Preamble

Vol. 43, No. 3

September, 2025

Progress in Astronomy

doi: 10.3969/j.issn.1000-8349.2025.03.06

An AGN Jet Model for the Sgr A Lobes at the Galactic Center

LI Sida1,2, GUO Fulai1,2

(1. Shanghai Astronomical Observatory, Chinese Academy of Sciences, Shanghai 200030, China;
2. School of Astronomy and Space Science, University of Chinese Academy of Sciences, Beijing 100049, China)

Abstract

The Sgr A lobes are a pair of 15-parsec-scale bipolar bubbles oriented perpendicular to the Galactic plane and symmetric about the Galactic center. X-ray observations reveal sharp edges, suggesting they are enclosed by forward shocks driven by an explosive event near the Galactic center. Outflows from central black hole activity provide a viable formation mechanism, making the bubble formation history crucial for understanding Galactic center evolution and high-energy astrophysical processes. Through hydrodynamic simulations, we investigate a model where short-timescale active galactic nucleus (AGN) jets create the bubbles. Our numerical results demonstrate that a 500-year jet episode can reproduce the observed bubble morphology, density, temperature, and X-ray surface brightness. However, current results cannot exclude alternative formation models, such as tidal disruption event outflows. Future multi-wavelength observations will impose tighter constraints on the bubble origin.

Keywords: active galactic nucleus; jets and outflows; bubbles; interstellar medium

1 Introduction

Multiple observational lines of evidence indicate that the Galactic center hosts a supermassive black hole of approximately $4 \times 10^6 M_\odot$ (Sgr A) \cite{1,2}. The evolution of supermassive black holes involves many extreme high-energy physical processes, possibly including outflow phenomena that inject substantial mass and energy into the surrounding environment. Although recent observations suggest that Sgr A is currently in a relatively quiescent state, its past active phases may have significantly influenced the gas distribution in the Galaxy across various temporal and spatial scales. A notable example is the famous Fermi bubbles, which extend to heights of approximately 10 kpc and may have formed from AGN-generated outflows propagating through the Galactic circumgalactic medium (CGM) several million years ago \cite{3--7}.

On smaller spatial scales, recent X-ray and radio observations have discovered a pair of 15 pc tall elliptical bubbles \cite{8--11}. These structures are commonly referred to as the Sgr A lobes. Both Chandra and XMM-Newton telescopes have observed and analyzed the X-ray emission from hot gas in the bubble region, revealing temperatures of 0.7–1.0 keV \cite{8--10}. The X-ray surface brightness and pressure of the hot bubble gas decrease with Galactic latitude, indicating that the outflowing gas is moving away from the Galactic plane. The bubbles also exhibit very sharp boundaries, suggesting that their outer edges may be shocks produced by an outflow. Considering their orientation perpendicular to the Galactic plane and symmetry about the Galactic center, outflows from regions very close to the central black hole represent a plausible mechanism for generating this bubble pair.

Potential outflow phenomena capable of producing such bubble structures fall into two categories: quasi-steady outflows, including stellar winds from stars near the central black hole and past AGN jets; and intermittent explosive events, including tidal disruption events (TDEs) of the central black hole and supernova explosions of individual stars near the Galactic center \cite{12}. Several hydrodynamic simulation studies have analyzed the evolution of supernova explosions in the Galactic center environment \cite{13,14}. Their results show that the energy injected by a single supernova explosion is sufficient to produce bubble structures of approximately 15 pc in size. However, because supernova explosions involve nearly spherically symmetric energy injection, it is difficult to explain the formation of the highly symmetric pair of elliptical Sgr A lobes, even when constrained by other stellar winds or gas disks. Given that the timescale for stellar winds in the Galactic center far exceeds the age of the Sgr A lobes, models where bubbles form through continuous energy injection from stellar winds also face challenges. Black hole tidal disruption events can produce powerful winds and/or jets, injecting sufficient energy into the surrounding environment to form the Sgr A lobes within a short period, making them a possible mechanism for explaining the origin of the Sgr A lobes \cite{15}. No specific studies have yet examined the AGN jet model for these bubbles, so we choose to investigate a model where shocks from jets launched by the central black hole cause the bubbles, similar to models used to explain the famous Galactic Fermi bubbles \cite{6}.

X-ray observations indicate that the internal energy of the Sgr A lobes is on the order of $10^{43}$ J \cite{8,10}. Since the specific outflow velocity is unknown, we cannot determine the precise age of the bubbles, but assuming they expand at the sound speed $c_s \approx 500$ km s$^{-1}$, their age would be approximately $3 \times 10^4$ years \cite{10}, providing an upper limit for constraining model parameters. In theoretical studies, AGN jets typically last for millions of years, but the jet duration required to explain these bubbles is only a few hundred years—far below conventional values. However, recent observational evidence suggests that AGNs can indeed produce short-timescale, low-energy jet phenomena \cite{16}. X-ray observations of molecular clouds toward the Galactic center also indicate that X-ray photons from these clouds may originate from the black hole region, reaching Earth after reflection off molecular clouds, suggesting that the central black hole may have experienced a highly active phase recently, with X-ray luminosity variation timescales far shorter than millions of years \cite{17,18}. X-ray observations have also identified jet structure candidates near the Galactic center black hole \cite{19,20}. These observational lines of evidence all suggest that AGNs can produce the jets required by our model.

Another aspect of the AGN jet model's validity concerns the supply of accretion disk material. The numerous Wolf-Rayet stars in the Galactic center region represent a possible source of matter. These stars are in a phase with very strong stellar winds, typically lasting $10^5$ years. Studies of winds from these Wolf-Rayet stars indicate that they may have played important roles in past activity of the central black hole and influenced the distribution of hot gas within the central parsec \cite{21--23}. Additionally, supermassive black holes can capture material from molecular clouds or massive stars to form accretion disks. In this work, we temporarily assume such an accretion disk exists without discussing the specific material source in detail.

In this study, we use hydrodynamic numerical simulations to investigate the evolution of short-timescale AGN jets in the Galactic center environment and discuss whether they can serve as the formation mechanism for the Sgr A lobes. As a basis for our investigation, we compare the simulated bubbles with the Sgr A lobes in terms of temperature, density, morphology, and X-ray surface brightness distribution. Section 2 describes the basic numerical simulation setup, Section 3 compares simulation results with observations, and Section 4 summarizes and discusses issues related to the AGN jet model.

2.1 Basic Simulation Setup

We assume the entire simulation system is axisymmetric and use the hydrodynamic simulation code ZEUS-MP \cite{24} to solve the equations in two-dimensional spherical coordinates:

$$
\begin{cases}
+ \rho \nabla \cdot \mathbf{v} = 0 \
d\mathbf{t} = -\nabla P - \rho \nabla \Phi + \nabla \cdot (e\mathbf{v}) \
= -P \nabla \cdot \mathbf{v}
\end{cases}
$$

where $\rho$, $\mathbf{v}$, $P$, $\Phi$, $e$, and $t$ represent density, velocity, pressure, gravitational potential, internal energy density, and time, respectively. Since the Sgr A lobes are oriented perpendicular to the Galactic plane and symmetric about the Galactic center, we assume the AGN jet injection is symmetric about the Galaxy's rotation axis. Considering the density and temperature ranges of hot gas in the Galactic center and bubble gas, the radiative cooling timescale would far exceed the bubble age, so we neglect the radiative cooling term in the energy equation. Finally, we assume all gas is ideal, satisfying:

$$P = k_B T \frac{\rho}{\mu m_\mu} = k_B T n$$

where $k_B$ is the Boltzmann constant, $m_\mu$ is the atomic mass unit, $\mu = 0.61$ is the mean molecular weight per particle, and $T$ and $n$ represent gas temperature and number density, respectively.

During the solution process, we establish 600 exponentially growing grids in the $r$ direction and 400 uniform grids in the $\theta$ direction. In the $r$ direction, we place the central black hole Sgr A* at the coordinate origin, with an inner boundary at 0.1 pc and an outer boundary at 30 pc, where $\Delta r_{i+1}/\Delta r_i = 1.007$, meaning each radial grid width is 0.7% larger than the previous one. At the inner boundary, we apply inflow boundary conditions during jet injection and switch to outflow boundary conditions (matching the outer boundary) after jet injection ends. Due to system symmetry, we only need to simulate the evolution of a single jet, so the $\theta$ direction grid spans $0^\circ \sim 90^\circ$, with reflective boundary conditions at both inner and outer boundaries.

2.2 Gravitational Potential and Initial Gas Distribution

Observations indicate the Sgr A lobes have a height of approximately 15 pc. Within 20 pc of the black hole, the gravitational potential is dominated by the black hole itself and the nuclear stellar cluster, so we adopt a constant gravitational field composed of these two components in our simulations. For the black hole potential, we use the Newtonian gravitational field produced by a point source with mass $M_{\rm BH} = 4 \times 10^6 M_\odot$. For the nuclear stellar cluster, we adopt the gravitational potential model from reference \cite{25}:

$$\Phi = \frac{1}{2} v_0^2 \ln(R_c^2 + r^2)$$

where $v_0 = 98.6$ km s$^{-1}$, $R_c = 2$ pc, and $r$ represents the radial coordinate.

In the Galactic center environment, pre-existing diffuse gas and material contributed by stellar winds from the nuclear stellar cluster interact to form the circumnuclear medium (CNM) assumed to exist before jet injection. Current observations cannot provide information about the CNM distribution before bubble formation. For simplicity, we assume this medium was in hydrostatic equilibrium before jet injection, with density following an exponential distribution:

$$n(r) = n_0 \left(\frac{r}{r_0}\right)^{-0.7}$$

where $n_0 = 50$ cm$^{-3}$ and $r_0 = 0.1$ pc. When solving for hydrostatic equilibrium, we assume a gas temperature of $5 \times 10^6$ K at $r = 50$ pc. This density distribution and the resulting temperature distribution are consistent with X-ray observations and Wolf-Rayet star wind simulations in this region within 1 pc \cite{21,23,26}. We note that the exponent in equation (2) lacks firm observational constraints. The initial density distribution directly affects the simulated bubble density and radiation intensity, leading to changes in required jet parameters. In the $0.1 \sim 1.0$ pc range, steady-state solutions assuming Wolf-Rayet star winds give an exponent of approximately $-2$. We adopt $-0.7$ in this work to ensure the simulated bubble gas density matches observed values, representing a weak constraint on gas density in the simulation region. If we used the $-2$ exponent, the bubble density would be significantly lower than observed, and the radiation would be substantially reduced. Therefore, we select the parameter combination in equation (2) such that hot gas density matches observed or wind-solution values in the inner region, while providing sufficient material beyond 1 pc to ensure the resulting bubble density and radiation match observations across most of the bubble region. More stringent constraints on the initial density and temperature distribution await further research.

2.3 Jet Configuration

We implement jet injection using the inflow boundary condition in the ZEUS-MP code. During the jet duration, we set the inner boundary condition in the $r$ direction to inflow and specify jet density, energy density, and velocity in virtual grid cells within the jet half-opening angle. After jet injection ends, we change the inner boundary condition to outflow.

In our primary AGN simulation, we assume the jet direction is perpendicular to the Galactic plane. Black hole jets are related to black hole spin, but the spin orientation of the Galactic center black hole is difficult to determine, which may pose a problem for the AGN jet model. The jet duration is set to 500 years, with a jet half-opening angle of $10^\circ$. Within the $0^\circ \sim 10^\circ$ grid range, we set jet material density $\rho = 5 \times 10^{-24}$ g cm$^{-3}$, temperature $10^8$ K, and constant $r$-direction velocity $1.1 \times 10^{10}$ cm s$^{-1}$.

Jet parameters are selected primarily to satisfy observed properties of the Sgr A lobes. With these settings, the total jet mass is $4 \times 10^{-3} M_\odot$ and total energy is $4.8 \times 10^{43}$ J.

3.1 Bubble Evolution

Density and temperature distributions during the early and middle simulation stages are shown in Figure 1 [FIGURE:1]. After jet injection, a clear shock front forms, with its outer contour corresponding to the bubble boundary in the simulation. During jet injection, jet material undergoes multiple recollimations, forming several recollimation shocks. At $t = 600$ years after jet injection ends, the shock height in the $z$ direction exceeds 10 pc, with the shock front exhibiting a slightly wider bottom and overall elongated shape, similar to most jet simulation results. Without subsequent energy injection, jet material velocity decreases significantly, gradually concentrating at the shock front, after which material continues to flow back toward the bubble bottom. By $t = 1500$ years, a complex low-density region exists in the lower bubble half, formed by multiple complex recirculation events. Overall, although shock height increases only modestly, the proportion of energy used for lateral expansion increases because jet material accumulates at the top with accompanying backflow and thermalization. Consequently, shock width increases noticeably during this period, and the overall morphology gradually approaches the elliptical shape observed for the Sgr A lobes.

When jet evolution reaches $t = 3500$ years, the shock front reaches 15 pc height, with overall morphology closely resembling the observed bubble shape. Figure 2 [FIGURE:2] shows density and temperature distributions of the simulated bubble at this time. The high-density region at the bubble outer contour represents a dense shell formed by shock-compressed hot gas sweeping through the CNM, while the bubble interior forms a high-temperature, low-density cavity.

The simulated bubble contour still differs from the actual observed contour, likely due to uncertainties in jet parameters. Shock width is influenced by multiple factors. For example, in our parameter survey, we found that jets with larger total mass and lower velocity decelerate and thermalize at higher altitudes, producing backflow that broadens the shock front. Large-scale AGN jet studies also indicate that variations in jet power can cause shock width changes \cite{27}.

3.2 Comparison with X-ray Observations

In this section, we calculate thermal X-ray emission from the simulated bubble and compare the resulting X-ray surface brightness with observations to discuss the viability of the AGN jet model as the bubble formation mechanism. We assume hot gas in the simulation is optically thin and in collisional ionization equilibrium, using the astrophysical plasma emission code (APEC) to retrieve X-ray emission coefficients for plasma in the $2 \sim 4.5$ keV energy range \cite{28}. X-ray surface brightness is calculated as:

$$
I(x, z) = \int n_e n_H \epsilon(T, Z) \, dy
$$

where $n_e$ and $n_H$ are electron and hydrogen ion number densities, respectively, and $\epsilon$ is the emission coefficient from APEC, a function of temperature $T$ and metallicity $Z$ (here taken as solar metallicity). To calculate X-ray surface brightness, we select a space of $-10 \sim 10$ pc in the $y$ direction, project fluid density onto a uniform Cartesian grid with 0.05 pc width, and integrate along the $y$ direction as the line of sight. To compare with XMM-Newton observations, we convert X-ray radiation units to "$5 \times 10^{-5}$ s$^{-1}$ pixel$^{-1}$". In this conversion, we adopt XMM-Newton telescope parameters: angular area per pixel of $4'' \times 4''$, and effective area in the $2 \sim 4.5$ keV range approximated as a constant 1000 cm$^2$. Since neutral hydrogen along the line of sight absorbs most X-ray radiation, we assume 75% of photons are absorbed. This absorption fraction corresponds to a neutral hydrogen column density range of $N_{\rm HI} = (5 \sim 7) \times 10^{22}$ cm$^{-2}$ \cite{9}, similar to values used in XMM-Newton observations ($N_{\rm HI} = (6.3 \sim 8.0) \times 10^{22}$ cm$^{-2}$) \cite{8}. Different observational methods yield different neutral hydrogen column density estimates; our adopted value is slightly lower than the column density estimated by Ponti et al. through dust measurements \cite{9}.

Figure 3 [FIGURE:3] shows our calculated two-dimensional X-ray surface brightness distribution, with white highlights representing the Sgr A lobe contours observed by Chandra. Radiation calculated from simulation results is comparable to XMM-Newton observations, and bubble contours agree well with observations. Using the analysis method of Heard and Warwick \cite{8}, we select radiation data within a $20^\circ$ half-opening angle around the $z$ axis, average it to obtain surface brightness variation in the $z$ direction, and compare it with XMM-Newton observations (Figure 4 [FIGURE:4]). The simulated surface brightness decay in the $z$ direction matches observations. Higher observed values in the innermost region likely result from the extreme Galactic center environment, where multiple mechanisms produce X-ray radiation in addition to the Sgr A lobes, leading to observed values exceeding our calculated radiation. Radiation values beyond 12 pc are lower than observed, possibly because XMM-Newton observations include contributions from ambient gas, where observed values represent superposition of radiation from bubble tops and environmental gas.

After calculating radiation values, we can also provide radiation-weighted average temperature and density of the bubble: $T_{\rm ave} = 1.02$ keV and $n_{\rm ave} = 6.23$ cm$^{-3}$, both consistent with X-ray observations \cite{8,10}.

4 Summary and Outlook

In this work, we analyze the possibility of the AGN jet model as a formation mechanism for the Galactic center Sgr A lobes through hydrodynamic numerical simulations. A 500-year AGN jet episode produces a shock front that, by 3500 years of evolution, can reproduce observed bubble properties in terms of morphology, temperature, density, and X-ray surface brightness. Based on current observational information, the AGN jet model can be considered a candidate formation mechanism for the Sgr A lobes.

However, our simulation results alone cannot definitively prove that Sgr A lobe formation originates from AGN jet shock evolution. On one hand, TDE outflows also have the capability to produce bubbles on the scale of the Sgr A lobes. On the other hand, X-ray observations reveal several bright spot structures in the bubbles that are approximately symmetrically distributed about the Galactic center. The specific origin of these radiation structures requires further in-depth analysis, and our thermal X-ray radiation model alone is insufficient to explain them. Additionally, the short-timescale AGN jet model faces an important issue: the required 500-year jet duration in our model is shorter than timescales given by most AGN observations. Although some observations suggest the Galactic center black hole may have experienced a rapidly varying active phase recently, this is insufficient to demonstrate that Sgr A* produced a jet lasting only 500 years in the past. The accretion disk producing the jet can be fed through various channels, such as winds from the series of Wolf-Rayet stars near the Galactic center or partial stripping of material from massive stars captured by the black hole. Assuming 10% energy conversion efficiency (i.e., 10% of accreted material energy converted to jet kinetic energy), the corresponding black hole accretion rate is approximately $10^{-5} M_\odot$/yr, corresponding to an Eddington ratio of $10^{-4}$. Star wind simulations of Wolf-Rayet stars indicate that the Wolf-Rayet phase typically lasts $10^5$ years, and this wind material may have participated in past activity of the central black hole \cite{23}. However, based on current observational evidence, we cannot determine the specific origin of the accretion disk.

References

[1] Genzel R, Eisenhauer F, Gillessen S. Reviews of Modern Physics, 2010, 82: 3121
[2] Morris M, Baganoff F, Muno M, et al. Astronomische Nachrichten Supplement, 2003, 324: 167
[3] Su M, Slatyer T R, Finkbeiner D P. ApJ, 2010, 724: 1044
[4] Guo F, Mathews W G. ApJ, 2012, 756: 181
[5] Mou G, Yuan F, Bu D, et al. ApJ, 2014, 790: 109
[6] Zhang R, Guo F. ApJ, 2020, 894: 117
[7] Yang H Y K, Ruszkowski M, Zweibel E G. Nature Astronomy, 2022, 6: 584
[8] Heard V, Warwick R S. MNRAS, 2013, 434: 1339
[9] Ponti G, Morris M R, Terrier R, et al. MNRAS, 2015, 453: 172
[10] Ponti G, Hofmann F, Churazov E, et al. Nature, 2019, 567: 347
[11] Zhao J H, Morris M R, Goss W M. ApJ, 2016, 817: 171
[12] Markoff S. Proceedings of the National Academy of Science, 2010, 107: 7196
[13] Yalinewich A, Piran T, Sari R. ApJ, 2017, 838: 12
[14] Ehlerová S, Palouš J, Morris M R, et al. A&A, 2022, 668: A124
[15] Li S, Guo F. https://uiadsabsharvardedu/abs/2024arXiv240410205L, 2024
[16] Yang X, Yao S, Gallo L C, et al. ApJ, 2024, 966: 151
[17] Churazov E, Khabibullin I, Ponti G, et al. MNRAS, 2017, 468: 165
[18] Terrier R, Clavel M, Soldi S, et al. A&A, 2018, 612: A102
[19] Li Z, Morris M R, Baganoff F K. ApJ, 2013, 779: 154
[20] Zhu Z, Li Z, Morris M R, et al. ApJ, 2019, 875: 44
[21] Quataert E. ApJ, 2004, 613: 322
[22] Ressler S M, Quataert E, Stone J M. MNRAS, 2018, 478: 3544
[23] Calderón D, Cuadra J, Schartmann M, et al. ApJL, 2020, 888: L2
[24] Hayes J C, Norman M L, Fiedler R A, et al. ApJS, 2006, 165: 188
[25] Stolte A, Ghez A M, Morris M, et al. ApJ, 2008, 675: 1278
[26] Baganoff F K, Maeda Y, Morris M, et al. ApJ, 2003, 591: 891
[27] Whitehead H W, Matthews J H. MNRAS, 2023, 523: 2478
[28] Smith R K, Brickhouse N S, Liedahl D A, et al. ApJL, 2001, 556: L91