The oldest lunar highlands are the visible remains of a planetary-scale sorting process. Soon after the Moon formed, much of its rocky exterior was molten. Cooling did not simply freeze that material in place. Crystals appeared in a sequence, some sinking into the interior and others rising toward the surface. The result was a layered Moon capped by a bright, feldspar-rich primary crust.
This lunar magma ocean model explains several major observations at once: the widespread anorthosite of the highlands, chemical reservoirs sampled by later volcanic rocks, and the concentration of unusual elements in the final remnants of melt. Its simplest version is elegant, but newer samples, age measurements, and geophysical data show that crust building was more prolonged and complicated than a single flotation event.
A Moon Hot Enough to Melt
The collision associated with lunar origin released enormous energy. Debris struck and merged, rock was compressed, and radioactive elements supplied additional heat. These processes could melt the Moon to great depth, perhaps across most of its silicate portion. A global or nearly global magma ocean would have convected vigorously as it lost heat to space.
The term “ocean” describes molten rock rather than water. It may also oversimplify the later stages. The early reservoir could have begun mostly liquid and then become a churning suspension of crystals and melt. Its depth, cooling rate, and duration depend on the Moon’s initial temperature, composition, atmosphere of rock vapor, and the insulating effect of any early lid.
Dense Crystals Sank First
Minerals do not all crystallize from magma at the same temperature. Early in the cooling sequence, magnesium- and iron-bearing minerals such as olivine and pyroxene began to form. Because these crystals were denser than the liquid around them, they tended to descend. Their accumulation helped construct chemically distinct layers inside the Moon.
As crystallization continued, the remaining liquid changed composition. Elements readily accepted into early minerals were removed, while others became concentrated in the melt. This process, known as fractional crystallization, means the last liquid can be dramatically different from the original magma even though both belong to the same cooling system.
The neat picture of crystals dropping immediately through a fully liquid ocean is not guaranteed. Strong convection can keep small crystals suspended, and a high crystal fraction turns liquid magma into a viscous slush. Sinking, mixing, compaction, and melt migration may all have operated at different stages.
Feldspar Rose to Make the Highlands
After much of the magma ocean had crystallized, calcium-rich plagioclase feldspar began to form in large quantities. Unlike the earlier dense minerals, plagioclase was less dense than the residual melt. It could rise, collect near the exterior, and solidify into anorthosite, a rock composed overwhelmingly of plagioclase.
This flotation crust provides the foundation of the Moon’s ancient highlands. Their bright appearance reflects the optical properties of feldspar-rich material, especially when compared with the darker, iron-richer basalts that later flooded some large impact basins. Crustal fragments returned from the Moon gave direct support to the magma ocean idea because it is difficult to produce such widespread, plagioclase-dominated terrain through isolated small magma chambers alone.
Modern gravity and topographic measurements indicate that the crust averages only a few tens of kilometers thick and varies substantially between regions. Large impacts have excavated, fractured, mixed, and redistributed it, so the exposed surface is not a pristine flotation layer. Some basins cut deeply enough to reveal material from the lower crust or mantle.
The Last Melt Became Chemically Extreme
Elements that do not fit easily into the structures of the common early minerals remained behind as crystallization proceeded. The final melt became enriched in potassium, rare-earth elements, and phosphorus, producing a component known by the acronym KREEP. It also concentrated heat-producing elements such as uranium and thorium.
This residue accumulated beneath or within the early crust and later participated in magmatism. Its uneven distribution is particularly important on the lunar nearside, where enriched material is associated with a broad region that also experienced extensive volcanism. Heat released by radioactive decay may have helped parts of the interior remain capable of producing melt long after the primary crust formed.
Late-stage dense minerals created another consequence. Once the magma ocean had crystallized, some upper cumulates may have been gravitationally unstable above lighter material. Portions of the mantle could overturn, mixing layers and moving heat-producing reservoirs. Subsequent impacts and basaltic eruptions further obscured the original arrangement.
Why the Simple Model Is Being Revised
The classic sequence—dense minerals sink, plagioclase floats, and KREEP remains—is still a powerful framework. However, lunar anorthosites vary more in composition and apparent age than a brief, uniform crust-forming episode would predict. Samples indicate that related crustal rocks may have formed across a span of hundreds of millions of years. Some material from beyond the Apollo sites is more magnesium-rich or otherwise different from the familiar ferroan anorthosites.
One proposed solution treats the cooling interior as a long-lived slush. Vigorous motion initially suspends crystals. When the solid fraction becomes high enough, viscosity rises sharply, a rigid lid develops, and buoyant melt rich in feldspar components moves upward through the crystal framework. Repeated melt extraction and crustal magmatism could then produce diverse anorthosites over a longer interval rather than assembling the crust in one rapid float-out.
No single sample can decide between these mechanisms, because lunar rocks have often been broken, heated, and welded together by impacts. Far-side anorthosite fragments returned from the South Pole–Aitken region nevertheless provide a powerful new comparison. Their mineralogy and composition resemble ferroan anorthosites from the nearside, supporting the formation of a broadly comparable magma-ocean crust across both hemispheres. One fragment records KREEP-related thermal reworking about 4.41 billion years ago, showing that later modification began extremely early.
The strongest reconstruction combines mineral chemistry, radiometric ages, meteorites, seismic constraints, gravity mapping, and samples from widely separated terrain. What survives beneath the craters is a record of the Moon’s first great transformation: a molten body becoming an internally layered world with a durable crust.