[gpaw-users] Correct settings for core hole calculations?
Toma Susi
toma.susi at aalto.fi
Wed Apr 24 15:42:10 CEST 2013
On 24.4.2013, at 16:03, Mathias Ljungberg wrote:
> Hi Toma,
Hi Mathias!
>
>
> On Apr 24, 2013, at 1:25 PM, Toma Susi wrote:
>
>> Dear gpaw-users,
>>
>> I am doing core hole calculations to look at the core binding energies of doped graphene. In the GPAW Wiki for XAS [1], it is mentioned that the excited state should be spin polarized, and thus one needs to set the magnetic moment on the core hole atom to 1, and fix the total magnetic moment.
>>
>> -Does this apply to any calculations with a core hole (specifically I'm currently looking at the simple core level binding energies for simulating XPS), or just for XAS? Is there a physical picture why this is necessary, or is it just a computational detail?
>
> The physical picture is that an core electron of spin alpha is excited to an unoccupied level with spin alpha for XAS.
> For XPS we can just remove one electron in the excited state, but the core hole still has spin alpha. For molecules the total energy differences can change by 10 eV or so by using spin polarization, for solids it is more complicated. It doesn't make sense to force the spin occupation for a metal but for an insulator it might. You are kind of pioneering this type of calculation so I guess the best way would be to try different things and see what happens….
Okay, thanks for the clarification.
Preliminarily it seems that forcing the magnetic moment of the core hole atom raises the energy (=smaller negative energy) of the core hole system compared to the reference system, meaning the total energy difference and thus the binding energy becomes larger.
In the cases I've studied, this brings the binding energies into better agreement with experimental values. But if you think this is unphysical for XPS and/or not advisable for a metallic system such as graphene, perhaps it's better not to use it.
>
>
>> -Furthermore, is there anything beyond the setup of the core hole that one should take into account when calculating 2p core levels for more massive elements (as compared to calculating the 1s levels)?
>
> The aim of the implementation was never to calculate 2p core level spectra since spin-orbit coupling becomes important in this case. But you can always try it with total energies (I am not sure the transition dipole elements are implemented though)
Preliminarily I can say that the 2p core hole calculation seems to work without a hitch, no issues whatsoever with convergence etc. And the few energy values I've calculated are actually even closer to experimental ones than for most 1s levels I've studied.
But of course, this can just be luck in these few cases, I'll have to make a more systematic study to see if this holds.
In any case, do you think that the core hole shouldn't be used for the 2p core levels?
>
>>
>> -Finally, is there a way to simulate core level electron energy loss spectra (EELS) with GPAW?
>
> I don't think so.
Any way to do this indirectly?
>
> Best regards.
> Mathias Ljungberg
>
>
http://physics.aalto.fi/personnel/?id=322
http://mostlyphysics.wordpress.com/
"Love is the only emotion that enhances our intelligence."
-Humberto Maturana
More information about the gpaw-users
mailing list