P-039
Gabrielius Mockus
gabrielius.mockus@ftmc.lt
Justinas Šlepavičius, Deimantas Šmigelskas, Rokas Silkinis, Linas Vilčiauskas
Center for Physical Sciences and Technology (FTMC), Lithuania
Analysis of phosphoric acid based systems using machine learning interatomic potentials
Established methods like Density Functional Theory (DFT) are of great importance in computational chemistry, as prominent software packages like the Vienna Ab initio Simulation Package (VASP) and others are built for it. However, even for relatively small systems (e.g., less than 100 atoms), such methods show exponential scaling and can quickly exhaust all available computational resources. This is where Machine Learning Interatomic Potentials (MLIPs) show promise.
Empirical interatomic potentials are fast, but usually cannot provide accuracy comparable to DFT methods, whereas DFT methods are more accurate, but slow and expensive. MLIPs have the potential to bring the best of both worlds: offering faster and lower-cost simulations with accuracy like that of DFT methods. The aim of our research was to construct an MLIP for orthophosphoric acid and systems similar to it. The reason for this is that orthophosphoric acid is one of the best known proton conductors. Analyzing the mechanism behind proton conduction in H3PO4 systems using computational methods may offer insights into understanding the role of proton transfer in living organisms, as well as aid the efforts of creating better fuel cells.
In this research, we have successfully parametrized a reactive MLIP. The results are analyzed in terms of structural (RDF, hydrogen bonding) and dynamical properties (diffusion coefficients). Our results indicate that MLIPs are a viable option for molecular dynamics simulations, providing greater speed than DFT methods with comparable accuracy.
References:
[1] J. Hafner, Journal of Computational Chemistry, 2008, 29 (13), 2044-2078.
[2] R. Jinnouchi, Physical Chemistry Chemical Physics, 2022, 24 (25), 15522-15531.
[3] L. Vilčiauskas, M. Tuckerman, G. Bester, S. J. Paddison, K.-D. Kreuer, Nature Chemistry, 2012, 4 (6), 461–466.
[4] A. Mikalčiūtė, L. Vilčiauskas, Physical Chemistry Chemical Physics, 2021, 23 (10), 6213-6224.