An accretion disk is a rotating distribution of gas, often accompanied by dust, in which matter gradually moves toward a central gravitating object. Such disks surround young stars, white dwarfs, neutron stars, and black holes, and can also form around planets and within systems containing two orbiting stars. Their defining feature is not simply a flattened shape: an accretion disk transports angular momentum, allowing some material to move inward while other material, or an outflow, carries angular momentum away.
Accretion converts gravitational binding energy into motion, heat, radiation, and sometimes powerful outflows. Around compact objects, it can release a much larger fraction of the supplied matter’s rest energy than stellar nuclear fusion. Accretion disks therefore power many of the brightest persistent sources in the universe and provide laboratories for plasma physics, magnetic turbulence, and strong gravity. Around young stars, related disks supply the material from which planetary systems emerge.
The appearance and behavior of an accretion disk depend on its mass supply, opacity, magnetic field, and central object. A cool, dusty disk extending over hundreds of astronomical units differs greatly from an X-ray-emitting flow only tens of kilometres from a stellar-mass black hole. Both nevertheless obey the same underlying constraints of mass, energy, and angular-momentum conservation.
Formation, circularization, and radial extent
A disk forms when infalling material has too much angular momentum to fall directly onto the central object. Gas drawn from a collapsing cloud, a companion star, or the interstellar medium generally possesses some transverse motion. As it falls inward, conservation of angular momentum increases its rotational speed. Collisions and shocks dissipate energy associated with intersecting streams, but they do not eliminate the total angular momentum. The gas consequently tends to settle into an approximately rotationally supported configuration.
For material orbiting a point mass , a useful estimate of the circularization radius is
where is the specific angular momentum and is the gravitational constant. This is the radius of a circular Newtonian orbit with the same specific angular momentum. The estimate assumes that the central mass dominates the gravitational field and that additional torques during circularization are unimportant. It explains why relatively modest differences in the angular momentum of the incoming material can produce large differences in disk size.
In a close binary star, a donor can fill its Roche lobe, the region within which matter is gravitationally associated with that star in the rotating binary potential. Gas then passes through the inner Lagrange point toward the companion. Unless the companion’s surface or magnetic field intercepts the stream first, the stream’s angular momentum usually leads to disk formation. Impact between the incoming stream and the outer disk produces a localized bright region, especially conspicuous in some white-dwarf binaries. Capturing a companion’s stellar wind can also produce a disk, although the captured angular momentum is more sensitive to the wind geometry.
Disks around young stars arise during the collapse of rotating molecular-cloud cores. They are continuously reshaped by infall, magnetic torques, companions, and their own gravity. In a tidal disruption event, a star passing close to a massive black hole is torn apart by differential gravity. Bound debris returns toward the black hole, and collisions between returning streams can convert eccentric trajectories into an accretion flow. This circularization need not be immediate or complete.
Neither disk edge is universal. A stellar surface, a magnetosphere, or a relativistic transition to plunging motion can determine the inner boundary. Tidal torques, the angular momentum of supplied gas, or gravitational instability can limit the outer disk. Accretion also does not require every parcel to move inward: an isolated viscously evolving disk commonly spreads outward while most of its mass moves inward, leaving a smaller amount of material at large radii to carry much of the original angular momentum.
Angular-momentum transport and magnetic turbulence
Gas on a circular orbit cannot move steadily inward unless it loses angular momentum. In a nearly Keplerian disk, the angular frequency is
so inner rings rotate faster than outer rings. A stress coupling neighboring rings can transfer angular momentum outward while extracting energy from this differential rotation. Part of the extracted energy becomes heat; another part may be carried away electromagnetically or mechanically. Microscopic viscosity is generally too weak to explain the accretion rates inferred in many astrophysical disks.
A widely used description is the alpha prescription introduced by Nikolai Shakura and Rashid Sunyaev in 1973. It represents the effective kinematic viscosity as
where is a characteristic sound speed, is the vertical pressure scale height, and is a dimensionless measure of transport efficiency. The prescription summarizes unresolved stresses rather than identifying their cause. Values inferred or obtained in models vary with physical conditions; is not a universal material constant. Equivalently, the magnitude of the radial–azimuthal stress is often written as a fraction of an appropriate local pressure.
In sufficiently ionized disks, the magnetorotational instability provides a powerful transport mechanism. A weak magnetic field linking fluid elements at different radii acts somewhat like an elastic connection. The inner element, which rotates faster, transfers angular momentum to the outer one. Losing angular momentum makes the inner element move inward, while the outer element moves outward, amplifying their separation. Steven Balbus and John Hawley established the importance of this instability for accretion disks in 1991, building on earlier work on the stability of magnetized rotating fluids.
The resulting turbulence produces correlated velocity fluctuations and magnetic stresses. In many ionized-disk simulations, magnetic stress supplies the larger contribution. Its strength depends on magnetic-field geometry, the net magnetic flux through the disk, and the ability of turbulence to regenerate magnetic fields. A disk threaded by substantial large-scale vertical flux can behave very differently from one containing only small magnetic loops.
Magnetic turbulence is not the only route to accretion. Spiral density waves and gravitational torques can dominate in sufficiently massive disks. Companion-induced tides can remove angular momentum near an outer edge. Magnetically driven winds can extract it vertically rather than transporting it radially through the disk. In weakly ionized regions, particularly in young stellar systems, finite electrical conductivity and drift between charged particles and neutral gas alter or suppress the magnetorotational instability. Consequently, a single disk may contain zones governed by different transport mechanisms.
Vertical structure and the thin-disk model
The simplest quantitative models assume a disk whose vertical thickness is much smaller than its radius, . Vertical pressure gradients then approximately balance the vertical component of the central gravitational force. If gas pressure provides the main support, the characteristic relation is
Cooling maintains a small sound speed relative to the orbital speed, keeping the disk geometrically thin. Radiation pressure, magnetic pressure, or inefficient cooling can increase its thickness.
A thin disk has strong radial and vertical gradients. Its midplane is usually denser than its surface layers, and the location from which radiation escapes depends on opacity. The surface temperature inferred from emitted radiation is not generally the midplane temperature. In an optically thick disk, photons undergo many interactions before escaping, so the interior can be substantially hotter than the effective radiating surface. Opacity changes caused by ionization, atomic transitions, or dust destruction can strongly modify the vertical structure.
For a steady Newtonian disk with an inward mass flow rate , local radiative cooling, and vanishing stress at an inner radius , the outward energy flux from each face is
This expression follows from mass and angular-momentum conservation together with the work done by internal torques. It is not simply the local change in gravitational potential energy: stresses also transport energy between neighboring radii. The inner-boundary factor reflects the assumed zero torque and changes if the boundary exerts a significant stress.
Defining an effective temperature through blackbody radiation,
gives sufficiently far outside the inner boundary. Integrating over both disk faces yields
under the same Newtonian assumptions. The other half of the gravitational energy released by material descending from large radius remains as orbital kinetic energy at the inner edge. When gas subsequently settles onto a slowly rotating stellar surface, much of that remaining energy can be released in a boundary layer.
The thin-disk approximation separates three characteristic timescales. The orbital or dynamical time is of order ; the thermal adjustment time is approximately ; and the radial evolution time is approximately . Thus
A thin disk can orbit many times before its mass distribution changes substantially. This hierarchy makes simplified models tractable, but it can fail in thick, rapidly evolving, strongly magnetized, or violently disturbed flows.
Radiatively efficient, hot, and supercritical flows
The rate at which a disk loses heat is as important as the rate at which it receives matter. If dissipated energy escapes rapidly, the gas remains relatively cool and the disk can be thin. If cooling is inefficient, internal energy accumulates, pressure support becomes important, and the flow thickens. Accretion states therefore cannot be classified solely by the identity of the central object. The same black hole can support markedly different flows as its mass supply and magnetic configuration change.
A standard luminosity scale is the Eddington luminosity. For fully ionized hydrogen with opacity dominated by electron scattering,
where is the proton mass, is the speed of light, and is the Thomson cross section. This expression balances gravity against the force of an isotropic radiation field on ionized gas. Disk radiation is neither spherical nor necessarily isotropic, so the Eddington value is a reference scale rather than an absolute luminosity ceiling.
At relatively low mass supply, a hot, optically thin, radiatively inefficient accretion flow may develop. Ions and electrons can have different temperatures because collisions exchange their energy slowly. Energy deposited mainly into ions may not reach the radiating electrons before the gas moves inward. Such flows are often geometrically thick and can lose a large fraction of their supplied mass in outflows. For a black hole, energy carried inward can cross the event horizon; for a material surface, energy reaching the surface can ultimately be radiated.
At high mass supply, the gas may instead be optically very thick. Radiation then diffuses outward slowly enough that part of it is carried inward with the accreting matter, a process called photon trapping. The resulting slim disk is an extension of thin-disk theory that includes radial transport of heat and departures from strictly Keplerian motion. Its name does not imply an extremely thin geometry; at sufficiently high supply, the flow can become substantially thick and develop strong winds.
The relation , with radiative efficiency , is useful only when the definition of is explicit. If winds remove mass, the rate supplied at large radius can greatly exceed the rate reaching the central object. Efficiency also depends on whether energy emerges as radiation, mechanical power, or electromagnetic power. Supercritical supply therefore does not require luminosity to rise in direct proportion to the supplied mass, while weak luminosity does not necessarily imply little gas is present.
Disks around young stars, compact stars, and galactic nuclei
Young stellar disks link accretion to the formation of planets. Their inner regions can be hot and substantially gaseous, while cooler outer regions contain dust grains and molecules. A protoplanetary disk can simultaneously feed its star and allow solids to grow, drift, and become incorporated into larger bodies. Gas disks around young stars commonly persist for a few million years, although lifetimes and accretion histories vary considerably. Observed rings and gaps need not all be produced by planets: changes in dust growth, gas pressure, condensation, and magnetic structure can also generate substructure.
In white-dwarf binaries known as cataclysmic variables, a donor star transfers gas to a compact companion. The disk commonly emits strongly in visible and ultraviolet light, while the inner boundary region can produce ultraviolet or X-ray emission. A sufficiently strong stellar magnetic field can truncate the disk and channel gas toward magnetic poles. Some strongly magnetic systems do not form a conventional disk at all. Accretion must also be distinguished from the thermonuclear burning responsible for classical nova eruptions, although the accumulated fuel was supplied by accretion.
In X-ray binaries, the accretor is a neutron star or black hole. Their deeper gravitational potentials heat inner flows to temperatures associated with X-ray emission. Neutron stars possess surfaces and may possess magnetic fields strong enough to disrupt the inner disk. Surface emission, boundary or spreading layers, and magnetically channeled accretion can therefore accompany disk radiation. A black hole lacks a material surface, so radiation must be produced before gas crosses its horizon or in associated plasma outside it.
Accretion onto supermassive black holes powers active galactic nuclei, including quasars. A radiatively efficient disk around a massive black hole is cooler than a comparable disk around a stellar-mass black hole. At fixed Eddington-scaled accretion rate and fixed radius in gravitational units, the characteristic effective temperature scales approximately as . Much of the thermal disk emission therefore appears in the optical and ultraviolet rather than at X-ray energies. Additional hot plasma produces higher-energy emission, while surrounding dust absorbs and re-emits part of the luminosity in the infrared.
The large-scale supply problem differs across these environments. A close binary can deliver gas directly through a stream, whereas a galactic nucleus must bring gas inward across a vast range of radii. Stellar torques, bars, interactions, turbulence, and gravitational instabilities can operate at different stages. The small-scale accretion disk does not by itself explain transport from an entire galaxy to its central black hole. Connecting these spatial scales is essential for understanding how nuclear activity is fueled and how central black holes gain mass.
Radiation, relativistic effects, and observational diagnostics
Most accretion disks are too small or distant to resolve directly, so their properties are inferred through spectroscopy, brightness variations, eclipses, and polarization. An optically thick disk radiates from annuli with different temperatures, producing a broader spectrum than a single-temperature blackbody. Under idealized thin-disk assumptions, an intermediate-frequency portion of the spectrum follows . Real spectra depart from this relation because of atmospheric opacity, irradiation, finite disk boundaries, and contamination by other emitting regions.
A hot, relatively tenuous region often called the corona can scatter lower-energy disk photons to higher energies. In inverse Compton scattering, energetic electrons transfer energy to photons, producing a hard X-ray continuum. Some of these X-rays illuminate the disk and are absorbed, scattered, or re-emitted. The resulting reflection spectrum can contain iron fluorescence near 6.4–6.97 kiloelectronvolts, depending on ionization, together with an enhanced continuum at higher X-ray energies. These features carry information about the illuminating source and the disk surface.
Near a black hole, general relativity alters orbital motion and the propagation of radiation. The innermost stable circular orbit lies at for a nonrotating black hole. For an ideal maximally rotating Kerr black hole, the corresponding limiting radius is for a prograde equatorial orbit and for a retrograde one. These values concern test-particle circular orbits; a real magnetized fluid can maintain stresses and emit radiation inside the nominal stability boundary.
Orbital Doppler shifts, gravitational redshift, relativistic beaming, and gravitational lensing combine to distort the disk’s appearance and spectral lines. The approaching side is generally enhanced relative to the receding side, and radiation from the inner region can acquire a strongly extended red wing. Modeling the thermal continuum and the relativistically broadened reflection spectrum provides two major methods for estimating black-hole spin. Their results depend on assumptions about disk thickness, inclination, atmospheric structure, illumination, and the location of the inner radiating edge.
Time-dependent observations add spatial information. X-ray reverberation measures delays between changes in the illuminating continuum and the disk’s response; optical and ultraviolet reverberation probes larger reprocessing regions. Eclipse mapping reconstructs brightness distributions in favorable binaries, while interferometry can resolve some nearby young stellar disks. The Event Horizon Telescope images published for M87* in 2019 and Sagittarius A* in 2022 resolved horizon-scale emission around black holes. Their bright rings arise from strongly lensed emission in the surrounding accretion flow, not from a directly photographed event horizon or a universally thin disk.
Instabilities, outbursts, and changing accretion states
Accretion disks often vary because heating, cooling, opacity, and transport respond differently to changes in local conditions. A particularly important mechanism involves partial ionization of hydrogen. In a suitable temperature range, increasing temperature changes opacity and the ability of the disk to radiate its dissipated energy. Local thermal-equilibrium solutions can then form a curve with a cool stable branch, a hot stable branch, and an unstable intermediate branch. A disk need not remain at a single steady temperature for a fixed external supply.
The disk-instability model explains many dwarf-nova outbursts in white-dwarf binaries. During quiescence, matter can accumulate in a relatively cool disk because inward transport is slow. Once a critical surface density is reached, a heating front propagates, converting disk regions to a hotter, more rapidly accreting state. The disk brightens as stored material moves inward. A cooling front subsequently returns it toward quiescence. This cycle is distinct from a classical nova, whose principal energy source is unstable nuclear burning on the white dwarf.
Related thermal-viscous processes contribute to outbursts of transient X-ray binaries. Irradiation from the inner accretion flow can keep the outer disk ionized, prolonging an outburst and changing its decay. Some black-hole binaries also undergo transitions between a hard state, dominated by high-energy emission from hot plasma, and a soft state, in which thermal disk radiation is prominent. The transition luminosity can differ between the rising and declining phases, producing hysteresis. Such behavior indicates that instantaneous mass supply alone does not fully determine the accretion geometry.
Rapid variability probes much smaller radii. Broadband fluctuations may originate at many disk locations and propagate inward, modulating the emission of faster inner regions. Quasi-periodic oscillations appear as peaks rather than perfectly sharp lines in variability spectra. Proposed mechanisms include orbital and epicyclic motions, disk oscillations, and precession of a tilted inner flow. No single interpretation applies securely to every observed class. Characteristic frequencies tend to be lower for more massive central objects when the relevant radii are comparable in gravitational units.
In massive, cool disks, self-gravity introduces a different instability. The Toomre parameter for a thin gaseous disk is
where is the epicyclic frequency and is surface density. Values near or below unity indicate susceptibility to local axisymmetric gravitational instability under the model’s assumptions. Whether a disk develops sustained spiral transport or fragments into bound objects also depends on cooling and global dynamics. This distinction matters both for young stellar disks and for the outer regions of disks around supermassive black holes.
Winds, jets, warps, and environmental effects
An accretion disk exchanges matter, momentum, and energy with its surroundings rather than functioning as a closed conveyor belt. Winds may be launched thermally when irradiated gas becomes hot enough to escape, radiatively when photons transfer sufficient momentum, or magnetically when field stresses accelerate matter away from the disk. These mechanisms can overlap. Absorption lines shifted toward shorter wavelengths reveal outflowing gas in many accreting systems, although converting such spectra into a mass-loss rate requires knowledge of the wind’s density, geometry, ionization, and location.
Magnetically driven winds can remove angular momentum directly from the disk. In an idealized magnetocentrifugal picture, matter moves along an ordered, inclined magnetic field line, drawing energy from disk rotation. As the outflow is accelerated and the field twists, magnetic stresses transmit angular momentum to larger distances. Such a wind can permit accretion even when radial turbulent stresses are relatively weak. Its effectiveness depends on magnetic flux, field geometry, and how well charged and neutral particles are coupled.
Highly collimated astrophysical jets are associated with some young stars, X-ray binaries, and active galactic nuclei. Their emission often includes synchrotron radiation from relativistic electrons in magnetic fields. Jets and broader winds are physically distinct, even when both are present. In black-hole systems, a jet can draw energy from the rotating accretion flow and, under suitable conditions, from black-hole rotation through magnetic fields threading the horizon. The latter possibility is described by the Blandford–Znajek mechanism, proposed in 1977.
A sufficiently large concentration of magnetic flux near a black hole can produce a magnetically arrested disk. Magnetic pressure then intermittently obstructs inflow, while gas penetrates through a strongly time-dependent flow. Such states can generate powerful jets. Efficiencies expressed relative to the rest-mass energy entering the black hole can exceed unity when the outflow additionally extracts black-hole rotational energy. This does not violate energy conservation: the black hole’s rotational energy provides the extra contribution.
Disks can also become warped, with orbital planes changing as a function of radius. A spinning black hole causes frame-dragging precession of tilted orbits, while internal stresses communicate the resulting warp. Companion torques and radiation can produce additional distortions. Outflows and jets carry the consequences of accretion far beyond the disk: they excavate cavities, shock surrounding gas, and redistribute energy. In galaxies and clusters, these processes participate in feedback between central accretion activity and the gas that supplies future star formation and black-hole growth.
Development of the theory and numerical modeling
Modern disk theory developed from several initially separate problems: mass transfer in binary stars, the energetics of quasars, and the formation of stars and planetary systems. In 1964, Edwin Salpeter and Yakov Zeldovich independently discussed accretion onto massive compact objects as a source of quasar energy. Donald Lynden-Bell’s 1969 work connected quasars and galactic nuclei with accretion onto massive black holes. These arguments made gravitational energy release central to the interpretation of exceptionally luminous astronomical sources.
The 1973 Shakura–Sunyaev model provided a practical description of thin disks by relating stress to local pressure. In the same year, Igor Novikov and Kip Thorne formulated a relativistic thin-disk model around black holes. These theories established relations among mass supply, radial temperature profiles, radiative efficiency, and emitted spectra. Their enduring usefulness comes from reducing a complex three-dimensional flow to equations that can be solved with a manageable set of assumptions.
The physical origin of the effective stress remained a major difficulty. The 1991 work of Balbus and Hawley identified the magnetorotational instability as a natural source of transport in ionized disks. Numerical studies then followed its nonlinear evolution into turbulence. Local “shearing-box” calculations isolate a small co-rotating patch with differential rotation, allowing detailed investigation of magnetic amplification and dissipation. Their controlled geometry is useful, but it cannot directly represent global features such as radial magnetic-flux accumulation, the inner boundary, or a large-scale wind.
Global simulations now solve magnetohydrodynamics, sometimes coupled to radiation transport and general relativity. General relativistic magnetohydrodynamic calculations describe plasma moving in strongly curved spacetime, while radiation-coupled models track energy and momentum exchange between gas and photons. Synthetic spectra and images are produced by transporting radiation through simulated flows. This approach allows direct comparison with observations, including polarization and horizon-scale structure, but the quality of a comparison depends on the assumptions used to connect fluid properties to emitting particles.
Several important processes remain below the scales accessible to global calculations. Weakly collisional plasma can distribute dissipated energy differently between ions and electrons, strongly affecting predicted luminosity. Magnetic reconnection, particle acceleration, nonideal magnetic effects, dust evolution, and detailed atomic opacity may require additional physical prescriptions. Numerical resolution and imposed boundary conditions can also influence turbulence and magnetic-flux transport. Current research therefore tests simulations against several observables at once—spectra, timing, polarization, resolved structure, and outflow properties—rather than treating agreement with a single measurement as a complete description of the disk.