The roasting process is one of the critical points to obtain a product of the highest quality with certain sensorial properties including the aroma of coffee. Samples of coffee beans were roasted at different thermal treatment intensities with the aim of obtaining aromatic compounds detected with an electronic device. Sensory analysis, volatile compound profiling, and electronic nose analysis were carried out. Through principal component analysis (95.8% of the total variance of the data was explained by PC1 and PC2) and partial least squares discriminant analysis (the sum of the diagonal elements gave a hit rate of 94%), it could be demonstrated that the E-nose is able to discriminate roasted coffee beans subjected to different thermal treatments. Aromatic profiling was carried out by a testing panel and volatile compounds (VOCs) for the discrimination of roasted coffee samples. Alcohols, aromatics, esters, ketones and furanone were found in higher proportions in samples at the lowest thermal treatment. The VOCs with positive attributes were 1-(4-nitrophenyl)-3-phenylamino-propenone, carboxylic acids, 2-methoxy-4-vinylphenol, and 2-phenylethyl alcohol, while the compounds with negative ones were 2-methyl-furan, 2,5-dimethyl-pyridine, 2-methyl-butanal, and 2-furfurylthiol. The PLS model allows for the quantification of the positive and negative aromas (RCV2 = 0.92) of roasted coffee by using the E-nose. Therefore, the E-nose, that is, an inexpensive and nondestructive instrument, could be a chemometric tool able to discriminate between different qualities of coffee during processing.
Keywords: coffee quality; electronic nose; roasted coffee beans; sensory analysis; volatile compounds.