Modeling Local Structure and Configurational Free Energy of Multi-Component Alloys Below the High-Temperature Limit

We present a method for modeling alloys as a statistical ensemble of randomly sampled ordered configurations on a given lattice. Accurately modeling the structure of alloys can require using large supercells to accommodate atomic disorder. However, with increasing supercell size, the set of possible configurations quickly becomes too large to exhaustively simulate; this is especially true for high-entropy compositions. In this work, we show that randomly sampling from the set of all possible configurations allows for the efficient modeling of alloy systems. This is done while still capturing the structural and energetic complexities found in exhaustive sampling. Furthermore, the statistical treatment naturally produces configurational free energy and local atomic structure as a function of the synthesis temperature. The high-temperature limit, implying perfect configurational disorder, is not assumed; neither is the final structure of the alloy. Thus, our method is particularly useful for studying the relationships between the synthesis temperature, short-range order, and local structure. The high-entropy pseudohexnary system, PbTe-PbSe-PbS-GeTe-GeSe-GeS, will be studied as a case example. Trends in both the miscibility and structural deviation from the rocksalt structure-type will be discussed, as well as their agreement with experiment.