Highly effective design of high GFA alloys with different metal-based and various components by machine learning

The glass-forming ability (GFA) is of great significance for the development of novel functional metal-based metallic glasses. In this study, seven popular machine learning (ML) algorithms were employed to design novel M-based (M = Fe, Co, Ni, Ti, Zr, and rare earth metal (RE)) and X-component ( X = 2, 3, 4, 5,6, and >6) alloys with excellent GFA. A GFA containing 6957 data points with structural analysis was established. Feature engineering was used to analyze the importance and correlation of features. ML algorithms were utilized for GFA prediction, revealing that Xtreme Gradient Boosting Trees exhibited the strongest predictive capability, achieving a high accuracy of 94.0%, a true positive rate of 97.6%, and a root mean squared error of 0.3705 across the entire dataset. Subsequently, the GFA of ternary to hexahydroxy alloys based on Fe, Co, Ni, Zr, Ti, and Y was predicted using all possible compositions generated through Python. Finally, a series of alloys with good GFA was successfully designed and prepared. The present work suggests that the proposed ML method can be utilized to design novel multiple-M-based amorphous alloys with high GFA.