Analysis of total phenolic compounds and caffeine in teas using variable selection approach with two-dimensional fluorescence and infrared spectroscopy

This paper reports the quantification of total phenolic compounds and caffeine in green and black tea using fluorescence, medium (MIR), and near (NIR) infrared spectroscopy combined with spectral variables selection tools (wavenumber or excitation/emission pairs). Nineteen tea samples were analyzed with a traditional analytical method for phenolic (Folin-Ciocalteu) and the content was expressed as milligrams of Gallic Acid Equivalent per gram of the sample (mg GAE g(-1) dry basis). Caffeine quantification was performed by high-performance liquid chromatography. To optimize the variable selection and predict total phenolic and caffeine content, we applied a Pure Spectra chemometric Modeling (PSCM). The predictive quality of the models was assessed using metrics: Root Mean Square Error (RMSE), Mean Absolute Error (MAPE) and Determination Coefficient (R-2). The spectroscopic techniques fluorescence, MIR, and NIR presented comparable metrics. Using only 4 variables, RMSE was less than 5.82 mg GAE g(-1) for the Test subset to predict total phenolic compounds and 1.79 mg g(-1) to predict caffeine. This work demonstrated that fluorescence, MIR, and NIR spectroscopy with the PSCM algorithm could be used to analyze the content of total phenolic compounds and caffeine quickly and non-destructively, using a small number of variables.

Automated summary

This study quantified total phenolic compounds and caffeine in green and black tea using fluorescence, MIR, and NIR spectroscopy combined with spectral variables selection tools. The results showed that these spectroscopic techniques, with the PSCM algorithm, could accurately predict total phenolic and caffeine content using a small number of variables.