santu chakraborty wrote:
I am a senior research fellow working in the Department of Physics,
Jadavpur University, Kolkata700032, India. My area of research is
ab-initio structure determination via X-ray powder diffraction
techniques. Presently I am working with an organic compounds. I have
indexed the powder diffraction pattern in the Monoclinic system. The
program EXPO-2004 indicates P21/a as possible space group. But the unit
cell volume clearly indicates that the asymmetric unit contains two
molecules. Now the problem is that whenever I try to solve the structure
in “ Direct Space” by simulated annealing or parallel tempering methods
using the software “FOX” (after initial energy minimization by “MOPAC”),
it shows large no of short contacts between the two molecules the
asymmetric unit. Could any one please let me know the appropriate steps
to overcome the problem or if there are any other softwares which will
serve my purpose.
Dear Santu,
there are many things that could have gone wrong.
To answer your second question: yes, there are several other packages
that can solve crystal structures from powder diffraction data. FOX,
TOPAS, PowderSolve and DASH are probably the best known direct-space
packages. Of these, only FOX is free, but there are other packages, some
of them free (PSSP), some commercial (Endeavour). DASH is sold by the
Cambridge Crystallographic Data Centre.
But switching software package is not necessarily a good approach. It is
more important to understand if there is a good reason why the solutions
are wrong.
How good is your data? Is there preferred orientation, for example?
How sure are you that your unit cell is correct? Did you use Detlef
Hofmann's average atomic volumes (D.W.M. Hofmann (2002). Acta Cryst.
B58, 489-493.) to get a good estimate of the expected unit-cell volume?
How sure are you that your space group is correct?
How sure are you that your molecular model is correct? Did you miss out
a water molecule? A counter ion (a chloride)? Do you have the right ring
conformation?
How sure are you that you have given FOX enough time to find the global
minimum? How many degrees of freedom does the system have? Probably at
least twelve with Z' = 2; if it is over 20, you'll probably need some
patience (an overnight run or longer). If it's over 30, finding the
global minimum might become very difficult. How reproducible is your
wrong solution? (If it is very reproducible, it may well be the global
minimumwithin your model, which indicates that you model is wrong.) What
is the gap with the next best structure interms of goodness of fit?
I can have a look at your data (with DASH) if you'd like.
Best wishes,
--
Dr Jacco van de Streek
Research Scientist
Cambridge Crystallographic Data Centre
Cambridge, United Kingdom
