2006 Wood et al., Accretion of the Earth and segregation of its core (1119308), страница 3
Текст из файла (страница 3)
Theliquid metal ponded at the base of the magma ocean and subsequently descended in large diapirs to the growing core withoutfurther equilibration with the surrounding silicate (Fig. 3). Highpressure core formation can explain, by metal–silicate equilibrium,the partitioning of many elements between core and mantle withoutrecourse to ad hoc assumptions about the oxidation state of thegrowing Earth. The ‘late veneer’ of chondrite-like material is,however, still required to explain the concentrations of the highlysiderophile elements in the silicate Earth.The ‘deep magma ocean’ model of core formationExplanations for the ‘excess siderophile problem’As the Earth grew, the gravitational energy deposited by accretingplanetesimals increased and at about 10% of its current size45 wassufficient for significant melting to occur.
After this point theThe W-isotopic anomalies of iron and silicate meteorites show thatasteroidal bodies segregated metallic cores very rapidly, within about5 Myr (ref. 2) after the beginning of the Solar System. The Earth, onthe other hand, appears to have undergone more protracted coreformation, the W-isotopic composition of the mantle implyingsingle-stage separation of the core about 30 Myr (refs 2 and 19)after the beginning of the Solar System or, using an exponentialaccretion model33, a mean-life of accretion and core-separation of12 Myr. The implications of the W-isotopic data are that theasteroidal cores re-equilibrated with the silicate Earth duringaccretion.
If re-equilibration had not occurred, the silicate Earthwould have inherited the low-pressure core–mantle partitioning ofthe small bodies and also their large positive W-isotopic anomalies.Given that the ‘excess siderophile problem’ was not inherited, butarose from the manner of core formation on the Earth, severalpossible explanations have been proposed:(1) Inefficient core formation34,35. This is a disequilibrium modelin which incomplete core separation left some fraction of corematerial mixed with the mantle. Later re-oxidation and re-mixingof this metal led to the observed elevated mantle concentrations ofsiderophile elements. Although this model can fit many of theelemental abundances, it cannot explain those of the weak side-Figure 2 | The effect of pressure on Ni and Co partitioning. The liquid½imetal) for Ni and Co aremetal–liquid silicate partition coefficients (Di ¼ ½isilicatethose at 2,123–2,750 K and oxygen fugacity corresponding to current FeOcontent of the mantle.
Data are from ref. 4 and ref. 38. Note that valuesapproach those required for core–mantle equilibrium at the current FeOcontent of the mantle when pressure is ,28 GPa.© 2006 Nature Publishing Group827REVIEWSNATURE|Vol 441|15 June 2006Earth would have periodically had an extensively molten outer layer(a magma ocean) of variable thickness. In some cases the impactenergy would have been sufficient to melt both the impactor and theproto-Earth46,47.
The fates of pre-existing cores depended on the sizesof the Earth and impactor at the times of impact. Large impacts mayhave allowed the accreting core to combine directly with that of theproto-Earth. This cannot have been common, however, because ofthe evidence of metal–silicate re-equilibration on the Earth. Mostimpactors disaggregated and metallic iron sank though the moltensilicate layer in droplets which, because of Rayleigh–Taylor instabilities44, would have been about 1 cm in diameter.
Droplets of thissmall size would have re-equilibrated with silicate melt as they fellthrough depths of only 60 m (ref. 44). Liquid metal and silicatetherefore continued to re-equilibrate until the former either reachedthe core–mantle boundary (if the mantle were completely molten) orcollected at a level above a solid, high-viscosity lower layer (Fig.
3). Inthe latter case the metal layer would cease re-equilibrating after it hadreached about 5 km in thickness44 and because of the high viscosity ofthe lower layer would segregate as large diapirs to the core48.Numerical simulations49,50 indicate that a magma ocean extendingto the core–mantle boundary would be short-lived and that the lowermantle would crystallize in a few thousand years. A shallower,partially molten layer would crystallize much more slowly, however,and could remain as a mixture of crystals and melt for 100 Myr(refs 49, 50).
Considering these results and the energetics of impactand core segregation leads to a dynamic view of the growing Earth inwhich the outer molten part deepened and shallowed many timesafter episodic impact. The pressures and temperatures recorded bycore–mantle partitioning are therefore values averaged over numerous cycles of metal accumulation and segregation such as thatdepicted in Fig.
3.Inspection of Fig. 3 and consideration of pressures within aFigure 3 | The deep magma ocean model. Impacting planetesimalsdisaggregate and their metallic cores break up into small droplets in theliquid silicate owing to Rayleigh–Taylor instabilities. These droplets descendslowly, re-equilibrating with the silicate until they reach a region of highviscosity (solid), where they pond in a layer.
The growing dense metal layereventually becomes unstable and breaks into large blobs (diapirs), whichdescend rapidly to the core without further interaction with the silicate.Note that the liquidus temperature of the silicate mantle should correspondto pressure and temperature conditions at a depth above the lower solidlayer and plausibly within the metal layer as indicated.828growing planet lead to several conclusions about the conditions ofcore segregation.
First, accumulation at the base of a completely orpartially molten layer implies temperatures close to or slightly belowthe liquidus of mantle peridotite. Second, average pressure shouldhave increased as the planet grew. Third, because the liquidus has apositive pressure–temperature slope, the average temperature of coresegregation should also have increased as the planet grew. Havingaccepted these principles, we can use the dependences of D i onpressure and temperature6,39,40,51 to develop constraints on the coresegregation process.
As shown in Fig. 2, the metal–silicate partitioning behaviour of both Ni and Co is strongly dependent on pressure,with temperature effects being subordinate6,52 while partitioning ofV and Si, for example, depends predominantly on temperature6.Therefore, by simultaneous consideration of a number of elements ofdiffering chemical behaviour, it is possible to test potential pressure–temperature paths of the accretion and core-segregation process.Accretion and core-segregation processesFor the simplest end-member case of single-stage equilibration ofmantle and core, current partitioning data for Ni and Co yieldpressures close to 40 GPa while temperature sensitive elementssuch as V and Cr yield temperatures of about 3,750 K (ref. 6).These conditions result in a core with about 10.5 wt% Si as thedominant light element6 but, as illustrated in Fig.
4, correspond totemperatures about 650 K above the liquidus of the mantle. Suchtemperatures are physically implausible, however, because the base ofthe magma ocean must be close to the mantle liquidus (Fig. 3). If theFeO content of the mantle (and hence the oxygen fugacity) is fixed,then forcing core–mantle equilibration to remain on the peridotiteliquidus results in an overall vanadium (V) core–mantle partitioncoefficient of about 0.4, a factor of four lower than that required.Because V partitioning is insensitive to pressure6,51, this is purely atemperature effect—peridotite liquidus temperatures are too low topartition the requisite amount of V into the core.
V partitioning intothe metal can be enhanced by assuming that the silicate is a crystal–liquid mixture but the effect is small. Furthermore, crystal–liquidmixtures tend to partition too much Nb and W into the core. TheFigure 4 | Conditions yielding correct core–mantle partitioning ofsiderophile elements during accretion. Oxygen fugacity (relative to IW orFe–FeO) is plotted as a function of fraction accreted, for Earth accretionmodels that yield the expected D values of Table 1 for Fe, Ni, Co, V, W, Nb,Cr, Mn, Ga, Si and P. Accreting metal was assumed to equilibrate with thesilicate mantle at the base of the magma ocean, which deepens as the Earthgrows. Note that at fixed oxygen fugacity (or FeO content of the mantle)during accretion, a temperature about 650 K above the silicate liquidus isrequired.
Metal–silicate equilibration models fixed at the silicate liquidustemperature during accretion require pressures (from Ni and Copartitioning) corresponding to 30–40% of the depth to the core–mantleboundary in the growing planet and also that oxygen fugacity increases asthe planet grows.© 2006 Nature Publishing GroupREVIEWSNATURE|Vol 441|15 June 2006simplest solution6, as illustrated in Fig. 4, is to allow oxygen fugacity(as represented by the FeO content of the mantle) to increase asaccretion and core-segregation progress.Figure 4 illustrates a continuous process of accretion and coresegregation in 1% intervals.
![](https://s.studizba.com/z.php?f=/templates/si/i/noavatar.png&w=100&h=100&t=1"){kind=avatar}
