[sdpd] Re: Is it just awfully quiet these days?

[Armel Le Bail <armel... @fluo.univ-lemans.fr>]

Roger M. Sullivan wrote:

>       In the interest of stimulating some discussion and seeking some help
>let me lay out the problem I am working on.  I am a fourth year graduate
>student and this is the first crystal structure I have ever tried to solve.

In a sense, you are not lucky to start in this specialty by a so difficult
case.
So that you should not be ashamed to fail up to now.

I downloaded your data and I think I probably failed too, up to now.

Why is it difficult ?

1-  FWHM  ~ 0.13 °(2-theta) with 0.94 A wavelength, synchrotron data,
this is equivalent to 0.21° with CuKalpha, conventional diffractometer.
Knowing that  it is now possible to attain FWHM ~0.045° with a
conventional diffractometer, then, either your sample produces line
broadening, or your synchrotron (or detector) is inefficient.

2- Indexation difficulty is of course dependent of the resolution.
None of the three main indexing programs (TREOR, DICVOL, ITO)
 is able to propose a cell with figure of merit M(20) larger than 20. 
Discarding the very weak reflection at 16.50°, all three programs 
converge to propose an orthorhombic cell which can pass the 'cell 
constrained whole profile fitting without structure' test (Pawley or 
Le Bail methods). Below is the summary of the fit (using FULLPROF)
in P212121 :

 => Cell parameters      :
                             12.65496   0.00060
                              7.21407   0.00043
                             12.19336   0.00047
 
 => overall scale factor :    0.001000000   0.000000000
 => Eta(p-v) or m(p-vii) :    0.02780   0.00840
 => Halfwidth parameters :    0.00200   0.00000
                              0.00042   0.00000
                              0.01839   0.00015
 => Zero-point:  -0.0403  0.0005
 ==> RELIABILITY FACTORS FOR POINTS WITH BRAGG CONTRIBUTIONS:

 => N-P+C:  1499
 => Rp:0.909     Rwp: 1.59     Rexp:    6.05 Chi2: 0.691E-01  L.S. refinement
 => Conventional Rietveld R-factors ==>
 => Rp: 9.89     Rwp: 8.56     Rexp:   32.55 Chi2: 0.691E-01
 => Bragg R-factor: 0.154       Vol: 1113.178( 0.095)  
 => Rf-factor=0.154             
 => Run finished at:     Date: 30/08/1999  Time:
13:45:24.500                              

So, Rp < 10% (background subtracted)

The problem is now that this cell is almost C centered (a fit in a C
space group discards only weak reflections), and moreover, the a/b
ratio is near of 1.732 (square root of 3), so that many cells are also
suggested by the indexing programs with V/2=556 A**3 and cell
parameters like 7.28 12.19 7.21 beta=119.84 and so on.

With such a a/b coincidence, it is extremely difficult to decide of
the extinctions in the orthorhombic cell. Which space group is
the good one ? I don't know, and even I am not really sure that
the orthorhombic cell is the good one.

A scratch test with ESPOIR in P212121 with Z=4 (C4H8CuAlCl4, only
10 non-hydrogen atoms to find) produces Rp(F) = 17%. I did not look
accurately to this proposition which may be completely false, of course.

The conclusion would be : instead of wasting more time, try to
obtain a pattern with better resolution, since you have a pattern
with FWHMs more than 10 times larger than the best synchrotron
possibilities. But maybe this is not possible ? I think that this
is probably possible because the  image plate detectors have 
distinctly lower resolution than step by step or multidetectors
(BM16 or BM01 at ESRF for instance, provide FWHM ~ 0.01°).

Anyway, good luck !


Armel Le Bail - Université du Maine, Laboratoire des Fluorures,
CNRS ESA 6010, Av. O. Messiaen, 72085 Le Mans Cedex 9, France
http://www.cristal.org/