Development of\na Machine Learning Potential to Study\nthe Structure and Thermodynamics of Nickel Nanoclusters

Machine learning (ML) potentials such as the Gaussian\napproximation\npotential (GAP) have demonstrated impressive capabilities in mapping\nstructure to properties across diverse systems. Here, we introduce\na GAP model for low-dimensional Ni nanoclusters and demonstrate its\nflexibility and effectiveness in capturing the energetics, structural\ndiversity, and thermodynamic properties of Ni nanoclusters across\na broad size range. Through a systematic approach encompassing model\ndevelopment, validation, and application, we evaluate the model’s\nefficacy in representing energetics and configurational features in\nlow-dimensional regimes while also examining its extrapolative nature\nto vastly different spatiotemporal regimes. Our analysis and discussion\nshed light on the data quality required to effectively train such\nmodels. Trajectories from large-scale MD simulations using the GAP\nmodel analyzed with data-driven models like graph neural networks\nreveal intriguing insights into the size-dependent phase behavior\nand thermomechanical stability characteristics of porous Ni nanoparticles.\nOverall, our work underscores the potential of ML models, which coupled\nwith data-driven approaches serve as versatile tools for studying\nlow-dimensional systems and complex material dynamics.