Accepted wisdom accounts for the bulk of lavas and intrusive igneous rocks that build island arcs and probably much of the continental crust by what is known as wedge melting. As old, cold and wet ocean lithosphere descends subduction zones, metamorphic reactions in the top layer of basalts and sediments (oceanic crust) release water-rich fluids. These depress the temperature at which melting can begin when they permeate the overriding mantle wedge. The water-releasing reactions involve dehydration of altered ocean floor that work to create the garnet-pyroxene assemblages that characterize eclogites, and drive the top slab of the lithosphere further from conditions under which it will begin to melt. Formation of abundant garnet and pyroxene also imparts the density jump that helps make oceanic lithosphere founder at destructive plate margins. The less there is, the lower the angle of subduction. Whether or not dense eclogite forms depends on the temperature at which lithosphere enters a subduction zone, and temperature depends to a large extent on how old the consumed lithosphere is. As sea-floor spreading shoves newly created lithosphere sideways from oceanic ridge systems, it slowly cools by conduction and interaction with permeating seawater. The faster the spreading or the smaller the plates involved, the sooner lithosphere can reach a subduction zone. Both factors can give rise to shallow-angled subduction.
The Earth loses heat that radioactive decay generates in the mantle by sea-floor spreading. Going back in time, there were more undecayed heat-producing isotopes, so more heat had to be lost. In the Archaean Aeon (more than 2500 million years ago) heat production was perhaps 2 to 3 times higher, and either spreading was much faster or there were more plates. Most geologists now accept that low-angled subduction was a common characteristic of Archaean geological processes. That is highly significant, because such conditions drive the top slab of oceanic crust towards melting, and the melts produced are very different from the basalts and andesites produced by modern wedge melting. They are much more silica-rich, and crystallize to form the trondhjemites, tonalites and dacites that are so common in Archaean continental crust.
Today, plate movements are sluggish, and though slab melting has been detected it was long thought to be rare, taking place only where very young oceanic lithosphere (less than 5 million years old) entered the mantle. Recent work by French geochemists (Gutscher, M-A. et al., 2000. Can slab melting be caused by flat subduction? Geology, 28, p. 535-538) showed that such occurrences relate to subduction of lithosphere as old as 45 million years. Their model to explain such Archaean-like processes involves a transformation from normal steep subduction to a phase involving almost horizontal movement of the descending lithosphere. The density reduction that this demands stems from the heating effect of the asthenosphere through which the plate travels. Wedge melting is generally close to the site of subduction, marked by an oceanic trench. Modern slab melting, however, needs a lengthy period of heating in a flat subduction zone, so the volcanoes that it produces lie much further away from the trench. Eventually the asthenosphere itself is cooled by the advancing plate and volcanism stops because the slab begins to dehydrate and to lose the potential for partial melting. This explains the lack of volcanoes over most of the known areas of flat subduction, as in the Andes of central Chile.
Plate tectonics is not the be all and end all of how the world works. It is merely the expression of the Earth’s overall behaviour by the thin surface rind of lithosphere. Almost certainly, all rocky planets behave similarly, in the sense of producing energy by decay of radioactive isotopes inside, and losing this energy by transport to the surface, where it escapes by radiation. How planets do this determines to a major degree the geological processes that go on at their surface. Clearly, there are subtle differences among the Inner Planets, because only the Earth shows signs of active plate movements that give it both a geological and, in its case, a biological life.
Why the Earth is so odd depends on its internal processes, so geochemists and geophysicists have spent 30 years seeking ways of unravelling how the mantle behaves. As well as a battery of geochemical methods to distinguish different kinds of mantle whose melting contributes to crust formation in different tectonic settings, the main arm in geodynamics is using earthquake waves in a manner akin to body scanning to image the deep interior. This seismic tomography is just beginning to resolve some of the widely divergent views about deep-Earth processes. So, a review of the state of the geodynamicists’ art in a recent issue of Science makes for compulsory reading (Tackley, P.J., 2000. Mantle convection and plate tectonics: toward an integrated physical and chemical theory. Science, 288, p. 2002-2007).
The geochemists’ problem, having discovered three chemically fundamental kinds of mantle that basalt magma production stems from, is to decide how they are arranged. They have at least 6 basic models. Before seismic tomography, each was as believable as the others. Through reviewing 3-D images of where hot and cold materials sit in the mantle – the key to motions within it – Tackley shows how some of the geochemical models must probably bite the dust, and the directions that research will take in future. There is still no self-consistent model for whole-mantle behaviour, but it is beginning to look like the various views of convection as simple cells, either from top to bottom of the mantle, or decoupled into lower and upper systems must give way to something much more complex. What does seem well established is that many subducted slabs find their way right down to the core-mantle boundary. The most primitive mantle ‘reservoirs’, from which the ocean island basalts over hot-spots stem in part, have an excess of 3He (formed only in stars and therefore locked in the Earth when it formed) over 4He (released by decay of radioactive uranium and thorium and so changing with time). These reservoirs are now probably in two gigantic, hot bulges rising from the core-mantle boundary, that dominate the most tectonically active parts of the lithosphere. Cooler mantle lies beneath more inert lithosphere. It has a composition from which mid-ocean ridge basalts emerge, and which signifies its loss over time of the materials that now make up the continents.
The most important possibility emerging from growing knowledge of the deep Earth is that Earth scientists might have to break from James Hutton’s 200 year old notion that the present is the key to the past. The plate-mantle system is something likely to change dramatically over time, and the Earth is currently in one form of many different kinds of possible behaviour.
In the same issue of Science is a review of how motions in the Earth’s core generate the geomagnetic field (Buffett, B.A., 2000. Earth’s core and the geodynamo. Science, 288, p. 2007-2012).