Biotite/melt trace-element, lithium, and F-OH partitioning in silicate magmas

Biotite is a key hydrous silicate mineral in evolved magmatic systems, but its control on the behaviour of minor- and trace-elements, in particular Li, Nb, F and the REE is not well understood. Here, we quantify that control in sodic (per)alkaline H2O-saturated magmas with variable F-content through crystallisation experiments at 650–800◦C and 200 MPa total pressure, at log f O2 ≈ FMQ +1. Biotite-glass pairs from tephriphonolite to phonolite fall deposits from Tenerife, Canary Islands, and a broad compilation from literature, complement our experimental data set. The new biotite-melt KdF/OH exchange coefficients are 2.9–47.0, typically 6.5–12.5, with minima for Al-rich, Mg-poor biotite. Nernst partition coefficients (D) for lithium are 0.24–32.8 with minima in F-poor biotite formed at high temperatures from peraluminous silicate melt. D values for the large alkali metal ions Na–Cs define Onuma parabolae consistent with their incorporation on the biotite XIIA-site. Niobium partition coefficients are 0.1–1.2, and highest in Ti-rich biotite. The REE and actinides have D values less than 0.01 as their ionic radii fall between the size of the XIIA-site and VIM-sites of biotite. Our data, alongside a literature compilation, constrain empirical models that: (1) describe the exchange of F and OH between the silicate melt and the biotite W-site; (2) predict the partitioning of 1+ cations Li–Cs between silicate melt and the biotite A- and M-sites; (3) predict DNb values. The models use the major-element composition of biotite and silicate melt, pressure and temperature as input. Models are calibrated for use over a wide range of pressure, temperature and bulk composition (P-T-X) and can be used to interrogate biotite from natural systems to determine the composition of their source melt, or to forward model the trace-element evolution of mafic to evolved peralkaline syenite or peraluminous granite systems at upper-mantle to crustal pressures.