Ultrasensitive Isomer Discrimination: A Joint Surface-Enhanced Raman Scattering (SERS) Spectroscopy and Machine Learning Strategy

Abstract

Isomer discrimination is of paramount importance across various sectors, including pharmaceuticals, agriculture, and the food industry, owing to their unique physicochemical characteristics. Because of their extremely similar characteristics, traditional analytical methods fail or encounter severe limitations in isomer discrimination. To overcome this grand challenge, a novel sensing strategy is proposed based on surface-enhanced Raman scattering (SERS) substrates (i.e., plasmonic platforms) combined with machine learning algorithms. These plasmonic platforms exhibit exceptional signal uniformity across wide regions and sensitivity, enabling the discrimination of structural isomers (hydroquinone, resorcinol, pyrocatechol), geometric isomers ((Z/E)-stilbene, (Z/E)-resveratrol), and optical isomers (R/S-ibuprofen). Notably, for the analysis of optical isomers, 1-naphthalenethiol is employed as a probe to facilitate specific isomer orientation on the surface of the plasmonic platform through, for the first time, π–π interactions. The integration of machine learning methodologies, such as Partial Least Squares Regression and Artificial Neural Networks, significantly enhances both quantitative analysis and classification accuracy, achieving detection limits as low as 2 × 10⁻⁸ m. Validation with commercially available ibuprofen samples shows excellent agreement with traditional circular dichroism results, highlighting the method's robustness and precision. The strategy provides a versatile, ultrasensitive, and reliable solution for isomer discrimination, with broad applications in pharmaceuticals, environmental monitoring, and clinical diagnostics.