[QE-users] incorrect LO-TO splitting for CaO and MgO

[Casey Brock]

I am using pw.x -> ph.x -> dynmat.x to get the LO and TO phonon frequencies at gamma for rock-salt CaO. The value for the LO phonon I am getting differs from experiment and other DFPT calculations in the literature by 5 THz (167 cm^-1). Perhaps this is a problem with Quantum Espresso, but it’s very possible I’m doing something wrong. I’ve included results and input files below, and any advice would be greatly appreciated.

The TO and LO frequencies in THz are given below. Notice that Quantum Espresso using DFPT gives a value that is too low.
experiment: 8.9, 17.1
frozen phonon (Phonopy + Quantum Espresso): 8.71, 17.1
DFPT (Quantum Espresso, pw.x->ph.x->dynmat.x):  8.53, 12.2

If I use Phonopy (frozen phonon technique) combined with Quantum Espresso (pw.x and ph.x) I get the correct splitting at gamma. I have tried the same calculation with MgO and I see the same problem, where the frozen phonon technique works while DFPT does not. Here are some things I have tried: different PAW pseudopotentials, increasing ecutwfc and ecutrho, increasing k-points, reducing conv_thr, and reducing tr2_ph.

Thanks in advance,
Casey Brock

input to pw.x
-------------------------------------------
&control
    calculation='scf'
    restart_mode='from_scratch'
    prefix='cao'
    pseudo_dir = './'
    outdir='./output/'
    tprnfor = .true.
    tstress = .true.
/
&system
    ibrav = 0
    nat = 2
    ntyp = 2
    ecutwfc = 70.0
    ecutrho = 280.0
/
&electrons
   conv_thr =  1.0d-10
   mixing_beta = 0.7
/
K_POINTS {automatic}
 12 12 12 0 0 0

CELL_PARAMETERS bohr
 4.45 4.45 0
 0 4.45 4.45
 4.45 0 4.45
ATOMIC_SPECIES
 O  16.00 O.pz-n-kjpaw_psl.0.1.UPF
 Ca 40.08 Ca.pz-spn-kjpaw_psl.1.0.0.UPF
ATOMIC_POSITIONS crystal
Ca  0.0  0.0  0.0
O   0.5  0.5  0.5

input to ph.x
---------------------------------------
Phonon for CaO
&inputph
 prefix='cao',
 outdir='./output/',
 fildyn='dyn.G'
 epsil = .true.
 tr2_ph=1.0d-14,
/
0 0 0

input to dynmat.x
---------------------------------------
&input
 fildyn = 'dyn.G'
 asr = 'simple'
 q(1)=0.0, q(2)=0.0, q(3)=1.0
/

-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.quantum-espresso.org/pipermail/users/attachments/20190327/5eb63315/attachment.html>