One possibility is to remove the ferrite lamellae at the surface by deep etching. This will facilitate  indexing the remaining cementite lamellae.

It is definitely not related to the structure factor of cementite since in case of isometric grains, e.g. in iron meteorites, the patterns look perfec, cf. the attached imaget. For meteorites cementite is called cohenite and this phase forms perfect grains as visible in the attached FSE image (left side). Unfortunately I do not find a high resolution pattern but one with a standard resolution of 160x120 I attached as well.  

I guess that the perlite lamellae in steel are simply too thin and remarkable overlayed by the surrounding metallic matrix, especially if they are not vertically aligned. I would recomment you to further test on vertical  lamellae exclusively, or a try of TKD. This would be a nice opportunity to prove how beneficial TKD really is. 

As metionned by Gert TKD could be efficient to index cementite. We have planned such experiments by the end of the year with our new EBSD system.

But I think "EBSD" in TEM with systems such as ASTAR could also be very efficient! We already performed such experiments with Edgar Rauch system on large round cementite particles (diamater of around 0.3µm) and it was working!

Did you try long final polishing using colloidal silica under low pressure instead of slight etching? This may avoid the formation of oxide scale on cementite lamellae while dissolving the surrounding ferrite matrix at the same time.

I had successfuly experienced a quite similar structure made of fine lamellea Cr-rich M23C6 carbides within an austenite matrix (that had been dissolved during the final super-polishing). The carbides having less than 0.5 µm thickness were very well indexed.

@Jerome: I guess we all have a somehow different understanding of "well indexed". Could you post a map, the system you used and some other experimental conditions you were using for this map (step width, acceleration voltage, collection speed, frame averaging etc.)? Nevertheless, I agree with you that etching is perhaps not the best solution, already because of the formed topography.

@Gert: Thanks for your recommendation. I am sure TKD would benefit in this case, but it needs new software adaptation and kind of TEM sample preparation, which is not possible at the moment for us. That is why I am trying to do it by normal EBSD, if possible. Your suggestion on focusing on Vertical Lamellae seems promising. Will try.

@ Jerome: I have not used etching. Interestingly in our material without etching I could see lamellae structure. Anyways, prolonged OPS polishing usually sounds good. But as Gert metnioned, it can cause edging effect at the cementite lamellae and thus it might not be possible to get a "flat-cementite" lamellae. But worth trying once! Thanks.

Worth noting to everybody that to show the lamellae structure in the SEM image, I normally need to slightly etch the samples to reveal the structure. Although this sample did not need etching.

@Volker: So you mean cementite does not contain any LAGBs? Then how does this phase bend? without dislocation arrangements?

Hi, the difference between EBSD and TKD is not worth to mention. At least I do not see any point since the only difference is that the sample is not or only slightly tilted. The signal formation is exactly the same (only on the back side of the sample), takes a bit more time (since you loose a part of useful electrons for the diffraction phenomenon), and the pattern center could be a bit higher than usually applied. Therefore there is no special software required. The only thing is that the orientation determination is incorrect since the 70 degree tilt is wrong and/or the sample orientation is unknown.

There is even not TEM sample necessary if you are able to thin a sample (as used for TEM before the invented FIB preparation). You could even try a wedge shape preparation. I am sure there are many old samples still available which have been investigated 10-50 years ago by TEM and waiting for TKD testings.

@Ehsan: Only because a lamellae looks bended it must not change its orientation. It can but it doesn't has to. In so far one should be careful with such kind of conclusions. However, also the opposit may happen: you are changing the direction of the driving force (e.g. temperature field) and you can play with the orientaion of the crystal. But even in this case the crystallization speed is imortant. It is a very sensitive construct :-). However, as you mentioned, then dislocations are required since the translation symmetry for very long distances is lost. Folloing strict crystallographic definitions this shouldn't even be a crystal, but technically it is. It is like for the definition of grains for EBSD. There huge accumulated orientation changes can appear and it is still assumed as single grain although waves of different places would never interfere anymore since there coherence is lost. 

@Gert: Thanks for the info. Can you please introduce me a reference for the TKD method you mentioned? Seems worth trying.

Brodusch, N.; Demers, H. & Gauvin, R. Nanometres-resolution Kikuchi patterns from materials science specimens with transmission electron forward scatter diffraction in the scanning electron microscope. J. Microsc., 2013

or

Trimby, P. W.; Cao, Y.; Chen, Z.; Han, S.; Hemker, K. J.; Lian, J.; Liao, X.; Rottmann, P.; Samudrala, S.; Sun, J.; Wang, J. T.; Wheeler, J. & Cairney, J. M. Characterizing deformed ultrafine-grained and nanocrystalline materials using transmission Kikuchi diffraction in a scanning electron microscope. Acta Mater., 2014, 62, 69-80

or

 Keller, R. & Geiss, R. Transmission EBSD from 10 nm domains in a scanning electron microscope. J. Microsc., 2012, 245

Hi Ehsan.

I've read some papers in this field, maybe useful.

Y. D. Zhang et al, New insights into crystallographic correlations between ferrite and cementite in lamellar eutectoid structures, obtained by SEM–FEG/EBSD and an indirect two-trace method, Journal of Applied Crystallography, 2007.

N. GUO & Q. LIU, Back-scattered electron imaging combined with EBSD technique

for characterization of pearlitic steels, Journal of Microscopy, Vol. 246, Pt 3 2012, pp. 221–228.

Thanks Habib;

Nice papers, but non of them reliably indexed the Cementite phase itself.

Thanks for sharing your interesting paper Farangis.

Have you tried this with the new Oxford detector that is much faster? (well... I have not tried the detector myself; they can say it is 2-3 times faster but is it with sacrificing the pattern resolution or not, I dont know...)

Farangis Ram, did you manage to get any nice results from the line scans?