Offset (Interpolated) MAN backgrounds

[BenH]

For those of us that don't have a perfect set of standards I would like to request something strange.

In many cases my standards have minor and element impurities in them that happen to be elements that I am interested in.  That makes the standard completely useless for a MAN backgrounds.  In yet other cases there are spectral interferences that make the standard useless for MAN curves. 

One way to get around this is to use off peak analyses to avoid the minor element impurities or the spectral interferences.  The MAN value would simply be the interpolated background value.  Would this be possible to implement?

[Probeman]

For those of us that don't have a perfect set of standards I would like to request something strange.

In many cases my standards have minor and element impurities in them that happen to be elements that I am interested in.  That makes the standard completely useless for a MAN backgrounds.  In yet other cases there are spectral interferences that make the standard useless for MAN curves. 

One way to get around this is to use off peak analyses to avoid the minor element impurities or the spectral interferences.  The MAN value would simply be the interpolated background value.  Would this be possible to implement?

Hi Ben,
So you saying, measure normal off-peaks on the MAN standards and utilize the interpolated backgrounds for the MAN curve in order to perform MAN corrections on unknowns? That's a really interesting idea as it would avoid issues where the MAN standard does not contain a zero concentration of the element, or has on-peak interferences for that emission line on the MAN curve.  Here are some nice examples of such contamination and interference issues on MAN curves:

http://probesoftware.com/smf/index.php?topic=4.msg1070#msg1070

http://probesoftware.com/smf/index.php?topic=4.msg5136#msg5136

http://probesoftware.com/smf/index.php?topic=4.msg5295#msg5295

That is one nice thing about the MAN method, it lets one know if one's standards are truly clean or not!

Not sure how hard it would be to implement, but here's an even stranger idea that doesn't require any code changes: simply acquire your MAN standards with the on-peak positions slightly to one side of the actual peak positions.  It might not work at low spectrometer angles where the continuum is highly sloped, but at medium to high sin thetas, the backgrounds are relatively flat and would probably give an accurate measurement of the continuum intensities for the MAN fit.

In other words, since Probe for EPMA allows different sample setups for the standards and unknowns you would acquire MAN backgrounds for both your MAN standards and your unknowns, but the on-peak positions would be "offset" for the MAN standards, while the unknowns (and primary standards) would utilize the correct on-peak positions for the MAN corrected samples.

Or you could break down and purchase some decent pure element and oxide standards and do the MAN backgrounds properly!   
john

[John Donovan]

One problem with the above idea for using "offset" on-peak positions for the MAN standards and normal on-peak positions for the unknowns (and primary stds), is that you won't be able to use any standards in the MAN fit, that are also used as primary standards.

Normally PFE automatically uses a primary standard as an MAN standard if the element in question is not present.  So for example, one can use FeS2 as a primary standard for Fe or S, but one can also use it as a MAN standard for trace elements like Ag, Cu and Ag, if they are not present in detectable amounts. So having a standard acquired with both "offset" and normal on-peak positions is a problem with Probeman's idea.

Basically I think Ben's original idea for using the interpolated off-peak intensities to create the MAN calibration curve is a good idea if I can think of a way to implement it.  One can create an "off-peak" MAN curve easily by simply setting the peak intensity equal to the interpolated background intensity.  But when it comes to calculating the unknown composition we don't want to overwrite the on-peak intensity with the background!

This will take some more thought...
john

[John Donovan]

I still need to think about how to apply Ben Hanson's idea to normal samples, but I was able to modify the MAN calibration dialog code so that it loads either normal on-peak (only) intensities for the MAN calibration curve for off-peak sample acquisitions as seen here:



or, loads the interpolated off-peak background intensities for the MAN calibration curve as seen here:



One interesting thing is that the variance is slightly worse for the interpolated off-peak MAN intensities compared to the on-peak only (off-peak) MAN intensities, as expected, since the interpolated intensity statistics are worse  (1/2 the counting time for each off-peak, compared the on-peak counting time)

I think this is going to work...  what a great idea from Ben Hanson!    It's very cool to see a new approach that should avoid not only problems with standards that are contaminated with trace amounts of the element of interest, but also have on-peak spectral interferences.  The point being that once this off-peak interpolated MAN calibration curve is done, one can simply measure on-peak (MAN) samples (unknowns) and get better statistics in 1/2 the time.

[BenjaminWade]

I concur...this is a great idea as I run into this sometimes as well. It would be great if there was a check box somewhere next to the standards list somewhere to let it know which ones to use the off peak interpolation for.

As you say you would have to be careful that you don't have to go miles out to measure interference free background positions where the continuum might be quite different.

[John Donovan]

One application that this new method might work well with is trace O ka. Normally it's difficult to find standards that do not contain oxygen (most metals have some trace amount of oxygen from a native oxide layer), and this method avoids that problem because it uses the interpolated off-peak intensity, instead of the on-peak intensity, for the MAN calibration.

[John Donovan]

Hi Ben,
Some of this may be confusing to others so let's define our terms a bit.  I propose the following:

Ben Hanson originally used the term "offset" MAN backgrounds to mean the interpolated off-peak intensities calculated from the hi and lo off-peak positions as measured in the MAN standards by acquiring them as off-peak standards. These *interpolated* off-peak intensities would then be applied to the MAN background calibration intensities by simply setting the interpolated on-peak intensity to the calculated (interpolated) off-peak intensity.

This is opposed to the existing practice of just using the measured *on-peak* intensities from off-peak measurements on the standards used in the MAN calibration (and ignoring the off-peak measurements altogether). That is using the Analytical | Use Off-Peak Elements for MAN Fit menu option as discussed here:

http://probesoftware.com/smf/index.php?topic=4.msg2255#msg2255

Therefore I think it is less confusing to use the term *interpolated* off-peak intensities to describe this idea of Ben Hanson's which avoids the problem he mentioned where an MAN standard may have a trace amount of the element present (at the peak position).

Instead I would like to propose that the term *offset* MAN backgrounds refer to the "stranger" suggestion I made here:

http://probesoftware.com/smf/index.php?topic=987.msg6444#msg6444

where one might "detune" the on-peak position for standards utilized in the MAN calibration (to a slightly offset peak position) so that one might avoid trace contamination that might be present in one's MAN standards. Of course this idea of mine has it's issues (you can't use the same standard as an MAN standard and also as a primary standard, and one has to modify the peak positions to be different for these MAN standards), but this "stranger" idea has the benefit of not requiring any code changes!   

So let's call Ben Hanson most excellent idea the *interpolated* off-peak MAN method. In code, this basically consists of setting the on-peak intensity of the MAN standards (acquired using the normal off-peak acquisition method), to the *interpolated* (that is calculated) off-peak intensity from the measured off-peak intensities.

So to respond to your points here:

I concur...this is a great idea as I run into this sometimes as well. It would be great if there was a check box somewhere next to the standards list somewhere to let it know which ones to use the off peak interpolation for.

If I understand you correctly, there is really is no need for this as the software will automatically figure out which MAN standards to perform this interpolation for.

As you say you would have to be careful that you don't have to go miles out to measure interference free background positions where the continuum might be quite different.

I assume you are referring to  my "stranger" idea of offsetting the on-peak position to avoid trace (on-peak) contamination.  Yes, if we were to utilize this method you'd want to not go too far from the on-peak position so that the continuum intensity is close to the intensity underneath the peak at the on-peak position.

But in the end, I would again suggest that it would probably be preferable to just get pure standards for use with the MAN calibrations!

[John Donovan]

Ok, I've tested Ben Hanson's new "interpolated" off-peak MAN method and it seems to be working. Please let me know what you all find.

Here are the steps:

1. Acquire intensities on your MAN standards but using normal off-peak measurements instead of the normal MAN acquisition method.

2. Select the Analytical | Use Off-Peak Elements For MAN Fit menu as seen here:



This makes sure that these off-peak standards will be found by PFE when you go to make your MAN assignments for each element.

3. Assign MAN standards as usual using the Analytical | Assign MAN Fits menu as usual. In the Assign MAN Fits dialog you will see displayed the *on-peak* intensities (ignoring the off-peak intensities for now!) for those standards that *do not* contain the element of interest).  If your standards are not perfectly pure, you may see that some standards lie above the general trend line from this contamination (of course this could be due to on-peak spectral interferences also).

4. Now we want to analyze our unknown (which was measured using normal MAN acquisition methods), to see the effect where we have subtracted too much background from our on-peak measurements, so we perform a quant on the unknown sample as seen here:

Un   11 pyrite #730
TakeOff = 40.0  KiloVolt = 20.0  Beam Current = 100.  Beam Size =    0

Un   11 pyrite #730, Results in Elemental Weight Percents

ELEM:       Au      Au      Cu      Ag
BGDS:      MAN     MAN     MAN     MAN
TIME:   400.00  400.00  400.00  400.00
BEAM:   100.49  100.49  100.49  100.49

ELEM:       Au      Au      Cu      Ag   SUM 
XRAY:     (la)    (ma)    (ka)    (la)
   627   -.005   -.002   -.010   -.004 100.066
   628   -.002    .003   -.007    .004 100.083
   629    .004   -.001   -.008   -.004 100.078
   630    .000   -.004   -.010    .004 100.077
   631    .002    .004   -.009   -.001 100.082
   632    .001    .000   -.011    .001 100.077

AVER:     .000    .000   -.009    .000 100.077
SDEV:     .003    .003    .001    .004    .006
SERR:     .001    .001    .001    .001
%RSD:     ----    ----  -15.53    ----
STDS:      579     579     529     547

STKF:   1.0001  1.0000   .9975   .9920
STCT:  15401.2  5782.1 13188.3  3764.1

UNKF:    .0000   .0000  -.0001   .0000
UNCT:       .0      .0    -1.0      .0
UNBG:    168.3    12.1    36.4     8.1

ZCOR:   1.4520  1.0366  1.1574  1.3373
KRAW:    .0000   .0000  -.0001   .0000
PKBG:     1.00    1.00     .97    1.00
BLNK#:      11      11    ----      11
BLNKL: .000000 .000000    ---- .000000
BLNKV: -.01444 .010240    ---- .009146

Note that the Cu concentration in our pyrite sample is negative because the MAN trend line was too high. So now we activate Ben Hanson's new interpolated off-peak MAN method by going to the Analytical | Analysis Options menu dialog and check the new option seen here:



Now we re-run our unknown sample quant and we obtain these results:

Un   11 pyrite #730
TakeOff = 40.0  KiloVolt = 20.0  Beam Current = 100.  Beam Size =    0

Un   11 pyrite #730, Results in Elemental Weight Percents

ELEM:       Au      Au      Cu      Ag
BGDS:      MAN     MAN     MAN     MAN
TIME:   400.00  400.00  400.00  400.00
BEAM:   100.49  100.49  100.49  100.49

ELEM:       Au      Au      Cu      Ag   SUM 
XRAY:     (la)    (ma)    (ka)    (la)
   627   -.005   -.001    .005   -.004 100.081
   628   -.002    .003    .008    .004 100.099
   629    .004   -.001    .007   -.004 100.093
   630    .000   -.004    .005    .004 100.092
   631    .002    .004    .006   -.001 100.097
   632    .001   -.001    .005    .001 100.092

AVER:     .000    .000    .006    .000 100.093
SDEV:     .003    .003    .001    .004    .006
SERR:     .001    .001    .001    .001
%RSD:  1437.74    ----   22.70 8127.41
STDS:      579     579     529     547

STKF:   1.0001  1.0000   .9975   .9920
STCT:  15401.2  5782.1 13188.3  3764.1

UNKF:    .0000   .0000   .0001   .0000
UNCT:       .0      .0      .7      .0
UNBG:    168.9    11.5    34.7     8.0

ZCOR:   1.4520  1.0367  1.1585  1.3373
KRAW:    .0000   .0000   .0001   .0000
PKBG:     1.00    1.00    1.02    1.00
BLNK#:      11      11    ----      11
BLNKL: .000000 .000000    ---- .000000
BLNKV: -.02000 .019429    ---- .011628

And all we did in the code was to set the on-peak intensity equal to the interpolated off-peak intensity for the standards used in the MAN calibration curve.

[BenH]

YES!  Oxide layers on my metal standards make MAN backgrounds impossible for oxygen.  This will be a big help to me.
Thanks again John for doing this!

By the way...I will just count longer on my backgrounds.  Shouldn't sacrifice anything in terms of precision.

 

[John Donovan]

YES!  Oxide layers on my metal standards make MAN backgrounds impossible for oxygen.  This will be a big help to me.
Thanks again John for doing this!

Hi Ben,
Are you quant mapping oxygen?  Using MAN backgrounds for quant mapping is a big time saver of course because one doesn't need to acquire off-peak maps with the MAN method.

By the way, I will modify the code so it automatically sets the "Use Off-Peak Elements For MAN Fits" menu if this new "Use Interpolated Off-Peaks for MAN Intensities" checkbox is checked when closing the Analysis Options dialog.
john

[BenjaminWade]

I concur...this is a great idea as I run into this sometimes as well. It would be great if there was a check box somewhere next to the standards list somewhere to let it know which ones to use the off peak interpolation for.

If I understand you correctly, there is really is no need for this as the software will automatically figure out which MAN standards to perform this interpolation for.

It's probably me not understanding correctly. For example taking oxygen, I could have standard X that is displaying a poor MAN fit as its oxidised while all others are OK. So I would visually determine that by reviewing the current MAN fit, and would enable the "interpolated" MAN measurement only for that standard X.

Or is it that the "interpolated" MAN is either on or off for all standards at once?

Cheers

[John Donovan]

Hi Ben,
Yes, please let's keep the interpolated off-peak MAN as all off or all on for the standards!   

I think it will work fine like this and one can still add/remove individual standards from the MAN fit manually as before.  Here's an example from a recent trace Au in pyrite run with the new interpolated off-peak MAN flag *off*:



Note that because of the interference of Fe Ka III on Au Ma, the standards containing Fe plot above the MAN trend line.  This is because the original default off-peak MAN method utilized the on-peak intensities from the off-peak acquired standards (ignoring the off-peak measurements). 

And here is the same MAN standard data, but with the interpolated off-peak MAN flag turned *on* in the Analytical | Analysis Options dialog:



Because we can now optionally utilize (thanks to Ben Hanson), the off-peak measurements by interpolation for the MAN calibration, any on-peak interference (and possible contamination effects) are removed from the MAN calibration. So now we can deal with standards that either have on-peak interferences, and/or are contaminated by the element of interest, and still obtain an accurate MAN curve.

Of course, with the interpolated off-peak MAN method we can still get off-peak interferences, just as we do for normal off-peak measurements, so be forewarned!

Never the less, pretty cool.
john

[Probeman]

So our EPMA class looked at another run we did on some Al, V, Fe, Cr alloys (also analyzing for oxygen), where we'd acquired off-peaks on the standards so we could compare the normal (on-peak) MAN method with the "interpolated" off-peak MAN method as proposed by Ben Hanson and suggested by Ben Wade as possibly interesting for oxygen analysis because the interpolated MAN would avoid the native oxide coating that is normally found on pure metals.

In this case we could only use pure Si and Fe as MAN standards for O Ka as the other metal standards had oxygen declared as a contaminant. Plotting up the MAN curve, it's just a straight line fit as seen here:



But as we were only interested in seeing the magnitude of the difference between normal on-peak MAN and interpolated off-peak MAN this will work fine.

So here is an analysis of our Cr metal standard using the normal on-peak MAN:

St  524 Set   1 Chromium metal, Results in Elemental Weight Percents
 
ELEM:       Al      Cr      Fe       V       O
BGDS:      MAN     MAN     MAN     MAN     MAN
TIME:    30.00   30.00   30.00   30.00   30.00
BEAM:    30.14   30.14   30.14   30.14   30.14

ELEM:       Al      Cr      Fe       V       O   SUM
    13   -.002 100.056    .018    .002    .803 100.876
    14   -.009 100.507    .007    .010    .802 101.317
    15   -.011 100.034   -.001    .006    .829 100.857

AVER:    -.007 100.199    .008    .006    .811 101.017
SDEV:     .004    .267    .010    .004    .015    .260
SERR:     .003    .154    .006    .002    .009
%RSD:   -60.66     .27  121.54   68.48    1.90


So about 0.8 wt% oxygen.  And here is the analysis using the interpolated off-peak MAN fit:

St  524 Set   1 Chromium metal, Results in Elemental Weight Percents
 
ELEM:       Al      Cr      Fe       V       O
BGDS:      MAN     MAN     MAN     MAN     MAN
TIME:    30.00   30.00   30.00   30.00   30.00
BEAM:    30.14   30.14   30.14   30.14   30.14

ELEM:       Al      Cr      Fe       V       O   SUM
    13    .015 100.121    .039    .001   1.073 101.248
    14    .009 100.572    .029    .009   1.072 101.690
    15    .007 100.099    .020    .004   1.099 101.229

AVER:     .010 100.264    .029    .004   1.081 101.389
SDEV:     .004    .267    .010    .004    .015    .260
SERR:     .003    .154    .006    .002    .009
%RSD:    43.34     .27   32.72   89.82    1.43


So it's about a 0.2 wt% difference in the measured oxygen, with the larger value using the interpolated off-peak MAN, which makes sense because the on-peak MAN would be affected by the native oxide layer on the Si and Fe standards used for the MAN calibration.  Thus causing a higher background intensity to be subtracted from the sample, but not so for the interpolated off-peak MAN method.

Interesting.
john

[Probeman]

St  524 Set   1 Chromium metal, Results in Elemental Weight Percents
 
ELEM:       Al      Cr      Fe       V       O   SUM
    13   -.002 100.056    .018    .002    .803 100.876
    14   -.009 100.507    .007    .010    .802 101.317
    15   -.011 100.034   -.001    .006    .829 100.857

AVER:    -.007 100.199    .008    .006    .811 101.017
SDEV:     .004    .267    .010    .004    .015    .260
SERR:     .003    .154    .006    .002    .009
%RSD:   -60.66     .27  121.54   68.48    1.90


St  524 Set   1 Chromium metal, Results in Elemental Weight Percents (interpolated off-peak MAN)
 
ELEM:       Al      Cr      Fe       V       O   SUM
    13    .015 100.121    .039    .001   1.073 101.248
    14    .009 100.572    .029    .009   1.072 101.690
    15    .007 100.099    .020    .004   1.099 101.229

AVER:     .010 100.264    .029    .004   1.081 101.389
SDEV:     .004    .267    .010    .004    .015    .260
SERR:     .003    .154    .006    .002    .009
%RSD:    43.34     .27   32.72   89.82    1.43

Upon a second glance I note that the other elements also showed interesting changes going from normal on-peak MAN fitting to the new interpolated off-peak MAN option.   The Al concentration went up from -0.007 to 0.01, the Fe went up from 0.008 to 0.029 and the V went down but by less than a single standard deviation.   

This suggests to me that me that the interpolated off-peak MAN is doing exactly what it is supposed to do, by avoiding small (undeclared) concentrations present in one's standards. Whether one knows they are there or not! 

This means to me that the background intensity decreased for these trace elements because the interpolated off-peak MAN avoided small contaminates or spectral interferences...

This might mean that this interpolated MAN method might be good enough for trace elements to remove the need for a blank correction.  I need to look at some more data!
john

[Probeman]

In thinking about this a bit more, it occurs to me that not only is this interpolated off-peak MAN measurement method likely to be as accurate as normal off-peak measurements, and maybe even a little better, since it's based on a regression of multiple standards (and because off-peak interferences are less likely in the simple metal and oxide standards that are usually utilized for the MAN calibration fit), but the typical time savings and better sensitivity of the MAN method will still apply. That is because we are only utilizing off-peak measurements for our MAN standards, not for the unknown samples using MAN (on-peak only) measurements.

Is anyone interested in writing up a short abstract on this with me?   I'd ask Ben Hanson who came up with the idea in the first place, but he always tells me that Corning Glass won't let him publish anything at all.  We'll certainly acknowledge him of course.
john