We recently discovered a phase transformation during a careful metallographic EBSD preparation. Does anyone recongnized this, or does this effect has been reported already somewhere? The material is a common commercial 304 stainless steel.
it is common problem during metallographic preparation and the most susceptible are 301, 201 and 304 steels, 316L is less and 310 is almost not transforming at all. This phenomenon is called strain-induced martensitic transformation. I am working on the reversion of strain-induced martensite:
Article The Investigation of Strain-Induced Martensite Reverse Trans...
Article Strain-induced martensite reversion in 18Cr–8Ni steel – tran...
annealed 304 is very soft and during cutting (machining) and grinding you can induce FCC to HCP to BCC transformation. I wonder how careful is your preparation. Usually this transformation is induced at the beginning during cutting or machining of the specimen so most crucial I think is grinding with low force to remove more material.
There is a chance that your sample is not transformed during preparation but you need give us more details about the sample history or the morphology of the BCC phase that you see.
I imagine that a 304 or 18/10 or 18/8 -Cr/Ni stainless steel can show such a behavior. It strongly depends on the processing history of your sample (In fact this is what has recently changed or is about to be changed by most manufacturers of such steels) . I think yours was quite severely plastically deformed in the austenitic state, hence it has a high dislocation density and high internal stresses (e.g. Taylors relation of plastic yield stress). The alloying system is in a range, where the austenite could be stable at room temperature, but only slight modifications such as a mechanical preparation (stress) could lead to a transformation.
What is always difficult for martensite is to make quantitative predictions for energies, since it is quite a heavy multiscale problem. Actually, this is the reason why there are people doing modelling =)
Well, everything is relative. But up to know we are quite used to scratch-free preparation :-) . Certainly, this was no perfect preparation but higher deformation etc we can exclude as well as high deformation density... However, the samples have been machined before (what should not be unusual since everyone uses 304 and does not expect that after caughing fcc transformed to martensite).
I wonder, how much delat-ferrite 304 contains, or is 304 free of any bcc-like phase?
You can simply check the effect of cutting using neodymium magnet (if transformation occurred it will stick harder to the cut surface than the others.) You may find few % of delta ferrite. If it is sheet or rod/wire the delta ferrite will be in form of stringers elongated with long axis of sample and it should have more Cr than the austenite while SIM (strain-induced martensite will have the same chemical composition as austenite). Can you show us SE/BSE or EBSD map image of that phase?
"what should not be unusual since everyone uses 304 and does not expect that after caughing fcc transformed to martensite" - ... what was the specimen history before "machining"?
Is the new phase readily indexed? Because strain induced martensite (which is certainly possible in SS304) would generally be hard to index and is often identified by locally poor band contrast. On the other hand, a clear BCC phase could be delta ferrite remnant from high temperature processing (this also is expected in SS304, usually ~4-6%). If possible, could you put up the EBSD maps? I am interested in the latter possibility.
As for how much strain induced martensite could form, that really depends on the degree of strain...at strains of 30-50%, SS304 could form 20-40% martensite. While polishing certainly some regions will experience higher strains than that...but how much area fraction those regions occupy (in the EBSD map) is anybody's guess.
It is sheet material but the grains are equiaxed, i.e. it has been recrystallized (?). I was simply irritated by the high amount of "delta ferrite" since we found roughly 30% bcc-like phase. The funny thing is that it transforms often entire austenite grains and has a unique orientation (if martensite I would expect the 24 variants...which also somewhere appear, but with a minor fraction.)
Below the phase maps and as example the IPFs for the three already applied preparations steps. The last we performed by ion etching.
It is practically the same position. Because of the new preparation Z is of course different so that grains become bigger or smaller... Green is austenite, red the bcc phase
It doesn't look like strain-induced martensite or delta ferrite. It is interesting. What parameters have you used during ion etching: ion type, energy, incidence angle? Some ions may induce such transformation.
Ion etching is only the last (third) prep since we always found scratches after polishing so that we assume some carbides or anything else makes trouble. Regarding the parameters I have to ask. We are etching with Ar, incident angle as highest as possible (I guess 50°), and maybe 4kV. But this I really have to ask...
Thank you for the images. These are not delta ferrite grains. The small martensite spots scattered around are quite reasonable- mechanical polishing will cause such transformations in localised areas (unlike mechanical testing, which yields more recognisable martensite morphology). Another clue is that these many of these localised areas are removed by ion milling (incidentally, have you also seen such changes after electropolishing?). A third clue, qualitative, is that most of these local areas are green in the IPF map (conventionally (110)), which is characteristic of deformation in FCC, hence indirectly indicating that these areas are heavily strained. This last point could be confirmed by some other measure of local strain- but I don't know how accurate those will be in such small and scattered regions.
As for the entire grains which appear to be converted to martensite- that is indeed puzzling. I wonder if it could be due to misindexing?
Pawel Nowakowski at Fischione showed how low-energy argon ion milling can improve the surface preparation and reduce the amount of strain-induced martensite .Conference Paper Microstructural changes in austenitic steels caused by sampl...
We get this all the time with low-carbon steels with about 5-20 vol-% austenite content. We consistently get lower fractions of retained austenite measured with EBSD when compared to XRD. Final polishing step with a low force and with a non-corrosive colloidal silica like Mastermet helps a bit.
I suppose even "opening up" a free surface with polishing might change the stress state of austenite grains, possibly resulting in a stress-induced martensite transformation at the surface.
Hi Cyril, as you see....the Ar milling at 5.5 kV improves it step by step. I nearly believe already that finally everything will disappear.
@ Tuomo: In our case it does not look like the typical martensite since entire grains or parst of it transform as grain(s) of uniform orientation which is quite unusual for martensite which selects different variants to minimize the surface strain. If you measure with XRD your phase content....how do you prepare the samples and why do you think you see there already too much bcc as well. The interaction depth with Co should be lower that 3µm...which is with W certainly higher but if you do not carefully prepare these samples as well, I can imagine that the transformation depth can reach much deeper than here observed after silica polishing.
@ M. Petersmann: You are right, but as the name modelling already explains, it is based on models, which can be entirely or partially wrong, or based on data which are partially or entirely wrong, or simply to imprecise. Therefore, we are actually doing experiments since you cannot trust models (see the weather forecast :-)).
Seriously, modelling is an important part and has the aim to safe experimental time. I only want point out that (from my point of view) everywhere any (essential) modeling will be double-checked by experiments (and not vice versa). Some problems you cannot check before they happen so that this prognisis you can believe or not (see weather forecast, global warming, earch quakes, tsunami...) so that in that case even the nowadays rough prediction is already better than nothing...
This is a misunderstanding :-). Sorry, but if the sample is not tilted in PECS it is called 0°. The maximum tilt angle is 60° (if we rotate our sample) what we are using, i.e. it is already the highest tilt angle. Sometimes I would wish that we could tilt higher but this also would reduce the etching rate drastically. It is possibly as with EBSD. Not the perfect angle but a compromise between all. And...we do not generate bcc during the ion etching, so that we are on the safe side...
Dear colleagues, over the last two days we found out that the 304 we were investigating does not contain any remarkable content of ferrite or martensite if one apply electro-polishing for about 1min. Then only some residual martensites still exist.
This gave rise to the question of whether the observations are the result of a gradient of the phase distribution. As you see from the image attached the surface of the sheet is indeed mainly covered by martensite. We assume that it will be (at least partly) removed during first grinding, but during grinding and diamond polishing it might happen that martensite on the surface is formed again as thinner layer. The observed microstructure is therefore a mixture of the original fcc with a locally appearing primary martensite and secondarily formed martensite which does not show the typical split-up in up to 24 variants but obviously prefers only one variant with apparently strong but homogeneous orientation changes.
What do you think? Is this interpretation reasonable?
Since I would not call myself an expert in texture of metals I would like to ask, how you would interprete the shown pole figures? I checked the literature for a while but did not find in this time a comparable behaviour of fcc or bcc (or I overlooked them).
The pole figure are derived from a map which only collected orientations from the non-polished surface.