Peak Shift in Mg Ka in EPMA: High Resolution X-ray Spectrometer Results for Silicate and Oxide Minerals

Recent experiments with a high resolution x-ray spectrometer have confirmed the suggestion made by Fournelle (2006), of the existence of chemical shifts in the Mg Ka peaks of various Mg-oxides and Mg-rich silicate minerals of significance in mineralogic and petrologic studies. The instrument (Bonnelle et al, 1994) utilizes a high resolution Johann-type beryl crystal (10-10; 2d of 15.9 Å) in a 500 mm diameter Rowland Circle configuration. Ten oxides and silicate minerals, in addition to a Mg-alloy reference standard, were excited with 8 keV electrons produced by a Pierce (klystron) electron gun with a Y2O3 cathode. The beam current used was 400 or 800μA in a spot size of approximately 1 cm diameter. The first order Mg Ka peak region was scanned slowly, so that at the peak maxima there were 50K-100K counts/channel for most samples. The detector is a gas-flow proportional counter with P10 gas and polypropylene window. Uncertainity on the energy measurement is ±0.05 eV. Using a Mg/Zr metal as a reference, the following shifts (in eV) were recorded, from minimum to maximum: spinel +0.07, chromite +0.10, MgO +0.15, hornblende +0.23, Fo90 olivine +0.26, diopside +0.27, biotite +0.35, prochlorite +0.37, pyrope +0.38, garnet Alm47Pyr41Gr11 +0.39. These shifts record the slight differences in the inner shell (core) levels for each mineral, due to the differences between minerals in the the nearest neighbors to the Mg atoms: i.e. the exact pairing of atoms with Mg, the number of such neighbors to Mg atoms, and their distances from Mg atoms localizes their wave functions and thusly the emitted photon's energy. In addition, the full width at half maximum (FWHM) was tabulated, with the width being an indication of more than one environment for the Mg atoms. These shifts may seem subtle for many practioners of EPMA, and for some instruments they may not be observable as not all crystals are created equal, and here "less sharp" crystals have a benefit. However, for some instruments, potential errors of a few per cent may occur if peak centering is off center, particularly used quick automated peaking without some verification. With modern electron probes and assumptions of spectrometer reproducibility, one peak position setting is commonly assumed to be valid for all cases; however, as indicated here, the 'extreme' cases -- spinel/chromite/MgO vs garnet -- warrant special attention. The old saying "use standards similar to your unknowns" has meaning here. Refs: C. Bonnelle, F. Vergand, P. Jonnard, J.-M. André, P. Avila, P. Chargelègue, M.-F. Fontaine, D. Laporte, P. Paquier, A. Ringuenet, B. Rodriguez, 1994, Rev. Sci. Instrum. 65, 3466-3471. J. Fournelle, 2006, Eos Trans. AGU, 87(36).