Rates of dehydration of olivines from San Carlos and Kilauea Iki

Abstract The rate at which H + moves into, through, and out of olivine is of great interest for understanding the deep water cycle, assessing the fidelity of olivine-hosted melt inclusions as recorders of pre-eruptive melt compositions, and estimating the decompression rate of magma during volcanic eruptions. Here we conduct a series of experiments and observations on natural Fe-bearing olivine to better understand and quantify this H + mobility, with a particular focus on the rate at which H + diffuses out of olivine. We have carried out sequential dehydration experiments on single crystals of natural, Fe-bearing olivine polished and oriented along three crystallographic orientations. These efforts include four separate parts: (1) the partial hydration of polished blocks of xenolithic San Carlos olivine at 800 °C and 1000 °C, 1 GPa, and oxygen fugacity (fO 2 ) at the Ni-NiO buffer (NNO); (2) sequential dehydration of a homogeneous block of San Carlos olivine at 800 °C, 1-atm, and fO 2 of NNO-2.6; (3) sequential dehydration of a polished block of an olivine phenocryst from Kilauea Iki at 800 and 1000 °C, 1 atm, and fO 2 held at NNO-2.6 until the final step, which was conducted at NNO + 1.9; and (4) the measurement by secondary ion mass spectrometry of hydrogen zonation profiles in an olivine phenocryst from Kilauea Iki coupled with Monte Carlo modeling to determine total H diffusivities in a natural setting. Hydrogen zonation profiles were measured in all three crystallographic directions a , b , and c through the uncut blocks before and after each time step for all experiments by Fourier transform infrared spectroscopy using light polarized || a and modeled in 3 dimensions to determine both total and peak-specific dehydration, producing over 50 sets of 3 apparent diffusivities. The results show that the apparent diffusivity of total H + varies both over time and for olivines with different defect populations, and thus H + does not diffuse at inherently site-specific rates. These variations are due to rapid reactions that may occur between sites as a function of changing defect concentrations. Our experiments reveal three new phenomena: (1) H + that is associated with the Ti-clinohumite defect, [Ti-2H], can enter and exit olivine at the fastest rate measured. (2) The rate of H + loss from specific infrared absorption peaks, and in total, may change with progressive dehydration due to the buildup of Fe 3+ and redistribution of defects, particularly the rehydration of [Ti-2H]. (3) The following Arrhenius laws appear to apply to total H + diffusion out of most natural olivines that have been studied experimentally and naturally during dehydration: D a  = 10 −5.4 exp(−130/RT); D b  = 10 −6.9 exp(−130/RT); and D c  = 10 −6.6 exp(−130/RT), where R is the gas constant 8.314 J/mol K; T is the temperature, the units of the diffusivities ( D ) are in m 2 /s; and the activation energy is 130 kJ/mol. This rate is slower than proton-polaron redox rate but faster than proton-metal vacancy diffusion and appears to characterize olivine of Fo ∼90 composition in a wide range of conditions, from self-diffusion, to melt inclusion re-equilibration, to mantle xenolith and volcanic phenocryst zonation during ascent.