AGN Radiation-Driven Winds: How Quasar Light Blows Gas Out of Galaxies

What it is: photons that push gas

Radiation carries momentum: a photon of energy E carries momentum E/c. When light from an accretion disk is absorbed or scattered by surrounding gas, that momentum is deposited, exerting an outward radiation pressure. In an active galactic nucleus (AGN), the central engine is a supermassive black hole of 10⁶–10¹⁰ M_☉ accreting through a disk that radiates from the infrared to hard X-rays, often near the Eddington luminosity L_Edd ≈ 1.26×10³⁸ (M/M_☉) erg/s.

An AGN radiation-driven wind is the gas that this radiation pushes off the disk and out through the galaxy. It is one of two broad flavors of AGN feedback — the other being relativistic jets (the "radio/kinetic mode"). Radiation driving is the natural "quasar/radiative mode": it dominates when the black hole accretes rapidly and shines brightly. The crucial insight is that the same luminosity that defines the Eddington limit for pure electron scattering becomes vastly more effective once you add line opacity or dust, so radiation can expel gas even when it seems formally sub-Eddington.

The mechanism: three ways light grabs the gas

Radiation couples to gas through opacity, and the outward force per unit mass is a_rad = κ L / (4π r² c), where κ is the relevant opacity. Three channels dominate, in order of increasing grip:

• Electron scattering (κ_es ≈ 0.34 cm²/g for ionized H). This sets the classic Eddington limit but is too weak on its own to expel much gas unless accretion is super-Eddington.
• UV line driving. In partly ionized gas, resonant bound-bound transitions of abundant ions (C IV 1549 Å, N V, O VI) absorb UV photons enormously more efficiently than electrons. The force multiplier M(t) can boost the effective opacity by 10–1000×, launching a disk-atmosphere wind. This is the leading model for ultra-fast outflows and broad-absorption-line (BAL) quasars.
• Dust driving. Dust grains have κ ~ 10²–10³ times the electron value, so even modest luminosities push dusty gas. A quasar at the Eddington limit for ionized gas is at the effective Eddington limit for a dusty system ~1000× more massive — the key to galaxy-scale feedback.

Wind material self-shields: as it accelerates the ionization drops, keeping lines effective — but too much X-ray flux over-ionizes the gas, killing line driving, so real disk winds thread a narrow "survival" corridor.

Key quantities and a worked example

The natural momentum scale is Ṗ ≈ L/c: if every photon is absorbed once, the wind gains momentum L/c per second ("single scattering"). For L = 10⁴⁶ erg/s that is L/c ≈ 3×10³⁵ dyne. Observed molecular outflows often show a momentum boost of ~10–20 × L/c, the signature of an energy-driven phase where a hot shocked bubble does PdV work.

• Ultra-fast outflows (UFOs): v ≈ 0.1–0.3c, column N_H ~ 10^23–24 cm⁻², launched at ~10–100 gravitational radii, kinetic power 10⁴²–10⁴⁵ erg/s.
• Feedback threshold: theory (Hopkins & Elvis 2010; Di Matteo et al. 2005) finds only ~0.5–5% of the bolometric luminosity need become mechanical wind power to regulate the black hole and quench the bulge.

Worked case: A 10⁸ M_☉ black hole at Eddington shines at L ≈ 1.3×10⁴⁶ erg/s. A UFO carrying dM/dt ≈ 1 M_☉/yr at v = 0.1c has kinetic power ½(dM/dt)v² ≈ 3×10⁴⁴ erg/s — about 2% of L, right in the feedback-relevant band.

How it's observed and detected

Radiation-driven winds are read off as blueshifted absorption imprinted on the AGN's own light — the gas is moving toward us, in front of the source. Different velocity regimes appear in different bands:

• X-ray Fe-K UFOs: highly ionized iron (Fe XXV/XXVI) absorption lines near 7–10 keV, blueshifted to imply v ~ 0.03–0.3c. Discovered systematically with XMM-Newton and Suzaku (Pounds, Tombesi, Reeves and collaborators, 2000s–2010s); PDS 456 and IRAS 13224−3809 are landmark cases.
• UV BAL troughs: broad, deep C IV, Si IV, and N V absorption in ~15–20% of quasars, tracing 5,000–60,000 km/s winds (SDSS, HST/COS).
• Galaxy-scale ionized/molecular outflows: [O III] 5007 Å blue wings and CO/OH P-Cygni profiles seen with integral-field units (MUSE), ALMA, and Herschel, tracing 100–1000 km/s winds extending to kpc scales.

Connecting a fast nuclear UFO to a slow, massive kpc-scale outflow in the same object (e.g., recent JWST/VLT work) is the observational holy grail — direct evidence that the small wind entrains and drives the big one.

How it compares to related regimes

Radiation-driven winds sit within a broader family of black-hole outflows; distinguishing them matters:

• vs. magnetohydrodynamic (MHD) winds: MHD/magnetocentrifugal winds are flung out along disk field lines and can reach similar speeds without any line opacity. Disentangling radiation from magnetic launching in a given UFO is genuinely unsettled.
• vs. relativistic jets: jets are collimated, magnetically dominated, and reach bulk Lorentz factors of several to tens — a fundamentally different (kinetic-mode) feedback channel, favored in low-accretion-rate systems.
• vs. momentum- vs. energy-driven phases: if the shocked wind cools efficiently it is momentum-driven (Ṗ ≈ L/c, confined to the inner ~few hundred pc, giving M_BH ∝ σ⁴). If cooling is inefficient the hot bubble expands adiabatically — energy-driven — boosting momentum ~10–20× and clearing gas to kpc scales.
• vs. stellar winds & supernovae: stellar feedback shapes low-mass galaxies, but only AGN winds carry enough energy to unbind gas from the deep potentials of massive ellipticals.

Significance, famous cases, and open questions

Radiation-driven winds are the front-runner explanation for why galaxies are not more massive than they are. In cosmological simulations (EAGLE, IllustrisTNG, SIMBA), AGN feedback is required to shut down star formation in massive galaxies and reproduce the observed galaxy stellar-mass function and the M–σ relation (M_BH ∝ σ^~4–5). The King & Pounds analytic framework ties this scaling directly to a momentum-driven, Eddington-limited wind.

Famous cases: PDS 456, the most luminous quasar in the local universe, drives a wide-angle 0.25c UFO carrying ~15–20% of L in momentum. IRAS 13224−3809 shows UFO variability that responds to disk luminosity in near-real time. Markarian 231 and IRAS F11119+3257 link a nuclear UFO to a massive molecular outflow.

Open questions: Is line driving or MHD the true launcher of UFOs? How much of the fast wind's momentum actually couples to galaxy-scale gas (the "boost factor")? Does over-ionization by X-rays cripple line driving in the most luminous quasars? And does AGN feedback suppress or sometimes trigger star formation? These remain active, contested frontiers.

The three radiation-driving channels and how their opacity, location, and speed compare

Driving channel	Opacity source	Where it acts	Typical outflow speed
Electron (Thomson) scattering	Free electrons, κ_es ≈ 0.34 cm²/g	Fully ionized inner disk; sets Eddington limit	Marginal unless super-Eddington
UV line driving	Bound-bound metal ion lines (C IV, N V, O VI); force multiplier 10–1000×	Warm, partly ionized disk atmosphere at ~10–1000 r_g	0.1–0.3c (ultra-fast outflows, BALs)
Dust (IR) driving	Dust grains, κ_dust ≈ 10²–10³ × κ_es	Dusty ISM / torus, ~1 pc to kpc	100–1000 km/s (molecular & ionized outflows)
Comparison: magnetic (MHD) wind	Magnetocentrifugal launch, not radiation	Threaded disk field lines	Variable, up to ~0.1c