Supported Ionic Liquid Membranes and Ion-Jelly® Membranes with [BMIM][DCA]: Comparison of Its Performance for CO2 Separation

In this work, a supported ionic liquid membrane (SILM) was prepared by impregnating a PVDF membrane with 1-butyl-3-methylimidazolium dicyanamide ([BMIM][DCA]) ionic liquid. This membrane was tested for its permeability to pure gases (CO2, N2 and O2) and ideal selectivities were calculated. The SILM performance was also compared to that of Ion-Jelly® membranes, a new type of gelled membranes developed recently. It was found that the PVDF membrane presents permeabilities for pure gases similar or lower to those presented by the Ion-Jelly® membranes, but with increased ideal selectivities. This membrane presents also the highest ideal selectivity (73) for the separation of CO2 from N2 when compared with SILMs using the same PVDF support but with different ionic liquids.


Introduction
Supported Ionic Liquid Membranes (SILMs) have attracted great attention in the past years, due to their ease of preparation and versatility. SILMs can be prepared by impregnating a porous membrane OPEN ACCESS the membrane without significant loss of ionic liquid, while presenting permeabilities to pure gases (H2, N2, O2, CO2 and CH4) similar to those obtained with SILMs using the corresponding IL, although with lower ideal selectivities.
In this work we have prepared a SILM based on a PVDF flat sheet membrane and 1-butyl-3-methylimidazoilum dicyanamide ([BMIM][DCA]) ionic liquid and studied its application as a membrane for the separation of gases (CO2, N2 and O2). Furthermore, we compare its performance, in terms of permeability and selectivity, with those obtained with Ion-Jelly ® membranes [23].

Results and Discussion
The permeabilities of CO2, N2 and O2 through the PVDF [BMIM][DCA] membrane as well as the calculated ideal selectivities for these gases are presented in Table 1. In Table 2 are presented some properties of the IL [BMIM][DCA] to help in the following discussion. The SILM presents a CO2 permeability two orders of magnitude higher than to the other gases, which may be due to its high solubility in [BMIM][DCA] when compared with the other gases [29]. Consequently, the calculated ideal selectivities are higher for separating CO2 from N2 or O2 than for separating O2 from N2, with the highest ideal selectivity being that between CO2 and N2.   Table 1 also presents the results obtained in a previous work [23] for an Ion-Jelly ® membrane prepared using the same ionic liquid ([BMIM][DCA]), for comparison. We can observe that the PVDF SILM presents lower N2 and O2 permeability values than the Ion-Jelly ® one. This difference may be due to the large amount of water in the Ion-Jelly membrane ® when compared with the SILM. Regarding the ideal selectivity results, the presence of water induces a less selective microenvironment for the transport of gases, and as a result a lower ideal selectivity for the gases in the Ion-Jelly ® membrane was observed.
In Figure 1 are plotted the gas permeabilities against the Lennard-Jones diameter of the gas molecules. We can observe that the PVDF SILM presents the same trend in permeability as that previously observed for the Ion-Jelly ® membrane, with O2 presenting a higher permeability than N2, due to its smaller size, while CO2 presents the highest permeability value, despite its larger size, due to its high solubility in the ionic liquid used.  [4]. When comparing the ideal selectivities between CO2 and N2 obtained in both works, it is possible to observe that [BMIM][DCA] presents the highest ideal selectivity of all the ILs tested (  [29], at 323K), which corresponds to a lower solubility value, the CO2 diffusion coefficient is higher. Morgan et al. [9] developed a correlation for the diffusivity of gases in imidazolium ionic liquids at 303.15 K which is shown in Equation (1): where D12 is the diffusivity of gases in RTILs (cm 2 ·s −1 ); µ2 is the RTIL viscosity (mPa·s) at 303.15 K; and V1 is the gas molar volume (cm 3 ·mol −1  Table 4. It can be noticed that they are always higher for BMIMDCA, due to its lower viscosity.   [31] obtained a permeability value of 200 barrer for CO2 and an ideal selectivity of 78.9 for the CO2/N2 separation, values which are similar to ours in the case of the ideal selectivity but higher in the case of the permeability. This difference in permeability could be a result of differences in the water content of the ionic liquids used to prepare the membranes or in the quantity of IL effectively impregnated in the membrane.

Preparation of Membranes
A disk was cut from PVDF flat sheet membrane, and was placed in the bottom of a high pressure stainless steel vessel. 1 mL of [BMIM][DCA] ionic liquid was spread on top of the membrane, and the vessel was closed. CO2 was added to the vessel, at a pressure of 0.2 MPa, in order to force the ionic liquid to flow into the pores of the membrane. After one hour the vessel was opened and the excess ionic liquid remaining on top of the membrane was gently wiped with absorbing tissue. To make sure there was no remaining ionic liquid, the membrane was set upright and left to drip overnight.

Single Gas Permeabilities
The pure gas permeability of the membrane for N2, O2 and CO2 was determined by using the experimental apparatus shown in Figure 3. This rig is composed by a stainless steel cell with two identical compartments separated by the membrane. The effective membrane area was 4 cm 2 . Each individual gas permeability was evaluated by pressurizing both compartments (feed and permeate) with the pure gas, and after opening the permeate outlet, establishing a driving force of around 0.07 MPa between the feed and the permeate compartments. The pressure change in both compartments over time was followed using two pressure transducers (Druck, PDCR 910 models 99166 and 991675, Leicester, UK). All measurements were performed at a constant temperature of 303 K by using a thermostatic bath (Julabo, Model EH, Seelbach, Germany). The measurements are reproducible and have an average standard error of 5%.
The permeability of a pure gas through the membrane was calculated from the pressure data measured over time on both compartments (feed and permeate) according to the following equation [33]: where pfeed and pperm are the pressures in the feed and permeate compartments (Pa), respectively; P is the membrane permeability (m 2 ·s −1 ); t is the time (s); and l is the membrane thickness (m). The indicator 0 refers to the conditions at t = 0. The geometric parameter β (m −1 ) is characteristic of the geometry of the cell and is given by: where A is the membrane area (m 2 ) and Vfeed and Vperm are the volumes of the feed and permeate compartments (m 3 ), respectively. The β value calculated in this way for the test cell used in this work was 41.97 m −1 . The data can be plotted as 1/β ln(∆p /∆p) versus t/l, and the gas permeability is obtained from the slope of this representation. The ideal selectivity (αA/B) is the ratio of the permeabilities of two different pure gases (A and B) through a given membrane.

Conclusions
A supported ionic liquid membrane was prepared by impregnating [BMIM][DCA] ionic liquid in a PVDF flat sheet membrane, and its permeability to CO2, N2 and O2 was tested and ideal selectivities were calculated. The performance of this membrane was compared to that of Ion-Jelly ® membranes and SILMs using PVDF impregnated with other ILs studied in previous works.
The permeability to CO2 obtained with this membrane is similar to that obtained with an Ion-Jelly ® membrane using the same IL (145 and 120 barrer, respectively) but one order of magnitude lower for N2 and O2. Consequently, the calculated ideal selectivities are higher for the PVDF membrane impregnated with [BMIM][DCA] studied in this work. The worse performance of the Ion-Jelly ® membrane can be a consequence of its higher content in water.
When comparing this membrane with similar membranes using PVDF impregnated with different ILs, it was found that [BMIM][DCA] presents the highest ideal selectivity for the separation of CO2 from N2 (73), due to a higher diffusion coefficient of CO2 in this IL.