Accurate Surface and Finite Temperature Bulk Properties of Lithium Metal at Large Scales using Machine Learning Interaction Potentials

• DOI: 10.48550/arxiv.2305.06925
• Publication Date: 2023-01-01
• Authors: Mgcini Keith Phuthi; Archie Mingze Yao; Simon Batzner; Albert Musaelian; Boris Kozinsky; Ekin D. Cubuk; Venkatasubramania...
• Keywords: Chemical Physics (physics.chem-ph); FOS: Computer and information sciences; Condensed Matter - Materials Science; Computer Science - Machine Learning; Physics - Chemical Physics; Materials Science (cond-mat.mtrl-sci); FOS: Physical sciences; Computational Physics (physics.comp-ph); 7. Clean energy; Physics - Computational Physics; Machine Learning (cs.LG)

The properties of lithium metal are key parameters in the design ion and batteries. They difficult to probe experimentally due high reactivity low melting point as well microscopic scales at which exists batteries where it is found have enhanced strength, with implications for dendrite suppression strategies. Computationally, there a lack empirical potentials that consistently quantitatively accurate across all ab-initio calculations too costly. In this work, we train Machine Learning Interaction Potentials (MLIPs) on Density Functional Theory (DFT) data state-of-the-art accuracy reproducing experimental results wide range simulations large length time scales. We accurately predict thermodynamic properties, phonon spectra, temperature dependence elastic constants various surface inaccessible using DFT. establish Bell-Evans-Polanyi relation correlating self-adsorption energy minimum diffusion barrier Miller index facets.