Applications of High-Speed CMOS Cameras for EBSD Micro-Structural Analysis

In this interview, Matt Nowell, the EBSD product manager at EDAX, talks to AZoM about applications of high-speed CMOS cameras for EBSD micro-structural analysis.

Could you give us a brief overview of EBSD detector history?

We're getting closer and closer to the 100-year anniversary mark of the initial diffraction experiments which would lead to what we now term Electron Backscatter Diffraction (EBSD) patterns. The initial EBSD patterns were captured using film. You'd put this into the chamber, expose it, bring it out, develop it, and look at it. This is a much slower collection time than we're used to now. You were collecting patterns from single points or single grains. When you look at these patterns, you'd have to deal with things like intensity gradients and image distortions, but you could get some nice-looking patterns with interesting crystallographic information. However, they were not a practical technique for orientation mapping over a large number of grains.

The next step in the evolution of this technology that started to bring it to where it is today was the integration of cameras with personal computers. David Dingley pioneered the use of online indexing, where he would couple a silicon-intensified target camera into a video capture board, into a PC, get the pattern into the PC and then do the crystallographic indexing by clicking on known zone axes in the pattern. The next step was the automated indexing. Rather than manually having to click on a zone axis, the computer could recognize the band positions within the pattern and determine the crystallographic orientation. We call this process indexing.

The third step in this development process was full automation. This was done by Dr. Stuart Wright in 1992, creating orientation imaging maps, or OIM maps, where each pixel can be colored according to orientation. With these cameras, we were limited to TV rate imaging, taking the analog signal from the camera into the frame grabber.

If we look at indexing speeds over the years, in the mid-'90s when we had analog cameras and the maximum frame rate was somewhere around 25 to 30 frames per second. 2001 was the introduction of digital CCD cameras, where we initially jumped up to about 40 patterns per second, and that continually improved until around 2006 with the introduction of high-speed digital CCD cameras. That led to a jump, initially of around 200 frames per second. That technology continued to evolve until we achieved speeds of around 1,500 patterns per second.

Then, a couple of years ago was the introduction of high-speed digital CMOS cameras. That led to a leap to around 3,000 patterns per second and now looking at speeds with our latest cameras of 4,500 index patterns per second or greater.

That brings us up to EDAX’s new Velocity™ EBSD Camera Series. Can you tell us a bit more about these cameras?

We have two different models. We have our Velocity™ Plus, which has an indexing speed up to about 3,000 index points per second. Then, the Velocity™ Super, which has speeds of up to 4,500 index points per second. This is powered by a high-speed, low-noise CMOS sensor. The imaging sensor has a 640 by 480-pixel resolution sensor. At the highest speeds, we operate at a 120 by 120-pixel image resolution, which is achieved by binning the original image into a lower resolution image.

At this resolution, not only do we get faster camera operation, but the patterns are higher pixel resolution than we had with CCD sensors at high speeds. At the 1,500 points per second maximum speeds with CCDs, we were operating at 30 by 30 pixels. With the CMOS sensor at 4,500, we're now at 120 by 120 pixels, so we're getting more information faster with a CMOS sensor. This pixel resolution improves band detection, which in turn improves indexing performance and also improves orientation precision even at the high speeds we can now obtain.

At full resolution of the camera, we get a 480 by 480-pixel resolution image. We can bin this down. With 2x2 binning we're averaging together a two by two block of pixels to go to 240 by 240-pixel resolution. We can bin to four by four, where we get 120 by 120 pixels. That's our standard-setting for scanning. Then we have an eight by eight-pixel binning set, where it's a 60 by 60 pixels.

The camera itself can operate at full speed and full resolution. It can collect frames at this rate, but the image transfer to the computer is limited by the size of those images. We can get a faster transfer from the camera into the computer as we increase binnings. At four by four binning, we can get 4,500 index points per second. The eight by eight binning does not increase scanning speed. That's just done to reduce the file size if you want to save the patterns for several different applications.

In short, the Velocity™ Camera Series can provide quality indexing with speeds up to 4,500 index points per second and can be used on a range of materials, including deformed materials, non-cubic materials, and multi-phase materials. It's ideal for things like in-situ and 3D EBSD applications.

What are the differences between CCD and CMOS sensors?

The biggest change that has occurred during this transition to high-speed cameras is the change from a traditional CCD sensor into the high-speed CMOS sensors. Both sensors use silicon as the photodiode material. As the light photon hits the silicon, it's converted into electron charges to determine how much light signal we're capturing per pixel.

CCD is an older technology and, generally, people say they use a single readout amplifier, but in practice sometimes a double readout amplifier is used. This acts as a bottleneck. With CMOS sensors, each pixel has its readout amplifier, so we’re are going from 1-2 amplifiers to 307,200, allowing us to get significantly faster readouts and faster frame rates.

On the CMOS, the charge is collected at each pixel and then passes through the amplifier. Once it's been through that processing that it can be more quickly transferred through the vertical and horizontal signal lines. Each pixel has its own dedicated electronics, this results in faster frame rates.

It's important to realize that the sensor is only one part of the detector. The sensor itself is a high-speed sensor that's been optimized specifically for our EBSD applications, but the entire system has been optimized for EBSD performance for this application.

The phosphor screen itself, we want it to be bright. We have a limited number of electrons fixed by the beam current on the SEM, so we want to be able to get as much light from the number of electrons we're using, but we also want it to be fast. The decay time of the phosphor has to be able to enable our high-speed mapping, so we've picked a phosphor screen that's both bright and fast. The lens we use in the Velocity™ is custom-designed for both sensitivity and performance, as well as low distortion, so it's a very fast, very efficient lens.

Can you tell us a bit more about the Pattern Region of Interest Analysis System (PRIAS™)?

PRIAS™ is an imaging technique that uses the phosphor screen and regions within the phosphor screen as different imaging channels. It has three different modes. One we call PRIAS™ Live, where we just set up a five by five array of regions of interest and can take a quick image. We have PRIAS™ Collection, where we monitor three different regions while we're collecting our EBSD data. We have a top channel, a middle channel, and a bottom channel. We can also save all the patterns and freely define different ROIs with what we call PRIAS™ Analysis.

Do EDAX’s detectors have any other functions that you can tell us about?

In addition to being able to just get our orientation maps at high speeds, we're also able to couple these with our EDAX EDS detectors. We have our Octane Elect and Octane Elite detectors that provide the highest input and output count rate.

Another unique function and feature we have is something that we call NPAR™, which is our Neighbor Pattern Averaging and Re-indexing. This is an approach to improve the signal-to-noise of the patterns while we're scanning. We save each pattern and then we reprocess them by taking a pattern and taking the surrounding kernel of patterns, averaging them together to improve the signal-to-noise, and then reindexing this lower-noise pattern. Of course, you are now effectively making your spatial acquisition volume a little bit larger but, oftentimes the trade-off of having a stronger pattern for that kernel area outweighs the fact that it's a larger sampling volume.

I want to say a little bit about what the implications of the speed are and what we could use it for. What we found, over the years generally, as we've made detectors faster and faster, people have just collected larger and larger data sets. We can get a higher density of sampling points in the scan grids so we can better define grains, grain boundary, and substructure.

We can look at larger scan areas and lower magnifications, giving us better statistics, better texture, or preferred orientation determination. We can get more pixels for improving our angular precision but, with bigger datasets, just remember, we are dealing with larger storage requirements. Your computational time just starts to get larger as you're processing on from 1 million to 10 million to 50 million-point datasets, it does take a little bit longer to churn.

The reason I mentioned that is, bigger isn't necessarily always better. You have to think about what you're trying to measure, what the goal of your experiments is.

How does Orientation Imaging Microscopy (OIM) Analysis™ make data handling easier?

There are several features within OIM Analysis™ to assist with handling larger datasets. First, it’s a 64-bit application, which means it can handle larger datasets. Second, the analytical functionality has been multi-threaded, which allows for the computing requirements to be distributed over the available CPU hardware. Third, there are a number of functions which are specifically implemented to preview the analysis function to determine if it’s worthwhile to proceed with the full analysis. These features take advantage of the statistical nature of the EBSD data.

Why is high-speed collection so important?

High speeds are ideal for dynamic experiments because we're able to collect data at 3,000 to 4,000 points per second, we can get more grains in the same time slides and less temporal change across the map.

The other application where the high-speed collection is essential is 3D EBSD, where we take an image and then we take a slice. This is generally done with a focused ion beam but can also be done with broad beam sources or mechanical polishing. Then, we take another EBSD image. Traditionally, EBSD has often been considered the limiting factor, but now with fast cameras, it's not. Having a high-speed camera allows you to get more slices, large area per slice or you get each slice faster.

About Matt Nowell

Matt is the EBSD Product Manager at EDAX and has a passion for EBSD and microstructural characterization. Matt joined TexSEM Labs (TSL) upon graduation from the University of Utah in 1995 with a degree in Materials Science and Engineering. At TSL, he was part of the team that pioneered the development and commercialization of EBSD and OIM. After EDAX acquired TSL in 1999, he joined the applications group to help continue to develop EBSD as a technique, and integrate structural information with chemical information collected using EDS.

Within EDAX, Matt has been involved in a number of roles, including product management, business development, customer and technical support, engineering, and applications support and development. Matt has published over 70 papers in a variety of application areas. He greatly enjoys the opportunity to interact with scientists, engineers, and microscopists to help expand the role that EBSD plays in materials characterization. In his spare time, Matt enjoys playing golf and pondering if changing the texture of his clubs will affect his final score.