Nd EVIDENCE FOR EARLY EARTH DIFFERENTIATION

It is widely accepted that the lunar mantle went through a stage of generalized melting, the so-called magma ocean, which gave rise to the radial minera-logical layering of the planet. Similar evidence is totally missing on Earth for which no remnants of a very early crust are known and mostly indirect arguments. Because of the large Earth's radius, the amount of energy liberated by impactors and by core differentiation seems to have been sufficient to melt the terrestrial mantle down do several hundreds of kilometers [1-2]. The Earth's strong gravity field and the higher water contents of the terrestrial mantle restrict the presence of plagioclase on the liquidus of basaltic melts to shallower levels than on the Moon [3]. The petrological nature (anorthosite vs granite) of a buoyant crust blan-keting a molten terrestrial upper mantle, and even its very existence, therefore remains problematic. A strong argument in favor of early differentiation of the Earth would be the presence of extinct radioactivity anomalies in the mantle source of basalts. Chronometers with a half-life in the range of 1 to100 million years would fractionate in different mantle and crustal layers with parent/daughter ratios that would strongly depend on when the fractionation took place. Samarium-146 decays to 142 Nd with a half-life of 103 My and fractionation of Sm from Nd by crystallization of high-pressure silicates from a magma ocean, notably garnet, majorite, and perovskite, makes this chro-nometer a potential indicator of the presence of a terrestrial magma ocean. A sizeable deficit of 142 Nd relative to the bulk planetary composition formed during condensation has already been identified in various types of meteorites, notably chondrites, mesosiderites, and eucrites [4-7]. This deficit of about 250 ppm (or 2.5 epsilon units) corresponds to a 146 Sm/ 144 Sm ratio at the time the Solar System formed of about 0.0076 ± 0.0005, a value that is divided by two every 100 My. Harper and Jacobsen [8] identified a measurable excess of 142 Nd in a metasediment from the Early Ar-chean supracrustals of Isua, dated at ca. 3.8 Ga by a number of methods and concluded that the terrestrial mantle went through a stage of major differentiation about 4.50 ± 0.10 Gy ago. Although this measurement was replicated many times in the same laboratory, a single anomalous sample and the small amplitude of the anomaly at the limit of the analytical uncertainty (35 ppm) called for a …