STUDY OF THE CHEMICAL PROFILE OF BRAZILIAN BEERS: AN ASSESSMENT BETWEEN ARTISANAL AND INDUSTRIAL DRINKS

Beer is commonly produced with four inputs: water, malt, hops, yeast and, in some cases, brewing adjuncts. Industrial and artisanal processes essentially follow the same steps and may undergo minor changes due to the style of beer. Despite this, many consumers attribute higher quality to craft beer compared to industrial ones. In this study, we applied the analytical techniques of infrared (IR), hydrogen nuclear magnetic resonance (¹H NMR) and mass spectrometry (ESI(±)FT-ICR MS) to discriminate craft beers from industrial ones (brands common to the Brazilian market). In addition, Partial Least Squares (PLS) and Support Vector Machine (SVR) regression were used to estimate some beer properties (pH, total acidity (meq·L^-1, color (EBC), bitterness (IBU), alcohol content (%v/v) and density (g·mL^-1)). The use of the IR allowed the identification of vibrations attributed to chemical compounds common to beverages, such as water (3320 and 1640 cm^-1), carbohydrates (1500 cm^-1) and ethanol (1050-1040 and 875-865 cm^-1). ¹H NMR showed good applicability in identifying classes of organic compounds in the beverage, where signs attributed to alcohols, organic acids, carbohydrates and amino acids were observed. The ESI(+)FT-ICR MS allowed the identification of chemical compounds present in beverages, allowing the construction of a fingerprint of the beers. In addition, the application of chemometric tools enabled the prediction of physicochemical properties, presenting promising results in the prediction of alcohol content (RMSEC 0.2430 and RMSEP 0.3929) and bitterness (RMSEC 1.3022 and RMSEP 1.6008), and also in the classification regarding the manufacturing process (craft and industrial beer).

Keywords:
chemometrics; regression; PLS; SVR; PCA