Differentiation of Black Tea According to Country of Origin Using the μ-CTE/TD/GC-MS Method Combined With Decision Tree-Optimizable Neural Network Analysis

Abstract

BACKGROUND: Accurate discrimination of the country of origin of teas is critical to determine their actual commercial value, to meet consumer preferences, and to ensure compliance with labeling regulations. Therefore, in this study, we developed a new approach to accurately discriminate the country of origin of teas in the Turkish market. RESULTS: A thermal desorption/gas chromatographic–mass spectrometric (TD/GC-MS) method combined with optimizable neural networks (ONN) was developed to analyze the volatile organic compounds (VOCs) of tea samples subjected to infusion or grinding pretreatments. Prior to GC-MS analysis, the conventional thermal desorption method was applied to VOCs in the powdered teas, while VOCs in the infused teas were adsorbed on Tenax-TA sorbent tubes attached to a micro-chamber/thermal extractor (μ-CTE) and then thermally desorbed. Using a feature selection technique, a total of 11 VOCs from infused tea samples, 21 VOCs from ground tea samples, and 18 VOCs from both groups were identified as specific VOCs that critically affect the classification of the teas. As a result of ONN classification of selected VOCs from only ground tea samples and infused tea samples, 95.51% and 96.7% accuracy was obtained, respectively, while 100% classification accuracy was achieved by ONN classification of VOCs from both sample groups. CONCLUSION: The results showed that different pretreatments applied to Turkish and Ceylon teas caused the release of different volatile compounds, resulting in more specific VOC profiles. In addition, the developed μ-CTE/TD/GC-MS method allowed a more accurate classification of the black tea samples than the TD/GC-MS system alone.