Membrane based separations are widely used in diverse applications across various chemical industries. The design of membranes for selective separation of specific components from gaseous mixtures, requires an accurate evaluation of both the membrane permeability and selectivity. Using all atom molecular dynamics simulations and trajectory extending kinetic Monte Carlo (TEKMC) we have recently shown that evaluating the self-diffusivity in CO2/N2 gas mixtures in carbon molecular sieves, alters the membrane performance on Robeson plots [J. Phys. Chem. B 2023, 127, 9841−9849]. However the influence of the transport or mutual diffusivity on the membrane performance has not been evaluated. The membrane permeability is related to the mutual diffusion coefficients of the different components in the gas mixture with contributions from both the self diffusivities and distinct diffusivities. However, many studies rely primarily on the pure component self-diffusivity to estimate the membrane permeability. To address this, we perform all-atom molecular dynamics simulations to compute the mutual diffusion coefficients using the Maxwell–Stefan approach, based on the Einstein relations for self- and cross-correlation of particle displacements in CO₂/N₂ and CO₂/CH₄ mixtures confined within carbon molecular sieve membranes. Our study examines the need for incorporating accurate values of the mutual diffusivities while assessing the performance on Robeson plots and aid in the design of efficient membranes for gas separation applications.
