An Efficient Method to Calculate the Diffusivity of Non-Spherical Molecules in Nanoporous Materials

Conventional methods for predicting diffusion coefficients in nanoporous materials are computationally expensive and the difficulty increases for non-spherical gas molecules and for materials with extremely small pores. Whereas the problem is partially alleviated by recent development in fast computers, the data-driven approach to materials design often requires a large amount of data over a broad range of conditions. To develop a more efficient method to calculate the diffusion coefficient, we extend the sting method for predicting the minimum free energy pathway (MFEP) by introducing collective variables applicable to non-spherical molecules. The combination of MFEP with the transition-state theory (TST) allows us to predict diffusion coefficients in good agreement with molecular dynamics simulation but with much less computational cost. The new theoretical procedure is promising for applications to data-driven approach to inverse design of nanoporous materials for membrane-based separation.