Phase Prediction via Crystal Structure Similarity\nin the Periodic Number Representation

The\nperiodic number (PN) representation of the chemical systems,\nintroduced by Dmitri Mendeleev, uncovers the fundamental principle\nof chemical similarity in a straightforward way. In this framework,\nthe rows correspond to the principal quantum numbers of the elements’\nelectronic configurations when considered isolated atoms. This systematic\narrangement allows for a deeper understanding of the relationships\nand patterns among the elements. In this study, we propose a novel\nstrategy for structure type (prototype) prediction by utilizing the\nPN concept to identify possible modifications and phase stability\nof unexplored chemical systems. Our PN-based crystal structure prediction\n(PNcsp) program, which evaluates similarity through PN neighboring\nin the phase map, provides the most probable prototypes for unknown\nand unreported modifications of given phases in binary and higher\norder chemical systems. We applied PNcsp to 59 distinct chemical systems\nwhose equimolar phases are indicated in the respective phase diagrams\nbut lack accurate experimental structure determination. Our methodology\nidentified 93 prototypes for these 59 equiatomic phases, of which\n47 exhibit mechanical and dynamic stability. Notably, this approach\ndiscovered 19 entirely novel, fully stable polymorphic phases, thereby\nexpanding the known landscape of potential materials. Furthermore,\nwe demonstrated that this method is also effective for nonequimolar\nand higher order systems.