The composition of hydrous partial melt at 410 km: Geodynamic implications (Invited)

Geophysical observations of low S-wave velocities and high electrical conductivities at 410 km depth have been interpreted as evidence for the presence of silicate melt on top of the Earth’s transition zone. It has been suggested that the difference in the water storage capacity of upper mantle versus transition zone minerals may cause dehydration melting as material up-wells across the 410. A key question is whether this hydrous partial melt is neutrally buoyant in the mantle at these conditions. In order to assess these possibilities it is important to determine the likely composition of small degree hydrous melts at these conditions and to measure the H2O contents of mantle minerals coexisting with this melt phase. The composition of a hydrous melt in equilibrium with a mantle peridotite composition has been determined at conditions equivalent to the top of the 410 km seismic discontinuity i.e. 13 GPa and 1550 oC. Sandwich experiments were performed in the Fe-free CMASH system where an “initial-guess” hydrous melt composition was equilibrated with 50% anhydrous peridotite. Fe-free compositions ensured that H2O was not lost during the experiments as H2, because H2O contents must be determined through mass balance using the starting bulk composition. The resulting melt composition was used to assemble a further melt, which was then equilibrated in the same way. After several iterations it was possible to derive a melt composition, which was in equilibrium with a mineral assemblage identical to that observed for an anhydrous peridotite composition at the same conditions. In addition we performed further experiments to assess the Fe-Mg partitioning between hydrous silicate melt and mineral phases at the same conditions. With these partition coefficients the composition of the melt phase could be calculated. We found the equilibrium melt to contain 7 wt % H2O but to have a lower FeO content then previously predicted. The results of melt density calculations indicate that this melt would not be neutrally buoyant at conditions compatible with the base of the upper mantle.