Assessing the Accuracy of Density Functional Approximations for Predicting Hydrolysis Reaction Kinetics

Hydrolysis reactions are ubiquitous in biological, environmental, and industrial chemistry. Density functional theory (DFT) is commonly employed to study the kinetics and reaction mechanisms of hydrolysis processes. Here, we present a new dataset, Barrier Heights for HydrOlysis - 36 (BH2O-36), to enable the design of density functional approximations (DFA) and the rational selection of DFAs for applications in aqueous chemistry. BH2O-36 consists of 36 diverse organic and inorganic forward and reverse hydrolysis reactions with reference energy barriers calculated at the CCSD(T)/CBS level. Using BH2O-36, we evaluate 63 DFAs. In terms of mean absolute error (MAE) and mean relative absolute error (MRAE), wB97M-V is the best-performing DFA tested, while MN12-L-D3(BJ) is the best-performing pure (non-hybrid) DFA. Broadly, we find that range-separated hybrid DFAs are necessary to approach chemical accuracy (0.043 eV). Although the best-performing DFAs include a dispersion correction to account for long-range interactions, we find that dispersion corrections do not generally improve MAE or MRAE for this dataset.