New techniques to simulate electrified solid-liquid interfaces better and faster

Abstract

Electrochemistry is at the heart of many processes that are key for the present and the foreseeable future of humanity. Batteries rely on electrochemical reactions to store and release energy, and are currently key for the transition to renewable energy sources. Electrocatalysis uses electrochemical reactions to drive chemical changes that synthesize valuable products or destroy hazardous ones (e.g. in water treatment). Corrosion is often an electrochemical process that needs to be minimized to increase the lifetime of materials and infrastructure. Seeing the importance of these processes and the significant success that society has had in controlling them, one might think that we have a good understanding of the underlying physics. However, most advances in these fields have been based on empirical observations that we don't fully understand and despite that they result in very succesful applications. One of the main reasons behind this lack of understanding is that the processes are hard to simulate computationally because (1) often voltage needs to be applied across the system to simulate realistic scenarios and (2) systems required to simulate the processes are big. In this thesis I tackle both problems by using the Non-Equilibrium Green's Functions formalism to solve the first problem and equivariant Graph Neural Networks that predict electron densities to solve the second one. In the last chapter, I merge both solutions in a promising proof of concept application that hopefully will be continued by future researchers.

Date of Award	2 Dec 2024
Original language	English
Supervisor	Pablo Jesús Ordejón Rontome (Director) & Jose Miguel Alonso Pruneda (Director)