Total charge of an anion is different from ideal

Top Page > Browsing

Total charge of an anion is different from ideal

Date: 2025/07/25 12:21
Name: Hiroki Akutsu <akutsu@chem.sci.osaka-u.ac.jp>: Dear OpenMX staffs,

I usually calculate organic conductors consisting of donors (BEDT-TTF) and anions, where the anion has closed shell so that the net charge should be integer, -1, -2, etc. A calculation of (BEDT-TTF)3(m-C6H4(SO3)2) includes 4 dianions (m-C6H4(SO3-)2) in the unit cell, therefore the total charge should be -8. However, the sum of Mulliken population indicates that the charge of the 4 dianions is -6.38, -1.60 a dianion. In the circumstance, the band filling of the conducting donor layer is also strange. I have checked 2 other calculations, both of which have the same problem.

Is it possible to fix the charge of anions? Is it my mistake? Or, is it intrinsic problem of the software and/or first principle calculations?

I am looking forward to getting your answer.

Cheers,

Hiroki

Page: [1]

Re: Total charge of an anion is different from ideal ( No.1 )

Date: 2025/07/31 18:37
Name: T. Ozaki

Hi,

The Mulliken populations are calculated by decomposing charge distribution
based on atomic orbitals associtated with each atom.
https://www.openmx-square.org/tech_notes/tech9-1_0/tech9-1_0.html

The atomic orbitals have long tail, and overlap with atomic orbitals
associated with other atoms. So, the overlap causes deviation from
the nominal charges. Even if you calculate NaCl bulk, the net Mulliken
populations deviate from +1 and -1.

Regards,

TO

Re: Total charge of an anion is different from ideal ( No.2 )

Date: 2025/08/09 11:55
Name: Hiroki Akutsu <akutsu@chem.sci.osaka-u.ac.jp>

Dear Ozaki-sensei,

Thank you for your answer!

You mean that Hirshfeld or more sophisticated population analysis might provide better charges close to integer.

For checking the band filling problems, we calculated the same material which does not include anion molecules and used the command,

scf.system.charge 8.0

The resultant Fermi surfaces are almost the same as the previous surfaces, suggesting that the band filling of the previous calculation is almost the same as the calculation without anions. In the circumstance, the different shape of the Fermi surface from that of the result of the tight binding calculation maybe caused by the effect of the electron correlation.

Hiroki

Page: [1]