High-throughput compatible approach for entropy estimation in magnetocaloric materials: FeRh as a test case

Abstract Aiming to predict new materials for magnetic refrigeration from high-throughput calculations asks for an accurate, transferable, and resource-wise balanced approach. Here, we analyze the influence of various approximations on the calculation of key properties of magnetocaloric materials, while revisiting the well-known FeRh system for benchmarking our approach. We focus on the entropy change and its contributions from the electronic, lattice, and magnetic degrees of freedom. All approximations considered are based on first-principles methods and have been tested, and compared for FeRh. In particular, we find that in this context, the Debye approximation for the lattice entropy fails, due to the presence of soft phonon modes in the AFM phase. This approximation is frequently used in the literature as a simple alternative to full phonon calculations, but since soft modes are likely to occur also among promising magnetocaloric materials where structural transformations are common. Therefore, the use of the Debye approximation should be discarded for these systems treatment. This leaves the calculations of the lattice contribution the most demanding task from the computational point of view, while the remaining contributions can be approximated using more efficient approaches. The entropy change Δ S shows a peak around 370 K, for which the total entropy change is given by 24.8 J K − 1 k g − 1 ( Δ S e l e = 7.38, Δ S l a t = 7.05, Δ S m a g = 10.36 J K − 1 k g − 1 ) in good agreement with previous theoretical and experimental findings.