Multiwfn forum

saeed_E

Member

Registered: 2019-12-21

Posts: 240

MP2 or dispersion-corrected DFT-functionals?

Dear Tian,
If you kindly let me, I like to ask a short question regarding using MP2 method versus dispersion-corrected DFT functionals within geometry optimization.
In an article (while I could not find that, it had been published in J. Phys. Chem. A if I am not wrong), it was stated that dispersion effects are very important in non-covalent interactions such as intermolecular hydrogen- or halogen-bonding and must be strongly be considered. In such cases, MP2 is quite preferred over dispersion-corrected DFT functionals such as B3LYP-D3(BJ) for geometry optimizations since MP2 is much more robust than dispersion-corrected DFT functionals to consider dispersions. On the other hand, while MP2 is a generalized method and always (in any case of non-covalent interactions) gives reasonable results for dispersion effects, the dispersion-corrected DFT functionals are highly case-dependent methods. For instance, if B3LYP-D3BJ gives good results for Cl...Cl interaction, it does not necessarily gives good results for N...Br interaction as well. Please let me know whether or not you completely agree and confirm reasonableness of this statement.

In advance, too many thanks for your kind attention.

Sincerely,
Saeed

sobereva

Tian Lu (Multiwfn developer)

From: Beijing

Registered: 2017-09-11

Posts: 1,777

Website

Re: MP2 or dispersion-corrected DFT-functionals?

Dear Saeed,

MP2 never performs so well as you stated. For example, it is well known that MP2 tends to strongly overestimate pi-pi stacking interaction.

B3LYP-D3(BJ) works well for geometry optimization for broad range of weaklying interacting systems. For example, the high-quality HB test set HB300SPX (J. Chem. Theory Comput. 2020, 16, 10, 6305–6316) employs B3LYP-D3(BJ) for geometry optimization, and J. Chem. Theory Comput., 11, 1481 (2015) showed that B3LYP-D3(BJ) performs nicely in optimization of many different kinds of dimers.

MP2 is much more expensive for geometry optimization and especially frequency analysis purposes, generally I do not recommend to use it. If an absolutely reliable and robust result is needed, I prefer to use double-hybrid functionals, such as revDSD-PBEP86-D3(BJ).

Best,

Tian

saeed_E

Member

Registered: 2019-12-21

Posts: 240

Re: MP2 or dispersion-corrected DFT-functionals?

Dear Tian,
Too many thanks for your highly kind attention to provide such very valuable, informative, and very nice comments. As always, I really extremely enjoyed and very learned from your comments. I will carefully study the article you so kindly did introduce me, my nice friend.

But, there is is challenging question in my mind! Professor Ibon Alkorta works on non-covalent interactions for years and he is quite expert in this field. In the almost all his study he employs only MP2 method in conjunction with a very large basis set such as aug-cc-pVTZ for geometry optimization purpose in the intermolecular XB- or HB-bonding mostly published in ASC journals! Could you please let me know, considering such very valuable advantages of DFT methods,why he always use MP2?

Yours sincerely,
Saeed

PS. I also asked you another question in "quantum chemistry forum" regarding energy decomposition using extralink 608. I would be highly grateful if you kindly answer this question as well.

Last edited by saeed_E (2023-03-01 08:54:44)

sobereva

Tian Lu (Multiwfn developer)

From: Beijing

Registered: 2017-09-11

Posts: 1,777

Website

Re: MP2 or dispersion-corrected DFT-functionals?

Dear Saeed,

You should ask him. In my opinion, MP2 is out-of-date and never the preferential choice nowadays.

I am too busy to carefully look at the L608 problem...

Best,

Tian

saeed_E

Member

Registered: 2019-12-21

Posts: 240

Re: MP2 or dispersion-corrected DFT-functionals?

Dear Tian,
My highest and warmest gratitude is presented to you for your valuable and nice comments; thank you very very much.

Sincerely yours,
Saeed