Fig. 7
From: Predicting phase behavior of grain boundaries with evolutionary search and machine learning
Clustering identifies GB phases observed at high temperature. Metastable grain boundary phases identified by the clustering analysis at 0 K become stable at high temperature. Evolutionary search a and clustering b, c for the Σ29(520) [001] predict three grain boundary phases: d Kites, e Split Kites, and f Distorted Kites. Kite phase is the ground state at 0 K. Split Kite phase is a high energy state at 0 K, but becomes a ground state with the lowest free energy at finite temperature, as demonstrated by a transformation in MD simulations at 900 K
