Convection currents in Earth’s mantle are driven by heat escaping from the planet’s interior, generated by two main sources: the slow decay of radioactive elements scattered throughout the rock, and primordial heat left over from Earth’s formation roughly 4.5 billion years ago. That heat creates temperature differences between the deep mantle and the surface, and because hot rock is less dense than cooler rock, the result is a slow churning motion where warmer material rises and cooler material sinks. But the details of how this works are far stranger and more contested than the tidy loop diagrams in most textbooks suggest.
Where the Heat Comes From
Earth’s mantle convection needs a fuel source, and it has two. The first is primordial heat, the thermal energy trapped inside the planet when it formed from colliding dust and rock in the early solar system. The violence of that accretion, along with the heat released when iron sank to form the core, left the interior extraordinarily hot. The second source is radiogenic heat, produced by the ongoing decay of unstable isotopes of uranium, thorium, and potassium embedded in mantle rock. Together, these sources sustain the temperature gradients that keep the mantle in motion.
The relative contribution of each source has been debated for decades. Modeling work has shown that if you try to explain Earth’s current heat output using only primordial heat, you get unrealistically high temperatures just 1.7 billion years ago, which contradicts geological evidence. Adding radiogenic heating pushes those extreme temperatures further back in time and produces thermal histories that match the geological record. The best estimates suggest that radioactive decay currently accounts for somewhere between 45 and 65 percent of the heat Earth loses at its surface, with the remainder being primordial cooling.1Journal of Geophysical Research: Solid Earth. Thermal histories of convective Earth models and constraints on radiogenic heat production in the Earth A more recent analysis frames these two sources as jointly fueling mantle convection, plate tectonics, and even the geodynamo that generates Earth’s magnetic field.2Earth and Planetary Science Letters. Quantifying Earth’s radiogenic heat budget
There is also a meaningful contribution from the core itself. The core is hotter than the overlying mantle, so heat flows outward across the core-mantle boundary, providing an additional push from below. Whole-mantle convection models suggest this core heat flux is substantial: one study found that whole-mantle circulation removes six times more heat from the core over Earth’s history than a layered convection scenario would.3Journal of Geophysical Research: Solid Earth. Modes of mantle convection and the removal of heat from the Earth’s interior The heat flowing out of the core is not uniform, either. It varies dramatically depending on what sits above a given patch of the core-mantle boundary, with heat flux in some spots being close to zero while other areas see enormous outflows.4Geophysical Journal International. Changes in core–mantle boundary heat flux patterns throughout the supercontinent cycle
How Solid Rock Can Flow
One of the most common misconceptions about mantle convection is that the mantle is molten. It is not. Apart from small pockets of melt near the surface, the mantle is solid rock all the way down. What allows it to convect is that at the temperatures and pressures found deep inside Earth, solid rock can deform and flow extremely slowly, on the order of centimeters per year. This is roughly the speed your fingernails grow, except the flowing material is a column of rock thousands of kilometers tall.
The mechanism that makes this possible is called creep, a process where atoms within mineral grains gradually rearrange under sustained stress. Under most mantle conditions, the dominant form of creep involves the movement of defects in the crystal lattice, rather than grains sliding past one another.5Tectonophysics. Mechanisms of high-temperature, solid-state flow in minerals and ceramics and their bearing on the creep behavior of the mantle The practical upshot is that the mantle behaves like an incredibly viscous fluid over geological time. Pour syrup on a cold day and watch it creep across your plate: the mantle does the same thing, just billions of times more slowly.
The viscosity of the mantle is not constant. It changes dramatically with temperature, pressure, composition, and even water content. Laboratory experiments on the mineral olivine, a major component of the upper mantle, have shown that dissolved water can reduce the rock’s viscosity by up to three orders of magnitude, meaning a factor of a thousand.6Earth and Planetary Science Letters. The coupled effects of mantle mixing and a water-dependent viscosity on the surface ocean These viscosity contrasts matter because they change how vigorously the mantle convects: wetter, hotter rock flows more easily, while colder, drier rock resists deformation. This variability helps explain why convection is not a simple, uniform churning but a complex, uneven process with fast lanes and sluggish zones.
Sinking Slabs and the Pull From Below
Textbook diagrams often show convection as a symmetrical loop: hot stuff rises in the middle, cold stuff sinks at the edges. In reality, the sinking side of the circulation dominates. When an oceanic tectonic plate cools as it moves away from a mid-ocean ridge, it becomes denser than the hot mantle beneath it. Eventually, at a subduction zone, the plate plunges back into the mantle. That cold, dense slab sinking under its own weight is arguably the single most powerful force driving plate motions and mantle circulation.
High-resolution flow models have confirmed that slab pull in the uppermost mantle is a major plate-driving force, and that the geometry of plate boundaries strongly influences both the direction and speed of plate motions.7Journal of Geophysical Research: Solid Earth. High‐Resolution Mantle Flow Models Reveal Importance of Plate Boundary Geometry and Slab Pull Forces on Generating Tectonic Plate Motions Earlier global models also found that the net driving forces from subducting slabs and the resistance they encounter during collision roughly balance, suggesting that slab pull is not just one force among many but a central organizer of the whole system.8Journal of Geophysical Research: Solid Earth. A simple global model of plate dynamics and mantle convection
This makes mantle convection a feedback loop rather than a one-way process. The heat from below softens and buoys the rock, but the cold plates at the surface actively participate by sinking and pulling more material along with them. Without that active participation from the plates themselves, the style of convection Earth has today would look very different.
The 660-Kilometer Barrier
About 660 kilometers down, a major phase transition occurs in mantle minerals. At this depth, the crystal structure of certain silicate minerals rearranges due to the increasing pressure. This transition is endothermic, meaning it absorbs heat, and that property creates a physical barrier to vertical flow. When a cold slab sinks toward this boundary, the phase transition is pushed deeper (because the cold slab keeps the mineral in its shallower-depth crystal form to greater depths), producing a mass anomaly that resists the slab’s descent. The same thing happens in reverse for hot upwellings: the phase boundary shifts upward and impedes their rise.9Reviews of Geophysics. Mantle dynamics: Influence of the transition zone
Seismic imaging confirms that many subducting slabs do stall at or near this boundary rather than punching straight through into the lower mantle.10Earth and Planetary Science Letters. Linked influences on slab stagnation: Interplay between lower mantle viscosity structure, phase transitions, and plate coupling Tomographic images show strong, wide high-velocity anomalies (the seismic signature of cold, stiff material) pooling in the transition zone under subduction regions, suggesting the slabs pile up before eventually collapsing downward under accumulated gravitational instability.11Physics of the Earth and Planetary Interiors. Global tomographic images of mantle plumes and subducting slabs: insight into deep Earth dynamics
This raises a big question: does the mantle convect as a single layer from top to bottom, or as two separate layers divided at 660 kilometers? Modeling work strongly favors whole-mantle convection, at least as the dominant mode. Layered convection models struggle to cool Earth efficiently enough because the thermal boundary layer that develops between the two layers acts as an insulator. The parameter range that satisfies known constraints on Earth’s thermal history is far wider for whole-mantle models, and may not exist at all for strictly layered ones.12Geochemistry, Geophysics, Geosystems. Cooling of the Earth: A parameterized convection study of whole versus layered models The current consensus is that the 660-kilometer boundary slows and complicates flow but does not permanently separate the upper and lower mantle into independent convection systems.
Mantle Plumes and Rising Hot Rock
While sinking slabs dominate the downwelling side, the upwelling side features mantle plumes, narrow columns of unusually hot material that rise from the deep mantle toward the surface. These plumes are responsible for volcanic hotspots like Hawaii and Iceland, which sit far from any plate boundary and cannot be explained by plate tectonics alone.
Seismic tomography now images plume-like slow anomalies (indicating hotter, less rigid rock) beneath major hotspot regions, including Hawaii, Iceland, the South Pacific, and Africa.11Physics of the Earth and Planetary Interiors. Global tomographic images of mantle plumes and subducting slabs: insight into deep Earth dynamics Under the South Pacific, a large plume head or cluster of several plumes appears to originate in the lowermost mantle and impinge on the 660-kilometer discontinuity from below.13Earth and Planetary Science Letters. Deep mantle plumes and convective upwelling beneath the Pacific Ocean In the Atlantic, full waveform seismic imaging has revealed that some volcanic island chains are fed by periodically spaced flow channels beneath the lithosphere, each connected to a separate group of plumes rooted at the core-mantle boundary and carrying distinct geochemical signatures.14PubMed Central. Deep mantle plumes feeding periodic alignments of asthenospheric fingers beneath the central and southern Atlantic Ocean
Plumes are not a mirror image of slabs. Slabs are wide, cold sheets; plumes are narrow, hot columns. The asymmetry exists because cold material is stiff and holds together as it sinks, while hot material is less viscous and tends to rise in focused conduits. The overall pattern is that most of Earth’s surface heat loss happens at mid-ocean ridges and subduction zones (driven by slab dynamics), while plumes contribute a smaller but geologically dramatic share of the heat budget, producing island chains, flood basalts, and other volcanic phenomena.
Giant Structures on the Core-Mantle Boundary
Two continent-sized blobs of anomalous material sit on the core-mantle boundary beneath Africa and the Pacific. Known as large low shear-wave velocity provinces, or LLSVPs, these structures are thousands of kilometers wide and extend roughly 1,800 kilometers upward from the base of the mantle.15Geochemistry, Geophysics, Geosystems. Compositional layering within the large low shear‐wave velocity provinces in the lower mantle They slow seismic waves passing through them, which indicates they are hotter, compositionally distinct, or both, compared to the surrounding mantle.
These structures are not passive bystanders in convection. They shape the flow pattern in the deep mantle, redirect plumes, and influence where heat escapes from the core. Convection models that include thermochemical piles matching the LLSVPs produce surface geoid patterns (broad-scale variations in Earth’s gravitational field) that match observations well, with geoid highs over both the Pacific and African LLSVPs.16Journal of Geophysical Research: Solid Earth. The long‐wavelength geoid from three‐dimensional spherical models of thermal and thermochemical mantle convection Adding realistic chemical density variations in the continental lithosphere further improves the match between modeled and observed surface topography.17Journal of Geophysical Research: Solid Earth. Modeling Geoid and Dynamic Topography From Tomography‐Based Thermo‐Chemical Mantle Convection
Research suggests the LLSVPs may be compositionally layered, with a primordial bottom domain near the core and a shallower domain of accumulated subducted basaltic material above it, separated by an abrupt transition.15Geochemistry, Geophysics, Geosystems. Compositional layering within the large low shear‐wave velocity provinces in the lower mantle These piles may also be enriched in heat-producing elements like uranium and thorium, which would make them internal heat sources in their own right. Simulations show that such enriched piles can remain stable for hundreds of millions of years or longer, depending on the balance between their intrinsic density and the extra buoyancy created by their internal heating.18Geochemistry, Geophysics, Geosystems. Effects of Heat‐Producing Elements on the Stability of Deep Mantle Thermochemical Piles At the very base of the mantle, ultralow velocity zones near hotspots reveal even more extreme local thermal and chemical variations, likely caused by internal convection within the LLSVPs themselves.19Journal of Geophysical Research: Solid Earth. Ultralow Velocity Zones at the Core‐Mantle Boundary Near the Caroline Hotspot
How Continents Rearrange the Flow
Continents are thick, buoyant, and thermally insulating compared to oceanic plates. When a large continent or supercontinent sits in one place for a long time, it acts like a blanket, trapping heat beneath it. This insulating effect causes the mantle underneath to warm, eventually spawning focused upwellings. Two-dimensional convection models with moving continents show that if a continent is wider than the natural wavelength of convection cells (roughly 3,000 kilometers, about the thickness of the mantle), it forces neighboring deep upwellings to merge into a single, stronger plume. That focused upwelling can produce a temperature anomaly about twice as large as what develops under typical oceanic lithosphere.20International Geology Review. Cold cratonic roots and thermal blankets: How continents affect mantle convection
This mechanism is thought to play a role in the supercontinent cycle. As heat builds beneath a supercontinent, the resulting upwellings produce flood basalts and continental uplift and eventually rip the supercontinent apart. The breakup redistributes the continental fragments, allowing the mantle to cool again in a different pattern, which in turn reorganizes subduction zones and sets the stage for future continental assembly. In this way, the continents do not simply ride on top of convection; they actively modulate it.
The heat flux at the core-mantle boundary also responds to surface rearrangements. Modeling shows that the locations of mantle upwellings and downwellings create localized heat flux anomalies at the core-mantle boundary that change drastically over Earth’s history as subduction zones shift and supercontinents assemble and break apart.4Geophysical Journal International. Changes in core–mantle boundary heat flux patterns throughout the supercontinent cycle Because the core’s heat loss pattern influences the geodynamo, convection in the mantle indirectly shapes Earth’s magnetic field as well.
Convection on a Cooling Planet
Earth has been losing heat since it formed, and that cooling has changed the style of convection over billions of years. In the Hadean and Archean eons, when the mantle was significantly hotter, it was also less viscous and probably convected more vigorously. Numerical models of early-Earth convection suggest that the tectonic style before modern plate tectonics was dominated by small-scale drips and rifts rather than the large, rigid plates we see today. As the mantle cooled, its viscosity increased and the lithosphere thickened and stiffened, eventually allowing rigid plates to form and modern-style subduction to begin.21Geology. Correlating mantle cooling with tectonic transitions on early Earth
Another factor in the transition to plate tectonics may have been chemical depletion of the earliest mantle. When the first continental roots formed through extensive melting, the leftover mantle material was stiffer than average, which resisted plate-boundary formation. As radioactive heat production declined over time, this stiffening effect weakened, potentially opening the door to true plate tectonics.22Journal of Geophysical Research: Solid Earth. Craton Formation in Early Earth Mantle Convection Regimes The upshot is that convection is not a static process. It evolves as Earth cools, and the style of surface tectonics we see today is a product of the mantle being at a particular temperature and viscosity, not something inevitable for any rocky planet.
Why Earth Has Plate Tectonics and Other Planets Do Not
Venus and Mars have mantles and internal heat, yet neither shows Earth-style plate tectonics. The reason highlights what makes Earth’s convection special. Whether convection can break the surface into moving plates depends on the balance between the stresses that convection generates and the strength of the outer shell. Both of those quantities depend on the planet’s size, thermal state, and cooling history.23Earth and Planetary Science Letters. Conditions for the onset of plate tectonics on terrestrial planets and moons
Mars is smaller than Earth, so it has cooled faster and its lithosphere has grown thick and strong relative to the convective forces beneath it. Venus is closer to Earth’s size but appears to lack plate tectonics today, possibly because its extremely hot surface keeps the lithosphere too weak to form rigid plates, or because the absence of surface water changes the rheology in ways that prevent subduction. Buoyancy-based modeling has estimated maximum mantle temperatures compatible with plate-like behavior: roughly 1,500 °C for Earth, around 1,450 °C for Venus under present conditions, and 1,300 to 1,400 °C for Mars.24Physics of the Earth and Planetary Interiors. Plate tectonics on the terrestrial planets When simulations use strongly temperature-dependent viscosity on a planet without Earth-like plate tectonics, the result is a “stagnant lid” mode: a thick, immobile shell with numerous small upwelling plumes poking at it from below, but no subduction and no large-scale plate motion.25Journal of Geophysical Research: Solid Earth. Low‐degree mantle convection with strongly temperature‐ and depth‐dependent viscosity in a three‐dimensional spherical shell
Earth’s convection is distinctive, then, not just because of the heat driving it but because the surface participates. The plates crack, subduct, and recycle, creating a two-way coupling between the surface and the deep interior. On a planet with a stagnant lid, convection still occurs in the mantle but never efficiently communicates with the surface, leading to a very different geological character.
Seeing Convection Through Seismic Eyes
Nobody has drilled more than about 12 kilometers into the Earth, so almost everything we know about mantle convection comes from indirect observations, especially seismic tomography. Earthquakes send waves through Earth’s interior, and the speed at which those waves travel depends on the temperature, composition, and rigidity of the rock they pass through. By compiling millions of seismic recordings, researchers build three-dimensional maps of where mantle material is seismically fast (usually cold and dense, associated with subducted slabs) and seismically slow (usually hot and less dense, associated with plumes and LLSVPs).
These tomographic images have become increasingly detailed. Fast anomalies throughout the mantle spatially correlate with the expected locations of subducted slabs, a relationship widely used in plate reconstructions and geodynamic modeling.26PubMed Central. Full-waveform inversion reveals diverse origins of lower mantle positive wave speed anomalies Slow-wavespeed domains account for the two antipodal LLSVPs in the lowermost mantle, as well as rising hot material above them and discrete mantle plumes.27EGUsphere. Ambient lower mantle structure and composition inferred from seismic tomography, convection models, and geochemistry The match between reconstructed ancient subduction zones and the positions of deep fast anomalies is one of the strongest pieces of evidence that slabs do sink all the way through the mantle over geological time, even if they stall at the 660-kilometer boundary for millions of years first.
Surface gravity measurements provide complementary evidence. Mantle convection creates broad-scale variations in the geoid, Earth’s gravitational field, because rising hot rock and dense sinking slabs create mass anomalies that tug on the surface. Numerical convection models that are constrained by tomographic data can reproduce the observed geoid pattern remarkably well, which is a strong check that our picture of mantle flow is broadly correct.17Journal of Geophysical Research: Solid Earth. Modeling Geoid and Dynamic Topography From Tomography‐Based Thermo‐Chemical Mantle Convection Between tomography and gravity, the invisible engine beneath our feet has become surprisingly well mapped, even if plenty of details remain fuzzy.