Experimental methods

Experiments were carried out at pressures of 16, 23, 25 and 45 GPa and temperatures ranging between 1500 and 1700 °C using different starting materials. Equilibrium (4.1) was studied at 16 and 23 GPa employing a carbonated harzburgite starting composition (Eq1) in the system Fe-Mg-Si-O-C assembled from a mineral mix of San Carlos olivine, San Carlos enstatite, synthetic magnesite and pure graphite, in the molar ratio 3:2:1:1. In addition, in each experiment a layer of olivine mixed with enstatite and magnesite (2:1:1) was added sandwiched between two layers of the Eq1 mix to facilitate Mössbauer measurements. 5 wt. % iridium metal powder (≤ 5 µm) was added to all layers to act as a sliding redox sensor.

Equilibrium (4.2) was studied at 25 and 45 GPa employing a mineral mixture in the system Fe-Mg-O-C (Eq2) comprising synthetic MgCO3, pre-synthesized ferropericlase (Fe/{Fe+Mg}=0.17) and pure graphite powder. 5 wt. % pure iridium metal was also added.

At 45 GPa one run (M140) was also performed with Al-bearing glass powder added to the Eq2 mixture, in order to crystallize aluminous magnesium silicate perovskite with a composition similar to that crystallised from pyrolitic bulk compositions (Kesson et al., 1998). This glass was prepared from high purity MgO, Al2O3, Fe2O3 and SiO2. After weighing, the oxide mixture was ground for 1 hour under ethanol in an agate mortar. The dried mixture was decarbonated by heating at a rate of 150

°C/hour to 1000 °C and then glassed at 1650 °C. The recovered and ground glass was then reduced in a

controlled H2-CO2-Ar atmosphere at 1000 °C at an oxygen fugacity equivalent to the Ni-NiO oxygen buffer for 24 hours.

Further experiments at 25 and 45 GPa employed the Eq2 bulk composition but 3 wt. % Ni metal was added instead of Ir metal in order to test a second type of sliding redox sensor.

Experiments between 16 and 25 GPa were performed in 1000 and 1200 tonne Kawai-type multianvil presses at the Bayerisches Geoinstitut.

Tungsten carbide anvils of 8 and 4 mm truncation edge length (TEL) were used with standard 14 and 10 mm edge length of Cr2O3-doped MgO octahedra. Starting powders were loaded into a graphite container that was enclosed in a rhenium foil capsule. The capsule was placed in the central portion of a straight lanthanum chromite furnace, surrounded by an MgO sleeves and spacers. The temperature was monitored with a W97Re3–W75Re25 (D type) thermocouple inserted within an alumina sleeve, with the junction in contact with the top of the capsule. Experiments were performed between 1500-1700 °C for 1-15 hours.

Experiments at 45 GPa (M131 and M140) were carried out using the MADONNA D-DIA (1500 tons) press installed at the Geodynamics Research Center (Ehime University, Japan) employing 14 mm edge length sintered diamond anvils. In these runs the starting powder was placed in a graphite capsule and compressed inside an MgO pressure medium (OEL 2.0 mm) doped with 5 wt. % Cr2O3. Fired pyrophyllite gaskets were used. Further experimental details are described by Tange et al. (2008). High temperatures were reached using a cylindrical LaCrO3 furnace. Temperature was measured with aW97Re3−W75Re25 thermocouple in contact with the bottom of the graphite capsule. The experiments were run for approximately 1 hour during which the temperature was manually controlled to within ±5

°C. After quenching, a long decompression time (30 hrs) was employed in order to prevent breakage of the sintered diamond anvils.

Pressure calibrations at the Bayerisches Geoinstitut are reported in Keppler and Frost (2005).

Sample pressures in experiments performed at 45 GPa were calibrated as a function of oil pressure based on in situ synchrotron X-ray diffraction experiments performed using the SPEED-Mk.II multianvil press installed at the BL04B1 beam line in SPring-8 (Katsura et al., 2004). Au was used as an internal pressure standard (Tsuchiya, 2003) and calibrations were performed up to ~62 GPa between 1700 and 2100 K. This calibration was further tested off-line by monitoring semiconductor to metal transitions in ZnS (15.5 GPa; Onodera and Ohtani, 1980), GaP (22.5 GPa; Dunn and Bundy, 1978), Zr (α-γ and β-γ transitions) and Fe2O3. Off-line calibration was consistent with synchrotron experiments (figure 4.1).

Fig 4.1 Pressure calibration for the 1.5 mm truncated edge length (TEL) assembly using laboratory and in situ X-ray diffraction experiments and the equation of state of gold (Tsuchiya, 2003). HT means high temperature. See text for further details.

Recovered samples were mounted in epoxy resin, sectioned and polished parallel to the axial furnace direction under ethanol to preserve carbonate phases. Textural observations of the recovered

run products were performed using FEG Scanning Electron Microscopy (Bayerisches Geoinstitut) and Field Emission Scanning Electron Microscopy (Geodynamic Research Center-Ehime), while the chemical composition of each phase was obtained using a Jeol JXA-8200 electron microprobe equipped with five wavelength dispersive spectrometers. An accelerating voltage of 15 kV and a beam current of 5-20 nA was employed and counting times varied between 20 and 10 s on peak and background (30 and 15 s for Ir-Fe metal) with a focused electron beam. Standards were natural silicates and metals (Ir, Fe and Ni). The PRZ correction was applied. The carbon content of metallic phases in the recovered samples was also randomly detected following the procedure described in Dasgupta and Walker (2008) and employing vitrified carbon (purity >99.99 % provided by Alfa) as standard.

The ferric/ferrous ratios within layers of wadsleyite (at 16 GPa) and Al-bearing perovskite (at 45 GPa) were determined at room temperature and pressure using ⁵⁷Mössbauer spectroscopy in transmission mode on a constant acceleration with a nominal 370 MBq ⁵⁷Co high specific activity source. The effective Mössbauer thickness was estimated from the composition. The velocity scale was calibrated relative to 25 μm thick α-Fe foil. Spectra were fitted to a Lorentzian line-shape using the fitting program NORMOS written by R.A. Brand (distributed by Wissenschaftliche Elektronik, Germany). Spectra were collected for periods of between 3-7 days over a sample region of approximately 250 µm in diameter.

Additionally, the iron valence state in minerals from experiment M140 were measured by Electron Energy Loss Spectroscopy on a thin film extracted using QUANTA 3D FEG focused ion beam. The Fe L2,3-edge energy-loss near-edge structure (ELNES) of the sample was examined using a 200 kV analytical scanning transmission electron microscope equipped with a parallel electron energy-loss spectrometer. ELNES spectra were collected in diffraction mode with convergence and collection semi angles of α = 4 mrad and β = 1.45 mrad, an energy dispersion of 0.1 eV per channel and 5 - 10 seconds integration time per read-out. The incident beam current was about 16 nA, and the fluence rate was

1.26x10³ e/Ų/sec. To reduce electron irradiation damage during EELS measurements especially on perovskite grains, the TEM thin foil was cooled to nearly liquid nitrogen temperature (ca. -160 °C) in a Gatan cooling holder. To evaluate the intensity ratio of L2,3 –edges, the energy resolution of 0.8-0.9 eV was measured as the width of the zero-loss peak (ZLP) at half height. The energy scale of the core-loss spectrum was fixed using the Fe L3-edge maxima of predominantly Fe²⁺- and Fe³⁺-bearing phases, ferropericlase and magnesium silicate perovskite, at 707.8 eV and 709.5 eV, respectively. Spectral processing included subtraction of the dark current, alignment and summation of offset spectra to reduce channel-to-channel gain variations, background subtractions of the form AE-g, deconvolution of the ZLP to remove the multiple scattering effect on the core-loss edge, and subtraction of the continuum intensity beneath the Fe L2,3 edge using a double arctangent function with two fixed inflection points at 708.65 eV and 721.65 eV. Quantification of the Fe L2,3-ELNES was made using an empirically-calibrated universal curve (van Aken and Liebscher, 2002).

4.3 Results

Experimental conditions and recovered phase assemblages are listed in Table 4.1, with full chemical analyses reported in Table S4.1 and S4.2 in the Appendix. In all experiments the graphite capsule transformed to polycrystalline diamond. Experiments were performed below the carbonated solidus for the various compositions and no indication of melting was observed (see figure 4.2).

At 16 GPa products from the study of equilibrium (4.1) contained wadsleyite, clinoenstatite, magnesite and diamond in addition to Ir-Fe alloy. At 23 GPa in the Ca-Mg-Si-Fe-O-C system ringwoodite and stishovite supplanted wadsleyite and clinoenstatite and coexist with Mg- and Ca-perovskite. At 25 and 45 GPa products included ferropericlase, magnesite, diamond and Ir-Fe alloy.

Experiment M140 at 45 GPa additionally contained Al-perovskite, while run H2981 at 25 GPa and

M131 at 45 GPa, which initially contained Ni instead of Ir, contained no metal phase after the experiment as all Ni oxidized and partitioned mainly into ferropericlase.

Tabella 4.1 Experimental conditions and run products Run no

P

(GPa) T (°C) Time

(hr) phases

S4226 16 1500 12 wads Cl-En mst C S4278 16 1550 12 wads Cl-En mst C H3102 23 1600 1 ringw Mg/Ca-Pv mst C Stish H2946 25 1500 12 Fe-pc - mst C H2982 25 1500 16 Fe-pc - mst C H2887 25 1550 12 Fe-pc - mst C S4807 25 1600 1 Fe-pc - mst C H2981* 25 1500 12 Fe-pc - mst C M131* 45 1700 1 Fe-pc - mst C M140 45 1700 1 Fe-pc Mg-Pv mst C

Notes: Wads = wadsleyite, Rinw = ringwoodite, Cl-en = clino-enstatite, C = diamond, Mst = magnesite, Mg/Ca-Pv = Mg, Ca-bearing perovskite, Stish = stishovite, Fe-pc = ferropericlase. * Runs containing nickel.

Fig 4.2: a) recovered sectioned run product S4226 (16 GPa/1500 °C) showing the wadsleyite-rich layer. b) H2946 run (25 GPa/1500 °C) showing a heterogeneous mineral assemblage with: magnesite (mst), Fe-periclase (Fe-pc), diamonds (dark grains) and the iron-iridium alloy (bright phase). c) Assembly used for experiments with sintered diamond anvils. d) Run M140 (45GPa/1700°C) showing ferropericlase (Fe-pc), magnesite (Mst), diamonds (C) and Mg-perovskite (Pv).

The TEM image in figure 4.3 shows part of the product from M140 at 45 GPa where coarse-grained ferropericlase is in contact with smaller well crystallized perovskite grains. Although only run for one hour M140 shows evidence for textural equilibrium. In addition the determined KDFe/Mg between perovskite and ferropericlase (=

{ _X

^Pv_Fe

_X

^Pv_Mg

} { _X

^Fp_Fe

_X

^Fp_Mg

}

) of 0.54 is in excellent agreement with values determined by Irifune et al. (2010) at similar pressures and temperatures (figure 4.4a). The Fe-Mg partition coefficient between magnesite and silicate minerals was found to decrease as the silicate phases transform to progressively higher-pressure polymorphs as shown in figure 4.4b. The Fe-Mg partition coefficient between magnesite and ferropericlase remained constant between 25 and 45 GPa, which seems to preclude any significant effects resulting from changes in the spin state of Fe in ferropericlase on Fe-Mg partitioning over this pressure range (Badro et al., 2003).

Fig 4.3: Bright-field image of the film recovered from run M140 along a [1-10] zone axis. Inclusions are shown (Mg-Pv magnesio-perovskite, Fe-pc ferropericlase and iridium-iron alloy), while dark lines and dots are clearly distributed on the ferropericlase grain and representative of dislocation features. On upper left part a SAED pattern is shown for Mg-Pv grain used for EELS analysis.

a

b

Fig 4.4: a) Shown is the Fe/Mg partition coefficient calculated for Mg-Pv and ferropericlase as compared to data by Irifune et al. (2010); b) the Fe/Mg KD is shown between magnesite and silicates (clinoenstatite and Mg-Pv) and between magnesite and ferropericlase as function of pressure.

Mössbauer spectra for wadsleyite and aluminous silicate perovskite from 16 and 45 GPa respectively are shown in fig. 4.5a-b. Spectra for wadsleyite (S4226, S4278) were fitted to one quadruple doublet for Fe²⁺ in M1 and M2 site and one for Fe³⁺ in the M3 site. The Fe³⁺/∑Fe ratio of wadsleyite was found to be 0.04 (2) in the two samples examined. Spectra for aluminous perovskite (M140) were fitted to

four quadrupole doublets, two for Fe²⁺ and one for Fe³⁺ in perovskite and a further doublet for Fe²⁺ in coexisting ferropericlase. The corresponding Fe³⁺/∑Fe ratio is 0.79 (±0.1). Further measurements were made on both perovskite and ferropericlase from the same sample using EELS (Fig. 4.5c-d).

Measurements on perovskite grains yielded a Fe³⁺/∑Fe ratio of 0.68(3), which is just within the uncertainty of the Mössbauer measurement. The Fe³⁺/∑Fe ratio for coexisting ferropericlase was 0.02(5).

In each experiment the oxygen fugacity was buffered by the coexistence of diamond and magnesite through the Fe-bearing equivalents of equilibrium (4.1) at 16 and 23 GPa and equilibrium (4.2) at 25 and 45 GPa. The oxygen fugacity was then measured in each experiment using a sliding redox buffer based on the Fe content of Ir-Fe alloy measured after the experiment (Stagno and Frost, 2010). The redox sensor in the wadsleyite stability field employs the equilibrium,

Fe2SiO4 = SiO2 + 2Fe + O2 (4.3)

wadsleyite alloy

where the oxygen fugacity is therefore calculated through,

[ ]

_RT

a

^{WadsFe SiO}

a

^SiO

a

^metalFe Ir-Fe alloy and silica respectively. The silica activity is determined from the equilibrium,

Im Dokument The carbon speciation in the Earth’s interior as function of pressure, temperature and oxygen fugacity (Seite 89-97)