Gradient Boosted Machine Learning Model to Predict H₂, CH₄, and CO₂ Uptake in Metal-Organic Frameworks Using...

Area of Science:

• Materials Science
• Computational Chemistry
• Machine Learning

Background:

• Predictive screening of metal-organic frameworks (MOFs) for gas uptake has relied on simulated data, limiting real-world applicability.
• Existing models' accuracy is constrained by the performance of underlying simulation techniques.

Purpose of the Study:

• To develop a machine learning model for predicting gas uptake in MOFs using experimental data.
• To provide a reliable tool for estimating H₂, CH₄, and CO₂ uptake in MOFs across various conditions.

Main Methods:

• A Gradient Boosted Tree model was constructed using experimentally derived gas uptake data for MOFs.
• Descriptors were sourced from existing literature, eliminating the need for new computational modeling.
• The model's performance was validated through repeated runs, assessing accuracy and error.

Main Results:

• The developed model achieved an average R² of 0.86 and a mean absolute error (MAE) of ±2.88 wt % across multiple runs.
• The model accurately predicts gas uptake for H₂, CH₄, and CO₂ at varying temperatures and pressures.
• Utilizing experimental data enhances the predictive accuracy compared to simulation-based approaches.

Conclusions:

• This model offers a practical, one-stop solution for predicting MOF gas uptake based on real-world experimental observations.
• The findings facilitate more reliable material selection and design for gas storage applications.
• The focus on experimental data integration makes the model directly applicable by practitioners in the field.