Is the total energy for charged systems reliable?

[Nico Holmberg]

Hi,

The same compensating homogeneous background charge is (implicitly) applied 
in CP2K if your system is charged. The compensating background charge is 
required because the electrostatic energy of a periodic charged system 
diverges. This is standard operation in most codes. 

A direct comparison of energies calculated with different charges is 
therefore meaningless because the total energy of a charged system 
converges very slowly w.r.t. to system size. You can however apply a 
finite-size correction to the energy of the charged system. It should be 
safe to compare energy differences after the correction. Some codes apply a 
correction automatically but with CP2K you have to do it manually. Applying 
e.g. a Madelung correction to the energy should be straightforward. There 
should be plenty of literature regarding finite-size corrections in charged 
vacancy modelling and even comparisons of different methods. This paper 
<https://journals.aps.org/prb/abstract/10.1103/PhysRevB.86.045112> could be 
a useful starting point.



BR, 

Nico

tiistai 20. kesäkuuta 2017 13.43.56 UTC+3 Cristiana Passiu kirjoitti:
>
> Dear all,
> the linked paper, from 2010, (
> http://aip.scitation.org/doi/10.1063/1.3491271), in which calculations 
> have been performed with the CRYSTAL package, states in the second 
> paragraph: "the total energies of charged systems have no physical meaning 
> in CRYSTAL06 because of the interaction with the balancing background of 
> charge".
> Do you know if this also applies to the cp2k package? I would like to 
> compare the stability of a neutral crystal with those of the same crystal 
> with positive or negative charges, which I'm using to model charged oxygen 
> vacancies. Does it make sense to compare the total energies obtained after 
> geometry optimization?.
>
> Many thanks,
> Cristiana
>
>
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <https://lists.cp2k.org/archives/cp2k-user/attachments/20170620/b5fbec0b/attachment.htm>