Composition, Mixing, and Melting of the Mantle I: Physical and Chemical Properties of the Lowermost Mantle
Presiding: S Akber-Knutson, California Institute of Technology; C Lee, Rice University; M Jellinek, University of British Columbia
V41A-01 08:30h
Extinct Nuclides and the Origin and Earliest History of the Earth
There are well established and measurable 129Xe/130Xe, 136Xe/130Xe, 182W/183W and 142Nd/144Nd isotopic variations in the Earth due to decay of the now extinct nuclides 129I, 244Pu, 182Hf and 146Sm. These nuclides record primarily differentiation processes which acted during the first ~300 million years of Earth history (4.57-4.25 Ga). Tracing isotopic variations of these systems through the geologic history may be used to study re-mixing of early formed reservoirs. This may, for example, provide a simple way of determining the rate of crustal recycling into the mantle. Among these systems the 182Hf-182W system is the only one directly related to the core formation process (and, by inference, accretion) of Earth. The 182Hf-182W system provides strong constraints which yield a mean time of accretion and core formation of ~10 Ma. The coupled 146Sm - 147Sm chronometer makes it possible to constrain the age of the initial silicate differentiation in the mantle source region of some of the Earth's oldest surviving crustal rocks to ~4.47 Ga. This also puts an upper limit on the total accretion interval of ~100 Ma which is consistent with the extraction of a primordial crust enriched in LREE from the mantle after the last giant impact on Earth. The I-Pu-Xe system is fundamentally unreliable as a chronometer of accretion. The presence of a large 129Xe excess in the deep Earth is consistent with a very early formation and a short time interval for the accretion of the Earth, but may primarily date a combination of gas/dust fractionation in the solar nebula and the closure of the atmosphere, rather than accretion itself.
V41A-02 INVITED 08:45h
Lower Mantle Tomography and D" Triplications
A relatively strong secondary phase (Scd) occurs between S and ScS beneath regions with relatively fast velocities in tomographic images; Alaska, Northern Eurasia, India, Central America, etc. The latter images are usually interpreted as cool slab material subducted around the circum-Pacific. Sidorin et al. (1999) argued that the triplication (Scd) is controlled by the combination of these enhanced velocity gradients (Grand's model) and a small positive velocity jump (1.5%) induced by a phase-change, occurring at a reference depth of 200 km with a clapreyron slope of about 6MPa/k (possible Post-Perovskite). Assuming such a mechanism, we can predict the elevation of the discontinuity above the CMB globally using thermal anomalies inferred from tomographic models. Many of the observed variations in travel time and amplitude Scd (triplicated phase) can be predicted by this model where the differential times reach a few seconds and the amplitude (Scd/S) ratios reach 50%. Strengths in the Pacific fall below 10% where only stacking can resolve such small features making it difficult to map coherent boundaries. After reviewing the 2D analyses producing these features, we present some new waveform modeling results introduced by more complex velocity-temperature mapping and explore 3D effects beneath Central America, where vary rapid changes in Scd are observed.
V41A-03 INVITED 09:05h
Iron/Manganese Variations in the Mantle
Can thermochemical models of mantle convection be tested by geochemical data? The answer is: Yes, but with real difficulty. The bulk of the mantle is inaccessible to direct laboratory analysis: mantle samples are limited to xenoliths (lithosphere), abyssal peridotites (MORB), and alpine peridotites (MORB or arc). An indirect sample of the mantle is provided by melts (basalts, picrites, etc.). The major breakthroughs in Mantle Geochemistry of the past 3-4 decades have come from measurements of isotope ratios and incompatible trace element abundances in melts. The links between major element variations (Fe/Mg, Si/Mg) that control the geophysical observables of the mantle and the geochemical observables (FOZO, etc.) are weak. Combined geodynamic and tomographic models of mantle convection indicate variations of 10% in the Fe content of the mantle. The previously published range of Fe/Mn in mantle peridotites is 50-70, and showed no correlation with MgO. We have developed a significantly more precise method of Fe/Mn ratio determination by ICP mass spectrometry. New data for mantle xenoliths show a strong positive correlation between Fe/Mn and MgO, and indicate that the primitive mantle has Fe/Mn 61. Hawaiian basalts and picrites have significantly higher Fe/Mn (67) than MORB (56), or Icelandic picrites (58). Assuming that the mantle source beneath Hawaii has a higher Fe content, these results imply an iron excess of 15-20% relative to ambient mantle. However, the Fe/Mn ratio depends on both the molar Fe/Mg ratio and on the Si/Mg ratio, because of the compatibility of Fe and Mn in olivine vs. pyroxenes. Recent studies of Ni in olivine in Hawaiian basalts have been taken to imply a higher Si content for their mantle source. Thus, it is not yet possible to uniquely distinguish between source variations of Fe/Mg and Si/Mg from the Fe/Mn ratios of Hawaiian melts. Either interpretation implies that the sources of some plumes (Hawaii, but not Iceland) are higher in Fe or Si than ambient mantle (MORB). Mantle geochemistry is finally coming to grips with the single most important question that geophysicists may ask of it: major element variation in mantle plumes that reach the surface to form volcanic islands.
V41A-04 09:25h
Could Ultra-Low-Velocity Zones be Subducted Banded Iron Formations?
Ultra-low velocity zones (ULVZs) are regions of the Earth's core-mantle boundary (CMB) of about 1 to 10 km thick which exhibit seismic velocities which are lower than radial reference models by about 10 to 20% in Vp and 10-30% in Vs, together with a possible increase in density of 0-20%. A number of origins for ULVZs have been proposed, such as ponding of dense silicate melt, core-mantle reaction zones, or underside sedimentation from the core. Here we suggest that ULVZs are relics of banded iron formations (BIFs) subducted to the CMB 2.8 to 1.8 billion years ago. Consisting mainly of interbedded iron oxides and silica, BIFs were deposited in the world's oceans during the late Archaean and early Proterozoic. BIF has a mean density some 25% higher than the mantle, all the way from the Earth's surface to the CMB. As part of the ocean floor, BIFs would therefore be recycled into the Earth's interior by subduction. We estimate that as much as 500 m thickness of BIF could have precipitated on the ocean floor between 2800 and 1800 billion years ago. Such a volume of BIF would produce a 1.3 km thick layer surrounding the outer core, or a 10 km thick layre if it was concentrated into the regions where ULVZs are observed, consistent with estimated ULVZ volume. Recent models suggest that the core may be close to oxygen saturation, in which case BIF could stably pond on top of the outer core. Even if the core is undersaturated in O, capillary rise and diffusion arguments suggest that a km-thick BIF body could survive at the CMB for billions of years. If our model, which we argue is at least as likely as other candidate mechanisms, is correct the presence of subducted BIF in the form of FeO and post-stishovite would explain many of the physical (seismic wave speed, low Vs/Vp, high electrical conductivity) and chemical (high Pt content reservoir below deep plume sources) properties invoked for regions of the CMB. It is perhaps ironic that one of the most enigmatic features of the deepest interior of the Earth may owe its existence to some of the earliest and most primitive life.
V41A-05 INVITED 09:40h
A Geodynamical Study into the Development of Thermochemical Piles Beneath Africa and the Pacific
The large low-velocity seismic anomalies under the Pacific and Africa are often interpreted as being piles of more-dense material. We have performed numerical modeling of thermochemical convection in a three-dimensional spherical geometry in order to determine whether the presence of a dense chemical component can lead to the formation of two large antipodal piles in the Earth's lower mantle. We find that without imposing plate motions on the surface of the model, the dense material forms a network of linear ridges which are passively swept around by downwellings and that thermochemical structures are generally controlled by the geometry of the downwelling system. If plate motion history from 119 Ma to the present was imposed, we find that large thermochemical piles form in the lower mantle under Africa and the Pacific. Furthermore, the general shape of these structures are such that a ridge-like pile forms under Africa and a more rounded pile develops under the Pacific, consistant with seismic tomography. These general features of our models hold for a wide choice of density contrast and initial layer thickness.