Alloy design and properties optimization of multi-component alloy based on solidification characteristics

The current design strategies focus on the solidification characteristics to discover new MACs with promising properties in an enormous search space composed of considerable composition combinations in a specific system. In the present study, the associated correlations between alloying elements and solidification interval characteristics are uncovered via machine learning (ML) in the multi-component systems. Given the Fe–Cr–Ni–Al system, testing accuracy value using FNN model in identifying the input and output features is above 89%, and the Cr and Al are the critical elements. With these understanding, FeCrNiAl_0.8 alloy was successfully designed and prepared, and the properties are tested to confirm the reliability of the alloy design strategies. The experimental investigations not only provide strong evidence for the predicted accuracy, but also find superior mechanical properties. The designed FeCrNiAl_0.8 alloy exhibits fracture strength and plastic strain of 2839 MPa and 41%, respectively, accompanied with remarkable work hardening capacity. Beyond that, the extraordinary specific strength make it competitive against virtually known high performance metals and polycrystalline superalloys even in high temperature applications. Our work provides new ideas to search for compositionally complex alloys and is also applicable to assist in developing advanced MACs for engineering applications.