High-temperature Compression Experiments of Ferropericlase and the Temperature Effect on Iron Spin Transition
The precise density profile of the mantle based on laboratory measurements is important for understanding the composition of the mantle. Iron-bearing lower mantle minerals have been proposed to undergo the spin transition at certain P-T conditions. One of the remarkable effects of the spin transition is on the density since this transition is accompanied by a volume shrinkage. Previous experimental studies have been mostly focused on the spin transition at room temperature. In order to collect the density data and understand the nature of the spin transition of an iron-bearing lower mantle mineral at high temperatures, we operated compression experiments of ferropericlase (Fp) with a composition of (Mg0.81,Fe 0.19)O with in-situ X-ray diffraction method at three different temperatures. In a diamond anvil cell (DAC) which can generate the lower mantle pressures, we produced high temperatures by an externally- or laser-heating system. In the externally-heated DAC, we collected the volume data from 19 to 85 GPa at a constant temperature of about 865 K. In the laser-heated DAC, we developed a new system for in-situ X-ray diffraction experiment, which is a membrane system. In this system, we are able to regulate the gas pressure in the membrane of the DAC and, therefore, compress the sample at a high temperature during the laser-heating. We collected the volume data at a high temperature of 1600 K from 20 to 120 GPa. Room-temperature experiments with a laser-annealing technique were also carried out on the same material. Anomalous volume reductions that cannot be explained by normal compression behavior were observed at 65-100, 60-82, and 58-64 GPa at 1600, 865, and 300 K, respectively. These volume reductions are likely related to the spin transition of ferrous iron in Fp. Thus, the spin transition pressure interval expands with increasing temperature. The observed density changes across this spin transition at high temperatures and 300 K are about 1.5% and 0.6%, respectively. We will discuss the temperature effect of the spin transition in Fp from our compression experiments at three different temperatures.
Potential effects of the spin crossover transition in ferropericlase on mantle velocities
The thermoelastic properties of ferropericlase Mg1-xFexO (x = 0.1875) throughout the iron spin crossover have been investigated by first principles at Earth's lower mantle conditions. The transition has important consequences for the elasticity such as a substantial bulk modulus reduction. At room temperature the transition is quite sharp in pressure but broadens with increasing temperature. Along a typical geotherm the transition should occur across most of the lower mantle with a more significant bulk modulus reduction around 1400-1600 km depth. This crossover transition is yet another ingredient, in addition to changes in temperature, chemical composition, and mineralogy that can introduce velocity heterogeneities in the mantle. We compare predictions of the effect of this spin transition alone on the elastic properties of homogeneous aggregates with the elastic properties of the lower mantle extracted from seismic tomography and PREM.
Mapping of thin ultra-low velocity zones at the CMB
Ultra-Low Velocity Zones (ULVZ) are localized features at the core mantle boundary (CMB) characterized as thin layers (<40 km), with strong reductions in seismic velocities (in the range of 10 to 30 %) for both P- and S- waves. Restrictions in available source-receiver combinations and few seismic probes sensitive to ULVZ structure leave large patches of the CMB unprobed for ULVZ existence. Probed areas show evidence for patchy ULVZ with many regions not showing any evidence for ULVZ in the seismic data. Lack of evidence for ULVZ structure either indicates that it is absent or that it is too small/thin (or mild in properties) to be detected with current seismic probes. The seismic properties of ULVZ appear most consistent with partially molten material which could consist either of mantle material or material chemically distinct from the bulk of the mantle. Recent high resolution waveform studies also find evidence that the ULVZ material is denser than the surrounding mantle which raises questions about the stability of ULVZ. On the other hand, geodynamical calculations suggest that the ULVZ material can remain stable in distinct pockets even with large density increases in exceed of 10 %. However, viscosity plays an important role: if the ULVZ viscosity is significantly lower than the surrounding mantle, other mechanisms may be needed to keep ULVZ material in isolated pockets, and preventing it from draining out of the mantle and forming a thin layer on the CMB. We will present recent seismological evidence for a preference of ULVZ for locations towards the edges of the large low shear velocity province beneath the Pacific with little ULVZ evidence outside of this province. Further evidence indicates ultra-thin (approximately 2 km thick) ULVZ in close proximity to regions showing seismic evidence for approximately 10 km thick ULVZ. High resolution geodynamical modeling shows that dense thermo-chemical piles (such as found beneath the central Pacific and southern Africa) might play an important role in ULVZ dynamics, including stabilizing ULVZ into lenses or ridges towards the perimeter edges of the piles.
D" Anisotropy Beneath the Caribbean, Central America and the East Pacific
Whilst the majority of the Earth's lower mantle appears to be relatively homogeneous, by contrast the few hundred kilometres above the core-mantle boundary (CMB) are host to a region of probable large chemical and thermal heterogeneity. Seismic observations of this region---known as D"---include a large increase in S-wave velocity that can vary in depth laterally over distances of <~100~km and significant seismic anisotropy (the variation of wavespeed with direction). The most recent candidate to explain these features in D" (including its anisotropy and bounding discontinuity) is the experimentally observed transformation of MgSiO3-perovskite to a post-perovskite structure at near CMB pressures and temperatures. As the phase change has a positive Clapeyron slope, regions where the geotherm is colder than average at the CMB---such as areas beneath long-term subduction---should show evidence of such a discontinuity and, depending on the alignment of mantle minerals or other structure, should also exhibit seismic anisotropy. We study the D" region beneath the Caribbean, Central America and the east Pacific using S and ScS phases mainly from deep-focus earthquakes with magnitude >~Mw~5.5 and depths >~550 km. Our method allows the incorporation of previous estimates of source-side upper mantle anisotropy, and by comparing the splitting parameters of the two phases (thus correcting for anisotropy in the upper mantle below the receiver), we obtain measurements of splitting in ScS alone; hence measuring the anisotropy in the lowermost mantle. The S and ScS phases are detected on around 450 seismic stations in Canada and the US (including Hawaii), yielding over 270 measurements of anisotropy in D". The measurements cover an area ∼4,000~km by ∼2,000~km centred on the CMB beneath Central America, and exhibit ∼1% S-wave anisotropy. In the Caribbean, they show a small but detectable departure from the first-order transverse isotropy with a vertical axis of symmetry (VTI) which can be explained as the same but with a tilted axis of symmetry (TTI). Here this dips a few degrees to the west; beneath Central America it dips to the east. Previous waveform studies agree with our results (e.g., Maupin et al., JGR, 2005). Beneath the east Pacific, where global S-wave models show a much less positive shear velocity anomaly, measurements show a significant degree of TTI, probably dipping by ∼30° to the east or southeast. Our interpretation (similarly to previous studies) of these features proposes that this is a result of the dynamics of the interaction of slab material with that already present at the base of the mantle, leading to deformation into 'ridges' aligned roughly perpendicular to the direction of palaeo-subduction over short scales (∼100 km and less) and the subsequent alignment of the crystals, melt pockets or other features which give rise to the TTI.
Re-access the D" discontinuity beneath the Cocos Plate: Finite difference modeling of USArray data
For the last two decades, seismic waveform modeling, stacking and migration have revealed a heterogeneous nature of the D" beneath the Cocos Plate in Central America. Previous estimates on the shear velocity jump across the D" vary from 0.5% to 3% over a length scale of a few hundred kilometers, assuming a 1-D wave propagation. In addition, there are increasingly accumulated evidence showing a large topographic offset in the D" of about 100 km in the region. However, it is not clear if 1-D wave propagation can reconcile such a rapid topographic offset and diverse estimates of shear velocity jump across the D" in a self-consistent manner. In this study, I collect high quality USArray data sampling the D" beneath the Cocos Plate and discuss waveform behavior in the context of 2-D wave propagation. I show that lateral variations in the strength and timing of the Scd arrival exist and it is a natural consequence of a strong (e.g. shear velocity jump of about 3%) but rough D" discontinuity (e.g. a topograhic offset of about 100 km). A weak (e.g. shear velocity jump of about 1.5%) but rough D" is not capable of reproducing the amplitude of these arrivals. A weak and flat D" can certainly reproduce the amplitude and timing of some weak Scd arrivals, but it is not consistent with the presence of strong Scd arrivals and topographic offset mapped from modeling, migration and imaging. Finally, I note that shear velocity jump estimated from 2-D wave propagation only serves as a lower bound if the D" topography is 3-D and changing over a length scale less than a few hundred kilometers in the out-of-plane direction.
Thermal and Elastic Structure of the Deep Mantle Inferred from Advanced Mantle Circulation Models with Pyrolite Composition
Arguably the most profound contribution by geodynamicists on constraining deep mantle structure and dynamics is to arrive at a better understanding of radial and lateral temperature heterogeneity in this remote region of our planet. Here we review the current understanding of lateral mantle temperature variations. Starting from simple scaling arguments on plume excess temperature which involve mantle non-adiabaticity and steeper adiabatic gradients within hot mantle upwellings we infer the existence of substantial lateral temperature heterogeneity in the deepest mantle. The lateral deviations from the mean are likely to exceed 1000 K within upwelling plumes and cold downwellings, suggesting a total thermal heterogeneity on the order of 2000 K in the lowermost mantle. We explore the dynamic consequences of this observation within advanced mantle circulation models capable of fully resolving the vigorous thermal regime of global mantle flow. Coupling the models to published thermodynamically self-consistent simulations of mantle mineralogy in the pyrolite composition, moreover, allows us to arrive at quantitative predictions on the amount of elastic heterogeneity expected for simple isochemical whole mantle flow. These results are in excellent agreement with a number of tomographic inferences of elastic lower mantle heterogeneity, both for P and S-wave velocity, including the observation of sharp lateral gradients which have been used to advance a role for chemical mantle heterogeneity. Our results are further supported by applying a tomographic filter, corresponding to seismic model S20RTS, to our mantle circulation models, and by exploring the geoid and True Polar Wander implied by our simulations. While we cannot preclude the existence chemical mantle variation, we conclude that their contribution must be minor and that isochemical whole mantle flow is agreement with a wide range of observational constraints on deep earth structure and dynamics.
High-resolution, tomography-based modelling of convective flow and deformation in the lowermost mantle.
Studies of seismic anisotropy in the deep mantle provide relatively sparse, but fundamentally important in-situ constraints on mantle flow and deformation patterns in the lowermost mantle. To date, tomography-based numerical investigations of the convective flow in the deep mantle have used long wavelength global tomography models (e.g. Ritsema et al. 1999), that resolve structures with scale lengths generally in excess of 1000 km. In addition, these previous models do not provide explicit constraints on non-thermal (e.g. compositional) heterogeneity that is known to have an important impact on mantle convection dynamics. Furthermore, the spatial resolution of these tomography models may be insufficient to establish a clear connection between the local inferences of deep-mantle deformation as seen in seismic anisotropy studies and the predictions derived from the tomography-based flow modelling. Substantial progress has recently been made in deriving seismic tomography models that approach the horizontal resolution needed to address these challenges. In this study we focus on the geodynamic implications of a new joint inversion of combined global seismic and surface geodynamic data sets, in which mineral physical constraints on mantle thermal properties are also included (Simmons et al. 2009). These tomographic inversions yield 3-D distributions of mantle density anomalies that include both thermal and compositional heterogeneity and therefore allow us to explicitly incorporate the stabilising effect of compositional buoyancy in the lower mantle. We employ these high resolution inferences of mantle density heterogeneity to provide new insights into the pattern and amplitude of the convective flow in the lowermost mantle and the implications for the associated flow-induced anisotropy.
Velocity Changes Across the High-Spin Low-Spin Transition in Ferropericlase at High Pressures up to 61 GPa
An electronic transition in Fe, from a high-spin state to low-spin state is promoted by pressure and has been observed in the Fe-bearing lower mantle minerals ferropericlase and silicate perovskite. This phase transition is accompanied by a volume change and presumably by changes in the elastic properties. Previous light scattering measurements have suggested that there is softening of the bulk and shear moduli in the mixed phase region across the HS-LS transition in ferropericlase. The softening was reported to be evident as decreased Vp and Vs within the mixed phase region, but has thus far not been verified by independent experiments. The electronic spin transition of iron in FP can occur above 40 to 90 GPa at room temperature, depending on the Fe content. Its effect on physical properties may profoundly affect our current understanding of lower mantle average composition and heterogeneity. Al will be incorporated into ferropericlase in realistic lower-mantle mineral assemblages, and could cause significant changes of its physical properties. However, the effect of Al on the HS-LS transition and properties of ferropericlase are unknown. We have performed Brillouin scattering experiments on Al-bearing ferropericlase at high pressures up to 61GPa. At ambient condition, incorporation of aluminum into ferropericlase results in a slight drop in the shear modulus, and no observed change in bulk modulus. Our results show a "softening" effect of the spin transition on the bulk modulus of Al-bearing ferropericlase at pressures over 40 GPa. This suggests that the incorporation of aluminum increases the pressure of spin transition. Our high pressure single-crystal velocity data shows that softening is most pronounced in the elastic modulus C12, which decreases significantly within transition region with mixed spin states. C11 is also likely to soften, although the effect appears to be less pronounced than for C12, whereas C44 does not display any softening within the pressure range of our observations. The shear modulus and Vs do not soften across the transition, within the resolution of our measurements.