Redox Potentials with COSMO-RS: Systematic Benchmarking with Different Databases

Recent techniques of computational electrochemistry can yield redox potentials with accuracy as good as 0.1 V. Yet, many such methods are not universal, easy to use, or computationally efficient. Herein, we provide a systematic benchmarking of a relatively cheap and straightforward computational approach for fairly accurate computations of redox potentials. It is based on a combination of the conductor-like screening model for real solvents (COSMO-RS) and the density functional theory (DFT). The benchmarking is done with databases covering diverse redox systems, including transition-metal aquacomplexes and various organic and inorganic compounds. In addition, we also present our own test set aiming at maximum chemical diversity and maximum range of redox potential values. We assess the performance of the fairly efficient computational protocol combining the COSMO-RS with the BP86 DFT functional. This is done by calibrating it against different high-level state-of-the-art techniques, in particular, polarizable continuum model (PCM) connected to composite G3(MP2,CC)(+) method, domain-based pair natural orbital implementation of coupled cluster theory, or complete basis set CBS-QB3 method. We also elaborate on the absolute reduction potential value of standard hydrogen electrode to be used with COSMO-RS, and we propose the value of approx. 4.4 V. The COSMO-RS/BP86-D3(BJ) combination outperforms other methods on a wide range of redox systems. However, we show that its accuracy is based on a balanced error cancelation and, therefore, it cannot be further systematically improved. As a result, the proposed procedure represents a pragmatic choice for large-scale screening, yet it could be combined with more advanced methods for detailed studies.