Particle Analysis on Table Top Scanning Electron Microscopes

Introduction

Particles in powder samples come in all shapes, sizes and compositions, and the requirements for their analysis and characterisation can be just as varied. Particle analysis tasks may require only morphological measurements from a single field of view or could require the combination of morphological and chemical data from many thousands of particles from a large sample area. A Table Top Microscope (TTM) equipped with an AZtecFeature X-Ray Energy Dispersive Spectroscopy (EDS) system provides an extremely versatile platform for performing the full range of particle analysis applications.

AZtecFeature uses the capabilities of the TTM to acquire images which are used to locate particles in a sample. The AZtec X-Ray EDS detector is then used to collect a spectrum from each located particle. By controlling the TTM's motorised stage, AZtecFeature performs this analysis over large areas of the sample, either to increase the population of particles analysed or to find particles that are sparsely distributed over the sample. The particles can be classified in terms of their morphological and/or chemical parameters, either during a run or after it ends. The classification results obtained during a run provide feedback of the types of particles being detected. These classifications can also be used to determine if additional analysis steps for individual particles are added or skipped. Finally they can be used to control the termination of an analysis session, when specific analysis targets are achieved.

Metal Powders – Determining Particle Size Characteristics and Contamination Levels

A Ti based powder was spread over a carbon adhesive tab on a sample stub and mounted in the TTM. The aim of the analysis was to characterise particle shapes and sizes and determine the extent and type of any contamination present.

AZtecFeature locates the particles in the electron image by applying one or more grey level thresholds. Setting the grey level thresholds in AZtecFeature is an interactive process, allowing the adjustment and tuning of the grey level threshold ranges to achieve the optimum detection of particles in the sample images.

In this case, a single grey level threshold was applied to the backscattered electron (BSE) image to locate the particles and determine their shapes. Figure 1 shows a single field of view BSE image (left half), which is used for the adjustment of the grey level thresholds. The right half of this image shows the result of the grey level thresholding as an overlay and represents the particles detected by AZtecFeature.

Fig. 1. Left - Backscattered electron image of particles. Right - Grey level thresholding, colouring each particle that has been located – note that adjacent particles have clearly been separated.

Fig. 2. (A) Thresholded image of a particle from which its morphological parameters are measured. (B) Morphological values for the particle in Fig. 2A.

Fig. 3. Histogram of the equivalent circular diameter of all powder particles detected.

Fig. 4. Particle image of complete powder sample, colour coded by class: Small Particles – Red, Large Particles – Blue and Tungsten Particles - Green.

Fig. 5. Histogram of shape for Large Particles in powder sample.

During the run the morphological data is measured for each particle from the thresholded image, Fig. 2. Then for each particle, an X-Ray spectrum is collected to determine its chemical makeup.

Stage automation extends the collection of particle data to multiple fields of view, allowing the analysis of whole areas of the sample. For this sample, data were collected from 12 fields of view, covering an area 2.67 mm x 2.63 mm, in which 3283 particles were detected.

In order to build a suitable classification scheme for this sample, the data collected during the automated run was reviewed to establish the overall characteristics of the particles observed. Viewing a histogram of the equivalent circular diameter (ECD) measurements for all the particles, two groups of particle sizes were found; the smaller with a mean of 9 (± 5) μm and the larger with a mean ECD of 70 (± 21) μm. This plot suggests that a classification scheme with two particle sizes Small (ECD <30 μm) and Large (ECD >30 μm), would be effective at separating the two set of particles.

A further class was added to identify the contaminant particles, which contain tungsten. Using this classification scheme, the particles were colour coded, according to their assigned class; Small Particles – Red, Large Particles – Blue and Tungsten Particles – Green.

This powder should be made only of titanium particles. When tungsten particles occur as a contaminant, these can introduce weaknesses in the final product. The identification of the number of tungsten particles provides a valuable tool for the assessment of the quality of the powder.

Another important characteristic of metal powders is the shape of the particles, as this has a direct effect on the way the powder flows and how it can be deposited. The 'Shape' parameter recorded by AZtecFeature measures how irregularly shaped a particle is. By plotting the shape parameter for the set of Large Particles the variation in particle shape can be easily evaluated. The resulting histogram, Fig. 5, shows that the majority of the particles have a value around 1, indicating that they are circular in shape but that there are also a significant number of irregularly shaped particles present. This measurement provides a quantification of what can be qualitatively seen of the Large Particles (coloured Blue) in Fig. 4, which mostly appear to be circular in shape.

Analysing Wear Debris

The operation of machines with engines or gears leads to wear of the moving components and creates wear particles. Analysis of these particles can identify which components are wearing and when the machine may need maintenance. The wear particles can be collected by filtering them out from lubricants and provide a method of evaluating the state of the system without having to dismantling the machine.

Wear particles were mounted on a carbon adhesive tab and analysed directly with the TTM and AZtecFeature. User defined analysis areas can be set up to scan whole samples or a specific area selected by the user.

A consequence of using multiple fields of view to analyse large areas, is that some particles may fall across the boundaries of individual images and so appear as multiple fragments, which are analysed in different fields of view. By utilising overlapping fields of view, AZtecFeature is able to automatically identify and reconstruct these fragmented particles to re-build the original particle. This can become an important part of the processing when the accurate counting of particle numbers is required, as it avoids multiple analyses of the same particle, which would alter the particle count. In this example, the original analysis located 7719 particles. Following reconstruction, the total number of particles was reduced by 105 to 7614.

A classification scheme was built to identify particles based on the major elements contained in them (> 60 Wt%); other classes were created for less specific particle types, based on combinations of elements with lower concentrations. Applying this classification scheme, the particles were colour coded according to their class and identified in the particle image, Fig. 6. Such a view can indicate if there are any specific areas of the analysed sample that could cause concern. For example, too high a density of particles, creating a situation where the image thresholding process is unable to separate out the individual particles. Advanced functionalities are built into AZtecFeature in order to overcome this scenario – see FeaturePhase application notes.

The classification classes can be used to filter the data set, allowing the selection of a specific class of particles that can then be subjected to further detailed analysis. In this example, the Fe-Rich particles are of particular interest – the dataset can be filtered; these particles are displayed in red in the montaged electron image (Fig. 7A).

The morphology of wear particles can be a specific indicator of the mode of wear. As an example, high aspect ratio particles may indicate the cutting or stripping of materials. In order to establish if this is the case for this sample, the aspect ratios of the Fe-rich class of particles were plotted to see if there were significant numbers of particles with high aspect ratios (Fig. 7B). This approach makes it possible to investigate the different mechanisms of wear which may be affecting different parts of a system. In this case the majority of the particle's aspect ratios have a relatively low value, with only a small number having a higher aspect ratio. Monitoring this distribution regularly would indicate any changes in wear occurring, indicating when maintenance should be considered.

The classification scheme used here was built on prior knowledge of the sample. Despite this, 949 particles were left unclassified - coloured green in the particle image - Fig. 6. As these particles may hold significant information on how wear is occurring, it was necessary to classify them. This task is simply done by clicking on the feature in the particle image to view a summary of its information - Fig. 8. In this case, the particle would be attributed as being a silicon fragment. If the identification of this type of particle is significant to the overall analysis, the classification scheme can be edited to add a new class. This would then cause the data to be immediately reclassified to identify any other particles which fall into that class.

Fig. 8. Summary of information for a single particle that was unclassified.

Conclusion

The TTM equipped with AZtecFeature provides a versatile tool for the routine analysis of particle samples. It is capable of automatically locating and collecting morphological and chemical data from particles in order to characterise them. The two particle analyses presented in this application note, illustrate how the range of available data can be employed to solve the analytical problems posed by these samples. Key to this is the flexibility of the classification scheme and the range of data it can utilize as classification criteria. Using these tools, analysts can customise the analysis of particle samples and determine the specific particle characteristics required from such samples.