How do I choose the best exchange correlation functional for running a DFT calculation?
Iam running different attacks on this
You can safely used M06-2X, many benchmarks have shown that M06-2X is one of the best and robust functional for organic type of systems, wB97XD is also a viable choice. Although ωB97M-V should perform better than M06-2X, it has not been supported by most well-known quantum chemistry codes yet.
If you require higher accuracy, double-hybrid functional could be taken into account. PWPB95-D3 (supported by ORCA) is strongly recommended. The B2PLYP-D3(BJ) supported by Gaussian is also a good choice.
Dear Lester,
Everything mentioned so far is correct and these are very capable functionals. However, these functionals, in particular double-hybrids, require large basis sets to yield accurate results. That is, triple zeta quality (cc-pVTZ, def2-TZVP) for structures and quadruple zeta (cc-pVQZ, def2-QZVPP) or better complete-basis extrapolation for energies if you want to be quantitative. Otherwise, basis-set superposition error (BSSE) will be quite significant. Hence, these calculation, if done correctly, are rather expensive.
Here, Grimmes composite method PBEh-3c provides a cost-effective alternative, as it provides reliable geometries (and energies) with a small def2-mSVP (similar to def2-SV(P)) basis and includes not only dispersion corrections (D3-BJ), but also the geometric counter-poise correction (gCP) to eliminate BSSE. The method is implemented in the free ORCA program package, as well as in Q-Chem and Turbomole. You can use it to quickly scan a large number of reactions, and then refine the results using single-point calculations with one of the above-mentioned functionals in a large basis.
Article Consistent structures and interactions by density functional...
thank you for the answers, im using DMOl3 as it is built in materials studio version 5. only basic functionals are available ( no hybrid functionals, no metaGGA, and no hartree fock)
LDA:
-PWC
-VWN
GGA
-PW91
-BP
-PBE
-BLYP
-B0P
-VWN-BP
-RPBE
-HCTH
also, numerical basis sets are only available, namely:
-DN
-DND
-DNP
-TNP.
If this is what you have available, I'd recommend any of the GGAs. However, all of these functionals will only provide qualitative results. To get anywhere near quantitative, you wont get around using a hybrid with a large basis and at least an estimate of thermochemical contributions. All of this is available e.g. in the free ORCA software (https://orcaforum.cec.mpg.de/), which is very straightforward to install and use. For example, the input-file for an optimization with the previously mentioned PBEh-3c approach would look like:
!PBEh-3c OPT
*xyz 0 1
molecule xyz coordinates
*
and that's it. Give it a try!
- Badges
- Science topic
- Similar topics
- Chemistry