Machine learning assisted design of BCC high entropy alloys for room temperature hydrogen storage

Abstract

Body-centered cubic (BCC) alloy systems can theoretically store double amounts of hydrogen compared with commercial metal hydrides at room temperature, and BCC high entropy alloys (HEAs) have shown the potential to reach this theoretic limit. However, the high thermodynamic stability of the dihydrides formed during hydrogen storage results in high operating temperatures. Here, by employing multi-objective Bayesian optimization-aided density functional theory calculations, we discovered 8 new HEA candidates for hydrogen storage, including the VNbCrMoMn HEA that can store 2.83 wt% hydrogen at room temperature and atmospheric pressure, vastly exceeding the hydrogen capacities of 1.38 wt% and 1.91 wt% for commercial LaNi5H6 and TiFeH2. Such a high performance of VNbCrMoMn is ascribed to the optimized hydrogen absorption thermodynamics, which is achieved under the guidance of interpretable machine learning which revealed that the thermodynamics of the first and second stages of absorption are largely determined by the bulk modulus and the number of states in the d-band, respectively.