2006 Wood et al., Accretion of the Earth and segregation of its core (1119308), страница 4
Текст из файла (страница 4)
New material added to the Earth wasassumed to be mixed with the pre-existing mantle, but isolated fromthe proto-core. The core was segregated in 1% steps, each aliquotbeing in equilibrium with the entire mantle at the time of segregationat pressures constrained by Ni and Co partitioning and temperatureson the peridotite liquidus6. The latter were found to be consistentwith magma oceans extending to 30–40% of the depth of the growingmantle. A large number of oxygen fugacity paths (of which three areillustrated) yield the partitioning of V, W, Nb, Ni, Co, Mn, Cr, Ga, Pand Si, consistent with the estimates of Table 1.
All require, however,that most of the Earth formed under conditions more reducing thanthose implied by the current FeO content of the mantle and that theEarth became more oxidized towards the end of accretion.Additionally, although this is an area of continuing debate53, theseaccretion models require addition of 0.5% of chondritic materialafter cessation of core formation to generate the chondritic ratios ofthe highly siderophile elements such as Pd, Ir and Pt in the silicateEarth (Fig. 1). The current ‘best explanation’ is therefore an amalgamof ‘deep magma ocean’ and ‘heterogeneous accretion’ hypotheses,that is, core segregation at the base of a deepening magma oceanunder progressively more oxidizing conditions6.
The importantquestions then are: what was the mechanism of oxidation andwhen did it operate? That oxidation has occurred is demonstratedby the current high oxygen fugacity of the mantle54.growth the proto-Earth consumed bodies in a narrow feeding zone(0.01 astronomical units, AU ) but that the source of accreting bodiesdramatically widened in the later stages of planetary growth1. It isthus conceivable that late accretion added more oxidized materialsfrom the current position of the asteroid belt1.Dissociation of H2O and escape of H2 from the Earth’s atmosphere55 may also have caused oxidation.
If this is an importantoxidation mechanism, however, it is difficult to see why the mantle ofMars, a more volatile-rich planet than the Earth, with a higher FeOcontent37, appears to be more reduced than the Earth’s mantle56.Finally, a recent hypothesis6,7 (elaborated below) is that the Earthsimply self-oxidized through crystallization of magnesium silicateperovskite.At depths below 660 km in the present-day Earth, the stable phasesare, assuming peridotitic composition, (Mg,Fe)SiO3, magnesiumsilicate perovskite (79% by volume, vol.%; ref. 57), (Mg,Fe)Omagnesiowüstite (16 vol.%) and CaSiO3 perovskite (5 vol.%).
Animportant property of magnesium silicate perovskite is that itdissolves the 4% Al2O3 present in peridotite by a coupled substitution with Fe3þ (refs 58 and 59) as follows:Mg2þ Si4þ $ Fe3þ Al3þIt has recently been discovered that this substitution mechanism is sostable that it forces ferrous iron to disproportionate to ferric iron plusiron metal7:3Fe2þ $ 2Fe3þ þ Fe0ð3ÞOr, in terms of oxide components:3FeO þ Al2 O3 ¼ 2FeAlO3 þ Fe0silicate meltOxidation of the Earth during and after accretionThe oxidation state of the Earth would have increased duringaccretion if late-arriving planetesimals had higher FeO/Fe ratiosthan the early ones8, thus increasing the oxidized iron content ofthe mantle and generating a concomitant increase in oxygen fugacity.This idea, although impossible to quantify, is broadly consistent withplanetesimal theory, which states that during the period of oligarchicFigure 5 | Sketch of a possible mechanism by which the mantle may haveself-oxidized via perovskite crystallization.
Lower-mantle perovskite (Pv)generates Fe3þ and Fe0 (metal) from disproportionation of Fe2þ. Fe metalis swept to the core by descending metal diapirs while the vigour of accretionand core segregation causes dissolution and recrystallization of perovskitewith release of Fe3þ to the overlying magma ocean.ð2Þð4ÞperovskiteþmetalThis means that, as perovskite began to crystallize in the extensivelymolten Earth, it dissolved ferric iron as FeAlO3 component andproduced Fe metal.
The perovskite is stable above 23 GPa, so thisprocess took place over a substantial part of the history of the Earth’saccretion and core formation. The implications are that metalsinking through the lower mantle to the core would inevitably havedissolved some of this internally produced Fe, resulting in a perovskitic layer which was relatively oxidized. Given the gravitationalinstability of any metal layer (Fig. 3), accretional energies and heatloss, the depth of the magma ocean must have fluctuated continuously, thereby generating fronts of dissolution and precipitation atthe lower boundary, as depicted in Fig.
5.Perovskite dissolution and re-precipitation acted as a ferric oroxygen ‘pump’, releasing ferric iron to the magma ocean duringdissolution and producing more during precipitation. As the Earthgrew, ferric iron released to the magma ocean would have beenconsumed by reduced species such as CH4. Droplets of Fe metalfalling through the magma ocean would also have reacted with ferriciron, driving reaction (3) to the left and producing more Fe2þ todissolve in the silicate melt. Hence the oxygen fugacity and thecontent of oxidized iron in the mantle (magma ocean) increasednaturally as a consequence of perovskite crystallization and dissolution.
In the very final stages of earth accretion this mechanism mayhave caused sufficient oxidation to halt metal segregation, setting thestage for the ‘late veneer’ of chondritic or similar material to addthe highly siderophile elements exclusively to the mantle. Theimportance of this self-oxidation mechanism is that it is experimentally observed to take place when magnesium silicate perovskitecrystallizes7.The self-oxidation mechanism removes the need to make ad hocassumptions about the compositions of the materials added to theEarth during accretion.
The oxidation-state changes required bypartitioning data are a simple consequence of the size of the Earth.Any planet in which magnesium silicate perovskite is an importantcrystallizing phase should undergo the same process. It could explainwhy the Earth’s mantle is more oxidized than that of Mars56, a planetin which perovskite is either absent or only stable very close to the© 2006 Nature Publishing Group829REVIEWSNATURE|Vol 441|15 June 2006core–mantle boundary.
Because of its small size, Mars cannot haveundergone the same period of extensive perovskite crystallizationduring accretion as did the Earth.Consequences of ‘self-oxidation’ and the age of the coreThe timing of core segregation can, in principle, be determined from aradioactive decay system in which either the parent or the daughterisotope is siderophile, so that core formation established the currentparent/daughter ratio in the silicate Earth. The systems 182Hf–182Wand 238,235U–206,207Pb are potential core chronometers.
The daughters,Wand Pb, are siderophile and their current isotopic constitutions yieldmodel-dependent ‘ages’ of core formation. Ages are obtained byassuming, for example, instantaneous core segregation at a fixed timeor, more realistically, continuous core segregation as the Earth grew.If it is assumed that all accreting material mixed isotopically withthe silicate Earth, that accretion rate decreased exponentially withincreasingly larger impacts, as predicted from the planetesimaltheory and that the core segregated continuously with fixed parentand daughter partitioning, a time constant for core formation can beobtained (Fig.
6):F t ¼ 1 2 e2lDtð5Þwhere F t is the cumulative fractional mass of the Earth at time t relativeto the present day, l is the time constant for core formation ( ¼ 1/t,where t ¼ the mean life) and Dt ¼ t 0 2 t (where t 0 is the age of theSolar System). On this basis, both Hf–W and U–Pb timescales shouldbe the same, but as can be seen from Fig. 6, the mean-life of coreformation is about 12 Myr using the Hf–W system and 28 Myr usingU–Pb. Changing the model (for example, to single-stage equilibration)alters both timescales, but the discrepancy between them remains.The U–Pb timescale is always more protracted than that of Hf–W.The only way in which both Hf–W and U–Pb timescales could beFigure 6 | Two estimates of the timing of accretion and core formation onthe Earth.
Both estimates assume an exponential growth model (equation(5) with progressively increasing masses of impactors). a, The result thatsatisfies the W isotopic composition of the bulk silicate Earth9. b, A resultthat satisfies recent estimates of the Pb isotopic composition of the bulksilicate Earth9. The Hf–W age of the Moon corresponding to the Moonforming giant impact (T GI) appears to lie in the range 40 to 50 Myr (ref. 61).As discussed in the text, the W result dates the principal phase of coreformation, while the Pb result plausibly dates a late-stage segregation ofsulphide after the giant impact.830correct is if Pb entered the core at a later stage than W, after theformation of the Moon9. The giant impact believed to have formedthe Moon46,60 defined the last major growth stage of the Earth, addingabout 10% of its mass.
The timing of the impact is given by the Hf–Wage of the Moon which appears to lie between 40 and 50 Myr(Fig. 6)61. Under the reducing conditions of early accretion (Fig. 4)W would be a siderophile element while Pb partitioning into themetal phase would be weak9,62. Conditions appear to have becomemore oxidizing during accretion, however, so that, as found in modernMORB63, an Fe–Ni sulphide became the only stable ‘metallic’ phase(Fig. 7). Figure 7 shows the effect when ferric iron produced fromperovskite crystallization is recycled back into a melt of peridotiticcomposition under upper mantle conditions.
![](https://s.studizba.com/z.php?f=/templates/si/i/noavatar.png&w=100&h=100&t=1"){kind=avatar}
