Chaotic Cold Accretion (CCA)

• CCA is a SMBH fueling mode where hot, stratified atmospheres cool nonlinearly into multiphase cold clouds and filaments triggered by turbulence.
• It enables rapid, chaotic accretion by broadening the angular-momentum distribution and enhancing feeding rates up to 100 times the traditional Bondi rate.
• CCA links a top-down condensation cascade—from soft X-ray plasma to molecular clouds—with self-regulated AGN feedback that maintains thermal balance.

Searching arXiv for recent and foundational papers on Chaotic Cold Accretion to ground the article in cited literature. Chaotic Cold Accretion (CCA) is a multiphase supermassive black hole (SMBH) fueling mode in which a hot, stratified atmosphere, maintained near global thermal balance but stirred by turbulence, undergoes nonlinear condensation into cold clouds and filaments that rain toward the nucleus. In the inner region, recurrent cloud-cloud, filament-filament, and cloud-torus interactions broaden, mix, and partially cancel angular momentum, so the SMBH accretion rate can approach the cooling rate and exceed hot Bondi-like inflow by up to two orders of magnitude (Gaspari et al., 2013, Gaspari, 2015). In a broader historical sense, CCA also overlaps with earlier analytic arguments that SMBHs are fed predominantly by cold clouds or filaments with low and rapidly varying angular momentum rather than by coherent kiloparsec-scale discs, although those precursor treatments did not yet contain the later precipitation-based thermodynamic framework (Nayakshin et al., 2012).

1. Conceptual development and definition

The modern CCA framework emerged from the convergence of three lines of argument. First, analytic work on SMBH feeding geometry showed that large planar feeders such as galaxy-scale discs or bars couple inefficiently to AGN feedback, so feedback-regulated SMBHs fed in that way should be overmassive at fixed velocity dispersion; the observed tendency for pseudobulges and barred systems to host smaller, not larger, SMBHs therefore favors stochastic feeding by cold clouds or filaments with nearly random angular-momentum directions (Nayakshin et al., 2012). Second, large-dynamical-range hydrodynamic simulations established that when a hot halo is allowed to cool while being heated and stirred, the flow does not remain a smooth hot Bondi inflow but fragments into a multiphase medium whose cold phase dominates the effective feeding (Gaspari et al., 2013). Third, the self-regulated feedback interpretation connected this feeding mode to the broader AGN baryon cycle in galaxies, groups, and clusters: condensation boosts accretion, accretion powers mechanical outflows, and those outflows restore global thermal balance without erasing the cool core (Gaspari, 2015).

CCA is therefore distinct from both classical hot accretion and coherent cold-disc accretion. It differs from Bondi accretion because the atmosphere is stratified, turbulent, cooling, heated, and multiphase rather than adiabatic, spherical, and steady (Gaspari et al., 2013). It differs from a thin cold disc because the condensed gas is not primarily organized into a long-lived, coherent, rotationally supported structure; instead, turbulence broadens the angular-momentum distribution, condensation produces both prograde and retrograde components, and the cold phase remains collisional and time-variable (Gaspari et al., 2014).

Later work extended the framework from a primarily two-phase hot-plus-cold picture to a top-down multiphase cascade in which soft X-ray plasma condenses into warm $10^4$ K filaments, then into neutral structures, and finally into molecular clouds below $50$ K, with the phases remaining cospatial and kinematically coupled (Gaspari et al., 2016). Recent “BlackHoleWeather” studies further recast CCA as a multiscale weather process in which turbulence controls whether the halo realizes a compact rainy state or an extended stormy state, and in which the meso-scale between halo rain and sub-pc feeding becomes the central organizing layer (Barbani et al., 26 May 2026, Barbani et al., 26 May 2026).

2. Thermodynamic pathway and condensation criteria

CCA begins in a hot atmosphere with local cooling, gravitational stratification, and non-negligible turbulence. The standard cooling time is written as

$$t_{\rm cool}\equiv \frac{3}{2}\frac{n k_{\rm B}T}{n_{\rm e} n_{\rm i}\Lambda},$$

while the free-fall time is

$t_{\rm ff}\equiv \left(\frac{2r}{g}\right)^{1/2}.$

The classical precipitation condition used throughout the CCA literature is

$\frac{t_{\rm cool}}{t_{\rm ff}} \lesssim 10,$

which identifies the radial range in which nonlinear thermal instability and extended multiphase condensation can develop (Gaspari et al., 2013, Gaspari, 2015, Gaspari et al., 2014).

A later refinement replaces the purely gravitational comparison with a turbulence-aware local criterion. In the cluster-core context, the condensation radius is defined by

$$\mathcal{C}\equiv \frac{t_{\rm cool}}{t_{\rm eddy}}=1,$$

with

$$t_{\rm eddy}=2\pi \frac{r^{2/3}L^{1/3}}{\sigma_{v,L}},$$

so that actual condensation occurs where cooling can outrun turbulent mixing on the local eddy timescale (Wang et al., 2023). The same logic appears in the top-down condensation model, which argues that the relevant nonlinear condition is whether a density peak cools faster than local turbulent stirring disperses it, rather than whether a linear mode satisfies a universal $t_{\rm cool}/t_{\rm ff}$ threshold (Gaspari et al., 2016).

The condensed gas follows a top-down thermodynamic cascade. In massive galaxies and group or cluster cores, hot X-ray plasma cools first through the soft X-ray and UV regime, forming warm ionized filaments at $\sim 10^4$ K, then neutral structures, and finally molecular clouds. In this picture, the soft X-ray phase acts as the immediate gateway of condensation, while the warm, cold, and molecular phases occupy progressively denser parts of the same multiphase complex (Gaspari et al., 2016, Barbani et al., 26 May 2026). This distinguishes CCA from monolithic cooling-flow collapse: global heating prevents the entire core from collapsing, but local overdensities still condense nonlinearly.

3. Angular-momentum transport and accretion regimes

The dynamical core of CCA is angular-momentum redistribution in a clumpy, collisional cold phase. In non-rotating or weakly rotating atmospheres, turbulence broadens the angular-momentum distribution of the hot gas, so the condensed cold phase inherits both prograde and retrograde motion. As the clouds and filaments fall inward, especially within the inner $\sim 1$ kpc and most strongly in the inner few hundred parsecs, recurrent inelastic collisions reduce $50$0, disrupt coherent circularization, and convert what would otherwise be rotationally supported inflow into rapid radial feeding (Gaspari et al., 2014, Gaspari et al., 2016).

A useful organizing parameter in rotating atmospheres is the turbulent Taylor number

$50$1

When $50$2, turbulence dominates coherent rotation and CCA can prevail; when $50$3, rotation dominates, retrograde condensation becomes rare, collisions are less effective, and the system tends toward a cold-disc mode (Gaspari et al., 2014, Gaspari, 2015).

Before summarizing the regimes, two quantitative results are central. In the hot mode, rotation or turbulence suppresses accretion relative to Bondi; in the cold chaotic mode, the accretion rate can reach $50$4 (Gaspari et al., 2013, Gaspari et al., 2014). In the top-down condensation model, the inner cloud ensemble can be approximated with a collisional mean free path $50$5 pc and an effective collisional viscosity $50$6, yielding a quasi-spherical viscous estimate

$50$7

which reproduces the simulated BH feeding rate (Gaspari et al., 2016).

Regime	Conditions	Accretion behavior
Hot mode	Adiabatic or weakly cooling; pressure-supported; turbulence or rotation present	$50$8 suppressed below or to $50$9; with rotation, $$t_{\rm cool}\equiv \frac{3}{2}\frac{n k_{\rm B}T}{n_{\rm e} n_{\rm i}\Lambda},$$0
Cold thin-disc mode	Cooling-dominated; coherent rotation; $$t_{\rm cool}\equiv \frac{3}{2}\frac{n k_{\rm B}T}{n_{\rm e} n_{\rm i}\Lambda},$$1	$$t_{\rm cool}\equiv \frac{3}{2}\frac{n k_{\rm B}T}{n_{\rm e} n_{\rm i}\Lambda},$$2; gas circularizes into a disc
CCA	Cooling + heating + turbulence; $$t_{\rm cool}\equiv \frac{3}{2}\frac{n k_{\rm B}T}{n_{\rm e} n_{\rm i}\Lambda},$$3	$$t_{\rm cool}\equiv \frac{3}{2}\frac{n k_{\rm B}T}{n_{\rm e} n_{\rm i}\Lambda},$$4

This regime structure is central to the interpretation of CCA as a distinct feeding mode rather than a generic synonym for all cold accretion.

4. Observational diagnostics and empirical support

CCA is constrained by both direct observations of nuclear cold gas and ensemble thermodynamic scalings. A particularly direct case is the Abell 2597 brightest cluster galaxy, where ALMA detected three narrow redshifted CO(2–1) absorption features at $$t_{\rm cool}\equiv \frac{3}{2}\frac{n k_{\rm B}T}{n_{\rm e} n_{\rm i}\Lambda},$$5, $$t_{\rm cool}\equiv \frac{3}{2}\frac{n k_{\rm B}T}{n_{\rm e} n_{\rm i}\Lambda},$$6, and $$t_{\rm cool}\equiv \frac{3}{2}\frac{n k_{\rm B}T}{n_{\rm e} n_{\rm i}\Lambda},$$7 against the AGN continuum. Their linewidths are $$t_{\rm cool}\equiv \frac{3}{2}\frac{n k_{\rm B}T}{n_{\rm e} n_{\rm i}\Lambda},$$8, implying compact clouds likely a few to tens of parsecs across, with masses of order $$t_{\rm cool}\equiv \frac{3}{2}\frac{n k_{\rm B}T}{n_{\rm e} n_{\rm i}\Lambda},$$9 and inferred columns $t_{\rm ff}\equiv \left(\frac{2r}{g}\right)^{1/2}.$0. Together with VLBA H I absorption and geometric arguments, these data place the clouds within the inner $t_{\rm ff}\equiv \left(\frac{2r}{g}\right)^{1/2}.$1 pc to few hundred pc and show that cold clumpy gas is moving inward toward the SMBH (Tremblay et al., 2016).

Cluster-core studies provide a complementary thermodynamic diagnostic. In a sample of 37 massive, X-ray-bright, morphologically regular clusters, the condensation radius defined by $t_{\rm ff}\equiv \left(\frac{2r}{g}\right)^{1/2}.$2 peaks at $t_{\rm ff}\equiv \left(\frac{2r}{g}\right)^{1/2}.$3, remains below $t_{\rm ff}\equiv \left(\frac{2r}{g}\right)^{1/2}.$4, and is systematically smaller than the larger quenched cooling-flow radius defined by $t_{\rm ff}\equiv \left(\frac{2r}{g}\right)^{1/2}.$5. Typically $t_{\rm ff}\equiv \left(\frac{2r}{g}\right)^{1/2}.$6 is about three times larger than $t_{\rm ff}\equiv \left(\frac{2r}{g}\right)^{1/2}.$7, with an average relation $t_{\rm ff}\equiv \left(\frac{2r}{g}\right)^{1/2}.$8 (Wang et al., 2023). In this interpretation, the broader cool core traces the long-term “macro weather,” whereas the smaller $t_{\rm ff}\equiv \left(\frac{2r}{g}\right)^{1/2}.$9 marks the effective multiphase rain and CCA zone.

CCA also predicts characteristic X-ray and optical morphologies. In simulations of rotating atmospheres, CCA produces inner flat X-ray temperature profiles and density profiles approaching $\frac{t_{\rm cool}}{t_{\rm ff}} \lesssim 10,$0, contrasted with the cuspy central temperature rise of hot-mode/Bondi-like accretion. The same simulations reproduce a morphological dichotomy in H$\frac{t_{\rm cool}}{t_{\rm ff}} \lesssim 10,$1: filamentary nebulae in turbulence-dominated CCA states, and compact rotating discs in more quiescent rotation-dominated states (Gaspari et al., 2014). The top-down condensation model further connects X-ray, H$\frac{t_{\rm cool}}{t_{\rm ff}} \lesssim 10,$2, [C II], H I, and ALMA molecular structures as cospatial manifestations of the same multiphase rain (Gaspari et al., 2016).

At the luminous AGN end, large-sample scaling relations have been used to discriminate feeding modes statistically. A uniform sample of 1,729 unobscured blue quasars yields a near-linear bolometric relation,

$\frac{t_{\rm cool}}{t_{\rm ff}} \lesssim 10,$3

and a shallower hard-X-ray relation,

$\frac{t_{\rm cool}}{t_{\rm ff}} \lesssim 10,$4

In that analysis, classical hot-mode/Bondi accretion underpredicts the observed luminosities by $\frac{t_{\rm cool}}{t_{\rm ff}} \lesssim 10,$5 dex at the high-mass end and by $\frac{t_{\rm cool}}{t_{\rm ff}} \lesssim 10,$6 at lower masses, whereas a precipitation-linked CCA prescription reproduces both the normalization and the near-linear slope (Fiore et al., 2 Jun 2026). This result does not probe cloud-scale morphology directly, but it strongly supports the proposition that luminous SMBH feeding is tied to halo thermodynamics rather than local spherical hot capture.

5. Self-regulated feedback and multiscale “weather”

CCA is ordinarily embedded in a feedback loop rather than treated as an isolated inflow solution. In the canonical self-regulated picture, the atmosphere cools until local regions satisfy the condensation criterion, multiphase clouds and filaments form, the SMBH feeding rate rises toward the cooling rate, and the resulting mechanical outflows inject power

$\frac{t_{\rm cool}}{t_{\rm ff}} \lesssim 10,$7

with $\frac{t_{\rm cool}}{t_{\rm ff}} \lesssim 10,$8, into the core. Weak shocks, bubbles, uplift, and turbulence then reheat the atmosphere, suppress further condensation, and restart the cycle once the cold supply is exhausted (Gaspari, 2015).

Recent meso-scale studies have refined this into a “weather” taxonomy. In jet-regulated CCA, low ambient turbulence yields a rainy state: condensation starts earlier, remains coherent and centrally confined, and sustains a longer-lived inner cold reservoir. Stronger turbulence yields a stormy state: condensation starts later, becomes more radially extended and filamentary, and produces burst-dominated fueling; in some runs it later transitions to a cloudy state with substantial cold gas but inefficient central delivery (Cammelli et al., 26 May 2026). In turbulence-driven no-jet baselines, the same weak-versus-strong stirring contrast appears as compact rainy CCA versus extended stormy CCA, while the SMBH accretion rate in both cases remains recurrently boosted by up to $\frac{t_{\rm cool}}{t_{\rm ff}} \lesssim 10,$9 above the hot-mode Bondi baseline (Barbani et al., 26 May 2026).

The meso-scale literature also shows that “more turbulence” is not synonymous with “more effective feeding.” Strong stirring produces more extended cold mass and broader thermodynamic distributions, yet innermost accretion can remain similar to the weak-stirring case because feeding depends on how efficiently the multiphase structures couple to the central inflow rather than on total condensed mass alone (Barbani et al., 26 May 2026, Barbani et al., 26 May 2026). A further refinement is the torque-coherence measure

$$\mathcal{C}\equiv \frac{t_{\rm cool}}{t_{\rm eddy}}=1,$$0

which quantifies whether successive CCA torque episodes accumulate coherently or cancel vectorially. In spin-coupled simulations, stronger turbulence fragments the inflow and enhances cancellation, whereas weaker turbulence preserves longer feeding bridges and drives faster spin evolution (Piana et al., 26 May 2026).

6. Scope, related usages, and unresolved issues

CCA has a clear modern meaning, but adjacent literatures use related language differently. The analytic $$\mathcal{C}\equiv \frac{t_{\rm cool}}{t_{\rm eddy}}=1,$$1 argument for chaotic feeding by cold clouds or filaments with nearly random angular momentum is best regarded as a precursor or supporting argument for CCA-like geometry, not as a full precipitation model (Nayakshin et al., 2012). Likewise, “chaotic accretion” in high-redshift SMBH growth studies usually denotes a sequence of randomly oriented thin-disc episodes that regulate black-hole spin and radiative efficiency; that framework is related through stochastic orientation, but it does not include multiphase halo condensation, cloud rain, or the hot-halo feedback loop and is therefore not CCA proper (Zubovas et al., 2021). Quasar-shocked shell fragmentation can also generate cold filaments and chaotic inflows, but that is a distinct route to multiphase feeding rather than the canonical hot-halo precipitation pathway (Nayakshin et al., 2012).

Several unresolved issues concern how CCA couples to the innermost accretion flow. In hydrodynamic horizon-scale simulations of an M87-like elliptical, multiphase condensation and chaotic cold streams do occur, but a rotationally supported cold disc or torus forms easily and dominates the cold-gas dynamics most of the time; the genuinely chaotic stage occupies $$\mathcal{C}\equiv \frac{t_{\rm cool}}{t_{\rm eddy}}=1,$$2 of the time, and the smallest radii are usually fed mainly by hot virialized gas with $$\mathcal{C}\equiv \frac{t_{\rm cool}}{t_{\rm eddy}}=1,$$3 (Guo et al., 2022). This suggests that the large-radius presence of cold rain does not automatically guarantee persistent horizon-scale cold feeding.

Other work indicates that omitted plasma physics may be decisive. In giant ellipticals, weak magnetic fields can be amplified in precipitating cold gas and extract angular momentum through magnetic braking, thereby preventing the formation of massive central cold discs; in CR-dominated jet models, including CR streaming and streaming heating restores self-regulation and suppresses cooling catastrophes (Wang et al., 2019). A plausible implication is that CCA is robust as a halo- and meso-scale condensation process, while its translation into sustained horizon-scale feeding remains sensitive to magnetic stresses, CR transport, unresolved disc physics, and the detailed filtering of torque between sink scale and ISCO.

For that reason, CCA is best treated not as a single closed formula for SMBH accretion, but as a multiscale paradigm: hot halos condense nonlinearly into a multiphase rain, the rain transports mass and angular momentum inward through collisions and torques, and feedback reorganizes the thermodynamic background that sets the next episode. Within that paradigm, the main open questions concern the duty cycle of truly chaotic feeding, the relative importance of hot versus cold gas at the smallest radii, and the extent to which MHD, CR, and relativistic disc physics preserve, suppress, or transform the CCA cascade.