Fermi Bubbles

Two balloons hiding in the gamma-ray sky

For decades the Milky Way's gamma-ray sky looked like a thin, bright band along the Galactic plane — the diffuse glow of cosmic rays smashing into interstellar gas — plus a scatter of pulsars and blazars. Then in 2010 Meng Su, Tracy Slatyer and Douglas Finkbeiner subtracted that expected foreground from the first two years of Fermi Large Area Telescope data and found something left over: two colossal lobes of emission billowing straight out of the Galactic Center, one toward the north Galactic pole and one toward the south. Each reaches roughly 8 to 10 kiloparsecs from the plane, spanning about 50 degrees in latitude apiece so that the full pair stretches roughly 100 degrees across the sky — projected onto a structure some 25,000 light-years tall per lobe.

What made them startling was not just their size but their cleanness. The bubbles have a roughly uniform surface brightness across their faces rather than a centrally peaked glow, and their edges are sharp — the emission drops off over a small fraction of the bubble radius. A slowly diffusing fog of cosmic rays would have soft, fuzzy boundaries. Crisp walls mean a shock or contact discontinuity: something was driven outward fast and is plowing into the surrounding halo. The Fermi Bubbles are, in effect, the smoke ring left by an explosion at the center of our own Galaxy.

Why they glow: leptonic versus hadronic gamma rays

Gamma rays at GeV energies are not thermal. No reasonable gas temperature radiates a billion-electron-volt photon as a blackbody; you need relativistic particles. Two channels can do it, and which one dominates inside the bubbles is still debated.

Leptonic (inverse Compton). Relativistic electrons scatter low-energy seed photons — starlight, infrared dust emission, and the cosmic microwave background — boosting them to gamma-ray energies. A single scattering kicks a seed photon of energy ε up to roughly

E_γ ≈ γ² ε

for γ ≈ 10⁶ (a TeV electron) scattering a CMB photon (ε ≈ 6 × 10⁻⁴ eV)
→ E_γ ≈ 10¹² × 6 × 10⁻⁴ eV ≈ 0.6 GeV

The catch is that TeV electrons cool fast — their inverse-Compton and synchrotron lifetime is short — so a leptonic model needs the electrons to have been accelerated recently or to be re-accelerated in situ. That naturally favors a young, energetic event.

Hadronic (pion decay). Cosmic-ray protons collide with ambient hydrogen nuclei, producing neutral pions that immediately decay into two gamma rays:

p + p → π⁰ + X
π⁰ → γ + γ        (each γ near 70 MeV in the pion rest frame)

Protons cool slowly, so a hadronic model can store the energy of an ancient outburst for a long time — but it requires more target gas than the tenuous halo readily supplies, and it predicts a high-energy neutrino counterpart that has not been clearly detected. The truth may be a mix: leptonic emission dominating the bulk of the GeV signal, with a hadronic contribution in the densest regions near the base.

What powered the outburst

The energy bookkeeping is enormous. Inflating two 8–10 kpc lobes against the weight of the Galactic halo and heating the gas at their walls requires a total energy injection of order 10⁵⁴ to 10⁵⁷ erg — the equivalent of 10³ to 10⁶ supernovae, delivered into the central few hundred parsecs over a span of about a million years. Two physical engines can supply that.

The first is Sagittarius A*, the Milky Way's central supermassive black hole of mass 4.3 × 10⁶ M☉. Today it accretes at a trickle, glowing at less than 10⁻⁸ of its Eddington luminosity. But if it briefly accreted near the Eddington rate a few million years ago — swallowing a tidally disrupted molecular cloud or a star — it could have launched a fast jet or wide-angle wind. The Eddington luminosity of Sgr A* is

L_Edd = 1.26 × 10³⁸ (M/M☉) erg/s
      = 1.26 × 10³⁸ × 4.3 × 10⁶
      ≈ 5.4 × 10⁴⁴ erg/s

Running at that luminosity for just 10⁴–10⁵ years delivers 10⁵⁶–10⁵⁷ erg — more than enough. The second engine is a nuclear starburst: a burst of star formation in the central molecular zone whose massive stars exploded as supernovae and drove a hot thermal wind. The Galactic Center does host a young population (the Arches and Quintuplet clusters), and starburst-driven superwinds are seen blowing kiloparsec-scale lobes out of galaxies like M82. Distinguishing the two histories — black-hole flare versus stellar superwind — is one of the central open questions about the bubbles.

The bubbles by the numbers

Property	Value	Note
Discovery	2010	Su, Slatyer & Finkbeiner, Fermi-LAT
Height of each lobe	~8–10 kpc	≈ 25,000–33,000 ly from the plane
Angular extent	~50° per lobe	≈100° north + south, on the sky
Maximum width	~6 kpc	Widest cross-section
γ-ray spectrum	dN/dE ∝ E⁻²	Hard, flat in νFν, ~1–100 GeV
1–100 GeV luminosity	~4 × 10³⁷ erg/s	Diffuse, near-uniform brightness
Edge sharpness	Drop over ≲ a few %	Implies a driven shock/contact wall
Total energy injected	10⁵⁴–10⁵⁷ erg	= 10³–10⁶ supernovae
Age	~1–6 Myr	From dynamical models
Outflow speed	~900–1,000 km/s	From entrained high-velocity clouds

For scale: a luminosity of 4 × 10³⁷ erg/s is about 10⁴ Suns radiated entirely in gamma rays, spread over a volume tens of thousands of light-years across. The bubbles are diffuse and faint per unit area — which is exactly why they hid inside the Galactic foreground until someone modeled and removed that foreground carefully.

The eROSITA bubbles: the same scar in X-rays

The Fermi Bubbles never lived alone. The microwave "haze" seen by WMAP and Planck — a region of hard-spectrum synchrotron emission — coincides with the bubble bases, telling us the same relativistic electrons radiate in radio as well as gamma rays. The North Polar Spur and the radio Loop I have long been mapped at the bubbles' edge. Then, in 2020, the SRG/eROSITA all-sky X-ray survey revealed the decisive image: two vast soft-X-ray bubbles, north and south, enclosing the gamma-ray Fermi Bubbles and extending even farther — about 14 kpc from the plane.

The interpretation is clean and satisfying. The X-rays trace shock-heated gas at roughly 10⁶–10⁷ K piled up against the outer walls of the outflow; the gamma rays trace the relativistic particles filling the interior. Two windows onto a single, galaxy-scale energy injection. The combined eROSITA + Fermi structure pushes the inferred energy budget to the upper end, ~10⁵⁶ erg, and strengthens the case that this was a genuine nuclear outburst rather than steady-state activity.

How they compare to other galactic outflows

Structure	Scale	Driver	Best seen in	Relation to Fermi Bubbles
Fermi Bubbles	8–10 kpc / lobe	Sgr A* flare or starburst	GeV gamma rays	The phenomenon itself
eROSITA bubbles	~14 kpc / lobe	Same outburst	Soft X-rays	Outer shock-heated shell
Microwave haze	~Bubble base	Same electrons	Microwave synchrotron	Radio counterpart
North Polar Spur / Loop I	~30° arc	Same complex	Radio / soft X-ray	Edge of the structure
M82 superwind	~few kpc	Nuclear starburst	Hα / X-ray	Starburst-analog comparison
Radio-galaxy lobes (e.g. Cygnus A)	~10–100 kpc	AGN jet	Radio synchrotron	AGN-jet analog at larger scale
Galactic chimney / fountain	~1 kpc	Clustered supernovae	Hα / H I	Smaller disk-driven outflow

Seen in this company, the Fermi Bubbles look like a scaled-down, nearby version of the radio lobes that AGN routinely blow — close enough that we can resolve their interior structure and measure the relativistic-particle content directly, something impossible in a galaxy hundreds of megaparsecs away. The Milky Way is, in a modest and temporary way, an active galaxy whose most recent outburst we happen to be watching cool.

Was Sgr A* recently active? Independent clues

The bubbles are not the only fossil of a more energetic Galactic Center. Several independent lines of evidence point to recent activity in the inner Galaxy:

• X-ray light echoes. Molecular clouds in the central few hundred parsecs (e.g. Sgr B2) emit fluorescent 6.4 keV iron-line X-rays that fade over years — consistent with reflecting a luminous flare from Sgr A* roughly 100–300 years ago, when it was perhaps a million times brighter than today.
• The Magellanic Stream H-alpha glow. Part of the trailing Magellanic Stream, tens of kpc above the disk, shows Hα emission that requires it to have been illuminated by a Seyfert-level ionizing flare from the Galactic nucleus a few million years ago — the same epoch the bubbles point to.
• The young nuclear stellar disk. A few-Myr-old population of massive stars orbits within a parsec of Sgr A*, evidence that a substantial accretion event delivered gas to the very center recently enough to form stars there.
• Entrained high-velocity clouds. Ultraviolet absorption lines toward background quasars reveal cold clouds being carried outward through the bubbles at ~900–1,000 km/s, a direct kinematic measurement of the wind that is inflating them.

None of these alone proves the origin, but together they sketch a Galactic Center that flickered to life a few million years ago, briefly behaving like a low-luminosity AGN, and left the bubbles as its longest-lived monument.

Common misconceptions and edge cases

• They are not the Galactic dark-matter halo, and not dark matter. Early on, some asked whether annihilating dark-matter particles produced the emission. But dark matter would give a centrally peaked, roughly spherical signal with a spectrum tied to a particle mass — not flat-topped lobes with sharp edges and a hard power-law spectrum. The morphology is the giveaway: the bubbles trace an outflow.
• They are not the same as the "Galactic Center GeV excess." A separate, more compact and roughly spherical GeV excess sits within a few degrees of Sgr A* and is debated as either millisecond pulsars or dark matter. That is a distinct structure from the kiloparsec-scale Fermi Bubbles, though both live in the inner Galaxy and both required careful foreground modeling to isolate.
• "Above and below" is real, not a projection artifact. The bipolar, mirror-symmetric geometry across the Galactic plane is intrinsic — exactly what a collimated outflow from the center produces, and a strong argument against any foreground or instrumental origin.
• The hard E⁻² spectrum is the clue, not a footnote. A flat νF_ν spectrum (equal power per logarithmic energy interval) is the signature of efficient particle acceleration — shocks or turbulence — not of a relaxed thermal or steady-diffusion population. It is why the bubbles read as the aftermath of a violent event.
• They will not last. At a few Myr old and expanding at ~1,000 km/s, the bubbles are transient. As the driving stops, the relativistic electrons cool, the shock weakens, and the gas mixes back into the halo over tens of millions of years. We are seeing a snapshot of a feature that is geologically young by Galactic standards.