Rebound Migration in Protoplanetary Disks

• Rebound Migration is the outward movement of planets induced by an expanding inner disk cavity during late-stage protoplanetary disk dispersal.
• It arises from a steep density gradient at the cavity edge that creates a strong positive corotation torque, reversing typical inward migration.
• Multi-planet simulations reveal that rebound migration can break resonances, widen period ratios, and trigger dynamical instabilities based on planet mass and disk dispersal rates.

Searching arXiv for the specified paper and closely related work on rebound migration. Rebound migration is an outward migration regime induced by a moving inner boundary, and in planetary dynamics it denotes the outward “surfing” motion of a planet, or a group of planets, when they encounter the rapidly evolving inner edge of a disk cavity during the late dispersal phase of a protoplanetary disk. In the specific sense developed for photo-evaporating disks, stellar X-ray photoevaporation removes gas from the inside out, an inner cavity opens and expands outward, and a planet near the cavity edge experiences a very steep surface-density gradient that can make the corotation torque strongly positive and reverse the usual inward migration. The recent extension to multi-planet systems shows that this process can substantially reshape final architectures, including resonant breaking, widened period ratios, and dynamical instability, with outcomes that depend critically on planet mass, mass ordering, and the disk dispersal timescale (Liu et al., 9 Jun 2026).

1. Definition and physical scope

In the usage of hydrodynamical planet-disk interaction studies, rebound migration refers broadly to any outward migration associated with the moving cavity edge, including cases where the planet is effectively carried outward by the retreating inner disk boundary (Liu et al., 9 Jun 2026). This definition emphasizes that the mechanism is tied not merely to a static torque reversal at a disk edge, but to the coupled evolution of a planet and an expanding cavity during late disk clearing.

The physical setting is a protoplanetary disk undergoing inside-out dispersal. As stellar X-ray photoevaporation removes gas from the inner regions, an inner cavity opens and expands outward. Near that cavity edge, gas is depleted interior to the planet but remains exterior to it, so the local disk environment becomes highly asymmetric. In this regime, the positive corotation torque can overcome the usual inward Lindblad torque, and the net migration reverses (Liu et al., 9 Jun 2026).

This basic concept has a close analogue in the earlier “magnetospheric rebound” literature, where the inner disk is truncated by the stellar magnetosphere and the cavity expands outward as the disk accretion rate declines. There too, planets trapped near the inner edge can be displaced outward with the moving cavity, altering period ratios and breaking resonances (Liu et al., 2017, Liu et al., 2017). A later N-body study similarly treats rebound-driven expansion of the disk’s inner edge as a mechanism that destabilizes resonant chains of super-Earths and mini-Neptunes (Pan et al., 9 Sep 2025). The photoevaporative formulation differs in the driver of cavity expansion, but the shared dynamical idea is an outward migration response to a retreating inner boundary.

2. Torque asymmetry at an expanding cavity edge

The decisive condition for rebound migration is a positive net disk torque near the cavity edge. In the multi-planet hydrodynamical treatment, the torque on a planet is written as

$$\Gamma = \int \Sigma \frac{\partial \Phi_{\rm p}}{\partial \phi}\, r\,dr\,d\phi = \sum_r T(r),$$

where the torque density $T(r)$ is integrated over the disk; rebound occurs when this net $\Gamma$ becomes positive close to the cavity edge (Liu et al., 9 Jun 2026).

The mechanism is an asymmetric torque balance generated by the steep density gradient at the cavity boundary. Gas is strongly depleted interior to the planet while remaining exterior, so the positive corotation component can dominate. For low-mass planets, the effect is mainly attributed to the steep vortensity gradient and the resulting corotation torque. For more massive planets, the same paper states that gap-edge asymmetries and even eccentric disk structures can also contribute (Liu et al., 9 Jun 2026).

This torque-level description is consistent with the single-planet photoevaporation study, which reports that super-Earth and Neptune-mass planets experience a strong positive corotation torque along the cavity edge that leads to sustained outward migration, whereas Saturn-mass planets do not because significant gas depletion in their co-orbital regions weakens the corotation torque (Liu et al., 8 Sep 2025). For Jupiter-mass planets, modest outward migration can still occur, but the mechanism is attributed mainly to disk eccentricity rather than the standard low-mass corotation-torque picture (Liu et al., 8 Sep 2025).

A plausible implication is that “rebound migration” is best regarded not as a single torque formula, but as a family of outward responses generated by an evolving inner boundary, with the detailed torque source depending on planet mass and on whether the flow remains in a low-mass corotation-dominated regime or a gap-perturbed giant-planet regime.

3. Hydrodynamical formulation in photo-evaporating disks

The multi-planet study employs 2D hydrodynamical simulations with Dusty FARGO-ADSG, modeling only the gas disk and neglecting disk self-gravity (Liu et al., 9 Jun 2026). The governing equations are the continuity equation with a photoevaporative sink term and a momentum equation with pressure, gravity, and viscosity:

$$\frac{\partial \Sigma}{\partial t} + \nabla \cdot (\Sigma \vec{v}) = \dot{\Sigma}_{\rm PE}, \qquad \frac{\partial \vec{v}}{\partial t} + \vec{v}\cdot\nabla\vec{v} = -\frac{\nabla P}{\Sigma} - \nabla\Phi + \vec{f}_\nu .$$

The disk is locally isothermal, and viscosity is prescribed by

$\nu = \alpha c_s H,$

with $\alpha = 10^{-3}$ and aspect ratio $h_0 = 0.05$ (Liu et al., 9 Jun 2026).

The numerical workflow is staged. The disk is first evolved in 1D without planets until an inner cavity forms, and is then restarted in 2D with planets inserted at full mass (Liu et al., 9 Jun 2026). Photoevaporation follows the X-ray prescription of Owen et al., with mass loss peaked around $\sim 2$–$3$ au, producing inside-out clearing (Liu et al., 9 Jun 2026). The grid spans $0.2r_0$ to $T(r)$0 with $T(r)$1 au, and both standard and high-resolution meshes are used (Liu et al., 9 Jun 2026).

This setup extends the earlier single-planet hydrodynamical study, which also evolves the disk in 1D until a cavity forms and then restarts in 2D with a planet near the cavity boundary, explicitly targeting the photo-evaporation-dominated phase (Liu et al., 8 Sep 2025). The continuity between the two papers is important: the earlier work establishes the viability of rebound outward migration in 2D hydrodynamics for one planet, while the later work shows how the same mechanism operates in two-planet and three-planet systems and interacts with resonance dynamics (Liu et al., 8 Sep 2025, Liu et al., 9 Jun 2026).

4. Multi-planet dynamics: resonance breaking, divergence, and instability

The extension from one planet to multiple planets is the principal contribution of the 2026 study. It considers two-planet and three-planet systems with masses ranging from $T(r)$2 and $T(r)$3 up to $T(r)$4, $T(r)$5, and $T(r)$6, with initial configurations placed near the 3:2 or 2:1 mean-motion resonances (Liu et al., 9 Jun 2026).

For low-mass pairs, planets are first captured into resonance, but as one planet encounters the expanding cavity edge it can reverse migration while the companion continues differently. This divergent migration widens the period ratio and can break the original resonance. In the $T(r)$7–$T(r)$8 cases, the resonant angle switches from libration to circulation and the final period ratio becomes much larger than the initial resonant value (Liu et al., 9 Jun 2026). When the more massive planet is the one near the cavity edge, the rebound is stronger; when the less massive planet is near the edge, rebound is weaker and the pair may remain only modestly displaced from resonance (Liu et al., 9 Jun 2026).

Mixed-mass systems exhibit more contingent behavior. A $T(r)$9 inner planet with a $\Gamma$0 outer planet can bypass the 2:1 resonance and become trapped in 3:2 resonance, which may survive disk dispersal. Reversing the mass ordering changes the result: if the massive planet is inside, cavity expansion can drive divergent motion and leave the pair clearly displaced from resonance (Liu et al., 9 Jun 2026). The paper therefore identifies mass ordering as a control parameter, not merely total mass.

Massive pairs perturb the disk strongly. Gap opening and disk eccentricity alter the torque profile away from a simple two-lobed structure, yet rebound can still occur if the planets open a common gap and the inner massive planet retains a net positive torque from the residual inner disk (Liu et al., 9 Jun 2026). In some cases, the pair migrates outward together before the inner giant falls into the cavity and the outer planet continues outward alone (Liu et al., 9 Jun 2026). This is presented as consistent with earlier work on outward migration of gap-opening pairs, now embedded in a photoevaporating disk with an expanding cavity.

The three-planet simulations display the same underlying mechanism in a less regular setting. In the fiducial, more massive disk, convergent migration first traps the outer planets into resonance, their motion leads to scattering of the innermost planet, and later the outermost planet undergoes rebound as the cavity expands. The final configuration is a resonant inner pair plus a more distant isolated planet (Liu et al., 9 Jun 2026). This suggests that rebound migration can interact with resonant capture and planet-planet scattering to yield hybrid architectures rather than a single stereotyped outcome.

5. Timescales, disk dependence, and suppression of rebound

A central result of the 2026 simulations is that rebound migration is not guaranteed whenever a cavity expands. In lower-mass disks, the cavity expansion can be too rapid for planets to surf it effectively, so rebound is suppressed and resonant chains survive (Liu et al., 9 Jun 2026). The paper states explicitly that rebound migration does not happen for arbitrarily rapid cavity expansion: if the cavity moves outward too fast, planets cannot maintain torque balance with the moving edge; if it moves too slowly, the system can evolve differently as well (Liu et al., 9 Jun 2026).

This timescale sensitivity aligns with the earlier single-planet photoevaporation study, which argues that moderate photoevaporation rates are optimal. If the cavity expands too quickly, the planet is left behind; if too slowly, the corotation torque does not remain strongly asymmetric over the relevant libration time, and rebound becomes ineffective (Liu et al., 8 Sep 2025). That paper derives a characteristic critical mass for rebound,

$\Gamma$1

to summarize the competition between cavity expansion and torque response (Liu et al., 8 Sep 2025).

The same dependence on disk properties also appears in the older magnetospheric rebound framework. There, migration is substantial in a massive disk and minor in a light disk, and longer disk depletion times produce larger final orbital periods and wider period ratios (Liu et al., 2017, Liu et al., 2017). The photoevaporative and magnetospheric versions therefore share a common structural feature: rebound efficacy is set by the relative rates of cavity motion, torque-mediated migration, and gas depletion.

A common misconception is that a moving cavity edge necessarily drives outward migration. The hydrodynamical results do not support that simplification. Instead, they indicate a restricted dynamical window in which the cavity expansion timescale, disk mass, and planet mass jointly permit the planet to remain torque-coupled to the edge (Liu et al., 9 Jun 2026, Liu et al., 8 Sep 2025).

6. Relation to resonant-chain evolution and observed exoplanet architectures

The broader significance of rebound migration lies in its role as a late-stage architecture-setting process. The 2026 multi-planet study argues that rebound migration provides a natural mechanism for transforming initially compact, resonant multi-planet systems into widely separated, often non-resonant systems commonly observed around Sun-like stars (Liu et al., 9 Jun 2026). By generating divergent migration near disk dispersal, it can break mean-motion resonances, enlarge period ratios, and sometimes trigger dynamical instability and scattering (Liu et al., 9 Jun 2026).

This conclusion is continuous with the magnetospheric rebound program initiated in 2017. The original two-planet model was motivated by the discrepancy between migration theory, which tends to trap super-Earths into resonance, and Kepler statistics, which do not show strong pile-ups at exact resonances. The proposed solution was that late disk dispersal and cavity expansion move planets out of resonance after the earlier migration stage (Liu et al., 2017). The follow-up statistical assessment found that magnetospheric rebound tends to erase differences between migration and in-situ initial conditions, widening period-ratio distributions in both cases (Liu et al., 2017).

A later N-body study on compact chains of super-Earths and mini-Neptunes strengthens that interpretation. It reports that rebound-driven cavity expansion strongly increases the fraction of systems that lose resonance chains relative to a no-rebound control, and that the main effect is early disruption of resonant chains rather than major changes in the ensemble distributions of orbital period ratio, radius uniformity, or the radius valley (Pan et al., 9 Sep 2025). This suggests that rebound migration may strongly affect dynamical history while leaving several bulk observables comparatively unchanged.

Taken together, these results support a specific picture of late-stage planet formation: disk clearing is not a passive shutdown of migration but a phase in which the moving inner boundary can actively remap a system’s spacing and resonant structure (Liu et al., 9 Jun 2026). A plausible implication is that non-resonant final architectures need not imply non-migratory formation histories, because resonant capture and subsequent rebound-driven divergence can occur sequentially.

7. Conceptual boundaries and cross-disciplinary uses of the term

The term “rebound migration” is not unique to planet formation. In microhydrodynamics, a related phrase is used for wall-bounded soft-particle motion, where force-induced lateral migration toward a wall can be counteracted by wall repulsion, yielding a stable off-center equilibrium position interpreted as a rebound-like migration (Förtsch et al., 2017). In fluid-structure interaction, contactless rebound in a viscous incompressible fluid describes a body that migrates away from a wall without topological contact because elastic energy storage and shape-dependent drag asymmetry generate reverse motion (Gravina et al., 2020). These usages retain the general idea of reverse or outward motion induced by an interaction with a boundary, but they are mechanically distinct from the disk-cavity problem.

Within exoplanet dynamics, however, the technical meaning is comparatively specific. Rebound migration denotes outward orbital migration caused by the recession of an inner disk edge, whether that edge is set by stellar magnetospheric truncation or by photoevaporative cavity expansion (Liu et al., 2017, Liu et al., 9 Jun 2026). The photoevaporative hydrodynamical simulations refine this picture by showing that the effect persists in two-planet and three-planet systems and that its consequences include resonance breaking, widened period ratios, and instability, but only within a restricted domain of mass and dispersal timescale (Liu et al., 9 Jun 2026).

The present evidence therefore situates rebound migration as a late-disk, cavity-coupled transport mechanism with architecture-scale consequences. Its defining features are a moving inner boundary, a strongly asymmetric local torque environment, and a dynamical competition between outward cavity recession and the planet’s ability to remain torque-coupled to that boundary (Liu et al., 9 Jun 2026, Liu et al., 8 Sep 2025).