Explain This If You Can!

[Probeman]

No speculations on what could be causing these different TDI trends on 3 simultaneous spectrometers...?

https://smf.probesoftware.com/index.php?topic=144.msg12205#msg12205

speculations... there is a bit too little information on other stuff to exclude the hardware effects. In case there is no hardware effect, I would speculate that maybe, maybe, this is crystallographic orientation thing. I cant remember now where I had read this, but as far I could remember it was somewhere shown that analyzing apatite-F at different orientations shows sever F loss where at other orientations shows nearly no F loss. I could get from that experiment the mind-shortcut that it is e-beam--crystallographic orientation controlled F loss. But Your experiment makes me start to think that maybe there is no loss at all (it was always for me illogical that negative (-1 valence) F ion would move toward negatively charged sample surface! There is probably only crystallographic repositioning of F, where for some spectrometer it gets hidden behind other atoms in lattice and thus absorption significantly goes up (decreasing intensity), while for different angle spectrometer they stay completely unhindered.

I think it would be easy to throw this hypothesis out the window If same behavior would be observed after redoing experiment with rotated sample.

This is a reasonable explanation I think.

The only thing that bothers me with this idea is that this would assume that the orientations of both the BaF2 and the CaF2 would be similar, yes?  Also I note that these are both cubic compounds, so not sure how orientation would affect ion migration...  but it's an interesting idea.

Here is a drawing of the std mount with the positions of CaF2 and BaF2 outlined and the locations of the Cameca spectometers 1, 2 and 4 shown:



The thing that makes a little more sense to me would be that perhaps the carbon coat connection to the mount might be only in one spot (towards spec 1?), so the sub surface negative charge would flow towards that point, and therefore "push" the F- ions away from spec 1, causing the large intensity drop over time, whereas spec 4 might see the least reduction in F- ions because it was getting more F- ions "pushed" towards it over time?

How does that sound as an explanation?

[sem-geologist]

easy to check (both hypothesis), as it is 1 inch round mount. I suggest to rotate it about 45 degrees and retry the experiment. If result will be the same – then our both hypotheses are failed, and the reason is buried somewhere in hardware.

[Probeman]

Yeah I thought of that, but I would have to ensure that conductive contact is made at specific points relative to the spectrometer orientation.  I think it might be easier to try using a thin insulator on the top of the mount and then bridge that insulator in specific places using a small drop of carbon.  Maybe this weekend...

[Probeman]

This is a close up of the 4096 x 4096 quant x-ray maps that Radek acquired:



https://smf.probesoftware.com/index.php?topic=1791.msg13743#msg13743

It's slightly amazing to me how one can obtain reasonable sensitivity at only 15 millisec per pixel at 100nA.  Accuracy for major elements is maintained by using the logarithmic dead time correction which ensures accuracy up to 300 to 400 kcps.

The image reminds me a bit of van Gogh's "Starry Night"...

[Les Moore]

4K image eh, really nice.
You'll have to upload it to your home TV to see the full resolution. Not many computer monitors have 4K resolution and within a normal business, none.

The danger here is that smaller pixels accurately capture the interaction volume issues:
1. Are the turquoise rings another phase or just a representation of sharp transition smoothed out by electron volume effects.    (MC X-Ray, DTSA etc)
2. Are the turquoise blobs yellow blobs just under the surface.  (MC X-Ray, DTSA, Casino etc)
3. Are the edges due to a substrate fluoresence issue (Penepma)
4. More I haven't thought about....

So you map at lower and lower kV's and get better resolution but poorer S/N
Paradoxically, a mental image of the best spatial resolution equaling the electron beam spot size occurs right at the excitation voltage as there is no excitation from deeper in the sample; but you get almost no counts.

   

[John Donovan]

The danger here is that smaller pixels accurately capture the interaction volume issues:

1. Are the turquoise rings another phase or just a representation of sharp transition smoothed out by electron volume effects.    (MC X-Ray, DTSA etc)
2. Are the turquoise blobs yellow blobs just under the surface.  (MC X-Ray, DTSA, Casino etc)
3. Are the edges due to a substrate fluoresence issue (Penepma)
4. More I haven't thought about....

So you map at lower and lower kV's and get better resolution but poorer S/N
Paradoxically, a mental image of the best spatial resolution equaling the electron beam spot size occurs right at the excitation voltage as there is no excitation from deeper in the sample; but you get almost no counts.

Sure, the same thing can happen at "normal" spatial resolution at high accelerating voltages.  Interpretation of quantitative results in multi-phase materials is always a challenge.

As you point out, the secondary fluorescence effects from nearby phases is particularly challenging. Though we have implemented a boundary fluorescence correction in Probe for EPMA for point analyses as seen here:

https://smf.probesoftware.com/index.php?topic=1545.0

we have not tested it for x-ray maps though the same code should work for both points and pixels (in CalcImage).  But either way, it's important to keep in mind that the location of the boundary relative to the points (or pixels), is critical for defining the SF boundary correction, as one must define one phase as the beam incident phase (the measured phase), and a single other phase as the boundary phase. See here for more discussion on this important aspect:

https://smf.probesoftware.com/index.php?topic=1545.msg12941#msg12941

Either way, it's a challenge not only because of the physics, but also the geometry, e.g., what is the phase boundary angle:

https://smf.probesoftware.com/index.php?topic=1545.msg13373#msg13373

and also from other artifacts such as Bragg defocus effects:

https://smf.probesoftware.com/index.php?topic=1545.msg13222#msg13222

In any case, the purpose of this 4K mapping attempt was simply to test the memory limitations of our mapping software and subsequent quantification, so hopefully there will be no attempt to interpret these quantitative maps.