[Pw_forum] How to distinguish phonon modes?

[Lorenzo Paulatto]

On 12/02/2013 03:28 AM, Kyeong-hyun Park wrote:
> I was able to plot along different directions, but how would I extract a
> specific phonon mode (for example, LA or TA1) from the frequency data
> file. I need to use only one mode of frequencies to do my calculation.


Dear Kyeong-hyun,
there are different way to do what you ask, each one has advantages and 
disadvantages. I'm going to list them briefly afterwards, but notice 
that I'm not discussing how to find out which is mode X or Y (X,Y = TA, 
LA, breathing,... ) as this depend on the material.

The matdyn code can sort the bands in different ways, depending on it 
input parameters. When you run matdyn it will create a file *.freq which 
can be post-processed by plotbands. This is required in case 3, in the 
other cases you can directly plot the .frq.gp file.

1. (default) The bands are sorted from the lowest to the higher in order 
of energy. As long as you bands never cross this is ok, it is also 
simpler and tidier.

2. (eigen_similarity=.true.) Order the phonons by energy at the first 
point, then follow up by continuity of the displacements. Pro: the order 
of the bands is kept when they cross. Cons: sometimes this goes wrong 
and produces ghost crosses, you then need to increase the number of q 
points. Also if the two bands crossing have the same character (i.e. ZA 
crossing ZO) this method won't be able to tell them apart. It only works 
for band paths, no band dispersions.


3. Using symmetry to discriminate between band crossind and avoided 
crossings. This requires that you modify the code of matdyn which 
inexplicably only enables this feature when the dynamical matrix is in 
xml format, add
      xmlifc=.true.
at line 311, just before the line
      CALL recips ( at_blk(1,1),at_blk(1,2),at_blk(1,3),  &
           bg_blk(1,1),bg_blk(1,2),bg_blk(1,3) )
With this modification when running matdyn (without special options) it 
will create a file .freq.rap which contains the modes symmetries, this 
file is the used by plotbands (to which you still feed the .freq file) 
to reorder the bands according to symmetry.

Pros: it does not have the ghost crossings of eigen_similarity.
Cons: It still only works for continuous q-point paths: if you require a 
broken path (i.e. G->X->L) each segment will be stored in a different 
file (fildyn.1.1, fildyn.2.1, etc.) and continuity between segments is 
not guaranteed.

4. In general there is no way to sort out a single band in the entire 
3-dimensional space. The problem is often of principle: a band that 
looks well distinct along a high symmetry line may cross out of that 
line. yet, if a band DOES have a well distinct character you CAN 
discriminate it, but it must be done by hand on a case-per-case basis. 
In other words: you have to write your own code. I.e. you may select the 
ZA and ZO bands of graphene by checking whether their displacements have 
a component along Z, but this approach won't work for other materials.


I hope this helps.

-- 
Dr. Lorenzo Paulatto
IdR @ IMPMC -- CNRS & Université Paris 6
+33 (0)1 44 275 084 / skype: paulatz
http://www-int.impmc.upmc.fr/~paulatto/
23-24/4é16 Boîte courrier 115, 4 place Jussieu 75252 Paris Cédex 05

-------------- next part --------------
A non-text attachment was scrubbed...
Name: smime.p7s
Type: application/pkcs7-signature
Size: 2956 bytes
Desc: S/MIME Cryptographic Signature
URL: <http://lists.quantum-espresso.org/pipermail/users/attachments/20131202/3ef9ecd1/attachment.p7s>