Atmospheric Escape

The intuition: a planet's air is always on a slow leak

Every atmosphere is a balance between two competing tendencies. Gravity pulls gas down toward the surface; heat keeps it bouncing around. At the very top of the atmosphere — the exobase, the altitude above which a particle is unlikely to collide with another before flying a long way — these two forces have their final argument. Most molecules are moving too slowly to leave and fall back. But the speed distribution always has a high-velocity tail, and a particle in that tail moving upward faster than the planet's escape velocity is simply gone, never to return.

For Earth this leak is a trickle: a few kilograms of hydrogen per second, negligible over the age of the solar system. But turn up the heat — move the planet close to its star, or make the star a young, ultraviolet-blazing furnace — and the trickle becomes a torrent. The upper atmosphere can inflate, heat to thousands of kelvin, and pour off the planet as a comet-like tail. Over a few hundred million years this can strip a Neptune-sized world down to its bare rocky core. Atmospheric escape is how a planet loses its air to space, and it is one of the most important forces shaping which planets exist at all.

Thermal escape and the Jeans parameter

The classical picture is Jeans escape, named for James Jeans. Gas at temperature T has a Maxwell-Boltzmann distribution of particle speeds, with the most probable speed set by

v_thermal = √(2 k T / m)

where k is Boltzmann's constant and m the particle mass. A particle escapes if it is above the exobase, moving outward, and faster than the local escape velocity

v_esc = √(2 G M / R)

The single number that governs the leak is the dimensionless Jeans (escape) parameter, the ratio of gravitational binding energy to thermal energy:

λ = G M m / (k T R) = (v_esc / v_thermal)²   (up to a factor 2)

The Jeans flux scales as e^(−λ)(1 + λ), so escape is exquisitely sensitive to λ. When λ is large (heavy molecules, cool exobase, massive planet) the exponential crushes the rate to nothing; when λ approaches a few, the tail of the distribution is no longer negligible and gas drains quickly. At Earth's exobase (T ≈ 1000 K, R ≈ 6900 km) λ for atomic hydrogen is about 8, so hydrogen escapes slowly; λ for molecular nitrogen is several hundred, so nitrogen is permanently bound. This single ratio explains why small, hot worlds keep no air while large, cool ones keep it forever.

When the leak becomes a wind: hydrodynamic escape

Jeans escape assumes the atmosphere stays in hydrostatic equilibrium and only the tail leaks. That assumption breaks when heating is strong. If stellar radiation deposits enough energy that the thermal energy of the upper atmosphere becomes comparable to its gravitational binding (λ falls toward unity), the gas can no longer remain static. It accelerates outward as a bulk flow, passing through a sonic point exactly as the solar wind does — a transonic planetary wind. This is hydrodynamic escape, and it is the dominant loss channel for hydrogen-rich worlds near active stars.

Crucially, hydrodynamic escape is not selective the way Jeans escape is. A fast bulk outflow of hydrogen exerts drag on heavier species and can carry oxygen, carbon, even metals out with it — a process called hydrodynamic drag or "blow-off." This is how a primordial hydrogen envelope can sweep away a planet's heavier volatiles, and why the chemistry of a stripped planet can differ wildly from what it started with.

Photoevaporation: the XUV-driven engine

The energy source that powers hydrodynamic escape on close-in exoplanets is the star's extreme-ultraviolet and X-ray (XUV) flux. Photons below about 91 nm ionise hydrogen; the photoelectrons heat the gas to 5,000–10,000 K, inflating it into an extended thermosphere that flows away. This XUV-driven outflow is called photoevaporation. In the widely used energy-limited approximation, the mass-loss rate is

Ṁ ≈ η π F_XUV R_p³ / (G M_p K)

where F_XUV is the XUV flux at the planet, R_p and M_p the planet's radius and mass, η ≈ 0.1–0.2 a heating efficiency (the fraction of absorbed energy that goes into unbinding gas rather than radiating away), and K ≤ 1 a correction for Roche-lobe geometry. The cubic dependence on radius is decisive: a puffy, low-density planet presents a huge cross-section and is very easy to evaporate, while a compact rocky planet of the same mass loses gas far more slowly.

The other lever is time. Young stars are XUV monsters. A Sun-like star at an age of 100 million years emits roughly 100–1000 times more XUV than today's Sun, and stays "saturated" at high activity for the first ~100 Myr before its rotation and magnetic activity spin down. Most of a planet's lifetime mass loss therefore happens in its first few hundred million years, against a star that has since calmed down. The escape we see today is the faint aftermath of a far more violent youth.

A field guide to escape regimes

"Atmospheric escape" is an umbrella over several physically distinct mechanisms. Which one dominates depends on the planet's gravity, its magnetic field, the gas species, and the stellar environment.

Mechanism	Type	Driver	Selective?	Where it dominates
Jeans escape	Thermal	Tail of Maxwell-Boltzmann distribution	Yes — light species	Earth (H), Titan
Hydrodynamic / photoevaporation	Thermal	XUV heating → transonic wind	No — drags heavies	Hot sub-Neptunes, hot Jupiters
Core-powered mass loss	Thermal	Cooling luminosity of rocky core	No	Close-in sub-Neptunes
Ion pickup	Non-thermal	Stellar-wind electric field	Yes — ions	Unmagnetised Mars, Venus
Sputtering	Non-thermal	Impacting ions eject neutrals	Partly	Mars, Mercury, Io
Dissociative recombination	Non-thermal	O₂⁺ + e⁻ → fast O + O	Yes — O, C, N	Mars oxygen loss
Polar / ion outflow	Non-thermal	Field-aligned currents	Yes — O⁺, H⁺	Earth polar caps
Impact erosion	Mechanical	Giant impacts blast off air	No	Early solar system

The non-thermal channels matter most for worlds that are too cold for vigorous thermal escape but lack a protective magnetic field. A magnetosphere deflects the stellar wind around the planet; without one, the wind couples directly to the ionosphere and erodes it. This is the central difference between Earth, which keeps its air behind a global dipole field, and Mars and Venus, which do not.

The numbers: who keeps their air and who loses it

Escape is ultimately a contest between escape velocity and exobase temperature. The table below collects the headline figures across the solar system and a few famous exoplanets.

Body	Mass (M⊕)	v_esc (km/s)	Exobase / wind T	Loss state
Earth	1.00	11.2	~1000 K	~3 kg/s H (negligible)
Venus	0.815	10.4	~300 K (cold, no field)	Non-thermal H, O
Mars	0.107	5.0	~300 K (no field)	~2–3 kg/s (MAVEN)
Titan	0.0225	2.6	~150 K, thick N₂	Slow hydrodynamic H₂, CH₄
HD 209458b	~220	~42	~10,000 K wind	~10¹⁰ g/s (H Ly-α tail)
GJ 436b (hot Neptune)	~22	~26	XUV-driven wind	Giant H coma, ~10⁹ g/s
WASP-107b (super-puff)	~30	~19	XUV-driven wind	Comet-like He tail
TRAPPIST-1 inner planets	~0.3–1.4	~7–10	M-dwarf XUV	Possible H loss, O build-up

For scale: a mass-loss rate of 10¹⁰ g/s is 10⁷ kg/s, about 3 × 10¹⁴ kg/yr. An Earth mass is 6 × 10²⁴ kg, so even at that ferocious rate it takes ~2 × 10¹⁰ years to lose one Earth mass of gas — but hot Neptunes can have only a few Earth masses of envelope to begin with, and the rate was far higher when the star was young, so they evaporate within ~10⁸ years. The arithmetic is what makes photoevaporation a population-shaping force rather than a curiosity.

The radius valley: escape's fingerprint in the exoplanet census

The most striking evidence for atmospheric escape is statistical. When Kepler's measured planet radii were corrected for accurate stellar parameters (Fulton et al. 2017), the distribution of close-in small planets showed a clear gap — the radius valley or "Fulton gap" — a deficit of planets with radii near 1.8 R⊕. The population splits into two peaks: rocky super-Earths near 1.3 R⊕ and gas-enveloped sub-Neptunes near 2.4 R⊕.

Photoevaporation explains this cleanly. A planet with a thin hydrogen-helium envelope (a percent or two of its mass) sitting on a rocky core is on a knife's edge. If XUV strips that envelope, the planet shrinks to the size of its bare core and joins the lower peak. If the envelope survives, the planet stays puffy and joins the upper peak. The intermediate state — a planet still carrying a partial envelope at exactly 1.8 R⊕ — is unstable: stripping a little gas raises the planet's bulk density and deepens its gravity well, but the dominant effect is that the radius collapses quickly once the envelope thins, so planets sweep through 1.8 R⊕ rather than lingering there. The valley is evacuated. The location of the valley, its slope with orbital period, and its dependence on stellar mass all match photoevaporation models — though a competing mechanism, core-powered mass loss (driven by the rocky core's own cooling luminosity rather than stellar XUV), predicts a nearly identical valley, and separating the two is one of the liveliest debates in exoplanet science.

Famous examples and direct detections

• HD 209458b ("Osiris"). The first exoplanet caught evaporating. In 2003, Hubble measured a Lyman-α transit depth of ~15% — ten times the planet's 1.5% optical transit — implying a vast neutral-hydrogen cloud overflowing the planet's Roche lobe. The archetype of an evaporating hot Jupiter.
• GJ 436b. A warm Neptune trailing an enormous comet-like coma of neutral hydrogen, so large it produces a transit lasting hours longer in Lyman-α than in visible light. A textbook case of a slowly evaporating ice giant.
• WASP-107b. A low-density "super-puff" whose escaping atmosphere was detected via the metastable helium 10830 Å triplet — a line observable from the ground that has become the workhorse tracer of escaping winds. WASP-107b shows a helium tail streaming away from the planet.
• The hot-Neptune desert. A near-complete absence of Neptune-mass planets on the hottest, closest orbits (periods under ~3 days). These worlds are too small to hold their envelopes against the intense XUV at such proximity, so escape clears them out — leaving behind only stripped rocky cores or the more robust gas giants.
• Mars. The clearest solar-system case. Having lost its magnetic field ~4 Gyr ago, Mars has been eroded by ion pickup, sputtering, and dissociative recombination; NASA's MAVEN orbiter has been measuring the loss directly since 2014. Isotopic enrichment of heavy argon and deuterium records the cumulative loss of a once-thicker atmosphere.
• TRAPPIST-1. Seven Earth-sized planets around an M dwarf that was extremely XUV-active for its first ~1 Gyr. Whether the inner planets retained any atmosphere — or were stripped and left with abiotic oxygen built up from water photolysis — is a key target for JWST.

Common misconceptions and edge cases

• "A magnetic field always protects an atmosphere." Not necessarily. A magnetosphere shields against stellar-wind erosion, but it also opens polar "funnels" along open field lines through which ionised gas can escape — Earth loses significant O⁺ through polar outflow. Some models even suggest a magnetosphere can increase net loss for certain planets by enlarging the interaction cross-section.
• "Escape velocity is a hard wall." Escape is not "the gas heats above v_esc and leaves." It is a statistical tail effect (Jeans) or a hydrodynamic flow through a sonic point. Plenty of gas escapes while the bulk atmosphere is far below the escape temperature, because the speed distribution always has a tail.
• "Hot Jupiters quickly evaporate to nothing." Most do not. A Jupiter-mass planet has an escape velocity of ~60 km/s and an enormous gas reservoir; even at 10¹⁰–10¹¹ g/s it loses only ~0.1–1% of its mass over its lifetime. It is the low-mass sub-Neptunes, not the giants, that escape strips down to cores.
• "Closer always means more escape." Geometry matters too. Very close in, the planet can fill its Roche lobe, and gas crossing the inner Lagrange point is lost to Roche-lobe overflow rather than thermal escape — a different regime with its own scaling.
• "Heavy isotopes prove escape; light ones don't." The signature is the ratio. Because lighter isotopes (¹H over ²H, ³⁶Ar over ³⁸Ar) escape preferentially, an atmosphere enriched in the heavy isotope is a fossil record of fractional loss. Mars's high D/H ratio (~5× Earth's) is among the strongest evidence that it lost most of its water.