Mercury’s present-day surface preserves the aftermath of processes that began when the innermost planet was still largely molten. A new experimental and numerical study has reconstructed how a sulfur-rich Mercurian magma ocean may have crystallized into a chemically and mineralogically layered mantle, potentially determining where later melting occurred and how Mercury developed its volcanic crust.
Researchers Fabrizio Saracino, Bernard Charlier, Yishen Zhang and Olivier Namur combined high-pressure crystallization experiments with fractional crystallization modeling to investigate the early differentiation of Mercury. Their results show that the structure produced as the magma ocean cooled depended strongly on its initial magnesium-to-silicon ratio, while sulfur affected both the minerals that formed and the density of the remaining liquid.
Recreating Mercury’s Early Magma Ocean
Mercury is thought to have experienced extensive melting early in its history as energy from accretion, large impacts, metal-silicate separation and radioactive decay heated the young planet. The resulting global silicate magma ocean eventually cooled and crystallized, creating the primordial mantle from which later volcanic activity could draw material.
Reconstructing that process is difficult because Mercury is chemically unusual. Measurements from NASA’s MESSENGER mission revealed a surface that is extremely poor in iron oxide but comparatively rich in sulfur, indicating that the planet formed and differentiated under highly reducing chemical conditions.
To reproduce those conditions, the researchers conducted experiments on sulfur-rich silicate melts at temperatures between 1,125°C and 1,525°C and pressures from 0.5 to 1.5 gigapascals. The experiments reached oxygen fugacities from about 3.7 to 8.4 log units below the iron-wüstite equilibrium, representing the highly reduced environment expected for Mercury.
The experiments were integrated with previous high-pressure results and a numerical model that followed the magma ocean as crystals were progressively removed from the remaining liquid. The modeled magma ocean initially extended to approximately 480 kilometres depth.
Two Starting Compositions Produce Very Different Deep Mantles
A major uncertainty is the exact composition of Mercury’s original silicate portion, known as Bulk Silicate Mercury. The study therefore examined two possible compositions inspired by enstatite chondrites and distinguished mainly by their magnesium-to-silicon ratios.
The lower-Mg/Si composition, called Mer8, has an Mg/Si ratio of 0.88. In this case, enstatite crystallizes first and produces an approximately 90-kilometre-thick basal orthopyroxenitic layer.
The higher-Mg/Si composition, Mer15, has an Mg/Si ratio of 1.16. It begins by crystallizing forsterite and produces a much thicker basal layer dominated by this mineral, corresponding to roughly 300 kilometres of dunite in the model.
Despite these very different starting paths, the compositions of the residual liquids eventually converge as crystallization proceeds. Once this happens, forsterite and enstatite crystallize together and the later evolution of the two magma-ocean scenarios becomes increasingly similar.
A Fertile Upper Mantle Emerges as the Planet Cools
The experiments reveal a distinct sequence of minerals appearing as the remaining magma becomes progressively more evolved. Clinopyroxene joins the crystallizing assemblage when the residual melt fraction falls to about 0.40 in Mer8 and 0.35 in Mer15. Quartz appears later, at melt fractions of approximately 0.24 and 0.28, respectively, followed by plagioclase at about 0.14 and 0.19.
The appearance of clinopyroxene is particularly important because the researchers use it to distinguish between two major mantle reservoirs. The deeper portion of the primordial mantle is comparatively refractory, meaning that it requires higher temperatures to begin melting. Above it, clinopyroxene-bearing rocks form a more fertile mantle that can generate melt more readily.
In the models, clinopyroxene-bearing fertile mantle begins forming at approximately 160 to 190 kilometres depth. This leaves a roughly 140-kilometre-thick fertile reservoir in the upper part of the modeled primordial mantle.
The Fertile Mantle Could Explain Mercury’s Volcanic Crust
Mercury’s surface contains extensive volcanic terrains, demonstrating that substantial amounts of mantle material melted after the original magma ocean had solidified. The new model suggests that the fertile portion of the primordial mantle alone could have produced a major share of that crust.
The researchers calculated how much crust could result if the approximately 140-kilometre-thick fertile reservoir underwent partial melting. At a melt fraction of 0.30, the model produces enough material to generate a volcanic layer about 40 kilometres thick.
That result is consistent with a previously published estimate of Mercury’s crustal thickness of 35 ± 17 kilometres. If higher degrees of partial melting of around 0.50 to 0.60 occurred, the calculated crustal thickness rises to approximately 60–80 kilometres.
The calculation does not require the entire Mercurian mantle to have melted uniformly. Instead, it supports a scenario in which the more easily melted upper reservoir became a principal source of Mercury’s volcanic crust.
The authors caution that Mercury’s mantle may subsequently have experienced density-driven overturn. Such large-scale rearrangement could have moved or mixed the original mantle layers, potentially creating additional melt sources that were not represented by the initial post-magma-ocean structure.
Sulfur Changes the Crystallization Path
Sulfur is not simply an additional chemical ingredient in the model. The experiments indicate that it substantially changes how Mercury’s magma ocean evolves.
Under the highly reduced conditions investigated, sulfur lowers the effective activities of magnesium oxide and calcium oxide in the melt. This slightly delays the formation of clinopyroxene and plagioclase. At the same time, sulfur increases silica activity, allowing quartz to appear relatively early during magma-ocean crystallization.
This produces a liquid evolution that differs substantially from sulfur-free models. Even magma-ocean compositions that begin with relatively high Mg/Si ratios can evolve toward residual liquids that are increasingly silica-rich and magnesium-poor.
The result may help explain why models that omit Mercury’s unusually high sulfur abundance can predict different primordial mantle mineralogies.
Dense Sulfides May Have Been Stored Inside Mercury
The study also examined how sulfur changes the density of the remaining silicate liquid. The calculations suggest that dissolved sulfur makes Mercury’s magma ocean less dense. In one example, approximately 10 weight percent sulfur lowers the calculated silicate-liquid density by about 210 kilograms per cubic metre.
The authors emphasize that the volumetric behaviour of sulfur in reduced silicate melts remains poorly constrained, so this part of the model carries significant uncertainty and requires further experimental testing.
Even so, the density calculations produce an important consequence. Graphite remains less dense than the residual magma and could therefore float, supporting earlier models in which graphite contributed to Mercury’s primordial flotation crust.
Quartz and plagioclase, by contrast, remain denser than the modeled sulfur-bearing liquid and would not be expected to form a similar flotation crust.
Magnesium- and calcium-bearing sulfides are also generally dense relative to the evolving magma. Rather than rising toward the surface, these sulfides would tend to become incorporated into or sink through the crystallizing mantle. In the high-Mg/Si scenario, some sulfides near the core-mantle boundary can initially have densities very close to that of the liquid, but the calculated density contrast is small and the study argues that efficient flotation would have been difficult.
Sulfur May Have Reshaped Mercury’s Radioactive Heat Budget
Where sulfides accumulated could also affect Mercury’s long-term thermal evolution because uranium, thorium and potassium supply heat through radioactive decay.
The researchers modeled how these heat-producing elements would be distributed as the magma ocean crystallized. In most scenarios, uranium, thorium and potassium behave largely as incompatible elements, meaning they preferentially remain in the residual liquid rather than entering the earliest crystallizing minerals.
As crystallization progresses, the late-stage liquid therefore becomes increasingly enriched in heat-producing elements. The study proposes that this process could have produced an upper-mantle reservoir analogous to a KREEP-like layer on the Moon, positioned between late-forming mantle cumulates and Mercury’s primordial graphite crust.
Exactly how uranium and thorium partition into Mercury’s unusual sulfide minerals remains uncertain. If these elements entered sulfides more readily than some experiments indicate, appreciable amounts of radioactive material could instead have been stored deeper in the intermediate and lower mantle. The paper therefore treats the distribution of heat-producing elements as a set of possible scenarios rather than a uniquely determined structure.
BepiColombo Could Test the Interior Model
The study provides predictions that can ultimately be compared with improved measurements of Mercury’s surface chemistry and mineralogy. The authors identify observations from the ESA-JAXA BepiColombo mission, including compositional and mineralogical measurements, as a way to test models of the planet’s differentiation and crustal formation.
Such observations could help constrain which modeled Bulk Silicate Mercury composition most closely resembles the planet’s original mantle and clarify how strongly sulfur influenced its evolution.
A More Complex Picture of Mercury’s Interior
The results do not establish a single definitive structure for Mercury’s mantle. The two starting compositions intentionally represent different plausible chemical scenarios, and important uncertainties remain in sulfur-bearing melt density, sulfide behaviour, heat-producing-element partitioning and any mantle overturn that occurred after magma-ocean crystallization.
Nevertheless, the experiments produce a consistent physical picture in which Mercury’s early magma ocean generated a deeply refractory mantle, a shallower fertile reservoir capable of producing volcanic crust, and sulfur-bearing cumulates that may have controlled both mineral distribution and radioactive heat storage. The work provides an experimentally anchored framework for connecting Mercury’s ancient molten state with the planet observed today.