1 676 cm-1in NMP is split to two peaks and a new peak at lower wavenumber appearing, which is identified at 1 622 cm-1in NMP-FeCl3. It is caused by the formation of coordination bond between the metal ions of Fe and the oxygen atom of C=O in NMP[25]. It can be seen from Figure 8(b) that when FeCl3·6H2O is added into NMP, the peak occurring at 1 622 cm-1in Figure 8(b) almost disappears (with the intensity of the peak greatly weakening), which might be attributed to water in NMP-FeCl3·6H2O which could inhibit the formation of a coordination bond between the metal ions of Fe and the oxygen atom of C=O in NMP.

3.6 High resolution XPS spectrum of NMP-FeCl3-H2O

In order to explore the coordination mode of NMP-FeCl3-H2O, the high resolution XPS spectrum of NMP- FeCl3-H2O was investigated. The Fe2p spectrum is shown in Figure 9 and the O1s spectrum is shown in Figure 10.

Figure 9 XPS high-resolution spectrum of Fe2p of NMP-FeCl3-H2O

Figure 10 XPS high-resolution O1s spectrum of NMP-FeCl3-H2O

Upon studying the Fe2p spectrum shown in Figure 9,two characteristic peaks of Fe 2p3/2 and Fe 2p1/2 are observed at 711.3 eV and 724.5 eV, respectively. The Fe3+cation of (FeCl3)2(DMF)3.3(H2O)2.7reported by Young Jeon Kim and Chong Rae Park[29]also showed two characteristic peaks of Fe 2p3/2 and Fe 2p1/2 at 711.6 eV and 724.8 eV, respectively. They considered that the characteristic peaks of Fe3+cation at 711.6 eV and 724.8 eV nearby demonstrated the existence of Fe3+-O.Figure 9 clearly shows that the two characteristic peaks coincide with those of Fe3+cation mentioned in the above literature, which indicates the existence of Fe3+-O in NMP-FeCl3-H2O. In NMP-FeCl3-H2O, there are only two possible coordination states of Fe3+-O, namely, the coordination of the Fe3+ion with the oxygen atom of C=O in NMP or the coordination of the Fe3+ion with the oxygen atom in water. The FITR spectrum of NMPFeCl3-H2O in Section 3.6 has showed that water in NMPFeCl3-H2O could inhibit the formation of a coordination bond between the metal ions of Fe and the oxygen atom of C=O in NMP. Therefore, it is highly possible that the Fe3+ion may coordinate with the oxygen atom in water.

Furthermore, in order to prove that the Fe3+-O in NMPFeCl3-H2O indeed originates from the coordination between Fe3+ion and oxygen atom in water, it is thus necessary to examine closely the O 1s spectra of the NMPFeCl3-H2O as shown in Figure 10. The O 1s spectra of(FeCl3)2(DMF)3.3(H2O)2.7reported by Young Jeon Kim and Chong Rae Park[29]apparently showed a peak at 532.0 eV because of the coordination between Fe3+ion and oxygen atom in water, along with another peak at 530.9 eV because of the coordination between Fe3+ion and oxygen atom of C=O. Figure 10 only shows a peak at 532.1 eV (close to 532.0 eV) because of the coordination between Fe3+ion and the oxygen atom in water, but does not show another peak at around 530.9 eV as reported in the literature. This result indicates that in the NMP-FeCl3-H2O based IL, the Fe3+ion only coordinates with the oxygen atom in water.

3.7 Desulfurization mechanism activated by water in aqueous NMP-FeCl3 IL

According to experimental results, and the FITR spectrometry and XPS spectrometry analyses, we speculate that the desulfurization mechanism of aqueous NMP-FeCl3IL (NMP-FeCl3-H2O) might be stated by means of the reactions shown in Figure 11. On the one hand, the oxygen atom of C=O in NMP molecule carries negative charge and is easy to form hydrogen bond association with a hydrogen atom (Number 1) in H2O molecule, which can prevent the Fe3+ion of FeCl3from coordinating with the oxygen atom of C=O in NMP. Thus,the empty orbital of Fe3+ions could coordinate with the solitary electrons on the oxygen atom in H2O molecule.As a result, the electron cloud of the oxygen atom in H2O inclines to the Fe3+ion, and the electron cloud of another hydrogen atom in the H2O molecule inclines to the oxygen atom, which can lead to a decreased electron cloud density of hydrogen atom (Number 2) in H2O molecule. Therefore, the hydrogen atom (Number 2)is activated. The activated hydrogen atom is easier to form hydrogen bond association with the sulfur atom of thiophene. On the other hand, because the oxygen atom of C=O in NMP can easily form a hydrogen bond association with hydrogen atom (Number 1) in H2O, the electron cloud of oxygen atom of C=O inclines to hydrogen atom.Then the electron cloud of the nitrogen atom in NMP inclines to the oxygen atom of C=O. The electron cloud of the methyl group in NMP inclines to the nitrogen atom,which can lead to a decreased electron cloud density of hydrogen atom in methyl group and a weak hydrogen bond association between the hydrogen atom of methyl group in NMP and sulfur atom of thiophene.

Figure 11 The possible desulfurization mechanism of NMP-FeCl3-H2O

The desulfurization performance of NMP-FeCl3-H2O is better than that of NMP-FeCl3, which may be related to the difference of desulfurization mechanisms between these two ILs. The desulfurization mechanism of NMP-FeCl3-H2O has been discussed above, while the desulfurization mechanism of NMP-FeCl3has been proposed in previous research and is stated as follows. (1)The oxygen atom of C=O in NMP molecule coordinates with the empty orbit of Fe3+ion of FeCl3, then the electron cloud of carbonyl oxygen atom in NMP inclines to Fe3+ion, the electron cloud of nitrogen atom in NMP inclines to oxygen atom of C=O, and the electron cloud of methyl in NMP inclines to nitrogen atom[29]. This leads to a decreased electron cloud density of hydrogen atom in methyl group and a weak hydrogen bond association between hydrogen atom of methyl group in NMP and sulfur atom of thiophene[37]. (2) The unoccupied empty orbit of Fe3+ion could coordinate with the solitary electrons of sulfur atom of thiophene[25]. The difference of desulfurization mechanisms between these two ILs is that the hydrogen bond association between the activated hydrogen atom in NMP-FeCl3-H2O and the sulfur atom of thiophene is stronger than the coordination between the unoccupied empty orbit of Fe3+in NMP-FeCl3and the solitary electrons of sulfur atom of thiophene.

The selectivity of aqueous ionic liquid is higher than that of anhydrous ionic liquid, which might be closely related to the polar difference between ionic liquid and simulated FCC gasoline. According to the detailed description of the COSMO-RS model referred to in the literature, the simulated FCC gasoline components in this paper are located almost fully in the nonpolar region, indicating their strong hydrophobic character[31]. The hydrophobicity of the simulated FCC gasoline components decreases in the following order: 1-octane ＞ 1-octene ＞ toluene＞ thiophene[31]. Therefore, the hydrophobicity of the simulated FCC gasoline is stronger than that of thiophene.The addition of strong polar water into NMP-FeCl3increases the polarity of ionic liquid, while enlarging the polar difference between simulated FCC gasoline and ionic liquid. Thus, the simulated FCC gasoline is more difficultly dissolved in the ionic liquid than thiophene, resulting in a decreased solubility of fuel in the ionic liquid.

4 Conclusions

A series of novel aqueous ionic liquids (NMP-FeCl3-nH2O) were prepared. Adding a small amount of water into the ionic liquid NMP-FeCl3could effectively improve the desulfurization rate and selectivity, and a small amount of water could activate the desulfurization performance of the ionic liquid NMP-FeCl3. When the added amount of anhydrous FeCl3was equal to 10% of the mass of NMP, the optimal amount of added water was 5%―10% of NMP. The NMP-FeCl3-H2O had good extraction performance on thiophene, benzothiophene,and 2-methylthiophene, and the desulfurization rate and the selectivity of NMP-FeCl3-H2O were better than those of NMP-FeCl3. The results of FTIR spectrometry and XPS spectrometry analyses showed that Fe3+ion does not coordinate with the oxygen atom of C=O in NMP but coordinates with the oxygen atom in water. It is possible that the desulfurization mechanism activated by a small amount of water in NMP-FeCl3-H2O could be attributed to the fact that a hydrogen atom in water is associated with the oxygen atom of C=O in NMP, the oxygen atom in water could coordinate with the Fe3+ion in FeCl3,resulting in the activation of another hydrogen atom in water. The activated hydrogen atom in water is easily associated with the sulfur atom of thiophene to improve the desulfurization performance.

Acknowledgments:This work was financially supported by the science and technology guidance plan project through the China Petroleum and Chemical Industry Federation (Contract No.2014-02-01).