ΔDFT Predicts Inverted Singlet–Triplet Gaps with Chemical Accuracy at a Fraction of the Cost of Wave Function-Based Approaches

Efficient OLEDs need to quickly convert singlet and triplet
excitons
into photons. Molecules with an inverted singlet–triplet energy
gap (INVEST) are promising candidates for this task. However, typical
INVEST molecules have drawbacks like too low oscillator strengths
and excitation energies. High-throughput screening could identify
suitable INVEST molecules, but existing methods are problematic: The
workhorse method TD-DFT cannot reproduce gap inversion, while wave
function-based methods are too slow. This study proposes a state-specific
method based on unrestricted Kohn–Sham DFT with common hybrid
functionals. Tuned on the new INVEST15 benchmark set, this method
achieves an error of less than 1 kcal/mol, which is traced back to
error cancellation between spin contamination and dynamic correlation.
Applied to the larger and structurally diverse NAH159 set in a black-box
fashion, the method maintains a small error (1.2 kcal/mol) and accurately
predicts gap signs in 83% of cases, confirming its robustness and
suitability for screening workflows.