Machine-learning-guided discovery of the gigantic magnetocaloric effect in HoB2 near the hydrogen liquefaction temperature

Magnetic refrigeration exploits the magnetocaloric effect, which is the entropy change upon the application and removal of magnetic fields in materials, providing an alternate path for refrigeration other than conventional gas cycles. While intensive research has uncovered a vast number of magnetic materials that exhibit a large magnetocaloric effect, these properties remain unknown for a substantial number of compounds. To explore new functional materials in this unknown space, machine learning is used as a guide for selecting materials that could exhibit a large magnetocaloric effect. By this approach, HoB2 is singled out and synthesized, and its magnetocaloric properties are evaluated, leading to the experimental discovery of a gigantic magnetic entropy change of 40.1 J kg−1 K−1 (0.35 J cm−3 K−1) for a field change of 5 T in the vicinity of a ferromagnetic second-order phase transition with a Curie temperature of 15 K. This is the highest value reported so far, to the best of our knowledge, near the hydrogen liquefaction temperature; thus, HoB2 is a highly suitable material for hydrogen liquefaction and low-temperature magnetic cooling applications. A material for magnetically cooling hydrogen to its liquid form has been identified by a data-driven approach. Some materials get colder when they are exposed to an alternating magnetic field. This so-called magnetocaloric effect enables refrigeration to within one thousandth of a degree of absolute zero. Trial and error have uncovered many magnetocaloric materials, but Pedro Baptista de Castro, from the National Institute for Materials Science in Tsukuba, Japan, and co-workers have instead approached material discovery in a more systematic way using machine learning. They trained their algorithm to screen prospective compounds using data from the scientific literature. In this way they identified, and then experimentally confirmed, that holmium boride, HoB2, has a giant magnetocaloric effect at temperatures around 15 Kelvin (–258 °C), near the liquefaction point of hydrogen. Magnetic refrigeration, which is based on the magnetocaloric effect (MCE), is an emerging pathway for environment-friendly refrigeration. In this work, we performed a machine learning based approach to discover experimentally that HoB2 exhibits |ΔSM| = 40.1 J/kg K (0.35 J/cm−3 K) for μ0ΔΗ = 5 Τ at second order transition of TC ~ 15 K, having the largest |ΔSM| around this temperature region. Thus, HoB2 is a highly suitable material for hydrogen liquefaction and low-temperature magnetic cooling applications. Our study also sheds light on the machine learning approach as an effective method for searching functional materials characterized by complex physical properties.