Predicting rate constants of hydroxyl radical reactions with organic pollutants: Algorithm, validation, applicability domain, and mechanistic interpretation

The reaction with hydroxyl radical (OH) is the most important removal process in the daytime for organic pollutants in the atmosphere, thus the OH reaction rate constants (kOH) are important to assessing the fate of organic pollutants in the troposphere. In this study, experimental data for log kOH of 722 organic chemicals were employed to develop quantitative structure–activity relationships (QSARs) for kOH, applying 22 molecular structural descriptors and partial least squares (PLS) regression. The QSAR development followed the OECD guidelines, with special attention to validation, applicability domain and mechanistic interpretation. For the established model, the leave-many-out cross-validated QCUM2 = 0.865, R2 = 0.878, and RMSE = 0.391 log units, indicating good robustness and predictivity. The predictive capability was also evaluated by external validation with QEXT2 = 0.872. The applicability domain of the model is composed of compounds containing C, H, N, O, S, F, Cl, Br, I, and Si atoms in various functional groups and analyzed by Williams plot. The main molecular structural factors governing kOH are the compactness of the molecule, the molecular ability of donating electrons and the number of halogen atoms in a molecule.