估算多种有机物的有机碳-水分配系数线性自由能模型

Linear Free Energy Relationships for Evaluating Organic Carbon-water Distribution Coefficients of Various Kinds of Organic Compounds

准确估算有机污染物的有机碳-水分配系数(Koc)对评价其环境归宿具有重要意义。广泛收集了文献中烷烃、醇类、胺类、苯酚类、卤代苯、多环芳烃及农药等十多类有机化合物(共214种)的实测lgKoc数据与分子结构参数(E、S、A、B、V),将数据集随机分为建模组与验证组。对建模组171种化合物采用多元线性回归方法建立了估算lgKoc的线性自由能模型(LFER)。模型的内部检验结果表明:预测值与实测值残差呈正态分布,留多法交叉验证相关系数为0.95,均方根误差为0.29,模型的显著性和稳定性较好。利用验证组43种化合物对模型进行外部验证,验证组化合物的实测值与预测值标准误差为0.26,证实了模型预测的准确性。最后,基于LFER模型,从分子间作用力的角度分析了有机物的吸附机理。空穴作用和色散作用是疏水性化合物被吸附的主要驱动力,空穴作用和氢键作用在极性化合物分配中起主要作用。

Organic carbon-water distribution coefficient （Koc） is significant for assessing the fate of chemicals in the environment. The data on molecular structure parameters （E, S, A, B, V） and Koc of 214 organic pollutants from peer-reviewed literature were carefully collected. These compounds can be classified into a dozen groups including alkanes, alcohols, amines, phenols, halogenated benzene, polycyclic aromatic hydrocarbons and pesticide and so on. These compounds were randomly divided into calibration group and validation group. The linear free energy relationship （LFER） equation of calibration group including 171 compounds was established by multiple linear regression method. The inter-test of the model showed that the predicted and the measured values followed a normal distribution. The leave-many out cross-validation correlation coefficient was 0. 95, the root mean square error was 0. 29. This indicates that the model is robust. The experimental Koc values of 43 compounds in the validation group were compared with the predicted values with LFER model in this work to assess its apphcability as exter-valida-tion, the standard error is 0.26. The model can also provide insight for understanding the sorption mechanisms of compounds in terms of molecular interaction with solutes. Cavity effect and dispersion effect were the dominant sorption mechanism for the hydrophobic compounds, H-bond interaction and cavity effect played an important role in the sorption of polar compounds.