Resumo:
The objective of this work was to control the purity level of canephora coffees (Conilon coffee) from the state of Espírito Santo (Brazil), as a function of the addition of grounds produced by the coffees themselves using a portable microspectrometer in the infrared region. (microNIR), associated with chemometric tools. The PCA model grouped the coffees according to the level of adulterant in the samples and showed that the presence of moisture and polysaccharides is important for the discrimination between pure and adulterated samples. PLS-DA and SIMCA classification models were built with commercial samples and resulted in the discrimination of authentic and non-authentic samples with accuracy values above 90% and 82% for PLS-DA and SIMCA, respectively. The quantification results presented by the PLS model were satisfactory for all models, reaching the best results at 75 °C with RMSEP values of 6.13 wt%, R2p = 0.9917; and LOD in the range of 5.6-8.9 wt%; LOQ = 17-26.8 wt%, according to IUPAC; relative standard deviation for repeatability of 1.0 and 1.7 wt%; and intermediate precision of 3.4 and 1.1 wt% for low (<10 wt%) and high (≈ 90 wt%) levels of adulterants, respectively. Therefore, this work represents the first exploratory study to analyze adulterations with coffee grounds by portable NIR, demonstrating the potential of the equipment against this type of adulteration.
