GND density calculation excluding grain boundaries?

[MTEXNewbie]

Is there a way I can calculate the GND density excluding any GNDs within a couple pixels of the grain boundary established by segmentation angle? This will enable to have GND in grain interiors without the influence of HAGB misorientation. The sample has around 100 grains.

Thank you.

[ruediger Kilian]

Hi,
actually it should be taken care of. The orientation gradient of EBSD points belonging to a grain boundary are set to NAN. However, this only applies to entire closed boundaries, so boundaries above the angle threshold but which were not closed during segmentation (grains.innerBoundary) can be an exception and one could think of a way overcoming this as well.

Do you have an example where this fails?

Cheers,
Rüdiger

[MTEXNewbie]

Hi Rüdiger,

Some basic clarifications first -

1. When we set a segmentation angle, all boundary segments with an angle higher than that defined misorientation angle are considered grain boundaries?

2. Orientation gradient of EBSD data points at those grain boundaries become NaN, and GND density is calculated by MTEX considering the grain interior orientation data?

3. If 1 and 2 are correct then it would be better to choose a low segmentation misorientation angle (<5 degrees) than the standard 15 degrees? This will help to identify the LAGBs (in addition to HAGBs) and set their orientation gradient to NaN, consequently excluding LAGBs and HAGBs from the GND calculation?

[ruediger Kilian]

> 1. When we set a segmentation angle, all boundary segments with an angle higher than that defined misorientation angle are considered grain boundaries?

If closed, they are grain boundaries, if not closed, they are stored in innerBoundary

> 2. Orientation gradient of EBSD data points at those grain boundaries become NaN, and GND density is calculated by MTEX considering the grain interior orientation data?

Yes. Have a look at the demo.

> 3. If 1 and 2 are correct then it would be better to choose a low segmentation misorientation angle (<5 degrees) than the standard 15 degrees? This will help to identify the LAGBs (in addition to HAGBs) and set their orientation gradient to NaN, consequently excluding LAGBs and HAGBs from the GND calculation?

Depends on the material and what sort of structures are actually represented by low angle boundaries. In the case you'd only want to see the effect of continuous curvature, that might be an option, however, your bulk result will obviously still depends on the segmentation angle.

Cheers,
Rüdiger

[MTEXNewbie]

1. All boundaries in my sample are closed, so all is fine there.

2. Apologies, would it be possible to point out at which line number the segmented boundaries are set to NaN?  - https://github.com/mtex-toolbox/mtex/blob/develop/doc/TensorAnalysis/GND_demo.m

3. My samples are metals that are already strained, this pre-strain will generate enough dislocations to form cellular dislocation walls evolving into sub-grains or LAGBs (<5 degrees). If I segment using 15 degrees, LAGBs present in grain interior will not set to NaN and will increase the calculated GND density for the bulk result. Hence the importance of excluding all boundaries (HAGBs and LAGBs) from the GND density calculation.

4. If orientation data at the boundaries are set to NaN by MTEX, then how is it possible for MTEX to calculate GND density at the boundaries? The image is taken from the demo - https://mtex-toolbox.github.io/files/doc/GND_demo_07.png

[ruediger Kilian]

> 1. All boundaries in my sample are closed, so all is fine there.

In mtex, all grain boundaries are closed by definition - those which are not closed are grains.innerBoundary

> 2. Apologies, would it be possible to point out at which line number the segmented boundaries are set to NaN? - https://github.com/mtex-toolbox/mtex/blob/develop/doc/TensorAnalysis/GND_demo.m

see here:
https://github.com/mtex-toolbox/mtex/blob/819cf3bded9076d028994951a3399844a86b2844/EBSDAnalysis/%40EBSDsquare/EBSDsquare.m#L63

In the demo, you can see that there are no crazy high gradients between different grains.

>
> 4. If orientation data at the boundaries are set to NaN by MTEX, then how is it possible for MTEX to calculate GND density at the boundaries? The image is taken from the demo - https://mtex-toolbox.github.io/files/doc/GND_demo_07.png

It's not the orientations being set to NaN but the gradient.

Cheers,
Rüdiger

[MTEXNewbie]

My confusion came from the colormap used in the demo, I thought the boundaries have highest GND density since it is white, whereas the boundaries are NaN so the white color means no data. This colormap shows it clearly - https://i.imgur.com/D87mBIz.jpg

MTEX excludes boundaries by default, so no need for additional steps. Many thanks to Rüdiger.

[MTEXNewbie]

Hi Rüdiger,

I get higher GND (average value) if I use segmentation angle 2 degrees instead of 2.5 degrees as used in GND_demo, is there a guideline what could be the minimum segmentation angle that can be used for FCC metals/alloys?

[Rüdiger Kilian]

Hi,
actually one would expect the opposite and that's what one can observe: the average gnd decreases with decreasing segmentation angle.

Cheers,
Rüdiger

[MTEXNewbie]

Hi Rüdiger,

I got it fixed, thanks. I was wondering what is being meant by The total dislocation energy at the end of GND_demo - https://mtex-toolbox.github.io/files/doc/GND_demo.html

The map at the very end of the document is a GND map or dislocation energy map?