Total magnetic moment: INFO.OUT vs LSJ.OUT

[Harry K]

Hi All,

I'm a little confused about the moments that are included in the INFO.OUT file. I interpreted the 'moments' section of INFO.OUT to be the entire magnetic moment in the unit cell. so for the following antiferromagnetic output for example:

Moments :
 interstitial                :  -0.6070203810E-16
 muffin-tins
  species :    1 (U)
   atom    1                 :   0.6865555152    
   atom    2                 :  -0.6865555152    
   atom    3                 :   0.6865555152    
   atom    4                 :  -0.6865555152    
  species :    2 (Au)
   atom    1                 :   0.6560020819E-21
   atom    2                 :  -0.6560020819E-21
   atom    3                 :   0.6560020819E-21
   atom    4                 :   0.6560020819E-21
   atom    5                 :   0.6560020819E-21
   atom    6                 :  -0.6560020819E-21
   atom    7                 :   0.6560020819E-21
   atom    8                 :   0.6560020819E-21
 total in muffin-tins        :   0.1110249265E-15
 total moment                :   0.5032288837E-16

I thought that meant that the total moment in each of the U muffin tins, orbital and spin combined, is +/- 0.6866. However the LSJ.OUT file suggests this isn't correct

Species :    1 (U)
 atom :    1
  L :   0.2156325880E-15 -0.1548721231E-19  -3.521125802    
  S :  -0.1046960262E-16  0.1052929878E-16  0.3431836098    
  J :   0.2051629854E-15  0.1051381157E-16  -3.177942192    
 atom :    2
  L :  -0.6468461323E-15 -0.1548721231E-19   3.521125802    
  S :   0.5249747355E-16  0.1052929878E-16 -0.3431836098    
  J :  -0.5943486588E-15  0.1051381157E-16   3.177942192    
 atom :    3
  L :   0.2156325880E-15 -0.1548721231E-19  -3.521125802    
  S :  -0.1046960262E-16  0.1052929878E-16  0.3431836098    
  J :   0.2051629854E-15  0.1051381157E-16  -3.177942192    
 atom :    4
  L :  -0.6468461323E-15 -0.1548721231E-19   3.521125802    
  S :   0.5249747355E-16  0.1052929878E-16 -0.3431836098    
  J :  -0.5943486588E-15  0.1051381157E-16   3.177942192  

Now there's a massive orbital contribution that dominates the total moment. So are the moments in INFO.OUT just the spin contributions? And how does this translate to comparison to experiment? Say i was trying to compare to a total magnetic moment measured in an experiment, should I compare the moment in INFO.OUT or should I sum those in LSJ.OUT?

I know this sort of stuff has come up a lot on the forum in the past, but I'm struggling to piece it together. Particularly why we would choose not to include the orbital contribution to the total moment in INFO.OUT. Any help is much appreciated!

Harry

[mfechner]

Dear Harry,

there is somewhere already a post of this to summarize

in INFO.OUT you find the spin magnetic moments

in LSJ.OUT you find the expectations values of L and S

to get the total magnetic moments you have

m=2*<S>+<L>

or the spin magnetic moment

ms=2*<S>

please not that the <L> value is not added in INFO.OUT. Since in your case its considerable large you will have to add this.

best regards
Michael

PS: Just as an adivce you migtht want to check the MT radius to see how good your values are ''converged'' since L and S expectation values are only computet within MT.

[Harry K]

Hi Michael,

Thanks for your detailed response :) So the total moment in the cell would be

m=2<S>+<L>+(interstitial moment) ?

where I get the interstitial moment from INFO.OUT? (Assuming it's not zero as in the above example).
Also, I know this is more a physics question than about elk, but how might I calculate the total moment using <J>?

And regarding your comment on muffin tins, you're suggesting I vary the muffin-tin radius and observe the convergence of the value in LSJ.OUT? Do I vary this from the species file? Should I accordingly vary other parameters like rgkmax or lmaxapw, and will it matter if I'm using DFT+U? And should I expect a better convergence for smaller or larger MT radius? My instincts tell me smaller.

Sorry for so many questions, thanks for all of your help already.

Regards,
Harry