'Cryptic' concentration dependent diffusion of a trivalent trace element (Sc3+) in olivine: implications for diffusion chronometry

Abstract:

Olivine-based diffusion chronometry often considers only the major and minor elements (i.e. Fe-Mg, Ni). Questions remain regarding the applicability of experimental diffusion coefficients to natural systems, thus the accuracy of resulting timescales.

Trace elements can therefore be very useful – it should be possible to find trace elements whose diffusivity is not affected by parameters that might affect major element diffusion. However, these bring their own complications, which also need to be addressed.

In this study, the diffusivity of scandium (Sc) in pure, synthetic forsterite (Mg₂SiO₄) is considered. Sc source powders were prepared using the sol-gel method from SiC₈H₂₀O₄, Sc₂O₃ and Mg. These powders contained forsterite plus protoenstatite (Mg₂Si₂O₆), plus various Sc contents, from ~10 wt. ppm to several wt. %. The mix with the highest Sc contents also formed thortveitite (Sc₂Si₂O₇). The powders were coupled to forsterite crystals polished on their (001) plane, then annealed at 1400 °C for 550 hours. They were then sectioned perpendicular to (001), and profiles ||[001] were measured by scanning laser ablation inductively coupled plasma mass spectrometry.

Whilst the majority of profiles conform broadly to the error function, suggesting concentration-independent diffusion, there is a positive correlation between the interface Sc concentration and the profile lengths, which range from 200-800 µm. Close inspection of the profiles show small negative deviations from the error function at low Sc contents (<~50 wt. ppm). Profiles with the lowest interface concentrations (>20 wt. ppm) show Sc profiles with a near flat plateau close to the interface, then a drop to background concentrations. These profiles are mirrored by Al out-diffusion profiles – Al is present at ~20 wt. ppm in the starting material.

Taken together, these results suggest 'cryptic' concentration dependence, i.e. concentration-dependent diffusion that is not visible by looking at the profile shapes alone. This is likely due to two processes, one in which low Sc contents are associated with low vacancy contents, given that two Sc are charge-balanced by an M-site vacancy, and the other where small amounts of Sc move into coordination with tetrahedral Al³⁺. Such behaviour is likely possible in natural systems, which might complicate the use of trivalent cations (e.g. Cr³⁺, Y³⁺, REE³⁺) for diffusion chronometry.