Gravitoturbulence in Astrophysical Disks

• Gravitoturbulence is defined as the self-regulated turbulent state in massive, rotating disks where gravitational instability is balanced by slow radiative cooling.
• It drives persistent spiral density waves and a multiscale turbulent cascade, effectively transporting angular momentum and influencing planetesimal formation.
• The phenomenon is marked by non-fragmenting conditions, dominance of solenoidal motions, and interactions with magnetic fields and dust evolution.

Gravitoturbulence refers to the self-regulated turbulent state arising in sufficiently massive, differentially rotating astrophysical disks when gravitational instability (GI) is present but rapid fragmentation is averted by sufficiently slow radiative cooling. This phenomenon is fundamental to the angular momentum transport, disk evolution, planetesimal formation, and dynamo activity in a range of systems, notably protoplanetary and circumplanetary disks, as well as in outer regions of AGN disks. At its core, gravitoturbulence results from the interplay between disk self-gravity, differential rotation, and the cooling timescale, producing persistent spiral density waves, large-scale and small-scale turbulence, and a rich set of secondary phenomena.

1. Instability Criteria, Onset, and Turbulent State

The onset of gravitoturbulence is governed by the Toomre parameter:

$Q \equiv \frac{c_s \kappa}{\pi G \Sigma}$

where $c_s$ is the local sound speed, $\kappa$ the epicyclic (orbital) frequency (typically $\Omega$ for Keplerian disks), $G$ the gravitational constant, and $\Sigma$ the surface density. Disks are unstable to axisymmetric self-gravity when $Q \lesssim 1$ (Booth et al., 2018, Klee et al., 2019, Shi et al., 2014, Riols et al., 2016, Keith et al., 2014). Non-axisymmetric gravitational instability persists for $Q\lesssim1.4$–1.7 depending on disk thickness and other effects.

Instability alone is not sufficient to sustain gravitoturbulence. For this, radiative cooling ($t_{\mathrm{cool}}$) must be slow enough to prevent immediate fragmentation into bound clumps. The key dimensionless parameter is the cooling time in orbital units, $\beta \equiv t_{\mathrm{cool}}\Omega$. If $c_s$0 is too short, fragmentation dominates; for moderate or long $c_s$1, the disk self-regulates into a quasi-steady, turbulent state with $c_s$2 (Booth et al., 2018, Klee et al., 2019).

Gravitoturbulence self-consistently balances shock heating from spiral arms and compressional work against imposed cooling, determining both the amplitude of turbulence and effective angular momentum transport.

2. Statistical, Morphological, and Spectral Properties

Gravitoturbulent disks are characterized by strong, large-scale, transient spiral density waves and a multiscale turbulent cascade.

Key features:

• Spiral Wakes: Dominant, non-axisymmetric spiral density features with measured pitch angles $c_s$3–$c_s$4 and azimuthal wavelengths $c_s$5 ($c_s$6 is disk scale height) (Booth et al., 2018, Béthune et al., 2021).
• Spatial Power: On large scales ($c_s$7), turbulence exhibits quasi-2D Kolmogorov-like spectra $c_s$8 in the velocity field, with in-plane motions dominating. On smaller scales ($c_s$9; $\kappa$0), the flow becomes nearly isotropic (Booth et al., 2018, Riols et al., 2017).
• Solenoidal Dominance: Decomposition of the turbulent velocity shows $\kappa$170% of the power is in solenoidal (incompressible) modes at all resolved scales, consistent with a turbulent cascade driven by inertial waves (Booth et al., 2018, Riols et al., 2017).
• Locality: In thin disks ($\kappa$2), angular-momentum and energy transport is largely local in both time and space when time-averaged over several orbits or scale heights. The Shakura–Sunyaev $\kappa$3 parameter, measured as $\kappa$4stress$\kappa$5, closely tracks the inverse of the effective cooling time: $\kappa$6 (Xu et al., 26 Apr 2025, Béthune et al., 2021).
• Clumpy Morphology: In saturated gravito-turbulence, spiral structure is dominated by recurrent, high-contrast clumps at corotation, stemming from nonlinear mode coupling across a range of azimuthal wavenumbers. These structures have typical scales $\kappa$7 in both $\kappa$8 and $\kappa$9, with lifetimes of $\Omega$0–$\Omega$1 orbits (Xu et al., 26 Apr 2025).

3. Fragmentation, Convergence, and Domain Size

Fragmentation occurs when cooling is sufficiently rapid ($\Omega$2 in 3D; higher in some 2D studies), so that heating via shocks cannot balance energy loss, and overdensities collapse gravitationally on a dynamical timescale (Booth et al., 2018, Klee et al., 2019). However, for marginally longer cooling ($\Omega$3–$\Omega$4), stochastically triggered fragmentation may occur, typically at highest numerical resolutions. For $\Omega$5, disks are robustly stable against fragmentation for simulation times of hundreds of orbits (Booth et al., 2018, Klee et al., 2019).

Numerically, convergence of the fragmentation boundary and turbulent properties requires large simulation domains and high resolution. In 3D shearing-box setups, the azimuthal box size must satisfy $\Omega$6 in order to resolve the dominant spiral wavelength and avoid artificial burstiness or suppressed turbulence. Similarly, in 2D, box sizes $\Omega$7 are required (Booth et al., 2018, Riols et al., 2017). Adequate resolution, generally $\Omega$8 zones per $\Omega$9, is necessary to properly capture the critical scale-height and Toomre length turbulent components that mediate fragmentation (Klee et al., 2019).

4. Vertical Structure, Parametric Instabilities, and Energy Transport

Gravitoturbulent velocity fluctuations are nearly vertically uniform, rising by only a factor of $G$0 from the midplane to several scale heights above, in stark contrast with MRI-driven turbulence where vertical gradients in velocity amplitude can exceed factors of $G$1 (Shi et al., 2014, Riols et al., 2018). Such vertical uniformity arises because even non-self-gravitating material at $G$2 is strongly forced by midplane overdensities through the long-range nature of gravity.

Three-dimensional simulations reveal the existence of poloidal, baroclinically-generated rolls associated with spiral wakes, with vertical velocities $G$3–$G$4 and typical widths $G$5–$G$6 (Riols et al., 2018). Parametric instabilities in the upper layers, triggered by large-scale epicyclic oscillations, drive nearly isotropic inertial-wave turbulence at $G$7–$G$8, seen as fine-scale, high-wavenumber kinetic motions (Riols et al., 2017). These features enhance vertical mixing, support the disk against collapse, and can hinder inward solid sedimentation.

5. Gravitoturbulence and Dust Evolution

Gravitoturbulent structure critically alters the collision velocities, spatial distribution, and growth conditions of solid particles:

• Diffusion and Forcing: Small dust grains ($G$9; Stokes number) couple to the turbulent gas and diffuse with $\Sigma$0–$\Sigma$1, with relative velocities scaling as $\Sigma$2 (Booth et al., 2018, Shi et al., 2016).
• Large Solids: Particles with $\Sigma$3 undergo strong stochastic gravitational stirring, leading to elevated radial diffusion and high eccentricities, raising destructive collision rates (Shi et al., 2016).
• Filament Formation: Intermediate-size particles ($\Sigma$4–$\Sigma$5) are uniquely susceptible to being swept into narrow, dense filaments by the interaction of spiral gravitational forcing and gas drag. Within these filaments, local dust-to-gas ratios can reach $\Sigma$6–$\Sigma$7 (Shi et al., 2016, Baehr et al., 2022).
• Collapse to Planetesimals: When the local dust density in filaments exceeds the Hill/Roche limit, direct gravitational collapse into planetesimals or planetary embryos is triggered, with bound objects up to several $\Sigma$8 for $\Sigma$9 at distances $Q \lesssim 1$0~AU (Baehr et al., 2022). Dust-gas backreaction modulates clump mass and multiplicity but does not prevent embryo formation.
• Vertical Mixing: Poloidal rolls and inertial-wave turbulence loft small grains into the disk atmosphere ($Q \lesssim 1$1 for $Q \lesssim 1$2), impeding efficient midplane sedimentation and thereby potentially delaying or modifying planetesimal formation (Riols et al., 2018).

6. Interaction with Other Angular Momentum Transport Mechanisms and Dynamos

Gravitoturbulence coexists and interacts with magnetorotational instability (MRI) and large-scale magnetic fields.

• Angular Momentum Transport: In pure GI, spiral wakes and correlated overdensities drive gravitational $Q \lesssim 1$3 Reynolds stresses, measured via the effective $Q \lesssim 1$4-parameter, typically $Q \lesssim 1$5–$Q \lesssim 1$6 for $Q \lesssim 1$7 (Booth et al., 2018, Béthune et al., 2021, Riols et al., 2016).
• MRI Suppression: Strong gravito-turbulence suppresses canonical zero-net-flux MRI for $Q \lesssim 1$8, yet can sustain its own spiral-wave dynamo, amplifying magnetic fields to near-equipartition levels via midplane vortex stretching and $Q \lesssim 1$9-effects (Riols et al., 2017, Béthune et al., 2022).
• Dynamo Activity: The GI-driven dynamo, distinct from classic mean-field $Q\lesssim1.4$0-$Q\lesssim1.4$1 mechanisms, relies on turbulent electromotive force arising from velocity-magnetic field correlations induced by spiral wakes and poloidal rolls. The process is efficient at moderate cooling ($Q\lesssim1.4$2–$Q\lesssim1.4$3) and moderate magnetic Reynolds numbers ($Q\lesssim1.4$4–$Q\lesssim1.4$5), saturating when Maxwell stresses and energy reach those of gravitational stresses, at which point the background $Q\lesssim1.4$6 and disk scale height increase beyond the instability threshold (Béthune et al., 2022, Riols et al., 2017).

7. Observational and Astrophysical Implications

Gravitoturbulence imprints distinctive signatures on disk morphology, kinematics, and solid body evolution:

• Observational Diagnostics: Turbulent velocity fields with nearly flat vertical profiles may be revealed via multi-line molecular spectroscopy, distinguishing gravito-turbulence from MRI-driven turbulence (Shi et al., 2014). Filamentary, clumpy spiral structure can masquerade as planets or companions in continuum imaging when telescope resolution is comparable to $Q\lesssim1.4$7 (Xu et al., 26 Apr 2025).
• Variability and Accretion: Stochastic dissipation in transient spiral wakes can cause significant accretion-rate variability on orbital timescales (Béthune et al., 2021, Xu et al., 26 Apr 2025).
• Planet and Core Formation: Direct embryo formation in filaments or clumps—implicated for the earliest stages of core and giant planet formation at large orbital separation—proceeds rapidly, providing seeds for later growth via pebble or gas accretion (Baehr et al., 2022, Shi et al., 2016).
• Circumplanetary and AGN Disks: In circumplanetary disks, gravitoturbulence dominates transport in the outer disk ($Q\lesssim1.4$8), sets the effective viscosity, prevents fragmentation for high cooling times, and determines the disk mass and temperature structure (Keith et al., 2014). Analogous mechanisms operate in the outer regions of AGN disks, with potentially similar implications for black hole feeding and star formation.

---

References:

• (Booth et al., 2018) Booth & Clarke, "Characterizing gravito-turbulence in 3D..."
• (Klee et al., 2019) Klee et al., "Closing the gap to convergence of gravitoturbulence in local simulations"
• (Shi et al., 2014) Shi & Chiang, "Gravito-Turbulent Disks in 3D: Turbulent Velocities vs. Depth"
• (Riols et al., 2017) Riols et al., "Gravito-turbulence and the excitation of small-scale parametric instability..."
• (Riols et al., 2018) Riols & Latter, "Spiral density waves and vertical circulation in protoplanetary discs"
• (Béthune et al., 2021) Riols et al., "Spiral structures in gravito-turbulent gaseous disks"
• (Xu et al., 26 Apr 2025) Zhu et al., "Global Simulations of Gravitational Instability in Protostellar Disks..."
• (Shi et al., 2016) Shi et al., "Dust dynamics in 2D gravito-turbulent disks"
• (Baehr et al., 2022) Baehr et al., "Direct Formation of Planetary Embryos in Self-Gravitating Disks"
• (Hirose et al., 2017) Hirose & Shi, "Gravito-turbulence in irradiated protoplanetary discs"
• (Keith et al., 2014) Keith & Wardle, "Accretion in giant planet circumplanetary disks"
• (Riols et al., 2016) Riols & Latter, "Gravitoturbulence in magnetised protostellar discs"
• (Béthune et al., 2022) Riols et al., "Gravitoturbulent dynamo in global simulations of gaseous disks"
• (Riols et al., 2017) Riols & Latter, "Magnetorotational instability and dynamo action in gravitoturbulent astrophysical discs"