Phosphorus Coupling Obfuscates Lithium Geospeedometry in Olivine

Abstract

Lithium zoning in zircon and olivine provides a powerful record of crystal histories and magmatic processes in volcanic systems. Characterizing Li behavior in olivine is important because it is one of the few elements in basaltic systems that track short-duration magmatic processes (e.g., less than a few days). However, the potential for trace element coupling and the competing effects of crystal growth and subsequent diffusion obfuscate interpretations of Li zoning patterns and their use for geospeedometry. Here, we use diverse analytical techniques (EPMA, LA-ICPMS, nanoSIMS) to untangle records of growth and diffusion in carefully oriented olivine crystals from Kīlauea (Hawai‘i). Li, P, and Al are targeted because they show correlations despite contrasting behaviors during growth and diffusion. Lithium zoning exhibits two styles: (1) non-coupled, wherein Li (1–3 ppm) shows diffuse zoning over 10s–100s of μm between the crystal core and rim with low (10s of ppm) and homogeneous P, and (2) coupled, wherein sub-ppm Li enrichments <40 μm wide are correlated with P-rich zones (100s of ppm) that preserve a blueprint of crystal growth. Non-coupled Li zoning modeled for diffusive re-equilibration yields hours–days timescales, reflecting magma-mixing events that primed the system for eruption. Coupled Li peaks consistently occur where analyses traverse the dendritic framework of P zoning that reflects rapid crystal growth. This correlation results from Li+ ions partially satisfying charge balancing requirements of P5+ ions in the olivine lattice. Aluminum balances most of the P budget in olivine but has non-systematic correlations with Li peaks. Transects can exhibit both non-coupled and coupled Li behaviors, preserving histories of both crystal growth and diffusive re-equilibration. Modeling a near-rim Li enrichment peak with the Li diffusion coefficient yields a timescale of <3 min, too short to represent the time elapsed following crystallization. If the near-rim Li and P peaks are modeled using the diffusion coefficient for P, congruent timescales of 5–11 days suggest that they have the potential to record magmatic processes. Several caveats such as poorly constrained initial conditions and analytical challenges limit the practicality of enrichment peak geospeedometry. Only broad zoning of non-coupled Li should be modeled for timescales of magmatic processes.