Benchmark Computations of Nearly Degenerate Singlet and Triplet states of N-heterocyclic Chromophores : I. Wavefunction-based Methods

Abstract

In this paper we investigate the role of electron correlation in predicting the S$_1$-S$_0$ and T$_1$-S$_0$ excitation energies and hence, the singlet-triplet gap ($\Delta$E$_{ST}$) in a set of cyclazines which act as templates for potential candidates for 5th generation Organic Light Emitting Diode (OLED) materials. This issue has recently garnered much interest with the focus being on the inversion of the $\Delta$E$_{ST}$, although experiments have indicated near degenerate levels with both positive and negative being within the experimental error bar (\emph{J. Am. Chem. Soc. 1980, 102: 6068 , J. Am. Chem. Soc. 1986, 108: 17} ). We have carried out a systematic and exhaustive study of various excited state electronic structure methodologies and identified the strengths and shortcomings of the various approaches and approximations in view of this challenging case. We have found that near degeneracy can be achieved either with a proper balance of static and dynamic correlation in multireference theories or with state-specific orbital corrections including its coupling with correlation. The role of spin contamination is also discussed. Eventually, this paper seeks to produce benchmark numbers for establishing cheaper theories which can then be used for screening derivatives of these templates with desirable optical and structural properties. Additionally we would like to point out that the use of DLPNO-STEOM-CCSD as the benchmark for $\Delta$E$_{ST}$ (as used in \emph{J. Phys. Chem. A 2022, 126: 8: 1378, Chem. Phys. Lett. 2021, 779: 138827}) is not a suitable benchmark for this class of molecules.