[QE-users] Difference in phonon frequencies produced by matdyn.x

Stefano de Gironcoli degironc at sissa.it
Thu Aug 16 00:03:04 CEST 2018


I think the reason of the difference is the presence of macroscopic 
electric fields that makes the phonon dispersion close to gamma non 
-analytic (it depends on the direction  of approach to Gamma point). 
Accounting for this non-anayticity is the reason why one computes the 
effective charges and the dielectric matrix.

when you select the path (with matdyn.x) the code knows which direction 
you are approaching Gamma and includes the non analytic terms 
accordingly. you get therefore the LO-TO splitting of modes that would 
be otherwise degenerate (174.65). The TO mode stays there while the LO 
is shifted up to 322). The other modes are not changed because they 
don't have a macroscopic polarization.

When calculation the Dos the code does not know which direction to adopt 
and does not include the LO-TO splitting .. the TO matrix is 
diagonalized that have additional symmetry. This is in principle wrong 
but the gamma point is just a single point in the DoS. neighboring 
points and the rest of the points in the BZ should be computed correctly.

HTH

stefano



On 15/08/2018 23:24, Jie Peng wrote:
> Dear QE users:
>
> I have been running phonon calculations of HfS2 using quantum 
> espresso. I followed the steps of relaxing to the equilibrium atomic 
> configuration using pw.x, computing dynamical matrices on a k point 
> mesh grid using ph.x, producing interatomic force constant matrix 
> using q2r.x.
>
> Now I want to plot phonon dispersion along high symmetry direction 
> Gamma-M-K-Gmma in the HCP (hexagonal closed pack) HfS2 lattice. So I 
> used the following input file for matdyn.x:
>
> / &input/
> /    asr='crystal'/
> /    amass(1)=178.49,/
> /    amass(2)=32.065,/
> /    ntyp=2/
> /    flfrc='HfS2.fc', flfrq='HfS2.freq', q_in_band_form=.true.,/
> / //
> / 4/
> / 0 0 0 40/
> /0.5 -0.2887 0 40/
> /0.6667 0 0 40/
> /0 0 0 1/
>
> which gives me phonon frequencies at Gamma point as:
>
> /   diagonalizing the dynamical matrix .../
> /
> /
> / q =       0.0000      0.0000      0.0000/
> / **************************************************************************/
> /     freq (    1) =      -0.000000 [THz] = -0.000001 [cm-1]/
> / (  0.545987  -0.000000    -0.167781  -0.000000 -0.084152  -0.000000   )/
> / (  0.545987  -0.000000    -0.167781  -0.000000 -0.084152  -0.000000   )/
> / (  0.545987  -0.000000    -0.167781  -0.000000 -0.084152   0.000000   )/
> /     freq (    2) =       0.000000 [THz] =  0.000004 [cm-1]/
> / (  0.166188   0.000000     0.552429   0.000000 -0.023176  -0.000000   )/
> / (  0.166188   0.000000     0.552429   0.000000 -0.023176  -0.000000   )/
> / (  0.166188   0.000000     0.552429   0.000000 -0.023176   0.000000   )/
> /     freq (    3) =       0.000000 [THz] =  0.000008 [cm-1]/
> / (  0.087255   0.000000    -0.002306   0.000000  0.570714   0.000000   )/
> / (  0.087255   0.000000    -0.002306   0.000000  0.570714   0.000000   )/
> / (  0.087255   0.000000    -0.002306   0.000000  0.570714   0.000000   )/
> /     freq (    4) =       5.235913 [THz] =  174.651242 [cm-1]/
> / (  0.123126   0.000000     0.213241   0.000000 -0.000000  -0.000000   )/
> / ( -0.342690  -0.000000    -0.593505  -0.000000  0.000000   0.000000   )/
> / ( -0.342690  -0.000000    -0.593505  -0.000000  0.000000   0.000000   )/
> /     freq (    5) =       7.799484 [THz] =  260.162777 [cm-1]/
> / ( -0.000000   0.000000     0.000000  -0.000000  0.000000   0.000000   )/
> / (  0.707104  -0.000013     0.001900  -0.000000 -0.000000  -0.000000   )/
> / ( -0.707104   0.000013    -0.001900   0.000000 -0.000000   0.000000   )/
> /     freq (    6) =       7.799484 [THz] =  260.162777 [cm-1]/
> / ( -0.000000  -0.000000    -0.000000  -0.000000 -0.000000   0.000000   )/
> / (  0.001900   0.000000    -0.707104  -0.000000  0.000000  -0.000000   )/
> / ( -0.001900  -0.000000     0.707104   0.000000  0.000000   0.000000   )/
> /     freq (    7) =       9.162102 [THz] =  305.614822 [cm-1]/
> / ( -0.000000   0.000000     0.000000   0.000000 -0.246235   0.000000   )/
> / (  0.000000  -0.000000    -0.000000   0.000000  0.685335  -0.000000   )/
> / (  0.000000   0.000000    -0.000000  -0.000000  0.685335   0.000000   )/
> /     freq (    8) =       9.170687 [THz] =  305.901189 [cm-1]/
> / ( -0.213241   0.000000     0.123126  -0.000000  0.000000  -0.000000   )/
> / (  0.593505  -0.000000    -0.342690   0.000000 -0.000000   0.000000   )/
> / (  0.593505  -0.000000    -0.342690   0.000000 -0.000000   0.000000   )/
> /     freq (    9) =      10.162710 [THz] =  338.991522 [cm-1]/
> / (  0.000000   0.000000    -0.000000  -0.000000 -0.000000  -0.000000   )/
> / ( -0.000000  -0.000000     0.000000   0.000000 -0.707107   0.000000   )/
> / ( -0.000000  -0.000000    -0.000000   0.000000  0.707107   0.000000   )/
> / **************************************************************************/
>
> However, when I tried to compute phonon DOS in which a mesh grid 
> rather than a list of high symmetry kpoints was declared, a different 
> set of Gamma point phonon frequencies appeared. The input file for DOS 
> calculation is:
>
> / &input/
> /    asr='crystal'/
> /    amass(1)=178.49,/
> /    amass(2)=32.065,/
> /    ntyp=2/
> /    flfrc='HfS2.fc', flfrq='HfS2_DOS.freq'/
> /    dos=.true./
> /    fldos='HfS2.dos'/
> /    deltaE=3/
> /    nk1=30, nk2=30, nk3=30,/
> / //
>
> The Gamma point frequencies are:
>
> /diagonalizing the dynamical matrix .../
> /
> /
> / q =       0.0000      0.0000      0.0000/
> / **************************************************************************/
> /     freq (    1) =      -0.000000 [THz] = -0.000004 [cm-1]/
> / (  0.025759  -0.000000    -0.007450  -0.000000 -0.576727  -0.000000   )/
> / (  0.025759  -0.000000    -0.007450  -0.000000 -0.576727  -0.000000   )/
> / (  0.025759  -0.000000    -0.007450  -0.000000 -0.576727   0.000000   )/
> /     freq (    2) =      -0.000000 [THz] = -0.000003 [cm-1]/
> / ( -0.568506   0.000000    -0.097733   0.000000 -0.024130  -0.000000   )/
> / ( -0.568506   0.000000    -0.097733   0.000000 -0.024130  -0.000000   )/
> / ( -0.568506   0.000000    -0.097733   0.000000 -0.024130   0.000000   )/
> /     freq (    3) =       0.000000 [THz] =  0.000005 [cm-1]/
> / ( -0.097317  -0.000000     0.568969  -0.000000 -0.011697  -0.000000   )/
> / ( -0.097317  -0.000000     0.568969  -0.000000 -0.011697  -0.000000   )/
> / ( -0.097317  -0.000000     0.568969  -0.000000 -0.011697   0.000000   )/
> /     freq (    4) =       5.235913 [THz] =  174.651242 [cm-1]/
> / ( -0.000025   0.000000     0.246235   0.000000 -0.000000  -0.000000   )/
> / (  0.000070  -0.000000    -0.685335  -0.000000  0.000000   0.000000   )/
> / (  0.000070  -0.000000    -0.685335  -0.000000  0.000000   0.000000   )/
> /     freq (    5) =       5.235913 [THz] =  174.651242 [cm-1]/
> / (  0.246234   0.000623     0.000025   0.000000  0.000000  -0.000000   )/
> / ( -0.685333  -0.001733    -0.000070  -0.000000 -0.000000  -0.000000   )/
> / ( -0.685333  -0.001733    -0.000070  -0.000000  0.000000   0.000000   )/
> /     freq (    6) =       7.799484 [THz] =  260.162777 [cm-1]/
> / (  0.000000  -0.000000     0.000000   0.000000  0.000000   0.000000   )/
> / (  0.707105  -0.000010     0.001673  -0.000000 -0.000000  -0.000000   )/
> / ( -0.707105   0.000010    -0.001673   0.000000 -0.000000   0.000000   )/
> /     freq (    7) =       7.799484 [THz] =  260.162777 [cm-1]/
> / (  0.000000  -0.000000    -0.000000  -0.000000 -0.000000   0.000000   )/
> / (  0.001673   0.000000    -0.707105  -0.000000  0.000000  -0.000000   )/
> / ( -0.001673  -0.000000     0.707105   0.000000  0.000000   0.000000   )/
> /     freq (    8) =       9.672630 [THz] =  322.644206 [cm-1]/
> / (  0.000000   0.000000    -0.000000   0.000000 -0.246235   0.000000   )/
> / (  0.000000  -0.000000     0.000000   0.000000  0.685335  -0.000000   )/
> / ( -0.000000   0.000000     0.000000  -0.000000  0.685335   0.000000   )/
> /     freq (    9) =      10.162710 [THz] =  338.991522 [cm-1]/
> / ( -0.000000  -0.000000    -0.000000   0.000000 -0.000000  -0.000000   )/
> / ( -0.000000  -0.000000     0.000000   0.000000 -0.707107   0.000000   )/
> / (  0.000000  -0.000000    -0.000000   0.000000  0.707107   0.000000   )/
> / **************************************************************************/
>
> I summarize the phonon frequencies for easier comparison:
> *Gamma-M-K-Gamma:*   -0.0000    0.0000    0.0000 174.6512  260.1628  
> 260.1628  305.6148  305.9012  338.9915
> *DOS: *-0.0000   -0.0000    0.0000  174.6512 174.6512  260.1628  
> 260.1628  322.6442  338.9915
>
> As can be seen, the phonon calculation along high symmetry point shows 
> a degenerate frequency at ***260.1628* while the DOS calculation shows 
> degeneracy at *174.6512 *and *260.1628.* Moreover, a new phonon mode 
> with frequency *322.6442 *appears in the DOS calculation that is 
> absent from calculation along high symmetry directions.
>
> Why should the Gamma point frequency depend on the list of kpoints one 
> is diagonalizing dynamical matrices on? It does not seem to be 
> numerical error since the difference is so huge. Can anyone help me 
> understand it? Thank you very much
>
> FYI, input file for ph.x is shown below:
>
> /
> phonons of HfS2
>  &inputph
> prefix='Hf'
> epsil=.true.
> fildyn='HfS2.dyn',
>  outdir='./tmp/'
> tr2_ph=1.0d-12
>   ldisp=.true
>   nq1=4
>   nq2=4
>   nq3=4
>  /
> /
> -- 
> ------------------------------------------------------------------------------------------------------------------------
> Jie Peng
> PhD student
> 2134 Glenn Martin Hall, Mechanical Engineering, University of Maryland
> College Park, Maryland, USA
> Phone:(+1) 240-495-9445
> Email: jiepeng at umd.edu <mailto:jiepeng at umd.edu>
>
>
>
> _______________________________________________
> 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/20180816/c7179403/attachment.html>


More information about the users mailing list