EfficientAb initioModeling of Random Multicomponent Alloys

We present in this Letter a novel small set of ordered structures (SSOS) method that allows extremely efficient ab initio modeling of random multicomponent alloys. Using inverse II-III spinel oxides and equiatomic quinary bcc (so-called high entropy) alloys as examples, we demonstrate that a SSOS can achieve the same accuracy as a large supercell or a well-converged cluster expansion, but with significantly reduced computational cost. In particular, because of this efficiency, a large number of quinary alloy compositions can be quickly screened, leading to the identification of several new possible high-entropy alloy chemistries. The SSOS method developed here can be broadly useful for the rapid computational design of multicomponent materials, especially those with a large number of alloying elements, a challenging problem for other approaches.