H diffusion in orthopyroxene and the retention of mantle water signatures

Abstract

Natural gem quality orthopyroxene crystals were experimentally dehydrated between 720 °C and 1020 °C, varying silica activity (buffered by olivine-orthopyroxene or orthopyroxene-quartz), oxygen fugacity (log fO2 relative to the quartz-fayalite-magnetite buffer (QFM) of around +1, −1 and −7) and composition (two different orthopyroxene types) to study these parameters' effects on both the diffusivity of hydrogen and the Fourier transform infrared (FTIR) spectra in orthopyroxene. Hydroxyl profiles were measured across partially dehydrated grains, from which H diffusion coefficients were extracted. Fitting these H diffusion coefficients to the isobaric Arrhenius relationship, log D = log D0 – Ea/(2.303RT), gives log D0 of −5.11, −4.93 and −5.52 m2/s, and Ea of 139, 149 and 125 kJ mol−1 for the [100], [010] and [001] axes, respectively, where D0 is the pre-exponential factor (m2/s), Ea is the activation energy (kJ mol−1), R is the universal gas constant (kJ mol−1 K−1) and T is the temperature (K). Hydrogen diffusivities at ~ΔQFM-1 and 700 °C to 1000 °C are between log D = −12.5 to −10.4 m2/s. This is similar to the fastest H diffusivity measured in olivine, suggesting diffusion primarily by a ‘proton-polaron’-type mechanism, whereby the movement of protons through the crystal is associated with Fe redox, with the net result being that dehydration should be accompanied by Fe oxidation. The determined diffusion coefficients are ~0.3 log units faster than those determined in H-2H self-diffusion experiments (Stalder and Behrens, 2006). The effect of aSiO2 and ƒO2 (between ~ΔQFM-1 to ~ΔQFM + 1) is only minor or within uncertainty. These data were applied to a natural sample from a xenolith from Las Cumbres in Chilean Patagonia, where orthopyroxene shows > 500 µm H loss profiles in length, and olivine shows 50–200 µm Fe-Mg, Y, Ti, V and Sc profiles on millimeter sized crystals. Modelled timescales calculated from a H profile in orthopyroxene, using our H diffusivities along with published values, are considerably shorter (minutes to days) compared to those derived from Fe-Mg inter-diffusion in olivine (weeks to years) from the same xenolith using Fe-Mg diffusivities in olivine. This discrepancy is likely due to H-loss in the natural orthopyroxene representing the timescale between the onset degassing of the host magma and emplacement, whereas Fe-Mg inter-diffusion records the time elapsed between xenolith entrainment and final emplacement. This supports previous assertions that H diffusion in orthopyroxene is sufficiently fast such that many xenolith-hosted orthopyroxene crystals may have suffered significant re-equilibration with the host magma H2O prior to later degassing-induced H-loss.