Abstract

We have measured Fe-Mg interdiffusion coefficients, DFe-Mg, parallel to the three main crystallographic axes in two natural orthopyroxene single crystals [approximately En98Fs1 (XFs = XFe = 0.01) and En91Fs9] using diffusion couples consisting of a 20–90 nm thick silicate thin film deposited under vacuum on polished and oriented pyroxene single crystals. The thin films were prepared using pulsed laser ablation of polycrystalline olivine pellets (composition: Fo30Fa70). Samples were annealed for 4–337 h at 870–1100 °C under atmospheric pressure in a continuous flow of CO + CO2 to control the oxygen fugacity, fO2, between 10−11 and 10−7 Pa within the stability field of pyroxene. Film thickness and compositional profiles were measured using Rutherford backscattering spectroscopy (RBS) on reference and annealed samples, and Fe concentration depth profiles were extracted from the RBS spectra and fitted numerically considering a compositional dependence of DFe-Mg in orthopyroxene. We obtain an Arrhenius relationship for both types of crystals, but only for the more Fe-rich composition a dependence on fO2 could be clearly identified. For diffusion along [001] in the composition Fs9, least-squares regression of the log DFe-Mg vs. reciprocal temperature yields the following Arrhenius equation:  
DFe-Mg[m2/s]=1.12×106(fo2[Pa])0.053±0.027exp[308±23[kJ/mol]/(RT)].
DFe-Mg in Opx with XFe = 0.01 obeys a relationship that does not depend on fO2:  
DFe-Mg[m2/s]=1.66×104exp[377±30[kJ/mol]/(RT)].

Diffusion along [001] is faster than diffusion along [100] by a factor of 3.5, while diffusion along [010] is similar to that along [001]. Comparison of DFe-Mg and rates of order-disorder kinetics indicates that for fO2 around the IW buffer and lower, diffusion in natural orthopyroxene becomes insensitive to fO2, which could be related to a transition in the diffusion mechanism from a transition metal extrinsic (TaMED) domain to a pure extrinsic (PED) domain. This behavior is analogous to that observed for Fe-Mg diffusion in olivine and this complexity precludes the formulation of a closed form expression for the composition and fO2 dependence of DFe-Mg in orthopyroxene at present. We were not able to quantitatively constrain the dependence of DFe-Mg on the XFs content from the profile shapes, but consideration of the experimentally measured diffusion coefficients along with the data for order-disorder kinetics suggests that the compositional dependence is weaker than previously estimated, at least for orthopyroxene with XFe < 0.5. For the major element compositional T and fO2 range of available experimental data, Fe-Mg interdiffusion in orthopyroxene is slower than in olivine and aluminous spinel, comparable to garnet, and faster than in clinopyroxene.

You do not currently have access to this article.