Structure characterization and predictability by Voronoi analysis

The unique ability of tessellation methods to characterize atomic structures is highlighted, with the Voronoi tessellation, by virtue of its freedom from constraining assumptions, emphasized as pre-eminent among these methods. It is demonstrated how to extract information from an atomic structure by constructing its Voronoi tessellation and analysing the results. In so doing, insight is gained into the coordination patterns of the atoms and how their positions are influenced by neighbours. As an example of such a procedure, typical Voronoi volumes are calculated for 72 atom types found in a set of 249 binary intermetallic and ionic compounds. Whereas the various atom types have broadly characteristic volumes, these are also sensitive to the other atom types with which they are combined in a given compound, i.e. whether the latter behaves as a metal, semimetal, semiconductor or insulator. Further, smaller variations in volume occur for different atom types within these four categories. Thus, structural information obtained from the Voronoi tessellation may possibly be correlated directly with atomic properties.