Ag nanoparticle-engineered carbon nitrogen as a high performance SERS platform for detecting organic pollutants assisted with machine learning

Surface-enhanced Raman scattering (SERS) has emerged as a highly effective technique for ultra-sensitive chemical analysis, enabling the detection of trace amounts of organic pollutants (OPs) through detailed spectral information. In this study, we propose a simple in situ deposition method to fabricate a hybrid SERS substrate combining graphite-phase carbon nitride (g-C 3 N 4 ) with Ag nanoparticles (Ag NPs). The size and distribution of the Ag NPs were finely tuned by adjusting the mass of AgNO 3 . The optimized g-C 3 N 4 /Ag composite exhibited excellent performance for SERS detecting methylene blue with a detection limit of as low as 2.81 × 10 −9 M. The enhanced SERS activity can be attributed to the plasmonic effect of the densely packed Ag NPs and the adsorption properties of g-C 3 N 4 . Furthermore, the substrate demonstrated excellent reproducibility, good uniformity and long-term stability. Additionally, when integrated with machine learning using a Random Forest algorithm, the system accurately identified four similar OPs—methylene blue, crystal violet, rhodamine 6 G, and thiram —with a classification accuracy of above 99.5 %. This work highlights the potential of combining SERS with machine learning for effective environmental and food safety monitoring.