Predicting Molecular Diffusion in Flexible Metal Organic Frameworks

Membrane-based separations based on nanoporous materials rely crucially on molecular diffusion in these materials. Unfortunately, knowledge of diffusion is still limited to small group of the very large number of nanoporous materials that are interesting candidates for membrane separations. To diversify nanoporous materials-based separation based on MOFs we built up a large diverse library and used molecular dynamics (MD) simulation to comprehensively understand how the pore structures and molecule species influence molecular diffusion. 18 representative MOFs were chosen from the CoRE MOF database and 12 adsorbate molecules covering different size, shape, polarity and flexibility are selected. UFF4MOF was employed to describe the interactions between the structure atoms. Comparisons with experimental crystal structures and with DFT-relaxed structures indicated that UFF4MOF gave satisfactory predictions of the MOF structures. Dynamically corrected transition-state theory methods (dcTST) was used to predict diffusivities of slow diffusing systems. Umbrella sampling combined with the one-dimensional weighted histogram analysis method (WHAM) was employed to calculate the diffusion free energy barriers. All of the MD simulations included flexibility of the MOFs, which is critical in making accurate predictions.

Our results make it possible for the first time to systematically compare molecular diffusivities for a library of molecules in a diverse range of MOFs with calculations that fully incorporate framework flexibility. The physical origins of the trends found in this diverse data set will be discussed. These observations suggest paths forward to developing methods that could be used in accurate high-throughput screening of large libraries of nanoporous structures and molecules for kinetic separations.