Supramolecular Plasmonic Nanogap Engineering for Machine Learning Decoded Multiplex Chiral Amino Acid SERS Analysis

Chiral recognition sensing is pivotal in pharmaceutical synthesis and biomedical applications, yet three key challenges persist: the inability to simultaneously achieve high-resolution identification and characterization of enantiomers, insufficient understanding of the molecular mechanisms governing chiral recognition, and limited efficacy in analyzing complex mixtures of chiral compounds. Here, we present a supramolecular plasmonic nanoparticle-on-mirror (NPoM) platform for surface-enhanced Raman scattering (SERS) chiral sensing. By employing sulfhydryl-functionalized β-cyclodextrin (β-CD) as a supramolecular core for selective chiral amino acid capture, the platform facilitates the precise formation of a ∼1.2 nm plasmonic nanogap hotspot between gold nanoparticles (AuNPs) and a gold film (AuF). This configuration significantly enhances enantiomer-specific Raman signals, thereby enabling high-resolution identification and enantioselective discrimination of aromatic amino acid enantiomers. Density functional theory (DFT) calculations reveal the chiral recognition mechanism: β-CD forms diastereomeric complexes with d- and l-enantiomers through distinct spatial configurations and differential binding affinities, resulting in characteristic spectral differences that enable precise enantioselective identification. Furthermore, integration with machine learning (ML) algorithms facilitates rapid, highly accurate enantiomer classification and robust blind prediction of unknown samples. Notably, the proposed platform demonstrates exceptional performance in the simultaneous discrimination of complex enantiomeric mixtures comprising multiple chiral aromatic amino acids, providing a powerful strategy for advanced chiral analysis.