The Earth's Lower Mantle

The Earth’s lower mantle (extending from a depth of 660–2900 km) comprises more than half (~56%) of the total volume of the Earth. Most of the Earth’s geodynamic processes stem from the mantle, which is reflected in the Earth’s structure. However, its composition and structure are not yet well known.

The most popular model of the lower mantle is the Preliminary Reference Earth Model (PREM), derived from seismic observations assuming the pyrolitic composition of the lower mantle. The uppermost part of the lower mantle (~660–770 km deep) has a steep velocity gradient, reflecting the mineral structure transformation from ringwoodite to bridgmanite and ferropericlase, after which gradual increase in both the compressional velocity (V _p ) and shear velocity (V _s ) reflects the near adiabatic compression of mineral phases. The adiabatic geothermal gradient within the upper mantle decreases with increasing depth without phase transitions. Subducting lithospheric slabs may significantly cool temperature profiles, particularly in the upper part of the lower mantle. However, results of experiments on the density of natural peridotite, performed within the range of entire lower-mantle pressures along the geotherm, demonstrated their significant mismatch with the PREM density model. This implies that the upper and the lower mantle must have different chemical compositions, i.e. the mantle is chemically stratified, with the inference of a non-pyrolitic composition of the lower mantle. The diapason of oxygen fugacity within the entire sequence of lower-mantle region may reach ten logarithmic units, varying from below the IW buffer to the FMQ buffer values.

There are three major sources of information about the composition of the lower mantle: high P–T experiments, theoretical calculations, and geological observations. Experimental data, based on the use of diamond-anvil cell technique (DAC), and theoretical calculations demonstrate that silicates, occurring in the upper mantle and the transition zone, are replaced by predominantly perovskitic assemblage in the lower mantle. Depending on the starting substrate composition, two mineral associations should occur at pressures corresponding to the lower-mantle conditions: ultramafic (bridgmanite + CaSi-perovskite + ferropericlase) and mafic (bridgmanite + CaSi-perovskite + ferropericlase + silica + Al-phase). Both iron-containing lower-mantle minerals, bridgmanite and ferropericlase, should be magnesium-rich. In recent decades, lower-mantle minerals were found as inclusions in diamonds from Brazil, Guinea, Canada, Australia and South Africa. They confirm the presence of ultramafic, mafic and carbonatitic mineral associations. Geological samples differ notably from the lower mantle compositions suggested on the basis of experimental and theoretical data for the pyrolitic composition. First, ferropericlase is the most common in the lower-mantle ultramafic association (averaging 55.6%), while bridgmanite comprises only 7.5%, about ten times lower than has been suggested (c. 70–74%) in the lower mantle. Second, silica inclusions were identified in all sets of lower-mantle minerals observed in diamond from all regions and areas. Third, wide variations in ferropericlase compositions, reaching an iron index of up to fe = 0.64 were observed. Minerals from the ultramafic association overwhelmingly predominate in the lower mantle samples; only two samples of mafic mineral phases, phase Egg and δ−AlOOH are found to date.

The juvenile ultramafic lower mantle is composed of the mineral association: bridgmanite + ferropericlase + CaSi-perovskite + free silica. Bridgmanite, with mg = 0.84–0.96 forms two compositional groups: low-Al and high-Al. High-Al bridgmanite is richer in Fe and infers the characteristic of deeper layers in the lower mantle. The crystal structure of bridgmanite is orthorhombic through the entire lower mantle down to the D″ layer. The chemical composition of ferropericlase is different from the predicted one with the magnesium index mg varying from 0.36 to 0.90. Low-Fe ferropericlase has a cubic rocksalt structure, which is stable throughout the entire lower mantle. Iron contents in both ferropericlase and bridgmanite and ferropericlase increase with pressure indicating the increase of Fe concentration in the lower mantle with depth. CaSi-perovskite is remarkably clean in its chemical composition with only minor admixtures of Ti, Al and Fe, but is enriched in trace elements. CaSi-perovskite within the lower mantle has a cubic structure which at low temperatures (in subsolidus conditions) may transfer into a tetragonal or orthorhombic structure. The presence of free silica in the lower mantle was identified in geological samples from all areas. In the upper part of the lower mantle it is represented by stishovite; at a depth of 1600–1800 km stishovite transforms into the CaCl₂-structured polymorph; and at the CMB, into a α-PbO₂ phase seifertite. In addition to the major minerals, a variety of other mineral phases occurs in the lower mantle: Mg–Cr–Fe, Ca–Cr and other orthorhombic oxides, jeffbenite, ilmenite, native Ni and Fe, moissanite and some others.

Mafic mineral association in the lower mantle is subordinate to the ultramafic one. It includes bridgmanite, CaSi-perovskite, SiO₂ and anhydrous aluminous phases. The former three are the same as that observed in the ultramafic association; but their chemical compositions differ from those in the ultramafic association, mainly in the significant enrichment of Al. Among aluminous phases the NAL phase occurs at low-pressure conditions and is replaced by a CF phase at a depth of 800–1200 km depth. NAL phase is also concentrated in Na and K, while CF phase does not contain K. The partition coefficient of aluminium between bridgmanite and the NAL phase vary from 0.10 to 0.26, demonstrating that the Al enrichment in bridgmanite occurs at the expense of the Al decrease in the NAL phases. The Al concentration in the CF phase remains constant and the Al concentration in bridgmanite, after reaching maximal concentrations (24–25 wt%) with disintegration of the NAL phase, remains constant as well. In addition to the major minerals, phase Egg, δ-AlOOH, and a series of dense hydrous magnesium silicates (DHMS) are expected to be present in the mafic association. Among these DHMS, Phase D and Phase H are most likely to occur in the subducting slabs within the lower mantle. Some of these minerals (phase Egg and δ-AlOOH) are observed in natural geological materials; the others have only been synthesized in laboratory experiments.

In addition to ultramafic and mafic associations, a primary natrocarbonatitic association occurs in the lower mantle. To date, it was identified as inclusions in diamonds from the Juina area, Mato Grosso State, Brazil. It comprises almost 50 mineral species: carbonates, halides, fluorides, phosphates, sulphates, oxides, silicates, sulphides and native elements. In addition, volatiles are also present in this association. Among oxides, coexisting periclase and wüstite were identified, pointing to the formation of the natrocarbonatitic association at a depth greater than 2000 km. Some iron-rich (Mg,Fe)O inclusions in diamond are attributed to the lowermost mantle. The initial lower-mantle carbonatitic melt formed as a result of low-fraction partial melting of carbon-containing lower-mantle material, rich in P, F, Cl and other volatile elements at the core–mantle boundary. During ascent to the surface, the initial carbonatitic melt dissociated into two immiscible parts, a carbonate-silicate and a chloride-carbonate melt. The latter melt is parental to the natrocarbonatitic lower-mantle association. Diamonds with carbonatitic inclusions were formed in carbonatitic melts or high-density fluids.

Diamond contains mineral inclusions of all three lower-mantle associations, juvenile ultramafic, mafic and carbonatitic; it is also an accessory mineral in all these associations. While the first two associations coexist with diamond, the carbonatitic association is a parental medium for the lower-mantle diamond. Physical and chemical characteristics of lower-mantle diamond differ from ones of lithospheric origin. Most of the lower-mantle diamonds are ‘nitrogen-free’ Type II variety. The others are usually low-nitrogen stones with the average nitrogen aggregation rate of 94%. The high proportion of nitrogen-aggregated diamonds suggests that they had a prolonged residence in the lower mantle under high-T conditions, which resulted in an almost complete transformation of single-atomic and paired nitrogen centers into polyatomic complexes. In contrast to lithospheric diamonds, almost all analyzed lower-mantle ones (70–89%) have noticeable levels of hydrogen centers (up to 4–6 cm⁻¹). The isotopic compositions of lower-mantle diamonds are located within a narrow range: from −5.45 to −1.26‰ δ ¹³C VPDB, with an average value of −4.36‰ ± 2.28‰ (2σ). It may be considered as the juvenile lower-mantle carbon isotopic composition. The isotopic composition of nitrogen for lower-mantle diamonds is located within a close range, from −5.2 to −1.0‰ δ ¹⁵N_atm, with an average value of δ ¹⁵N_atm = −3.00‰ ± 2.37‰ δ ¹⁵N_atm. Lower-mantle diamond was formed in carbonate-oxide parental melts and fluids, which experienced fractional crystallization with the decrease of temperature and changes in the melt composition. The most important role in this process belongs to the carbonate component in the parental melt.

Extremely high P–T conditions in the lower mantle affect some basic physicochemical properties of elements in minerals. One of the most important is a spin transition, which significantly changes the properties of iron-containing minerals in lower-mantle associations. The iron high-spin to low-spin transition in ferropericlase occurs at depths of 1000–1500 km. It is accompanied by the reduction of the unit cell volume and corresponding seismic velocity variations. However, the spin crossover in ferropericlase is a seismologically transparent transition owing to its gradual nature. Incorporation of iron in bridgmanite is more complex than in ferropericlase. Fe²⁺ in the A site remains in the HS state at all mantle conditions. In contrast, Fe³⁺ undergoes a spin transition in the entire range of lower-mantle conditions. The iron spin transition in bridgmanite does not change the existing seismological model down to the D″ layer. Under high-pressure conditions, chemical elements can obtain dramatic new properties in the lower mantle, including the formation of unexpected crystal structures and completely new counter-intuitive compounds. Some of these compounds are confirmed experimentally. Most of these transformations may occur within the lower mantle in specific compositions, which may produce only accessory mineralization. However, they may play a significant role in the Earth’s balance of light elements, in the formation of the primordial carbonatitic association, and influence some major lower-mantle phases, such as periclase with the formation of magnesium peroxide MgO₂.

The Dʺ layer is a ~200 km layer at the bottom of the lower mantle (at ~2700–2900 km depth). It has low S-wave velocity gradients and increased scatter in travel times and amplitudes. Compositionally, there are two sources for the Dʺ layer: the oxide lower mantle and the outer core. Oxides from the lower mantle experience phase and physical changes within the Dʺ layer; the outer core delivers the metallic part to the Dʺ layer composition. The transformation of bridgmanite in post-perovskite, creating a 1–1.5% shear velocity increase, is the major effect for distinguishing of the Dʺ layer. Post-perovskite may incorporate 1–2 wt% H₂O and thus may store significant amounts of hydrogen within the Dʺ layer. The transformation of bridgmanite in post-perovskite is accompanied with the transition of the orthorhombic CaCl₂-structured SiO₂ in seifertite, α-PbO₂-structured SiO₂, resulting in a slight decrease in bulk sound speed by ~0.4% and shear wave decrease. Iron-rich liquid metal from the outer core (containing 5–10% light elements), namely C, N, O and Si infiltrates into the lowermost mantle and forms a series of native Fe ⁰, iron carbides and nitrides, and silicon carbide. Of particular importance is the presence of Fe₇C₃ and Fe₇N₃. When in association with diamond, these are the first solidus phases crystallizing from the metallic liquid in the D″ layer. The presence of iron nitrides in the Dʺ layer is closely related to their suggested presence in the inner core and helps to solve the problem of ‘missing nitrogen’ in the Earth’s nitrogen balance.

Over the last 20 years, global seismology has made significant progress in mapping the deep interior of the Earth. Tomographic studies identified variations in lower-mantle chemistry and phase transitions with depths of observed seismic heterogeneities occupying the entire range of the lower mantle. Three major zones of seismic heterogeneities can be outlined. The upper (shallow) zone from 660–1300 km includes ~70% of all heterogeneities, observed almost equally near subduction zones and beneath the tectonic plates. The middle zone, from 1300 to 1900 km, includes ~20% of all heterogeneities, which are observed entirely near subduction zones. The lower zone, from 1900 km to the border of D″ layer at 2700 km includes only a small number of heterogeneities. The deepest seismic heterogeneities, identified within the central parts of the Eurasian and North American plates, are located at depths of ~2630 km and ~2400 km. No correlation between the observed seismic heterogeneities and major mineral phase transitions and spin crossover were identified. The seismic heterogeneities, most likely, reflect local and regional chemical variations within the lower mantle.