Related Experiment Video
Updated: Aug 6, 2026
09:46
Adsorption Device Based on a Langatate Crystal Microbalance for High Temperature High Pressure Gas Adsorption in Zeolite H-ZSM-5
Published on: August 25, 2016
Interpretable Machine Learning Framework for Gas Adsorption Prediction and Screening on Transition Metal
1College of Artificial Intelligence, Tianjin University of Science and Technology, Tianjin300457, China.
Langmuir : the ACS Journal of Surfaces and Colloids
|July 18, 2026
Summary
Machine learning now predicts gas adsorption on doped transition metal dichalcogenides (TMDs). This framework accelerates the discovery of new materials for hazardous gas sensing applications.
Area of Science:
- Materials Science
- Computational Chemistry
- Chemical Engineering
Background:
- Two-dimensional transition metal dichalcogenides (TMDs) show potential for gas sensing.
- Predicting gas adsorption on doped TMDs is challenging due to complex interactions.
Purpose of the Study:
- Develop a machine learning framework to predict adsorption energy on doped TMDs.
- Screen doped TMD materials for gas sensing applications.
Main Methods:
- Constructed a dataset of 354 first-principles adsorption entries for six gases, four TMD hosts, and metal dopants.
- Utilized 12 adsorption-configuration-independent descriptors.
- Employed boosted-tree regression models (GBR, XGB) and SHAP analysis for interpretation.
Main Results:
- Achieved high prediction accuracy (R2 > 0.95) with boosted-tree models.
- Identified key descriptors like gas-phase zero-point energy and dopant valence electron count.
- Successfully classified adsorption regimes.
Conclusions:
- The developed machine learning framework enables efficient screening of doped TMDs for gas sensing.
- Provides a thermodynamics-guided approach for materials discovery.
