- Physical and chemical constraints on core - mantle differentiation in terrestrial planets. (2007)
- In this study a physical mechanism and geochemical parameters have been examined in high pressure (P) and high temperature (T) experiments in order to place constraints on the conditions and the manner by which core-mantle differentiation occurred on Earth and terrestrial planets. The wetting characteristics of liquid Fe-Si alloys in a matrix of the respective predominating stable silicate mantle mineral (forsterite, silicate perovskite) at pressures of 2-5 and 25 GPa and temperatures of 1600-2000°C have been studied by determining the liquid metal - solid silicate contact angles. The median angle values from texturally-equilibrated samples were found to be independent of P, T, silicate mineralogy and the Si content in the metal fraction and range between 130° and 140° which is far above the critical wetting boundary of 60°. Therefore, within the studied range of conditions dissolved Si does not lower the surface energies between Fe-rich liquids and silicate mantle grains. As a consequence, under reducing conditions the presence of Si in the metal phase of planetary bodies would not have enhanced percolative flow as an effective metal-silicate separation process. The effects of P, T and oxygen fugacity on the liquid metal - liquid silicate partitioning behaviour of the elements Ta, Nb, V, Cr, Si, Mn, Ga, In and Zn have been studied experimentally over a wide range of high-P and high-T conditions of 2-24 GPa, 1750-2600°C and at low oxygen fugacities of -1.3 to -4.2 log units below the iron wüstite buffer. With the derived parameters the respective element depletions in the mantle can be tested under various conditions suggested in core formation models. These data indicate that Nb can serve as an important constraint on oxygen fugacity and P for metal-silicate equilibration. Core formation must have occurred at conditions significantly greater than 20 GPa in order for Nb not to have been massively depleted under conditions necessary to deplete the weakly siderophile element V. Moreover, our study shows that the volatile elements Mn and Ga, would experience strong fractionations in any core-mantle equilibration scenario at pressures below 60 GPa and temperatures at least as high as the peridotite liquidus, while their observed abundances in the mantle is near-chondritic. To a more extreme extent such an observation has been made for the elements Zn and In for which pressures over 80 GPa may be required to explain their near-chondritic ratio in the mantle. Based on these observations we find strong support for the existence of a deep magma ocean during metal-silicate separation, which is an essential component in current polybaric multi-stage core formation models. Although these models succeed in reproducing the observed mantle abundances of many siderophile elements, and can be constrained based on the partitioning behaviour of elements such as Nb, the observed behaviour of the volatile elements Mn, Ga, Zn and In may call for an additional process. Such a process may be the late accretion of volatiles in material that did not undergo core-mantle separation or strong fractionation processes in the condensing nebula that are reflected in the meteorite record. In a third study, the first liquid metal-liquid silicate partitioning data at high pressures up to 18 GPa and high temperatures up to 2500 °C have been obtained for the highly siderophile elements (HSEs) Ru, Rh, Pd, Re, Ir and Pt. This group of elements presents a number of experimental and analytical difficulties, mainly due to their extreme metal-silicate partition coefficients. In addition to refining the experimental technique we have succeeded in producing suitable standards for trace analysis of these elements in quenched silicates using LA-ICP-MS. This study shows that both increasing P and T would decrease the partition coefficients of all HSEs examined in a way similar to the P effect observed for the siderophile elements Ni and Co. This involves two pressure regimes with a strong decrease of the partition coefficients at < 6 GPa, but only a weak P dependence at higher pressures. This difference in P effect can most likely be assigned to structural changes in the silicate melt. In order for the determined partition coefficients to be used quantitatively in models for the Earth, the data have to be corrected from the experimental level of large HSE-concentrations in the metallic phase to infinite dilution. Using Rh as an example for which data exist to perform such a correction, it can be shown that the principal P and T trend does not change significantly once the correction for dilution is performed. From this we can conclude that the P-effect would not be sufficient to decrease the partition coefficients to a degree that the mantle concentrations of the HSEs could be explained. Therefore, a process such as the accretion of an undifferentiated late veneer seems to be necessary.