The presence of silicate melt may enhance rates of cation diffusion in olivine

Abstract

Olivine is commonly used as a ‘crystal clock’ to extract timescales relevant to pre-eruptive perturbations within mafic magmatic systems. Diffusion chronometry applications require accurate calibrations for the rates at which Fe-Mg or other commonly measured elements like Ni, Mn, and Ca diffuse through the crystal lattice. In the past, these rates have been mainly characterized using solid-solid diffusion couple experiments involving olivine single crystals, thin films, or powder sources. Despite the presence of melt surrounding olivine in natural magmatic systems, very few experiments involving magma have been performed, largely because controlling interface reactions is difficult. For this study, we carried out olivine-melt diffusion experiments as a test of the diffusion chronometry method, and to determine whether the presence of melt influences the calculated timescales. To approximate a natural system, we incorporated small natural Kīlauea and San Carlos olivine seeds within a natural Kīlauea basalt and tracked diffusive re-equilibration through time. To better control interface reactions, after some equilibration period at an initial superliquidus temperature of 1290°C, the runs were rapidly cooled to form a rim and left to dwell at various final temperatures (1200, 1220, 1240, 1255°C) for 6–84 h. Concentration gradients for Fe-Mg, Mn, Ni, Ca were measured, and the step-wise nature of the core-rim transition was ascertained using slow diffusing elements like P or Al. When these gradients are modeled using published diffusivities, the timescales retrieved are typically 10 times longer than the actual experiment durations. Thus, measured diffusivities are an order of magnitude faster than those previously obtained in olivine-solid source experiments, but they are in excellent agreement with the only two other melt-olivine datasets. We explore reasons for why melt-bearing olivine diffusion experiments tend to yield faster rates. The possible effects of (1) growth during diffusion, (2) diffusion during any initial dissolution step, and (3) extended tube or planar defects at the interface on calculated diffusivities are all considered but found to be inconsequential. Instead, we argue that additional point defects (vacancies) are likely created at the interface by higher concentrations in elements like Al or H in the basalt melt compared to other solid couple diffusant sources. Future applications of diffusion chronometry in olivine may require a complete re-evaluation of published diffusivities using melt-bearing experimental configurations.