The chemical composition of Earth's lower mantle can be constrained by combining seismological observations with mineral physics elasticity measurements. However, the lack of laboratory data for Earth's most abundant mineral, (Mg,Fe,Al)(Al,Fe,Si)O bridgmanite (also known as silicate perovskite), has hampered any conclusive result. Here we report single-crystal elasticity data on (Al,Fe)-bearing bridgmanite (MgFeSiAl)O measured using high-pressure Brillouin spectroscopy and X-ray diffraction. Our measurements show that the elastic behaviour of (Al,Fe)-bearing bridgmanite is markedly different from the behaviour of the MgSiO endmember. We use our data to model seismic wave velocities in the top portion of the lower mantle, assuming a pyrolitic mantle composition and accounting for depth-dependent changes in iron partitioning between bridgmanite and ferropericlase. We find excellent agreement between our mineral physics predictions and the seismic Preliminary Reference Earth Model down to at least 1,200 kilometres depth, indicating chemical homogeneity of the upper and shallow lower mantle. A high Fe/Fe ratio of about two in shallow-lower-mantle bridgmanite is required to match seismic data, implying the presence of metallic iron in an isochemical mantle. Our calculated velocities are in increasingly poor agreement with those of the lower mantle at depths greater than 1,200 kilometres, indicating either a change in bridgmanite cation ordering or a decrease in the ferric iron content of the lower mantle.

A new suite of 173 clinopyroxene grains from heavy-mineral concentrates of the diamondiferous Novinka kimberlite (Upper Muna field, Yakutia) has been analyzed for major and minor elements with an electron microprobe to perform a thermobarometric study and model the thermal structure of the Archean Upper Muna lithospheric mantle. Scrupulous evaluation of propagation of analytical uncertainties on pressure estimates revealed that (1) the single-clinopyroxene geobarometer can be very sensitive to analytical uncertainties for particular clinopyroxene compositions, and that (2) most clinopyroxenes from Novinka have compositions that are sensitive to analytical uncertainties, notwithstanding their apparent compositional suitability for single-clinopyroxene thermobarometry based on previously proposed application limits. A test on various mantle clinopyroxenes containing different proportions of the sensitive elements Cr, Na, and Al allowed us to identify clinopyroxene compositions that produce unacceptably high propagated errors and to define appropriate analytical conditions (i.e., higher beam currents and longer counting times for specific elements) that allow precise P-T estimates to be obtained for sensitive compositions. Based on the results of our analytical test, and taking into account the intrinsic limitations of the single-clinopyroxene thermobarometer, we have designed a new protocol for optimum thermobarometry, which uses partly revised compositional filters. The new protocol permits precise computation of the conductive paleogeotherm at Novinka with the single-clinopyroxene thermobarometer of Nimis and Taylor (2000). Thermal modeling of the resulting P-T estimates indicates a ~34 mW/m2 surface heat flow, a thermal lithosphere thickness of ~225 km, and an over 100 km thick “diamond window” beneath Novinka in the middle Paleozoic (344–361 Ma). We estimate that appropriate analytical conditions may extend the applicability of single-clinopyroxene thermobarometry to over 90% of clinopyroxene-bearing garnet peridotites and pyroxenites and to ~70% of chromian-diopside inclusions in diamonds. In all cases, application to clinopyroxenes with Cr/(Cr+Al)mol < 0.1 is not recommended. We confirm the tendency of the single-clinopyroxene barometer to progressively underestimate pressure at P > 4.5 GPa