Abstract: Interatomic energies (IAEs) are important intrinsic quantities for understanding, designing and regulating materials’ properties, typically calculated using density functional theory (DFT). However, exhaustive DFT calculations of IAEs between an object element (E_obj) and all property-toward potential alloying elements in the periodic table are time-consuming and expensive. Therefore, this work proposes an IAEs-based digital modelling strategy combining DFT calculations and active machine learning, which enables rapid selection of alloying elements for target properties of interest. The strategy was applied to a cast C70600 alloy used for marine engineering equipment manufacturing to screen its property-toward alloying elements (E) throughout the periodic table. With a small number of DFT calculations, three high-accuracy IAE models involving binding energy between Ni-E atoms, NaCl adsorption energy of E-dissolved Cu₂O, and formation energy of E_xO_y were established. Based on these models, the low-cost and environment-friendly elements Si, Al, and Sn were efficiently identified to be capable of simultaneously enhancing mechanical strength, corrosion resistance, and high-temperature oxidation resistance. Experimental validations exhibited that adding 3.4 at.% Si, Al, or Sn individually to C70600 cast alloy led to 114 %-276 % increases in yield strength, 6 %-25 % reductions in corrosion rate after 7 days in 3.5 % NaCl solution, and 10 %-16 % decreases in oxidation weight gain after 10 h at 800 °C in air. The proposed strategy could reduce the DFT calculations for the 118 periodic table elements by >85 %, and offers a new route for developing alloys with multiple target properties.

Key words: Interatomic energy, DFT calculations, Active machine learning, Cu-Ni alloys, Property-toward element selection