High performance solvent vapor identification with a two sensor array using temperature cycling and pattern classification

Abstract

Temperature modulation of semiconductor gas sensors is a powerful strategy to improve selectivity and stability of gas sensor arrays in applications where different gases have to be identified [IEEE Sens. J. 1 (3) (2001) 207; Sens. Actuators B 40 (1997) 33; Sens. Actuators B 43 (1997) 45; Anal. Chim. Acta 361 (1998) 93; Anal. Chem. 68 (1996) 2067; Sens. Actuators B 33 (1996) 142]. A recent review can be found in [Sens. Actuators B 60 (1999) 35]. We present an array composed of two commercial metal oxide gas sensors allowing discrimination of six organic solvents over a wide concentration range from 2 to 200ppm in air. Temperature cycling reduces sensor baseline drift considerably over a test period of several months. Additional signal pre-processing suppresses the influence of humidity and leads to further drift reduction. The system comprises a hierarchical pattern classification evaluating shape features generated from the sensor response curve during temperature cycling, which are compared with other feature generation methods (like FFT and wavelet, see also [Sens. Actuators B 41 (1997) 105]). The classification requires comparatively little computing power and allows flexible adaptation to different operating environments, for example, to suppress false alarms from interfering gases.Reproducibility of sensor performance was checked using four systems with identical sensors. The same features can be used for the classification with all systems but the areas for classification have to be adapted to the individual sensors.