[SVC41-01] Degassing processes of silicic magma estimated from multivolatile analysis of obsidian pyroclast
★Invited Papers
Keywords:magma, degassing, volatile
The degassing of silicic magma during ascent is important in the control of final eruption style. However, what actually occurs in ascending magma has not been well constrained because of lack of clear evidence. One possible way to investigate this is to analyse obsidian pyroclasts, which are ejected together with pumice and ash during an explosive eruption, and have been considered as essential magma fragments that were quenched at various depths of the conduit (e.g., Newman et al., 1988). In this study, I analysed volatile contents of rhyolitic obsidian pyroclasts from AD 886 eruption of Mukaiyama volcano and discuss the degassing process.
The obsidian pyroclast from Mukaiyama volcano was dense glassy fragment containing a small number of bubbles (<0.1 vol%). The major composition of obsidian pyroclast was identical to that of pumice. FTIR analysis revealed that the H2O content of glass was 1.1–2.4 wt% and CO2 content was <2 ppm. Assuming gas-melt equilibrium, I estimated the gas-saturation pressure to be 9–36 MPa by using solubility law of Liu et al. (2005). The initial H2O content of the Mukaiyama magma was estimated to be ~4 wt% based on phenocryst assemblage. Therefore, the obsidian pyroclasts are considered to have experienced open-system gas loss, because they are almost bubble-free but have only a part of initial water.
The Cl-content mapping analysis revealed that all samples have high-Cl stripes running in one direction. The width of each stripe was ~10 microns. Bubbles were often aligned in the same direction, and they were connected through the high-Cl stripe. The Cl content was elevated near the bubble interface. These observations indicate that bubbles are resorbing (Yoshimura et al., in prep.), and that the high-Cl stripes are the welded interface of the resorbing bubbles. Number density of the stripes, which may be proxy for bubble number density before resorption, was higher for lower-H2O clasts (150 stripes/mm for 1.1 wt%) and lower for higher-H2O samples (25 stripes/mm for 2.4 wt%). This indicates that vesicularity of the obsidian pyroclast had been pressure-dependent before bubbles resorbed. Timescale of bubble resorption was estimated to be a few minutes based on Cl diffusion profiles across the high-Cl stripes. Such rapid bubble resorption may have been caused by repressurisation. Another possibility that bubbles resorbed because of water depletion, which was caused by continuing open-system gas loss (Yoshimura et al., in prep.), is unlikely, because this process occurs only locally near open channels. The reason why repressurisation occurred in ascending magma is currently under investigation.
The obsidian pyroclast from Mukaiyama volcano was dense glassy fragment containing a small number of bubbles (<0.1 vol%). The major composition of obsidian pyroclast was identical to that of pumice. FTIR analysis revealed that the H2O content of glass was 1.1–2.4 wt% and CO2 content was <2 ppm. Assuming gas-melt equilibrium, I estimated the gas-saturation pressure to be 9–36 MPa by using solubility law of Liu et al. (2005). The initial H2O content of the Mukaiyama magma was estimated to be ~4 wt% based on phenocryst assemblage. Therefore, the obsidian pyroclasts are considered to have experienced open-system gas loss, because they are almost bubble-free but have only a part of initial water.
The Cl-content mapping analysis revealed that all samples have high-Cl stripes running in one direction. The width of each stripe was ~10 microns. Bubbles were often aligned in the same direction, and they were connected through the high-Cl stripe. The Cl content was elevated near the bubble interface. These observations indicate that bubbles are resorbing (Yoshimura et al., in prep.), and that the high-Cl stripes are the welded interface of the resorbing bubbles. Number density of the stripes, which may be proxy for bubble number density before resorption, was higher for lower-H2O clasts (150 stripes/mm for 1.1 wt%) and lower for higher-H2O samples (25 stripes/mm for 2.4 wt%). This indicates that vesicularity of the obsidian pyroclast had been pressure-dependent before bubbles resorbed. Timescale of bubble resorption was estimated to be a few minutes based on Cl diffusion profiles across the high-Cl stripes. Such rapid bubble resorption may have been caused by repressurisation. Another possibility that bubbles resorbed because of water depletion, which was caused by continuing open-system gas loss (Yoshimura et al., in prep.), is unlikely, because this process occurs only locally near open channels. The reason why repressurisation occurred in ascending magma is currently under investigation.