An Electron Microprobe Study of Synthetic Aluminosilicate Garnets

The aluminosilicate garnets represent an important mineral group. Common end-members are given by E3Al2Si3O12, where E=Fe2+ (almandine), Mn2+ (spessartine), Mg (pyrope), and Ca (grossular). End-members have been synthesized, but their exact compositions and stoichiometries are generally unknown. Synthetic aluminosilicate garnet can possibly contain minor Fe3+, Mn3+, F- and OH- and possibly vacancies. Slight atomic disorder over the 3 different cation sites may also occur. Natural crystals are considerably more complex. Electron probe microanalysis (EPMA) provides a method to determine garnet chemistry and stoichiometry. However, accurate determinations are not always a simple matter and uncertainties exist. We have started a study on well-characterized synthetic aluminosilicate garnets in order to i) determine more exactly their compositions and stoichiometries and ii) better understand possible complications in EPMA. Synthetic almandine, spessartine, pyrope, and grossular samples were synthesized under varying conditions both hydrothermally and dry and with different starting materials. A closed thermodynamic system was present and the bulk starting material composition represented the exact stoichiometric end-member garnet that was desired. IR, Raman and Mössbauer spectroscopy in some cases and X-ray diffraction were used to characterize the samples. Synthetic pyrope has been investigated with a SX51 with simple oxide/silicate standards (Fo90 olivine for Mg, wollastonite for Si, and both Al2O3 and kyanite for Al). Previously observed problems were reproduced: low stoichiometry for Al and high for Si and Mg. Fournelle (2007, AGU Fall Mtg) noted chemical peak shifts for Al and Mg Ka in garnets; this effect was eliminated here by proper peaking. Earlier suggestions for issues with mass absorption coefficients were not seen, and Probe for EPMA software demonstrated there was not much difference between the most recent FFAST values vs. the older Heinrich values. Similarly, a matrix correction based on CITZAF was compared with PAP, with little difference in Al (both low) and Mg values (both high), though PAP had higher Si values and CITZAF had lower ones. An assumption in EPMA is that the intensity of a single peak channel is representative of the integral of all x-ray counts under the total peak. It is known that this is not true for “light elements” (Be-F). We performed detailed wavescans of the complete peaks of Si, Al and Mg Ka of both standards and pyrope. Mg and Si peak scans showed little or no difference between pyrope and standard, but the Al scans had reproducible differences of 3% between the Al2O3 standard and unknown. This yielded an “area peak factor” correction of 1.03, which brings the Mg closer, but not all the way, to a stoichiometry of 2, Si to almost 3, and Mg closer to 3 but still too high. One possibility to be evaluated is whether there may be solid solution between pyrope and a minor majorite (Mg3(Mg,Si)Si3O12) component involving coupled substitution 2Al = Mg,Si at the octahedral site. In summary: Is this an EPMA analytical issue (peak fine structure differences), or an issue regarding a small amount of an unforseen component (majorite) in the garnet?