Finding the low-energy structures of Si[001] symmetric tilted grain boundaries with a genetic algorithm

Abstract

We developed a global structure optimization method, genetic algorithm, for a fast and efficient prediction of grain-boundary structures. Using this method we predicted the most stable structures and a number of low-energy metastable structures for Si[001] symmetric tilted grain boundaries with various tilted angles. We show that most of the grain-boundary structures can be described by the structural unit model with the units being the dislocation cores and perfect-crystal fragments. The energies of the grain-boundary structures obtained from the genetic algorithm optimization are evaluated by tight-binding calculations using the environment-dependent Si tight-binding potential developed previously and found to be in very good agreement with the first-principles calculation results.