Dense Hydrous Magnesium Silicate Research Articles

Crystal chemistry of dense hydrous magnesium silicates: The structure of phase H, MgSiH2O4, synthesized at 45 GPa and 1000 C

The crystal structure of the dense hydrous magnesium silicate phase H, MgSiH 2 O 4 , synthesized at 45 GPa and 1000 °C, was investigated by single-crystal X-ray diffraction. Although showing a deterioration process under the X-ray beam, the compound was found to be orthorhombic, space group Pnnm (CaCl 2 -type structure), with lattice parameters a = 4.733(2), b = 4.3250(10), c = 2.8420(10) A, V = 58.18(3) A 3 , and Z = 1. The structure was refined to R 1 = 0.0387 using 53 observed reflections [2σ( I ) level]. Magnesium and silicon were found to be disordered at the same octahedral site (with a mean bond distance of 1.957 A). Hydrogen was not located in the difference Fourier maps, but it is very likely disordered at a half-occupied 4 g position. The centrosymmetric nature of the structure of phase H is examined in relation to that reported for pure δ-AlOOH at ambient conditions (non-centrosymmetric, P 2 1 nm ), and the possibility that these two compounds can form a solid solution at least at high pressure is discussed.

Melting of phase D in the lower mantle and implications for recycling and storage of H2O in the deep mantle

We determined the melting phase relations and conditions of the dense hydrous magnesium silicate phase D (nominally MgSi2O4(OH)2) on composition in MgO–SiO2–H2O (MSH), MgO–Al2O3–SiO2–H2O (MASH), and FeO–MgO–Al2O3–SiO2–H2O (FMASH), and on a mixture of phase D+olivine+enstatite (MSH) at 22–32GPa and 1000–1800°C. Contrasting to previous studies, we performed H2O-undersaturated experiments. Bulk compositions were synthetic mixtures of brucite+silica or brucite+olivine+enstatite on the silica-rich side of the tie-line perovskite–H2O. At 22–24GPa, the maximum thermal stability of phase D is between 1350 and 1400°C in MSH and FMASH, but 1600°C at 24GPa in the Fe-free, Al-bearing bulk composition (MASH). Apparently, addition of Al2O3 increases the stability field of phase D by 200°C, an effect that is counter balanced by addition of FeO. At 32GPa, the stability of phase D (MSH and FMASH) is between 1350 and 1400°C. At 22GPa, phase D melts to a Mg-rich melt coexisting with MgSi-ilmenite+stishovite, whereas at 24–32GPa melt coexists with perovskite and stishovite. Even melts from bulk compositions in the silica-rich part of the MSH system (molar bulk Mg/Si<0.5) are magnesian-rich (Mg/Si molar ratio of 2–5) and are distinct from aqueous fluids and hydrous melts at lower pressures.The temperature stabilities determined in this study indicate that slabs that thermally relax when stagnating on top of the 660-km discontinuity or penetrating into the lower mantle will have their last dense hydrous magnesium silicate phase, i.e., phase D, melting and producing a magnesian and hydrous melt that will rise through the transition zone. Such a melt could be responsible for observed low velocity zones, and may be neutrally buoyant at the 410-km discontinuity and will affect the structure and dynamics of the mantle.

Stability and crystal chemistry of iron-bearing dense hydrous magnesium silicates

Dense hydrous magnesium silicates (DHMS) are supposed to be key phases in planetary water cycles because of their ability to carry water to deep mantle regions in subduction slab environments. In order to understand water cycles in iron-enriched planetary systems such as Mars knowledge of the water content and stability of iron-bearing DHMS is required. Iron-bearing DHMS were synthesized based on two starting compositions, MgFeSiO4+9.5wt% H2O system and a simple hydrous Martian mantle composition containing Fe, Mg, Al and Si+12.35wt% H2O (hydrous FMAS system). Compared to literature data on phase D, iron-bearing phase D shows analogous variations in water contents as Mg-phase D but appears to be stable at higher temperatures than Mg-phase D for both starting compositions used in this study. Iron-bearing superhydrous phase B contains up to 7wt% H2O and shows an extended thermal stability in the hydrous FMAS system. The high-temperature stability of iron-bearing DHMS with a Mars-like bulk composition indicates that these hydrous phases could host significant amounts of water at core-mantle boundary conditions (1500°C and 23GPa) in a hydrous Martian mantle.

Transmission electron microscopy characterization of the dislocations and slip systems of the dense hydrous magnesium silicate superhydrous B

Superhydrous B (shy B) is a dense hydrous magnesium silicate (DHMS) phase, for which experimental studies have indicated a large stability field along cold subduction geotherms in the Earth’s transition zone and top region of the lower mantle. The rheological properties of shy B may therefore be relevant to the rheology and possibly the deep seismic characteristics of subduction zones. Samples of shy B have been synthesized and deformed in a multianvil apparatus at 20 GPa and at temperatures 1000–1100 °C. The resulting dislocations have been characterized by transmission electron microscopy (TEM) with the weak-beam dark-field (WBDF) and the large-angle convergent-beam electron diffraction (LACBED) techniques. We emphasize the role of (010) plane in the plasticity of phase shy B. We report that [100] and [001] glide in (010), and also climb of [100]. We also observe 〈101〉 glide in (010) and {111} planes under high stress conditions.

First principles prediction of a new high‐pressure phase of dense hydrous magnesium silicates in the lower mantle

First principles calculations have shown that the high‐pressure phase transition of phase D MgSi2O6H2 to the new high‐pressure hydrous phase MgSiO4H2 plus stishovite occurs at about 40 GPa. The crystal structure of the new phase is very similar to that of CaCl2‐type SiO2 and has monoclinic symmetry with space group P2/m. This new hydrous phase is stable up to about 52 GPa and then dissociates into MgSiO3 perovskite and H2O (ice VIII) at static 0 K conditions. The Clapeyron slope between phase D and the new phase + SiO2 is positive; about 6.4 MPa/K at 1000 K. This new hydrous phase further extends the stability field of dense hydrous magnesium silicates and also opens the pathway toward the existence of a high‐pressure phase of ice at the lower mantle conditions.

Electrical conductivity of dense hydrous magnesium silicates with implication for conductivity in the stagnant slab

Electrical conductivities of dense hydrous magnesium silicates (DHMS), phase A, super-hydrous phase B (SuB) and phase D, were measured by means of impedance spectroscopy in the frequency range of 10−1–106Hz at temperatures up to 775, 700 and 700K and pressures of 10, 18 and 22GPa, respectively. For all phases, the increase in electrical conductivity (σ) with temperature follows the Arrhenian formula: σ=σ0exp(−(ΔH/kT)). The pre-exponential factors (σ0) and activation enthalpies (ΔH) of phase A, SuB and phase D yield values of 7.28±0.82S/m and 0.77±0.01eV, 292±48S/m and 0.83±0.01eV and 1342±154 and 0.75±0.01eV, respectively. Higher pressure DHMS phases show higher conductivity values. The electrical conductivities of phase D and super hydrous phase B are about two and one orders of magnitude higher than that of phase A in the same temperature range, respectively. Although the proton conduction is considered to be a dominant mechanism, there is no clear relationship between water content and conductivity. Rather the conductivity of DHMS phase is closely related to the O⋯O distance. The conductivity–depth profiles for a cold subduction zone were constructed based on the phase proportion predicted in the descending slab. The results show distinctly lower conductivity values than those geophysically observed beneath the northeastern China and the Philippine Sea, where the cold slab stagnates in the mantle transition zone. Consequently, the DHMS phases themselves cannot be a main contributor to enhance the conductivity in the stagnant slab. Dehydration of the stagnant slab would strongly enhance the conductivities in the transition zone beneath northeastern China and Philippine Sea.

Crystal defects in dense hydrous magnesium silicate phase A deformed at high pressure: characterization by transmission electron microscopy

Dense hydrous magnesium silicate (DHMS) Phase A has been deformed at 11 GPa, 700 and 400 °C in the multianvil apparatus. Transmission electron microscopy (TEM) characterizations, using the weak-beam dark-field (WBDF) technique, have shown that at 700 °C, two different types of perfect dislocations are activated with ⅓〈2110〉 and ⅓〈2113〉 Burgers vectors. The ⅓〈2110〉 dislocations glide in the basal plane, but for ⅓〈2113〉 dislocations, no glide plane could be identified. At 400 °C, dissociation of ⅓〈2110〉 dislocations is observed in the basal plane. Furthermore, prismatic and pyramidal glide planes were identified, with dissociations in pyramidal planes, at this temperature. Also a growth reticular merohedral twin has been observed; the twin law results from a rotation of 180° around the [1540] axis.

The 3.65 A phase in the system MgO-SiO2-H2O: Synthesis, composition, and structure

The 3.65 A phase, a member of the family of dense hydrous magnesium silicates (DHMS) in the system MgO-SiO 2 -H 2 O, was synthesized in a 77 h multi-anvil press experiment at conditions of 10 GPa and 425 °C by using a gel of composition MgSiO 3 plus water in excess as starting materials. From our multi-methodical study including SEM, TEM, EMP, IR, and Raman analysis, we determined the composition of the 3.65 A phase to be MgSi(OH) 6 . Powder XRD combined with Rietveld refinement revealed the 3.65 A phase to be isostructural with δ-Al(OH) 3 . The 3.65 A phase can be described as a hydrous A-site vacant perovskite with probably long-range random distribution of Si and Mg at octahedral sites. Locally, some ordering of Mg and Si might exist, as indicated from the spectroscopic measurements. The 3.65 A phase represents the second DHMS with exclusively sixfold-coordinated Si, the other being phase D. The 3.65 A phase is stable at pressures above about 9.0 GPa and temperatures below 500 °C. This limited P-T stability together with its high water content makes the 3.65 A phase an unrealistic mantle component. If at all, it might only occur under hydrous conditions in the coldest parts of deeply and extremely fast subducted old oceanic slabs.

Seismic evidence of negligible water carried below 400-km depth in subducting lithosphere

Strong evidence exists that water is carried from the surface into the upper mantle by hydrous minerals in the uppermost 10-12 km of subducting lithosphere, and more water may be added as the lithosphere bends and goes downwards. Significant amounts of that water are released as the lithosphere heats up, triggering earthquakes and fluxing arc volcanism. In addition, there is experimental evidence for high solubility of water in olivine, the most abundant mineral in the upper mantle, for even higher solubility in olivine's high-pressure polymorphs, wadsleyite and ringwoodite, and for the existence of dense hydrous magnesium silicates that potentially could carry water well into the lower mantle (deeper than 1,000 km). Here we compare experimental and seismic evidence to test whether patterns of seismicity and the stabilities of these potentially relevant hydrous phases are consistent with a wet lithosphere. We show that there is nearly a one-to-one correlation between dehydration of minerals and seismicity at depths less than about 250 km, and conclude that the dehydration of minerals is the trigger of instability that leads to seismicity. At greater depths, however, we find no correlation between occurrences of earthquakes and depths where breakdown of hydrous phases is expected. Lastly, we note that there is compelling evidence for the existence of metastable olivine (which, if present, can explain the distribution of deep-focus earthquakes) west of and within the subducting Tonga slab and also in three other subduction zones, despite metastable olivine being incompatible with even extremely small amounts of water (of the order of 100 p.p.m. by weight). We conclude that subducting slabs are essentially dry at depths below 400 km and thus do not provide a pathway for significant amounts of water to enter the mantle transition zone or the lower mantle.

The structure of a super-aluminous version of the dense hydrous-magnesium silicate phase D

The dense hydrous-magnesium silicate phase D, which has the ideal formula MgSi2H2O6, may be an important link in a chain of hydrous phases that carry H2O in the ultramafic portions of subducting lithosphere, into the Earth’s lower mantle. We have synthesized a new Al-rich form of phase D, containing up to 50 wt% Al2O3, using a multi-anvil device at ~1300 °C and 25 GPa. The phase, with the formula Mg0.2Fe0.15Al1.8H1.8SiO6, was initially produced in a bulk composition designed to synthesize Al- and Fe-rich magnesium silicate perovskite with a composition similar to that produced in experiments on mid-ocean ridge basalt bulk compositions at lower mantle conditions. Further experiments using a starting mixture based on the composition of this Al-rich phase resulted in the synthesis of 60–70 μm long single crystals at similar conditions. The recovered crystals were slightly richer in H2O (Mg0.2Fe0.12Al1.5Si0.92H3.1O6) and their unit-cell parameters were similar to those of MgSi2H2O6 phase D. A refinement of the crystal structure was carried out in the P 31 m space group and revealed a more disordered cation distribution than magnesium silicate phase D. All cation-oxygen distances are similar, suggesting a high degree of Si/Al disorder. Although the stability field of this new variant of phase D is yet to be determined, this phase may be an important host for H2O within portions of subducted oceanic crust in the lower mantle.

Investigation of structure and hydrogen bonding of superhydrous phase B (HT) under pressure using first-principles density functional calculations

High-pressure behaviour of superhydrous phase B (high temperature; HT) of Mg10Si3O14(OH)4 (Shy B) is investigated with the help of density functional theory-based first-principles calculations. In addition to the lattice parameters and equation of state, we use these calculations to determine the positional parameters of atoms as a function of pressure. Our results show that the compression induced structural changes involve cooperative distortions in the full geometry of the hydrogen bonds. The bond-bending mechanism proposed by Hofmeister et al. (Vibrational spectra of dense hydrous magnesium silicates at high pressure: Importance of the hydrogen bond angle, Am. Miner. 84 (1999), pp. 454–464) for hydrogen bonds to relieve the heightened repulsion due to short H- - -H contacts is not found to be effective in Shy B. The calculated O–H bond contraction is consistent with the observed blue shift in the stretching frequency of the hydrogen bond. These results establish that one can use first-principles calculations to obtain reliable insights into the pressure-induced bonding changes of complex minerals.

Sound velocity measurement in liquid water up to 25 GPa and 900 K: Implications for densities of water at lower mantle conditions

We extended the pressure range of sound velocity measurements for liquid water to 25 GPa and 900 K along the melting curve using a laser heated diamond anvil cell with a combined system of Brillouin scattering and synchrotron X-ray diffraction. Experimental pressure and temperature were obtained by solving simultaneous equations: the melting curve of ice and the equation of state for gold. The sound velocities obtained in liquid water at high pressures and melting temperatures were converted to density using Murnaghan's equation of state by fitting a parameter of the pressure derivative of bulk modulus at 1 GPa. The results are in good agreement with the values predicted by a previously reported equation of state for water based on sound velocity measurements. The equation of state for water obtained in this study could be applicable to water released by dehydration reactions of dense hydrous magnesium silicate phases in cold subducting slabs at lower mantle conditions, although the validity of Murnaghan's equation of state for water should be evaluated in a wider pressure and temperature ranges. The present velocity data provides the basis for future improvement of the accurate thermodynamic model for water at high pressures.

High-pressure infrared spectroscopy of the dense hydrous magnesium silicates phase D and phase E

Phases D and E are hydrous magnesium silicates that are stable at high-pressure–temperature conditions and could serve as H 2O carriers in the Earth's mantle, especially in subducting slabs. Using synchrotron infrared (IR) spectroscopy, we measured the infrared spectra of polycrystalline samples of phases D and E to 42 and 41 GPa, respectively. For both phases, at least three broad OH stretch vibrations were observed at elevated pressures indicating that each phase has multiple hydrogen positions that exhibit disorder. No structural phase transition or amorphization was observed for either phase over the measured pressure range. The mode Grüneisen parameters of phases D and E are in the range of −0.12 to 1.14 and −0.17 to 0.83, respectively, with mean values of 0.41 (phase D) and 0.31 (phase E). Using empirical correlations of OH frequency and O⋯H and O⋯O bond lengths; the six OH vibrations of phase D at ambient pressure have corresponding O⋯H and O⋯O bond distances in the range of 1.519–1.946 Å and 2.225–2.817 Å, whereas the four OH vibrations of phase E have the corresponding O⋯H and O⋯O bond distances in the range of 1.572–2.693 Å and 2.557–2.986 Å. These ranges encompass values reported from single-crystal X-ray diffraction measurements. At high pressures, the observable OH stretching vibrations exhibit both positive and negative pressure slopes. Our high-pressure infrared spectra for phase D do not support the occurrence of hydrogen symmetrization as predicted by first-principles calculations.

高圧含水マグネシウムケイ酸塩D相及びsuperhydrous B相の構造:1・2次元 1 H、 29 Si NMRによる情報

高圧含水マグネシウムケイ酸塩D相及びsuperhydrous B相の構造:1・2次元 1 H、 29 Si NMRによる情報

Sound velocities and elasticity of DHMS phase A to high pressure and implications for seismic velocities and anisotropy in subducted slabs

Dense hydrous magnesium silicate (DHMS) phase A forms in cold subducted slabs after the breakdown of antigorite serpentine, and may play an important role in the transport of water within the upper mantle. In this paper we present acoustic velocities and the single-crystal elastic properties of Fe-bearing phase A, (Mg 0.981Fe 0.019) 7Si 2O 8(OH) 6, measured by Brillouin spectroscopy on a sample compressed to 12.4(2) GPa in a diamond anvil cell. A fit to the acoustic data using a 3rd order finite-strain EOS yields the following adiabatic bulk ( K S) and shear ( μ) moduli and their pressure derivatives: K S = 106(1) GPa, (∂ K S/∂ P) T0 = 5.8(3), μ = 61(1) GPa, (∂ μ/∂ P) T0 = 1.8(1). Within the experimental resolution, the pure longitudinal elastic constants, C 11 and C 33, and the off-diagonal C 12 constants exhibit positive linear pressure dependence, whereas C 44, C 66 and C 13 increase with a quadratic dependence on pressure. The axial compressibility of phase A remains highly anisotropic in the investigated pressure range, with the a-axis being 15% more compressible than the c-axis at 12.4(2) GPa. Compared to forsterite, the aggregate compressional ( V P) and shear ( V S) acoustic velocities of phase A are 7% slower at room pressure. Although the velocity contrast diminishes to 3.5% for V P, it is maintained for V S over the investigated pressure range. Phase A has high shear wave anisotropy ( A S) and shear-wave polarization anisotropy ( A S P 0 ) of A S = 20% and A S P 0 = 18 % , and a more moderate compressional wave anisotropy A P = 12% at room pressure. The A P of phase A decreases to 8% at 12.4(2) GPa, remaining significantly lower than that of forsterite, whereas the shear anisotropy is nearly constant at ∼20% over the same pressure range and exceeds that of forsterite by 12% at 12.4(2) GPa. At upper mantle pressures, the shear wave splitting in phase A ( A S P 0 ) is 20% higher than in forsterite. The results of this study were used with thermoelastic data for other relevant minerals to compute the density, seismic velocities and V P/ V S ratios of subducted garnet–harzbugite and moderately depleted harzburgite assemblages with various degrees of hydration as a function of pressure along a slab isotherm at 1073 K. The results suggest that the seismic velocities of dry and water-saturated harzburgites (44.5 vol% phase A) may be indistinguishable at upper mantle P–T conditions because of the increasing concentration of high-pressure orthopyroxene upon hydration. This phase displays high seismic velocities that offset the decrease in velocities due to phase A, rendering hydration difficult to detect (anelastic attenuation is not considered). Combined observations from the analysis of seismic parameters indicate that significant shear wave anisotropy, accompanied by high V P/ V S and Poisson's ratios and pronounced shear wave splitting, could be major diagnostic features for identifying phase A-bearing assemblages at depth (180–350 km) in cold subducted slabs.

Dense hydrous magnesium silicates, phase D, and superhydrous B: New structural constraints from one- and two-dimensional 29Si and 1H NMR

To gain new structural insights into phase D and superhydrous B, two phases of potential mantle water reservoir, we have applied a range of one- (1D) and two-dimensional (2D) 1H and 29Si NMR techniques, as well as Raman spectroscopy, to samples synthesized at 24 GPa and 900~1100 °C. These data have revealed that phase D is characterized by disordered and varying local structures around both Si and H. The 29Si NMR spectra of phase D contain a nearly symmetric, broad peak near -177.7 ppm, attributable to octahedral Si with local structural disorder. The high-resolution 1H CRAMPS spectra of phase D contain a main broad peak near 12.6 ppm with shoulders near 10 and 7 ppm, suggesting a distribution of hydrogen bonding distances. For superhydrous B, our comprehensive 2D 1H and 29Si NMR results have clearly revealed that it contains dissimilar hydrogen (H1-H2) pairs and one tetrahedral Si site, consistent with space group Pnn2.

Synchrotron X-ray analysis of norbergite, Mg2.98Fe0.01Ti0.02Si0.99O4(OH0.31F1.69) structure at high pressure up to 8.2 GPa

The natural norbergite, Mg2.98Fe0.01Ti0.02Si0.99O4(OH0.31F1.69) is examined by synchrotron X-ray diffraction analysis at pressures up to 8.2 GPa. The measured linear compressibilities of the crystallographic axes are βa = 2.18(4) × 10−3, βb = 2.93(7) × 10−3, and βc = 2.77(7) × 10−3 (GPa−1), respectively and the calculated isothermal bulk modulus of the norbergite is KT = 113(2) GPa based on the Birch–Murnaghan equation of state assuming a pressure derivative of K′ = 4. The crystal structures of norbergite are refined at room temperature and pressures of 4.7, 6.3, and 8.2 GPa, yielding R values for the structure refinements of 4.6, 5.3, and 5.3%, respectively. The bulk moduli of the polyhedral sites are 293(15) GPa for the tetrahedron, 106(5) GPa for the M2 octahedron, 113(2) GPa for the M3 octahedron, and 113(3) GPa for the total void space. The bulk modulus exhibits a good linear correlation with the filling factor for polyhedral sites in structures of the humite minerals and forsterite, reflecting the Si4+ + 4O2− ⇔ □ + 4(OH, F)− substitution in the humite minerals. Moreover, two simply linear trends were observed in the relationship between bulk modulus and packing index for natural minerals and dense hydrous magnesium silicate minerals. This relationship would reflect that the differences in compression mechanism were involved with hydrogen bonding in these minerals.

Nanoinclusions of high-pressure hydrous silicate, Mg3Si4O10(OH)2 · nH2O (10Å-phase), in mantle olivine: Mechanisms of formation and transformation

The 10A-phase, Mg3Si4O10(OH)2 · nH2O, where n = 0.65÷2, belongs to the group of dense hydrous magnesium silicates (DHMS), which were produced in experiments and are regarded as hypothetical mineral carriers for H2O in the mantle. However, DHMS were almost never observed in nature. The only exception is the finding of the 10A-phase as nanoinclusions in olivines from mantle nodules in kimberlites. The inclusions with sizes of a few ten nanometers have a pseudohexagonal habit and are characterized by the presence of voids free of solids. The 10A-phase fills the equatorial parts of the inclusions, and, in the majority of inclusions, it is replaced by the low-pressure serpentine + talc assemblage. Based on the analysis of electron microscope images, a model was proposed for the solid-state formation of inclusions, the precursory material of which was transformed to the 10A-phase with the liberation of a water fluid. According to this model, the formation of hydrous nanoinclusions and their subsequent autoserpentinization occurred without the influx of H2O from the external medium through the mobilization of intrinsic hydroxyl-bearing point defects trapped during olivine crystallization. The subsequent autoserpentinization of the inclusions occurred during decompression owing to interaction between the inclusion material and the host olivine matrix. The process was accompanied by the partial exhaustion of the fluid phase and the replacement 10A-phase + H2O = Serp + Tc. The criterion for the credibility of the model is the conservation of the volume of material during the reaction at P = const and T = const.

Predicted elastic properties of the hydrous D phase at mantle pressures: Implications for the anisotropy of subducted slabs near 670-km discontinuity and in the lower mantle

The dense hydrous magnesium silicate called the D phase is the most likely candidate to recycle hydrogen into the lower mantle in subduction zones. As seismology represents the preferred method to detect this mineral in subduction zones, the single crystal elastic tensor has been calculated using first principle methods. A triple cell ( P 3 ¯ m 1 ) was used to account for an ordered ideal composition of the D phase to lower mantle pressures. The elastic tensor of the mineral at ambient conditions is predicted to be C 11 = 387.7, C 33 = 287.7, C 44 = 100.4, C 12 = 108.0, C 13 = 51.1, C 14 = − 14.6 in GPa with bulk modulus at zero pressure K o = 163 GPa and a statistical error of about 4.0 GPa for all constants. The pressure derivatives of the elastic constants have been determined to pressure of 80 GPa. Our tensor correctly predicts the published experimental measurements of linear compressibility along the a- and c-axes, the logarithmic pressure derivative of the c/ a axial ratio and bulk modulus at zero pressure. It contradicts the experimentally measured elastic tensor reported by (Liu, L.-G., Okamoto, K., Yang, Y.-J., Chen, C.-C., Lin, C.-C., 2004. Elasticity of single-crystal phase D (a dense hydrous magnesium silicate) by Brillouin spectroscopy. Solid State Commun. 132, 517–520), which fails to reproduce the experimental compressibility data. Using the elastic constants and density predicted to 85 GPa, the seismic anisotropy of P- and S-waves has been calculated as a function of pressure. The single crystal has high P-wave velocities and shear wave splitting in the basal plane below 20 GPa. At high pressures, the maximum P-wave velocities have threefold symmetry normal to a-axes in the basal plane, whereas the maximum shear wave splitting is parallel to a-axes with twofold symmetry. The D phase has high Vp and Vs anisotropy at ambient pressure, 17.6% for Vp and 19.9% for Vs. The Vp anisotropy decreases with increasing pressure to about 30 GPa, while it remains constant at 8% from 30 to 85 GPa. The Vs anisotropy decreases to 17% at 20 GPa and then increases to 22% at 85 GPa. The isotropic velocity ratios of S to P ( Rs/ p), bulk sound to S ( Rϕ/ s) and density to S ( Rρ/ s) are calculated as a function of pressure. The Rs/ p shows very little variation with pressure with low value of 0.6, whereas both Rϕ/ s and Rρ/ s increase with pressure to peak values of 2.8 and 2.0 respectively at about 60 GPa. The presence of the D phase in subducted horizontal ‘stagnant’ plates in the circum Pacific could contribute to a number of seismic observations, such as the observed velocity heterogeneity at depths between 500 and 1000 km and the fast S-waves travelling horizontally with horizontal polarization in this depth range. Further the lower density of the D phase compared with anhydrous minerals could influence buoyancy of the hydrated ‘stagnant’ slabs. The transformation of the D phase to perovskite plus ferripericlase during thermal annealing may be responsible for horizontal slab seismogenesis.

The composition of liquids coexisting with dense hydrous magnesium silicates at 11–13.5 GPa and the endpoints of the solidi in the MgO–SiO 2–H 2O system

High-pressure liquids in the MgO–SiO 2–H 2O (MSH) system have been investigated at 11 and 13.5 GPa and between 1000 and 1350 °C. A bulk composition more magnesian than the tie-line forsterite–H 2O was employed for the study. Rocking multi-anvil experiments were combined with a diamond trap set-up. After termination of the experiments, the liquid trapped in the diamond layer was analysed by laser ablation ICP-MS using the ‘freezing’ technique. At 11 GPa, liquids coexist with one or two of phase A, clinohumite, chondrodite, and forsterite. A marked discontinuity in the evolution of liquid compositions near 1100 °C is observed at 11 GPa. A step of ∼13 wt% H 2O and 13 wt% MgO is interpreted to result from overstepping the fluid-saturated solidus reaction mass balanced to 1.00(18) phase A + 1.07(4) fluid = 0.63(15) chondrodite + 1.44(2) melt. At 13.5 GPa liquids coexist with one or two of hydrous wadsleyite, clinohumite, superhydrous B, phase B, and forsterite. The discontinuity in liquid composition is no longer present, indicating that the second critical endpoint of the solidus has been overstepped. Thus, hydrous melts in the Mg-rich part of the MSH system (molar bulk Mg/Si > 2) are chemically distinct from aqueous fluids at pressure up to 11 GPa. Convergence of fluid and melt compositions along the solidus resulting in a supercritical liquid occurs between 11 and 13.5 GPa, at which pressure the entire MSH system becomes supercritical.