Study of the ion mobility in defect-laden ZrO 2 under an electric field using neural network with predictions for Born effective charges

Unusual mass transport behavior in tetragonal ZrO 2 ceramics has attracted attention under flash events induced by strong electric fields. However, this observation cannot be attributed solely to Joule heating, suggesting the importance of understanding the ion behaviors associated with defective states under a strong electric field. Previous studies have studied the impact of an external electric field but were typically limited to fixed formal charges for the ions. In this work, to incorporate the response of ions to an electric field, we calculate Born effective charges, and use them in addition to the energy and forces to train neural network potentials. Our molecular dynamics simulations using trained models show that under an applied electric field, the diffusivity of oxygen ions is enhanced in defect-laden ZrO 2 with a preexisting oxygen vacancy, which could be associated with the observed unusual mass transport behavior. This is a milestone towards the accurate description of defect-laden materials under an applied electric field.