High-Throughput Computational Exploration of MOFs for Short-Chain PFAS Removal

Short-chain per- and polyfluoroalkyl substances (PFASs) are increasingly replacing regulated long-chain PFASs, yet they remain challenging to remove from water due to their high persistence, mobility, and weak affinity toward conventional adsorbents. In this work, we developed a hybrid high-throughput computational screening (HTCS) strategy to identify high-performance MOFs for the selective adsorption of perfluorobutanoic acid (PFBA), a representative short-chain PFAS, from water. The workflow begins with a curated MOF dataset and employs Monte Carlo (MC) simulations based on synergistic use of a classical universal force field (UFF) and a universal machine-learned interatomic potential (u-MLIP), enabling scalable and quantitatively accurate prediction of adsorption across large MOF databases. A set of promising MOFs initially identified using UFF-based HTCS, that combine strong PFBA affinity and high PFBA/H2O selectivity were re-evaluated with u-MLIP to refine adsorption predictions and to assess guest-induced framework flexibility, enabling the exclusion of materials with unfavourable high water-framework interactions. Ultimately, four high-performance MOFs were identified that optimally balance strong PFBA interactions, high PFBA selectivity over water, and practical considerations including sustainability, water stability, and synthetic feasibility. This study demonstrates that combining classical force fields with u-MLIPs enables scalable, quantitatively accurate MOF adsorption screening and establishes transferable principles for the rational design of adsorbents targeting short-chain PFAS.