Sample preparation

Seeing is Believing?

Dr. René de Kloe, Applications Specialist, EDAX

A few weeks ago, I participated in a joint SEM – in-situ analysis workshop in Fuveau, France with Tescan electron microscopes and Newtec (supplier of the heating-tensile stage). One of the activities during this workshop was to perform a live in-situ tensile experiment with simultaneous EBSD data collection to illustrate the capabilities of all the systems involved. In-situ measurements are a great way to track material changes during the course of an experiment, but of course in order to be able to show what happens during such an example deformation experiment you need a suitable sample. For the workshop we decided to use a “simple” 304L austenitic stainless-steel material (figure 1) that would nicely show the effects of the stretching.

Figure 1. Laser cut 304L stainless steel tensile test specimen provided by Newtec.

I received several samples a few weeks before the meeting in order to verify the surface quality for the EBSD measurements. And that is where the trouble started …

I was hoping to get a recrystallized microstructure with large grains and clear twin lamellae such that any deformation structures that would develop would be clearly visible. What I got was a sample that appeared heavily deformed even after careful polishing (figure 2).

Figure 2. BSE image after initial mechanical polishing.

This was worrying as the existing deformation structures could obscure the results from the in-situ stretching. Also, I was not entirely sure that this structure was really showing the true microstructure of the austenitic sample as it showed a clear vertical alignment that extended over grain boundaries.
And this is where I contacted long-time EDAX EBSD user Katja Angenendt at the MPIE in Düsseldorf for advice. Katja works in the Department of Microstructure Physics and Alloy Design and has extensive experience in preparing many different metals and alloys for EBSD analysis. From the images that I sent, Katja agreed that the visible structure was most likely introduced by the grinding and polishing that I did and she made some suggestions to remove this damaged layer. Armed with that knowledge and new hope I started fresh and polished the samples once more. And I had some success! Now there were grains visible without internal deformation and some nice clean twin lamellae (figure 3). But not everywhere. I still had lots of areas with a deformed structure and whatever I tried I could not get rid of those.

Figure 3. BSE image after optimized mechanical polishing.

Back to Katja. When I discussed my remaining polishing problems she helpfully proposed to give it a try herself using a combination of mechanical polishing and chemical etching. But even after several polishing attempts starting from scratch and deliberately introducing scratches to verify that enough material was removed we could not completely get rid of the deformed areas. Now we slowly started to accept that this deformation was perhaps a true part of the microstructure. But how could that be if this is supposed to be a recrystallised austenitic 304L stainless steel?

Table 1. 304/304L stainless steel composition.

Let’s take a look at the composition. In table 1 a typical composition of 304 stainless steel is given. The spectrum below (figure 4) shows the composition of my samples.

Figure 4. EDS spectrum with quantification results collected with an Octane Elite Plus detector.

All elements are in the expected range except for Ni which is a bit low and that could bring the composition right at the edge of the austenite stability field. So perhaps the deformed areas are not austenite, but ferrite or martensite? This is quickly verified with an EBSD map and indeed the phase map below confirms the presence of a bcc phase (figure 5).

Figure 5. EBSD map results of the sample before the tensile test, IQ, IPF, and phase maps.

Having this composition right at the edge of the austenite stability field actually added some interesting additional information to the tensile tests during the workshop. Because if the internal deformation in the austenite grains got high enough, we might just trigger a phase transformation to ferrite (or martensite) with ongoing deformation.

Figure 6. Phase maps (upper row) and Grain Reference Orientation Deviation (GROD) maps (lower row) for a sequence of maps collected during the tensile test.

And that is exactly what we have observed (figure 6). At the start of the experiments the ferrite fraction in the analysis field is 7.8% and with increasing deformation the ferrite fraction goes up to 11.9% at 14% strain.

So, after a tough start the 304L stainless steel samples made the measurements collected during the workshop even more interesting by adding a phase transformation to the deformation. If you are regularly working with these alloys this is probably not unexpected behavior. But if you are working with many different materials you have to be aware that different types of specimen treatment, either during preparation or during experimentation, may have a large influence on your characterization results. Always be careful that you do not only see what you believe, but ensure that you can believe what you see.

Finally I want to thank the people of Tescan and Newtec for their assistance in the data collection during the workshop in Fuveau and especially a big thank you to Katja Angenendt at the Max Planck Institute for Iron Research in Düsseldorf for helpful discussions and help in preparing the sample.

Avoid a Distorted View

Dr. Stuart Wright, Senior Scientist EBSD, EDAX

In the world of “fake news” and “alternative facts”, it is important that we dig a little deeper than the headlines to understand the world around us and to avoid a distorted view those in power often want to give us. Ironically, the same is true at the microscale. I recently ran into some work concerning the effects of sample prep on x-ray measurements. It made me reflect on some early work we did to explore the effects of sample prep on EBSD results.

In order to prepare EBSD samples properly it is important to understand that surface finish is not the whole story. It is important that the layer of material sampled by EBSD be distortion free. Charts shown in many metals preparation handbooks clearly show that there can be significant deformation imparted into the sub-surface of a material during preparation. Consider the following chart adapted from a figure in a classic EBSD sample preparation paper: D. Katrakova & F. Mücklich (2001) “Specimen preparation for electron Backscatter Diffraction. Part I: Metals” Praktische Metallographie. 8:547-65. This plot clearly shows why sample prep for EBSD needs to be meticulous.

My longtime colleague, Matt Nowell, did a nice study comparing by grinding two samples, one ground to 240 grit and one to 1200 grit. He then cross-sectioned these samples and carefully prepared the cross-sectioned surfaces. Matt then did OIM scans on the two surfaces. Using a Kernel Average Misorientation (KAM) map, the degree of deformation in the 240 grit sample is clearly more pronounced that in the 1200 grit sample. Matt and I have always wanted to repeat this measurement for more grits and materials but have never found the time to pursue it again.

Many times, students who have asked me “which grinding and/or polishing steps can I skip?” Or, “how many times can I really use a grinding paper?” (I remember as a student we got one paper for each grit for the semester and we would hang them from a wire with clothes pins in the sample prep lab!). Or, “can’t I just do the final grinding step for a longer time and skip the coarser grinding steps?” One thing we’ve learned on our own and in conversations with the sample prep vendors is that the recipes developed with several steps for what intuitively may feel like short times really are the steps that lead to the best results -basically confirming the plot shown above.

The improvement in cameras, image processing and particularly NPAR™ should not be used as an excuse to take shortcuts in sample prep. While it may be possible to get patterns and reasonable maps, are you really looking at the representative microstructure of interest or a distorted version resulting from deformation induced by sample prep?

I believe EBSD has had a positive impact on the metallography community. EBSD has forced us to be more careful in sample preparation over that typically done for light microscopy or even scanning electron microscopy. Hopefully that extra care has resulted in more representative microstructural characterization.