Process parameters and magnetic properties (coercivity, remanence, squareness, maximum energy product) of data-driven fabrication of Nd-Fe-B anisotropic magnets by direct hot extrusion.

We implemented an active learning pipeline assisted by machine learning and Bayesian optimization (ALMLBO) for predicting magnetic properties (coercivity, remanence, squareness) of Nd-Fe-B anisotropic magnets fabricated by direct hot extrusion, using a fixed commercial Nd14Fe76Co3.4B6Ga0.6 (at%) powder (MQU-F™), from their process parameters, and propose optimal processing leading to high coercivity and remanence, simultaneously. ALMLBO allowed us to optimize the process to exhibit high coercivity, 𝞵0Hc ~ 1.7 T, and remanence, 𝞵0Br ~ 1.4 T, resulting in an excellent maximum energy product, (BH)max ~ 380 kJ/m3. Therefore, the following dataset contains the process parameters (hot-press temperature and load; hot extrusion temperature, ram speed, and load limit; the identification number of the dies used at the extrusion exit) as well as the magnetic properties (coercivity, remanence, squareness, maximum energy product) of Nd-Fe-B anisotropic magnets, fabricated by direct hot extrusion, obtained before and during 3 cycles of ALMLBO.