A general-purpose machine learning Pt interatomic potential for an accurate description of bulk, surfaces and nanoparticles

TL;DR

This paper introduces a machine learning interatomic potential for platinum that accurately models bulk, surface, and nanoparticle properties, enabling efficient simulations with DFT-level accuracy.

Contribution

A novel Gaussian approximation machine learning potential for platinum trained on diverse DFT data, offering high transferability and accuracy across different platinum structures.

Findings

Excellent agreement with DFT for bulk and surface properties

Accurate modeling of nanoparticle stability and behavior

Enabled advanced simulations like phase diagrams and crystallization studies

Abstract

A Gaussian approximation machine learning interatomic potential for platinum is presented. It has been trained on DFT data computed for bulk, surfaces and nanostructured platinum, in particular nanoparticles. Across the range of tested properties, which include bulk elasticity, surface energetics and nanoparticle stability, this potential shows excellent transferability and agreement with DFT, providing state-of-the-art accuracy at low computational cost. We showcase the possibilities for modeling of Pt systems enabled by this potential with two examples: the pressure-temperature phase diagram of Pt calculated using nested sampling and a study of the spontaneous crystallization of a large Pt nanoparticle based on classical dynamics simulations over several nanoseconds.