Novel estimation method for anisotropic grain boundary properties based on Bayesian data assimilation and phase-field simulation

Utilizing the data assimilation and multi-phase-field grain growth model, this study proposes a novel framework of measuring anisotropic (nonuniform) grain boundary energy and mobility. The framework can evaluate a large number of boundary properties from typical observations of grain growth without requiring specifically designed experiments or calculations. In this method, by optimizing the multi-phase-field model parameters such that the simulation results are in good agreement with the observation data, the energies and mobilities of multiple individual boundaries are directly and simultaneously estimated. To validate the method, numerical tests on boundary property estimation were performed using synthetic microstructure dataset generated from grain growth simulations with a priori assumed property values. Systematic tests on simple tricrystal systems confirmed that the proposed method accurately estimates each boundary energy and mobility within an error of only several % of their assumed true values even for conditions with strong property anisotropy and grain rotation. Further numerical tests were conducted on a more general multi-grain system, showing that our method can be successfully applied to complicated polycrystalline grain growth. The obtained results demonstrate the potential of the proposed method in extracting a large dataset of grain boundary properties for arbitrary boundaries from actual grain growth observations.