Towards physically more comprehensive AGN modelling in cosmological simulations: A MACER-based modification of IllustrisTNG

Abstract

Active galactic nuclei (AGN) feedback plays a significant role in many aspects of galaxy formation and evolution and has become a key ingredient in cosmological simulations. However, the subgrid models of AGN feedback in cosmological simulations such as IllustrisTNG (hereafter TNG) often overlook recent progress in the small-scale modelling of black hole (BH) accretion and AGN physics. In this study, we improve on this by incorporating central aspects of the MACER model, a framework that treats AGN physics in greater detail, into the TNG feedback implementation. Specifically, we adopt MACER-prescriptions for feedback output for high and low accretion rates in a new model while the estimation of the accretion rate remains unchanged. We test this updated scenario both for idealized elliptical galaxies and for a cosmological box. Compared to the original TNG model, the MACER-based simulation leads to a higher star formation rate (SFR) and BH accretion rate in ellipticals, yielding a gas density profile in better agreement with observations. In the cosmological simulations, the time evolution of the SFR density, galaxy stellar mass function at $z=0$, and $M_{\star}-M_{\rm BH}$ relation at $M_{\star}>10^{10.5}\,{\rm M_{\odot}}$ are similar in both models. The MACER model better reproduces low-mass BHs in low-mass galaxies, and yields milder quenching in massive galaxies, although this is accompanied by the absence of a pronounced colour bimodality. Still, the similarity of the outcomes underlines the self-regulated nature of BH feedback: for different feedback energetics, the accretion rate tends to adjust such that a similar total AGN feedback energy is released.

Figures (18)

• Figure 1: Upper left: Feedback efficiency versus Bondi accretion rate in the high–accretion-rate regime (quasar/cold mode) for basic-MACER and TNG. Dotted curves (different colours) show basic-MACER at different $M_{\rm BH}$; the solid curve shows TNG for $M_{\rm BH}\approx2.236\times10^8\,{\rm M_{\odot}}$. The feedback efficiency is defined as $\dot{E}/(\dot M_{\rm Bondi} c^2)$. Upper right: Feedback efficiency at low accretion rates (radio/hot mode): basic-MACER with and without jets (black/gray dotted), and TNG (solid). The definition of feedback efficiency is the same as in the upper left panel. Lower left: Transition accretion rate in units of the Eddington rate between feedback modes as a function of $M_{\rm BH}$ in basic-MACER and TNG. Lower right: Accretion fraction versus accretion rate in basic-MACER and TNG.
• Figure 2: Bondi accretion rate ( upper panel), SFR ( middle panel) and AGN energy output rate ( lower panel)
• Figure 3: Radial profiles of density, temperature and cooling time for the simulated elliptical galaxy in NoAGN, basic-MACER and TNG. The colour of the lines represents the time at which the radial density profile was calculated. The figure shows that the simulated galaxy with the TNG model has a lower central density and longer cooling time compared to the one with the basic-MACER model, indicating that the AGN feedback in the TNG model is significantly stronger than in the basic-MACER model.
• Figure 4: The galaxy stellar mass function at $z=0$ comparing the basic-MACER and TNG models across different resolutions. To illustrate the resolution dependence, higher-resolution results (basic-MACER_HighRes and TNG_HighRes with $2\times512^3$ particles in a $25\,h^{-1}{\rm cMpc}$ box) are shown with thick solid lines, and standard-resolution results with dotted lines. The dark green points and dashed line represent the observational result from baldry12, which is shown for reference.
• Figure 5: Star formation rate density (SFRD) as a function of redshift. The points with error bars are observational SFRD determinations from cucciati12, magnelli13, gruppioni13, and bouwens14abouwens14b. The SFRDs of the basic-MACER and TNG simulations agree quite well with each other and are broadly consistent with the data. There is a more pronounced late-time reduction of the SFRD in the TNG model, reflecting the overall stronger impact of its AGN feedback model.