Here's a variation on a crazy idea from Donovan...
As we know, trace element accuracy is possible in EPMA but there are a number of artifacts that can make such analyses problematic. Depending on the specific matrix involved there may be issues with off-peak background positions, on-peak interferences, sample (and detector) absorption edges and beam sample sensitivity to name just a few.
Therefore many of us are seeking secondary standards for checking our trace element accuracy in EPMA. The problem of course is that getting homogeneous (and accurate) trace element standards is difficult. For both natural and synthetic materials we have problems with inclusions or contamination and heterogeneity. Then there are the issues with knowing what the trace element level actually is. Is it 110 PPM or is it 115 or 105 PPM? Who knows? Every technique has its systematic errors. This is discussed in some length in this topic here:
http://probesoftware.com/smf/index.php?topic=928.0But there are two values that we can know with very high accuracy, First we can know something is 100% with extremely high accuracy. For example, if our Si or SiO2 EPMA standard has been analyzed for possible contaminants and all are less than a few PPM, then we can be pretty confident that our Si is 99.99% Si or our SiO2 is 99.99% SiO2 (or however many 9s we checked for).
Likewise, we can also know that something is *zero* with extremely high levels of accuracy. In fact in these cases we can have trace element accuracy *equal* to our measurement precision level, by simply using a matrix matched pure synthetic blank standard.
For example, if we use highly sensitive techniques such as ICP-MS on a pure synthetic standard material, and we know that our possible contaminants (that is, trace elements of interest) are well below the detection limit of the microprobe (let's say 1 PPM), then by measuring a pure synthetic standard which is similar to our matrix and carefully analyzed so we know the traces are below EPMA sensitivity, we can now check the accuracy of our EPMA trace element measurement by simply seeing if we can measure zero. If we do not obtain a zero measurement, then something is wrong, perhaps the background is curved (it always is!). Or there's a spectral interference, etc., etc. This of course is where the blank correction in PFE can be applied to correct our unknowns.
For very simple unknowns such as measuring traces in say quartz, we are already fine, because there are many highly pure SiO2 materials available at low cost. But that is not the case for other materials of interest.
Let's take a simple case of measuring traces in say olivine. We'd like to have matrix similar to our unknown. But growing a homogeneous olivine of intermediate composition is difficult. There's always some zoning in an MgFeSiO4 composition, but wait!
We don't care what the major element composition is, we only care about the trace elements in some olivine composition. That is we really only care that the trace elements Mn, Ni, Cr, Al, Ca, etc. are below the detection limit for EPMA. That way we can check our trace element accuracy in a matrix matched standard because we know regardless of the actual Fe-Mg ratio, we should obtain zero +/- the measurement precision.In fact, we would actually *want* our olivine trace element standard to have a range of composition! Why? Because then we can find an appropriate Mg rich or Fe rich (matrix matched) area of our synthetic olivine to test our zero measurement accuracy!
Now let's take a nasty case of trace U and Pb in monazite. Traditionally we would try and obtain a perfect homogeneous synthetic monazite that is accurately doped with the trace elements of interest. But that is extremely difficult to synthesize. It is usually highly zoned in major element composition (and trace elements) and therefore it can not really be considered a standard.
But, as long as the trace elements in question (say U and Pb), are *not* present at EPMA detection limits- it does not matter! Again, by having a highly zoned synthetic monazite that we know is free from U and Pb, we have the perfect standard to check our trace element accuracy and we can probably find an intermediate composition of Ce, Nd, Gd, Sm, Th, etc. phosphate, that is a close matrix match to our actual unknown and we have high accuracy for zero concentrations of U and Pb.
We can continue the same line of thinking to traces in alloys, traces in sulfides, etc. Again: the actual composition doesn't have to be an exact match (or even known accurately for that matter). What really matters is that the trace elements of interest are below EPMA detection limits in a similar matrix.
So what I think we really need are highly purified starting materials, and then synthesize some olivines, monazite, etc. for use as secondary trace element standards, to test our trace accuracy determinations in a close matrix match material.
As has been said before: "if you can't measure something, try measuring nothing, because if you can't measure nothing, you can't measure anything". And I would take it one step further: " for best accuracy, try measuring nothing, in a matrix matched pure synthetic material".
Hey I told you it was crazy!
What do you think?
Poll