Active machine learning guides discovery of giant magneto-caloric high-entropy MnFeCoVPSiB-based materials with ultra-low hysteresis

In recent years, computational-driven discovery of novel materials with targeted functionalities is a highly active research area. In this project, we have applied active-learning methods, coupled with closed-loop experimentation, to guide the discovery of high-entropy magnetocaloric materials (HE-MCMs). For instance, it is predicted that (Mn,Fe,(Co/V))2(P,Si,B)- based materials show negligible hysteresis, which will be beneficial for the improvement of energy efficiency for real prototypes. The above findings can accelerate the development and further unlock the perspective for GMCE applications, e.g., in solid-state magnetic refrigeration, magnetic heat pumps and thermomagnetic generators to convert waste heat into electricity. However, to further understand the physical mechanism of the predicted interesting compounds, there are several open questions that still need to be resolved like the changes in magnetic moments and coordination environment corresponding to different external dopants. Combined with other advanced structural characterizations, the complementary X-ray magnetic circular dichroism (XMCD) utilizing synchrotron radiation is expected to probe the intrinsic magnetic properties of the selected constituent atoms (Mn/Fe/Co/V). Systematic experiments will provide a detailed insight in the nature of the magneto-elastic transition in these high-entropy (Mn,Fe,(Co/V))2(P,Si,B) compounds guided by the discovery of active machine learning, which will be essential to further boost the discovery of more interesting HE-MCMs.