Skip to Content.
Sympa Menu

forum - Re: [abinit-forum] Problem in GW calculation of GaAs

forum@abinit.org

Subject: The ABINIT Users Mailing List ( CLOSED )

List archive

Re: [abinit-forum] Problem in GW calculation of GaAs


Chronological Thread 
  • From: Zeila Zanolli <zeila.zanolli@uclouvain.be>
  • To: forum@abinit.org
  • Subject: Re: [abinit-forum] Problem in GW calculation of GaAs
  • Date: Fri, 2 Oct 2009 12:36:57 +0200

Hi Sankeerth,

getting the gap of GaAs 'right' is not as easy as it might seem, even in the GW approximation (and even if everything is well converged).

To cut a long story short, the problem is essentially related to the presence of the Ga 3d states:
they need to be included among the valence states since they have a spatial superposition with the 4s and 4p
but the resulting LDA gap is too small compared to the experimental one,
and the GW calculation (who is ment to be a 'small' correction of the LDA one) can't open it enough to get the experimental value.

I've tried to address this problem some time ago, also going beyond a 'one-shot' GW calculation.
The result (together with many references, and a better explanation of the problem) can be found in the paper:

Phys. Rev. B {\bf 75}, 245121 (2007)
"Model GW band structure of InAs and GaAs in the wurtzite phase."
Zanolli, Fuchs, Furthmueller, von Barth, Bechtstedt

(Despite the title refers to 'wurtzite', results for ''zincblende" GaAs are also reported)


All the best,
Zeila



On 2 Oct 2009, at 04:57, ธนูสิทธิ์ บุรินทร์ประโคน wrote:

Dear Sankeerth

I feel that the ecutwfn and ecutsigx you used for the GW calculation in dataset4, and the ecutwfn and ecuteps you used for the screening calculation in dataset3 may not the well converged values (too small) for GaAs. You may create a new KSS file with the nbandkss2 set to 200. Then carefully carry out the convergence test of GW corrections with respect to the mentioned parameters, by following the tutorials.

Hope this help.

Thanusit


On September 26, 2009 3:34:33 AM ICT, "sankeerth rajalingam" <sankeerth.rajalingam@gmail.com>, "sankeerth rajalingam" <sankeerth.rajalingam@gmail.com> wrote:


Dear ABINIT users,

            I am a new ABINIT user. I am learning to do GW calculations to achieve correct bandgap. I have worked on basic tutorials and also band structures.
I have performed GW correction for GaAs bandstructure at the gamma point, in a way similar to the one specified in the tutorial, using Troullier-Martins pseudopotentials.

The output of the calculation, the corrected bandgap was 1.203 eV, which is significantly different from the following values:


1.58 eV, in Godby RW, Schluter M and Sham LJ, "Self-energy operators and exchange-correlation potentials in semiconductors", Phys. Rev. B. 37, 10159 (1988).


1.52eV (experimental value from TABLE 1), in Remediakis IN and Kaxiras E, “Band-structure calculations for semiconductors within generalized-density-functional theory”, Phys. Rev. B. 59, 5536 (1999).

 

Can someone please specify the reasons for the difference in the output, when compared to the value in the reference?


Also, please suggest me if I need to make any changes in the input file.

 



-------------------------INPUT FILE BEGINS---------------------------


The input file used for the calculation is shown below.

ndtset      4

 shiftk2  0.0 0.0 0.0  # This grid contains the Gamma point
             0.0 0.5 0.5
             0.5 0.0 0.5
             0.5 0.5 0.0
istwfk2  19*1                    # Option needed for Gamma
iscf2    -2             # Non self-consistent calculation
getden2  -1             # Read previous density file
nband2   9
nbandkss2 100        # Number of bands to store in KSS file

# Dataset3: Calculation of the screening (epsilon^-1 matrix)
optdriver3  3        # Screening calculation
getkss3     -1       # Obtain KSS file from previous dataset
nband3       150      # Bands to be used in the screening calculation
ecutwfn3     5     # Planewaves to be used to represent the wavefunctions
ecuteps3     7     # Dimension of the screening matrix
ppmfrq3    16.7 eV  # Imaginary frequency where to calculate the screening


# Dataset4: Calculation of the Self-Energy matrix elements (GW corrections)
optdriver4  4        # Self-Energy calculation
getkss4     -2       # Obtain KSS file from dataset 1
getscr4     -1       # Obtain SCR file from previous dataset
nband4      200      # Bands to be used in the Self-Energy calculation
ecutwfn4    6      # Planewaves to be used to represent the wavefunctions
ecutsigx4   7      # Dimension of the G sum in Sigma_x
                     # (the dimension in Sigma_c is controlled by npweps)
nkptgw4      1                # number of k-point where to calculate the GW correction
kptgw4                       # k-points
  0.000    0.000    0.000    # (Gamma)
bdgw4       4  5             # calculate GW corrections for bands from 4 to 5


# Definition of the unit cell: fcc
acell  3*10.683187931        # This is equivalent to   10.217 10.217 10.217
rprim  0.0  0.5  0.5   # FCC primitive vectors (to be scaled by acell)
          0.5  0.0  0.5
          0.5  0.5  0.0

# Definition of the atom types
ntypat 2          # There is only one type of atom
znucl 31 33          # The keyword "znucl" refers to the atomic number of the
                  # possible type(s) of atom. The pseudopotential(s)
                  # mentioned in the "files" file must correspond
                  # to the type(s) of atom. Here, the only type is Silicon.

# Definition of the atoms
natom 2           # There are two atoms
typat  1 2        # They both are of type 1, that is, Silicon.
xred              # Reduced coordinate of atoms
      0.0  0.0  0.0
      0.25 0.25 0.25

# Definition of the planewave basis set (at convergence 16 Rydberg 8 Hartree)
ecut 8.0          # Maximal kinetic energy cut-off, in Hartree

# Use only symmorphic operations
symmorphi 0

# Definition of the SCF procedure
nstep   10        # Maximal number of SCF cycles
diemac  12.0      # Although this is not mandatory, it is worth to
                  # precondition the SCF cycle. The model dielectric
                  # function used as the standard preconditioner
                  # is described in the "dielng" input variable section.
                  # Here, we follow the prescription for bulk silicon.
tolwfr  1.0d-10

# This line added when defaults were changed (v5.3) to keep the previous, old behaviour
  iscf 5

------------------------------INPUT FILE ENDS--------------------------

 
 

----------------------------OUTPUT FILE BEGINS-----------------------

The output of the calculation is shown below.

 k =    0.000   0.000   0.000

  Band   E0   <VxcLDA>  SigX  SigC(E0)   Z    dSigC/dE  Sig(E)  E-E0       E
    4  -0.428 -11.190 -12.443   0.749   0.771  -0.298 -11.578  -0.389  -0.816
    5   0.216 -10.251  -7.238  -2.794   0.784  -0.275 -10.080   0.171   0.387

 E^0_gap          0.643
 E^GW_gap         1.203
 DeltaE^GW_gap    0.560

--------------------------------OUTPUT FILE ENDS-------------------------


 

 Thank you.

 Regards,
 Sankeerth Rajalingam
 Graduate Research Assistant

 




---------------------------------------------------------------------------------------------
Dr. Zeila Zanolli

Université Catholique de Louvain (UCL)
Unité Physico-Chimie et de Physique des Matériaux (PCPM) 
Place Croix du Sud, 1 (Boltzmann)
B-1348 Louvain-la-Neuve, Belgium
Phone: +32 (0)10 47 3501 
Mobile: +32 (0)487 556699
Fax: +32 (0)10 47 3452
---------------------------------------------------------------------------------------------







Archive powered by MHonArc 2.6.16.

Top of Page