Faceted gold nanocrystal film over nanosphere for sensitive SERS distinction of fungicides by deep learning analysis

Metal film over nanosphere (MFON) as a classical SERS-active substrate has attracted numerous attention and applications due to its excellent sensitivity and signal uniformity. However, such MFON substrates prepared by metal deposition require strictly controlled experimental conditions with high cost to obtain high batch reproducibility. In addition, the precise adjustment of the metal particle fineness and crystal facet is hard to conduct, limiting the SERS signal for further improvement. In this work, we propose multifaceted Au nanocrystal over nanosphere (AuNCFON) structures based on facile and robust interfacial assembly technique. Notably, the preparation method could reduce significant loss of metallic materials without the requirement of large-scale instrumentation, and allows precise control of the metal film thickness and the nanocrystal facet. Specifically, the size and facet of nanocrystals on PS spheres are systematically regulated and the Au dodecahedral films with {110} facet on PS arrays exhibit optimal SERS activity. The SERS intensity is amplified by an enhancement factor of 0.13×107 with excellent signal homogeneity and batch reproducibility. Finally, the sensitive and accurate distinction of similarly structured molecules including crystal violet and malachite green (fungicides) is demonstrated assisted by principal component analysis-partial least squares (PCA-PLS) algorithm. The AuNCFON SERS substrate combined with deep learning analysis has a broad application prospect in reliable identification of chemical pollutants.