No. 030 October 17, 2026 9 min read

Nobody Has Seen the Carrier

Interstellar comet 3I/ATLAS put nickel vapour into a coma far too cold to vaporise nickel. The leading explanation is a molecule never detected in any comet — and one that, in the laboratory, does not break apart the way the explanation needs it to.

On 23 July 2025, with the interstellar comet 3I/ATLAS still 3.78 astronomical units from the Sun, the VLT’s X-Shooter spectrograph found twenty-two emission lines of neutral nickel in its coma, strongest at 3414.8 and 3524.5 Å. It found no lines of neutral iron at all. A body at that distance has a surface somewhere around 150 K. Metallic nickel does not become a gas at 150 K, and it does not become a gas at ten times 150 K. Something out there was handing free nickel atoms to the sunlight, and whatever it was, it was not nickel.

3.78 au

where the nickel appeared

nucleus around 150 K

22 to 0

nickel lines to iron lines

in the first clear detection

1.6 : 1

nickel to iron by 2.17 au

the solar ratio is 1 : 18

carbonyls ever detected in a coma

any comet, this one included

I started reading about an interstellar object and ended up reading about a molecule. The interesting thing about 3I/ATLAS is not that it came from another star. It is that it did the same strange thing our own comets have been quietly doing for twenty years, only louder, and that the best explanation for all of it is a compound nobody has ever seen in a comet.

The thing our comets were already doing

In 2021 Jean Manfroid and colleagues went back through two decades of UVES spectra from the Very Large Telescope — about twenty comets, at heliocentric distances from 0.68 to 3.25 au — and found neutral iron and neutral nickel emission in essentially all of them. The equilibrium surface temperature across that range runs from roughly 340 K down to about 150 K. To liberate atomic iron and nickel from refractory dust, metallic grains, or sulfides, you need temperatures hundreds to thousands of degrees higher. Sulfide-based routes alone want something above 600 K. Nothing in a comet at 3 au is anywhere near that.

Worse, the ratio was wrong. In the comae the average was log(Ni/Fe) = −0.06 ± 0.31 — nickel and iron in roughly equal numbers. The solar value is −1.25 ± 0.04: eighteen iron atoms per nickel. Comets were running about fifteen times nickel-rich relative to the Sun, with a wide spread between individuals, 73P at +0.60 and 103P at −0.64. The paper ends on the admission that “important constituents of the nucleus or processes in the coma are still missing.”

In the same issue of Nature, Piotr Guzik and Michał Drahus reported atomic nickel in the coma of 2I/Borisov, the previous interstellar comet, at 2.32 au — equilibrium temperature near 180 K, 0.9 ± 0.3 × 10²² atoms per second — and inferred a short-lived parent with a lifetime of about 340 seconds at 1 au. The problem was never parochial.

The answer that is almost too good

There is a class of compounds that behaves exactly as required: metal carbonyls. Nickel tetracarbonyl, Ni(CO)₄, sublimates somewhere between 74 and 82 K. Iron pentacarbonyl, Fe(CO)₅, between 97 and 108 K. For scale, carbon dioxide ice goes at 72 K and water ice at 152 K. A molecule that volatile sitting in a cometary nucleus would come off early, come off cold, and come off nickel-first, because Ni(CO)₄ is the more volatile of the pair by roughly a factor of ten in sublimation rate. The observed nickel excess falls out of the vapour pressures almost for free.

Sublimation temperature, K · log scale

CO ice

CO₂ ice

Ni(CO)₄

74–82

Fe(CO)₅

97–108

H₂O ice

152

Nickel-rich sulfides

at least 600

Nickel tetracarbonyl has a history worth pausing on. Ludwig Mond, with Carl Langer and Friedrich Quincke, made it in 1890 by passing carbon monoxide over nickel metal. It is a colourless liquid boiling at 43 °C, and appallingly toxic — the occupational exposure limit is 0.001 ppm as an eight-hour average. Heat the vapour past roughly 180 °C and it decomposes back into carbon monoxide and pure nickel. That reversibility is the Mond process: the industrial route to high-purity nickel.

Which is where the popular version of this story went sideways. Avi Loeb lists the nickel detection among his anomalies for 3I/ATLAS, noting correctly that “the existence of nickel without iron is a result of the carbonyl pathway” and that the pathway is used “for the industrial production of nickel alloys,” then concluding, pressed by Josh Winn, that “we have never seen it in comets so it is an anomaly” and “it is a technological signature.” The first clause is the problem: we have seen it in comets, in about twenty of them, going back to 2002 and published in Nature four years before 3I/ATLAS was found. And carbonyl chemistry turning up in both a refinery and a cold comet is not a coincidence of purpose. It is the same fact about the molecule in both places — Ni(CO)₄ is the most volatile nickel-bearing species anyone knows of. Humans use that to separate nickel from ore; a comet would use it to lose nickel before it loses iron. No intent required at either end.

What the spatial profiles buy you

In July 2026, Astronomy & Astrophysics published a measurement I find delightful on instrumental grounds alone. A. C. Mura and colleagues took three 1200-second spectra of 3I/ATLAS at 2.17 au with the 1.22 m Galileo telescope at Asiago — a small, old instrument, its slit long enough to reach some 2.8 × 10⁵ km from the nucleus — and pulled Haser scale lengths for the metals out of them. Nickel: 8.26 (+2.12/−1.48) × 10⁵ km. Iron: 2.48 (+0.57/−0.42) × 10⁵ km. Production rates of (1.13 ± 0.18) × 10²⁴ nickel atoms per second against (7.06 ± 1.11) × 10²³ for iron, giving log(Ni/Fe) = 0.20 ± 0.11, about 1.6 to one.

The shape of the profiles is the evidence. The metal distributions show no sign of an extended source — no slow release from dust grains out in the coma — which is what a short-lived parent molecule released from the nucleus would look like. The authors are careful about what that licenses: metal carbonyls, they write, “remain a possible but still unconfirmed source of cometary Fe and Ni atoms.” They also flag that the seeing disc, roughly 6000 km projected at the comet, is hard to disentangle from a parent scale length, so the parent could be shorter-lived than the fit suggests; and that the standard outflow velocity law may simply not hold for this object, since their radial slopes come out steeper than the model wants.

The asymmetry that went away

Here is the part that most of the coverage missed. Damien Hutsemékers and colleagues tracked both metals with UVES from 4 December 2025 through 22 February 2026, from 1.88 out to 4.35 au on the way back out. Iron production fell from log Q ≈ 24.3 at 1.88 au to 22.1 by 3.50 au; nickel from 24.0 to 22.4. And the ratio moved: log Q(Ni)/Q(Fe) was about 0.04 near 1.93 au — ordinary, indistinguishable from solar-system comets — climbing to roughly 0.7 by 3.2–3.5 au, and iron dropped out of the spectra entirely after 7 February.

So “nickel without iron” was never a composition. It was a slope. At large heliocentric distance the more volatile carrier is already active and the less volatile one is not yet, and the iron lines sit below the detection threshold; closer in, both are running and the ratio relaxes to something unremarkable. The single most quoted anomaly about this object is an artefact of where in the orbit you happened to look.

What does not relax is the magnitude. Near 2 au, 3I/ATLAS was putting out at least an order of magnitude more metal than a typical solar-system comet, and 2I/Borisov did not — Borisov’s nickel-to-iron ratio and absolute output were both ordinary. One interstellar comet is loud and the other is not, and nobody has explained why.

Where this is thinner than it reads

I went looking for the detection underneath all this and it is not there.

No carbonyl has ever been directly observed in a cometary coma. Not Ni(CO)₄, not Fe(CO)₅, not in 3I/ATLAS, not in any of Manfroid’s twenty. The entire mechanism is reconstructed from sublimation temperatures, from the absence of an extended source in the spatial profiles, and from the fact that the nickel’s inferred activation energy — 0.20 to 0.27 eV in the VLT analysis — is far too low for metal sublimation or solar-wind sputtering and fits a volatile instead.

And there is a specific laboratory problem sitting underneath it. Ultrafast infrared work on gas-phase Ni(CO)₄ photodissociation at 261 nm finds the molecule shedding ligands in steps: prompt dissociation to singlet Ni(CO)₃, which loses another CO over tens of picoseconds, with a parallel concerted channel to triplet Ni(CO)₂ that then likely goes to Ni(CO). Bare nickel atoms are not the reported product. The observed cometary emission comes from free Ni I. If the carbonyl is the carrier, something has to take the last ligands off, and the photochemistry that does it has not been measured — the cross-sections for the intermediate fragments are the unverified link in the chain.

The modelling has had to add machinery, too. Hutsemékers finds that desorption of carbonyls from sublimating CO₂ and H₂O ice is negligible, which pushes the release point several centimetres or more below the surface, and the early nickel excess wants a transient heat source around 100–140 K, possibly the amorphous-to-crystalline ice transition. That may well be right. It is also three further assumptions stacked on a parent nobody has detected, and the same paper concedes that superheated metallic nanoparticles cannot be excluded. It is worth noting that this paper, the most complete dataset in the story, is an arXiv preprint from May 2026 whose journal status I could not confirm — the tightest constraint on the Ni/Fe evolution is the least certified piece of evidence for it.

What would settle it

A direct detection. Ni(CO)₄ has mid-infrared vibrational bands, and JWST already looked at this object: Cordiner and colleagues found a coma dominated by carbon dioxide at 3.32 au, CO₂/H₂O = 7.6 ± 0.3, about eighteen times the expected value for that distance and the most extreme such ratio in any comet bar C/2016 R2. That is a carbon-rich environment, which is encouraging for carbonyl chemistry and not evidence of it. A mid-infrared limit on Ni(CO)₄ in a coma would be worth more here than another spectrum of nickel lines — and a laboratory measurement of how the last two CO ligands come off under solar UV would be worth more than both.

Until then, the honest statement is narrower than the headlines in either direction. We know there is cold metal vapour in comets, including comets from other stars. We know the nickel runs ahead of the iron in a way that tracks volatility. We have a molecule that fits so well it is hard to look past, and no observation of it, and a lab result suggesting it does not quite do the final step we need it to do.