About this Abstract |
Meeting |
2024 TMS Annual Meeting & Exhibition
|
Symposium
|
Thermodynamics and Kinetics of Alloys II
|
Presentation Title |
Efficient Ab Initio Estimation of the High-temperature Liquidus Curve |
Author(s) |
Shibo Tan, Joshua Willwerth, Wenhao Sun |
On-Site Speaker (Planned) |
Wenhao Sun |
Abstract Scope |
Melting temperature is a fundamental consideration in materials synthesis and processing, as well as in operational stability. The liquidus curve is usually obtained from the Calculation of Phase Diagrams (CALPHAD) approach, but thermodynamic data only exists for a limited set of chemical systems. Although phase diagrams can now be predicted using first-principles DFT calculations, DFT is a T = 0K method and cannot efficiently predict melting temperatures. Here, we present a thermodynamic referencing scheme to combine DFT convex hulls with CALPHAD approaches, such that we can rapidly estimate the high-temperature liquidus curves of phase diagrams, at a computational cost low enough to integrate into high-throughput DFT databases. Using the ASM phase diagram database as a training set, we build a machine-learned algorithm for non-ideal liquid mixing free energies, which enables the ab initio prediction of liquidus curves in systems where experimental data currently does not exist. |
Proceedings Inclusion? |
Planned: |