eROSITA Bubbles: Galactic Halo Feedback

• eROSITA Bubbles (RBs) are giant soft-X-ray lobes centered on the Galactic Center, acting as key indicators of episodic nuclear feedback and energy injection into the halo.
• They exhibit sharp, well-defined X-ray edges with multiphase thermal structures and variable shock strengths, which help constrain their age and underlying dynamics.
• Multi-wavelength analyses, including X-ray, gamma-ray, and radio observations, demonstrate the RBs' role in shaping cosmic-ray transport and mediating metal mixing in the inner circumgalactic medium.

The eROSITA bubbles (RBs) are a pair of giant soft-X-ray lobes centered on the Galactic Center, extending symmetrically above and below the Milky Way disk and now widely treated as a primary nearby manifestation of Galactic nuclear feedback. In the discovery analysis, the structures span roughly $80^\circ$–$85^\circ$ in projection, enclose the smaller Fermi-LAT $\gamma$-ray lobes, and imply an energy injection of order $10^{56}\,\mathrm{erg}$ into the inner circumgalactic medium (CGM) (Predehl et al., 2020). Subsequent work has established the RBs as large, distant Galactic-halo structures rather than local foreground shells, while leaving open major questions concerning their detailed thermodynamic state, shock strength, age, asymmetry, and physical driver (Liu et al., 2024, Sarkar, 2024).

1. Discovery and observational identification

The RBs were first isolated in the eROSITA all-sky survey as two giant soft-X-ray “bubbles” permeating the Milky Way halo and extending symmetrically above and below the Galactic Center. In the original interpretation, the bright northern limb is the high-latitude analogue of the North Polar Spur, while the newly traced southern annulus mirrors it with fourfold symmetry about Sgr A*. One-dimensional surface-brightness cuts through the $0.6$–$1.0\,\mathrm{keV}$ maps show sharp, well-defined edges, and the overall morphology places the RBs in direct spatial correspondence with the pre-existing Fermi bubbles (Predehl et al., 2020).

A central early controversy concerned distance. One class of interpretations treated the northern structure as a local supernova-related shell, whereas another treated the full system as a Galactic-center outflow on $\mathrm{kpc}$ scales. Morphological work using three-dimensional dust maps identified three isolated clouds at $500$–$800\,\mathrm{pc}$ whose projected shadows coincide with defining X-ray-dark features of the North Polar Spur and the Lotus Petal Cloud. Because these clouds fully absorb the soft X-ray emission in projection, the corresponding X-ray structures must lie behind them, yielding a conservative lower limit of nearly $1\,\mathrm{kpc}$ and disfavoring local $85^\circ$0 models for the main northern arc (Liu et al., 2024).

That same morphological analysis further argued that the North Polar Spur and the Lotus Petal Cloud are not independent distant features but two limbs of one giant bubble rooted in the Galactic Center. Polarized radio arcs in the X-ray-dark gap between them align with the outer border of the X-ray structure and provide a way to define a continuous northern bubble boundary, strengthening the case that the RBs are giant and distant structures rather than coincidental local foreground features (Liu et al., 2024).

2. Geometry and three-dimensional structure

Under the discovery paper’s approximate-sphericity deprojection, each RB lobe has a radius of $85^\circ$1, so that the hot rims reach $85^\circ$2 above and below the plane. The sharp X-ray edges are most closely reproduced by a thick-shell geometry with inner radius $85^\circ$3 and outer radius $85^\circ$4, inferred from one-dimensional surface-brightness cuts through the $85^\circ$5–$85^\circ$6 maps (Predehl et al., 2020).

Later geometric reconstructions replaced simple spherical caps with more flexible parameterizations. Liu et al. modeled the northern structure as the line-of-sight tangent of a three-dimensional skewed cup rooted in the Galactic Center, with best-fit radii $85^\circ$7, $85^\circ$8, $85^\circ$9, skew factors $\gamma$0 and $\gamma$1, and a cup-cutoff height $\gamma$2. A comparable southern fit yielded a more prolate and less tilted bubble, again sharing a Galactic-center root (Liu et al., 2024).

A more recent eROSITA morphology analysis of the western Galactic hemisphere found that the horizontal size of both bubbles is comparatively well constrained, with semi-minor axes of order $\gamma$3–$\gamma$4, whereas the vertical extent is much less secure because the observed X-ray emission is nearly insensitive to whether a high-latitude cap exists and where it is located. In that fit, the northern RB requires a tilt of $\gamma$5 toward $\gamma$6, while the southern bubble requires only a small tilt (Yeung et al., 4 May 2026). This suggests that the silhouette of the RBs is controlled more tightly by their lateral shell geometry than by the poorly constrained high-latitude cap.

3. X-ray spectroscopy, plasma phases, and abundances

In the discovery analysis, the rims of the RBs are dominated by thermal plasma at $\gamma$7. A preliminary spectral fit to the bright southwestern rim yielded an absorption-corrected column $\gamma$8–$\gamma$9, and converting the eROSITA count rate into emission measure under the assumption of $10^{56}\,\mathrm{erg}$0 abundances gave characteristic electron densities $10^{56}\,\mathrm{erg}$1–$10^{56}\,\mathrm{erg}$2 for a $10^{56}\,\mathrm{erg}$3-thick shell. Integrating over the full sky extent gave $10^{56}\,\mathrm{erg}$4 in the $10^{56}\,\mathrm{erg}$5–$10^{56}\,\mathrm{erg}$6 band and a total thermal energy $10^{56}\,\mathrm{erg}$7 (Predehl et al., 2020).

Suzaku analyses complicated this picture by showing that a single-temperature shell model is not generally adequate. Gupta et al. reported that a standard three-component soft diffuse X-ray background model systematically underpredicts the flux at $10^{56}\,\mathrm{erg}$8–$10^{56}\,\mathrm{erg}$9 and $0.6$0–$0.6$1, whereas a two-temperature thermal model markedly improves the fits. In their bubble sightlines, the warm-hot component was fitted with $0.6$2 and $0.6$3, while the hot component had $0.6$4 and $0.6$5. The corresponding halo temperatures outside the shells were similar, but the shell emission measures were higher by factors of $0.6$6–$0.6$7 (Gupta et al., 2022).

The same Suzaku study emphasized chemical structure. In the warm-hot phase of the bubbles, $0.6$8 solar, about ten RB sightlines require $0.6$9 solar, and one sightline shows $1.0\,\mathrm{keV}$0 solar. Gupta et al. interpreted the supersolar $1.0\,\mathrm{keV}$1 and $1.0\,\mathrm{keV}$2 ratios as favoring a stellar-feedback origin over an AGN wind whose abundances would reflect the interstellar medium (Gupta et al., 2022).

Spatially resolved eROSITA spectroscopy of the western hemisphere also found a two-component thermal structure, but with somewhat different values: a hotter component at $1.0\,\mathrm{keV}$3 and a colder one at $1.0\,\mathrm{keV}$4, with the colder component’s emission measure larger by a factor of about five on average. That analysis found sub-solar abundances in the interior, $1.0\,\mathrm{keV}$5, and no conclusive evidence for $1.0\,\mathrm{keV}$6-element enhancement there, whereas the North Polar Spur showed higher abundances, $1.0\,\mathrm{keV}$7, which at face value disfavors a simple common origin for the North Polar Spur and the western eRObub interior (Yeung et al., 4 May 2026). Taken together, these studies indicate that the RB plasma is at least multiphase and that abundance inferences are region dependent.

4. Shock physics, age estimates, and dynamical state

The discovery paper interpreted the outer edge of the RBs as a non-radiative, collisionless forward shock. Using the contrast between an interior temperature $1.0\,\mathrm{keV}$8 and an ambient halo temperature $1.0\,\mathrm{keV}$9, and the Rankine-Hugoniot temperature jump relation

$\mathrm{kpc}$0

it inferred a weak Mach number $\mathrm{kpc}$1. For an upstream temperature corresponding to $\mathrm{kpc}$2, the implied shock velocity is $\mathrm{kpc}$3, which gives a characteristic expansion age of $\mathrm{kpc}$4 for a $\mathrm{kpc}$5 bubble (Predehl et al., 2020).

Gupta et al. argued that this weak-shock interpretation is not generally consistent with the broader Suzaku data set. Using

$\mathrm{kpc}$6

with $\mathrm{kpc}$7, they obtained $\mathrm{kpc}$8. Adopting an upstream density $\mathrm{kpc}$9 yields a compression ratio $500$0, which is the theoretical maximum for an infinitely strong adiabatic shock, yet they found no corresponding post-shock temperature increase. They therefore concluded that the shells cannot be simple adiabatic shocks and that their X-ray brightness traces denser gas rather than hotter gas (Gupta et al., 2022).

Other analyses recover stronger shocks, especially at high latitudes and in selected sectors. Keshet and Ghosh used eROSITA edge detection combined with stacked radio and $500$1-ray profiles to detect both bubble tips at $500$2 in both radio and $500$3-rays. Interpreting the downstream radio spectra with diffusive shock acceleration, they found $500$4 and $500$5, corresponding to $500$6 and $500$7, respectively. In their interpretation, both tips robustly exceed $500$8 and remain $500$9, while the southern bubble is fainter because its edge propagates into a medium roughly half as dense (Keshet et al., 20 Jan 2026).

A focused study of the south-eastern bubble likewise treated that sector as a propagating forward shock. Using a post-shock temperature $800\,\mathrm{pc}$0, it derived a current expansion speed of $800\,\mathrm{pc}$1, a physical radius $800\,\mathrm{pc}$2–$800\,\mathrm{pc}$3, and an age $800\,\mathrm{pc}$4–$800\,\mathrm{pc}$5. The same analysis inferred an upstream CGM density $800\,\mathrm{pc}$6 with a factor-of-two systematic uncertainty, and a heavy-element abundance $800\,\mathrm{pc}$7, emphasizing that this region may provide an effectively in situ CGM measurement (Churazov et al., 21 Mar 2026).

By contrast, the western-hemisphere eROSITA morphology fit recovered more modest shell Mach numbers, $800\,\mathrm{pc}$8 and $800\,\mathrm{pc}$9, while also stressing that cap geometry is degenerate (Yeung et al., 4 May 2026). The literature therefore does not support a single universal shock characterization: some analyses favor weak or modest shocks, while others infer $1\,\mathrm{kpc}$0 at high-latitude tips or in the south-eastern sector.

The dynamical smoothness of the rims adds a further constraint. Time-dependent Rayleigh-Taylor calculations applied to the RBs showed that the observed smooth outer shell implies an age significantly below the fragmentation time. In the $1\,\mathrm{kpc}$1-model halo considered there, the current age was estimated as about $1\,\mathrm{kpc}$2, while shell fragmentation times are $1\,\mathrm{kpc}$3–$1\,\mathrm{kpc}$4, explaining why the rims remain intact (Schulreich et al., 2021).

5. Origin scenarios and competing formation models

The most basic conclusion of the discovery paper was that large energy injections from the Galactic Center are the most likely cause of both the X-ray and $1\,\mathrm{kpc}$5-ray bubbles. It explicitly left open whether the engine was stellar feedback or supermassive-black-hole activity. In that comparison, a central starburst wind was described as being at the upper limit of what could supply the required impulse, though not impossible if the inner $1\,\mathrm{kpc}$6 experienced a star-formation rate of a few $1\,\mathrm{kpc}$7. A Seyfert-like episode of radiatively inefficient accretion onto Sgr A* lasting $1\,\mathrm{kpc}$8–$1\,\mathrm{kpc}$9 with bolometric $85^\circ$00 was said to provide the required energy more easily (Predehl et al., 2020).

AGN-jet models have been especially influential. Yang et al. used three-dimensional hydrodynamic plus cosmic-ray simulations with FLASH and CRSPEC to model a bipolar AGN jet from Sgr A* with $85^\circ$01, $85^\circ$02, duration $85^\circ$03, and total energy $85^\circ$04–$85^\circ$05. In their model, the shock reaches $85^\circ$06 vertically and $85^\circ$07 laterally at $85^\circ$08, while the same event reproduces the morphology and multi-wavelength spectra of the RBs, the Fermi bubbles, and the microwave haze (Yang et al., 2022).

A related but explicitly two-episode version was advanced by Zhang et al., who used PLUTO simulations to argue that a first jet pair launched $85^\circ$09 ago created the RBs, now extending to $85^\circ$10 with gas temperatures of $85^\circ$11, while a second jet pair launched $85^\circ$12 ago produced the Fermi bubbles, currently reaching $85^\circ$13 with temperatures of $85^\circ$14–$85^\circ$15. In that picture, the sharp edges of both systems are forward shocks (Zhang et al., 18 Jul 2025). Keshet and collaborators similarly argued that nested Fermi and eROSITA bubbles require two distinct but very similar collimated Galactic-center outbursts, each carrying $85^\circ$16, separated by roughly $85^\circ$17 (Ghosh et al., 30 Jan 2026).

A different recurrent-transient channel is provided by tidal disruption events. Scheffler et al. modeled regular TDEs at Sgr A* occurring every $85^\circ$18–$85^\circ$19, each depositing $85^\circ$20 or $85^\circ$21 so that the mean luminosity is $85^\circ$22. In their $85^\circ$23-model halo, a superbubble blown for $85^\circ$24 by injections every $85^\circ$25 reaches $85^\circ$26–$85^\circ$27, develops a coherent $85^\circ$28–$85^\circ$29 shell, and yields synthetic X-ray maps with bright rims at the observed longitudes and latitudes (Scheffler et al., 30 Jan 2025).

Stellar-feedback models remain competitive, but the mechanism is formulated in more than one way. Gupta et al. argued from chemical abundances that the RB shells are most naturally explained by massive-star feedback in the Galactic Center (Gupta et al., 2022). Shimoda and Asano instead proposed that the Fermi and eROSITA bubbles are persistent structures generated by a steady cosmic-ray-driven Galactic wind associated with ordinary star formation. In that scenario, cosmic-ray heating dominates radiative cooling for densities $85^\circ$30, keeping gas near the virial temperature of $85^\circ$31; inner-disk material falls back as a fountain, outer-disk material escapes, and the resulting structures persist for at least the last $85^\circ$32 rather than tracing a single $85^\circ$33 event (Shimoda et al., 2024).

No consensus origin has therefore emerged. The present literature supports at least four live classes of explanation: short AGN-jet episodes, recurrent TDE driving, stellar-feedback or starburst driving, and quasi-steady cosmic-ray-driven winds. The decisive discriminants differ among these models and include abundances, shock speeds, temporal sequencing of the nested Fermi and eROSITA systems, and the degree to which the RBs are transient versus persistent.

6. Asymmetry, multi-wavelength counterparts, and astrophysical significance

Although the RBs are globally bipolar and approximately symmetric about the Galactic Center, several studies emphasize north-south and east-west asymmetries. Hydrodynamic simulations by Wang et al. argued that the northern bubble’s westward tilt and eastern X-ray brightening are best explained by a CGM wind of about $85^\circ$34–$85^\circ$35 entering from “east by north,” compressing the windward eastern edge, tilting the northern bubble by $85^\circ$36 to the west, and helping to make the southern bubble $85^\circ$37–$85^\circ$38 times fainter in $85^\circ$39–$85^\circ$40. That model also inferred an accretion rate $85^\circ$41–$85^\circ$42 (Mou et al., 2022). Keshet et al., however, argued that the asymmetry favors an eastern ambient-density gradient over western-wind suggestions, because a hypothetical lateral wind would be difficult to reconcile with the strong nested shocks (Ghosh et al., 30 Jan 2026). The asymmetry problem therefore remains open.

Across wavebands, the RBs are tightly coupled to the Fermi bubbles. The discovery paper emphasized that the RB shells neatly enclose the Fermi-LAT lobes, with the X-ray rim tracing the forward shock in halo gas and the $85^\circ$43-ray edge plausibly marking the contact discontinuity between shocked ambient medium and shocked outflow (Predehl et al., 2020). Yang et al. further showed that the same jet event can reproduce the microwave haze through synchrotron emission from the same cosmic-ray electron population that produces the Fermi-bubble inverse-Compton emission (Yang et al., 2022).

Possible particle counterparts now extend beyond photons. A study of AMS-02 data identified positron-fraction features near $85^\circ$44, $85^\circ$45, and $85^\circ$46 and argued that the $85^\circ$47 feature corresponds to an event age of $85^\circ$48–$85^\circ$49, consistent with independent bubble age estimates; that work proposed these features as counterpart signals of the Fermi bubbles, of substructures in them, or of the RBs (Cholis et al., 2022). At much higher energies, IceCube HESE 12-year analyses found $85^\circ$50 neutrino detections of the high-latitude Fermi-bubble shells and preliminary evidence for neutrinos from the larger RB sectors, especially the X-ray-bright northeastern shell. In that interpretation, a typical nonthermal ion energy of order $85^\circ$51 is carried in each shell, corresponding to an ion acceleration efficiency of order $85^\circ$52 for a $85^\circ$53 outburst (Keshet et al., 21 Jun 2026).

The RBs have also been proposed to affect Galactic cosmic-ray transport on much larger scales. In a phenomenological transport model, an advective outflow boundary associated with the eROSITA bubbles reproduces the $85^\circ$54 hardening in secondary-to-primary ratios, with best-fit parameters $85^\circ$55 for the inner halo half-height and $85^\circ$56 for the bubble outflow speed (Schroer, 2 Apr 2026). This suggests that the RBs may act not only as tracers of a past outflow but also as a macroscopic boundary condition for present-day Galactic cosmic-ray escape.

In the broadest astrophysical sense, the RBs demonstrate that the Milky Way has reheated its CGM episodically, even though it is a quiescent disk galaxy. The discovery paper emphasized that a blast of $85^\circ$57 at $85^\circ$58 likely drives turbulence, buoyant motions, and metal mixing through the inner halo, temporarily halting the slow cooling flow that otherwise replenishes the disk (Predehl et al., 2020). This suggests that the RBs are a nearby empirical instance of the feedback cycle usually discussed in the context of galaxy evolution.

Several proposed observational tests are now direct. Wang et al. argued that Doppler mapping of $85^\circ$59–$85^\circ$60 lines with Athena or HUBS could detect bulk CGM motions of $85^\circ$61–$85^\circ$62 and map the asymmetry-producing wind or density redistribution (Mou et al., 2022). Churazov et al. proposed that soft X-ray bolometers with $85^\circ$63 could resolve the double-horn velocity profile of an expanding shell and decisively test whether the south-eastern bubble is a forward shock (Churazov et al., 21 Mar 2026). Because the present literature contains mutually inconsistent inferences for shock strength, metallicity, and age, such kinematic measurements are likely to be central to the next phase of RB studies.