機械学習を用いた光酸化濃縮蛍光膜によるガス状水銀検出の強化

Severe mercury pollution from coal-fired flue gas drives the need for robust, cost-efficient, and high-fidelity detection. To address the challenges of complex processes and low accuracy in existing detection techniques, we develop a B,N-doped carbon dots-AgCl/Ag fluorescent membrane sensor (CDs-AgCl/Ag) based on a photo-controlled oxidation and enrichment strategy, integrated with machine learning (ML) to enhance detection precision. Upon visible-light excitation, in situ oxidation of Hg⁰ occurs via the surface plasmon resonance effect of Ag nanoparticles, while B,N-doped carbon dots capture oxidized Hg²⁺ to induce fluorescent responses. The color signal features of fluorescence images are analyzed by multiple ML models. The results show that both linear regression (Linear) and support vector regression (SVR) models exhibit excellent fitting performance for detecting Hg⁰, achieving a detection limit of 3.2 × 10^-7 g m^-3, a 310-fold sensitivity increase, and 97% accuracy. To the best of our knowledge, this work presents the first composite fluorescent membrane sensor integrated with ML for gaseous mercury detection in flue gas. In addition to superior sensitivity, our system shows clear advantages over conventional methods with lower cost and environmental impact, offering great potential for practical environmental monitoring.