“Houston, we have a problem” – Jack Swigert, Apollo 13
We also have a problem, though hopefully not one of life or death. However, it is a serious problem and one that requires our collective attention. It is a question of accuracy in the field of microanalysis.
Let's start by asking what might be the largest source of inaccuracy today in microanalysis.
Some of us would say: by *not* using standards. That is, standardless EDS analysis. Unfortunately we seem to have reached an impasse on what can be done about this situation (aside from getting EDS vendors to remove the "Quant" button and getting every SEM lab to obtain proper standards!), so let's put this aside for now, more on this later.
Then what might be the second largest source of inaccuracy in microanalysis today? We would suggest: by using standards! Now what could we possibly mean by this statement?
Well, 40 years ago there was another problem in microanalysis. Namely that the analytical physical models for matrix corrections at that time were simply not very good. To address this issue, various empirical and semi-empirical methods were developed and tried (e.g., alpha factors, calibration curves, ZAF, etc.). But most of the time, for high accuracy work in these early days, it was usually necessary to utilize "matrix matched” standards. That is, find some "known" material that was similar to our unknowns, in order to minimize the matrix correction extrapolations from the standard to the unknown. As we all know, if the unknown is the same composition as the standard, the matrix correction is exactly 1.000.
So what did people do? Well, heroes such as Gene Jarosewich and others went into their mineral collections and picked out some large, well crystallized natural specimens and diligently performed wet chemistry and other characterization on the pulverized material. Many of us have made similar efforts in our own laboratories to find such “matrix matched” standards.
Unfortunately these natural materials often had trace and minor element impurities, were inhomogeneous and zoned, containing various inclusions of different minerals and like the restaurant review we've all heard about: the food was terrible, and the portions small. That is, when a request for standard material was made, we would often only receive tiny flyspecks.
Since that time we have been the beneficiaries of many advances in the physical models for matrix corrections (e.g., phi-rho-z models such as PAP, XPP, Brown, Armstrong, improved mass absorption coefficients, improved fluorescence corrections, etc.), so that today we can often obtain accuracy better than 2% relative in most matrices. Sometimes, especially for minor and trace elements, our accuracy is only limited by our measurement precision! So such “matrix matched” standards are often no longer necessary in many cases. But what is necessary are large quantities of high purity, high accuracy standards!
But here we are today still relying on these sometimes poorly characterized, natural, impure, heterogeneous, inclusion ridden standard materials (and such small portions!), which are now arguably a major source of inaccuracy in our field. See Fournelle, J., & Scott, J. (2017).
Note: due to differences in chemical bonding and coordination, there can be subtle peak shift and shape effects, and therefore we may still be required to utilize oxide and silicate standards for analyzing oxide and silicate materials, and sulfide and sulfosalt standards for analyzing sulfide and sulfosalt materials, etc. We can probably live with that!For possible evidence of these issues see Gale et al., (2013) “The mean composition of ocean ridge basalts” and also Yang et al. (1996) "Experiments and models of anhydrous, basaltic olivine-plagioclase-augite saturated melts from 0.001 to 10 kbar" where they state: "An interlaboratory comparison has been made (Reynolds 1995) including MIT, the Smithsonian Institution in Washington, Lamont Doherty and University of Hawaii. It is the practice in our laboratory to correct microprobe data obtained elsewhere to an MIT reference before making thermobarometric or modeling analyses (see Table 1). Although Grove et al. (1992) neglected to discuss this issue, the Smithsonian data discussed in that paper was corrected before plotting and estimation of crystallization pressure. Failure to do so can result in significant errors, and is most commonly evident as a discrepancy in the pressures estimated from the different equations". Also please see figure 1 below from Penny Wieser for a graphical representation of these various interlaboratory biases.
So what can be done about this situation? Well, notable efforts have been made at NIST to synthesize mineral glass standards, e.g., K-411, K-412 and more recently at the USGS basaltic glasses BCR-2G, BHVO-2G and BIR-1G and these glasses be very useful standards if produced in sufficient (kilogram) quantities so that every microanalysis lab on the planet can obtain them. Compositional characterization of such glasses is however quite non-trivial, but can be done with enough effort. If only such glasses were available in kilogram quantities and freely available. Indeed if they are available in such large quantities, they should be part of every standard collection, but apparently they are not. So what might we do?
We propose that a modest to moderate investment by our international microanalysis community can provide high purity, high accuracy standards for current and future generations of microanalysts.
We propose by utilizing high purity synthetic single crystal materials produced in kilogram quantities every microanalytical laboratory in the world could have access to the same standards. Such end-member single crystals of high purity can be, unlike glass standards which require further compositional characterization, already known in composition!
Note: some questions have been raised as to the degree of, or closeness to, stoichiometry of industrially-produced synthetic materials. Specifically, to what accuracy can the chemical stoichiometry of such single crystals be determined? For example, if a high purity single crystal is homogeneous on the micro-scale, is it also likely to be chemically stoichiometric? This will require further investigation.
E.g., high purity, single crystal Mg2SiO4 should be exactly Mg: 34.550 Si: 19.962 O: 57.143 weight percent (assuming accepted terrestrial isotopic distributions!). And it is grown industrially today as a laser material.
We propose to invest in high purity, stoichiometric (thermodynamically constrained end-member), synthetic standard materials produced in kilogram quantities. Specifically, pure enough single crystals so that homogeneity is not in question (though both purity and homogeneity can be checked), and enough quantity so that *every* microanalytical laboratory in the world has access to the *same* primary standards. In other words, the global standardization of microanalysis, much as was done hundreds of years ago for the metric system, when there were no global standards for commercial or scientific weights and measures. Call it the metrification of microanalysis standards if you will.
We do not propose that these materials be produced in academic/government laboratories; most do not appear set up for kilogram production quantities. However we would very much depend on the expertise of those individuals among our colleagues who are experienced in the growth of such materials to advise us as to what synthetic minerals may be commercially possible.
Instead, we propose that there are sufficient industrial/commercial resources capable of producing semi-conductor and optical/electronic materials, so that we could contract out the production of such high purity single crystal boules for these standard materials. What standards should we invest in producing? That is a good question. We think this is for us as a community to decide. Some polling on this question should be organized.
We might guess that the average cost of the synthesis per kilogram of such high purity synthetic standard materials might average around $10K each (pers. comm., Marc Schrier, Calchemist). Some materials are already available (e.g., SiO2, MgO, Al2O3, MgAl2O4, Mg2SiO4, YAG, YIG, Fe2O3, TiO2, SrTiO3, RbTiOPO4, KTiPO4, MnO, Fe3O4, NiO, ZnO, LaAlO3, MnPSe3, LiTaO3, etc.) and will be a fraction of this cost and can be bought "off the shelf".
Other synthetic minerals may require further research and development, e.g., ZrSiO4 (zircon), ZrO2 (zirconia), HfSiO4 (hafnon), HfO2 (hafnia), ThSiO4 (tetragonal thorite), ThSiO4 (monoclinic huttonite), Fe2SiO4 (fayalite), Mn2SiO4 (tephroite), CaMgSi2O6 (diopside), Al2SiO5 (sillimanite), NaAlSiO4 (nepheline), KAlSi3O8 (sanidine), KAlSi2O6 (leucite), KAlSi3O8 (orthoclase), NaAlSi3O8 (albite), CaAl2Si2O8 (anorthite), Fe3Al2Si3O12 (almandine), PbSiO3 (alamosite), CaAl2O4 (krotite), CaAl4O7 (grossite), CaAl12O19 (hibonite), CaSiO3 (wollastonite), MgSiO3 (enstatite), FeSiO3 (ferrosilite), sulphides (which are seriously lacking since the pioneering days at the USGS-Reston in the 1970s by researchers including Barton, Skinner, Czamanske, Bethke, Toulmin, among others), tellurates, arsenides, niobates, tantalates, etc., may cost 2 or 4 or 10 times this. Let’s do more research on what might be possible at a reasonable cost.
The point being that with further research we believe that other high purity single crystal materials can be identified, developed, characterized and included as useful microanalysis standards in large quantities for use worldwide.
As former directors and presidents of several microanalysis societies, we know the money is available. For example we believe that the Microbeam Analysis Society has accumulated an order of magnitude more money in their funds than would be necessary to fund such a project. By spending around 10 to 30% of just the MAS funds we could secure the global future of high accuracy microanalysis for generations. If several other national microanalysis societies join this effort, the cost to each society will be an even smaller percentage, and all will benefit.
It should also be noted that such high purity materials could also serve additional purposes such as:
1. Primary standards to check the compositions of the current standards in every microanalytical lab. If we all are not utilizing the *same* primary standards, what is the point of comparing them?
2. Primary and secondary standards as a test bed for the community consensus k-ratio database as proposed by Nicholas Ritchie. This means we should strive for at least two standards per element in this effort!
3. "Blank" materials for trace element analysis and also for mean atomic number (MAN) background standard materials. Six or more “nines” purity is required for use as a trace element blank.
4. Having these end member high purity synthetics (and maybe some glasses) will really stress our EPMA matrix corrections, dead time calibrations, beam current (Faraday Cup) linearity, not to mention effective takeoff angles and stage tilt on SEM instruments. Such failure mode analysis is essential if we are to make progress in improving these areas of instrumental calibration.
5. It should also be noted that unlike the “historical accidents” of many of our current standards available today (which are very unlikely to ever be re-created with the same exact compositions), the future production of high purity, single crystal, and thermodynamically constrained standard compositions can always be repeated in the future if necessary. E.g., high purity MgAl2O4 will always be high purity MgAl2O4.
Some possible other items to consider:
6. The MASFIG committee should establish the minimum qualifications for a candidate standard material to be included in the archive: e.g., characterization by XRD for a crystalline material; independent elemental analysis for a glass; trace measurements by WDS/ICP-MS to establish minimum detectable limit for a specified suite of elements.
7. It should be noted that in the area of synthetic minerals there are basically two types of candidates: (a) materials already produced at an industrial scale and readily available in kilogram quantities at a fairly reasonable prices (e.g., MgO, Al2O3, MgAl2O4, TiO2, SrTiO3, etc.), and (b) those that are only produced in experimental laboratories in limited (e.g, grams to tens of gram) amounts (e.g., Mg2SiO4, Fe2SiO4, ZrSiO4, Al2SiO5, CaMgSi2O6, etc.).
It must be said that we should probably first concentrate on those materials that are already available in sufficient quantities with reasonable prices to begin with, and then follow up with consultation and investigation of other possible synthetic minerals based on their feasibility of being produced in sufficient quality and quantities.
8. Establish an on-line database for the information on each standard material, perhaps supported by a non-fungible digital token (NFT) that documents the composition and any other issues, e.g., dose sensitivity, surface layers, etc. This database could include approved additions of information to the analytical record for each material supplied by users. FIGMAS already has a framework for this process.
9. Establish a site for the repository of the materials, located at a university, museum or national laboratory.
10. Establish a strong mechanism for making these standard materials available to customers worldwide, e.g., create working relationships with the vendors who currently provide prepared microanalysis standards. A participating vendor would be given a quantity of the standard material that could be included in that vendor’s prepared microanalysis standards for distribution. A portion of the material supplied to the vendor should also be available for interested customers to purchase (at a nominal cost to cover the vendor’s expenses) individual rough pieces suitable for mounting and polishing by the customer.
11. These materials may also be useful for other methods of characterization, i.e., Raman spectroscopy, Infrared specular reflectance spectroscopy, Infrared ATR spectroscopy (as powdered material), etc.
Regardless, this is a global analytical issue affecting the microanalysis community. Every microanalysis lab should be able to reference the same primary standard materials if we are to attempt to properly compare our data and results. If such standard materials are readily available in kilogram quantities, then not only every EPMA lab, but every SEM lab should be able to utilize the same reference materials. Now that would be something worth having for a truly global science of microanalysis.
We are currently in the gathering ideas phase. This effort is clearly one that will foster lots of interest from our community and beyond (as we should hope, with a project such of large scope as this). Please post your comments and ideas to this topic and let’s begin the discussion on how to finally move forward on this critical aspect of our field.
This is an investment not only for ourselves, but for the future of our science, so please join us in these efforts and change the world for future generations (of analysts) to come. They will thank us!
Signed,
Marisa Acosta, University of Lausanne
Dave Adams, Auckland University
Julien Allaz, ETH Zurich
Renat Almeev, Hannover of University
Paul Asimov, California Institute of Technology
Aaron Bell, University of Colorado
Joseph Boro, University of Hawaii
Scott Boroughs, Washington State University
Emma Bullock, Carnegie Institution of Science
Paul Carpenter, Washington University
Henrietta Cathey, Queensland University of Technology
Dave Crabtree, Ontario Geological Survey
Joel Desormeau, University of Nevada, Reno
John Donovan, University of Oregon
Mike Dungan, University of Oregon
Paul Edwards, Strathclyde University
Jon Fellowes, University of Manchester
John Fournelle, University of Wisconsin
Zack Gainsforth, University of California at Berkeley
Raynald Gauvin, McGill University
Karsten Goemann, University of Tasmania
Stacia Gordon, University of Nevada, Reno
Dick Grant, Sandia National Laboratory
Juliane Gross, Rutgers University
Jakub Haifler, Masaryk University
John Hanchar, Memorial University of Newfoundland
Jason Herrin, Nanyang Technological University
Heidi Hoefer, Frankfurt University
Julia Hammer, University of Hawaii
Eric Hellebrand, University of Utrecht
Dominik Hezel, University of Frankfurt
Raymond Jeanloz, University of California, Berkeley
Mike Jercinovic, University of Massachusetts
Brian Joy, Queen’s University
Stuart Kearns, University of Bristol
Adam Kent, Oregon State University
Michael Lance, Oak Ridge Laboratory
Donovan Leonard, Oak Ridge National Laboratory
Yanan Liu, University of Toronto
Xavier Llovet, University of Barcelona
Andrew Locock, University of Alberta
Heather Lowers, United States Geological Survey
Chi Ma, California Institute of Technology
Danny MacDonald, Dalhousie University
Ryan McAleer, USGS, Reston
Mike Matthews, Atomic Weapons Establishment
Francis McCubbin, NASA, Johnson Space Center
Andrew Mott, Texas A&M
Aurelien Moy, University of Wisconsin
Timothy Murphy, Macquarie University
Will Nachlas, University of Wisconsin
Owen Neill, University of Michigan
Angus Netting, University of Adelaide
Dale Newbury, National Institute of Standards and Technology
Phil Orlandini, University of Texas, Austin
Changkun Park, Korea Polar Research Institute
Anne Peslier, NASA, Johnson Space Center
Glenn Poirier, University of Ottawa
Xiaofei Pu, Idaho National Laboratory
Ron Rasch, University of Queensland
Minghua Ren, University of Nevada, Las Vegas
Paul Renne, Berkeley Geochronology Center
Nicholas Ritchie, National Institute of Standards and Technology
Malcolm Roberts, University of Western Australia
George Rossman, California Institute of Technology
Dawn Ruth, USGS Menlo
Gareth Seward, University of California, Santa Barbara
Lang Shi, McGill University
Tom Sisson, USGS Menlo
Giovanni Sosa-Ceballos, , National Autonomous University of Mexico
John Spratt, London Museum of Natural History
Frank Tepley, Oregon State University
Edward Vicenzi, Smithsonian Institution
Anette von der Handt, University of Minnesota
Benjamin Wade, University of Adelaide
Richard Walshaw, University of Leeds
Penny Wieser, Oregon State University
Axel Wittmann, Arizona State University
Karen Wright, Idaho National Laboratory
Panseok Yang, University of Manitoba
Shui-Yuan Yang, China University of Geosciences
Marty Yates, University of Maine
Keewook Yi, Korea Basic Science Institute
Ying Yu, University of Queensland
Zhou Zhang, Zhejiang University
Ryan Ziegler, NASA, Johnson Space Center

Fig 1 - Assessing the effect of interlaboratory biases on Cpx-only and Cpx-Liq thermobarometery using the average reported Cpx and Liq composition from the experiments of Krawcyznski et al. (2012) analyzed on the MIT microprobe.
(a-b) Interlaboratory correction factors for glass from Gale et al., (2013) relative to the Lamont microprobe (plotting at 1, 1).
(c-d) Calculated Cpx-only and Cpx-Liq pressures and temperatures for the average reported composition from Experiment 41c-106, corrected as if these materials were measured in the different laboratories shown in a-b. We assume the Cpx and Glass offsets between different laboratories are identical, as to our knowledge no Cpx round robin has ever taken place.
(e-f) as for c-d, using experiment 41c-108b.
Calculating pressures can vary by ~4 kbar and 50 K just depending on which microprobe analyses were performed on. These systematic offsets between laboratories likely increase the amount of noise in experimental datasets compiled from different laboratories when calibrating different thermobarometric expressions. Similar systematic offsets in pressure and tempreature space can be expected for different groups measuring Cpx and Glass compositions in natural samples to calculate pressures and temperatures.
For example, for a given natural Cpx composition, the MIT microprobe might yield ~12.5 kbar, while the Lamont microprobe would yield ~10 kbar (c). These potential offsets largely cannot be corrected retrospectively, as there is insufficient data on the magnitude of offsets between different EPMA laboratories for different geological materials.
See attached pdf (please login to see attachments).