Ionic Liquid Confined in Mesoporous Polymer Membrane with Improved Stability for CO2/N2 Separation

Supported ionic liquid membranes (SILMs) have a promising prospect of application in flue gas separation, owing to its high permeability and selectivity of CO2. However, existing SILMs have the disadvantage of poor stability due to the loss of ionic liquid from the large pores of the macroporous support. In this study, a novel SILM with high stability was developed by confining ionic liquid in a mesoporous polymer membrane. First, a mesoporous polymer membrane derived from a soluble, low-molecular-weight phenolic resin precursor was deposited on a porous Al2O3 support, and then 1-ethyl-3-methylimidazolium tetrafluoroborate ([emim][BF4]) was immobilized inside mesopores of phenolic resin, forming the SILM under vacuum. Effects of trans-membrane pressure difference on the SILM separation performance were investigated by measuring the permeances of CO2 and N2. The SILM exhibits a high ideal CO2/N2 selectivity of 40, and an actual selectivity of approximately 25 in a mixed gas (50% CO2 and 50% N2) at a trans-membrane pressure difference of 2.5 bar. Compared to [emim][BF4] supported by polyethersulfone membrane with a pore size of around 0.45 μm, the [emim][BF4] confined in a mesoporous polymer membrane exhibits an improved stability, and its separation performance remained stable for 40 h under a trans-membrane pressure difference of 1.5 bar in a mixed gas before the measurement was intentionally stopped.


Introduction
Membrane-related processes including gas separation membranes and supported ionic liquid membranes (SILMs) are considered to be very promising technologies. Owing to their high permeability and selectivity of CO 2 , SILMs have a promising prospect of application in biogas purification and flue gas separation [1][2][3]. Usually, SILM supports are microfiltration membranes made from polyethersulphone (PES) [3][4][5][6][7], polyvinylidene fluoride (PVDF) [7][8][9][10][11], polypropylene (PP) [7,12], nylon [6,13], ceramics [14][15][16], and their pore sizes are in the range from 0.1 to 0.45 um. For example, Adibi et al. prepared an SILM by impregnating alumina porous supports in ionic liquid [1,3-di (3-methyl-imidazolium) propane bis(trifluoromethylsulfonyl) imide], for which the ideal CO 2 /CH 4 selectivity was 27.1 at 27 • C [16]. However, the application of SILMs in industry is still limited because ionic liquid in this kind of membrane is mainly immobilized in the support by capillary forces, and can easily leak out from the larger pores of the support [6,7,17], whose pore size distribution is wide under a high trans-membrane pressure difference. For instance, Neves et al. observed that the membrane weight of an SILM in a hydrophilic support decreased continuously at a trans-membrane pressure difference of 1 bar [18]. Albo et al. introduced four kinds of ionic liquid with phosphonium cation into either hydrophilic or hydrophobic PVDF and found that the membrane weight loss did not reach a stable value at a 1 bar pressure difference [19].
According to the Young-Laplace equation, the maximum trans-membrane pressure difference that SILMs can withstand is relevant to the pore size of the support membrane. Decreasing the pore size will increase capillary forces, thereby enhancing the stability of an SILM. Tsuru et al. immobilized a high CO 2 solubility ionic liquid (IL) [emim][Ac] into mesoporous TiO 2 with an average pore size of approximately 20 nm [20] and nanoporous TiO 2 and SiO 2 -ZrO 2 layers with average pore sizes of approximately 2.5 and 1 nm, respectively [21]. The stability of the membranes was ensured for 25 h at an applied pressure of 4 bar in the separation of CO 2 and N 2 . Obviously, a narrow pore size distribution of the support will clearly have a positive effect on the stability of an SILM [22,23].
In the past decades, significant efforts have been devoted to developing mesoporous polymers with large surface areas and very narrow pore size distributions. The evaporation induced self-assembly (EISA) method was widely used to obtain mesoporous polymers [24][25][26]. Mesoporous phenolic resin with different structures and uniform pore size can be generated by simply adjusting the ratio of the phenol/template and the calcination temperature [25].
Obviously, a mesoporous polymer with uniform and smaller pore sizes is quite suitable to use as the support of an SILM. In this study, ionic liquid 1-ethyl-3-methylimidazolium tetrafluoroborate [emim] [BF 4 ] confined in a mesoporous polymer membrane with high stability was developed. A mesoporous polymer membrane was deposited on a porous Al 2 O 3 support by self-assembly of triblock copolymers with soluble, low-molecular-weight phenolic resin precursors, followed by high-temperature treatment under Ar flow. The ionic liquid [emim] [BF 4 ] was immobilized in the mesoporous polymer membrane, forming a novel SILM. Effects of trans-membrane pressure difference on SILM separation performances were investigated. By confining ionic liquid into the mesopores of the polymer, an improved stability was achieved.  4 ] with a thickness of around 500 nm. EDXS was used to analyze the composition of the upper layer ( Figure 2c). The characteristic peaks attributed to elements C (Figure 2d), B (Figure 2e), and F ( Figure 2f) were found in the survey spectrum. From the corresponding elemental mapping, it can be seen that C, B, and F were mainly distributed on the top of the membrane, as indicated by the white dash line. According to SEM and elemental mapping images, it can be concluded that [emim] [BF 4 ] confined in the mesoporous polymer membrane was successfully grown on the Al 2 O 3 support.     In order to observe the pore size of the phenolic resin derived from the resol precursor, the polymer membrane was scratched from the surface of the alumina support before introducing ionic liquid. Transmission electron micro (TEM) was used to observe the pore size of the phenolic resin. It can be seen from Figure 3 that the pore size is around 7 nm, which is in accordance with the previous reports [25]. Thus, the transport of gases through such a porous membrane should follow the mechanism of Knudsen diffusion. Figure 4 shows the permeances and CO 2 /N 2 ideal selectivities of the mesoporous polymer membrane on the Al 2 O 3 support. At a trans-membrane pressure difference of 1 bar, the CO 2 permeance is as high as 1500 mol/m 2 ·h·bar, and the CO 2 /N 2 selectivity is only 0.9, which is quite close to the theoretical Knudsen separation factor of 0.8. Obviously, the phenolic resin-based membrane with a pore size of around 7 nm could not well separate the CO 2 from the gas mixture of CO 2 /N 2 . In order to observe the pore size of the phenolic resin derived from the resol precursor, the polymer membrane was scratched from the surface of the alumina support before introducing ionic liquid. Transmission electron micro (TEM) was used to observe the pore size of the phenolic resin. It can be seen from Figure 3 that the pore size is around 7 nm, which is in accordance with the previous reports [25]. Thus, the transport of gases through such a porous membrane should follow the mechanism of Knudsen diffusion. Figure 4 shows the permeances and CO2/N2 ideal selectivities of the mesoporous polymer membrane on the Al2O3 support. At a trans-membrane pressure difference of 1 bar, the CO2 permeance is as high as 1500 mol/m 2 ·h·bar, and the CO2/N2 selectivity is only 0.9, which is quite close to the theoretical Knudsen separation factor of 0.8. Obviously, the phenolic resinbased membrane with a pore size of around 7 nm could not well separate the CO2 from the gas mixture of CO2/N2.   In order to observe the pore size of the phenolic resin derived from the resol precursor, the polymer membrane was scratched from the surface of the alumina support before introducing ionic liquid. Transmission electron micro (TEM) was used to observe the pore size of the phenolic resin. It can be seen from Figure 3 that the pore size is around 7 nm, which is in accordance with the previous reports [25]. Thus, the transport of gases through such a porous membrane should follow the mechanism of Knudsen diffusion. Figure 4 shows the permeances and CO2/N2 ideal selectivities of the mesoporous polymer membrane on the Al2O3 support. At a trans-membrane pressure difference of 1 bar, the CO2 permeance is as high as 1500 mol/m 2 ·h·bar, and the CO2/N2 selectivity is only 0.9, which is quite close to the theoretical Knudsen separation factor of 0.8. Obviously, the phenolic resinbased membrane with a pore size of around 7 nm could not well separate the CO2 from the gas mixture of CO2/N2.   After introducing ionic liquid into the pores of the phenolic resin, the pores will be full of ionic liquid under vacuum, forming a novel SILM, [emim] [BF 4 ] confined in the mesoporous polymer membrane on the Al 2 O 3 support. The permeation of gases through the novel SILM follows the solution-diffusion mechanism. Figure 5a presents the gas permeance and ideal selectivity of the novel SILM for pure CO 2 or N 2 gas at different trans-membrane pressures. The permeance of CO 2 is significantly higher than that of N 2 , and the ideal selectivities of CO 2 /N 2 based on the permeation tests are in the range of 32-40. When the trans-membrane pressure difference is 2.5 bar, the ideal selectivity of CO 2 /N 2 is about 40. High permeability of CO 2 should result from its high solubility and diffusivity in ionic liquid. N-containing organic heterocyclic was reported to play an important role in interacting with CO 2 , thus increasing its solubility [28]. Kazarian et al. found that CO 2 could also interact with BF 4 − owing to a Lewis acid/base interaction, and thus facilitate CO 2 transport [29].

Results and Discussion
Lim et al. pointed out that CO 2 molecules can easily squeeze into the inter-ion space between the cation and anion [30]. Blanchard et al. [31] and Shim et al. [30] further explained that this behavior is due to the existence of "void space" or "free volume" in the ionic liquid. Figure 5b presents the gas permeance and actual selectivity of the novel SILM for mixed CO 2 /N 2 gas (50% CO 2 and 50% N 2 ) at different trans-membrane pressures. The actual selectivities of CO 2 /N 2 based on the permeation tests are in the range of 23-25. When the trans-membrane pressure difference is 2.55 bar, the actual selectivity for mixed CO 2 /N 2 is about 25. Obviously, the ideal selectivity is much higher than the actual selectivity of CO 2 /N 2 ; this is because the actual selectivity for mixed CO 2 /N 2 gas is related to the ideal selectivity of CO 2 /N 2 and the pressure of the permeate side and the feed side [3,32]. The higher the ideal selectivity and the ratio of the pressure in the feed side to the permeate side are, the closer the ideal selectivity and the actual selectivity will be [3,32]. As shown in Figure 5, the CO 2 /N 2 selectivity increased slightly with the increase of the trans-membrane pressure difference gradient across the membrane. It is necessary to point out that the novel SILM shows a high CO 2 /N 2 selectivity at the high trans-membrane pressure difference of 2.55 bar, and the permeation flux of N 2 is still at a very low level, indicating the good stability of the membrane. The ideal CO 2 /N 2 selectivity of the novel SILM is comparable with other SILMs [20,21,33] and found to be very close to the intrinsic value of the [emim]BF 4 ] (ideal selectivity: 44 [3]) using a lag-time technique [34], although the permeance of the mixed gas for the novel SILM is lower than that reported in the literature [3].
Nanomaterials 2017, 7, 299 5 of 11 After introducing ionic liquid into the pores of the phenolic resin, the pores will be full of ionic liquid under vacuum, forming a novel SILM, [emim][BF4] confined in the mesoporous polymer membrane on the Al2O3 support. The permeation of gases through the novel SILM follows the solution-diffusion mechanism. Figure 5a presents the gas permeance and ideal selectivity of the novel SILM for pure CO2 or N2 gas at different trans-membrane pressures. The permeance of CO2 is significantly higher than that of N2, and the ideal selectivities of CO2/N2 based on the permeation tests are in the range of 32-40. When the trans-membrane pressure difference is 2.5 bar, the ideal selectivity of CO2/N2 is about 40. High permeability of CO2 should result from its high solubility and diffusivity in ionic liquid. N-containing organic heterocyclic was reported to play an important role in interacting with CO2, thus increasing its solubility [28]. Kazarian et al. found that CO2 could also interact with BF4 − owing to a Lewis acid/base interaction, and thus facilitate CO2 transport [29]. Lim et al. pointed out that CO2 molecules can easily squeeze into the inter-ion space between the cation and anion [30]. Blanchard et al. [31] and Shim et al. [30] further explained that this behavior is due to the existence of "void space" or "free volume" in the ionic liquid. Figure 5b presents the gas permeance and actual selectivity of the novel SILM for mixed CO2/N2 gas (50% CO2 and 50% N2) at different trans-membrane pressures. The actual selectivities of CO2/N2 based on the permeation tests are in the range of 23-25. When the trans-membrane pressure difference is 2.55 bar, the actual selectivity for mixed CO2/N2 is about 25. Obviously, the ideal selectivity is much higher than the actual selectivity of CO2/N2; this is because the actual selectivity for mixed CO2/N2 gas is related to the ideal selectivity of CO2/N2 and the pressure of the permeate side and the feed side [3,32]. The higher the ideal selectivity and the ratio of the pressure in the feed side to the permeate side are, the closer the ideal selectivity and the actual selectivity will be [3,32]. As shown in Figure 5, the CO2/N2 selectivity increased slightly with the increase of the trans-membrane pressure difference gradient across the membrane. It is necessary to point out that the novel SILM shows a high CO2/N2 selectivity at the high trans-membrane pressure difference of 2.55 bar, and the permeation flux of N2 is still at a very low level, indicating the good stability of the membrane. The ideal CO2/N2 selectivity of the novel SILM is comparable with other SILMs [20,21,33] and found to be very close to the intrinsic value of the [emim]BF4] (ideal selectivity: 44 [3]) using a lag-time technique [34], although the permeance of the mixed gas for the novel SILM is lower than that reported in the literature [3].  For comparison, the ionic liquid [emim][BF4] was introduced into the porous PES membrane with a mean pore size of around 0.45 μm. Figure 6 presents the CO2 permeance of [emim][BF4] confined in the PES membrane and mesoporous polymer membrane at different trans-membrane pressure differences using the mixed gas (50% CO2 and 50% N2) as feed gas. The CO2 permeance increases suddenly from 0.015 to 3.59 mol/m 2 ·h·bar as the trans-membrane pressure difference increases from 0.75 to 1 bar, which might be caused by the leakage of [emim][BF4] from the pores of the PES membrane. This indicates that the ionic liquid [emim][BF4] supported by the PES membrane with a pore size of around 0.45 μm cannot tolerant the trans-membrane pressure difference of more than 1 bar. In contrast, the CO2 permeances almost remain constant at about 0.024 mol/m 2 ·h·bar when the trans-membrane pressure difference increases from 0 to 2.5 bar, indicating the better stability of the novel SILM. It can be seen from Figure 6 that the CO2 permeance of the novel SILM is larger than that of the ionic liquid [emim][BF4] supported by the PES membrane, which is probably caused by the difference of the thickness of the IL layer between the two kinds of membranes.   4 ] was introduced into the porous PES membrane with a mean pore size of around 0.45 µm. Figure 6 presents the CO 2 permeance of [emim] [BF 4 ] confined in the PES membrane and mesoporous polymer membrane at different trans-membrane pressure differences using the mixed gas (50% CO 2 and 50% N 2 ) as feed gas. The CO 2 permeance increases suddenly from 0.015 to 3.59 mol/m 2 ·h·bar as the trans-membrane pressure difference increases from 0.75 to 1 bar, which might be caused by the leakage of [emim][BF4] from the pores of the PES membrane. This indicates that the ionic liquid [emim][BF 4 ] supported by the PES membrane with a pore size of around 0.45 µm cannot tolerant the trans-membrane pressure difference of more than 1 bar. In contrast, the CO 2 permeances almost remain constant at about 0.024 mol/m 2 ·h·bar when the trans-membrane pressure difference increases from 0 to 2.5 bar, indicating the better stability of the novel SILM. It can be seen from Figure 6 that the CO 2 permeance of the novel SILM is larger than that of the ionic liquid [emim] [BF 4 ] supported by the PES membrane, which is probably caused by the difference of the thickness of the IL layer between the two kinds of membranes. For comparison, the ionic liquid [emim][BF4] was introduced into the porous PES membrane with a mean pore size of around 0.45 μm. Figure 6 presents the CO2 permeance of [emim][BF4] confined in the PES membrane and mesoporous polymer membrane at different trans-membrane pressure differences using the mixed gas (50% CO2 and 50% N2) as feed gas. The CO2 permeance increases suddenly from 0.015 to 3.59 mol/m 2 ·h·bar as the trans-membrane pressure difference increases from 0.75 to 1 bar, which might be caused by the leakage of [emim][BF4] from the pores of the PES membrane. This indicates that the ionic liquid [emim][BF4] supported by the PES membrane with a pore size of around 0.45 μm cannot tolerant the trans-membrane pressure difference of more than 1 bar. In contrast, the CO2 permeances almost remain constant at about 0.024 mol/m 2 ·h·bar when the trans-membrane pressure difference increases from 0 to 2.5 bar, indicating the better stability of the novel SILM. It can be seen from Figure 6 that the CO2 permeance of the novel SILM is larger than that of the ionic liquid [emim][BF4] supported by the PES membrane, which is probably caused by the difference of the thickness of the IL layer between the two kinds of membranes.  In addition, permeation tests with binary gas mixtures (CO 2 /N 2 ) were carried out to assess the long-term stability of the novel SILM. As shown in Figure 7, under the trans-membrane pressure of 1.5 bar, the permeances of CO 2 and N 2 almost keep constant for 40 h, indicating that the membrane can withstand a high trans-membrane pressure difference without performance degradation.

For comparison, the ionic liquid [emim][BF
In addition, permeation tests with binary gas mixtures (CO2/N2) were carried out to assess the long-term stability of the novel SILM. As shown in Figure 7, under the trans-membrane pressure of 1.5 bar, the permeances of CO2 and N2 almost keep constant for 40 h, indicating that the membrane can withstand a high trans-membrane pressure difference without performance degradation.

Preparation of Ionic Liquid Confined in Mesoporous Polymer Membrane
The porous Al2O3 support was obtained using a method similar to that described in the literature [35]. In a typical process, 0.45 g alumina powders were pressed into 18 mm disk at 8 MPa. Then, a porous Al2O3 support was obtained after sintering at 1400 °C for 10 h. The resol precursor was obtained using a soft-template method described in the literature [24,25]. Typically, phenol (0.61 g) was melted at around 40 °C in a beaker. Under stirring, 0.13 g NaOH aqueous solution (20 wt %) was added slowly. Then, 1.05 g formalin (37 wt %) was added dropwise, and the reaction mixture was stirred at 70 wt % for 60 min. After cooling the mixture to room temperature, the pH of the reaction mixture was adjusted to neutral (7.0) using 0.6 M HCl solution. Water was removed under vacuum below 45 °C. The above resol precursor was dissolved in ethanol, forming 20 wt % ethanolic solution, which was further added to the ethanol solution of pluronic F127 under stirring with the molar composition of phenol/formaldehyde/NaOH/F127 of 1:2:0.1:0.003-0.008. Then, 0.2 mL of the above homogeneous resol solution was spread on the surface of the Al2O3 support, followed by the solvent evaporation in a petri dish for 3 h, and subsequently in the oven at 100 °C for 24 h. It is necessary to point out that the drying processes probably have an important influence on the structure of the

Preparation of Ionic Liquid Confined in Mesoporous Polymer Membrane
The porous Al 2 O 3 support was obtained using a method similar to that described in the literature [35]. In a typical process, 0.45 g alumina powders were pressed into 18 mm disk at 8 MPa. Then, a porous Al 2 O 3 support was obtained after sintering at 1400 • C for 10 h. The resol precursor was obtained using a soft-template method described in the literature [24,25]. Typically, phenol (0.61 g) was melted at around 40 • C in a beaker. Under stirring, 0.13 g NaOH aqueous solution (20 wt %) was added slowly. Then, 1.05 g formalin (37 wt %) was added dropwise, and the reaction mixture was stirred at 70 wt % for 60 min. After cooling the mixture to room temperature, the pH of the reaction mixture was adjusted to neutral (7.0) using 0.6 M HCl solution. Water was removed under vacuum below 45 • C. The above resol precursor was dissolved in ethanol, forming 20 wt % ethanolic solution, which was further added to the ethanol solution of pluronic F127 under stirring with the molar composition of phenol/formaldehyde/NaOH/F127 of 1:2:0.1:0.003-0.008. Then, 0.2 mL of the above homogeneous resol solution was spread on the surface of the Al 2 O 3 support, followed by the solvent evaporation in a petri dish for 3 h, and subsequently in the oven at 100 • C for 24 h. It is necessary to point out that the drying processes probably have an important influence on the structure of the mesoporous membrane and thus its performance. Here, we employed the common drying procedure. After further treating in flowing argon in a quartz tube furnace at 350 • C for 5 h with a ramping rate of 1 • C/min, the mesoporous polymer membrane on the Al 2 O 3 support was obtained. The coating, drying, and calcination processes were repeated three times to repair the possible defects.
The supported ionic liquid membrane was prepared as follows. First, 0.15 g [emim] [BF 4 ] was dropped onto the surface of the mesoporous membrane, then introduced into the mesopores of the phenolic resin on the Al 2 O 3 support under a vacuum of 0.03 MPa, forming ionic liquid confined in the mesoporous polymer membrane on the Al 2 O 3 support. The excess ionic liquid was carefully wiped off using a filter paper. Figure 8 shows the schematic flowcharts of the membrane permeation measurement. The system consists of a feed cell and a permeate cell separated by an SILM, which was sealed by O-rings. CO 2 and N 2 were fed with a flow rate of 6 mL/min, while CH 4 was always fed to the permeate side as a sweep gas with a flow rate of 1 mL/min under atmospheric pressure. CH 4 , CO 2 , and N 2 flow rates were controlled by mass-flow controllers (Bronkhorst, Ruurlo, The Netherlands). Trans-membrane pressure was monitored using a pressure gauge at the feed side. The gases in the permeation side included the sweep gas and the permeated gas. The flow rate of the gas at the outlet in the permeation side was measured by a bubble flow meter, while gas composition at the outlet was determined by a gas chromatograph (Agilent 7890, Agilent Technology, Santa Clara, CA, USA) equipped with a thermal conductivity detector (TCD). mesoporous membrane and thus its performance. Here, we employed the common drying procedure. After further treating in flowing argon in a quartz tube furnace at 350 °C for 5 h with a ramping rate of 1 °C/min, the mesoporous polymer membrane on the Al2O3 support was obtained. The coating, drying, and calcination processes were repeated three times to repair the possible defects.

Gas Permeation Measurements
The supported ionic liquid membrane was prepared as follows. First, 0.15 g [emim] [BF4] was dropped onto the surface of the mesoporous membrane, then introduced into the mesopores of the phenolic resin on the Al2O3 support under a vacuum of 0.03 MPa, forming ionic liquid confined in the mesoporous polymer membrane on the Al2O3 support. The excess ionic liquid was carefully wiped off using a filter paper. Figure 8 shows the schematic flowcharts of the membrane permeation measurement. The system consists of a feed cell and a permeate cell separated by an SILM, which was sealed by O-rings. CO2 and N2 were fed with a flow rate of 6 mL/min, while CH4 was always fed to the permeate side as a sweep gas with a flow rate of 1 mL/min under atmospheric pressure. CH4, CO2, and N2 flow rates were controlled by mass-flow controllers (Bronkhorst, Ruurlo, The Netherlands). Trans-membrane pressure was monitored using a pressure gauge at the feed side. The gases in the permeation side included the sweep gas and the permeated gas. The flow rate of the gas at the outlet in the permeation side was measured by a bubble flow meter, while gas composition at the outlet was determined by a gas chromatograph (Agilent 7890, Agilent Technology, Santa Clara, CA, USA) equipped with a thermal conductivity detector (TCD). The selectivity of CO2/N2 is defined as the ratio of permeance of CO2 to that of N2. The ideal selectivity of CO2/N2 can be calculated by Equation (1):

Gas Permeation Measurements
where and are the volumetric flow rates of the permeate side, which are composed of membrane permeance and sweep gas and are determined by a bubble flow meter with the feed of CO2 and N2, respectively. yi is the fraction of the component i in the permeate side determined by the gas chromatograph (i = CO2 and N2).
The actual selectivity of CO2/N2 is computed by Equation (2): where xi is the fraction of the component i in the feed side determined by the gas chromatograph (i = CO2 and N2). The selectivity of CO 2 /N 2 is defined as the ratio of permeance of CO 2 to that of N 2 . The ideal selectivity of CO 2 /N 2 can be calculated by Equation (1): where J CO 2 and J N 2 are the volumetric flow rates of the permeate side, which are composed of membrane permeance and sweep gas and are determined by a bubble flow meter with the feed of CO 2 and N 2 , respectively. y i is the fraction of the component i in the permeate side determined by the gas chromatograph (i = CO 2 and N 2 ). The actual selectivity of CO 2 /N 2 is computed by Equation (2): where x i is the fraction of the component i in the feed side determined by the gas chromatograph (i = CO 2 and N 2 ).

Characterization
The Fourier-transform infrared (FTIR) spectra of the Al 2 O 3 support, the mesoporous polymer membrane on the Al 2 O 3 support, [emim] [BF 4 ] confined in the mesoporous polymer membrane on the Al 2 O 3 support, and pure ionic liquid [emim] [BF 4 ] were obtained on a Nicolet iN10 spectrophotometer (Thermo fisher Scientific, Waltham, MA, USA). A KBr disc was used as a specimen substrate to collect the FTIR spectrum of pure ionic liquid [emim] [BF 4 ]. Scanning electron microscopy (SEM) images and elemental mapping spectra were carried out on Hitachi S-4800 microscopy (Hitachi, Tokyo, Japan). Transmission electron microscopy (TEM) images were performed on an H-7650 microscope (Hitachi, Tokyo, Japan) under an accelerate voltage of 100 kV.

Conclusions
In summary, a novel supported ionic liquid membrane with good stability was developed by confining ionic liquid [emim][BF 4 ] in a mesoporous polymer membrane on an Al 2 O 3 support. The supported ionic liquid membrane exhibits a high ideal CO 2 /N 2 selectivity of 40 based on the permeances of pure CO 2 and pure N 2 , and an actual selectivity of around 25 in a mixed CO 2 /N 2 gas at a trans-membrane pressure difference of 2.5 bar. In contrast to ionic liquid [emim] [BF 4 ] confined in a macroporous PES membrane, the ionic liquid confined in a mesoporous polymer membrane with a uniform pore size of 7 nm exhibits good stability, which was steadily operated for more than 40 h under a trans-membrane pressure difference of 1.5 bar in a mixed CO 2 /N 2 gas.