Although Fe is a major element in mantle silicates and spinel, it is their Fe3+ content that determines whether Fe3+/Fe2+ equilibria can be effective redox buffers under 'normal' mantle conditions. A recent study by Canil et al. (1994) of sub-continental mantle xenoliths concluded that the budgets of Fe3+, C, H and S were sub-equal and that the upper mantle O2 was unlikely to be generally controlled by any single buffer system. To judge whether this assessment is valid for other tectonic environments, we determined the O2 and Fe2O3 the contents of 17 spinel peridotites from the Beni Bousera (Morocco), Ronda (Spain), Lherz (France) and Caussou (France) lherzolite massifs. These massifs have geochemical affinities to sub-oceanic mantle and are considered to have originated in the asthenosphere (Kormprobst, 1969; Obata, 1980. Whereas the Lherz and Caussou massifs have also undergone a subsequent stage in the sub-continental lithosphere, these massifs provide further information about sub-continental mantle processes. On the other hand, parts of Beni Bousera and Ronda appear to have undergone much less late-stage modification and still record many characteristics of sub-oceanic mantle.
The Fe3+/ Fe of spinel (sp), clinopyroxene (cpx) and orthopyroxene (opx) were determined by Mössbauer spectrometry. Olivine (ol) from one sample contained no detectable Fe3+. The Fe3+/ Fe of sp and cpx in each sample are comparable, ranging from the detection limit of the Mössbauer method (0.01-0.02) to 0.30. There are no systematics to this general relationship. Opx has a lower Fe3+/ Fe, with ×0.01 to 0.07. The Fe3+ contents are significantly different from those computed directly from microprobe analyses, without secondary standards. The unreliability of using normal microprobe analyses for calculation of Fe3+ contents is well known (Canil et al., 1994; Wood and Virgo, 1989; McGuire et al., 1991; Luth and Canil, 1993; Woodland et al., 1992).
logO2 (relative to FMQ) were computed for each sample with both the sp-opx-ol (Bryndzia, 1990) and cpx-opx-ol (Luth and Canil, 1993) oxygen barometers, using two-pyroxene temperatures, assuming 15 kb pressure (9 kb for Ronda) and neglecting the Fe3+ in the opx. logO2 values from the sp- and the pyroxene-based oxygen barometers are satisfyingly consistent with each other and are in agreement with published values reported for these massifs (Woodland and Kornprobst, 1992). This consistency further supports the interpretation that the redox values represent true mantle conditions. Based on the cpx-opx-ol equilibrium, Beni Bousera and Ronda record low logO2s: FMQ-4.1 to FMQ and FMQ-2.3 to FMQ, respectively. The most oxidised sample from Ronda was plagioclase-bearing. The Lherz and Caussou samples are relatively oxidised: FMQ+0.3 to FMQ+1.1, with the most oxidised sample being strongly metasomatised. The reduced values recorded at Beni Bousera and Ronda are consistent with the notion that Fe3+ is behaves as an incompatible element during partial melting (Bryndzia, 1990). We find, however, no distinct relationships between logO2 and either major or trace element contents of the spinel or silicate phases.
Fe2O3 contents derived from the Mössbauer data, probe compositions and mineral modes reveal that cpx tends to dominate the total Fe2O3 budget in our peridotites. Sp accounts for only 20-50% of the total in agreement with other studies (Canil et al., 1994; McGuire et al., 1991; Luth and Camil, 1993). Although opx has a low Fe3+/ Fe, its relative abundance and Fe-rich composition means that opx can make a significant contribution to the total Fe2O3 content. At Beni Bousera, the Fe2O3 content ranges from ×0.01 to 0.13 wt%, the lowest being a harzburgite. This is much lower than the 'fertile mantle' value of Canil et al. (1994) (0.30 wt%) and is about half the amount present in several samples from Ronda (0.20-0.24 wt%). It appears unlikely that such small Fe2O3 contents could effectively buffer O2 in the sub-oceanic mantle. Whether enough C is present to buffer O2 in such situations remains open. The very low Fe2O3 in harzburgites implies that they are particularly susceptible to redox changes from interaction with infiltrating fluids or melts. There is no correlation between the total Fe2O3 content and the calculated logO2.
Bryndzia, L.T., & Wood, B.J., Am. J. Sci., 290, 1093-1116 (1990).
Canil, D. et al., Earth Planet. Sci. Lett. 123, 205-220 (1994).
Kornprobst, J., Contr. Miner. Petrol. 23, 283-322 (1969).
Luth, R.W. & Canil, D., Contr. Miner. Petrol. 113, 236-248 (1993).
McGuire, A.V., Dyar, M.D. & Nielson, J.E., Contr. Miner. Petrol. 109, 252-264 (1991).
Obata, M., J. Petrol., 21, 533-572 (1980).
Wood, B. & Virgo, D., Geochim. Cosmochim Acta, 53, 1277-1291 (1989).
Woodland, A.B., Kornprobst, J. & Wood, B.J., J. Petrol. 33, 203-230 (1992).