Atomistic simulations of Phase Transitions and Stabilization Mechanisms in Ferroelectric Hafnium Dioxide

Hafnium dioxide (HfO₂) exists in several polymorphs, including the ferroelectric phase of Pca2₁ symmetry, which has gained much attention due to its potential applications in electronics and memory devices, hence requiring further research for understanding of mechanisms for stabilization. This study focuses on developing a machine-learning interatomic potential (MLIP) based on density functional theory (DFT) data to study phase transitions and mechanical properties of HfO₂ under varying temperature and pressure conditions. The MLIP is validated by its good agreement with DFT predictions for lattice parameters, equations of state, elastic constants, and bulk and shear moduli across multiple structures and pressures. The MLIP is then employed to investigate phase transitions of ferroelectric HfO₂ (Pca2₁) under two ensembles. Our calculations reveal that under the isotropic NPT ensemble, HfO₂ transitions from the orthorhombic ferroelectric Pca2₁ phase to a mixture of tetragonal and orthorhombic Pbcn phases at elevated temperatures, while the anisotropic NPT ensemble drives a transition to the Pbcn phase, showing the potential influence of directional stress. We further report the development of a ML model to identify local symmetries to study structural phase transition mechanisms.