High-precision authentication of cigarettes using surface-enhanced Raman scattering spectroscopy combined with machine learning

Abstract

Rapid and reliable authentication of genuine versus counterfeit cigarettes is challenging because products vary across brands and production batches. Here, we develop a surface-enhanced Raman scattering (SERS)-based screening method for cigarette authentication and evaluate it across 17 commercial brands using tobacco water extracts from genuine and confirmed counterfeit products. Method optimization identified citrate-reduced silver nanoparticles (AgC NPs) as the most reproducible SERS substrate, and an extraction temperature of 40 °C provided the best balance of signal stability and class separability. Under these conditions, eight machine-learning classifiers were benchmarked on SERS spectral fingerprints; a quadratic-kernel support vector machine (SVM) achieved the best overall performance, combining high accuracy with strong generalization while limiting overfitting. The optimized model reached 99.7% accuracy, 99.7% sensitivity, and 99.9% specificity for discriminating genuine versus counterfeit cigarettes across the 17 brands. In an external blind-sample evaluation using ten unknown cigarettes from previously uncharacterized batches, the model maintained >80.0% authentication accuracy, indicating tolerance to batch-to-batch variability. Overall, this SERS-machine learning strategy enables rapid, high-accuracy cigarette authentication and shows promise for field-deployable screening.

Abstract as supplied by the indexing source. © the authors / publisher.