High Accuracy Real-Time Multi-Gas Identification by a Batch-Uniform Gas Sensor Array and Deep Learning Algorithm.

Area of Science:

• Materials Science
• Sensor Technology
• Artificial Intelligence

Background:

• Semiconductor metal oxide (SMO) gas sensors are crucial for environmental monitoring but suffer from low selectivity and accuracy issues.
• Existing electronic nose (E-nose) systems face challenges with sensor non-uniformity and real-time detection limitations.

Purpose of the Study:

• To overcome the selectivity and accuracy limitations of SMO gas sensors.
• To develop a system for real-time, selective gas detection using uniform sensor arrays and deep learning.

Main Methods:

• Fabrication of uniform nanocolumnar metal oxide (SnO2, In2O3, WO3, CuO) gas sensor arrays using glancing angle deposition.
• Application of a convolutional neural network (CNN) for pattern recognition of sensor array responses.
• Preprocessing of sensor response data for input into the CNN.

Main Results:

• Achieved high batch uniformity in fabricated gas sensor arrays.
• Demonstrated real-time selective gas detection for CO, NH3, NO2, CH4, and acetone with high accuracy (98%).
• Reported minimum response times of 1, 8, 5, 19, and 2 seconds for the respective gases.

Conclusions:

• The combination of uniform SMO sensor arrays and CNN-based deep learning effectively resolves selectivity and accuracy issues in gas sensing.
• This approach enables reliable, real-time detection of multiple gases, paving the way for advanced environmental monitoring applications.