[QE-users] Acuostic Sum Rule:-Regarding

Lorenzo Monacelli lorenzo.monacelli at roma1.infn.it
Sun Sep 20 11:55:00 CEST 2020


Hi,

I believe that to claim if a structure is stable or not it is unsafe to 
rely on the Fourier interpolation (q2r and PlotPhon code), specially 
when they depend so much on the application of the sum rule. Instead, 
you should run a calculation on the specific q point that is unstable. 
This calculation, unless at Gamma where ASR is trivial on the acoustic 
modes, is unaffected by ASR.

If you find differences between ASR in the interpolation, try increasing:

1) The parameters as tr2_ph (I usually use 1d-18 but it may depend on 
the system)  and conv_thr in the pw (I never use values bigger than 
1d-14 for phonons, sometimes even smaller values are required)

2) the Q grid in which you ran the calculation is too coarse grain, this 
creates problems in the Fourier transform operated by q2r, try to use 
finer meshes.

One of the strenght of quantum-espresso PHonon package is that it can 
compute the phonons at any q point without using a commensurate supercell.

So, to indentify an instability you should see an imaginary mode 
directly in the first output of the phonon package, not only in the 
interpolated band structure (you can grep the 'freq' keyword on the 
output dynamical matrix, remember that you will have 3 acoustic modes at 
gamma with random values, you can use them to check the convergence of 
the parameters).

So if you find a q point at which you have an imaginary frequency after 
the Fourier interpolation, and this q point was not present in the 
original q mesh you gave, try running a ph.x with just that q point.

Moreover, pay attention to the structure and cell relaxation before 
running the phonons, in many cases I found systems in which you must 
specify a forc_conv_thr much smaller than the default value (e.g. 1d-5) 
to avoid fake imaginary modes appearing in the calculation.

Bests,

Lorenzo


Il 20/09/20 15:16, Suresh A ha scritto:
> Dear All,
>              I am asking this question to know whether I have made a 
> mistake in my calculations during my run using phonon code in pwscf. I 
> also went through previous posts but could not understand as I am not 
> good at phonon theory.
> I have run calculations for three alloys. Two of them give positive 
> frequencies for  zasr='crystal' as well as for zasr='simple' but for 
> one alloy it gives negative frequency for zasr='simple' and positive 
> frequency for zasr='crystal'. Due to curiosity I just simply increased 
> the convergence threshold for phonon calculation alone (not in any 
> other part in scf) from  tr2_ph = 1.0d-12, to tr2_ph = 1.0d-14, Now 
> the results are reversed. I got positive frequency for zasr='crystal' 
> and negative frequency for zasr='simple. I want to ask two questions 
> from this,
>
> 1. May I claim my alloy is a stable one, because the previous report 
> says that it is an unstable structure?
> 2.I want to know whether I have made any mistake in my run?
> Following is my procedure I used for my calculations:
> 1. SCF
> 2. Phonon run
> 3. q2r with zasr = simple and zasr= crystal
> 4. PlotPhon code for postprocessing (I have run for other two alloys 
> for which there is no negative frequency for zasr= simple and 
> crystal.) Here I simply used this code immediately after  I ran q2r by 
> simply changing asr='crystal' to simple in the folder PlotPhon-> 
> Scripts-> Plot_run.
> Did i make any mistake here?
> Thanks and regards
> Yours Sincerely,
>
>  A.Suresh,
> Post-Doctoral Fellow
> Department of Physics
> National Institute of Technology
> Tiruchirappalli
> Tamil Nadu,
> India
>
> _______________________________________________
> Quantum ESPRESSO is supported by MaX (www.max-centre.eu/quantum-espresso)
> users mailing list users at lists.quantum-espresso.org
> https://lists.quantum-espresso.org/mailman/listinfo/users
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.quantum-espresso.org/pipermail/users/attachments/20200920/2e01205e/attachment.html>


More information about the users mailing list