Thermodynamic Modeling of Pure Elements from 0 K with Uncertainty Quantification using PyCalphad and ESPEI

TL;DR

This work implements thermodynamic models of pure elements into open-source tools PyCalphad and ESPEI, enabling systematic comparison, parameter evaluation, and uncertainty quantification down to 0 K for CALPHAD modeling.

Contribution

It introduces a systematic implementation and evaluation of pure element models in open-source software, facilitating high throughput CALPHAD modeling with uncertainty analysis.

Findings

Successfully modeled 41 pure elements with various thermodynamic models.

Enabled quantitative comparison and evaluation of models using open-source tools.

Demonstrated the use of MCMC in ESPEI for uncertainty quantification.

Abstract

Thermodynamic modeling of pure elements is the foundation of the CALPHAD modeling of engineering materials. Recently, multiple physics-based models have been proposed to describe Gibbs energy of pure elements down to 0 K, extending from 298.15 K in the current CALPHAD modeling. To enable their systematic and quantitative comparison and adoption, those thermodynamic models of pure elements are implemented into the open-source software packages PyCalphad and ESPEI in the present work for evaluation of model parameters and model fitness. PyCalphad and ESPEI are suitable tools for implementation of these models for high throughput CALPHAD modeling of multicomponent materials. Particularly, Markov Chain Monte Carlo used in ESPEI allows for uncertainty quantification of model parameters and model predictions. Through the remodeling of 41 pure elements, the present work demonstrates the quantitative comparison of modeling of pure elements with different models and enables the efficient development of multicomponent systems with continuously improved CALPHAD description of pure elements.