Introduction

The Earth's mantle is a thick silicate-rock shell lying between the crust and the outer core and constitutes a principal component of the planet's internal structure. It has a mass of 4.01×10^24 kg—about 67% of Earth's total mass—and extends roughly 2,900 km in depth, equivalent to ≈46% of Earth's radius and nearly 84% of its volume. Compositionally solid, mantle rocks deform and flow on geological timescales as a very slow, viscous material, enabling long‑term convective movement. Partial melting within the mantle is the source of new crust: decompression melting beneath mid‑ocean ridges produces oceanic crust, while melting associated with subduction processes contributes to the generation of continental crust.

The rheological structure of the upper mantle is characterized by a mechanical division between a stiff, cold lithospheric mantle and an overlying, mechanically weaker asthenosphere, separated by the lithosphere–asthenosphere boundary (LAB). Together with the crust, the lithospheric mantle constitutes the lithosphere that behaves as rigid tectonic plates riding atop the more ductile asthenosphere; beneath the asthenosphere mantle material generally regains a comparatively rigid response with increasing depth and pressure.

Oceanic and continental lithospheres differ substantially in thickness: oceanic plates (crust plus lithospheric mantle) are typically on the order of 100 km, whereas continental lithosphere commonly reaches 150–200 km, reflecting contrasts in thermal structure, composition and tectonic history that influence plate buoyancy and strength.

Seismology divides the mantle into upper mantle, transition zone and lower mantle on the basis of abrupt velocity discontinuities that signal changes in mineralogy and physical state. The upper mantle overlies the Mohorovičić discontinuity (Moho) — the crust–mantle boundary located roughly 7–35 km depth — and extends downward to the strong 410 km seismic discontinuity.

The transition zone, between about 410 and 660 km depth, is dominated by pressure‑induced phase transformations of olivine‑group minerals; these transformations (e.g., to wadsleyite in the ~410–520 km interval and to ringwoodite in the deeper part of the zone) produce the pronounced seismic velocity contrasts that define the zone’s bounds.

The lower mantle spans roughly 660 to 2,891 km and is controlled by high‑pressure silicate phases. Bridgmanite is the principal phase through much of the lower mantle (~660–2,685 km), while a transition to post‑perovskite is inferred at the deepest part of the lower mantle (~2,685–2,891 km), indicating further changes in crystal structure and physical properties with depth.

The lowermost ~200 km of the lower mantle, known as the D″ region, exhibits anomalous seismic signatures and hosts distinctive features such as large low‑shear‑velocity provinces (LLSVPs) and ultra‑low velocity zones (ULVZs). These heterogeneities mark a complex boundary layer that interfaces with deeper planetary structure and influences mantle dynamics and thermal evolution.

The seismic Moho marks the top of the mantle by a pronounced jump in seismic-wave speeds and thus separates the crust from a mantle dominated by peridotitic lithology. Upper-mantle peridotite is composed principally of olivine, clinopyroxene, orthopyroxene and an aluminous phase whose identity changes with pressure: plagioclase in the shallowest mantle gives way to a spinel-structured phase at greater depth and to garnet below roughly 100 km. With increasing depth and pressure pyroxenes become progressively unstable and are partly transformed into majoritic garnet, reflecting continuous reworking of the peridotitic assemblage.

Within the transition zone olivine undergoes isochemical, high-pressure polymorphic transformations to wadsleyite and then ringwoodite; these phases retain the bulk mantle chemistry while adopting crystal structures stable at elevated pressures. Unlike nominally anhydrous olivine, wadsleyite and ringwoodite can incorporate substantial amounts of hydroxyl in their crystal lattices, a capacity that supports the hypothesis that the transition zone may host considerable quantities of water. Near the base of the transition zone ringwoodite breaks down to form bridgmanite (previously termed magnesium silicate perovskite) and ferropericlase, and garnet ceases to be stable—together these reactions signal a marked shift in mineralogy entering the lower mantle.

The lower mantle is therefore dominated by bridgmanite and ferropericlase, with minor constituents such as calcium perovskite (and related Ca–Fe oxides) and stishovite reflecting the high‑pressure mineralogy of Earth’s deep interior. In the lowermost ≈200 km of the mantle bridgmanite undergoes an additional isochemical transition to the post‑perovskite structure; this late transition is implicated in the pronounced seismic and dynamic contrasts observed at the base of the mantle.

Possible remnants of Theia collision

Seismic imaging of the lowermost mantle — the layer immediately above the core–mantle boundary — reveals two continent‑sized anomalous provinces characterized by markedly reduced seismic wave speeds. Despite their seismic “slowness,” these regions exhibit higher bulk densities than the surrounding mantle, a combination that argues against a purely thermal explanation and instead indicates a distinct material composition.

Their greater density and compositional contrast imply that these provinces function as chemically isolated reservoirs at the base of the mantle. Such domains are likely resistant to entrainment by typical convective overturn and therefore can persist for very long geological intervals. Each anomaly’s lateral extent, on the order of thousands of kilometres, is large enough to perturb deep‑mantle flow and to affect heat transfer across the core–mantle boundary.

One proposed origin for these dense, seismically slow zones is that they are buried vestiges of Theia’s mantle — material delivered by the giant impact thought to have produced the Moon. If this interpretation is correct, it demonstrates that large‑scale compositional heterogeneities created during the Moon‑forming collision can survive incomplete mixing and remain sequestered at the mantle base for billions of years, thereby constraining models of early Earth accretion and post‑impact dynamics.

The existence of such compositional domains has significant geodynamic and geochemical consequences: they can alter mantle convection patterns and the initiation and composition of mantle plumes, modify thermal and mass flux across the core–mantle boundary, and serve as potential sources for atypical isotopic and elemental signatures observed at Earth’s surface.

Composition (Earth’s mantle)

Fragments of mantle rock delivered to the surface—most conspicuously green peridotite xenoliths found in volcanic hosts (e.g., in Arizona)—provide direct, in situ windows into mantle mineralogy and chemistry. Such xenoliths are discrete pieces of mantle lithology entrained by ascending magmas (basalts, kimberlites) and are complemented by mantle sections exposed in ophiolites, where slices of oceanic lithosphere have been emplaced onto continental crust. Because they are physical samples of otherwise inaccessible depths, these occurrences are essential for constraining upper‑mantle mineralogy and geochemistry, even though they sample only limited places and depths.

A common compositional reference for the uppermost mantle is the “depleted MORB” (dMORB) residue, expressed in oxide mass percent: SiO2 44.71; MgO 38.73; FeO 8.18; Al2O3 3.98; CaO 3.17; Cr2O3 0.57; NiO 0.24; MnO 0.13; Na2O 0.13; TiO2 0.13; P2O5 0.019; K2O 0.006. This composition represents the depleted residue remaining after extraction of mid‑ocean ridge basalt and serves as a standard for the chemistry of the uppermost mantle.

Interpretation of mantle composition must account for important limitations. Most quantitative constraints derive from samples that sample only the shallowest mantle (xenoliths and ophiolitic peridotites), producing a sampling bias and leaving open whether the lower mantle shares the same bulk composition as the upper mantle. In addition, the mantle’s average chemistry has not been static: repeated melting and segregation of magmas to form oceanic and continental crust progressively depleted the convecting mantle in certain elements, so bulk composition has evolved over geologic time.

Recent mineral‑physics and inclusion studies have also revealed more exotic phases that may occur within mantle domains. For example, analyses of water‑bearing fluid inclusions in diamonds suggest that supercritical water trapped at great depth can crystallize as the high‑pressure polymorph ice VII when pressure–temperature conditions change during ascent and cooling of the host diamond, illustrating how unusual aqueous phases can persist within deep‑mantle materials.

Temperature within the mantle increases strongly with depth, rising from roughly 500 K (≈230 °C) at the crust–mantle interface to about 4,200 K (≈3,900 °C) at the core–mantle boundary. This overall vertical gradient is punctuated by relatively thin, vigorous thermal boundary layers immediately beneath the crust (near the Moho) and adjacent to the core–mantle boundary, where temperature changes much more abruptly than in the more gradually warming mantle interior.

Although representative mantle rock (peridotite) has a melting temperature near 1,500 K at surface pressure, mantle material at depth commonly attains temperatures far above that value yet remains predominantly solid. The key control is lithostatic pressure: pressure rises with depth from a few hundred megapascals at the Moho to about 139 GPa at the core–mantle boundary, and the melting point (solidus) of mantle minerals increases with pressure. The combination of a depth-dependent solidus, the mantle’s vertical thermal profile, and the steep temperature changes confined to the top and bottom boundary layers explains why high internal temperatures do not produce wholesale melting of the mantle.

Movement

Mantle convection arises from the thermal contrast between the cool surface and the hot core and from the capacity of crystalline mantle rocks to deform slowly over geological time. Heat input at the core–mantle boundary produces thermal expansion of lowermost material that reduces its density and drives buoyant upwellings (plumes), while surface cooling produces dense, sinking lithosphere; numerical convection models commonly depict these contrasts with warm (red) and cool (blue) domains in single time slices. Because mantle rocks flow by very slow, permanent deformation—accommodated by the motion of point, line and planar defects through mineral crystals—the mantle behaves as a fluid on long timescales rather than as an instantaneous viscous liquid.

Convective circulation concentrates descending material at convergent plate margins (subduction zones) and predicts elevated topography and hotspot volcanism above rising plumes. An alternative explanation for intraplate volcanism is the plate hypothesis, which attributes surface volcanism to passive lithospheric extension enabling magma ascent rather than to deep-mantle plume roots. Mantle convection is chaotic in the fluid‑dynamic sense and is a fundamental driver of lithospheric plate motions; nevertheless, the rigid‑plate movements commonly referred to as plate motion are distinct from the historical concept of continental drift, even though lithosphere and mantle motions are tightly coupled through processes such as slab descent.

Rheological properties vary strongly with depth. Viscosity generally increases downward but the relation is non‑linear and interrupted by layers of much lower viscosity—notably within portions of the upper mantle and close to the core–mantle boundary—creating mechanical contrasts that steer convective patterns and influence where plate boundaries nucleate. A pronounced seismic and compositional region exists roughly 200 km above the core–mantle boundary, the D″ layer (term coined by K. E. Bullen), which may contain subducted, ponded slab material or high‑pressure polymorphs such as post‑perovskite.

Seismicity patterns record the mantle’s thermal and rheological state. Shallow earthquakes result from brittle failure, but increasing temperatures and pressures below roughly 50 km suppress ordinary brittle faulting as mantle rocks transition to viscous behavior. Yet subduction zones commonly generate earthquakes to depths of about 670 km; proposed mechanisms that allow seismic rupture at such depths include dehydration embrittlement, thermal runaway, and mineral phase transformations. Cold, descending lithosphere steepens local geothermal gradients and thereby strengthens surrounding mantle rocks, permitting intermediate‑depth seismicity between approximately 400 and 670 km.

Physical conditions in the lower mantle are extreme: pressure rises with depth to about 136 GPa near the base of the mantle. Estimates of mantle viscosity span many orders of magnitude—roughly 10^19 to 10^24 Pa·s in the upper mantle—reflecting sensitivity to depth, temperature, composition and stress state. These large viscosities imply exceedingly slow mantle flow, although the uppermost mantle can weaken under high stresses; such localized weakening is important for the initiation and localization of tectonic plate boundaries.

Exploration

Because oceanic crust is substantially thinner than continental crust, the seafloor has long been the preferred access point for direct sampling of the upper mantle beneath the lithosphere. Early attempts to reach mantle depths—including the pioneering but ultimately unsuccessful Project Mohole, terminated in 1966 after limited penetration (~180 m) and escalating costs—demonstrated both the scientific promise and the technical difficulty of deep drilling.

From 1968 onward, coordinated international ocean‑drilling programs yielded the most sustained progress. The Deep Sea Drilling Project (DSDP; 1968–1983), managed scientifically through Scripps Institution of Oceanography and guided by a large advisory community organized under JOIDES, produced empirical evidence crucial to seafloor spreading and plate‑tectonic theory. Successive programs (the Ocean Drilling Program, 1985–2003, and the Integrated Ocean Drilling Program thereafter) have maintained continuous multinational efforts in scientific ocean drilling. Notable operational achievements include a 2005 borehole by the JOIDES Resolution reaching 1,416 m below the seafloor and a 2007 RRS James Cook expedition to an area of exposed mantle on the Atlantic seafloor some 3 km beneath the ocean surface, where recovery of in situ mantle samples was planned.

Parallel technological ambitions have pushed into new domains: Japan’s Chikyu Hakken initiative equipped the drillship Chikyū to attempt unprecedented seabed penetration (targeting up to ~7,000 m below the seafloor), and more speculative concepts—such as a 2005 design for a radioactively heated, self‑melting probe—have been proposed to traverse crust and mantle autonomously. Computational work has also advanced understanding of mantle history and resource distribution; for example, 2009 supercomputer simulations traced the long‑term evolution and isotopic patterns of mantle minerals back to the early Earth.

Recent field results illustrate both progress and continuing ambiguity in mantle sampling. In 2023 the JOIDES Resolution recovered several hundred metres of core from the Atlantis Massif (maximum borehole depth 1,268 m, with 886 m of peridotite recovered) interpreted as upper mantle material. Although these cores are regarded as closer analogues to mantle rock than magmatic xenoliths—having not been melted and recrystallized—scientific debate persists over whether pervasive seawater alteration has transformed some samples into deep lower‑crustal material. Together, these historical and contemporary efforts underscore that, while ocean drilling remains the most viable route to direct mantle study, technical constraints and post‑recovery alteration continue to shape interpretation and drive innovation.