Combining fluorescence anisotropy and parallel factor analysis to estimate the degree of polymerization in red wine

Tannins in red wine are polymer structures of flavanols, such as catechin and epicatechin, and their degree of polymerization affects sensory attributes, especially bitterness and astringency. Traditional analytical methods for assessing the degree of polymerization are tedious and involve extensive sample preparation. In this study, we have investigated fluorescence spectroscopy with and without polarization as a faster alternative. Two sample sets were used to study this: a simple sample system consisting of six chemical standards and a set of six red wine samples. The samples were diluted prior to fluorescence excitation-emission matrix measurements, and the three-way data were decomposed using the trilinear model Parallel Factor Analysis (PARAFAC). With the non-polarized measurements we show that when flavanols polymerize the excitation/emission position of the fluorescent signal does not change, but the quantum yield decreases. For the polarization measurements (anisotropy), the PARAFAC model succeeded in resolving the flavanol fluorescent component, which was used to calculate the polarization of the samples. The polarization results showed a strong correlation (r² = 0.94) with the traditional reference method for estimating the degree of polymerization. Our approach combines fluorescence anisotropy with PARAFAC decomposition for calculating polarization. This is a promising novel method for estimating the mean degree of polymerization, both in simple sample systems and in real red wine samples.