Intermolecular charge-transfer is a highly important process in biology and energy-conversion applications where generated charges need to be transported over several moieties. However, its theoretical description is challenging since the high accuracy required to describe these excited states must be accessible for calculations on large molecular systems. In this benchmark study, we identify reliable low-scaling computational methods for this task. Our reference results were obtained from highly accurate wavefunction calculations that restrict the size of the benchmark systems. However, the density-functional theory based methods that we identify as accurate can be applied to much larger systems. Since targeting charge-transfer states requires the unambiguous classification of an excited state, we first analyze several charge-transfer descriptors for their reliability concerning intermolecular charge-transfer and single out DCT as an optimal choice for our purposes. In general, best results are obtained for orbital-optimized methods - and among those, IMOM proved to be the most numerically stable variant - but optimally-tuned range-separated hybrid functionals combined with rather small basis sets proved to yield surprisingly good results. This makes these fast calculations attractive for high-throughput screening applications.
