Clues on black hole feedback from simulated and observed X-ray properties of elliptical galaxies

The centers of elliptical galaxies host supermassive black holes that significantly affect the surrounding interstellar medium through feedback resulting from the accretion process. The evolution of this gas and of the nuclear emission during the galaxies’ lifetime has been studied recently with high-resolution hydrodynamical simulations. These included gas cooling and heating specific for an average AGN spectral energy distribution, a radiative efficiency declining at low mass accretion rates, and mechanical coupling between the hot gas and AGN winds. Here we present a short summary of the observational properties resulting from the simulations, focussing on 1) the nuclear luminosity; 2) the global luminosity and temperature of the hot gas; 3) its temperature profile and X-ray brightness profile. These properties are compared with those of galaxies of the local universe, pointing out the successes of the adopted feedback and the needs for new input in the simulations.

💡 Research Summary

This paper investigates how feedback from supermassive black holes (SMBHs) in elliptical galaxies shapes the X‑ray observable properties of both the nuclear source and the surrounding hot interstellar medium (ISM). The authors employ high‑resolution one‑dimensional hydrodynamical simulations that self‑consistently treat stellar mass loss, Type Ia supernova heating, radiative cooling, and both radiative and mechanical AGN feedback. Radiative feedback is based on an average quasar spectral energy distribution; the radiative efficiency ε varies with the Eddington‑scaled accretion rate ˙m, reaching ε≈0.1 at high ˙m and declining as ε≈10 ˙m for ˙m≲0.01, mimicking radiatively inefficient accretion flows (RIAFs). Mechanical feedback is modeled as a broad‑line region (BLR) wind whose kinetic efficiency scales with the Eddington ratio, peaking at a few ×10⁻³ for l≈2. Jet feedback is not included in the present runs but is noted as a future improvement.

A representative galaxy is chosen with stellar velocity dispersion σ=260 km s⁻¹, B‑band luminosity L_B=5×10¹⁰ L_⊙, stellar mass‑to‑light ratio M_*/L_B=5.8, effective radius R_e=6.9 kpc, and a dark‑matter halo that supplies an equal mass to the stars within R_e. The computational grid extends from 2.5 pc to 250 kpc with logarithmic spacing. Simulations start at an age of ~2 Gyr (z≈2) and follow the evolution for ~10 Gyr, during which multiple AGN outbursts occur.

Key results:

1. Nuclear Luminosity Evolution – Early in the simulation the SMBH undergoes brief, near‑Eddington outbursts driven by the collapse of a dense cold shell formed from stellar mass loss. As the galaxy ages, the stellar mass‑loss rate declines, the replenishment time lengthens, and the average accretion rate drops to ˙M≈0.01 M_⊙ yr⁻¹ (˙m≈1.1×10⁻³). The radiative efficiency then settles at ε≈0.01, yielding a bolometric nuclear luminosity L_bol≈2×10⁴³ erg s⁻¹ and an Eddington ratio l≈2×10⁻⁴ at the present epoch (~12 Gyr). Compared with the Palomar spectroscopic survey and Chandra observations of nearby ellipticals, this luminosity is still higher than typical LINER/LLAGN nuclei (L_X,nuc≈10³⁸–10⁴² erg s⁻¹, L_X,nuc/L_Edd≈10⁻⁶–10⁻⁸). The authors suggest that additional mechanical feedback (e.g., jets or nuclear winds) would be required to further suppress the accretion rate.

2. Global Hot‑Gas X‑ray Emission – The total ISM X‑ray luminosity L_X,ISM peaks during outbursts and then declines, reaching ≈10⁴⁰ erg s⁻¹ at the present epoch. This lies at the low end of the ROSAT‑derived distribution for galaxies with the same L_B (10⁴⁰–10⁴² erg s⁻¹). The authors interpret this as overly efficient gas “degassing” in the simulations, possibly because the model galaxy is isolated. They propose that confinement by an external intra‑group/cluster medium or a reduction in the assumed SNIa heating could raise the gas content to observed levels.

3. Temperature Evolution – Emission‑weighted temperatures in the soft (0.3–2 keV) and hard (2–8 keV) bands evolve together, with hard temperatures briefly spiking (up to several keV) at the onset of an outburst when a very hot central bubble forms. The soft‑band temperature averages ≈0.5–0.7 keV within R_e, consistent with Chandra measurements for galaxies of similar σ (≈260 km s⁻¹). The model reproduces the observed correlation between σ and gas temperature, though the simulated temperature is slightly lower than the best‑fit relation (T≈0.7 keV).

4. Radial Temperature Profiles – During quiescent periods the projected temperature declines monotonically from ≈1 keV at ~100 pc to ≈0.4–0.5 keV at ~20 kpc, a negative gradient typical of inflowing gas in steep potentials. This matches the majority of observed profiles from Chandra, where many ellipticals show a factor‑two drop from the centre to the outskirts. A few observed systems display central temperature peaks (~1 keV), which are only weakly reproduced in the simulations.

5. Surface Brightness Profiles – The X‑ray surface brightness rises sharply toward the centre and flattens around 5–10 kpc, resembling the “core–cusp” morphology seen in high‑resolution Chandra images. Because the simulations lack an external confining medium, the overall brightness is lower than in many observed galaxies, especially those embedded in groups or clusters.

Overall, the study demonstrates that the adopted AGN feedback scheme successfully prevents runaway cooling flows, keeps the SMBH mass within the observed range, and yields temperature and surface‑brightness gradients broadly compatible with observations. However, the nuclear X‑ray luminosity remains higher than typical low‑luminosity AGN, and the hot‑gas content is under‑abundant, indicating that additional mechanical feedback (jets), environmental confinement, or adjustments to SNIa heating are required for a more faithful reproduction of the local elliptical galaxy population.

The authors conclude that future work should incorporate jet feedback, explore a wider range of mechanical‑feedback efficiencies, and model galaxies within realistic group/cluster environments to resolve the remaining discrepancies.