Partitioning Behaviors of Cobalt and Manganese along Diverse Melting Paths of Peridotitic and MORB-Like Pyroxenitic Mantle

The Co, Mn, Fe, and Ni contents of olivine phenocrysts and host basalts are sensitive to source mantle lithology, which suggests they may be used to constrain the processes of mantle melting and identify basalts formed from non-peridotitic (i.e. pyroxenitic) mantle sources. Here, we use a new comprehensive, forward model involving multiple parameters to simulate partitioning of Co and Mn during partial melting of the mantle in different tectonic settings: (1) polybaric continuous melting of peridotite mantle in mid-ocean ridges can generate melts that show decreasing Co and Mn with increasing degrees of melting so that the mid-ocean ridge basalts (MORBs) contain similar to 39-84 mu g/g Co and similar to 900-1600 mu g/g Mn; (2) flux-melting of the mantle wedge in subduction zones tends to produce a melt that has Co increasing from similar to 24 to 55 mu g/g and Mn from similar to 500 to 1110 mu g/g with increasing temperature; (3) melts produced by isobaric melting of the subcontinental lithospheric mantle are also sensitive to increasing temperature and have similar to 35-160 mu g/g Co and similar to 800-2600 mu g/g Mn; (4) decompression melting of peridotite related to the mantle plume generates melts containing similar to 45-140 mu g/g Co and similar to 1000-2000 mu g/g Mn, and the abundances of these metals decrease with increasing degrees of melting; and (5) partitioning behaviors of Co, Mn, and Ni during decompression melting of MORB-like pyroxenite contrast with those during decompression melting of peridotite due to the different mineralogy and compositions in mantle lithologies, and the MORB-like pyroxenite-derived melt is metal-poor with similar to 25-60 mu g/g Co, similar to 290-1600 mu g/g Mn, and similar to 160-340 mu g/g Ni. Although high-Ni, low-Mn forsteritic olivine phenocrysts and high melt Fe/Mn ratio have been proposed as diagnostic indicators of pyroxenitic components in the mantle, our models show that these features can be also generated by melting of peridotite at greater depth (i.e. a high pressure and temperature). To quantify the effect of high-pressure melting of peridotite on these diagnostic indicators, we modeled the correlations of melt Fe/Mn and olivine Co, Mn, and Ni contents with melting depth along the decompression melting path of a thermal plume. When Fe/Mn ratios of basalts and/or compositions of olivine phenocrysts deviate significantly from our modeled correlation lines, high-pressure melting of peridotite cannot explain these data, and the existence of pyroxenitic component in the mantle source is likely required. The pyroxenite-derived melt is modeled to be Ni-poor, but mixing with a peridotite-derived melt can strongly increase the partition coefficient of Ni between olivine and mixed melt, resulting in the generation of high-Ni olivine phenocrysts in plume-associated magmatic suites.