Slowly, geochemists as well as planetary scientists have been taking up the implications of a likely infernal origin for the Earth-Moon system that resulted from a Mars-size planet colliding with the proto-Earth, shortly after planetary accretion.  The chemistries of both Earth and Moon have sufficient similarities for a common origin to be almost certain.  There is one difference: lunar rocks are more depleted in volatiles than those accessible on the Earth.  Terrestrial rocks were at some stage in their evolution purged of some volatile elements.  The Moon’s early history seems to be extraordinarily simple.  It is recorded in the pale rocks of the lunar highlands that are made dominantly of feldspars.  Their low density and abundance suggest that feldspars floated to the top of completely molten rock, in much the same way as similar anorthosites on Earth seem to have formed in large magma chambers. The difference is that lunar anorthosites probably once formed the entire crust of the early Moon, and formed by simple differentiation of a deep, all-encompassing magma ocean.  The late Dennis Shaw applied this simple notion to the Earth’s earliest evolution during the 1970s, but his vision was largely ignored by his geochemist peers.  A mantle-wide zone of complete melting was resurrected when William Hartmann’s giant impact theory appeared: the energy involved seems to make this an inevitable corollary of his idea.

Indirect analysis of the mantle from the geochemistry of its basaltic products has shown that the mantle is not homogeneous.  Some has been partially stripped of basalt-forming elements, and there are other chemical heterogeneities.  However, examined from the standpoint of isotopes of neodymium (¹⁴²Nd and ¹⁴⁴Nd) more or less every magmatic rock has been considered to have been ultimately derived from material with the same isotopic composition as chondritic meteorites, and by extension, that of the Galaxy in the vicinity of what became the Solar System.  That observation has been a major counter argument to the notion of an early terrestrial magma ocean. Differentiation of such a fundamentally molten Earth would have separated some of the samarium-146 (the source of ^142‑Nd through radioactive decay) from ¹⁴⁴Nd, thereby imparting different growth histories for ¹⁴²Nd/¹⁴⁴Nd ratios to different mantle ‘reservoirs’.  The half-life of ¹⁴⁷Sm is about 100 million years, so that radiogenic ¹⁴²Nd would accumulate most in Earth’s early history, thereafter tending towards a constant proportion of neodymium, unlike the ¹⁴³Nd used in radiometric dating that accumulates much more slowly from decay of ¹⁴⁷Sm (half life about 100 billion years).

There was a flaw in this counter argument.  The similarity of chondritic and terrestrial Nd isotope patterns might have stemmed from isotopic measurements that were insufficiently precise to detect significant differences. Mass spectrometry has undergone a near-quantum leap in precision.  Applied to the chondrite-Earth rock comparison, the neodymium data for chondrites remains as determined earlier, but the ¹⁴²Nd/¹⁴⁴Nd ratios of terrestrial rocks turn out to be 20 parts in a million higher than for chondrites (Boyet, M & Carlson, R.W. 2005. ¹⁴²Nd Evidence for Early (>4.53 Ga) Global Differentiation of the Silicate Earth.  Science, Published online June 16 2005; 10.1126/science.1113634).  That doesn’t seem very much, but quite sufficient to suggest plausibly that indeed the Earth’s mantle did indeed evolve from a magma ocean.  Its upper part was enriched in samarium by its fractionation as a solid that probably crystallised downwards.  Whatever was left of the original liquid would be at the base of the protomantle, and in it many other elements that favoured melt over crystals – so-called ‘incompatible’ elements – would have been enriched.  Boyet and Carson suggest that such a deep, enriched layer may amount to between 5 to 30% of the current mass of the mantle. 

The implications, if the ideas are confirmed, are enormous, because geochemists up to now have taken the bulk of the mantle that supplies basalt magmas – and whose composition is quite well constrained – to represent the whole silicate Earth.  That may satisfy geochemical parameters, but worries geophysicists.  The ‘standard’ Earth has insufficient radioactive uranium, thorium and potassium to account for the heat that flows to the surface. In fact it generates about a half, leaving the rest to speculation. One school looks to supposed gravitational potential energy locked in the core when it formed by inward collapse of iron-nickel alloy and slowly released thereafter.  Another theorises about radioactive potassium-40 combined in sulphides of the core, which also ‘leaks’ out.  The possible existence of the last dregs of an early magma ocean, near the core-mantle boundary (CMB), would not only account for 43% of surface heat flow, but might also drive convection in the liquid outer core as a means of generating Earth’s magnetic field.  Even more important, it might fuel the rise of plumes from the CMB that are increasingly implicated in periodic repaving of the Earth’s surface by flood-basalt volcanism.  Since flood basalts are a popular source for mantle geochemists’ data, why are the signs of such a peculiar source region not clear in their analyses?  Either they are not looking with the requisite precision, or the source itself does not move with plumes, merely setting them in motion.  Eminent geochemists see a bit of a hectic time ahead….. 

See also: Kerr, R.A. 2005. New geochemical benchmark changes everything on Earth.  Science, v. 308, p. 1723-1724.