Combining results collected at multiple excitation voltages

[Warren Straszheim]

I've had occasion to do some analyses on a borosilicate glass. There was interest in measuring boron directly, which led me to use 8 kV for analysis to get better excitation of boron. I also used a long processing time which included conditions optimized for light x-rays (fwiw, PT=6 using Oxford's Aztec system). Yes, I know that this is challenging.

It then became apparent that I would need to use a higher voltage to measure some of the minor elements. In particular, I wanted to measure a low level of Sr. But the Sr-L line is at 1.81 keV and falls on the shoulder of the very strong Si-K line. Thus, I needed to use the Sr-K line at 14.2 keV, for which I used an excitation voltage of  30 kV. So, how can one combine the results of the two analyses? Is there any published work describing that approach for EDS or EPMA?

My approach so far was to measure the beam current to match the standards to my conditions and then to turn off normalization of the results. I measured  major and minor elements from B through Ca at 8 kV, then from Mg through Sr at 30 kV. I noted the B/Si ratio from the 8 kV analysis and used it to calculate the B level from the Si level in the 30 kV analysis. That did much to account for the missing material in the 30 kV analysis. 

It seemed to be the best that I could do under the circumstances. But now I am getting push-back with criticism that that was not a fair way to proceed (i.e., it was unfounded).

Any comments from the collective knowledge base?

Warren Straszheim
Materials Analysis and Research Lab
Iowa State University
515-294-8187

[John Donovan]

I would just say that from a physics perspective there is nothing problematic about using different accelerating voltages for different emission lines in the same interaction volume.

In fact many people do this all the time to optimize our EPMA measurements for sensitivity. In the case of WDS measurements in Probe for EPMA we simply modified the physics code (many years ago) to allow different accelerating voltages for each emission line. That is, don't simply assume that every element (emission line) is at the same accelerating voltage!

But of course you'll surely want to acquire both the unknown and standard for that emission line at the same accelerating voltage.  I assume you're using standards, correct?

PFE also supports different take detector off angles for each emission line which can be useful if the sample is tilted:

https://smf.probesoftware.com/index.php?topic=1579.0

I also see no problem with different accelerating voltages for EDS elements either, but of course that would require multiple spectra per sample.  Really just a housekeeping issue.

I don't know if anyone else supports multiple accelerating voltages like PFE, but originally we did it for multi-voltage thin film analysis.  This allows one to obtain both thickness and composition of thin films at the same time using Pouchou's method.

The physics code is on the Open MicroAnalysis site:

https://github.com/openmicroanalysis/calczaf

and our free CalcZAF app supports multiple accelerating voltages also and can be downloaded here:

https://www.probesoftware.com/resources/

Note the Combined Conditions button shown here which allows different keVs and detector take off angles for each element:



Also:
https://smf.probesoftware.com/index.php?topic=1466.msg11339#msg11339