The Deep Earth


Mean radius = 6371 km

Mass = 5.97 x 1024 kg

Density surface rocks = 2-3 x 103 kg m-3

Average density = 5.52 x 103 kgm-3

The deepest man made hole is about 12 km in the Russian Kola peninsula (1994).


In the Pechenga Ni-ore province (now known as an impact site!)



Volcanoes bring material to the surface from depths of ~100 km.


Diamonds may come from ~400 km.

diamond%20in%20kimberlite diamonds


The rest of our information about the Earth's interior must come from study of Earthquake (or nuclear) shock waves, complemented by cosmochemistry and mineral physics.


By studying how long it takes waves to pass through Earth (travel times) can work out a model for the internal structure of the planet.




Body waves - "P-waves" and "S-waves"

Vp = [K + 4/3 /ρ] 0.5               Vs = [  / ρ] 0.5

Vp,s = velocity; K = bulk modulus, a measure of how material compresses under P; = shear modulus; ρ = density

Since K > 0, Vp > Vs, P = primary. S = secondary or shear


Vp, etc as a function of depth can be obtained from travel-time curves:


The bending ray follows a ray path, which is characterised by the ray parameter (p), where:

p = r0/V0 = r sin(i)/V

This is the Benndorf relationship, which relates the distance from the centre of the Earth at which the ray starts to return to the surface (r0) and the speed it is traveling with at that depth (V0). These two terms then also define its speed at shallower depths (r), the angle of incidence (i) and the velocity (V) at that depth.


Seismic Structure of the Earth

In Earth, seismic velocity (V) varies with depth (Z). Seismic waves can be reflected and refracted at interfaces (cf light). The surfaces between layers are curved, but principles of refraction and reflection can be used to infer how V varies with Z, and the depths of layers.

Major result is a V v Z plot obtained from travel-times.


Vp and Vs tend to vary smoothly over a large part of the Earth but have a number of discontinuities.  

     Mohorovicic discontinuity:

~10-60 km - marks crust/upper mantle boundary.

At this depth there a change of seismic wave velocity and also a change in chemical composition.

Named after Andrija Mohorovičić, the Istrian seismologist who discovered it.


The boundary is ~25-60 km deep beneath the continents and ~5-8 km deep beneath the ocean floor. 



     Low velocity zone:

~ 50-200 km













Shear wave velocity profile showing LVZ beneath Tanzanian craton.


     Lehmann discontinuity:

~ 220 km depth. Increase in Vp and Vs by 3-4%. It may not be ubiquitous. It is sometimes called the after Inge Lehman (who more famously discovered the presence of an inner core).


     Transition zone:

~400-670 km with a number of sharp increases in Vp and Vs



     Lower mantle:

~670-2885 km monotonic increase in Vs and Vp.




At CMB have D an anomalous region just above the CMB with seismically fast and slow regions, Ultra Low Velocity zones (ULVZ) possibly due to partial melt; this region also possible slab graveyard, possible perovskite to post-perovskite transition, possible repository of primordial material area of very active research.


fuzzy core-mantle boundary better size and labels

     Core-mantle boundary (CMB):

~2885 km, Vs = 0, Vp drops.


Beno Gutenberg, who first established the depth of the CMB to be 2880km.


     Outer Core:

~2885-5145 km


Outer core liquid (S-waves not possible), and has a lower Vp velocity

The region that extends from 103 to 143 from the epicenter of an earthquake and is marked by the absence of P waves. The P-wave shadow zone is due to the refraction of seismic waves in the liquid outer core.

The region within an arc of 154 directly opposite an earthquake's epicenter that is marked by the absence of S waves. The S-wave shadow zone is due to the fact that S waves cannot penetrate the liquid outer core.

Lehmann saw P wave arrivals in P-wave shadow zone to infer presence of IC.


     Inner Core:

~5145-6371 km, Vp increases and Vs inferred > 0, -> solid inner core. Even inner core is not homogeneous. Layered and anisotropic.



Density of the Earth


From the seismic data, it is also possible to work out the density of the Earth as a function of depth, via data on K, g .


..and the Adams-Williamson relationship:




where F is the seismic parameter (= VF2 = K/r).

Thus, considering only density changes with depth in the Earth,


From the hydrostatic law,


The seismic parameter is defined by


However, it is also a known function of seismic S and P velocities, so it can be measured with depth. Plugging (2) and (3) into (1) gives


For smooth (but not necessarily for discontinuous) r(r), this can be integrated, using the total mass and moment of inertia as boundary conditions. This is the Adams-Williamson equation.



This shows that the density increases from about 3.3 g/cc in the upper mantle, and reaches about 5 g/cc at the CMB.




Here there is a major discontinuity with a jump to ~ 10 g/cc in the outer core and which rises to ~13 g/cc at the centre of the Earth.


The average density of the core is approximately 10.8 g/cc.

The inferred V and r curves are average values for a given depth.

This also gives P as a function of Depth:


Seismic Tomography

Now know that there are many seismic ray paths:




We can calculate how long they should take to travel certain paths PREM model (via Preliminary Reference earth Model):



When measured for any given earthquake, the waves may be faster or slower than expected:


A full detailed 3 dimensional set of travel time differences, gives a seismic tomographic image that reveals local variations in V and r due to variations in chemical composition or thermal structure.

Generally blue = faster = colder

Red = slower = hotter




The composition of the Earth

Seismology shows that the Earth is layered, that it is largely solid and crystalline (LVZ close to melting, outer core liquid), and that it has a complex but well defined density structure.

What are the chemical and mineralogical make up of the following layers:






transition zone






In general:
  - continental - inhomogeneous, high SiO2, t = 35 km (25-70 km), mixed ages (oldest 3,800 my)
  - oceanic - layered basalts and gabbros, t = 6 km, orderly in age and structure, young (< 200my)

In places Moho is a chemical change:


but it may also be due to a phase change (e.g. basalt -> eclogite)



The Upper Mantle

Upper mantle can be sampled directly via:-

 (i) Ophiolites and tectonic slices.


 (ii) Inclusions brought up in volcanics, kimberlites, etc.


Typical rock a garnet peridotite made of Mg,Fe silicates


   60% olivine (Mg,Fe)2SiO4
  18% orthopyroxene (Mg,Fe)SiO3
  12% garnet (Ca,Mg,Fe)3Al2Si2O12
  10% clinopyroxene Ca(Mg,Fe)Si2O6

Garnet peridotite (similar to experimentalists pyrolite) partially melts to give basalt and so is a suitable candidate for upper mantle on petrological grounds as we know basalt liquid comes from upper mantle to form oceanic crust.


The Upper Mantle is heterogeneous (with eclogite, dunite, etc.), because of melting to give basalts and residual rocks.

Density, Vp and Vs of minerals give an excellent fit to density, Vp and Vs of upper mantle, so garnet peridotite also satisfies geophysical constraints.


LVZ may be due to geotherm approaching the solids of slightly hydrous peridotite. Pre-melting gives rise to anomalous properties.


Transition Zone

At 400 km have discontinuity in Vp and Vs.  Density increases, this could be due to:-

 (1) same minerals but with higher molecular weight (i.e. more Fe, less Mg).
 (2) structural phase change to a more densely packed structure.
 (3) a combination of (1) and (2).


What happens to olivine if it is subjected to P + T of transition zone?

At about 120 kb + 1400C (400 km)

Forsterite -> Beta-Mg2SiO4 (wadsleyite)




At P, T about 550 km:

Beta-Mg2SiO4 -> Spinel-Mg2SiO4 (ringwoodite)




Wadsleyite and spinel are both spinelloid minerals.




Told apart by XRD:



Forsterite transforms to a denser polymorph at high P. 

Beta-phase and spinel-Mg2SiO4 are the minerals of the transition zone.

P/T of transformation match those of seismic discontinuity.


Density, Vp and Vs of beta- and spinel-Mg2SiO4 are exactly compatible with transition zone seismic data.


Also find in this P/T zone

pyroxene ->  garnet (majorite)


Lower Transition Zone about 60% spinel, 40% garnet structured (Mg, Fe) silicates.


Lower Mantle

More difficulty to be sure what is responsible for 670 km discontinuity.


P about 250 kbar  )       Difficult to achieve
T about 1800 C  )       by experiment


Can be obtained using a Multi Anvil Cell or the Diamond Anvil Cell and Laser heating (P > 1 Mbar; T > 3000 K). 


Both of these techniques are used in research in UCL-Bbk.

Multi-anvil cells need a large load frame:




Problem with MAC difficult to do in situ studies.

In situ possible in DAC, but very small sample volume (<10-2 mm3 for DAC).



High P generated by very small area of diamond tip (c.f. stiletto heels).


At P/T of 670 km discontinuity have spinel structure polymorph disproportionation to perovskite structure MgSiO3 + MgO. :

Mg2SiO4  ->    MgSiO3     +    MgO
spinel            perovskite        periclase


NB: Si coordination change: Si [IV] in spinel -> Si [VI] in perovskite


Also at ~25 GPa

Garnet  ->    Perovskite


Lower mantle composed of (Mg,Fe)SiO3 perovskite plus (Mg,Fe)O - magnesiowustite, plus minor phases such as CaSiO3-perovskite.


Ca-perovskite is cubic (or almost), Mg-peroskite is orthorhombic:


Because it is so difficult to make, we do not know Vs, Vp for MgSiO3 perovskite very well still the basis of active research.


670 km discontinuity is likely to be an isochemical phase transformation, but lower mantle could be richer in Fe or Si than transition zone.  Still not sure.


Other phases will occur in mantle because of subduction, etc, basalt in slab will change:



Have SiO2 phases here in LM subducted slab.


SiO2 phases are complex:


Core mantle boundary D is a complex region perhaps melting, perhaps reaction zone, perhaps slab grave yard.


Subject of active research with specific seismic ray paths, e.g.:


D probable origin of plumes:


*         ULVZ could be due to melting of SiO2 rich pods.

*         Whether there is a reaction between silicates and oxides of D and core depends on the chemistry of the core.

*         In 2004 a new phase transitions was found, when perovskite transforms to a post-perovskite phase (see Iitaka et al, 2004):

*         The structure is iso-structural with CaIrO3 and is characterised by having edge sharing SiO6 octahedra.


*         Perovskite will transform into the new phase at a P which corresponds to the D boundary (see Oganov et al 2004):

*         and Tsuchiya et al 2004:


Lower mantle now seen as:

PosterPov Mode

*         Still the subject of active research, but thought to explain reflector at 2650km in cold regions, and no reflections in hot regions:

Wookey Geotherm


*          D Vp, Vs, Vbulk and Density for hot and cold geotherms. Perovskite as solid lines. The effect of post-perovskite shown in dotted colour lines. PREM black dotted:

Wookey Vp-Vs

*          Note Vs and Vbulk anti-correlated. Vp not greatly affected.

*          D and post-perovskite only present in cold regions:


*          also not in early earth.


PosterPov Dprime


Believed to be Fe rich on basis of

 (1) Cosmic abundances.



 (2) Iron meteorites.


 (3) Seismic Characteristics:


 (4) Metallic conductor to give magnetic field.


P + T of core very high.  P > 1.5 Mbar, T about 5000-6000 K. 

Phase diagrame of Fe, suggests that Fe in the core is hcp-Fe:



Vp, Vs of Fe at these pressures not easy to determine. 

Can be obtained from Shock-Wave experiments, inelastic scattering or theory:


Pure Fe considered too dense for outer core.  Must be alloyed with lower density elements - Si, S, C or O?

S found in iron meteorites.  Fe-S outer core fits density data for 9-12% S by wt.

Is the outer core - inner core boundary isochemical or is there any compositional change?

Shock data suggest IC a little less dense than pure Fe. 

Could be an Fe-Ni alloy (if meteorites).  In this case the OC/IC boundary is a chemical discontinuity.

Recently Alf et al suggest:

OC : 82 mole% Fe, 10% S, 8% O

IC: 89.5% Fe, 10% S, 0.5% O

Probable model is that the IC is crystallising from OC. T of ICB is close to T melt of Fe:


Crystallisation occurs as core cools below Tm.


Outer core is enriched in light elements as they are more soluble in liquid than solid Fe:


High P phase diagram not know in detail. Only low P studies:


Details of the core not well established and still open to revision (see recent paper by Vočadlo on Fe in the core).

Thermal structure of the Earth can be obtained from P-T points of discontinuities, linked to known phase relations:


Will return to this when we look at Heat in the Earth.


Click here for more detailed notes of data analysis of deep Earth seismic waves and here for more on structure.

Click here for a practical on Seismology & Earth Structure.

Back to contents