What's wrong with bader analysis in my calculation?

[wot19920302]

Dear quantumwise staffs
        I want to analyse spin distribution using bader charge analysis. It seems that maxima number can't correspond to atom number. Boron atoms have 0 charge. However, boron atoms have charge when using mulliken population. I have read  this link: http://docs.quantumwise.com/manuals/Types/Bader/Bader.html#bader-c . It says the results may be inaccurate if we use DFT because of not considering electrons near nuclei. Is it DFT that cause unreasonable results in my calculation? How do we perform bader analyse in ATK except FHI-AIM?
        P.S. attachments are bader and mulliken analyse in my calculation.
         Best 

[Ulrik G. Vej-Hansen]

As you can see on the manual page, Bader analysis is unfortunately not supposed to work with ATK-DFT, because the all-electron density is not provided: http://docs.quantumwise.com/manuals/Types/Bader/Bader.html

You could in principle still get qualitatively usable results, but this requires that there is a local maximum in the density near the nuclei, and I suspect that in this case, the electrons from B are simply too delocalized to form local maxima which the Bader analysis can detect.

[wot19920302]

sir, another question about bader analysis:  can we get spin.up and spin.down electron distribution around atoms just like mulliken population?

[zh]

It cannot be done.

I recommend the use of critic2 to perform the Bader charge analysis even if the all-electron density is provided by the ATK-DFT code.
The critic2 (http://schooner.chem.dal.ca/wiki/Critic2) provides the superposition all-electron charge of atoms.  One can also use the ELF(electron localization function) as reference to determine the basin, local maximum and so on.

[wot19920302]

sir, can critic2 or vasp  get bader analysis for different spin just like mulliken population?

[zh]

The ATK-DFT or VASP can generate the spin-up and spin-down charge densities. You need to store these charge densities separately into two different files.  And then use the same reference (ELF or all-electron charge density) to get the local volumes (basin) of every atom by using critic2 . In this way, you may be able to the local charge of spin-up and spin-down components for every atom. 

[wot19920302]

thank you sir,I will try