DI23A-1747
Is the mantle transition zone globally dry?: Observational evidence from electrical conductivity
As recent studies pointed out (e.g., Bercobici and Karato, 2002; Ohtani et al., 2004), the mantle transition zone is characterized by the high water storage capacity (higher than the upper or lower mantle). Because the presence of water is an important factor in geodynamic processes, it is worth for making efforts to reveal how water distributes in the transition zone. Geomagnetic induction method is one of possible geophysical methods that enables us to approach this problem in terms of electrical conductivity. Here we focus how the conductivity distributes in the transition zone from a global view point by examining one-dimensional (1-D) mean conductivity profile. The laterally heterogeneous conductivity will be examined later in the future, as it is highly anticipated if we consider the importance of the up- and down-welling of mantle flow in the Earthfs water cycle and small diffusion length expected by mineral physics. We analyzed the dataset of geomagnetic field and cable voltage observations in the north Pacific region and renewed one-dimensional conductivity model. Difference from the past models (Utada et al., 2003; Kuvshinov et al., 2005) is two folds: additional time series data till the end of 2006 were used while data till the end of 2000 were used in our previous study, and ocean induction effects were carefully corrected more than the previous case. Now our revised model profile is compared with the experimental values by Yoshino et al. (2008) . It was found that the new conductivity profile is more or less consistent with experimental values for dry samples down to the lower mantle, except those at the shallower part of the transition zone where observational values are higher than experimental ones by about a factor of three. This result infers that the transition zone beneath the North Pacific Ocean contains about 0.3-0.5 wt % of water at its shallower part.
DI23A-1748
Is there significant amount of water in the transition zone? Perspective of electrical conductivity
Water in the earth's mantle plays important roles in the mantle dynamics. Although knowledge of storage capacity of mantle minerals has been obtained by many studies, water content in the mantle cannot be directly estimated from these studies. Electrical conductivity profiles of the Earth's mantle obtained from the electromagnetic studies can provide constraints on the thermal and chemical states of the mantle. Especially, electrical conductivity is very sensitive to small amount of hydrogen contents in nominally anhydrous minerals. Accurate knowledge of electrical conductivity of mantle minerals is needed to constrain water contents in the mantle as a function of depth. The mantle transition zone could be a large water storage because wadsleyite and ringwoodite can store significant amounts of water in their crystal structure. Water content in the transition zone could be quantified using electrical conductivities of hydrous wadsleyite and ringwoodite. Although the correct interpretation of water content depends on accurate knowledge of two conduction mechanisms (small polaron and proton conductions) in these minerals, early studies failed to distinguish them. In the present study, electrical conductivities of wadsleyite, ringwoodite and majorite garnet, which are main constituent minerals in the mantle transition zone, were measured as functions of water content and temperature, rendering it possible to separate contributions of small polaron and proton conduction mechanisms. The starting materials were powder of San Carlos olivine (Fo91) for wadsleyite and ringwoodite and powder of synthetic glass with a composition of pyrolite minus olivine for majorite garnet. The electrical conductivities of the samples were measured at various pressures from 16 to 23 GPa and temperatures up to 2000K. For all the dry samples, electrical conductivity displays Arrhenian behavior over the entire investigated temperature range. In the high temperature range above 1700 K, activation energies (more than 1.5 eV) tend to be higher than those in the lower temperature range (less than 1.5 eV). The absolute values of electrical conductivity (S/m) increase in order of wadsleyite, majorite garnet and ringwoodite. From these measurements we noted that the electrical conductivities of dry wadsleyite and ringwoodite (less than 100 wt. ppm of water) are much lower than those previously reported. While conductivities of samples with certain amounts of hydrogen are comparable to that of the dry one, conductivity increases with increasing hydrogen concentrations. Activation energies of hydrogen-bearing nominally anhydrous minerals decreases with increasing hydrogen concentration from nearly 1 to 0.5 eV. The contributions of proton conduction are small at temperatures corresponding to the mantle transition zone. For the mantle transition zone, conductivity jumps in association with the olivine-wadsleyite and wadsleyite- ringwoodite transitions have similar magnitude (~0.7 log unit). The presence of majorite garnet has little effect on the conductivity-depth profile of the upper mantle. The dry mantle model well explains the current semi-global conductivity-depth profiles obtained from the recent geoelectromagnetic studies. There is no necessity to introduce water in the mantle transition zone.
DI23A-1749
Electrical conductivity of pyrope garnet at high temperature and high pressure
Garnet is one of the important constituent minerals in the upper mantle and the transition zone of the Earth's
mantle. However, there were very little previous works on its electrical conductivity. We have measured the
electrical conductivity of single crystal garnet under the conditions of 4-16 GPa, 873-1473 K and frequency
range from 10-2 to 106 Hz, with a range of water content (from less than 10 to 8000 H/106Si).
A KAWAI-type multi-anvil apparatus and Solarton-1260 Impedance/Gain Phase analyzer are used in this
study. The impedance spectra show two circles correspondent to the intrinsic resistivity of the crystal and the
effects of charge accumulation at the electrodes. The DC electrical conductivity was determined by the
impedance spectroscopy. Molybdenum and molybdenum oxide solid buffer was selected to control the
oxygen partial pressure. Results on hydrous and anhydrous samples were compared to determine the
influence of water content on the electrical conductivity of single crystal garnet. Under anhydrous conditions,
the electrical conductivity of pyrope garnet increases with temperature and decreases with pressure. When
we used a thermal activation parameterization,
σ ∝ exp(-(E*+PV*)/RT), E*≈ 150
kJ/mol and V*≈ 2 cm3/mol.
Hydrous garnet crystals have significantly higher electrical
conductivity with different temperature and pressure sensitivity, and the conductivity in these samples
increases with the water content. The results can be summarized as
σ ∝ [Cγw]exp(-
(E*+PV*)/RT) with γ≈ 0.6, E*≈ 70 kJ/mol and V*≈ -0.6
cm3/mol. These results are similar to those obtained in our group for olivine, wadsleyite and ringwoodite
and we suggest that the mechanism of hydrogen conduction is likely common to these minerals. We conclude
that hydrogen enhances the electrical conductivity of pyrope garnet and its effect increases with pressure
but decreases with temperature. At a typical pressure and temperature in the upper mantle, the influence of
water is substantial.
DI23A-1750
Deformation of dry Wadsleyite up to 18 GPa and 2100 K Using a Rotational Drickamer Apparatus
Shear deformation experiments on dry polycrystalline wadsleyite have been conducted at 14.7-18.0 GPa, 1690-2100 K and strain rates of 2.6-16 x 10-5 s-1 using a rotational Drickamer apparatus (RDA) at a synchrotron facility. The stress was measured from the orientational dependence of lattice spacing for the (013), (211), (141), (240) and (244) planes. Based on the mechanical and microstructural observations, we infer that deformation occurs by power-law creep involving dislocation glide and climb under relatively high temperature conditions, whereas deformation is due to exponential creep through the Peierls mechanism at relatively low temperatures. A sample deformed at the highest temperature showed significant grain-size reduction, and most of small grains are dislocation free. We interpret that these observations are evidence for dynamic recrystallization and that diffusion creep (and grain-boundary sliding) plays an important role after dynamic recrystallization. Consequently, the strength determined in the highest temperature condition provides a lower limit for the power-law dislocation creep. We conclude that the strength of wadsleyite in the power-law dislocation creep is higher than or comparable to that of olivine. The strength of wadsleyite in the Peierls regime is similar to that of olivine.
DI23A-1751
Single-crystal Elasticity of Wadsleyite With 1.7 wt % H 2 O to 11 GPa by Brillouin Scattering
Polymorphs of olivine have the greatest water storage capacity among all the nominally anhydrous mantle minerals (e.g. Bolfan-Casanova et al., 2000). Wadsleyite (β-Mg 2SiO 4) is able to contain up to 3.3 wt% H2O (Smyth et al., 1987). This phase is considered as the dominate mineral in the mantle from 410 km to 520 km depth. A previous study showed that water decreases the elasticity of wadsleyite strongly at ambient conditions (Mao et al., 2008), but has no detectable effect on the pressure derivatives of the bulk and shear moduli of wadsleyite containing 0.84 wt% H2O (Mao et al., in press). The effect of H 2O content on high-pressure elasticity has not been investigated for samples with larger water contents. In this study, we performed high-pressure Brillouin measurements on a single crystal of wadsleyite with 1.7 wt% H2O to obtain the elastic tensor, hence the aggregate elastic moduli and their pressure derivatives. Two crystals were cut and polished into 30~40μm-thick platelets. Single- crystal x-ray diffraction was conducted at x17C of Brookhaven National Laboratory to determine the orientations of the two platelets. The samples were measured at 7 pressures steps up to 11 GPa. For each platelet, 19 spectra were collected at 10° intervals at each pressure step in order to cover a range of 180° degrees. Preliminary result shows that the elastic constants, Cij, have the similar trends as anhydrous and 0.84 wt%-H2O wadsleyite while increasing pressure. Pressure derivatives of bulk and shear moduli are 4.1 (2) and 1.3 (1) for 1.7 wt%-H2O wadsleyite, which are not different within uncertainty from those of anhydrous and 0.84 wt%-H2O wadsleyite. (Zha et al., 1997; Mao et al., in press). This suggests that the pressure derivatives of elastic moduli do not change with water content at least up to 1.7 wt% H2O. It further confirms the previous inference that at least 1 wt% H2O in wadsleyite at 410 km is required for a pyrolite composition (60 vol% olivine) to match seismic models (Mao et al., in press).
DI23A-1752
Fine-Scale P-wave Structure in the Upper Mantle North of Australia
We study the P-wave field recorded by the high frequency (centering near 1 Hz) Warramunga seismic array (WRA) in Australia. WRA has an aperture of about 20 kilometers and is equipped with 20 short-period vertical component instruments. Nearly 2000 earthquakes at all source depths are studied, which span an epicentral distance range of roughly 12 to 25 degrees. Our initial work focused on array methods to identify relatively weak but coherent P-wave triplication arrivals due to the upper discontinuities near 410 and 660 km depth. However, significant waveform complexities and additional unpredicted seismic arrivals are commonly present, suggesting small scale structural complexities either associated with the discontinuities, or heterogeneity along the path. The additional energy does not appear source related, nor clearly attributable to any particular wave path geometry or region. Here we explore back-projection schemes to map arrivals to possible scattering or reflection locations. A challenge with this approach is confident identification of a reference phase; thus, we proceed using only the highest quality data. Nth-root stacking and fk-analysis are used to determine slowness, delay time, and back azimuth, which are used to estimate reflection locations. Results will be presented in the context of the tectonically complex region, replete with past and present subduction, and hence potentially the presence of associated chemical heterogeneity.
DI23A-1753
Amplitude Ratios of Reflected Phases: Implications for the Aluminum Content of the Mantle at the 660 km Discontinuity
The depth and sharpness of the 660 km seismic discontinuity arising from solid-state phase transformations at the base of the upper mantle are strongly dependent on the composition and thermal state of the mantle. This discontinuity is generally attributed to the phase transformation of ringwoodite to Mg-perovskite plus magnesiowuestite at 660 km depth in the Earth, though there is also likely a contribution from the phase transformation of majorite to Al-bearing perovskite near this depth. Transformation of the majorite phase contributes to the magnitude of the P-wave velocity contrast across this boundary and is particularly sensitive to mantle composition; if the Al content at 660 km depth is above several percent, and the substitution of Al into perovskite is associated with formation of oxygen vacancies, the bulk modulus of perovskite can be lowered by as much as 10%. Lateral perturbation in mantle temperature reduces the effect of aluminum on bulk modulus, and strengthens the contrast across this boundary. Here, we investigate the velocity contrast across the 660 km discontinuity by examining the amplitudes of seismic waves that reflect off of the underside of the discontinuity, referenced to waves that reflect of the underside of Earth's surface (the former arrive first, and are called precursors). Our dataset consists of over 150,000 seismograms that sample beneath the Pacific Ocean and the South American continent. We combine the seismic observations with mineral physics modeling of the bulk modulus contrast at 660 km depth for Al-bearing perovskite. The amplitude ratio of the 660 km discontinuity precursor is much lower than predicted for the reference Earth model PREM, requiring a reduction in the bulk modulus contrast across this boundary, relative to PREM, by over 50%. We find that the relatively small, observed amplitude ratios are most readily explained by the presence of perovskite with a bulk modulus that is markedly depressed relative to that of Al-free perovskite. Accordingly, our results indicate that the predominant perovskite phase at the top of the lower mantle is softened by vacancy- associated substitution of Al, and our comparison with mineral physics models provide constraints on not only the bulk modulus contrast across the 660 km discontinuity, but also mantle temperature and composition at this depth.
DI23A-1754
Splitting of the 520 and 660 km discontinuities in receiver functions and implications for mantle dynamics
The existence and nature of the transition zone discontinuities is important for our understanding of the internal dynamics of the Earth, in particular the mixing between the upper and lower mantle. Here, we use seismic Pds receiver functions as they very suitable to study the detailed characteristics of discontinuities at a range of frequencies and locations. Our global data set shows that the 660 km discontinuity is split into double peaks in many different regions. The splitting is not correlated to tectonic areas, and can be found in both subduction zone areas as well as in less tectonically active regions. A previous study showed that the 520 km discontinuity was split in SS precursors (Deuss & Woodhouse, 2001). Here, we confirm the splitting of the 520 km discontinuity in receiver functions and show several examples where, for the same location, both the receiver function and SS precursor stacks show robust splitting. Splitting of transition zone discontinuities requires phase transitions in garnet in addition to olivine phase transitions. However, phase transitions alone cannot explain the regional variation of our observations. Widespread splitting on a global scale can be explained by the presence of accumulated slab material in the transition zone. The 660 km discontinuity would, at least partially, provide a boundary allowing some differentiated material to accumulate either above it (splitting the 520 km discontinuity), or below it (splitting the 660 km discontinuity). This is in agreement with recent thermo-chemical mantle convection models.
DI23A-1755
The role of thermal expansion and volume changes on the slab stress distribution
The distribution of the stress within the subducted lithosphere, especially at the depth of the transition zone, is still poorly understood and is crucial for the understanding of deep focus earthquakes. Using a 2D Cartesian finite-volume code, we present numerical models of subduction with emphasis on the slab's stress distribution. Using a non-linear viscoelastic rheology, we solve the thermomechanical problem for a subducting slab with a simplified mineralogical composition and take into account the effects of major phase transitions at 410 km and 660 km. In particular we describe the stress field in relation to the thermal expansion from the warming of the slab and to the volumetric changes induced by phase transitions, which, so far, have only been considered within the framework of strongly simplified kinematic subduction models.
DI23A-1756
Model for Transition Zone Formation from Upwelling Thermo-Chemical Plumes of Intermediate Rheology
The mantle transition zone has been limited to a layer of approximately 250 km at the base of the upper mantle, identified by studies in seismology and mineral physics. However, there are many uncertainties as to the nature of formation of this mid-mantle layer, its evolution over geological time, physical properties, and its role facilitating or inhibiting whole mantle flow. Here, we conduct laboratory fluid experiments using high viscosity corn syrup fluids and liquid gallium to study mantle convection processes in the early Earth. Specifically, we consider early core formation events involving metal-silicate plumes which sink following impact events and entrain magma ocean material from the surface during descent. Preliminary studies indicate that low viscosity, buoyant material, that makes up the model magma ocean near the surface is entrained in conduits that form behind quickly descending liquid metal plumes to the base of the lower mantle. This low density material brought to the base of the model lower mantle becomes buoyant and subsequently rises back up to the top of the fluid box forming a new intermediate material that has experienced both chemical and thermal diffusion along its mantle pathway and empties at the top of the lower mantle or base of a magma ocean. Two-component fluid experiments are considered in the presence of a hot lower thermal boundary layer at Rayleigh numbers of 103 to 105 and low Reynolds number flow, and indicate that upwelling thermo-chemical plumes may form following core formation events. This new third fluid layer of intermediate rheology is considered as a model for the mantle transition zone. Shadow graph images indicate a sharp density contrast with surrounding fluids that persists for long times, consistent with seismic discontinuities observed for the Earth's transition zone. We will present quantitative estimates of material rheology, density, and flow properties for scaling with a silicate mantle, geophysical, and geochemical studies.
DI23A-1757
Trace Elements in Basalts From the Siqueiros Fracture Zone: Implications for Melt Migration Models
Incompatible trace element (ITE) ratios in MORB from a variety of locations may provide insights into the melt migration process by constraining aggregated melt compositions predicted by mantle melting and flow models. By using actual plate geometries to create a 3-D thermodynamic mantle model, melt volumes and compositions at all depths and locations may be calculated and binned into cubes using the pHMELTS algorithm [Asimow et al., 2004]. These melts can be traced from each cube to the surface assuming several migration models, including a simplified pressure gradient model and one in which melt is guided upwards by a low permeability compacted layer. The ITE ratios of all melts arriving at the surface are summed, averaged, and compared to those of the actual sample compositions from the various MOR locales. The Siqueiros fracture zone at 8° 20' N on the East Pacific Rise (EPR) comprises 4 intra-transform spreading centers (ITSCs) across 140 km of offset between two longer spreading ridges, and is an excellent study region for several reasons. First, an abundance of MORB data is readily available, and the samples retrieved from ITSCs are unlikely to be aggregated in a long-lived magma chamber or affected by along-axis transport, so they represent melts extracted locally from the mantle. Additionally, samples at Siqueiros span a compositional range from depleted to normal MORB within the fracture zone yet have similar isotopic compositions to samples collected from the 9-10°EPR. This minimizes the effect of assuming a uniform source composition in our melting model despite a heterogeneous mantle, allowing us to consistently compare the actual lava composition with that predicted by our model. Finally, it has been demonstrated with preliminary migration models that incipient melts generated directly below an ITSC may not necessarily erupt at that ITSC but migrate laterally towards a nearby ridge due to enhanced pressure gradients. The close proximity of the ITSCs at Siqueiros to the large ridges bounding the fracture zone provide a good opportunity to model this phenomenon and may help explain the variable ITE ratios found between samples collected within the transform and those near the ridges.
DI23A-1758
Si-29 NMR Results on Forsterite, Wadsleyite and Ringwoodite: Structural Disorder and Effects of Magnetic Cations
The most abundant mineral in the lower part of the transition zone of the Earth's mantle is probably ringwoodite (spinel structured [Mg,Fe]2SiO4). Recent theoretical calculations (Panero, JGR, in press) have shown that significant, thermally activated Mg,Si disorder should occur in ringwoodite at mantle pressures and temperatures, with important consequences for elastic constants and seismic wave velocities. However, direct characterization of the extent of this disorder, marked by tetrahedral Mg and octahedral Si, has been difficult because of the similar x-ray scattering of Mg and Si. Iron-free, Si-29 enriched, anhydrous synthetic forsterite starting material was used to synthesize ringwoodite and wadsleyite in the 5000-tonne multi-anvil press at the Bayerisches Geoinstitut, at 1500 +- 50 deg C and 24 and 20 GPa, respectively. Estimated temperature-quench times from 1500 to <400 deg C are less than 5 s. We report here detailed Si-29 NMR studies on the forsterite starting material, ringwoodite, and wadsleyite. With the isotopically enriched starting material, the obtained very high signal-to-noise (S/N) ratios (>1000) would allow octahedral Si to be detected at the level of a fraction of 1 %. However, only tetrahedral Si was observed in all samples. This suggests that Mg,Si site disorder present in ringwoodite at high temperature may revert to an ordered distribution on cooling, as has been observed directly for the highest temperature component of the Mg,Al disorder in MgAl2O4 spinel. An additional, small, broad SiO4 peak seen in ringwoodite may be a result of imperfect re-ordering during the rapid quench process. The effects of minor components of paramagnetic dopant cations (e.g. 0.2 % added Co2+ and "natural" impurities in the reagents) on the NMR spectra and on spin-lattice relaxation have also been studied to better understand the small details of the spectra revealed by such high S/N. As is typical for Si-29 in silicates, relaxation is clearly non-exponential and instead initially follows a power law, as expected when controlled by direct coupling to magnetic impurities. Remarkably, in the forsterite, at least 12 small, "extra" NMR peaks are observed, many of which are at chemical shifts well above and well below any previously observed for SiO4 groups in oxides (e.g. -28 and -129 ppm). It is likely that dilute paramagnetic ions, at discrete (second cation neighbor?) distances and site orientations from the SiO4 sites, give rise to these features, probably via a through-space, dipolar "pseudocontact" shift mechanism (Grey et al., JACS 1990), which has apparently not been previously reported in silicates. These "extra" peaks are absent in spectra of ringwoodite, presumably because of the higher symmetry of its structure. Full characterization of such features could potentially allow the "mapping" of the distribution of such ions in complex silicate structures.
DI23A-1759
alpha to beta to gamma transformations in Mg2SiO4 and mantle discontinuities
Phase relations in Mg2SiO4 have been investigated by first principles quasiharmonic calculations. The computed phase boundaries obtained using the local density approximation (LDA) and the generalized gradient approximation (GGA) bracket the experimental ones, with LDA (GGA) calculations giving the lowest (highest) bound, while the Clapeyron slopes are in good agreement with the experimentally determined ones. This is the same trend displayed by previous similar computations of polymorphic phase boundaries. Further analyzes reveal that despite the uncertainties in phase boundary determination, the calculated discontinuities in density, bulk modulus, and bulk sound velocity are quite insensitive to pressure and have small uncertainties and useful accuracy to discriminate potential sources of discontinuities in the mantle. We verify that ~3% density discontinuity at 410-km depth can be produced primarily by the alpha to beta transition in an aggregate with pyrolite composition. However, the 1.3--2.9% density discontinuity observed in some places at 520-km depth cannot be accounted for solely by the beta to gamma transition and requires also changes in the coexisting pyroxene/garnet/Ca-perovskite system. Research supported by NSF/EAR 0635990, and NSF/ITR 0428774 (VLab). Computations were performed at the Minnesota Supercomputing Institute.