Juske,
One point you should know the background in a TOF powder diffraction
pattern is not high - that is a nearly constant background multiplied by
a strongly varying incident spectrum which is much higher at short TOF
than at long TOF. Do a divide by incident spectrum and you wll see
what I mean. Moreover, if you plot delta/sig vs TOF (in POWPLOT - GSAS)
the resulting plot should be uniformly noisy as a function of TOF - that
assures that the weights properly include the incident spectrum. If not
- talk to your instrument scientist. As for your chi**2 question - a
high value, assuming the difference curve shows that you have no
unsatisfied systematic error, can simply be from counting too long. If
the difference curve is OK the structures are chemically sensible, I'd
not worry about it (and reviewers shouldn't either).
Bob Von Dreele
R.B. Von Dreele
IPNS Division
Argonne National Laboratory
Argonne, IL 60439-4814
-Original Message-
From: Horita, Juske [mailto:[EMAIL PROTECTED]
Sent: Monday, March 12, 2007 8:09 AM
To: Rietveld_l@ill.fr
Subject: N-TOF
Dear all:
We're in the midst of Rietveld refinements of neutron TOF data
acquired at LANSCE NPDF. The NPDF gives very high backgrounds in low d-
values (TOF time), which drop quickly with d-value. After many
iterations, the fit between the data and model looks very reasonable,
including the background. The model is very close to what has been
reported before and it makes sense. However, the Chi^2 doesn't come down
close to one. We refined using the Le Bail method, and got a very
similar result, including the Chi^2 value (about 20). We feel
comfortable with our model, but a high Chi^2 is bothering. Reading the
literature, high Chi^2 can occur where data are collected to very high
precision. We think this might be our case, but how can we prove so?
Any comments and suggestions are appreciated.
Best,
Juske
*
Juske Horita
Chemical Sciences Division, Oak Ridge National Laboratory
P.O.Box 2008, MS 6110
Oak Ridge, TN 37831-6110
(865) 576-2750; Fax (865) 574-4961
For express-mail:
Juske Horita
Chemical Sciences Division, Oak Ridge National Laboratory
Bethel Valley Rd, 4500S, S214
Oak Ridge, TN 37830