Graph-Based Modeling and Molecular Dynamics for Ion Activity Coefficient Prediction in Polymeric Ion-Exchange Membranes

Abstract

The partitioning of ions between polymeric ion-exchange membranes (IEMs) and the surrounding liquid is governed by the activity coefficients of the ions, which, in turn, significantly impact various ion transport processes within these membranes, notably conductivity. This study introduces a computational framework to predict ions’ activity coefficients in charged ion-exchange membranes (IEMs). This method employs a machine learning (ML) model using molecular-scale characteristics obtained from molecular dynamics (MD) simulations, particularly by emphasizing solvation properties within the context of IEMs. Specifically, the framework utilizes graph convolutional networks (GCN) to establish connections between the chemical structure of the polymer and the molecular-level attributes. This ultimately leads to determining macroscopic attributes, such as the activity coefficient, across a range of IEM materials having random copolymer and block copolymer systems. Furthermore, saliency maps were generated to identify the critical features of polymer molecules that correlate with the ion activity coefficients. The graph-based prediction strategy proved highly accurate in predicting ion activity coefficients within IEMs, even with relatively small training data sets.