FLIP through the pages of any geology textbook, and you鈥檒l see the classic
diagram of planet Earth sliced open to reveal distinct layers like a giant
onion. Bright colours mark these out: the thin crust on which we live, the rocky
mantle, the outer core of swirling liquid iron, and the hard nugget of iron that
is the inner core. The outermost layers can be a little smudgy: though the
mantle rock is solid, it can still creep, and every so often it bursts through
the fragile crust, melting as it does so to create flows of fresh lava. But
farther down, geophysicists have long believed that the transitions from solid
mantle to liquid outer core to solid inner core are sharply defined, with
precise mineral structures that are forced into being by the extreme pressures
and temperatures.
However, surprising discoveries in the past few years have challenged this
neat picture. Nearly 3000 kilometres beneath the planet鈥檚 surface, at the base
of the supposedly solid mantle, researchers have found vast blobs of gooey
semi-molten rock up to 40 kilometres thick and as wide as good-sized continents.
Any one of them would hold enough melted rock to dwarf the largest volcanic
reservoir near the Earth鈥檚 surface.
Geophysicists are beginning to discover what effects this mush may have on
the inner Earth鈥攁nd perhaps even on the ground beneath our feet. For
instance, they say, a semi-molten blob might draw enough heat out of the core to
spawn the 鈥渉ot spots鈥濃�攔ising plumes of hot mantle鈥攖hat created
volcanic islands such as Iceland, Hawaii and the South Pacific鈥檚 paradises.
What鈥檚 more, iron from the core may dissolve into the semi-molten rock to form
electrically conducting patches that alter wobbles in the planet鈥檚 spin and
define the edges of the paths taken when the magnetic poles change places
(鈥淲hen north flies south鈥�, 快猫短视频, 30 March 1996, p 24).
The zones may even be the fading embers of an ocean of molten rock that once suffused the
young Earth鈥攁nd whose remnants remained trapped between the liquid core
and the solid mantle as the planet inexorably cooled.
Advertisement
鈥淭his shows the top of the planet is not as special as we thought,鈥� says
Quentin Williams, a mineral physicist at the University of California, Santa
Cruz. 鈥淚t changes all the rules for how we think about the bottom of the
mantle.鈥� Fellow mineral physicist Raymond Jeanloz of the University of
California, Berkeley, agrees: 鈥淲e used to think of the mantle as a thick ceramic
shell isolated from a liquid steel outer core,鈥� he says. 鈥淲e are now thinking
诲颈蹿蹿别谤别苍迟濒测.鈥�
The interface between the mantle and the core has long been known as one of
the planet鈥檚 wildest boundaries, with pressures more than a million times higher
than at the surface, a temperature jump at the core鈥檚 edge of some 1000 掳C,
and an increase in density greater than that between air and soil. Since
researchers can鈥檛 drill deep enough to probe this region, they rely on seismic
waves generated by earthquakes to work out what鈥檚 there. By recording the path a
wave takes near the surface, extrapolating it back through the Earth, and
gauging whether it has sped up or slowed down, seismologists can deduce much
about the composition and structure of the hidden layers.
Most seismic waves zip through the mantle, but there is one particular
path鈥攃alled a 鈥渄iffractive phase鈥濃�攁long which the wave can sometimes
get bogged down, the way a muddy track slows a thoroughbred. After a large
earthquake, a wave travelling this path dives into the planet and ripples for
some distance along the mantle side of the core-mantle boundary, like a stone
skimming across a lake before sinking. The wave then dips into the outer core,
re-enters the mantle and races back to the surface. Seismologist Ed Garnero
learnt of this phase from German scientists in the late 1980s while working as a
graduate student at Caltech with seismologist Don Helmberger. 鈥淔or Fiji
earthquakes recorded in the eastern US, the phase was bigger and arrived later
than it should,鈥� says Garnero, now at Berkeley. But neither Helmberger nor
Garnero immediately realised the obscure wave鈥檚 significance: 鈥淚 showed it to
Don, who basically said, `Hmmm, interesting, now go back to work鈥�,鈥� says
Garnero.
Into the slow zone
Seismologist Steve Grand of the University of Texas, Austin, rediscovered the
phase in microfiche files while on sabbatical at Caltech in 1992. He, Helmberger
and Garnero took a closer look and decided that something strange was slowing
seismic waves by a few per cent in the bottom 100 kilometres of the mantle. Such
velocity reductions were noteworthy but not earth-shattering, and the group鈥檚
first paper hardly caused a ripple.
But during the next two years Garnero found more examples of the sluggish
waves, including many that probed a broad swath of mantle under the
south-central Pacific Ocean. Seismic velocity reductions of a mere few per cent
could not explain the new observations, Garnero realised. So Helmberger put
forward a bold suggestion: perhaps the slow layer at the bottom of the mantle
was much thinner than they had thought鈥攋ust tens of kilometres thick. A
layer that thin would better explain the details of the seismograms, Helmberger
said, but it would require that the seismic waves were hitting something truly
odd that retarded their speeds by 10 per cent or more. A slowdown of that
magnitude was unheard of in the deep Earth. However, Garnero overcame his
initial scepticism and worked out the first sketch of what he called the
鈥渦ltra-low velocity zone鈥� or ULVZ.
At first, no one could imagine what this layer was made of. But when Garnero
moved north to Santa Cruz in 1995, he worked in earnest with Williams to find
out what could make seismic waves drag so drastically. One possibility was a
chemical change in the rock. Some years earlier, Jeanloz and mineral physicist
Elise Knittle had shown that chemical reactions may occur between liquid iron in
the outer core and silicates in the lower mantle, a claim that remains in
dispute. Could such reactions change the chemistry of the rock enough to put the
brakes on seismic waves? 鈥淎ny time silicates and iron get together, they mess
each other up,鈥� says Williams. However, this effect still wasn鈥檛 big enough to
explain the seismic slowdowns.
Hot Swiss cheese
To come up with other explanations, Williams and Garnero bounced some
outlandish ideas around. The most heretical was partial melting. Most
researchers did not believe that any ingredient of the lower mantle could melt
under the intense pressures there, even at temperatures of 3000 掳C or more.
However, Williams and Garnero maintained that the only way to account for the
sluggish seismic waves in such a thin zone was if between 5 and 30 per cent had
melted. At the low end of that range, the material might resemble packed ball
bearings greased by a sheen of oil. A higher ratio of melt could create pockets
of liquid rock within a solid matrix, like a hellish Swiss cheese.
Just a dash of partial melt is enough to change the rock鈥檚 character
profoundly. For instance, Williams and his colleagues think the ULVZ flows and
deforms about a trillion times more readily than the solid mantle above it, like
molasses on steel. Such pliable material would be far more efficient at
extracting heat from the red-hot core and circulating it upwards into the rest
of the mantle, perhaps forming the roots of warm plumes that ascend to the
surface. However, the scientific community remained unconvinced. The notion that
the lower mantle insulated the core like a thick ceramic Thermos was an
entrenched part of the geophysical canon of inner Earth. 鈥淭he general knee-jerk
reaction was that extreme proposals need extreme documentation,鈥� Jeanloz says.
鈥淲e all said, `You have to prove this鈥�.鈥�
So Garnero and several co-workers extended their studies of diffractive waves
to cover just under half of the entire core-mantle boundary. They found that
blobs of ULVZ fill about one-quarter of their study region, proving that the
swath under the south-central Pacific is no fluke. Thick patches also lie under
Iceland, Alaska and Africa.
Whiplash waves
Seismologists John Vidale of the University of California, Los Angeles, and
Justin Revenaugh of UC Santa Cruz used a different kind of seismic wave, one
that bounces off the core-mantle boundary, to probe the ULVZ. Helmberger and
Revenaugh see hints in these reflections that seismic shear waves鈥攚hich
have the same motion as a snapping whip鈥攄awdle through the ULVZ,
travelling perhaps 30 to 50 per cent slower than usual.
Only partial melt could make them do that, but the shear wave evidence is
indirect and notoriously hard to extract. Even so, says geophysicist Thomas
Ahrens of Caltech, 鈥減artial melt is a sound conclusion and a damned interesting
discovery鈥�. While a few researchers harbour doubts, Ahrens says there is a
growing consensus.
Attention is now turning to the vexed issues of what the ULVZ is made of and
how it might affect the behaviour of the Earth above. Those issues, it turns
out, are intertwined. If the ULVZ contains iron from the outer core, for
instance, that could influence Earth鈥檚 magnetic field. Also, radioactive
elements in the ULVZ might heat it up and help send plumes rising through the
mantle. But this is all speculation, because no one has the slightest idea what
the ULVZ contains. 鈥淪eismology is a very blunt tool,鈥� says geophysicist Thorne
Lay of UC Santa Cruz. 鈥淭hese are beautiful and uniquely clean seismic
observations, and we know they must sample the core-mantle boundary. They tell
us the ULVZ is a big, kick-ass structure. But they don鈥檛 tell us squat about
chemical composition.鈥�
Fortunately, there are other ways to tackle the question. Researchers are
exploring two scenarios. One is that a section of the lower mantle unpolluted by
iron from the core melts at exactly the temperature and pressure found in that
region. Recent work by mineral physicist Reinhard Boehler of the Max Planck
Institute for Chemistry in Mainz, Germany, supports that view. Boehler notes
that temperature estimates for the top of the core have converged in recent
years on about 3700 掳C. In the lab, when Boehler crushes silicon dioxide,
magnesium oxide and other likely ingredients of the deep mantle, part of the
mash starts to melt near that temperature. That鈥檚 consistent with a thin ULVZ,
he observes, because the temperature of the mantle falls rapidly with increasing
distance from the core. 鈥淚t may seem a beautiful coincidence,鈥� Boehler admits.
鈥淏ut it can explain the ULVZ without resorting to an exchange of molten iron and
silicate at the boundary.鈥�
Recent research by Knittle at UC Santa Cruz, however, suggests that iron does
indeed react with some of the mantle constituents. The liquid part of the tiny
quantities of semi-molten mineral stew that she has cooked up in the lab under
pressure seems denser than the solid it leaves behind. Ordinarily, molten rock
is more buoyant than solid rock, which explains why volcanoes point up. But in
this case, some heavy ingredient鈥攑erhaps iron from the core鈥攎ay
weigh down the melt and pin it at the core-mantle interface. Knittle鈥檚 work also
hints that uranium may separate into the molten fraction and heat it via
radioactive decay. To resolve these issues scientists will have to gauge the
chemical compositions of samples weighing just a few billionths of a gram in
flecks less than a thousandth of a millimetre wide. 鈥淭he experiments pose an
incredible technical challenge,鈥� Lay observes.
A different line of inquiry may help settle this dispute. Geophysicist Bruce
Buffett of the University of British Columbia, Vancouver, studies tiny wobbles
in Earth鈥檚 axis of rotation, called 鈥渘utations鈥�, which result from the combined
gravitational pulls of the Earth, Sun and Moon on the Earth鈥檚 interior. The
liquid core鈥檚 electrical and magnetic interactions with the overlying mantle
determine its degree of freedom in responding to these forces. 鈥淚t appears there
is much more friction between the core and mantle than we would have guessed,鈥�
Buffett says. 鈥淭o explain it, we need a highly conducting layer at the base of
the mantle.鈥�
Iron entrained from the core into the lower mantle in a global layer just 200
metres thick would do the trick, he says. Such an iron-rich layer probably would
not alter the planet鈥檚 magnetic field in any way that could be detected at the
surface. However, there are hints that Earth鈥檚 magnetic poles travel along
preferred routes from north to south when they flip every million years or so.
Those reversal paths, it seems, may avoid the most conductive patches where the
ULVZ is thickest.
Volcanic hot spots
Magnetic fields are not the only planetary patterns that the ULVZ may drive.
Williams, Revenaugh and Garnero point to a correlation between ULVZ patches and
volcanic hot spots. Hawaii and Iceland lie above especially prominent blobs of
ULVZ. Over the entire globe, the matches between hot spots and ULVZ are by no
means perfect, but they have less than a 1 per cent chance of arising randomly,
say the researchers.
Other geophysicists find this statistical link intriguing but not compelling.
Vidale, for one, thinks that ULVZ enthusiasts may overestimate the true extent
of the slow-velocity blobs. 鈥淢aps show the ULVZ covering a bit more than
one-quarter of the core-mantle boundary region that has been searched, but I鈥檓
sceptical of those,鈥� he says. Localised bits of very thick ULVZ would retard
seismic energy just as surely as broad regions of thinner semi-molten material,
he notes. In that case, the ULVZ might not underlie all the hot spots that
Williams and his colleagues claim.
Regardless of the ULVZ鈥檚 extent, a tough question remains: could its
influence extend all the way to the Earth鈥檚 surface? Rock in the upper mantle
gradually circulates over geologic time, driving the motion of the continental
plates. But debate has long raged over whether those motions are confined to the
upper mantle or persist all the way down to the core-mantle boundary. Recent
research suggests that the latter is true. If so, the hot ULVZ may generate
plumes of warm rock that rise right up to the surface. But the devil is in the
details. So little is known about the ULVZ鈥檚 properties, says geodynamics
modeller Louise Kellogg of the University of California at Davis, that such
schemes require vigorous arm-waving.
For instance, lab tests on partly melted rock hint that the ULVZ may flow a
trillion times more readily than the surrounding mantle, but that figure could
vary by a factor of a thousand or more. The ULVZ鈥檚 chemistry and thickness could
change markedly within tens of kilometres. Until researchers fill these gaps in
their knowledge, there is little to gain from devising complex computer models
to see whether the ULVZ would generate plumes that come all the way up, Kellogg
says.
Geophysicist Michael Wysession of Washington University in St Louis agrees
that it鈥檚 premature to assign sweeping roles to the ULVZ in whole-Earth
processes. But maps of the ULVZ, he observes, suggest tantalisingly that beneath
major subduction zones鈥攚here one cold tectonic plate slides underneath
another, and plunges into the planet鈥檚 interior鈥攖he layer either doesn鈥檛
exist or is too thin to see. And that鈥檚 just what you鈥檇 expect if the surface
were dynamically connected to the depths. What鈥檚 more, he says, 鈥渢here鈥檚 been no
subduction above the mid-Pacific for hundreds of millions of years. That stuff
has just been sitting on top of the core-mantle boundary and warming up. And
it鈥檚 the biggest ULVZ there is.鈥�
Even as geophysicists debate these issues, the ULVZ has opened a new window
on the delicate physics of Earth鈥檚 inner layers. 鈥淭he story of the Earth is this
continuous dance between pressure and temperature,鈥� says Wysession. 鈥淭emperature
keeps increasing and wants things to melt, but increasing pressure fights back
and wants to keep things solid. This dance occurs all the way down. You just add
a little bit of chemistry here and there, and `poof!鈥�, you can make it melt.鈥� It
may take many years for researchers to learn the secrets of this latest
geological magic trick.
