News:

:) Remember, you need to be logged in to see posted attachments!

Main Menu

Nasty Boundary Fluorescence Analytical Situations

Started by John Donovan, September 18, 2013, 09:56:09 AM

Previous topic - Next topic

Probeman

Quote from: Probeman on April 03, 2024, 10:31:03 PM
Note that if you have Probe for EPMA you can perform a boundary fluorescence correction directly on your unknown data from the Analyze! Window:

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

Note that the SF boundary correction does not yet incorporate a Bragg defocus calibration, so at the moment it is a worse case scenario correction. Which depends on the orientation of the boundary relative to the Bragg crystal on the spectrometer being utilized.

But I believe a Bragg defocus calibration is currently being developed...

Until this Bragg defocus correction is implemented and one wants to correct for secondary boundary fluorescence (as opposed to FIBing out the sample and mounting the individual grains to avoid fluorescence from nearby phases), you should orient your sample in the stage so that the spectrometer making the trace element is aligned with the boundary. That is the phase boundary you are making measurements adjacent to, should point towards the spectrometer making the trace measurement.  Then as one moves the stage away from the adjacent boundary, there is little to no Bragg defocus effects.

And because the Bragg defocus effects are minimized, the boundary fluorescence model calculated from PENFLUOR/FANAL will not over correct the k-ratios.

Of course the opposite approach can also be attempted whereby one orients their specimen in the sample stage so that the adjacent boundary points directly *away* (at 90 degrees) from the spectrometer making the trace measurement.  And then one limits their trace measurements to points being at least 50 or 100 um away from the adjacent boundary.  Then one might hope that the WDS Bragg defocus reduces the detection of the boundary fluorescence emissions and no boundary fluorescence correction is necessary.

Of course with EDS there are no Bragg defocus effects, so the model from PENFLUOR/FANAL should apply as is for elements measured by EDS.
The only stupid question is the one not asked!

Probeman

#91
Quote from: Facundo Bilbao on June 04, 2024, 06:50:19 AM
Quote from: Probeman on October 20, 2022, 01:33:05 PM
Quote from: Yishen Zhang on October 20, 2022, 12:50:51 PM
I went through the SF correction and have successfully modeled my own standards file (so many hours!)

So you have run your standards to create .PAR files for your standard compositions using Standard.exe?  Nice!


I have a question regarding the .PAR files used in the FANAL program. What parameters exactly does Standard simulate for set file for an specific material and what information is in the .PAR file? And why does it take a long time to simulate those parameters?

The Standard application merely provides the GUI for the PENFLUOR/FANAL FORTRAN programs.  The PAR calculations are performed in PENFLUOR (the left side of the GUI in Standard), while the boundary fluorescence k-ratios are extracted by FANAL (the right side of the GUI in Standard). 



Note that if the same material is specified for both the beam incident and boundary phases, the k-ratios calculated are for a bulk sample:

https://smf.probesoftware.com/index.php?topic=152.0

I've attached the paper on the FORTRAN programs and also the source code  below.  One helpful thing is to look at the output in Excel by checking the "Send To Excel" checkbox. That would the the output to the k-ratios2.dat file and also check here:

https://smf.probesoftware.com/index.php?topic=58.msg5895#msg5895

There are some minor modifications to the FANAL code by Donovan, so I've provided both the modified and original code from Llovet.
The only stupid question is the one not asked!

John Donovan

Quote from: Facundo Bilbao on June 04, 2024, 06:50:19 AM
...And why does it take a long time to simulate those parameters?

The PENFLUOR code is basically the PENELOPE/PENEPMA Monte Carlo code modified to simulate multiple keVs.  Monte Carlo can take a long time to obtain reasonable precision on a PC.  The default is 3600 seconds per keV, but one can reduce that time if precision is not important.

The advantage of PENFLUOR/FANAL is that once the PAR file has been calculated one can extract k-ratios for any keV and element/x-ray line and distance from the boundary in seconds.  Previously we had to run PENEPMA hours for each boundary distance.
John J. Donovan, Pres. 
(541) 343-3400

"Not Absolutely Certain, Yet Reliable"

Ben Buse

Quote from: Probeman on March 29, 2022, 04:27:59 AM

So if indeed there are epoxy formulations with several percent Cl, then the boundary fluorescence issue of this element at trace concentrations (near epoxy boundaries) could become a much larger concern for us.

I'll attempt to calculate a PAR file with "several percent chlorine" and present these results...

Just wondering if you ever did this, and whether par files or images available

Probeman

Quote from: Ben Buse on October 03, 2024, 05:24:53 AM
Quote from: Probeman on March 29, 2022, 04:27:59 AM

So if indeed there are epoxy formulations with several percent Cl, then the boundary fluorescence issue of this element at trace concentrations (near epoxy boundaries) could become a much larger concern for us.

I'll attempt to calculate a PAR file with "several percent chlorine" and present these results...

Just wondering if you ever did this, and whether par files or images available

Ha, I don't think I did.

But it only takes a few seconds using the PENFLUOR-FANAL GUI in Standard...

If one selects compounds already calculated in the PENEPMA12 distribution, e.g., you can find an orthoclase (high K so possibly fluorescing Cl), an Epon epoxy (0.3 wt% Cl) and a Cl std (pure Cl):



Looks like it's less than 10 PPM or so a few microns away from the epoxy boundary, so not too much of a problem it would seem.  A much bigger problem for trace Cl was this incident that occurred a few years ago in our lab:

https://smf.probesoftware.com/index.php?topic=423.msg2288#msg2288
The only stupid question is the one not asked!

John Donovan

#95
We have modified the secondary boundary fluorescence code in CalcZAF so that one can perform the correction on more than a single element at a time (though this particular example does not show this), but the new element list does allow for multiple elements to be corrected for secondary boundary fluorescence at a time:



Note that if the contribution of secondary fluorescence is less than a one percent or so, running the boundary correction in CalcZAF in probably not necessary as the matrix correction will not be affected much. Therefore one can simply subtract the contribution from secondary fluorescence using the k-ratios2.dat file in Excel or other post processing application.

But it depends on the physics details. If performing this subtraction manually be sure to utilize the total concentration (beam incident material or MatA PLUS boundary material or Mat B) MINUS the concentration only from the beam incident material (MatA). This is to ensure you only subtract the contribution from the boundary material from your measurements in the beam incident material.

Exceptions to this rule would include things like correction of an element that is interfering with another element and of course secondary fluorescence artifacts that are larger than a percent or so.

These situations are more common than people realize. For example here is the secondary fluorescence of Fe measured in carbon adjacent to Fe metal:



And it gets worse:



In the previous Fe in carbon example one might get away with a simple subtraction, but in the Cr in Fe example, one must perform the secondary boundary fluorescence correction in CalcZAF, inside the matrix correction (or using Probe for EPMA where it is very easy).
John J. Donovan, Pres. 
(541) 343-3400

"Not Absolutely Certain, Yet Reliable"