welcome in the club!
I'm pleased that the fundamental parameter approach for modeling peak
profiles is now available in another new Rietveld program. I understand yo=
ur
enthusiasm because we got the same feeling some years ago.
For already 3 years, some of us have used the newly developed Rietveld
program BGMN (Author: Joerg Bergmann; C language) for quantitative phase
analysis and structure refinement. It implements another fundamental
parameter approach for modeling peak profiles, but has the same advantages
as Koalariet and some new features:
- absolute stability and convergence due to the new profile model and a
refinement algorithm, which allows the definition of physically
meaningful parameter ranges -> lattice constants and site occupancies
for substitutional series can be refined in an easy way, no more
negative temperature factors etc.!
- no refinement of instrumental profile necessary;
the program handles not only strongly asymmetric Bragg-Brentano
profiles, but also transmission geometry profiles (flat sample/capillary=
);
of course soller slits and fine detector slits are no prerequisite for
BGMN refinements which improves the particle statistics dramatically
- refinement of multiple and strong preferred orientation with built-in
spherical harmonics (good refinement results eg. for feldspars and
phyllosilicates); automatical prediction of the suitable order
- refinement of max. 3 parameters for crystallite size/microstrain per
phase (size, size distribution, microstrain);
the line broadening can in common be anisotropic
- powerful formula interpreter which allows the free definition of new
parameters and constraints between any parameters
- modeling and refinement of strongly disordered structures for chlorites
and kaolinite without changes in the compiled code using the interpreter
- convenient operation under Dos, Windows95/NT and OS/2
The stability of the peak model allowes for example the refinement of
domains of one phase with different degrees of real structure and (nearly)
the same lattice constants.
For "structure-refiners", there are functions for modeling force fields
between all atoms in the unit cell and between neighboring cells.
Additionally the formulation of chain models is possible in an easy way.
These features were successfully used for structure investigations of
polymers and other molecular crystals.
The program including some references is presented some time next week und=
er
http://www.mineral.tu-freiberg.de/mineralogie/bgmn
In the next days also a slightly limited demo-version of the program with
some examples will be available on this web-site.
With best regards,
Thomas Taut
> While the following opinions may put my head on the chopping block more
> than is healthy, the above is required for the new Rietveld program
> Koalariet
> for Win95 written by Alan Coehlo. This uses the fundamental parameters
> approach of Coehlo and Cheary.
> The software is available via anonymous ftp to:
> ftp://ftp.minerals.csiro.au/pub/xtallography/koalariet
> In trying to find analogies between this Koalariet Rietveld program and
> others, I have one of the following 3 to yet decide upon:
> 1) An early jet engine of the 1940's vs piston driven propeller
> aircraft.
> 2) An early screw driven ship vs the paddle variety
> 3) The coming of modern poetry at the start of the 20th century
> ("But Mr Eliot.....it doesn't rhyme????!!!!")
> A personal summary is that the Koalariet software does away with
> the typical mathematical peak fits and uses fundamental parameters to
> model
> the contribution of the diffractometer and other effects. Thus, for
> instance, in modeling peak width/shape, you have the choice of directly
> refining on:
> Crystallite Size and/or (if necessary)
> Gaussian Strain and/or
> Leorentzian Strain
> All other effects such as low angle peak asymmetry, profile contribution
> by
> the diffractometer, high angle peak asymmetry and peak shifts due to
> linear
> absorption, etc; are automatically modeled as part of the geometry and
> sample
> (the diffractometer, linear absorption parameters are refinable/checkabl=
e
> if
> you are in the mood).
> Thus there is a much better chance that any misfit of the peaks can
> be traced to real physical effects of the sample or diffractometer
> as all the main parameters profile parameters are defined as physically
> meaningful measurements. This compares again to mathematical fits that
> are physically meaningless(?) and cannot be pursued further if there are
> misfits. The Koalariet program can even model rather bizarre peaks shap=
es
> obtained by removing both incident and diffracted beam Sollers slits -
> without any requirement for extra mathematical fit parameters.
> ----
> A bonus advantage of this is you can get rid of the diffracted beam
> Sollers slit, use a coarser receiving slit to double the intensity seen
> by the detector and improve particle statistics - the program has
> no trouble modeling effects such as the extra peak asymmetry. This
> compares with the typical 3 width, 3 shape approach and up to a dozen(?)
> extra parameters to improve the mathematical fit of the peaks(?).
> Combined with being completely written anew in C, the program is extreme=
ly
> stable. (An example on the ftp site shows that Peak anisotropy on a
> sample is more likely due to a distribution of unit-cells).
> ----
> Another major feature of the program is an in-built macro/equation edito=
r
> in
> the input files. Thus any parameter can be refined or linked to any oth=
er
> parameter in an arbitrary manner (i.e., link the elemental substitution =
to
> the unit cell shift for quantitative analysis purposes)
> New parameters/definitions/macros can be defined in the input file witho=
ut
> having
> to get a change in the source code:
> (i.e., modeling preferred orientation on multiple vectors
> modeling peak anisotropy (presently as mathematical fits) on
> multiple
> vectors
> check for peak intensity variations due to non-infinite thickness=