Multiwfn forum / Basis set requirement for anions under very high level of theory

Re: Basis set requirement for anions under very high level of theory

dummy@example.com (wham09) — Sun, 15 Mar 2026 16:31:48 +0000

I appreciate your replies!

Re: Basis set requirement for anions under very high level of theory

dummy@example.com (sobereva) — Sat, 14 Mar 2026 00:24:22 +0000

1 Yes, but the larger the basis set, the smaller the accuracy improvement from adding diffuse functions

2 In principle, ma-TZVPP/ma-QZVPP is not suitable for basis set extrapolation, because the diffuse function part is not constructed in a systematic manner.

Re: Basis set requirement for anions under very high level of theory

dummy@example.com (wham09) — Fri, 13 Mar 2026 00:23:50 +0000

Thank you for the reply. To further clarify, if the conditions you explained (significantly negative charge on certain atoms, and their participation in the reaction) are met, addition of diffuse functions to those atoms is required even if the basis set is already QZ or 5Z quality?

Additional: I notice that in the Multiwfn’s orca input generating function, the option for DLPNO-CCSD(T)/CBS (3,4/def2) disappears when I activate the diffuse function option, and only the option for extrapol(3,4/aug-cc) remains. Is this because ma-TZVPP/ma-QZVPP are somewhat less appropriate for extrapolation?

Re: Basis set requirement for anions under very high level of theory

dummy@example.com (sobereva) — Thu, 12 Mar 2026 21:34:29 +0000

The answer depends on if there are atoms with significant negative charge (e.g. ADCH atomic charge < -0.5) and meantime directly participating in the reaction, it is not dependent of theoretical method. Only those atoms need diffuse functions to improve result.
However, note that if basis set extrapolation is used, the necessity of adding diffuse functions is somewhat weakened.

Basis set requirement for anions under very high level of theory

dummy@example.com (wham09) — Thu, 12 Mar 2026 17:27:09 +0000

Dear Prof. Lu,

I wonder if diffuse functions are "always" required for reaction energy calculation of anions, even if the calculation level is already very high. For example, in the following 4 scenarios, should diffuse functions be added?

1. ordinary hybrid functional (e.g. PBE0-D4 or wB97M-V) with def2-QZVPP: Should change to ma-QZVPP?
2. double hybrid functional (e.g. PWPB95-D4 or wB97M(2)) with def2-QZVPP: Should change to ma-QZVPP?
3. DLPNO-CCSD(T) with CBS (def2-TZVPP / def2-QZVPP): Should change to ma-TZVPP -> ma-QZVPP?
4. DLPNO-CCSD(T) with CBS (cc-pVTZ / cc-pVQZ): Should change to aug-cc-pVTZ -> aug-cc-pVQZ?