Basal Magma Oceans: Formation & Dynamics

• Basal magma oceans are dense, long-lived molten silicate layers at or near the core–mantle boundary formed through primary crystallization and secondary remelting processes.
• Variations in compressibility, crystallization mode, and composition critically determine their thickness, longevity, and role in planetary differentiation.
• Advanced modeling frameworks reveal their impact on volatile sequestration, magnetic field generation, and seismic structures in Earth, Mars, and rocky exoplanets.

Searching arXiv for papers on basal magma oceans and closely related topics. A basal magma ocean is a molten silicate layer at or near the core–mantle boundary that is gravitationally retained at the base of a rocky planet’s mantle. In the classical picture, it originates during crystallization of a global magma ocean when high-pressure melt becomes denser than, or comparably dense to, coexisting solids and segregates downward; in other formulations it can also arise secondarily through overturn of iron-enriched cumulates or by bottom-up remelting driven by a superheated core after a giant impact. Across Earth, Mars, and rocky exoplanets, the concept refers not to a single universal structure but to a family of basal melt reservoirs whose thickness, longevity, composition, and dynamical role depend on melt–solid density relations, crystallization mode, volatile content, and the thermal coupling between core, mantle, and surface (Nakagawa et al., 12 Dec 2025, Boley et al., 2023, Ballmer et al., 2018).

1. Definition and conceptual scope

In its most general usage, a basal magma ocean is a dense, long-lived molten silicate layer at or near the core–mantle boundary formed during the crystallization of an early global magma ocean (Nakagawa et al., 12 Dec 2025). For super-Earth lava worlds, a more specific definition is used: the basal magma ocean is the melt layer that forms and remains gravitationally stable at the base of the silicate mantle when a planet’s pressure–temperature structure intersects the mantle solidus at both low and high pressures but not in between, producing a layered mantle with a shallow surface magma ocean, an intervening solid rock layer, and a basal magma ocean. That three-layer configuration is denoted MOSMO, for surface Magma Ocean–Solid mid-mantle–basal Magma Ocean (Boley et al., 2023).

The term also has a secondary usage in terrestrial-mantle studies. Even if a “primary” basal magma ocean did not form directly during initial magma-ocean solidification, a “secondary” basal magma ocean can form later when iron-rich cumulates overturn, sink, and partially remelt near the core–mantle boundary. In that setting the basal melt need not be global or continuously present; it may be intermittent and spatially localized, fed episodically by melts derived from small Fe-rich diapirs (Ballmer et al., 2018).

These usages are compatible rather than contradictory. They describe different physical pathways by which silicate melt becomes stabilized at the base of a mantle. This suggests that “basal magma ocean” is best treated as a dynamical end state—persistent basal silicate melt—rather than a single formation mechanism.

2. Formation pathways

The classical mechanism is the liquid–solid density crossover at high pressure. Silicate melts are less dense than solids at low pressure, but they are more compressible, so with increasing pressure their density can equal or exceed that of the crystalline assemblage. In the exoplanet context, this compressibility contrast is explicit: at approximately $5$ GPa melts exhibit approximately $20\%$ compression relative to zero-pressure volume versus approximately $6\%$ for solids, and in super-Earth mantles with core–mantle boundary pressures of several hundred GPa, melt densities can overtake solid densities (Boley et al., 2023). The review literature summarizes the same mechanism as crystallization beginning in the mid–lower mantle where melt becomes denser than coexisting solids, generating an initial basal magma ocean thickness of about $300$–$900$ km depending on crystallization mode (Nakagawa et al., 12 Dec 2025).

A second pathway is cumulate overturn during fractional crystallization. If the global magma ocean crystallizes under fractional conditions, early cumulates are relatively Fe-poor while late-stage cumulates become strongly Fe-enriched and gravitationally unstable. Rayleigh–Taylor overturn transports that dense material downward; partial remelting during descent or at the core–mantle boundary then yields a basal iron-rich melt reservoir (Ballmer et al., 2018). The compositional framework used in the review is the standard partition relation

$$D_i=\frac{C_{\text{solid},i}}{C_{\text{liquid},i}},$$

with incompatible elements $(D_i<1)$ concentrating in the residual melt during fractional crystallization. The residual melt therefore becomes progressively enriched in Fe, heat-producing elements, noble gases, and volatiles, all of which favor a dense and chemically distinctive basal reservoir (Nakagawa et al., 12 Dec 2025).

A third pathway is superheated-core-driven secondary mantle melting after a giant impact. Smoothed Particle Hydrodynamics simulations of Moon-forming collisions show that the impact can generate a superheated outer core and a large upward thermal contrast across the core–mantle boundary. In the canonical Moon-forming scenario, the core-side core–mantle boundary temperature is approximately $11{,}663$ K and the mantle-side temperature approximately $4{,}111$ K, giving $\Delta T \approx 7{,}552$ K. Parameterized melting calculations then yield secondary mantle melting within approximately $20\%$0–$20\%$1 years and a basal melt layer approximately $20\%$2 km thick in the low-viscosity case (Zhou, 18 Nov 2025). In that framework the basal melt layer is the precursor of a basal magma ocean if it remains ponded and gravitationally stable.

3. Physical controls on stability, crystallization, and longevity

Three controls recur across the literature: compressibility, composition, and thermal isolation. Compressibility determines whether a melt can pond at depth at all. Composition then modulates both density and melting relations. Iron enrichment is especially important because iron is incompatible in lower-mantle silicates, partitions into the melt, increases melt density, and lowers melting temperature, thereby stabilizing a basal melt layer over time (Dragulet et al., 1 Aug 2025).

Crystallization mode controls how strong that enrichment becomes. Rapid cooling without an insulating atmosphere favors batch or equilibrium crystallization, whereas slower cooling with an atmosphere can lead to fractional crystallization in late stages, enriching the residual melt in incompatible components and creating strong geochemical and density gradients (Nakagawa et al., 12 Dec 2025). In the terrestrial cumulate-overturn models, the late-stage cumulates can reach $20\%$3, but preservation of a thick global layer with such extreme enrichment is inconsistent with seismic constraints on the present mantle (Ballmer et al., 2018).

Longevity follows from thermal isolation by the overlying solid mantle. Once surface melt has solidified, the remaining basal melt cools slowly because heat must be extracted through solid-state mantle convection. The review expresses this with the scaling

$20\%$4

and notes present-day estimates of core–mantle boundary heat flow $20\%$5 TW (Nakagawa et al., 12 Dec 2025). In Earth-focused transport calculations, the electronic thermal conductivity of iron-enriched silicate liquid increases with iron content and temperature but remains modest enough that conductive heat loss need not suppress convection: using $20\%$6 W/m/K, $20\%$7 K/km, and $20\%$8 km gives $20\%$9 TW, below plausible total BMO heat flows of approximately $6\%$0–$6\%$1 TW (Dragulet et al., 1 Aug 2025).

These results support a common interpretation: a basal magma ocean is most persistent when dense-enough melt is produced early, isolated beneath a high-viscosity mantle, and compositionally driven toward greater iron enrichment as crystallization proceeds.

4. Modeling frameworks and quantitative regimes

The modern literature treats basal magma oceans with several distinct model classes. One-dimensional exoplanet-interior calculations use ExoPlex, a thermodynamically self-consistent solver that integrates mass conservation, hydrostatic equilibrium, an adiabatic temperature profile, Gauss’s law of gravity, and thermally dependent equations of state. In that framework the mantle is Earth-like pyrolite, the core is liquid iron, and any shell above the solidus is treated as fully molten. The model includes anhydrous pyrolite melt, hydrous pyrolite melt with $6\%$2 wt\% $6\%$3, and carbonated pyrolite melt with $6\%$4 wt\% $6\%$5 (Boley et al., 2023).

That model yields three mantle structures: a fully molten mantle magma ocean, a surface magma ocean overlying a solid mantle, and MOSMO. For a $6\%$6 planet, core–mantle boundary pressure is $6\%$7–$6\%$8 GPa; the first appearance of a basal magma ocean occurs at $6\%$9–$300$0 K, melting initiates near the base at approximately $300$1 GPa, and the top of the basal magma ocean shoals to approximately $300$2 GPa by $300$3 K. The mantle becomes fully molten at $300$4 K for anhydrous and carbonated compositions and $300$5 K for the hydrous case (Boley et al., 2023).

A second class comprises moving-boundary geodynamic models of cumulate growth and overturn. In the terrestrial study of magma-ocean crystallization, convection is solved in a two-dimensional box $300$6 km deep and wide while the crystallization front advances at prescribed velocity $300$7. Instability onset follows the scaling

$300$8

and the resulting small late-stage diapirs thermally equilibrate on

$300$9

For $900$0 km, $900$1 Myr, so diapirs equilibrate, melt, and react with ambient mantle during descent rather than reaching the core–mantle boundary intact (Ballmer et al., 2018).

A third class is coupled thermal–magnetic evolution models for Earth-like planets. In that approach, the planet’s internal structure is integrated together with energy budgets for mantle, basal magma ocean, and core. The nominal Earth-like model adopts an initial core–mantle boundary temperature $900$2 K, an initial BMO thickness $900$3 km, and a BMO Fe partition coefficient $900$4. The model predicts a transient basal magma ocean dynamo followed by a core dynamo after $900$5 billion years, with BMO solidification at $900$6 Gyr by the model’s termination criterion (Lherm et al., 2024).

5. Planetary expressions: Earth, Mars, and rocky exoplanets

On Earth, basal magma oceans are invoked to explain early deep-mantle differentiation, but their present-day expression is disputed. Strongly fractionated scenarios predict a thin basal sub-layer with density anomalies at or above $900$7, yet present-day seismic constraints on large low shear-wave velocity provinces indicate density anomalies of only approximately $900$8–$900$9 kg/m$$D_i=\frac{C_{\text{solid},i}}{C_{\text{liquid},i}},$$0, or about $$D_i=\frac{C_{\text{solid},i}}{C_{\text{liquid},i}},$$1–$$D_i=\frac{C_{\text{solid},i}}{C_{\text{liquid},i}},$$2, and volumes of approximately $$D_i=\frac{C_{\text{solid},i}}{C_{\text{liquid},i}},$$3 of the mantle. Ultra-low velocity zones comprise only approximately $$D_i=\frac{C_{\text{solid},i}}{C_{\text{liquid},i}},$$4 by volume. The terrestrial geodynamic inference is therefore that a thick, globally preserved, extremely Fe-rich basal layer is inconsistent with seismic data, whereas moderately Fe-enriched hybrid assemblages plus intermittent basal partial melt are compatible with LLSVPs and some ULVZs (Ballmer et al., 2018). The review reaches a parallel conclusion: the most extreme fractional-crystallization scenarios appear inconsistent with present-day lowermost-mantle geometry and heat-flow requirements, implying moderated enrichment, mixing with recycled crust, or hybrid basal structures (Nakagawa et al., 12 Dec 2025).

Mars is treated as a contrasting case. The review cites seismological support for a present-day BMO approximately $$D_i=\frac{C_{\text{solid},i}}{C_{\text{liquid},i}},$$5 km thick above Mars’ core (Nakagawa et al., 12 Dec 2025). A parameterized-convection study explains the Earth–Mars contrast through compositional buffering by partial melting in the first solid mantle. For the reference static comparison, Earth with bulk $$D_i=\frac{C_{\text{solid},i}}{C_{\text{liquid},i}},$$6, mantle depth $$D_i=\frac{C_{\text{solid},i}}{C_{\text{liquid},i}},$$7 km, $$D_i=\frac{C_{\text{solid},i}}{C_{\text{liquid},i}},$$8 m s$$D_i=\frac{C_{\text{solid},i}}{C_{\text{liquid},i}},$$9, and core temperature approximately $(D_i<1)$0 K yields buffered magma-ocean $(D_i<1)$1 and density anomalies typically $(D_i<1)$2 kg m$(D_i<1)$3, favoring entrainment into thermochemical piles. Mars with bulk $(D_i<1)$4, mantle depth $(D_i<1)$5 km, $(D_i<1)$6 m s$(D_i<1)$7, and core temperature approximately $(D_i<1)$8 K yields buffered $(D_i<1)$9 and $11{,}663$0 kg m$11{,}663$1, stabilizing a global basal silicate layer that may be partially molten (Córdoba et al., 6 May 2026).

Rocky exoplanets extend the parameter space further. For Earth-like compositions at $11{,}663$2 K, magma-rich mantles become denser than solid-only mantles above a crossover of approximately $11{,}663$3 and $11{,}663$4 for anhydrous magma, and approximately $11{,}663$5 and $11{,}663$6 for carbonated magma. This is consistent with the broader conclusion that Earth-like planets with magma oceans larger than approximately $11{,}663$7 or approximately $11{,}663$8 tend to be modestly denser than their solid-only counterparts (Boley et al., 2023). The same study identifies exoplanetary case studies in which most melt resides at depth: at $11{,}663$9 K, WASP-47 e is MOSMO and approximately $4{,}111$0 molten by mantle volume, with approximately $4{,}111$1 of that melt volume in the basal magma ocean; at $4{,}111$2 K, K2-141 b is MOSMO with approximately $4{,}111$3 total melt and approximately $4{,}111$4 of the melt in the basal reservoir (Boley et al., 2023).

6. Volatile, magnetic, tidal, and observational implications

Basal magma oceans are major volatile reservoirs in both exoplanetary and terrestrial-habitability models. In the super-Earth MOSMO regime, the intervening solid layer can mechanically isolate the basal melt from the surface. For a $4{,}111$5 planet, the modeled sequestration capacity exceeds $4{,}111$6 times the mass of water in Earth’s present-day oceans for the $4{,}111$7 wt\% $4{,}111$8 melt equation of state and approximately $4{,}111$9 times Earth’s surface-plus-crust carbon for the $\Delta T \approx 7{,}552$0 wt\% $\Delta T \approx 7{,}552$1 case, with sequestration occurring on billion-year timescales if the MOSMO architecture persists (Boley et al., 2023).

In Earth-mass planets around M-dwarfs, the volatile role is more ambivalent. A $\Delta T \approx 7{,}552$2-D reservoir model distinguishes the surface magma-ocean phase from a later basal magma-ocean reservoir $\Delta T \approx 7{,}552$3. The BMO injects water into the mantle at a constant rate $\Delta T \approx 7{,}552$4, and the resulting surface effect depends strongly on $\Delta T \approx 7{,}552$5. A short-lived basal magma ocean $\Delta T \approx 7{,}552$6 Gyr) can be beneficial to surface habitability because injection into a hot, degassing mantle can temporarily increase surface water. A long-lived basal magma ocean $\Delta T \approx 7{,}552$7–$\Delta T \approx 7{,}552$8 Gyr) instead tends to sequester water in the mantle and diminish surface habitability potential. The study concludes that magma oceans and deep-water cycling can allow inner-habitable-zone late-M-dwarf planets that would otherwise be desiccated if they formed with less than approximately $\Delta T \approx 7{,}552$9 Earth oceans to maintain or recover surface water, but very long-lived BMOs tend to lock water away from the surface (Moore et al., 2023).

Magnetic consequences have become increasingly central. In the coupled Earth-like evolution model, the dynamo criterion is expressed by both positive available entropy $20\%$00 and magnetic Reynolds number

$20\%$01

with a sustained large-scale dynamo requiring $20\%$02. In the nominal case, a BMO with $20\%$03 S m$20\%$04 satisfies $20\%$05 for the first $20\%$06 Gyr, generates a typical internal field $20\%$07 mT, and contributes an average surface dipole field $20\%$08T before the core dynamo becomes dominant after approximately $20\%$09 Gyr (Lherm et al., 2024). Ab initio transport calculations strengthen that possibility by showing that electrical conductivity increases strongly with iron enrichment: at $20\%$10, bulk-silicate-Earth-like melts already exceed $20\%$11 S/m at BMO conditions, whereas at $20\%$12 electrical conductivity exceeds $20\%$13 S/m, electrons carry $20\%$14 of total conductivity, and the predicted silicate-dynamo lifetime extends to approximately $20\%$15 Gyr under mixing-length scaling (Dragulet et al., 1 Aug 2025).

A distinct mechanical consequence is tidal coupling. In a two-layer fluid Earth with a liquid-metal core overlain by a less dense silicate basal magma ocean, lunar tides act directly on the density contrast $20\%$16 at the core–BMO interface. The semidiurnal response resonates when $20\%$17, where $20\%$18 is the natural frequency of the interfacial mode. Near resonance, the model predicts enhanced core-boundary ellipticity, core-flow speed, magnetic Reynolds number, and instability metrics; for plausible early-Earth parameters, $20\%$19–$20\%$20 and laminar viscous dissipation in the BMO reaches approximately $20\%$21–$20\%$22 TW near resonance (Kiernan et al., 15 Jun 2026). This suggests that a basal magma ocean could have augmented mechanically driven pre–inner-core dynamo action.

Observationally, the basal magma ocean remains a constrained but unresolved hypothesis. Seismic low-velocity structures, heat-flow budgets, magnetic history, and atmospheric absence on some hot rocky planets are all consistent with some form of basal melt reservoir, but each dataset permits multiple interpretations. The review literature accordingly identifies several open questions: the long-term coupling between a BMO and mantle convection, the precise relation of BMOs to LLVPs and ULVZs, the time history of $20\%$23, the initial thermal and compositional state after magma-ocean solidification, and the roles of redox evolution and deep volatile cycling (Nakagawa et al., 12 Dec 2025). The present evidence supports basal magma oceans as plausible and, in some planetary regimes, robust outcomes of differentiation; it does not yet reduce them to a single canonical morphology or evolutionary trajectory.