MR43A-1795
Melting Behavior and Chemical Properties of the Iron-Carbon System
One of the most challenging experiments related to the laser heating technique in the diamond anvil cell (DAC) is an unambiguous x-ray based detection of melting by recording high quality diffuse x-ray scattering from molten materials at high pressure. Employing a newly developed, advanced, flat top laser heating system at GSECARS, we were able to perform on-line melting experiments at pressures up to 60 GPa. The capability to maintain the molten sample in the DAC for a relatively long time (at least 60 s) allowed us to collect high quality x-ray scattering data suitable for structure analysis even from low-Z molten materials, such as Si, Ge, Fe, Fe3C, Fe7C3 etc. In this work, we focused on the melting behavior and chemical properties of the iron-carbon system at pressures up to ~170 GPa studied with on-line micro x-ray diffraction in a double sided laser heated DAC at GSECARS (Sector 13, APS). Iron carbides (Fe3C, Fe7C3) were synthesized in- situ in the DAC from various mixtures of Fe and C powders with different atomic ratios. We have found that the chemical reaction between iron and carbon takes place independent of the structure of starting phases of iron (fcc or hcp) and carbon (graphite or diamond). The reaction temperature increased gradually from ~1000 K to ~1700 K as pressure increased from 6 GPa to 155 GPa. The melting temperature of iron carbide was found to be systematically lower than for iron by ~300-400 degrees in the pressure range 20-60 GPa. The experimentally measured structure factor and related pair distribution function of iron carbide melt were analyzed and compared with pure iron data at related pressures. High- pressure, high-temperature stability of iron carbide phases at the relevant Earth's mantle-core conditions and physical/chemical properties of iron-carbon melts provide important constraints on models of the formation of D" layer and interactions at the core mantle boundary. Implications of these results for the composition and structure of the Earth's interior will be discussed.
MR43A-1796
Density and structure of basaltic magma under high pressure and high temperature
The density of magma is one of the important properties for discussing evolution of magma ocean at the early history of the planets and magmatic activity in the planetary mantle. We have measured the density of basaltic melt at high temperature and high pressure by X-ray absorption method. The experiments were carried out using a DIA-type cubic press at BL22XU of the SPring-8. X-ray absorption method is accurate method for the density measurement under desired pressure and temperature compared to the other methods. This method for density measurements was originally developed by Katayama et al. (1993). It is based on the Lambert-Beerfs law. The sample was placed in a diamond capsule to calibrate the sample thickness and the X-ray absorption profile of the sample was measured by ion chambers. We succeeded in measuring the density of basaltic melt up to 4.6 GPa and up to 2000 K. We obtained the compression curve of basaltic melt by using the Birch-Murnaghan equation of state with a negative pressure derivative of bulk modulus (dK/dP). A negative dK/dP might be caused by the structural change of the silicate melts, although it is unlikely in crystals. The structure of magma is based on continuous three-dimensional networks of corner-sharing SiO4 and AlO4 tetrahedra, as being derived from a network of tetrahedrally coordinated Si and Al atoms each linked to four others through a shared O atom. The principal mechanisms of compression for silicate melts involve continuous changes in T-O-T bond angles and bond lengths. Silicate melts might undergo continuous and gradual changes in topology and cation coordinations. In order to further understand these changes and how they are affected by the microscopic structure, we have conducted the energy-dispersive X-ray diffraction to determine the structure of the basaltic melt up to 5 GPa. High pressure and high temperature X-ray diffraction experiments on basaltic melts were carried out by the energy dispersive method using SPEED Mk-II apparatus installed at the BL04B1 of SPring-8. Diffracted X- rays were collected at 2theta of 4, 5, 7, 9, 11, 15, 18 and 22 using a solid state detector system. We obtained the structure factor, S(Q), from the raw X-ray diffraction data using an analytical program developed by Funakoshi (1997). The first peak of the S(Q) reflects the -T-O-T- intermediate-range silicate network. In this study, the first peak shifts to higher Q with increasing pressure, while the second peak does not change. This result indicates that the intermediate-range network in basaltic melt changes to denser structure (e.g., squeezing the open space and shrinking the network rings) without shortening the intra- tetrahedral atomic distance with increasing pressure.
MR43A-1797
Chemistry of Low Degree Hydrous Partial Melt at High Pressures
Geophysical evidence hints towards the possibility that silicate melt may reside on top of the 410 km seismic discontinuity in some regions of the mantle. Based on the difference in the water storage capacity of upper mantle (dominantly olivine) and transition zone minerals (primarily wadsleyite), it has been suggested that if there is any melting at the top of the transition zone, it is likely to be hydrous melt. This hydrous melt phase might be able to reconcile the differences between geochemists view of stratified mantle (based on trace element differences in MORB and OIB) as opposed to geophysicsts view of whole mantle convection (based on seismic tomography of slabs). Fundamental questions remain unanswered: what is the composition of low degree hydrous partial melts at these conditions? Is this melt gravitationaly stable? Could this melt be detectable geophysically? In this study, we are determining the compositions of low degree hydrous partial melts that would be in equilibrium with the mineral assemblage at the base of the upper mantle. Initial experiments employ an estimated melt composition in the system CaO-MgO-Al2O3-SiO2- H2O which was sandwiched between layers of peridotite (CAMS). By examining the shift in the composition of melt and mineral phases as the melt tries to equilibrate with the solid phases, the melt composition can be iteratively modified until the melt composition remains unchanged. Experiments are performed at 14 GPa and 1723 K i.e, conditions pertinent to the top of the transition zone. The effect of iron on the melting relations is also being explored.
MR43A-1798
Detailed structure of the carbonated peridotite solidus ledge in the system CaO-MgO- Al2O3-SiO2-CO2
The presence of carbonatitic melts in the sub-oceanic mantle has been inferred on the basis of geochemistry of ocean-island basalts and xenoliths, CO2-vesicles in voluminous mid-ocean ridge basalts, geophysical observations, and experimental petrology. The carbonate ledge somewhere between 2.0-3.0 GPa is a prominent feature of the solidus of carbonated peridotite. The ledge is formed by a precipitous temperature drop of the mantle solidus as the solubility of CO2 in the melt rapidly increases with increasing pressure, leading to a negative Clapeyron slope of the mantle solidus over some pressure interval. It is terminated by the appearance of crystalline carbonate, on the high-pressure side, at the expense of CO2 vapor that exists on the low pressure side. The solidus temperature of carbonated peridotite decreases by roughly 200 C at the ledge. The carbonate ledge was first noted at the solidus of carbonated peridotite in the system CaO- MgO-SiO2-CO2 [CMS-CO2], and it persists in the more complex model system CaO-MgO-Al2O3-SiO2-CO2 [CMAS-CO2]. In the system CMAS-CO2, the ledge appears to be bounded by two invariant points consisting of spinel-garnet peridotite+CO2+melt, on the low pressure side, and garnet peridotite+CO2+dolomite+melt, on the high pressure side. While the lower pressure invariant point is created by the intersection of the spinel-garnet (sp-gt) peridotite subsolidus boundary curve with the solidus of vapor-bearing peridotite, the higher pressure invariant point occurs where the CO2-bearing phase coexisting with the garnet peridotite phase assemblage changes from vapor to dolomite. The precise position and shape of the ledge and the position of the two invariant points is very significant because the temperatures at which the mantle melts and the melt compositions are dependent on the behavior of CO2, the depth interval in the Earth where carbon exists as CO2, and the onset of transformation of CO2 to crystalline carbonate in the Earth. Due to these important implications, we are conducting high-pressure experiments to determine the melting relations at the solidus of CO2-bearing spinel-garnet peridotite in the system CMAS-CO2 at 2.1 to 3.0 GPa. We have found that the higher pressure invariant point, involving fo+opx+cpx+gt+CO2+dolomite+melt phase assemblage, occurs at 3 GPa/1225 C. The melt is calciocarbonatitic with roughly 40-42 wt.% CO2. Going along the solidus toward lower pressure, we have determined the following isobaric invariant points: 2.8 GPa/1275 C, 2.6 GPa/1375 C, 2.4 GPa/1400 C, 2.3 GPa/1425 C, and 2.1 GPa/1425 C. The first four points involve fo+opx+cpx+gt+CO2+melt, whereas the last one involves the fo+opx+cpx+sp+CO2+melt phase assemblage. Melt compositions at these isobaric invariant points resemble calciocarbonatites. These five points span a univariant curve in P-T space, lack crystalline carbonate, and hence define the famous carbonate ledge in the CMAS-CO2 system. Significantly, since two different isobaric invariant assemblages occur at 2.3 and 2.1 GPa, the lower pressure invariant point must occur at about 2.2 GPa/1425 C, consisting of fo+opx+cpx+sp+gt+CO2+melt. Experiments are currently under way to understand the melting relations where CO2 vapor coexists with the spinel lherzolite phase assemblage.
MR43A-1799
Water-rich carbonatites at low pressures and kimberlites at high pressures
Trace-element chemistry of carbonatites found on the Earth's surface indicates their general derivation from the mantle. However, while most documented carbonatites (both from the surface and in diamonds from the continental mantle) have calcic compositions, melting experiments on anhydrous, carbonated mantle peridotite produce magnesiocarbonatitic melts. Moderately calcic melts are produced in dry experiments, but even these melt compositions are fairly far removed from resembling calciocarbonatites. Explanations for this discrepancy include derivation of calciocarbonatites from more Mg-rich carbonatite melts by crystal fractionation or wall-rock reactions in the mantle. While these two mechanisms might well work, at least in some cases, the emplacement of carbonatites seems to occur at temperatures considerably less than 1000 C. This low temperature compared to the melting temperature of pure carbonates can, in part, be explained by the presence of alkalis and halogens in natural carbonatites. However, there is also ample evidence for the presence of significant amounts of H2O, and melting experiments indicate its high solubility in molten sodium carbonate melts. In addition, in other model systems, most notably MgO-SiO2-H2O (MSH), supercritical behavior (no difference between melt and fluid) has been demonstrated to occur at pressures somewhere around 11-13 GPa. Owing to different physical and chemical properties of melts and fluids, there is considerable petrological interest toward understanding supercritical behavior in the Earth. Here we present melting relations for hydrous, carbonated garnet lherzolite in the system CaO-MgO-Al2O3-SiO2- CO2-H2O (CMAS-CO2-H2O) from 2.5 to 7 GPa. With fo+opx+cpx+gt+carbonate+melt+fluid present in this system, the solidus is isobarically invariant, and is a curve in P-T space. Major findings can be summarized as follows: (1) at 2.5-3 GPa, the solidus temperature of hydrous, carbonated garnet lherzolite is lower by 225-250 C than for water-free carbonate-bearing garnet lherzolite; (2) at these pressures, there is roughly 17-20 wt percent dissolved water in the melts, suggesting that carbonatitic melts can incorporate large amounts of water; (3) from 2.5 to roughly 3.5 GPa, melts coexisting with fo+opx+cpx+gt+dolomite+fluid are highly calcic and partly overlap calciocarbonatites found in nature; (4) a P-T invariant point occurs at 3.7 GPa/1125 C, at which fo+opx+cpx+gt+dolomite+magnesite+melt+fluid coexist, marking the beginning of the stability of magnesite at the hydrous, carbonated peridotite solidus; (5) with increasing pressure starting at this invariant point, the fluid-saturated solidus becomes considerably closer to the water-free, carbonated solidus in the model system CMAS-CO2. For instance, at 7 GPa, it lies only 125 C lower than that of water-free carbonated peridotite. At 6 and 7 GPa, the melt coexisting with the fo+opx+cpx+gt+magnesite+fluid phase assemblage, contains about 5-7 wt percent water, and is more akin to kimberlite (all in wt percent: 20-25 SiO2, 30-32 MgO, 19-20 CaO, 2-3 Al2O3) than carbonatite. At this stage it is not entirely clear what changes in the phase relations cause melts to attain this character.
MR43A-1800
HIGH PRESSURE TOMOGRAPHY IN STUDIES OF CORE FORMATION MECHANISMS
Permeability of Fe-rich melts in the silicate Earth plays a critical role in core formation. In past experimental studies, this problem was approached by examining microstructure of silicate samples containing Fe-rich melts recovered from high pressure and high temperature deformation experiments. Recently, x-ray computed microtomography (CMT) was successfully used to generate three-dimensional (3D) images on samples recovered to ambient conditions [Roberts et al, 2007, GRL 34]. Based on the high resolution 3D images, it is now possible to predict quantitatively melt permeability, which can then be applied to models of melt segregation to understand coalescing processes of planetesimals. With the development of high- pressure x-ray tomography microscope (HPXTM) [Wang et al., 2005, RSI 76], the time is ripe for in-situ tomography at high pressure (P) and high temperature (T), with controlled deformation. Here we report recent progress in this development. Powdered San Carlos olivine was sintered with 4 and 12 wt percent FeS at 1 GPa and 1620 K for 24 h, in a graphite capsule, yielding an equilibrium texture. The samples were then loaded in a Drickamer anvil cell in the HPXTM and compressed to 6 GPa, heated to between 700 and 800 K. The samples were deformed by rotating the top and bottom Drickamer anvils in opposite directions. 3D tomography data were collected prior to deformation and periodically at various P, T and shear strain conditions. At high pressure olivine was deformed in the ductile regime, and a clear microstructural evolution was observed. Detailed data analysis will be presented and the feasibility of studying core-formation mechanisms using the HPXTM discussed. Work supported by NSF EAR-0711057. Part of the work was performed under the auspice of the U.S. department of Energy by Lawrence Livermore National Laboratory under Contract DE-AC52-07NA27344.
MR43A-1801
Constraint of sulfur influence on bulk modulus of liquid Fe
Sulfur has been widely considered as a candidate of light element to account for the density deficit in the Earth's core relative to pure liquid iron because it is depleted in the Earth's mantle, easily alloys with iron, and moderately siderophile during iron-silicate interaction at high pressures. Knowledge about density change caused by alloying S into Fe is therefore critical for estimating the required partitioning and understanding the reality of existence of sulfur in the Earth's core. Currently available data of sulfur influence on bulk modulus of liquid iron are very limited, and due to technical challenges, the pioneer experiments have produced quite devise results indicating that sulfur content can reduce the bulk modulus by 1.7GPa/1wt%S to 3.5GPa/1wt%S. This will result in a 5wt%S uncertainty in estimating the core composition. Using in situ synchrotron x-ray radiograph technique, we measured equation of state of liquid Fe-36wt%S system in a multi-anvil press up to 6 GPa. Fitting the experimental data to a second order Birch- Murnaghan equation of state with a fix K'= 4 produces a bulk modulus of 15.4 GPa which corresponds to an influence of sulfur content of 1.8 GPa/1wt%S on the bulk modulus of the system. This value is consistent with the lower bound of previously defined range of sulfur influence on the bulk modulus of liquid iron. Impact of this result on the core composition modeling will be discussed.
MR43A-1802
A Thermodynamic Theory for Dense Silicate Liquids That Includes Explicit Provision for Variation in Composition and Fluid Structure, Derived From the Rosenfeld-Tarazona Potential Energy-Temperature Scaling Law
Rosenfeld and Tarazona (1998, Molecular Physics 95:141) derive an expression from a fundamental- measure energy functional for hard spheres and thermodynamic perturbation theory for the scaling of the potential energy (U) on temperature (T) in a dense classical liquid: U = a + bT3/5, where a and b are functions of volume (V). From this expression we have formulated a general theory of the thermodynamic properties of dense liquids, and have applied this theory successfully to both polymerized and depolymerized silicate melts over the temperature range 2000-6000 K and at pressures (P) up to 130 GPa, using data from molecular dynamics (MD) simulations of molten MgSiO3, Mg2SiO4, CaMgSi2O6, and CaAl2Si2O8. In this work we extend the thermodynamic treatment to account for the compositional variability of multicomponent liquids and to deal with the mixing of melt species characterized by both composition and distinct cation-oxygen coordination number (CO-CN). We posit that the relation V = Σ Vi Xi holds, where Xi is the mole fraction of an endmember component with a particular CO-CN and Vi is the volume of that species; note that Vi is only a function of T and P, and that the posited relation embodies the concept of ideal volume-mixing of melt species. In combination with Rosenfeld-Tarazona theory, a self-consistent formalism for the solution thermodynamics is obtained, and in particular the equilibrium distribution of species of fixed stoichiometry but variable CO-CN is determined from the condition of homogeneous equilibrium. The resulting thermodynamic model must be calibrated from data documenting both the variation in macroscopic properties (U, V) as well as structural characteristics (CO-CN) of liquids as a function of T and P. We demonstrate application of the theory to a set of MD simulation data on MgSiO3 liquid. Unique measures of structural variation are explored for this and other MD data sets to ascertain an optimal set of endmember species that should be adopted to formulate a comprehensive model for multicomponent systems.
MR43A-1803
Geophysical Evidence for Magmatic Transport in the Lower Crust in Iceland
An event characterized by deep seated earthquake activity, started near Upptyppingar, N-Iceland, in early 2007 and has been ongoing episodically since. The area is part of the Kverkfjöll Volcanic System, lying in the fissure swarm extending northeast. Iceland GeoSurvey has carried out a resistivity survey, measuring two parallel, approximately 20 km long E-W profiles with 5 km apart, centred immediately north of Upptyppingar area. Using both TEM and MT method, the resistivity structure of the area was modelled to 30 km depth below the profile. Two pronounced horizontal low resistivity layers are observed; the upper at 1-2 km depth and the lower at 7-13 km depth. Beneath the deeper layer a low resistivity column is observed right under the Upptyppingar area, extending as deep as can be detected. The upper low resistivity layer reflects clay rich rock caused by thermal alteration. Its lower boundary corresponds to the change in dominant alteration minerals at temperatures above 230°C. The deeper low resistivity layer appears to be the same as present beneath most of Iceland according to the results of previous MT-surveys. We interpret this layer to mark the boundary between the brittle and the ductile crust for normal strain rates, supposedly close to the 700°C isotherm. Contrary to most earthquakes in Iceland, most of the earthquake hypocenters beneath Upptyppingar are below the low resistivity layer, i.e. at 10 to 20 km depth beneath the surface, moving upwards with time. They occur within the deep low resistivity column and line up at the lower boundary of the deeper conductor. Very few earthquakes occur within the low resistivity layer itself, but some activity is above it. The low resistivity column probably represents hotter rocks than the surroundings, due to recent magma intrusions. The composition of the crust below the low resistivity layer must then be different and mechanically stronger than above. Resistivity data along with the earthquake distribution suggest magma upwelling from the mantle to the lower part of the crust, causing the deep earthquakes due to high strain rates. Near the lower boundary of the deep conductor the rock becomes more ductile and the high density magma accumulates as intrusions.
MR43A-1804
Viscoelasticity of partially molten rocks at mantle P-T
The mechanical properties of a partially molten rock are crucial for interpreting the attenuation and velocities of seismic waves. While partial melting is often associated with low seismic shear wave velocity zone (LVZ) (Anderson and Sammis, 1970); studies have also suggested that partial melting may not be required to interpret the large velocity variation in the upper mantle (Faul and Jackson, 2005; Karato and Jung, 1998; Stixrude and Lithgow-Bertelloni, 2005). Extensive studies (Faul et al., 2004; Gribb and Cooper, 2000; Jackson et al., 2004) have been carried out on the attenuation behavior of partially-molten samples using melt-bearing olivine sample but at low-pressure (less than 300 MPa). Our recent developments (Li and Weidner, 2007) have allowed us to study energy dissipation in materials at high P-T (up to 15 GPa-2000K) and seismic frequencies. We have performed some experiments to characterize attenuation and dispersion in a model mantle composition (Kilborne Hole peridotite nodule, KLB- 1) at 1-10 GPa pressures. We use the multi-anvil high pressure device D-DIA (Durham et al., 2002; Wang et al., 2003) coupled with synchrotron X-ray radiation (Weidner et al., 2005). Mechanical testing is done by applying a uniaxial forced oscillation on a sample typically of 1 mm length and 1-1.5 mm diameter. The preliminary results raised many interesting issues. The detected Q and Young's modulus were found to be dependent on the pre-annealing conditions of the sample which may be a result of unrelaxed differential stresses. The geometry of the melted phases highly depends on the magnitude of the stress and melt- fractions. Microscopic analysis of the recovered sample will also be discussed.
MR43A-1805
High-temperature ultrasonic measurement in multi-anvil high-pressure apparatus under the melting temperatures of mantle materials
Measurement of the elastic wave velocities of partially or completely molten mantle materials is important to understand the nature and dynamics of the Earthfs upper mantle, such as those relevant to subduction volcanism, lithosphere-asthenosphere boundary and/or possible melt layer above the 410 km discontinuity. Some attempts to measure elastic wave velocities of silicate melts have been made at high temperatures and atmospheric pressure (e.g. Ai and Lange, 2008), but the corresponding measurement under the upper mantle pressure conditions has not yet reported to date. We therefore need to develop the experimental technique for elastic wave velocity measurement at high-pressures and at temperatures greater than the melting temperatures of mantle materials (higher than 1400 °C) to address the above issues. We recently expanded both pressure and temperature conditions of ultrasonic measurement in multi-anvil apparatus to 1400 °C and around 19 GPa (Higo et al., 2008; Irifune et al., 2008), and here we report some preliminary results of the measurement at further extended temperatures to 1700 °C at high pressure. Graphite or LaCrO3 were used as the heater instead of platinum used in our previous studies, and zirconia was used as thermal insulator. We also adopted highly-sintered zirconia as the buffer rod for transmitting ultrasonic waves to the sample. Some heating tests at ~8-12 GPa demonstrated that temperatures to 1800 and 1900 °C were achieved with graphite and LaCrO3 heaters, respectively. Ultrasonic measurement was carried out for a molybdenum rod at around 8 GPa using graphite heater. We successfully obtained clear elastic wave signals reflected at both ends of the molybdenum sample up to 1700 °C, which is higher than the melting temperatures of most hydrous silicates (e.g. hydrous Mg-silicates and/or hydrous MORB) and of some anhydrous silicate minerals (e.g. albite). The present experimental setup enables us to determine elastic wave velocities of mantle materials at these temperatures. Further technical developments are currently being pursued toward the ultrasonic measurements at higher temperatures in multi-anvil apparatus.
MR43A-1806
EXPERIMENTAL STUDIES ON SEISMIC ANISOTROPY ENHANCED BY MELT
It is commonly agreed that the seismic anisotropy, most likely caused by aligned minerals, is a very important indicator of intracrustal deformation. Ultrasonic velocity measurements on the schists, gneisses, migmatites, amphibolites and mylonites from Higher Himalayan Crystallines (HHC) and Honghe strike-slip fault zone in the south-western China show that the average anisotropic magnitude of them is about 4 % (the maximal magnitude of them is less than 8 %). However, a series of measurements of surface-wave dispersion inversions and waveform inversions of P to S conversions produced at seismic impedance contrasts conducted in Tibet indicate marked anisotropy with the magnitude ranging from 4 - 18% within the middle to lower crust. What causes the abnormal large anisotropy within Tibetan crust? The peak metamorphism temperatures and pressures of HHC reached 800 -850°C and varied from 0.8 GPa to 1.2 GPa, and part of HHC rocks to some extent had undergone granulitic-grade metamorphism which temperature and pressure were in excess their solidus. Localized melt bands and melt pocket preferred orientation (MPO) are present in HHC pervasively. If melt is oriented, the melt is expected to be an extremely important factor to induce anisotropy. Experimental results indicated that MPO can result in an extra anisotropy, whose magnitude can be increased by a factor of 2 for a melt fraction of 5 - 10 vol% with an aspect ratio of 0.1 - 0.5. The contribution of MPO on the anisotropy is likely comparable to or larger than that induced by lattice preferred orientation of major minerals, possibly amphiboles and micas, in middle to lower crust. Therefore, we attribute the abnormal anisotropy observed within Tibetan middle to lower crust to the aligned melt.
MR43A-1807
Effect of Water and Mg Number on the Density of Ultramafic Silicate Melts in the Deep Upper Mantle
Density of silicate melts plays a very important role in controlling the chemical differentiation of the Earth, since the density contrast between the melts and the surrounding solids determines the direction of material circulations in the Earth. Melts formed in the deep upper mantle likely contain some water, which may have a relatively large effect on melt density compared to other chemical components in the melt as demonstrated by previous studies (Matsukage et al., 2005; Sakamaki et al., 2006; Agee, 2008). However, experimental data are still very sparse to fully constrain the effect of water on melt density. In addition to water, our detailed analysis of the experimental data using the third-order Birch-Murnaghan equation of state shows that the Mg # of the melt is another very important parameter that controls melt density. In this study, sink/float experiments are performed using a Kawai-type multianvil apparatus to determine the density of hydrous ultramafic silicate melts with different water content and Mg # at high pressures. In the sink/float experiments, density markers that are made of single-crystal diamonds are loaded into sample capsules. The sample is then completely melted under high-pressure and high-temperature conditions. After quenching the density of the melt can be bracketed by the sinking or floating of the density markers in each experiment. Previously, Matsukage et al. (2005) studied four melt compositions (s3a, s7a, s5a, and s6a) with 5 wt% water but different Mg #. The strategy of the current study is to vary water content in these compositions to be able to calculate the partial molar volume of water at high pressure directly by comparing the densities of melts with different water contents. Therefore we determine the density of three hydrous melt compositions, s63, s73 with 3 wt% water and s67 with 7 wt% water by adding or subtracting water from s6a and s7a in Matsukage et al. (2005). Two dry compositions s6 and s7 which are the dry part of s6a and s7a in Matsukage et al. (2005) are also studied. Experimental conditions are from 8 to14 GPa and from 2173 to 2373 K. The densities of these melts are compared to constrain the partial molar volume of the water as well as the effect of Mg #. Our preliminary results are consistent with previous studies (Matsukage et al., 2005; Sakamaki et al., 2006; Agee, 2008). Given these results, we will report the conditions under which the density crossover between hydrous melts and the surrounding solids at 410 km could occur.
MR43A-1808
Experimental Constraints on the Partitioning and Valence of V and Cr in Garnet and Coexisting Glass.
A series of experiments with garnet and coexisting melt have been carried out across a range of oxygen fugacities (near hematite-magnetite (HM) to below the iron-wustite (IW) buffers) at 1.7 GPa to study the partitioning and valence of Cr and V in both phases. Experiments were carried out in a non-end loaded piston cylinder apparatus, and the run products were analyzed with electron microprobe and xray absorption near edge structure (XANES) analysis at beamline 13-ID at the Advanced Photon Source of Argonne National Lab. The valences of vanadium and chromium were determined using the intensity of the K pre-edge peaks, calibrated on a series of Cr and V-bearing standard glasses. This technique has been applied to V and Cr in glasses and V in spinels previously, and in these isotropic phases there are no orientation effects on the XANES spectra (Righter et al., 2006, Amer. Mineral. 91, 1643-1656). We also now demonstrate this to be true for V and Cr in garnet. Also, previous work has shown that V has a higher valence in the glass (or melt) than in the coexisting spinel. This was also true for V in garnet-glass pairs in this study. Vanadium valence in garnets varied from 2.7 below the IW buffer to 3.7 near HM, and for coexisting glass it varied from 3.2 to 4.3. Vanadium valence measured in some natural garnets from mantle localities indicated V is slightly more reduced than 3. In contrast, Cr valence measured in garnet and coexisting glass for all experimental and natural samples was 2.9-3.0, suggesting that the valence of Cr does not vary within either phase across a large fO2 range. These results demonstrate that while V varies from slightly more reduced than 3+ to slightly more oxidized than 4+ in garnet-melt systems, Cr does not, and this will ultimately affect the partitioning behavior of these two elements in natural systems. Garnet/melt D(Cr) are between 12 and 17 across this range of fO2, whereas D(V) has the highest partition coefficient ~3, near the IW buffer where the valence of V is almost entirely 3+.
MR43A-1809
Pair distribution function study on compression of liquid gallium
Integrating a hydrothermal diamond anvil cell (HDAC) and focused high energy x-ray beam from the superconductor wiggler X17 beamline at the National Synchrotron Light Source (NSLS) at the Brookhaven National Laboratory (BNL), we have successfully collected high quality total x-ray scattering data of liquid gallium. The experiments were conducted at a pressure range from 0.1GPa up to 2GPa at ambient temperature. For the first time, pair distribution functions (PDF) for liquid gallium at high pressure were derived up to 10 Å. Liquid gallium structure has been studied by x-ray absorption (Di Cicco & Filipponi, 1993; Wei et al., 2000; Comez et al., 2001), x-ray diffraction studies (Waseda & Suzuki, 1972), and molecular dynamics simulation (Tsay, 1993; Hui et al., 2002). These previous reports have focused on the 1st nearest neighbor structure, which tells us little about the atomic arrangement outside the first shell in non- crystalline materials. This study focuses on the structure of liquid gallium and the atomic structure change due to compression. The PDF results show that the observed atomic distance of the first nearest neighbor at 2.78 Å (first G(r) peak and its shoulder at the higher Q position) is consistent with previous studies by x-ray absorption (2.76 Å, Comez et al., 2001). We have also observed that the first nearest neighbor peak position did not change with pressure increasing, while the farther peaks positions in the intermediate distance range decreased with pressure increasing. This leads to a conclusion of the possible existence of !!locally rigid units!!L in the liquid. With the addition of reverse Monte Carlo modeling, we have observed that the coordination number in the local rigit unit increases with pressure. The bulk modulus of liquid gallium derived from the volume compression curve at ambient temperature (300K) is 12.1(6) GPa.
MR43A-1810
Non-bridging Oxygens in Calcium Aluminosilicate Glass From Per-calcic to Peraluminous Compositions
The role of non-bridging oxygen (NBO) and its effects on the thermodynamic and transport properties of aluminosilicate melts are not fully understood, although this species clearly must have a major influence on configurational entropy, viscosity, etc. Its existence along metaluminous joins in alkali- and alkaline-earth aluminosilicates was first postulated from viscosity measurements (Toplis et al., 1996, 2004) and then directly observed in several metaluminous calcium aluminosilicates by 17O nuclear magnetic resonance (NMR) spectroscopy. Much of the recent work has concentrated on glasses with an M+n/(M+nAl) ratio greater than or equal to 0.5 (metaluminous to peralkaline or per-alkaline earth); however, the observed viscosity maxima in several ternary systems occur when this ratio is less than 0.5 (peraluminous). Using NMR spectroscopy, this study investigates the effects of the Ca/Al ratio on the amount of NBO present in calcium aluminosilicate (CAS) glasses. 17O MAS NMR spectra of glasses with 60 mol% SiO2 show a decrease in NBO as the ratio R=Ca+2/(Ca+2Al) decreases, from 6.9% at R=0.56 to 1.0% at R=0.44. Measurable amounts of NBO thus persist well into the peraluminous region of the CAS system, but the species becomes undetectable (<0.5%) when R reaches 0.38 and 0.33. 27Al MAS NMR spectra of these glasses show an increase in the amount of five-coordinated aluminum as compositions become more peraluminous, as is well-known from previous studies (Neuville et al. 2006). Comparison with published viscosity measurements measured at both higher and lower mol % SiO2 (Toplis et al. 2004) suggests that the viscosity maximum does not correspond exactly with the disappearance of NBO from the glasses, but effects of temperature on speciation will need to be taken into account to accurately link glass structure with melt properties: recent work has shown, for example, that NBO content increases with temperature in CaAl2Si2O8 melt (Stebbins et al. 2008).
MR43A-1811
Temperature Effects on Aluminoborosilicate Glass and Melt Structure
Quantitative determination of the atomic-scale structure of multi-component oxide melts, and the effects of temperature on them, is a complex problem. Ca- and Na- aluminoborosilicates are especially interesting, not only because of their major role in widespread technical applications (flat-panel computer displays, fiber composites, etc.), but because the coordination environments of two of their main network cations (Al3+ and B3+) change markedly with composition and temperature is ways that may in part be analogous to processes in silicate melts at high pressures in the Earth. Here we examine a series of such glasses with different cooling rates, chosen to evaluate the role modifier cation field strength (Ca2+ vs. Na+) and of non-bridging oxygen (NBO) content. To explore the effects of fictive temperature, fast quenched and annealed samples were compared. We have used B-11 and Al-27 MAS NMR to measure the different B and Al coordinations and calculated the contents of non-bridging oxygens (NBO). Lower cooling rates increase the fraction of [4]B species in all compositions. The conversion of [3]B to [4]B is also expected to convert NBO to bridging oxygens, which should affect thermodynamic properties such as configurational entropy and configurational heat capacity. For four compositions with widely varying compositions and initial NBO contents, analysis of the speciation changes with the same, simple reaction [3]B = [4]B + NBO yields similar enthalpy values of 25±7 kJ/mol. B-11 triple quantum MAS NMR allows as well the proportions of [3]B boroxol ring and non-ring sites to be determined, and reveals more [3]B boroxol ring structures present in annealed (lower temperature) glasses. In situ, high-temperature MAS NMR spectra have been collected on one of the Na-aluminoborosilicate and on a sodium borate glass at 14.1 T. The exchange of boron between the 3- and 4-coordinated sites is clearly observed well above the glass transition temperatures, confirming the importance of such local structural dynamics in controlling the bulk viscosity.
MR43A-1812
Strain Localization and Melt Segregation in Partially Molten Rocks During Torsion Experiments
We will describe the mechanism of strain localization coupled with the conditions for generation and segregation of melt in partially molten synthetic metapelites. Torsion experiments were performed at 750° C, 300 MPa confining pressure, constant shear strain rate of 3x10-4 s-1 for shear strain ranging from 0.5 to 5.0 on synthetic very fine grained aggregates of muscovite and quartz (respectively 0.3 and 0.7 by volume). Static experiments were also conducted on the same materials under same P and T conditions for the same duration of the dynamic runs (1, 2 and 5 hours). The starting material, which was fabricated by first uni-axial pressing of the powdered material at 200 MPa at room temperature and then hot isostatic pressing at 160 MPa at 580° C for 6 hours, had a porosity of about 17-20% and a strong foliation defined by muscovite shape orientation. The torsion experiments were conducted on cylindrical samples of 10mm diameter with the foliation either parallel or normal to the cylinder axis. The stress vs. strain relation of the deformed material showed initially a steep hardening stage (peak stress ~ 120 MPa), after the yield (~ 70 MPa) and then a sharp linear strain weakening followed by steady state flow at significantly low shear stress (~ 15 MPa). In the samples with foliation parallel to the cylinder axis, at low shear strains (γ= 0.5 - 1.0) the foliations rotated about 25° towards the direction of shear and long muscovite grains were boudinaged across the extension direction. At γ ~ 1.5, the shear localization was first observed in brittle-ductile mode at very low angles (20-22°) to the bulk shear plane. With progressive shearing, the fractures dilated across their width and developed a discreet damage zone. Melting of quartz and muscovite is obvious only after γ ~ 2, where melt begin to collect in the dilated damage zones. Comparison between the static and dynamic experiments revealed that melting is greatly enhanced by dynamic experiments. When extrapolated to nature, deformation enhanced melting might explain rapid segregation and melt percolation to form large S-type plutons such those exposed in the Himalaya.