Disproportionation of Fe²⁺ in Al-free silicate perovskite in the laser heated diamond anvil cell as recorded by electron probe microanalysis of oxygen

Abstract

Experimental evidence is reported for Fe²⁺ disproportionation in Al-free perovskite (Pv), when submitted to large temperature gradients (i.e., under off-equilibrium conditions) in a laser heated diamond anvil cell (LHDAC). To enable this effect, the experimental procedure was designed to produce large radial and axial temperature gradients. In the Pv and ferropericlase (Fp) assemblage synthesized after dissociation of natural olivine, the three chemical states of iron (i.e., Fe⁰, Fe²⁺ and Fe³⁺) could be evidenced by electron probe microanalysis (EPMA), through variations of oxygen contents attached to the Fe cations. Despite inherent difficulties for determination of O-contents and Fe³⁺/ΣFe ratios using EPMA, we recorded significant changes in iron oxidation state across the laser-heated strip. These changes are correlated with variations in composition for the major elements (Fe, Mg, and Si), which evidences that the Pv/Fp assemblage experienced large segregation under the strong temperature gradients. Grains of metallic iron were detected in parts of the laser-heated strip coexisting with a Pv phase with Fe/(Mg + Fe) = 6 at% and most of its iron as Fe³⁺. This Fe²⁺-disproportionation reaction involves insertion of Fe³⁺-defects in the Pv lattice. This Fe³⁺-bearing Pv phase is presumably unstable and decomposes into a mineral assemblage including magnesioferrite, which is detected at the border of the laser-heated strip.

Appendix: Criteria for selecting the plots of Fig. 6

A two stage protocol has been established to discard EPMA data affected by topographical effects due to the non-flatness of the sample surface and the presence of voids and cracks onto: (1) analytical wt%-totals, including measured oxygen content, should be encompassed within the interval delimited by the wt%-totals with oxygen calculated by stoichiometry of the measured cations, with Fe oxide as FeO (lower limit), and the same with Fe oxide as Fe₂O₃ (upper limit) (Fig. 7a); (2) the wt%-total values that are calculated to be different by more than 1wt% compared to the mean wt%-total value for the whole sample are discarded (Fig. 7b). The selected plots encompassed within the regions, where vertical hatched bands overlap the horizontal band (Fig. 7b). We observe that no data were selected for the first part of the line scan, due to underestimated oxygen content correlated presumably with a round-shaped surface in contact with epoxy. We also note that the region corresponding to the open symbols (Fig. 7a) is to be discarded. Nevertheless, they were selected as a record of the presence of the metallic phase forming the peak #1 region of the line scan.

Fig. 7

a Differences in the wt%-totals as deduced from analyses with oxygen either measured or calculated by stoichiometry of the measured cations. Diamonds with oxygen attached to iron as Fe²⁺, squares with oxygen attached to iron as Fe³⁺. Hatched area delimit the regions of the linescan where plots have undergone a first selection to construct Fig. 6 (see text). Open symbols presence of a metallic phase (Fe⁰) and should not be discarded. b Final selection for plots exhibiting wt%-totals lying between 96 and 98%. Vertical hatched regions are the same as (a)