Show Summary Details

Page of

Printed from Oxford Research Encyclopedias, Planetary Science. Under the terms of the licence agreement, an individual user may print out a single article for personal use (for details see Privacy Policy and Legal Notice).

date: 10 April 2021

Composition of Earthfree

  • H. PalmeH. PalmeSenckenberg Research Institute and Natural History Museum

Summary

Early models of the composition of the Earth relied heavily on meteorites. In all these models Earth had different layers, each layer corresponded to a different type of meteorite or meteorite component. Later, more realistic models based on analyses of samples from Earth began with Ringwood’s pyrolite composition in the 1960s. Further improvement came with the analyses of rare MgO rich peridotites from a variety of occurrences all over the Earth, as xenoliths enclosed in melts from the upper mantle or as ultramafic massifs, tectonically emplaced on the Earth’s surface. Chemical systematics of these rocks allow the determination of the major element composition of the primitive upper mantle (PUM), the upper mantle after core formation and before extraction of basalts ultimately leading to the formation of the crust. Trace element analyses of upper mantle rocks confirmed their primitive nature. Geochemical and geophysical evidence argue for a bulk Earth mantle of uniform composition, identical to the PUM, also designated as “bulk silicate Earth” (BSE). The formation of a metal core was accompanied by the removal of siderophile and chalcophile elements into the core. Detailed modeling suggests that core formation was an ongoing process parallel to the accretion of Earth. The composition of the core is model dependent and thus uncertain and makes reliable estimates for siderophile and chalcophile element concentrations of bulk Earth difficult.

Improved stable isotope analyses show isotopic similarities with noncarbonaceous chondrites (NCC), while the chemical composition of the mantle of the Earth indicates similarities with carbonaceous chondrites (CC). In detail, however, it can be shown that no single known meteorite group, nor any mixture of meteorite groups can match the chemical and isotopic composition of Earth. This conclusion is extremely important for any formation model of the Earth.

Of all regions of the Earth none invites speculation more than that which lies beneath our feet, and in none is speculation more dangerous (Oldham, 1906).

The assumption that elemental abundances in meteorites are similar to those in the Earth has provided a fruitful basis for the construction of Earth models (Ringwood, 1961).

Historical Perspective

Early Views and the 19th Century

The chemical composition of the Earth became a scientific issue only in the 19th and 20th century. Earlier scientists were not concerned with the composition of bulk Earth; they rather emphasized the physical state of the interior of the Earth. Because of the widespread occurrence of volcanism on the surface of Earth, they thought that the interior must be very hot. In addition, theories of the interior of the Earth had to account for earthquakes and magnetism.

A good example is the Earth model of Descartes (1596–1650). A fire-fluid, sunlike core is overlain by a spherical shell of metal. This solid shell is enclosed by two other shells, one consisting of water and another of air. Above this last shell there is the freely floating Earth crust, broken up into several pieces. Earthquakes as well as eruptions of volcanoes are caused by accidental sparks, which are inflamed by sulfuric vapor in crevices and caves of the Earth’s crust. Thereby mountains and seas formed (Oeser, 1992).

Athanasius Kircher (1602–1680), a Jesuit scholar and polymath, described in his book, Mundus Subterraneus, the interior of the Earth. He postulated a huge fire in the center of the Earth, feeding numerous smaller fire seats in the mantle. These fires were connected with the central fire and interconnected through a system of channels. Some of these channels reached the surface of the Earth, feeding volcanoes. The theory of Kircher was very popular, partly because of his superior drawings of the interior of the Earth.

Things changed at the end of the 18th and the beginning of the 19th century. In 1794 the German physicist E. F. F. Chladni (1756–1827) demonstrated that meteorites are of extraterrestrial origin, by showing that there is no other way to account for the falls of meteorites upon the surface of the Earth (Chladni, 1794). The first chemical analyses of meteorites were published by the English chemist E. Howard (1774–1816) in 1802, and shortly afterward by M. E. Klaproth (1743–1817), a professor of chemistry in Berlin. These early investigations led to the important conclusion that meteorites contained the same elements that were known from analyses of terrestrial rocks. By the year 1850, 18 elements had been identified in meteorites: C, O, Na, Mg, Al, Si, P, S, K, Ca, Ti, Cr, Mn, Fe, Co, Ni, Cu, and Sn (Burke, 1986). A popular hypothesis, which arose after the discovery of the first asteroid Ceres on January 1, 1801, by Piazzi (1746–1826), held that meteorites came from a single disrupted planet in the asteroid belt, between Mars and Jupiter.

In 1847 the French geologist A. Boisse (1810–1896) proposed a more sophisticated model, in an attempt to account for all known types of meteorites from a single meteorite planet. He envisioned a body with layers in sequence of decreasing densities from the center to the surface. The core of the planet consisted of metallic iron surrounded by a mixed iron-olivine zone. The region overlying the core contained material similar to stony meteorites with ferromagnesian silicates and disseminated grains of metal gradually extending into shallower layers with aluminous silicates and less iron. The uppermost layer consisted of metal-free stony meteorites, that is, eucrites or meteoritic basalts. About 20 years later G.­A. Daubrée (1814–1896) carried out experiments by melting and subsequent cooling of meteorites. On the basis of his results he came to similar conclusions as Boisse, namely that meteorites come from a single, differentiated planet with a metal-rich core, a silicate mantle, and a crust. Both Daubrée and Boisse also expected that Earth is composed of a similar sequence of concentric layers as the meteorite planet (see Burke, 1986; Marvin, 1996).

One of the problems in comparing meteorites with Earth was the dominance of olivine and the presence of metallic Ni-bearing iron in most meteorites and the absence of metal in most terrestrial rocks. Some authors, however, thought that the interior of the Earth, where basalts formed, could be of very different composition. The famous German chemist and mineralogist C. Rammelsberg (1813–1899) probably was the first to think in the right direction. In 1872 he wrote a brochure (Rammelsberg, 1872) with the title Über die Meteorite und ihre Beziehungen zur Erde (About meteorites and their relation to Earth). The paper ends with the following statement: “The praehistoric Eifel volcanoes have ejected rounded masses, so called ‘bombs,’ consisting of olivine, bronzite, augite and chromite, the same minerals that are repeatedly found in meteorites. The question therefore arises: are these rocks perhaps samples of the unaltered core of the Earth?” He talks about terrestrial rocks that we today use to derive elemental abundances representative of the silicate Earth. Rammelsberg also speculated that metallic iron, so characteristic of most meteorites may be present in the interior of the Earth.

Earth Models in the First Half of the 20th Century

At the beginning of the 20th century more analyses of meteorites and their major phases became available, elemental abundance patterns for meteorites and terrestrial rocks were established, and the presence of a metal core of Earth was supported with evidence provided by studies of seismic waves.

In 1901 Farrington (1901) made a detailed comparison of the mineralogy and chemistry of the Earth’s crust and meteorites, and he found differences between crustal rocks and meteorites. “Looked at quantitatively then it may be said that terrestrial rocks abound in free silica, lime, alumina, and alkalies, while meteorites abound in iron, nickel and magnesia. Whether these quantitative differences would be maintained if the constitution of the Earth as a whole could be compared with that of meteorites, is, as hinted at the beginning, doubtful.”

The next step was to include the iron core into Earth models. If the Earth is compositionally similar to the meteorite planet, it should have a high content of iron. The presence of a central metallic core to the Earth was firmly established by Wiechert (1861–1928) in 1897, primarily because of the high density of Earth. As pointed out before, earlier models also had included iron cores, but, as Brush (1980) emphasized, these models had a continuous decrease in density from core to crust, whereas in Wiechert’s model there was, for the first time, a clear boundary between core and mantle. Using the study of seismic wave propagation Oldham (1858–1936) confirmed the existence of the core in 1906. In 1914 Beno Gutenberg (1889–1960), Wiechert’s student, located the boundary of the core accurately at a depth of 2,900 km (Gutenberg, 1914). In 1926 the fluidity of the outer core was finally accepted, and in 1937 Inge Lehmann (1888–1993) detected the small solid inner core (see Brush, 1980, for details).

Similar to Farrington (1901), Harkins (1873–1951) at the University of Chicago thought that meteorites would provide a better estimate for the bulk composition of the Earth than the terrestrial rocks collected at the surface as we have only access to the “mere skin” of the Earth (Harkins, 1917). Harkins made an attempt to reconstruct the composition of the hypothetical meteorite planet by compiling compositional data for 125 stony and 318 iron meteorites, and mixing the two components in ratios based on the observed falls of stones and irons. For his meteorite planet Harkins calculated Mg/Si, Al/Si, and Fe/Si atomic ratios of 0.86, 0.079, and 0.83, very closely resembling corresponding ratios of the average solar system based on presently known element abundances in the Sun and in CI-meteorites (see Burke, 1986).

In 1925, H. S. Washington (1867–1934) published a paper with the title “The Chemical Composition of the Earth” (Washington, 1925). In the center of the Earth was an iron core with a radius of about 2,900 km compositionally similar to iron meteorites. Overlying the core was a metal-silicate layer with some 700 km, resembling pallasites in composition and with decreasing depth, gradually becoming enriched in silicates and depleted in metal. Washington assumed that this layer has approximately the composition of chondrites. The uppermost part of the Earth has a basaltic or gabbroic composition that grades into a layer with granitic and granodioritic composition.

Ida and Walter Noddack (Noddack & Noddack, 1930), the discoverer of the element Re, also analyzed many meteorites in order to obtain the composition of the meteorite planet. They determined the composition of the silicate part of the meteorite planet by removing metal from chondritic meteorites and analyzing the rest as silicate. To this they added the results of analyses of metal and troilite (FeS), which they had separated from iron meteorites. As in other reconstructions of the meteorite planet the problem was to determine the fraction of silicate to metal and sulfide in the bulk planet. Harkins had simply used the frequency of stony and iron meteorites. The Noddacks used the average density of the inner planets to define the metal content of the meteorite planet.

In 1922 Victor Moritz Goldschmidt introduced his zoned Earth model (Goldschmidt, 1922). In a more comprehensive paper in 1930 he published details (Goldschmidt, 1930). Goldschmidt assumed, as other modelers did, that the Earth was initially completely molten and on cooling separated into three immiscible liquids. The iron metal liquid formed the core (3,500–6,700 km), which was overlain by a sulfide-oxide shell (1,700–3,500 km) covered by an eclogite shell (1,100–1,700 km), followed by an outer shell of uncompressed silicates, and on top of the uppermost crust was the biosphere and finally the gaseous atmosphere, which was the result of outgassing of the cooling Earth. During fractional crystallization of a liquid, precursor elements partitioned into the various layers according to their geochemical character. Goldschmidt distinguished four groups of elements: siderophile elements preferring the metal phase, chalcophile elements preferentially partitioning into sulfide, lithophile elements remaining in the silicate shell, and atmophile elements concentrated in the atmosphere. The geochemical character of each element was determined from a variety of sources, behavior in experiments, theoretical considerations and abundance in the corresponding phases of meteorites. Goldschmidt does not explain how he arrived at the thicknesses of the various layers. He was more concerned with the partitioning of trace elements into the various layers than the bulk Earth composition. Goldschmidt attributed the high concentrations of U, Th, and light REE in crustal rocks to their incompatibility with common rock forming minerals, whereas the compatible elements Mg and Cr are concentrated in silicates of the deeper Earth.

Goldschmidt commented on similar elemental abundance patterns of heavy elements in the silicate Earth, average meteorites, and the Sun, as well as some neighboring stars, which were spectroscopically analyzed by Payne (1925).

In a review article, the astronomer N. H. Russell (1941) concluded: “The average composition of meteorites differs from that of the Earth’s crust significantly, but not very greatly. Iron and magnesium are more abundant and nickel and sulfur rise from subordinate positions to places in the list of the first ten. Silicon, aluminum, and the alkali metals, especially potassium, lose what the others gain.” And Russell continued: “The composition of the Earth as a whole is probably much more similar to the meteorites than that of its ‘crust.’” Russell concludes this paragraph with a statement on the composition of the core: “The known properties of the central core are entirely consistent with the assumption that it is composed of molten iron—though not enough to prove it. The generally accepted belief that it is composed of nickel–iron is based on the ubiquitous appearance of this alloy in metallic meteorites,” and, it should be added, also on the abundances of iron and nickel in the sun. This basic picture outlined by Russell is still valid.

A summary of models proposed in the first half of the 20th century is depicted in Figures1 and 2 taken from Buddington (1943). All models have the core beginning at a depth of 2,900 km, except for the Wiechert (1897) and Suess (1909) models. Layers above the core are very different, primarily in the amount of sulfide, oxide, and metal. Various silicates are proposed for upper mantle and crust. High pressure phases were unknown at his time.

Figure 1. Various geophysical (upper part) and corresponding chemical and mineralogical (lower part) structures of Earth (from Buddington, 1943).

Figure 2. Additional geophysical (upper part) and corresponding chemical and mineralogical (lower part) structures of Earth (from Buddington, 1943).

Summary of Early Models

All these early models have a few features in common:

1.

They all have an FeNi core of a size accurately determined by Gutenberg (1914), except for two earlier models.

2.

There is a more or less extensive layer of mixed metal-silicates (pallasites, mesosiderites) or sulfides and oxides between core and crust.

3.

The source of all meteorites was a single, disrupted planet, which was similarly structured to Earth but smaller than Earth.

4.

Elemental concentrations in the various layers of Earth were assumed analog to the distribution of elements in different phases of meteorites.

5.

The total Fe content of Earth was not evaluated. Prior’s early findings (Prior, 1916) that the amount of Fe in bulk meteorites is often constant, only the ratio of oxidized to metallic Fe is variable, were largely ignored.

6.

There were no attempts to calculate the bulk composition of the Earth, by mass balance. People ignored the possibility of Earth to compositionally match chondritic meteorites. Chondrites were rather one component of the meteorite planet.

Progress in the Second Half of the 20th Century

Major progress in instrumental and analytical techniques, as well as in theoretical concepts resulted in a better understanding of the origin and composition of Earth and the relationship of Earth to meteorites. The major points will be mentioned here and, in some cases, more detailed discussions are given in the next sections.

Many more chemical analyses of chondritic meteorites were performed, and their more or less uniform composition and the approximate agreement with solar abundances was taken as evidence for a common origin of solid matter in the early solar system, including Earth. In 1961 Ringwood (1961) wrote: “Apart from certain volatile components and trace elements, chondrites are extremely uniform in chemical composition. This average composition provides the basis for a satisfactory geochemical model of the Earth, and, with the exception of iron, is in reasonable agreement with the solar elemental abundances.”

Ringwood (1961) proposed to make ordinary chondrites from carbonaceous chondrites by reduction, and he emphasized the work of Prior (1920), who noticed a more or less constant Fe-content of meteorites.

MacDonald (1959) pointed out that the average heat flux through the surface of the Earth corresponds, within a factor of two, to the heat produced by the decay of long-lived radioactive elements of a planet with chondritic composition (i.e., chondritic abundances of K, U, Th), confirming the chemical similarity of Earth and meteorites. However, Wasserburg, MacDonald, Hoyle, and Fowler (1964) noted a considerably lower K/U ratio of terrestrial rocks compared to meteorites, because of lower K or higher U or both. A lower K/U ratio would produce a different heat flux through the surface of the Earth and could thus confirm a non-chondritic composition of the Earth. The problem, however, is the unknown fraction of heat resulting from the initial accretion and core formation. The ratio of radioactive heat compared to the total heat produced within the Earth, the Urey ratio, is estimated at about 0.38 (see Arevalo, McDonough, & Luong, 2009), for a recent discussion), making it difficult to precisely constrain the K/U ratio of Earth from heat flux through Earth’s surface.

The lower than chondritic K/U ratios of Earth rocks were confirmed by new data on K Rb, Cs, U, Ba, and Sr, as well as Sr-isotopes, in a variety of terrestrial rocks obtained by Gast (1960) with the newly developed isotope dilution mass spectrometry. This author concluded that the upper mantle is depleted in the relatively volatile alkali-elements compared to the refractory Ba, Sr, and U. Gast (1960) concluded that this is either a typical signature of bulk Earth or that the lower mantle accounts for a complementary signature, that is, high alkali/refractory element ratios. Arguments based on the ratio of incompatible volatile to non-volatile elements played and still play an important role in estimating the composition of bulk planets, including Earth (see, e.g., Arevalo et al., 2009).

The idea of a single meteorite planet was finally given up. Improvement in the study of asteroids and the realization of a huge diversity of asteroids made a common origin of all these objects improbable. The final evidence was provided by oxygen isotopes (see Clayton, 1993, for a summary). In a δ17O versus δ18O diagram, samples from a single planet or planetesimal lie approximately on a slope ½ line. Samples from Earth and Moon plot along such a line. Most meteorites do not plot on the Earth-Moon line, but on lines parallel to the Earth-Moon line. Meteorites are different in oxygen isotopes from Earth and also among each other. Hence, they cannot come from a single meteorite planet, nor can the Earth be made of a single type of meteorite, with the exception of enstatite chondrites. This case will be discussed in later sections in connection with other stable isotopes.

In summary, although there is certainly not a single planet delivering all extraterrestrial materials, the major element composition of most undifferentiated extraterrestrial materials and reconstructed bulk planet compositions are roughly similar to the nonvolatile element composition of the solar photosphere (e.g., Palme & Zipfel, 2017). The major variables are the extent of oxidation, reflected in the ratio of oxidized to metallic iron, as originally envisioned by Prior (1920), and the contents of volatile and refractory elements.

Hot Origin of Earth

A hot origin of Earth was suggested by Eucken (1944). Eucken assumed that Earth formed on cooling of an initially hot gas sphere of solar composition. Eucken’s conclusions were based on calculations by Emden (1907) on fractional condensation of an initially hot gas sphere (see Fegley, Lodders, & Jacobson, 2020, for details). On the basis of Eucken’s model Turekian and Clarke (1969) proposed an inhomogeneous accretion model for the Earth. Iron was assumed to have a higher condensation temperature than Mg-silicates. This is indeed the case at a total pressure above 7.1*10−5 atm (Grossman, 1972). Iron would therefore condense in an Emden sphere before olivine and pyroxene and form the core of Earth upon which Mg-silicates would condense. This is the essence of the inhomogeneous accretion model. Turekian and Clarke (1969) wrote:

Thus the order of condensation coincides grossly with the stratification observed in the Earth and inferred in other planets. Such stratification is usually attributed to the settling of the densest material towards the planetary center. But high density turns out to be associated with low volatility, enabling planets to accrete in a way that is automatically stable gravitationally. The iron body that is now the Earth’s core formed by accumulation of the condensed iron-nickel.

In this model, core and silicate mantle are not in thermodynamic equilibrium, and there is no need to reduce metallic iron with H2 or carbon. There is, however, overwhelming evidence for the opposite, thermodynamic equilibrium between core and mantle. Metallic elements were extracted from the mantle and transported into the core, as will be discussed. This does not necessarily exclude the inhomogeneous accretion model, as core mantle equilibration could have occurred after the end of Earth formation. Lewis (1972) envisioned a global picture of planet formation, with condensation from a gas of solar composition and decreasing equilibration temperatures from the sun outward. While Mercury, closest to the Sun, was predicted to have an FeNi-core, Venus had, besides metal, some oxidized Fe but no FeS; at the heliodistance of Earth, metallic FeNi and FeS formed a metal-sulfide core and left some FeO for the mantle; and the most distant terrestrial planet, Mars, had an FeS dominated core. In these early models Earth is made directly from the cooling solar nebula, whereas newer models emphasize the collisional growth of Earth from small aggregates to large planetesimals, involving significantly radial mixing.

Cameron (1962) made a first attempt to model the protosolar nebula (a mixture of gas and solids) from which meteorites and planets formed. In his calculations, temperatures were high enough to vaporize the solids in the gaseous accretion disk. Cameron’s models were refined (Cameron & Pine, 1973) and led to the viscous accretion models, where the infalling interstellar material redistributes mass and angular momentum through turbulent viscosity (see Wood & Morfill, 1988). From 1960s to the early 1990s, cosmochemists interpreted their data in terms of a hot solar nebula where preexisting solids evaporated and condensed on cooling to components found in meteorites.

In the 1990s and later high global temperatures in the solar nebula were in disfavor. Widespread isotope anomalies in meteorites and their components, the presence of presolar grains, the presence of local enrichments of 22Ne, a decay product of 22Na with a half-life of 2 years, and so forth cast doubt on a globally hot solar nebula (e.g., McSween & Huss, 2010).

It now appears that fractionation of volatile elements requires high temperatures for most solid matter analyzed in the solar system. Moderately and highly volatile elements are partially missing in all solid solar system materials (Palme, Larimer, & Lipschutz, 1988). Loss of volatiles is ubiquitous and not a local process. There is no material with enrichment in volatiles indicating mere redistribution of volatile elements. Volatiles must have been lost from the inner solar system. High temperatures of 1,800 to 2,000 K are required for fractionation of refractory elements and formation of chondrules. This could be the result of local heating processes or formation near the Sun. In this case, the high temperature material must have been transported to larger heliocentric distances. None of the models of the solar nebula produces such high temperatures at 1 AU and beyond (e.g., Cassen, 2001).

Low Temperature Accretion

Early on, Urey (1952) proposed low temperatures during formation of planets and meteorite parent bodies. In his 1952 paper he concluded: “It is difficult to understand how any condensation process from high temperature gases could possibly lead to this result” (accumulation of planetesimals).

Modern theories of planet formation start with the solar nebula, a mixture of gas and dust, surrounding the growing Sun. Planets are then made in three steps: (a) dust settles into the mid-plane and accretes to increasingly larger objects ending with km-sized planetesimals; (b) this stage is followed by the collisional accretion of planetesimals to produce planetary embryos, moon to mars sized objects; finally (c), the interaction between embryos becomes the dominant factor as they perturb each other onto crossing orbits, thus producing accretion via violent collisions (e.g., Bond, Lauretta, & O’Brien, 2010). After their formation, the terrestrial planets began to heat up, melt, and form metal cores by the settling of heavy metallic FeNi-alloys, whereas low-density partial melts from the interior of the planets rose to the surface and formed planetary crusts, processes that are collectively called planetary differentiation. The heat required for melting and differentiation was provided by accretion (collisions of large bodies); in addition, the decay of short-lived radioactive nuclei, such as 26Al (half-life 71,000 years) and 60Fe (half-life 2.6 million years), may have contributed, depending on the timescale of accretion. These kinds of model do not require high temperatures, but they also do not exclude high temperatures during early nebular processes that lead to losses of volatiles before or during formation of solid grains (e.g., Humayun & Cassen, 2000). To what extent losses of volatiles during later reheating contributed to the overall volatile element depletions is presently unclear. The lack of isotopic fractionations during evaporation of, for example K, speaks against this possibility (Humayun & Cassen, 2000). In addition, the approximately chondritic ratio of Na/Mn in chondritic meteorites and in the Earth argues against evaporation (O’Neill & Palme, 1998, 2008). It is much easier to lose Na on heating than Mn (Gellissen, Holzheid, Kegler, & Palme, 2019). This would produce solids with high Mn/Na ratios, such as is the case for the Moon and the eucrite parent body, but not for the Earth.

Another possibility is the inheritance of volatile depletion from the interstellar medium (Palme, Lodders, & Jones, 2014; Yin, 2005). In the interstellar medium (ISM), about 1% of the total matter is in grains—the rest is in the gas phase. Refractory elements, Mg and Si, are quantitatively in solids; moderately volatile elements are partly in solids; and highly volatile elements are entirely in the gas phase. If the solar nebula has inherited this structure, no further evaporation processes are required to explain the volatile element depletions.

Improvements in Seismic Wave Studies

Improved techniques for the registration of seismic waves allow better definition of seismic discontinuities. Together with high-pressure mineral physics experiments it was found that mantle discontinuities are associated with solid–solid phase transitions and not with compositional changes. The major upper mantle mineral olivine undergoes a change to a spinel structure termed wadsleyite at 410 km depth. Wadsleyite itself undergoes a less dramatic transition to ringwoodite at around 510 km and, then, at 660 km depth, ringwoodite changes to a combination of perovskite and bridgmanite (e.g., Bercovici, 2015).

Studies of convection in the presence of phase changes show that they may impede convection temporarily but not permanently. With the help of seismic tomography, it is possible to observe slabs penetrating the 660 km discontinuity into the lower mantle, which is generally taken as indicating whole mantle convection (van der Hilst, Widiyantoro, & Engdahl, 1997).

Whether the bulk mantle has on average a uniform composition, or whether chemical differences between upper and lower mantle can persist for billions of years is not finally resolved. Geochemical arguments for a well-mixed mantle are very strong. They will be discussed in another section.

Stable Isotope Analyses

New developments in analytical instrumentation have led to revolutionary discoveries in cosmochemistry, in particular in isotope geo- and cosmochemistry. One of the most promising new instruments is the multi collector inductively coupled mass spectrometer (MC-ICPMS), which significantly improved and still improves the sensitivity and accuracy of the determination of radiogenic and stable isotope ratios (Zinner, Moynier, & Stroud, 2011). This has led to (a) an increase in spatial resolution of the analysis of isotope ratios, allowing for the study of increasingly smaller samples, and (b) an increase in the precision of isotopic analysis that allows more precise dating, improves the study of isotopic heterogeneity and provides clues for the nuclear origin of characteristic isotope ratios. The new data, together with oxygen isotopes, mentioned earlier, have established that meteorites come from a variety of parent bodies. Furthermore, it is now clear that chondritic meteorites are not isotopically equilibrated. They contain elements with different isotopic composition in different phases, clearly excluding any melting.

Ringwood’s New Approach

Perhaps the most important change in estimating the bulk composition of Earth came with the pioneering work of Ringwood. For the first time, he tried to estimate the bulk silicate Earth composition from the analysis of terrestrial rocks. Although earlier workers noted that the composition of the crust is very different from most meteorites, they had only a vague idea about the composition of underlying Mg- and Fe-rich layers. Ringwood invented the term “pyrolite” to characterize a complementary relationship of basalt and residual peridotitic mantle. Basalts, partial melts from the mantle, are enriched in Si and Al, ultimately forming the crust. The upper part of the mantle, the source region of basalts, has to be accordingly depleted in basaltic elements. The deeper mantle represents the undisturbed mantle. This concept was very successful and is the basis for all further estimates of upper mantle composition. To reiterate: Ringwood believed, as others, that the composition of the Earth is in a broad sense chondritic, but he attempted to estimate the composition of the mantle of the Earth from actual chemical analyses of rocks found on the surface of the Earth (Ringwood, 1962a).

The Chemical Composition of Chondritic Meteorites and the Cosmochemical Classification of Elements.

Undifferentiated Meteorites

The thermal history of smaller planetesimals, the parent asteroids of meteorites, depends on the timescale of accretion and the amount of incorporated short-lived radioactive nuclei, primarily 26Al with a half-life of 7.1 × 105 years. In some cases, there was sufficient heat to completely melt and differentiate the planetesimals into a metal core and a silicate mantle, while in other cases even the signatures of a modest temperature increase are absent. Undifferentiated meteorites derived from unmelted parent bodies are called chondritic meteorites. Their chemical composition derives from the solar composition more or less altered by processes in the early solar nebula, but essentially unaffected by subsequent planetary processes, that is, melting and crystallization. Their textures, however, may have been modified by thermal metamorphism or aqueous alteration since their formation from the solar nebula. Chemically, chondritic meteorites are characterized by approximately similar numbers of atoms of silicon, magnesium, and iron, indicating the absence of the two fundamental planetary differentiation processes, metal separation, which leads to low Fe/Si in residual silicates and partial melting, which produces low Mg/Si melts and high Mg/Si residues. The roughly solar composition is, however, only a first-order observation. Nebular processes, primarily fractionation during condensation and/or aggregation, have produced a range of variations in the chemistry of chondritic meteorites. The extent to which individual elements are affected by these processes depends mainly on their volatility under nebular conditions.

Differentiated Meteorites

The powerful radioactive heat sources (26Al, 60Fe) in the early solar system have led in some cases to melting and differentiation of smaller planetesimals. The most prominent examples are iron meteorites that formed early as segregated metal cores in differentiated planetesimals and were dated by the Hf-W chronometer (Krujier et al., 2014). In later formed planetesimals heat sources were largely decayed and there was not enough energy for melting and they were only heated to temperatures of up to 1,100°C, leading to various degrees of thermal metamorphism in the solid planetesimals. This led to the current paradox that undifferentiated, chondritic meteorites are younger than iron meteorites, products of melting of differentiated planetesimals. Some iron meteorites are nearly as old as refractory (high temperature) inclusions of some chondritic meteorites, the oldest objects of the solar system. This indicates that planetesimal melting and differentiation must have occurred very early, when the concentrations of short-lived radioisotopes were high. Corresponding silicate phases, as old as iron meteorites, have not yet been identified.

Figure 3. Element/Si mass ratios of characteristic lithophile elements (except Fe) in the major groups of chondritic (undifferentiated) meteorites, the solar photosphere, and the Earth’s mantle. Meteorite groups are arranged according to decreasing oxygen content. The best match between solar abundances and meteoritic abundances is with CI-meteorites. Meteorite data: Wolf and Palme (2001), Wasson and Kallemeyn (1988); solar photosphere: Palme et al. (2014); Earth’s mantle: Palme and O’Neill (2014).

Figure is from Palme and O’Neill (2014).

Chemistry of Major Chondrite Groups

The major chondrite classes shown in Figure 3 are carbonaceous chondrites with CI, CM, CO, and CV subgroups, ordinary chondrites with H, L, and LL subgroups, and enstatite chondrites with EH and EL subgroups. These are the most populated chondrite classes. The large number of meteorites recovered from Antarctica and from hot deserts have led to a significant increase in the number of new groups of chondritic meteorites (not shown in Figure 3). For a detailed discussion the reader is referred to Krot, Keil, Goodrich, Scott, and Weisberg (2003).

Ultimately, the goal is to compare the composition of the Earth with the composition of chondritic meteorites, in order to establish (or discard) a possible genetic relationship of bulk Earth with one or more groups of chondritic meteorites. But, in contrast to chondrites, Earth is a differentiated planet with a metal/sulfide core of largely unknown composition. Therefore, study begins with comparing the silicate fraction of Earth with the silicate fraction of chondritic meteorites and siderophile and chalcophile elements are neglected. If agreement of silicate Earth with silicates of a certain group of chondrites or a mixture of chondrites is found, the composition of Earth’s core could be calculated by using the composition of the metallic fractions of these meteorites.

A detailed comparison of lithophile elements in Earth and meteorites requires some knowledge of the cosmochemical and geochemical properties of lithophile elements.

Condensation Temperatures and Cosmochemical Components

The element ratios plotted in Figure 3 are representative of different cosmochemical components. Each of these components has a characteristic formation temperature indicated by its condensation temperature. Condensation temperatures are calculated by assuming thermodynamic equilibrium between solids and a cooling gas of solar composition (see Lodders, 2003, for details). The cosmochemical components are attributed to distinct regimes of condensation temperatures between 1825 and >0 Kelvin (K) at 10−4 bar (10 Pa). Accordingly, with decreasing condensation temperature, elements representative of the various cosmochemical components are assigned to six groups, which account for the bulk of chondritic meteorites:

(1)

Refractory component: The first phases to condense from a cooling gas of solar composition are Ca, Al-oxides and silicates rich in trace elements, such as the REE (rare Earth elements), Zr, Hf, and Sc. These elements are named refractory lithophile elements (RLE); they condense before the condensation of Mg-silicates. The refractory siderophile elements (RSE) comprise metals with low vapor pressures, for example, W, Os, Ir, that condense before condensation of the major fractions of Fe and Ni in multicomponent metal alloys (Palme & Wlotzka, 1976). The refractory component makes up about 5% of the total condensable matter. Variations in Al/Si ratios (Figure 3) are produced by addition or loss of a refractory component. The CV carbonaceous chondrites have the highest and enstatite (EL and EH) chondrites the lowest Al/Si ratios. The difference between the highest and lowest ratio is about 50%. A very similar pattern as for aluminum would be obtained using other refractory elements (Ca, Ti, Sc, REE, etc.), because the ratios among refractory elements in meteorites are constant in all classes of chondritic meteorites, at least within about 5–15% (see, e.g., Pack, Russell, Shelley, & van Zuilen, 2007).

(2)

Mg-silicates: The major fraction of condensable matter is associated with the three most abundant elements heavier than oxygen: Si, Mg, and Fe. Magnesium and Si condense as forsterite (Mg2SiO4), which converts to enstatite (MgSiO3) at lower temperatures by reaction with gaseous SiO. Variations in Mg/Si ratios of bulk meteorites are produced by the incorporation of variable amounts of early condensed forsterite. The total range of Mg/Si ratios is about 25%. Again, the CI ratio fits best with the solar ratio (Figure 3).

(3)

Metallic iron condenses as FeNi alloy at about the same temperature as forsterite (Grossman, 1972). The sequence depends on total pressure. It is important to realize that refractory metals such as Os, Ir, Pt, and so forth condense independently and ahead of the main FeNi-metal phase. Variations in the concentrations of Fe and other siderophile elements in meteorites are produced by the incorporation of variable amounts of metal. Variations in Fe are larger than those of Al and Mg. In Figure 3 CI chondrites and EH chondrites have about solar Fe/Si ratios. All other chondrites have lower Fe/Si ratios. There are new rare meteorite groups with excess Fe, such as the CB and CH chondrites (Fe/Mg in CH = 1.7 x CI, CB = 7.8–9.9 x CI) and a new group of Fe-rich chondrites with affinities to E-chondrites (Weisberg, Ebel, Nakashima, Kita, & Humayun, 2015).

(4)

Moderately volatile elements have condensation temperatures between those of Mg-silicates and FeS (troilite). The most abundant moderately volatile element is sulfur, which starts to condense by reaction of gaseous H2S with solid Fe metal at 710 K, independent of total pressure. Half of all sulfur is condensed at 664 K. Moderately volatile elements are distributed among sulfides, silicates, and metal. The amount and the relative abundances of these elements in meteorites are probably the result of removal of volatiles during condensation and/or incomplete condensation (Palme et al., 1988). The elements Na, Zn, and S (Figure 3) belong to the group of moderately volatile elements. Their concentrations in chondritic meteorites are significantly more variable than those of Al and Mg. Here the agreement in composition between CI chondrites and the Sun is particularly noteworthy and confirms the unique position of CI-chondrites. It is also important to note that all other groups of chondritic meteorites have, with very few exceptions, lower volatile element contents than CI chondrites when normalized to Si or Mg. Basically, there are no meteorites with excess volatiles (see Palme et al., 1988).

(5)

Highly volatile elements have condensation temperatures below those of FeS and above water ice. The group of highly volatile elements comprises elements with very different geochemical affinities, such as the chalcophiles Pb and In and the lithophiles Cs and I. The abundances of the highly volatile elements vary with the petrologic type of the chondrite. The higher the metamorphic grade the lower the trace element abundances, presumably because of losses caused by metamorphic heating. The contents of the moderately volatile elements S, Zn, and Te vary little among the different metamorphic types of ordinary chondrites, in contrast to the highly volatile elements Cs, Br, Cd, Bi, Tl, and In, whose abundances vary by orders of magnitudes between type 3 and type 6 metamorphic grade. The variations are primarily in the higher petrologic types, while their abundances in type 3 ordinary chondrites are similar to those of the moderately volatile elements (Keays, Ganapathy, & Anders, 1971; Schaefer & Fegley, 2010). Internal heating of small parent bodies may lead to loss of highly volatile elements by evaporation. There are no highly volatile elements shown in Figure 3.

(6)

Ultra volatile elements have condensation temperatures below that of water ice. This group includes H, C, N, O, and the noble gases. The O/Si ratios decrease monotonously in the sequence of chondritic meteorites shown in Figure 3. This can, however, not be interpreted as a depletion trend, as oxygen and all other elements of this group are not fully condensed even in CI chondrites (Table 1).

In summary, Figure 3 gives some indication for the range of elemental variations in chondritic meteorites. Despite these variations, the sequence of decreasing element abundances is the same in all chondritic meteorites and the Sun with the most abundant elements Si, Mg, and Fe, followed by Al and Ca. Planetary melting processes such as core formation or partial melting lead to rocks with completely different compositions. We assume that minor variations among chondritic meteorites are produced by nebular processes that occurred before accretion and formation of their parent bodies. But it means that the bulk Earth is of solar composition modified by early solar nebula processes. It appears that the same processes that produced the variations in chondritic meteorites are responsible for variations in the compositions of the terrestrial planets.

In Figure 3 the composition of the Earth’s mantle, as derived by Palme and O’Neill (2014), has been added. Refractory elements such as Al are similarly enhanced in Earth as in CV-meteorites and moderately volatile elements are similarly depleted. The Mg/Si ratio is also not very different from the carbonaceous chondrite Mg/Si ratios. Iron and S are low in the BSE (bulk silicate Earth) because they reside predominantly in the core. Two other moderately volatile elements, Zn and Na, are depleted in Earth and meteorites, reflecting the general volatile element depletion in most of solar system rocky materials. For the low S content in the mantle of the Earth, both core formation and the general depletion of volatiles are responsible. Dreibus and Palme (1996) estimated the S content of the core of the Earth by assuming that S and Zn are similarly depleted in bulk Earth, as suggested by similar condensation temperatures.

Some researchers assumed that the components that make up the inner planets are the same as the constituents of chondritic meteorites. Variations in the compositions of planets or planetesimals are then produced by variable fractions of cosmochemical components in planets or planetesimals. This is the basis for Anders’s seven component models of Earth and Mars (Ganapathy & Anders, 1974).

Earth

The principal division of the Earth into core, mantle, and crust is the result of two fundamental processes: (a) The formation of a metal core very early in the history of the Earth. If the core formed in a single event, core formation occurred at around 34 ± 3 million years after the beginning of the solar system, based on Hf-W dating (Kleine & Walker, 2017). (b) The formation of the continental crust by partial melting of the silicate mantle. This process has occurred with variable intensity throughout the history of the Earth.

According to the chondritic Earth model, the bulk silicate Earth should have a Mg/Si wt. ratio of around 0.90 (see Figure 3). Most rocks on the surface of the Earth have much lower ratios. The crust of the Earth has a Mg/Si ratio of about 0.05, mid-ocean-ridge basalts have about 0.2. There are, however, uncommon MgO-rich rocks on the surface of the Earth with Mg/Si ratios between 1 and 1.3, close to the ratio in chondrites. They occur in a variety of geologic settings and range from lherzolites to harzburgites and dunites with a modal content of clinopyroxenes from below 1% to slightly above 20%. Ringwood (1962a) suggested that these are rocks from the mantle of the Earth, emplaced on the surface by tectonic processes or as xenoliths enclosed in rapidly ascending basanitic magmas. Many of these rocks are very low in Al and Ca and have high Mg/Si ratios. They are apparently lacking a basaltic component. Ringwood (1962a, 1962b) reconstructed the primitive upper mantle (the mantle before melt extraction) by mixing 4 parts of peridotites (depleted upper mantle) and one part of basalt (a pyroxene-olivine rock), and he named this hypothetical rock type pyrolite, representing the mantle of the Earth before extraction of a partial melt. The pyrolite composition was very influential, and many studies in experimental petrology simulating the chemical composition of the upper mantle were performed with this composition (Green & Falloon, 1998).

Rocks From the Mantle of Earth

The term “peridotite” comprises a variety of rocks collected on the surface of the Earth. Their olivine content is above 40% and they have variable amounts of orthopyroxene (opx) and clinopyroxene (cpx) and small amounts of an Al-bearing phase, such as spinel or garnet. Peridotites with the lowest MgO contents have, in general, the highest concentrations of Al2O3, CaO, TiO2, and other incompatible elements that preferentially partition into the liquid phase during partial melting. With a maximum of about 20% cpx, peridotites are often termed “fertile,” emphasizing their ability to produce basalts on melting. Most peridotites have below 20% cpx; those with less than 5% cpx are called harzburgites. The Mg/Si ratios of peridotites are between 1.0 (high cpx) and 1.4 (low cpx). Dunites, which may form as cumulates, have Mg/Si ratios significantly higher than 1.0 (Su, Chen, Guo, & Liu, 2016). It is remarkable that we have rocks on the surface of the Earth with Mg/Si ratios close to the Mg/Si ratio of the average solar system ratio (CI-ratio) with 0.89. From no other planet do we have rocks with MgO contents as high as terrestrial peridotites. The close to chondritic Mg/Si ratios of peridotites, along with other chemical characteristics, qualifies these rocks as mantle rocks, and they are therefore used to derive information about the chemical composition of the mantle of the Earth. The “primitive” major element chemistry is a prerequisite for an undifferentiated rock to represent a piece of the mantle of the Earth. Additional chemical and petrological evidence, such as chondritic ratios of refractory lithophile elements, unfractionated patterns of highly siderophile elements, and the ability of the mantle to produce basalts on partial melting will be discussed later (Green & Falloon, 1998; Palme & O’Neill, 2014). Physical properties of these rocks, such as density and seismic wave speeds, are required for mantle rocks to satisfy geophysical constraints (O’Neill & Palme, 1998).

Several abbreviations are used to describe the mantle of the Earth. Bulk silicate Earth (BSE) is the most general expression. It includes the chemical composition of the total silicate Earth including the crust, but without the core. A frequently used term is PUM, which stands for primitive upper mantle, and is restricted to the upper mantle. Sometimes PM is used for primitive mantle. This is the composition of the bulk fertile mantle, the mantle composition before crust formation, but after core formation, equivalent to BSE. Since it is argued here that the primitive upper mantle is compositionally identical to the bulk mantle, BSE, PUM, and PM are equivalent.

Occurrences of Upper Mantle Rocks

Rocks and rock fragments from the Earth’s mantle occur in a variety of geologic settings (e.g., see O’Neill & Palme, 1998): (a) as mantle sections in ophiolites representing suboceanic lithosphere; (b) as massive peridotites, variously known as Alpine, orogenic, or simply high-temperature peridotites; (c) as abyssal peridotites, dredged from the ocean floor—the residue from melt extraction of the oceanic crust; (d) as spinel- (rarely garnet-) peridotite xenoliths from alkali basalts, mostly from the subcontinental lithosphere; and (e) as garnet peridotite xenoliths from kimberlites and lamproites—these fragments sample deeper levels in the subcontinental lithosphere and are restricted to ancient cratonic regions.

Peridotites have physical properties such as density and seismic velocity propagation characteristics that match the geophysical constraints required of mantle material. Another obvious reason for believing that these rocks come from the mantle is that the constituent minerals have chemical compositions, which show that the rocks have equilibrated at upper mantle pressures and temperatures. In the case of the xenoliths their ascent to the surface of the Earth was so fast that minerals had no time to adjust to the lower pressure and temperature on the Earth’s surface. Most of the information used for estimating the chemical composition of the mantle is derived from spinel-lherzolites originating from a depth of 40 to 60 km. Garnet lherzolites, which sample the mantle down to a depth of ~200 km and a temperature of 1,400°C, are much rarer. It is clear that claims that rocks from such shallow depth chemically represent the whole mantle of the Earth down to core-mantle boundary at 2,900 km depth are not self-evident and require justification.

Major Element Composition of Upper Mantle Rocks

Maaløe and Aoki (1977) compiled data on the chemical composition of continental (302 analyses) and oceanic (82 analyses) lherzolites from all over the world. They noted good correlations of MgO with Al2O3, CaO, and so forth and negative correlations with NiO. The average of all analyzed lherzolites was taken as average for the upper mantle composition. The results gave 41 MgO and 2.46% Al2O3 (Table 1).

In 1979, Jagoutz et al. (1979) reported the results of analyses of six spinel lherzolites with high Al2O3 and CaO contents and with some 20% clinopyroxene, the major host phase of CaO. Jagoutz et al. (1979) argued that this should be a better estimate for the bulk mantle composition, since they may represent the mantle of the Earth before partial melting.

Figure 4. Plot of Mg/Si vs Al/Si (weight ratios), similar to Figure 2 in Jagoutz et al. (1979). Peridotite samples: Siberia Xenoliths from Ionov and Hofmann (2007); CDOB (central Dinaric ophiolite belt, Yugoslavia) from Lugovic et al. (1991); Ronda peridotite from Frey et al. (1985); fertile xenoliths from Jagoutz et al. (1979; sample Ka168 from this paper is not plotted, because of excessive Al); Sun1 is the newest photospheric abundanc (Scott et al. 2015); Sun2 represents earlier photospheric abundances listed in Scott et al. (2015). P&N is the PUM value obtained by Palme and O’Neill (2014); other PUM compositions are listed in Table 1; CI, CM, CO, CV data are from Wolf and Palme (2001); H, l, LL data are from Jarosewich (1990); E chondrites are from Jarosewich (1990), Mason (1966), and von Michaelis et al. (1969).

The estimated mantle composition is accordingly lower in MgO (38.3%) and higher in Al2O3 (4.02%) than the estimate of Maaloe and Aoki (1977).

Figure 4 shows a plot of Al/Si versus Mg/Si of upper mantle rocks and chondritic meteorites, similar to Figure 2 in Jagoutz et al. (1979). The following points should be noted:

1.

The upper mantle array in Figure 4 is defined by peridotites from very different geologic settings, as xenoliths from Siberia (Ionov & Hofmann, 2007), as ophiolites from the central dinaric ophiolite belt (CDOB) in the former Yougoslavia (Lugovic, Altherr, Raczek, Hofmann, & Majer, 1991), and as a peridotite massif from Ronda (Spain), uplifted from the upper mantle to the surface of the Earth (Frey, Suen, & Stockman, 1985). All samples in Figure 4 show the same Mg-Al correlation. This and other correlations, such as Ca, Ti, V, Sc versus Mg follow worldwide trends, independent of the geologic setting and the mineralogy of samples. Samples of massive peridotites and of spinel and garnet lherzolite xenoliths from worldwide localities plot on the same Mg/Si versus Al/Si correlation (Figure 4), as well as on similar plots of Al, Ca, Ti, Na, and so forth versus Mg, not shown here (Basaltic Volcanism Study Project [BVSP], 1981; McDonough, 1990; McDonough & Sun, 1995; O’Neill & Palme, 1998; Palme & O’Neill, 2014). It is particularly noteworthy that trends for xenoliths and massive peridotites are statistically indistinguishable (McDonough & Sun, 1995). This is impressively shown with hundreds of samples by Pearson, Canil, and Shirey (2003) and Bodinier and Godard (2003).

2.

Jagoutz et al. (1979) reported the results of analyses of six more or less fertile spinel lherzolites. The term “fertile” means that these compositions are able to produce basalts on melting. The fertile xenoliths plot in the lower (Al-rich) part of the Mg/Si versus Al/Si correlation. One sample of the Jagoutz et al. (1979) suite, Ka168, is not plotted because of excess Al, introduced by a high abundance of spinel, that is, non-representative sampling.

3.

The Sun1 point with error bars in Figure 4 represents the newest estimate of photospheric abundances by Scott et al. (2015). The new data point is further away from CI-chondrites than earlier estimates, given in Scott et al. (2015) and plotted as Sun2 in Figure 4.

4.

The red points indicate primitive upper mantle estimates from various authors. The P&ON composition is from Palme and O’Neill (2014). This composition is used as reference for further calculations.

5.

The positive trend in Al/Si versus Mg/Si ratios of chondritic meteorites is not really understood. Dauphas, Poitrasson, Burkhardt, Kobayashi, and Kurosawa (2015) suggested that early separation of condensed forsterite, reflected in 30Si isotopes, was responsible for a global Mg/Si fractionation of planetesimals and planets including the Earth. The considerable spread in Al/Si ratios of carbonaceous chondrites may be ascribed to variable additions of Ca, Al-rich inclusions (CAI).

6.

The position of Earth’s mantle with regard to meteorites is uncertain. Assuming that the bulk silicate Earth (BSE) plots along the Mg/Si versus Al/Si correlation it plots significantly above the CI-point, which is believed to represent the bulk solar system. It would require an Al/Si ratio above that of CV-meteorites to match the solar system Mg/Si ratio. The BSE compositions indicated in Figure 4 are all above the CI-composition, suggesting that the BSE has a higher Mg/Si ratio than the average solar system. There are several explanations for the low Si-content of the upper mantle of the Earth.

The Composition of the Mantle

If peridotites are relied on for the closest match to the mantle of the Earth, the mantle should be expected to plot on the Mg-Al correlation of Figure 4, a kind of universal plot for Earth mantle rocks. Peridotites with the lowest MgO contents have the highest concentrations of Al2O3, CaO, and other refractory compatible elements. Most peridotites are, however, depleted in incompatible elements to various degrees, that is, they have lower contents of CaO, Al2O3, and so forth than a fertile mantle would have. As mentioned, Maaløe and Aoki (1977) used an average of all peridotites, from depleted to fertile, for representing the bulk silicate Earth. Ringwood (1962a, 1962b) had recombined depleted mantle (low in Al, Ca, Ti) with basalts to estimate the composition of the primitive mantle. Jagoutz et al. (1979) used an average of rather fertile xenoliths as upper mantle composition. Palme and Nickel (1985) based their estimate on chondritic ratios of refractory elements. The Sc/Yb ratio, for example, is only chondritic for MgO contents of 35 to 39%. At higher MgO contents, the Sc/Yb ratio is higher, because Yb is more incompatible than Sc and is thus more effectively extracted from the mantle. Including Al and Ti, Palme and Nickel (1985) calculated 35.5% MgO and 4.75% Al2O3 for the primitive upper mantle. The modal content of this composition is 43.45% olivine, 32.1% opx, 23.1% cpx, and 1.3% spinel. Palme and Nickel (1985) also noted a 15% enhanced Ca/Al-ratio compared to the CI-ratio in spinel lherzolite suites worldwide. A higher than CI-chondritic Ca/Al ratio was also observed by Pearson et al. (2003) in their compilation of data of nearly 600 xenoliths (see their Figure 9). On the other hand, there seem to be suites of xenoliths with chondritic Ca/Al ratios, as for example a suite of 46 xenoliths from Siberia, with a chondritic Ca/Al ratio of 1.09 ± 0.18 (Ionov & Hofmann, 2007). The issue is not yet clear.

Hart and Zindler (1986) based their estimate for PUM on 33 fertile lherzolites and chondritic Nd/Ca ratios. These authors obtained 37.8% MgO and 4.06% Al2O3 for the primitive upper mantle (PUM).

In 1995 McDonough and Sun (1995) published a widely cited paper on the abundances of major, minor, and trace elements in the silicate Earth. Similar to Palme and Nickel (1985) they used variations of refractory elements with MgO in peridotites to fix the mantle composition at chondritic ratios. Their MgO content of PUM was 37.8% with a corresponding Al2O3 value of 4.06%.

In the same year Allègre, Poirier, Humler, and Hofmann (1995) published a PUM composition based on similar datasets and using the “least differentiated” sample. Their composition was very similar to the McDonough and Sun (1995) composition.

O’Neill and Palme (1998) used mass balance and molar 100*Mg/(Mg+Fe), that is, Mg# ratios to define a PUM composition. The molar ratio, Mg#, reaches a minimum value of 89 for the most primitive peridotites. This is significant because FeO stays essentially constant during partial melting and Mg# numbers are the same in melt and residue. From Mg# and mass balance equations O’Neill and Palme (1998) calculated a PUM composition with 36.64% MgO and 4.73% Al2O3. A slightly modified, but essentially similar procedure was used by Palme and O’Neill (2014) with nearly the same results. The various estimates of the major element compositions of the Earth’s mantle are summarized in Table 1. The table also contains a mantle composition by Lyubetskaya and Korenaga (2007), which is, compared to the other compositions listed, fairly MgO-rich. This composition is not an extrapolation to the fertile mantle, but it is based on the average peridotite composition and therefore more MgO-rich than other compositions, similar to the Maaløe and Aoki (1977) composition (not listed in Table 1).

Table 1. Major Element Composition of the Primitive Mantle by Palme and O’Neill (2014) and Comparison With Estimates From the Literature

Palme and O’Neill (2014)

Ringwood (1979)

Jagoutz et al. (1979)

Wänke et al. (1984)

MgO

36.77 ± 0.44

38.1

38.3

36.9

Al2O3

4.49 ± 0.37

3.3

3.97

4.19

SiO2

45.40 ± 0.30

45.1

45.1

45.6

CaO

3.65 ± 0.31

3.1

3.50

3.54

FeOt

8.10 ± 0.05

8.0

7.82

7.58

Total

98.41 ± 0.10

97.6

98.7

97.8

(RLE/Mg)N

1.22 ± 0.10

0.94

1.09

1.16

mg#

0.890 ± 0.001

0.895

0.897

0.897

Palme and Nickel (1985)

McDonough and Sun (1995)

Allègre et al. (1995)

Lyub. and Kor. (2007)

MgO

35.5

37.8

37.77

39.50

Al2O3

4.75

4.45

4.09

3.52

SiO2

46.2

45.0

46.12

44.95

CaO

4.36

3.55

3.23

2.79

FeOt

7.40

8.05

7.49

7.97

Total

98.2

98.9

98.7

98.7

(RLE/Mg)N

1.43

1.17

1.07

0.88

mg#

0.891

0.893

0.900

0.898

mg#—molar Mg/Mg+Fe; FeOt—all Fe as FeO. (RLE/Mg)N—Refractory lithophile elements normalized to Mg and CI-chondrites. Lyub. and Kor. (2007)—Lyubetskaya and Korenaga (2007).

Minor and Trace Element Contents of the Mantle of the Earth

Once the major element composition of the mantle is fixed minor and trace elements can be calculated from correlations with major elements and in some cases from element ratios. The composition of the primitive Earth mantle is given in Table 2 and plotted in Figure 5. Table 2 also contains element abundances in CI-chondrites, which were taken from Palme et al. (2014).

The abundances of refractory, lithophile elements, such as Ti, Sc, heavy REE, Hf, and Zr, can be calculated from correlations with MgO. The ratios among compatible refractory elements are essentially chondritic. (e.g., Jochum, McDonough, Palme, & Spettel, 1989). Peridotites are often contaminated with incompatible refractory lithophile elements (light REE, Ba, Ta, etc.) as well as non-refractory incompatible elements (K, Rb, Cs, etc.). These elements are introduced by cryptic metasomatism, a kind of metasomatism where primarily incompatible elements are introduced with a fluid phase. The nature of the fluid providing the incompatible elements is unknown, as no major element enhancements are noticeable. The incompatible refractory lithophile elements in PUM were assumed to occur in chondritic ratios to the compatible refractory elements. The two refractory lithophile elements V and Nb show some affinity to metal at reducing conditions in metal-silicate partition experiments. Their abundances are lower than those of other refractory lithophile elements. Details are given in Palme and O’Neill (2014).

Abundances of non-refractory incompatible lithophile elements (K, Rb, Cs, etc.) or partly siderophile/chalcophile incompatible elements (W, Sb, Sn, etc.) are calculated from correlations with RLE of similar compatibility. Early on, Gast (1960) noticed the depletion of alkalis relative to the much less volatile U, Sr, and Ba in a large number of terrestrial rocks. Wasserburg et al. (1964) concluded that the excellent correlation of K and U in terrestrial rocks defines the bulk Earth K/U ratio, which is a factor of eight lower than the CI-chondrite ratio. This approach was used by Wänke et al. (1973) and others to estimate the abundances of several incompatible volatile elements such as K, Rb, or Cs in the Moon. In addition, it was found that some incompatible siderophile elements correlate with incompatible lithophile elements (e.g., W and La), which allowed the estimation of the contribution of the siderophile element to core and mantle, assuming chondritic ratios in bulk Earth of, for example, W and La, a reasonable assumption as both elements are refractory (Rammensee & Wänke, 1977). This powerful method has been subsequently applied to Earth, Mars, and Vesta. The procedure is, however, only applicable if the two correlated elements have the same degree of incompatibility, that is, the two elements do not fractionate from each other during igneous processes, such as K/La or K/U. However, care must be taken to ensure that all important reservoirs are included in the dataset (see O’Neill & Palme, 1998, for details).

Another example is the Sm versus Sn correlation of Jochum, Hofmann, and Seufert (1993). Both basalts, partial melts from the mantle and mantle samples, representing the depleted mantle after removal of melt, plot on the same correlation line. From the Sm/Sn ratio of 0.32 obtained by Jochum et al. (1993), a mantle abundance of 144 ppb Sn is calculated, reflecting a 35-fold depletion of Sn relative to CI chondrites in Earth’s mantle. The origin of this depletion is twofold: (a) a general depletion of moderately volatile elements in Earth and (b) some partitioning of the siderophile element Sn into the core of Earth. Many of the abundances of moderately volatile lithophile, siderophile, and chalcophile elements listed in Table 2 were calculated from such correlations with RLE of similar compatibility. Well-known examples are K/U, K/La, W/Th, P/Nd, Rb/Ba, and so forth. Some trace elements (As, Cd, In) were analyzed in individual minerals of peridotites and their bulk content was obtained by adding up their mineral fractions with their corresponding element contents (Witt-Eickschen, Palme, O’Neill, & Allen, 2009). For further details the reader is referred to O’Neill and Palme (1998) and Palme and O’Neill (2014).

The refractory siderophile elements Re, Os, W, Ir, Mo, Ru, Pt, and Rh behave in chondritic meteorites similar to refractory lithophile elements. They should be enriched in the bulk Earth. But, except for W Mo, the refractory siderophile elements are all highly siderophile elements (HSE) with metal silicate partition coefficients above 10,000. They quantitatively partitioned into the core. The present inventory of these elements in the mantle of the Earth was added to Earth at a late stage of accretion, after core formation had ceased (Chou, 1978). This late accretional component is also known as late veneer. The abundance pattern of the HSE, including Pd, as a moderately volatile element, is not strictly chondritic. There are small anomalies in Ru, Rh, and Pd, indicated in Figure 5 (Becker et al., 2006; Schmidt, Palme, Kratz, & Kurat, 2000). New data on sulfide-silicate partition coefficients suggest that Ru and Pd being the least chalcophile elements were less effectively transported to the core than Ir and Pt. Some residual Ru and Pd remained in the mantle, that is, they were not completely extracted into the core, and produced the presently observed anomalies (Laurenz, Rubie, Frost, & Vogel, 2016).

The two refractory siderophile elements W and Mo have lower metal/silicate partition coefficients than the HSE. They are only partly extracted into the core (Figure 5). The behavior of W is similar to that of Fe; Mo is somewhat more siderophile (Newsom & Palme, 1984).

Moderately volatile lithophile elements define a depletion sequence, indicated in Figure 5. The condensation temperatures used in Figure 5 are from Wood, Smythe, and Harrison (2019). The largest difference from the Lodders (2003) condensation temperatures is the condensation temperature of Cl, which is, according to Wood et al. (2019), 462°K at 10−4 bar (10 Pa), instead of 948°K listed by Lodders (2003). The depletion pattern of volatile elements in PUM with strong depletions from Mg to Zn, followed by a plateau, indicated here by In is, according to new measurements, similar to that found in carbonaceous chondrites (Braukmüller, Wombacher, Funk, & Münker, 2019).

Table 2. Composition of CI-Meteorites and the Primitive Mantle of Earth, According to Palme and O’Neill (2014)

Z

E

CI

s.d

PUM

s.d.

Z

E

CI

s.d

PUM

s.d

1

H

%

1.97

10

0.012

20

56

Ba

ppm

2.42

5

6.85

15

3

Li

ppm

1.45

13

1.6

20

57

La

ppm

0.2414

3

0.6832

10

4

Be

ppm

0.0219

7

0.062

10

58

Ce

ppm

0.6194

3

1.7529

10

5

B

ppm

0.775

10

0.26

40

59

Pr

ppm

0.09390

3

0.2657

15

6

C

%

3.48

10

0.010

u

60

Nd

ppm

0.4737

3

1.341

10

7

N

%

0.295

15

0.0002

u

62

Sm

ppm

0.1536

3

0.4347

10

8

O

%

45.90

10

44.33

2

63

Eu

ppm

0.05883

3

0.1665

10

9

F

ppm

58.2

16

25

40

64

Gd

ppm

0.2069

3

0.5855

5

11

Na

ppm

4962

9

2590

5

65

Tb

ppm

0.03797

3

0.1075

15

12

Mg

%

9.54

43

22.17

1

66

Dy

ppm

0.2558

3

0.7239

10

13

Al

%

0.840

6

2.38

8

67

Ho

ppm

0.05644

3

0.1597

15

14

Si

%

10.70

3

21.22

1

68

Er

ppm

0.1655

3

0.4684

10

15

P

ppm

985

8

87

15

69

Tm

ppm

0.02609

3

0.07383

15

16

S

%

5.35

5

200

40

70

Yb

ppm

0.1687

3

0.4774

10

17

Cl

ppm

698

15

30

40

71

Lu

ppm

0.02503

3

0.07083

15

19

K

ppm

546

9

260

15

72

Hf

ppm

0.1065

3

0.3014

10

20

Ca

%

0.911

6

2.61

8

73

Ta

ppm

0.015

10

0.043

5

21

Sc

ppm

5.81

6

16.4

10

74

W

ppm

0.096

10

0.012

30

22

Ti

ppm

447

7

1265

10

75

Re

ppm

0.0400

5

0.00035

20

23

V

ppm

54.6

6

86

5

76

Os

ppm

0.495

5

0.0039

15

24

Cr

ppm

2623

5

2520

10

77

Ir

ppm

0.469

5

0.0035

10

25

Mn

ppm

1916

6

1050

10

78

Pt

ppm

0.925

5

0.0076

20

26

Fe

%

18.66

4

6.30

1

79

Au

ppm

0.148

12

0.0017

30

27

Co

ppm

513

4

102

5

80

Hg

ppm

0.35

50

0.006

u

28

Ni

ppm

1.091

7

1860

5

81

Tl

ppm

0.140

11

0.0041

25

29

Cu

ppm

133

14

20

50

82

Pb

ppm

2.62

8

0.185

10

30

Zn

ppm

309

4

53.5

5

83

Bi

ppm

0.110

9

0.003

u

31

Ga

ppm

9.62

5

4.4

5

90

Th

ppm

0.0300

7

0.0849

15

32

Ge

ppm

32.6

9

1.2

20

92

U

ppm

0.00810

7

0.0229

15

33

As

ppm

1.74

9

0.068

30

34

Se

ppm

20.3

7

0.076

u

35

Br

ppm

3.26

15

0.075

50

37

Rb

ppm

2.32

8

0.605

10

38

Sr

ppm

7.79

7

22.0

5

39

Y

ppm

1.46

5

4.13

10

40

Zr

ppm

3.63

5

10.3

10

41

Nb

Ppm

0.283

10

0.595

20

42

Mo

ppm

0.961

10

0.047

40

44

Ru

ppm

0.690

5

0.0071

20

45

Rh

ppm

0.132

5

0.0012

20

46

Pd

ppm

0.560

4

0.0071

20

47

Ag

ppm

0.201

9

0.006

50

48

Cd

ppm

0.674

7

0.035

20

49

In

ppm

0.0778

5

0.018

20

50

Sn

ppm

1.63

15

0.14

30

51

Sb

ppm

0.145

14

0.0054

40

52

Te

ppm

2.28

7

0.009

u

53

I

ppm

0.53

20

0.007

u

55

Cs

ppm

0.188

6

0.018

50

Data from Palme and O’Neill (2014); CI-data from Palme et al. (2014); s.d.—standard deviation; E—element.

Figure 5. Abundance pattern of elements in the silicate Earth. Elements are normalized to CI-chondrites and Mg and arranged in order of decreasing condensation temperatures, taken from Wood et al. (2019). For most refractory elements between Hf and Ba (e.g., Al, Ti, Ca, and REE), element symbols are not given. Earth depletion trend for moderately volatile lithophile elements is indicated. The high abundance of In, an element with siderophile and chalcophile tendencies, is remarkable. The Figure is modified from Palme and O’Neill (2014).

Geological Processes Affecting Mantle Rocks

Aside from correlations of refractory lithophile elements with MgO there is also a more or less universal correlation of MgO with Ni: the more MgO the more Ni. The concentrations of FeO (FeO includes all Fe-bearing species) in various peridotites are independent of MgO. In addition, the FeO contents of mantle rocks are with 8.1% FeO constant at all locations. The silicon content of peridotites only slightly decreases with increasing MgO content (BVSP, 1981; McDonough, 1990; McDonough & Sun, 1995; O’Neill & Palme, 1998).

All these features are consistent with extraction of a partial melt from a fertile mantle. In detail, however, the picture is not so simple. All mantle peridotites (whether massive peridotites or xenoliths) are metamorphic rocks that may have had a complex subsolidus history after melt extraction ceased. As well as subsolidus recrystallization, massive peridotites may also have undergone enormous amounts of strain during their emplacement in the lithosphere. They sometimes show modal heterogeneity on the scale of centimeters to meters, caused by mobilization of melts of primarily chromium diopside composition and thus producing inhomogeneities. Such rocks would not plot on the correlations mentioned (see Palme & O’Neill, 2014). The Ca/Al-ratio would not be constant, as Ca is almost entirely in cpx, whereas a large fraction of Al is in opx and spinel. Similarly, the amount of olivine determines the Ni content. Variable ol/opx and opx/cpx ratios would disturb Al-Ni and Ca/Ni anti-correlations. It is thus important to avoid apparently processed samples, and a fairly large rock sample should be homogenized to obtain a representative bulk analysis.

In addition, as mentioned, peridotites may have suffered metasomatism, which may re-enrich the rock in incompatible components subsequent to depletion by melt extraction. Where this is obvious (e.g., in reaction zones adjacent to later dikes) it may be avoided easily by appropriate sampling, but often the metasomatism is cryptic, as mentioned, in that it has enriched the peridotite in incompatible trace elements without significantly affecting major-element chemistry (Frey & Green, 1974). Thus, cryptic metasomatism cannot be avoided and its signature is often seen in harzburgites with high light rare earth elements (LREE) and K enrichments.

Significance of Al-Mg and Ni-Mg Correlation

Despite these problems it appears that the basic process producing the MgO-Al2O3 and other correlations is the same for most peridotites, partial melting of a fertile mantle with partial or complete removal of melt and subsequent re-equilibration (see discussion in Frey et al., 1985). An example is shown in Figure 6.

Figure 6. Chemical compositions of a suite of peridotites from the Tariat region (Mongolia) using data of Ionov and Hofmann (2007). The x-axis indicates MgO contents of samples, whereas on the y-axis elements are normalized to the abundances in PUM (Table 2). This procedure allows the simultaneously plotting of Al and Ni in a single plot. The MgO PUM concentration by Palme and O’Neill (2014) is indicated with a vertical line. The correlations of MgO with Al and Ni intersect at this MgO content.

The results of the chemical analysis of 45 xenoliths from central Asia (Ionov & Hofmann, 2007) are shown in Figure 6. Most xenoliths were larger than 10 cm in diameter and in most cases several hundred grams of the rocks were homogenized to obtain representative samples. Major elements were determined with carefully calibrated XRF-procedures. Part of the results are plotted in Figure 4. In order to show the correlations of Ni with Mg and with Al in a single plot, we have normalized the Al and Ni data to the mantle composition of Palme and O’Neill (2014) as given in Table 2. The correlations of Ni, Al, and also SiO2 with MgO are excellent. Least square fit calculations show that the Ni-Mg and the Al-Mg correlation intersect at an MgO content of 36.32% and at 1.0007 above the zero point of the ordinate. This is very close to the MgO content of 36.77% of Palme and O’Neill (indicated in Figure 4) and nearly at the zero points for Al and Ni.

There is one sample with anomalous high Al and low MgO (sample with lowest MgO in Figure 6), which does not fit well into this picture. Ionov and Hofmann (2007) suggest that this sample is anomalous and contains some fraction of a pyroxene vein. The regular pattern obtained with all other samples suggests a uniform process, such as partial melting producing these correlations, as argued in some detail by Ionov and Hofmann (2007). Melt extracted from a mantle with 36.88% MgO would simultaneously remove Al and increase Ni and MgO in the residual mantle. The gains of Ni and MgO are determined by the mass loss of the Al and Si rich melt. It is important to emphasize that the Al and Ni correlations only converge to a single point on the MgO line if the Palme and O’Neill (2014) normalization for Al and Ni is used. If another normalization is used, for example the Lyubetskaya and Korenaga (2007) mantle composition, the Al, Ni, and MgO lines would not meet in a single point.

Figure 7. The same type of Al and Ni correlations with MgO as in Figure 6. Samples are from the Rond peridotite massif in Spain. The slopes of the Al and Ni correlations with MgO are the same as in Figure 6. The two correlations intersect the MgO axis at about the PUM MgO content of Palme and O’Neill (2014).

It appears that the Al and Ni correlations in Figure 6 are typical of all occurrences of mantle rocks. As an example, the data for the Ronda peridotite massif in Spain by Frey et al. (1985) are shown in Figure 7. The crossover point at the MgO-axis is at 36.11%. The slopes of the Ni and Al correlations are within error the same as in the Siberian xenoliths in Figure 6. This is a worldwide trend, which implies a uniform process for producing these patterns. It should be noted that the data on Ronda include plagioclase, spinel, and garnet lherzolites. According to Frey et al. (1985) the pattern of the Ronda peridotite in Figure 7 was produced by varying degrees of partial melting of a primitive mantle with, in some cases, incomplete removal of melt, and subsequent re-equilibration was involved. There is, however, no evidence from chemistry for the so-called refertilization. Starting with a harzburgite, proponents of refertilization believe that a harzburgitic mantle became more fertile by reacting with melts from deeper layers and further geologic processes. Le Roux et al. (2007) pointed out:

Combined with previously published indications of refertilization in orogenic peridotites, our new observations in Lherz suggest that most lherzolite massifs represent secondary (refertilized) rather than pristine mantle. Together with geochemical data on mantle xenoliths, this indicates that melt transport and melt-rock reaction play a key role on the rejuvenation and erosion of the lithospheric mantle.

The chemical data for peridotites from Lherz presented by Le Roux et al. (2007) follow the same trends as those in Figures 6 and 7. There is therefore no doubt that a single process is responsible for the observed correlations. The observed Al- and Ni-MgO correlations cannot be the result of refertilization of undefined entities of molten material. It is difficult to envision that xenoliths (Figure 6) and samples from peridotite massifs (Figure 7) should follow exactly the same trend. In summary, there are good arguments that the PUM composition, derived by Palme and O’Neill (2014) and others, is representative of the upper mantle and probably of the whole mantle of Earth.

Is the Upper Mantle Composition Representative of the Bulk Earth Mantle?

It has been proposed that there is a substantial difference in major element chemistry, particularly in Mg/Si and Mg/Fe ratios, between the upper mantle above the 660 km seismic discontinuity and the lower mantle below this discontinuity. From Table 2 and Figure 4 it is clear that the upper mantle of the Earth has a higher than chondritic Mg/Si ratio and that Al and other refractory elements are enriched relative to Mg and Si. If the bulk mantle is CI-chondritic (i.e., solar) then the lower mantle should have higher Si/Mg and lower Al/Mg ratios to compensate the chemistry of the lower mantle.

Chemical layering may have occurred either as a direct result of inhomogeneous accretion without subsequent mixing (e.g., Turekian & Clark, 1969) or as the result of layering after crystallization of a magma ocean and without later mixing. Models with separate convection in upper and lower mantle (Hofmann, 1997) are increasingly unlikely. Seismic tomography indicates penetration of upper mantle slabs into the lower mantle, and this requires corresponding transport of lower mantle material into the upper mantle (Agrusta, van Hunen, & Goes, 2018; Fukao & Obayashi, 2013; van der Hilst, Widiyantoro, & Engdahl, 1997). As mentioned, direct geophysical evidence for a Si-enriched lower mantle is unlikely (Ballmer, Houser, Hernlund, Wentzcovitch, & Hirose, 2017; Hyung, Huang, Petaev, & Jacobsen, 2016).

There is, however, evidence from geochemistry that a chemically stratified mantle is impossible. The key point is chondritic ratios of compatible refractory elements in fertile mantle peridotites. Ratios among heavy REE, Ca, Al, Sc, and so forth approach chondritic values with increasing Al/Mg and decreasing Ni/Mg ratios until MgO contents of 36–38% MgO are reached (Hart & Zindler, 1986; McDonough & Sun, 1995; Palme & Nickel, 1985). Chondritic ratios of compatible refractory elements are not expected for a differentiated mantle, when Si-rich and Si-poor phases crystallize from a global magma ocean. Kato, Ringwood, and Irifune (1988) and Corgne, Liebske, Wood, Rubie, and Frost (2005) have shown from experimental studies that fractional crystallization of high-pressure phases from a hypothetical magma ocean would disturb many of the ratios of RLE (refractory lithophile elements) away from chondritic values.

A difference in 142Nd/144Nd ratios between average ordinary and carbonaceous chondrites and all terrestrial rocks was suggested as evidence for an early differentiation event and long-term isolation of an incompatible element–rich reservoir at the base of the mantle (Boyet & Carlson, 2005). Burkhardt et al. (2016), however, showed that the 142Nd offset between the accessible silicate Earth and chondrites reflects a higher proportion of s-process neodymium in the Earth and not an early differentiation process. Thus, there is no need for a hidden-reservoir or super-chondritic Earth models, consistent with a bulk Earth chondritic Sm/Nd ratio. The ubiquitous excess of 142Nd in all accessible samples from the silicate Earth rather supports a uniform composition of the upper and lower mantle (Bouvier, Vervoort, & Patchett, 2008).

A compositionally uniform mantle is not consistent with the model of Javoy et al. (2010). Because the stable isotope composition of some elements, such as O, N, and Cr are identical in Earth and E-chondrites, Javoy et al. (2010) concluded that Earth is made of enstatite chondrites (EC). There are several elements with differences in isotopic composition between Earth and E-chondrites: Si (Fitoussi & Bourdon, 2012), Ti (Zhang, Dauphas, Davis, Leya, & Fedkin, 2012), 92Mo, and 142Nd (Render, Fischer-Gödde, Burkhardt, & Kleine, 2017). In addition the difference between the chemical composition of the PM and EC is very large (see Figure 3). Javoy et al. (2010) had to postulate higher Si/Mg and lower Al/Mg ratios in the lower mantle to balance the Si-poor and A-rich upper mantle and produce a bulk Earth with high Si and low refractory element contents, typical of EC. Complex processes are required to produce such a compositionally layered mantle. It seems impossible to keep refractory element ratios chondritic during major igneous fractionation on Earth, required to satisfy the Javoy model.

The Mg/Si Problem of Earth’s Mantle, Si and Cr in the Core?

The superchondritic Mg/Si ratio of the mantle of the Earth is well established and was discussed by several authors (e.g., Jagoutz et al., 1979; Ringwood, 1979).

In Figure 8 Si/Mg versus Al/Mg has been plotted for terrestrial rocks. It is more convenient to have Mg in the denominator, because Mg does not partition into metal, while Si does, given sufficiently reducing conditions. There are data of some 350 peridotites and several hundred MORB (mid ocean ridge basalts) glasses plotted. On this scale the differences between PUM, CI-chondrites, and the Sun plot cannot be resolved. On the other hand, there is a large difference in chemistry between MORB and the mantle of the Earth. Data for partial melts from experimental petrology by Baker and Stolper (1994) are also plotted. The large gap in chemistry between upper mantle and basaltic rocks (natural and produced by experimental petrology) is the result of eutectic melting of the mantle. There is no compositional continuum between mantle rocks and melts produced by partial melting of the mantle. As pointed out before it should be emphasized that there are rocks on the surface of the Earth with nearly solar ratios of Mg, Si, and Al. Such rocks are not available from other planets. The rocks that were analyzed from Moon, Mars, and Venus are far from primitive, chondritic rocks with regard to Mg, Si, and Al. Their MgO contents are too low. Even the Chassigny meteorite, a Martian ultramafic rock, has only 31.6% MgO and 0.69% Al2O (Burghele et al., 1983). It is important to note that the bulk chemical composition of Earth, including major, minor, and trace elements, is much better known than estimated bulk silicate composition of other planets.

Figure 8. Si/Mg versus Al/Mg on a log–log scale. The upper mantle rocks are represented by 365 xenoliths data from the GEOROC (Sarbas & Nohl, 2008). They are representative of peridotitic rocks from worldwide localities. The PM composition is plotted at the high Al/Mg end (this work). The composition of 2,912 mid-ocean ridge basalt (MORB) glasses is indicated (GEOROC ; Sarbas & Nohl, 2008). The results of partial melting experiments with PM-like compositions at 10 kbar and 1,270 and 1,390°C are shown for comparison (Baker & Stolper, 1994). The composition of the continental crust is indicated (Rudnick & Gao, 2003).

Figure is from Palme and O’Neill (2014).

Figure 9 is a close-up of Figure 8. It demonstrates the depletion of Si in Earth’s mantle relative to chondritic meteorites. Chondritic meteorites and the Sun have higher Si contents than rocks from the mantle of the Earth.

Figure 9. Si/Mg versus Al/Mg plot for upper mantle rocks and chondritic meteorites. Two PM compositions from this work and from McDonough and Sun (1995) are indicated. For EC and ordinary chondrites (OC), only averages are plotted (Wasson & Kallemeyn, 1988). The CC data are from Wolf and Palme (2001). They show a large spread in Al/Mg but have fairly constant Si/Mg ratios. Earth’s mantle is clearly enriched in Al (and other refractory elements) and depleted in Si, relative to the Sun or CI meteorites. With 7% Si in the core, the bulk Earth composition would plot on the CI Si/Mg ratio. The solar data plotted are from Lodders et al. (2009). The newest solar data from Scott et al. (2015) are different but overlap within error bars with other solar data (see Figure 3). The figure is from Palme and O’Neill (2014).

Ringwood (1989) suggested that the Si/Mg ratio of Earth’s mantle is the same as the solar ratio, implying that CI chondrites have a non-solar Si/Mg ratio. This is very unlikely in view of the excellent agreement between elemental abundances of CI chondrites and the Sun (Palme et al., 2014). Other attempts to explain the low Si/Mg ratio of Earth’s mantle include volatilization of Si from the inner solar system by high temperature processing connected with the early evolution of the protosun and the solar nebula (Ringwood, 1979, 1991). As mentioned, Dauphas et al. (2015) found a correlation of δ30Si with the Mg/Si ratios of chondrites and planets. These authors conclude that fractionation of olivine in the early solar system could produce the observed variations in δ30Si as well as the correlated changes in Mg/Si ratios. Another possibility is partitioning of some Si into the metal core of Earth (e.g., Hillgren, Gessmann, & Li, 2000 and references therein). The latter proposition has the advantage that it would also reduce the density of the outer core, which is about 10% too low for an FeNi alloy (Poirier, 1994). That Earth’s core contains some Si is now assumed by most authors, but the amount is unclear. About 18% Si would be required to produce the 10% density deficit. This number is far too high, as argued by O’Neill and Palme (1998), who concluded that other light elements must be involved in reducing the density of the core. The problem may be somewhat less severe in light of density estimates of liquid FeNi alloys at high pressures and temperatures by Anderson and Isaak (2002), leading to a density deficit of only 5%. Based on these data, McDonough (2014) estimated a core composition with 6% Si and 1.9% S as light elements. Earlier models by Wänke (1981) had 12.5% Si in the core, and 7.3% in the Allègre et al. (1995) model, respectively. In a recent accretion model by Rubie et al. (2011), Earth’s core contains 8 wt% Si. Since about 60% of the total Cr of the Earth is in the core, it is not surprising that some Si is in the core; both elements have siderophile tendencies, Cr being more more siderophile than Si (Mann, Frost, & Rubie, 2009).

Figure 10. Cr/Mg versus Al/Mg in upper mantle rocks and chondritic meteorites. The low Cr/Mg ratios of Earth’s mantle reflect the presence of Cr in the core. This requires reducing conditions during core formation in the early history of Earth (see text). The sources of data are the same as in Figures 14 and 15. The figure is from Palme and O’Neill (2014).

A comparable, yet even more dramatic effect than for Si can be seen for Cr. The Cr content of the upper mantle rocks is fairly constant and independent of the Al/Mg ratio (Figure 10). The constant Cr of peridotites indicates an average Cr solid/liquid partition coefficient of about one. Chondritic meteorites have a factor of 2 higher Cr/Mg ratios than upper mantle rocks (Figure 10). Both elements, Cr and Mg, are reasonably well correlated in chondritic meteorites. They have similar condensation temperatures, although their condensation behavior is very different, Mg is calculated to condense as forsterite at 1336 K (10−4 bar, 10 Pa) and Cr in solid solution with metal at 1296 K and 10−4 bar pressure (Lodders, 2003). The obvious explanation for the low Cr/Mg ratios in upper mantle rocks is that a major fraction of the terrestrial Cr is in the core. From mass balance and the assumption of a chondritic Cr/Mg ratio for the bulk Earth (around 36), one calculates that about 60% of the total Cr of Earth is in the core. The Cr deficit in Earth’s mantle has been known for a long time and it has always been interpreted as indicating reducing conditions during formation of Earth’s core, at least during its initial stage (O’Neill, 1991; Ringwood, Kato, Hibberson, & Ware, 1990; Wänke, 1981). Recent modeling of the accretion history of Earth including core formation also postulates accretion with initially reduced materials to accommodate the low Cr and also V in mantle rocks (Rubie et al., 2011; Wade & Wood, 2005). At the oxygen fugacity of the present mantle with 8% FeO, negligible amounts of Cr would partition into the core. Siebert, Badro, Antonangeli & Ryerson (2013), however, have challenged this conclusion. Based on high pressure partition data with the laser heated diamond anvil cell, they found at high temperatures and high pressures unexpectedly high metal-silicate partition coefficients for Cr and V, even at oxidizing conditions. These high partition coefficients are not the result of direct extrapolation of low temperature partition data. This does not necessarily exclude the possibility of changes of initially reducing to more oxidizing conditions during accretion, but if true, the strong argument for reducing conditions early in the accretion history of the Earth is weakened. It leaves open the possibility of chemical exchange between core and mantle late in the accretion history of Earth. In any case the too low Cr and V contents of the Earth’s mantle is a powerful argument for core mantle interaction.

The Bulk Composition of the Earth

So far, the mantle of the Earth, which accounts for 32.5% of the total mass of Earth, has been considered. When comparing the Earth’s mantle composition with chondritic meteorites the most obvious “anomaly” in the mantle composition is the low content of iron and other siderophile elements. As described, this led early on to the suggestion that the core of the Earth consists of iron. As meteoritic metal is always accompanied by Ni and Co, most authors concluded that the core of the Earth must contain several percent of Ni and a corresponding fraction of Co, so that the bulk Earth Fe/Ni and Ni/Co are CI-chondritic (Table 2). This view has not changed and an FeNi core of the Earth is almost universally accepted.

The bulk Earth composition is the sum of the composition of BSE and the core. The Fe and Ni contents of the Earth maybe calculated with some confidence. But the trace element contents of the core, including S, are largely unknown or, to say it positively, model dependent. Bulk Earth compositions based on BSE and core composition are given by several authors: Kargel and Lewis (1993), McDonough (2014), Wang, Lineweaver, and Ireland (2018).

The Iron Content of the Earth

The bulk Fe content of Earth can be calculated by mass balance from the FeO content of the PM and the Fe content of the core. For example, assuming 75% Fe in the core yields an Fe/Mg ratio of bulk Earth of 1.91, compared to 1.96 for CI-chondrites. After the addition of about 5% Ni, 20% of the mass of the core would be left for the light element(s), which seems far too much. A higher Fe content of the core leads to higher than CI bulk Earth Fe/Mg ratios. Type 3 carbonaceous chondrites have lower Fe/Mg ratios,1.53 on average (Wolf & Palme, 2001), but EH-chondrites have a significantly higher Fe/Mg ratio of 2.74 (Wasson & Kallemeyn, 1988). Anderson and Isaak (2002) estimated only 5% for the light element in the core, which would require an Fe content of the core of 90%, which leads to an Fe/Mg bulk Earth ratio of 2.24. McDonough (2014) presents two models with 85.5 and 88.3% Fe in the core. The lower value yields a bulk Earth Fe/Mg ratio of 2.11 and the higher Fe content to 2.20 corresponding to 11 and 14% higher Fe/Mg ratios than CI chondrites. It is thus very unlikely that bulk Earth has a CI-chondritic Fe/Mg ratio. Most other groups of chondrites have even lower Fe/Mg ratios than CI chondrites. Exceptions are H chondrites with similar Fe/Mg ratios as CI chondrites and EH chondrites with Fe/Mg ratios exceeding the CI ratio. But the high Si/Mg ratios of EH chondrites disqualify them as proto-Earth materials, as discussed. The conclusion, therefore, is that Earth must have an Fe excess of at least 10%. O’Neill and Palme (2008) suggested that the high Fe content of the Earth is the result of collisional erosion of the silicate shell of the Earth, whereby the Earth has lost a significant fraction of silicates, but almost no iron, which largely resides in the core of the Earth.

The Composition of the Core

The low contents of siderophile elements in the Earth’s mantle (see Figure 5) is generally ascribed to the removal of siderophile elements from the mantle by metal segregation to the core. Good examples are Ni and Co. Both elements are significantly depleted in PUM relative to chondritic abundances, that is, the Mg/Ni ratio in PUM is about a factor of 10 higher than in chondritic meteorites (see Figure 5). Experimentally determined metal/silicate partition coefficients at atmospheric pressure would predict much lower Ni contents in PUM and a Ni/Co ratio much lower than that observed, as Ni is much more siderophile than Co (e.g., Schmitt, Palme, & Wänke, 1989). In their experiments Li and Agee (1996) found that pressure and temperature have an enormous effect on Ni, Co metal/silicate partition coefficients. With increasing pressure and temperature, Ni and Co become less siderophile and their metal/silicate partition coefficients converge. Subsequent work has confirmed this and it was also found that a sudden decrease of Ni and Co partition coefficients at 3–5 GPa, probably related to the structural position of Ni and Co in silicate melts, significantly contributed to this behavior (Kegler et al., 2008 and references therein). A study by Siebert, Badro, Antonangeli, and Ryerson (2012) using diamond anvil cells found agreement of observed Ni and Co mantle contents with contents predicted by partition coefficients at a pressure of 50–60 GPa and a temperature of 3,000 to 4,00K. The metal in equilibrium with silicate melt at these conditions had about 1.3% oxygen and 6% silicon.

The concentrations of other siderophile or chalcophile elements in the mantle of the Earth depend on their metal-silicate or sulfide-silicate partition coefficients and their pressure and temperature dependence. With the knowledge of such partition coefficients reasonable core formation models can be constructed (Mann et al., 2009).

Rubie et al. (2015) made extensive calculations to simulate the growth of the Earth and the formation of the core. Six simulations were selected from a large suite of Grand Tack N-body accretion simulations by Jacobson and Morbidelli (2014). All the models began with 68 to 213 different embryos (Moon to Mars sized differentiated objects) and around 2,000 smaller planetesimals, which made up the Earth and the other inner planets by collisions 100 to 150 million years after formation of the solar system. The starting planetary bodies were spread out from near the Sun to Jupiter, with chondritic bulk compositions, but enhanced refractories and increasing fractions of volatiles with increasing distance from the Sun. Depending on the size of the bodies, magma oceans were formed and liquid metal settled to the core. The simulations, based on the metal-silicate partition coefficients by Mann, Frost and Rubie (2009), led to the following results: (a) accretion of Earth was heterogeneous, from initially reducing to later oxidizing, (b) metal–silicate equilibration pressures increased as accretion progressed and were, on average, 60–70% of the core–mantle boundary pressures at the time of each impact, and (c) a large fraction (70–100%) of the metal of impactor cores equilibrated with a small fraction of the silicate mantles of proto-planets during each core formation event. The models with the approximately correct FeO in the BSE predict the Ni, Co, Cr, V, and Nb abundances in the silicate mantle correctly.

Moderately volatile and highly volatile elements (Te, Sn, Pb, Tl, etc.) mostly have siderophile and/or chalcophile tendencies. It is assumed that their abundances initially followed the volatility trend marked in Figure 5. During metal and sulfide segregation they were partly extracted into the core. Using their initial abundances calculated from the volatility trend (Figure 5) and their present mantle abundances, the concentrations in the core can be calculated. McDonough (2014) has performed such calculations and gives core concentration for these elements. Another interpretation is suggested by Braukmüller, Wombacher, Funk, and Münker (2019). These authors believe that the mantle of the Earth initially had a hockey stick pattern of volatile and highly volatile elements, a pattern typical of carbonaceous chondrites (Braukmüller, Wombacher, Hezel, Escoube, & Münker, 2018) with a continuous CI-normalized decrease from Mg to Zn and a flat pattern from Zn to In. Core formation removed elements between Zn and In (Sn, Cd, Tl, Se, Te; Figure 5) from the mantle to various degrees. Only Zn and In were not affected by core formation. The high abundance of In in the mantle of the Earth has been a problem for some time (Witt-Eickschen et al., 2009). It does not fit into a continuous decrease of volatiles with decreasing condensation temperatures. The CI/mantle concentration ratio for In is 4.3, for Zn 5.7, compared to 19.3 for Cd, 12 for Sn, and 250 for Te (Table 2). Indium has siderophile and chalcophile tendencies and yet it does not seem to be much influenced by core formation (see discussion in Witt-Eickschen et al., 2009).

Another group of elements, the highly siderophile elements with metal-silicate partition coefficients above 104 are strongly depleted in the Earth’s mantle. They were more or less quantitatively extracted from the mantle with core forming metal and later added with a chondritic veneer. The metal-silicate partition coefficients are different for different HSE and the only way to explain their approximately chondritic relative abundances is to extract in a first step the HSE quantitatively and add them in a second step with chondritic material (Mann, Frost, Rubie, Becker, & Audétat, 2012).

A table of the elemental composition of the core requires certain assumptions. It is particularly important to get some idea about the amount of S in the core. Based on S abundances in massive peridotites, the Earth’s mantle has about 200 ppm S (Table 2). The core is certainly higher in S. Dreibus and Palme (1996) estimated the bulk Earth S content by using the PUM Zn content as a proxy, because Zn and S have similar condensation temperatures indicating similar volatilities. It is, in addition, assumed that Zn is entirely lithophile. The calculated bulk Earth S-content is 0.56% and the S-content of the core is 1.7%. Newer studies indicate, however, that Zn has some siderophile tendencies (Corgne, Keshav, Wood, McDonough, & Fei, 2008; Mann et al., 2009). Corgne et al. (2008) estimate a maximum of 30 ppm in the core. This would increase the Zn content of bulk Earth by 30% and yield a S content of 2.2% for the core. An estimate for the composition of the core would be: 85% Fe, 7% Si, 5% Ni, 2% S, 0.75% Cr, 0.24% Co. This would amount to 100% in total. More precise estimates make little sense, in view of the uncertainties involved, in particular with regard to the light element(s) in the core. A good candidate is oxygen. In the calculations of Rubie et al. (2015), there is 2.58 to 3.81% oxygen in the cores of Earth-like planets. If there is oxygen in the core all other constituents would decrease correspondingly.

The uncertainties involved in the composition of the core propagate to the bulk composition of the Earth, if a substantial fraction of an element is in the core. The abundances of strictly lithophile elements in the bulk Earth are simply reduced by 32.5%, compared to BSE compositions. For all other elements the fractions in the metal core and the silicate mantle have to be added up to the bulk Earth concentrations.

The Isotopic Composition of Earth

The most abundant element in the Earth (by atoms) is oxygen. In a diagram of δ17O versus δ18O, the oxygen isotopic compositions of terrestrial rocks plot along a line with a slope of 0.5, designated as the terrestrial fractionation line (TFL; Clayton, 1993). Samples of other planetary bodies plot along parallel lines with the same slope. The vertical distance of these lines to the TFL is termed Δ17O, so that Δ17O is zero for TFL. It is generally assumed that samples plotting along a single slope 0.5 line belong to the same parent body, while parallel lines indicate different parent bodies. The work of Clayton (1993 and references therein) has demonstrated that our solar system contains more than 50 such parent bodies. There are attempts to model the Earth and Mars as mixtures of several meteorite parent bodies, using oxygen isotopes (Lodders, 2000) and other stable isotopes (Fitoussi, Bourdon, & Wang, 2016).

Figure 11. Δ17O of the various groups of chondritic meteorites. In OC and EC, Δ17O increases with increasing oxygen content, reflecting increasing reaction with an 17O-rich reservoir. For CC the situation is more complicated: CRs contain abundant metals,; whereas CMs have a few percent of water, both have similar Δ17O. The most reduced chondrites (EL) and the most oxidized chondrites (CI) have similar Δ17O. This is also the Δ17O of Earth (reference line).

Figure 11 shows the Δ17O values of the various groups of chondritic meteorites. On the left side of the diagram are members of non-carbonaceous chondrites (NCC), as defined by Warren (2011). There is a clear trend of increasingly heavier oxygen with increasing total oxygen as reflected in the FeO contents of olivine and pyroxene from EH to R chondrites. This could be viewed as reflecting increasing additions of 16O poor oxygen (see also Figure 3). The CC on the right side of Figure 11 does not follow such a simple trend. Meteorite groups are arranged in order of increasing content of water. There is no correlation of Δ17O with the amount of oxygen. The CR-group meteorites contain abundant metal and have much less total oxygen than CM-chondrites with several percent of water, although they have similar Δ17O. The Earth, at Δ17O = 0, could claim membership of CC or NCC. It has the same oxygen isotopic composition as enstatite chondrites but is also not far from type 1 carbonaceous chondrites.

Isotopic signatures of other elements (e.g., 50Ti, 54Cr, 62Ni) confirm the dichotomy of meteorite types with NCC and CC (Warren, 2011). For most isotopes the Earth falls into the group of NCC. As an example, a 50Ti vs Al/Si plot is shown (Figure 12). Earth has the same 5OTi signature as ordinary and enstatite chondrites but with higher abundances of refractory elements (Al/Si ratio). Earth fits chemically better with carbonaceous chondrites. This is a general observation. The chemical composition of Earth seems to extend trends in CV-chondrites, but the isotopic composition is more similar to H- and E-chondrites (NCC).

Figure 12. The 50Ti deviations from the terrestrial standard versus the Al/Si ratio. EC and OC plot on or slightly below the terrestrial reference, while CC have higher 50Ti and are more enriched in refractory elements, also typical of the BSE. This highlights the isotopic similarity of Earth with NCC and the chemical similarity with CC. 50Ti data are from the supplemental material of Dauphas and Schauble (2016).

Figure 13 shows compositional trends of CC and NCC in comparison with the composition of Earth. All data are normalized to CI and Si. Aluminum and all other refractory elements plot below the CI-line, whereas CC have increasing refractory element contents in the sequence CI, CM, CO, CV. The Earth seems to extend this trend to still higher refractory element contents. Normalizing to Mg would produce a lower increase in refractory elements, because PUM has a low Si/Mg ratio. The strong depletion in Mn and other moderately volatile elements in CC seem to continue to even stronger depletions in Earth. Again, NCCs are less depleted in Mn. Sodium and K follow the same trend as Mn. The quintessence is that Earth is chemically similar, but not identical, to CC, but with regard to stable isotopes follows NCC.

Figure 13. The similarity of the chemistry of silicate Earth with carbonaceous chondrites is evident in this figure. CC have slightly higher Mg, significantly higher Al, and lower Mn than NCC. Earth chemistry appears as extension of CC trends.

It is not easy to reconcile these apparently contradicting findings. The Morbidelli, Libourel, Palme, Jacobson, and Rubie (2020) paper offers a possibility. In this model Earth has elevated concentrations of refractory elements and is low in moderately volatile elements, such as Mn, because Earth or the material that made the Earth is incompletely condensed. Thus the reason for the chemical similarity of carbonaceous chondrites and Earth is not that Earth is largely composed of CC material, but processes involved in making the Earth produced fractionations similar to those observed in CC. The Morbidelli et al. model requires then, that the anhydrous CI-component which makes up most of OC and EC has a similar isotopic composition as the material parental to the Earth.

Other stable isotopes also provide evidence that Earth is not simply a mixture of different types of chondritic meteorites, as has been suggested earlier on the basis of oxygen isotopes (Fitoussi et al., 2016; Lodders, 2000). Burkhardt, Kleine, Oberli, Pack, and Bourdon (2011) showed that terrestrial Mo and Ru are richer in s-process isotopes than any known meteorites. This implies that Earth cannot be represented by any combination of known types of meteorites. If NCC or CC contribute to Earth then material richer in s-process Mo than Earth is required to balance the CC and NCC Mo. More recent analyses have shown that this argument can be extended to Ru and Nd (Render et al., 2017).

Conclusions

It took some 200 years to create a reasonable composition of the upper mantle of the Earth. This composition derived from the analyses of upper mantle rocks is most likely representative of the bulk Earth mantle, as there is increasing evidence for a well-mixed mantle. Although Earth’s mantle has roughly chondritic composition of nonmetallic elements, it is, in detail, different in chemistry from known meteorites, a conclusion reached earlier by some authors (McDonough & Sun, 1995). Recent isotopic evidence confirms this conclusion and firmly excludes an origin of Earth by any mixing of chondritic meteorites. The consequences of this finding for the origin of the Earth are presently unclear. If the Earth has collected material from a variety of heliocentric distances it should chemically and isotopically represent a mixture of presently observed meteorites and remnants from the early solar system, as suggested by current Earth making models (e.g., Rubie et al., 2011, 2015). Or did Earth form in a completely different way, at different locations and by different processes than meteorites? These are major questions, which need to be answered by future research.

Acknowledgments

The author is grateful to Bill McDonough for his very helpful comments.

References

  • Agrusta, R., van Hunen, J., & Goes, S. (2018). Strong plates enhance mantle mixing in early Earth. Nature Communications, 9, 1–10.
  • Allègre, C. J., Poirier, J. P., Humler, E., & Hofmann, A. W. (1995). The chemical composition of Earth. Earth and Planetary Science Letters, 134, 515–526.
  • Anderson, O. L., & Isaak, D. G. (2002). Another look at the core density deficit of Earth’s outer core. Physics of the Earth and Planetary Interiors, 131, 10–27.
  • Arevalo, R. A., McDonough, W. F., & Luong, M. (2009). The K/U ratio of the silicate Earth: Insights into mantle composition, structure and thermal evolution. Earth and Planetary Science Letters, 278, 361–369.
  • Baker, M. B., & Stolper, E. M. (1994). Determining the composition of high-pressure mantle melts using diamond aggregates. Geochimica et Cosmochimica Acta, 58, 2811–2827.
  • Ballmer, M. D., Houser, C., Hernlund, J. W., Wentzcovitch, R. M., & Hirose, K. (2017). Persistence of strong silica-enriched domains in the Earth’s lower mantle. Nature Geoscience, 10, 236.
  • Becker, H., Horan, M. F., Walker, R. J., Gao, S., Lorand, J. P., & Rudnick, R. L. (2006). Highly siderophile element composition of the Earth’s primitive upper mantle: Constraints from new data on peridotite massifs and xenoliths. Geochimica et Cosmochimica Acta, 70, 4528–4550.
  • Bercovici, D. (2015). Mantle dynamics past, present, and future: An introduction and overview. In G. Schubert (Ed.), Treatise on Geophysics: Volume 7. Mantle dynamics (D. Bercovici, Ed.) (pp. 1–30). Oxford, UK: Elsevier-Pergamon.
  • Bodinier, J. L., & Godard, M. (2003). Orogenic, ophiolithic, and abyssal peridotites. In R. W. Carson (Ed.), The mantle and Core: Volume 2. Treatise in geochemistry (H. D. Holland & K. K. Turekian, Eds.) (pp. 103–170). Oxford, UK: Elsevier-Pergamon.
  • Bond, J., Lauretta, D. S., & O’Brien, D. P. (2010). Making the Earth: Combining dynamics and chemistry in the solar system. Icarus, 205, 321–337.
  • Bouvier, A., Vervoort, J. D., & Patchett, P. J. (2008). The Lu-Hf and Sm-Nd isotopic composition of CHUR: Constraints from unequilibrated chondrites and implications for the bulk composition of terrestrial planets. Earth and Planetary Science Letters, 273, 48–57.
  • Boyet, M., & Carlson, R. W. (2005). 142Nd evidence for early (> 4.53 Ga) global differentiation of the silicate Earth. Science, 309, 576–581.
  • Braukmüller, N., Wombacher, F., Funk, C., & Münker, C., (2019). Earth’s volatile element depletion pattern inherited from a carbonaceous chondrite-like source. Nature Geoscience, 12, 564–568.
  • Braukmüller, N., Wombacher, F., Hezel, D. C., Escoube, R., & Münker, C. (2018). The chemical composition of carbonaceous chondrites: Implications for volatile element depletion, complementarity and alteration. Geochimica et Cosmochimica Acta, 23, 17–48.
  • Brush, S. G. (1980). Discovery of the Earth’s core. American Journal of Physics, 48, 705–723.
  • Buddington, A. F. (1943). Some petrological concepts and the interior of the Earth. American Mineralogist, 28, 119–140.
  • Burghele, A., Dreibus, G., Palme, H., Rammensee, W., Spettel, B., Weckwerth, G., & Wänke, H. (1983). Chemistry of the Shergottites and the Shergotty Parent Body (SPB): Further evidence for the two component model of planet formation. Lunar and Planetary Science, 14, 1041 (abstract).
  • Burke, J. G. (1986). Cosmic debris: Meteorites in history. Berkeley: University of California Press.
  • Burkhardt, C., Borg, L. E., Brennecka, G. A., Shollenberger, Q. R., Dauphas, N., & Kleine, T. (2016). A nucleosynthetic origin for the Earth’s anomalous 142Nd composition. Nature, 537, 394–398.
  • Burkhardt, C., Kleine, T., Oberli, F., Pack, A., & Bourdon, B. (2011). Molybdenum isotope anomalies in meteorites: Constraints on solar nebula evolution and origin of the Earth. Earth and Planetary Science Letters, 312, 390–400.
  • Basaltic Volcanism Study Project (BVSP). (1981). Basaltic volcanism on the terrestrial planets. New York, NY: Pergamon Press.
  • Cameron, A. G. W. (1962). The formation of the sun and planets. Icarus, 1, 13–69.
  • Cameron, A. G. W., & Pine, M. R. (1973). Numerical models of the primitive solar nebula. Icarus, 18, 377–406.
  • Cassen, P. (2001). Nebular thermal evolution and the properties of primitive planetary materials. Meteoritics and Planetary Science, 36, 671–700.
  • Chladni, E. F. F. (1794). Über den Ursprung der von Pallas gefundenen und anderer ihr ähnlicher Eisenmassen und über einige damit in Verbindung stehende Naturerscheinungen [About the origin of the iron mass found by Pallas and similar ones to it, and on some connected natural phenomena]. Riga, Latvia: J. F. Hartkoch. Translated in 1803–1804, Journal of Mines, 15, 286–320, 466–485.
  • Chou, C. L. (1978). Fractionation of siderophile elements in the Earth’s upper mantle. Proceedings of the Lunar Planetary Science Conference, 9, 219–230.
  • Clayton, R. N. (1993). Oxygen isotopes in meteorites. Annual Review of Earth and Planetary Sciences, 21, 115–149.
  • Corgne, A., Keshav, S., Wood, B. J., McDonough, W. F., & Fei, Y. W. (2008). Metal–silicate partitioning and constraints on core composition and oxygen fugacity during Earth accretion. Geochimica et Cosmochimica Acta, 72, 574–589.
  • Corgne, A., Liebske, C., Wood, B. J., Rubie, D. C., & Frost, D. J. (2005). Silicate perovskite-melt partitioning of trace elements and geochemical signature of a deep perovskitic reservoir. Geochimica et Cosmochimica Acta, 69, 485–496.
  • Dauphas, N., Poitrasson, F., Burkhardt, C., Kobayashi, H., & Kurosawa, K. (2015). Planetary and meteoritic Mg/Si and δ‎30Si variations inherited from solar nebula chemistry. Earth and Planetary Science Letters, 427, 236–248.
  • Dauphas, N., & Schauble, E. A. (2016). Mass fractionation laws, mass-independent effects, and isotopic anomalies. Annual Review of Earth and Planetary Sciences, 44, 709–783.
  • Dreibus, G., & Palme, H. (1996). Cosmochemical constraints on the sulfur content in the Earth’s core. Geochimica et Cosmochimica Acta, 60, 1125–1130.
  • Emden, R. (1907). Gaskugeln Anwendungen der mechanischen Wärmetheorie auf kosmologische und meteorologische Probleme. Druck und Verlag von B.G., Teubner.
  • Eucken, A. (1944). Über den Zustand des Erdinnern. Naturwissenschaften, 14, 112–121.
  • Farrington, O. C. (1901). The constituents of meteorites. Journal of Geology, 9, 522.
  • Fegley, B., Lodders, K., & Jacobson, N. S. (2020). Volatile element chemistry during accretion of the Earth. Geochemistry, 80(1), 125594.
  • Fitoussi, C., & Bourdon, B. (2012). Silicon isotope evidence against an enstatite chondrite Earth. Science, 335, 1477–1480.
  • Fitoussi, C., Bourdon, B., & Wang, X. (2016). The building blocks of Earth and Mars: A close genetic link. Earth and Planetary Science Letters, 434, 151–160.
  • Frey, F. A., & Green, D. H. (1974). The mineralogy, geochemistry and origin of lherzolite inclusions in Victorian basalts. Geochimica Cosmochimica Acta, 38, 1023–1059.
  • Frey, F. A., Suen, C. J., & Stockman, H. W. (1985). The Ronda high temperature peridotite: Geochemistry and petrogenesis. Geochimica et Cosmochimica Acta, 49, 2469–2491.
  • Fukao, K., & Obayashi, M. (2013). Subducted slabs stagnant above, penetrating through, and trapped below the 660 km discontinuity. Journal of Geophysical Research Solid Earth, 118, 5920–5938.
  • Ganapathy, R., & Anders, E. (1974). Bulk compositions of the moon and Earth, estimated from meteorites. Proceedings of the 5th Lunar Science Conference, 2, 1181–1206.
  • Gast, P. W. (1960). Limitations on the composition of the upper mantle. Journal of Geophysical Research, 65, 1287–1297.
  • Gellissen, M., Holzheid, A., Kegler, P., & Palme, H. (2019). Heating experiments relevant to the depletion of Na, K and Mn in the Earth and other planetary bodies. Geochemistry, 79, 1–13.
  • Goldschmidt, V. M. (1922). Über die Massenverteilung im Erdinnern, verglichen mit der Struktur gewisser Meteoriten. Naturwissenschaften, 10, 918–920.
  • Goldschmidt, V. M. (1930). Geochemische Verteilungsgesetze und kosmische Häufigkeit der Elemente. Naturwissenschaften, 18, 999–1013.
  • Green, D. H., & Falloon, T. (1998). Pyrolite: A Ringwood concept and its current expression. In I. Jackson (Ed.), The Earth’s mantle: Structure, composition and evolution—The Ringwood volume (pp. 311–378). Cambridge, UK: Cambridge University Press.
  • Grossman, L. (1972). Condensation in the primitive solar nebula. Geochimica et Cosmochimica Acta, 36, 597–619.
  • Gutenberg, B. (1914). Über Erdbebenwellen VIIA. Nachr. Ges. Wiss. Gottingen Math. Physik. Kl, 166. Nachrichten von der Königlichen Gesellschaft der Wissenschaften zu Goöttingen, Mathematisch-Physikalische Klasse, 125, 116–176.
  • Harkins, W. D. (1917). The evolution of the elements and the stability of complex atoms. Journal of the American Chemical Society, 39, 856–879.
  • Hart, S. R., & Zindler, A. (1986). In search of a bulk-Earth composition. Chemical Geology, 57, 247–267.
  • Hillgren, V. J., Gessmann, C. K., & Li, J. (2000). An experimental perspective on the light element in the Earth’s core. In R. M. Canup & K. Righter (Eds.), Origin of the Earth and Moon (pp. 245–263). Tucson: University of Arizona Press.
  • Hofmann, A. W. (1997). Mantle geochemistry: The message from oceanic volcanism. Nature, 385, 219–229.
  • Humayun, M., & Cassen, P. (2000). Processes determining the volatile abundances of the meteorites and the terrestrial planets. In R. M. Canup & K. Righter (Eds.), Origin of the Earth and Moon (pp. 3–23). Tucson: University of Arizona Press.
  • Hyung, E., Huang, S., Petaev, M. I., & Jacobsen, S. B. (2016). Is the mantle chemically stratified? Insights from sound velocity modeling and isotope evolution of an early magma ocean. Earth and Planetary Science Letters, 440, 158–169.
  • Ionov, D. A., & Hofmann, A.W. (2007). Depth of formation of subcontinental off-craton peridotites. Earth and Planetary Science Letters, 261, 620–634.
  • Jacobson, S. A., & Morbidelli, A. (2014). Lunar and terrestrial planet formation in the Grand Tack scenario. Philosophical Transactions of the Royal Society A, 372, 20130174.
  • Jagoutz, E., Palme, H., Baddenhausen, H., Blum, K., Cendales, M., Dreibus, G., . . . Wänke, H. (1979). The abundances of major, minor and trace elements in the Earth’s mantle as derived from primitive ultramafic nodules. Proceedings of the Lunar Planetary Science. Conference, 10, 1141–1175.
  • Jarosewich, G. (1990). Chemical analyses of meteorites: A compilation of stony and iron meteorite analyses. Meteoritics, 25, 323–337.
  • Javoy, M., Kaminski, E., Guyot, F., Andrault, D., Sanloup, C., Moreira, M., . . .Jaupart, C. (2010). The chemical composition of the Earth: Enstatite chondrite models. Earth and Planetary Science Letters, 293, 259–268.
  • Jochum, K. P., Hofmann, A. W., & Seufert, H. M. (1993). Tin in mantle-derived rocks: Constraints on Earth evolution. Geochimica et Cosmochimica Acta, 57, 3585–3595.
  • Jochum, K. P., McDonough, W. F., Palme, H., & Spettel, B. (1989). Compositional constraints on the continental lithospheric mantle from trace elements in spinel peridotite xenoliths. Nature, 340, 548–550.
  • Kargel, J., & Lewis, J. (1993). The composition and early evolution of Earth. Icarus, 105, 1–25.
  • Kato, T., Ringwood, A. E., & Irifune, T. (1988). Experimental determination of element partitioning between silicate perovskites, garnets and liquids: Constraints on early differentiation of the mantle. Earth and Planetary Science Letters, 89, 123–145.
  • Keays, R. R., Ganapathy, R., & Anders, E. (1971). Chemical fractionations in meteorites—IV Abundances of fourteen trace elements in L-chondrites: Implications for cosmothermometry. Geochimica et Cosmochimica Acta, 35, 837–868.
  • Kegler, P., Holzheid, A., Frost, D. J., Rubie, D. C., Dohmen, R., & Palme, H. (2008). New Ni and Co metal-silicate partitioning data and their relevance for an early terrestrial magma ocean. Earth and Planetary Science Letters, 268, 28–40.
  • Kleine, T., & Walker, R. J. (2017). Tungsten isotopes in planets. Annual Review of Earth Planetary Science, 45, 389–417.
  • Krot, A. N., Keil, K., Goodrich, C. A., Scott, E. R. D., & Weisberg, M. K. (2003). Classification of meteorites. In A. M. Davis (Ed.), Meteorites, comets and planets: Vol. 1. Treatise on Geochemistry (H. D. Holland & K. K. Turekian, Eds.) (pp. 83–128). Oxford, UK: Elsevier.
  • Kruijer, T. S., Touboul, M., Fischer-Gödde, M., Bermingham, K. M., Walker, R. J., & Kleine, T. (2014). Protracted core formation and rapid accretion of protoplanets. Science, 344, 1150–1153.
  • Laurenz, V., Rubie, D. C., Frost, D. J., & Vogel, A. K. (2016). The importance of sulfur for the behavior of highly-siderophile elements during Earth’s differentiation. Geochimica et Cosmochimica Acta, 194, 123–138.
  • Le Roux, V., Bodinier, J.-L., Tommasi, A., Alard, O., Dautria, J.-M., Vauchez, A., & Riches, A. J. V. (2007). The Lherz spinel lherzolite: Refertilized rather than pristine mantle. Earth Planetary Science Letters, 259, 599–612.
  • Lewis, J. S. (1972). Metal/silicate fractionation in the solar system. Earth and Planetary Science Letters, 15, 286–290.
  • Li, J., & Agee, C. B. (1996). Geochemistry of mantle–core differentiation at high pressures. Nature, 381, 686–689.
  • Lodders, K. (2000). An oxygen isotope-mixing model for the accretion and composition of rocky planets. Space Science Reviews, 92, 341–354.
  • Lodders, K. (2003). Solar system abundances and condensation temperatures of the elements. Astrophysical Journal, 591, 1220–1247.
  • Lodders, K., Palme, H., & Gail, H. P. (2009). Abundances of the elements in the solar system. In J. E. Trümper (Ed.), Landolt-Börnstein, New Series, Vol. VI/4B (Chap. 4.4, pp. 560–630). Berlin, Germany: Springer-Verlag.
  • Lugovic, B., Altherr, R., Raczek, I., Hofmann, A. W., & Majer, V. (1991). Geochemistry of peridotites and mafic igneous rocks from the Central Dinaric Ophiolite Belt, Yugoslavia. Contributions to Mineralogy and Petrology, 106, 201–216.
  • Lyubetskaya, T., & Korenaga, J. (2007). Chemical composition of Earth’s primitive mantle and its variance: 1. Methods and results. Journal of Geophysical Research, 112, B03211, 1–21.
  • Maaløe, S., & Aoki, K.–I. (1977). The major element composition of the upper mantle estimated from the composition of lherzolites. Contributions to Mineralogy and Petrology, 63, 161–173.
  • MacDonald, G. J. F. (1959). Calculations on the thermal history of the Earth. Journal of Geophyical Research, 64, 1967–2000.
  • Mann, U., Frost, D. J., & Rubie, D. C. (2009). Evidence for high-pressure core–mantle differentiation from the metal–silicate partitioning of lithophile and weakly siderophile elements. Geochimica et Cosmochimica Acta, 73, 7360–7386.
  • Mann, U., Frost, D. J., Rubie, D. C., Becker, H., & Audétat, A. (2012). Partitioning of Ru, Rh, Pd, Re, Ir and Pt between liquid metal and silicate at high pressures and high temperatures—Implications for the origin of highly siderophile element concentrations in the Earth’s mantle. Geochimica et Cosmochimica Acta, 84, 593–613.
  • Marvin, U. B. (1996). Ernst Florens Chladni (1756–1827) and the origins of modern meteorite research. Meteoritics and Planetary Science, 31, 545–588.
  • Mason, B. (1966). The enstatite chondrites. Geochimica et Cosmochimica Acta, 30, 25–39.
  • McDonough, W. F. (1990). Constraints on the composition of the continental lithospheric mantle. Earth and Planetary Science Letters, 101, 1–18.
  • McDonough, W. F. (2014). Compositional model for the Earth’s core. In H. D. Holland & K. K. Turekian (Eds.), The mantle and core: Treatise on geochemistry (2nd ed., vol. 3, pp. 559–576). Oxford, UK: Elsevier.
  • McDonough, W. F., & Sun, S.–S. (1995). The composition of the Earth. Chemical Geology, 120, 223–253.
  • McSween, H., Jr., & Huss, G. (2010). Cosmochemical models for the formation of the solar system. In Cosmochemistry (pp. 484–517). Cambridge, UK: Cambridge University Press.
  • Morbidelli, A., Libourel, G., Palme, H., Jacobson, S. A., & Rubie, D. C. (2020). Subsolar Al/Si and Mg/Si ratios of non-carbonaceous chondrites 2 reveal planetesimal formation during early condensation in the 3 protoplanetary disk. Earth and Planetary Science Letters, 538, 116220.
  • Newsom, H. E., & Palme, H. (1984). The depletion of siderophile elements in the Earth’s mantle: New evidence from molybdenum and tungsten. Earth and Planetary Science Letters, 69, 354–364.
  • Noddack, I., & Noddack, W. (1930). Die Häufigkeit der chemischen Elemente. Naturwiss, 18, 757–764.
  • O’Neill, H. St. C. (1991). The origin of the moon and the early history of the Earth—A chemical model. Part 2: The Earth. Geochimica et Cosmochimica Acta, 55, 1159–1172.
  • O’Neill, H. St. C., & Palme, H. (1998). Composition of the silicate Earth: Implications for accretion and core formation. In I. N. S. Jackson (Ed.), The Earth’s mantle: Composition, structure, and evolution (pp. 311–380). Cambridge, UK: Cambridge University Press.
  • O’Neill, H. St. C., & Palme, H. (2008). Collisional erosion and the nonchondritic composition of the terrestrial planets. Philosophical Transactions of the Royal Society A: Mathematical, Physical and Engineering Sciences, 366, 4205–4238.
  • Oeser, E. (1992). Historical earthquake theories from Aristotle to Kant. Historical Earthquakes in Central Europe, Vol. 1 (R. Gutdeutsch, G. Grünthal, & R. Musson, Eds.) (pp. 11–31). Vienna, Austria: Geologische Bundesanstalt.
  • Oldham, R. D. (1906). The constitution of the interior of the Earth, as revealed by earthquakes. Quarterly Journal of the Geological Society, 62, 456–475.
  • Pack, A., Russell, S. S., Shelley, J. M. G., & van Zuilen, M. (2007). Geo- and cosmochemistry of the twin elements yttrium and holmium. Geochimica et Cosmochimica Acta, 71, 4592–4608.
  • Palme, H., Larimer, J. W., & Lipschutz, M. E. (1988). Moderately volatile elements. In J. F. Kerridge & M. S. Matthews (Eds.), Meteorites and the early solar system (pp. 436–461). Tucson: University of Arizona Press.
  • Palme, H., Lodders, K., & Jones, A. (2014). Solar system abundances of the elements. In H. D. Holland & K. K. Turekian (Eds.), Treatise on geochemistry (2nd ed., vol. 2, pp. 15–36). Oxford, UK: Elsevier.
  • Palme, H., & Nickel, K. G. (1985). Ca/Al ratio and composition of the Earth’s upper mantle. Geochimica et Cosmochimica Acta, 49, 2123–2132.
  • Palme, H., & O’Neill, H. St. C. (2014). Cosmochemical estimates of mantle composition. In H. D. Holland & K. K. Turekian (Eds.), Treatise on geochemistry (2nd ed., vol. 3, pp. 1–39). Oxford, UK: Elsevier.
  • Palme, H., & Wlotzka, F. (1976). A metal particle from a Ca-Alrich inclusion from the meteorite Allende, and condensation of refractory siderophile elements. Earth and Planetary Science Letters, 33, 45–60.
  • Palme, H., & Zipfel, J. (2016). The Earth contains a large fraction of material not represented by meteorites. In Lunar and Planetary Science, XLVI, #2252. The Lunar and Planetary Institute, Houston (CD-ROM).
  • Palme, H., & Zipfel, J. (2017). The chemistry of solar system materials: Sun, planets, asteroids, meteorites and dust. In J. M. Trigo-Rodríguez, ‎M. Gritsevich, & ‎H. Palme (Eds.), Assessment and mitigation of asteroid impact hazards (pp. 33–53). New York, NY: Springer.
  • Payne, C. H. (1925). Stellar atmospheres: A contribution to the observational study of high temperature in the reversing layers of stars. In H. Shapley (Ed.), Harvard observatory monograph No. 1. The Observatory, Cambridge, MA.
  • Pearson, D. G., Canil, D., & Shirey, S. B. (2003). Mantle samples included in volcanic rocks: Xenoliths and diamonds. In R. W. Carson (Ed.), The mantle and core: Volume 2. Treatise in geochemistry (H. D. Holland & K. K. Turekian, Eds.) (pp. 172–276). Oxford, UK: Elsevier-Pergamon.
  • Poirier, J.­P. (1994). Light elements in the Earth’s outer core: A critical review. Physics of the Earth and Planetary Interiors, 85, 319–337.
  • Prior, G. T. (1916). On the genetic relationship and classification of meteorites. Mineralogical Magazine, 18, 26–43.
  • Prior, G. T. (1920). The classification of meteorites. Mineralogical Magazine, 19, 51–63.
  • Rammelsberg, C. (1872). Über die Meteoriten und ihre Beziehung zur Erde (p. 31). Berlin, Germany: C. B. Lüderitz’sche Verlagsbuchhandlung.
  • Rammensee, W., & Wänke, H. (1977). On the partition coefficient of tungsten between metal and silicate and its bearing on the origin of the moon. Proceedings of the 8th Lunar Science Conference, 2, 309–409.
  • Render, J., Fischer-Gödde, M., Burkhardt, C., & Kleine, T. (2017). The cosmic molybdenum-neodymium isotope correlation and the building material of the Earth. Geochemical Perspectives Letters, 3, 170–178.
  • Ringwood, A. E. (1961). Chemical and genetic relationships among meteorites. Geochimica et Cosmochimica Acta, 24, 159–197.
  • Ringwood, A. E. (1962a). A model for the upper mantle. Journal of Geophysical Research, 67, 857–867.
  • Ringwood, A. E. (1962b). A model for the upper mantle 2. Journal of Geophysical Research, 67, 4473–4477.
  • Ringwood, A. E. (1979). Origin of the Earth and Moon (p. 295). Springer-Verlag.
  • Ringwood, A. E. (1989). Significance of the terrestrial Mg/Si ratio. Earth and Planetary Science Letters, 95, 1–7.
  • Ringwood, A. E. (1991). Phase transformations and their bearing on the constitution and dynamics of the mantle. Geochimica et Cosmochimica Acta, 55, 2083–2210.
  • Ringwood, A. E., Kato, T., Hibberson, W., & Ware, N. (1990). High-pressure geochemistry of Cr, V and Mn and implications for the origin of the Moon. Nature, 347, 174–176.
  • Rubie, D. C., Frost, D. J., Mann, U., Asahara, Y., Nimmo, F., Tsuno, K., . . . Palme, H. (2011). Heterogeneous accretion, composition and core-mantle differentiation of the Earth. Earth and Planetary Science Letters, 301, 31–42.
  • Rubie, D. C., Jacobson, S. A., Morbidelli, A., O’Brien, D. P., Young, E. D., de Vries, J., . . . Frost, D. J. (2015). Accretion and differentiation of the terrestrial planets with implications for the compositions of early-formed Solar System bodies and accretion of water. Icarus, 248, 89–108.
  • Rudnick, R. L., & Gao, S. (2003). Composition of the continental crust. In R. K. Rudnick (Ed.), Treatise on geochemistry: Vol. 3. The crust (pp. 1–64). Oxford, UK: Elsevier-Pergamon.
  • Russell, H. N. (1941). The cosmical abundance of the elements. Science, 94, 375–381.
  • Sarbas, B., & Nohl, U. (2008). The GEOROC database as part of a growing geoinformatics network. In S. R. Brady, A. K. Sinha, & L. C. Gundersen (Eds.), Geoinformatics 2008—Data to knowledge, Proceedings: U.S. 95 Geological Survey Scientific Investigations Report 2008–5172 (p. 76).
  • Schaefer, L., & Fegley, B., Jr. (2010). Volatile element chemistry during metamorphism of ordinary chondritic material and some of its implications for the composition of asteroids. Icarus, 205, 483–496.
  • Schmidt, G., Palme, H., Kratz, K. L., & Kurat, G. (2000). Are highly siderophile elements (PGE, Re and Au) fractionated in the upper mantle? New results on peridotites from Zabargad. Chemical Geology, 163, 167–188.
  • Schmitt, W., Palme, H., & Wänke, H. (1989). Experimental determination of metal/silicate partition coefficients for P, Co, Ni, Cu, Ga, Ge, Mo, and W and some implications for the early evolution of the Earth. Geochimica et Cosmochimica Acta, 53, 173–185.
  • Scott, P., Grevesse, N., Asplund, M., Sauval, A. J., Lind, K., Takeda, Y., . . . Hayek, W. (2015). The elemental composition of the Sun. I. The intermediate mass elements Na to Ca. Astronomy and Astrophysics, 573, 1–19.
  • Siebert, J., Badro, J., Antonangeli, D., & Ryerson, F.J. (2012). Metal–silicate partitioning of Ni and Co in a deep magma ocean. Earth and Planetary Science Letters,321–322, 189–197.
  • Siebert, J., Badro, J., Antonangeli, D., & Ryerson, F.J. (2013). Terrestrial accretion under oxidizing conditions. Science, 339, 1194–1197.
  • Su, B., Chen, Y., Guo, S., & Liu, J. (2016). Origins of orogenic dunites: Petrology, geochemistry, and implications. Gondwana Research, 29, 41–59.
  • Turekian, K. K., & Clark, S. P. (1969). Inhomogeneous accretion of the Earth from the primitive solar nebula. Earth and Planetary Science Letters, 6, 346–348.
  • Urey, H. (1952). Chemical fractionation in the meteorites and the abundance of the elements. Geochimica et Cosmochimica Acta, 2, 269–282.
  • van der Hilst, R., Widiyantoro, S., & Engdahl, E. (1997). Evidence for deep mantle circulation from global tomography. Nature, 386, 578–584.
  • von Michaelis, H., Ahrens, L. H., & Willis, J. P. (1969). The composition of stony meteorites II. The analytical data and an assessment of their quality. Earth and Planetary Science Letters, 5, 387–394.
  • Wade, J., & Wood, B. J. (2005). Core formation and the oxidation state of the Earth. Earth and Planetary Science Letters, 236, 78–95.
  • Wang, H. S., Lineweaver, C. H., & Ireland, T. (2018). The elemental abundances (with uncertainties) of the most Earth-like planet. Icarus, 299, 460–474.
  • Wänke, H. (1981). Constitution of terrestrial planets. Philosophical Transactions of the Royal Society, Series A, 303, 287–302.
  • Wänke, H., Baddenhausen, H., Dreibus, G., Jagoutz, E., Kruse, H., Palme, H., . . . Teschke, F. (1973). Multielement analyses of Apollo 15, 16, and 17 samples and the bulk composition of the moon. Proceedings of the Fourth Lunar Science Conference, 2, 1461–1481.
  • Wänke, H., Dreibus, G., & Jagoutz, E. (1984). Mantle chemistry and accretion history of the Earth. In A. Kröner, G. Hanson, & G. Goodwin (Eds.), Archaean geochemistry (pp. 1–24). Berlin, Germany: Springer.
  • Warren, P. H. (2011). Stable-isotopic anomalies and the accretionary assemblage of the Earth and Mars: A subordinate role for carbonaceous chondrites. Earth and Planetary Science Letters, 311, 93–100.
  • Washington, H. S. (1925). The chemical composition of the Earth. American Journal of Science, Series 5, 9, 351–378.
  • Wasserburg, G. J., MacDonald, G. J. F., Hoyle, F., & Fowler, W. A. (1964). Relative contributions of Uranium, Thorium, and Potassium to heat production in the Earth. Science, 143, 465–467.
  • Wasson, J. T., & Kallemeyn, G. W. (1988). Compositions of chondrites. Phiosphical Transactions Royal Society of London, A325, 535–544.
  • Weisberg, M. K., Ebel, D. S., Nakashima, D., Kita, N. T., & Humayun, M. (2015). Petrology and geochemistry of chondrules and metal in NWA 5492 and GRO 95551: A new type of metal-rich chondrite. Geochimica et Cosmochimica Acta, 167, 269–285.
  • Witt-Eickschen, G., Palme, H., O’Neill, H. St. C., & Allen, C. M. (2009). The geochemistry of the volatile trace elements As, Cd, Ga, In and Sn in the Earth’s mantle: New evidence from in situ analyses of mantle xenoliths. Geochimica Cosmochimica Acta, 73, 1755–1778.
  • Wolf, D., & Palme, H. (2001). The solar system abundances of P and Ti and the nebular volatility of P. Meteoritics and Planetary Science, 36, 559–572.
  • Wood, B. J., Smythe, D. J., & Harrison, T. (2019). The condensation temperatures of the elements: A reappraisal. American Mineralogist, 104, 844–856.
  • Wood, J. A., & Morfill, G. E. (1988). A review of solar nebula models. In J. F. Kerridge & M. S. Matthews (Eds.), Meteorites and the solar system (pp. 329–347). Tucson: University of Arizona Press.
  • Yin, Q. (2005). From dust to planets: The tale told by moderately volatile elements. In A. N. Krot, E. R. D. Scott, & B. Reipurth (Eds.), Chondrites and the protoplanetary disk Astronomical Society of the Pacific Conference series (Vol. 341, pp.632–644). San Francisco.
  • Zhang, J., Dauphas, N., Davis, A. M., Leya, I., & Fedkin, A. (2012). The proto-Earth as a significant source of lunar material. Nature Geoscience, 5, 251–255.
  • Zinner, E. K., Moynier, F., & Stroud, R. M. (2011). Laboratory technology and cosmochemistry. Proceedings of the National Academy of Sciences of the United States of America, 108, 19135–19141.