Electrochemical construction of Turing-patterned plasmonic MOF substrate for machine learning-assisted SERS analysis of volatile organic compounds

Plasmonic MOF materials offer a promising solution to the low affinity of volatile organic compounds (VOC) toward conventional pure-metal surface-enhanced Raman scattering (SERS) substrates. However, the widely adopted bottom-up synthesis approaches often suffer from the irregular aggregation of metal nanostructures, impairing signal reproducibility, and the formation of flexible MOF with ligand defects, compromising selectivity. Here, we report a purely electrochemical top-down strategy to fabricate a Turing-patterned plasmonic MOF SERS substrate, in which the electrochemically sculptured Turing-patterned Ag enables highly reproducible Raman signal amplification, while the subsequently current-grown rigid ZIF-8 layer—with restricted ligand dynamics—selectively enriches volatile organic compounds via hierarchical porosity and non-covalent interactions. Notably, the Turing-patterned Ag@ZIF-8 enables instantaneous readout of the molar ratios of chloroform, benzene, and benzaldehyde in the gaseous mixtures solely by their Raman intensity ratios. By integrating spectral collection via portable Raman spectrometers with machine learning-assisted spectral analysis, we can readily distinguish colorectal cancer-related VOC profiles from healthy ones in simulated breath samples with an accuracy up to 94.0%. This work presents a novel practical approach for constructing low-cost, high-performance plasmonic MOF SERS substrates, advancing point-of-care diagnostics by leveraging SERS technique and artificial intelligence. © 2026 Elsevier B.V.