Fermi/eROSITA Bubbles in the Milky Way

• Fermi/eROSITA Bubbles are vast, bilobate high-energy structures extending tens of kiloparsecs from the Galactic Centre, with sharp γ-ray and X-ray edges.
• They exhibit multiple formation scenarios—from starburst winds and AGN jets to tidal disruption events and persistent cosmic-ray driven outflows—each with distinct energetic signatures.
• Multiwavelength analyses reveal in-situ particle acceleration at the bubble shells, explaining the hard spectral profiles and shock properties observed across the structures.

The Fermi/eROSITA Bubbles are a pair of giant, bilobate structures extending up to tens of kiloparsecs above and below the Galactic Centre (GC) of the Milky Way. They are observed as hard-spectrum γ-ray emission (the Fermi Bubbles) detected by the Fermi-LAT telescope, and as sharply bounded soft X-ray shells (the eROSITA Bubbles) revealed by the eROSITA X-ray observatory. These features, highly symmetric about the Galactic rotation axis and plane, have become the focus of intense research due to their implications for feedback processes, cosmic ray (CR) transport, and the energetics of the circumgalactic medium (CGM) (Sarkar, 14 Mar 2024, Predehl et al., 2020, Tank et al., 10 Nov 2025). While their ultimate origin—whether in star formation-driven winds, AGN outbursts, tidal disruption events (TDEs), or persistent galactic winds—remains debated, recent high-resolution, multi-wavelength analyses and simulation campaigns have elucidated critical aspects of their structure, energetics, physical conditions, and formation pathways.

1. Morphology, Multiwavelength Structure, and Spatial Extent

The Fermi Bubbles (FBs) are characterized by sharp-edged γ-ray lobes, each subtending approximately ±50° in Galactic latitude (corresponding to ~10 kpc at 8 kpc distance) and ±20° in longitude, with nearly flat surface brightness and pronounced symmetry about the disk and rotation axis (Sarkar, 14 Mar 2024, Predehl et al., 2020). The eROSITA Bubbles (EBs) enclose the FBs, extending to |b| ≃ 80°, i.e. ~14 kpc above and below the plane, and show bright, thick-shell geometry in X-rays, with narrow shell thickness compared to their extent (outer radius ~7 kpc, shell thickness ~2 kpc). This nested, bilobate morphology is robustly axisymmetric and persists even when the GC jet axis is not strictly perpendicular to the disk; hydrodynamic simulations of "failed" tilted jets, which are disrupted by the central ISM, demonstrate that buoyancy in the stratified halo can produce vertical, symmetric bubbles regardless of initial jet inclination (Tseng et al., 31 May 2024).

The edges of the EBs are marked by forward shocks, traced by abrupt enhancements in soft X-ray brightness, while the FBs are bounded by a contact discontinuity inside the shocked shell. Radio and microwave observations detect associated nonthermal emission ("microwave haze") with matching spatial profiles, and polarized radio lobes extending tens of degrees beyond the γ-ray rims are attributed to aged CR electron populations (Sarkar, 14 Mar 2024).

2. Physical Conditions, Energetics, and Emission Mechanisms

The EBs' shells have post-shock electron temperatures kT ≈ 0.2–0.3 keV and electron densities n_e ≃ 2–4×10⁻³ cm⁻³ (Predehl et al., 2020). The implied shock Mach numbers are M ≈ 1.5–2, corresponding to shock velocities v_s ≃ 340–1000 km/s. The total thermal energy content of each shell is ~10⁵⁶ erg, while the nonthermal (CR) energy in the FBs is an order of magnitude lower, E_FB ≃ 10⁵⁵ erg. Cooling times for the hot shell (t_cool ≃ 2×10⁸ yr) are much longer than their inferred dynamical ages (~20 Myr), so the shocks are non-radiative (Predehl et al., 2020).

The FB γ-ray emission can be explained through either leptonic (inverse-Compton up-scattering by CR electrons) or hadronic (π⁰ decay from CR protons) processes. In leptonic models, CR electrons must have a hard power-law distribution (N(γ) ∝ γ^–s with s ≈ 2.1) with cutoff E_cut ~1 TeV, and the observed hard γ-ray spectrum (dN/dE ∝ E^–2.1) persists without significant latitude-dependent softening except at the very highest latitudes, where spectral hardening is seen [(Yang et al., 2014); (Tank et al., 10 Nov 2025)]. This limb hardening is indicative of in-situ particle acceleration, likely at the turbulent shell or at weak shocks. For a purely hadronic scenario, the required CR proton energy and cutoff are similar, but these models generally fail to reproduce the associated microwave haze without invoking an additional primary electron population (Yang et al., 2018, Tank et al., 10 Nov 2025).

The eROSITA Bubbles' X-ray emission is consistent with an optically thin, thermal plasma of subsolar metallicity (Z ≈ 0.2–0.7 Z_⊙), with integrated luminosity L_X ≃ 10³⁹ erg/s (Predehl et al., 2020, Sacchi et al., 30 Jan 2025). Surface brightness profiles and sharp limbs favor a scenario where the EBs trace the forward shocks of expanding superbubbles, and the FBs the inner contact discontinuity.

3. Spatio-Spectral Morphology and Implications for Cosmic-Ray Populations

A template-free, pixel-by-pixel spectral analysis of the Fermi Bubbles demonstrates that both hadronic and leptonic models can formally fit the spatially resolved γ-ray data, provided the parent CR populations follow either broken power-law (BPL) or exponential cutoff (EPL) spectra (Tank et al., 10 Nov 2025). In the BPL case, below the break energy (~100 GeV), the CR spectral indices harden (Δα ≈ 0.2–0.5) towards the bubble edges, with high-latitude caps exhibiting the hardest spectra. The amplitude of the CR electron population required in leptonic fits increases towards the edges, yielding a "limb-brightened" spatial profile for the electron energy density u_e(r), peaking at >1 eV/cm³ near the shell and closely correlated with spectral hardening.

Critically, the cooling time for TeV-scale electrons (~0.5–1 Myr for IC losses) is orders of magnitude shorter than plausible advection times from the GC, strongly disfavoring simple advective transport as the source of these high-energy electrons. Instead, distributed in-situ acceleration—either by turbulence or weak shocks at the shell boundary—is required to maintain the observed electron spectrum at the periphery. For protons, the much longer pp-loss timescales (t_pp ~10¹⁰ yr) allow accumulation over Gyr intervals, but the observed spatially variable spectral indices are not naturally reproduced without postulating strong confinement or non-steady injection (Tank et al., 10 Nov 2025).

4. Formation Scenarios: Starburst, AGN, Tidal Disruption, or Persistent Winds

Extensive multiwavelength modeling and numerical simulations support several physically plausible formation scenarios, each with distinguishing features and constraints:

• Star Formation/Starburst Wind: Mechanical feedback from SNe and stellar winds in a nuclear starburst, with SFR ≳ 0.3–0.5 M_⊙/yr over 20–40 Myr, can deliver a time-averaged mechanical power L_wind ≈ (5–7)×10⁴⁰ erg/s, sufficient to drive a Mach ~1.5 shock to ~14 kpc in ~30 Myr. Such winds reconcile the metallicity, bubble morphology, shell expansion velocities, and total energy (Sarkar, 14 Mar 2024, Sacchi et al., 30 Jan 2025, Shimoda et al., 27 Mar 2024).
• AGN Jet or Accretion Wind: An Eddington-level accretion episode (duration 0.1–2 Myr, power L_jet ∼ 10⁴⁴ erg/s) can account for both the required energetics and the short dynamical ages suggested by the sharp edges and uniform spectrum. Simulations demonstrate that even jets misaligned by ≤45° with respect to the disk will, after initial ISM disruption, form vertically oriented, axisymmetric bubbles via buoyancy (Tseng et al., 31 May 2024, Sarkar et al., 2022). Double-episode AGN jet models, with two successive outbursts separated by ~10 Myr, can explicitly accommodate the nested eROSITA (older, larger, cooler) and Fermi (younger, smaller, hotter) bubbles (Zhang et al., 18 Jul 2025).
• Tidal Disruption Event (TDE) Outflow: Regular TDEs occurring every 10–100 kyr, each injecting ≃10⁵³ erg, over a ~16 Myr epoch, result in a total injected energy ~1.5×10⁵⁶ erg and can naturally generate both the observed X-ray EBs (via the forward shock) and FBs (via turbulent acceleration and hadronic CR interactions at the contact discontinuity). The synthetic X-ray/gamma-ray maps, sizes, ages, Mach numbers (M ~2–3), and surface brightness profiles quantitatively match the observations (Scheffler et al., 30 Jan 2025, Dogiel et al., 22 Nov 2024).
• Persistent, CR-Driven Galactic Wind: An alternative model posits the bubbles as quasi-steady, persistent structures maintained by the long-term star formation rate (SFR ~3 M_⊙/yr) of the Galaxy over ≥1 Gyr, with cosmic-ray pressure and Alfvénic heating supporting fountain-like outflows to ~10–14 kpc. Such winds provide a natural mechanism for metal removal (maintaining disk metallicity), produce the correct X-ray/gamma-ray surface brightness, and explain the nested, long-lived morphology without requiring fine-tuned, short-lived outbursts (Shimoda et al., 27 Mar 2024).
• Hybrid/Turbulent Acceleration Models: Rayleigh–Taylor (RT) instability and hydrodynamic turbulence at the expanding shell naturally generate Kolmogorov‐type turbulence, feeding the acceleration of CRs (second‐order Fermi process) and supporting a hard, uniform γ-ray spectrum across the FB volume (Dogiel et al., 22 Nov 2024). The smoothness of the observed rims implies the current age (<20 Myr) is well below the fragmentation timescale for RT instability in a β-model halo (Schulreich et al., 2021).

5. Constraints from Simulations and Extragalactic Comparisons

Cosmological simulations (e.g., TNG50) of MW-mass disk galaxies routinely produce X-ray/γ-ray bubbles with physical, morphological, and energetic properties similar to those observed in the Milky Way, driven by episodic, low-accretion SMBH feedback. In these simulations, the characteristic energy injection per bubble event is 10⁵⁶–10⁵⁸ erg, kinetic power 10⁴¹–10⁴³ erg/s, and expansion velocities 100–2000 km/s. Such bubbles possess Mach numbers M ~2–4, X-ray luminosities L_X ~10³⁹–10⁴⁰ erg/s, and metallicity Z ~0.5–2 Z_⊙, closely matching empirical measurements. However, these models also predict larger (≳50 kpc) AGN-driven bubbles in high-M_BH galaxies, whereas the MW's bubbles are confined to ~14 kpc, a scale that is replicated only in galaxies with high SFR-driven anisotropy (Pillepich et al., 2021, Sacchi et al., 30 Jan 2025).

Stacked Chandra analyses of external edge-on disk galaxies reveal that significant soft X-ray (0.3–2 keV) surface brightness enhancements, extending to ~14 kpc, are present only in systems with high SFR, consistent with a starburst-driven origin for MW-like bubbles (Sacchi et al., 30 Jan 2025). However, AGN-driven outflows, TDEs, or low-luminosity magnetically dominated jets remain viable for lower SFR or more massive bulges as alternative energy sources.

Model Class	Energy Source	Power/Duration	Bubble Scale/Morphology	Key Constraints
Starburst/SN Wind (Sarkar, 14 Mar 2024, Sacchi et al., 30 Jan 2025)	SNe/starburst	10⁴⁰–10⁴¹ erg/s / 20–40 Myr	14 kpc, axisymmetric, X-ray/γ-ray edges	SFR, metal loading, timescale
AGN Jet/Accretion Wind (Zhang et al., 18 Jul 2025, Tseng et al., 31 May 2024)	SMBH jet/accretion	10⁴⁴ erg/s / 0.1–2 Myr	Nested, sharp-edged, symmetric lobes	Jet alignment, timing, OVIII/OVII ratio
TDE-driven Outflow (Scheffler et al., 30 Jan 2025, Dogiel et al., 22 Nov 2024)	GC Tidal Disruptions	3×10⁴¹ erg/s / 16 Myr	Cocoon with forward shock (EB), inner contact (FB)	TDE rate, energy coupling
Persistent Wind (Shimoda et al., 27 Mar 2024)	CR-driven Galactic wind	10⁴⁰ erg/s / ≥1 Gyr	Long-lived, quasi-steady multi-scale	SFR history, metal ejection
Hybrid/Turbulent (Tank et al., 10 Nov 2025, Dogiel et al., 22 Nov 2024)	Turbulent shell accel.	–	In-situ CR reacceleration at shell edges	Spectral hardening, limb energetics

6. Key Observational Tests, Open Questions, and Prospects

Multiwavelength observational constraints sharpen several remaining issues:

• The measured O VIII/O VII X-ray line ratios and the temperature of the shock-compressed shell (kT ~0.2–0.3 keV, Mach ≲2) disfavor blastwaves from super-Eddington jet outbursts (which would yield kT ≳1 keV) (Sarkar et al., 2022).
• UV and radio spectroscopy of embedded clouds in the Bubbles determines outflow velocities (≳300 km/s), kinematic ages (6–9 Myr), chemical abundances (Z ≳ 0.5 Z_⊙), and mass outflow rates (≲0.5 M_⊙/yr), supporting both starburst and AGN scenarios but arguing for a relatively recent, short-duration energy injection (Fox et al., 2019).
• The uniformity and sharpness of the rims, both in X-rays and γ-rays, as well as the spatially resolved spectral hardening at high latitudes, require ongoing in-situ CR acceleration efficiently distributed at the shell boundary (Tank et al., 10 Nov 2025).
• The nested, two-shell morphology—with eROSITA Bubbles fully enclosing Fermi Bubbles—places important chronological and dynamical constraints; two-episode jet or wind models can naturally explain the observed structures (Zhang et al., 18 Jul 2025).

Open questions persist regarding the coupling between the GC energy source (either AGN, SFR, or TDEs) and the large-scale CGM, the transport/acceleration mechanisms for relativistic particles, the origin of metallicity gradients, and the relation between MW bubbles and those found in external galaxies. Future high-resolution X-ray (Athena/XIFU, XRISM), UV (LUVOIR), and radio (SKA) facilities will provide decisive tests, especially by measuring velocity fields, shock structure, ionization states, and detailed CR distribution in the bubbles (Sacchi et al., 30 Jan 2025, Fox et al., 2019).

7. Synthesis: Unified Physical Picture of the Fermi/eROSITA Bubbles

The accumulated evidence supports a model in which the Fermi/eROSITA Bubbles are the relics of a powerful, weakly supersonic outflow launched from the Milky Way's central few hundred parsecs within the last 10–20 Myr. The shocks now seen as eROSITA shells are driven into the ambient CGM, while the contact discontinuities bounding the interior identify the Fermi Bubbles. The observed γ-ray and multi-frequency synchrotron emission (microwave haze) arise from distributed acceleration of CR electrons, predominantly in-situ at the boundary, by turbulence or mild shocks. While the precise roles of star formation, AGN activity, and TDEs remain to be fully disentangled, all scenarios require sustained energy injection ≳10⁴⁰ erg/s and efficient coupling to the halo. The smooth, sharply defined rims, matching symmetry, and multiwavelength spectra—in concert with the inability of advective transport alone to supply TeV electrons at high-latitude—strongly constrain the physical parameters and demand models featuring rapid, spatially distributed CR acceleration and halo-wide turbulent coupling (Tank et al., 10 Nov 2025, Sarkar, 14 Mar 2024, Scheffler et al., 30 Jan 2025, Zhang et al., 18 Jul 2025, Tseng et al., 31 May 2024).