Neptune’s Inner Moons Are Smashed Worlds

Triton outweighs all fifteen of Neptune’s other moons put together, two hundred times over, so what happened to the moons that came before it?

Welcome to FreeAstroScience. We opened a new Science Advances paper expecting Neptune’s small moons to be made of ice, and instead JWST found clay, a mineral family that needs millions of years in warm liquid water, on bodies sitting near 50 kelvin. What follows is our best account of where that water once was.

Neptune’s inner moons and rings are reaccreted fragments of one or more large icy bodies destroyed when Triton arrived, most likely Neptune’s original regular satellites. James Webb Space Telescope spectra published in Science Advances on 29 July 2026 show magnesium-rich clay minerals on Larissa, Galatea, and the rings, minerals that form only after millions of years in liquid water inside a large differentiated body. None of the observed moons or rings shows any trace of water ice, and that absence remains unexplained.

Fifty kelvin is colder than any freezer humans have built for a kitchen, and it is roughly the temperature at which Neptune’s inner moons have spent their whole existence. Serpentine clays want the opposite: liquid water at up to 300 or 400 kelvin, held steady for somewhere between one and ten million years. Finding such minerals on Larissa and Galatea is like finding river mud caked onto a comet. Something was warm for a very long time, and M. Ryleigh Davis and colleagues at Caltech argue in the new paper that the warmth belonged to the deep interiors of worlds which no longer exist.

What did JWST see on Neptune’s inner moons?

NIRSpec, Webb’s near-infrared spectrograph, stared at Proteus for 39 minutes, at Larissa for 92, at Galatea for 136, and at the faint rings for 3 hours and 14 minutes combined. Coverage ran from 0.6 to 5.3 micrometers. Every spectrum came back odd in the same two ways. All four surfaces absorb close to 70 percent of the light near 3 micrometers, seven photons in every ten, and in the paper’s comparison chart of surfaces without an ice signature, that depth makes the Neptunian set the clear outlier. Yet the bands frozen water always leaves are missing: nothing at 1.5, 1.65, 2.0, or 4.5 micrometers, and no Fresnel peak at 3.1.

On the two smaller moons and the rings, one further absorption sits on the short-wavelength side of that hydration band: a sharp, checkmark-shaped dip at 2.72 micrometers, the signature of magnesium-rich serpentine clays. Inside the RELAB meteorite library, its closest matches are CM carbonaceous chondrites such as Meteorite Hills 00639 and Cold Bokkeveld. Those are petrologic type 2.0 to 2.1, the most thoroughly water-processed members of their class. Closer still, in band center and shape, is the asteroid Ceres. Proteus mostly sits this out. Its clay contribution is weak or absent, and it alone carries a faint carbon dioxide band at 4.27 micrometers, marginally stronger on its trailing side.

Table 1 — What JWST NIRSpec found on Neptune’s inner satellites and rings (Davis et al., Science Advances, 29 July 2026)

Clay like this cannot form where these moons live

Magnesium-rich serpentine grows when rock soaks in liquid water for one to ten million years, at temperatures no higher than about 300 to 400 kelvin. Nothing in Neptune’s neighborhood offers those conditions today. Could the moons have brewed it internally? Interior models of the Uranian moons by Castillo-Rogez and colleagues (2023) supply the benchmark. Miranda, at roughly 470 kilometers across, will never generate enough heat to melt its own water ice under most formation scenarios, and Larissa and Galatea are about 2.5 times smaller than Miranda.

Impacts do not rescue the idea either. Shock experiments by Furukawa and colleagues (2011) show collisions can hydrate rock locally. But a crater 20 to 30 kilometers wide, already a sixth of Larissa’s diameter, raises local temperatures by only 100 to 200 kelvin. Starting from 50 kelvin, that still fails to melt ice. Hydrothermal plumbing of the kind such craters feed on Earth also dies out orders of magnitude too fast to build CM2-grade clay, and any impactor big enough to do better would have shattered a moon this small instead of warming it.

So the clay was inherited. Davis and colleagues note that the same 2.72-micrometer band appears on Ceres and on the big, dark main-belt asteroids Hygiea and Pallas, whose clays formed deep inside large differentiated parents. Neptune’s version reads the same way: this material was cooked inside something far bigger than today’s moons, then thrown out and reassembled. Everywhere else we study icy interiors from the outside, the way ground-based antennas sound Jupiter’s moon in our report on Europa’s radar characterization.

At Neptune, the interior became the surface.

Which worlds were smashed to make them?

Triton has been the prime suspect since at least 1989, the year Voyager 2 flew past Neptune and Goldreich’s team published its dynamical account of the system. Neptune keeps 16 known moons today, yet Triton carries more than 99.5 percent of their combined mass, two hundred kilograms for every one in all the others together. A moon that big, circling its planet backward on a steep tilt, reads as a captured Kuiper Belt object rather than a homegrown satellite. Capture is violent. If Neptune once owned regular satellites, the tidy kind born in a disk around their planet, Triton’s arrival would have driven them into crossing orbits and collisions, grinding them into a debris disk. We looked at that quieter formation channel in our assessment of the first exomoon candidate around a brown dwarf. Today’s small moons and rings would be the recycled leftovers, with the parents’ insides exposed.

“If Neptune once had a system of moons that looked something like what we see at Uranus today, we expect it would’ve been completely destroyed by the process of Triton getting captured,” Davis, who led the work, told Universe Today on 1 August.

There is a rival story, and it has numbers of its own. In simulations by Hyodo and colleagues, a large differentiated Kuiper Belt object, an order of magnitude more massive than Pluto, wanders inside Neptune’s Roche limit and is torn apart by tides. Somewhere between 0.1 and 10 percent of its mass stays behind as a ring of mostly rocky debris. Davis and colleagues take that scenario seriously and then set it aside, for two reasons with names attached. A companion JWST study led by Matthew Belyakov found Nereid, Neptune’s distant third-largest moon, coated in crystalline water ice. No observed Kuiper Belt object looks like that, but a surviving member of an original satellite family would. On top of this, a second massive visitor would likely have destabilized the fragile orbit Triton was still settling into, while multiplying assumptions the satellite story never needs. We find that reasoning convincing, with the caveat the authors themselves print. If the shredding happened before Triton showed up, the original satellites could have been built from dwarf-planet rubble, and the two histories become spectroscopically identical.

Temperature narrows down the victims. Because the clay survived, the bulk of this material was never heated past about 700 kelvin, not during the breakup and not afterward. Pluto’s interior, in the thermal models of Robuchon and Nimmo that the paper cites, climbs beyond 1100 kelvin, hot enough to bake serpentine back into dry rock. The destroyed bodies were most plausibly moon-sized, big enough to melt water in their bellies yet small enough never to scorch the clay that water made. There is something quietly humbling in that arithmetic, because the only icy-world interiors anyone can inspect directly belong to worlds that no longer exist.

The ice that should be there and isn’t

Both origin stories predict water ice, and neither gets any. Large satellites that melt inside are expected to refreeze ice-rich outer shells, which is exactly what the big moons of Uranus wear. A shredded, super-Pluto Kuiper Belt object should carry a hefty ice fraction too. On Proteus, Larissa, Galatea, and the rings: nothing. Davis and colleagues call the absence unexpected in their own discussion. We would push back on any summary, including stretches of Universe Today’s, that files the missing ice as one more line of support. It is the loose thread. Either the reaccreted material was volatile-poor from the start, or these moons lost their water ice in the collisional grinding that followed, and the paper cannot yet say which.

Whichever it is, a great deal has gone missing. Today’s inner satellite system holds roughly 1 percent of the combined mass of Uranus’s five large regular satellites, assuming a rock-like density of 2 grams per cubic centimeter. If Neptune’s original system was anything similar, about ninety-nine parts in a hundred of it, ice included, has been removed. Proteus is its own riddle, hydrated more deeply than the others yet carrying the least clay. Its material may never have passed through warm water at all. The paper also allows that the impact which chipped the 35-kilometer moon Hippocamp off its flank later baked Proteus dry, or that the outermost moon simply reaccreted from a different region of the debris disk than Larissa and Galatea did. Davis and colleagues list the options and choose none.

Honesty asks for one more admission: the deep 3-micrometer band itself still has no name. Whatever hydrated mineral carries it matches nothing in the four spectral libraries the team searched, the USGS collection, the Winnipeg facility, RELAB, and SSHADE. Even the closest natural comparison, the Neptune Trojan 2006 RJ103, fails in detail. A weak dip near 3.07 micrometers hints at ammonia-bearing material, as on Ceres, but it shows strongest on clay-poor Proteus, which argues against ammoniated clays as the carrier. We cannot tell you what this mineral is, and neither can the authors, because laboratory spectra of the right candidates under outer solar system conditions have never been taken. Webb keeps doing this to us lately. It handed Titan and Pluto a shared, unidentified absorber earlier this summer, the case we covered in our piece on the JWST mystery signal at Titan and Pluto. Now Neptune’s small moons carry an unidentified absorber of their own.

Where the smashed-worlds reading stands

• Clay, but no ice.
• The 2.72-micrometer band matches the most water-processed CM2 chondrites and Ceres, minerals that demand one to ten million years in liquid water inside a large body.
• Nereid’s crystalline ice, from Belyakov’s companion study, tips the verdict toward destroyed original satellites rather than a shredded dwarf planet.
• The missing water ice stays unexplained, and the 3-micrometer carrier stays unidentified.

That is a lot of history resting on one sharp band, which is why we walked you through the crater arithmetic and the Nereid evidence instead of just handing you the verdict. Translating a dense spectroscopy paper into words you can carry out of the room, from our desk here in Rimini, is the whole point of FreeAstroScience. Keep your reason awake while you carry them, because a mind that dozes is where nonsense moves in. Test us instead: check the band positions against the meteorite record, and tell us where our preference for the destroyed-satellite reading bends. The paper’s own wish list names what comes next, laboratory spectra of candidate minerals at outer solar system temperatures. When a match for that 3-micrometer carrier finally lands, we will run it against these moons right here. FreeAstroScience, Rimini. Gerd Dani.

Sources

1. Davis, M. R., Belyakov, M., Wong, I., Milby, Z., Brown, M. E. (2026). Neptune’s inner moons and rings are exposed icy body interiors. Science Advances, 12 (31), eaeb1437, AAAS. Published 29 July 2026. https://www.science.org/doi/10.1126/sciadv.aeb1437
2. Belyakov, M., Davis, M. R., Wong, I., Batygin, K., Brown, M. E. (2026). Nereid as a regular satellite of Neptune. Science Advances, 12, eaeb1429, AAAS.
3. Hyodo, R., Charnoz, S., Ohtsuki, K., Genda, H. (2017). Ring formation around giant planets by tidal disruption of a single passing large Kuiper belt object. Icarus, 282, 195-213, Elsevier.
4. Castillo-Rogez, J., Weiss, B., Beddingfield, C., Biersteker, J., Cartwright, R., Goode, A., Melwani Daswani, M., Neveu, M. (2023). Compositions and interior structures of the large moons of Uranus and implications for future spacecraft observations. Journal of Geophysical Research Planets, 128, e2022JE007432, AGU.
5. Robuchon, G., Nimmo, F. (2011). Thermal evolution of Pluto and implications for surface tectonics and a subsurface ocean. Icarus, 216, 426-439, Elsevier.
6. Furukawa, Y., Sekine, T., Kakegawa, T., Nakazawa, H. (2011). Impact-induced phyllosilicate formation from olivine and water. Geochimica et Cosmochimica Acta, 75, 6461-6472, Elsevier.
7. Tognetti, L. (2026). Smashed Ice Worlds Formed Neptune’s Inner Moons and Rings. Universe Today. Published 1 August 2026.