From: "Radaelli, PG (Paolo) " [email protected]
To: "'[email protected]'" [email protected]
Subject: RE: One GSAS question.
Date: Thu, 17 May 2001 18:18:54 +0100

Dear Xiangyun Qiu,

the easiest thing is to create a new purely magnetic phase containing only
Mn and with P1 symmetry.  Since Mn is in a special position, you don't need
to refine the coordinates.  You'll have to constrain the reciprocal lattice
tensor parameters and the phase fraction between the two phases.  The latter
are inversely proportional to the cell volume.  Tricky but a good
exercise...

Paolo Radaelli

From: Xiangyun Qiu [email protected]
To: "'[email protected]'" [email protected]
Subject: RE: One GSAS question.

Thanks a lot for the answer!

Apologize for not making myself very clear. Actually I am refining some
layered manganites data, I have non special postion Mn atoms(4 in one unit
cell for the magnetic phase of Space group P1), I need to make them in 
the same position as the Mn in the atomic structure phase(only one in the
unit cell (I4/mmm) ). 

Thanks again!

Best,

Xiangyun Qiu


On Thu, 17 May 2001, Radaelli, PG (Paolo)
wrote:

> Dear Xiangyun Qiu,
> 
> the easiest thing is to create a new purely magnetic phase containing only
> Mn and with P1 symmetry.  Since Mn is in a special position, you don't need
> to refine the coordinates.  You'll have to constrain the reciprocal lattice
> tensor parameters and the phase fraction between the two phases.  The latter
> are inversely proportional to the cell volume.  Tricky but a good
> exercise...
> 
> Paolo Radaelli
> 

Xiangyun 

*********************
Xiangyun Qiu

Dept. of Physics & Astronomy
MSU East Lansing, MI
Http://www.msu.edu/~qiuxiang

*********************

Date: Fri, 18 May 2001 10:03:02 +0200
To: [email protected]
From: Jonathan WRIGHT [email protected]
Subject: RE: One GSAS question.

Better to work with a single phase and use the red and black symmetry
operators, in my opinion. You might have to lower the symmetry for some
magnetic structures and add contraints on the positions accordingly, but
this saves a lot of headaches, even if it creates a few by making you think
about the symmetry of the magnetic structure. Prodd and fullprof both
handle magnetic structures and multipattern ToF data with slightly more
flexibility than GSAS in terms of the kinds and symmetry of the structures
allowed.

If you insist on P1 then just put all the coordinates which are related
into the same constraint. The fourfold axis will relate x and y for the
atoms in the P1 space group to x or y from the tetragonal. However, if you
have a fourfold site in I4/mmm then it's either 0,1/2,0; 0,1/2,1/4 or 0,0,z
and you only need to constrain z for the latter and fix x and y. In expedt
the constraint would be something like:
1 z 1 1         ! Phase one (I4/mmm), z co-ord, atom one, value 1
2 z 1 1         ! Phase 2 (P1) atom at 0,0,z
2 z 2 -1        ! Phase 2 atom at 0,0,-z
2 z 3 1         ! Phase 2 atom at 1/2,1/2,z+1/2
2 z 4 -1        ! Phase 2 atom at 1/2,1/2,-z+1/2

Essentially theres a line added for each thing you want to constrain
together. It gets a bit hair raising if you try to put the same variable in
two different constraints - you can (and should) always put all the related
variables into the same one, this avoids confusion with disappearing or
'ignored' constraints. Remember to enter fixes (expedt; L A F) for the x
and y co-ordinates. I've assumed the magnetic cell is the same size as the
crystallographic. So it could be either ferromagnetic or antiferromagnetic
and the body centring involves time reversal, eg I4/mmm with k=(0,0,1). Or
some evil combination of the two :)

For "simple" structures you can get away with just refining the atomic
positions from the nuclear phase and updating them in the magnetic phase.
The magnetic model won't have a big impact on the crystal structure if you
have good (any?) data at short d-spacings, unless you are looking for small
structural distortions. If so, then it is indeed a tricky problem.

Best of luck,

Jon