At 10:45 AM 3/26/99 +0000, you wrote:
> I want to spend a few words about
>plotting data vs. Q.
I second Paolo's suggestion of plotting data in Q space. Besides all the
advantages he mentioned of Q space plots. I'd like to add one more minor
advantage, but one I use frequently for preliminary data analysis. That is
that manual or visual indexing in Q space is much easier (as Q(2h,2k,2l) =
2Q(h,k,l)).
I might go further and say that 2theta is not a very useful or even
meaningful quantity anymore. In the ancient days of D-S cameras, I can
understand use of 2theta as it was the direct observable. But in these days
of automated diffractometers and stepper motors the most primitive - i.e.,
the direct "machine" - observables are motor steps. I can see some
justification in plotting data as a function of motor steps - as it is a
direct observable, but why 2theta. Not that I am suggesting plotting data as
a function of motor steps. I am pointing out that 2theta plots are
visulally as unfriendly as motor-step plots would be. As we have decided to
transform direct observables -motor-steps - into some other quantity, why
not transform it into something universal as Q space?
I'd like to take this even a step further and suggest that not only should
we plot data in Q space, but even collect it in Q space. I think there are
several advantages to it. In fact, we often setup our scans for resonance
scattering in Q space. We find this very convient as the scan parameters do
not have to be recalculated for every change in energy.
But I'd like to extend this mode of collection for powder patterns for
Rietveld refinement. The advantage there is that the instrumental
resolution of most x-ray powder diffractometers I have used decreases (i.e.
width increases) at high angle as measured in 2th units. Thus, a constant
2theta width scan selected with step width to adequately cover the low angle
peaks is an overkill for the high angle region. It would be more profitable
to increase the 2theta step size at high angles and spend that saved time to
count longer. A constant Q step scan will accomplish this.
The question then is will the standard Rietveld codes accept data in const Q
space? (More specifically a question for Bob; does GSAS have - or how
difficult would it be to implement - a BINTYPE called constQ?)
Apurva
************************************************************************
Apurva Mehta
SSRL/SLAC
PO Box 4349, MS 69
Stanford, CA 94309
(650) 926 4791
(650) 926 4100 - FAX
************************************************************************
