Kinetic Monte Carlo simulations were made of the self-diffusion of a methane/perfluoromethane mixture in silicalite. The hopping rates and model set-up were taken to match previously published molecular dynamics simulations and PFG-NMR data. In the case of the present kinetic Monte Carlo simulation the acceleration/deceleration of particles with varying mixture composition could be attributed to correlation effects. The logarithmic interpolation rule for mixture self-diffusion coefficients reported by Snurr and Karger was validated for a broad range of loadings. However, a deficiency of  the present kinetic Monte Carlo model was that it was not able to cope with different saturation capacities of the two species. The influence of different saturation capacities upon mixture diffusion was accounted for by the Maxwell-Stefan theory. The Maxwell-Stefan formulation of diffusion in multi-component mixtures was used to obtain explicit formulas for calculating the diffusivities of binary mixtures within a zeolite matrix. The theoretical development permitted the estimation of the mixture diffusivities on the basis of the pure component diffusivities at zero loading. Several mixture rules for the estimation of the exchange coefficient D12 were considered. Though none of them was fully satisfactory, it was demonstrated that the D12 value should have the same order of magnitude as the pure component Maxwell-Stefan diffusivities in order to account properly for acceleration/deceleration behavior.

Diffusion of Binary Mixtures in Zeolites: Kinetic Monte Carlo versus Molecular Dynamics Simulations. D.Paschek, R.Krishna: Langmuir, 2001, 17[1], 247-54