Effectiveness of an E-Nose Based on Metal Oxide Semiconductor Sensors for Coffee Quality Assessment

Coffee quality, which ultimately is reflected in the beverage aroma, relies on several aspects requiring multiple approaches to check it, which can be expensive and/or time-consuming. Therefore, this study aimed to develop and calibrate an electronic nose (e-nose) coupled with chemometrics to approach coffee-related quality tasks. Twelve different metal oxide sensors were employed in the e-nose construction. The tasks were (i) the separation of Coffea arabica and Coffea canephora species, (ii) the distinction between roasting profiles (light, medium, and dark), and (iii) the separation of expired and non-expired coffees. Exploratory analysis with principal component analysis (PCA) pointed to a fair grouping of the tested samples according to their specification, indicating the potential of the volatiles in grouping the samples. Moreover, a supervised classification employing soft independent modeling of class analogies (SIMCA), partial least squares discriminant analysis (PLS-DA), and least squares support vector machine (LS-SVM) led to great results with accuracy above 90% for every task. The performance of each model varies with the specific task, except for the LS-SVM models, which presented a perfect classification for all tasks. Therefore, combining the e-nose with distinct classification models could be used for multiple-purpose classification tasks for producers as a low-cost, rapid, and effective alternative for quality assurance.