High‐Throughput Data Generation and Transfer Learning Enabled Microstructure‐Property Integrated Design of Nickel‐Based Powder Metallurgy Superalloy

Nickel-based powder metallurgy (PM) superalloys are indispensable for aero-engine turbine disks, but their design is constrained by complex multicomponent interactions and the difficulty in predicting long-term service performance. Herein, we present a data-driven framework integrating high-throughput thermodynamic calculations, diffusion-multiple experiments, and transfer learning to predict long-term microstructural stability and mechanical properties. By calibrating computational models with sparse experimental data, our transfer learning approach enables accurate prediction of microstructural features, which further serve as inputs for mechanical property modeling. From a screening of 10⁵ compositions, we identify a promising low-density (8.33 g/cm³) alloy, designated USTB-PM750. This alloy exhibits a yield strength of 1138 MPa and a creep life of 141 h to 0.2% strain at 750°C under 480 MPa. Microstructural analysis reveals that its superior performance stems from a low stacking-fault energy, which promotes the formation of stacking faults and microtwins. These defects subsequently evolve into dense networks of Lomer-Cottrell locks, further reinforced by solute-segregation-induced local phase transformations. This approach significantly improves alloy design efficiency and offers a promising high-performance candidate material for turbine-disk applications in advanced aero-engines.