Dear Charles,

you may share your fluorescence spectrum

(please from the lowest photon energy you can measure at all).

What target material does the x-ray tube have?

 Thanks in advance.

My dear,

I  just had a look at the specifications of your XRF system. Your x-ray tube is an Rh based one ...........and you wonder that you see a lot of Rh when you do not use a filter?

Compton scatter of 2,7keV (at around 180°) ends up at 2,67keV. So it is 30 eV below your net 2,7keV energy. Your energy resolution of the detector  is around 140eV, so you cannot destinguish between 2,67 and 2,7 keV.; that's it.

For the case of Rh k-alpha (~20keV ) you will end up around 1,5 keV smaller  Compton scattered photon energy. Therefore you are able to separate the tube's Rh k-alpha from the sample's Rh k-alpha via a filter.  

So your 2,7keV Rh signal arises mainly directly from your tube itself.

As I mentioned, we know our normal Rh level, because we already know the specs for our device. We have a background reading for the device, and this is consistent with the great majority of our scans.

This is something different completely. It only shows up in certain rocks, specifically with high gypsum and dolomite content. I'll post a spectrum as soon as I can. Perhaps a few from two different sources, so you can compare previous issues with our current issue concerning dolomite-gypsum combinations.

What we're looking for is a reason for Rh to be enhanced, not the reason why Rh is present. Nearly 100% of our samples previously scanned have a similar Rh signature slightly lower than 2.7keV. We're so confident with our Rh background level that we can quantify Cl from our Rh peak on a regular basis. Except in high gypsum and high dolomite materials. But, not all dolomite, and not all gypsum.

That's why this is confusing. I'm thinking along the lines of enhancement, because Ca is elevated when both gypsum and dolomite is present in the same sample, but we don't know the math behind why. Our spectra software is automatically removing the Ca escape peak, which isn't in that position anyway.

We want to make sure that there IS math behind this enhancement, and classify it specifically, OR determine if another element is present which our device cannot detect the Ka peak ranges, but that we can see L or M peaks for that element. We've examined the range of elements around At which have non-alpha peaks around this region, and we're not convinced that those elements could be possible either.

Another alternative we've been wondering about is the combination of the Ca escape peak with other pile-up peaks. Is this possible, and if so, could it cause a peak at 2.7 when combined with our background Rh peak? What would be the probability of that occurring? Our device is not the best device for quantifying Mg, so we're also not sure if we could extract Mg from pileup peaks if this were the case.

Dear Charles,

what are the type of samples which do not show the 2,7keV peak. What is the approximate main chemical composition/ density?

Compton scatter /backscatter is proportional to the density. 

And what is about moisture or the surface of your sample. Low elements are good Compton scatterers: H2O ,-SO4,  -CO3 (Gypsum, Dolomite). Do your previous sample do also contain these molecules?

Pile up: due to the very low photon energy I would exclude that the  2,7keV peak is a result of pile up. Each component in such a case has to have a very low energy, so passing the air with high propability  is very unlikely for each component.

Just a test: please perform an XRF scan form the surface of a pan filled with water ( there is enough H2O) and look if  the 2,7keV peak is stiil remaining. (But with your filters removed, such as in the case of the above dicussed experiments).

The samples with the lower and constant peak height are primarily Calcite, with little to no dolomite and gypsum. They also have widely varying iron (as oxide or sulfide), silica, and strontium (as celestite), as well as chloride. These other samples don't seem to have any effect on the peak heights in that area. With helium flushing, we are reducing our argon counts to 0 or 1 per 60 seconds, which gives much clearer readings at eV lower than argon.

As for density, the background level of the entire spectrum is rarely higher or lower. We judge our matrix density on the position of that background level. As we encounter more porous material, this background level goes up (to about 500% standard). With more crystalline material, it goes down (to about 10% standard). The exception so far being when porous material appears solid because it is filled with hydrocarbons. The math we've read about so far says this level is an effect of the average atomic number, which makes sense when we encounter high concentrations of hydrocarbons, and also air-filled matrix.

In the case of these high Rh peak samples, the background level is not substantially going up or down from our standard set. The Rh peak is also not directly tied to the fluctuation in background level.

As for moisture, these samples are dehydrated. They're kept in air-sealed bags until being scanned. They're scanned in a cool dry lab. But gypsum does contain water.

One of the reasons I suspected pileup is that this device is not great at detecting Mg, but this sample set contains very high magnesium dolomite. Because our device is not made to specifically target this range, I wondered if the device may not be able to differentiate and remove excessive pileup from magnesium. These are the highest magnesium dolomites we've ever scanned.

In previous samples, when scanning gypsum nodules directly, we do see elevated background levels. And I suspect this is due to scanning the water included in the gypsum. This background change does elevate our Rh peak near 2.7 in that case, but it also elevates everything else. That is not the case with this sample set. Only Rh is elevated, and it's elevated far more than the background level. In some scans, it appears that Rh is made a major component, but we know this is not the case, because additional Rh is not in the sample.

It is very confusing. I'll be at the lab later and can compile an image of our usual spectral range, as well as a set of spectra for our standard range, and the spectra for this set.

Thank you for your help so far.

Dear Charles,

just a simple trick for identifying pile-up peaks in XRF.

a) for a non pile-up peak the peak height is proportional to the intensity of the exciting radiation; thus in case of an x-ray tube the peak height is proportional to (or linear with) the mA setting of the tube. Thus doubling the mA will double the peak height; or when mA is lowered to 1/2 the peak will go down to 1/2 of its former value.

b) for a pile-up peak based on coincident detection of two photons  the peak height is proportional to the intensity of both of the two contributors; i.e. the peak height is proportional to the square of the mA setting. Thus doubling the mA will increase the peak height by a factor of 4 (but pay attention on dead time effects); or when going to 1/2 of mA the peak will go down to 1/4.

c) for a pile-up peak based on triple coincidence of photons the peak height is proportional to cubic (3rd power) mA setting;  thus doubling the mA will increase the peak height by a factor of 8; or when going to 1/2 of the mA setting the peak height will go down to 1/8 of its primary value.

So you are easily able to identify whether your 2,7 keV peak is a pile-up based XRF peak and at the same time you can easily identify the pile-up order via the peak height dependence on the mA setting of the tube.

 So go ahead and good luck. 

I will try that tonight. Thanks for the trick. I had not thought of that, but it seems obvious now.

That being said, I have serious questions about our previous chemical analysis of the samples. I found an escape peak for Rh at 0.95 in our highest Rh valued sample. At 13x Rh intensity, the escape peak becomes quite clear.

Am I correct in thinking that pileup peaks which get past the hardware cannot then make an escape peak? It seems like the probability of that would be extremely low, if it can actually happen.

You are right, the propability for escape peaks arising  from pile up peaks is very, very low. One of the two coinciding photons has to  undergo an escape process; altogether very low propability for that.

I've added an assortment of files and images to the top question. Given the first two images and the attached spectra file, it certainly looks like we have a rock set with relatively high Rh. I did however find that some of our other rock sets do jump back and forth between Rh and Cl, but this is probably only because as Cl rises, it adds to Rh until it overtakes Rh and becomes the primary peak in that region. I think though that some of the rise might be attributed to actual small quantity Rh in the sample. I cannot be sure from the second and third images.

Dear Charles,

from your displayed spectra it is clear to me that the peak at 2,7keV is due to Rh. The 2,8keV one of Rh  is also present. So the 2,7keV one is no pile up peak.

The question is: where is the Rh?

If you do not have got an equivalent Rh k-alpha peak from your sample the low energy Rh signal(s) will arise from the primary tube spectrum as I mentioned above.

Filtration will kill these peaks but also most of the excitating radiation for the detection of Cl.

You should go to another tube ( e.g. Cu based). Cu has got no lines in that energy region. But I do not know whether you XRF system can be provided with an Cu x-ray tube.

Have you already tested your system by scanning water (demi -water, definitely without Rh)?

Additional remark: I cannot open/read  your PDZ file.

We test our system with water DI water, an alloy disc provided with the device, and a standard rock of our choosing. To know we're in the right settings and that the gun is functioning properly, scans must register the same with these spectra daily. Which is why I was amazed that so much Rh was showing up in the spectra for these rocks. It's supposed to be extremely rare, and it didn't show up on our other analyses. Like you said, it should also register differently at the Ka1 peak, but it does not.

Scanning di water shows us the background parts of the gun, including the stainless steel casing inside. At Bruker's last inspection, we were told we had one of the cleanest devices inside given our level of use. We really do pamper it with a nice cool, dry, and fairly dust-free room.

I was really hoping to determine if this was some kind of mistake, rather than having to find more funding for alternative devices or external examination.

I've also been reading about the prolene window sticker we put over the end of the device for protection. We've been told by Bruker that it has to come off to perform helium flushing scans to allow the helium to actually flow. Without this removed, or without helium flushing, we know that argon amplifies many lines and confuses others. From what I understand now, the prolene window can somewhat normalize Rh values from the gun based on the thickness of the window. I'm going to try a few more tests with the window today by placing a fresh window between the gun and the sample, but without preventing helium flow.

I haven't been able to determine accurately if the Rh peak is a linear function using reduced input energy, or by changing the distance of the target sample from the tip of the gun. So far, all peaks have dropped off too quickly and are also buried promptly by the calcium and sulfur left foot, depending on sample scanned.

I'll post more spectra images later with the results of these other three tests, and a water scan for comparison.

Again, thanks for all your input on this matter.