Problems with EPW phonon dispersion

[jhein]

Dear all,

I'm currently trying to calculate electron-phonon-coupling in lithium azide, but I'm already running into problems when comparing the phonon dispersion gained from Quantum Espresso with the results obtained by the EPW calculation at the Gamma point (unusual softening of the lowest
acoustic mode) which does not appear in other calculations.

qevsepw.png (47.62 KiB) Viewed 5756 times

Gamma is at ~0.85

I have tried other paths and a higher point density along the path with the strange softening still appearing around Gamma.
Is this a problem in the interpolation?
And is there a way to fix it?

I would be grateful for any help!

Cheers
Jakob

scf input:

Code: Select all

&control
calculation = 'scf',
restart_mode='from_scratch',
prefix = 'lin3',
pseudo_dir = './',
outdir = './',
verbosity='high',
/
&system
ibrav = 0,
celldm(1)= 10.63348890,
nat=4,
ntyp=2,
ecutwfc=60,
ecutrho=480,
nbnd=26
/
&electrons
conv_thr=1.0d-10
/
CELL_PARAMETERS alat
   0.497678237  -0.295480812  -0.001804250
   0.497678237   0.295480812  -0.001804250
  -0.256137630  -0.000000000   0.846376674
ATOMIC_SPECIES
 Li  6.9675  Li.pbe-s-kjpaw_psl.1.0.0.UPF
   N   14.00650   N.pbe-n-kjpaw_psl.1.0.0.UPF
ATOMIC_POSITIONS crystal
N             0.6075314159        0.6075314159        0.7438150802
N             0.3924685841        0.3924685841        0.2561849198
N             0.5000000000        0.5000000000        0.5000000000
Li           -0.0000000000       -0.0000000000       -0.0000000000
K_POINTS automatic
7 7 5 1 1 1

ph input:

Code: Select all

&inputph
prefix='lin3',
fildyn='lin3.dyn',
fildvscf='dvscf',
ldisp=.true.,
outdir='./',
nq1=3,
nq2=3,
nq3=3,
tr2_ph=1.0d-12
/

epw input:

Code: Select all

&inputepw
  prefix      = 'lin3'
  amass(1)    = 6.9675
  amass(2)    = 14.00650
  outdir      = './'

  iverbosity  = 0

  elph        = .true.
  epbwrite    = .true.
  epbread     = .false.

  epwwrite    = .true.
  epwread     = .false.

  nbndsub     =  13

  wannierize  = .true.
  num_iter    = 500
  iprint      = 2
  dis_win_max = 16  
  dis_froz_max= 13.3  
  wdata(1)= 'dis_num_iter=600'
  proj(1) = 'N:s,p'
  proj(2) = 'Li:s'

  elecselfen  = .false.
  phonselfen  = .true.
  a2f         = .false.

  band_plot = .true.
  wannier_plot= .true.

  temps       = 0.01 
  degaussw    = 0.1 ! eV

  dvscf_dir   = '../ph/save'
  filukk      = './lin3.ukk'
  filqf       = 'path.dat'
 
  nk1         = 6
  nk2         = 6
  nk3         = 6

  nq1         = 3
  nq2         = 3
  nq3         = 3
 /

[stiwari]

Hi Jakob,

It seems that your phonon grid is a bit coarse (3X3X3) which could lead to not so great Wannierization, leading to a not so great fine-grid interpolation (phonon dispersion). It will be good if you can increase the coarse grid (in ph.in) to 6X6X6 for better interpolation.

Best regards,
Sabya.