(556h) Using Trajectory-Extending Kinetic Monte Carlo Simulations to Reassess Membrane Selectivities for Gas Mixtures in Carbon Molecule Sieves

A reliable estimation of diffusion coefficients from molecular dynamics simulations is required for evaluating membrane permeability and performance for a wide range of adsorbents for gas separations. Determining self diffusivities and transport diffusion coefficients in gas mixtures from equilibrium molecular dynamics simulations is challenging due to the long time scales involved in accessing the diffusive limit. Gases diffusing through dense carbon molecular seive and polymeric membranes experience extended subdiffusive regimes, which hinders reliable extraction of diffusion coefficients from mean squared displacement data. We improve the sampling of the diffusive landscape by implementing the trajectory-extending kinetic Monte Carlo (TEKMC) technique to efficiently extend molecular dynamics trajectories from ns to μs time scales. The TEKMC algorithm is used to evaluate the mixture diffusivities in binary mixtures to determine the permselectivity of CO₂ in CH₄ and N₂ mixtures in dense carbon molecular sieve membranes. Interestingly Robeson plot comparisons show that the permselectivity obtained from pure gas diffusion data is significantly lower than that predicted using mixture diffusivity data. Specifically, in the case of the CO₂/N₂ mixture, we find that using mixture diffusivities led to permselectivities lying above the Robeson limit highlighting the importance of using mixture diffusivity data for an accurate evaluation of the membrane performance. Combined with gas solubilities obtained from grand canonical Monte Carlo (GCMC) simulations, our work shows that simulations with the TEKMC method can be used to reliably evaluate the performance of materials for gas separations. Additionally, the evaluation of membrane performance from pure component diffusion coefficients has to be carefully reassessed in light of our results.