F Ka ROI artifact

[Probeman]

Hi Ben,
This is interesting.  I ran some scans on my old SX50 on TiO2 and fluor-phlogopite and here is the scan on TiO2 on PC1 (I don't have a PC0 crystal):



Looks very much like yours, no visible interference. Now here is a scan using the TAP crystal on the fluor-phlogopite:



The peak shape is a bit funky probably because the TAP crystal is 30 years old!  But it's at the peak position.   Note that this is a pretty high precision scan because although I only ran at 15 keV, 20 nA, the count time per point was 80 secs.

Now here is another scan on the TAP crystal but this time on the TiO2 std:



Whoa, so there is definitely an interference there and it doesn't look like a fluorine peak.  I can't see what element could be however (Mo?).  I'm going to run an EDS scan for a while and see if anything pops up that shouldn't be there...

[Probeman]

Here is an EDS scan on my synthetic TiO2.  Not much besides the Ti Ka/Kb (and sum peaks). A little Si and Al.

That tiny peak visible in the WDS is barely visible in the EDS spectrum as a change in slope where fluorine is. I wonder what it could be?  I guess that just demonstrates how sensitive WDS is. And how sensitive MAN is to the background.

See attached (login to see attachments).
john

[BenjaminWade]

Hi John
Cool! Very interesting...
I am running some longer scans on my PC0 and TAP on my TiO2 again to see if I can replicate what you see. It must be there I guess, just that my scans were not long enough. Very quick google on manufacturing of TiO2 xtals, sounds like there can be a number of impurities depending of course on the starting material, but also on the type of manufacturing process. Just to list a few...
(Fe,Nb,Cr,Ta,V from starting materials, and Cl and/or S from some of the acidic manufacturing processes).

Looking at the positions of these relative to the F peak though, really only Ta matches up (sort of) peak position wise. Its a really obscure II order Ta intf though, and I would be surprised if there was enough Ta in the TiO2 to produce that intf...the test would be to do a wavescan over the Ta La (on LIF) position in the rutile (There is a III order intf of Ti on Ta Ma on TAP).

Will post my wavescans when its done.

Cheers

[Malcolm Roberts]

Hi John,
As requested. Here are two plots. The first includes std 140 Rutile. The second includes Mn metal and spessartine. I have noticed that Mn is the most significant thing to avoid when calibrating the MAN fit for F on our LDE1 crystal.
Cheers,
Malc.

[BenjaminWade]

Hi Malcolm
Thanks for that. Interesting! Your rutile std (140) also plots slightly above the fit. Would love to see the full list of standards you use in those plots as well. For a second I was about to scroll up/down in that standard list on the image...

Cheers

[Malcolm Roberts]

Hi Ben
That very much depends on the job. These plots come from two different packages. One for apatite and one for phylosilicates. The apatite run does not make use of the rutile standard for obvious reasons - we don't analyse Ti. However, there is a certain uniformity in what I get up to!!
Apatite run = 12 off = anhydrite, wilb ap, fluorite, magnetite, periclase, pyrope, tugtupite, wollastonite, dur ap, jadeite, mn metal, spessartine
bt, mus run = 21 off = wilb ap, barite, clinochlore, cr2O3, fluorite, magnetite, periclase, rutile, tugtupite, V metal, wollastonite, dur ap, corundum, jadeite, Mn metal, orthoclase, SC Ol, Kak pyrope, spessartine, biotite, muscovite.

Note.... I insert a bunch of things as refs as you might notice.
Cheers,
Malc.

[BenjaminWade]

Great, thanks for that Malcolm. Definitely some others in there I am going to try now and see how they go.

John, my scans on my TiO2 at F Ka peak position are attached for my LTAP and one of my other TAP xtals. I am impatient so only did 40sec count times, but the ?intf? peak is definitely there. Looking at it properly, it actually more lines up with Nb than Ta. But this is a higher order Nb IV interference. The element itself would make sense in a natural rutile given there is a niobian rutile end member. Perhaps the starting material for synthetic TiO2 isn't very pure.

The scan on my PC0 isn't finished yet.

Cheers

[Malcolm Roberts]

And the other thing I would like to raise is just how handy the MAN approach can be to dealing with interferences and such things. You will have seen Sobolev's work on trace elements in olivine. He runs up an elaborate technique to deal with Fe interference on Co, looking at counts at the Co peak position using Fe-only bearing standards and modifying the Co on peak counts off line. This interference cannot be dealt with simply by the usual correction procedure, but what Sobolev is doing (and may not know it) is simply a MAN background correction. if one sets up the MAN correction for Co in olivine using Fe only standards (Fe metal, pyrite, magnetite etc. etc) you can do precisely the same job at the press of a button. I'm not going to go into the utility of John's approach for traces....... It's pretty clear I think. The absolute flexibility of this approach to background correction is astounding if set up correctly and with sufficient forethought.

[Malcolm Roberts]

Lads.....
Just working on this new run for apatite. it's a more comprehensive thing involving some REE as well. Three plots: initial unplayed with, then including rutile (Ben ...interesting eh?), finally the working plot.

John... you might recall way back in the dark mists of time I suggested that a good addition might be to be able to specify  Zed range rather than having to deselect standards from the list? Quite a lot of standards had to be deassigned to get me to the zed range I was interested in. Any chance something like this might be implemented??

[Probeman]

Hi John,
As requested. Here are two plots. The first includes std 140 Rutile. The second includes Mn metal and spessartine. I have noticed that Mn is the most significant thing to avoid when calibrating the MAN fit for F on our LDE1 crystal.
Cheers,
Malc.

Hi Malc,
Cool. So that is consistent with Ben's MAN plots in that there doesn't seem to be a problem with the LDE/PC crystals, just possibly the TAP crystals.

I suspect this is because the interfering line we saw in my TAP scan on Tio2 is a higher order reflection which are suppressed with the multi-layer Bragg crystals.  I mean what element could it be if it's a first order reflection?  I guess it could be fluorine except that the peak shift is quite large.

Do you have a TAP MAN curve for F ka with TiO2 (or Ti metal) that you can share?

[Malcolm Roberts]

Hi John
It would take an awful lot of digging around to find this. I abandoned TAP for F fairly early on as I was getting much better results from LDE1, e.g., higher on peak count rates and better LLD........
I'll see what I can find..
Watch this space.....
Cheers,
Malc.

[Probeman]

Great, thanks for that Malcolm. Definitely some others in there I am going to try now and see how they go.

John, my scans on my TiO2 at F Ka peak position are attached for my LTAP and one of my other TAP xtals. I am impatient so only did 40sec count times, but the ?intf? peak is definitely there. Looking at it properly, it actually more lines up with Nb than Ta. But this is a higher order Nb IV interference. The element itself would make sense in a natural rutile given there is a niobian rutile end member. Perhaps the starting material for synthetic TiO2 isn't very pure.

The scan on my PC0 isn't finished yet.

Cheers

Hi Ben,
Very cool that your natural TiO2 shows the same peak as my synthetic!

I guess there could be a trace Nb in my (our) TiO2.  The EDS scan shows two tiny, tiny bumps at 2.1 and 2.2 keV, but they are very much in the noise.

We need to do a high precision wavescan of TiO2 at the Nb La region.
john

[Probeman]

And the other thing I would like to raise is just how handy the MAN approach can be to dealing with interferences and such things. You will have seen Sobolev's work on trace elements in olivine. He runs up an elaborate technique to deal with Fe interference on Co, looking at counts at the Co peak position using Fe-only bearing standards and modifying the Co on peak counts off line. This interference cannot be dealt with simply by the usual correction procedure, but what Sobolev is doing (and may not know it) is simply a MAN background correction. if one sets up the MAN correction for Co in olivine using Fe only standards (Fe metal, pyrite, magnetite etc. etc) you can do precisely the same job at the press of a button. I'm not going to go into the utility of John's approach for traces....... It's pretty clear I think. The absolute flexibility of this approach to background correction is astounding if set up correctly and with sufficient forethought.

Malcolm,
I think you have put your finger on it exactly.

That, and I would also note, this again reinforces Paul Carpenter's admonition that by using the MAN method, we learn things about our standards that we might never have suspected...
john

[Malcolm Roberts]

Hi John
As requested here are some plots for F on TAP. The first has everything. Note big problem with Cr. The second shows remaining bother from Mn and V. Note 828 is Kak hbl which has no F recorded in the database! I would remove it in real life of course!!

[BenjaminWade]

Yes Kakanui Hbd definitely has fluorine in it. With Fe and Mg intf on I still return anywhere from 500-1000 ppm in mine.

Latest on the mystery peak, definitely not Nb. Detailed wavescan at Nb La peak position on the rutile returns nothing, and given the size of the "intf" peak at F Ka peak position, if it was Nb it should be quite a reasonable peak.

The detailed wavescan at F peak position on rutile on PC0 also finished and is attached, but its confusing me. I don't see any interfering peak there at all like we do in the TAP scans, yet in my MAN fits for PC0 my rutile definitely sits above the fitted line, and those MAN fits were constructed using only peak counting times of 12 seconds, so I should definitely see it in the wavescan where it was 40 seconds per point? I am now thoroughly confused...

Although my Titanium metal is probably oxidised, I am going to run detailed scans on that as well to see if I see any peak on PC0/TAP. Given there isn't any problem with oxygen apparent from the rutile scans I am assuming my oxidised Ti metal will be OK...