Zircon saturation re-revisited

Improvements in experimental, analytical and computational methodologies together with published studies yielding seemingly contradictory results prompted us to return to the determination of zircon stability in the range of felsic to intermediate melts expected in continental environments. We (re-)analyzed both the run products from the zircon crystallization study of Watson and Harrison (1983) and a new style of zircon dissolution experiments (up to 25 kbar) using a large radius ion microprobe to constrain a refined zircon solubility model. The new data yield broadly similar patterns as before when arrayed for temperature and confirm that the parameter M [=(Na + K + 2Ca)/(Al·Si)] is an appropriate compositional proxy for the chemical interactions through which zircon is dissolved. We used a Bayesian approach to optimize the calculation of the coefficients in the zircon solution model, which is given by: ln D Zr = 10108 ± 32 / T K − 1.16 ± 0.15 M − 1 − 1.48 ± 0.09 where DZr is the distribution coefficient of Zr between zircon and melt and the errors are at one sigma. Sensitivity tests indicate that temperature and composition are the two dominant controls on zircon solubility in crustal melts with no observable effects due to pressure (up to 25 kbar) or variable water content. Comparison of the down-temperature extrapolation with natural examples confirms the validity of the model at ca. 700 °C.