Study On Structure-Property Relationship Of Critical Solution Temperature Of Binary Polymer Solutions

The phase behavior of binary polymer solutions such as lower critical solution temperature(LCST)and the upper critical solution temperature(UCST)has an important role in the development and design of polymer-related processes.Partially miscible polymer solutions usually have two solubility boundaries,the θ(LCST)and the θ(UCST),which can be used to realize the temperature usage limits for polymer processing in the form of solution.Although the experiment is an effectual method to obtain the critical solution temperature data of polymer solution,it is not only time and material consuming,but also very difficult to measure especially for extreme critical solution temperatures.In this case,establishing a mathematical model to estimate the critical solution temperatures of polymer solutions could be an important alternative method.Quantitative structure-property/activity relationship(QSPR/QSAR)method combines the molecular structure information of compounds with its property or activity data to establish a mathematical model,so as to predict the target property or activity of unknown compounds.In this work,based on the concept of norm descriptor proposed by our group,the critical properties of binary polymer solution are quantitatively studied by QSPR method,which is mainly divided into the following two aspects:Study on the structure-property relationship of binary polymer solution of θ(LCST):the topological LCST-QSPR model and spatial LCST-QSPR model are established by using 177 θ(LCST)data points.Based on the topological norm descriptors obtained from topological molecular structures,the topological LCST-QSPR model was established using multiple linear regression(MLR)method.The model can be established quickly in this process because there is no need for structural optimization.The squared correlation coefficients(R~2)of training set and test set were 0.9406 and 0.9379,respectively.Based on the topological norm descriptors,a series of quantum chemical norm descriptors obtained from stable optimized molecular structures were added to establish the spatial LCST-QSPR model with higher accuracy.The R~2 of the training set and test set of the model were 0.9423and 0.9392,respectively.The results of various validation methods and statistical parameters show that the LCST-QSPR models based on norm descriptors have good fitting effect and have good stability,prediction ability and application prospect.Study on the structure-property relationship of binary polymer solution of θ(UCST):the topological UCST-QSPR model and spatial UCST-QSPR model are established by using 109 θ(LCST)data points.The topological UCST-QSPR model was established based on the topological norm descriptors by MLR method.The R~2 of the training set and test set of this model were 0.8604 and 0.8785,respectively.After adding the quantum chemical norm descriptors,the spatial UCST-QSPR model with higher accuracy was established.The R~2 of the training set and test set were 0.8826 and 0.8869,respectively.The results show that the two UCST-QSPR models established by using norm descriptors have good stability and prediction performance,and the models with strong reliability and wide applicability.The topological and spatial QSPR models established in this work have good fitting effects,indicating that topological norm descriptors and quantum chemical norm descriptors are effective in predicting θ(UCST)and θ(LCST)of binary polymer solutions.And the addition of quantum chemical norm descriptors improves the accuracy of models to a certain extent.By multiple validation methods,it is found that the experimental and predicted values of the four models established in this study have a good correlation,which shows that the norm descriptors have good applicability and high accuracy in predicting the phase behavior data of binary polymer solutions.