Parametrized, Transferable Classical Density Functional Theory for Alkane/Alkene Separation in Cationic Zeolites

We present a classical density functional theory (DFT) enhanced with a binary solid-fluid interaction parameter, <i>k_si</i>, for the accurate and efficient prediction of the adsorption of small alkane/alkene mixtures in cation-exchanged zeolites. The parameter <i>k_si</i> scales the van der Waals cross-interaction energies as a factor (1 - <i>k_si</i>) to the Berthelot combining rule, and it turns out to be transferable among zeolite structures. Focusing on sodium-exchanged Faujasite (FAU) and Linde Type A (LTA) zeolites, we have parametrized <i>k</i>_Z,π and <i>k</i>_Na,π to account for interactions between the propene double bond and framework zeolite atoms or the extraframework cations, respectively. Each parameter was fitted to a single grand canonical Monte Carlo (GCMC) propene isotherm. The resulting classical DFT predictions demonstrate excellent agreement with GCMC simulations across a wide range of conditions, including variations in temperature, Si/Al ratio, aluminum distribution, topology, and in propane/propene mixtures. We also evaluated the applicability of the method to calcium-exchanged zeolites, revealing the limitations of the fixed-cation approximation in systems where the effects of cation mobility and pore accessibility are significant. These findings demonstrate that classical DFT with transferable <i>k_si</i> parameters is a robust alternative to empirical adsorption isotherm models, although it has limitations. It combines predictive accuracy with a computational cost that is orders of magnitude lower compared to molecular simulations. This approach allows for high-throughput screening and interpolation or extrapolation of pure and mixed-gas adsorption isotherms in monovalent cationic zeolites. There is potential to extend this approach to other porous materials, such as metal-organic frameworks, or to other weakly polar adsorbate systems.