V24A-01
New Insights Into the Heat Sources of Mantle Plumes, or: Where Does all the Heat Come From, Heat Producing Elements, Advective or Conductive Heat Flow?
Melting anomalies in the Earth's upper mantle have often been attributed to the presence of mantle plumes that may originate in the lower mantle, possibly from the core-mantle boundary. Globally, mantle plumes exhibit a large range in buoyancy flux that which is proportional to their temperature and volume. Plumes with higher buoyancy fluxes should have higher temperatures and experience higher degrees of partial melting. Excess heat in mantle plumes could reflect either a) an enrichment of the heat producing elements (HPE: U, Th, K) in their mantle source leading to an increase of heat production by radioactive decay or b) advective or conductive heat transport across the core-mantle boundary. The advective transport of heat may result in a physical contribution of material from the core to the lower mantle. If core material is incorporated into the lower mantle, mantle plumes with a higher buoyancy flux should have higher core tracers, e.g. increased 186Os and Fe concentrations. Geophysical and dynamic modelling indicate that at least Afar, Easter, Hawaii, Louisville and Samoa may all originate at the core-mantle boundary. These plumes encompass the whole range of known buoyancy fluxes from 1.2 Mgs -1(Afar) to 6.5 Mgs -1 (Hawaii) providing evidence that the buoyancy flux is largely independent of other geophysical parameters. In an effort to explore whether the heat producing elements are the cause of excess heat we looked for correlations between fractionation corrected concentrations of the HPE and buoyancy flux. Our results suggest that there is no correlation between HPE concentrations and buoyancy flux (with and without an additional correction for variable degrees of partial melting). As anticipated, K, Th and U are positively correlated with each other (e.g. Hawaii, Iceland and Galapagos have significantly lower concentrations than e.g. Tristan da Cunha, the Canary Islands and the Azores). We also find no correlation between currently available Fe concentration data and buoyancy flux. The apparent lack of correlation suggests that excess heat may be a result of conductive heat contribution from the core. Additional precise 186Os and Fe data are needed to further assess these conclusions.
V24A-02
Effect of Water Activity on Mineral Reaction Overstepping at High P and T: Experimental Approach and Petrological Implications
It has long been considered, on the basis of laboratory experiments, that mineral transformation kinetics are high compared to the rates of P-T changes responsible for metamorphic processes. Natural examples, however, clearly show that average mineral transformation kinetics are much slower than expected. One way to reconcile natural and experimental data is to consider fluid availability (especially water) as the limiting factor. There is a growing number of field and geophysical evidences for prograde metamorphic reaction overstepping, especially with respect to eclogite formation which is of importance for the Earth's dynamics due to the associated density changes. The role of water availability on reaction overstepping in metamorphic processes is a challenging experimental field. Reaction overstepping means that nucleation is delayed and therefore the role of water upon passing mineral nucleation barriers is the issue to be addressed. A way to favor, experimentally, nucleation over growth is to investigate systems which are far from equilibrium. Using the MAX80 cubic multi-anvil press installed on the German synchrotron (HASYLAB-DESY, Hamburg), we have monitored using in-situ X-ray diffraction, the progress of reaction: Ca(OH)2 + MgCO3 = CaCO3 + Mg(OH)2, at 1.7 GPa for temperatures below 600°C (i.e., in the field of aragonite). These experiments were performed either under controlled dry conditions (low H2O activity) or under the intrinsec humidity of the pressure assembly; the Ca(OH)2 + MgCO3 starting material, composed of fine powder, being initially submitted to the ambient air moisture. Under these pressure and temperature conditions, the free energy of this exchange reaction is of -30 to -35 kJ.mol.-1. Additional experiments were performed under excess water in a piston-cylinder apparatus (starting material sealed together with water in a gold capsule) at 1.7 GPa and 150°C for five different run durations. For all experiments, the plot of the reaction progress as a function of time displays shapes which are typical of nucleation-dominated reactions. In order to attain the same nucleation kinetics under dry and excess-water conditions, two highly contrasted temperatures are required (550 and 150°C, respectively). Furthermore, under dry conditions, Arrhenius plot shows a high activation energy of ca. 570 kJ.mol-1 (in the 550 - 600°C range), in line with the notion of overstepping.
V24A-03
Fast Kinetics of Dehydration of Hydrous minerals at Subduction Zone Conditions
In a number of subduction zones, earthquakes are not located on a single upper Benioff plane, but on two dipping planes separated by 20-40 km, defining a double seismic zone (DSZ). The likelihood origin for those earthquakes is hydraulic embrittlement, due fluids released by dehydration reactions. Such a model relies on the coincidence between the loci of the dehydration reaction and hypocenters, on the rheology of the rocks, and on their dehydration rate, which needs to be fast enough to locally raise fluid pressure. From phase equilibrium experiments, antigorite, talc, chlorite and 10Å phase appear as likely candidate minerals, since their dehydration occur also at P-T conditions compatible with those of the DSZ earthquakes. However, little was known about the speed of those reactions. Hence, we have measured the kinetics of dehydration of antigorite, talc, and 10Å phase, under H2O unsaturated and saturated conditions, in order to decide whether their dehydration might trigger earthquakes in subduction zones. Measurements were done by real-time in situ X-ray diffraction at HP and HT in the Paris-Edinburgh press at the European Synchrotron Radiation Facility (ESRF, ID27), with time resolution shorter than 1min. Hydrous minerals were loaded in a cold-sealed double wall capsule of h-BN outside, and Ti inside closed by a thin Au foil at both ends. They were pressurized, and then subsequently heated in the Paris-Edinburgh press. P and T conditions were calculated from the unit-cell parameters of Au and NaCl, according to the cross- calibration method. While approaching the dehydration reactions, X-ray diffraction pattern were acquired and inspected for new diffraction lines to identify the onset of reaction. As dehydration began, P-T conditions were held constant, until transformation was significant, if not completed. The extent of transformation as a function of time was deduced from the relative intensity of the diffraction peaks of the reaction products vs the reactants. The transformation-time data were analysed in terms of the Avrami's theory. Grain size and shape were inferred from SEM observations, on the recovered samples. This allowed to calculate the rate of release of aqueous fluids by the dehydration of the various minerals as a function of depth, to compare dehydration and viscous deformation rates, and to discuss the possibility of fluid embrittlement.
V24A-04
The Effect of ~290 ppmw H2O on the Rate of Olivine-Ringwoodite Transformation at 18 GPa
It has been suggested that in the interior of cold subducting lithosphere, the transformation of olivine to its high- pressure polymorphs, wadsleyite and ringwoodite, may be kinetically inhibited, leading to the subduction of metastable olivine deep into the mantle transition zone. The rapid transformation of this metastable olivine has been identified as a possible origin for deep-focus earthquakes. In the coldest subduction zones, olivine is predicted to transform directly to ringwoodite. Converging lines of evidence from seismology, mineral physics, and petrology suggest that subducting mantle olivine may be partially hydrated. This research investigates the effect of H2O on the kinetics of the olivine to ringwoodite transformation. We used a two-part methodology to study the transformation of hydrated olivine crystals to ringwoodite: a hydration experiment, in which olivine spheroids are hydrated in water during piston-cylinder experiments; and a multi-anvil experiment that partially transforms two of these spheroids in each run. Benefits of this methodology include the ability to consistently and uniformly hydrate olivine for numerous transformation experiments, greater potential to control hydrogen content, and increased opportunity for pre-transformation characterization. Transformation of hydrated olivine was performed in the ringwoodite stability field at 18 GPa and temperatures of 700oC, 900oC, and 1100oC for various run durations. The addition of approximately 290 ppmw H2O to olivine is shown to promote its transformation to ringwoodite by enhancing growth rates, lowering the activation enthalpy for growth, and weakening the ringwoodite to allow for dissipation of elastic strain energy sufficient to maintain constant growth rates throughout transformation. SIMS analyses of partially transformed samples indicate that hydrogen diffuses into the ringwoodite rim during transformation, progressively depleting the olivine center of hydrogen during transformation. Olivine in partially transformed samples contain as little as 50 ppmw H2O without a corresponding decrease in transformation rate, indicating that between approximately 50 and 290 ppmw H2O, transformation rate is independent of hydrogen content. This suggests that extrapolations of kinetic data from more hydrated samples underestimate the effect of 50 ppmw H2O and that hydrogen contents consistent with values observed in mantle-derived olivine are sufficient to promote olivine to ringwoodite transformation in the transition zone.
V24A-05
Looking for Critical Endpoints in Aqueous Fluids and Rocks: Input and Then Separation of Supercritical Fluids Underneath Volcanic Arcs
Aqueous fluids dissolve significant amounts of silicates under high-T and high-P condition. Silicate components dissolved in aqueous fluids coexisting with mantle peridotite change their major element chemistry from andesitic at 1-2 GPa to peridotitic at 3 GPa and higher pressures (Ayers et al., 1997, EPSL; Stalder et al., 2001, CMP; Mibe et al., 2002, GCA; Kawamoto et al., 2004, AM). Observations of unmixing and mixing between aqueous fluid and silicate melts by use of synchrotron X-ray radiography with multi-anvil type high-T and high-P apparatus at SPring-8 allow us to estimate PT conditions of critical endpoint, where critical temperature becomes equal to H2O-saturated solidus temperature (Mibe et al., 2004, GCA). We observe aqueous fluid and a silicate melt coexisting in pressures lower than a pressure, while above the pressure, we observe only one fluid phase, and then interpret it to be the pressure of a critical endpoint. We found a systematic increase in pressure of critical endpoint with decreasing SiO2 concentrations: sediment (63 SiO2 wt. %, 2.6 GPa), high-Mg-andesite (60 SiO2 wt. %, 2.9 GPa), MORB (50 SiO2 wt. %, 3 GPa, Mibe et al., 2007 under review), and peridotite (44 SiO2 wt. %, 3.8 GPa, Mibe et al., 2007 JGR). The chemistry of silicate components in aqueous fluids derived from downgoing slabs can vary from silica-rich in sediment and basalt layers to magnesium-rich in peridotite layers. The present set of data of critical endpoint pressures covers a wide range of chemical compositions and suggests that slab-derived fluids should be under supercritical conditions at the downgoing slabs beneath the volcanic arcs (Tatsumi and Eggins, 1995, Blackwell). This means a continuous change from hydrous melts to aqueous fluids at the base of mantle wedge underneath volcanic arcs. Whether the slab-derived fluids have chemical characteristics like a partial melt or an aqueous fluid depends on the temperature; slab derived-supercritical fluids in relatively warm regions can dissolve more silicate components than slab derived-supercritical fluids in relatively cold regions. The melt-like supercritical fluid formed at the base of the warm mantle wedge will separate into a melt phase and a fluid phase when the supercritical fluid meets its critical curve during its ascent. Partitioning of elements between aqueous fluids and hydrous silicate melts should play a significant role when the supercritical fluid would separate into them. Adakites are characterized by high Sr/Y ratio and low Y, which can be formed by garnet residue through a partial melting of downgoing basaltic crust (Defant and Drummond, 1990, Nature). Partitioning data between alkali chloride fluids and an andesite melt (Keppler, 1996, Nature) indicate high Sr/REE in alkali chloride fluids at 0.3-2 GPa. If this is the case in the nature, Sr/Y of fluids becomes higher than a coexisting silicate melts, and Y remains to be low. Adakites can be produced not by partial melting but by fluids-induced melting; large chemical variations observed in adakites with respect to Sr/Y and Y can be formed by the separation of supercritical fluids into fluids and silicate melts.
V24A-06
Water content of magma generated just above the 410 km seismic discontinuity
Seismological observations have revealed the existence of low velocity and high attenuation zones just above the 410 km seismic discontinuity in the Earth's mantle. It has been suggested that the existence of a small amount of melt could be responsible for such anomalies. The density of silicate melt under dry conditions has been measured under high pressure and found to be denser than the surrounding mantle, thereby allowing the melt to exist at this depth. However unless water exists, it is impossible to melt the mantle in the average temperature of ~1400°C at 410 km. Sakamaki et al. (2006) examined the density of hydrous magma, and claimed that the hydrous magma may be gravitationally stable in the case that the water content of the magma is less than 6.7 ± 0.6 wt% H2O. Nevertheless, the actual water content generated in this condition in the mantle is still unknown. Therefore we have determined the water content in the hydrous melt generated in the condition of 410 km depth at various temperatures. Two starting materials, pyrolite-2.9 wt% H2O and 8.3 wt% H2O, were used. The MA-8 type high pressure apparatus was adopted in the high pressure and high temperature experiments. The experiments were carried out at pressure of 13.5 GPa and temperatures from 900°C to 1600°C. Micro-Raman spectroscopy and EPMA were used for identification and chemical analysis of the recovered samples. The amounts of melt were determined from the cross sections of the recovered samples. We found that clear melt segregations occurred above 1400°C in the 2.9 wt% H2O system and above 1100°C in the 8.3 wt% H2O system. The amount of melt increased with increasing water content and temperature, from ~6% at 1400°C to ~44% at 1500°C in 2.9 wt% H2O system, and from ~30% at 1100°C via ~63% at 1400°C to ~95% at 1600°C. One of the most important results is that the water content of magma is more than 10 wt% H2O at 1400°C. Therefore, it may be difficult that such hydrous melt is gravitationally stable just above the 410 km seismic discontinuity.
V24A-07
Defining the chemical role of H2O in mantle melts: Effect of melt composition and H2O content on the activity of SiO2
Quantifying the influence of volatiles (H2O, CO2) on the chemistry of mantle melts is a critical aspect of understanding the petrogenesis of arc magmas. A significant amount of experimental work done on the effect of H2O on the solidii of various mantle compositions, as well as on multiple saturation points of various primitive melts, has shown that H2O stabilizes olivine with respect to orthopyroxene. Or, in other words, at constant activity of SiO2, the presence of H2O decreases the activity coefficient of SiO2 in the melt, potentially leading to mantle melts that have suprisingly high SiO2 contents (Carmichael, 2002). Quantification and modelling of this behavior in hydrous silicate melts in equilibrium with the mantle have proven problematic, due mainly to a relatively small set of experiments that allow this type of thermodynamic analysis, and because of the experimental and analytical difficulties of dealing with hydrous high P-T samples (e.g. quench to a glass, rapid melt-solid reaction on quench, electron beam sensitivity of resulting glass, volatile content determination, etc). A further complication in the existing data includes co-variance of important experimental parameters (e.g. T and H2O content), making robust statistical regression analysis difficult and potentially misleading. We present here results of high P-T experiments conducted at a single pressure and temperature (1.0 GPa, 1200 deg C) that have the specific goal of quantifying the effect of H2O, as well as other melt components, on the activity coefficient of SiO2 in mantle melts. Using a "sandwich" type experiment, basaltic melts are saturated with an olivine plus orthopyroxene mineral assemblage with varying H2O and CO2 contents. The resulting samples have their bulk solid phase and glass compositions determined using EPMA, and the volatile content of the glass is determined by FTIR. The activity of SiO2 is then calculated using the olivine and orthopyroxene compositions. This value is then used, along with the mole fraction of SiO2 that is measured in the glass, to calculate an activity coefficient for SiO2 in that particular melt. The results show that for two starting compositions, H2O clearly has a strong negative effect on the activity coefficient of SiO2, consistent with some earlier intepretations. Further work is being conducted on differing starting compositions, as well as increasing the range of volatile contents, in order to better quantify their influence on this important chemical parameter of mantle melts. Ultimately, these experiments will help determine whether hydrous arc lavas, including high-Mg andesites, can be attributed to a primitive mantle origin, or whether other magmatic processes are necessary to generate their observed bulk compositions. It will also quantify the amount of H2O necessary to generate such magmas, giving insight into the potential H2O content present in the sub-arc mantle source regions, and allowing a more precise estimate of volatile fluxes in volcanic arc settings.
V24A-08 INVITED
Generation of High-Silica Melts From the Mantle: Effects of Alkalis and Water
It seems to be generally accepted that addition of water to mantle compositions has a dramatic effect on the compositions of the silicate melts produced. Thus, for example, experiments at close to H2O saturation in the 70's demonstrated expansion of the liquidus field of olivine and implied that liquids of "andesitic" composition could be generated by partial melting of water-saturated peridotite. At that, time, however, experimental difficulties cast doubt on some of the conclusions. As part of a study of high Mg-andesites from arc settings, we have re-visited the question of the effects of P, T, H2O (and other minor components) on the compositions of mantle melts. We began with dry lherzolite melting in the systems CMAS, NCMAS and natural bulk compositions. Simple and systematic variations in SiO2 and MgO contents of lherzolite-saturated melts with pressure and alkali content can be observed. Specifically, SiO2 concentration increases with decreasing pressure and increasing total alkalis while MgO shows exactly opposite behavior. The effects of H2O, at least up to ~10 weight per cent are much less obvious. There is a slight depression of MgO content and, surprisingly, little systematic effect on SiO2. It appears, therefore that, in the concentration range of relevance to arc settings, H2O may not be particularly important in mantle melting except for its effect on liquidus temperature. We then turned to the question of how to generate high Mg-andesites which precipitate olivines of Fo90 composition if water isn't important. Harzburgitic rather than lherzolitic residue is the most obvious answer. Loss of clinopyroxene from the residue means that Ca and Al contents of the melts decrease while Si and Mg increase. In terms of oxides we find that MgO and SiO2 increase in the approximate ratio 3:1 after cpx is lost. Thus a "typical" high Mg-andesite with 55 SiO2, 10 MgO and 3.5 per cent total alkali oxide could be generated at around 1 GPa without water in equilibrium with a harzburgite residue.