Methodology for the selection of suitable sensors for incorporation into a gas sensor array

An investigation into suitable mathematical techniques which can be used to select sensor components for a gas sensor array is reported. Data from a tin dioxide Taguchi semiconductor sensor array were obtained individually for various organic solvents and analysed using multivariate techniques, including principal component analysis, cluster analysis and star symbol plots. It was shown that the array data produced a series of characteristic response patterns for the analytes. It was also found that analytes of similar chemical nature had similar response patterns, indicating a correlation between sensor interaction and the chemical functional groups of the analyte. The multivariate techniques used proved to be very useful in enabling a suitable selection of the components of the array to be made by identifying which of the components or sensors were acting independently.