A Robust SPME Method for the Analysis of Wine Volatiles based on Multiple Internal Standards and Multivariate Regression

The analytical signals obtained by SPME are extremely sensitive to experimental parameters affecting liquid-gas and gas-solid distribution coefficients, such as alcoholic degree, non-volatile composition or wine content in major fermentation volatiles. In order to measure the weight of these factors and to design a robust calibration, synthetic but real-like wines containing selected analytes, a cocktail with 13 different internal standards and variable amounts of ethanol, non-volatile constituents, and major volatile compounds were prepared. Data revealed that the wine content in major volatile constituents affected the extraction of most analytes, while ethanol level and matrix composition specifically affected the least volatile components. In 29 out of 65 cases, a single internal standard provided a robust calibration, while in 30 other cases a mathematical combination of the 13 internal standards obtained by PLSR provided maxima accuracy.