Data-Driven Design and Discovery of Metal-Organic Framework/Polymer Mixed Matrix Membranes

Abstract

Metal-organic frameworks (MOFs) are one of the most promising classes of porous materials thanks to their tunable chemistry, structural diversity, and a variety of useful structural properties such as very high porosities and extraordinarily large surface areas. Mixed-matrix membranes (MMMs) that incorporate MOF fillers into polymers have emerged to overcome the selectivity-permeability trade-offs inherent to traditional polymer membranes for gas separations. MOF/polymer MMMs are currently at the forefront of membrane research thanks to their ability to bridge the performance gaps of pure polymers and the practical limitations of pristine MOFs. While experimental studies have demonstrated the strong benefit of using of MOF/polymer MMMs for many different types of separations, the rational design of novel MMMs remains a complex, multi-scale challenge. Integration of machine learning (ML) to current experimental and computational studies will be central to unlocking the industrial potential of MOF/polymer MMMs by guiding materials selection, predicting membrane performance, and even synthesis conditions. This perspective explores how ML is reshaping the design and discovery of MOF/polymer MMMs by discussing current progress, opportunities, and challenges of uniting computational innovation with experimental validation to create the next generation of MOF-based MMMs.