[gpaw-users] Convergence Problem with Core-Hole Calculations

[Mathias Ljungberg]

On May 24, 2010, at 6:53 PM, Hongliang Xin wrote:

> Dear Mathias,
>
> Thanks for your information. I have to confirm with you several  
> points below to make sure I understand it correctly.
> 1. For core-level shift calculations with periodic boundary  
> condition, the system is not well defined as mentioned in previous  
> post. Putting the extra electron in the valence band will probably  
> over screen the core-hole, so the big unit cell can relieve this  
> problem.
>

Yes, that is right. For a metallic system at least the system with an  
extra electron and the one without it should converge to the same  
energy for a large enough cell. For metallic screening the XAS and XPS  
energies coincide. For insulators I am not really sure but it might be  
the same even there.


> 2. For calculations with core-hole, do we always need to use spin- 
> polarized calculator and fix the magmom on the excited atom to be 1?
>

Fixing the magnetic moment is only really feasible for non-magnetic  
systems. For a magnetic system the occupations of different spin  
orbitals will reorganize themselves in response to the core hole so  
fixing the magnetic moment seems like a bad idea in this case. Anyway,  
to get all the relaxation a spin polarized calculation is necessary,  
however I haven't really looked into energy differences between core  
holes  when doing a spin paired calculation.

Mathias


> Thanks,
>
> Hongliang
>
> On Mon, May 24, 2010 at 11:18 AM, Mathias Ljungberg <mathias.ljungberg at physto.se 
> > wrote:
> Hi Hongliang,
>
> First of all your system is charged. Because of a the core hole you  
> need to put charge=-1 to get an extra electron in the valence band - 
> if that is what you want. Maybe this is the reason the dipole  
> correction does not work. If you have a neutral system I think mixed  
> periodic boundary conditions should work as well. Anyway, the  
> convergence of the total energy of the core hole state is pretty  
> sensitive to the grid spacing and size of the box, I attach some  
> plots of tests I did for a water molecule (without any dipole  
> correction). The first plot shows the convergence with grid spacing  
> and the XAS case has got an extra electron in the valence band, XPS  
> does not. The second plot shows the convergence with box size. If  
> you are looking for relative shifts maybe you can get away with less  
> converged parameters. One more thing, to get reliable total energies  
> you should use a spin polarized calculation.
> <dks_grid.eps><dks_box_size.eps>
>
> Best Regards,
> Mathias
>
>
>
>
>
>
>
> On May 19, 2010, at 3:39 AM, Hongliang Xin wrote:
>
>> Here is the input file and output. I am not exactly calculating the  
>> delta Kohn-Sham energy. I am trying to evaluate the final state  
>> effect of core-level shift. I am not very convinced about what I am  
>> doing though. I just thought that the calculated delta energy  
>> should be converged with respect to vacuum space. For some reason,  
>> it did not allow me to put non-periodic on z direction with core- 
>> hole.
>>
>> Any suggestions would be very helpful.
>>
>> Thanks,
>>
>> Hongliang
>>
>> On Tue, May 18, 2010 at 6:56 PM, Mathias Ljungberg <mathias.ljungberg at physto.se 
>> > wrote:
>> Hi Hongliang,
>>
>> Can you send a script that illustrates your problem, including the  
>> text output? I think the delta Kohn-Sham energy should converge  
>> without the dipole correction.
>>
>> Best Regards,
>> Mathias Ljungberg
>>
>>
>>
>>
>> On May 18, 2010, at 9:56 PM, Hongliang Xin wrote:
>>
>>> Dear All,
>>>
>>> I have been experiencing the convergence problem with the core- 
>>> hole calculations. Without dipole correction, the delta Kohn-Sham  
>>> energy is not converging with respect to the vacuum space. Adding  
>>> dipole correction as instructed before, the SCF convergence is  
>>> very very fluctuating due to the WFS problem as mentioned on  
>>> previous Dipole Correction question. I have tried to reduce the  
>>> spacing to 0.15, or change the mode to "lcao", and the problem is  
>>> still there.
>>>
>>> Any suggestions with this situation would be highly appreciated.
>>>
>>>
>>> Thanks,
>>>
>>> Hongliang
>>>
>>> -- 
>>> Hongliang Xin
>>> Ph.D. Candidate
>>> Dept. of Chemical Engineering
>>> University of Michigan
>>> 3166 HH Dow
>>> 2300 Hayward
>>> Ann Arbor, MI 48109
>>> Phone: (734) 647-8051
>>> E-mail: hxin at umich.edu
>>> _______________________________________________
>>> gpaw-users mailing list
>>> gpaw-users at listserv.fysik.dtu.dk
>>> https://listserv.fysik.dtu.dk/mailman/listinfo/gpaw-users
>>
>>
>>
>>
>> -- 
>> Hongliang Xin
>> Ph.D. Candidate
>> Dept. of Chemical Engineering
>> University of Michigan
>> 3166 HH Dow
>> 2300 Hayward
>> Ann Arbor, MI 48109
>> Phone: (734) 647-8051
>> E-mail: hxin at umich.edu
>> < 
>> OAgfcc 
>> .txt 
>> ><Script.py><atoms.nc>_______________________________________________
>>
>> gpaw-users mailing list
>> gpaw-users at listserv.fysik.dtu.dk
>> https://listserv.fysik.dtu.dk/mailman/listinfo/gpaw-users
>
>
>
>
> -- 
> Hongliang Xin
> Ph.D. Candidate
> Dept. of Chemical Engineering
> University of Michigan
> 3166 HH Dow
> 2300 Hayward
> Ann Arbor, MI 48109
> Phone: (734) 647-8051
> E-mail: hxin at umich.edu

-------------- next part --------------
An HTML attachment was scrubbed...
URL: http://listserv.fysik.dtu.dk/pipermail/gpaw-users/attachments/20100524/528f21eb/attachment.html