Evolution and tectonics - Mantle Convection for Geologists - page 134

in geochemical estimates, but a higher uranium content of the Earth implies other
puzzles about Earth's composition.
In the second option, the transient cooling might have lasted for several hundred
million years without implying excessively high mantle temperatures in the past,
but any longer would raise questions about geological processes. If the mantle has
only 10 TW of radiogenic heating plus 7 TW from the core, and a mantle heat loss of
35 TW, then 18 TW has to come from cooling. This implies a cooling rate of about
140 ◦ C/Gyr, compared with the cooling rate in the models of the previous sections
of around 35 ◦ C/Gyr. If such a cooling rate had been sustained for 500 Myr, then
the mantle would have been 70 ◦ C hotter in the Palaeozoic than now. A temperature
that high might imply magmatism that ought to be evident in the geological record,
but our calibrations of the mantle models and of mantle petrology are not so secure
as to be able to strongly rule out the possibility.
Some fluctuations in the cooling rate are expected to have occurred just due to
normal fluctuations in the age-area distribution of oceanic plates [143]. However,
these may not have a large enough amplitude [144], or might imply higher rather
than lower heat flow in the recent past [145]. A more interesting possibility is
that plate tectonics might have been intermittent, as suggested by Silver and Behn
[146, 147]. This would cause larger fluctuations in heat loss, and would allow
the mantle to maintain its temperature over the longer term despite the present
imbalance.
The third option is that the heat imbalance is long term and the mantle has been
cooling more rapidly for a long time. A mechanism that might have this result has
been advocated by Korenaga [148, 149]. Korenaga argues that plate velocities are
limited not just by the viscous resistance to flow that we invoked in Chapter 5, but
by the energy it takes to bend the lithosphere as it enters a subduction zone. He
argues as well that the plate thickness is controlled not just by conductive cooling
but also by small-scale convection under the plate and by the dehydration that
accompanies melting under mid-ocean ridges, which increases the local viscosity.
The latter assumption leads to plates that are thicker in a hotter mantle, rather than
thinner. This increases the bending resistance to the plates, which slows them and
prevents the heat loss from being large. The result can be a mantle temperature
that peaked around 3 Gyr ago. An example of this kind of evolution is shown in
Figure 9.4 (black curves). This solution assumes a radius of curvature of subducting
plates of 200 km.
Solution '200' has a present cooling rate of 200 ◦ C/Gyr and a present Urey ratio
of 0.3. Early in Earth history the heat loss (Figure 9.4(b)) is less than the heat
generation (plus heat released by core cooling, not shown) and the mantle warms.
The (upper) mantle temperature peaks at about 1750 ◦ C about 3.4 Gyr ago, at which
time the heat loss and plate velocities reach minima because the plates reach their