Enhancing Stability, Magnetic Anisotropy, and Coercivity of $τ$-L$1_0$ MnAl: Machine Learning, $\textit{Ab Initio}$, and Micromagnetic Modeling

Churna Bhandari,Gavin N. Nope,T. Lograsso,Durga Paudyal

2024-07-31

Abstract:The binary manganese aluminium (MnAl) alloy with L$1_0$ crystal structure is a promising rare earth element-free permanent magnetic material because of its exceptional magnetic properties. However, experimentally synthesizing it in a stable bulk form is extremely challenging. Here, an alternative method of stabilizing the material is proposed and theoretically verified by partially substituting Mn and Al sites with Fe and Ni and identifying its enhanced phase stability, magnetic anisotropy, and coercivity from density functional theory (DFT), machine learning (ML) crystal graph convolution neural network (CGCNN), and micro-magnetic modeling. When considering a fixed ($50\%$)-Ni, the magnetic anisotropy increases with the increasing Fe content but decreases the formation energy. The calculated formation energies, elastic constants, and phonon frequencies demonstrate that all the binary and quaternary compositions are stable. Most importantly, the magnetic moment and magnetic anisotropy constants in $50\%$-Fe substituted composition (equiatomic phase) increase significantly compared to the MnAl. The predicted coercivity of the equiatomic phase is larger than the parent compound calculated by combining DFT computed parameters with micromagnetic simulations.

Materials Science