H2 in

Jun 28, 2010 - Functional Materials for Gas Storage. Part II: .... Experiences with the publicly available multipurpose simulation code, Music ... A n...
0 downloads 0 Views 5MB Size
12158

J. Phys. Chem. C 2010, 114, 12158–12165

Molecular Simulation for Adsorption and Separation of CH4/H2 in Zeolitic Imidazolate Frameworks Hai-chao Guo,† Fan Shi,‡,§ Zheng-fei Ma,*,† and Xiao-qin Liu† State Key Laboratory of Materials-Oriented Chemical Engineering, Nanjing UniVersity of Technology, Nanjing, 210009 China, and The Chemical Engineering Department, UniVersity of Michigan, Ann Arbor, Michigan 48109 ReceiVed: September 16, 2009; ReVised Manuscript ReceiVed: June 2, 2010

In this work, the grand canonical Monte Carlo method was employed to study the adsorption and separation characteristics of CH4/H2 on MOF-5 and five zeolitic imidazolate frameworks (ZIFs), including two sodalite (SOD), ZIF-8 and -67, two merlinoite (MER), ZIF-10 and -60, and one DFT, ZIF-3. Simulations show that more CH4 molecules are adsorbed in all frameworks than H2, which is consistent with a higher pure gas isosteric heat of adsorption of CH4 as compared with that of H2. For both gases, adsorbed amounts primarily rely on the physical and chemical parameters of the adsorbent. Results of density distribution profiles and equilibrium snapshots of the ZIFs indicate that the most preferential gas adsorption sites for both CH4 and H2 are the positions near linkers. At high pressures, CH4 begins to fill up in the center of the SOD cage. We also found that the selectivity for CH4 increased with the difference between the isosteric heats of adsorption of CH4 and H2, ∆qst, but decreased to some extent due to the packing effect. Both the isosteric heats of adsorption and the packing effect are mainly influenced by the topology of the framework. 1. Introduction Zeolitic imidazolate frameworks (ZIFs), a new class of nanoporous materials based on zeolitic topologies, consist of transition-metal ions (zinc/cobalt) and imidazolate/imidazolatetype linkers.1-3 They appear to be promising materials for the storage and separation of gases by adsorption due to their exceptional chemical and thermal stabilities and possess more tunable structures as compared with traditional zeolites.4,5 To date, several studies have reported the experimental adsorption on ZIFs with pure components, such as CH4,2,6-8 CO2,1,2,8 CO,1 N2,1,3-5 and H2.5,6,9 The Yildirim group6 measured the adsorption of CH4 and H2 on ZIF-8 over large temperature and pressure ranges. They7,9 also analyzed adsorption sites for CD4 and D2 molecules on ZIF-8 using neutron powder diffraction at low temperatures and found that the imidazolate linker is primarily responsible for adsorption. However, owing to the fact that getting experimental adsorption isotherms for mixtures is more complicated than that for pure components, separation characteristics, which are crucial for applying materials for practical separation in industries, are difficult to obtain from experimental methods. For example, to provide pure H2 for use in fuel cell applications, CH4 from either methane steam reforming or methane dry reforming10,11 has to be separated from H2. So far, most of the experimental zeolite membrane data on mixed gas separations indicate that there is no correlation between single gas and binary gas separations. Kalipcilar et al.12 experimentally measured a SSZ-13 zeolite membrane and determined that the separation factor for a CH4/ H2 mixture is 8. Hong et al.13 separated H2 from CH4 with a SAPO-34 membrane, achieving a separation factor of about 20. This is because the pore size of SAPO-34 is 0.38 nm, similar * To whom correspondence should be addressed. E-mail: [email protected]. † Nanjing University of Technology. ‡ University of Michigan. § Current address: URS Washington Group/U.S. D.O.E., National Energy Technology Laboratory, 626 Cochrans Mill Road, Pittsburgh, PA 15263.

to the kinetic diameter of CH4 (0.38 nm) but larger than that of H2 (0.29 nm), indicating that SAPO-34 can capture CH4 molecules strongly, leaving H2 molecules diffuse in pores. A novel NaA zeolite membrane,14 modified by eliminating the eight-membered rings, shows very high separation factors for H2/CH4 (>1200), H2/N2 (>1600), and H2/CO (>660) due to the existence of four- and six-membered rings, which favor the permiability of H2. The H2 separation performance and mechanisms of zeolite membranes were summarized by Ockwig et al.,11 who predicted that zeolite membranes will be put at the forefront of separation technology owing to their chemical, mechanical, and thermal stabilities and high selectivities. The microscopic information regarding the adsorption and separation of gas mixtures using zeolite membranes, however, is difficult to obtain. To further understand the mechanism of separation, a combination of experimental methods and computational simulation is necessary. In the past few years, the adsorption and separation of CH4 and H2 in inorganic porous materials and metal-organic frameworks (MOFs) were simulated successfully using the grand canonical Monte Carlo (GCMC) method. Mitchell et al.15 investigated the separation of CH4/H2 in three different titanosilicates (zorite, ETS-4, and ETS-10) at 500 °C via a combination of GCMC and molecular dynamics (MD). Their results demonstrated that these titanosilicates showed higher loading and self-diffusivity for H2 than CH4. They also concluded that zorite and ETS-4 could be efficient sieves for H2/CH4 mixtures because CH4 does not show appreciable displacement in these sorbents; ETS-10 could be a promising adsorbent for fast industrial separations of H2 and CH4 due to high self-diffusion coefficients for H2. Gallo et al.16 calculated the adsorption and separation of CH4/H2 on four typical MOFs and suggested that the CH4 selectivity depends more on adsorbate-framework interactions than on adsorbate-adsorbate interactions. Liu et al.17 simulated the separation characteristics of CH4/H2 in three interpenetrated IRMOFs (IRMOF-9, -11, and -13) and compared

10.1021/jp908978q  2010 American Chemical Society Published on Web 06/28/2010

Adsorption and Separation of CH4/H2 in ZIFs

J. Phys. Chem. C, Vol. 114, No. 28, 2010 12159

Figure 1. Two-dimensional view of a unit cell of (a) ZIF-8, (b) ZIF-67, (c) ZIF-10, (d) ZIF-60, and (e) MOF-5 and 2 × 2 × 1 cells of (f) ZIF-3: C, gray; N, blue; H, white; O, red; Zn, yellow; Co, orange.

them to three noninterpenetrated counterparts (IRMOF-10, -12, and -14). The results indicated that CH4 selectivity was enhanced on the interpenetrated frameworks due to their small pores and abundance of adsorption sites. Using GCMC, Yang and Zhong18 found that CH4 selectivity in Cu-BTC (benzene-1,3,5-tricarboxylate) is larger than that in MOF-5 because of stronger interactions in the complex structure of Cu-BTC as compared with those in MOF-5. In this work, GCMC simulations were performed to study the adsorption and separation properties of CH4/H2 at 300 K on MOF-5 and five ZIFs with three types of zeolite topologies,19 including two sodalite (SOD), ZIF-8 and -67, two merlinoite (MER), ZIF-10 and -60, and one DFT, ZIF-3. The experimental data6 on the adsorption of pure CH4 and H2 on ZIF-8 will be used to validate our model for CH4/H2 adsorption. The adsorption isotherms, isosteric heat, and adsorption selectivity for CH4 and the effects of the topology of sorbents to the selectivity will be discussed. In addition, preferential adsorption sites in ZIFs will be studied. The goal of this work is to evaluate ZIFs as promising candidates for the storage and adsorptive separation

of CH4/H2 and to provide valuable information for applying ZIFs for such processes in the future. 2. Materials and Methods 2.1. ZIF Structures. The unit cell frameworks of the of ZIFs and MOF-5 (as shown in Figure 1) were constructed from experimental powder X-ray diffraction (PXRD) data.1,5,20 The unit cell of both ZIF-8 and ZIF-67 contains six four-membered rings and eight six-membered rings, which corresponds to the SOD zeolite-type topology. The cubic framework of ZIF-8 consists of zinc and 2-methylimidazolate (2-MeIM) linkers, with a lattice constant of 16.9910 Å. The ZIF-67 sphere framework was linked by cobalt and 2-MeIM linkers with a lattice constant of 16.9589 Å, which is slightly smaller than that of ZIF-8. The ZIF-3 framework consists of zinc and imidazolate (IM) linkers, with a zeolite DFT topology.5 The unit cell of ZIF-3 is a cuboid with dimensions of 18.970 × 18.970 × 16.740 Å3. The ellipsoidal cages of ZIF-3 contain four eight-membered rings and two four-membered rings. By examining the location of the four-membered rings, the ZIF-3 cages could be grouped into

12160

J. Phys. Chem. C, Vol. 114, No. 28, 2010

Guo et al.

TABLE 1: Structural Properties for the ZIFs and MOF-5 zeolite daperture topology (Å) dpore (Å) ZIF-81 ZIF-671 ZIF-31 ZIF-101 ZIF-601 MOF-520 a

SOD SOD DFT MER MER

3.4 3.4 4.6 8.2 7.2 11.2

11.6 11.6 6.0 12.1 9.4 18.5

cage

chemical formula

[46.68] [46.68] [42.84] [412.86] [412.86] [46]

Zn(MeIM)2 Co(MeIM)2 Zn(IM)2 Zn(IM)2 Zn(IM)1.5(MeIM)0.5 Zn4O(bdc)3a

bdc ) 1,4-benzenedicarboxylate.

2.3. Simulation Methods. The adsorption and separation characteristics of CH4/H2 were calculated using Monte Carlo (MC) simulations in the grand canonical ensemble (µVT) with the code MuSiC,26 where the chemical potential, temperature, and volume are constant. Chemical potentials were converted to fugacity with the Peng-Robinson equation of state (PREOS).27 In many studies,28-30 absolute adsorbed amounts obtained by MC simulation are converted to excess adsorbed amounts to compare with experimental data. In this paper, the amount of excess adsorbed molecule Nex is calculated using eq 4:

TABLE 2: LJ Potential Parameters for Gases and Frameworks in This Work force field type adsorbate framework22

23

CH4 H224 C H N O Zn Co

σ (Å)

ε/kB (K)

3.73 2.59 3.43 2.57 3.26 3.12 2.46 2.56

148 12.5 52.84 22.14 34.72 30.19 62.4 7.05

two types: cage A (with the four-membered rings at the ends of the ellipsoid) and cage B (with the four-membered rings at the middle of the ellipsoid), as shown in Figure 1f. ZIF-10 and ZIF-60 belong to the MER topology, containing twelve fourmembered rings and six eight-membered rings. In the ZIF-10 framework, each Zn atom is bridged by four IM linkers, whereas in ZIF-60, each Zn atom is connected with three IM linkers and one MeIM linker. The differences in organic ligands are illustrated in Figure 1c,d. The structure of MOF-5 is described elsewhere.21 The structural properties for ZIFs1 and MOF-520 are summarized in Table 1. 2.2. Force Field. van der Waals (VDW) interactions between frameworks and adsorbates were treated with a 12-6 LennardJones (LJ) potential (see eq 1). The cross-interaction parameters betweendifferentatomswerecalculatedusingtheLorentz-Berthelot mixing rules (see eqs 2 and 3).

uLJ ) 4εij

[( ) ( ) ] σij rij

12

-

σij rij

6

(1)

1 σij ) (σi + σj) 2

(2)

εij ) √εiεj

(3)

In this work, the potential parameter values for frameworks were from the UFF force field of Rappe et al.22 The potential parameter values for CH4 and H2 molecules were from TraPPEUA23 and the work of Buch24 (see Table 2), in which the real molecules CH4 and H2 were described as pseudoatoms. The cutoff radii for ZIFs and MOF-5 were set to be 7.8 and 12.8 Å, respectively. The quadrupole moment of H2 is quite weak. Garberoglio et al.25 showed that the effect of electrostatic interactions for the adsorption of H2 was essentially negligible at room temperature and pressures up to 5 MPa. In addition, CH4 is a nonpolar molecule. Consequently, electrostatic interactions were also ignored in this work.

Nex ) Nab - VpF

(4)

where Nab is the amount of absolute adsorbed molecules, Vp represents the pore volume of adsorbent, and F is the density of the adsorbate calculated using the Peng-Ronbinson equation of state at a given temperature and pressure. Because the framework structures were assumed to be rigid, the potential energies for gas molecules/frameworks were pretabulated on a series of nodes spaced 0.2 Å apart in the three dimensions. During the simulation, the potential energies at nontabulated points were calculated by hermite interpolation between the nodes. The numbers of unit cells of frameworks adopted in this simulation were 2 × 2 × 2, and periodic boundary conditions were applied in all three dimensions. The frameworks were treated as rigid with frozen atoms during simulation. In this work, a total of 3 × 107 steps were used; the first half of these moves was used for equilibration, and the remaining steps were used for calculating the ensemble averages. For pure CH4 and H2, three types of moves (translation, random insertion, and deletion) were used. For binary mixtures, an additional move named identity-swap was included in order to accelerate the convergence and reduce statistical errors.31 Every possible move was given equal probability. In a dilute (ideal gas) phase, the isosteric heat of adsorption qst was calculated from eq 532

qst ) RT -

ν> ( ∂∂ )

T

(5)

where is the average potential energy of the adsorbed phase, is the average number of molecules adsorbed, T is the temperature, and R is the ideal gas constant. To estimate the pore volume (Vp) for these materials, we employed an approach proposed by Talu and Myers.33 They suggested simulating the amount of helium molecules contained in a unit mass adsorbent (Na) from GCMC simulations at low pressures (P) and ambient temperature (T0). Helium is assumed to be a nonadsorbing gas, and its pore volume can be calculated from eq 6

Vp )

NakT0 P

(6)

where k refers to the Boltzmann constant. The calculated pore volumes are shown in Table 3. For the adsorptive separation of binary mixtures, the selectivity for component 1 over component 2 is defined by eq 7

Adsorption and Separation of CH4/H2 in ZIFs

S)

J. Phys. Chem. C, Vol. 114, No. 28, 2010 12161

x1 y2 x2 y1

(7)

where yi and xi are the mole fractions of component i (i ) 1 or 2) in the bulk and adsorbed phases, respectively. The density distribution profiles were used to characterize the adsorption sites of the different materials. Jiang et al.34,35 demonstrated the molecular level interactions of alkanes in IRMOF-1 and carbon nanotube bundles using the density distribution profiles. Yang and Zhong36 simulated the local density function F(r) of CO2 adsorbed in COF-8 and methane adsorbed in COF-10, expressed as F(r) ) /(2πLr∆r), assuming hundreds of concentric cylindrical layers as unit cells in parallel around the axis of the pore. Along with the evidence of the snapshots at different pressures, they explained that stepped isotherms are the results of the multilayer adsorption at the temperatures studied. In this work, density distribution profiles were computed to quantitatively describe adsorption. The local density in every layer is expressed in eq 8

F(r) )

4πr2∆r

(8)

where represents the average adsorbed molecules located in r to r + ∆r and r is the radial distance from the center of a unit cell. A unit cell is divided into approximately 150 concentric spherical layers along the radial direction, simulating a nearly spherical cage. 3. Results and Discussion 3.1. Validation of the Force Field. To validate the employed force field based on the LJ potential model, simulated adsorption excess isotherms of CH4 and H2 on ZIF-8 and MOF-5 at 300 K were compared to experimental data.6 As shown in Figure 2, good agreement between the simulation results and the experimental data was observed. Therefore, this force field based on the LJ potential model will be utilized for further study in this work. 3.2. Adsorption of Pure Components. A high isosteric heat of adsorption indicates strong interactions between the adsorbate and the adsorbent. Isosteric heats of adsorption of CH4 and H2 for five ZIFs and MOF-5 at 1000 kPa were calculated, as summarized in Table 3. The isosteric heats of CH4 adsorption are all greater than those of H2, indicating stronger interactions between CH4 and the frameworks than those of H2. Additionally, we determined that the isosteric heat of adsorption for the different frameworks decreases in the following order: ZIF-3 (DFT) > ZIF-8 and -67 (SOD) > ZIF-10 and -60(MER) > MOF5. This demonstrates that the isosteric heats of adsorption are similar between ZIFs sharing the same topology but differ for ZIFs with dissimilar topologies. It is also interesting to find that TABLE 3: CH4 and H2 Adsorption Isosteric Heats at 1000 kPa and the Pore Volumes for Five ZIFs and MOF-5 framework

Vp (cm3/g)

qst(CH4) (kcal/mol)

qst(H2) (kcal/mol)

MOF-5 ZIF-3 ZIF-8 ZIF-10 ZIF-60 ZIF-67

1.39 0.64 0.52 0.82 0.77 0.54

2.40 3.63 3.29 2.86 2.91 3.26

1.09 1.44 1.37 1.19 1.20 1.36

Figure 2. Comparison of the simulated and experimental6 excess isotherms on ZIF-8 and MOF-5 at 300 K for (a) CH4 and (b) H2.

generally the isosteric heats increase with the decreasing cage volume, with the exception of ZIF-3, which has the smallest dp and a relatively small daperture. Therefore, the isosteric heat of adsorption is strongly affected by the ZIF topology, specifically, with respect to the physical properties of pores and the number and type of linkers. The relationship between qst and pressure is explained in the Supporting Information. Adsorption isotherms of CH4 and H2 for ZIFs and MOF-5 at 300 K were computed using GCMC, as shown in Figure 3. Figure 3a,b shows that the adsorbed amount increases gradually with rising pressure for CH4 and H2, respectively. It is noteworthy that MOF-5 shows the largest uptake for both CH4 and H2 but has the smallest isosteric heats, whereas ZIF-3 has the largest isosteric heats and a relatively small uptake. These findings indicate that the uptake of gas molecules for adsorbents not only depends on the isosteric heat but also strongly on the pore volume. Large pore volumes can accommodate more gas molecules, which results in a large uptake of a particular gas component. In addition, for ZIFs sharing the same topology (e.g., SOD ZIF-8 and -67 or MER ZIF-10 and -60), the isotherms and pure gas isosteric heats of adsorption of CH4 and H2 are close. This indicates that adsorption behaviors depend on not only the type of metal ions but also the ZIF topology itself. Small discrepancies in either the isotherms or the isosteric heats may be caused by small variances in the organic ligands. We also found that the amount of adsorbed CH4 exceeds the amount of H2 for all frameworks. The difference could be mainly attributed to the energetic effects. 3.3. Separation of Binary Mixtures. In this section, the CH4 selectivity for CH4/H2 mixtures with different gas compositions and pressures at a constant temperature37-39 are discussed. Figure 4a-c illustrates the effect of pressure on CH4 selectivity on different frameworks at three gas compositions. It shows that CH4 selectivities are very close between ZIFs with

12162

J. Phys. Chem. C, Vol. 114, No. 28, 2010

Guo et al.

Figure 3. Simulated isotherms on five ZIFs and MOF-5 for (a) CH4 and (b) H2.

the same topology according to following order: ZIF-3 > ZIF-8 and -67 > ZIF-60 > ZIF-10 > MOF-5. It can be seen that the behavior of pressure-dependent CH4 selectivity could be categorized into three groups: (1) For SOD and MER types of ZIFs, CH4 selectivity remains nearly unchanged or slightly decreases along with pressure while yCH4 e 0.5. In a CH4-rich mixture, CH4 selectivity gradually decreases along with rising pressure. (2) For DFT ZIF-3, CH4 selectivity decreases with increasing pressure. With increasing CH4 concentrations, CH4 selectivity declines more quickly. (3) For MOF-5, with increasing CH4 composition, CH4 selectivity increases to a certain extent along with increasing pressure. Therefore, it could be concluded that the topology of ZIFs has a greater effect on CH4 selectivity than pressure and gas composition. To explain the effect of ZIF topology on CH4 selectivity, the difference between the isosteric heats of adsorption of CH4 and H2,40 ∆qst, was calculated. Generally, a large ∆qst indicates a good CH4 selectivity. Figure 5 illustrates ∆qst over different sorbents for equimolar CH4/H2 mixtures at different pressures. Again, ∆qst for the frameworks follows the order ZIF-3 > ZIF-8 and -67 > ZIF-60 > ZIF-10 > MOF-5, which is the same as that of CH4 selectivity. Therefore, ∆qst provides a good measure of how selectivity varies with different framework topologies. For ZIF-8 and -67, the simple SOD topology results in a steady CH4 selectivity of about 11. As to MER ZIF-10 and -60, the CH4 selectivities are low due to a small ∆qst. However, the 2-MeIM linkers in ZIF-60 enhanced the ∆qst, which provided a slightly better CH4 selectivity as compared with ZIF-10. ZIF-3 had the largest ∆qst, resulting in a higher CH4 selectivity than other frameworks. However, with high loadings or pressures, packing effects on ZIF-3 become significant, which decreases the CH4 selectivity. This effect will be discussed in the following section. For MOF-5, which exhibits the smallest ∆qst and possesses a simple framework configuration, the CH4 selectivity is the

Figure 4. Influence of pressure on CH4 selectivity for three gas compositions: (a) yCH4/yH2 ) 0.05:0.95, (b) yCH4/yH2 ) 0.5:0.5, and (c) yCH4/yH2 ) 0.95:0.05.

Figure 5. Differences of isosteric heats (∆qst) between CH4 and H2 over different frameworks for equimolar CH4/H2 mixtures.

lowest among all frameworks examined, with a CH4 selectivity value near 5. However, it is notable that the CH4 selectivity of MOF-5 increases slowly at high pressures or high CH4 partial pressures. Yang and Zhong18 suggested that the packing effect

Adsorption and Separation of CH4/H2 in ZIFs

J. Phys. Chem. C, Vol. 114, No. 28, 2010 12163

Figure 6. Equilibrium snapshots of CH4 adsorbed at 100 kPa on (a) ZIF-8, (c) ZIF-3, (e) ZIF-10, and (g) ZIF-60 and at 3000 kPa on (b) ZIF-8, (d) ZIF-3, (f) ZIF-10, and (h) ZIF-60 (where x, y ∈ [0,1], z ∈ [0.4,0.6] for ZIF-8, -10, and -60 and x, y ∈ [0,1.5], z ∈ [0.4,0.6] for ZIF-3).

favoring H2 is insignificant and energetic effects still dominate in the studied pressure range. 3.4. Preferential Adsorption Sites. To understand the adsorption behavior of CH4 molecules on sorbents with different topologies, preferential adsorption sites for CH4 over SOD ZIF8, DFT ZIF-3, and MER ZIF-10 and -60 were investigated. Figure 6 shows equilibrium snapshots of adsorbed CH4 in simulation unit cell(s) where x, y ∈ [0, 1] (x, y ∈ [0, 1.5] for ZIF-3) and z ∈ [0.4, 0.6] over ZIF-8, -3, -10, and -60 at 100 and 3000 kPa, respectively. Although the aperture size, daperture, of ZIF-8 is smaller than the kinetic diameter of CH4, Bux et al.41 believed that large molecules, such as CH4, can still enter slowly into the cage of ZIF-8 through the flexible apertures. For SOD ZIF-8, at a low pressure, as shown in Figure 6a, CH4

molecules occupy four corners of a square space near the 2-MeIM in the cage, which agrees with the discussion of Wu.7 At high pressures, in addition to the sites near linkers, CH4 molecules start to fill up the center of the SOD cage, as shown in Figure 6b. These CH4 adsorption behaviors in ZIF-8 are very similar to the results of H2 in ZIF-8 reported by Zhou et al.42 For DFT ZIF-3, the adsorption patterns in cages A and B are comparable: most of the CH4 was adsorbed close to linkers. However, the available adsorption space in cage A is wider and shorter than that in cage B. At 3000 kPa, it seems that CH4 molecules are more crowded in cage B than in cage A. For MER ZIF-10 and -60, for a better understanding of the CH4 adsorption behaviors, we studied adsorption sites inside the channel connecting two cages. At 100 kPa, most of the CH4

12164

J. Phys. Chem. C, Vol. 114, No. 28, 2010

Guo et al.

Figure 7. Density distribution profiles on ZIF-8 for (a) CH4 and (b) H 2.

molecules would like to occupy the four corners inside the channel near the organic linkers, as shown in Figure 6e,g. However, because of the presence of a MeIM linker in the ZIF60 structure, the adsorbed CH4 molecules were compressed to a smaller space than that in ZIF-10. In addition, an octagontype adsorption space was observed in ZIF-60. At 3000 kPa, as shown in Figure 6f,h, octagon-shaped adsorption spaces were found in both ZIF-10 and -60. However, in ZIF-10, there are four strong adsorption sites for CH4 at 45, 135, 225, and 315° of the octagon. In ZIF-60, there is no significant difference in the adsorption sites around the octagon, which is probably due to the compressed adsorption space by the MeIM linker. Compared with the CH4 adsorption in the ZIF-8 cage, the adsorption sites in the center of the MER ZIF channels are not favorable, even at 3000 kPa. The primary reason is because the pore volume and daperture of ZIF-10 and -60 are much larger than those of SOD ZIF-8. Therefore, CH4 molecules are more likely to stay closer to the linkers than to the center. In addition to the equilibrium snapshots of CH4 adsorption, Figure 7 gives examples of the density distribution of CH4 and H2 in SOD ZIFs (e.g., ZIF-8 and ZIF-67). With regard to CH4 in ZIF-8, as shown in Figure 7a, the local density at approximately 4.3 Å and near the 2-MeIM linkers increases sharply with rising pressure. This finding confirms the early snapshot conclusions shown in Figure 6a,b. Furthermore, an adsorption peak near 0 Å is found at 3000 kPa, indicating that the CH4 begins to fill up the center of the SOD cage, which has also been found in Figure 6b. Figure 7b shows that the local density of H2 always reached a peak at about 4.3 Å, while exhibiting considerable randomness from 0 to 3 Å. These findings indicate that H2 prefers to adsorb at the sites near the 2-MeIM linkers and then fills up the cage progressively, which is consistent with the experimental observations of Wu et al.7,9 Similar conclusions can be drawn from the density distribution profiles of CH4 and H2 for ZIF-67, as shown in the Supporting Information. These similarities may be due to similar topologies,

Figure 8. Equilibrium snapshots of an equimolar CH4/H2 mixture adsorbed in ZIF-3 at: (a) 100 kPa and (b) 3000 kPa. Blue dot, CH4; red dot, H2 (where x, y ∈ [0,1.5], z ∈ [0.4,0.6]. For a better view of the density distribution of gas molecules in cages A and B, neighbor cages are omitted). Highlighted areas are accessible adsorption sites for CH4 and H2 at different pressures.

linkers, and cage sizes. The different metal ions in ZIF-8 and -67 seem to have very limited influence on the preferential adsorption sites for both CH4 and H2. 3.5. Influence of Packing Effect on CH4 Selectivity. At high pressures, in addition to the adsorption sites near linkers, H2 randomly adsorbed in the cages, as indicated by Figure 7b. Generally, at high pressures, because of packing effects, H2 will be preferentially adsorbed as compared to CH4 and ultimately will decrease the CH4 selectivity. Figure 8 gives an example of adsorption of an equimolar CH4/H2 mixture in DFT ZIF-3 by presenting equilibrium snapshots at 100 and 3000 kPa. At 100 kPa, both CH4 and H2 favorably occupied sites near fourmembered rings with strong interactions (see highlighted areas in Figure 8a). At 3000 kPa, H2 molecules begin to pack the space (see highlighted areas in Figure 8b) where large-sized

Adsorption and Separation of CH4/H2 in ZIFs CH4 molecules will not fill. This is because cages in ZIF-3 are split into several discrete ellipsoids by the small diameter (dpore ) 6 Å) of the framework. At a high pressures, these cages would rather accommodate small H2 (2.89 Å) molecules as compared with large CH4 (3.8 Å) molecules, which results in a decrease in CH4 selectivity. Keffer et al.43 also suggested that small guest molecules are entropically favored in spaces with high pore densities, where they can achieve higher packing densities. For SOD-type ZIF-8 and -67, their frameworks have plenty of capacious spheres with a diameter of 11.6 Å, in which nearly all the space is accessible to H2 and CH4 gas molecules. Thus, the packing effect on these two SOD ZIFs is not significant unless at a very high pressures or high CH4 partial pressures. For MER-ZIF-10 and -60, large pores and simple structures eliminate the packing effect at a high pressure or high CH4 partial pressure; instead, the energetic effects still play a significant part in CH4 selectivity. Consequently, CH4 selectivity could be reduced via the packing effect, which is principally influenced by the topology structure of frameworks but not the type of metal ions. 4. Conclusions Adsorption and separation of CH4/H2 in MOF-5 and DFT ZIF-3, SOD ZIF-8 and -67, and MER ZIF-10 and -60, were simulated by GCMC in this work. The effects of different topologies on adsorption behavior have been investigated. For pure components, the amount of CH4 adsorbed is always larger than that of H2 for the four different frameworks due to strong interactions between CH4 and the frameworks, as indicated by the isosteric heats of adsorption. However, the uptake of gas molecules for each adsorbent depends mainly on the framework topology and the physical and chemical properties of pores. Density distribution profiles of ZIF-8 and ZIF-67 indicated that the preferred adsorption sites are close to the 2-MeIM linkers in the cage, and the secondary sites for CH4 are in the center of the SOD cage. Therefore, the organic linkers in ZIF-8 and ZIF67, rather than the type of metal ions, strongly influence the adsorption and separation of CH4 and H2. With regard to binary mixtures, CH4 selectivities largely increase with increasing ∆qst but decrease with the packing effect. Both the ∆qst and the packing effect are influenced strongly by the topology of the frameworks, including pore volume, pore size, and linker properties. This work provides useful macroscopic and microscopic information for a complete understanding of adsorption and separation properties for CH4/H2 in ZIFs. The simulation results could be used to help design new ZIFs possessing large pore volumes and strong interactions for practical applications. Acknowledgment. The financial support by the Major Basic Research Project of the Natural Science Foundation of Jiangsu Province Colleges (No. 08KJA530001) is greatly appreciated. The authors wish to thank Prof. Randall Snurr and his colleagues for their guidance about the usage of the MuSiC code. Supporting Information Available: Figures showing absolute isotherms on ZIF-8 and MOF-5, isosteric heat versus pressures for CH4 and H2, and density distribution profiles on ZIF-67 for CH4 and H2. This material is available free of charge via the Internet at http://pubs.acs.org.

J. Phys. Chem. C, Vol. 114, No. 28, 2010 12165 References and Notes (1) Banerjee, R.; Phan, A.; Wang, B.; Knobler, C.; Furukawa, H.; O’Keeffe, M.; Yaghi, O. M. Science 2008, 319, 939. (2) Hayashi, H.; Cote, A. P.; Furukawa, H.; O’Keeffe, M.; Yaghi, O. M. Nat. Mater. 2007, 6, 501. (3) Wang, B.; Cote, A. P.; Furukawa, H.; O’Keeffe, M.; Yaghi, O. M. Nature 2008, 453, 207. (4) Morris, W.; Doonan, C. J.; Furukawa, H.; Banerjee, R.; Yaghi, O. M. J. Am. Chem. Soc. 2008, 130, 12626. (5) Park, K. S.; Ni, Z.; Cote, A. P.; Choi, J. Y.; Huang, R. D.; UribeRomo, F. J.; Chae, H. K.; O’Keeffe, M.; Yaghi, O. M. Proc. Natl. Acad. Sci. U.S.A. 2006, 103, 10186. (6) Zhou, W.; Wu, H.; Hartman, M. R.; Yildirim, T. J. Phys. Chem. C 2007, 111, 16131. (7) Wu, H.; Zhou, W.; Yildirim, T. J. Phys. Chem. C 2009, 113, 3029. (8) Banerjee, R.; Furukawa, H.; Britt, D.; Knobler, C.; O’Keeffe, M.; Yaghi, O. M. J. Am. Chem. Soc. 2009, 131, 3875. (9) Wu, H.; Zhou, W.; Yildirim, T. J. Am. Chem. Soc. 2007, 129, 5314. (10) Nenoff, T. M. Defect-free thin film zeolite membranes for hydrogen separation and isolation. DOE/H2 Annual ReView Meeting, Berkeley, CA, May 19, 2003. (11) Ockwig, N. W.; Nenoff, T. M. Chem. ReV. 2008, 107, 4078. (12) Kalipcilar, H.; Bowen, T. C.; Noble, R. D.; Falconer, J. L. Chem. Mater. 2002, 14, 3458. (13) Hong, M.; Li, S. G.; Falconer, J. L.; Noble, R. D. J. Membr. Sci. 2008, 307, 277. (14) Nishiyama, N.; Yamaguchi, M.; Katayama, T.; Hirota, Y.; Miyamoto, M.; Egashira, Y.; Ueyama, K.; Nakanishi, K.; Ohta, T.; Mizusawa, A.; Satoh, T. J. Membr. Sci. 2007, 306, 349. (15) Mitchell, C. M.; Gallo, M.; Nenoff, T. M. J. Chem. Phys. 2004, 121, 1910. (16) Gallo, M.; Glossman-Mitnik, D. J. Phys. Chem. C 2009, 113, 6634. (17) Liu, B.; Yang, Q. Y.; Xue, C. Y.; Zhong, C. L.; Chen, B. H.; Smit, B. J. Phys. Chem. C 2008, 112, 9854. (18) Yang, Q. Y.; Zhong, C. L. J. Phys. Chem. B 2006, 110, 17776. (19) Lewis, D. W.; Ruiz-Salvador, A. R.; Gomez, A.; Rodriguez-Albelo, L. M.; Coudert, F. X.; Slater, B.; Cheetham, A. K.; Mellot-Draznieks, C. CrystEngComm 2009, 11, 2272. (20) Eddaoudi, M.; Kim, J.; Rosi, N.; Vodak, D.; Wachter, J.; O’Keeffe, M.; Yaghi, O. M. Science 2002, 295, 469. (21) Babarao, R.; Hu, Z. Q.; Jiang, J. W.; Chempath, S.; Sandler, S. I. Langmuir 2007, 23, 659. (22) Rappe, A. K.; Casewit, C. J.; Colwell, K. S.; Goddard, W. A.; Skiff, W. M. J. Am. Chem. Soc. 1992, 114, 10024. (23) Martin, M. G.; Siepmann, J. I. J. Phys. Chem. B 1998, 102, 2569. (24) Buch, V. J. Phys. Chem. 1994, 100, 7610. (25) Garberoglio, G.; Skoulidas, A. I.; Johnson, J. K. J. Phys. Chem. B 2005, 109, 13094. (26) Gupta, A.; Chempath, S.; Sanborn, M. J.; Clark, L. A.; Snurr, R. Q. Mol. Simul. 2003, 29, 29. (27) Peng, D. Y.; Robinson, D. B. Ind. Eng. Chem. Fundam. 1976, 15, 59. (28) Duren, T.; Snurr, R. Q. J. Phys. Chem. B 2004, 108, 15703. (29) Garberoglio, G. Langmuir 2007, 23, 12154. (30) Liu, L. C.; Fu, J.; Sun, H. Sci. China, Ser. B: Chem. 2008, 51, 760. (31) Fuchs, A. H.; Cheetham, A. K. J. Phys. Chem. B 2001, 105, 7375. (32) Snurr, R. Q.; Bell, A. T.; Theodorou, D. N. J. Phys. Chem. 1993, 97, 13742. (33) Talu, O.; Myers, A. L. AIChE J. 2001, 47, 1160. (34) Jiang, J. W.; Sandler, S. I.; Schenk, M.; Smit, B. Phys. ReV. B 2005, 72, 045447. (35) Jiang, J. W.; Sandler, S. I. Langmuir 2006, 22, 5702. (36) Yang, Q. Y.; Zhong, C. L. Langmuir 2009, 25, 2302. (37) Frost, H.; Duren, T.; Snurr, R. Q. J. Phys. Chem. B 2006, 110, 9565. (38) Karra, J. R.; Walton, K. S. Langmuir 2008, 24, 8620. (39) Martin-Calvo, A.; Garcia-Perez, E.; Castillo, J. M.; Calero, S. Phys. Chem. Chem. Phys. 2008, 10, 7085. (40) Sicar, S.; Cao, D. V. Chem. Eng. Technol. 2002, 25, 945. (41) Bux, H.; Liang, F.; Li, Y.; Cravillon, J.; Wiebcke, M.; Caro, J. J. Am. Chem. Soc. 2009, 131, 16000. (42) Zhou, M.; Wang, Q.; Zhang, L.; Liu, Y. C.; Kang, Y. J. Phys. Chem. B 2009, 113, 11049–11053. (43) Keffer, D.; Davis, H. T.; McCormick, A. V. J. Phys. Chem. 1996, 100, 638.

JP908978Q