absolute orientations obtained from some EBSD systems

[ruediger Kilian]

Dear list,
we posted the following mail on the geo-tectonics mailing list to make users aware of some inconsistencies of orientation and corresponding maps obtained on certain ebsd systems.

Because here on the mtex list, it was also previously asked how to import .ctf or .crc into mtex and how it should look like, this might be relevant.

In short: Depending on you choice of ‘xAxisDirection’ and ‘zAxisDirection’ mtex does a perfect job in correctly importing the data. However, as in our test cases, this also means that the pole figures in mtex will appear oriented differently (but correct) from the way as they appear in the Aztec gui (and if not taken care of also in Channel5), as the latter do not necessarily represent absolute orientations, i.e are not related to the map.



Dear list, especially EBSD users,

at the beginning of 2015 we became aware of certain inconsistencies in EBSD data e.g. compared to U-stage data. EBSD measurements on an obliquely cut quartz single crystal used as a reference sample indicated that there was a rotation of orientations by 180° around the sample normal at this particular EBSD system (Aztec/Oxford).

Subsequently, Michel has tested this issue on four different Oxford EBSD / SEM systems (see end of the mail), also with a quartz single crystal. All tests confirm that the derived crystal orientations (e.g. represented by pole figures) for each system are rotated (180° rotation around the normal to the sample surface/center of the pole figure) with respect to the EBSD-map and scanning electron image (EBSD-map coordinates and scanning electron image are consistent to each other).

Now we are afraid that this orientation rotation might also exist for other EBSD systems of HKL Channel-5 Flamenco and Oxford instruments AZtec, in combination with different SEM systems (SEM manufacturer). There was an EGU poster of 2014 (EGU2014-3057) where a benchmark quartzite sample was measured in different EBSD laboratories and by optical methods, it turned out that the involved Oxford EBSD system (none of the above) also produced orientations that were rotated in comparison with all other results. There are two additional laboratories who confirmed this behaviour in their setup.

The issue arises from the fact that the EBSD and the SEM do not necessarily follow the identical conventions and, additionally, that the EBSD setup seems not to be checked to produce absolute orientations (pole figure presentation is consistent with orientation map/electron image) during installation. Oxford Instruments confirmed to us that they will provide a set of documents with better explanations of the internal definitions of their machine setup to all their customers. This will hopefully make users and operators aware of the possibility that their orientations might not necessarily be absolute orientations and might have to be rotated by 180° or 90° around the sample normal.

In several use cases such data can lead to erroneous interpretations. Hence we simply want to raise awareness of the issue and would also be interested to see if other laboratories have done any of these tests and would be keen sharing their experiences (also with OIM – former TSL; Bruker etc.). If you are operating an EBSD facility and you are in doubt, it might be worth checking with a reference sample that is capable to detect those rotations (e.g. similar to the test-sample Michel has used). In case you detect the issue, please try to inform all your users that have been collecting data in the past. Nothing is worth spending heaps of time, drawing conclusions, maybe writing a paper based on ill-defined data. Maybe it might be worth checking old data sets as well.

There will be a poster (EGU2016-8221) at the EGU this year (17-22 April, Vienna) discussing the implications of this issue and what kind of interpretation derived by orientation data will be affected, so if you'd be interested, we'd be happy to have a chat there.


Best wishes


Michel, Rüdiger




The tests with a single quartz crystal (glued with one of the rhomboeder {r} planes onto the glass holder. C-axis dips with ca. 37.5° towards South-East; lower right corner of the thin section) were carried out on the following systems.

SEM Oxford Instrument Acquisition software

1. Tescan (Vega) Flamenco

2. Zeiss (Cross Beam 1540Esb) Flamenco

Fast Acquisition software

3. FEI (Helios 600i) AZtec



4. Zeiss (Merlin) AZTec

Indirect check (same microstructure was scanned with Zeiss (Cross Beam 1540Esb - Flamenco)

All test data and a sketch of the sample orientation can be downloaded from following link.

https://faubox.rrze.uni-erlangen.de/dl/fiGutJDktC35dVXL73QHhZvR/EBSD_symmetry_test_compiled_Sept-2015_180rotation.zip



—

Rüdiger Kilian
University Basel
Basel - Switzerland

Michel Bestmann
GeoZentrum Nordbayern
Erlangen - Germany

[bjørn eske sørensen]

Dear Ruediger

This is an important issue. We are using a TSL system for 'online' collection  here at NTNU and our own software (NORDIF) to collect 'offline' data, which are then 'offline' indexed with the TSL software. We would gladly test out our system with a similar setup as you. We are also running a workshop on EBSD in june, related to the scandem conference and it would be quite relvant to show such a test then.

Sincerely yours 

Bjørn Eske Sørensen

Norwegian University of Technology and Science

[ruediger Kilian]

Hi Bjørn,
if you have a look at the geo-tectonics list thread, some people have had this issue also on TSL, but basically related to software updates which had overwritten the preferences.
In Aztek systems most of the misunderstandings come from the fact that at first, orientations are related to the ebsd-camera reference system (“seen from the camera”) while the map is presented form the beam view (most of the time looking down the sample, top edge of the image is lowest part of sample) and the user has to take care that things are rotated later on into a coherent reference frame. There are certainly systems that do it differently. However, as the manual was more then “cryptic” - as some might call it - we figured doing a test is actually the easiest way to find confidence in the data.It just has to be a reference sample that is sensitive to this sort of rotation.
Cheers,
Rüdiger