COMPUTER-ASSISTED STUDY OF THE RELATIONSHIP BETWEEN MOLECULAR-STRUCTURE AND SURFACE-TENSION OF ORGANIC-COMPOUNDS

Computer-assisted methods are applied to the study of the relationship between molecular structure and observed surface tension of small organic alkanes, alkyl esters, and alkyl alcohols. Features of these molecules are encoded using a wide variety of topologic, geometric, and electronic descriptors. The simple correlation between these descriptors and observed surface tension values is examined to gain insight as to which molecular features most influence the observed surface tension. Multivariate linear regression models for each functional group class are also examined. The results of the examination of both the simple correlations and the regression models suggest that molecular surface area is an important feature. The results also show that many descriptors provide surface area information which is specific to particular portions of the molecule, and that this information provides better results in modeling surface tension than the van der Waals or solvent-accessible surface area. Finally, a multiple linear regression model is developed for a combined set of alkanes, alkyl esters, and alkyl alcohols which yields good results for predicting the surface tension of similar compounds.