Formation of Low-Δ18o Rhyolites After Caldera Collapse at Yellowstone, Wyoming, USA
Total Page:16
File Type:pdf, Size:1020Kb
Formation of low-δ18O rhyolites after caldera collapse at Yellowstone, Wyoming, USA Ilya N. Bindeman John W. Valley Department of Geology and Geophysics, University of Wisconsin, 1215 West Dayton Street, Madison, Wisconsin 53706, USA ABSTRACT We present a new model for the genesis of low-δ18O rhyolites of the Yellowstone caldera based on analyses of zircons and individual quartz phenocrysts. Low-δ18O rhyolites were erupted soon after the massive caldera-forming Lava Creek Tuff eruption (602 ka, ~1000 km3) and contain xenocrysts of quartz and zircon inherited from precaldera rhyolites. These zircons are iso- topically zoned and out of equilibrium with their host low-δ18O melts and quartz. Diffusion modeling predicts that magmatic disequilibria of oxygen isotopes persists for as much as tens of thousands of years following nearly total remelting of the hydrothermally altered igneous roots of the depressed cauldron, in which the alteration-resistant quartz and zircon initially retained their δ18O values. These results link melting to caldera collapse, rule out rapid or catastrophic magma–meteoric water interaction, and indicate wholesale melting rather than assimilation or partial melting. Keywords: Yellowstone, zircon, oxygen isotopes, caldera, low δ18O. INTRODUCTION rock major and trace element composition is simi- (Spicuzza et al., 1998b). We measured four to Meteoric water plays an important role in the lar to that of isotopically normal high-silica rhyo- seven aliquots of UWG-2 garnet standard on each genesis of ore deposits, explosive volcanism, and lites of precaldera lavas or lavas erupted simul- day of analysis. Nine analyses of NBS-28 quartz hydrothermal activity. Low values of δ18O un- taneously outside the caldera. yielded an average value of 9.55‰. Air abrasion of ambiguously provide evidence of this process. We report analyses of more than 200 indi- zircon fractions was done in a corundum mortar Although the alteration of rocks by meteoric water vidual quartz phenocrysts and zircon separates and normally took from 1 to 5 days. Selected ion is well understood, the incorporation of meteoric from the Lava Creek Tuff and younger intra- microprobe analyses were performed on 20-µm- water into magma remains controversial. Studies caldera lavas (602–70 ka). The ability to analyze diameter spots, 3 µm into prismatic crystal growth of low-δ18O volcanic and shallow plutonic mag- samples smaller than an individual quartz pheno- faces of zircon using a Cameca ims-4f at the Uni- mas in which tens of percent of their total oxygen cryst allows us to measure the phenocryst-scale versity of Edinburgh, Scotland, with a Cs+ primary must have been ultimately derived from meteoric (<1 mm) variability, which was not possible in beam and energy filtering (Valley et al., 1998). or surface water provide important insight into earlier studies. mechanisms of fluid-magma interaction (Fried- ZONING AND VARIABILITY OF δ18O IN man et al., 1974; Muehlenbachs et al., 1974; Criss SAMPLE PREPARATION AND INDIVIDUAL QUARTZ PHENOCRYSTS and Taylor, 1986; Taylor, 1987). The genesis of ANALYTICAL TECHNIQUE AND ZIRCONS low-δ18O magmas has been variously proposed to Individual phenocrysts of quartz and sanidine, Values of δ18O in individual obsidian spheres require: (1) direct, sometimes catastrophic mete- and obsidian spheres, were handpicked from (Table 11) within several lava flows were measured oric water addition into magma (Friedman et al., hand specimens or from crushed rock. Quartz was and are homogeneous (δ18O = ±0.3‰). These 1974; Hildreth et al., 1984); (2) more gradual, purified in fluoroboric acid to dissolve feldspars values are similar to those reported by Hildreth multistage assimilation of hydrothermally altered and glass. Special attention was paid to glass et al. (1984), demonstrating a remarkable homo- rocks (Taylor and Sheppard, 1986; Balsley and inclusions and phenocryst morphology to assure geneity of δ18O in magmas parental to low-δ18O Gregory, 1998); or (3) partial melting of the walls that all analyzed phenocrysts are primary and not lavas. In contrast, all low-δ18O rhyolites erupted of the magma chamber (Bacon et al., 1989). the result of precipitation from a secondary hydro- immediately after the Lava Creek Tuff are dis- Classic examples of low-δ18O lavas are found in thermal fluid. Zircons were separated from 20 kg tinctly variable in the δ18O values of individual the Yellowstone Plateau volcanic field, Wyoming, samples of rock using standard techniques of quartz phenocrysts (Fig. 1). Variations of 1.0‰– USA, where abrupt 18O depletions (accompanied density and magnetic separations. The extracted 1.5‰ are common among quartz phenocrysts by changes in Sr and Pb isotopes) in magmas 50–250 mg of zircons were purified from admixed from a single hand specimen, and variation is in occurred after caldera-forming eruptions at 2.0 and minerals and glass by cold HF. excess of 2‰ for the Canyon flow and the South δ18 δ18 0.6 Ma. The most dramatic drop in O values The University of Wisconsin CO2-laser mass- Biscuit Basin flow. The highest values of O (by ~5.5‰) of quartz phenocrysts occurred after spectrometer system (Valley et al., 1995) allows among quartz phenocrysts in the South Biscuit the last caldera-forming eruption, which deposited precise analysis of 1–2 mg samples with the Basin flow and the Canyon flow approach +6‰, the Lava Creek Tuff and formed the Yellowstone average reproducibility of better than ±0.10‰ 1 caldera (Hildreth et al., 1984, 1991). The >40 km3 (1 standard deviation, st.d.). Quartz was analyzed GSA Data Repository item 200079, Oxygen iso- δ18 δ18 tope ratios,Yellowstone caldera, is available on request of postcaldera, low- O lavas with O quartz by the rapid heating technique (Spicuzza et al., from Documents Secretary, GSA, P.O. Box 9140, (Qz) <3‰ appear exclusively within the Yellow- 1998a). For analyses of more reactive sanidine and Boulder, CO 80301-9140, [email protected], or stone caldera. It is remarkable that their whole- obsidian, we used an airlock sample chamber at www.geosociety.org/pubs/ft2000.htm. Data Repository item 200079 contains additional material related to this article. Geology; August 2000; v. 28; no. 8; p. 719–722; 4 figures. 719 LCT values more than 1‰ higher than either the cal- a SBB 6 Precaldera Quartz culated rim composition based on zircon abra- WY AC CF NBB sion, the ion microprobe analyses of zircon rims, SCL ML or zircon in equilibrium with their host obsidian. 4 Precaldera Zircons This finding suggests that even the smallest MBB grains of zircon are isotopically zoned and con- 2 DR tain higher δ18O cores. Such zoning would sug- gest that much of the total mass of zircons was in- 0 Quartz herited and that diffusive exchange with δ18 O (‰, VSMOW) O (‰, low- O magma was an important process, Zircons: 18 δ cores affecting a larger mass percentage of smaller -2 >105, >149 µm crystals because of their larger surface area/mass. bulk <53 µm It is remarkable that all studied zircons from -4 rim, calculated low-δ18O rhyolites younger than the Lava Creek Tuff show evidence of δ18O zoning, whereas all b abraded zircon fractions from precaldera and Equilibrium ∆ (Qz-Zrc) extracaldera rhyolite rocks are homogeneous. O (Qz-Zrc), ‰ 2 The Scaup Lake flow and other voluminous 18 ∆ rhyolites erupted after an ~200 k.y. quiescence 0 in eruptive activity inside the caldera show no significant heterogeneity in quartz phenocrysts, no isotopic zoning in zircon, and no ∆(Qz-Zrc) 600 550 500 450 200150 100 Age, k.y. disequilibria (Fig. 1). Two processes operated to reequilibrate Figure 1. Oxygen isotope ratios of larger and smaller zircons (Zrc), measured compositions of zircon, quartz, and magma: (1) diffusive oxygen abraded cores, and calculated compositions of rims in low-δ18O rhyolites from Yellowstone caldera,Wyoming. Boxes show ranges of quartz (Qz) δ18O values.These data show (1) variability exchange, and (2) solution reprecipitation on of δ18O values among individual quartz phenocrysts, especially for low-δ18O lavas erupted after inherited cores and new overgrowth in equilib- Lava Creek Tuff (LCT) at 550–450 ka; (2) variability of different size zircons and air-abraded rium with the melt. Oxygen diffusion coefficients cores; (3) disequilibria between average quartz and zircon in most low-δ18O lavas. CF—Canyon in zircon in Yellowstone water-bearing magmas flow, DR—Dunraven Road flow, SBB, MBB, and NBB—South, Middle, and North Biscuit Basin are on the order of 10–17–10–18 cm2/s (water- flows, SCL—Scaup Lake flow, ML—Mallard Lake flow, WY—West Yellowstone flow, AC—Aster Creek flow. Ages are from Gansecki et al. (1996, 1998) and Obradovich (1992). Plotted ages of saturated experiments at 750–850 °C, Watson post-Lava Creek Tuff lavas are not exact to prevent overlap of data points. VSMOW is Vienna and Cherniak, 1997), making it possible to equili- standard mean ocean water. Note that oxygen isotopes permit making of stratigraphic distinc- brate a typical 50-µm-radius zircon grain after a tions. Our results show that three separate outcrops (hills), southern, middle, and northern, few tens of thousands of years (Fig. 2). Sluggish along Firehole River of Upper Geyser Basin, defined earlier as single unit of Biscuit Basin flow, represent three independent lava flows, which have distinctly different isotope compositions of rates of zircon growth and dissolution in magmas –15 –17 phenocrysts. We call them South, Middle, and North Biscuit Basin flows. Only Middle Biscuit (10 –10 cm/s, at the same temperatures and Basin was sampled by Hildreth et al. (1984). NBB and SBB flow ages were not determined. rate of cooling or heating of 0.01 °C/yr, Watson, 1996) suggest that it may take at least tens of thousands of years to form a 50-µm-thick zircon similar to normal-δ18O quartz in precaldera rocks cons have exchanged oxygen as a function of sur- overgrowth.