> Matthew Henrichsen <[EMAIL PROTECTED]> wrote 20 Jan 1999:
> >Why does x-ray powder diffraction fail? Why do the others work?
>
> If you read Rod Hill's intercomparison (J. Appl. Cryst. (1992) 25, 589-610)
> you will see that it is not only thermal parameters that are better determined
> by single crystal and neutron powder methods :-)
>
> There are simply more systematic errors with X-ray powder diffraction,
> and the on-going discussion about preferred orientation illustrates this.
> With neutrons you have much larger samples and it is less of a problem.
>
> You also have intensities that don't fall off rapidly with angle due to the
> X-ray structure factor (neutron scattering is from the almost point nucleus)
> Accurate Debye-Waller factors need good high angle data. Absorption
> is also much larger for X-rays, and an error there translates directly into
> an error in the overall thermal parameter.
>
> I guess one can debate all these points, but its difficult to argue with the
> numbers from the intercomparison.
>
> And hey, if you could do it all with X-rays, we neutron types would be
> out of a job !
>
> Alan H.
>
> Alan Hewat, ILL Grenoble, FRANCE <[EMAIL PROTECTED]> tel (33) 4.76.20.72.13
> ftp://ftp.ill.fr/pub/dif fax (33) 4.76.48.39.06 http://www.ill.fr/dif/
While neutrons do seem nicer (if you have easy access to
them) I agree with Joerg that if due care is taken, you
can still do quite well with Powder X-ray data - depending
on the problem at hand. Though this is not a 5 minute
job and might not be a popular option(?).
It most likely requires more patience, willingness if
needed to perform multiple sample synthesis/preparation
experiments and carefully examining the results (not
just releasing anisotropic code words willy nilly on
a single sample preparation and data collection).
Sounding like a cracked record of old, difference Fourier
electron density maps can be indispensable in visually
examining the results to help figure out if the extra
parameters are refining on noise, thermal motion or disorder.
Regular visual examination and comparison of the ORTEPs is
also useful (in the below cited case using Eric Dowty's commercial
ATOMs for Windows software - which can directly import a
variety of Rietveld formats - and does very well generating
Polyhedra - http://www.tricon.net/comm/shape/)
(Refer fig 8 of below cited J. Acta Cryst paper)
Refinement of light atom/Oxygen occupancies of heavy atom
structures may also be possible.
----
Excuse the self citation but a recent example of this is
viewable:
"Structure Analysis of the 6H-Ba(Ti, Fe3+, Fe4+) O(3-o)
Solid Solution", Journal of Solid State Chemistry, 135, 312-321
(1998) Grey, Li, Cranswick, Roth and Vanderah.
"Accurate Site Occupancies for Light Atoms from Powder X-ray
Data?" Oxygen/Vacancy Ordering in 6H-BaFe0.67Ti0.33O(3-o)
(o = 0.08 and 0.32), Supposedly in the latest J. Appl Cryst
(only have copies of the proofs) Grey, Cranswick and Li.
(In the case of trying to refine accurate Oxygen site
occupancies from X-ray Powder data, this included;
experimenting with low angle data truncation to
minimise the influence of the valence states of
the atoms; collecting and comparing both fixed
count time/variable count time data collection;
synthesise multiple batches of samples to
examine reproducability - not just repacking the
same sample).
---
Possibly repeating myself from a previous post
before publication, for those who remember the
talk on this at IUCr in Seattle; Oxygen occupancies were
originally derived exclusively from Powder XRD
refinements. This caused negative comments from
the audience and in the final publication, validating
Neutron refinements are presented as well.
I don't think the above will close down any
research reactors in the foreseeable future-
and there was an additional benefit of
validating with Neutron data:
Another observation with this paper was the benefit
of refining "separately" on the neutron and the X-ray
data; which displayed different information on
the iron/titanium metal sites. The total interpretation
of these results showing different local displacement
of Iron and Titanium in the M2 site.
A combined neutron/X-ray refinement would average
this information out. (refer Fig 7b of the J. Acta
Cryst paper)
Lachlan.
PS: The above results were not performed with a Fundamental
Parameters Rietveld (LHPM/Riet7/CSRIET was used). Though
a personal non-rigorous comparison of structure
refinement (isotropic thermals) using both an FP
Rietveld and a non-FP Rietveld did not give significant
differences in the final structure refinement. The version
of Koalariet (FP Rietveld) available from the CSIRO and CCP14
sites cannot generate Fourier electron density maps which I
would regard as a major asset when performing anisotropic
refinements and investigating disorder in many
circumstances. Plus this version does not have an option
for anisotropic refinement of thermals.
I would assume a FP profile model should be able to get
slightly more information out of the power pattern with
respect to attempting to refine subtle parameters such as
anisotropic thermals in a stable and reliable manner(?)
--
Lachlan M. D. Cranswick
4th Jan - 16th March 1999 Visiting Scientist
CSIRO Division of Minerals, Melbourne, Australia
Tel: (613) 9545 8500 (switch)
Fax: (613) 9562 8919 [EMAIL PROTECTED]
Collaborative Computational Project No 14 (CCP14)
for Single Crystal and Powder Diffraction
Daresbury Laboratory, Warrington, WA4 4AD U.K
Tel: +44-1925-603703 Fax: +44-1925-603124
E-mail: [EMAIL PROTECTED] Ext: 3703 Room C14
NEW CCP14 Web Domain (Under heavy construction):
http://www.ccp14.ac.uk
