Energy-dispersive X-ray spectroscopy (EDS) is an analytical method used for chemical characterization or elemental analysis of a sample.
It can be challenging to perform EDS analysis on a sample with a rough surface, such as a fracture surface, due to shadowing effects as shown in the schematic in figure 1.
Further complexities occur from surfaces considerably inclined towards the detector and from other surfaces sloping away from the detector.
Challenges with Performing EDS Analysis on Rough Surfaces
The schematic in Figure 1 shows the scattering of X-rays at a single point where a hill obstructs the X-rays from being identified by the EDS detector. When surfaces are considerably inclined towards the detector and other surfaces are sloping away from the detector, further complexities occur.
Figure 1. (a) Schematic of X-Ray scattering at a single point where a "hill" prevents X-Rays from reaching the EDS detector.
Figure 2 shows an example of a fracture surface in sandstone. Shadowing can be clearly observed from the EDS Si map. While the shadowing is evident, it is not clear if a dark region in the map is due to shadowing or because of low Si content.
Figure 2. Secondary electron image (left) and a silicon EDS map (right) of a sandstone fracture surface.
It is possible to alleviate the shadowing effects in two ways:
- Collect a set of maps from an area, rotate the sample 180° and then collect the same set of maps over the same area, rotate the results and merge with the original scan results.
- Use two well-balanced detectors simultaneously with different azimuthal angle and then merge the two sets of maps.
An issue with the first technique is that it is time consuming and often results do not merge successfully.
Figure 3 shows an example of two EDS detectors mounted on an SEM. Using dual EDS detectors allows the shadowing effects to be mitigated. This technique is not only more efficient as the area needs to be mapped only once, but also reduces any errors related to registering the two sets of maps using a single detector.
Figure 3. Interior view from below of an SEM (JEOL 700F) with dual silicon drift detectors.
However, simply summing the two datasets together is not sufficient to properly capture the missing data. A better approach would be to combine the data by choosing only the maximum signal at each pixel.
Usually, a normalizing procedure is required to eliminate topography effects, as the count rates may be below average when viewed by one detector while the other detector is completely shadowed.
A commonly available normalization procedure is to analyze the merged dataset with a ZAF procedure or a k-ratio.
A backscattered electron image (BEI) from a fracture in an aluminum alloy can be seen in Figure 4. The surface of the sample is very rough. Several different phases seem to be evident in the sample as a dendritic structure can also be seen.
Figure 4. Backscattered Electron Image (BEI) of a fracture surface in an aluminum alloy sample.
Examples of EDS Maps from Two Detectors
EDS maps from the two different detectors are shown in Figure 5. Here, the full spectrum constitutes the region of interest (ROI). Again, the shadowing is clearly seen in these maps.
Figure 5. EDS maps of fracture surface obtained using two different detectors.
Figure 6 shows an EDS map created by simply summing data from both detectors. Areas that provided signal to both the detectors are represented in white, those that did not provide signal to either of the detectors are black and the areas that provided signal to just one of the two detectors are gray.
Since the ROI is for the entire spectrum, the contrast in the map is an artifact arising from the signal distribution between the two detectors and is not chemical.
Figure 6. Sum of the EDS maps of the fracture surface obtained using two different detectors.
The sum map for the full spectrum ROI can be used in conjunction with EDS maps for real elemental ROIs. In these elemental maps, however, instead of summing the ROI data at each pixel from both detectors together, the maximum counts between the two ROI data sets are used.
These maps are overlaid on an inverted and binarized sum map for the full spectrum ROI. In these combined maps, white points are regions that cannot be detected using either of the two detectors. The black points indicate areas where the element is not present. An example for two elements, silicon and copper, is shown in Figure 7.
Figure 7. Maximum signal EDS maps for Cu and Si from the fracture surfaces.
However, this map still has effects of topography in it. The mid-toned areas may be due to topographic effects or points with less content of the specified element. To resolve this ambiguity, the maximum signal data must be normalized and a ZAF correction has been used.
In Figure 8, the results of the ZAF normalization on a map where all of the elements are combined together in a single map is shown The areas for which it was not possible to obtain any signal using either detector is now shown in cyan.
Figure 8. Maximum signal EDS maps are shown in overlay mode both before (left) and after (right) ZAF normalization.
Dual detectors are very useful for resolving the topographic and shadowing effects associated with rough surfaces. This can help in a range of application areas such as the characterization of wear surfaces, fracture surfaces or as-machined surfaces.
In order to have the map with the clearest impression of the chemistry without the ambiguity of sample topography, the maps should be merged using a maximum signal function and the dataset should then be normalized in some way.
This information has been sourced, reviewed and adapted from materials provided by EDAX Inc.
For more information on this source, please visit EDAX Inc.