The dominant role of dark matter halo in quenching central galaxies

Abstract

Understanding the quenching of star formation in central galaxies remains a core challenge in galaxy evolution. Two decades ago, the concept of halo quenching was introduced as a dominant mechanism, positing that massive central galaxy quenching is governed by the thermodynamics of gas predominantly influenced by dark matter halos. However, a vastly increasing body of observational evidence consistently indicates that quenching correlates predominantly with central properties like velocity dispersion, bulge mass, and black hole mass. When these properties are controlled, halo mass appears to show weak influence, supporting AGN feedback as the primary mechanism. A recurring key issue, however, is that these studies rely on halo masses derived via abundance matching (AM). Direct observational measurements from weak lensing, satellite kinematics, and galactic dynamics reveal that AM systematically overestimates halo masses of star-forming centrals while underestimating those of passive ones. To accurately assess the true role of halo mass. we re-estimated halo masses for SDSS groups; the resulting halo mass function and stellar-to-halo mass relations (SHMRs) for both populations match theoretical predictions and weak lensing measurements. Using these improved masses, we find direct observational evidence that halo mass is the dominant factor in quenching central galaxies, with a clear threshold at $M_{h}\sim 10^{12.1}M_\odot; $. By applying a simple correction to AM data using weak lensing-derived SHMRs, we demonstrate that previous claims regarding the dominance of central properties stem primarily from systematic biases in AM halo masses. Our results suggest that the significance of AGN feedback is primarily manifested in halos above this mass threshold, in galaxies already primed for quenching. In other words, AGN feedback appears to become effective in halos above this mass threshold.

Figures (4)

• Figure 1: The important role of halo mass in central galaxy quenching. The median sSFR (left) and $\Delta$MS (right) of central galaxies are shown as a function of halo mass and central velocity dispersion. Halo masses are estimated using the machine learning (ML) method described in Zhao et al. 2025ApJ...979...42Z. Colors indicate the median value shown in each panel, and the black contours trace iso-levels of the same quantity. The vertical and horizontal green lines ($M_\mathrm{h} \sim 10^{12.1} M_\odot$ and $V_{\rm disp} \sim 85\, \rm km\,s^{-1}$) define a characteristic L-shaped threshold. Crucially, galaxies with halo masses below this threshold remain star-forming regardless of their central velocity dispersion, whereas quenching dominates only in massive halos coupled with high central velocity dispersion.
• Figure 1: Robustness of the halo mass threshold. The median sSFR is plotted as a function of halo mass and gravitational potential (left) or bulge mass (right), analogous to the primary analysis in Fig. \ref{['fig:ML_halo_mass']}. The vertical green line indicates the same halo mass threshold ($M_\mathrm{h} \sim 10^{12.1} M_\odot$) identified in the main text.
• Figure 2: Comparison of sSFR variations using AM-derived halo masses. Analogous to the left panel of Fig. \ref{['fig:ML_halo_mass']}, but using halo masses derived via the AM method based on total stellar mass ranking (left) and total $r$-band luminosity ranking (right) 2007ApJ...671..153Y. The magenta line at $V_{\rm disp}\sim125~\rm km\,s^{-1}$ marks the apparent quenching threshold reported in previous studies. The green lines correspond to the L-shaped threshold identified in Fig. \ref{['fig:ML_halo_mass']}. Data are restricted to galaxies with $M_\mathrm{h} > 10^{11.8}M_\odot$, corresponding to the low-mass cutoff of the AM halo-mass estimates in the group catalogue.
• Figure 3: Origin of the discrepancy: systematic bias in abundance matching.Left: SHMRs for star-forming (blue) and passive (red) galaxies. Solid and dashed lines represent results from our ML method 2025ApJ...979...42Z and the AM method with total stellar mass ranking 2007ApJ...671..153Y, respectively. The upturn of the AM results (dashed lines) at the low-mass end is an artifact caused by the halo mass cutoff at $M_\mathrm{h} \sim 10^{11.8}M_\odot$. Data points with error bars represent independent weak lensing measurements from Mandelbaum et al. 2016MNRAS.457.3200M and Bilicki et al. 2021AA...653A..82B. Shaded regions and error bars indicate the 16--84th percentile ranges. Right: The median sSFR distribution re-calculated using AM halo masses after applying a systematic correction ($\pm 0.15$ dex) derived from the weak lensing results. This simple correction largely recovers the L-shaped threshold (green lines) observed in Fig. \ref{['fig:ML_halo_mass']}.