[QE-users] Systematically improve surface slab phonons - any "trick"?
wolf.christoph at qns.science
Thu May 3 18:59:27 CEST 2018
Thanks to help of Thomas Brumme and others I got some decent results for
phonons of 2D systems, but it is a bit of a hit or miss approach and I am
not sure how to systematically improve the results ("phonon dispersions").
There are a few things that seem to be quite reliable
- vc-relax cells give better phonons than equation of state derived ones
- lowering scf convergence to 10e-16 definitely improves
- as soon as the potential is flat (pp.x plot_num=11 and then average.x...)
increasing vacuum does not seem to do much. Alternatively
assume_isolated='2D' also helps but I want to avoid it because other codes
don't seem to work (EPW)
but there are a few things that puzzle me, in particular
- increasing ecutwfc does not systematically improve phonons (the scf
energy seems to converge at ~60 Ry cutoff but I increased it further to 120
to see if it has any effect, does not seem to be the case though..)
- increasing in-plane q point sampling does not systematically improve
phonons (e.g. going from 4x4x1 to 6x6x1 to 8x8x1 might first improve, then
worsen the result)
- last but not least, matdyn.x seems to have a problem with some supercells
(wrong charge, e.g 15.8 instead of 16) and increasing the very-very safe
range of 2*nr does not seem to be the trick. Occurrence is a bit random and
I was hoping that someone with experience of 2D systems could share his or
her "magic ingredient". Since dispersions are quite costly and soft modes
appear at the weirdest parts of the BZ. In particular I am interested in
oxides on metal slabs (this is a bit tricky as the oxides have large gaps
but the stacked system becomes a metal the supercell..).
Any help is very much appreciated.
Thanks in advance!
Center for Quantum Nanoscience, Institute for Basic Science
Ewha Womans University, Seoul, South Korea
-------------- next part --------------
An HTML attachment was scrubbed...
More information about the users