Deep learning for CALPHAD modeling: Universal parameter learning solely based on chemical formula

2023/07/10 by Hong, Qi-Jun
#FOS: Physical sciences #Materials Science (cond-mat.mtrl-sci)

paper · doi:10.48550/arxiv.2307.04283

Abstract

Empowering the creation of thermodynamic and property databases, the CALPHAD (CALculation of PHAse Diagrams) methodology plays a vital role in enhancing materials and manufacturing process design. In this study, we propose a deep learning approach to train parameters in CALPHAD models solely based on chemical formula. We demonstrate its application through an example of calculating the mixing parameter of liquids. This work showcases the integration of CALPHAD and deep learning, highlighting its potential for achieving automated comprehensive CALPHAD modeling.

Related

• Construction and Tuning of CALPHAD Models Using Machine-Learned Interatomic Potentials and Experimental Data: A Case Study of the Pt-W System
• Amending CALPHAD databases using a neural network for predicting mixing enthalpy of liquids
• Generalizable Prediction Model of Molten Salt Mixture Density with Chemistry-Informed Transfer Learning
• Differentiable thermodynamic modeling
• Physics-informed machine learning for predicting temperature-dependent chemical properties