QSARs for Rate Constants of OH Reactions with Organic Pollutants: Algorithm, Application Domain, Validation, and Mechanistic Interpretation

摘要

The reaction of organic pollutants with ·OH in the atmosphere is of great concern because it is the primary process for their degradation and transformation in the daytime. The ·OH reaction rate constants (kOH) are of great importance to assessing the behavior and fate of organic pollutants in the troposphere. Literature data for logkOH of 722 organic chemicals were used to develop QSAR models for logkOH, employing 22 molecular structural descriptors and partial least squares (PLS) regression. For the established model, the leave-many-out cross-validated Q2cum=0.848, R2=0.857, and SE=0.431 log units, indicating good robustness and predictive capability of the model. The predictive capability was also evaluated by external validation. The application domain was defined on the basis of descriptor space and the structural domain comprised of compounds containing C, H, N, 0, S, F, Cl, Br, I, and Si atoms in various functional groups (>C=C<, -C≡C-, -OH, -CHO, -O-,>C=O, -COOH, -C≡N, -NH2, -NO2, -SH, -SO3H, -X, etc.). The model suggested that the main molecular structural factors governing kOH are molecular ability of donating electrons, the compactness of the molecules, molecular size and number of halogen atoms.