Available online at www.sciencedirect.com
Procedia Computer Science 4 (2011) 1193–1202
International Conference on Computational Science, ICCS 2010
Understanding Adsorption and Separation behavior of shorter chain alkane mixtures in Zeolitic Imidazolate Frameworks by molecular simulations Li zhang, Qingwen Li, Zhengjie Lu, Xinping Wang* Department of Chemistry, Zhejiang Sci-Tech University, XiaSha Higher Education Zone, Hangzhou, 310018, China. Key Laboratory of Advanced Textile Materials and Manufacturing Technology of Education Ministry
Abstract
Novel Zeolitic imidazolate frameworks (ZIFs) have attracted a lot of attentions in adsorption and catalysis owing to their superior properties to the traditional nanoporous materials. Grand canonical Monte Carlo combined with the configurational-bias Monte Carlo simulation was used to evaluate the adsorption isotherms of short linear alkane mixtures in ZIF-8 and-10.It is found that adsorbed amount of alkane molecules increases with the pressure increasing, the longer chain alkane molecules, the higher adsorption amount, and the adsorbed amounts of alkane molecules in ZIF-10 are larger than that in ZIF-8. The selectivities of the ZIFs for different components were further investigated and discussed, it is found that the selectivity of ZIF-10 is little higher than that of ZIF-8, but it is much lower than other zeolites. The adsorption sites of alkane mixtures were also obtained by mCT images, the adsorption mechanism of C1-C3 alkane mixtures in the ZIFs was analyzed, it may be helpful to design new ZIF materials with good adsorption and separation capability. Key words˖Zeolitic imidazolate frameworks (ZIFs); adsorption isotherm; adsorption sites; relative selectivity, alkane mixtures
Introduction Microporous zeolite materials have been widely explored by experimental and theoretical techniques due to the broad application in petrol chemical cracking, water purification and softening [1-2], and gas exchange and storage.[34] since 1990. Lots of zeolites with different topologies and structures were used for adsorption and separation of different mixtures.[5-6] In the recent years, the attention has turned to metal organic frameworks (MOFs),it seems to be an attractive way of combining this structural diversity with embedded metallic functions. Recently, a novel class of subfamily of MOFs, with structures similar to zeolites were synthesized, it is composed of tetrahedral cluster of MN4 (M=Co, Cu, Zn, etc.)unites covalently linked by simple imidazolate ligands(IM). Given that the M-IM-M angle is near145°, which is coincident with the Si-O-Si angle preferred and commonly found in many zeolites. [78] Therefore, this kind porous crystalline material was named as Zeolitic Imidazolate Frameworks (ZIFs). It exhibits the advantages of high adsorption capacity, tunable pore size, light weight, and chemical functionality of classical MOFs, these characteristics combined with their high thermal and chemical stability and different structures similar * Corresponding author. Tel.: +86-571-86843600; fax: +86-571-86843600. E-mail address:
[email protected]
1877–0509 © 2011 Published by Elsevier Ltd. Selection and/or peer-review under responsibility of Prof. Mitsuhisa Sato and Prof. Satoshi Matsuoka doi:10.1016/j.procs.2011.04.128
1194
Li zhang, et al. / Procedia Computer Science 4 (2011) 1193–1202
to zeolites, make ZIFs promising alternatives to zeolites and other nanoporous materials for adsorption, separation, and catalysis. It is well known that adsorption and separation behavior of alkane mixtures in microporous materials are of great importance in the petroleum and petrochemical industries. To design adsorption processes needs a great deal of adsorption and separation data. But it is time-consuming to obtain adsorption data by experiments alone, especially for mixtures. Molecular simulation provides an attractive way to acquire adsorption data and explore the adsorption mechanism at molecular level. In fact, many research groups have employed Monte Carlo ( MC ) and molecular dynamic simulation (MD) to investigate the adsorption and diffusion mechanism of alkane mixtures in zeolites and other microporous materials, such as MOFs.[9-13 ]For example, Lu et al[14] have discussed the adsorption and separation C4-C7 alkane isomer in zeolites with different topologies by using configurational-bias Monte Carlo technique, it was found that the selectivity of zeolites are closely related to their topologies. Düren and Snurr investigated the adsorption behavior of CH4/n-C4H10 mixtures and the selectivity in different IRMOFs, the results indicated that the selectivity for n-butane varied strongly with the MOF linker molecule, they have also proposed some newly designed materials that show higher selectivity.[15]Besides, they also simulated multicomponent mixture of C5-C7 in MOF-1, the result shows that MOFs are highly selective for separating alkane mixtures based on the degree of branching in mixtures[13]. The adsorption of pure linear and branched alkane and their mixtures in IRMOF-1 have been studied by Jiang et.al[11], it was found that competitive adsorption occurs in alkane mixtures as a consequence of size and/or configurational differences between the components. In our previous works, we also investigated the adsorption and separation of alkane isomer mixtures(C4-C6) in IRMOF-1 and -6, it is illustrated that the longer alkane molecules, especially the branched molecules cannot be very close to the inorganic corner owing to the stronger repulsive interaction and the space hindrance of IRMOF-6[16], which is consist with the viewpoint proposed by smit [17], that is, the adsorption behavior of mixtures would be affected by size entropy effect configurational entropy effects and length entropy in company with each other.[17] Despite several experimental and simulation studies focused on adsorption and separation behavior of light gases and alkane mixtures in zeolitesǃcarbon materials and MOFs in the past few years. There are almost no simulation and experimental studies focusing on the adsorption of alkane mixtures in ZIFs, which are considered as the most promising alternatives to zeolites. The adsorption mechanism of alkane mixtures in ZIFs is poorly understood, and the relationship between the structure and adsorption is still lacking, which may provide useful information for designing new ZIFs with good adsorption and separation capacity. In this work, the grand canonical Monte Carlo (GCMC) in combination with the Configurational-bias Monte Carlo (CBMC) technique was employed to investigate the adsorption isotherm of short linear alkane mixtures in ZIF-8 and-10,which are consist of the same ZnN4 unites linked by different simple imidazolate ligands with different topologies. The relative selectivities of ZIFs were also explored to reveal the separation behavior of adsorbent. Besides, the Computer Tomography for materials (mCT) technique was introduced to investigate the adsorption mechanism of alkane mixtures in ZIF-8 and ZIF-10.
Models and Simulation details In this paper, two typical ZIFs, ZIF-8 and ZIF-10 were constructed from the X-ray diffraction (XRD) data [18]. The Lennard-Jones potential was employed to describe the interactions between the adsorbed molecules and frameworks of ZIFs as well as that between adsorbate. The adsorbate molecules were described with a united atom (UA) force field, but the atomistic model was employed for ZIF frameworks. The potential parameters for alkane molecules were taken from Vlugt et al[19].while for the MOF frameworks, the Dreiding force field was adopted[20]. The cut off radius was set to be 13.8 Å, and the cross interaction parameters between different units were calculated from the Jorgensen mixing rules [21] The simulation boxes are constructed by eight unite cell of ZIFs, The GCMC simulations were carried out at 298 K, 2×105 MC cycles were performed for alkane mixtures. 1×105 simulation steps were performed for equilibration, and another 1×105 steps to sample the data. Molecular simulation calculated the absolute amount of the adsorbed molecules, while the experimentally measured one is the excess amount of the adsorbed molecules. In order to compare with the experiment data, the excess molecules, nex is calculated by equation:
n ex
n abs V g U g
where Vg is the pore volume of adsorbent, and ȡg represent the bulk gas density of adsorbed molecules, it is calculated with the Peng-Robinson equation of state.
1195
Li zhang, et al. / Procedia Computer Science 4 (2011) 1193–1202
Results and Discussion 3.1
Comparison with the experimental
In our previous studies, we have investigated the adsorption isotherm of hydrogen and alkane mixtures in IRMOFs by GCMC simulations [16, 22], it suggest that the simulation results agree well with the experimental data. In order to further verify the model and the programs used in this work, the adsorption isotherms for pure hydrogen in ZIF-8 at 77K was calculated for alternative and compared with Yaghi’s experimental data owing to the absence of the experiment data for alkane molecules in ZIFs. From Fig 1, it is found that the simulation results are reasonably agreement with the experimental data, which demonstrates the simulation model and the force field parameters used in this work. 160
120
3
adsorbed amount/ cm g
-1
140
100 80 60 40
simulation experiment
20 0
0.0
0.2
0.4
0.6
0.8
1.0
P / bar Figure 1. Comparison of the simulation adsorption isotherm for hydrogen with experiment in ZIF-8 at 77K 3.2
Adsorption isotherm
As we known, it is more convenient to simulate the adsorption isotherms of mixtures at various pressures rather than that obtained by experiment. The adsorption isotherms of equimolar methane-ethane-propane ternary mixtures in ZIF -8 were plotted in Fig 2, it can be observed that the longer chain component is preferentially adsorbed at low pressure, when the pressure increases, the adsorption amount of propane reaches the maximal value and then decreases. As to the shorter chain component, the adsorbed amounts start to increase at higher pressure, then increase with pressure increasing. The adsorption isotherms of C1–C3 alkane mixtures in ZIF-10 were also presented in Fig. 3, every component is adsorbed independently at different pressures. As the pressure increase, due to the larger adsorption heat between the longest chain component and the frameworks of ZIFs, the propane first increase to maximum then decrease. The longer chain alkane molecules seem to be squeezed out the pore channels by relative shorter chain alkane at higher pressures. As can be seen in Fig 2 and Fig 3, the adsorbed amounts of methane/ethane/propane in ZIF-10 are larger than that in ZIF-8, which can be attributed to the lager pore channels of ZIF-10. The adsorbed amount of propane reaches the maximal around 1000 KPa, while other components do not reach their peaks at this pressure, which suggests that the zeolitic imidazolate frameworks may display good separation ability at this condition. Therefore, the separation behavior of alkane mixtures discussed below would be investigated at this condition.
Li zhang, et al. / Procedia Computer Science 4 (2011) 1193–1202
adsorbed amount/ mol. kg
-1
1196
methane ethane propane
3 2 1 0 -3
-2
10
-1
10
0
10
10
1
2
10 10 P / kPa
3
10
4
10
5
10
adsorbed amount/ mol. kg
-1
Figure 2. Adsorption isotherms of equimolar methane-ethane-propane ternary mixtures at 298K in ZIF-8 8
methane ethane propane
6 4 2 0 -2
10
-1
10
0
10
1
2
10 10 P / kPa
3
10
4
10
5
10
Figure 3. Adsorption isotherms of equimolar methane-ethane-propane ternary mixtures at 298K in ZIF-10 3.3
Separation behavior
As discussed in our previous study, the most interested and important factor is the selectivity of the nonporous materials for different components. The relative selectivity ( D ) of adsorbent is determined by the ratio product of mole fractions of components A to B in adsorption phase and in gas phase. In this work, the relative selectivities of ZIF-8 and ZIF-10 for different mole fraction of mixtures were calculated and shown in Fig. 4. The mole fraction of propaneǃethaneǃand methane in gas phase is fixed to 0.33, respectively, and then the mole fraction of other two components varied correspondingly. The results show that all the selectivity of ZIFs are greater than 1.0, it suggests that ZIFs can store more longer chain component, which can be attributed to the stronger interaction between the longer chain component and ZIFs. As shown in Fig 4a, when the molar fraction of propane fixed, the relative selectivity of ZIF-8 and ZIF-10 increase rapidly with the molar fraction of methane in gas phase increase. Similar to that shown in Fig 4a, the relative selectivity of ZIFs for
1197
Li zhang, et al. / Procedia Computer Science 4 (2011) 1193–1202
propane/methane system (xethane fixed to 0.33) also increase, its value is much larger than that shown in Fig 4a, which can be explained from the adsorption isotherm. From Fig 2 and Fig 3, it can be observed that the adsorption amount of propane have reach the maximum at the pressure of 1000Kpa, while the methane molecules have just began to be adsorbed at that pressure. These results suggest that ZIF-8 and ZIF-10 could be considered as one of the most potential separation material. As to the propane/ethane system, the selectivity change trend of ZIF-8 is similar to that discussed above, it increases with the mole fraction increasing. Whenas the change trends of ZIF-10 is quite different from ZIF-8, it decreases to the minimum than increases, which may be attributed to the competition of size effect and entropy effect. In summary, the selectivity of ZIF-10 is little higher than that of ZIF-, but it is much lower than other zeolite, it may caused by the relative larger pore channels of ZIFs.
5.5
ZIF-8 ZIF-10
4.5
2
6
DC H /CH
4
5.0
4.0 a
3.5 0.0
0.1
0.2
0.3
0.4
y CH
0.5
0.6
4
24 ZIF-8 ZIF-10
16
3
8
DC H /CH
4
20
12 8 4 0.0
b 0.1
0.2
0.3
yCH
4
0.4
0.5
0.6
1198
Li zhang, et al. / Procedia Computer Science 4 (2011) 1193–1202
4.0
ZIF-8 ZIF-10
3
8
2
DC H /C H
6
3.5
3.0
c 0.0
0.1
0.2
0.3
yC H
0.4
0.5
0.6
2 6
Figure 4. The Selectivity of methane-ethane-propane system in ZIF-8 and ZIF-10 when fixed one of the component, (a) the mole fraction of propane fixed to 0.33, (b) the mole fraction of ethane fixed to 0.33, (c) the mole fraction of methane fixed to 0.33. 3.4
Adsorption sites
It is well known that the adsorption and separation behavior of mixtures is closely influenced by the adsorption mechanism. Therefore, with the aim of furtherly understanding the adsorption mechanism from molecular level, a technique used in our preview work [22] named ‘Computer Tomography for materials’(mCT) was introduced to investigate the adsorption sites of alkane mixtures in ZIF-8 and ZIF-10 at different pressures. In our previous works[22], we have proven that the adsorption sites of hydrogen molecules in MOFs obtained by mCT technique is well in agreement with that obtained by XRD and neutron powder diffraction. Recently, it is also employed to reveal the hydrogen adsorption sites in ZIF-8 by zhou et. al,[24] and the results is also consistent with the neutron diffraction experiment result published by Yildirim[23], which suggests that the mCT technique is an effective way to illustrate adsorption mechanism of adsorbed molecules in ZIFs. It is very convenient to investigate the adsorption sites for adsorbate in microporous from the mCT imagine. During the simulation process, the entire configuration of adsorbed molecules were recorded every 1000 Monte Carlo steps, and then the density distribution of adsorbate was calculated statistically on the basis of all of these configurations, finally the mCT images were obtain by compute imaging. From mCT images, the redder, the higher densities of alkane molecules in this region, the stronger alkane adsorb ability of
the adsorption sites. The mCT images of methaneǃethane and propane in ZIF-8 at low pressure(0.1Kpa) were calculated and plotted in Figure 5, respectively, it can be easily found that there is nearly no methane molecules adsorbed in ZIF-8. As shown in figure 2, the adsorption amount of ethane and propane at low pressure is just a few, therefore, the adsorption sites of them can not be clearly observed from the mCT images. When the pressure increases to 100KPa, more and more alkane molecules adsorbed in ZIF-8, the adsorption sites become more and more legible.(Figure.6) It can be observed that the methane molecules was gradually adsorbed, most of them are preferential adsorbed around the 2-methylimidazolate organic linkers of ZIF-8 and partial to the C=C bond, which is consistent with the adsorption mechanism of hydrogen in ZIF-8 discussed by Yildirim through neutron diffraction. [23] However, the alkane adsorption sits in ZIFs are greatly different from those in MOFs. In our previous work [22], we have revealed that the predominant binding site of alkane isomer mixtures in IRMOFs are near the inorganic secondary building units, then adsorbed around the organic linker. As to the alkane adsorbed in ZIFs, the primary adsorption site located
1199
Li zhang, et al. / Procedia Computer Science 4 (2011) 1193–1202
around the organic linkers and close to C=C bond, which is contrary to that in MOFs, it can be attributed to the steric effect of the methyl group of the organic linker of 2-methylimidazolate as proposed by Zhou et.al[24]. The adsorption mechanism of ethane and propane molecules are similar to that of methane molecules shown in Figure 6a, there is also some alkane molecules adsorbed around the ZnN4 cluster. When the pressure continue increases to10000Kpa (Figure.7), the mCT images of alkane molecules were also similar to that shown in Figure 6, in other words, alkane molecules still adsorbed around the preferential adsorption sites even the pressure as high as 10000Kpa. It can be easily concluded that there is no competition of energy effect and size effect between these alkane molecules by comparing the mCT images of methaneǃethaneǃpropane molecules in ZIF-8 at different pressure. The adsorption behavior of all alkane molecules are controlled by the energy effect, which is quite different from that in Cu-BTC discussed in our previous works[25], they indicated thatthere is a competition between the energy effect and the size effect during the alkane mixtures adsorbed in Cu-BTC. This difference maybe attributed to the different structure of microporous materials, nearly all the place of the spherical pore of ZIF-8 is equivalent to each other, and all the C=C bonds of the 2-methylimidazolate face to the spherical pore of ZIF-8. Thus , due to the strong interaction between adsorbed molecules and ZIF frameworks and the steric effect of the methyl group, all the alkane molecules can be adsorbed in the spherical pore of ZIF-8 and closer to the C=C bond of 2-methylimidazolate
(a)
(b)
(c)
Figure 5. mCT images of methane(a),ethane(b)and propane(c)for methane-ethane-propane mixtures in ZIF-8 at 0.1 kPa
(a)
(b)
(c)
Figure 6. mCT images of methane(a)ˈethane(b)and propane(c)for methane-ethane-propane mixtures in ZIF-8 at 100 KPa
1200
Li zhang, et al. / Procedia Computer Science 4 (2011) 1193–1202
(a)
(b)
(c)
Figure 7. mCT images of methane(a),ethane(b)and propane(c)for methane-ethane-propane mixtures in ZIF-8 at 105 KPa The adsorption sites of methaneǃethane and propane molecules in ZIF-10 at different pressure was also explored and plotted in Figure 8-10, respectively. Form Figure 8, it can be observed that the density distribution of alkane molecules is not so concentrated as that in ZIF-8, it may be caused by the relative larger pore of ZIF-10, some alkane molecules are adsorbed dispersedly around the ZnN4 cluster. When the pressure increases to 100Kpa, all the adsorbed alkane molecules are firstly preferential adsorbed in the four-ring aperture, then some other molecules adsorbed around the metal-organic corner. The relative strong adsorption sites in the four-ring pore channel might result form the coupling of four ZnN4 cluster that close to each other. Some ethane and propane molecules are also adsorbed around the ZnN4 cluster of the eight-ring and formed a adsorption sites circle in the eight-ring aperture. As shown in Figure 9b and 9c, the mCT images of ethane and propane is similar to each other. It is well known that the adsorption heat for methane is smaller than that of ethane and propane, the adsorption amount of alkane mixture increase with the pressure increasing, some methane molecules are gradually extruded from the four-ring pore channel owing to their weaker interaction, that is, the adsorption behavior is mainly controlled by the energy effect. As the pressure increase to 10000kPa (Figure10), more and more alkane molecules adsorbed around the ZnN4 cluster and the six-ring and eight-ring pore channels were also occupied, while the methane molecules are squeezed out the four-ring channel completely by other short channel alkane molecules. In other words, ZIF-10 have a distinguish effects for CH4, Whenas, ZIF-8 do not. it can be effectively employed to explain different selectivity ZIF-10 and ZIF-8 discussed above. Through the analysis of the adsorption sites at different pressures by mCT technique, it can be found that all the alkane molecules are preferential adsorbed around the 2-methylimidazolate organic linkers of ZIF-8 and partial to the C=C bond, there is no competition between energy effect and the size effect during the adsorption in ZIF-8. As to ZIF-10, methaneǃethaneǃpropane molecules were firstly preferential adsorbed in the four-ring aperture, then adsorbed around the ZnN4 cluster and the six-ring and eight-ring pore channel. As the pressure increases, the energy effect is preferential to the size effect, methane is extruded from the four-ring pore owing to the relative weak interaction, and the methane molecules mainly located around the imidazolate organic linkers of eight-ring pore channel, the six-ring pore channel is also occupied, i.e the adsorption behavior of methane molecules is controlled by energy effect.
1201
Li zhang, et al. / Procedia Computer Science 4 (2011) 1193–1202
(a)
(b)
(c)
Figure 8. mCT images of methane(a),ethane(b)and propane(c)for methane-ethane-propane mixtures in ZIF-10 at 1 KPa
(a)
(b)
(c)
Figure 9. mCT images of methane (a),ethane (b) and propane(c) for methane-ethane-propane mixtures in ZIF-10 at 100 KPa
(a)
(b)
(c)
Figure 10. mCT images of methane (a), ethane (b) and propane(c) for methane-ethane-propane mixtures in ZIF-10 at 105 KPa
Conclusion The grand canonical Monte Carlo combined with the configurational-bias Monte Carlo simulation technique was employed to study the adsorption isotherm of short linear alkane mixtures in ZIF-8 and-10.It is found that the adsorbed amounts of alkane molecules in ZIF-10 are larger than that in ZIF-8, and the adsorbed amount of alkane molecules increases with pressure increasing, the longer chain alkane molecules, the higher adsorption amount. The
1202
Li zhang, et al. / Procedia Computer Science 4 (2011) 1193–1202
selectivity of the ZIFs for different components were further investigated and discussed, the selectivity of ZIF-10 is little higher than that of ZIF-8, but it is much lower than other zeolite. The adsorption sites of alkane mixtures obtained by mCT images, it can be found that all the alkane molecules are preferential adsorbed around the 2methylimidazolate organic linkers of ZIF-8 and partial to the C=C bond, there is no competition between energy effect and the size effect during the adsorption in ZIF-8. As to ZIF-10, methaneǃethaneǃpropane molecules were firstly preferential adsorbed in the four-ring aperture, then adsorbed around the ZnN4 cluster and the six-ring and eight-ring pore channel. When the pressure increases, the energy effect is preferential to the size effect, methane is extruded from the four-ring pore owing to the relative weak interaction, and the methane molecules mainly located around the imidazolate organic linkers of eight -ring pore channel, the six-ring pore channel is also occupied, the adsorption behavior of methane molecules is controlled by energy effect.
Acknowledgements This work was financially supported by the National Natural Science Foundation of China (Grant No. 20904048) and the Scientific Research Fund of Zhejiang Provincial Education Department(No. Y200805980).It is also supported by Science foundation of Zhejiang SCI-Tech University(ZSTU)under Granted No 0813825-Y and the Program for Changjiang Scholars and Innovative Research Team in University (Grant No. IRT 0654)
References 1. 2. 3. 4. 5. 6. 7. 8. 9. 10. 11. 12. 13. 14. 15. 16. 17. 18. 19. 20. 21. 22. 23. 24. 25.
H.Van Bekkum, E.M. Flanigen, P.A.Jacobs, J.C. Jansen (eds), Introduction to Zeolite Science and Practice Elsevier, Amsterdam, 2001 D.W.Breck, Zeolite Molecular Sieves Wiley, New York, 1974. G. T. Palomino, M. R. L. Carayol, C. O. Areán, J. Mater. Chem. 16 (2006) 2884 W. Yang, X. Lin, J.H. Jia, A.J. Blake, C. Wilson, P. Hubberstey, et.al Chem. Commun, (2008) 359 S.B. Pu, Y.Tanaka,T. Inui, Sep. Techno. 6 (1996)189 R. Krishna, J.M. Baten, Sep. Purif. Techno. 55 (2007) 246 X.C. Huang, Y.Y. Lin, J.P. Zhang, X.M. Chen, Angew. Chem. Int.Ed. 45 (2006) 1557. K.S. Park, A.P. Cote, J.Y. Choi, R. Huang, F.J. Uribe-Romo, H.K. Chae,et.al Proc. Nat. Acad. Sci. 103 (2006). 10186. B. Smit, R.Krishna, Chem. Eng. Sci. 58 (2003) 557-568 S. Chempath, J.F.M. Denayer, K.M.A. De Meyer, G.V. Baron, R.Q. Snurr, Langmuir. 20 (2004) 150. J.W. Jiang, S.I. Sandler, Langmuir. 22 (2006) 5702. J.W. Jiang, S.I. Sandler, M. Schenk, B. Smit, Phys. Rev. B. 72 (2005) 045447. D. Dubbeldam, C.J. Galvin, K.S. Walton, D.E. Ellis, R.Q. Snurr, J. Am. Chem. Soc. 130 (2008) 10884 L.H. Lu, Q. Wang, Y.C. Liu, J. Phys. Chem B. 109 (2005) 8845. T. Duren, R.Q. Snurr, J. Phys. Chem. B. 108 (2004) 15703. L. Zhang, Q. Wang, T. Wu, Y.C. Liu, Chem. Eur. J. 13(2007) 6387. R. Krishna, B.Smit, S. Calero, Chem. Soc. Rev. 31 (2002) 185 R. Banerjee, A. Phan, B.Wang, C. Knobler, H. Furukawa, M. O!Keeffe, et. Al .Science. 319 (2008) 939-943 T.J.H. Vlugt, W.Zhu, F.Kapteijn, J.A. Moulijn, B. Smit, R. Krishna, J. Am. Chem. Soc., 120 (1998) 5599. S.L. Mayo, B.D. Olafson, W.A. Goddard III, J. Phys. Chem. 94(1990) 8897 W.L. Jorgensen, J.D. Madura, C.J. Swenson, J. Am. Chem. Soc. 106 (1984) 6638. L. Zhang, Q. Wang, Y.C. Liu, J. Phys. Chem. B. 111 (2007) 4291. Z .Wei, W. Hui, T.J. Udovic, J.J. Rush, T. Yildirim, J. Phys. Chem. B. 112 (2008) 12602. M.Zhou, Q.Wang, L. Zhang, Y.C. Liu, Y. Kang, J. Phys. Chem. B. 113 (2009) 11049. D.Chen, L. Zhang, Y.C. Liu, Q.Wang, Acta Chimcia Sinica. 66 (2008) 2227.