Hello I am also interested in it. As the EBSD suppliers claim that their device is able to distinguish two (same type) lattices if they differ at least by 5% in parameter. I doubt that it is really possible in in many real specimens. So I would not expect so precise value of lattice parameter.

Thanks Jaroslav, I guess their statements are more related to a) perfect pattern in b) high resolution, and c) for single pattern analysis, i.e. not during a map. For a map I would say, that even 10% sounds challenging. The tendency perhaps, but not a real discrimination. But my question was also related to single patterns. 

Hi Gert,

I believe, it can be easily better than 5% using something like internal standard (something, on which you measured lattice parameter by XRD). When mounted in the same holder as sample, correction on the center of gnomonic projection can be done for both material and standard. Using the same voltage and distance between the sample and the detector being calibrated through the standard, lattice parameter of the sample can be established. Theoretically, I think, it must depends on resolution of Kikuchi pattern - the angular width of pixel.

Practically, EBSD in Hough transform evaluated mapping it is insensitive. The lattice parameter of ferrite can be lowered about 1 angstrom without important change of evaluated points.

Hello,

Yesteraday I tried silicon sample mapping and it was sometimes indexed as cubic titanium lattice parameter 5.43 vs 3.29. But 4x4 binning was used and the system is quite old. but as yout told Gert in this case 10% seems to be challenging. Maybe for single pattern purpose larger detector distance could be usefull because it may increase the angular with of pixel (as Jaromír suggested).

Which system you are using? If Channel5 there is a way to use band widths (TSL also should have it). Otherwise every system ONLY uses the band position, i.e. a increase of the resolution doesn't matter since for all cubic phases the band positions are identical. Only if the visibility of bands is different, the system easily differentiates them (ferrite, austenite). You can also index silicon by austenite or another fcc phase.  If you have e.g. nitrides in your steel (e.g. TiN) you cannot separate it without EDS if austenite if it is in the phase list.

In Chanel5 it is called O-lock. The indexing takes longer but it might be possible to differentiate roughly Si and Ti. (I guess gamma-Ti which is actually tetragonel :-)... Ok, pseudocubic) 

Yes we have Channel 5, but it was in 4x4 binning so the accuracy is limited. Anyway as I told the 5% (or 10%) resolution is challenging from my point of view even when lower binning is used. Maybe as Jaromir suggested the angular with of pixel could be improved by larger detector distance  (or maybe lower acceleration voltage). For single patterns I guess it could be Ok.

The larger detector distance does not help that much because of the band profiles  which are even in ideal case (simulation) "blurred", i.e. you don't see more since you only spread your blurred profile. Moreover, there are many deviations between visible band edge and a geometrical approximation assuming a hyperbolic function. For phases forming very sharp band edges (low Z elements which reduces the dynamical effects dramatically) it might, however, help. But then I am not sure how much the small solid angle limits the lattice parameter accuracy. It is similar as for indexing...you don't see that many zone axes anymore. 

With HOLZ rings you can even see directly the lattice vector length, but...a) there are only some phases which produce pronounced HOLZ rings, and as Langer and Daebritz showed, even in a single patterns HOLZ rings of different [uvw] end in (slightly) different lattice parameters. Therefore, I am not sure whether HOLZ rings are the best solution, obviously not a general one. Where it works, fantastic... no doubt!

Hi Ben, how you would estimate the accuracy of EBSD, and from which number you would call it useful? If you are searching in a database for a phase (just try AMCSD) how much you would setup the tolerance for what you would still accept a hit as possible candidate? If you look for solid solution (many alloy but also a huge number of minerals) for the same class of material a surprisingly wide range of lattice parameters exists. And if I look for the databases delivered together with EBSD systems... are they really useful because you can select any phase available?

There are numerous applications where we perhaps do not need the accurate lattice parameters as delivered in XRD but only a good guess because a) the composition is anyway different, the phase is not yet described in crystal structure databases (only in PDF with highly speculative intensities), or simply to get a hint which symmetry this phase has. Especially for small precipitates you can get a nice Kikuchi pattern but the EDS signal is more from the matrix and therefore not very trustful.

I know, Phase ID for EBSD is not simple, but a really good texture analysis is (from my point of view) much more complicated if we really want formulate reliable conclusions from a few tens or hundreds grains. In comparison, the interpretation of a single pattern seems to be easy, and any conclusion is probably not worse than an orientation map containing wrong or not-identifies phases and a questionable texture interpretation.

No doubt, any high accuracy is always welcome, and lattice parameter determination is no exception, but hundred years ago people even couldn't measure them and used lattice parameter ratios which are perhaps more useful for an identification of a mineral class. And ratios are quite nicely to determine from EBSD, or am I wrong?

Do we really need the absolutely exact lattice parameters?

Dear Gert Nolze,

My contribution to this discussion follows a side track.

1. For the study of preccipitation/transformation the study of lattice parameter changes, e.g. as a function of time, may provide valuable information.

2. For such purposes, high accuracy values of lattice parameters are more than welcome.

3. DeBije-Scherrer photographs combined with Nelson-Riley extrapolation provide lattice determinations with high accuracy, a tool that is often overlooked.

References:

1. Ageing of liquid-quenched ans solid-quenched Aluminium base alloys: analysis of lattice parameter variations, i.e. my thesis available via ResearchGate

2. a text book on X-ray diffractometry, e.g. H.P. Klug & L.E. Alexander, X-ray diffraction procedures, latest edition (2nd ed. John Wiley and Sons, New Yorik, 1974)

https://www.researchgate.net/profile/P_Mourik/publications?pubType=thesis

Hi Ben,

you are right. Threading a question is somehow a pain. I also often miss reactions on my answers. 

Which TEM technique you have in mind? I know, CBED is very sensitive but not always possible. SAED? How accurate this can be?

EDX: we all know the limitations so that it is some kind of discriminator but accurate information especially for small precipitates in a matrix...

XRD: Well, micro-XRD...but also there the problem arises, who has it and what are you doing with small precipitates of less than 1vol%? TEM again :-) ?

Thanks for your answer and best greetings

Gert