Anomalous Decay of the matrix element and the Very large electron-phonon coupling

[qiaoqiao66]

Deal all,
I am currently trying to compute the electron-phonon coupling of some compounds. The test-suite under the root directory of Quantum ESPRESSO were tested successfully, and other compounds have also been calculated successfully. However, There have been some unusual problems in the recent calculations.
1. The calculated electron-phonon coupling strength is far too large.
Electron-phonon coupling strength = 969.4177308

Estimated Allen-Dynes Tc = 54.853462 K for muc = 0.10000

Estimated BCS superconducting gap = 0.008319 eV

2. I have plot the decays of the Hamiltonian, Dynamical matrix, and matrix elements, results are as follows. Is the turn in the figure unusual？

3. I have checked the electronic and phonon band structure band structure after wannier interpolation. And the band structure and phonon dispersion from wannier interpolation match well with those of QE. Results are as follows.


My questions is:
What can lead to such behavior? I don't believe the parameters of my calculations are unreasonable, maybe there is some principle error or some kind of clue that can point on.

The input file for the final calculation is:
--
&inputepw
prefix = 'Ta'
amass(1)=28.0855,
amass(2)=14.0067,
amass(3) =180.95
outdir = './'
dvscf_dir = './save'
asr_typ = 'crystal'
!lifc=.true.

etf_mem = 0
elph = .true.
kmaps = .false.
epbwrite = .true.
epbread = .false.
epwwrite = .true.
epwread = .false.
lphase = .true.
use_ws =.true.

system_2d=.true.

vme = .false.

wannierize = .true.
nbndsub =12
bands_skipped = 'exclude_bands = 1:20'
num_iter = 300

dis_froz_max= -4
proj(1)='Si:s'
proj(2) ='N:s'
proj(3)='Ta:s'
proj(4)='Ta:d'
wdata(1) = 'bands_plot = .true.'
wdata(2) = 'begin kpoint_path'
wdata(3) = 'G 0.00 0.00 0.00 Y 0.50 0.00 0.00'
wdata(4) = 'Y 0.50 0.00 0.00 M 0.3333333333 0.3333333333 0.3333333333'
wdata(5) = 'M 0.3333333333 0.3333333333 0.3333333333 X 0 0.00 0.00'
wdata(6) ='end kpoint_path'
wdata(7) ='guiding_centres=.true.'
wdata(8) = 'use_ws_distance = T'
wdata(9) = 'bands_num_points =50'


fsthick = 1
degaussw = 0.01

iverbosity=0

delta_smear = 0.01 ! eV
degaussq = 0.2! meV
nqstep = 500
ep_coupling = .true.
ephwrite = .true.
eliashberg = .true.
!laniso = .true.
!imag = .true.
!lpade = .true.
!nsiter = 500
conv_thr_iaxis = 1.0d-5
wscut = 1 ! eV
muc = 0.1
mp_mesh_k = .true.
nkf1 = 200
nkf2 = 200
nkf3 = 1
nqf1 = 20
nqf2 = 20
nqf3 = 1

nk1 = 10
nk2 = 10
nk3 =1
nq1 = 10
nq2 = 10
nq3 = 1
/

[hpaudya1]

Hi qiaoqiao66,

I also had a same issue few years back, and later I found that the problem was because of a .dvscf_qX file for a specific q-point. I solved it by re-running a phonon calculation for that specific q-point. In your case, I do not know which q-point has as issue, so I would recommend you to re-run your full phonon and use new set of .dyn, dvscf, and patterns files.

Hope this helps.

Best,
Hari

[qiaoqiao66]

Dear Hari，
Thank you very much for your reply and your suggestions. I have re-running a full phonon phonon calculation with different nq, then use new set of .dyn, dvscf, and patterns files re-running the epw calculation. But the problem in the above email still appears. Additionally, I also tried to do it again with a different pseudo potential, the problem still exists. But may I ask you, how did you find the faulty .dvscf_qX file for a specific q-point?

Thank you very much again for your reply!

Best,
Zhihong Yuan

[hlee]

Dear Zhihong Yuan:

system_2d=.true.

In addition to Hari's answer, I would like to emphasize again that the current version of EPW doesn't support the 2D case; I am not 100% sure, but it might be related to your current issue. Probably, 25 Ang might be related to the cell length along the surface perpendicular direction.

Sincerely,

H. Lee

[qiaoqiao66]

Dear H. Lee，
Thank you very much for your reply and your suggestions. I have also calculated a 2D system before, and successfully obtained the electron-phonon coupling strength.
Recently, I meet the similar problem in a 3D system.

Sincerely,

ZhiHong Yuan

[hlee]

Dear ZhiHong Yuan:

Even though you obtained some numbers, officially the current version of EPW doesn't support the 2D case (viewtopic.php?f=3&t=1311) and the implementation related to the input of system_2d is not complete.

Sincerely

H. Lee

[qiaoqiao66]

Dear H. Lee,

Thank you very much for your reminding. Then I will use it to do the calculation of 3D system.

But in the calculation of 3D system, I have also encountered the Anomalous Decay of the matrix element, which have shown in the last email, what could be the cause of this phenomenon? Thank you very much!

Sincerely

ZhiHong Yua

[hlee]

Dear ZhiHong Yuan:

I think that the tendency is different.

For the 2D case, as you can see in all four plots, the shrinked versions of decaying behaviors are repeated from the same position of 25 Ang.
But, in the plots for the 3D case, we can't see this tendency; I think that you need to find its origin in your Wannierization.

Sincerely,

H. Lee

[qiaoqiao66]

Dear H. Lee:

Thank you for your reply and suggestions.
I have checked the Wannierization, the Wannier Function centers and spreads are displayed below. The band structure from wannier interpolation match well with those of QE, which also is shown below.

Sincerely,
ZhiHong Yuan

[mxh]

Hi qiaoqiao66,

I also have a same issue. The electron-phonon coupling strength is very strange, about 1000. I adjusted "fsthick", wscut" and" muc", and the electro-acoustic coupling was reduced to 283. Obviously, this value is still very strange. Have you solved your problem? How do you solve it?
Thank you in advance for your reply.

Best,
mxh