Role of Dielectric Screening in Calculating Excited States of Solvated Azobenzene: A Benchmark Study Comparing Quantum Embedding and Polarizable Continuum Model for Representing the Solvent

The low energy excited states of the conformational isomers of solvatedazobenzene are calculated with several DFT methods accounting for the solute-solventinteraction implicitly with the polarizable continuum model or explicitly with subsystemDFT. For the latter, embedding potentials are calculated for 21 sampled snapshots of thesolvent molecules. First, wefind that accounting for the solvent implicitly or explicitly haslittle effect on the predicted cis-trans S1excitation energy gap. Second, wefind thatazobenzene'sS1cis and trans energies are accurate as long as a screened range-separatedhybrid exchange-correlation functional is employed. Finally, we also tested a simplifiedworkflow whereby a single, averaged, embedding potential is used. Unfortunately, wefindlarger deviations against the experiment for the simplified workflow. This highlights a basicflaw in the approach, where the time scale of solvent averaging is much longer than that ofthe solute's electronic polarization.

Automated summary

The low energy excited states of solvated azobenzene's conformational isomers were calculated using various DFT methods, considering the solute-solvent interaction implicitly or explicitly. The results show that the solvent has little effect on the predicted cis-trans S1 excitation energy gap, and the accuracy of azobenzene's S1 cis and trans energies relies on using a screened range-separated hybrid exchange-correlation functional. A simplified workflow with a single averaged embedding potential was also tested, but it led to larger deviations from experimental results, indicating a fundamental flaw in the approach where the solvent averaging time scale is much longer than the solute's electronic polarization time scale.