High CO₂ solubility, permeability and selectivity in ionic liquids with the tetracyanoborate anion

Abstract

Five different ionic liquids containing the tetracyanoborate anion were synthesized and evaluated for CO₂ separation performance. Measured CO₂ solubility values were exceptionally high compared to analogous ionic liquids with different anions and ranged from 0.128 mol L⁻¹ atm⁻¹ to 0.148 mol L⁻¹ atm⁻¹. In addition, CO₂ permeability and CO₂/N₂ selectivity values were measured using a supported ionic liquid membrane architecture and the separations performance of the ionic liquid membranes exceeded the Robeson upper bound. These results establish the distinct potential of ionic liquids with the tetracyanoborate, [B(CN)₄], anion for the separation of CO₂.

---

Introduction

The selective separation of specific gases from a complex mixture has long been a critical component of many industrial and energy generation processes. Though found in a wide range of applications, the evolution of separations technology has largely been driven by needs and advances in the petroleum and petrochemical industries, such as in the removal of acid gases including CO₂ and H₂S from natural gas streams. More recently, the elimination of CO₂ from postcombustion flue gas emitted by large scale energy generating power plants represents a significant separations challenge with significant potential consequences.

Traditionally, liquid absorbents such as aqueous amines have been used to eliminate CO₂ from gas streams, either through chemical or physical absorption. However, issues such as solvent loss, degradation and high regeneration energy have led to increased interest in advanced materials and alternative separations methods to enhance or replace these traditional processes. Membranes are part of this growing segment of research because they are reliable, require a smaller footprint, and use less energy for the separation. Room-temperature ionic liquids (RTILs), which are popularly defined as a class of materials that are liquid at or below room temperature and exist completely as ions, have received attention recently as both a liquid absorbent and as a membrane material for gas separations.^1–3

Driven by several key properties such as high thermal and chemical stability, a wide electrochemical potential, high ionic conductivity, and negligible vapor pressure, hundreds of ionic liquids have been synthesized and reported, along with associated benefits and potential application.⁴ Perhaps one of the most important properties of ionic liquids is the ability to tailor many of their physical and chemical properties by combining different cation/anion pairs or by adding functional groups to either the cation or anion. This synthetic variability opens up considerable opportunities to create customized solvents for a particular application or task. For gas separations applications, the negligible vapour pressure and the ability to tailor the interaction of CO₂ with the ionic liquid are the most significant properties.

Supported ionic liquid membranes (SILMs) offer a unique separations platform because they combine the advantages of membranes with the ability to tailor the ionic liquid for optimum performance. In principle, a SILM is based on the selective absorption of CO₂ into the ionic liquid followed by CO₂ transport and desorption.⁵ Based on the solution-diffusion mechanism, improvements in the performance of a SILM are dependent on balancing the CO₂ solubility in the ionic liquid with fast diffusion and desorption. While significant improvements in the chemical sorption of CO₂ in ionic liquids have been reported,^6,7 it is critical to select ionic liquids that have high physisorption of CO₂ rather than high chemisorption as well as fast diffusion.

Though much of the work reported for ionic liquids in CO₂ separation has focused on the imidazolium-based ionic liquids, a number of cations including pyrrolidinium, pyridinium, and phosphonium have been explored in an effort to enhance CO₂ solubility and CO₂/N₂ selectivity.⁸ In addition to the cation, a variety of anions such as [PF₆], [BF₄], [N(CN)₂], and [Tf₂N] have been considered as a means to improve performance and advance our understanding of gas transport in ionic liquids.^9–11 Ionic liquids that contain the [B(CN)₄] anion are a set of liquids that have been somewhat underexplored despite their unique physicochemical properties. Interest in these [B(CN)₄] RTILs has generally been limited to use as electrolytes and in dye-sensitized solar cells.^12,13 Recently, however, we reported a high CO₂ solubility and CO₂/N₂ selectivity in an imidazolium-based ionic liquid with a [B(CN)₄] anion.¹⁴

Since this first report of the high CO₂ solubility, permeability and CO₂/N₂ selectivity of the [B(CN)₄] anion, interest in this RTIL for gas separations has grown; quantum mechanical calculations and molecular dynamics simulations revealed a weak interaction between the cation and the [B(CN)₄] anion responsible for the high CO₂ solubility supporting the potential of this RTIL for CO₂ separations.¹⁵ In this work, we build on these prior results and explore the separations properties of a series of [B(CN)₄]-containing RTILs coupled to a variety of cations including imidazolium, pyrrolidinium, and piperidinium. Effects of alkyl chain length and methyl substitution as well as ring structure on the CO₂ solubility, permeability and CO₂/N₂ selectivity were examined. As a class, these RTILs show high CO₂ solubility as well as excellent permeability and selectivity values that exceed the Robeson upper bound. Finally, the high CO₂ solubility exhibited by these RTILs is examined within the current mechanistic views of gas solubility in ionic liquids.

Results and discussion

Physical properties

The structures of the imidazolium, pyrrolidinium, and piperidinium cations and the [B(CN)₄] anion used in this work are shown in Fig. 1. For the imidazolium RTILs, two different alkyl chain lengths were studied and the addition of a methyl group on the C2 hydrogen of the ring was also investigated. The structure of the ionic liquids was verified using NMR spectroscopy (see Supplementary Information). Decomposition temperatures, which were measured by thermogravimetric analysis (TGA), are given in Table 1 and ranged from 361 °C for [C₄MPyrr] to 396 °C for [emmim]. These decomposition temperatures are slightly lower than but comparable to those observed in fluorinated anions such as [Tf₂N].^16–18 In contrast to [Tf₂N]-based ionic liquids, both of the cyclical cations, [C₄MPyrr] and [C₄mpip], with the [B(CN)₄] anion displayed lower thermal stabilities than the imidazolium-based RTILs.

Fig. 1 The structures of the five cations and the [B(CN)₄] anion. The abbreviations are as follows: a) [emim], b) [bmim], c) [emmim], d) [C₄MPyrr], e) [C₄mpip], f) [B(CN)₄].

Table 1 Physical properties of the [B(CN)₄]-containing ionic liquids

Ionic Liquid	Molar Volume^a (cm^3 mol^−1)	Molecular Weight (g mol^−1)	Viscosity^b (cP)	Ea (kJ mol^−1)	Tdec (°C)
a Obtained for 25 °C. b Acquired at 20 °C
[emim][B(CN)4]	219.46	226.04	19.5	23.3	396
[bmim][B(CN)4]	245.04	254.11	44.9	32.6	393
[emmim][B(CN)4]	226.91	240.07	36.0	29.0	396
[C4MPyrr][B(CN)4]	262.66	257.14	58.9	30.1	361
[C4mpip][B(CN)4]	248.72	266.13	111	36.5	373

---

The viscosity of RTILs, which are generally high compared to typical organic solvents, is a critical parameter from a practical sense because a high viscosity will tend to reduce the diffusion of gas through the ionic liquid which can ultimately reduce the performance of the membrane. Many of the models reported to rationalize the diffusivity of gases through RTILs show an inverse relationship between viscosity and diffusivity.^19,20 Clearly, for high permeability it is important to use ionic liquids with low viscosity.

The temperature dependent viscosities are shown in Supplementary Information (Table S1†). As expected, the viscosity decreased nonlinearly as the temperature increased. Table 1 shows that the [emim][B(CN)₄] exhibited the lowest viscosity of all the RTILs tested while [C₄mpip][B(CN)₄] showed the highest viscosity. Similar to previously reported results, lengthening the alkyl chain on the cation, e.g., going from ethyl to n-butyl, increased the viscosity.²¹ The addition of a methyl group on the C2 position of the imidazolium ring also resulted in an increased viscosity in agreement with results previously reported by Noack et al.²² In addition, Crosthwaite et al. measured the viscosities and saw nearly a factor of two increase when the methyl group was added to the C2 position for [hmmim][Tf₂N] compared to [hmim][Tf₂N].²¹ In general, use of the [B(CN)₄] anion drastically reduced the viscosity compared to [Tf₂N] congeners. For example, [bmim][B(CN)₄] has a viscosity of 44.4 cP which is nearly 12% lower than the reported [Tf₂N] congener²³ while [emmim][B(CN)₄] has a viscosity of 36.0 cP which is 244% lower than the [Tf₂N]-based RTIL.²⁴

Fig. 2 presents the Arrhenius plot for the five ionic liquids. The curves are highly linear confirming that these RTILs follow Arrhenius behaviour. The activation energies for viscous flow given in Table 1 are comparable to literature values.¹⁹ These values are included because of applicability in calculating solubility parameters.

Fig. 2 Arrhenius plot of the [B(CN)₄]-containing ionic liquids. Solid lines are the best fit lines used to calculate the activation energies.

CO₂ permeability and CO₂/N₂ selectivity

Gas transport in a SILM follows the typical solution/diffusion mechanism where permeability is related to solubility and diffusivity via the following equation:

P = S D (1)

To maximize gas permeability, or membrane productivity, it is important to use materials with both high solubility as well as high diffusivity. The ideal selectivity is then given by ratio of the pure gas permeabilities:

α_ij = P_i/P_j = (S_i/S_j)·(D_i/D_j) (2)

Given the comparatively low viscosities of the [B(CN)₄] RTILs, the permeabilities were generally expected to be high. Table 2 shows the CO₂ permeability values and the CO₂/N₂ selectivity values for the [B(CN)₄] ionic liquids. Note that the [emim][B(CN)₄] has been previously reported.¹⁴ Traditionally, [emim][Tf₂N], which has a permeability of 1702 barrer, has been used as a benchmark for high permeability ionic liquids since it exhibits one of the highest CO₂ permeability values.⁸ With the exception of [C₄mpip], all of the [B(CN)₄] RTILs investigated show permeability values that are comparable to or exceed [emim][Tf₂N]. The lowest viscosity RTIL, [emim][B(CN)₄], showed the highest permeability at 2040 barrer while the highest viscosity RTIL showed the lowest permeability. In general, the permeability shows an inverse relationship with the viscosity which is in agreement with previously reported trends in permeability and viscosity.⁸ The CO₂/N₂ selectivity values for the [B(CN)₄]-based RTILs are also shown in Table 2 along with permeability and selectivity values for [emim][BF₄] and [emim][TfO] for additional comparison. In general, the selectivity values of the [B(CN)₄] RTILs are quite high, with values ranging from 38 to 53. Interestingly, the N₂ permeability in [emim][B(CN)₄] is lower than in [emim][Tf₂N] and this is at least partially responsible for the high selectivity. A similar effect has been reported by Carlisle for the addition of a nitrile group to an imidazolium cation with the [Tf₂N] anion. In that case, the N₂ solubility was reduced compared to the unfunctionalized analog leading to higher CO₂/N₂ selectivity.²⁵

Table 2 CO₂ permeabilities and CO₂/N₂ selectivities

Ionic Liquid	Permeability (barrer)		CO2/N2 Selectivity
	CO2	N2
a Taken from [Ref. 8].
[emim][B(CN)4]	2040 ± 60	38 ± 4	53 ± 6
[bmim] [B(CN)4]	1755 ± 50	44 ± 3	40 ± 2
[emmim] [B(CN)4]	1721 ± 80	38 ± 4	46 ± 5
[C4MPyrr] [B(CN)4]	1633 ± 60	44 ± 3	38 ± 2
[C4mpip][B(CN)4]	961 ± 25	26 ± 4	37 ± 5
[emim][Tf2N]^a	1702	73.6	23
[emim][BF4]^a	968.5	21.8	44.5
[emim] [TfO]^a	1171.4	28.9	40.5

---

As previously stated, a number of reports have been published describing the relationship between gas diffusion through ionic liquids and viscosity of the liquid. Though the details of the different empirical models as well as the properties included vary, there is general agreement that diffusivity decreases as viscosity increases. As will be discussed in more detail in the next section, the CO₂ solubility values for the [B(CN)₄] RTILs are high but fairly similar. Differences in gas diffusivity are likely responsible for the relatively large variations in the CO₂ permeabilities measured for these RTILs, particularly as the viscosities of the RTILs vary from 19.5 cP to 111 cP.

The use of the selectivity vs. permeability curve, generally referred to as the Robeson plot, has become prevalent in gas separations literature as a method to evaluate membrane materials. This plot, which was first reported as an empirical upper bound for polymeric membranes but was subsequently placed on a firmer theoretical foundation,^26,27 demonstrates the compromise that typically exists between high permeability and high selectivity. Most membranes fall below the upper bound line for gas pairs such as CO₂/N₂, which has provided significant motivation to discover membrane materials that exceed this limit. Fig. 3 shows the Robeson plot with the most recent upper bound line.²⁸ The [B(CN)₄] RTILs all exceed the Robeson upper bound line which illustrates the remarkable potential of these ionic liquids for CO₂ separation.

Fig. 3 Robeson plot showing the [B(CN)₄] anion RTILs in comparison to the upper bound line. The open circles are polymers, filled squares correspond to ionic liquids, and the open triangles are the [B(CN)₄] anions in this work. The line is the upper bound.

Fig. 4 shows the permeability as a function of driving pressure for the [B(CN)₄] RTILs. Clearly, there is no dependence of the permeability on the driving pressure which indicates that there is no facilitated transport within the ionic liquids. This confirms that the CO₂ is physically absorbed into the ionic liquid and there is no chemisorption of the CO₂ molecule to either the cation or the [B(CN)₄] anion.

Fig. 4 The effect of pressure difference across the membrane on the measured permeability. The lack of pressure dependence suggests non-facilitated gas transport through the ionic liquids.

CO₂ solubility

Given the impact of CO₂ solubility on gas separations performance, we measured solubility isotherms for the [B(CN)₄] RTILS (see Fig. 5) where the ordinate is given in terms of volumetric solubility. The absorption isotherms are highly linear and the slopes appear to be fairly comparable. Quantitatively, the slope of each isotherm was obtained using a linear least squares fit and these values are presented in Table 3. We use volume-based solubility because practical applications of gas capture technologies require knowledge of the amount of CO₂ that can be absorbed by a fixed volume of ionic liquid. However CO₂ solubility values are often reported on a mole fraction basis in the form of Henry's Law constants according to the following:

where x is the mole fraction of CO₂ absorbed in the ionic liquid and p is the pressure. To facilitate comparison with literature values, Henry's Law constants are also included in Table 3.

Fig. 5 CO₂ absorption isotherms for the [B(CN)₄]-containing ionic liquids acquired at 25 °C.

Table 3 CO₂ solubilities and Henry's Law constants for the [B(CN)₄]-containing ionic liquids

Ionic Liquid	CO2 Solubility^a (mol L^−1 atm^−1)	HCO2 (atm)	COSMO-RS Prediction		FFV
			CO2 Soubility	HCO2
a Acquired at 25 °C with ±3% uncertainty
[emim][B(CN)4]	0.128	38.9	0.19	25	0.1323
[bmim] [B(CN)4]	0.132	32.3	0.21	20	0.1451
[emmim] [B(CN)4]	0.148	31.9	0.31	17	0.1390
[C4MPyrr] [B(CN)4]	0.130	30.4	0.26	15	0.1520
[C4mpip] [B(CN)4]	0.128	30.2	0.26	15	0.1480

---

In general, all of the [B(CN)₄] RTILs exhibited exceptional CO₂ solubility values that were larger than the prototypical ionic liquid, [emim][Tf₂N], which has a CO₂ solubility of 0.103 mol L¹ atm⁻¹.²⁹ The [emmim][B(CN)₄] RTIL had the highest solubility value at 0.148 mol L⁻¹ atm⁻¹. Interestingly, the lowest CO₂ solubility was measured for [emim][B(CN)₄], which also had the highest CO₂ permeability. This is primarily due to a higher CO₂ diffusivity in the [emim][B(CN)₄]. The Henry's Law constants show a similar trend as the volumetric solubility though there are some differences. For example, [emmim] showed a higher volumetric solubility than [C₄MPyrr] but a higher Henry's constant (higher Henry's constants generally equate to lower solubility). These differences are relatively small and can be attributed to the varying molar weights of the different ionic liquids.³⁰ In fact, the volumetric solubilities for the [emim], [C₄MPyrr], and [C₄mpip] are very similar.

The three imidazolium RTILs are of particular interest because they allow us to explore the impact of alkyl chain length and methyl substitution on CO₂ solubility. Increasing the alkyl chain length from [emim] to [bmim] resulted in a slight increase in CO₂ solubility. Furthermore, replacing the C2 hydrogen with a methyl group on the imidazolium ring also improved the CO₂ solubility, with the value increasing from 0.128 to 0.148 mol L⁻¹ atm⁻¹. This differs from previous results for [emim][PF₆] and [emim][Tf₂N] where replacing the C2 hydrogen with a methyl group slightly decreased the CO₂ solubility.³¹

To further probe these results, the COSMOtherm package, which has gained popularity as a method to rapidly estimate properties such as density, free volume and CO₂ solubility,³² was used to calculate the CO₂ solubility values for each of the RTILs and the results are shown in Table 3. Optimized molecular geometries for each RTIL were used to calculate molecular surfaces (or COSMO surfaces) and thermodynamic properties.

CO₂ solubility values predicted using the COSMOtherm package are in very good agreement with the measured values. For example, COSMOtherm predicted that substituting the methyl group at the C2 position would increase the CO₂ solubility which is consistent with the measured results and validates the effect of the methyl group for the [B(CN)₄] anion. Furthermore, though the predicted Henry's constants are not identical to the measured values, the trend is in excellent agreement, [emim] > [bmim] > [emmim] > [C₄MPyrr] > [C₄mpip]. In terms of volumetric solubility, COSMOtherm predicted the largest CO₂ solubility for [emmim][B(CN)₄] which is also in agreement with measured results. Increasing the alkyl chain length from [emim] to [bmim] resulted in a slight increase in both the measured and calculated solubility values in contrast to the alkylimidazolium [Tf₂N] RTILs which show decreasing solubility with alkyl chain length.^33,34

The high CO₂ solubilities for the [B(CN)₄] illustrate the unique nature of this anion and reaffirm the ongoing debate over the mechanism governing gas solubility in ionic liquids. While there is general agreement that CO₂ solubility in ionic liquids is affected by cation-anion, CO₂-anion, and CO₂–CO₂ interactions, there is some disagreement on the relative importance and magnitude of each interaction.^11,35,36 In a recent review, Hu et al. suggested that the combination of two effects, i.e., CO₂ insertion into void space (or free volume) in ionic domains of the ionic liquid followed by CO₂-ion interactions, is responsible for CO₂ solubility in ionic liquids.³⁷ Increasing the size, flexibility or asymmetry of the anion can lead to increased free volume in the ionic regions and enhanced CO₂ solubility.

Despite the high symmetry of the [B(CN)₄] anion which might suggest low free volume in the ionic domain, free volume can be increased if the ionic liquid can rearrange and accommodate the absorption of CO₂. One explanation then for the high CO₂ solubility in these RTILs could be the ability of the ionic liquid to rearrange due to reduced interaction between the cations and the [B(CN)₄] anion compared to other anions such as [PF₆] or [Tf₂N]. This is supported by a recent report by Babarao et al. who showed that there is a weaker cation-anion interaction for [emim][B(CN)₄] than for [emim][BF₄], [emim][PF₆], or [emim][Tf₂N],¹⁵ which could lead to an enhanced ability to rearrange and greater CO₂-anion interaction and higher solubility.

Bara et al. recently noted the importance of fractional free volume in estimating CO₂ solubility in RTILs.³⁸ The fractional free volume (FFV) was calculated using the COSMOtherm program according to the following equation:

where V_m is the molar volume and V_COSMO is the volume calculated within the COSMOtherm program. Examination of Table 3 shows that though there is no direct correlation between the FFV and the solubility within the [B(CN)₄] set of ionic liquids, the trend seems to be that the Henry's Law constants decrease as the FFV increases (see Fig. 6).

Fig. 6 Henry's Law constants as a function of the fractional free volume (FFV).

An interesting comparison is given in Fig. 7 which shows the solubility as a function of molar volume along with the Camper Model for CO₂ solubility in RTILs.³³ Though the Camper Model was developed to estimate the solubility for imidazolium-based ionic liquids, it is instructive to show as a comparison for the [B(CN)₄] ionic liquids. The Camper Model relates the solubility of CO₂ in an RTIL to the molar volume only. Here we see that the CO₂ solubility in the [B(CN)₄] ionic liquids significantly exceed the molar volume dependence, again indicating that additional factors play a role in determining gas solubility.

Fig. 7 Comparison of the solubility of the [B(CN)₄]-containing ionic liquids to the Camper Model.

Conclusions

CO₂ permeability, CO₂/N₂ selectivity and CO₂ solubility values were measured for ionic liquids that contain the [B(CN)₄] anion and various cations. CO₂ permeability and selectivity values exceeded the Robeson upper bound line for all of the [B(CN)₄] ionic liquids, suggesting their significant potential as supported ionic liquid membranes. In addition, all of the ionic liquids exhibited high CO₂ solubilities ranging from 0.128 mol L⁻¹ atm⁻¹ to 0.148 mol L⁻¹ atm⁻¹. Based on the current views of gas solubility in ionic liquids, we postulate that the high CO₂ solubilities result from the combination of reduced cation-anion interactions and enhanced CO₂-anion interactions. These results firmly establish [B(CN)₄] as a unique anion in offering exceptional separations performance for CO₂ when combined with various types of cations to form [B(CN)₄]-containing ionic liquids.

Experimental

Gas solubility

Gas solubility measurements were obtained using a gravimetric microbalance (Hiden Isochema, IGA). Approximately 60 mg of the RTIL was loaded into the sample container and sealed in the stainless steel chamber. The RTIL was dried and degassed at a temperature of 80 °C and a vacuum pressure of 1 mbar for a minimum of 4 h before recording the dry mass. Mass measurements were then acquired at various CO₂ pressures up to 10 atm. The temperature of the sample was maintained using a constant-temperature recirculating water bath.

Viscosity

Temperature dependent viscosity measurements were acquired using a cone/plate viscometer (Brookfield, DVII + Pro), which allowed the use of small sample volumes of < 1 mL. The sample was kept at the desired temperature with a recirculating water bath (VWR) for at least 30 min prior to measurement and five separate measurements at each temperature were obtained and averaged.

Permeability

The RTIL was loaded into a polyethersulfone membrane with 100 nm pore size by immersing the membrane in the liquid and placing it in a vacuum desiccator overnight. Permeabilites were measured using a constant-volume system consisting of two chambers (permeate and retentate) separated by the membrane. The membrane was positioned on a porous stainless steel support for mechanical stability and loaded into the membrane test chamber where it was evacuated to a pressure of 25 mTorr overnight before measurements were acquired in order to equilibrate the membrane. The retentate side was pressurized with each gas and the pressure rise in the permeate chamber was measured with a Baratron pressure gauge (MKS Instruments) as a function of time to obtain the permeability using the following equationwhere d is the membrane thickness, τ is the tortuosity, ϕ is the porosity, and the permeance is determined by the following equationwhere V is the permeate volume, R is the ideal gas constant, T is the absolute temperature, A is the membrane area, ΔP is the pressure difference across the membrane, and dP/dt is the pressure rise on the permeate side. All permeability values were the result of single gas measurements. The thickness of the RTIL was assumed to be equivalent to the thickness of the membrane.

Calculations

The COSMOtherm program was used to predict gas solubility and Henry's Law constants in the ionic liquids. The COSMOtherm program³⁹ is based on the COSMO-RS (COnductor-like Screening MOdel for Real Solvents)⁴⁰ theory which links the quantum mechanically calculated, solvated molecular surfaces from the COSMO method to interacting molecular surfaces and macroscopic properties such as solubility via a statistical thermodynamics model. Full COSMO optimization of the molecular structure was obtained with Turbomole V 6.0 at the level of density functional theory (DFT) with the BP86 functional for electron correlation and exchange⁴¹ and using the resolution-of-identity (RI) technique and the TZVP basis set.

Synthesis of ionic liquids

Sodium and potassium tetracyanoborate were purchased from SelectLab Chemicals. All other chemicals were purchased from Sigma-Aldrich in the highest available purity and used as received. Starting bromide salts, namely 3-ethyl-1,2-dimethyl-1H-imidazol-3-ium bromide,⁴² 3-butyl-1-methyl-1H-imidazol-3-ium bromide,⁴³ 1-butyl-1-methylpiperidin-1-ium bromide,⁴⁴ 1-butyl-1-methylpyrrolidin-1-ium bromide,⁴² were synthesized and purified according to established literature procedures. ¹H and ¹³C NMR spectra were recorded on a Bruker Avance 400, at 400 MHz and 100 MHz respectively. A JEOL (Peabody, MA) JMS-T100LC (AccuTOFTM) orthogonal time-of-flight (TOF) mass spectrometer was used to characterize the compounds. Thermal stabilities were measured using a Thermal Advantage 2950 Analyzer with platinum pans under a nitrogen atmosphere (10 °C min⁻¹ heating to 750 °C). More detailed synthesis procedures are included in the Supplementary Information.†

Acknowledgements

S.M.M, J.S.Y., D.J., and S.D. were sponsored by the Division of Chemical Sciences, Geosciences, and Biosciences, Office of Basic Energy Sciences, U.S. Department of Energy. P.C.H. was supported by the Advanced Research Projects Agency – Energy, U.S. Department of Energy.

References

1. J. E. Brennecke and B. E. Gurkan, J. Phys. Chem. Lett., 2010, 1, 3459–3464.
2. C. M. Wang, S. M. Mahurin, H. M. Luo, G. A. Baker, H. R. Li and S. Dai, Green Chem., 2010, 12, 870–874.
3. R. D. Noble and D. L. Gin, J. Membr. Sci., 2011, 369, 1–4.
4. T. Welton, Chem. Rev., 1999, 99, 2071–2083.
5. Q. Gan, Y. R. Zou, D. Rooney, P. Nancarrow, J. Thompson, L. Z. Liang and M. Lewis, Adv. Colloid Interface Sci., 2011, 164, 45–55.
6. E. D. Bates, R. D. Mayton, I. Ntai and J. H. Davis, J. Am. Chem. Soc., 2002, 124, 926–927.
7. C. M. Wang, X. Y. Luo, H. M. Luo, D. E. Jiang, H. R. Li and S. Dai, Angew. Chem., Int. Ed., 2011, 50, 4918–4922.
8. P. Scovazzo, J. Membr. Sci., 2009, 343, 199–211.
9. P. Scovazzo, D. Havard, M. McShea, S. Mixon and D. Morgan, J. Membr. Sci., 2009, 327, 41–48.
10. M. B. Shiflett, D. W. Drew, R. A. Cantini and A. Yokozeki, Energy Fuels, 2010, 24, 5781–5789.
11. J. L. Anthony, J. L. Anderson, E. J. Maginn and J. F. Brennecke, J. Phys. Chem. B, 2005, 109, 6366–6374.
12. M. Marszalek, Z. F. Fei, D. R. Zhu, R. Scopelliti, P. J. Dyson, S. M. Zakeeruddin and M. Gratzel, Inorg. Chem., 2011, 50, 11561–11567.
13. D. B. Kuang, P. Wang, S. Ito, S. M. Zakeeruddin and M. Gratzel, J. Am. Chem. Soc., 2006, 128, 7732–7733.
14. S. M. Mahurin, J. S. Lee, G. A. Baker, H. M. Luo and S. Dai, J. Membr. Sci., 2010, 353, 177–183.
15. R. Babarao, S. Dai and D. E. Jiang, J. Phys. Chem. B, 2011, 115, 9789–9794.
16. C. P. Fredlake, J. M. Crosthwaite, D. G. Hert, S. Aki and J. F. Brennecke, J. Chem. Eng. Data, 2004, 49, 954–964.
17. J. G. Huddleston, A. E. Visser, W. M. Reichert, H. D. Willauer, G. A. Broker and R. D. Rogers, Green Chem., 2001, 3, 156–164.
18. H. Tokuda, K. Hayamizu, K. Ishii, M. Susan and M. Watanabe, J. Phys. Chem. B, 2005, 109, 6103–6110.
19. S. S. Moganty and R. E. Baltus, Ind. Eng. Chem. Res., 2010, 49, 5846–5853.
20. D. Morgan, L. Ferguson and P. Scovazzo, Ind. Eng. Chem. Res., 2005, 44, 4815–4823.
21. J. M. Crosthwaite, M. J. Muldoon, J. K. Dixon, J. L. Anderson and J. F. Brennecke, J. Chem. Thermodyn., 2005, 37, 559–568.
22. K. Noack, P. S. Schulz, N. Paape, J. Kiefer, P. Wasserscheid and A. Leipertz, Phys. Chem. Chem. Phys., 2010, 12, 14153–14161.
23. J. N. C. Lopes, M. F. C. Gomes, P. Husson, A. A. H. Padua, L. P. N. Rebelo, S. Sarraute and M. Tariq, J. Phys. Chem. B, 2011, 115, 6088–6099.
24. K. Fumino, A. Wulf and R. Ludwig, Angew. Chem., Int. Ed., 2008, 47, 8731–8734.
25. T. K. Carlisle, J. E. Bara, C. J. Gabriel, R. D. Noble and D. L. Gin, Ind. Eng. Chem. Res., 2008, 47, 7005–7012.
26. L. M. Robeson, J. Membr. Sci., 1991, 62, 165–185.
27. B. D. Freeman, Macromolecules, 1999, 32, 375–380.
28. L. M. Robeson, J. Membr. Sci., 2008, 320, 390–400.
29. S. M. Mahurin, J. S. Yeary, S. N. Baker, D. E. Jiang, S. Dai and G. A. Baker, J. Membr. Sci., 2012, 401–402, 61–67.
30. J. E. Bara, T. K. Carlisle, C. J. Gabriel, D. Camper, A. Finotello, D. L. Gin and R. D. Noble, Ind. Eng. Chem. Res., 2009, 48, 2739–2751.
31. C. Cadena, J. L. Anthony, J. K. Shah, T. I. Morrow, J. F. Brennecke and E. J. Maginn, J. Am. Chem. Soc., 2004, 126, 5300–5308.
32. J. Palomar, M. Gonzalez-Miquel, A. Polo and F. Rodriguez, Ind. Eng. Chem. Res., 2011, 50, 3452–3463.
33. D. Camper, J. Bara, C. Koval and R. Noble, Ind. Eng. Chem. Res., 2006, 45, 6279–6283.
34. P. K. Kilaru, P. A. Condemarin and P. Scovazzo, Ind. Eng. Chem. Res., 2008, 47, 900–909.
35. P. J. Carvalho and J. A. P. Coutinho, J. Phys. Chem. Lett., 2010, 1, 774–780.
36. T. Seki, J. D. Grunwaldt and A. Baiker, J. Phys. Chem. B, 2009, 113, 114–122.
37. Y. F. Hu, Z. C. Liu, C. M. Xu and X. M. Zhang, Chem. Soc. Rev., 2011, 40, 3802–3823.
38. M. S. Shannon, J. M. Tedstone, S. P. O. Danielsen, M. S. Hindman, A. C. Irvin and J. E. Bara, Ind. Eng. Chem. Res., 2012, 51, 5565–5576.
39. F. Eckert and A. Klamt, COSMOlogic GmbH & Co. KG, Leverkusen, Germany, 2011.
40. F. Eckert and A. Klamt, AIChE J., 2002, 48, 369–385.
41. R. Ahlrichs, F. Furche and S. Grimme, Chem. Phys. Lett., 2000, 325, 317–321.
42. X. W. Chen, X. H. Li, H. B. Song, Y. Qian and F. R. Wang, Tetrahedron Lett., 2011, 52, 3588–3591.
43. A. K. Burrell, R. E. Del Sesto, S. N. Baker, T. M. McClesky and G. A. Baker, Green Chem., 2007, 9, 809–809.
44. Z. B. Zhou, H. Matsumoto and K. Tatsumi, Chem.–Eur. J., 2006, 12, 2196–2212.

---

Footnote

† Electronic Supplementary Information (ESI) available: temperature dependent viscosity values, detailed description of synthesis and NMR results are included. See DOI: 10.1039/c2ra22342b

---