Room-Temperature Trace NO2 Monitoring System Based on Two-Dimensional Heterostructures and Integrated with Deep Learning.

Accurate detection of trace NO2 at room temperature is crucial for air quality control and the early diagnosis of respiratory diseases. High-precision detection of low-concentration gases has long been a research focus, with material optimization and deep learning algorithms emerging as effective strategies to enhance sensor accuracy. Herein, a remote NO2 monitoring system for low-concentration detection is proposed, based on Bi2S3/WO3 heterostructures, wireless communication modules, and deep learning techniques. The sensor exhibits a high response of 17.9 to 5 ppm NO2, a sensitivity of 3.84/ppm, rapid response/recovery times (27/110 s), excellent selectivity, and reliable stability. These superior performances are attributed to the enhanced charge transfer, carrier separation, and the generation of active oxygen species of the formed heterostructures. To overcome the limitations posed by data scarcity, a 1D-CNN/LSTM deep learning model is introduced, achieving accurate ppb-level regression with an R2 value of 0.9826 after data augmentation. This model significantly improves detection accuracy at low concentrations. Furthermore, by integrating wireless communication modules, the system supports real-time monitoring, multichannel operation, and intelligent alarming, offering a novel strategy for high-precision gas detection at room temperature.