Nanoporous materials are excellent candidates of materials membrane separation. Their ordered and atomic scale porous networks are size and shape selective to different guest molecules, performing effective and low energy cost separations of mixtures. In this thesis we present a general methodology for predicting the macroscopic flux of multicomponent mixtures through porous membranes with information obtained from detailed atomistic simulations. The feasibility of such atomistic simulations is based on the known detailed atomic level structure of the porous materials.

We first applied atomically detailed simulations to predict the binary permeance of CH₄/H₂ mixtures through defect-free (10,10) single walled carbon nanotubes acting as membranes at room temperature. Earlier simulations indicated that single-component gas diffusion in carbon nanotubes is extremely rapid compared to other known nanoporous materials. Our results show that this observation also holds for binary mixtures adsorbed in carbon nanotubes. As a result, carbon nanotube membranes are predicted to exhibit extremely high fluxes. Our results predict that carbon nanotubes are strongly selective for CH₄ over H₂ when a mixture of these gases permeates through a membrane.

We extended our study to CH₄/CO₂ adsorption and diffusion in DDR zeolite. The comparison of the diffusivities between experimental measurement and Molecular Dynamic (MD) simulation results lead to an analysis on the sensitivity of diffusion to the DDR structure. We developed a lattice model to study the CH₄ and CO₂ diffusion in DDR. The qualitative agreement between the results from MD simulation and lattice model was the basis of our future work.

Most molecular simulations of binary adsorption are performed using grand canonical Monte Carlo (GCMG) to independently examine a range of discrete state points of interest. We presented a Transition Matrix State Monte Carlo (TMMC) algorithm for the calculation of binary mixture adsorption isotherms in porous materials. At the completion of a TMMC simulation, the adsorption isotherm for all possible bulk phase compositions and pressures is available without data fitting or interpolation. Finally, we examined the intrinsic accuracy of Ideal Adsorbed Solution Theory (IAST) by an algorithm based on TMMC and histogram reweighting.