QSAR modeling for reaction rate constants of eaq− with diverse organic compounds in water

Abstract

The hydrated electron (e_aq⁻)-based advanced reduction processes (ARPs) are recognized as a reliable and feasible approach for decontamination of water pollutants. The reaction rate constant (k_eaq⁻) of e_aq⁻ with organic pollutants is a crucial parameter for assessing their removal efficacy and designing the ARPs. In this work, quantitative structure–activity relationship (QSAR) models for predicting the k_eaq⁻ values of structurally diverse compounds were developed using both multiple linear regression (MLR) and support vector machine (SVM) methods. The developed models covered a larger dataset (>600 compounds) and exhibited wider applicability domains than existing models, as well as possessing adequate goodness-of-fit (MLR: R_tr² = 0.712; SVM: R_tr² = 0.805) and predictability (MLR: R_ext² = 0.816, Q_ext² = 0.724; SVM: R_ext² = 0.830, Q_ext² = 0.769). Mechanistic analysis revealed that the reactivity of compounds towards e_aq⁻ was highly related to energy of the lowest unoccupied molecular orbital (E_LUMO), percentage of halogen atoms (X%), the most negative charge on Br atom (q_min,Br) and hardness (η). Additionally, the functional group analysis indicated that the compounds with electron withdrawing groups tended to have increased reactivity towards e_aq⁻ compared to those with electron donating groups. The developed QSAR models may serve as a valuable strategy for assessing the removal efficiency and understanding the fate of various pollutants in e_aq⁻-based ARPs.