From Purgatory to Paradise: The Volatile Life of Hawaiian Magma

Abstract:

Variations in radiogenic isotope ratios and magmatic volatile abundances (e.g., CO₂ or H₂O) in Hawaiian lavas reveal key processes within a deep-seated mantle plume (e.g., mantle heterogeneity, source lithology, partial melting, and magma degassing). Shield-stage Hawaiian lavas likely originate from a mixed plume source containing peridotite and recycled oceanic crust (pyroxenite) based on variations of radiogenic isotopes (e.g., ²⁰⁶Pb/²⁰⁴Pb). The mantle source region may also be heterogeneous with respect to volatile contents, yet the link between pre-eruptive volatile budgets and mantle source lithology in the Hawaiian plume is poorly constrained due to shallow magmatic degassing and mixing. Here, we use a novel approach to investigate this link using Os isotopic ratios, and major, trace, and volatile elements in olivines and mineral-hosted melt inclusions (MIs) from 34 samples from Koolau, Mauna Loa, Hualalai, Kilauea, and Loihi. These samples reveal a strong correlation between volatile contents in olivine-hosted MIs and Os isotopes of the same olivines, in which lavas that originated from greater proportions of recycled oceanic crust/pyroxenite (i.e. ‘Loa’ chain volcanoes: Koolau, Mauna Loa, Loihi) have MIs with the lower H₂O, F, and Cl contents than ‘Kea’ chain volcanoes (i.e. Kilauea) that contain greater amounts of peridotite in the source region. No correlation is observed with CO₂ or S. The depletion of fluid-mobile elements (H₂O, F, and Cl) in ‘Loa’ chain volcanoes indicates ancient dehydrated oceanic crust is a plume component that controls much of the compositional variation of Hawaiian Volcanoes. The presence of dehydrated recycled mafic material in the plume source suggests that subduction effectively devolatilizes the mafic part of the oceanic crust. These results are similar to the observed shifts in H₂O/Ce ratios near the Easter and Samoan hotspots [1,2]. Thus, it appears that multiple hotspots may record relative H₂O depletions and possibly other volatiles. [1] Dixon et al. 2002, Nature 420:385-89 [2] Workman et al. 2006, EPSL 241:932-51