Machine Learning Vacancy Formation Energy in Nickel-Based Superalloys

Conference · Tue Mar 17 00:00:00 EDT 2026

DOI:https://doi.org/10.2172/3023552· OSTI ID:3023552

Sundar, Aditya ^[1]; San, Saro ^[1]; Gao, Michael ^[2]

---

1. NETL Site Support Contractor, National Energy Technology Laboratory
2. NETL

Thermal vacancies play a critical role in high-temperature Ni-based superalloys and influence various properties such as creep resistance, oxidation, etc. This study systematically investigates the impact of commonly used transition metals (Cr, Co, Fe), refractory metals (Nb, Ta, Mo, W) and other elements (Al, Cu, Ti, Mn) on the thermodynamic stability of 36 binary, 20 ternary, 11 quaternary, 9 quinary, and 3 senary FCC Ni-based alloys covering various elemental combinations. Density functional theory-based studies on Ni-X binary alloys show that higher concentrations of Cr, Nb, Ta, Al, and Ti introduce significant lattice distortions and broaden the distribution of vacancy formation energies (standard deviation up to 0.15 eV). These elements partially donate electrons, reducing their self-consistent chemical potentials relative to single-element reference values and lowering vacancy formation energies, while Co, Fe, Mo, and W show lower charge localization. These trends extend from 3-6 element alloys, where Cr, Nb, and Ta-rich compositions have low-energy states (~0.5 eV) that increase vacancy concentrations. Finally, graph neural network models are developed to screen over 5000 virtual alloys. Eleven leading compositions are identified with mean vacancy formation energy higher than 1.75 eV and vacancy concentration ~2 orders of magnitude lower than pure Ni at 1000 K. These results provide valuable guidelines to achieve controlled defect engineering in structural alloys.

• Research Organization: National Energy Technology Laboratory (NETL), Pittsburgh, PA, Morgantown, WV, and Albany, OR (United States)
• Sponsoring Organization: USDOE Office of Fossil Energy and Carbon Management (FECM); USDOE Office of Fossil Energy and Carbon Management (FECM), Office of Carbon Management (FE-20)
• OSTI ID: 3023552
• Resource Type: Conference presentation
• Conference Information: Conference Name: TMS 2026 Annual Meeting & Exhibition Location: San Diego, CA, United States Start Date: 3/15/2026 12:00:00 AM End Date: 3/19/2026 12:00:00 AM
• Country of Publication: United States
• Language: English

Related Subjects

DFT
machine learning
Ni superalloys
Vacancies