New Features/Improvements in Standard/CalcZAF

[Probeman]

Gareth Seward has been working steadily over the summer and has got some of our graphics converted over to the new Pro Essentials graphics library.

See the following attachments (you must be logged in) and please note the new zoom graphics capabilities now available...

[UofO EPMA Lab]

I notice that one needs to run the CalcZAF.msi installer again in order to get the updated example (NISTBIN, Pouchou, etc) binary k-ratio input files for CalcZAF error distribution calculations, now that additional x-ray lines are supported.  Basically the "absorber only" integer flag is now 13 rather than 7.  These files are updated:

Directory of C:\ProgramData\Probe Software\Probe for EPMA

09/27/2015  04:27 PM               276 auagcu.dat
09/27/2015  04:27 PM             1,102 auagcu2.dat
09/27/2015  04:28 PM           132,323 nistbin.dat
09/27/2015  04:32 PM            21,892 nistbin2.dat
09/27/2015  04:32 PM            19,350 nistbin3.dat
09/27/2015  04:32 PM            37,870 nistbina20.dat
09/27/2015  04:32 PM            26,412 nistbinz10.dat
09/27/2015  04:33 PM            63,504 pouchou.dat
09/27/2015  04:34 PM           219,716 pouchou2.dat
09/27/2015  04:36 PM             6,088 Pouchou2_Au,Cu,Ag_only.dat
09/27/2015  04:37 PM             8,855 pouchoua20.dat
09/27/2015  04:38 PM             8,274 pouchouz10.dat

[Ben Buse]

Hi,

I'm having some problems with calczaf v11.0.8 using the MAC tables.

(1) Typed emitter pair C Ka Ag fine,
(2) then typed emitter pair C Ka Au - subscript out of range,
(3) then selected MAC table - file already open.
(4) Select MAC table again and it works.

Actually I think it boils down to why is C kA Au emitter/absorber pair not in the mac pairs - although its present when viewed as a table?

Thanks

Ben

[Probeman]

Hi,

I'm having some problems with calczaf v11.0.8 using the MAC tables.

(1) Typed emitter pair C Ka Ag fine,
(2) then typed emitter pair C Ka Au - subscript out of range,
(3) then selected MAC table - file already open.
(4) Select MAC table again and it works.

Actually I think it boils down to why is C kA Au emitter/absorber pair not in the mac pairs - although its present when viewed as a table?

Thanks

Ben

Oh... my... gosh... an actual bug!     

Thanks Ben, I apparently broke this MAC emitter-absorber pair code when I when I added support for the Ln, Lg, Lv, Ll, Mg and Mz lines.   I already fixed it and I will upload a new version tonight.  Thanks again!

MAC value for C ka in Au =   15151.17  (LINEMU   Henke (LBL, 1985) < 10KeV / CITZMU > 10KeV)
MAC value for C ka in Au =   15210.00  (CITZMU   Heinrich (1966) and Henke and Ebisu (1974))
MAC value for C ka in Au =        .00  (MCMASTER McMaster (LLL, 1969) (modified by Rivers))
MAC value for C ka in Au =   16876.73  (MAC30    Heinrich (Fit to Goldstein tables, 1987))
MAC value for C ka in Au =        .00  (MACJTA   Armstrong (FRAME equations, 1992))
MAC value for C ka in Au =   16642.76  (FFAST    Chantler (NIST v 2.1, 2005))

A very nasty absorption correction...

[Ben Buse]

Thanks John, yes so I'm going to use silver instead

Ben

[Probeman]

Thanks John, yes so I'm going to use silver instead

Ben

Hi Ben,
There actually is an old post on this exact issue:

http://probesoftware.com/smf/index.php?topic=48.msg959#msg959

john

[John Donovan]

Thanks John, yes so I'm going to use silver instead

Ben

Hi Ben,
The latest version (11.0.9) fixes this MAC pair display menu issue in CalcZAF.
john

[John Donovan]

Here is the new Penepma12 GUI in Standard, showing a log plot of HfSiO4 at 15 keV (modeled down to 400 eV):

[Probeman]

I notice that one needs to run the CalcZAF.msi installer again in order to get the updated example (NISTBIN, Pouchou, etc) binary k-ratio input files for CalcZAF error distribution calculations, now that additional x-ray lines are supported.  Basically the "absorber only" integer flag is now 13 rather than 7.  These files are updated:

Directory of C:\ProgramData\Probe Software\Probe for EPMA

09/27/2015  04:27 PM               276 auagcu.dat
09/27/2015  04:27 PM             1,102 auagcu2.dat
09/27/2015  04:28 PM           132,323 nistbin.dat
09/27/2015  04:32 PM            21,892 nistbin2.dat
09/27/2015  04:32 PM            19,350 nistbin3.dat
09/27/2015  04:32 PM            37,870 nistbina20.dat
09/27/2015  04:32 PM            26,412 nistbinz10.dat
09/27/2015  04:33 PM            63,504 pouchou.dat
09/27/2015  04:34 PM           219,716 pouchou2.dat
09/27/2015  04:36 PM             6,088 Pouchou2_Au,Cu,Ag_only.dat
09/27/2015  04:37 PM             8,855 pouchoua20.dat
09/27/2015  04:38 PM             8,274 pouchouz10.dat

I'm attaching the Pouchou2.dat CalcZAF input file, without B Ka and Cu La k-ratios, for matrix correction evaluations below. The file format for these CalcZAF binary k-ratio input files are described here:

An example of the file format is seen here:

       79     29     2    5    15.     52.5    .8015   .1983   .7400   .0
       79     29     2    5    15.     52.5    .6036   .3964   .5110   .0
       79     29     2    5    15.     52.5    .4010   .5992   .3120   .0
       79     29     2    5    15.     52.5    .2012   .7985   .1450   .0

The data file format assumes one line for each binary. The first two columns are the atomic numbers of the two binary components to be calculated. The second two columns are the x-ray lines to use ( 1 = ka, 2 = kb, 3 = la, 4 = lb, 5 = ma, 6 = mb, 7 = Ln, 8 = Lg, 9 = Lv 10 = Ll, 11 = Mg, 12 = Mz, 13 = specified, that is, do not calculate the intensity). The next two columns are the operating voltage and take-off angle. The next two columns are the weight fractions of the binary components. The last two columns contains the k-exp values for calculation of k-calc/k-exp. If the second element x-ray line is "specified" (iray = "13"), then no experimental k-ratio value is required.

Note that although the k-ratio file format example above has two emitting elements (so that two k-ratio error sets will be calculated), only one emitting element per line is actually required, as in the following example (where the x-ray line of "13" in the fourth column indicates an "absorber only", specifically an element by difference from 1.0):

       13    26     1    13     20     52.5    0.241     .0    0.124    .0
       13    26     1    13     25     52.5    0.241     .0    0.098    .0
       13    26     1    13     30     52.5    0.241     .0    0.083    .0
       26    13     1    13     20     52.5    0.759     .0    0.736    .0
       26    13     1    13     25     52.5    0.759     .0    0.742    .0
       26    13     1    13     30     52.5    0.759     .0    0.748    .0

[Probeman]

Here's the sexy new histogram and concentration vs error plots in CalcZAF (see attached because we're moving to our own image server and I can't stand to use TinyPic any longer!).

[John Donovan]

I added the spectrometer position equations as graphics to the latest version of CalcZAF:

[Probeman]

Here is the calculate temperature rise dialog in CalcZAF (see Run menu). It is based on electron dose and thermal conductivity.



You can download the latest CalcZAF here:

http://www.probesoftware.com/Technical.html

john

[John Donovan]

All,
Because correction of fluorescence by (and for) beta emission lines is now the default in CalcZAF (and Probe for EPMA), some people with older versions of CalcZAF or Probe for EPMA have reported getting warnings that some absorption edges are missing from the absorption edge table (XEDGE.DAT).

These tables were updated around the time that I implemented this more rigorous fluorescence correction, but the CalcZAF.msi installer update from the Help menu *does not* overwrite these existing .DAT files.  Why? Because some people have edited these .DAT files manually and don't want them updated...

If you are seeing such warnings in the CalcZAF or Probe for EPMA log window such as these:

WARNING in ZAFFlu2- K  l3 absorption edge is zero
WARNING in ZAFFlu2- Mn l2 absorption edge is zero
WARNING in ZAFFlu2- Ag m3 absorption edge is zero
WARNING in ZAFFlu2- Cd m3 absorption edge is zero
WARNING in ZAFFlu2- Cd m5 absorption edge is zero
WARNING in ZAFFlu2- Cs m3 absorption edge is zero
WARNING in ZAFFlu2- Mo m3 absorption edge is zero
WARNING in ZAFFlu2- Pd m3 absorption edge is zero
WARNING in ZAFFlu2- Rh m3 absorption edge is zero
WARNING in ZAFFlu2- Ru m3 absorption edge is zero
WARNING in ZAFFlu2- Sn m3 absorption edge is zero


first of all, don't panic. These secondary fluorescence effects from beta lines are trivial in almost all instances and the warnings can be ignored.  Second, if you want you can download the .DAT files attached below and overwrite your existing .DAT files that are in the

C:\ProgramData\Probe Software\Probe for EPMA

folder for Win 7 and later, or the

C:\Documents and Settings\All Users\Application Data\Probe Software\Probe for EPMA

folder for XP and older.  Please note that by default Microsoft makes these folders invisible for some silly reason- so you'll have to go into the folder options tab and make them visible (yet the application folder which is not writable, is visible!).

Replacing your existing .DAT files with these newer files will fix the warnings about missing absorption edges. Please let me know if you have any trouble at all and I (or your Probe Software specialist) will be happy to assist you.
john

[John Donovan]

I changed the maximum allowable incident beam energy in CalcZAF and Standard from 50 keV, to 100 keV.  Why? Because Paul Carpenter asked why not!

I am not quite sure why that limitation was there to begin with, but I think the analytical expressions will continue to perform OK at these higher beam energies (because some physics gets simpler at very high energies?). Do I have any empirical evidence to support that guess? No.  But if someone has any empirical measurements in bulk materials at energies above 50 keV, please feel free to share them here with us.

As for the Penepma Monte Carlo calculations in Standard, the Penepma 2012 full spectrum simulations should be fine, with the possible exception of the boundary fluorescence simulations using the Penfluor/Fanal codes, which by default only simulates electron beam energies between 5 keV and 50 keV.  So again, if anyone has synthetic boundary (couples) in bulk samples, please share them here so we can test the codes at these higher beam energies.

The original boundary fluorescence paper here:

http://epmalab.uoregon.edu/publ/Llovet,%20et%20al.,%20Secondary%20fluorescence%20in%20electron%20probe%20microanalysis%20of%20couple%20materials.pdf

noted some divergence in the continuum fluorescence at higher beam energies, so that is one possible area of concern.  So for example, measurements of an Al or Mg couple adjacent to Zn or Cu would be very interesting to compare to the Monte Carlo calculations (since this is pure continuum fluorescence situation).
john

[Paul Carpenter]

John,
Thank you for making this change, as it really was an artificial limitation in the code. Current microprobe and SEM instruments are probably limited to operating voltages less than 50 kV (I know that on the JEOL the HV system can indeed run at 50 but the system is now limited to 40 kV I think, to reduce the possibility of arcing in the HV system).

X-ray photon absorption knows no boundaries. X-ray fluorescence spectrometry can be performed on instruments up to ~ 70 kV (WDS systems). The absorption correction would use the same code we use for EPMA with the same mass absorption coefficients being applicable. High energy electron beams are used for various procedures up to MeV range, and clearly there a different set of macs would apply.

The SDD X-ray spectrometer has decreasing efficiency above ~15 kV but in principle one could analyze K-lines using the EDS detector for systems where analysis using L or M lines is problematic; this requires analysis at higher accelerating voltage.

A point worth considering is that almost none of the macs we use were actually measured, they are fitted values. So the "range" of applicability does not exist. Equally, low kV analysis is a regime for which the correction algorithms were not developed but that does not prohibit their use, and indeed the phi-rho-z algorithm works well with most attention on the accuracy of mac values at low kV.

Paul