Extra charge

[Fadil]

Dear users,

Is it possible to calculate dielectric function of a material with additional charge?
Actually, I tried to include extra charge with 'tot_charge' in quantum espresso, but GW+BSE calculation gives negative dielectric function.
When I skip the gw step and calculate only bse, the results are more or less reasonable.
Since the system is metallic with extra charge, is there anything special I need to do to get the dielectric function?
I would like to know that if it is a standard method to obtain the dielectric function by including extra charge in the quantum
espresso calculations, and are the results reliable? You can see my report files.

Best

[Daniele Varsano]

Dear Fadil,
note, if the system with extra charge is metallic, then consider that intraband transitions are not included in the screening, unless you use a full frequency methods and include them as a Drude model.

When I skip the gw step and calculate only bse, the results are more or less reasonable

Well, this sounds something strange happened in the GW correction, have you checked if your QP corrections are reasonable.

In any case, if the system is metallic, there is a particular reason to do a BSE calculations, for metals, independent particle calculations are usually reasonable.

Best,
Daniele

[Fadil]

Dear Daniele,

I found the problem in GW correction, but this time bse calculation get stucks in 'Wave-Function Phases'.
I am getting similar error in other calculations too. (Please see my report files.)

my bsebands;
% BSEBands
165 | 176 | # [BSK] Bands range

The error is;
[ERROR] STOP signal received while in :[08.03.02] Wave-Function Phases
[ERROR]EMPTY WF index @ b:181 k:2 spin:1 ID:0

Is it related with the extra charge inclusion?

My system is originally semiconductor. Inclusion of extra charge bring metallic states.
But the system has excitons with large binding energies. Thats why I am using bse.

Yes, I know 'intraband transitions' are not included. How can I use the drude model?

[Daniele Varsano]

Dear Fadıl,
can you also post the input file showing the problem?

The Drude term can be added setting the

Code: Select all

DrudeWXd

variable, you need to assign the Drude plasmon frequency by hand (not easy in your case). Note that it works only for full frequency method and it cannot be used with the plasmon-pole approximation. This makes the calculation very cumbersome as it needs to be converged wrt the number of frequencies. For your system it could be prohibitive.

Finally let me suggest you to update to the newer version of the code:

Code: Select all

git clone git@github.com:yambo-code/yambo.git yambo

Best,
Daniele

[Fadil]

Dear Daniele,

You can find my input and report files in the attachment.
Thank you for your help

Best

[Daniele Varsano]

Dear Fadil,
the problem is related to the fact that after QP correction you have several metallic bands, up to index 182.

Code: Select all

 [K]States summary         : Full        Metallic    Empty
                                 0001-0170   0171-0182   0183-0500
  [K]N of el / N of met el  :  340.5000    0.7981

As in your BSE you are including bands up to 180 (or 176) the BSE cannot be solved as there are not empty states considered.
You can either include more bands in the BSE, or maybe try to lower the electronic temperature:

Code: Select all

ElecTemp=

but I'm not sure this will take effect in your configuration as the Fermi level is recalculated and possibly it is not possible to lower it.

Best,
Daniele