Electron-phonon matrix elements with spin orbit coupling (SO
Posted: Wed Jul 12, 2017 5:10 pm
Dear EPW developers,
I would like to ask a question concerning the calculation of electron-phonon matrix elements with spin-orbit coupling (SOC).
I have calculated with and without SOC (denoted as SOC and NSOC) the electron-phonon matrix elements interpolated along the high-symmetry k-point path for the cubic symmetry.
The matrix elements for NSOC diverge as expected for q -> 0 for the long range component.
However, this was not the case for matrix elements with SOC. Instead, for q -> 0, the matrix elements are suppressed to zero, as if the long range component was not included.
Please find a plot with the magnitude of the electron-phonon matrix elements at the k = Gamma point for averaged over transitions involving the 3-fold degenerate bands here:
I have separately checked the norm-conserving pseudopotentials used against other pseudopotentials and other codes, and found them to be consistent with each other in terms of effective band masses and the SO slitting of the bands.
I have also checked the wannierization for NSOC and SOC and find the band structure to agree well with the original QE calculations.
Please find below the input files for running scf, nscf, and epw.
Could you advise on where a possible error might be (e.g., input) or if this is a known issue?
Thank you for your time.
Sincerely,
Wennie
------------------------------------------------------------------------------------------------------
epw.in
------------------------------------------------------------------------------------------------------
------------------------------------------------------------------------------------------------------
scf.in
------------------------------------------------------------------------------------------------------
I would like to ask a question concerning the calculation of electron-phonon matrix elements with spin-orbit coupling (SOC).
I have calculated with and without SOC (denoted as SOC and NSOC) the electron-phonon matrix elements interpolated along the high-symmetry k-point path for the cubic symmetry.
The matrix elements for NSOC diverge as expected for q -> 0 for the long range component.
However, this was not the case for matrix elements with SOC. Instead, for q -> 0, the matrix elements are suppressed to zero, as if the long range component was not included.
Please find a plot with the magnitude of the electron-phonon matrix elements at the k = Gamma point for averaged over transitions involving the 3-fold degenerate bands here:
I have separately checked the norm-conserving pseudopotentials used against other pseudopotentials and other codes, and found them to be consistent with each other in terms of effective band masses and the SO slitting of the bands.
I have also checked the wannierization for NSOC and SOC and find the band structure to agree well with the original QE calculations.
Please find below the input files for running scf, nscf, and epw.
Could you advise on where a possible error might be (e.g., input) or if this is a known issue?
Thank you for your time.
Sincerely,
Wennie
------------------------------------------------------------------------------------------------------
epw.in
------------------------------------------------------------------------------------------------------
Code: Select all
--
&inputepw
prefix = 'WO3'
amass(1) = 183.84
amass(2) = 15.999
outdir = './'
elph = .true.
ep_coupling = .true.
kmaps = .true.
epbwrite = .false.
epbread = .false.
etf_mem = .false.
lpolar = .true.
!! ww add for printing out matrix elements
lgelph = .true.
lfmt = .true.
epwwrite = .false.
epwread = .true.
!ephwrite = .true. ! write el-ph elements of fine mesh to file, unformatted
nbndsub = 42
nbndskip = 0
wannierize = .false.
num_iter = 500
dis_win_min =-75.000
dis_win_max = 20.000
dis_froz_min= -5.000
dis_froz_max = 10.000
!dis_win_max = 11
!dis_froz_max= 11
proj(1) = 'W:s,p,d'
proj(2) = 'O:s,p'
wdata(1) = 'bands_plot=.true.'
wdata(2) = 'spinors=.true.'
wdata(3) = 'begin kpoint_path'
wdata(4) = 'G 0.0000 0.0000 0.0000 X 0.000 0.5000 0.0000'
wdata(5) = 'X 0.0000 0.5000 0.0000 M 0.500 0.5000 0.0000'
wdata(6) = 'M 0.5000 0.5000 0.0000 G 0.000 0.0000 0.0000'
wdata(7) = 'G 0.0000 0.0000 0.0000 R 0.500 0.5000 0.5000'
wdata(8) = 'R 0.5000 0.5000 0.5000 X 0.000 0.5000 0.0000'
wdata(9) = 'end kpoint_path'
elecselfen = .false. ! scattering rate -> ~ tau
phonselfen = .false.
a2f = .false.
vme = .false. ! .true. does not work
parallel_k = .true.
parallel_q = .false.
fsthick = 5.0
eptemp = 300.0
degaussw = 0.1
efermi_read = .true.
fermi_energy= 6.7 !approx mid-gap; VBM = 6.1500, CBM = 6.8526
dvscf_dir = './save'
filqf = 'testqpt'
filkf = 'testkpt'
nk1 = 6
nk2 = 6
nk3 = 6
nq1 = 6
nq2 = 6
nq3 = 6
/
20 cartesian
0.000000000 0.000000000 0.000000000 1
...
------------------------------------------------------------------------------------------------------
scf.in
------------------------------------------------------------------------------------------------------
Code: Select all
&CONTROL
calculation = 'scf'
restart_mode = 'from_scratch',
prefix = 'WO3',
tstress = .true.
tprnfor = .true.
pseudo_dir = './',
outdir = './'
wf_collect = .true.
verbosity = 'high'
/
&SYSTEM
ibrav = 1,
celldm(1) = 7.1749995265,
nat = 4,
ntyp = 2,
ecutwfc = 125.00,
nbnd = 48,
lspinorb = .true.,
noncolin = .true.
la2f = .true.
/
&ELECTRONS
diagonalization = 'cg'
mixing_mode = 'plain'
mixing_beta = 0.3
conv_thr = 1.0d-10
diago_full_acc = .true.
/
&IONS
ion_dynamics = 'damp'
/
&CELL
cell_dynamics = 'bfgs'
/
ATOMIC_SPECIES
W 183.84 W_frl.upf
O 15.999 O_frl.upf
ATOMIC_POSITIONS (crystal)
W 0.000000000 0.000000000 0.000000000
O 0.500000000 0.000000000 0.000000000
O 0.000000000 0.500000000 0.000000000
O 0.000000000 0.000000000 0.500000000
K_POINTS automatic
6 6 6 0 0 0