John
You are probably fully aware of this, but I wanted to get your opinion and that of the others on the forum. I set up a probe run using the JEOL software using the FFAST MACs and ZAF procedure. I analyzed some olivine and pyroxenes. I can get 10 of 11 elements coming out correctly when I analyze my reference standards, but MgO is always low. Pete McSwiggen logged onto my probe yesterday and did some testing. He is telling me that JEOL has two MAC sets (FFAST and modified Heinrich) and three ZAF procedures (ZAF, Armstrong and XPP). He did a calibration and recalculated my standards using the six variations of the MACs and the ZAF procedures. The best combination came out with the modified Heinrich and ZAF. The olivines came out fine, but MgO on my Lunar Crater Augite and which has 17.32 wt% MgO, only comes out with 17.1 MgO. My Cameca SX-100 calculated these perfectly and I know I used PAP as the correction procedure, but I am not sure about the MAC set used by Cameca. This basically means if I were to use the JEOL software, I would have to post-process the data for mobile elements and then post-process MgO on all of my pyroxenes - which is ludicrous.
Pete mentioned Armstrong talked about this issue last year at the M&M meeting and it is not unique to JEOL. It mainly comes down to the MACs and the correction procedures being chosen and PfE has these same MACs and ZAF procedures (along with a host of others).
Obviously, I am jumping to PfE, but what is your take on this? I am again floored that Cameca seems to have solve this issue 30 years ago and JEOL has not. Is there a best set of MACs and a preferred correction procedure for typical geological phases (glass, olivine, pyroxene, amphibole, feldspar, phosphates, etc)?
Thanks
Joe
Quote from: Joe Boesenberg on December 12, 2024, 09:34:11 AMJohn
You are probably fully aware of this, but I wanted to get your opinion and that of the others on the forum. I set up a probe run using the JEOL software using the FFAST MACs and ZAF procedure. I analyzed some olivine and pyroxenes. I can get 10 of 11 elements coming out correctly when I analyze my reference standards, but MgO is always low. Pete McSwiggen logged onto my probe yesterday and did some testing. He is telling me that JEOL has two MAC sets (FFAST and modified Heinrich) and three ZAF procedures (ZAF, Armstrong and XPP). He did a calibration and recalculated my standards using the six variations of the MACs and the ZAF procedures. The best combination came out with the modified Heinrich and ZAF. The olivines came out fine, but MgO on my Lunar Crater Augite and which has 17.32 wt% MgO, only comes out with 17.1 MgO. My Cameca SX-100 calculated these perfectly and I know I used PAP as the correction procedure, but I am not sure about the MAC set used by Cameca. This basically means if I were to use the JEOL software, I would have to post-process the data for mobile elements and then post-process MgO on all of my pyroxenes - which is ludicrous.
Pete mentioned Armstrong talked about this issue last year at the M&M meeting and it is not unique to JEOL. It mainly comes down to the MACs and the correction procedures being chosen and PfE has these same MACs and ZAF procedures (along with a host of others).
Obviously, I am jumping to PfE, but what is your take on this? I am again floored that Cameca seems to have solve this issue 30 years ago and JEOL has not. Is there a best set of MACs and a preferred correction procedure for typical geological phases (glass, olivine, pyroxene, amphibole, feldspar, phosphates, etc)?
Thanks
Joe
How do we know what this lunar augite composition really is? How homogeneous is it? 17.3 vs. 17.1 is pretty tight... what sort of variances are you seeing in these analyses?
I'm guessing the lunar augite was not analyzed by wet chemistry. Was it an average of probe analyses? Try this test instead to check accuracy: use an Mg2SiO4 synthetic as a primary Mg standard. We know for sure what that chemistry is. Will Nachlas has this material as part of the FIGMAS effort:
https://smf.probesoftware.com/index.php?topic=1415.0
And for a secondary Mg standard I would utilize the SRM K-412 and K-411 mineral glasses. Those values are also very well known and homegeneous. When we've run this test we've gotten results within 1% accuracy (with accurate dead time corrections). Will Nachlas hopes to provide a primary and secondary standard in his FIGMAS mounts for the most common geological elements, so this sort of accuracy testing will be easier in the future. I'd contact him about some of these synthetic materials.
That said, I've noted in the past that the Heinrich MAC for Mg Ka by Fe is about 20% too high (compared to other tabulated values):
https://smf.probesoftware.com/index.php?topic=239.msg3042#msg3042
https://smf.probesoftware.com/index.php?topic=598.msg7759#msg7759
This is why traditionally people have obtained garnet analyses that have high totals... here are some other interesting posts related to this question:
https://smf.probesoftware.com/index.php?topic=239.msg3045#msg3045
There's also the possibility of "empirical" MACs that can be determined using multi-voltage analysis:
https://smf.probesoftware.com/index.php?topic=1340.msg9630#msg9630
Normally these empirical MACs are utilized for low Z emitters (O, N, C, etc.), but there are other situations in which they can be useful, for example Si Ka which is close to the Hf absorption edge.
I would also suggest trying the "Use All Matrix Corrections" checkbox in the Analyze! window to compare the various matrix corrections with your measurements in Probe for EPMA. Of course there are also 6 different MAC tables in Probe for EPMA... see here for an example running all matrix corrections and MAC tables in CalcZAF:
https://smf.probesoftware.com/index.php?topic=890.msg11956#msg11956
This comparison of matrix correction is exactly what Andrew Lock recently did for the NIST Au-Ag alloys:
https://smf.probesoftware.com/index.php?topic=40.msg13051#msg13051
The Lunar Crater augite is a historical standard analyzed by Gene Jarosewich at the Smithonian by wet chemistry in the 80's. He called it Chromian Augite in his 1987 Geostandards Newsletter. I wouldn't rely on random probe analyses. I have analyzed this standard several thousand times in the last 30 years with four different microprobes: a JEOL 8600 at Rutgers, an ARL at AMNH and two SX-100's at AMNH and Brown. This is a very homogeneous standard. The only time this standard is off is when the calibration is off. However, besides this standard, the other Ca-Al bearing pyroxenes I have, all come out with the wrong MgO, and all low. I have not tried low-Ca pyroxenes yet. They may come out better (like the olivine). I am trying to PfE, so have put this aside for the moment.
I used pure forsterite (Fo100) as the primary Mg standard for the calibration. I also tried a Fo97. Both are synthetic. Both gave the same answer. I also changed spectrometers for Mg. Still got the same wrong answer. When Pete McSwiggen at JEOL did his tests, he did 5 analyses on different grains of LC Augite. All came out low. None were near 17.3. They were all clustered tightly around 17.08 MgO (range of 17.03 to 17.13) (when using the best MAC and ZAF combination available on the JEOL probe).
Ok. I will try your suggestions when I can run PfE. Using PfE, may just solve the problem, if this is in fact a JEOL software issue. No one at JEOL has told me yet, if anyone else has experienced this issue. I find it hard to believe I am the first one to notice this in the software. From what I understand, this software has been around for some time. Thanks.
OK, cool. So this is one of the Smithsonian standards which is OK, but as you know, some of those standards are known to be heterogeneous and have inclusions:
https://smf.probesoftware.com/index.php?topic=301.msg5207#msg5207
See this video which discusses some of the issues:
https://www.youtube.com/watch?v=mOca7-G4FvQ&ab_channel=ProbeSoftwareInc
But if you have confirmed the homogeneity and purity of this augite that's good enough for me (the problem being that wet chemistry does an average of 10 grains which can bias the analysis of individual grains). In fact Gene has stated that if these standards are utilized in a standardization, they should be averaged over multiple grains, which of course none of us bother with... :D
But it will be interesting to see what you come up with for sure.
John
Once I learn PfE and retackle this issue, one of the things I want to test is whether I can duplicate my results from the Cameca SX-100 using PAP. This would give me a lot more confidence in my new JEOL probe analytical abilities. Are all the MACs used for PAP in PfE? I know that Pouchou and Pichoir 1991 used the MAC30 mass absorption coefficients, but had a supplementary table for the light elements and replaced some specific MAC values for the heavier elements. I saw PAP is in PfE and I saw that MAC30 is in there, but is the complete combination used by Pouchou and Pichoir in PfE? Thanks.
Quote from: Joe Boesenberg on December 16, 2024, 05:32:49 PMJohn
Once I learn PfE and retackle this issue, one of the things I want to test is whether I can duplicate my results from the Cameca SX-100 using PAP. This would give me a lot more confidence in my new JEOL probe analytical abilities. Are all the MACs used for PAP in PfE? I know that Pouchou and Pichoir 1991 used the MAC30 mass absorption coefficients, but had a supplementary table for the light elements and replaced some specific MAC values for the heavier elements. I saw PAP is in PfE and I saw that MAC30 is in there, but is the complete combination used by Pouchou and Pichoir in PfE? Thanks.
Good question.
The MAC tables in Probe for EPMA (and CalcZAF) are shown here:
(https://smf.probesoftware.com/gallery/395_18_12_24_8_22_47.png)
The MAC30 tables are the values from Pouchou so far as I know, but there are updated values I obtained some time ago for boron through fluorine emitters which are attached. I can't remember who gave me this file, it could have been Philip Poeml.
These values appear to be somewhat different from the MAC30 values, so you can edit the MAC30 table (MAC30.DAT) using CalcZAF if you want from the X-Ray | Edit MAC Table menu.
But the PAP_MAC.txt file only goes up to F, not Mg.
See attached.