ED-XRF spectra - recurrant artifact below 0 keV?

[zappacake]

Hello,

I can't seem to find a comprehensive explanation for this recurring phenomenon on my XRF spectra graphs. It's a high intensity peak that occurs seemingly below the 0 keV energy mark.

And it seem to be common to many graphs, yet nowhere seems to say what it actually is, aside from a rather vague 'electronic noise peak' or 'pulses', neither of which make sense, as noise is created via the fluorescence of materials in the X-ray tube, in the vicinity of the sample and in the detector, and other interactions in the sample. Noise creates actual transitions peaks as it results from the fluorecsence of actual elements - just not the sample elements!

So what the hell IS that damn 0 keV peak?!

 

[mfb]

How do you measure the photon energy? Which measurement where could be interpreted as negative value?

 

[zappacake]

When photons enter the detector unit of the XRF instrument, they are converted to a series of electrical charges until the photon absorption occurs. The charge are moved to the sides of the detector, in turn creating a deltaV that can be amplified. In this process, detector output voltage moves from background levels, to some maximum peak voltage, before returning to its original level, creating a pulse. A higher energy photon leads to more chargecharges, and creates a larger overall voltage pulse in the detector.
None of this seems to me as if it could be interpreted as a negative value in any way. I've managed to find some amptek manuels that cites series and parallel electronic noise from input capacitance and thermal leakage current, respectively. But it's says this 'noise corner' occupies the lower energy end of the resulting spectrum, rather than the...minus end! But I suppose I should dig a bit deeper with the above, when time allows...

 

[M Quack]

What kind of detector? Silicon drift diode (SDD), Si(Li), Ge or something like this?

The energy calibration is not completely linear. It is quite possible that an integrated charge (peak height) of zero does not correspond to zero keV. Thermal noise would create tiny peaks with almost zero charge, and the non-linear calibration would then give a fake energy below zero. In any case, try to adjust the threshold for minimum peak height, that should solve the problem.

 

[mfb]

Fluctuations in your background could occur in both directions - more electrons or less than average. These fluctuations could be interpreted as small signals, positive and negative. Your software should remove these artefacts.

 

[zappacake]

Thanks M Quack and mfb.

Yes, it's an Si(Li) detector. It does seem plausible that it's something to do with thermal leakage noise from the X-ray tube.

Mfb - what do you mean? If we consider only bremsstrahlung radiation, then surely the resulting spectral continuum curve (I love using the word 'continuum' in a sentence...!) results directly from the non-characteristic photons that reach the detector, rather than on an averaged value. This will be a constant presence that will fluctuate of course, but how would that result in a negative signal? I'm still very much in the process of learning about the instrument, in practical terms, and yeah, further work with the software would help immensely, but I'm afraid my time with the machine is up! I got some very clear spectra from mixed metal samples, but comparing glass samples was very much more difficult. I suspect that this is a direct result of lower atomic number elements, lower energy k-alpha transitions, and thus less intense peaks due to reduced probability of characteristic x-rays being able to escape the surface of the sample. Thanks for your help!
Lee.

 

[mfb]

I think that your events are not related to any photons (of significant energy) at all.
"Electronic noise" is noise in your detector, due to thermal noise or other effects.

Consider this situation:

While you get clean signals from high-energetic photons, your readout has some noise, which gives small positive and negative fluctuations, compared to the average.

 

[M Quack]

I agree with mfb. this is most probably noise from the detector itself. Try to cover the detector with a piece of lead and start anxacquisition. Is the ghost peak still there?

Usually such noise is screened out in the analog electronics, before the pulse height digitizer. That avoids saturating the counting chain.