An accurate and transferable machine learning interatomic potential for nickel

This paper presents a magnetism-hidden machine learning Deep Potential (DP) model for nickel, trained using spin-polarized density functional theory (DFT) calculations. The model accurately predicts a wide range of properties for both FCC and HCP phases, including lattice parameters, elastic constants, phonon spectra, and various defect properties. The model's accuracy and transferability are demonstrated through investigations of the FCC-HCP allotropic phase transition under uniaxial tensile loading, revealing a high critical strain and an atypical orientation relationship.

Xiaoguo Gong, Zhuoyuan Li, A. S. L. Subrahmanyam Pattamatta, Tongqi Wen, David J. Srolovitz