[Pw_forum] Structural Optimization Calculation

[Maciej Szary]

Dear Cameron,

> I expect this simulation to allow the _atomic positions_ to relax 
> leaving the actual lattice constant unchanged. I also expect the 
> atomic positions to relax in all 3 directions (hence the 1 1 1 after 
> each coordinate specification) but I have seen the x and y coordinates 
> remain relatively unchanged. Again I am really after the Se-Se distance.
>
Why would you expect in-plane relaxation of atoms? You're calculating 
1x1 surface cell, so any Se in-plane movement wouldn't change anything, 
next set of neighbour Se atoms would still be arrange respectively with 
presented lattice constant. Of course, in-plane position of Mo atom 
could change with the respect to Se however, it would distorted the 
MoSe2 structure. Normally from the top view you see hexagonal structure 
for TMDC materials and a change of in-plane relative positions of Mo and 
Se would affect this symmetry.

I don't recall MoSe2 reconstructions, but maybe literature say other 
ways. If so you should just repeat relaxation if right supercell.

> My questions are: is this relaxation calculation doing what I expect 
> it to do? And is the equilibrium structure guaranteed if convergence 
> is achieved? The motivation for the second question is that, I have 
> run a relaxation calculation with this input file, it converges, but 
> then I find negative frequencies in the phonon dispersion particularly 
> with the ZA and TA modes (the ZA mode being the characteristic 
> quadratic mode found in 2D monolayers such as graphene and would be 
> most susceptible to variations in the Se-Se distance). As the 
> simulation experienced no interruptions I suspect the negative 
> frequencies to be a result of numerical issues with the atomic 
> positions and/or the MP grid size (note I used an 8 8 1 grid size for 
> the phonon calculation, not 6 6 4 as mentioned above)

If you think that atomic positions could be the source of your problems, 
than maybe should try different functionals (PBE should provide 
different distances).

Regards,
Maciej

Maciej Szary,
PhD student,
Computational Physics and Semiconductors Division,
Poznan University of Technology,
Poland
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.quantum-espresso.org/pipermail/users/attachments/20160217/dc2a42b4/attachment.html>