Often EBSD maps are performed using "integer values" like 50nm for step width (or pixel dimension) instead of 53nm. Is there any meaningful reason why one has to do this or is this simply a matter of habit? Does it have any meaningful influence if grain size determination is carried out using for one map e.g. 50nm and for another map 53nm?
To my knowledge there is no real reason behind this. As long as the step size is at least approximately 5 times smaller that the grain size, 3 nm is negligible.
The step width should be adjusted depending on your use of the EBSD result.
If your interest resides only in the fabric (what cystallographic preferred orientation), you finally use one point per grain, so you will use a step as small as possible so that the measurement takes as little time as possible (=> step 3-5 times smaller than average grain size is good, especially if grain size is uniform).
If you are particularly interested in shape fabric, grain size distribution and other stuff where the map needs to be precise enough to nicely reflect the thin section, 5-8 times smaller than grain size is more adequate (depending on your field of research I presume. I am into geology).
Why odd numbers such as 53 nm instead of 50 or 55? Sometimes it is just a matter of using the time the machine is available, with the most efficient and practical way. If you start measurements at 10PM and want to collect them the next morning at 7AM, you should calculate what is the best step for the measurements to finish somewhere between 6:30 and 7AM and not at 4AM so that the electron beam does not stand 3hours above the same point as it may alter your material (you may end up with a map more precise than really needed but that is not a bad thing)
Just a comment on David's answer. A very prolonged data collection in EBSD can lead to unreliable EBSD result, which is due to beam instability of SEM in too long imagings.
BTW, in updated versions of data collection software you can adjust in the software that when the scan is finished, turn off the HT Gun in the SEM (At least OIM (TSL) has this option), so you do not damage your sample because of extra beam-ing.
As long as I know Channel5 (I guess from 2002) there was also an option to switch off the cathode (W-SEM) or the high voltage (FE-SEM). Bruker even offers an automatic retraction of the detector with switching off the camera. I am not sure what Thermo is doing in their system but I would assume that they all learned from their EDX mapping function or feature analysis (e.g. GSR) and are all able to control the SEM.
My question was mainly related to the often existing opinion that one should always the same step with for grain size determination, but as David already mentioned, it is actually a question of the microstructure and actually needs to adapt.
Although from a practical point of view there is no need to stick with rounded step size values, any change in step size will affect your error budget. So if you are applying some error analysis on your data, this will be affected by a change in step size which might complicate comparing different data sets. So if you are going to compare two microstructures based on differences in the morphology or the size of the grains, I would recommend to stick with the same step size.
@ Frederik: you certainly implement some uncertainties, and I agree that one prevents any problem which is related to the step size if one does not change it, but I wonder whether you are running in another problem when the grain size is changing by a factor which is in the scale of 2 and more. I want give an example: if you select a step width of x and your grain size is in the scale of 5x, are the data not only comparable if the step size is adatpted as well, i.e. your grain size would be 5y so that you should change the step width to y? I know, the problem is that you don't know the grain size before the scan. But it means if you scan the grains with approximate grain size of 5y you also implement an error if you scan with step width x. As long as y and x are not very different I believe that it will not affect your results remarkably. But if your grain sizes are remarkably different (e.g. 5 times bigger or even more), does is makes sense to measure always with x (or y) since you did this always? I think not... If I should be right, the conclusion would be that the step width should be adapted on the grain size as David already proposed. The error (standard deviation) is anyway that big that effects resulting from small changes in step width are perhaps neglegtable. Let me give a practical example: If I would have two samples (one before, one after recrystallization) I would try to select magnifications which visually reflect a similar number of grains for both microstructures. Using the same pixel resolution of the image (and the EBSD scan) the respective pixel size is then already adapted so that one can use this pixel size already as optimized step width, no matter which number is given (e.g. 50 or 53. To change then the step with to a common one (regarding a sometimes given statement: "We always use 5µm for grain size measurements!") is from my point of view not useful. What do you think?
@ Gert: I agree with what you are saying! I was just trying to introduce some caution in that there might be an influence in certain specific cases, especially if there is some absolute value parameter that is used to base conclusions on. If you have very different step sizes, at some point, the results from the small step size measurement might be in the uncertainty range of the results of your second large step size measurement (even if the step sizes themselves seem appropriate based on a simple resolution style reasoning), leading to an inability to draw any statistically meaningful conclusions.
- Badges
- Science topic
- Similar topics
- Physical Sciences
- Materials Science