No abstract

Terrestrial planets are thought to have formed through collisions between large planetary embryos of diameter ∼1,000-5,000 km. For Earth, the last of these collisions involved an impact by a Mars-size embryo that formed the Moon 50-150 million years (Myr) after the birth of the Solar System. Although model simulations of the growth of terrestrial planets can reproduce the mass and dynamical parameters of the Earth and Venus, they fall short of explaining the small size of Mars. One possibility is that Mars was a planetary embryo that escaped collision and merging with other embryos. To assess this idea, it is crucial to know Mars' accretion timescale, which can be investigated using the (182)Hf-(182)W decay system in shergottite-nakhlite-chassignite meteorites. Nevertheless, this timescale remains poorly constrained owing to a large uncertainty associated with the Hf/W ratio of the Martian mantle and as a result, contradicting timescales have been reported that range between 0 and 15 Myr (refs 6-10). Here we show that Mars accreted very rapidly and reached about half of its present size in only 1.8(+0.9)(-1.0) Myr or less, which is consistent with a stranded planetary embryo origin. We have found a well-defined correlation between the Th/Hf and (176)Hf/(177)Hf ratios in chondrites that reflects remobilization of Lu and Th during parent-body processes. Using this relationship, we estimate the Hf/W ratio in Mars' mantle to be 3.51 ± 0.45. This value is much more precise than previous estimates, which ranged between 2.6 and 5.0 (ref. 6), and lifts the large uncertainty that plagued previous estimates of the age of Mars. Our results also demonstrate that Mars grew before dissipation of the nebular gas when ∼100-km planetesimals, such as the parent bodies of chondrites, were still being formed. Mars' accretion occurred early enough to allow establishment of a magma ocean powered by decay of (26)Al.

The Earth formed by accretion of Moon- to Mars-size embryos coming from various heliocentric distances. The isotopic nature of these bodies is unknown. However, taking meteorites as a guide, most models assume that the Earth must have formed from a heterogeneous assortment of embryos with distinct isotopic compositions. High-precision measurements, however, show that the Earth, the Moon and enstatite meteorites have almost indistinguishable isotopic compositions. Models have been proposed that reconcile the Earth-Moon similarity with the inferred heterogeneous nature of Earth-forming material, but these models either require specific geometries for the Moon-forming impact or can explain only one aspect of the Earth-Moon similarity (that is, O). Here I show that elements with distinct affinities for metal can be used to decipher the isotopic nature of the Earth's accreting material through time. I find that the mantle signatures of lithophile O, Ca, Ti and Nd, moderately siderophile Cr, Ni and Mo, and highly siderophile Ru record different stages of the Earth's accretion; yet all those elements point to material that was isotopically most similar to enstatite meteorites. This isotopic similarity indicates that the material accreted by the Earth always comprised a large fraction of enstatite-type impactors (about half were E-type in the first 60 per cent of the accretion and all of the impactors were E-type after that). Accordingly, the giant impactor that formed the Moon probably had an isotopic composition similar to that of the Earth, hence relaxing the constraints on models of lunar formation. Enstatite meteorites and the Earth were formed from the same isotopic reservoir but they diverged in their chemical evolution owing to subsequent fractionation by nebular and planetary processes.