|
Organic carboxylic acid is an important chemical raw materials and intermediates , the complexation extraction is an important method for separation and purification of organic carboxylic acid dilute solution , the equilibrium constant of complexation extraction thermodynamic properties . In this paper, the atom type of electric topological index (ETSI) to describe the sub- effective characterization of the structure of organic molecules . 658 kinds of organic matter and its literature for the study of the nature of the solubility in water , its structure - the nature of the associated mathematical model to multiple linear regression (MLR) . The results show that The ETSI descriptor to fully distinguish the studied compounds ; model 14 descriptors as independent variables , the model estimated correlation coefficient γ and LOO cross-validation correlation coefficient q 0.9074,0.9019 . Trioctylamine (TOA) as the extraction agent, chloroform , octanol , toluene, and xylene as a diluent on the glycolic acid was extracted ; octanol as a diluent for the formic acid , acetic acid, propionic acid, butyric acid , lactic , acrylic acid , malonic acid , succinic acid, malic acid , tartaric acid , maleic acid , salicylic acid, and m- nitro benzoic acid, etc. carboxylic acid acetate. Discuss the various factors affecting the extraction . Extraction mechanism based on the assumption of the extraction process to create a mathematical model of linear and non-linear regression through the model to the experimental data , draw the complexation extraction equilibrium constant K the 11 , K < sub > 12 . The results show that , the hypothetical extraction mechanism with the experimental data in good condition , the model can be used to predict the extraction . The same time, this paper studied the relationship (QSAR) of the organic carboxylic acid complexation extraction equilibrium constant and its molecular structure , its multiple linear regression (MLR) model to the six kinds ETSI descriptors as independent variables , the model estimated related coefficient , and LOO cross-validated coefficient q respectively 0.9987,0.9721 . The results show that the model has a good performance estimates and projections .
|