General EPMA > EPMA Standard Materials
An Open Letter to the Microanalysis Community
Probeman:
--- Quote from: crystalgrower on December 17, 2021, 11:41:22 AM ---The real intent of this Open Letter may be to acquire some piles of materials not able to be sold for other reasons. A pitch to ask for business leftovers should have been stated up front, rather than all the noise about money possibly sitting in the bank accounts of nonprofits.
--- End quote ---
Asking for commercial quotations, seeking state/government sources, asking for "leftovers" and custom synthesis are all activities that can be pursued in parallel, they are not exclusive pursuits.
Nachlas et al. are currently writing grant proposals, and matching funds may be required from "money in bank accounts of non profits", though exactly how that will be leveraged is up to those society directors.
Probeman:
There has been some discussion in this topic on whether we really require so called "matrix-matched" primary standards for high accuracy EPMA. That is, do our primary standards really need to be similar in composition (and also valence and coordination), to our unknown materials?
So aside from the questions regarding the accuracy of our compositional matrix correction physics, it's a valid question since as we know from multiple studies that in the case of light elements at least, we experience peak shift and shape effects that can result in accuracy problems when not utilizing integrated area scan acquisitions for elements such as oxygen, nitrogen, carbon, boron, etc. Even sulfur k-alpha can have a significant peak position shift (though not a shape change) depending on the oxidation state of the sulfur:
https://probesoftware.com/smf/index.php?topic=127.0
So, even if our compositional matrix corrections were perfect, we would still need to ascertain the magnitude of the peak shift and shape effects due to chemical states. And I think it's still an open question whether we can accurately extrapolate from one material to another when the element (mission transitioning from the valence shell) in our primary standard is in a difference valance state and/or coordination than that of our unknown.
And that is one reason why we selected as our first test sample for the high purity synthetic standard round robin three materials: MgO, Al2O3 and spinel (MgAl2O5). The idea being that these pairs (Mg Ka in MgO to MgAl2O3 and Al Ka in Al2O3 to MgAl2O5), are not only very different in composition, but also somewhat different in their chemical states. And those materials were readily available as high purity synthetics! :)
We are still awaiting the results from this first round robin, but I decided to share another test of this concern, that is Si Ka in SiO2 compared to some common silicates. This was a test I ran recently looking further into the problem of measuring trace Sr and Rb in silicates, but let's ignore those trace results for now and focus on the Si and Al major elements. Unfortunately I didn't have an Al2O3 standard in the standard mount (and wasn't running this test for the Si and Al concentrations as they were only being measured for the interference corrections), but still the Si data might be helpful regarding these major elements accuracy issues.
So using SiO2 as the primary standard for Si (and nepheline as the primary standard for AL), we obtain these results for labradorite:
ELEM: Sr Rb Si Al Ca Na K Fe Mg O SUM
1379 .055 -.013 24.495 16.458 9.577 2.841 .100 .319 .084 46.823 100.739
1380 .051 -.001 24.060 16.495 9.577 2.841 .100 .319 .084 46.823 100.350
1381 .061 -.004 24.038 16.513 9.577 2.841 .100 .319 .084 46.823 100.351
1382 .070 -.016 23.908 16.544 9.577 2.841 .100 .319 .084 46.823 100.250
AVER: .059 -.008 24.125 16.502 9.577 2.841 .100 .319 .084 46.823 100.423
SDEV: .008 .007 .256 .036 .000 .000 .000 .000 .000 .000 .216
SERR: .004 .003 .128 .018 .000 .000 .000 .000 .000 .000
%RSD: 13.91 -82.23 1.06 .22 .00 .00 .00 .00 .00 .00
PUBL: n.a. n.a. 23.957 16.359 9.577 2.841 .100 .319 .084 46.823 100.060
%VAR: --- --- .70 .88 .00 .00 .00 .00 .00 .00
DIFF: --- --- .168 .143 .000 .000 .000 .000 .000 .000
STDS: 251 1023 14 336 --- --- --- --- --- ---
So well within 1% relative accuracy on both Si and Al. Now for the nepheline (just looking at Si because this is the primary standard for Al):
ELEM: Sr Rb Si Al Na K Fe O Ca SUM
1383 .009 .030 20.553 17.872 12.552 4.657 .155 44.418 .075 100.322
1384 -.002 .045 19.924 17.774 12.552 4.657 .155 44.418 .075 99.598
1385 .006 .029 20.594 17.954 12.552 4.657 .155 44.418 .075 100.440
1386 .002 .025 20.422 17.857 12.552 4.657 .155 44.418 .075 100.163
AVER: .004 .032 20.373 17.864 12.552 4.657 .155 44.418 .075 100.131
SDEV: .005 .009 .308 .074 .000 .000 .000 .000 .000 .373
SERR: .002 .004 .154 .037 .000 .000 .000 .000 .000
%RSD: 134.64 27.41 1.51 .41 .00 .00 .00 .00 .00
PUBL: n.a. n.a. 20.329 17.868 12.552 4.657 .155 44.418 .075 100.054
%VAR: --- --- .22 (-.02) .00 .00 .00 .00 .00
DIFF: --- --- .044 (.00) .000 .000 .000 .000 .000
STDS: 251 1023 14 336 --- --- --- --- ---
Again excellent accuracy extrapolating from SiO2. Now our orthoclase standard:
ELEM: Sr Rb Si Al Fe K Na Ba O SUM
1387 .005 .120 29.905 8.844 1.461 12.859 .675 .054 45.798 99.721
1388 -.011 .121 29.736 8.860 1.461 12.859 .675 .054 45.798 99.553
1389 -.001 .087 30.128 8.792 1.461 12.859 .675 .054 45.798 99.853
1390 -.010 .117 30.202 8.773 1.461 12.859 .675 .054 45.798 99.929
AVER: -.004 .111 29.993 8.817 1.461 12.859 .675 .054 45.798 99.764
SDEV: .008 .016 .212 .041 .000 .000 .000 .000 .000 .165
SERR: .004 .008 .106 .021 .000 .000 .000 .000 .000
%RSD: -172.78 14.61 .71 .47 .00 .00 .00 .00 .00
PUBL: n.a. .027 30.286 8.849 1.461 12.859 .675 .054 45.798 100.009
%VAR: --- 311.36 -.97 -.36 .00 .00 .00 .00 .00
DIFF: --- .084 -.293 -.032 .000 .000 .000 .000 .000
STDS: 251 1023 14 336 --- --- --- --- ---
Again within 1% relative accuracy for both.
In other prior work I've seen similar accuracy extrapolating from MgO to other Mg silicates, so I do believe these extrapolations are feasible, though we will see in the case of Al Ka since we already know from work by Fournelle that there are subtle Al peak position shifts in feldspars at least.
By the way, these measurements were performed at 50 nA because the purpose was to look at the trace elements, but the beam was defocused to 10 um to minimize TDI effects. Never the less, some intensity changes over time were observed as shown below, but only for the Si Ka emissions!
The above being a normal exponential TDI fit. A hyper-exponential fit might be worth trying even though it appears to overfit, the intercepts are probably more accurate and that's what we utilized here:
This resulted in a TDI correction of around 1.7% +/- 0.2 for Si Ka and 0.34% +/- 0.1 for Al Ka in the labradorite
Anyway, bottom line is that major element matrix correction extrapolations can be quite accurate even when extrapolating from a pure oxide to a silicate mineral, and the valence and coordination effects seem to be minimal.
Probeman:
I just wanted to address a question on the purity required for synthetic standards that we utilize as primary standards for major element analysis. In other words, how pure do these synthetic mineral standards really need to be?
My answer would be that I think we should be initially focused on major element standards. The main reason being to guide analysts away from the seduction of so called "matrix matched" standards of questionable accuracy (and availability). Instead what we need are high accuracy (and high availability) major element standards, e.g., MgO, Mg2SiO4, MgAl2O4, SiO2, Fe2SiO4, Fe3O4, etc., etc.
Trace element (doped) standards are really not necessary in my opinion. The only thing we need for trace element accuracy are pure metals or simple oxides for primary standards (and their accuracy is not all that important!). What's important for trace elements is the background modeling! And optionally also high purity blank (possibly matrix matched) materials for the application of a blank correction. Doped trace standards are for dopes! 😁
Back to major elements, our preliminary MgO, Al2O3, MgAl2O4 FIGMAS data reported by Will shows that with accurate dead time calibrations we can obtain better than 1 to 2% accuracy extrapolating from simple compounds to significantly different compositions ( I obtained similar results myself with the test mount). And also my own long term experience with other high concentration primary standards for other elements, and a recent test using SiO2 as a primary standard here:
https://probesoftware.com/smf/index.php?topic=1415.msg10574#msg10574
shows similar sub 1% accuracy. The good news is that our modern matrix corrections are pretty damn good, the weak point being the dead time calibrations on our instruments. And those can be easily fixed in a couple of hours work:
https://probesoftware.com/smf/index.php?topic=1160.0
So our concerns regarding trace elements in these synthetic materials I think should only be relative to the extent to which these traces affect the accuracy of the theoretical (major element) stoichiometry of these synthetic compositions. So to me that means 99.99% purity is just fine. A 100 ppm variance on our major elements is not a significant concern.
Probeman:
Attached below is the latest draft of the action plan of global standards (login to see attachments).
If anyone has additional information (e.g., additional suppliers of high purity synthetic single crystals), just send me the information and I will add it to the document.
Thanks!
Navigation
[0] Message Index
[*] Previous page
Go to full version