Multigas Identification by Temperature-Modulated Operation of a Single Anodic Aluminum Oxide Gas Sensor Platform and...

Area of Science:

• Materials Science
• Chemical Sensing
• Artificial Intelligence

Background:

• Semiconductor metal oxide (SMO) gas sensors offer high sensitivity and cost-effectiveness for gas detection.
• A key limitation of SMO sensors is their poor selectivity, hindering accurate identification of specific gases.
• Existing methods to improve selectivity, such as new materials and filters, have proven insufficient.

Purpose of the Study:

• To address the selectivity challenge in SMO gas sensors.
• To implement temperature-modulated operation on a single SMO gas sensor using an anodic aluminum oxide (AAO) microheater platform.
• To develop a deep learning model for accurate gas classification and concentration estimation.

Main Methods:

• Utilized an AAO microheater platform for stable temperature modulation of an SMO gas sensor.
• Applied a staircase waveform with six distinct temperature conditions.
• Collected gas response data for acetone, ammonia, ethanol, and nitrogen dioxide.
• Employed a convolutional neural network (CNN) for pattern recognition and prediction.

Main Results:

• Achieved a high gas classification accuracy of 97.0%.
• Obtained mean absolute percentage errors (MAPE) for concentration estimation: acetone (13.7%), ammonia (19.2%), ethanol (19.8%), and nitrogen dioxide (19.4%).
• Successfully differentiated between similar odors, outperforming human olfactory capabilities.

Conclusions:

• Temperature modulation combined with CNN-based deep learning significantly enhances SMO gas sensor selectivity and accuracy.
• This integrated approach provides a robust solution for precise gas identification and quantification.
• The method demonstrates potential for advanced gas sensing applications, including distinguishing complex odor mixtures.