The critical role of clumping in line-driven disc winds

TL;DR

The study addresses the overionization problem in line-driven disc winds by introducing microclumping in a self-consistent Monte Carlo radiation-hydrodynamics framework. By varying the clumping factor up to $f_ ext{cl}\\sim 100$ (equivalently $f_V\sim 0.01$–$0.1$), it demonstrates that modest clumping dramatically lowers the ionization state, restores the line-driving force, and yields powerful winds with $rac{ ext{M}_ ext{wind}}{ ext{M}_ ext{acc}} \gtrsim 10^{-4}$, along with UV resonance lines absent in smooth models. The clumped winds also reprocess a significant fraction of the disc luminosity, reshaping the broad-band SED via enhanced bound-free opacities and disc backwarming, while obeying energy conservation. Together, these findings provide the first robust, self-consistent demonstration that clumping reconciles line-driven wind theory with observations for AWDs and AGNs, and they highlight clumping as a central ingredient for realistic disc-wind models.

Abstract

Radiation pressure on spectral lines is a promising mechanism for powering disc winds from accreting white dwarfs (AWDs) and active galactic nuclei (AGN). However, in radiation-hydrodynamic simulations, overionization reduces line opacity and quenches the line force, which suppresses outflows. Here, we show that small-scale clumping can resolve this problem. Adopting the microclumping approximation, our new simulations demonstrate that even modest volume filling factors ($f_V \sim 0.1-0.01$) can dramatically increase the wind mass-loss rate by lowering its ionization state -- raising $\dot{M}_{\rm wind}$ and yielding $\dot{M}_{\rm wind}/\dot{M}_{\rm acc}\!\gtrsim\!10^{-4}$ for such modest filling factors. Clumpy wind models produce the UV resonance lines that are absent from smooth wind models. They can also reprocess a significant fraction of the disc luminosity and thus dramatically modify the broad-band optical/UV SED. Given that theory and observations indicate that disc winds are intrinsically inhomogeneous, clumping offers a physically motivated solution. Together, these results provide the first robust, self-consistent demonstration that clumping can reconcile line-driven wind theory with observations across AWDs and AGNs.

Figures (5)

• Figure 1: Time evolution of the wind mass-loss rate, $\dot{M}_{\mathrm{wind}}$, measured at the outer boundary for three models: no clumping ($f_{\rm _V} = 1.0$), moderate clumping ($f_{\rm _V} = 0.1$), and strong clumping ($f_{\rm _V} = 0.01$). For all models, $\dot{M}_{\mathrm{wind}}$ approaches a quasi-steady state by 300 s. The high-density disc region is excluded from the computation of $\dot{M}_{\mathrm{wind}}$.
• Figure 2: Density and poloidal velocity fields at $t = 850 \,\mathrm{s}$ for the three models: no clumping (left; similar to Model A of Mosallanezhad2025), moderate clumping (middle), and strong clumping (right). The colormap shows the logarithmic density, while overlaid velocity vectors—normalized so that the longest corresponds to $v_{p}^{\max} = 3500\,\mathrm{km\,s^{-1}}$—illustrate the flow structure. Grey curves trace streamlines, and the solid white curve marks the Mach 1 surface. Animated versions of each panel are provided in the Supplementary Material.
• Figure 3: Mean ionization state of oxygen for three models: no clumping (left), moderate clumping (middle), and strong clumping (right). Ionization stages are labeled using the standard astronomical convention (neutral = I). The black solid line marks the representative Mach number $M\!=\!1$ (sonic) surface.
• Figure 4: Synthetic UV spectra for a $30^\circ$ inclination, generated from a snapshot of each model using Sirocco. Model spectra are shown for the smooth wind (green), moderate clumping (blue), and strong clumping (red) prescriptions. For comparison, we include the observed ultraviolet spectrum of the archetypal high-state cataclysmic variable RW Sex (grey; HST program 14637, PI: Long), which has a similar inclination ($i \simeq 30^\circ$). All spectra are normalized to a distance of 100 pc. The positions of the Lyman limit and several key UV resonance lines are marked by light-grey vertical dashed lines.
• Figure 5: Angle-averaged spectral energy distributions (SEDs) for three clumping prescriptions. Solid coloured curves show the Emergent Spectrum: green = no clumping ($f_{\rm _V}=1.0$), blue = moderate clumping ($f_{\rm _V}=0.1$), red = strong clumping ($f_{\rm _V}=0.01$). The black curve is the Pure Disc (without irradiation/reprocessing) used as the input multi-temperature blackbody. Vertical markers indicate the H, He i, and He ii ionization edges (labels aligned on a common horizontal level). EUV/UV/Optical bands are indicated by dashed separators with double-headed arrows; an auxiliary top axis shows wavelength in Å. The plotted range is limited to $3\times10^{14}\!-\!3\times10^{16}$ Hz, with $\nu L_{\nu}$ capped at $\le 10^{36}$ erg s$^{-1}$.