Electrospun nylon/Ag@ZIF-8 nanofibers for robust SERS detection of gaseous molecules assisted by deep learning

The detection of gaseous molecules using surface-enhanced Raman spectroscopy (SERS) remains challenging due to the rapid diffusion and low intensity of gas molecules. In this study, we developed a flexible and breathable SERS substrate by in-situ growth of Ag@ZIF-8 nanoparticles on electrospun nylon 6/6 fibers. The microchannels in the fiber membrane, combined with the high porosity of ZIF-8, enable effective capture and enrichment of volatile organic compounds (VOCs) in SERS hot spots. The optimized substrate demonstrated a detection limit of 6.64 × 10 −8 M for dithiohydroquinone (DP) gas, along with excellent reproducibility, linear correlation (R 2  = 0.99), and stability over 30 days. Furthermore, using a residual neural network (RNN) model for deep learning analysis, the substrate achieved 100 % accuracy in detecting gaseous glutaraldehyde (GA), a typical VOC found in disinfectants. This study showcases the potential of nylon/Ag@ZIF-8 substrates, enhanced by artificial intelligence, for practical and precise detection of gaseous molecules in various environments.