THE VAST, grinding machine called plate tectonics carries continents thousands of kilometres across the Earth. It creates and consumes oceans, and recycles chemicals that would otherwise build up and suffocate life. But why it is still going is a mystery. By rights, this machinery ought to have seized up.
Without some means of stress release, the tectonic plates should have long ago jammed together in a global gridlock. If our planet could not crack through these plates, as soon the gears clashed, all motion would grind to a halt. But what could produce forces strong enough to crack these immense rafts of rock? The process doesnât happen on any other planet in the solar system â even Venus, our near twin, doesnât have plate tectonics. What keeps Earth moving?
Suspicion is growing that the secret ingredient may be water deep in the Earthâs mantle, because the planet is so wet. Even so, it is hard to understand how just a few drops of water could affect the solid planet so profoundly. After all, you canât crack a boulder by spitting on it.
Advertisement
Yet researchers are gradually beginning to understand how the tiny amounts of moisture inside the Earth might have such a profound effect. Among them are Klaus Regenauer-Lieb and colleagues of the CSIRO, Australiaâs leading scientific research organisation in Perth. Their theory not only explains the Earthâs unique behaviour, it makes some startling predictions. If they are right, then in less than a million years giant earthquakes will be rocking the eastern coast of North America. Soon afterwards a great mountain range will begin to grow, stretching from Newfoundland to Florida, and volcanoes will spew lava and belch ash into the North American skies.
Such a dramatic series of events will follow something even more spectacular â the breaking of a tectonic plate. Each plate consists of a layer of brittle crust a few kilometres thick supported by a strong slab of rigid mantle up to a few hundred kilometres thick. Together they are known as the lithosphere. For the plates to keep moving, there have to be places where new crust is born and old crust destroyed. New crust is generated at mid-ocean ridges, where two plates pull apart from one another and the gap is filled by magma from below. Old crust is destroyed at the other end of the conveyor belt, when ocean lithosphere sinks underneath another plate. The weight of this sinking edge pulls the whole plate along behind it, like a chain sliding over the edge of a table. Itâs called subduction.
Continental flotsam
It is always ocean crust that gets destroyed at subduction zones. Continents are the accumulated flotsam that has proved too light to be subducted over billions of years, so they are far too light to sink into the mantle. And that presents a problem. Eventually, a subduction zone eats its way right through the edible ocean crust until it meets another continent or an island chain that just wonât go down. Suddenly, subduction shuts down. âSubduction zones are thought to die when something too buoyant clogs them,â says Chad Hall, a geologist at California Institute of Technology in Pasadena.
For example, more than 40 million years ago, the Indian plate was moving towards north Asia and eventually hit it, creating the Himalayas. The ocean between the two continents was swallowed by an ancient subduction zone that is not there today.
If that were the end of the story, then within a couple of hundred million years, there would be no subduction zones left on Earth.
That would be bad news for us, because volcanoes keep spewing out sulphur and carbon dioxide and we rely on subduction to carry these chemicals back into the mantle, locked within ocean sediments. Without that waste-disposal mechanism, our planet would probably end up like Venus, with a hot, thick, choking atmosphere.
So for the past four billion years, something has been creating new subduction zones. But what force can crack through a tectonic plate at least 70 kilometres thick and 1000 kilometres long, and then bend it downwards so it can slide into the mantle? According to simple calculations, forces from plates pushing on each other, or the weight of sediments dumped by rivers at the edge of continents, are far too weak to crack a plate.
The mystery deepened in 1990, when the Magellan spacecraft returned radar images from Venus. Although its atmosphere is different, Venusâs diameter is only 5 per cent smaller than Earthâs and it is made of almost exactly the same stuff. Most planetary scientists thought that Venus would have plate tectonics too. Magellan found otherwise. Venus has a single, fixed surface and the pattern of craters shows that the whole surface is about the same age, around 500 million years old. Scientists now think that Venus goes through long periods of relative quiet, punctuated by sudden volcanic cataclysms that resurface the whole planet.
Planetary scientists began to suspect that water held the key to the difference between these twins. Could it be that water can soften the hard, cool rock of the lithosphere? To find out, geophysicists had to know how water affects rocks under the pressures typical many kilometres underground.
In 2000, Shenghua Mei and David Kohlstedt conducted definitive experiments at the University of Minnesota in Minneapolis. They tested a mineral called olivine that makes up more than 50 per cent of the rocks of the upper mantle, and provides most of their strength. In their experiment, a piece of olivine about the size of a little finger was squeezed by high-pressure gas and heated in a furnace. Sure enough, Mei and Kohlstedt found that adding water to the sample made it much softer.
The pair found that at the temperatures and pressures typical 40 kilometres underground, wet olivine deforms 10 times as fast as dry olivine. Most importantly, the olivine doesnât need to be very wet. Add just 20 parts per million of water, and the mineral oozes easily. The presence of just a few hydrogen ions from water introduces defects in the crystalline structure of the olivine, weakening it.
Along with other researchers, Regenauer-Lieb saw this as a possible solution to the conundrum. He teamed up with David Yuen, also at the University of Minnesota, to devise a computer simulation showing how water could make subduction possible.
Regenauer-Lieb and Yuen looked at the join between continent and ocean crust. Here, sediments washed off the continent gradually pile up, putting the plate under stress. In previous simulations, even a pile of sediment 10 or 15 kilometres thick was not enough to punch through a plate. But the new model used Mei and Kohlstedtâs data to see if adding a little water to the plate could do the trick.
Sure enough, the simulations showed that a wet plate can break (Science, vol 294, p 578). As the pressure piles on, a fault slices catastrophically through the lithosphere, simultaneously snapping the crust and rigid upper mantle. Once severed, the oceanic side of the plate bends under the weight of sediments and begins to sink down into the mantle.
If this model is right, then the next place to experience such a cataclysm will be the east coast of North America. There, the ocean crust is the oldest in the world, and has accumulated a colossal 15 kilometres of sediment, making the lithosphere sag beneath it. Once it begins to give way, it will take less than a million years before the Atlantic Ocean crust starts to slide underneath America. Africa could be next in line, as the Atlantic coast of the Sahara has the second heaviest sediment load. As the Atlantic starts to close, Europe and America will finally begin to come together.
This is Regenauer-Liebâs conclusion, but other geologists have developed different models. Michael Gurnis, a geologist at Caltech, thinks that plates are far more likely to crack on old fault lines. At mid-ocean ridges, new crust forms as the plates move apart. The ridges donât run in continuous lines, but look more like a line of steps, displaced suddenly sideways every few hundred kilometres in what are called transform faults. A young plate emerging from such a broken source is riven by fracture zones, because a strip of plate streaming out of one step of the ridge canât be seamlessly joined to a strip coming from the neighbouring step. These weakened fracture zones divide two pieces of crust with different ages. Perhaps the older part, being colder and denser, could trigger subduction by sinking and sliding underneath the younger?
In July, Gurnis, with Caltech colleagues Chad Hall and Luc Lavier, and Maria Sdrolias and Dietmar Mueller from the University of Sydney, showed how that could happen (Earth and Planetary Science Letters, vol 212, p 15). When an old fracture zone is squeezed by plates on either side moving together, the older side of the plate gets pushed underneath the younger, and youâve got subduction again.
In this model, the tectonic machine still depends on water because the simulations assume that the friction in a fracture zone is practically zero. High-pressure water could act as a lubricant. âThe only way I can really think of to get strengths this low is to have elevated pore pressure of water,â Hall says. Gurnis adds that water could also transform the olivine in the fracture zone into a weaker mineral called serpentine.
Regenauer-Lieb doesnât think this model fits the facts. If subduction starts out at transform faults, he says, then the subduction zone will run perpendicular to âisochronesâ, lines of constant age on the crust. But subduction zones are more commonly parallel to isochrones; that is, the crust being consumed at a zone is all of much the same age. That fits Regenauer-Liebâs model, where the oldest ocean crust goes down first because it is sitting next to a continent, picking up sediment. Gurnis counters that there are places where subduction is far from parallel to the isochrones. And in any case, plates move around and change direction, so the present alignment of subduction and isochrones doesnât necessarily reflect the alignment when subduction started.
Deciding who is right could be tricky because subduction destroys the evidence. The rocks that were around when todayâs subduction zones began have long since melted into the mantle. âThe thing about subduction zones is that they flush the geological record down the toilet,â Gurnis says. In fact, consensus has yet to favour one model over another. âThe initiation of subduction is an unsolved problem,â says Gerald Schubert, who has studied the tectonic differences between Earth and Venus at the University of California, Los Angeles. âThe models work as models but it is unclear if the Earth works the way the models do.â
Tectonic oil
There is one big problem facing both these theories. When ocean plates are formed they are very dry, because almost all the water is squeezed out as the magma solidifies to form lithosphere. If water really oils the tectonic gears, there has to be some way for it to seep back into the plate, where it can moisten fracture zones or soften the mantle enough for it to give way.
Gurnis thinks the water could be seeping down from the oceans and circulating through cracks in the rock in a deep hydrothermal system. But more than a few kilometres down into the crust the pressure may become so high that there are no holes for water to run through. Regenauer-Lieb thinks instead that any fresh water would have to come from below, from the mantle itself. The mantle is plastic enough to convect very slowly, and this month (Mineralogical Magazine, vol 67, p 697) Regenauer-Lieb and Thomas Kohl of the Swiss Federal Institute of Technology in Zurich suggest that a blob of mantle that is warmer and wetter than surrounding rock could rise up and slowly insinuate itself into an ocean plate over tens of millions of years. This could be primordial water that has been there since the Earthâs formation, or have come down through previous subduction.
Having a wet mantle plume come up right under a huge sediment load sounds like quite a a coincidence, but there is a hint that it might be happening. Suzan van der Lee at Northwestern University in Evanston, Illinois, has analysed seismometer readings to look for evidence of wet rocks under the ocean of the North American plate. Seismic waves from earthquakes around the plate are delayed by regions of wetter, softer rock, and there are indeed such hints of wet rocks below 150 kilometres under the US east coast. The data is too poor to reveal whether these are plumes, however. Van der Lee thinks they may be relics of the Atlanticâs ancestor, an ancient sea called the Iapetus Ocean, which was subducted around 400 million years ago.
Getting better data will mean dropping seismometers onto the Atlantic ocean floor, and the chances of funding are slim for now. The current seismic network might also be able to pick up signs of earthquakes as the plate begins to give way. Eventually the evidence will be incontrovertible. âAn obvious clue would be volcanoes emerging near the Atlantic coast of the USâ, Regenauer-Lieb says.
Whichever theory you back, a little water makes all the difference. Which raises the possibility that, in the very distant future, our planet could lose its water and seize up entirely. Venus may once have had plate tectonics like the Earth. Perhaps, a runaway greenhouse effect meant that clouds rose to the upper atmosphere, where ultraviolet radiation from the sun broke up the water molecules. Hydrogen, being so light, leaked away into space. Gradually the oceans dried up, and the mantle lost all its water through volcanoes. And maybe, the desiccated lithosphere became too strong to break, so no more subduction, and no more plate tectonics. Could it happen here? âYes, we could âgo Venusâ,â says Regenauer-Lieb. Without plate tectonics to recycle volcanic gases, Earth could end up with mountains iron pyrites covered in sulphur snow.
