94 Earth & Planetary Science It is widely accepted that at least the Earth’s uppermost mantle, perhaps down to the top of the mantle transition zone, has a peridotitic (pyrolitic) bulk composition, on the basis of petrological evidence. On the other hand, a variety of chemical compositions, ranging from peridotitic to chondritic, have been proposed for the lower mantle. This has long remained controversial owing to the lack of conclusive arguments. The mineral assemblage of the lower mantle has been examined from density measurements under high P-T conditions. However, recent computational simulations demonstrated that such experimentally derived density and bulk moduli do not place unique constraints on the mantle composition because of their intrinsic uncertainties, whereas shear velocity ( V S ) data strongly constrain the lower mantle models. It is thus crucial to obtain reliable V S data of the major lower mantle constituents, silicate perovskite (pv) and (Mg,Fe)O ferropriclase (fp), under relevant high P-T conditions. Recent progress in Brillouin scattering spectroscopy optimized for extreme high-pressure conditions has so far enabled us to measure V S of MgSiO 3 -pv [1], post-perovskite [2], and MgO periclase [3] up to 172 GPa. Although those sound velocity data indeed provide valuable information on the lower mantle mineralogy, the effects of chemical impurities, such as iron and aluminum, and high temperature remain unsettled. More importantly, sound velocity measurements of pv and fp have never been performed under simultaneously high- P and high- T conditions corresponding to the lower mantle environment. Here, we determined V S of aluminous silicate pv and fp up to 124 GPa at 300 K. The high P-T measurements were also conducted on MgSiO 3 pv and MgO at a temperature of 2700 K up to 91 GPa using the newly developed Brillouin scattering system at beamline BL10XU [4,5]. We obtained very sharp Brillouin peaks from the transverse acoustic modes of each phase over the entire P-T range we explored (Fig. 1). No significant peak broadening was observed with increasing pressure. The angle-dispersive synchrotron X-ray diffraction measurements were conducted simultaneously to determine the volume of the sample and the pressure at BL10XU, the synchrotron X-ray source of SPring-8 in the energy range of 30–50 keV. The X-ray diffraction peaks for each sample were very sharp, and two-dimensional X-ray diffraction images showed clear circular Debye rings with fairly uniform intensity distribution along the circle, indicating that the sample did not undergo significant grain growth or lattice preferred orientation under high pressure and high temperature. On the basis of the fit of the finite strain to the P-V S profiles of Al-pv and fp, we obtained the best-fit values of G 0 = 166(1) GPa (shear modulus) and G 0 ' = 1.57(5) (pressure derivative of shear modulus) for Al-pv. The G 0 ' of Al-pv coincides with the 1.56(4) of pure MgSiO 3 pv, indicating a minimal effect of Al on G 0 '. For fp, the finite strain was fitted separately for low- (5 – 40 GPa) and high- (60 – 121 GPa) pressure ranges, because anomalous behavior was observed around 50 GPa owing to spin crossover. The fitting result gives G 0 = 113(2) GPa and G 0 ' = 2.15(5) for the HS state, and G 0 = 130(2) GPa and G 0 ' = 2.04(5) for the LS state. Extrapolation of the HS data to high pressure does not reproduce the LS data, supporting the claim that the spin crossover of iron is not associated with the elastic softening. These G 0 and G 0 ' values for Al-pv and fp are plotted together with previous data in Fig. 2 as a function of iron or aluminum content. With such relationships, we can estimate G 0 and G 0 ' for the representative mantle composition ( X Mg = 79 for fp and X Al 2 O 3 = 4 wt% for pv). The high P-T measurements on pure MgSiO 3 pv and MgO showed the velocity reduction by ~4% and ~6%, respectively, on average at 2700 K from the room- T data at lower mantle pressures. A linear fitting of the shear moduli against pressure, combined with the reported values of G 0 and G 0 ' [1,3], provides temperature derivatives of the shear modulus: d G /d T = –0.020(1) GPa/K for both pv and MgO. The shear strain derivatives of the Mineralogical model of lower mantle inferred from high-pressure and -temperature sound velocity data Fig. 1. Brillouin spectra of lower mantle phases: (a) MgSiO 3 (+4 wt% Al 2 O 3 ) pv at 95 GPa/300 K; (b) (Mg 0.92 Fe 0.08 )O at 86 GPa/300 K; (c) MgSiO 3 pv at 91 GPa/2700 K; and (d) MgO at 48 GPa/2700 K. TA and LA indicate transverse and longitudinal acoustic modes of Brillouin shift, respectively. (b) (d) 25 20 15 10 5 0 -5 -10 -15 -20 -25 Frequenty Shift (GHz) 25 20 15 10 5 0 -5 -10 -15 -20 -25 Frequenty Shift (GHz) Intensity TA TA TA TA TA TA TA TA diamond TA diamond TA ghost peak diamond TA ghost peak ghost peak diamond TA diamond TA diamond TA diamond TA NaCl TA NaCl TA NaCl LA NaCl LA ghost peak ghost peak ghost peak ghost peak diamond TA ghost peak Rayleigh peak Al-MgSiO 3 Pv 95 GPa, 300 K 86 GPa, 300 K 91 GPa, 2700 K 48 GPa, 2700 K MgSiO 3 Pv (Mg, Fe)O MgO (a) (c) Rayleigh peak Rayleigh peak Rayleigh peak 95 Grüneisen parameter ( g ) are estimated to be η S0 = 2.4(2) for pv and 3.0(3) for MgO. Present measurements performed over a wide P-T range that covers almost the entire range of lower mantle conditions allow us to constrain the lower mantle mineralogy. We model the lower mantle as a two-phase mixture of pv and fp in the SiO 2 -MgO-FeO- Al 2 O 3 system, in which (Al,Fe)-bearing pv contains 4 wt% Al 2 O 3 with X Mg = 94 and fp has X Mg = 79. A constant Mg-Fe partitioning coefficient between pv and fp is assumed for the entire lower mantle. The V S profiles of these pv and fp were calculated along the typical temperature profiles (Fig. 3(a)). The PREM is best fitted by the mixture of 95% pv and 5% fp in volume ( X Pv = 0.95) (Fig. 3(b)). The velocity of fp increases steeply by ~ 4% across the spin crossover, however, such an anomalous feature is not clear in the calculated profile of the pv+fp mixture, suggesting that the spin crossover of iron in fp may be seismologically invisible. On the other hand, the V S profile for a peridotitic (pyrolitic) mantle ( X Pv = 0.80) is shown to be lower by up to 3.2% than the PREM throughout the pressure range of the lower mantle, indicating that the conventional peridotitic model is incompatible with the seismological observations. The present results indicate that the conventional peridotitic mantle model is not compatible with the seismic properties of the lower mantle and suggest that the lower mantle is dominated by perovskite (~95 vol%) and therefore is silica-rich in comparison with the upper mantle. The different chemical compositions between the upper and lower mantle could be a consequence of fractional crystallization of the magma ocean extending to the deep lower mantle in the early history of the Earth. The primordial chemical stratification may have been preserved through the subsequent solid-state convection until the present day. The layered mantle convection is presumed to have been predominant in early Earth, indicating limited mass transport between the upper and lower mantle. The seismic tomography images of subducting slabs or upwelling plumes penetrating the upper/lower mantle boundary may not represent whole-mantle convection but intermittent/transitional stages between layered and whole mantle-convections. Motohiko Murakami Department of Earth and Planetary Materials Sciences, Tohoku University Email: motohiko@m.tohoku.ac.jp References [1] M. Murakami et al .: Earth Planet. Sci. Lett. 256 (2007) 47. [2] M. Murakami et al .: Earth Planet. Sci. Lett. 259 (2007) 18. [3] M. Murakami et al .: Earth Planet. Sci. Lett. 277 (2009) 123. [4] M. Murakami et al .: Phys. Earth Planet. Inter. 174 (2009) 282. [5] M. Murakami, Y. Ohishi, N. Hirao and K. Hirose: Nature 485 (2012) 90. G ' 0 G 0 (GPa) 0.5 1.0 1.5 2.0 (c) (d) (a) (b) 150 140 160 170 180 190 200 1.4 1.6 1.8 2.0 2.2 2.4 2.6 80 90 100 120 110 130 140 150 160 (Mg, Fe)O Al-MgSiO 3 Pv 0 1 2 3 4 5 X Al2 O 3 (wt%) X Mg ((100 * Mg/(Mg+Fe)) mol ) This study (LS) This study This study (HS) Jacobsen et al. , 2002 Jacobsen et al. , 2005 Murakami et al. , 2007 Jackson et al. , 2006 Kung et al. , 2002 Kennett & Jackson, 2009 76 80 84 88 92 96 100 Fig. 2. Effect of Al and Fe on shear modulus ( G 0 ) and its pressure derivative ( G 0 '). (a, b) (Mg,Fe)O fp, and (c, d) Al-bearing MgSiO 3 pv. Stars correspond to G 0 or G 0 ' for representative iron/alumina content of X Mg = 79 mol% for fp and X Al 2 O 3 = 4 wt%. Lines show the best-fit trends. Blue and red lines in (a) and (b) indicate the trends of (Mg,Fe)O at HS and LS states of iron, respectively. 5.2 5.6 6.0 6.4 6.8 7.2 7.6 Shear Velocity (km/s) Temperature (K) Brown & Shankland, 1981 Shear wave velocity profile Lower Mantle Geotherm 30 40 50 60 70 80 90 100 110 120 130 Pressure (GPa) Pressure (GPa) 1600 1800 2000 2200 2400 2600 2800 3000 (a) (b) 30 40 50 60 70 80 90 100 110 120 130 V s T PREM Fp ( X Mg =0.79) Pyrolite ( X Pv =0.80) Best Fit ( X Pv =0.95) Pv ( X Mg =0.94, 4wt% Al 2 O 3 ) Fig. 3. Representative lower mantle geotherm (a) and calculated shear wave velocity profiles of fp (black lines) and pv (blue lines) (b) . Velocity profile for the pyrolite model is shown as the green line. Best-fit model to the PREM is shown as a red curve.
home
- JP
- EN