The <B>DuPont Challenge</B> has been won!  Unfortunately, the lessons learned
are not the ones that we set out to learn, namely "are there better techniques
that can be used to solve structures from poor quality powder diffraction
patterns?".  First let me congratulate <B>Leonid Solovyov and Sergei Kirik</B>
 from the
Institute of Chemistry and Chemical Engineering, Russian Academy of Sciences in
Krasnoyarsk, Russia, on their solution which fits both the synchrotron and
neutron data with Bragg R values at approximately the 5-6% level as presently
refined.  Basically, the stoichiometry of the compound was found to be
incorrect: the stoichiometry is not HAlF4, as alleged by the chemist, but 
PyH(Al3F10), i.e. 1/3 of a pyridinium fluoride molecule per Al. The structure 
consists of [Al3F10 (-1)]n layers (which by the way several entries reported 
correctly) separated by layers of pyridinium counterions. The unit cell was 
essentially that given in the announcement and the space group remains C2/m, 
assuming that the pyridinium lies in a crystallographic mirror plane and is 
disordered about a crystallographic 2/m site.  The [Al3F10]n layer does 
consist of both edge- and corner-sharing AlF6 units as suspected earlier. 
The atomic coordinates as presently refined with GSAS are given below:<P>
<pre>
    TYPE     X       Y       Z      FRAC  NAME      UISO

   1 AL   0.00000 0.00000 0.00000 1.00000 AL(1)    0.00893 
   2 AL  -0.06940 0.50000 0.10612 1.00000 AL(2)    0.01353
   3 F   -0.06736 0.20284 0.09729 1.00000 F1       0.01954 
   4 F    0.20749 0.00000 0.10936 1.00000 F2       0.00299
   5 F   -0.00960 0.50000 0.29383 1.00000 F3       0.04092
   6 F    0.13722 0.50000 0.07366 1.00000 F4       0.00488
   7 C   -0.18553 0.00000 0.45726 1.00000 C1       0.06713
   8 C   -0.07358 0.00000 0.36959 1.00000 C2       0.06460
   9 C    0.08740 0.00000 0.40801 1.00000 C3       0.05691
  10 H   -0.28904 0.00000 0.44136 1.00000 H1       0.04121
  11 H   -0.16309 0.00000 0.25294 1.00000 H2       0.26650
  12 H    0.14114 0.00000 0.31076 1.00000 H3       0.08751
</pre><P>

At this point, the contribution from the N atom, which probably is disordered
over the 3 carbon sites (note Uiso's for the 3 C atoms are all the same) has 
not been included in the calculated structure factors.<P>

        Although I, and many of you who attempted to solve this structure,
would probably like to "impeach" the chemist at this point, let me point out
that the TGA weight loss from the PyHAlF4 precursor on heating is quite
consistent with the formation of HAlF4, and that the loss of Py was noted
in the TGA-MS which was used to study the off-gases. Unfortunately, the Py is
being evolved from the precursor right up until the pyH(Al3F10) material itself
decomposes and was thought to be the source of the Py observed in the TGA-MS.
Also, why would one run a C/H/N analysis on HAlF4 - of course it is going to 
be contaminated with some pyridine material from the precursor!???  Well, now 
we know why!  In fact, the C/H/N analysis has now been completed and the 
results are consistent with the PyH(Al3F10) formula given above.<P>

   So, there are no lessons to be learned except the one that
crystallographers have learned over and over : never trust a chemist!<P>

Congratulations again to Leonid Solovyov and Sergei Kirik!<P>

Richard Harlow (and a chastened Norman Herron)
CRD, DuPont