How to include spin orbit LS coupling effects?

Message

kzhuo3 · #1 Post by **kzhuo3** » Wed Jul 03, 2013 6:05 am

I am working on WS2. I don't think magnetic effects are important here but W is heavy and I would like to include LS coupling.

From the manual, I know for sure that I need
LORBIT = .TRUE.
SAXIS = 0 0 1

But what about MAGMOM? Do I need this at all? If needed, what would be a good initial guess? e.g. MAGMOM = 0 0 0 0 0 0 0 0 0
or MAGMOM = 0 0 2 0 0 -1 0 0 -1

Are there any other flags that I need to add?

#2 Post by **admin** » Wed Jul 03, 2013 11:25 am

If LS coupling is on (LSORBIT=.TRUE.) non-collinear magnetism is assumed automatically. Therefore use the MAGMOM with initial values as recommended (1.2-1.5 times larger than experimental values).

Torbjorn · #3 Post by **Torbjorn** » Fri Aug 02, 2013 6:24 pm

The spin-orbit terms are of course independent of whether you have an exchange splitting of the potential, so there is no danger in principle with starting from zero moments, SO is included as it should. If you expect there to be a magnetic moment you should of course give it a kick to get started (as usual). Remember that you cannot use a binary compiled with the NGZhalf flag when running SO or noncollinear magnetism.