Reactive Phase Behavior of Molten Alkali Carbonates and Hydroxides using Molecular Simulations and Ab Initio-based Machine Learning Models

Alkali-metal carbonates and hydroxides are widely used in energy and environmental applications due to their appealing properties. At high temperatures, pure carbonates and hydroxides decompose into vapor carbon dioxide and water, respectively. Modeling the reactive vapor-liquid equilibrium of these systems can be crucial in designing their applications. Classical molecular dynamics (MD) simulations are a computationally efficient tool for predicting system properties at different conditions. However, they are not able to model chemical reactions except via detailed, predefined reaction mechanisms. Ab initio molecular dynamics (AIMD) simulations can accurately predict chemical reactions in the systems, but this approach is computationally demanding limiting system sizes and time scales. Machine learning models can overcome these challenges by training neural networks on quantum chemical data. They have shown good results in retaining the accuracy of the underlying ab initio methods while being comparable in efficiency to classical MD simulations. In the current work, we generate machine-learning models for lithium carbonates and hydroxides to study their multiphase equilibria. We perform direct coexistence simulations to analyze dissociation reactions at different conditions. We evaluate our results in terms of system composition, lifetimes, and partial pressures of vapor species. In general, our machine learning model predictions agree well with available experimental results.