unidentified peaks in WDS scan on Al2O3

[DirkMueller]

Hi all,

I am looking for traces in Al2O3. So I did a full spectra WDS scan on synthetic Al2O3 and found some peaks I cannot (reasonably) identify. Attached you can find two images showing the region around the Al Ka peak. Some Si might be reasonable, but the other two peaks? Krypton would somehow fit, but...

Machine: Cameca SX-100
Source: LaB6
Crystal: TAP (on Spec 1)
Beam: 15 kV, 200 nA
Steps: 2000
Dwell time: 2000 ms per step

Thanks in advance
Dirk

[sem-geologist]

Just to be sure it is not Kr, as its major lines are not there. 1) most probably what You see is just spectral artifacts of diffracting TAP crystal. Check the SiO2 or Si ("metalic") with similar conditions, You will see those peaks around Si Ka too. 2) or those are some not-described satellite peaks. I see those on our WDS scans on TAP too (interestingly for Si also). Albeit the position shifts depending from type of mineral thus I am more keen to think it is some satellite peaks.

[DirkMueller]

1) On LIF there is no Kr Ka peak. -> no Kr
2) I run the same scan on quartz and found similar peak(s). Attached the image. There is only one small peak on the right side of the Si Ka. The left one might be overlapped by the Si Kb.

So, whatever it is - satellite or diffracting artifact - it's not an additional trace element.

Thank you for your hints!

[Probeman]

1) On LIF there is no Kr Ka peak. -> no Kr
2) I run the same scan on quartz and found similar peak(s). Attached the image. There is only one small peak on the right side of the Si Ka. The left one might be overlapped by the Si Kb.

So, whatever it is - satellite or diffracting artifact - it's not an additional trace element.

Thank you for your hints!

Just out of curiosity I tried the same test on our instrument and see the same "Kr" peaks, but I agree they must be some kind of artifact (see attached images).

On a slightly related topic, do you also see an elevated "background" around the peak on your LTAP crystals?

[sem-geologist]

Yes I confirm that background is elevated for LTAP, it makes it useless for us as in our workflow (we analyse most often minerals with 30-40 elements) there simply is no space for background measurement. It can be nice crystal for material sciences (few elements to measure),but not for geology. The same problems goes for LPET's, and additionally to broader background tails, it has another horrendous artifacts - I guess the reflection from back side of crystal (some broad not very intensive peaks nearby 2nd order peaks), but normal PET has not these. The only large crystal which I have nothing bad to say about is LLIF (broadening is there, but it is not such a deal-breaker).

P.S. sorry, no pictures to share, as currently our extended spectrometer is down (which has LTAP).

[Probeman]

Yes I confirm that background is elevated for LTAP, it makes it useless for us as in our workflow (we analyse most often minerals with 30-40 elements) there simply is no space for background measurement. It can be nice crystal for material sciences (few elements to measure),but not for geology. The same problems goes for LPET's, and additionally to broader background tails, it has another horrendous artifacts - I guess the reflection from back side of crystal (some broad not very intensive peaks nearby 2nd order peaks), but normal PET has not these. The only large crystal which I have nothing bad to say about is LLIF (broadening is there, but it is not such a deal-breaker).

P.S. sorry, no pictures to share, as currently our extended spectrometer is down (which has LTAP).

I would not think that this peak broadening is due to reflection from the back of the crystal- if that were the case, I would expect that we'd see the same broadening from normal size PET crystals.  In addition I think emission lines measured on PET would not be penetrating that far!  That's millimeters of thickness!

But in any case, if this peak broadening is indeed causing difficulty in selecting off-peak background positions, one might consider utilizing an MAN type background correction. We utilize MAN backgrounds almost all the time for low to moderate Z materials such as silicates and oxides:

https://probesoftware.com/smf/index.php?topic=1378.msg9994#msg9994

Would certainly speed up your acquisitions (and improve precision).

[sem-geologist]

I would not think that this peak broadening is due to reflection from the back of the crystal-...

;
You understand me wrong, broadening is due to imperfection in bending of large XTAL. The "back-side reflection" which I created is my hypothetical cause for observed additional very broad, low amplitude peaks near 2nd order real peaks for LPETs (however, not seen on PET). While these artifacts are faint and would not affect the minor/major element analysis, the high current scans shows that they are for all lines and follow the general characteristics of 2nd order lines (i.e. if it is Ka Kb, then artifacts appear as two bumps spaced and weighted accordingly; it is is L lines, then artifacts also will resemble L lines). It can be only partly filtered with PHA diff mode, (same as 2nd order). These are very faint and strongly broadened, that's why I think It could be back-side reflection. Why we don't see those on PET? I don't know. I should mention that all our LPET's on two machines (SX100, SXFiveFE) were mounted at nearly same time (While getting a new SXFiveFE, w got SX100 expanded with additional spectrometer with large XTAL's). Maybe there is something wrong in that production series of LPET's. (had to be produced before 2014, the year it was mounted). I don't know the answers, albeit I know that artifacts are making work with LPETS often more complicated than it could be. There is not only no place for background. often these "bump" artifacts is overlapping with peaks which needs to be measured, and so Peak interference corrections needs to be applied. That is critical for some trace measurements. (Particularly if someone relies on spreadsheets or other systems for finding interference, those artifacts are absent in those systems; only high current slow WDS scan on high concentration material can reveal those artifacts).

[Probeman]

You understand me wrong, broadening is due to imperfection in bending of large XTAL. The "back-side reflection" which I created is my hypothetical cause for observed additional very broad, low amplitude peaks near 2nd order real peaks for LPETs (however, not seen on PET).

I agree that spectral artifacts are produced "due to imperfection in bending of large XTAL". I doubt that "back side reflection" of the crystals are involved in any PET artifacts. A more likely cause is mis-orientation of micro domains due to the polygonization process produced during thermal cycling of these crystals after bending. This thermal cycling is done at the factory to improve the reflectivity of the Bragg crystals after they are plastically deformed from bending. Indeed these micro mis-orientations are the cause of the long tails we observe for all WDS peaks.

[Mike Jercinovic]

To get back to Dirk's original question for Al and Si, I believe these satellites at low energy (relative to Ka emission) are due to the radiative Auger process (KLL RA in this case).  Few X-ray catalogues report any satellites other that the typical ones that are from multiple ionizations and their effect on the Ka emission and are at higher energy than the Ka line.  These light element spectra have low fluorescence yields, so much of the relaxation ends up being through electron emission, radiative and non-radiative Auger.  Part of the radiative Auger process involves emission of an X-ray at or near the Auger electron energy.  There is a peak in the Auger spectrum (measured by electron spectroscopy), and the radiative Auger spectrum (as measured by X-ray spectroscopy) for Si that is just above 1600eV, so at just above 30000 sin theta on TAP.  Although this line is very subtle, expected to be at or below 0.01% of the Ka1,2 line, you can easily see it, particularly with a large crystal.  When I measure the intensity on our LTAP, it is about 0.005% of the Ka emission.  Note that this is a useful and interesting thing!  Jun Kawai, I think, introduced the term EXEFS for this... Extended X-ray Emission Fine Structure.  There are potential uses as details of this emission will be effected by band structure/bonding environment considerations.

[AndrewLocock]

To get back to Dirk's original question for Al and Si, I believe these satellites at low energy (relative to Ka emission) are due to the radiative Auger process (KLL RA in this case).  Few X-ray catalogues report any satellites other that the typical ones that are from multiple ionizations and their effect on the Ka emission and are at higher energy than the Ka line.  These light element spectra have low fluorescence yields, so much of the relaxation ends up being through electron emission, radiative and non-radiative Auger.  Part of the radiative Auger process involves emission of an X-ray at or near the Auger electron energy.  There is a peak in the Auger spectrum (measured by electron spectroscopy), and the radiative Auger spectrum (as measured by X-ray spectroscopy) for Si that is just above 1600eV, so at just above 30000 sin theta on TAP.  Although this line is very subtle, expected to be at or below 0.01% of the Ka1,2 line, you can easily see it, particularly with a large crystal.  When I measure the intensity on our LTAP, it is about 0.005% of the Ka emission.  Note that this is a useful and interesting thing!  Jun Kawai, I think, introduced the term EXEFS for this... Extended X-ray Emission Fine Structure.  There are potential uses as details of this emission will be effected by band structure/bonding environment considerations.

The K-L_2,3L_2,3 radiative Auger transitions correspond in energy (within a few eV) to the non-radiative Auger transitions.
The major Auger transitions are tabulated at https://srdata.nist.gov/xps/
A more comprehensive catalog of Auger transitions is given by:
Coghlan, W.A. & Clausing, R.E. (1973). Auger catalog calculated transition energies listed by energy and element. Atomic Data 5, 317-469.

Examples of Mg K-LL RAE lines are in the attached file MgO-TAP-JEOL.pdf
Three TAP-crystal spectrometers were used on a JEOL8900R electron microprobe to scan the range 0.96 to 1.2 keV.
Both the K-L_2,3L_2,3 and K-L₁L_2,3 radiative Auger transitions of Mg are observed.

However, the same experiment using our Cameca SX100 gives an important difference!
Spectra obtained with 3 TAP-crystal spectrometers on our Cameca X100 over the range 0.96 to 1.2 keV are in the attached file:
MgO-TAP-Cameca.pdf
Both the K-L_2,3L_2,3 and K-L₁L_2,3 radiative Auger transitions of Mg are observed.
Questions:
- What is the broad asymmetric peak at ~1.025 keV?
- Why does this peak have very different intensities on the 3 TAP-crystal spectrometers of the Cameca SX100?

This peak at ~1.025 keV is not a diagram line, not a satellite line, not a RAE line, and not a higher-order diffraction line.
It does not appear to be a multiple diffraction line (Renninger effect).
It is observable on Zn, Mg, Al, Si, and Zr elemental reference materials.
In these materials, it occurs at a fixed fraction of the energy of the principal diagram line of each element.

I would be very interested to know if others have observed this with TAP-crystal spectrometers.

Thanks, Andrew

[sem-geologist]

Examples of Mg K-LL RAE lines are in the attached file MgO-TAP-JEOL.pdf
Three TAP-crystal spectrometers were used on a JEOL8900R electron microprobe to scan the range 0.96 to 1.2 keV.
Both the K-L_2,3L_2,3 and K-L₁L_2,3 radiative Auger transitions of Mg are observed.

However, the same experiment using our Cameca SX100 gives an important difference!
Spectra obtained with 3 TAP-crystal spectrometers on our Cameca X100 over the range 0.96 to 1.2 keV are in the attached file:
MgO-TAP-Cameca.pdf
Both the K-L_2,3L_2,3 and K-L₁L_2,3 radiative Auger transitions of Mg are observed.
Questions:
- What is the broad asymmetric peak at ~1.025 keV?
- Why does this peak have very different intensities on the 3 TAP-crystal spectrometers of the Cameca SX100?

This peak at ~1.025 keV is not a diagram line, not a satellite line, not a RAE line, and not a higher-order diffraction line.
It does not appear to be a multiple diffraction line (Renninger effect).
It is observable on Zn, Mg, Al, Si, and Zr elemental reference materials.
In these materials, it occurs at a fixed fraction of the energy of the principal diagram line of each element.

I would be very interested to know if others have observed this with TAP-crystal spectrometers.

Thanks, Andrew

SX100, and SXFive - KLL are visible, however that weird line is not here on any of our TAP crystals on both of Cameca probes. It is clearly some kind of sdtrange artifact in your probe.

[AndrewLocock]

SX100, and SXFive - KLL are visible, however that weird line is not here on any of our TAP crystals on both of Cameca probes. It is clearly some kind of sdtrange artifact in your probe.

Thanks.
Yes, also, when I look at Sandrin Feig's EPMA-Method Development Tool, I do not see this feature in those scans either.

One wonders what is going on...

[Mike Jercinovic]

Oh man, this is really interesting Andrew!  So let me get this right, these lines occur only on low pressure spectrometers with TAP (LTAP, etc.)?  And, these lines always occur at about 0.225 keV below the main diagram line?  If you go to some other crystal in the same spectrometer, do you see the same thing?  Seems like you and sem-geologist now confirm is that this is not a TAP thing.  One of the many interesting things here is that the line intensities seem pretty different from one crystal to another on the SX100... looks like relatively lowest on the LTAP (PHA or pressure related?).  The fact that they appear a fixed eV below the main diagram line is most curious as they are not emission lines, and that they are not showing up in the JEOL or the sem-geologist SXs.  This seems to suggest something to do with the rest of the detection system, like detectors, but something particular to yours.  Seems like it could relate to an inelastic x-ray scattering feature like X-ray Raman where a percentage of incoming x-rays lose a quantized amount of energy, and maybe this can be related somehow to the detection components, but it has to be something common to all of your low P spectrometers, like the P10. I wonder if the P10 is okay in your machine?  Just guessing.  I need to look and see what we get on our systems too.  Good puzzler.

And thanks for posting the Auger references!  Unfortunately, our library only covers Atomic Data back to about 1993, but I'll figure it out.

Mike J.

[John Donovan]

This peak at ~1.025 keV is not a diagram line, not a satellite line, not a RAE line, and not a higher-order diffraction line.
It does not appear to be a multiple diffraction line (Renninger effect).
It is observable on Zn, Mg, Al, Si, and Zr elemental reference materials.
In these materials, it occurs at a fixed fraction of the energy of the principal diagram line of each element.

I would be very interested to know if others have observed this with TAP-crystal spectrometers.

This unidentified peak on TAP is closer to the F Ka peak so maybe or maybe not related:

https://probesoftware.com/smf/index.php?topic=992.0

[AndrewLocock]

Oh man, this is really interesting Andrew!  So let me get this right, these lines occur only on low pressure spectrometers with TAP (LTAP, etc.)?  And, these lines always occur at about 0.225 keV below the main diagram line?  If you go to some other crystal in the same spectrometer, do you see the same thing?  Seems like you and sem-geologist now confirm is that this is not a TAP thing.  One of the many interesting things here is that the line intensities seem pretty different from one crystal to another on the SX100... looks like relatively lowest on the LTAP (PHA or pressure related?).  The fact that they appear a fixed eV below the main diagram line is most curious as they are not emission lines, and that they are not showing up in the JEOL or the sem-geologist SXs.  This seems to suggest something to do with the rest of the detection system, like detectors, but something particular to yours.  Seems like it could relate to an inelastic x-ray scattering feature like X-ray Raman where a percentage of incoming x-rays lose a quantized amount of energy, and maybe this can be related somehow to the detection components, but it has to be something common to all of your low P spectrometers, like the P10. I wonder if the P10 is okay in your machine?  Just guessing.  I need to look and see what we get on our systems too.  Good puzzler.

And thanks for posting the Auger references!  Unfortunately, our library only covers Atomic Data back to about 1993, but I'll figure it out.

Mike J.

I have only looked for this behavior (these lines) with the TAP crystals on my Cameca SX100; all of these are low-pressure spectrometers.
In detail, these features occur at about 81.7% of the energy of the principal diagram lines.
Here, I refer to these anomalous lines as low-energy non-diagram features (LEND features), as follows from the averages of observations in the pure elements:

   Main Line (keV)   LEND (keV)
Zr Lα   2.0417   1.6701
Si Kα    1.7398   1.4224
Al Kα    1.4866   1.2157
Mg Kα   1.2536   1.0243
Zn Lα   1.0118   0.8276

Interestingly, when one looks at Zn and Zr, there are also features that correspond exactly to the same energy-fraction of the Lβ₁ lines.

In terms of Cameca units (100,000sinθ), these features are at Kα /0.817 or Lα / 0.817, etc.

My understanding of X-ray Raman peaks is that those lines should be very close to the diagram lines, which these LEND features are not.
Similarly, the Compton shift would be very tiny at this energy range.

I am now wondering if I have some thin film coating on my TAP crystals, causing hybrid diffraction (I am grasping at straws).

With regard to the 1973 Auger catalog, I can probably email it to you.

Cheers,
Andrew