What can we really learn about the mantle from water contents of minerals in xenoliths?

Mantle xenoliths are pieces of the interior of the Earth that have been entrained by magma and brought to the surface, frequently quite rapidly. As such they often contain geochemical signatures characteristic of their source regions, although interactions with the magma prior to and during transport and depressurization can result in alteration and overprinting. A number of recent studies have investigated the water contents in the nominally anhydrous minerals within xenoliths from a range of localities and attempted to provide constraints on the water content of the mantle source regions. They have made use of the experimentally observed rates of water loss from the various minerals to argue that several minerals retain their water contents during transport, while olivine grains, in particular, are significantly dehydrated. Using these observations, estimates have been made of rates of ascent from the mantle source region and of mantle water content. At issue is the robustness of the arguments for water loss based on experimental studies and their true applicability to the issue of xenolith entrainment and transport. In olivine, the subject of most experimental study for water diffusion and solubility, diffusive transport of water-derived species occurs by 2 processes: a redox process that is rate limited by self-diffusion of hydrogen, and a slower process whereby defect associates composed of a hydrogen ion and a metal vacancy diffuse cooperatively, rate limited by the diffusivity of the metal vacancies. Water loss by the redox process is limited by the number of iron atoms that can change from ferrous to ferric; many more hydrogen ions can be lost as hydrogen-vacancy associates. It is anticipated that both processes also occur in the pyroxenes and garnet, although only the redox process has been studied in detail. A further complication to the use of experimentally determined hydrogen diffusivities results from the demonstrated compositional effects on diffusion. In low-iron clinopyroxene, diffusion is much slower than in high-iron clinopyroxene, resulting from lower fluxes of polarons at lower iron concentrations. While these issues must be considered in any analysis of mantle xenoliths, judicious use of experimental data does provide useful constraints on mantle water content and on xenolith ascent rates.