Core Accretion

A planet built core first

Jupiter contains more than twice the mass of every other planet in the Solar System combined — about 318 Earth masses, the overwhelming majority of it hydrogen and helium gas. The central puzzle of giant-planet formation is not where the gas came from; the young Sun was surrounded by a disk with plenty of it. The puzzle is how to gather that much gas onto a single body fast enough, when the gas itself only sticks around for a few million years. Core accretion answers that puzzle with a two-step strategy: first build a solid seed massive enough to bend the surrounding gas to its will, then let gravity do the rest.

In the core-accretion picture, a giant planet begins as dust. Micron-sized grains in the protoplanetary disk collide and stick, growing through pebbles to kilometre-scale planetesimals to a solid protoplanetary core. When that core reaches a critical mass of roughly ten Earth masses, a thermodynamic threshold is crossed: the core can no longer hold up the gas envelope around it in static balance. The envelope contracts, fresh gas rushes in, and the planet enters runaway gas accretion, swelling to hundreds of Earth masses. The whole sequence is a race against the clock, because the gas disk that supplies the envelope disperses at an age of 3 to 10 million years. Win the race and you get a gas giant; lose it and you are left with an ice giant or a bare super-Earth.

How it works: dust to pebbles to planetesimals to a core

The build-up proceeds across many orders of magnitude in size, and the physics changes character at each stage.

Dust to pebbles. Sub-micron silicate and ice grains, well coupled to the gas, collide gently and stick through van der Waals and electrostatic forces, growing to millimetre and centimetre aggregates. Around centimetre size, growth stalls at the so-called bouncing and fragmentation barriers, and the aerodynamically marginal "pebbles" begin to feel gas drag strongly enough to drift radially inward.

Pebbles to planetesimals. Crossing from pebbles to kilometre-scale bodies is the hardest jump — a metre-sized boulder would spiral into the star in a few hundred years from gas drag (the "metre barrier"). The modern resolution is collective gravitational collapse: the streaming instability concentrates pebbles into dense filaments that collapse directly into 10–100 km planetesimals, bypassing the dangerous intermediate sizes.

Planetesimals to a core. Once planetesimals exist, gravitational focusing takes over. The largest bodies grow fastest because their cross-section is enhanced by their own gravity — this is runaway growth, and it produces a few large embryos. The embryos then enter oligarchic growth, each clearing its own annular feeding zone and growing toward an isolation mass. In the giant-planet region beyond the snow line, where ices add solid surface density, the isolation mass can reach several Earth masses, providing the seed for a true core.

The envelope. A core of even a few Earth masses already has enough gravity to capture a tenuous atmosphere of disk gas. As long as solids keep raining down, the energy they deposit heats this envelope and holds it up in hydrostatic equilibrium. The envelope grows slowly and quietly — until the core crosses the critical mass and the equilibrium fails.

The critical core mass and why it is near ten Earth masses

The heart of core accretion is a stability condition on the gas envelope. Consider a core of mass M_core surrounded by an envelope of disk gas. While planetesimals or pebbles fall onto the core at a rate Ṁ_solid, they release gravitational luminosity

L ≈ G · M_core · Ṁ_solid / R_core

That luminosity must be transported out through the envelope. The envelope settles into a static structure determined by hydrostatic equilibrium and radiative (or convective) energy transport. There is, however, a maximum gas mass that a given core can support in such a static structure. As M_core grows, this maximum at first grows too — but it has a ceiling. Beyond a critical core mass, no static envelope solution exists: any envelope you try to build must contract instead of sitting still. The standard estimate, from Mizuno (1980) and refined by Stevenson (1982), gives a critical core mass of order

M_crit ≈ 10 M⊕ · ( Ṁ_solid / 10⁻⁶ M⊕/yr )^(3/(s+...)) · (κ / κ₀)^q

where κ is the envelope opacity and the exponents are modest. The practical takeaway is robust even though the precise prefactor is not: the critical mass lands in the range 5–15 Earth masses, commonly quoted as ~10 M⊕. It rises if solids fall faster (more support luminosity) or the envelope is more opaque (harder to radiate), and falls if the planetesimal supply dwindles or grain opacity is low. This is why the giant planets all have cores near this value, and why a lone super-Earth that never reached it stays gas-poor.

The runaway: from core to gas giant

Once the core exceeds M_crit, the envelope can no longer remain static. It contracts, and contraction is self-amplifying. As the envelope shrinks, its outer boundary pressure drops below the disk pressure, so disk gas flows in to fill the gap. The added gas increases the envelope mass, which deepens the potential well and forces more contraction. The system passes through the crossover mass, where the envelope mass equals the core mass — for a Jupiter analog, a total of roughly 16–30 Earth masses — and then the gas accretion rate climbs steeply.

At this point the planet's own thermodynamics no longer limit growth; the bottleneck becomes how fast the disk can deliver gas. The planet accretes hundreds of Earth masses in a span short compared with the disk lifetime, until it grows massive enough to open an annular gap in the disk and starve itself, setting its final mass. A body that was a quiet ten-Earth-mass core for millions of years becomes a 300-Earth-mass gas giant in a geologically brief burst. This sharp two-phase character — slow core build-up, then explosive gas capture — is the signature of core accretion.

Worked example: building Jupiter against the clock

Take the canonical Pollack et al. (1996) calculation of a Jupiter analog at 5.2 AU and follow the mass budget. The model breaks neatly into three phases.

Phase 1 — Core build-up (planetesimal runaway/oligarchic growth)
  Core mass rises rapidly to the local isolation mass.
  Reaches M_core ≈ 11 M⊕,  M_env ≈ 1 M⊕   after ~0.5 Myr

Phase 2 — Slow envelope growth (the long plateau)
  Core accretion stalls (feeding zone depleted); envelope grows
  quasi-statically while the core inches up.
  Lasts several Myr — the rate-limiting bottleneck.
  Ends near crossover:  M_core ≈ M_env ≈ 16 M⊕   at ~7–8 Myr

Phase 3 — Runaway gas accretion
  Past crossover the envelope collapses; gas pours in.
  M_planet: ~30 M⊕ → ~318 M⊕ in well under 1 Myr
  Final Jupiter mass ≈ 318 M⊕ = 1 M_Jupiter

The lesson hides in Phase 2. The classic planetesimal-only model spent several million years on the slow envelope-growth plateau, so Jupiter was not finished until roughly 8 Myr — dangerously close to, or beyond, the observed 3–10 Myr disk-dispersal window. If the disk had cleared at 5 Myr, this Jupiter would have been stranded as a Neptune-class object. That tension is real and quantitative, and it is the single strongest motivation for the pebble-accretion refinement: feed the core with drag-captured pebbles instead of slow planetesimals, and Phase 1 plus the start of Phase 2 collapse from millions of years to a few hundred thousand, finishing the planet with time to spare.

Variants and regimes

Core accretion is a family of models, not a single recipe. The main regimes differ in how the core is fed and where in the disk the planet sits.

• Planetesimal accretion (classic). The core grows by capturing 1–100 km bodies through gravitational focusing. Robust physics, but slow in the outer disk — the source of the timescale problem.
• Pebble accretion (modern). Gas drag lets the core capture inward-drifting cm–dm pebbles from a wide annulus, with an enormous effective cross-section. Builds ~10 M⊕ cores in < 1 Myr and is now the default fast channel.
• In-situ versus migration. A core may grow where it forms or migrate inward through the disk (Type I migration) while accreting, which changes the local supply of solids and gas and can deliver giants to short-period orbits — relevant to hot-Jupiter migration.
• Arrested cores (ice giants). A core that reaches ~10–15 M⊕ but never crosses into runaway — because the gas dispersed first or the outer disk was sparse — ends as an ice giant like Uranus or Neptune: a core with only a thin envelope.
• Super-Earths / sub-Neptunes. Cores that stayed well below critical, or formed late, keep only a percent-level hydrogen envelope. These dominate the exoplanet census discovered by transit surveys.

Core accretion versus disk instability

The principal rival to core accretion is gravitational disk instability, in which a massive, cold disk fragments and collapses directly into gaseous clumps. The two mechanisms make sharply different predictions, and the table below summarizes where each is favored.

Property	Core accretion	Disk instability
Direction	Bottom-up (solids first)	Top-down (gas collapses directly)
Timescale	~10⁵–10⁷ yr (race vs. disk)	A few orbital periods (~10³ yr)
Solid core	~10 M⊕ core required	Little or no initial core
Heavy-element content	Naturally enriched (matches Jupiter/Saturn)	Near-solar unless later polluted
Favored location	Few–tens of AU, near/beyond snow line	Cold outer disk, tens–hundreds of AU
Disk mass needed	Ordinary minimum-mass disk works	Massive, marginally stable disk
Best explains	Jupiter, Saturn, ice giants, super-Earths	Massive wide-orbit imaged companions

Core accretion is the dominant paradigm because it accounts for the heavy-element enrichment of the Solar System's giants, the smooth continuum of planet masses from super-Earths up to gas giants, and the strong observed correlation between giant-planet occurrence and host-star metallicity (more solids, easier cores). Disk instability remains viable for a minority of systems — chiefly very massive planets on wide orbits, such as some of the bodies seen by direct imaging.

Observational status and tests

Core accretion is not just theory; it makes testable predictions, and several lines of evidence support it:

• The giant-planet metallicity correlation. Fischer & Valenti (2005) and later RV/transit surveys found that the frequency of giant planets rises steeply with host-star metallicity. More solid material means cores form faster and reach critical mass before the disk clears — a direct prediction of core accretion, and one that disk instability does not naturally make.
• Heavy elements in Jupiter and Saturn. The Galileo probe and Juno mission show Jupiter is enriched in heavy elements relative to the Sun, with a diluted "fuzzy" core rather than a sharp boundary — consistent with a solid core that accreted gas and was later mixed.
• Disk lifetimes. Spitzer and ground-based infrared surveys of young clusters (Haisch et al. 2001; Hernández et al. 2007) measure the fraction of stars retaining inner gas disks as a function of age, giving the 3–10 Myr dispersal window that core accretion must beat. ALMA now resolves the dust in these disks directly.
• Gaps and rings in protoplanetary disks. ALMA images of disks such as HL Tau and TW Hya show concentric gaps widely interpreted as carved by forming planets — possibly cores in the act of accreting their gas — within the relevant protoplanetary disk lifetime.
• The super-Earth / sub-Neptune population. Kepler revealed that small planets with modest gas envelopes are extraordinarily common, exactly the population expected from cores that grew but never reached runaway gas accretion.

Common pitfalls and misconceptions

• "The critical mass is exactly 10 M⊕." It is an order-of-magnitude threshold, not a constant. The actual value runs from about 5 to 15 Earth masses depending on envelope opacity, the rate at which solids fall in, and the local disk conditions. Treat "~10" as a benchmark, not a law.
• "Runaway gas accretion means the core keeps growing fast." No — the runaway is in the gas envelope. The solid core is largely finished by the time runaway begins; what explodes is the hydrogen-helium mass.
• "The gas disk lasts forever." The opposite is the central constraint. The gaseous disk is gone in 3–10 Myr, which is why building the core in time is the whole game.
• "Core accretion and disk instability are mutually exclusive." Both may operate, in different regimes. Core accretion dominates for typical giants; disk instability may form rare massive wide-orbit planets.
• "Pebble accretion replaced core accretion." Pebble accretion is a mechanism within core accretion for feeding the core faster — it is the modern fix to the timescale problem, not a competing theory.
• "Ice giants formed a different way." Uranus and Neptune fit naturally as core-accretion outcomes that reached ~10–15 M⊕ cores but ran out of gas (and time) before runaway, so they captured only modest envelopes.

Quantitative analysis: the race condition

The whole of core accretion can be compressed into a single inequality — the build-up time must beat the disk-dispersal time:

t_core + t_envelope  <  t_disk ≈ 3–10 Myr

The core build-up time depends on the solid surface density Σ_s and the growth mechanism. For oligarchic planetesimal growth at orbital frequency Ω, the time to reach a core mass M_core scales roughly as

t_core ∝ M_core^(1/3) / ( Σ_s · Ω )

so it grows with orbital distance (Ω falls as a^(−3/2)) and shrinks with more solids. This is precisely why the outer Solar System is hard for the planetesimal model: at 5 AU and beyond, Ω is small and t_core balloons toward the disk lifetime. Pebble accretion changes the scaling because the capture cross-section is set by gas drag rather than by gravitational focusing of fast bodies. In the relevant regime the pebble accretion rate is

Ṁ_pebble ≈ 2 · R_acc · ρ_peb · v_rel     (R_acc enlarged by gas drag)

which can be one to two orders of magnitude faster than planetesimal accretion for the same disk, dropping t_core from millions of years to a few hundred thousand. Once the core reaches M_crit, the second term t_envelope is short — the runaway is fast by construction — so beating the deadline reduces to building the core quickly. To make the numbers concrete, plug in a Jupiter-region core:

Target:   M_core = 10 M⊕ at a = 5.2 AU
Planetesimal-only:  t_core ≈ several Myr  → finishes near 8 Myr  (marginal)
Pebble-fed:         t_core ≲ 0.5 Myr      → finishes well before 3 Myr (safe)
Then runaway: ~30 M⊕ → ~318 M⊕ in <1 Myr either way

The arithmetic shows why both ingredients matter: the ~10 M⊕ critical mass sets what you must build, and the 3–10 Myr disk lifetime sets how fast. Core accretion succeeds wherever the disk hands over enough solids, fast enough, to win that race.