Real-Time Gas Identification at Room Temperature Using UV-Modulated Sb-Doped SnO₂ Sensors via Machine Learning.

Area of Science:

• Materials Science
• Chemical Sensing
• Machine Learning

Background:

• Conventional gas sensors often require multiple parameters like response and recovery times, limiting real-time applications.
• Thermally activated sensors have inherent limitations in speed and energy efficiency for continuous monitoring.

Purpose of the Study:

• To develop a novel, real-time gas identification method at room temperature.
• To utilize UV-modulated Sb-doped SnO₂ sensors and machine learning, relying solely on gas response (R).

Main Methods:

• Employed UV-modulated Sb-doped SnO₂ sensors with five distinct UV intensity levels (5-30 mW/cm²).
• Generated a five-dimensional optical fingerprint based on gas response (R) under varying UV illumination.
• Applied machine learning algorithms, including Support Vector Machine (SVM) and K-Nearest Neighbors (KNN), for gas discrimination.

Main Results:

• Achieved nearly 100% accuracy in discriminating between oxidizing (O₃, NO₂) and reducing (NH₃, H₂) gases.
• Demonstrated the effectiveness of using gas response (R) as the sole metric for classification.
• Validated the method's capability for instantaneous response detection.

Conclusions:

• The proposed method enables real-time gas identification without relying on time-dependent sensor parameters.
• This approach simplifies sensor operation and enhances detection speed, crucial for rapid monitoring systems.
• Paves the way for advanced, efficient gas sensing technologies.