【摘要】 根据MOPAC2009软件包中PM6算法得到的分子描述符研究半透膜仪（SPMDs）采集大气中多氯联苯（PCBs）采样速度(R_air)的定量构效关系（QSPR）模型,并分析影响R_air的关键因素。以半经验PM6算法得到的分子量子化学描述符作为预测变量,采用偏最小二乘算法（PLS）构建了R_air的QSPR模型。根据交叉验证,所得到的最佳模型中PLS成分解释的因变量的累积变异(Q²_cum)为0.683,这表明该模型具有良好的预测能力和稳健性。通过外部验证和将实验测得的R_air与预测得到的R_air进行比较,对所构建模型的稳定性和可靠性进行了验证,结果表明无论是训练组还是预测组,其预测值与实测值间均具有较好的线性关系,线性相关系数均大于0.8376。对PCBs采样速度R_air的主要影响因素为PCBs与SPMDs中甘油三油酸酯分子间的相互作用大小和为将PCBs溶解在甘油三油酸酯中形成洞穴所需能量要求。更多还原

【Abstract】 The purpose of this paper is to develop Quantitative structure-property relationship（QSPR） models for air sampling rates(R_air) of polychlorinated biphenyls（PCBs） with semipermeable membrane devices（SPMDs） based on molecular descriptors derived from the PM6 method in MOPAC2009 software package,and analyze the key factors influencing the R_air.Partial least square（PLS） regression was used to develope QSPR model for R_air.Quantum chemical descriptors computed by semi-empirical PM6 method were used as predictor variables.The cumulative variance of the dependent variable explained by the PLS components and determined by cross-validation(Q²_cum),for the optimal model is 0.683,indicating that the model has good predictive ability and robustness, and could be used to estimate R_air values.The stability and validity of the model was also tested by external validation testing and by comparing calculated R_air with experimental data.The linear correlation coefficients between predicted and experimental values were all larger than 0.8376 for both training set and prediction set.The main factors governing R_air values of PCBs are intermolecular interactions between PCBs and triolein in SPMDs and the energy required for cave-forming in dissolution of PCBs into triolein of SPMDs.更多还原