Experimental investigation of hydrogen isotope fractionation during hydration of olivine-hosted melt inclusions: Implications for D/H in Baffin Island picrites

Olivine-hosted melt inclusions are used to investigate the hydrogen isotope compositions (D/H) of Earth's mantle reservoirs. Studies have shown, however, that hydrogen in melt inclusion can equilibrate rapidly in response to changes to the external environment via the diffusion of protons (H⁺) and deuterons (D⁺) through the host olivine. Given that protons diffuse faster than deuterons, a kinetic fractionation of hydrogen isotopes is expected to accompany both the hydration and dehydration of melt inclusions. Other volatile species in the melt inclusion may also be affected by changes to the internal pressure and/or oxygen fugacity (fO₂). Here we report results from experiments designed to investigate the behaviors of volatiles and hydrogen isotopes during the hydration of olivine-hosted melt inclusions. We show that the concentration of H₂O in initially H₂O-poor inclusions increases rapidly (up to ∼4 wt.% within 24 h) when the host olivine is in contact with aqueous fluid at 1200 °C and 300 MPa. The extent of hydration is controlled by time, temperature, melt inclusion volume, and H⁺ diffusion distance. Hydrogen isotopes initially become lighter (i.e., D/H decreases) as hydration proceeds, defining a negative correlation with H₂O concentration. This trend reverses with increasing hydration as the inclusions must eventually equilibrate with the external fluid. These experimental results agree well with diffusion calculations carried out using a spherical geometry and a lattice diffusivity of 10^–11.2±0.2 m²/s for H⁺ at 1200 °C. Therefore, anomalously light hydrogen isotopes in olivine-hosted melt inclusions from Baffin Island may not be taken as representative of the composition of the mantle source unless post-entrapment hydration can be excluded as a possibility. An increase in CO₂ concentration and a significant drop in sulfur concentration accompany hydration of the melt inclusions in our experiments. The former is consistent with an observed decrease in vapor bubble size and results from a hydration-induced internal pressure increase. The latter is ascribed to the exsolution of molten sulfide from the silicate melt, which might be related to a lower fO₂ in the experiments as compared to the starting materials. We find no evidence for exchange of F or Cl between the melt inclusion and external fluid.