Abstract
The accuracy of dispersion-corrected calculations (DFT-D2, DFT-D3 and DFT-NL) is assessed here, with large basis sets (def2-QZVP) to avoid incompleteness effects, for the most stable structure of a real-world polyphenol dimer chosen as an appropriate model. Natural polyphenols form such complexes with π-stacking playing a key stabilizing role. Our benchmark calculations predict its existence favored by 22-24 kcal/mol with respect to the isolated monomers, mainly driven by both π-π and H-bonding interactions. The adequate comparison of lower-cost DFT-based methods allowed bracketing their expected accuracy. These results thus pave the way towards reliable studies of challenging aggregation processes of natural products.