Heat Flow and Thermodynamics of the Lithosphere - Quantitative Plate Tectonics - page 326

these variations give rise to thermal subsidence,
they cannot be considered as the ultimate
cause of the instability that leads to subduction
initiation of old oceanic lithosphere. In fact, the
most reliable picture of the thermal history of
oceanic plates, which is given by PCM, predicts
asymptotic thermal and isostatic equilibrium as
t !1 . In addition, a detailed petrological and
geophysical analysis of the density structure of
the oceanic lithosphere has shown that the depth-
averaged value of ¡ at t D 90 Ma (including 7 km
thick crust with lower density ¡ D 2,900 kg m -3 )
is between 3,310 and 3,312 kg m -3 (Afonso
et al. 2007 ). In this instance, the density contrast
with respect to the surrounding asthenosphere
would be ¡ D 35.5 kg m -3 , which is less
than the commonly assumed value, but more
interestingly the predicted depth-averaged
density would be slightly lower than the density
of the asthenosphere immediately below the
compensation depth z c (¡ a Š 3,330 kg m -3 ).
Therefore, a hypothetical gravitational instability
of the old oceanic lithosphere could hardly
explain the initiation of subduction, a process
that is still poorly known. However, differently
from the unsubducted lithosphere, slabs in the
upper mantle are cooler than the surrounding
mantle and have positive density contrasts up to
200 kg m -3 , although the portions subducted
at depths close to the 670 km discontinuity may
attain hydrostatic equilibrium (Ganguly et al.
2009 ). It is interesting to note that the model of
Ganguly et al. ( 2009 ) predicts that a fragment
of subducted lithosphere may have positive (i.e.,
upward directed) buoyancy in the uppermost
lower mantle. In fact, in the case of old slabs
the endothermic phase transitions of ringwoodite
into perovskite and magnesiowüstite occur at
depths significantly greater than the 670 km
discontinuity. In this instance, we have that
the 9 % negative density anomaly associated
with the mineralogical contrast relative to the
surrounding mantle, which gives a positive
contribution to buoyancy, would prevail over the
negative contribution associated with the thermal
anomaly (see Sect. 1.6 ) .
The increased density of the subducted litho-
sphere is mostly due to the metamorphism of the

oceanic crust at high temperatures and pressures
and to the effect of phase transitions in the upper
mantle (see Sect. 1.6 ) . In fact, shortly after down-
ward bending, the hydrated basalts and gabbros
of the oceanic crust are converted to their high-
pressure eclogitic phases, with release of substan-
tial amounts of H 2 O and a consequent increase of
the crustal density to values in excess of 3,500 kg
m -3 in a few Myrs (Ahrens and Schubert 1975 ;
Kirby et al. 1996 ; Peacock and Wang 1999 ).
Regarding the effect of phase transitions, we have
seen in Chap. 1 that the exothermic phase tran-
sition of the olivine phase in peridotite to wads-
leyite increases the density by 6 %. Therefore,
the olivine of a slab segment located just above
the 410 km depth discontinuity will be subject
to premature phase transition to wadsleyite. Con-
sequently, the density contrast and the negative
slab buoyancy will increase further at this depth.
Schubert and Turcotte ( 1971 ) estimated that the
total body force exerted on a descending slab due
to the shallower phase transition was nearly as
large as the force on the slab due to thermal con-
traction. Tassara et al. ( 2006 ) in a study about the
Nazca plate estimated an average density contrast
¡ D ¡ l - ¡ a Š 90 kg m -3 in the asthenosphere,
while Ganguly et al. ( 2009 ) propose even larger
values of ¡, in the case of old plates, in the deep
asthenosphere just above the transition zone. In
summary, our present knowledge of the complex
thermochemical processes that accompany the
penetration of slabs into the mantle allows to
say that once started, the “subduction factory”
can proceed autonomously, driven by the negative
buoyancy and passive sinking of the slabs, at least
down to the transition zone.
Today most geoscientists accept the idea that
the pull exerted by the subducting slabs on the
tectonic plates is the dominant force driving
plate motions. This concept was originally
proposed by Richter ( 1973 ) on the basis of
theoretical arguments. However, it grew into
a widespread theory after that an empirical
analysis about the relative importance of the
different torques exerted on tectonic plates
confirmed that the negative buoyancy of slabs
played a major role (Forsyth and Uyeda 1975 ).
In

their analysis, Forsyth and Uyeda ( 1975 )