From discovery to optimization: Data-driven development of NH₃ sensing materials validated by experiments and DFT.

Area of Science:

• Materials Science
• Environmental Science
• Chemical Engineering

Background:

• Ammonia (NH3) is a hazardous air pollutant with significant agricultural and industrial sources.
• NH3 exacerbates haze formation and poses risks to human health, necessitating sensitive monitoring.
• Current gas sensor development is inefficient, and existing machine learning (ML) approaches lack practical focus.

Purpose of the Study:

• To develop an integrated framework for accelerated discovery of NH3 gas-sensing materials.
• To utilize ML, literature mining, experimental validation, and DFT analysis for practical sensor development.
• To identify novel SnO2-based NH3 sensors with enhanced performance.

Main Methods:

• A Stacking ML model was employed to predict NH3 sensing performance of SnO2-based materials.
• Literature mining and ML were combined to guide experimental design and parameter selection.
• Density Functional Theory (DFT) calculations were used to elucidate sensing mechanisms.
• Experimental validation confirmed ML predictions and sensor performance.

Main Results:

• The Stacking model achieved high accuracy (RMSE of 0.389 ppm⁻¹ and R² of 0.874) in predicting NH3 sensing.
• ML predictions for modifiers, loadings, and temperatures showed <20% deviation from experimental results.
• Tantalum-loaded SnO2 was identified as a novel NH3 sensor with a record-low detection limit of 9.7 ppb.
• DFT revealed that tantalum enhances NH3 adsorption through electronic state redistribution and increased charge transfer.

Conclusions:

• The integrated framework effectively accelerates the discovery and development of gas-sensing materials.
• Tantalum-loaded SnO2 represents a promising material for highly sensitive NH3 detection.
• This study offers a practical strategy for advancing gas sensor technology through combined computational and experimental approaches.