About this Abstract |
Meeting |
2023 TMS Annual Meeting & Exhibition
|
Symposium
|
Materials Genome, CALPHAD, and a Career over the Span of 20, 50, and 60 Years: An FMD/SMD Symposium in Honor of Zi-Kui Liu
|
Presentation Title |
Melting Temperature Prediction via Integrated First Principles and Deep Learning |
Author(s) |
Qijun Hong |
On-Site Speaker (Planned) |
Qijun Hong |
Abstract Scope |
We build a rapid and accurate tool for melting temperature prediction, by integrating methods we have developed based on density functional theory and deep learning. DFT generates highly accurate data points and iteratively improves the deep learning model, while the latter provides speed and efficiency. On the DFT side, we have built an accurate and cost-effective method for first principles melting temperature calculation. The method is implemented in the SLUSCHI package, an automated tool for DFT melting point calculation. On the deep learning side, we have built a melting temperature database that contains both experiment and computation. Based on the data, we then built a graph neural networks model that rapidly predicts melting temperature. We present examples of applications, such as the design and discovery of high-melting-point materials, and the evolution of mineral's melting temperature. |
Proceedings Inclusion? |
Planned: |
Keywords |
Computational Materials Science & Engineering, Machine Learning, Phase Transformations |