V12A-01 INVITED
U-series disequilibria in crystals: ages as tracers
U-series disequilibria offer a unique perspective on the fates of crystals within magmatic systems. In addition to delimiting the timescales of magmatic processes, crystal ages can be used as a tracer of different crystal populations even in the case where only subtle differences exist between major- and trace-element chemistries of populations. For example, $^{226}$Ra-230Th ages of crystals in Mt St Helens lavas erupted since 2 ka are in some cases several kyr older than eruption ages which, when combined with significant Ra-Th disequilibria in the whole-rocks, suggests protracted crystal storage and entrainment in subsequent batches of magma passing through the reservoir. More broadly, in many cases 230Th-$^{238}$U and $^{226}$Ra-230Th ages measured in the same crystals are discordant. This pattern likely indicates progressive and/or episodic crystal growth where the Th-U ages more closely represent average crystallization ages while Ra-Th ages are weighted toward recent growth, suggesting in turn that some significant fraction of the mass of crystals represent xenocrysts or "antecrysts" recycled from earlier generations of magmas within the same system. Conversely, in cases where ages of different parent-daughter pairs are concordant, mineral separates must be dominated by crystal growth within a relatively narrow time interval relative to the half-life of the shortest-lived daughter isotope. The duration of the crystal record within a given magma can be complicated by crystal recycling and obscured by average ages derived from measurement of bulk mineral separates. One way to extract more information about the proportion and ages of older and younger parts of the crystal population(s) is to analyze different size fractions within the same sample; for example, analyses of different sizes of plagioclase from the ongoing eruption at Mt St Helens are in progress. U-series ages and other crystal-scale geochemical information can also be a powerful combination. For example, preservation of major- or trace-element disequilibrium between zones within crystals limits the duration of crystal residence at high temperature; when combined with absolute age information from U-series disequilibria, these data can provide clues about the thermal conditions of crystal storage and thus whether such storage is likely to have occurred in a mostly-liquid or mostly-crystalline part of the magma system.
V12A-02 INVITED
Oxygen isotope heterogeneity of igneous phenocrysts: the rule and not the exception?
Oxygen isotope geochemistry of igneous rocks relies on analyses of bulk of resistant or ''refractory'' phenocrysts as a tool to measure δ18O value of equilibrium melt and interpret its origin. However, recent study of individual "phenocrysts" and their size fractions for oxygen isotope composition by laser fluorination, and core and rim study by SIMS, reveal isotope zoning and intracristalline variability, which exceeds analytical uncertainty, and often reach several permil. Air abrasion helps in retrieving cores of crystals, and has been used to obtain cores of quartz, zircon, sphene, and olivine. Air abrasion of individual crystals followed by laser fluorination, is the most precise (better than ñ1%) approach. SIMS analysis has greater spatial resolution but worse precision. I present examples of 2 to 5 permil oxygen isotope zoning in zircons from Yellowstone and Timber Mt calderas, 3permil zoning in olivines from Laki, Iceland, and 0.6 permil zoning in olivines from Kamchatka, and review other examples from literature. In all studied cases, isotope zoning results from entrainment of normal-δ18O crystals into either low- or high-δ18O melt. The magnitude of isotope disequilibria between melt and different minerals in the same rock, and between core and rim of the same mineral, permits estimate of time since the beginning of exchange ("entrainment") and the eruptive quench. Complementary tools, such as trace elemental zoning profiles in phenocrysts, ion microprobe dating of zircon cores and rims, and 210Pb-226Ra disequilibria dating of melt, provide similar timescales. The oxygen diffusive and model time estimates call for short magma segregation and eruption timescales of 10-1000 years. Therefore, isotopically-zoned crystalline material in magmas is often recycled cumulates and xenocrysts in secular equilibrium, rather than phenocrysts. Long-enough residence of crystals in melt will erase their isotopic zoning and lead to equilibrium Δ18O(mineral-melt) so xenocrysts can then adequately reflect δ18Omelt. An igneous rock should be studied by analysis of several individual crystals.
V12A-03
Phenocryst Dynamics in the Magma Chamber of the Guadeloupe Soufriere From U-Th-Ra Disequilibria in 1440A.D. Lavas
U-Th and Ra analyses of andesitic to dacitic products of the Soufriere last eruption (1440+/-50 A.D.) were performed in order to constraint the time scales and the nature of magmatic processes preceeding eruption. The stratigraphic sequence indicates that, dacitic lavas were first erupted followed by more andesitic lavas, similar to that of the actual dome. Most whole rocks show $^{238}$U and $^{226}$Ra excess relative to 230Th, like many others arc lavas. Their ($^{226}$Ra/230Th) activity ratios decrease with differentiation degree from 1.507 to 0.983, whereas the (230Th/$^{238}$U) ratios increase from 0.942 to 1.003. The groundmasses of the least differentiated lava show a ($^{226}$Ra/230Th) ratio higher than 1, but all the others are at secular equilibrium. Time dependent fractional crystallization models assuming a common andesitic parental liquid in a vertically zoned magma chamber can explain the compositional variations and U-Th-Ra disequilibria of 1440 Soufriere lavas and groundmasses. The timescales of magma differentiation inferred by fractional crystallization models are of a few ten thousands years for a crystallization rate ranging from 1 to 2.10$^-5$ y$^-1$. The 1440 Soufriere lavas show phenocrysts out of equilibrium with their host melt (complex zoning and overgrowth rims). U-Th disequilibria have been measured in one of the least differentiated lava and provide precious information on their age of crystallization, their origin and the solids dynamics in magma chamber. The crystallization age of 35.0ñ5.9 ky obtained for pyroxenes using U-Th internal isochron is older than the differentiation time of their host lava given by crystallization models but younger than the differentiation time required to get the most evolved composition. That suggests old cumulates remobilization (recycling), whereas the liquids must have evolved separately. Moreover, the magnetite and plagioclase separates attest of a U-Th disturbance. In fact, they do not plot on the isochron defined by the whole rock, the groundmass, the clinopyroxenes and the orthopyroxenes. The magnetite phenocrysts show a very particular U-Th signature. Taken published partitioning coefficient data, they have too high U/Th ratio to have crystallized from any magma of the Soufriere or of Lesser Antilles arc. It thus requires an episode of U-enrichment after crystallization, sufficiently old to allow a strong increase of (230Th/$^{232}$Th) ratio. Probably dragged from an altered wall-rock, the magnetites suggest an assimilation process. By comparing the ($^{226}$Ra/230Th) ratios of whole rocks and groundmasses, ($^{226}$Ra/230Th) ratios of mineral assemblages can be inferred. Greater than 1 for most lavas, they show that some plagioclases have crystallization ages younger than 8000 years which is shorter than the differentiation timescale of their host lavas. It must result of a late crystallisation in their host lava or of a crystallisation from a magma batch injected later than their host lava and followed by their ascent through chamber by flotation. The U-Th-Ra disequilibria of minerals also show evidence for minerals exchanges between the different magma batches by sinking or flotation and of surrounding assimilation.
V12A-04
Deciphering Multistage Crystal Histories in Arc Magmas
Discrepancy between crystal ages derived by short-lived chronometers with vastly differing half-lives is one manifestation of the potential for complex, multistage evolution of phenocrysts in arc magmatic systems. Deciphering these processes is critical for estimating realistic crystal histories and, ultimately, the physical mechanisms of differentiation. Some of the biggest chronological discrepancies are evident in the andesitic compositional range, the most ubiquitous material erupted at arcs. In some systems, such as Sangeang Api in the Sunda arc, U-Th and Ra-Th systematics of bulk plagioclase separates are not in conflict and indicate that differentiation occurred over several 1000 years via crystallization due to cooling in the lower crust. Here, $^{210}$Pb data indicate significant degassing occurred in the decade prior to eruption but post-dated phenocryst growth and magma differentiation. Combined textural and U-Th-Ra isotope approaches often, however, provide compelling evidence that plagioclase phenocrysts contain old cores and thus are zoned in both age and composition. One of the best examples of apparently conflicting time-scale information comes from Soufriere volcano on St. Vincent in the Lesser Antilles. U-Th isotopes analyses of bulk plagioclase separates conflict with whole-rock and mineral Ra-Th disequilibria and attest to non-linear growth histories, and involvement of recycled cumulates upon which renewed crystal growth has taken place. We augment this well-constrained case study with new in situ Sr isotope analyses for one of the Soufriere lavas and a cumulate xenolith erupted in 1979. Significant isotope heterogeneity is observed, and complimentary isotope variations exist between cumulate xenolith and lava plagioclase phenocryst cores, lending further support to the model of heterogeneous core-rim evolution in the Soufriere system. We conclude that mineral time scales should always be cross-examined with other textural and/or isotope techniques where possible.
V12A-05
Open-system degassing, gas fluxing, and the growth and entrainment of olivine during small volume mafic eruptions: A case study of Volcan Jorullo, central Mexico
Olivine-hosted melt inclusions from mafic arc volcanoes commonly contain highly variable H2O and CO2 contents that are best explained by olivine crystallization over a wide range of depths during ascent and degassing. An excellent example of this is found in deposits from the 1759-1774 eruption of Jorullo volcano in the subduction-related Trans-Mexican Volcanic Belt. Jorullo's earliest lavas are primitive (9.3 wt% MgO, Fo$_{86-91}$ olivine), and lava compositions evolved over time as a result of crystal fractionation (Luhr and Carmichael, 1985). We analyzed melt inclusions in Mg-rich olivine from a 5-meter-thick proximal ash fall sequence. Melt inclusions from the basal tephra have variable H2O (<1-5.2 wt%) and CO2 (44-900 ppm), corresponding to crystallization pressures of <100 bars to 4.2 kb. This indicates that olivine crystallized over a range of depths extending from the mid-crust to very shallow levels beneath the volcano. Relations between H2O and incompatible K, Ti, and P, and comparison to experimental phase equilibria demonstrates that olivine crystallization is driven primarily by the effects of H2O loss from the melt during ascent. A puzzling feature of the H2O and CO2 data at Jorullo and other mafic arc volcanoes is that CO2 concentrations do not decrease as rapidly with degassing as is predicted by degassing models. A likely explanation involves open-system degassing in which relatively CO2-rich vapor fluxes upwards through the conduit. This vapor is probably released by crystallizing and degassing magma at greater depths within the system. Melt inclusions from the upper part of the tephra section at Jorullo have lower H2O (0.2-1.4 wt%) than inclusions from the basal tephra and no detectable CO2, suggesting shallow crystallization of degassed magma toward the end of the violent-Strombolian-style eruptions. The change in olivine crystallization depths over time probably results from evolution of the conduit system. During the early, pre-eruption stage, as magma fractures its way upward and creates a conduit system, olivine crystallizes over a wide range of depths, trapping variably degassed melts. These crystals are probably stored in the complex dike and sill network that is growing beneath the volcano. When magma finally breaks out at the surface, initiating the eruption, magma flux increases, and magma moving upward through the conduit system entrains olivine crystals stored in the dike and sill network. Over time, the conduit system becomes localized, so early-formed olivine from greater depths are no longer entrained, and crystallization occurs mainly at shallow depths beneath the volcano. Such a model may have general applicability to many mafic eruptions and helps to explain how olivine crystals formed at diverse depths and of sometimes diverse compositions become incorporated into a batch of erupting magma.
V12A-06
Old phenocrysts gather on the shores of the Aegean
Information derived from zircon crystals has proven to be extremely useful in deciphering magmatic processes. In particular, their crystallization age can be deduced from U-Th-Pb dating due to the slow diffusion of U, Th and Pb in the zircon lattice, leading to a closure temperature above the magmatic temperature for most silicic magmas. This characteristic of zircon U-Th-Pb dating establishes it as one of the best tools to assess the duration of assembly and longevity of silicic magma bodies above their solidi. It also provides a means of recognizing xenocrystic material in the form of cores inherited from assimilated lithologies. The ~150 km3 rhyolitic Kos Plateau Tuff eruption occurred ~160 ka ago (40Ar/39Ar sanidine age) in the eastern part of the Aegean Arc, leaving a >20 km diam. caldera off shore of the island of Kos (Greece). SHRIMP U-Th-Pb dating of zircons from multiple samples at different stratigraphic levels records an age distribution from ~160 ka (indistinguishable from eruption age by 40Ar/39Ar) to ~500 ka and a peak of zircon crystallization from 200 to 250 ka. These results highlight two important observations: (1) large silicic magma bodies can form and reside above their solidi for >300 ka, (2) the absence of U-Pb ages older than 500 ka, together with the whole-rock isotopic ratios of this rhyolite (ε_{Nd}$ = 0.2-0.9 and $^{87}$Sr/$^{86}$Sr = 0.7040-0.7044) in an area where Palaeozoic crust is exposed at the surface, implies minimal crustal contamination. The low basalt flux, hence low thermal flux, of the dying Aegean subduction zone may explain the growth of this voluminous rhyolitic body without significant mass contribution from the surrounding pre-existing crust.
V12A-07
Strange Attractors: Symbiosis in Magma Crystallization
Volcanic products commonly contain crystal clots and preferential associations of solid phases. These mineral associations respond to the chemical environment during crystal nucleation, growth, and coarsening. Interpreting this record and assessing the implications for mush zone dynamics requires an understanding of the energetics of co-crystallization. Intimate intergrowth of compositionally and structurally dissimilar phases may arise if chemical potentials are elevated by gradients in the concentrations of rejected components, i.e., constitutional supercooling, CS. Alternatively, mineral intergrowths may ameliorate the phase transformation by providing energetically favorable sites for nucleation, e.g., epitaxy. In order for compositional gradients to develop and cause CS crystallization, crystal growth must be rate-limited by diffusion in the melt. Secondary phase nucleation is homogeneous in the boundary layer of a primary phase during CS. Thus, no particular crystallographic orientation relationship is expected between the phases, and all faces of the primary crystal advancing at similar rates should provide equally probable sites for nucleation of a secondary phase. Rejected components are mutually beneficial in pairs related by CS; this mechanism may be important if the frequency of mineral associations correlates with degree of compositional dissimilarity. In contrast, epitaxial relationships indicate heterogeneous nucleation of a secondary crystal at a crystal-melt interface in response to large interfacial energy (σ). A preferred orientation relationship may occur among phases if σ is strongly anisotropic. Finally, epitaxy is expected to be important if co-crystallizing minerals share lattice characteristics. We examine crystal intergrowths in the groundmasses of natural and experimental samples using BSE imaging, EPMA, and EBSD orientation mapping to assess preferential mineral pairings, size relationships, compositional complements, and crystallographic alignments. Grain segmentation facilitated by orientation mapping suggests that crystal number densities may be overestimated using conventional BSE imaging. Preliminary results from near-liquidus basalts and hornblende reaction rims in Mount St. Helens dacite indicate a strong tendency for pyroxene and titanomagnetite to co-crystallize by the CS mechanism.
V12A-08
What's in a Whole Rock Analysis? Integrating Crystal Size Distributions and Micro-Scale Isotopic Variations at Stromboli Volcano
Analysis of feldspar crystals by microsampling at high spatial resolution has allowed the construction of a detailed crystal isotopic stratigraphy for a lava from the lower Vancori sequence (26,000 years before present) of Stromboli volcano. The crystals record the magmatic evolution of the Stromboli system prior to the eruption of the sample. $^{87}$Sr / $^{86}$Sr isotope ratios at nucleation were elevated, with $^{87}$Sr / $^{86}$Sr of 0.70650 relative to $^{87}$Sr / $^{86}$Sr of 0.70617 in the groundmass glass, with gradual mixing between core and rim towards the glass values. Quantifying the crystal size distribution data then provides a time aspect for the isotopic evolution of the crystal population. Integrating these two types of analysis allows the investigation of changes in the chemical budget through time as the crystal population evolved, a powerful tool for investigating magmatic processes. The presence of isotopically distinct cores and rims shows significant retention of crystals occurring beneath Stromboli and the subsequent incorporation of these crystals into erupted materials. These isotopic core-rim relationships are found consistently between different crystals of the same population. That the variations are correlated across crystals shows that the system was both simple in structure and well-mixed in the Vancori period, and that crystal inheritance was likely between different magma batches at Stromboli. After establishing a timeline of variation in the isotopic data, the isotopic variation can be integrated through the CSD analysis to determine the mean isotopic ratio of the crystal population. This can then be compared with the groundmass glass values and, when mixed, reproduces the whole rock isotopic ratio. That the measured whole-rock ratio does indeed reflect the theoretical sum of its component parts confirms the relative simplicity of the feldspar population as well as the validity of assumptions inherent in the CSD analysis of a simple, linear growth rate.