The mantle is the thick layer of predominantly solid rock between Earth’s outer crust and its metallic core. Extending from a few kilometres beneath parts of the ocean floor to approximately 2,890 kilometres below the surface, it accounts for about 84% of Earth’s volume and 67% of its mass. Although mantle rocks behave elastically over the short intervals during which seismic waves pass through them, they deform and flow over geological time. This combination of solidity and slow motion distinguishes the mantle from both the rigid surface plates and the liquid outer core.
Mantle circulation transfers heat from Earth’s interior, participates in plate tectonics, and supplies the material from which most magmas and much of the crust ultimately form. It also exchanges water, carbon, and other chemical components with the surface. Knowledge of the mantle comes from seismic observations, mantle-derived rocks, laboratory experiments at high pressures and temperatures, and physical models. These approaches reveal a chemically and dynamically heterogeneous interior whose mineral structures change profoundly with depth.
Boundaries and internal divisions
The mantle’s upper boundary is the Mohorovičić discontinuity, usually shortened to the Moho. Across this boundary, seismic velocities generally increase as relatively silica-rich crustal rocks give way to denser mantle rocks. Beneath ordinary oceanic crust, the Moho lies approximately 6–7 kilometres below the seafloor; beneath continents, its depth is commonly 30–50 kilometres, increasing to more than 70 kilometres beneath some major mountain belts. These are typical values rather than fixed limits. The seismic boundary can also differ locally from the precise chemical boundary, particularly where alteration has changed the properties of uppermost mantle rocks.
The lower boundary is the core–mantle boundary, about 2,890 kilometres deep. Here solid silicate rock overlies liquid iron-rich alloy. The contrast is pronounced: standard reference models place density just above the boundary near , compared with roughly immediately below it. The disappearance of transmitted shear waves in the outer core, together with the refraction and reflection of compressional waves, identifies this boundary. It is simultaneously a major chemical interface, a mechanical boundary, and a region of heat exchange.
The mantle is conventionally divided into an upper mantle extending to approximately 660 kilometres and a lower mantle beneath it. Within the upper mantle, the interval between about 410 and 660 kilometres is called the transition zone. These depths correspond primarily to pressure-induced changes in mineral structure rather than to global changes in bulk chemical composition. Another discontinuity near 520 kilometres is detectable in many regions, although it is less consistently expressed. Local temperatures, chemical differences, and the presence of descending slabs cause these boundaries to depart from their nominal depths.
A separate classification concerns mechanical behavior. The lithosphere consists of the crust and the mechanically strong uppermost mantle. Beneath it, the asthenosphere is relatively weak and accommodates deformation associated with plate motion. These divisions do not coincide with the upper-mantle and lower-mantle boundaries. Oceanic lithosphere thickens as it cools and ages, while continental lithosphere varies greatly; beneath some ancient cratons, chemically depleted mantle roots extend to depths exceeding 200 kilometres.
The asthenosphere is not a continuous ocean of magma. Its reduced strength chiefly reflects high temperature and the effects of grain size, water content, and deformation mechanisms; small amounts of melt can contribute in some settings. Similarly, the seismic low-velocity zone commonly associated with it is not a uniquely defined global layer. A chemical boundary, a seismic discontinuity, and a mechanical boundary describe different properties and need not occupy the same depth.
Chemical composition and mineral transformations
The mantle consists mainly of silicate minerals containing magnesium and iron. Oxygen, silicon, magnesium, and iron dominate its elemental composition, with smaller but important amounts of calcium and aluminium. Its composition differs from the crust, which is generally richer in silica and in elements preferentially concentrated into melts. Most accessible upper-mantle samples are varieties of peridotite, an ultramafic rock dominated by olivine together with orthopyroxene, clinopyroxene, and an aluminium-bearing mineral such as spinel or garnet.
Peridotite is a rock family rather than a single fixed composition. Lherzolite contains appreciable amounts of both major pyroxenes and can represent comparatively fertile mantle capable of producing basaltic melt. Harzburgite contains little clinopyroxene and commonly records previous melt extraction. Dunite is overwhelmingly olivine-rich and may form through extensive melting or reactions between migrating melt and surrounding rock. These variations preserve evidence of mantle processing rather than requiring entirely separate primordial layers.
With increasing pressure, atoms rearrange into denser crystal structures. Near 410 kilometres, olivine transforms into wadsleyite; around 520 kilometres, wadsleyite changes into ringwoodite. Near 660 kilometres, ringwoodite breaks down into bridgmanite, a magnesium-rich silicate with the perovskite structure, and ferropericlase, a magnesium–iron oxide. These phase transitions explain much of the observed change in seismic properties across the transition zone. Other mantle minerals undergo their own transformations, so the actual reactions in natural rock are more complex than those in pure magnesium silicate.
The lower mantle is dominated by bridgmanite, with ferropericlase and a calcium-silicate perovskite phase as other major constituents. Their exact proportions depend on assumed bulk composition. Bridgmanite is generally regarded as Earth’s most abundant mineral by volume, even though ordinary surface exposures do not preserve it: it is stable only at high pressure and usually transforms during decompression. In the deepest mantle, sufficiently high pressure stabilizes a denser post-perovskite structure in some temperature and compositional conditions. Its occurrence helps explain some seismic features near the core–mantle boundary.
Models of average mantle composition are constrained by mantle rocks, basalt chemistry, meteorites, and cosmochemical estimates. A widely used model composition, pyrolite, approximates a fertile mantle source capable of yielding basalt and a depleted peridotitic residue. It is a reference model, not evidence that the mantle is homogeneous. Recycled basaltic crust, depleted residues, chemically modified lithosphere, and potentially ancient reservoirs all introduce heterogeneity. The distinction between average composition and locally sampled composition is essential because surface-accessible mantle material disproportionately represents shallow regions that have experienced melting and deformation.
How the mantle is investigated
Seismology provides the most extensive direct physical evidence about the mantle. Compressional waves respond to both resistance to compression and rigidity, whereas shear waves depend on rigidity and cannot propagate through an ordinary liquid. For an isotropic elastic material, their speeds are
where is the bulk modulus, the shear modulus, and the density. Travel times, reflected and converted waves, and Earth’s free oscillations constrain the depth dependence of these properties. Their interpretation must account for temperature, composition, mineral structure, and the distribution of these properties in space.
A foundational discovery came in 1909, when Andrija Mohorovičić interpreted earthquake arrivals in terms of a marked velocity increase beneath the crust. Subsequent work established the major mantle discontinuities and the core boundary. The Preliminary Reference Earth Model, published by Adam Dziewonski and Don Anderson in 1981, became a widely used radial model of density, seismic velocities, and attenuation. Such reference models describe average structure; they do not capture all regional variation.
Seismic tomography uses large collections of seismic observations to map three-dimensional variations in wave speed. Many fast anomalies correspond to relatively cold descending slabs, while some slow anomalies are consistent with hotter material. These associations are physically useful but not unique: composition, phase, anisotropy, and partial melt can also affect velocity. Resolution is uneven because earthquakes and recording stations are geographically concentrated. Long-wavelength structures are generally more securely resolved than small features deep beneath poorly sampled regions.
Rocks provide complementary evidence. Ophiolites expose fragments of oceanic lithosphere tectonically emplaced on land, including portions of the uppermost mantle. Xenoliths, fragments carried upward in volcanic magmas, sample mantle beneath both continents and oceans. Basalts reveal their sources indirectly through major elements, trace elements, and isotopic compositions. Mineral inclusions in diamonds can preserve evidence from much greater depths, including the transition zone and lower mantle, although they represent exceptionally small and selective samples.
Laboratory experiments recreate mantle conditions using large-volume presses, shock compression, and laser-heated diamond-anvil cells. X-ray diffraction identifies high-pressure crystal structures, while measurements of elasticity, melting, and chemical partitioning connect mineral properties to geophysical observations. Atomistic calculations extend this work into conditions difficult to sustain experimentally. Independent constraints come from Earth’s gravity field, rotational properties, surface heat flow, electrical conductivity, and deformation after ice-sheet loss. Agreement among these different observations is more informative than any single measurement because several combinations of temperature and composition can produce similar seismic signals.
Temperature, heat transport, and deformation
The mantle contains heat inherited from planetary formation and differentiation, together with heat released by the decay of radioactive isotopes, chiefly those of uranium, thorium, and potassium. It also receives heat from the core. Earth’s total surface heat loss is estimated at roughly 47 terawatts, but this is not exclusively mantle heat: it includes heat generated in the crust and heat transmitted upward from deeper reservoirs. The relative contributions of mantle radioactivity, secular cooling, and core heat remain important constraints on thermal evolution.
Temperature rises with depth, but not at a constant rate. Within vigorously convecting regions, it approximately follows an adiabatic gradient: a moving parcel becomes warmer when compressed without needing to receive external heat. Steeper gradients occur in thermal boundary layers, especially the lithosphere and the lowermost mantle. Upper-mantle potential temperatures inferred for typical mid-ocean-ridge sources are commonly about 1,300–1,400 °C. Potential temperature is the temperature a parcel would have if brought adiabatically to a reference pressure, not its actual temperature at depth. Temperatures near the mantle’s base are much less tightly constrained, with many estimates lying around 3,500–4,000 kelvin.
Despite these temperatures, most mantle material remains below its melting temperature because the solidus generally increases with pressure. A hot solid can nevertheless deform through the movement of crystal defects. In diffusion creep, atoms move through crystals or along grain boundaries; in dislocation creep, defects migrate through crystal lattices. Grain-boundary sliding and other mechanisms can also contribute. Their relative importance depends on temperature, stress, mineralogy, grain size, and water content.
Mantle viscosity is therefore an effective material property rather than one universal constant. Many geodynamic models employ values of approximately –, with substantial regional and depth-dependent variation. Estimates come partly from postglacial rebound: the delayed rise of land after removal of large ice sheets. The lower mantle is generally inferred to be more viscous than much of the upper mantle, but detailed profiles differ among models. Temperature-dependent viscosity allows cold slabs and warm surroundings to have very different strengths.
This behavior reconciles two seemingly contradictory descriptions. On timescales of seconds to minutes, the mantle supports shear waves and behaves chiefly as an elastic solid. Over thousands to millions of years, accumulated creep allows large deformation. The resulting flow is extremely slow and typically has negligible inertia. Accordingly, mantle dynamics is usually described by a creeping-flow approximation, not by the turbulent fluid motion familiar from rivers or the atmosphere. Heat is carried efficiently through the interior by this slow circulation, while thermal conduction becomes especially important across boundary layers.
Convection and its relationship to plate tectonics
Mantle convection arises because variations in temperature and composition create buoyancy differences under gravity. Cooling near the surface makes oceanic lithosphere denser, while heating at depth can make mantle material more buoyant. Whether a particular region rises or sinks also depends on its composition, mineral phases, and mechanical connection to surrounding material. The mantle is consequently not well represented by a few permanent, geometrically regular circulation cells.
The tendency toward thermal convection can be expressed by a Rayleigh number,
where is gravitational acceleration, thermal expansivity, a characteristic temperature difference, layer thickness, thermal diffusivity, and dynamic viscosity. Estimates for the mantle greatly exceed the onset threshold for convection in simple layers. However, a single Rayleigh number does not capture its strong viscosity contrasts, spherical geometry, phase transitions, internal heating, or mobile surface plates.
Subduction is a central part of this circulation. As oceanic plates age, they cool and thicken; where they descend, their negative buoyancy exerts a major force on plate motion. The sinking plate draws surrounding mantle into motion and creates circulation above and around the slab. Plate motion and mantle flow are therefore dynamically coupled rather than separate systems in which a conveyor belt simply carries passive plates. Forces associated with slab descent, gravitational sliding away from elevated ridges, plate-boundary resistance, and stresses beneath plates all contribute.
The transition zone complicates the passage of slabs into the lower mantle. The phase transformations near 660 kilometres can oppose or modify vertical motion, and an increase in viscosity adds resistance. Seismic images show that some slabs flatten or accumulate near this depth, while others penetrate well into the lower mantle. This combination is inconsistent with a universally impermeable boundary between two independently convecting layers. It also shows that whole-mantle exchange need not occur uniformly at all subduction zones or throughout geological time.
Rising flow takes several forms. Beneath mid-ocean ridges, much upwelling is driven by plate separation and the need to replace material moving laterally away. More localized buoyant upwellings are described as mantle plumes. The plume hypothesis, developed prominently by W. Jason Morgan in 1971, links some intraplate volcanism to hot material rising from depth. Hawaii is a major example, but volcanic hotspots do not all share an established common origin. Tomography, geochemistry, and plate reconstructions support deep sources for some systems, while the depth, shape, and persistence of individual upwellings remain subjects of investigation.
Partial melting and the production of crust
Mantle melting ordinarily begins when local conditions cross the solidus, the temperature at which the first liquid becomes stable for a given pressure and composition. It is usually partial: minerals melt in different proportions, and the first liquid differs chemically from its source. Fertile upper-mantle peridotite commonly produces basaltic melts, leaving a residue increasingly depleted in components extracted into the liquid. The resulting separation of melt and residue is a major mechanism of planetary chemical differentiation.
At mid-ocean ridges, the principal mechanism is decompression melting. Mantle rises as plates separate, and its pressure decreases more rapidly than its temperature falls along the ascent path. It can therefore cross the solidus without an external heat source raising its temperature. Melting occurs across a broad region beneath the ridge rather than exclusively inside a narrow vertical conduit. Melt segregates from the solid framework, becomes focused toward the ridge, and feeds crustal magma systems. The ordinary oceanic crust produced through this process is approximately 6–7 kilometres thick.
Subduction zones have a different setting. The descending slab releases water-rich fluids as hydrous minerals become unstable. These fluids enter the overlying mantle wedge and lower the temperature required for melting. The wedge, rather than the slab itself, supplies much of the primary magma in many volcanic arcs. Subsequent crystallization, crustal interaction, and magma mixing produce the diverse rocks erupted by arc volcanoes. Direct slab melting can occur under particular thermal conditions, but it is not the general explanation for arc magmatism.
Melt extraction is a physical as well as a chemical process. At low melt fractions, liquid can occupy connected grain-edge networks where mineral–melt geometry permits. Pressure gradients and buoyancy drive migration through the deformable solid matrix. Chemical reactions can enlarge preferred pathways, and fractures or dikes can carry melt more rapidly at shallower levels. The permeability, compaction behavior, and reactive evolution of the mantle determine how effectively small melt fractions are collected into larger magma bodies.
Trace elements provide sensitive records of these processes. Incompatible elements preferentially enter the melt rather than remain in the main mantle minerals, so even a small degree of melting can strongly enrich the liquid relative to its source. Rare-earth-element patterns help constrain melting depth and residual mineralogy, particularly whether garnet was present. Radiogenic isotope systems distinguish source histories that cannot be inferred from major-element chemistry alone. Mid-ocean-ridge and ocean-island basalts display systematic differences, but neither group represents a chemically uniform reservoir. Repeated melting, melt infiltration, and recycling create a mantle whose components can retain distinct histories while remaining dynamically mixed.
Water, carbon, and exchange with the surface
Mantle water is mostly not stored as free liquid. Hydrogen occurs in crystal defects, commonly associated with oxygen as hydroxyl, and in hydrous minerals under appropriate conditions. The concentration of hydrogen in nominally anhydrous minerals may be small, yet it can strongly affect melting, deformation, and electrical conductivity. Consequently, “water content” in mantle studies often denotes the equivalent amount of represented by structurally bound hydrogen rather than a separate fluid phase.
Wadsleyite and ringwoodite can incorporate substantially more water than the dominant minerals above and below the transition zone under suitable laboratory conditions. This gives the transition zone considerable potential storage capacity. A hydrous ringwoodite inclusion reported from a diamond in 2014 provided direct evidence that at least part of the transition zone contains appreciable water. It did not establish that the whole region is uniformly hydrated. Storage capacity, actual concentration, and total inventory are different quantities, and global estimates remain dependent on indirect observations and models.
Subduction carries water downward in altered oceanic crust, sediments, and hydrated mantle rocks. Much is released at relatively shallow depths through metamorphic reactions, feeding fluids and magmatism above the slab. Some can travel deeper in cold slabs, where particular hydrous phases remain stable. In the opposite direction, mantle-derived magmas return water to the crust and ultimately the surface. This exchange forms a deep component of the water cycle, although its rates and the balance between inward and outward transport vary over time.
Carbon likewise moves between the surface and mantle as part of the carbon cycle. Subducted material includes carbonate minerals and organic carbon, while the mantle stores carbon in forms determined by pressure, temperature, composition, and oxidation state. Possible hosts include carbonates, diamond, graphite, and carbon-bearing melts or fluids. Oxidized carbon-bearing melts can form at temperatures below those required for extensive silicate melting, making small carbon concentrations important to mantle melting behavior. Carbon’s chemical form changes as material enters regions with different redox conditions.
These volatile exchanges connect mantle processes with the long-term evolution of oceans and atmosphere, but the mantle is not a single well-mixed tank. Different reservoirs have different residence times, and surface outputs sample them unevenly. Volcanic gas measurements, basalt compositions, mineral inclusions, and experiments constrain parts of the system. Noble gases provide additional tracers: helium isotope ratios, for example, can help identify source components with different histories of radiogenic helium accumulation and degassing. Such signatures constrain reservoir evolution but do not uniquely specify source depth or prove that a sample came from an untouched primordial region.
The lowermost mantle and deep heterogeneity
The deepest few hundred kilometres of the mantle are often called the D″ region. This is a seismological designation for a complex basal region, not a globally uniform layer with a single composition or sharply fixed top. It contains strong lateral variations in seismic velocity, discontinuities, anisotropy, and pronounced temperature gradients. Because it lies above the hot core, it is a major thermal boundary region and a likely source of some buoyant upwellings.
The discovery of the post-perovskite transition in magnesium silicate in 2004 supplied a mineral-physics explanation for some D″ observations. Under suitable conditions, bridgmanite transforms into a layered crystal structure with different elastic properties. The depth of this transition depends on temperature as well as pressure, so it may occur at different heights above the core or be absent where conditions do not favor it. Mineral alignment during deformation can also generate directional differences in wave speed. Neither post-perovskite nor temperature alone explains every feature of the lowermost mantle.
Two enormous regions of relatively low shear-wave velocity occupy much of the mantle base beneath Africa and the Pacific. They are commonly termed large low-shear-velocity provinces. Their broad outlines recur in many tomographic models, although estimates of their detailed geometry differ. Proposed explanations include elevated temperature, distinct chemical composition, or combinations of the two. Models involving long-lived thermochemical structures are compatible with several observations, but their density, origin, longevity, and internal organization are not uniquely established.
Smaller features called ultralow-velocity zones occur in some locations immediately above the core. They can be only tens of kilometres thick and show unusually large reductions in seismic speed. Partial melting, iron enrichment, and other compositional effects have been proposed to explain them. Different zones may have different origins. Their detection depends strongly on the availability of seismic waves sampling the appropriate portions of the boundary, so an apparent absence can reflect limited coverage rather than genuine uniformity.
The core–mantle boundary also connects mantle dynamics to the evolution of Earth’s magnetic field. Flow in the electrically conducting outer core generates the field, but the mantle controls the spatial pattern and long-term rate of heat removal from the core. Uneven cooling can influence core convection without requiring the mantle itself to generate the field. Possible chemical exchange at the boundary adds further complexity: high-pressure experiments indicate that reactions and elemental partitioning are possible, but their global extent is difficult to establish. Seismic structure, mineral stability, and core energetics therefore provide complementary constraints on this inaccessible interface.
Origin, long-term evolution, and other planetary mantles
Earth’s mantle emerged during accretion and differentiation approximately 4.5 billion years ago. As metallic material separated and sank to form the core, the remaining silicate portion developed into the mantle and crust. Large impacts, gravitational energy release, and radioactive heating produced extensive melting, possibly including one or more global or nearly global magma oceans. Metal–silicate equilibration during this period helped establish the mantle’s elemental inventory, particularly the abundance of elements that preferentially enter metal.
Crystallization of a magma ocean need not produce an initially uniform solid mantle. Different minerals crystallize at different stages, leaving the residual liquid chemically changed. Some crystallization sequences can create unstable density arrangements that later overturn; others can establish persistent compositional contrasts. Subsequent impacts, melting, and convection further modify these structures. Present-day observations permit several evolutionary scenarios, and no single detailed reconstruction of early mantle layering is established.
Over geological time, the mantle has cooled while continually exchanging material with the crust. Ancient komatiites, unusually magnesium-rich volcanic rocks especially common in the Archean record, provide evidence for higher mantle melting temperatures in Earth’s past. Their interpretation depends on source composition and volatile content as well as temperature. Cooling has changed melting productivity, lithospheric strength, and the conditions under which slabs can descend. The onset and early character of plate tectonics remain debated because ancient rocks preserve incomplete and sometimes ambiguous records of the processes that formed them.
Mantles are not unique to Earth. The Moon, Mars, Venus, and Mercury all possess silicate mantles, although their thicknesses, compositions, thermal histories, and relationships with crust and core differ. Mercury’s unusually large metallic core leaves a comparatively thin silicate shell. The Moon’s small size has favored cooling and the development of a thick rigid outer layer. Mars preserves extensive evidence of mantle-derived volcanism but lacks Earth’s present global network of active plate boundaries. Venus is similar to Earth in size, yet its surface deformation and volcanic history do not demonstrate a currently operating Earth-like plate-tectonic system.
Planetary comparisons show that a convecting mantle does not automatically produce plate tectonics. Interior temperature, water content, lithospheric strength, planet size, and the history of surface cooling all affect whether the outer shell breaks into mobile plates or behaves more like a stagnant lid. For rocky exoplanets, mantle structures are inferred from planetary mass, radius, stellar composition, and high-pressure models rather than direct samples. At pressures exceeding those within Earth, familiar silicate minerals may undergo additional transformations, so Earth’s mineral layering cannot simply be extended unchanged to larger rocky worlds.