Re: Sample height vs Zero point (continued)
Bob Von Dreele ( vondreele@popler.lansce.lanl.gov )
Thu, 02 Apr 1998 10:07:46 -0700
At 10:49 AM 4/2/98 +0100, you wrote:
>
>Dear All
>
>Further to the recent discussions on corrections for zero point errors
>and/or sample height, i did some comparison between the two refinement
>strategies (on data collected on a D5000 Cu-alpha1) and compared the
>results with one of the world best neutron diffractometers, HRPD at the
>ISIS facilty, UK. For this I used two samples, both high quality oxides
>prepared using standard solid state techniques.
>
>so the results:
>compound 1
>zero point 5.71356(6) 21.45782(30) 606.639(18)
>sample shift 5.71199(4) 21.45190(28) 606.138(15)
>HRPD 5.71200(1) 21.45223(7) 606.148(2)
>
>compound 2
>zero point 3.83386(14) 27.8166(12) 408.86(4)
>sample shift 3.83303(13) 27.8105(11) 408.59(4)
>HRPD 3.83393(1) 27.81471(13) 408.849(2)
>
>in both cases the zero point error was about 0.05deg 2theta, or 2 to 3
>step sizes.
>Of course the ESD's resulting from the measurement on HRPD are somewhat
>smaller than those collected on the D5000. But what is surprising is that
>for compound 1 the agreement between the two machines is *best* if you
>refine the sample height, however for sample two, it is vice versa.
>So, if we are trusting HRPD for a moment, than there is no *best*
>strategie based the results of this small number of compounds.
>What are we going to believe, what are the real SD's and how are we going
>to measure it? Friends what are your thoughts?
OK, my two cents worth on this. I run a TOF powder diffractometer (a lot
lower resolution than HRPD!) so I must comment that there can be a
systematic bias in lattice parameters determined from such instruments. The
problem is that the location of the scattering center in the sample on a
TOF neutron powder diffractometer is not easily determined (except by use
of internal standards) and this can create a small bias in the lattice
parameters determined on such instruments because of the small change in
the neutron flight path and scattering angle from sample to sample. So the
differences seen above may arise from a shift in scattering center between
the two TOF experiments. If I had to put my money on one of these results
as "most accurate", I'd choose the sample shift ones because the parameter
being refined most closely describes the actual physical situation; i.e.
the sample plane is displaced from the surface of the focussing circle.
This assumes of course that the diffractometer was properly aligned so that
the zero point was +/-0.001 or less. I might also add that peaks in
Bragg-Brentano diffraction patterns can have an additional systematic
displacement due to sample transparency for samples with low absorption. In
GSAS this is handled with the "trns" profile parameter.
Bob Von Dreele