The buffering capacity of lithospheric mantle: implications for diamond formation

Current models for the formation of natural diamond involve either oxidation of a methane-bearing fluid by reaction with oxidized mantle, or reduction of a carbonate-bearing fluid (or melt) by reaction with reduced mantle. Implicit in both models is the ability of the mantle with which the fluid equilibrates to act as an oxidizing or reducing agent, or more simply, to act as a source or sink of O-2. If only redox reactions involving iron are operating, the ability of mantle peridotite to fulfill this role in diamond formation may not be sufficient for either model to be viable. Using the recent experimental recalibration of olivine-orthopyroxene-garnet oxybarometers of Stagno et al. (2013), we re-evaluated the global database of similar to 200 garnet peridotite samples for which the requisite Fe3+/Fe2+ data for garnet exist. Relative to the previous calibration of Gudmundsson and Wood (1995), the new calibration yields somewhat more oxidized values of Delta log fO(2) (FMQ), with the divergence increasing from <0.5 units of log fO(2) at similar to 3 GPa to as much as 1.5 units at 5-6.5 GPa. Globally, there is a range of similar to 4 log units fO(2) for samples from the diamond stability field at any given pressure. Most samples are sufficiently reduced such that diamond, rather than carbonate, would be stable, and CHO fluids at these conditions would be H2O-rich (>60 mol%), with CH4 being the next most abundant species. To ascertain the capacity for mantle peridotite to act as a source or sink of O-2, we developed a new model to calculate the fO(2) for a peridotite at a given P, T, and Fe3+/Fe2+. The results from this model predict 50 ppm or less O-2 is required to shift a depleted mantle peridotite the observed four log units of fO(2). Coupled with the observed distribution of samples at values of fO(2) intermediate between the most reduced (metal-saturated) and most oxidized (carbonate-saturated) possible values for diamond stability, these results demonstrate that peridotites are very poor sinks or sources of O-2 for possible redox reactions to form diamond. A corollary of the poor redox buffering capacity of cratonic peridotites is that they can be employed as faithful indicators of the redox state of the last metasomatic fluid that passed through them. We propose that diamond formation from CHO fluids is a predictable consequence either of isobaric cooling or of combined cooling and decompression of the fluid as it migrates upward in the lithosphere. This establishes a petrological basis for the observed close connection between subcalcic garnet and diamond: based on high solidus temperatures of harzburgite and dunite effectively precluding dilution of CHO fluids through incipient melts, such highly depleted cratonic peridotites are the preferred locus of diamond formation. Due to a rapid increase in solidus temperature with increasing CH4 content of the fluid, diamond formation related to reduced CHO fluids may also occur in some cratonic lherzolites.