Article pubs.acs.org/JPCC
Ionic Liquid Membranes Supported by Hydrophobic and Hydrophilic Metal−Organic Frameworks for CO2 Capture Krishna M. Gupta, Yifei Chen, and Jianwen Jiang* Department of Chemical and Biomolecular Engineering, National University of Singapore, 117576 Singapore S Supporting Information *
ABSTRACT: Atomistic simulation is performed to investigate CO2 capture in ionic liquid (IL) membranes supported on metal−organic frameworks (MOFs). The IL considered is 1-n-butyl-3-methylimidazolium thiocyanate [BMIM][SCN], while hydrophobic ZIF-71 and hydrophilic Na-rho-ZMOF with the same topology and similar pore size are used as supports. The [SCN]− anion prefers to locate near the metal cluster of ZIF-71 and the Na+ ion of Na-rho-ZMOF, in contrast to the bulky and chainlike [BMIM]+ cation that resides in the open cage. In both membranes, the [SCN]− interacts more strongly than the [BMIM]+ with the MOF supports. With regard to CO2 capture from CO2/N2 mixture, CO2 adsorption is greater than N2, while CO2 diffusion is slower in both membranes, particularly in [BMIM][SCN]/ ZMOF because the Na+ ions in Na-rho-ZMOF act as strong binding sites for CO2. The permselectivity of CO2 over N2 is governed by adsorption selectivity, as diffusion selectivity remains a constant over the pressure range examined. Compared to many polymer membranes and polymer-supported ILs, [BMIM][SCN]/ZMOF exhibits higher permeability and permselectivity, and also surpasses the Robeson’s upper bound. On the basis of the two MOF-supported [BMIM][SCN] membranes examined for CO2 capture, the simulation study suggests that hydrophilic support is superior to the hydrophobic counterpart. imidazolium-based ILs.7 For CO2/N2 and CO2/CH4 mixtures, Bara et al. found the ideal solubility selectivities in imidazoliumbased ILs with one, two, or three oligo(ethylene glycol) substituents are 30−75% higher than in alkyl counterparts.8 Shi and Maginn computed the sorption of CO2 and gas mixtures in 1-n-hexyl-3-methylimidazolium bis(trifluoromethylsulfonyl)imide.9 To screen ILs for CO2 capture, Zhang et al. implemented the COSMO-RS method to calculate the Henry’s constants of CO2 in 408 ILs, and [FEP]−-based ILs were identified to be promising candidate.10 The COSMO-RS method was also used by Gonzalez-Miquel et al. to estimate CO2/N2 selectivity in 224 ILs, and [SCN]−-based ILs were predicted to outperform over other counterparts.11 Furthermore, as introduced by Bates et al.,12 task-specific (functionalized) ILs have been proposed for CO2 capture.13,14 Membrane-based separation possesses advantages in terms of energy efficiency, capital cost, separation performance, and continuous operating mode.15 Consequently, ILs have been developed as supported IL membranes (SILMs) for CO2 capture.15,16 Compared to neat ILs, SILMs not only reduce the cost and viscosity of ILs, but also enhance mechanical strength and separation efficiency. Noble and co-workers examined the performance of poly(ether sulfone) SILMs for CO2/N2 and CO2/CH4 separation and observed the ideal selectivity versus permeability exceeds the Robeson’s upper
1. INTRODUCTION Combustion of fossil fuels (coal, oil, and natural gas) is the key contributor to ∼80% CO2 emissions worldwide.1 The emissions are expected to increase because cheap and abundant fossil fuels will continue to be a substantial fraction of energy portfolio. The International Panel on Climate Change (IPCC) predicts that the atmospheric CO2 would increase up to 570 ppm in 2100, leading to a global temperature rise of 1.9 °C and a sea level increase of 38 cm.2 Consequently, CO2 capture from emissions has become a prime issue, and there is an urgent need to reduce the carbon footprint for environmental protection and sustainable development. Several techniques such as amine scrubbing, cryogenic distillation, sorbent adsorption, and membrane separation have been proposed for CO2 capture. Among these, ionic liquids (ILs) have received considerable interest. As a unique class of organic slats, ILs possess melting point lower than 100 °C, negligible volatility, nonflammability, and high thermal stability.3 The first study using ILs for CO2 capture was attempted in 1999.4 Since then, several experimental and theoretical studies have been performed in this area. For example, Brennecke and co-workers demonstrated that anions play a dominant role in CO2 dissolution, particularly the hierarchy of CO2 solubility in methylimidazolium-based ILs follows [NO3]− < [DCA]− < [BF4]− ∼ [PF6]− < [TfO]− < [Tf2N]− < [methide]−.5 In a separate experiment, they found fluoroalkyl chains in either cation or anion could improve CO2 solubility in ILs.6 On the basis of simulation, Cadena et al. observed that CO2 is strongly associated with [PF6]− in © 2013 American Chemical Society
Received: December 17, 2012 Revised: February 25, 2013 Published: February 26, 2013 5792
dx.doi.org/10.1021/jp312404k | J. Phys. Chem. C 2013, 117, 5792−5799
The Journal of Physical Chemistry C
Article
with parameters from the universal force field (UFF).28 A number of simulation studies have demonstrated that UFF can accurately predict adsorption and diffusion in various MOFs.29 Figure S3 and Table S1 give the atomic charges and LJ parameters of the framework atoms in ZIF-71 and rho-ZMOF, respectively. Figure 1 shows the molecular structure of [BMIM][SCN]. The atomic charges were adopted from our previous studies.24
bound.17 Gas permeation was measured by Neves et al. in 1-nalkyl-3-methylimidazolium ILs supported on different (hydrophilic and hydrophobic) poly(vinylidene fluoride) (PVDF), and hydrophobic PVDF-SILMs were found to be more stable with higher CO2 affinity.18 Recently, Close et al. prepared SILMs using alumina supports of different pore sizes and tested for CO2/N2 separation.19 All studies to date have used polymeric and inorganic supports to prepare SILMs.20,21 Emerged as a special class of porous materials, metal−organic frameworks (MOFs) could be potential supports for ILs because the pore size, volume, and functionality of MOFs are readily tunable in a rational manner.22 In this perspective, we proposed IRMOF-1 as a support for [BMIM][PF6] membrane and computationally examined its performance for postcombustion CO2 capture.23 In addition, different [BMIM]+-based IL membranes supported on IRMOF-1 were examined, with a focus on anion effects.24 The ILs considered were [BMIM][BF4], [BMIM][PF6], [BMIM][SCN], and [BMIM][Tf 2 N]; [BMIM][SCN]/ IRMOF-1 was predicted to be superior to other SILMs in CO2 permeability/selectivity and surpass the Robeson’s upper bound. In this study, we further conduct molecular simulation to investigate CO2 capture in MOF-supported IL membranes. While [BMIM][SCN] is used as in our previous study,24 two MOFs are considered here, namely ZIF-71 and Na-rho-ZMOF. They possess the same topology and similar pore size, but ZIF71 is hydrophobic and Na-rho-ZMOF is hydrophilic. Therefore, molecular insight into the role of hydrophobic/hydrophilic framework in MOF-supported IL membranes can be provided. Following this section, the models and simulation methods are briefly described in section 2. In section 3, the structural properties of [BMIM][SCN] in ZIF-71 and Na-rho-ZMOF are presented. Thereafter, the performance of CO2/N2 separation in [BMIM][SCN]/ZIF-71 and [BMIM][SCN]/ZMOF membranes is discussed, followed by the concluding remarks in section 4.
Figure 1. Molecular structure of [BMIM][SCN].
Figure S4 and Table S2 list the atomic types and charges in [BMIM][SCN]. In addition, the bonded (stretching, bending, and torsional) and LJ potentials were included and represented by the Amber force field.30 The LJ parameters in [BMIM][SCN] are listed in Table S3. Two [BMIM][SCN] membranes were examined: one supported on ZIF-71 and the other on Narho-ZMOF. In each membrane, the weight ratio of [BMIM][SCN] to MOF was 0.15. On this basis, the desired number of [BMIM][SCN] was added randomly into the pore of each MOF. Then, the confined [BMIM][SCN] was subjected to energy minimization with the steepest descent method. Finally, molecular dynamics (MD) simulation was performed in a canonical (NVT) ensemble to further equilibrate the confined [BMIM][SCN]. Temperature was maintained at 300 K by velocity-rescaled Berendsen thermostat with a relaxation time of 0.1 ps.31 The LJ interactions were evaluated with a cutoff of 14 Å, and the electrostatic interactions were calculated using particle-mesh Ewald method32 with a grid spacing of 1.2 Å and a fourth-order interpolation. The equations of motion were integrated with a time step of 2 fs by the Leapfrog algorithm.33 Total MD simulation duration was 20 ns, in which the last 10 ns was used for analysis. Gromacs 4.5.3 was implemented for the minimization and MD simulation.34 The MOF structures were assumed to be rigid during the simulation. Figure 2 illustrates the equilibrated snapshots of [BMIM][SCN]/ZIF-71 and [BMIM][SCN]/ZMOF membranes. It can be seen that [BMIM][SCN] in each MOF is mostly located in the open αcage and proximal to the framework. Particularly, [SCN]− anion in Na-rho-ZMOF is preferentially bound onto Na+ ions. The detailed structural properties of [BMIM][SCN] in the membranes will be discussed below. 2.2. Adsorption, Diffusion and Permeation of CO2/N2 Mixture. To evaluate the separation performance of [BMIM][SCN]/ZIF-71 and [BMIM][SCN]/ZMOF membranes, the adsorption, diffusion, and permeation of CO2/N2 mixture were simulated at 300 K. Specifically, the grand canonical Monte Carlo (GCMC) method was used to evaluate the adsorption at different pressures with a modified version of Bigmac.35 The composition of CO2/N2 in bulk phase was assumed to be
2. MODELS AND METHODS 2.1. [BMIM][SCN] Membranes Supported on MOFs. Two different MOFs (ZIF-71 and Na-rho-ZMOF) are considered as shown in Figure S1 of the Supporting Information. Both MOFs possess rho-type topology. ZIF-71 25 has a space group of Pm3m ̅ and a lattice constant of 28.554 Å. One unit cell of ZIF-71 contains a truncated cuboctahedron (αcage) with 48 Zn atoms. Each Zn atom is coordinated with four N atoms of 4,5-dicholoroimidazolate ligands to form fourcoordinated molecular building block. Na-rho-ZMOF has a space group of Im3̅m and a lattice constant of 31.062 Å.26 It was synthesized by metal−ligand-directed assembly of In atoms and 4,5-imidazoledicarboxylic acid (H3ImDC). Each In atom is coordinated to four N atoms and four O atoms of four separate doubly deprotonated H3ImDC ligands, respectively, to form eight-coordinated molecular building block. Both MOFs consist of truncated cuboctahedra (α-cages) that are connected by double eight-member ring (D8MR). Figure S2 illustrates the pore morphologies and radii in ZIF-71 and Na-rho-ZMOF.27 The cage radii are approximately 8.4 and 9.1 Å, while the window radii are 2.4 and 2.8 Å in ZIF-71 and Na-rho-ZMOF, respectively. The atomic charges of ZIF-71 and rho-ZMOF framework atoms were calculated by density-functional theory as described in our previous study.27 The dispersion interactions were represented by Lennard-Jones (LJ) potential 5793
dx.doi.org/10.1021/jp312404k | J. Phys. Chem. C 2013, 117, 5792−5799
The Journal of Physical Chemistry C
Article
Figure 2. (a) [BMIM][SCN]/ZIF-71 and (b) [BMIM][SCN]/ZMOF membranes at a weight percentage WIL/MOF = 0.15. C of [BMIM]+ and framework: gray; C of [SCN]−: cyan; S: yellow; N: blue; Zn: orange; Cl: light green; In: green; Na+: purple; O: red; H: white. The metal clusters are shown as polyhedra.
Figure 3. Radial distribution functions of (a) [SCN]− and (b) [BMIM]+ around the Zn, Cl, and N atoms of ZIF-71 in [BMIM][SCN]/ZIF-71.
0.15:0.85, representing a typical flue gas. CO2 was mimicked as a there-site model, and its intrinsic quadrupole moment was described by a partial charge model.36 The partial charges on C and O atoms were qC = 0.576e and qO = −0.288e (e = 1.6022 × 10−19 C), with C−O bond length of 1.18 Å and bond angle ∠OCO of 180°. N2 was represented as a two-site model with the LJ potential parameters fitted to experimental bulk properties.37 Table S4 lists the LJ parameters of CO2 and N2. The Lorentz−Berthelot combining rules were used to evaluate cross-interaction parameters. To evaluate the LJ interactions, a spherical cutoff of 15 Å was used, while the electrostatic interactions were calculated using Ewald sum. The real/ reciprocal space partition parameter and the cutoff for reciprocal lattice vectors were 0.2 Å−1 and 8, respectively, to ensure convergence. In the GCMC simulation, the number of trail move was 2 × 107. The first 107 moves were used for equilibration and the subsequent 107 moves for ensemble averages. Five types of trail moves were randomly attempted, including displacement, rotation, partial regrowth at a neighboring position; complete regrowth at a new position; swap with reservoir. Their probabilities were 0.1, 0.1, 0.1, 0.1, and 0.6, respectively. Thereafter, MD simulation was conducted to estimate the diffusion of CO2/N2 mixture in each membrane. The initial configuration for the MD simulation was from the GCMC simulation. At each pressure, the MD simulation duration was 50 ns with the last 40 ns for analysis. The solubility and diffusion coefficients of CO2 and N2 were calculated from the GCMC and MD simulations. Subsequently, the permeabilities were calculated and compared to the
reported values in other membranes. Finally, the adsorption, diffusion, and permeation selectivities for the separation of CO2/N2 mixture were estimated.
3. RESULTS AND DISCUSSION 3.1. Structures of [BMIM][SCN] in Membranes. The structures of [BMIM][SCN] in the two MOF-supported membranes are characterized by radial distribution function g(r) gij(r ) =
Nij(r , r + Δr )V 4πr 2ΔrNN i j
(1)
where r is the distance between atoms i and j, Nij(r,r+Δr) is the number of atom j around i within a shell from r to r + Δr, V is the system volume, and Ni and Nj are the numbers of atoms i and j, respectively. The centers of masses of [BMIM]+ and [SCN]− were considered respectively to calculate the g(r). Figure 3a shows the g(r) of [SCN]− anion around the Zn, Cl, and N atoms of ZIF-71 in [BMIM][SCN]/ZIF-71 membrane. Sharp peaks are observed, particularly around the Zn atom at r = 4.2−4.5 Å. The reason is [SCN]− has a negative charge and interacts favorably with the positively charged Zn atom. The Cl atom is present in the open α-cage and more easily accessible than the N atom by [SCN]−; consequently, higher peaks are seen in the g(r) around the Cl atom than N atom. As shown in Figure 3b, the g(r) of [BMIM]+ cation appears at r = 3.2 Å around the Cl atom but at r > 4.8 Å around the Zn and N atoms. This is attributed to bulky and chainlike structure of 5794
dx.doi.org/10.1021/jp312404k | J. Phys. Chem. C 2013, 117, 5792−5799
The Journal of Physical Chemistry C
Article
[BMIM]+, which is preferentially reside in the α-cage (close to the Cl atom) rather than the metal cluster (ZnN4). Comparing Figures 3a and 3b, we can infer that [SCN]− has a stronger interaction than [BMIM]+ with ZIF-71 framework. Figure 4 shows the g(r) of cation−anion, anion−anion, and cation−cation in [BMIM][SCN]/ZIF-71 membrane. A pro-
is seen at r > 4.0 Å in the g(r) of [BMIM]+−Na+. The g(r) of [BMIM]+−O and [BMIM]+−N appear at shorter distance compared to [BMIM]+−Na+ and [BMIM]+−In. This indicates [BMIM]+ is proximal to the organic linker rather than the metal cluster and Na+ ion. With a chainlike structure, [BMIM]+ prefers to reside in the α-cage because small corner near the metal cluster cannot accommodate [BMIM]+. As also observed in ZIF-71 (Figure 3), [SCN]− interacts more strongly with Narho-ZMOF compared to [BMIM]+. Figure 6 plots the g(r) of cation−anion, anion−anion, and cation−cation in [BMIM][SCN]/ZMOF membrane. A high
Figure 4. Radial distribution functions of cation−anion, anion−anion, and cation−cation in [BMIM][SCN]/ZIF-71.
nounced peak is observed at r = 4.0 Å for [BMIM]+−[SCN]¯. The g(r) of [SCN]−−[SCN]− and [BMIM]+−[BMIM]+ are much lower than that of [BMIM]+−[SCN]¯ due to unfavorable electrostatic repulsion. As a comparison, the g(r) in bulk phase of [BMIM][SCN] are shown in Figure S5. Upon confined in ZIF-71, the magnitude of g(r) is elevated as attributed to geometric constraint. Particularly, [BMIM]+−[SCN]− in the membrane exhibits the largest elevation, whereas the peak of g(r) for either [SCN]−−[SCN]− or [BMIM]+−[BMIM]+ is also elevated from bulk phase. The peak position is generally shifted to a longer distance compared to bulk phase, indicating the distance of [SCN]−−[SCN]− or [BMIM]+−[BMIM]+ increases in the membrane. This is because each ion is surrounded by more counterions than in bulk phase, a consequence of confinement effect. A similar finding was also observed in our previous study for IMROF-1 supported different IL membranes.24 Figure 5 plots the g(r) of [SCN]− and [BMIM]+ around the In, O, and Na atoms, and Na+ ion of Na-rho-ZMOF in [BMIM][SCN]/ZMOF membrane. The g(r) of [SCN]−−Na+ exhibits a sharp peak at r = 2.4 Å because Na+ ions act as strong binding sites for [SCN]−. In contrast, other framework atoms interact much more weakly with [SCN]− and substantially lower g(r) are observed at r > 3.2 Å. For [BMIM]+, a low peak
Figure 6. Radial distribution functions of cation−anion, anion−anion, and cation−cation in [BMIM][SCN]/ZMOF.
peak is seen in the g(r) of [SCN]−−[SCN]− at 3.6 Å, remarkably different from that in [BMIM][SCN]/ZIF-71 membrane. The reason is Na+ ions are preferential binding sites for [SCN]− as depicted in Figure 5a; thus, [SCN]− anions mostly reside near Na+ ions and are close to each other. Another consequence is that [BMIM]+ and [SCN]− are largely separated with a low peak in the g(r) of [BMIM]+−[SCN]−, which is less pronounced compared to those in [BMIM][SCN]/ZIF-71 (Figure 4) and in bulk [BMIM][SCN] (Figure S5). 3.2. Separation of CO2/N2 Mixture. Figure 7 shows the adsorption isotherms of CO2/N2 mixture as a function of total pressure in [BMIM][SCN]/ZIF-71 and [BMIM][SCN]/ ZMOF membranes. Apparently, the extent of CO2 adsorption is greater than N2 at any given pressure because CO2 is a threesite molecule and has a stronger interaction than N2 with the membrane. Compared to [BMIM][SCN]/ZIF-71, [BMIM][SCN]/ZMOF has a greater adsorption capacity for CO2. The reason is that Na+ ions in Na-rho-ZMOF are strong binding
Figure 5. Radial distribution functions of (a) [SCN]− and (b) [BMIM]+ around the In, O, N, and Na+ of Na-rho-ZMOF in [BMIM][SCN]/ZMOF. 5795
dx.doi.org/10.1021/jp312404k | J. Phys. Chem. C 2013, 117, 5792−5799
The Journal of Physical Chemistry C
Article
Figure 7. Adsorption isotherms of CO2/N2 mixture (15:85) in (a) [BMIM][SCN]/ZIF-71 and (b) [BMIM][SCN]/ZMOF.
Figure 8. Radial distribution functions of CO2 around the S and N atoms of [SCN]−, the N1 and N2 atoms of [BMIM]+, (a) the Zn and Cl atoms of ZIF-71 in [BMIM][SCN]/ZIF-71, and (b) the In and Na atoms of ZMOF in [BMIM][SCN]/ZMOF. The total pressure of CO2/N2 mixture is 100 kPa.
Figure 9. Diffusion coefficients of CO2/N2 mixture (15:85) in (a) [BMIM][SCN]/ZIF-71 and (b) [BMIM][SCN]/ZMOF.
phenomenon is observed in [BMIM][SCN]/ZMOF membrane (Figure 8b). A very sharp peak is seen at 3.7 Å in the g(r) around Na+ ions of Na-rho-ZMOF, revealing Na+ ions are the most favorable sites. The S and N atoms of [SCN]− exhibiting lower peaks between 3.6 and 4.6 Å can be considered as the secondary favorable sites. The diffusion coefficients of CO2 and N2 in [BMIM][SCN]/ ZIF-71 and [BMIM][SCN]/ZMOF membranes were calculated from
sites for CO2 as further explained below. As a result, N2 adsorption in [BMIM][SCN]/ZMOF is weak and even lower than that in [BMIM][SCN]/ZIF-71. To identify favorable adsorption sites for CO2, Figure 8 shows the g(r) for the carbon atom of CO2 around the anion and cation of IL as well as the framework atoms in the two membranes. In [BMIM][SCN]/ZIF-71 membrane (Figure 8a), a sharp peak is observed at 3.2 Å in the g(r) around the S atom of [SCN]−. Though the N atom of [SCN]− also exhibits a peak at 3.2 Å, the peak height is lower. The g(r) around the Zn atom of ZIF-71 has pronounced and broad peak at 5.5−6.5 Å. For [BMIM]+, the peaks around the N1 and N2 atoms are lower and at longer distance. Therefore, [SCN]− and Zn atom are the most favorable sites for CO 2 adsorption. A different
N
1 d D= lim ∑ ⟨|ri(t ) − ri(0)|2 ⟩ 6N dt t →∞ i = 1 5796
(2)
dx.doi.org/10.1021/jp312404k | J. Phys. Chem. C 2013, 117, 5792−5799
The Journal of Physical Chemistry C
Article
where N is the number of gas molecules of the same type and ri(t) is the position of the ith molecule at time t. Figure 9 plots the diffusion coefficients of CO2/N2 mixture at different total pressures in the two membranes. In each membrane, DCO2 is smaller than DN2 because CO2 has a larger molecular weight and a stronger interaction with the membrane. With increasing pressure, DN2 decreases because of enhanced steric hindrance as more gas molecules are adsorbed. Nevertheless, DCO2 increases due to screening effect. This trend of CO2 diffusion versus pressure was also observed in neat ZIFs.38 With a large quadrupole moment, CO2 is strongly bound onto ZIF-71 or Na+ ions of Na-rho-ZMOF. With increasing pressure, the most favorable sites are gradually occupied (screened) and CO2 is adsorbed onto less favorable sites; consequently, CO 2 experiences relatively weaker interaction and hence DCO2 increases. The magnitude of DCO2 in [BMIM][SCN]/ZIF-71 is 10−5 cm2/s, 1 order of magnitude smaller than in neat ZIFs.38 This is primarily attributed to the reduction of free volume in ZIF-71 upon adding [BMIM][SCN]. Nevertheless, DCO2 here is significantly larger than 10 −8 cm 2 /s in polyimide membranes.39 Compared to [BMIM][SCN]/ZIF-71, DCO2 and DN2 in [BMIM][SCN]/ZMOF are 1 order of magnitude smaller. For CO2, this is because of the strong interaction with Na+ ions. For N2, the reason is primarily the steric hindrance as more molecules are adsorbed in [BMIM][SCN]/ZMOF than in [BMIM][SCN]/ZIF-71 (Figure 7). Based on adsorption and diffusion, the permeability of gas i was calculated by Pi = SiDi, in which Si is the solubility coefficient (ci/pi, ci is loading and pi is partial pressure) estimated from the isotherms in Figure 7. Tables 1 and 2 list
membrane; it decreases with increasing pressure because adsorption occurs in less favorable sites at high pressures. For CO2, S decreases more rapidly than the increase of DCO2; thus, PCO2 also decreases. Nevertheless, both S and D for N2 decrease with increasing pressure leading to decrease in PN2. It is worthwhile to compare PCO2 in the two membranes with other materials. For example, compared to IRMOF-1 supported [BMIM][SCN] membrane,24 PCO2 is larger in [BMIM][SCN]/ ZIF-71 but smaller in [BMIM][SCN]/ZMOF. Both membranes here exhibit larger PCO2 than several polymer-supported IL membranes at 1 bar.40,41 In poly(ether sulfone)-supported [EMIM][B(CN)4] membrane, PCO2 was measured to be 2040 barrer,42 which is nearly 21 and 14 times smaller than those in [BMIM][SCN]/ZIF-71 and [BMIM][SCN]/ZMOF (at 1 bar), respectively. The separation performance of both membranes for CO2/N2 mixture is quantified by permeation selectivity PCO2 PN2
gas
solubility coeff
diffusion coeff
permeability
100
CO2 N2 CO2 N2 CO2 N2
0.4752 0.0326 0.4040 0.0288 0.2967 0.0216
9.01 24.60 10.20 22.22 11.79 18.78
42795 8042 41193 6406 34971 4057
200 600
Table 2. Solubility Coefficients [cm3 (STP)/cm3 (membrane) (cmHg)], Diffusion Coefficients [10−6 cm2/s], and Permeabilities (barrer) of CO2 and N2 in [BMIM][SCN]/ZMOF press. (kPa)
gas
solubility coeff
diffusion coeff
permeability
100
CO2 N2 CO2 N2 CO2 N2
4.9971 0.0162 2.9860 0.0142 1.2494 0.0104
0.55 2.28 0.83 2.13 0.93 1.05
27584 370 24814 302 11470 118
200 600
SCO2 DCO2 S N2 DN2
(3)
where SCO2/SN2 is adsorption selectivity calculated from (xi/xj)/ (yi/yj) and xi and yi are the mole fractions of gas i in adsorbed and bulk phase, respectively. DCO2/DN2 is diffusion selectivity. As shown in Figure 10, SCO2/SN2 is within 13−15 in [BMIM][SCN]/ZIF-71 and within 110−310 in [BMIM][SCN]/ZMOF over the pressure range examined. Upon comparison, SCO2/SN2 is ∼25 in [BMIM][SCN]/IRMOF-1.24 With increasing pressure, SCO2/SN2 decreases due to the reduced number of favorable sites for CO2 adsorption. On the other hand, DCO2/DN2 remains mainly constant with only marginal enhancement when pressure increases. More specifically, DCO2/ DN2 lies within 0.3−0.6 and 0.2−0.9 in [BMIM][SCN]/ZIF-71 and [BMIM][SCN]/ZMOF, respectively. Therefore, PCO2/PN2 is primarily dominated by SCO2/SN2 rather than DCO2/DN2. This finding is in significantly contrast to gas separation in glassy polymer membranes where the performance is largely based on diffusion selectivity.43 Figure 11 plots the permselectivity PCO2/PN2 versus CO2 permeability PCO2 in MOF-supported IL membranes. The experimental data in polymer membranes44 and polymersupported ILs45 are included. [BMIM][SCN]/ZMOF exhibits higher PCO2/PN2 and PCO2 compared to polymer membranes and polymer-supported ILs and also surpasses the Robeson’s upper bound.44 Despite higher PCO2, [BMIM][SCN]/ZIF-71 has lower PCO2/PN2. Therefore, the simulation results suggest that Na-rho-ZMOF is a better candidate than ZIF-71 as a support for [BMIM][SCN] membrane and that [BMIM][SCN]/ZMOF might be interesting for CO2 capture from flue gas.
Table 1. Solubility Coefficients [cm3 (STP)/cm3 (membrane)(cmHg)], Diffusion Coefficients [10−6 cm2/s], and Permeabilities (barrer) of CO2 and N2 in [BMIM][SCN]/ZIF-71 press. (kPa)
=
4. CONCLUSIONS CO2 capture has been simulated in [BMIM][SCN] membranes supported on two different MOFs (hydrophobic ZIF-71 and hydrophilic Na-rho-ZMOF). Because of the confinement effect, [SCN]− and [BMIM]+ in each membrane are more packed compared to bulk phase. In addition, [SCN]− has a stronger
the solubility coefficients, diffusion coefficients, and permeabilities of CO2 and N2 in [BMIM][SCN]/ZIF-71 and [BMIM][SCN]/ZMOF membranes, respectively. Physically, S is a characteristic of interaction strength between gas and 5797
dx.doi.org/10.1021/jp312404k | J. Phys. Chem. C 2013, 117, 5792−5799
The Journal of Physical Chemistry C
Article
Figure 10. Adsorption, diffusion and permeation selectivities of CO2/N2 mixture in (a) [BMIM][SCN]/ZIF-71 and (b) [BMIM][SCN]/ZMOF.
within 0.3−0.6 and 0.2−0.9. Consequently, permselectivity PCO2/PN2 is dominated by SCO2/SN2, and the separation of CO2/ N2 mixture is driven by adsorption. Compared to polymer membranes and polymer-supported ILs, [BMIM][SCN]/ ZMOF possesses higher PCO2/PN2 and PCO2, and surpasses the Robeson’s upper bound. This study provides atomistic insight into the microscopic properties of IL and the separation of CO2//N2 mixture in IL membranes supported on MOFs and reveals that the hydrophilic support (Na-rho-ZMOF) outperforms the hydrophobic counterpart (ZIF-71) in separation performance.
■
ASSOCIATED CONTENT
* Supporting Information S
Crystal structures, pore morphologies, and radii in ZIF-71 and Na-rho-ZMOF, atomic charges in ZIF-71 and Na-rho-ZMOF, LJ parameters in ZIF-71 and Na-rho-ZMOF, atomic types and charges in [BMIM][SCN], LJ parameters in [BMIM][SCN], LJ parameters of CO2 and N2, radial distribution functions in bulk phase of [BMIM][SCN]. This material is available free of charge via the Internet at http://pubs.acs.org.
■
Figure 11. CO2/N2 permselectivity versus CO2 permeability in MOFsupported IL membranes. The red circles are experimental data in polymer membranes, and the line is Robeson’s upper bound.44 Also illustrated are the data in polymer-supported ILs.45
AUTHOR INFORMATION
Corresponding Author
*E-mail:
[email protected].
interaction than [BMIM]+ with the MOF supports. Specifically, negatively charged [SCN]− preferentially interacts with the Zn atom in ZIF-71 and the Na+ ion in Na-rho-ZMOF. However, bulky and chainlike [BMIM]+ mostly reside in the open α-cage because small corner near the metal cluster cannot accommodate [BMIM]+. At a given pressure, CO2 is more adsorbed than N2 in each membrane, particularly in [BMIM][SCN]/ZMOF as attributed to the presence of Na+ ions in ZMOF that act as strong binding sites for CO2. This also leads to smaller DCO2 than DN2. With increasing pressure, DCO2 increases because of screening effect while DN2 decreases due to steric hindrance. Furthermore, DCO2 and DN2 in [BMIM][SCN]/ZMOF are 1 order of magnitude smaller than in [BMIM][SCN]/ZIF-71. PCO 2 in both membranes is larger compared to several polymer-supported IL membranes. Over the pressure range under study, the estimated adsorption selectivity SCO2/SN2 ranges within 13−15 and 110−310 in [BMIM][SCN]/ZIF-71 and [BMIM][SCN]/ ZMOF, respectively, while diffusion selectivity DCO2/DN2 is
Notes
The authors declare no competing financial interest.
■
ACKNOWLEDGMENTS The authors are grateful for the support from the National University of Singapore (R-279-000-297-112) and the National Research Foundation of Singapore (R-279-000-261-281).
■
REFERENCES
(1) Quadrelli, R.; Peterson, S. The Energy-Climate Challenge: Recent Trends in CO2 Emissions from Fuel Combustion. Energy Policy 2007, 35, 5938−5952. (2) Stewart, C.; Hessami, M. A. A Study of Methods of CO2 Capture and Sequestration - the Sustainability of a Photosynthetic Bioreactor Approach. Energy Convers. Manage. 2005, 46, 403−420. (3) Plechkova, N. V.; Seddon, K. R. Applications of Ionic Liquids in the Chemical Industry. Chem. Soc. Rev. 2008, 37, 123−150. (4) Blanchard, L. A.; Hancu, D.; Beckman, E. J.; Brennecke, J. F. Green Processing Using Ionic Liquids and CO2. Nature 1999, 399, 28−29. 5798
dx.doi.org/10.1021/jp312404k | J. Phys. Chem. C 2013, 117, 5792−5799
The Journal of Physical Chemistry C
Article
(26) Liu, Y. L.; Kravtsov, V. C.; Larsen, R.; Eddaoudi, M. Molecular Building Blocks Approach to the Assembly of Zeolite-like MetalOrganic Frameworks with Extra-Large Cavities. Chem. Commun. 2006, 14, 1488−1490. (27) Nalaparaju, A.; Zhao, X. S.; Jiang, J. W. Molecular Understanding for the Adsorption of Water and Alcohols in Hydrophilic and Hydrophobic Zeolitic Metal-Organic Frameworks. J. Phys. Chem. C 2010, 114, 11542−11550. (28) Rappe, A. K.; Casewit, C. J.; Colwell, K. S.; Goddard, W. A.; Skiff, W. M. Uff, a Full Periodic Table Force Field for Molecular Mechanics and Molecular Dynamics Simulations. J. Am. Chem. Soc. 1992, 114, 10024−10035. (29) Jiang, J. W.; Babarao, R.; Hu, Z. Q. Molecular Simulations for Energy, Environmental and Pharmaceutical Applications of Nanoporous Materials: From Zeolites, Metal-Organic Frameworks to Protein Crystals. Chem. Soc. Rev. 2011, 40, 3599−3612. (30) Cornell, W. D.; Cieplak, P.; Bayly, C. I.; Gould, I. R.; Merz, K. M.; Ferguson, D. M.; Spellmeyer, D. C.; Fox, T.; Caldwell, J. W.; Kollman, P. A. A Second Generation Force Field for the Simulation of Proteins, Nucleic Acids, and Organic Molecules. J. Am. Chem. Soc. 1995, 117, 5179−5197. (31) Berendsen, H. J. C.; Postma, J. P. M.; Vangunsteren, W. F.; Dinola, A.; Haak, J. R. Molecular-Dynamics with Coupling to an External Bath. J. Chem. Phys. 1984, 81, 3684−3690. (32) Darden, T.; York, D.; Pedersen, L. Particle Mesh Ewald - An N.logN Method for Ewald Sums in Large Systems. J. Chem. Phys. 1993, 98, 10089−10092. (33) Hockney, R. W. Potential Calculation and Some Applications. Methods Comput. Phys. 1970, 9, 136−211. (34) van Der Spoel, D.; Lindahl, E.; Hess, B.; Groenhof, G.; Mark, A. E.; Berendsen, H. J. C. Gromacs: Fast, Flexible, and Free. J. Comput. Chem. 2005, 26, 1701−1718. (35) Vlugt, T. J. H.; Smit, B. http://molsim.chem.uva.nl/bigmac. (36) Hirotani, A.; Mizukami, K.; Miura, R.; Takaba, H.; Miya, T.; Fahmi, A.; Stirling, A.; Kubo, M.; Miyamoto, A. Grand Canonical Monte Carlo Simulation of the Adsorption of CO2 on Silicalite and NaZSM-5. Appl. Surf. Sci. 1997, 120, 81−84. (37) Murthy, C. S.; Singer, K.; Klein, M. L.; McDonald, I. R. Pairwise Additive Effective Potentials for Nitrogen. Mol. Phys. 1980, 41, 1387− 1399. (38) Liu, J.; Keskin, S.; Sholl, D. S.; Johnson, J. K. Molecular Simulations and Theoretical Predictions for Adsorption and Diffusion of CH4/H2 and CO2/CH4 Mixtures in ZIFs. J. Phys. Chem. C 2011, 115, 12560−12566. (39) Zhang, L. L.; Xiao, Y. C.; Chung, T. S.; Jiang, J. W. Mechanistic Understanding of CO 2-Induced Plasticization of a Polyimide Membrane: A Combination of Experiment and Simulation Study. Polymer 2010, 51, 4439−4447. (40) Scovazzo, P. Determination of the Upper Limits, Benchmarks, and Critical Properties for Gas Separations Using Stabilized Room Temperature Ionic Liquid Membranes for the Purpose of Guiding Future Research. J. Membr. Sci. 2009, 343, 199−211. (41) Kim, D. H.; Baek, I. H.; Hong, S. U.; Lee, H. K. Study on Immobilized Liquid Membrane Using Ionic Liquid and PVDF Hollow Fiber as a Support for CO2/N2 Separation. J. Membr. Sci. 2011, 372, 346−354. (42) Mahurin, S. M.; Hillesheim, P. C.; Yeary, J. S.; Jiang, D.; Dai, S. High CO2 Solubility, Permeability and Selectivity in Ionic Liquids with Tetracyanoborate Anion. RSC Adv. 2012, 2, 11813−11819. (43) Maier, G. Gas Separation with Polymer Membranes. Angew. Chem., Int. Ed. 1998, 37, 2960−2974. (44) Robeson, L. M. The Upper Bound Revisited. J. Membr. Sci. 2008, 320, 390−400. (45) Bara, J. E.; Carlisle, T. K.; Gabriel, C. J.; Camper, D. E.; FInotello, A.; Gin, D. L.; Noble, R. D. Guide to CO2 Separations in Imidazolium-Based Room-Temperature Ionic Liquids. Ind. Eng. Chem. Res. 2009, 48, 2739−2751.
(5) Aki, S. N. V. K.; Mellein, B. R.; Saurer, E. M.; Brennecke, J. F. High-Pressure Phase Behavior of Carbon Dioxide with ImidazoliumBased Ionic Liquids. J. Phys. Chem. B 2004, 108, 20355−20365. (6) Muldoon, M. J.; Aki, S. N. V. K.; Anderson, J. L.; Dixon, J. K.; Brennecke, J. F. Improving Carbon Dioxide Solubility in Ionic Liquids. J. Phys. Chem. B 2007, 111, 9001−9009. (7) Cadena, C.; Anthony, J. L.; Shah, J. K.; Morrow, T. I.; Brennecke, J. F.; Maginn, E. J. Why is CO2 So Soluble in Imidazolium-Based Ionic Liquids? J. Am. Chem. Soc. 2004, 126, 5300−5308. (8) Bara, J. E.; Gabriel, C. J.; Lessmann, S.; Carlisle, T. K.; FInotello, A.; Gin, D. L.; Noble, R. D. Enhanced CO2 Separation Selectivity in Oligo(Ethylene Glycol) Functionalized Room-Temperature Ionic Liquids. Ind. Eng. Chem. Res. 2007, 46, 5380−5386. (9) Shi, W.; Maginn, E. J. Atomistic Simulation of the Absorption of Carbon Dioxide and Water in the Ionic Liquid 1-n-Hexyl-3Methylimidazolium Bis(trifluoromethylsulfonyl)Imide. J. Phys. Chem. B 2008, 112, 2045−2055. (10) Zhang, X. C.; Liu, Z. P.; Wang, W. C. Screening of Ionic Liquids to Capture CO2 by COSMO-RS and Experiments. AIChE J. 2008, 54, 2717−2728. (11) Gonzalez-Miquel, M.; Palomar, J.; Omar, S.; Rodriguez, F. CO2/ N2 Selectivity Prediction in Supported Ionic Liquid Membranes by COSMO-RS. Ind. Eng. Chem. Res. 2011, 50, 5739−5748. (12) Bates, E. D.; Mayton, R. D.; Ntai, I.; Davis, J. H. CO2 Capture by a Task-Specific Ionic Liquid. J. Am. Chem. Soc. 2002, 124, 926−927. (13) Xu, A. R.; Wang, J. J.; Wang, H. Y. Effects of Anionic Structure and Lithium Salts Addition on the Dissolution of Cellulose in 1-Butyl3-Methylimidazolium-Based Ionic Liquid Solvent Systems. Green Chem. 2010, 12, 268−275. (14) Wang, C. M.; Luo, X. Y.; Luo, H. M.; Jiang, D. E.; Li, H. R.; Dai, S. Tuning the Basicity of Ionic Liquids for Equimolar CO2 Capture. Angew. Chem., Int. Ed. 2011, 50, 4918−4922. (15) Noble, R. D.; Gin, D. L. Perspective on Ionic Liquids and Ionic Liquid Membranes. J. Membr. Sci. 2011, 369, 1−4. (16) Lozano, L. J.; Godinez, C.; de los Rios, A. P.; HernandezFernandez, F. J.; Sanchez-Segado, S.; Alguacil, F. J. Recent Advances in Supported Ionic Liquid Membrane Technology. J. Membr. Sci. 2011, 376, 1−14. (17) Scovazzo, P.; Kieft, J.; Finan, D. A.; Koval, C.; DuBois, D.; Noble, R. D. Gas Separations Using Non-Hexafluorophosphate [PF6]− Anion Supported Ionic Liquid Membranes. J. Membr. Sci. 2004, 238, 57−63. (18) Neves, L. A.; Crespo, J. G.; Coelhoso, I. M. Gas Permeation Studies in Supported Ionic Liquid Membranes. J. Membr. Sci. 2010, 357, 160−170. (19) Close, J. J.; Farmer, K.; Moganty, S. S.; Baltus, R. E. CO2/N2 Separations Using Nanoporous Alumina-Supported Ionic Liquid Membranes: Effect of the Support on Separation Performance. J. Membr. Sci. 2012, 390−391, 201−210. (20) Scovazzo, P. Determination of the Upper Limits, Benchmarks, and Critical Properties for Gas Separations Using Stabilized Room Temperature Ionic Liquid Membranes for the Purpose of Guiding Future Research. J. Membr. Sci. 2009, 343, 199−211. (21) Le Bideau, J.; Viau, L.; Vioux, A. Ionogels, Ionic Liquid Based Hybrid Materials. Chem. Soc. Rev. 2011, 40, 907−925. (22) Eddaoudi, M.; Kim, J.; Rosi, N.; Vodak, D.; Wachter, J.; O’Keefe, M.; Yaghi, O. M. Systematic Design of Pore Size and Functionality in Isoreticular MOFs and Their Application in Methane Storage. Science 2002, 295, 469−472. (23) Chen, Y. F.; Hu, Z. Q.; Gupta, K. M.; Jiang, J. W. Ionic Liquid/ Metal-Organic Framework Composite for CO2 Capture. J. Phys. Chem. C 2011, 115, 21736−21742. (24) Gupta, K. M.; Chen, Y.; Hu, Z.; Jiang, J. Metal−Organic Framework Supported Ionic Liquid Membranes for CO2 Capture: Anion Effects. Phys. Chem. Chem. Phys. 2012, 14, 5785−5794. (25) Banerjee, R.; Phan, A.; Wang, B.; Knobler, C.; Furukawa, H.; O’Keeffe, M.; Yaghi, O. M. High-Throughput Synthesis of Zeolitic Imidazolate Frameworks and Application to CO2 Capture. Science 2008, 319, 939−943. 5799
dx.doi.org/10.1021/jp312404k | J. Phys. Chem. C 2013, 117, 5792−5799
