Co-assembly SERS strategy for trace detection and intelligent recognition of controlled-drug mixtures

Surface-enhanced Raman scattering (SERS) is promising for trace chemical analysis, drug detection, and illicit-substance screening. However, conventional assembly-based substrates generally rely on post-assembly soaking, which limits analyte access to nanoscale hot spots and compromises sensitivity and mixture recognition. Here, a co-assembly SERS strategy is proposed by introducing analyte molecules during noble-metal nanoparticle assembly, enabling synchronous enrichment of targets within plasmonic hot-spot regions and enhanced Raman responses. Rhodamine 6G, scarlet and benzimidazole were used as model molecules to verify the superior enhancement of the co-assembly method over the conventional soak mode for cationic, anionic, and drug-like molecules. Sufentanil and midazolam were further selected as representative drug molecules for Raman fingerprint acquisition and low-concentration detection, with lowest tested concentrations of 5 ppm and 0.5 ppm, respectively. To address peak overlap, intensity fluctuation and manual interpretation challenges in mixed-drug spectra, a rule-based multidimensional spectral matching algorithm was developed by integrating peak-position, peak-intensity-ratio and local waveform-similarity matching through confidence-score fusion. This work provides a proof-of-concept for combining co-assembly-enhanced SERS with rule-based multidimensional spectral matching for the recognition of representative controlled-drug molecules in model mixtures.