Hot Jupiter Migration

The 1995 puzzle

On the 6 October 1995, Michel Mayor and Didier Queloz announced in Nature the detection of a Jupiter-mass planet around the Sun-like star 51 Pegasi. Its orbital period was 4.23 days. Its semi-major axis was 0.052 AU — eight times closer to its star than Mercury is to ours. Nothing in planet-formation theory at the time predicted such an object should exist. Cold Jupiters at Solar-System-like distances had been the universal expectation; the very first confirmed extrasolar planet around a main-sequence star turned out to be the exact opposite.

The discovery split into two questions that have driven the field ever since. First, can a gas giant form in place at 0.05 AU? The unanimous answer is no — the inner regions of a protoplanetary disk are too hot, too gas-poor, and too small a reservoir of solids for core accretion. Second, if hot Jupiters cannot form there, how did they get there? The leading answers — Type II disk migration, high-eccentricity tidal migration, and secular chaos — are different physical pathways that all end the same way: a gas giant on a short-period orbit. Distinguishing between them is now the central observational problem.

Mayor and Queloz shared the 2019 Nobel Prize in Physics for the discovery. Hot Jupiters have since become the best-characterised class of exoplanets: bright host stars, deep transits, short repeat cadence, and atmospheres warm enough to drive strong day-night thermal contrasts. They are simultaneously the most studied and the least typical of known exoplanets, occurring around only about 1 percent of Sun-like stars.

Why hot Jupiters cannot form in situ

The core-accretion model of gas-giant formation goes like this. Dust grains in a protoplanetary disk grow by sticking and gravitational settling. Once a body reaches a few hundred kilometres, runaway accretion produces protoplanets of order an Earth mass. Beyond the snow line — the radial location where the disk is cold enough for water and other volatiles to condense as ices — the surface density of solids jumps by roughly a factor of four. Cores can grow to 10 Earth masses there in 1 to 3 million years, fast enough to capture a runaway hydrogen-helium envelope before the gas disk dissipates after about 5 to 10 million years.

For a Sun-like star, the snow line sits near 2.7 AU. Inside it, two things conspire against in-situ giant formation. The available solid surface density falls because ices have evaporated. And the total mass within 0.05 AU of any plausible protoplanetary disk — even the high-end "minimum mass extrasolar nebula" reconstructions — is only of order a tenth of an Earth mass, two orders of magnitude short of a single gas-giant core. There is simply not enough rock to build a Neptune there, let alone a Jupiter.

The alternative is gravitational instability — direct fragmentation of a massive disk into self-gravitating clumps. This can produce gas giants quickly, but it requires the disk to be cold and massive, which favours formation at tens to hundreds of AU, even further out than core accretion. It cannot produce a hot Jupiter in place either. Whatever the formation channel, the conclusion is the same: gas giants are born far out and must move in.

Type II disk migration

The first migration mechanism proposed in detail is now known as Type II disk migration, laid out for the 51 Peg b context by Douglas Lin, Peter Bodenheimer and Daniel Richardson in a 1996 Nature paper. A massive planet embedded in a viscous gas disk exerts strong tidal torques on the gas. When the planet exceeds a critical mass — roughly

M_gap / M_⋆ ≳ 40 (ν / r² Ω) × (H/r)² ≈ a few × 10⁻⁴

for typical disk parameters — its torque overpowers the local viscous spreading and carves an annular gap in the surrounding gas. The planet then sits in the centre of its own clearing, locked to the gas viscously: as the disk drains onto the star on the viscous timescale, it carries the planet inward with it.

The viscous timescale at radius r is

t_visc ≈ r² / ν = (1 / α Ω) × (H/r)⁻²
       ≈ 10⁶ years   for α = 10⁻², r = 5 AU around a Sun-like star

A million years is comfortably shorter than the disk lifetime, so a forming gas giant can drift from 5 AU down to less than 0.05 AU before the disk dissipates and the migration stops. The natural stopping mechanism is the inner edge of the disk, where the disk truncates because of stellar magnetic fields (the magnetospheric cavity, around 0.04-0.05 AU for typical T Tauri parameters) or where the local viscous timescale becomes shorter than the planet's tidal response.

Type II migration has a clean prediction: it preserves the planet's orbital alignment with the disk, which itself is generally aligned with the star's spin. So hot Jupiters that arrived by Type II migration should have low stellar obliquities. They should also tend to be alone and on circular orbits, since the disk damps eccentricity.

High-eccentricity tidal migration

The alternative pathway needs no gas disk. Long after the disk has dissipated, a gas giant on a wide (1 to 10 AU) circular orbit can be perturbed onto a highly eccentric orbit by a few different mechanisms. The eccentric orbit grazes the star at periastron, where strong tidal dissipation drains orbital energy. With each periastron passage the orbit shrinks; angular momentum is approximately conserved as the orbit circularises. The end state is a hot Jupiter on a circular orbit with semi-major axis

a_final ≈ a_initial × (1 − e_initial²) ≈ 2 × r_peri

so a wide-orbit Jupiter excited to e = 0.98 settles at twice its periastron distance, which can be of order 0.05 AU if the periastron grazed inside about 0.025 AU.

Three mechanisms can excite the necessary high eccentricities.

• Planet-planet scattering. A system of two or more gas giants on initially close orbits is dynamically unstable. Gravitational scattering ejects one and leaves the survivor on a highly eccentric orbit. Simulations reproduce the observed eccentricity distribution of cold Jupiters this way.
• Kozai-Lidov oscillations. If the gas giant is in a hierarchical triple with a distant binary stellar companion on an inclined orbit, secular perturbations drive coupled oscillations between the inner orbit's eccentricity and inclination. For initial mutual inclinations above about 39.2°, the cycles drive the inner planet's eccentricity to near unity. Periastron eventually drops low enough for tidal dissipation to take over and circularise the orbit. The mechanism is named for Yoshihide Kozai (1962) and Mikhail Lidov (1961).
• Secular chaos. Even in a system with only planet-mass perturbers, slow secular interactions can chaotically pump the eccentricity of one giant to extreme values over hundreds of millions of years. Wu and Lithwick (2011) showed this is a viable channel in multi-Jupiter systems.

All three high-e pathways share a key signature: they couple the inner planet's orbit to its inclination. The resulting hot Jupiters are predicted to have a broad distribution of stellar obliquities, including significantly misaligned and even retrograde orbits.

Measuring stellar obliquity: the Rossiter-McLaughlin effect

The Rossiter-McLaughlin (RM) effect is the cleanest observational handle on which migration pathway a given hot Jupiter took. A rotating star has one hemisphere coming toward us (blueshifted) and one receding (redshifted). When a planet transits the disk, it occults a small piece of one hemisphere at a time. The disc-integrated radial-velocity curve develops an anomaly: a redshift dip while the planet covers the blueshifted side, then a blueshift dip while it covers the redshifted side. The shape and skew of the anomaly depend on the geometry — specifically on λ, the sky-projected angle between the planet's orbital normal and the star's spin axis.

ΔRV(t) ∝ (R_p / R_⋆)² × v sin(i) × f(impact parameter, λ)

Holt predicted the effect for eclipsing binaries in 1893; Rossiter and McLaughlin published independent measurements for Algol-type binaries in 1924. The technique was adapted to transiting hot Jupiters by Queloz and others starting in 2000. The first measurement that delivered a striking surprise was HD 80606b in 2010 — clearly retrograde, with λ near 180°. Many more have followed.

The picture that emerged is roughly bimodal. Around hot, rapidly rotating stars (T_eff above about 6,250 K), hot Jupiter obliquities are broadly distributed — many systems are significantly tilted, some retrograde. Around cooler stars below 6,250 K, the obliquities are mostly small. The interpretation, developed by Winn et al. (2010), is that tidal dissipation in the convective envelopes of cool stars realigns the host's spin with the planet's orbit on a Gyr timescale, washing out the dynamical history of high-e migrants. Hot stars have radiative envelopes that resist tidal coupling, so the primordial obliquity survives. About half of all measured hot Jupiters across the full sample are significantly misaligned — strong evidence that a substantial fraction got there via high-eccentricity migration.

Other diagnostics that pin down a migration pathway

• Wide-orbit companion gas giants. If a hot Jupiter system contains a second, distant gas giant on a wide orbit, planet-planet scattering or Kozai-Lidov are plausible. Type II migration tends to leave the gas giant alone or in a resonant chain with smaller bodies, not a wide companion.
• Stellar binary companions. A wide binary tertiary is suspicious for Kozai-Lidov. Surveys find stellar companions to hot-Jupiter hosts at rates broadly consistent with the binary population, but with hints of an excess in the right geometric configuration.
• Inner small planets. Hot Jupiters almost always lack inner companions in their immediate vicinity. Type II migration sweeps inner planets up or scatters them out, predicting exactly this. High-e migration is even more violent and gives the same prediction. Either way, the loneliness of hot Jupiters is itself a diagnostic.
• Eccentricity remnants. Some hot Jupiters retain non-zero eccentricities, indicating they have not finished circularising. The eccentricity distribution as a function of period can be inverted to a tidal Q factor and a migration timescale.
• Sub-saturn-mass desert. Between roughly 0.1 and 1 Jupiter masses at short orbits there is a deficit — the "sub-Saturn desert" — interpreted as evidence that only objects above a critical mass survive runaway gas accretion or evaporation during migration.

Worked example: tidal circularization timescale

Consider a Jupiter-mass planet pumped to eccentricity 0.98 with periastron at 0.025 AU around a Sun-like star. The tidal circularization timescale for the planet's own tides is

τ_circ ≈ (4 / 63) × (Q_p / k_2,p) × (M_p / M_⋆) × (a / R_p)⁵ × (1 / Ω_orb)
       × (1 − e²)^(13/2) / [f₃(e) − (11/18) f₄(e) Ω_p/Ω_orb ]

For order-of-magnitude purposes the leading scaling is

τ_circ ∝ (a / R_p)⁵ × (1 − e²)^(13/2)

The catastrophic factor (1 − e²)^(13/2) — about 5 × 10⁻¹² for e = 0.98 — is what makes high-eccentricity migration work. A planet on a moderately eccentric orbit (e = 0.3) at 5 AU has τ_circ many orders of magnitude longer than the age of the universe. The same planet excited to e = 0.98 with periastron at 0.025 AU circularises in a fraction of a billion years, ending up on a circular orbit at a ≈ 0.05 AU. The strong nonlinearity is the reason the high-eccentricity pathway is "all or nothing" — only orbits driven to near-unity eccentricity end up as hot Jupiters.

The death of hot Jupiters — WASP-12b

Once parked on a short circular orbit, a hot Jupiter is not necessarily safe. If its orbital period is shorter than the host star's rotation period — which is generally the case — the tidal bulge it raises on the star leads, dragging the star to spin up and the orbit to decay. The orbital evolution is governed by the stellar tidal quality factor Q_⋆, which is poorly constrained but observationally lies between 10⁵ and 10⁷ for Sun-like stars.

WASP-12b is the textbook case of an observed orbital decay. Discovered in 2008 with a 1.09-day period around a slightly evolved F-type star, its transit timings shift by about 30 milliseconds per year, implying a current orbital decay rate of about 30 ms/yr — corresponding to Q_⋆ near 2 × 10⁵ and an in-fall timescale of roughly 3 million years. The planet is also being tidally disrupted: ultraviolet observations reveal a comet-like tail of escaping atmosphere, with mass loss rates around 10⁹ g/s. Within a few million years, WASP-12b will either be fully consumed or stripped to its core.

The implication is that the population of hot Jupiters we observe is in steady state: some are being delivered (by ongoing high-e migration in older systems, or stragglers from disk migration in young systems), and some are being destroyed. The observed ~1 percent occurrence rate around Sun-like stars is the equilibrium between supply and sink.

Comparing the three migration pathways

Mechanism	Driver	Timescale	Predicted obliquity	Companion signature
Type II disk migration	Disk gas torques in viscous disk	~ 10⁶ yr (in-disk)	Low (aligned with disk/star)	None required
Planet-planet scattering	Mutual gravity of multi-giant system	~ 10⁷–10⁸ yr (high-e then tidal)	Broadly distributed, can be retrograde	Often a wide-orbit Jupiter survivor
Kozai-Lidov + tides	Distant inclined binary	~ 10⁸–10⁹ yr	Bimodal; often misaligned	Wide stellar binary
Secular chaos	Long-term secular forcing in multi-planet system	~ 10⁸–10⁹ yr	Often misaligned	Multiple giants on wide orbits
In-situ formation	Direct accretion at 0.05 AU	n/a	Low	Inner small planets preserved

No single mechanism explains every hot Jupiter. Most likely the observed population is a mix: low-obliquity systems orbit cool stars where realignment may have erased the original obliquity, and the high-obliquity systems are mostly survivors of high-e migration around hot stars whose radiative envelopes cannot realign on Gyr timescales. The exact balance of channels is an active question for missions like ARIEL, PLATO, and the Twinkle obliquity survey.

Notable systems

• 51 Pegasi b. The first confirmed exoplanet around a main-sequence star. 0.46 Jupiter masses, 4.23-day orbit, 0.052 AU. Discovered by Mayor and Queloz (1995). 2019 Nobel.
• HD 209458b. "Osiris". The first transiting exoplanet (Charbonneau et al. 2000). Demonstrated that 51-Peg-class objects are genuinely gas-giant in radius, not low-mass stars on near-edge-on orbits. Famous for the detection of an evaporating atmospheric envelope.
• WASP-12b. The clearest case of observed tidal orbital decay, with timing shifts of about 30 ms/yr. A few-Myr lifetime remaining.
• HD 80606b. 4-Jupiter-mass planet on a remarkably eccentric 111-day orbit (e ≈ 0.93) — caught mid-pathway through high-eccentricity migration. The Rossiter-McLaughlin measurement showed a retrograde projected obliquity.
• WASP-17b. An inflated hot Jupiter on a confirmed retrograde orbit, λ ≈ 150°. A canonical demonstration that primordial alignment is not preserved in many hot-Jupiter systems.
• KELT-9b. The hottest known hot Jupiter, with a day-side temperature near 4,600 K — hotter than a K dwarf. Atomic Fe and Fe⁺ have been detected in its atmosphere.

Common pitfalls

• Treating "hot Jupiter" as a formation class. Hot Jupiter is an end-state, not a birth state. Two planets that look identical today may have arrived by completely different pathways. The distinguishing evidence is in the system architecture, not the planet itself.
• Equating obliquity with formation channel directly. Tidal realignment around cool stars can erase a primordial high obliquity. A low obliquity is consistent with either Type II migration or with a high-eccentricity migrant that has been realigned. Only the obliquity distribution as a function of stellar temperature lets you back out the formation rate.
• Forgetting the disk-edge stopping mechanism. Type II migration is often described as "the planet drifts inward with the gas". But the gas in the inner cavity drains onto the star, so the migration must stop somewhere, typically the magnetospheric truncation radius near 0.04 AU. The pile-up of observed periods just above ~3 days reflects this disk edge.
• Confusing Type I and Type II. Type I migration is the linear regime for low-mass planets that do not open a gap; the planet drifts on a fraction of the viscous timescale and direction depends on disk thermodynamics. Type II is the gap-opening regime for massive planets. Hot Jupiters are Type II migrants; super-Earths and mini-Neptunes are Type I.
• Assuming Kozai-Lidov needs special tuning. The critical mutual inclination is only 39.2°, and any sufficiently distant inclined stellar binary triggers the oscillations. The mechanism is generic in hierarchical triples; what is non-generic is having the inner orbit's periastron driven low enough for tidal dissipation to lock it in as a hot Jupiter.