DI34A-01
Evolution of Mantle Reservoirs and Heterogeneity
Mantle heterogeneities arise from differentiation at mid--ocean ridges, subduction zones, and melting associated with hotspots. The residues from these processes together with the recycled products are circulated in the mantle where they evolve isotopically, mechanically and spatially. The change in length scale of heterogeneities depends on their effective viscosity (dehydrated melting residues are strong, recycled crust and unerupted plume material is weak) as well as their location in the mantle; deformation rates are high in the upper mantle, and lower in the higher viscosity deeper mantle. Melting residues and recycled material circulate to varying depths depending on the details of mantle flow, and heterogeneities may follow a range of paths. Melting at ridges samples the upper mantle on a time scale of ~300 My; processing the volume of the mantle takes ~10 times longer. This leads to a preferential processing of the upper mantle. Exchange with the deeper mantle is from deep subduction, and from plume conduits that rise from depth, as well as convective stirring. The increase in viscosity in the lower mantle, together with the migration of subduction zones limits deep subduction as slabs slow down, thicken and remain in the transition zone and the top of the lower mantle. There is also substantial recirculation of subducted material in the upper mantle. Folded slabs may trap other heterogeneities within them. Models of deformation in the mantle and vertical transport of heterogeneities are based on geographically realistic mantle flow models with plates. A statistical assemblage of mantle heterogeneities is sampled to produce MOR basalts and continental crust, which reflect averages of the component heterogeneities. The model successfully reproduces observations of Pb, Nd and Sr isotopes There is a ubiquitous component in the upper mantle similar in composition to FOZO or C that is made up of heterogeneities with small length scales (<15 km). This mixes with residues of melting (which are larger) to produce observed data arrays. The lower mantle contains larger heterogeneities and it evolves and exchanges relatively slowly with the upper mantle and surface. The model allows the effects of heterogeneity properties, deformation rates, sampling rates and material transport rates to be assessed, and it captures the essentially statistical nature of mantle evolution and sampling.
DI34A-02 INVITED
Helium in a convecting mantle
We reconcile the 3He/4He evidence for a lower mantle with high primordial 3He with seismic and geochemical inferences that the lower mantle has been subjected to convective flow and has undergone significant mass exchange with the upper mantle over Earth's lifetime. He is highly incompatible and is extensively lost from the mantle during melting. Consequently, lithospheric slabs that penetrate into the lower mantle, as inferred from seismic tomography, imply that the lower mantle has been processed and outgassed by partial melting beneath mid-ocean ridges and hotspots. Paradoxically, high 3He/4He ratios measured in ocean-island basalts are typically thought to indicate that these lavas are derived from a relatively undegassed mantle sourve that is located presumably in the lower mantle. We show that mixing of noble-gas depleted lithospheric slabs within a convecting lower mantle decreases the degassing efficiency by slowly diluting He concentrations in this mixed mantle reservoir. Despite extensive processing by melting throughout Earth's history, this results in present-day 3He with several tens of percent of primordial concentrations. We therefore find no inconsistencies between mantle-derived He in oceanic basalts and geophysical evidence for significant mass exchange between lower and upper mantle. Moreover, we see no need to invoke hidden or other convectively isolated mantle reservoirs as sources for high 3He/4He ocean island basalts.
DI34A-03
Secular changes in the style of mantle melting and mantle differentiation as constrained by the depths and temperatures of magma genesis
The Earth's mantle differentiates mainly by decompression partial melting induced by solid-state convection. Owing to their low densities, these magmas rise to the surface and drive the formation of oceanic and continental crusts. Because many trace elements, including the heat-producing elements, are partitioned into liquids, the extraction of melts to the surface concentrates these elements to the surface of the Earth and depletes the mantle. However, at high pressures, e.g. greater than 9 GPa, magmas become dense enough to be negatively buoyant. High pressure melting occurs only if the Earth was sufficiently hot to allow for deep intersection of the solidus with the adiabat. To assess how the depth and temperature of melting has changed through time, we compiled an updated experimental dataset to calibrate new SiO2-based and Mg- based thermobarometers for mafic to ultramafic magmas multiply saturated in olivine and orthopyroxene at depth. Mid-ocean ridge basalts yield equilibration Ps and Ts of 0.37-1.2 GPa and 1300-1400 C. Island arc basalts yield similar Ps and Ts but anomalously wet arcs yield slightly lower temperatures and pressures. The deepest basaltic magmatism at present occurs in hotspot regions; for example, post-shield magmas in Hawaii yield Ps and Ts as high as 5 GPa (150 km) and 1600 C. Thus, in the modern Earth, melting is limited to the uppermost 200 km of the mantle. However, in the Earth's first Gy, it may have been hotter than 1700 C as constrained by thermobarometry on 3.5 Gy Barberton komatiites. Initial melting depths of these magmas may have been as high as 8 GPa. We show that melt compositions formed at 9 GPa or greater are negatively buoyant, suggesting that prior to 3.5 Gy ago, melting occurred deep enough to generate sinking magmas, which percolated downwards to impregnate the underlying mantle. After a critical fraction of melt impregnation is met, the aggregate would convectively sink into the lower mantle. Because high P and T melts have high Fe and are enriched in incompatible trace elements, including K, U, Th, and primordial He, their downward transport would lead to the generation of a dense and stable chemical boundary layer enriched in the same incompatible elements that today are extracted to the Earth's surface. These chemical piles would have negative shear velocity, positive density, and nearly negligible bulk sound velocity anomalies, consistent with the lack of correlation of these parameters in the deep mantle. The tops of these chemical piles could be the roots of thermal plumes that give rise to hotspots. The most important consequence of our model is that generation of a primitive-like lower mantle reservoir with an "undegassed" signature is the result of processing the entire mantle through melting, thus, there unprocessed portions of the mantle are not needed to explain the apparent elemental imbalances of the Earth. Finally, generation of this chemical boundary layer could have left behind an upper mantle with non-chondritic relative abundances of the elements, suggesting that models of the bulk silicate Earth based on upper mantle rocks and chondritic references may need revision.
DI34A-04 INVITED
A comprehensive approach to seismic anisotropy as a tool to understand the origin and dynamics of the asthenosphere
Isotropic velocity anomalies as mapped throughout the mantle are broadly consistent with global flow model predictions for descending slabs as well as rising plumes connected to ocean island chains. We discuss progress on the interpretation of seismically anisotropic maps of the upper mantle. An integrated approach that combines mineral physics, seismology, and geodynamics indicates that seismic anisotropy can put quantitative bounds on mantle flow and rheology. A synoptic picture from lab-samples to plate-scales indicates how deformation over the last tens of million years is recorded underneath oceanic plates, while Wilson-cycle collisions are frozen into old continental lithosphere and keels. Flow models can roughly match patterns of azimuthal anisotropy in oceanic plates, radial anisotropy averages in the upper mantle, and radial anisotropy patterns. These observables provide new constraints on dynamics, including the partitioning between diffusion and dislocation creep, the lack of decoupling between upper and lower mantle, and azimuthal anisotropy imposes a speed limit on net rotations of the whole lithosphere. We argue that second order features such as variations in olivine textures as a function of water content may be inferred from residual anisotropy, i.e. the deviations of mapped anisotropy from geodynamic background flow models. Geodynamics can provide a priori information to tie together seemingly distinct seismological observations, and seismic anisotropy promises to evolve into a key quantitative tool to understand the origin and role of the asthenosphere for both oceanic and continental plate dynamics.
DI34A-05 INVITED
Compositional Stratification in the Upper Mantle: Seismic Evidence and Geodynamic Implications
The study of interior dynamics requires knowing the current physical properties (e.g., rheology, elasticity) and understanding the relations between the driving forces and their effects, which lead to the current conditions. Mineral physics provides a description of these physical properties as a function of composition (C), temperature (T) and pressure (P). The knowledge of T-C conditions in the Earth relies on the interpretation of geophysical data based on mineral physics. Recently, we inverted long-period seismic waveforms directly for temperature or composition. We found that lateral variations of temperature can explain a large part of the data in the upper mantle. However, the radial average VS profile we obtain cannot be explained with a homogenous composition. A relative enrichment with depth in garnet and pyroxenes is consistent with seismic data. A similar compositional structure characterizes recent geodynamic models that include the phase transitions of the non-olivine system. Here, we use the same physical properties for the seismic inversion and to model the thermochemical evolution of the mantle with the code STAGYY. The flow field is computed with a finite-volume multigrid solver and tracers are used for composition. Multi-component phase changes and melting-induced compositional differentiation are included. The principal features of the modeled T-C structure are compared with the one inferred from observations and the presence of compositional stratification in the upper mantle is examined in details. In addition, the thermo-chemical models give insights on small-scale heterogeneities that are not resolved from the geophysical observations. In order to account for the large uncertainties of some key mineral physics parameters at mantle conditions (e.g., viscosity and seismic attenuation, density relation between depleted and enriched compositions), the seismic inversion and the modeling are repeated using different possible parameters.
DI34A-06 INVITED
The Nature of Lower Mantle Seismic Anomalies
Lateral variation of temperature is the dominant influence on seismic anomalies in the upper mantle through contributions from the intrinsic temperature derivatives, velocity dispersion as a result of anelasticity, and phase equilibria. Although intrinsic temperature derivatives are suppressed at higher pressures, high velocitiy anomalies are expected to persist within the cold cores of slabs subducted into the lower mantle. The apparent general failure of this expectation has fueled continued speculation for over a decade concerning the fate of subducted slabs and the nature of mantle circulation. The discovery of a high-spin to low spin transition in iron-bearing lower mantle phases provides a mechanism to further suppress intrinsic temperature derivatives of elasticity. Key here is the thermodynamic nature of spin transitions and its connection to elasticity. In a pressure regime of 40 to 60 GPa all three elastic constants of ferropericlase smoothly pass through minima with respect to an extrapolation of the pure high-spin state. This behavior is consistent with a macroscopic thermodynamic description that predicts no sharp changes in spin fractions and dependent physical properties. A significant consequence is that in the spin transition pressure regime, normal temperature derivatives of elasticity are reduced. Within Earth, a minimum in the temperature derivatives caused by the transition is predicted to occur at a depth of 1500 km. This correlates well with the depth where seismic tomographic results show minimum structure and model power. Thus, the spin transition may serve to mask or even invert velocity anomalies associated with lateral temperature differences.
DI34A-07 INVITED
The role of iron chemistry on the interpretation of lower mantle heterogeneities
Iron is a major element in the mantle and its chemical behavior (partitioning, spin transition..) affects the physical and transport properties of the phases which host it. Such variations can provide explanations of major heterogeneities observed in the mantle. Magnesium silicate perovskite (Mg,Fe)SiO3 (Mg-pv) and ferropericlase (Mg,Fe)O (fp) are the dominant phases in the lower-mantle and can potentially host significant amount of iron. It is thus of prime importance to constrain element partitioning at high pressure for improving models of the deep Earth. We investigated iron partitioning between Mg-pv and fp synthetised under lower- mantle conditions (up to 115 GPa and 2200 K) in a laser heated diamond anvil cell (LH-DAC). Recovered samples were thinned to electron transparency by focussed ion beam (FIB) and characterized by analytical transmission electron microscopy (ATEM). Addititional informations on trace elements were provided by measurements using nanometer scale ion probe (nanoSIMS). Iron concentrations in both phases were obtained from EDX measurements and nanoSIMS and are in excellent agreement. Our results are the first to show that recently reported transitions in the lower-mantle directly affect the evolution of Fe-Mg partitioning between both phases. Mg-pv is increasingly iron-depleted above 70-80 GPa possibly due to the high spin- low spin transition of iron in fp. Conversely, the perovskite to post-perovskite transition is accompanied by a strong iron enrichment of the silicate phase. We will discuss the implications of these partitioning variations in terms of potential heterogeneities. We will also address shortly the early history of the Earth, as the observation of nanoparticles of metallic iron in the Mg-pv bearing runs suggests the disproportionation of ferrous iron. These particles were not observed in post-perovskite (ppv) bearing sample. Implications on the oxidation state of the Earth and core segregation will be discussed.
DI34A-08 INVITED
Thermal vs. Elastic Heterogeneity in High-Resolution Mantle Circulation Models with Pyrolite Composition: High Plume Excess Temperatures in the Lowermost Mantle
We study a new class of high-resolution mantle circulation models and predict their corresponding elastic heterogeneity. Absolute temperatures are converted to seismic velocities using published thermodynamically self-consistent models of mantle mineralogy for a pyrolite composition. A grid spacing of ~25 km globally allows us to explore mantle flow at earth-like convective vigor so that modeled temperature variations are consistent with the underlying mineralogy. We concentrate on isochemical convection and the relative importance of internal and bottom heating in order to isolate the thermal effects on elasticity. Models with a large temperature contrast on the order of 1000 K across the core-mantle boundary, corresponding to a substantial core heat loss of up to 12 TW, result in elastic structures that agree well with tomography for a number of quantitative measures: These include spectral power and histograms of heterogeneity as well as radial profiles of root-mean-square amplitudes. In particular, high plume excess temperatures of +1000--1500 K in the lowermost mantle lead to significant negative anomalies of shear wave velocity of up to -4%. These are comparable to strong velocity reductions mapped by seismic tomography in the prominent low- velocity regions of the lower mantle. We note that the inference of a large core heat flux is supported by a number of geophysical studies arguing for a substantial core contribution to the mantle energy budget. Additionally, we find significant differences in the characteristics of thermal and elastic heterogeneity in the transition zone due to phase transformations of upper mantle minerals. Our results underline the necessity to include mineral physics information in the geodynamic interpretation of tomographic models.