Identification of baijiu flavor types and brands by NMR spectroscopy and multivariate statistical analysis

The classification of baijiu types is of great significance in distinguishing counterfeits. In this work, a robust and efficient classification model has been established to distinguish 3 major baijiu flavor types and three brands in strong flavor type by quantitatively nuclear magnetic resonance (NMR) coupled with multivariate chemometric analysis. A total of 48 commercial baijiu samples were analyzed to represent those three flavor types. The ratios between hydrogen and deuterium isotope were quantified by NMR, in the CH₃- and -CH₂- groups of ethanol. The relative contents of nine key flavor-related compounds were determined by ¹H NMR at the same time. Initially, unsupervised principal component analysis, hierarchical cluster analysis and heatmap analysis were performed on the NMR dataset, proving preliminary clustering patterns among the baijiu samples. Subsequently, orthogonal partial least squares discriminant analysis was applied to build a classification model, to show clear differentiation among baijiu flavor types and to find key molecular markers for these distinctions. NMR spectroscopy, combined with multivariate statistical analysis, was demonstrated to be an effective method for baijiu flavor types identification and brand discrimination. The Variable Importance in Projection Plots show that the hydrogen and deuterium isotope ratios of ethanol the CH₃- and -CH₂- groups could be used to discriminate baijiu categories and brands. This method has great potential to identify other liquor and alcohol-containing drinks. It also could be broadly used as a classification and quality evaluation strategy in complex food and beverage systems.