DI41B-01 INVITED
The Strong Effect of H2O On Olivine Transformation Kinetics Suggests That Some Subducting Slabs Are Dry
Because of the high solubilities of hydrogen in nominally anhydrous mantle and transition-zone minerals, it is possible that much of Earth's H2O resides in the upper mantle and transition zone. If Earth's mantle is part of its water cycle, subduction zones would provide the vehicle for transporting H2O from Earth's surface to the transition zone. Seismic and petrologic evidence has been used to argue that the mantle portion of subducting oceanic lithosphere can be hydrated, which would allow hydrous olivine to be transported to the transition zone. In fast and cold subducting slabs, where temperatures may be low enough to inhibit equilibrium transformation of olivine, a wedge of metastable olivine is thought to extend deep into Earth's transition zone. Olivine transformational faulting in such a metastable wedge provides an explanation for deep focus earthquakes. Previous kinetic experiments show that large amounts of H2O in olivine enhance growth rates in the olivine-wadsleyite and olivine-ringwoodite phase transformations. Here we present new experimental results on the transformation of hydrous (100 and 300 wt-ppm H2O) San Carlos olivine to ringwoodite at 18 GPa. Our experiments show that small amounts of H2O enhance ringwoodite growth rates and greatly decrease activation enthalpies relative to anhydrous samples. Even at temperatures as low as 700 °C, we obtain measurable growth rates, implying rapid ringwoodite growth at temperatures corresponding to the interiors of cold subducting slabs. Combining our kinetic data with thermal modeling of subduction zones shows that even for very old and fast subduction zones, small amounts of H2O would eliminate the metastable olivine wedge. Seismic evidence for a metastable olivine wedge in the Mariana subduction zone suggest that the olivine in this slab has less than 100 wt-ppm H2O. If deep-focus earthquakes are caused by olivine transformational faulting, then subduction zones that exhibit these earthquakes must be nearly dry.
DI41B-02 INVITED
Melting in the Deep Upper Mantle Oceanward of the Honshu Slab
Melting in the Deep Upper Mantle Oceanward of the Honshu Slab Utilizing P-wave tomography, Obayashi et al. (2006) discovered a low velocity zone (LVZ) oceanward of the northern Honshu slab. They concluded that the LVZ was due to a 200 °C thermal/chemical anomaly in combination with either a localized increase in iron content or a 50 km thick layer of partial melt above the 410-km discontinuity. We re-examine the region studied by Obayashi et al. (2006) using multiple ScS reverberations, SH- polarized shear phases that are reflected off of the core-mantle boundary and discontinuities within the mantle. Shear waves are more sensitive to changes in shear strength than compressional waves, making them well suited for distinguishing between diffuse thermal/chemical anomalies and regions of partial melt. Our paths sample the region studied by Obayashi et al. (2006) as closely as allowed by the differing source-receiver geometries required by the two different methods. The region where Obayashi et al. (2006) observed the LVZ is well sampled by our data and our results are similar to theirs, with the exception of the region directly east of Taiwan. The Hales, Gutenberg, Lehmann, and 520-km discontinuities are found in many of the paths at various depths. All of the paths contain the 410-km and 660-km discontinuities. Three of the paths on the oceanward side of the Honshu slab contain a LVZ above the 410-km discontinuity, with an average depth of 356 km and a thickness that ranges from 50-75 km. The top of the LVZ is marked by an impedance decrease of -1.2 to -2.0 %. We interpret the LVZ as a layer of negatively buoyant partial melt produced by the combined effects of water and temperature, and estimate the temperature anomaly to be 155 °C with a corresponding maximum water content of 0.100 wt %. If water content varies with depth, some melts may be buoyant and others dense, helping to explain both the LVZ at depth and the surface volcanism in the area. Obayashi, M., Sugioka, H., Yoshimitsu, J., Fukao, Y., 2006. High temperature anomalies oceanward of subducting slabs at the 410-km discontinuity. Earth and Planetary Science Letters. 243, 149-158.
DI41B-03
Ubiquitous Low-Velocity Layer Atop the 410-km Discontinuity Beneath the Northern Rocky Mountains
The 410-water-filter model is a provocative new model which predicts that a spatially localized layer of hydrous partial may pool atop the 410-km discontinuity ("410") as a result of the difference in water solubility between the mantle transition zone and the overlying upper mantle [Bercovici and Karato, 2003]. If the transition zone is sufficiently hydrated then upwelling mantle crosses the wet-olivine solidus and produces a hydrous partial melt. The density of this partial may be intermediate to that of the uppermost transition zone and deep upper mantle causing the melt to pond atop the 410. The model suggests that decoupling of incompatible elements into the melt dynamically creates MORB and OIB geochemical signatures obviating the need for pre-existing chemical reservoirs. Details of the 410-water-filter and its potential operation are the source of much debate and research [Hirschmann et al., 2006; Karato et al., 2006; Leahy and Bercovici, 2007]. However, a small number of seismic observations detect a low-velocity layer above the 410 and are interpreted as a manifestation of the 410-water filter [Song & Helmberger, 2004; Gao et al., 2006; Vinnik et al., 2003, 2004]. More recently a ubiquitous low velocity layer was detected with P-S receiver functions atop the 410 beneath three dense PASSCAL arrays in the northern Rocky Mountains, with the data requiring a 22-km thick layer with an 8.9% shear velocity reduction and sharp (<6.4 km) top gradient [Jasbinsek and Dueker, 2007]. Here we report observations of a similar low velocity layer atop the 410 beneath the RISTRA and CD-ROM arrays in the southern Rocky Mountains. Velocity modeling with this data does not significantly differ from previous results, implying a coherent structure to the melt layer atop the 410 beneath the Rocky Mountains.
DI41B-04 INVITED
Static Compression of Hydrous Silicate Melts and Density Crossovers in the Mantle
High pressure experiments using the sink/float method have bracketed the density of hydrous iron-rich ultrabasic silicate melt from 1.35 to 10.0 GPa at temperatures from 1400 to 1860C. The silicate melt composition was a 50- 50 mixture of natural komatiite and synthetic fayalite. Water was added in the form of brucite Mg(OH)2 and was present in the experimental run products at 2 wt% and 5 wt% levels as confirmed by microprobe analyses of total oxygen. Samples were contained in compression-sealed molybdenum capsules. Sink/float marker spheres implemented were gem quality synthetic forsterite (Fo100), San Carlos olivine (Fo91), and several types of natural pyrope-almandine garnet crystals with compositions in the range Py74-62. Experimental run times were 30 seconds, thus minimizing sphere-liquid reactions and liquid reaction with capsule and pressure media. 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 density of the silicate melt with 5 wt% water at 2 GPa and 1500C is 0.192 g/cc less than the anhydrous form of this melt at the same P and T. This density difference gives a partial molar volume of water in silicate melt of approximately 7 cc/mol, which is similar to previous studies at high pressure. However, much work is still needed to determine the effect of pressure and composition on the partial molar volume of water in magma. Future studies should require precise density measurements of the same melt composition with and without water, observing numerous sink/float brackets, over a wide pressure range. The komatiite-fayalite liquids with 0 and 2 wt% H2O, have extrapolated density crossovers with equilibrium liquidus olivine at 8 and 9 GPa respectively, but there is no crossover for the liquid with 5 wt% H2O. These results are consistent with the hypothesis that dense hydrous melts could be gravitationally stable atop the 410 km discontinuity in the Earth. The results also support the notion that equilibrium liquidus olivine could float in a FeO- rich hydrous martian magma ocean. Extrapolation of the data suggests that FeO-rich hydrous melt could be negatively buoyant in the Earth's D-double-prime region or atop the core-mantle-boundary (CMB), although experiments at higher pressure are needed to confirm this prediction.
DI41B-05
Stability of a Compressible Hydrous Melt Layer Above the Transition Zone
The transition zone water-filter model (Bercovici and Karato, Nature, 2003) provides a possible explanation for the conflicting geochemical and geophysical evidence about layering and chemical heterogeneity in the mantle. The model proposes that ambient upwelling mantle undergoes dehydration melting upon leaving the highly water- soluble transition zone, and the resulting melt filters out incompatible elements leaving a relatively dry and depleted source region for MORBs It is crucial to the model that the melt phase be denser than the solid phase such that it becomes trapped above the 410-km boundary where it can be entrained by slabs and recirculated into the deeper mantle. Because the melt phase is significantly more compressible than the solid phase it is expected that a density crossover occurs with increasing pressure. There is evidence to suggest this density crossover exists above the transition zone and as such the melt phase will be more dense than the solid. However, the depth of the density crossover is not well constrained. Thus, we investigate the stability of a compressible melt layer which intersects the density crossover. Analytic models of Rayleigh-Taylor type instabilities are used to determine the effects of compressibility and crossover location on the instability growth rates. Subsequently, more realistic models incorporating continual injection of new melt from below and entrainment of melt by slabs are employed to determine the overall effect on layer stability and evolution that results from an intrinsically unstable segment at the melt layer's top.
DI41B-06
Water Induced Instabilities at the top of Stagnant Slabs in the Transition Zone: Context and Consequences
Water enters the Earth's mantle at trenches by subduction of oceanic lithosphere. Most of this water immediately returns to the atmosphere through arc volcanism, but a part of it, retained in Dense Hydrous Magnesium Silicates (DHMSs) and Nominally Anhydrous Minerals (NAMs) like olivine, is expected as deep as the mantle transition zone (410-660 km depth). There, slabs can be deflected and linger before sinking into the lower mantle. Because it lowers the density and viscosity of mantle minerals in the transition zone, water is likely to affect the dynamics of the stagnant slab. The consequence of water's presence on the dehydration of a stagnant slab is explored. In particular, we focus on the possible onset of small-scale convective instabilities despite the 'adverse' thermal gradient (i.e., mantle above the floating slab is cooled from below). The competition between the thermal and hydrous effects on the density, and thus on the convective stability of the top part of the slab, is investigated using a numerical model including water dependent density and viscosity and temperature dependent water-solubility. For relatively high initial water content in the floating slab (≥ 1wt%), small-scale convection is likely to occur. Other important controlling parameters are the water dependence of density and viscosity. If small scale convection occurs at the top of a stagnant slab it enhances the rate of slab dehydration (otherwise controlled by the slow diffusion of water) and provides an efficient way to heat up the slab. Model results suggest that young and wet stagnant slabs are unlikely to reach the lower mantle because they would thermally equilibrate relatively quickly with the surrounding transition-zone mantle.
DI41B-07
Carbonated Eclogite Solidus Between 14 and 20 GPa: Results from the Model CMAS-CO2 System and Contrasting Solidus Behavior to Carbonated Peridotite
The carbonate ledge at ~2.0 GPa is a pronounced feature of the carbonated peridotite solidus. At the ledge, where the CO2-bearing phase changes from vapour to carbonate, the melt composition becomes carbonatitic. After this drop, the solidus of carbonated peridotite gradually rises in P-T space, up to at least 12 GPa. Between 14 and 16 GPa, Keshav et al. (2007) reported another drop in the solidus of carbonated peridotite in the model CMS-CO2 system. Similar to the lower-pressure topology, the solidus at higher pressure resumes a positive slope between 16-20 GPa, and seems to flatten between 22 and 26 GPa. Concomitant with this second drop, the melts become extremely calcic (Ca/Ca+Mg, Ca no.-0.62) at 16 and 20 GPa, but attain more magnesio-carbonatitic (Ca no.-0.40) character both at shallower or greater depths than the transition zone. Clearly, the second drop in the carbonated peridotite solidus has tremendous consequences for geological processes in the deep mantle. The other major rock-type presumed to be present in the mantle is eclogite of broadly basaltic composition. Clarifying the solidus topology of carbonated eclogite in model systems over a similar pressure range is also an important task, because the solidus topology affects the fate of subducted carbonate in the deeper mantle. The position of the solidus of carbonated eclogite will address its impact on local or extensive melting (if it occurs), the possible relationship between the carbonated peridotite and carbonated eclogite solidi at these depths (400-600 km), their respective incipient melts, and ultimately the possibility of carbonate survival at these and greater depths. With these issues in mind, we have determined the solidus of model carbonated eclogite in model CMAS- CO2 system between 14 and 20 GPa. At 14 and 16 GPa, the melts are in equilibrium with cpx, majoritic garnet, stishovite, and magnesite. At 20 GPa, the melts are in equilibrium with calcium-perovskite (capv), garnet, stishovite, and magnesite. From average calculated melting reactions along these isobarically univariant curves, stishovite is produced upon melting at all pressures investigated. Significantly, cpx at 14 and 16 GPa and capv at 20 GPa are the dominant contributors toward melt production/composition, in contrast to lower pressures (3-8 GPa) where carbonate dominantly contributes toward melt generation/composition at the solidus. The solidus of model carbonated eclogite at 14, 16, and 20 GPa, lies at 1350, 1450, and 1600 degrees C, respectively, and is nearly linear in P-T space. Melts in equilibrium with all the crystalline phases are highly calcic (Ca no.-0.70), resembling calcio-carbonatites. When magnesite is exhausted from the crystalline assemblage, the melts become slightly less calcic (Ca no.-65). The model carbonated eclogite solidus is always lower than the model carbonated peridotite solidus in the same pressure range. The most remarkable feature of this work is the absence of a drop in the solidus of model carbonated eclogite between 14 and 16 GPa, a result that is in stark contrast to that observed for the model carbonated peridotite at identical pressures. Therefore, even though the solidus temperatures in both carbonated peridotite and eclogite are strongly influenced by the presence of crystalline carbonate, melt compositions and the shape of the solidus in the pressure range investigated seem to be dominantly controlled by the silicate component of the rock in question. Given these results, it is fair to say that a wide range of petrological and geochemical processes operate at these depths in the mantle, and that we have barely scratched the surface in our investigation.
DI41B-08
Ca(Ti,Si)O3 Diamond Inclusions Crystallized From Carbonate Melts in the Transition Zone: Experimental Constraints
Composite diamond inclusions consisting of coexisting endmember CaSiO3 and CaTiO3 are rare but occur in diamond populations from Juina, Brazil1-2. Phase relations show that above ~9 GPa (at 1500 K) a perovskite-structured solid solution exists between these endmembers, while at lower pressures intermediate compositions produce coexisting CaTiO3-perovskite and CaSiO3 in the walstromite structure3. Inclusions with `perovskite' stoichiometry are commonly interpreted as fragments of solid mantle from the transition zone or lower mantle4-6. Here we report on two composite diamond inclusions from Juina kimberlite, and can effectively eliminate a subsolidus origin on the basis of experimental mineral phase relations. Instead, based on new melting experiments we find that the inclusions most likely crystallized directly from Ca-rich carbonate melts. Like other workers1-2 we interpret the composite inclusions as exsolution products of a high-pressure Ca(Ti,Si)O3 perovskite stable in the transition zone. Our bulk inclusion compositions are estimated to contain 50- 65 mol% CaTiO3, and are remarkably low in MgSiO3 component at less than 0.2 mol%. Experiments have shown that in peridotite or eclogite lithologies, Ca-rich perovskite in equilibrium with an MgSiO3-phase (majorite or Mg-perovskite) have about 3 to 7 mol% MgSiO37-8. Here we report on new subsolidus laser-heated diamond anvil cell experiments at 20-50 GPa in the ternary system CaSiO3-CaTiO3-MgSiO3 that bracket the CaTi-rich limb of the solvus between Ca- and Mg-rich perovskites. All experiments were made at 2000 (±200) K for 45-75 min, and were analysed using synchrotron micro-focus X-ray diffraction. We find that the solubility of MgSiO3 in CaTi-perovskite solid solutions increases significantly with increasing CaTiO3 component. Thus, Ti-rich calcium perovskite in peridotite or eclogite lithologies should have very high, not exceptionally low, MgSiO3 component. Accordingly, a subsolidus paragenesis is unlikely for the Juina inclusions, suggesting that melt may be involved in their formation. The MgSiO3-content of calcium perovskite in nominally volatile-free melting experiments on natural peridotite and eclogite compositions8-9 is also much higher than that of the Juina inclusions (>3 mol%). We have made melting experiments on model carbonated peridotite (CMS-Ti- CO2) and eclogite (CMAS-Ti-CO2) compositions at 20 GPa using multi-anvil techniques at the Bayerisches Geoinstitut. We find that liquidus Ca(Ti,Si)-perovskite coexisting with Ca-carbonatite melt is remarkably depleted in MgSiO3 component, and that the Ti-rich diamond inclusion compositions are well-matched by perovskite crystallized from melt derived from a model eclogite source. We suggest that diamond and CaTi-perovskite crystallized syngenetically from a melt derived from carbonated eclogite in the transition zone.1. Hayman, Kopylova, & Kaminsky, CMP 149, 430-445 (2005). 2. Kaminsky et al., CMP 140, 734-753 (2001). 3. Kubo, Suzuki, & Akaogi, PCM 24, 488-494 (1997). 4. Brenker et al., EPSL 236, 579-587 (2005). Harte et al., GS Spec. Pub. #6, 125-153 (1999). 6. Stachel et al, CMP 140, 16-27 (2000). 7. Hirose et al.. Nature 397, 53-56 (1999). 8. Irifune & Ringwood, EPSL 117, 101-110 (1993). 9. Corgne et al., GCA 69, 485-496 (2005).