MicroNIR spectrometer and multivariate classification in coffee fraud: towards a portable and agile method to legitimize Brazilian Conilon with indication of origin

Resumo

Portable analysis methods are at the pinnacle of next generation measurement
tools and are seen as a functional issue for the beverage and food industry. In this work,
a portable spectrometer takes measurements of solid coffee samples in the near-infrared
region, from 906 to 1650 nanometers. The miniaturized spectrometer is powered by the
notebook and allows each sample to be evaluated instantaneously, providing a unique
chemical fingerprint in seconds. These spectral fingerprints are analyzed and interpreted
by chemometrics methods. This method development was motivated by the rise of
Conilon coffee in the specialty coffee scenario in the face of a new industry perception
and by the need that coffee producers and associations have to legitimize their coffee, that
in this case, is their own responsibility. Conilon from Espírito Santo was registered under
geographical indication (GI) in Brazil, with the production located predominantly in the
northern region of state. Other regions in the south of the state also produce Conilon,
which makes its sensory profile diverse. In a real case study with Brazilian Conilon coffee
with indication of origin, the spectrometer showed promising results. A set of 124 Conilon
samples from Espírito Santo State with indication of origin and another of 75 Conilon
coffees from different regions and named in the study as “samples without geographical
indication” were analyzed. The chemometric method named Partial Least SquaresDiscriminant Analysis (PLS-DA) was used to build a multivariate classification model.
The result of the final predictive model showed 100% sensitivity and specificity, which
means that no sample was misclassified. To conclude the work, the spectral variables that
most influence of differentiation between GI Conilon and non-GI Conilon were
interpreted, allowing say that they differ in terms of carbohydrates, chlorogenic acids,
lipids, caffeine, and trigonelline contents. The method could be a quality control tool for
the inspection of fraud with respect to the certification of origin of Conilon coffee with
this specification.