DISCRIMINATION OF LIQUOR AROMAS BY PATTERN-RECOGNITION ANALYSIS OF RESPONSES FROM A GAS SENSOR ARRAY

A sensor array composed of six semiconductor gas sensors was applied to the discrimination of liquor aromas. A semi-automatic headspace concentrator utilizing porous polymers was used for pretreatment of sample aromas in order to remove excess amounts of ethanol and to standardize aroma introduction to the sensor array. The differences in response patterns for liquor samples were not so conspicuous because of the non-selectivity of the gas sensors. After normalizing the sensor responses to eliminate the effects of absolute magnitude, pattern recognition analysis was applied to the resulting six-dimensional data matrices. Cluster analysis succeeded in classifying eight liquors. Five spirits (cognac and four different brands of whisky) were correctly classified by both linear discriminant analysis and cluster analysis.