A K-beta source may indeed be useful. In grad school I once used Cu K-beta to resolve the superlattice reflections in beta-brass. These Bragg peaks cannot be observed with Cu K-alpha (or most other radiations) because the atomic scattering factors of Cu and Zn are too similar, so the superlattice lines have negligible intensity. But Cu K-beta provides a stronger contrast between Cu and Zn (because the energy lies between the absorption edges of Cu and Zn), so you can see the correct Simple Cubic pattern instead of what otherwise appears to be BCC.

With a synchrotron this is a popular technique to sort out site occupancies, since you can select your wavelength to take advantage of anomalous scattering.

A K-beta source may indeed be useful. In grad school I once used Cu K-beta to resolve the superlattice reflections in beta-brass. These Bragg peaks cannot be observed with Cu K-alpha (or most other radiations) because the atomic scattering factors of Cu and Zn are too similar, so the superlattice lines have negligible intensity. But Cu K-beta provides a stronger contrast between Cu and Zn (because the energy lies between the absorption edges of Cu and Zn), so you can see the correct Simple Cubic pattern instead of what otherwise appears to be BCC.

With a synchrotron this is a popular technique to sort out site occupancies, since you can select your wavelength to take advantage of anomalous scattering.

In my lab, researchers often use K-beta radiation instead of K-alpha mainly because of "single" peak of the K-beta comparing to the doublet K-alpha1/K-alpha2, that can hide some small peaks.

You have to choose between intensity (less in K-beta) or to be sure not miss some peaks...

thanx to: Edward Payzant, Nicolas Guiblin, and Volker Klemm for the answers its really illuminate my mind

Cheers

@ Nicolas: This sounds interesting. Until now I thought that the quite exact knowledge of the alpha doublet profile combined with the higher intensity compared to beta, and the commonly used refinement procedures are that well-developed that you don't need a single profile, especially if you have to consider that also beta is no single line (as Volker pointed out) and contains a few with often less accurate intensity ratio. I can imagine that there are very special cases, e.g. as described by Edward, but they are usually not because of the doublet structure but because if the extinction behaviour... Or do you use the beta because of the shorter wavelength?

As i know, some researchers use K-alpha and K-betta both in experiment (conjoined). Than you can use the ratio of K-alpha intensity and K-betta intensity to eliminate the tool factor. In this way you will have dependence of ratio (K-alpha/K-betta) as function of the angle. But if you use XRR for example (or another measurement technique with full-profile analysis) you need a special software to simulate such kind of intensity (basically software simulate just K-alpha and K-betta, not a ratio)