Author Topic: Utilizing an element by difference for quantitative trace element mapping  (Read 380 times)

John Donovan

  • Administrator
  • Emeritus
  • *****
  • Posts: 2487
  • Other duties as assigned...
    • Probe Software
If we are interested in mapping trace elements in relatively simply matrices (quartz, zircon, rutile, pyrite, etc), we can utilize the element (or formula) by difference option to allow the quantification in CalcImage (or Probe for EPMA) to perform a rigorous matrix correction for the trace elements. That is unless we willing to "burn" a spectrometer just to measure Si (and any other matrix elements).  In the case of say, Ti Ka in quartz, the matrix correction is approximately 20% at 15 keV, so it is important that even trace elements be matrix corrected for proper quantification.

The element (or formula) by difference calculation is the same for both CalcImage and Probe for EPMA, though the Calculation Options dialog in CalcImage is somewhat simplified compared to PFE:

Once we have specified Si as the element by difference (and the calculation of stoichiometric oxygen for SiO2), one may also specify (if desired), to output a map for the element by difference:

When the quantification is complete one obtains the following output using the oxide output maps (thank-you Ben Buse for noting the oxide element by difference output issue):

Note however that even if the Calculate "Totals" Image option is checked, CalcImage (and Surfer) will not display the totals image by default.  This is because when we are calculating an element (or formula) by difference, the totals image is pretty uninteresting. Or is it?

Here we've loaded the totals image by browsing to it using the File | Open GRD File menu in CalcImage:

The upper left image is the totals image, and using the mouse cursor, the display always shows 100.0000 wt. % for every pixel.  So what is going on here?  If we next open the classify .DAT output file in Excel we can immediately see the issue:

Most pixels are 100 wt. %, but some are 99.99999 wt. %, and in fact a few are 99.99998 wt. %.  Well that's math on computers for you!    ::)
« Last Edit: December 19, 2018, 12:06:06 pm by John Donovan »
John J. Donovan, Pres. 
(541) 343-3400

"Not Absolutely Certain, Yet Reliable"