Serum-Stabilized Pickering Emulsions as Reproducible Liquid SERS Analyzer for Direct Metabolic Fingerprinting

Abstract

Reliable Surface-enhanced raman spectroscopy (SERS) analysis of serum remains challenging due to strong matrix interference and the poor stability and reproducibility of conventional solid substrates. Herein, we report a reproducible, liquid-phase, label-free SERS analyzer based on serum-stabilized Pickering emulsions. In this system, intrinsic serum components regulate interfacial assembly and nanoparticle distribution, enabling direct detection of molecular fingerprints in complex serum matrices. By directly using human serum as the aqueous phase and optimizing emulsification conditions with a n-hexadecane–ethanol mixed oil phase, a robust oil-in-water emulsion substrate was rapidly constructed, exhibiting excellent signal uniformity with relative standard deviations below 7%. The Pickering emulsion effectively enriches serum metabolites while minimizing macromolecular background interference, enabling label-free spectral acquisition. Our analysis reveals significant variations in amino acids, glucose, and purine-related metabolites, underscoring the analyzer's sensitivity for metabolic profiling. Multivariate techniques, including principal component analysis and a feedforward neural network, enhance spectral feature extraction and metabolic state classification. This liquid SERS strategy combines high reproducibility, minimal sample pretreatment, and compatibility with data-driven analysis, establishing a versatile tool for serum metabolite fingerprinting and metabolic studies in complex biofluids.