One of the most well-studied families of complex hydrides, the borohydrides of alkaline earth metals, displays an impressive diversity of structures, some of which are extremely complex[1]. Given the significance of these and other complex hydrides (e.g., alanates, silanes, mixed anion hydrides) for hydrogen storage and other applications, a computational tool for designing such materials in silico is highly desirable. Existing crystal structure prediction (CSP) methods struggle to accomplish this, as the complexity increases exponentially with the number of degrees of freedom. For example, Be(BH)2 contains over 200 atoms in the unit cell.

Computational discovery of novel materials for hydrogen storage

In this research, we address this problem by advancing the USPEX CSP algorithm [2] through the introduction of complex ions as building blocks. Specifically, we use well-known complex anions (BH4, NH4, SiH4) and metal cations as elementary structural units.

Practically, this integrated strategy has predicted a number of complex hydrides as well as a novel two-dimensional polymorph in calcium borohydride. The layered morphology of this polymorph opens new possibilities for applications, particularly in hydrogen storage and catalysis. These findings underscore the potential of our approach, demonstrating its impact on the design of new materials.