V11D-01 INVITED
Enhancing in-situ U-series accessory mineral chronometry to constrain pace and processes of magma accumulation, storage, and eruption
Over the past decade, deciphering absolute crystallization ages at micro-scales through in-situ U-series dating has spearheaded conceptual changes in the understanding of how magmas evolve en route to the surface. U-Th and U-Pb zircon geochronology by secondary ionization mass spectrometry (SIMS) now routinely accesses a continuous age spectrum extending well into the Holocene with ka precision for individual analysis spots ~25 - 40 μm in diameter. SIMS depth profiling further increases spatial resolution at least tenfold. These developments are epitomized by studies that document multiple generations of zircon growth, predating eruptions by 10's to 100's of ka, as common features in silicic volcanics. Recently, these methods have been enhanced by (1) linking ages to Ti-in-zircon thermometry and trace element or oxygen isotopic fingerprinting; (2) tracking compositional changes with age in accessory minerals sensitive to magma differentiation, such as allanite; and (3) combining U-Th zircon ages with other chronometers, such as (U-Th)/He zircon or 226Ra-230Th-238U plagioclase dating. The emerging picture reveals a wide gamut of processes acting in magma bodies, including piecemeal, compartmentalized magma aggregation, protracted and thermally retrograde crystal storage coupled with magma differentiation, thermal rejuvenation and recycling of crystal mushes and solidified intrusions or magma chamber rinds, and wall rock assimilation. For the future, experimental constraints on accessory mineral-melt partitioning (e.g., Ayers and Luo, 2008: Geochim Cosmochim Acta, v. 72, p. A39) promise to put interpretation of zircon compositions on firmer footing, as is the case for additional accessory mineral saturation calibrations and improved modeling of existing ones (e.g., Harrison et al., 2007: Geology, v. 35, p. 635–638). By the same token, new experimental constraints on U-series element partitioning (e.g., Miller et al., 2007: Am Min, v. 92, p. 1535-1538) should enhance precision and applicability of mineral chronometers. Moreover, analytical advances in high spatial resolution SIMS hold promise for utilizing melt inclusions in zircon as an additional repository for constraining the chemical and thermal evolution of magmas.
V11D-02
Do 226Ra-230Th isochrons provide realistic crystallization ages?
In this contribution we investigate the timescales of melt evolution and crystal growth in the Mt Erebus magmatic system using measurements of 238U-230Th-226Ra-210Pb-210Po and 232Th-228Ra-228Th. Our sample suite consists of 22 historic bombs, ranging from 1972-2005; and 5 anorthoclase megacrysts separated from historic bombs for the years 1984, 1989, 1993, 2004, 2005.These samples 238U-230Th and 230Th-226Ra are significant and uniform over the 36 year historical record. The anorthoclase megacrysts and phonolite glass show complimentary 226Ra/230Th disquilibria. In all samples, 210Pb/226Ra are in secular equilibrium for both phases. For the phonolite glass 228Ra/232Th is in equilibrium, whereas in the anorthoclase megacrysts 228Ra/232Th is significantly greater than unity. For the 2005 bomb, whose eruption date is known explicitly, 210Po was not completely degassed. In-situ ion probe measurements of Ba and Th in the anorthoclase and phonolite glass show that our anorthoclase and phonolite glass separates are pure. Instantaneous crystal fractionation, with long magma residence time (greater than 100 years and less than 3 kyrs, depending on DBa/Ra), can account for the 238U-230Th-226Ra-210Pb systematics. However, the significant 228Ra/232Th disequilibria in the anorthoclase megacrysts preclude this simple interpretation. To account for all of the decay-series and trace element data we have developed a continuous crystallization model, which incorporates both nuclide in-growth and decay during crystallization. Our model can successfully reproduce all of the measured 238U and 232Th decay series disequilibria and is consistent with numerous petrological observations, which suggest that anorthoclase crystal growth was episodic and continued right up to eruption. More importantly, this model shows that when the timescale of crystallization is comparable to the half-life of 226Ra, the simple 230Th-226Ra isochron techniques typically used in most U decay series studies provide erroneous ages.
V11D-03
Integrated Isotopic Dating and Geospeedometry: Timescales of Rhyolite Rejuvenation and Crystallization at Yellowstone
Two techniques have been primarily used to quantify the timescales associated with the evolution and residence of voluminous rhyolitic magmas: 1) isotopic dating, especially using U-Th-Pb dating of accessory minerals, and 2) "geospeedometry" using the diffusional relaxation of intra-crystal zoning of trace element or isotope composition. These methods can resolve different aspects magma evolution over short and long timescales, yet few studies have combined both. To evaluate the timescales associated with crystallization, storage, and rejuvenation of rhyolitic magma at Yellowstone caldera, we have applied ion microprobe 238U-230Th dating of zircon and geospeedometry using zoned sanidine and pyroxene phenocrysts from the ~260 ka Scaup Lake flow (SLF). The SLF is one of the youngest and most-evolved lavas of the Upper Basin Member, rhyolites which mainly erupted 100 kyr after the 0.6 Ma caldera-forming Lava Creek Tuff, and well before the effusion of the voluminous Central Plateau Member lavas between 170 and 70 ka. SLF zircons yield a 230Th-corrected U-Pb age of 312±45 ka (2σ, n=19, MSWD=2.7), suggesting magma generation on the order of 104-105 years before eruption. Most SLF sanidines have significant intra-crystal zoning with multiple resorption surfaces and rims with relatively high Ba and Sr concentrations. Clinopyroxene cores contain exsolution lamellae and are overgrown by inclusion-free rims. Quartz phenocrysts have rims with higher Ti concentrations than their cores. These petrographic observations suggest recycling of subsolidus residue from earlier intrusions and renewed crystallization in new and hotter melt. Diffusion profiles for trace elements across resorption boundaries near sanidine cores yield timescales on the order of 103-104 years, assuming temperatures of 850° C. Profiles across near-rim boundaries suggest insignificant diffusion and that rims formed immediately prior to eruption. Timescales from Fe-Mg diffusion between clinopyroxene cores and rims are similar to those from sanidine. Application of both isotopic dating and geospeedometry to SLF minerals yields timescales over several orders of magnitude, which reflect different aspects of magma evolution. The ages of SLF zircons are likely to reflect a mixture of recycled crystal residue and renewed crystallization. Contrasting timescales from geospeedometry are dependent on petrographic context and reflect different evolutionary "milestones" in the chronology of the SLF magma.
V11D-04 INVITED
The ICSD Method: Integrating Crystal Size Distributions (CSD) With Isotopic Microanalysis
Integration of state-of-the-art micro-geochemical and textural analysis techniques is a powerful way to
constrain igneous systems. This presentation will give a summary of the ICSD method and utility of
integrating CSD and micro-isotopic data. This allows us to reconcile the textural information with isotopic
measurements to trace inheritance and magma mixing processes. Further, we can use the CSD to perform
an integration of isotopic ratio through the crystal population, yielding the isotopic ratio of the crystal
population as a whole and how it has developed through time. A test conducted on samples from Stromboli
allows us to explore the crystal inheritance process and to correct out the inherent sampling bias in
microanalysis towards analysing larger crystals. Combined with a growth rate, we can then explore aspects
of the timing of magma mixing and isotopic ratio shifts due to changing magma supply at Stromboli volcano.
These are similar to those timescales observed directly in the modern system. Coupling these two
techniques is therefore of potential use in constraining the dynamics of volcanic processes that have been
highlighted by isotopic variations within single crystals.
We will further present similar considerations for Santorini volcano and discuss the type of analyses needed
for meaningful ICSD analysis.
http://dx.doi.org/10.1016/j.epsl.2007.05.037
V11D-05
How much warning before a supereruption? Studies of Ti-zoned quartz in the Bandelier Tuff
It is widely recognized that silicic eruptions may be triggered by magmatic recharge and thermal rejuvenation of magma resident in shallow chambers. The time delay between the recharge event and the eruption is not well constrained, particularly for caldera-forming supereruptions (CFSE), but Wark et al. [1] used a promising approach to estimate a <100-year timescale for recharge and eruption of the Bishop Tuff employing recent advances in Ti-in-quartz thermometry and kinetics. Their method relied on measuring the width of Ti zone boundaries in quartz to estimate maximum residence times of quartz grains at magmatic temperatures, with the assumption that a major step in Ti content and hence CL brightness (a proxy for Ti content) observed in quartz grains from late-erupted pumice represents the recharge event. We have followed the same approach to estimate the residence time of Ti- and CL-zoned quartz in the Bandelier Tuff CFSE at 1.6 Ma from the Valles caldera, New Mexico. The tuff exhibits internal isotopic and trace-element variations consistent with recharge. Late-erupted quartz grains in the Otowi Member have thick CL-bright rims with ~50 ppm Ti, overgrown on CL-dark cores with ~22 ppm Ti. The overgrowth boundary truncates CL growth zoning within the core and hence represents a heating and resorption event that we associate with recharge. Using the recently-installed Jeol 8500F field-emission microprobe at Washington State University, we have measured the CL-brightness zone boundaries in several grains and find them to be at most 2 microns wide. This constrains zoned quartz residence times to a maximum of 20 years at 750°C using the Ti diffusivity of Cherniak et al. [2]. The time that elapsed between recharge and the onset of quartz regrowth is still unconstrained. This result nonetheless invites the question of how much warning society can expect before the next CFSE, even if the geographic site is known and a subsurface recharge event can be recognized as such when it occurs. [1] Geology 35, 235-238 (2007) [2] Chem. Geol. 236, 65-74 (2007).
V11D-06
Prolonged Storage of Andesitic Recharge Magmas Inferred From 238U-230Th-226Ra Mineral Separate Data
Over the last decade studies in arc settings have shown evidence for fast transport rates of magmas through the crust (Turner et al., 2001) and rapid crystallization by decompression during the magma ascent through the crust (Blundy and Cashman, 2005). While for island arcs with a thin crust source-to-surface magma transport may be rapid, continental arcs may be more likely to experience stalling and differentiation of magma batches at multiple levels. As a consequence, basaltic magmas are rare in many continental arcs. Instead, basaltic andesites comprise the most mafic magmas that erupt at the surface or mix in the shallow crust with more evolved magmas to produce hybrid silicic andesites and mingled dacites. However, it remains unclear how quickly these mafic magmas ascend through the crust, and how long dacitic magmas are stored prior to eruption. To address these questions we studied the crystal residence ages of their plagioclase phenocrysts. In particular, we investigated basaltic andesites from Volcan Quizapu that mixed and mingled with large volumes of dacite magma during the large (~ 5 km3) 1846/47 AD eruption (Hildreth and Drake, 1992). We dated bulk mineral separates in andesitic enclaves, hybridized silicic andesites and mingled dacites using 238U-230Th-226Ra disequilibria. All plagioclase-glass pairs yield model ages of several thousand years for the 230Th-226Ra system when corrected for Ra/Ba fractionation in plagioclase. The hybridized silicic andesites and mingled dacites yield crystal residence ages of ~ 3000 yrs. The two analyzed enclaves of andesitic recharge magmas have distinct crystal ages: one enclave separate (~ 57 wt.% SiO2 shows a 230Th-226Ra age of ~ 2000 years for the plagioclase- groundmass pair, whereas the second, more mafic enclave separate (~ 54 wt.% SiO2) has a crystal residence time of ~ 5200 years. This suggests that (1) the andesite recharge magmas were stored prior to their mixing with the main dacite magma for up to several thousand years, (2) various andesitic magmas of distinct age coexisted in the plumbing system prior to mixing and subsequent eruption, and (3) dacitic magmas are stored for thousands of years prior to mixing and eruption. The potential for contamination of the andesitic plagioclase separate by glass or other plagioclase populations must be carefully accounted for. The more mafic enclave sample has been extensively studied by electron microprobe, and thin section plagioclase populations have been compared to grain mounts of a split of the dissolved bulk separate. In thin section plagioclase and groundmass appear to be co-genetic and only one plagioclase population exists. The analyzed bulk plagioclase separate from this sample also contains sodic plagioclase (~ An35) originating from the dacite end-member during the mineral separation procedure; however, the amount of contamination was quantified through electron microprobe work on grain mounts. Correcting the model ages to account for the presence of dacitic plagioclase or glass in the analyzed plagioclase separate results in even longer residence times for calcic plagioclase, thus the implications of our results for the storage of andesite recharge magma underneath Volcan Quizapu are robust to these effects.
V11D-07
Homogenization processes in silicic magma chambers by stirring and latent heat buffering
Volcanic eruptions commonly sample chemical and thermal heterogeneities as many deposits from large explosive events show gradients in composition, crystallinity and temperature. However, the most viscous of all magmas - crystal-rich rhyolites-dacites and large silicic plutons - are, in contrast, strikingly homogeneous. This observation is in stark contradiction to the common assumption that the main homogenizing process in magma chambers is mechanical mixing by convection. During transient convection, the total amount of strain determines whether homogeneity can be attained. Here, we show that a convective magma body needs 5-10 overturns following the introduction of heterogeneities to be homogenized to lengthscales such that diffusion is effective, irrespective of the vigor of the convection. For heterogeneities that are continuously re-established or introduced by the convective process itself (such as density instabilities generated by crystallization at the cooling boundaries of magma chambers), convective stirring is unable to lead to complete homogeneity at crystal fraction < 40 vol%. Therefore, to explain the low variability in major element whole-rock composition in crystal-rich dacitic/rhyolitic ignimbrites and silicic plutons, we propose that another mechanism of homogenization acts to decrease thermal (and related crystallinity) variability at high crystal fraction (> 50 vol%). It is induced by latent heat buffering of silicic magmas close to the haplogranitic eutectic, and leads to the rapid equilibration of temperature (and crystal fraction) throughout the magma reservoir. As this process drives the magma body towards a uniformly high crystallinity ("crystal mush"), we refer to it as "mushification". We use a combined numerical method based on stagnant-lid convection at low crystal fraction (< 45 vol%) and a lattice Boltzmann thermal conduction with phase change at high crystallinities (> 45 vol%) to show the strong dependence of the mushification process on the slope of the melt fraction-temperature relationship close to the solidus.
V11D-08
Batholith Maturation as Inferred From Zircons From the Aucanquilcha Volcanic Cluster, Northern Chile
New in situ (SHRIMP-RG) U/Pb and trace element data from zircons from the Aucanquilcha Volcanic Cluster (AVC), in northern Chile, help constrain the evolution of the magmatic underpinnings of a long-lived (11 Ma- present), intermediate volcanic complex through progressive stages of development. Large age spectra (<2.5 my) are observed in the samples from the beginning and waning stages of the AVC, times characterized by low eruption rates, while small age spectra (<500 ky) are typical of samples from the middle stage of the AVC, a time characterized by high eruptive output. Samples with large age spectra likely represent a cooler magmatic setting where discrete zircon age domains are preserved. Samples with smaller age spectra represent a hotter stage of the magma system during which crystals were largely equilibrated with ~cogenetic magmatic activity. Crystallization temperatures for AVC zircons, calculated using the Ti-in-zircon thermometer, range from 689 to 911 °C, with ~80 °C variation in a single sample being typical. Samples from the beginning of the eruptive flare-up (with zircon crystallization Ts of < 911 °C) are hotter than the rest of the samples (Ts of < 815 °C), suggesting a thermal driver for the eruptive increase. Temperature spectra from AVC zircons likely reflect variable zircon saturation histories typical of mafic to intermediate arc magmas. Crystal recycling enhances this effect by mixing together populations of zircons that crystallized from magmas that experienced different (if only slightly) fractionation histories. Such variation is apparently not expected in granites, where magmas are interpreted to have been generated at eutectic melt conditions and thus represent a highly regulated, repeatable process (Harrison et al., 2007). A composite view of all zircon spot ages and the cumulative volume vs. time plot of the AVC show that zircon crystallization and eruptive episodes are temporally correlative, suggesting that the timing of plutonic and volcanic events correspond. The oldest (antecrystic) grains analyzed correspond to the earliest eruptions. Also, no zircons analyzed hail from a roughly two million year eruptive gap from ~8 Ma to ~6 Ma. Samples from the last stage of AVC volcanism contain antecrysts that date back to the period of high eruptive output-- but not older--suggesting a system-wide eradication of zircons during the thermal maturation of the AVC. Alternatively, older sectors of the evolving AVC batholith are less likely to be sampled with progressive mush development as the magmatism becomes more centralized in the complex. This apparent masking of magmatic activity in the plutonic setting via thermal overprinting—with only small, likely marginal zones (or even sparse antecrystic grains) as remnant physical documents of the true life spans—may be a common occurrence and thus promote underestimates of the life spans of large, silicic to intermediate plutonic igneous systems. The effect would be more pronounced with intermediate systems, as these magmas are hot and more likely to dissolve extant zircon upon entry into the batholith.