Author Topic: Nasty Boundary Fluorescence Analytical Situations  (Read 53710 times)

Probeman

  • Emeritus
  • *****
  • Posts: 2798
  • Never sleeps...
    • John Donovan
Re: Nasty Boundary Fluorescence Analytical Situations
« Reply #15 on: October 05, 2014, 12:55:35 PM »
The latest update of CalcZAF (and Probe for EPMA) has a small change in the output of the Fanal "couple" calculations.

First the output folder name now contains the takeoff angle for the model as seen here:



And the actual conditions and materials are now documented in an additional Fanal.txt text file as seen here:



The contents of this file as as follows:

"SiO2.par"                      <- Material A (beam incident phase)
"TiO2.par"                      <- Material B (boundary phase)
"Ti.par"                        <- Material BStd (standard)
40                              <- take off angle (surface takeoff though Fanal input uses polar takeoff)
15                              <- beam energy in keV
22                              <- emitted element atomic number
1                               <- emitted element x-ray
The only stupid question is the one not asked!

Probeman

  • Emeritus
  • *****
  • Posts: 2798
  • Never sleeps...
    • John Donovan
Re: Nasty Boundary Fluorescence Analytical Situations
« Reply #16 on: November 06, 2014, 06:39:03 PM »
I have mentioned previously an issue with using Penflur/Fanal in Standard.exe to model secondary fluorescence for low energy emission lines such as oxygen Ka as seen here:

http://probesoftware.com/smf/index.php?topic=119.msg479#msg479

The same thing applies to modeling low energy x-ray lines for secondary fluorescence from boundary phases such as the L family of first row transition elements. In this example we look at Fe La (not Ka) in Ni adjacent to pure Fe. But because the Fe L edge is below the default minimum electron/photon energy of 1 keV, the Fanal application is unable to correctly model the production of Fe La and we obtain the following bogus SF artifact as seen here:



However by recalculating the Fe-Ni system to below the Fe L edge energy we can obtain a more accurate calculation for secondary fluorescence from L family lines as seen here:



The good news is that I have modified the Standard GUI for Penfluor/Fanal so that the program will *automatically* adjust the minimum electron/photon energy in order to include the K emission lines for elements below Na.

For modeling of higher Z elements with emission line edge energies below 1 keV, you will still need to recalculate the PAR file by setting the minimum electron/photon energy as usual. The modified window is seen here:

The only stupid question is the one not asked!

Probeman

  • Emeritus
  • *****
  • Posts: 2798
  • Never sleeps...
    • John Donovan
Re: Nasty Boundary Fluorescence Analytical Situations
« Reply #17 on: March 23, 2016, 04:20:41 PM »
I need a favor from CalcZAF/Standard/Probe for EPMA users.

If you've utilized the Standard program GUI for modeling secondary fluorescence across boundaries (this topic) or the "fast Monte-Carlo" matrix corrections in CalcZAF (or Probe for EPMA), described here:

http://probesoftware.com/smf/index.php?topic=47.0

please reference the paper by Llovet et al. 

I would like Xavier to update the Penfluor/Fanal code to implement the new Penepma 2014 code, but he says he hasn't seen much in the way of citations for this paper, and so he's not sure there is enough interest in the method.  I really only contributed the GUI and the matrix method (matrix.mdb), but think it is an outstanding contribution and I also would like to see it cited, if anyone has utilized the Standard GUI for secondary fluorescence modeling or the fast MC methods in CalcZAF or PFE.

The citation would be:

Llovet, et al., "Secondary fluorescence in electron probe microanalysis of couple materials", J. Appl. Phys., (2012)

Thanks and please feel free to post your publication in this topic as well!
john
« Last Edit: March 23, 2016, 04:25:51 PM by Probeman »
The only stupid question is the one not asked!

John Donovan

  • Administrator
  • Emeritus
  • *****
  • Posts: 3251
  • Other duties as assigned...
    • Probe Software
Re: Nasty Boundary Fluorescence Analytical Situations
« Reply #18 on: April 26, 2016, 01:43:52 PM »
Hi John,
Thanks for your posts on this very interesting topic. I will try those corrections/steps and see how it goes.
Meanwhile, I am attaching a BSE image showing the Qti data I measured sometime back on a Qtz_Ti couple (the grains glued next to each other and polished).

The picture (attached below) describes it all.

Hi Ravi,
Thanks for posting your data. Should we assume the center of the first point (using a 10 um beam size) was 6 um from the boundary?

I'm not sure what beam size Wark and Watson used for their measurements, but your data is roughly comparable to theirs as plotted here:



Good work!
john

Edit by John: there is a silly typo in the plot above. It says that if the boundary axis is at right angles to the spectrometer axis, there will be little or no defocusing. But...

It should say "if the boundary axis is parallel to the spectrometer axis, there will be little or no Bragg defocusing"!  That is, little or no Bragg defocus effect on the measured intensity of the secondary fluorescence due to the boundary.

Attached below is a corrected version of the plot above
« Last Edit: April 26, 2016, 02:40:26 PM by John Donovan »
John J. Donovan, Pres. 
(541) 343-3400

"Not Absolutely Certain, Yet Reliable"

Probeman

  • Emeritus
  • *****
  • Posts: 2798
  • Never sleeps...
    • John Donovan
Re: Nasty Boundary Fluorescence Analytical Situations
« Reply #19 on: December 14, 2016, 10:35:53 PM »
I'm at AGU this week to give my trace MAN talk, but I've seen a few talks where I wonder if secondary fluorescence boundary effects have been considered enough.

On that note I've attached a few plots from the Standard.exe that model some trace element situations used to estimate thermometry and/or barometery in geological  systems. Remember to login to see attachments!
john
The only stupid question is the one not asked!

Probeman

  • Emeritus
  • *****
  • Posts: 2798
  • Never sleeps...
    • John Donovan
Re: Nasty Boundary Fluorescence Analytical Situations
« Reply #20 on: December 14, 2016, 10:48:18 PM »
And remember, secondary fluorescence boundary effects can also be negative as shown in the attachments below.

In these examples, the loss in intensity as the boundary is approached is *not* due to electrons leaking into the boundary phase. The intensity loss is because the emission line in question is fluoresced by its own matrix, and because the boundary phase does not contain the measured element, there is less self-excitation of the emitted line as the boundary is approached.

As John Fournelle has demonstrated, this self excitation intensity loss is also seen in epoxy mounted grains. The point being that no matter how small one's electron excitation volume is, the x-rays produced within that volume still travel tens to hundreds of microns further.
john
« Last Edit: December 15, 2016, 08:11:29 AM by Probeman »
The only stupid question is the one not asked!

Ben Buse

  • Professor
  • ****
  • Posts: 488
Re: Nasty Boundary Fluorescence Analytical Situations
« Reply #21 on: January 26, 2017, 01:53:33 PM »
Just corrected data for Ca secondary fluorescence in olivine - where olivine contains Ca.

Fanal model. Olivine ca. 300ppm Ca.



Calczaf works really well! Exported data from PFE to CalcZAF. Loaded an image, then drew the boundary. Calculated results:



Thanks John!

Ben
« Last Edit: April 14, 2020, 12:02:36 PM by John Donovan »

John Donovan

  • Administrator
  • Emeritus
  • *****
  • Posts: 3251
  • Other duties as assigned...
    • Probe Software
Re: Nasty Boundary Fluorescence Analytical Situations
« Reply #22 on: January 26, 2017, 02:58:04 PM »
Just corrected data for Ca secondary fluorescence in olivine - where olivine contains Ca.

Fanal model. Olivine ca. 300ppm Ca.

Thanks John!

Ben

Hi Ben,
You are more than welcome.    I appreciate you working with me!

We have to find a way to get information on these secondary boundary fluorescence artifacts more visible to geologists (and metallurgists for that matter!), and have them start checking for these effects (in Standard.exe) and if necessary, correct for them as well (in CalcZAF)! 

Since CalcZAF (and Standard) are freely downloadable apps, there really is no excuse for not doing so (except perhaps because the Llovet, et al. paper on this was published in the Journal of Applied Physics!)    :-\

I'm no geologist, but it seems to me that these trace element diffusion profiles at mineral grain rims are being significantly degraded by including these fluorescence artifacts in the thermodynamic chemical models for estimating the pressure/temperature/duration conditions of formation...  are there any modelers out there that can speak to this question?
john
John J. Donovan, Pres. 
(541) 343-3400

"Not Absolutely Certain, Yet Reliable"

Ben Buse

  • Professor
  • ****
  • Posts: 488
Re: Nasty Boundary Fluorescence Analytical Situations
« Reply #23 on: January 26, 2017, 11:26:42 PM »
Hi John,

I agree, when we (Bristol, Leeds, Oxford, Cambridge and Manchester) teach our NERC (research body) ATSC (Advanced Training Short Course) on EPMA we cover secondary fluroscence modelling using Standard.exe.

I should also mention for anyone with sharp eyes - the above data is a first approximation - corrected using already simulated materials (although a reasonable match) - for the final correction the actual compositions will be simulated where important/materials deviating significantly.

I guess also for clarity - in the data given above - its important to mention that it does not matter that the modelled material which the beam is hitting contains a small and possibly different about of Ca - for the calczaf correction subtracts the amount of Ca generated by secondary fluorescence. 

« Last Edit: January 27, 2017, 12:51:53 AM by Ben Buse »

Probeman

  • Emeritus
  • *****
  • Posts: 2798
  • Never sleeps...
    • John Donovan
Re: Nasty Boundary Fluorescence Analytical Situations
« Reply #24 on: January 27, 2017, 09:14:43 AM »
I guess also for clarity - in the data given above - its important to mention that it does not matter that the modelled material which the beam is hitting contains a small and possibly different about of Ca - for the calczaf correction subtracts the amount of Ca generated by secondary fluorescence.

Hi Ben,
That is actually a very good point.  I hadn't quite considered that implication, but you are correct. 

As you said, in CalcZAF (and soon PFE!), the correction for secondary fluorescence by a boundary phase is based on the contribution only from the boundary material. Therefore the beam incident material's composition only matters so far as the production of characteristic and continuum x-rays produced by the beam incident phase stimulates secondary fluorescence in the boundary material.

That said, it is important when subtracting larger secondary fluorescence effects to perform this correction during the sample matrix iteration, because if you are subtracting 2 or 3 wt% of an SF artifact from your sample matrix, the quant matrix correction needs to be recalculated, and especially if the fluoresced element is one of the elements in an interference correction. 

For example, if you are measuring trace Ni in a Cu alloy adjacent to a Co rich phase, you will measure a large Co signal (up to 4 wt.%) near the Cu-Co boundary and since Co Ka/Kb interferes with Ni Ka, you would be over correcting the interference correction of Co on the trace Ni in the Cu alloy. 

However, since I implemented both the spectral interference correction and the secondary boundary fluorescence correction within the matrix iteration, all these issues are handled automatically!
john
« Last Edit: January 27, 2017, 03:41:20 PM by Probeman »
The only stupid question is the one not asked!

Probeman

  • Emeritus
  • *****
  • Posts: 2798
  • Never sleeps...
    • John Donovan
Re: Nasty Boundary Fluorescence Analytical Situations
« Reply #25 on: February 08, 2017, 03:10:59 PM »
Maybe this will only confuse things, but I think some further explanation might be helpful with regard to the correction for secondary fluorescence from nearby phases. Here is an example of Zr diffusion from a zircon crystal into the surrounding glass.  The investigator in this case wanted to be sure that the halo of Zr we see around the zircon crystal is real diffusion and not simply an artifact of Zr fluoresced by x-rays emitted from the glass phase.



Note that the units here are log(wt.%) as a normal wt.% plot would not show the low level variation in Zr around the zircon crystal. 

Below is a plot of the wt% Zr (from point A to point B in the above image), showing the actual EPMA measurement, and also the secondary fluorescence modeling from the Standard application, assuming the glass composition as the beam incident phase (with ~1400 PPM Zr), and zircon as the boundary phase. Note that point A in the plot below is on the left and point B is on the right.

Now, if the boundary artifact is relatively large and the matrix composition changes significantly, then one should perform the secondary boundary fluorescence correction in CalcZAF so the matrix correction is recalculated automatically as the boundary artifact is subtracted.  But if the boundary artifact is relatively small (as in this Zr in glass example), then a simple subtraction is more than sufficient accuracy.



The "hatched" area is the degree of actual diffusion of Zr in to the glass phase. To obtain this concentration we can simply subtract the boundary artifact from the EPMA measurement and we have the actual Zr diffusion profile...
« Last Edit: April 14, 2020, 12:01:47 PM by John Donovan »
The only stupid question is the one not asked!

Probeman

  • Emeritus
  • *****
  • Posts: 2798
  • Never sleeps...
    • John Donovan
Re: Nasty Boundary Fluorescence Analytical Situations
« Reply #26 on: February 13, 2017, 05:05:02 PM »
This is an "old chestnut", but worth pulling out once in a while because I occasionally still see investigators thinking this effect is compositional, when it is simply a secondary fluorescence edge artifact. Here we see intensity as a function of distance in a beam incident composition of 1% Fe 99% Ni adjacent to a boundary material of pure Ni.



Note that all electrons come to rest in the beam incident material in this model, so there is no "electron leakage" going on here.  Question for discussion: why does the Fe Ka intensity drop off as the edge is approached?  And before we answer consider that we see a similar situation in pure Fe adjacent to pure Ni:



So what is going on here?
« Last Edit: April 14, 2020, 12:01:21 PM by John Donovan »
The only stupid question is the one not asked!

Ben Buse

  • Professor
  • ****
  • Posts: 488
Re: Nasty Boundary Fluorescence Analytical Situations
« Reply #27 on: February 22, 2017, 01:10:28 AM »
Hi John

Yes this was a timely reminder reading this yesterday - because it's a small effect and only effects limited number of elements - but could potentially be important in analysing small inclusions in a host - where getting accurate stoichiometry is important. This absence of expected self-fluorescence - in a bulk Fe would be fluorescenced by the Fe adjacent to the excitation volume - seems only to effect heavy elements like Fe - I guess where self-fluorescence from kb? is important. It has no impact on lighter elements

Ben

Probeman

  • Emeritus
  • *****
  • Posts: 2798
  • Never sleeps...
    • John Donovan
Re: Nasty Boundary Fluorescence Analytical Situations
« Reply #28 on: February 22, 2017, 08:02:24 AM »
Yes this was a timely reminder reading this yesterday - because it's a small effect and only effects limited number of elements - but could potentially be important in analysing small inclusions in a host - where getting accurate stoichiometry is important. This absence of expected self-fluorescence - in a bulk Fe would be fluorescenced by the Fe adjacent to the excitation volume - seems only to effect heavy elements like Fe - I guess where self-fluorescence from kb? is important. It has no impact on lighter elements

Hi Ben,
My understanding is that the effect is from additional Fe K shell fluorescence from continuum fluorescence produced in the beam incident sample (Fe), and because there is no Fe in the boundary phase to be fluoresced by the emitted continuum radiation as the boundary is approached. Yes, the same thing will occur for inclusions mounted in epoxy as John Fournelle and Phil Gopon pointed out a few years ago when studying iron silicide inclusions in epoxy.

Note that Fe Kb cannot fluoresce the Fe K shell.

The same thing happens for pure Si, but to a smaller degree:
john

« Last Edit: April 14, 2020, 12:00:42 PM by John Donovan »
The only stupid question is the one not asked!

Probeman

  • Emeritus
  • *****
  • Posts: 2798
  • Never sleeps...
    • John Donovan
Re: Nasty Boundary Fluorescence Analytical Situations
« Reply #29 on: February 22, 2017, 09:56:19 AM »
I should also mention that because of this decrease in self fluorescence near boundaries for elements such as Fe, Phil Gopon and I spent considerable time and effort adding 6 additional "non-traditional" lines for quantitative analysis in probe for EPMA as seen here:

http://probesoftware.com/smf/index.php?topic=152.msg3498#msg3498

These are the Ln, Lg, Lv, Ll, Mg, Mz lines.
john
The only stupid question is the one not asked!