| Feb 20, 2025 |
AI system achieves near-perfect accuracy in detecting toxic gases |
| (Nanowerk Spotlight) When a chemical plant leaked toxic gas in Texas in 2014, first responders couldn’t quickly determine which specific compounds had been released. This delayed evacuation decisions and proper medical treatment. Current gas detection technology often fails to distinguish between similar chemicals in real-time or identify complex mixtures of gases without costly, time-consuming analysis, leaving critical gaps in environmental monitoring, industrial safety, and medical diagnostics. |
| Standard electronic “noses” can detect the presence of gases but struggle to tell apart compounds with similar structures—like benzene versus toluene, which differ by just one carbon atom. Even advanced laboratory equipment like mass spectrometers, while precise, requires hours of analysis and trained operators. Meanwhile, biological noses achieve remarkable sensitivity through millions of specialized receptor cells, but even they sometimes confuse similar molecules. |
| The fundamental challenge lies in capturing a gas’s unique signature. Most sensors measure only how strongly a gas binds to a surface. But gases can interact with materials in multiple ways—through charge transfer, electron doping, Schottky barrier modulation, dipole-dipole interactions, or subtle structural changes. A truly effective detection system needs to capture these diverse interactions and process the complex data efficiently. |
| A team of researchers from the Chinese Academy of Sciences has created an artificial intelligence system that captures these subtle differences by combining specialized organic sensors with advanced pattern recognition. Their system measures not just the strength of a gas’s interaction with sensor materials, but also the direction of the change and how the signal evolves over time – similar to how biological noses process multiple aspects of a scent. |
| The team reported their findings in Advanced Materials ("Bionic Olfactory Neuron with In-Sensor Reservoir Computing for Intelligent Gas Recognition"). |
 |
| a) Schematical illustration of the relationship between gas sensors and sustainable development of human society. b–d, h) The gas discrimination processes in mammalian olfactory systems. e–h) The gas discrimination processes by the proposed intelligent olfactory systems in this work. (Image: Reprinted with permission by Wiley-VCH Verlag) (click on image to enlarge) |
| The system uses nine different organic semiconductors – materials that can conduct electricity under specific conditions – arranged in an array. Each semiconductor responds differently to various gases, creating a unique pattern of electrical changes. When a gas molecule interacts with these materials, it can transfer electrons, create areas of positive and negative charge, or cause the material to expand slightly. The variety of these interactions helps create distinct signatures for each gas. |
| The researchers paired these sensitive materials with a computational method called reservoir computing, which efficiently processes complex patterns while using less power than traditional artificial neural networks. To classify the detected gases, the system employs a machine learning technique called k-nearest neighbors (KNN), which groups similar patterns together for rapid and accurate identification. This combination achieved 100% accuracy in identifying eight different gases, including industrial chemicals like methanol and acetone, as well as more dangerous compounds like nitrobenzene and nerve agent simulants. |
| More impressively, when challenged with a library of 26 different gases—including mixtures and nearly identical chemical variants—the system maintained 99.04% accuracy. This level of precision in distinguishing similar compounds is among the highest reported for electronic nose technologies. |
| To demonstrate practical applications, the scientists mounted their sensor array on a robotic dog designed to detect water pollutants. The robot successfully identified several hazardous compounds dissolved in water, including carcinogens and nerve agent simulants, by analyzing the gases released from contaminated samples. The system could distinguish these compounds both individually and in mixtures—a crucial capability for real-world environmental monitoring. |
| The technology addresses key limitations of current gas detection methods. Unlike laboratory equipment, it provides rapid results without requiring extensive training to operate. The system processes data directly on the device rather than sending information to a central server, making it suitable for distributed sensor networks monitoring air quality or industrial emissions. |
| The researchers aim to test the system under more challenging conditions, where changing humidity, temperature variations, and interfering compounds could affect performance. They also plan to expand the range of detectable gases and further improve the system’s energy efficiency. |
| By combining sensitive organic materials with efficient artificial intelligence, this technology offers a new approach to detecting and identifying gases in complex environments. The high accuracy in distinguishing similar compounds could transform how we monitor air quality, detect industrial leaks, and diagnose diseases through breath analysis. |
By
Michael
Berger
– Michael is author of four books by the Royal Society of Chemistry:
Nano-Society: Pushing the Boundaries of Technology (2009),
Nanotechnology: The Future is Tiny (2016),
Nanoengineering: The Skills and Tools Making Technology Invisible (2019), and
Waste not! How Nanotechnologies Can Increase Efficiencies Throughout Society (2025)
Copyright ©
Nanowerk LLC
|
Become a Spotlight guest author! Join our large and growing group of guest contributors. Have you just published a scientific paper or have other exciting developments to share with the nanotechnology community? Here is how to publish on nanowerk.com. |
