EXPERIMENTAL CONSTRAINTS ON MELTING CONDITIONS RELEVANT TO CORE FORMATION

摘要

Our recent melting experiments on the major mantle and core materials at high pressure have shown that the very high melting temperatures in the lower mantle preclude models of deep magma oceans in the lower mantle unless the Earth was heated substantially by giant impacts. The melting temperatures of the Fe-O-S system are slightly above and nearly parallel to the present-day geotherm in the lower mantle. Thus, at the time of core formation, temperatures in the lower mantle most likely exceeded not only the Fe-O-S solidus but also the melting temperatures of endmembers in this system, thus providing simple conditions for metal segregation. Eutectic lowering of the melting temperatures is not required. There is evidence that the previously observed chemical reactions between Mg-Si-perovskite and liquid iron were at least partly due to the presence of moisture in the samples, because under dry conditions such reactions could not be observed visually, in stark contrast to undried conditions. Reaction zones of yet unknown nature between iron and MgO, however, are clearly visible even after moderate heating. The core-mantle boundary and D '' region most likely reflect a combination of large thermal and chemical gradients. Thermal discontinuities are inferred from both observed radial and lateral velocity gradients and new experimental results: sound velocity measurements at mantle pressures show an increase in dln/dlnv with pressure, which when combined with the strong decrease in the thermal expansion coefficient with pressure imply large lateral temperature variations near the base of the lower mantle. Indeed, the melting relationship measurements of iron and iron-oxygen-sulfur compounds measured at pressures up to 2 Mbar imply a temperature discontinuity across the core-mantle boundary in excess of 1300 K. Melting temperatures of Mg,Fe,Si-perovskite and magnesiowustite are estimated to lie above 7000 and 5000 K, respectively, thus allowing large temperature variations at the bottom of the mantle without large scale melting and normal viscosity-rheology. [References: 30]