R10 rutile?

[fenheee]

Hi all,

We are trying to do some provenance analysis and EPMA/LA-ICP-MS characterization of detrital rutile, mainly targeting Nb and Cr.

I'm trying to track down whether the R10 rutile (known for elevated Nb) is still available anywhere for EPMA calibration. If R10 is no longer obtainable, I'd be very interested to hear if anyone knows of Nb-bearing alternatives that could serve as reliable standards/secondary standards for unknowns.

At present, I have access to R632, Sugluk-4, and PCA-S207, but we've found that their Nb contents are either too low or not sufficiently homogeneous for this purpose.

Any suggestions, leads, or experience would be greatly appreciated!

[Probeman]

Hi all,

We are trying to do some provenance analysis and EPMA/LA-ICP-MS characterization of detrital rutile, mainly targeting Nb and Cr.

I'm trying to track down whether the R10 rutile (known for elevated Nb) is still available anywhere for EPMA calibration. If R10 is no longer obtainable, I'd be very interested to hear if anyone knows of Nb-bearing alternatives that could serve as reliable standards/secondary standards for unknowns.

At present, I have access to R632, Sugluk-4, and PCA-S207, but we've found that their Nb contents are either too low or not sufficiently homogeneous for this purpose.

Any suggestions, leads, or experience would be greatly appreciated!

Sorry to say, but this approach, while traditional, is less than ideal for high accuracy, high sensitivity trace element analysis on modern EPMA instruments.

In the past because of problems with background correction methods and matrix corrections, people assumed that they must use matrix matched standards with a similar concentration to their unknowns.

But this traditional method for trace and minor elements is problematic for several reasons, some of which you already alluded to (e.g., standard heterogeneity). Today our modern matrix corrections are very accurate, so we can easily extrapolate from a pure metal or pure oxide primary standard to low concentrations in a unknown with a different matrix. 

If we want to check for zero accuracy, we can run a blank, which is a material with a (roughly) similar matrix (depending on whether the background artifacts are due to the instrument or the matrix), but containing less than 1 PPM of the element in question. In other words, we can use a blank material to determine our trace accuracy because we can know exactly what zero is if the element is question is below our detection limits.

So, can you measure zero accurately for this element in this matrix, is the ultimate question for determining trace element accuracy. Using a standard with an approximately known concentration (is the concentration 1000 PPM or 1100 PPM?) just introduces more inaccuracy.

In addition, modern background methods, e.g., multi-point backgrounds, are considerably more accurate than traditional two background linear fit methods.

See here for a quick explanation:

https://smf.probesoftware.com/index.php?topic=610.msg11833#msg11833

And here for a more detailed explanation:

https://www.youtube.com/watch?v=9KM5lU403VY&t=567s&ab_channel=ProbeSoftwareInc

[Probeman]

To further explain why a pure metal or pure oxide standard is best for trace element work, see a typical detection limit equation such as this:



Note that the standard intensity is in the denominator, while the variance of the background is in the numerator. So we want the ratio of the background variance to the standard intensity to be a small as possible to obtain the highest sensitivity. This will generally be the case for a primary standard that is a pure element or pure oxide (or sulfide).

In fact, getting the variance of the background as small as possible (e.g., increased counting times) also improves trace element sensitivity. That is why the MAN background correction has improved trace element sensitivity. The variance on the background is smaller than for off-peak measurements because the variance of the MAN correction depends on major element statistics, not continuum statistics.

In addition, the accuracy of a pure metal or pure oxide (or pure sulfide) is well known. So by utilizing a pure metal or pure oxide (or pure sulfide) as a primary standard, we also obtain improved accuracy for our trace elements.

[John Donovan]

We are trying to do some provenance analysis and EPMA/LA-ICP-MS characterization of detrital rutile, mainly targeting Nb and Cr.

Based on Probeman's suggestions above it seems quite straight forward... for your primary standards for Nb and Cr, looking at what is available here:

https://smf.probesoftware.com/index.php?topic=1771.msg13588#msg13588

your best choice for a primary standard for Nb would be Nb metal, or if prefer an oxide matrix you can buy some LiNbO4 synthetic, which is commercially available. For Cr, your choices would likewise be Cr metal or Cr2O3 synthetic. All should be easy to obtain and maybe you already have them. Just be sure they are at least 99.9% pure. If they are, you now know their composition within 0.1% accuracy!

From CalcZAF, assuming pure element standards extrapolating to TiO2, we get this:

LINEMU   Henke (LBL, 1985) < 10KeV / CITZMU > 10KeV

  Z-LINE   X-RAY Z-ABSOR     MAC
      Ti      ka      Ti  1.0869e+02
      Ti      ka      O   6.5919e+01
      Ti      ka      Nb  6.7886e+02
      Ti      ka      Cr  1.3625e+02
      O       ka      Ti  2.1976e+04
      O       ka      O   1.1999e+03
      O       ka      Nb  1.5203e+04
      O       ka      Cr  3.1438e+03
      Nb      la      Ti  7.8700e+02
      Nb      la      O   5.5824e+02
      Nb      la      Nb  7.2394e+02
      Nb      la      Cr  1.0124e+03
      Cr      ka      Ti  5.7731e+02
      Cr      ka      O   3.7620e+01
      Cr      ka      Nb  4.1530e+02
      Cr      ka      Cr  8.2659e+01

 ELEMENT  ABSFAC  ZEDFAC  FINFAC STP-POW BKS-COR   F(x)e
   Ti ka  1.0288 4.36228 4.48781  .20699   .9030   .9720
   O  ka  1.4270 3.91542 5.58728  .24381   .9546   .7008
   Nb la  1.2397 4.73358 5.86808  .15882   .7518   .8067
   Cr ka  1.0205 4.38010 4.46993  .20735   .9082   .9799

SAMPLE: 32767, ITERATIONS: 0, Z-BAR: 16.42514

 ELEMENT  ABSCOR  FLUCOR  ZEDCOR  ZAFCOR STP-POW BKS-COR   F(x)u      Ec   Eo/Ec    MACs
   Ti ka   .9951   .9998  1.0858  1.0804  1.1247   .9655   .9768  4.9670  3.0199 92.3768
   O  ka  6.6071  1.0000   .8908  5.8852   .8465  1.0522   .1061   .5317 28.2114 13673.8
   Nb la   .9809   .9887  1.1546  1.1197  1.3894   .8311   .8224  2.3710  6.3264 697.117
   Cr ka  1.0671  1.0000  1.0952  1.1687  1.1406   .9602   .9183  5.9900  2.5042 361.664

 ELEMENT   K-RAW K-RATIO ELEMWT% OXIDWT% ATOMIC% FORMULA TAKEOFF KILOVOL                                        
   Ti ka  .00000  .55491  59.950   -----  33.307    .998   40.00   15.00                        
   O  ka  .00000  .06805  40.050   -----  66.613   1.995   40.00   15.00                        
   Nb la  .00000  .00089    .100   -----    .029    .001   40.00   15.00                        
   Cr ka  .00000  .00086    .100   -----    .051    .002   40.00   15.00                        
   TOTAL:                100.200   ----- 100.000   2.995

Note that the matrix correction (ZAFCOR) extrapolating from pure metals for Nb and Cr is ~12% and 17% respectively.  For primary standards LiNbO4 and Cr2O3, the extrapolation would be much smaller due to both unknown and standard having an oxygen matrix (note that the ZEDCOR correction is most of the correction when the pure metals are utilized).

But since we know that modern pr(z) matrix corrections are quite accurate (better than 2%) the benefit of accuracy using a pure element or oxide primary standard will be more important.

As for testing our trace element accuracy in rutile, a synthetic TiO2 material is very easy to obtain for this purpose. See here for examples using the blank correction for ultimate trace element accuracy:

https://smf.probesoftware.com/index.php?topic=454.msg6694#msg6694

The cool thing about having a suitable blank material such as synthetic TiO2, is that our measurement precision becomes equal to our trace element accuracy:

https://smf.probesoftware.com/index.php?topic=29.msg387#msg387

We just need to be sure that all trace elements (being measured) in our blank material are 1 PPM or lower. That is, anything below 1 PPM is essentially a zero for EPMA or SEM. However, you can document a non-zero blank level for the blank material in the blank correction. For example, in my synthetic SiO2, we have 1.42 PPM Ti (from ICP-MS), so I just enter 0.000142 (wt.%) for the blank level when measuring Ti in quartz and running my synthetic quartz as an unknown:



This level is then automatically incorporated into the blank correction during the matrix iteration.

[fenheee]

Thank you so much for your detailed replies and for sharing your experience — this is really helpful for my rutile trace element work!

I am currently working mainly with the JEOL native software (without Probe for EPMA). May I ask: are some of these procedures you described — for example, using a pure oxide/metal primary standard, defining a "zero" blank material, and extrapolating to trace element concentrations in rutile — also feasible within the JEOL software environment? If so, which parts can be implemented directly, and which parts would require additional tools like Probe for EPMA?

I really appreciate your guidance and the insights you have shared here.

[John Donovan]

Thank you so much for your detailed replies and for sharing your experience — this is really helpful for my rutile trace element work!

I am currently working mainly with the JEOL native software (without Probe for EPMA). May I ask: are some of these procedures you described — for example, using a pure oxide/metal primary standard, defining a "zero" blank material, and extrapolating to trace element concentrations in rutile — also feasible within the JEOL software environment? If so, which parts can be implemented directly, and which parts would require additional tools like Probe for EPMA?

Yes, you can absolutely use pure element or pure oxide standards as your primary standards in the JEOL software as the difference in the matrix corrections are small, and you will certainly obtain higher accuracy and also higher sensitivity trace element measurements. Run some experiments, and see for yourself!

As for the blank correction which is built into Probe for EPMA, you can perform such blank corrections manually in a spreadsheet after you export your results in Excel or other software.  And as long as you have suitable blank materials, which in your case would be a high purity synthetic TiO2.

The reason the blank correction in built into Probe for EPMA is mostly for convenience, but also for ultimate accuracy as when the blank correction is applied to the unknown intensities, there will be a small change in the matrix correction. By including the blank correction during the matrix correction this is automatically compensated for.

But in most cases the blank correction is so small (less than hundreds of PPM), that the change in the matrix correction is negligible. Therefore it could be performed afterwards with only a very small error by simply noting the concentration difference from zero measured in your blank material, and applying that offset manually to your unknowns.  Read this paper and make sure you understand the method and try it:

J. J. Donovan, et. al., "Improved Electron Probe Microanalysis of Trace Elements in Quartz", American Mineralogist, 96, 274-282, 2011

Also be aware of the importance of spectral interferences. When I started in the field everyone was reporting a small amount of vanadium in their rutiles, so we published a paper on how this should be corrected for:

J. J. Donovan, D. A. Snyder and M. L. Rivers, "An Improved Interference Correction for Trace Element Analysis"
Microbeam Analysis, 2: 23-28, 1993

However, since the JEOL software has no spectral interference correction you will have to make sure that you are avoiding these situations. Typically this is done by measuring a high purity material (TiO2), and checking that you are seeing no trace element concentrations statically above zero (or whatever your ICP-MS has determined).  In Probe for EPMA this interference measurement can be automatically applied to your unknowns...

https://smf.probesoftware.com/index.php?topic=69.0

Here is a nice slide explaining the important considerations for trace elements (note also secondary boundary fluorescence effects, if you are measuring within a few hundred microns of another phase boundary which contains the element of interest:



Another good topic:

https://smf.probesoftware.com/index.php?topic=928.0

Happy to answer any further questions.