Page 1 of 1
GW correction
Posted: Wed Sep 25, 2013 10:05 am
by sunnyone
Dear All!
I just want to ask about the GW correction, i have calculated the band gap of a compound
LDA 0.781
GW 1.444
while the experimental is about 2.5 ev, is there any way to get a better correction ???. or what are the parameters that affecting the result??
kind regards
Re: GW correction
Posted: Wed Sep 25, 2013 9:02 pm
by gabriel.antonius
Hi sunnyone,
First, you should make sure that all parameters are converged in your calculation. These parameters include the k-point grid, the number of bands in sigma, the number of bands in the screening, the size of the dielectric matrix, ... This is not simple since the parameters are inter-dependent. Have a look at this
post which address the cross-convergence of the parameters. Also, make sure you have read the GW tutorial.
Still, it is possible that even after converging all parameters, the band gap be too small compared to experiments. This has to do with LDA overly screening the coulombian interaction. There are some materials where this problem is important, and understanding it is a difficult topic.
Depending on your material, you could consider improving the LDA starting point for the GW calculation. For example, introducing a Hubbard term "U" can improve the positioning of some bands with respect to the fermi level, which has an impact on the band gap.
Re: GW correction
Posted: Fri Oct 04, 2013 7:49 pm
by sunnyone
Dear gabriel.antonius,
I am trying to make convergence through tutorial tgw1_3 and also for my compound and i got this error:
wfk_read_ene.F90:123:ERROR
hdr_io returned fform/=0
so, is there any solution for this problem?
i am using Abinit version 7.0.5, if you can make the convergence study for the parameters for me i will send you the details of my compound
Kind regards.