Multiscale Modeling of Zr Corrosion: Predicting Microstructures of Oxide Films

Abstract

A first-principles-informed multiscale modeling approach is reported for oxide formation during corrosion of zirconium. The present simulations reproduce the columnar corrosion film microstructure and grain size distribution seen in micrographs of oxidized Zr surfaces. This structural feature emerges from a hexagonal ZrO layer adjacent to the $α$-Zr/ZrO$_{2-x}$ interface, which is dynamically stabilized by continuous heterogeneous grain nucleation and grain boundary mass diffusion. This ZrO layer is identified as a significant oxygen diffusion barrier and a potential contributor to sub-parabolic pretransition corrosion kinetics. The simulations also reproduce the undulatory nature of the metal/oxide interface, leading to regions with tensile stress components capable of nucleating cracks, despite the overall state of compressive hydrostatic stress within the oxide film. First principles density functional theory (DFT) computations of fundamental ZrO$_x$ properties were combined with molecular dynamics simulations using a machine-learned potential based on a DFT training set to provide extensive input to an advanced phase field model for oxide formation and its microstructure, as implemented in the new MedeA PhaseField module. The phase field simulations reported herein incorporated computed stoichiometry-dependent thermochemical data, anisotropic elastic properties, interface energies, bulk and grain boundary mass transport processes, multiphase heterogeneous grain nucleation, and strain energy-dependent phase stabilities.