ALBERT

Description: We derive a novel method for determining the oxidation state of a magma as zircon crystallized, with a standard error of ±0·6 log unit ƒO2, using ratios of Ce, U, and Ti in zircon, without explicit determination of the ionic charge of any of them, and without independent determination of crystallization temperature or pressure or parental melt composition. It yields results in good agreement with oxybarometry on Fe–Ti oxide phenocrysts and hornblende phenocrysts quenched in eruptive I- and A-type dacites and rhyolites, but our zircon oxybarometer is also applicable to slowly cooled plutonic rocks and applicable to detrital and xenocrystic zircons. Zircon/melt partition coefficients of Ce and U vary oppositely with ƒO2 variation in the silicate melt. The Ce/U ratio in zircon also varies with the Ce/U element ratio in the silicate melt. During mafic-to-felsic magmatic differentiation, Ce and U are incorporated mainly in calcium-dominated lattice sites of clinopyroxene, hornblende, apatite, and occasionally titanite and/or allanite, all of which have a similar degree of preference for Ce over U. We employ the U/Ti ratio in zircon and in silicate melt as a magmatic differentiation index. Convergent- and divergent-plate-margin differentiation series consistently follow the relation log (Ce/U) ≈ –0·5 log (U/Ti) + C' in silicate melts of basaltic to rhyolitic composition. That correlation permits thermodynamic derivation of the oxybarometry relation among those elements in zircon: log fO2(sample)−log fO2(FMQ)≈42n+1log[Ce/(Ui×Ti)z]+C, wherein Ui denotes age-corrected initial U content, FMQ represents the reference buffer fayalite + magnetite + quartz, superscript z denotes zircon, and n varies with the average valence of uranium in the zircon’s parental silicate melt. We empirically calibrate this relation, using 1042 analysed zircons in 85 natural populations having independently constrained log ƒO2 in the range FMQ – 4·9 to FMQ + 2·9, to obtain the equation log fO2(sample)−log fO2(FMQ)=3·998(±0·124) log[Ce/(Ui×Ti)z]+2·284(±0·101) with a correlation coefficient R = 0·963 and standard error of 0·6 log unit ƒO2 in calc-alkalic, tholeiitic, adakitic, and shoshonitic, metaluminous to mildly peraluminous and mildly peralkaline melts in the composition range from kimberlite to rhyolite. Thermodynamic assessment and empirical tests indicate that our formulation is insensitive to varying crystallization temperature and pressure at lithospheric conditions. We present a revised equation for Ti-in-zircon thermometry that accounts appropriately for pressure as well as reduced activity of TiO2 and SiO2 in rutile- and quartz-undersaturated melts. It can be used to retrieve absolute values of ƒO2 from values of ΔFMQ obtained from a zircon analysis.