Characterization of Encapsulated Flavor Systems by NIR and Low-field TD-NMR: A Chemometric Approach

A quantitative method for measuring simultaneously the flavor and water contents in model spray-dried flavor delivery systems was developed using spectroscopic techniques and chemometrics. Nine encapsulated systems were prepared, consisting of a solid carrier (maltodextrin and gum arabic) and varying the amounts of water and flavor. The model flavors used in this work were a hydrophobic (limonene) and a more hydrophilic (2,5-dimethylpyrazine) single components. Near-infrared (NIR) and low-field time-domain nuclear magnetic resonance (low field TD-NMR) data were acquired on each system and analyzed using multivariate chemometric techniques to develop optimal prediction models. Partial least squares regression models showed good predictive ability, with coefficients of determination (R²) between 0.81 and 1.00 and low root mean square error of cross-validation values compared to the range of concentrations. The predictive ability of the chemometric models computed using the NIR spectra improved significantly when data were pre-processed using multiplicative signal correction. The development of good prediction models (i.e., robust models resulting in accurate predictions for water and flavor content) from the NMR relaxation data spectra was successful only for the hydrophobic limonene systems, yielding prediction models whose performance was better than the models obtained using the NIR data. Overall, NIR spectroscopy and NMR relaxometry were identified as complementary techniques rather than competitive methods in the characterization of encapsulated flavor systems.