Question about EPW 4.0 for 2D materials

[leedairo]

Dear Dr. Samuel,

I want to calculate electronic self energy for some 2D materials using EPW 4.0.
But, I cannot get resonable result for 2D materials.

First time, I try to run this code using example (EPW/testst/Inputs/t01), I could get same result compare to Refs.
And, I run this code for a bulk silicon, I also get resonable result.
(I can plot a correct band structure using linewidth.elself file in both cases)

But, I test this code for a graphene, I got unphysical result.
(Also I can plot a band structure using linewidth.elself file. But the band structure is wrong)

I checked my output file, I find that "WF centre and spread" is not converged, in graphene.wout file
I try to some test changing input option (proj, dis_mix_ratio), but I cannot get converged value.

I don't know what is wrong in my input file.
Could you please give me some tips or example using EPW4.0 for 2D materials?

With best regards,
Seungjun Lee.

Code: Select all

##############  graphene.wout  ##################

+--------------------------------------------------------------------+<-- CONV
 | Iter  Delta Spread     RMS Gradient      Spread (Ang^2)      Time  |<-- CONV
 +--------------------------------------------------------------------+<-- CONV
      0     0.702E+03     0.0000000000      701.8628183025      19.23  <-- CONV
      1    -0.678E+02    83.9381167195      634.0914986042      19.24  <-- CONV
      2    -0.494E+02    60.0839851770      584.6450389953      19.25  <-- CONV
      3    -0.381E+02    63.3536083613      546.5396878428      19.25  <-- CONV

...

    297     0.368E+01   147.4933146472      221.8698468338      20.92  <-- CONV
    298     0.430E+00    62.9800156447      222.2996362739      20.93  <-- CONV
    299    -0.218E+01    41.2376546125      220.1232541305      20.94  <-- CONV
    300    -0.170E+01    20.5509863985      218.4280169276      20.94  <-- CONV

###############  epw_scf.in ##################

 &control

    calculation     = 'scf'

    prefix          = 'graphene'

    restart_mode    = 'from_scratch'

    wf_collect      = .false.

    pseudo_dir      = './'

    outdir          = './'

    tprnfor         = .true.

    tstress         = .true.

 /

 &system

    ibrav           = 4

    a               = 2.467625

    c               = 20

    nat             = 2

    ntyp            = 1

    ecutwfc         = 60

    occupations     = 'smearing'

    smearing        = 'mp'

    degauss         = 0.02

    nbnd            = 4

 /

 &electrons

    diagonalization = 'david'

    mixing_beta     = 0.7

    conv_thr        = 1.0d-20

 /

ATOMIC_SPECIES

  C  12.01078  C.pbe-nc.UPF

ATOMIC_POSITIONS crystal

  C   0.3333333333  0.6666666666  0.00

  C   0.6666666666  0.3333333333  0.00

K_POINTS automatic

6 6 1 0 0 0 

 

###############  epw.in ##################

--

&inputepw

  prefix      = 'graphene'

  amass(1)    = 12.01078

  outdir      = './'




  iverbosity  = 0




  elph        = .true.

  epbwrite    = .true.

  epbread     = .false.




  epwwrite    = .true.

  epwread     = .false.




  nbndsub     =  8

  nbndskip    =  0




  wannierize  = .true.

  num_iter    = 300

  iprint      = 2

!  dis_win_max = 12

!  dis_froz_max= -9

  proj(1)     = 'C:sp2;pz'

  wdata(1)    =  'dis_mix_ratio = 0.2'

 

  elinterp    = .true.

  phinterp    = .true.




  tshuffle2   = .true.

  tphases     = .false.




  elecselfen  = .ture.

  phonselfen  = .false.

  a2f         = .false.




  parallel_k  = .true.

  parallel_q  = .false.




  fsthick     = 15 ! eV 

  eptemp      = 300 ! K (same as PRB 76, 165108)

  degaussw    = 1.0! eV




  dvscf_dir   = './save'

  filukk      = './graphene.ukk'

  filkf       = './path.dat'

 

  nqf1        = 30

  nqf2        = 30

  nqf3        = 1

  

  nk1         = 6

  nk2         = 6

  nk3         = 1




  nq1         = 6

  nq2         = 6

  nq3         = 1

 /

      7 cartesian

           0.0000000   0.0000000   0.0000000   0.0555556

           0.0000000   0.1924501   0.0000000   0.3333333

           0.0000000   0.3849002   0.0000000   0.3333333

           0.0000000  -0.5773503   0.0000000   0.1666667

           0.1666667   0.2886751   0.0000000   0.3333333

           0.1666667   0.4811252   0.0000000   0.6666667

           0.3333333   0.5773503   0.0000000   0.1111111

[sponce]

Dear Seungjun,

Thank you for posting your question on the forum.

It seems you problem is linked with Wannier90.

I personally have never done 2D calculations yet. Therefore I'm not sure if something specific needs to be used within Wannier.

To solve this issue you should probably try to run your calculations using Wannier90 separately and when you found the correct Wannier parameters you can input them
into EPW using the wdata input variable. Having correctly locallized Wannier function for your system is a requirement before using EPW.

Wannier90 can be downloaded directly from QE by issuing make w90 or from their website http://www.wannier.org/.

Best,

Samuel

[roxana]

Dear Seungjun,

As Samuel pointed out your problem is linked with Wannier90.

In the Wannier90 code there is a very good example (example10) on graphite that can be used as a starting point.
You can also find a discussion on this topic on the Wannier90 forum

http://mailman.qe-forge.org/pipermail/w ... 00541.html

Below is a set of parameters that you can use as a starting point with Wannier90 or EPW

nbndsub = 7
nbndskip = 0

wannierize = .true.
num_iter = 1000
dis_win_max = 20.0
dis_froz_max= 3.2
proj(1) = 'f=0.5,0.5,0.0:l=0,mr=1'
proj(2) = 'f=0.5,1.0,0.0:l=0,mr=1'
proj(3) = 'f=0.0,0.5,0.0:l=0,mr=1'
proj(4) = 'f=0.333333333,0.666666667,0.0:l=1,mr=1'
proj(5) = 'f=0.666666667,0.333333333,0.0:l=1,mr=1'
proj(6) = 'f=0.0,0.0,0.157:l=0,mr=1'
proj(7) = 'f=0.0,0.0,-0.157:l=0,mr=1'
wdata(1) = 'dis_num_iter = 2000'
wdata(2) = 'dis_mix_ratio = 1.0'
wdata(3) = 'guiding_centres = .true.'

Best,
Roxana

[leedairo]

Thank you for kind reply.

But, I cannot solve this problem yet,

I check my output file and test agian.

1) I try to run your calculations using Wannier90 separately on graphene (Input parameter choose by slightly modified example10 in wannier90 code).
I can get resonable value. (correct band structure and converged "WF centre and spread" value)

2) I try to run EPW 4.0 using same input compare to 1), but I cannot get resonable value.

3) I try to EPW 4.0 using Roxana's advice, but I also failed.


I guess that large vacuum distance is cause of this problem.

I compared two result 1) and 2), the value of "Final Omega_I" has a big difference between two results.
I think that "Final Omega_I" and "WF spread" are too large in my case, I cannot get converged value.

Could you please check my problem agian?

With best regards,
Seungjun Lee.


####################################
In my case, unit vector (graphene)
Lattice Vectors (Ang)
a_1 2.467625 0.000000 0.000000
a_2 -1.233813 2.137026 0.000000
a_3 0.000000 0.000000 20.000000 << vacuum distance
#####################################
"Final Omega_I" value in graphene.wout file

using wannier90
- vacuum distance 20 ~ Final Omega_I 2.50620629 (Ang^2)
- vacuum distance 10 ~ Final Omega_I 2.38801114 (Ang^2)

using EPW 4.0
- vacuum distance 20 ~ Final Omega_I 91.34943611 (Ang^2)
- vacuum distance 10 ~ Final Omega_I 53.98999704 (Ang^2)
- vacuum distance 8 ~ Final Omega_I 49.48208581 (Ang^2)

##############################################
& input of wannier parameter in epw.in

nbndsub = 5
nbndskip = 3
wannierize = .true.
num_iter = 2000
iprint = 2
proj(1) = 'f=0,0,0.5:pz'
proj(2) = 'f=0.33333333,0.66666667,0.5:sp2;pz'
wdata(1) = 'guiding_centres = true'

##############################

using EPW 4.0 - vacuum distance 20 ~ Final Omega_I 91.34943611 (Ang^2)

graphene.wout file

I think that "Spread (Ang^2)" is too large and unphysical value in my case

*------------------------------- WANNIERISE ---------------------------------*
+--------------------------------------------------------------------+<-- CONV
| Iter Delta Spread RMS Gradient Spread (Ang^2) Time |<-- CONV
+--------------------------------------------------------------------+<-- CONV

------------------------------------------------------------------------------
Initial State
WF centre and spread 1 ( 0.234285, 0.180038, 9.545852 ) 81.42372867
WF centre and spread 2 ( -0.532017, 0.891559, 10.264635 ) 68.77537924
WF centre and spread 3 ( 0.137071, 1.634878, 11.339782 ) 67.21109473
WF centre and spread 4 ( -0.129081, 1.052533, 10.134970 ) 72.93287358
WF centre and spread 5 ( -0.086439, 2.509243, 10.066337 ) 76.95293358
Sum of centres and spreads ( -0.376181, 6.268251, 51.351576 ) 367.29600979

0 0.367E+03 0.0000000000 367.2960097475 14.01 <-- CONV
O_D= 275.1960238 O_OD= 0.7505499 O_TOT= 367.2960097 <-- SPRD

....


Final State
WF centre and spread 1 ( 6.736970, 1.328636, 2.084576 ) 23.42472151
WF centre and spread 2 ( 0.514513, -1.170222, -9.895328 ) 19.36693459
WF centre and spread 3 ( 1.487075, -1.089430, -3.327999 ) 22.74290625
WF centre and spread 4 ( 5.287497, -4.539413, -8.454906 ) 21.85244158
WF centre and spread 5 ( 1.831841, 1.652049, 6.767832 ) 21.44244409
Sum of centres and spreads ( 15.857896, -3.818380,-12.825825 ) 108.82944802

Spreads (Ang^2) Omega I = 91.349436109
================ Omega D = 16.864229949
Omega OD = 0.615781535
Final Spread (Ang^2) Omega Total = 108.829447593

###########################################################

using wannier90 - vacuum distance 20 ~ Final Omega_I 2.50620629 (Ang^2)

graphene.wout

*------------------------------- WANNIERISE ---------------------------------*
+--------------------------------------------------------------------+<-- CONV
| Iter Delta Spread RMS Gradient Spread (Ang^2) Time |<-- CONV
+--------------------------------------------------------------------+<-- CONV

------------------------------------------------------------------------------
Initial State
WF centre and spread 1 ( 0.355888, 1.843582, 5.000000 ) 0.49500717
WF centre and spread 2 ( 0.355887, 0.616419, 5.000000 ) 0.49500889
WF centre and spread 3 ( 1.418643, 1.230001, 5.000000 ) 0.49500885
WF centre and spread 4 ( 0.710141, 1.230000, 5.000000 ) 0.80215312
WF centre and spread 5 ( 0.000000, 0.000000, 5.000000 ) 0.80216772
Sum of centres and spreads ( 2.840559, 4.920002, 25.000000 ) 3.08934574

....

Writing checkpoint file graphene.chk... done

Final State
WF centre and spread 1 ( 0.355710, 1.843890, 5.000000 ) 0.46786416
WF centre and spread 2 ( 0.355711, 0.616112, 5.000000 ) 0.46786420
WF centre and spread 3 ( 1.418997, 1.230000, 5.000000 ) 0.46786415
WF centre and spread 4 ( 0.710141, 1.230000, 5.000001 ) 0.80215045
WF centre and spread 5 ( 0.000000, 0.000000, 5.000000 ) 0.80216452
Sum of centres and spreads ( 2.840560, 4.920001, 25.000000 ) 3.00790749

Spreads (Ang^2) Omega I = 2.388011141
================ Omega D = 0.002921331
Omega OD = 0.616975020
Final Spread (Ang^2) Omega Total = 3.007907492

[sponce]

Dear Seungjun Lee,

If you can get reasonable values with Wannier90, you should also be able to get the same with EPW.

Indeed the EPW code calls Wannier through a library.

You should therefore have the exact same results for the Wannier part.

Be sure that you have the same input though.

[anindya@iiita]

Dear Seungjun Lee,
I am facing the same problem with graphene.In the wannierized band structure a small amount of band gap is seen at the K point.I don't know how to remove the gap.Did you solve the problem?Please let me know how did you do it?

Thanks and regards,
Anindya Bose