MR33C-01 INVITED
Experimental Compressibility of CO2 in Silicate Melts and the Effect on Planetary Differentiation
High pressure experiments using the sink/float method have bracketed the density of carbonated partial melt of peridotite and carbonated Apollo 14 black glass melt at high pressures and temperatures. The experiments were designed to determine the compressibility of CO2 in silicate melts and allow prediction of crystal-liquid density crossovers in CO2-bearing planetary magmas. The silicate melt compositions were synthetic mixtures of reagent oxides with CO2 added in the form of CaCO3. Samples were contained in compression-sealed molybdenum capsules. Sink/float marker spheres implemented were gem quality synthetic forsterite (Fo100) and San Carlos olivine (Fo91). Experimental run times were 30 seconds, thus minimizing sphere-liquid reactions and liquid reaction with capsule and pressure media. CO2 (total) in the quench melt run products was estimated by electron microprobe analyses of carbon and oxygen. All experiments were carried out in a Walker multi-anvil apparatus or a Quick Press piston-cylinder device at the Institute of Meteoritics, University of New Mexico. The densities of peridotite partial melt with 5 wt % CO2 and the same peridotite partial melt with no CO2 were determined at 4.3 GPa and ~1825 C. Using the density difference between the carbonated and non-carbonated melts we calculate a partial molar volume of CO2 in peridotite partial melt of approximately 18 cm3mol-1 at these conditions. This represents a 35% decrease in VCO2 compared to estimates for VCO2 at 1-bar (Liu and Lange, 2003), indicating a high compressibility for CO2 in silicate melt over the range 0-4 GPa. Our value of VCO2 is similar to that of estimated for basalt at 19.5 GPa (Ghosh et al., 2007), suggesting that the compressibility of CO2 in silicate melt decreases significantly with pressure, although more experiments are needed to confirm this possibility. Our experiments on Apollo 14 black glass are the first of their kind to simultaneously determine melt density and CO2 solubility in silicate melt at high pressure. This study is also the first to examine the effect of high titanium concentrations on the properties of carbonated silicate melts. We bracketed the density of this silicate melt + 5 wt % CO2 at 1 GPa and 1315 C. and observed the presence of an exsolved fluid phase in the quenched run products, in accord with CO2 super-saturation. At higher pressure and temperature (4.7 GPa and 1600 C, equivalent to the pressure at the center of the Moon) no fluid exsolution was observed in the run product indicating that 5 wt % CO2 was dissolved in the black glass melt at these conditions. Apollo 14 black glass represents the densest known magma in the solar system. Presence of a low density volatile propellant in its mantle source region is required to account for it as a presumptive lunar fire fountain eruptive. Our initial results are consistent with a dissolved CO2 propellant at high pressure in black glass magma that buoyantly drives it to the lunar surface and degasses efficiently by decompression.
MR33C-02 INVITED
The effect of water on the partial melting of peridotite at 3 GPa.
We have investigated the influence of water on partial melting of fertile garnet peridotite by performing experiments with a hydrated synthetic peridotite at 3 GPa. The starting material consists of a synthetic KLB-1 peridotite analog mixed in varying ratios with a synthetic hydrous olivine composition (Mg# of 83 + 10% H2O) to produce starting materials with 1, 1.5, and 2.5 wt.% H2O. Experiments were performed in a Walker-style multianvil apparatus in AuPd and Fe-presaturated AuPd capsules at 1250-1475 °C. Resulting charges consist of liquid+ol+opx±cpx±gt and all phase compositions were analyzed by electron microprobe. Though liquids quench to glass in some experiments, they are more commonly preserved as 20×20 to 100×300 micron pools of heterogeneous quench crystals. For these, extensive analyses are required to reconstruct partial melt compositions. Consequently, we perform ~150 individual microprobe analyses on the quench for each experiment and accept only those analyses (typically 10 to 40 analyses) that yield olivine-liquid Fe-Mg KD values between 0.30 and 0.35. Inferred equilibrium liquid compositions range from basaltic at high temperature to nephelinitic at low temperature. Melt fractions, F, are calculated by mass balance and range from 0.12 to 0.31. As an index of the influence of H2O on melting, we calculate ΔT, which is the difference between the temperature of the experiment and that required to generate that melt fraction under dry conditions, which are inferred from the experiments of Walter (1998) and from those of Davis and Hirschmann (in prep). For experiments in AuPd capsules, small corrections for iron loss are also required. Calculated values of ΔT as a function of dissolved H2O concentration are inferred by mass balance, and conservation of H2O during experiments can be investigated by FTIR on glassy quench products. The ΔT vs. H2O in the melt trend shows excellent agreement with previous determinations at lower pressure. This simple relation allows calculation of the influence of flux melting in arc and back-arc settings, and, when extrapolated to low bulk H2O contents, to the influence of H2O on partial melting in the source regions of oceanic islands and mid-ocean ridge basalts.
MR33C-03 INVITED
Melting at High Pressures
At high pressures, melts tend to become more similar to the crystalline solid phases. In general, the change in volume with melting, Δ V, becomes small and the entropy of melting, Δ S, becomes constant leading to a melting curve that bends over and approaches a constant. [1] Solids near melting also show approaching dynamical instabilities near melting, such as enhanced diffusivities in a premelting region. [2] Some materials display different behavior. Na, for example, shows a melting curve with a maximum followed by a negative pressure slope down to low temperatures. [3] Raty et al. proposed that the electronic structure of liquid and solid sodium are different, due to opening of a pseudogap in liquid Na, leading to increased density of the liquid and a negative melting slope. [4] We have performed first-principles molecular dynamics simulations for solid and liquid sodium as a function of P and T, and find no evidence of a pseudogap or electronic transition in Na. Rather we find that liquid Na is denser due to closer first neighbors with icosohedral packing due to softening of the potential such as occurs in a Gaussian core potential. We are also performing first-principles MD for Mg2SiO4 liquid to understand diffusivity and dynamical properties of the melt using the QBOX code. Initial results show D=(14, 3.2, 16) 10-6 m2/s for (Mg,Si,O) at P=0 and 6000K and D=(5.7, 1.9, 9.2) 10-6 m2/s at 5000K. Lower temperature and higher pressure simulations are in progress. [1] R. E. Cohen, and Z. Gong, Phys. Rev. B 50, 12301 (1994). [2] R. E. Cohen, and J. Weitz, in Properties of Earth and Planetary Materials at High Pressure and Temperature, edited by M. H. Manghnani, and T. Yagi (AGU, Washington, D.C., 1998), pp. 185. [3] E. Gregoryanz et al., Phys. Rev. Lett. 94, 185502 (2005). [4] J.-Y. Raty, E. Schwegler, and S. A. Bonev, Nature 449, 448 (2007). [5] F. H. Stillinger, and P. G. Debenedetti, Biophysical Chemistry 105, 211 (2003).
MR33C-04 INVITED
Melting of Fe and Fe120Si8 at the Earth's Core Pressures by ab Initio Molecular Dynamics
The solid Earth's inner core (IC) consists mainly of iron likely alloyed with some light elements. At low
temperature iron is stable in hexagonal close packed (hcp) phase up to very high pressures.
However, there is an accumulating evidence that under pressures (~ 364 GPa) and temperatures
(above 6000 K) in the Earth's IC iron, either pure or alloyed with light elements (e.g. Si), might be stable
in the body-centred cubic (bcc) phase1,2. The melting temperature of this phase in the IC is
unknown. Conditions of the IC are not achieved in experiment. Previous theoretical studies concentrated
mostly on the melting of the hcp phase3. We show, by combination of ab initio molecular
dynamics and Z-method4 that pure bcc Fe melts at at the pressure in the center of IC at
~7000 K. Iron, alloyed with 6.25% of Si, melts at a temperature of ~7200 K. While light elements
depress hcp Fe melting temperatures5, we show that Si addition has opposite effect on
bcc Fe. Melting temperatures of bcc and hcp 2,3 are within mutual error bars,
even though bcc melts at a higher temperature. However, the melting temperature of Si-alloyed
bcc iron is clearly above that of Si-alloyed hcp phase5. This is because of different
bonding of Si-Fe within the bcc as compared to the hcp structure. Therefore, the existing
estimates of core temperatures have to be corrected upwards. 1. Brown, J.M. & McQueen, R.G. J.
Geophys. Res. 91, 7485(1986).
2. Belonoshko, A.B., Ahuja, R. & Johansson, B. Nature 424,
1032(2003); Belonoshko, A.B., Skorodumova, N.V., Rosengren, A. & Johansson, B. Science 319,
797(2008).
3. Belonoshko, A.B., Ahuja, R. & Johansson, B. Phys. Rev. Lett. 84, 3638(2000); Alfé,
D., Gillan, M.J. & Price, G.D. Nature 401, 462(1999).
4. Kresse, G. & Furthmüller, J. J. Phys.
Rev. B 54, 11169(1996); Belonoshko, A.B., Skorodumova, N.V., Rosengren, A. & Johansson, B. Phys. Rev. B 73, 012201(2006).
5. Alfé, D., Price, G.D. & Gillan, M.J. Cont. Phys. 48, 63
(2007).
MR33C-05 INVITED
In-situ Elasticity and Density Measurements on Melts and Amorphous Materials at High Pressures
Physical properties of melts are of great importance for understanding the dynamics and differentiation at various stages of the Earth's evolution. While elasticity and density equation of state of crystalline phases now can be consistently measured to the conditions of the core-mantle boundaries, precise measurements of these properties on melts at high pressure are still under extensive investigations. With the application of synchrotron X-radiation sources, new developments have emerged to facilitate the study properties and structures of melts and amorphous materials at the conditions of the deep interior of the Earth. For example, the application of X-radiography/ X-ray tomography enables the in-situ measurement of viscosity and density in multi-anvil and diamond anvil cell apparatus using falling sphere/sink-float, X-ray absorption and X-ray microtomography methods. In this study, we present a new technique for density measurements on melts and amorphous materials by using simultaneous synchrotron X-radiation and ultrasonic interferometry measurements in a cubic type multi-anvil apparatus installed at X-17B2, NSLS of Brookhaven National Lab. By directly measuring the length(thickness) of the sample using X-radiography, a precise determination of sound velocities can be obtained at high pressures, from which the density of melts and/or amorphous materials at high pressures can be calculated through an integration with respect to pressure. This technique opens a unique opportunity for the study of melts and amorphous materials, which will be demonstrated using results obtained from our recent studies on ceramic and metallic materials.
MR33C-06 INVITED
The Role of Cation Coordination Change on Topological Mechanisms of Compression in Silicate Melts at Low Pressure
It has long been recognized that silicate melts are several times more compressible than corresponding crystals at one bar. However, this enhanced compressibility of liquids relative to crystals decreases rapidly with pressure. By 6 GPa, the difference may be <20%. The enhanced compressibility of silicate liquids at one bar is directly attributable to topological mechanisms of compression only available to liquids (and not crystals). However, these topological mechanisms appear to be affected by the coordination (IV, V, or VI) of the network modifying cations (e.g., Si4+, Al3+, Ti4+, Fe2+). Before exploring this connection further, what is meant by "topological" mechanisms of compression is first explained, which requires a comparison between the compression of crystals vs. liquids. The principal mechanisms of compression for crystals (also available to liquids) involve changes in either bond lengths or T-O-T bond angles. Crystals may also undergo an abrupt, first-order phase transition to a higher-density structure, which corresponds to an abrupt change in topology, with or without cation coordination change (e.g., coesite to stishovite vs. quartz to coesite). Both of these phase transitions involve bond breaking and thus a change in topology. In contrast to crystals, liquids may undergo continuous and gradual changes in topology with pressure, owing to the dynamic character of liquids where bonds are continuously broken and reformed. These topological mechanisms of compression in liquids play a profoundly important role in promoting an increase in the density contrast between melts and crystals at low pressure (5-0 GPa) and thus partial melting and melt segregation in the upper mantle. The connection between Si4+ coordination and topological mechanisms of compression in pure SiO2 liquid is discussed in Treve et al. (2002), where they show that defect concentrations in VSi enhance the network connectivity (increase in density through topological change). Here, the connection between Ti4+, Fe2+ and Al3+ coordination change and topological mechanisms of compression are illustrated. For example, a series of 1 bar sound speed and density measurements on 10 Na2O-TiO2-SiO2 liquids (Liu et al., 2007) and 4 CaO-MgO- TiO2-SiO2 liquids, indicate that the partial molar compressibility of TiO2 varies dramatically as a function of Ti coordination. TiO2 compressibility is 9.9± 0.1, 18.7± 0.2, and 6.3± 0.2 × 10- 2 GPa-1 at 1673 K when Ti is 6-, 5-, and 4-fold coordinated, respectively. Similarly, in CaO-FeO- SiO2 liquids, the partial molar compressibility of FeO varies from ~7.0 ± 0.1 to ~5.0 ± 0.2 × 10-2 GPa-1 at 1673 K as Fe2+ changes from 4- to 5-fold coordination. Finally, the partial molar compressibility of Al2O3 at 1673 K is striking different in CaO-MgO-SiO2 liquids (4.51 ± .04 × 10-2 GPa-1; Ai and Lange, 2008) vs. Na2O-SiO2 liquids (-3.96 ± 0.22 × 10-2 GPa-1; Kress et al., 1988), which is likely related to the occurrence of small concentrations of five-coordinated Al in the CaO-MgO-SiO2 liquids at one bar. Causes for these variations in compressibility (dominated by topological mechanisms) with cation coordination change are explored.
MR33C-07 INVITED
Melt Viscosity at High Pressures: A complementary strategy.
The pressure dependence of the viscosity of silicate melts continues to be one of the most important outstanding questions in the area of transport and state in the deep earth. The experimental challenges to viscosity determination at extreme pressures are multiple, primarily involving the feasibility of achieving large isothermal domains of liquid silicate where the significant strain involved in most viscometry methods can be adequately observed. Driven by the scientific questions regarding the nature and efficiency of processes at the estimated extreme thermal regime of the deep earth, the experimental progress has been dominated by the superliquidus temperature range of experimental viscosity determinations. The work is certainly delivering increasingly adequate coverage of the very high P high T range of melt viscosity – but it is not and cannot provide a comprehensive picture of melt rheology that is sufficient for the purpose of developing generalisable viscosity-composition relationships. This is because the very strongly non-Arrhenian temperature-dependence of viscosity observed universally for molten silicates at low pressure and anicipated at high pressure as well. Thus I propose thta the experimental task of describing melt viscosity at very high pressures. I propose therefore that the experimental strategy to obtain melt viscosity at very high pressures be complemented by lower temperature techniques. In this manner the complete viscosity-temperature relationships of molten silicates at very high pressures may become accessible. This strategy has been the key to the development of recent high-quailty multicomponent models of melt viscosity such as the GriD model (Giordano, Russell, Dingwell, EPSL, 2008). I urge the experimental testing of the feasibility of techniques aimed at obtaining glass transition temperatures and relaxation time determinations at temperatures just above this transition. I will present some possible lines of approach for such experimental strategies including the experience gained from lower pressure for handling such materials under what are metastable conditions.
MR33C-08 INVITED
In Situ Raman Spectroscopic Investigation of the Structures of Aqueous Fluid, Hydrous Melt, and Supercritical Fluid in the System KAlSi3O8-H2O up to 900 °C and 2.3 GPa
In subducting slabs, the P-T conditions evolve to the point where aqueous fluid and hydrous silicate melt become completely miscible and form a supercritical fluid. Although this transformation is well documented, the properties of these slab-derived fluids are not yet well understood. Water affects physical (e.g., viscosity and density) and chemical properties (e.g., element partitioning) of silicate melts and supercritical fluids mainly by changing their microscopic structures. Therefore, investigation of the structures of these fluids under high P-T conditions and the formulation of a dissolution mechanism for H2O in silicate melt are crucial for understanding the properties of slab-derived fluids in subduction zones. Here we report an in situ Raman spectroscopic investigation of the structures of aqueous fluid, hydrous melt, and supercritical fluid in the system KAlSi3O8-H2O, which represents an end-member analog of the composition of experimentally determined subduction-zone fluids. The experiments were performed in a hydrothermal diamond-anvil cell [Bassett et al., 1993] equipped with relatively low-fluorescence type Ia diamonds with 750-1000 μm culets. Iridium gaskets with 125- μm initial thickness were used to contain the sample. A piece of KAlSi3O8 glass, which was preformed into a disk shape, was put in the central hole of the gasket together with de-ionized water. Molybdenum or chromel wires were used for heating. During operation, the cell was flushed with an argon- hydrogen (2 %) mixture to prevent oxidation of diamonds and heaters. Temperature was measured by a pair of type-K thermocouples which were attached to both the upper and lower diamond anvils. Pressure of the sample was determined using the equation of state of pure water. The error in pressure and temperature is estimated to be ~± 0.2 GPa and ± 2 ° C, respectively. The unpolarized Raman spectra were acquired by a HORIBA Jobin Yvon confocal LabRAM HR800 spectrometer (800 mm focal length), using 532.06 nm (frequency doubled Nd:YAG) laser excitation source operating at ~20 mW, 100 μm confocal aperture, and a 1800 gr/mm grating. The structures of aqueous fluid and hydrous melt become closer when conditions approach the second critical endpoint (~1.5 GPa and 800 °C). Almost no three-dimensional network was observed in the supercritical fluid above 2 GPa. However, at these P-T conditions, a large amount of silicate component is dissolved, suggesting that the physical and chemical properties of these phases change drastically at around the second critical endpoint. This kind of highly concentrated but low-viscosity fluid is expected to play an important role in transferring the materials in subducting slabs to the mantle wedge efficiently upon dehydration of hydrous minerals. Our experimental results indicate that the fluids released from a subducting slab change from aqueous fluid to supercritical fluid with increasing depth under the volcanic arcs.
MR33C-09 INVITED
The structure of high pressure structure of magnesium silicate liquids: insight from in situ diffraction studies.
Liquids in the MgO-SiO2 system provide a good approximation for the liquids that are produced when the mantle minerals from planetary interiors melt, however because of their refractory nature, direct study of liquid structures is logistically difficult. Magnesium silicates do not form glasses particularly easily and specialised synthesis techniques are required to provide vitreous forms. Combined neutron and high energy X-ray diffraction studies for a range of magnesium silicate glasses has however shown a change in structure between 38 and 33 % SiO2. This is an increase in coordination number from 4.5 to 5.0 and a limit to the formation of a polymerised silicate network. These changes coincide with a change in the rheology and similar changes might be anticipated at high pressure in more silica-rich compositions. Diffraction studies at high pressure and temperature provide a direct measure of the changing structure of liquids and amorphous materials that result in changes in their density and bulk thermodynamic properties. In this presentation the results of neutron and X-ray scattering experiments on magnesium silicate liquids and glasses are presented. Neutron diffraction studies on a single composition magnesium silicate glass from ambient pressure to 8.7 GPa show changes in the amorphous structure. There is a change in the oxygen environment consistent with a collapse of a relatively open silicate framework. Between 5.7 and 8.6 GPa changes in the local environment of magnesium is identified consistent with a change from a low pressure glass structure with a sparse silicate network to a higher pressure structure in which connected magnesium-oxygen polyhedra adopt the network-forming role. Further insight into this high pressure behaviour is provided by ambient pressure diffraction measurements of liquid magnesium silicates are high temperature. These measurements using high energy X-rays and levitation techniques show changes in the local environment surrounding magnesium and changes in structure when the liquids are cooled to form glasses. Intriguingly, the glass structures are not necessarily direct representatives of the liquid. Although both the high temperate and high pressure experiments are challenging, these in-situ studies reveal a wealth of structural complexity in geophysically important liquids. As new facilities become commissioned and new techniques are developed, studies of these and similar liquids at elevated pressures and temperatures will soon be achievable.
MR33C-10 INVITED
Structural Relaxation of High-Pressure Silicate Melts During Quenching And Decompression
Quenched silicate glasses recovered from high pressure have been used to apply structural study of densified silicate melts. Densification mechanism of silicate melts has been discussed on the basis of study on the quenched glass. It is, however, known that the structure of recovered glass is affected not only by fictive temperature during quenching but also by decompression process from high-pressure to ambient condition. It is important to understand how structure of silicate glass relaxes during quenching and decompression, in order to evaluate the structure of silicate melts under pressure. We studied the structure of some aluminosilicate melts and glasses under pressure by in situ x-ray diffraction experiments using synchrotron radiation. Structural change during quenching and decompression was evaluated from both structure factor and radial distribution function with the aid of molecular dynamics simulation. Quenching from melt to glass under pressure does not affect much the nearest neighbor environment but change significantly the intermediate range structure, such as network of TO4 tetrahedra, During quenching, the FSDP of structure is sharpened and the corresponding RDF peaks, such as O-O and T-T peaks, become clear independent peaks, which indicate strong ordering in the network of TO4 tetrahedra. On the other hand, decompression has a little effect on the glass structure, which is mainly caused by elongation of atomic distances. This decompression effects must, however, affects the number of oxygen coordinated aluminum and the distribution of bridging and non bridging oxygen.