Graph-Based Modeling\nand Molecular Dynamics for Ion\nActivity Coefficient Prediction in Polymeric Ion-Exchange Membranes

The partitioning of ions between polymeric ion-exchange\nmembranes\n(IEMs) and the surrounding liquid is governed by the activity coefficients\nof the ions, which, in turn, significantly impact various ion transport\nprocesses within these membranes, notably conductivity. This study\nintroduces a computational framework to predict ions’ activity\ncoefficients in charged ion-exchange membranes (IEMs). This method\nemploys a machine learning (ML) model using molecular-scale characteristics\nobtained from molecular dynamics (MD) simulations, particularly by\nemphasizing solvation properties within the context of IEMs. Specifically,\nthe framework utilizes graph convolutional networks (GCN) to establish\nconnections between the chemical structure of the polymer and the\nmolecular-level attributes. This ultimately leads to determining macroscopic\nattributes, such as the activity coefficient, across a range of IEM\nmaterials having random copolymer and block copolymer systems. Furthermore,\nsaliency maps were generated to identify the critical features of\npolymer molecules that correlate with the ion activity coefficients.\nThe graph-based prediction strategy proved highly accurate in predicting\nion activity coefficients within IEMs, even with relatively small\ntraining data sets.