Tunning CO2 Separation Performance of Ionic Liquids through Asymmetric Anions

This work aims to explore the gas permeation performance of two newly-designed ionic liquids, [C2mim][CF3BF3] and [C2mim][CF3SO2C(CN)2], in supported ionic liquid membranes (SILM) configuration, as another effort to provide an overall insight on the gas permeation performance of functionalized-ionic liquids with the [C2mim]+ cation. [C2mim][CF3BF3] and [C2mim][CF3SO2C(CN)2] single gas separation performance towards CO2, N2, and CH4 at T = 293 K and T = 308 K were measured using the time-lag method. Assessing the CO2 permeation results, [C2mim][CF3BF3] showed an undermined value of 710 Barrer at 293.15 K and 1 bar of feed pressure when compared to [C2mim][BF4], whereas for the [C2mim][CF3SO2C(CN)2] IL an unexpected CO2 permeability of 1095 Barrer was attained at the same experimental conditions, overcoming the results for the remaining ILs used for comparison. The prepared membranes exhibited diverse permselectivities, varying from 16.9 to 22.2 for CO2/CH4 and 37.0 to 44.4 for CO2/N2 gas pairs. The thermophysical properties of the [C2mim][CF3BF3] and [C2mim][CF3SO2C(CN)2] ILs were also determined in the range of T = 293.15 K up to T = 353.15 K at atmospheric pressure and compared with those for other ILs with the same cation and anion’s with similar chemical moieties.


Introduction
Although Ionic Liquids (ILs) were first introduced as a green alternative to conventional organic solvents, mainly due to their vanishing vapor pressure [1], it was their unusual set of properties, such as low volatility, high thermal stability, low flammability [2,3] together with the possibility to tune the ILs properties through the combination of endless number of cations and anions [4], that afforded their popularity. Today, it is well known that ILs are not intrinsically green, not only due to their fossil fuel source, but also their complex, expensive, and non-sustainable synthesis, purification processes, and their recalcitrant nature. Nevertheless, ILs can provide the implementation of more sustainable processes, being at the center of clever solutions to common problems.
Ionic liquid-based membranes have won the interest of the scientific community since they gather the benefits of membrane separation, together with the unique characteristics of ILs. Several studies show the feasibility of using these membranes at an industrial [FPFSI] [39], CF3SO2-N-SO(CF3)=N-SO2CF3 [40], R-BF3 (R = CnH2n+1, n = 1-5) [41], Rf-BF3 (Rf = CnF2n+1, n = 1-4) [42], and the synthesis of ILs on their basis. Taking this into account, in a recent work [43,44] [TFSAM] displays higher CO2 permeability, diffusivity, and solubility than SILMs with the selected ILs mixtures. Although, it should be mentioned that the CO2/N2 selectivity of both SILMs was similar and can be positioned on top of or slightly above the Robeson plot [45].
In this work, we step forward in the design of new asymmetrical anions (Scheme 1). Two anions, namely trifluoro(trifluoromethyl)borate [CF3BF3]− and dicyano((trifluoromethyl)sulfonyl)methanide [DCATfM]− or [CF3SO2C(CN)2]−, were prepared in the form of their alkali salts and used for the synthesis of [C2mim]+ containing Ils. While [DCATfM]-anion was synthesized for the first time, for the preparation of K[BF3CF3] salt, we propose an improved approach in comparison with literature procedure [46], which avoids the use of corrosive hydrogen fluoride and toxic hexamethyltin (Scheme 2  25.59 and 25.00 mPa.s at 298.15 K, respectively. The obtained viscosity values seem to be relatively low, since typically imidazolium-based ILs show dynamic viscosities in the range between 0.841 and 2.57 × 10 5 mPa.s, as shown in recent works on the predictive estimation of pure ILs viscosities [47]. They were further used in the preparation of new SILMs and study of their gas separation properties.

Synthesis of New ILs
Potassium (trifluoromethyl)trifluoroborate K[CF3BF3] is an air-and water-stable solid that was first reported in 1960 by Chambers, Clark, and Willis [48]. The synthesis was carried out in two steps (Scheme 2). On the first step, the trimethyltrifluoromethyltin reacted with boron trifluoride with the formation of a 1:1 adduct, that precipitated immediately from the tetrachloride solution. The second step consisted of the ion exchange with potassium fluoride in an aqueous medium, where the insoluble trimethyltin fluoride was precipitated and filtered off, while the evaporation of the

Synthesis of New ILs
Potassium (trifluoromethyl)trifluoroborate K[CF 3 BF 3 ] is an air-and water-stable solid that was first reported in 1960 by Chambers, Clark, and Willis [48]. The synthesis was carried out in two steps (Scheme 2). On the first step, the trimethyltrifluoromethyltin reacted with boron trifluoride with the formation of a 1:1 adduct, that precipitated immediately from the tetrachloride solution. The second step consisted of the ion exchange with potassium fluoride in an aqueous medium, where the insoluble trimethyltin fluoride was precipitated and filtered off, while the evaporation of the resulting solution provided potassium trifluoromethylfluoroborate. Despite the good purity of obtained K[CF 3 BF 3 ], the use of this method is limited by the high toxicity of the tin compounds. Later on, in 2003, Molander [46] suggested an improved method for the preparation of K[CF3BF3]. Instead of toxic trimethyltrifluoromethyltin, it was suggested to start the reaction with Ruppert's reagent, (trifluoromethyl)trimethylsilane, which was treated with trimethoxyborane in the presence of potassium fluoride (Scheme 3). Lately, the aqueous hydrogen fluoride was added to the resulting intermediate, and the title compound was isolated in 85% overall yield. Although, both the yield and purity of K[CF3BF3] were sufficiently high, this method suffered from the usage of corrosive hydrogen fluoride and thus, from the impossibility to carry out the reaction in glass reactors and the necessity for special vessels made of copper or Teflon. Scheme 3. Synthetic pathway for the preparation of K[CF3BF3] following Molander [46].
In this work, we introduced a new method for the synthesis of K[CF3BF3] (Scheme 4). As in the Molander approach [46], the first step consisted in the formation of so-called "ate" complex, although the subsequent fluorination with hydrogen fluoride was replaced with a substitution with BF3·Et2O, followed by the removal of B(OCH3)3 by distillation. Thus, the suggested approach differs from the previous ones [46,48]  Later on, in 2003, Molander [46] suggested an improved method for the preparation of K[CF 3 BF 3 ]. Instead of toxic trimethyltrifluoromethyltin, it was suggested to start the reaction with Ruppert's reagent, (trifluoromethyl)trimethylsilane, which was treated with trimethoxyborane in the presence of potassium fluoride (Scheme 3). Lately, the aqueous hydrogen fluoride was added to the resulting intermediate, and the title compound was isolated in 85% overall yield. Although, both the yield and purity of K[CF 3 BF 3 ] were sufficiently high, this method suffered from the usage of corrosive hydrogen fluoride and thus, from the impossibility to carry out the reaction in glass reactors and the necessity for special vessels made of copper or Teflon.
resulting solution provided potassium trifluoromethylfluoroborate. Despite the go purity of obtained K[CF3BF3], the use of this method is limited by the high toxicity of tin compounds. Later on, in 2003, Molander [46] suggested an improved method for the preparat of K[CF3BF3]. Instead of toxic trimethyltrifluoromethyltin, it was suggested to start reaction with Ruppert's reagent, (trifluoromethyl)trimethylsilane, which was treated w trimethoxyborane in the presence of potassium fluoride (Scheme 3). Lately, the aqueo hydrogen fluoride was added to the resulting intermediate, and the title compound w isolated in 85% overall yield. Although, both the yield and purity of K[CF3BF3] w sufficiently high, this method suffered from the usage of corrosive hydrogen fluoride a thus, from the impossibility to carry out the reaction in glass reactors and the necessity special vessels made of copper or Teflon. Scheme 3. Synthetic pathway for the preparation of K[CF3BF3] following Molander [46].
In this work, we introduced a new method for the synthesis of K[CF3BF3] (Scheme As in the Molander approach [46], the first step consisted in the formation of so-cal "ate" complex, although the subsequent fluorination with hydrogen fluoride w replaced with a substitution with BF3·Et2O, followed by the removal of B(OCH3)3 distillation. Thus, the suggested approach differs from the previous ones [46,48]  In this work, we introduced a new method for the synthesis of K[CF 3 BF 3 ] (Scheme 4). As in the Molander approach [46], the first step consisted in the formation of so-called "ate" complex, although the subsequent fluorination with hydrogen fluoride was replaced with a substitution with BF 3 ·Et 2 O, followed by the removal of B(OCH 3 ) 3 by distillation. Thus, the suggested approach differs from the previous ones [46,48] 2 ] showed a singlet at −81.5 ppm assigned to CF 3 group. Both CF 3 and CN groups were found in 13 C NMR as a quadruplet at 125. 6-115.9 and as singlet at 115.5 ppm, respectively. FTIR spectra of [C 2 mim][CF 3 SO 2 C(CN) 2 ] contained the following characteristic absorption bands: 3158, 3119 and 2990 cm −1 attributed to CH stretching; 1350 and 1181 cm −1 assigned to asymmetric and symmetric vibrations of S=O bond; and finally 1209 and 1071 cm −1 bands that were designated to the CF vibrations.  Figure 1. The values are listed in Table S1 in Supporting Information.

Termophysical Properties of ILs: Density, Molar Volume, and Viscosity
Density and viscosity are very important properties of ILs used in gas separation since they have a direct influence on gas permeation properties. Both density and viscosity for [C2mim][CF3BF3] and [C2mim][CF3SO2C(CN)2] were measured within the range of 293.15 K up to 353.15 K at ambient pressure and their behaviour with temperature is presented in Figure 1. The values are listed in Table S1 in Supporting Information.   Figure 1. The values are listed in Table S1 in Supporting Information.   [53] and 668.15K [54], it can be concluded that these new ILs show a loss thermal stability.

Termophysical Properties of ILs: Density, Molar Volume, and Viscosity
Density and viscosity are very important properties of ILs used in gas separation since they have a direct influence on gas permeation properties. Both density and viscosity for [C2mim][CF3BF3] and [C2mim][CF3SO2C(CN)2] were measured within the range of 293.15 K up to 353.15 K at ambient pressure and their behaviour with temperature is presented in Figure 1. The values are listed in Table S1 in Supporting Information.

T (K)
). The errors bars are smaller than the symbols used to represent the experimental data. From Figure 1 it is possible to see that the temperature dependence of density for where T is temperature in K and a, b, and c are adjustable parameters, which are listed in Table 1. For [C 2 mim][CF 3 BF 3 ], a polynomial second-order fitting [55,56] was used, while for [C 2 mim][CF 3 SO 2 C(CN) 2 ] a linear equation was well describing the experimental data (c = 0), as it can be observed from the correlation coefficient R 2 listed in Table 1.

Linear Fitting
The isobaric thermal expansion coefficients (α P ) were calculated using density experimental data through Equation (2).
where ρ is the density in g·cm −3 and P is the pressure. IL, much higher (50% higher) thermal expansion coefficients were measured, indicating a larger change in volume with temperature compared to other ILs. The molar volumes V M (cm 3 ·mol −1 ) were also calculated for the two studied Ils, in the same range of temperatures using Equation (3).
where ρ corresponds to the density (g·cm −3 ) and M is the molar mass (g·mol −1 ). The calculated molar volumes are listed in Table S2 and represented in Figure S1.
Where η corresponds to the viscosity in mPa.s, Ea is the activation energy, R is the ideal gas constant, and T is the temperature in K. The values η∞ and Ea for the two ILs under study are listed in Table 2.  Figure 1. The values are listed in Table S1 in Supporting Information.

T (K)
). The errors bars marked are smaller than the symbols used to represent the experimental data.
where η corresponds to the viscosity in mPa·s, E a is the activation energy, R is the ideal gas constant, and T is the temperature in K. The values η ∞ and E a for the two ILs under study are listed in Table 2.   (1), it was also possible to obtain the solubilities (S) of the mentioned three gases, which are also depicted in Figure 4. The values of the gas permeabilities, diffusivities, and solubilities at 293.15 K and 1 bar of feed pressure are presented in Table 4, whereas for 308.15 K and 1 bar of feed pressure are listed in Table S5 3 ], were also introduced in Figures 3 and 4.
The common gas permeability trend for ILs, consisted in the following order: P CO2 >> P CH4 > P N2 and reported previously [44] was found to be fair for ILs studied in this work as well ( Figure 3). The new SILMs demonstrated high performance in terms of carbon dioxide permeability as compared to other SILMs given for comparison ( Figure 3   The common gas permeability trend for ILs, consisted in the following order: PCO2 >> PCH4 > PN2 and reported previously [44] was found to be fair for ILs studied in this work as well (Figure 3). The new SILMs demonstrated high performance in terms of carbon dioxide permeability as compared to other SILMs given for comparison (Figure 3) Figure 4.  [58] For both new SILMs, the gas diffusivity followed the following gas order: CO2 > N2 > CH4, which is in agreement with the gas kinetic diameters, CO2 (3.30Å) < N2 (3.64Å) < CH4 (3.80Å) [59]. It can also be observed that [C2mim][CF3SO2C(CN)2] presents a higher diffusivity for all gases when compared to [C2mim][CF3BF3], following the same trend observed for gas permeability (Figure 4). Surprisingly, CH4 diffusivity for [C2mim][CF3SO2C(CN)2] filled SILM (1313 × 10 −12 m 2 s −1 ) was found to be the highest among studied SILMs and was significantly higher than for the other gases. It should be stated, that the standard deviation obtained from the three independent measurements was very small, indicating the accuracy of the mentioned result. It can also be observed that, for the three studied gases, SILM with Although it was previously stated that there is an inverse proportionality relationship between viscosity and gas diffusivity [31,32], this cannot be used to explain the lower gas diffusivity of
For both new SILMs, the gas diffusivity followed the following gas order: CO 2 > N 2 > CH 4 , which is in agreement with the gas kinetic diameters, CO 2 (3.30Å) < N 2 (3.64Å) < CH 4 (3.80Å) [59]. It can also be observed that [C 2 mim][CF 3 SO 2 C(CN) 2 ] presents a higher diffusivity for all gases when compared to [C 2 mim][CF 3 BF 3 ], following the same trend observed for gas permeability (Figure 4). Surprisingly, CH 4 diffusivity for [C 2 mim][CF 3 SO 2 C(CN) 2 ] filled SILM (1313 × 10 −12 m 2 ·s −1 ) was found to be the highest among studied SILMs and was significantly higher than for the other gases. It should be stated, that the standard deviation obtained from the three independent measurements was very small, indicating the accuracy of the mentioned result. It can also be observed that, for the three studied gases, SILM with [C 2 mim][CF 3 Figure 4. The following trend for the solubility of gases was observed: CO 2 >> CH 4 > N 2 , which was found to correlate with the dependence noticed above for gas permeabilities. This indicates that the gas permeability in SILMs is governed by gas solubility [31,32,60]. It is interesting to observe that despite their quite different chemical structures, both ILs under study show a similar value for CO 2 solubility (26.  Table S2. As for CH 4  , taking into account that the cation is the same for both Ils. However, when CO 2 solubility was considered, an almost linear relationship could be obtained, as also found for other ILs with similar cation [33]. Although it is not possible to directly compare the CH 4

Temperature Effect on Permeation Properties
It is well known that temperature has a direct and visible effect in gas permeation properties through ILs [61]. The effect of temperature was also here studied for [C 2 mim][CF 3 BF 3 ] and [C 2 mim][CF 3 SO 2 C(CN) 2 ] filled SILMs. The values for gas permeability, diffusivity, and solubility at 308.15 K, for the three studied gases, are listed in Table S5 and compared with those obtained at 293.15 K in Figure 5.

Temperature Effect on Permeation Properties
It is well known that temperature has a direct and visible effect in gas permeation properties through ILs [61]. The effect of temperature was also here studied for [C2mim][CF3BF3] and [C2mim][CF3SO2C(CN)2] filled SILMs. The values for gas permeability, diffusivity, and solubility at 308.15 K, for the three studied gases, are listed in Table S5 and compared with those obtained at 293.15 K in Figure 5. As expected, Figure 5a shows that an increase in temperature generally leads to high gas permeability for all the gases for the two studied ILs.  Figure 5c, this parameter tends generally to decrease with increase in temperature. An average decrease of 37% was obtained for both SILMs independently As expected, Figure 5a shows that an increase in temperature generally leads to high gas permeability for all the gases for the two studied ILs. There is only one exception for CO 2 Figure 5c, this parameter tends generally to decrease with increase in temperature. An average decrease of 37% was obtained for both SILMs independently of the nature of the gas under study. The exception was noticed for N 2 solubility in [C 2 mim][CF 3 SO 2 C(CN) 2 ], where an increase of 129% was indicated with the increase in temperature.
In conclusion, the increase in gas permeability with temperature can be ascribed to the enhanced molecular diffusion through the SILMs, according to the solution-diffusion model [62]. As the temperature increases, the diffusion of gas molecules through the SILM increases due not only to the higher gas kinetic energy, but also because of the decrease in the viscosity of ILs [63].

CO 2 Separation Performance
The CO 2 /N 2 and CO 2 /CH 4 ideal permselectivities of the studied SILMs at temperatures of 293.15 K and 308.15 K are summarized in Table 5. As expected from the single gas permeabilities discussed before for the three studied gases through the new SILMs, a higher ideal permselectivity was obtained for the CO 2 /N 2 gas pair in comparison to CO 2 /CH 4 . This behavior is generally observed for ILs [64].  2 ] SILMs decreased with the increase in temperature for both CO 2 /N 2 and CO 2 /CH 4 gas pairs. Specifically, the permselectivity of [C 2 mim][CF 3 BF 3 ] for both gas pairs showed a decrease of 24% and 22%, respectively. For [C 2 mim][CF 3 SO 2 C(CN) 2 ] SILM with an increase in temperature, a 25% decrease for CO 2 /N 2 and an 18% increase for CO 2 /CH 4 permselectivities was observed. The CO 2 /N 2 permselectivity decrease with temperature stems from the solubility selectivity and, in fact, gas solubility decreases with the rise in temperature [65].

Membranes Performance Comparison
To better interpret the permeability together with the ideal permselectivity results for CO 2 /N 2 and CO 2 /CH 4 pair of gases, the Robeson plots, presented in Figure 6a,b, were applied. The Robeson plot upper limit for each gas pair represents an empirical limit proposed in 2008 by Robeson, through the correlation of a substantially amount of gas permeability and selectivity data. The area in the right hand corner corresponding to high CO 2 permeability and simultaneously high CO 2 /N 2 or CO 2 /CH 4 selectivity is the target region, where in 2008 no data were yet available.  [45]. Literature data for other SILMs was used for comparison [28,29,57,64,[66][67][68][69][70][71][72][73].
It can be seen from Figure 6a  Overall, this SILM is on top of the Robeson plot line, meaning that it shows a good selectivity when compared to the state-of-the-art membranes, and demonstrates an improvement in terms of CO2 permeation when compared of other ILs, representing a step forward to overcome the low gas flux of IL-based membranes.
It can be seen from Figure 6a 2 ]. Overall, this SILM is on top of the Robeson plot line, meaning that it shows a good selectivity when compared to the stateof-the-art membranes, and demonstrates an improvement in terms of CO 2 permeation when compared of other ILs, representing a step forward to overcome the low gas flux of IL-based membranes.
In what concerns the CO 2 /CH 4 separation, and as discussed earlier, the fact that [C 2 mim][CF 3 SO 2 C(CN) 2 ] has a higher CH 4 diffusivity, and consequently higher permeability, than those SILMs used for comparison, greatly affects CO 2 /CH 4 selectivity. So, despite the great increase in CO 2 permeability, its permselectivity suffers a great reduction. For example, comparing SILMs permselectivity filled with [C 2 mim][CF 3 SO 2 C(CN) 2 ] to that of  3 , 98.0%, Sigma Aldrich), (trifluoromethyl)trimethylsilane (99%, Sigma Aldrich), boron trifluoride diethyl etherate (BF 3 ·Et 2 O, Sigma Aldrich), and trifluoromethanesulphonyl fluoride (CF 3 SO 2 F, 98% ABCR, Karlsruhe, Germany) were used without further purification. Carbon dioxide (CO 2 ), nitrogen (N 2 ), and methane (CH 4 ) were all supplied by Air Liquide and were of, at least, 99.99% purity. Gases were used as received without further purification. Tetrahydrofuran (THF) was purified by refluxing over the deep purple sodium-benzophenone complex. Propionitrile (EtCN, 99%, Sigma Aldrich) was distilled over CaH 2 . Triethylamine (99.5%, Merck, Darmstadt, Germany) was distilled under inert atmosphere over metal Na. Malononitrile (99%, Sigma Aldrich) was distilled under reduced pressure. Potassium fluoride (KF, 99.97%, Sigma Aldrich) was dried in a stainless-steel pan using a hot plate (Severin, Sundern, Germany, 1500 W) set to maximum heating (T > 200 • C) for 2-3 h with stirring. After cooling to~70 • C, it was quickly transferred to the heat-treated glass flask, then stored under inert atmosphere. Anhydrous magnesium sulfate was prepared in-house from MgSO4·7H 2 O (MgSO 4 , >99%, Sigma Aldrich) by heating a saturated aqueous solution of magnesium sulfate in a stainless-steel pan (~500 mL) using a hot plate (Severin 1500 W) set to maximum heating. Following evaporation of all liquid water, heating was continued for another~6-8 h (T > 200 • C) to obtain an anhydrous cake. The cake was allowed to cool to~70 • C, broken into~1-cm sized pieces, then stored in a sealed glass jar prior to use. Porous hydrophobic poly(vinylidene fluoride) (PVDF) Durapore ® supports, with a pore size of 0.22 µm and average thickness of 125 µm, were provided by the Merck Millipore Corporation.

Potassium Trifluoro(Trifluoromethyl)Borate (KCF 3 BF 3 )
The 150 mL of anhydrous THF, freshly dried KF (10.00 g, 172.3 mmol), trimethyl borate (19.68 g, 189.3 mmol), and (trifluoromethyl)trimethylsilane (26.90 g, 189.3 mmol) were added to the four-necked round bottom flask equipped with a reflux condenser, thermometer, mechanical stirrer, and dropping funnel under inert atmosphere. The suspension was slowly heated to 50 • C and the reaction was continued for 5-6 h at 50 • C until the formation of a clear transparent solution. The solution was further cooled down to -25 • C and after the dropwise addition of BF 3 ·Et 2 O (21.35 g, 150.5 mmol), the reaction was allowed to warm up to RT. The stirring was continued at RT for 3 h, whereupon the reflux condenser was replaced by a distillation system and 50 mL of THF were distilled off. The 50 mL of i-PrOH were added to the mixture and the THF residue was evaporated until the vapor temperature reached 100 • C. The reaction mass was cooled down to RT, the precipitate was filtered, collected, and dried at 25 • C/0.1 mbar for 2 h. Yield: 19.9 g (60%); T m > 350 • C; 13

1-Ethyl-3-Methyl-Imidazolium Trifluoro(Trifluoromethyl)Borate [C2mim][CF 3 BF 3 ]
The 1-ethyl-3-methyl-imidazolium bromide (16.95  Malononitrile (4.07 g, 61.6 mmol) and freshly distilled triethylamine (13.72 g, 135.6 mmol) were dissolved in 60 mL of anhydrous EtCN under inert atmosphere at RT. The reaction solution was cooled down to −40 • C in the cold bath (CCl 4 + dry ice). The trifluoromethanesulphonyl fluoride (10.31 g, 67.8 mmol) was distilled in the pre-weighted thick-walled glass trap from the balloon and then redistilled in the dropping funnel containing 15 mL of anhydrous EtCN and equipped with a cooling jacket precooled to −78 • C. The dropping funnel with the CF 3 SO 2 F solution was attached to the reaction flask and was added dropwise under inert atmosphere. The reaction mixture was stirred at −40 • C for 1 h and then allowed to warm up to RT. Stirring was continued at RT for an additional 12 h, whereupon the solvent was removed under reduced pressure at 35 • C/10 mbar and the obtained brown oil was dried at 35 • C/0.1 mbar for 3 h. The residual brownish oil was dissolved in 90 mL of DCM and washed with water (4 × 90 mL). The organic layer was dried over anhydrous MgSO 4 , which was further filtered off and the solvent was removed at reduce pressure. The product in a form of light-brown oil was dried at 55 • C/0.1 mbar for 5 h with a special flask filled with P 2 O 5 and introduced into the vacuum line. Yield: 15 The suspension of LiH (0.86 g, 107.6 mmol) in 20 mL of anhydrous THF was slowly added to the solution of triethylammonium dicyano((trifluoromethyl)sulfonyl)methanide (10.74 g, 35.9 mmol) in 50 mL of anhydrous THF preliminary cooled down to 0 • C under inert atmosphere (Ar). Caution: Immediately after the start of the reaction, gas evolution was observed. The reaction mixture was warmed up to room temperature and stirred at 35 • C for 2 h. The excess of LiH was filtered off, the solvent was removed at reduced pressure, and the residue was dried at 60 • C/0.1 mbar for 3 h. The anhydrous CH 2 Cl 2 was added to the crude product representing viscous brown oil and the mass was stirred mechanically under the inert flow till the formation of the yellow powder. The powder was collected by filtration and dried at 60 • C/10 mbar for 2 h, whereupon it was dissolved in 90 mL of anhydrous CH 3 CN and refluxed with carbon black for 2 h. The charcoal was filtered off, the solvent was removed under reduced pressure, and the light-yellow viscous oil was dried at 60 • C/10 mbar for 2 h. The anhydrous DCM was added to the oil and the mass was stirred mechanically under the inert flow till the formation of the pale-yellow powder. The powder was collected by filtration and dried at 60 • C/0.1 mbar for 12 h. Yield: 6.31 g (86%); T m = no melting point determined (decomposition started >260 • C); 13

Water Content
In order to reduce the content of water and other volatile substances, all IL samples were additionally dried at approximately 1 Pa and 318 K for at least 4 days and their H 2 O content was determined by Karl Fischer titration using an 831 KF Coulometer (Metrohm).

Spectroscopical Properties
NMR spectra were recorded on AMX-400 and AMX-600 spectrometers (Bruker, Germany) at 25 • C in the indicated deuterated solvent and are listed in ppm. The signal corresponding to the residual protons of the deuterated solvent was used as an internal standard for 1 H and 13 C NMR, while for 19 F NMR, the C 6 F 6 (−162.5 ppm) was utilized as an external standard. IR spectra were acquired on a Magna-750 (Nicolet Instrument Corporation) or on Tensor 27 (Brucker, Germany) Fourier IR-spectrometer using ATR technology (128 scans, resolution is 2 cm −1 ) and Spectragryph optical spectroscopy software [74].

Thermal Properties
Thermal gravimetric analysis (TGA) was carried out in air on a TGA2 STARe System (Mettler Toledo, Switzerland) applying a heating rate of 5 • C min −1 . The onset weight loss temperature (T onset ) was determined as the point in the TGA curve at which a significant deviation from the horizontal was observed. The resulting temperature was then rounded to the nearest 5 • C. For Differential Scanning Calorimetry (DSC) measurements, all samples were hermetically sealed in Al pans inside the argon-filled glove-box (MBRAUN MB-Labstar, H 2 O and O 2 content < 0.5 ppm). DSC of [C2mim][CF 3 BF 3 ] and [C2mim][CF 3 SO 2 C(CN) 2 ] samples was performed on a DSC3+ STARe System (Mettler Toledo, Switzerland) differential calorimeter applying a heating rate of 2 • C min −1 in the range of −90 to 150 • C. The glass transition (T g ) and melting (T m ) temperatures were determined during second heating cycle. Crystallization temperature (T cr ) was determined during second cooling cycle. For simplicity and convenience, the temperatures of thermal transitions were measured as maxima of the peaks corresponding to the endothermic or exotermic heat effects.

Thermophysical Properties
Measurements of viscosity (η) and density (ρ) were carried out in the temperature range from 293.15 K up to 353.15 K and at atmospheric pressure using an Anton Paar (model SVM 3000) automated rotational Stabinger viscometer-densimeter, with a temperature uncertainty of ±0.01 K. The relative uncertainty of the dynamic viscosity is ±0.25%, and the absolute uncertainty of the density is ±0.0005 g·cm −3 . Triplicate measurements were carried out and standard deviations were calculated.

Supported Ionic Liquid Membranes (SILMs) Preparation
The studied ILs were supported on porous hydrophobic PVDF supports using vacuum, as previously described [30,75]. In short, the porous membrane filter was firstly placed inside a vacuum chamber for about 1 h in order to remove any impurities and/or the air within the pores. Afterwards, while keeping the vacuum in the chamber, a few drops of the studied ILs were carefully placed on the membrane surface. To ensure proper impregnation, the SILM was left for over an hour inside the vacuum chamber, after which it was taken out and the IL's excess was wiped with paper tissue.

Gas Permeation Experiments
Ideal CO 2 and N 2 permeabilities and diffusivities through the prepared membranes were measured at 293.15 K and 308.15 K using an apparatus with a time-lag method implemented, whose detailed description was previously reported [76]. The SILMs were inserted into the permeation cell and degassed under vacuum (<0.1 kPa) for 12 h. Then, gas permeation experiments were conducted in such a way that three CO 2 and N 2 gas independent measurements were carried out for each membrane sample. In between each run, the permeation cell and lines were thoroughly vacuumed.
The solution-diffusion mass transport model was used to describe the gas transport through the prepared SILMs. Consequently, after measuring the permeability (P) and the diffusivity (D), the solubility (S) can be calculated using Equation (1).
A critical parameter to quantify membrane performance is the ideal permselectivity, α i,j , that can be calculated as shown in Equation (2): where i is the most permeable gas and j is the less permeable gas.

Conclusions
In the present study, two ILs containing common cation ([C 2 mim] + ) and two asymmetric anions, namely [C 2 mim][CF 3 SO 2 C(CN) 2 ] and [C 2 mim][CF 3 BF 3 ], were successfully prepared. The alkali dicyano((trifluoromethyl)sulfonyl)methanide salt and the [C 2 mim][CF 3 SO 2 C (CN) 2 ] IL were here synthesized for the first time. The density and viscosity of both ILs were measured in the temperature range between 293.15 K and 353.15 K and atmospheric pressure and compared to those properties of ILs bearing structurally similar symmetric anions. Finally, two new ILs were further used as liquid phases for the formation of SILMs and their CO 2 /N 2 and CO 2 /CH 4 gas separation properties were studied in detailed.
The application of "asymmetric principle" for the design of new anions led to the mixed results that were found to be dependent on the type of the functional group introduced in the anion. The substitution of one fluorine atom with the CF 3 group in the BF 4 anion resulted in a pronounced decrease at melting point (from 15 to −17.9 • C), a reduction of thermal stability (from T onset = 420 to 215 • C). In terms of permeation properties, a decrease in CO 2 , N 2 , and CH 4 single gas permeabilities (from 968, 44 3 ]. Thus, the substitution of the CN group with the CF 3 SO 2 group led to the disappearance of crystallization or melting processes, and an increase in thermal stability (from T onset = 270 to 300 • C). In terms of permeation properties, an increase in CO 2 and CH 4 permeabilities (1095 and 667, 152 and 34.4 Barrer at 293.15 K, and 1 bar of feed pressure, respectively), being this last one due to a remarkable increase in the CH 4 diffusivity.
In terms of viscosity, two different behaviors were here observed, as the introduction of CF 3 in [C 2 mim][BF 4 ] leads to an [C 2 mim][CF 3 BF 3 ] IL with slightly higher viscosity than the former, probably due to the bulkiness and rigidity of the CF3 group, while [C 2 mim][CF 3 SO 2 C(CN) 2 ] presents a lower viscosity than the corresponding IL with symmetrical anion structure [C 2 mim][NTf 2 ]. The introduction of CN groups in the IL's anion to achieve asymmetry has already been shown to be a valuable strategy to provide ILs with low viscosity.
To conclude, the change from [BF 4 ]to [CF 3 BF 3 ]anion does not represent a step forward in the improvement of SILMs, neither in terms of CO 2 permeability nor in CO 2 /N 2 selectivity. The [C 2 mim][CF 3 SO 2 C(CN) 2 ] SILM was found to be on top of the Robeson plot line, showing good (CO 2 /N 2 ) permselectivity when compared to the state-of-theart membranes, and demonstrating an improvement in terms of CO 2 permeation when compared of other ILs, representing a significant advancement in overcoming the low gas flux of IL-based membranes.