Magnetic-assisted microfluidic SERS platform integrated with deep learning for rapid identification of single and mixed antibiotic systems

Magnetic enrichment and oscillation were integrated into a microfluidic SERS chip. • Magnetic regulation enhanced SERS sensitivity and spectral reproducibility. • Chemical washing coupled with magnetic oscillation facilitated rapid substrate regeneration. • Deep learning-assisted spectral analysis achieved accurate recognition of single and mixed antibiotics. To improve the sensitivity, signal uniformity, and recognition capability of conventional microfluidic SERS (MF-SERS) systems, a magnetic-assisted MF-SERS platform integrating magnetic enrichment, magnetic oscillation, and deep learning-based recognition was developed. Highly sensitive Ag@Fe 3 O 4 @SiO 2 core-shell magnetic substrates were prepared and characterized. COMSOL simulations revealed that magnetic aggregation shortened the inter-microsphere spacing and enhanced plasmonic coupling between Ag nanoparticles on adjacent microspheres, resulting in a maximum local electric field intensity of 138.55 V/m and a maximum local theoretical electromagnetic enhancement of 3.68 × 10 8 . Benefiting from magnetic enrichment, the lowest experimentally detectable concentration of rhodamine 6 G (R6G) was improved from 10 −10 to 10 −11 mol/L. Magnetic oscillation further promoted substrate−analyte contact and improved adsorption uniformity, leading to a continuous decrease in the relative standard deviations (RSDs) of characteristic Raman peaks with increasing oscillation time. In addition, the combination of magnetic oscillation and chemical washing enabled efficient removal of antibiotic residues and rapid substrate regeneration. By integrating a multilayer perceptron model, the platform achieved accurate classification of 11 single-component and multicomponent antibiotic samples with an average test accuracy of approximately 95%. This work provides a promising strategy for constructing MF-SERS platforms with high sensitivity, improved signal uniformity, rapid regeneration, and intelligent analytical capability. Download: Download high-res image (142KB) Download: Download full-size image