Selection of molecular descriptors used in quantitative structure-gas chromatographic retention relationships: I. Application to alkylbenzenes and naphthalenes

The descriptors used in quantitative structure-gas Chromatographic retention relationships are chosen among the physico-chemical properties and calculable indices of the compounds studied. In order to overcome the problems related to this choice, a suitable classification of molecular descriptors would be beneficial. The classification in this study is based on the extent of correlation of the corresponding descriptor with the retention by linear modelling. The idea is illustrated with a set of more than 35 descriptors and gas chromatographic retention data for 16 alkylbenzenes and 13 naphthalenes chromatographed on three stationary phases with different polarities. Four groups, namely substantial, important, likely and specific descriptors, were defined. The descriptors from the substantial group correlate with the retention index of the compounds studied to a very high extent, above 0.99. Some of the descriptors from the next group, important, are suitable for modelling, if tuned with the descriptors from the specific group. None of the descriptors from the likely group is able to be combined either with substantial or specific descriptors in order to give an adequate model, despite the proved applicability of many of them in quantitative structure-activity relationships. Very good correlations between some indicator variables and quantum chemical indices were also established. Owing to the fact that the indicator variables (descriptors) can be easily designed and interpreted, their use instead of the more complicated quantum chemical indices is promising, thus making modelling with them well suited for practical chromatography.