The effect of water on Si and O diffusion rates in olivine and implications for transport properties and processes in the upper mantle

We performed piston cylinder experiments (1200–1350 °C, 2 GPa) to determine the diffusion rates of Si and O in mantle olivine under water undersaturated (brucite absent, 45 ppm H2O in olivine) as well as close to water-saturated (brucite present, ∼370 ppm H2O in olivine) conditions. Diffusion couples consisted of oriented and polished San Carlos olivine cylinders coated with thin (∼few 100 nm) films of the same composition enriched in 29Si and 18O, with a protective coating of ZrO2 on top. Relationships between water solubility in olivine and water fugacity, combined with thermodynamic equilibrium calculations, indicate fH2O ∼1 GPa, fO2 ∼IW buffer for brucite absent and fH2O ∼9 GPa, fO2 ∼QFM buffer for brucite present experiments. We find that under hydrous conditions DSi ≈ DO and diffusion anisotropy is weak to non-existent. Fitting the raw data at 2 GPa and fH2O ∼0.93 GPa yields Arrhenius parameters [Do and Ep in D = Do exp(−Ep/RT)] of: 1.68 (±3.52) × 10−7 m2 s−1 and 358 ± 28 kJ mol−1 for Si, and 1.43 (±1.80) × 10−4 m2 s−1 and 437 ± 17 kJ mol−1 for O, respectively (1 sigma errors). D (2 GPa, fH2O = 0.97 GPa, 1200 °C): D (1 atm., dry, 1200 °C) is 1000 for Si and 10 for O, respectively. Equations incorporating explicitly the effect of water are discussed in the text. is of our data suggests that O diffuses by an interstitial mechanism whereas Si diffuses via vacancy complexes. The relation between the water fugacity and the Si diffusion rates seems to obey a power law with a water fugacity exponent of 0.2–1. The amount of H incorporated into olivine at the experimental conditions is orders of magnitude higher than the likely concentration of Si vacancies. Therefore, a small fraction (∼0.01%) of the total incorporated H in olivine suffices to considerably enhance the concentration of Si vacancies, and hence diffusion rates. Activation energies for O diffusion under dry and wet conditions are similar, indicating that the mechanism of this diffusion does not change in the presence of water. This inference is consistent with results of computer simulations. ation creep in olivine under wet conditions appears to be controlled by both, Si as well as O diffusion. Absolute creep rates can be calculated from the diffusion data if it is assumed that climb and glide of dislocations contribute equally to creep. Finally, analysis of the various transport properties indicate that <10 ppm of water in olivine is sufficient to cause a transition from “dry” to “wet” laws for most processes. As these water contents are even lower than the observed water contents in most mantle olivines (i.e. minimum values measured at the surface), we conclude that results of water present but undersaturated kinetic experiments are directly applicable to the mantle. Indeed, “wet” kinetic laws should be used for modeling geodynamic processes in the upper mantle, even if the mantle is thought to be undersaturated with respect to water.