Black-Hole Binaries and Multiple Systems

A black-hole binary is an evolving physical system, not just two masses on a relativistic orbit. Its history may include stellar evolution, mass transfer, a galaxy merger, dynamical friction, three-body scattering, a circumbinary disk, or a tertiary companion before radiation reaction takes control. This chapter organizes those mechanisms by scale and by what is conserved, measured, or still uncertain. The recurring task is to identify the current bottleneck rather than attach one migration law to every binary.

Helpful background. Black-Hole Geometry supplies relativistic orbital scales; Relativistic Matter and Compact Objects supplies stellar and gas physics; Black-Hole Astrophysics supplies source and host context; Two-Body Dynamics supplies binary energy and angular momentum; Gravitational-Wave Astronomy supplies detector and population inference.

Six routes through binary evolution

The chapter follows six distinct but connected questions.

1. Stellar-Mass X-Ray Binaries joins a donor’s orbit and evolution to mass transfer, X-ray state, and radio jet while sharing distance, inclination, and absorption nuisance parameters.
2. Massive Black-Hole Binary Formation follows two galactic nuclei through stripping and dynamical friction until the black holes form a resolved bound pair.
3. Stellar-Dynamical Hardening derives how repeated three-body encounters extract binary energy and how an empty or refilled loss cone changes the rate.
4. Circumbinary Disks and Gas Torques closes gravitational, advective, accretion, and magnetic angular-momentum fluxes before assigning inspiral or expansion.
5. Triples and Secular Dynamics compares Kozai–Lidov torque with relativistic precession, octupole structure, instability, and direct three-body evolution.
6. Final-Parsec Evolution and Coalescence combines stellar, gaseous, geometric, and radiation-reaction mechanisms into a cosmic-time bridge to merger.

The first route concerns a compact stellar binary whose visible donor is part of the measurement. Routes two through four build a massive binary inside a galaxy. The fifth is a branch that can act on either stellar or massive systems. The sixth asks whether all environmental stages actually connect to the radiation-dominated end state.

The common bookkeeping

For an isolated Keplerian binary with total mass $M$, reduced mass $\mu$, semimajor axis $a$, and eccentricity $e$,

$$E_b=-\frac{G\mu M}{2a}, \qquad J_b=\mu\sqrt{GMa(1-e^2)}.$$

An environment changes the orbit only by carrying away or adding energy and angular momentum. Stars gain positive energy in slingshot encounters. Gas transports angular momentum through gravitational stress, advection, accretion, and magnetic or viscous stress. A tertiary exchanges angular momentum secularly while leaving the two semimajor axes approximately fixed in the ideal double-averaged limit. Gravitational waves carry both energy and angular momentum to infinity.

These mechanisms cannot be compared with unlabeled “timescales.” A useful massive-binary evolution equation is

$$\dot a = \dot a_{\rm df} -s_\star a^2 +\dot a_{\rm gas} +\dot a_{3} +\dot a_{\rm GW},$$

but it is stage-dependent. Dynamical friction ceases to be the right resolved description after a hard binary forms. A smooth $\dot a_3$ exists only in a stable secular regime. Each term requires a companion eccentricity equation, and each numerical control volume must close the corresponding energy and angular-momentum flux.

The stellar-mass case adds mass exchange. If matter moves from a donor to a black hole or escapes in a wind, $M$, $\mu$, and $J_b$ all evolve. A radial-velocity mass function constrains a combination of mass and inclination; it does not by itself determine the accretor mass. Likewise, an X-ray state identifies a repeatable phenomenology, not a unique disk geometry. The page-level joint model keeps these shared variables from being counted twice.

A scale-dependent bottleneck

The relevant environment changes as separation falls.

At kiloparsec to tens-of-parsec scales, two black holes generally remain attached to merging galactic components. Stripping changes the effective satellite mass, and a low-density core can weaken dynamical friction. A simulation must resolve the wake and the nuclear density rather than reposition black-hole particles by hand and call the result a pairing time.

Once the pair is hard, stars passing within its interaction region can be ejected. The reference full-loss-cone rate,

$$\frac{d}{dt}\left(\frac1a\right) \simeq H\frac{G\rho}{\sigma},$$

is sustainable only if new low-angular-momentum stars arrive. Spherical collisionless systems deplete this supply; relaxation, massive perturbers, rotation, and especially triaxial centrophilic orbits can refill it. Merritt and Milosavljević 2005, §§ 4–6 explains why loss-cone supply, rather than the local encounter coefficient alone, controls long-term hardening.

Gas introduces a different bottleneck: material must reach the nucleus, avoid premature star formation or expulsion, cross the cavity, and transfer net angular momentum. Simulations disagree when disk thickness, thermodynamics, magnetic topology, sink rules, boundary supply, and duration differ. The sign of one torque component is not the sign of $\dot a$.

At sufficiently small $a$, the radiation term rises approximately as $a^{-3}$ for fixed eccentricity while constant stellar hardening gives $\dot a_\star\propto-a^2$. Radiation reaction therefore wins sharply. The scientific uncertainty is usually the time needed to reach that transition, not whether a sufficiently compact isolated binary ultimately radiates.

Multiplicity is a mechanism, not a label

A stable tertiary can drive large inner eccentricity through orbit-averaged torque. In the test-particle quadrupole limit,

$$\sqrt{1-e_{\rm in}^2}\cos i_{\rm mut} =\text{constant},$$

but octupole effects, rapid relativistic apsidal motion, and finite hierarchy can invalidate that simple picture. Naoz 2016, §§ 2–4, pp. 445–468 lays out the secular expansion and its breakdown.

Observation of a comoving tertiary establishes multiplicity only after calibration and a chance-alignment model. It does not establish mutual inclination, the past outer orbit, or an active secular cycle. Conversely, a system can have experienced tertiary-driven evolution even if later mass loss widened the outer orbit. The chapter’s V404 Cygni case keeps the Gaia DR3 association likelihood separate from the evolutionary interpretation.

Unstable triples belong to direct few-body dynamics rather than the secular expansion. They can eject one member, exchange companions, or force a rapid encounter. Calling every three-body-assisted merger “Kozai–Lidov” erases this boundary.

Evidence, simulations, and population claims

Three levels of evidence recur. An individual system constrains masses, motion, gas, or a companion under a source model. A controlled calculation tests a mechanism and its convergence. A population inference asks how often the mechanism operates after formation and observational selection. Success at one level does not establish the next.

The computational pages therefore pin initial models, resolution, softenings or regularization, boundaries, duration, and seeds. They recover an analytic friction or scattering benchmark before interpreting a galaxy merger. They test particle-number dependence and conserved budgets before extrapolating a hardening rate. Frontier pages retain competing positions and explicit failure cases.

The observational pages keep the native data products and likelihood visible. Radial velocities, light curves, Gaia astrometry, imaging, spectra, and timing have different calibrations and selections. Shared distance, inclination, epoch, and source-state parameters belong in one generative model rather than in multiplied posteriors that reuse the same prior.

The observational and frontier comparisons include results through 2026-08-18. Conservation laws and limiting scalings are more durable. Candidate associations, calibration releases, gas-torque signs, host-shape distributions, and population delay models can change as new observations or calculations test their assumptions.

Connections to other topics

This chapter develops source evolution and the environmental transport that changes a black-hole pair. For vacuum orbital mechanics, continue to Two-Body Dynamics; for waveform and selection inference, see Gravitational-Wave Astronomy. Disk thermodynamics is developed in Accretion Flows and Disks, and the emitted spectra in Radiation, Spectra, and Variability.

Find the active hardening channel

Consider an equal-mass $10^8M_\odot+10^8M_\odot$ circular binary with $K_{\rm GW}=8.60\times10^{-7}\,{\rm pc^4\,Gyr^{-1}}$ and stellar hardening $s_\star=20\,{\rm pc^{-1}\,Gyr^{-1}}$. A gas model contributes $\dot a_{\rm gas}=+0.05\,{\rm pc\,Gyr^{-1}}$ at $a=0.10\,{\rm pc}$. Compute the stellar and radiation terms and identify the net direction. Then set $a=0.020\,{\rm pc}$ and $\dot a_{\rm gas}=+0.001\,{\rm pc\,Gyr^{-1}}$; which channel controls the second state?

Solution

At $0.10\,{\rm pc}$,

$$\dot a_\star=-s_\star a^2=-0.20\ {\rm pc\,Gyr^{-1}}, \qquad \dot a_{\rm GW}=-\frac{K_{\rm GW}}{a^3} =-8.60\times10^{-4}\ {\rm pc\,Gyr^{-1}}.$$

Including gas gives $\dot a=-0.151\,{\rm pc\,Gyr^{-1}}$ to the shown precision. Stars dominate and the gas delays, but does not reverse, the inspiral.

At $0.020\,{\rm pc}$, $\dot a_\star=-0.0080\,{\rm pc\,Gyr^{-1}}$, while

$$\dot a_{\rm GW} =-\frac{8.60\times10^{-7}}{(0.020)^3} =-0.108\ {\rm pc\,Gyr^{-1}}.$$

The small positive gas term leaves $\dot a\simeq-0.115\,{\rm pc\,Gyr^{-1}}$. Radiation reaction now controls the shrinkage. The calculation compares signed rates at one state; a merger time still requires integration with the evolving eccentricity and environment.

References

• Merritt, David, and Miloš Milosavljević. “Massive Black Hole Binary Evolution.” Living Reviews in Relativity 8 (2005): 8. doi:10.12942/lrr-2005-8.
• Naoz, Smadar. “The Eccentric Kozai–Lidov Effect and Its Applications.” Annual Review of Astronomy and Astrophysics 54 (2016): 441–489. doi:10.1146/annurev-astro-081915-023315.