15 years after the introduction of CCD-based EBSD detectors a new camera technology enters the EBSD world. It is hard to believe, the formerly demonised CMOS chips are now the saviours of the EBSD market. In contradiction to all prevously made statements CMOS are now not only faster but also more sensitive and less noisy. Well, companies are (primarily) not founded to make scientists happy. They need to sell in order to survive (or make their shareholders happy), and if the market does not increase as expected a new technology is very welcome to increase the business ono this way.
In fact, I believe that the spec sheets of the new detectors are credible, but how the detectors really work in practice? Are there some users which have direct comparisons? How profitable is the change if one considers the comparatively huge price?
Dear Gert, I had a privilege to test the Symmetry detector from Oxford instruments recently, on the same samples that I did the analysis with our old CCD system. The first impression was: that's fast! I mean - the huge areas can be scanned at high resolution in a matter of 10s of minutes, what I was doing before overnight. The other thing was of course the software - joining the EDS and EBSD data through AZtecHKL EBSD system can really open up whole new chapter of research, and there was no more playing with the background subtraction, binning etc. From the performance point of view, this is a huge step forward, since you're not occupying your SEM for hours anymore (also the drift is negligible at such speeds).
On the other hand, I think making things too simple can lead to the similar difficulties as with the EDS mappings: hundreds of users not knowing what they are doing, and just following a (personal) manual, without any background knowledge. Don't get me wrong - everybody has to start somewhere, but oversimplifying things can have serious impact on the whole field. But this is another topic, and is not related to your question :-)
Dear Gert, I had a privilege to test the Symmetry detector from Oxford instruments recently, on the same samples that I did the analysis with our old CCD system. The first impression was: that's fast! I mean - the huge areas can be scanned at high resolution in a matter of 10s of minutes, what I was doing before overnight. The other thing was of course the software - joining the EDS and EBSD data through AZtecHKL EBSD system can really open up whole new chapter of research, and there was no more playing with the background subtraction, binning etc. From the performance point of view, this is a huge step forward, since you're not occupying your SEM for hours anymore (also the drift is negligible at such speeds).
On the other hand, I think making things too simple can lead to the similar difficulties as with the EDS mappings: hundreds of users not knowing what they are doing, and just following a (personal) manual, without any background knowledge. Don't get me wrong - everybody has to start somewhere, but oversimplifying things can have serious impact on the whole field. But this is another topic, and is not related to your question :-)
The speed might be really one argument... On the other hand, does speed really helps? For drift, 3D measurements (although I am not sure how useful they often really are), sensitive sample etc. However, my impression is that we already now generate too many measurements without analyzing them carefully enough. If something does not look like it should: take another measurement :-)
Anyway, with which detector did you measure before (or still) in your institute? I mean, I wonder when I hear that you still do overnight measurements. We don't do this anymore for a decade. And huge maps only make trouble when you want (re)analyze them. Therefore, my common maps are 640x480 or often simply 400x300. This takes already with an old detector about 30min (if you really squeeze the last accuracy out), i.e. not speed but precision optimized measurements. Otherwise it would take only a few minutes with the previous detector generation.
I completely agree - there is also a serious problem with data storage (also for just writing the data on the disk) and of course post-processing, which would almost demand a computer cluster just to re-open the map. Which leads to exactly what you mention - we're just making nice colourful images for a high-impact comic book journals.
The old rig is Nordlys 2s detector with always crashing Channel5 post-processing software. But the main problem is the lack of meaningful scientific questions we should answer with EBSD, therefore it is usually used inappropriately. Unfortunatelly.
Well, since I am back in BAM I exclusively store the patterns (like in EDS the spectra, of course). This takes much longer...but I never had trouble with post-processing (crashes). Only their software is quite poor. However, there are good and free alternatives like ATOM or MTEX (which requires a Matlab license). Regarding other companies I've heard complains about their software as well, especially if bigger maps are processes. The more I am surprised about the huge maps people collect. For what? Who they want impress? :-)
Anyway, I also know that many EBSD investigations are not the paper worth on they are printed. Nevertheless, I am not that pessimistic in general: EBSD data can help to understand materials. We only have to ask the right questions. The answers are then often not that difficult. Instead, in many papers only colorful maps are drawn (and the best: multiphase IPF maps where each phase has another color key :-) ) Such maps nobody can interpret, at least not me (some very rare cases excluded of course like single crystal maps). There is a lot to do. Independent from my question: Presently speed is seriously not the dominant problem. Maybe for the rare application in industry it is but there I do not see a market because of the weak linkage between EBSD maps and practial conclusions.
- Badges
- Science topic
- Similar topics
- Physical Sciences
- Materials Science