Every rocky world in our solar system began its life as a hellscape. In the wake of accretionary impacts and the heat released by radioactive decay of short-lived isotopes like aluminum-26, the outer hundreds of kilometers of terrestrial planets and large moons existed as global oceans of silicate melt. These magma oceans were not metaphor but reality, persisting for timescales ranging from a few million years on smaller bodies to potentially over 100 million years on Earth, shielded beneath dense protoatmospheres.

What happened during this brief but consequential interval set the stage for everything that followed. The initial differentiation of mantle and crust, the partitioning of volatiles between interior and atmosphere, the establishment of compositional heterogeneities that would later drive mantle convection patterns, all of these emerged from the physics of cooling silicate melts. The Moon's bright highlands, Mercury's strange volatile-rich crust, and the apparent compositional dichotomy of Mars all trace their origins to this primordial epoch.

Yet our understanding remains frustratingly incomplete. The very processes that built planetary architecture have often obscured the evidence of their own operation. By comparing how different worlds preserved or erased their magma ocean signatures, we begin to decode a universal grammar of planetary beginnings, one that may apply equally to exoplanets we will never visit but can characterize through their atmospheric chemistry.

As a magma ocean cools, it does not freeze uniformly. The crystallization sequence depends on pressure, composition, and oxygen fugacity, producing a stratified cumulate pile whose mineralogy varies dramatically with depth. In the deep mantle, bridgmanite and ferropericlase dominate; at intermediate depths, majoritic garnet and ringwoodite appear; and only in the upper few hundred kilometers do olivine and pyroxene become the principal cumulate phases.

The thermodynamics here are unforgiving. As crystals settle out, they extract specific elements from the melt, leaving residual liquids enriched in incompatible elements: potassium, rare earth elements, phosphorus, and the heat-producing isotopes of uranium and thorium. This residual liquid, often called KREEP in the lunar context, becomes progressively more chemically exotic as solidification proceeds. The last few percent of melt to crystallize contains a disproportionate fraction of the planet's incompatible inventory.

Critically, cumulate density does not increase monotonically with depth of crystallization. Iron-rich phases that precipitate late in the sequence, particularly ilmenite-bearing assemblages, can be denser than the underlying cumulates they overlie. This creates the gravitational instability that defines the next chapter of planetary evolution.

The bottom-up versus top-down debate about magma ocean solidification remains unresolved for Earth. Whether crystallization initiated at the core-mantle boundary or at mid-mantle depths, where the adiabat first intersected the liquidus, has profound implications for how primordial heterogeneities were preserved. Seismic anomalies at the base of Earth's mantle, the large low-shear-velocity provinces, may be the fossilized remnants of basal magma ocean crystallization.

On smaller bodies, lower pressures collapse this complexity. The Moon's magma ocean crystallized largely from the bottom up through olivine and orthopyroxene before transitioning to plagioclase-saturated assemblages, producing the simpler but equally informative stratigraphy we sample today.

Takeaway

Crystallization fractionates not just minerals but planetary destinies. The same incompatible elements rejected by early crystals become the engines of later geologic activity.

The lunar highlands present one of comparative planetology's most striking signatures: a globe-encircling crust of anorthosite, a rock composed of more than 90 percent calcium-rich plagioclase feldspar. This is not subtle geochemical evidence requiring inference. It is a primary crust, kilometers thick, visible from Earth as the bright regions distinguishing themselves from the basaltic maria.

The mechanism is elegant. When the residual liquid of a crystallizing magma ocean becomes saturated in plagioclase, the resulting crystals are less dense than the melt surrounding them. Rather than sinking like olivine and pyroxene before them, they float upward, accumulating beneath the cooling surface to form a buoyant lid. The Moon's low pressures kept plagioclase stable to considerable depths, allowing this floatation crust to grow thick and laterally continuous.

On larger bodies, this process is suppressed in several ways. Higher pressures destabilize plagioclase in favor of denser aluminous phases like spinel and garnet, shifting the aluminum budget away from a floating mineral. Vigorous convection in deeper magma oceans may also entrain plagioclase faster than it can rise, preventing the accumulation of a coherent crustal layer.

Mars likely produced some plagioclase floatation, but the evidence is fragmentary, possibly preserved in the ancient southern highlands or buried beneath later volcanic and impact deposits. Mercury skipped this stage almost entirely, its anomalously low iron content and high sulfur suggesting a reduced magma ocean that crystallized a graphite-bearing flotation crust instead, a fundamentally different chemistry yielding a fundamentally different planet.

These divergent outcomes underscore that primary crusts are not universal features but contingent products of size, pressure, redox state, and composition. The exoplanet population almost certainly includes worlds whose primary crusts we cannot yet imagine.

Takeaway

Buoyancy is destiny in a cooling planet. Whether a crystal sinks or floats determines whether it becomes mantle or crust, and that single choice echoes for billions of years.

A magma ocean that crystallizes from the bottom upward leaves behind a stratigraphy that is gravitationally unstable. Late-crystallizing cumulates, enriched in iron and titanium-bearing phases, are denser than the magnesium-rich cumulates beneath them. The mantle of a freshly solidified rocky planet is essentially an inverted density structure waiting to invert itself.

When this overturn occurs depends on the rheology of the cumulate pile, which in turn depends on temperature, grain size, and water content. On the Moon, evidence from KREEP-rich basalts and the asymmetric distribution of titanium-rich mare suggests that overturn was substantial but incomplete, with dense ilmenite-bearing cumulates sinking toward the core-mantle boundary while bringing heat-producing elements deep into the interior.

On Mars, the hemispheric dichotomy and the persistence of distinct geochemical reservoirs in shergottite and nakhlite meteorites argue for either incomplete overturn or for cumulate diapirs that rose without thoroughly mixing the mantle. The Martian mantle preserves chemical fingerprints from magma ocean solidification that Earth's vigorous convection long ago homogenized.

Earth's overturn history is essentially invisible to direct observation, overwritten by four billion years of plate tectonics and mantle stirring. Yet isotopic anomalies in tungsten-182 and neodymium-142 in Archean rocks hint at preserved domains that escaped this mixing, possibly the deep mantle structures we image seismically today. Some of the oldest material on Earth may be magma ocean cumulates hiding in plain sight.

Understanding cumulate overturn matters beyond Earth. It controls the initial thermal state of the mantle, the timing of secondary crust production through partial melting, and the distribution of heat-producing elements that determine whether a planet sustains long-term geological activity.

Takeaway

Planets do not simply solidify and then begin their geological lives. The transition itself is a violent rearrangement, and what survives that rearrangement shapes everything afterward.

The first billion years of a rocky planet are not prelude but foundation. The crystallization of a magma ocean establishes the initial mantle stratigraphy, the primary crust, the depth distribution of heat sources, and the volatile budget of the atmosphere. Every subsequent process, plate tectonics or stagnant lid, persistent volcanism or geological death, plays out against the template laid down in those early millions of years.

Comparative planetology reveals just how sensitive these outcomes are to initial conditions. A slightly different size, a different oxidation state, a different impact history, and the resulting world diverges dramatically. The Moon, Mars, Mercury, and Earth all passed through magma ocean stages, yet emerged as profoundly different worlds.

As we begin to characterize the atmospheres of rocky exoplanets, we are implicitly probing their magma ocean histories. The volatile species we detect, the absence of atmosphere where we expect one, the thermal emission spectra of hot lava worlds: all are echoes of crystallization sequences playing out across the galaxy.