Abstract: In order to investigate the physicochemical properties of polybrominated diphenyl ethers, a topological distance feature location index (T) was proposed based on the characteristic value of molecules vertex atoms (group) (δ_i) and calculated by a new method. A series of quantum chemical parameters was obtained by the Gaussian 03 program and molecular electronegativity distance vector (M_t) was obtained by the program in literature. Three quantitative structure-retention/property relationship (QSRR/QSPR) models between relational descriptors and chromatographic retention time, octanol/air distribution coefficient (lgK_OA), and super cold fluid steam pressure at 298 K (lgP_L) of PBDEs were set up by best subsets regression method. The correlation coefficients (R) were 0.9953, 0.9983 and 0.9982, the cumulative cross-validation coefficients (Q) were 0.9948, 0.9975 and 0.9975, and the standard deviations (SD) were 0.0222, 0.0754 and 0.0868, respectively. The result illustrates that the established models are reliable and better than those in literature.

Key words: pollution, computer simulation, model, prediction, polybrominated diphenyl ethers, quantitative structure property relationship

摘要： 为研究多溴联苯醚（PBDEs）理化性质，提出其分子中顶点原子（基团）特征值（δ_i）新的计算方法，并在修正的距离矩阵基础上构建了拓扑距离特征定位指数（T）；采用文献程序软件和Gaussian 03程序软件分别计算，得到其分子电性距离矢量（M_t）和多种量子化学参数。用最佳变量子集回归方法建立了相关参数与PBDEs的色谱保留时间（RRT）、正辛醇/空气分配系数（lgK_OA）和298 K超冷流体蒸气压（lgP_L）的3个定量构效关系模型。模型复相关系数（R）分别为0.9953、0.9983和0.9982，逐一剔除法交叉验证相关系数（Q）分别为0.9948、0.9975和0.9975，标准偏差（SD）分别为0.0222、0.0754和0.0868，均优于前人的工作。研究结果表明，所建模型具有良好的稳健性和预测能力，拓扑距离特征定位指数对PBDEs性质的表征是有效的。

关键词: 污染, 计算机模拟, 模型, 预测, 多溴联苯醚, 定量结构性质相关