Geoscience Reference
In-Depth Information
and probably also CO
2
still have an important
effect on the melting regime (Asimow & Lang-
muir, 2003). Due to the considerable depression
of solidus temperatures, small amounts of hy-
drous melt already form before the dry solidus is
reached. While the corresponding melt fractions
are small, this mechanism allows incompatible
trace elements to be extracted from a much
larger volume of upper mantle than in the ab-
sence of water and melting begins at much greater
depth. Total melt production and crustal thick-
ness increase, while the mean extent of melting
decreases, since the melting occurs over a much
larger depth interval.
The seismic low-velocity zone (LVZ) in the up-
per mantle is probably due to the presence of a
small degree of partial melt (Lambert & Wyllie,
1970; Mierdel
et al
., 2007). While a local mini-
mum in seismic velocities can also be produced by
purely solid-state effects without the presence of
melt (Karato & Jung, 1998; Stixrude & Lithgow-
Bertelloni, 2005), the sharpness of shear wave
velocity drop at the upper boundary of the LVZ
together with the high and strongly anisotropic
electrical conductivity can probably only be ex-
plained by the presence of melt (Rychert
et al
.,
2005; Yoshino
et al
., 2008). Since the temperature
in the LVZ is below the dry peridotite solidus,
melting can only occur because solidus temper-
atures are depressed by water and possibly also
by CO
2
.Mierdel
et al
. (2007) demonstrated that
the depth of the LVZ precisely coincides with a
minimum in water solubility in the minerals of
the upper mantle (Figure 1.3, above). This means
that at this depth, water partitions more strongly
into silicate melts and therefore stabilizes a small
melt fraction. Hirschmann (2010), using a some-
what different line of reasoning, also concluded
that the presence of H
2
O together with the effect
of CO
2
will stabilize a small fraction of melt in the
LVZ, although the fraction may be smaller than
required to explain the geophysical observations.
Estimating the stability of partial melt and
the extent of melting in the upper mantle re-
quires data on (1) bulk water contents, (2) the
effect of small amounts of water on the peri-
dotite solidus and (3) water partitioning between
melt and upper mantle minerals. All of these
data are presently subject to considerable uncer-
tainty. Liu
et al
. (2006) concluded that water
decreases the melting temperature by 45
◦
Cper
wt % water in the melt, while the parameteriza-
tion of Hirschmann (2010) suggests a somewhat
smaller effect. Numerous data have recently been
produced on the partitioning of water between
silicate melts and olivine, pyroxenes, and garnet
(Koga
et al
., 2003; Aubaud
et al
., 2004; Kohn
& Grant, 2006; Hauri
et al
., 2006; Grant
et al
.,
2007; Tenner
et al
., 2009). Plausible values range
from 2
10
−
3
- 0.03
for orthopyroxene, 0.01- 0.03 for clinopyroxene,
and 1
10
−
4
-3
10
−
3
·
·
for olivine, 3
·
10
−
3
for garnet. The partition co-
efficients for pyroxenes strongly increase with Al
content, consistent with solubility data. How-
ever, even taking this effect into account, there
is still quite a large scatter in the partitioning
data. Moreover, most of the partition coefficients
were measured by SIMS, which cannot distin-
guish between water truly dissolved in the crystal
lattice and water present as mechanical impuri-
ties, such as fluid or melt inclusions. Some of the
data reported in the literature should therefore be
considered as upper limits of the true partition
coefficients. Mineral/melt partition coefficients
of water are expected to depend on temperature
and particularly on pressure; however, the limited
amount of data available does not yet allow us to
properly evaluate these effects.
10
−
3
-3
·
·
1.3.6 Earth's deep water cycle
A traditionally held view is that the oceans are
the products of outgasing of the mantle, although
alternative models have been proposed (Albar ede,
2009). Evidence for the existence of oceans already
in the Archaen implies that most of the outgas-
ing must have occurred very early, in agreement
with evidence from noble gases that also sug-
gests very early loss of volatiles from the mantle
(Marty & Yokochi, 2006). For a long time, it was
believed that degassing of the Earth is essentially
a one-way process, leaving behind a dehydrated
mantle. This view has been changed by the recog-
nition that traces of water in the nominally
