Disk Instability Planet Formation

Why disk instability matters

For decades, the standard story of how giant planets are born was core accretion: dust sticks into pebbles, pebbles into planetesimals, planetesimals into a solid core of roughly 5–15 Earth masses, and only then does the core capture a runaway envelope of gas. The trouble is timing. That whole sequence has to finish before the star blows away its gas — and typical protoplanetary disks dissipate in only about 3 million years. In the cold outer disk, tens of astronomical units out, orbital periods are so long that core accretion simply cannot build a core in time. Yet direct imaging keeps finding massive planets exactly there.

• It solves a speed problem. Fragmentation completes in ~10³–10⁴ years, sidestepping the millions of years core accretion requires far from the star.
• It explains wide-orbit giants. Massive planets at 30–100+ AU, like the four planets of HR 8799, are hard to make in place any other way.
• It links planets to stars. The same gravitational collapse physics that fragments a molecular cloud into stars operates one scale down inside a disk.
• It is a testable dichotomy. Disk-instability planets form from bulk gas and should be metal-poor relative to core-accretion giants — a fingerprint spectroscopy can chase.
• It bridges to brown dwarfs. The upper end of the process blurs into star formation, connecting planet birth to the Jeans instability.

How it works, step by step

Disk instability is a competition between three forces acting on a parcel of disk gas: self-gravity tries to pull the gas together, while thermal pressure and rotational shear try to hold it apart. The theory is essentially the bookkeeping of who wins.

1. Build a massive, cold disk. The instability needs a lot of mass — a disk that is a substantial fraction (roughly 10% or more) of the star's mass — and it needs to be cold, so that sound speed and pressure support are low.
2. Drive the Toomre Q below 1. As the disk grows in surface density Σ or cools (lowering c_s), the local Toomre parameter drops. Once Q < ~1, self-gravity beats pressure and rotation at some radius.
3. Grow spiral arms. The disk does not fragment uniformly; it first develops strong self-gravitating spiral density waves, funneling gas into dense filaments. ALMA has photographed exactly this pattern in the disk around Elias 2-27.
4. Cool fast enough to fragment. A dense clump heats as it compresses. It only survives if it radiates that heat away in less than a few orbits — the Gammie criterion β = t_cool·Ω ≲ 3. This works in the cold, optically thin outer disk, not the warm interior.
5. Fragment into bound clumps. A spiral arm pinches off into a self-gravitating clump of many Jupiter masses of gas. This is the actual "instant" of planet formation — over in a few orbital periods.
6. Contract into a protoplanet. The clump slowly contracts and heats over the following ~10⁴–10⁵ years, potentially with dust grains sedimenting toward its center — a possible route to a solid core.

The governing equation: Toomre Q

The single most important number in disk instability is the Toomre stability parameter, first derived by Alar Toomre in 1964 for the stability of galactic gas disks:

Q = c_s κ / (π G Σ)

Every symbol has a clear physical meaning and units:

Symbol	Quantity	Role / units
cs	Gas sound speed	Pressure support (m/s); larger = more stable
κ	Epicyclic frequency (≈ Ω for a Keplerian disk)	Rotational/shear support (s⁻¹); larger = more stable
Σ	Disk surface density	Self-gravity driver (kg/m²); larger = more unstable
G	Gravitational constant	6.674 × 10⁻¹¹ m³ kg⁻¹ s⁻²
Q	Toomre parameter	Dimensionless; Q < 1 → unstable

Read it as a tug-of-war: the numerator (c_sκ) is stabilizing — hot, fast-spinning gas resists collapse — and the denominator (πGΣ) is destabilizing — a heavy disk collapses under its own weight. A cold (small c_s), massive (large Σ), slowly-rotating outer disk (small κ) drives Q down. In razor-thin 2D theory the threshold is exactly Q = 1; real 3D simulations show fragmentation kicking in closer to Q ≈ 1.4–1.7, but the low-Q requirement is the same. Crucially, Q alone is not enough — Gammie (2001) showed a clump also needs a short cooling time, β = t_cool·Ω ≲ 3, or it just re-inflates. Both conditions must hold at once, which is why disk instability is confined to the cold, distant regions of unusually massive disks.

Disk instability vs. core accretion

The two theories are not rivals so much as different channels that dominate in different regimes. This is the comparison every planet-formation student learns first:

Property	Disk Instability	Core Accretion
Direction	Top-down (gas fragments)	Bottom-up (solids grow first)
Timescale	~10³–10⁴ years	~1–10 million years
Where it works best	Cold outer disk, ~30–100+ AU	Inner disk, inside the snow line to ~10s AU
Disk mass needed	Massive (≳ 10% of star)	Modest
Key requirement	Q < 1 AND fast cooling (β ≲ 3)	Reach critical core mass (~10 M⊕) before gas disperses
Typical products	Massive giants, brown dwarfs (~1–13+ M_J)	Sub-Neptune to Jupiter-class giants
Heavy-element content	Lower (bulk gas)	Higher (built on a solid core)
Champions	Kuiper 1951; Boss 1997	Safronov 1969; Pollack et al. 1996

A worked intuition: where does Q dip below 1?

Consider a disk around a solar-mass star. At radius r the orbital frequency is Ω = √(GM★/r³), so κ ≈ Ω shrinks steeply as you move outward — at 50 AU the disk turns roughly thirty times more slowly than at 5 AU. Temperature also falls with distance, driving down the sound speed c_s ∝ √T. Both effects push Q lower in the outer disk. Meanwhile the surface density Σ, though it declines outward, does so more gently than κ drops. The net result: for a sufficiently massive disk, Q is comfortably above 1 in the warm inner regions but sags toward and below 1 somewhere beyond ~30–50 AU. That is precisely the zone where wide-orbit giants like the HR 8799 planets (at ~16, 27, 43, and 68 AU) sit — and precisely where core accretion runs out of time. Add the cooling requirement, which is also easiest to satisfy in the cold, optically thin outer disk, and the two conditions overlap in the same distant annulus. That overlap is the whole physical argument for disk instability.

A short history

Gerard Kuiper first suggested in 1951 that a massive disk might fragment gravitationally into protoplanets. Alar Toomre's 1964 stability analysis — done for galaxies, not planets — gave the field its central diagnostic, the Q parameter. The idea was largely eclipsed by the success of core accretion in the 1970s–80s, until Alan Boss revived it in 1997 with 3D hydrodynamic simulations showing that a marginally unstable disk really could spawn Jupiter-mass clumps in a few orbital periods. Charles Gammie's 2001 shearing-box study then supplied the missing ingredient — the cooling-time criterion — explaining why fragmentation is restricted to the outer disk and settling a long dispute about whether clumps survive. Modern radiation-hydrodynamics and ALMA imaging of spiral-armed disks (Elias 2-27 in 2016; the AB Aurigae candidate protoplanet in 2022) keep the theory alive as a real, if secondary, formation channel.

Common misconceptions

• "Low Q alone makes a planet." No — a disk with Q < 1 that cannot cool fast enough just settles into steady gravitoturbulence and self-regulates back to Q ≈ 1. You need Q < 1 and β = t_cool·Ω ≲ 3.
• "It replaces core accretion." It doesn't. Most giant planets, especially close-in ones, are still best explained by core accretion. Disk instability is a complementary channel for the wide, massive outliers.
• "It can make terrestrial planets." It fragments gas, so it makes gas giants, not rocky worlds. Any rocky remnant would require the speculative tidal-downsizing route (grain sedimentation plus tidal stripping).
• "It works everywhere in the disk." The inner disk is too warm to cool efficiently and too fast-rotating (high κ) to reach low Q. Fragmentation lives in the cold outer disk.
• "Spiral arms mean a planet is forming." Self-gravitating spiral arms are the precursor, but many spiral disks never fragment — they simply transport angular momentum and stay marginally stable.
• "It's been directly observed." Not caught in the act. We see unstable, spiral-armed disks and candidate wide protoplanets, but no unambiguous clump-to-planet event yet.