Use of an array of metal oxide sensors coupled with solid phase microextraction for characterisation of wines Study of the role of the carrier gas

This paper investigates the behaviour of metal oxide (MOX)-based sensors exposed to wines in the presence of different gas carrier backgrounds including air (oxygen), inert gases (nitrogen and helium) and mixtures of air/inert gas. The influences of the gas on the resistance of the sensors, and on the intensity, kinetics, repeatability and reproducibility of the responses have been analysed. The sensors exposed to nitrogen or helium show a higher conductivity due to the displacement of the oxygen adsorbed on the surface of the sensors. Signals registered in an oxygen background are faster than those observed in the presence of N-2 or He, indicating that it is easier to displace the oxygen than inert gases from the sensor surface. In addition, inert gases cause irreproducibility of the responses, particularly in the case of helium that decreases drastically the lifetime of the sensors. A purposely designed system of MOX connected to solid phase microextraction (SPME) has been developed that makes it possible for mixtures of air/inert gas to reach the sensor chamber instead of the pure inert gas. Using such mixtures, a drastic improvement of the kinetics and the reproducibility of the responses has been attained. The optimised system SPME-MOX has permitted to discriminate wines with similar characteristics such as red wines elaborated with the same variety of grape but aged using different types of oak woods. Moreover, the improvement of the stability and reproducibility of the signals has allowed monitoring the ageing of wines. The system has demonstrated its ability in discriminating and recognising among wines after 3 and 6 months of permanence in oak wood barrels. Both the calibration and the validation values obtained by using a partial least squares (PLS2) regression method indicate a good-quality model performance (slope near 1, offset near 0 and large correlation between sensors and categorised variables). (C) 2008 Elsevier B.V. All rights reserved.