SANTA CRUZ, CA--A peculiar type of seismic wave has unveiled a major surprise deep within the earth: The base of the planet's mantle, long assumed hard as a rock, instead may be partially molten.
Researchers at the University of California, Santa Cruz, studied seismic waves that skim along the sharp boundary between the earth's outer core and lower mantle. Under the south-central Pacific Ocean, something bogs down the speed of the waves by about 10 percent--a huge amount by geophysical standards. The most likely cause, the researchers claim, is that melted material bathes the mantle rock and transforms it into a thick mush.
If true, this conclusion violates the geophysical canon, which holds that the earth consists of four distinct layers: crust, solid rocky mantle, outer core of liquid iron, and inner core of solid iron. The upper mantle melts in regions where hot material rises to the planet's surface, but no one suspected the lower mantle--even in the cauldron of the inner earth--might behave the same way.
"This layer could fundamentally change our understanding of the mechanics of the core-mantle boundary," says mineral physicist Quentin Williams, lead author of the study and associate professor of earth sciences at UC Santa Cruz. "Suddenly, the bottom of the mantle might not be as stable as we thought. It's a provocative result."
Williams and seismologist Edward Garnero, a postdoctoral researcher at UCSC, published their results in the September 13 issue of the journal Science.
The seismic evidence points to a partially molten layer between 5 and 40 kilometers thick in the zone under the South Pacific. Williams and Garnero believe the layer probably encircles the globe, but it may be as thin as a few kilometers or even a few hundred meters in most areas. That's a tiny fraction of the 2,900- kilometer-thick mantle. But because the suspected layer touches the dynamic outer core, the consequences would be dramatic.
For instance, says Williams, the layer may help heat flow from the core into the mantle, a process that ultimately drives the relentless march of earth's crustal plates. Indeed, other seismologists recently found a broad swath of unusually warm mantle under the South Pacific. That swath lies directly above the layer identified in the new study, a correlation that Williams calls "extremely exciting."
Further, a partially molten layer would conduct electricity more readily than solid rock. That would influence the planet's magnetic field, the workings of which are still a mystery.
The Santa Cruz researchers acknowledge that other geophysicists may not readily accept their interpretation of the slow seismic waves. Rather than partially melted rock, some scientists will favor chemical reactions between the outer core and the lower mantle. Iron from the liquid outer core may interact with silicates and other minerals at the bottom of the mantle to form alloys--essentially, "rusting" on a planetary scale. This change in the property of the rocks could affect the way seismic waves travel through them.
Williams and Garnero do not rule out this possibility in their paper. However, they view it as far less likely than a small fraction of melt. "Reactions must occur between the core and the mantle," Williams says. "Any time silicates and iron get together, they mess each other up. But it simply wouldn't depress the seismic velocities to the extent that we see."
Gauging the sluggishness of the seismic waves was an arduous task for Garnero and several colleagues, notably seismologist Donald Helmberger of Caltech. They examined records of about two dozen large earthquakes in the southwest Pacific as measured by seismic stations in North America. The seismic energy they pinpointed was a "diffractive" wave, so named because it ripples for some distance along the mantle side of the core-mantle boundary. Some waves got delayed by slogging through the slow layer; those delays showed up as extra bumps in the seismograms.
Enter Williams, who studies the properties of inner-earth minerals under high pressures and temperatures. His calculations showed that to account for the seismic slowdown, between 5 and 30 percent of material within the layer must be molten. At about 5 percent, the melt would spread evenly among the rock grains to create slippery, hot conditions, much as a thin sheen of oil greases a container of ball bearings. If the fraction is higher, the melted material might occupy molten pockets within otherwise solid rock, like a hellish Swiss cheese.
Garnero says the conditions at the core-mantle boundary beneath the south-central Pacific are not unique; researchers believe the temperature there is the same worldwide. "We would hypothesize that if this partial melt occurs anywhere, it would occur everywhere," he says. "But if it's thinner than a few kilometers in other parts of the globe, we don't have the resolution with our seismic techniques to detect it."
Both researchers note that their theory is testable, via both mineral-physics experiments in the lab and further seismic studies. Garnero and seismologist John Vidale of UCLA are now exploring another phase of seismic wave that would slow down even further-- as much as 30 percent--in a partially melted layer.