Accretion
Radiation-Pressure Disk Instability: Why Bright Accretion Disks Flicker
Every 50 seconds, the black-hole binary GRS 1915+105 lets out a sharp, double-peaked X-ray pulse — a repeating "heartbeat" that astronomers have watched for decades. That rhythm is the signature of a disk of gas spiralling toward the black hole that literally cannot sit still: at high luminosity the inner disk builds up, ignites, blows itself apart in radiation, drains, and refills, over and over.
The radiation-pressure disk instability is the physics behind this flicker. When an accretion disk gets bright enough that the pressure of its own trapped photons exceeds ordinary gas pressure, the standard thin-disk equilibrium becomes both thermally and viscously unstable. Instead of steadily glowing, the inner disk oscillates on a thermal-viscous limit cycle, producing large-amplitude luminosity variations that recur on timescales of seconds to years depending on the mass of the central object.
- TypeThermal + viscous instability of accretion disks
- RegimeRadiation-pressure-dominated inner disk, L ≳ few × 0.1 L_Edd
- PredictedShakura & Sunyaev 1976; Lightman & Eardley 1974
- Onset (10 M_sun BH)L ≳ 3 × 10^37 erg/s (~a few % of L_Edd)
- Key relationP_rad ∝ T^4, viscous stress ∝ P_total → heating steeper than cooling
- Observed inGRS 1915+105, IGR J17091-3624; candidate driver in AGN/changing-look quasars
Interactive visualization
Press play, or step through manually. The visualization is yours to drive — try it before reading on.
Watch the 60-second explainer
A condensed visual walkthrough — narrated, captioned, under a minute.
What the instability is: pressure in a thin accretion disk
A geometrically thin, optically thick accretion disk — the Shakura-Sunyaev (1973) α-disk — supports itself vertically against gravity through pressure. That pressure has two contributions: ordinary gas pressure, P_gas = ρ k T / (μ m_H), and radiation pressure from the trapped thermal photons, P_rad = (1/3) a T^4, where a is the radiation constant.
These scale very differently with temperature. Gas pressure rises only linearly in T, but radiation pressure rises as T^4. So as you push more matter through the disk and the inner regions heat up, radiation pressure eventually overtakes gas pressure. Once P_rad ≫ P_gas, the disk enters the radiation-pressure-dominated regime — and that is precisely where the standard equilibrium becomes unstable.
- Thermal instability (Shakura & Sunyaev 1976): a small temperature bump heats faster than it can cool, running away.
- Viscous instability (Lightman & Eardley 1974): the mass-flow rate decreases with surface density, so the disk clumps into rings instead of smoothing out.
The mechanism: why heating outruns cooling
In the α-prescription, the viscous stress that heats the disk scales with the total pressure: stress ≈ α P_total. When radiation dominates, P_total ≈ P_rad ∝ T^4, so the heating rate per unit area, Q+, is an extremely steep function of temperature.
The cooling rate, Q−, is set by radiative diffusion out through the disk surface. In an optically thick, radiation-supported layer the vertical flux depends only weakly on the midplane temperature. The result is the classic imbalance:
- Perturb the disk hotter → Q+ (∝ T^4-ish) shoots up faster than Q− → net heating → runaway. Thermally unstable.
- The steady-state relation between accretion rate and surface density Σ has a negative slope (dṀ/dΣ < 0) on the radiation branch, so the viscous diffusion coefficient is effectively negative → matter piles into rings. Viscously unstable.
Plotted as accretion rate versus Σ at fixed radius, the equilibria trace an S-curve. The middle, negatively-sloped branch is unstable; the disk cannot rest there and instead jumps between the hot upper branch and a cooler lower branch — a limit cycle.
Characteristic numbers and a worked example
For a black hole of mass M, the Eddington luminosity — where radiation pressure on electrons balances gravity — is L_Edd ≈ 1.26 × 10^38 (M/M_sun) erg/s. For a 10 M_sun black hole that is ~1.3 × 10^39 erg/s.
The radiation-pressure region appears well below Eddington. For a 10 M_sun black hole the innermost disk becomes radiation-dominated and unstable once the luminosity exceeds roughly 3 × 10^37 erg/s, i.e. only a few percent of L_Edd. So many bright X-ray binaries sit squarely in the unstable regime.
- Instability zone radius: only the inner few to tens of gravitational radii (R_g = GM/c^2 ≈ 15 km for 10 M_sun).
- Limit-cycle period scales with the local viscous/thermal time. In GRS 1915+105 the ρ 'heartbeat' recurs every ~10–100 s (often quoted near 50 s).
- AGN scaling: timescales grow roughly with M, so a 10^7–10^8 M_sun supermassive black hole would flicker on years-to-millennia — a proposed driver of intermittent quasar activity.
How it's observed: the X-ray 'heartbeat'
The cleanest signature is GRS 1915+105, a ~12 M_sun black-hole binary discovered in 1992. Its 'ρ' variability class shows sharp, double-peaked X-ray flares recurring roughly every 50 seconds. The spectral evolution across each cycle — the inner disk brightening, expanding, cooling, and collapsing — matches the predicted march up and down the S-curve, tracking a local Eddington limit in the inner disk.
Its only known 'twin' is IGR J17091-3624, whose heartbeats repeat on even shorter, ~5–70 s timescales (about ten times faster, consistent with a lower mass and/or higher accretion rate). Both objects display a whole zoo of related quasi-periodic states.
The theoretical backbone was built by numerical work: Honma, Matsumoto & Kato (1991) first computed the disk's time evolution through the instability, and Szuszkiewicz & Miller (1998) followed several consecutive outbursts, confirming genuine limit-cycle behavior rather than a one-off transient.
How it differs from its close cousins
The radiation-pressure instability is often confused with the more familiar hydrogen-ionization (thermal-viscous) instability that powers dwarf-nova and soft X-ray transient outbursts. Both use an S-curve and a limit cycle, but the physics is different:
- Radiation-pressure instability is driven by radiation pressure exceeding gas pressure in the hot inner disk, near the Eddington limit; recurrence is seconds-to-minutes for stellar-mass black holes.
- Hydrogen-ionization instability is driven by the sharp opacity change when hydrogen partially ionizes near ~6,500 K, at intermediate radii and lower accretion rates; recurrence is days-to-weeks and drives full 100–1000× outbursts.
It is also distinct from the magneto-rotational instability (MRI), which is not a limit-cycle at all but the small-scale turbulence that provides the disk's viscosity in the first place. Crucially, whether the α-stress scales with total pressure (unstable) or only gas pressure (stable) is exactly what MRI simulations are trying to pin down.
Significance, open questions, and famous cases
The instability matters because it connects a clean piece of textbook disk theory to real, observable variability — and because it exposes a tension. Standard theory predicts that every luminous accretion disk should flicker violently, yet many high-luminosity X-ray binaries and AGN look remarkably steady. Why?
- The stabilization problem: if the effective viscous stress scales with gas pressure (or magnetic pressure) rather than total pressure, the runaway is suppressed. MRI simulations give mixed results, so the correct closure is still debated.
- The heartbeat systems GRS 1915+105 and IGR J17091-3624 are the strongest evidence that the instability is real and operates as a limit cycle — they are the field's benchmark cases.
- Scaling up: radiation-pressure limit cycles are invoked to explain intermittent radio galaxies, and increasingly the dramatic changing-look AGN and quasar variability, where a supermassive black hole's disk brightens or dims by large factors over years.
Resolving why some disks flicker and others don't remains one of the open problems in accretion physics.
| Property | Radiation-pressure instability | Hydrogen-ionization instability |
|---|---|---|
| Trigger | Radiation pressure > gas pressure in hot inner disk | Partial ionization of hydrogen near T ≈ 6,500 K |
| Unstable quantity | Both thermal (Q+ vs Q-) and viscous (negative diffusion) | Thermal-viscous S-curve, opacity-driven |
| Location in disk | Innermost radii, only at high L (near Eddington) | Intermediate radii, at low-to-moderate accretion rate |
| Recurrence time (stellar BH) | Seconds to minutes (e.g. ~50 s in GRS 1915+105) | Days to weeks (dwarf-nova / X-ray outbursts) |
| Amplitude | Factor of a few in L, quasi-periodic 'heartbeats' | Full outburst: factor 100-1000 in optical/X-ray |
| Best-known systems | GRS 1915+105, IGR J17091-3624 | SS Cygni, dwarf novae, soft X-ray transients |
Frequently asked questions
What causes the radiation-pressure disk instability?
It arises when radiation pressure (∝ T^4) exceeds gas pressure (∝ ρT) in the hot inner region of an accretion disk. Because the viscous heating rate then scales with the total pressure, heating rises far more steeply with temperature than cooling does. A small temperature increase runs away, making the disk thermally unstable, while the mass-flow rate falling with surface density makes it viscously unstable too.
Who discovered the radiation-pressure disk instability?
The viscous version was identified by Alan Lightman and Douglas Eardley in 1974, and the thermal version by Nikolai Shakura and Rashid Sunyaev in 1976, building on their 1973 α-disk model. The nonlinear limit-cycle behavior was later computed numerically by Honma, Matsumoto & Kato (1991) and by Szuszkiewicz & Miller (1998).
What is the 'heartbeat' state of GRS 1915+105?
It is the 'ρ' variability class of the black-hole binary GRS 1915+105, in which X-rays show sharp, double-peaked flares recurring roughly every 50 seconds. The pattern matches the predicted thermal-viscous limit cycle of the radiation-pressure instability, with the inner disk repeatedly building up, brightening past a local Eddington limit, collapsing, and refilling.
At what luminosity does the instability turn on?
The inner disk becomes radiation-pressure-dominated and unstable at a luminosity well below the Eddington limit. For a 10 M_sun black hole that threshold is around 3 × 10^37 erg/s — only a few percent of its ~1.3 × 10^39 erg/s Eddington luminosity — so many bright X-ray binaries sit in the unstable regime.
How is it different from the dwarf-nova instability?
Both are thermal-viscous limit cycles described by an S-curve, but the triggers differ. The dwarf-nova (hydrogen-ionization) instability is driven by the opacity jump when hydrogen partially ionizes near 6,500 K, at intermediate radii, producing outbursts every days-to-weeks. The radiation-pressure instability is driven by radiation pressure near the Eddington limit in the innermost disk, producing much faster seconds-to-minutes flickering.
Why don't all bright accretion disks flicker if the theory predicts instability?
This is an open problem. Standard α-disk theory predicts that any radiation-dominated disk should be unstable, yet many luminous disks appear steady. The likely resolution is that the real viscous stress does not scale with the total pressure — if it tracks gas or magnetic pressure instead, the runaway is suppressed. MRI turbulence simulations are still working out which scaling is correct.