Quantitative structure-property relationship modeling of phosphoric polyester char formation

A required characteristic for almost all commercial polymers is to be flame retardant. This can be achieved if they have intrinsic flame-retardant behavior, or by incorporating in them flame-retardant materials. Polyphopshpoesters can be used as flame-retardant materials for other polymers. In the present work, the multiple linear regression (MLR) technique was used for a quantitative structure-property relationship study of char yields of 32 polyphosphonates and polyphosphates. The polyphosphoesters were modeled by their mer units, which were pre-optimized using the MMFF94 force field. The molecular descriptors derived from the optimized structures were calculated for the minimum energy conformers, using the InstantJChem and Dragon programs. The polymer char yield was related to the structural descriptors, using MLR calculations, which were combined with a genetic algorithm for variable selection. Several stable MLR models were obtained, which were externally validated using seven compounds, as a test set. Model equations emphasize the important influence of structural descriptors to the polymer charring behavior. The best-developed regression models contain Randic molecular profiles, Geometry, Topology and Atom-Weights Assembly, and 3D Molecule Representation of Structures based on Electron diffraction descriptors, which influence positively the char formation. These models were used for the prediction of potential char formation of nine polyphosphonates and polyphosphates.