China’s Zhurong rover crossed a thin layer of flat, hydrated rocks in southern Utopia Planitia. New analysis argues that their branching internal patterns and chemistry are best explained by large selenite crystals, a clear, well-formed variety of gypsum that grows directly from concentrated water.
If that interpretation is right, Mars maintained an active water system far later than the era of its famous rivers and lakes. The crystals may also contain fluid inclusions, microscopic droplets sealed during growth. No such inclusion has been observed on Mars, however, and no droplet was measured by the rover. This is one study, not settled consensus.
The distinction matters. Zhurong found evidence for the mineral and its crystal habit at rover scale. The tiny archive imagined inside it remains a testable prediction, one that would require microscopy, in-place chemical analysis or sample return.
Zhurong combined crystal shape with chemistry
The Nature Astronomy paper reports decimetre-scale crystal forms inside a laterally extensive layer of platy rock only 5 to 10 centimetres thick. Earlier observations had established that these rocks contained hydrated material, but the exact phase was uncertain.
Zhurong’s short-wave infrared spectrometer recorded absorption bands near 1.45 and 1.95 micrometres, associated with water and hydroxyl, plus a set of additional features that matched laboratory spectra of gypsum. Its laser-induced breakdown spectrometer detected sulfur, calcium and hydrogen in every platy-rock target. The sulfur and calcium peaks closely resembled an onboard gypsum reference, while their correlated abundances supported calcium sulfate dihydrate, CaSO4·2H2O.
The images supplied another line of evidence. Some rock interiors displayed radiating crystal clusters. Others showed branching “Christmas-tree” patterns resembling the fishtail twinning of terrestrial selenite. The material appeared internally uniform and lacked the sand-rich rosette structure expected of gypsum desert roses. A detailed 2026 conference summary lays out how crystal habit, texture, infrared signatures and elemental chemistry converge on the selenite interpretation.
A primary evaporite is more than a sulfate stain
Gypsum has been identified elsewhere on Mars, but sulfate can appear in several geological roles. Mineral-rich groundwater can cement pre-existing sediment, fill fractures or leave fine-grained alteration products. None of those settings necessarily requires a pond at the surface when the mineral formed.
The new paper interprets Zhurong’s layer as a primary evaporite. The large euhedral crystals, uniform composition and thin, continuous geometry are more consistent with growth at or near the bottom of a briny water body than with later cement or fracture filling. “Euhedral” means the crystals developed recognisable external faces rather than simply occupying whatever pore space was available.
That interpretation strengthens, but does not replace, earlier reports. A 2022 Science Advances study used Zhurong images and spectra to identify hydrated sulfate or silica materials in duricrusts. Work published the following year found potential Amazonian brine activity and present water-vapour cycling in data from the rover’s first 110 sols. SpaceDaily also reported Zhurong’s evidence for much younger saline-water effects on nearby dunes. The selenite claim concerns a different layer and a different episode.
The water could have arrived many times
A mass-balance calculation links the mineral layer to a substantial amount of water. Depending on the initial brine concentration, precipitating 5 to 10 centimetres of gypsum would require a cumulative water-equivalent column of at least 6.25 to 25 metres. “Cumulative” is essential: the site did not need a 25-metre-deep lake all at once.
The authors propose sustained or episodic upwelling into a shallow basin. At any one time, the liquid layer may have been a metre deep or less and mostly covered by ice. Repeated delivery and freezing could process a much larger total volume through that small body.
Southern Utopia Planitia contains pitted cones, troughs, polygonal ground and structures interpreted as mud volcanism or dike systems. In the proposed sequence, magma intruded into a volatile-rich cryosphere, melted buried ice and helped drive ion-rich groundwater toward the surface. The model is plausible within that geological setting, but Zhurong did not observe the ancient pond or its plumbing directly.
Freezing provides the concentrating mechanism. As relatively pure ice forms, most dissolved ions remain in the liquid. The residual water becomes progressively saltier, a process called cryoconcentration. Magnesium in the brine can suppress the nucleation of calcium sulfate, allowing fewer crystals to grow larger before the last liquid disappears or drains away.
About 757 million years is recent only for Mars
The time claim requires care. Crater counts assign the underlying Vastitas Borealis Formation a model age near 3.2 billion years. Counting smaller craters yields a resurfacing age of about 757 million years, with the conference analysis reporting an uncertainty of roughly 66 million years.
The gypsum-bearing layer lies immediately beneath a thin sand mantle. The researchers did not see a thick overburden that would suggest ancient material had been deeply buried and then excavated. They therefore associate the near-surface evaporite with the younger resurfacing event rather than the much older substrate.
This is not a radiometric age measured from an individual crystal. Crater-count dating estimates how long a surface has accumulated impacts and depends on models that translate crater density into time. The link between the resurfacing and crystal growth is a geological interpretation. “Roughly 760 million years old” is a defensible shorthand for the host unit, not a direct laboratory date for the gypsum.
Even with that qualification, the timing is striking. Seven hundred and sixty million years is ancient by terrestrial standards, but it falls late in the Martian Amazonian period, long after Mars lost the warm, persistently wet conditions usually associated with its early history. The result points to local, episodic liquid water in an otherwise cold world, not a return to a planet-wide temperate climate.
The microscopic brine remains a prediction
Large gypsum crystals on Earth commonly trap small portions of their parent solution as they grow. These fluid inclusions can line growth bands or occupy tiny cavities inside the crystal. If they remain closed, they preserve a sample of water rather than merely the solid minerals left after water vanished.
A review of fluid inclusions in Mars-analogue chemical sediments explains why they attract astrobiologists. Terrestrial gypsum and halite inclusions can retain dissolved salts, gases, organic compounds and, in some environments, cellular material. They are compact records of the hydrosphere and sometimes the biosphere at the moment of crystallisation.
Zhurong did not have the optical or sampling equipment needed to see such pockets in these rocks. It did not cut a transparent section, focus through a crystal, pierce an inclusion or analyse its liquid. The paper predicts that large Martian selenite is likely to contain inclusions because comparable terrestrial crystals commonly do. “May contain” is the scientifically important wording.
The same caution applies to biology. No organism, fossil, organic molecule or biosignature was detected in the proposed selenite. A brine could have been chemically habitable without ever being inhabited. Finding a sealed droplet would first be a geological and geochemical result.
A returned crystal could test the whole story
An intact inclusion could reveal the water’s salinity, acidity, dissolved elements and gases. Isotope ratios could help distinguish shallow melted ice from deeper groundwater and reconstruct how freezing or evaporation modified the original solution. Multiple inclusions along growth zones might record changes during successive episodes.
Testing that archive would be difficult. Researchers would need to establish that an inclusion formed with the crystal rather than in a later fracture, that it remained sealed, and that any organics were Martian rather than contamination from spacecraft or laboratory handling. Gypsum can also dehydrate or recrystallise, processes that may disturb an old inclusion.
The shallow setting cuts both ways. Large crystals and a 5-to-10-centimetre layer provide some shielding, but material close to the surface is exposed to oxidants and ionising radiation. The uppermost gypsum is also relatively transparent to ultraviolet light. Pockets deeper inside a crystal or layer would offer the better preservation target.
A Mars Reconnaissance Orbiter view of Utopia Planitia shows how varied this broad northern basin is; the image is regional context, not the Zhurong outcrop itself. Locating, approaching and sampling the rover-scale selenite layer would require a future mission designed for fine mineralogy and clean collection.
For now, the result is a chain of increasingly ambitious inferences: the spectra and chemistry indicate hydrated calcium sulfate; the morphology points to primary selenite; primary selenite implies concentrated liquid water; terrestrial selenite suggests fluid inclusions may survive inside. Every link is testable. Only the final step, opening or scanning a Martian crystal, can show whether a microscopic sample of that late water is still there.