Development of a model for predicting hydroxyl radical reaction rate constants of organic chemicals at different temperatures

- Two MLR models were developed for predicting OH reaction rate constants (kOH).
- Temperature-dependent model can be used to predict kOH at different temperatures.
- A model is specially developed for predicting kOH at the room temperature (298 K).
- The two models cover a larger dataset and have broader applicability domains.

The reaction rate constants of hydroxyl radicals with organic chemicals (kOH) are of great importance for assessing the persistence and fate of organic pollutants in the atmosphere. However, experimental determination of kOH seems fairly unrealistic, due to the soaring number of the emerging chemicals additional to the large number of existing chemicals. Quantitative structure-activity relationship (QSAR) models are excellent choices for evaluating and predicting kOH values. In this study, a QSAR model that can predict kOH at different temperatures was developed by employing quantum chemical descriptors and DRAGON descriptors. The adjusted determination coefficient Radj2 of the model is 0.873, and the external validation coefficient Qext2 is 0.835, implying that the model has satisfactory robustness and good predictability. Additionally, a QSAR model was also built for kOH prediction at room-temperature (298 K). The development of the two models followed the guidelines for development and validation of QSAR models proposed by the Organization for Economic Co-operation and Development (OECD). The applicability domains of the current models were extended to several classes of compounds including long-chain alkenes (C8C13), organophosphates, dimethylnaphthalenes, organic selenium and organic mercury compounds that have not been covered in the previous studies.

242