The role of galactic winds fueling central starbursts and quasars in the FIRE cosmological simulations

TL;DR

This study uses FIRE-2 cosmological simulations and a novel PReS particle-tracking algorithm to propose a wind pileup fueling mechanism: the last major stellar-feedback outflow evacuates the inner galaxy and piles up gas in the inner CGM, which later reaccretes coherently to drive extreme inflows into the nucleus. The authors quantify how pileup gas can supply >50% of central gas within tens of Myr after reformation, with inflow rates reaching $\sim$10$^2$–10$^3$ $M_{\odot}$ yr$^{-1}$ down to $<100$ pc and central densities exceeding $10^{11} M_{\odot}$ kpc$^{-3}$. Angular momentum evolution appears largely unaffected by pileup, suggesting fueling is dominated by reservoir buildup and rapid delivery rather than angular-momentum loss. The mechanism provides a natural prelude to rapid BH growth and AGN feedback at cosmic noon, and offers observable predictions (cavity/shell signatures, central gas compaction, and post-outflow inflows) testable by JWST and ALMA observations.

Abstract

Central starbursts and Active Galactic Nuclei (AGN) are thought to be fueled by either galaxy interactions or secular processes in gravitationally unstable discs. We employ cosmological hydrodynamic simulations from the Feedback in Realistic Environments (FIRE) project to propose a new nuclear fueling scenario based on the transition that galaxies undergo from bursty to smooth star formation and from prominent global galactic winds to inefficient stellar feedback as they grow above $M_{\star}\sim 10^{10-10.5}\,{\rm M}_{\odot}$: the last major galactic wind event shuts down star formation, evacuates gas from the galaxy, and slows down gas accretion from the circumgalactic medium (CGM), creating a $\sim$$10^{10}\,{\rm M}_{\odot}$ pileup of gas in the inner CGM which later accretes coherently onto the galaxy, achieving a tenfold increase in inflow rate over pre-outflow conditions. We explicitly track the accumulation of gas along the outflow pathway owing to hydrodynamic interactions and show that $\sim$50% of gas fueling the central $\sim$10-100$\,{\rm pc}$ over the subsequent $\sim$15$\,{\rm Myr}$ can be traced back to pileup gas having experienced $>$50% change in infall velocity owing to the wind interaction. This galactic wind pileup effect may thus represent a significant fueling mode for compact starbursts and luminous AGN. Galactic winds at earlier times or AGN-driven outflows can have qualitatively similar effects, but the pileup of gas driven by the last major galactic wind event refuels the galaxy precisely when the deepening stellar potential prevents further gas evacuation by stellar feedback, providing the ideal conditions for quasar fueling at the time when AGN feedback is most needed to regulate central star formation in massive galaxies at their peak of activity.

Figures (11)

• Figure 1: Flowchart illustrating the PReS algorithm for identifying gas particles affected by the galactic wind pileup effect, including the decision-making criteria and the sequential steps involved. The initial set of "pileup particles" is defined as all gas within the central 2 kpc of the galaxy in the snapshot before the last global galactic outflow ($\Delta t \equiv -20\,{\rm Myr}$), corresponding to gas fully ejected by the time ($\Delta t \equiv 0\,{\rm Myr}$) at which the global galactic wind is first identified. Surrounding gas elements can be flagged in subsequent snapshots by interacting hydrodynamically with previously identified pileup particles. See text for details.
• Figure 2: Normalized probability distribution of radial velocity change measured at $\Delta t\sim 0\,{\rm Myr}$ for (1) all unaffected gas within $R_{\rm stall} < R \leq 40\,{\rm kpc}$ (grey), (2) pre-selected gas identified inside of the smoothing kernel of pileup gas (light blue), and (3) newly flagged pileup gas corresponding to different velocity thresholds (coloured lines). For pre-selected and pileup gas, velocity changes are calculated over the corresponding interaction time $\tau_{\rm int}$ for each particle ($0.2\,{\rm Myr} \lesssim \tau_{\rm int} \lesssim 5\,{\rm Myr}$). For comparison, velocity changes of unaffected gas are computed using $\tau_{\rm int} = 2\,{\rm Myr}$. Particles near pileup gas (light blue) exhibit larger changes in radial velocity on average compared to the background gas distribution (grey). Additionally, flagged pileup gas shows larger positive radial velocity changes with increasing velocity threshold, with average changes ranging from $\sim$$100\,{\rm km}\,{\rm s}^{-1}$ ($\beta =0.1$) to $\sim$$300\,{\rm km}\,{\rm s}^{-1}$ ($\beta=4$) and maximum velocity changes reaching $\sim$$2000\,{\rm km}\,{\rm s}^{-1}$.
• Figure 3: Projected face-on gas mass surface density distribution and kinematics of gas in a 16 kpc-wide region around a simulated massive, star-forming galaxy ($M_{\rm star} \sim 10^{10.5}\,{\rm M}_{\odot}$, ${\rm SFR} \sim 300\,{\rm M}_{\odot}\,{\rm yr}^{-1}$) at $z \sim 2.28$ over $\sim$70 Myr. Inflowing gas is represented in blue hues, outflowing gas is represented in red hues, and the colour value and saturation correspond to the gas surface density logarithmically scaled. The last major galactic-scale outflow driven by stellar feedback fully evacuates the ISM within 1 kpc at $\Delta t\equiv 0$ Myr, expanding to larger scales while interacting with inflowing gas in the inner CGM (see larger cavity at $\Delta t=10$ Myr). Once the outflow stalls (at $\Delta t\sim 20$ Myr), the inflow component eventually dominates, and the deepening stellar potential enables the formation of a compact, high surface density gas disc at $\Delta t \sim 50$ Myr which can no longer be disrupted by stellar feedback owing to gravitational confinement of winds.
• Figure 4: Top: Radial profiles of the gas density, where the colour scale indicates time evolution within each phase and the black line is the radial density profile before the outflow event (at $\Delta t = -20$ Myr; reproduced in the three panels for reference). Bottom: Cavity radius as a function of time, following the expansion of the kpc-scale outflow up to the stalling radius $R_{\rm stall}\sim 4$ kpc (" Buildup" phase; red), the subsequent inflow of gas and reformation of the gas disc in the galaxy (" Infall" phase; blue), and the ultimate fueling of the nuclear region (" Fueling" phase; green). The dashed horizontal line denotes $R_{\rm stall}$. The x markers denote the times that are analysed in more detail in Figures \ref{['fig:flagged_maps']}, \ref{['fig:decomposed_densityradprofiles']}, and \ref{['fig:decomposed_flowradprofiles']}. The nuclear gas density on $<$100 pc scales during the Fueling phase greatly exceeds that of pre-outflow conditions.
• Figure 5: Expansion of the global galactic wind and interaction with the infalling CGM, creating a pileup of gas that later accretes onto the galaxy and enables the formation of an ultra-dense nuclear gas disc. The middle panels show the projected gas mass surface density distribution (background grey scale) in the central 60 kpc region over a 60 Myr period since the onset of the last major galactic outflow. The propagation of galactic wind pileup gas is indicated by the purple-to-yellow colour scale, corresponding to gas particles flagged by the PReS algorithm as having been impacted by the pileup effect (in this case for $\beta=0.5$, i.e. experiencing $>$50% change in radial velocity). The top and bottom rows show close-up views of the central 10 kpc region at $\Delta t = 0$ Myr and $\Delta t=60$ Myr, respectively, in both the face-on and edge-on projections, showing that a substantial amount of gas hydrodynamically impacted by the wind contributes to the sub-kpc nuclear gas reservoir at later times.