Oxygen isotope constraints on the origin of high-Cr garnets from kimberlites

The association between diamonds and high-Cr, sub-calcic garnets from kimberlites suggests that garnets have experienced high pressure (and presumably, high temperature) conditions in the subcontinental lithosphere mantle (SCLM). The oxygen isotope compositions of these garnets are generally reported to be lower than average values of mantle olivines (similar to 5.2 parts per thousand)-opposite the sense of oxygen isotope fractionation expected at high-temperature equilibrium, and to correlate negatively with their Cr-contents. These observations were interpreted to indicate the sub-calcic garnets either are derived from melting Cr-enriched but O-18-depleted source in the SCLM, or have experienced cryptic metasomatism by high-Cr, high-Mg#, but O-18-depleted fluids/melts in the SCLM or during their ascent to the surface. We investigate the possibility that oxygen isotope characteristics of these garnets instead reflect an analytical artifact that reduces the measured 880 of garnets in proportion to its Cr content-a so-called 'Cr-effect'. Mixtures of garnet standards (UWG-2) plus various amounts of pure Cr-metal were measured for delta O-18-values by infrared-laser fluorination technique (ILFT). Our results show that oxygen isotope compositions of sample mixtures vary systematically as a function of Cr-content and O-2-yield. Mass-balance analyses indicate that the correlation between the measured delta O-18-value and Cr-content is an analytical artifact mostly due to isotopic fractionation between the extracted-O-2 and chromium oxyfluorides left as residues in the sample chamber during the ILFT analyses, which amounts to a decrease in the measured delta O-18 values of similar to 0.064 parts per thousand for the increase of every wt.% Cr2O3 present in garnets. After applying our experimentally-calibrated correction factor to garnet samples in garnet peridotite xenoliths from South Africa, our study reveals an equilibrium oxygen isotope fractionation between the Cr-poor garnet and coexisting olivine (i.e., garnet >= olivine) with a fractionation factor of similar to 0.28 parts per thousand. More importantly, high-Cr garnets (Cr2O3>5%), after correction for the 'Cr-effect', are actually enriched, rather than depleted, in O-18 compared with low-Cr garnets, suggesting the involvement of O-18-enriched crustal materials during their formation. We hypothesize a fluid-flow model using depth-dependent Cr partitioning. This model simultaneously explains relatively high-delta O-18-values and Cr and Mg enrichment in garnets, implying that the reduced C-H-O fluids that are responsible for the formation of diamonds in both eclogite and peridotite xenoliths in kimberlites, are also responsible for transporting Cr to oxidized locations where the Cr partition coefficient between garnet and fluids is high. (C) 2011 Elsevier B.V. All rights reserved.