An exciting development over the past few decades has been the use of high-throughput computational screening as a means of identifying promising candidate materials for a variety of structural or functional properties. Experimentally, it is often found that the highest-performing materials contain substantial atomic site disorder. These are frequently overlooked in high-throughput computational searches, however, due to difficulties in dealing with materials that do not possess simple, well-defined crystallographic unit cells. Here we demonstrate that the screening of magnetocaloric materials with the help of the density-functional-theory-based magnetic deformation proxy can be extended to systems with atomic site disorder. This is accomplished by thermodynamic averaging of the magnetic deformation for ordered supercells across a solid solution. We show that the highly nonmonotonic magnetocaloric properties of the disordered solid solutions $\mathrm{Mn}({\mathrm{Co}}_{1-x}{\mathrm{Fe}}_x)\text{Ge}$ and $\left({\mathrm{Mn}}_{1-x}{\mathrm{Ni}}_x\right)\text{CoGe}$ are successfully captured using this method.

• Received 27 November 2019
• Accepted 22 January 2020

DOI:https://doi.org/10.1103/PhysRevMaterials.4.024402