Seismic Observations of Mantle Discontinuities and Their Mineralogical and Dynamical Interpretation - Physics and Chemistry of the Deep Earth - page 307

see it at long period. However, as this is not the
case, there is a contradiction between the PP and
P P observations of the 660 km discontinuity
which is currently unexplained. Some studies
have suggested that the difference in angle of
incidence between PP and P'P' can explain the
contradiction (Estabrook & Kind, 1996), but it
can still not explain the complete disappearance
of the '660' in PP precursors. And even though
the '660' is seen in short period P P precursors,
an array study of short period PP precursors in
the Pacific did not find any observations of the
'660' (Rost & Weber, 2002).
SS precursors always show a clear 660 km
discontinuity and the peak has a simple shape
in the majority of locations. Receiver function
observations, on the other hand, reveal a more
complicated structure than the SS precursors,
with single and double reflections ranging in
depth from 640 to 720 km (Simmons & Gurrola,
2000). An example of double reflections is shown
in the Pds receiver function in Figure 10.5b with
peaks at 660 and 750 km depth. In a global study
it was found that even the single reflections have
an asymmetric shape in the receiver functions
and there are no stations with simple reflections
from the 660 km discontinuity (Andrews &
Deuss, 2008).
Most seismic studies used to interpret the
660 km discontinuity in terms of phase transi-
tions in olivine from ringwoodite to perovskite
and magnesiow ustite/ferropericlase (also called
the post-spinel transition). This phase transition
is generally regarded as being sharp, occurring
over less than a few km in depth, making it dif-
ficult to explain why it is often not seen in PP
precursors. One phase transition cannot explain
the occurrence of multiple discontinuities either.
Additional phase transitions in majorite garnet
and ilmenite are needed to explain the more com-
plex features found in the seismic data (Vacher
et al ., 1998; Weidner &Wang, 2000; Hirose, 2002).
For a pyrolite mantle model, there will be
phase transitions in both olivine and garnet. The
interaction between the post-spinel transition
in olivine and garnet phase transitions depends
on the temperature structure (see Figure 10.6).

For low-temperature geotherms, the post-spinel
transition with its negative Clapeyron slope is
dominant at 660 km depth, leading to a sharp dis-
continuity visible in both short and long period
data. Additional transitions in the residuum
from majorite garnet to ilmenite to perovskite
at 610-640 km depth lead to multiple seismic
discontinuities, which have indeed been seen in
regional seismic studies (Simmons & Gurrola,
2000; Andrews & Deuss, 2008). At higher
temperatures, the transition frommajorite garnet
to perovskite becomes dominant with a positive
Clapeyron slope. There is no post-spinel transi-
tion and ilmenite is no longer stable. Majorite
garnet coexists with perovskite over a relatively
large pressure and temperature range and the
transition is more diffuse, resulting in a smaller
seismic discontinuity at 660 km depth followed
by a strong gradient. This wider interval might
explain why PP precursors do not always see the
660 (Deuss et al ., 2006), although it would suggest
that the absence of P660P would be limited to hot
regions, which is not the case. Aluminium con-
tent has also been shown to have a strong effect
on the width and temperature regime of the phase
transitions (Weidner &Wang, 2000), which would
be an alternative way to explain the disappearance
of the 660 km discontinuity in PP precursors.

10.3.4 Transition zone thickness

The most robust measurement that can be made
for the transition zone discontinuities is the tran-
sition zone thickness, which is defined as the
distance between the 410 and 660 km disconti-
nuities. SS precursors have predominantly been
used to make global maps of the transition zone
thickness, (Flanagan & Shearer, 1998a; Gu et al .,
2003; Houser et al ., 2008; Lawrence & Shearer,
2008; Deuss, 2009), but receiver functions can be
used on the continents and on the ocean islands
(Chevrot et al ., 1999; Li et al ., 2003; Lawrence
& Shearer, 2006b; Tauzin et al ., 2008). Quite
significant variations exist between different SS
precursor thickness maps (Deuss, 2009), so we
are still far from being able to robustly inter-
pret the transition zone thickness in terms of