V31F-01 INVITED
Rare earth element partition coefficients in zircon/melt systems
In principle, rare earth element (REE) concentrations in magmatic zircon crystals can be used to reconstruct growth conditions, provided accurate information is available about the partitioning of REEs between zircon and magma. Over the past twenty-five years, there have been several studies published that used natural and synthetic samples to determine zircon-melt rare earth element (REE) partition coefficients (D's) in geological systems. These data show a remarkable range in values covering several orders of magnitude for some REEs, pointing to as yet unquantified effects of pressure, temperature and/or composition on zircon-melt D's. This calls into question the validity of using these data to calculate the REE composition of co-existing melts. To be able to quantify the effects of P, T and X on zircon-melt partitioning, a theoretical framework to interpret zircon-melt D's is required. As a first step towards this framework, we assessed literature zircon-melt REE D's using the lattice strain model proposed by Blundy and Wood (1994). This model assumes that the primary factor influencing elemental partitioning into a mineral is the strain caused by substituting a trace element into a crystal lattice site. Unconstrained fits to available data give physically unrealistic values for Ro [optimum radius of the zircon Zr site]; Do [(theoretical) strain-free partition coefficient]; and E [apparent Young's Modulus]. Using the well-constrained partitioning data from Hanchar et al. (2001), we derive improved constraints on the relation between E and r0, allowing far more realistic fits to literature data. From the resulting fits it is clear that many LREE D's show significant positive deviations from trends predicted by the lattice strain model. This may reflect analytical problems or mineral inclusions (e.g., monazite), and makes them unsuitable for use in geochemical modelling. The data reported by Sano et al. (2001) for an ion-probe study of REE partitioning between silicate glass, zircon and apatite appear most consistent with lattice-strain based expectations. References Blundy, J., and Wood, B. (1994) Prediction of crystal-melt partition coefficients from elastic moduli. Nature, 372, 452-454. Hanchar, J.M., Finch, R.J., Hoskin, P.W.O., Watson, E.B., Cherniak, D.J., and Mariano, A.N. (2001) Rare earth elements in synthetic zircon. 1. Synthesis and rare earth element and phosphorus doping. American Mineralogist 86: 667-680. Sano, Y., Terada, K., Fukuoka, T. (2001) Ion microprobe analysis of rare earth elements in silicate glass, apatite, and zircon. Eleventh Annual V.M. Goldschmidt Conference, Abstract # 3310, LPI Contribution No. 1088, Lunar and Planetary Institute, Houston (CD-ROM).
V31F-02
Ti Diffusion in Zircon
Diffusion of Ti under anhydrous conditions at 1 atmosphere and under fluid-present conditions at 1.1-1.2 GPa has been measured in natural zircon. The source of diffusant for 1-atm experiments was a ZrO$_{2}$- TiO$_{2}$-ZrSiO$_{4}$ mixture, with experiments run in Pt capsules. Diffusion experiments conducted in the presence of H$_{2}$O-CO$_{2}$ fluid were run in a piston-cylinder apparatus, using a source of ground TiO$_{2}$, ZrSiO$_{4}$ and SiO$_{2}$, with oxalic acid added to produce H$_{2}$O-CO$_{2}$ vapor and partially melt the solid source material, yielding an assemblage of rutile + zircon + melt + vapor. Resonant nuclear reaction analysis (NRA) with the nuclear reaction $^{48}$Ti(p,$\Gamma$)$^{49}$V was used to measure diffusion profiles for both sets of experiments. The following Arrhenius relation was obtained for Ti diffusion normal to c over the temperature range 1350-1550C at one atmosphere: D$_{Ti}$ = 3.3x10$^{2}$ exp(-754 $\pm$ 56 kJ mol$^{-1}$ /RT) m$^{2}$sec$^{-1}$ Ti diffusivities were found to be similar for experiments run under fluid-present conditions. A fit to all of the data yields the Arrhenius relation D = 1.3x10$^{3}$ exp(-741 $\pm$ 46 kJ mol$^{-1}$ /RT) m$^{2}$sec$^{-1}$. These data suggest that zircon should be extremely retentive of Ti chemical signatures, indicating that the recently developed Ti-in-zircon crystallization geothermometer (Watson and Harrison, 2005; Watson et al., 2006) will be quite robust in preserving temperatures of zircon crystallization. Titanium diffuses somewhat faster in zircon than larger tetravalent cations U, Th, and Hf, but considerably more slowly than Pb, the REE, and oxygen; hence Ti crystallization temperatures may be retained under circumstances when radiometric ages or other types of geochemical information are lost. {\it Watson EB, Harrison TM (2005) Science 308, 841-844. Watson EB, Wark DA, Thomas JB (2006) CMP(in press).}
V31F-03 INVITED
Interpretation of Ti-in-zircon Thermometery in Plutonic Rocks
The advent of the empirical Ti-in-zircon thermometer resulted in its immediate application to a large number of geological problems as well as on-going tests of its validity. Among the consequences of note is a new appreciation that zircon growth in slowly crystallizing systems is a complex process rather unlike the widely adopted model in geochronology where a U-Pb zircon date is ascribed to a single temperature. Tests of the thermometer also require recognition of the interplay between zircon solubility and melt chemistry in any liquid- line-of-descent. Possibly complicating such tests is the use of zircon saturation thermometry based on whole rock [Zr] data as a benchmark for comparison with the zircon thermometer. Take, for example, tonalite 101 of Carroll and Wyllie (J. Pet. 46, 275, 1990) with M = (K+Na+2Ca)/(Si Al) =1.9. Assuming [Zr] = 150 ppm (Condie, Chem. Geol. 104, 1, 1993), a completely molten rock of this composition would saturate in zircon at 748$^{o}$C. However, crystallization of modal phases prior to the onset of zircon stability drives up melt [Zr] while lowering M and thus zircon solubility. By parameterizing the Carroll and Wyllie glass compositions (2.4-5.2% H$_{2}$O) with temperature, approximating the crystallization sequence between 1100-700$^{o}$C as an error function, and assuming negligible Zr partition into modal phases, we find that zircon would first begin to grow at 827$^{o}$C when the melt reached 280 ppm Zr. The predicted form of the zircon crystallization temperature distribution depends on specific assumptions regarding the crystallization path, but in our case yields a broad spectrum declining with temperature with less than 20% of zircon growing between 750-700$^{o}$C. If grown as rims on large zircons, this potentially corresponds to an even smaller fraction of area exposed in sectioned grains. We previously proposed the existence of wet, minimum melting conditions throughout the Hadean Eon based on a distinctive peak in crystallization temperatures in >4 Ga zircons at 680\pm25$^{o}$C, with a small fraction of grains yielding temperatures from 750-1100$^{o}$C. This pattern has persisted in subsequent analyses with over 250 zircons now analyzed. While it has been argued that the small overlap between this peak and the broad distribution of zircon temperatures from mafic rocks obscures this distinction, we note that the 680\pm25$^{o}$C distribution could not be plausibly derived from slowly cooled mafic or intermediate rocks without subsequent action by an unknown sorting mechanism. The above calculation formalizes this by illustrating the inevitability of magmas cooling from high temperature producing very broad temperature distributions with the bulk of the zircon mass appearing at temperatures above that estimated from bulk saturation calculations. The simplest explanation for the dominant low temperature Hadean peak is that it reflects prograde melting conditions under near water-saturated conditions. In effect, as soon as the protolith reaches anatectic conditions, the majority of melt fertility is lost due to the presence of excess water. That no subsequent peaks are yet observed corresponding to various vapour-absent melting equilibria supports this conclusion. To test the prograde vs. retrograde thermal history hypotheses, we are undertaking high resolution ion imaging of $^{48}$Ti$^{16}$O$^{- }$ in Hadean zircons to assess whether a broad or narrow spectrum of magmatic temperatures are characteristic of these ancient grains.
V31F-04
Elemental Analysis of Zircon by High Mass Resolution USGS-Stanford SHRIMP-RG: Measuring and Evaluating Ti-in-zircon Temperatures and Compositional Characteristics
High mass resolution of SHRIMP-RG permits measurement of a large set of trace elements for zircon, including 48Ti, Sc, and Nb (requiring better than 9,000 MR) and Be, B, F, P, 49Ti, V, Y, all the REE, Hf, Th, and U (Mazdab and Wooden 2006). A 15-20 micron spot allows analysis of numerous discrete CL zones from single zircons with minimal contributions from unknown material below the exposed surface. Data from suites of zircons from more than 20 individual granitoid samples suggest several general observations: (1) Temperatures calculated by Ti-in- zircon (Watson et al 2006) are entirely compatible with petrologic constraints; uncertainty in a(TiO2) introduces uncertainty in calculated T, but for reasonable values between 0.5 and 0.8 T's consistently fall between 650 and 900 C, mostly in the lower half of the range; (2) T can vary by 150-200 C within suites of zircons from individual samples and even in single zircons, where zonation may be normal (high to low, core to rim), reverse (low to high) or fluctuating; (3) Hf concentrations increase with decreasing T because of Zr/Hf fractionation between zircon and melt (Claiborne et al in press); (4) Many elements and element ratios show a co-variation with T and Hf concentration � e.g., Th/U and MREE/HREE decrease with increasing Hf and decreasing T. Hf concentrations can continue to increase after a minimum T is reached, indicating continuing zircon growth from remaining (near eutectic?) melt. Yb/Gd (steepness of the HREE pattern) is an excellent monitor of fractionation, particularly at lower T (below 750 C) where the ratio increases rapidly. This trend may result from co- fractionation of accessory minerals and/or be driven by the thermodynamics of crystal growth, and/or may involve other factors and processes as yet poorly understood. Magmatic zircons commonly have a negative Eu anomaly of about 0.5 or lower which may change little or become more pronounced with falling T; anomalies probably reflect feldspar fractionation rather than magmatic oxidation conditions. Zircons typically have positive Ce anomalies that rise as T falls and Hf increases. This reflects either fractionation of minerals that incorporate Ce+3 but little Ce+4, or oxidation. U and Th concentrations are typically highest in low-T zones but often show very irregular patterns with T and Hf. Molar ratios of total 3+ ions over P are mostly 1-5, suggesting charge compensation other than the xenotime substitution (Mazdab and Wooden 2006). Random analyses of zircons for T and composition are of limited use given wide variation within single zircons. Process interpretations should be based on trends observed in multiple zircons from individual samples, as many samples have characteristics distinct from general trends. Hydrothermal zircon (Hoskin 2005) is not unusual as a rim zone and may reflect a fine intergrowth of other minerals (apatite, titanite, oxides) or unusual late stage growth conditions. Ti temperatures from these zones are often unreliable, and all analyses for Ti and trace elements should include screens (i.e. F, Al, Ca, Fe) for Ti-bearing minerals and other accessories. While trace element concentrations of zircons may not be diagnostic of rock types in general, careful analyses as described above provide invaluable information about magmatic and metamorphic processes.
V31F-05
New Approaches to Characterizing Thermal Histories of Silicic Magma Systems
We are evaluating applications of two new experimentally based thermometers, combined with a refined TiO$_{2}$ solubility model, to constraining thermal histories of silicic magma systems. The Zr-in-sphene and Ti- in-quartz (TitaniQ) thermometers are expected to be useful tools in crustal petrology and have certain advantages over other commonly applied thermometers. Perhaps the most notable advantage is the potential for sphene and quartz to record thermal histories because cation diffusivities are quite low (~10$^{-22}$ m$^{2}$/s at $750\deg$C), compared to the rapid reequilibration that Fe-Ti oxides and feldspars experience in response to temperature changes. Another advantage of the sphene and quartz thermometers is that they require the analysis of only a single phase. The TitaniQ thermometer does require knowledge of TiO$_{2}$ activity if rutile is absent, which is often the case in rhyolites, but this can be determined using the TiO$_{2}$ solubility model. We have applied the two thermometers and the TiO$_{2}$ solubility model to rhyolites from six different sources that contain both sphene and quartz and have determined crystallization temperature using two methods. First, we measured Zr concentrations in sphene and directly calculated a temperature from the sphene thermometer. Next, this temperature can be used with the TiO$_{2}$ solubility model to determine TiO$_{2}$ activity in the melt as a function of temperature and melt composition. The TitaniQ temperature can now be determined for the correct TiO$_{2}$ activity. The alternative method is to combine the TitaniQ and TiO$_{2}$ solubility equations to create an equation for temperature that depends only on Ti content of quartz and Ti content and composition of the host glass. There are relatively large errors associated with this approach due to the similarity of the slopes for the equations describing Ti solubility in quartz and in melts. Resulting sphene temperatures all fall in a very restricted range between 735 - $755\deg$C. The range of TitaniQ temperatures (715 - $750\deg$C) based on the first application method generally overlap those of sphene, but are systematically higher (760 - $815\deg$C) using the alternative method. The small variation of the sphene temperatures is an illustration of the precision of the thermometer; the $20\deg$ difference between the minimum and maximum temperatures reflect a 200 ppm difference in Zr concentration. This precision implies that the small variations in temperature evident in sphene are real, and that they all crystallized in a very narrow temperature range.
V31F-06
Rutile Saturation in Magmas Revisited: Siliceous, Alkali-Rich Compositions
A compositional feature common to subduction-related lavas and the bulk continental crust is depletion of the high-field-strength elements (e.g., Zr, Hf, Nb, Ta) relative to mid-ocean ridge basalts. Because the high-field- strength elements are strongly compatible in rutile, it has been posited that their depletion in island arc basalts is due to its presence as a residual phase, either in the mantle wedge or the subducted oceanic crust. Ryerson and Watson (1987) experimentally investigated rutile saturation in silicate melts and showed that it is unlikely to be a residual phase in the mantle wedge because the solubility of TiO$_{2}$ in basalt at near-solidus conditions is ~7 to 9 wt%. However, they left open the possibility that the wedge is metasomatized by low-degree partial melts of rutile-bearing eclogite derived from the subducted slab. Here we present results from new experiments that investigate rutile saturation in siliceous, alkali-rich melts compositionally similar to low-degree partial melts of eclogite. Results from these experiments demonstrate that the pressure dependence of rutile saturation is strong for these melt compostions, most likely due to a decreased coordination state for Ti$^{4+}$ in the melt. Rutile saturation along the H$_{2}$O-saturated solidus of eclogite requires less than 100 ppm TiO$_{2}$ in the melt at upper mantle conditions. Experiments were carried out on two SiO$_{2}$-Al$_{2}$O$_{3}$-MgO-CaO-Na$_{2}$O-K$_{2}$O base melt compositions (rhyodacite; basalt) at temperatures of 1150-1450$\deg$C and pressures of 1 bar to 35 kbar. Rutile saturation was achieved by addition of 10-40 wt% TiO$_{2}$. Low-pressure experiments were carried out in a vertical gas-mixing furnace. High pressure experiments were carried out using a solid-medium piston-cylinder device. The major element composition of glass and rutile were determined by electron microprobe. Molar rutile-melt partition coefficients from 21 rhyodacite experiments were fit to a thermodynamically derived functional form to allow extrapolation to pressure-temperature conditions along the H$_{2}$O-saturated basalt solidus of Lambert and Wyllie (1972). Temperature dependence was fit to an Arrhenius-type relationship using 1 bar experiments. Pressure dependence was fit using a polynomial in pressure and temperature for the change in molar volume. The resulting equation indicates that rutile saturation at 1 bar and 1075$\deg$C requires ~2 wt% TiO$_{2}$, but that the TiO$_{2}$ content of rutile-saturated melts decreases rapidly along the H$_{2}$O- saturated basalt solidus to a minimum of ~13 ppm at 15 kbar and 640$\deg$C. At 30 kbar and 750$\deg$C the concentration of TiO$_{2}$ in a rutile-saturated siliceous melt is ~20 ppm. Dissolved H$_{2}$O will increase these concentrations somewhat, but this effect appears to be weak; i.e., the presence of ~10 wt% H$_{2}$O increases the concentration of TiO$_{2}$ in the melt by less than a factor of 2 at 10 kbar and 1250$\deg$C. Therefore, partial melts of rutile-bearing eclogite are a viable metasomatizing agent for the source region of island arc basalts. References: Lambert, I. B., and Wyllie, P. J. (1972) {\it J Geol} 80, 693-708; Ryerson, F. J. and Watson, E. B. (1987) {\it Earth Planet Sci Lett} 86, 225-239.
V31F-07
A Tale of Two Plutons: Using Monazite to Reconstruct the Fluid History of Contact Metamorphic Aureoles
The rare earth phosphate monazite can be a powerful tool for characterizing the timing and spatial extent of fluid infiltration during contact metamorphism. We used an ion microprobe to investigate how wallrock monazite responded to the intrusion of two different granitic plutons, the Cretaceous Birch Creek Pluton (BCP) in the White Mountains of eastern California and the Miocene Searchlight pluton in southern Nevada. The contact metamorphic aureoles of both plutons contain monazite and display evidence of alteration by acidic magmatic fluids (sericitization). Contact metamorphism occurred at conditions under which monazite in granitic systems has been shown to be susceptible to hydrothermal alteration: mildly acidic fluids at temperatures of ~250-400C and pressures of ~0.15-0.4 GPa. Monazite from the hydrothermal aureole of the BCP records the infiltration of magmatic fluids into the Early Cambrian Deep Spring Formation (DSF) containing metaquartzites and metasandstones. Monazites in the DSF < 0.6 km from the contact show patchy zoning and have Th-Pb ages and oxygen isotope compositions similar to monazites in the Birch Creek granite but different from monazites > 0.6 km from the contact, suggesting that they dissolved and reprecipitated in infiltrating magmatic fluids. In contrast, the stable isotope compositions of monazites and host rocks (Proterozoic gneisses and the Cretaceous Ireteba granite) on the flanks of the Searchlight pluton do not show evidence of hydrothermal alteration, even though many of the monazites display patchy zoning and ages corresponding to the Searchlight intrusion. The Searchlight gold-silver-copper mining district is primarily located in the roof of the Searchlight intrusion, which contains rocks that were intensely hydrothermally altered but contain no monazite. Searchlight magmatic fluids were strongly focused into the roof zone, with little or no fluid escaping out the sides of the Searchlight pluton during crystallization. Monazite has proven useful for characterizing the where and when of magmatic fluid infiltration.