Circumgalactic Medium

A galaxy is mostly the gas you cannot see

When you picture a galaxy you picture the disk — the spiral arms, the bulge, the swarm of stars. But that luminous structure is the small bright core of something far larger. Surrounding every galaxy is a tenuous, near-invisible atmosphere of gas that fills the dark-matter halo and stretches out hundreds of kiloparsecs. This is the circumgalactic medium, the CGM. For the Milky Way it reaches roughly to the virial radius at about 250 kpc — more than ten times the radius of the stellar disk — and it contains as much ordinary matter as all the stars you can see, possibly more.

The CGM is not just a passive shroud. It is the galaxy's lungs and stomach at once. Fresh gas accretes through it from the cosmic web and sinks inward to form stars; supernovae and active nuclei blow enriched gas back out into it; some of that gas cools and rains back down again. Nothing enters or leaves a galaxy without crossing the CGM. If you want to understand why galaxies keep forming stars for billions of years instead of running dry in one, or why some galaxies abruptly "quench" and turn red, the answer lives out here in the halo, not in the disk.

Why the halo is hot: the virial temperature

When gas falls into a massive dark-matter halo, it falls fast — and where infalling streams collide and shock, that bulk kinetic energy thermalizes. The gas settles toward the halo's virial temperature, the temperature at which the thermal energy per particle balances the gravitational binding. Setting the thermal energy of a proton against the depth of the potential at the virial radius gives

k_B T_vir ≈ (1/2) μ m_p G M_halo / R_vir
T_vir ≈ (μ m_p / 2 k_B) · (G M_halo / R_vir)
      ≈ 6 × 10^5 K × (M_halo / 10^12 M☉)^(2/3)

where μ ≈ 0.6 is the mean molecular weight for fully ionised gas of cosmic composition, m_p is the proton mass, and the last line uses the scaling R_vir ∝ M_halo^(1/3) at fixed redshift. For a Milky-Way-mass halo of M_halo ≈ 1–2 × 10¹² M☉ this works out to roughly 10⁶ K — soft-X-ray-emitting plasma so hot that hydrogen and helium are fully ionised and the only spectral fingerprints come from highly stripped heavy ions like O VII and O VIII.

This single relation explains a fundamental divide between galaxies. In low-mass halos (≲ 10¹¹·⁵ M☉) the virial temperature is low enough that infalling gas never settles into a stable hot atmosphere — it cools faster than it can shock, accreting "cold" along filaments. In massive halos a stable, hot, hydrostatic atmosphere forms — the classic picture of a galactic corona. The transition between these two accretion modes, near 10¹²ish solar masses, is one of the most important boundaries in galaxy formation.

A medium with no single temperature

If the CGM were simply gas at the virial temperature, the story would be tidy. It is not. Observations reveal a genuinely multiphase medium, with cold and hot gas coexisting in the same volume across more than two orders of magnitude in temperature:

• Cool phase (~10⁴ K). Photoionised clouds traced by H I Lyα, Mg II, Si II/III, and C II/III. These are the clumps that can sink inward and feed star formation. COS-Halos found this cool gas out to ~150 kpc around essentially every galaxy it probed.
• Warm-hot phase (~10⁵–10⁵·⁵ K). Collisionally ionised gas traced by the O VI doublet at 1032/1038 Å — the signature ion of the CGM. O VI peaks in abundance near 3 × 10⁵ K, exactly the temperature where cooling is most efficient, so it marks gas in transition between phases.
• Hot phase (~10⁶ K). The virialised X-ray-emitting corona, traced by O VII and O VIII absorption and (faintly) emission. This phase carries much of the mass in massive halos but is hardest to detect.

The reason for the coexistence is thermal instability. Hot halo gas cools radiatively; where a parcel's cooling time drops below the local free-fall time, it loses pressure support faster than it can be replenished and condenses into a cool cloud. The controlling ratio is t_cool/t_ff: simulations and the "precipitation" framework (Voit and collaborators) find that when this ratio falls below roughly 10, the hot atmosphere precipitates cold clouds that rain onto the galaxy — a thermostat that ties the CGM directly to the fuel supply.

How we see the invisible: quasars as flashlights

The CGM is fatally faint in emission — its density is only 10⁻⁴ to 10⁻³ atoms per cubic centimetre, and surface-brightness from such thin gas falls below most instruments' sensitivity. The breakthrough technique is absorption-line spectroscopy: find a bright background quasar whose sightline happens to pass within a few tens of kiloparsecs of a foreground galaxy, and read off the shadows the halo gas casts on the quasar's spectrum.

Each ion absorbs at a known rest wavelength, so its presence, column density, and velocity are all measured directly:

H I  Lyα   1215.67 Å    neutral hydrogen, all cool gas
Mg II      2796/2803 Å  cool ~10^4 K photoionised clouds
C IV       1548/1551 Å  warm ~10^5 K
O VI       1031.9/1037.6 Å  warm-hot ~3×10^5 K   ← CGM signature
O VII      21.6 Å (X-ray)   hot ~10^6 K virial gas
O VIII     18.97 Å (X-ray)  hottest virial gas

Hubble's Cosmic Origins Spectrograph (COS), installed in 2009, opened the UV window that holds Lyα, C IV, and O VI for low-redshift galaxies, and the COS-Halos survey (Tumlinson, Werk, Prochaska and collaborators, 2011 onward) systematically mapped ~40 galaxy halos this way. Because a single sightline only samples one impact parameter, the field builds statistical halo "maps" by stacking many galaxy–quasar pairs. Newer integral-field instruments (MUSE on the VLT, KCWI at Keck) now detect extended Lyα emission directly around some galaxies, and the kinematic Sunyaev-Zeldovich effect and fast-radio-burst dispersion measures probe the elusive hot, ionised baryons.

The missing-baryon problem

Cosmology tells us precisely how much ordinary matter the universe contains: the baryon-to-matter ratio measured from the cosmic microwave background and big-bang nucleosynthesis is Ω_b/Ω_m ≈ 0.157. A halo of mass M_halo should therefore have formed with a baryon budget of

M_baryon, expected ≈ 0.157 × M_halo
                   ≈ 1.6 × 10^11 M☉   for M_halo = 10^12 M☉

Yet when astronomers count the baryons actually locked into a Milky-Way-like galaxy — stars, the interstellar medium, dust — they find only a fraction of that, perhaps 20–30%. The rest are the "missing baryons." The CGM is the leading hiding place. Absorption surveys infer 10¹⁰–10¹¹ M☉ of gas in the cool and warm-hot halo phases alone, and the hard-to-detect hot virial phase may hold more still. Add the diffuse warm-hot intergalactic medium beyond the halo and the local baryon census largely closes. The missing baryons were never missing; they were simply too diffuse and too hot to shine.

The baryon cycle: inflow, outflow, recycling

What makes the CGM the master regulator of galaxy growth is that it sits on a continuous conveyor belt of gas. Three flows define the baryon cycle:

Flow	Direction	Driver	Speed / signature	Consequence
Cosmological accretion	Cosmic web → CGM	Gravity	Cold streams, ~100–200 km/s	Fresh fuel supply
Cool inflow / precipitation	CGM → disk	Cooling + gravity	Inflow, ~100 km/s; redshifted absorption	Sustains star formation
Supernova-driven outflow	Disk → CGM	Stellar feedback	~100–500 km/s; blueshifted Mg II/O VI	Enriches & heats halo
AGN-driven outflow	Disk → CGM / beyond	Black-hole feedback	up to ~1000s km/s	Can unbind gas, quench
Galactic fountain	CGM → disk (return)	Cooling of fallback	~100–200 km/s; mixed metallicity	Recycles enriched gas
Hot atmosphere (static)	—	Virial shock	10⁶ K, X-ray	Pressure reservoir / quenching

The single number that ties it all together is the depletion time: at the Milky Way's star-formation rate of ~1–2 M☉/yr and disk gas mass of a few × 10⁹ M☉, the disk would run out of fuel in only 1–2 Gyr. Galaxies have been forming stars for over 10 Gyr. The deficit is made up by the CGM continuously resupplying gas — and the chemical fingerprint of that resupply (intermediate metallicities, a mix of pristine and enriched gas) is exactly what absorption surveys see. The CGM is the buffer tank that keeps the engine running.

The CGM by the numbers

It helps to see how the CGM compares to the other gaseous reservoirs a galaxy interacts with. The contrasts in density and scale are enormous:

Reservoir	Density (cm⁻³)	Temperature	Extent	Typical tracer
Molecular ISM	10² – 10⁶	10 – 100 K	disk midplane	CO, dust
Warm/cold ISM	0.1 – 100	10² – 10⁴ K	disk, ~0.3 kpc thick	H I 21 cm, Hα
Cool CGM	10⁻³ – 10⁻²	~10⁴ K	tens of kpc	Mg II, Lyα
Warm-hot CGM	~10⁻⁴	10⁵ – 10⁵·⁵ K	up to ~150 kpc	O VI, C IV
Hot CGM (corona)	~10⁻⁴ – 10⁻⁵	~10⁶ K (Tvir)	to Rvir ~250 kpc	O VII / O VIII (X-ray)
Intergalactic (WHIM)	10⁻⁶ – 10⁻⁴	10⁵ – 10⁷ K	cosmic web	O VI, broad Lyα

A useful intuition: the cool CGM is roughly 10²² times less dense than the air you are breathing — far emptier than any vacuum made on Earth — yet because it fills a sphere ~250 kpc across — a volume of order 10⁷ cubic kiloparsecs — the integrated mass rivals everything the galaxy has turned into stars over its entire history. Diffuse does not mean unimportant.

Where the CGM shows up

• The Milky Way's own halo. We sit inside our CGM, which complicates the view but lets us study it in exquisite detail. High-velocity clouds (HVCs) like Complex C are cool CGM gas falling toward the disk at ~100–200 km/s; the Magellanic Stream is a 200°-long ribbon of gas stripped from the Magellanic Clouds and now part of the halo; O VI and O VII absorption against background AGN traces the warm-hot and hot phases.
• COS-Halos and the cool-gas ubiquity. Hubble/COS found O VI and Lyα absorption around nearly every star-forming L* galaxy probed within ~150 kpc, establishing that a massive cool/warm-hot CGM is a universal feature of galaxies, not an exception.
• Lyα blobs and "Slug"-class nebulae at cosmic noon. Around z ≈ 2 quasars, KCWI and MUSE reveal CGM gas glowing in Lyα over hundreds of kiloparsecs — the giant "Slug" nebula around the quasar UM287 (Cantalupo et al. 2014) is a famous example, lit up by the quasar's radiation.
• Quenched ellipticals and hot coronae. Massive red galaxies sit in hot, X-ray-bright halos near 10⁶·⁵–10⁷ K. Here the CGM is stable, hot, and unable to cool efficiently — feedback keeps it that way — so accretion is throttled and star formation is shut down.
• Ram-pressure stripping in clusters. When a galaxy plunges through the hot intracluster medium, its CGM (and even its ISM) is stripped away, severing the fuel supply and quenching the galaxy from the outside in — a vivid demonstration of how losing the CGM changes a galaxy's fate.

Common misconceptions and edge cases

• "The CGM is just empty space." Its density is tiny, but it holds as much baryonic mass as the galaxy's stars. Empty in density, not in mass.
• "O VI tells you the hot virial gas." O VI peaks near 3 × 10⁵ K — the warm-hot, transitional phase — not the ~10⁶ K virial corona. The truly hot gas needs O VII / O VIII in X-rays. Confusing the two badly mis-estimates the mass budget.
• "All the halo gas was blown out by the galaxy." Much CGM gas is pristine, freshly accreted from the cosmic web and never inside a star. Absorption surveys see a mix of metallicities precisely because inflow and recycled outflow are both present.
• "Quenching means the gas is gone." In a massive quenched galaxy the CGM can be enormous — it's just hot and stable, unable to cool and fall in. The fuel is there; the thermostat is stuck on "hot."
• "A single quasar sightline maps a halo." One sightline samples one impact parameter and one velocity slice. CGM maps are statistical, built from hundreds of galaxy–quasar pairs at different impact parameters — never assume one pencil-beam represents the whole halo.
• "The CGM and the ISM are the same gas just farther out." They differ by orders of magnitude in density and by the physics that dominates: the ISM is star-forming and self-gravitating in places; the CGM is shaped by the halo potential, virial shocks, and large-scale flows.