As a classically trained chemist, most of my career has been spent looking into the challenges presented within the materials science world. Not so coincidentally, I never spent much time thinking about the problems that face our scientific brethren in the life sciences. Cells, tissues, biomolecules – these things were the squishy equivalent of a foreign language to me.
In the past few years, while working in the Atomic Force Microscopy space and the Electron Microscopy (EM) space, I kept coming across researchers from the life sciences who asked questions that sounded more and more like traditional materials science questions. What is the structure of my biomolecule? What elements are present, and in what proportion in this tissue or vesicle? What are the mechanical properties of my cell? And I thought, what kinds of questions should we be asking, and where can we use classic materials analysis methods (in this case, Energy Dispersive Spectroscopy (EDS)) to answer questions in the life sciences? Like any good researcher, this led me to a literature search to see what’s been done before. I’d like to share what I found and what problems we can potentially solve with EDS.
Taking a half-step back, I should briefly explain the principles of EDS as a technique to make sure we’re all on the same page. The too long; didn’t read (TL;DR) explanation of EDS goes something like this. When the electron beam interacts with a sample within an electron microscope, X-rays are emitted. Each element releases X-rays with a unique energy signature, proportionally increasing as a function of their atomic number (e.g., carbon X-rays are lower energy than iron X-rays, which are lower still than lead X-rays). An EDS X-ray detector (like EDAX manufactures) captures these X-rays, identifies the elements present in the sample, and quantifies their concentration. The big advantage of an EDS detector in your SEM is identifying which elements are present. The disadvantages are that it doesn’t discriminate the structure of how these elements are bonded to one another, what molecules or more complex compounds might be present, and historically, EDS hasn’t been tremendously efficient at detecting or quantifying elements lighter than, say, fluorine (the latest generation EDS detectors by-and-large have overcome this, and now routinely measure elements as light as boron, beryllium, and in the right conditions, lithium).
The latter disadvantage has historically been only part of the limiting factor in the general acceptance of EDS in the life science community. Biological systems are basically a collection of carbon, nitrogen, hydrogen, and oxygen, and the structural arrangement of these elements is what matters. EDS didn’t do that, unlike, say, vibrational spectroscopy (Fourier-transform infrared spectroscopy (FTIR) or Raman) or mass spectroscopy, and therefore was discounted for lack of usefulness to the standard biologist. But, with the novel researcher looking for any edge they can find to learn more about their system, new attention is being given to attributes where heavy element detection or accumulation is present in biological systems.
Mineralization and bioaccumulation of materials in tissues and systems is a perfect example of how EDS provides new insights into biological processes. Heavier elements like calcium, phosphorus, and potassium are accumulated and concentrated in easily detectable amounts in tissues leading to the formation of biominerals like kidney stones, sclerotic materials, and bone spurs. EDS provides a simple and clear method to visualize and quantify how these elements are distributed.
Figure 1. STEM EDS images of human sclerotic tissues to show elemental concentrations.
There are numerous examples in the literature of using nanoparticles from a host of different elements to understand cellular and biomechanistic behaviors better. From tumor growth studies using iron and copper nanoparticles to track the deposition of drugs within a cancer cell, to using zinc and iron nanoparticles to understand biomaterial scaffolding, to using silver and gold nanoparticles to understand the efficacy of erectile dysfunction drugs, nanoparticles allow for a targeted tracking of materials using EDS. 
Toxicological leaching of biological implants for dental and orthopedic research to novel biomaterials always require a toxicological study to ensure that the materials used do not leach or otherwise migrate during in vivo applications. EDS is a suitable tool to evaluate lifetime studies looking for the flow of titanium, nickel, iron, or other metallic elements during post-mortem analysis of implanted structures. In addition, environmental leaching of hazardous materials, which are accumulated in plant life, can also be measured via SEM EDS. The ability of different phenotypes of plants to absorb iron or manganese from the soil and concentrate it in the cellular structure can be measured effectively.
Figure 2. EDS spectra and cartoon characterization showing changes in Fe or Mn uptake and concenration as a function of phenotype of Arabidopsis thaliana cotelydons seed pods. 
These are just a few examples of the numerous ways that SEM, STEM, and TEM-based EDS can be used to complement research in the life sciences. As we continue to see a blending of the materials and biological worlds, I look forward to seeing more examples of elemental analysis being used to further scientific discovery.
Satoshi Hara, E. et al, Nanostructural analysis of distinct nucleation sites in pathological mineralization. RSC; Mater. Adv., 2021, 2, 4423.
As an applications engineer, it is always fun to play with cutting-edge products. Last year, I got an exciting new lab partner, an Orbis PC Micro X-ray Fluorescence (Micro-XRF) Analyzer, which is an excellent complement to Scanning Electron Microscopy – Energy Dispersive Spectroscopy (SEM-EDS)-based X-ray microanalysis.
Figure 1. The Orbis PC Micro-XRF Analyzer.
For those of you who are not familiar with Micro-XRF, it is a technique similar to Energy Dispersive Spectroscopy (EDS) in that they both detect generated X-rays after interaction with the sample. For EDS, X-rays are generated by electrons boarding the sample, while in a Micro-XRF unit, fluorescent X-rays are excited by high-energy X-rays emitted from the X-ray tube. Silicon Drift Detectors (SDDs) are used for X-ray detection in modern EDS and Micro-XRF systems. Data collection is also similar because it is possible to use either one to do qualitative and quantitative analysis, mapping, and linescan.
This benchtop Orbis PC analyzer utilizes the benefits of conventional XRF while implementing micro-spot X-rays down to 30 μm by employing a polycapillary technique with a moveable stage. For higher Z elements, it improves the detection limits ten times or more than SEM-EDS. It uses higher-energy X-rays to generate lines that are not detectable with EDS, such as Sr L, Zr K, and Ag K, which is useful when lower energy lines overlap in the EDS spectrum. The industry-exclusive motorized turret, integrating video and X-ray optics, provides coaxial X-ray analysis and sample view perpendicular to the sample surface for more accurate sample positioning and no shadowing of the X-ray beam. The analysis is non-destructive, with no beam damage to the sample, and minimal sample preparation is required.
Grinding and polishing of the sample are not generally required, and conductivity is not an issue. Sample loading is flexible in that thicker samples can be loaded directly on the stage, and thinner samples, particulates, and fibers can be mounted. The sample shape and height can be irregular, and the large sample chamber in this benchtop Micro-XRF unit can accommodate a wide range of sample sizes. Samples can be run either in low-vacuum mode or air mode, allowing the analysis of liquids or samples that will dehydrate in a vacuum. An SEM-based Micro-XRF system does not have many of the benefits brought by this benchtop unit. Once the sample is loaded in an SEM chamber, all the requirements of SEM samples apply. The chamber size and stage of an SEM largely limit the sample dimensions, and non-conductive samples must be coated. The ability to analyze samples that cannot tolerate a vacuum atmosphere is also lost.
Figure 2. The unique four-position turret. Position 1 is the high magnification video, and position 2 is the 30 μm polycapillary X-ray optic. Position 3 and 4 are 1 mm and 2 mm collimators, respectively.
Figure 3. The video (green) and X-ray (dark red) paths of the Orbis Analyzer are coaxial and perpendicular to the sample surface, which means the X-ray path is observable in the video, and there is no shadowing of the X-ray beam. If the X-ray path is non-coaxial (red), it can be blocked by the high topography object.
Since SEM-EDS and Micro-XRF share many similarities and work together to accomplish the complete needs of spectral analysis (see the How to Correlate Micro-XRF and SEM-EDS for Optimal X-ray Characterization of Materials article in the March 2022 issue of the Insight newsletter), I always like to correlate them from every aspect. The absorption edge is the most recent one that caught my attention. For EDS users, if you ever take a close look at the Bremsstrahlung background modeling in the APEX™ software, it is not a smooth curve but exhibits sharp edges (e.g., Figure 4). These are absorption edges, indicating the minimum energy required for an element to eject an electron from its core orbital to create a vacancy. For example, the absorption edge of Ni K lies at approximately 8.33 keV. As the electron energy reaches this value, there is a huge spike in energy attenuation because this is the point that the excitation of Ni K lines begins (Figure 5). The Mass Absorption Coefficient quantitatively represents each element’s absorption of energy. The self-absorption of X-ray photons in a specimen is the dominant effect in EDS, as well as the Bremsstrahlung background distribution shape. The mass absorption coefficient jumps visibly influence the spectrum, mainly in the soft X-ray region. Our Bremsstrahlung background modeling includes these absorption edges for fine control and accurate background correction.
Figure 4. The absorption edge of Fe K at approximately 7.11 keV in an EDS spectrum.
Figure 5. The absorption edges of Cr, Fe, and Ni K lines.
The absorption edge plays an extraordinary role in Micro-XRF since the design of primary beam filters employs the knowledge of absorption edges. The Orbis PC unit is equipped with six primary beam filters to preferentially absorb X-rays at certain ranges to reduce the background to improve detection limits and eliminate artifact peaks. The filter wheel is placed between the X-ray tube and X-ray optic, so the X-rays scattered by the filter do not reach the sample (Figure 6). Only X-rays focused by the optic or collimated by the collimator reach the sample for accurate sample targeting. Figure 7 shows the background in the spectrum if the X-rays generated from the X-ray tube are exposed to a Ni filter. There is a strong correlation between the background in this figure and the graph illustration in Figure 5. The X-ray attenuation decreases as the absorption edge at approximately 8.33 keV is approached in Figure 5. This coincides with more and more of the tube X-rays penetrating through the filter and being present in the spectrum. Once the energy reaches 8.33 keV, there is a sudden increase in the absorption of X-rays shown in Figure 5, and this is why a huge amount of the background signal is absent in Figure 7 since most of the signal is now absorbed by the Ni filter. After the significant jump at the absorption edge, the attenuation continues to decrease as energy increases in Figure 5. This correlates to the background getting higher and higher in Figure 7 since more and more tube X-rays continue to penetrate through the Ni filter. The area with the lowest background signal in Figure 7 is the high-sensitivity region where the Ni filter cleans up the spectrum, allowing true elemental peaks of interest to show up. Figure 8 is an example of the detection limits of As in an As2O3 sample. The Al-heavy and Ni filters significantly increase the peak-to-background ratio to push the detection limit to a single-digit ppm-level.
Figure 6. Schematic of filter wheel design in Orbis system.
Figure 7. Background spectrum from the X-ray tube after being exposed to a Ni filter.
Figure 8. A spectrum overlay of As2O3 was collected using an Orbis PC without a filter (red), with Al-heavy (blue), and Ni (green) filters.
Expect a few new application notes and experiment briefs from this unique lab partner!
Back in my early days of installing some of the first EBSD systems in the world, one of the issues I had was figuring out how to demonstrate the system’s performance and how to help users operate their systems to get that same performance. As EBSD users know, this technique requires a certain level of sample preparation to obtain useable patterns and good quality maps. Because of this, I would bring my own previously prepared samples to set up a system. However, I generally would not leave these behind. This encouraged customers to figure out sample preparation before using their EBSD system.
After a few of these visits, we decided it would be beneficial to provide users with standard samples that could be left with the systems. To do this, we selected the material, prepared it for EBSD, and then packaged it for delivery. The question at that point was, ” what material do we use?”
We wanted something that would produce good EBSD patterns, not significantly degrade over time, and was something we could prepare ourselves. One of the materials EBSD had consistent success with early on, and still do today, is nickel-based superalloys. These materials have a higher average atomic number than aluminum alloys for stronger EBSD pattern intensity, large enough grains for work on both tungsten and FEG source SEMs, and can sit in a lab for years while still producing good EBSD patterns after the initial preparation. This led us to select Inconel 600 as our standard material.
It also led to it being one of the most well-characterized alloys by EBSD globally, even if the results are not all published. We have used our nickel standard to test all our detectors, from early SIT video cameras, to the first DigiView CCD cameras, the high-speed Velocity CMOS cameras, and now the Clarity Super direct-detector system. This material gives us a consistent reference point to better understand performance. We have also used the material for validating PRIAS™ imaging, NPAR™ processing, and OIM Matrix™ indexing.
Figure 1. An EBSD IQ map with random grain boundaries drawn as black lines, primary twins drawn as red lines, and secondary twins drawn as blue lines.
Several interesting microstructural features can be measured with these alloys. First, a high fraction of twin boundaries are typically present within the nickel samples. Figure 1 shows an EBSD Image Quality (IQ) map with random grain boundaries drawn as black lines, primary twins drawn as red lines, and secondary twins drawn as blue lines. We can also show grain maps with this high twin fraction, where grains are determined from the measured orientations and then randomly colored while including and excluding the twins in the grain grouping algorithm. Figure 2a shows the grain map, including the twin boundaries, while Figure 2b shows the grain map excluding the twin boundaries. There is a significant difference in effective grain size between these two microstructure views. Finally, we know that the twinning plane in face-centered cubic nickel alloys is the (111) plane. We can display the (111) plane trace on both sides of the twin boundaries, as shown in Figure 3.
Figure 2. a) A grain map that includes the twin boundaries. b) A grain map excluding the twin boundaries.
Figure 3. Combined IQ and IPF orientation map with (111) plane traces shown on both sides of selected twin boundaries.
Now you have some idea of what you can measure with your EDAX EBSD nickel standard.
Ever since I started university and later began my graduate research work on energy-related topics, global warming and renewable energy are two subjects that appear frequently in papers and conferences. To mitigate and avoid the potential climate catastrophes that global warming may cause, governments and companies have invested heavily in renewable energy research over the years. Lithium batteries are one of the renewable energy technologies that are commonly used for cars and appliances. As you may know, many governments have implemented laws to ban fossil fuel cars sales in the foreseeable future and have encouraged companies like Telsa, Nio, and BYD to make these batteries more readily available.
However, charging an automobile is not as convenient as adding gasoline. And if you’ve ever driven an electric car, you’re probably aware of how much the mileage varies between summer and winter. But electric cars are the future. As universities, research institutes, and enterprises troubleshoot issues like these, I think the future of battery technology will be bright and more surprises will show up.
Since I joined Gatan, I have also been responsible for some of EDAX products. Gatan and EDAX are both scientific equipment providers of material characterization solutions for electron microscopy. For lithium batteries, we have a series of products that cover users’ application needs in one way or another. Last year, we introduced a joint characterization solution for lithium using the EDAX Octane Elite Energy Dispersive Spectroscopy (EDS) Detector and Gatan OnPoint™ Backscattered Electron (BSE) Detector. With this solution, we can reduce the detection limit of lithium by nearly ten times, compared with current schemes, to a single-digit mass percentage. At the same time, the characterization ability is not affected by the oxidation state of lithium.
Figure 1. The lithium mapping from joint characterization of the EDAX Octane Elite EDS Detector and Gatan OnPoint BSE Detector.
Many users wonder why it is difficult to characterize lithium as a light metal (whether elemental or ionic) with an EDS detector alone. The reasons behind this are related to the mechanism by which X-rays are generated in electron microscopy and the window material of the EDS detector. Long story short, the generation of EDS signals requires the electron beam to knock out the electrons in the inner shell of an element, and then the vacancies cause the electrons from the outer shell to refill. After refilling the vacancy, due to the difference in energy levels of the two electron shells, an EDS signal corresponding to this energy difference is generated.
Figure 2. Characteristic X-ray production using Si K_α as an example. Adapted from myscope.training.
So, in this over simplified scheme, EDS can only detect lithium metal, and cannot detect lithium ions (just have two electrons in the K shell, no electron to refill the hole). In addition, due to the fact that the characteristic X-ray energy of lithium is only 55 eV, the common thick polymer window in EDS detectors absorbs low-energy X-rays heavily. However, the unique ultra-thin Si3N4 window material in EDAX EDS detectors provides higher X-ray transmittance at the low-energy range (see red line in Figure 3). Therefore, EDAX helps.
Figure 3. The low energy X-ray transmission rate comparison between EDAX Si3N4 window material (in red) and commonly used polymer window (in green).
Gatan’s image filter (GIF) system offers a different solution from another technical point of view on the same lithium detection issue, and the electron energy loss spectroscopy (EELS) spectrum is much better and easier at detecting lithium. In contrast to the generation process of EDS signals, EELS signals begin to generate in the first step (inelastic scattering), namely the interaction of the electron beam with electrons outside the nucleus. The signal counts of EELS are much stronger than that of EDS, and the characterization of lithium is naturally much more convenient than that of EDS. Of course, lithium or battery materials, as a whole, are very sensitive and are not resistant to the electron beam, which creates additional requirements for Gatan’s imaging filter system. It needs to be fast, have high sensitivity, and low noise.
Figure 4. Gatan GIF Continuum K3 System.
The figure above shows the Gatan GIF Continuum K3 System, which has high sensitivity from the K3 direct detection camera. It can also collect data at high speeds with little noise. Last November, professor Meng Gu’s team at the Southern University of Science and Technology (SUST) in China published a paper on Matter. They used an extremely low beam dose (10 pA) to successfully characterize lithium and acquire the fine structure of the lithium element from electron loss near edge structure (ELNES) spectra. Then, they mapped out lithium metal and surficial oxidized lithium in their battery material using the MLLS function in the Gatan DigitalMicrograph® Software. The GIF Continuum K3 not only detects lithium but also identifies lithium in different chemical valence states. This work has important values for studying the “dead lithium” problem.
However, for lithium-ion battery research, the detection of lithium is only the first step. The more important content is about studying the transport pathways of lithium ions, and these pathways determine the energy density, capacity, and life span of a battery. But how do we characterize the flow of those ions? This problem corresponds with figuring out how to characterize the grain structure inside the cathode material of a battery. There is a correlation between the grain size of a cathode material, the specific crystal plane, grain boundaries, and the transport tendencies of lithium ions. In an ACS Nano article published at the end of last year, Yuki Nomura from Panasonic Company of Japan employed both precession electron diffraction (PED), a crystallographic characterization method similar to Electron Backscatter Diffraction (EBSD) but on a transmission electron microscope (TEM)), and the Gatan Quantum Imaging Filter Series, taking data from the same region of electrode material on an in-situ TEM. The results show the relationship between the real-time distribution of lithium at different stages during a charging reaction and certain grain boundaries and crystal planes block the movement of lithium ions. For a particular crystal orientation, lithium ions have a clear tendency to move through during charging, while some other crystal planes and grain boundaries have obvious resistance to the movement of lithium ions. Personally speaking, it is believed that from Yuki’s work, there will be more relevant research published in this field in the future. As a result, researchers are helping to achieve a more reasonable design for battery material’s crystal structure and chemical composition.
It’s hard not to think of the EBSD technology on a scanning electron microscope (SEM) after looking through the PED used in Panasonic’s paper. After all, EBSD can do all the functions that PED can achieve on TEM, except spatial resolution, on scanning electron microscopy, or even better (for example, angular resolution). Given the electron beam dose issue on battery materials, the main CMOS scintillator-based EBSD detectors on the market may have some difficulty with characterization. In response to this problem, EDAX has an EBSD product based on direct detection technology, the Clarity™.
Figure 5. a) Inverse pole figure (IPF) of lithium battery cathode material using normal EBSD experimental conditions, HV: 20 kV, beam current: 1.6 nA. Many unindexed points in cathode particle; b) IPF of the same cathode material but on a different region using the Clarity EBSD Detector, HV: 10 kV, beam current: 400 pA. where more structural details are disclosed; c) The Clarity EBSD Detector.
In August 2020, Donal Finegan’s team at the Renewable Energy National Laboratory (NREL) in the USA used Clarity to obtain orientations, grain boundaries, and morphologies information about NMC electrode material for lithium-ion batteries. This ample structural information helps researchers identify the mechanism by which intergranular cracks occur to understand the transport pathway of lithium ions, and the reduction of battery capacity caused by the expansion of cathode material lattice during charging and discharging processes. Previously, many publications only showed that polycrystalline, small grain cathode materials contributed to better battery performance. Still, the performance advantage caused by specific polycrystalline materials or those characteristics in small grains is not clear. Finegan’s work, through Clarity EBSD, helps us find the grain boundary structure that could be potentially beneficial thereby this work can guide people to designing more accurate battery materials. In addition, EBSD has another advantage. Counting on the material processing capabilities of Focused Ion Beam (FIB) electron microscopy, we can also achieve 3D-EBSD characterization and study grains on a three-dimensional scale. This feature is nearly impossible for PED. I believe that more research on grain size and boundaries based on three dimensions is the future, and will bring us more surprises.
As an application scientist works who for a scientific instrument company, I enjoy thinking deeply about the field of equipment applications and taking the practical problems from our users’ research as opportunities for us to improve our technical knowledge and demonstrate the superior performance of our equipment. In the future, I look forward to seeing our Gatan and EDAX equipment shine in the fields of renewable energy, additive manufacturing, ultrafast electron diffraction, cryo-EM coronavirus research, and other research fields. And I also look forward to, through my own learning and improvement, bringing more inspiration and thinking to our users from the application perspectives so that our users can not only use our equipment properly but also use our equipment in a more advanced way.
 Han, Bing, et al. “Conformal Three-Dimensional Interphase of Li Metal Anode Revealed by Low Dose Cryo-Electron Microscopy.” Matter (2021).
 Nomura, Yuki, et al. “Lithium Transport Pathways Guided by Grain Architectures in Ni-Rich Layered Cathodes.” ACS nano (2021).
The new year is here. And with it, we look for ways to start our research on the right foot.
Over the past two years, I’ve traveled less and been fortunate to spend more time supporting customers online and working in the lab analyzing samples. Whether it’s a customer’s sample or my own, my goal is to push the limit by using different conditions across various samples to get the best results. Using this approach, I can always get good Electron Backscatter Diffraction (EBSD) indexing results when analyzing challenging samples. It is the highlight of my day when I see the colorful Inverse Pole Figure (IPF) maps and feel that my hard work has paid off.
On the other hand, my experience in technical support provides me with some tips. After all, most people’s mistakes are similar. I want to take this opportunity to discuss this.
In many cases, analysis results are sub-optimal because experimental details are not well controlled and are not a reflection of the product age or technology. Recently, I have seen many papers (such as the deformed Ti on the cover of Science) where they did not use the latest hardware/technology, but the results are excellent. If you take EBSD as a whole process and properly deal with all influencing factors, then any shortcomings will affect the final result. Many factors need to be accounted for, such as the preparation of the sample, sample mounting, the input signal strength or weakness, and then it comes to the EBSD operation itself.
Figure 1. September 17, 2021 issue of Science magazine featuring an EBSD orientation map of cryoforged Titanium.
Since I have a Gatan Ilion® II (model 697) in my sample preparation room, I no longer worry about the sample preparation process. Ion polishing is the best method, and it can achieve the requirements most of the time. Of course, vibration or electrolytic polishing is also a suitable method; just pay attention to the choice of parameters.
Figure 2. Gatan Ilion II System
Often it is attention to the small details that pay the biggest dividends. When mounting the sample for EBSD, we want to eliminate sources of physical sample drift due to the effects of a 70° sample tilt. I use a mechanical method or choose a liquid glue when performing this step. This becomes more important as the sample size and mass increase. Also, use an appropriate beam current selected for EBSD. Beam currents used for high-resolution SEM imaging are often lower than required for traditional EBSD detectors. Make sure the image is in focus and is properly stigmated. I once demonstrated the effect of focus on EBSD Image Quality (IQ) values, and the people present were astonished. When needed, use the dynamic focus correction on the SEM to keep the focus constant across the tilted surface.
Be aware of the different parameters that can be set for the EBSD system as well as the SEM. From my personal experience, the selection and optimization of these parameters can easily increase the speed and quality of your data.
Once you get to this point, the process is almost complete. The EBSD parameters are pretty simple, as long as the signal can be reached. Think about the number of grains to analyze and the SEM magnification required for this field of view. Then select a step size appropriate for the average grain size and type of analysis. Selecting this requires thinking about our desired acquisition time, the speed of the detector, and the details of the microstructure, and the APEX software can recommend different values. If every step is done well, then this process should be perfect. Then, just do it.
In the end, these proven approaches can be applied to your existing or new instrumentation to achieve your best results. I hope these thoughts are helpful to ensure your work goes smoothly in the new year.
Last year we released APEX 2.0 software, which had grown considerably into a more complete microanalysis package to offer both Energy Dispersive Spectroscopy (EDS) and Electron Backscatter Diffraction (EBSD) characterization capability. APEX 2.1 was delivered earlier this year to include support for dual EDS detectors on the EDS side. Towards the end of this year, we added another major feature, Full Standards Quant, and multiple other new functionalities with the release of APEX 2.2. APEX 2.2 has been out for more than a month, and many users have enjoyed the higher-level features and enhancements brought by the latest release. In this blog post, I will go over what is new in APEX 2.2 for EDS.
Spectrum and Image Annotation
With the new floating Annotation Toolbar, the user can easily add text and shapes in images and spectra. A ruler option is available to measure any feature in the image, and the measurement is saved with the image. All of the annotations can be edited or deleted after creation. Annotations are saved with HDF5 files and included in the report. Spectra can be saved as images with annotations.
Figure 1. left) Floating toolbar and annotations in the spectrum. right) Annotations in the SEM image.
Previously, multiple spectra were normalized with respect to the highest energy peak by default. The new Spectrum Normalization feature gives the user more flexibility. The spectra can be normalized based on any peak by selecting the region or element or drawing in the spectrum overlay. The ratio information is shown on the overlay, and the normalized energy range is displayed in the report.
Figure 2. Spectra normalized with respect to the Si K alpha peak. The ratio information is shown at the top-right corner.
Quant Montage Maps
In APEX 2.0, Montage mapping was released as a key feature that allows precise large-area imaging and EDS and EBSD mapping. It uses stage movements to collect individual high-resolution images and maps through a grid pattern over a large sample surface and stitches them into montages. In APEX 2.2, we took one more step forward, and now the user can rebuild Montage maps as NET and ZAF (Wt% and At%) maps. The workflow is the same as rebuilding single-field EDS maps for ease of use.
Full Standards Quant
The most exciting new functionality in this release is Full Standards Quant. We calculate a unique reference value from a single element standard spectrum to determine the product of the beam current and detector solid angle. The reference value will be burned with every subsequent spectrum collected at the given parameters. It normalizes out any differences in beam current and detector geometry. As long as the standard and unknown spectra were collected at the same accelerating voltage, the standard can be applied to quantify the unknown sample. There are three modes available in Full Standards Quant. The Standard mode is suitable for those users who want to take complete control of the standard selection. The MultiStandards mode uses the mean values of the loaded standards. The SmartStandards mode is an intelligent approach that automatically picks the best fitting standards using an iterative internal assessment algorithm. The Full Standards Quant chart (Figure 3) gives a visual representation of how close your standards are to your unknown. It also generates a curve to show the calculated element concentration versus net counts based on the loaded standards.
Figure 3. Full Standards Quant chart of Si K. The yellow diamonds show standards, and the “X” indicates the unknown sample. The calculated curve gives the element concentration versus net counts based on the loaded standards.
APEX 2.5 with new features and productivity enhancers for EDS and EBSD is on the horizon in 2022, followed by more integrations in the APEX platform. As always, minor version upgrades of APEX are free, so experience the power of APEX 2.2 now and prepare for the benefits of APEX 2.5 soon. It will be an exciting year for APEX!
After finishing my Ph.D. in structural geology at Utrecht University, I joined EDAX in 2001 to become an EBSD Application Specialist. At the time, I expected that my experience with investigating microstructures of rocks would be equally applicable to metals and ceramics. After all, grains are grains, and deformation structures in rocks and ceramics are not all that different. At least, that was my impression when I visited the International Microscopy Conference (IMC14) in Cancun, Mexico, in 1998. My poster on melt and deformation microstructures in partially molten olivine-orthopyroxene rocks was the only geological contribution in a metallurgical section. At first glance, nobody even noticed that my TEM images were not on metal, and I had many interesting discussions on potential deformation mechanisms and how metallurgists and geologists could work together and learn from each other.
Quickly after joining EDAX, I recognized that there is a fundamental difference between the work of a material scientist and that of a geologist. To generalize, a material scientist works on developing materials like metal alloys or ceramics for a specific application and typically has a pretty good idea of what kind of processing has been applied to a material. As a geologist working with natural rocks, you have to work with the material and try to unpick its formation history that may span millions of years by observations alone. This turned out to be a helpful skill in looking at customer samples that have come my way over the years. In many cases, I only have very limited information on the background of a sample. Then, I have to rely on my observations to unravel the sequence of events and select the analytical options for the successful characterization of the microstructure. Observing materials without assuming that you know what happened allows you to catch unexpected events in production processes, and that makes a very useful connection between characterizing materials in geology and materials science.
For example, during a recent training course, we worked on a stud welded sample. This was a new technique for me, and a first-practice EBSD map showed a very interesting microstructure. We used that map during the training, but since then, I have taken the time to do a complete characterization to see if I could identify what really happens during these short milliseconds that it takes to weld a stud to a substrate.
Online, I found a general description of the welding technique (Figure 1). In the stud welding process, an electrical arc is generated between the tip of the stud and the substrate so that a small melt pool forms at the contact. The stud is then plunged into the melt pool, which solidifies within milliseconds, creating a strong bond. Can we use EBSD to tell us what really happens?
Figure 1. Stud welding procedure, 1) The stud is placed close to the substrate. 2) An electric arc is created to melt the contact area on both sides. 3) The stud is pushed into the melt pool. 4) After a few ms of cooling time, the weld is complete.
And I start just like I would when looking at a rock outcrop. First, take a step back and look at the entire structure. Once we have seen that, we can zoom in on key areas to investigate what has happened.
The SEM image (Figure 2) does not clearly show how large the area affected by the welding process is in the cross-section. The expected microstructural changes can be illustrated using an EBSD Image Quality (IQ) map. An IQ map shows the contrast of the bands in the diffraction patterns where recrystallized areas with good quality patterns are bright. In contrast, deformed areas producing poor quality patterns appear dark. Therefore, any changes in the crystalline microstructure are likely to be clearly visible in the EBSD results.
Figure 2. a) The polished sample in 3 cm resin mount. b) An SEM secondary electron montage image of the stud weld contact. (1806 fields, 15.7 x 10.7 mm)
Before we look at the weld structure itself, we need to check the starting microstructure of the base materials to recognize what changes during welding. The threaded stud on top consists of deformed austenitic steel. At the top of the IQ map (Figure 3a), individual grains can be seen along the stud’s center axis. With increasing deformation towards the threads, the grains become smaller and EBSD patterns degrade, resulting in very dark IQ values at the stud surface.
Figure 3. a) An IQ map and b) a phase map of a montage EBSD scan with 221 fields, 45 million points @ 1.5 µm steps. Red is ferrite and green is austenite. c) An IPF map on IQ showing the crystal directions parallel to the sample normal direction. The scalebar is 4 mm.
This structure is related to the wire drawing of the stud and the shaping of the threads.
The ferrite substrate appears homogeneous with an equiaxed grain structure and consistently good quality patterns. However, the change in color from green to purple in the IPF map (Figure 3c) shows a gradient in the dominant grain orientation distribution towards the interface. This does not appear to be related to the welding process and is probably introduced during the production of the ferritic steel sheet. The welding process dramatically changes the IQ appearance of both the stud and the substrate (Figure 3a). In the welding zone, the austenitic material becomes bright while the ferrite goes very dark. Between these two areas, a band of material is present that has been melted and then quickly solidified. In this solidified material, there are swirls of darker IQ values that appear indicative of the melt’s movement during the plunge phase when the stud gets pushed into the melt pool. The EBSD phase map in Figure 3b indicates that these darker bands in the melted zone consist of ferrite grains. The main components of the weld zone are shown in Figure 4.
Figure 4. A weld structure IQ map with the main structures indicated.
To see what causes these changes in the IQ map, we have to look more closely at the EBSD results. The bright IQ band in the austenite stud consists of small equiaxed grains that are fully recrystallized and produce good quality patterns (Figure 5, 6a).
Figure 5. An EBSD montage map of the entire weld zone, IPF on IQ map, 133 fields and 91M points @ 750 nm steps. The scalebar is 4 mm.
Figure 6. a) A detailed IPF on PRIAS center map of the recrystallized austenite layer. b) An IPF on PRIAS center map showing the columnar austenite grains crystallized from the former melt pool.
This structure suggests that the temperature in this band has been high enough to allow recrystallization of the deformed structure. Still, the metal remained solid and cooled down too fast to enable significant grain growth.
The fine-grained austenite microstructure is in stark contrast to the area that has been melted. Atomic movements in the melt were so fast that upon cooling, very large columnar grains could form in two bands with a seam in the middle (Figure 6b). This double band structure indicates solidification that started on both sides of the melt pool on the solid metal surfaces until the grains met in the middle.
The structure on the ferrite side is more complex. In the middle of the weld contact, a triangular area of apparent columnar grains has formed with a void at the bottom tip (Figure 7). Underneath these larger grains is a fine-grained band over the full width of the weld. In both zones, the EBSD patterns have deteriorated compared to the original ferrite matrix structure.
Figure 7. An IPF on PRIAS center map showing only the ferrite phase.
The substrate’s ferrite structure has a complicating factor when compared to the austenitic steel stud. When ferrite is heated close to the melting temperature, the crystal structure changes into that of austenite. This means that the columnar grains in the ferrite’s center area actually crystallized from the melt pool as large austenite grains, similar to those shown in Figure 6b. However, in this case, the austenitic crystal structure is not stable at room temperature, and upon cooling, these grains changed back into ferrite. This back transformation does not simply change the structure of the entire grain into the ferrite structure. Instead, it creates organized clusters of ferrite grains with different “child” orientations inside each original austenite grain. At low magnifications, these clusters give the impression of columnar grains with a single orientation. In reality, each column may contain many small grains with up to 12 different orientations (Figure 8).
Figure 8. A close up of the columnar grains in the ferrite area.
Below the melt pool, the original small ferrite grains also transform into austenite during the welding process, but the temperature does not get high enough to enable significant grain growth while the metal remains solid (Figure 9). When these small austenite grains transform back into ferrite during cooling, the original grains also get split up into many small ferrite children, which are a little deformed. That is what causes the dark appearance of this zone in the IQ map in Figure 3a.
Figure 9. An IPF on PRIAS center map of the fine-grained ferrite zone below the melt pool.
The latest version of OIM Analysis™ contains a new tool to investigate this phase transformation from high-temperature austenite into the low-temperature ferrite phase. The clusters of child ferrite grains can now be fit together into the original high-temperature austenite structure that was present during (weld) processing. This is extremely helpful in understanding the contact between the ferrite substrate and the crystallized melt pool.
The EBSD map in Figure 10a shows the compound columnar grains in the ferrite substrate with the two bands of columnar grains that were formed from the melt on top. There is no obvious microstructural correlation between the ferrite and the first layer of columnar grains from the melt in this image. However, when the high-temperature austenite microstructure is reconstructed, the columns in the ferrite substrate coalesce into single austenite grains that match the bottom layer’s orientation perfectly.
Figure 10. An IPF map of the ferrite-melt contact zone. a) The microstructure as measured and b) the microstructure after reconstruction of the high-temperature austenitic microstructure. The scalebar is 1 mm.
This means that these columnar grains grew continuously from the melt at high temperature when the weld was formed then separated into the fragmented ferrite structure and intact austenite structure upon cooling. The exact location of this boundary between the final ferrite and austenite phases is determined by the chemical composition, especially the Ni content (Figure 11). The ferrite substrate has little Ni, which causes the austenite to transform back into ferrite upon cooling.
Figure 11. A simultaneously collected EDS map for Ni. The scalebar is 1 mm.
Figure 12. An IQ detail map of the left melt pool, 2.5M points @ 1.5 µm steps. The scalebar is 900 µm.
Finally, the Ni distribution in the melt pool also explains the occurrence of the dark bands or swirls in the IQ map in Figure 12. When the stud was pushed into the melt pool, the melt fractions from the stud and ferrite substrate mixed, but not thoroughly. On the left side, an area that is somewhat depleted in Ni can be recognized in the melt pool, and where the Ni content drops below a threshold, these portions of the austenite grains that grew from the melt are not stable and transform into ferrite (Figure 13).
Figure 13. a) A phase map of the left melt pool area. Green is austenite and red is ferrite. b) An IPF on PRIAS center map of the left melt pool as measured. The scalebar is 900 µm.
The parent grain reconstruction (Figure 14) confirms that all the small ferrite grains exhibit orientations derived from the coarse austenite grain structure.
Figure 14. An IPF on PRIAS center map after parent grain reconstruction.
Just like the investigation of natural rocks, a detailed analysis of the microstructure of a metallurgical sample like this weld contact can provide a comprehensive image of the active microstructural processes with only minimal prior knowledge. Starting with an overview of the entire sample, followed by targeted scans of distinct parts of the microstructure using EBSD and EDS, a timeline of events can be reconstructed that produced a strong bond between two materials.
And what does it matter if the creation of a structure takes milliseconds or perhaps many millions of years? In the end, everything is connected.
Dr. Jordan Moering, Mid-Atlantic Sales Manager, EDAX
Like any nerdy high schooler growing up in the suburbs of North Carolina, I had a lot of weird hobbies growing up. Some of these turned into scientific interests that brought me to grad school, and others just turned into party tricks (Rubik’s cube, anyone?). I’ve recently been thinking about one of these hobbies, especially after hearing our recent webinar on OIM Matrix and Forward Modeling.
When I was 15, I thought that rubber band balls were really cool. I’m not sure why, but I really enjoyed the twists and turns of the multiple layers of bands on top of each other. The aspect ratio of the band thickness to the curvature and angle of the bands was something I found really fascinating. Fast forward a few decades and what used to be a softball-sized rubber band ball is now a 12 kg behemoth that I pass every day while walking to my office. I’ve been giving a lot of thought to this ball recently because of how it mimics the dynamical diffraction simulation used in dictionary indexing.
Figure 1. (left) My 12 kg ball of rubber bands. (right) Simulated Kikuchi patterns for EBSD analysis.
I suppose what really fascinates me the most about these projections is how accurate they are. The fundamentals of diffraction can be so simply described in Bragg’s law, but the implications for these phenomena are profound. Because different crystallographic indices diffract incoming electrons at different intensities, the collected image shows the orientation of the crystal wherever the electron beam was parked. The resulting bands (called Kikuchi lines) are a direct representation of the material’s crystal structure.
Now, I’m not an expert on diffraction, but I find all this to be fascinating. What’s cool to me is that recent developments in computing and modeling have enabled new types of indexing. This includes Dictionary Indexing which utilizes an entirely synthesized library of diffraction patterns to correlate the likely orientation of every collected pattern when obtaining EBSD scans. What’s fascinating to me is that these simulations historically struggle to account for artifacts in the Kikuchi patterns like lens blurring, phosphor illumination, etc. With the advent of direct detection cameras however, there is no need to account for these as individual electrons create the image on the sensor. New techniques like forward model-based indexing are only accelerating the adoption of this new technique. And at the core of these new modeling techniques are simulations – simulations of Kikuchi patterns.
So yeah, I see my rubber band ball every day and think about simulated diffraction patterns. I suspect that it is a very low symmetry system based on the geometry.