Teaching PHA

[Ben Buse]

PHA is hard to teach.

Previously I've done a wavescan with open PHA - showing both 1n and higher orders. Then set PHA for 1n - repeat wavescan. Then set PHA for high orders repeat wavescan.

I was playing around with Stephen Reed's virtual wds demonstration version - and it does PHA really nicely - it shows the concepts well because the PHA resolution is better than an instrument - its idealised.

The example of Hf LA1 and Zr KA1 2nd order is show below. You can change the PHA in virtual WDS and see the effect.



I also tried to repeat it on the Cameca - it works quite well but the high background means you can't distinguish Hf LA1 from continuum x-rays. At 30kV - the column wouldn't go any higher.



Although it does mean that for Zr KA1 2nd order you can set the PHA to the 2nd order energy and substantially lower the continuum background increases p/b ratio.

[Probeman]

PHA is hard to teach.

Hi Ben,
I agree!

My position has been to just teach the users to generally accept the default detector (high voltage) bias value and then only adjust the gain to get the PHA peak around 3-5 volts on the JEOL, or 2-3 volts on the Cameca and leave detector in differential mode with the baseline/window values at 0.5/9.5 volts for JEOL and 0.5/4.5 volts for Cameca.  Assuming one is tuning on a high concentration of the element in question.

My reasoning is that if one has a high intensity of the element, the PHA peak can only shift slightly to the right as the intensity drops (e.g., minor/trace elements).

The fact that there might be an escape peak present is OK, as those photons do represent the photons of interest. The idea being that if one can, one should always include the escape peak in the PHA range because that improves one's statistics. Trying to exclude the escape peak will often introduce a non-linear response from the detector as the escape peak is partially removed, due to the shifting of the peak from different count rates.

Also, if one attempts to try and set tight baseline/window values to avoid high order Bragg interferences, again one will often only introduce non-linear responses from the detector, due to the shifting of the peak from different count rates.

I find it better to "let them all in" and then the interference correction sort things out! 

The only time I find I actually need to mess with the baseline/window settings is for an interference of say, Na ka II on O ka. Simply because it's difficult to find an interference standard that contains Na, but no oxygen...

Of course, once we all go to pin diode detectors for our WDS spectrometers things may get easier!

Just me two cents.
john

[Ben Buse]

Yes EDS detectors instead of porportional counters in WDS spectrometers are the way to go - allowing complete removal of 2nd order interferences (Ken Moran, Rich Wuhrer work).

[Ben Buse]

Hf La & Zf Ka II example - are particularly good on the Jeol/PFE - for unlike on the cameca software the counts are not normalized - giving the double peak - good to demonstrate the principle of PHA

[Probeman]

Hi Ben,
What exactly do you mean by the peak labeled "X-ray continuum" in the Zr metal?

I wonder if that peak in the Zr metal isn't also a Hf La peak?  The peak in the Zr metal is shifted slightly to the right compared to the HfO2 scan, but that would make sense since the high count rate in the HfO2 would shift the peak to the left compared to a low count rate.  I suspect that commercially available Zr metal has some Hf in it (except for nuclear grade Zr).  They are difficult to separate chemically.

I could be wrong...   but try measuring Hf in your Zr metal using the Hf Ma line also.
john

[Probeman]

Yes EDS detectors instead of porportional counters in WDS spectrometers are the way to go - allowing complete removal of 2nd order interferences (Ken Moran, Rich Wuhrer work).

Hopefully these solid state detectors in WDS spectrometers won't be affected by gain shifting with different count rates. 

That way we'll be able to use PHA discrimination effectively. In the meantime, the interference correction works well.

[Ben Buse]

Hi John,

So the x-ray continuum is the bremsstrahlung. I found I had a PHA scan for Cu metal at the Hf La peak position - so just background with no peak - again you see the 1st order PHA peak generated by the bremsstrahlung. Its particularly pronounced because I'm using a light element crystal, with a high beam current.



Ben

[Probeman]

Hi Ben,
I'd never noticed this before!   Very cool.

Makes sense now that I think about it because as you said, the continuum isn't really all that "continuous" for a WD spectrometer!      That is because WD only diffracts a very narrow energy range which would show up similar to an "emission" line.

Nice example for teaching PHA.
john

PS If I had nothing else better to do with my time, I would try to model the PHA distribution as a function of composition (and spectrometer position) in the PFE simulation mode...