magnetization

Queries about input and output files, running specific calculations, etc.


Moderators: Global Moderator, Moderator

Post Reply
Message
Author
apple
Jr. Member
Jr. Member
Posts: 99
Joined: Tue Feb 16, 2010 7:09 pm

magnetization

#1 Post by apple » Sat Jan 07, 2012 2:02 am

hi,
I am looking at oxygen vacancies in a bulk system with a energy gap of ~2.5 eV.
The total energy I get is -241.4019 eV and the magnetization is 1.71.
If I constrain the magnetic moment to 1 (by setting the NUPDOWN flag to 1), the total energy is -241.400 eV and the magnetization is 1 (of course).
If I constrain the magnetic moment to 2 (by setting the NUPDOWN flag to 1), the total energy if -241.388 eV and the magnetization is 2 (of course).
Why are the total energies insensitive to the magnetic moment, i.e. why are they equal even though the magnetizations are different? I used the same smearing method in all cases.
Can someone please explain it to me. I am completely lost.
Thanks.
Last edited by apple on Sat Jan 07, 2012 2:02 am, edited 1 time in total.

Danny
Full Member
Full Member
Posts: 201
Joined: Thu Nov 02, 2006 4:35 pm
License Nr.: 5-532
Location: Ghent, Belgium
Contact:

magnetization

#2 Post by Danny » Sun Jan 08, 2012 1:25 pm

degeneracy of those two magnetic states? Do a series NUPDOWN=0..5 for example, then you will generally see different behavior. With regard to your system, just check the standard spins for the elements involved an see if multiple combinations are possible. Note that some elements have multiple valencies. Take for example TiO2. O has valence -2, Ti has valence +2,+3,+4. In TiO2 the formal valence of Ti will be +4, if you make an O vacancy, then you can consider one unitcell to contain TiO. Using all possible valencies for Ti, you find that mag=0,1 and 2 are possible. This might also be the case in your system. For the TiO2 I would then predict that mag=0,1,2 can be quite close toghether in values, however mag=3 will differ quite a lot (and should be less stable in this simple view).

Danny
Last edited by Danny on Sun Jan 08, 2012 1:25 pm, edited 1 time in total.

apple
Jr. Member
Jr. Member
Posts: 99
Joined: Tue Feb 16, 2010 7:09 pm

magnetization

#3 Post by apple » Mon Jan 09, 2012 5:15 am

Danny,
thanks for replying. Following your suggestions, I am running a series of NUPDOWN jobs from 0-5. While still waiting for them to converge, I can see that mag=0,1,2 results are close in energy, while mag=3-5 results are higher in energy. Now, given the degeneracies in my magnetic solutions, should I consider calculations for charged vacancies? (I have never worked on defects, or charged defects for that matter, so please excuse my perhaps stupid questions). I would appreciate your help. Thanks.
Last edited by apple on Mon Jan 09, 2012 5:15 am, edited 1 time in total.

Post Reply