Machine learning-assisted photoelectrochemical/fluorescence dual-mode sensor for ultra-sensitive glyphosate detection

Xiaoyan Wang, Zhen Cao, Baolong Shi, An Zhao, Jidong Dai, Sheng-Tong Wu, Dalei Wang, Wei Liu,
Machine learning-assisted photoelectrochemical/fluorescence dual-mode sensor for ultra-sensitive glyphosate detection,
Chemical Engineering Journal,
Volume 525,
2025,
170337,
ISSN 1385-8947,
https://doi.org/10.1016/j.cej.2025.170337.
(https://www.sciencedirect.com/science/article/pii/S1385894725111819)
Abstract: The abuse of glyphosate has generated substantial risks to natural environments and public health, emphasizing the critical role of precise and timely identification of glyphosate in diverse ecological and food sources. In this work, a dual-mode photoelectrochemical (PEC) and fluorescence (FL) sensor based on a CdTe/Bi4O5Br2 heterostructure was constructed, and machine learning was embedded into the PEC and FL signal fusion process for the first time. Experiments demonstrate that machine learning-based fusion of photocurrent and fluorescence information reduces the detection limit for glyphosate to 1.27 fM, which was approximately 3.4-fold and 5.4-fold improvements over standalone PEC and FL detection modes, respectively. The recovery rates in the three types of matrices (apples, lettuce, and tap water) were 97.68–104.94 %, and the K-Nearest Neighbor model had a classification accuracy rate of 100 % for the spiked samples. The dual-mode sensor combines fast response, resistance to environmental interference and ease of operation, which offers a novel instrument for real-time detection of minute pesticide residues in complex environments.
Keywords: Dual-mode sensor; Photoelectrochemical; Fluorescence; Glyphosate; Machine learning