Room-Temperature Gas Identification via Photochemically Activated p-type CuCrO<sub>2</sub> Nanostructured Films Using Machine Learning
Yan-Fong Lin, Zi-Chun Tseng, Wen–Jeng Hsueh, Jiann-Heng Chen, Chun‐Ying Huang
Abstract
Operating gas sensors at different temperatures is a common approach to improve selectivity, but it often compromises long-term stability and hinders real-time detection. In this study, we introduce photochemically activated p-type CuCrO 2 nanostructured thin films, which generate distinct optical fingerprints under different UV light intensities, enabling real-time gas identification. The hexagonal-like nanostructure of the CuCrO 2 film increases the surface area, significantly enhancing the sensor’s sensitivity and selectivity. The sensor demonstrated a gas response of approximately ∼3.6 for various concentrations of N -propanol, NO 2, NH 3, ethanol, methanol, and formaldehyde under 20 mW/cm 2 UV illumination, with response times ranging from 50 to 200 s. By applying machine learning techniques, we achieved a classification accuracy exceeding 90%. Furthermore, regression analysis enhanced the predictive accuracy of gas concentrations, with R 2 values above 0.9. The sensor also exhibited excellent long-term stability, with only 7.4% degradation over 90 days and consistent performance under varying humidity. This work introduces a strategy to address selectivity and stability challenges, making these sensors suitable for environmental monitoring and industrial safety applications.