SOC removes LO-TO splitting in GaN

[Vahid]

Dear Samuel,

I ran the phonon dispersion with qe-6.2.1 using both the matdyn.x code and EPW-4.3.0 for GaN example case. I used 6x6x6 coarse k/q grids. The PSP is given earlier on this thread. It is a fully relativistic PSP. Tags lpolar and epsil, lspinorb and noncolin are all set to true.

In the EPW input, I used lifc=.true. and asr_typ='simple'. I used 'simple' asr in matdyn.in and q2r.in as well and renamed the force constant file to ifc.q2r.xml in the save folder.

Here is a comparison of the two phonon dispersions (https://drive.google.com/file/d/1t-Mnw- ... q-v04/view). The EPW bands have higher energies especially for the upper optical modes. Is this behaviour expected?

Thanks,

Vahid

[sponce]

Hello,

Could you try asr_typ='crystal' ?

Also, the simple sum rule is the default one in EPW. Therefore if you do not need the crystal sum rule (usually better than simple), then you can just do
lifc=.false.

In principle lifc = false and lifc = .true. + asr_typ='simple' should get the same results.
There might be a bug if it is not the case.

Could you do those 2 additional calculations and tell us your results ?
Calculation 1: same but with asr_typ='crystal
Calculation 2: lifc = false and asr_typ does not need to be specified. You do not need ifc.q2r.xml file in this case.

Thanks,
Samuel

[Vahid]

Dear Samuel,

The setting asr_typ=crystal (https://drive.google.com/open?id=1F6Ell ... uFYdXAtM2A) and lifc=.false. (https://drive.google.com/open?id=1LzpJ0 ... XLreEdk4UZ) are consistent with the lifc=true and asr_typ=simple result reported earlier. In all three cases, EPW is consistently higher than matdyn. And in all three cases, EPW dispersion is the same.

I also tried asr=crystal in matdyn.in and q2r.in and the matdyn dispersion is the same for both asr settings.

Best,

Vahid

[sponce]

Hello,

Are you sure you did not do a silly mistake and that the input are consistent ?

I suspect something like a mistake in the value "amass" for example.

This would results in such a shift up.

Best,
Samuel

[Vahid]

Hello,

I will try to debug the SOC part of the code. I will keep you posted as to what I find.

Best,

Vahid

[carla.verdi]

Hi Vahid,

Do you mean that if you turn off SOC in your GaN calculation you get the right phonon dispersion?

Carla

[Vahid]

Hi Carla,

I have not tried running GaN without SOC. I have run other polar materials without SOC and there were no discrepancies between matdyn and EPW phonon dispersion.

Fo this thread, I was merely trying to run on GaN with SOC. I checked to see if the phonon dispersions are consistent but they appear not to be. The masses are consistent in all the input files. So are the lattice parameters.

I will keep looking and will report back when I find something.

Best,

Vahid

[Vahid]

Hi Samuel and Carla,

I found out why the matdyn and EPW phonon dispersions did not match. In converting from meV to wavenumber, I used a factor of 0.12 whereas I should have used 0.12398. My bad!

Best,

Vahid

[sponce]

Dear Vahid,

Glad to hear that it now works for you.
Thanks for letting us know.

Best,
Samuel