Qspr Study Of The Nature Of Some Typical Persistent Organic Pollutants (pcde And Pbde) Environment Allocation

This thesis introduced the concept of persistent organic pollutants (POPs), Quantitative Structure-Property Relationships (QSPR) for Pops, partial least squares (PLS) method and chemical descriptors.The polychlorinated diphenyl ethers (PCDEs) and polybrominated diphenyl ethers (PBDEs) were used to develop the QSPR models in the thesis. Total 18 theoretical molecular structural descriptors as predictor variables computed by PM3 Hamiltonian. And the regression was performed by the use of PLS method .The QSPR models were successfully developed on sub-cooled liquid vapor pressures (PL), noctanol/water partition coefficients (kow) and sub-cooled liquid water solubility (Sw, L) of PCDEs, sub-cooled liquid vapor pressures (PL) of PBDEs. And the models of quantitative relationships between structures, environmental temperatures and properties (QRSETPs) on octanol-air partition coefficient (KOA) were also successfully developed. By the use of the MNDO and AMI Hamiltonian, The QSPR models were successfully developed on (PL) of PCDEs. And the results were comparative with the models of PM3.The researches show that the Q2cum values were high, indicating a good predictive ability and robustness of the models. The models were tested as acceptable for prediction of PL, Kow and KOA by validation sets, and the consistency between observed and predicted values is good. So the developed models can used to estimate the relevant properties of PCDEs or PBDEs.In the model of PL, intermolecular dispersive interactions play a leading role in governing the values of PL, then comes electrostatic, dipole-dipole and dipole-induced dipole interactions. The cross-validated Q2cum value for the model is 0.993, indicating a good predictive ability and stability. Using the model, the values of PL of other were estimated.The models on log KOA of PBDE congeners were developed for predicting log KOA values of PBDEs at different temperatures. The optimal model contains nine theoretical molecular descriptors and 1/T as predictor variables. The cross-validated Q2cum value for the model is 0.975, indicating a good predictive ability and stability.The models identify intermolecular dispersive interactions as playing a key role in governing the magnitude of PI, KOW and Koa.By using of AMland MNDO Hamiltonians, QSPR models for logpl, logkow and logSw, L were obtained. The results were comparative with the model of PM3, indicating that the consistency between observed and predicted values of PM3 is the best, followed by AMland MNDO. Although while the results produces by different semi-empirical methods are not generally comparable, they do often reproduce similar trends. The models of three methods have good predictive ability and robustness and can be used for estimation.