By J. William Bell, NCSA -- Geophysicists at the Carnegie Institution of Washington take a crack at the chemistry of minerals deep in the Earth's mantle.
Comparison between the crystal structure of silica. The left panel is the structure under static conditions (0 Kelvin and 120 gigapascals). The right panel is a snapshot of the molecular dynamics run showing the dynamic structure of the same material under thermodynamic conditions similar to the Earth's deep mantle.
In the late 1950s and early 1960s, "Project Mohole" attempted to retrieve a sample of mantle by drilling a hole through the Earth's crust at the bottom of the ocean. Run by an informal band of scientists called the American Miscellaneous Society and the National Science Foundation, the project returned important information on the crust, but it was discontinued before it breached the Mohorovicic Discontinuity that marks the boundary between crust and mantle. Project Mohole made it to a depth of about 180 meters, and the record stood at about 2,100 meters in 2005. A Japanese scientific vessel, the Chikyu, is currently being outfitted to close the final 800 meters to the goal. With 50 years of attempts, it's clear that understanding the Earth's mantle is a challenge that will never be met by direct observation alone (or even direct observation primarily). Once the Mohorovicic Discontinuity is tapped, researchers will still only kiss the outer realms of mantle that continues to a depth of 2,900 kilometers. Instead, geophysicists rely on indirect means. Razvan Caracas and Ronald Cohen of the Carnegie Institution of Washington in Washington, D.C., plumb the mantle's greatest depths by predicting properties of minerals and melts at high pressures and temperatures using NCSA's Cobalt and Copper supercomputers. They perform first-principles computations within density functional theory. These calculations are called "first-principles" because there is no experimental input -- they use only fundamental physics as described by quantum mechanics and electrodynamics in the form of the density functional theory. Properties of Earth's materials are obtained from simulating the behavior of electrons and nuclei under certain thermodynamic conditions. "We can't go out and look at samples," says Cohen. "We only have seismological studies and experiments and simulation. We have no other information on how the deep Earth might work." With their studies, they provide key data about the physical properties of minerals under extreme conditions of temperature and pressure. These data are further used by other geoscientists to interpret observations, to plan experiments, or to build geodynamical models of the interior of the planets. "The properties that can be seen [by experimentalists or by those doing computer simulations] are limited. We try always to work closely with one another and put constraints on one another...Sometimes [a feature] is first predicted by experiment. Other times, in theory. But the constraints give us an idea of what to look for, and one method is often an immediate help to the other," says Caracas, a Carnegie Fellow at the institution who runs the simulations.
Absolute seismic anisotropy of the MgSiO3 and FeSiO3 perovskite and post-perovskite as computed at 120 GPa. (Absolute seismic anisotropy is the absolute difference between the velocities of shear waves propagating in the vertical and in the horizontal planes.) Post-perovskite has a lager seismic anisotropy and its presence in the D'' layer may explain similar features recorded from that part of our planet.
A puzzle at 20 million psi
Caracas and Cohen focus not on the upper boundary of the mantle but on the opposite side: the core-mantle boundary. This boundary between the solid, silicate-based mantle and the liquid, iron-based core is "the most dramatic interface in the whole Earth," according to Caracas. The contrast in chemistry, mineralogy, viscosity, and seismic wave velocities is greater even than the contrast between ocean water and the solid bottom. Pressure in this region is more than 130 gigapascals, or about 20 million pounds per square inch. Temperature is more than 2,500 Kelvin. The mantle side of the core-mantle boundary, the D'' (dee double prime) region has "puzzled geophysicists for decades," according to a 2005 article in Geophysical Research Letters by Caracas and Cohen. As with other boundary areas in the Earth's mantle, seismic waves behave differently when they cross it. The why of this puzzle lies in the peculiarities and the complexity of the signals we receive from this region and in its remoteness and extreme nature of its thermodynamic conditions. For years, it was thought that the bulk of this lowermost mantle was composed of an iron-bearing magnesium silicate with perovskite structure. Perovskite (MgSiO3) is not found on the surface, but it makes up about 70 percent of the mantle. However, diamond-anvil cell experiments, which compress a sample between two diamonds to recreate the extreme pressures inside planets, complicated the conventional wisdom in 2003. They showed that the silicate undergoes a phase transition at about 125 gigapascals and 2,500 Kelvin. The crystal structure of perovskite is altered, and it transforms to what is called post-perovskite. This physical change offers possible explanations for the mystifying behaviors ascribed to the lowermost mantle; seismic waves traveling through crystal structures in the post-perovskite formation would have a different signature than waves traveling through perovskite, for example. But without a clearer understanding of what influences the change and what impacts it has on the mantle's character, those explanations remain broad and subject to further change. That's where Caracas, Cohen, and their collaborators come in. Pick up the change
One of the prime targets of their simulations is the effect that the addition of certain atoms has on perovskite's transition into post-perovskite. The team has shown that aluminum oxide, which is known to be present in the lowermost mantle, slightly increases the pressure at which the transition takes place. The introduction of additional iron ions, meanwhile, considerably reduces the transition pressure. These results were published in Geophysical Research Letters and presented at the Annual Meeting of the American Geophysical Union in Fall 2005. More thorough calculations of these changes, now underway at NCSA, will allow the team to build complete phase diagrams of the perovskite to post-perovskite transition and to assess how iron influences that transition. A phase diagram shows the point at which the transition occurs at different temperatures and pressures. At a higher temperature, the required pressure might be lower, for example. The team is also calculating the elastic constants of the crystals. This feature equates to the stiffness of the crystals as they deform and return to their original shape under the sort of stress that might be generated by a seismic wave. In 2006, meanwhile, the team published on the Raman spectra of perovskite and post-perovskite -- the first time these spectra had ever been derived for a mineral by simulation using density functional theory. Raman spectroscopy is an experimental technique used to study the vibrational characteristics of materials. When light hits a molecule or a crystal, the incoming photons interact with particular atomic vibrations and change their original frequency. This shift is a direct expression of the atomic vibrational frequency. It is measurable by Raman spectroscopy and is very specific. Using it, scientists can determine the interatomic chemical bonds that are present in the molecule or crystal and can use it as a nondestructive identification tool. The team's article, which also appeared in Geophysical Research Letters, showed substantial differences between the spectra exhibited by perovskite and those for post-perovskite. This finding tells experimentalists that Raman spectroscopy is a good means of testing whether the transition has taken place in a sample, because the disparity between the two spectra is so pronounced. "NCSA's entire approach was essential to delivering these findings," says Caracas. "They helped with setup and compiling our codes, and their flexibility in using the machines and giving us dedicated time was key." This research is supported by the National Science Foundation's Division of Earth Sciences. For further information:
http://www.gl.ciw.edu/
http://www.gl.ciw.edu/~r.caracas/ Team members:
Ronald Cohen
Razvan Caracas
Burkhard Militzer
