V51B-01 INVITED
On the Deformation of UHP Eclogite: From Laboratory to Nature
Eclogites from orogens often show strong plastic deformation together with the surrounding country rocks represented by felsic gneisses. Eclogite rheology is an important piece of information for understanding the subduction of continental crust to great depth and then return to the surface. In the past two decades, there have been numerous studies of naturally deformed eclogites using a combination of optical microscopy, scanning electron microscopy (SEM), electron back-scattered diffraction (EBSD) and transmission electron microscopy (TEM). Most of these studies suggest that omphacite accommodates most of the strain imposed on eclogite and develops strong lattice preferred orientations (LPOs); garnet crystals generally behave as rigid grains. However, there have been some recent controversies regarding the ductility of garnet and LPO development in omphacite. Natural garnets deformed by dislocation creep and by grain boundary processes have both been proposed based on a variety of observations and other evidence. The variation of S-type and L-type fabric in omphacite has been attributed to drastically different mechanisms, such as dislocation creep, diffusion-assisted grain growth and space-group transformation. We present here the comparison side by side of deformation microstructures, fabrics, and seismic properties between naturally deformed and experimentally deformed eclogites synthesized from the same natural rocks. Our results suggest remarkable similarities despite the large difference in conditions (e.g. strain rates, stress, temperature, etc.) responsible for producing them. We conclude that: 1) In dry eclogite, omphacites accommodate most of the deformation while garnets behave as essentially rigid bodies; 2) In wet eclogite, the deformation is accommodated by shape change in both garnet and omphacite; 3) In both cases, omphacites develop pronounced S-type or L-type microfabrics arising from dislocation creep and variation of the geometry and orientation of the finite strain ellipsoid; 4) Garnets develop no LPO in either case; 5) The seismic properties of eclogites are controlled by both garnet and omphacite but their anisotropy patterns of P-wave velocity are controlled only by the LPO of omphacite.
V51B-02 INVITED
Dehydration Embrittlement and the H2O Content of Subducting Lithosphere
Dehydration embrittlement has been demonstrated in a variety of hydrous phases and is strongly implicated as the trigger mechanism of intermediate-depth earthquakes (70-300km). Antigorite serpentine is capable of initiating such a shearing instability during dehydration at pressures from 0.1 to 6 GPa in the laboratory, a range over which the volume change accompanying dehydration changes from positive to negative, yet the shearing instability occurs under all conditions. Similarly, exsolution of very small quantities of H2O from nominally anhydrous phases can also trigger instability in the laboratory. It is thus highly likely that dehydration under stress of any reasonably abundant phase in subducting lithosphere would trigger earthquakes. Here I will use this logic to argue that subducting lithosphere is progressively "wrung dry" over the depth interval 100-400 km and that only very small amounts of H2O can exist in such lithosphere below that depth. The evidence is the following: (1) Earthquake frequency declines exponentially between 100 and 300 km (suggesting that the cause of the instability is being exhausted); (2) the resurgence of earthquakes in the transition zone could, in principle, be triggered by dehydration of dense hydrous magnesium silicates, the "alphabet phases", but the continuous production of earthquakes with a maximum at ~ 600 km and the sudden termination before 700 km is inconsistent with the conditions under which the "alphabet phases" exhibit mineral reactions (indicating that this mechanism is unlikely to be involved in deep earthquakes); (3) the pressure/temperature paths followed in subduction zones is such that each of the "alphabet phases" is replaced by another that has a higher H2O content (hence release of free H2O is highly unlikely); (4) the "alphabet phases" are only stable if the highly abundant phases wadsleyite and/or ringwoodite are fully saturated with H2O, which would require so much water that the lithosphere would have been so buoyant above 400 km that it would not have entered into the transition zone (hence saturation is extremely unlikely); (5) if significant H2O is present in ringwoodite, wherever lithosphere passes through into the lower mantle there should be an abundance of earthquakes as that H2O is released during ringwoodite breakdown (the lack of such earthquakes implies that even small amounts of H2O in ringwoodite are unlikely); (6) if H2O gets passed from one "alphabet phase" to another, there should be a flurry of earthquakes in the lower mantle during the dehydration of phase D, the last of these phases (such earthquakes are absent, strongly suggesting that phase D is also absent). The lack of earthquakes in these circumstances where they would be expected if H2O is significantly present implies that (i) dehydration embrittlement is at best a minor trigger of earthquakes in the mantle transition zone and (ii) subduction zones deeper than ~400 km are essentially dry. A corollary is that subduction does not significantly recycle H2O into the deep mantle, at least not at this time.
V51B-03 INVITED
Subducted lithosphere at the core-mantle boundary? Possible rheological and chemical consequences for the D" region.
Recent studies using seismic tomography datasets and horizontally polarized shear waves (e.g. Hutko et al., Nature, v. 441, 332-337, 2006) have suggested that not only is the core-mantle boundary a complex rheological and chemical environment, but conclude that subducted lithosphere may ultimately be emplaced in the lowermost mantle, the D" region. Analyses of these data sets suggest that this region is significantly dynamic in terms of temperature variations, associated up- and downwellings and overall rheologic behavior. Mineral physics studies of high pressure transitions have added to the discussion of the core mantle boundary (CMB) region in terms of first discovering the post-perovskite phase which may help our understanding of observed seismic anisotropy in addition to modifying chemistry so that the area becomes enriched in Fe which in turn influences heat flux mechanisms. Among many key areas for study, is investigating the rheological behavior of perovskite/post-pervoskite during shear deformation and the possibility that during shear events outer core material may be entrained into D". Downwelling cold slabs may drive deformation in D" as they impinge directly upon the CMB and can set up transient shear environments. Early experimental results on post-perovskite analog material (Yamazaki et al., EPSL v. 252, 372-378), have shown that post-perovskite may deform by dislocation, not diffusion, creep. In a set of deformation experiments under high pressure, olivine and FeS responded such that molten FeS migrated into zones of relatively lower pressure, driven by transient pressure gradients set up while the material undergoing shear. Continued experimental study at high pressure integrated with observational sets of data and numerical modeling is critical in furthering our understanding of how slab penetration into D" determines the rheological and chemical heterogeneity currently observed in the D" region.
V51B-04
Hydrogen Diffusion in Olivine: Implication for Point Defects Models in Olivine
The knowledge of hydrogen diffusion laws in olivine is essential to explain the zonations observed in some mantle xenoliths [1,2] and to understand the origin of hydrogen-enhanced electrical conductivity and plastic deformation in this mineral [3]. However, despite its importance to understand the mechanisms of hydrogen diffusion no study on the rate of isotopic diffusion of hydrogen is available [4]. Only hydrogen uptake experiments have been performed and were used to infer the diffusivities of hydrogen and of metal vacancies [5,6]. We present here the results of an isotope diffusion study (H-D exchange experiments) undertaken in natural iron-poor olivine single crystals. The results show that diffusion is highly anisotropic with diffusion along [100] almost one hundred times faster than along the two other crystallographic directions. The diffusion laws are more than thirty times slower than the previously measured rates of hydrogen uptake in San Carlos olivine [5]. These results disagree with earlier interpretations. A new point defect model is proposed and metal vacancies diffusivities in olivine are reassessed. [1] Demouchy, S. et al. (2006) Geology, 34, 429-432. [2] Peslier, A.H. and Luhr, J.F. (2006) Earth Planet. Sci. Lett. 242, 302-319. [3] Wang, D. et al. (2006) Nature 443, 977-980. [4] Ingrin, J. and Blanchard, M. (2006) Reviews in Mineralogy &. Geochemistry. 62, 291-320. [5] Kohlstedt, DL. and Mackwell, SJ. (1998) Z. Phys Chem., 207, 147-162. [6] Demouchy, S. and Mackwell SJ. (2006) Phys. Chem. Minerals, DOI 10.1007/s00269- 006-0081-2.
V51B-05 INVITED
Traces of H2O in Ultrahigh-Pressure Metamorphic Rocks
Ultrahigh-pressure (UHP) metamorphic rocks accommodate a significant amount of H2O at high pressures and temperatures during their deep subduction. Fluid-driven processes are responsible for mineral reactions; they may trigger phase transformations and provide a decisive weakening effect on the rheological behavior of the rocks during deep subduction, or they may lead to brittle failure and earthquakes. Dehydration reactions, producing fluid (i.e., H2O, CO2) during regional metamorphism at low-to-mid crustal levels (P ~ 0.1 - 1 GPa), are reasonably pressure insensitive. At these levels, the maximum dehydration occurs at the greatest temperatures experienced by buried rocks. In the deep subduction zone (> 120 - 150 km), where both high pressures and temperatures operate, the situation is different. There, on the one hand, the water becomes chemically bonded and incorporated into the structure of both nominally hydrous and anhydrous minerals. On the other hand, the aqueous fluid (a supercritical fluid) dissolves a considerable amount of mineral components at high pressures and temperatures, and a solute concentration increases as pressure is increased. Microstructural observations on ultrahigh-pressure minerals from eclogites and metasediments provide convincing evidence of fluid involvement as deep as the upper mantle and possibly, the mantle transition zone. Diamond is one of the minerals of great importance because it unambiguously records the high pressure (minimum 4 GPa and possibly > 4GPa) at which the host rocks were recrystallized. We present here the results of studies of nano-inclusions associated with dislocations of growth and/or with interstitial defects of carbon in diamond structure obtained with transmission electron microscopy, microRaman, and microInfrared synchrotron assisted spectroscopy. A diverse composition of multicomponent fluid and crystalline inclusions and characteristic of nitrogen aggregations, provide evidence that the diamonds were crystallized from a supercritical C-O-H fluid during a UHP metamorphism related to continental collision. These observations are also consistent with diamonds synthesized at high pressure and high temperature from graphite, amorphous carbon, and coal in the presence of H2O. The crustal signature of carbon isotopes (δ 13C) in the diamonds, together with their multiphase fluid- solid inclusions, provides evidence of a pathway by which organic carbon and H2O were subducted to the mantle depths and returned back to the Earth's surface. Microstructural patterns such as healed cracks, microfabrics, etc., observed in other minerals co-existing with diamonds, can therefore be recognized and attributed to UPH metamorphism events and may cast a light on the deformation and rheology of UHP metamorphic rocks.
V51B-06
Describing Ancient Forearcs Through the Study of Mafic and Ultramafic Metamorphic Rocks: a Case Study from the US Appalachians
When assessing the role of subduction in ancient orogenic rocks, metamorphism associated with collision can obscure original geochemical signatures. In the Blue Ridge province of the US Appalachians, small, discontinuous mafic and ultramafic rock exposures form a broad chain near the center of the belt, a map pattern that has led past workers to suggest they represent an ancient ophiolite (Misra and Keller 1978; Misra and McSween 1984). The focus of our 1997-2002 Blue Ridge REU Site Program was to examine the field relations, structure and geochemistry of a number of bodies in fine detail, toward understanding what these rocks represented in a convergent plate boundary setting of the late Precambrian. Our results indicate two classes of units: a) multi-lithologic bodies with contact relations and chemical signatures consistent with mafic and ultramafic cumulate rocks, and immobile trace element patterens indicating oceanic origins. These bodies are enclosed in pelitic metasediments and proximal to major faults or terrane boundaries, supporting an origin as fragments of a subducting oceanic plate. b) Podiform ultramafic and mafic bodies of varied size and lithologic complexity. The ultramafic bodies are dunites with minor pyroxenite that preserve U-shaped REE patterns and enriched isotopic signatures. The mafic rocks show major and immobile element systematics consistent with andesitic protoliths. These units occur in several Blue Ridge terranes characterized by block-in matrix rocks (Hatcher et al 2005; Raymond et al 1989). Relict eclogites point to deposition in melange sequences, suggesting that these bodies represent entrained crustal and mantle blocks from the overriding plate. While LILE patterns in such ancient rocks have been reset, their major and immobile trace element signatures are similar to those in modern subduction systems. Thus, subduction in the late Precambrian appears to have functioned much as it does today.
V51B-07
Strontium Isotopic Evidence for Episodic Discharge of Slab Fluids to Mud Volcanos in the Marianas Forearc
Subduction of the Pacific plate beneath the Mariana forearc releases hydrous fluids which move upward along fractures and react with fault gouge and fractured harzburgitic mantle rocks to produce serpentinite "mud" that discharges with cold water as springs on the ocean floor. This mud accumulates as seamounts, tens of km wide and up to two km high. As part of Leg 195 of the Ocean Drilling Program, we measured the isotopic composition of Sr in water squeezed from intervals of the cores, the serpentine mud, leaches of the serpentine mud, and in entrained harzburgitic clasts. Sr87/Sr86 in the pore fluids is constant at ~ 0.7056 except for just below the seawater-mud interface where a ratio of 0.7091 and Sr concentration of ~80 ug/L indicate seawater is the major pore fluid. Strontium concentrations abruptly decrease below this interface to ~6 umol/kg, and 87Sr/86Sr decreases to ~0.7056 through the remainder of the penetrated section except where seawater -precipitated aragonite has equilibrated with the rising fluids, raising their isotopic composition to near seawater values, and at one horizon where isotopic composition and strontium concentrations briefly rise to near seawater values. Because the upward flow of fluids is on the order of cm/yr, Sr isotopic equilibration between rising fluids and the carbonate precipitates must have occurred within a maximum of a few hundred years. In contrast, the strontium isotopic compositions of leached serpentine mud, and of harzburgite clasts entrained in the mud, is always significantly greater that that of the pore fluids. In harzburgite clasts the ratio is as great as 0.7088, almost as high as that of seawater. Whereas strontium in aragonite clearly equilibrated isotopically within a few hundred years, strontium in harzburgite clasts has not re-equilibrated with adjacent pore fluids in the same time. The harzburgite clasts and associated serpentine mud must have been near the sea floor, unburied, for a yet undetermined but much longer period of time to have equilibrated from ~ 0.704 to 0.709 prior to subsequent burial and variable equilibration with poer fluids. These variations in the strontium isotopic composition of solids and pore waters suggest episodic rather than steady state expulsion of fluids from the subduction zone . It may be possible to characterize at least the periodicity of fluid release in the mud volcano setting by investigating zonation of strontium isotopic composition of hartzburgite clasts throughout the 60 meter deep composite cores.
V51B-08
Izu Backarc Extension-related Basalts: The Importance of Accessory Phases in Island arc Magma Genesis
Studies of Izu's Oligocene-Miocene backarc and Quaternary volcanic front suggest that slab-derived components have been consistently partitioned between the rear arc that receives slab melts versus the volcanic front that receives only slab fluids. Th/La ratios increase whereas Ba/La ratios and Sr, Pb, Nd and Hf isotope ratios all decrease across the arc. The volcanic front mantle source is also more depleted than the mantle source of the rear arc volcanoes. More recent and widespread basaltic volcanism (3-0 Ma) is associated with “backarc” extension and rifting between the volcanic front and rear arc. The spatial and temporal variability of these basalts is used to constrain the sources and processes of magmagenesis. Extension-related backarc basalts have major element systematics that are MORB-like and similar to those of Manus and Lau"-two backarc basins with the fastest spreading rates and highest inferred mantle potential temperatures"-even though spreading is not yet occurring at Izu. REE patterns are also MORB-like. However, ratios of fluid-mobile elements to fluid- immobile elements (e.g., Ba/Nb) and Pb isotope ratios suggest variable but minor amounts of slab-derived fluid components. Nd and Hf isotope systematics and Hf*/Hf and Ce*/Ce ratios are consistent with some slab-derived sediment melt contribution but, if so, then Hf*/Hf, Zr/Hf, La/Nb, and Th/La ratios require trace accessory phases (e.g., ilmenite, rutile, and zircon) to be present in either the subducting slab or mantle wedge, or both. Distinguishing between pre-existing and subduction-related mantle heterogeneities is still problematic, given the small contributions of sediment inferred from slab-mantle mixing calculations, but trace accessory phases such as ilmenite would be required in the mantle wedge even if sediment is not the source of elevated REE/HFSE ratios.