Dynamic Monte Carlo simulations were used to examine binary diffusion in ZSM-5. Diffusion in zeolites was strongly influenced by dynamic and static heterogeneity. The former described the concentration dependent influence of diffusing molecules on each other within the restricted environment of the zeolite pore network. The latter was related to the presence of adsorption sites of different strengths, at which molecules adsorbed for different average times, before moving on to a neighboring adsorption site. Diffusion quantum Monte Carlo simulations were an effective tool for studying the influence of both forms of heterogeneity upon diffusion. The self-diffusivities, as determined from diffusion quantum Monte Carlo, depended strongly upon the pore network topology, the average residence times of both species on each type of site and the concentrations of those species. The Maxwell-Stefan approach was known to predict rather well diffusion in silicalite-1, the statically homogeneous version of ZSM-5. However, it was not rigorous in predicting the self-diffusivity in the presence of any form of strong static heterogeneity. The ability of the Maxwell-Stefan approach to model the self-diffusion of binary mixtures in ZSM-5 was compared to diffusion quantum Monte Carlo results; which served as a benchmark. Alternative theories (effective medium approximation, mean field) which accounted for the discrete sites of the lattice were considered for their ability to predict the self-diffusivity in binary mixtures. The effective medium approximation was shown to be a promising method in situations where the Maxwell-Stefan approach could not work due to the presence of strong correlation effects.

Dynamic Monte Carlo Simulations of Binary Self-Diffusion in ZSM-5. X.Liu, D.Newsome, M.O.Coppens: Microporous and Mesoporous Materials, 2009, 125[1-2], 149-59