Mercury’s surface may contain only about 37% silicon dioxide by mass—up to 25% less than previous estimates—suggesting its ancient lavas came from a hotter, more extensively melted mantle.

Mercury’s volcanic surface may contain substantially less silicon dioxide than the values long used to describe its crust. A new laboratory calibration, applied to an existing mid-infrared observation, places the abundance at about 37.4 percent by mass. Earlier estimates derived from MESSENGER measurements lay between 49 and 60 percent.

The difference is geologically consequential. Silica is not simply an ingredient to be counted; its concentration records how mantle rock melted and how magma changed before reaching the surface. A silica-poor volcanic crust is consistent with hotter mantle material undergoing a larger degree of melting, potentially at greater depth. It is not, however, a direct temperature measurement or a unique reconstruction of Mercury’s interior.

The estimate comes from a peer-reviewed letter in Planetary Research led by Christian Renggli of the Max Planck Institute for Solar System Research. This is one study, not settled consensus. The calibration was tested against the Moon, where returned rocks offer ground truth, but the Mercury number currently depends on a single Earth-based measurement of one spectral feature. BepiColombo should turn that point estimate into a much more demanding global test.

Mercury’s chemistry has to be read from a distance

No mission has landed on Mercury and returned a rock. Almost everything known about its surface chemistry has been inferred from reflected or emitted radiation, particle interactions and measurements made by passing or orbiting spacecraft. MESSENGER transformed that picture by mapping elemental ratios, but converting elements into actual minerals and rock types requires assumptions about how those elements are bonded.

That distinction is unusually important on Mercury. The planet formed in highly reducing, oxygen-poor conditions. Silicon that would ordinarily be counted as silicon dioxide in an Earth-like rock can instead occur in metallic or carbide phases. A measurement of total silicon is therefore not automatically a measurement of SiO2.

Mercury’s volcanic plains cover much of the surface and preserve eruptions from early planetary history. Volcanism declined dramatically within roughly the first billion years after formation, leaving ancient lavas exposed to impacts and space weathering. Their composition is one of the few surviving records of the mantle that produced them.

The Christiansen Feature acts like a compositional marker

Renggli and colleagues used the Christiansen Feature, or CF, in the mid-infrared. At this wavelength, a material’s refractive index approaches that of its surroundings. The feature appears as a reflectance minimum or an emission maximum, and its precise position shifts with the structure and chemistry of silicate material.

Silicon and oxygen bonds strongly influence the spectrum between roughly 7 and 14 micrometres. In the glass standards assembled for the study, the CF moved from about 7.4 micrometres in material containing 97.6 percent SiO2 to about 9.9 micrometres in material containing only 0.5 percent. Lower silica generally pushed the feature to longer wavelengths.

Earlier work often represented this relationship with a straight line. The expanded dataset shows that the extremes curve away from that approximation. The authors fitted a second-order polynomial with an R-squared value of 0.957, compared with 0.926 for a linear fit to their internally consistent data and 0.772 for one earlier linear dataset.

Those statistics do not make the method exact. Scatter around the curve corresponds to an estimated uncertainty of plus or minus 3.4 percentage points of SiO2. Environmental effects and uncertainty in the observed CF wavelength must also be considered when the method is applied to another world.

Seven glass beads extended the calibration

The team synthesised seven usable glass-bearing samples from reagent-grade oxides. Their compositions spanned almost the entire possible range, from 0.5 to 97.6 percent SiO2. To reach the most extreme compositions, a carbon-dioxide laser focused onto powdered material and heated it above 2,000 degrees Celsius for about 30 seconds.

The resulting beads were around 500 micrometres in diameter. Some contained unmelted material or crystals formed during rapid cooling, so the researchers used electron microscopy and Raman spectroscopy to locate homogeneous, fully glassy areas. Micro-FTIR measurements then recorded spectra from spots only 25 to 40 micrometres across, averaging 512 spectra for each measurement.

Glass is useful here because it isolates broad compositional changes without a forest of overlapping mineral peaks. It is also relevant to airless bodies: impacts, explosive volcanism, lava and space weathering all produce glass on the Moon, and related processes operate on Mercury. It is still an analogue, not a physical sample of Mercury’s regolith.

The Moon served as the calibration’s reality check

A relationship measured in a laboratory matters only if it performs on a planetary surface. The team therefore applied its equation to corrected CF data from the Diviner radiometer aboard NASA’s Lunar Reconnaissance Orbiter. The source map covered every longitude from 70 degrees south to 70 degrees north at 32 pixels per degree.

The derived silica map recovered the Moon’s familiar two-part structure. Dark volcanic maria averaged 45.8 plus or minus 0.9 percent SiO2, while highland terrain averaged 50.9 plus or minus 1.6 percent. It also identified locally evolved volcanic regions, reaching as high as 76 percent at Lassell Massif.

Most importantly, the orbital estimates could be compared with material brought to Earth by Apollo, Luna and Chang’e missions. Some sites matched closely. Others differed because an orbital averaging circle covers tens of thousands of square kilometres and mixes local soils with distant ejecta, whereas a returned sample records a much smaller place. The method reproduced the broad mare-highland contrast and the expected compositional ranges without being tuned to every landing site.

The spectra, calibration inputs and lunar map are preserved in the authors’ open Zenodo dataset. That lunar exercise is the strongest part of the validation. It does not eliminate the extra uncertainty involved in transferring the same relationship to Mercury.

One 8.5-micrometre observation becomes 37.4 percent

Mercury does not yet have a comparable global CF map. The paper uses an Earth-based mid-infrared observation published in 2000 that placed the feature near 8.5 micrometres. Insert that wavelength into the new calibration and the result is 37.4 percent SiO2 by mass.

The previous range of 49 to 60 percent was not a direct oxide measurement. It was calculated from elemental ratios recorded by MESSENGER’s X-ray spectrometer, combined with an absolute silicon abundance from its gamma-ray spectrometer. That silicon value, 24.6 plus or minus 7 percent, carried a large uncertainty. Converting the elemental inventory into oxides also required assumptions about valence states.

The Max Planck Institute’s account of the study describes the new value as up to 25 percent lower than previously thought. The underlying numbers deserve precision. A value of 37.4 is about 24 percent lower than 49, and about 38 percent lower than 60. In absolute composition, the difference is roughly 12 to 23 percentage points. The headline formulation follows the institution’s description; the paper supplies the fuller comparison.

There is another reason the earlier calculation may run high. MESSENGER found an unusually low oxygen-to-silicon ratio. If some surface silicon is hosted in reduced metal or carbide rather than oxidised silicate, treating all of it as SiO2 overstates the oxide abundance. The new infrared result is consistent with that reduced chemistry.

Why less silica implies more extensive melting

Silica concentration changes as a mantle and its magmas evolve. During progressive cooling and crystallisation, early minerals remove some components while SiO2 becomes increasingly concentrated in the remaining liquid. Later, smaller-degree melts can inherit that enriched signature. Conversely, very extensive melting of hot mantle material draws in more of the source and can produce more silica-poor magma.

This is why the authors interpret 37.4 percent as evidence that Mercury’s surface lavas came from mantle material melted more thoroughly, at higher temperature and possibly greater depth than previous estimates suggested. Space Daily covered an earlier experimental study arguing that Mercury’s lavas rose from deep within the planet. The new CF calibration offers a separate line of evidence that pushes in a similar direction.

The inference should not be mistaken for a thermometer. The paper does not assign a new mantle temperature, melting pressure or eruption depth. Those values depend on the starting composition, pressure, oxygen availability, sulfur and carbon chemistry, and the minerals left behind in the mantle. A lower silica estimate narrows the acceptable histories; it does not select one by itself.

Nor does 37.4 percent mean Mercury is missing the balance of its elemental silicon. The number describes silicon dioxide as an oxide fraction in the optical surface. Reduced silicon can be present in other chemical forms, while the crust and mantle beneath the observed skin may differ.

MESSENGER’s chemical provinces do not disappear

One global-looking number can obscure a geologically diverse planet. MESSENGER found that magnesium, aluminium, sulfur, calcium and silicon ratios vary across Mercury. When Space Daily reported the mission’s new surface-composition maps in 2015, the important result was the separation of distinct geochemical terranes rather than a single average rock.

The new work does not replace those maps. It supplies an independent conversion between a mid-infrared feature and SiO2 abundance. X-ray spectroscopy is sensitive to elemental ratios; MERTIS-style infrared spectroscopy constrains minerals and bonds. If the methods can be reconciled, they should show both how much silicon is present and how much of it actually occurs as silica-bearing silicate.

Regional variation is one of the major unresolved questions. The 8.5-micrometre CF is not a high-resolution global survey. Mercury could contain areas substantially richer or poorer in silica than the point estimate suggests, just as the lunar map contains basaltic maria, highlands and rare silica-rich volcanic centres.

BepiColombo can turn the estimate into a planetary map

The decisive observations should come from the Mercury Radiometer and Thermal Infrared Spectrometer, MERTIS, aboard ESA’s Mercury Planetary Orbiter. The instrument covers 7 to 14 micrometres, including the full region needed to locate the CF, and can reach a spectral resolution of about 90 nanometres. Its objectives include mapping rock-forming minerals, surface temperature and thermal inertia.

Space Daily reported MERTIS’s first thermal-infrared observations of Mercury during BepiColombo’s cruise. Those flyby measurements demonstrated the instrument, but routine orbital mapping will be far more powerful. The mission’s transfer module separated successfully on September 3, 2026, and ESA currently lists orbital capture for November 21.

After the two orbiters separate and the Mercury Planetary Orbiter reaches its final path, routine science is scheduled to begin on April 6, 2027. The ESA BepiColombo factsheet lists MERTIS alongside MIXS, an imaging X-ray spectrometer that will produce new elemental-ratio maps. Used together, the instruments can test the oxide assumptions that complicate the MESSENGER-era values.

The strongest result is a prediction that can soon fail

The CF position is affected by more than composition. Temperature, vacuum, grain size, irradiation, viewing geometry and surface texture can shift the signal. The lunar input map was corrected for illumination, topography, geometry and space-weathering effects, yet artefacts remained at high latitudes. Mercury’s surface reaches very different temperatures and has endured a harsher solar environment.

The study’s use of homogeneous glass also simplifies a real regolith made from minerals, impact products and mixtures at many grain sizes. That simplification helped isolate the SiO2 relationship, but it means the plus-or-minus-3.4-point calibration uncertainty is not the complete uncertainty budget for Mercury.

The 37.4-percent estimate is therefore best treated as a well-motivated prediction from a newly tested method, not a final inventory of the planet. Its credibility comes partly from being exposed to a near-term check. If BepiColombo maps a CF consistent with low silica across the volcanic plains, Mercury’s apparently cold, static surface will preserve evidence of a hotter and more thoroughly melted ancient mantle. If it does not, the global data will show where the one-measurement inference broke down.