Stratified Accretion Disk Wind

Updated 1 January 2026

Stratified accretion disk wind is a physically layered, multi-zone outflow from accretion disks around compact objects, characterized by clear gradients in density, velocity, and ionization.
Observations across X-ray, UV, and optical bands reveal distinct absorption and emission features that help constrain wind properties and differentiate launching mechanisms.
Advanced MHD and radiative simulations, combined with high-resolution spectroscopy, are essential for modeling wind structure and assessing their impact in AGN, XRBs, and protoplanetary disks.

A stratified accretion disk wind is a physically layered, large-scale outflow originating from the surface of an accretion disk around a compact object, characterized by spatial gradients in physical parameters—including density, velocity, and ionization state—as functions of both radius and polar angle. Observational and theoretical studies across black hole X-ray binaries (XRBs), active galactic nuclei (AGN), protoplanetary disks, and young stellar objects demonstrate that these winds exhibit multi-zone or continuous stratification, with distinct kinematic, thermodynamic, and radiative properties at different disk radii and heights. This stratification results from a combination of disk structure, radiative and magnetic driving mechanisms, and local physical conditions, and is crucial for interpreting wind signatures in high-resolution X-ray and optical/UV spectra (Fukumura et al., 22 Oct 2025, Marziani et al., 27 Dec 2025, Collaboration et al., 18 Sep 2025, Miller et al., 2016, Matthews et al., 2020, Fukumura et al., 2013).

1. Physical Structure and Parameterization

Stratified disk wind models invoke radial and poloidal variation of density, velocity, and ionization. A typical parameterization for the electron density and velocity in an MHD-driven wind is:

$n(r,\theta) = n_0\,g(\theta)\,f_{\rm den}\left(\frac{r}{r_0}\right)^{-p} \quad v(r,\theta) = v_0\,h(\theta)\,f_v\left(\frac{r}{r_0}\right)^{-1/2}$

with $r$ the spherical radius, $\theta$ the polar angle measured from the disk axis, $n_0$ (density normalization), $v_0$ (velocity normalization), $r_0$ (inner launch radius), and $p\sim1-1.4$ (density power-law slope). The angular functions $g(\theta)$ and $h(\theta)$ encode MHD geometry, with values near unity in the disk midplane and $\ll1$ at the poles, reflecting poloidal collimation (Fukumura et al., 22 Oct 2025, Fukumura et al., 2013, Bai et al., 2013). Stratification naturally emerges in both bi-conical and vertically-stratified geometries.

A summary of scale-dependent properties:

Radial Zone	Typical Launch Radius ( $r$ )	Density ( $n$ )	Outflow Velocity ( $v$ )	Ionization ( $\log \xi$ )
Innermost	$\sim$ few– $10^3$ $r_g$	$10^{12}$ – $10^{16}$ cm $^{-3}$	0.01–0.1 $c$	4–5.5
Intermediate	$10^3$ – $10^4$ $r_g$	$10^{10}$ – $10^{13}$ cm $^{-3}$	$10^3$ – $10^4$ km/s	3–5
Outer (Wind base)	$10^4$ – $10^5$ $r_g$	$10^8$ – $10^{10}$ cm $^{-3}$	$10^2$ – $10^3$ km/s	1–4

This stratification results in distinct observational zones: inner ultra-fast outflows (UFOs), intermediate X-ray warm absorbers, and outer thermal/Compton-driven winds or failed wind regions (Marziani et al., 27 Dec 2025, Braito et al., 2020, Miller et al., 2016).

2. Observational Diagnostics and Signatures

Stratified disk winds are revealed through multi-component absorption and emission lines across the X-ray, UV, and optical bands. High-resolution X-ray spectra (e.g., XRISM/Resolve, Chandra/HETG, XMM-Newton) detect multiple zones characterized by different ionization parameters, column densities, and velocities via features such as Fe XXVI Ly $\alpha$ doublets and Fe K-shell transitions (Fukumura et al., 22 Oct 2025, Braito et al., 2020, Marziani et al., 27 Dec 2025, Miller et al., 2016).

Characteristic diagnostics:

Fe XXVI Ly $\alpha_{1,2}$ doublet: Line profile (single broad, canonical 2:1 doublet, or double-peaked 1:1) directly constrains wind velocity dispersion $\sigma_v$ and hence the local turbulence and density (Fukumura et al., 22 Oct 2025).
UV/Optical lines: Blueshifts and asymmetries of high-ionization (e.g., C IV $\lambda$ 1549), intermediate-ionization (Al III $\lambda$ 1860), and low-ionization (Mg II $\lambda$ 2800, H $\beta$ ) lines map wind stratification in AGN (Marziani et al., 27 Dec 2025).
X-ray Warm Absorbers/UFOs: Observed as multiple absorption zones with $v/c\sim 0.01$ –0.3 and $\log\xi\sim3$ –$5$, often contemporaneously (Braito et al., 2020, Miller et al., 2016, Fukumura et al., 2013).

XRISM/Resolve and Chandra/HETG have directly resolved the spectral signatures of stratified wind regimes, confirming the model predictions (Fukumura et al., 22 Oct 2025, Miller et al., 2016).

3. Launching Mechanisms and Theoretical Models

The physical drivers of stratified winds include magnetohydrodynamic (MHD) stresses, radiative line-driving, thermal (Compton) driving, and combinations thereof. The relative importance of each mechanism varies by object class, luminosity, and launching radius:

Inner disk (few– $10^3$ $r_g$ ): Magnetic driving (both pressure and magnetocentrifugal) dominates; supported by high densities, short launching radii, and high velocities inconsistent with thermal/radiative-only scenarios (Miller et al., 2016, Fukumura et al., 2013). The canonical MHD solution gives $v\propto r^{-1/2}$ and $n\propto r^{-p}$ , with opening angles set by magnetic field geometry and specific angular momentum.
Intermediate/outer disk ( $10^4$ – $10^5$ $r_g$ ): Thermal-Compton driving becomes significant, producing equatorial, slow (few $10^2$ km/s) outflows. Stratification arises as localized irradiation increases temperature and pressure, launching distinct wind slices (Her X-1, GX 13+1) (Collaboration et al., 18 Sep 2025, Kosec et al., 2023).
Radiative line-driving: In high-Eddington AGN disks, radiation pressure on lines accelerates high-ionization, low-column gas to high velocities near the continuum source, but deeper layers become bound at high column densities ( $N_H\gtrsim10^{23-24}\,\mathrm{cm}^{-2}$ ) (Marziani et al., 27 Dec 2025, Matthews et al., 2020).

Global 2D/3D MHD and radiation-hydrodynamics simulations reproduce the observed stratified flows and their transition from magnetically confined, fast, ionized layers to outer, thermally driven slow winds (Collaboration et al., 18 Sep 2025, Kosec et al., 2023, Matthews et al., 2020).

4. Ionization, Density, and Velocity Profiles

The radial and angular stratification governs the observables. Linear scaling laws predict:

$\xi(r,\theta) = \frac{L_{\mathrm{ion}}}{n(r,\theta)\,r^2}$

$v(r) = v_\infty(1 - R_0/r)^\beta$

with $\xi$ the ionization parameter, $L_{\mathrm{ion}}$ the ionizing luminosity, $v_\infty$ the terminal velocity, $R_0$ launch radius, and $\beta$ the velocity-law slope ( $\sim$ 0.5–1; lower for more magnetically dominated flows, higher for radiatively driven winds) (Marziani et al., 27 Dec 2025). Density typically falls as $n\propto r^{-p}$ , with $p=1$ –1.5.

Observational and modeling work reveals:

Steep radial density profiles ( $p>1$ ) yield a large range of $U_H$ , $\xi$ satisfying both high-ionization (inner launch) and low-ionization (failed wind/BLR base) requirements (Matthews et al., 2020).
In multi-phase AGN winds, distinct phases with $v/c\sim0.07$ and $0.2$ ("Zones 1 and 2") correspond to different launch radii and densities, with the faster, higher- $\xi$ phase less persistent (Braito et al., 2020).

Angular stratification (bi-conical, polar, or equatorial) determines which lines of sight intercept Compton-thick wind structure and influences observed absorbing column densities and blueshifted lines (2207.13731, Matthews et al., 2020).

5. Variability, Regimes, and Observational Implications

Stratified winds are dynamic: both temporal variability and clumping/fractal structure have been observed. Key regimes and transitions:

Three turbulence regimes (Fe XXVI doublet):
- $\sigma_v\gtrsim 5,000$ km/s (broad, blended trough)
- $\sigma_v\sim 1,000$ –$5,000$ km/s (canonical 2:1 profile)
- $\sigma_v\lesssim 1,000$ km/s (double-peaked, 1:1) [(Fukumura et al., 22 Oct 2025), XRISM/Resolve]
Variability: Intermittent appearance (or disappearance) of fast wind zones in AGN and black hole XRBs on timescales of days–years, with Compton-thick absorbers appearing/disappearing as the wind geometry and column change (Braito et al., 2020).
Spectral modification: Compton scattering by stratified winds can systematically down-scatter disk photons, producing up to $\sim$ 70% flux suppression—biasing black hole spin measurements based on continuum fitting if not properly modeled (Fukumura, 2 Jan 2025).
Failed winds: Sub-escape-velocity, stratified/wind regions can dominate broad-line emission in AGN, supporting models in which much of the observable BLR arises from failed or slowly rising wind layers rather than true outflows (Matthews et al., 2020).

The table below summarizes observable regimes for the Fe XXVI doublet:

Regime	$\sigma_v$ (km/s)	Profile	Physical Interpretation
I	$\gtrsim 5,000$	Blended, single broad trough	Strong turbulence, MHD/shocks
II	$1,000$–$5,000$	Standard 2:1 doublet	Moderate shear/thermal turbulence
III	$\lesssim 1,000$	Narrow, nearly 1:1 doublet	Quiescent flow, weak turbulence

6. Multiwavelength and Multiphase Contexts

Stratified disk winds appear across all disk-fed accreting systems:

XRBs and AGN: Show multi-zone absorbers (fast zones: UFOs; slow, persistent: warm absorbers); major scale-dependent similarities and differences arise in density, velocity, and total kinetic power (Fukumura et al., 22 Oct 2025, Braito et al., 2020).
Protoplanetary Disks: Vertically stratified, magnetocentrifugal winds drive angular momentum loss, with suppression of MRI turbulence and accretion proceeding in thin, offset current layers. These winds are vital for disk evolution and planetesimal formation (Bai et al., 2013).
Young Stellar Objects: FU Ori–type outbursts produce stratified winds, with a fast, collimated inner jet and a slower, wide-angle outer wind, each with distinct launch radii, acceleration, and temperature gradients (Carvalho et al., 2024).
Observational "X-ray tomography": Precessing/warped disks (e.g., Her X-1) allow tomography of the vertical wind structure, showing sharp density declines with height and fragmentation into clumps at large $z$ (Kosec et al., 2023).

The stratified wind paradigm unifies these phenomena—quantitatively capturing the transition from compact, fast, high-ionization outflows near the disk to slow, massive, lower-ionization layers at larger radii.

7. Modeling Methodologies and Future Prospects

Recent advances in physically self-consistent, multi-dimensional Monte Carlo radiative transfer, coupled with machine-learning-based emulators, now enable rapid, accurate modeling of stratified disk winds and their spectra across large parameter spaces (e.g., XRADE emulator) (2207.13731). These techniques incorporate local ionization balance, velocity fields, and complex emission/absorption processes.

Physical grid construction: $n(r, \theta)$ , $v(r, \theta)$ , $\xi(r, \theta)$ precomputed on multi-dimensional grids.
Non-spherical geometry & self-shielding: Biconical/polar flows with self-shielding and multiple scattering included (Matthews et al., 2020, 2207.13731).
Predictive diagnostics: Emulators enable direct fitting of observed high-resolution spectra, distinguishing among wind launching mechanisms and constraining disk physics in both AGN and XRBs.
Interpretation of feedback and evolution: Stratified wind energetics and coupling to molecular outflows establish vital connections between nuclear activity and host galaxy/star-disk evolution (Braito et al., 2020, Collaboration et al., 18 Sep 2025).

Ongoing and future high-sensitivity observations (e.g., XRISM/Resolve, Athena/X-IFU) will further test predictions of stratified wind theory, allowing time-resolved, spatially sensitive mapping of wind structure and accretion feedback across all compact object classes.

Key references: (Fukumura et al., 22 Oct 2025, Marziani et al., 27 Dec 2025, Collaboration et al., 18 Sep 2025, Matthews et al., 2020, Braito et al., 2020, Fukumura et al., 2013, Bai et al., 2013, 2207.13731, Miller et al., 2016, Kosec et al., 2023, Fukumura, 2 Jan 2025, Carvalho et al., 2024)