What is EBSD?
EBSD, Electron Backscatter Diffraction, is a well established accessory for the SEM which is applied to characterise microstructure. Typically, it is integrated with chemical analysis through Energy Dispersive Spectrometry (EDS). The integration of EBSD with EDS extends the breadth of sample information you can gather in the SEM, giving greater material insight.
In this webinar we covered the basics of EBSD and detail the most popular applications with worked examples.
View this educational webinar if you want to:
Make better use of the SEM and understand what EBSD can offer
Understand what microstructure is - and why it is important
Identify phases or compounds in your samples in addition to chemical composition
Characterise grain structure
Review the questions & answers that were raised at the end of the webinar
(To view in HD click the button at bottom right in YouTube).
Download the presentation (PDF)
Questions & Answers
Here are some of the questions that were raised during the webinar - and the answers. If you'd like to ask our webinar team a further question click here.
Talking about integrated EBSD and EDS, but EBSD data is collected at high tilt. How accurate is EDS at high tilt?
When you collect EDS data from a sample at high tilt and compare it to a flat sample you can see that there is a difference in the profile of the spectrum. But this is not a problem for us, the quant routines in the AZtec software uses the tilt value in the quantification routine and so the quantification is still accurate. This has been tested on a range of standard material.
How long does it take to collect EBSD data?
With modern hardware and software it is possible to collect EBSD data at fast speeds. The NordlysMax detector can collect EBSD patterns at speeds in excess of 1500 patterns per second. This means you can collect EBSD maps in a matter of minutes.
You showed some examples of texture, can EBSD identify the different texture types in a sample?
We showed texture represented on orientation maps. In addition through our software we can plot texture on pole figures and inverse pole figures. This then enables us to identify what a specific texture is and compare the texture in the material to reference or ideal textures. We can also plot specific areas of specific texture on a map to quantify area fractions of different texture components in a sample
When Grain size is measured using image analysis it typically follow ASTM methods, can you follow similar methods with EBSD?
Yes in the AZtec software the ASTM can be followed, and the grain size number can be calculated. In addition, if some element of the standard is not followed correctly a warning is given.
How different do crystal structures or phases need to be if they are to be separated by EBSD?
When using EBSD only for the phases to be separated on crystal structure alone the there needs to be a 10% difference in lattice parameter and you can find out what the lattice parameters are by looking at crystallographic data bases, and so see whether the phases you have can be separated. Also, if the crystal structure is similar but the chemistry is unique then phases can be separated using combined EBSD and EDS. So there are different techniques you can employ to separate phases depending on what the challenges are.
What sample preparation is needed for EBSD?
To get the best out of EBSD the sample surface should not have contamination or damage. There are various techniques for EBSD sample preparation, which will largely depend on the material being looked at: These include typically metallographic techniques such as Mechanical grinding and polishing. In addition, you can use FIB or ion beam milling, or electrolyte polishing. For more details, please see www.EBSD.com.
What are the spatial resolution limits for EBSD?
Spatial resolution you can achieve will partially depend on the material being analysed, and the beam conditions being used. In general when using conventional EBSD you can achieve spatial resolution better than 100nm, this can be optimised by tailoring analysis conditions such as reducing beam energy. As we showed in the presentation the technique of Transmission Kikuchi Diffraction can improve this by an order of magnitude achieving resolution in the 10nm range.
What is the orientation accuracy that EBSD can achieve?
Orientation accuracy or angular resolution is a measure of how accurate the EBSD orientation measurements are. It is important because with better orientation accuracy smaller changes in orientation – such as sub grain structures - can be measured.
We have a tool called AZtec Refined Accuracy which can deliver angular resolution down to 0.05deg and this is a benefit when looking at strain and subgrain structures
You showed a map of grain boundaries, can EBSD detect other types of boundaries in a material?
Using EBSD we can detect and visualise special boundaries, such as twin boundaries. These are measured by looking at the misorientation between points and calculating axis angle pairs. Those which fit the twin relationship are then defined as twin boundaries.
What is the benefit of using EBSD over other techniques such as XRD?
There are similarities to EBSD and XRD, in that they can both be applied to identify phases and examine texture. However EBSD offers some advantages. With EBSD you get a spatial visualization of phases, grains and any local/global texture. This enables differences to be seen between different regions of a sample – such as in the weld we looked at earlier. The angular resolution In EBSD enables subtle changes in the sample to be seen. Also, EBSD will assist in detecting and identifying minor phases or impurities in a sample which could be missed by XRD.
How long does it take to record the orientation map of the quarter steel rod with 6 mm radius?
The data from the quarter steel rod consisted of 132 fields collected with a 10% overlap to help with the montage. Each field acquired using a step size of 200nm in order to resolve the smaller grains in the material. In total, including the stage movements it took 19 hours to acquire and montage this data.
Can EDS reliably detect element Hydrogen with 1.01 EV XRF line?
No, Hydrogen can not be detected by use of EDS. Recently there has been made significant improvements of EDS detectors in order to improve the performance for light element detection, however Hydrogen is still not possible. In the example Zirconium example in the webinar the Zirconium and Zirconium hydride could not be separated by EDS, only by EBSD.
I am wondering what is the smallest grain size that could be measured by this technique? As for example, can grains of size 5-20 nm, which may form on small interconnect lines in a transistor, be measured? Or, what changes need to be implemented in order to measure such small nano-grains using a FEI system?
In order to determine grain size accurately we typically aim at using a step size of approximately 1/10 of the average grain size diameter. Resolving grains with a diameter of the order 5-20nm would be in the range of what is possible using TKD, as shown on the TKD slides in the presentation. For conventional EBSD the grain size you mention would be too small. For TKD you need to work with electron transparent samples, so there is an added element of sample preparation and you will need a holder in order to position the sample for TKD. The data acquisition and analysis can be done with a conventional EBSD system, so if you already have an EBSD system then you just need a TKD holder for your SEM.
How many grains should be taken into account for a valid texture analysis?
For analysis it is important that the collected data is representative of the material, this is also the case for texture analysis. If the material only has a small variation in grain size then you can use the number of detected grains as a reference and typically we would aim for at least 1000 grains. However if texture is the main interest then it benefits to collect the data using a larger step size in order to get data from a larger area instead of focussing on getting each grain well resolved. This is particular beneficial if there are large variations in grain size as you don’t want a few very large grains to dominate the texture analysis.
For strain measurement, what is the best sample surface preparation conditions for EBSD analysis? How would I know the strain or distortion is not introduced by the polishing or sample preparation?
The EBSD patterns might be diffuse either due to strain in the material or poor sample preparation and in some cases it can be difficult to know if the deformation is from sample preparation or from the material itself. If you have established that a certain preparation method works on an unstrained material then I would start with using the same method on the strained material. If the sample is well prepared then it should be possible to get a clear image of the microstructure using FSD or BSE detectors, but this will not be possible if the sample preparation is not good enough.
Another approach is to follow the evolution of the EBSD pattern during the preparation. If the quality of the pattern doesn’t improve by extending the preparation time then that implies that the material itself is strained and that it is not due to the deformation layer from the preparation.
Can you specify your sample preparation method for grinding/polishing?
To get the best out of EBSD the sample surface should not have contamination or damage. To get the best results it is important that the sample surface is free of damage. During the preparation process it is important at each step to remove the damage layer introduced by the previous step in the process. The actual time for each step will depend on the material. By using an optical microscope it is easy to check if scratches from the previous step have been removed or not.
Typically we start with grinding using grid paper, at each step changing to finer and finer paper. Next step polishing using diamond suspensions starting at 9μm and going down to 1μm in small steps. After the 1μm polish we typically finish with colloidal silica for a few minutes.
There are specific examples provided on www.EBSD.com for different materials.
How can I speed-up EBSD measurements without losing resolution?
To get the data acquired as fast as possible, it is important to first think about what the data is to be used for and to make sure that the samples are well prepared. If the sample preparation is poor then it will lead to poor quality and weak EBSPs which will limit the speed of the data acquisition. For large grained materials the probe current can often be increased which will increase the speed of the measurements, without loss of angular resolution.
In order to get the best spatial resolution the SEM parameters (kV and probe current) will be limited but the data acquisition can be speeded up by collecting EBSPs at a lower resolution (binning) or by bring the detector closer to the sample to increase the intensity of the signal and optimize the geometry to ensure that detector is collecting as much signal as possible. If the EBSD patterns are of good quality then it is also possible to run the EBSD detector at a lower exposure level (under saturated), which will speed up the analysis.
EBSD and surface roughness correlation.
Most sample preparation methods lead to a relatively flat surface, for EBSD the surface does not have to be perfectly flat. In order to get a good quality diffraction pattern it is important that the surface of the material is damage free. Surface roughness is not necessarily a big problem in itself, as long as it doesn’t cause shadowing of the beam going into the sample or the diffracted signal going out of the sample.
What is the confidence level of the data?
The determination of the phase and the orientation is done by comparing against existing databases. This means that in order to get correct data it is important to be comparing against the correct reference. When the phase and orientation is being determined by the AZtec software, there are a number of checks being used by the software to ensure that the data is correct. This means that we can have a high level of confidence in the data before doing any processing.
Is EBSD any help with characterising polymers?
EBSD is an electron diffraction technique. In order for an EBSD pattern to be generated the structure must fulfil the Bragg conditions by having regular spacing between the atomic planes and having several atomic planes stacked within the interaction volume. A polymer contains repeating units however the overall structure of an individual molecule and the structure generated by several molecules doesn’t lead to generation of a diffraction pattern, so EBSD will not help to characterise polymers.
For TKD, does the sample have to be horizontal?
No it doesn’t have to be horizontal.
The main drive behind TKD is to improve the spatial resolution. One of the issues with conventional EBSD at high tilt is that the spatial resolution in the downhill direction on the sample is reduced by nearly a factor of 3 due to the sample tilt. By having the sample positioned horizontal the spatial resolution is the same in the x and y directions of the map.
What about sample preparation? How smooth should be the sample surface be?
To get the best out of EBSD the sample surface should not have contamination or damage. To get the best results it is important that the sample surface is free of damage, so during the preparation process it is important at each step to remove the damage layer introduced by the previous step in the process. There are various techniques for EBSD sample preparation, which will largely depend on the material being looked at: These include typically metallographic techniques such as Mechanical grinding and polishing. In addition, you can use FIB or ion beam milling, or electrolyte polishing. For more details, please see www.EBSD.com.
If the sample is not conductive, is it possible to take EBSD without any coating?
Collecting mapping data from a non conductive sample can be challenging as it is likely to cause drift problems. Coating the sample is one solution to this problem, however it is not always possible.
One alternative is to use VP mode on the SEM, which is widely used as means of charge neutralisation, thereby solving the drift problem caused by the sample being non-conductive.
Another option is to make a conductive path from the SEM stage to the area which is to be analysed, using silver paint or copper tape, and thereby reduce the charging problem.
A third option is to reduce the probe current or the kV in order to minimize the charging problems. The NordlysNano is well suited for this type of work due to its high sensitivity which allow analysis to take place using low kV. There are several application examples of EBSD on non conductive material on www.EBSD.com.
For strain visualization. Are you able to quantify the strain difference between different regions of the material?
The maps shown in the webinar was showing strain variations by assigning a colour based on the level of misorientation within individual grains or from pixel to pixel. The selected colour key means that regions with different levels of strain/misorientation will be coloured differently and by interrogating the actual data values you can quantify the difference in the strain levels. There are a number of other visualisation tools relating to strain, all depending on what it is that you want to display.
Can we do EBSD on fractured surfaces? What would be the complications? Especially when the samples are porous.
It is possible to use EBSD on fractured surfaces. The main issue would be to get the surface oriented so that the beam can get to the surface and out again without too much shadowing. This also means that for fractured surfaces it is not always a good idea to use mapping but instead just collect data from a few individual points. If the fracture has generated clear flat facets then the facet area can often be mapped.
For nano-sized samples, such as some metal interconnects other than Copper (e.g., Ruthenium), the chemical and mechanical polishing method is not exactly optimized yet. As such, some roughness, usually of the order of ~1-2 nm, is prevalent in these samples. Could these be an issue for obtaining good grain mapping for SEM-TKD?
A surface roughness of 1-2nm would not be a problem for TKD, unless you are trying to resolve grains of a similar grain size.