MR42A-01
CaSiO3-CaTiO3-MgSiO3 Phase Relations from 20-50 GPa and the Formation of Ca- perovskite Inclusions in 'Ultradeep' Diamonds
MR42A-02 INVITED
The Perovskite to Post-Perovskite phase transition in Al-bearing (Mg,Fe)SiO3: A XANES in-situ analysis at the Fe K-edge
Phase transition from perovskite (Pv) to Post-Pv (PPv) phase in MgSiO3 has been studied by many groups since its discovery in 2004 (1,2) and the different studies find similar transition pressures. The effect of Al and Fe on the phase transition remains more controversial. The most recent studies suggest an increase of the transition pressure with increasing Fe-content (3,4), but other experimental work (5) as well as ab-initio calculations (6) show the opposite effect. The effect of Al was reported to increase slightly the pressure transition to the CaIrO3 form (4,7), but its influence on the Fe3+ content in the PPv phase has not been documented yet. By means of in situ study of the Fe K-edge fine structures (XANES), we investigated the phase relations between Pv and PPv phases for three different Al-(Mg,Fe)SiO3 compositions. For this, we synthesized various Pv and PPv mixtures using laser-heated diamond anvil cell (DAC) for pressures between 60 and 170 GPa. The sample's mineralogy, i.e. the Pv and PPv phase fractions, was determined using in-situ X-ray diffraction at the ID27 beamline of the ESRF (8,9). Then, we probed the Fe speciation, i.e. the Fe concentration in each phases, in-situ in the DAC using the ĩ-XANES mapping technique available at the ID24 beamline (10,11). Both pieces of information were combined to retrieve the Fe partitioning coefficient between the two high-pressure phases. Our results show that Fe partitions strongly into the PPv phase, which implies a very large binary loop of coexistence of the two phases. Thus, at the core-mantle boundary pressure (135 GPa), the Pv and PPv phase always coexist for all geophysically relevant Al-(Mg,Fe)SiO3 compositions, and the Fe-content in the PPv-phase is only a few percent. References: 1. M. Murakami, K. Hirose, K. Kawamura, N. Sata, Y. Ohishi, Science 304, 855 (2004). 2. A. R. Oganov, S. Ono, Nature 430, 445 (2004). 3. S. Tateno, K. Hirose, N. Sata, Y. Ohishi, Phys. Earth Planet. Inter. 160, 319 (2007). 4. D. Nishio-Hamane, T. Nagai, K. Fujino, Y. Seto, N. Takafuji, Geophys. Res. Lett. 32, L16306 (2005). 5. W. L. Mao et al., PNAS 101, 15867 (2004). 6. J. P. Brodholt, A. R. Oganov, personal communication. 7. S. Ono, A. R. Oganov, T. Koyama, H. Shimizu, Earth Planet. Sci. 246, 326 (2006). 8. N. Guignot, D. Andrault, G. Morard, M. Mezouar, Earth Planet. Sci. 256, 162 (2007). 9. E. Schultz et al., High Press. Res. 25, 71 (2005). 10. S. Pascarelli, O. Mathon, M. Muņoz, T. Mairs, J. Susini, J. Synch. Rad. 13, 351 (2006). 11. M. Muņoz et al., Geochemistry Geophysics Geosystems 7, Q11020 (2006).
MR42A-03
Thermodynamic Phase Relations in the MgO-FeO-SiO2 System in the Lower Mantle
The perovskite (Pv) to post-perovskite (PPv) phase transition at pressures near the Earth's core-mantle boundary (CMB) is currently the favored candidate for explaining most, if not all, of the peculiarities of the D" layer (~200 km region above core) [1, 2]. Additionally, the pressure- and temperature-dependence of this phase boundary in the Earth provides the possibility of an important new thermo-barometer at the bottom of the convecting mantle. The post-perovskite phase boundary in pure MgSiO3 is fairly well known, but the experimental and calculated results on the partitioning of Fe among the stable coexisting phases and its influence on the transition pressure are currently contradictory [3, 4, 5, 6]. Using density functional theory (DFT), we investigate the MgO-FeO-SiO2 ternary system over the temperatures and pressures relevant to the core-mantle boundary. We use DFT to calculate the energies of the relevant stable phases (Pv, PPv, (Mg,Fe)O magnesio-wustite, and SiO2 stishovite) for a range of Fe compositions along the Mg-Fe binary. These results are fit with a Vinet equation of state, allowing us to parameterize the effect of both pressure and Fe composition. The effect of temperature is modeled using a Grüneisen thermal correction, where the vibrational heat capacities are determined using DFT perturbation calculations and the quasi-harmonic approximation. These pressure-, temperature-, and composition-dependent equations of state are then used to explore the predicted phase relations. The results of this investigation are a complete thermodynamic description of the stable phases for this simplified chemistry and a theoretical prediction for iron partioning in the lower mantle. In particular, we find that there may be a coincidence point (azeotrope) on the Pv-PPv phase loop, across which the sense of Fe-partitioning changes sign, as well as significant immiscibility between Mg-rich and Fe-rich post-perovskite. These findings help explain many of the seemingly incongruent experimental results on the effect of iron and are also useful in the planning of future experiments. [1] Murakami et al. (2004) Science 304, 855-858. [2] Oganov & Ono (2004) Nature 430, 445-448. [3] Caracas & Cohen (2005) GRL 32, L16310. [4] Kobayashi et al. (2005) GRL 32, L19301. [5] Auzende et al. (2007) Goldschmidt Abstract. [6] Caracas and Cohen (2008) PEPI 168, 147.
MR42A-04 INVITED
On the Spatial Extent of the Anticorrelation of Bulk Sound Speed and Shear Velocity in the Lower Mantle
The anticorrelation of bulk sound speed and shear velocity at the base of the mantle is now well-documented and is a feature of models generated using both travel times and mode structure coefficients. What is less well-determined is the spatial extent of this anticorrelation with some models showing an anticorrelation throughout the whole lower mantle while others have the anticorrelation confined to the bottom 500km or so of the mantle. Shear velocity models of the lower mantle are now very robust but bulk sound speed models are generally not so, partially because of their sensitivity to earthquake (mis)location. We demonstrate how different methods of handling earthquake location in the tomographic inversion can lead to quite different results for bulk sound speed with some common techniques leading to erroneous models with strong negative correlations throughout the lower mantle. More robust techniques (including projection methods and simultaneous inversion for location and structure) lead to strong anticorrelations in the bottom 500km of the lower mantle with weak positive correlations above that. Velocity anomalies have low amplitudes in the mid- lower mantle and it turns out that the correlation between bulk sound speed and shear velocity in this region is not robust with the data allowing a positive correlation characteristic of thermal effects. Our ray theory inversions indicate that the anticorrelation of bulk sound speed and shear velocity extends well above the region where post-perovskite could exist. If corroborated by inversions using finite-frequency kernels, we will have to appeal to chemical heterogeneity in at least the bottom 500km of the lower mantle to explain the signal.
MR42A-05
Constraints on the I-W and C-CO Mineral Redox Buffers at Lower Mantle Conditions
The oxidation state of the Earth's lower mantle is an area of great interest in petrology and mineral physics, as it plays a key role in governing mantle mineralogy. It is well known that the amount of available oxygen in a system (oxygen fugacity) can dictate the system's mineralogy, as displayed by the reaction between oxygen and iron to form wustite: Fe + 1/2O2 = FeO (I-W buffer), or the reaction between carbon (graphite or diamond) and oxygen to form a carbonate ion: C + 3/2O2 = CO3 2- (C-CO buffer). However, the redox state of the lower mantle is uncertain, particularly in light of a recently reported crystal- chemically controlled self-redox reaction in iron (Frost et al., 2004) and debates on the oxidation state of carbon in the mantle (Brenker et al., 2007; McCammon et al., 2004). We present results of experiments to constrain the I-W mineral redox buffer relative to the C-CO mineral redox buffer at mantle P-T conditions. Samples prepared according to the balanced reaction FeCO3 + 2Fe = 3FeO + C were loaded in the laser-heated diamond anvil cell and heated to 1700-2200 K at pressures of 23-70 GPa. Bulk phase relationships were determined by x-ray diffraction at the National Synchrotron Light Source. Recovered samples were then analyzed by TEM and EDS coupled with Focused Ion Beam Milling (FIB) to further quantify the Fe, C, and O content of the metal and oxide phases to determine the relative fugacities. At 29 GPa and 2000 K, the IW buffer is approximately 1.2 log units below the C-CO buffer. Effects of pressure, temperature, and the spin crossover in FeO on the relative buffers will be discussed.
MR42A-06
Atomistic Origins of Densification in Oxide Melts in Earth's Mantle: Insights from Synchrotron X-ray Raman Scattering
Whereas the structure of oxide melts at high pressure is essential for understanding chemical evolution of the Earth system in the magma ocean in the mantle, little is known about their liquid structures and the densification mechanism due to the inherent structural disorder and the lack of suitable experimental probes at high pressures. We have recently shown that x-ray Raman scattering combined with diamond anvil cell technique, can yield a new opportunity to study the bonding changes in amorphous systems at high pressure (e.g. Lee SK et al.. Phys. Rev. Lett. 2007, 98, 105502; Lee SK et al. Proc. Nat. Aca. Sci. 2008, 105, 7925). Here, we explore the pressure-induced structural changes in model oxide melt phases in mantle, such as borates, germanates, Na- and Mg- silicates up to 50 GPa using x-ray Raman scattering. The O-Kedge spectra for model oxide melts elucidate the marked difference in densification behavior with varying composition but all the oxide glasses studied here presents the first experimental evidence for the formation of the triply coordinate oxygen above 20 GPa. The increase in the fraction of the triply coordinated oxygen results in a reduced free volume needed to host elements that are more incompatible, leading to an increase in the crystal-melt partitioning coefficient of elements, such as radioactive nuclides thereby significantly affecting the process of the chemical differentiation in the Hadean magma oceans. Its formation can be an efficient densification mechanism in the oxide melt in Earth's mantle and explain the atomistic origin of the high-density Mg-silicate melts at the core-mantle boundary.
MR42A-07
Partitioning of H2O in the mantle transition zone and lower mantle
Water is the most abundant volatile component in the Earth, and the presence of H2O into the mantle minerals affects the elastic and rheological properties. It is well known that wadsleyite and ringwoodite, which are the most abundant minerals in the mantle transition zone, can accommodate significant amount (~3 wt%) of H2O in the crystal structures (e.g. Inoue et al., 1995; Kohlstedt et al, 1996). On the other hand, majorite garnet is the second abundant mineral in the mantle transition zone, and the H2O content was reported to be ~0.2 wt% (Katayama et al., 2003). It is important to know the maximum H2O storage capacity in the minerals, and also to know the partitioning of H2O in the mantle minerals to evaluate the water content in the Earth with temperature and pressure dependence. We have conducted high pressure experiment to determine the partitioning of H2O between wadsleyite, ringwoodite, perovskite and garnet. High-pressure experiments were conducted by MA-8 type (Kawai-type) high-pressure apparatus in Ehime University. We used pyrolite composition which was approximated with respect to five major components, CaO, MgO, FeO, Al2O3 and SiO2. Three starting H2O contents, 2.9, 8.3 and 15.6 wt% were selected. The experimental P-T conditions were 14-23 GPa and 1200-1700°C. The recovered samples were polished and then the chemical compositions were determined by EPMA in Ehime University and the water contents of minerals were measured by SIMS in Hokkaido University. The H2O contents of wadsleyite and ringwoodite decreased with increasing temperature, which is consistent with Ohtani et al. (2000). However the H2O content of majorite did not change so much with increasing temperature. As the results, the partition coefficients between wadsleyite and majorite, and between ringwoodite and majorite decreased with increasing temperature. Further details will be presented.
MR42A-08
Effect of Water on topography of 660 km Seismic Discontinuity: Experimental View
The 660 km discontinuity is one the most important structural boundaries in the Earth's interior. It divides upper and lower mantle and is usually attributed to the dissociation of Mg2SiO4 (spinel) to MgSiO3 (perovskite) and MgO (periclase) (known as post-spinel transition, PST). Determination of a phase boundary of PST is important for developing the model of the Earth's interior. Previous studies of PST in dry system (both simplified and complex systems) have shown that the claypron slope (dP/dT) is very gentle and negative. In this study, we report new data on Mg2SiO4-H2O up to 24 GPa, covering the pressure range of the base of the upper mantle. The experiments were performed in synchrotron facility 'SPring-8' at Hyogo prefecture, Japan, using white X-ray radiation. We used Mg2SiO4 + 2 wt% H2O and Mg2SiO4 as the starting materials. AgPd and graphite capsule were used as a sample containers in hydrous and dry systems, respectively. The generated pressure was calculated from Au equation of state (EOS) proposed by Anderson (1989). Our results show that the PST boundary in Mg2SiO4 + 2 wt% H2O shifts to higher pressure by 1.4 GPa at 1400°C compare to anhydrous Mg2SiO4 system in the same study. The new data are used to model the topography of 660 km seismic discontinuity and the subducting slab to the deeper part of the mantle and provide tight constraints on its physical and chemical state.