XTREME Database

[John Donovan]

We all know that the Pouchou k-ratio database has been of enormous value in the evaluation of various matrix correction models, both analytical and Monte-Carlo derived methods. However, this database could be updated in several ways. For example there is no facility for handling compounds with more than 2 elements. Also, the standards in the Pouchou database are always assumed to be pure elements which greatly restricts some comparisons.

I would like to propose a project for all of us, that is to begin the creation of a new k-ratio data base that we might call the XTREME database. That is:

X-ray
Table of
Ratios with
Exceptional
Matrix
Effects

Another suggestion would be to include a text field for "attribution", that is, the person and lab that made those measurements. If we allow compound materials, we will need fields to include the standard compositions, so that std k-factors can be calculated as well.

The idea being that we would focus on materials and beam voltages that show "extreme" matrix corrections. If our models can handle these "extreme" situations, they can handle normal situations mostly likely also. 

What do you all think about this? Please chime in with ideas, problems, suggestions, whatever.

Also Probe Software will be an exhibitor at AGU next month so feel free to drop by our booth and let us know what you think.

[pgopon]

I think this is a great idea.  One problem I foresee however, is that in some of the more EXTREME cases our models will never be able to do the physics justice.  I look forward to talking about this at AGU with you.

phil

[Probeman]

Hi Phil,
I agree- but "failure modes" can still be instructive!

I've attached a text file that contains the results of running all my Penepma calculations through my binary alpha factor database looking for "xtreme" absorption or fluorescence cases. It is true that the fluorescence model is analytical using Salvat and Llovet's Fanal FORTRAN, where both the beam incident and boundary materials are the same... See:

http://probesoftware.com/smf/index.php?topic=47.0

but the absorption and atomic number effects are modeled using the full Penepma Monte-Carlo code (Penfluor) at 11 different beam energies. Then I filtered those for between 5 and 30 keV and ZAF corrections greater than 8 (800% absorption) or less then 0.6 (40% fluorescence) and those are attached.

A few selections from the text file show that there are some interesting cases that could be empirically measured for comparison as seen here:

Warning: very large matrix correction (ZAF= 6.12903, K= 1.63158, C=10) for Mg ka in Zn at 26 keV : 12-30_40.txt...
Warning: very large matrix correction (ZAF= 6.42811, K= .777833, C=5) for Mg ka in Zn at 26 keV : 12-30_40.txt...
Warning: very large matrix correction (ZAF= 6.63289, K= .150764, C=1) for Mg ka in Zn at 26 keV : 12-30_40.txt...
Warning: very large matrix correction (ZAF= 6.44948, K= 1.55051, C=10) for Mg ka in Zn at 27 keV : 12-30_40.txt...
Warning: very large matrix correction (ZAF= 6.15973, K= .162345, C=1) for Al ka in Ge at 30 keV : 13-32_40.txt...

Warning: very large matrix correction (ZAF= .580743, K= 1.72193, C=1) for Si ka in Pb at 5 keV : 14-82_40.txt...
Warning: very large matrix correction (ZAF= .598947, K= 1.66960, C=1) for Si ka in U  at 5 keV : 14-92_40.txt...
Warning: very large matrix correction (ZAF= .550399, K= 1.81687, C=1) for Ti kb in Fe at 7 keV : 22-26_40.txt...
Warning: very large matrix correction (ZAF= .392112, K= 2.55029, C=1) for Cr ka in Mn at 7 keV : 24-25_40.txt...
Warning: very large matrix correction (ZAF= .377312, K= 2.65033, C=1) for Cr kb in Mn at 7 keV : 24-25_40.txt...
Warning: very large matrix correction (ZAF= .489611, K= 2.04244, C=1) for Cr ka in Mn at 8 keV : 24-25_40.txt...
Warning: very large matrix correction (ZAF= .475308, K= 2.10390, C=1) for Cr kb in Mn at 8 keV : 24-25_40.txt...
Warning: very large matrix correction (ZAF= .567872, K= 1.76096, C=1) for Cr ka in Mn at 9 keV : 24-25_40.txt...
Warning: very large matrix correction (ZAF= .555012, K= 1.80176, C=1) for Cr kb in Mn at 9 keV : 24-25_40.txt...
Warning: very large matrix correction (ZAF= .369880, K= 2.70358, C=1) for Cr ka in Fe at 7 keV : 24-26_40.txt...
Warning: very large matrix correction (ZAF= .576173, K= 8.67795, C=5) for As ka in Pb at 13 keV : 33-82_40.txt...
Warning: very large matrix correction (ZAF= .561766, K= 1.78010, C=1) for As ka in Pb at 13 keV : 33-82_40.txt...

Lots more where these came from!

So, if you have a binary compound of any of these element pairs, but don't have one of these specific (1:99, 5:95, 90:10, 8-:20, 60:40, 50:50, etc.) stoichiometries, say 50:50 NaF (alas, because although I have already calculated a 50:50 binary, it's 50:50 by weight!), just let me know and I'll calculate the Penepma k-ratio for your actual specimen, and we'll have an empirical comparison!

[Probeman]

Here's a more recent summary of binary compositions with large matrix corrections after scanning the Penfluor/Fanal calculations from the latest matrix.mdb databse (see attached below).
john