Negative and positive charge density isosurface mix up?

Message

kzhuo3 · #1 Post by **kzhuo3** » Fri Jul 15, 2011 11:18 am

I used to do calculations with hexagonal nanowire cross sections in rectangular cuboid unit cells. The nanowires were about 9A apart. Realising the lack of c3v symmetry, I changed the unit cells to rhomboids.

Unfortunately, the CHGCARs came out with predominantly negative charge density isosurfaces (as seen in VESTA) after I started using the more symmetric rhomboidal unit cells. This was no issue when I was using the cuboidal unit cells.

Furthermore, the isosurface shapes for a particular isosurface magnitude are practically identical in both cases. It's as though only the positive and negative charge signs got mixed up between both cases. Any help here?

Things I've tried:
increasing ENCUT, NG(X,Y,Z), PREC
using different pseudopotentials
post production runs with LPARD = .TRUE. to recalculate CHGCAR from WAVECAR

Some other people have encountered a similar issue before ( see http://cms.mpi.univie.ac.at/vasp-forum/ ... php?4.1783 ) but it hasn't received any solid conclusion yet.

kzhuo3 · #2 Post by **kzhuo3** » Sun Jul 24, 2011 9:22 am

I now think this is actually a VESTA bug. The latest VESTA beta version 2.90.1b, built on May 16 2011 has the problem but the older 2.1.6, built on Jan 24 2011 seems to display the CHGCAR fine.

kzhuo3 · #3 Post by **kzhuo3** » Tue Jul 26, 2011 12:06 pm

Yup, I have confirmation that this is a VESTA problem from the developers themselves. They promised to fix it in the next release. So VASP is not at fault this time. Sorry about that.