How can we calculate the magnetic moment associated with the net up-spin and net down-spin from density of states (DOS)? I use vasp for spin polarized calculations.
Secondly, how can we define correct bandgap in ferromagnetic semiconductor? The two different spin channels open different bandgaps.
Hello!
You can find the local magnetic moments with a decomposition into orbitals in the OUTCAR, if you set LORBIT = 11. Of course you can also integrate the DOS in both spin channels up to the Fermi energy and subtract them. That way you do not miss the interstitial regions.
For your second question I attached a short phyton script that calculates the bandgap(s) from a OUTCAR file. However, be advised that it is rather difficult to obtain correct bandgaps in DFT. With standard GGA functionals you will almost always underestimate gaps. Hybrid functionals can help but are computationally costly. GGA+U is a very tricky matter and as you have to choose a value for U (and J), predictive power is also limited unless you can calculate these parameters with one method or another. Ideally you have an experimental value of the bandgap to at least know if your chosen level of theory gets you close to that.
Cheers, Michael
Hi,
I think that the first answer from @Michael Wolloch is for calculation using VASP.
In the SIESTA output AND if you did spin polarized calculations you will have the answer right away:
siesta: Total spin polarization (Qup-Qdown) = 2.63014.
To your second question: for me, any bandgap must be calculated for the same type of band: Gap(up)= Conduction(up)-Valence(up) and Gap(down)= Conduction(down)-Valence(down). So, you will have two bandgaps.
The answer from @Michael Wolloch give you some insides about the band calculation. This is a major problem of DFT. For some materials, LDA or GGA are ok. For other ones, you must use hybrid, for other ones any actual functional works, and so far. You have to search the literature for your material type and pick the right functional or at least, a good one (of course, you always can use the scissors operator).
I have observed a difference between the magnetic moment in OSZICAR and OUTCAR, which value of magnetic moment is correct and why?
Hi,
You are probably comparing two differnet things. The magnetization from OSZICAR should be the integrated spin up DOS minus the integrated spin down DOS. From OUTCAR I guess you read the added up local moments at the end if you set LORBIT=11. This should look like this:
magnetization (x)
# of ion s p d tot
----------------------------------------
.
.
.
64 0.005 0.140 0.000 0.144
------------------------------------------------
tot 0.454 4.111 34.506 39.071
Here you have the moments only in spheres around the atoms and you are gonna miss some information of space in between them (you can play around with RWIGS to minimize empty space between atoms). Thus the values are different. Both are correct thou, but one is the total magnetization (which is of course also written in the OUTCAR, just search for magnetization) and the other is the sum of the local moments.
Hope that clears things up,
cheers, M
Hello All,
I have a query here. I understand that magnetic moment will be the difference of spin up and spin down area of DOS plot. As I know area of DOS plot will be unitless. In that case, I am confused about what will be the unit of moment calculated from the net area. Is it Bohr Magneton or I have to do some conversion?
FYI: I used SPRKKR for the calculation of DOS. I don't have the OSZICAR or OUTCAR file. I have to find the moment from the DOS plot.
Thanks for your time and cooperation.
- Badges
- Science topic
- Similar topics
- Chemistry
- Nanotechnology
- Nano