Development of a group contribution method to predict aqueous phase hydroxyl radical (HO*) reaction rate constants

The hydroxyl radical (HO*) is a strong oxidant that reacts with electron-rich sites of organic compounds and initiates complex chain mechanisms. In order to help understand the reaction mechanisms, a rule-based model was previously developed to predict the reaction pathways. For a kinetic model, there is a need to develop a rate constant estimator that predicts the rate constants for a variety of organic compounds. In this study, a group contribution method (GCM) is developed to predict the aqueous phase HO* rate constants for the following reaction mechanisms: (1) H-atom abstraction, (2) HO* addition to alkenes, (3) HO* addition to aromatic compounds, and (4) HO* interaction with sulfur (S)-, nitrogen (N)-, or phosphorus (P)-atom-containing compounds. The GCM hypothesizes that an observed experimental rate constant for a given organic compound is the combined rate of all elementary reactions involving HO*, which can be estimated using the Arrhenius activation energy, E(a), and temperature. Each E(a) for those elementary reactions can be comprised of two parts: (1) a base part that includes a reactive bond in each reaction mechanism and (2) contributions from its neighboring functional groups. The GCM includes 66 group rate constants and 80 group contribution factors, which characterize each HO* reaction mechanism with steric effects of the chemical structure groups and impacts of the neighboring functional groups, respectively. Literature-reported experimental HO* rate constants for 310 and 124 compounds were used for calibration and prediction, respectively. The genetic algorithms were used to determine the group rate constants and group contribution factors. The group contribution factors for H-atom abstraction and HO* addition to the aromatic compounds were found to linearly correlate with the Taft constants, sigma*, and electrophilic substituent parameters, sigma+, respectively. The best calibrations for 83% (257 rate constants) and predictions for 62% (77 rate constants) of the rate constants were within 0.5-2 times the experimental values. This accuracy may be acceptable for model predictions of the advanced oxidation processes (AOPs) performance, depending on how sensitive the model is to the rate constants.