spectrometer background - a power law?

[Ben Buse]

Hi,

Using dataset of MAN curves I can calculate the spectrometer background for a given Z, at various spectrometer positions. (Thereby allowing calculation of element error and detection limit)

I have limited data but it appears to be a power law! And I've just checked this against a complete wavescan. 1st order approx to power law (obviously will also have absorption edges etc.)

Presumably where intensity is a function of source to crystal distance (i.e. L-value). I better get reading!



here's wavescan

[Probeman]

Hi Ben,
Are you saying that you are getting similar curves for wavescans compared to MAN intensities plotted as a function of spectrometer position (for a given average Z)?

If so, that would be as expected because the MAN calibration curves (bgd intensity vs. average atomic number or Z-bar), are simply attempting to predict the continuum intensity at a given spectrometer position (for a range of sample average Z).

And yes, background intensity as a function of spectrometer position is exponential.  Or perhaps I am missing something you are saying?  Can you explain in more detail?

[Ben Buse]

No your right, nothing ground breaking, I'd just never realised the spectrometer background fits a power curve. It also allows interpolation of MAN data for elements on which it has not been collected - to estimate det lim and errors offline.

Ben

[Probeman]

Ah, OK.    Of course one can also just use PFE to calculate MAN detection limits!   

But if you do "roll your own" MAN detection limit method, I would be very interested to see what you come up with, because of course normal (off-peak) sensitivity estimate methods don't apply for MAN backgrounds (the MAN method has significantly better trace sensitivity).

You might want to read our discussion on MAN sensitivity calculations in the 2016 Amer. Min. paper.  See the discussion starting on page 1851 in the section titled "Calculation of detection limits with the MAN background method".

http://epmalab.uoregon.edu/publ/A%20new%20EPMA%20method%20for%20fast%20trace%20element%20analysis%20in%20simple%20matrices.pdf

I'd be very interested in what you think.
john