The solid Earth - Geochemistry: An Introduction - page 238

the 147 Sm/ 144 Nd ratios of these reservoirs. This is not so. Typical 147 Sm/ 144 Nd ratios mea-
sured for continental rocks and the mantle are 0.12 and 0.25, respectively. This indicates
that reservoirs corresponding to mantle and continents are open to mutual exchanges and
have the effect of raising the apparent 147 Sm/ 144 Nd ratio of crustal rocks and lowering
that of mantle rocks. Both the growth of continents from the mantle and the recycling of
continental crust by mantle convection have been caught red-handed!
Although there is apparent consensus that recycling of continental crust corresponds
essentially to subduction of detrital sediments resting at the top of lithospheric plates, the
inverse process of extraction of crustal material is not fully understood. As with convec-
tion, we can consider that there has been a canonical model of crustal growth based on
the addition of andesite at converging plate boundaries. The Andean volcanoes, Mount
Pinatubo or Mont Pelée, seem ideal agents for supplying mantle to the continents. This
seems to be confirmed by the similarity of the chemical composition of lava from these
volcanoes (including the famous Nb deficit anomaly) and the isotopic compositions of
many elements, especially O, Sr, and Nd, to the composition of continental rocks.
Once again, superficial similarities mislead observers. A first alarming indication is the
histogram of the age of formation of continental crust. No matter which method was con-
sidered, the first histograms of rocks ages produced in the 1960s (e.g. Gastil, 1960 ) and the
1970s were already extremely clear: continents form well-defined age provinces (Hurley
et al. , 1962 ) and age histograms define very sharp peaks. The advent of dating using ion
probes and laser-ablation ICP MS now allows the rapid production of large numbers of
reliable zircon ages. Bedload sediments represent excellent samples of the crust from river
watersheds and the detrital zircons they contain make great samples for the formation ages
of the local crust. Hundreds of zircons from the bedload of four major rivers of the world
(the Yangtze, the Mississippi, the Congo, and the Amazon) have been analyzed by Iizuka
et al. (submitted). The concordant U-Pb ( 207 Pb*/ 206 Pb*) ages define very sharp peaks cor-
responding to well-known orogenic episodes. These are “magic” ages of very rapid crustal
growth around 300, 600, 1100, 1400, 1800, 2100, and 2700 million years ago ( Fig. 11.17 ).
Such well-defined histograms are very unlikely to be the result of differential erosion: by
any means, global processes such as erosion and sedimentation can be suspected to even
out frequency peaks, not to create them. Geochronology also indicates that the time taken
for these new continental expanses to form can be comparatively short. Millions of square
kilometers of continent arose to form West Africa between 2170 and 2090 million years
ago. Similarly, huge swaths of continental crust formed within a few tens of millions of
years around 2700 Ma ago in the Superior Province of Canada with little crustal growth
between the spikes. How can these characteristics be attributed to subduction, which is
in essence a constant phenomenon in response to the formation of new oceanic crust,
and therefore to the progressive outflow of internal heat? We can, of course, imagine that
cold oceanic lithosphere occasionally collapses into the deep mantle during brutal events
referred to as an avalanche. The thermal effects and mechanical consequences of such a
process appear too violent, though, compared with the relatively constant character of the
geological record since the earliest times (lava chemistry, sediment mineralogy).
Steady subduction in itself appears insufficient to account for crustal growth. Most sub-
duction zones are associated with rather small amounts of lavas and most of the andesitic