About this Abstract |
| Meeting |
2027 TMS Annual Meeting & Exhibition
|
| Symposium
|
Computational Thermodynamics and Kinetics
|
| Presentation Title |
Accelerated Prediction of Solute Diffusion in Ni-based
substitutional solid solutions via First-Principles-Informed Machine
Learning |
| Author(s) |
Jinyoung Kwon, Najla Zulfikar, Shailendra Kumar Verma, Kyoungdoc Kim |
| On-Site Speaker (Planned) |
Jinyoung Kwon |
| Abstract Scope |
Abstract: Diffusion of alloying elements critically governs microstructural
evolution in
Ni-based superalloys, particularly γ′ coarsening and high-temperature
creep. Herein, we combine first-principles calculations with Monte Carlo
simulations and the five-/fourteen-frequency models of Lidiard and Bocquet
to
examine the effects of solute pairs, including Al, Ti, and Co, on
vacancy-mediated impurity diffusion in dilute fcc Ni. The calculated
diffusivity enhancement follows the order Ti-X > Al-X > Co-X, indicating
that
Ti-containing solute pairs most strongly promote impurity diffusion. This
behavior is mainly controlled by activation-energy variations associated
with
impurity atomic volume and strain, as well as by diffusion-prefactor
changes
reflecting solute-vacancy bond directionality. To clarify the role of
d-electron filling, we calculate the migration and formation energies of
Mo-containing solute pairs and train a crystal graph convolutional neural
network (CGCNN) for machine-learning-accelerated diffusivity prediction.
The
resulting dataset provides valuable kinetic parameters for modeling creep
and
γ′ coarsening in Ni-based superalloys. |
| Proceedings Inclusion? |
Planned: |