A new high-pressure experimental apparatus to study magmatic processes at precisely controlled redox conditions

Abstract

Abstract Oxygen fugacity (fO2) is typically controlled in high P-T experiments by using solid-state redox buffer assemblages. However, these are restricted to impose discrete fO2 values, often with significant gaps between neighboring assemblages. Semi-permeable hydrogen membranes (Shaw 1963) are often used in internally heated pressure vessels for more flexible fO2 control in hydrous experiments; however, their implementation in more widely available externally heated pressure vessels has not yet gained space. We propose a prototype molybdenum-hafnium carbide (MHC) pressure vessel apparatus that simultaneously allows rapid quenching and flexible, precise, and accurate redox control via a custom-designed hydrogen membrane. Test runs with two membranes at a time, one imposing and another one monitoring fH2, demonstrated that 95% of the imposed hydrogen pressure was attained inside the pressure vessel within 2 h at 800–1000 °C, after which a steady state equilibrium was established. Furthermore, experiments comparing redox-dependent Cu solubility in silicate melts at fO2 imposed by the fayalite-magnetite-quartz, Re-ReO2, and MnO-Mn2O3 buffers and identical target fO2 imposed by the hydrogen membrane confirmed consistency between the two methods within 0.25 log units fO2 deviation at T = 900 °C and P = 2000 bar. This powerful yet cost-effective and low-maintenance apparatus may open up new pathways for studying redox reactions in hydrous magmas and magmatic fluids. As a proof of concept, we conducted near-liquidus phase-equilibrium experiments with H2O-saturated calc-alkaline basalt and shoshonite melt compositions at five different fO2 values equally distributed between half log unit below the Ni-NiO buffer (NNO-0.5) and NNO+2.7. Most experiments crystallized olivine, clinopyroxene, and Ti-magnetite. The Mg# of the olivine increased with fO2, and the Fe3+/Fetotal ratios in the silicate melt were determined based on Fe(II)-Mg exchange between olivine and melt. The Fe3+/Fetotal ratios in the shoshonite melt were systematically higher by about 0.06 ± 0.01 than those in the calc alkaline basalt melt at identical fO2. The values determined for the basaltic melt were consistent within 1σ error (<0.033 deviation) from those predicted by the equation of Kress and Carmichael (1991). The Fe-Ti exchange coefficient between magnetite and silicate melt increases from 1.73 ± 0.19 (1σ) at NNO –0.5 to +7.12 ± 0.36 at NNO+2.7 for shoshonite and has a similar range for the calc-alkaline basalt.