Gradual oxidation of the mantle through water accretion . 28

For this project, the possibility that the accretion of H₂O-rich material led to the gradual oxidation of the mantle was explored. This would have only

been possible provided that this process could have continued once the oxygen fugacity had risen above the level where metallic iron would be stable, implying that the core-forming material would have been an FeS sulfide. There is a fair amount of evidence that as the final magma ocean cool and crystallised, FeS may have exsolved (the “Hadean matte”, e.g., O’Neill, 1991; Rubie et al., 2016) and been removed to the core.

In order to examine whether a Hadean matte could have removed hydrogen to the core, experiments were performed to equilibrate a mid ocean ridge basaltic melt with an FeS composition at 3 GPa, in multianvil presses at the Bay-erisches Geoinstitut. The experiments contained 0.2 - 0.5 wt % H₂O, and graphite capsules were employed inside outer welded AuPd capsules to min-imise water loss during the experiments. Experiments were run at 3 GPa to ensure that a silicate glass was produced upon quenching of the experi-ments. The resulting run products comprise a large pool of silicate glass and quenched FeS melt. The oxygen fugacity in the experiments was adjusted by changing the bulk FeO content of the silicates and by adding varying amounts of Fe to the experiments.

Hydrogen concentration of the run products was determined with elastic re-coil detection analysis (ERDA), performed with a nuclear microprobe at CEA Saclay, France. Hydrogen partition coefficients for the experimental run prod-ucts were then calculated and the resulting data were used to assess the potential oxidation of the mantle through interaction of H₂O and FeS in a pri-mordial magma ocean. In chapter 5 I present results from hydrogen molten silicate/ liquid sulfide partitioning experiments and discuss the plausibility of water accretion as an oxidation mechanism for the mantle.

2.1 High-pressure experimental techniques

This thesis mainly describes experiments performed at high pressures and temperatures in multianvil presses. In a multianvil apparatus, high pres-sures are achieved with two stages of anvils to direct the force of up to 5000 tonnes of hydraulic force onto a small sample volume. The first large-volume high-pressure apparatus was developed during the 1950s by Tracy Hall, and featured a tetrahedral design (Hall, 1958). Some years later, the split-sphere multianvil design was introduced by Kawai and Endo (1970), which featured a steel sphere split into 6 wedge-shaped anvils. These were held in a pressurised oil reservoir, and contained an inner array of 8 tungsten carbide anvils. Pres-sure from the oil was directed through the 2 stages of anvils onto the sample.

The design was later changed to supply pressure with a hydraulic press.

All experiments for this study were accomplished using presses of either this design (Kawai-type) or of a second type, introduced in 1991 (Walker, 1991) that features 6 outer cylindrical anvils, coupled with the same array of 8 inner tungsten carbide anvils. Further details as to the history and design of the multianvil apparatus can be found in the literature (e.g., Keppler and Frost, 2005; Liebermann, 2011).

Figure 2.1: The experimental setup of a multianvil press apparatus. (A, B) The sample, along with cylindrical heating elements, MgO spacers, and a thermocouple is placed in an octahedral pressure medium. (C) This is surrounded by an array of 8 tungsten carbide cubes, which have truncated corners (3 cubes missing to show the geometry). (D) This array of cubes is placed in the press, with epoxy sheets insulating the cubes from the outer steel anvils. (E) The hydraulic press compresses the entire setup.

The 8 inner tungsten-carbide anvils have precisely machined truncations of their corners, such that when placed in a cubic array an octahedral cavity is

created in the center, which is where the high-pressure assembly is placed (figure 2.1). The sample is contained within an octahedral pressure medium, which for all experiments performed for this work was made of MgO doped with 5 wt% Cr₂O₃. The octahedron has a hole drilled through, which is fitted with a zirconia sleeve, into which the sample, heating element and a thermocouple are placed. The different experimental requirements called for small variations in the setup within the assembly, and so further details (including figures) of each setup are given in the individual chapters.

Pyrophilite gaskets and the pressure medium itself flow under pressure, cre-ating a quasai-hydrostatic pressure on the sample. Further details of the multianvil technique, including pressure calibrations, can be found in Kawai and Endo (1970), Walker et al. (1990), Rubie (1999), Frost et al. (2004b) and Keppler and Frost (2005). Four different presses were used for this work, with varying maximum loads of 500, 1200, and 5000 tonnes. Three are at the Bayerisches Geoinstitut, and the fourth is installed at beamline 13-ID-D at the Advanced Photon Source (APS), Argonne National Laboratory for use with synchrotron radiation.

A variety of assembly sizes and tungsten carbide anvil truncation edge lengths were used for this study, the most common being an “18/11”, which indi-cates an 18-mm edge-length octahedron, used with tungsten-carbide cubes with 11 mm corner truncations. In the 500 (Walker-type, Voggenreiter) and 5000 (Kawai-type, Zwick) tonne presses, this setup can produce pressures up to 6 and 18 GPa, respectively. For higher pressures (up to 23 GPa), 10/4 and 10/5 assemblies were used. The pressure calibrations employed were those routinely used at the BGI, and are based on univariant phase transitions that occur at known pressures and temperatures. Frost et al. (2004b) contains details on the pressure calibration of the 5000 tonne Zwick press, and cali-brations for the other presses are reported in Keppler and Frost (2005).

Heating was achieved through the use of both conductive (graphite or rhe-nium) and resistive (lanthanum chromite) heaters, depending on the exper-imental conditions. Temperature is increased by raising the current that is passed through the assembly, and quenched by shutting off the power. For all experiments except those performed at the APS (see chapter 3 for details of

the setup used at the synchrotron), temperature was monitored with an axi-ally inserted, W₉₇Re₃ - W₇₅Re₂₅ (type D) thermocouple, in which the wires are held in a 4-bore alumina tube. For experiments during which a thermocouple failed, temperature was estimated from previous temperature-power relations from experiments performed with the same type of assembly.