Understanding Dynamics of Heterogeneous Ferroelectric Oxides at the Nanoscale using Graph Neural Networks on Reactive Force-Field Simulations

Ferroelectrics are a technologically important class of materials for next generation microelectronics platforms, such as ferroelectric memristors, which show spontaneous long-range ordering of electric polarization. Recent advances have even shown formation of novel nanoscale chiral polar structures, such as polar skyrmions in heterogeneous oxides, that open new possibilities for storing and manipulating information at the nanoscale. While formation of long-range polar order as well as polar skyrmions is understood to be a delicate balance between coupling between microscopic degrees of freedom, there is a critical need to understand dynamics of polarization switching in such heterogeneous materials under high fields. In this talk, we will present insights into the dynamics of polarization switching in defective BaTiO₃, obtained from combining large-scale atomistic reactive simulations with dynamical graph neural network approaches. Specifically, we will focus on how polar-structure and dynamics changes around point-defects, and how this interaction influences domain-wall dynamics.