Super-Eddington Accretion Flows

Updated 17 January 2026

Super-Eddington accretion flows are astrophysical systems where mass accretion rates surpass the Eddington limit, resulting in thick, turbulent disks with significant photon trapping.
They launch strong, radiation-pressure-driven winds and broad-angle outflows that carry substantial mass and kinetic energy at speeds up to 0.3c.
Multidimensional GRRMHD simulations reveal that magnetic fields and turbulent energy transport enhance radiative efficiency and produce anisotropic emission profiles.

Super-Eddington accretion flows refer to astrophysical accretion systems in which the mass accretion rate onto a compact object (such as a black hole or neutron star) exceeds the Eddington rate—the theoretical limit at which radiation pressure outward equals gravitational attraction inward. In these flows, radiative and mechanical feedback, photon trapping, and multidimensional instabilities produce a qualitatively distinct global structure from thin-disk or sub-Eddington paradigms. Super-Eddington accretion is central to the phenomenology of ultraluminous X-ray sources, tidal disruption events, ultra-luminous X-ray pulsars, and the rapid early growth of black hole seeds in the high-redshift universe.

1. Fundamental Theory and Global Structure

The Eddington accretion rate $\dot{M}_{\rm Edd}$ is defined by balancing gravity and the outward force of Thomson-scattered radiation, such that $\dot{M}_{\rm Edd} = L_{\rm Edd}/(\eta c^2)$ with $L_{\rm Edd} = 4\pi GM m_p c / \sigma_T$ and $\eta$ the radiative efficiency. In standard $\alpha$ -disk theory, flows remain thin ( $H/R \ll 1$ ) and radiatively efficient ( $\eta \sim 0.06$ –0.1). When $\dot{M} \gg \dot{M}_{\rm Edd}$ , however, the "slim disk" solution applies: the vertical disk thickness inflates to $H/R \sim 0.3$ –1, radial advection of heat becomes dynamically important, and radiative efficiency decreases logarithmically with mass accretion rate ( $L_{\rm rad} \simeq L_{\rm Edd}[1 + \ln (\dot{M}/\dot{M}_{\rm Edd})]$ ) (Lasota et al., 2015, Mayer, 2018, Jiang et al., 2024).

Multi-dimensional general-relativistic radiation magnetohydrodynamic (GRRMHD) simulations demonstrate that photon trapping and turbulent advection dominate over diffusive radiative transport in the inner disk, setting a trapping radius $R_{\rm trap} \sim (\dot{M}/\dot{M}_{\rm Edd}) R_S$ inside which photon diffusion time exceeds inflow time (Jiang et al., 2014, Zhang et al., 2 Jun 2025, Zhang et al., 12 Sep 2025). The disk forms a geometrically thick, radiation-pressure-supported structure, with strong turbulence, MRI-driven angular-momentum transport, and a low-density polar funnel through which radiation escapes (Jiang et al., 2024).

A key result of both analytic and numerical models is that even at high $\dot{M}/\dot{M}_{\rm Edd}$ , the aspect ratio $H/R$ saturates at $0.5$–$1$ rather than inflating further (Lasota et al., 2015). The radiative efficiency in 2D/3D flows is consistently higher than predicted by 1D slim-disk models due to vertical buoyant advection and multidimensional photon leakage, yielding $\eta \sim$ a few percent at $\dot{M} \sim 10$ – $100\,\dot{M}_{\rm Edd}$ and falling to $\eta \lesssim 1\%$ for $\dot{M} \gtrsim 100\,\dot{M}_{\rm Edd}$ (Jiang et al., 2014, Jiang et al., 2017, Zhang et al., 2 Jun 2025, Jiao, 2023).

2. Outflows, Winds, and Angular Dependence

Super-Eddington flows ubiquitously generate strong radiation-pressure-driven winds, with mass outflow rates that can approach or even exceed the net accretion rate onto the compact object (Kitaki et al., 2021, Yoshioka et al., 2024, Jiang et al., 2017, Takeuchi et al., 2013). The outflow geometry typically consists of two components:

A broad-angle, uncollimated wind (opening half-angle $\sim$ 30°–60°) carries a characteristic velocity $v \sim 0.1$ –$0.3c$ and is launched from radii inside the trapping radius but outside $R \sim 0.5\,R_{\rm trap}$ (Kitaki et al., 2021, Yoshioka et al., 2024).
A low-density, optically thin polar funnel ( $\theta \lesssim 10$ –15°) through which radiation and, if present, relativistic jets can escape (Jiang et al., 2014, Jiang et al., 2024, Curd et al., 2022).

Quantitatively, large-scale simulations find that for $\dot{M}_{\rm BH} \sim 100\,L_{\rm Edd}/c^2$ the true outflow rate (mass flux escaping the system) is $\dot{M}_{\rm outflow} / \dot{M}_{\rm BH} \sim 0.1$ –$0.5$ (Kitaki et al., 2021, Jiang et al., 2017). The mechanical energy flux carried by outflows is typically $\sim$ 5%–30% of the observed radiative luminosity, peaking at polar and intermediate angles, as derived from angular flux decompositions in both RHD and GRRMHD simulations (Kitaki et al., 2021, Jiang et al., 2017, Yoshioka et al., 2024).

In the outer regions, radiation-driven instabilities (primarily Rayleigh–Taylor and photon bubble–analogues) induce the formation of clumpy, marginally optically thick outflow structures, with covering factors $C \sim 0.3$ (Takeuchi et al., 2013). These clumps modulate emergent spectra and can explain rapid AGN variability, as well as potentially contributing to the broad-line region (Takeuchi et al., 2013).

3. Multidimensional Energy Transport and Magnetic Effects

Photon trapping, energy advection, and vertical (buoyant) transport are central to the thermodynamics of super-Eddington disks. In 1D treatments, nearly all dissipation in the inner disk is advected inward, drastically reducing emergent flux. However, in higher-dimensional solutions, vertical advection (e.g., through magnetic buoyancy and turbulent motion) enables much of the dissipated energy to be delivered to the surface and radiated before being accreted (Jiao, 2023, Jiang et al., 2014, Jiang et al., 2024). This results in a radiative efficiency $\eta$ larger than predicted by slim-disk theory (by factors 2–5 at fixed $\dot M$ ).

The launching, structure, and efficiency of outflows and jets are sensitive to magnetic topology. The presence of net vertical flux (MAD regime) and high black hole spin can yield powerful Blandford–Znajek jets, with jet efficiencies reaching several percent of $\dot{M}c^2$ , and efficient evacuation of the polar funnel (Zhang et al., 2 Jun 2025, Zhang et al., 12 Sep 2025, Curd et al., 2022). In the SANE regime (lower net flux), powerful radiatively driven outflows dominate, and jets are weaker or absent (Curd et al., 2022, Zhang et al., 2 Jun 2025). Maxwell stress is the primary channel for angular-momentum transport in both regimes, with Reynolds stress becoming significant only in the presence of spiral shocks or low magnetic pressure (Jiang et al., 2017, Zhang et al., 12 Sep 2025).

4. Spectral Properties, Coronae, and Anisotropy

The emergent spectra of super-Eddington accretors differ fundamentally from those of sub-Eddington systems. A signature diagnostic is the development of an optically thick ( $\tau \gtrsim 3$ ), low-temperature ( $k_B T_e \lesssim 10\,{\rm keV}$ ) corona, formed by radiation-pressure–driven winds above the disk and heated by magnetic reconnection (Kawanaka et al., 2020). This corona efficiently upscatters seed disk photons to produce a Comptonized hard X-ray tail with a sharp cutoff at $\sim 10$ –30 keV.

Radiation is strongly beamed along the polar direction, leading to a marked $\theta$ -dependent anisotropy. Face-on observers measure radiative fluxes exceeding $L_{\rm Edd}$ by factors of 2–10, while edge-on luminosity can be suppressed by orders of magnitude (Kitaki et al., 2021, Jiang et al., 2024, Qiao et al., 5 May 2025). Viewing angle also determines the relative contribution of X-ray and optical/UV emission due to reprocessing in the wind or envelope, as shown in simulations of TDEs (Qiao et al., 5 May 2025). X-ray reverberation mapping in the Fe K $\alpha$ region unambiguously distinguishes super-Eddington flows through a combination of short lag timescales, mass-linear lag scaling, and frequency-independent lag-energy spectra, reflecting reflection off the wind funnel instead of the disk surface (Thomsen et al., 2021).

5. Scaling Laws and Unified Parameterization

Axisymmetric and 3D simulations across black hole mass scales ( $M = 10\,M_\odot$ to $10^8\,M_\odot$ ) demonstrate that, in the electron-scattering–dominated regime, the normalized radiative and mechanical luminosities as functions of accretion rate $\dot{m} \equiv \dot M c^2 / L_{\rm Edd}$ are nearly universal:

$L_{\rm rad}/L_{\rm Edd} \approx 0.03 \,\dot{m}$ for $\dot{m} \gtrsim 100$ (Yoshioka et al., 2024).
$L_{\rm mech}/L_{\rm Edd} \approx 0.06\,(\dot{m}/100)^{1.7}$ .
The ratio $L_{\rm mech}/L_{\rm rad} \sim 0.05$ at $\dot{m} \sim 100$ matches observed kinetic–to–bolometric power in systems like NLS1 galaxies (Yoshioka et al., 2024, Kitaki et al., 2021).
The isotropic equivalent mechanical luminosity at polar angles exhibits a broken power-law dependence on $\dot{m}$ with a transition at $\dot{m} \sim 400$ , reflecting disk inflation and wind collimation.

The mass outflow rate, wind velocities, and the relative importance of equatorial winds versus polar outflows are parametrized robustly across all simulated mass scales (Kitaki et al., 2021, Yoshioka et al., 2024, Jiang et al., 2017).

6. Super-Eddington Magnetized Neutron Stars

For accretion onto strongly magnetized neutron stars, the super-Eddington flow is truncated at the magnetospheric radius $R_m$ , where magnetic pressure balances ram and radiation pressure. Analytical and simulation-based models demonstrate that:

The disk remains a slim disk with $h = H/R \sim 0.2$ –0.8, loses mass in a wind, and transfers angular momentum to the NS primarily via matter torque at $R_m$ (Chen et al., 2024, Inoue et al., 2024).
Accretion transitions to narrow columns (dipole geometry) or belts (quadrupole geometry) inside $R_m$ . Radiation and strong fields together allow super-Eddington $\dot{M}$ to penetrate the magnetosphere, set the spin-up torque, and regulate pulsar spin evolution.
At high $\dot{M}$ , radiative forces can drive quasi-periodic instabilities in the magnetospheric curtain, especially when the luminosity from the accretion column locally exceeds $L_{\rm Edd}$ (Mushtukov et al., 2024). The resulting high-frequency QPOs match observed variability in ultraluminous X-ray pulsars.

Emergent luminosities in these flows match $L \sim L_{\rm Edd}[1+\ln(\dot{M}/\dot{M}_{\rm Edd})]$ (Inoue et al., 2024, Chen et al., 2024), and the total energy output and pulse profiles depend sensitively on accretion rate, magnetic geometry, and multipolar field structure.

7. Applications: Tidal Disruption Events, High-Redshift BH Growth, and Observational Probes

Super-Eddington accretion is the physical basis for:

Tidal Disruption Events (TDEs), where fallback rates can reach $\dot{M} \sim 10$ – $100\,\dot{M}_{\rm Edd}$ (Qiao et al., 5 May 2025). Simulations incorporating the expected time-dependent fallback rate show that outflows reprocess X-rays into optical/UV, explaining the observed diversity of TDE light curves and spectral properties. The observed luminosity, temperature, and emission radii are set by both $\dot{M}(t)$ and the observer's inclination.
ULXs and ULXPs, where supercritical accretion onto stellar-mass black holes or neutron stars yields luminosities $L \gg L_{\rm Edd}$ via beaming and efficient wind reprocessing (Kitaki et al., 2021, Kawanaka et al., 2020, Inoue et al., 2024).
Rapid supermassive black hole growth in high- $z$ galaxies, enabled by sustained super-Eddington accretion rates in dense nuclear disks. Three-dimensional simulation-based efficiencies ( $\eta \sim 0.01$ –0.05) and strong outflows enable massive black hole buildup even in the presence of modest feedback (Mayer, 2018).
Radio jets and polarimetric signatures, which depend on both flow magnetization and disk orientation, as explored in global GRRMHD models for next-generation VLBI observations (Curd et al., 2022).

Continued simulation and analytic progress is clarifying the parameter space for super-Eddington engines, their instabilities, observable spectra, feedback on host galaxies, and their impact on the broader landscape of accreting compact objects.

References