Toomre Q Stability

The tug-of-war in one number

Picture a thin disk of gas and stars whirling around a galaxy. Gravity is always trying to gather material into denser knots; if it ever wins locally, the disk fragments into clumps and arms. Two forces fight back. On small scales, random motion resists collapse — pressure for a fluid, velocity dispersion for a swarm of stars — because a tiny over-dense patch simply disperses before it can pull itself together. On large scales, the disk's own rotation resists collapse, because differential rotation shears any extended over-density apart faster than gravity can assemble it.

The trouble is that these two defenders cover opposite ends of the wavelength spectrum. Pressure protects short wavelengths; rotation protects long ones. The dangerous region is in between. Alar Toomre's insight in 1964 was that the disk is stable everywhere only if the two defenders' coverage overlaps — if the largest wavelength pressure can kill is bigger than the smallest wavelength rotation can kill, there is no gap for gravity to exploit. The Q parameter is precisely the ratio that measures whether that overlap exists. Q greater than 1 means the defenders overlap and the disk is smooth; Q less than 1 means a window of vulnerable wavelengths is open and the disk fragments; Q near 1 is the knife-edge where marginal structure — flocculent spiral arms, rings, beads-on-a-string — emerges.

The criterion and its two forms

For a razor-thin, differentially rotating stellar disk, Toomre's local linear analysis gives the stability parameter

Q = σ_R κ / (3.36 G Σ)        (collisionless stellar disk)

where σ_R is the radial velocity dispersion of the stars, κ is the epicyclic frequency, Σ is the disk surface mass density, and G is Newton's constant. The disk is stable to all axisymmetric (ring-shaped) perturbations when Q > 1 and unstable when Q < 1. For a gaseous (fluid) disk the same physics yields a slightly different coefficient,

Q = c_s κ / (π G Σ)           (gaseous fluid disk)

with c_s the gas sound speed in place of the dispersion, and π ≈ 3.1416 replacing 3.36. The difference comes from how a collisionless star gas responds to a perturbation (Landau-damped, phase-mixed) versus how a collisional fluid does (a pressure wave). Because 3.36 > π, a stellar disk is marginally easier to destabilise than a fluid disk with the same numerator — gas pressure is a slightly more effective stabiliser than random stellar motions.

When Q falls below 1, instability does not act on all scales equally — it lives in a band bracketed by two characteristic lengths. The longest unstable (critical) wavelength is set by rotation alone balancing self-gravity, and the fastest-growing (most unstable) wavelength, which sets the size of the clumps that actually appear, is exactly half of it at the marginal Q = 1 point:

λ_crit = 4π² G Σ / κ²          (longest unstable / critical wavelength)
λ_max-growth = 2π² G Σ / κ²    (fastest-growing wavelength at Q = 1)

This is why fragmenting disks produce clumps of a preferred size rather than collapsing wholesale — the instability has a built-in length scale set by the competition between gravity, rotation, and surface density. The Toomre mass that fragments out is roughly M_T ≈ Σ λ_crit².

Where the epicyclic frequency comes from

The κ in the numerator is the heart of the rotational stabilisation. The epicyclic frequency is the rate at which a star nudged radially off a circular orbit oscillates in and out — the frequency of the little ellipse (the epicycle) it then traces relative to its guiding centre. It is set entirely by the shape of the rotation curve:

κ² = (2Ω / R) d(R² Ω)/dR = 4Ω² + 2Ω R (dΩ/dR)

where Ω(R) is the orbital angular velocity. Three reference cases are worth memorising:

Rotation law	Ω(R)	κ in terms of Ω	Where it occurs
Solid body	Ω = const	κ = 2Ω	Inner galaxy bulge, rigid cores
Flat rotation curve	v = const, Ω ∝ 1/R	κ = √2 Ω	Bulk of spiral disks
Keplerian	Ω ∝ R⁻³ᐟ²	κ = Ω	Protoplanetary disks, AGN disks
Declining (rare)	steeper than Kepler	κ < Ω	(unstable to ring modes — generally not seen)

The larger κ is, the stiffer the rotational restoring force and the more strongly long wavelengths are stabilised. Because most of a galaxy disk has a flat rotation curve, κ = √2 Ω is the workhorse value, and it makes the Toomre criterion straightforward to evaluate from observable kinematics.

From Jeans to Toomre: adding rotation

The Toomre criterion is the rotating-disk descendant of the classic Jeans instability. In a non-rotating, uniform medium, Jeans showed that self-gravity overcomes pressure above a critical length λ_J = c_s √(π / G ρ): anything bigger collapses, anything smaller bounces back as a sound wave. There is no upper bound on the unstable scale — the biggest blobs are the most unstable.

Put that medium into a rotating disk and the picture inverts at the long-wavelength end. Pressure still protects short wavelengths exactly as before, but now rotation introduces a second restoring force that kills the long wavelengths the Jeans analysis left wide open. The dispersion relation for axisymmetric perturbations in a thin gas disk makes this explicit:

ω² = κ² − 2π G Σ |k| + c_s² k²

Here k is the wavenumber of the perturbation. The κ² term (rotation) dominates at small k (long wavelength); the c_s² k² term (pressure) dominates at large k (short wavelength); the destabilising self-gravity term −2π G Σ |k| sits in between. Instability (ω² < 0) requires that the gravity term win somewhere in the middle. Minimising over k shows it can only win if

c_s κ / (π G Σ) < 1     i.e.     Q < 1

So Q is not an ad-hoc guess — it falls straight out of demanding that the discriminant of the dispersion relation stays positive. As κ → 0 the rotational term vanishes, Q → 0, and you recover an everywhere-unstable, Jeans-like medium.

Quantified examples across the disk zoo

Q is a local number, so it is most useful as a profile Q(R) across a disk and a comparison across object classes. The table below gives representative values; the cold, gas-rich systems cluster near the marginal Q ≈ 1 boundary where structure grows, while hot, old, gas-poor disks sit safely above it.

System	Component	Typical Q	State
Milky Way solar neighbourhood	Old thin-disk stars	≈ 2 – 2.5	Axisymmetrically stable
Milky Way solar neighbourhood	Cold atomic + molecular gas	≈ 1 – 2	Marginal; hosts star formation
Nearby spiral (e.g. M51) arms	Gas in arms	≈ 0.7 – 1.3	Locally fragmenting → GMCs
High-redshift clumpy disk (z ≈ 2)	Gas-rich turbulent disk	≈ 0.5 – 1	Violently unstable → 10⁸–10⁹ M☉ clumps
Protoplanetary disk, outer region	Cold gas + dust	≈ 1 – few	Marginal; possible direct GI planets
Elliptical galaxy / hot bulge	Hot pressure-supported stars	≫ 1 (disk Q n/a)	No disk instability

A concrete solar-neighbourhood estimate for the old stellar disk: with σ_R ≈ 38 km/s, the local epicyclic frequency κ ≈ 37 km/s/kpc, and surface density Σ ≈ 50 M☉/pc², plugging into Q = σ_R κ / (3.36 G Σ) gives Q ≈ 2.0. The same calculation for the cold gas, using c_s ≈ 6–8 km/s for the molecular phase and Σ_gas ≈ 10–13 M☉/pc² with the gas coefficient π, lands in the Q ≈ 1–2 range — much closer to the marginal boundary, which is where star formation is observed to concentrate. The disk is not uniformly stable; it is the marginal gas that builds the spiral arms.

Why disks live at Q ≈ 1

One of the most striking facts about real disks is that they are not found scattered across all values of Q — they pile up near Q ≈ 1. This is not a coincidence; it is a thermostat. If a patch of disk drops below Q = 1, it goes unstable, fragments, and forms structure. That structure (clumps, spiral shocks, young stars, supernovae) stirs up the gas, raising its velocity dispersion σ_R or c_s and, because Q ∝ σ, pushing Q back up toward and above 1. Conversely, if a region is too hot (Q ≫ 1), nothing forms, the gas radiates away its turbulence, σ falls, and Q drifts back down. The disk is held at the boundary by negative feedback — gravitational heating from instability balanced by radiative cooling.

This self-regulation has a profound consequence for galaxy evolution: the star-formation rate of a disk is set, in part, by how it sheds the energy that keeps it marginally stable. The empirical Kennicutt-Schmidt star-formation law and the existence of a star-formation threshold surface density (below which disks turn off, observed in the outskirts of spirals) are both natural readings of a disk regulating itself to Q ≈ 1.

Where Toomre Q shows up

• Spiral arms. Marginally stable disks (Q just above 1) are fertile ground for swing-amplified, shearing spiral waves. The grand-design two-armed spirals and the flocculent multi-armed spirals are both consequences of disks sitting near the instability boundary, where non-axisymmetric modes grow even though axisymmetric ones are quenched.
• Giant molecular clouds. The ≈ 10⁶ M☉, ≈ 100 pc clouds that dominate molecular gas in spiral disks have masses and spacings close to the Toomre mass M_T ≈ Σ λ_crit² and the Toomre wavelength — direct fingerprints of the gravitational instability that births them along arms.
• Clumpy high-redshift galaxies. At z ≈ 1–3, gas-rich turbulent disks have Q well below 1 and fragment into a handful of giant kpc-scale clumps of 10⁸–10⁹ M☉ each, glowing with vigorous star formation — the "chain" and "clump-cluster" galaxies seen by Hubble and JWST. These clumps may migrate inward and help build galactic bulges.
• Protoplanetary disks. In the cold outer regions of a massive young disk, Q can approach 1, raising the possibility of direct gravitational-instability formation of giant planets and brown dwarfs at tens of AU — an alternative to slow core accretion, invoked for wide-orbit companions like those imaged around HR 8799.
• AGN and accretion disks. The outer, self-gravitating parts of accretion disks around supermassive black holes drop to Q ≈ 1 beyond a self-gravity radius of order 0.01–0.1 pc, where the disk fragments into stars instead of accreting smoothly — a leading explanation for the puzzling disk of young, massive stars observed at ≈ 0.04–0.5 pc from Sgr A* in the Galactic Centre.

Refinements and modern extensions

• Two-component (stars + gas) Q. Real disks have both a hot stellar component and a cold gas component, and they destabilise together. The combined criterion (Wang & Silk 1994; Romeo & Wiegert 2011) is always more unstable than either component alone — a disk can be gravitationally unstable even when both Q_star and Q_gas individually exceed 1.
• Thickness correction. Toomre's disk is razor-thin. Finite vertical thickness dilutes the in-plane self-gravity and stabilises the disk, raising the effective Q by a factor of roughly (1 + k h) for a disk of scale height h — typically a 10–50% effect.
• Swing amplification and Q_eff ≈ 2. Toomre (1981) showed that non-axisymmetric trailing waves can be transiently amplified by factors of 10–100 for Q up to ≈ 2. So the practical threshold for visible spiral structure is Q ≈ 1.5–2, not the strict axisymmetric Q = 1.
• Magnetic and viscous effects. In magnetised gas disks, magnetic tension modifies the dispersion relation and the magnetorotational instability provides a parallel route to turbulence; in viscous disks, gravitational instability can saturate into a self-sustained, marginally-stable turbulent state (gravitoturbulence).

Common misconceptions and edge cases

• "Q > 1 means perfectly stable." It means stable to axisymmetric ring modes only. Trailing spiral (non-axisymmetric) waves swing-amplify up to Q ≈ 2, which is why disks with Q ≈ 1.5 still show prominent spiral arms. Axisymmetric stability is necessary, not sufficient, for a featureless disk.
• Confusing Q with the Jeans criterion. The Jeans length has no upper cutoff — the largest blobs are most unstable. Toomre adds rotation, which stabilises the largest scales and leaves only an intermediate band vulnerable. The two are not interchangeable; Q reduces to a Jeans-like situation only as κ → 0.
• Using the wrong coefficient. The stellar value 3.36 and the gas value π are not interchangeable. Using 3.36 for gas understates its instability; using π for stars overstates it. Pick the coefficient that matches the component you are analysing — and for mixed disks use the two-fluid combined Q, which is more unstable than either.
• Treating Q as a global property. Q is local. A galaxy can be stable in its hot inner disk (high κ, high σ) and unstable in a cold gas ring at larger radius. The meaningful object is the profile Q(R), not a single number for the whole galaxy.
• Forgetting the most unstable wavelength. Even a Q < 1 disk does not collapse monolithically — it fragments at λ_T = 4π² G Σ / κ², producing clumps of a characteristic mass and spacing. The instability selects a scale; it does not simply "collapse the disk."