Biomimetic SERS platform with machine learning for sensitive detection of pesticides in complex food matrices

Shuangyun Li, Ting Cao, Meifeng Xu, Mengqi He, Xin Cai, Yao Lu, Yan Jin, Muzi Han, Hanwen Liu, Chaonan Wang,
Biomimetic SERS platform with machine learning for sensitive detection of pesticides in complex food matrices,
Optics & Laser Technology,
Volume 192, Part C,
2025,
113756,
ISSN 0030-3992,
https://doi.org/10.1016/j.optlastec.2025.113756.
(https://www.sciencedirect.com/science/article/pii/S0030399225013477)
Abstract: The widespread overuse of pesticides in recent years has posed severe threats to both public health and ecosystems.Herein, we report a biomimetic Ag NPs@PDMS substrate fabricated by microstructure replication from rose petals onto polydimethylsiloxane (PDMS) and subsequent Ag deposition via thermal evaporation. The optimized substrate demonstrated satisfactory SERS enhancement with an enhancement factor (EF) of 1.70 × 107, exceptional uniformity with an RSD of 5.92 %, excellent reproducibility, and good stability. Machine learning algorithms were employed to classify SERS spectra collected on the proposed substrate, during which the spectral data of different substances were mixed together. The algorithms demonstrated robust spectral classification capability, achieving 99.50 % accuracy not only for binary molecular systems but also for complex datasets containing spectral information from crystal violet (CV), rhodamine 6G (R6G), thiabendazole (TBZ), Sudan III, melamine, and malachite green (MG). In real-world application scenarios, the developed substrate was able to detect TBZ in the complicated tea solution at a concentration as low as 0.1 ppm, and when combined with machine learning algorithms, the system achieved remarkable classification accuracy exceeding 97.00 % for identifying TBZ-contaminated tea samples. This machine learning-enhanced substrate platform shows considerable promisefor food quality monitoring and control applications.
Keywords: SERS; PDMS; Ag; Pesticides; Machine learning