High Resolution Stage Mapping

[John Donovan]

Attached is a scan using Probe Image on my Sx100 with 200 nm pixels on a relatively vertical interface on a Cu-Al eutectic alloy (Cu-Al adjacent to Al).  Remember that one needs to be logged in to see attachments!

As you can see the Sx100 linear encoders provides excellent reproducibility on the 32 scan lines.

Note the band of totals over 100% at the interface in the Total wt% image.  This is due to the fact that the electron interaction volume in creating emissions from both phases resulting in a large matrix correction error for Al ka by Cu.  For further information see Julie Barkman's M&M poster "Electron Probe Quantitative Mapping vs. Defocused Beam Analysis" which is attached below.

[John Donovan]

However, no stage is mechanically perfect.

The following images show the stage reproducibility of running two maps on exactly the same area using exactly the same conditions for the BSE image.

All replicate maps were 256 x 256 using 0.5 um stage steps and 20, 150 and/or 400 msec dwell per pixel. ImageJ was used to create the animated GIFs using the Stack menu.

Remember, Cameca stages scan in the X axis, JEOL stages scan in the Y axis!

The worst example, first attachment below was Wash. U.'s 8200 which was greatly improved by having the service engineer in to tighten the stage mechanicals.

After the stage was repaired, the improved reproducibility can be seen in the second attachment below.

[John Donovan]

Here's a pair of replicate stage maps from the UofO SX100 using 0.5 um pixels.

[John Donovan]

Here is a replicate (4 maps?) stage scan from the JEOL 8900 at the USGS.

[John Donovan]

Here are some from the Sx100 in Tasmania.

[Ben Buse]

A few of maps of plagioclase I acquired using Probe Image whilst testing the instrument. Maps plotted in imagej. Taken at 7kV, step size 0.3um. 40nA, bit noisy as I didn't have to long, dwell times 50ms for 512 image, 20 ms for 756 image and 10ms for 1024 image. Maps are for Na

[John Donovan]

A few of maps of plagioclase I acquired using Probe Image whilst testing the instrument. Maps plotted in imagej. Taken at 7kV, step size 0.3um. 40nA, bit noisy as I didn't have to long, dwell times 50ms for 512 image, 20 ms for 756 image and 10ms for 1024 image. Maps are for Na

Hi Ben,
I like the middle one the best... very pretty.

Can't wait for you to quant them!

When you do quant them, try the log weight % output which will suppress the Na "hot spots" and bring out the Na zoning details in the plag as seen here:

http://probesoftware.com/smf/index.php?topic=107.msg402#msg402
john

[Ben Buse]

Hi John

Here's the quant one - plus as apfu which is great! I've interpolate pixels in surfer reducing noise. And Ca is on two spectrometers in view of its low overvoltage ratio.





Ben

[Les Moore]

Hi guys,  lovely images.  Been struglling down this path for a while.
Registration issues between passes are a nightmare when you have some regions that charge.
Stage reproducibility is not the issue, beam deflection is...
It is local and mucks up the registration locally; this means that some parts wil register and others not.

It gets back to what you really need for registration (make sure these are on the same pass)
What you need to use to achieve the yield you need.
I've been trying to look for trace element changes in thin films at low kV and correlate trace element variations.
5kV, 20-40nA etc etc.
Secret stuff so can't say much more apart from I occasionally wish I had more spectrometers around the column to overcome these microregistration issues. 

[Probeman]

Hi guys,  lovely images.  Been struglling down this path for a while.
Registration issues between passes are a nightmare when you have some regions that charge.
Stage reproducibility is not the issue, beam deflection is...
It is local and mucks up the registration locally; this means that some parts wil register and others not.

It gets back to what you really need for registration (make sure these are on the same pass)
What you need to use to achieve the yield you need.
I've been trying to look for trace element changes in thin films at low kV and correlate trace element variations.
5kV, 20-40nA etc etc.
Secret stuff so can't say much more apart from I occasionally wish I had more spectrometers around the column to overcome these microregistration issues.

Hi Les,
One thing that works well for multiple spectrometer passes which have stage (or beam) drift, is to "align and crop" the images. 

CalcImage now has a very easy to use feature which allows one to utilize any map as a base or reference (but usually one would use the BSE or SEI map) and then using the mouse or cursor keys, align a single map (again usually the BSE or SEI map) from each subsequent spectrometer pass acquisition, to create a table of "offsets" for each spectrometer pass, which the software then uses to automatically crop all the maps in those multiple passes to the same size and registration:

http://probesoftware.com/smf/index.php?topic=41.msg5752#msg5752

It works way better than I thought it would and yes, I did have help with the code!   
john