Disk Winds and Black-Hole Outflows

Disk winds reveal themselves through blueshifted absorption, emission, scattering, and state-dependent obscuration, but a line velocity alone does not identify what launched the gas. Thermal pressure, continuum or line radiation force, and magnetic stress can coexist. The discriminants are a joint set: ionization, column, density, velocity structure, spectral energy distribution, launch radius, geometry, and variability, all folded through the instrument response and an explicit selection procedure. Even then the result is conditional because density-sensitive levels, saturation, partial covering, and an unseen ionizing continuum can move the inferred radius enough to change the preferred mechanism.

Required background. Classical Field Theory supplies radiation forces and magnetic stresses.

Helpful background. Black-Hole Geometry and Mechanics supplies orbital and escape scales; Relativistic Matter and Compact Objects supplies thermodynamics and compact-binary context; and Fluid and Kinetic Theory supplies wind dynamics, ion populations, shocks, and transport.

From an absorption spectrum to wind parameters

For a photoionized zone of hydrogen number density $n$ at radius $r$, define

$$\xi=\frac{L_{\rm ion}}{nr^2}, \qquad N_H=\int n\,ds.$$

Here $L_{\rm ion}$ must be integrated over the same spectral band used by the photoionization grid. Ionic ratios constrain $\xi$ only after the ionizing spectral energy distribution, abundances, atomic rates, and density effects are specified. A metastable-to-ground line ratio can constrain $n$, giving

$$r=\left(\frac{L_{\rm ion}}{n\xi}\right)^{1/2},$$

but radiative pumping and cascades can populate the same level that would otherwise be interpreted as a collisional density diagnostic. Line centroids give a line-of-sight velocity only after the wavelength scale, systemic velocity, blends, and relativistic Doppler convention are fixed. Equivalent width ceases to grow linearly with column when a line saturates; the line-spread function can hide a black intrinsic core.

For a steady sector subtending solid angle $\Omega$ in steradians, with volume filling factor $C_V$, mean mass per hydrogen $\mu_m m_p$, and radial speed $v$, a common estimate is

$$\dot M_w\simeq \Omega C_V\mu_m m_p nr^2v =\Omega C_V\mu_m m_p\frac{L_{\rm ion}}{\xi}v, \qquad \dot E_k=\frac12\dot M_wv^2.$$

The apparent cancellation of radius does not remove the geometry problem: $\Omega$, $C_V$, velocity projection, radial stratification, and duty cycle remain. Treating $C_V=1$ as a measurement can overstate the mass loss by orders of magnitude.

Thermal, radiative, and magnetic launching

The Compton temperature $T_C$ of the illuminating spectrum defines

$$R_C=\frac{GM\mu_{\rm part}m_p}{T_C}$$

in the site’s $k_B=1$ units. Here $\mu_{\rm part}m_p=\rho/(n_e+n_{\rm ions})$ is the mean mass per free particle, approximately $0.61m_p$ for fully ionized gas of near-solar composition. It differs from $\mu_m m_p=\rho/n_H$ in the mass-loss formula. Restoring $k_{\rm B}$ gives $R_C=GM\mu_{\rm part}m_p/(k_{\rm B}T_C)$: the radius where the thermal scale $k_{\rm B}T_C/(\mu_{\rm part}m_p)$ matches $GM/R$. Outside a fraction of $R_C$, sufficiently heated gas can form a thermal wind. Radiation reduces effective gravity; residual line and bound–free opacity add force. The dynamics depend on luminosity and spectrum Begelman, McKee, and Shields 1983, §§ II–IV, pp. 72–82, with binary observations reviewed in Díaz Trigo and Boirin 2016, §§ 2–4.

Magnetocentrifugal launching can operate when a sufficiently inclined ordered poloidal field lets matter move outward along a rotating line while extracting disk angular momentum. In the ideal cold-disk construction, the line must tilt more than $30^\circ$ from the vertical at the surface; magnetic pressure gradients and toroidal fields allow broader configurations. The derivation and its assumptions appear in Blandford and Payne 1982, §§2–3. This is a dynamical possibility, not an observational label: an absorption spectrum rarely measures the field geometry directly.

Mechanism	Necessary physical scale	Spectral or temporal leverage	Principal degeneracy
Thermal	Disk extends to an appreciable fraction of $R_C$	Response to $T_C$, luminosity, disk size, and state	Unseen ionizing SED and radiative assistance
Radiative	Sufficient luminosity and residual opacity	Momentum flux, ion stages, UV/X-ray continuum changes	Shielding, saturation, and force-multiplier model
Magnetic	Ordered or turbulent field supplies stress	Launch inside a robust thermal radius; velocity stratification	Density, geometry, and no direct field measurement

A credible comparison fits these mechanisms to the same continuum, atomic data, and response. Rejecting thermal driving using an observed luminosity while allowing an optically thick wind to attenuate that luminosity is internally inconsistent.

Dataset, calibration, selection, and likelihood

The evidence cutoff for this page is 2026-08-18. A concrete high-resolution reference is the public Chandra ACIS/HETGS observation of GRO J1655−40, ObsID 5461, obtained on 2005 April 1 for 62.15 ks in graded continuous-clocking mode Neilsen and Homan 2012, §2. A reproducible frozen reduction uses the Transmission Grating Catalog products, combines the positive and negative first-order HEG spectra with CIAO 4.13, and carries their response matrices and effective-area files into SPEX 3.05 with its pion model, as documented by Tomaru, Done, and Mao 2023, §§2–3. Their atomic comparison also records CHIANTI 10.0.1 and shows why radiative cascades and electron scattering change the density and intrinsic continuum.

Calibration is part of the inference. Preserve the release-specific HEG wavelength solution and line-spread function, ACIS gain and effective area, order-sorting and bad-pixel masks, exposure and dead time, background extraction, plus/minus-order registration, and any RXTE or broadband cross-normalization. Do not fit an unfolded spectrum as though its bins were independent measurements. For predicted source counts

$$\lambda_i(\vartheta)=t_{\rm exp} \int R_i(E)A(E)S(E;\vartheta)\,dE+b_i,$$

use the Poisson count likelihood for source and background. Parameters include continuum and SED shape, $\xi$, $N_H$, $n$, velocity, turbulent width, abundances, covering fraction, and one or more zones. Priors on logarithmic scale parameters and geometry must be stated; shared continuum and abundance nuisance parameters must remain shared across lines and orders.

Line selection also changes the answer. State the blind wavelength range, continuum windows, binning, masked detector features, time or flux intervals, treatment of blended doublets, detection threshold, and upper-limit construction. A search over many centroids and widths needs a global false-positive calibration from simulated continuum-only spectra, not a local $\Delta C$ threshold. ObsID 5461 was highlighted because it is unusually line rich, so it cannot by itself determine wind incidence. Population statements must include nondetections and the inclination, spectral-state, exposure, and count-rate selection function.

An adversarial reanalysis of one famous wind

The original detailed fit of ObsID 5461 used three photoionization codes and inferred a dense, highly ionized wind at a small radius, arguing that thermal driving was inadequate Miller et al. 2008, §§2–5, pp. 1361–1372. The result was scientifically valuable because the Fe XXII metastable line appeared to supply density information, but it was never independent of the atomic population model and intrinsic SED.

A later count-space reanalysis included radiative cascades, electron scattering, partial covering, and alternative unobserved soft-X-ray continua. It inferred a density more than an order of magnitude lower and showed that an optically thick thermal–radiative model could reproduce much of the spectrum Tomaru, Done, and Mao 2023, §§3–4 and §7. That calculation has its own limitations: its preferred inclination and optical depth are imperfect, and its radiation-hydrodynamic model does not close every scattering feedback. The defensible conclusion at the cutoff is therefore narrower than either slogan: this observation does not require a uniquely magnetic wind, and thermal–radiative and magnetic models should be compared with matched atomic physics, attenuation, geometry, and line-profile predictions.

Reproduction begins from the Chandra Data Archive observation search or the named TGCat products, records checksums and all response files, reruns the plus and minus orders separately, and injects lines into continuum-only simulations to calibrate trials. It then refits at least two continua and two atomic/photoionization treatments. This page does not claim a fresh numerical reproduction of the published fit; it specifies the software and data boundary needed to perform one honestly.

An absorption-line wind constrains matter leaving the disk; it does not by itself identify the source of a relativistic jet’s power. That separate electromagnetic problem is developed in The Blandford–Znajek Process, while Relativistic Matter and Compact Objects develops the broader plasma and radiative closure used by wind models.

Exercises

Inference and filling-factor diagnostic. Suppose $L_{\rm ion}=10^{38}\ {\mathrm{erg}\,\mathrm{s}^{-1}}$, $\xi=10^4\ {\mathrm{erg}\,\mathrm{cm}\,\mathrm{s}^{-1}}$, $n=10^{12}\ {\mathrm{cm}^{-3}}$, $N_H=3\times10^{22}\ {\mathrm{cm}^{-2}}$, $v=5\times10^7\ {\mathrm{cm}\,\mathrm{s}^{-1}}$, $\Omega=2\pi$, $C_V=0.1$, and $\mu_m=1.23$. Find $r$, the slab thickness $\Delta r=N_H/n$, $\dot M_w$, and $\dot E_k$. What dominates the systematic range if $C_V$ is only known to lie between $0.01$ and $1$?

Solution

The ionization relation gives

$$r=\sqrt{\frac{10^{38}}{10^{12}10^4}}=10^{11}\ {\mathrm{cm}}.$$

The thickness is $\Delta r=3\times10^{10}\ {\mathrm{cm}}=0.3r$. With $m_p=1.673\times10^{-24}\ {\mathrm{g}}$,

$$\dot M_w =(2\pi)(0.1)(1.23)(1.673\times10^{-24}) \frac{10^{38}}{10^4}(5\times10^7) \simeq6.5\times10^{17}\ {\mathrm{g}\,\mathrm{s}^{-1}},$$

and $\dot E_k\simeq8.1\times10^{32}\ {\mathrm{erg}\,\mathrm{s}^{-1}}$. Because both scale linearly with $C_V$, the allowed filling-factor interval alone spans a factor of one hundred. A small statistical error on $\xi$ cannot compensate for that geometric systematic.

References

• Begelman, Mitchell C., Christopher F. McKee, and Gregory A. Shields. “Compton Heated Winds and Coronae above Accretion Disks. I. Dynamics.” The Astrophysical Journal 271 (1983): 70–88. DOI.
• Blandford, Roger D., and David G. Payne. “Hydromagnetic Flows from Accretion Discs and the Production of Radio Jets.” Monthly Notices of the Royal Astronomical Society 199 (1982): 883–903. DOI.
• Díaz Trigo, María, and Laurence Boirin. “Accretion Disc Atmospheres and Winds in Low-Mass X-Ray Binaries.” Astronomische Nachrichten 337 (2016): 368–374. DOI.
• Miller, Jon M., et al. “The Accretion Disk Wind in the Black Hole GRO J1655−40.” The Astrophysical Journal 680 (2008): 1359–1377. DOI. Open preprint.
• Neilsen, Joseph, and Jeroen Homan. “A Hybrid Magnetically/Thermally Driven Wind in the Black Hole GRO J1655−40?” The Astrophysical Journal 750 (2012): 27. DOI.
• Tomaru, Ryota, Chris Done, and Junjie Mao. “What Powers the Wind from the Black Hole Accretion Disc in GRO J1655−40?” Monthly Notices of the Royal Astronomical Society 518 (2023): 1789–1801. DOI.