Double-hybrid density functionals applied to a large set of INVEST systems: validating the (SOS1-)PBE-DH-INVEST expressions†

We thoroughly assess here the recently developed PBE-DH-INVEST expression and its SOS1-PBE-DH-INVEST variant, both belonging to the family of double-hybrid (DH) density functionals, against the NAH159 dataset of organic molecules displaying a very low (positive or even negative) energy gap, ΔE_ST, between the lowest-energy excited-state of singlet (S₁) and triplet (T₁) multiplicity. The NAH159 dataset comprises a large set of substituted derivatives of azulene, azupyrene, isopyrene, heptalene, cyclazine (or monoazaphenalene), pentazine (or pentaazaphenalene), and heptazine (or heptaazaphenalene) systems, thus covering most of the chemical templates so far discovered displaying ΔE_ST < 0 values. The performance of any model able to deliver correct ΔE_ST values, both in sign and magnitude, is critical for further studies in OLEDs and related applications. Given the robustness and accuracy of the results obtained by the (SOS1-)PBE-DH-INVEST functionals, together with their moderate basis set dependence, we can recommend them as an alternative to more costly wavefunction-based methods or other DH density functionals.