A self-enriching capillary SERS platform combined with CNN for rapid detection and classification of micro/nano plastics in water

Micro/nanoplastics particles (M/NPPs) have been increasingly detected in various organisms, including humans, raising growing concerns about their environmental and health impacts. This study presents a simple and effective capillary tube-based surface-enhanced Raman scattering (SERS) method for the rapid and ultra-sensitive direct detection of M/NPPs in water without drying. A high-performance SERS substrate was fabricated by uniformly modifying the capillary inner wall with a monolayer of silver nanoparticles (MAgNPs), creating densely distributed “hot spots”. Using rhodamine 6 G, the substrate achieved a detection limit of 10 −13 M with an enhancement factor of 5.81 × 10 10 . Capillary action drew polystyrene (PS) NPPs into the capillary tube, where flotation and enrichment at the tube top significantly improved detection efficiency. Both experiments and simulations confirmed this enrichment mechanism. The method detected 70 and 300 nm PS NPPs in lake water samples using spiking experiments at concentrations as low as 0.001 mg/mL. By combining SERS spectra with a convolutional neural network (CNN), mixed M/NPPs were accurately classified and identified, achieving 99.05 % recognition accuracy. This work offers a promising approach for rapid detection and precise identification of M/NPPs in environmental monitoring.