A nanoengineered Au@Ag@PVP SERS solid-phase substrate empowered by machine learning for rapid and trace-level screening of pesticide residues in tea

Pesticide residues in tea pose significant health risks to consumers. This research focuses on developing a surface-enhanced Raman scattering (SERS) sensor using a solid-phase substrate made of Au@Ag@PVP nanoparticles (NPs), combined with machine learning chemometric techniques, for the rapid detection of trichlorfon, carbendazim, and clothianidin in tea. Au@Ag@PVP NPs were immobilized onto anodic aluminum oxide (AAO) filter membranes via vacuum filtration, yielding densely packed NPs that generate strong electromagnetic hot spots, thereby increasing SERS sensitivity. Raman spectra of tea infusions containing different concentrations of pesticides were acquired and analyzed using three machine-learning algorithms. Among these models, competitive adaptive reweighted sampling-partial least squares (CARS-PLS) demonstrated superior predictive performance, with Rc = 0.9950, Rp = 0.9942, and RPD = 9.2893 for trichlorfon; Rc = 0.9955, Rp = 0.9942, and RPD = 9.1010 for carbendazim; and Rc = 0.9981, Rp = 0.9957, and RPD = 9.7523 for clothianidin. This approach enables the rapid, quantitative detection of pesticide residues with minimal sample preparation and high sensitivity, representing a practical method for monitoring tea safety.