Strategies for Improving Accuracy Using Trace Elements Standards

[Mike Jercinovic]

Well, hopefully a wt% Or 2 wt.% U is something humans  could measure, but its more about reconciling actininide counts with all those pesky transitions...

[Probeman]

Well, hopefully a wt% Or 2 wt.% U is something humans  could measure, but its more about reconciling actininide counts with all those pesky transitions...

Well there it is.  We've come full circle.   
 
Yes, well you might be forgiven to think it's something that humans could measure, but as you know it gets pretty difficult when there is inhomogeneity present at the micro scale.  And my understanding from Hanchar was that doping U or Pb in synthetic monazite always produced zoned crystals.  We see the same darn thing when doping Ti in quartz, and I suspect other trace elements in other synthetic crystals.

Hence the rationale for my plea/rant for matrix matched zero blank standards...   at least we know a priori exactly what the concentration is, when the element is not present!   

[Mike Jercinovic]

Sorry, the point here is not really the measurement of uranium in this case, but the influence of the presence of uranium on thorium in particular. That is, measuring zero thorium in a blank with lots of uranium present.. just like measuring zero uranium with a hi Th blank. We do see the occasional monazite with much more uranium than thorium, so just trying to get the acitinides right.

[Probeman]

Sorry, the point here is not really the measurement of uranium in this case, but the influence of the presence of uranium on thorium in particular. That is, measuring zero thorium in a blank with lots of uranium present.. just like measuring zero uranium with a hi Th blank. We do see the occasional monazite with much more uranium than thorium, so just trying to get the acitinides right.

Ah, I misunderstood. OK that makes sense.

So as you say, for monazite accuracy one really needs *two* blank standards.  One with REEs and Th as a zero blank for the U and Pb measurement. And a second blank standard standard with lots of REEs and U, but with zero Th present.

It's a difficult material. I'm glad you're the one who is working all this out.      But seriously I'd like a piece of the U and Pb blank synthetic monazite just to test my own efforts.  If you have some available.   

[Probeman]

Recently I've been developing a method for measuring a slew of trace elements in ilmenites and sulfides and I need some help with what appear to be spectral artifacts in my measurements. The customer wanted the following trace elements measured:

pt ma   pd la   ga ka   ag la   au la    v ka   lu la    f ka   hf la   cr ka   as ka   nb la   ta ma   zr la

Which is a just fine, but we also need to measure the major elements to deal with possible interferences from Fe, Ti, S, Cu and while we're at it probably Si, Ca, Al, etc. as well.  So I decided to measure the remaining elements using our Thermo Pathfinder EDS system, because measuring 14 trace elements on 5 spectrometers is time consuming enough as it is!  So we measured a full EDS spectrum in PFE along with the WDS trace elements. This allows us to correct for interferences on the WDS trace elements using the (major/minor) element concentrations determined from our EDS spectra.

All fine and good. Though a quick reminder, when one has this many elements by EDS and lots of points, it's really important to note this feature for utilizing "stored" EDS net intensities, in order to speed the quant calculations up considerably:

https://probesoftware.com/smf/index.php?topic=40.msg8637#msg8637

Anywho, during the method development I acquired some high precision wavescans on of their ilmentites and noticed this odd "peak" in the Hf La region on an LiF crystal:



You can see (most) of the MPB (multi point background) measurement positions that I decided to utilize for this effort. But the thing is this peak is *way* too wide to be a normal spectral emission peak and I have no idea what it is, but it creates an "interference" of around 0.3 wt% Hf in pure iron metal.

Then last night I extended the range a bit more to see the remaining MPB position and increased the acquisition time for better precision and now we see this:



So what the heck is this "peak"? Has any body seen this on their instruments/crystals?

[Probeman]

I'm beginning to think I need a new microprobe!

Here's a similar looking artifact, but this time on a PET crystal and sitting right on top of the Ag La emission line position!  It's hard to see in this scan but look closely:



Again it seems to be associated with the presence of iron, because here is a higher precision scan on our pure Fe std:



Has anyone seen anything like this before?

[Anette von der Handt]

It looks similar to what we in JEOL land (Karsten G., Dave A. and I) called "the hump" and was caused by BSE electrons making it into the detector despite the electron trap. It showed at the same position on all crystals and scaled with BSE yield of whatever you ran on. Not sure if its the same as I lost my ability to read Cameca positions but run a scan on something like gold and see if it gets worse?

[Probeman]

Interesting.

It seemed to me to scale with Fe concentration (iron was the highest Z element I was measuring), but if you JEOL types are also seeing this "hump" at the Fe Ka position, that would make some sense, as that's about where it is. Treating it as an Fe interference seems to deal with it pretty well actually.

And no need to rely on your memory of Cameca spectrometer positions (or JEOL spectrometer positions if you're a Cameca type), as you can use this cute window in CalcZAF to calculate them:

https://probesoftware.com/smf/index.php?topic=598.msg4347#msg4347

Anyway, I'm running a scan tonight on a gold standard, so I'll let you know.

[Probeman]

It looks similar to what we in JEOL land (Karsten G., Dave A. and I) called "the hump" and was caused by BSE electrons making it into the detector despite the electron trap. It showed at the same position on all crystals and scaled with BSE yield of whatever you ran on. Not sure if its the same as I lost my ability to read Cameca positions but run a scan on something like gold and see if it gets worse?

Hi Anette (and other JEOL types!),
Can you show us an example of this "hump"?  I totally believe it but on my Cameca instrument this "hump" seems to be related to Fe content as shown in my previous scans.

Here's a scan on my gold standard as you suggested:



Same conditions, but nada on the "hump".  Compare this to the 2nd scan in this above post:

https://probesoftware.com/smf/index.php?topic=928.msg8730#msg8730

[Probeman]

Here is a scan from last night on my pyrite std:



This "hump" artifact on the Cameca definitely seems correlated to Fe concentration (compare with above Ag La scans on ilmenite and Fe metal). The good news is that it's only a few hundred PPM in magnitude and I can utilize the interference correction to remove this artifact (even though I don't quite see how it could be related to Fe emissions as even plotting up 20th order reflections shows nothing with KLM markers).  Possibly a secondary Bragg diffraction from the PET lattice?

I assume the "hump" that you and Dave and Karsten are seeing on your JEOL instruments (because you said it is unrelated to composition, only Z), can be dealt with using the blank correction?  What is its approximate magnitude in PPM?  Do you have a scan handy you can show us?

I suppose we (both JEOL and Cameca) people should be glad that this WDS spectral artifact is only a few hundred PPM in magnitude (the "hump" near Ag La is around 260 PPM in the Fe std without an interference correction).

Though a possibly related spectral artifact on LiF near the Hf La peak position (as posted above) is a bit larger, around 2800 PPM in pure Fe.  Here it is from the pyrite scan from last night:



You can imagine my customer's excitement when he initially thought there might be several thousand PPM of Hf in there ilmenites, only to have it disappear after I also noticed Hf in my Fe std and I applied an interference correction.

Well at least the vanadium signal is still strong in their ilmenite samples, even after applying the Ti -> V interference correction!   

[BenjaminWade]

Just in case you thought you were going crazy, I have the cursed hump too...

[Probeman]

Just in case you thought you were going crazy, I have the cursed hump too...

Hi Ben,
Thanks. That is good to know. And your SXFive instrument is quite a bit newer than my SX100 (so maybe it's not a Bragg crystal degradation, but more a secondary diffraction artifact).  Does the "hump" seem to be related to Fe content? Maybe also scan a gold or other high Z std and see if the "hump" disappears.  You should also scan the region around Hf La on an LiF crystal too, on an Fe std. 

By the way, please move your in-line images to the user forum gallery:

https://probesoftware.com/smf/index.php?topic=2.msg4040#msg4040

Using external image hosting sites means that the links to your images will eventually break.  Some users that were using TinyPics to host their images lost all their images when TinyPic shutdown recently.

Just use the Gallery feature from your user forum login and your image links will be permanent.

[Probeman]

It looks similar to what we in JEOL land (Karsten G., Dave A. and I) called "the hump" and was caused by BSE electrons making it into the detector despite the electron trap. It showed at the same position on all crystals and scaled with BSE yield of whatever you ran on. Not sure if its the same as I lost my ability to read Cameca positions but run a scan on something like gold and see if it gets worse?

So Ben Hanson (Corning) reported to me that he sees something on his JEOL instrument that might be related. He says:

I saw something similar while doing boron.  I couldn’t figure it out for the life of me until I did some SCA scans.  It seems the low-energy noise tail moved across my baseline at the “peak” position, but not at the positions just away from the “peak”.  I couldn’t fathom this and thought that I must be dreaming.  WTF could affect the low end noise by simply moving the spectrometer?  I don’t know.  May be worth a check?  I know that I’m insane but I did check this a bunch of times.

Here is a slide he sent me:



I'm wondering if this is related to the stray electrons Anette mentions.  He suggests running a PHA scan at different wavelengths.

[Karsten Goemann]

So on the JEOL probes the "hump" (a wide, asymmetrical peak) was caused apparently by some BSE striking the detector (bypassing the monochromator crystal). Only visible on the P10 counters. Assumedly the thick Xe counter entrance window absorbs these. Presumably they were high energy BSEs as they made it through the P10 counter thin window and the intensity could be reduced with differential PHA, whereas changing the baseline had no effect. It got worse the higher Z the material (e.g. gold) and the higher the accelerating voltage.

JEOL re-designed shields in the spectrometer so we don't see this hump anymore. On our machine we had these at around 155 mm L value in the vicinity of Rh La on PET. You can use your tool to convert to Cameca units :-)

Here is an old wavescan at 30 kV showing this hump (and a few other things like some small characteristic peaks and the Ar K edge):


But on the Cameca the WDS are separated from the chamber with only the small "separation window" for the x-rays to pass through, so I don't see how your artefact can be the same thing. I assume your scans are from a high pressure P10 counter, so it's not very likely any BSEs make it through that thick separation window? I don't think it is the same thing as it appears related to the sample material which ours isn't.

[Probeman]

Hi Karsten,
Thanks for this data.  That is good news that JEOL was able to modify the WDS detector to remove this spectral artifact.

155 mm on a JEOL instrument (according to CalcZAF Calculate Spectrometer Position) is closer to Ru La:

Spectro position for ru la on PET (140 mm), is 155.0865 (with refractive index correction, k= 0.000144)

On a Cameca instrument that corresponds to spectrometer position of 55000, so that can't be related to the "hump" Ben Wade and I see at 47500 or so. And as you point out yours seems to be strictly Z (BSE) dependent.

Would you be willing to run a high sensitivity scan on your PET on the JEOL around the region of Ag La? That would be ~133 mm on your JEOL instrument? Please do the scan on an Fe metal or other Fe bearing standard.