Measuring Elemental Distributions in a Sample

SEM/EDS (Scanning Electron Microscopy and Energy Dispersive Spectroscopy) mapping is utilized to measure elemental distributions in a sample. Usually this method generates an analysis of one field of view by scanning the electron beam in a grid pattern to create a map.

This article discusses utilizing Thermo Scientific™ Pathfinder™ X-ray Microanalysis System to control SEM sample stage motors to gather extremely large X-ray maps. This is done by attaining multiple fields of view in a grid pattern and combining those maps into one large X-ray map.

This expanded field of view is much simpler to understand, captures X-ray maps larger than the field of view provided by the SEM, and can capture features which can be missed when analyzing only a limited field of view

For more than five decades, element distribution maps created by SEM/EDS have been collected by analysts. In the beginning, as the electron beam was scanned, the method was to put a white dot on a sheet of photographic film every time an X-ray for a given element was detected. In the early 1980s this gave way to digital X-ray maps. A digital X-ray map from ca 1985 can be observed in Figure 1.

It is worth noting that this required around an hour to gather relatively few X-ray counts. Every aspect of SEM/EDS mapping has improved a great deal since then and SEMs are much more capable of focusing many electrons into a small spot.

Furthermore, a lot of modern SEMs are equipped with motorized stages that can be controlled by the EDS system. Storage systems, computers, and displays are all much more capable. Processing just one spectrum used to take more than five minutes, but today many thousands of spectra are processed per second.

Figure 1. Digital X-ray map from ca 1985. This required about an hour for acquisition.

Crucially, the EDS detector itself is far quicker than before. The Silicon Drift Detector (SDD) has superseded the Si(Li) detector design and even the SDD has had many generations of improvement. Presently, the Thermo Scientific™ UltraDry Silicon Drift Detector can gather X-ray data at the rate of several hundred thousand counts per second while delivering resolution equal to or better than the Si(Li) detector.

Additionally, newer UltraDry models provide large solid angles that give more X-ray counts at lower beam current. By utilizing two SDDs at the same time, and thereby doubling the throughput, even faster measurements can be acquired.

Acquiring Very Large Area Maps

The entire SEM/EDS system, combined with the newest Pathfinder software electronics, can now generate high quality X-ray maps in minutes. Through the decades the constant theme is that the analyst selects a particular field of view to measure and report to those requesting data.

The much improved capabilities of modern equipment give the analyst the power to expand the view presented to requestors and gather extremely large X-ray maps by collecting lots of individual maps and merging those into one large map. The final maps can be over tens of millimeters on one side.

The level of detail and size of the map correlates with the time required to create a wide area map. The assertion here is that very large maps can be produced in the reasonable time of between one and three hours, a half day of instrument time. In the instance of a valuable sample which could have taken some length of time to gather and a day or more to mount and polish, this is a reasonable amount of time considering the value of the data acquired.

Analysts often collect maps from only a small field of view because of the difficulties in collecting information quickly. To see how this bit of information fits into the overall structure of the sample it takes some imagination. A large field of view produces a scene which is much simpler to interpret, particularly for those not accustomed to observing SEM/EDS data.

Broad distributions of features in a sample can be observed with a larger field of view, not just one or two examples as has been the case in the past. Relatively rare features can be observed that might easily be missed when examining just one or two areas. This is crucial because the composition of a sample is usually not clear by just viewing the electron micrograph.

It also maps distributions which are bigger than the field of view supplied by the SEM, delivering views that were not possible otherwise. Large area mapping by SEM/EDS is much quicker than analysis by some other methods because most of the spatial information is collected by moving the electron beam which is extremely quick, relatively little slower stage motion is needed.

Lastly, this rich data set can be revisited in the future to find the composition at any point in the region analyzed, as Pathfinder software mapping saves X-ray spectra at each point in the map.

Figure 2. Array of three rows and four columns of images representing 12 SI maps acquired in a grid pattern.

Technique

Analysis Automation is a feature of Thermo Scientific™ Pathfinder™ X-ray Microanalysis Software which controls the SEM stage whilst gathering Spectral Imaging (SI) X-ray maps. Analysis Automation moves the sample stage in a grid pattern or it can use a random set of locations. The particular set of locations can be saved for future utilization.

In order to collect a large map, the analyst sets up the system in the same way as for collecting a single map. The SEM beam current and kV are set, the associated image size is set, and the map array size is set, then the pulse processor time constant is chosen. After that, the analyst sets a termination criterion to instruct the system how much data to gather in each map. Maps can be stopped after acquiring a specific amount of data or after a fixed time.

Lastly, Analysis Automation is used to set the pattern of locations to analyze and then the start button is pushed. The data are gathered without intervention and kept on the mass storage system. The entire data set can be processed as one large map or individual maps may be processed.

The analyst selects elements, data type and related parameters in order to create a tiling of the whole data set, just as when observing a single map. Results can be calculated as weight percent, atomic percent, or simply net counts. Next, batch processing is invoked and the name of the data set is chosen, Pathfinder software then calculates a result.

The numerous individual maps are treated as one large map and the result is a series of images in one data set. There is one image for each element selected and one for the electron image. Each image is as large as all of the individual maps tiled together and the element maps can be overlaid onto the image. The color, brightness and contrast of each map can be altered.

Example Results

An analysis of a ceramic brake pad sample is shown in Figure 2. There is a grid of four rows and three columns of images which represents twelve SI (Spectral Imaging) maps. Each map is a 512 x 384 array of spectra. These were gathered at 15 kV and each map took about five minutes to attain. After merging the individual images, the resulting image is shown in Figure 3.

The net counts maps for all of the maps calculated from this data set are shown in Figure 4. Net counts maps have the X-ray continuum background removed and separate overlapped peaks. There is so much data that it is useful to have a large monitor to view it. Otherwise, printing the data as a poster can be extremely helpful as it allows several engineers to view and compare multiple maps whilst discussing the results.

Figure 3. Image formed by merging together of the separate images shown in Figure 2.

Figure 4. Net count maps of a ceramic brake pad sample. The map consists of an array of spectra 2,048 across and 1,536 high.

Some selected elemental maps of the same sample overlaid on the SEM micrograph can be observed in Figure 5. This shows a number of features of the sample and it is immediately clear that materials containing Ba are extremely common. The Cu threads and Cl rich material (possibly a PVC) though large, are quite rare.

Selecting a field of view that omitted these materials would have been easy to do. The Sb rich particles (actually SbS) are small and distributed unevenly. Imagine gathering the same twelve maps but printing them in a booklet form. The end requestor would have to flip through these one after another without the advantage of the overall view. It is much easier to extract detail from a large, complex view than from a sequence of images which are ordered in time.

Pathfinder software has two more tools, in addition to calculating element maps, COMPASS and Phase, which extract multi element phases from SI maps. Two phases from the brake pad sample, (Ti, K oxide and SbS) overlaid on the micrograph are shown in Figure 6. This shows at a glance the usual sizes, shapes, and distribution of these materials.

Two phases are shown in Figure 7, a C and Cl (possibly PVC) phase and a Ca ceramic (Mg, Al, Ca silicate) phase. So again, at a glance it is easy to see the typical shape, size, and distribution of these materials. Even though the C and Cl phase is quite big it would be easy to ignore it.

Many analysts would actually ignore it intentionally as an artifact and focus on the smaller, more common features because it is so large. This is a benefit of acquiring data from a large field of view. It shows large areas of the sample, not just those selected by the analyst. What might be dismissed as an artifact may, in fact, be crucial information.

Figure 5. Selected elemental maps of the ceramic brake pad overlaid on the micrograph. The field of view is about 2.8 mm wide and 1.5 mm high.

Figure 6. Phase overlays of a Ti, K oxide ceramic shown in cyan, and SbS particles (shown in gold) overlaid on the micrograph.

Figure 7. Phase overlays of Ca rich ceramic (Mg, Al, Ca Silicate) phase shown in purple, and a CCl (possibly PVC) phase shown in green.

Figure 8. Silicon solar cell. The thin traces are composed of Ag paste.

Figure 9. Segment of the Te map overlaid on the micrograph showing one Te inclusion.

A photograph of a silicon solar cell is shown in Figure 8. Large area mapping was employed to analyze the composition of one of the metallic traces over a long distance. These are nominally composed of Ag paste. In fact, analyzing a number of typical fields of view would show only Ag. Small areas of Te inclusions appeared only when measuring over more than a centimeter of the trace.

This measurement was similar to gathering a line scan, 257 maps were gathered in one row and each individual map was only 64 x 54 spectra. Each image is 512 x 442 with a resolution of around 0.27 microns. The pixel resolution is 2.1 microns. The complete size of this map is about 3.5 cm by 115 microns. The entire acquisition time was about three hours and each map with image took under a minute to collect.

A section of the Te map overlaid on a micrograph can be seen in Figure 9. This is net counts data with the background signal removed. There were large parts of this sample which showed no evidence of Te. An analyst could have inspected a number of areas and not having seen the Te simply reported the trace as pure Ag if they were asked to characterize this sample.

It was noted that Te was visible mainly where the trace appeared damaged. Figure 10 is a spectrum gathered from one inclusion confirming the presence of Te. This type of information is extremely useful for an engineer familiar with this product, as it shows the types of defects present, their frequency and their size. This is difficult to gather from viewing selected areas. This data set has an aspect ratio of around 305 to 1.

Enlarging the image to be 10 cm high makes the image look more than 30 m wide. Yet, if there are any flaws in this material due to a manufacturing flaw or corrosion, then this detailed view is of huge value in locating them. This example illustrates the brilliant magnitude of data acquisition possible with the Thermo Scientific™ UltraDry EDS Detector, and Pathfinder software with Analysis Automation.

Figure 10. Spectrum from one inclusion confirming the presence of Te.

Summary

Analysts now possess another powerful tool in their arsenal to study samples. Large area maps whether in a linear array, rectangular grids, or as random locations, permit the analyst to gather vast amounts of data from samples with little more effort than required to gather a single X-ray map.

The high throughput of the UltraDry Silicon Drift Detector, improvements in SEM technology, the speed of the Pathfinder software electronics and the enhanced power of modern computers all combine to make this method accessible to analysts. These large data sets can be acquired in a reasonable length of time because of this improved technology.

For many samples, generating a map of a wide field of view makes the result more understandable to those who are observing the data, and also makes it more inclusive by covering a bigger area of the sample.

The result is also more reliable because alterations in composition are not always indicated in the SEM micrograph, so covering more area in a map heightens the chance of a more comprehensive analysis. All of these advantages improve the quality of the work provided by the analyst.

This information has been sourced, reviewed and adapted from materials provided by Thermo Fisher Scientific – Materials & Structural Analysis.

For more information on this source, please visit Thermo Fisher Scientific – Materials & Structural Analysis.