Detection and Pattern Recognition of Chemical Warfare Agents by MOS-Based MEMS Gas Sensor Array

Area of Science:

• Chemical sensing
• Materials science
• Microelectromechanical systems (MEMS)

Background:

• Chemical warfare agents (CWAs) pose significant public safety risks.
• Developing rapid and reliable detection technologies for CWAs is crucial for early warning and public safety.

Purpose of the Study:

• To develop a compact, low-power sensor array for detecting multiple CWAs.
• To evaluate the sensor array's performance in terms of detection range, limit, and accuracy.
• To apply pattern recognition algorithms for CWA classification and identification.

Main Methods:

• Fabrication of a 24-sensor array using metal oxide semiconductor (MOS)-based MEMS technology.
• Testing sensor responses to various CWAs including hydrogen cyanide (AC), GB, GD, VX, and HD.
• Feature extraction from kinetic and dynamic sensing response characteristics.
• Application of Principal Component Analysis (PCA) and machine learning (narrow neural network) for classification.

Main Results:

• The MOS-based MEMS sensor array demonstrated good sensing performance for all tested CWAs.
• Detection limits were below Immediately Dangerous for Life and Health (IDLH) levels, enabling early warning.
• PCA showed distinct groupings for different CWAs.
• A narrow neural network achieved nearly 100.0% prediction accuracy for CWA classification.

Conclusions:

• The developed MOS-based MEMS sensor array is suitable for qualitative detection and early warning of CWAs.
• Integration with pattern recognition algorithms allows for rapid, in situ, and on-site CWA detection and identification.
• The sensor array shows great potential for portable CWA detection devices.