Abstract
The determination of the origin of coffee beans by NMR fingerprinting has been shown promising and classification has been reported for samples of different countries and continents. Here we show that this technique can be extended and applied to discriminate coffee samples from one country against all others, including its closest neighbors. Very high classification rates are reported using a large number of spectra (>300) acquired over a two-year period. As original aspects it can be highlighted that this study was performed in fully automatic mode and with non-deuterated coffee extracts. This is achieved using a series of experiments to procure a robust suppression of the solvent peaks. As is, the method represents a cost effective opportunity for countries to protect their national productions. (C) 2014 Elsevier Ltd. All rights reserved.
