EPMA Teaching Examples

[Probeman]

I started this topic so we can post interesting examples of EPMA issues for educational purposes...

Here is an example.  I knew that the U Ma peak is close to the Ar absorption edge, but I did not realize that the Cd La peak is also near this absorption edge. 



This is why it is important to scan the region around your analytical peak, to not only avoid interfering secondary emission lines on the off-peak positions, but also to avoid interpolating our background fit across absorption edges from our detectors.
john

[Probeman]

Here is a microanalysis example I sometimes give my students to make sure they are considering the physics properly.

This example was created using the "demo" EDS simulation mode in Probe for EPMA which utilizes Penepma to generate an EDS spectra in real time. One could also do it on an actual instrument, but I find it useful to have each student download PFE and have them install it on each of their laptops.  That way they can explore the possibilities on their own.

This spectrum is simply an EDS acquisition in PFE on a pure Mo standard at 20 keV.  I ran it for 300 seconds and the spectrum looks quite good.  But just to prove the point I then ran it for 3000 seconds and that what I'll show below, though even a 1 min EDS acquisition makes the point clearly.  Here is the full EDS spectrum "acquired" by Penepma:



We can then zoom in on the Mo La line and see that the KLM markers line up nicely with the L emission family:



Penepma is so cool!   

Now we zoom in on the Mo Ka ROI and what do we see?



Nothing!   Now we ask our students: OK, so why don't we see the ~17 keV Mo Ka emission?  Hint: we're using a 20 keV beam energy...  then I ask them to look up the Mo K edge energy...

I'm sure there lots of such entertaining and educational examples for students, so please feel free to share your favorites.

[Probeman]

Teaching probabilities can be fun.  I just found these old apps I had slapped together for my Weird Science freshman seminar class (for non-science majors) about 5 years ago, so nothing fancy. In fact you can probably find better apps elsewhere on the net.  But they have their appeal.   Also probably good for high school students and science minded middle schoolers even.

Dice.exe is a simple app that allows one to specify the number of dice, and the number of faces for each die.  Default is 2 dice with 6 faces! 



Here's a chart I found that explains what you're seeing with two dice with 6 faces each:



Toss.exe just tracks random coin flips. Run it a few times by clicking the Toss button. You might be surprised.



The exe files are attached below (login to see), along with the MS chart control (MSChrt20.ocx). Just copy the two .exe files and the .ocx file to any folder and they should run fine from there.

If anyone is interested in the source code I'd be happy to share it.

[Probeman]

This is a very simple app to demonstrate Brownian motion to students.  It is not at all scientific as it's completely schematic, but it does help to illustrate the effects of density and temperature on Brownian motion I think.

The Brownian.exe and the associated color palette files (which can be modified as necessary) are attached below (remember to login to see attachments).

To use the app, copy the exe and the .fc color palette files to any folder, then double-click the Brownian.exe file, and then click the Start menu. There should be no other dependencies on most PCs, but please let me know if you see any problems.

To adjust the temperature (drawing speed), one can use the Options menu or simply tap the up/down cursor keys.  Up for higher temperatures, down for lower temperatures.  To adjust the mean free path (distance), again use the Options menu or simply use the right/left cursor keys. Right to increase mean free path and left to decrease mean free path.

It is not possible to show the temperature effects in a screen shot, but here is the app with a long mean free path (low density):



and here is the app with a short mean free path (high density):



The palette color represents the Z dimension... maybe this app might be useful for middle and high schoolers?

[Probeman]

I recently was testing some new code over the weekend, so running Probe for EPMA in "demo" or simulation mode and acquiring some multi-point and off-peak bgds on standard samples.  As many of you know, this PFE simulation mode is what we use in the classroom to teach EPMA to our students as the simulation is quite realistic and that way they can all follow along on their laptops. In fact, this simulation mode is almost too realistic as I am about to explain!

So I acquired some standard data and just for fun I clicked the Analyze button and much to my surprise I saw some quite questionable results as seen here for my Orthoclase standard:

ELEM:       Si       K      Al      Mg      Fe      Ca      Mn       O       H      Na      Ba   SUM 
    63  30.153  12.932   7.960    .014   1.679   -.027    .030  45.798    .000    .675    .054  99.268
    64  30.275  12.863   7.804    .024   1.687    .021   -.084  45.798    .000    .675    .054  99.118
    65  30.207   7.435   7.829    .009   1.612    .004   -.035  45.798    .000    .675    .054  93.588

AVER:   30.212  11.077   7.864    .016   1.659   -.001   -.030  45.798    .000    .675    .054  97.324
SDEV:     .061   3.154    .084    .008    .041    .024    .057    .000    .000    .000    .000   3.237
SERR:     .035   1.821    .048    .004    .024    .014    .033    .000    .000    .000    .000

What is going on? 

Then it struck me (ouch!).  Now I must explain by going back to actual instrument measurements a bit: as we know, our various EPMA softwares all have some idea of where the emissions lines should appear in our WDS spectrometer ranges, but due to the fact that our EPMA instruments are not mechanically perfect, we know that we generally need to start our new probe runs by tuning our spectrometers so our on-peak measurement positions are right at the top of the emission peaks, where a slight variance in the positioning of the WDS spectrometer will not cause a severe change in the recorded intensity. Just imagine how much the intensity would vary if we happened to be measuring our peak intensity on the side of our emission lines!  A slight change in spectrometer position will produce a large change in the measured intensity. 

For EDS measurements the actual emission peak positions are not an issue for two reasons, one, the EDS detectors are relatively stable, so the peaks don't move around much, and two, the spectral resolution of the EDS detectors are so poor, any detector or electronics instability is usually masked by the extreme widths of the measured emission lines.

Now, back to my demo simulation run. So it turns out that in my haste to test my code, I neglected to peak up the spectrometers before acquiring the standard intensities. Now you might say, but this is just a simulation, doesn't the software know where the emission lines will appear in the simulated spectrometer range?  And yes, the software does know where they should appear, but the spectral simulation software (in this case the Penepma Monte Carlo physics package), has slightly different theoretical emission positions in its emission energy databases compared to the Armstrong emission line database that PFE has utilized historically.  Also, there are several other factors such as the Bragg refractive index correction which is applied to each Bragg crystal (even in simulation mode) and probably a few other factors that I'm not even thinking of.

In addition, the simulation mode in PFE is designed so when one moves to a specified stage or spectrometer position, the software introduces a small amount of error, so just as in an actual EPMA instrument, the actual position arrived at is not exactly at the target position, because, you know, reality.

So the result is that the simulated WDS spectrometer emission lines do not appear exactly at the expected Bragg angles (just as they don't appear exactly where they are expected on your actual instrument!).  Now one could say this is a bug, but I prefer to think of this as a simulation feature since it more closely resembles the performance of an actual instrument.

So I then ran the peaking procedure on all the elements and lo and behold, we now obtain these quantitative results:

ELEM:       Si       K      Al      Mg      Fe      Ca      Mn       O       H      Na      Ba   SUM 
    96  30.395  13.004   8.973    .041   1.706    .015    .014  45.798    .000    .675    .054 100.676
    97  30.160  12.873   8.975    .021   1.668   -.003    .039  45.798    .000    .675    .054 100.261
    98  30.191  12.888   8.933    .001   1.704   -.006    .017  45.798    .000    .675    .054 100.256

AVER:   30.249  12.922   8.961    .021   1.693    .002    .023  45.798    .000    .675    .054 100.397
SDEV:     .128    .072    .024    .020    .021    .011    .014    .000    .000    .000    .000    .241
SERR:     .074    .042    .014    .012    .012    .006    .008    .000    .000    .000    .000

The K Ka measurements are now quite a bit better.  So this is actually a "teaching moment" to help remind your students that yes, with WDS spectrometers, we do need to tune the spectrometer positions to the actual measured positions, even if we are running in simulation mode. The simulation mode in PFE is that realistic!

[Probeman]

As we all know there is a relationship between wavelength and energy.

And in fact we know that angstroms = 12,398 / electron volts (e.g., 1 angstrom = 12.398 keV), but where does that 12,398 number come from?  It all comes down to Planck's constant and the speed of light.

Here is a short explanation that might be useful and/or fun for students when teaching your EPMA class/lab:



I've attached the explanation as a Word document below if you'd like to edit it a bit for teaching (login to see attachments).

[Brian Joy]

And in fact we know that angstroms = 12,398 / electron volts (e.g., 1 angstrom = 12.398 keV), but where does that 12,398 number come from?  It all comes down to Planck's constant and the speed of light.

No, length does NOT equal energy.  YOU NEED TO INCLUDE THE UNITS!!!

λ [Å] = 12.398 Å·keV / E [keV]

Also, E = hν= hc/λ

[Probeman]

Hi Brian,
Calm down, dude (were all caps really necessary?).     Here is a modified version:



Thank-you though for catching the other typo.  Nice to have you to keep us on our toes!   

The edited Word document is attached below for those that are interested (I also updated the Microsoft equations since they weren't editable anymore in the 25 year old document).