Related Experiment Video
Updated: Aug 16, 2025
07:23
Using Insect Electroantennogram Sensors on Autonomous Robots for Olfactory Searches
Published on: August 4, 2014
23.1K
Ultra-Low-Power E-Nose System Based on Multi-Micro-LED-Integrated, Nanostructured Gas Sensors and Deep Learning
Kichul Lee1, Incheol Cho1, Mingu Kang1
1Department of Mechanical Engineering, Korea Advanced Institute of Science and Technology (KAIST), Daejeon 34141, Republic of Korea.
ACS Nano
|December 19, 2022
Summary
Researchers developed an ultra-low-power electronic nose (e-nose) using UV micro-LED gas sensors and AI. This novel system achieves high accuracy in gas detection, offering a power-efficient solution for environmental monitoring and IoT applications.
Area of Science:
- Materials Science and Engineering
- Sensor Technology
- Artificial Intelligence
Background:
- Growing demand for air quality monitoring and industrial safety drives the need for advanced gas sensors.
- Semiconductor metal oxide (SMO)-type sensors offer advantages but lack selectivity, necessitating electronic nose (e-nose) systems.
- Conventional e-nose systems face challenges with high power consumption as sensor count increases.
Purpose of the Study:
- To develop an ultra-low-power e-nose system utilizing ultraviolet (UV) micro-light-emitting diode (μLED) gas sensors and a convolutional neural network (CNN).
- To overcome the selectivity limitations of traditional gas sensors through an integrated sensor array and pattern recognition approach.
- To demonstrate a power-efficient and highly accurate gas sensing solution for environmental Internet of Things (IoT) applications.
Main Methods:
- Fabrication of monolithic photoactivated gas sensors by depositing nanocolumnar In2O3 film with plasmonic metal nanoparticles (NPs) directly onto μLEDs.
- Development of an e-nose system comprising two distinct μLED sensors with silver and gold NP coatings.
- Application of pattern recognition and CNN algorithms to analyze sensor responses for gas classification and concentration regression.
Main Results:
- Achieved a total power consumption of 0.38 mW, a significant reduction compared to conventional heater-based e-nose systems.
- Demonstrated a gas classification accuracy of 99.32% for five different gases (air, ethanol, NO2, acetone, methanol).
- Obtained a mean absolute error of 13.82% for gas concentration regression, indicating high selectivity and accuracy.
Conclusions:
- The developed μLED-based e-nose system offers a highly selective, real-time, and ultra-low-power solution for gas sensing.
- The system's low power consumption enables stable, long-term battery operation, making it suitable for environmental IoT deployments.
- This technology represents a significant advancement in gas sensing for applications requiring continuous monitoring and high accuracy.
