We will probably never see it or be able to use it. But billions of droplets of water buried hundreds of kilometers below us hold precious secrets of their own. Lou Bergeron reports
DEEP inside the Earth, the pressure is excruciating. Squeezed into strange shapes and forms, the rocks are so hot that they crawl like super-thick treacle. It is an inferno worthy of Dante, but it also contains something surprising. What's the last thing you would expect to find in this hellish environment? Water. Vast amounts of the stuff. In fact, more than 400 kilometers inside the Earth there may be enough water to replace the surface oceans more than ten times.
But this water is not a series of immense seas. Rather, it is scattered in droplets, some as small as a single molecule, with most trapped inside crystal lattices of rare minerals that only form under intense pressures. How much there is down there is still fiercely debated. But these inner "oceans" could help to explain long-standing puzzles about Earth's formation, the causes of deep earthquakes hundreds of kilometers inside the Earth, and why massive volcanic outbursts suddenly flood hundreds of thousands of square kilometers with lava. They may even give a glimpse of what the future holds for the Earth's climate--and if we might ever be drowned from below.
Many different strands of evidence point to the existence of this hidden water. One is the Earth's lack of water, compared with meteorites. For years, geologists have puzzled over this anomaly.
The Earth formed 4·5 billion years ago from a swirling cloud of dust, and gases such as water vapor, carbon dioxide, ammonia and methane. The cloud gradually formed tiny clumps of matter that grew into larger planetesimals, some 10 kilometers across. These stuck together and drew in more dust and gas. Eventually, the Earth had enough gravity to hold onto the gases in the planetesimals, and stop them escaping into space.
According to Thomas Ahrens, a geophysicist at the California Institute of Technology in Pasadena, you can work out how much water was around in the early Solar System by examining ancient meteorites--leftovers from the birth of the Solar System--that still bombard the Earth. Check the water in the meteorites and do the sums, he says, and you find out that the Earth should contain about 3 per cent water. In fact, adding up all the water in the oceans and atmosphere gives you just a tiny fraction of a per cent.
So where did all the water go? Well, many scientists believe that shortly after the Earth formed, a Mars-sized object slammed into it, stripping off its atmosphere and ejecting a chunk of the planet to create the Moon. So, presumably, much of the missing water could have been blown out into space at the same time. But there are clues that some water could still be trapped deep inside the Earth.
One clue comes from the helium emitted during volcanic eruptions. Helium has two isotopes. Helium-4 is formed by radioactive decay, while helium-3 comes from the time of birth of the Universe. According to Ahrens, helium-3 is plentiful in the volcanic rocks spewed up by so-called mantle plumes--upwellings of magma from deep inside the Earth that give rise to island chains such as the Hawaiian islands.
Helium is very volatile--much more so than water. If the Earth has held on to some of its helium-3 for this long, why shouldn't it have have held onto some of its water as well. Helium-3 "gives a clear signal that the Earth really does contain these kinds of material", says Ahrens.
Another tantalizing piece of evidence comes from kimberlites, rocks rich in iron and magnesium that travel along narrow channels from the mantle up to the Earth's surface. The rising rocks drag along other minerals, notably diamonds, which only form at depths below 180 kilometers. They also carry mica-like rocks, which are rich in water but would be unstable down at the depths where diamonds form, and so only exist nearer the surface.
But according to Raymond Jeanloz of the University of California at Berkeley, that doesn't mean that the water in the mica-like minerals didn't come up from greater depths. "These minerals may be the decompression products of higher pressure, water-rich phases that we never get the chance to see," says Jeanloz.
Meanwhile, Stephen Haggerty of the University of Massachusetts at Amherst, has found evidence that some kimberlites contain material from deep down in the mantle. Haggerty discovered that some kimberlites contain remnants of a mineral called majorite, which forms between about 300 and 670 kilometers down. So the water in kimberlites may have come from 670 kilometers down--the boundary between the upper and lower mantle.
Seismic data also suggest that there may be significant amounts of water in the mantle. Water has the effect of softening many rocks, slowing the speed of seismic waves passing through. And geologists have found that seismic waves do move unusually slowly in parts of the mantle. But despite these hints, until the late 1980s most researchers believed that it is basically pretty dry inside the Earth. Yes, there probably is some water near the surface, they reasoned, but go deeper than around 200 kilometers and there's nowhere to store water. The rocks, they presumed, would be just too hot to hold onto it. Then Joseph Smyth of the University of Colorado in Boulder made a startling discovery while studying a mineral called wadsleyite. The mineral is made up of silicon, magnesium and oxygen. Geologists believe that wadsleyite sits deep inside the Earth in roughly the top half of the transition zone between the upper and lower mantles, which spans depths of 400 to 700 kilometers.
But researchers can't dig down into the mantle to see what's happening there, so they have to rely on secondary tools. One way of working out which minerals may lurk deep inside the Earth is to take common upper-mantle rocks and squeeze and heat them until they reach the temperatures and pressures found inside the Earth.
That's how wadsleyite was made back in the 1960s. But the early researchers started out with dry olivine, so they ended up with dry wadsleyite. Smyth's breakthrough was to discover that even though wadsleyite only exists at temperatures well above 1000 ºC, it can still hold water.
At the same time, researchers were investigating whether other minerals under intense pressures and temperatures could hold water. One group of minerals being tested had first been created in the mid-1960s by Ted Ringwood and Alan Major at the Australian National University in Canberra. The minerals, dubbed phases A, B and C, had been synthesized in anhydrous conditions. Then, in 1987, Lin-Gun Liu and Ringwood not only found that phase C could hold water, but also synthesized a new mineral, christened phase D, which could also store water under the immense pressures and temperatures of the mantle.
Suddenly, there was somewhere to put water deep inside the mantle. "You can have oceans and oceans of water stored in the transition zone," says Jay Bass of the University of Illinois in Urbana-Champaign. "It's sopping wet stuff." Researchers think wadsleyite can hold as much as 3·3 per cent water by weight. It may not sound like much, but there could be an awful lot of wadsleyite.
According to Smyth, models of the mantle's composition suggest that at the depths where wadsleyite is stable, between half and three-quarters of the material is the right stuff for making this mineral. "If the region between 400 and 525 kilometers were, say, 60 per cent wadsleyite, and that phase was saturated at 3·3 weight per cent, that's ten oceans of water," says Smyth. But Dan Frost, an experimental petrologist at the Carnegie Institution of Washington's Geophysical Laboratory in Washington DC, thinks the mantle could contain even more water.
Frost says that solidified lava that has erupted at mid-ocean ridges contains glass that can be analyzed for water content. His research team has calculated how much water the lava's parent material in the mantle must have contained. "It ends up being between 100 and 500 parts per million," he says. And if the whole mantle contained 500 parts per million of water, Frost calculates that would be the equivalent of 30 oceans of water.
But there is a catch--mid-ocean ridge basalts form by melting just the top part of the mantle. "The question is, does that reflect the bottom part of the mantle?" says Frost. He believes that the whole mantle is relatively homogeneous in its composition, and that only the mineral structures change with depth. But no one can say for certain whether this is the case.
New water-bearing minerals are still being found. Earlier this year, Smyth's group published their discovery of wadsleyite II, another hydrous phase that may be stable even deeper into the mantle than the first wadsleyite.
But as Smyth notes, just because all these phases can hold water doesn't mean that they actually do. For that, you need to check what is really going on in the mantle. The main tool for probing rocks in the mantle is seismology. When an earthquake sends out seismic waves, geophysicists measure how fast the waves pass through mantle rocks. Given the speed of the seismic waves, and some information about how different minerals transmit them, geologists can work out what minerals are present.
And since water often slows seismic waves down, they may even be able to tell whether the minerals are wet or dry. The problem is that nobody knows the seismic properties of the hydrated version of the new minerals. "So we have no way of looking for seismic evidence of whether the phase is saturated or not," says Smyth. Many different groups are currently trying to pin down the seismic properties of the new hydrous minerals.
Meanwhile, water locked deep inside the Earth may be having significant effects on the surface through spectacular events such as the creation of island chains and massive outpourings of volcanic lava. Both features are examples of "hot spot" volcanism, which researchers believe is caused when a massive plume of hot material wells up from the mantle, melting rock which erupts through the crust.
For island chains like those around Hawaii, the hot spot is thought to be stationary while the tectonic plate slowly moves over it, producing one volcanic island after another. But some researchers believe that such plumes of material may not be primarily temperature driven. As Mark Richards at the University of California at Berkeley says, "hot spot volcanism could be triggered not by blobs of material that are anomalously hot rising through the mantle, but blobs of material that are anomalously wet". Richards points out that the presence of volatiles such as water in a mass of rock would make it buoyant, causing it to rise. "The presence of volatiles also lowers their melting point, so that when they rise to the surface, you get massive melting," he says.
Water could also explain puzzles about some of the occasional vast outpourings of lava that have taken place from time to time in Earth's history, and covered hundreds of thousands of square kilometers of land. The problem is with the lithosphere--the cold, solid outer shell of the Earth that includes the crust as well as the very uppermost mantle, and usually extends to about 70 kilometers below the surface. According to Chris Hawkesworth at the Open University in Milton Keynes, flood basalts erupted in some regions--for instance the famous Deccan Traps in India--have a composition that shows they formed from the asthenosphere, the mantle region below the lithosphere where the material is already thought to be partially molten. In these cases, the continent was already rifting apart when the plume arrived, and the magma had a fairly easy time passing up through the lithosphere.
But some continental flood basalts, in Brazil for instance, and in Namibia, are chemically different--richer in silica, and with different ratios of trace elements and isotopes. This suggests that they came from melting not of the asthenosphere but the lithosphere itself--an astonishing feat since the cold root of these continents extends down to around 170 kilometers.
In 1992, Hawkesworth and Kerry Gallagher at University College in London proposed that water could explain the mystery. Hawkesworth and Gallagher calculated that if a plume of hot material were to reach the bottom of the continental lithosphere, the heat could cause the lithosphere to dehydrate and to melt at lower temperatures. "If you want to melt the lithosphere, all models will agree that you have to have some volatiles present. And the experimental data will say that you need on the order of 0·3 or 0·4 per cent," says Hawkesworth. That works out to between 3000 and 4000 parts per million.
The Earth's inner oceans could also explain how earthquakes happen deep in the mantle--a long-standing puzzle ("Shaken to the core", New Scientist, 9 December 1995, p 42). Charles Prewitt, of the Geophysical Laboratory in Washington DC, thinks that water being squeezed out of minerals in the transition zone could cause the unexplained earthquakes that occur there. "Minerals could dehydrate and release a fluid, and that could essentially lubricate a fault. The fault could slip and the earthquake could occur," he says.
Don Weidner at the State University of New York in Stonybrook is also interested in transition-zone earthquakes, but sees a possible catch in the dehydration theory. He points out that generating an earthquake requires a build-up of stress followed by a sudden release, and suspects that this would be difficult if there is free water available. "If we're looking at a zone where water is being released, maybe the material's too weak to store the energy to have an earthquake," he says.
Another aspect of deep water that is hotly debated is the question of how much the water content of the mantle varies with time. Volcanoes are constantly spewing out water that they have ferried up from below, and water could also be sent back down into the Earth via subducting slabs. But how efficient are the slabs at carrying water into the mantle?
Much of the water-saturated sediment that accumulates on the seafloor as it moves along under the ocean is scraped off the descending slab by the tectonic plate that overrides it in the subduction zone. How much of the sediments actually slip into the mantle is hard to say. A more formidable barrier is the heat inside the mantle.
Many scientists used to believe that as a tectonic plate descends, no matter how much water is in it, all that water will be driven off by the heat of the mantle and disappear into magma which heads back up to the surface. But gradually evidence has accumulated that some water is making it down. Guust Nolet, at Princeton University, has found what he believes are traces of an ancient subducted plate under Central Europe, 300 kilometers below the surface. Though the slab itself has sunk from view--since subduction stopped 400 million years ago, Nolet says that seismic velocities are markedly slow. This is surprising under an old continent, where the relatively cold rock should produce fast seismic speeds. Nolet's interpretation is that the region is still enriched with water from the passing slab.
Although it is not easy to estimate how much water goes into the mantle through subduction compared with how much comes back out via volcanoes it seems that the whole system is roughly in balance. But what if the balance were to shift, and more water come out than goes in? Obviously the oceans would rise, but the more important effects would be in the atmosphere. "Water is the primary greenhouse gas," notes Jeanloz. If there were a massive build-up of greenhouse gases, he says, it could have a devastating effect on every living creature on Earth. But a sudden outpouring of water, Noah-style, is not likely even if the balance does tilt to a greater outflow. Rather it would be a gradual change on geological timescales, which would affect only our most distant descendants. Perhaps by then they will have evolved gills.
Lou Bergeron is a science writer based in Santa Cruz, California
"The Hidden Shore" by Raymond Jeanloz, The Sciences January/February 1993; "Water in the Earth's Upper Mantle" by Alan Bruce Thompson, Nature, vol 358, 23 July 1992
From New Scientist, 30 August 1997