Diffusion of CH4 and H2 in ZIF-8
Highlights
► First molecular dynamics simulations of a fully flexible ZIF-8. ► Inclusion of lattice flexibility proves to be absolutely necessary. ► Reasonable agreement between simulation and experimental measurements is obtained.
Introduction
Metal-organic frameworks (MOFs) are very promising for numerous industrial applications such as membrane gas separation, gas storage or heterogenous catalysis. Several simulation studies [1] provide new insights into the diffusion and adsorption of molecules within these fascinating structures. The high flexibility of MOFs can lead to surprising effects [2], [3] which indicate that the common approximation of a rigid lattice may not be generally applicable for all host–guest combinations on MOFs. This holds true especially for guest molecules with a kinetic diameter similar to the diameter of the windows or channels of that particular MOF.
Due to exceptional thermal and chemical stability, the MOF subgroup of zeolitic imidazolate frameworks (ZIFs) [4], [5] is of particular interest. These frameworks are based on metal ions connected to imidazolate type linkers and frequently they resemble zeolitic structures. As one of the many different ZIFs that resulted from this approach, ZIF-8 crystallizes in sodalite topology (SOD) and consists of Zn(II) tetrahedrically coordinated by four 2-methylimidazolate linkers. For the diffusion of guest molecules, the passage through the eight six-membered windows connecting each cavity, is the rate limiting step. The window size is estimated to 3.4 Å by X-ray diffraction structure analysis [4]. Bux et al. [6] succeeded in synthesizing the first ZIF-8 membrane (Fig. 1) and measured a relatively high selectivity for H2 in H2/CH4 mixtures. For a molecular understanding of the permeabilities of H2 and CH4 through ZIF-8 membranes, knowledge about the microscopic details of diffusive mass transport are highly desirable. As a proof for molecular sieving, this new ZIF-8 membrane (Table 1) shows selectivities in the binary gas separation which are much higher than the Knudsen selectivity .
However, the measured selectivity is not infinite, which would be expected in the case of ideal molecular sieves. The CH4 molecules with a kinetic diameter of about 3.8 Å should not be able to enter the pore system with crystallographic windows of only 3.4 Å diameter. In contrast, actual diffusion measurements with infra-red microcopy (IRM) on single ZIF-8 crystals confirmed sorption uptake of CH4 into the ZIF-8 (Fig. 5).
Section snippets
Simulation methods and theory
We designed our ZIF-8 model based on the X-ray diffraction data of the Cambridge Structure Database and our simulation cell consisted of 2 × 2 × 2 unit cells with two cavities in each unit cell (1 uc = Zn12(mim)24 with mim = 2-methylimidazolate). Fig. 2 shows the simulation cell and the structure of the windows of the ZIF-8 crystal. The software package DL_POLY (see http://www.cfs.dl.ac.uk/tutorials/dlpoly/dlpoly.tutorial.pdf) was used for the MD simulations. Since the H2 molecule is small and because
Results
While DREIDING has already shown to fail in flexible nonbonded approaches [20], it also exhibits problems for our bonded system. In the way that is provided by the DL_POLY – tool DREIDING suggests angles around zinc and nitrogen that slightly differ from those of the structure determined by X-ray diffraction. This results in a turning of the linkers which effectively increases the window size. In addition, the Zn–N bond force constant given by DREIDING is about 18 times higher than the one
Conclusions
In order to understand the high H2/CH4 selectivity of ZIF-8 membranes, we performed molecular dynamics simulations in an entirely flexible model of the ZIF-8 structure. The simulations indicate that this system is much more sensitive with respect to the choice of interaction parameters than e.g. the more rigid zeolites. For the AMBER parameter set we even found, that flexibility as the crucial factor made it possible for the CH4 molecules to change cavities. The forecasted membrane selectivity
Acknowledgements
The authors would like to thank J. Kärger for enlightening discussions and Msc. K. Seeharmat for technical support. We thank the DFG for funding within the framework of SPP1362.
References (32)
- et al.
Micropor. Mesopor. Mater.
(2010) - et al.
Micropor. Mesopor. Mater.
(2010) - et al.
J. Mol. Struct.
(2000) - et al.
Chem. Eng. J.
(2007) - et al.
Micropor. Mesopor. Mater.
(2009) - et al.
J. Mol. Struct.
(2000) - et al.
Chem. Soc. Rev.
(2009) - et al.
diffusion-fundamentals.org
(2009) - et al.
Proc. Natl. Acad. Sci. U.S.A.
(2006) - et al.
Acc. Chem. Res.
(2010)