Direct quantification of the percentage of arabica in roasted and ground coffee blends of arabica/robusta by High Resolution Nuclear Magnetic Resonance spectroscopy (NMR), in solution and multivariate statistical analysis. Particularly, the metabolites content present in water extracts of coffee is analyzed and compared with NMR data of our database. This latter includes data concerning several roasted and ground arabica/robusta coffee blends from different geographical origin and with different roasting degrees, with arabica composition ranging between 0 and 100% in weight. A statistical regression protocol performed on NMR data allows to predict the percentage of arabica in roasted and ground coffee blends of unknown composition.
The proposed method allows to monitor different classes of chemical compounds within a single experiment, obtaining a metabolic profile of the roasted coffee blends analyzed. This represents an advantage in the prediction of the percentage of arabica compared to chromatographic techniques that evaluate the arabica/robusta composition by monitoring single compounds such as 16-O-methylcafestol (present in robusta in variable amount based on the geographical origin and in a small amount in arabica) or D-5-avenasterol (present in both arabica and robusta but in a larger amount in the latter) which could be added or subtracted. Furthermore, our database including coffee blends of different geographical origins and different roasting degrees, presents a natural heterogeneity that mediates metabolic differences attributable to both provenance and roasting conditions of coffee.
Italy