OPTIMAL SELECTION OF GAS-CHROMATOGRAPHIC COLUMNS FOR THE ANALYTICAL CONTROL OF CHEMICAL WARFARE AGENTS BY APPLICATION OF INFORMATION-THEORYTO RETENTION DATA

Retention data can be used to characterize and identify the components of a mixture. The identification power of a single column is rather l imited, however. With several columns of different retention character istics, a high identification power can be achieved. In order to apply chromatographic retention data for the identification of chemical spe cies it is therefore important to select columns with low correlation of their retention characteristics. The retention behavior of five gas chromatographic columns was compared in pairs by linear regression of the retention data of a test set of compounds. The correlation coeffi cient of the two sets of retention data is an estimate of the degree o f correlation of the columns. It was found that five pairs of columns had correlation coefficients of less than 0.900, the lowest value bein g 0.769. A more exact approach for the selection of columns for the ch romatographic identification of compounds is based on the application of information theory. The information content of retention data (in b its) can be calculated from their probability distribution function. T he information content was calculated for the same five gas chromatogr aphic columns with the same retention data used in the calculation of the correlation coefficients. The optimal choice with respect to infor mation power and effort is a combination of three columns having an in formation content of 19.8 bits. The loss of information caused by the correlation of the column retention amounts -2.7 bits in this case. Th e method of column selection was applied for the development of a gas chromatographic screening method for the identification of chemical wa rfare agents, precursors, and decomposition products. A test set of 16 chemically different species of such compounds was used.