Sublimation Jets: Ice Turning Straight to Gas and Steering a Comet

Solid to vapour: why comet ice never melts

A comet nucleus is what Fred Whipple in 1950 called an icy conglomerate: water ice, more volatile ices, silicate dust and organics, weakly bound. Rosetta refined that into an icy dirtball. 67P/Churyumov-Gerasimenko has a bulk density of just 533 ± 6 kg/m³ — half that of water, implying 70-75% pore space — a mass of 9.98 x 10¹² kg, and a dust-to-gas mass ratio near 4. It reflects about 6% of incident light; 1P/Halley ~4%, 19P/Borrelly ~3%. Comets are among the darkest objects in the Solar System, which is why they heat efficiently.

The decisive fact is water's phase diagram. Its triple point sits at 611.66 Pa and 273.16 K, and below that pressure liquid water has no stable field at all. On a comet the ambient pressure is essentially the ice's own vapour pressure — about 0.16 Pa at 200 K, four thousand times too low. Melting is not merely unlikely on a comet; it is thermodynamically forbidden.

The governing relation is a surface energy balance on illuminated ice:

• (1 − A) S_☉ cosθ / r² = εσT⁴ + L Z(T) + conduction, with A ≈ 0.05, S_☉ = 1361 W/m², r in AU, and latent heat of sublimation L = 2.83 x 10⁶ J/kg.
• The flux obeys Hertz-Knudsen, Z(T) = P_vap(T) √(m / 2πkT). Because P_vap follows Clausius-Clapeyron and climbs exponentially, Z acts as a switch rather than a dial.

At 1 AU a dark subsolar ice surface absorbs ~1300 W/m² but radiates only ~90 W/m² at 200 K, so its temperature rises until sublimation carries the rest. The balance lands near 200 K, stripping a few times 10⁻⁴ kg m⁻² s⁻¹ (~10¹⁸ molecules cm⁻² s⁻¹). Ice is a thermostat, pinning the surface near 200 K instead of the ~390 K a dry black rock would reach. Move outward and the thermostat fails, since the blackbody equilibrium temperature of a rapid rotator falls as T ≈ 278 K / √r(AU): the ~150 K needed for vigorous water sublimation arrives near 3 AU, which is why comets turn on inside that distance. Beyond it, CO₂ (from ~10-13 AU) and CO (sublimating near 25 K) take over.

Why the outflow is a jet and not a wind

Sublimation happens where the photons land. The cosθ term means mass flux tracks the subsolar point, so a rotating nucleus switches sources on and off every few hours: Rosetta watched 67P's Hapi neck brighten and fade with the 12.4-hour spin, thermal lag pushing peak output past local noon.

The dust mantle sharpens that localisation. Grains too heavy to lift fall back as an insulating blanket, the water front retreats millimetres to centimetres below the surface, and gas must diffuse out through porous rubble. MIRO and VIRTIS measured a very low thermal inertia for 67P, ~10-50 J m⁻² K⁻¹ s^−1/2, giving a diurnal skin depth of ~2 cm and a seasonal one of 1-2 m. Activity is a thin-skin phenomenon concentrated where the mantle is cracked, freshly collapsed or absent — fissures, cliff faces, pit walls. Only a small fraction of the surface is active: ~10% for Halley per Giotto, a few percent for a typical Jupiter-family comet, effectively over 100% for hyperactive 103P/Hartley 2, whose coma is full of icy grains that keep sublimating after they leave.

Gas departs at roughly its thermal speed, a few hundred m/s, accelerating into vacuum to a terminal 0.5-1 km/s. Near perihelion the coma stays collisional out to hundreds of kilometres, but the gas thins so fast that dust is effectively accelerated only in the innermost tens of kilometres. Inside that drag-coupling zone, aerodynamic drag — gas density times velocity squared times grain cross-section — overwhelms a nucleus gravity whose escape speed is under 1 m/s, so micron grains leave easily, decimetre chunks are marginal, and the largest fall back as airfall.

Freed dust then feels radiation pressure, quantified by β, its ratio to solar gravity, roughly 5.7 x 10⁻⁵ Q_pr/(ρa) for grain density in g/cm³ and radius in cm; β nears 1 for sub-micron grains. Grains of different size leave at different times and are pushed by different amounts, fanning out along Finson-Probstein syndynes and synchrones into the broad, curved dust tail. Meanwhile solar ultraviolet dissociates and ionises the gas — a water molecule survives ~10⁵ s against dissociation, ~10⁶ s against ionisation at 1 AU — and the ions are picked up by the draped interplanetary magnetic field and swept anti-sunward at tens to hundreds of km/s as the straight blue ion tail, glowing in the CO⁺ comet-tail bands near 427 nm. Ludwig Biermann used exactly this acceleration in 1951 to infer the solar wind; Hannes Alfvén supplied the magnetic draping in 1957.

Jets as rockets: non-gravitational forces

Every kilogram of vapour carries momentum, and the reaction pushes back. Uniform sources would cancel; localised sources on a sunlit, rotating, rough surface do not.

Scale it for 67P. Near its 13 August 2015 perihelion, Rosetta measured peak water production of a few times 10²⁸ molecules per second, of order 10³ kg/s. At an exhaust speed of ~500 m/s that is ~5 x 10⁵ N gross; divided by a 10¹³ kg nucleus, 5 x 10⁻⁸ m/s² if perfectly collimated, and of order 10⁻⁹ m/s² after realistic cancellation. Over the months around perihelion that adds only centimetres to decimetres per second — negligible for a spacecraft, glaring to astrometry, because a small along-track error compounds into days of timing error over decades.

Orbit determination handles this with the formalism of Brian Marsden, Zdenek Sekanina and Donald Yeomans (1973): coefficients A1 (radial), A2 (transverse) and A3 (normal), each multiplied by an empirical g(r) ∝ (r/r₀)^−2.15[1 + (r/r₀)^5.093]^−4.6142 with r₀ = 2.808 AU — a curve deliberately shaped like the water sublimation rate. Typical |A1| ~ 10⁻⁸ AU/day², or ~2 x 10⁻⁷ m/s².

A2 changes the clock, because a transverse force adds or removes orbital energy. Johann Franz Encke noticed in 1819 that comet 2P returned about 2.5 hours early every 3.3-year orbit — the first non-gravitational effect ever measured. Friedrich Bessel proposed the rocket explanation in 1836; Whipple's 1950 ice model supplied the working substance. The sign of A2 depends on spin direction, since thermal lag shifts peak outgassing to the afternoon side — a direct analogue of the Yarkovsky effect on asteroids.

1P/Halley is the showcase: outgassing delays its perihelion by roughly four days per 76-year revolution. Yeomans and Kiang fitted A1 and A2 in 1981 and integrated back to the 240 BC apparition, matching Chinese records — which is why the 9 February 1986 perihelion was known years ahead to the day, and Giotto could be aimed within 596 km of a nucleus nobody had seen. The tradition began with Edmond Halley's 1705 prediction of the 1758 return, refined by Alexis Clairaut, Joseph Lalande and Nicole-Reine Lepaute to within a month of the 13 March 1759 perihelion. The modern flashpoint is 1I/'Oumuamua, found by Robert Weryk with Pan-STARRS1 on 19 October 2017: Micheli and colleagues reported a radial excess acceleration of ~5 x 10⁻⁶ m/s² scaled to 1 AU, falling roughly as 1/r² — the signature of solar-driven outgassing, except that Spitzer detected no dust at all.

Torque, spin change and splitting

Thrust applied off the centre of mass is a torque, so jets rewrite rotation as well as orbit. Rosetta watched 67P's 12.4-hour spin period shorten by roughly 20 minutes across the 2015 perihelion, with a measurable shift of the spin axis; a comparable change had occurred in 2009. The record holder is 41P/Tuttle-Giacobini-Kresák, whose period more than doubled from ~20 hours to 46-60 hours within weeks in 2017 — the fastest known spin change of any Solar System body.

When torque spins a comet up instead, the endpoint is destruction. A cohesionless rubble pile sheds mass below roughly P_crit ≈ 3.3 h / √ρ (ρ in g/cm³), about 4.5 hours at 533 kg/m³. Analyses of 67P's overhangs and collapsed cliffs imply tensile strengths of only ~1-10 Pa, so splitting is common: 73P/Schwassmann-Wachmann 3 began fragmenting in 1995 and had produced dozens of pieces by 2006, and C/1999 S4 (LINEAR) disintegrated entirely in July 2000. These differ from tidal disruptions such as D/1993 F2 Shoemaker-Levy 9, torn apart by Jupiter in July 1992.

Between routine activity and destruction sit outbursts. Rosetta catalogued 34 in the three months bracketing 67P's 2015 perihelion. The cleanest causal chain came on 10 July 2015, when a ~70 m fracture along the Aswan cliff gave way: OSIRIS images showed the collapse, freshly exposed bright water ice and a coincident dust plume — the first outburst tied to a specific landslide. On a grander scale, 17P/Holmes brightened by a factor of roughly a million, magnitude 17 to 2.8, in about 42 hours in October 2007.

How sublimation jets are actually measured

Flybys and rendezvous. The 1986 Halley armada — Vega 1 and 2, Suisei, Sakigake and ESA's Giotto, which passed 596 km from the nucleus on 14 March 1986 — returned the first resolved cometary nucleus: a 15 x 8 km peanut of albedo ~0.04, with dust jets from a small sunward fraction of the surface. Deep Space 1 imaged a strongly collimated jet at 19P/Borrelly in 2001, and Stardust flew through 81P/Wild 2 in January 2004, returning grains in 2006. Deep Impact drove a 370 kg impactor, fronted by 113 kg of copper, into 9P/Tempel 1 at 10.3 km/s on 4 July 2005; its extended mission, EPOXI, found at 103P/Hartley 2 on 4 November 2010 that CO₂ jets from the small lobe were hurling out centimetre-scale icy chunks while water sublimated from the smooth waist.

Rosetta. ESA's orbiter escorted 67P from 6 August 2014 to 30 September 2016. OSIRIS imaged jets, cliff collapse and airfall at metre resolution; ROSINA's mass spectrometers tied coma composition and production rate to rotation phase; MIRO measured surface and subsurface temperatures and water output; VIRTIS mapped the diurnal water-ice cycle, watching frost condense at night and vanish after dawn; GIADA weighed individual grains; and the RSI radio-science experiment pinned the nucleus mass to better than 0.1%.

Remote sensing. Water production is inferred from the OH radical — 18-cm lines at Nançay, the 308 nm ultraviolet band — from the hydrogen coma in Lyman-alpha with SOHO/SWAN, or directly from near-infrared water lines with Keck/NIRSPEC and IRTF/iSHELL. JWST measures H₂O, CO₂ and CO at once, which is how the CO₂-rich coma of the third interstellar object, 3I/ATLAS, was characterised after its discovery on 1 July 2025. Ground-based jet structure is enhanced with the Larson-Sekanina rotational gradient filter, which manufactures spurious spirals if the centroid is off. The non-gravitational coefficients come from orbit fits to decades of astrometry, tabulated in the JPL small-body database.

Look-alikes and persistent misconceptions

• Comets do not melt. There is no meltwater or slush. Descriptions involving melting have the wrong phase transition and the wrong energy budget: sublimation costs 2.83 MJ/kg, about eight and a half times the 0.334 MJ/kg of fusion.
• Not cryovolcanism. Enceladus's south-polar plumes, found by Cassini in 2005, vent a pressurised subsurface ocean through the tiger stripes, powered by tidal heating. Cometary jets are sunlight landing on near-surface ice. The plumes look alike; the physics is unrelated.
• A jet in an image is not proof of a nozzle. One of Rosetta's deflating results is that much apparent collimation is geometric: concave alcoves focus outflow, self-shadowing sharpens edges, and line-of-sight integration through a broad but structured coma manufactures filaments — as Jean-François Crifo had argued from gas dynamics for years beforehand. Discrete sources exist, but a bright ray is a projection first and a vent second.
• Tails do not trail the comet. Both point broadly anti-sunward, so an outbound comet leads with its tails. The occasional sunward anti-tail, as on Comet Arend-Roland in 1957, is a perspective effect from large, low-β grains in the orbit plane; Hale-Bopp in 1997 even showed a third, neutral sodium tail.
• 'Oumuamua's acceleration was not a technosignature. A radial, roughly 1/r² non-gravitational force is the standard fingerprint of outgassing, seen in hundreds of ordinary comets. The real puzzle is the missing dust, which motivates volatiles that leave no visible tail: sublimating H₂ ice, N₂ ice chipped from a Pluto-like body, or hydrogen released from amorphous water ice. The debate is about chemistry, not propulsion.

What remains unsolved

The dust-lifting problem. Near perihelion the sublimation pressure available to lift material is of order 0.01-0.1 Pa, while tensile strengths inferred from collapsing overhangs are ~1 Pa or more — yet Rosetta routinely watched decimetre-scale chunks leave 67P. Candidate fixes include gas pressure building beneath a sealing mantle, thermal-fatigue cracking, the low effective cohesion of fluffy aggregates, and the weakness of re-deposited airfall.

Outburst triggers. Cliff collapse explains some events but not those far from the Sun. Crystallisation of amorphous water ice, exothermic at ~9 x 10⁴ J/kg and running away near 130-140 K, is the leading candidate for distant activity such as the roughly seven outbursts a year of 29P/Schwassmann-Wachmann 1 at ~6 AU. Whether comets actually retain amorphous ice, and at what depth, is hard to test.

Isotopic heterogeneity. Rosetta measured a D/H ratio in 67P's water of 5.3 x 10⁻⁴, about three times Earth's ocean value, while Herschel found an Earth-like 1.6 x 10⁻⁴ at 103P/Hartley 2 — scatter that complicates any simple account of comets delivering Earth's water.

Better thrust physics. The Marsden g(r) curve encodes water sublimation and quietly fails for comets active far beyond 3 AU, for hyperactive objects and for interstellar visitors. Fixing it matters: non-gravitational forces are the dominant error term in long-term impact-hazard prediction for active bodies, and in targeting missions such as ESA and JAXA's Comet Interceptor, which will wait at Sun-Earth L2 to ambush a dynamically new comet from the Oort cloud.

What drives cometary activity at different distances from the Sun

Volatile or process	Free-sublimation temperature	Approximate switch-on distance	What it does
Water ice (H2O)	~150-200 K	~3 AU inbound	Dominant engine near perihelion; lifts most of the dust that makes the visible coma and dust tail
Carbon dioxide (CO2)	~80-90 K	~10-13 AU	Keeps comets weakly active far from the Sun; at 103P/Hartley 2 it flings out centimetre-sized icy chunks
Carbon monoxide (CO)	~25 K	>50 AU for free ice; observed at 5-25 AU when released from depth	Powers distant comae, e.g. 29P/Schwassmann-Wachmann 1 at ~6 AU and C/2017 K2 seen active near 24 AU
Amorphous-to-crystalline ice transition	~130-140 K	~4-6 AU	Exothermic (~9 x 10^4 J/kg); releases trapped CO and CO2 and is a leading trigger for distant outbursts
Clathrate breakdown / pressurised pockets	varies	variable	Episodic release; invoked for sudden outbursts that no steady sublimation curve predicts