Gentlemen:
After the last long period argument, I can conclude the key is "if
you prefer the success to the truth when you do SDPD or Rietveld refinement,
particularly for using restraint (or contraint)". Unfortunately, in fact,you
can be happy for structure solving with restraint but you can't say it is true.
On the other side, you still need to go on after structure solving without
restraint, because we all know others can get the same or better R factors to
give so called reliable structure but here or there different from you, without
restraint,too! After all, when we obtain a reliable result, we don't sure this
is the best and the truest for every parameter! There is not a STANDARD R or
e.s.d value to tell you that you reach the end!
Most of people agree the accuracy of single crystal diffraction, but
here I would like to cite the words from Sheldrick (the author of Shelx97 and
2003)"with the right restraints, you can fit an elephant to any data". (Crystal
structure refinement: a crystallographer’s guide to SHELXL,P. Müller, editor,
P199). So, if you only need the success, you can use as many restraints as you
like, even more than the data, Not problem! Restraint is really a shortcut for
structure solving.
But if you prefer for the truth, sorry, you have to certificate the
restranited structure by other characterization when you think you can't obtain
them directly from PXRD.
So it is meaningless for us to say allowable or not to use
restraint, or for the dicaussion of the futurute of using restraint or
unrestraint in SDPD. The thing we only need to do now is to remember and
re-hold the gold rule of Rietveld refienment: if you can obtain the accurate
information from other tests, you must not use refinement for them. In fact,
this is also a CERTIFICTATION rule for the restraints you used!
In all, we are not the Gold, so it is not wisdom for us to stop
others to the success as well as make them only for the truth! After all,a
"successful" structure can be published easily if the eidtors and referees
agree, only tested by several people before go to the databases! But the true
structure migh wait for seveal decades of hard work!
Well, I think it time for me to go on the strtucutre of Bi4Ge3O12
now. Yes, I can obtain the so called standard strcuture from ICSD, but till
now, it is still not clear because the recorded structure doesn't including the
antisite defect,which affects the S.O.F of Bi and Ge from single crystal
diffraction, if I can or must restraint the S.O.F to 1 when I try it by PXRD or
neutron data? :-))
By the way, a "successful" structure is of course better than
obtaining nothing, do you agree?
Good luck!
Haohong
= = = = = = = = = = = = = = = = = = = = = =
Haohong Chen
Shanghai Institute of Ceramics
Chinese Academy of Science
China
= = = = = = = = = = = = = = = = = = = = = =
======== 2013-08-09 00:22:00 您在来信中写道： ========
Dear colleagues,
Since the restrained SDPD future has been announced in the respective list
without a chance to go back:
http://tech.groups.yahoo.com/group/sdpd/message/3016
It's my turn to put forward an alternative unrestrained variant that seems a
bit more interesting and not so hopeless, IMHO.
This variant consists in a more accurate approach to the experiment and the
modeling instead of compensating the inaccuracy by restraints.
In the discussion of the restrained Rietveld:
http://dx.doi.org/10.1107/S1600536813017522
proposed as an instance of the SDPD future trend, I presented an unrestrained
variant of the refinement:
http://tech..groups.yahoo.com/group/sdpd/message/3023
and noted that Future improvements are possible if one exerts further efforts
to the sample preparation, measurements and the refinement model adjustment.
Spending an extra couple of hours for the data analysis, an additional
micro-admixture has been identified whose inclusion into the model resulted in
further notable improvements of the structure geometry:
http://sites.google.com/site/ddmsuite/home/vn2075-DDM.cif
http://sites.google.com/site/ddmsuite/home/vn2075-DDM.png
Characteristic distances and their expectations:
Pos. Dist. Exp.
Co-O1 1.92(2) 1.91
Co-O2 1.94(2) 1.91
Co-N1 1.98(1) 1.96
Co-N2 1.95(2) 1.96
Co-N3 1.92(2) 1.96
Co-N4 1.99(1) 1.96
C1-O1 1.31(4) 1.32
C1-O2 1.34(4) 1.32
C1-O3 1.25(3) 1.23
N5-O4 1.27(3) 1.24
N5-O5 1.23(3) 1.24
N5-O6 1.25(3) 1.24
All distances and intramolecular angles are in agreement with the expected
values within their uncertainties. Future improvements are still possible,
being determined mostly by the sample preparation and measurements now.
Of course, in REALLY complex and HOPELESS cases a minimal number of constraints
may be imposed if the refinement convergence of atomic positions is REALLY
impossible without them. In the above example, rigid constraints were imposed
on H-atoms, which is, though, not always necessary.
The use of soft restraints should be avoided as much as possible since a
generally acceptable methodology of their safe application has not been
developed yet.
Best unrestrained regards,
Leonid
*******************************************************
Leonid A. Solovyov
Institute of Chemistry and Chemical Technology
660049, K. Marx 42, Krasnoyarsk, Russia
http://sites.google.com/site/solovyovleonid
*******************************************************
++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++
Please do NOT attach files to the whole list <alan.he...@neutronoptics.com>
Send commands to <lists...@ill.fr> eg: HELP as the subject with no body text
The Rietveld_L list archive is on http://www.mail-archive.com/rietveld_l@ill.fr/
++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++
