Earth's core–mantle boundary is a place none of us will ever go, but researchers using a special high-velocity cannon have produced results showing there may be molten rock at this interface at about 1,800 miles. Further, this molten rock may have rested peacefully at the core-mantle boundary for eons.
In a presentation at the fall meeting of the American Geophysical Union (AGU) today, California Institute of Technology geophysics professor Tom Ahrens reports new measurements of the density and temperature of magnesium silicate--the stuff found in Earth's interior--when it is subjected to the conditions that exist at the planet's core-mantle boundary.
The Caltech team did their work in the institute's shock wave laboratory, where an 80-foot light-gas gun is specially prepared to fire one-ounce tantalum-faced plastic bullets at mineral samples at speeds up to 220 thousand feet per second--about a hundred times faster than a bullet fired from a conventional rifle. The 30-ton apparatus uses compressed hydrogen as a propellant, and the resulting impact replicates the 1.35 million atmospheres of pressure and the 8,500 degrees Fahrenheit temperature that exist at the core–mantle boundary.
The measurements were conducted using natural, transparent, semiprecious gem crystals of enstatite from Sri Lanka, as well as synthetic glass of the same composition. Upon compression, these materials transform to a 30–percent denser structure called perovskite, which also dominates Earth's lower mantle at depths from 415 miles to the core–mantle boundary.
According to Ahrens, the results "have significant implications for understanding the core–mantle boundary region in the Earth's interior, the interface between rocky mantle and metallic core." The report represents the work of Ahrens and assistant professor of geology and geochemistry Paul Asimow, along with graduate students Joseph Akins and Shengnian Luo.
The researchers demonstrated by two independent experimental methods that the major mineral of Earth's lower mantle, magnesium silicate in the perovskite structure, melts at the pressure of the core–mantle boundary to produce a liquid whose density is greater than or equal to the mineral itself. This implies that a layer of partially molten mantle would be gravitationally stable over geologic times at the boundary, where seismologists have discovered anomalous features best explained by the presence of partial melt.
Two types of experiments were conducted: pressure-density experiments and shock temperature measurements. In the pressure-density experiments, the velocity of the projectile prior to impact and the velocity of the shock wave passing through the target after impact are measured using high-speed optical and x-ray photography. These measurements allow calculation of the pressure and density of the shocked target material. In shock temperature measurements, thermal emission from the shocked sample at visible and near-infrared wavelengths is monitored with a six-channel pyrometer, and the brightness and spectral shape are converted to temperature.
In both types of experiments, the shock wave takes about one ten-millionth of a second to pass through the dime-sized sample, and the velocity and optical emission measurements must resolve this extremely short duration event.
The pressure-density experiments yielded a surprising result. When the glass starting material is subjected to increasingly strong shocks, densities are at first consistent with the perovskite structure, and then a transition is made to a melt phase at a pressure of 1.1 million atmospheres. As expected for most materials under ordinary conditions, the melt phase is less dense than the solid. Shock compression of the crystal starting material, however, follows a lower temperature path, and the transition from perovskite shock states to molten shock states does not occur until a pressure of 1.7 million atmospheres is reached. At this pressure, the liquid appears to be 3 to 4 percent denser than the mineral. Like water and ice at ordinary pressure and 32 °F, under these high-pressure conditions the perovskite solid would float and the liquid would sink.
Just as the negative volume change on the melting of water ice is associated with a negative slope of the melting curve in pressure-temperature space (which is why ice-skating works-- the pressure of the skate blade transforms ice to water at a temperature below the ordinary freezing point), this result implies that the melting curve of perovskite should display a maximum temperature somewhere between 1.1 and 1.7 million atmospheres, and a negative slope at 1.7 million atmospheres. This implication of the pressure-density results was tested using shock temperature measurements. In a separate series of experiments on the same starting materials, analysis of the emitted light constrained the melting temperature at 1.1 million atmospheres to about 9,900 °F. However, at the higher pressure of 1.7 million atmospheres, the melting point is 8,500o F. This confirms that somewhere above 1.1 million atmospheres, the melting temperature begins to decrease with increasing pressure and the melting curve has a negative slope.
Taking the results of both the pressure-density and shock temperature experiments together confirms that the molten material may be neutrally or slightly negatively buoyant at the pressure of the base of the mantle, which is 1.35 million atmospheres. Molten perovskite would, however, still be much less dense than the molten iron alloy of the core. If the mantle were to melt near the core–mantle boundary, the liquid silicate could be gravitationally stable in place or could drain downwards and pond immediately above the core–mantle boundary. The work has been motivated by the 1995 discovery of ultralow velocity zones at the base of the Earth's mantle by Donald Helmberger, who is the Smits Family Professor of Geophysics and Planetary Science at Caltech, and Edward Garnero, who was then a Caltech graduate student and is now a professor at Arizona State University. These ultralow velocity zones (notably underneath the mid-Pacific region) appear to be 1-to-30-mile-thick layers of very low-seismic-velocity rock just above the interface between Earth's rocky mantle and the liquid core of the Earth, at a depth of 1,800 miles.
Helmberger and Garnero showed that, in this zone, seismic shear waves suffer a 30 percent decrease in velocity, whereas compressional wave speeds decrease by only 10 percent. This behavior is widely attributed to the presence of some molten material. Initially, many researchers assumed that this partially molten zone might represent atypical mantle compositions, such as a concentration of iron-bearing silicates or oxides with a lower melting point than ordinary mantle--about 7,200 oF at this pressure.
The new results, however, indicate that the melting temperature of normal mantle composition is low enough to explain melting in the ultralow velocity zones, and that this melt could coexist with residual magnesium silicate perovskite solids. Thus the new Caltech results indicate that no special composition is required to induce an ultralow velocity zone just above the core–mantle boundary or to allow it to remain there without draining away. The patchiness of the ultralow velocity zones suggests that Earth's lowermost mantle temperatures can be just hotter than, or just cooler than, the temperature that is required to initiate melting of normal mantle at a depth of 1,800 miles.
