News

European legislation permits the inclusion of non-cocoa vegetable fats up to 5% into milk chocolate recipes.  Anything beyond this is adulteration.

In this paper (purchase required) the authors report the use of proton NMR combined with chemometrics to discriminate between milk fats, cocoa fats and non-cocoa vegetable fats (“cocoa butter equivalents”, CBE).  They prepared known mixes (0-100%) of different fats.  They used both a targeted and an untargeted approach.  The targeted approach used the integrals of the signals belonging to ω-3, ω-6, ω-9, and saturated fatty acids.  The untargeted approach used the spectra as fingerprints.

The authors reported that the untargeted partial least-squares discriminant analysis model (PLS-DA) recognized the type of CBE in blends with sensitivities in prediction higher than 75%. The targeted PLS-DA model was capable of recognizing non-adulterated milk chocolate fats with 100% sensitivity and specificity in prediction. Conversely, low percentages in sensitivity were achieved for most of CBEs. Both targeted and untargeted PLS regression models efficiently determined the amount of CBE in blends. Fingerprinting models showed better results both in the classification and quantification of CBEs.  They conclude that this proves the applicability of ¹H NMR in milk chocolate quality control.

Photo by Kaffee Meister on Unsplash