(230w) A Graph-Based Method for Structure Classification in Amorphous Materials

Identifying structures of amorphous systems has received a great deal of interest because of its theoretic importance for understanding the nature of glass as well as its potential application in developing glassy materials, including polymers and metallic glasses. In this work, we propose a graph theory-based tool to identify structural motifs in glassy system and also compute the structural entropy. In short, we group together local structures by comparing the associated Voronoi diagrams using a graph theoretic technique -- graph isomorphism. In comparison with other cluster identification approaches that rely on predefined structures, our method is completely automated. We apply the method to dense colloidal suspensions via computer simulation and identify a plenty of distinct clusters, among which icosahedron-like structures are the most common. In contrast to crystals that dominated by fcc order, less than 10 % atoms of the system belong to the center of icosahedral clusters, as a result of geometric frustration. These common clusters grow in general as glass transition is approached through densification, with icosahedral order showing the most significant increase. Finally, we obtain the structural entropy for hard disks and hard spheres by counting the number of distinct graphs of subregions. Our results are consistent with the configurational entropy via thermodynamic integration.