Magnetic Moment in LMD state, Improvement on Curie Temperature, Atoms position in the unit cell

Posted on : January 09, 2019 (Wed) 00:05:52

by Mohammad Saidur Rahman

Hello Expert(s),

I'm a new learner in this field and have already posed several problems, which is obvious. As not everything is given in the available manual book, it's difficult to resolve the problems on my own. I have already taken a look at the blog posts. It's very helpful and to the best of my knowledge it's the only one of its kind. I tried to access to other sites, but those are all in Japanese. Anyway, here are my questions:

1. When I calculated the LMD energy using the input file that's available within the package for "fe", the magnetic moment turned out to be 0.000. Is it good?
What is the unit of magnetic moment that is being used in this package?

FYI, I have first calculated SCF (performed using "go" mode) for ferromagnetic Fe, then created the potential file using ./fmg for LMD state calculation, exactly what you previously suggested in the blog posts. Then I went on to calculate LMD energy.

2. My calculated Curie temperature (T_c) for "fe" was 1265.2836 K, which is greater than experimental value 1043 K. Is it good? Can I get more close to the accurate (experimental) value by changing something in the input file?
Interestingly, when I took into account the relativistic treatment (sra), T_c was like 1186. What's going on? Does Mean Field Theory treat the "fe" system relativistically? If not, then putting "sra" in the input file is not justifiable, right?

3. My last question: When I use the conventional unit cells like bcc/fcc, I am having problems with atom numbers and their positions in the unit cell (in the input file). An example may help you to understand what I'm implying:

Suppose I'm working with GaAs semiconductor crystal system. It has zinc-blend structure. In the package's input file it's given that-

c atmicx atmtyp
0 0 0 Ga
0.25 0.25 0.25 As
0.5 0.5 0.5 Vc1
0.75 0.75 0.75 Vc2
c-----------------------------------------

Now if there were no defects included, there'd have been 4 atoms in the unit cell according to above.
But any crystal with zinc-blend structure should have 8 atoms in a unit cell. Like 4 atoms come from (8 corners and 6 sides) and other 4 come from (0.25a,0.25a,0.25a), (0.75a,0.75a,0.25a),(0.75a,0.25a,0.75a),(0.25a,0.75a,0.75a). [a=lattice constant] Am I right?

If so, then are you using a different zinc-blend representation that I'm not aware of? Care to tell me?

I've also seen that you guys used primitive fcc/bcc in the input files. Is it because you want to minimize the amount of atoms in a unit cell for being computationally less expensive?

I hope you have got what I've been facing problems with.

Sincerely,
Mohammad Saidur Rahman

[Re:02] Magnetic Moment in LMD state, Improvement on Curie Temperature, Atoms position in the unit cell

Posted on : January 09, 2019 (Wed) 23:47:49

by Mohammad Saidur Rahman

Hello Mr. Hitoshi GOMI,

Thanks a lot for your answers.

Don't you think "number of atoms in the unit cell" for the acronym "natm", as stated in the manual, for a crystal structure system in concern, is slightly misleading?

Other questions: About those two vacancies in GaAs system- those are definitely empty spheres, right? Or are they defects?
Can I place them in other places in the unit cell except (0.5,0.5,0.5) and (0.75,0.75,0.75) or introduce more empty spheres other than only two?

Sincerely,
Mohammad Saidur Rahman

[Re:01] Magnetic Moment in LMD state, Improvement on Curie Temperature, Atoms position in the unit cell

Posted on : January 09, 2019 (Wed) 19:22:32

by Hitoshi GOMI

Hello Mohammad Saidur Rahman,

>the magnetic moment turned out to be 0.000. Is it good?

Total (bulk) magnetic moment should be zero.
Local magnetic moment should remain.


>What is the unit of magnetic moment that is being used in this package?

Bohr magneton


>1265.2836 K, which is greater than experimental value 1043 K. Is it good?

I think so.


>Can I get more close to the accurate (experimental) value by changing something in the input file?

I am not familiar with the issue, but the MFA sometimes overestimates the Tc.
See https://arxiv.org/abs/cond-mat/0408450


> But any crystal with zinc-blend structure should have 8 atoms in a unit cell.

The crystal structure is specified by the combination of a Bravais lattice and a basis.
The zinc-blend structure can be represented by the fcc Bravais lattice and the basis, which have two atoms.
It can also be represented by the simple cubic Bravais lattice and the basis, which have eight atoms.
See also http://kkr.issp.u-tokyo.ac.jp/bbs/thread.php?id=297

Hitoshi GOMI