Interference Corrections on Standards

[MatSci]

Hello All,
  This is beginner question:  I am analyzing a 16 component glass, and have a run that consists of 25 standards and 6 unknowns.  I analyzed extra standards so I could use them for interference corrections and to also help collect sufficient data for MAN background corrections.  I collected standards before and after each unknown.  I collected all of the data last week, and now I am going through the analysis process.  The data from the standards looks reasonably close to the "published values", but there are some obvious outliers - elements reported that don't exist, or concentrations that are off from the reported values (e.g. B and O).

Question 1:  As part of the analysis process, is one supposed to go through all of the measurements on all of the standards and perform interference corrections to address discrepancies between the measured and the reported values for the standards? 

Question 2: What about selecting different reference standards for the same element on the different standards in my run?  Is that analytically appropriate?  Is it necessary?  For example, I'm trying to measure oxygen.  I have selected a fully-characterized multi-component glass as my reference standard for oxygen for all of the standards and all of the unknowns in my run.  In some cases when measuring the concentration of oxygen in a mineral standard (e.g. Albite), I obtain better oxygen values when using another mineral standard (e.g. Orthoclase) as the reference for oxygen in Albite, as compared to using my glass standard or a simple oxide like Al2O3 as the oxygen standard for measuring oxygen in Albite.  I'm assuming the variations in local chemical environment are responsible for the variations in calculated concentrations.

Question 3: If making interference corrections and adjustments to the measurements on the standards is an appropriate practice, what is the easiest way to apply the interference corrections and variations in reference standard to all of the sets of standard measurements in my run?  (I have 6 sets of standards measurements in this run).  I'm assuming that the "Set-up" feature is the path to pursue, but I haven't used that before, and I'm not sure if I need to use an "Elemenl Set-up" or a "Sample Set-up".

Thank you for your patience.  And, I apologize if this is an obvious question.

[John Donovan]

Hello All,
  This is beginner question:  I am analyzing a 16 component glass, and have a run that consists of 25 standards and 6 unknowns.  I analyzed extra standards so I could use them for interference corrections and to also help collect sufficient data for MAN background corrections.  I collected standards before and after each unknown.  I collected all of the data last week, and now I am going through the analysis process.  The data from the standards looks reasonably close to the "published values", but there are some obvious outliers - elements reported that don't exist, or concentrations that are off from the reported values (e.g. B and O).

Question 1:  As part of the analysis process, is one supposed to go through all of the measurements on all of the standards and perform interference corrections to address discrepancies between the measured and the reported values for the standards? 

Hi Brad,
These are excellent questions, that should be discussed more often.  It really comes down to "how confident can we be about accuracy in EPMA"?  I'm going to start at the beginning, so bear with me while I discuss basic things I know you already understand quite well.

The fact that we can analyze each standard (as though it were an unknown), which we have acquired data for, is I think, one of the more powerful features in Probe for EPMA.  The idea being that if our standards don't agree with each other, why even bother looking at actual unknown samples?

So, yes, one should definitely examine each standard that we've acquired in our probe runs, to be sure that not only does the analyzed value agree with the published value (assuming that we have a secondary standard to check against!). But also, if the standard database reports a zero concentration for that element, can we actually measure a zero concentration for that element? Remembering that the primary standard is the standard assigned to that element, and a secondary standard is any standard that contains a non-zero concentration of the element, that is not assigned as the primary standard for that element. These standard assignments are on an element by element basis so a standard can be the primary standard for some elements, and also a secondary standard for other elements.

So there are really two questions here: one, using a secondary standard that is (ideally) similar to our unknown matrix, how close are we to the published value? This is the accuracy check for EPMA. Here is an example of a secondary standard sample, being analyzed from the Analyze! window:

ELEM:       Na      Si       K      Al      Mg      Ca      Ti      Mn      Fe       P      Cr   SUM 
   365    .037  21.171    .007   4.807  11.702  10.871   -.009    .050   7.870    .015    .012 100.130
   366    .044  21.212    .012   4.798  11.702  10.861    .014    .062   7.635    .008    .009  99.955
   367    .042  21.316    .016   4.822  11.638  11.074    .011    .069   7.697    .018    .004 100.304

AVER:     .041  21.233    .012   4.809  11.681  10.935    .005    .060   7.734    .014    .009 100.130
SDEV:     .004    .075    .004    .012    .037    .120    .013    .010    .122    .005    .004    .175
SERR:     .002    .043    .003    .007    .021    .069    .007    .006    .070    .003    .002
%RSD:     8.65     .35   37.11     .25     .31    1.10  246.35   16.28    1.57   37.98   45.57

PUBL:     .043  21.199    n.a.   4.906  11.657  10.899    n.a.    .077   7.742    n.a.    n.a. 100.120
%VAR:    -4.54     .16     ---   -1.98     .20     .33     ---  -21.72    -.10     ---     ---
DIFF:    -.002    .034     ---   -.097    .024    .036     ---   -.017   -.008     ---     ---
STDS:      336     162     374     336     162     162      22      25     162     285     396

In this case we are using the NIST K-411 SRM (std #162) as a primary standard, to analyze the NIST K-412 SRM (std #160) as a secondary standard. The lines highlighted in red are the most important for our evaluation of accuracy.

Specifically the AVER: line is what the secondary standard analyzes as in elemental weight percent. Note also in the line labeled STDS: that K-411 (#162) is the assigned (primary) standard for Si, Mg, Ca, Fe. So next we look at the PUBL: line which is what our standard database is telling what we *should* be getting.  To evaluate our accuracy, PFE conveniently displays the %VAR: line, which is the relative error between the AVER: line and the PUBL: line. The DIFF: line being the absolute difference between the two. Typically in EPMA we'd like to see accuracy in the 1 to 3% range, ideally below 2% relative accuracy.

What about our primary standard for Si, Mg, Ca and Fe:

ELEM:       Na      Si       K      Al      Mg      Ca      Ti      Mn      Fe       P      Cr   SUM 
   368    .014  25.269    .008    .016   8.873  10.995    .017    .117  11.210    .017   -.007 100.087
   369    .003  25.547    .008    .011   8.875  11.167   -.006    .109  11.324   -.001   -.019 100.575
   370    .008  25.357    .007    .020   8.833  11.077   -.029    .081  11.124    .007    .001 100.045
   371    .007  25.329    .001    .019   8.804  11.014    .006    .122  11.176    .007   -.001 100.043

AVER:     .008  25.376    .006    .016   8.846  11.063   -.003    .107  11.209    .007   -.007 100.187
SDEV:     .005    .120    .003    .004    .034    .077    .020    .018    .084    .007    .009    .259
SERR:     .002    .060    .002    .002    .017    .039    .010    .009    .042    .004    .005
%RSD:    57.93     .47   56.19   25.65     .38     .70 -643.83   16.86     .75   96.67 -138.78

PUBL:     n.a.  25.382    n.a.    .053   8.847  11.057    n.a.    .077  11.209    n.a.    n.a. 100.183
%VAR:      ---  (-.02)     ---  -69.51  (-.01)   (.06)     ---   39.36   (.00)     ---     ---
DIFF:      ---  (-.01)     ---   -.037   (.00)   (.01)     ---    .030   (.00)     ---     ---
STDS:      336     162     374     336     162     162      22      25     162     285     396

Note that %VAR: and DIFF: values for those elements are in parentheses. This is to remind the user that these elements are assigned as the primary standard to this standard, so of course they look good!  But note that the other elements (Na, K, Ti, P and Cr) in both standards are also a check for accuracy, however with the exception of Al, the other elements are essentially trace elements.  So what accuracy is this checking? This is checking the accuracy of our background measurement! 

That is, if the PUBL: value for an element is listed as zero (often n.a. for not analyzed), we should observe a value that is statistically zero. That is within a few standard deviations of zero as we can see for example in the case of K which is within 2 standard deviations of zero.

But not only does this check our background measurement, but to (finally!) get to your question it also checks for possible interferences from other elements. That is, if we are getting a non-zero value for an element which should be zero, we need to check for a possible interference using the Standard Assignments dialog as discussed here:

https://probesoftware.com/smf/index.php?topic=626.msg8019#msg8019

Now there's also the possibility that this standard has a small contamination of the measured element, so we need to "know" our standards to "trust" them. 

Question 2: What about selecting different reference standards for the same element on the different standards in my run?  Is that analytically appropriate?  Is it necessary?  For example, I'm trying to measure oxygen.  I have selected a fully-characterized multi-component glass as my reference standard for oxygen for all of the standards and all of the unknowns in my run.  In some cases when measuring the concentration of oxygen in a mineral standard (e.g. Albite), I obtain better oxygen values when using another mineral standard (e.g. Orthoclase) as the reference for oxygen in Albite, as compared to using my glass standard or a simple oxide like Al2O3 as the oxygen standard for measuring oxygen in Albite.  I'm assuming the variations in local chemical environment are responsible for the variations in calculated concentrations.

Yes, you are exactly correct. This is because the energy distribution of low Z emission lines are affected by the chemical state of the other elements bonded to that low Z element.

So in additional to all the the above issues regarding accuracy, oxygen and other low energy emission lines are subject to chemical state issues, which can affect the accuracy because the peak intensity is not accurately characterizing the elemental concentration. This is the idea behind Bastin's Area Peak Factors (APFs):

https://probesoftware.com/smf/index.php?topic=536.msg2993#msg2993

One can also investigate Area Peak factors as described here, but it's ma bit of a research project:

https://probesoftware.com/smf/index.php?topic=536.msg2946#msg2946

Another alternative for APFs is to utilize the "integrated peak intensities" acquisition option. This tells Probe for EPMA to acquire a full scan on each peak, as opposed to just the peak intensity. That is to utilize the integrated intensity as opposed to just the peak intensity to deal with chemical states causes shape changes in the peak. This option is here in the Elements/Cations dialog:



The basic idea is to acquire a scan from the low off-peak to the high off-peak to capture the entire integrated intensity of the peak.  It works! But it is more time consuming than just measuring the peak intensity.  Here is an example measuring sulfur which is a somewhat different animal because changes in the sulfur chemical state cause a change in the sulfur peak *position*, but not their *shape*:

https://probesoftware.com/smf/index.php?topic=733.msg4660#msg4660

Question 3: If making interference corrections and adjustments to the measurements on the standards is an appropriate practice, what is the easiest way to apply the interference corrections and variations in reference standard to all of the sets of standard measurements in my run?  (I have 6 sets of standards measurements in this run).  I'm assuming that the "Set-up" feature is the path to pursue, but I haven't used that before, and I'm not sure if I need to use an "Elemenl Set-up" or a "Sample Set-up".

Thank you for your patience.  And, I apologize if this is an obvious question.

If you're simply adding, removing or changing interference corrections to the standards (and hopefully also to the unknowns), just use the Select All button in the Analyze! window when the All Samples option is selected.

As for acquisition of new samples, the software automatically utilizes previously assigned interferences in the last unknown, or from the Automate! based on the Sample Setup, if Sample Setups were assigned there.

I probably missed some important questions in your post, so please feel free to ask further questions.  Getting really accurate data isn't easy, but it can be done. Getting good data on oxygen and other low energy emission lines is even more challenging.

You might also want to investigate the application of empirical MACs as described here:

https://probesoftware.com/smf/index.php?topic=8.msg5258#msg5258

Finally, here is a link to a report I wrote up in my attempts to measure boron in various magnesium boride copmpounds. It might provide you with some ideas:

https://epmalab.uoregon.edu/reports/Preliminary%20work%20on%20MgB2%20and%20MgB4.pdf

[Anette von der Handt]

Hi Brad,

I agree with John and I would say that you should always correct for your interferences: if you need to use an interfered standard as a primary standard for the interfered element you obviously need to correct its intensity, if you are using the interfered standard as a secondary standard it allows you to assess the accuracy of the interference correction. In both cases valuable things to do.

To add to your question 3:
Whenever you need to apply certain settings (standard assignments/interference corrections, calculation options, disable elements, and so forth) post-analysis and you only want to apply changes to a subset of your analyses (potentially dispersed throughout your run): if you select multiple entries, PFE will apply the current settings in the last unknown of the list to all the others in whatever option you have opened. I usually start processing from the bottom up so that I can make use of this and use my lowest as a template.

If you want to use variable settings for your analyses (like standards versus unknowns or you have very different phases) you want to use sample set-ups and you have to create them by each analysis setup in the "Analyze" window and clicking the yellow "Add to setups". The "Save setups" slightly confusingly, would create element setups for all the elements in that analysis and add them to the global element database. You can either load in a sample set-up then when you create a new unknown in the Acquire window for one-by-one analysis or you can assign sample setups to digitized position in the Automate window.

The difference between sample setup and element setup in a nutshell: Element set-up is just the settings for an element X-ray line on a particular spectrometer and would include background positions, detector settings etc. Essentially everything that you set in the "Elements/Cations" dialog. Sample set-up goes beyond that and would include, in addition to the element conditions also standard assignments, acquisition options etc. Essentially everything that you set in the "Acquire" window.