metal ions in cp2k

[Matt W]

Hi Qiang,

I think that dissociation into partially charged ions is the 'correct' 
incorrect behaviour using standard functionals - see

'@article{ruzsinszky2006spurious,
  title={Spurious fractional charge on dissociated atoms: Pervasive and 
resilient self-interaction error of common density functionals},
  author={Ruzsinszky, Adrienn and Perdew, John P and Csonka, G{\'a}bor I 
and Vydrov, Oleg A and Scuseria, Gustavo E},
  journal={The Journal of chemical physics},
  volume={125},
  number={19},
  pages={194112},
  year={2006},
  publisher={AIP Publishing}
}'

for instance. 

Also, it is quite hard to converge wavefunctions for the stretched molecule 
- I'm not sure your set up really converges. I suspect that Gaussian 
converges to a different solution - though you don't say what setup you are 
using there.

So, actually what you are trying to do is quite difficult, rather than an 
easy task. A better approach is to model the NaCl molecule near equilibrium 
then calculate the Na and Cl atoms separately for most thermodynamic 
applications.

Matt

On Sunday, November 13, 2016 at 7:47:13 AM UTC, Qiang Li wrote:
>
> Dear Vladimir,
>
> The UKS does not help to separate the charges, as I expected. I did a 
> calculation with same system by Gausssian 09 for a test, and the Mulliken 
> charge analysis shows that the Na(+1)Cl(-1) are well separated. I do expect 
> similar result from CP2k, since it also describes the valence electrons 
> with gaussian basis sets.  I am uploading the input file for my cp2k run, 
> which you can have a quick look whether I did something wrong or not.
>
> Thank you for being replying.
>
> Best regards,
> Qiang
>
> 在 2016年11月10日星期四 UTC+8下午7:51:26，Vladimir Rybkin写道：
>>
>> Hi Li,
>>
>> this means you have a "covalent" dissociation into atoms, not ions. Try 
>> to do a spin-unrestricted calculation, i.e. UKS (alternative name - LSD) 
>> calculation in the DFT section.
>>
>> Yours,
>>
>> Vladimir
>>
>> четверг, 10 ноября 2016 г., 12:01:16 UTC+1 пользователь Qiang Li написал:
>>>
>>>
>>> Hi, Vladimir,
>>> Thanks for your reply, that was quite helpful.
>>> I did some tests before I reply back. Simply, I calculated Na+Cl- system 
>>> (distance is set to 10 angstrom, using a 20x10x10 box), and did the 
>>> Mulliken charge analysis. I found that using  GPW method, i.e. using the 
>>> GTH potential and DZVP basis set (same method in this paper: 
>>> http://dx.doi.org/10.1088/2053-1583/3/2/025026), the charges was 
>>> Na(0.5)Cl(-0.5), which is very strange to me. Alternatively, the charges 
>>> are well separated when I apply GAPW method (Na+Cl-).
>>> What's the physics behind this?
>>> Best regards,
>>> Li
>>>
>>>
>>> 在 2016年11月8日星期二 UTC+8下午11:03:55，Vladimir Rybkin写道：
>>>>
>>>> Dear Li,
>>>>
>>>> there are Mulliken and Hirshfeld population analyses in CP2K, which 
>>>> provide partial charges on atoms. 
>>>> To keep charges separated you need a large simulation cell. 
>>>> Alternatively, you may consider a charged cell (no counter ions). This is 
>>>> simpler to do, although, the large cell is even more a must.
>>>>
>>>> Yours,
>>>>
>>>> Vladimir 
>>>>
>>>> вторник, 8 ноября 2016 г., 15:31:07 UTC+1 пользователь Qiang Li написал:
>>>>>
>>>>> Dear all,
>>>>> I found several publications dealing with metal ions in the AIMD 
>>>>> calculations, Zn(2+) for example. The way I see it is that the authors use 
>>>>> Cl- counter ions to neutralize the system. However, how one can make 
>>>>> sure that the charge of the two ions are well separated, i.e. Zn (2+) and 
>>>>> Cl (-1) ? is there any charge analysis available  in CP2K 4.1 ? Any 
>>>>> suggestions are appreciated.
>>>>> All the bests,
>>>>> Li
>>>>>  
>>>>>
>>>>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <https://lists.cp2k.org/archives/cp2k-user/attachments/20161116/ce7b501c/attachment.htm>