Quantum Chemistry-ANN Methods Study on <i>T</i>g of Polyacrylates

Abstract

Themechanismandaffectingfactorsoftheglasstransitionforpolymershavebeenanalyzed.The structuralunitsofthirty鄄eightpolyacrylateshavebeenoptimizedandtheirquantumchemicaldescriptorshave beenobtainedbyDFT/6鄄31G(d) method.Thecalculatedresultsindicatethatthelengthofsidechain,numberofside chains,polarizability,dipolemoment, E HOMO , heatcapacityatconstantvolume,andthermalenergyarethemain factorsaffectingglasstransitiontemperature(T g). Theregularityof T g forpolyacrylatesarediscussedbythepattern recognitionmethod (PLS)withquantumchemicaldescriptorsasfeatures.Thetwoclassesofpolymerswith different T g distributeindifferentregions.Thequantitativerelationshiphavebeenstudiedbetweenthese descriptorsand T g bystepwiseregressionandBP鄄ANN(backpropagationartificialneuralnetwork)methods.The correlationcoefficientsbetweenthepredictedandexperimental T g forthetwomethodsare0.9753and0.9985,and thestandarddeviationsare18.42andresults4.25,respectively.