Pressure Effect on Partitioning of Nickel, Cobalt, and Sulfur During Mantle-Core Formation

Carl B. Agee Dept. of Earth and Planetary Sciences, Harvard University, Cambridge, MA 02138, USA

Jie Li Dept. of Earth and Planetary Sciences, Harvard University, Cambridge, MA 02138, USA


Isothermal experiments up to 20 GPa show a dramatic influence of pressure on partitioning of Ni and S between molten Fe-alloy and silicate melt. Both Ni and Co become less siderophile with pressure. Pressure affects Ni much more than Co such that their decreasing partition coefficients (D alloy melt / silicate melt) reach values of 29 and 26 respectively at an extrapolated pressure of ~28 GPa. The observed abundances of Ni and Co and their near chondritic ratio can be explained by alloy-silicate chemical equilibrium at high pressure during core extraction in an magma ocean. The partitioning behavior of sulfur is the opposite of Ni and Co, it becomes more siderophile with pressure. Sulfur's enhanced affinity for Fe-alloy with depth should make it the dominant light element in the Earth's core.


Experiments were performed at pressures of 2, 5, 10, 12, 14, and 20 GPa, under isothermal conditions of 2000šC, with a Walker-style octahedral multi-anvil device. Details of the experimental design and technique can be found in Agee et al. (1995). Starting material was finely ground powder from a split of the Allende meteorite from The Smithsonian Institution. Sample capsules were fashioned from high purity MgO rod. All experiments contained coexisting immiscible liquids of silicate and Fe-rich alloy. Both silicate and alloy liquids, upon quenching, formed discrete, relatively large masses of crystals and glass. The average composition of these domains were determined by multiple broad beam analyses using a Cameca electron microprobe.


In the range 2 to 20 GPa the Ni partition coefficient decreases more than five fold from 318 to 59. Over the same range the Co partition coefficient decreases by less than half from 45 to 27. The trend for sulfur partitioning is the opposite, it increases by seven fold from 74 to 519. Oxygen fugacity was calculated to be near or slightly below the iron-wuestite buffer and varied by only a half log unit in the experiments. The present results are in good agreement with the work of Thibault and Walter (1995) who studied Ni and Co partitioning in a sulfur free system up to 12 GPa.

Implications for Mantle-Core Formation

The high pressure partitioning data can explain one of the most enigmatic aspects of the "excess siderophile problem", namely the near chondritic Ni/Co in the upper mantle. The results are consistent with molten Fe-alloy equilibrating at the base of a silicate magma ocean with depth 750-1100 km. The data do not favor a magma ocean that reached depth of ~2900 km (the present core-mantle boundary) because mantle Ni/Co would become strongly super-chondritic. Shallow magma oceans (<700 km) are also ruled out because of the sub-chondritic Ni/Co produced at lower pressures. The data can also be applied to a scenario in which core formation commenced when the Earth reached one tenth of its present mass or about half of its present radius. In this case, the upper and the lower mantles could have a similar Ni/Co signature. We can also consider implications for sulfur becoming more siderophile with pressure. Because of its moderately volatile nature, the fate of sulfur during accretion is more difficult to assess than elements such as Ni and Co. It is generally agreed that the sulfur content of the upper mantle is two orders of magnitude lower than the concentration of sulfur in chondrites. Such large depletion can be partly explained by volatile loss to space during a high temperature stage of Earth accretion. Alternatively, a significant fraction of the chondritic sulfur may reside in the Fe-alloy that makes up the Earth's core. Our experimental results argue in favor of a sulfur depleted mantle by alloy/silicate equilibria and in turn support the hypothesis that sulfur is the dominant light element in the molten outer core.


Agee, C.B., Li, J., Shannon, M.C. & Circone, S., J. Geophys. Res., 100, 17,725-17,740, (1995).

Thibault, Y. & Walter, M.J., Geochim. Cosmochim. Acta, 59, 991-1002, (1995).