(62e) Something Old Something New: Molecular Understanding of Interfacial (electro)Chemistry with Machine Learning and Liquid State Theory

Chemistry at liquid interfaces plays a crucial role from electrocatalysis to atmospheric chemistry. The interfacial environments where these reactions occur can differ substantially from bulk liquid, displaying unique chemical compositions, molecular configurations, and interfacial electric fields that can alter reaction rates and products. In this talk, I will discuss new methods for the molecular simulation of interfacial chemistry that provide the microscopic insight needed to optimize and control these reactions. First, I will describe the application of new approaches in liquid state theory to predict how the thermodynamics of a bulk liquid reaction change when we bring this reaction to the interface, allowing us to separately quantify the contributions from direct solute-interface interactions, interfacial electric fields, and solvent interactions such as the interfacial hydrophobic effect. Applying these tools to the paradigmatic example of NaCl ion pair dissociation, we are able to quantitatively replicate the results of Molecular Dynamics simulations with simple continuum models while simultaneously decomposing the thermodynamics of interfacial reactivity. Second, I will present a new computational method combining Machine Learning based force fields with Constant Potential Molecular dynamics to perform reactive molecular dynamics simulations of working electrochemical cells. These methods enable previously intractable electrochemical simulations by providing (1) the ab-initio accuracy of interatomic interactions needed to describe the surface adsorption and changes in bonding that occur during electrochemical reactions, (2) the fluctuations in electron numbers that occur near a working electrode as current flows through the system, and (3) the efficiency of classical molecular dynamics needed to sample the explicit solvent fluctuations and to apply the rare-event methods needed to describe reaction statistical mechanics.