The process in question is called homogeneous ice nucleation. In most planetary atmospheres, water vapour does not condense into ice on its own. Instead, it freezes onto airborne motes of dust, salt or other tiny solids, a process known as heterogeneous nucleation. These foreign particles lower the energetic barrier to ice formation, allowing ice to grow at modest levels of supersaturation, the condition in which the air holds more water vapour than it can stably support. Homogeneous nucleation, by contrast, requires water molecules to organise themselves spontaneously into a stable ice embryo purely by chance collisions. The energy cost of creating a new ice surface from nothing is so high that the vapour must be supersaturated by many hundreds of percent before this becomes statistically likely. On Earth, such extreme supersaturation is essentially never reached in the lower atmosphere, which is why the process is considered a curiosity rather than a working mechanism.

There is, however, one place on Earth where homogeneous nucleation is thought to operate: the polar summer mesosphere, at altitudes near eighty kilometres, where temperatures plunge below minus one hundred and fifty degrees Celsius and noctilucent clouds of tiny ice crystals form. Theoretical studies have also suggested that ephemeral ice clouds could nucleate homogeneously in the upper mesosphere of Venus. Yet homogeneous nucleation from water vapour has never been directly observed in nature on any planet, and atmospheric scientists have generally regarded it as too demanding to matter under realistic conditions. The new work challenges that assumption by pointing to a planet where the conditions are, paradoxically, far more favourable than on Earth, despite Mars having much less water overall.

Mars’s atmosphere is thin, cold and nearly dust-dominated, and it sits well below the triple point of water, meaning liquid water cannot persist and water cycles almost entirely between vapour and ice. Crucially, the Martian atmosphere contains very few condensation nuclei at high altitude. Airborne dust exists, but its abundance drops sharply with height, and the number of available particles per cubic centimetre in the upper troposphere can be minuscule compared with Earth’s aerosol-rich air. When an air parcel rich in water vapour is rapidly cooled in such a particle-poor environment, supersaturation can climb to extreme values before any heterogeneous freezing gets a chance to consume the excess vapour. In that regime, homogeneous nucleation becomes not just possible but competitive, and the model developed by Hernández-Bernal and colleagues shows that it can take over entirely.

The team embedded a detailed scheme of homogeneous ice nucleation from water vapour into meteorological models of Mars, including the Mars Planetary Climate Model developed at the Laboratoire de Météorologie Dynamique and a mesoscale model capable of resolving the winds around individual volcanoes. The mesoscale simulations capture a key dynamical ingredient of the AMEC: a cold pocket of air that forms on the western, leeward side of Arsia Mons in the early morning hours, where orographic flow, gravity waves and the volcano’s enormous height combine to chill the air dramatically. As air parcels traverse this cold pocket, temperatures fall low enough, and supersaturation rises high enough, for water vapour to freeze spontaneously into vast numbers of minuscule ice crystals, each only a fraction of a micrometre across.

The results were striking. With homogeneous nucleation switched on, the model reproduced the distinctive characteristics of the AMEC that had eluded conventional microphysics: the cloud’s extreme elongation, its recurrent diurnal cycle of growth and dissipation, its high altitude near the top of the tropospheric boundary, and its optical brightness. The homogeneously nucleated particles, being numerous and tiny, scatter sunlight efficiently and give the cloud its characteristic appearance, while the dynamics of the cold pocket stretch the freshly formed ice into the long tail that streams away from the volcano. Simulations using only heterogeneous nucleation on dust, the standard assumption for Martian water ice clouds, failed to match the observed cloud, regardless of how the properties of the dust nuclei were tuned. Sensitivity tests showed that the outcome depends delicately on thermodynamic parameters such as the surface energy of ice and the vapour pressure over ice at very low temperature, with small adjustments shifting the cloud’s brightness and extent in ways consistent with the observational constraints.

Those observational constraints come from an unusually rich set of spacecraft data, all openly available. The Visual Monitoring Camera on the European Space Agency’s Mars Express orbiter, originally a humble engineering webcam, has provided a long record of the cloud’s daily evolution, complemented by the high-resolution imagery of the HRSC instrument on the same spacecraft. Images from the Mars Colour Camera aboard the Indian Space Research Organisation’s Mars Orbiter Mission and from the Emirates eXploration Imager on the United Arab Emirates’ Hope probe added further coverage. This multi-mission archive allowed the team to pin down the cloud’s morphology, timing and radiative properties with enough precision to make the model comparison meaningful, and the modelling code and datasets have been released openly for other researchers to scrutinise and extend.

The implications extend well beyond one photogenic cloud. If homogeneous nucleation from water vapour is operating on Mars, then the standard picture of cloud formation on that planet, in which dust grains almost always serve as the scaffolding for ice, is incomplete. High-altitude water ice clouds influence the Martian climate in important ways, affecting the vertical distribution of water vapour, the radiative budget of the atmosphere and even the formation of snowfall. Clouds that nucleate homogeneously consist of far more, and far smaller, particles than heterogeneously nucleated ones at the same water content, which changes how they scatter light and how they sediment. Climate models that omit this pathway may therefore misrepresent the water cycle and its climatic feedbacks, particularly in the cold, clean air above the great volcanoes and in the polar regions.

For Earth, the finding is a provocative nudge rather than a direct overturning. Terrestrial clouds below the mesosphere almost certainly form heterogeneously, thanks to the planet’s abundant aerosols, and the extreme supersaturations required for homogeneous freezing of water vapour are not reached in the troposphere. But the Mars result demonstrates that the process is viable under real planetary conditions, not merely in laboratory chambers, and it encourages a re-examination of assumptions about where and when homogeneous nucleation might matter, including in the mesospheres of Earth and Venus and potentially in the atmospheres of exoplanets with cold, clean, water-bearing air. The study also highlights how laboratory measurements of the vapour pressure and surface tension of supercooled water and ice, performed over the past decade, now feed directly into planetary-scale conclusions.

There is also a lesson in how the discovery was made. The AMEC was long known to observers and had been modelled before, but only by combining a decade of serendipitous spacecraft imagery with high-resolution mesoscale dynamics and a microphysical scheme that most modellers had set aside as irrelevant did the pieces fall into place. A cloud that looked like a mere curiosity of Martian meteorology has turned out to carry a fundamental message about how matter changes phase in extreme environments. As missions continue to monitor Mars’s atmosphere with ever finer instruments, the elongated cloud of Arsia Mons may prove to be the first confirmed natural laboratory for a process that textbooks had filed away as theoretically possible but practically impossible, a reminder that the universe often finds ways to cross thresholds we assumed were out of reach.

Subject of Research: Homogeneous ice nucleation from water vapour in the Martian atmosphere

Article Title: Homogeneous ice nucleation from water vapour suggested by elongated clouds on Mars

Article References: Hernández-Bernal, J., Määttänen, A., Spiga, A., & Forget, F. (2026). Homogeneous ice nucleation from water vapour suggested by elongated clouds on Mars. Nature Geoscience, 19(10), 1213-1217. https://doi.org/10.1038/s41561-026-02089-9

Image Credits: AI Generated

DOI: 10.1038/s41561-026-02089-9

Keywords: Mars, Arsia Mons Elongated Cloud, homogeneous ice nucleation, cloud microphysics, water ice clouds, planetary atmospheres, supersaturation, Nature Geoscience, Martian climate, mesoscale modelling, heterogeneous nucleation, atmospheric science

The team, led by Lamei Zhang and corresponding author Xiaoyu Cai, used a cold metal transfer WAAM system, a variant of the technique that keeps heat input low and reduces spatter and residual stress. Their feedstock was a 1.2-millimetre nickel-aluminium bronze wire deposited onto a 5-millimetre substrate, building single-pass multilayer walls measuring 200 by 6 by 65 millimetres. The innovation lay in how the titanium was introduced. Rather than pre-alloying the wire, the researchers prepared ethanol solutions containing 50-nanometre titanium powder at concentrations of 0.03, 0.05, and 0.1 weight percent, ultrasonically dispersed them, and brushed the solution onto each layer between deposition passes. Once the ethanol evaporated, the next arc pass remelted the surface, dissolving the nanoparticles directly into the molten pool. This in-situ strategy merges manufacturing and modification into a single integrated process, sidestepping the need for separate post-print heat treatments that have historically been the main tool for homogenising WAAM bronzes.

The results revealed a strikingly non-monotonic relationship between titanium content and grain structure, one that defies the simple intuition that more refinement agent always means finer grains. At 0.03 weight percent titanium, the opposite of refinement occurred: average grain size ballooned to 75.2 micrometres, nearly double the 39.1 micrometres of the unmodified printed alloy, and columnar grains proliferated. The explanation lies in a nucleation threshold. Using classical solidification theory, the team calculated that at this low loading, roughly 65 percent of the titanium particles dissolve in the ferociously hot arc molten pool, dropping the effective nucleation site density below the critical value of about 2 times 10 to the 12 per cubic centimetre. With too few active nucleants and a solute drag force far too weak to impede grain boundary migration, the surviving particles could not arrest growth, and the microstructure coarsened.

At 0.05 and 0.1 weight percent, however, the physics flipped decisively. Enough titanium particles survived the thermal assault to push nucleation density past the threshold, triggering prolific heterogeneous nucleation that shattered the coarse columnar regime. Average grain sizes plummeted to 17.3 and 21.6 micrometres respectively. The team also invoked Zener pinning theory, showing that the calculated limiting grain size of about 20.6 micrometres at 0.05 weight percent, after applying a particle distribution correction, matches the measured 17.3 micrometres remarkably well. Electron backscatter diffraction added another dimension: the strong (100) crystallographic texture of the low-titanium sample, with a pole density peak of 7.13, weakened dramatically to 3.13 and then 2.91 as titanium content rose, signalling grains oriented far more randomly and uniformly through the build.

Crucially, transmission electron microscopy confirmed that the titanium did not create any new phases or disrupt the alloy’s delicate multiphase constitution. The printed bronze retained its characteristic alpha matrix, martensitic beta-prime phase, spherical kappa-II precipitates, lamellar kappa-III precipitates, and coffee-bean-shaped kappa-IV iron-rich particles. Energy-dispersive spectroscopy showed the titanium dissolving predominantly into the alpha matrix at concentrations matching the added amounts, while the lamellar kappa-III phase shrank in axial size as titanium content climbed. This means the modification works through grain refinement, solid-solution strengthening, and dislocation pinning rather than by interfering with the thermodynamics of precipitation, preserving the corrosion advantages inherent to conventional nickel-aluminium bronze chemistry.

The mechanical consequences were substantial. Hardness rose by 6 to 8 percent across all titanium-modified samples, reaching peaks near 195.6 HV0.3 compared with 181.2 HV0.3 for the unmodified deposit, and remained remarkably uniform along the build height because each Vickers indentation samples many grains and phases at once. More importantly, tensile testing in both longitudinal and transverse directions showed that the strength-orienting effects of columnar growth had been tamed. The 0.1 weight percent titanium alloy delivered the best overall package: a longitudinal ultimate tensile strength of 718.2 megapascals, up from 664.2 megapascals in the as-deposited alloy, with a yield strength of 372.5 megapascals and 28 percent elongation. The gap between testing directions narrowed to within experimental error, evidence that mechanical anisotropy, the Achilles heel of printed metals, had been effectively suppressed.

Fractography told a coherent mechanistic story. The 0.03 weight percent sample, cursed with coarse columnar grains and a preferred crack path, failed in a quasi-cleavage brittle mode in the transverse orientation, posting the lowest elongation at 20 percent and a yield strength scatter of plus or minus 33.4 megapascals that betrayed deep microstructural heterogeneity. In contrast, the higher-titanium alloys fractured in a ductile, dimpled fashion even in transverse specimens. Pores and cracks in all samples originated at hard precipitate phases and then extended into the alpha matrix, but the refined and spheroidised precipitate morphology in the 0.1 weight percent alloy improved resistance to crack propagation without altering the fundamental initiation mechanism.

Corrosion performance, arguably the decisive metric for marine service, tracked the microstructural improvements with impressive fidelity. After 21 days of immersion in 3.5 weight percent sodium chloride solution, the 0.1 weight percent titanium alloy showed the lowest corrosion rate at 0.167 grams per square metre per day, with mass loss stabilising near 3 milligrams. Its corrosion product layer thinned to 1.83 micrometres, compared with 2.43 and 2.15 micrometres for the lower-titanium variants, and its corrosion interface was straight and uniform, indicating stable passivation. Electrochemical testing confirmed the trend: corrosion current density fell 15.6 percent to 8.49 microamperes per square centimetre, corrosion potential shifted positively by 4.34 millivolts, and the open-circuit potential stabilised in just 18 seconds, the fastest of any sample, reflecting rapid formation of a dense passive film. X-ray diffraction identified the protective products as aluminium oxide and cuprous oxide, the same chemistry that makes bronze a marine workhorse, but now deployed across a finer, more homogeneous microstructure that lengthens the path corrosive ions must travel.

What makes this study resonate beyond one alloy system is the demonstration that microalloying can break the traditional trade-off in which improving one property degrades another. Previous composition tweaks to nickel-aluminium bronze, whether raising nickel content, adjusting the aluminium-to-nickel ratio, or adding tin or chromium, typically boosted strength at the expense of ductility or corrosion behaviour. Here, the coordinated action of grain refinement, precipitate morphology optimisation, and texture weakening allowed strength, ductility, and corrosion resistance to rise together, a synergy the authors connect to the broader concept of escaping performance trade-offs through multi-microstructural feature cooperation. For naval architects and offshore engineers, the practical message is that a titanium addition of 0.05 to 0.1 weight percent, applied during printing rather than after, offers a low-cost route to printed bronze components that could finally match or exceed their cast and forged ancestors, opening the door to large, complex, corrosion-proof marine hardware built directly at the arc.

Subject of Research: Titanium nanoparticle microalloying of wire-arc additive manufactured nickel-aluminium bronze

Article Title: Effect of low-level Ti microalloying on the microstructure, mechanical properties, and corrosion behaviour of WAAM-processed nickel-aluminium bronze

Article References: Zhang, L., Cai, X., Song, B., Dong, B., Lin, S., & Khoshnaw, F. (2026). Effect of low-level Ti microalloying on the microstructure, mechanical properties, and corrosion behaviour of WAAM-processed nickel-aluminium bronze. Advanced Materials Joining, 1(1), Article 17. https://doi.org/10.1007/s44500-026-00016-x

Image Credits: AI Generated

DOI: 10.1007/s44500-026-00016-x

Keywords: nickel-aluminium bronze, wire arc additive manufacturing, titanium nanoparticles, microalloying, grain refinement, corrosion resistance, mechanical properties, cold metal transfer, heterogeneous nucleation, marine engineering, texture weakening, passivation