Abstract
Nickel is a strongly compatible element in olivine, and thus fractional crystallization of olivine typically results in a concave-up trend on a Fo–Ni diagram. ‘Ni-enriched’ olivine compositions are considered those that fall above such a crystallization trend. To explain Ni-enriched olivine crystals, we develop a set of theoretical and computational models to describe how primitive olivine phenocrysts from a parent (high-Mg, high-Ni) basalt re-equilibrate with an evolved (low-Mg, low-Ni) melt through diffusion. These models describe the progressive loss of Fo and Ni in olivine cores during protracted diffusion for various crystal shapes and different relative diffusivities for Ni and Fe–Mg. In the case when the diffusivity of Ni is lower than that for Fe–Mg interdiffusion, then olivine phenocrysts affected by protracted diffusion form a concave-down trend that contrasts with the concave-up crystallization trend. Models for different simple geometries show that the concavity of the diffusion trend does not depend on the size of the crystals and only weakly depends on their shape. We also find that the effect of diffusion anisotropy on trend concavity is of the same magnitude as the effect of crystal shape. Thus, both diffusion anisotropy and crystal shape do not significantly change the concave-down diffusion trend. Three-dimensional numerical diffusion models using a range of more complex, realistic olivine morphologies with anisotropy corroborate this conclusion. Thus, the curvature of the concave-down diffusion trend is mainly determined by the ratio of Ni and Fe–Mg diffusion coefficients. The initial and final points of the diffusion trend are in turn determined by the compositional contrast between mafic and more evolved melts that have mixed to cause disequilibrium between olivine cores and surrounding melt. We present several examples of measurements on olivine from arc basalts from Kamchatka, and published olivine datasets from mafic magmas from non-subduction settings (lamproites and kimberlites) that are consistent with diffusion-controlled Fo–Ni behaviour. In each case the ratio of Ni and Fe–Mg diffusion coefficients is indicated to be <1. These examples show that crystallization and diffusion can be distinguished by concave-up and concave-down trends in Fo–Ni diagrams.
Brain water diffusion coefficients and diffusion anisotropy in non-demented patients with diffuse leuko-araiosis.
