Fermi level for metallic system by GW

[yping]

Dear developer,

I am trying to do G0W0 calculation for the metallic system by Yambo. I have two questions:

first, I am also interested to know the Fermi level shift relative to LDA/GGA by G0W0. But I didn't find it in the Yambo output (except the original LDA/GGA Fermi level). Is there a way to obtain the GW corrected Fermi level as well?

second, I read the implementation about the intraband transition for dielectric matrix according to :
http://journals.aps.org/prb/pdf/10.1103 ... .64.195125
It seems a jellium model is used in the code in practice. I just wonder whether this will work for more complicated metallic system, for example metallic transition metal oxide, where the system is not a simple metal.
And in which situation, the intraband transition is important to be taken into account in a GW calculations for metallic systems?

Thank you so much!

Yuan Ping
Materials Postdoc Fellow
Joint Center for Artificial Photosynthesis (JCAP)
Lawrence Berkeley National Laboratory &
California Institute of Technology

[Daniele Varsano]

Dear Yuan,
the Fermi level variation is not reported as in line of principle one may be interested in the GW correction for selected k/bands only. In order to look at it you can for instance perform an Independent Particle absorption calculation assigning the QP correction to the response function (yambo -o c -V qp) using the XfnQPdb or XfnQP_E variables. In this way the occupations are recalculated according to the QP correction and the Fermi shift is reported.
About intraband transitions, if I remember correctly the ab-initio calculation reported in the paper you mentioned (Eq. 15-17) is not in the code anymore, and they are taken into account via a Drude model (Eq.16). The Drude plasmon energy has to be given in input. I'm not expert on transition metal oxides but I've seen in literature that the model has been used also in strongly correlated metals (see eg Appl. Opt. 37(22), 5271-5283 (1998); http://www.spectro.ethz.ch/doc/pdf/AnnPhys190.pdf). Similarly, not being expert on that, I do not have a rule of thumb to estimate when intraband contribution in the epsilon are relevant in the GW calculation.
Best,
Daniele

[yping]

Dear Daniele,

Thank you for your reply. This is really helpful. The suggestion for outputing Fermi level works well.

I have a technical problem when I use the Drude model in Yambo. Simply by modifying the Al example in the tutorial:

#
# /$$ /$$ /$$$$$$ /$$ /$$ /$$$$$$$ /$$$$$$
# | $$ /$$//$$__ $$| $$$ /$$$| $$__ $$ /$$__ $$
# \ $$ /$$/| $$ \ $$| $$$$ /$$$$| $$ \ $$| $$ \ $$
# \ $$$$/ | $$$$$$$$| $$ $$/$$ $$| $$$$$$$ | $$ | $$
# \ $$/ | $$__ $$| $$ $$$| $$| $$__ $$| $$ | $$
# | $$ | $$ | $$| $$\ $ | $$| $$ \ $$| $$ | $$
# | $$ | $$ | $$| $$ \/ | $$| $$$$$$$/| $$$$$$/
# |__/ |__/ |__/|__/ |__/|_______/ \______/
#
# GPL Version 3.4.1 Revision 3187
# http://www.yambo-code.org
#
em1d # [R Xd] Dynamical Inverse Dielectric Matrix
gw0 # [R GW] GoWo Quasiparticle energy levels
HF_and_locXC # [R XX] Hartree-Fock Self-energy and Vxc
EXXRLvcs= 941 RL # [XX] Exchange RL components
Chimod= "hartree" # [X] IP/Hartree/ALDA/LRC/BSfxc
NGsBlkXd= 27 RL # [Xd] Response block size
% QpntsRXd
1 | 29 | # [Xd] Transferred momenta
%
% BndsRnXd
1 | 20 | # [Xd] Polarization function bands
%
% DrudeW
27.0000 | 0.05000 | eV
%
% EnRngeXd
0.213329 | 9.714396 | eV # [Xd] Energy range
%
% DmRngeXd
0.10000 | 0.10000 | eV # [Xd] Damping range
%
ETStpsXd= 18 # [Xd] Total Energy steps
% LongDrXd
0.1000E-4 | 0.000 | 0.000 | # [Xd] [cc] Electric Field
%
% GbndRnge
1 | 20 | # [GW] G[W] bands range
%
GDamping= 0.10000 eV # [GW] G[W] damping
dScStep= 0.10000 eV # [GW] Energy step to evalute Z factors
DysSolver= "n" # [GW] Dyson Equation solver (`n`,`s`,`g`)
%QPkrange # [GW] QP generalized Kpoint/Band indices
1| 29| 1| 20|
%
%QPerange # [GW] QP generalized Kpoint/Energy indices
1| 29| 0.0|-1.0|
%


It seems the part for Drude mode is not reading:
" [WARNING] The system is a metal but Drude term not included."

My guess is my format for Drude frequency is not correct. I followed the format on Yambo website.
Do you have an idea what could be the problem?

Another question is when I change between "Chimod= "IP" and "Chimod= "Hartree", the output basicly didn't change. Is this a real effect of the local field effect/ Hartree part in the dielectric matrix not important
or something else?


Thanks!

Yuan Ping
Materials Postdoc Fellow
Joint Center for Artificial Photosynthesis (JCAP)
Lawrence Berkeley National Laboratory &
California Institute of Technology

[Daniele Varsano]

Dear Yuan,
the reference variable for Drude contribution in the webpage can be misleading, thanks for reporting:
You need to use DrudeWXd, DrudeWXs, DrudeWBS in order to include the Drude contribution in the dynamical/static dielectric matrix or BSE, so in your case the variable to use is DrudeWXd

About the local field effects, do you have numerically strictly *identical* results of "IP" and "Hartree" or just very similar? In metal systems it can happen that the local fields do not play a role and the independent particle it is a good approximation.
Best,
Daniele

[yping]

Dear Daniele,

Thanks for your reply.

I have tried DrudeWXd as follows for Al example:
#
# /$$ /$$ /$$$$$$ /$$ /$$ /$$$$$$$ /$$$$$$
# | $$ /$$//$$__ $$| $$$ /$$$| $$__ $$ /$$__ $$
# \ $$ /$$/| $$ \ $$| $$$$ /$$$$| $$ \ $$| $$ \ $$
# \ $$$$/ | $$$$$$$$| $$ $$/$$ $$| $$$$$$$ | $$ | $$
# \ $$/ | $$__ $$| $$ $$$| $$| $$__ $$| $$ | $$
# | $$ | $$ | $$| $$\ $ | $$| $$ \ $$| $$ | $$
# | $$ | $$ | $$| $$ \/ | $$| $$$$$$$/| $$$$$$/
# |__/ |__/ |__/|__/ |__/|_______/ \______/
#
# GPL Version 3.4.1 Revision 3187
# http://www.yambo-code.org
#
em1d # [R Xd] Dynamical Inverse Dielectric Matrix
gw0 # [R GW] GoWo Quasiparticle energy levels
HF_and_locXC # [R XX] Hartree-Fock Self-energy and Vxc
EXXRLvcs= 941 RL # [XX] Exchange RL components
Chimod= "hartree" # [X] IP/Hartree/ALDA/LRC/BSfxc
NGsBlkXd= 27 RL # [Xd] Response block size
% QpntsRXd
1 | 29 | # [Xd] Transferred momenta
%
% BndsRnXd
1 | 20 | # [Xd] Polarization function bands
%
% EnRngeXd
0.213329 | 9.714396 | eV # [Xd] Energy range
%
% DmRngeXd
0.10000 | 0.10000 | eV # [Xd] Damping range
%
ETStpsXd= 18 # [Xd] Total Energy steps
% LongDrXd
0.1000E-4 | 0.000 | 0.000 | # [Xd] [cc] Electric Field
%
% GbndRnge
1 | 20 | # [GW] G[W] bands range
%
GDamping= 0.10000 eV # [GW] G[W] damping
dScStep= 0.10000 eV # [GW] Energy step to evalute Z factors
DysSolver= "n" # [GW] Dyson Equation solver (`n`,`s`,`g`)
%QPkrange # [GW] QP generalized Kpoint/Band indices
1| 29| 1| 20|
%
%QPerange # [GW] QP generalized Kpoint/Energy indices
1| 29| 0.0|-1.0|
%
%DrudeWXd
27.0000 | 0.00500 | eV
%

It seems the code still doesn't read DrudeWXd:
" [WARNING] The system is a metal but Drude term not included. " and X Drude frequency : 0.00 0.00
Not sure what's going on. I have tried code (the Version 3.4.1) compiled on my workstation and supercomputing center. It doesn't seem to me a compiling issue.

For the local field effect, I did get identical results between Chimod= "hartree" and Chimod= "IP" for the Al. I can do a more complicated case than metal Al but this is strange.
I haven't found anywhere in the output shows the hartree is included in Chi or not.
Just want to make sure.

Thanks!

Yuan Ping
Materials Postdoc Fellow
Joint Center for Artificial Photosynthesis (JCAP)
Lawrence Berkeley National Laboratory &
California Institute of Technology

[Daniele Varsano]

Dear Yuan,
the correct syntax is:

Code: Select all

DrudeWXd= ( 0.00     , 0.00     )  eV  # [Xd] Drude plasmon

If you are not sure of the syntax you can activate the variable using more verbosity using the -V option when building the input in this case:

Code: Select all

yambo -g n -d -V resp

activates more verbosity for response function. In general, -V all, activates all the advanced option.

About the Local field effect, no idea, they should be similar but not identical: try to assign "Hartree" instead of "hartree", even if this should not be an issue, and verify also that in the output/report the variable NGsBlkXd is correctly read and different from 1 (default) (if NGsBlkXd=1 then the calculation is essentially the IP case), in the report and output the input file is reported at the end. Moreover the output file is named in different way: o.eps_q1_IP-RPA and o.eps_q1_inv_rpa_dyson.


Best,
Daniele