Historical Geology/Structure of the Earth
In this article we shall review some key facts on the structure of the Earth, and discuss how they are known. We shall refer back to facts already discussed in the articles on igneous rocks, seismic waves, and physical properties of rocks; readers may wish to refresh their memories on these subjects before reading further.
Structure of the Earth
The Earth can be divided by composition into the crust, mantle, and core, as shown to scale in the diagram to the right.
Crust comes in two varieties: continental and oceanic. Continental crust consists mainly of felsic rocks such as granite, and is about 30-50 km thick, varying from place to place; oceanic crust consists mainly of somewhat more mafic rocks such as basalt, and is about 5-10 km thick.
The nature of these rocks suggests the reason why there is a crust as such. The reader may at this point find it useful to review the main article on igneous rocks. To summarize the most important points, felsic rocks (i.e. those that have a high silica content) also have a lower density and a lower melting point than more mafic rocks.
The minerals known as feldspar, for example, form 60% of the Earth's crust; their melting points range from 600°C - 1000°C, depending on their exact chemical composition, and a density between 2.55 and 2.76. Compare this to the ultramafic olivine of the upper mantle, with a melting point ranging between 1200°C - 1900°C and a density of 3.27–3.37.
This immediately suggests why the Earth should have a crust as such: that is, a region chemically distinct from the mantle. The more felsic minerals in the mantle would melt, because they have a lower melting point than the major constituents of the mantle; they would rise, being less dense by virtue both of their composition and their molten state; once they had erupted on the surface and cooled, they would "float" on the mantle as a result of their lower density. So even if the earth started off in a relatively homogeneous state (as geologists think it did) this process, known as differentiation, would ensure that the Earth ended up with a crust of rocks composed of minerals more felsic than those in the mantle.
The mantle consists of a further 2890 kilometers of denser ultramafic rocks.
It is sometimes wrongly stated that the mantle consists of molten rock. We know that this cannot be true because S-waves pass through the mantle, which would not be possible if it was a fluid (see the main article on seismic waves for more details). One reason for the popular belief in a liquid mantle is that, after all, lava, a liquid, erupts from out of the mantle. However, this lava is produced by partial melting of the solid mantle. This, incidentally, is why it has a different composition from the mantle, being more felsic. Although the mantle is not liquid, it does flow: technically, it is a ductile solid, as explained in the article on physical properties of rocks.
The core is the innermost part of the Earth, having a radius of 3,400 km. It can be divided into the outer core, which is molten, and the inner core, with a radius of 1,220 km, which is solid. (This arrangement may seem strange at first, but recall that the inner core is at greater pressure and so will have a higher melting point.)
The core is made mainly of iron. As with the existence of the crust, this can be explained on the hypothesis of differentiation: just as the light substances rose to the top, so the denser substances would sink to the bottom.
The lithosphere and athenosphere
The division into crust, mantle, and core partitions the Earth according to the composition of the rocks. Sometime, however, it is more useful to group the crust and the uppermost layer of the mantle together as the lithosphere. What these have in common is that they are brittle and elastic, as opposed to the plastic and ductile rocks in the rest of the mantle. The lithosphere ranges in thickness from 40-200 km, varying from place to place; it is thicker under continental crust. The concept of the lithosphere is especially important in the context of plate tectonics, as the plates in plate tectonics are not (as is sometimes stated) plates of the Earth's crust: they are plates of the lithosphere.
The portion of the mantle immediately below the lithosphere is called the athenosphere. This is the weakest part of the mantle, because although it is at a lower temperature than the deeper rocks, it is also at a lower pressure.
How do we know?
We shall now, as usual in these articles, sketch out how the knowledge described in the preceding sections was obtained. Let us first of all consider the facts that constrain any attempts to make a model of the physical properties of the Earth's interior.
vP and vS
As was explained in the article on seismic waves, it is possible by studying earthquakes to discover the velocities of P- and S-waves at various points within the Earth. The results are summarized in the chart to the right.
As explained in the previous article, we call a property spherically symmetric if it varies only with the distance from the center of the Earth and not with latitude and longitude. Probably no geological property (except distance from the center itself) is exactly spherically symmetric; but many of them can be demonstrated to be very nearly so: in what follows we shall use "spherically symmetric" to mean "spherically symmetric to a good degree of approximation".
As discussed in our article on seismic waves, the velocities of P- and S-waves (vP and vS) are spherically symmetric properties. It would be very remarkable if this was the case, and yet the properties of the Earth on which these velocities depend was not.
In the case of density, there is good evidence that it is spherically symmetric: for if it was then the force of gravity at the surface of the Earth would be (to a good degree of approximation) the same at any point on the surface; which it is. So the evidence is that density (which we shall denote by the Greek letter ρ) must be spherically symmetric. It immediately follows that the same must be true for pressure, since this can be calculated from density.
Now consider the fact that for any point in the Earth the velocity of S-waves (vS) is given by the formula vS = √μ/ρ, where μ is the rigidity and ρ is the density. So given that vS and ρ are both spherically symmetric, it follows that μ must be also. Furthermore, the velocity of P-waves (vP) is gven by vP = √(κ + 4μ/3)/ρ. So given that vP, μ, and ρ are spherically symmetric, it follows that so is κ.
By reasoning of this sort, exploiting the inter-relatedness of the properties that interest us, we can show that they are all spherically symmetric. So to construct a first approximate model of the Earth, we only need to associate each depth within the Earth with a value for gravity, density, incompressibility, and so forth.
We know the strength of gravity at the surface of the Earth, because we can measure it directly; we also know the gravity at the center of the Earth, since in any spherically symmetric body this must be precisely 0. We know the temperature at the surface and the rate of heat flow. We know the pressure at the surface: 1 atmospheric pressure. We know the mass of the Earth, which can be easily deduced from experiments measuring Newton's constant G. Since we know this and the volume of the Earth, we also know its average density.
These all serve as constraints on any Earth model. For example, if we think we know a function relating density to depth, we can easily calculate what the surface gravity of the Earth should be if this function was correct.
We have already used the formulas relating the properties to deduce the spherical symmetry of some properties from the spherical symmetry of other properties. But the relationships between them allow us to be much more precise than that. For example, if we know that vS = √μ/ρ, then knowing the relationship between depth and vS, and the relationship between depth and ρ, we automatically know the relationship between depth and μ. And knowing this, and knowing the relationship between depth and vP, we can exploit the formula vP = √(κ + 4μ/3)/ρ to tell us the relationship between depth and κ — and so forth.
This means that it is not necessary or even possible for us to form separate hypotheses as to the values of the various physical properties that interest us. The values that we do know (vP and vS) place constraints on the values that we would like to know.
To summarize: any model of the values of physical variables within the Earth must be constrained by:
- The known values of vP and vS
- Boundary conditions
- Spherical symmetry
These constraints are sufficient for geologists to work out figures for pressure, density, gravity, incompressibility, and so forth. Some results are shown in the graphs to the right.
Here μ represents rigidity; κ incompressibility; P pressure; ρ density; and g the force of gravity.
Hypotheses about the mineral composition of the Earth must of course be constrained by our model of its physical properties: the minerals must have the right density, rigidity, etc to account for these properties.
To date, it has been possible to drill a little over 12 km into the crust and take samples and make temperature measurements. The results are that continental crust, beneath any layers of sediment that have been deposited, is indeed composed of felsic granites, gneisses and so forth; and the oceanic crust of more mafic rocks such as basalt and gabbro. We can also study sections of ocean crust that have been thrust up onto the land — ophiolites. These will be the subject of a subsequent article.
There are a number of clues to the composition of the mantle.
- Volcanic eruptions sometimes bear up to the surface fragments of peridotite; their broken jagged shapes indicate that they must have been torn from the parent rock by the force of the eruption. Unfortunately, such eruptions originate from a maximum depth of about 180 km, so they only give us a sampling of the upper mantle.
- Except at subduction zones (where volcanoes recycle the material of the crust) volcanoes and rifts generally emit basaltic lava; in can be experimentally shown that this is just what would be produced by partial melting of peridotite.
- The base of ophiolites is serpentinite, a rock produced from peridotite in the presence of heat and water; that is, under the conditions present in the upper mantle.
- Peridotite has the right density to account for the values of ρ inferred from the seismological data.
- Minerals change their phase with pressure: the same elements in the same proportion adopt a more compact configuration. For example, at the pressures found at a depth of about 400 km, the mineral olivine (the main mineral constituent of peridotite) changes to wadsleyite. The seismological data indicate just the change of density at this depth that we would expect if such a change of phase took place. The other discontinuities in the density of the upper mantle are explicable in a similar manner.
- We do not see any discontinuities that would definitely indicate the substitution below some depth in the mantle of some different material altogether.
Although a cautious geologist would not claim absolute certainty as to the composition of the lower mantle, there is general agreement that the mantle is made of more or less the same stuff from the top of the mantle down to the top of the core.
Finally, the core. The differentiation of the core into an outer liquid core and an inner solid core is based on the study of S- and P-waves. S-waves don't travel through the outer core, proving that it is liquid; P-waves travel faster through the inner core, showing an abrupt transition that corresponds well to a change of phase from solid to liquid.
Given the density of the core, it must be composed mainly of iron. To be sure, it could in principle be composed of a mixture of something much heavier than iron, such as gold, mixed with something much lighter. However, iron is the only element in the Earth, in the Solar system, or in the Universe generally that is both dense enough and common enough to account for the mass.
The iron probably contains an admixture of about 8% nickel, since these elements are usually found in association in these proportions. However, this iron-nickel mix would actually be somewhat too heavy to account properly for the mass of the Earth; it would follow that there must also be a proportion of lighter elements. The abundant light elements silicon and oxygen are favorite candidates for this role.