The rate of water loss from olivine-hosted melt inclusions

Diffusive water loss from olivine-hosted melt inclusions has been reported previously. This process must be considered when interpreting melt inclusion data. This study measured the rate of water loss from olivine-hosted melt inclusions during heating-stage experiments to test a previous diffusive reequilibration model and the hydrogen diffusion mechanism that controls the rate. Olivine-hosted melt inclusions were heated to a constant temperature in reduced Ar gas in a heating stage for a few hours, and unpolarized Fourier transform infrared spectra were repeatedly measured through the inclusions. Water loss occurred rapidly in the experiments. Within a few hours, the water absorbance at 3,500 cm(-1) wavenumber decreased by half. The observed water loss rate can be explained by the diffusive reequilibration model and hydrogen diffusion in olivine coupled with metal vacancy. The beginning of water loss was different in the low- and high-temperature experiments. At low temperatures (1,423 and 1,437 K), water loss did not occur in the initial 1 or 2 h. At high temperatures (1,471-1,561 K), water loss began immediately. The initial time period without water loss at low temperatures may be explained by a hydrogen fugacity barrier in the host olivine. At low temperatures, the internal pressure may be lower than the equilibrium pressure of melt inclusion and olivine, causing lower hydrogen fugacity in the melt inclusion than in the olivine, which will delay the water loss from the melt inclusion. The tested model and diffusivity were used to estimate the rate of water loss during homogenization experiments and magma eruption and cooling. For 1-h homogenization experiment, the model shows that large inclusions (50 mu m radius) in large olivines (500 mu m radius) are robust against water loss, while large or small inclusions (50-10 mu m radius) in small olivines (150 mu m radius) may suffer 30-100% water loss. For natural samples, the correlation between water concentration and melt inclusion and olivine sizes may be helpful to infer the initial water concentration, degree of diffusive reequilibration, and magma cooling rate.