ABSTRACT
Besides selectivity and sensitivity constraints, sensor drift constitutes the main limitation of current gas sensor array instruments [I]. In order to improve baseline stability, it is customary to expose the sensor array to a sequence of wash and reference gases prior to sampling a target d o u r . Since washlreference (WR) gases can be assumed to remain stable over time, variations in sensor response during these stages may be utilized, at no extra cost, as measures of the drift associated with each experiment and, therefore, compensated for in software.
