Tracing AGN Feedback Power with Cool/Warm Outflow Densities: Predictions and Observational Implications

TL;DR

This paper investigates how AGN-driven winds imprint on the cool, observable gas in galaxy discs by linking the density of cool outflowing clouds to the power of the hidden hot wind. Using high-resolution AREPO simulations with a targeted refinement scheme, the authors show that cool gas densities scale as $n_{ m H}\propto L_{ m AGN}^{1/2}$ and cloud sizes scale as $R_{ m CGC}\propto L_{ m AGN}^{-1/6}$, with a break at $L_{ m AGN}\gtrsim 10^{46}$ erg s$^{-1}$ due to ablation; core densities continue to follow the same scaling. The results imply that observational outflow rates inferred from density tracers can be severely biased if the luminosity dependence of density is ignored, potentially over- or underestimating true mass fluxes by orders of magnitude and flattening the Mdot–$L_{ m AGN}$ relation. By bridging the visible cool phase with the elusive hot wind, the work provides a robust diagnostic of energy-driven feedback and has direct implications for interpreting AGN-driven outflows across the luminosity spectrum.

Abstract

Winds launched at the scale of the accretion disc or dusty torus in Active Galactic Nuclei (AGN) are thought to drive energy-conserving outflows that shape galaxy evolution. The key signature of such outflows, the presence of a hot ($T \gtrsim 10^9 \, \rm K$), shocked wind component, is hard to detect directly. Observations of AGN outflows typically probe a separate outflow phase: cool/warm gas with $T \lesssim 10^5 \, \rm K$. Here, we show that the density of cool outflowing gas scales with AGN luminosity, serving as an indirect diagnostic of the elusive hot, shocked wind. We use hydrodynamic simulations with the moving-mesh code AREPO to target the interaction between a small-scale AGN wind of speed $\approx 10^4 \, \rm km \, s^{-1}$ and galactic discs containing an idealised, clumpy interstellar medium (ISM). Through a new refinement scheme targeting rapidly-cooling, fast-moving gas, our simulations reach a resolution of $\lesssim 0.1 \, \rm pc$ in the cool, outflowing phase. We extract an ensemble of cool clouds from the AGN-driven outflows produced in our simulations, finding that their densities increase systematically with AGN wind power and AGN luminosity. Moreover, the mass distribution and internal properties of these cloudlets appear to be insensitive to the initial properties of the ISM, and shaped mainly by the dynamics of radiative, turbulent mixing layers. The increase in cool outflow density with kinetic wind power and AGN luminosity has profound implications for observational estimates of outflow rates and their scaling with AGN luminosity. Depending on the available outflow and density tracers, observationally-derived outflow rates may be overestimated by orders of magnitude.

Figures (18)

• Figure 1: In this test, the initial condition featured a flow with two distinct fluids—a central layer immersed within a surrounding background fluid—each possessing different densities and moving at different speeds (see Section \ref{['subsubsec:refinement-tests']}). The left panel displays a mass resolution of $10^5 \ {\rm M_\odot}$, while the right panel has a mass resolution of $10^2 \ {\rm M_\odot}$, 1000 times smaller. The central panel implements our additional refinement technique, with a resolution boost factor of $\beta = 1000$ which enhances the resolution in the regions of interest, for the faster and denser gas. The panels show the gas $20 t_{\rm KH}$ after the start, both the High-Resolution and the new refinement technique simulations exhibit the expected ripples resulting from the Kelvin-Helmholtz instability, which are not captured in the Low-Resolution simulation. Notably, the central panel achieved these results in approximately 37% of the time required for the right panel, demonstrating the efficiency of the enhanced refinement scheme.
• Figure 2: Density (top) and velocity (bottom) distribution for the Low-Resolution (dark red line), High-Resolution (black line) and the intermediate simulation using Our Method (grey area). Our Method presents a similar distribution to the high-resolution simulation, for the fast and dense gas. Overall, the shape of the grey area, resembles the black line (high-resolution) more than the red line (Low-Resolution).
• Figure 3: Total number of cells for the three Kelvin-Helmholtz test simulations. Our Method refines cells extensively after $t > t_{\rm KH} \approx 0.32$ Myr, rapidly increasing the total cell number, roughly matching the High-Resolution simulation value. By $15 \ t_{\rm KH}$, simulations performed with Our Method have about half the number of cells in the High-Resolution simulation.
• Figure 4: The top panel shows a density-weighted projection with a depth of 100 pc, centred on a massive outflowing cool gas cloud (CGC) with $M_{\rm CGC} \sim 10^{5} \ {\rm M_\odot}$, extracted from simulation L45_R64. This density map highlights its complex internal structure, with an internal density ranging from $1$ cm$^{-3}$ to $3 \times 10^{4}$ cm$^{-3}$. For the same CGC, we show the density distribution (second row), cumulative mass distribution (third row), and cumulative volume distribution (last row). The density distribution follows an approximately log-normal distribution (green dash-dotted curve). The dotted blue curve shows the initial cool gas distribution, for comparison. In the third and fourth panels, the coloured area highlights the CGC cores, comprising the top 10% densest cells that account for more than 25% of the total CGC mass comprised in less than 7% of the total CGC volume.
• Figure 5: Density-weighted projections showing gas pressure (red/orange colour) and density (blue/green colour) for simulations with $L_{\rm AGN} \, = \, 10^{44} \, \rm erg \, s^{-1}$ (bottom, left section), $L_{\rm AGN} \, = \, 10^{45} \, \rm erg \, s^{-1}$ (bottom, right section) and $L_{\rm AGN} \, = \, 10^{46} \, \rm erg \, s^{-1}$ (top section). The circular inset plots surrounding the central panel show density projections that zoom-in on some of the outflowing clouds found in these simulations. As the AGN luminosity and wind kinetic power increase, outflowing clouds become denser as their surrounding pressure grows, and shatter into smaller cloudlets. The spatial scale is the same for all inset plots.