Interstellar Medium of Galaxies

The interstellar medium is the thin soup of gas and dust between a galaxy's stars — where its baryons spend most of their idle life waiting to become stars. The mean particle density is about one hydrogen atom per cubic centimetre — a better vacuum than any laboratory, nineteen orders of magnitude thinner than sea-level air — yet across the disk this tenuous matter is roughly a tenth of the visible baryonic mass and supplies every gramme of every future star. By mass it is about 70% hydrogen, 28% helium, and a few percent heavier elements, with about a hundredth locked into solid dust grains. The one fact that organizes everything else is that the ISM is not a single gas but a set of coexisting phases over six decades of temperature, from ten-degree molecular clouds to million-degree supernova bubbles, held together by one shared quantity: pressure. Because the phases share $P/k\sim3000\ \mathrm{K\,cm^{-3}}$ , cold gas must be dense and hot gas rarefied, and — remarkably — the microphysics of heating and cooling makes two of these phases stable at the same pressure at once. That is the Field thermal instability, the spine of this article: it explains why the neutral gas sorts itself into cold clouds and warm intercloud gas rather than a uniform haze. Around it we hang the 21-cm line that maps the atomic gas, the CO-to- $\mathrm{H_2}$ conversion that weighs the molecular gas, the extinction law that betrays the dust, and the Jeans criterion that turns cold gas into stars.

Key takeaways

The interstellar medium (ISM) is the gas and dust between stars — about 10–15% of the disk's baryonic mass, mean density ∼1 H atom cm⁻³, spanning six decades in temperature and density.
It lives in coexisting phases at a common pressure $P/k\sim3000\ \mathrm{K\,cm^{-3}}$ : molecular clouds (10–50 K), cold and warm neutral atomic gas (CNM/WNM), warm ionized gas (WIM), and a supernova-inflated hot ionized medium (HIM, $\sim10^6$ K).
The Field thermal instability makes the equilibrium curve $P(n)$ non-monotonic, so two neutral phases (dense cold, diffuse warm) are stable at one pressure — the origin of the multiphase structure.
Each phase has its probe: the 21-cm H I line (atomic), CO via $X_{\rm CO}\approx2\times10^{20}$ (molecular), Hα and free-free (ionized), dust extinction with $R_V\approx3.1$ (grains).
Star formation drains the coldest, densest molecular gas once it exceeds the Jeans mass; winds and supernovae return heated, enriched gas — the baryon cycle.
Key idea: the ISM is a multiphase, roughly pressure-confined medium whose thermal state is set by heating (photoelectric, cosmic-ray) balanced against radiative line cooling, producing coexisting phases that mediate the conversion of gas into stars and back.

The core insight

The ISM is a multiphase, roughly pressure-confined medium whose thermal state is set by the balance of heating (photoelectric emission from dust, cosmic rays) against radiative cooling (atomic and molecular line emission), producing distinct coexisting phases that together mediate the conversion of gas into stars and back.

Where this lives today

The multiphase ISM is one of the most actively surveyed systems in astrophysics because it sits at the choke point of galaxy evolution. All-sky 21-cm surveys (HI4PI, GALFA-HI) map the atomic gas and its cold/warm split, while ALMA resolves molecular clouds from the Milky Way out to the reionization epoch, stress-testing the $X_{\rm CO}$ factor of Example 3. Planck turned the sky into a dust map, and Gaia parallaxes now yield three-dimensional extinction maps of the solar neighbourhood in place of the old two-dimensional screens. On the theory side the frontier has moved from static equilibrium to dynamics: simulations like TIGRESS and FIRE resolve how supernova feedback stirs the phases, self-regulating star formation so galaxies convert only a few percent of their gas per free-fall time — the Kennicutt–Schmidt law emerging from the balance these equations sketch. Open problems stay sharp: the true filling factor and pressure of the hot phase, the abundance of "CO-dark" gas, the cosmic-ray and magnetic pressure that equations (1) and (5) omit, and how the same physics scales in the low-metallicity gas JWST now dissects.

The physical picture

Imagine a room in which very hot air and very cold air share one pressure. Because pressure is roughly common, the hot parcels must be diffuse and the cold dense — and both can sit side by side indefinitely. That is the ISM in a sentence. What the analogy hides is why the gas prefers two temperatures rather than a smooth gradient, and the answer is radiative cooling: atoms and molecules shed heat efficiently only where they have accessible transitions — fine-structure lines like [C II] 158 μm in cool neutral gas, forbidden lines in ionized gas, Lyman-α near $10^4$ K. Where cooling outruns heating the gas condenses to a cold branch; where it cannot cool (very hot, very rarefied gas) it lingers hot for tens of millions of years. Selective cooling is the sieve that sorts a stirred medium into discrete phases, mapped in the first figure onto the density–temperature plane.

Dust is the ISM's catalyst and thermostat in one. Grains soak up starlight and re-emit it in the far infrared, shield cloud interiors from dissociating ultraviolet, and provide the only surfaces where two hydrogen atoms can combine into $\mathrm{H_2}$ — a reaction that essentially cannot happen in the gas phase. Photoelectrons kicked off those grains are the dominant heating source for the neutral gas. The deepest point in the subject is that the balance between this heating and line cooling is not single-valued: over a range of pressures it permits two stable densities at once. The second figure is that equilibrium curve — the pressure-versus-density relation whose S-shape is why the neutral ISM is two-phase rather than one.

Figure

The interstellar thermal-equilibrium curve P/k versus n where heating balances cooling. The S-shape has a segment with dP/dn negative (red, thermally unstable); a horizontal line at P/k ≈ 3000 K cm⁻³ meets two stable branches — warm neutral gas near n ≈ 0.3 cm⁻³ and cold neutral gas near n ≈ 30 cm⁻³ — so both coexist at one pressure

The core equations

One shared pressure. The most useful number for the diffuse ISM is its thermal pressure, because it is what the phases have in common: sound-crossing times are short compared with the phases' evolution, so parcels equalize pressure long before temperature.

\frac{P}{k} = n\,T \approx 3\times10^3\ \mathrm{K\,cm^{-3}} \qquad (1)

where: $P$ — gas thermal pressure; $k$ — Boltzmann constant; $n$ — total particle number density; $T$ — kinetic temperature. At fixed $P$ , low $T$ forces high $n$ and high $T$ forces low $n$ , so $n_{\rm CNM}T_{\rm CNM}\approx n_{\rm WNM}T_{\rm WNM}$ — cold phase dense, warm phase diffuse, both on one isobar in the first figure.

The 21-cm line. Neutral atomic hydrogen is the workhorse tracer of the diffuse ISM because it emits a single sharp line from the hyperfine splitting of its ground state — the spin-flip of the electron relative to the proton.

\lambda_{21} = \frac{hc}{\Delta E} = 21.106\ \mathrm{cm},\qquad \nu_{21}=1420.406\ \mathrm{MHz} \qquad (2)

where: $\Delta E\approx5.9\times10^{-6}\ \mathrm{eV}$ — energy split between the parallel and antiparallel electron–proton spin states; $h$ — Planck constant; $c$ — speed of light. The transition is so forbidden (lifetime ∼11 Myr) that no laboratory can compete, yet the sheer column of hydrogen makes it bright — and its Doppler shift maps Galactic rotation straight through the dust.

Extinction and reddening. Dust removes starlight by absorption and scattering, preferentially at short wavelengths, so background stars are both dimmed and reddened — both quantified in magnitudes.

A_\lambda = 1.086\,\tau_\lambda,\qquad E(B-V)=A_B-A_V,\qquad R_V\equiv\frac{A_V}{E(B-V)}\approx3.1 \qquad (3)

where: $A_\lambda$ — extinction (dimming) in magnitudes at $\lambda$ ; $\tau_\lambda$ — dust optical depth; $E(B-V)$ — colour excess (extra reddening between blue and visual bands); $R_V$ — total-to-selective ratio, ≈3.1 for diffuse dust and larger in dense clouds where grains have grown. Measure how much redder a star looks than its type predicts, multiply by $R_V$ , and you recover the total dimming $A_V$ .

Gas follows dust. Dust is mixed into the gas at a nearly constant ratio, so a background star's reddening measures the total hydrogen column in front of it — the handiest way to weigh a sightline.

N_{\rm H} \approx 5.8\times10^{21}\,E(B-V)\ \mathrm{cm^{-2}\,mag^{-1}} \qquad (4)

where: $N_{\rm H}$ — total hydrogen column density (atoms cm⁻²); $E(B-V)$ — reddening in magnitudes. This Bohlin–Savage–Drake relation bridges an optical colour to a gas mass, which is why extinction maps double as gas maps.

Heating equals cooling. A diffuse parcel's temperature is set not by adiabatic history but by the instantaneous balance of heating against radiative cooling, both fast compared with dynamical times.

n\dot{u} = \Gamma - n^2\Lambda(T),\qquad \Gamma_{\rm pe}\approx10^{-25}\ \mathrm{erg\,s^{-1}}\ \text{per H} \qquad (5)

where: $\Gamma$ — volumetric heating rate, dominated by photoelectrons ejected from dust grains by starlight (plus cosmic rays); $\Lambda(T)$ — cooling function per particle pair; $n\dot u$ — net internal-energy change per volume. Heating scales as $n$ (one photoelectric event per atom) while cooling scales as $n^2$ (two-body collisions), so their balance pins a density-dependent equilibrium temperature $T_{\rm eq}(n)$ .

Field's instability criterion. Fed back into the pressure, the equilibrium isobar $P_{\rm eq}(n)$ can bend backwards — and wherever it does, the gas is thermally unstable to condensation at constant pressure.

\left.\frac{\mathrm{d}P_{\rm eq}}{\mathrm{d}n}\right|_{\rm eq} < 0 \;\;\Longrightarrow\;\; \text{thermally unstable} \qquad (6)

where: $P_{\rm eq}(n)=n\,k\,T_{\rm eq}(n)$ — pressure along the heating/cooling equilibrium locus. On the red branch of the second figure a small compression lowers the equilibrium pressure, so the parcel is over-pressured relative to its surroundings and keeps collapsing until it reaches the cold branch — the runaway that splits one gas into two.

Jeans mass. Once gas is cold and dense enough, thermal pressure can no longer support it against its own gravity; the threshold is the Jeans mass.

M_J = \left(\frac{5kT}{G\mu m_{\rm H}}\right)^{3/2}\left(\frac{3}{4\pi\rho}\right)^{1/2} \qquad (7)

where: $M_J$ — minimum mass for collapse; $T$ — temperature; $\rho=\mu m_{\rm H}n$ — mass density; $\mu$ — mean molecular weight (≈2.3 for molecular gas); $m_{\rm H}$ — hydrogen mass; $G$ — gravitational constant. Since $M_J\propto T^{3/2}\rho^{-1/2}$ , cold dense gas has a small Jeans mass — exactly why stars form in molecular clouds and nowhere else.

Weighing the invisible molecules. The bulk of a molecular cloud is $\mathrm{H_2}$ , which has no permanent dipole and does not radiate at 10 K, so we weigh it through the CO molecule riding along with it.

N(\mathrm{H_2}) = X_{\rm CO}\,W_{\rm CO},\qquad X_{\rm CO}\approx2\times10^{20}\ \mathrm{cm^{-2}\,(K\,km\,s^{-1})^{-1}} \qquad (8)

where: $N(\mathrm{H_2})$ — molecular hydrogen column; $W_{\rm CO}$ — velocity-integrated CO(1–0) intensity; $X_{\rm CO}$ — empirical Milky-Way conversion factor. CO is optically thick, so its brightness measures cloud area and number, not column — the near-constancy of $X_{\rm CO}$ is an empirical near-miracle carrying factor-of-two uncertainty.

The core derivations

Derivation AThermal bistability and the two-phase neutral medium

The goal is to show, from heating and cooling microphysics alone, that the neutral interstellar gas does not settle into one uniform temperature but splits spontaneously into a cold dense phase and a warm diffuse phase coexisting at one pressure. We balance heating against cooling to get an equilibrium temperature versus density, build the isobar, apply the Field criterion to find the unstable stretch, and read off the two stable phases the second figure displays.

Step 1. Set the net heating rate of equation (5) to zero: $\Gamma(n)=n^2\Lambda(T_{\rm eq})$ . Heating (grain photoelectrons, cosmic rays) is roughly $\propto n$ while collisional cooling is $\propto n^2$ , so the balance defines an equilibrium temperature $T_{\rm eq}(n)$ that falls as the gas is compressed. Why: making cooling win at high density lets a compressed parcel keep cooling instead of heating adiabatically.

Step 2. Put in the cooling function's shape. Below $10^4$ K the coolants are fine-structure lines ([C II] 158 μm, [O I] 63 μm); toward $10^4$ K Lyman-α collisional excitation switches on and $\Lambda(T)$ soars, capping $T_{\rm eq}$ . Why: this steep rise near $10^4$ K bends the equilibrium curve back on itself.

Step 3. Build the isobar $P_{\rm eq}(n)=n\,k\,T_{\rm eq}(n)$ . Because $T_{\rm eq}$ drops while $n$ rises, the product is non-monotonic — it climbs to $P_{\rm max}$ , falls, then climbs again, tracing the S-shape of the second figure. Why: the two rising limbs are the future phases; the falling limb between them is the barrier that keeps them apart.

Step 4. Apply the Field criterion (6). On the falling limb a parcel squeezed slightly needs less pressure than its neighbours provide, so it is over-pressured, contracts, and runs away to the dense limb; the two rising limbs are stable. Why: stability is just the sign of the pressure response to a compression.

Step 5. Read off coexistence. For any ambient pressure between $P_{\rm min}$ and $P_{\rm max}$ the isobar cuts both stable limbs at once — WNM ( $n\sim0.3\ \mathrm{cm^{-3}}$ , $T\sim8000$ K) and CNM ( $n\sim30\ \mathrm{cm^{-3}}$ , $T\sim80$ K) at the same $P/k\sim3000\ \mathrm{K\,cm^{-3}}$ . This is what 21-cm surveys see: warm gas in emission, cold gas in absorption, along the same sightlines. Why: two-phase H I is the direct fingerprint of the non-monotonic isobar. *The two stable branches land at CNM conditions (80 K, 30 cm⁻³) and WNM conditions (8000 K, 0.3 cm⁻³) at $P/k\sim3000\ \mathrm{K\,cm^{-3}}$ — exactly what 21-cm surveys observe.*

Go deeper · Derivation B · 1 minJeans mass from the virial criterion

Derivation BJeans mass from the virial criterion

The goal is to find the minimum mass a cloud of cold gas must exceed before its self-gravity overwhelms its thermal pressure and it collapses to form stars — the quantitative reason star formation is confined to the coldest, densest phase. We compare the gravitational and thermal energies of a uniform sphere via the virial theorem, impose collapse, eliminate the radius for density, and evaluate for real clouds.

Step 1. State the virial condition: a self-gravitating cloud is bound and contracts when twice its thermal energy fails to balance its gravitational binding, $2K+U<0$ . Why: the virial theorem is the cleanest statement of the tug-of-war between pressure and gravity.

Step 2. Write the two energies for a uniform sphere of mass $M$ , radius $R$ : $U=-\tfrac{3}{5}GM^2/R$ , and $K=\tfrac{3}{2}\,\tfrac{M}{\mu m_{\rm H}}\,kT$ for its $M/(\mu m_{\rm H})$ particles. Why: these are the standard uniform-sphere expressions, exact up to the $3/5$ geometric factor.

Step 3. Impose $2K<|U|$ , giving $\dfrac{3MkT}{\mu m_{\rm H}}<\dfrac{3}{5}\dfrac{GM^2}{R}$ , a relation between mass, temperature, and radius at threshold. Why: at equality the cloud is marginally bound; exceeding it tips into collapse.

Step 4. Eliminate the radius with $R=\left(\dfrac{3M}{4\pi\rho}\right)^{1/3}$ and solve for the critical mass: $M_J=\left(\dfrac{5kT}{G\mu m_{\rm H}}\right)^{3/2}\left(\dfrac{3}{4\pi\rho}\right)^{1/2}$ , equation (7). Why: trading $R$ for $\rho$ turns a mass–radius statement into a clean mass–density threshold.

Step 5. Evaluate. Since $M_J\propto T^{3/2}\rho^{-1/2}$ , only cold, dense gas has a small enough $M_J$ to collapse. A diffuse cloud at $n\sim10^3\ \mathrm{cm^{-3}}$ gives $M_J\sim17\ M_\odot$ ; a dense core at $n\sim10^5\ \mathrm{cm^{-3}}$ drops to $\sim1.7\ M_\odot$ . Why: one formula explains both why whole clouds are unstable and why they fragment to stellar masses. *The formula gives $M_J\approx17.5\ M_\odot$ at $(10\ \mathrm{K},\ 10^3\ \mathrm{cm^{-3}})$ and $\approx1.75\ M_\odot$ at $(10\ \mathrm{K},\ 10^5\ \mathrm{cm^{-3}})$ , confirming that denser cores fragment to stellar masses.*

Limiting cases

The character of the ISM changes utterly across the density–temperature plane; each phase is a limiting regime of the same balance equations.

Case	Result	Meaning
Molecular clouds	$n\sim10^{2}$ – $10^{6}\ \mathrm{cm^{-3}},\ T\sim10$ – $50\ \mathrm{K}$	Cold, dense, self-gravitating; traced by CO; the only sites of star formation. Much of the ISM mass in tiny volume, above the diffuse isobar.
Cold neutral medium (CNM)	$n\sim30\ \mathrm{cm^{-3}},\ T\sim80\ \mathrm{K}$	Diffuse atomic H I in 21-cm absorption; the dense stable branch of the bistable neutral gas.
Warm neutral / ionized medium	$n\sim0.1$ – $1\ \mathrm{cm^{-3}},\ T\sim6000$ – $10^4\ \mathrm{K}$	The diffuse stable branch (WNM, 21-cm emission) plus the ionized Reynolds layer (WIM, Hα); large volume-filling factor.
Hot ionized medium (HIM)	$n\sim10^{-3}$ – $10^{-2}\ \mathrm{cm^{-3}},\ T\sim10^6\ \mathrm{K}$	Coronal gas shock-heated by supernovae; fills much of the disk volume; radiates in X-rays and O VI; persists for tens of Myr.

Common trapDo not confuse a phase's mass with the volume it fills

The tempting mistake is to picture the ISM as a smooth gas at its mean ∼1 cm⁻³. It is nothing of the sort — it is violently inhomogeneous, and a phase's volume-filling factor has almost no relation to its mass fraction. The hot medium fills much of the disk volume yet holds negligible mass; molecular clouds are the reverse, cramming much of the mass into a sliver of volume. Averaging the density smears away exactly the structure that matters. A second trap hides in the molecular gas: $\mathrm{H_2}$ is nearly undetectable at 10 K, so molecular masses come from the empirical CO-to- $\mathrm{H_2}$ factor $X_{\rm CO}$ , uncertain at the factor-of-two level and known to go "CO-dark" in low-metallicity or strongly-irradiated gas. Quote molecular masses with that caveat.

Worked examples

1 · Will a cold cloud collapse? (fully worked)

A molecular cloud has $T=10$ K. Estimate its Jeans mass at a diffuse $n=10^3\ \mathrm{cm^{-3}}$ and inside a dense core at $n=10^5\ \mathrm{cm^{-3}}$ ( $\mu=2.3$ ). What does each tell you?

Step 1. Density: $\rho=\mu m_{\rm H}n=2.3\times1.673\times10^{-24}\times10^3=3.85\times10^{-21}\ \mathrm{g\,cm^{-3}}$ . Thermal factor: $5kT/(G\mu m_{\rm H})=5(1.38\times10^{-16})(10)/[(6.67\times10^{-8})(3.85\times10^{-24})]=2.69\times10^{16}\ \mathrm{g\,cm^{-1}}$ .

Step 2. Raise to the powers of equation (7): $(2.69\times10^{16})^{3/2}=4.41\times10^{24}$ , $(3/4\pi\rho)^{1/2}=7.88\times10^{9}$ , so $M_J=3.47\times10^{34}\ \mathrm{g}$ — $M_J\approx17\ M_\odot$ at $n=10^3\ \mathrm{cm^{-3}}$ .

Step 3. At $n=10^5\ \mathrm{cm^{-3}}$ only $\rho$ changes ( $\times100$ ), and $M_J\propto\rho^{-1/2}$ , so it drops tenfold to $M_J\approx1.7\ M_\odot$ .

Step 4. Interpret. A giant molecular cloud of $10^4$ – $10^6\ M_\odot$ dwarfs the 17- $M_\odot$ diffuse Jeans mass, so it is wildly unstable. But as its cores reach $n\sim10^5\ \mathrm{cm^{-3}}$ the local Jeans mass falls to roughly a solar mass, so the cloud does not collapse to one monster but fragments into many star-sized pieces — why a cloud makes a cluster. (The textbook shorthand "a few $M_\odot$ " holds only for dense cores; the diffuse value is genuinely ∼17 $M_\odot$ .)

2 · Dereddening a star and weighing its sightline (fully worked)

A star is observed with colour excess $E(B-V)=0.30$ mag. Using $R_V=3.1$ , find the total visual extinction, the fraction of light reaching us, and the hydrogen column in front of it.

Step 1. Total dimming: $A_V=R_V\,E(B-V)=3.1\times0.30=\mathbf{0.93\ mag}$ — nearly a magnitude on top of distance.

Step 2. Fraction transmitted: $I/I_0=10^{-A_V/2.5}=10^{-0.372}=\mathbf{0.42}$ — only 42% of the visual light survives; optical depth $\tau_V=A_V/1.086=0.86$ .

Step 3. Hydrogen column from equation (4): $N_{\rm H}=5.8\times10^{21}\times0.30=\mathbf{1.7\times10^{21}\ cm^{-2}}$ — a typical diffuse sightline, a few hundred parsecs of ordinary ISM.

Step 4. Why it matters: uncorrected, the star is catalogued 0.93 mag too faint and too red, corrupting its distance, luminosity, and temperature. Every stellar-population and distance-ladder measurement in the plane rests on exactly this calculation.

3 · Weighing a molecular cloud with CO

A CO(1–0) map gives a velocity-integrated intensity $W_{\rm CO}=50\ \mathrm{K\,km\,s^{-1}}$ . Find the $\mathrm{H_2}$ column and mass surface density.

Step 1. Column from equation (8): $N(\mathrm{H_2})=X_{\rm CO}W_{\rm CO}=(2\times10^{20})(50)=\mathbf{1.0\times10^{22}\ cm^{-2}}$ .

Step 2. Mass per area with helium ( $\mu_{\rm H_2}\approx2.8\,m_{\rm H}$ ): $\Sigma=N(\mathrm{H_2})\,\mu_{\rm H_2}m_{\rm H}=0.047\ \mathrm{g\,cm^{-2}}=$ $224\ M_\odot\,\mathrm{pc^{-2}}$ , canonical for a cloud; a radius-10-pc cloud then holds $M\sim7\times10^4\ M_\odot$ — a GMC weighed from a line tracing gas we never directly see.

Physics of the portrait

Astronomers had long known there was gas between the stars — stationary absorption lines, dark clouds, glowing nebulae — but no framework tied the cold clouds to the warm intercloud gas until George B. Field showed that thermal physics forces the gas into distinct coexisting phases, the insight this article is built around.

Origins · 1 minGeorge B. Field and how this was discovered

George B. Field

1929–2019 · thermal instability and the multiphase interstellar medium

George Brooks Field was an American astrophysicist whose 1965 paper on the thermal instability of interstellar gas gave the field its organizing principle. He showed that a diffuse medium heated and cooled by radiation is generically unstable to breaking into phases: wherever equilibrium pressure falls as density rises, small condensations run away, so the gas cannot stay uniform. Four years later, with Donald Goldsmith and Harm Habing, he turned this into the two-phase model of the neutral ISM — cold dense clouds and a warm diffuse intercloud medium in pressure equilibrium — the picture 21-cm studies would confirm and McKee and Ostriker would extend in 1977 with the supernova-driven hot phase. Field's reach went further: the intergalactic medium, the 21-cm spin temperature underpinning today's cosmological 21-cm experiments, galaxy formation. He was the founding director of the Harvard–Smithsonian Center for Astrophysics and chaired the 1980s "Field Report" decadal survey. His signature idea — that ISM structure is thermodynamically compelled, not merely observed — is why we teach the subject as a physics problem rather than a catalogue.

History

1904

— Hartmann finds stationary Ca II absorption lines, the first evidence of gas between the stars

1930

— Trumpler establishes interstellar dust and extinction from open-cluster distances

1944

— van de Hulst predicts the 21-cm line; Ewen and Purcell detect it in 1951

1965

— Field derives the thermal instability of interstellar gas, formalized as the two-phase model with Goldsmith and Habing in 1969

1977

— McKee and Ostriker add the supernova-driven hot phase, completing the three-phase model

Going further

Texts. Carroll and Ostlie, An Introduction to Modern Astrophysics, 2nd ed., Ch. 12 · Draine, Physics of the Interstellar and Intergalactic Medium (2011), the standard graduate reference · Mo, van den Bosch and White, Galaxy Formation and Evolution, Ch. 8–9 · Binney and Tremaine, Galactic Dynamics, 2nd ed., Ch. 1 and 8 · Spitzer, Physical Processes in the Interstellar Medium (1978), the founding monograph. The anchor numbers used throughout are: 21-cm frequency 1420.406 MHz ( $\lambda=21.106$ cm); $R_V\approx3.1$ ; $N_{\rm H}$ – $E(B-V)$ coefficient $5.8\times10^{21}$ cm⁻² mag⁻¹ (Bohlin, Savage and Drake 1978); $X_{\rm CO}\approx2\times10^{20}$ cm⁻² (K km s⁻¹)⁻¹ (Bolatto, Wolfire and Leroy 2013); and $P/k\sim3000\ \mathrm{K\,cm^{-3}}$ . The two-phase and thermal-instability framework follows Field (1965) and Field, Goldsmith and Habing (1969), extended by McKee and Ostriker (1977). Key spot-checks: the Jeans mass at (10 K, $10^3$ cm⁻³) $\approx17\ M_\odot$ and at $10^5$ cm⁻³ $\approx1.7\ M_\odot$ ; the conversion factor $1.086=2.5\log_{10}e$ ; and the dereddening chain $A_V=0.93$ mag, transmitted fraction 0.42, $N_{\rm H}=1.7\times10^{21}$ cm⁻² for $E(B-V)=0.30$ .

Prerequisites. Structure of the Milky Way Galaxy · Galactic Coordinate System and the Solar Neighborhood · Stellar Populations I, II, and III. Leads to. Neutral Hydrogen 21-cm Line and Galactic Kinematics · Spiral Galaxies and Density Wave Theory · Galactic Rotation Curves.

Ask the tutor: Ask me to walk through why the heating–cooling balance bends backwards to make two phases, or to work any dereddening, Jeans-mass, or CO-to-H₂ estimate with your own numbers.

Is the interstellar medium really almost empty, and if so why does it matter?

Yes and no. Its mean density of about one atom per cubic centimetre is a harder vacuum than anything made on Earth. What makes it matter is volume: over a galactic disk, that whisper of gas is roughly ten percent of the visible baryonic mass, and it is the only reservoir from which new stars form. So the ISM is at once nearly empty and utterly decisive — empty locally, dominant globally. Its low density is also why its physics is exotic: collisions are rare, so gas can sit far from equilibrium and radiate through forbidden transitions that never appear in a dense laboratory gas.

What actually holds the different phases at the same pressure?

Sound. A pressure imbalance is smoothed out by pressure waves travelling at roughly the thermal speed. In the diffuse ISM the sound-crossing time of a region is short compared with the time it takes to heat, cool, or move, so parcels equalize pressure long before anything else. Temperature and density are then free to differ enormously between phases as long as $nT$ matches — which is why the first figure strings the phases along a single diagonal isobar. Pressure is the common currency, set by the weight of the disk and the stirring of supernovae, while each phase picks the temperature its cooling physics allows.

Why is the neutral gas two phases instead of a smooth range of temperatures?

Because the heating/cooling equilibrium is not single-valued. Plot the pressure a parcel needs to sit in balance against its density and the curve bends backwards — part of it has pressure falling as density rises. Gas on that falling stretch is unstable: squeeze it and it wants less pressure than its surroundings supply, so it collapses onto the cold branch; stretch it and it runs onto the warm branch. Only the two rising branches are stable, so the gas is driven to one temperature or the other, leaving a cold dense phase and a warm diffuse phase with almost nothing in between.

How do we know how much molecular gas a galaxy has if H₂ is invisible?

We cheat, carefully. H₂ at 10 K has no permanent dipole and its lowest radiating transitions need hundreds of kelvin, so cold clouds are dark in H₂. But carbon monoxide, about one CO per ten thousand H₂, has a rotational line at 2.6 mm that collisions easily excite. Decades of cross-checks — against dust, cosmic-ray gamma rays, dynamical masses — show CO brightness tracks the H₂ column through a roughly universal $X_{\rm CO}\approx2\times10^{20}$ . It works because CO lines are optically thick, so their integrated brightness counts clouds and velocity spread rather than a true column. The method is indispensable but imperfect: a factor-of-two scatter, and it fails in metal-poor or strongly-irradiated gas where CO is destroyed but H₂ survives — the "CO-dark" gas.

Why does dust redden starlight rather than just dimming it evenly?

Because interstellar grains are comparable in size to the wavelength of visible light — a few tenths of a micron — and such grains scatter and absorb blue light more efficiently than red. A 400 nm photon is far likelier to be knocked out of the beam than one at 700 nm, so a star seen through dust loses more blue light and looks redder than its type predicts. That colour shift, $E(B-V)$ , sits atop an overall dimming; the ratio $R_V=A_V/E(B-V)\approx3.1$ in diffuse gas encodes the grain sizes, climbing toward 5 in dense clouds where grains coagulate. A grey screen of large pebbles would dim without reddening — reddening is the fingerprint of small grains.

What inflates the million-degree hot phase, and why does it survive so long?

Supernova blast waves. When a massive star explodes it drives a shock that heats gas to $\sim10^6$ K, carving a low-density bubble. At the ambient pressure such hot gas must be extremely rarefied (recall $nT$ is roughly fixed), $\sim10^{-3}$ – $10^{-2}$ atoms per cubic centimetre, so it fills much of the disk volume while holding almost no mass. It survives because its cooling time is long: cooling scales as density squared, and this gas is so thin it radiates its heat only over tens of millions of years, while fresh supernovae keep re-heating it. The interconnected bubbles can break out of the disk to feed a hot Galactic corona and drive fountains that rain back down — the plumbing of the baryon cycle.