V31F-01 INVITED
Experimental Constraints on Trace Element Partitioning During Dehydration and Melting of K-free MORB at 4-6 GPa and 700-1200 0C
Many of the distinctive chemical characteristics of island arc lavas are attributed to the melt generation in mantle source regions metasomatized by aqueous fluids and/or hydrous melts released from the oceanic subducting slab. Constraints on the composition and nature of the liquid phase (aqueous fluid, hydrous melt, supercritical liquid) during the dehydration of the subducting slab are critical in determining the partitioning behavior of key trace elements. We have experimentally determined the major and trace element composition and nature of liquids in equilibrium with K-free eclogite at 4-6 GPa and 700-1200°C. Two techniques were used. Diamond-trap, fluid-saturated experiments were conducted using the rocking multi-anvil to enhance homogeneity in the experimental charge. After the termination of the experiment, the liquid phase was analyzed using LA-IPC-MS while freezing the capsule to prevent fractionation of the liquid during preparation of the capsule. The occurrence of both aqueous fluid (at lower temperatures) and hydrous melt (higher temperatures) at 4 and 5 GPa as opposed to a single supercritical liquid at 6 GPa is reflected in the trace element partitioning behavior. Aqueous fluids are characterized by higher compatibility of U than Th in the fluid, and moderately steep REE patterns. Uranium is more compatible in an eclogitic mineralogy in equilibrium with hydrous melts, which also exhibit a steeper REE pattern. Supercritical liquids display a melt-like signature at all temperatures. Calculated lattice strain model parameters of garnet and cpx are linked to the nature of the coexisting liquid phase: the size of the optimal cation entering the X-site of garnet is larger when garnet coexists with a fluid compared to garnet coexisting with a hydrous melt. At 6 GPa, garnet is in equilibrium with a supercritical liquid and has similar characteristics to that in equilibrium with a hydrous melt. The changes in the size of the optimal cation entering the X-site in the garnet affects its partitioning coefficient, Do(3+). The Do(3+) values are nearly an order of magnitude higher in garnets coexisting with fluid at lower temperature compared to those coexisting with hydrous melts at higher temperature and supercritical liquids at all temperatures.
V31F-02 INVITED
HFSE Processing During Subduction and the Consequences for Nb/Ta and Zr/Hf Ratios in the Mantle
High-precision (MC-ICP-MS) Nb-Ta concentration ratios in Silicate Earth reservoirs (mantle and crust) are consistently sub-chondritic (<19.9; Münker et al., 2003). Various models have been proposed to explain this observation and include hidden reservoirs in the silicate Earth or Nb fractionation into the metal core. Nb becomes siderophile at high pressure and thus the core is a potential reservoir for the missing Nb (Wade & Wood, 2001). This model implies Nb depletion of the silicate portion of the Earth soon after, or even during accretion by a selective, pressure driven partitioning of Nb into the metal phase. As a consequence the bulk-silicate Earth acquired a Nb/Ta ratio of ~14 instead of ~20 as suggested by chondrites (Münker et al., 2003). In contrast, Zr/Hf likely remained chondritic (~35). As shown by the correlated Nb/Ta - Zr/Hf array (terrestrial fractionation array), subsequent second-order silicate differentiation that generated Earth's crust and mantle fractionated Nb/Ta concomitantly with Zr/Hf and produced complementary reservoirs with respect to Nb/Ta (crust ~12-13; mantle ~16). Although the mechanisms that fractionate Nb/Ta are poorly understood, a key role is attributed to the processes taking place during subduction of oceanic lithosphere, i.e. fractionation during dehydration and partial melting of eclogite or garnet amphibolite in the presence of Ti-phases with high D-values for the HFSE. Some hotspot lavas bear signatures of eclogite derived melts in that they have slightly higher Nb/Ta but lower Lu/Hf ratios than expected from melting of primitive mantle peridotite independent of whether rutile is present in the eclogitic residue or not. Eclogite melting, however, is not suitable to explain low Nb/Ta in the continental crust. Therefore, significant portions of the continental crust may have been produced early in Earth's history by amphibolite dominated melting in subduction zones or within thickened Archean mafic crust, as melts in equilibrium with garnet amphibolite are expected to have low Nb/Ta (e.g. Foley et al., 2002; Condie, 2008). In addition to fractionation by eclogite or garnet amphibolite melting, metasomatism at the asthenosphere - lithosphere boundary beneath continental plates may affect Nb/Ta concomitantly with Lu/Hf. Some evidence for this can be observed in rutile-bearing eclogite xenoliths that show evidence that metasomatism produces elevated Nb/Ta ratios in the lithospheric mantle along with low Lu/Hf and thus unradiogenic 176Hf/177Hf over time (Aulbach et al., 2008). This observation is supported by HFSE compositions of continental basalts, which display some of the highest Nb/Ta ratios measured so far by high-precision methods. References: Münker et al. 2003, Science 301, 84-87; Wade & Wood 2001, Nature 409, 75-78; Foley et al. 2002, Nature 417, 837-840; Condie 2008, Geology 36, 611-614; Aulbach et al. 2008, Nature Geoscience 1, 1-5.
V31F-03
Fractionation of High Field Strength Elements During Water-Saturated Partial Melting of Eclogite: Can Subduction Produce a Mantle Reservoir With Superchondritic Nb/Ta ?
An examination of the relative abundances of Ti, Zr, Nb and Ta in the depleted mantle and continental crust suggests the existence of an additional reservoir characterized by superchondritic Nb/Ta and Ti/Zr ratios [1]. While it has been shown that rutile-bearing refractory eclogite produced by partial melting of subducted oceanic crust is a potential candidate for this reservoir [2], experimental studies have consistently produced rutile-melt partition coefficients that cannot generate the required high field strength element fractionations [3]. Here I present new experimental results demonstrating that the temperature dependences of rutile-melt partitioning of Ta and Nb differ enough to produce DNb/DTa ~1.4 during H2O-saturated partial melting of eclogite at 3 GPa. Therefore, partial melting of subducted oceanic crust at these conditions is capable of producing rutile-bearing refractory eclogite with superchondritic Nb/Ta. Experiments were carried out on model rhyodacite and haplobasalt base melt compositions in the system SiO2-Al2O3-MgO-CaO-Na2O-K2O. Rutile saturation was achieved by adding 10- 40 wt% TiO2. Each starting composition was doped with ZrO2, Nb2O5, HfO2, and Ta2O5. Low-pressure experiments were carried in a vertical quenching furnace, and high pressure experiments were conducted in a solid-medium piston-cylinder device. The major element compositions of glass and rutile, as well as the trace element content of the rutile, were determined by electron microprobe. The trace element content of the glass was determined SIMS. These experimental results place new constraints on the HFSE fractionations produced by partial melting of subducted oceanic crust. At the low temperatures (~750°C) that characterize the H2O- saturated basalt solidus at 3 GPa [4], partial melting at the amphibolite/eclogite transition will produce a siliceous melt with strong depletions in Ti, Zr, Nb, Hf, and Ta. Further, a Ta/Nb compatibility crossover at 3 GPa and ~1000°C causes these melts to have larger Nb/Ta ratios than the residual eclogite. References: [1] McDonough (1991), Philos Trans R Soc London Ser A, 335:407-418; [2] Rudnick et al. (2000), Science, 287:278-281; [3] Schmidt et al. (2004), Earth Planet Sci Lett, 226:415-432; [4] Lambert and Wyllie (1972), J Geol, 80:693-708.
V31F-04 INVITED
Rutile is holding Nb and Ta in the mantle, negligible Nb and Ta in the core
Continental Crust has a significant depletion in Nb and Ta relative to La that has been attributed to convergent margin (arc) magmatism and greater retention of Nb and Ta in the mantle source. This depleted pattern is a consequence of the plotting order of elements, which has been established by the relative partitioning behavior of elements during MORB-OIB genesis. It is our hypothesis that rutile in subducting slabs or delaminated lower continental crust is the important phase causing Nb(Ta)-depletion in the continental crust, as well as lowering its Nb/Ta. Experimental studies reveal a range of Nb/Ta fractionation responses in residual rutile depending on temperature and phase relations(melting vs dehydration). Examples of rutile-bearing, refractory eclogites have been identified that serve as analog materials for residues of the continental crust. These rutiles have radiogenic Hf isotopes (Vervoort, unpubl. data), and thus are not recent precipitates from metasomatic melts, as has been recently suggested. What remains is to understand the total silicate Earth's mass balance. In this regard, it is worth noting that early Archean Barbarton komatiites possess chondritic La/Nb ratios, and Nb/Ta ~15, a value comparable to Allende CV3 chondrite. This observation is not consistent with the storage of Nb (or Ta) in the core and suggests that the silicate Earth controls the planetary budget of Nb and Ta. Constraints on the amount of Nb in the core must be evaluated by multi-element approaches, using ratios of refractory lithophile elements. Chondritic ratios for La/Nb and Nb/Ta are not defined simply as a single value with a restricted range and are not always constant, with examples of both negligible and distinct differences between groups of chondrites. The database for chondrites is still small for Nb and Ta.
V31F-05 INVITED
Titanium: A Tracer of Mafic Crust in Ocean Island Basalts?
The recycling of mafic oceanic crust and its re-sampling in ocean island basalts (OIB) is a long-standing notion (e.g., 1). However, arguments based on trace element ratios and long-lived isotopic measurements fall prey to the uncertain make-up of the mantle source, time-integration and other added 'flavours' the OIB source may contain. Accordingly, the paradigm of crustal recycling continues to be debated (e.g., 2,3). With the advent of large online geochemical databases such as PetDB and GEOROC, major elements are yielding new insight to mantle melting processes. I review why the high field strength elements (HFSE) and titanium in particular appear to be a unique tracers of mafic oceanic crust in the source of ocean island basalts (e.g., 4). I will further explore the use of simple Ti abundances combined with other geochemical parameters in a well-characterised OIB suite from Pico, Azores to estimate the amount of recycled crust in the source. 1. Hofmann and White, 1982; 2. Niu and O'Hara, 2003; 3. Pilet et al. 2008; 4. Prytulak and Elliott, 2007
V31F-06
High-Nb Basalts Provide Evidence for Rapid Recycling of High Field Strength Elements?
The majority of arc lavas are depleted in high field strength elements (HFSE) relative to other incompatible trace elements, suggesting that HFSE behave conservatively during subduction zone processing of crustal materials injected into the mantle. However, rare basaltic rocks enriched in Nb and other HFSE occur in a few arcs. These high-Nb basalts and/or Nb-enriched basalts have been proposed to be due to melting of portions of the mantle wedge that have been metasomatized by partial melt of the subducted slab. Slab melt supposedly stabilizes amphibole which then breaks down and enriches the mantle wedge with HFSE. If this is correct, then high-Nb basalts are probes of occasional, rapid recycling of HFSE in subduction zones and hold the key to a better understanding of complex subduction zone processes. High-Nb basalts occur in the Sulu Arc (southern Philippines, western Pacific) and Baja California (western Mexico, eastern Pacific). In both places, high-Nb basalts are mildly to moderately alkalic and have major and trace element compositions that greatly overlap with those of intraplate seamount basalts onboard the subducting slabs. More important, their Sr, Nd and Pb isotopic ratios are EM2- and HIMU-like, similar to the respective isotopic signatures of western and eastern Pacific seamount basalts. Thus data clearly indicate that high-Nb and seamount basalts come from similar mantle sources. These sources have not recently been metasomatized by slab melt and most probably represent relatively long-lived heterogeneous components that are inherently but variably present in the upper mantle. The heterogeneous upper mantle can be trapped in nascent subduction zones and/or can upwell through windows in subducting slabs to produce high-Nb basalts.
V31F-07
Fertility Pulses in the Iceland Plume
Diachronous V-shaped ridges (VSRs) in basement topography straddling the Reykjanes Ridge south of Iceland, expressed most clearly in the gravity field, reflect ±10% variation in crustal thickness. They appear to be produced by pulses of excess magma productivity propagating from Iceland at ~200 km/Ma with a periodicity of ~5 Ma. Three possible explanations for these crustal thickness variations are: (1) pulses in mantle temperature of ~20°C; (2) pulses of excess fertility; or (3) the regulation of mantle flow by rift relocations in Iceland. If the VSRs are due to pulses in mantle temperature and/or fertility, then this ought to be recorded in the Tertiary lava successions in east and west Iceland. Rift relocation, on the other hand, should leave no such record. Variations in mantle temperature of ~20°C will cause variations in degree of mantle melting and this should affect ratios of trace elements (e.g. Zr/Y) that are sensitive to degree of melting of garnet lherzolite. Icelandic basalt and normal mid-ocean ridge basalt (N-MORB) form distinct and parallel linear arrays on a logarithmic plot of Nb/Y vs. Zr/Y suggesting that the Icelandic basalt source is relatively enriched in Nb. Recycled subducted ocean crust is a plausible source of relative Nb enrichment, and the ΔNb parameter (the deviation, in log units, from a reference line separating the Iceland and NMORB arrays) serves as a proxy for this recycled component. ΔNb is insensitive to degree of melting and subsequent fractional crystallisation. Thus variations in ΔNb will reflect mantle fertility while variation in Zr/Y will reflect mantle temperature. The east Iceland Tertiary lava succession shows a steady decrease in Zr/Y from 14 to 2 Ma, implying a ~30% increase in degree of melting over this period but the scatter in the data is greater than the predicted variation caused by small temperature pulses. By contrast, ΔNb shows clear peaks and troughs that coincide closely with peaks and troughs in the gravity field across the VSRs adjusted to the time when they left Iceland. Thus fertility pulses in the outflow from the Iceland plume is the likely cause of the VSRs, but we cannot rule out small thermal pulses.
V31F-08
A Metasomatic Origin for the Enriched Walvis Ridge Basalts
he isotopic composition of the EMI and HIMU are consistent with them containing a component of recycled crust. For HIMU it is suggested that subducted oceanic crust plays an important role and for EMI it is argued that pelagic sediments is present in small but important amounts. One of the "type localities" of the EMI source is the Walvis Ridge, and the most enriched basalts from this locality define the EMI endmember. Models that explain the isotopic composition of the Walvis Ridge basalts need in addition to subducted oceanic crust with sediment a component of subcontinental lithosphere (SCL) [1, 2]. New and more complete data on Walvis Ridge basalts allows a better constrain on its source evolution. The approach has been to explain both the isotopic composition as well as the trace element composition with an as-simple-as possible model. The parent-daughter ratios of the U-Th-Pb, Lu-Hf, Sm-Nd and Rb-Sr system provide a large part of the incompatible trace element pattern. Starting with a bulk silicate Earth composition the isotopic composition can be modeled by a three stage evolution: first as Bulk Silicate Earth which ended by an 4.2 Ga old addition of 1.5% of a low degree garnet melt (1%). This enrichment raised the μ of the system to 10 during the second stage of the sources' history. This enrichment was followed by a 3.9 Ga old depletion equivalent to the extraction of 0.5% melt, again in the garnet stability field. This source then melts to 10% partly in the garnet and partly in the spinel stability field to generate the trace element pattern of the basalts. Because the nuclides represent a large range in half-lives the age constraints are surprisingly strong and especially the early fractionation cannot be younger then 4.1 Ga. Similarly the duration of the second stage is constrained to plus or minus 0.1 Ga. This purely metasomatic origin explains the enriched Walvis ridge basalt source evolution from "cradle to grave" with garden variety processes. Metasomatism at the mid- ocean ridge or at deeper in the mantle as proposed for E-MORB [3] are possible enrichment and depletion processes. We can also calculate a similar "cradle to grave" evolution of the less enriched end of the Walvis Ridge array which is similar to FOZO. It is proposed that FOZO is a mixture of depleted mantle, subducted oceanic crust and sediments. To generate this material in six dimensional isotope space and have the appropriate trace element compositions to be a source for the basalts the mixture needs to be 90% depleted mantle with 10% subducted crust with a 2% sediment component. This source then melts to a relatively high degree (15%). It should of course also be emphasized that these are not unique solutions, but only possible ones. [1] S.A. Gibson, et al. , Earth Plan. Sci. Lett. 237(2005). [2]A. Stracke, et al. , G-cubed 6(2005). [3] K.E. Donnelly, et al. , Earth Plan. Sci. Lett. 226(2004).