Re: Magnetic refinements in GSAS

Paolo G. Radaelli ( (no email) )
Wed, 09 Sep 1998 14:07:23 +0100

Dear Paz:

In Gsas, there are two ways to carry out a magnetic refinement:

1) By the Shubnikov groups approach. Shubnikov groups are subgroups of the
regular space groups, whereby time reversal symmetry operators are allowed.
You can learn about them in:

W. Opechowski and R. Guccione, in Magnetism, edited by G. T.
Rado and H. Suhl Academic, New York, 1963, Vol. II A, p.
105.

In GSAS, you cannot input a Shubnikov group symbol, but you can construct
them by making certain symmetry operators "red" (time-reversal) or "black"
(non-time-reversal). This approach is very beautiful for those (like me)
that believe that the only interest in crystallography is symmetry, but is
limited, because only certain spin orientations are allowed. For instance,
if a spin sits on a 3- or 4-fold axis, it is only allowed to be parallel to
it. This is because the introduction of a spin perpendicular to such axis
will eliminate the symmetry operator altogether. In reality, if such thing
happens, the magneto-elastic coupling will lower the crystal symmetry as
well, but these effects are usually too small to be seen and refined.
Therefore, it is usually convenient to keep the crystal symmetry as is.
Also, the way it is implemented in GSAS, this approach cannot handle
magnetic supercells (no cell translation operators can be red) or
incommensurate magnetic structures.

2) By introducing a purely magnetic phase with the desired symmetry, which
can be different from the crystal symmetry. At the limit, you can refine
the magnetic structure in P1. This two-phase model is required when you
work with magnetic superstructures (with the exception, perhaps, of single
crystals, but I don't know enough about this option). This works fine as
long as you set the correct scale factors for the crystal and magnetic
phases, otherwise you will get the wrong moments. To do this, just remember
that the scale factor in GSAS is proportional to the number of unit cells in
your sample. Therefore, if you double the unit cell you need to cut the
phase fraction of the magnetic phase in half. Also, you will need to set
constraints on symmetry-related spins if you use a symmetry lower than the
true magnetic symmetry.

There are other ways to describe magnetic structures, namely using
representation theory. These are not currently implemented in GSAS. If you
need to work with incommensurate structures, for instance, you are better
off with FULLPROF. Also, note that there is a fundamental flaw in the way
GSAS handles magnetic structures with the moments not parallel to a crystal
axis, in that the components of the magnetic moments are treated
independently. This rapidly leads to divergences if the deviation from the
axis is small. The correct way is to treat the magnitude and the angle as
variables (somebody ought to point this out to Bob Von Dreele, sometimes).
In general, GSAS is pretty unstable for non-collinear structures, and you
may easily miss the solution. Having done hundreds of magnetic refinements
with GSAS and a few with FULLPROF, my advice is: use GSAS for simple,
non-collinear structures with high magnetic symmetry and FULLPROF for the
more complex structures.

Paolo
Dr. Paolo G. Radaelli
ISIS Facility
Rutherford Appleton Laboratory, Bldg. R3
Chilton, Didcot
Oxon. OX11 0QX
United Kingdom

Phone : (+44) 1235-44 5685
FAX : (+44) 1235-44 5642
e-mail: P.G.Radaelli@rl.ac.uk
pgr@isise.rl.ac.uk