Unraveling Separation Mechanisms of Propanol Isomers and Water in LTA Zeolites via Deep Potential Molecular Dynamics

High-purity isopropanol (IPA) is one of the most widely used solvents in numerous industry processes, and its energy-efficient separation from n-propanol (NPA) and water via molecular sieves is of significant practical importance. Although linear NPA is conventionally expected to permeate molecular sieves more readily than branched IPA due to reduced steric hindrance, experiments show that IPA permeation lies between that of water and NPA in Linde Type A (LTA) membranes. Here, we investigate the propanol–water separation process within LTA using deep learning-based molecular dynamics (DPMD) simulations. The results reveal that water molecules not only traverse the nanoscale pore of molecular sieves more readily than alcohol molecules but also promote the migration of Na+ ions toward the eight-membered ring windows, dynamically blocking the pores and modulating the electrostatic environment within the channels. Under these dynamically evolving conditions, branched IPA exhibits lower hindrance and compact conformation, enabling better orientational adaptation within the fluctuating pore environment and faster diffusion compared to linear NPA.