V11C-01 08:00h
Lateral Offset of the Volcanic Front: Implications for Fluid Pathways in the Mantle Wedge
One of the most striking common features of subduction zones worldwide is the appearance of the volcanic front at a height of approximately 120 km above the subducting slab. The water-rich compositions of the lavas erupted at the volcanic front suggest that melting is initiated by dehydration of hydrous phases in the slab, primarily amphibole. However, the location of the front is offset considerably from the predicted origin of fluids due to amphibole dehydration at $\sim$80km slab depth. The lateral offset at the surface varies with subduction angle, but generally the predicted site of fluid release is $\sim$20-70 km trench-ward of the actual volcanic front. Many of the proposed mechanisms for generating this offset involve stalling the fluid in the mantle such that it is drawn down and/or back into the mantle wedge due to viscous flow in the solid mantle. For example, the fluid may re-crystallize as phlogopite and pargasitic amphibole in the portion of the mantle that is viscously coupled to the subducting slab. These newly formed hydrous minerals have higher breakdown pressures than glaucophane, the dominant hydrous mineral in the slab, and could explain the offset to deeper apparent depths of dehydration. However, processes that rely upon solid mantle flow are very slow. For a slab descending at 100mm/yr, 5x10$^{5}$ years are required to descend 40 km vertically ($\sim$50 km along-slab). Such long periods of time spent in transport in the mantle are seemingly contradicted by strong U-series isotopic disequilibria in arc lavas. Although special circumstances may be evoked in order to allow U-series disequilibria to be extended in time, it is also possible that reaction rates in the cold descending slab are sluggish to the point that fluids are not released at the expected depth of 80 km, but instead are retained to greater depths where increasing temperatures allow reactions to proceed. In this scenario, the fluids would then be able to proceed relatively quickly to the region of melting, preserving their isotopic disequilibria.
V11C-02 INVITED 08:15h
Beryllium Isotope and Combined Be and U-series Isotope Studies of Volcanic Arcs: Implications for Fluid and Melt Transport Through the Mantle Wedge
Beryllium isotope and combined studies of $^{10}$Be/$^{9}$Be and U-series isotopes in volcanic arcs can 1) map transport of demonstrably slab derived elements through the mantle wedge; 2) certify the relationship of U series isotopes to slab derivation; 3) identify multiple stages in subduction modification of the mantle and constrain their timescales; and 4) speak to element partitioning into fluids and melts from the slab. In the Kurile, Aleutian and Bismarck arcs, $^{10}$Be/$^{9}$Be ratios for lavas from behind the volcanic front are often comparable to, and sometimes greater than, those at the volcanic front, despite the longer path to rear-arc locations, along which $^{10}$Be is decaying in transit. Be/Zr ratios show a similar pattern of across-arc increase, without increases in enrichments of Mo and Sn, species which would be mobilized with Be if F-bearing fluids were present. The simplest interpretation is that sediment melting, and its contribution to the mantle wedge, is greater behind the front than at the volcanic front. Despite evidence for an increasing sediment melt contribution behind the front, volcanoes from the Kuriles contain progressively less B, Pb, As and Sb with increasing depth to the slab, indicating that fluid processes updip of about 180 km (beginning in the shallow forearc) strip these elements nearly quantitatively from the sedimentary portion of the downgoing slab. For studies published to date (Aleutians, Central America, S. Chile, Bismarck, Mariana) $^{10}$Be/$^{9}$Be ratios are generally highest for samples plotting furthest from the $^{238}$U-$^{230}$Th equiline (i.e. highest $^{238}$U/$^{232}$Th, $^{230}$Th/$^{232}$Th, or both). In lavas from the Southern Volcanic Zone (SVZ) of S. Chile (Sigmarsson et al., EPSL 2002), U excess (U$_{xs}$), Ra excess (Ra$_{xs}$) and $^{10}$Be/$^{9}$Be are strongly correlated (r$^{2}$=0.81-0.94). This argues that U enrichment and in some cases Ra enrichment in arc lavas is related to slab processes that are capable of mobilizing $^{10}$Be out of the sediment column, rather than reflecting only dynamic melting processes or element partitioning in a hydrous mantle wedge. If U and Ra are thought to be carried in a fluid from the altered oceanic crust, that fluid must also carry sediment-derived elements, at least in the SVZ. In Nicaragua, the Aleutians and S. Chile, volcano $^{10}$Be/$^{9}$Be ratios correlate well with $^{230}$Th/$^{232}$Th, $^{143}$Nd/$^{144}$Nd and U$_{xs}$ and Ra$_{xs}$, respectively. The well constrained mixing lines require that both the subducted component and the mantle to which it is added be relatively homogenous for these slab-derived tracers. At zero $^{10}$Be (i.e. prior to sediment addition within the last 1-2 Ma), the Nicaraguan mantle is characterized by high Ba/La, $^{87}$Sr/$^{86}$Sr and $^{230}$Th/$^{232}$Th significantly elevated above MORB or OIB values, attributed to earlier subduction modification of the mantle (Reagan et al., GCA 1994). In the SVZ of Southern Chile, at zero $^{10}$Be, the $^{230}$Th/$^{232}$Th of the inferred mantle source is ~0.9, implying sediment addition prior to the more recent event that introduced $^{10}$Be. By contrast, at zero $^{10}$Be, the inferred Aleutian mantle has a $^{143}$Nd/$^{144}$Nd ratio of $\sim$ 0.5131, suggesting little or no prior sediment addition to the mantle.
V11C-03 08:30h
Metasomatic Mechanisms for Fractionation of Slab-Derived B Isotope Signatures Within the Mantle Wedge: Evidence From HP-Metasomatized Peridotites
We present B and $\delta^{11}$B data for metasomatic zones within exotic peridotite blocks of the Franciscan Complex, CA, that further constrain the behavior of this valuable tracer in subduction zones. These blocks preserve a series of metasomatic zones (serpentinite/talc schist/tremolite schist) imposed by Si-enriched hydrous fluids and are thought to be analogous to those developed in the mantle wedge adjacent to the slab-mantle interface. $\delta^{18}$O data indicate that these zones equilibrated with a single fluid at temperatures of $450-500\deg$C; this metasomatism is akin to that producing serpentinite seamounts in the Mariana subduction zone, although at higher temperature and with greater extents of reaction progress. Incipiently serpentinized protoliths contain 10-50ppm B, which overwhelms the likely initial $\delta^{11}$B of these peridotites. $\delta^{11}$B for these least-altered peridotites ranges from +6 to +8$\permil$, which agrees well with estimates for slab-derived fluid $\delta^{11}$B for the temperature recorded by these samples. Serpentinites formed after peridotite contain ~10-80ppm B, and increasing B concentrations exhibit a marked positive correlation with serpentine $\delta^{18}$O and whole-rock SiO$_{2}$ indicating that progressive serpentinization correlates with increasing B concentrations. However, as B contents increase, $\delta^{11}$B becomes negative (0 to -10$\permil$). As increasing B supports progressive removal of B from fluids, negative $\delta^{11}$B indicates a fractionation process. The direction of this fractionation is the opposite of that expected for a Rayleigh distillation process, and indicates that a shift in the $\delta^{11}$B composition of the fluid phase during serpentinization is most likely. Formation of a talc-dominated rock from serpentinites resulted in drastically reduced B concentrations of 3-20ppm, but with similar $\delta^{11}$B. This reaction, although driven by an influx of fluid, is a dehydration reaction and reduced B concentrations imply remobilization of B by dehydration fluids, but no fractionation of $\delta^{11}$B due to dehydration is apparent. Tremolite-dominated samples formed after talc schists likewise have similar B concentrations and $\delta^{11}$B, indicating that this reaction has little effect on B systematics. Data in the literature strongly indicate that subduction zone fluids have basic pH. If metamorphic silicates faithfully record only the isotopic composition of B speciated tetrahedrally within fluids, our data suggest that the pH of serpentinizing fluids is lowered within the serpentinite metasomatic zone. The most likely agent for a shift in fluid chemistry during serpentinization is the formation of magnetite, in which silicate ferrous iron is oxidized; a by-product of this reaction is probably H$_{2}$ gas, which would act to lower fluid pH. The evolution of $\delta^{11}$B to negative values during serpentinization therefore suggests that B and its isotopes can be significantly fractionated by metasomatic processes in the mantle wedge, and that true "slab-derived" $\delta^{11}$B may not be sampled by arc volcanics.
V11C-04 08:45h
Lithium Isotopic Composition of Guatemalan Jadeitites
Lithium (Li) isotopes have been measured on Guatemalan jadeitites using MC-ICP-MS, to better understand the evolution of fluids and fluid exchange during subduction. These jadeitites are high-pressure ($>$6 Kbar), low-temperature (200 to 400C) rocks in serpentinite melange that have metasomatic signatures (including high Li concentrations), and are interpreted as having crystallized from slab-derived fluids. They occur on both sides of the Motagua Fault Zone of Guatemala, which is where the North American and Caribbean plates meet. We have measured the Li isotopic compositions of both jadeite and phengitic muscovite from jadeitites occurring both north and south of the fault. The range in Li isotopes is 0 to -4 \permil, relative to the standard LSVEC, and the precision is 0.5 \permil. There is no large systematic offset in the isotopic ratio north and south of the fault, nor is there a fractionation in isotopic ratio between jadeite and phengite within a given rock. These initial values may suggest that the fluids being sampled by these HPLT phases are isotopically light, compared to the Li isotopic ratio of the mantle (+4 \permil), that of seawater (+30 \permil) or that of altered ocean crust or sediments (\sm 0 to +25 \permil). Further analyses are needed to constrain the source and nature of these fluids as well as the process that generates such a large isotopic fractionation in precipitates of a fluid that equilibrated with an isotopically heavy protolith.
V11C-05 09:00h
In situ determination of Pb, Sr, Rb, Zr partitioning between hydrous melts and aqueous fluids at high pressure and temperature
The combination of experiments performed in diamond anvil cells and in situ synchrotron light source measurements represents a powerful tool for studying the petrogenesis of subduction-related magmas. We have undertaken a program of partitioning experiment coefficient determination between silica-rich hydrous melts and aqueous fluids of various compositions (± NaCl) for Pb, Sr, Rb and Zr at high pressures (up to 1.5 GPa) and temperatures (up to 900°C), using the X-ray microprobe of the ID22 ESRF beamline. Experiments were carried out in a Bassett-modified diamond anvil cell. We measured the equilibrium between aqueous fluids and silicate melts before, and/or after, their complete miscibility was reached, allowing the calculation of partition coefficients for the studied element (i): Difluid/melt=Cifluid/Cimelt. Our experimental results show that for the studied elements the partitioning process strongly depends on the composition of the aqueous phase in equilibrium with the melt. Indeed, the presence of Cl in the starting aqueous solution favours a significant affinity of Pb, Rb, Sr for the melt phase, whereas without Cl, there is no strong partitioning of these elements into either the aqueous or the hydrous silicate phase. Difluid/melt values will be presented together with elemental distribution maps obtained inside the cell. This strong affinity for silicate melt in the presence of an NaCl aqueous solution implies that saline aqueous fluids may play a key role in the generation of arc magmas. These fluids may also act as initiators of the geochemical fingerprint of arc-related magmas. Trace element transport by silica-rich hydrous fluids may also be considered as an efficient metasomatic agent in the mantle during elemental recycling in subduction zones.
V11C-06 09:15h
Trace Element Partitioning Between Coexisting Silicate Melts: the Effect of Melt Composition
Experiments on coexisting immiscible liquids allow to nail down the influence of melt structure/composition on the partitioning of trace elements without having to deal with crystal compositions changing with melt composition. Piston cylinder experiments at 0.3-0.7 GPa, 1050-1240 $\rm ^o$C in the K$_2$O-FeO-Al$_2$O$_3$-SiO$_2$ system yield a Fe-gabbroic coexisting with a Fe-rich granitic melt $\pm$ quartz and fayalite. The suitable temperature range is limited by massive crystallization of fayalite and quartz below 1120 $\rm ^o$C and by the abrupt closure of the miscibility gap above 1180 $\rm ^o$C. Typical textures yield droplets of gabbroic melt in granitic melt, in the presence of fayalite, these droplets coalesce with the crystals. In all cases, the gabbroic droplets have internal droplets of granitic melt, zones of quench-fayalite, and sometimes crystallize wuestite. In order to produce two melts measurable by LA-ICP-MS, the piston cylinder is mounted onto a centrifuge, after 48 hours of equilibration, we segregate the gabbroic melt completely from the granitic melt at typically 6-12 hours at 700-1000 {\em g}. Each of the melts was then free of droplets, the gabbroic melt has sometimes quench-fayalite but other quench textures are at the submicron scale. The resulting partition coefficients agree with the few previous studies (Watson 1976, CMP: Ryerson & Hess 1978, GCA) performed on few elements at microprobe concentration level, but deliver a very surprising element pattern. The alkalis and earth alkalis have D's (gabbro/granite) raising regularly with field strength or electronegativity from 0.1 to 5; only Cs, Rb, K, and Na partitioning into the granitic melt. A second group of elements (transition metals, REE, commonly used HFSE) partition into the gabbroic melt at fairly similar D's (e.g. 5-11) without correlation with r, Z/r, electronegativity or whatsoever. There is a faint trend to higher D's with higher oxygen coordination number and there are distinct small differences between elements (e.g. Nb/Ta) but the proposed dependence of D's with field strength or a similar parameter does at least not exist for these elements. A third group of elements with partition coefficients of 1-0.07 is comprised by the amphoteric elements which either substitute Si in the tetrahedra polymerizing the melt or form low ($\leq$ 4) coordination polyhedra with oxygen mostly due to lone electron pairs (e.g. Sn, Pb, As, Sb). The observed behavior can be best described by a 3 oxygen-type site model. The melt-effect on partition coefficients of commonly used trace elements in the investigated compositional range is thus up to 2 orders of magnitude, but apart from a few cations (e.g. alkalines and Ba,Sr,Pb,Al), the relative fractionation is less than a factor of 2. With a narrowing immiscibility gap, a straightforward correlation between gap width and (decreasing) D's is observed.
V11C-07 INVITED 09:30h
Second Critical Endpoint in Peridotite-H2O System and its Bearing on the Magmatism in Subduction Zones
Migration of fluid/melt through mantle wedge plays an important role on the magmatism in subduction zones. In general, both the solubility of water in silicate melts and the solubility of silicate materials in aqueous fluid increase with increasing pressure. This could suggest that, above a certain critical pressure and temperature (the second critical endpoint), silicate melts and aqueous fluid become indistinguishable from each other. Before discussing the movement of these two mobile phases, therefore, the stability field of each phase should be investigated. In order to determine the second critical endpoint in peridotite-H2O system, experiments were conducted using X-ray radiography technique together with Kawai-type double-stage multi-anvil high pressure apparatus (SPEED-1500) installed at SPring-8, Japan. Direct X-ray beam, which passes through the anvil gaps of SPEED-1500 and sample under high pressure, is observed with an X-ray camera. The sample container should not react with hydrous samples, but should be x-ray transparent. We, therefore, developed a new sample container, which is composed of a metal tube and a pair of single crystal diamond lids put on both ends of metal tube. The sample in the metal container can directly be observed through the diamond lids with X-ray radiography. The experimental conditions are at pressures from 1.7 to 4.0 GPa and at temperatures up to about 1400 deg. C. Pressure is applied first, and then temperature is increased. At around 1000 deg. C and 1.7 GPa, a dark gray sphere appeared in the light gray matrix. The light gray matrix that absorbed less X-ray is considered to be an aqueous fluid phase, whereas the dark gray sphere is silicate melt. With further increasing temperature, the drastic overturn was observed. In the experiments up to 3.6 GPa, two phases (fluid and melt) were observed. At 4.0 GPa, we could not distinguish two phases in the radiographic images. Our experimental results indicate that aqueous fluid and silicate melt can coexist up to 3.6 GPa and there is no difference between these two phases above 4.0 GPa. It could be concluded that the second critical endpoint in the system peridotite-H2O occurs at pressure between 3.6 and 4.0 GPa. At pressures below the second critical endpoint, chemical differentiation could be occurred by the fluid-liquid immiscibility. On the other hand, no such differentiation could happen above the second critical endpoint because the only supercritical fluid is stable. Our experimental results suggest important implications for the magmatism under hydrous conditions, such as the subduction zone magmatism.
V11C-08 09:45h
Mantle Diapirs and Genesis of Arc Magmas: Evidence From the Sumisu Caldera Volcano, Izu-Bonin arc, Japan.
Are arc basalts dry, wet or both? What are the relationships between flux melting and pressure release melting beneath arc volcanoes? The example we discuss is the Sumisu caldera volcano, Izu-Bonin arc. We present evidence that (1) there exist two kinds of basalts in the volcano, which have petrographical, mineralogical and geochemical differences. (2) These systematic and interrelated differences could have been resulted from different water contents in these magmas and differing degrees of melting in the source mantle, and thus dry and wet basalts can coexist in a single volcanic system, and (3) wet basalts have been derived from a more depleted source than dry basalts, which resulted from different degrees of melting. These degrees of melting are $\sim$20 % and $\sim$10%, respectively. We further present a mantle diapir model beneath the volcano, which could produce dry and wet basalts simultaneously in the same volcano. Our findings may have relevance to magma genesis models in other subduction zones. Basalt-basaltic andesite ($<$55 wt % SiO$_{2}$) and dacite-rhyolite (66-74 wt % SiO$_{2}$) are predominant eruptive products in Sumisu caldera volcano, Izu-Bonin arc, Japan. The most-magnesian basalt (8.5 % MgO), as well as some of the other basalts, contain low Zr (20-30 ppm), which cannot yield basalts containing higher Zr (30-40 ppm) through fractionation and/or assimilation. On the other hand, we recognised that high- and low-Zr basalts have differing phenocryst assemblages, distinct phenocryst chemistries of olivine, plagioclase and pyroxene, different depletion of REE (rare earth element) patterns, and differing fluid mobile-element/immobile-element ratios. Estimated primary olivine compositions are more magnesian ($>$Fo$_{91}$) and thus more depleted in low-Zr basalts compared to those in high-Zr basalts (Fo$_{90}$). Low-Zr basalts contain up to 5 vol % augite, but many high-Zr basalts are free of augite, which appears only in their evolved stage. Hydrous basalts crystallize olivine followed by augite and plagioclase, producing the former assemblage. Moreover, the low-Zr basalts have higher Ba/La and Ba/Zr ratios than the high-Zr basalts. We suggest that both dry and wet primary basalts existed in the Sumisu magmatic system, each having different trace element concentrations and mineral chemistry and assemblages. The lower content of Zr and light REE and magnesian primary olivines in the wet basalt could therefore have resulted from a higher degree of partial melting ($\sim$20 %) of a hydrous source mantle compared to $\sim$10 % melting of a dry source mantle. Interestingly, Sr, Nd and Pb isotopes between these wet and dry basalts are similar and are limited in range. These lines of evidence indicate that mantle diapir model might be applicable to satisfy the configuration of such a mantle source region beneath a single volcanic system such as Sumisu.