Regulation-based rapid and accurate determination of ethanol content in alcoholic beverages using Near-InfraRed Spectroscopy and data processing tools

The beverage industry is rapidly expanding, and consequently, fast and accurate quality control methods ensure production process efficiency and compliance with national regulations. This study introduces, for the first time, the construction of a model that offers performance comparable to the traditional HPLC technique for various types of alcoholic beverages, serving as the reference standard test method concerning national regulations using a combination of Near-InfraRed Spectroscopy (NIRS) with Partial Least Squares Regression (PLS) determining the ethanol content. The developed PLS models demonstrated robust performance, and optimal results were achieved using the full spectral range, coupled with standard normal variate (SNV) transformation and the second derivative of Savitzky-Golay smoothing, achieving prediction biases of -0.0056% and 0.0124% alc/vol for Mahalanobis and wavelength selection methods, respectively. Given its high accuracy and rapid analysis time, the NIRS-PLS methodology presents a compelling alternative to HPLC for routine quality control in the alcoholic beverage industry. • NIRS combined with chemometrics allows accurate ethanol content detection • Results follow the Canadian Association of Liquor Jurisdictions • The multi-product model analyzed 101 unique commercial liquors ans spirits • Different possibilities to create a prediction model using NIRS were discussed • NIRS is suitable for complex systems and implementation in routine QC for beverages