Machine Learning-Assisted Design of Metal Hydride Alloys for Hydrogen Applications using CALPHAD Predictive Modelling - HYPHAD

Project summary

The HYPHAD project aims to exploit the recently developed universal machine learning potential in conjunction to CALPHAD, experiments, and large-scale synthesis to overcome the complex multi-scale nature of metal hydrides and rapidly screen diverse potential metal hydride materials for hydrogen applications. HYPHAD will address the needs of the industry partner in this proposal, namely, (1) discovering new metal hydrides to at least partially replace the costly and rare Ti, Zr and V elements of commercially used TiMn2 and FeTi and (2) finding economic, optimal alloying compositions for TiMn2 and FeTi. The proposed machine learning potential-based muti-scale discovery strategy is novel and broadly impactful to the materials science community. In addition to storage, hydrides can also be used for the compression and purification of hydrogen. The HYPHAD will contribute to Sustainable Development Goal 7 and the priority area of "Sustainable advanced materials for energy."