A REE-based zircon geothermometer based on improved lattice strain modeling of zircon-melt REE partition coefficients

Several methods for tracing geological processes based on the rare earth element (REE) concentrations in zircon have been proposed in recent years. Application of these methods requires quantitative knowledge of the partition coefficients (D) for the REE between zircon and co-existing phases, which may be melt, fluid, or other minerals. Precisely determining the D_REE of zircon, however, is challenging. This is especially true for the light rare earth elements, which generally occur in very low concentrations in zircon. This difficulty can be overcome by modeling D_REE using a prediction function based on the crystal lattice strain model. The prediction function is defined by three empirically derived lattice strain parameters: the apparent Young's Modulus E of the Zr site in which the REE resides; the ionic radius r₀ of a (fictive) REE optimally fitting into this site; and the maximum partition coefficient D₀ of this optimally fitting REE. In principle, these three parameters can be determined by fitting the prediction function to sets of measured D_REE. The ionic radii of the REE, however, are all larger than r₀ for zircon, commonly resulting in strain model fits yielding physically unrealistic values of r₀, in particular. Here, we use a compilation of zircon-melt REE partition coefficients from the literature to establish a new empirical relationship between r₀ and E, which is used as an additional constraint during lattice strain model curve fitting. Lattice strain parameters for literature D_REE data sets of natural and synthetic zircon samples are calculated using our constraint resulting in a new parametrization of the inverse temperature dependence of D₀. This parametrization enables quantitative estimates of REE-based crystallization temperatures of magmatic zircons, complementing estimates based on Ti-in-zircon geothermometry.