V13B-0523
Contrasting Ti Zoning Patterns in Quartz Phenocrysts from the Bishop and Fish Canyon Tuffs: An Expression of Volatile Composition in Recharge Melts
Mafic recharge is increasingly recognized as playing an important role in the thermal and chemical evolution of silicic magma systems. Even in the absence of direct interaction between mafic and silicic magmas, volatiles released from recharge melts may be effective agents of heat exchange ("gas sparging" of Bachmann and Bergantz, 2003). These volatiles will also variably affect - depending on their abundance and composition - phase relations and physical properties of the overlying silicic magma. Addition of H2O to rhyolite melt should decrease solidus temperatures and melt viscosity; addition of CO2-rich bubbles (into which H2O will migrate from the melt) will have the opposite effects. We interpret differences in quartz zoning patterns from the Bishop and Fish Canyon Tuffs - each of which has a near-eutectic, rhyolitic melt composition - to reflect differences in the composition of volatiles added by recharge melts. Late-erupted Bishop Tuff contains phenocrysts of quartz that are reversely zoned with respect to Ti content: low-Ti cores (~40-50 ppm) are separated from high-Ti rims (80-100 ppm) by a sharp boundary that appears to be a dissolution feature. This pattern is best explained by recharge involving CO2-rich basalts typical of the region. Quartz that originally grew at 740°C partially dissolved due to heating and was later reprecipitated at the higher temperatures (780-800°C) of the CO2-elevated solidus. In contrast, Fish Canyon quartz is strongly resorbed, with variable Ti contents (90-120 ppm) but no systematic zoning. Partial dissolution of quartz without growth of high-Ti rims can be explained by the addition of heat and H2O-rich volatiles (which lowered the solidus temperature) released from recharge melts like those extruded after Fish Canyon eruption (the Huerto Andesite).
V13B-0524
The Origin of Quartz Phenocrysts in the Bishop Tuff Rhyolitic Magma Based on Trace Tlements in Melt Inclusions and Pumice Matrix Glass
The origin of crystals in silicic volcanic rocks has been challenged by textural and isotopic studies of phenocrysts and melt inclusions in some silicic ignimbrites. These studies suggest that quartz and feldspar are 'xenocrysts' derived either from re-melting of previously solidified magma bodies or disaggregation from crystal-rich rinds along the margins of the magma body. To investigate the origin of quartz in the Bishop Tuff, we have analyzed trace elements in the matrix glass of pumice clasts and in quartz-hosted melt inclusions. Our results suggest contrasting origins for quartz in different parts of the Bishop Tuff. In all samples from the early part of the eruption (F1-F7, terminology of Wilson and Hildreth, 1997) and tephra fall samples from the middle part (F9), trace element compositions of matrix glasses are similar to but slightly more evolved than quartz-hosted melt inclusions. This indicates a cogenetic relationship between quartz crystals and their surrounding matrix glass, consistent with in situ crystallization. The range of incompatible element concentrations in melt inclusions and matrix glass from single pumice clasts requires 16-20 wt.% in situ crystallization. This is greater than the actual crystal content of the pumices (<15% crystals). Interestingly, pyroclastic flow samples from the middle part of the eruption (Ig2Ea, Ig2Eb) show contrasting trends: in some clasts the matrix is more evolved than the inclusions whereas in other clasts the matrix is less evolved than the inclusions. In the late Bishop Tuff (Ig2Na, Ig2NWa) all crystal-rich samples have matrix glasses that are significantly less evolved than quartz-hosted inclusions. Trace element abundances indicate that most of the quartz in the late Bishop Tuff crystallized from more differentiated rhyolitic magma similar in composition to early erupted Bishop Tuff. We used cathodoluminescence images of melt inclusion-bearing quartz crystals to correlate growth in the host crystal with compositional data from melt inclusions to assess the chronological evolution of melt inclusion compositions. All quartz crystals show a variety of oscillatory zoning textures. Melt inclusions in samples from the early part of the eruption and tephra fall samples from the middle part are less evolved in the core of the crystal and are more evolved closer to the crystal rim, consistent with in situ crystallization of quartz. However, quartz phenocrysts from late-erupted samples have bright luminescent rims that are not observed in early and middle erupted samples (Peppard et al., 2001, Am. Min.). Highly differentiated melt inclusions similar to those in early samples occur within the dark cores whereas less differentiated inclusions are associated with the bright rims. Our results for late erupted Bishop Tuff are compatible with either crystal settling of quartz from shallower levels within the magma body or late-stage mixing of a new batch of less differentiated rhyolitic magma in the deeper levels of the Bishop magma chamber.
V13B-0525
Phase Equilibria Experiments on Nisyros, Greece:''the Right Stuff" in -''the Right Stuff" out
To extract meaningful information from phase equilibrium experiments the starting composition must be made of ''the right stuff'' and the results compared with the ''real phenocrysts''. Nisyros island is one of the most active volcanic centres on the South Aegean Volcanic Arc. Pleistocene to recent volcanism has evolved from submarine to sub-aerial in style, culminating in the large caldera-forming eruptions at ca. 33 ka. Post caldera activity has been confined to the extrusion of viscous domes, which fill almost half of the caldera, and hydrothermal eruptions. Studies on the sub-volcanic plumbing dynamics of Nisyros have shown that it is a complex and poorly constrained system, and the need to constrain these parameters is gaining impetus given that Nisyros is showing signs of unrest. The domes are dacitic (SiO2 68 wt. %). Phenocryst phases in decreasing order of abundance are plagioclase, clinopyroxene, orthopyroxene and Fe-Ti oxides. Amphibole is present only as a xenocryst phase, derived from the disaggregation of andesitic magmatic inclusions which are prominent in the dome rocks. There are no eruption dates for the post-caldera activity, but the domes are probably Holocene in age. Age differentiation between the domes is based solely on the extent of weathering and position relative to adjacent domes. Preliminary U/Th investigations of zircons from the ca. 33 ka caldera phase by ion probe have shown that these are inherited, with ages of 92, +35/-27 ka. Low zircon saturation temperatures ($731-769 ^{\circ}\mathrm{C}$) for the post-caldera domes suggest similarly that any zircons present will have been inherited. Extensive geochemical evidence for magma mixing indicates the need for caution in choosing an appropriate starting composition for the phase equilibria experiments. Simple major element modelling suggests limited bulk mixing between silicic and andesitic end-members. There is also little evidence to suggest that the dacite has been hybridised by the andesitic inclusions, therefore a natural composition (SiO2 69 wt. %) is being used as a starting composition. This composition represents that of the most recent volcanic eruption on the island. A series of phase equilibria experiments are currently being conducted in Cold Seal Pressure Vessels over a temperature range of $750-900 ^{\circ}\mathrm{C}$, at varying pressures and water contents and a range of oxygen fugacities. Results from these will be presented in comparison with the natural assemblage in a search for the real storage conditions for the post-caldera dacites.
V13B-0526
Constraining Pre-Eruptive Storage Conditions at Nea Kameni, Santorini
Santorini is one of the most active volcanic centres on the South Aegean Volcanic Arc. Historic volcanism culminated in the catastrophic eruption in 1600BC which destroyed the Minoan civilisation and generated a water filled caldera 8 km long and 4 km wide. Recent activity has been confined to formation of two intra-caldera shield volcanoes, the Kameni islands, the last eruption occurring in 1950. The Kameni lavas are sparsely porphyritic dacites containing phenocrysts of labradorite, augite, hypersthene and magnetite. Xenocrysts of olivine, anorthite and Mg-rich augite are derived from the disaggregation of mafic xenoliths. The most striking feature of these lavas is their relatively homogeneous nature despite 2000 years of activity. The low variability in silica content (64-68 wt. %) has prompted suggestions that the chamber may be chemically or thermally buffered. Even with numerous petrological studies, little is known about the dynamics of the system. In this instance experimental phase equilibria are unequivocally one of the best ways of establishing the 'real' phenocryst assemblage. This research uses phase equilibria experiments to constrain pre-eruptive storage conditions at Nea Kameni. Using a natural sample from the 1866-1870 Georgios lavas as a starting composition, a series of water-saturated experiments have been undertaken in a rapid quench Cold Seal Pressure Vessel over a temperature and pressure range of 800-900$^${$\circ$}$mathrm{C}$ and 80-150 MPa. Comparison of the experimental phase equilibria with the natural assemblage suggests magma storage at water saturated conditions at >900 $^{\circ$}$mathrm{C}$ and >100 MPa. These results suggest the magma chamber is at a depth of ~2.8-3.5 km, which is comparable to the calculated depth estimates of earlier studies (2-4 km). Further experiments at higher temperatures will constrain the system more closely and these results will also be presented.
V13B-0527
Light Elements in Plagioclase Phenocrysts of the Nea Kameni Dacites: What do they say about processes during and after crystallization, and just how fast crystallization can take place?
Nea Kameni is the larger of two intra-caldera islands in Santorini Volcano, the most active volcanic center of the Hellenic Arc in Greece. Phenocrysts of samples collected from the 1939-1941 dacite lava flows were analyzed for major and trace element compositions using the EMP and SIMS, with particular focus on concentrations of Li, Be and B. Three types of plagioclase phenocrysts are highlighted in this study. Type N is normally-zoned in terms of An content, with values ranging from 62 in the core to 40 at the rim. A general increase in Li, Be, B from core to rim characterizes this plagioclase, with values of 6.6-7.6 μg/g for Li, 0.37-0.75 μg/g for Be, and 0.10-0.18 μg/g for B. Type O crystals exhibit only oscillatory variations in An content, which ranges from 40 to 57. In terms of Li, however, two subtypes have been identified for this population. Type O1 crystals show Li concentrations that markedly decrease from core to rim, with a maximum value of 23.4 μg/g in the core and a minimum of 6.5 μg/g at the rim. Type O2 crystals are more common, where Li patterns are rather flat and have values within 6.6-11.0 μg/g. Be and B in both subtypes do not vary systematically, except for the slight inverse correlation with An content, and have values of 0.54-0.96 μg/g and 0.10-0.24 μg/g, respectively. Type C phenocrysts have inherited cores with higher An content compared to the outer portions. Within the core, An is between 54 and 63 and Li, Be, B tend to increase from the central portion outwards, with values of 6.5-9.6 μg/g, 0.31-0.60 μg/g, and 0.10-0.16 μg/g, respectively. Towards the rim, the profiles are disrupted: An content drops to 40-54, whereas light elements, especially Be, are elevated at the boundary but decrease gradually toward the rim. Values lie within 6.8-13.9 μg/g for Li, 0.61-0.97 μg/g for Be, and 0.14-0.24 μg/g for B. The different plagioclase types exhibit the complexity of processes involved in the formation of these phenocrysts: from normal crystallization with its expected element concentration trends, to degassing of magma causing dramatic decreases in water-soluble elements, to diffusion of trace elements that modify primary compositions within the crystals. Also, magma mixing and entrainment of earlier-formed phases may have developed phenocrysts with distinct chemical disparities. Li profiles of certain crystals yield astonishing results about the residence time of plagioclase. Simple linear diffusion calculation gives spans of seconds to a few hours, compared to the hundred-year scale previously derived from diffusion modelling of other elements (e.g., Sr by Zellmer et. al, 1999; Mg by Costa et al., 2003). This suggests that crystallisation may have been driven by decompression and degassing during eruption, and that these combined processes occurred extremely fast. Although quite tentative, this method challenges our concept of maximum magma residence times and their implication on volcanic hazards. References: Costa F, Chakraborty S, Dohmen R (2003). GCA 67(12): 2189-2200. Zellmer GF, Blake S, Vance D, Hawkesworth C, Turner S (1999). CMP 136: 345-357.
V13B-0528
Nevado de Longavi Dacites: Adakite Genesis by Intra-Crustal Fractional Crystallization in a Cold Subduction Zone
Holocene dacites (5000-7000 y.B.P.; 63-65 wt% SiO2) of Nevado de Longav'{i} volcano (NLV, 36.2°S, Chile) are the only occurrence of adakitic magmas in the central Chilean Andes, showing high Na2O (4-4.6 wt%) and low K2O/Na2O (0.25-0.33), coupled with high Sr/Y (62-81), La/Yb (10-15), and exceptionally low Y and HREE (Y= 7-9 ppm; Yb=0.6-0.9 ppm). NLV dacites have low abundances of incompatibles elements (K, Rb), inconsistent with crustal assimilation, whereas Sr and Nd isotopic ratios suggest a mantle origin for these melts. The mineral assemblage of these dacites involves plagioclase, amphibole, and minor orthopyroxene and Fe-Ti oxides (cpx absent). Similar assemblages, including high volumes of amphibole (30 vol%) are found in quenched mafic enclaves (53-58 wt% SiO2) in dacites, suggesting a broadly co-genetic relationship. Adakitic dacites can be derived from hydrous enclave magmas by 50% fractionation in the deep crust (0.46 amph+0.38 plag+0.07 oliv+0.02 cpx+0.07 opx+0.04 mt+0.08 ap+0.015 garnet). This model is consistent with amphibole-rich cumulate xenoliths in NLV lavas, and garnet fractionation at deep crustal pressures (35-40 km) is supported by experimental results on water-saturated arc-basalt melts [1, 2, and 3]. High B abundances in dacites (37-49 ppm) and mafic enclaves (17-33 ppm) relative to other southern Andean magmas support a scenario in which NLV adakitic dacites are related to transient high inputs of slab-derived fluids from the oceanic Mocha Fracture Zone, perhaps by serpentinite dehydration [4]. Fractional crystallization is a viable mechanism for adakite generation in cold-subduction zones where large fluxes of slab-derived fluids occur. [1] Grove, T.; Elkins-Tanton, L.; Parman, S.; Chatterjee, N.; Muentener, O.; Gaetani, G. 2003. Contrib. Mineral. Petrol., 145, p.515-533; [2] Muentener, O.; Kelemen, P.; Grove, T. 2001. Contrib. Mineral. Petrol., 141, p.643-658; [3] Ulmer, P.; Muentener, O.; Alonso-Perez, R. 2003. Geophys.l Res. Abstract, Vol.5, 08308. [4] Ulmer, P.; Tromsdorff, V. 1995. Science, 268, p.858.
V13B-0529
U-series crystal ages in historic eruptions of Volcan Quizapu, Chile
The time scales of generation, homogenization, and storage of large volumes of intermediate to silicic magmas have been the subject of much recent debate. Crystal ages determined by U/Th, Rb/Sr and U/Pb dating can help delimit these time scales, and for many of these systems are on the order of several hundred thousand years. Voluminous eruptions with SiO2 contents of 65-70 wt.% that are recent enough to employ $^{226}$Ra-230Th dating are rare. We present preliminary U-Th-Ra crystal ages for historic eruptive products of Volcan Quizapu in Central Chile, a satellite vent of Cerro Azul. Quizapu erupted 4-5 km3 of dacite in 1846/47 and again in 1932, of almost identical bulk composition (67-68 wt.-%). The 1932 dacite was accompanied by volumetrically minor rhyodacite and by an initial and terminal andesite, which most likely was not stored in the main magmatic reservoir (Hildreth and Drake, 1992, Bull. Volc. 54:93-125). Plagioclase and glass separates for the 1932 rhyodacite and the 1846 dacite show 230Th-$^{232}$Th ratios indistinguishable from each other and within error of 230Th-$^{238}$U radioactive equilibrium. When an amphibole separate is included, the data for the rhyodacite yield a zero age isochron indicating crystallization within few thousand years of eruption. In addition, preliminary $^{226}$Ra-230Th data for plagioclase in the 1932 rhyodacite and the 1846 dacite show $^{226}$Ra excess. Taking differential partitioning of Ra and Ba into account and assuming that plagioclase crystallized from the entraining glass separates yield crystal ages of less than ~3000 years and less than ~5600 years, respectively, consistent with the 230Th-$^{238}$U isochron age for the rhyodacite sample. Additional analyses are in progress, including plagioclase and amphibole separates from the 1932 dacite. Based on the interpretation from previous studies that the main magmatic system remained almost compositionally homogenous over a timescale of 80 years and that the crystallization and removal of major phases was minor, it is likely that plagioclase phenocrysts stayed dispersed in the magma between the two major eruptions. We therefore predict that crystal ages in the two dacites will be similar and young. This would imply that homogenous intermediate magma bodies on the order of ~10 km3 may evolve and establish within a few ten thousand years and may not need long timescales as might be inferred from older magmatic systems.
V13B-0530
Xenocrystic Material in Arc Volcanics: Example from Recent Merapi Lavas
Recent basaltic-andesite from Merapi volcano contains abundant complexly zoned plagioclase crystals, which we have analysed for petrographic textures, major element composition and Sr isotope composition. Electron microprobe traverses of these phenocrysts have revealed individual crystals with variations in An content between 40 and 95 mol% across resorption surfaces. In addition, there appears to be a negative correlation between An content and other indicators of magmatic evolution, such as MgO wt% and FeO wt%. In-situ Sr isotope analyses of these samples show Sr ratios of discrete zones in plagioclase phenocrysts that range from 0.70568 to 0.70627 (± 1). This range notably exceeds that seen in the recent host basaltic-andesite whole rocks (< 0.705737). Often, zones with the highest An content also have the highest radiogenic Sr values, and low MgO, which is indicative of a Ca-rich, MgO-poor, high radiogenic Sr contaminant. Recent Merapi deposits contain numerous xenoliths of metamorphosed limestone with feldspar of up to An95 and whole rock Sr ratios of 0.70584 to 0.70786 (± 2). The presence of these inclusions coupled with the Sr ratios observed in crystal zones of individual lava phenocrysts, indicates that magma crust interaction may be far more significant than previously thought, as numerous crystal cores in the lava appear to be inherited from a foreign source. This poses consequences for eruptive behaviour, mass-balance calculations, and hazard assessment at Merapi and similar island arc volcanic systems elsewhere.
V13B-0531
Long-term Magma Recharge Recorded in 87Sr/86Sr Zonation in Plagioclase Phenocrysts, El Chichon, Mexico
During the last 8,000 years, El Chichon volcano has erupted relatively similar trachyandesitic magma 11 times. This compositional trend likely indicates that long-term, steady state processes operate in the El Chichon magmatic system. Previous studies found that plagioclase crystals in magma erupted in 1982 A.D. are zoned in anorthite content (An) and $^{87}$Sr/$^{86}$Sr ratio, recording magma recharge, differentiation, and changes in volatile concentration; we extend those studies to examine plagioclase crystals from six eruptions dating back to 3100BP. $^{87}$Sr/$^{86}$Sr ratios and major-element compositions were measured with laser ablation ICP-MS and electron microprobe, respectively, along core-to-rim transects of crystals. $^{87}$Sr/$^{86}$Sr ratios vary from 0.7040 to >0.7060 ±0.0002 and An contents range from ~An$_{35}$ to ~An$_{70}$. In some crystals, An content and $^{87}$Sr/$^{86}$Sr ratio correlate negatively. Not all An zonations are, however, coincident with $^{87}$Sr/$^{86}$Sr zonations; this is particularly true for An zones smaller than 30 um. Interestingly, crystals from the same eruption (even the same pumice) do not record the same zonation patterns. Indeed, some crystals from the 1250 BP pumice show an overall trend of increasing $^{87}$Sr/$^{86}$Sr and decreasing An, whereas other crystals show $^{87}$Sr/$^{86}$Sr ratios and An content that, although variable, maintain approximately constant baseline values. Our results contrast with Sr isotopic analyses of 1982 El Chichon plagioclases that showed systematic, core-to-rim decreases in $^{87}$Sr/$^{86}$Sr independent of An zonation. Multiple magmatic processes are required to produce the different plagioclase zonation patterns. Negatively correlated Sr and An zones probably reflect replenishment by mafic magmas with low $^{87}$Sr/$^{86}$Sr ratios, followed by magma differentiation and possible assimilation of high-$^{87}$Sr/$^{86}$Sr material. Zonations in An content unaccompanied by $^{87}$Sr/$^{86}$Sr variations result from either changes in temperature or volatile content of the magma. Different zonation patterns in phenocrysts from the same eruption may indicate incomplete mixing during magma recharge. Plagioclase phenocrysts thus record multiple episodes of magma recharge, differentiation, and possibly changes in temperature or volatile concentration that have maintained a relatively uniform long-lived magmatic system at El Chichon volcano.
V13B-0532
El Chichon Phenocrysts: A Study in the Variability of Excess Argon Within Plagioclase and Hornblende
El Chichon volcano has produced 11 eruptions in the last 8,000 years, all of which display a relatively constant trachyandesitic composition. The constancy of the eruptive products attests to the tapping of a long-lived magma chamber and suggests a system held in steady state between the influx of basaltic magma and differentiation of resident magma. What is poorly understood about such systems, though, is the interval between such injections and eruptions. We have sampled pyroclastic and lava flows from six of these eruptions, subjecting their dominant phenocryst phases (plagioclase and hornblende) to argon isotopic analyses in order to investigate the dynamics of a steady state magmatic system. Plagioclase from the 1982, 550 ybp, 1500 ybp and 3100 ybp eruptions display variable enrichment of 40Ar ("excess" argon), whereas hornblende from the same eruptions shows little or no enrichment. Modeling of argon diffusion in these mineral phases indicates variable but brief intervals between injection and eruption (on the order of years) and rules out xenocrystic contamination as the source of excess argon. Mafic enclaves were found within these eruptive units, and the source of this argon is most likely the same as the source of this basaltic magma injected prior to eruption. In contrast, both plagioclase and hornblende phenocrysts from the 1250 ybp and 900 ybp eruptions have argon isotopic ratios near atmospheric values. This lack of excess argon is indicative of a longer time interval between injection of new material and eruption, which allowed the phenocrysts to re-equilibrate with atmospheric argon.
V13B-0533
Petrological and Experimental Constraints on the Recent Magma Plumbing System at Okmok Volcano, Alaska, USA
The most recent eruption of Okmok volcano, Alaska, USA, occurred in February-May 1997, after ca. 9 yr. period of quiescence. The eruption was mainly Strombolian, producing multiple ash plumes, with heights up to 9000 meters ASL, and a 6-km-long lava flow. Here, we use petrological and experimental data to constrain the recent magma plumbing system at Okmok. The whole-rock composition of the 1997 Okmok basalt is homogeneous, and averages 52.27±0.18 wt.% SiO2. The mineral assemblage includes plagioclase (14 vol.%), olivine (3 vol.%), clinopyroxene (<0.3 vol.%), and Ti-magnetite (<0.3 vol.%). Plagioclase phenocrysts are strongly bi-modal: their cores are calcic (An$_{85-95}$), whereas the outermost rims (and microlites) are more sodic (An$_{66-68}$). Similarly, cores of olivine phenocrysts are Mg-rich (Fo$_{83-85}$), whereas their rims and (microlites) are more Fe-rich (Fo$_{73-76}$). Glass inclusions in phenocrysts contain ca. 2.5 wt.% volatiles. Compositions of the matrix glass and the outermost rims of phenocrysts are most closely reproduced by phase-equilibria experiments at 1015-$1030° C and 62-76 MPa water pressure. Compositions of cores of plagioclase and olivine phenocrysts have not been reproduced in our experiments up to 100 MPa, which suggests that they possibly originated from a deeper source. We tentatively suggest that the basaltic magma erupted in 1997 was first stored at deep depths, and then migrated and re-equilibrated at depths of about 3.3-4.1 km beneath the floor of Okmok caldera. Detailed InSAR studies indicate that the 1997 Okmok eruption was preceded by at least 5 years of a nearly continuous uplift of the caldera floor (Lu et al., 2005 and references therein). The deformation was interpreted as the result of gradual pressurization of a shallow magma reservoir by repetitive magma inputs at a depth of 3.6 km, which agrees well with our petrologic estimates. Our ongoing petrological experiments will be used to further constrain the pre-eruptive conditions for the 1997 Okmok basalt and to estimate probable magma ascent rates during the eruption.
V13B-0534
Soufrière Hills Plagioclase: Postcards From the Edge.
Secondary Ion Mass Spectrometry (SIMS) can provide sub-micron depth resolution for analyzing products of volcanic eruptions. SIMS was used to examine the outer rims of plagioclase phenocrysts derived from both explosive and effusive eruptions of the Soufrière Hills Volcano (SHV), Montserrat. Phenocrysts were separated from the host igneous rock by crushing with a mortar and pestle and then cleaned with a Branson Sonifier. A 12.5 kV O2$^{+}$ primary ion beam was used to examine the variation in ten elements (Ca, Na, Si, Al, Ti, Zr, K, Fe, Sr, Li) through a crystal depth of 5-9 microns. Plagioclase crystals separated from explosively produced pumice clasts show increasing anorthite (An) content with depth into the crystal surface, starting at ~10% An at the surface and reaching a constant composition of ~45% An at 2-4 microns depth. According to experimentally determined estimates of plagioclase growth rates for the SHV magma (Couch et al. 2003; J. Petrology 44, 1477-1502), the 2-4 microns depth over which An changes corresponds to 1-7 hours of growth. Sr also shows a general increase with depth into the crystal. K shows a rapid decrease in abundance with depth. Fe shows more complex patterns that may indicate late-stage crystallization of magnetite. Plagioclase derived from exogenous dome samples also have surface compositions of ~10% An increasing with depth to ~30% An, but rather than plateau, the values begin to decrease again at 2-5 microns depth. This fluctuating abundance of An may reveal the presence of micron-scale decompression-induced growth zones that have not been previously documented due to limitations in the spatial resolution of conventional analytical techniques. Explosive and effusive samples exhibit conflicting Li trends. The explosively derived plagioclase have elevated surface Li concentrations while the dome derived plagioclase have low surface Li concentrations. These differing trends may provide evidence of closed system vs. open system degassing as a function of eruptive style. Geochemical analyses of igneous phenocrysts using the SIMS depth-profiling technique can be used to constrain the style of magma decompression and eruption. Additional analyses are currently being performed on an expanded suite of samples in order to confirm these results and to relate crystal-edge chemistry to other parameters such as quench pressure and degree of magma degassing.
V13B-0535
Geochemical, Textural and Petrographic Indicators for Eruptive Behavior at Merapi Volcano, Indonesia
Over the 40,000 years if its activity, Merapi volcano has exhibited a wide variety of eruptive style. Recent eruptive activity is dominated by growth and subsequent gravitational collapse of small lava domes; however, the tephrostratigraphic record reveals a history that includes mild effusive activity and is punctuated by highly explosive events (to VEI 4). The long-term geochemical evolution at Merapi is characterized by a gradual increase in silica content from basalt to andesite. The most interesting variation at Merapi is observed in K2O content, with eruptive products belonging to both medium- and high-K series. An abrupt change from medium- to high-K lavas occurred with the somma-forming eruption of the Tegalsruni tephra ~100 A.D. (Andreastuti 1999). Our work indicates that similar compositional shifts occurred in the earlier history of the Merapi system, although these events are not well-constrained stratigraphically. We have sought to interpret these compositional shifts in terms of changes to the magmatic plumbing system of Merapi as revealed by petrographic and textural analysis. Plagioclase feldspar phenocrysts and microphenocrysts within Merapi lavas display two contrasting styles of crystal size distributions (CSDs) and compositional zoning patterns, which in turn indicate distinct plumbing systems and transport. Briefly, basaltic lavas from throughout Merapi's history show kinked CSD patterns that suggest magma mixing in at least two crustal chambers prior to emplacement. In contrast, recent basaltic andesite dome lavas have smoothly curved CSD patterns that suggest prolonged residence at near-surface conditions under steady-state open system behavior. The high crystallinity of recent Merapi basaltic andesites (~60 vol.%) corresponds to a viscosity approaching the critical rheological locking limit (Marsh, 1981). Current Merapi eruptive behavior oscillates about this rheological limit, with temporary crystalline plugs in the eruptive conduit disrupted by repeated efflux of magma. A slight change in the reservoir-conduit dynamics could interrupt this balance and modify the current eruptive style. We propose that this transition from steady-state to highly explosive activity has happened during Merapi's history, and that such events disrupt the plumbing system to the extent that magmas from different K-affinity are tapped.
V13B-0536
Plagioclase Crystal Populations in Lavas from Calbuco Volcano, Chile
Calbuco Volcano in Southern Chile is regionally unique in that it is predominantly andesitic, whereas surrounding volcanic centers are basaltic in composition. Lava from Calbuco is also hornblende-bearing, phenocryst-rich, and it contains crystal clots that are interpreted as cumulate minerals entrained in the magma. The phenocryst assemblage of the lava, and the mineral assemblage of the crystal clots, is: plagioclase, hornblende, orthopyroxene, clinopyroxene, olivine and oxide minerals. Three different species of plagioclase crystals are found in Calbuco andesite: zoned phenocrysts in the lava, unzoned grains in the crystal clots, and high-An megacrysts. The phenocrysts are predominantly anhedral and normally zoned with An mol% in the core ranging from An$_{68-91}$, and at the rim from An$_{58-82}$. Most of these crystals range in size from 50 to 700 microns, with rare grains larger, and nearly all contain melt inclusions. Plagioclase grains within the crystal clots are not zoned and have An mol% of An$_{70-93}$, which overlaps the An mol% of the zoned phenocryst cores. These grains range in size from 100 to 500 microns. Megacrysts are 2000 microns and larger in size, and are unzoned, except for the rim of the grains. They have very high An mol% of An$_{88-92}$ in the core. The 10-60 micron rims are normally zoned. Based on the overlapping An mol% contents of zoned plagioclase cores and the plagioclase from crystal clots, it is probable these two populations of crystals have formed as phenocrysts, crystallizing directly from the basaltic andesite to andesite magma. The origin of the megacrysts is more puzzling. Periodic replenishment by a more mafic magma and magma mixing could introduce these large An-rich plagioclase crystals, or they could be cumulate crystals that formed from a more mafic magma and were later entrained in andesitic magma. In either case, the narrow rim of normal zoning suggests a relatively short time span between entrainment and eruption.
V13B-0537
Distinguishing Phenocrysts From Xenocrysts; Dating the Onset of Volcanic Activity on the Isle of Rum, Scotland.
Major volcanic activity on the Isle of Rum started with eruption of rhyodacite, now preserved as intrusive and extrusive parts of the intra-caldera succession. A thick sequence of intra-caldera ignimbrites and sedimentary breccias are preserved in the north and south of the island. Twenty feldspar crystals of the rhyodacite were dated using Ar/Ar and yielded a mean apparent age of 60.83 ± 0.27Ma (MSWD = 3.4), consistent with previously published dates for the crosscutting (i.e. younger) ultrabasic-layered intrusion of 60.53 ± 0.08Ma, Hamilton et al., (1998, Nature). On an age versus probability plot the feldspars do not, however, show a simple Gaussian distribution, but a major peak at 60.33Ma and two smaller shoulders at approx 61.4Ma and 63Ma. Our preliminary interpretation of the older ages is that they include a) xenocrysts derived from earlier Tertiary plutonics (61.4Ma peak) and b) older feldspars that have largely re-equilibrated within the rhyodacite magma chamber, possibly derived from Lewisian gneiss (63Ma peak). This may imply a residence time of these xenocrysts in the magma chamber of up to several years; cf. Gansecki et al., (1996, Earth Planet Sci. Lett.). The youngest and strongest age peak at 60.33Ma is suggested to represent the rhyodacite event. The oxygen isotope composition of the rhyodacite feldspars (6.88 ‰) is in the range of magmatic phenocrysts (6-7.5 ‰) and Lewisian gneisses (5-8 ‰) and well above the very low oxygen isotope values usually associated with high-T alteration. Hydrothermal overprint due to the layered ultrabasic intrusion was therefore probably minimal. We suggest the 60.33 ± 0.21Ma crystal age represents the rhyodacite eruption/intrusion event, implying that the ultrabasic-layered suite was already forming at depth and emplaced at shallow structural levels quickly thereafter. These new age dates tie in very well with recent work by Chambers et al., (2005, Lithos), highlighting a very quick succession of events: doming, violent ignimbrite eruption, intrusion of granite and ultrabasic rocks, followed by the rapid erosion of the Rum super-volcano, all in probably less than 1 million years.
V13B-0538
Water Concentrations and Distribution in Evolving Melts as Suggested by Melt Inclusions and Matrix Glasses
Water concentrations in glasses and feldspar crystals from volcanic centers associated with the Torfajokull central volcano of southern Iceland were determined using Fourier transform infrared spectroscopy (FTIR). FTIR spectra were collected using synchrotron-generated radiation at the National Synchrotron Light Source at Brookhaven National Laboratory. The bright synchrotron-generated IR beam permits the analysis of spots as small as 10 mm and collection of maps of water concentration across areas of interest. The goal of this study was to explore the distribution of water between glasses and coexisting feldspar crystals in a range of mingling and non-mingling magmas in a large bimodal volcanic complex. With distance from phenocryst borders, water concentrations of matrix glass generally decrease, but some also subsequently increase, from initial relative highs closest to crystal edges. Water concentrations of matrix glasses vary from a high of ~ 1 wt % in clean glass to a low of ~ 0.5 wt % in more microcrystalline glass, and concentrations grade between adjacent glasses. Differences in water concentration between the glasses may be a result of crystallinity, glass composition, or both. Average water concentrations in melt inclusions vary slightly between different feldspar crystals and throughout single feldspar crystals. Averages of water concentration in melt inclusions in single phenocrysts range from 0.24 to 0.77 wt %, vary within phenocrysts by as much as ~ 0.7 wt %, and are generally higher in inclusions in hornblende. Variations in water concentration amongst melt inclusions and matrix glasses may reflect water behavior during melt evolution as crystallization, and mingling if present, proceed. The determination of water concentration in conjunction with composition in these glasses and crystallizing phases may permit a quantitative understanding of water distribution throughout the crystallization history of complex melts.
V13B-0539
Time scales of assimilation and mixing at Krafla volcano, Icleand
Assimilation of crustal material is an important process affecting the composition of magmas erupted from volcanoes in a wide variety of tectonic settings, yet the timescales over which assimilation occurs are difficult to measure directly. One particularly striking example of crustal assimilation is Krafla volcano in Iceland, where a systematic decrease in δ18O with decreasing MgO in erupted lavas has been previously interpreted as evidence for assimilation of hydrothermally-altered crustal rocks. We present $^{238}$U-230Th-$^{226}$Ra and oxygen-isotope data for groundmass and plagioclase separates of two lavas representing different phases (Jan-Feb 1980 and Nov 1980) of the most recent eruption of Krafla in order to directly link crystal ages with variations in oxygen-isotope composition and by extension mass balance of assimilation. Multiple size fractions of plagioclase separates were analyzed for both U-series disequilibria and oxygen isotopes. Oxygen isotope measurements of 2-5 aliquots of each size fraction of plagioclase vary significantly outside of analytical uncertainty (standard deviation of replicate measurements is 0.14-0.27 permil compared to typical s.d. of 0.03-0.05 permil for homogeneous samples) and also differ between samples, with a total range in δ18O of over 1 permil. These variations are systematic with grain size and eruptive sequence, with δ18O increasing with decreasing grain size in plagioclase from the Jan-Feb 1980 lava, and higher but roughly constant δ18O in all size fractions of plagioclase from the Nov 1980 lava. Collectively these data suggest mixing of plagioclase (and presumably host liquids) with differing influence of assimilated crustal material. $^{238}$U-230Th disequilibria of plagioclase and groundmass yield zero-age isochrons for both samples, indicating that crystallization dominantly occurred within the past 10 kyr. This is consistent with $^{226}$Ra-excesses measured in all plagioclase separates [($^{226}$Ra)/(230Th) = 1.08-1.98]. A difference in age between large and small plagioclase fractions of the Jan-Feb 1980 samples may be suggested by the smaller $^{226}$Ra-excess in the large fraction. In-situ trace-element measurements in plagioclase will allow correction for impurities in the mineral separates, and when combined with U-series measurements currently in progress for corresponding whole-rock and groundmass separates, will quantify differences in average age between the different plagioclase separates.
V13B-0540
The Magmatic Plumbing System of Quaternary Basalts in the Main Ethiopian Rift
Quaternary basalts from the Silti-Debre Zeyit Fault Zone, in the central Main Ethiopian Rift, provide an opportunity to investigate the magmatic plumbing system of continental rift basalts. We investigated olivine, clinopyroxene and feldspar phenocrysts and xenocrysts hosted in MgO-rich basalts to reveal information concerning the history of the host basalt en-route to the surface. These histories may be used to constrain some of the complex lithospheric processes that occur in continental rifting. Two end-members were identified with distinctly different magmatic ascent histories: one group that ascended rapidly to the surface and another that rose slowly or paused in the mid- to upper crust. The rapidly-ascending end-member is typified by the presence of skeletal olivine, dynamically recrystallised grains of olivine, kink-banding in clinopyroxene, and clinopyroxene crystallization pressures of 8-9 kbar. Electron microprobe analysis revealed reverse zoning of olivine xenocrysts in these lavas. The depths of clinopyroxene equilibration in rapidly-ascending basalts correspond to the base of the crust and indicate magma ponding at the crust-mantle boundary. In the shallowly-equilibrated basalts, the common stress features observed in the previous group were absent, and clinopyroxene equilibrium pressures were generally between 2 and 4 kbar. In these shallowly-equilibrated basalts, feldspar phenocrysts were pervasive and often nucleated on pre existing clinopyroxene and olivine xenocrysts. Olivine xenocrysts exhibit chemical re-equilibriation with the host magma. Evidence of rapidly ascending magmas in this area is consistent with surface morphology as explosive maars are observed along the Silti-Debre Zeyit Fault Zone. The presence of two basalt groups with distinctly different magmatic ascent histories and separated by a distance of less than 5 km argues strongly against the existence of large mid-crustal magma chambers in this region. Instead we propose a system of dykes and small magma chambers that traverse the lithosphere (e.g. Rooney et al., 2005). This petrographic model can be used in geochemical analysis of the regional basalts by constraining likely crustal contamination of those samples that have ascended slowly or ponded within the crust.
V13B-0541
Textural and micro-Sr isotopic investigations of feldspar phenocrysts from the Teide-Pico Viejo stratovolcanic complex, Tenerife (Canary Islands)
As part of the ongoing EU-funded ERUPT (European Research on Understanding Processes and Timescales in magma systems) project, we have engaged in a detailed petrologic and geochemical study of samples from the Teide-Pico Viejo stratovolcanic complex, Tenerife (Canary Islands). Located within Las Cañadas caldera, the Teide-Pico Viejo complex represents a large component of the most recent (0.18 Ma - present) volcanic activity on Tenerife. Therefore, through in-depth investigation of eruptive products from the Teide-Pico Viejo complex we hope to gain a better understanding of the current state of the Tenerife magmatic system. The study includes: a) whole-rock isotopic (Sr-Nd-Hf-Pb) analyses, b) microsampling of individual feldspar phenocrysts for $^{87}$Sr/$^{86}$Sr via TIMS analysis, and c) quantitative textural (CSD) analysis of feldspar phenocrysts. Previous work and our own initial examination of individual feldspar phenocrysts via electron microprobe analysis have revealed the presence of considerable core-to-rim variations in major and trace element contents. Likewise, preliminary results of $^{87}$Sr/$^{86}$Sr microanalysis of the same feldspar phenocrysts indicate relatively small (but still detectable) accompanying variations in $^{87}$Sr/$^{86}$Sr from ~ 0.7031-0.7032. The most significant shifts in $^{87}$Sr/$^{86}$Sr within individual feldspar phenocrysts often correspond with obvious textural boundaries (e.g., resorption surfaces). In addition, petrographic observations of some Teide-Pico Viejo lavas suggest a sequential process by which large complexly zoned feldspar megacrysts are formed. This process appears to involve initial amalgamation of individual elongate feldspar phenocrysts into large clusters followed by subsequent crystallization of a rim that entirely surrounds each cluster. These early results suggest a complex crystallization history for at least some feldspar phenocrysts indicative of an open and dynamic magmatic system beneath the Teide-Pico Viejo complex.
V13B-0542
Compositional and Textural Evidence for Crystal Ancestry in the Voluminous Silicic Magmas of the Bruneau-Jarbidge Eruptive Center, Yellowstone Hotspot
The Bruneau-Jarbidge eruptive center (BJEC) of southern Idaho, USA, offers an opportunity to address the problem of crystal ancestries in large silicic volcanic systems and their origins as phenocrysts, xenocrysts, or antecrysts. This large intracontinental system was powered by the Yellowstone hotspot from 12.7 to approx. 8 Ma and lies in contradistinction to both the present day Yellowstone system and voluminous arc systems in that its estimated discharge rate of rhyolitic magma is exceptionally high (2300 km3/Ma), and the mineral assemblages reflect equilibrium conditions and near liquidus temperatures of 900-1000° C. Detailed observations of mineral compositions and textures in tuffs and lavas from this system bear on current ideas concerning the relative roles of rejuvenation and multiple source inputs in magma generation and their relevance to the problem of magma residence times. Silicic magmas of the BJEC are represented by a suite of more than twenty large-volume crystal poor rhyolitic tuffs (100-1000 km3 ea) and lavas (up to 200 km3 ea). The tuffs (known collectively as the Cougar Point Tuff) and lavas share a sparse anhydrous mineral assemblage and show striking similarities to one another with respect to mineral textures and compositions. These similarities demand an exceptional but systematic petrogenesis that is poorly understood. Many tuffs share identical doublets or triplets of glass and pyroxene compositional modes, and some units are nearly indistinguishable from one another despite significant intervals between eruptions (up to 1 Ma). Multiple discrete modes are present at the hand specimen level consistent with the syn-eruption of multiple discrete magma volumes. However, lavas succeeding the tuffs are compositionally unimodal, with a collective compositional spectrum of discrete modes that overlaps that of the tuffs. Mineral textures observed in thin section support the coexistence of phenocrysts and inherited material that may derive from source rocks and/or rejuvenated mush from earlier magma reservoirs. However, mineral compositions confound the effort to detect different crystal histories because mineral phases in contrasting textural contexts have homogeneous, identical compositions. Also, multiple modes are not found in crystal clusters, xenoliths, or glomerocrysts, which might be expected if magmas had mingled or mixed prior to eruption. Such compositional homogeneity despite textural variety, especially in the lavas, suggests a common source (represented by xenoliths and xenocrysts) rather than multiple sources, and perhaps that the source and the magmas with their crystal cargo were thermally equilibrated and spatially juxtaposed prior to eruptions.
V13B-0543
A Textural and Microanalytical Evaluation of Magma Mixing in Detroit Seamount Lavas.
Basalts from DSM are incompatible trace element and isotopically depleted relative to other tholeiitic basalts from the Emperor Seamount Chain (ESC) and the Hawaiian Islands. Sources of this depletion have been debated, and may include variable partial melting of depleted components in the Hawaiian plume source and/or magma mixing that involved an exchange of liquid and crystals between plume and MORB magmas. To evaluate what roles, if any, magma mixing played in the genesis of DSM basalts we conducted a quantitative textural and microanalytical study focused on plagioclase phenocrysts. We measured plagioclase crystal size distributions (CSDs) in three pillow basalts from DSM, ODP Site 1203 Units 3, 14, and 31 at the top middle and bottom of the cored section, respectively. All three basalts have concave up plagioclase CSDs, although upward deflections of the Unit 14 and 31 CSDs are more pronounced. Based upon CSD slopes we suggest each of the three basalts contains mixed populations of plagioclase crystals indicative of magma mixing (upward CSD deflection may be due to textural coarsening, overgrowth, or magma mixing). From the textural data a minimum of two crystal populations are present in each basalt: Population A = smaller crystals; Population B = larger crystals. To test the magma mixing hypothesis, we quantified major and trace elements in crystals from each population. Unit 3 crystals from population A are An$_{62-88}$ and population B are An$_{71-81}$. Population A crystals from Units 14 and 31 are An$_{66-75}$ and population B are An$_{80-87}$. Unlike the distinct An contents of population A and B crystals of Units 14 and 31, the overlapping An content of Unit 3 crystals is inconsistent with a mixed crystal population. Parent magma compositions for populations A and B were calculated using partition coefficients that account for An content and crystallization temperature. Parent magmas of population A and B from Unit 3 have overlapping Ti, Sr, Y, Ba, and LREE abundances. Parent magmas of population A and B crystals from Units 14 and 31 are distinct from one another. In both Units 14 and 31 population A parent magmas have lower Sr and Y. Unit 14 population B parent magmas tend to be less LREE enriched, whereas Unit 31 population B parent magmas are more LREE enriched than those of respective population A crystals. Parent magmas of all Unit 3 crystals are more LREE depleted than those of Unit 14 and 31 crystals. This trace element evidence is used to support our initial interpretation that only Units 14 and 31 basalts contain both textural and compositional evidence of magma mixing. While the CSD of the Unit 3 basalt is slightly concave up, major and trace element signatures recorded in plagioclase indicate that the upward CSD deflection is attributed to greater error associated with shape measurements of larger crystals. Magma mixing appears to have played a role in the petrogenesis of Detroit Seamount lavas.
V13B-0544
Pseudobrookite-Magnetite-Ilmenite Assemblages in Highly Oxidized Hornblende Andesite of the Coleman Pinnacle Flow, Mount Baker, Cascade Range, WA
Pseudobrookite microphenocrysts occur in cognate xenoliths in a 305 ka hornblende andesite flow from Coleman Pinnacle, Mount Baker, WA. Pseudobrookite (Psb)-ferropseudobrookite (Fpb)-karrooite (Kar) solid solutions are associated with hornblende or clusters of orthopyroxene, clinopyroxene, plagioclase, ilmenite (Ilm)-hematite (Hem)-geikielite (Gk) solid solutions, ulvospinel (Usp)-magnetite (Mt) solid solutions, apatite and zircon in a matrix of rhyolitic glass. Grains of Psb$_{46-49}$Fpb$_{25-28}$Kar$_{24-28}$ occur with Ilm$_{47}$Hem$_{42}$Gk10 rims or intergrown with Ilm$_{45-47}$Hem$_{40-43}$Gk12. These textures are analagous to those observed between armalcolite and ilmenite in lunar basalts. In a unique occurrence, Psb$_{48}$Fpb$_{20}$Kar$_{32}$ and Usp$_{22}$Mt$_{78}$, an assemblage previously unrecorded in natural systems, form an apparently stable symplectitic intergrowth surrounding a core of Ilm$_{43}$Hem$_{42}$Gk14. Mass balance calculations show that the Psb$_{ss}$ + Usp$_{ss}$ assemblage is not an isochemical decomposition of ilmenite and the following explanation for the unique intergrowth is proposed. In the Fe-Ti-O system (Mg-free), the lower stability limit of Psb$_{ss}$ lies at a temperature higher than that of the Ilm-Hem solvus, most likely precluding the stable existence of a Psb$_{ss}$ + Usp$_{ss}$ assemblage. However, all three phases (Ilm$_{ss}$, Psb$_{ss}$, Usp$_{ss}$) in the symplectitic assemblage contain significant amounts of Mg, e.g., the Psb$_{ss}$ contains 32 mol% MgTi2O$_{5}$ (karrooite) component. The effect of addition of Mg to the Fe-Ti-O system is to increase the temperature of the solvus and tricritical point along the Ilm-Hem-Gk join. The symplectitic Psb$_{ss}$ has a lower stability limit of ~710°C based on linear interpolation between the lower stability limits of Fpb (1140°C), Psb (585°C) and Kar (646°C), allowing the Psb$_{ss}$+Usp$_{ss}$ tie line to lie inside the solvus and thereby stabilizing the Mg-bearing Psb$_{ss}$ + Usp$_{ss}$ assemblage relative to an exsolved Mg-bearing Ilm$_{ss}$ + Hem$_{ss}$ pair. Experiments are being conducted to test this model. The compositions of Ilm$_{ss}$ + Usp$_{ss}$ pairs coexisting with Psb$_{ss}$ are used to calculate the temperatures and oxygen fugacities of formation. The most recent refinement of the Fe-Ti oxide geothermometer-oxygen barometer (Mark Ghiorso, personal communication) applied to the Psb$_{ss}$-bearing assemblages yield T = 990-1130°C and fO2 = NNO +1.5 to +1.75. Coexisting orthopyroxene and clinopyroxene in the Psb$_{ss}$-bearing cognate xenoliths record a temperature of formation of ~960°C. The calculated fO2 range is consistent with the composition of the Ilm$_{ss}$ in equilibrium with Psb$_{ss}$ ± rutile and is also consistent with the presence of Fe$^{3+}$-rich oxyhornblende phenocrysts (Fe$^{3+}$/σFe = 0.8) in the andesite.
V13B-0545
Crystals in Kimberlites: Where are the Phenocrysts?
Kimberlites are among the igneous rocks that we know less about their petrogenesis. This is in part due to their high crystal contents (mainly olivine) and varied provenance of their crystal populations, so that true liquid compositions are difficult to identify. Distinguishing between the different sources of crystals (e.g., phenocrysts vs. xenocrysts) is very important for understanding the formation kimberlites and their associated diamonds. Here we report the results of a petrological and diffusion modeling study of olivine and monticellite from Lac de Gras kimberlites (Canada) which bear implications for the mode and timescales of kimberlite petrogenesis. The rocks contain olivine (Ol) of varied morphology and sizes, from angular through euhedral to rounded, and from about 0.5 cm to 50 μm. In contrast, monticellite (Mo) is euhedral and about 100 μm in diameter. Many large Ol display deformation features such as undulose extinction or kink-bands. Such observations are common to many kimberlites. We distinguish3 types of Ol: (1) unzoned at about Fo$_{93}$, (2) crystals with cores at Fo$_{86-89}$ and rims at Fo$_{91}$, and (3) crystals with cores at ~ Fo$_{93}$ and rims at Fo$_{91}$. In addition, any type of crystal may also have a thin (<50 μm) rim of very Mg-rich Ol (Fo$_{98}$). Despite the different core compositions, all Ol have rims at an intermediate composition (Fo$_{91}$). This can be the result of mixing between two components and might reflect mantle metasomatic processes related to kimberlite formation. Mo has rather homogenous composition at Fo$_{95-96}$. Low pressure (< 1 kbar) thermodynamics of Ca-Fe-Mg olivines indicate that this Mo is only in equilibrium with the very Mg-rich Ol (Fo$_{98}$) found as small rims. Thus, Mo crystallized late, probably when the kimberlite magma was en route to the surface. Diffusion modeling of the Fe-Mg gradients between Ol cores and rims indicates times of a few months, whereas those at the very Fo-rich rims indicates times of a few days. This last estimate overlaps with that of other authors for transfer times of kimberlite magmas from the mantle to the surface. In contrast, our new estimate of several months between magma mixing and eruption have not been yet recognized. Finally, it seems very difficult to attach any genetic information to the word phenocryst in these rocks. The compositional and textural features of the majority of Ol are either the result of open-system processes or occurred very late (Mo and very Fo-rich rims), when the magma was almost at the surface.
V13B-0546
Mineral-Scale Sr Isotopic Study of Plagioclase in the Mafic Dikes of the North American Wall and the Diorite of the Rockslides, Yosemite Valley, California
The North American Wall mafic dikes and the diorite of the Rockslides mafic complex in the intrusive suite of Yosemite Valley show evidence of mixing with their host granites as well as with earlier components. Rocks sampled are mostly gabbro to gabbroic diorite in composition. Whole rock major element variation diagrams indicate the mafic magma mixed with a more silicic component, but extrapolation to the silica end member does not yield the same result for each element. Mafic rocks show a wide variation in trace element concentrations, especially in Cr and Ni, with two samples showing enrichment in Cr (>300 ppm) and Ni (~44 ppm) compared to other samples (Cr =13-94 ppm; Ni = 5-26 ppm). These two samples also have the most primitive whole rock isotopic values determined to date for Yosemite Valley rocks, with ε_{Nd}$ = -3.3 to -3.5 compared to -5 to -7 for other mafic rocks of the suite. Plagioclase phenocrysts within each unit display bimodal compositional populations. Subhedral to euhedral partially resorbed calcic cores (mode = An$_{84-88}$) are reminiscent of a mafic magma, while sodic rims (mode = An$_{48-50}$) are the product of a more silicic component. Very little to no intermediate zoning is present between cores and rims. Mineral-scale LA-ICP-MS $^{87}$Sr/$^{86}$Sr analysis of plagioclase cores and rims are consistent with previously published enriched bulk-rock ratios for the suite (0.7065-0.7078), but are unable to distinguish between mixing components. The data show no direct evidence for a depleted mantle melt component contaminated by crustal assimilation. The mafic rocks are comparable to high-alumina basalts, whose generation involves crystal fractionation and magma mixing/crustal assimilation. The evolution of these high-alumina basalts may have provided an opportunity for assimilation to take place before plagioclase crystallization, thus explaining why plagioclase core-rim analysis could not distinguish between mixing components. Therefore, it is possible but not necessary to derive the rocks from an enriched mantle source.
V13B-0547
The origin of high-Mg olivines and Cr-rich spinel inclusions from Mediterranean Tertiary lamproites: the case study from Vera, Spain
Si-rich Mediterranean type lamproites (ca. 55-56 wt. % SiO2) are olivine-phyric, having extremely Fo-rich millimetre-sized olivine macrocrysts that host high-Cr spinels. This is one of the most typical features of the whole lamproite group. The origin of olivine in Si-oversaturated lamproitic melts is an unresolved question related with lamproite genesis; previous studies have interpreted such high-Fo olivines as mantle xenocrysts. However, in most Mediterranean lamproites, the olivine is idiomorphic and tabular (magmatic) with consistent chemical compositions indicating its phenocrystal character, showing that the interpretation as xenocrysts is inappropriate. Si-rich lamproites from the Vera lava flow in Southeast Spain contain two types of olivine macrocrysts: the first are phenocrysts, and the second are mantle xenocrysts. In this contribution, we present major and trace element data for both types of olivine, and show that Vera may be a key lamproite locality where the origin of olivine may be resolved. (1) Tabular idiomorphic olivine phenocrysts up to 1.0 x0.3 mm show extremely high Fo (up to 96 %) and very high NiO contents up to 1 wt. %. In terms of measurable trace elements, the most noteworthy is the enrichment of Li and B up to 65 and 10 ppm, respectively. Cr\# of tiny idiomorphic high Mg-chromite inclusions in the olivines reach up to 0.96. The olivine-spinel pairs plot in the most refractory part of the olivine-spinel mantle array diagram. (2) In contrast, xenomorphic mantle olivine xenocrysts show Fo components not exceeding 92 % and NiO contents less than 0.5 wt. %. Trace element analyses show considerably lower amounts of Li and B less than 10 and 5 ppm, respectively. They bear large spinel inclusions with considerably less refractory compositions with Cr\# (86-88). We consider two different models which may explain the origin of the phenocrystal olivine: cognate phenocrysts crystallized from the host magma, and crystallization from parental melts of olivine lamproite composition. The extreme compositions of olivine-spinel pairs convey information about the extremely refractory character of the subcontinental lithosphere. It resembles harzburgitic suprasubduction-type orogenic peridotites, which have experienced twofold depletion in addition to the normal depleted upper mantle signature: first during the production of MORB, and secondly associated with intraoceanic subduction. We argue that the refractory component of the Mediterranean lamproites may be derived from an island-arc oceanic lithosphere accreted to the continent during the Alpine geotectonic and collisional processes. This best explains the extreme depletion in non-cratonic environments such as the Mediterranean area.
V13B-0548
Magmatic Processes for Producing the Variation From Ankaramites to Phenocryst-Poor Basalts: an Example From St. Helena Island
St. Helena is a volcanic island situated on 39 m.y. old lithosphere, approximately 750 km east of the Mid-Atlantic Ridge. The subaerially exposed portion of the island is dominated by a pile of basaltic lava flows erupted from two main centers (older NE volcano and younger SW volcano), associated with subordinate pyroclastics and trachytic to phonolitic intrusions. The K-Ar ages of 7-14 Ma have been reported for the volcanics comprising the NE and SW volcanoes [Baker et al., 1967; Abdel-Monem and Gast, 1967]. Rock samples collected from all stratigraphical units found above sea level form moderately alkalic suite ranging from ankaramite through trachyandesite to phonolite. These rocks show continuous compositional trends, but the change in slope is observed at ca. 5 wt% MgO. Basaltic rocks with more than 5 wt% MgO are mainly classified into ankaramite and olivine basalt. Ankaramite (8-16 wt% MgO) is characterized by abundance (50-10 vol.%) of coarse (15-3mm in longest dimension) olivine and clinopyroxene phenocrysts. Kink banded olivines, which have identical composition to kink band-free olivine phenocrysts, comprises 4-57% of the total olivine phenocryst population. In contrast, olivine basalt (5-8 wt% MgO) is the most abundant rock type in the subaerial portion of St. Helena Island. They have olivine microphenocryst and subordinate clinopyroxene and plagioclase microphenocrysts. Unlike ankaramite, mafic phenocrysts are less abundant (<ca. 10 vol.%) and small (mainly about 0.5 mm in size). Petrological features of these alkali basalts are summarized as follows: 1) Bulk MgO concentration shows a good correlation with amount of olivine and clinopyroxene phenocrysts. 2) Compositional trend from ankaramite to olivine basalt can be explained by addition of olivine and clinopyroxene into a magma and/or subtraction from it. 3) The maximum forsterite content in olivine core in ankaramites is almost constant at Fo85-87 irrespective of the variation in bulk FeO/MgO from 0.6 to 1.3. The most Mg\#-rich olivine (~ Fo87) can be precipitated from a melt with FeO/MgO of 0.9, assuming a Kd of 0.3+/-0.03 [Roeder and Emsile, 1970]. 4) Groundmass compositions of ankaramites have transitional chemical compositions between ankaramites and olivine basalts. This is further supported by compositions of olivine phenocrysts in ankaramites; the minimum forsterite contents (Fo74-80) in phenocryst core from individual ankaramite samples can be in equilibrium with a melt with FeO/MgO=1.5-2.1, which is comparable to the composition of olivine basalts. These features indicate that alkali basalts from ankaramites to olivine basalts would be basically produced by crystallization of a parental magma with FeO/MgO~0.9 and subsequent magma extraction from the partially crystallized magma chamber. The difference in the proportion of melt to crystal in the extracted magmas controlled the variations from ankaramite to mafic olivine basalts. Kink-banded olivines and fractured coarse phenocrysts commonly found in ankaramites are likely to represent the crystals disrupted from a deformed rigid crust and/or crystal-mush zone present within the magma chamber.
V13B-0549
Relationship Between SEM-Cathodoluminescence and Trace Element Chemistry of Quartz in Granitic Igneous Rocks of the Oslo Continental Rift
Numerous granitic intrusions were emplaced in the Oslo area during the main rifting episode (275-240 Ma). The Drammen biotite-granite and Eikeren-Skrim alkali-feldspar granite (ekerite) represent two of the most evolved rock types related to this period of plutonic activity. This study documents the textural and chemical features of the different generations of quartz (Qz) present in the two granites, in relation with the igneous and hydrothermal processes at the origin of their formation. Both granites suffered pervasive alteration by magmatic and hydrothermal (sub-solidus) fluids. Contrary to other minerals, primary igneous quartz is well preserved and revealed by SEM-cathodoluminescence (SEM-CL) images. Several generations of secondary quartz are also recognisable and clearly related to late magmatic and/or subsolidus fluid infiltrations. Four quartz types can be categorised by SEM-CL luminescence and texture: Type 1: Bright primary quartz frequently showing oscillatory growth zoning from core to rim, in the form of alternating bright and dark ring-liked sectors reflecting compositional variations of the crystallising melt. Type 2: Light grey up to hundreds μm wide diffuse alteration zones, which follow grain boundaries and open cracks cutting through grains of Qz type 1. Type 3: Usually darker than Qz type 2. Contrary to Qz type 2, not necessarily directly associated with macroscopic fracturing features. Qz type 3 is featured by irregular patterns cutting through Qz type 1 and 2 and seems to be the result of dissolution/recrystallisation processes, possibly through a network of micro cracks related to the major fluid channelling fractures. Type 4: Narrow cracks and patches of black Qz crosscutting all the other types. EPMA in situ analysis of the different quartz generations confirm that the intensity of luminescence of quartz is strongly positively correlated to the presence of specific trace elements in quartz like Ti and Al. Hence the different quartz types mentioned before can be distinguished by levels of luminescence and trace element chemistry. Qz type 1 contains about 250 (200-350) ppm Ti and 130 (80-160) ppm Al, whereas Qz type 2 and 3 have overlapping concentrations averaging 60 (10-120) ppm Ti and 90 (50-130) ppm Al. Qz type 4 was not analysed due to the narrowness of the structures. These results document that quartz alteration by percolating fluids has a purifying effect, leaching Ti and Al out of the quartz by diffusion (Qz type 2)and dissolution/recrystallisation (Qz type 3). SEM-CL textures and chemistry of quartz help to understand the complex magmatic and hydrothermal history of these granites, whose original magmatic picture is strongly overprinted and blurred by pervasive fluid-related alteration features.
V13B-0550
Evolution of the Campanian Ignimbrite Magmatic System I: Constraints on Compositional Zonation and Eruption Probability Imposed By Phase Equilibria
The eruption and deposition of the ~39.3 ka Campanian Ignimbrite (CI), a large volume (~200 km3 DRE) trachytic to phonolitic ignimbrite, is the dominant event in the history of the Campi Flegrei volcanic field near Naples, Italy. In an effort to comprehend its petrological evolution, we have conducted ~~110 MELTS (Ghiorso, 1997) phase equilibria simulations of the major element evolution of parental CI magma. The goals of this work are to approximate oxygen fugacity (f$_{O2}$), initial dissolved water content and pressure at which isobaric closed system fractional crystallization of parental melt most accurately captures the observed liquid line of descent and to study the implications of heat extraction from parental CI magma with respect to the origin of compositional zonation and the probability of explosive eruption. Although the CI magma body did not evolve as a perfectly closed system, this assumption allows quantitative insight into magma-host rock mass exchange using trace element and isotopic data (see companion contribution by Bohrson et al.). The parental melt composition was reconstructed using data for melt inclusions trapped within CI clinopyroxene phenocrysts reported by Webster et al. (2003), while allowing for reaction between parental melt and clinopyroxene host. The inferred parental melt is a basaltic trachyandesite. The search space for pressure, (f$_{O2}$) and initial dissolved H2O was 0.1-0.5 in 0.05 GPa increments, QFM-1 to QFM+3 and 1, 2 and 3 wt. % H2O, respectively. The criteria used to judge the quality of a simulation include correspondence of the MELTS prediction with CI liquid and phenocryst compositions. Results indicate that a good first-order model involves evolution from a basaltic trachyandesite parent by isobaric (~0.15 GPa) crystal fractionation initially containing ~3 wt% dissolved H2O along the QFM+1 buffer. H2O first saturates at 1127°C at 0.15 GPa when the dissolved water content is ~4 wt %. A striking result is the discovery of a pseudo-invariant point at ~883°C (T$_{ip}$) and 0.15 GPa. The fraction of melt changes abruptly from ~0.5 to ~0.1 at T$_{ip}$ due to the simultaneous crystallization of alkali feldspar, plagioclase, spinel, biotite and apatite. At T$_{ip}$, there is a dramatic decrease in the viscosity of melt (by a factor of four) and magma density (~5%) and an increase in the dissolved H2O content of the melt (from 4.4-5.1 wt%) and in the volume fraction,$\theta$, of supercritical fluid in the multiphase system. In particular, $\theta$ increases from ~0.05 at 885°C to ~0.6 at 882°C. The liquid composition also changes discontinuously at T$_{ip}$ with Si, Na, and H2O increasing and K and Al decreasing as temperature falls below T$_{ip}$. The marked variations in composition and properties of volatile-saturated melt and magma were the trigger that led to the catastrophic eruption and formation of the compositionally-zoned CI magma. Because phase equilibria modeling provides information on the enthalpy changes associated with fractional crystallization and because the dimensions of the CI magma chamber and heat extraction rate can be approximated, a time scale for CI magmatic evolution can be derived. The estimated crystallization duration ($\tau$) is10-100 ka and 75% of $\tau$ is spent at or near T$_{ip}$.
V13B-0551
Petrogenesis of Andesites and Dacites From the Southern Juan de Fuca Ridge
The Cleft segment and the ridge-transform intersection (RTI) of the Southern Juan de Fuca Ridge have been investigated during three cruises of MBARI's R/V Western Flyer beginning in 2000. A total of 53 rock cores and 276 precisely located rock or glass samples were collected during sixteen dives with the ROV Tiburon. These ROV dive samples and observations allow us to test models regarding the magmatic evolution of this segment and the relationships between specific tectonic and morphologic features and magmatic processes. An extremely wide range of N-type MORB lavas were recovered which are, on average, more evolved (lower MgO) off-axis, away from the present neovolcanic zone, and towards the RTI. ROV dive T735 investigated a region of unfaulted, curved volcanic ridges that overshoot the Blanco Transform and recovered 39 samples ranging from ferrobasalt to andesite and dacite (SiO2 = 50.1 to 66.9 wt.%; Mg\# = 49.9 to 9.7). The highly evolved lavas were recovered from two large constructional domes comprised of unusually large pillow flows, and extremely blocky, vesicular flows - similar to some terrestrial silicic domes. Some of the andesite-dacite hand samples are extremely vesicular with elongate vesicles (1-10 cm) in a glassy matrix. The dacitic melts have an H2O content of ~1.7% with a vapor saturation pressure of ~240 bars. The large vesicles appear to be the result of the high H2O content, causing the exsolution of H2O and CO2 vapor. The andesitic and dacitic lavas also have elevated Cl- levels that range from 4000-6000 ppm. Mineral assemblages are dominated by microphenocrysts of ferroaugite and ferropigeonite, with lesser amounts of sodic plagioclase and FeTi oxides. Rare zircon, disequilibrium fayalite and myrmekitic textures of plagioclase and quartz are present. A few of the more magnesian phenocrysts (xenocrysts?) exhibit fine normal zoning whereas more Fe-rich crystals exhibit fine reverse zoning. Additionally, inclusions of quenched basaltic material appear within some of the evolved lavas. These samples represent an extensive and unique set of some of the most highly fractionated ocean floor rocks that have ever been recovered; particularly from such a well-documented setting. Fractional crystallization models which predict over 80% crystallization do not adequately explain the major element chemistry of the silicic lavas and most incompatible trace elements exhibit significant enrichments relative to predicted concentrations. The highly evolved nature of the dacites, crystal zoning patterns and the presence of basaltic inclusions suggest the lavas are the result of mixing between two crystal bearing end-members, i.e. a typical basalt and a rhyolite likely generated from partial melting of oceanic crust. Isotopic analyses (Sr, Nd, Pb) are in progress in order to better constrain the sources involved. Periodic propagation of mantle-derived melts or down-ridge diking events into the static RTI environment may lead to the generation of such silicic oceanic lavas.
V13B-0552
Evolution of the Campanian Ignimbrite Magmatic System II: Trace Element and Th Isotopic Evidence for Open-System Processes
The Campanian Ignimbrite, a large volume (~200 km3 DRE) trachytic to phonolitic ignimbrite was deposited at ~39.3 ka and represents the largest of a number of highly explosive volcanic events in the region near Naples, Italy. Thermodynamic modeling of the major element evolution using the MELTS algorithm (see companion contribution by Fowler et al.) provides detailed information about the identity of and changes in proportions of solids along the liquid line of descent during isobaric fractional crystallization. We have derived trace element mass balance equations that explicitly accommodate changing mineral-melt bulk distribution coefficients during crystallization and also simultaneously satisfy energy and major element mass conservation. Although major element patterns are reasonably modeled assuming closed system fractional crystallization, modeling of trace elements that represent a range of behaviors (e.g. Zr, Nb, Th, U, Rb, Sm, Sr) yields trends for closed system fractionation that are distinct from those observed. These results suggest open-system processes were also important in the evolution of the Campanian magmatic system. Th isotope data yield an apparent isochron that is ~20 kyr younger than the age of the deposit, and age-corrected Th isotope data indicate that the magma body was an open-system at the time of eruption. Because open-system processes can profoundly change isotopic characteristics of a magma body, these results illustrate that it is critical to understand the contribution that open-system processes make to silicic magma bodies prior to assigning relevance to age or timescale information derived from isotope systematics. Fluid-magma interaction has been proposed as a mechanism to change isotopic and elemental characteristics of magma bodies, but an evaluation of the mass and thermal constraints on such a process suggest large-scale fluid-melt interaction at liquidus temperatures is unlikely. In the case of the magma body associated with the Campanian Ignimbrite, the most likely source of open-system signatures is assimilation of partial melts of compositionally heterogeneous basement composed of older cumulates and intrusive equivalents of volcanic activity within the Campanian region. Additional trace element modeling, explicitly evaluating the mass and energy balance effects that fluid, solids, and melt have on trace element evolution, will further elucidate the contributions of open vs. closed system processes within the Campanian magma body.
V13B-0553
Experimental results on decompression crystallization in an Aleutian basaltic-andesite
In 1991-92, Westdahl volcano in the Aleutians erupted basaltic andesitic magma, producing lava fountaining and ash clouds to 2 km asl, resulting in a lava flow that extended about 7 km from vent. In order to understand the magmatic ascent rates during that relatively non-explosive eruption, this study compares plagioclase and orthopyroxene microlite textures in the basaltic andesite to crystallization kinetics of plagioclase and pyroxene in laboratory experiments. The experiments used melt-rich basaltic andesite that was hydrated at 150 MPa, 1000° C, and an oxygen fugacity of NNO. Before decompression, the starting material was mainly water-saturated melt that contained a few percent of Fe-Ti oxides and pargasite. Pieces of the starting material were then decompressed rapidly to 25 MPa, and held for various times. Preliminary results show that when held at low pressure for less than 1 hour no crystallization occurred, except for some growth around existing pargasite and oxides. In runs held for about an hour both enstatite and plagioclase nucleate and grow, and are skeletal. In runs held from 4 to 24 hours, plagioclase growth dominates, as enstatite remains a minor phase. The single-step decompressions will be compared with runs using multiple steps to examine how the kinetics change. We can then compare the results with observed crystallization textures in the Westdahl lava and tephra to infer the path and rate of magma ascent during that eruption.
V13B-0554
U-series Disequilibria in Continental Arcs: NE Japan Case
Basalts and andesites from continental arcs typically are close to equilibrium between ($^{238}$U) and (230Th), and have subdued excesses of $^{226}$Ra compared to oceanic arcs (Turner et al., 2003). There is ambiguity whether these and other geochemical features derive from subducted sediment, subcontinental lithosphere, or the crust. We report new $^{238}$U-$^{234}$U-230Th-$^{226}$Ra data for 20 historical to Holocene samples from Asama in the south of NE Japan to Tarumai in the north. Most straddle the equiline with <5% excess U or Th, yet have 10-60% excess $^{226}$Ra. The only significant excess U occurs at Usu and Iwaki in the north where it correlates with higher (230Th)/($^{232}$Th) as in other arcs (e.g., Mariana). Otherwise (230Th)/($^{232}$Th) ratios are 0.75-1.05 which is less than in the surrounding Izu and Kurile oceanic arcs. Most ratios <0.95 are associated with trace element evidence of crustal contamination that drags samples down the equiline. The exception is Iwate where Th/U ratios are ~4.0 even in depleted basalts. Excess $^{226}$Ra is greater in basalts than andesites. The excess $^{226}$Ra, plus elevated 10Be and Th, is attributed to ubiquitous slab inputs that are overprinted by even higher Th contents from a crustal source with Th/U >4.
V13B-0555
Two-Pyroxene Thermometry's Inability to Recover Temperatures of Hydrous Arc Andesites
Two-pyroxene thermometry is one of the most commonly used methods for constraining crystallization temperatures of igneous rocks. It is based primarily on the distribution of Ca, Mg, and Fe between orthopyroxene and clinopyroxene. Numerous variants of the two-pyroxene thermometer have been calibrated over a wide range of temperatures ($700-1500° C) and compositional conditions (e.g. Davidson and Lindsley, 1985; Bertrand and Mercier, 1985; Brey and Kohler, 1990; Fonarev and Grapchikov, 1991). However, based on comparison with recent experimental results, the applicability of the two-pyroxene thermometer for hydrous arc andesites is refuted, and its use elsewhere is questioned. The run products of several water-saturated phase equilibria experiments on a calc-alkaline basaltic andesite (55.3 wt% SiO2), andesite (62.6 wt% SiO2), and two high-Mg andesites (62.2 and 62.0 wt% SiO2) from western Mexico were fully characterized in detailed microprobe analyses by Moore and Carmichael (1998) and Blatter and Carmichael (2001). Experiments were performed at temperatures between 930 and $1000° C and pressures between 441 and 2280 bars. Experimental equilibration was tested by comparing long duration runs, in which all phases were homogeneous in composition and had euhedral shapes. The starting materials were phenocryst-poor and relict crystals were not observed in the run products. The experimental run temperatures could not be recovered using the compositions of the co-existing pyroxenes and subsequent calculations from QUILF (Andersen et al., 1993) or iterative determinations from MELTS (Ghiorso and Sack, 1995). The crystallization temperatures yielded by QUILF were typically too high by $~100° C, but ranged from -186 to +$177° C. The crystallization temperatures obtained from MELTS were more problematic, with several experiments not converging on a unique temperature and others off by up to $800° C. These results suggest that these two-pyroxene thermometers may not be adequately calibrated for hydrous andesites. As seen in Lindsley's (1983) graphical representation of two-pyroxene thermometry, the isotherms for the Ca component in orthopyroxene are very tightly spaced, so the temperature calculation is controlled primarily by the Ca component in the clinopyroxene. However, in the temperature interval of the phase-equilibria experiments ($930-1000° C), the Ca component does not appear to be correlated with temperature. This suggests that not only are previous calibrations of two-pyroxene thermometry not applicable for arc magmas, but that the attempt to find a calibration based on Ca content in this temperature range is a flawed approach. The experience of this study leads us to call into question the reliability of two-pyroxene thermometry for any igneous rocks which are not almost identical to those used in experimental calibrations.
V13B-0556
U-series disequilibrium of basaltic rocks from Kick'em-Jenny submarine volcano, Lesser Antilles island arc
Kick'em Jenny (KEJ) submarine volcano located 9 km to the north of Grenada in the Lesser Antilles volcanic arc produces lavas ranging in composition from high MgO basalts to moderately evolved andesites. We have determined U-series disequilibria in 12 porphyritic lavas erupted from KEJ volcano by TIMS and MC-ICP-MS methods to constrain the timing and identify the processes creating the magma diversity observed. The SiO2 contents of samples studied here vary from 47 to 55 wt.% SiO2 while REE patterns evolve from slightly LREE enriched, MREE/HREE = 1 patterns to strongly LREE enriched, MREE depleted concave-Up patterns. Separate dissolutions of sample KEJ100 indicate an external reproducibility (1s) of 0.7% for (230Th/$^{238}$U) (n=4), 0.8% for (230Th/$^{232}$Th) (n=4) and 0.6% for ($^{226}$Ra/230Th) (n=3), respectively. For all sample, ($^{234}$U/$^{238}$U) lies within 0.7% of unity, suggesting that secondary alteration by seawater has not disturbed the U-series data significantly. Sample ages for these submarine erupted samples are unknown, resulting in uncertain values for initial ($^{226}$Ra/230Th); however, 10 out of 12 of the measured ($^{226}$Ra/230Th) range between 3.16 and 1.13 and are thus unequivocally young with respect to decay of 230Th and $^{231}$Pa since eruption. The U (0.535 - 4.876 ppm) and Th (1.25 - 10.78 ppm) concentrations increase with SiO2 contents. (230Th/$^{232}$Th) has a restricted range, varying from 0.994 to 1.093 with the exception of one sample. (230Th/$^{238}$U) ranges from 0.684 to 0.875 while ($^{231}$Pa/$^{235}$U) ranges from 1.76 up to 2.84, among the highest $^{231}$Pa excess in island arcs yet reported. These data confirm previous observations of the unusual behavior of KEJ lavas relative to global observations in having both large $^{238}$U and $^{231}$Pa excesses. Combined with ($^{226}$Ra/230Th), these disequilibria observations require that $^{238}$U excesses reflect more than solely fluid addition to the mantle wedge from the subducted oceanic slab.
V13B-0557
The 226Ra-230Th-238U disequilibria of enigmatic magmas from Piton de la Fournaise Volcano, Reunion Island (1950-1998)
We present a detailed study of the 226Ra-230Th-238U disequilibria and selected incompatible trace-element abundances of lavas from Piton de la Fournaise, a frequently active ocean-island volcano, using high-precision MC-ICP-MS. The samples erupted between 1950-1998 from vents within the summit caldera (Enclos Fouque) or along the rift zones of the volcano. The lavas display a significant range in 226Ra-230Th disequilibria (25-33% excess 226Ra) and 230Th-238U disequilibria (14-20% excess 230Th). The (230Th/232Th) ratios of the lavas are relatively constant (0.9% range) and do not correlate with either Th/U or (230Th/238U). A strong correlation (R2=0.97) between the (230Th/238U) and Th/U ratios suggests that most of the Th-U fractionation in these lavas occurred recently compared to the 76 kyr half-life of 230Th [because the effects of this fractionation are not yet reflected in the (230Th/232Th) ratios]. In contrast, the 226Ra-230Th disequilbria do not correlate with the Ba/Th ratios of the lavas (a geochemical analog). Thus, the range in Ba/Th (6.1%) and (226Ra/230Th) ratios probably results from either more than one magmatic process or a single process operating over a time scale that is longer than the half-life of 226Ra (1600 years). Unlike Hawaiian shield volcanoes, which display systematic temporal geochemical variations on a time scale of decades to centuries, the fluctuations in lava chemistry at Piton de la Fournaise do not display any coherent trends over the last 50 years. Some geochemical parameters [e.g., Ba/U, Nd/Sm or (230Th/238U)] do vary systematically over a period of a few years, but these trends are not observed consistently for the same samples and different geochemical parameters. The origin of these complex variations in lava chemistry at Piton de la Fournaise is enigmatic, but probably includes a combination of mantle (Albarede and Tamagnan, 1988, J. Petrol. v. 29) and crustal (Sigmarsson et al., 2005, EPSL v. 234; Vlastélic et al., 2005, J. Petrol. v. 46) processes. Possible explanations for the data will be explored at the meeting.