Does anyone characterize non-cubic SiC-polytypes by EBSD? How would you define the necessary requirements?
A problem with distinction of polytypes (as in SiC) by EBSD can be that polytypism is often accompanied by layer faulting. And layer faulting will typically blurr the Bragg reflections which allow for distinction of the different polytypes (those with h-k unequal 3n in hexagonal setting; sometimes some special of these). This will also have its effects on Kikuchi patterns.... Depending on how "clean" your polytypes are... These reflections which are in common for all polytypes will give clear Kikuchi lines. Automatic indexing might give under certain conditions results which seem to be better than they are....
Gert! Would that be like SiC 6H & SiC 4H?
It is so (discernible) with XRD! Why wouldn't it be so with EBSD?
I'll be glad to be a "fly on the wall" for this discussion. I'm very interested :-)
Hi Ravi,
EBSD is a fantastic technique but the "required" speed is often very restrictive. In fact I guess you can separate between them. However, EBSD is very bad in lattice parameter sensitivity. 10% difference are practically not hard to see which is ridicules for XRD. Moreover, all high-order interferences superimpose so that superstructure reflections cannot be discovered (what is no problem for TEM and in principle also for XRD possible). Additionally, all these polytypes have (also in P-XRD) planes which generate interferences which appear nearly at the same Bragg position (or as band in a pattern). Since these are the most intense bands only the minor bands finally decide whether two phases can be distinguished or not...and in case of "high-speed" measurements they are not often ignored so that accidentally or sometimes also systematically another phase is "identified". You need to consider, that commonly only 5-7 bands are used for identification which is not much since there can be e.g. already 4 or more {110} planes appear with their band. For orientations absolutely OK, but for phase recognition at mutiphase materials hard on the edge.
high-speed should mean that you can index the patterns. For SiC I would say: 40...60 Kikuchi patterns per second.
That is high speed alright! What is the detectable change in crystal orientation that can be quantified voxel to voxel? Arc minutes, Arc sec, what? What is the approximate VOXEL dimension? "Tear drop" dimension for EBSD?
the angular resolution using images in the size of 160x120 pixels only cannot be that precise. One can show that the precision is typically around 0.5° what is for orientation measurements unnecessary high, especially since often a "smearing" of orientation is used in a postprocessing to change from discrete orientation is a orientation distribution function. Commonly the FWHM of the used Gaussian is around 10° (!). For misorientation measurements, however, the precision cannot be high enough. Pattern matching algorithms instead of the commonly used Hough transform for band detection is clearly better and can push the precision at least a factor two using the same pattern, but it is also clearly slower. Why only 160x120 pixels? :-) It is all a question of time and intensity. A higher resolved image needs more time for the acquisition in similar quality and a clearly higher processing time for band detection. The acquisition scales with the square whereas the Hough transform is even more time consuming. Therefore, even higher resolved images are commonly not much better in orientation determination since the the applied Hough resolution is not scaled in the same precision as recommended by the higher pattern resolution. Problematic: this is not clearly recognizable by the user so that the disappointment is quite big if the investment in a higher pattern resolution does not result in any improvement of orientation precision. The term voxel is used for 3D-EBSD using serial sectioning. Because already in 2D the drift during the EBSD investigation is already non-predictable and very inhomogeneous, you can images that merging serial sections is quite speculative and needs a adjustment not only of adjacent sections but even by adjacent voxels. This is one reason why 3D-EBSD data often only look nice and impressive, but a quantification of data regarding a 3D-orientation investigation is because of the inaccuracy of x,y,z of each collected orientation at least questionable. In so far: expensive technique, time consuming, and often overestimated regarding possible conclusions. However, nice to look at and very impressive to demonstrate at conferences or when visitors coming :-).
A problem with distinction of polytypes (as in SiC) by EBSD can be that polytypism is often accompanied by layer faulting. And layer faulting will typically blurr the Bragg reflections which allow for distinction of the different polytypes (those with h-k unequal 3n in hexagonal setting; sometimes some special of these). This will also have its effects on Kikuchi patterns.... Depending on how "clean" your polytypes are... These reflections which are in common for all polytypes will give clear Kikuchi lines. Automatic indexing might give under certain conditions results which seem to be better than they are....
- Badges
- Science topic
- Similar topics
- Physical Sciences
- Materials Science