Siderophile geochemistry of ureilites: A record of early stages of planetesimal core formation

New bulk-compositional data, including trace siderophile elements such as Ir, Os, Au, and Ni, are presented for 25 ureilites. Without exception, ureilites have siderophile abundances too high to plausibly have formed as cumulates. Ureilites undoubtedly underwent a variety of “smelting,” by which C was oxidized to CO gas while olivine FeO was reduced to Fe-metal. However, pressure-buffered equilibrium smelting is not a plausible model for engendering the wide range (75–96 mol%) of mafic-silicate core mg among ureilites. The smelting reaction produces too much CO gas. Even supposing a disequilibrium process with the smelt-gas leaking out of the mantle, none of the ureilites, least of all the ureilite with the most “reduced” (highest) olivine-core mg (ALH84136), has the high Fe-metal abundance predicted by the smelted-cores model. In principle, the Fe-metal generated by smelting could have been subsequently lost, but siderophile data show that ureilites never underwent efficient depletion of Fe-metal. Ureilites display strong correlations among siderophile ratios such as Au/Ir, Ni/Ir, Co/Ir, As/Ir, Se/Ir, and Sb/Ir. Ureilite siderophile depletion patterns loosely resemble siderophile fractionations, presumably nebular in origin, among carbonaceous chondrites. However, Zn, for an element of moderate volatility, is anomalously high in ureilites. A tight correlation between Au and Ni extrapolates to the low-Ni/Au side of the compositional range of carbonaceous chondrites. From this mismatch, mild but nonetheless significant depletions of refractory siderophile elements such as Ir and Os, and moderate depletions of strongly siderophile, weakly chalcophile elements such as Ni and Au, we infer that the ureilite siderophile fractionations are largely the result of a non-nebular process, i.e., removal of S-rich metallic melt, possibly with minor entrainment of Fe-metal. Several lines of trace-element evidence indicate that melt porosity during ureilite anatexis was at least moderate. The ureilite pattern of very mild depletions of extremely siderophile elements, but much deeper depletions of moderately siderophile, chalcophile elements, suggests that asteroidal core formation probably occurs in two discrete stages. In general, separation of a considerable proportion (several wt%) of S-rich metallic melt probably occurs long before, and at a far lower temperature than, separation of the remaining S-poor Fe-metal. Apart from the Fe-metal itself, only extremely siderophile elements wait until the second stage to sequester mainly into the core.