EARTH'S CONTINENTAL CRUST 145

the sunlight it receives. Many atmospheric constituents can Helium isotopes

absorb portions of this low energy radiation - these, in recent Hydrologic cycle
Organics: contemporary degradation and preservation
times, have been called the 'greenhouse gases'. Water vapor Natural gas
and C0 2 are the most important such gases, but 0 3 , CH 4 , and Nitrogen cycle
N 20 also contribute substantially to the greenhouse effect. The Oxygen
man-made chlorofluorocarbons CC1 3 F (F-11) and CC1 2 F 2 Precambrian atmosphere
(F-12) are among the most potent greenhouse gases known, Volcanic gases
but their concentrations are so small that at present levels their
greenhouse effect is negligibly small. All the greenhouse gases
are present at very low concentrations relative to the major EARTH'S CONTINENTAL CRUST
constituents of the atmosphere (Table El ). After a layer of the
atmosphere absorbs this heat from the Earth, it radiates it Introduction
back, half outward and half back to the Earth's surface. So at
the Earth's surface we receive this additional energy from the The continental crust constitutes only 0.40% of the Earth's
atmosphere, that makes the surface warmer than it would be mass (0.62% of the crust-mantle system) but has a geochemical
if there were no gases that absorbed infrared radiation - this significance of far greater importance than such trivial numbers
is the greenhouse effect. would indicate. For elements most incompatible with mantle
or core mineralogy, such as Cs, Rb, K, U and Th, the continen-
tal crust contains in excess of one-third of the Earth's budget.
Anthropogenic perturbations of atmospheric Accordingly, the continental crust is an extremely important
composition geochemical reservoir; understanding its chemical composition
Human activities have always had an influence on the Earth's and chemical evolution are first order problems for geochemis-
atmosphere. In the past however, this influence was limited to try. Recent reviews relevant to the geochemistry and geochemi-
affecting naturally occurring gases, and that too was not sig- cal evolution of the continental crust may be found in Taylor
nificant on the global scale. In the last century, human activities and McLennan (1985, 1995, 1996), McLennan (1988, 1989),
have altered the composition of the atmosphere, both by Condie (1992), McLennan and Taylor (1996), Rudnick and
adding to the naturally occurring atmospheric constituents and Fountain (1995), Christensen and Mooney (1995) and
by introducing entirely new gases into the atmosphere which Wedepohl (1995).
have no natural sources. First, many of the gases added, C0 2 ,
CH 4 , and N 2 0 for example, can cause global warming by Physical characteristics of the continental crust
increasing the greenhouse effect. Second, exotic gases such as The continental crust is the exterior layer of the Earth that lies
the chlorofluorocarbons mentioned above can cause a deple- above the Mohorovicic seismic discontinuity, where compres-
tion of the stratospheric ozone layer. Ozone, as mentioned sional seismic wave velocities (Vp) increase abruptly downward
earlier, absorbs high energy solar radiation in the ultraviolet from about 7-8 km/s. It comprises low-density, geochemically
part of the sun's spectrum. A thinner ozone layer means more and mineralogically differentiated rocks with relatively high
ultraviolet radiation reaching the surface, where it can be altitude, thus distinguishing it from the low-lying more mafic
harmful to human health and the environment. These global oceanic crust that dominates the ocean basins. The continental
environmental problems have demonstrated that atmospheric crust includes most of the ten major continents and several
constituents present in very minute amounts (trace gases) can submerged microcontinents and is defined seismically in these
have profound influences on the environment. It is partly crustal blocks as the uppermost layer of the Earth with
because these gases are so rare that human activities can affect VP ~ 7.8 km/s (average of 6.45 km/s) and shear wave velocities
their concentrations so significantly. CVs) ~ 4.3 km/s (average of 3.65 km/s).
Primarily on the basis of seismic velocity data, it is possible
M.A.K. Khalil to model the density profile of the continental crust. The base
of the crust typically has a density of approximately 3.1 g/cm 3
Bibliography which generally diminishes towards the surface to about
2.6-2.8 g/cm 3 . Although the heterogeneity of the continental
Barry, R.G. and Chorley, R.J. (1992) Atmmphere, Weather and
Climate. 6th edn. New York: Routledge. crust near the surface is well known, the crust is probably very
Charlson, R.J. (1992) The Atmosphere, in Global Biogeochemical complicated throughout. At the base, high velocity mantle-like
Cycles (eds Butcher et al.). New York: Academic Press. rocks and low velocity continental rocks are probably interlay-
Intergovernmental Panel on Climate Change. 1990. Climate Change. ered. Accretion of mantle-derived ultramafic to mafic magmas
New York: Cambridge University Press. at the base of the continental crust (underplating) is also
Khalil, M.A.K. (1995) Greenhouse gases in the Earth's atmosphere, in
thought to be an important process at least locally, further
Encyclopedia of Environmental Biology, Vol. 2 (ed. W.A.
Nirenberg). New York: Academic Press. complicating the density structure of the lowermost crust. The
Wuebbles, D.J. and Edmonds, J. (1991) Primer on Greenhouse Gases. average density of the continental crust recently has been esti-
Chelsea, MI: Lewis Publishers. mated to be 2.83 g/cm 3 (Christensen and Mooney, 1995).
The continental crust has an area of 2.10 x 10 8 km 2 (41.2%
Cross-references of the Earth's surface) 71.3% of which lies above sea level. The
thickness of the continental crust is highly variable, ranging
Biogeochemistry
Carbon cycle
from about 10 to > 70 km, making estimates of an average
Elements: distribution difficult. Over the past decade or so, estimates of mean conti-
Elements: noble gases nental thickness have varied by about 15%, the most recent
Geochemistry: low temperature being 41 km (Christensen and Mooney, 1995), suggesting a
146 EARTH'S CONTINENTAL CRUST

volume of 8.61 x 10 9 km 3 and, for a density of 2.83 g/cm 3 , a Table E2 Some physical characteristics of the continental crust
mass of 2.44 x 10 25 g.
Area 2.10 X 10 8 km 2
Heat flow from the continental crust is an important parame-
Fraction of Earth's area 0.412
ter for constraining the abundance and the vertical distribution
Mean thickness 41 km
of heat producing elements (K, Th and U) through the crust Range of thicknesses 10-75 km
(e.g. McLennan and Taylor, 1996). Continental heat flow Average thickness at sea level 28 km
averages 65 ± 1.6 mW/m 2 (Pollack et a!., 1993) resulting in a Volume 8.6] X 109 km 3
heat loss through the continents of about 13.7 x 10 12 W. A Mean density 2.83 gjcm3
significant amount of continental heat is associated with rela- Mass 2.44 X 1025 g
tively young orogenic activity. This contribution to heat decays Fraction of Earth's mass 0.0040
over several hundred million years to an equilibrium value Fraction of Earth's mass, excluding core 0.0062
which is controlled by crustal radiogenic heat ( 60%) and the Cunent growth rate 2.0 ± 1.0 x 10 15 gjyear
heat flux from the underlying mantle or 'reduced heat flow' Growth rate over past 2.5 Ga 3.0 ± 1.0 x 10 15 gjyear
(40%). Between Archean and younger terranes, there appears Growth rate over past 4.0 Ga 6.1 x 10 15 gjyear
to be a rather abrupt offset in average continental heat flow Fraction of area above mean sea level 0.713
from tectonically stabilized terranes. Average heat flow Mean elevation 125m
is 41 ± 1 mW/m 2 for stabilized Archean terranes and Mean elevation lying above 200m isobath 690m
48 ± 1 mW/m 2 for stabilized continental crust in general Mean heat flow 65mW/m2
(Nyblade and Pollack, 1993). This offset is due to some combi- Total heat loss 1.37 X 10 13 W
nation of Archean crust being of different composition (more Average heat production 0.70 11W/m3
mafic) and the fact that subcontinental lithosphere is thicker Temperature at base of shields 350--400'C
beneath Archean cratons resulting in thermal resistance or Temperature at base of orogenic zones 500-600"C
Temperature at base of rift zones 650-750°C
deflection of heat.
Average near surface geothennal gradient 30°Cjkm
The continental crust, because of its great antiquity, provides Range of near surface geothermal gradients I0-50°C/km
the primary record of most of the Earth's geologic history. In
contrast to oceanic crust, all of which has an age less than Mean compressional wave seismic velocity (Vp) 6.45 kmjs
Mean shear wave seismic velocity (V,) 3.65 kn1js
0.20 Ga, continental crust has a mean age of about 2.5 Ga and
Mean Poisson's ratio (a-) 0.27
contains fragments in excess of 4.0 Ga. Thus, continental crust
evolves on time scales of 10 9 years in contrast to oceanic crust Mean age (age of Earth is 4.56 Ga) 2.5 ±0.5 Ga
which evolves on time scales of 10 7 -10 8 years. The Earth's Range of ages 0.00-3.96 Ga
continental crust is probably the only example within the solar
Continents and microcontinents (Area [10 6 km 2 ], mean thickness [km]): Eurasia
system of an ancient, multiply differentiated ('tertiary') crust. (62.160, 38), N. America (39.792, 35), Africa (35.722, 24), S. America (23.982,
This is due to a combination of plate tectonic processes, allow- 42), Antarctica (18.984, 33), Australia (14.163, 36), India (4.648, 38), Arabia
ing for the rapid recycling of the mantle and ocean crust and (4.595, 23), New Zealand (3.980, 21), Central America (1.344, 18), Rockall
(0.459, 33), Seychelles (0.386, 32), Agulhas (0.134, 12), Jan Mayen (0.055, -).
the planet's huge volumes of free water influencing the condi-
tions of melting within the Earth. Data sources: Cogley, 1984; Taylor and McLennan, 1985; Pollack eta/., 1993;
Christensen and Mooney, 1995; Zandt and Ammon, 1995; McLennan and
A summary of the physical characteristics of the continental Taylor, 1996.
crust is given in Table E2.

Origin of the continental crust both within orogenic zones and elsewhere. The overall signifi-
cance of such processes represents a major uncertainty in
Estimates of the composition and chemical evolution of the models of crustal composition, both because the mass fraction
continental crust are dependent on models for its origin. This is not known and because the primary compositional relation-
is because it is not possible to sample the deep crust in a ships of such rocks are not fully understood. Other minor
representative and precise manner, and constraints from more sources of continental crust include obduction of oceanic crust
indirect geophysical data (e.g. heat flow and seismic measure- (ophiolites) and mantle derived magmas produced at continen-
ments) are themselves model-dependent to varying degrees. tal rifts.
Large scale sampling programs are restricted to the near sur- Although the processes described above adequately explain
face and thus have been successful in providing information the production of continental crust during the past 2.0-2.5
only on the upper continental crust. billion years, there is good reason to believe that well over
The most important mechanism for the formation of conti- 50% of the continental crust accreted prior to that time (i.e.
nental crust is the accretion of magmatic rocks derived from during the Archean), and was produced by somewhat different
the upper mantle at subduction zones. Mantle melting is initi- mechanisms. The Earth has cooled over geologic time and the
ated by fluid released from the downgoing slab during subduc- mantle was significantly hotter early in its history. A likely
tion. Accordingly, new crust results from volcanic and plutonic consequence is that at Archean subduction zones mantle melt-
rocks, formed at island arcs and continental arcs, and much ing took place at higher temperatures and lower pressures,
of the intrusive rock that is underplated onto the crust at the resulting in significantly different igneous compositions. It is
base of arcs. A fraction of continental crust ( < 5-10%) is also possible that subducting oceanic lithosphere itself may have
produced by intraplate volcanism associated with the accretion partially melted and that such melts contributed to new
of oceanic islands and intrusion of mantle-derived magmas Archean continental crust. Although our understanding of
within continental blocks. Intrusion of mafic to ultramafic Archean igneous rocks is at an early stage, abundant geochemi-
magmas at the base of the crust (underplating) takes place cal features are consistent with such scenarios, and must be
 EARTH'S CONTINENTAL CRUST 147

incorporated into estimates of average crustal composition (see granulite terranes may not be fully representative of the
Precambrian geochemistry). lowermost continental crust.
Seismic models, the most commonly employed technique for
Composition of the continental crust estimating crustal compositions, are based on seismic refrac-
tion profiles through the continental crust. Lithological propor-
Bulk crust tions are estimated based on laboratory studies of seismic
Estimating the chemical composition of the continental crust velocities in rocks and minerals, and the chemical composition
can lead to very different results (Figure El). Tectonic models is determined from analyses of representative rock types.
include identifying the plate tectonic process responsible for Difficulties with this approach include possible non-uniqueness
the formation of continents and assigning the composition of of lithological interpretations of seismic velocity profiles.
the average or most common lithology within that environ- The most commonly cited model of crustal composition
ment. The best known of these is the Andesite model (see Taylor incorporates aspects of heat flow data, the Andesite model and
and McLennan, 1985), but this has been questioned as being continental growth history, such that Archean crustal composi-
inadequate to explain the detailed petrology, geochemistry and tions are estimated independently (Taylor and McLennan,
age distribution of the continents. A recent model proposes 1985, 1995). Archean and post-Archean crustal composition
that the continents grow mainly through the accretion of ocean estimates are determined independently from various con-
islands (Abbott and Mooney, 1995) and would be basaltic in straints and then are 'mixed' in proportions determined from
character. Basaltic compositions for the continental crust isotopic and other constraints on crustal growth history. This
encounter several geochemical difficulties, not least being an composition is presented in Table E3.
inability to generate the well constrained granodioritic com-
position for the upper 25-35% of the crust without appealing Upper crust
to ad hoc mechanisms.
Geologic models rely on mapped geologic distributions of Only the uppermost part of the continental crust is exposed
lithologies in exposures oflow through highest grade metamor- and accessible to sampling. Accordingly, upper crustal com-
phic rocks and chemical analyses of representative examples. positions are best constrained and also provide considerable
Such models lead to the most felsic compositions, as extreme insight into the processes that ultimately differentiate the conti-
as dacite. In turn, these models are the least reliable because nental crust. Three basic techniques have been employed in
they depend solely on exposed metamorphic terranes for esti- estimating upper crustal abundances, including: (a) estimating
mates of mid- to lower crustal abundances. It is now well the mass balance among various igneous rock types exposed
known that highly metamorphosed crustal xenoliths, an addi- at the surface of the continental crust; (b) direct measurement
tional probe to lower crustal lithologies, tend to show systemat- from large scale sampling programs, in regions of stabilized
ically more mafic compositions than do rocks from exposed crust; (c) average sediment compositions to estimate relatively
high grade terranes (Rudnick and Fountain, 1995) and so 'immobile' elements. The agreement among these various
methods is remarkable. A general consensus regarding the
average composition of the upper continental crust approxi-
mates the igneous rock type granodiorite (Table E3 ). Such a
3.0 composition is likely representative of about the upper
{? Tectonic 10-15 km or 25-35% of the entire crust.
2.5 ~ Geological The use of sedimentary compositions to estimate upper

•
crustal abundances provides a mechanism for tracing the evolu-
EB Seismic
tion of upper crust over geologic time. A number of elements,
2.0 most importantly Th and rare earth elements (REE) La-Lu,
Y, and Sc, are present at extremely low concentration in most
natural waters and thus are carried almost exclusively in sands
1 1.5
and muds during sedimentary transport. Since shales are the
most abundant sedimentary rock type, average shale may be
0C\J 1.0 equated to the average upper crust for these elements, with
~ only minor corrections. Although only a few elements are
represented, it is possible to extend the list using well estab-
0.5 lished relationships among elements (Th/U = 3.8, K/U = 10 4 ,
VOLCANIC ROCKS
LOW-K CALC-ALKALINE K/Rb vs. K) in crustal rocks. Thus, upper crustal compositions
0. 0 +-..--,-......-.-.-.,....,-..,...--.--.-..--,--.---.--.-...---.-+ from sediment data compare extremely well to those deter-
48 50 52 54 56 58 60 62 64 66 mined from the sampling programs (Table E4).
Most shales have remarkably similar REE patterns
Si0 2 (wt. %) (Figure E2), generally accepted as being parallel to the average
upper crustal pattern. Relative to chondri tic meteorites (repre-
Figure E1 Estimates of the bulk composition of the continental crust sentative of the bulk Earth for these elements), the upper crust
on a plot of K2 0 vs Si0 2 . Superimposed is the geochemical
is enriched in the light REE, La-Sm, consistent with overall
classification of calc-alkaline volcanic rocks. Crustal composition
models taken from Weaver and Tarney, 1984; Shaw eta/., 1986
enrichment of incompatible elements in the crust, particularly
(geologic models); Wedepohl, 1995; Rudnick and Fountain, 1995; the upper crust. The pattern is also anomalously depleted in
Christensen and Mooney, 1995 (seismic models); Taylor and europium. Eu-depletion results from fractionation of plagio-
Mclennan, 1985; Abbott and Mooney, 1995; Mclennan and Taylor, clase, a mineral that is strongly enriched in Eu, during partial
1996 (tectonic models, Taylor and Mclennan model). melting and fractional crystallization. A complementary
148 EARTH'S CONTINENTAL CRUST

Table E3 Chemical composition of the continental crust

Upper Bulk Lower Average Lower Archean

continental continental continental andesitic crustal bulk
Element crust crust crust crust granulites crust

Li (ppm) 20 13 11 10 6
Be (ppm) 3.0 1.5 1.0 1.5
B (ppm) 15 10 8.3
Na (wt%) 2.89 2.30 2.08 2.60 1.9 2.23
Mg (wt%) 1.33 3.20 3.80 2.11 4.3 3.56
AI (wt%) 8.04 8.41 8.52 9.50 8.78 8.04
Si (wt 0lr•) 30.8 26.8 25.4 27.1 24.4 26.6
p (ppm) 700 400
K (wt%) 2.80 0.91 0.28 1.25 0.50 0.75
Ca (wt%) 3.00 5.29 6.07 5.36 6.7 5.22
Sc (ppm) 11 30 36 30 31 30
Ti (wt%) 0.30 0.54 0.60 0.48 0.5 0.60
v (ppm) 60 230 285 175 195 245
Cr (ppm) 35 185 235 55 215 230
Mn (ppm) 600 1400 1700 llOO 800 1500
Fe (wt'/o) 3.50 7.07 8.24 5.83 6.5 7.46
Co (ppm) 10 29 35 25 38 30
Ni (ppm) 20 105 135 30 88 130
Cu (ppm) 25 75 90 60 26 80
Zn (ppm) 71 80 83 78
Ga (ppm) 17 18 18 18 13
Ge (ppm) 1.6 1.6 1.6
As (ppm) 1.5 1.0 0.8
Se (ppm) 50 50 50
Rb (ppm) 112 32 5.3 42 11 28
Sr (ppm) 350 260 230 400 348 215
y (ppm) 22 20 19 22 16 19
Zr (ppm) 190 100 70 100 68 100
Nb (ppm) 25 II 6 ll 5
Mo (ppm) 1.5 1.0 0.8
Pd (ppb) 0.5 I I
Ag (ppb) 50 80 90
Cd (ppb) 98 98 98
In (ppb) 50 50 50
Sn (ppm) 5.5 2.5 1.5
Sb (ppm) 0.2 0.2 0.2
Cs (ppm) 3.7 1.0 0.1 1.7 0.3
Ba (ppm) 550 250 150 350 259 220
La (ppm) 30 16 II 19 8 15
Ce (ppm) 64 33 23 38 20
Pr (ppm) 7.1 3.9 2.8 4.3 2.6 3.7
Nd (ppm) 26 16 12.7 16 II 16
Sm (ppm) 4.5 3.5 3.17 3.7 2.8 3.4
Eu (ppm) 0.88 1.1 1.17 1.1 1.1 1.1
Gd (ppm) 3.8 3.3 3.13 3.6 3.1 3.2
Tb (ppm) 0.64 0.60 0.59 0.64 0.48 0.59
Dy (ppm) 3.5 3.7 3.6 3.7 3.1 3.6
Ho (ppm) 0.80 0.78 0.77 0.82 0.68 0.77
Er (ppm) 2.3 2.2 2.2 2.3 1.9 2.2
Tm (ppm) 0.33 0.32 0.32 0.432 0.32
Yb (ppm) 2.2 2.2 2.2 2.2 1.5 2.2
Lu (ppm) 0.32 0.30 0.29 0.30 0.25 0.33
Hf (ppm) 5.8 3.0 2.1 3.0 1.9 3
Ta (ppm) 2.2 1.0 0.6 0.6
w (ppm) 2.0 1.0 0.7
Re (ppb) 0.4 0.4 0.4
Os (ppb) 0.05 0.05 0.05
Ir (ppb) 0.02 0.1 0.13
Au (ppb) 1.8 3.0 3.4
Tl (ppb) 750 360 230
Pb (ppm) 20 8.0 4.0 10 4.2
Bi (ppb) 127 60 38
Th (ppm) 10.7 3.5 1.06 2.5 1.2 2.9
u (ppm) 2.8 0.91 0.28 1.0 0.2 0.75

Adapted from Taylor and McLennan, 1995 with minor alterations. Values for average lower crustal granulites from Rudnick and Fountain, 1995.
 EARTH'S CONTINENTAL CRUST 149

Table E4 Comparison of upper crust composition determined from An intriguing observation is that large regions, representing
sedimentary rock data and from direct large scale sampling residues of partial melting, though predicted in many models
measurements of crustal origin and differentiation, have not been observed.
This has not yet been reconciled with evidence that intracrustal
From Basis of Direct
sediment estimate measure
partial melting is a dominant mechanism in generating upper
crustal compositions. In Table E3, an estimate of average lower
La (ppm) 30 0.8* Shale 32 crustal composition is given, based on the assumption that the
Sm (ppm) 4.5 0.8*Shale 4.5 upper crust represents 25% of the entire crust. Also shown for
Yb (ppm) 2.2 0.8* Shale 1.5 comparison is the average of available lower crustal rocks
Sc (ppm) II 0.8*Shale 10 at the granulite grade of metamorphism (Rudnick and
Th (ppm) 10.7 0.8*Shale 10.5 Fountain, 1995).
u (ppm) 2.8 Th/U = 3.8 2.5
K 20 (wt%) 3.4 K/U = 10,000 3.30
Rb (ppm) 112 K/Rb = 250 110 Mineralogy of the crust
The average mineralogical composition of the continental crust
Adapted from Taylor and McLennan, 1985; McLennan, 1989.
is a difficult parameter to estimate and, due to sampling restric-
tions, is limited to the upper crust. In Table ES, two estimates
of average upper crustal mineralogy are tabulated. The first is
an estimate, based on a ratio of metamorphic to plutonic rocks
of 1.9. The second estimate is essentially representative of only
(J) exposed surface rocks and includes a significant proportion of
2 glass, preserved in extrusive rocks. Also shown in Table E5 are
'§ 100 the calculated normative mineralogies for the crustal composi-
<ll
a; tions listed in Table E3. An important feature is that the lower
~ crust has little normative quartz, on average, and abundant
normative plagioclase. This is in contrast with a common
•,
0
:;:::; ''
·;:: previously held view that the lower crust is quartzo-feldspathic

.
'0
c '' in nature. Instead, the lower crust on average is basaltic in
£
0
'' character.

. •- ... ··-..... -·- ·-..

()
E Upper \ ¥..~..;;:,;::--­
a..
a.. 10
Crust ..
Table ES Mineralogical composition (%) of the continental crust
E
a..
a.. UPPER CRUST MINERALOGY
Upper Exposed
crust 1 crust 2

LaCe Pr Nd Sm Eu Gd Tb Dy Ho Er Tm Yb Lu Quartz 23.2 20.3

Plagioclase 39.9 34.9
Figure E2 Chrondrite normalized rare earth element diagram Glass 0.0 12.5
showing the compositions of various shale composites and the Orthoclase 12.9 11.3
average composition of loess deposits. Data are from Taylor eta/., Biotite 8.7 7.6
1983; Gromet eta/., 1984; Taylor and Mclennan, 1985; Goldstein Muscovite 5.0 4.4
and jacobsen, 1988; Mclennan, 1989. Also shown is the estimate Chlorite 2.2 1.9
of the average rare earth element abundances for the upper Amphibole 2.1 1.8
continental crust from Taylor and Mclennan, 1985; see Table E3. Pyroxene 1.4 1.2
Olivene 0.2 0.2
Oxides 1.6 1.4
En-enrichment is predicted for the lower crust. Since plagio- Other 3.0 2.6
clase is stable only up to about 10 kbar of pressure ( 40 km Total 100.2 100.1
depth on Earth), the fractionation of Eu is a relatively shallow
process and the differentiation of the crust is thought to be NORMATIVE MINERALOGY
dominated by intracrustal igneous and metamorphic processes. Upper Bulk Lower
crust crust crust
Lower crust Quartz (Q) 16.1 6.6 3.7
Lower crustal samples are preserved as high grade amphibolite Orthoclase (Or) 20.1 6.5 2.0
and granulite exposures or lower crustal xenoliths carried to Albite (Ab) 33.0 26.2 23.7
the surface in volcanic pipes. However, it is not clear that Anorthite (An) 13.9 26.2 30.4
either provides an unbiased sampling of the lower crust because Diopside (Di) 5.8 8.7 9.8
Hypersthene (Hy) 10.0 24.1 28.6
special tectonic conditions are responsible for their presence
Ilmenite (II) 1.0 1.7 1.9
at the surface and because these independent 'samples' provide Total 99.9 100.0 100.1
quite different results. In spite of any shortcomings, it is clear
that studies of such rocks indicate that the lower continental Upper crust mineralogy from Nesbitt and Young, 1984; 'assumes
crust is both lithologically and compositionally complex. metamorphic/intrusive rock ratio of 1.9; 2 represents exposed crust surface.
150 EARTH'S CONTINENTAL CRUST

Evolution of the continental crust that is added in each episode appears to diminish over time
(probably correlating with the cooling of the Earth) such that
A great deal of controversy exists regarding the evolution of
the period of growth at about 2.0-1. 7 Ga is the second most
the continental crust over the history of the Earth. A major
important period of crustal growth after that of the Late
source of this controversy deals with uncertainty in the amount
Archean. There is no evidence for changing crustal composi-
of continental crust that may be recycled back into the mantle,
tions or changes in the fundamental processes giving rise to
mainly in the form of subducted sediment, and how this value
continental crust or causing differentiation of the crust since
changes over time (McLennan, 1988; Armstrong, 1991).
the Archean.
Depending on the exact nature of continental crust-mantle
recycling, the mass of the continents may grow continuously
Scott M. McLennan and Stuart Ross Taylor
or episodically over geologic time or may even remain in
steady state.
Shortly after early core formation, the Earth was struck by Bibliography
a giant planetesimal at least the size of Mars, at about 4.45 Ga, Abbott, D. and Mooney, W. ( 1995) The structural and geochemical
which probably resulted in whole Earth melting. As the Earth evolution of the continental crust: Support for the oceanic plateau
cooled and began to solidify, it is unlikely that a lithosphere model of continental growth. National Report to I.U.G.G. Rev.
stabilized to the point of sustaining any form of crust much Geophys. Supp., 231-42.
Armstrong, R.L. ( 1991) The persistent myth of crustal growth. Aust.
before 4.3-4.4 Ga. The earliest formed crusts were probably J. Earth Sci., 38, 613-30.
of ultramafic to mafic in composition, although there is some Christensen, N.J. and Mooney, W.D. ( 1995) Seismic velocity structure
evidence that regions of this crust may have been enriched in and composition of the continental crust: a global view. J. Geophys.
the most incompatible elements. The occurrence of 4.2 Ga Res., 100, 9761-88.
zircons as detrital components in younger sedimentary rocks Cogley, J.G. (1984) Continental margins and the extent and number of
indicates that at least minor amounts of relatively felsic crust continents. Rev. Geophys., 22, 101-22.
Condie, K.C. ( 1992) Chemical composition and evolution of the upper
(continental affinity?) had formed by this time. The oldest continental crust: contrasting results from surface samples and
preserved rocks of clear continental affinity are small exposures shales. Chon. Geol., 104, 1-37.
of highly metamorphosed tonalitic and granodioritic rocks Goldstein, S.J. and Jacobsen, S.B. (1988) Rare earth elements in river
(now high grade gneisses), in the range 3.95-3.7 Ga, found in waters. Earth Planet. Sci. Lett., 89, 35-47.
various shield areas around the world. Such areas probably Gromet, I.P., Dymek, R.F., Haskin, L.A. and Korotev, R.L. (1984)
The 'North American shale composite': its compilation, major and
constituted no more than several percent of the Earth's surface, trace element characteristics. Geochim. Cosmochim. Acta, 48,
and all crust formed prior to about 3.8 Ga was massively 2469-82.
disturbed by intense meteorite bombardment that was endemic McLennan, S.M. (1988) Recycling of the continental crust. Pure Appl.
in the early solar system. Geophys., 128, 683-724.
Crustal growth was especially rapid during the latest McLennan, S.M. (1989) Rare earth elements in sedimentary rocks:
influence of provenance and sedimentary processes. Rev. Mineral.,
Archean (2.8-2.6 Ga); 50-60% of the continental crust was in 21, 169-200.
place by 2.5 Ga. Continental crust formed during this time McLennan, S.M. and Taylor, S.R. (1996) Heat flow and the chemical
was probably of different bulk composition to the more composition of continental crust. J. Geo/., 104, 369-77.
recently formed crust on account of higher temperatures on Nesbitt, H.W. and Young, G.M. (1984) Prediction of some weathering
the early Earth. It is likely that melting within the mantle to trends of plutonic and volcanic rocks based on thermodynamic and
kinetic considerations. Geochim. Cosmochim. Acta, 48, 1523-34.
form continental crust took place at higher temperatures and Nyblade, A.A. and Pollack, H.N. ( 1993) A global analysis of heat flow
lower pressures (depth), thus affecting phase relations. In addi- from Precambrian terrains: implications for the thermal structure
tion to mantle melting induced by fluid release from subducting of Archean and Proterozoic lithosphere. J. Geophys. Res., 98,
slabs, melting of the subducting lithosphere itself may also 12207-18.
have contributed to the formation of new continental crust Pollack, H.N., Hurter, S.J. and Johnson, J. R. (1993) Heat flow from
during the Archean. Many detailed geochemical relationships the earth's interior: analysis of the global data set. Rev. Geophys.,
31, 267-80.
were affected but, in general, Archean crust is somewhat less Rudnick, R.L. and Fountain, D.M. ( 1995) The lower crust. Rev.
enriched in incompatible elements than Proterozoic and Geophys., 33, 267-309.
Phanerozoic crust. Shaw, D.M., Cramer, J.J., Higgins, M.D. and Truscott, M.G. (1986)
A critical difference between Archean and younger continen- Composition of the Canadian Precambrian shield and the continen-
tal crust is that the Archean upper crust (as sampled by sedi- tal crust of the earth. Geol. Soc. Land. Special Pub/., 24, 275-82.
Taylor, S.R. and McLennan, S.M. ( 1985) The Continental Crust: Its
mentary rocks) is far less depleted in the rare earth element Composition and Evolution. Oxford: Blackwell Scientific
Eu. Archean sedimentary rocks with the characteristic Publishers, 312 pp.
En-depletion seen in younger sedimentary rocks appears to be Taylor, S.R. and McLennan, S.M. ( 1995) The geochemical evolution
the exception. This indicates that the Archean crust was not of the continental crust. Rev. Geophys., 33, 241-65.
differentiated into upper and lower crust in the same manner Taylor, S.R. and McLennan, S.M. ( 1996) The evolution of continental
crust. Sci. Am., 274, 60-5.
as during the post-Archean, where intracrustal melting leads Taylor, S.R., McLennan, S.M. and McCulloch, M.T. (1983)
to strong vertical variations in chemistry. Accordingly, the Geochemistry of loess, continental crustal composition and crustal
Archean/Proterozoic transition is a first order benchmark in model ages. Geochim. Cosmochim. Acta, 47, 1897-905.
the chemical evolution of the continental crust. Weaver, B.L. and Tarney, J. (1984) Empirical approach to estimating
Throughout the post-Archean, the crust appears to have the composition of the continental crust. Nature, 310, 575-7.
grown in an episodic fashion, probably related to the cycles of Wedepohl, K.H. (1995) The composition of the continental crust.
Geochim. Cosmochim. Acta, 59, 1217-32.
accumulation and dispersion of supercontinents. Thus, addi- Zandt, G. and Ammon, C.J. (1995) Continental crust composition
tional major episodes of crustal growth are seen at about constrained by measurements of crustal Poisson's ratio. Nature,
2.0-1.7 Ga, 1.3-1.1 Ga and 0.5-0.3 Ga. The amount of crust 374, 152-4.
 EARTH'S CORE 151

Cross-references by mass in the terrestrial planets, leads one to conclude that

the Earth's core is dominantly composed of iron.
Carbonate sediments
Earth's formation and geochemical evolution The coefficient of the moment of inertia for the Earth is a
Earth's mantle geochemistry number that depends principally on the distribution of mass
Earth's oceanic crust within the planet with respect to its axis of rotation. The Earth
Elements: alkali and alkaline earth possesses an equatorial bulge and shows flattening at the poles.
Elements: distribution The rotational flattening, combined with the precession of the
Elements: halogens
Elements: trace equinoxes, establishes the value for the coefficient of the
Fluid-rock interactions moment of inertia for the Earth. The Earth cannot, therefore,
Geochemical tectonics be assumed to be a homogeneous sphere with a uniform distri-
Geochemistry of sediments bution of mass throughout, and there is a marked concen-
Magmatic processes tration of mass (the iron rich core) in the Earth's center.
Metamorphic environments
Precambrian geochemistry Further evidence for the existence of a metallic core comes
from the Earth's geomagnetic field. This field originates in the
fluid outer core. The convection of iron, which is a good
electrical conductor, in the outer core creates a self exciting
EARTH'S CORE dynamo and produces the Earth's magnetic field. The forces
driving this convection are still being discussed, but are likely
due to thermal or chemical (or both) gradients between the
Introduction
inner-outer core boundary and the core-mantle boundary.
The solid Earth is made up of three fundamental layers, the
core, mantle and crust, which are the products of planetary
differentiation. The core, which is about half an Earth radius Cosmo- and geochemical considerations
wide or only about 1/8 of an Earth volume, makes up about Studies of meteorites help to constrain the composition of the
1/3 of the Earth's mass and is composed of ~ 85% iron, ~ 5% bulk Earth, silicate Earth, and the Earth's core. The composi-
nickel and lesser amounts of other elements that readily alloy tion of the Earth's mantle is not too dissimilar to that of some
with iron. The mantle and crust make up the other half of the stony, metal-depleted, meteorites (achondrites) and thus it is
Earth's radius, and are dominantly composed of silicates of assumed that the core's composition is comparable to that of
magnesium, iron, aluminum, and calcium. iron meteorites. The stony achondrites and iron meteorites are
Our understanding of the Earth's core is primarily estab- derived from differentiated planetismals, whereas more primi-
lished from remote geophysical measurements, including tive meteorites, the chondrites (meteorites composed of stony
studies of the Earth's seismological profile and its magnetic and iron components), are believed to be their undifferentiated
field, and from studies of the Earth's orbital behavior, which precursors and can be surmised as analogs for planetary com-
provides us with a coefficient of the moment of inertia for the positions. By combining these meteoritical data with con-
Earth. Secondary constraints on the nature of the Earth's core straints on the composition of the Earth's mantle one can
come from studies of meteorites and models for the bulk establish a model composition for the planet and by difference
composition of the Earth and its primitive mantle (i.e. the its core.
early Earth mantle that existed prior to the formation of the Planetary compositional models based on primitive meteor-
crust). Together these data sets provide important insights into ites relate elemental abundances in the planet to those observed
the nature and composition of the Earth's core and the pro-
in chondritic meteorites in general, particularly the CI carbon-
cesses that lead to its formation.
aceous chondrites, the most primitive of the chondritic meteor-
ites. These meteorites possess the highest abundances of the
Geophysical constraints moderately volatile and volatile elements relative to the refrac-
The time it takes a seismic wave to pass through the Earth tory elements. The CI carbonaceous chondrites have a com-
depends on the temperature and elastic properties of its internal position that closely matches that of the solar photosphere
layers, which are in turn a function of density. Thus a density (Anders and Grevesse, 1989), thus providing a scale with
structure for the Earth can be established from seismological which we can compare other meteorites. Many question the
data. Travel time curves for both P (compressional) and assumptions involved in planetary models built from chondri tic
S (shear) seismic waves show a shadow zone (a zone where no data. Do the chondritic meteorites, which are believed to be
seismic waves occur because of wave refraction) and is evidence fragments remaining from the initial accretionary material that
of a fundamental discontinuity (boundary) in the Earth at formed during the early development of the solar nebula
2900 km depth (i.e. the core). Depth to the inner-outer core (Wasson, 1985), reflect the degree of chemical and isotopic
boundary ( 5120 km) is more difficult to establish, although the heterogeneity in the early solar nebula? Do comets, which have
existence of a solid inner core was recognized by the presence a higher volatile to refractory element component compared
of a distinct family of P waves that pass through it. The bulk to CI chondrites, possess more primary signatures (Ringwood,
sound velocity of a material is directly correlated to its average 1989)? This heterogeneous collection of meteoritic materials is
density and this in turn can be related to its mean atomic what we have to construct and test our models of planetary
number (Birch, 1952). Based on this, it is observed that the compositions, including the Earth's composition.
core has a mean atomic number that is much higher than that Some 4.6 billion years ago the sun and our system of planets
of the mantle and that the outer liquid core is ~ 10% less dense began forming from the gravitational collapse of a rotating
than that of pure iron when compared to the inner core. This, molecular cloud. This early formational process was a stochas-
plus the fact that iron is likely to be the most abundant element tic one, initially involving the coalescing of small particles
152 EARTH'S CORE

followed by the later coalescing of planetesimals forming plane- CI carbonaceous chondrites (the most primitive of the chon-
tary embryos and the subsequent collision of these embryos to dritic meteorites), or (2) those that model the silicate Earth as
form planets. The chondri tic meteorites record much of these being depleted in Mg and Si relative to the refractory litho-
early processes and thus document, to some extent, the nature phile elements.
of material that went into planet building. Differentiated mete- The seismic velocity structure of the silicate Earth, in combi-
orites (e.g. iron and stony meteorites), most with ages compar- nation with mineral physics data for phases at the appropriate
able to that of our solar system, provide information about pressures and temperatures, provides important information
the early timing and processes of core-mantle separation on about the average density and from this the bulk composition
small rapidly formed planetesimals. of the combined crust-mantle system. Our understanding of
Establishing the composition of the Earth's core is a three the lower mantle, beneath the 660 km seismic discontinuity, is
step process: ( 1) establish the composition of the silicate frac- generally restricted to information derived from seismology,
tion of the Earth (crust plus mantle), (2) given this, establish mineral physics and fluid dynamic experiments. There still
a planetary composition, based on a series of orbital and remains, however, some doubt that the bulk composition of
geophysical observations and assumptions about the planet the mantle above the 660 km discontinuity is similar to that
from the data of chondritic meteorites, and (3) then subtract of the lower mantle below this discontinuity. This is due to
the mantle from the bulk planet to estimate the composition uncertainties in the experimental mineral physics data and their
of the core. The core composition, thus derived, can be com- extrapolation to appropriate pressure and temperature condi-
pared with geophysical data of the Earth's interior and cosmo- tions (Jackson, 1983; Weidner, 1986). These uncertainties
chemical data on iron meteorites to test for consistency. Each allow for alternative model compositions (Agee, 1993;
of these steps is reviewed below. Anderson, 1983) that argue against bulk chemical homogeneity
Knowing the composition of the Earth's mantle is essential in the mantle. Recently, however, a growing body of seismolog-
to establishing the details of the core's composition as well as ical evidence demonstrates that subducted oceanic plates penet-
constraining the bulk composition of the Earth. Meteorites rate the 660 km discontinuity (Grand, 1994; Van der Hilst
provide a first order representation of how elements are distrib- et al., 1997). These observations, combined with data for the
uted between the core and mantle of a planet. Elements geoid and constraints from fluid dynamics (Davies and
enriched in the core are siderophile (affinities for iron), whereas Richards, 1992; Hager and Richards, 1989), indicate that there
those enriched in the mantle are lithophile (rock forming ele- is and has been considerable mass transfer between the upper
ments). The chalcophile elements (affinities for sulfur) are also and lower mantle. Thus, it follows that the compositions of
mostly concentrated in the cores of planets, whereas the the upper and lower mantle are similar.
atmophile elements are concentrated in the exosphere of the Investigations of mantle-derived samples are an important
planet. Iron meteorites are depleted in lithophile elements, and avenue for establishing the Earth's primitive mantle composi-
thus the Earth's core is assumed to be depleted in these ele- tion (Hart and Zindler, 1986; Jagoutz et al., 1979; McDonough
ments. The Earth's mantle contains only a fraction of the and Sun, 1995). Mantle samples include peridotite xenoliths
planet's budget of siderophile and chalcophile elements. Based (mantle rock fragments brought by kimberlites and basalts),
on a wide body of data, these same element distribution pat- massif peridotites (large mountain-sized bodies of the mantle
terns appear to be true for Mars, Venus, the Moon, and the that have been thrust upon the Earth's crust) and primitive
parent bodies of the achondrite meteorites. Elements are also high temperature magmas (e.g. basalts and komatiites). Studies
classified according to their condensation temperatures as of peridotites provide direct information on the nature and
being refractory or volatile, with subdivisions of the latter into composition of the upper mantle. Primitive basalts and komati-
transitional, moderately and highly volatile. The refractory ites, ranging in age from Archean to modern, are in large
elements are assumed to be in chondritic relative proportions degree partial melts of the mantle that have not experienced
in all models of planetary composition. significant chemical differentiation or crustal contamination.
They provide additional, although less direct, data about the
initial composition of their source regions back to about 3.8 Ga
The composition of the Earth's mantle and can be used to construct a time-integrated evolutionary
There have been many attempts to estimate the composition model of the mantle. This information is useful in constraining
of the silicate Earth (the Earth's primitive mantle: the com- the amount of material exchange between the core and mantle
bined crust and mantle composition), and most models agree through time.
within about 10% at the major element level (Allegre et al., Overall, compositional comparisons of mantle samples of
1995; Anderson, 1983; Ganapathy and Anders, 1974; Jagoutz modern to Archean age reveal that there has been relatively
et al., 1979; McDonough and Sun, 1995; Ringwood, 1966). little change in the bulk composition of the mantle at the major
All models assume that the refractory lithophile elements are element level. Thus, models that add basaltic melts back to
in chondritic proportions and are excluded from the core. depleted peridotites in order to re-establish chondritic ratios
Beyond this, however, the models diverge in approach and of the refractory lithophile elements provide us with robust
assumptions. There is no consensus as to the absolute abun- compositional models for the mantle. Such a model composi-
dances of the refractory lithophile elements (e.g. Ca, AI, Ti, tion for the silicate Earth is presented in Table E6. This model
Zr, Sr, REE) in the primitive mantle. Moreover, there is even is consistent with existing petrologic and isotopic data, as well
less agreement as to the relative and absolute abundances of as with those limits established by seismology, mineral physics,
the moderately volatile (e.g. Na, K, Rb, As, Sb, Ge) and geodynamics and heat flow.
volatile elements (e.g. S, Se, Cd, Pb, Bi, Tl) in the primitive
mantle. In addition, compositional models for the Earth usu- The composition of the Earth
ally fall into one of two groups: ( 1) those that assume the In the above model the silicate Earth possesses chondritic
silicate Earth has a complement of Mg and Si equal to that in proportions of the refractory lithophile elements at ~2.75
 EARTH'S CORE 153

Table E6 Composition of the bulk Earth, core, and silicate Earth

Element Bulk Earth Core Silicate Earth Element Bulk Earth Core Silicate Earth
H 260 600 100 Rh 0.24 0.74 0.001
Li 1.1 0 1.6 Pd 1 3.1 0.004
Be 0.05 0 0.07 Ag 0.05 0.15 0.008
B 0.2 0 0.3 Cd 0.08 0.15 0.04
c 730 2000 120 In 0.007 0 0.01
N 55 170 2 Sn 0.25 0.5 0.13
0 (%) 29.7 0 44 Sb 0.05 0.13 0.006
F 10 0 15 Te 0.3 0.85 0.012
Na (%) 0.18 0 0.27 I 0.05 0.13 0.01
Mg (%) 15.4 0 22.8 Cs 0.035 0.065 0.021
AI(%) 1.59 0 2.35 Ba 4.5 0 6.6
Si ('Yo) 16.3 6.4 21.0 La 0.44 0 0.65
p 1100 3200 90 Ce 1.13 0 1.68
s 6345 19000 250 Pr 0.17 0 0.25
Cl 76 200 17 Nd 0.84 0 1.25
K 160 0 240 Sm 0.27 0 0.41
Ca (01<,) 1.71 0 2.53 Eu 0.10 0 0.15
Sc 10.9 0 16 Gd 0.37 0 0.54
Ti 810 0 1200 Tb 0.067 0 0.10
v 95 120 82 Dy 0.46 0 0.67
Cr 4700 9000 2625 Ho 0.10 0 0.15
Mn 720 50 1045 Er 0.30 0 0.44
Fe(%) 31.9 85 6.26 Tm 0.046 0 0.068
Co 880 2500 105 Yb 0.30 0 0.44
Ni 18220 52000 1960 Lu 0.046 0 0.068
Cu 60 125 30 Hf 0.19 0 0.28
Zn 40 0 55 Ta 0.025 0 0.037
Ga 3 0 4 w 0.17 0.47 0.029
Ge 7 20 1.1 Re 0.075 0.23 0.0003
As 1.7 5 0.05 Os 0.9 2.8 0.003
Se 2.7 8 O.D75 Ir 0.9 2.6 0.003
Br 0.3 0.7 0.05 Pt 1.9 5.7 0.007
Rb 0.4 0 0.60 Au 0.16 0.5 0.001
Sr 13 0 20 Hg 0.02 0.05 0.01
y 2.9 0 4.3 Tl 0.012 0.03 0.004
Zr 7.1 0 10.5 Pb 0.23 0.4 0.15
Nb 0.44 0 0.66 Bi 0.01 0.03 0.003
Mo 1.7 5 0.05 Th 0.055 0 0.08
Ru 1.3 4 0.005 u 0.015 0 0.02

All abundances given in f.lg/g unless otherwise stated. Compositional models for the bulk Earth, core, and silicate Earth are modified after McDonough
and Sun (1995).

times their abundance levels in CI carbonaceous chondrites. Earth. A significant density increase is calculated from the
This enrichment factor is due to the fact that these chondrites jump in seismic P wave velocities at the inner-outer core
are rich in water and C0 2 , whereas the Earth is not and to boundary (Shearer and Masters, 1990). This, combined with
the fact that the silicate Earth experienced core subtraction. It laboratory studies on the compression of Fe-Ni alloy at core
has long been established that the silicate Earth is depleted in pressures (Brown et al., 1984; Brown and McQueen, 1986;
a number of volatile lithophile elements (e.g. K and Rb) rela- Mao et al., 1990), indicate that the outer core possesses a
tive to primitive chondritic meteorites. When compared with significant proportion ( ~ 10 ± 4%) of a light element (i.e. an
CI carbonaceous chondrites the silicate Earth is depleted in element with a much lower atomic number than Fe) or a
Mg and Si relative to the refractory lithophile elements and combination of light elements (e.g. Si, S, 0). Thus, assuming
shows a uniform decrease in the abundances of the moderately a chondritic Fe/Ni value for the Earth, the core has an iron
volatile and volatile lithophile elements with decreasing con- content of 85% and the bulk Earth has an Fe/Al of ~ 20 ± 2
densation temperature. This volatility trend is surmised to (Table E6). (Aluminum is a proxy for the refractory lithophile
reflect the nature of the nebular material in the planetary elements, and it is considered least likely of the major lithophile
feeding zone of the proto-Earth at !AU (Figure E3). elements (e.g. Si, Mg, Ca) to be incorporated as the light
The composition of the Earth's core is poorly constrained element in the core.) This is an important result; it provides
beyond its major constituents (i.e. an Fe-Ni alloy). The planet us with an estimate of the composition of the planet that does
is assumed to possess chondritic proportions of Fe and Ni not require knowledge of the nature of the light element com-
(value of Fe/Ni varies by only 10-15% in chondri tic meteorites) ponent in the core. From this we can evaluate the different
and, given the low Ni content of the mantle(~ 2000 ~tg/g), the groups of chondri tic meteorites as analogs to the bulk Earth.
core is calculated to be the dominant reservoir for Ni in the Chondritic meteorites display a range of Fe/Al ratios with
154 EARTH'S CORE

mineral physics data. It is recognized that the lighter the ele-

~ a~undances relative to that in CI chondrites and Mg normalized
ment, the less is required to compensate for the density contrast
1.0 p-g 8 A~ : ~ t•llli II Ull'ill!!! Illig~~~~ llli lllillli (see discussion in Poirier, 1994). In order to characterize the
~ sp ~ CueMnfl~~
e SbKA 0 ""'lP'
~ ~~ !9 ~ ° ~ light element component we must construct models of the bulk

....
Au
NaeA"tl Ge
Earth's composition and this in turn depends on our models
of core formation.
0.1
e
llli
Earth
CM chondrites
AgGaRb~·
.Ci
8 Cs
Sn"
:.sczn
Te s
~dl Pb Bi Tl

ln~ 1 ~g
)? 0

e
Possible processes of core formation in the early Earth are
many and range from metal-sufide equilibrium at low pres-
t> CO chondrites sures to the high pressure solubility of FeO in Fe. The physics
0 CV chondrites c "
eN of planetary accretion and core separation are presently
eH
0.01

1200 1000 800 600 400 200

• insufficiently understood to rule out effectively some competing
hypotheses (Stevenson, 1981 ). An estimate for the composition
Temperature K (at 50% condensation for nebular at I0-3 atmospheres) of the core, based on an assumed light element mixture of
10%, is presented in Table E6. The principal observations used
Figure E3 A plot of the abundances of elements in the Earth to construct this estimate are the relatively constant siderophile
relative to those in primitive meteorites (i.e. Cl carbonaceous ratios in chondri tic meteorites, the composition of the mantle,
chondrites) versus their condensation temperatures, expressed as
the bulk Earth volatility trend for the lithophile elements
their 50% condensation temperatures in a solar nebula that is
assumed to be at 10- 3 atm. Elemental abundances are shown (Figure E3) and the compositional observations on iron
relative to the Mg contents of the most primitive class of meteorites meteorites.
(CI carbonaceous chondrite); doing this allows for a better The core is the dominant repository of the siderophile ele-
comparison between the different types of meteorites and their ments in the Earth. Hence, the limited variation in Fe/Cr,
vastly different water and C0 2 contents. Condensation temperatures Cr/V, Ni/Co and Ni/P in chondritic meteorites and the relative
are from Wasson (1985), or are interpolated where estimates are abundances of these elements in the mantle provide for an
uncertain; abundance data for the meteorites are from Wasson and estimate of these minor to trace elements in the core. By setting
Kallemeyn (1988) and Table E6 for the Earth. The meteorites CM,
CO and CV chondrites are three different varieties of carbonaceous
an upper limit of ~ 10% for the light element content of the
chondrites, which are more evolved than Cl chondrites. core, the abundances of the major and minor siderophile ele-
ments are established. This yields a core composition with
~85% Fe, ~5% Ni, ~0.9% Cr and 0.3% P and Co, along
many having a value close to 20 (Allegre et al., 1995). However, with trace quantities of other siderophile and chalcophile ele-
the high Fe/Al values (35) ofEH chondrites and the low Fe/Al ments (Table E6). The elements Cr and V are often considered
values ( 13-15) for CV and CK chondrites make these meteor- to be lithophile; however, based on their abundances in the
ite groups unlikely analogs for the bulk Earth. mantle, it is assumed that a fraction of these elements was
The significant aspects of the Earth's compositional model partially incorporated into the core as a result of high pressure
in Table E6 are: ( 1) Fe/Al ~ 20, (2) chondri tic proportions of metal-silicate equilibria during core formation.
the refractory elements, with absolute abundances comparable The limits of our understanding of the core's composition
to those in other volatile-depleted, carbonaceous chondrites, (including the light element component) depend on models of
and (3) the relative abundances of the moderately volatile and core formation and the class of chondritic meteorites chosen
volatile lithophile elements in the mantle decrease with models for constructing models of the bulk Earth's composition. Based
for the element's nebular condensation temperature, demon- on the Earth's volatility curve (Figure E3), which is established
strating that the Earth has a volatile depletion pattern similar from the lithophile elements and carbonaceous chondrites, we
to some of the carbonaceous chondrites (Figure E3). can constrain the planetary budget for the volatile chalcophile
elements in the planet. From this, one predicts that the bulk
The composition of the Earth's core of the planetary budget for sulfur (and selenium and tellurium,
The core, or metallic Earth, is subdivided into an outer liquid other strongly chalcophile elements) is held within the core
sphere and inner solid region. It is dominantly composed of a and not the mantle, given its marked depletion in the latter
mixture of iron and nickel, with other siderophile elements (Table E6). Nonetheless, this preferential partitioning of sulfur
making up < 3% by mass. In addition to this, the outer core into the core can only account for <2% of the light element
has ~ 10% of a low atomic weight component, which accounts component in the outer core. The corollary is that there is
for its lower than expected density when compared to the more than one light component that accounts for the lower
density of the solid inner core and the changes that accompany density of the outer core.
such a liquid-solid transition. The nature of this light element The estimate for Si in the core is based on the volatility
in the outer core has received considerable attention and its curve for lithophile elements in the bulk Earth (Figure E3 ),
identity is not well constrained. and this is not well constrained, particularly given the marked
Since Birch's early work (Birch, 1952) there have been many variations in Si contents of chondritic meteorites. O'Neill
suggestions for what might be the light element alloying with (1991) argued that a limited amount of Si (:0::5%) entered the
iron in the Earth's outer core. Birch and others suggested H, core, and invoked additional components to satisfy fully the
C, N, 0, Mg, Si, S (or a combination of some of these elements) density requirements. The recent suggestion by O'Neill and
as possible candidate light elements in the outer core, in which Palme ( 1997) that the volatility of Mn is comparable to that
to satisfy the density increase at the outer-inner core boundary. of Na, may explain the low Mn content of the planet. The
It is unfortunate, however, that the exact proportion and char- high abundances of V and Cr in the Earth's core (Table E6)
acter of the light element component in the outer core cannot can be accounted for by their depletions in the silicate Earth.
be constrained presently with our existing seismological and The estimate of the Mg/V ratio for the bulk Earth (0.16) is
 EARTH'S CORE 155

within the range for chondritic meteorites (0.15-0.24), assum- the Earth's initial ratio of U/Pb, an essential piece of informa-
ing no Mg in the core. Assuming no V in the Earth's core, the tion for dating core formation. Moreover, it appears that the
bulk Earth would have a Mg/V ratio (0.28) that is greater bulk of the Earth's Pb budget has remained in the mantle,
than that seen in all chondri tic meteorites. Thus, we can safely based on the Earth's volatility trend, with little or no Pb
assume that there is a significant quantity of V in the Earth's partitioned into the core. These features make this isotope
core. This, along with recent experimental work on Ni and Co system less than favorable for constraining the timing of core
partitioning (Li and Agee, 1996), argues for high pressure formation.
equilibration of metal and silicate in the Earth's mantle during Core separation is considered to have been wholly completed
core formation. Ringwood (1984) has argued that under high after about the first half billion years of Earth history, based
pressure some amount of FeO may also have been dissolved on the composition of the earliest rocks derived from the
in the metal. Thus, oxygen may likely be part of the light mantle (circa 4.0 billion years ago). Komatiites, ancient high
element component in the outer core, although it is not incor- magnesium lavas that were generated by large degrees of melt-
porated into this model. McDonough and Sun (1995) noted ing of the mantle, and basalts carry with them a compositional
that there are anomalous depletions of Cs, Cl, Br and I in the signal of the Earth's mantle at the time of their eruption.
silicate Earth, indicating that under high pressure Cs, and Similar ratios of siderophile to lithophile elements (e.g. Fe/Mg,
possibly the heavy halides, may have been partially sequestered Fe/Mn, and P/Nd) in 3.5 billion years old and later komatiites
into the core. Alternatively, these halides may have been and in modern basalts demonstrate that the relative abun-
strongly partitioned into the Earth's exosphere early on and dances of elements concentrated in the core (siderophile) and
lost with the rest of the proto-atmosphere during the sun's those concentrated in the mantle (lithophile) have not changed
T-Tauri phase. significantly in the mantle throughout much of Earth's history.
Presently, there are considerable efforts being invested in Moreover, peridotite fragments with ages that span the last
gaining further insights into the nature of the light element 3.5 billion years of Earth's history also possess constant values
component in the Earth's core. Until we can, with considerable of key lithophile/siderophile ratios (e.g. Mg/Ni), further indi-
certainty, establish the number of light element component(s) cating that negligible amounts of material exchange have
in the Earth's outer core, we must entertain multiple occurred between the core and mantle since early core
hypotheses. formation.

The timing of core formation William F. McDonough

Chemical and isotopic observations indicate that much of the
core's formation post-dates the dominant stage of planetary Bibliography
accretion, and that core separation appears to be completed Agee, C.B. (1993) Petrology of the mantle transition zone. Annu. Rev.
within the first few hundred million years (or less) of the Earth Planet. Sci., 21, 19-41.
Earth's history. The timing of core formation is not well estab- Allegre, C.J., Poirier, J.-P., Rumler, E. and Hofmann, A.W. (1995)
lished and depends on separation of a short-lived radioactive The chemical composition of the Earth. Earth Planet. Sci. Lett.,
e
isotope of hafnium 82 Hf, a lithophile element) from its decay 134, 515-26.
e
product isotope of tungsten 82 W, a siderophile element) Anders, E. and Grevesse, N. (1989) Abundances of the elements:
meteoritic and solar. Geochim. Cosmochim. Acta., 53, 197-214.
during core formation (Halliday et al., 1996; Harper and Anderson, D.L. (1983) Chemical composition of the mantle.
Jacobsen, 1996; Lee and Halliday, 1995). The extinct nuclide, J. Geophys. Res., 88 (Suppl.), B41-52.
182 Hf, which decays to 182W with a half-life of about 9 million Birch, F. (1952) Elasticity and constitution of the earth's interior.
years, was present 4.6 billion years ago during planetary accre- J. Geophys. Res., 57, 227-88.
tion. The isotopic composition of W in the mantle is similar Brown, J.M. and McQueen, R.G. (1986) Phase transitions, Giineisen
to that in primitive chondritic meteorites and is unlike that in parameter and elasticity for shocked iron between 77 GPa and
400 GPa. J. Geophys. Res., 91, 7485-4.
iron meteorites; this indicates that core separation on the Earth Brown, J.M., Ahrens, T.J. and Shampine, D.L. (1984) Hugoniot data
must post-date the decay of most of the short-lived 182Hf (circa for pyrrhotite and the Earth's core. J. Geophys. Res., 89, 6041-8.
20-50 million years). In contrast, depletions of 182W in the Davies, G. and Richards, M. (1992) Mantle convection. J. Geol., 100,
W isotopic spectrum of iron meteorites is firm evidence for the 151-206.
early separation of a W-bearing iron core from a Hf-bearing Ganapathy, R. and Anders, E. ( 1974) Bulk composition of the moon
and earth, estimated from meteorites. Proc. Fifth Lunar Sci. Conf,
mantle in these planetesimals. This is also consistent with the 2,1181-1206.
much quicker isotope system of Mn-Cr (3.7 million year half- Grand, S.P. ( 1994) Mantle shear structure beneath the Americas and
life), which shows that iron meteorites separated from stony surrounding oceans. J. Geophys. Res., 99, I 1591-621.
achondrites on 1-10 million years or less time scales (Lugmair Hager, B. and Richards, M. (1989) Long-wavelength variations in the
etal., 1994). Earth's geoid: physical models and dynamical implications. Phil.
Another isotope system often considered useful for con- Trans. R. Soc. Lond., A328, 309-27.
Halliday, A.N., Rehkiimper, M., Lee, D.-C. and Yi, W. (1996) Early
straining the timing of core formation is the U-Pb system. evolution of the Earth and Moon: new constraints from Hf-W
This is because Pb is considered to be chalcophile and may isotope geochemistry. Earth Planet. Sci. Lett., 142, 75-89.
have been sequestered into the core during its formation, while Harper, C.L. and Jacobsen, S.B. ( 1996) Evidence for 182Hf in the early
U is lithophile and was concentrated in the mantle. As noted Solar system and constraints on the time scale of terrestrial accre-
above, the Earth shows abundant evidence for depletions in tion and core formation. Geochim. Cosmochim. Acta, 60, I 131-53.
moderately volatile and volatile lithophile elements. Pb is a Hart, S.R. and Zindler, G.A. (1986) In search of a bulk-earth composi-
tion. Chem. Geol., 57, 247-67.
volatile chalcophile element and it, along with other chalco- Jackson, I. (1983) Some geophysical constraints on the chemical com-
phile elements, displays depletions comparable to those seen position of the Earth's lower mantle. Earth Planet. Sci. Lett., 62,
in the volatile lithophile elements. Thus, it difficult to establish 91-103.
156 EARTH'S FORMATION AND GEOCHEMICAL EVOLUTION

Jagoutz, E., Palme, H., Baddenhausen, H. et a!. (1979) The abun- tectonic cycle, plume related magmatism and convection in the
dances of major, minor and trace elements in the earth's mantle as deep interior of the Earth. As the cooling of a planet is propor-
derived from primitive ultramafic nodules. Proc. Lunar Planet. Sci.
Conf, 10, 2031-50. tional to its surface area ( ~ r2 , where r =radius of planet) and
Lee, D.-C. and Halliday, A.N. ( 1995) Hafnium-tungsten chronometry heating by radioactive decay to the volume ( ~ r 3 ), the internal
and the timing of terrestrial core formation. Nature, 378, 771-4. activity of a planet increases with its size and declines with the
Li, J. and Agee, C. B. ( 1996) Geochemistry of mantle-core differentia- time scale of the decay oflong-lived radioactive isotopes, which
tion at high pressure. Nature, 381, 686-9. is on the order of billions of years. A large planet such as the
Lugmair, G.W., Macisaac, C. and Shukolyukov, A. (1994) Small time Earth is still active today, but smaller planets such as the Moon
differences recorded in differentiated meteorites. Meteoritics, 29, have cooled to a degree where any internal activity has ceased.
493-4.
Mao, H.K., Wu, Y., Chen, L.C. and Shu, J.F. (1990) Static compres- The quantitative assessment of the differentiation of the Earth
sion of iron to 300 GPa and Fe 0 .8 Ni 0 . 2 to 260 GPa: implications for requires the knowledge of the initial composition of the Earth.
composition of the core. J. Geophys. Res., 95, 21737-42. Current theories of Earth formation indicate that the Earth
McDonough, W.F. and Sun, S.-s. (1995) The composition of the did not grow by collecting fine dust particles from space, but
Earth. Chem. Geol., 120, 223-53. by a hierarchical accumulation process from small to increas-
O'Neill, H. S.C. (1991) The origin of the Moon and the early history of ingly larger planetesimals. Thus the Earth was made by con-
the Earth - A chemical model. Part 2: The Earth. Geochim.
Cosmochim. Acta, 55, 1159-72.
structive collisions, from the same population of planetesimals
O'Neill, H. S.C. and Palme, H. ( 1998) Composition of the silicate that produced meteorites by destructive collisions. The com-
Earth: implications for accretion and core formation, in The Earth's position of meteorites therefore provides important clues to
Mantle: Composition, Structure and Evolution (ed. I. Jackson). New the bulk composition of the Earth.
York: Cambridge University Press, pp. 3-126.
Poirier, J.-P. (1994) Light elements in the Earth's outer core: A critical Formation of the Earth
review. Phys. Earth Planet. Interiors, 85, 319-37.
Ringwood, A. E. ( 1966) The chemical composition and origin of the Star formation is the result of the collapse of dense, cold,
Earth, in Advances in Earth Sciences (ed. P.M. Hurley). Cambridge: interstellar molecular clouds, perhaps triggered by a supernova.
MIT Press, pp. 287-356. The solar nebula, a rotating disk of a mixture of gas and dust,
Ringwood, A.E. (1984) The Earth's core: its composition, formation separated from the interstellar medium about 4.6 billion years
and bearing upon the origin of the Earth. Proc. R. Soc. Lond. A, ago. Material within the rotating disk lost, through viscous
395, 1-46. dissipation or by other processes, angular momentum and
Ringwood, A.E. (1989) Significance of the terrestrial Mg/Si ratio.
Earth Planet. Sci. Lett., 95, 1-7.
accreted to the growing central star, the Sun. Most of the mass
Shearer, P. and Masters, G. (1990) The density and shear velocity of the solar nebula ultimately found its way into the Sun. The
contrast at the inner core boundary. Geophys. J. Int., 102,491-8. planets, including the Earth, Mars, Jupiter, asteroids etc. are
Stevenson, D.J. ( 1981) Models of the Earth's core. Science, 214, made of the tiny fraction of matter (0.14%) that remained
611-19. unaccreted to the Sun.
Vander Hilst, R., Widiyantoro, S. and Engdhal, E.R. (1997) Evidence The oldest dated solid materials formed within the solar
for deep mantle circulation from global tomography. Nature, 386, nebula are Ca-Al-rich inclusions of the Allende meteorite.
578-84.
Wasson, J.T. ( 1985) Meteorites. Their Record of Early Solar-System These inclusions are thought to have condensed from the cool-
History. New York: W.H. Freeman, 267 pp. ing solar nebula. Their age, based on the isotopic composition
Wasson, J.T. and Kallemeyn, G.W. (1988) Compositions of chon- of lead, is 4 566 109 years, with an uncertainty of only 2 million
drites. Phil. Trans. R. Soc. Lond., A325, 535-44. years (Manhes eta!., 1988). This marks the beginning of the
Weidner, D.J. ( 1986) Mantle model based on measured physical prop- formation of solid bodies in the solar system. Within a short
erties of minerals, in Chemistry and Physics of Terrestrial Planets time span of a few million years small planetesimals, the parent
(ed. S.K. Saxena). New York: Springer-Verlag, pp. 251-74.
bodies of meteorites, were accreted and, in some cases, melted
and differentiated. The formation of the larger inner planets,
Cross-references including the Earth, took much longer and was more compli-
Earth's continental crust cated. It involved, according to Wetherill (1994), two steps.
Earth's formation and geochemical evolution The first was rapid growth of protoplanets (embryos) with
Earth's mantle geochemistry masses of about 1/100 of the mass of the Earth within 100 000
Earth's oceanic crust years from 'local' material, i.e. material derived from the helio-
Elements: distribution
Elements: transitional
centric distance at which the planets formed. In a second step
Geochemical tectonics lager planets such as the Earth formed by collisions of embryos.
Sulfides This step took millions of years and involved significant radial
Sulfides in mafic and ultramafic rocks mixing. Embryos from various heliocentric distances may have
contributed to the growing Earth. This has had important
implications for the bulk chemical composition of the Earth.
EARTH'S FORMATION AND GEOCHEMICAl It is often assumed that the solar system is compositionally
stratified, with increasingly more volatile rich and more oxi-
EVOlUTION dized material with increasing distance from the Sun. If so,
late Earth accreting material from beyond Mars' orbit may
In discussing the geochemical evolution of the Earth it is often have provided significant amounts of volatile constituents, such
assumed that the Earth was initially hot and chemically homo- as water, to the growing Earth, more than local material
geneous, and that it differentiated, on cooling, into core and could provide.
mantle. The combination of retained initial thermal energy
and energy produced by the decay of the long-lived radioactive Structure of the Earth
nuclei 4 °K, 235 U, 238 U and 232Th provided the heat required The division of the Earth into distinct layers has been inferred
for further differentiation, such as crust formation, the plate from the results of analyses of travel times of seismic waves
 EARTH'S FORMATION AND GEOCHEMICAL EVOLUTION 157

oceanic crust continental crust older than 2 Ga in contrast to the thin (7 km) and young
(6 km) (40 km) oceanic crust of basaltic composition.

;:;_;:;;:;_;;_~U;!:;P.§~R;_;~~A-~N;lJ~§:~_:_::_::::;:~., -~ Mohorovicic Bulk chemical composition of the Earth

- - \ Discontinuity Calculating bulk Earth compositions requires the knowledge
410 km
660 km
\. low velocity of the composition of the core and the mantle. Mantle composi-
\. zone tions are derived from a combination of chemical data of rocks
transition from the Earth's upper mantle and of meteorite data which
zone
are primarily used to define ratios among cosmochemical
LOWER MANTLE related groups of elements (Jagoutz et al., 1979; McDonough
and Sun, 1995). Uncertainties in the bulk composition of the
Earth arise from uncertainties in the composition of the lower
mantle and the composition of the core. Compositional uni-
formity of upper and lower mantle is often assumed. In the
2885 km
chemical composition of upper mantle rocks there is no evi-
dence for global fractionation processes. A lower mantle frac-
tionated in refractory elements (e.g. Hf and Sm) as a result of
crystallization of a magma ocean would require a complemen-
tary upper mantle signature, which is not observed (Kato
et al., 1988). Others, however, believe, based on seismic data,
that the lower mantle has a higher Si/Mg (more pyroxene) and
more Fe than the upper mantle (Anderson, 1989). Although
it is generally accepted that the major constituent of the core
is metallic iron plus about 5% Ni, data obtained by shock
experiments on metallic iron are interpreted to indicate that
the observed density is lower than the density of a pure FeNi-
metal core. The nature of the light element that is required to
balance the density of the outer core is unknown. Suggestions
Figure E4 Structure of the Earth (after Anderson, 1989). have been made, from hydrogen, oxygen, silicon to sulfur, but
no consensus has emerged. A combination of several elements
triggered by earthquakes (Anderson, 1989). Three principal may be the most likely possibility (Poirier, 1994; O'Neill and
units are distinguished (Figure E4 ), the core with 32% of the Palme, 1997).
mass of the Earth, the overlying mantle (67.3%) and a thin Bulk Earth abundances will be shortly discussed in terms of
outer crust (0. 7%). cosmochemically related element groups.
The solid inner core ( 5155 km) is assumed to have frozen
out of the liquid outer core, extending from the inner core Volatile elements
boundary (ICB) to the core-mantle boundary (CMB) at The bulk chemical composition of the Sun is reasonably well
2885 km. The growth of the inner core is an ongoing process. known from absorption spectroscopy of the solar photosphere.
The outer core is assumed to be fluid, based on the absence Primitive, undifferentiated or chondri tic meteorites, i.e. meteor-
of shear waves and because rapid convection is required to ites derived from planetesimals that have not undergone
drive the geodynamo and produce the magnetic field of the internal melting processes, have chemical compositions broadly
Earth. Above the outer core there is a thin layer, 100-300 km consistent with the composition of the Sun (for elements heav-
thick, with significant variations in seismic velocities. The so ier than 0 and except for rare gases). A small group of meteor-
called D" layer is thought to be chemically distinct and acts as ites, the CI-chondrites, approach the composition of the Sun
a thermal boundary layer bridging the temperature difference more closely than others. The compositional differences
between outer core and lower mantle which may be as high as between CI-chondrites (with solar abundances) and other
1350°C (Boehler, 1996). Although the core is losing heat to undifferentiated meteorite groups are primarily in the abun-
the overlying mantle it appears to be chemically isolated. The dances of volatile elements, such as Na, K, Zn, S, Pb etc.
lower mantle extends to a depth of 660 km, accounting for Analyses of samples from the upper mantle of the Earth show
49.2% of the mass of the Earth. The seismic discontinuity at that the bulk Earth has significantly lower volatile element
660 km reflects the transition of intermediate density minerals abundances than CI -chondrites. This is not unusual, as most
(ringwoodite and majorite) to the high density minerals perov- meteorites (except, of course, CI-chondrites) have lower than
skite and magnesiowuestite, the dominating minerals in the solar abundances of volatile elements. The depletion of volatile
lower mantle. If the 660 km discontinuity also indicates a elements in most meteorites and in planets is the result of
compositional change between upper and lower mantle is not processes in the solar nebula, probably by incomplete conden-
clear. Geochemical arguments support a chemically homogen- sation, i.e. before accretion (Palme et al., 1988). The generally
eous mantle. The zone from 660 km to 410 km is called the low content of volatile elements in the Earth implies low bulk
transition zone and marks the transition of olivine to wadsley- Earth S-contents. This limits the S-content of the Earth core
ite and orthopyroxene to majorite, a mineral with garnet to a few percent (Dreibus and Palme, 1996).
structure. The crust is separated from the mantle by the
Mohorovicic discontinuity (Moho) reflecting a sudden increase Refractory elements
in seismic velocity. The continental crust is on average about Refractory elements (Ca, AI, Ti, REE, U, Th etc.) condense
40 km thick, has andesitic composition and is in large areas from a cooling gas of solar composition before the major
158 EARTH'S FORMATION AND GEOCHEMICAL EVOLUTION

phases, metal and Mg-silicates. Refractory elements occur in

all chondritic meteorites in abundance ratios in which they
occur in the Sun. This unique pattern is also assumed for the
Earth, implying, for example, that the REE pattern in the bulk 1.2
Earth is the same as in the Sun and chondritic meteorites. "'
0
-~
Other important chondritic (or solar) ratios among refractory 1-<

elements in the bulk Earth are AljCa, AljU, Sr/Nd, Nb/Ta, ..c
bD
1.0
Zr/Hf, Lu/Hf, Th/U. The hypothesis of constant refractory ';;)

element ratios in the bulk Earth is extremely important for ~

estimating bulk Earth compositions. A postulated chondritic, i:/3
i.e. solar, Hf/Sm ratio in the bulk Earth and a measured ---
bD
0.8
:OS
chondritic ratio in the upper mantle have been used to argue
for a chemically identical upper and lower mantle ( Kato eta!.,
0.6
1988). So far no evidence has been found that would not be
consistent with this hypothesis, with the exception of the Th/U
ratio which is sometimes assumed to be above chondri tic. Also, 0.05 0.1
V abundances in samples from the upper mantle of the Earth AlI Si (weight ratios)
are considerably lower than those calculated from refractory Figure E5 Mg/Si vs. AI/Si ratios in rocks from the upper mantle of
element ratios. Since V has siderophile tendencies, the missing the Earth (full symbols) and chondritic meteorites (E: enstatite
V is probably in the core (see Earth's core). chondrites; other meteorite types as indicated). Data for the solar
Similar assumptions as for refractory lithophile elements are photosphere are included. Two average primitive upper mantle
made for refractory metals, i.e. metals condensing at temper- compositions are marked by stars (MD + S: McDonough and Sun,
atures above FeNi. Although the Earth's core is the major 1995; P + N: Pal me and Nickel, 1985). Adapted from Pal me
reservoir for the refractory siderophile elements W, Mo, Ir, (1990)
Os, Re, Ru, Pt and Rh, their ratios in the Earth mantle provide
important constraints for the core formation process (see ratios. Solar bulk Earth ratios would require additional reser-
below). voirs for Si and AI. Even if some Si were in the core the AI/Mg
ratio would still be non-solar. The other possibility is a lower
mantle with higher Si and lower Al to compensate the low Si
Mg, Si, Fe and high AI of the upper mantle. Differences in chemistry
The four elements 0, Fe, Si and Mg make up about 95% of between upper and lower mantle are, however, unlikely.
the bulk Earth (in atoms). However, there is some ambiguity The bulk Earth composition probably does not match the
as to the bulk Earth concentrations of these elements. In composition of any group of chondri tic meteorites. What can
Figure E5 Mg/Si ratios of upper mantle rocks and meteorites be said, however, is that the kind and extent of deviations of
are plotted against their AljSi ratios. The upper mantle rocks the bulk Earth composition from the solar composition is
are either mantle xenoliths or samples from massive peridotite similar to variations found among chondritic meteorites
bodies (see Palme and Nickel, 1985, for details). The trend (O'Neill and Palme, 1997).
observed for the ultramafic terrestrial rocks is typical of hun- The structure and composition of the Earth as outlined
dreds of such analyses of rocks in continental areas throughout above is the result of three major differentiation processes:
the world (McDonough and Sun, 1995; O'Neill and Palme, (a) the formation of the Earth's core during or very soon after
1997). the end of accretion, (b) the formation of the continental crust
The decrease in Mg/Si with increasing AI/Si in Figure E5 is with a mean age of about 2-2.2 Ga and (c) the formation and
qualitatively understood as melt extraction trend where the recycling of oceanic crust with a characteristic mixing time of
peridotitic rocks plotted reflect residues after extraction of about 1.0 Ga.
variable melt fractions. Increasing removal of AI into the par-
tial melt is accompanied by increasing Mg in the residue. The Formation of the core
primitive upper mantle (PUM) composition, i.e. the composi- The first major differentiation event that has affected the Earth
tion of the upper mantle before any melt extraction, should is the formation of the FeNi core. The gravitational energy
plot somewhere at the end of this trend with high AI/Si and liberated by core formation would, if released instantaneously,
low Mg/Si. Different choices of upper mantle rocks lead to raise the temperature of the mantle to about 2300°C. However,
small differences in the estimates of PUM. The pyrolite com- details of this process, whether it occurred continuously during
position, a mixture of depleted mantle and basalt, is given for accretion or whether a single vigorous late event caused
comparison (Ringwood, 1979). The composition of chondri tic metal to segregate into the core of the Earth, are unclear
meteorite groups and the solar photospheric ratios are indi- (Stevenson, 1990).
cated. Apparently there is a considerable variation in AI/Si The composition of the mantle in particular, the abundances
and Mg/Si ratios in chondritic meteorites with a trend roughly of siderophile elements, such as Ni, Co, W, Ir, Pt, Co, Au, As,
opposite to the terrestrial trend. According to Figure E5 the Ge, Ir, Os etc. contain important information on details of the
Mg/Si and Al/Si ratios of the upper mantle are within the core formation process. To what extent did core forming metal
range of CV-chondrites but clearly separated from the solar equilibrate with silicates? Under what conditions (temperature,
or the CI ratio. pressure, oxygen fugacity) did the equilibration occur? What
If the lower mantle has the same Mg/Si and AI/Si ratios as was the size of the metals that equilibrated with silicate? Was
the upper mantle then the bulk Earth should have non-solar there enough time to achieve equilibrium between massive
 EARTH'S FORMATION AND GEOCHEMICAL EVOLUTION 159

metal blobs and silicates? What were the phases that equili- under these conditions a Ni content of the Earth mantle of
brated: solid or liquid metal with liquid silicates, relevant for 40 ppm would be expected instead of the 1900 ppm derived
equilibration in the upper mantle, or liquid metal with solid from the analyses of upper mantle rocks (e.g. McDonough
silicate reflecting lower mantle conditions? Each of these possi- and Sun, 1995). Results of recent experiments, however, show
bilities will imprint a different siderophile element signature on that metal/silicate partition coefficients for Ni decrease with
the residual mantle. This signature is the only long lasting increasing temperature and pressure (Li and Agee, 1996).
effect that constrains the infinite number of geophysical mecha- Equilibration of core-forming metal and mantle silicates at
nisms that could possibly lead to separation and collection of great depth within the Earth would imprint higher Ni concen-
metal required for the formation of the core. trations on the mantle and could at a certain depth be com-
Any quantitative prediction for the residual pattern of sider- patible with equilibrium distribution between metal and
ophile elements in the Earth's mantle requires the knowledge silicate. Equilibration at lower mantle conditions requires the
of metal/silicate coefficients as function of temperatures, pres- knowledge of partition coefficients of liquid metal with Mg-
sures, oxygen fugacities and compositions. In Figure E6 an perovskite and Mg-wiistite. Recent high pressure experiments
overview is given of a large number of experimentally deter- have produced such data (Ohtani eta!., 1997). It appears that
mined metal/silicate partition coefficients at one atmosphere global core-mantle equilibration is consistent with the abun-
pressure. Stippled lines indicate extrapolations from experi- dances of some siderophile elements in the Earth's mantle
mental data. Elements above Fe are siderophile with increasing considering pressure dependence of metal/silicate partition
siderophility and those below are lithophile. coefficients (Righter and Drake, 1997). However, it remains
The relevance of these partition coefficients will be demon- to be seen whether this model accounts for the mantle abun-
strated with Ni. Experimentally determined metal/silicate parti- dances of all moderately siderophile elements.
tion coefficients for Ni are around 1300, at I atm pressure, a In alternative models for explaining siderophile element
temperature of 1600°C and at an oxygen partial pressure abundances in the Earth's mantle, the initial silicate/core-form-
appropriate for core-mantle equilibrium (Holzheid and Palme, ing metal equilibration occurs at very reducing conditions
1996). If core-forming metal equilibrated with mantle silicates removing siderophile elements quantitatively from the mantle.
The present mantle abundances are produced by addition of
later oxidized components that would only partly equilibrate
with core metal (Wanke eta!., 1994; O'Neill, 1991). Further
IW-2.3 (1600°C) IW experimental results may permit us to distinguish between the
t global equilibrium model and the inhomogeneous accretion

* ---- ------- Pt
models.
The discrepancy between observed mantle abundances and
the mantle-core distribution derived from 1 atm metal-silicate
partition coefficients is extreme for elements with strong met-
-------------------lr allic character, such as the noble metals Ir, Os, Pt, Ru, Pd,
etc. These elements have metal-silicate partition coefficients in
excess of 10 6 (Figure E6) while core mantle equilibrium would
require partition coefficients of about 800.
· - - - - - - - - - - - - - - - - - - - - Os However, metal-silicate partition coefficients of the noble
·--- metals do not decrease sufficiently with either temperature or

~;~
with pressure in order to reach the equilibrium ratio (Borisov
eta!., 1994; Holzheid et a!., 1997). In addition ratios among
the highly siderophile elements in the Earth's mantle are essen-

Fe~~~ tially the same as in chondri tic meteorites. Thus the overabun-
dance of the highly siderophile elements in the Earth's mantle
is best explained by the addition of a late accretionary compo-
~~~---~a nent (about 0. 7% of CI-chondrites). This late chondri tic veneer
:-;;;-~--~
-------
-===~rMn was added to the Earth after the end of core formation and
thus without later equilibration with core metal. Segregation
- - - - - Si of core-forming metal had removed the highly siderophile
T = 1600'C ',, elements before the arrival of the late veneer. Recent analyses
T= 1300'C ', of the isotopic composition of Os demonstrate that the RejOs
(lr, Au, Pd, Pt} ........... Ta
ratio of the mantle of the Earth is within the narrow range of
T = 1400'C chondritic ratios (Meisel et a!., 1996), supporting the late
(Os)
veneer concept. It is thus clear that models of core formation
10-1 0 L-----L---'--'-----'---'-----'----'----'--~-~ are intimately related to the accretion history of the Earth.
11.5 10.5 9.5 8.5 7.5 New data on the timing of core formation obtained from
-log f02 e
W-isotope analyses 82 Hf decays with 1.5 million years half-
life into 182W) indicate that core formation occurred late or
Figure E6 Metal-silicate partition coefficients for a variety of lasted until about 60 million years after the formation of the
elements at one atmosphere (see O'Neill and Palme, 1997, for solar system. Similar information is obtained from Pb isotopes.
details). Full lines indicate ranges in f 0 covered by experiments. Removal of Pb into the core approximately 50 million years
Dashed lines reflect extrapolations. ' after accretion is consistent with the evolution of Pb isotopes
160 EARTH'S FORMATION AND GEOCHEMICAL EVOLUTION

in the history of the Earth (Lee and Halliday, 1995; Galer and are incompatible 4°K, 87 Rb, 147 Sm, 238 U, 232 Th and to a lesser
Goldstein, 1996). extent 176 Lu. They are often concentrated in small melt frac-
The relevance of siderophile elements in the mantle of the tions decoupled from the major mantle phases. The less incom-
Earth is currently an active area of research with ongoing patible Re and the compatible Os make the Re-Os system
experimental work, in particular at high pressures and temper- more useful for dating events involving the mantle.
atures. These studies are important because their results have Although oceanic crust is also formed by partial melting of
implications for models of the accretion of the Earth. the mantle it differs in two respects from continental crust.
First, partial melting occurs in a part of the mantle which has
Crust formation experienced an earlier event of crust formation and is therefore
depleted in incompatible elements. Second, oceanic crust is
The formation of the Earth's crust is the result of partial
through the plate tectonic cycle processed and reinjected into
melting of the mantle. The changes in chemistry associated
the mantle. It has a comparatively short life time of about
with crust formation are apparent from Figure E5. The partial
I billion years, whereas continental crust has a mean age of
melt is rich in Al and low in Mg, the residual mantle shows
about 2.23 billion years.
the opposite signature, low Al and high Mg. According to this
simple scheme two groups of elements may be distinguished:
The formation of continental crust (see Earth's
(a) incompatible elements, which are not accommodated in the
minerals of upper mantle rocks and therefore partition into continental crust)
the partial melt (similar to AI). These elements are enriched The first several hundred million years of the history of the
in the crust. (b) Compatible elements are well accommodated Earth, generally referred to as 'Hadean', are shrouded in dark-
in upper mantle minerals. They follow Mg and their concen- ness. Large parts of the Earth were partially or completely
trations in the crust are low (see Geochemical classification of molten as a result of heat generated by accretion, core forma-
elements). In Table E7 concentration ratios of elements tion, enhanced decay of radioactive isotopes and impacts of
between continental crust (see Earth's continental crust) and large bodies. The latter point is particularly noteworthy since
primitive upper mantle (McDonough and Sun, 1995) are given. we know from the ages of lunar rocks that many of the huge
These ratios reflect the degree of incompatibility. Typical lunar basins on the front side of the Moon were formed by
incompatible elements are Rb, Cs, U, Th, light REE, Ta and impacts of large planetesimals at around 3.9 Ga. The Earth
K, while Ni, Cr and Mg are compatible elements with low could hardly have escaped such a bombardment. However, in
concentrations in the crust. Most radionuclei used for dating contrast to the Moon traces of these impacts have not yet been

Table E7 Enrichment of elements in the bulk continental crust over the primitive mantle

BCC PUM BCC/PUM BCC PUM BCC/PUM

Rb 32 0.6 53.33 Dy 3.7 0.674 5.49

Cs 1 0.021 47.62 Ho 0.78 0.149 5.23
u 0.91 0.0203 44.83 Er 2.2 0.438 5.02
Th 3.5 0.0795 44.03 Yb 2.2 0.441 4.99
K 9100 240 37.92 Tm 0.32 0.068 4.71
Ba 250 6.6 37.88 y 20 4.3 4.65
w 1 0.029 34.48 In (ppb) 50 11 4.55
B 10 0.3 33.33 Ga 18 4 4.50
Ta I 0.037 27.03 Ti 5400 1205 4.48
La 16 0.648 24.69 Lu 0.3 0.0675 4.44
Bi (ppb) 60 2.5 24.00 AI(%) 8.41 2.35 3.58
Be 1.5 0.068 22.06 v 230 82 2.80
Mo I 0.05 20.00 Cu 75 30 2.50
As 1 0.05 20.00 Cd (ppb) 98 40 2.45
Ce 33 1.675 19.70 Ca (%) 5.29 2.53 2.09
Sn 2.5 0.13 19.23 Sc 30 16.2 1.85
Nb 11 0.658 16.72 Re (ppb) 0.5 0.28 1.79
Pr 3.9 0.254 15.35 Ge 1.6 1.1 1.45
Sr 260 19.9 13.07 Zn 80 55 1.45
Nd 16 1.25 12.80 Mn 1400 1045 1.34
Hf 3 0.283 10.60 Si (%) 26.8 21 1.28
Ag (ppb) 80 8 10.00 Fe(%) 7.07 6.26 1.13
Zr 100 10.5 9.52 Co 29 105 0.28
Sm 3.5 0.406 8.62 Pd (ppb) I 3.9 0.26
Na 23000 2670 8.61 Mg (%) 3.2 22.8 0.14
Li 13 1.6 8.13 Cr 185 2625 0.070
Eu 1.1 0.154 7.14 Ni 105 1960 0.054
Gd 3.3 0.544 6.07
Tb 0.6 0.099 6.06

Concentrations in ppm nnless otherwise indicated.

BCC, bulk continental crust (Taylor and McLennan, 1985).
PUM, primitive mantle (McDonough and Sun, 1995).
 EARTH'S FORMATION AND GEOCHEMICAL EVOLUTION 161

detected on the surface of the Earth. The oldest dated minerals material into the lower mantle and a corresponding reflux from
on the Earth so far are mineral grains of zircon (ZrSi0 4 ), the lower into the upper mantle (Christensen, 1995).
ranging in age from 4.0 to 4.3 Ga. These zircons are found as
inclusions in gneisses of younger age. However, as zircon crys- The complementary nature of crust and mantle
tallization probably occurred from relatively SiOrrich magmas Continental crust formed by partial melting of the mantle.
the presence of SiOrrich rocks as old as 4.3 Ga, or even older, This process extracted incompatible elements from the mantle
must be inferred. Also, from the rounded shape of these zircons into the crust and must therefore have affected the chemical
the presence of running water prior to 4 Ga has been inferred composition of a large fraction of the mantle. The chemical
(see Rogers, 1994, and references therein). Clearly some kind composition of the crust is rich in AI and Si but low in Mg,
of crust, although of unknown extent, must have existed prior complementary to the mantle from which it formed (Figure E5,
to 4 Ga. The oldest dated rocks on Earth are slightly younger Table E7). Although partial melting occurs at relatively shal-
than 4 Ga. Rocks with ages of around 3.8 Ga are common in low depth (from 70 to 30 km) by decompression melting, mate-
many of the Earth's Archean shields. The total area of these rial from the deep interior, which is hotter and therefore lighter
Archean shields is small and the extent of this early crust is than the surrounding mantle of the Earth, will rise to the
unknown. The beginning of the growth of continental crust surface within mantle plumes. This and mantle mixing by
was probably at around 4 Ga. Crust formation was either convection will result in a general depletion of the mantle in
rapid and more or less complete at this time or it occurred by incompatible elements, including the heat-producing elements
episodic or continuous growth of new crust over geologic time, U, Th and K. The enrichment of the crust in incompatible
steadily increasing in volume. In the first case subduction of elements must be balanced by a corresponding depletion in
continental sediments is balanced by the accretion of island the mantle. For example, the estimated average U content of
arcs to the continents. However, current thinking is that geo- the crust is 0.91 ppm, compared to 20.3 ppb for the primitive
logic, geochemical and isotopic constraints favor more or less upper mantle of the Earth (PUM). This is the concentration
steady crustal growth, interrupted by periods of accelerated of U in a hypothetical chemically uniform upper mantle, i.e.
crust formation, e.g. from 3.8-3.5 Ga and 3.2-2.5 Ga. At before crust formation. The U crust-enrichment factor is then
around 2.5 Ga more than 50% of the present continental crust 45. The maximum enrichment factor is 150, corresponding to
may have formed (Taylor and McLennan, 1985). The present- the mantle-crust mass ratio. Thus either U has been completely
day production rate of continental crust is about 1 km 3 /year, extracted from one-third of the Earth's mantle and has the
low compared to oceanic crust formation encompassing a primordial abundance in the other two-thirds, or the mantle
volume of about 20 km 3 . Although crust formation occurred has lost one-third of its U more or less uniformly to the crust.
by partial melting of the mantle, major intracrustal melting Since upper mantle and transition zone together account for
processes may have had a significant impact on the composi- 27% some part of the lower mantle must have been involved
tion of the uppermost crust (Taylor and McLennan, 1985). in having delivered U to the crust, particularly when con-
sidering that samples recovered from the upper mantle contain
some U. The basic uncertainty in this argument is in the
The formation of the oceanic crust (see Earth's oceanic average crustal abundance which is difficult to obtain, con-
crust) sidering in addition that upper and lower crust differ in their
The plate tectonic cycle of oceanic crust formation and subduc- chemical composition (see Earth's continental crust). A more
tion involves three steps: formation of the crust at mid-ocean reliable estimate for the fraction of depleted mantle comes
ridges, movement of plates away from the ridges and subduc- from 40Ar in the Earth's atmosphere. Argon is an incompatible
tion of the cold lithospheric plates back into the mantle. The element and there is a net flux of Ar through mantle to the
atmosphere. The amount of atmospheric 40 Ar, a decay product
time scale for processing oceanic crust is 1-1.3 Ga. The
of 4 °K, requires that about 50% of the total 40Ar produced
descending oceanic plate will be gradually heated and dehy-
since 4.5 Ga is degassed from the mantle and that the rest is
drated. This triggers melting and gives rise to andesitic volca-
still within the mantle. More or less complete degassing of the
nism at convergent plate margins. Penetration of the
upper mantle requires that at least 23% of the lower mantle
lithospheric plate through the 660 km discontinuity into the
is, in addition, completely degassed and the rest has not lost
lower mantle has been postulated on the basis of seismic data,
any 40 Ar or that a larger fraction of the lower mantle has been
implying exchange of material between lower and upper mantle
affected to a lesser extent (Allegre eta!., 1996).
(van der Hilst et al., 1997). That the upper mantle is depleted in incompatible elements,
In addition, mantle plumes are sometimes thought to have as expected from crust formation, is amply documented by the
formed deep in the lower mantle, at the mantle-core interface composition of basalts presently erupting at mid-ocean ridges
(see Hofmann, 1997). They also have to cross the 660 km (MORB). The signature of incompatible elements in MORB
discontinuity and transport lower mantle material into the is characteristic of a source region depleted in incompatible
lithosphere, where melting occurs and gives rise to intraplate elements by an earlier event. A two-step model with melt
volcanism. This would suggest a whole mantle convection with extraction from a primitive mantle as a first step and a second
considerable mixing between lower and upper mantle. On the extraction from the depleted mantle successfully reproduces
other hand, the endothermic transition of y-phase to perovskite the composition of MORBs (Hofmann, 1988).
and magnesiowtistite at the upper mantle-lower mantle trans-
ition acts as a barrier for transport of material from the lower H. Palme
to the upper mantle (Christensen, 1995). This thus favors
layered convection in separate convection cells for the upper Bibliography
and lower mantle. Mixing of both systems may occur from Allegre, C.J., Hofmann, A.W. and O'Nions, R.K. (1996) The argon
time to time through catastrophic penetration of upper mantle constraints on mantle structure. J Geophys. Res., 23, 3555-7.
162 EARTH'S MANTLE GEOCHEMISTRY

Anders, E. and Grevesse, N. (1989) Abundances of the elements: Ringwood A.E. (1979) Origin of the Earth and Moon. Berlin:
Meteoritic and solar. Geochim. Cosmochim. Acta, 53, 197-214. Springer-Verlag.
Anderson, D.L. (1989) Theory o.lthe Earth. MA: Blackwell Brookline Rogers, J.J.W. ( 1993) A History of the Earth. Cambridge: Cambridge
Village, 388 pp. University Press, 295 pp.
Borisov, A., Palme, H. and Spettel, B. (1994) Solubility of palladium Stevenson, D.J. (1990) Fluid dynamics of core formation, in Origin of
in silicate melts: Implications for core formation in the Earth. the Earth (eds H.E. Newsom and J.H. Jones). New York: Oxford
Geochim. Cosmoclzim. Acta, 58, 705-16. University Press, pp. 231-49.
Boehler, R. ( 1996) Melting temperature of the Earth's mantle and Taylor, S.R. and McLennan, S.M. (1985) The Continental Crust: Its
core: Earth's thermal structure. Annu. Rev. Earth Planet. Sci., 24, Composition and Evolution. Oxford: Blackwell Scientific
15-40. Publications, 312 pp.
Christensen, U. ( 1995) Effects of phase transitions on mantle convec- van der Hilst, R.D., Widiyanatoro, S. and Engdahl, E.R. (1997)
tion. Annu. Rev. Earth Planet. Sci., 23, 65-87. Evidence for deep mantle circulation from global tomography.
Dreibus, G. and Palme, H. (1996) Cosmochemical constraints on the Nature, 386, 578-84.
sulfur content in the earth's core. Geochim. Cosmochim. Acta, 60, Waenke, H., Dreibus, G. and Jagoutz, E. (1984) Mantle chemistry and
1125-30. accretion history of the Earth, in Archean Geochemistry (ed.
Galer, S.J.G. and Goldstein, S.L. (1996) Influence of accretion on lead A. Kroener). Berlin: Springer-Verlag, pp. 1-24.
in the Earth, in Earth Processes, Reading the Isotope Code (eds Wetherill, G.W. ( 1994) Provenance of the terrestrial planets. Geochim.
A. Basu and S. Hart), Geophysical Monograph, American Cosmochim. Acta, 58,4513-20.
Geophysical Union, pp. 75-89.
Hofman, A.W. ( 1988) Chemical differentiation of the Earth: the rela-
tionship between mantle, continental crust, and oceanic crust. Earth Cross-references
Planet. Sci. Lett., 90, 297-314.
Hofmann, A.W. ( 1997) Mantle geochemistry: the message from oce- Earth's atmosphere
anic volcanism. Nature, 385, 219-29. Earth's continental crust
Holzheid A. and Palme H. (1996) The influence ofFeO on the solubil- Earth's core geochemistry
ity of Co and Ni in silicate melts. Geochim. Cosmochim. Acta, Earth's mantle geochemistry
60, 1181-93. Earth's ocean geochemistry
Holzheid A., Sylvester P., Palme H., O'Neill H.St.C. and Rubie D.C. Earth's oceanic crust
(1997) High pressure Pd-metaljsilicate partition coefficients: con- Elements: actinide series
firmation of the late veneer hypothesis. Lunar Planet. Sci., XXVIII, Elements: alkali and alkaline earth
587-8. Elements: distribution
Jagoutz, E., Palme, H., Baddenhausen, H. et a!. ( 1979) The abun- Elements: halogens
dances of major, minor and trace elements in the Earth's mantle as Elements: heat producing
derived from primitive ultramafic nodules. Proc. Lunar Planet. Sci. Elements: lanthanide series, rare earths
Conf, 10, 2031-50. Elements: radioactive
Kato, T., Ringwood., A.E. and Irifune, T. ( 1988) Experimental deter- Elements: trace
mination of element partitioning between silicate perovskites, gar- Geochemical classification of the elements
nets and liquids: constraints on early differentiation of the mantle. Geochemical tectonics
Earth Planet. Sci. Lett., 89, 123-45. Precambrian atmosphere
Li, J. and Agee, C.B. (1996) Geochemistry of mantle-core formation Precambrian geochemistry
at high pressure. Nature, 381, 686-9. Precambrian organic matter
Lee, D.-C. and Halliday, A.N. (1995) Hafnium-tungsten chronometry
and the timing of terrestrial core formation. Nature, 378, 771-4.
Manhes, G.C., Goepel, C.J. and Allegre, E. (1988) Systematique U-Pb
dans Ies inclusions refractaires d'AIIende: Ie plue vieux materiau
solaire, C.R. ATP Planetol., 323-7. EARTH'S MANTLE GEOCHEMISTRY
McDonough W.F. and Sun S.-s. ( 1995) The composition of the Earth.
Chem. Geol., 120, 223-53. Definition and scope of topic
Meisel, T., Walker, R.J. and Morgan, J.W. ( 1996) The osmium isotopic
composition of the Earth's primitive upper mantle. Nature, 383, The Earth's mantle is that part of the interior between the
517-20. outer liquid core and the crust. It extends from a depth of
Ohtani, E., Yurimoto, H. and Seto, S. ( 1997) The element partitioning ~ 2900 km to ~ 5 km beneath the oceans and ~ 30-50 km
between molten iron, silicate melt, and lower-mantle minerals:
implications for core formation of the Earth. Phys. Earth Planet. beneath the continents. It represents 67% of the Earth mass
Interiors, 100,97-114. with a density range from ~3.3 to 5.7 g/cm 3 .
O'Neill, H.St.C. (1991) The origin of the Moon and the early history The mantle may be subdivided into concentric shells, princi-
of the Earth - a chemical model. Part 2: the Earth. Geochim. pally composed of silicate minerals, defined by seismic velocity
Cosmochim. Acta, 55, 1159-72. discontinuities which reflect changes in mineralogy and/or bulk
O'Neill, H. St. C. and Palme, H. ( 1997) Composition of the silicate
Earth: implications for accretion and core formation, in: The
chemistry. The liquid core-mantle boundary is the Gutenberg
Earth's Mantle: Structure, Composition and Evolution - the discontinuity, initially considered to be the boundary between
Ringwood Volume (ed. I. Jackson). Cambridge: Cambridge the mantle and a solid core. The crust-mantle interface
University Press (in press). is a narrow zone of seismic discontinuity known as the
Palme, H. (1990) Back to the Earth's beginning. Nature, 343, 23-4. Mohorovicic discontinuity or 'Moho'. It marks an increase in
Palme, H. and Nickel, K.G. (1985) Ca/AI ratio and composition of the seismic velocities due to the mineralogical transitions from the
Earth's upper mantle. Geochim. Cosmochim. Acta, 49, 2123-32.
Palme, H., Larimer J.W. and Lipschutz, M.E. (1988) Moderately vola- mafic lower crust to the ultramafic mantle. The rigid uppermost
tile elements, in Meteorites and the Early Solar System (eds J.F. mantle and crust which undergo brittle failure form the litho-
Kerridge and M.S. Matthews). Tucson, University of Arizona sphere. The underlying mantle is substantially weaker than the
Press, pp. 436-61. lithosphere, may undergo plastic deformation, and comprises
Poirier, J.-P. (1994) Light elements in the earth's outer core: A critical the asthenosphere. The lithosphere and asthenosphere may be
review. Phys. Earth Planet. Interiors, 85, 319-37.
Righter, K. and Drake, M.J. (1997) Metal-silicate equilibrium in a separated by a zone where shear-wave velocities decrease sub-
homogeneously accreting earth: new results for Re. Earth Planet. stantially, known as the low-velocity zone, which may be par-
Sci. Lett., 146, 541-54. tially molten.
 EARTH'S MANTLE GEOCHEMISTRY 163

The thicknesses of the lithosphere and asthenosphere vary to depths of ~ 200 km because: ( 1) the mantle may be tectoni-
because of lateral changes in the depth of the low-velocity cally exposed at the Earth's surface in ultramafic massifs;
zone. At mid-ocean ridges the low-velocity zone is essentially (2) ultramafic xenoliths, which are direct samples of the
at the Earth's surface, thus the lithosphere is absent. Beneath mantle, are transported to the Earth's surface in ascending
old oceanic crust the low-velocity zone is at ~ 70 km depth, magmas in both oceanic and continental regions; ( 3) common
and 200-250 km beneath the continental cratons. However, in mafic and rare ultramafic magmas, which are the products of
some continental regions undergoing active extension, e.g. partial melting within the mantle at relatively shallow depths,
Basin and Range Province of the western USA, the low-veloc- are exposed at the Earth's surface; ( 4) conditions of temper-
ity zone appears to coincide with the Moho, thus the mantle ature and pressure within the mantle are easily attained in
lithosphere may be very thin or absent. experimental studies, thus allowing the testing of mantle
Rapid increases in seismic wave velocities at ~400 km and models; ( 5) mantle structure can be determined more precisely
~ 670 km depth are superimposed on a region of gradual by geophysical studies.
change in velocity gradient between 400 km and ~ 1000 km,
known as the transition zone which geophysically separates the Mantle chemistry
upper mantle from the lower mantle. However, thermal models The first-order separation of the crust of the Earth's lithosphere
frequently refer to upper and lower mantie convection (the into oceanic and continental parts also reflects lateral composi-
two-layer model) in which the upper/lower mantle boundary tional difference within the upper mantle. Consequently, it is
is at the 670 km discontinuity. The 400 km discontinuity has appropriate to discuss the oceanic and continental mantle sepa-
been interpreted as a phase transition from olivine to spinel rately. The discussion below frequently employs the concepts
and from orthopyroxene to majorite. The 670 km discontinuity of Earth chemical 'reservoirs', as further discussed by the
may represent the conversion of silicates to the perovskite Geochemical Earth Reference Model (GERM) Initiative
structure, but may also represent a mineralogically complex (WEB site: http://www-ep.es.llnl.govfgerm/germ-home.html ).
repository for lithospheric material recycled through subduc- These concepts have also been used to provide estimates of
tion zones. the bulk chemical composition of the Earth (Allegre et al.,
1995; McDonough and Sun, 1995; Tables E8 and E9).
Sources of information Estimates of the Earth's mantle composition rely on two classes
Geophysical data can provide fundamental information of model; the pyrolite model and the CI carbonaceous chon-
regarding the Earth's composition. A non-rigid globe with the drite model. The pyrolite model calculates an average composi-
same size, mass, and angular velocity as the Earth would have tion of the mantle based on xenolith studies then adds back a
a flattening of ~ 1/230 whereas the actual value is 1/298. This melt component, based upon the chemistry of basalts and
discrepancy can only be resolved if there is a large increase in ultramafic magmas. The CI chondrite model assumes the bulk
density toward the center of the Earth, and is thus consistent Earth accreted with the major element composition of CI
with a geochemically zoned Earth. carbonaceous chondrites but was depleted in volatile trace
Limitations can be placed on some aspects of upper mantle elements.
geochemistry from seismic data, e.g. Fe/Mg ratio, and by the
beginning of the 19th century the peridotitic and/or eclogitic The oceanic mantle
composition of the upper mantle had been demonstrated. Little The chemistry of modern oceanic mantle can be inferred indi-
is known about the transition zone or lower mantle, supposi- rectly at mid-ocean ridges where new lithosphere is created.
tions being based upon the combination of seismic information Beneath mid-ocean ridges, upwelling asthenosphere decom-
and ultra-high temperature and pressure experimental studies, presses adiabatically, and partially melts over a range of pres-
cosmochemical considerations, and elemental partitioning sures, depending upon the temperature at which the diapir
between the core and mantle (McFarlane and Drake, 1990). crosses the mantle solidus to produce mid-ocean ridge basalt
More is known about the geochemistry of the upper mantle (MORB). Off-ridge magmatism, associated with mantle hot

Table E8 Postulated upper mantle compositions

Si0 2 Ti0 2 Al 2 0 3 FeO MgO CaO K2 0 Na 2 0

Primitive ( 1) 46.38 0.18 4.1 7.62 38.11 3.24 0.03 0.33

Primitive (2) 44.5 0.15 2.55 8.65 41.7 2.25 0.12 0.25
Primitive (pyrolite-high Ti) (3) 43.95 0.57 3.88 8.1 39 2.6 0.22 0.6
Primitive (pyrolite-low Ti) (4) 45.6 0.2 3.3 8.1 38.5 3.1 0.4
Primitive ( 5) 46.2 0.23 4.75 7.7 35.5 4.36 0.4
Average garnet lherzolite ( 6) 45.89 0.09 1.57 6.91 43.46 1.16 0.12 0.16
Average spinel lherzolite ( 6) 44.2 0.13 2.05 8.29 42.21 1.92 0.06 0.27
Spinel lherzolite ( 7) 44.48 0.16 3.59 8.1 39.22 3.44 0.02 0.3
Depleted spinel lherzolite ( 8) 45.31 0.11 0.43 6.52 46.03 0.56 0.17 0.13
Depleted mantle (9) 46.32 0.17 3.94 7.65 38.44 3.19 0.01 0.28
Fertile mantle ( 10) 45 0.15 3.3 8 39.8 2.6 0.02 0.34
Primitive mantle (I 0) 49.9 0.16 3.64 8 35.1 2.89 0.02 0.34

(I) Hart and Zindler (1986); (2) Wylie (1971); (3) Green and Ringwood (1967); (4) Ringwood (1979); (5) Palme and Nickell985); (6) Maaloe and Aoki (1977);
(7) Takahashi (1986); (8) Carswell (1980); (9) Kinzler and Grove (1992); (10) Taylor and McLennan (1985).
164 EARTH'S MANTLE GEOCHEMISTRY

Table E9 Best estimate of chemical composition of silicate Earth based on the pyrolite model (after McDonough and Sun, 1995)

Pyrolite Pyrolite
(normalized (normalized
Element CI Pyrolite to Mg and CI) +/~ Element CI Pyrolite to Mg and CI) +/~

Li (ppm) 1.5 1.6 0.45 30 Pd 550 3.9 0.003 80

Be 0.025 0.068 1.16 20 Ag 200 8 0.017 F3
B 0.9 0.30 0.14 F2 Cd 710 40 0.024 30
c 35000 120 0.0015 F2 In 80 II 0.058 40
N 3180 2 0.0003 F2 Sn 1650 130 0.033 30
F 60 25 0.17 F2 Sb 140 5.5 0.017 50
Na 5100 2670 0.22 15 Te 2330 12 0.002 F2
Mg (%) 9.65 22.8 1.00 10 I 450 10 0.009 F3
AI(%) 0.860 2.35 1.16 10 Cs 190 21 0.047 40
Si (%) 10.65 21.0 0.83 10 Ba 2.40 6600 1.16 10
p (ppm) 1080 90 0.035 15 La 237 648 1.16 10
s 54000 250 0.002 20 Ce 613 1675 1.16 10
CI 680 17 0.011 F2 Pr 92.8 254 1.16 10
K 550 240 0.18 20 Nd 457 1250 1.16 10
Ca (%) 0.925 2.53 1.16 10 Sm 148 406 1.16 10
Sc 5.92 16.2 1.16 10 Eu 56.3 !54 1.16 10
Ti 440 1205 1.16 10 Gd 199 544 1.16 10
v 56 82 0.62 15 Tb 36.1 99 1.16 10
Cr 2650 2625 0.42 15 Dy 246 674 1.16 10
Mn 1920 1045 0.23 10 Ho 54.6 149 1.16 10
Fe(%) 18.1 6.26 0.15 10 Er 160 438 1.16 10
Co 500 105 0.089 10 Tm 24.7 68 1.16 10
Ni 10500 1960 0.079 10 Yb 161 441 1.16 10
Cu 120 30 0.11 15 Lu 24.6 67.5 1.16 10
Zn 310 55 0.075 15 Hf 103 283 1.16 10
Ga 9.2 4.0 0.18 10 Ta 13.6 37 1.16 15
Ge 31 1.1 0.015 15 w 93 29 0.13 F2
As 1.85 0.05 0.011 F2 Re 40 0.28 0.003 30
Se 21 0.075 0.002 70 Os 490 3.4 0.003 30
Br 3.57 0.050 0.006 F2 Ir 455 3.2 0.003 30
Rb 2.30 0.600 0.11 30 Pt 1010 7.1 0.003 30
Sr 7.25 19.9 1.16 10 An 140 1.0 0.003 F2
y 1.57 4.30 1.16 10 Hg 300 10 0.014 F4
Zr 3.82 10.5 1.16 10 Ti 140 3.5 0.011 40
Nb (ppb) 240 658 1.16 15 Pb 2470 ISO 0.026 20
Mo 900 50 0.024 40 Bi 110 2.5 0.010 30
Ru 710 5.0 0.003 30 Tb 29 79.5 1.16 15
Rh 130 0.9 0.003 40 u 7.4 20.3 1.16 20

From Li to Zr element concentrations are given in ppm; Nb to U are given in ppb; and Mg, AI, Si, Ca and Fe are in wt'Yo. The +I- column is a subjective
judgment of the uncertainty of this estimate. Uncertainties are expressed in %, unless otherwise stated; F ~factor (F2 ~we know this estimate to within a factor
of 2). Most of the major and minor elements and a number of the refractory lithophile elements are known to within ± 10% better.

spots, builds volcanic islands and seamounts composed of Ca, AI, Na, Ti. However, xenolith studies indicate that mantle
oceanic island basalt (OIB) and derivatives. Mantle xenoliths lherzolites have magmaphile trace element abundances, e.g.
are frequently associated with OIBs and are dominated by rare-earth elements, high field strength elements, that may vary
spinel harzburgites. Other rock types include spinel and garnet over at least an order of magnitude. These differences are
lherzolites, websterites, wehrlites and pyroxenites. Based upon reflected in the common use of'depleted' and 'enriched' (synon-
chemical studies of xenoliths and high-pressure experiments it ymous with 'fertile') mantle, although these are relative terms
is generally accepted that the uppermost oceanic mantle is that have not been rigorously defined. For instance, the terms
ultramafic. Mantle composition and mineralogy depend upon may be defined in terms of a reference mantle material having
the degree to which the mantle has been processed through chondritic abundances of magmaphile trace elements. These
partial melting. High-pressure experimental studies on ultra- observations are also consistent with the compositions of
mafic rocks containing alumina show that the mineralogy MORB and OIB, which show quite different magmaphile trace
changes from plagioclase-, to spinel- to garnet-peridotite as a element abundances that cannot be reconciled with a single
function of increasing pressure. All of these rock types are mantle source. The reasonable assumption is that MORB melts
observed as xenoliths implying that the upper mantle shows a are derived from mantle dominated by lherzolite relatively
similar mineralogic zonation as a function of depth. The major depleted in magmaphile trace elements whereas OIB melts are
element chemistry of all mantle-derived basalts indicates that derived from more enriched lherzolite. This is clearly evident
their source regions must be largely lherzolitic in mineralogy from an examination of rare-earth elements (REE; Figure E7)
to provide the required concentrations of major elements, e.g. in MORB and OIB. The overall REE concentrations in MORB
 EARTH'S MANTLE GEOCHEMISTRY 165

for MORB and OIB when plotted on Nd-Sr-Pb isotopic

1000
diagrams can only be explained by mixing of at least four
discrete mantle reservoirs (Allegre and Turcotte, 1985) defined
as follows: MORE or DM- a component characterized by
depletion of magmaphile trace elements (hence the acronym
DM for 'depleted mantle'), low 87 Sr/86 Sr and 206 Pbj2° 4 Pb, high
143 Nd/ 144 Nd that is apparent in mid-ocean ridge basalts (hence
100 .
the MORB acronym); HIMU- a component characterized by
~
"0
high 206 Pbj2° 4Pb, moderate 143 Nd/ 144 Nd, and low 87 Sr/86Sr.
This signature is observed in mantle-derived basaltic rocks
c
0
.c from a few oceanic islands, e.g. St. Helena, Tubuaii, Mangaia
&2
w and reflects high U/Pb, Th/Pb, and low Rb/Sr mantle environ-
w ments. The pseudo-acronym derives from the 'high p' values
~ 10
in these rocks, where p = 238 Uj2° 4Pb; EMJ and EM2 - two
'enriched mantle' (hence the acronym EM) components that
are characterized by 143 NdF 44 Nd and/or 87 Sr/86 Sr that are less
and more radiogenic, respectively, than the bulk silicate Earth
• TRANSITIONAL BASALTS (OIB) (BSE) values, indicating a high Rb/Sr and low Sm/Nd
environment.
• BASANITES (OIB)
Detailed studies of these isotope systems, together with
"' MID-OCEAN RIDGE BASALTS (MORB) 3He/ 4 He, and magmaphile trace elements indicate that the
HIMU component is probably partly altered oceanic litho-
La Ce Pr Nd Sm Eu Gd Tb Dy Ho Er Tm Yb
sphere recycled through subduction zones into the mantle; a
Figure E7 Log-normal distribution of rare earth element (REE) process that has probably occurred for most of Earth history,
concentrations in various types of oceanic basalt. The although the recycling times are poorly constrained (Hanan
concentrations have been normalized to average abundances in and Graham, 1996). However, the origin of the EM! and EM2
ordinary chondrites to eliminate the 'odd-even atomic number' components is less understood. EM! is commonly considered
distribution of the REE observed in nature. The depletion in light to reflect recycled terrigenous sediments or continental crust,
REE for MORB and the overall low concentrations of the REE imply
whereas EM2 may represent recycled pelagic sediments or
a mantle source strongly depleted in these elements. The LREE
fractionation observed in OIB cannot be produced through partial
continental lithosphere. These chemical components may
melting of the MORB source, but could be produced from a mantle reside in physically distinct reservoirs which implies some
source having approximately chondritic abundances of the REE. The degree of mantle stratification. High 3 He/4He values in some
depletion in heavy REE in OIB relative to MORB indicates the oceanic basalts suggest the involvement of the lower mantle
presence of garnet (which strongly fractionates the REE) in the OIB which may be less outgassed than the dynamic upper
source and, thus, partial melting at depths in excess of 70 km. mantle. Physical mixing may occur in ascending hot plumes
that arise at thermal boundary layers (the core-mantle bound-
are low, often approaching those of chondrites, and the chon- ary or a mantle seismic discontinuity) and may sample and
drite-normalized REE patterns show substantial depletions in mix the lower and upper mantle.
light REE. These patterns and overall concentrations can only
be explained if the mantle source has been depleted by one or The continental mantle
more partial melting events that have preferentially extracted The nuclei of the continental crust (cratons) and keels of thick
the light REE, and by inference other magmaphile elements. lithospheric mantle, approaching 200 km in thickness, formed
In contrast, many OIBs show a strong enrichment in REEs during the Archean and have been stable features of the Earth
and a preferential enrichment in light REE which imply an surface after that time. Extraction of the Archean crust from
undepleted, chondrite-like mantle source. the mantle, including large volumes of high-temperature ultra-
The isotopic compositions of MORB and OIB have been mafic melts (komatiites) involved the efficient transport of
studied for several decay schemes, e.g. 87 Rb- 87 Sr, 147 Sm- 143 Nd, magmaphile elements out of the upper mantle. The magnesian-
23su_zo6Pb, 23su_zo1pb, z32Th-zosPb, n6Lu-176Hf. The ranges rich, harzburgitic, low density, Archean lithospheric keel was
of Sr and Nd isotope values, from a variety of tectonic settings, gravitationally stabilized within the mantle, which may account
are shown in Figure E8. The radiogenic isotopes in basalts are for the enduring nature of the cratonic crust. However, post-
particularly instructive in understanding mantle chemistry and Archean magmatic processes have radically changed the chem-
evolution because individual isotopes are not fractionated istry and mineralogy of the sub-cratonic harzburgitic litho-
during melting or crystal fractionation. Various trace element sphere. These processes involve the transfer of fluids and low-
ratios may also provide petrogenetic information as substitutes volume melts from the underlying asthenosphere, and have
for radiogenic isotopes (Weaver, 1991 ). The remarkable diver- become better understood through the study of xenoliths,
sity exhibited in Figure E8 reflects isotopic variability in the kimberlites and their inclusions, especially diamonds (Navon
mantle source regions and indicates that the upper mantle is et al., 1988). Infiltration of small volume, alkaline, gas-rich
isotopically heterogeneous at various scales (Zindler and Hart, fluids from the asthenosphere, or even the lower mantle, has
1986). Based on the isotopic systematics of MORB and OIB, caused extensive metasomatism and introduction of new mag-
which represent mantle-derived silicate melts, variability has maphilic elements. New material has also been added in the
been explained by mixing of discrete mantle components, form of basaltic melts, perhaps derived through plume magma-
sometimes referred to as 'reservoirs'. The physical meaning of tism, that have suffered heat death within the thick Archean
each component remains unclear. The scatter of data points lithosphere.
166 EARTH'S MANTLE GEOCHEMISTRY

143 144 SR-ND MANTLE ARRAY FOR MORB AND om

ND/ ND
Mij'd-Adantic Ridge
0.5133 IIJ.!·-----,___ Iceland
: --.'
0.5132

0.5131 Hawaiian Galapagos Islands

Society
0.5130 ;Islands

0.5129

0.5128 Canary
Islands

0.5127 St. Helena

Koolau
0.5126 (Hawaii)

0.5125

0.5124
Tristan
0.5123 daCunha

0.7025 0.7035 0.7045 0.7055 0.7065

0.7030 0.7040 0.7050 0.7060

87 86
SR/ SR
Figure E8 The range of 143 Nd/' 44 Nd and 87 Sr/" 6 Sr observed in MORB and OIB. The rocks are all sufficiently young that growth of 143 Nd and
87 Srafter the basalts were erupted can be ignored. Hence, the measured isotopic values represent the mantle values at the site of melting.
The roughly linear array of MORB samples is extended considerably by the addition of OIB samples. However, the dispersion from a linear
trend requires mixing of more than two mantle sources with different Nd and Sr isotopic compositions. The approximate 'end-member'
compositions (DM, HIMU, EM1, EM2) are shown, but the physical meaning of these components remains debatable. The substantial ranges
in both Nd and Sr isotope values, provide a clear indication of mantle isotopic heterogeneity at both regional and local scales. The
negative correlation between 143 Nd!' 44 Nd and 87 Sr/86 Sr reflects the integrated effects of melt extraction on mantle geochemistry. Rb and Nd
are more magmaphile than Sr and Sm so that Rb/Sr ratios are lowered and Sm/Nd ratios are increased in residual mantle. The long half-
lives of these isotopic systems require that mantle variations in Sm/Nd and Rb/Sr must have existed over long periods of geologic time
(probably measured in billions of years) to produce the observed isotopic variations.

A peculiarity of Archean magmatic processes was the extrac- Studies of ratios of elements that have disparate geochemical
tion of large volumes of ultramafic magma (komatiite) from behavior within the Earth provide important information on
the upper mantle, together with komatiitic basalt and tholeiite. the early evolution of the mantle. For instance, P and Ti have
These rocks are preserved in greenstone belts, albeit commonly similar partitioning behavior during partial melting of the
fragmented, deformed and metamorphosed. Greenstone silicate mantle, thus the Ti/P ratio remains largely unmodified
assemblages > 3.0 Ga are rare, e.g. Pilbara Block, Western during such events. In contrast, P is a strongly siderophilic
Australia (3.55 Ga); Barberton, South Africa (3.46 Ga); element, whereas Ti is not Assuming that the Earth accreted
Minnesota River terrain, USA (3.3-3.6 Ga), but preservation with chondritic abundances of the elements, then the Ti/P ratio
of younger Archean greenstone assemblages is substantially would change dramatically during formation and separation
better, e.g. Abitibi Belt, Canada (2.7 Ga); Kambalda, Western of an immiscible metal phase. Archean and Proterozoic komat-
Australia (2.7 Ga). A major change in Earth processes iites and basalts have Ti/P ratios of~ 10 (Sun, 1987) compared
occurred after the formation of Archean lithosphere. to chondritic values of ~I, indicating that the iron core must
Ultramafic rocks are rare in Proterozoic and younger assem- have been completely separated from the mantle prior to the
blages suggesting a decrease in mantle heat flow, and develop- time of Archean volcanism.
ment of post-Archean crust dominated by subduction-related Radiogenic isotope studies of Archean and post-Archean
processes. Consequently, mantle beneath post-Archean crust mantle-derived volcanics have played a key role in understand-
is fundamentally different from Archean-age mantle and has ing the early melting history of the continental asthenosphere.
the chemical characteristics associated with the modern oceanic Post-emplacement metamorphism of Archean and Proterozoic
upper mantle. lavas renders Rb-Sr isotopic studies suspect because of the
 EARTH'S MANTLE GEOCHEMISTRY 167

potential mobility, and therefore fractionation, of these ele- A similar argument applies to the low 87 Sr/86 Sr ratios of mid-
ments during alteration. The trivalent REE are generally ocean ridge basalts.
immobile during chemical alteration, or at least are not easily Direct evidence for metasomatism can be found in mantle-
fractionated from each other, thus the radiogenic systematics derived composite xenoliths composed of Type I assemblages
of the Sm-Nd system provide useful information regarding that contain Cr-diopside and Type II assemblages that contain
early mantle evolution. 143 Nd is a radiogenic daughter product Al-augite. Type I xenoliths represent mantle ultramafic assem-
of 147 Sm with a radiogenic half-life of 1.06 x 10 11 years; thus, blages, whereas Type II xenoliths are interpreted as assem-
the 143 Ndj 144 Nd ratio may be a sensitive indicator of mantle blages produced by the high-pressure crystallization of basaltic
processes occurring over a long period of mantle evolution. melts (Frey and Prinz, 1978). Type I selvages to Type II
The 143 Ndjl 44 Nd of the system under study is commonly com- xenoliths may show evidence for melt-related metasomatism.
pared to the time-integrated 143 Nd/ 144 Nd of the bulk silicate Relative to discrete Type I xenoliths, i.e. those not physically
Earth, and is expressed in 8 units (parts per ten thousand associated with Type II xenoliths, metasomatism causes
difference from the bulk silicate Earth value). The bulk silicate increases in CaO, Al 2 0 3 , Ti0 2 and the FejMg ratio, but may
Earth e value is always zero, because the notation is time- show increases in alkali elements only if these are introduced
integrated and thus requires that the age of the system under in new hydrous phases, e.g. amphibole and mica.
study be known. Fluid inclusions within mantle minerals, e.g. olivine, dia-
Many continental basalts, of different ages, have positive mond, provide evidence for C0 2 -bearing fluids within the
e values indicating that the mantle source areas beneath the mantle, that would be highly reactive with mantle silicate
continents have had Sm/Nd ratios greater than the bulk Earth minerals. C0 2 may be dissolved in silicate and carbonatite
value (assumed to be chondri tic) for sufficient time to measur- melts, in supercritical fluids, or in carbonate minerals (Eggler,
ably increase the 143 Nd/ 144 Nd ratios above the time-integrated 1987). C0 2 -rich fluids have the potential for decoupling the
chondritic value (DePaolo, 1988). This observation implies REE from other lithophile trace elements by stabilizing REE
that the mantle must have lost a component with a Sm/Nd carbonate complexes, whereas H 2 0-rich fluids have no such
ratio substantially less than the chondritic values, although the effect.
data can be interpreted in terms of a gradual loss of a low
Sm/Nd component or a major mantle event at ~ 1.5-2.0 Ga.
Evidence for the former comes from studies of Archean, Synopsis
Proterozoic and younger mantle-derived lavas. Archean The chemistry of the Earth's mantle has been evolving from
komatiites have small, but measurable, positive e values and the formation of the planet at ~4.56 Ga. Current models of
these values increase in younger lavas. Even the 3.8 Ga meta- the early chemical evolution of a chondritic mantle, that
volcanics from !sua, Greenland have 8 ~ + 2. Collectively, assume an initially molten Earth, predict large scale chemical
these data indicate that the mantle had undergone substantial and mineralogic fractionation during cooling and crystalliza-
fractionation prior to 3.8 Ga, and this process has been con-
tion. If correct, then the lower mantle is relatively silica-rich
tinuous for much of Earth's history.
and dominated by MgSi0 3 perovskite, whereas the upper
mantle is relatively silica-poor. Direct information from xeno-
Mantle metasomatism liths, and indirect information from mantle-derived partial
There is clear physical evidence from mantle xenoliths, and melts (komatiites and basalts) provide constraints on upper
chemical evidence from trace elements and radiogenic isotopes mantle composition from the Archean (younger than
~ 3.8 Ga) to the present. These studies indicate the develop-
in mantle-derived volcanics, that both the oceanic mantle
(Le Roux, 1987) and continental mantle (Kempton, 1987) ment of a highly refractory, thick Archean lithospheric keel
have been subjected to metasomatism. Mantle metasomatism, beneath the continental cratons that formed in response to
using a broad definition, includes alteration processes brought high heat flow and removal of large volumes of komatiitic
about by chemical and isotopic interaction of mantle minerals melt. The chemical composition of Proterozoic and younger
with silicate-rich liquids (melts) and/or hydrothermal fluids continental lithosphere appears to be similar to modern oceanic
that may be enriched in H 20 and C0 2 , or both. Thus, metaso- lithosphere, implying a lower global heat flow and crustal
matism is distinct from the physical emplacement of basaltic recycling through subduction zones.
magmas as dykes, veins, and irregular bodies within the mantle, Chemical and isotopic heterogeneities can be documented
although this process also changes the bulk chemical and iso- within the upper mantle since at least the Archean. These
topic composition of the mantle. The resulting metasomatism observations testify to the thermodynamic nature of the mantle
may be 'patent' (synonymous with 'modal') in which there is resulting in mass and energy transfer. Materials may be trans-
visible petrographic evidence for metasomatism, or 'cryptic', ported from the lower to the upper mantle as a result of plume
in which there are trace element and/or isotopic changes unac- activity, and from the asthenosphere to the lithosphere, result-
companied by mineralogic changes (Dawson, 1984). ing in extensive metasomatism of the latter. In this manner
The decoupling of trace element abundance and isotopic 'barren' lithosphere may be chemically rejuvenated. Surface
composition in mantle-derived volcanics provides indirect evi- processes thermally driven by the motion of the asthenosphere
dence for mantle metasomatism. For instance, oceanic and result in recycling of fresh and altered lithosphere. This may
continental alkali basalts are characterized by strong light rare- be the single most important mechanism for developing chemi-
earth element (LREE) enrichment relative to the heavy rare- cal and isotopic heterogeneities within the mantle.
earth elements (HREE), but are enriched in 143 Nd relative to The nature, scale, and origin of elemental and isotopic
the bulk silicate Earth. This trace element-isotope decoupling heterogeneity within the mantle continues to be a productive
implies a relatively recent (probably < 200 Ma) introduction research focus, and developing quantitative models that
of LREE-enriched material into the LREE-depleted mantle. account for multi-isotope variations is a challenge. Observed
168 EARTH'S OCEAN GEOCHEMISTRY

isotopic vanatrons in mantle-derived rocks require the exis- Takahashi, E. ( 1986) Melting of a dry peridotite KLB-1 up to 14 Gpa:
tence of multiple 'end-member' compositions, although their implications on the origin of the peridotitic upper mantle.
J. Geophys. Res., 91, 9367-82.
number and physical meaning remain unclear. Isotopic varia- Taylor, S.R. and McLennan, S.M. (1985) The Continental Crust: its
tions on the scale of kilometers are being documented, particu- Composition and Evolution. Oxford: Blackwell Scientific
larly along mid-ocean ridges, and their explanation requires Publications, 312 pp.
the cooperation of geochemistry and high-resolution geophys- Weaver, B.L. (1991) The origin of oceanic island basalt end-member
ics. The role of the continental crust in determining the chemis- compositions: trace element and isotopic constraints. Earth Planet.
Sci. Lett., 104, 381-97.
try of mantle-derived melts, and hence the ability to predict
Wyllie, P.J. (1971) The Dynamic Earth. New York: John Wiley and
mantle chemistry, is a contentious subject that has raised the Sons. 416 pp.
question of crustal contamination in ocean island basalts. Zindler, A and Hart, S. (1986) Chemical Geodynamics. Annu. Rev.
Earth Planet. Sci., 14, 493-571.
W. Ian Ridley
Cross-references
Bibliography
Earth's continental crust
Allegre, C.J. and Tnrcotte, D.L. (1985) Geodynamic mixing in the Earth's core geochemistry
mesosphere boundary layer and the origin of oceanic islands. Earth's formation and geochemical evolution
Geophys. Res. Lett., 12,207-10. Earth's oceanic crust
Allegre, C.J., Ponrier, J.-P., Rumler, E. and Hofmann, A.W. (1995) Elements: alkali and alkaline earth
The chemical composition of the Earth. Earth Planet. Sci. Lett., Elements: distribution
134, 515-26. Elements: halogens
Carswell, D.A. (1980) Mantle derived lherzolite nodules associated Fluid-rock interactions
with kimberlite, carbonatite and basaltic magmatism. Lithos, 13, Geochemical tectonics
121-38. Helium isotopes
Dawson, J. B. ( 1984) Contrasting types of upper man tie metasomatism, Mid-ocean ridge basalts (MORE)
in Kimberlites- II. The Mantle and Crust/Mantle Relationships (ed. Precambrian geochemistry
J. Kornprobst). Amsterdam: Elsevier, pp. 289-94. Volcanic gases
DePaolo, D.J. (1988) Neodymium Isotope Geochemistry. Berlin: Volcanism
Springer-Verlag, 187 pp.
Eggler, D.H. (1987) Solubility of major and trace elements in mantle
metasomatic fluids: experimental constraints, in Mantle
Metasomatism (eds M.A. Menzies and C.J. Hawkesworth). EARTH'S OCEAN GEOCHEMISTRY
London: Academic Press, pp. 21-41.
Frey, F.A. and Prinz, M. (1978) Ultramafic inclusions from San
Carlos, Arizona: petrologic and geochemical data bearing on their Major constituents of the ocean
petrogenesis. Earth Planet. Sci. Lett., 38, 129-76.
Green D.H. and Ringwood, A.E. ( 1967) The genesis of basaltic Composition and consequences
magmas. Contrib. Mineral. Petrol., 15, 103-90.
Hanan, B.B. and Graham, D.W. ( 1996) Lead and helium isotope evi-
Ever since the studies of Dittmar in the late 1890s it has been
dence from oceanic basalts for a common deep sonrce of mantle established that many of the major ionic constituents of sea
plumes. Science, 272, 991-5. water have a remarkably constant ratio to that of the chloride
Hart, S.R. and Zindler, A. ( 1986) In search of a bulk-earth composi- content. The latter is the major anion in sea water (558 mM),
tion. Chon Geol., 57, 247-67. with a mean oceanic residence time of well over 100 million
Kempton, P.D. ( 1987) Mineralogic and geochemical evidence for years. This provides a useful reference for the other dissolved
differing styles of metasomatism in spinel lherzolite xenoliths:
enriched mantle source regions of basalts? in Mantle Metasomatism constituents. The observed constancy is particularly true for
(eds M.A. Menzies and C.J. Hawkesworth). London: Academic the cations sodium, potassium, lithium, and magnesium, as
Press, pp. 45-89. well as for the anions bromide and sulfate. Table ElO gives the
Kinzler, R.J. and Grove, T.L. ( 1992) Primary magmas of mid-ocean typical concentrations (mmol/dm 3 ) of the major constituents
ridge basalts 2. Applications. J Geophys. Res., 97, 6907-26. of sea water. These constituents represent well over 99% of the
Le Roux, A.P. ( 1987) Source regions of mid-ocean ridge basalts: evi- dissolved salts.
dence for enrichment processes, in .Mantle Metasomatism (eds M.A.
Menzies and C.J. Hawkesworth). London: Academic Press. Some of the major constituents, however, are not uniformly
pp. 389-422. distributed in the ocean; they can vary with both location and
Maaloe, S. and Aoki, K. (1977) The major element composition of the depth. Typically, dissolved calcium in North Pacific deep
upper mantle estimated from the composition of lherzolites. waters is about 0.8% higher in its Ca/Cl ratio than in North
Contrib. Mineral. Petrol., 63, 161-73. Atlantic surface water (Figure E9). Alkalinity and total C0 2
McDonough, W.F. and Sun, S.-S. (1995) The composition of the
Earth. Chem. Geol., 120, 223-53.
contents may change by as much as 8 and 15%, respectively.
McFarlane, E.A. and Drake, M.J. ( 1990) Element partitioning and the
early thermal history of the Earth, in Origin of the Earth (eds H.E.
Newsom and J. H. Jones). New York: Oxford University Press, Table E10 Major ion composition of sea water
pp. 135-50.
Navon, 0. et al. (1988) Mantle-derived fluids in diamond micro-inclu- Cations (mM) Anions (mM)
sions. Nature, 335, 784-9.
Palme, H. and Nickel, K.G. (1985) Ca/Al ratio and composition of the Na+ 480 cl- 558
earth's upper mantle. Geochim. Cosmochim. Acta, 49, 2123-32. K+ 10.4 so~- 29
Ringwood, A. E. ( 1979) Origin of the Earth and Moon. New York: Li+ 0.027 HCO:J 2
Springer-Verlag, 295 pp. Ca 2 + 10.45 co~- 0.2
Sun, S.-S. ( 1987) Geochemical characteristics of Archean komatiites: Mg2+ 54 Br- 0.84
implications for the Earth and mantle evolution. J. Volcano!. Sr 2 + 0.087
Geotherm. Res., 32, 67-82.