2006 Wood et al., Accretion of the Earth and segregation of its core (1119308), страница 2
Текст из файла (страница 2)
The mostimportant point is that, although the bulk Earth does not have anexactly CI composition28, refractory lithophile elements such as Ca,Sc, Ti and the rare earths are present in the silicate Earth in thesame relative proportions as in the carbonaceous chondrites3,26,29.Carbonaceous chondrites represent undifferentiated protoplanetarymaterial, so the implication is that the bulk Earth contains the samerelative proportions of all refractory elements as carbonaceouschondrites.
This is the basis of the chondritic reference model.As can be seen on the left-hand side of the diagram, there is adecreasing relative abundance of elements in the silicate Earth withdecreasing condensation temperature or increasing volatility. Thisdemonstrates that the Earth is depleted in volatile elements withrespect to the chondritic reference, although the correlation withincreasing volatility is purely qualitative. Depletions of the silicateEarth in refractory siderophile elements such as W, Mo, Re and Os aredue to partial extraction of these elements into the core.
The extent ofextraction can be estimated by comparing silicate Earth concentrations(ratioed to CI chondrites) to those of refractory lithophile elementssuch as the rare-earth elements. This enables us to calculate, for eachelement, a core–mantle partition coefficient D i defined as follows:½icoreDi ¼½isilicate Earthwhere [i] is the concentration of element i. Lithophile elements have Dvalues close to zero. Table 1 shows that siderophile elements exhibit avery wide range of core–mantle partitioning behaviour, reflecting theirdifferent chemical properties and the conditions under which coresegregation took place.Core–mantle partitioning is best defined for those refractoryelements that are weakly or moderately siderophile and which, likeNi and Co, are compatible in solid mantle silicates.
The concentrations of these elements vary little in mantle samples, which meansthat D i values based on the chondrite model have small uncertainties.Highly siderophile refractory elements such as the Pt group (Fig. 1)are slightly less well-constrained. The abundances of these elementsin mantle samples are very low and, since they enter minor sulphideTable 1 | Core–mantle partition coefficientsElementFigure 1 | Elemental abundance in the silicate Earth versus temperature of50% condensation. Elemental abundances in the silicate Earth are ratioedEarthto those in CI carbonaceous chondrites3 and normalized to ½Mg½MgCI ¼ 1:0.The abundances are plotted against the temperature by which 50% of theelement would have condensed from a gas of solar composition at a totalpressure of 1024 bar (ref.
27). We note that depletions of the silicate Earth inrefractory ‘Siderophile’ and ‘Highly siderophile’ elements relative to‘Refractory lithophile’ elements are due to sequestration in the core. Corecontents of volatile elements that condense at low temperatures are moredifficult to constrain. REE, rare-earth elements.826D FeD NiD CoDVDWD PdD IrD PtD NbD CrD MnD SiDSD GaDPRefs 3 and 29Ref. 26Likely rangeLow-pressure experimental D i(refs 30 and 31)13.6626.523.81.8316800800800NE3.40.290.2976NE2213.6524.424.7NENENENENENE2.950.34NENE4513.6523–2723–271.5–2.215–22600–1,000600–1,000600–1,0000.2–0.80.5–3.5*0.2–2.0*0.1–0.35*50–100*0–1.5*20–50*13.654,9006800.0237 £ 10510114 £ 106NE0.20.0061025501530NE, not estimated.*Value uncertain due to volatility.© 2006 Nature Publishing GroupREVIEWSNATURE|Vol 441|15 June 2006minerals, rather variable.
It is generally accepted, however, that theirratios to one another in the silicate Earth are, as shown in Fig. 1,approximately chondritic.Core concentrations of volatile siderophile elements such as Si, Sand P are difficult to estimate because their bulk Earth contents arenot well constrained. This means, for example that depletions in Siand Cr (Fig. 1) could be due to volatility or to sequestration in thecore or both. The lack of constraint means that the uncertainties in D iare large (Table 1).Segregation of the reduced core from the oxidized mantle tookplace at high temperatures as shown above, and element partitioningbetween the two depended on oxygen fugacity, as can be seen fromthe redox reaction:nð1ÞMOn=2 ¼ M þ O24This reaction represents reduction of the element as it transfers fromits normal oxidation state (n) in the silicate to oxidation state zero inthe metal.
Given a concentration of FeO in the mantle of about 8weight per cent (wt%) and of Fe in the core of 85 wt% (refs 3, 26, 29),reaction (1) implies that the core separated from the mantle at anoxygen fugacity approximately 2 log f O2 units below Fe–FeO (ironwüstite, IW) equilibrium.Table 1 shows a comparison of the calculated core (metal)–mantle(silicate) partition coefficients and those obtained experimentally atlow pressures (0–2 GPa), high temperatures (,1,800 K) and oxygenfugacity corresponding to 8 wt% FeO in the mantle30,31. As can beseen, the observed core–mantle partition coefficients are, for many ofthe siderophile elements, much (orders of magnitude) smaller thanthose determined experimentally.
This manifests itself in the concentrations of many elements of concern (such as Ni, Co, Pd) beingmuch greater in the mantle than would be predicted from theexperimental data32, the so-called ‘excess siderophile problem’. Ofequal importance are the weakly siderophile elements such as V andCr for which the mantle concentrations are lower than predictedfrom the experimental data. Clearly, the mantle concentrations ofthese two groups of elements cannot be explained by an equilibriumwith the metallic core at the fixed pressure, temperature and oxygenfugacity conditions of Table 1. We would expect, however, that earlycore segregation in small planetesimals and planetary embryos,including the proto-Earth, would have resulted in siderophileelement abundances generally consistent with the D values given inTable 1.
The Earth has apparently inherited little of the geochemistryof these earlier events.rophiles V and Cr30. A further objection is the enormous amount ofwater, about 15 times the current mass of the hydrosphere, whichwould be required to re-oxidize the ‘retained’ metal8.(2) Heterogeneous accretion8,36,37. This model proposes that theconditions of core segregation changed from reducing at the beginning of accretion to more oxidized towards the end. During theearliest, reduced, phase, highly and moderately siderophile elements(Ni, Co and the Pt group, for example) would have been completelyextracted to the core, together with some fraction of the weaksiderophiles such as Si, V and Cr. As conditions became moreoxidizing, core extraction of siderophile elements would have ceasedprogressively, beginning with the weakest siderophiles and endingwith more strongly metallic elements.
The final phase, after coreformation had ceased, was the addition of a ‘late veneer’ of about0.5% of chondritic material. The latter raised the concentrations ofhighly siderophile elements in the mantle and, because core formationhad ceased, ensured their approximate chondritic proportions to oneanother (Fig. 1). This model solves the ‘excess siderophile problem’,but, being a disequilibrium model with multiple steps, is verydifficult to test.(3) High-pressure core formation. Extension of metal–silicatepartitioning experiments to pressures above 3 GPa (refs 4, 5, 38–41)demonstrated that the partition coefficients of some siderophileelements (most notably Ni and Co; Fig. 2) change with increasingpressure such that their mantle abundances may be explained bymetal–silicate equilibrium at very high pressures and temperatures(see refs 4 and 42 for example).
These observations led to the ‘deepmagma ocean’ hypothesis to explain the ‘excess siderophile’ problem.As the Earth grew, it is argued (see ref. 43 for example), droplets ofmetallic liquid descended through a 700–1,200-km-deep (28–40 GPa)magma ocean, equilibrating with the silicate liquid as they fell44.
![](https://s.studizba.com/z.php?f=/templates/si/i/noavatar.png&w=100&h=100&t=1"){kind=avatar}
