Abstract: Polychlorinated diphenyl ethers (PCDEs) have received more and more
concerns as a category of potential persistent organic pollutants (POPs).
Modeling its environmental fate and exposure assessment require a number of
fundamental physico-chemical properties. However, the experimental data are
currently limited due to the difficulty in analysis caused by the complexity of
PCDE congeners. As an alternative, the quantitative structure property
relationship (QSPR) approach could be used. In this paper, twelve kinds of
molecular connectivity indices (MCIs) of all 209 possible molecular structure
patterns of PCDEs were calculated. Based on 106 PCDEs with three observed
physico-chemical properties-vapour pressure (P^o_L), aqueous
solubility (S_w) and n-octanol/water
(K_{ow}) and their MCIs data, a series of QSPR equations
were established using multiple linear regression (MLR) method. As a result,
three equations with best performance were selected mainly from the view of
high regression coefficient (R)and low standard error (SE). All of them showed
significant relationship and high accuracy. With these equations the properties
of other 103 patterns of PCDEs without the reported observed values were
predicted. Further more,three partition properties for PCDE congeners-Henry's
Law constants (H), partition coefficients between gas/water
(K_{gw}) and gas/n-octanol (K_{go}) were
calculated according to the internal relationship among these six
properties. These observed and predicted values, in contrast with the criteria
listed in the Stockholm treaty about POPs which has been signed by more than
ninety countries in May 2001, illustrated that most of PCDEs congeners are
potential persistent organic pollutants. As all descriptors/predictors are
derived just from the molecular structure itself and without the import of any
empirical parameters, this method is impersonal and promising for the
estimation of physico-chemical properties of PCDEs.
Keywords: PCDEs, POPs, QSPR, molecular connectivity indices
