Enantioselective SERS Sensing of Amino Acids by Self-Assembled Chiral Au Nanorods

Abstract

Chiral discrimination is essential in pharmaceutical and biochemical analysis but remains technically demanding due to the subtle physicochemical differences between enantiomers. Surface-enhanced Raman scattering (SERS) has emerged as a promising tool for chiral sensing, yet conventional SERS substrates suffer from poor structural homogeneity, unstable signal readouts, and elusive enantioselective mechanisms. Here, we present a flexible label-free enantioselective SERS platform based on macroscopic, self-assembled monolayers of chiral Au nanorods (CNRs) transferred onto a flexible polydimethylsiloxane (PDMS) substrate. By exploiting the intrinsic Raman peak of PDMS (488 cm–1) as a built-in internal standard, the platform drastically suppresses spatial inconsistencies, yielding a high and reproducible enantiomeric Raman intensity ratio (Il-Phe/Id-Phe) with a relative standard deviation of less than 9%. Principal Component Analysis (PCA) and a supervised Support Vector Regression (SVR) machine learning model were integrated, enabling accurate, label-free quantification of enantiomeric excess in binary mixtures with a high coefficient of determination (R2 = 0.9962) and low root-mean-square error (RMSE = 2.05%). Finite-difference time-domain (FDTD) simulations and Density functional theory (DFT) calculations reveal that the enantioselectivity originates from local chiral fields of CNRs and selective adsorption energies of enantiomers on high-index crystal facets. Collectively, this work establishes an effective strategy for chiral SERS sensing, highlighting the synergy between nanoscale self-assembly, plasmonic chirality, and machine learning for advanced enantioselective analysis.