Earth's internal structure

No one has ever drilled more than about 12 kilometers into the Earth, yet geologists know the detailed composition, temperature, and physical state of every layer down to the center some 6,371 kilometers below the surface.¹ This knowledge comes almost entirely from seismology — the study of how vibrations from earthquakes travel through the planet's interior. The speed at which seismic waves move and the paths they follow depend on the density and elasticity of the material through which they pass. When waves accelerate, slow, or bend sharply, they are crossing a boundary between materials of fundamentally different character. By recording those signals at seismometer networks around the globe and working backward through the physics of wave propagation, scientists have built a portrait of Earth's interior whose broad outlines have not changed since the mid-twentieth century and whose fine detail continues to sharpen with each new generation of instruments and computing power.^{1, 6}

Seismology as a probe of the deep Earth

Earthquakes radiate two principal types of body waves, each sensitive to different properties of the material they traverse. P-waves (primary or compressional waves) are longitudinal: they push and pull the rock in the same direction they travel, much like sound through air. Because compression can propagate through any medium — solid, liquid, or gas — P-waves pass through all parts of Earth's interior. S-waves (secondary or shear waves) shake the rock perpendicular to their direction of travel. Shear requires a rigid medium, so S-waves cannot travel through liquids. This fundamental property is what allowed seismologists to identify Earth's liquid outer core: a global "shadow zone" exists in which S-waves are absent from seismometers placed roughly 103 to 143 degrees of arc from an earthquake epicenter, precisely consistent with a fluid region that absorbs shear energy.^{3, 1}

P-waves also cast a shadow, but a subtler one. Because waves refract — bending toward regions of lower velocity — the liquid outer core deflects P-waves inward, leaving a zone of reduced P-wave amplitude between about 103 and 143 degrees. The German-American seismologist Beno Gutenberg mapped this P-wave shadow in 1914 and from it inferred the depth of the core-mantle boundary with remarkable accuracy: 2,900 kilometers below the surface, a figure confirmed by subsequent studies to within a few kilometers.³

Beyond direct wave analysis, seismologists use a technique called seismic tomography, which applies the same mathematical principles as medical CT scanning. By comparing the actual arrival times of thousands of seismic waves at hundreds of stations worldwide against predicted arrival times for a uniform reference Earth, researchers build three-dimensional maps of slow (hot or compositionally different) and fast (cold or denser) regions in the mantle and core. Tomographic models have imaged cold subducting slabs sinking deep into the mantle, hot plumes rising from near the core-mantle boundary, and broad low-velocity provinces that may represent ancient chemical heterogeneities accumulated over billions of years.⁷

The layered structure of Earth

Earth's interior is organized into four concentric shells distinguished by composition, density, and physical state. The boundaries between them are marked by sharp changes in seismic wave velocities, each named for the scientist who identified it. Moving from the surface inward, these shells are the crust, the mantle, the outer core, and the inner core.^{6, 1}

Principal layers of Earth's interior^{6, 8, 19, 24}

The crust and the Mohorovičić discontinuity

The outermost shell of Earth is the crust, the thin, rigid layer of silicate rock on which all surface geology occurs. It is not uniform. Oceanic crust, generated continuously at mid-ocean ridges and recycled back into the mantle at subduction zones, is relatively thin — averaging 5 to 10 kilometers — and composed predominantly of basalt and gabbro, dense iron- and magnesium-rich rocks.²⁴ Continental crust is far thicker, averaging 35 to 40 kilometers and reaching 70 kilometers beneath some mountain ranges such as the Himalayas, and is dominated by granitic rocks rich in silicon, aluminum, potassium, and sodium. Because continental crust is less dense than the underlying mantle, it is buoyant and is not recycled on geologically short timescales; some fragments survive for more than 4 billion years.²⁴

The boundary between the crust and the mantle, the Mohorovičić discontinuity (universally abbreviated to the Moho), was discovered in 1909 by the Croatian geophysicist Andrija Mohorovičić. Analyzing seismograms from an earthquake in the Kupa Valley of what is now Croatia, he noticed that waves arriving at distant stations appeared to travel faster than expected, as if an additional set of faster waves were arriving alongside the direct crustal waves. He correctly interpreted this as evidence of a compositional boundary at depth, below which seismic velocities increased sharply as waves entered denser material. P-wave velocities jump from roughly 6 km/s in the lower crust to about 8 km/s in the uppermost mantle at this interface.² The Moho is not a uniform global horizon — it is shallower beneath ocean floors and considerably deeper beneath mountain belts — but it is globally continuous and is detected wherever seismic surveys are conducted.⁵

The mantle: composition, convection, and seismic structure

The mantle extends from the Moho to the core-mantle boundary at 2,891 kilometers depth and constitutes roughly 84 percent of Earth's volume.¹⁹ It is composed almost entirely of silicate minerals. The uppermost mantle is dominated by peridotite, a coarse-grained rock rich in the mineral olivine along with pyroxenes and garnet. These minerals are stable at the pressures and temperatures of the upper mantle, but as depth increases, pressure forces mineral grains to reorganize into denser crystal structures. At around 410 kilometers depth, olivine transforms into a denser polymorph called wadsleyite, producing a global seismic discontinuity detectable in wave reflection studies. A second major phase transition at approximately 660 kilometers depth, where wadsleyite breaks down into bridgmanite (formerly called perovskite) and ferropericlase, marks the boundary between the upper and lower mantle.^{6, 13} These phase transitions create barriers that influence, though do not absolutely prevent, the vertical flow of material within the mantle.

The mantle is solid in the ordinary mechanical sense — seismic S-waves traverse it — but on geological timescales of millions to hundreds of millions of years it behaves as an extremely viscous fluid and flows by solid-state creep, a process in which atoms slowly diffuse through crystal lattices under sustained stress and heat.¹¹ This slow convective flow, driven by heat escaping from Earth's interior, is the engine that moves tectonic plates, opens and closes ocean basins, builds and erodes mountain ranges, and cycles material between the crust and the deep interior. Average mantle flow rates are measured in centimeters per year, closely matching the observed drift rates of tectonic plates as measured by GPS.¹¹

A key distinction within the upper mantle is between the lithosphere and the asthenosphere. The lithosphere is the mechanically strong, rigid outer shell of Earth that includes the crust and the uppermost, cooler part of the mantle. It is not defined by composition but by mechanical behavior: the lithosphere moves as rigid plates over geological time rather than flowing. Oceanic lithosphere is typically 50 to 100 kilometers thick; continental lithosphere can exceed 200 kilometers in old, stable regions called cratons.²³ Beneath the lithosphere lies the asthenosphere, a zone where temperatures are close enough to the melting point of mantle rock that the material weakens substantially and can flow more readily. Seismically, the asthenosphere appears as a low-velocity zone for both P and S waves, reflecting its reduced rigidity. This contrast in mechanical behavior — rigid lid floating on a ductile substrate — is what permits the lateral movement of tectonic plates over the deeper mantle.¹²

A long-standing debate in mantle geodynamics concerns whether convection is stratified or whole-mantle. In the layered convection model, the 660-kilometer phase transition is a sufficiently strong density barrier to keep upper-mantle and lower-mantle material largely separate, with heat transferred across the boundary mainly by conduction. In the whole-mantle model, convection cells span the full depth from core-mantle boundary to the base of the lithosphere, allowing chemical exchange between all mantle depths. Seismic tomographic images now show cold subducting plates penetrating through the 660-kilometer boundary into the lower mantle in many locations, suggesting that whole-mantle convection operates at least periodically, though some chemical stratification persists.^{13, 7}

At the very base of the mantle sits a region of special complexity designated the D″ layer (pronounced "D double-prime"), a zone approximately 200 to 300 kilometers thick just above the core-mantle boundary.¹⁴ First identified from anomalous seismic waveforms in the early 1980s by Thorne Lay and Donald Helmberger, the D″ layer exhibits heterogeneous seismic velocities, patches of ultralow-velocity zones interpreted as partially molten rock, and evidence of thermal and chemical exchange with the liquid core below.^{14, 21} In 2004, high-pressure experiments revealed that the dominant lower-mantle mineral bridgmanite transforms into a denser form called post-perovskite at conditions matching those at the base of the mantle, providing a phase-transition explanation for the seismic discontinuity at the top of the D″ layer.¹⁵ The D″ layer is also the proposed source region for deep mantle plumes — columns of anomalously hot material that rise buoyantly through the mantle and, after tens of millions of years, reach the surface to produce hotspot volcanism such as that beneath Hawaii and Iceland.²⁰

The core: Gutenberg, Lehmann, and the iron heart of the planet

Earth's core is an iron-dominated body divided into two fundamentally distinct regions separated by the Lehmann discontinuity at approximately 5,150 kilometers depth. The outer core, extending from the core-mantle boundary at 2,891 kilometers to 5,150 kilometers depth, is liquid. The inner core, from 5,150 kilometers to the center at 6,371 kilometers, is solid. These facts are known because S-waves cannot propagate through the outer core but do traverse the inner core, as confirmed by the detection of waves that pass through the inner core (designated PKIKP waves) at seismometer stations on the opposite side of the planet from an earthquake.²²

The existence of a major seismic discontinuity near the center of Earth was first inferred by Beno Gutenberg from the P-wave shadow zone he analyzed in 1914, establishing the core-mantle boundary depth.³ But Gutenberg's model placed a fluid core that extended all the way to Earth's center, which left certain anomalous seismic arrivals unexplained. In 1936, the Danish seismologist Inge Lehmann, working from careful analysis of P-wave arrivals in the shadow zone that should have been absent if the entire core were liquid, proposed that a solid inner core existed within the liquid outer core, from which waves could reflect and refract at unexpectedly steep angles to reach stations inside the shadow zone.⁴ Subsequent seismological observations and computational modeling fully confirmed Lehmann's hypothesis; the inner-core boundary is now one of the best-constrained features of Earth's deep interior. Lehmann's discovery is considered one of the most remarkable achievements in geophysics, accomplished using only paper records and slide-rule calculations.

The core is composed primarily of iron alloyed with nickel and small amounts of lighter elements — candidates including sulfur, oxygen, silicon, carbon, and hydrogen that lower the density below that of pure iron at core pressures.^{19, 25} The identity and abundances of these light elements remain an active research question because they are inferred indirectly: by comparing the seismically measured density of the core against the density predicted by high-pressure iron experiments and by matching the bulk composition of iron meteorites, which represent the cores of disrupted planetesimals from the early solar system. The outer core liquid contains a higher proportion of light elements than the inner core because as the inner core gradually solidifies from the outside in, it preferentially incorporates iron and nickel while expelling lighter elements into the liquid above.^{8, 25}

Temperatures in the core are extreme by any surface standard. At the inner-core boundary, estimates based on melting experiments at multimegabar pressures cluster around 5,000 to 6,000 °C — comparable to the surface of the Sun.⁸ This heat is a mixture of primordial heat inherited from Earth's accretion and differentiation over 4.5 billion years ago, radioactive decay of elements (primarily potassium-40 and possibly uranium and thorium at trace levels in the core), and the latent heat and gravitational energy released as the inner core slowly crystallizes.¹⁸

The geodynamo and magnetic field

The most consequential product of the liquid outer core is Earth's magnetic field. Electric currents circulating in the electrically conducting iron-nickel liquid generate a large-scale magnetic field through a self-sustaining process called the geodynamo. The mechanism was first made physically plausible by Walter Elsasser and Edward Bullard in the 1940s and 1950s, but a fully self-consistent numerical simulation of the geodynamo was not achieved until 1995, when Gary Glatzmaier and Paul Roberts solved the coupled equations of fluid dynamics and electromagnetism for a rotating, convecting spherical shell of conducting fluid and produced a field closely resembling Earth's observed dipole field — including spontaneous reversals of magnetic polarity.⁹

The driver of convection in the outer core is twofold. Thermal convection arises from the temperature difference between the hotter core-mantle boundary below and the cooler base of the mantle above. Compositional convection arises because the freezing of iron-nickel alloy at the inner-core boundary releases lighter elements into the overlying liquid, creating buoyant upwellings of lighter, less-dense fluid that rise and drive circulation. Earth's rotation shapes the convective flow through Coriolis forces into columnar vortices aligned roughly parallel to the rotation axis, a geometry that efficiently generates and sustains the large-scale magnetic dipole field.^{9, 10}

Earth's magnetic field is not static. On geological timescales, the dipole axis wanders and the field's intensity fluctuates. More dramatically, the polarity of the field reverses — magnetic north becomes magnetic south and vice versa — on an irregular basis averaging roughly once every several hundred thousand years over the Cenozoic, though intervals between reversals have ranged from tens of thousands to tens of millions of years.¹⁰ The paleomagnetic record of reversals, preserved in volcanic rocks and marine sediments, has been critical evidence for the reality of seafloor spreading and plate tectonics: symmetrical striped patterns of normal and reversed polarity flanking mid-ocean ridges proved that new seafloor was being created and magnetized at the ridge axis and spreading away over millions of years.¹⁰

The geomagnetic field performs a vital planetary service by deflecting much of the solar wind — the continuous stream of charged particles emanating from the Sun — far from Earth's surface, creating the magnetosphere. Without this field, solar wind would erode the upper atmosphere over geological timescales, as has apparently occurred on Mars, which lost its global dynamo early in its history and subsequently lost much of its early thick atmosphere.⁹

Geothermal gradient and heat flow

Heat continuously escapes from Earth's interior to the surface. The global average heat flow through the ocean floor and continents combined is approximately 47 terawatts — a quantity comparable in magnitude to human civilization's total energy consumption, though far too diffuse at any given surface location to harvest usefully without geothermal drilling.¹⁶ The geothermal gradient — the rate at which temperature increases with depth — averages roughly 25 to 30 °C per kilometer in the uppermost crust, though it varies enormously depending on location and tectonic setting: gradients exceed 100 °C/km in volcanic arcs and rifts, while ancient cratonic shields may show gradients as low as 10 °C/km.¹⁶

The gradient is not constant with depth. In the crust and uppermost lithosphere, heat escapes largely by conduction — the slow transfer of thermal energy through solid rock. Deeper in the mantle, conduction alone is too inefficient to transport the required heat; instead, the material itself moves, carrying heat upward by convection. This is why the mantle can remain near but below its melting point throughout its great depth: convection is a far more efficient heat-transfer mechanism than conduction for a mobile solid such as the mantle.^{11, 18}

Earth's heat has two principal sources. Primordial heat — the thermal energy stored during the planet's accretion from the solar nebula and during the differentiation event in which a dense iron core sank through the early magma ocean — accounts for roughly half of the current heat flow. The other half derives from the continuous radioactive decay of long-lived isotopes, principally uranium-238, uranium-235, thorium-232, and potassium-40, distributed through the mantle and crust.¹⁸ This combination of stored primordial heat and ongoing radiogenic heating has kept Earth's interior warm enough to sustain mantle convection, plate tectonics, and the geodynamo across the entire 4.5-billion-year history of the planet.^{16, 18}

Approximate heat flow by tectonic setting¹⁶

Reference Earth models and ongoing research

The quantitative foundation for understanding Earth's interior is the Preliminary Reference Earth Model (PREM), published by Adam Dziewonski and Don Anderson in 1981. PREM is a one-dimensional mathematical model that specifies seismic wave velocities, density, pressure, and elastic moduli as functions of depth for a spherically symmetric, non-rotating Earth in hydrostatic equilibrium.⁶ It was constructed by inverting an enormous dataset of seismic travel times, normal-mode frequencies of Earth's free oscillations, and astronomically derived constraints on Earth's mass and moment of inertia into a self-consistent set of depth profiles. PREM remains the standard reference against which seismic anomalies are measured, and most seismic tomography studies express their results as deviations from PREM velocities.

PREM's one-dimensional simplicity is both a strength and a limitation. It accurately captures the global average structure but by design cannot represent the three-dimensional heterogeneity of the real Earth. Modern research in seismology focuses on resolving finer structures: the detailed morphology of subducting slabs in the transition zone, the geometry of superplumes rising from the core-mantle boundary, the seismic anisotropy of the inner core (which appears to transmit P-waves faster along Earth's rotation axis than perpendicular to it, possibly reflecting the preferential alignment of iron crystals as they grow under the influence of the magnetic field), and the physical properties of ultralow-velocity zones at the very base of the mantle.^{21, 22}

Complementing seismology, mineral physicists recreate deep-Earth conditions in diamond-anvil cells and multi-anvil presses, squeezing tiny samples to pressures exceeding 1.5 million atmospheres while heating them with lasers, then measuring their densities, sound velocities, and phase transitions. These experiments calibrate the interpretations of seismic data by establishing which mineral phases are stable at given pressures and temperatures and what seismic velocities they produce.^{17, 15} Together, seismology and mineral physics have transformed the study of Earth's interior from informed inference into a quantitative science capable of resolving features hundreds of kilometers across at depths inaccessible by any other means.