LAUSR

Transition metal complexes are a promising class of redox mediators for targeting redox flow batteries due to the tunability of their electrochemical potentials. However, reliable time-efficient tools for the prediction of their reduction potentials are needed. In this work, we establish a suitable density functional theory protocol for their prediction using an initial experimental data set of aqueous iron complexes with bidentate ligands. The approach is then cross-validated using different complexes found in the redox-flow literature. We find that the solvation model affects the prediction accuracy more than the functional or basis set. The smallest errors are obtained using the COSMO-RS solvation model (mean average error (MAE) = 0.24 V). With implicit solvation models, a general deviation from experimental results is observed. For a set of similar ligands, they can be corrected using simple linear regression (MAE = 0.051 V for the initial set of iron complexes).