small Ni-phosphide

[Axel Wittmann]

Dear EPMA Forum Folks,

Trying to beat the physical limitations of EMPA: I am asked to measure the composition of ~500 nm Ni-phosphide crystals that were separated from a meteorite and transferred onto a carbon planchette. The grains are euhedral but show some uneven surfaces. Although the semi-quantitative EDS results look promising, we need more quantitative data to back up structural data for a submission as a new natural mineral.

Any suggestions how to approach this problem with the electron microprobe ? At 12 KeV, the results were not very confidence-inspiring.

Axel Wittmann

[Probeman]

Trying to beat the physical limitations of EMPA: I am asked to measure the composition of ~500 nm Ni-phosphide crystals that were separated from a meteorite and transferred onto a carbon planchette. The grains are euhedral but show some uneven surfaces. Although the semi-quantitative EDS results look promising, we need more quantitative data to back up structural data for a submission as a new natural mineral.
Any suggestions how to approach this problem with the electron microprobe ? At 12 KV, the results were not very confidence-inspiring.

Hi Axel,
This is a tough one, but it should be doable. Start here:

https://probesoftware.com/smf/index.php?topic=281.0

And here for a worked example for a CaF2 particle that shouldn't be any easier from the perspective of the differences in emission line energies of Ni ka and P ka:

https://probesoftware.com/smf/index.php?topic=40.msg1368#msg1368

[Nicholas Ritchie]

You might try simulating the measurement using the Monte Carlo in DTSA-II.  Select the inclusion model and provide your best estimates of the size and composition of the inclusion and the substrate.  The images will give you some idea of whether the beam will remain in the sample.   The spectra will give you an idea of whether secondary emission will complicate matters.  Since the volume is so small, you might lose some signal due to diminished secondary emission. 

All of this for a few minutes of effort.

[Probeman]

You might try simulating the measurement using the Monte Carlo in DTSA-II.  Select the inclusion model and provide your best estimates of the size and composition of the inclusion and the substrate.  The images will give you some idea of whether the beam will remain in the sample.   The spectra will give you an idea of whether secondary emission will complicate matters.  Since the volume is so small, you might lose some signal due to diminished secondary emission. 

All of this for a few minutes of effort.

That is a good idea and very much worth doing.

I should add however, that the Armstrong particle correction method in Probe for EPMA does *not* require that the interaction volume remain completely within the particle. 

Armstrong's method is a slightly related treatment to Pouchou's thin film quantification approach as implemented in STRATAGem/BadgerFilm, in which the beam goes through the volume of interest and into the substrate. The geometric parameters handle the non-infinite interaction volume for quantification. And as with Pouchou's method, it is sometimes worth analyzing the particle using multiple beam energies.

But obviously, one should choose beam conditions in which the interaction volume is at least "mostly" within the particle just to improve sensitivity. And utilize a low Z substrate, e.g., carbon planchette or Si wafer (note: P Ka fluoresces Si, but Si Ka does not fluoresce P).

[Axel Wittmann]

Thank you Probeman & NicholasRitchie !
Really appreciate that you suggest constructive paths forward.
Given the magnitude of the problem, I was almost expecting replies that suggest it can't be done (yet).
Most annoyingly, some of the particles started to move during the analyses.
Best regards,
Axel

[Nicholas Ritchie]

Another alternative is a peak-to-background analysis using EDS.  In fact, I'd strongly suggest using EDS for this analysis regardless.  As Armstrong mentions in his article in the Green Book, if you use WDS, you should either 1) collect all elements on the same detector; or 2) rotate the sample to keep the same face towards the measuring detector.  Either way you are collecting one element at a time.  With EDS, all elements are always collected from the same vantage point and, better yet, excited by the same electrons.

The only reason I'd even consider WDS for a particle was for trace elements.  And even then, I'd only use WDS for the trace element and EDS for everything else.

With P-to-B, you'd need to mount the particles on a TEM grid or other thin-film substrate.   Bulk substrates won't work.  Bruker has a good implementation of P-to-B.