Qspr Study On Part Of The Polymers

With the rapid development of science and technology, the polymersplay a most important role in our daily life. It also is important to dosome tests at this level. However, it is unrealistic to measure the all databy the experiments. Therefore, it is necessary to establish thequantitative structure-property relationship (QSPR) for chemists,materials scientists, and environmental scientists. QSPR is verypractical value and guiding significance, especially for unknownpolymers.There were some novel molecular structure parameterizationmethods, which had been used to describe the polymers of chemicalstructures. Here some QSPR models were established by multiple linearregression (MLR). At the same time, both good prediction capabilitiesand favorable estimation stability were made sure by internal andexternal checking. And the results showed that these novel descriptorscould preferably descript information of molecular structure in ourresearch.(1) The molecular electronegativity-interaction vector (MEIV) wasused to describe the chemical structures of121polymers. To study themolecular structure in forecasting of the refractive index ofunconjugated polymers, the MLR models were created with variablescreening by the stepwise multiple regression (SMR) technique andstatistics. Then the models examined by both internal and externalvalidation on their stability. The correlation coefficients of the builtmodels, leave-one-out cross-validation (LOOCV) and predicted valuesversus experimental ones of external samples are0.916,0.884and0.923. The satisfactory results showed that information related to theactivity of the unconjugated polymers could preferably be expressed byMEIV.(3) The study on QSPR of polyacrylic acid ester would be important in researching for polymeric compounds. MEIV was used to describethe chemical structures of22polyacrylic acid esters. To study themolecular structure in forecasting of the glass transition temperature (Tg)of polyacrylic acid esters, various MLR models were created withvariable screening by the SMR technique and statistics. Then themodels were examined by both internal and external validation on theirstability. The correlation coefficients of the built models, LOOCV andpredicted values versus experimental ones of external samples are0.982,0.971and0.922. The satisfactory results showed that informationrelated to activity of polyacrylic acid esters could preferably beexpressed by MEIV.(4) Quantitative structure property relationships were systematicallystudied for glass transition temperature simulation of polyacrylate. Anewly developed descriptor-generalized correlative index (GCI) wasused for the expression of chemical structure of polyacrylate. Amultiple linear regression model was built after screening someinsignificant parameters with the stepwise multiple regressiontechnique. The correlation coefficient (Rcum) and cross-validatedcorrelation coefficient (RCV) were0.991and0.985. Furthermore, thesuperior performance of the MLR model was tested by the external set,the correlation coefficient is Qext=0.920. In conclusion, the generalizedcorrelative index descriptor can be used for estimating and predictingglass transition temperature of polyacrylate.(5) A study on structural parameter, topological descriptors, wasused to describe the chemical structure of the part of the typicalpolymer. After the structural characterization, the descriptors obtainedwere screened by MLR. The limit characteristic ration multiplecorrelation coefficient (Rcum), leave-one-out cross-validated (RCV) andpredicted values versus experimental ones of external samples (Qext)were0.960,0.930and0.949. The result showed that the model hadfavorable stability and good prediction capabilities, topologicaldescriptors could be used for estimating and predicting limitcharacteristic of the flexible part of the polymers (6) A study on structural parameter,3D Holographic Vector of AtomicInteraction Field (3D-HoVAIF), was used to describe the chemicalstructure of the part of the typical polymer. After the structuralcharacterization, the descriptors obtained were screened by MLR. Thelimit characteristic rations of multiple correlation coefficient (Rcum),leave-one-out cross-validated (RCV) and predicted values versusexperimental ones of external samples (Qext) were0.948,0.889and0.790. The result showed that the model had favorable stability andgood prediction capabilities,3D-HoVAIF descriptor could be used forestimating and predicting Q-e parameters of ponomers in free-radicalcopolymerizations.