Benchmarking Machine Learning Algorithms for the Prediction of Gas Adsorption in Metal-Organic Framework Materials

Author

Juan G. Benitez, The University of Texas Rio Grande Valley

Date of Award

8-1-2024

Document Type

Dissertation

Degree Name

Doctor of Philosophy (PhD)

Department

Mathematics

First Advisor

Haoyuan Chen

Second Advisor

Zhijun Qiao

Third Advisor

Hansapani Rodrigo

Abstract

Metal-organic frameworks (MOFs) are nanoporous materials with a structure and chemistry that can be tailored with precision to support a variety of applications, such as gas storage, drug delivery, catalysis, and sensing. Central to the realization of these applications is the need to model adsorption in MOFs. However, if the goal is to screen a large number of the hundreds of thousands of MOFs discovered, conventional high-throughput molecular simulations are prohibitively demanding in terms of computational resources. Therefore, efficient approaches are required to efficiently identify suitable MOFs for specific applications. In our work, machine learning (ML) models to predict adsorption of nitrogen and carbon dioxide in MOFs are tested. The comparison of three main ML paradigms (linear regression, neural networks and tree-based algorithms) together with an appropriate feature engineering setup, shows that modeling can be performed efficiently and with notable accuracy. The features used for modeling possess physical interpretation and can be meaningfully measured against ML predictions using SHAP values as a measure of feature importance, providing insights that align with first-principles adsorption mechanisms.

Comments

Copyright 2024 Juan G Benitez. https://proquest.com/docview/3098420242

Recommended Citation

Benitez, Juan G., "Benchmarking Machine Learning Algorithms for the Prediction of Gas Adsorption in Metal-Organic Framework Materials" (2024). Theses and Dissertations. 1585.
https://scholarworks.utrgv.edu/etd/1585