PATTERN-RECOGNITION STUDIES OF HALOGENATED ORGANIC-COMPOUNDS USING CONDUCTING POLYMER SENSOR ARRAYS

Direct measurement of volatile and semivolatile halogenated organic co mpounds of environmental interest was carried out using arrays of cond ucting polymer sensors. Mathematical expressions of the sensor arrays using microscopic polymer network model is described. A classical, non parametric, and unsupervised technique of cluster analysis was used to discriminate between polychlorinated organic phenol vapor response ve ctors in 2-dimensional space and to identify clusters or groups to whi ch unknown vectors are likely to belong. The characteristic pattern fo r each sample was used to generate the database employed in the determ ination of the Euclidean distances between two given patterns and the normalized sensor response, as well as to develop a 2-dimensional mapp ing from a multidimensional space for quantifying the distinction of t he samples. Results obtained demonstrated that conducting polymer sens or arrays can be utilized in the identification and quantitation of ch lorinated organic phenols based on the differences in their Euclidean distances. The systematic differences, qualitatively defined by the Eu clidean difference measurements, were most clearly visible when the na ture and the function of the functional groups were considered.