Inferring Magma Ascent Times and Conduit Processes for Rhyolitic Eruptions Using Diffusive Loss of Hydrogen From Melt Inclusions

Abstract:

Magma volatile contents and ascent rates primarily determine whether rhyolites erupt explosively or effusively. Techniques for assessing ascent rates include experimental calibrations of microlite growth and hornblende breakdown, and diffusion of volatiles in reentrants (embayments). We present a new method based on diffusive loss of H from quartz-hosted, rhyolitic melt inclusions (MIs) in the Huckleberry Ridge Tuff (HRT). The initial HRT fall deposits show evidence for syn-eruptive reworking, reflecting episodicity during the opening explosive phases. Quartz-hosted MIs from 9 levels in the fall deposits have U, Cl, and B contents that form correlated arrays versus Rb, indicative of the effects of crystallization differentiation. However, H and Li, which diffuse much faster in quartz, plot as scattered clusters, suggesting that diffusive changes occurred post-entrapment. In each fall horizon sampled, wide ranges of H₂O (~1.0-4.5 wt%) but restricted ranges in CO₂ occur in sealed MIs. In contrast to the wide H₂O ranges, incompatible elements in the MIs from each sample vary by less than a factor of 2, consistent with diffusive loss of H from MIs. We infer that the highest H₂O values reflect the magma H₂O content at the pre-eruptive storage depth, and lower values reflect variable diffusive losses, implying that at a given stratigraphic horizon, co-deposited crystals experienced different ascent histories. Timescales for H₂O loss from a diffusion model¹ and experimentally calibrated H diffusivities² yield decompression times of <12 hours to 5 days, with the lowest H₂O values (~1 wt%) requiring ~2 weeks. HRT initial magma ascent was complex in that single fall layers contain crystals from both rapidly ascending, deeply derived magma and magma that had decompressed to varying degrees over varying times within the conduit system. The timescales for individual stratigraphic horizons are of the same magnitude as the estimates based on reworking in individual layers. The field evidence for multiple time breaks and the MI evidence for hours to weeks ascent timescales of co-erupted crystals in single layers requires that magma ascended episodically over the weeks-months timescale represented by the early HRT fall deposit.

¹Cottrell, E. et al. (2002) G³ 3:1026; ²Severs, M.J. et al. (2007) Chem Geol 237:358.