Mining for elastic constants of intermetallics from the charge density landscape

There is a significant challenge in designing new materials for targeted properties based on their electronic structure. While in principle this goal can be met using knowledge of the electron charge density, the relationships between the density and properties are largely unknown. To help overcome this problem we develop a quantitative structure-property relationship (QSPR) between the charge density and the elastic constants for B2 intermetallics. Using a combination of informatics techniques for screening all the potentially relevant charge density descriptors, we find that C₁₁ and C₄₄ are determined solely from the magnitude of the charge density at its critical points, while C₁₂ is determined by the shape of the charge density at its critical points. From this reduced charge density selection space, we develop models for predicting the elastic constants of an expanded number of intermetallic systems, which we then use to predict the mechanical stability of new systems. Having reduced the descriptors necessary for modeling elastic constants, statistical learning approaches may then be used to predict the reduced knowledge-based required as a function of the constituent characteristics.