Polymerized 4-Fold Coordinated Carbonate Melts in the Deep Mantle

The deep carbon cycle and carbon reservoir in the Earth’s lower mantle are of great interest in high-pressure mineral physics since they are thought to have important contributions to the dynamical and chemical evolution of the solid Earth. A high-pressure silicate-carbonate, CaSiC2O7, has recently been discovered, but its stability in reactions with surrounding mantle materials remains unclear. To investigate this, we perform ab initio density functional calculations assuming several reactions with deep mantle mineral components. The calculated results show that this new class of carbon-bearing silicate compound is stabilized by the reactions of CaCO3 and MgCO3 with SiO2, CaSiO3 perovskite, and MgSiO3 bridgmanite at different pressures and inversely inhibits the stabilization of CaCO3, MgCO3, and CaC2O5 above 60 GPa. These results suggest that CaSiC2O7 represents an important host for carbon in the Earth’s deep mantle and influences the form in which carbon is stored and transported under lower mantle conditions.

Carbonates play a key role in the chemistry and dynamics of our planet. They are directly connected to the CO2 budget of our atmosphere and have a great impact on the deep carbon cycle. Moreover, recent studies have shown that carbonates are stable along the geothermal gradient down to Earth's lower mantle conditions, changing their crystal structure and related properties. Subducted carbonates may also react with silicates to form new phases. These reactions will redistribute elements, such as calcium (Ca), magnesium (Mg), iron (Fe) and carbon in the form of carbon dioxide (CO2), but also trace elements, that are carried by the carbonates. The trace elements of most interest are strontium (Sr) and rare earth elements (REE) which have been found to be important constituents in the composition of the primitive lower mantle and in mineral inclusions found in super-deep diamonds. However, the stability of carbonates in presence of mantle silicates at relevant temperatures is far from being well understood. Related to this, very little is known about distribution processes of trace elements between carbonates and mantle silicates. To shed light on these processes, we studied reactions between Sr- and REE-containing CaCO3 and Mg/Fe-bearing silicates of the system (Mg,Fe)2SiO4 - (Mg,Fe)SiO3 at high pressure and high temperature using synchrotron radiation based μ-X-ray diffraction (μ-XRD) and μ-X-ray fluorescence (μ-XRF) with μm-resolution in a laser-heated diamond anvil cell. X-ray diffraction is used to derive the structural changes of the phase reactions whereas X-ray fluorescence gives information on the chemical changes in the sample. In-situ experiments at high pressure and high temperature were performed at beamline P02.2 at PETRA III (Hamburg, Germany) and at beamline ID27 at ESRF (Grenoble, France). In addition to μ-XRD and μ-XRF, ex-situ measurements were made on the recovered sample material using transmission electron microscopy (TEM) and provided further insights into the reaction kinetics of carbonate-silicate reactions. Our investigations show that CaCO3 is unstable in presence of mantle silicates above 1700 K and a reaction takes place in which magnesite plus CaSiO3-perovskite are formed. In addition, we observed that a high content of iron in the carbonate-silicate system favours dolomite formation during the reaction. The subduction of natural carbonates with significant amounts of Sr leads to a comprehensive investigation of the stability not only of CaCO3 phases in contact with mantle silicates but also of SrCO3 (and of Sr-bearing CaCO3). We found that SrCO3 reacts with (Mg,Fe)SiO3-perovskite to form magnesite and gained evidence for the formation of SrSiO3-perovskite. To complement our study on the stability of SrCO3 at conditions of the Earth's lower mantle, we performed powder X-ray diffraction and single crystal X-ray diffraction experiments at ambient temperature and up to 49 GPa. We observed a transformation from SrCO3-I into a new high-pressure phase SrCO3-II at around 26 GPa with Pmmn crystal structure and a bulk modulus of 103(10) GPa. This information is essential to fully understand the phase behaviour and stability of carbonates in the Earth's lower mantle and to elucidate the possibility of introducing Sr into mantle silicates by carbonate-silicate reactions. Simultaneous recording of μ-XRD and μ-XRF in the μm-range over the heated areas provides spatial information not only about phase reactions but also on the elemental redistribution during the reactions. A comparison of the spatial intensity distribution of the XRF signal before and after heating indicates a change in the elemental distribution of Sr and an increase in Sr-concentration was found around the newly formed SrSiO3-perovskite. With the help of additional TEM analyses on the quenched sample material the elemental redistribution was studied at a sub-micrometer scale. Contrary to expectations from combined μ-XRD and μ-XRF measurements, we found that La and Eu were not incorporated into the silicate phases, instead they tend to form either isolated oxide phases (e.g. Eu2O3, La2O3) or hydroxyl-bastnasite (La(CO3)(OH)). In addition, we observed the transformation from (Mg,Fe)SiO3-perovskite to low-pressure clinoenstatite during pressure release. The monoclinic structure (P21/c) of this phase allows the incorporation of Ca as shown by additional EDX analyses and, to a minor extent, Sr too. Based on our experiments, we can conclude that a detection of the trace elements in-situ at high pressure and high temperature remains challenging. However, our first findings imply that silicates may incorporate the trace elements provided by the carbonates and indicate that carbonates may have a major effect on the trace element contents of mantle phases.

Water, in trace amounts, can greatly alter chemical and physical properties of mantle minerals and exert primary control on Earth’s dynamics. Quantifying how water is retained and distributed in Earth’s deep interior is essential to our understanding of Earth’s origin and evolution. While directly sampling Earth’s deep interior remains challenging, the experimental technique using laser-heated diamond anvil cell (LH-DAC) is likely the only method available to synthesize and recover analog specimens throughout Earth’s lower mantle conditions. The recovered samples, however, are typically of micron sizes and require high spatial resolution to analyze their water abundance. Here we use nano-scale secondary ion mass spectrometry (NanoSIMS) to characterize water content in bridgmanite, the most abundant mineral in Earth’s lower mantle. We have established two working standards of natural orthopyroxene that are likely suitable for calibrating water concentration in bridgmanite, i.e., A119(H2O) = 99 ± 13 μg/g (1SD) and A158(H2O) = 293 ± 23 μg/g (1SD). We find that matrix effect among orthopyroxene, olivine, and glass is less than 10%, while that between orthopyroxene and clinopyroxene can be up to 20%. Using our calibration, a bridgmanite synthesized by LH-DAC at 33 ± 1 GPa and 3,690 ± 120 K is measured to contain 1,099 ± 14 μg/g water, with partition coefficient of water between bridgmanite and silicate melt ∼0.025, providing the first measurement at such condition. Applying the unique analytical capability of NanoSIMS to minute samples recovered from LH-DAC opens a new window to probe water and other volatiles in Earth’s deep mantle.

A simple four‐parameter model for calculating thermoelastic properties of MgSiO3 perovskite is presented based on the Vinet model for static lattice and the Debye approximation for lattice vibration. The input parameters are the volume of the unit cell, V0, the bulk modulus, K0, its pressure derivative, K0′, and the Debye temperature, Θ0, in the static lattice at zero pressure. For V0, K0 and K0′ the theoretical values by Stixrude and Cohen [1993] are used and Θ0 is determined to reproduce the experimental value at ambient conditions, 980 K, by Akaogi and Ito [1993]. The resulting isobars are in good agreement with experimental data to 1300 K and 11 GPa by Wang et al. [1994], with those to 1200 K at 20 GPa by Utsumi et al. [1995], and with those to 1500 K and to 2000 K, respectively, by Kato et al. [1995] and Funawori et al. [1996] both at 25 GPa. Using the present equation of state together with the method for calculating adiabatic Lamé constants λS and μS for isotropic medium given in the present paper, density ρ, and sound velocities νp and νs of MgSiO3 perovskite under lower mantle conditions have been calculated where the constant‐entropy model is assumed with the temperature at the core‐mantle boundary being taken to be 3000 K. The results for ρ, νp, and νs are in agreement with the preliminary reference Earth model (PREM) within −2.4%∼ −3.7%, +3.3%∼+1.1%, and +0.8%∼−6.8%, respectively, over the lower mantle from 670 to 2891 km in depth. The calculated thermal expansivity under lower mantle conditions is in good agreement with that of the lower mantle estimated by Anderson [1982]. Using the present model with the parameters determined from experimental data at room temperature by Knittle and Jeanloz [1987], assuming Θ0 to be the same as that of MgSiO3 perovskite, thermoelastic properties of (Mg0.9, Fe0.1)SiO3 perovskite under lower mantle conditions have been calculated. The density becomes in much better agreement (+0.4%∼−0.8 %) with PREM and νp and νs remain almost unchanged from those of MgSiO3 perovskite.

The regularities of the joint genesis of diamonds and their paragenetic inclusions under lower mantle conditions are controlled by the melting relations at the liquidus of the multicomponent diamond-forming system. The boundary compositions of this system are evident from the generalized data on the analytical mineralogy of paragenetic inclusions in lower-mantle diamonds. The structure of the liquidus of the diamond-forming system was studied in a physicochemical experiment for P–T parameters typical of the depths of 670–800 km. The compositions of parental melts/solutions for diamonds and paragenetic inclusions correspond to the multicomponent system MgO–FeO–СаО–SiO2–MgCO3–FeCO3–CaCO3–Na2CO3–C. Its primary melting is controlled by the peritectic phase relations at the solidus, which was shown as a result of experimental studies of polythermal sections of the system during a study of their phase diagrams. Of key importance is the effect of the “stishovite paradox,” i.e., the peritectic reaction between the ultrabasic bridgmanite phase and melt with the formation of basic oxide associations of periclase–wustite solid solutions and stishovite. The peritectic reaction of bridgmanite is a fundamental feature of the diamond-forming system and determines the major peculiarity in its liquidus structure. Structure of the peritectic liquidus provides the physicochemical basis for the evolution of growth melts of diamonds and their paragenetic minerals. Based on the experimental data, we have constructed a fractional diagram of syngenesis of diamonds and inclusions clearly illustrating the solution–melt mechanism of diamond genesis and sequence of growth trapping of primary inclusions by diamonds under the lower-mantle conditions. The physicochemical factors of the genesis of diamonds and primary inclusions are agreed and generalized in a compositional diagram of lower-mantle diamond-forming media, and they provide a natural basis for the genetic classification of inclusions of rock-forming and accessory minerals in diamonds

Phase relationships in KLB‐1 peridotite, normal‐type mid‐oceanic ridge basalt (NMORB) and K‐rich basalt compositions were investigated experimentally from 28 to 143 GPa and 2050 to 3000 K. A laser‐heated diamond anvil cell with the synchrotron X‐ray diffraction was developed to perform in situ observations of samples at simultaneous high pressures and temperatures. We confirmed that the peridotite composition crystallized into an assemblage of Mg perovskite + Ca perovskite + magnesiowüstite at pressures below 97 GPa and that NMORB and K‐rich basalt compositions both crystallized as Mg perovskite + Ca perovskite + stishovite + an aluminous phase with a CaFe2O4‐type structure at pressures below 68 GPa. The orthorhombic Mg perovskite is stable in natural rock compositions at pressures less than 97 GPa, corresponding to about 2200 km depth. Phase transition of silica phase from stishovite to CaCl2‐type phase was observed at 78 GPa in NMORB composition. Moreover, we observed that NMORB crystallized in to an assemblage of CaIrO3‐type (Mg,Fe)SiO3 + Ca perovskite + α‐PbO2‐type silica + an aluminous phase with a calcium CaTi2O4‐type structure at 143 GPa, corresponding to the core‐mantle boundary depth. We measured the unit cell volumes of minerals in the NMORB composition at high P‐T and thereby estimated the density of subducted MORB to a depth of 1800 km in the lower mantle. In this depth regime, MORB is denser than model mantle densities, e.g., PREM and AK135. The estimated density of MORB may contribute to an understanding of dynamics of the subducted slab in the deep lower mantle.

High-temperature equation of state of CaSiO 3 perovskite and its implications for the lower mantle

Carbonate, the major carbon reservoir on Earth’s surface, can enter into the mantle by subduction. However, evidence for recycled surficial carbonates in the deep mantle is still scarce. Ocean island basalts from Cook-Austral islands and St. Helena Island, widely called HIMU basalts because of their high μ = 238U/204Pb sources, are thought to be fed by mantle plumes originating in the lower mantle. Here we report exceptionally high δ66Zn values (δ66Zn = 0.38 ± 0.03‰) of these HIMU lavas relative to most published data for oceanic basalts (δ66Zn = 0.31 ± 0.10‰), which requires a source contributed by isotopically heavy recycled surficial carbonates. During subduction of the oceanic lithosphere, melting of mixed surficial carbonates and basaltic crust in the deep mantle generates carbonatite melts, which metasomatizes the nearby mantle and the resultant carbonated mantle ultimately evolves into a high-δ66Zn HIMU source. High-δ66Zn signatures of HIMU basalts, therefore, demonstrate that carbonates can be transported into Earth’s deep mantle.

The presence of water in lower mantle minerals is thought to have substantial effects on the rheological properties of the Earth's lower mantle in what is generally known as “hydrolytic weakening”. This weakening will have profound effects on global convection, but hydrolytic weakening in lower mantle minerals has not been observed experimentally and thus the effect of water on global dynamics remains speculative. In order to constrain the likelihood of hydrolytic weakening being important in the lower mantle, we use first principles methods to calculate the partitioning of water (strictly protons) between mineral phases of the lower mantle under lower mantle conditions. We show that throughout the lower mantle water is primarily found either in the minor Ca-perovskite phase or in bridgmanite as an Al3+–H+ pair. Ferropericlase remains dry. However, neither of these methods of water absorption creates additional vacancies in bridgmanite and thus the effect of hydrolytic weakening is likely to be small. We find that water creates significant number of vacancies in bridgmanite only at the deepest part of the lower mantle and only for very high water contents (>1000 ppm). We conclude that water is thus likely to have only a limited effect on the rheological properties of the lower mantle.

Strong constraints can be placed on the origin of heterogeneity of seismic wave velocities and density if the observed ratios of various parameters are compared with mineral physics predictions. They include the shear to compressional wave velocity heterogeneity ratio,Rs/p≡ δ logVs/δ logVp, the bulk sound to shear wave velocity heterogeneity ratio,Rϕ/s≡ δ logVϕ/δ logVs, and the density to velocity heterogeneity ratio,Rρ/s,p≡ δ log ρ/δ logVs,p. Using mineral physics considerations, we calculate these ratios in the lower mantle corresponding to the thermal and chemical origin of velocity and density heterogeneity. Both anharmonic and anelastic effects are considered for thermal origin. Anharmonic effects are estimated from the theoretical calculations as well as from laboratory measurements which show a marked increase inRs/pwith pressure from ∼1.5 to ∼2.1 in the lower mantle. Such a trend is marginally consistent with seismological observations showing an increase inRs/pwith depth (from ∼1.7 to ∼3.2 in the lower mantle). However, anharmonic effect alone cannot explain inferred lowRρ/s(<0.2–0.4), and the values ofRs/ppredicted from anharmonic effects are systematically smaller than seismological observations. Anelastic effects must be included to account for these geophysical observations. A new formulation of anelastic effect is developed in which the role of nonlinear temperature dependence of anelasticity is included. When anharmonic and anelastic effects are combined, thermal effects on velocity and density can account for most of the geophysical observations in Earth's lower mantle. Effects other than temperature variations (such as chemical heterogeneity) are required only in the very deep portions of the mantle (deeper than ∼2000–2300 km). A puzzling observation is very large values ofRs/p(>2.7) and corresponding negative values ofRϕ/s(andRρ/s) in the deep lower mantle which cannot be accounted for by thermal or simple chemical heterogeneity such as the heterogeneity in the Fe/(Fe + Mg) and/or Mg/(Mg + Si) ratios. Possible causes of anomalies in this region are discussed, including the role of anisotropy and a combined effect of heterogeneity in Fe and Ca content.

Carbonatic components of parental melts of the deeper mantle diamonds are inferred from their primary inclusions of (Mg, Fe, Ca, Na)-carbonate minerals trapped at PT conditions of the Earth’s transition zone and lower mantle. PT phase diagrams of MgCO3–FeCO3–CaCO3–Na2CO3 system and its ternary MgCO3–FeCO3–Na2CO3 boundary join were studied at pressures between 12 and 24 GPa and high temperatures. Experimental data point to eutectic solidus phase relations and indicate liquidus boundaries for completely miscible (Mg, Fe, Ca, Na)- and (Mg, Fe, Ca)-carbonate melts. PT fields for partial carbonate melts associated with (Mg, Fe)-, (Ca, Fe, Na)-, and (Na2Ca, Na2Fe)-carbonate solid solution phases are determined. Effective nucleation and mass crystallization of deeper mantle diamonds are realized in multicomponent (Mg, Fe, Ca, Na)-carbonatite–carbon melts at 18 and 26 GPa. The multicomponent carbonate systems were melted at temperatures that are lower than the geothermal ones. This gives an evidence for generation of diamond-parental carbonatite melts and formation of diamonds at the PT conditions of transition zone and lower mantle.

Shock-induced phase transitions in siderite up to 90 GPa and implications for deep carbon cycle

Al2O3 is estimated to total 4 to 5 mol% in all mantle compositional models, and is believed to be incorporated into (Mg,Fe)SiO3‐perovskite at lower mantle conditions. Using synchrotron X‐ray diffraction, we have measured the 300 K equation of state of a perovskite with XAl=Al/(Al+Mg+Si)=.077, up to 32 GPa. A least squares refinement of two independent data sets yields the following equation of state parameters V0=163.52(3) ų, K0=229(4) GPa, and K0=2.5(4). These values allow us to assess the most probable substitution mechanism for aluminium into perovskite at pressure conditions of the shallow lower mantle. Al3+ is likely to be incorporated in place of Si4+ in the octahedral site of perovskite, hence requiring the creation of oxygen vacancies for charge balance. As a consequence, aluminous perovskites may have a high affinity for water. The enhanced compressibility of aluminous perovskite certainly has also strong geophysical and geochemical implications, as it may revive perovskite‐rich lower mantle models.

δs, a measure of the sensitivity of the adiabatic bulk‐modulus to isobaric density changes, controls lateral variations of P‐wave velocities. To estimate the effect of pressure on δs, we have performed lattice dynamic calculations of the cubic alkaline earth oxides. We used a many‐body parametric Hamiltonian that reproduces the available laboratory data on thermoelastic properties. The calculations were carried out for P‐T conditions that correspond to those of the lower mantle. We find that δs decreases from 4±1 at standard conditions to ≤ 2 at lower mantle conditions. The latter value is consistent with D.L. Anderson's estimates based on geophysical data.With δs ≤ 2, and a corresponding parameter for shear modulus not decreasing with compression, no partial melting is required to account for the observation that ΔlnVs/ΔlnVp ≥ 2 in the lower mantle. Although more data and calculations on additional minerals are needed to draw any firm conclusions, our results indicate that the relatively small lateral variations of lnVp, when compared to lnVs, are due entirely to compressional effects.

Lower mantle electrical conductivity based on measurements of Al, Fe-bearing perovskite under lower mantle conditions