Accurate AutoID and tools for reliable identification of elements in a sample

We at Oxford Instruments were interested in the recent publication of an article by Dale Newbury of NIST in Microscopy and Microanalysis entitled 'Misidentification of Major Constituents by Automatic Qualitative Energy Dispersive X-ray Microanalysis: A Problem That Threatens the Credibility of the Analytical Community' (published Dec 2005, Volume 11 545-561).

For those of you who have seen this article we would like to reassure you about the accuracy of your INCAEnergy system for element identification. We would also like to remind all our users about two simple tools you can use to make sure you identify the elements in your spectra correctly.

For those unfamiliar with this article, it tests the results of four commercially available EDS software platforms for the reliability of their auto peak identification software. This paper concluded that software platforms were unable to reliably identify even major elements in some samples, and would result in inexperienced users making false assumptions about the elements in their samples. Therefore the conclusion in the article is that this 'threatens the credibility of the analytical community'.

We are confident that our INCAEnergy system was not amongst the systems tested as we have a large development team, who have extensively tested the product, including the materials used in these tests.

The INCAEnergy software uses an autoID algorithm which has been developed over 15 years to include not only peak position but also peak shape to ensure correct AutoID, even where peaks overlap. This algorithm is much more reliable and sophisticated than the algorithms Dr. Newbury criticizes in his paper. Therefore the good news is your INCA system will provide you with reliable automatic peak ID. It looks like some EDS users may not be so lucky!

INCAEnergy provides two tools you can use to check the accuracy of element identification.
These tools are found on the 'Confirm Elements' screen in the Point&ID or Analyzer navigators.
They are:

1. Check Total

2. Spectrum Reconstruction Overlay

To enable these tools, simply check the 'Overlay Spectrum Reconstruction' box, or select Reconstruction in the 'Overlay' drop down, situated in the top left corner of the 'Confirm Elements' screen.

Check Total gives a number in %. If the number is less than 80% or greater than 120% then there is a possible problem with the spectrum collected or the elements identified. This could be:

1. A major peak in the spectrum is incorrectly identified
2. The sample is charging
3. The area being analysed is not homogeneous
4. The kV saved with the spectrum is not correct

Therefore if the check total is outside the optimum range (80%-120%), check the accuracy of the element identification.

This is a simple task using the Spectrum Overlay tool.

Spectrum overlay is a red line spectrum, which is overlaid onto your spectrum. It shows what the spectrum should look like based on the elements identified. If these elements are not correct then there will be a bad match between the spectrum and the red overlay.


One Example - KBr (Potassium Bromide)

KBr was one sample for which the EDS systems tested by Dr. Newbury had difficulty identifying the elements correctly. This was because the autoID routines tested could not always differentiate between Br and Al. The autoID function in INCAEnergy correctly identifies K and Br, however, for the sake of this discussion, assume that the peaks were misidentified manually by the operator.

With INCAEnergy it is easy to see that Br has been misidentified as Al when analysing KBr. As shown in the image below:

1. The check total is less than 80% (only 37%), therefore you should check the overlay
2. The overlay on the peak labelled Al does not fit well, particularly the small peak to the left of the peak labelled Al is not shown by the overlay at all.

If there is a major error in the element identification of a spectrum, using the check total and spectrum overlay tools will quickly show the problem.

A correctly identified KBr spectrum is shown in the image below, as determined using the INCAEnergy AutoID software.



Conclusion

The conclusion we can take from Dr. Newbury's paper is that:

• Not all EDS systems are capable of performing even the basic demands of the EDS user i.e. identifying elements correctly.
• However, as INCAEnergy users you can be confident that the results obtained from your INCAEnergy system are accurate and of the highest quality.