RESUMEN
The applicability of 1H NMR spectroscopy coupled with chemometric in the quality control of dark chocolate was investigated for the first time to detect cocoa-butter equivalents (CBEs) above the allowed limit by European regulation. Blends of chocolate-fats with CBEs in the range 0-50 % were prepared and analyzed by 1H NMR spectroscopy. Datasets composed of peaks' areas or spectral variables (fingerprinting) in glycerol region were tested for the creation of multivariate statistical models. Partial least-squares discriminant analysis (PLS-DA) and regression (PLS-R) methods were used to correctly identify the type of CBE and quantify its concentration respectively. The performances of the models created on the two datasets were evaluated in terms of chemometric indicators and compared. The robustness of models was investigated through the analysis of test sets and random permutation tests. Fingerprinting models revealed fruitful results in classifying and quantifying CBEs in blends demonstrating the applicability of NMR in chocolate quality control.
