不同工艺酱香型白酒挥发性物质差异分析

Differences in Volatile Compounds in Maotai-Flavored Baijiu Produced by Different Processes

利用气相色谱及气相色谱-质谱联用技术对两种工艺酱香白酒中挥发性物质进行定性定量分析,结合香气活性值和化学计量学方法构建鉴别模型并找出关键差异物质。结果表明:碎沙酒样中各类挥发性物质的质量浓度均低于稛沙酒样,但其酯类、酸类和芳香族化合物占其所有挥发性物质的相对含量高于稛沙酒样,说明白酒品质可能与各物质配比相关。以4/5酒样为训练集构建鉴别模型,1/5酒样为测试集验证模型,测试正确率为93.33%,可以基本实现两种工艺酒样的鉴别。9种不同比例混合酒样在模型中的分布情况表明,该模型还可用于混合酒样的初步识别,验证了该模型的适用性。同时筛选出14个潜在差异物质,并利用聚类热图进一步证实这14种物质对区分两种不同工艺酒样的有效性。其中,糠醛的变量投影重要性的得分值最高,同时糠醛在两种酒样中的香气活性值差异也较大,说明糠醛可能是造成稛沙和碎沙酒样之间差异的重要物质。

Gas chromatography(GC)and gas chromatography-mass spectrography(GC-MS)were used to identify and quantitate the volatile substances in Maotai-flavoured baijiu produced by two different processes.A discriminant model was established based on odor activity value(OAV)by chemometrics and the key differential substances were identified.The results showed that the contents of volatile compounds in baijiu made from ground sorghum were lower than those in baijiu made from whole sorghum,but the proportions of esters,acids and aromatic compounds were higher,indicating that baijiu quality may be related to the proportion of each compound.Baijiu blends at different ratios were prepared to validate the applicability of the proposed model,four-fifths of which were used as training set,and the remaining one-fifth were used as testing set.The overall accuracy rate was 93.33%,indicating that the model could discriminate the two baijiu samples.The distribution of 9 different proportions of mixed samples in the model showed that it could also be used for preliminary identification of mixed samples,which verified the applicability of the model.In addition,14 potential differential compounds were identified,and their effectiveness in discriminating the two processes was confirmed by the clustering heatmap.Among these compounds,the variable importance in the projection(VIP)value of furfural was the highest,and the OAV of furfural differed greatly between the two baijiu samples,indicating that furfural may be an important substance causing the difference between the baijiu samples.