Determination of toxic gases based on the responses of a single electrocatalytic sensor and pattern recognition techniques

Abstract

A response from an electrocatalytic gas sensor contains fingerprint information about the type of gas and its concentration. As a result, a single gas sensor can be used for the determination of different gases. However, information about the type of gas and its concentration is hidden in the unique shape of the current–voltage response and it is quite difficult to explore. One of the ways to get precise information about the measured gas is to use multivariate data analysis and pattern recognition techniques. In this paper we present the results of an investigation based on a combination of the cyclic voltammetry measurement technique and chemometric analysis methods such as principal components analysis, linear discriminant analysis and the k-nearest neighbors algorithm classifier for distinguishing different types of toxic gases. The responses of the single electrocatalytic gas sensor to 20 ppm of ammonia, nitrogen dioxide, sulfur dioxide and various concentrations of hydrogen sulfide in balance with synthetic air have been measured and analyzed. The reduction of measurement points in the data set used for multivariate analysis was evaluated.