Mapping the 'materials gene' for binary intermetallic compounds—a visualization schema for crystallographic databases

This paper provides a new approach for mapping the relative stability of intermetallic compounds. We quantitatively assess the collective role of numerous chemical and bonding parameters that govern the stability of these compounds (which we call 'genes') using the principles of information entropy. It is shown that one can establish a quantitative scaling parameter, in terms of Shannon entropy, that permits one to map the relative contributions of these parameters on to a single map. This new 'structure map' provides a means of exploring a multivariate array of attributes associated with structural stability and of discerning the efficacy of classical classification mappings for crystal chemistry. We used a binary AB2 intermetallics database as a platform for developing a classification scheme of phase stability based on the concept of Shannon information entropy. We have integrated a metric of information entropy into a recursive partitioning classifier for projecting high-dimensional data manifolds on to a low-dimensional structure map, hence providing a new visualization scheme of complex and high-dimensional crystallographic data sets.