[CP2K-user] Hybdrid functionals

Mikael Unge mikae... at gmail.com
Mon Jun 7 14:25:25 UTC 2021


Hi Frederick

Thanks, with HF section set it works better.
PBE0  or HSE06 now give larger band gaps.

Regards,
Mikael

fredag 4 juni 2021 kl. 17:03:55 UTC+2 skrev Frederick Stein:

> Hi Mikael,
>
> Probably, you are missing the Hartree-Fock section in the XC section (&HF) 
> as CP2K does not set up Hartree-Fock sections of LibXC functionals 
> automatically.
> If you would like to use the PBE implementation of CP2K, use the PBE 
> section in the XC_FUNCTIONAL section, for LibXC you have to set &LIBXC with 
> FUNCTIONAL GGA_XC_PBE.
>
> Best,
> Frederick
>
> mi... at gmail.com schrieb am Freitag, 4. Juni 2021 um 15:25:35 UTC+2:
>
>> Hi community
>>
>> I do DFT simulations of polyethylene (PE). PE has a wide band gap (~8.8 
>> eV) but with most xc funtionals the band gap is calculated to 6-7 eV.
>>
>> We have seen that hybrid functionals like HSE03, HSE06 and PBE0 give 
>> quite good values (7.97-8.75 eV) plane wave DFT
>> https://aip.scitation.org/doi/full/10.1063/1.4983650
>> However, when I try these functionals at DFT-GAPW level in CP2K (using 
>> libxc) I get band gaps in the order of 4.8-5.3 eV for a small amorphous 
>> system. 
>>
>> Is there some studies done with DFT-GAPW and hybrid functionals?
>>
>> I use version 8.1 of cp2k.
>>
>> In the input file I adress libxc functionals as in the HSE06 example below
>>      &XC_FUNCTIONAL
>>           &LIBXC
>>     FUNCTIONAL HYB_GGA_XC_HSE06 
>>           &END LIBXC     
>>       &END XC_FUNCTIONAL
>> is this the correct way?
>>
>>
>> In the manual of 8.0 it is mentioned that PBE0 should be possible to
>> use but the PBE0 keyword can not recognized by cp2k.
>>
>> https://manual.cp2k.org/cp2k-8_1-branch/CP2K_INPUT/FORCE_EVAL/DFT/XC/XC_FUNCTIONAL.html
>> Would it be any difference in using PBE0 via keyord to XC_functional 
>> compared to using libxc version of PBE0?
>>
>> Regards,
>> Mikael
>>
>>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <https://lists.cp2k.org/archives/cp2k-user/attachments/20210607/475b99af/attachment.htm>


More information about the CP2K-user mailing list