Probe Software Users Forum
General EPMA => Discussion of General EPMA Issues => Topic started by: Rom on January 16, 2022, 08:45:49 PM
-
Hi dear colleagues,
Our issue is indetermination on 0.5-1wt.% metals (Fe, Cu,...) in simple stoichiometric oxides if we use standards pure metals.
We didn't find peak shifts when come from metals to oxides.
K-line, Lif.
To simplify our task we use specify (not measured) oxygen concentration.
What direction should we look? These are strong lines and use APF or MAC corrections looks unusual.
Thank you.
-
Our issue is indetermination on 0.5-1wt.% metals (Fe, Cu,...) in simple stoichiometric oxides if we use standards pure metals.
We didn't find peak shifts when come from metals to oxides.
K-line, Lif.
To simplify our task we use specify (not measured) oxygen concentration.
What direction should we look? These are strong lines and use APF or MAC corrections looks unusual.
1. Please explain "indetermination".
2. One should not observe peak shifts with high energy emission lines (e.g., Fe Ka, Ni Ka, etc.). Peak shifts due to chemical states are usually only seen when the transition shell is also the bonding shell (e.g., C Ka, N Ka, Si Ka etc.).
3. Either pure metal (if unoxidized) or oxide primary standards should be fine as the most important correction is the background correction at low concentrations:
4. Please explain "What direction should we look? These are strong lines and use APF or MAC corrections looks unusual."
5. Here are some additional readings for you:
https://probesoftware.com/smf/index.php?topic=1404.0
https://probesoftware.com/smf/index.php?topic=1378.0
-
1.Indetermination: the measured concentration of metal in oxide is lower in 0.5-1 wt% its stoichiometric (real) concentration.
2. Yes
3. Yes
4. I just kindly asked where should I looking for my mistake (because 2,3 - yes).
5. Thank you
-
1.Indetermination: the measured concentration of metal in oxide is lower in 0.5-1 wt% its stoichiometric (real) concentration.
Did you mean to say: "the measured concentration of metal in oxide is lower than 0.5-1 wt% which is its stoichiometric (real) concentration."?
-
Yes. You are right.
-
For minor elements of first row transition elements I wouldn't be too worried about MACs (certainly not APFs), but did you perform careful wavescans to check the background positions? That's going to have the biggest effect on accuracy for minor elements.
Also are the totals good? Did you include oxygen from ferric iron in the calculation?
https://probesoftware.com/smf/index.php?topic=92.msg8593#msg8593
What is the oxide matrix? What are the minor elements in question?
-
I didn't understand about minor elements. I told about metal measure in pure oxides, which are very close to stoichiometric compound (Fe2O3, Cu2O etc.)
Yes, I performed careful wavescans to check the peak shifts and background positions.
Totals are 98.5-99.5. I used both: measured or specify (not measured) oxygen concentration.
My issue is absolutely the same question of your topic! Thank you!
But my results didn't change if I use "calculate excess oxygen from Ferrous/Ferric ratio" checkbox (attached file). In excel file I see zeroes in FeO, Fe2O3, excess oxygen etc. columns.
-
There are 3 questions:
1. What I should check If I want to use "calculate excess oxygen from Ferrous/Ferric ratio" - how can I start this options;
2. How I can start this options for system Cu-O or Sn-O etc., without Fe (it is necessary if we research alloys after quenchingfrom liquid for instance).
3. What is the fundamental base of "the measured concentration of metal in oxide is lower than 0.5-1 wt% which is its stoichiometric (real) concentration"? Everything is clear if we discuss Oxygen concentration but why this correction affects on metal concentration?
Thank you a lot.
-
It appears no one has responded to your last post here, but I will begin again with a quote from your post yesterday in the other topic:
Actually my target is simple. I try to understand why the measured content of Fe in Fe2O3 standards (Taylor and SPI) is 2wt% lower than it should be.
...
Fe standard - Fe metal in the same block with Fe2O3 (carbon coating is the same).
O standard - Fe2O3 or MgO at the same block.
Doesn't matter: oxygen measured or calculated, use or not Fe+2/+3 correction, width and shape BG (from detail WSs analysis).
FeKa is a strong line, so APFs for Fe wont affect.
MAC, APF for O change O. Of course Fe changes a bit as well but not so successfully as it needs.
The only point I thought was a wrong peaking. But also not, everything is good.
Certainly, my affords with searching the solution give me new knowledge but...
You and your colleagues many times recommended to look at results of calculations with different corrections. I did. But what should I see is not clear. For instance here are results for calculation with default MAC table (LINEMU Henke...) and FFAST table. Empirical MAC and APF values are not use in the calculations.
Let's start by assuming you are measuring Fe Ka using Fe metal as a primary standard and Fe2O3 as a secondary standard and calculating oxygen by stoichiometry (2:3). Then your only concern should be the measurement of Fe Ka.
The following should not be concerns:
1. Peaking issues should not be an issue (little to no peak shifts for Fe Ka)
2. MAC issues should not be a concern as MACs are very small for Fe Ka in oxygen as shown here:
MAC value for Fe ka in O = 22.55 (LINEMU Henke (LBL, 1985) < 10KeV / CITZMU > 10KeV)
MAC value for Fe ka in O = 22.20 (CITZMU Heinrich (1966) and Henke and Ebisu (1974))
MAC value for Fe ka in O = 22.25 (MCMASTER McMaster (LLL, 1969) (modified by Rivers))
MAC value for Fe ka in O = 22.26 (MAC30 Heinrich (Fit to Goldstein tables, 1987))
MAC value for Fe ka in O = 22.25 (MACJTA Armstrong (FRAME equations, 1992))
MAC value for Fe ka in O = 20.85 (FFAST Chantler (NIST v 2.1, 2005))
MAC value for Fe ka in O = 22.00 (USERMAC User Defined MAC Table)
3. Carbon coating differences should not be an issue as long as the over voltage is reasonable (15 keV or higher).
4. Background corrections should not be an issue using the LiF Bragg crystal. However analyzing Fe Ka on a PET Bragg crystal could be very problematic for the background will be very curved at such a low sin theta.
5. Surface polish should not be an issue for an energetic line such as Fe Ka, but always worth making sure the samples are well polished.
However, these issues could be problematic:
1. Your Fe metal standard could have surface oxidation, but if so this would tend to raise the concentration of Fe in your Fe2O3 secondary standard (the primary standard intensity is in the denominator of the k-ratio).
2. I note that you have only analyzed for Fe cations. Have you checked that your Fe2O3 standard from SPI/Taylor is actually 99.99% pure? Is it natural or synthetic? Common natural impurities are Si, Ti, Al, Mn, H2O. This alone could explain your observations.
Again, as I have said before, we need to identify, obtain and distribute at least two high purity synthetic minerals for each of the (at least to begin with, common) geological elements, on a global basis, so that we can actually begin to rigorously compare our results with each other. This problem is described here for those that have not seen it:
https://probesoftware.com/smf/index.php?topic=1415.0
For example, Will Nachlas has documented that a so called 99.99% Rh metal standard in one of his commercial mounts actually has ~4 wt% Fe in it! :o
Quantitative analysis of oxygen as a major element is a whole separate endeavor, and here is a link to one of those discussions:
https://probesoftware.com/smf/index.php?topic=197.0
But I suggest that we first figure out what is going on when you try to analyze Fe2O3 using Fe metal and calculating oxygen by stoichiometry. I would start by analyzing for the trace and minor elements, but maybe start by examining the Fe2O3 with EDS using a long count time to obtain sufficient sensitivity.
-
Hi John,
The results below are with measured oxygen (Standard is MgO).
Yes, of course I utilizing the Fe Ka emission line.
What topic should I look?
Thank you!
-
Hi John,
The results below are with measured oxygen (Standard is MgO).
Yes, of course I utilizing the Fe Ka emission line.
What topic should I look?
Thank you!
Doesn't matter: oxygen measured or calculated, use or not Fe+2/+3 correction, width and shape BG (from detail WSs analysis).
You said (quoted above), that you saw the same low totals when calculating oxygen by stoichiometry. So let's deal with the quantification of Fe first and once that is working we can look at the measurement of oxygen, which is much more complicated.
Please share with us an example of measuring Fe in Fe2O3 using Fe metal as your primary standard and calculating oxygen by stoichiometry.
-
Please, see attached picture.
It should be noted that Fe met of the Standard includes (estimate) 0.6-0.7%wt of oxygen. But it doesn't matter on the current step.
Default MAC table (LINEMU Henke...), default ZAF correction
-
OK, it is clear that your Fe metal standard and your Fe2O3 standard do not agree, so there is definitely a problem. You didn't mention the conditions (keV, nA, etc.), so that would be worth knowing also...
By the way, the oxygen values you show are the same for every data point (all three) so they are not being calculated by stoichiometry, but rather being loaded from the standard database as a fixed concentration (note the "Published" value for oxygen). But no matter because the real problem is the measured Fe concentration in your Fe2O3 standard.
2. I note that you have only analyzed for Fe cations. Have you checked that your Fe2O3 standard from SPI/Taylor is actually 99.99% pure? Is it natural or synthetic? Common natural impurities are Si, Ti, Al, Mn, H2O. This alone could explain your observations.
But I suggest we start by trying to figure out what is going on by analyzing your Fe2O3 secondary standard using Fe metal as a primary standard and calculating oxygen by stoichiometry. I would begin by analyzing your Fe2O3 standard using EDS with a long count time time and check for trace/minor impurities. You won't be able to check for H2O, but it's a start.
I would start with examining your Fe2O3 standard and check it for impurities. It seems unlikely to have that much "undeclared" contamination, but it is possible.
Also there is an excellent application that comes with Probe for EPMA that can be used to check how well your standards agree with each other called Evaluate.exe. It was developed with John Fournelle and I think Dan Kremser back in the day, but is still very useful.
Basically you can select any PFE probe database file with standards that has been acquired and you can plot them up one element at a time, to see how well the standards all agree with each other. Here is a topic that describes the application:
https://probesoftware.com/smf/index.php?topic=340.msg10757#msg10757
Worth checking out.
-
Quite many directions. So lets move step by step.
1. 15 keV, 10 um diameter (we used from 0 to 100 um, in general all the same), 25 sec on peak, 8+8 sec BG, WS for Fe and Fe2O3 on the picture.
2. You a right (sorry), the oxygen has taken from the standard composition. On the picture lower - the oxygen, calculated from stoichiometry.
3. I would start with examining your Fe2O3 standard and check it for impurities. It seems unlikely to have that much "undeclared" contamination, but it is possible.
Yes. What I did.
-I measured two completely different Standards of Fe2O3 - in Taylor block and in SPI block. Both give close results. Both are natural.
-I checked WSs "from edge to edge" which I collected on Fe2O3 Taylor. Nothing found. But I'll check again Si, Ti, Al, Mn - thank you for show the direction of searching.
-May be I need to check energy limit with EDS?
-Ok, I'll scan one of my Fe2O3 standards using EDS with a long count time time and check for trace/minor impurities.
4. I didn't use Evaluate.exe. Hope the application is not too difficult for me.
-
1. 15 keV, 10 um diameter (we used from 0 to 100 um, in general all the same), 25 sec on peak, 8+8 sec BG, WS for Fe and Fe2O3 on the picture.
You should never use more than a 20 um beam diameter to avoid Bragg defocus effects.
But the most important condition parameter to my mind is the beam current. One possible concern is that you are using a very high beam current and if your dead time correction is out of calibration that could explain some of what you are seeing.
4. I didn't use Evaluate.exe. Hope the application is not too difficult for me.
The Evaluate application is very easy to use.
-
Forgot... we use 20nA
-
Forgot... we use 20nA
OK, so probably not a problem with the your dead time corrections.
Please check for impurities in your Fe2O3 and use Evaluate.exe to check on agreement with other Fe standards.
-
Ok, thank you. I am disappearing for several days to check everything.
-
Greetings!
1. I did not find some impurities in Fe2O3 Taylor with EDS spectrometer (15KeV, 20nA, 10 um, 15-20 min). Energy limit is 15 keV - see photos of screen.
2. I used Evaluate application for evaluate 4 standards with Fe we have in the Taylor block: FeCuS2(28), FeS(16), Fe2O3(33), Fe(32). Sample 7201 - Fe2O3
synthesized from Fe metal (1200C, CO2 atmosphere, 4 hours) in our lab. See addition picture.
-
I told you Evaluate was easy to use! :)
So both your Fe2O3 standards compare low in Fe compared to your Fe metal and also your Fe sulfide. Maybe they have excess H2O or oxygen?
What are the compositions provided by the commercial vendors?
-
Yes, the Evaluate is easy and useful application - thank you.
All these standards except 7201-Fe2O3 are in the Taylor block which was supply by commercial vendor.
I thought about O2 or H2O in Fe2O3 Taylor - it is possible but not 2%wt.
I can add to the Evaluate diagram results for extra one Fe2O3 Standard from SPI block which was supply by commercial vendor too.
Also we can see the same composition in synthetic fresh sample 7201. I can't image that this sample obtained in the furnace at the exactly known PO2 contains less then 69.5-70wt% of Fe.
-
I would not rely on your Fe2O3 being 'bang on' stoichiometric - it will almost inevitably contain some mixed valence Fe.
-
I would not rely on your Fe2O3 being 'bang on' stoichiometric - it will almost inevitably contain some mixed valence Fe.
To be clear, it's not my Fe2O3. It's mounted in a Taylor block from Rom's lab at the University of Queensland. Yet someone, somehow manages to call it a "standard" whatever that means! And advertise it for sale! :D
But yes, I agree that makes a lot of sense. Could this be related at all to the inclusion of H2O or OH in hematite?
-
one persons standard is another persons 'where did we find that again?'
;D
to be fair, as we all know some things are kinda hard to make/find as microanalytical standards, and something with the potential for mixed valence like this is always going to be suspicious. Trace metals in metals generally is another but by and large, thats the beauty of the matrix correction!
-
one persons standard is another persons 'where did we find that again?'
;D
to be fair, as we all know some things are kinda hard to make/find as microanalytical standards, and something with the potential for mixed valence like this is always going to be suspicious. Trace metals in metals generally is another but by and large, thats the beauty of the matrix correction!
True, but the matrix correction can only correct for what it knows about. That's why specifying unanalyzed elements is so important:
https://probesoftware.com/smf/index.php?topic=92.0
-
So what could you suggest me to check next?
I recap:
we have 2 commercial Standard blocks with Fe2O3: Taylor block (sorry if it is the fake block, it is out of my responsibility) and SPI block. Also we have "handmade" sample which composition is close to Fe2O3.
3 samples in total (actually we have a pile of samples like last one).
Measuring of the all 3 samples gives us very similar results: Fe is in 1.5-2 %wt. lower than we expect.
What should I check? It is not peaking issue, not dead time, not BG, not APF, not the standard issue (Fe metal), not energy limit... what else?
Thank you very much!
-
Can you do some Mossbauer to determine the Fe+2/Fe+3 ratios?
Try obtaining a synthetic high purity magnetite. That should be close to Fe3O4.
-
I will try to do something in this direction.
But it means that the main version of our issue is wrong compositions of Fe2O3 Standards.
And all 3+ absolutely independent substances (2 - commercial and 1...100+ samples from our lab) have the same issue - total contaminations are close to 2%wt. Also all binary compositions which close to Cu2O, NiO, ... (the topic starts from this) have the same issue. Everything is possible...
I'll update the topic when collect some new information.
Could you suggest me what I need to see and how I should analyze results of different ZAF calculations?
My question unfortunately drowned the discussion
https://probesoftware.com/smf/index.php?topic=1514.msg11698#msg11698
And extra one question - could you suggest the methodology of obtaining a synthetic high purity magnetite? I want to kill any questions to methodology we used )). Our main way is furnace, hang high purity Iron foil, 1200C, mixture of CO, CO2, 3-5 hours. - the same way we did Fe2O3. The only difference in furnace atmosphere.
Thank you!
-
But it means that the main version of our issue is wrong compositions of Fe2O3 Standards.
And all 3+ absolutely independent substances (2 - commercial and 1...100+ samples from our lab) have the same issue - total contaminations are close to 2%wt. Also all binary compositions which close to Cu2O, NiO, ... (the topic starts from this) have the same issue. Everything is possible...
We're all just trying to help figure out what is going on with your Fe2O3 analyses. But thinking about Jon's suggestion a bit more, I have to wonder if Fe+2/Fe+3 ratio cannot be the issue because Fe2O3 should already be all Fe+3, correct? If some of the Fe is Fe+2 then that would give us even less oxygen, right? I am not a mineralogist, so perhaps someone with that expertise could chime in on this question?
But that does still leave the question of OH and H2O as additional contaminants...
Could you suggest me what I need to see and how I should analyze results of different ZAF calculations?
My question unfortunately drowned the discussion
https://probesoftware.com/smf/index.php?topic=1514.msg11698#msg11698
Two points here: first, the results of all the matrix corrections (both ZAF and phi-rho-z) yield low Fe in your Fe2O3 samples relative to your Fe standard, so I don't think matrix corrections are the problem. Second, the historical ZAF corrections, e.g., Philibert, Love-Scott can probably be ignored. The Armstrong and XPP/PAP phi-rho-z corrections are probably the most accurate in general.
What I do find interesting is that in your Evaluate screen capture here:
(https://probesoftware.com/smf/gallery/1_28_03_23_4_18_20.png)
2. I used Evaluate application for evaluate 4 standards with Fe we have in the Taylor block: FeCuS2(28), FeS(16), Fe2O3(33), Fe(32). Sample 7201 - Fe2O3
I note that both your sulfide standards agree within 0.5% absolute with your Fe standard. This suggests to me that you might be doing things mostly correct, but that it's the Fe2O3 standards that are the problem.
And extra one question - could you suggest the methodology of obtaining a synthetic high purity magnetite? I want to kill any questions to methodology we used )). Our main way is furnace, hang high purity Iron foil, 1200C, mixture of CO, CO2, 3-5 hours. - the same way we did Fe2O3. The only difference in furnace atmosphere.
I am not suggesting that you grow your own Fe3O4. Rather I am suggesting that you try to obtain a synthetic single crystal Fe3O4 from a commercial crystal grower and see how that compares. The document attached to this post has a number of crystal growers listed in the appendix:
https://probesoftware.com/smf/index.php?topic=1415.msg10929;topicseen#msg10929
-
I've been running some tests of my own on this question from Rom on his efforts to utilize Fe metal as an Fe primary standard for various Fe oxides where the Fe is present in major concentrations.
As some of you may remember, Rom found that when using Fe metal as a primary standard, he was getting consistently low totals for his Fe oxide standards, but his sulfide standards seem to analyze close to their expected values. I can now report that I am seeing similar effects.
Why could this be? Well as discussed in the posts above, it could be a question of standard accuracy, e.g., his hematite standards could be contaminated with OH or H2O. But I am seeing similar issues with my own magnetite standard and that has been analyzed for Fe using wet chemistry and for ferric-ferrous ratios using colorimetry by Ian Carmichael many years ago.
It is also unlikely to be a matrix correction or background issue, as discussed above. Though it might be worth checking the dead time calibration, though at 20 nA that is unlikely to be a significant effect. So what else could it be?
Now we all know that when analyzing lower Z elements with low energy emission lines where the valence shell is one of the shells involved in the electron transition which produces an x-ray emission, there can be significant peak shape/shift effects from chemical bonding, such that we must usually utilize a primary standard that is at least somewhat similar to our unknown.
For analysis of light elements such as O, N, C or B it is well known that such chemical peak shift/shape effects can be quite large, even when utilizing relatively low resolution modern LDE diffractors. In such cases, the use of a close matrix matched primary standard is necessary, or one can utilize area peak factors (APFs) to account for these chemical bonding effects:
https://probesoftware.com/smf/index.php?topic=536.0
One can also utilize integrated intensities for the acquisition of WDS intensities which can be quite slow, but can handle these peak shift/shape effects automatically. See the Integrated Intensities options in the Elements/Cations dialog:
https://probesoftware.com/smf/index.php?topic=536.msg2992#msg2992
And even in the case of say Mg, Al and Si Ka, we would not want to utilize a metal primary standard when analyzing those elements in oxides or silicates. Instead we would utilize an oxide or silicate standard such as MgO, MgAl2O4 or Mg2SiO4 for Mg Ka analysis of oxides and silicates. An exact matrix match is not necessary for these emission lines, but we can't reliably extrapolate from a pure metal to the oxide/silicate.
However, for higher Z, higher energy emission lines, e.g., Fe Ka, I would have thought that these chemical peak shift/shape effects would be minimal, but perhaps that is not the case. For example, here is a measurement of Fe Ka using Fe metal as a primary standard, analyzing pyrite as a secondary standard:
St 730 Set 1 Pyrite UC # 21334, Results in Elemental Weight Percents
ELEM: Fe Mn Cr Si S Ti
TYPE: ANAL ANAL SPEC SPEC SPEC SPEC
BGDS: LIN LIN
TIME: 40.00 40.00 --- --- --- ---
BEAM: 29.89 29.89 --- --- --- ---
ELEM: Fe Mn Cr Si S Ti SUM
342 46.309 -.007 .000 .000 53.450 .058 99.810
343 46.229 -.012 .000 .000 53.450 .058 99.725
344 46.329 -.039 .000 .000 53.450 .058 99.797
345 46.304 .004 .000 .000 53.450 .058 99.816
346 46.323 .018 .000 .000 53.450 .058 99.849
AVER: 46.299 -.007 .000 .000 53.450 .058 99.799
SDEV: .040 .021 .000 .000 .000 .000 .046
SERR: .018 .010 .000 .000 .000 .000
%RSD: .09 -288.57 .00 .00 .00 .00
PUBL: 46.550 n.a. .000 n.a. 53.450 .058 100.058
%VAR: -.54 --- .00 --- .00 .00
DIFF: -.251 --- .000 --- .000 .000
STDS: 526 525 --- --- --- ---
Here we can successfully extrapolate from Fe metal to FeS2 (as Rom found). Another attempt:
St 730 Set 2 Pyrite UC # 21334, Results in Elemental Weight Percents
ELEM: Fe Mn Cr Si S Ti
TYPE: ANAL ANAL SPEC SPEC SPEC SPEC
BGDS: LIN LIN
TIME: 40.00 40.00 --- --- --- ---
BEAM: 29.89 29.89 --- --- --- ---
ELEM: Fe Mn Cr Si S Ti SUM
387 46.198 -.032 .000 .000 53.450 .058 99.674
388 46.255 -.034 .000 .000 53.450 .058 99.729
389 46.237 .005 .000 .000 53.450 .058 99.750
390 46.361 .024 .000 .000 53.450 .058 99.893
391 46.218 -.016 .000 .000 53.450 .058 99.710
AVER: 46.254 -.010 .000 .000 53.450 .058 99.751
SDEV: .064 .025 .000 .000 .000 .000 .084
SERR: .029 .011 .000 .000 .000 .000
%RSD: .14 -240.04 .00 .00 .00 .00
PUBL: 46.550 n.a. .000 n.a. 53.450 .058 100.058
%VAR: -.64 --- .00 --- .00 .00
DIFF: -.296 --- .000 --- .000 .000
STDS: 526 525 --- --- --- ---
Again within 1% relative accuracy. Now let's analyze the magnetite standard from Carmichael:
St 395 Set 1 Magnetite U.C. #3380, Results in Elemental Weight Percents
ELEM: Fe Mn Cr Si S Al Mg O
TYPE: ANAL ANAL SPEC SPEC SPEC SPEC SPEC SPEC
BGDS: LIN LIN
TIME: 40.00 40.00 --- --- --- --- --- ---
BEAM: 29.90 29.90 --- --- --- --- --- ---
ELEM: Fe Mn Cr Si S Al Mg O SUM
322 69.969 .046 .007 .000 .000 .201 .072 27.803 98.098
323 69.922 .017 .007 .000 .000 .201 .072 27.803 98.021
324 70.188 .036 .007 .000 .000 .201 .072 27.803 98.307
325 70.542 .056 .007 .000 .000 .201 .072 27.803 98.681
326 70.528 .006 .007 .000 .000 .201 .072 27.803 98.616
AVER: 70.230 .032 .007 .000 .000 .201 .072 27.803 98.345
SDEV: .296 .021 .000 .000 .000 .000 .000 .000 .297
SERR: .133 .009 .000 .000 .000 .000 .000 .000
%RSD: .42 64.40 .00 .00 .00 .00 .00 .00
PUBL: 72.080 .054 .007 .000 n.a. .201 .072 27.803 100.217
%VAR: -2.57 -40.75 .00 .00 --- .00 .00 .00
DIFF: -1.850 -.022 .000 .000 --- .000 .000 .000
STDS: 526 525 --- --- --- --- --- ---
Our relative error has now increased to ~2.5% which is unacceptable. Analyzing other Fe silicate (glass) standards we observe the same low accuracy, here for NIST SRM K-412:
St 160 Set 2 NBS K-412 mineral glass, Results in Elemental Weight Percents
ELEM: Fe Mn Cr Si S Mg Ca Al O
TYPE: ANAL ANAL SPEC SPEC SPEC SPEC SPEC SPEC SPEC
BGDS: LIN LIN
TIME: 40.00 40.00 --- --- --- --- --- --- ---
BEAM: 29.89 29.89 --- --- --- --- --- --- ---
ELEM: Fe Mn Cr Si S Mg Ca Al O SUM
352 7.563 .058 .000 21.199 .000 11.657 10.899 4.906 43.597 99.879
353 7.510 .047 .000 21.199 .000 11.657 10.899 4.906 43.597 99.816
354 7.498 .029 .000 21.199 .000 11.657 10.899 4.906 43.597 99.785
355 7.521 .078 .000 21.199 .000 11.657 10.899 4.906 43.597 99.857
356 7.579 .057 .000 21.199 .000 11.657 10.899 4.906 43.597 99.894
AVER: 7.534 .054 .000 21.199 .000 11.657 10.899 4.906 43.597 99.846
SDEV: .035 .018 .000 .000 .000 .000 .000 .000 .000 .045
SERR: .016 .008 .000 .000 .000 .000 .000 .000 .000
%RSD: .46 33.33 .00 .00 .00 .00 .00 .00 .00
PUBL: 7.742 .077 n.a. 21.199 n.a. 11.657 10.899 4.906 43.597 100.077
%VAR: -2.69 -30.21 --- .00 --- .00 .00 .00 .00
DIFF: -.208 -.023 --- .000 --- .000 .000 .000 .000
STDS: 526 525 --- --- --- --- --- --- ---
Again about 2.5% relative low for Fe. I did not have a Mn sulfide standard, but running Mn metal against MnO, I see a similar low accuracy:
St 25 Set 3 MnO synthetic, Results in Elemental Weight Percents
ELEM: Fe Mn Cr Si S O
TYPE: ANAL ANAL SPEC SPEC SPEC SPEC
BGDS: LIN LIN
TIME: 40.00 40.00 --- --- --- ---
BEAM: 29.89 29.89 --- --- --- ---
ELEM: Fe Mn Cr Si S O SUM
347 .008 74.428 .000 .000 .000 22.554 96.990
348 .015 75.138 .000 .000 .000 22.554 97.707
349 .024 74.848 .000 .000 .000 22.554 97.426
350 .026 74.670 .000 .000 .000 22.554 97.250
351 -.002 74.831 .000 .000 .000 22.554 97.384
AVER: .014 74.783 .000 .000 .000 22.554 97.351
SDEV: .012 .261 .000 .000 .000 .000 .262
SERR: .005 .117 .000 .000 .000 .000
%RSD: 81.45 .35 .00 .00 .00 .00
PUBL: n.a. 77.446 n.a. n.a. n.a. 22.554 100.000
%VAR: --- -3.44 --- --- --- .00
DIFF: --- -2.663 --- --- --- .000
STDS: 526 525 --- --- --- ---
About a 3.5% relative error. And here for Mn2SiO4 synthetic olivine:
St 275 Set 2 Mn2SiO4 (manganese olivine) synthetic, Results in Elemental Weight Percents
ELEM: Fe Mn Cr Si S O
TYPE: ANAL ANAL SPEC SPEC SPEC SPEC
BGDS: LIN LIN
TIME: 40.00 40.00 --- --- --- ---
BEAM: 29.89 29.89 --- --- --- ---
ELEM: Fe Mn Cr Si S O SUM
362 .019 52.526 .000 13.907 .000 31.688 98.140
363 .013 52.363 .000 13.907 .000 31.688 97.971
364 .008 52.843 .000 13.907 .000 31.688 98.446
365 .013 52.404 .000 13.907 .000 31.688 98.013
366 .016 52.316 .000 13.907 .000 31.688 97.927
AVER: .014 52.490 .000 13.907 .000 31.688 98.099
SDEV: .004 .212 .000 .000 .000 .000 .209
SERR: .002 .095 .000 .000 .000 .000
%RSD: 28.58 .40 .00 .00 .00 .00
PUBL: .000 54.406 .000 13.907 .000 31.688 100.001
%VAR: .00 -3.52 .00 .00 .00 .00
DIFF: .000 -1.916 .000 .000 .000 .000
STDS: 526 525 --- --- --- ---
Again about 3.5% relative accuracy. So not good for major element accuracy! >:(
Personally I've always utilized pure oxide standards for my primary standards for measurements in oxides and silicates, and so have never observed these effects previously. Has anyone else looked at these peak shift/shape effects for the first transition series elements? Please share some data with us...
Could these low values be from a subtle chemical peak shape/shift effect? I am running some detailed wavescans to check... but I suspect it's the most likely explanation.
But again, remember that these subtle primary standard matrix effects are something that may be important for major elements, but probably not for trace elements as discussed here:
https://probesoftware.com/smf/index.php?topic=610.msg11752#msg11752
A 2 or 3% relative error at 100 PPM is going to produce an absolute error of 2 or 3 PPM, generally not something we are concerned with. Even at 1000 PPM that's only an absolute error of 20 or 30 PPM, so still below the detection limit for many measurements. In other words if you want to measure some first series transition elements at trace levels in an oxide or silicate matrix, and you don't have a suitable pure oxide standard, you are probably OK to use a pure metal standard.
-
I have only one word: oxidation. Especially Pyrite. It is not coincidence that in geology we call it "fool's gold", We should call it "fool's standard" in microanalysis too as it is not suitable even lets say as tertiary standard. The only time I could get good pyrite analyses was then pyrite was polished and without any delay dried under vacuum, and then transfered to probe. Even few hour exposure to atmosphere (or even worse, i.e. warming it up in heater which we do to most of samples to get rid of moisture on its surface) will hasty oxidize the pyrite. And still pyrite and chalcopyrite are widely used as standards in many labs... :(
The metal is a similar story, maybe not so bad as pyrite and chalcopyrite, but still very susceptible (with exception of some noble metals - which are called noble not without a reason).
Try repeating the experiment at 25kV, the discrepancy should be much lesser.
I think synthetic oxides (or other compounds) is the golden spot as they very weakly react with atmosphere, while still having high concentration of target element. Metal while being pure element, is very easily affected by atmosphere. While not used it should be stored in vacuum or in Nitrogen-purged and filled box. Glasses - these are hard to assess how much it reacts with atmosphere (at least on metals we will see growing O Ka peak when looking for peak with lowering acceleration voltages.
-
I have only one word: oxidation.
I thought about oxidation, but I think that dog is barking up the wrong tree. :D
Here's why: if as you claim, the pyrite standard was oxidized and the Fe oxide standards are much less oxidized, then the Fe concentrations in the secondary oxide standards should be closer to the true values than expected. But we observe the opposite. It's the oxides that appear to be 2 or 3 percent (relative) lower than the sulfide (which is actually pretty darn close to the expected value).
Of course it's possible that both the pyrite and Fe metal standards are equally oxidized and that is why the sulfide standard analyzes quite well. But if the Fe metal standard is oxidized, then the secondary oxide standards should analyze "higher" for Fe content than than expected, but as mentioned already, they analyze considerably lower than expected.
But I also agree that a test at 25 keV would be worth doing, or at least a quantitative oxygen measurement. But I'm also going to continue with some detailed wavescans (using a stage increment to avoid beam damage issues).
-
OK, so I ran another test run of Fe and also a Mn oxide and silicate against Fe and Mn metal primary standards and got more results. Here are the Fe standards plotted in the Evaluate application:
(https://probesoftware.com/smf/gallery/395_25_04_23_10_06_12.png)
Key:
160 NIST K-412
162 NIST K-411
396 Chromite (UC # 523-9)
395 Magnetite U.C. #3380
730 Pyrite UC # 21334
526 Fe metal
As you can see, the pyrite standard agrees excellently with the Fe metal primary standard, but the oxide, silicate and glass standards show a consistent error trend (in absolute wt% units) as one would expect if this is due to a systematic error. Here is the analysis of magnetite:
St 395 Set 1 Magnetite U.C. #3380, Results in Elemental Weight Percents
ELEM: Fe Mn Cr Si S Al Mg O
TYPE: ANAL ANAL SPEC SPEC SPEC SPEC SPEC SPEC
BGDS: LIN LIN
TIME: 40.00 40.00 --- --- --- --- --- ---
BEAM: 29.87 29.87 --- --- --- --- --- ---
ELEM: Fe Mn Cr Si S Al Mg O SUM
1821 70.585 .038 .007 .000 .000 .201 .072 27.803 98.706
1822 70.551 .031 .007 .000 .000 .201 .072 27.803 98.665
1823 70.740 .049 .007 .000 .000 .201 .072 27.803 98.872
1824 69.997 .014 .007 .000 .000 .201 .072 27.803 98.094
1825 70.360 .043 .007 .000 .000 .201 .072 27.803 98.486
AVER: 70.447 .035 .007 .000 .000 .201 .072 27.803 98.564
SDEV: .285 .013 .000 .000 .000 .000 .000 .000 .297
SERR: .128 .006 .000 .000 .000 .000 .000 .000
%RSD: .41 38.54 .00 .00 .00 .00 .00 .00
PUBL: 72.080 .054 .007 .000 n.a. .201 .072 27.803 100.217
%VAR: -2.27 -35.41 .00 .00 --- .00 .00 .00
DIFF: -1.633 -.019 .000 .000 --- .000 .000 .000
STDS: 526 525 --- --- --- --- --- ---
Note the roughly 2% (relative) error in the magnetite Fe content using the Fe metal standard. Here is the NIST K-411 mineral glass:
St 162 Set 1 NBS K-411 mineral glass, Results in Elemental Weight Percents
ELEM: Fe Mn Cr Si S Mg Ca Al O
TYPE: ANAL ANAL SPEC SPEC SPEC SPEC SPEC SPEC SPEC
BGDS: LIN LIN
TIME: 40.00 40.00 --- --- --- --- --- --- ---
BEAM: 29.87 29.87 --- --- --- --- --- --- ---
ELEM: Fe Mn Cr Si S Mg Ca Al O SUM
1811 10.866 .112 .000 25.382 .000 8.847 11.057 .053 43.558 99.875
1812 10.816 .096 .000 25.382 .000 8.847 11.057 .053 43.558 99.809
1813 10.745 .085 .000 25.382 .000 8.847 11.057 .053 43.558 99.727
1814 10.892 .087 .000 25.382 .000 8.847 11.057 .053 43.558 99.876
1815 10.857 .103 .000 25.382 .000 8.847 11.057 .053 43.558 99.858
AVER: 10.835 .096 .000 25.382 .000 8.847 11.057 .053 43.558 99.829
SDEV: .057 .011 .000 .000 .000 .000 .000 .000 .000 .063
SERR: .026 .005 .000 .000 .000 .000 .000 .000 .000
%RSD: .53 11.87 .00 .00 .00 .00 .00 .00 .00
PUBL: 11.209 .077 n.a. 25.382 n.a. 8.847 11.057 .053 43.558 100.183
%VAR: -3.33 25.29 --- .00 --- .00 .00 .00 .00
DIFF: -.374 .019 --- .000 --- .000 .000 .000 .000
STDS: 526 525 --- --- --- --- --- --- ---
And the NIST K-412 mineral glass:
St 160 Set 1 NBS K-412 mineral glass, Results in Elemental Weight Percents
ELEM: Fe Mn Cr Si S Mg Ca Al O
TYPE: ANAL ANAL SPEC SPEC SPEC SPEC SPEC SPEC SPEC
BGDS: LIN LIN
TIME: 40.00 40.00 --- --- --- --- --- --- ---
BEAM: 29.86 29.86 --- --- --- --- --- --- ---
ELEM: Fe Mn Cr Si S Mg Ca Al O SUM
1806 7.534 .091 .000 21.199 .000 11.657 10.899 4.906 43.597 99.884
1807 7.562 .062 .000 21.199 .000 11.657 10.899 4.906 43.597 99.882
1808 7.562 .073 .000 21.199 .000 11.657 10.899 4.906 43.597 99.893
1809 7.563 .064 .000 21.199 .000 11.657 10.899 4.906 43.597 99.886
1810 7.555 .083 .000 21.199 .000 11.657 10.899 4.906 43.597 99.896
AVER: 7.555 .075 .000 21.199 .000 11.657 10.899 4.906 43.597 99.888
SDEV: .012 .012 .000 .000 .000 .000 .000 .000 .000 .006
SERR: .006 .006 .000 .000 .000 .000 .000 .000 .000
%RSD: .16 16.64 .00 .00 .00 .00 .00 .00 .00
PUBL: 7.742 .077 .000 21.199 .000 11.657 10.899 4.906 43.597 100.077
%VAR: -2.41 -2.95 .00 .00 .00 .00 .00 .00 .00
DIFF: -.187 -.002 .000 .000 .000 .000 .000 .000 .000
STDS: 526 525 --- --- --- --- --- --- ---
Again relative errors around 2 to 3% using Fe metal as the primary standard. Remember, as mentioned previously, if we were concerned that the Fe metal standard was oxidized, then we would predict that these Fe secondary standards would all analyze higher than expected. Though as suggested by SG, we should also run them at 20 keV to minimize surface effects, which I will try to do as soon as I get another crack at the instrument.
I also ran a Mn oxide and silicate against Mn metal and see similar results, here again displayed in the Evaluate application:
(https://probesoftware.com/smf/gallery/395_25_04_23_10_21_28.png)
Key:
160 NIST K-412
162 NIST K-411
396 Chromite (UC # 523-9)
395 Magnetite U.C. #3380
275 Mn2SiO4 synthetic
25 MnO
525 Mn meta
Unfortunately I don't have a Mn sulfide standard to try, but again we see a similar error trend in the Mn oxide and silicate, though even slightly larger than for the Fe analyses, of around 3 or 4% relative error using Mn metal as a primary standard. Here is the Mn oxide standard:
St 25 Set 1 MnO synthetic, Results in Elemental Weight Percents
ELEM: Fe Mn Cr Si S O
TYPE: ANAL ANAL SPEC SPEC SPEC SPEC
BGDS: LIN LIN
TIME: 40.00 40.00 --- --- --- ---
BEAM: 29.87 29.87 --- --- --- ---
ELEM: Fe Mn Cr Si S O SUM
1801 .025 74.462 .000 .000 .000 22.554 97.041
1802 .003 74.412 .000 .000 .000 22.554 96.969
1803 -.005 74.352 .000 .000 .000 22.554 96.901
1804 .002 74.236 .000 .000 .000 22.554 96.793
1805 .017 74.113 .000 .000 .000 22.554 96.684
AVER: .008 74.315 .000 .000 .000 22.554 96.878
SDEV: .012 .141 .000 .000 .000 .000 .142
SERR: .005 .063 .000 .000 .000 .000
%RSD: 143.58 .19 .00 .00 .00 .00
PUBL: .000 77.446 .000 .000 .000 22.554 100.000
%VAR: .00 -4.04 .00 .00 .00 .00
DIFF: .000 -3.131 .000 .000 .000 .000
STDS: 526 525 --- --- --- ---
And here the Mn silicate:
St 275 Set 1 Mn2SiO4 (manganese olivine) synthetic, Results in Elemental Weight Percents
ELEM: Fe Mn Cr Si S O
TYPE: ANAL ANAL SPEC SPEC SPEC SPEC
BGDS: LIN LIN
TIME: 40.00 40.00 --- --- --- ---
BEAM: 29.85 29.85 --- --- --- ---
ELEM: Fe Mn Cr Si S O SUM
1816 .004 52.066 .000 13.907 .000 31.688 97.664
1817 .004 52.123 .000 13.907 .000 31.688 97.722
1818 .020 52.390 .000 13.907 .000 31.688 98.005
1819 .005 52.425 .000 13.907 .000 31.688 98.025
1820 .001 52.584 .000 13.907 .000 31.688 98.180
AVER: .007 52.318 .000 13.907 .000 31.688 97.919
SDEV: .007 .218 .000 .000 .000 .000 .218
SERR: .003 .097 .000 .000 .000 .000
%RSD: 113.08 .42 .00 .00 .00 .00
PUBL: n.a. 54.406 n.a. 13.907 n.a. 31.688 100.001
%VAR: --- -3.84 --- .00 --- .00
DIFF: --- -2.088 --- .000 --- .000
STDS: 526 525 --- --- --- ---
So what could be causing this? Next I will show some careful wavescans that might indicate the source of the problem...
-
So, to summarize the previous post in this topic, we appear to see a systematic error when analyzing oxide and silicate Fe and Mn samples using an Fe metal primary standard. And this error appears to be somewhat larger for Mn Ka compared to Fe Ka. However, as Rom reported earlier, we do not see these accuracy errors when analyzing Fe sulfides using an Fe metal standard as seen here:
(https://probesoftware.com/smf/gallery/395_26_04_23_9_22_09.png)
Now let's examine some high precision wavescans I did over the weekend starting with Fe Ka on the standards containing major concentrations of Fe:
(https://probesoftware.com/smf/gallery/395_25_04_23_10_37_04.png)
These scans were run in my spec3 LLIF crystal (2 atm P-10) at 15 keV and aside from the strange artifacts in all the scans which we will discuss later, they all seems to be pretty consistent. Let zoom in a bit:
(https://probesoftware.com/smf/gallery/395_25_04_23_10_37_23.png)
The artifacts are more visible and I've already asked several colleagues off-line as to what these may be from, as we do not seem them in the other spectrometers, but I'm guessing that they are some sort of mechanical issue. These are step/count scans (not ROM or continuous) using a step size of 2 sin theta units and 40 second count time for each point.
But again, the scans appear to be very consistent with no obvious differences between the Fe metal and the other secondary standards. The Fe metal being the cyan triangle symbols.
But now let's look at the Mn Ka scans on Mn metal and a MnO and Mn2SiO4 secondary standards:
(https://probesoftware.com/smf/gallery/395_25_04_23_10_37_41.png)
There appears to be slight shift between the metal and the oxidized standards, and zooming in we can this shift a bit better:
(https://probesoftware.com/smf/gallery/395_25_04_23_10_38_01.png)
I don't know if this explains the discrepancy we are seeing in the Fe and Mn oxide and silicate standards when extrapolating from metal standards, but it's worth considering. I guess even knowing that Mg, Al and Si ka show significant peak shifting between metal and oxide chemistry, I'm just surprised that Fe Ka and Mn Ka would still experience a detectable peak shift. I haven't tried this same test using a PET crystal, but I'll bet the peak shift for Fe and Mn Ka would almost certainly be less detectable.
What do you all think?
-
Thought I'd add to this as I had the machine to myself for a change and I've seen something similar before on both our Cameca when working on wustite/magnetite samples standardised to either Fe metal or hematite, and on our JEOL when working with steel samples.
I ran a few WDS scans overnight on our SX100 on Fe metal, hematite Fe2O3 and fayalite Fe2SiO4. I used the same conditions John mentions: 15 kV, step size of 2 sin theta units and 40s count time. I used a 20 nA beam defocused to 5 microns. I didn't peak centre first, hence the offset.
Here's the whole scan:
(https://probesoftware.com/smf/gallery/796_27_04_23_2_46_51.jpeg)
And zoomed in to a similar region as above:
(https://probesoftware.com/smf/gallery/796_27_04_23_2_47_12.jpeg)
Couple of things to note:
- I can also see the "lumps", but much less pronounced
- the Fe Ka1 peak seems much more peak like - there isn't a flat top on mine?
- there doesn't seem to be a significant difference between the Fe Ka1 (K-L3) emissions for the various materials I've scanned
- there does seem to be a difference in the Fe Ka2 (K-L2) position as a function of bonding environment
I mentioned that I was looking at steels on our JEOL - a few years ago, I was struggling to get decent totals between the samples and our standards, and did a few wavelength scans to work out what was going on. This is what I found:
(https://probesoftware.com/smf/gallery/796_27_04_23_3_36_22.jpeg)
That's two WDS scans on each phase, going standard-sample-standard-sample, using the same conditions on the same spectrometer. Count rates weren't particularly different between sample/standard. Wasn't expecting such a shift in the Fe Ka, and never really got to the bottom of what caused it!
-
Interesting that you are seeing these same "lumps" on your Cameca. I am going to ask our instrument engineer to clean the spectrometer gearing and see if that helps.
So the JEOL scans were performed on your Fe metal standard and another Fe metal standard? If I saw such a shift like that I would assume it's some sort of mechanical glitch...
I would also be very interested in seeing if you can reproduce the systematic analytical errors we are seeing using Fe metal as a primary standard, and analyzing Fe oxide versus Fe sulfide secondary standards. Mn and/or Cr metal and standards also if you get a chance.
-
Hmm... I think I can have partial lead, albeit I would not rule out some other fundamental workings from the equation.
First of all,
JonF, what was that Fe phase (edit: yeah I get it - its steel, but metallic Fe tends to form some Fe minerals where some are magnetic) and how big it was?
I think the moral the story below will be "standards should be big, but not too big, especially magnetic standards".
So this year we got to do analysis of metallic meteorite sample as polished slab (not thin-section) something like 2.5cm x 4cm with about 1.5cm thickness.
My colleague asked me to take a look, as totals of analyses was "quite too far" from the expected analytical 100% (that was on SX100 but I don't think this matter). Then I noticed that optical image is extremely shifted from what we see on electronic image. The problem was that this Fe in meteorite formed some magnetic minerals, which because of huge thickness was producing strong enough magnetic field to significantly shift the beam up to and more than 100 µm!!! We looked through the sample and there had to be different crystals with different magnetic orientation (image shift relative to optical had different directions depending from place on the thin section), we looked for least "image-shifted" (or rather magnetically shifted) spot, and used microprobes built-in beam shift capability to compensate that shift further more and then we got analytical totals close to 100%.
WDS is very sensitive to geometry (that is why I am absolutely skeptical about replacing wire-based proportional counter with SDD based counter. Unless such SDD could be made as thin 100µm wide long (>2cm) wire?). That is advantage (better spectral resolution) but also disadvantage (if e-beam position is not guaranteed).
Just think about it. It is enough to move stage in Z direction out from Rowland circle few µm to get systematic lower totals. Thus shifting beam for more than 10µm in some directions can also move the geometry from perfect Rowland's circle significantly.
This also explains, why often verification of spectrometer "is our best Friend" - beam can drift during day in some semi-circle due to natural daily magnetic field fluctuations. Verification of spectrometers is important even if no crystal flip was done and from my experience in most of low analytical total situation it can fix that.
BTW special caveat is needed for labs which has huge solar panels installed (going greeeeeeeeeeeeen, huh!) on roof or is planned to be installed - Your probe is at a risk of being influenced by a huge DC field during the day, which can't be compensated anyhow (only nights are warrantied to be safe). In such unfortunate case the X-ray registration is not only susceptible to atmospheric front (low pressure GFPC sensitivity to humidity and atmospheric pressure changes), but also sensitive to clouds passing-by. Way to go :P.
However, this does not explain observations of probeman, as shift in his case was very minimal. I think there are some fundamental areas still not well understood or overlooked:
1. not efficient or not existent self-absorption correction
2. diverging probabilities of beam electron collision with different shell electrons depending from valence electron state of metal atom
3. the lax approach with dead time correction when historically collecting the database of MAC's - biased matrix correction.
...something else...
-
Hmm... I think I can have partial lead, albeit I would not rule out some other fundamental workings from the equation.
...
I think the moral the story below will be "standards should be big, but not too big, especially magnetic standards".
When I first read this I thought to myself, OK, this is interesting. By the way, here is a discussion on dealing with magnetic specimens in the microprobe, because these magnetic sample/beam deflection effects are quite real and can be quite serious especially with large samples:
https://probesoftware.com/smf/index.php?topic=354.0
Because I could see a situation where my primary Fe metal standard was slightly magnetic, and therefore the beam was deflected from its normal position. If one then aligned their peak position to this deflected beam position on the Fe metal primary standard, and then one went to a non-magnetic sample, I could imagine that the intensity on the non-magnetic secondary standard could be lower than expected, due to being beam moving back towards it's normal position, but now out of the previous Rowland circle geometry peaked on the primary Fe metal standard.
But then, thinking about it a bit more, I think this hypothesis fails for two reasons. First, my Fe metal standard is a 0.1mm wire that is mounted vertically and cross sectioned, so I think the magnetic field from this standard is quite minimal. Second, and more importantly, the pyrite standard is non-magnetic just as the oxides and silicates, yet pyrite analyzes perfectly using the Fe metal primary standard.
2. diverging probabilities of beam electron collision with different shell electrons depending from valence electron state of metal atom
Yeah, I've got to believe that this is some valence issue producing slight peak shape/shift effects.
Is it worth thinking about these valences effects and why Fe metal and FeS2 would have similar valence effects compared to oxides and silicates?
-
Oh probeman, You make me sound again like "dog barking to wrong tree"...
However, this does not explain observations of probeman, as shift in his case was very minimal. I think there are some fundamental areas still not well understood or overlooked:
I said that myself that this magnetic effect is rather not main cause for your observed discrepancies. My message was more aimed to JonF, and his graphics. Why, because metallurgic samples tends to be bigger pieces of metal, not thin sections. Also people tending to do a lot of analysis of metallurgy samples tends to have a custom made standard - metalurgic one of course. The first thing, which raises for me a red flag that in JonF situation this was a magnetic effect, is the identical shape of these peaks, where in valence shift we will have top shifted as there will be some minor energy drift of one of Ka1, Ka2, Ka3 and also its proportions can get distorted (Probably Ka3 the mostly) - Valency shifts often are also clear peak distortion. Also from JonF description of situation it looks like these were metalurgical samples.
Alongside magnetic effects another general caveat for analytical geometry is that apertures should be centered as much as possible across working range of currents! Because if not beam at 5nA and beam at 100nA can have significantly shifted peaks. In some cases (our FEG ) has far from ultimately precise centered column assembly, and it is impossible to achieve ideal beam position stability across 1 to 1000nA. The subtilities won't be seen in small area beam scan based mapping due to statistical noise, but wavescans can catch these small differences from my experience.
Yes I think I had read somewhere that at some physical conditions pyrite can behave a bit like metal (from conductivity perspective), it probably translates in similar workings at atomic level - that would explain why there is similarity... if valency having its workings here is the hypothesis on right track. BTW, while I am taking information in wikipedia with grain of salt, Pyrite has nice page there and indeed it stated there that it is semiconductor. It gets even more interesting at "Research" section, where it is stated that some researchers observed that pyrite can have voltage-induced transformation into ferromagnetic material... (Ok lets bark onto other tree, maybe this time this will be the right one :) ) I looked up the referenced sources - it looks they need some Ionic liquid (what ever it is) and applies 1V to do the transformation (which is reversible btw). While we have no ionic liquid in EPMA (I guess :o) we have much more than 1V :P beam. What if in reality this non magnetic mineral puts its "magnetic troll-face" when we hit it with beam and takes it off when we stop the beam? Could be that "Fool's gold" is "fooling around" with us :D
-
Hey, no worries, we're all just hypothesizing away! ;D
So until we have more/better data, I'd simply make the following tentative suggestion, and that is to only use pure metal standards when measuring major elements in alloys and sulfides, and stick with (high purity, synthetic) oxide/silicate standards for oxides and silicates.
Of course for trace elements a 2 or 3% relative error won't even be detectable, as previously described here:
https://probesoftware.com/smf/index.php?topic=610.msg11752#msg11752
so feel free to use, say, a pure Ni or Mn metal standard for trace elements in olivine/pyroxene. Just don't do as we've sometimes seen a few "analysts" do, using San Carlos olivine as a primary standard for Ni:
https://probesoftware.com/smf/index.php?topic=610.msg11833#msg11833
:o
-
I am very appreciate your efforts in this direction.
Some addition things which might give somebody good ideas. In all cases of the list lower the Standards - 100% metal.
1. Incorrect result of measuring metal in oxide/silicate we obtain for many (I suppose for all) metals.
2. More often the measured concentration of metal (Fe, Ni, Pb,....) in its oxide/silicate is lower then we expected but sometimes higher (Sb, Cu).
3. Differences between measured and expected (published) concentration can be from ~1-2 rel% (Sb in oxide) to ~2 rel% (Fe in oxides, Ni in silicates) and more: ~2-3 rel% (Pb in oxide, silicate).
4. Delta between measured and published concentration depends of electron energy (10-20 keV range). Energy increase - Metal relative intensity in oxide (I unknown, cps / I standard, cps) decrease. This dependency is more evident for K lines (Fe, Ni) and less evident or close to zero for L lines (Fe, Cu).
5. The incorrect result doesn't depend of wrong BG or peak shift. So it depends from the substance.
There was only my observations, not more. Firstly I accused in this issue secondary fluorescence but overdetermination of Sb and may be Cu in their oxides is opposite this assumption.