Desulfurisation of oils using ionic liquids: selection of cationic and anionic components to enhance extraction efficiency

Abstract

Extraction of dibenzothiophene from dodecane using ionic liquids as the extracting phase has been investigated for a range of ionic liquids with varying cation classes (imidazolium, pyridinium, and pyrrolidinium) and a range of anion types using liquid–liquid partition studies and QSPR (quantitative structure–activity relationship) analysis. The partition ratio of dibenzothiophene to the ionic liquids showed a clear variation with cation class (dimethylpyridinium > methylpyridinium > pyridinium ≈ imidazolium ≈ pyrrolidinium), with much less significant variation with anion type. Polyaromatic quinolinium-based ionic liquids showed even greater extraction potential, but were compromised by higher melting points. For example, 1-butyl-6-methylquinolinium bis{(trifluoromethyl)sulfonyl}amide (mp 47 °C) extracted 90% of the available dibenzothiophene from dodecane at 60 °C.

---

Introduction

The current state of the art for removing sulfur-containing compounds in refineries is catalytic hydrogenation.¹ However, there is an increased interest in alternative processes to hydrodesulfurisation driven by regulatory requirements to reduce the content of sulfur in fuels and the current limitations of hydrodesulfurisation technologies to fully remove refractory aromatic sulfur compounds, particularly alkylated benzothiophenes and dibenzothiophenes.² Approaches³ based on adsorption⁴ or complexation,⁵ oxidative extraction,⁶ biodesulfurisation,⁷ and ultrasonic treatments⁸ have been explored.

Liquid–liquid extractive desulfurisation (EDS) systems using ionic liquids as the extracting phase have been reported.^9–12 Many ionic liquids are immiscible, or only partially miscible, with hydrocarbons,¹³ and these studies have attempted to use the unique solvent characteristics of ionic liquids^13,14 to provide high extraction ratios and greater selectivity compared to molecular solvents.¹⁵ The principal focus of these studies is the extraction of refractory sulfur-containing polyaromatics, such as dibenzothiophene (DBT, Fig. 1) and 4,6-dimethyldibenzothiophene from dodecane as a model oil system.

Fig. 1 Chemical structure of dibenzothiophene (DBT).

The ionic liquids studied have been chosen principally for their ready availability at relatively low cost and environmental impact. Following the initial report by Bösmann et al.,⁹ Jess and co-workers have promoted 1,3-dialkylimidazolium alkylsulfates,¹⁰ whereas Li and co-workers¹² favour 1,3-dialkylimidazolium alkylphosphates as candidates for extractive desulfurisation technologies.

While simple hydrocarbons are poorly soluble in many ionic liquids,¹³ aromatic compounds, such as benzene and toluene, can be highly soluble,¹⁶ with interactions between the ionic liquid ions and aromatic solutes leading to liquid clathrate and solid-state inclusion complexes.^17,18 However, despite these favourable interactions,¹⁹ ionic liquids have not proven to be exceptional solvents for liquid–liquid separation of aromatics from aliphatic hydrocarbons.^20,21

For sulfur-containing aromatics, such as DBT, 1-butyl-3-methylimidazolium octylsulfate ([C₄mim][C₈H₁₇SO₄]) was reported¹⁰ to have the highest partition ratio (K_D) for DBT from dodecane (K_D = 1.9) of all the ionic liquids, with the exception of chloroaluminate ionic liquids, which are hydrolytically unstable and reactive to many substrates. An alternative reaction/extraction approach has also been described combining hydrogen peroxide/ethanoic acid oxidation of the aromatic sulfur compounds to sulfones and sulfoxides, which are then extracted with much higher partition ratios to the ionic liquid phase.²²

Interestingly, most of the published data relates to ionic liquids with imidazolium cations.^9–12 1-Alkylpyridinium salts, for example, are described in the patent literature only as organic salt additives in liquid–liquid separations.²³ With the variety of ionic liquids available, especially considering those with different types of cations, we reasoned that a more general study to examine the influence of different cationic and anionic species on the partitioning might lead to better ionic liquid extracting systems.

Here, we describe an investigation of the role of different cationic and anionic components of an ionic liquid extracting phase on the efficiency of partitioning of DBT from dodecane as a model for extractive desulfurisation of oils.

Results and discussion

Extraction of a model system consisting of dodecane containing DBT as a source of polyaromatic sulfur with a sulfur content of 500 ppm (2880 ppm, 0.0156 mole l^–1 of DBT) was studied with a range of ionic liquids, chosen in order to assess the role of the different cation classes (imidazolium, pyridinium, methylpyridinium, pyrrolidinium, see Fig. 2), and anion types, respectively, both with common counterions. This enabled the study of the respective roles of cation, anion or cation–anion pairs on the extraction profile.

Fig. 2 Chemical structure of seven of the ionic liquid cations studied here: (A) 1-butyl-3-methylimidazolium ([C₄mim]⁺); (B) 1-butylpyridinium ([C₄py]⁺); (C) 1-butyl-1-methylpyrrolidinium ([C₄mpyrr]⁺); (D) 1-butyl-4-methylpyridinium ([C₄⁴mpy]⁺); (E) 1-butyl-3-methylpyridinium ([C₄³mpy]⁺); (F) 1-butyl-3,4-dimethylpyridinium ([C₄^2,4dmpy]⁺); (G) 1-butyl-3,5-dimethylpyridinium ([C₄^2,5dmpy]⁺).

Extraction experiments were performed on a 4 cm³ scale using equal volumes (2 cm³ each) of ionic liquid as the extracting phase and dodecane containing DBT at 0.0156 molar concentration (corresponding to 500 ppm of sulfur) in the initial organic phase. The volume of the ionic liquid phase was not observed to change during the contact and mixing with dodecane, indicating low co-solubility of the phases. The hydrocarbon phase was analysed before and after extraction for DBT by GC-MS.

Initially, we compared selected systems from the literature to establish that the experimental protocol used in this work could replicate previously reported results at 25 and 60 °C from Jess and co-workers.^9,10 Comparing the data in Table 1 indicates a reasonable compatibility between the results for 1-butyl-3-methylimidazolium octylsulfate ([C₄mim][C₈H₁₇SO₄]), and 1-ethyl-3-methylimidazolium ethylsulfate ([C₂mim][C₂H₅SO₄]) at 25 °C and 1-butyl-3-methylimidazolium hexafluorophosphate ([C₄mim][PF₆]), 1-butyl-3-methylimidazolium tetrafluoroborate ([C₄mim][BF₄]), and [C₄mim][C₈H₁₇SO₄], at 60 °C. We obtained a reasonable correlation with the literature results‡ confirming the consistency of the experimental methodology, although we notably obtained a consistently lower K_D in the [C₄mim][C₈H₁₇SO₄] system.

Table 1 Comparison of DBT partition ratios (K_D), with literature data from references 9 and 10 in parentheses^a

Ionic liquid	KD
a General conditions: 25 or 60 °C, volume ratio model oil : ionic liquid = 1 : 1, mixing time 1 h, equilibration time 15 min, initial sulfur content 500 ppm as DBT.
	25 °C	60 °C
[C4mim][PF6], 4	—	0.8 (0.7/0.9)
[C4mim][BF4], 1	—	0.9 (1.0)
[C4mim][C8H17SO4], 2	1.7 (1.9)	1.6 (2.1)
[C2mim][C2H5SO4]	0.9 (0.8)	—

---

Having established that this protocol was consistent with the previous literature, subsequent screening experiments were performed at 40 °C. This enabled better experimental stability of temperature and gave improved mixing characteristics for some of the more viscous ionic liquids.

The ionic liquids used contained [C₄mim]⁺ (1–7), [C₄py]⁺ (8,9), [C₄⁴mpy]⁺ (10–13), [C₄³mpy]⁺ (14–17), [C₄^2,4dmpy]⁺ (18), [C₄^2,5dmpy]⁺ (19), and [C₄mpyrr]⁺ (20) cations (see Fig. 2); the extraction results for removing DBT (at 500 ppm sulfur content) from dodecane for each ionic liquid with the anions hexafluorophosphate [PF₆]^–, octylsulfate [OcSO₄]^–, trifluoromethanesulfonate [OTf]^–, tetrafluoroborate [BF₄]^–, bis{(trifluoromethyl)sulfonyl}amide [NTf₂]^–, thiocyanate [SCN]^–, and ethanoate [CH₃CO₂]^– are shown in Table 2 as percentage extracted and as the partition ratio K_D.

Table 2 Ionic liquids screened and extraction of dibenzothiophene (concentration equivalent to 500 ppm of sulfur) from dodecane at 40 °C using equal volumes of ionic liquid and dodecane, expressed as percent extracted and as partition ratio (K_D)^a

	Ionic liquid	Percent extracted (%)	KD	Standard error (%)
a General conditions: 40 °C, volume ratio model oil : ionic liquid = 1 : 1, mixing time 1 h, equilibration time 15 min, initial sulfur content 500 ppm as DBT
1	[C4mim][BF4]	47	0.9	0.6
2	[C4mim][C8H17SO4]	63	1.7	0.9
3	[C4mim][CF3SO3]	50	1.0	0.6
4	[C4mim][PF6]	53	1.2	0.5
5	[C4mim][NTf2]	50	1.0	2.8
6	[C4mim][SCN]	66	1.9	1.4
7	[C4mim][CH3CO2]	61	1.6	0.1
8	[C4py][NTf2]	55	1.2	3.4
9	[C4py][BF4]	43	0.8	1.1
10	[C4^4mpy][NTf2]	76	3.3	1.0
11	[C4^4mpy][BF4]	70	2.3	0.4
12	[C4^4mpy][SCN]	79	3.8	0.87
13	[C4^4mpy][CF3SO3]	72	2.6	0.58
14	[C4^3mpy][NTf2]	77	3.4	0.5
15	[C4^3mpy][BF4]	70	2.3	0.4
16	[C4^3mpy][SCN]	83	4.9	0.5
17	[C4^3mpy][CF3SO3]	69	2.3	1.6
18	[C4^2,4dmpy][NTf2]	83	4.9	0.28
19	[C4^2,5dmpy][NTf2]	81	4.0	0.53
20	[C4mpyrr][NTf2]	47	0.9	1.9

---

Anion dependence with common [C₄mim]⁺ cation

The influence of the ionic liquid anion on the partitioning was explored for [C₄mim]⁺ ionic liquids with the series of anions; [PF₆]^– (1), [C₈H₁₇SO₄]^– (2), [OTf]^– (3), [BF₄]^– (4), [NTf₂]^– (5), [SCN]^– (6), and [CH₃CO₂]^– (7). EDS data for the first four ionic liquids here have previously been reported (as described above); the last three ionic liquids have not been previously studied in the context of desulfurisation.

Comparison of the data for these ionic liquids shows that the choice of anion has little direct effect on the extraction ability in most cases (shown as percentage DBT extracted in Fig. 3). Only a relatively small variation in partition ratio with anion type was observed, with K_D in the range 0.9–1.9, which corresponds to DBT extraction between 47 and 66%. The ordering of the partition ratio with anion type is [BF₄]^– < [NTf₂]^– < [OTf]^– < [PF₆]^– < [C₈H₁₇SO₄]^– < [CH₃CO₂]^– < [SCN]^–. The four ionic liquids [C₄mim][PF₆], [C₄mim][C₈H₁₇SO₄], [C₄mim][OTf], and [C₄mim][BF₄] have similar partition ratios (K_D = 1–1.2), which are also comparable with the data from Jess and co-workers^9,10 for [C₄mim]X ionic liquids measured at 60 °C.

Fig. 3 Percentage DBT extracted by [C₄mim]X (1–7) from dodecane at 40 °C (1 : 1 volumes, 500 ppm sulfur in dodecane phase as DBT).

Interestingly, the ionic liquids with the ethanoate and thiocyanate anions ([C₄mim][CH₃CO₂] and [C₄mim][SCN]) gave the highest partition ratios of all the [C₄mim]⁺ salts studied here. Jess and co-workers^9,10 have suggested that anion size was an important factor in achieving better extraction, supported by the high value of K_D for [C₄mim][C₈H₁₇SO₄]. However, our data suggest that high relative ionicity, rather than the size, of the anion may be more significant in promoting extraction.

Cation dependence with common anion

Having established that changing the anion of the ionic liquid has only a small variation on K_D for dibenzothiophene, despite the significant variation in the usual physical properties of ionic liquids driven by the anion type,¹⁵ the influence of the cation type was investigated. The key common classes of cations used to prepare ionic liquids were studied as bis{(trifluoromethyl)sulfonyl}amide salts. Data for [C₄mim][NTf₂] (5), [C₄py][NTf₂] (8), [C₄⁴mpy][NTf₂] (10), [C₄³mpy][NTf₂] (14), [C₄^2,4dmpy][NTf₂] (18), [C₄^2,5dmpy][NTf₂] (19), and [C₄mpyrr][NTf₂] (20) gave a much wider distribution in K_D, ranging from 0.9 to 4.9 (47–83% extraction of DBT), than that observed from the change in anion type, and are shown graphically in Fig. 4.

Fig. 4 Percentage DBT extracted by [A][NTf₂] (5, 8, 10, 14, 18–20) from dodecane at 40 °C (1 : 1 volumes, 500 ppm sulfur in dodecane phase as DBT).

The ability of the ionic liquids to extract DBT follows the order, pyrrolidinium ≈ imidazolium ≈ pyridinium < methylpyridinium < dimethylpyridinium, with the two 1-butyl-dimethylpyridinium isomers ([C₄^2,4dmpy][NTf₂] and [C₄^2,5dmpy]][NTf₂]) giving the highest partition ratios for DBT observed for any of the systems. Both of these ionic liquids extracted between 81–83% of the DBT from dodecane, followed by the 1-butyl-methylpyridinium systems ([C₄⁴mpy][NTf₂] and [C₄³mpy][NTf₂]), which where able to extract 76–77% of the available DBT in a single contact. The three remaining bis{(trifluoromethyl)sulfonyl}amide ionic liquids ([C₄mim][NTf₂], [C₄py][NTf₂], and [C₄mpyrr][NTf₂]), showed poorer extraction ability, removing between 47–55% of DBT.

A comparable trend was also observed for the smaller set of corresponding tetrafluoroborate ionic liquids; [C₄mim][BF₄] and [C₄py][BF₄] gave similar, poor, extraction of DBT (K_D = 0.9 and 0.8, respectively), whereas [C₄⁴mpy][BF₄], and [C₄³mpy][BF₄] exhibited good extraction of DBT (K_D = 2.3 in each case), almost comparable to the high partition ratios measured in the corresponding bis{(trifluoromethyl)sulfonyl}amide containing ionic liquids, extracting 69–70% of the DBT from the dodecane phase.

Four additional ionic liquids with 1-butyl-4-methylpyridinium and 1-butyl-3-methylpyridinium cations; [C₄⁴mpy][SCN], [C₄⁴mpy][CF₃SO₃], [C₄³mpy][SCN], and [C₄³mpy][CF₃SO₃] also were examined. All showed good extraction characteristics, comparable to those of the corresponding bis{(trifluoromethyl)sulfonyl}amide analogues. [C₄⁴mpy][CF₃SO₃] and [C₄³mpy][CF₃SO₃] with trifluoromethanesulfonate anions extracted around 70% of the DBT with a single contact, comparable with the corresponding bis{(trifluoromethyl)sulfonyl}amide and tetrafluoroborate ionic liquids. Even higher partition ratios were measured with the thiocyanate–containing ionic liquids [C₄⁴mpy][SCN] and [C₄³mpy][SCN] (K_D = 3.8 and 4.9 respectively, corresponding to 79 and 83% extraction). For each group of cations examined, the ionic liquids with thiocyanate anions consistently showed the greatest ability to extraction DBT from dodecane. These results support the observations of cation dominance in the partitioning efficiency, and significantly less importance of the anion, although the effect of the anion is incontrovertible.

Influence of cation isomerisation

There are few data to differentiate between the extraction ability of the two sets of isomeric ionic liquids with 1-butyl-methylpyridinium cations ([C₄⁴mpy]⁺ and [C₄³mpy]⁺), or between the two results for 1-butyldimethylpyridinium isomers ([C₄^2,4dmpy][NTf₂] and [C₄^2,5dmpy][NTf₂]). Only small variations in K_D with the positions of the methyl groups on the ring with comparable partition ratios are observed for each pair of ionic liquids.

QSPR analysis of solvent group contributions to extraction

Explaining the differences in the extraction of DBT by the different ionic liquids is challenging. Factors such as the size/shape/aromaticity and/or charge distribution appears to be the important criteria. Both [C₄mim]⁺ and [C₄py]⁺ cations contain ten non-hydrogen atoms, compared to eleven for [C₄³mpy]⁺ and [C₄³mpy]⁺ and twelve for the [C₄^2,4dmpy]⁺ and [C₄^2,5dmpy]⁺ cations, which may lead to the small relative increase in affinity for aromatic components. However, the methyl group is not usually considered to have a significant inductive effect on aromatic rings. Similarly, the presence, or absence, of aromaticity may not be the only feature of importance; [C₄mpyrr]⁺ also contains ten non-hydrogen atoms, but in contrast is neither flat nor an aromatic cation.

Quantitative structure–property relationship (QSPR) approaches have been used with some success in predicting the physical properties of ionic liquids.^24–30 A QSPR analysis of the partitioning data was used to identify whether a correlation could be obtained, and which, if any, of the resulting descriptors could be related to the extracting ability of the ionic liquids. The study focused on the influence of the cation using data from twelve ionic liquids with the common bis{(trifluoromethyl)sulfonyl}amide anion (4, 8, 10, 14, 18–20 from Table 2 and five additional ionic liquids§).

An initial series of one-, two- and four-parameter correlations were screened using the heuristic method³¹ implemented in CODESSA,³² a complete description of the QSPR treatment and analysis is available as supplementary material.† The descriptors identified from the one-, two- and four-parameter correlations indicated that the size, shape and aromaticity of the cation were important contributors. However, with such a small number of observations (12), chance correlations are likely if the number of screened variables is large.³³ Notably, a number of descriptors from the four parameter correlation (WPSA₁, WPSA₂ and FPSA₃) were related and describe weighted positive charge size and distribution (suggesting that the cation size has a significant correlation with extraction ability). Topliss and Edwards suggest that all duplicate descriptors with intercorrelation r² > 0.8 should be discarded for the final modelling,³³ which also dramatically reduces the likelihood of chance correlations, since the number of descriptors screened is reduced.

The four-descriptor correlation was rescreened using a more stringent rejection criterion for duplicate descriptors (2-parameter intercorrelation r² ≥ 0.80); this reduced the descriptor pool to 31. The best regression correlation, shown in Fig. 5 (r² = 0.9835, cvr² = 0.9452) was:

%_Extract = 2.9504e01(ε_HOMO–1) – 1.3835e04 (P_σ–σ) – 7.9048e02 P′^C_AB – 1.2512e02(S_YZ^r) + 1.5082e04

Fig. 5 Comparison of experimental and calculated percentage DBT extraction from the regression analysis (r² = 0.9835, cvr² = 0.9452) for the ionic liquids 4, 8, 10, 14, 18–20 and five additional ionic liquids; (a) [CH₃OC₂H₄mim][NTf₂]; (b) [NMeBu(CH₂CH₂OH)₂] [NTf₂]; (c) [HOEtmmor][NTf₂]; (d) [C₄⁴CNpy][NTf₂]; (e) [C₆⁴CNpy][NTf₂].

ε_HOMO–1: HOMO-1 energy (2nd ionisation potential)

P_σ–σ: maximum σ–σ bond order

P′^C_AB: minimum (>0.1) bond order for a C atom

S_YZ^r: YZ shadow/YZ rectangle

With this correlation, the best interpretation of the descriptors is that the extraction is governed by a combination of solute–solvent interactions with the aromatic system (ε_HOMO–1, P_σ–σ, P′^C_AB) and is also controlled by the topology of the cation, with larger/flatter cations (ZY shadow/XY rectangle) performing best. However, these criteria can also be interpreted in a number of alternative ways: for example, in order to obtain a larger/flatter cation, additional carbon atoms can be added (pyridinium, methylpyridinium, dimethylpyridinium, etc.), which increases the number of carbon atoms and the reduction in sigma–sigma bond order all as intercorrelated descriptors.

Given the small number of observations, caution must be exerted when interpreting these results. Yet, where overall conclusions can be drawn, they point towards transfer interaction ability, aromaticity and the shape of the molecule as being influential with this extraction process and suggest that ionic liquids containing cations with greater aromatic character, for example containing polyaromatic cations, would give enhanced extraction. Ionic liquids containing a polyaromatic quinolinium and isoquinolinium cations are known, but tend to have relatively high melting points, a common feature of the larger ring structure.^34,35 Two examples, 1-butylisoquinolinium bis{(trifluoromethyl)sulfonyl}amide (mp 56 °C), and 1-butyl-6-methylquinolinium bis{(trifluoromethyl)sulfonyl}amide (mp 47 °C) were prepared and tested for extraction of DBT at 60 °C. Both ionic liquids performed exceptionally well, giving 80% extraction (K_D = 4) with 1-butylisoquinolinium bis{(trifluoromethyl)sulfonyl}amide and 90% extraction (K_D = 9) for 1-butyl-6-methylquinolinium bis{(trifluoromethyl)sulfonyl}amide, consistent with the QSPR predictions.

Practical considerations for implementation and scale-up

These results show how the partition ratio (K_D) for dibenzothiophene between dodecane and ionic liquids is strongly influenced, and can be controlled, by the selection of the cation and to a much lesser extent anion of the ionic liquid. Using these data to develop practical liquid–liquid extraction processes using real refinery feeds requires a number of additional considerations.

Real feeds (such as kerosene or diesel) contain a mixture of components, including paraffinic, aromatic, alkylaromatic and a range of polyaromatic hydrocarbons. The co-miscibility of the ionic liquids and feeds, selectivity and specificity of the extracting phase for polyaromatic sulfur compounds over hydrocarbons and polyaromatic hydrocarbons become important, and in this case should show high capacity with rejection of hydrocarbons. Initial tests have shown that the partition ratio for 4,6-dimethyldibenzothiophene (DMDBT) follows a similar trend to those for DBT, although with lower K_D (0.77 for DMDBT compared to 1.6 for DBT with [C₄⁴CNpy][NTf₂], (d), consistent with the data previously described by Eßer and co-workers¹⁰ for [C₄mim][C₈H₁₇SO₄] (K_D = 1.9 for DBT and 0.8 for DMDBT). These results are part of continuing work to examine the effects of specific target ionic liquids in real feeds, considering selectivity, extraction and recovery of the extractants from the ionic liquids.

Secondly, the question of removal (and recovery) of extracted materials from the ionic liquid phase needs to be considered. In a conventional liquid–liquid extraction process, secondary stripping stages would be used to recover extracted material. In practical terms, the need for efficient re-extraction often negates the advantages of high K_D values from the initial extraction. An alternative to extraction is distillation, however, this is not an appealing approach to recover high boiling polyaromatic sulfur compounds, such as DBT (bp 332–333 °C), even when the relatively high thermal stability of many ionic liquids is taken into consideration.

However, since liquid–liquid EDS is not currently considered competitive with conventional hydrodesulfurisation technologies,³ advancing this approach to desulfurisation may also require non-conventional re-extraction techniques. Nie and co-workers¹² have shown that polyaromatic sulfur compounds can be reprecipitated from water-soluble ionic liquids on addition of water, as one approach to recovery and reuse of ionic liquid phases. Electrochemical approaches to hydrodesulfurisation recovering the hydrocarbon-portion of the molecules may also be appropriate, especially if significantly higher concentration factors and loadings can be obtained using ionic liquids (or other) novel extracting phases.

Summary

Liquid–liquid extraction of dibenzothiophene from dodecane has been investigated for a range of ionic liquids with varying cation classes (imidazolium, pyridinium, pyrrolidinium, and quinolinium) and with a range of anion types. Partition ratios for dibenzothiophene to the ionic liquids ranged from 0.8 to 9, and showed clear variation with cation class. The polyaromatic quinolinium-based ionic liquids demonstrated the best extractive ability. However, these and other polyaromatic cations, such as 1,3-dialkylbenzimidazolium, are technically limited by their higher melting points and only tend to form low-melting point ionic liquids with highly flexible perfluorinated anions, such as bis{(trifluoromethyl)sulfonyl}amide. This highlights the importance of only using theoretical predictions in tandem with empirical data on rheological and thermophysical properties in order to obtain realistic target materials.

Amongst the monocyclic systems, the ionic liquids can be ranked by cation; methylpyridinium ≥ pyridinium ≈ imidazolium ≈ pyrrolidinium, with much less significant variation with anion type. The ionic liquids with [C₄^2,4dmpy]⁺ and [C₄^2,5dmpy]⁺ cations showed the best extraction performance with 81–83% of the DBT removed in a single contact. Ionic liquids containing 1-butyl-n-methylpyridinium cations (n = 3 or 4) gave good DBT extraction (70–83%), irrespective of the anion present.

Surprisingly, the ionic liquids with ethanoate and thiocyanate anions gave the best extraction performance with each cation, indicating that more benign (and much cheaper) anions could be used as alternatives to the perfluorinated bis{(trifluoromethyl)sulfonyl}amide and tetrafluoroborates, especially when coupled to cations, such as 1-butyl-3-methylpyridinium (83% extraction).

Experimental

Reagents

Dodecane and dibenzothiophene were purchased from Aldrich and used as received. Solutions of dibenzothiophene in dodecane (2880 ppm, 0.0156 mole l^–1 of DBT, corresponding to 500 ppm of sulfur) were prepared by dissolving dibenzothiophene (288 mg, 0.00156 mole) in dodecane (100 cm³).

Ionic liquids

The ionic liquids 2, 3, 6, 7, and 11 were purchased from Fluka, 9 was purchased from Acros. All were used as received. The ionic liquids 1, 4, 5, 8, 10, and 12–20 were synthesised by alkylation of the respective 1-methylimidazole, pyridine, 4-methylpyridine, 3-methylpyridine, 3,4-dimethylpyridine, 3,5-dimethylpyridine or 1-methylpyrrolidine heterocyclic bases with 1-bromobutane, followed by anion metathesis. 1-Ethyl-3-methylimidazolium ethylsulfate was prepared using the previously published synthetic route.³⁶ The ionic liquids were characterised by proton NMR spectroscopy, ion-chromatography and Karl Fisher titration.

Partitioning experiments

Extraction experiments were performed under standardised conditions: equal volumes (2 cm³) of dodecane (containing 500 ppm sulfur as dibenzothiophene) and ionic liquid were mixed for 1 h at 950 rpm in 25 × 75 mm vials with 10 mm triangular magnetic stirring bars, using an Anachem stem-block for 1 h at 25, 40, or 60 °C, followed by equilibration at temperature for 15 min without stirring. A sample (1 cm³) of the upper dodecane phase was taken by pipette, taking care not to disturb the phase boundary, and was analysed by GC-MS (PerkinElmer Clarus 500 with an ESGE SolGel column) for dibenzothiophene using direct injection of the samples, and comparing the integral with the phase before extraction. The percentage of DBT extracted and partition ratio of DBT to the ionic liquid were calculated by difference from the integral of the dibenzothiophene signal using the equation:

K_D = [DBT]_{ionic liquid}/[DBT]_dodecane

which reduces, when the phase volumes are equal, to

K_D = [I_initial–I_dodecane]/[I_dodecane]

where I is the peak integral analysed by GC-MS.

Measurements were performed in triplicate, and replicate experiments using different batches of ionic liquid, where available, were also performed. Mass balance was confirmed by analysis of both phases (dodecane and ionic liquid) using HPLC (Agilent 1200 series with RP18 column in acetonitrile/water) in a number of cases.

Acknowledgements

We thank BP (JDH, XZ, IL-M), QUILL and its Industrial Advisory Board, EPSRC through the Portfolio Partnership Scheme Grant EP/D029538/1 (KRS), and the Ministerio de Educación y Ciencia of Spain (grant CTQ2005-07676 to GS) for funding and support.

References

1. J. G. Speight, in ‘Kirk-Othmer Encyclopaedia of Chemical Technology’, ed. A. Seidel, Vol. 18, John Wiley & Sons, Inc., Hoboken, New Jersey, 2007, pp. 1–49.
2. G. C. A. Schuit and B. C. Gates, AIChE J., 1973, 19, 417; X. L. Ma, K. Y. Sakanishi and I. Mochida, Ind. Eng. Chem. Res., 1994, 33, 218.
3. E. Ito and J. A. R. van Veen, Catal. Today, 2006, 116, 446.
4. G. P. Khare, World Pat., WO0132805, 2001.
5. M. Macaud, M. Sévignon, A. Favre-Réguillon, M. Lemaire, E. Schulz and M. Vrinat, Ind. Eng. Chem. Res., 2004, 43, 7843; M. Sévignon, M. Macaud, A. Favre-Réguillon, J. Schulz, M. Rocault, R. Faure, M. Vrinat and M. Lamaire, Green Chem., 2005, 7, 413.
6. J. T. Sampanthar, H. Xiao, J. Dou, T. Y. Nah, X. Rong and W. P. Kwan, Appl. Catal., B, 2005, 63, 85.
7. F. Li, P. Xu, J. Feng, L. Meng, Y. Zheng, L. Luo and C. Ma, Appl. Environ. Microbiol., 2005, 71, 276; S. A. Denome, C. Oldfield, L. J. Nash and K. D. Young, J. Bacteriol., 1994, 176, 6707.
8. R. W. Gunnerman, US Pat. 6,827,844, 7 Dec 2004.
9. A. Bösmann, A. Datsevich, A. Jess, A. Lauter, C. Schmitz and P. Wasserscheid, Chem. Commun., 2001, 2494.
10. J. Eßer, P. Wasserscheid and A. Jess, Green Chem., 2004, 6, 316.
11. S. Zhang, Q. Zhang and Z. C. Zhang, Ind. Eng. Chem. Res., 2004, 43, 614.
12. Y. Nie, C. Li, A. Sun, H. Meng and Z. Wang, Energy Fuels, 2006, 20, 2083.
13. P. Wasserscheid and W. Keim, Angew. Chem., 2000, 112, 3926.
14. T. Welton, Chem. Rev., 1999, 99, 2071; J. D. Holbrey and K. R. Seddon, Clean Prod. Proc., 1999, 1, 223.
15. J. L. Anderson, J. Ding, T. Welton and D. W. Armstrong, J. Am. Chem. Soc., 2002, 124, 14247.
16. L. A. Blanchard and J. Brennecke, J. Ind. Eng. Chem. Res., 2001, 40, 287.
17. J. D. Holbrey, W. M. Reichert, M. Nieuwenhuyzen, O. Sheppard, C. Hardacre and R. D. Rogers, Chem. Commun., 2003, 476.
18. A. R. Choudhury, N. Winterton, A. Steiner and K. A. Johnson, CrystEngComm, 2006, 8, 742.
19. S. Tsuzuki, M. Mikami and S. Yamada, J. Am. Chem. Soc., 2007, 129, 8656.
20. G. W. Meindersma, A. J. G. Podt and A. B. de Haan, Fuel Proc. Technol., 2005, 87, 59.
21. A. Arce, M. J. Earle, H. Rodríguez and K. R. Seddon, Green Chem., 2007, 9, 70.
22. W.-H. Lo, H.-Y. Yang and G.-T. Wei, Green Chem., 2003, 5, 639.
23. Y. Horii, H. Onuki, S. Doi, T. Mori, T. Takeo, H. Sato, T. Ookuro and T. Sugawara, US Pat., 5,494,572, February 27, 1996.
24. A. R. Katritzky, A. Lomaka, R. Petrukhin, R. Jain, M. Karelson, A. E. Visser and R. D. Rogers, J. Chem. Inf. Comput. Sci., 2002, 42, 71.
25. A. R. Katritzky, R. Jain, A. Lomaka, R. Petrukhin, M. Karelson, A. E. Visser and R. D. Rogers, J. Chem. Inf. Comput. Sci., 2002, 42, 225.
26. D. M. Eike, J. F. Brennecke and E. J. Maginn, Green Chem., 2003, 5, 323.
27. S. Trohalaki, R. Pachter, G. W. Drake and T. Hawkins, Energy Fuels, 2005, 19, 279.
28. S. Trohalaki and R. Pachter, QSAR Comb. Sci., 2005, 24, 485.
29. N. Sun, X. He, K. Dong, X. Zhang, X. Lu, H. He and S. Zhang, Fluid Phase Equilib., 2006, 246, 137.
30. I. López-Martin, E. Burello, P. N. Davey, K. R. Seddon and G. Rothenberg, ChemPhysChem, 2007, 8, 690.
31. A. R. Katritzky, V. S. Lobanov and M. Karelson, Chem. Soc. Rev., 1995, 279.
32. A. R. Katritzky, V. S. Lobanov and M. Karelson, CODESSA: Reference Manual; Version 2, University of Florida, 1994.
33. J. G. Topliss and R. P. Edwards, J. Med. Chem., 1979, 22, 1238.
34. A. E. Visser, J. D. Holbrey and R. D. Rogers, Chem. Commun., 2001, 2484.
35. D. F. Wassell, PhD thesis, The Queen's University of Belfast., Sept 2003; D. F. Wassell and K. R. Seddon, Abstracts of Papers, 223rd ACS National Meeting, Orlando, FL, United States, April 7–11, 2002, 2002, IEC-147.
36. J. D. Holbrey, W. M. Reichert, R. P. Swatloski, G. A. Broker, W. R. Pitner, K. R. Seddon and R. D. Rogers, Green Chem., 2002, 4, 407.

---

Footnotes

† Electronic supplementary information (ESI) available: QSPR analysis of solvent group contributions to extraction. See DOI: 10.1039/b710651c

‡ The variation in partition ratio (K_D) quoted over the range 1–3 correspond to only small changes in actual percentages of DBT extracted, for example, the differences between K_D = 1.6 and K_D = 1.9, for DBT extraction with the ionic liquid [C₄mim][C₈H₁₇SO₄] here, and in the literature, represent percentage extracted of 62 and 66%, overall—only 4% variation, which is probably within the experimental errors between different methods. Similarly, this experimental variation is exemplified by the differences between results published in references 9 and 10 by the same researchers.

§ Additional ionic liquids incorporated in the QSPR data analysis, with experimentally determined DBT distribution ratio and percentage extracted: (a) 1-(methoxyethyl)-3-methylimidazolium bis{(trifluoromethyl)sulfonyl}amide ([CH₃OC₂H₄mim][NTf₂]), K_D = 0.6, 28%; (b) N-methyl-N-butyl-N,N-di(2-hydroxyethyl)ammonium bis{(trifluoromethyl)sulfonyl}amide ([NMeBu(CH₂CH₂OH)₂][NTf₂]) K_D = 0.26, 21%; (c) 1-(2-hydroxyethyl)-1-methylmorpholinium bis{(trifluoromethyl)sulfonyl}amide ([HOCH₂CH₂mmor][NTf₂]) K_D = 0.7, 41%; (d) 1-butyl-4-cyanopyridinium bis{(trifluoromethyl)sulfonyl}amide ([C₄⁴CNpy][NTf₂]) K_D = 1.6, 61%; (e) 1-hexyl-4-cyanopyridinium bis{(trifluoromethyl)sulfonyl}amide ([C₆⁴CNpy][NTf₂]) K_D = 2.0, 66%.

---