IMPROVED PARTITION COEFFICIENTS FOR ZIRCON/MELT FROM IN SITU ANALYSIS OF GLASS AND CRYSTAL RIMS: COHERENT BEHAVIOR, STRONG TEMPERATURE DEPENDENCE

Zircon has proven to be an invaluable repository of information about past environments in which it grew, commonly retaining information about settings and times for which there is little or no other record (e.g., the Hadean; the deep crust from which inherited zircons were extracted). Given a firm knowledge of zircon/melt Kds, zircon crystals have great potential for characterizing evolving compositions and conditions in magmas from which they grew, but reliable Kds have proved elusive (e.g. Hanchar & Van Westrenen 2007; Luo & Ayers 2009). Experiments are commonly conducted with materials that don’t reflect natural compositions and/or durations are too short for growth of large, fully equilibrated crystals. Most Kds determined from natural materials are hampered by the fact that melt compositions from which zircons grew are unknown. Glass-hosted zircon rims can provide measurable equilibrium crystal and melt compositions.

We calculated Kds from in situ zircon rim (SHRIMP) and host glass (LA-ICPMS) analyses in 16 rhyolite and dacite samples from a variety of settings (hot spot/spreading center, Iceland [tholeiites plus an anomalous calc-alkaline lava]; incipient continental extension, CA-NV-AZ [high-K calc-alkaline]; continental arc, Mt St Helens). Most rims were analyzed in cross section, but we also analyzed some crystal face surfaces. All glasses are rhyolite (SiO₂ 72-79 wt%), with Zr concentrations and estimated zircon saturation Ts (T_Zr; Boehnke et al 2013) ranging from 60-800 ppm and 660-940°C.

Kds range over more than an order of magnitude (e.g. La 0.0002-0.004; Lu 30-700; Th 1.5-50; U 7-150; Nb 0.1-1), but are very coherent. Notably, log_REE Kds vs ionic radii of all samples plot as parallel parabolic curves that are consistent with theoretical considerations (Blundy & Wood 2003). Kd magnitudes correlate very well with T indicators (Zr in glass, T_Zr, and Ti in zircon), but not with major element glass compositions. (Zr/Hf)_zircon/(Zr/Hf)_glass ranges from 1.2 to 1.8.

We suggest that coupling crystal rim and host glass analyses is the most effective way to determine reliable zircon Kds, and mineral Kds in general (Padilla & Gualda in rev). Furthermore, for quantitative applications it is critical to constrain conditions of crystallization: in the case of zircon, Kds differ by >10x as a function of T.