On the performance of DFT/MRCI for singlet–triplet gaps and emission energies of thermally activated delayed fluorescence molecules

Abstract

This work investigates the performance of the density functional theory multireference configuration interaction (DFT/MRCI) method for the donor–acceptor and multi-resonance thermally activated delayed fluorescence (TADF) emitters of the recent STGABS27 benchmark set [L. Kunze, A. Hansen, S. Grimme and J.-M. Mewes, J. Phys. Chem. Lett., 2021, 12, 8470–8480]. Comparing the accurate experimental singlet–triplet energy gaps (ΔE_ST) and fluorescence energies (E_EM) to values computed with DFT/MRCI reveals a robust performance without large or systematic errors. Specifically in the vertical approximation without a solvation model, DFT/MRCI achieves mean absolute deviations (MADs) for singlet–triplet gaps and emission energies of 0.06 eV and 0.21 eV, respectively. Surprisingly, these values do not improve systematically when geometric relaxation and state-specific solvation effects are included. Apparently, part of these effects are absorbed in the parameterization of DFT/MRCI and attempting to include them explicitly via a ROKS+PCM reaction field leads to an imbalanced treatment. As a result, the simplest approach of running calculations in the vertical approximation in gas phase turns out to be the most accurate. Albeit less accurate and more computationally demanding than state-specific orbital-optimized DFT, DFT/MRCI has the advantage that all low-lying excited states are obtained in a single calculation, including transition properties between them. At the same time, the aforementioned performance for the ΔE_ST and E_Em values is achieved without molecule-specific or state-specific adjustments like optimal tuning that is often necessary for time-dependent DFT. Hence, we conclude that DFT/MRCI is particularly useful during the initial stage of computational investigations of TADF emitters to screen for the ΔE_ST and identify the relevant states, whose energies can then be refined with accurate state-specific DFT methods like ROKS or (Δ)UKS with MADs for ΔE_ST below 0.03 eV.