• Keine Ergebnisse gefunden

Ferric iron content of garnets and attainment of redox equilibrium

a Wads Mg 2 SiO 4

5. Fe 3+ /Fe tot measurements of garnets equilibrated with carbon and carbonate in a peridotite assemblage

5.3.3 Ferric iron content of garnets and attainment of redox equilibrium

Results of Mössbauer determinations of the ferric/ferrous iron ratios in garnet-rich layers are shown in Table 5.3 as determined from the relative areas of Fe3+ and Fe2+ doublets; hyperfine parameters are also reported. When compared to garnet spectra analysed by both Luth et al. (1990) and Gudmundsson and Wood (1995), the I.S. and the Q.S. (isomer shift and quadrupole splitting) appear very similar (figure 5.4a), although some discrepancies arise most likely due to difficulties in fitting spectra with either small ferric iron concentrations or due to interference from the presence of other iron-bearing phases in the layer. In experiments where magnesite, opx and olivine were added to the layer garnet is assumed to be the only ferric iron-bearing phase in the assemblage. This is a safe assumption since no significant amount of Fe3+ has been previously reported either for olivine or orthopyroxene (O’Neill et al., 1993).

Table 5.3 Hyperfine parameter for Mössbauer measurement of ferric iron Sample Fe2+ Fe3+

Notes: I.S. stays for isomer shift, while Q.S. for quadrupole splitting. Both are reported in mm/s and measured relative to the α-Fe metal at 298 K. X2 is the chi-squared referred to the fitted spectra. Results of ferric iron in garnets are characterized by an uncertainty of ±0.02.

The accuracy of the ferric iron determinations was tested with run V587a by taking two different spectra: one measurement carried out over the entire cross section of the sample; a second measurement limited to the layer containing garnet. Fitted spectra are shown in figure 5.4b and both results are consistent and give about 6 % of ferric iron in garnet. The uncertainty in the ideal thickness calculated to reduce the saturation effect was tested collecting a few spectra after re-polishing the samples and resulted in a value within the uncertainty for the measured Fe3+/∑Fe (±2%). Finally, small amounts of ferric iron measured in recovered garnets (V621b and S4432b) seem to confirm the absence of significant Fe3+ in the starting glasses.

Figure 5.5 shows Fe3+/∑Fe ratios of garnets as a function of pressure compared to Fe3+/∑Fe ratios of garnet-bearing mantle peridotites from different localities. Fe3+/∑Fe ratios are also shown from the calibration experiments of Gudmundsson and Wood (1995). It is noteworthy that the ferric/ferrous contents of garnets equilibrated in this study cover a similar range to measurements made on garnets from mantle xenoliths. Fe3+/∑Fe garnet ratios from the calibrations of Gudmundsson and Wood (1995), on the other hand, are in general more oxidized than most natural samples.

The relationship between the garnet Fe3+/∑Fe ratio and the fo2 determined with the iridium redox sensor (FMQ normalized) is shown in figure 5.6. The results of the calibration experiments of Gudmundsson and Wood (1995) are also shown for comparison. In the study of Gudmundsson and Wood (1995) the final garnet Fe3+/∑Fe ratios were approached from natural garnet starting materials that had both higher and lower initial Fe3+/∑Fe ratios, as shown in figure 5.6. However, reversals were not performed in this previous study at the same oxygen fugacities and at the fo2 comparable with the current study Fe3+/∑Fe ratios were only approached from compositions with initial higher levels of Fe3+. Gudmundsson and Wood (1995) also state that reducing garnet ferric Fe contents was extremely sluggish and only a very limited shift in the garnet Fe3+/∑Fe ratio was possible. In the current study starting garnet glass compositions were employed for most experiments that had initial Fe3+ levels

below detection limits and thus, equilibrium was approached from the opposite direction to those of Gudmundsson and Wood (1995).

a b

Figure 5.4 (a) Mössbauer spectra of garnet V458a and H2759b at 298 K with ferric iron doublets shown in red. b) Spectra collected from V587a showing differences in line widths and relative areas between a spectrum collected over the entire sectioned sample (bulk) and the spectrum collected only from the garnet-bearing layer. Green and red doublets are representative of ferrous and ferric iron measured in garnet. Small white doublets are also shown, which represent ferrous iron-bearing phases (olivine and orthopyroxene).

Figure 5.5 Shown is the increase of ferric iron content of recovered synthetic garnets measured by Mössbauer spectroscopy as function of pressure. Ferric iron contents of natural garnets from mantle xenoliths (Kaapvaal Craton, Vitim Plateau, Slave Craton; see legend for references) are also shown for comparison. Green squares are experimental data used to calibrate eq. (5.1) and reported by Gudmundsson and Wood (1995).

As several different experimental configurations were employed in this study, by comparing the extent of garnet oxidation with the reversals of Gudmundsson and Wood (1995) it is possible to assess which of the configurations from this study are likely to have reached equilibrium. Firstly, some of the lowest garnet Fe3+/∑Fe ratios occurred in experiments where magnesite was not added to the garnet layer and the temperatures were too low for melting to occur. These experiments (i.e., V621b, H2759b) therefore, likely did not reach redox equilibrium with the buffering carbon-carbonate assemblage due to the lack of local oxidizing agent and are excluded from further consideration. Secondly, subsolidus experiments appear also to result in low Fe3+/∑Fe ratios that can be excluded (V606a). A number of experiments were performed at 1400 °C and 6-7 GPa, which although performed at almost identical conditions resulted in varying levels of garnet Fe3+ content. The highest levels of Fe3+ were reached when either H2O was added to the sample (S4402a) or when the sample was Ca-free (V493b).

However, in a few experiments it is unclear why similar conditions rendered garnet Fe3+ contents that were so different. Experiment V524b for example performed in a Ca free system at 7 GPa and 1500 °C produced garnet with a Fe3+/∑Fe ratio of only 0.055, where as experiment V493b, which was identical except being run at 100 °C lower in temperature, produced garnet with a Fe3+/∑Fe ratio of 0.126. In several instances, however, the highest levels of Fe3+ were reproducible from different starting materials such as at 6 GPa and 1350°C where garnet Fe3+/∑Fe ratios of 0.12 and 0.11 were obtained from synthetic glass and natural garnet starting compositions respectively. From the consistency of these repeated experiments it is assumed equilibrium was achieved.

-3 -2.5 -2 -1.5 -1 -0.5 0

0 0.05 0.1 0.15 0.2 0.25 0.3

Fe3+/Total Fe

Log fO2 (FMQ) G&W

G&W initial 7GPa 3GPa 7GPa Ca-free 6GPa

Figure 5.6 Fe3+/∑Fe ratio of garnets from experiments performed at different pressures plotted against oxygen fugacity determined from the Ir-alloy redox sensor normalized to FMQ. G&W refers to the previous measurements of Gudmundsson and Wood (1995) from samples equilibrated at 3 - 3.5 GPa and 1300 °C. Arrows indicate the approach to equilibrium for two experiments performed at approximately the same conditions from this study and that of Gudmundsson and Wood (1995). Open circles indicate G&W starting garnet compositions.

5.4 Oxygen-thermobarometry measurements on the experimental garnet peridotite