Coffee can be blended to create a variety of products to meet consumer's needs. In order to uncover the blending effect of coffee beans, we performed an experiment using principal component analysis (PCA). Twelve varieties of green beans were tested in 11 experimental groups, and the volatile compounds of the beans were analyzed. A total of 41 volatile compounds were identified. PCA was performed on 13 compounds that had a low odor threshold value or a high concentration among the identified compounds. PCA of total volatile compounds showed that principal component (PC) 1 and PC2 were extracted within 80% cumulative dispersion level. In PC1 and PC2, furfuryl alcohol and formic acid ethyl ester showed the greatest positive correlation coefficients among all the volatile compounds. The largest negative correlation coefficients in PC1 and PC2 were 4-hydroxy-2-butanone and 3-(ethylthio)propanal, respectively. Using PCA of the major volatile compounds in coffee, propanal and 1-methylpyrrole were found to have the largest positive correlation coefficients in PC1 and PC2, respectively. In the score plot of the major volatile components, 4 kinds of blended coffee were closely grouped, therefore showing similar aroma qualities. However, 5 kinds of other blended coffees showed a positive correlation with PC2. This is probably due to 3-(ethylthio)propanal acting as a specific value. The application of statistical methods to blended coffee allows for logical and systematic data analysis of data and may be used as a basis for quality evaluation.
Keywords: blended coffee; dynamic headspace analysis; espresso; principal component analysis; volatile compounds.