GRAIN BOUNDARY DIFFUSION OF SI, MG, AND O IN ENSTATITE REACTION RIMS: A SIMS STUDY USING ISOTOPICALLY DOPED REACTANTS

Abstract

Diffusion-controlled growth rates of polycrystalline enstatite reaction rims between forsterite and quartz were determined at 1,000 °C and 1 GPa in presence of traces of water. Iron-free, pure synthetic forsterite with normal oxygen and silicon isotopic compositions and quartz extremely enriched in 18O and 29Si were used as reactants. The relative mobility of 18O and 29Si in reactants and rims were determined by SIMS step scanning. The morphology of the rim shows that enstatite grows by a direct replacement of forsterite. Rim growth is modelled within a mass-conserving reference frame that implies advancement of reaction fronts from the initial forsterite–quartz interface in both directions. The isotopic compositions at the two reaction interfaces are controlled by the partial reactions Mg2SiO4 = 0.5 Mg2Si2O6 + MgO at the forsterite–enstatite, and MgO + SiO2 = 0.5 Mg2Si2O6 at the enstatite–quartz interface, implying that grain boundary diffusion of MgO is rate-controlling. Isotopic profiles show no silicon exchange across the propagating reaction interfaces. This propagation, controlled by MgO diffusion, is faster than the homogenisation of Si by self-diffusion behind the advancing fronts. From this, and using D Vol Si,En at dry conditions from the literature, results a D′ Si,En δ value of 3×10–24 m3 s–1 at 1,000 °C. The isotopic profiles for oxygen are more complex. They are interpreted as an interplay between the propagation of the interfaces, the homogenisation of the isotope concentrations by grain boundary self-diffusion of O within the rim, and the isotope exchange across the enstatite–quartz interface, which was open to 18O influx from quartz. Because of overlapping diffusion processes, boundary conditions are unstable and D́ Ox,En δ cannot be quantified. Using measured rim growth rates, the grain boundary diffusivity D́ MgO δ of MgO in iron-free enstatite is 8×10–22 m3 s–1 at 1,000 °C and 1 GPa. Experiments with San Carlos olivine (fo92) as reactant reveal lower rates by a factor of about 4. Our results show that isotope tracers in rim growth experiments allow identification of the actual interface reactions, recognition of the rate-controlling component and further calculation of D´δ values for specific components.