Multiwfn forum

Dirk

Member

Registered: 2021-04-02

Posts: 4

CM 5 charges from xtb

Dear Dr. Lu,

thank you very much for extending population analyses toward periodic systems.
I have a question relating to the calculation of CM5 charges from CP2K files:

CP2k integrates the Gfn-1-xtb tight binding approach from Grimme et al. and
calculation of periodic cells with 100s of atoms are quite fast and efficient.

Could the output from xtb calculations also be utilized to create CM5 charges,
and if so: are large deviations of the values in comparison to DFT calculations to be expected ?

With kind regards, Dirk

sobereva

Tian Lu (Multiwfn developer)

From: Beijing

Registered: 2017-09-11

Posts: 1,631

Website

Re: CM 5 charges from xtb

Dear Dirk,

Unfortunately, I found current version of CP2K is unable to output .molden file for GFN1-xTB calculation.

Since CP2K is quite fast, regular DFT calculation for ~100 atoms is very easy.

Best regards,

Tian

Dirk

Member

Registered: 2021-04-02

Posts: 4

Re: CM 5 charges from xtb

Dear Tian,

thank you very much for the fast reply. I am going to post a feature request to the dvelopers of CP2k.
Reading the manual of the recent version of the stand-alone version of xtb it appeared that the GFn-1 version
of the code produces Mulliken & CM-5 charges by default. The GFn-1 functional is capable of treating 3D periodic cell,
hence there exists an acceptable solution.

sobereva

Tian Lu (Multiwfn developer)

From: Beijing

Registered: 2017-09-11

Posts: 1,631

Website

Re: CM 5 charges from xtb

Dear Dirk,

I would like to mention that according to my some tests, quality of GFN1-xTB wavefunction is not quite satisfactory. For example, I compared electrostatic potential (ESP) distribution on molecular surface for a few polar systems, the difference between GFN-xTB ESP and DFT ESP is evident, they are only qualitatively in agreement with each other. Therefore, when common DFT calculation cost is acceptable, I always suggest using DFT wavefunction for deriving atomic charge purpose.

In fact I intended to publish a work aiming at comprehensively evaluating quality of wavefunction of GFN-xTB method long time ago, but I still do not find adequate time to do... I hope this work could be finished in the next few months.