I also obtained a similar fig. using Vashista potential. As you mentioned
I couldn't find any paper where the phonon band is calculated using the
The phonon bands of 3C SiC in the literature are usually calculated using
MD+DFT, and the result is different. The main difference is that in the
DFT case, a gap is opened in the Gamma point for the optical band, that it
is not found from MD simulations with these potentials.
On Thu, Oct 17, 2019 at 8:33 AM Bruce Fan <brucenju@xxxxxxxxx> wrote:
I have tried the Tersoff potential as used in the paper by Porter, Li, and
And I got the same results as in figure 1(b) of this paper. See the figure
below. There is a trick to obtain the correct results: the cutoff distance
in xyz.in should be set to between 1.86 and 3, because Si-Si and C-C
interactions should be excluded. The cutoff in phonon.in (for force
constants) should be larger to include these pairs (the dynamic matrix
element should include second-nearest neighbors). I used 2.5 A and 3.5 A
for the above two cutoff distances. See more discussion in section 2 of the
the above paper. The potential parameters were also taken from this paper.
So my conclusion is that there should be no problem for SiC with the
Tersoff potential using the phonon code in GPUMD. As for Vashishta, I have
not found a reference which has calculated the phonon dispersion. So I
don't know how to proceed. If there is nothing to compare, you are in an
unexplored island and I cannot say for sure the code will still work :-)
Note that the k-path in the figure below is not the same as in the example
I provided. I only learned the subtle difference between the two kinds of
On Thu, Oct 17, 2019 at 12:21 PM Bruce Fan <brucenju@xxxxxxxxx> wrote:
Thanks, but this is from experiments, right? We should compare to results
from the same potentials we used...
I remember Ju Li has a paper on SiC around 1997 using a Tersoff. How
about Vashishta, any published results?
Florencia Carusela <flor.caru@xxxxxxxxx> 于 2019年10月17日周四 12:17写道：
Here is the link of the paper
On Thu, Oct 17, 2019 at 5:29 AM Bruce Fan <brucenju@xxxxxxxxx> wrote:
Thanks for the question. This seems to be a difficult question. I have
not tested the phonon code with SiC before. I tried to find the paper you
referred to, but couldn't find it. Could you give me a link to the paper?
I have a few comments after a quick look at your inputs:
1) run.in is not needed to run the phonon code. It was my fault to put
it in the released example. That was used when I developed the code. Now
you only need the following files: phonon.in, kpoints.in, basis.in,
and xyz.in. Of course, including run.in in the input directory will
not cause any problem.
2) Your create_xyz.m Matlab script will produce an xyz.in file with
the first line looking like:
512 4 3 0 0 0 0
This means that you set the cutoff distance for neighbor list building
to 3 A, which is too short for the Vashishta potential.
I remember the cutoff distance for this potential is 7.35 A. Then you
can set it as 8 A. So the
first line can be changed to
512 300 8 0 0 0 0
Here, "300" is used to reserve enough memory of the neighbor list. I
know this number is difficult to choose, but I still leave it to the user
rather than using a fixed large number such as 1000. Sometimes, 4 or even 3
is enough and one can save memory by choosing it properly.
3) Accordingly, the cutoff in phonon.in should also be changed to a
larger number, say 8 A.
4) The k-path I provided in the example might not be the same as those
used in the references you were looking for. I need to study this in more
On Thu, Oct 17, 2019 at 12:20 AM Florencia Carusela <
I have problems to calculate the phonon bands of the 3C SiC. I am
using Vashishta potential. I checked your parameters and are ok according
to Vashishta's paper.
There are problems with the optical band, I cannot reproduce it (look
e.g. APPLIED PHYSICS LETTERSVOLUME 80, NUMBER 2310 JUNE 2002).
I tried different orientations and cutt off distances but still not
working. I don't know if I am doing something wrong or something I should
tune in the code. Did you ever try to calculate it?
By the way, the phonon bands for Si with Tersoff are ok! I also tried
Tersoff with 3C SiC but it is also worse.
I attach the input files