Composition of the mantle
What is the Earth's mantle made of?
The Earth's mantle is the starting point for nearly all considerations of the evolution of the Earth. As the product on the one hand of accretion and core segregation, it is ultimately derived somehow from the original material of the solar system and thus shares some#kinship with meteorites. On the other hand, the Earth's mantle has a long history since the formation of the Earth that has also influenced its present composition. Nonetheless, the concept of "the" primitive mantle as an important benchmark and starting point is still an important part of Earth science. Since we can't grind the earth up and analyze it, we must make an educated guess, based on the clues we have at hand.
Constraints from xenoliths and orogenic lherzolites
Volcanoes frequently entrain fragments of their wallrocks as they ascend through the crust and mantle. Aside from clearly crustal lithologies (e.g., granite gneiss) the xenoliths with high P-T mineral assemblages fall into two categories, eclogitic and peridotitic. The second constraint we have on the composition of the mantle is that it is "that which melts to form basalt". Though as we will see, that encompasses a large range of compositions, it is already an important observation and constraint. this line of reasoning ruled out eclogite as a mantle composition, since eclogite melts to form tonalite, not basalt. The melting source region of basalts must be in peridotitic material.
Another constraint comes from peridotite massifs thought to have been residues of melting. Figure 1 shows compositional trends in residual peridotite compositions from a variety of sources. These compositional trends are consistent with the extraction of a partial melt from some original compositions, but where do we place the primitive mantle in these diagrams?
Figure 1: Chemical trends in residual peridotites. Open circles are abyssal peridotites, Crossed squares are spinel lherzolite xenoliths, remaining symbols are residual compositions from melting experiments.
A logical approach would be to assign (much as is done with basalts) the most residual (i.e., Al-rich) composition as being the primary composition, from which all others are derived. But what if al mantle rocks undergo a certain minimum degree of melting? The primitive mantle could be much more Al-rich still, even to the point of being nearly basaltic, and there would be no way of detecting it. By the same token, what if some of the samples depicted started out as fertile peridotites that were infiltrated by basalt. Then the primitive mantle would have lower Al than the most Al-rich samples. Clearly, the primitive mantle should lie on these trends, but the trends themselves do not tell us where.
Constraints from seismology
One of the major constraints we have on the Earth's composition is the physical properties of the Earth's interior, as deduced from seismic data. Figure 2 shows the seismic velocity and density profiles of the Earth. Any candidate chemical composition for the Earth must be one that reproduces these physical properties at appropriate pressure and temperatures.
Figure 2: Seismic velocity and density as a function of radius.
Extraterrestrial evidence
The key to the use of extraterrestrial chemical compositions as a key to the composition of the Earth is the observed fact that the bulk compositions of the sun and the most undifferentiated meteorites known (the CI chondrites) coincide completely. The chondrites themselves define a trend in composition space which intersects the trands depicted in Figure 1. It is a reasonable and important conclusion that the bulk composition of the Earth lies at this intersection. Figure 3 illustrates the critical features of this observation.
Figure 3. Mg/Si vs. Al/Si for xenoliths and Chondrites (after Jagoutz, et al., 1979).
The black dots define a differentiation trend analogous to the ones shown in Figure 1 for the residual mantle. This trend is at least partly due to the extraction of partial melt from the mantle. The fields below show the so-called "cosmochemical fractionation trend" for chondritic meteorites. Where the two trends intersect is then the preferred composition of the Earth – note the nearly exact match to the composition of Pyrolite proposed by Green (1966).
Pyrolite
As a starting point for the examination of melting in complex experimental systems, Ted Ringwood in the middle 1960’s proposed the composition he called Pyrolite. This made the assumption that taking an average residual peridotite composition and mixing it with enough primitive basalt to produce the "right" calcium and aluminum contents would reflect more or less the compostiion of the primitive upper mantle. This approach has the advantage that the compostion is easy to define and produce, and almost certainly does have relevance to the upper mantle. The drawback is a certain circularity, since the "right" Al and Ca concentrations are defined based on mantle rocks themselves.
"Famous" mantle compositions
Here are some of the better known mantle composition estimates from the literature.
| Hofmann (1988) | Hart and Zindler (1986) |
Pyrolite (Ringwood, 1966) |
Mason (1966) | Hutchinson, (1974) | |
| SiO2 | 45.96 | 46.38 | 45.2 | 48.1 | 45.0 |
| MgO | 37.78 | 38.12 | 37.5 | 31.1 | 39.0 |
| FeO | 7.54 | 7.619 | 8.0 | 12.7 | 8.0 |
| Al2O3 | 4.06 | 4.097 | 3.5 | 3.1 | 3.5 |
| CaO | 3.21 | 3.239 | 3.1 | 2.3 | 3.25 |
| Na2O | 0.332 | 0.333 | 0.57 | 1.1 | 0.28 |
| Cr2O3 | 0.43 | 0.55 | 0.41 | ||
| MnO | 0.14 | 0.42 | 0.11 | ||
| P2O5 | 0.06 | 0.34 | |||
| K2O | 0.032 | 0.13 | 0.12 | 0.04 | |
| TiO2 | 0.181 | 0.182 | 0.17 | 0.12 | 0.09 |
| NiO | 0.25 |
Table 1. Mantle bulk major element compositions.
Mantle Mineralogy
The subsolidus mineralogy of the upper mantle can be seen directly in peridotite xenoliths in basalts, and can be experimentally verified in experiments. The uppermost mantle is comprised of a peridotite with about 60% olivine 30% enstatite 6% clinopyroxene, 2% plagioclase and 2% spinel. At slightly greater depth, the plagioclase disappears and is replaced by spinel and clinopyroxene. Deeper still, garnet appears. These phase relations are illustrated in Figure 4.
Figure 4. Mantle phase relationships. The solid lines shown above the solidus in Figure 4 show the disappearances of phases during adiabatic fractional melting.