CalcImage Quantitative X-ray Mapping Examples

[John Donovan]

Here is a pdf of my presentation yesterday on quantitative x-ray mapping for WDS EPMA (see attached file below if you are logged in).
john

[John Donovan]

We recently upgraded from Golden Software's Surfer 15 to Surfer 17.

One can choose the 32 bit version or the 64 bit version to install. We choose the 64 bit version. Interestingly the default installation folder that the Surfer installer offers you is "Surfer" as opposed to previous versions which always appended the Surfer version number to the folder by default, e.g., "Surfer 11" or "Surfer 15". We left the default installation folder as "Surfer" and proceeded with the installation which went fine.

Then we launched CalcImage and opened some existing maps, the first in Cameca (cartesian) stage mode as seen here for this alloy quantitative map in elemental weight percent:



and next in JEOL (anti-cartesian) stage orientation as seen here in this Mt St. Helens feldspar in oxide weight percent:



Bottom line, it appears that the CalcImage presentation output scripting still all works in the latest version (17) of Golden Software's Surfer app.

[John Donovan]

We did some additional testing using the latest Surfer 17 from Golden Software, checking that the interactive "slicing" features all still work and it looks like they work as expected.

Here for illustration, is a slice through a zircon showing the uranium distribution, first using traditional off-peak measurements:



And here is the same acquisition, but processed using the MAN background correction:



And you can see, the MAN processed data has better precision, and even more importantly for x-ray map acquisitions, acquiring this map would have taken only 1/2 the acquisition time, since one only needs to acquire an on-peak x-ray map when using MAN!

[Probeman]

I promised our EPMA students I would post this quant x-ray map they produced last quarter (better late than never I guess!):





15 keV, 30 nA, 100 ms per pixel

[John Donovan]

Here's a recent map our lab did on what I think is a granite.

Question to the geologists: I'm aware of what "graphic" granite looks like in hand specimen, but are the textures visible in these maps, an example of "micro graphic" granite?

[macosta]

I'd say so - there is a subtle difference probably not worth arguing over between a micrographic texture (properly microcrysts of skeletal quartz hosted in alkali feldspar or sodic plagioclase) and between granophyre  (angular intergrowths of quartz and alkali feldspar). Generally, micrographic textures are more well-defined than granophyre, but both are thought to indicate eutectic crystallization.

Personally, micrographic qz looks blebbier to me than does granophyric qz, so I would call it a granophyre.

The distinction is somewhat arbitrary though.

Here is a catalogue of graphic textures in thin section:
http://www.alexstrekeisen.it/english/pluto/graphic.php

Compared to granophyric textures in thin section:
http://www.alexstrekeisen.it/english/pluto/granophyric.php

[John Donovan]

Thanks Marisa!

Also, I love the technical term "blebbier".  And I mean that sincerely since apparently "blebbier" is found in a published textbook on Geology:

https://books.google.com/books?id=N7SWDwAAQBAJ&pg=PA175&lpg=PA175&dq=blebbier&source=bl&ots=kn38RpzLqb&sig=ACfU3U3kxm023Pj796RJ3H2RnyhJovPRWw&hl=en&sa=X&ved=2ahUKEwj5z4TT_tzpAhWwGTQIHbYGDmQQ6AEwBHoECAoQAQ#v=onepage&q=blebbier&f=false

Nice.   

[macosta]

It's actually a legitimate mineralogical descriptor: https://en.wikipedia.org/wiki/Bleb_(mineralogy)

[John Donovan]

It's actually a legitimate mineralogical descriptor: https://en.wikipedia.org/wiki/Bleb_(mineralogy)

Cool. And blebbier is the adjective!   

[Probeman]

Several people have asked for the link to our quantitative WDS x-ray mapping manuscript pre-print, so here it is:

http://www.minsocam.org/MSA/Ammin/AM_Preprints/7739DonovanPreprint.pdf

We are still waiting for proofs from Amer. Min., but hopefully it will be published in journal format in the next few months.

[John Donovan]

Recently Phil Orlandini reported a minor bug in CalcImage (which we immediately fixed) and in the data example he sent us we noticed that he ran the quant map using quite "mild" conditions, i.e., 20 msec per pixel and 20 nA beam current.

He said this was just a "quick and dirty" map as he wasn't "pushing" sensitivity, but he and I thought it would be instructive to share his results to show what one gets with such "mild" conditions. Here is the Surfer output where one can see the quite "grainy" appearance of the data:



With a pixel size of 679 x 372 pixels (times 20 msec) that's around 1.4 hours for this quant map.

In this 10 x 10 pixel extraction average we can see however, that the accuracy is still quite good, though the std err is still quite large given these "mild" conditions:



We just wanted to emphasize that while one can obtain good accuracy with these conditions, generally we should generally utilize higher beam currents to obtain better precision/sensitivity. This results in the same acquisition time, but improved statistics. Personally I often utilize 100 to 400 msec per pixel and at least 50 nA of beam current for some really nice quant x-ray maps.

[John Donovan]

Here is the public access link for our quantitative EPMA mapping paper published in 2021 in American Mineralogist:

http://www.minsocam.org/MSA/AmMin/Public_Access/2021_public/AM106P1717.pdf

[Julien]

If you want to know more about quantitative element mapping, and see a bunch of examples, feel free to watch my group meeting presentation available here: https://www.youtube.com/watch?v=A-IjFNRTofI.

If you want it without advertisement, you can watch the video on my laboratory website: https://highpressure.ethz.ch/facilities/analytical/electron-microscopy-lab.html

Best,

Julien

[John Donovan]

If you want to know more about quantitative element mapping, and see a bunch of examples, feel free to watch my group meeting presentation available here: https://www.youtube.com/watch?v=A-IjFNRTofI.

If you want it without advertisement, you can watch the video on my laboratory website: https://highpressure.ethz.ch/facilities/analytical/electron-microscopy-lab.html

I just watched this video and it is not only an excellent introduction to full pixel quantification of x-ray maps, but also a very complete treatment of advanced methods for pixel quantification (background, interferences, beam sensitivity, etc.).  I should add that this fully quantitative treatment of x-ray maps is only available in the Probe for EPMA software, other treatments using calibration curves are not fully quantitative.

Thank-you for posting this.   By the way, the full paper mentioned in the video is available here:

https://pages.uoregon.edu/donovan/download/Donovan_2021_Amer_Min_2021-7739.pdf

or here:

https://pubs.geoscienceworld.org/msa/ammin/article-abstract/106/11/1717/608663/Quantitative-WDS-compositional-mapping-using-the

This was described in the video, but I just want to emphasize that when performing pixel averaging to obtain improved precision, one should always utilize the standard error as opposed to standard deviation as Julien described. This is because the standard deviation only describes the precision of a single point (or pixel), while the standard error describes the precision of the average of the points (or pixels)!  In other words, the standard deviation does not improve as one increases the number of pixels averaged!  In CalcImage, one has the choice, but the standard error is the default option.