Prediction of Polymer Glass Transition Temperatures Using a General Quantitative Structure-Property Relationship Treatment

Quantitative structure-activity relationship (QSAR) correlations have been widely applied for biological activities over several decades.1-7 Also, many applications of quantitative structure-property relationships (QSPR) are known in analytical chemistry.3,8-14 For instance, we recently successfully used15 our CODESSA (ComprehensiVe Descriptors for Structural and Statistical Analysis) QSPR program16 to achieve the first comprehensive correlation of both GC retention times and response factors for a large and widely diversified set of organic compounds. The QSPR equations developed15 allow the prediction of each of these two quantities for unknown compounds with significant confidence. Considerable attention has been devoted to establishing quantitative structure-property relationships in polymers, as recently reviewed.17 As regards glass transition temperatures, the most successful approach for homopolymers has been that of Koehler-Hopfinger which relates Tg by multiple linear regression to the intramolecular flexibility represented by linear contributions of conformational entropies of the repeat units computed from torsional rotations about backbone and side-chain bonds and to intermolecular interactions represented by the dispersion, positive electrostatic, and negative electrostatic intermolecular energies.18,19 This model was recently simplified using the Triphos 5.2 Force field and for a set of 12 alkyl acrylates R2 of 0.902 was established with four descriptors [93MCMS261]. However, this method is not applicable to copolymers, and the scatter is greater for less closely related sets. Earlier work relied on group additive methods21,22 which are not applicable to predictions for polymers containing previously noninvestigated groups. The effect of molecular weight on Tg was rationalized by a “corresponding states” model.23 Another approach to the prediction of Tg for linear polymers, random copolymers, and cross-linked polymers based on the tabulated atomic incremental volumes has been published by Wiff, Altieri, and Goldfarb in preliminary form only.24 We now report that a QSPR treatment based on CODESSA methodology can successfully rationalize the glass transition temperature (Tg) values of a set of low and medium molecular weight homoand copolymers and that the derived multilinear dependencies can be used for the prediction of the Tg values for unknown polymers of similar structure.