Probe Software Users Forum
Software => CalcZAF and Standard => Topic started by: Les Moore on November 27, 2014, 12:50:44 PM
-
Hi John,
Perhaps you could check my value for S in C graphite. I came up with 1.177.
A couple of questions
1. When you quote alpha 1, 2 & 3, just what is the form of the expression these feed into (dumb bunny question)
2. To quote so many sig figures, do you use the uncertainty figure from the PENEPMA output or...
3. Do you use the sig error from the regression of various concentrations' Alpha factors.
Cheers
Les
-
Perhaps you could check my value for S in C graphite. I came up with 1.177.
Hi Les,
Can you let me know what the S-C ratio is and at what beam energy you are using (I'll assume 40 degrees takeoff!)? Remember though, if it's a trace element concentration the background correction is *much* more important than the matrix correction.
1. When you quote alpha 1, 2 & 3, just what is the form of the expression these feed into (dumb bunny question)
No problem. Attached below is an abstract just submitted for AMAS 2015. You might also want to check this link for a historical background:
http://epmalab.uoregon.edu/bence.htm
2. To quote so many sig figures, do you use the uncertainty figure from the PENEPMA output or...
3. Do you use the sig error from the regression of various concentrations' Alpha factors.
I am not propagating the Penepma uncertainties at the moment. The reason is that during the calculation of the 11 binary compositions one can't know in advance the composition of the emitter in question and therefore the error propagation gets complicated. In fact I'm not exactly sure how to do it!
-
Hi John,
I ran 0, 1, 2, 3, 4 & 5% S in graphite with a density of 2.2 (I think). I then regressed it to get an even simpler linear co-efficient = 1.177 * Ks.
I used 5kV as I was after high spatial resolution and didn't want a huge interaction volume considering the very light element substrate.
I understand the issue wrt accuracy is with the background but I figure as the graphite is 90+% C, the background variations should be minor unless..... the S changes the structure and density.
This appears to be the case as the BSE image shows a distinct change in contrast.
As always, a simple job is never so.
O'Reillies Law "Murphy was an optimist"
-
Hi Les,
OK on the 5 keV, but density is of course irrelevant for bulk analysis.
Are you trying to create a calibration curve? I haven't calculated the S ka in C binaries using Penepma Monte-Carlo yet (though it would only take a weekend), but using the Analytical | Calculate and Plot Binary Alpha Factors menu here is the alpha versus concentration plot using CalcZAF defaults:
(https://probesoftware.com/smf/oldpics/i59.tinypic.com/2jfwqb4.jpg)
It would appear that the correction is very small. Essentially S intensity (elemental k-ratio) equals S concentration. Maybe I don't understand the question?
john
-
Hi Les,
Ok, the C-S binary run in Penepma finished the first composition which just happens to be one of your compositions, that is, 1 percent sulfur in carbon (skill not luck!). ;)
(https://probesoftware.com/smf/oldpics/i62.tinypic.com/10ng5f5.jpg)
If I read your questions correctly, as you can see, both CalcZAF and Penepma agree closely (the red and green symbols), the matrix corrections are a relative non-issue (CalcZAF = 1.08, Penepma = 1.07) at 15 keV. I attach a detailed Excel output for the calculation below.
Now at 5 keV, the situation is... what just a darn tootin' minute! This is interesting: the matrix correction for 1% S in C is quite a bit larger at 5 keV than 15 keV, quite unintuitive! See here:
(https://probesoftware.com/smf/oldpics/i58.tinypic.com/2ugkih4.jpg)
I attach the 5 keV Excel sheet below also. I have to admit, this is surprising. If we just look at CalcZAF here are the 15 keV data:
ELEMENT ABSCOR FLUCOR ZEDCOR ZAFCOR STP-POW BKS-COR F(x)u Ec Eo/Ec MACs
S ka .9851 1.0000 1.0968 1.0805 1.1746 .9338 .9451 2.4720 6.0680 197.942
C ka 1.1108 1.0000 .9990 1.1096 .9984 1.0006 .4653 .2838 52.8467 2791.77
ELEMENT K-RAW K-VALUE ELEMWT% OXIDWT% ATOMIC% FORMULA KILOVOL
S ka .00000 .00925 1.000 ----- .377 .004 15.00
C ka .00000 .89219 99.000 ----- 99.623 .996 15.00
TOTAL: 100.000 ----- 100.000 1.000
And here is the same composition at 5 keV:
ELEMENT ABSCOR FLUCOR ZEDCOR ZAFCOR STP-POW BKS-COR F(x)u Ec Eo/Ec MACs
S ka .9979 1.0000 1.1684 1.1659 1.2104 .9653 .9931 2.4720 2.0227 197.942
C ka 1.0222 1.0000 .9986 1.0208 .9980 1.0006 .8693 .2838 17.6156 2791.77
ELEMENT K-RAW K-VALUE ELEMWT% OXIDWT% ATOMIC% FORMULA KILOVOL
S ka .00000 .00858 1.000 ----- .377 .004 5.00
C ka .00000 .96980 99.000 ----- 99.623 .996 5.00
TOTAL: 100.000 ----- 100.000 1.000
The matrix correction for S Ka in carbon is twice as large at 5 keV than 15 keV. As you are probably aware, usually the lower the beam energy the smaller the matrix correction since the average emitted x-ray absorption path length is shorter at lower beam energies. For example, here is 1% Si in Al at 15 keV:
ELEMENT ABSCOR FLUCOR ZEDCOR ZAFCOR STP-POW BKS-COR F(x)u Ec Eo/Ec MACs
Si ka 1.9811 1.0000 .9738 1.9291 .9811 .9925 .4571 1.8390 8.1566 3184.61
Al ka 1.0004 .9997 1.0003 1.0003 1.0002 1.0001 .8870 1.5600 9.6154 403.602
ELEMENT K-RAW K-VALUE ELEMWT% OXIDWT% ATOMIC% FORMULA KILOVOL
Si ka .00000 .00518 1.000 ----- .961 .010 15.00
Al ka .00000 .98971 99.000 ----- 99.039 .990 15.00
TOTAL: 100.000 ----- 100.000 1.000
Note the large matrix correction for Si Ka. And now at 5 keV, the Si Ka matrix correction is almost half as seen here:
ELEMENT ABSCOR FLUCOR ZEDCOR ZAFCOR STP-POW BKS-COR F(x)u Ec Eo/Ec MACs
Si ka 1.1170 1.0000 .9786 1.0931 .9837 .9948 .8832 1.8390 2.7189 3184.61
Al ka 1.0001 .9998 1.0002 1.0001 1.0002 1.0001 .9832 1.5600 3.2051 403.602
ELEMENT K-RAW K-VALUE ELEMWT% OXIDWT% ATOMIC% FORMULA KILOVOL
Si ka .00000 .00915 1.000 ----- .961 .010 5.00
Al ka .00000 .98993 99.000 ----- 99.039 .990 5.00
TOTAL: 100.000 ----- 100.000 1.000
Very interesting- I did not see that coming!
On the other hand what about time dependent intensity effects (TDI), such as losing sulfur during the acquisition? I've never tried analyzing sulfur in graphite, so I have no idea...
-
In looking at another binary similar to measuring sulfur in carbon, for example oxygen in MgO, we however *do not* see the same effect of the absorption increasing as the absorption path length is decreased at lower beam energies.
For example, here is MgO at 15 keV:
ELEMENT ABSCOR FLUCOR ZEDCOR ZAFCOR STP-POW BKS-COR F(x)u Ec Eo/Ec MACs
Mg ka 1.2458 1.0000 1.0211 1.2722 1.0337 .9879 .6955 1.3050 11.4943 1298.79
O ka 1.7616 .9981 .9663 1.6990 .9516 1.0154 .3978 .5317 28.2114 3603.22
ELEMENT K-RAW K-VALUE ELEMWT% OXIDWT% ATOMIC% FORMULA KILOVOL
Mg ka .00000 .47401 60.303 ----- 50.000 1.000 15.00
O ka .00000 .23366 39.697 ----- 50.000 1.000 15.00
TOTAL: 100.000 ----- 100.000 2.000
And here it is again but at 5 keV:
ELEMENT ABSCOR FLUCOR ZEDCOR ZAFCOR STP-POW BKS-COR F(x)u Ec Eo/Ec MACs
Mg ka 1.0347 1.0000 1.0285 1.0642 1.0394 .9895 .9465 1.3050 3.8314 1298.79
O ka 1.1173 .9990 .9608 1.0724 .9429 1.0189 .8448 .5317 9.4038 3603.22
ELEMENT K-RAW K-VALUE ELEMWT% OXIDWT% ATOMIC% FORMULA KILOVOL
Mg ka .00000 .56663 60.303 ----- 50.000 1.000 5.00
O ka .00000 .37018 39.697 ----- 50.000 1.000 5.00
TOTAL: 100.000 ----- 100.000 2.000
As I would think we would all expect, the absorption of both Mg and O in MgO, *decreases* as the beam energy is lowered.
Why then, not so for sulfur in carbon?
Here (again) is the Penepma matrix corrections from Penepma for S Ka and C Ka in 99% C and 1% S:
15 keV, S Ka in C-S_99-1: 1.06923
15 keV, C Ka in C-S_99-1: 1.11242
5 keV, S Ka in C-S_99-1: 1.24759
5 keV, C Ka in C-S_99-1: 1.02667
Why is S Ka in C behaving in this manner? Could it be due to the S L edge (165 eV) which can be fluoresced by carbon Ka (277eV)? But the fluorescence calculated from Fanal gives 0.9903 at 15 keV and 0.9878 at 5 keV, so that can't be it.
-
The same effect occurs in another system, e.g., Ti Ka in Mg (also 10 atomic numbers apart):
Here is Ti Ka in 99% Mg at 15 keV:
ELEMENT ABSCOR FLUCOR ZEDCOR ZAFCOR STP-POW BKS-COR F(x)u Ec Eo/Ec MACs
Ti ka 1.0239 1.0000 1.1500 1.1775 1.2143 .9470 .9493 4.9670 3.0199 212.202
Mg ka 1.0081 1.0000 .9988 1.0068 .9981 1.0007 .8595 1.3050 11.4943 515.582
ELEMENT K-RAW K-VALUE ELEMWT% OXIDWT% ATOMIC% FORMULA KILOVOL
Ti ka .00000 .00849 1.000 ----- .510 .005 15.00
Mg ka .00000 .98329 99.000 ----- 99.490 .995 15.00
TOTAL: 100.000 ----- 100.000 1.000
Here is Ti Ka in 99% Mg at 5 keV:
ELEMENT ABSCOR FLUCOR ZEDCOR ZAFCOR STP-POW BKS-COR F(x)u Ec Eo/Ec MACs
Ti ka 1.0004 1.0000 1.2380 1.2384 1.2380 1.0000 .9991 4.9670 1.0066 212.202
Mg ka 1.0012 1.0000 .9985 .9997 .9979 1.0006 .9782 1.3050 3.8314 515.582
ELEMENT K-RAW K-VALUE ELEMWT% OXIDWT% ATOMIC% FORMULA KILOVOL
Ti ka .00000 .00807 1.000 ----- .510 .005 5.00
Mg ka .00000 .99034 99.000 ----- 99.490 .995 5.00
TOTAL: 100.000 ----- 100.000 1.000
So, a similar effect as S Ka in 99% C.
So let's plot of the matrix correction vs. HV for a number of 1% emitters in a 99% matrix using Penepma (Penfluor/Fanal):
(https://probesoftware.com/smf/oldpics/i61.tinypic.com/2aiea2a.jpg)
I'm having some difficulty explaining these results but maybe there is a physicist "in the house"?
-
Hi John,
Glad my confusion has engendered others. I'd hate to be mystified alone.
Here (again) is the Penepma matrix corrections from Penepma for S Ka and C Ka in 99% C and 1% S:
15 keV, S Ka in C-S_99-1: 1.06923
15 keV, C Ka in C-S_99-1: 1.11242
5 keV, S Ka in C-S_99-1: 1.24759
5 keV, C Ka in C-S_99-1: 1.02667
In the above quote, your summary correction table seems to be different from the Output calcs
ELEMENT ABSCOR FLUCOR ZEDCOR ZAFCOR STP-POW BKS-COR F(x)u Ec Eo/Ec MACs
S ka .9979 1.0000 1.1684 1.1659 1.2104 .9653 .9931 2.4720 2.0227 197.942
C ka 1.0222 1.0000 .9986 1.0208 .9980 1.0006 .8693 .2838 17.6156 2791.77
ELEMENT K-RAW K-VALUE ELEMWT% OXIDWT% ATOMIC% FORMULA KILOVOL
S ka .00000 .00858 1.000 ----- .377 .004 5.00
C ka .00000 .96980 99.000 ----- 99.623 .996 5.00
TOTAL: 100.000 ----- 100.000 1.000
The latter being close enough to my regression slope value of 1.177 of the 5 different chemistries.
Have I missed something?
Les
-
Here (again) is the Penepma matrix corrections from Penepma for S Ka and C Ka in 99% C and 1% S:
15 keV, S Ka in C-S_99-1: 1.06923
15 keV, C Ka in C-S_99-1: 1.11242
5 keV, S Ka in C-S_99-1: 1.24759
5 keV, C Ka in C-S_99-1: 1.02667
In the above quote, your summary correction table seems to be different from the Output calcs
ELEMENT ABSCOR FLUCOR ZEDCOR ZAFCOR STP-POW BKS-COR F(x)u Ec Eo/Ec MACs
S ka .9979 1.0000 1.1684 1.1659 1.2104 .9653 .9931 2.4720 2.0227 197.942
C ka 1.0222 1.0000 .9986 1.0208 .9980 1.0006 .8693 .2838 17.6156 2791.77
ELEMENT K-RAW K-VALUE ELEMWT% OXIDWT% ATOMIC% FORMULA KILOVOL
S ka .00000 .00858 1.000 ----- .377 .004 5.00
C ka .00000 .96980 99.000 ----- 99.623 .996 5.00
TOTAL: 100.000 ----- 100.000 1.000
The latter being close enough to my regression slope value of 1.177 of the 5 different chemistries.
Have I missed something?
Les
The first set of calculations are from the Penepma/Penfluor Monte-Carlo calculation GUI in Standard.exe and the 2nd set of calculations are from analytical calculations from CalcZAF.
The C Ka matrix correction calculations at 5 keV are in almost perfect agreement (1.02), but yes the S Ka at 5 keV are a little different (1.24 vs. 1.16), and the physics is, as I said above, weird.
Frankly I trust Penepma a little more (especially when it gets "weird")... ;)
-
Hi Les,
Did you mean to "strike out" most of your post?
No, goodness knows how I managed it.
Are you trying to create a calibration curve?
Yes, a simple linear calibration curve so I can make my own calibrated maps.
I have now two factors - my own at 1.177 and yours as 1.24.
I like your excel spreadsheet format. I take it A refers to Sulphur and B to Carbon.
I tend to drive PENEPMA from the DOS prompt and use command files and exe code to strip out the lines of interest from the output files.
Les
-
Hi Les,
Did you mean to "strike out" most of your post?
No, goodness knows how I managed it.
Are you trying to create a calibration curve?
Yes, a simple linear calibration curve so I can make my own calibrated maps.
I have now two factors - my own at 1.177 and yours as 1.24.
I like your excel spreadsheet format. I take it A refers to Sulphur and B to Carbon.
I tend to drive PENEPMA from the DOS prompt and use command files and exe code to strip out the lines of interest from the output files.
Les
It looks like you wrote a "square bracket S square bracket" which is a "strikeout" format code.
You might give the Penepma GUI a go. I suspect you'll find it *much* easier to use.
john
-
I like your excel spreadsheet format. I take it A refers to Sulphur and B to Carbon.
Thanks!
The column headings are explained in the Probe for EPMA reference, just hit the F1 key in any of my apps and search for Penepma.
In the Penepma Penfluor/Fanal GUI, because it was designed for calculation of secondary fluorescence effects at boundaries, A refers to the beam incident material and B refers to the boundary material.
However for the purposes of the matrix correction calculations, we simply specify the same material for *both* material A and B and then we get a *matrix* calculation... cool! 8)
So for the two spreadsheets attached above (here):
http://probesoftware.com/smf/index.php?topic=47.msg2016#msg2016
both are S Ka in carbon, one at 5 keV and the other at 15 keV.
See here for easy modeling of matrix effects in Standard.exe:
http://probesoftware.com/smf/index.php?topic=152.0
-
Les,
In case you find it useful I've attached .PAR files for 11 compositions of the carbon-sulfur binary system in a zip file below. They are calculated down to 0.2 keV (for C Ka) even though it obviously doesn't matter for S Ka.
These can be utilized in the Penepma/Penfluor/Fanal GUI in Standard.exe as described here:
http://probesoftware.com/smf/index.php?topic=152.0
-
Thanks,
How do I download the latest version of your software?
I am encouraged by the ease that it does what I do with DOS batch files in a GUI. The version I have is quite old.
Les
-
How do I download the latest version of your software?
I am encouraged by the ease that it does what I do with DOS batch files in a GUI. The version I have is quite old.
Hi Les,
The best place to start is (appropriately enough) the Getting Started With CalcZAF topic here:
http://probesoftware.com/smf/index.php?topic=81.0
Penepma 2012 comes with the CalcZAF.msi installer. Just follow the defaults and prompts. Penepma 2012 is very excellent.
I'll be updating to the latest Penepam 2014/2015 (?) pretty soon, but Xavier and Salvat are fixing a small bug I found for B Ka, so I'll probably wait for that update.
Once CalcZAF/Penepma are installed on your computer, you can start with CalcZAF.exe and Standard.exe. Basically CalcZAF deals with unknowns and Standard deals with knowns. Here is the intro:
http://probesoftware.com/smf/index.php?topic=152.0
-
Thanks,
Now I know what I am doing over the Christmas break!
Les
-
First, the absorption correction acts as expected. It is smaller at 5 keV than 15 keV. Second, the increase in the total correction seems to result from an increase in the Z correction (both stopping power and backscatter correction.)
It is not obvious from staring at the Bethe energy loss equation why the difference would be so large for the difference in stopping power (see attached PDF). It is equivalently unintuitive why such a small amount of C has such a large influence on the backscatter correction (the mean atomic number does not change by much.) However, these calculations are much more challenging to understand than the absorption correction which simply depends upon MAC and path length.
Simulating this measurement in DTSA-II shows that the analytical matrix corrections are consistent with Monte Carlo calculations so if there is an error then it manifests itself in both analytical and Monte Carlo calculations.
Edit by John: Thanks Nicholas! A nice physics intuition mystery.
For those just reading this, here is the post containing the graph Nicholas and I are trying to explain:
http://probesoftware.com/smf/index.php?topic=398.msg2019#msg2019
-
hi John,
An intuitive thought: if the S is dilute, which it is, the electrons on average may interact with C before they get to the S atoms. This means that the average energy they have by the time they get to the sulphur atoms is lower than the beam energy and thus closer and closer to the minimum excitation energy. Given that, there should be no such effect at high S concentrations.
I would imagine that K in S would kick up at about 7kV and Fe in C at about 12 kV.
Les
-
There may be some confusion here.
Les was originally discussing an emitter (S ka) at the 1% level in a 99% matrix of carbon. So I modeled that as seen in the graph here:
http://probesoftware.com/smf/index.php?topic=47.msg2019#msg2019
along with some other emitter-absorber combinations. That's when I called in the "big gun" Nicholas Ritchie at NIST to help explain things.
When Nicholas chimed in with a post labeled "1% C in S at 5 keV and 15 keV" I thought, well maybe he misread our discussion, but I thought I might as well model the emitter at the 99% level even though the matrix effects will obviously be close to 1.
So here is that graph (99% emitters in a 1% absorber matrix):
(https://probesoftware.com/smf/oldpics/i57.tinypic.com/2hed5qv.jpg)
and the Ti ka matrix correction at 10 keV? Well they really are all slightly different- so I didn't use the same value accidentally.
But what the heck occurs at 10 keV for Ti ka in these absorbers?
-
Hi John,
Now I'm looking at condensation materials in the blast furnace and one of them is a graphite phase, unlike any seen so far in the blast furnace so far. It is mostly C with ~5%K. This might be a KxC60 buckyball condensate.
More modelling.
If the blast furnace does turn out to be a buckyball generator and we could get it out, it would be worth substantially more than the iron we produce ;D
Les