T31F-01 INVITED 08:00h
The transition-zone water-filter model of mantle convection and chemistry
The transition-zone water-filter model of mantle convection was recently proposed (Bercovici and Karato, Nature v425, p39, 2003), to reconcile geochemical observations of isolated chemical mantle reservoirs (evident in distinct chemical signatures of ocean island and mid-ocean ridge basalts, OIB and MORB, respectively) with geophysical, especially seismological evidence for whole mantle circulation and mixing. The model proposes that as the background ambient mantle upwelling (rising in response to the downward flux of subducting slabs) moves out of the high-water-solubility transition zone into the low-solubility upper mantle above 410km, it undergoes water supersaturation and partial melting that filters out incompatible elements; the remaining solid phase continues to ascend and supplies relatively dry and depleted materials to the MORB source region. The residual melt is presumed denser than the surrounding solid and is thus trapped at the 410km boundary until slab entrainment recirculates the enriched material back into the deeper mantle. The filtering effect is suppressed for hotter mantle plumes, because of their greater ascent rates and lower water solubility in the transition zone, allowing plumes to generate more enriched OIBs. Simple calculations of elemental-vs-compatability spectra ("spidergrams") using the two-stage melting of ambient mantle (melting in the filter mechanism and at ridges) and single stage melting of plumes (at hotspots only) reproduce chemical observations for mid-ocean ridges and Hawaiian plume spectra. Differences in radiogenic isotope ratios between MORB and OIB are explained by different compatibilities of parent and daughter products (causing their different depletion at the filter zone) and a short separation in isotopic evolution; the necessary compatibilities and parent:daughter ratios to yield the observed isotopic composition is consistent with field and laboratory measurements. The model also predicts that radiogenic internal heating would also be non-uniform, i.e., largely sequestered in the sub-410km mantle and even in the transition zone itself. Numerical models of convection with such heat-source distributions show that no layering or implausible circulation is induced since all boundaries of the mantle are permeable. Various seismic studies over the last decade indicate the presence of a melt zone above the 410km discontinuity, which provides partial support for our hypothesis. Further geodynamical studies of the horizontal structure of the 410-km melt zone and mechanisms for recirculating water through the transition zone are underway and will be discussed.
T31F-02 08:15h
VOLATILE MANTLE UPWELLINGS, SUBDUCTION AND EMERGENCE OF PLATE TECTONICS
We propose a scenario below where the Earth has episodically degassed over its history with the release of hydrogen and associated siderophile elements into the lithosphere of the Earth. Tomographic images of deep upper-mantle low-velocity anomalies now give some first clues as to how and where this may happen. Surprisingly, low-velocity anomalies are found to as deep as 400 km and deeper beneath early Paleozoic suture zones such as the Tornquist-Tesseire Zone in Eurasia (Nolet and Zielhuis, 1994) and the Appalachians in North America (Van der Lee and Nolet, 1997). Such suture zones have no reason to be hot and the low velocities are more likely explained by increased volatile content associated with metasomatism of the deep upper mantle. Computations on the residence time of water in olivine suggest that this metasomatized upper mantle could presently be diffusing and/or welling upwards, in which case it would hydrate the shallow lithosphere from below. The vertical extent and lateral bending of the low-velocity anomaly beneath the Appalachians is suggestive of such an upwelling. We have used a 3-D visualization package (AMIRA) to delineate this feature which has important implications for the near-future evolution of the Atlantic North-America plate margin. We are presenting the hypothesis where plate tectonics is lubricated by wet upper mantle plumes and sheets. These sheets are attracted and advected by the strong guidance of the top layer, the lithosphere, in much the same way as the hot spots are rooted on the D$^{II}$-layer and held fixed by the lower mantle. In the Archean the young Earth is expected to have been hotter and richer in volatiles. While sheet like upwellings might be linked to the presence of some style of subduction (A), circular plume-like upwellings are a robust feature of plume tectonics (B). The early Earth must thus have gone through a convective fluid mechanical (A) and a plate tectonic solid mechanical mode (B). Self-organization of plate tectonics is suggested to have emerged out of both modes A and B in transition.
T31F-03 INVITED 08:30h
Seismic Evidence for Water Atop the Mediterranean Transition Zone
We present seismic evidence for up to 1000 wt ppm of water near depths of 400 km below the Mediterranean region. This water has possibly been carried there by subducting oceanic plates, which release most of the subducted water at depths shallower than 150 km into the mantle wedge overlying the plate but could carry small amounts to as deep as 400 km or below. The water causes the olivine to wadsleyite transition at 410 km to broaden to a 20-30 km thick interval (Wood, 1995). This broadening is detected by receiver functions that we constructed from seismograms recorded by permanent stations in the Mediterranean region as well as seismograms from our temporary MIDSEA array, which samples new locations in and south of the Mediterranean Sea. The receiver functions highlight S waves that converted from P waves at upper-mantle discontinuities. The S waves that converted at the 410-km discontinuity decrease significantly in amplitude with increasingly high frequency content of the receiver functions. The most likely explanation for this observation is a 410-km discontinuity that is thicker than the global average and reaches thicknesses of 20 to $>$30 km below multiple Mediterranean locations. Discontinuity relief and split discontinuities are less likely explanations for these observations than increased water content of olivine, because of their lateral incoherency. The Mediterranean transition zone is densely populated with subducted lithospheric fragments (Marone et al., 2004), which could have supplied the water.
http://www.sg.geophys.ethz.ch/midsea/
T31F-04 08:45h
Temporal Evolution of the Earth's Deep Water and Carbon Cycles
Most investigation techniques for determining the Earth's hydration state only allow to constrain the present-day conditions. However, to understand the geodynamical and geochemical evolution of the Earth, an approach needs to be taken that includes time. We address the question of how the Earth's hydration state and CO$_2$ content may have evolved through time by the using geodynamic models. These models solve for the two main processes that control the global water and carbon cycles: outgassing at mid-ocean ridges and hotspots and deep recycling at subduction zones. We have recently presented a possible scenario on how the Earth's deep water cycle may have evolved through time (R\"upke et al., 2004). In that scenario the Earth's mantle has been largely, but not completely outgassed ($\sim$93%). Due to slab recycling, it contains about $\sim$49% of the exosphere's water content and $\sim$33% of its initial water content, of which $\sim$80% has been recycled back from the exosphere. Here we present new results from a follow-up study in which we have linked and put into relation our predicted mantle water concentrations to geochemically determined water concentrations in OIB and MORB as recently determined by (Dixon et al., 2002). We find that in a plum-pudding type mantle, the most primitive components (FOZO) are the wettest; recycling associated components like the EM&HIMU components are much dryer but contain still more water than the dry depleted mantle source that melts beneath mid-ocean ridges to make MORB. These findings are in striking agreement with the geochemical data, so that our modelling approach appears to be a valid approximation to the evolution of the Earth's water cycle. Furthermore, we will present first results from new models that will solve for the global carbon cycle. These new models will use a 2-D subduction zone model that solves for decarbonification (and thereby deep CO$_2$ recycling) at subduction zones and a parameterized mantle convection model to solve for CO$_2$ outgassing at mid-ocean ridges and hotspots. This type of modelling approach can help to reconcile data from different fields of geosciences in order to better understand how the Earth's hydration and carbonification state may have evolved through time to form the present-day conditions.
T31F-05 INVITED 09:00h
Water contents in the mantle transition zone beneath the north Pacific derived from the electrical conductivity anomaly
In the former work, we estimated the three-dimensional electrical conductivity anomaly in the mantle transition zone beneath the north Pacific by inversion of semi-global electromagnetic network data. In this paper we try to convert the result to anomalies of temperature or water contents by combining it with results of recent seismic tomography (Fukao et al., 2004), and with reference to physical parameters determined by laboratory experiments of mantle materials. First we assumed that both electrical conductivity and seismic P-wave velocity anomalies are simply corresponding to thermal anomaly. Under this simple assumption, conductivity and velocity models were independently converted to temperature field by using the physical relationship derived from the laboratory measurements (Karato, 1993; Xu et al., 2000). Temperature field models from electrical and seismic data show similarities as a whole. For instance, high temperature anomaly of about 200K is seen beneath Hawaii. However, significant discrepancy is found beneath Mariana islands where the seismic tomography indicates low temperature anomaly, while electromagnetic data shows high temperature anomaly. Although there still remains a problem of spatial resolution, this result indicates that this particular feature may not be explained only by thermal effect. Taking into consideration that this region is surrounded by subducted slabs, we further assumed that this discrepancy is caused by the existence of water dehydrated from those slabs. Under this assumption, by combining the Nernst-Einstein relationship (e.g. Karato, 1990) and the recent result of laboratory measurements of hydrogen diffusivity in wadsleyite (Ohtani et al., in preparation), the water contents anomaly was estimated from the electrical conductivity anomalies. As a result, the mantle transition zone beneath Mariana islands turned out to contain about 0.2 weight % more water than surrounding mantle which is expressed by a reference structure.
T31F-06 09:15h
The Effects of Hydrogen on Electrical Conductivity in Wadsleyite and Ringwoodite: Implications for Hydrogen Content in the Mantle Transition Zone
The transition zone of Earth's mantle ($\sim$410 to 660 km depth) can contain a large amount of hydrogen (water) but the exact amount of hydrogen content in the mantle transition zone is unknown. Here we determined the effects of hydrogen on electrical conductivity in wadsleyite and ringwoodite under controlled chemical environment to infer the water content in the mantle transition zone. Synthetic polycrystalline samples with varying hydrogen contents were prepared and their electrical conductivity was measured under the transition conditions using an AC impedance method for 10$^{2}$ to 10$^{6}$ Hz. Hydrogen content of each sample was measured both before and after the conductivity measurement. The change in hydrogen content was relatively small (less than $\sim$20%). The electrical conductivity in wadsleyite is similar to that of ringwoodite and is given by $\sigma$=A$\cdot$C$^{r}_{H}$$\cdot$ exp(-H$^{*}$/RT) with A=0.1$\pm$0.1 (S/m), r=0.68$\pm$0.05 and H$^{*}$=87$\pm$3 (kJ/mol) for wadsleyite and with A=1.0$\pm$0.3 (S/m), r=0.69$\pm$0.03 and H$^{*}$=104$\pm$2 (kJ/mol) for ringwoodite, where T is temperature (K), R the gas constant, $\sigma$ the electrical conductivity (S/m) and C$_{H}$ the molar concentration of hydrogen (H/10$^{6}$Si). This relation suggests that the dominant charge carrier in these minerals under the experimental conditions is free proton. The activation enthalpy determined is relatively small and consequently the hydrogen content can be well constrained with relatively small uncertainties associated with the uncertainties in temperatures. A comparison with the geophysically inferred conductivity values shows that the water content in the mantle transition zone to be $\sim$0.2$\pm$0.1 wt$%$ for the acceptable range of temperatures from 1750 to 1950 K. These values of water content significantly exceed those in the upper mantle, suggesting that not all of the hydrogen is transported with upwelling current across the 410 km discontinuity.
T31F-07 09:30h
Water Transport into the Transition Zone and Lower Mantle by High Pressure Hydrous Phases in the Slabs
Water can be transported into the transition zone and lower mantle by the high pressure hydrous phases in the slabs. The candidates for the carriers of water are superhydrous phase B and phase D (= phase G) which are stable in the peridotite or harzburgite layer of the slabs. Phase Egg (AlSiO3OH) and delta-AlOOH are also possible candidates for water carriers in the crustal components in the slabs. We have determined the stability field of these hydrous phases expected in the transition zone and lower mantle. The stability fields of superhydrous phase B and phase Egg were studied up to the pressure of 30 GPa and the temperatures form 1000 to 1400 K by using the conventional quenching method together with the in situ X-ray diffraction study at high pressure using synchrotron radiation from PF and SPring 8. Phase Egg decomposes to delta-AlOOH and stishovite at the top of the lower mantle. The decomposition boundary of superhydrous phase B into phase D (G) + perovskite + periclase has a negative slope and can be expressed by the following equation, P(GPa)=33.2-0.0037T(K). We clarified the stability field of hydrous phase D (G) by using the laser heated diamond anvil cell with the synchrotron X-ray radiation at PF up to a pressure of 60 GPa and 1500 K. Natural orthopyroxenes (Mg#=92-93) were used for the starting material. Fine grained Pt powder was mixed with the pyroxene sample for absorption of the laser power for generation of high temperatures up to 1500 C. Phase D(G) is stable up to c.a. 45 GPa and 1500 K and 50 GPa at 1300 K with a negative dT/dP, The phase boundary is expressed as P(GPa)=75.2-0.021T(K) which is consistent with the previous result by Shieh (1998). These results suggest that there are major dehydration regions associated with the subducting slabs; i.e., the bottom of the transition zone and the deep lower mantle. The fluid generated by the decomposition of the hydrous phases might be responsible for the seismic reflectors observed in some depths in the lower mantle.
T31F-08 09:45h
Water in Mantle-Derived Melts: Constraints on the Mantle Water Cycle
I review mantle water concentrations derived from study of mantle-derived melts. Average MORB mantle contains about 120 ppm H$_{2}$O, with estimates ranging from about 50 (NMORB) to 450 (EMORB) ppm. Mantle sources of oceanic island basalts (OIB) typically have water concentrations (about 400 to 1000 ppm) greater than MORB. Deep mantle sources with water concentrations greater than 1000 ppm have not been found. The amount of residual water that survives the "subduction gauntlet" and is retained in the down-going slab to enter the deep mantle has until recently been quite uncertain. Measurements of volatile contents of OIB containing components of previously subducted lithosphere (EM-type) contain roughly half the water found in other OIB mantle sources (FOZO-type), indicating that water is efficiently extracted during the subduction process. Thus, water in the subducted slab must be expelled through fluid venting in the fore-arc, expelled during island arc magmatic processes, or sequestered in the hydrated, overlying mantle wedge. Effective partitioning of water into the exosphere results in depletion of water relative to other incompatible elements in the recycled lithosphere, accompanying continental-derived materials, and ultimately the mantle. Water concentrations in arc and back-arc lavas and in melt inclusion in phenocrysts confirm that the source regions for arc lavas are anomalously wet (from 250 ppm to 1 wt% H$_{2}$O) compared to MORB and OIB sources. The question to be answered is: If the transition zone contains from 0.5 to 1 wt% H$_{2}$O, as suggested by solubility experiments and seismic velocities, why don't we see any evidence for highly hydrous, deep mantle-derived melts?