V13A-1436 1340h
Fluid-Metasedimentary Rock Interactions Within Subduction Zone M\'{e}lange: The Secret to Trace Element Enrichments in Arc Magmas?
Fluid alteration of deeply-subducted metasedimentary rocks on the island of Syros, Greece, was investigated using geochemical mass balance analysis to evaluate potential sources for trace elements in arc magmas. Concentrations of large ion lithophile elements (LILE; e.g., Ba, K, Rb, Cs, Ca, Sr), U, and Pb are generally elevated in arc magmas relative to more refractory elements like Ti, Th, Hf, Nb, and Zr. This geochemical ``fingerprint'' is considered to be a strong indicator of fluid addition to arc magma source regions. However, the source for trace elements and the fluids that transport them are controversial because experiments and field observations suggest that metamorphic dehydration of subducted mafic, ultramafic, and/or sedimentary rock reservoirs alone is unable to liberate and transport the required elements to the mantle wedge. New petrological and geochemical data from contacts between subducted metasedimentary rock and meta-ultramafic m\'{e}lange matrix on Syros reveal that multiple sources may be required. We propose that dehydration fluids from subducting slabs infiltrate and equilibrate with heterogeneous meta-ultramafic/metamafic rocks in m\'{e}lange zones near the slab-mantle interface. When these fluids gain access to subducted metasedimentary rock they can preferentially leach LILE, U, and Pb, without sediment melting, due mostly to the breakdown of phengite and epidote. Migration of these trace element-enriched fluids into the sub-arc mantle wedge could directly trigger partial melting and produce the elevated Ba/Th, Sr/Th, Pb/Th, U/Th and radiogenic Sr that are characteristic geochemical features of arc magmatism. Our findings provide a mechanism for the geochemical decoupling of Th from U and Pb consistent with the U-Th isotopic disequilibrium observed in many arcs. Infiltration at shallower depths (e.g., Syros) would produce metasomatized mantle that could be carried deeper by wedge corner flow to undergo partial melting when it reached sub-arc regions.
V13A-1437 1340h
Sr-Nd-Pb Isotope Geochemistry of Melange Formation: Implications for Identification of Fluid Sources in the Mantle Wedge and the Arc
Paramount to our ability to decipher the behavior of fluids and melts within the mantle wedge and the overall subduction system are the chemical compositions of rocks adjacent to the slab-mantle interface. Profound metamorphic and metasomatic alteration of pre-subduction lithologies to form melange along the slab-mantle interface may yield rock types inheriting mixed chemical compositions of diverse pre-subduction lithologies. Early work on melange geochemistry indicates competitive effects between mechanical mixing, metasomatism by fluids or melts, and mineral stabilities imposed by the resulting bulk composition. We have explored the Sr-Nd-Pb isotope geochemistry of low- to high-grade melange zones in the Catalina Schist, CA, to address this crucial missing component in studies of subduction-zone mass flux. The Catalina Schist contains lawsonite-albite (LA), lawsonite-blueschist (LB), and amphibolite (AM) facies melange zones, all with mineralogy dominated by talc, chlorite, and Na-Ca amphiboles, with additional minerals such as micas, rutile, zircon, and apatite stabilized based on bulk sample chemistry. Major element compositions vary, from strongly ultramafic in the AM melange, to more crustal-like compositions (i.e., more reminiscent of basaltic to sedimentary protoliths) for LA and LB melange. However, initial Sr and Nd isotope ratios for all grades of melange are essentially indistinguishable, displaying a wide variation from $^{87}$Sr/$^{86}$Sr=0.703-0.709 and $\epsilon$Nd= +15 to -15. Covariations are generally negative, similar to that of the mantle array, but with some samples extending to higher Sr ratios at constant $\epsilon$Nd that probably reflect inheritance of seawater Sr. No clear mixing relationships between $^{87}$Sr/$^{86}$Sr and 1/Sr exist, suggesting either localized buffering of Sr isotope ratios or that mixing relations are obscured by secondary devolatilization. However, a clear mixing trend for Nd indicates two end-members, one a high-concentration, positive $\epsilon$Nd source (AOC?), the other with low-concentration and negative $\epsilon$Nd (devolatilized sediments?). Likewise, initial Pb isotope ratios for all grades of melange form a single array independent of rock type or inferred protolith. Melange matrix of the Catalina Schist preserves initial $^{206}$Pb/$^{204}$Pb of 18.95-19.59, $^{207}$Pb/$^{204}$Pb of 15.61-15.68, and $^{208}$Pb/$^{204}$Pb of 37.85-39.05. Such elevated Pb ratios are typical of subducting oceanic sediments, but not of MORB-like oceanic crust or peridotites of the depleted mantle. The similarity of these initial ratios suggests pervasive alteration of Pb isotope signatures within diverse rock types by fluids during subduction. As Pb concentrations decline from LA/LB to AM melange, this suggests devolatilization of Pb from the ultramafic AM melange will transfer crustal-like Pb isotope ratios. Sr-Nd-Pb isotope systematics for arc volcanic rocks are commonly used as indicators of fluid sources from the subducting slab to the arc magma source region. Our results suggest such an assumption is extremely dangerous, as hybridization processes common to melange zones are more likely to occur along the slab-mantle interface than is preservation of a pre-subduction section. Such metamorphic mediation and buffering of "slab" compositions is essentially unknown, yet our data support an interpretation where these processes impart a fundamental control on the chemistry of fluids passed to the mantle wedge.
V13A-1438 1340h
Serpentinized Peridotites of the Stonyford Volcanic Complex: Melt Depletions and Enrichments in a Fore-arc Mantle Wedge.
The question of geochemical flux in the mantle wedge during subduction is critical to our understanding of arc volcanism, and forms an important aspect of the global geochemical flux. The MARGINS program attempts to understand these processes by studying active subduction zones, where direct observations can be made. An alternative approach is to examine outcrops of lithospheric mantle that underlie crust known to have formed by supra-subduction zone (SSZ) magmatism. This lithospheric mantle represents, in part, the source from which the overlying crust was extracted, and its mineralogy and composition reflect the processes that have affected it through time, including melt extraction, fluid phase enrichment, and subsequent interactions with melt derived from lower in the mantle. These processes have been frozen in place by cooling and emplacement of the mantle lithosphere and its overlying crust. Fore-arc peridotites are characterized by spinels with higher cr\#s than abyssal peridotites, indicating significantly higher fractions of partial melting compared to abyssal peridotites. High fractions of partial melting are confirmed by whole rock incompatible trace element concentrations, which are strongly depleted when compared to abyssal peridotites. Forearc peridotites near Stonyford, California, preserve evidence for variable melt extraction under different pressure regimes and subsequent melt reactions that have re-enriched most samples in LREE. Many of these samples are less than 50% serpentinized and many of the harzburgites contain pristine, optically clear Cpx, Opx, olivine, and spinel. Cr-spinels are characterized by a range in cr\#s, consistent with up to 40% partial melting in the most Cr-rich rocks. Spinels from two of the harzburgites have low cr\#s and high mg\#s that plot at the undepleted end of the abyssal peridotite array. Three harzburgites have spinels that plot below the Cr-rich end of the abyssal peridotite array, within the less-depleted end of the Izu-Bonin-Mariana fore-arc peridotite array. Spinels in the other peridotites plot within the most depleted end of the fore-arc peridotite array. These Cr-rich spinels are also comparable to chromite microphenocrysts in boninite lavas associated with extended fore-arcs. Pyroxenes in the depleted harzburgites have higher mg\#s and lower Al, Ti, and Na contents than pyroxenes from the abyssal harzburgites. Samples that preserve primary olivine have Cr-spinels that fall on the Olivine-Spinel Mantle Array (OSMA) and melting trend defined by Arai (1994). The fraction of melt extracted from a fertile MORB mantle, based on Arai's model, ranges from less than 5% in one of the enriched samples to 28%; spinels from several samples with no relict olivine have cr\#s that imply approximately 40% melt extraction. Our preliminary trace element data for the Stonyford peridotites show that melting in the depleted fore-arc peridotites initiated within the garnet facies and continued into the spinel facies. The peridotite compositions fall on melt reaction trends (pyroxene dissolved, olivine precipitated) that are distinct from simple melt addition trends, and imply percolation of melt through a porous medium with which it was not in equilibrium. The effects of this melt reaction can be seen in spider plot normalized to primitive mantle, in which the LREE are enriched in the depleted harzburgites and dunites, whereas all REE are elevated in the "enriched" harzburgites (those with aluminous spinel and enriched pyroxene). Preliminary ion probe data on pyroxene from one harzburgite suggest that this melt was an alkali basalt similar to those found within the Stonyford volcanic complex.
V13A-1439 1340h
Oxidized As (V) in fore-arc mantle serpentinites: Transfer of fluid-soluble elements from slabs to arc magmas
Fluids released from subducting slabs and sediments hydrate the overlying peridotites in mantle wedges. Such hydrated peridotites (serpentinites) are enriched in fluid-soluble elements, although insoluble elements show the refractory geochemical signature. The enrichment pattern of serpentinites is similar to that of arc magmas (Hattori & Guillot, 2003 in Geology). Arsenic is one of the most enriched soluble elements, reaching greater than 1000 times of the primitive mantle value. We examined the speciation and occurrence of As in serpentinites to understand how such a highly mobile element can be transferred from subducting slabs to arc magmas via mantle wedges. Our study used serpentinites associated with the Tso Morari eclogitic rocks in the Indus Suture Zone of Himalaya. They represent the hydrated peridotites at the base of mantle wedge beneath the margin of Eurasia and were exhumed from the depth of about 100 km during the active subduction of the Indian continental margin. The serpentinites are made up of antigorite, chromite, minor talc. It contains As ranging from 6 to 275 ppm and S up to 51 ppm, but most have S below detection limit, 4 ppm. X-ray absorption spectroscopy data show that As is mostly As(V) and combined with oxygen, although Minute grains of As-bearing sulphides and arsenides are identified in samples. The fractions of As(V), calculated from the X-ray absorption near-edge structure, are greater in samples with higher As contents. High proportion of As(V) in the fore-arc mantle serpentinites contasts with high As(III) in the serpentinite at the base of the Nidar ophiolite. The source of As (V) in the serpentinites is most likely As adsorped on Fe- and Mn-oxides in subducted sediments and slabs. It was released during the subduction of slab and sediments at low temperatures, \< $350\deg$ C, and shallow depths, \< 25 km. Continuous flux of water from slabs at the base of the mantle wedge likely maintained As in oxidized condition. Arsenic in the serpentinites is, then transported to deeper levels in the mantle by the downward movement of the host serpentinites along the subduction zone. Eventual dehydration of serpentinites discharges water and fluid-soluble elements, leading to partial melting and tne enrichment of these elements in partial melt
V13A-1440 1340h
Toward long-term geochemical sampling of gases and deep fluids in subduction zone fore-arcs: New instrument developments
We present preliminary results of an on-going instrument development study aimed at quantifying the rate of elemental loss to the ocean/atmosphere in active fore-arc margins. Work on subduction zones to date has focused on elemental fluxes associated with magmatism at the arc front. For example, the flux of carbon output along the strike of the Central America arc is $\sim$ 5 x 10$^{7}$ mol/yr/km, or ~14% of that potentially available by input via the trench (Shaw et al., EPSL, 2003). This result indicates that carbon is (a) efficiently recycled to the (deeper) mantle, i.e. the mantle beyond the zone of arc magma generation, and/or (b) lost in the fore-arc region. There are few constraints on elemental losses at the fore-arc region; the present work, therefore, is motivated by quantifying the flux of volatiles (and other species) lost in the early stages of the subduction cycle. This will allow a qualitative assessment of the importance of deep recycling and contribute to an increased understanding of the hydrogeology of active margins. The Chemical and Aqueous Transport (CAT) meters (Tryon et al., Deep Sea Research, 2001) used in this study record a time series of flow rates by injecting a tracer at a constant known rate into the flow stream through the instrument and by sampling downstream of this point for tracer dilution. They also collect a time series of seep fluids in copper coils and maintain them at seafloor pressure during recovery. The Optical Flow Meter (OFM) measures flow by determining the time-of-flight of a tracer pulse injected into the flow stream. An osmotic pump is used to sample fluids in a manner similar to the CAT meters. A series of tests utilizing both sets of instruments has been conducted at the Extrovert Cliffs site in Monterey Bay during 2004. Sites chosen range from diffuse flow sites with output rates of 10s of cm/yr to highly focused visibly flowing sites: all localities are covered by extensive microbial mats and chemosynthetic clams. Our coupled data sets of aqueous flow rate and geochemical/volatile composition allow us to estimate geochemical fluxes at these sites. Our next deployment of the instrumentation will target seeps off the Costa Rica margin.
V13A-1441 1340h
Decarbonation, Serpentinization, Abiogenic Methane, and Extreme pH beneath the Mariana Forearc
Low-chlorinity springs sampled from ten sites on nine serpentinite mud volcanoes show systematic chemical gradients across the outer Mariana forearc that result from progressive devolatilization of the subducting Pacific plate. Sites range from 50 to 90 km from the trench axis corresponding to depths to the top of the plate of $\sim$15 to 29 km. Dissolved sulfate, Na/Cl, K, Rb, Cs, and B in the springs all increase regularly with distance from the trench, leached from the subducting sediment and altered basalt in response to increasing temperature at depth from $\sim$100-$300\deg$C. Sites nearer the trench have high Ca (up to 75 mmol/kg) and Sr, low alkalinity, and pH 10.7, whereas sites farther from the trench have almost no Ca and Sr, alkalinity (some carbonate but mostly hydroxyl) as high as 69 meq/kg, and pH 12.5. Springs with high alkalinity also have high methane ($>$44 mmol/kg) that feeds sulfate-reducing microbial communities in the shallow subsurface and macrofauna at the seafloor. These distal springs form chimneys and crusts of CaCO$_{3}$, whereas the proximal springs form chimneys of brucite. High alkalinity at the distal sites apparently results from decarbonation at the top of the subducting plate; because serpentinization during ascent generates both high pH and H$_{2}$, the resulting dissolved carbonate is reduced to methane such that carbonate alkalinity is replaced by hydroxyl alkalinity: 4H$_{2}$ + CO$_{3}$$^{=}$ = CH$_{4}$ + H$_{2}$O + 2OH$^{-}$. This reaction can account for the much higher pH of the distal springs. Chlorinity of the springs varies from 234-546 mmol/kg and is related more to latitude N-S than to distance from the trench. Distal springs have otherwise similar compositions over this entire range of chlorinity, implying that chloride derives from depth rather than from mixing with seawater within the seamounts themselves. The range in chlorinity can readily be explained by serpentinization at reasonable water/rock mass ratios of 0.2-1.0 if 30-40% of the spring water originates as residual pore water in subducted sediment and basalt rather than as H2O$^{+}$ of dehydration. For convergence at 4.75 cm/y and per km of trench, the rate of subduction under the Mariana forearc is 24 km$^{3}$/My or 3.4 x 10$^{10}$ g/y of (dry) sediment 500 m thick, and 95 km$^{3}$/My or 2.7 x 10$^{11}$ g/y of altered basalt 2 km thick. For loss from sediment of 1% H2O$^{+}$ and 2% residual pore water, plus from altered basalt 2% H2O$^{+}$ and 1% pore water, going from lawsonite-albite to epidote-blueschist facies as in the Catalina schist (Bebout, 1995), this amounts to a water flux of 9 x 10$^{9}$ g/y per km of trench, upward into the mantle of the outer Mariana forearc. This water derives 90:10 from basalt:sediment and 63:37 from H2O$^{+}$: residual pore water. Corresponding vertical fluxes of sulfate, C, Na, K, Rb, Cs, B, Ca, and Sr in the forearc springs represent 0.05-3% of the amounts subducted, consistent with continued supply at greater depths.
V13A-1442 1340h
Chemical and isotopic compositions of thermal water related with possible ascending deep fluids in Kii Peninsula, SW Japan
Kii Peninsula is located in the fore-arc region of SW Japan, where the Philippine Sea plate has been subducting. Although there are no Quaternary volcanic activity, high-temperature thermal brine as well as various kind of hot springs well out in many places. Possible origin of these unusual hot springs is that fluids originated from dehydrated water from subducting slab are incorporated into the surface water (Kazahaya et al., 2003; Matsumoto et al., 2003). Salah and Zhao (2003) suggested the existence of fluids in the crust and the mantle wedge resulting from the dehydration of the subducting slab based on the detailed 3-D P and S wave velocity structures. We examined both chemical and isotopic features of hot springs to make distribution of these unusual hot springs and their relationship with fluids dehydrated from the subducting slab clearer. NaCl-type hot springs were significantly enriched in the heavy isotopes of oxygen and hydrogen, although those of Na(Ca)HCO$_{3}$-type were not shifted from meteoric water line. From helium isotopic compositions, the gas component derived from deeper region including upper mantle is main contributor for both type of hot springs. The $^{20}$Ne concentrations in NaCl-type hot springs are far lower than that in air saturated water (ASW), while most of Na(Ca)HCO$_{3}$-type hot springs have similar or higher $^{20}$Ne concentration level than ASW. From these results, we can get insights into the origin of each type hot spring. High temperature thermal brine with mantle derived component is incorporated into the surface meteoric water and, subsequently, suffered water-gas separation. The gases separated from the thermal brine were rich in noble gas with high $^{3}$He/$^{4}$He ratio and were dissolved into meteoric water. The former may correspond to NaCl-type hot spring, and the latter may correspond to Na(Ca)HCO$_{3}$-type one.
V13A-1443 1340h
Across-arc variation in ${}^{40}$Ar/${}^{36}$Ar ratios of olivines in volcanic rocks from the Izu-Ogasawara arc, Japan
The Izu-Ogasawara arc is located at an intra-oceanic convergent margin between the Pacific and Philippine Sea plates. This arc is suitable to investigate the recycling of volatile materials concurrent with subduction process, because contribution of continental crustal component in arc magma can be negligible. Here we discuss noble gas isotopic compositions of olivines in volcanic rocks from the volcanic front and back-arc regions of this arc to investigate the volatile behavior in slab-derived fluid during subduction processes. The samples from both regions have similar ${}^{3}$He/${}^{4}$He ratio of about 8.0 ${R}_{A}$, which is in the range of the MORB value ($8 \pm 1 {R}_{A}$), indicating that the contribution of helium in slab-derived fluid to the mantle wedge is negligible. However, ${}^{40}$Ar/${}^{36}$Ar ratios of samples from the back-arc region range from 380 to 620, whereas those from the volcanic front region are 300 to 320. The ${}^{40}$Ar/${}^{36}$Ar ratios of both regions are significantly lower than that of the MORB source (up to 40000) indicating that the contribution of slab-derived atmospheric argon is dominant in the mantle wedge. The difference in ${}^{40}$Ar/${}^{36}$Ar ratios in both regions may reflect the different contribution of slab-derived component. Assuming that atmospheric argon in the highest ${}^{40}$Ar/${}^{36}$Ar samples from each region are derived from subducting slab, but not due to the shallow level contamination, contributions of slab-derived ${}^{40}$Ar in each magma are about 90% in the volcanic front and 50% in the back-arc regions, respectively. If water content of source mantle which generates a primary melt is 0.14 wt%, and wedge mantle originally contains argon similar to the MORB source (${}^{40}$Ar/${}^{36}$Ar = 40000, ${}^{40}$Ar concentration = $7.4\times {10}^{-7}$ cc/g), slab-derived ${}^{40}$Ar/H${}_{2}$O ratios in the source mantle are estimated to be $5.6\times {10}^{-3}$ cc/g in the volcanic front and $4.7\times {10}^{-4}$ cc/g in the back-arc regions, respectively. These ${}^{40}$Ar/H${}_{2}$O ratios are comparable to those of air-saturated sea water, sediments and altered oceanic crust ($2\times {10}^{-5}$ to $3\times {10}^{-4}$ cc/g). This suggests that subducting atmospheric ${}^{40}$Ar is assumed to be introduced into the mantle wedge associated with slab dehydration process beneath the arc.
V13A-1444 1340h
Generation of high-Mg andesites in the Kueishantao volcano, the southernmost part of the Okinawa Trough
Kueishantao is an emerged volcanic islet located at the western end of the Southernmost Part of Okinawa Trough (SPOT). The Okinawa Trough, extending from SW Kyushu, Japan to NE Taiwan, is widely regarded as a backarc basin that is built behind the Ryukyu arc-trench system owing to subduction of the Philippine Sea plate underneath the Eurasian plate. The SPOT, however, is not a simple backarc basin but an embryonic rift zone in which early arc volcanism occurs as a result of the Ryukyu subduction (Chung et al., 2000). The Kueishantao is one of such volcanoes thus formed in the SPOT and consists mainly of andesitic lava flows dated to be $\sim$7000 yr old. In this study, we report whole rock major and trace element, and Sr-Nd-Pb isotope compositions of the Kueishantao andesites. The results indicate that some of the samples have unexpectedly high magnesium, with MgO $\geq$ 5 wt.$%$ and Mg\# $>$ 0.5, relative to their silica contents (SiO$_{2}$$\approx$ 60 wt.$%$), which allow them to be coined as high-Mg andesites (HMAs). In the incompatible element variation diagram, these Kueishantao HMAs exhibit enrichments in the large ion lithophile elements and Th, U and Pb, and depletions in the high field strength elements, features typical of arc lavas from the Ryukyu subduction zone as well as convergent margins worldwide. More interestingly, their overall geochemical compositions are very similar to those of the mean continental crust proposed by Rudnick and Fountain (1995). The Kueishantao HMAs have uniform isotope compositions, with low $\epsilon$Nd (-4.3 to -5.0), high Sr ($^{87}$Sr/$^{86}$SrÍ™$\approx$ 0.706) and Pb (18.75, 15.68 and 39.02 of $^{206}$Pb/$^{204}$Pb, $^{207}$Pb/$^{204}$Pb and $^{208}$Pb/$^{204}$Pb, respectively) ratios. Such continental isotopic signatures have led previous workers (Chen et al., 1995) to argue significant crustal contamination as a major petrogenetic process, but our evaluation shows that this simple binary mixing model fails to explain their geochemical and Pb isotope systematics. We propose, instead, that the Kueishantao HMAs result from partial melting of subducting sediments and subsequent melt-mantle interaction, an interpretation in consistency with seismic tomographic data beneath the SPOT characterized by a complex collision/extension/subduction tectonic context off NE Taiwan.
V13A-1445 1340h
Evidence for Dehydrated Slab Melt Components in the Sources of the Izu Bonin Arc and the Mexican Volcanic Belt
Understanding the pathways of fluids and melts in subduction zones relies on deciphering the processes that generate the unique trace elements patterns of arc magmas. While it is common consensus that these patterns reflect interaction of slab-derived fluid and/or melt components with the subarc mantle, the details of these processes remain highly controversial. One approach towards solving the problem is searching comprehensive high quality data sets that are now becoming available from the global spectrum of arcs. Should common signatures emerge through the veil of the structural and compositional diversity, those must have general meaning for arc magma genesis. Similar slab melt components appear to exist in the rear-arc region of the intra-oceanic Izu Bonin Volcanic Arc and the arc front region near Zitacuaro of the continental Mexican Volcanic Belt. These arcs are opposites in the global spectrum of volcanic arcs: they differ widely in a range of tectonic and compositional parameters, among them the thickness and composition of the crust, as well as the composition of the subarc mantle and the age and composition of the subducting slab. In either arc, a subordinate group of K-rich mafic lavas exist that - despite displaying certain key characteristics of oceanic and continental arc magmas - differ systematically from the dominant calc-alkaline and tholeiitic arc series. In both arcs, the K-rich subgroups show typical oceanic vs. continental arc differences in the abundances of CaO, Na$_{2}$O, FeO and highly incompatible trace elements. However, relative to the calc-alkaline and tholeitic series, the K-rich subgroups have always lower SiO$_{2}$, higher MgO and mg\#, as well as lower ratios of highly fluid mobile elements to rare earth elements (e.g. lower Pb/Nd, Li/Yb), higher LREE abundances, steeper REE patterns and often higher MREE/HREE ratios. However, their prominent negative Nb-Ta and Hf-Zr anomalies clearly attest to a subduction influence. In both arcs, the Pb-Sr-Nd isotope systematics are consistent with all arc magmas being mixtures of slab (sediment, oceanic crust) and mantle components. However, in the Nd/Pb vs Pb isotope diagram, the high-K subgroups are systematically displaced to higher Nd/Pb values at comparable Pb isotopes ratios relative to the calc-alkaline series. We suggest that the high-K subgroups originate from mantle source regions that were infiltrated by composite slab melts (sediment and oceanic crust) derived from a previously dehydrated slab. Since the contribution of the sediment to the composite slab melt is comparatively minor ($<$8% in Izu Arc, and $<$20% in the Mexican Volcanic Belt), the negative Nb-Ta and Hf-Zr anomalies cannot be inherited from the sediment melt component alone. Therefore, additional fractionation is required in order to generate the distinct trace element patterns of the high-K subgroups that may occur either during slab melting or during melt/rock interaction in the mantle. In contrast, the calc-alkaline and tholeiitic series show a stronger influence of slab fluid and sediment melt components. Owing to their end member character in either arc, the presence of the high-K subgroups support hence models of progressive slab and wedge evolution for arcs.
V13A-1446 1340h
Light Elements Across the Central American Arc, SE Guatemala
We have examined B-Be-Li systematics of volcanoes in southeastern Guatemala which represent a nearly continuous 120 km cross-arc transect of the active Central American Arc. Volcanoes behind the front are small monogenetic cinder cones erupting basaltic magmas that show compositional differences when compared to volcanic front lavas. Boron and Li abundances, B/Be (&<&20 behind the front versus 20-75 at the volcanic front) and Li/Yb (&<&5 behind the front versus 5-10 at the front) are generally higher in volcanic centers along the Guatemalan volcanic front, and are generally lower in behind the front volcanoes. All B/Be and Li/Yb ratios are elevated relative to MORB values, suggesting inputs from the slab both at the volcanic front and behind the front. No consistent B/Be or Li/Yb variation is observed with increasing distance from the front, as is seen in other arcs. B and Li systematics are similar to those of other fluid mobile elements in Guatemalan volcanoes. Be abundances, by contrast, are higher in behind the front volcanoes, and show systematics similar to those of Nb and other incompatible elements. Be/Nd ratios range from 0.03 - 0.08, similar to other calc-alkaline arc lavas. Be in Guatemalan lavas is thus controlled largely by changing extents of partial melting of a source with uniform Be contents. Some volcanoes at the front, from Pacaya to Fuego, record lower B concentrations and lower B/Be, similar to those of behind the front centers, suggesting overall lower slab inputs to the source regions in this section of the arc. Our data support the model of Walker et al (1995), in which melting processes abruptly change across the arc in Guatemala, from flux dominant melting at the front to decompression dominant melting behind the front. However, some level of slab input, as indicated by elevated B and Li ratios, is present behind the front. Li isotope variations are being examined in these lavas to try to identify the signatures in slab-modified mantle in both volcanic front and behind the front Guatemalan centers.
V13A-1447 1340h
Composition of fluids during serpentinite breakdown in subduction zones: Evidence for limited boron mobility
Subduction of serpentinized oceanic lithosphere provides an important mechanism by which H$_{2}$O and possibly boron is introduced to the deep mantle. Although it has been suggested that release of H$_{2}$O during serpentinite dehydration may govern certain physical processes in subduction zones, the geochemical signature of this fluid and the olivine-rich residue have largely been ignored in terms of their importance to arc magma genesis and the recycling of oceanic crust, respectively. In this work, we present results from novel piston-cylinder experiments designed to characterize the mobility of B and other trace elements during dehydration of serpentinite. Fluids derived from the serpentine breakdown reaction were drained into a porous diamond trap, thereby segregating fluid and solid domains. Following each run, the olivine/opx residue and the quench precipitates within the diamond trap were analysed for trace elements using laser ablation ICP-MS (LA). Three experiments were conducted at 30 kbar and 750$\deg$C for a duration of one week. LA analyses show that elements highly mobile in aqueous solutions (Li, As, Cs, Ba, Rb, Pb) are strongly depleted in the olivine-opx residue indicating that they were dissolved in the aqueous fluid produced by serpentine break down. From these data, we determined the fluid/residue partition coefficients ($^{F/R}$D) to be 10-250. The surprising exception is B, which is only enriched in the fluid by a factor of 2 to 4 times, with respect to the residue, corresponding to $^{F/R}$D = 3-5. This result is in marked contrast to previous studies on B partitioning, which suggest that K$_{D}$ for boron is $>>$ 1 during reaction at high temperatures. The difference in B behaviour observed in our experiments is likely due to the involvement of olivine, which is capable of incorporating significant boron. The results from our experiments have widespread implications with regard to boron recycling throughout the Earth and suggest that high-B olivines from xenoliths or ophiolitic sections are a fingerprint for metasomatized or previously serpentized mantle Our results also suggest that significant B is recycled into the deep mantle, and that fluids migrating into the olivine-rich mantle wedge may undergo partial B removal during recrystallization processes.
V13A-1448 1340h
The Role of Water in Arc Magmatism in Nicaragua and Costa Rica.
The Central American Volcanic Arc provides an excellent opportunity to study the effect of varying subduction parameters on arc magma geochemistry. Here we examine melt inclusions trapped in olivine phenocrysts in young mafic tephras from the active arc segments in Nicaragua and Costa Rica. These segments of the subduction zone provide starkly contrasting situations with differing oceanic crustal input (East Pacific Rise vs. oceanic crust overprinted by the Galapagos hot spot) and slab dip ($\sim70\deg$ vs. $\sim30\deg$) in Nicaragua and Costa Rica respectively. We collected major-, trace-, and volatile-element concentration data from olivine-hosted (Fo$_{68-87})$ melt inclusions from volcanoes Cerro Negro, Nejapa, Masaya, and Mombacho along the volcanic front in central Nicaragua and Irazu and Arenal in Costa Rica. In Nicaragua H$_{2}$O shows a general increase with decreasing Fo of the host olivine, reflecting incompatible behavior during differentiation from 1-3 wt. $%$ in samples from Fo$_{>85}$ to $\sim 5%$ in some of the slightly more evolved, yet still relatively undegassed, samples. This is best illustrated by samples from Mombacho and Cerro Negro. Comparison of the major element data with experimental melting studies indicates that the Nicaraguan melts with higher FeO and lower SiO$_{2}$ (at similar MgO contents) could have formed at greater depths and temperatures in the presence of less water than the Costa Rican melts with higher SiO$_{2}$ and lower FeO. The most primitive samples from Irazu (host olivine Fo$_{84-89)}$ are extremely rich in H$_{2}$O (3-5 wt. $%$). H$_{2}$O enrichment correlates with enrichment of some LILE, LREE (i.e., Ba, La, K), relative to HFSE for most samples from the Nicaraguan volcanoes. S/Nb and Cl/Nb also show some correlation with H$_{2}$O/Nb in Nicaraguan samples, with some exceptions possibly related to volatile loss is the less primitive samples. These correlations are generally not present in the Costa Rican samples, thus suggesting that despite apparent larger initial water contents in Costa Rican magmas, many Nicaraguan trace element enrichments are derived mainly through input of hydrous fluids whereas Costa Rican trace element budgets may require additional processes.
V13A-1449 1340h
Volatiles in a Subduction-Related Primitive Basaltic Cinder Cone: Investigating Volcan Jorullo, Mexico
To investigate volatiles in primitive subduction-related basaltic magmas, we have analyzed volatile (H$_{2}$O, CO$_{2}$, Cl, S) concentrations in olivine-hosted melt inclusions from the 1759-1774 eruption of Jorullo volcano in the central Trans-Mexican Volcanic Belt (TMVB). Jorullo's earliest lavas are primitive (9.3 wt% MgO, Fo$_{86-90}$ olivine), and lava compositions evolved over time as a result of crystal fractionation (Luhr and Carmichael, 1985, CMP). Tephra samples were collected from a 5-meter-thick proximal ash fall sequence. Olivine crystals from the base of the section are fractured and contain clusters and chains of Cr-spinel inclusions in addition to melt inclusions. Samples from the top of the section have more euhedral crystals with solitary Cr-spinel inclusions and fewer melt inclusions. Melt inclusions from the basal tephra have variable H$_{2}$O ($<$1-4.8 wt%) and CO$_{2}$ (34-770 ppm), corresponding to crystallization pressures of $<$100 bars to 3.7 kbars. This indicates that olivine crystallized over a wide range of depths, trapping variably degassed melts during magma ascent. Melt inclusions from the upper sample have lower H$_{2}$O (0.2-1.4 wt%) and no detectable CO$_{2}$, suggesting shallow crystallization of degassed magma toward the end of the violent-Strombolian-style eruptions. For Jorullo melt inclusions, the maximum H$_{2}$O contents (4-5 wt%), which should most closely represent primary values, compare with values of $\sim$4 wt% H$_{2}$O at nearby Paricutin (Luhr, 2001, CMP) and 1.3-5.2 wt% in cinder cones in the Chichinautzin volcanic field to the east (Cervantes and Wallace, 2003, Geology). Relatively high Ba/Nb in Jorullo lavas shows that the underlying mantle wedge in this region of the TMVB has been enriched by a subduction-derived component. Ratios of H$_{2}$O to incompatible trace elements follow similar patterns to those observed for Chichinautzin cinder cones, and thus further demonstrate a link between H$_{2}$O and trace element enrichment in the mantle wedge. The high H$_{2}$O in Jorullo basaltic magma is surprising given the low B content of whole rock samples (Hochstaedter et al., 1996, GCA). The latter has been attributed to subduction of young, relatively hot lithosphere beneath the TMVB, which should result in significant slab devolatilization beneath the forearc region. Our results show that despite early loss of B from subducted materials in such "hot" subduction zones, large amounts of H$_{2}$O continue to be released from the slab further down dip, resulting in flux melting of the mantle wedge beneath the volcanic arc.
V13A-1450 1340h
Cycling of Li, K, Rb, and Cs at Subduction Zones and Ridge Crests With Implications for Ocean Chemistry: Hydrothermal Experiments at 35-350$\deg$C and 600 bars
The alkali metals (Li, K, Rb, Cs) are excellent tracers of fluid-rock reactions and cycling because of strong partitioning into the fluid phase, particularly at moderate to high temperatures. As such, they can provide critical information on the process of fluid recycling at the principal plate boundary, the reaction site temperature, the nature of the reacting solid phases (i.e. the involvement of sediment in arc volcanoes), with links to seawater (SW) chemistry. Our hydrothermal experiments indicate that the reactivity of each alkali metal is distinct; each has a characteristic behavior with respect to partitioning into the fluid phase with temperature. The data thus suggest that alkali concentration ratios together with Li isotopes may be used for geothermometry. MORB-SW and smectite-SW hydrothermal experiments were conducted between $35-350\deg$C at $25\deg$C, 600 bars, and a water/rock mass ratio of $\sim$5, using the Dickson-type rocking autoclave. Run-times continued until steady-state partitioning was observed. In the MORB-SW experiments, fluid K/Cl molar ratios first increase slightly, from 35-$65\deg$C, then decrease to a minimum of 1.1 x10$^{-2}$ at $275\deg$C. At $>$$275\deg$C the ratio sharply increases to 6.0 x10$^{-2}$ at $350\deg$C. In contrast, Li/Cl, Rb/Cl, and Cs/Cl ratios all increase from 35-$350\deg$C, but each at a distinct slope, indicating significant fractionation between the alkali metals. The Li/Cl ratio gently increases between 35-$250\deg$C, and sharply increases to a ratio of $\sim$1.20 x10$^{-3}$ at $350\deg$C. Rb/Cl behaves similarly, except for a higher inflection temperature of $\sim$$300\deg$C, and a steeper slope between 300-$350\deg$C. Cs/Cl behaves distinctly, the ratio increasing steadily with temperature (i.e. no inflection point) to 1.8 x10$^{-8}$ at $350\deg$C. Consequently, Li/K and Cs/K ratios exhibit sharp maxima at $\sim$$275\deg$C, at 35 x10$^{-3}$ $&$ 10 x10$^{-6}$, respectively. In the smectite-SW experiments there is no reversal in slope, thus the sediment influence on the fluid chemistry can be clearly identified. Rb/K, in contrast to Li/K and Cs/K, do not have a maximum, but steadily increase from 35-$350\deg$C in the MORB experiments. In the smectite experiments, the fluid Rb/K ratios are considerably higher, hence, are as well indicators of sediment contribution. At $350\deg$C the enrichment factors (ER) of Li/Cl and Cs/Cl are higher in the smectite than in the MORB experiments ($\sim$70 & 150 vs. $\sim$20 & 30, respectively). The ER of Rb/Cl is $\sim$10 in both experiments, but that of K/Cl is higher in the MORB vs. smectite experiments (3.0 & 1.7, respectively). Like Sr, the fluid Li totally exchanges its isotopes with MORB or with the smectite. In the MORB experiments, the fluid $\delta$$^{7}$Li fall on a SW-MORB mixing line, shifting from 30.95$\permil$, the SW value, to17.4$\permil$ at $150\deg$C, and 7.33$\permil$ at $350\deg$C. Hence, $\delta$$^{7}$Li values together with Li/K, Li/Rb and Li/Cs ratios constrain the reaction temperature ($\pm$20-$30\deg$C) and the nature of the rock involved. Applications of the experimental data (1) to pore fluids from two subduction zones: the Middle America Trench offshore Costa Rica, and the Nankai Trough, and to an 'average' ridge crest hydrothermal fluid composition, and (2) to revised Rb and Cs ocean budgets, will be presented.
V13A-1451 1340h
The Influence of Subducted Sediment Composition on the Volcanic gas Output at Arcs: Comparing the Luzon and Bicol Arcs, the Philippines
We asses the influence of subducted sediment composition on the volcanic gas output (chemistry and isotope characteristics) by contrasting two juxtaposed volcanic arcs in the Philippines: the western Luzon Arc sampled at the Bulalo geothermal field (flank of Mt. Makiling) and Mt. Pinatubo, and the eastern Bicol Arc sampled at the Tiwi geothermal field (Mt. Malinao). Gases show no distinction in general chemical characteristics between the two arcs with overlapping values in CO$_2$/S ~ 2-37; S/HCl ~ 8-18; N$_2$/He <1,000; CO$_2$/N$_2$ ~ 70-1400 and CH$_4$/He ~67-195. These results are consistent with a varying hydrothermal influence superimposed on arc-type magmatic gas. $^3$He/$^4$He ratios are also similar between the two arcs (Bulalo ~ 6.9 R$_A$, n=6; Pinatubo ~ 7.1 R$_A$, n=2; and Tiwi ~ 6.3 R$_A$, n=3). There is, however, a marked difference between the arcs when comparing carbon characteristics. Geothermal gases from Bulalo (n=4) are tightly constrained in CO$_2$/$^3$He (8 -10 x 10$^9$) whereas Tiwi (n=4) displays much greater variability (11-42 x 10$^9$). In addition, $\delta$$^1$$^3$C values of Bulalo span the range typical of arc-derived carbon (-2.7 to -4.1$\permil$) whereas $\delta$$^1$$^3$C at Tiwi is significantly higher at -0.6 to -0.2$\permil$. Interestingly, Pinatubo gases show CO$_2$/$^3$He values closer to Tiwi (40 x 10$^9$) but $\delta$$^1$$^3$C values closer to Bulalo (-4.8$\permil$). Apportioning CO$_2$ between slab-derived limestone (L) and organic sediment (S), and mantle wedge (M) (see Sano and Marty, Chem. Geol., 1995) then the Luzon Arc is similar to other arcs worldwide (L = 70-80%; M = 10-20% and S = 8-12%). This likely reflects heterogeneity in sediment sources derived from Eurasia and subducted via the Manila Trench. In contrast, CO$_2$ from Tiwi can be described as a binary mixture between L and M (~88:12) with no evidence of organic sediment (S) in its source. Pelagic sediment of the Philippine Plate is the principal carbon-bearing species subducted via the East Luzon Trench and the organic C-content is low. Hence, the Philippines provide a clear example where source composition of the down-going sediment exerts the primary control on the carbon output characteristics: this is not the case for the general gas chemistry. We compare these findings with other arcs worldwide.
V13A-1452 1340h
Modeling Open System Metamorphic Devolatilization of Subducted Metabasalts
Fluids derived from the devolatilization of a subducting slab play a critical role in the melting of the mantle wedge and consequent arc volcanism. In this study we present results from thermodynamic modeling of subduction zone metamorphic devolatilization of metabasalts for pressures up to 6 GPa using an approach which considers fluid fractionation from source lithologies and infiltration from subjacent lithologies. This open system approach offers a more realistic model of subduction zone devolatilization than closed system models using similar computational techniques. Along high temperature geotherms, decarbonation is complete in the case where the maximum H$_{2}$O flux likely to be achieved in subduction settings is assumed. In the case of a dry mantle and lower crust, decarbonation along high temperature geotherms is limited to $\sim$20 wt.% of original carbonate content. Along low and intermediate temperature geotherms, decarbonation is insignificant except in the case of maximal H$_{2}$O infiltration where carbonate loss reaches a maximum of $\sim$20%. Relative to closed system behavior, our open system modeling leads to more extensive decarbonation along high and intermediate temperature geotherms, whereas decarbonation along low temperature geotherms is not appreciably different. Because our model assumes pervasive compaction-driven fluid flow, the effect of fluid infiltration is maximized and therefore it is possible that the results presented here overestimate the extent of infiltration induced decarbonation in subducting crust. Subsequently, fluid propagation models which account for channelized flow could diminish the magnitude of carbonate loss compared to what is predicted in this study. The extent of metabasalt dehydration for open system behavior is not significantly different when compared to closed system devolatilization for all geotherms considered in this study. Because metabasalts undergo appreciable dehydration under all but the coldest geotherms, this lithology provides a fertile H$_{2}$O source for infiltration driven decarbonation of overlying marine sediments as well as fluid induced partial melting in the mantle wedge. Due to limited decarbonation along low and intermediate temperature geotherms and limited dehydration along low temperature geotherms, oceanic metabasalts represent an effective lithology for recycling of volatiles into the deep mantle.
V13A-1453 1340h
Petrology and Geochemistry of Eclogite Xenoliths from the Colorado Plateau: Implications for the Evolution of Subducted Oceanic Crust
Eclogite xenoliths from the Colorado Plateau, interpreted as representing fragments of the subducted Farallon plate, are used to infer the evolution of trace element and isotopic compositions of oceanic crust subducted into the subarc mantle. The xenoliths consist of almandine-rich garnet, Na-clinopyroxene, lawsonite and zoisite with minor amounts of phengite, rutile, pyrite and zircon. They have essentially basaltic bulk-rock major element compositions, except for significantly higher Na$_{2}$O, but similar K$_{2}$O, with increasing SiO$_{2}$ contents, compared to altered MORB. These major element characteristics are explained by spilitization/albitization during hydrothermal alteration at a mid-ocean ridge and by subduction zone metasomatism in the forearc region. Whole-rock trace elements and Sr, Nd, and Pb isotopic compositions of the xenoliths are variable and exhibit significantly enriched characteristics compared to altered MORB, except for similar Zr/Hf ratios of 36.9 to 37.6. Mass balance for the Colorado Plateau eclogite xenoliths can be achieved for 26 trace elements, Rb, Cs, Sr, Ba, Y, REE, HFSE (Zr, Hf, Nb, and Ta), Pb, Th and U. The mass balance calculations and mineralogical observations corroborate that whole-rock chemistry of the xenoliths were contaminated by near-surface processes after eruption and limited interaction with the serpentinized ultramafic microbreccia host magma. Thus, Sr, Nd and Pb isotopic compositions of separated minerals from the xenoliths were measured to avoid these secondary effects; these separates yield distinctively enriched isotopic compositions in the range of 0.70502 to 0.70590 for $^{87}$Sr/$^{86}$Sr, -1.5 to -3.1 for $\epsilon$Nd and 18.928 to 19.052 for $^{206}$Pb/$^{204}$Pb. This suggests that the xenoliths were metasomatized by a fluid equilibrated with the sedimentary layer probably covering the Farallon plate in the forearc region. This metasomatism resulted in the xenoliths acquiring distinctively enriched isotopic compositions compared with those of altered MORB. Some of the distinct isotopic signatures observed in OIBs compared to those from MORBs have been interpreted as a result of oceanic sediment subducted deep into the mantle. Our results, on the contrary, suggest an alternative possibility that these anomalous isotopic reservoirs in the mantle are formed by the subduction of oceanic crust modified by the metamorphic fluid from the covering sedimentary rocks.
V13A-1454 1340h
Steam Cooking of the Andes
Subduction related orogeny at the pacific margin has lead to one of the largest and highest plateaus on Earth - the Central Andes. This high plateau is characterized by geophysical anomalies, such as a high heat flow density exceeding 100 mW/m$^{2}$, a crustal thickness of about 70 km, a high Buguer anomaly, an extreme electrical conductivity anomaly, and low velocity zones. To better understand the underlying processes generating this high plateau, we need to identify and quantify the controlling parameters of the Andean evolution. It has been shown by Springer 1999 that conductive heat transfer alone is not able to explain the observed heat flow density. Babeyko et al. 2002 discuss additional heat flow through an enhanced mantle heat flux in addition to crustal convection. However, a portion of the observed heat flux remains still unresolved. We'll present a simple model to explain the observed heat flux by advective heat of water through the mantle wedge. Part of the high amount of fluids which is subducted with the down-going slab is released beneath the mantle wedge. The ascending water is heated by the hot mantle wedge. The fluid transports heat into the crust leading to an enhanced heat flow density. This steam cooking hypothesis leads to a number of consequences, which allows to explain the triggering of the Central Andean evolution, the high heat flux, partial melting of the crust, and the strong variation from North to South of geological, petrological, and geophysical observations. Babeyko, AY et al. (2002), EPSL, 199, 373-388 Springer M (1999) Tectonophysics, 306, 377-395
V13A-1455 1340h
From source to surface: An Os isotope study of the transfer of mantle source signatures in subduction-related melts
Osmium isotopic data for subduction zone magmas span a wide range from mantle-like to very radiogenic values. While passage through continental crust likely influences a magma's $^{187}$Os/$^{188}$Os signature towards higher values, especially in evolved, low-Os magmas, what is unclear, in situations where slab-melting has been proposed, is the influence of the subducted slab on $^{187}$Os/$^{188}$Os at the magma source. Because osmium is much more concentrated in the mantle than in most subduction zone magmas, and because of the compatible nature of osmium in mantle peridotites, it is possible that any slab contribution to $^{187}$Os/$^{188}$Os will be completely obliterated during percolation of the magma through the mantle wedge, i.e., the magma reaching the crust is buffered to mantle wedge values. Examination of osmium isotopic systematics in peridotite-hosted pyroxenites in the Beni Bousera massif, Morocco suggests that this expectation may be unfounded. These pyroxenite layers have previously been shown to represent the crystallized products of magmas derived from remelted lithosphere in an ancient subduction zone (Pearson et al, 1993). While the $^{187}$Os/$^{188}$Os ratios in the margins of the pyroxenite layers have been buffered by interaction with the host peridotite, the interior portions of pyroxenite layers are generally much too radiogenic to have been significantly altered by mantle buffering. This indicates that while initial melts flowing through veins in the mantle interact and are affected by the peridotite wall rock, subsequent melt flow is armoured from these effects by the reacted margins. The highly radiogenic $^{187}$Os/$^{188}$Os of the pyroxenite interiors must reflect the composition of the source, rather than crustal contamination. These, together with previously published observations (Becker et al., 2004) show how highly radiogenic melt signatures in mantle melts may be transferred to the crust without losing their identity. Recent analyses of osmium isotopic compositions of adakites from Mindanao, Philippines reveal surprisingly low $^{187}$Os/$^{188}$Os ratios - too low to have been derived from remelting 50-million-year old oceanic crust. The results from the Beni Bousera pyroxenites indicate the likelihood that the mantle-like $^{187}$Os/$^{188}$Os signatures of the Mindanao adakites reflect their source composition. Becker et al., 2004. Chem. Geol. 208:141-156 Pearson et al., 1993 J. Petrol. 34:125-172
V13A-1456 1340h
Slab Devolatilization and Os and Pb Mobility in the Mantle Wedge of the Kamchatka Arc
Os-, Nd-, and Pb-isotope measurements were performed on a suite of eleven Kamchatka peridotite mantle xenoliths in andesitic volcanics from the southern (Avachinsky) and central (Sheveluch and Kharchinsky) segments of the Kamchatka peninsula. The peridotite xenoliths studied are mostly wehrlites, harzburgites and dunites. In our previous study of these xenoliths (Basu et al., 2004), petrographic evidence as well as He-, Sr-isotopic, CO$_{2}$ and trace element concentration data were interpreted to indicate that fluids derived from seawater-altered lithosphere moved through the mantle wedge beneath Kamchatka. Peridotites with higher Os content (1.3- 5.2 ppb) show a narrow $^{187}$Os/$^{188}$Os range between 0.1182 and 0.1272 while the xenoliths with lower Os content ($<$ 1 ppb) show a wider variation and more radiogenic $^{187}$Os/$^{188}$Os values (0.1287- 0.1585). $^{187}$Os/$^{188}$Os and $^{3}$He/$^{4}$He (R/Ra) values of the xenoliths indicate that the xenoliths are metasomatized by the slab-derived fluids from the subducting altered oceanic crust. The original Os signature of some of the mantle wedge xenoliths seems to be eradicated and imprinted by the slab fluids from the subducting crust. The Nd-isotopic (+8.7 - +9.3) and Pb-isotopic ratios ($^{206}$Pb/$^{204}$Pb = 18.25- 18.76; $^{207}$Pb/$^{204}$Pb = 15.43- 15.55; $^{208}$Pb/$^{204}$Pb = 37.68- 38.24) of the xenoliths are similar to those of the Pacific MORB. Pb and Ba contents of the xenoliths increase with increasing $^{187}$Os/$^{188}$Os . The Os- and Pb-isotopic data of these xenoliths combined with the He-isotopic data (Basu et al., 2004), indicate that the radiogenic Os-signature and the higher Ba and Pb concentration in these rocks are inherited from the devolatilized altered lithospheric slab. Thus, the mantle-wedge beneath Kamchatka was infiltrated by slab-derived fluids with Os, Pb, and other volatile and fluid-mobile elements.
V13A-1457 1340h
Along Strike Hf-Nd Isotopic Composition of Aleutian Arc Lavas
The Aleutian Arc provides an ideal tectonic setting to study Hf-Nd systematics due to its along-strike tectonic and geochemical variation. Sediment load and subduction rate decrease westward along the arc with an increasingly oblique subduction angle, causing a progressive decrease westward in the flux of sediment into the subduction zone. A suite of samples collected across the length of the arc were analyzed for Hf and Nd isotopic composition. A coupled decrease in both Hf and Nd isotopic composition was observed from east to west along the arc. Piip seamount samples at 193\deg W have the highest $\epsilon$$_{Hf}$20.3 and $\epsilon$$_{Nd}$10.4. Lavas from Great Sitkin (176\deg W) and Korovin (174\deg W) have intermediate compositions of 7.4-7.9 $\epsilon$$_{Nd}$and 13.6-14.4 $\epsilon$$_{Hf}$. The lowest $\epsilon$$_{Hf}$ values obtained (12.2-13.0) were from the Cold Bay Volcanic Complex (163\deg W). Various ashes of the DSDP site 178 of the Alaskan abyssal plain have approximately the same Hf and Nd isotopic composition as Cold Bay. Clay and turbidite sequences from site 178 have lower $\epsilon$$_{Hf}$0.8-7.8 and $\epsilon$$_{Nd}$2.8-0.82 values. The change in Hf and Nd isotopic composition along the arc is consistent with a decrease in sediment flux westward. A sediment melt component in the east could mobilize Hf and Nd into the mantle wedge. Fluxing of the subducting sediment could also incorporate Nd into the melting mantle. Both these processes would lower the radiogenic signature of the source region.
V13A-1458 1340h
Hf-Nd Isotope Characterization of the Tonga-Kermadec Sub-Arc Mantle
Subduction-related magmatism involves fluxing of the mantle wedge by a slab-derived fluid and/or melt. Uncertainty surrounds the thermal, physical and chemical state of the slab and slab-wedge interface beneath the volcanic front, creating doubt over the application of experimental data on slab dehydration and melting to models of magma genesis and mass balance. Constraining the composition of the wedge should provide an important control on such models. However, simple models of wedge convection from the backarc to sub-arc environment are clearly incomplete (e.g. do not explain the lack of correlation between depletion of the sub-arc mantle and either backarc width or extension rate). To investigate the dynamics of wedge convection, we undertook Hf-Nd isotope determinations for lavas spanning most of the Tonga-Kermadec arc. Advantages of these elements include their fluid-immobile character (Hf strongly, Nd moderately) relative to the highly mobile, and thus predominantly slab-derived, Sr and Pb. Previous Sr-Nd-Pb isotope studies of backarc Lau Basin lavas revealed that Indian MORB mantle (IMM) replaced Pacific MORB mantle (PMM) during basin opening, and that IMM underlies the central Tonga arc. In contrast, the backarc mantle south of Valu Fa Ridge and under the Havre Trough remains PMM. Thus, a transition from IMM to PMM was anticipated beneath the south Tonga arc. Hf isotope compositions of the Tonga-Kermadec arc lavas range from +12 to +16 $\epsilon$-units and are 1-2 $\epsilon$Hf units below those of the Lau Basin spreading centres. Tonga lavas are generally in the upper part of this range, whereas Kermadec lavas are more varied. All Tonga-Kermadec lavas plot within the IMM field in $\epsilon$Hf vs $\epsilon$Nd-space, indicating IMM-like mantle is under the entire arc from $15-35\deg$S. This result is further enhanced by subtracting the $\sim$20-40 % Nd added from the subducting slab (PMM-composition). $\epsilon$Hf exhibits a mild negative correlation with Hf/Yb and increasing latitude, consistent with a minor Hf contribution from subducting volcanigenic sediment in the Kermadec arc sector. Our data provide no evidence for a slab melt, even in the southern Kermadecs where the slab is warmer due to slow subduction. The occurrence of IMM beneath the Tonga-Kermadec arc, but PMM below the southern Lau Basin-Havre Trough, is inconsistent with simple models invoking progressive southeastward migration of IMM into the Tonga-Kermadec system accompanying the opening of the Lau-Havre backarc basin. Instead, considerably complexity is required in the geometry of the IMM-PMM interface, the sub-arc wedge flow, or the origin of the IMM-like signature.
V13A-1459 1340h
Experimentally Determined Trace Element Partitioning between Rutile and Melt: Implications for Subduction Zone Processes
Mineral/melt trace element partition coefficients were determined experimentally for rutile (TiO$_{2}$) for a large number of trace elements (Zr, Hf, Nb, Ta, V, Co, Cu, Zn, Sr, REE, Cr, Sb, W, U, Th). Whilst the high field strength elements (Zr, Hf, Nb, Ta) are compatible in rutile, most other studied trace elements are incompatible (Co, Zn, Sr, Th, REE). In all experiments we found D$_{Ta}$ $>$ D$_{Nb}$, D$_{Hf}$ $>$ D$_{Zr}$ and D$_{U}$ $>$ D$_{Th}$. Partition coefficients for some polyvalent elements (Sb, W, V) were, as expected, rather sensitive to oxygen fugacity. Melt composition exerts a strong influence on HFSE partition coefficients. With increasing polymerisation of the melt, rutile/melt partition coefficients for the high field strength elements Zr, Hf, Nb and Ta increase about an order of magnitude. However, D$_{Nb}$/D$_{Ta}$ and D$_{Hf}$/D$_{Zr}$ appear not to be affected by melt composition. Because D$_{U}$ $>$ D$_{Th}$, partial melting of rutile-bearing eclogite in subducted lithosphere may cause excesses of $^{230}$Th over $^{238}$U in some island arc lavas, whereas dehydration of subducted lithosphere may cause excesses of $^{238}$U over $^{230}$Th. From our partitioning results we infer partition coefficients for Pa which we predict to be much lower than previously anticipated. Contrary to previous studies, our data imply that rutile should not significantly influence observed $^{231}$Pa-$^{235}$U disequilibria in certain volcanic rocks.
V13A-1460 1340h
Subduction Zone Mass Fluxes
Recycled crustal material is often cited as a source for various mantle reservoir characteristics. These inferences are often based on ingoing slab compositions, although processes within the subduction zone could significantly change the chemistry of the slab before it is mixed into the mantle. Few studies have examined the composition of "residual" slab material. Using a mass-balance approach, we have estimated the chemical compositions of residual slabs at nine oceanic subduction zones (Aleutians, Central America, Izu-Bonin, Kurile, Mariana, Northern Lesser Antilles, Southern Lesser Antilles, Sunda-Java, and Tonga), representing approximately 30% of the total length of subduction zones worldwide. We remove an inferred mantle wedge component from fractionation-corrected average island arc volcanics (IAV) in order to estimate the composition and flux of the "slab component" in the IAV. We then calculate the residual slab composition by subtracting the "slab component" flux from the ingoing slab flux. We have also calculated slab-to-backarc fluxes at the Marianas and Tonga island arcs in order to more accurately constrain residual slab fluxes. Our results allow us to investigate the feasibility of various mantle evolution models. The HIMU mantle reservoir gives rise to basalts with very high $^{206}$Pb/$^{204}$Pb and somewhat low $^{87}$Sr/$^{86}$Sr. It has often been suggested that this reservoir represent anciently subducted oceanic crust. We use our calculated residual slab U/Pb, Rb/Sr, and Sm/Nd values to estimate the present isotopic composition of material subducted 1.8 Ga ago. Only two of the arcs (Izu-Bonin and the Marianas) have high enough residual slab U/Pb to produce HIMU-like Pb isotope signatures. However, Rb/Sr ratios in these arcs are too high to be consistent with HIMU characteristics, and $\epsilon_{Nd}$ is also inconsistent with the HIMU reservoir. Galer and O'Nions (1985) argued that to account for the difference between present $^{232}$Th/$^{238}$U ratios (kappa) and the time-integrated $^{232}$Th/$^{238}$U calculated from Pb isotope ratios ($\kappa _{Pb}$) in MORB, the residence time of Pb in the depleted mantle must be short (about 600 Ma). Others have argued instead that low present-day kappa in the depleted mantle reflects preferential recycling of U via subduction of low-Th/U altered crust. Our results indicate that subduction processing does not fractionate Th from U; therefore, a residual slab will only contribute low Th/U material to the mantle if the ingoing slab has a low Th/U ratio. Low crustal Th/U is only expected in oxidizing ocean conditions. However, recent studies indicate that the oceans may have remained fairly anoxic for much of Earth's history (until about 600 Ma). If this was the case, it is unlikely that oceanic crust could have caused the lowering of mantle Th/U ratios. The Pb/Ce ratio has been used as an indicator of recycled material in the mantle because of the large contrast in crust and mantle values (0.30 vs. 0.035). Most basalts derived from mantle plumes have Pb/Ce ratios similar to those of MORB (about 0.04). This observation is apparently inconsistent with the idea that mantle plumes contain recycled crustal material, since subducted slabs tend to have high Pb/Ce (greater than 0.20) due to their sedimentary component. Our results show that, although subduction zone processing can lower slab Pb/Ce ratios, residual slab compositions are still not consistent with the observed mantle plume compositions: the lowest residual slab Pb/Ce we observe is roughly 0.07. It therefore seems likely that other processes must contribute to the lowering of slab Pb/Ce ratios.