News:

:) If you are a member, please feel free to add your website URL to your forum profile

Main Menu

New Features in CalcImage

Started by John Donovan, August 20, 2013, 09:46:19 AM

Previous topic - Next topic

John Donovan

Did you know that you can classify your quant maps in CalcImage using any of the available output types (elemental or quant, oxide, atomic, k-ratio, net intensities, and formula basis)?
John J. Donovan, Pres. 
(541) 343-3400

"Not Absolutely Certain, Yet Reliable"

John Donovan

#1
Here's an example of calculating excess oxygen in magnetite as oxide formulas...
John J. Donovan, Pres. 
(541) 343-3400

"Not Absolutely Certain, Yet Reliable"

John Donovan

#2
Here's a cool feature that I implemented recently in CalcImage. It's called correlated pixel quantification (CPQ) and the idea is that instead of using the point intensities of standards acquired in the PFE MDB file, one instead acquires standard intensities as x-ray maps, exactly the same as the unknown maps.

Of course one can acquire the standard x-ray maps with different beam currents or pixel dwell times since the program will normalize for those parameters, but depending on what one is trying to correct for this may or may not be desired.

This method was developed for Philippe Pinard and Silvia Richter at Aachen for beam line scans of carbon in steel where the buildup of carbon around the point of acquisition causes the "baseline" of carbon to vary considerably over the length of a beam line scan as seen here:

http://smf.probesoftware.com/index.php?topic=48.msg160#msg160

The idea is that if one acquires the carbon background as an x-ray map with the same acquisition parameters as the unknown x-ray map, the carbon "baseline" variation will be normalized out. Of course there are many other parameters that will affect the rate of carbon buildup such as thermal conductivity and specimen vacuum, but it seems worth trying.

In fact, one can use a mix of MDB standard intensities and simply replace selected standard intensities as required with x-ray maps on standards for on-peak, background or interference standard intensities.

I decided to test this method by acquiring a low mag beam scan map on a pyrite grain and also x-ray maps with the same parameters on the standard to "normalize" out the effects of Bragg defocusing.  As one can see from the attached maps it works quite well although the imprecision in areas where the Bragg defocusing is severe, shows slightly greater errors.
John J. Donovan, Pres. 
(541) 343-3400

"Not Absolutely Certain, Yet Reliable"

John Donovan

#3
Here's a new feature "hot off the press" in CalcImage which now allows one to specify the calculation of detection limit and analytical sensitivity maps from quant x-ray maps.

See attached for an example.  Pretty cool! (if I do say so myself!).  ;D
John J. Donovan, Pres. 
(541) 343-3400

"Not Absolutely Certain, Yet Reliable"

Julien

Pretty cool indeed! Can you give us the analytical conditions?

John Donovan

Quote from: Julien Allaz on September 06, 2013, 07:51:00 PM
Pretty cool indeed! Can you give us the analytical conditions?

Julien is correct- I accidently used the std count times instead of the x-ray map pixel dwell times for the detection limit and analytical sensitivity calculation. All fixed now! Thanks, Julien!

The conditions were pretty normal: 15 keV, 30 nA, 0.2 seconds per pixel.
John J. Donovan, Pres. 
(541) 343-3400

"Not Absolutely Certain, Yet Reliable"

John Donovan

#6
And here's an example of the 1 sigma analytical sensitivity output from CalcImage. Note that output is skipped (pixel value set to blanking value) if the concentration is less than 1 wt%.  This threshold could be adjusted if necessary (user specified?), but this seems a reasonable default.

The blanked values are shown as "gray" to distinguish them from high precision pixels (idea from Julie Barkman).
John J. Donovan, Pres. 
(541) 343-3400

"Not Absolutely Certain, Yet Reliable"

John Donovan

OK, here is the mineral end-member quant mapping as requested.

By the way, I doubt that this mineral grain is actually a pyroxene, instead it is very probably an amphibole, but it suffices to demonstrate the method.

Note that pixels which cannot be calculated are "blanked" and appear gray. Please let me know how this works for your lab.
John J. Donovan, Pres. 
(541) 343-3400

"Not Absolutely Certain, Yet Reliable"

John Donovan

#8
OK, here's a honest to goodness igneous feldspar grain with some interesting zoning, both oxide and end-member basis.

Here is where one can specify all calculation output options in CalcImage (note that elemental quant data is *always* output automatically):

John J. Donovan, Pres. 
(541) 343-3400

"Not Absolutely Certain, Yet Reliable"

John Donovan

Oh, almost forgot, here's the formula maps which of course can be calculated on the basis of any element including the sum of cations.
John J. Donovan, Pres. 
(541) 343-3400

"Not Absolutely Certain, Yet Reliable"

John Donovan

I should also mention that one can also calculate the stoichiometric oxygen image and the total (sum) image.
John J. Donovan, Pres. 
(541) 343-3400

"Not Absolutely Certain, Yet Reliable"

John Donovan

#11
This is another output option in CalcImage that I wasn't sure would be useful, but if one is interested in a wide range of concentrations, the log wt% option seems to be useful as seen in these attachments comparing the wt% to log wt% for a banded quartz samples (with nice CL correlation but not from Ti).

But because of a number of higher concentration trace element mineral phases in the sample, the beautiful banding is not visible except in the log wt% output.

Note: I output these originally using the 4 plots per page CalcImage generated script and then combined them into a 5 plot per page by opening the automatically saved SRF files in Surfer and mulching the plot and titles around. To make the second plot I simply saved it to a different name and then just loaded the log wt% GRD files.

Easy!
John J. Donovan, Pres. 
(541) 343-3400

"Not Absolutely Certain, Yet Reliable"

Gareth D Hatton

I have used the log plot capability in CalcImage for looking at poisons in catalyst samples, they show up really well.
This is a fantastic addition that saves adjusting low concentration maps by hand in surfer.   :)

John Donovan

#13
Did you know that you can also perform phase classification in CalcImage using lists of quantitative analysis points from Probe for EPMA using this menu?



Then simply import the export file into CalcImage using this menu:



The resulting phase numbers are listed in the NK column. Furthermore you can "pre-process" quantitative image data from CalcImage in this window to extract ranges based on the sum of each pixel (to remove pixels of epoxy, etc) as seen here:

John J. Donovan, Pres. 
(541) 343-3400

"Not Absolutely Certain, Yet Reliable"

John Donovan

These are three small but nice features recently added to CalcImage:

1. If you hover the mouse cursor over an image loaded into CalcImage, it will popup a text field that displays the file name, x/y min/max and the image width/height in stage units.



2. If you load a quant dataset into the Classify Image window, the image size is displayed in microns at the top of the classified image.



3. In addition, the image width and height (in microns) are now output to the Surfer plots in the axis labels.

John J. Donovan, Pres. 
(541) 343-3400

"Not Absolutely Certain, Yet Reliable"