Benchmarking ionization potentials and electron affinities of potential photovoltaic molecules using DFT/QTP functionals and EOM-CC

Abstract

Accurate predictions of ionization potentials and electron affinities are essential for guiding the design of organic photovoltaic materials. In this study, we revisit a set of twenty molecules that have been explored in earlier studies and extend the analysis using our approaches: (1) CC theory in its ionization potentials (IPs)/electron affinities (EAs)-equation-of-motion (EOM)/coupled-cluster singles and doubles (CCSD) forms to offer a direct measure of IPs and EAs and (2) Kohn–Sham density functional theory (DFT) using the QTP family compared to a broad range of exchange–correlation functionals (from simple local density approximation/generalized gradient approximation forms through hybrid and range-separated variants). For consistency with prior calculations, one-shot G0W0 corrections are applied to each QTP result. To reduce the cost of coupled-cluster computations, we also investigate both standard and tailored frozen natural orbital (FNO) truncations. Our results confirm that local and semi-local DFT functionals exhibit significant errors, whereas global hybrids and range-separated hybrids offer improved accuracy. The QTP functionals stand out by matching or exceeding the performance of all other functionals. G0W0 on top of DFT starting points further refines orbital energies, bringing them into close agreement with coupled-cluster results. Tailored FNO truncations preserve coupled-cluster accuracy while reducing the virtual space by nearly half. Timing tests on anthracene demonstrate that QTP00 and G0W0@QTP00 workflows almost achieve coupled-cluster quality predictions in under a day, compared with week-long runtimes for full EA-EOM/CCSD.