Zwitterion Hydration and Ion Association Design Principles Elucidated by Simulation and Machine Learning

Design principles connecting material properties to underlying chemistry and processing are highly effective tools for aiding the discovery of materials with advanced and targeted properties. Recently, computational techniques, such as high-throughput simulations and machine learning, have accelerated the discovery of design principles and are being used to quantify the impact of changing molecular chemistry, processing conditions, etc. on resultant material properties. These methods can require a large volume of diverse data with standardized methodologies to produce accurate quantitative models. In this contribution, molecular dynamics simulations and machine learning models are implemented to examine the influence of molecular chemistry on hydration, ion association, and ionic conductivity properties of zwitterions and polyelectrolyte materials. Simulations are used to produce hydration and ion association data at the atomic scale for zwitterion molecules and polymers, then these data are incorporated with cheminformatic descriptors to train and evaluate machine learning models revealing quantitative relationships between design and property at the molecular, polymeric, and system scales.