A fast and highly efficient strategy for detection of camellia oil adulteration using machine learning assisted SERS
Peipei Xu, Qingling Nie, Runbing Huang, Jing Shi, Junjie Ren, Ruiyun You, Hengfang Wang, Yanlian Yang, Yudong Lu
Abstract
Camellia oil (CAO) is a high-quality edible vegetable oil, commonly known as “Oriental olive oil," with medical value and biological activity, but easily adulterated. Currently, we developed a method that combines Surface Enhanced Raman Spectroscopy (SERS) with machine learning for the effective identification of camellia oil. SERS is essential in this context because it significantly enhances the sensitivity and specificity of the detection process, allowing for the identification of even minor adulterations that traditional methods may overlook. We employed SERS spectra of both pure and adulterated camellia oil on an NPAg sheet coated with 4-thiobenzonitrile (4MBN) at a concentration of 0.02% for the machine learning analysis. The utilization of 4MBN for signal generation within the Raman silent region further enhances the stability of spectral acquisition and ensures more accurate results. The k nearest neighbors (KNN) model exhibited superior performance, achieving a test set accuracy of 97.24%. Consequently, the NPAg sheet@[email protected]%ER strategy, designed to amplify compositional differences in edible oils, emerges as an effective tool for rapidly verifying the authenticity of such oils. • Portable Raman for high volume operations. • Mass Production of Transparent SERS Substrates with Excellent Mechanical Stability. • The KNN model had the best performance with an accuracy of 97.24% on the test set.