Data‐driven “cross‐component” design and optimization of γ′‐strengthened Co‐based superalloys

The design of complex multicomponent superalloys has always been challenging due to the interaction of multiple elements and stringent requirements for various properties. Herein, an integrated approach to designing the high‐component (>7) γ′‐strengthened Co‐based superalloys with well‐balanced properties is developed by combining the diffusion multiples and machine‐learning models. A “cross‐component” prediction is achieved by the machine‐learning models, where two types of novenary superalloys are screened out for aeroengine and industrial gas turbine blades, respectively, based on the experimental database mainly consisting of six to seven elements. The method is verified to be effective or slightly more favorable than the Calculation of Phase Diagram (CALPHAD) in predicting the γ′ solvus temperature ( T γ′ ) and phase constituent of the high‐component alloys when reasonable data of low‐component alloys are just provided. Furthermore, the oxidation resistance and hardness of polycrystal superalloys as well as the compressive strength of single‐crystal superalloys are tested. Finally, some factors affecting the accuracy of “cross‐component” prediction are discussed. Expanding the compositional range and supplementing the critical interaction data of multiple elements in the database are beneficial for improving the accuracy of the “cross‐component” prediction.