Machine learning assisted biomimetic flexible SERS sensor from seashells for pesticide classification and concentration prediction

Complex surface structures are among the critical materials for fabricating sensors for surface-enhanced Raman scattering. In this study, a novel biomimetic sensor was fabricated by replicating the surface structure of seashells, and uniform silver nanoparticles were imparted to it via in-situ synthesis. The natural surface structure of seashells, characterized by numerous pores, folds, and protrusions, plays a pivotal role in creating electromagnetic ’hotspots’ that significantly enhance the excitation of Raman signals. Meanwhile, the synergistic interaction between the PDMS material and the seashell-like surface structure to form biomimetic sensors with excellent SERS performance, high uniformity, and stability. Moreover, to integrate Raman spectroscopy detection with spectral analysis, we established a machine-learning-based classification prediction model. The Raman spectra detected using the seashell biomimetic sensor were used for training, enabling the classification and prediction of the types and concentrations of target substances. The proposed detection and analysis system demonstrates promising potential for practical applications.