Disk Instabilities and State Transitions

Accretion disks change state because several clocks and feedback loops coexist. Local thermal or viscous instability can amplify a perturbation; hydrogen ionization can produce a global storage-and-release cycle; evaporation, condensation, irradiation, and magnetic-flux transport can change the inner geometry. No single mechanism presently explains every black-hole transient, its hysteresis, spectrum, and timing. The durable result is therefore a discriminating framework: derive each mechanism’s stability criterion and timescale, then require it to predict more than a luminosity threshold.

Required background. Fluid and Kinetic Theory: continuum kinematics and balance laws supplies local perturbations, transport, and thermal balance.

Helpful background. Black-Hole Geometry supplies inner orbital scales; Relativistic Matter and Compact Objects supplies opacity and radiation pressure.

The hierarchy of disk clocks

At radius $R$, define

$$t_{\rm dyn}=\Omega^{-1}, \qquad t_{\rm th}\sim\frac{1}{\alpha\Omega}, \qquad t_{\rm vis}\sim \left(\frac{R}{H}\right)^2t_{\rm th}.$$

The dynamical time restores orbital and vertical balance. The thermal time changes the internal energy at approximately fixed surface density $\Sigma$. The viscous time redistributes mass and angular momentum. For $\alpha<1$ and $H/R<1$, their ordering is

$$t_{\rm dyn}<t_{\rm th}<t_{\rm vis}.$$

A proposed mechanism must land on the observed transition clock after using the appropriate radius, thickness, and stress. Matching a duration by freely choosing $R$ is not a prediction.

For a local temperature perturbation at fixed $\Sigma$, thermal stability requires net cooling to increase with temperature:

$$\left. \frac{\partial(Q^- - Q^+)}{\partial T} \right|_\Sigma>0.$$

If the derivative is negative, a hotter patch gains heat faster than it loses it and runs away. Viscous stability instead asks how the equilibrium mass flux changes with $\Sigma$. These criteria are logically distinct: a branch can fail one without failing the other.

Abramowicz and Fragile 2013, §§ 8.1–8.3 reviews the thermal, viscous, and radiation-pressure criteria and emphasizes their dependence on the stress and cooling closure. The textbook radiation-pressure instability follows when heating scales with total pressure under a particular alpha prescription. Magnetic support, stress delay, vertical transport, advection, and winds change the derivatives. An unstable analytic annulus is therefore a hypothesis to test in a global radiation-MHD flow, not an automatic limit cycle.

Ionization cycles and hysteresis

At temperatures where hydrogen changes ionization state, opacity and cooling vary steeply. The local thermal equilibria form an S-shaped curve in $(\Sigma,\dot M)$ or $(\Sigma,T_{\rm eff})$: a cool neutral branch and a hot ionized branch can be stable, while the middle branch is unstable. A disk fed slowly on the cool branch stores mass until a heating front moves through it; the hot disk then drains until a cooling front returns it to quiescence.

This disk-instability model naturally separates the long recurrence/decay clock from the short local heating-front clock. Irradiation from the inner flow changes the hot-zone radius and prolongs decay. The donor supply, outer radius, evaporation, and truncation determine whether the inner disk is present at ignition. Done, Gierliński, and Kubota 2007, §§ 2.2–2.3 and 6.1 connects this cycle to black-hole binary outbursts.

Spectral-state hysteresis adds a stronger constraint. On the rise, a hard flow may persist to a higher luminosity than the luminosity at which a soft disk returns to the hard state on the decline. A rule depending only on instantaneous $\dot M$ cannot create two transition thresholds. At least one memory variable is needed: disk surface-density history, irradiation, coronal evaporation/condensation, net magnetic flux, or some coupled combination.

The measured transition should be defined in detector-independent physical terms wherever possible. A hardness ratio is bandpass and absorption dependent; timing noise and radio behavior provide additional state information. Rise and decline must be compared at matched luminosity rather than pooled into one track.

Evidence and competing mechanisms

The evidence cutoff for this frontier comparison is 2026-08-18. The unresolved question is not whether disks vary, but which feedback controls a particular transition and whether one formulation can reproduce the joint thermal, spectral, timing, polarization, and hysteresis data. The relevant parameter regime includes mass, radius, luminosity, thickness, opacity, and magnetic flux; each boundary condition for mass and flux supply is part of the claim.

Mechanism or formulation	Evidence and predicted result	Timescale or discriminant	Limitation and present status
Hydrogen-ionization disk instability with irradiation	Produces recurrent storage, heating fronts, and cooling fronts; explains much of the outburst envelope	Outer-disk viscous decay, front propagation, recurrence versus disk size and donor supply	Does not by itself select the inner hard/soft geometry or unique rise/decline thresholds
Total-pressure radiation instability	A local alpha closure can make radiation-dominated annuli thermally unstable	Thermal-time luminosity cycles and a radiation-pressure radial boundary	Stress closure, magnetic support, advection, and global radiation-MHD calculations can weaken or reshape the instability
Truncation with evaporation and condensation	Competing hot and cool solutions move the inner disk edge and can retain history	Spectral radius, reflection, lags, and transition luminosity on both branches	Microphysical heat conduction and coronal geometry are uncertain; several parameterizations fit similar tracks
Magnetic-flux accumulation or release	Changes stress, coronal power, jet activity, and possibly the hard-to-soft threshold	Polarization, jet power, inflow-time flux transport, and held-out outburst prediction	Net-flux supply and turbulent diffusion are poorly observed; cycles remain simulation- and boundary-dependent

These positions are credible competitors rather than mutually exclusive labels. The ionization cycle can set the global outburst while evaporation or magnetic flux controls the inner geometry. Conversely, a local radiation-pressure cycle may operate in a limited radial and luminosity regime without explaining a months-long binary outburst. Done, Gierliński, and Kubota 2007, §§ 4.1–4.3 and 9 reviews the coupled spectral and timing evidence; Abramowicz and Fragile 2013, § 12.3 summarizes the gap between local stability claims and global disk behavior.

A decisive comparison should calibrate each model on one subset of an outburst and predict a held-out transition. Use the same mass, distance, inclination, absorption, and instrument response; propagate cadence and nondetections; compare predicted front times, luminosity, spectral radius, variability, and jet or polarization change. A mechanism that explains only the fitted hardness track but misses the timescale or reverse transition has failed a joint test.

The interpretation could change after a resolved, multi-band state-transition campaign measures disk radius, lags, polarization, and jet evolution together; after a converged global radiation-MHD calculation produces a repeatable cycle across stress and boundary prescriptions; or after a held-out outburst falsifies one of the competing hysteresis mechanisms.

Boundaries and failure modes

The stability derivative is meaningful only after declaring what remains fixed and how heating responds. Holding $\Sigma$ fixed for a thermal perturbation is not the same as evolving a viscous annulus. Numerical cycles must survive grid, timestep, opacity-table, radiation closure, vertical-domain, and boundary refinement. A cycle driven by density floors or an imposed outer supply variation is not an intrinsic instability.

Detailed spectra and instrument likelihoods are outside this page’s boundary. Radiation, Spectra, and Variability develops state classification and spectral–timing systematics. The analysis here explains the physical feedbacks, stability tests, and testable timescale connections.

Common pitfalls

Equating variability with instability. A stable disk can be driven by variable supply, while an unstable annulus need not create an observable global cycle.

Using one Eddington ratio for both transition directions. Hysteresis is information about state history, not scatter to be averaged away.

Treating the alpha prescription as a measured constitutive law. It is a reduced stress closure whose pressure dependence must be tested against MHD dynamics.

Build a two-threshold cycle

For a $10M_\odot$ black hole at $R=100r_g$, take $\alpha=0.1$ and $H/R=0.05$. Compute $t_{\rm dyn}$, $t_{\rm th}$, and $t_{\rm vis}$. For a toy annulus with $Q^+\propto T^8$ and $Q^-\propto T^4$ at fixed $\Sigma$, determine its thermal stability. Finally, suppose the hard-to-soft transition occurs at $0.20L_{\rm Edd}$ on the rise and the soft-to-hard transition at $0.03L_{\rm Edd}$ on decline. Explain why one instantaneous threshold cannot reproduce the cycle.

Solution

For a Keplerian orbit,

$$t_{\rm dyn}=\Omega^{-1} =\left(\frac{R}{r_g}\right)^{3/2}\frac{GM}{c^3}.$$

Since $GM/c^3=4.9255\times10^{-5}\,{\rm s}$ for $10M_\odot$ and $100^{3/2}=1000$,

$$t_{\rm dyn}\simeq0.0493\,{\rm s},\qquad t_{\rm th}\simeq\frac{t_{\rm dyn}}{\alpha}=0.493\,{\rm s},$$ $$t_{\rm vis}\simeq \left(\frac{R}{H}\right)^2t_{\rm th} =400(0.493\,{\rm s})\simeq197\,{\rm s}.$$

Near equilibrium, a positive temperature perturbation changes heating as $8\,\delta T/T$ but cooling as only $4\,\delta T/T$. Thus $\partial(Q^- - Q^+)/\partial T<0$ and the toy branch is thermally unstable.

The two transition luminosities differ by a factor $0.20/0.03\simeq6.7$. A model whose state is a single-valued function of instantaneous luminosity must switch at the same value in both directions. Reproducing this loop requires a second, history-dependent variable or separate stable branches.

References

• Abramowicz, Marek A., and P. Chris Fragile. “Foundations of Black Hole Accretion Disk Theory.” Living Reviews in Relativity 16 (2013): 1. DOI.
• Done, Chris, Marek Gierliński, and Aya Kubota. “Modelling the Behaviour of Accretion Flows in X-Ray Binaries: Everything You Always Wanted to Know about Accretion but Were Afraid to Ask.” Astronomy and Astrophysics Review 15 (2007): 1–66. DOI.