T13C-1961 INVITED
Seismic Hygrometry of the Mantle
Somewhere on the order of 1012 kg of water is subducted yearly. Most is returned to the surface hydrosphere through arc volcanism and broadscale flow, but some amount, perhaps as much as a quarter, makes it into the deep upper mantle and beyond. The fate of that water, apart from its distribution in MORB source regions, is largely unknown. Our ignorance isn't for lack of interest: the geophysical effects of small amounts of water on mantle silicates are often subtle. In particular, the predicted seismic velocity decrements of lightly hydrated mantle rocks are easily masked by or mistaken for temperature and major element variability. Discrimination between water and everything else requires highly accurate estimates of seismic velocities and attenuation and still is non- unique. There are, however, other calling cards of water that seismologists can look to. These "hygrometers" exploit the effects of hydrogen on phase transitions and melting and include: width and elevation of the 410-km discontinuity, the reflection coefficient ratio of the 520-km to 410-km discontinuity, the presence of partial melt atop the 410-km discontinuity and, potentially, the strength and width of the seismic lid-LVZ transition. Together, these hygrometers provide a series of spot estimates of water in the upper mantle and transition zone and can be applied anywhere mantle phase transitions can be accurately mapped. We will demonstrate the use of these measures and others in discerning the degree of hydration of a broad swath of western and central Pacific upper mantle, revealing both wet and dry regions. We are quickly approaching the point at which seismology will be able to meaningfully estimate the ppm concentration of water in large portions of the mantle.
T13C-1962
Nature of a low-velocity zone atop the transition zone in Northwestern Canada
Seismic studies over the past decade have identified a low-velocity layer above the transition zone at various locations around the globe. This layer is hypothesized to be a lens of dense, fluid-rich silicate melt ponding atop the 410 km discontinuity, beneath the silicate melt-density crossover theorized to exist within the upper mantle. In order to quantify the nature and geographical extent of the layer in Northwestern Canada, we have assembled a P- and S-wave dataset from the Canadian National Seismograph Network Yellowknife Array (YKA), the Canadian Northwest Experiment (CANOE), and the POLARIS Slave array. P- and S- receiver functions (PRFs and SRFs) for all stations were generated via standard procedures utilizing earthquakes at epicentral distances of 30° to 95° for PRFs and 55° to 85° for SRFs. Totals of RFs computed include ~ 2500 (P) and ~ 400 (S) for YKA, ~ 7000 (P) and ~ 800 (S) for POLARIS Slave, and ~ 25000 (P) and ~ 1600 (S) for CANOE. In order to compute Poisson's ratio, an important discriminant of possible composition and/or fluid content, we generated a suite of 1-D velocity models based on IASP91, but with varying thicknesses and velocity ratios for a hypothetical layer above the 410 km discontinuity. From these models, we computed moveout curves from travel-times of five different scattering modes (Ps, Sp, Pps, Pss, & Ppp) for all combinations of source-receiver geometry represented in the data. A grid search was performed over all possible models, where we shifted and stacked autocorrelations to obtain an estimate of the model that most accurately predicts the interval thickness and Poisson's ratio. Results from geographic profiles of PRFs and SRFs and 1-D migrated depth stacks indicate upper mantle and transition zone arrivals with moveout compatible with both direct and reverbatory phases from the layer. We obtain estimates for the interval Poisson's ratio in the range 0.37 to 0.40, which falls well above the 1-D velocity model average of ~ 0.25 for this depth range and favours the presence of high melt or fluid fractions. The nature of these signals, their spatial distribution, and their implications for the nature of the low-velocity zone will be discussed.
T13C-1963 INVITED
Evidence From Seismology for Water in the Upper Mantle Transition Zone
We discuss two types of seismological observation that may provide evidence for water in the transition zone. First, we review constraints from (global) seismic tomography with P and S wave travel time data. For several subduction systems (e.g., western Pacific) it has been observed that P and S wavespeeds in the mantle wedge are anomalously slow to depths well into the transition zone. In regions where there is evidence for stagnant slabs (e.g., beneath Philippine Sea plate and SE Asia) the anomalous wavespeeds persist far away from the zone of active subduction. These regions are also marked by anomalous Vp/Vs ratios. These observations suggest the presence of compositional variations and, most likely, the presence of water that is recycled from the surface by subduction processes. Second, we present results from high resolution transition zone imaging with generalized Radon transform of the broadband wavefield containing SS precursors. We study the mantle beneath the central Pacific, around and west of Hawaii. In addition to pronounced, laterally continuous interfaces near 410, 520, and 660 km depth the images also reveal structure between 800-1000 km and near 370 km depth. These scatter zones may not form laterally continuous global interfaces but they may indicate hitherto unknown, or poorly understood, changes in composition or mineralogy. The complexity near the top of the transition zone is reminiscent of evidence from (long period) ScS reverberations for silicate melt atop the 410 km discontinuity beneath SE Asia (Revenaugh and Sipkin, 1994). This may suggest that silicate melt and, hence, water is present in the mantle transition zone beneath parts of the Pacific far away from present-day subduction.
T13C-1964
Geodynamical constraint on the history of Earth's ocean volume
The chemical composition of the bulk silicate Earth indicates that the present-day thermal budget is likely to be characterized by a significant excess of surface heat loss over internal heat generation. When combined with the notion of continental freeboard, this thermal budget places a tight constraint on the history of Earth's ocean volume. The capacity of Earth's oceans is determined by seafloor topography, which is in turn controlled by the vigor of mantle convection. The constancy of ocean volume thus implies a specific history of mantle convection, which can be shown to be inconsistent with the degree of secular cooling constrained by petrological data. The ocean volume needs to temporally variable, and a likely scenario of thermal evolution suggests that the Archean oceans had about twice as much water as today, and the mantle may have started as dry and have been gradually hydrated by subduction. The global water cycle may thus be dominated by regassing, rather than degassing.
T13C-1965 INVITED
Requirements for Hydrous Melting at 410 km
It has long been clear that some H2O is recycled into the mantle, as the continued flux of H2O out of the mantle via oceanic volcanism, coupled with near-constant continental freeboard, requires that the flux of subducted H2O at least match that which is outgassing at ridges and oceanic islands. On the other hand, petrologic and geochemical arguments indicate that H2O subduction is rather inefficient, with much of the H2O in the downgoing slab returning to the near-surface environment through metamorphic devolatilization and/or melting in forearc, sub-arc, and back-arc environments. Among the most interesting questions regarding H2O recycling in subduction zones are those related to the transport and storage of H2O in and near the transition zone and the possible role of hydrous partial melting in mass transfer out of the transition zone. In particular, there are now numerous seismic observations suggesting small amounts of melt atop the 410 km discontinuity. Most of these are associated with areas of present or recent subduction, suggesting a causal link. Although other origins, including possible roles for carbonated melts, may be considered, the most obvious hypothesis for the origin of such melts is dehydration melting. Examination of experiments on hydrous peridotite and thermodynamic calculations suggest that stabilization of melt at 400 km along a normal mantle adiabat requires about 15 wt.% dissolved H2O. Given appropriate mineral/liquid partition coefficients for H, DH, the H2O concentration in the peridotite residue can be calculated, and this gives both the minimum H2O required to incite melting as well as the concentration of H2O in the residue once melt has been extracted. Direct experimental determinations of DH, are limited to lower pressures (<5 GPa for pyroxene and garnet, <3 GPa for olivine). Application of these indicates that melts with 15 wt.% H2O are in equilibrium with a residue with 500 ppm H2O. However, thermodynamic models and inferences from olivine storage capacity experiments suggest that DH for olivine increases at high pressure, reaching approximately 0.01 at 400 km. This suggests that the threshold for melting just above the transition zone likely is ~1000-1500 ppm H2O. One of the implications of this is that rocks with ~1000 ppm H2O are delivered to the transition zone by subduction. Such modest quantities do not necessarily require stability of hydrous minerals along subduction geotherms, as much or all of this H2O may be carried by nominally anhydrous minerals.
T13C-1966
Water is not Recycled into the Deep Mantle in Subducting Lithosphere
Overwhelming evidence exists that H2O is carried downward from the surface into the upper mantle as hydrous minerals in the uppermost 10-12km of subducting lithosphere. Large amounts of that water are released as the lithosphere heats up, triggering earthquakes and fluxing arc and back-arc volcanism. Similarly, there is abundant experimental evidence documenting the high solubility of H2O in olivine, the most abundant phase of the upper mantle, and even higher solubility in its high-pressure polymorphs, wadsleyite and ringwoodite, as well as the existence of dense hydrous magnesium silicates (DHMS) that could be stable to pressures equivalent to well into the lower mantle. Here we combine experimental and seismic evidence to test whether the seismicity patterns in subducting lithosphere and the stability relations of these potentially relevant hydrous phases are consistent with a wet lithosphere. We show that there is nearly a one-to-one correlation between dehydrating minerals and seismicity at depths less than ~250 km and conclude that the former is the trigger of the instability that leads to the latter. At greater depths, however, we find no correlation between occurrences of earthquakes and depths where earthquakes are predicted based on expected breakdown of any of these hydrous phases. Lastly, we note that metastable olivine (which, if present, can explain the distribution of deep-focus earthquakes) is incompatible with even small amounts of H2O and that very strong evidence for such metastable olivine exists west of and within the subducting Tonga slab and also has been reported from three other subduction zones. We conclude that subducting slabs are wrung dry by 400 km and thus do not provide a pathway for water to the transition zone or lower mantle.
T13C-1967
Constraints on the H2O content of subducting slabs from olivine-ringwoodite transformation kinetics
It is possible that much of Earth's H2O resides in the upper mantle and mantle transition-zone in nominally anhydrous minerals (NAMs). If Earth's water cycle includes a deep mantle component, subduction would provide the transport mechanism from Earth's surface to the transition zone. How hydrated is subducting mantle lithosphere? Double seismic zones and petrologic evidence have been used to argue that breakdown of serpentines can cause intermediate-depth earthquakes by dehydration embrittlement. If this model is correct, then NAMs in the mantle portion of subducting oceanic lithosphere would interact with an H2O-rich fluid from 2 to 6 GPa, adding hundreds to a thousand wt ppm H2O subducting olivine. In fast and cold subducting slabs, where temperatures are low enough to inhibit equilibrium transformation of olivine, a wedge of metastable olivine is thought to survive deep into Earth's transition zone. Olivine transformational faulting provides an explanation for deep focus earthquakes if the metastable olivine can persist deep into the transition zone. The likelihood of metastable olivine in the transition zone has been evaluated using experimentally determined kinetic data combined with thermo-kinetic models. Our kinetic experiments show that 300 wt ppm H2O in olivine enhances ringwoodite growth rates so much that metastable olivine is unlikely to survive in cold and fast subduction zones if hydrated with 300 wt ppm H2O. Evidence for metastable olivine in the transition zone, including seismic data and deep focus earthquakes, suggest that olivine in subducting mantle lithosphere has less than 300 wt ppm H2O. Here we present experimental results on the transformation of hydrous (~30 wt-ppm H2O) San Carlos olivine to ringwoodite at 18 GPa. Transformation kinetics for olivine with 30 wt ppm H2O will be discussed to constrain how hydrated a subducting slab can be and still have metastable persistence of olivine in the transition zone.
T13C-1968
Does the Nazca Slab Beneath Central Argentina Influence the Water Content of the Adjacent Transition Zone?
When the Nazca flat-slab rolls over and plunges into the transition zone under Argentina, it appears to separate an electrically resistive transition zone to the west from an electrically conductive transition zone to the east. The simplest explanation for this is that the water content of the transition zone is much lower to the west than the east. The low conductivity to the west can be explained if anhydrous upper mantle mantle is being carried down into the transition zone by slab motion. The much higher conductivity to the east is beneath the Rio de la Plata Craton whose root almost certainly inhibits vertical motion east of the slab. Thus water injected by the descending slab is likely to accumulate in the transition zone. This idea was first presented in a Nature paper in 2004. Since then, we have collected more magnetotelluric data to the south where the slab dip is normal, but voluminous back-arc basaltic volcanism occurs and in the region where the slab is said to be flexing continuously between the two geometries. A goal of this work is to test whether the slab has a similar relation to transition zone conductivity along strike. The new data, originally collected along linear profiles perpendicular to the expected strike of the slab in the mantle clearly indicated that 2-D interpretation would be problematic. Indeed, analysis of new data in the flexure region using 2-D methods reveals a narrow, roughly east-west, near vertical resistive structure extending down to the top of a conductive transition zone. A possible, but controversial interpretation of this structure is that it is the signature of a slab tear rather than the widely-accepted continuous flexure geometry. If a tear is indeed correct, then there is an opportunity to test how the slab is influencing the transition zone conductivity and by inference the water content by looking at the southern edge of the plunging 'flat- slab' as it enters the transition zone. Since the original data were collected we have been building an array of sites in both regions to permit fully 3-D analysis. Arrays of sufficient scale to undertake 3-D inversion now exist and preliminary results will be presented. Whatever the outcome of our current work, we want to stress the importance of probing the electrical conductivity of the transition zone in regions where identifiable vertical motions can influence the water content.
T13C-1969
Continental Collision: A Novel Pathway for the Earth's Deep Water-Cycle
Subuction of oceanic lithosphere has been long considered as the main mechanism to carry water into the
deep mantle. Regardless of whether various dehydration reactions may leave subducted slabs dry, new
evidence indicates that continental collision is a viable pathway to bring water into the mantle transition zone
(TZ). Using high-resolution, triplicate P- and S-waveforms recorded by broadband seismic arrays, we show
that a corresponding anomaly of high S-wave speed (VS) is absent where an anomaly of high P-wave
speed (VP) was recently recognized in the TZ beneath central Tibet. A likely cause of the discrepancy
between anomalies in VP and VS is a minor amount of water in nominally anhydrous polymorphs of olivine.
Prior to thickening by continent-continent collision, the Tibetan lithospheric mantle was part of a mantle
wedge which has been hydrated during past episodes of subduction. Hydration of the sub-continental
lithospheric mantle (SCLM) provides a natural mechanism to facilitate ductile deformation, so rapid thickening
of the SCLM, which would have been hindered by advection of cold materials, can take place readily.
Convective instability would then lead to removal and sinking of thickened, cold SCLM, leaving a hydrous
remnant in the TZ detectable only in VP. This interpretation is consistent with occurrence of hydrous
minerals associated with recent magmatism in northern Tibet, interpreted as a direct consequence of
convective instability which took place about 15 Ma ago.
http://netfiles.uiuc.edu/tseng1/Reprint
T13C-1970 INVITED
Role of Hydrogen in stagnant slabs and big mantle wedge
Recent seismic tomography data imply that subducting slabs are stagnant at some regions such as beneath Japan and Northeast China [1, 2]. The stagnant slab can have an important effect on the overlying transition zone and upper mantle. A big mantle wedge (BMW) model has been proposed by Zhao [2], in which the stagnant slab in the transition zone could play an essential role in the intra-plate volcanic activities overlying the slab. Water released by the stagnant slab could be important for such igneous activities, such as Mt. Changbai in Northeast China. In cold subducting slabs, several hydrous minerals together with nominally anhydrous minerals accommodate OH and transport water into the transition zone [3]. The effect of dehydration of the stagnant slab has been analyzed by Richard et al. [4]. They argued that warming of the stagnant slab due to heat conduction could play an important role for the slab dehydration, and local oversaturation could be achieved due to decrease of the water solubility in minerals with temperature, and fluid can be formed in the overlying transition zone. We determined the hydrogen diffusion in wadsleyite and ringwoodite under the transition zone conditions in order to clarify the deep processes of the stagnant slabs, and found that diffusion rates of hydrogen are comparable with that of olivine [5]. We also determined the dihedral angle of aqueous fluid between wadsleyite grains and majorite grains under the transition zone conditions. The dihedral angles are very small, around 20-40 degrees, indicating that the oversaturated fluids can move rapidly by the percolation mechanism in the transition zone. The fluids moved to the top of the 410 km discontinuity can generate heavy hydrous melts due to a larger depression of the wet solidus at the base of the upper mantle [6]. Gravitationally stable hydrous melts can be formed at the base of the upper mantle, which is consistent with seismological observations of the low velocity beneath Eastern China, Europe, and United State (e.g., [7]). According to the BMW model by Zhao [2], the intra-plate volcanisms in Northeast China including Mt. Changbai are different from the hot plumes and they might be generated due to some processes related to the deep-seated dehydration from the stagnant slab. Recent geochemical studies on volcanic rocks and associated mantle xenoliths in Northeast China (e.g., [8]) indicated that there is no geochemical evidence for involvement of subducting slab in most basalts, i.e., no depletion of high field strength elements and no enrichment of large ion lithophile elements. There is no clear evidence for a high-3He/4He mantle plume component in these rocks, i.e., 3He/4He ratios are significantly below the high 3He/4He ratios of mantle plumes such as those beneath Hawaii and Iceland. The geochemical signatures of the deep dehydration should be different from those in the conventional mantle wedge, since the fluids generated at such depths are completely different from those at the shallow depths. Further studies including the element partitioning between fluids and mantle under the deep upper mantle and the transition zone conditions are necessary to clarify the possible role of the Big Mantle Wedge on the intra-plate volcanism. [1] Fukao, et al., J. G. R. 108, doi:10.1029/2001JB000989, 2003. [2] Zhao et al., Chin. Sci. Bulletin 49, 1401, 2004, [3] Ohtani, Elements, 1, 25, 2005. [4] Richard et al., EPSL, 251, 156, 2006. [5] Hae et al., EPSL, 243, 141, 2006. [6] Litasov and Ohtani, PEPI, 134, 105, 2002. [7] Song et al., Nature, 427, 530, 2004. [8] Chen et al., Lithos, 96, 108, 2007
T13C-1971
Sound Velocity Crossover of Hydrated Olivine at High Pressures
The presence of water in nominally anhydrous minerals can dramatically affect the elastic properties of mantle minerals (e.g. Mao et al., 2008). To determine the effect of water on the sound velocity of the upper mantle, we measured the single-crystal elasticity of forsterite, Mg2SiO4, with 0.9 wt.% H2O to 14 GPa by Brillouin scattering. At ambient conditions, the addition of this amount of H2O in forsterite decreases the aggregate bulk and shear moduli by 2.2-2.4% compared with the anhydrous phase (Jacobsen et al., 2008). However, the presence of 0.9 wt.% H2O increases the pressure derivatives of the bulk modulus of forsterite from 4.2 (2) (Zha et al., 1996) to 4.5(1), and the shear modulus from 1.4(1) to 1.8(1). As a result, the bulk and shear modulus of hydrous forsterite become greater than that of anhydrous forsterite at 14 GPa and 6 GPa, respectively. A similar crossover of the bulk modulus is observed in hydrous ringwoodite at around 12 GPa (Jacobsen and Smyth 2006). Using our measured elasticity of hydrous olivine, we examine the effect of water on the sound velocities of olivine with various water contents at typical mantle conditions. We assume a 1400°C adiabat, 10 mol% iron content, and that the temperature derivatives of the bulk and shear moduli of hydrous olivine are the same as those of the anhydrous phase. The compressional and shear velocities of hydrous olivine are offset to lower values at 0-km depth but become faster than those of the anhydrous phase beyond 150 km and 100 km, respectively. If mantle olivine is water saturated (~0.4 wt.% H2O), the corresponding compressional and shear velocities are 0.6% and 1.1% faster than those of anhydrous olivine at 400-km depth. A preliminary model based on the work of Karato (2006) has been constructed to account for effects of anelasticity due to temperature and H2O content. Our model indicates that the effect of anelasticity on the shear velocity of hydrous olivine with 0.9 wt% H2O may offset the increase in the observed anharmonic velocity components of hydrous olivine, and the net effect would be that seismic velocities in the hydrous olivine would be similar to those for dry olivine at 400 km.
T13C-1972
Structure and Compressibility of Iron- and Aluminum-bearing Phase D
We are studying the structure and spectroscopic properties of Al- and Fe-bearing phase D, a dense hydrous magnesium silicate and potential carrier of subducted water into the deep mantle. Gem-quality single- crystals of phase D measuring up to 0.5 mm in size were synthesized at 25 GPa and 1400 °C in a multianvil apparatus with approximate composition, MgFe0.15Al0.3Si1.5H2.5O6. Mössbauer spectroscopy indicates that all the iron is ferric, Fe3+. Unpolarized single-crystal Raman spectroscopy at ambient conditions reveals in the low-frequency range lattice modes at 202, 377, 491, 588, 682, 800, 1114 and 1277 cm-1. In the high-frequency range broad bands at 2204, 2487, 2863 and 3482 cm-1 can be observed. High-pressure synchrotron infrared spectroscopy shows little variation in the position of the main OH band at 2863 cm-1 up to the maximum pressure of 43.5 GPa, with dν/dP = -1.3 cm-1/GPa, which is not indicative of hydrogen bond symmetrization at high pressure. Thermal expansivity of the trigonal unit cell was measured between 140 and 300 K, giving linear coefficients of thermal expansion for the a-axis, c-axis, and volume of 4.8(6) × 10-6, 7.6(7) × 10- 6, and 17(2) × 10-6 K-1, respectively. Structure refinements were carried out by single- crystal X-ray diffraction at 273 and 100 K. A hydrogen position located between the SiO6 octahedral layers was refined at (0.54, 0, 0.09) with an O-H distance of 0.85 Å, similar to pure Mg-phase D [1]. Upon cooling to 100 K, the d(O...O) hydrogen bond distance decreases minimally from 2.646(1) Å to 2.641(1) Å, and the O-H distance is reduced to 0.81 Å at 100 K, further evidence that hydrogen bond symmetrization does not occur in phase D on contraction of the structure. In addition, we have identified a second hydrogen position from difference Fourier synthesis, located along the octahedral edge of the silicon site, suggesting a substitution of (Al, Fe)3+ + H+ = Si4+ on that site. Finally, the isothermal equation of state was studied using single-crystal synchrotron X-ray diffraction to 26 GPa in a diamond-anvil cell loaded in a helium pressure medium. We obtain a bulk modulus of about 164 GPa when the pressure derivative dK/dP = 4.4 is free to refine (where this deviation of dK/dP from 4.0 is statistically significant at the 98% confidence level). Here, we will discuss potential H-sites, substitution mechanisms as well as implications of the stability and properties of Al- and Fe3+- bearing phase D within cold subduction zones in the deep mantle.
T13C-1973
Effects of Hydration on the Elastic Properties of Upper Mantle and Transition Zone Minerals
Water may be incorporated as hydroxyl into both the hydrous and nominally anhydrous silicates of the upper mantle and transition zone. Incorporation of hydroxyl has significant effects on the elastic properties of the major and minor phases of this region such that it may be possible to detect the presence of water from seismic studies. We have been measuring the effects of hydration on the equation of state and elastic properties of hydrous and anhydrous forsterite, wadsleyite, ringwoodite. We have used these new studies together with existing data to model the velocity structure of the upper mantle and transition zone, based on a thermodynamic and empirical model of mantle mineralogy. We find that it is difficult to model the velocity structure of PREM or other seismic velocity models through the transition zone with an anhydrous pyrolite composition, whereas it is straightforward to match the spherical velocity model structures with the addition of a few thousand parts per million H2O by weight. Further, it is difficult to dehydrate a serpentinized subducting slab. Phase A and 10Å phase both result from partial dehydration of serpentine and are stable to depths in excess of 250km where olivine begins to incorporate significant amounts of hydroxyl. At 360 km depth olivine is capable of incorporating nearly one percent water by weight.
T13C-1974
Grain Boundary Structurally-Bonded Water in Olivine Aggregates
Water storage capacity of nominally anhydrous olivine has been extensively investigated because of its numerous geophysical and geochemical implications for the Earth's dynamic mantle. However, all previous experimental research has been concentrated on the water solubility in single crystals of olivine. Grain boundary as potential storage sites for water in the mantle has not been experimentally studied, in part because solubility experiments were always performed under water-saturated condition, rendering the examination of grain boundaries nearly impossible due to the presence of free water. In the present study we have conducted annealing experiments on forsterite at 5 - 6 GPa and 1200 °C and at water- undersaturated condition. Duration was typically 2 - 3 hours. A small amount of enstatite or periclase was added to the starting forsterite powder (including a few large olivine grains) to buffer the silica activity, while oxygen fugacity was controlled by using various capsule materials (Re, Fe, or BN). FTIR analyses were performed on both single crystal and polycrystalline olivine in doubly-polished thin section of each experimental charge. The results are as follows: (1) single crystal and polycrystalline olivine in the same charge always yielded similar IR pattern, indicating all absorption peaks are due to similar structurally-bonded water (i.e., hydroxyl); (2) water content of periclase-buffered (i.e., low silica activity) sample is at least one order of magnitude higher than those of enstatite-buffered and unbuffered (pure forsterite) samples; (3) under reducing environment (Fe or BN capsule), water content of polycrystalline olivine is always higher than that of single crystal by at lease a factor of 5, regardless of silica activity buffering. We therefore infer that large amount of structurally-bonded water is stored at grain boundaries; (4) with decreasing oxygen fugacity, IR spectra of olivine are increasingly dominated by an absorption peak centered around 3600 cm-1, indicating a gradual change in dominant water incorporation mechanism in olivine. These results strongly suggest that grain boundaries could be significant storage sites for water in the Earth's mantle, especially at locations where oxygen fugacity and silica activity are low.
T13C-1975
Crystal Chemistry of Hydrous Forsterite and its Vibrational Properties up to 41 GPa
Forsteritic olivine is a major component of the Earth's upper mantle and a major potential reservoir of H and
thus water in the Earth's interior. The crystal structure of hydrous pure magnesium forsterite with up to 9000
ppmw H2O synthesized at 12 GPa and 1250 °C, has been refined. The major hydration
mechanism appears to be M1 cation vacancy with protonation of the O1-O2 octahedral edge of M1. Raman
spectra up to 41GPa show strong coupling between the two Ag modes (824.4 cm-1 and 856.2 cm-
1). Mode Grüneisen parameters γ
T13C-1976
Compressibility and Structural Stability of Hydrous Olivine Fo97Fa3 up to 34 GPa by XRD and Raman Spectroscopy
Two hydrous olivines of composition Fo97Fa3 with 4883 ppmw (SZ0407A) and 8000 ppmw (SZ0407B) were synthesized at 12 GPa and 1250 °C. Single-crystal diffraction was used to determine the cell parameters at pressures up to 7.1 GPa at room temperature. Synchrotron powder X-ray diffraction and Raman scattering experiments were performed on sample SZ0407A in a diamond-anvil cell to 34 GPa at room temperature. For both samples, the compressibility is the largest along the b-axis and smallest along the a-axis. Using the compression (V/V0) vs pressure data to 29.08 GPa for sample SZ0407A, and the third-order Birch-Murnaghan equation of state, we calculate the isothermal bulk modulus and its pressure derivative as K0 = 119.4 ± 1.5 GPa and K'0 = 6.6 ± 0.5. Single- crystal compression data to 7 GPa for sample SZ0407A give K0 = 121.5 ± 0.6 GPa and K'0 = 5.7 ± 0.2; and for sample SZ0407B we obtain K0 = 122.2 ± 1.2 GPa and K'0 = 6.3 ± 1.0. High pressure Raman spectra for SZ0407A up to 34 GPa show a continuous shift of all the observed bands to higher frequency with increasing pressure. Although there is no indication of any first-order phase transition, the Raman results do indicate some subtle structural modification around 22 GPa, well above the stability field of olivine. In the low-frequency range between 200 cm-1 and 800 cm-1, M2 translation mode at 302.5 cm-1, SiO4 rotation mode at 325.2 cm-1, and ν4 mode at 545.4 cm- 1 disappear, whereas the ν4 mode at 606.9 cm-1 is gradually intensified in the same pressure range, indicating partial appearance of Si-O-Si linkages accompanied by a possible metastable increase in the silicon coordination.
T13C-1977
Orthopyroxene: the Most Hydrous Nominally Anhydrous Upper Mantle Phase?
According to a recent study by Mierdel et al. [1], orthopyroxene, the second most abundant phase in the upper mantle, can incorporate up to almost 1 wt % H2O at high P and T. The data indicate a rapid decrease in storage capacity with increasing pressure from 1.5 to 3.5 GPa, corresponding to the well-known trend of decreasing Al content with increasing pressure in opx. This could have important implications for stabilizing hydrous partial melts, commonly thought to be responsible for the Earth's asthenosphere [1]. However, the OH concentrations measured by Mierdel et al. conflict with several previous studies that obtained much lower amounts, at comparable Al contents (up to ~5 wt % Al2O3, e.g. [2-5]). To investigate this discrepancy, we conducted new experiments on a suite of 10 natural orthopyroxene starting materials with variable Fe/Mg ratios (Mg# 77.5-99.5) and Al contents (0.13-4.76 wt % Al2O3). The experimental conditions were 900 °C and 1.5 or 3 GPa, with fO2 fixed at NNO. Run durations were chosen so as to achieve equilibrium according to available diffusion data for H in opx. Retrieved crystals were analyzed for OH and major element concentrations, taking special care to check for concentration gradients. Whereas the rims of some crystals are characterized by Fe loss, there is remarkable consistency in OH concentrations among different crystals from the same charge, and a lack of gradients in OH concentration. Differences in IR spectra compared to the starting materials, combined with electron probe data, indicate that the coordination environment of Al associated with H in most samples changes from dominantly tetrahedral towards octahedral coordination at high pressure. We obtained a maximum concentration of about 650 ppm H2O at 3 GPa in opx with 4.76 wt % Al2O3 (compared to 400 ppm at 1.5 GPa in the same material). This contrasts with concentrations of 1420-6400 ppm H2O measured by Mierdel et al. for Fe-free opx with similar Al concentrations. The discrepancy between Mierdel et al.'s results and ours, as well as those of other studies on both Fe-free and Fe-bearing opx [2-5], remains unresolved. One possibility is that the diffusion of H in orthopyroxene occurs by multiple mechanisms, as in olivine. One mechanism clearly operates rapidly enough to eliminate concentration gradients in our samples, to within the 50-micron resolution of our technique. However, it is possible that an additional, slower mechanism is involved in the achievement of reversible equilibrium, perhaps leading to higher OH contents such as those observed by Mierdel et al. On the other hand, the extreme variations in Al concentration (up to 50%) and OH concentration (up to 40%) in crystals grown under the same conditions by Mierdel et al. suggest the possibility that rapid growth under non-equilibrium conditions could trap metastable concentrations of point defects. [1] Mierdel et al., 2007, Science, 315, 364-368 [2] Rauch and Keppler, 2002, CMP, 143, 525-536 [3] Koga et al., 2003, G3, 4. DOI: 10.1029/2002GC000378 [4] Stalder, 2004, EJM, 16, 703-711 [5] Stalder et al., 2005, CMP, 150, 473-485
T13C-1978
Diffusion profiles of OH towards melt inclusions in garnets in lherzolite xenoliths from the Victor diamond mine, South Africa
In diamond exploration it is important to understand the mechanism of the kimberlite eruption and dynamics of kimberlites in terms of the composition of the kimberlitic melt. Water is found as OH in nominally water free minerals (NAMS) caused by mantle metasomatism. The water has a higher solubility in the kimberlitic melt and thus is removed from the NAMS. The decrease of OH content in NAMS can thus be measured and used to find the composition of the kimberlitic melt. We present for the first time high resolution FT-IR based synchrotron measurements of OH concentrations towards melt inclusions in lherzolitic garnets from the Victor Diamond Mine in South Africa. We measured hydrogen profiles towards totally embedded microcracks and melt inclusions in these garnet crystals. The measurements show a strong variation in OH-concentrations and demonstrate that the amount of water stored in diamond bearing layers has been underestimated for a long time due to the loss of water during the uplift of the kimberlite.
T13C-1979
OH and CO2 diffusion profiles in garnets from eclogitic xenoliths from the Victor diamond mine, South Africa (UNESCO IGCP 557)
The mechanisms of kimberlite eruptions plays a significant role in diamond exploration. One major physical aspect in understanding the eruption dynamics of kimberlites is the composition of the kimberlitic melt. Water is commonly dissolved in nominally water free minerals (NAMS) as OH caused by mantle metasomatism. Due to the eruption of the kimberlite, water partitions prefentially into the kimberlitic melt, because due to the higher solubility of water in the melt compared to NAMS. The decrease of OH content in NAMS can thus be used, to quantify the composition of the kimberlitic melt. We present for the first time high resolution FT-IR based synchrotron measurements of OH and CO2 concentrations in eclogitic garnets from the Victor Diamond Mine in South Africa. We measured hydrogen and CO2 profiles towards totally embedded microcracks in these garnet crystals. The measurements in the studied garnets show a strong variation in OH and CO2 concentrations and demonstrate that the amount of water and carbondioxide stored in diamond bearing layers has been underestimated for a long time due to the loss of water and carbondioxide during the uplift of the kimberlite.