Machine learning analysis of alloying element effects on hydrogen storage properties of AB2 metal hydrides

Zirconium-titanium-based AB2 is a potential candidate for hydrogen storage alloys and NiMH battery electrodes. Machine learning (ML) has been used to discover and optimize the properties of energy-related materials, including hydrogen storage alloys. This study used ML approaches to analyze the AB2 metal hydrides dataset. The AB2 alloy is considered promising owing to its slightly high hydrogen density and commerciality. This study investigates the effect of the alloying elements on the hydrogen storage properties of the AB2 alloys, i.e., the heat of formation (ΔH), phase abundance, and hydrogen capacity. ML analysis was performed on the 314 pairs collected and data curated from the literature published during 1998–2019, comprising the chemical compositions of alloys and their hydrogen storage properties. The random forest model excellently predicts all hydrogen storage properties for the dataset. Ni provided the most contribution to the change in the enthalpy of the hydride formation but reduced the hydrogen content. Other elements, such as Cr, contribute strongly to the formation of the C14-type Laves phase. Mn significantly affects the hydrogen storage capacity. This study is expected to guide further experimental work to optimize the phase structure of AB2 and its hydrogen sorption properties.