The effect can be important (50%) in a fully ordered alloy or compound, compare:

A. J. BOURDILLONP,. G. SELF and W. M. STOBBS: Crystallographic orientation effects in energy dispersive X-ray analysis, PHILOSOPHICAL MAGAZINE A, 1981,VOL.44, No. 6,1335-1350

Use this link, if you have free access:

http://www.tandfonline.com/doi/pdf/10.1080/01418618108235813

Great!  Thanks!

Yes, EDS signal can be significantly affected by crystal orientation, as the work cited by Volker Klemm clearly showed.

The reason for this stems from the anisotropy of crystal structures (i.e. different ordering scheme of matter in the crystal structure depending on the crystal orientation). Therefore, crystals with very strong differences in their structure in certain orientations as compared with others, may be affected by large differences in the production of primary X-rays generated by the impinging electron beam in EDS analysis.

Such anisotropy can also be observed in the actual crystal morphology (or crystal habit). For instance platy crystals will certainly give different results if they are analysed parallel or perpendicular to the plates. The same may occur for fibrous morphology, etc. etc.

Generally speaking, equant crystal grains with cubic structure should present the least effect.

The answer, as for many other problems with EDS quantitation resides in the use of standards of known composition, similar crystal structure and crystal orientation as your unknowns.

In order to quantify the variation you may have in response of crystal orientation, try and analyse several grains with different known orientations and compare the results with models of the crystal structure. The aim is that of understanding the physics of the electron beam interaction in terms of depth distribution of the generated X-rays and their absorption paths before they reach the detector.