REDUCING UNCERTAINTIES IN ZR-IN-RUTILE THERMOMETRY (Invited Presentation)

The Zr-in-rutile thermometer (ZiR) enjoys widespread use, but the effects of different sources of uncertainties in calculated temperatures have not been fully assessed. In addition to calibration uncertainties, variations in Zr concentrations within and among rutile grains raise questions of how to identify composition(s) to use for thermometry. Here we assess temperature uncertainties in the context of new ZiR calibrations (Kohn, 2020; Am Min) and natural Zr variability, and present a method to identify distinct populations of rutile compositions that may be interpreted petrologically and used for more precise ZiR calculations.

New ZiR calibrations have propagated temperature uncertainties of ±20-30 °C (2σ) for an experimental-only dataset calibration, and ±10-15 °C (2σ) for a combined experimental and natural dataset. However, for rocks, how do we choose which rutile composition(s) to use for thermometry? Compositional criteria must be founded on rock-specific petrologic models of rutile stability and compositional equilibration and reequilibration. Nonetheless, many models assume that the maximum Zr concentration likely corresponds most closely with the peak of metamorphism. As discussed by Penniston-Dorland et al. (2018, EPSL), use of either an average of all data or a single highest-Zr measurement will introduce bias. Rather, their “mean-max method” uses either the average of the 4 highest concentrations (if they are each compositionally distinct), or a population of statistically indistinguishable high-Zr concentrations. To help identify single distinct populations from the total distribution of Zr concentrations in rutile, we have applied the generic IsoPlotR function Peakfit (Vermeesch, 2018, Geosci Front; analogous to the Unmix routine in IsoPlot, Ludwig, 2012, Berkeley Geochron) to identify distinct sub-populations, their mean values, and their corresponding standard errors. Applying Peakfit to rutile data can readily resolve a set of analytically indistinguishable high-Zr concentrations, in some cases yielding precisions as small as ±5 °C. Peakfit can be applied to any 1-D dataset, e.g., to identify maximum inclusion pressures from measurements of Raman spectra to calculate maximum entrapment pressures, minimum zircon U-Pb ages to identify likely eruption ages, etc.