Painstaking work on meteorites and their re-evaluation has only a small, non-specialist readership, but now and again developments in the science and its bearing on how the Solar System and its planets formed need a review. The latest of these (Wood, B.J. et al. 2006. Accretion of the Earth and segregation of its core. Nature, v. 441, p. 825-833) doesn’t deviate much from generally accepted ideas, except in detail. For a long while it has seemed inescapable that gravitational potential energy accumulated from accretion of mass, together with energy released by decaying short-lived isotopes formed by a supernova near the dust cloud that gave birth to the Solar System would have led to hot protoplanets. So core formation by segregation of dense immiscible metal and sulphide melts was likely to have been sooner rather than later – such melts form at lower temperatures than do those made of silicates.

The daughter isotope (¹⁸²W) of one short-lived isotope (¹⁸²Hf) is especially revealing in both meteorites and the Earth. Hafnium favours entry into silicates while tungsten has an affinity for metallic iron; they are siderophile. So, when metallic melts form in a silicate body the Hf/W ratio increases in the silicates. If that segregation occurs before most ¹⁸²Hf has decayed – within about 45 Ma – then the silicate part will express an excess of ¹⁸²W while metals have a deficiency. In the case of metallic meteorites, ¹⁸²W is so low as to indicate segregation of the metal from silicate within less than 5 Ma of the ultimate origin of the Solar System. Inevitably, Earth would have incorporated some of these early-formed metallic parts during its accretion. Tungsten isotopes from terrestrial rocks, however, suggest that core formation lasted about ten times longer, and imply that this early metal re-mixed with silicate in the mantle during accretion, and formation of the core was a secondary product of heating of the growing planet. The mantle has an excess of siderophile elements, which poses a problem. There are three possibilities: core formation was never completed, some of these elements remaining locked in silicate; it took place while overall chemical conditions changed from reducing to oxidizing, so that the most siderophile ended up in the core during the reduced phase then less siderophile elements progressively favoured silicate entry as conditions became oxidising; as the Earth grew the pressure under which segregation of core materials increased. The third scenario invokes a deep ‘ocean’ of magma through which droplets of metal fell, equilibrating with silicate melt and then forming a pond on the ‘ocean’ floor, ultimately to descend as large masses.

Wood et al. examine these three scenarios in the light of recent data and planetary modelling, suggesting that the second was the most likely by a process of ‘self-oxidation’ as its size increased, perhaps linked with the formation of perovskite in the deep mantle once a limiting radius had been achieved. Such a heterogeneous accretion and core segregation would explain the disparity between estimates of the timing of the core from tungsten and lead isotopes (~12 and ~28 Ma respectively)   They also revisit the oddly low density of the liquid outer core – about 8% less than expected of an iron-nickel alloy, ascribing it to a mixture of the low-atomic weight elements, silicon, sulphur, carbon and hydrogen, with an unknown proportion of oxygen.