U21D-01 INVITED
Coupled core-mantle thermal evolution after a giant impact
Various arguments suggest that the Earth's mantle was partially or completely molten at the end of planetary accretion (the magma ocean hypothesis). The early Earth's core was likely to be even hotter than the mantle, perhaps by thousands of degrees Kelvin. Here we address the following questions: What role did the core play in the evolution of the magma ocean and how hot was the core upon crystallization of the magma ocean? We model the coupled evolution of the core-mantle system immediately after the last giant impact. In the early stages of evolution, the core was cooling very fast. This generated a high heat flow at the base of the mantle and may have even caused additional melting of the mantle. When the average temperature of the lower mantle dropped below solidus, the heat flow from the core decreased but was still very high due to the remaining molten layer at the base of the mantle and the narrow melt channels formed throughout the mantle. The heat from the core could have also been removed by plumes produced by the instability of the thermal boundary layer at the base of the mantle. The presence of the initially hot core does not affect much the time it takes the lower mantle to crystallize and it does not affect the previously suggested conclusion that the lower mantle crystallized without any significant chemical differentiation. When the core-mantle boundary temperature decreased below the critical temperature for the rheological transition (about 60 percent crystal fraction), the heat loss rate of the core dropped by several orders of magnitude. The subsequent core cooling was relatively slow and the temperature at the core-mantle boundary may have never decreased below mantle solidus. This model is consistent with the estimates of the present-day temperature at the core-mantle boundary and with the estimates of the core temperature change during planetary evolution. A partially molten base of the present-day mantle is consistent with the existence of the ultra-low velocity zone at the base of the mantle.
U21D-02
Core Exsolution: A Likely Consequence of Giant Impacts and a Likely Energy Source for the Geodynamo
Exsolution (mantle underplating by material leaving the core) is an old idea but has been difficult to pin down, primarily because of our ignorance of the relevant phase diagrams. If exsolution occurs, it is roughly an order of magnitude more important per unit mass of exsolved material than the growth of the inner core and can therefore play a major role in the geodynamo even when only 0.5 percent of the core mass is exsolved. Some general principles can be identified even though many of the details remain obscure: (1) Hf/W data suggest that a substantial amount of core/mantle equilibration took place in the high T and high P core formation events that follow immediately after a giant impact, e.g., the Moon-forming impact. This mandates a significant amount of solution of mantle constituents into the core, including (very importantly) components that would not normally be viewed as very soluble, such as MgO. (2) It is precisely those low solubility (yet otherwise major) Earth constituents that are most likely to exsolve upon dilution and cooling in the core because of the Arrhenius form of the solubility law and the large enthalpy of solution. Of course, Gibbs phase rules preclude simply following the MgO, but simplified three component models suggest this is the most important factor. Extrapolation of experimental data together with core cooling models suggest (but does not prove) that the likely solubilities and subsequent exsolution of MgO are important for the geodynamo. (3) The post-giant impact model of a magma ocean at the base of the mantle (either through being very hot or through downward percolation of dense mantle melt) allows additional extraction of mantle constituents into the core through double diffusive convection that does work against gravity using the early primordial heat reservoir. This is an energy storage device useful for later core convection and geodynamo generation. (4) The consequences for the geochemistry we can sample (e.g., OIBs) is likely to be subdued or non-existent because the lowermost mantle is stably stratified and lacks efficient mixing with the shallow mantle. (5) The core composition (density deficit relative to pure iron) is a constraint on these models since it places a limit on the amount of equilibrated material subjected to severe T and P. The core would be even less dense than observed if this amount is too great. Thus, the core composition (and complementary mantle siderophiles) should be viewed as being on a mixing line between high T,P and lower T,P reservoirs at time of last equilibration. (5) This is an attractive alternative to K-40 for aiding the core geodynamo because it is complementary to inner core growth, i.e., acts early to mid-history when the inner core growth is possibly not present. It may still be mildly significant even now. (6) It can also help maintain the Mercury dynamo.
U21D-03 INVITED
Energy Sources for an Early Dynamo
Convection in the Earth's core is currently driven by a combination of thermal and compositional buoyancy. The compositional buoyancy is due to light elements, which segregate into the liquid core as the inner core grows. Prior to the formation of the inner core, it is thought that convection is sustained solely by thermal buoyancy as the core cools. However, the high heat flow needed to power an early dynamo raises questions about the primordial heat needed to maintain this high heat flow over the age of the Earth. Radiogenic elements, such as 40K, can provide another energy source which alleviates problems with the thermal history of the core. An alternate energy source involves changes in chemical equilibrium between the core and the base of the mantle. High-pressure experiments suggest that partitioning of oxygen and silicon between liquid iron and mantle minerals is strongly dependent on temperature. Cooling is expected to transfer light elements from liquid iron into the mantle minerals, leaving behind a dense, iron-enriched liquid at the top of the core. This additional source of compositional buoyancy can drive vigorous convection prior to the formation of the inner core and may be the primary energy source for an early dynamo.
U21D-04 INVITED
Accretion and Core Formation: Constraints From Metal-Silicate Partitioning Experiments
The concentrations of siderophile elements in the Earth's mantle provide, in principle, when compared to chondritic meteorites, the means to estimate the conditions under which the Earth's core segregated from the mantle. The additional requirements are experimental metal-silicate partitioning data for as many elements as possible over appropriate ranges of pressure, temperature and oxygen fugacity. The purpose of this presentation is to use data for a large number of elements of widely differing behavior (e.g V, Ni, Cr, Nb, Co, W, Si, Ga, Pb, Cu, Ag) to constrain the conditions under which the Earth accreted and segregated its core. If we assume that the core segregated continuously during accretion at the base of a homogeneous magma ocean then results indicate that the magma ocean would have, on average, extended to one-third of mantle depth on the growing planet. In order to match the V and Cr concentrations of the mantle, however, accretion must have begun under strongly reducing conditions with progressive oxidation occurring as the Earth grew. This model leads to a slightly subchondritic Nb/Ta ratio of the silicate Earth and an Si content of the core of about 5per cent. The latter is in accord with recent Si isotopic data.Modifications of the model lead to similar conclusions. If we assume that the magma ocean is not well-stirred and that metal equilibrates only locally with melt, then pressures and temperatures increase but progressive oxidation is still required. If we assume that metal equilibrates with a crystal-rich mush instead of melt, oxidation is required and it becomes very difficult to match the mantle concentrations of strongly incompatible elements such as Nb and W. Given that the Earth underwent an early reduced phase before becoming more oxidised, oxidation could have occurred by addition of more oxidised materials in the later stages of accretion. Another possibility is that the mantle "self-oxidised" through disproportionation of ferrous iron to ferric plus metal in the lower mantle. Whatever the explanation, the results imply that Si and S are the major light elements in the core and that, because of the strongly reducing conditions, the oxygen content of the core is low.
U21D-05 INVITED
A two phase model of core mantle segregation
The Hadean time during which the Earth was simultaneously growing by accretion and ongoing core segregation is still poorly known. Although in the first millions of years after the condensation of the first solids, radioactivities now extinct were heating the planet, the major source of heat came somewhat later, deposited near the surface by meteoritic impacts. The surface heating became important when gravity was itself important (say for a planet larger than 1000 km). A classical scenario of core-mantle segregation assumes that a hot shallow layer was formed, in which the iron (or the undifferentiated material) was melted. In this layer, the iron separated from the silicates, and formed metallic pounds on top of a still undifferentiated deeper and colder mantle. This metallic layer underwent an instability and sank into the deeper mantle as a diapir. To test this scenario we developed this code based on the two phase formalism of Bercovici et al. (2001). As the melting temperature of iron is lower than that of silicates we assume that iron can be present in both solid or liquid phases while silicates remain solid. When the metal is solid, the metal and the silicates are locked together and we treat their mixture as a single phase fluid where density is function of temperature and composition (iron/silicate proportions). When metal is liquid, it can separate from the silicates and the two phases interact through shear stress (e.g., Darcy flow) and normal stress. The evolution of the volume proportion of liquid iron (the porosity) is controlled by the difference of pressure between the two phases. The heat equation accounts for the release of potential energy that occurs during segregation. The model predicts an evolution significantly different from the simple original scenario. The increase of temperature due to segregation (release of gravitational energy) is comparable to the initial heat delivered by the impact so that the process of segregation, once started, is more or less self maintained. The first diapir that crosses the mantle leaves a cusp-like trail that connects the protocore to the near surface silicates across the undifferentiated material. Melting occurs continuously both in the shallow and in the deep mantle. The sinking of metallic diapirs is very fast (of order of 100 kyrs) as instead of deforming the surrounding material as in a usual Stokes flow, the undifferentiated material desegregates on the bottom side of the diapir, the silicates cross the metallic phase, and accumulate behind the sinking diapir. The first impact that melts the iron phase is therefore potentially ableto trigger the whole core-mantle segregation.
U21D-06
Linked magma ocean solidification, cumulate mantle compositions, and atmospheric growth
Early in terrestrial planet evolution energetic impact, radiodecay, and core formation may have created one or more whole or partial silicate mantle magma oceans. The time to mantle solidification and then to clement surface conditions allowing liquid water is highly dependent upon heat flux from the planetary surface through a growing primitive atmosphere. Here we model the time to clement conditions for 500, 1,000, and 2,000 km-deep magma oceans on Earth. Included in our calculations are partitioning of water and carbon dioxide between solidifying mantle cumulate mineral assemblages, evolving liquid compositions, and a growing atmosphere. Magma ocean solidification and subsequent planetary evolution proceeds through three major phases. First, the magma ocean solidifies, partitioning volatiles between solid cumulates, evolving liquids, and a growing primordial atmosphere. In these simplified models the timescale of solidification is limited by the emissivity of the atmosphere. This step produces cumulates with density that non-monotonically increases with radius, and are therefore gravitationally unstable to overturn. In step two, the unstable solidified silicate mantle cumulates overturn to a stable configuration. The overturn process creates a mantle that is gravitationally stable and therefore resistant to the onset of thermal convection, but that is laterally heterogeneous in composition and temperature. Hot cumulates that formed deep in the magma ocean rise to shallower depths during overturn and may melt adiabatically, producing the earliest planetary crusts. The surface of the planet is therefore heated. In step three, the planet conducts heat through its solidified mantle and radiates it to space through the primordial atmosphere formed in step one. We find that small initial volatile contents (0.05 wt% H2O, 0.01 wt% CO2) can produce atmospheres in excess of 100 bars, and that mantle solidification is 99% complete in less than 100,000 years. Subsequent cooling to clement surface conditions occurs in 10 to 50 Ma, not considering impact erosion of the atmosphere. Though the great majority of volatiles are degassed into the atmosphere, a geodynamically significant quantity is sequestered in the solid cumulates, as much as 750 ppm by weight OH in models beginning with 0.5 wt% water, and a minimum of 10 ppm by weight in the driest cumulates of models beginning with just 0.05 wt% water. Even these small water contents significantly lower the melting temperature of mantle materials, facilitating later volcanism.
U21D-07
Episodic Layering of the Early Mantle by the ‘Basalt Barrier' Mechanism: Implications for Tectonic and Geochemical Evolution
Episodically layered mantle convection is induced during the first 1.5-2 Ga of evolving numerical models by the ‘basalt barrier' mechanism, which is due to the buoyancy of subducted oceanic crust between depths of 660 km and about 750 km. During layered periods the upper mantle cools and the lower mantle warms, reaching temperature differences of up to 300°C. Also the oceanic crust is only a few kilometers thick, due to settling of the basaltic component through the upper mantle and consequent depletion of the uppermost mantle. This can explain the observed strong early depletion of the mantle, and the thin oceanic crust would facilitate subduction and plate tectonics. The layering inhibits heat loss, so that high mean mantle temperatures persist. The layering breaks down roughly every 100-150 Ma, at which time hot, fertile lower mantle floods the upper mantle and there is a dramatic burst of magmatism lasting a few million years and manifest in the models as oceanic crustal thicknesses of tens of kilometers. As radioactive heating declines and the mantle gradually cools, subducted plates eventually become thick and heavy enough to penetrate and disrupt the ‘basalt barrier', and no further layering occurs. These results suggest causes for the apparently episodic growth of continental crust, for the lack of evidence for strongly episodic behaviour over the past billion years or so, and for the distinct ‘vertical tectonics' of some Archean granite-greenstone terrains. The models yield mean residence times of mantle material consistent with the mantle lead-isotopic apparent age of about 1.8 Ga. Accumulations of (denser) basaltic material at the base of the models preserve a remarkably clear record of the early magmatic pulses. These accumulations persist strongly into the present. They are consistent with a recent proposal that helium isotopes record mantle melting events, and they would help to explain differences in Hf and Nd isotopes between the accessible mantle and chondritic meteorites. The accumulations, and ‘piles' of more-basaltic material under mantle upwellings, are broadly consistent the seismic D" zone and with the large, seismically slow zones deep under Africa and the western Pacific.
U21D-08
Varying rates of mantle convection in the early Earth
The transition from 142Nd/144Nd approximately 35 ppm higher than chondritic in early Archean rocks from Isua (Caro et al., GCA, 2006; Boyet and Carlson, EPSL, 2006; Bennett et al., 38th LPSC, 2007) to the 20 ppm excess measured in rocks younger than 3.5 Ga (Boyet and Carlson, Science 2005) suggests a period of mixing between LREE enriched and depleted reservoirs probably formed within 30-50 Ma of Earth formation. In order to reach 142Nd/144Nd as high as 35 ppm above chondritic requires a quite high Sm/Nd ratio (147Sm/144Nd = 0.22). This value is similar to the Sm/Nd ratio needed to explain the elevated initial 142Nd/144Nd and 143Nd/144Nd of ancient lunar crustal rocks (Boyet and Carlson, EPSL, 2007). This observation suggests that extensive differentiation of the Earth through a magma ocean event was completed before the Moon formed from a LREE-depleted terrestrial mantle. Between 4.5 and 3.8 Ga, mixing in the mantle must have been relatively minor in order to preserve the very high 142Nd/144Nd seen in the ancient rocks from Isua. Between 3.8 and 3.5 Ga, relatively rapid mixing between high- and low-Sm/Nd reservoirs produced during the magma ocean event must be invoked to lower the 142Nd/144Nd to the value measured in all rocks younger than 3.5 Ga. A possible explanation for this evolution in mantle composition notes geodynamic models that rapidly crystallize (within millions of years) a magma ocean to leave a buoyantly unstable cumulate pile that then overturns rapidly (e.g. Elkins-Tanton et al., EPSL, 2005). After the overturn, the mantle has cold dense material underlying buoyant hot material, which inhibits mantle convection until internal heating overwhelms the chemical buoyancy. The evidence for time varying 142Nd/144Nd in the early Archean suggests that by 3.8 Ga, this inhibition had been removed and that convection throughout a large portion of the mantle had begun. This model provides the unexpected prediction that the Hadean may have been a relatively quiescent time on Earth between the violence of Earth formation, global melting, magma ocean crystallization and overturn, and the resumption of wholesale mantle convection in the early-Archean that continues to the present day.