A catalogue of metallic glass-forming alloy systems

Rational materials design out of vast compositional space is attractive yet challenging. The data-driven approach has shown promise in accelerating the development of advanced multicomponent alloys, such as metallic glasses. However, data-driven development of glass-forming alloys is limited by the sparse and biased datasets. In this study, we establish the high-throughput experimental database (HED), featuring an unprecedented quantity and diversity of experimental data. This database, encompassing 15,080 materials from 33 alloy systems synthesized and characterized under consistent conditions, provides a robust dataset for the training of machine learning model. The developed model is validated by both literature data and high-throughput experiments, and enables the creation of a catalogue of metallic glass forming alloy systems. The catalogue would serve as a practical reference for efficient design of glass-forming alloys systems.