Geoscience Reference
In-Depth Information
the 660 km phase transition as the former case.
The HBL subducted below 660 km releases much
more water than that in the former case because of
the lower
C
max
the depth to which the water is conveyed by
generating upwelling plumes of the hydrated ma-
terials. This probably occurs in the transition
zone, when the large hydrated area in the transi-
tion zone is formed as the case with lower mantle
C
max
H
2
O
. Accordingly, most of the released
water remains in the transition zone, being aided
by the ascending return flow in the wedge mantle
to redistribute efficiently H
2
O into the transition
zone. In addition, retreat (backward migration) of
the slab and the trench occurs in this specific
model shown in Figure 13.7, and the hydration
spreads horizontally to the direction of retreat. It
is noted that the 100-km thick area above the slab
in the transition zone remains anhydrous because
dry mantle above the 410-km phase transition is
dragged into the transition zone by the subducted
slab. After 65 Myr, the water reaches the 410 km
phase boundary, and finally fills the wedge mantle
area.
Although in these models, instantaneous segre-
gation of fluid phase is assumed, fluid migration
during process-1 is limited by permeability or
width and number density of cracks. In practice,
it is impossible to separate the fluid perfectly from
the descending mantle flow because a low fluid
fraction gives low upward velocity. Scaling anal-
ysis suggests a critical fraction
φ
c
to be
H
2
O
of 0.001 wt %. In this case, the structure of
the hydrated area could be much different from
that derived from our simulation. The dehydra-
tion at the 660-km phase transition is supposed
to have similar effects to those from the stagnant
slab discussed in Richard and Iwamori (2010), as
was described in the former section. When most
of the water is transported into the lower mantle
as in Figure 13.6 the Rayleigh-Taylor (RT) insta-
bility due to its buoyancy is expected to play
a key role to separate the hydrated layer from
the descending slab (Gerya & Yuen, 2003). The
growth time of the RT instability depends on the
density change due to the water and the viscosity
of the lower mantle. Although the effects of the
hydration on the material property in the lower
mantle have not been revealed, we assume that
the effects on the density and the viscosity in the
lower mantle are similar to those used in Richard
and Iwamori (2010) for the upper mantle. Accord-
ing to Richard and Iwamori (2010), the growth
time is estimated to be 50 to 250 Myr for the
lower mantle viscosity of 10
22
Pa s (Okuno &
Nakada, 2001). Sinking speed of the lower man-
tle slab is estimated in the range from 1 to 6 cm
yr
−
1
inferred from seismic tomography (Ricard
et al
., 1993) or moving speed of the subducting
lithosphere (Billen, 2010). The time scale of the
RT instability growth is comparable to that of
the slab descent to the bottom of the lower man-
tle (40-200 Myr). We can therefore expect that
a considerable amount of water that crosses the
660-km phase transition is transported into the
deep lower mantle, although some of the water
may return to the transition zone from the in-
termediate depth of the lower mantle with the
hydrated ascending plume. Can we detect these
subducted fluid components by some means?
In the following final section, we discuss geo-
chemical evidences that could reflect large-scale
domains associated with such components in
the mantle.
10
−
4
∼
to 10
−
3
, assuming
μ
=
10
−
3
9.8 m s
−
2
,
Pa s,
g
=
10
3
kg m
−
3
,
B
(the constant involved in
permeability)
ρ
=
2.8
×
10
3
,
n
10
−
3
m (Iwamori,
1998). These parameters are appropriate for a
relatively shallow region (
<
100 km depth) of sub-
duction zones, and are likely to change for the
deep mantle:
μ
would increase and
ρ
would
decrease since the aqueous fluid may contain
a significant amount of silicate components be-
yond the second critical endpoint (Mibe
et al
.,
2007). In this case with a greater
μ
and a smaller
ρ
, the critical fraction may exceed 10
−
3
, indi-
cating that the order of 1000 ppm H
2
O could
be trapped in the descending mantle flow as
in Figure 13.6, even if the mineral phases are
intrinsically dry.
We have also made the other assumptions for
the physical parameters and model configura-
tions. Especially, we have not considered the
effects of the density and viscosity reduction
by hydration. These may significantly influence
=
=
3,
R
=
