In Situ Grown Ag NF/PDMS Film SERS Sensor Coupled with Machine Learning for Sensitive Detection of Fungicides in Environmental Water

Abstract In this work, a two-dimensional SERS sensor (Ag NFs/PDMS) was fabricated by growing silver nanoflower (Ag NF) arrays in situ on amine-functionalized poly(dimethylsiloxane) (PDMS) via a facile one-step reduction strategy. FDTD simulations confirmed a high density of interparticle ″hot spots″ in the optimized substrate, endowing it with exceptional signal uniformity (RSD ≤ 5.46%) and batch-to-batch reproducibility. The sensor enabled sensitive detection of fungicides, crystal violet (CV), and thiram (TRM), across multiple environmental water matrices (tap water, pond water, and lake water), with limits of detection as low as 1.13 × 10–9 M and spike recoveries of 92.3–115.4% (RSD ≤ 5.12%, n = 3), well below the U.S. EPA permissible levels. Integrating SERS with machine learning algorithms─support vector machine, random forest, and decision tree─achieved 100% classification accuracy (AUC = 1.0) for simultaneous discrimination of CV and TRM in spiked pond water. This platform offers a rapid, reliable, and user-friendly solution for fungicide monitoring in environmental applications.