There are 10 methods at least for Kalpha2 elimination, almost all published
in the J. Appl. Cryst., none seems to work really better than the simplest
which is the so called Rachinger method (Phys. Rev. 37, 1931, 154-156 and
J. Sci. Instrum. 25, 1948, 254-255) for alpha1-alpha2 separation
(deconvolution
is not appropriate).
In fact, the Rietveld method should always be applied to the original pattern.
This means that if you really dislike alpha2, then use an incident beam
monochromator or synchrotron data. Otherwise fit with the alpha1+alpha2
sum (and the software adds them, it does not convolute them).
Armel