Files and links (1)
PDFOpen Access (License Unspecified), Open Access
Thesis
Predicting Gibbs Free Energy Functions of Inorganic Compounds Using Machine Learning
Master of Science (M.S.), Drexel University
Jun 2023
DOI:
https://doi.org/10.17918/00001776
Abstract
The goal of this thesis is to develop a machine learning model that can accurately predict the coefficients of the Gibbs free energy function for stoichiometric inorganic compounds. The model relies on a set of physical descriptors readily available from an established materials database of 0K density function theory (DFT) calculations (i.e., the Materials Project database). Knowing the Gibbs free energy of the phases in a system as a function of temperature, pressure, and chemical compositions provides the basis to predict the thermodynamic stability of the phases and the associated phase equilibrium and phase diagrams. However, obtaining Gibbs free energy data for many solid stoichiometric compound phases across a wide temperature range is challenging due to the high experimental or computational demands. In this thesis, we develop a machine learning model to predict the Gibbs free energy functions for 155 different solid compounds phases. These compounds belong to various categories, including intermetallic compounds, binary oxides, and ternary oxides, and the model covers the temperature range from 300K up to the decomposition temperature of each compound. The development process employs the random forest algorithm, combined with a hyper-parameter optimization method using random search and three-fold cross-validation. To train and evaluate the model, we utilize data from the SSUB6 database, which provides Gibbs free energy functions derived from high-fidelity experimental thermochemical data. 26 physical descriptors are extracted from the Materials Project database to be considered as features in the machine learning model. This thesis also provides a framework for future work on expanding the coverage to the high-temperature stable phases and other classes of compounds (e.g., carbides, nitrides, etc.), making the model more robust.
Metrics
1031 File views/ downloads
56 Record Views
Details
- Title
- Predicting Gibbs Free Energy Functions of Inorganic Compounds Using Machine Learning
- Creators
- Tran Quang Thai
- Contributors
- Yong-Jie Hu (Advisor)Steven J. May (Advisor)
- Awarding Institution
- Drexel University
- Degree Awarded
- Master of Science (M.S.)
- Publisher
- Drexel University; Philadelphia, Pennsylvania
- Number of pages
- 65 pages
- Resource Type
- Thesis
- Language
- English
- Academic Unit
- Materials (Science and) Engineering (Metallurgical Engineering) [Historical]; College of Engineering (1970-2026); Drexel University
- Other Identifier
- 991021212414004721