UCLA

Grain growth plays a pivotal role in determining the macroscopic properties of several polycrystalline materials. It is predominantly controlled by grain boundary and triple junction mobility, and several atomistic and mesoscale models have been developed to study this phenomenon. In particular, multi-order parameter phase-field and vertex dynamics models have been extensively used to understand the grain evolution dynamics and the effect of triple junction drag on the growth kinetics. In the current work, we present a novel sequentially-coupled phase-field and vertex dynamics model for both isotropic and anisotropic grain growth simulations. The proposed approach, which uses a backpropagation neural network, image processing and mathematical techniques for accurate grain boundary curvature detection, takes the advantages of both the models and provides an efficient way to switch from phase-field to vertex dynamics model depending on the growth controlling mechanism. Our results suggest that phase-field generated microstructures can replace Voronoi tessellation as the input to vertex dynamics, with a reduction in computational expenses and the ability to simulate more realistic and complex microstructures.