IP spectra of bilayer graphene - choice of k-points

[chwolf]

Dear all,

I am working on bilayer graphene with finite electric field; this system has a "real" gap (~0.3-0.4 eV) and I want to see its low-energy optical spectrum (basically to see how doping affects the spectrum); however I find it extremely hard to find a k-point grid that gives a "converged" spectrum.

Since applying a field splits the "Dirac cone" into two separate "cones" in k-space I was first looking into finding the k-grid that gives the smallest band-gap (i.e. this really hits the VBM and CBM). All my grids are divisible by 3 (i.e. they also contain the K point which has a significant transition amplitude). For grids 9,33,66,72,78.. the band-gap (last column) is

Code: Select all

  41 2.06940 2.79530 0.72590
 545 2.27250 2.61450 0.34200
2180 2.31460 2.57420 0.25960
[b]2594 2.31650 2.57100 0.25450[/b]
3044 2.31000 2.57630 0.26630
4052 2.29280 2.59580 0.30300
4901 2.30970 2.57900 0.26930

Now the gap of grids with 66, 72 and 78 k-points look quite similar but their spectra look quite different; Part of it might just be due to smearing but I was wondering if there is a better way of testing this? Since I will change the field and the doping level testing this every time will be very cumbersome...

Thanks in advance for your help and insight!

Chris

Code: Select all

# GPL Version 4.2.1 Revision 110. (Based on r.14778 h.7b4dc3)
#                        MPI Build                          
#                http://www.yambo-code.org                  
#
optics                         # [R OPT] Optics
chi                            # [R CHI] Dyson equation for Chi.
Chimod= "IP"                   # [X] IP/Hartree/ALDA/LRC/BSfxc
NGsBlkXd= 1            Ry      # [Xd] Response block size
% QpntsRXd
    1 |  1 |                 # [Xd] Transferred momenta
%
% BndsRnXd
   1 |  120 |                   # [Xd] Polarization function bands
%
% EnRngeXd
  0.00000 | 5.00000 | eV      # [Xd] Energy range
%
% DmRngeXd
  $smear |  $smear | eV      # [Xd] Damping range
%
ETStpsXd= 1000                  # [Xd] Total Energy steps
% LongDrXd
 0.000000 | 0.000000 | 1.000000 |        # [Xd] [cc] Electric Field
%

[Daniele Varsano]

Dear Christoph,
actually, I do not have a clear suggestion for that. It really seems that you need a large k point sampling, besides that my suggestion is to increase the sampling by doubling it (when possible, ie having the smaller sampling as a subset of the larger one) and look at the spectrum considering larger smearing.

Best,
Daniele