MR32A-01 INVITED
Temperature effects on oxide melt structure: spectroscopic constraints and thermodynamic implications
Most oxide melts have significant configurational components to their thermodynamic properties (e.g. heat capacities well above the classical vibrational limit), requiring that their structures become increasingly disordered at higher T. In several oxide melt systems, spectroscopic and scattering studies have begun to quantify such structural changes. Comparison of apparently diverse systems can lead to new insights into underlying common features of interactions among structural species. In borate and borosilicate liquids, recent high T spectroscopic studies, as well as results on glasses with different fictive T's prepared by cooling at different rates, have shown that tetrahedral boron ([4]B) converts to trigonal boron ([3]B) at higher T, with a mechanism involving non-bridging oxygens (NBO) often suggested as [4]B = [3]B + NBO. We have recently directly detected the implied effects on oxygen speciation by O-17 NMR in a Ca- aluminoborosilicate. There is now good evidence, also largely from NMR studies, that this reaction occurs reverse for B, Al, and Si during pressure-induced densification, as [n]M + NBO = [n+1]M. However, the interaction of network cation coordination and NBO as a function of temperature in aluminosilicates has remained poorly constrained. In a recent study of Ca aluminosilicates, using in-situ high T Al-27 NMR and MAS and 3QMAS NMR on glasses with varying fictive T, we found clear evidence that the content of [5]Al increases with increasing T. The proportion of "anomalous" NBO (e.g. NBO in CaAl2Si2O8 glass) also increases slightly with T; but in NBO-rich compositions the NBO content seems to be decoupled from [5]Al. Understanding the relationship of this complexity to T and P effects in other systems requires formulations of reactions among species that account more completely for changes in oxygen speciation, particularly the fractions of oxygens with three network cation neighbors. In these systems as well as in boron-rich melts, observed structural changes can account for significant parts of the configurational properties, but other, as yet to be determined, changes much also be important or even predominant.
MR32A-02
Pressure-Induced Changes in Crystal-Melt Partitiong Coefficient between Silicate Melts and Crystals: A View from Solid-State NMR
Silicate melts in the Earth's interior played essential roles in physicochemical differentiation of the early Earth. Diverse macroscopic properties of silicate melts in the Earth's interior including crystal-melt partitioning depend on their atomic structures at high pressures. Despite their essential implications to many geophysical and geodynamic problems, little is known about the nature of silicate glasses at high pressure, including the densification mechanisms and the atomistic origins of the macroscopic properties at high pressure, mostly due to the lack of suitable high-pressure experimental probes. Recent advances in high-resolution multi-dimensional solid-state NMR and synchrotron x-ray raman scattering yielded improved insights into the structure of oxide glasses with varying pressure (e.g. Lee SK. Geochim. Cosmochim. Acta 2005, 69, p3695; Lee SK et al. Nature Materials 2005, 4, p851; Lee SK et al. Phys. Rev. Lett. 2007, 98, 105502)). The structural information from the above advanced spectroscopic tools has been useful to calculate thermodynamic properties including crystal-melt partitioning coefficient (Lee SK. Geochim. Cosmochim. Acta 2005). Here, we report the spectroscopic evidence of differential pressure dependence controlling the abundances of several types oxygen configuration in complex silicate glasses at high pressure. While all of the glasses studied exhibit a general trend of decreasing non-bridging oxygen concentration with pressure, the details of their pressure dependence is strongly dependent on the composition of melts. Several types of non- bridging oxygen in the divers glass also have varying pressure dependence. Fraction of [4,5]Al-O-[4,5]Al in aluminosilicate glasses, a measure of configurational entropy in aluminosilicate glasses decreases with pressure, suggesting a decrease in configurational disorder caused by the mixing among high energy framework units. We then calculated the crystal-melt partition coefficient from the structural information from solid-state NMR and inelastic x-ray scattering. Our results indicate that the detailed structural information around oxygen in silicate melts at high pressures should be taken into the modeling of their thermodynamic properties.
MR32A-03
Structure of Ca and Mg aluminosilicate melts under pressure
Viscosity and density are important physical properties to control the transportation of magma in the Earth's interior, which are affected by microscopic structure of magma. Magmas have been considered to be densified by structural change with increasing pressure. Recently, Allwardt et al. (2005) found the structure change in auminosilicate glasses quenched under pressure using the NMR spectroscopy, in which an average coordination number of aluminium increases from four to five up to about 10 GPa. We carried out direct observation of structural change in aluminsilicate melts by using in-situ X-ray diffraction method. High-pressure and high-temperature experiments were conducted using MAX80 installed at AR-NE5C of Photon Factory, Tsukuba, Japan. Energy dispersive X-ray diffraction method is applied to acquire diffraction patterns from molten samples. X-ray diffraction spectra were acquired for Ca3Al2Si6O18 and Mg3Al2Si6O18 composition melts up to 5 GPa and 2000 K. Signs of pressure-induced structural change are observed in radial distribution functions as well as in diffraction patterns. The first sharp diffraction peaks (FSDP) shift to higher-Q side with increasing pressure, indicating reduction of the size of the intermediate rage structure, such as rings or chains composed of SiO4 and AlO4 tetrahedra. Radial distribution function also changes with pressure. The nearest neighbour T-O peak (composed of Si-O and Al-O interactions) becomes asymmetric, indicating the coordination increase of the nearest neighbour ions. These observations are consistent with the densification of high-pressure quenched aluminosilicate glasses.
MR32A-04 INVITED
First-Principles Investigation of Melts in the MgO-SiO2 System
We have recently completed simulations of four melt compositions, namely, SiO2, MgSiO3, Mg2SiO4 and MgO within density functional theory. These results allow us to investigate the structural and thermodynamical properties of melts along the MgO-SiO2 join as a function of pressure. In particular, we have found that the mixing in MgO-SiO2 system is substantially nonideal at low pressures with negative excess volume and enthalpy of mixing. With increasing pressure, the volume of mixing decreases rapidly to a value close to zero at pressures above 50 GPa whereas the enthalpy of mixing remains negative. The radial distribution functions and coordination environments are found to show interesting changes with varying composition. Also, the effects of composition on diffusivity are shown to be substantial at low pressures whereas the effects are increasingly suppressed with increasing pressure.
MR32A-05 INVITED
Important yet Overlooked Topological Mechanisms of Compression in Silicate Liquids at Low Pressure (0-5 GPa).
The compressibility of silicate liquids has long been known to be larger than corresponding solids (especially pronounced at low pressures; 0-5 GPa), with important consequences for the increase in the density difference between partial melts vs. crystalline mantle with decreasing pressure, which in turn has consequences for decompressional mantle melting and an increased buoyancy drive for partial melts. Despite the importance of the enhanced compressibility of liquids relative to solids at low pressures, its cause has often been overlooked. Instead, most of the focus on understanding the structural mechanisms of liquid compressibility has been limited to Al3+ and Si4+ coordination change. Here, a discussion of the importance of topological mechanisms of compression (and expansion) in liquids at low pressure is presented. The principal mechanisms of compression for minerals (also available to liquids) involve changes in either bond lengths or T-O-T bond angles. Minerals may also undergo an abrupt phase transition (usually first-order) to a higher-density structure with pressure, which corresponds either to an abrupt change in topology (e.g., quartz to coesite) and/or cation coordination (e.g., coesite to stishovite), both of which can induce large changes in density. In contrast, liquids may undergo continuous and gradual changes in topology (network connectivity) and/or cation coordination, which requires bonds to be broken and reformed and thus reflects the dynamic character of liquids in contrast to solids. Thus, not only changes in cation coordination, but also changes in topology is a mechanism of compression uniquely available to liquids. The magnitude of these topological mechanisms of compression can be seen by comparing the compressibility of liquid KAlSi3O8 to that of sanidine as a function of pressure. At one bar, the compressibility of liquid KAlSi3O8 is 264% larger than that of its crystalline equivalent; this difference diminishes to a 19% difference by 6.5 GPa. There is little evidence for coordination change of either Al3+ or Si4+ in KAlSi3O8 liquid at low pressure, so that the enhanced liquid compressibility clearly involves topological mechanisms (e.g., changing from a tridymite to a feldspar topology) rather than one based on Si/Al coordination change. This mechanism of compression rapidly increases with decreasing pressure and facilitates the large density difference between liquid and crystal at one bar. Another example of the importance of topological mechanisms of compression (and expansion) in liquids is seen with the volume of fusion (18%) of diopside at one bar. The 18% increase in liquid volume cannot be explained by a change in Si4+ coordination number or by mechanisms available to crystalline diopside. In fact, crystalline diopside would have to be heated by more than 2000 degrees above the one-bar melting temperature in order to increase its volume by 18%. Thus, the only plausible explanation for the large volume of fusion is a change in average topology (e.g., from a pyroxene to a pyroxenoid topology). As liquid CaMgSi2O6 is then compressed with increasing pressure, a wide range of denser topologies, including one similar to pyroxene, can be accessed; this mechanism contributes substantially to the compressibility of liquid CaMgSi2O6 at low pressure. Other examples and the consequences for the efficient extraction of oceanic and continental crust out of the uppermost mantle will be given.
MR32A-06
Large Grüneisen Gamma of Dense Silicate Liquids: More Experiments and a First Self- consistent Model
The Grüneisen parameter, γ, of solid materials normally decreases upon compression, approximately as γρq = constant where q=1. However, multiple lines of evidence now indicate the opposite behavior in silicate liquids, in which γ increases upon compression (i.e., q<0). This was observed in shock-melted (Mg,Fe)2SiO4 liquid by Brown et al. [1] via comparison of the Hugoniot and release velocity. We observed the same behavior in Mg2SiO4 liquid (q ≤ -1.5) from comparison of the Hugoniots of forsterite and wadsleyite [2]. First-principles molecular dynamics simulations of MgSiO3 liquid [3] confirm that γ increases with density and show that γ in the liquid phase mimics solids with similar Si coordination state. Hence a continuous increase in γ of silicate liquids to lowermost mantle pressures, well beyond the range where transition to six-coordination of Si is complete, suggests that even higher-coordinated species are forming in the melt and by extension there may be 8- coordinated silicate minerals with stability fields beginning not very far above the Earth's core-mantle boundary pressure [4]. We present new experimental evidence for this behavior in another liquid composition. The Hugoniot of 1400°C anorthite-diopside eutectic liquid was measured at low pressure by Rigden et al. [5] and extended to 110 GPa by our recent work. We collected a Hugoniot point on a solid aggregate of the same composition initially at room temperature, shocked into the melt regime at 133 GPa. The difference in internal energy between this point and the hot liquid Hugoniot allows determination of the γ of this aluminosilicate liquid at 50% compression; the result fits q = -1.85±0.2, entirely consistent with the behavior of enstatite, forsterite, and Fe- bearing olivine liquids. We suggested on the basis of an approximate calculation that the large γ of dense silicate liquids yields a liquid isentrope steeper than the liquidus of a lower mantle magma ocean [2]. Here we show a preliminary self-consistent thermodynamic model of the MgO-SiO2 binary that matches the phase diagrams of MgO, Mg2SiO4, MgSiO3, and SiO2 in the lower mantle, that incorporates negative q in the γ model of the liquid, and that allows calculation of pressure-entropy diagrams showing how model isentropes behave during cooling. We find that for peridotite or chondritic compositions, perovskite crystallization begins at an entropy maximum near 60 GPa. The consequences for geochemical evolution depend on whether these crystals remain turbulently suspended or fractionate [6]; in the case of suspension our model shows that the mush transition affects the entire lower mantle over a rather narrow range in potential temperature. Below this point the solidus does not have a maximum and normal decompression melting behavior is observed. 1. Brown et al., in High-Pressure Research in Mineral Physics, M.H. Manghnani and Y. Syono, Editors. 1987, AGU: Washington, DC. p. 373-384. 2. Mosenfelder et al., J. Geophys. Res., 2007. 112: p. B06208. 3. Stixrude & Karki, Science, 2005. 310(5746): p. 297-299. 4. Akins & Ahrens, Geophys. Res. Lett., 2002. 29(10): 1394-1397. 5. Rigden et al. J. Geophys. Res. 1988. 93(B1): p. 367-382. 6. Solomatov & Stevenson. J. Geophys. Res., 1993. 98(E3): p. 5375-5390.
MR32A-07 INVITED
Silicate melt viscosity: from low to high pressure and the need for metastable state data.
The remarkable variation of viscosity with composition and temperature exhibited by silicate liquids has provided an enormous task for experimental petrology and volcanology. Such great strides have been made in their experimental parameterisation in the past decade that the description of the temperature-dependence and composition-dependence of liquid silicates at low pressure is now possible for many applications using new multicomponent non-Arrhenian approaches (e.g. Russell et al., this meeting). Those models serve not only for earth science applications. Rather, they are sufficiently general that they can be used to pose questions regarding the fundamental structural origins of multicomponent melt viscosity. Further, the models can be used in a practical matter to make predictions that help to guide further experimental approaches. The latter activity serves, in turn, to test the models themselves. A great underinvestigated terrain now separates such models from coping with melt viscosity at very high pressures. This gap must be closed in the coming years through well-selected experimental studies as well as the simulation of melt viscosity - both of these activities at high pressure. An essential precursor to the success of low pressure models has been the inclusion of low temperature, high viscosity data. These data are often obtained in the metastable state of supercooled liquids, at temperatures just above the glass transition. Further vital data come to us from glass transition temperatures themselves, also metastable state data. The full description of the pressure-dependence of multicomponent melt viscosities will inevitably require such metastable state data as well. How easy the experiments will be remains to be seen, but they must become a priority if we are to achieve a fully generalisable pressure-dependent model for melt viscosity. Certainly advances in the simulation of ever increasing viscosities will be a great help in this quest. Only when we can fully parameterise the influence of pressure on the full temperature-viscosity relationships of liquid silicates will we be in a position to grasp the structural basis of viscosity in these liquids under deep earth conditions.
MR32A-08
Pressure Effect on Hydrous Rhyolite Viscosity: a Model
Viscosity is the single most important property governing the efficiency, rates and nature of melt transport. In geological environments viscosity controls eruption dynamics and rates of physicochemical processes (e.g., degassing, crystallization) in magmas. Composition, volatile content, temperature, pressure, as well as crystal and bubble contents are parameters influencing the viscosity of silicate liquids to various extents. The prediction of viscosity over the range of conditions encountered in nature and, in particular, at conditions relevant to eruptive events, is still challenging. The influence of H2O-content on viscosity is dramatic (e.g., the addition of 1 wt% of water may decrease viscosity by 6 orders of magnitude). Rhyolitic melt at pressure conditions typical for magma chambers contain up to 10 wt% H2O and in this study we experimentally determined the effect of pressure on the viscosity of hydrous rhyolitic melts. The Newtonian viscosity of synthetic rhyolitic liquid (HGG) containing 0-5.25 wt% dissolved water was measured at pressures from 4 to 26 kbar. The combined use of a high T concentric cylinder and the high-T, high-P falling sphere techniques allowed viscosity measurements in the interval from 102 to 107 Pa s, pertaining to eruptive conditions. No previous experimental work has investigated such a wide range of viscosity. Concentric cylinder measurements were performed on dry melts up to 1650°C. The falling sphere technique on static and centrifuging piston-cylinders allowed to measure viscosities between 580 to 1350°C at different pressures. The increased acceleration field (50-1000 g), applied to the sinking sphere, enabled us to measure viscosities below the solidus temperature. As to be expected, viscosity decreases with water contents and with increasing temperature. The viscosity response of hydrous melt to pressure increase is different: as expected at high T the viscosity decreases, whereas at low T the viscosity increases. For dry compositions a decrease of viscosity is observed in the entire T-range, indicating that in the presence of H2O, T and P- induced speciations and structural variations significantly affect melt rheology. A combination of our and previous studies on rhyolitic melts were employed to calculate an empirical model reproducing the experimental data and predicting the viscosity of rhyolitic melts in the full range of T-P-H2O- X space for which data are available. The model is based on the Vogel-Fulcher-Tammann (VFT) equation. Our model reproduces the viscosity of about 500 viscosity data obtained on both synthetic and natural silicic melts from peraluminous to peralkaline with RMSE values of less than 0.25 log-units and less than 5% relative error on a logarithmic scale.