I am not sure what all the fuss on background subtraction is all
about, as long as the uncertainties in the data are unchanged. We
have to model background in all powder methods and add it to the
computed pattern. If you subtract something from the obs pattern
before you fit, the only things that changes is the normalization
term in the R factors. Since the purpose of the weighted R-factor is
to measure the quality of different models to the same data, this
really makes no difference. The unweighted r-factor is pretty much
meaningless, so again who cares?
I don't love the idea of modifying data, rather than applying a
correction to a calculated value -- observations should be
sacrosanct, IMHO, but having said that, more than once have I applied
an absorption correction directly to measured intensities.
The important point is that if you measure 100 counts at a point, the
uncertainty in that is 10. If you decide that 95 of those counts are
background, go ahead and subtract them if you must, but the
uncertainty on the observation remains 10, and does not drop to sqrt
(5) -- this means you must compute uncertainties before subtracting
and use a Rietveld code that will accept "xye" input, as Pam pointed
out.
Brian
********************************************************************
Brian H. Toby, Ph.D. office: 630-252-5488
Senior Physicist/Materials Characterization Group Leader
Advanced Photon Source
9700 S. Cass Ave, Bldg. 433/D003 work cell: 630-327-8426
Argonne National Laboratory secretary (Marija): 630-252-5453
Argonne, IL 60439-4856 e-mail: brian dot toby at anl dot gov
********************************************************************
"We will restore science to its rightful place, and wield
technology's wonders... We will harness the sun and the winds and the
soil to fuel our cars and run our factories... All this we can do.
All this we will do."
