Iron geochemistry of the mantle

The Fe/Mg ratio is an important constraint on the compositionally controlled density of the mantle. However, this ratio cannot be inferred from erupted lavas from OIB or MORB sources, but must be determined directly from mantle peridotites. Recently, the Fe/Mn ratio of erupted lavas has been used as an indicator of potential Fe variability in the mantle driven by core-mantle interaction, recycled oceanic crust, or even variations in the temperature of mantle melting. The classic compilation of McDonough & Sun (1995) provided the currently accepted Fe/Mn ratio of the upper mantle, 60±10. The uncertainty on this ratio allows for 15-30% variability in mantle iron abundances, which is equivalent to a density variation larger than observed by seismic tomography in the mantle. To better understand the relationship between mantle peridotites and erupted lavas, and to search for real variability in the Fe/Mn ratio of mantle peridotites, we report precise new ICP-MS measurements of the transition element geochemistry of suites of mantle xenoliths that have known Fe/Mg ratios. For 12 Kilbourne Hole xenoliths, we observe a clear correlation between Fe/Mn and MgO (or Fe/Mg) over an Fe/Mn range of 59-72. Extrapolation of this trend to a Primitive Mantle (PM) MgO content of 37.8 yields an Fe/Mn of 59±1 for the PM. Our new analyses of KLB-1 powder and fused glass beads yield an Fe/Mn of 61.4 for both samples, which plots on the Kilbourne Hole Fe/Mn vs. MgO trend. A set of ten xenoliths from San Carlos yield a wide range of Fe/Mn (56-65) not correlated with MgO content. The San Carlos xenoliths may have experienced a metasomatic effect that imprinted variable Fe/Mn. A clinopyroxene-rich lithology from San Carlos yields an Fe/Mn of 38, which plots on an extension of the Kilbourne Hole Fe/Mn vs. MgO trend. These new results, and those from other xenolith localities being measured in our lab, provide new constraints on the compositional variability of the Earth's upper mantle. McDonough, W. F. & Sun, S.-s. (1995) Chem. Geol. 120, 223-253.