Experimental Constraints on the H2O-Saturated Plagioclase Liquidus and the Storage Depth of the Izu-Oshima 1986B Basaltic Andesite Melt

Abstract

High-temperature melting and crystallization experiments were carried out at pressures from 1 atm to 196 MPa and under H2O-saturated conditions on the basaltic andesite melt of the Izu-Oshima 1986B eruption (i.e., the BM melt), using a 1-atmosphere fO2-controlled furnace and an internally heated pressure vessel. These data were used to constrain the H2O-saturated plagioclase liquidus (HSPL) of the melt. The fO2 conditions were controlled by a mixed H2-CO2 gas at the Ni-NiO (NNO) buffer for the 1 atm experiments, but were not controlled for the high-pressure experiments. Plagioclase is the liquidus phase at 1 atm, whereas early saturation of Fe-Ti oxide above the plagioclase liquidus occurred in the high-pressure experiments due to the elevated fO2 conditions. The HSPL temperature decreases from 1172 ± 8°C to 1030 ± 20°C as the pressure increases from 1 atm to 196 MPa. A combination of previously proposed models for the plagioclase liquidus and melt H2O-solubility can predict the experimentally determined HSPL temperatures, even if oxidation-induced magnetite crystallization occurs. Using these models and the previously reported pre-eruptive temperature of ∼1100 ± 30°C, we estimate the pre-eruptive pressure conditions of the BM melt to be 42-32+48 MPa, which corresponds to depths of 1.9-1.4+1.9 km. The estimated depth is consistent with that of the shallow active dikes previously identified from geophysical studies, suggesting that the BM melt was derived from a small, shallow magma chamber formed in the shallow dike region.