Fictituous oxygen peaks in graphite

[Les Moore]

Hi guys,

I have noticed a fictituous oxygen peak thrown up on my EDS spectra on a SDD detector - I know it is fictituous as WDS says so.  It is most likely the doubling peak. 

The problem is, it is still quite aparent at 5% DT. I use the axiom that doubling peaks aren't an issue until you get up to 20-30% DT.  If this is the case, is this an issue with the new SDD detectors - i.e. easy dounbling at low energies?
If you run the spectra through a stdless ZAF pkg, it comes up with 10% oxygen.

It is not very dependant on pulse processing setup.

I found the same on oxygen but other elements got in the way so I could not be sure.

Anyone have any experience on this?

Les Moore

[Probeman]

I have noticed a fictituous oxygen peak thrown up on my EDS spectra on a SDD detector - I know it is fictituous as WDS says so.  It is most likely the doubling peak. 

The problem is, it is still quite aparent at 5% DT. I use the axiom that doubling peaks aren't an issue until you get up to 20-30% DT.  If this is the case, is this an issue with the new SDD detectors - i.e. easy dounbling at low energies?
If you run the spectra through a stdless ZAF pkg, it comes up with 10% oxygen.

It is not very dependant on pulse processing setup.

I found the same on oxygen but other elements got in the way so I could not be sure.

Hi Les,
I have a difficult time seeing how this could be a sum peak because in steel what could be present at high concentrations enough to cause sum peaks (Fe La, but that's 0.7 keV, not 0.51 keV).  And carbon is 0.277 x 2= 0.55 keV so not that close though EDS might be fooled...

Could it be a surface oxidation peak?  What keV are you at?

Could it be a V La peak?  Even with WDS, the V La is almost on top of the O Ka...

[Les Moore]

Hi Probeman,

Not steel in this case.

This work was done on the graphite I was looking at in my other post.  The middle of some of the (micro) char particles have no significant O by wds i.e. no discernable peak above background yet the EDS showed heaps.  Our other machine (ASPEX) SDD EDS  also showed the same effect but this time on a analytical graphite crucible.

The other mystery is if it is there, why was it's ratio to the C Ka a function of total dead time rate. 

I suppose I could look at any doubling of the Fe La on steel to see if this has similar issues.

Just a bit of an oddity that worries me - oxygen in a "second phase particle in steel" analysis means a remnant deoxidation or reoxidation product, carbon means exactly the opposite.   

[Newbury]

Low photon energy peaks are especially prone to coincidence (and have always been, including for Si(Li)-EDS performance), because the charge pulse of a low photon energy x-ray is just above the noise floor and inspection/discrimination functions are ineffective.

There is a very easy way to determine if the suspect peak is due to coincidence: just progressively lower the beam current (e.g., successive factors of 2) and compare the spectra (scaled to the C peak in this case).  This experiment is illustrated in the attached pdf for the Bruker QUAD SDD that we have on our probe (feel free to share this pdf), as well as some nice examples of trace level SDD-EDS work in the presence of coincidence and severe overlap.

Dale Newbury
NIST

[Les Moore]

Hi Dale,

I initially did as you suggested and this is why I posted the query. 

My results: