First-Principles-Based Machine Learning Models for Phase Behavior and Transport Properties of CO₂

Abstract

In this work, we construct distinct first-principlesbased machine-learning models of CO 2 , reproducing the potential energy surface of the PBE-D3, BLYP-D3, SCAN, and SCAN-rvv10 approximations of density functional theory. We employ the Deep Potential methodology to develop the models and consequently achieve a significant computational efficiency over ab initio molecular dynamics (AIMD) that allows for larger system sizes and time scales to be explored. Although our models are trained only with liquid-phase configurations, they are able to simulate a stable interfacial system and predict vapor−liquid equilibrium properties, in good agreement with results from the literature. Because of the computational efficiency of the models, we are also able to obtain transport properties, such as viscosity and diffusion coefficients. We find that the SCAN-based model presents a temperature shift in the position of the critical point, while the SCAN-rvv10-based model shows improvement but still exhibits a temperature shift that remains approximately constant for all properties investigated in this work. We find that the BLYP-D3-based model generally performs better for the liquid phase and vapor−liquid equilibrium properties, but the PBE-D3-based model is better suited for predicting transport properties.

Molecular Simulation of Lithium Carbonate Reactive Vapor–Liquid Equilibria Using a Deep Potential Model

“…While dispersion corrections were included in the AIMD simulations and consequently in the calculation of energies and forces of the configurations in the training data set, the short-range nature of the DP model allows us to take into account long-range effects only implicitly. The short-range model can still accurately predict liquid properties of the systems due to screening effects; however, inclusion of the long-range effects can further improve vapor and interface properties. , Long-range effects can be considered using the DP long-range (DPLR) models. However, DPLR models are approximately 5 times slower than DP models considerably increasing the computational expenses of the simulations …”

“…Moreover, the box size should be large enough to accommodate vapor-phase components, making the use of this method with direct AIMD simulations very challenging. DP models have previously shown their efficiency in direct coexistence simulations of nonreactive systems, , but, to the best of our knowledge, have not been used to study reactions with vapor components. Here, we develop and use a DP model to show the dissociation of Li 2 CO 3 into vapor CO 2 and describe the details of the reaction process.…”