Semi-automated computational investigation of the oxidative degradation mechanisms of bisphenol A in Fenton-type processes

Abstract

Bisphenol A (BPA) is a widespread industrial contaminant and endocrine disruptor whose efficient removal remains challenging because multiple, competing radical channels operate under Fenton-type advanced oxidation conditions. Here, we present a semi-automated first-principles approach to elucidate this process, comprising: (i) exhaustive discovery of unimolecular radical transformations with AutoMeKin; (ii) targeted manual construction of initial •OH addition and hydrogen-abstraction transition states; (iii) DFT refinement at ωB97XD/def2-TZVPP with SMD solvation model; (iv) selective microsolvation (up to two water molecules) for high potential energy barriers; (v) transition state theory rate constants evaluation for all elementary steps, and unified statistical treatment of dual bottlenecks for bimolecular •OH reactions; and (vi) Kinetic Monte Carlo (KMC) simulations with Pilgrim to obtain product distributions. The reaction network maps all feasible early •OH additions (ipso/ortho/meta/para), phenolic O–H abstraction, multistep hydroxylations, attempted dehydration steps, epoxidation, ring opening, and C–C scission leading to hydroxylated, quinonoid, lactone, and cleavage products. Selective microsolvation lowers critical rearrangement barriers, converting otherwise rate-determining steps into kinetically viable channels. KMC analysis identifies a characteristic ≈2:1 [•OH]:[BPA] threshold. Below it, early hydroxylated and ketone intermediates persist (e.g., catecholic and cyclohexadienone forms), whereas above it they are rapidly converted into trihydroxylated derivatives, ring-cleavage fragments, and quinone products. A reduced mechanism derived from sensitivity analysis reproduces the kinetics of the full network while retaining only essential OH-addition and phenolic H-abstraction steps. This integrated workflow thus provides mechanistic insight and a predictive, computationally efficient kinetic model readily transferable to other organic contaminants in advanced oxidation processes.