[Wien] magnetic susceptibility for a ferromagnetic metal or for a ferromagnetic semiconductor

pieper pieper at ifp.tuwien.ac.at
Thu Jul 20 16:54:10 CEST 2017



>>> From what I have understood from userguide and Prof P Blaha's 
>>> replies;
>>> For semiconductor and insulator; there is the orbital part of the 
>>> magnetic susceptibility only
>>> but for the metals there is also the spin part

No! This is not correct. Let me expand a little on my view of this topic 
in the hope to clarify the problem at least somewhat:

In a nutshell, you want Wien2k to find the state minimizing the total 
energy of some solid because you suspect that this is the state whose 
properties you observe in some experiment, and someone told you that DFT 
is the method of choice to do the search.

Magnetic moments contribute M*B to that energy with M and B being 
vectors in a scalar product. In a first approximation two contributions 
from electrons to M can be considered separately, namely the spin 
moment, and the moment from an orbital momentum (or the current 
associated with a non-zero orbital momentum). Both parts contribute to 
the energy in the field. That is why Peter Blahas very first statement 
was:

A magnetic field influences the spin and orbital degrees of freedom.

It always does. Both contributions to M. No matter wether you have an 
insulator or a metal. The influence on specific properties you are 
interested in may be too small to bother, but somewhere else it will 
show up.

There are different ways to tell Wien2k where to look for the state with 
minimum energy in an applied magnetic field (or electric field). This is 
where symmetry enters stage: Wien2k searches the state with lowest 
energy in the subspace compatible with the symmetries specified by the 
input files. The symmetry of the crystal structure is in the struct 
file. Concerning magnetic moments you can tell Wien2k to

- consider only states that are symmetric with respect to spin, i.e. do 
a non-spin-polarized calculation. Doesn't make sense if you are 
interested in the effect of a magnetic field or want to model 
ferromagnets, but for the majority of materials and properties it's 
perfect to save time and calculate just spin +1/2, then carry the result 
over to spin -1/2.

- take into account the two spin channels separately, i.e. do a 
spin-polarized calculation, and introduce B as 'orbital potential' in 
the program orb. Note that a field you put into .inorb acts on the 
electrons in the single atomic shell you specify, and only for the part 
within the muffin-tin radius. Everything else is indirectly adjusted 
according to interactions. I am (of course) with Peter Blaha: One can 
expect very reasonable estimates for field induced moments at the given 
shell, at least if the field induced polarization of electrons not 
affected by .inorb is small. The option also shows a Zeeman splitting of 
spin-up and -down bands due to the interactions, but I expect that 
depending on the case this can rapidly turn into a paedagogical example 
rather than an estimate of the spin susceptibility.

- take into account the fact that the energy contribution M*B quite 
possibly reduced symmetry. M as well as B are vectors, after all, and 
transform as such under symmetry operations. You can switch on 
spin-orbit interaction to tell Wien2k about this. (Be aware that Wien2k 
always assumes that M and B are colinear). As Gerhard Fecher pointed 
out, a reduced symmetry in a magnetic field is NOT taken into account in 
the orb program. With only the orb option active Wien2k may search the 
needle (the energy minimum) in a completely wrong haystack (subspace).

- do a calculation of the linear response of the all-electron wave 
function (representing spins AND orbital moments) to an applied field. 
As I understand it this is what the NMR module does. However, it 
calculates only the local response at the nuclei, not the integrated 
macroscopic susceptibility.

Note that all this does not represent anything like the macroscopic 
susceptibility you will detect with a ferromagnetic (or any other 
magnetically ordered) material in your magnetometer. The susceptibility 
in these cases even at zero Kelvin (where DFT arguably applies) is 
dominated by things the spins can and will do, but Wien2k is completely 
unaware of (magnons, domain walls ...).

Finally, don't take the UG or this mailing list as a substitute for a 
textbook. That is not their intention. Follow the advice of Prof. Fecher 
and read up on the subject of magnetism in solids. Perhaps start with 
something on solid state physics in general, not a specialized treatment 
of magnetism. Personally I like the introduction to Solid State Physics 
by Ashcroft and Mermin.

Good luck

---
Dr. Martin Pieper
Karl-Franzens University
Institute of Physics
Universitätsplatz 5
A-8010 Graz
Austria
Tel.: +43-(0)316-380-8564


Am 19.07.2017 16:02, schrieb karima Physique:
> From what I have understood from userguide and Prof P Blaha's replies;
> For semiconductor and insulator; there is the orbital part of the
> magnetic susceptibility only but for the metals there is also the spin
> part and I ask Prof. P Blaha and Prof Gerhard Fecher  to confirm this
> answer or to correct it.
> 
> 2017-07-19 13:34 GMT+02:00 Wien2k User <wien2k.user at gmail.com>:
> 
>> I did not underestimate his answer and the proof I thanked him and I
>> apologize if I did not convey my message well
>> 
>> 2017-07-19 12:48 GMT+02:00 Wien2k User <wien2k.user at gmail.com>:
>> 
>> Dear Fecher, Gerhard
>> 
>> You can answer me directly instead of asking me all these questions
>> otherwise I thank you for your answer and I will look for this book
>> to read it and in the meantime I will wait for the answers of the
>> users and prof P. Blaha that I much prefer.
>> 
>> 2017-07-19 3:47 GMT+02:00 Wien2k User <wien2k.user at gmail.com>:
>> 
>> dear wien2k user
>> 
>> From the userguide we find how to calculate the magnetic
>> susceptibility for an insulator or a paramagnetic metal but how to
>> calculate the magnetic susceptibility for a ferromagnetic metal or
>> for a ferromagnetic semiconductor?
> 
> _______________________________________________
> Wien mailing list
> Wien at zeus.theochem.tuwien.ac.at
> http://zeus.theochem.tuwien.ac.at/mailman/listinfo/wien [1]
> SEARCH the MAILING-LIST at:
> http://www.mail-archive.com/wien@zeus.theochem.tuwien.ac.at/index.html
> [2]
> 
> 
> 
> Links:
> ------
> [1] http://zeus.theochem.tuwien.ac.at/mailman/listinfo/wien
> [2] 
> http://www.mail-archive.com/wien@zeus.theochem.tuwien.ac.at/index.html
> _______________________________________________
> Wien mailing list
> Wien at zeus.theochem.tuwien.ac.at
> http://zeus.theochem.tuwien.ac.at/mailman/listinfo/wien
> SEARCH the MAILING-LIST at:
> http://www.mail-archive.com/wien@zeus.theochem.tuwien.ac.at/index.html


More information about the Wien mailing list