Melting phase relations in the MgO–MgSiO3 system between 16 and 26 GPa: Implications for melting in Earthʹs deep interior

Melting experiments in the system MgO–MgSiO3 were performed to study the eutectic melt composition and phase relations between 16 and 26 GPa using the mutlianvil apparatus. By employing a multi-chamber capsule design, several different starting compositions along this binary join could be run at a single pressure and temperature and internally consistent phase relations and liquidus compositions were obtained. A single eutectic composition was identified on this join which becomes progressively more MgO-rich over the investigated pressure range. A simple thermodynamic model is developed to describe the melting phase relations based on literature models for melting curves of end-members and a symmetric liquid mixing model. shown that melting relations of a natural peridotite composition at high pressure can be reasonably approximated on the basis of the phase relations of this simple binary, and a very good agreement is found once the effects of FeO on phase relations and melting temperatures are considered. ermodynamic model, extrapolated to pressures covering the entire mantle, predicts that the eutectic composition becomes richer in MgO up to approximately 80 GPa, where it becomes near constant with pressure and has a Mg/Si ratio close to that of a peridotite composition. By applying further experimental results to account for the effect of FeO on the melting temperatures, the model predicts that the solidus and liquidus for a peridotitic composition at lower mantle pressures are never more than ∼250 K apart. results can be used to examine a partial melt origin for the existence of localised zones with ultra low shear wave velocities (ULVZ) at the core–mantle boundary (CMB). The solidus temperature of peridotitic mantle at the CMB is estimated to be 4400±300 K, which would require temperatures at the CMB to be at the very top end of the estimated range for melting to occur. The proximity of the solidus and liquidus temperature predicted by the model, however, implies that large melt fractions could form over small depth intervals as a result of relatively small increases in either temperature or FeO content. This is consistent with a seismically sharp transition in the upper boundary of ULVZ layers. Given the high temperatures required to melt mantle peridotite at the CMB, a partial melt origin due to raised FeO contents seems more plausible.