OIM Matrix™ is a powerful software tool available as an optional module with the OIM Analysis™ platform that allows users to simulate electron backscatter diffraction (EBSD) patterns using dynamical diffraction models to produce more realistic pattern simulations compared to traditional kinematic diffraction simulations. These more realistic simulations can be easily and accurately compared to experimental patterns to improve indexing performance. With the release of OIM Analysis 9, spherical indexing is available in the OIM Matrix module, which significantly improves the time to results and reduces the barrier of entry to using this powerful technology. Tips and tricks for getting started and using spherical indexing are in this article.
What do I need to run OIM Matrix?
OIM Matrix is an optional module within OIM Analysis. For operation, you need a valid license for both OIM Analysis and OIM Matrix. To use OIM Matrix spherical indexing, you must be running OIM Analysis 9 and need an appropriate graphics processing unit (GPU) card installed on your PC. OIM Analysis and OIM Matrix can be used on the EDAX PC connected to the EBSD detector and the scanning electron microscope (SEM), plus on a separate PC using a remote software license.
What kind of GPU card do I need?
The GPU implementation for spherical indexing requires a modern Nvidia GPU card. For EDAX PCs on the SEM, the RTX A4000 card is recommended as a high-performance single-slot GPU. For remote analysis PCs, you can also use this card. Additionally, you can utilize cards within the RTX 40 series. All these cards have at least 6 GB of GPU memory, sufficient for spherical indexing analysis. The relative performance of different GPUs is available on the PassMark software website using the G3D rating as a benchmark.
| Card |
|
Benchmark |
|
Slot requirement |
| RTX A4000 |
|
19354 |
|
1-slot |
| RTX 4060 |
|
19367 |
|
2-slot |
| RTX 4060 Ti |
|
22512 |
|
2-slot |
| RTX 4070 |
|
26857 |
|
2-slot |
| RTX 4070 Ti |
|
31680 |
|
Varies by manufacturer |
| RTX 4080 |
|
34704 |
|
3-slot |
| RTX 4090 |
|
38892 |
|
3-slot |
Note that OIM Analysis and OIM Matrix can be run on laptop PCs using laptop versions of the RTX 40 series cards.
How do I set up my EBSD data acquisition in APEX™ to use spherical indexing?
Spherical indexing is an offline analysis of saved EBSD patterns. To perform this work, the EBSD patterns must be saved during data acquisition. This is done by activating the Save Patterns option in the Scan Parameters window, as shown in Figure 1. The UP2 format is recommended as it provides the highest bit-depth images, but the UP1 format, with its corresponding lower bit-depth, can be used if file size considerations are important. This option will create a *.up2/.up1 pattern file and save it in the same folder as the APEX project file. When performing analysis away from the EDAX PC, it is essential to copy both the APEX project file and the required patterns files to the remote analysis PC.
Figure 1. The Scan Parameters window.
Spherical indexing will work with all EDAX EBSD detectors. For Velocity™ and Clarity™ detectors, 4 x 4 binning is recommended for most applications. The EBSD pattern pixel resolution (120 x 120 pixels for Velocity and 129 x 129 pixels for Clarity) provides sufficient resolution for spherical indexing and orientation refinement performance. Examples of this are found on the Being more precise and Being more precise again EDAX blog posts.
How do I set up spherical indexing in OIM Analysis?
When you open the dataset you want to use for spherical indexing, ensure the associated pattern file is recognized within OIM Analysis. If the patterns are available, you will see a Patterns section in the Sample Summary window that details the pixel size, image bit depth, and file name and location. This association of the scan and pattern files should occur automatically if both files are in the same Windows directory. If they are not, you can still manually associate a pattern file with the scan by right-clicking on the dataset in the Project Tree and selecting the Reindex Configuration\Associate Pattern file option.
Figure 2. The Reindexer button in the Utilities toolbar.
Select the Reindexer button on the Utilities toolbar, as shown in Figure 2. This will open the Reindexer window, as shown in Figure 3. Within the Reindexer, select Spherical as the indexing mode. Other reindexing modes available are dictionary indexing and Hough indexing that both require saved EBSD patterns; and Hough peaks indexing, which uses the Hough peak information saved during APEX EBSD mapping. The mode for orientation refinement is also available.
Figure 3. The Reindexer window.
A simulated master pattern should be associated with each phase of interest. Select the Phase option by pressing F6 or selecting the Phase option from the pulldown list shown in Figure 4. Once selected, this will show the phase(s) used during initial indexing in APEX. Select the phase(s), then select Load to associate a master pattern file. OIM Analysis 9 contains a preloaded library of master patterns for common materials located in Program Files\OIM Analysis 9\Materials. If your material of interest is unavailable in this library, a new master pattern can be generated using the Crystal Structure Builder tool on the Utilities toolbar.
Figure 4. The Phase option pulldown list.
Once the master pattern is associated, you can start the reindexing process by pressing the Re-Index Partition button in the bottom left portion of the Reindexer window, as shown in Figure 5. Note that this reindexing is based on partitions. In this example, I am using the All Data partition (seen in the Project Tree in Figure 2), which allows all data points to be reprocessed using spherical indexing. This partition-based approach allows for the reindexing of defined subsets. For example, only low-confidence index points or a selected phase could be reindexed. This increases the efficiency of the reindexing process by only using it when required.
Figure 5. The Re-Index Partition button in the Reindexer window.
What band width setting should I use?
Spherical indexing single parameter users need to select, termed band width. Higher band width values provide finer pattern details from the simulated master pattern used in the pattern-matching indexing. Band width values of 63 or 95 can be used to start your analysis for many samples and applications. The band width value is set in the Spherical Indexing section (F3), as shown in Figure 6. When trying to differentiate similar patterns in challenging applications like pseudosymmetric patterns, higher band widths can be useful. Note that the indexing time is a function of the band width parameter.
Figure 6. The band width is set in the Spherical Indexing section.
Can I use NPAR™ with spherical indexing?
Yes, NPAR can be combined with spherical indexing to provide higher signal-to-noise EBSD patterns in the indexing algorithm for improved indexing success rates. Additionally, other image processing options are available in the Image Processing panel, as shown in Figure 7. A circular filtering mask can be used for the traditional phosphor-based EBSD detectors, while a mask designed for the Clarity direct detector is also available. NPAR and its NLPAR extension are both available to improve EBSD signal-to-noise through local pattern averaging. There are several background correction options that can be used to optimize the background processing for both single and multi-phase samples. There is also an Advanced image processing toolbox with a range of tools for tackling less frequent problems.
The Adaptive Histogram Equalization is a normalization function recommended for spherical indexing.
Figure 7. The Image Processing panel.