Experimental constraints on the partitioning of Ru, Rh, Ir, Pt and Pd between chromite and silicate melt: The importance of ferric iron

We have performed partitioning experiments to assess the role of chromium-rich spinel in controlling the behavior of the platinum-group elements (PGEs) during igneous differentiation. Spinels were equilibrated with natural and synthetic iron-bearing basalt at 0.1MPa and 2GPa at 1400–1900°C over an fO₂ range of IW+1.6 to IW+7. Results from relatively reduced, graphite-encapsulated experiments done at 2GPa indicate that Ru is compatible in Cr-spinel (mineral/melt partition coefficient, D, of ~4), followed by Rh and Ir, which are moderately incompatible (D range of 0.04 to ~1), with Pt and Pd the most incompatible (D<0.2). Partition coefficients for Ir, Ru and Rh measured at more oxidizing conditions in this and previous studies are 10 to 1000 times higher than results from experiments using graphite capsules. We account for the variation in spinel–melt partitioning with a model which considers both the affinity of the PGE cation for a particular spinel lattice site, and the change in site occupancy accompanying the increase in ferric iron component with fO₂. Assuming that Ir and Rh are present as divalent species, with a strong affinity for VI-fold coordination, DIᵣ and DRₕ are predicted to rise rapidly with the ferric iron component, explaining the large D-values for magnetite-rich spinels. Model results indicate that DIᵣ≤20 and DRₕ are ≤100 for ferric-iron poor, Cr-rich compositions, as would crystallize in komatiites, some layered intrusions, and ophiolites. The overall compatibility of Ru for chromite is consistent with the predominance of Ru³⁺ at experiment conditions and the similarity in the size of Ru³⁺ to Cr³⁺ and Fe³⁺. The increase in DRᵤ with the ferric iron content of the spinel likely involves a strong effect of mineral composition superimposed on a change in melt speciation (Ru²⁺ to Ru³⁺) with increased fO₂. The effect of mineral composition is a consequence of the difference in octahedral site preference energy (OSPE) between Ru³⁺, Fe³⁺ and Cr³⁺, with stronger partitioning of Ru into Fe³⁺-rich compositions due to the enhanced reduction in energy gained by the Ru³⁺ substitution. Ru partition coefficients for ferric-iron poor spinel are expected to be ~30, which is somewhat lower than values estimated from natural samples obtained from in situ chromite analyses. Results indicate that the ferric iron content of chromite exerts a strong control on the partitioning of some PGEs which should be taken into account in both future experimental work and in models of igneous differentiation.