GSA Annual Meeting in Seattle, Washington, USA - 2017

Paper No. 47-9
Presentation Time: 4:55 PM


SHEA, Thomas, Geology & Geophysics, University of Hawaii at Manoa, 1680 East-west rd. POST 614B, Honolulu, HI 96822,

Olivine is the most commonly used mineral to disentangle the complex interplay between differentiation and mixing history of basaltic magmas from their source region to the surface. Based on morphological arguments and trace element (e.g. P, Al) distribution patterns, recent studies have suggested that olivine phenocrysts frequently preserve the blueprint of a rapid growth history. Comparatively, elements that diffuse relatively fast (Fe-Mg, Ni, Ca) have been inferred to mostly record subsequent re-equilibration of the crystal cargo during magma mixing via diffusion. These views of growth and zoning could fundamentally change the way petrologists utilize chemical information stored in olivine to retrieve source and timescale information.

Experimental studies have so far generally focused on olivine crystallization in melts with few components (e.g. CMAS), and chemical diffusion in olivine without surrounding melt. Series of 1-atm crystallization experiments involving high-MgO (~11.4 wt.%) Kilauea basalt were performed to investigate the development of chemical zoning during initial growth, and subsequent modifications by diffusion under isothermal conditions. Low to moderate degrees of undercooling (ΔT=10-60°C) were applied by cooling the runs rapidly from near-liquidus conditions to their final temperature. Results show that little to no zoning occurs in most major and minor elements (Fe-Mg, Ni, Ca, Cr, Mn) during rapid growth. In contrast, nominally incompatible trace elements like Al or P show striking variations in composition. Phosphorus appears to substitute for Si, and yields similar ‘apparent’ partition coefficients (~0.8). Diffusion in most major and minor elements is apparently much faster than predicted by relationships known from melt-free experiments, raising questions about the timescales that we can infer from diffusion modeling in natural crystals.