Coarsening kinetics of topologically highly correlated grain boundary networks

Abstract

We apply phase-field simulations in two dimensions to study the thermal coarsening of grain boundary (GB) networks with high fractions of twin and twin-variant boundaries, which for example are seen in grain-boundary-engineered FCC materials. Two types of grain boundary networks with similar starting special boundary fractions but different topological features were considered as initial conditions for the grain growth simulations. A lattice Monte Carlo method creates polycrystalline microstructures (Reed and Kumar (RK)), which exhibit hierarchical organization of random and special coincidence site lattice boundaries. The other type of microstructures (randomly distributed (RD)) contains random distributions of special boundaries subject only to crystallographic constraints. Under the assumption that random boundaries have larger energy and much higher mobility than special boundaries, simulations show that increasing the initial special boundary fraction in both microstructures slows down grain growth. However, the two starting microstructures exhibit very different behavior in the evolution of GB character and triple junction (TJ) distributions. The RD networks coarsened more slowly than the RK networks with comparable initial fractions of special boundaries. The observed trend in the evolution of the RK microstructures is explained by an extended von Neumann-Mullins analysis. This study demonstrates that the special boundary fraction is not a sufficient indicator of the coarsening behavior of twinned GB networks; the network topology must also be considered to correctly predict the grain growth kinetics.