208 Encyclopedia of the Solar System
and ocean island basalts—are used as strong arguments in
favor of the layered convection. New evidence, gathered
within the tomographic studies to be discussed later, gives
support to a significant impedance to the flow between the
upper mantle and lower mantle.
The whole mantle convection is favored by geodynam-
icists who develop kinematic and dynamic models of the
mantle flow. For example, the geometry and motions of
the known motions of the plates are much easier to explain
assuming whole mantle circulation. Evidence has been pre-
sented for penetration of slabs into the lower mantle, based
on the presence of fast velocity anomalies in the regions of
the past and current subduction. At the same time, there
is evidence for stagnation and “ponding” in the transition
zone of some of the subducted slabs. The recent results
from seismic tomography seem to support the concept of at
least partial separation of the upper and lower mantle flow.
In the early 1990s a model of mantle avalanches
was developed: the subducted material is temporarily
accumulated in the transition zone as the result of an en-
dothermic phase transformation at the 660-km disconti-
nuity. Once enough material with the negative buoyancy
collects, however, a penetration can occur in a “flushing
event,” where most of the accumulated material sinks into
the lower mantle. The calculations, originally performed in
two-dimensional geometry, indicated the possibility of such
events causing major upheavals in the Earth’s history. How-
ever, when calculations were extended to three-dimensional
spherical geometry, their distribution in space and time
turned out to be rather uniform.
The computer models of the mantle convection are
still tentative. There are many parameters that control the
process. Some, such as the generation of the plates and
plate boundaries at the surface, are difficult to model. Oth-
ers, such as the variation of the thermal expansion coef-
ficient with pressure-or temperature-dependent viscosity,
are poorly known; even one-dimensional viscosity variation
with depth is subject to major controversies.
The lower mantle appears mineralogically uniform, with
the possible exception of the uppermost and lowermost
100–150 km. There is a region of a steeper velocity gradient
in the depth range of 660–800 km, which may be an ex-
pression of the residual phase transformations. Also, at the
bottom of the mantle, there is a region of a nearly flat, pos-
sibly slightly negative gradient. This region, just above the
CMB, known as “D,” is the subject of intense research. Its
strongly varying properties, both radially and horizontally,
are being invoked in modeling mantle convection, chemical
interaction with the core, possible chemical heterogeneity
(enrichment in iron), and as evidence for partial melting.
In 2004, the existence of a new phase: “post-perovskite”
has been proposed; its existence may affect the complexi-
ties in the “D” region. The seismic velocities and density
throughout the bulk of the lower mantle appear to sat-
isfy the Adams–Williamson law, describing the properties
of the homogeneous material under an adiabatic increase in
pressure.6.5 Outer Core (2981–5151 km)
The outer core is liquid: it does not transmit shear waves.
Consideration of the average density and the moment of
inertia pointed to a structure with a core that would be
considerably heavier, possibly made of iron, judging from
cosmic abundances. We now know that the core is mostly
made of iron, with some 10% admixture of lighter elements,
needed to lower its density. It has formed relatively early in
the Earth’s history (first 50 Ma) in a melting event in which
droplets of iron gravitationally moved toward the center.
Though difficult to estimate, some current models place its
temperature in the range of 3000–5000K.
The presence of a liquid with a very high electrical con-
ductivity creates conditions favorable to self-excitation of a
magnetic dynamo. It is important to know that the magnetic
field we observe at the surface is only a small fraction of the
fields present in the core. Actually we see only one class of
the field: the poloidal, whereas the toroidal field, possibly
much stronger, is confined to the core.
Numerical models of the dynamo predicted several key
phenomena observed at the surface: the primary dipolar
structure with the alignment of the dipole axis close to the
axis or rotation of the Earth, the westward drift of secular
variations, and reversals of the polarity of the magnetic field.
The later phenomenon is the cause of the magnetic anoma-
lies on the ocean floor, which allowed estimating the rate
of ocean spreading. The two most widely known models of
dynamo are quite different in detail, with one by Glatzmaier
and Roberts having the strongest field deep in the core, and
one by Kwang and Bloxham being the strongest near the
surface of the outer core.
Seismological data are consistent with the model of the
core as that of a homogeneous fluid under adiabatic temper-
ature conditions. As often near major discontinuities, there
is some difficulty with pinning down the values near the
end of the interval: just below the CMB and just above the
inner core boundary (ICB).6.6 Inner Core (5251–6371 km)
An additional seismic discontinuity deep inside the core,
which came to be called the inner core, was discovered by
Inge Lehmann in 1936. The fact that it is solid was pos-
tulated soon afterwards, but satisfactory proof awaited an-
other 35 years, when observations and analysis of the free
oscillations of the Earth showed that it indeed must have a
finite rigidity.
It is believed that the inner core formed during the his-
tory of the Earth, perhaps some 2 Ga ago. As the Earth was
cooling, the temperatures at the Earth’s center dropped be-
low the melting point of iron (at the pressure of 330 GPa)