Li–Na interdiffusion and diffusion-driven lithium isotope fractionation in pegmatitic melts

In this study, we investigate the diffusion of Li and its stable isotopes (6Li and 7Li) in flux-rich (1.8 % Li2O, 2.6 % B2O3, 2.3 % P2O5 and 3 % F) pegmatitic melts in order to contribute to the understanding of Li enrichment in such systems. Two glasses were synthesized with a model pegmatitic composition, one of which is highly enriched in Li (> 1 wt %, PEG2-blue) and the other one essentially Li-free (PEG2-Li-free). Diffusion couple experiments were performed to determine the chemical diffusivity of Li in dry pegmatitic melts. Experiments were conducted using rapid-heat and rapid-quench cold-seal pressure vessels in a temperature range of 650–940 ∘C at 100 MPa with Ar as the pressure medium. We observed rapidly formed diffusion profiles, driven by an interdiffusive exchange of the monovalent alkalis Li and Na, while the other elements are immobile on the timescale of experiments (1–30 min). From these experiments, activation energies for Li–Na interdiffusion were determined as 99 ± 7 kJ mol−1 with a pre-exponential factor of log D0 = −5.05 ± 0.33 (D0 in m2 s−1). Li and Na partitioning between the stronger depolymerized PEG2-blue and the less depolymerized PEG2-Li-free leads to a concentration jump at the interface; i.e. Na is enriched in the more depolymerized PEG2-blue. Li–Na interdiffusion coefficients in the studied melt composition are in a similar range as Li and Na tracer diffusivities in other dry aluminosilicate melts, confirming little to no effect of aluminosilicate melt composition on Li diffusivity. Thus, added fluxes do not enhance the Li diffusivity in the same way as observed for H2O (Holycross et al., 2018; Spallanzani et al., 2022). Using melt viscosity as a proxy for the polymerization of the melt shows that water has a stronger potential to depolymerize a melt compared to other fluxing elements. Faster diffusion of 6Li compared to 7Li leads to a strong Li isotope fractionation along the diffusion profile, resulting in δ7Li as low as −80 ‰ relative to the diffusion-unaffected regions. This diffusive isotope fractionation can be quantified with an empirical isotope fractionation factor (β) of 0.20 ± 0.04, similar to previously observed β values for Li diffusion in melts. This suggests in accordance with previously published data that a β value of ca. 0.2 seems to be universally applicable to diffusive Li isotope fractionation in aluminosilicate melts.