Geoscience Reference
In-Depth Information
become denser than other upper-mantle assem-
blages. Garnet-poor and olivine-rich residues or
cumulates are likely to remain at the top of
the upper mantle since they are less dense than
parental peridotites and do not undergo phase
changes in the upper 300 km of the mantle.
The zero-pressure density of a typical eclogite is
about 15% and 3% denser than basalts and fer-
tile peridotites, respectively. With a coefficient
of thermal expansion of 3
STRATI-
FIED
MANTLE
GARNET
PERIDOTITE
MODEL
ECLOGITE
MODEL
0
ol
+
plaq
3 g /cm
3
cpx
+
BASALT
pyroxenite 3.25 g/cm
3
ol
+
opx
+
ga
3.4 g/cm
3
cpx
+
100
eclogite
3.45 g/cm
3
ol
+
10
−
5
◦
C, it would
require temperature differences of 1000--5000
◦
C
to generate similar density contrasts by thermal
effects alone, as in normal thermal convection,
or to overcome the density contrasts in these
assemblages. Some eclogites are less dense than
some mantle peridotites at depths greater than
some 200 km. If the intrinsic density of the deep
mantle is only 1% greater than the shallower lay-
ers, it could be permanently trapped because of
the very low expansivity at high pressure.
Simple Stokes' Law calculations show that
inhomogeneities having density contrasts of 0.1
to 0.4 g/cm
3
and dimensions of 10 km will sep-
arate from the surrounding mantle at velocities
of 0.5 to 2.5 m/yr in a mantle of viscosity 10
20
poises. This is orders of magnitude faster than
average convective velocities. Inhomogeneities of
that magnitude are generated by partial melting
as material is brought across the solidus in the
normal course of mantle convection. The higher
mantle temperatures in the past make partial
melting in rising convection cells even more
likely and the lowered viscosity makes separa-
tion even more efficient. It seems unlikely, there-
fore, that chemical inhomogeneities can survive
as blobs entrained in mantle flow for the long
periods of time indicated by the isotopic data.
Gravitational separation is more likely, and this
leads to a chemically stratified mantle like that
shown in Figures 25.3 and 25.4. The unlikely alter-
native is that the reservoirs differ in trace ele-
ments but not major elements, intrinsic density
or melting point. Small differences in bulk chem-
istry change the mineralogy and therefore the
intrinsic density and melting point; mineralogy
is more important than temperature in generat-
ing density inhomogeneities.
The density differences among basalt, deple-
×
200
+
pyroxene s.s.
+
ga s.s.
ga s.s.
+
cpx s.s.
300
3.6 g/cm
3
r
=
3.45
−
400
garnet s.s.
3.45 g/cm
3
500
b
spinel
−
+
ga s.s.
600
3.6 g/cm
3
3.6 g/cm
3
g
−
ga s.s.
3.7 g/cm
3
sp
+
700
perovskite
(Mg, Fe)0
4.0 g/cm
3
+
ilmenite
3.8 g/cm
3
Fig. 25.2
Approximate densities of basalt/eclogite (left) and
garnet peridotite (right). Eclogite (subducted oceanic
lithosphere or a cumulate in a deep magma ocean) is denser
than peridotite until olivine converts to
β
-spinel. Below some
400 km the garnet and clinopyroxene in eclogite convert to
garnet solid solution. This is stable to very high pressure,
giving the mineralogical model shown in the center, the
gravitationally stable configuration. The heavy line indicates
that the bulk chemistry varies with depth (eclogite or
peridotite).
peridotite. Seismic tomography and the abrupt
termination of earthquakes near 670 km depth
suggests that oceanic lithosphere can penetrate
to this depth.
Intrinsic density increases in the order basalt,
picrite, depleted peridotite, fertile peridotite,
garnet-rich eclogite. Some eclogites are less dense
than some peridotites. Basalts and picrites crys-
tallizing or recrystallizing below about 50 km can
ted
peridotite,
fertile
peridotite
and
eclogite
