I enjoyed this paper from the Grimme group:
Best-Practice DFT Protocols for Basic Molecular Computational Chemistry, Angew. Chem. Int. Ed., 2022, 61, e202205735.
The authors discuss the best modern methods for running a DFT computations – a guide through the jungle. For me the main conclusions regarding functionals are:
- The composite method r2SCAN-3c offers great cost-benefit a ratio and we are starting to adopt it as a default method for ground-state optimisations.
(This should be taken with a grain of salt, since it is the authors’ own method, but the arguments are sound.) - If you want to go beyond r2SCAN-3c, you have to try really really hard, as in range-separated hybrid meta-GGA with a triple/quadruple-zeta basis set. There is not really any reason to ever use double-zeta basis sets – better use a compound method.
- B3LYP/6-31G* is outdated.