Fundamental Equations of the Interstellar Medium

The matter between the stars — phases, radiation, dust, chemistry, turbulence, magnetism, shocks, and the road to new stars  ·  Summer 2026

0

What Is the Interstellar Medium?

Not empty space — the Galaxy's working fluid

Between the stars of the Milky Way sits roughly 7×10⁹ M☉ of gas and dust — about a tenth of the stellar mass — spanning seven orders of magnitude in temperature (10 K molecular cores to 10⁷ K supernova-heated plasma) and more than ten in density. It is the medium stars are born from, the repository they enrich when they die, and the screen through which we observe everything else. A "vacuum" by laboratory standards (the best lab vacuum is denser than a molecular cloud), it is nonetheless a chemically active, magnetized, supersonically turbulent fluid with weather, phases, and a life cycle.

The ISM's defining property is that it is a system far from equilibrium: heated by starlight, cosmic rays, and blast waves; cooled by spectral lines and dust; stirred faster than it can settle. Its physics is the physics of that imbalance. One warning about the bookshelf: this sheet is about the interstellar medium — the companion Interstellar sheet in this collection covers the physics of the film (Gargantua, wormholes, Miller's planet), a different beast entirely.

Multiphase structure — five coexisting phases in rough pressure balance. Radiative transfer — every observable is light filtered through the medium itself. Dust & chemistry — 1% of the mass, controlling heating, shielding, and H₂ formation. Turbulence & magnetism — supersonic, magnetized motions that resist and regulate collapse. Feedback — supernovae and H II regions that stir, heat, and eject the gas.

As on the companion sheets, every equation is paired with a plain-language reading of what it physically asserts, a Use in Research column with key references, and each section ends with the open unknowns. The Jeans criterion and the stellar side of star formation live on the stellar-astrophysics sheet; galactic-scale star-formation laws live on the galaxies sheet. Toggle the Dark theme at top-right for a dark background.

I

The Multiphase Medium

6 equations

The ISM is not one gas but five — molecular, cold neutral, warm neutral, warm ionized, and hot — coexisting at wildly different densities yet similar pressures. That arrangement is not an accident: it falls out of one heating–cooling balance with multiple stable solutions. These are the equations that decide which phase a parcel of gas belongs to.

NameEquationVariablesUse in Research
Interstellar Pressure \[ \frac{P}{k_{\rm B}} = n\,T \approx 3000\!-\!5000\ {\rm cm^{-3}\,K} \]
Phases at densities differing by factors of 100 coexist because pressure, not density, is what equilibrates. Cold-and-dense sits beside warm-and-tenuous at the same nT — a thermostat written into the phase diagram.
 n = total particle density; T = kinetic temperature; thermal part only (turbulent, magnetic, cosmic-ray parts are each comparable)
The organizing quantity of ISM structure: measured P/k pins the phase diagram, and its rough equality among thermal, turbulent, magnetic, and cosmic-ray components ("equipartition") is one of the deepest unexplained regularities in the field.
Key referencesSpitzer (1978); Jenkins & Tripp (2011); Draine (2011).
Heating–Cooling Balance \[ \Gamma\,n = n^2\,\Lambda(T) \]
Heating scales with the number of absorbers; cooling — collisions exciting ions that then radiate — scales with the number of colliding pairs. Equilibrium temperature is where the two curves cross, and the crossing depends on density: the origin of phases.
 Γ = heating rate per particle; Λ(T) = cooling function (line emission, mostly [C II] 158 µm and [O I] 63 µm in neutral gas)
The master balance behind every ISM temperature; [C II] 158 µm alone radiates away ~0.1–1% of the Milky Way's total luminosity — the single brightest emission line of the Galaxy, and the workhorse SFR tracer for high-z galaxies (ALMA).
Key referencesDalgarno & McCray (1972); Wolfire et al. (1995, 2003); Bennett et al. (1994).
Thermal Instability \[ \left(\frac{\partial \mathcal{L}}{\partial T}\right)_{\!P} < 0 \;\Rightarrow\; {\rm unstable} \]
If losing heat makes gas lose heat faster (at constant pressure), uniform gas cannot survive: it spontaneously separates into cold clouds and warm intercloud medium. Phases are not assembled — they condense, like fog.
 ℒ = net loss function nΛ − Γ; Field criterion, isobaric mode; unstable range ~300–5000 K for typical ISM
The mechanism that makes the two-phase medium: gas driven into 300 K < T < 5000 K by turbulence or spiral shocks splits into CNM and WNM in a cooling time — and simulations of ISM structure are largely simulations of this instability under stirring.
Key referencesField (1965); Field, Goldsmith & Habing (1969); Heiles & Troland (2003).
Photoelectric Heating \[ \Gamma_{\rm pe} \approx 10^{-26}\,\epsilon\, G_0\ \ {\rm erg\,s^{-1}\,per\,H} \]
Starlight heats interstellar gas by proxy: FUV photons strike dust grains and PAHs, kick out electrons, and the electrons share their energy collisionally. The gas is warmed by the smoke, not the fire — at a few percent efficiency.
 G₀ = FUV field in Habing units (local value ≈ 1.1); ε ≈ 0.01–0.05, efficiency (falls when grains charge up)
The dominant heat source of neutral atomic gas and PDR surfaces; its dependence on grain charging couples gas temperature to the electron density and radiation field — the feedback loop that sets where CNM can exist.
Key referencesWatson (1972); Bakes & Tielens (1994); Weingartner & Draine (2001); Hollenbach & Tielens (1999).
Supernova Energy Injection \[ \dot E_{\rm SN} \sim \frac{10^{51}\,{\rm erg}}{50\!-\!100\ {\rm yr}}\ \ ({\rm Galaxy\!-\!wide}) \]
Each supernova deposits ~10⁵¹ erg; the Galactic rate of a few per century makes the time-averaged power rival all other heating combined. Supernovae are the ISM's central heating plant — and its demolition crew.
 ~10–30% survives radiative losses as hot-gas thermal energy + turbulence + cosmic rays; per unit mass this is the ISM's dominant stirring term
The source of the hot (10⁶ K) phase, most ISM turbulence at large scales, the galactic fountain, and (via shock acceleration, Section IX) the cosmic rays. Modern galaxy simulations succeed or fail on how they handle exactly this injection.
Key referencesCox & Smith (1974); McKee & Ostriker (1977); Zucker et al. (2022).
The Phase Census \[ f_V({\rm HIM}) \sim 0.5;\quad f_M({\rm H_2 + CNM}) \sim 0.5 \]
Volume and mass live in different phases: hot gas fills perhaps half the disk's volume while holding ~1% of its mass; molecular gas holds a quarter of the mass in a fraction of a percent of the volume. The ISM is an ocean of foam with pebbles in it.
 f_V, f_M = volume/mass fractions; phases: H₂ (10–20 K), CNM (~80 K), WNM (~8000 K), WIM (~8000 K, ionized), HIM (~10⁶ K)
The bookkeeping frame for all ISM modeling; the hot-phase filling factor (McKee–Ostriker's central prediction) controls how supernova energy and metals propagate, and remains genuinely uncertain at the factor-of-two level.
Key referencesMcKee & Ostriker (1977); Reynolds (1989); Ferrière (2001); Draine (2011).
Open unknowns · The Multiphase Medium
Why Equipartition?
Thermal, turbulent, magnetic, and cosmic-ray pressures are all ~1 eV/cm³. Coincidence?
No first-principles argument requires four independently sourced energy reservoirs to equalize, yet they do to within factors of two. Self-regulation via shared instabilities is suspected but not demonstrated — this may be the ISM's most important unexplained fact.
The Hot-Gas Filling Factor
Does the 10⁶ K phase fill 20% or 70% of the disk?
Fifty years after McKee–Ostriker, the HIM's topology — connected tunnels versus isolated bubbles — is still unsettled; it controls supernova venting, metal mixing, and the disk–halo circulation. eROSITA maps constrain it locally but the Galaxy-wide answer is model-dependent.
What Regulates the Pressure?
Is midplane pressure set by gravity, feedback, or both in a fixed ratio?
Pressure-regulated, feedback-modulated models tie P/k to the weight of the gas column and star formation to the required support. The framework works remarkably well across galaxies — but whether it is the cause or a consequence of the phase structure is debated.
II

Radiative Transfer & the 21-cm Line

6 equations

Everything known about the ISM arrives as radiation that the ISM itself has emitted, absorbed, or scattered — the medium is both the subject and the instrument. Radiative transfer is the grammar of that self-portrait, and the 21-cm hyperfine line of hydrogen is its most productive single sentence.

NameEquationVariablesUse in Research
The Equation of Radiative Transfer \[ \frac{dI_\nu}{ds} = -\kappa_\nu\, I_\nu + j_\nu \]
Along every ray, intensity is depleted by absorption and replenished by emission. One linear ODE carries all of observational astrophysics — every spectrum is a solved or unsolved instance of it.
 I_ν = specific intensity; κ_ν = absorption coefficient; j_ν = emissivity; source function S_ν = j_ν/κ_ν
The engine inside every spectral-synthesis, PDR, and dust-continuum code; its formal solution I = I₀e^{−τ} + S(1−e^{−τ}) (for constant S) is the single most-used formula in ISM spectroscopy.
Key referencesSchwarzschild (1906); Chandrasekhar (1960); Rybicki & Lightman (1979).
Optical Depth & Column Density \[ \tau_\nu = \int \kappa_\nu\, ds = N\,\sigma_\nu \]
Opacity integrated along the line of sight — the e-folding count of how buried a source is. For a uniform absorber it collapses to column density times cross-section: the bridge from what we see to how much is there.
 N = column density (cm⁻²); σ_ν = cross-section per particle; τ = 1 defines every "photosphere"
The universal currency of ISM measurement: every abundance, mass, and extinction is an N inferred through some σ. The art of the field is finding transitions whose σ is known and whose τ is measurable.
Key referencesRybicki & Lightman (1979); Draine (2011).
Brightness Temperature \[ T_B \equiv \frac{c^2}{2k_{\rm B}\nu^2}\, I_\nu \quad (h\nu \ll k_{\rm B}T) \]
In the radio regime every intensity can be quoted as the temperature of the blackbody that would produce it. For optically thick thermal lines T_B is the gas temperature itself — radio astronomy's thermometer convention.
 I_ν = intensity; Rayleigh–Jeans limit; for a thermal line T_B = T(1−e^{−τ})
The native unit of all radio/mm observations (kelvin, not janskys, for extended emission); the T_B of optically thick CO reads cloud temperatures, and 21-cm absorption/emission pairs split HI into its cold and warm phases.
Key referencesRayleigh (1900); Dickey & Lockman (1990); Heiles & Troland (2003).
The 21-cm Line \[ N_{\rm HI} = 1.823\times10^{18} \int T_B\, dv \ \ {\rm cm^{-2}} \]
The hyperfine flip of hydrogen's electron spin — a transition so forbidden each atom waits ~11 Myr to emit — becomes, summed over interstellar columns, the most information-rich line in astronomy. Integrate the profile and you have weighed the hydrogen.
 T_B in K, v in km/s; valid for τ ≪ 1; harmonic-mean spin temperature corrections when thick
Mapped the Galaxy's spiral arms, warp, and rotation curve (the dark-matter problem's original data); modern all-sky surveys (HI4PI) and the 21-cm forest / cosmic-dawn signal extend the same physics from the local ISM to redshift 20.
Key referencesvan de Hulst (1945); Ewen & Purcell (1951); HI4PI Collaboration (2016).
Spin Temperature \[ \frac{n_1}{n_0} = 3\,e^{-T_*/T_{\rm spin}} \quad (T_* = 0.068\ {\rm K}) \]
The level ratio of the hyperfine states defines an excitation temperature that need not equal anything physical — it is set by a competition among collisions, the radio background, and Lyman-α scattering (the Wouthuysen–Field effect). Whether T_spin tracks the gas decides whether hydrogen is visible at all.
 T_spin = excitation temperature of the 21-cm doublet; collisions couple it to T_kin at CNM densities; Lyα pumping couples it in low-density gas
In the Galaxy, T_spin ≈ T_kin (dense enough) — so absorption measures real temperatures. At cosmic dawn the coupling history is the signal: the global 21-cm absorption trough encodes when the first stars' Lyα switched the coupling on.
Key referencesWouthuysen (1952); Field (1958); Bowman et al. (2018); Singh et al. (2022).
Emission Measure & Free–Free \[ {\rm EM} = \int n_e^2\, ds; \qquad \tau_{\rm ff} \propto\, {\rm EM}\ T^{-1.35}\nu^{-2.1} \]
Ionized gas announces itself by braking radiation: electrons deflected by ions emit a continuum whose strength counts electron–ion encounters — the square of the density, integrated along the ray. Hα emission obeys the same EM.
 n_e = electron density; EM in cm⁻⁶ pc; free–free is the flat-spectrum radio continuum of every H II region
The mass- and density-meter of ionized gas: radio free–free gives extinction-free SFRs, Hα surveys (WHAM) map the warm ionized layer, and the n_e² weighting (versus DM's n_e, Section VIII) measures how clumpy the ionized ISM is.
Key referencesOster (1961); Reynolds (1989); Haffner et al. (2003); Condon & Ransom (2016).
Open unknowns · Radiative Transfer & 21 cm
The Cosmic-Dawn Trough
Was the EDGES 21-cm absorption real — and if so, what cooled the gas?
A confirmed extra-deep trough requires baryons colder than adiabatic cooling allows (dark-matter scattering?) or an added radio background. SARAS 3 rejected the EDGES profile at 95%; next-generation global-signal and SKA experiments carry the question.
Self-Absorption Blind Spots
How much cold HI hides inside its own emission?
HI self-absorption and optically thick CNM bias column densities low by tens of percent along cold sightlines — the "opacity correction" to the Galactic gas census is still argued at the 10–30% level, propagating directly into dust-to-gas and X_CO calibrations.
The Wouthuysen–Field Era in Detail
Can 21-cm tomography actually deliver the first-stars power spectrum?
Foregrounds outshine the cosmological signal by 10⁴–10⁵; separating them relies on spectral smoothness assumptions that instrument systematics violate. HERA and SKA-Low are in a decade-long fight with this — the payoff (the entire z ~ 6–30 baryon history) justifies it.
III

Absorption-Line Diagnostics

5 equations

Point a spectrograph at a bright star or quasar and the intervening medium prints its inventory onto the continuum — element by element, cloud by cloud, at column densities emission could never reach. Absorption spectroscopy is the ISM's accounting department, from the local clouds to damped Lyman-α systems at high redshift.

NameEquationVariablesUse in Research
Equivalent Width \[ W_\lambda = \int \left(1 - \frac{F_\lambda}{F_{\rm cont}}\right) d\lambda \]
The area a line removes from the continuum, expressed as the width of a perfectly black rectangle — a measure immune to instrumental blurring. Whatever the spectrograph does to the profile, the missing light is conserved.
 F_λ = observed flux; F_cont = interpolated continuum; W in mÅ for ISM lines
The primary observable of absorption spectroscopy since the 1930s; resolution-independent, so 1930s photographic W values remain usable beside HST/STIS echelle data a century later.
Key referencesHartmann (1904); Adams (1949); Spitzer (1978).
The Curve of Growth \[ W \propto N \;\to\; \sqrt{\ln N} \;\to\; \sqrt{N} \]
How a line's strength grows with column: linearly while thin, logarithmically once the core saturates, then as √N when damping wings take over. Three regimes, three different sensitivities — and a trap for anyone who assumes the first.
 N = column density; regimes: linear (τ₀ ≲ 1), flat/saturated, damped; b-parameter controls the flat part
The decoder of every absorption survey: measure W for several lines of one ion with different fλ, place them on the curve, and N and b come out together. Saturated lines on the flat part are where abundance controversies go to hide.
Key referencesMinnaert (1930); Strömgren (1948); Spitzer (1978); Savage & Sembach (1996).
The Voigt Profile \[ \phi(\nu) = H(a, x): \ {\rm Gaussian\ core} * {\rm Lorentzian\ wings} \]
Every absorption line is a thermal/turbulent Gaussian convolved with the Lorentzian of natural broadening. The core carries the velocity dispersion (b-parameter); the far wings remember only quantum mechanics — and thus only N.
 a = damping/Doppler ratio; x = offset in Doppler units; b = √(2kT/m + b²_turb)
The fitting function of all quantitative absorption work (VPFIT and descendants); comparing b across species of different mass separates thermal from turbulent broadening — one of the few direct ISM temperature measurements that needs no excitation modeling.
Key referencesVoigt (1912); Carswell & Webb (2014, VPFIT); Draine (2011).
Damped Lyman-α Columns \[ N_{\rm HI} \gtrsim 2\times10^{20}\,{\rm cm^{-2}} \;\Rightarrow\; {\rm DLA} \]
At high column the Lyα line's damping wings dominate everything, and N reads directly off the wing shape regardless of saturation or kinematics. A DLA is a galaxy's gas disk backlit by a quasar — weighed by quantum mechanics alone.
 N_HI from Lorentzian wing fit; DLAs hold most neutral gas at every redshift
The census tool for cosmic neutral gas: Ω_HI(z) from DLA surveys tracks the fuel supply of cosmic star formation, and DLA metallicities are the chemical history of ordinary galaxies measured without ever seeing the galaxy.
Key referencesWolfe et al. (1986, 2005); Prochaska & Wolfe (2009); Linsky et al. (2019).
Elemental Depletion \[ \delta({\rm X}) = \log\!\left(\frac{N_{\rm X}/N_{\rm H}}{({\rm X/H})_\odot}\right) \]
Interstellar gas is missing metals relative to the Sun — because they are not gone, they are solid. The deficit pattern, element by element, is a chemical assay of the dust performed on the gas.
 δ(X) = logarithmic depletion; refractory elements (Fe, Ti, Ca) depleted 1–3 dex, volatiles (O, N, Zn) mildly
The only way to inventory dust composition sightline by sightline; depletions strengthen with cloud density (grain growth in situ) and weaken in shocked gas (sputtering) — a live tracer of the dust life cycle, systematized by Jenkins's F* parameter.
Key referencesField (1974); Savage & Sembach (1996); Jenkins (2009).
Open unknowns · Absorption-Line Diagnostics
The Diffuse Interstellar Bands
500+ absorption bands, known for a century — carried by what?
Only one DIB carrier is identified (C₆₀⁺, four bands, confirmed 2015). The rest — presumably large carbon molecules — remain the longest-standing unsolved identification problem in spectroscopy. Whoever solves it inventories a major organic reservoir of the Galaxy.
Missing Oxygen
Gas + known dust doesn't add up to the expected O budget in dense sightlines.
Up to ~20–30% of oxygen is unaccounted for toward dense clouds — not in gas, not in silicates or known ices. Large O-bearing organics? Unknown ice mantles? JWST ice surveys are the current best hope for closing the books.
Small-Scale Structure
Why does the ISM show AU-scale column variations it shouldn't sustain?
Binary-star and multi-epoch pulsar absorption reveal structure down to tens of AU with implied overpressures of 10–100×. Sheets and filaments seen edge-on, or genuine overpressured clumps? The geometry is still unresolved after three decades.
IV

H II Regions & Ionized Gas

6 equations

Around every O star the ISM is ionized, heated to 10⁴ K, and lit up in emission lines — nature's own plasma diagnostics laboratory. H II regions mark star formation across the universe, and their forbidden lines are how densities, temperatures, and abundances are measured everywhere from Orion to redshift 10.

NameEquationVariablesUse in Research
Strömgren Radius \[ R_S = \left(\frac{3\,Q_{\rm H}}{4\pi\, n^2\, \alpha_B}\right)^{1/3} \]
A hot star ionizes exactly as much gas as its photon output can keep ionized against recombination — a sphere with a razor edge, since the transition from ionized to neutral happens over one photon mean free path. Ionization is a budget, not a gradient.
 Q_H = ionizing photons/s (O5V: ~10⁴⁹·⁵); α_B ≈ 2.6×10⁻¹³ cm³/s at 10⁴ K; n = density
The zeroth-order model of every H II region and the reason ionized nebulae have sharp rims; generalized (with dust and winds) it sets the size–density relation used to classify ultracompact through giant H II regions.
Key referencesStrömgren (1939); Osterbrock & Ferland (2006); Churchwell (2002).
Photoionization Equilibrium \[ n_{\rm HI} \int_{\nu_0}^{\infty} \frac{4\pi J_\nu}{h\nu}\sigma_\nu\, d\nu = n_e\, n_p\, \alpha_B \]
Inside the nebula, every ionization is balanced by a recombination — a detailed-balance ledger that fixes the tiny neutral fraction and, through it, everything the nebula emits. The recombination cascade is the source of every hydrogen line you have ever seen from a nebula.
 J_ν = mean intensity; σ_ν = photoionization cross-section; Case B: recombinations to n ≥ 2 (Lyman photons reabsorbed)
The core loop of photoionization codes (CLOUDY, MAPPINGS) used for everything from planetary nebulae to AGN narrow-line regions; Case B Hα/Hβ = 2.86 is the universal extinction ruler (the "Balmer decrement").
Key referencesBaker & Menzel (1938); Osterbrock & Ferland (2006); Ferland et al. (2017, CLOUDY).
Recombination Timescale \[ t_{\rm rec} = \frac{1}{n_e\,\alpha_B} \approx \frac{1.2\times10^{5}}{n_e}\ {\rm yr\ (n_e\ in\ cm^{-3})} \]
Switch off the star and the plasma forgets it was ionized in one recombination time — centuries for a bright nebula, tens of Myr for the diffuse ionized layer. Ionization is a lease that must be continuously renewed.
 α_B at 10⁴ K; compare t_rec to source lifetime/variability to test equilibrium
Decides which nebulae are in equilibrium (dense H II regions: yes) and which are fossils (low-density gas remembers dead sources); fossil ionization underlies quasar light echoes and the interpretation of the WIM's scale height.
Key referencesOsterbrock & Ferland (2006); Lintott et al. (2009); Haffner et al. (2009).
Density Diagnostic: [S II] Doublet \[ \frac{I(6716)}{I(6731)}: \ 1.45 \;(n_e\!\to\!0)\ \longrightarrow\ 0.44 \;(n_e\!\to\!\infty) \]
Two lines from the same ion, from levels so close their ratio can't feel temperature — but with different critical densities, so collisions rebalance them as density rises. A density meter made of two wavelengths.
 Sensitive over n_e ~ 10²–10⁴ cm⁻³; [O II] 3726/3729 covers similar range; IR fine-structure pairs extend it
The standard n_e measurement of nebular astronomy, from Galactic H II regions to JWST spectra of z > 6 galaxies (where measured densities run ~10× higher than local — a real evolution in ISM conditions).
Key referencesOsterbrock & Ferland (2006); Kewley et al. (2019); Isobe et al. (2023).
Temperature Diagnostic: [O III] Ratio \[ \frac{I(4959)+I(5007)}{I(4363)} \approx \frac{7.9\, e^{3.29\times10^4/T_e}}{1+4.5\times10^{-4} n_e/T_e^{1/2}} \]
Two [O III] lines from levels 33,000 K apart in excitation: their ratio is a Boltzmann thermometer for the electrons. The faint 4363 Å line is hard to catch — and everything abundance-related hangs on it.
 T_e = electron temperature; typical H II regions: 7,000–12,000 K, hotter at low metallicity
The foundation of the "direct method" for nebular abundances: T_e sets every collisional emissivity, so metallicities without it (strong-line methods) inherit ~0.2-dex systematic disputes. JWST now measures 4363 routinely at high z.
Key referencesMenzel, Aller & Hebb (1941); Osterbrock & Ferland (2006); Izotov et al. (1999); Curti et al. (2023).
Expansion of an H II Region \[ R(t) = R_S\left(1 + \frac{7}{4}\frac{c_{\rm II}\, t}{R_S}\right)^{4/7} \]
The 10⁴ K bubble is overpressured ~100× against its surroundings, so after ionizing it starts pushing: a D-type front driving a shock into the neutral gas. H II regions are not just lit — they excavate.
 c_II ≈ 10 km/s, ionized-gas sound speed; Spitzer solution; stalls when pressures equalize or the star dies
The workhorse model of early stellar feedback: sets pre-supernova energy/momentum injection into clouds, drives champagne flows, and possibly triggers (or truncates) the next generation of star formation at the swept-up shell.
Key referencesKahn (1954); Spitzer (1978); Deharveng et al. (2010); Krumholz et al. (2019).
Open unknowns · H II Regions
The Abundance Discrepancy
Recombination lines give 2–3× higher metal abundances than collisional lines. Which is right?
The "ADF problem" afflicts every deep nebular spectrum: O²⁺ from faint recombination lines exceeds O²⁺ from [O III] systematically. Temperature fluctuations, metal-rich inclusions, or unrecognized physics — after 80 years (Wyse noted it in 1942), unresolved, and it caps the accuracy of all nebular abundances.
Leaking Photons
What fraction of ionizing photons escape their H II regions — and their galaxies?
The WIM requires ~15% of O-star photons to travel kiloparsecs; reionization requires a few percent to escape galaxies entirely. The escape fraction is the least constrained number in reionization modeling, and it is set by the pc-scale porosity of H II region walls.
Feedback Bookkeeping
What actually destroys molecular clouds — photoionization, winds, radiation pressure, or SNe?
Clouds disperse within ~10–30 Myr with star-formation efficiencies of only a few percent, and pre-SN feedback evidently dominates — but the division of labor among mechanisms flips between simulations. JWST + ALMA cloud-scale surveys (e.g. PHANGS) are adjudicating now.
V

Interstellar Dust

6 equations

One percent of the ISM's mass is in solid grains — and that one percent absorbs half the starlight ever emitted, catalyzes the molecule that makes clouds molecular, carries the heavy elements, and re-emits the Galaxy's energy budget in the infrared. Dust is the ISM's minority partner with veto power.

NameEquationVariablesUse in Research
Extinction & Reddening \[ A_\lambda = 1.086\,\tau_\lambda; \qquad R_V = \frac{A_V}{E(B-V)} \approx 3.1 \]
Dust dims in magnitudes proportional to optical depth, and dims blue light more than red — so stars behind dust are both faint and reddened. One parameter, R_V, summarizes the grain-size mix along the path.
 E(B−V) = color excess; R_V ≈ 3.1 diffuse ISM, → 5–6 in dense clouds (grain growth)
The correction applied to essentially every photometric measurement in astronomy; the full extinction curve A_λ (2175 Å bump, UV rise, IR power law) is the primary constraint on grain composition and sizes. Rajiv's own VarWISE dereddening pipeline lives downstream of exactly this physics.
Key referencesTrumpler (1930); Cardelli, Clayton & Mathis (1989); Fitzpatrick (1999); Schlafly & Finkbeiner (2011).
Gas-to-Dust Ratio \[ \frac{N_{\rm H}}{E(B-V)} \approx 5.8\times10^{21}\ {\rm cm^{-2}\,mag^{-1}} \]
Hydrogen column and dust reddening march in lockstep across the diffuse Galaxy — gas and dust are well mixed, at ~100:1 by mass. Measure the reddening of anything and you have weighed the hydrogen in front of it.
 Bohlin ratio (Copernicus/FUSE Lyα + H₂ vs. E(B−V)); rises in low-metallicity galaxies (less dust per H)
The conversion underlying every dust-based gas map (SFD, Planck, Bayestar, Edenhofer) and X-ray absorption column; its constancy is the empirical anchor of "dust traces gas," and its metallicity scaling calibrates ISM masses in other galaxies.
Key referencesBohlin, Savage & Drake (1978); Rachford et al. (2009); Rémy-Ruyer et al. (2014).
Grain Size Distribution (MRN) \[ \frac{dn}{da} \propto a^{-3.5}, \quad 0.005 \lesssim a \lesssim 0.25\ \mu{\rm m} \]
The extinction curve demands grains of all sizes with steeply more small ones — the fingerprint of a collisional cascade (compare Dohnanyi's −3.5 for asteroids on the planets sheet; same exponent, same physics, 10¹² times smaller). Most surface area in the smallest grains, most mass near the largest.
 a = grain radius; silicate + carbonaceous populations; extended at the small end by PAHs (~nm)
The default grain model of ISM physics for five decades (with Draine-era refinements); the small-grain end controls photoelectric heating and the mid-IR PAH bands that JWST now maps in every nearby galaxy.
Key referencesMathis, Rumpl & Nordsieck (1977); Draine & Li (2007); Witstok et al. (2023).
Dust Temperature & Thermal Emission \[ T_d \approx 16.4\, G_0^{1/6}\ {\rm K}; \qquad S_\nu \propto \nu^{\beta} B_\nu(T_d) \]
A grain balances absorbed starlight against its own inefficient far-IR radiation, sitting near 20 K in the diffuse ISM almost regardless of the field (the 1/6 power). Half the light ever emitted by stars has been reprocessed through this equation.
 G₀ = FUV field; β ≈ 1.5–2, emissivity index; modified blackbody peaks ~150 µm locally
The basis of all far-IR/submm dust masses (Herschel, Planck, ALMA continuum) and of the cosmic infrared background; β–T degeneracies in the fit are the standard systematic in every dust-mass paper.
Key referencesDraine & Lee (1984); Planck Collaboration XI (2014); Casey, Narayanan & Cooray (2014).
H₂ Formation on Grains \[ R_{\rm H_2} \approx 3\times10^{-17}\, n\, n_{\rm HI}\ \ {\rm cm^{-3}\,s^{-1}} \]
Two H atoms meeting in the gas cannot radiate away their binding energy — molecular hydrogen essentially cannot form in the gas phase. Grain surfaces hold one atom until the next arrives: every H₂ molecule in the universe since the first stars was born on a dust grain.
 n = total H density; rate coefficient from grain surface area × sticking; measured via FUSE H₂ absorption
The rate that gates the atomic→molecular transition (with self-shielding, Section VI): sets the HI column "ceiling" of ~10²⁰·⁷ cm⁻² at solar metallicity before clouds turn molecular, visible directly in HI saturation in nearby galaxies.
Key referencesGould & Salpeter (1963); Hollenbach & Salpeter (1971); Jura (1975); Wakelam et al. (2017).
Polarization & Aligned Grains \[ \frac{p_{\rm max}}{A_V} \lesssim 3\ \%/{\rm mag} \quad ({\rm Serkowski}) \]
Starlight emerges from dusty sightlines a few percent polarized: spinning non-spherical grains align their short axes with the magnetic field and preferentially block one polarization. Dust turns the ISM into a polarizer — and the polarization into a compass.
 p = polarization fraction; Serkowski law peaks near 0.55 µm; alignment via radiative torques (RATs)
The basis of all dust-based magnetic-field mapping: optical/IR polarization (absorption) and Planck/ALMA polarized emission trace plane-of-sky field orientation from the whole sky down to protostellar disks — feeding the Chandrasekhar–Fermi method (Section VIII).
Key referencesHiltner (1949); Hall (1949); Serkowski, Mathewson & Ford (1975); Lazarian & Hoang (2007); Planck Collaboration XXXV (2016).
Open unknowns · Dust
Where Is Dust Actually Made?
Stellar winds and supernovae can't obviously supply what sputtering destroys.
Grain destruction timescales (~few ×10⁸ yr) are shorter than injection timescales from AGB winds and SNe — implying most dust mass must regrow in the ISM itself, on seeds, by accretion. How grain growth works in cold clouds at 10 K is essentially open, and JWST keeps finding mature dust at z > 6 where AGB stars haven't had time.
What Carries the 2175 Å Bump?
The strongest dust feature still lacks a confirmed carrier.
Graphitic carbon? PAH mixtures? The bump's strength varies between sightlines and vanishes in some galaxies (SMC bar), yet its central wavelength barely moves — a strong constraint no proposed carrier fully satisfies.
Anomalous Microwave Emission
What emits the 20–60 GHz excess correlated with dust?
Spinning nanoparticles (PAHs? nanosilicates?) are the leading candidate, but AME strength doesn't track PAH abundance as first predicted. It matters beyond dust physics: AME is a foreground for CMB B-mode searches.
VI

Molecular Clouds & Astrochemistry

6 equations

Where dust shields the gas from starlight, chemistry switches on: hydrogen pairs up, CO forms, and over 300 molecules — up to buckyballs and branched organics — assemble at 10 K in a vacuum. Molecular clouds hold the Galaxy's star-forming reservoir, observed almost entirely through trace species standing in for invisible H₂.

NameEquationVariablesUse in Research
H₂ Photodissociation & Self-Shielding \[ f_{\rm shield}(N_{\rm H_2}) \;\to\; \left(\frac{N_{\rm H_2}}{10^{14}\,{\rm cm^{-2}}}\right)^{-3/4} \]
H₂ is destroyed by UV line absorption, not continuum — so once a thin skin of H₂ builds up, it absorbs those exact lines and shields everything behind it. The atomic-to-molecular transition is therefore a switch: sharp, column-triggered, almost binary.
 N_H₂ = molecular column; dissociation via Lyman–Werner bands (91.2–110 nm); dust shielding assists
Sets where clouds turn molecular: at solar metallicity the HI→H₂ transition completes near Σ ≈ 10 M☉/pc² — the observed HI saturation ceiling in galaxies, and the gateway condition for all star formation.
Key referencesHollenbach, Werner & Salpeter (1971); Draine & Bertoldi (1996); Savage et al. (1977); Krumholz, McKee & Tumlinson (2009).
The X-Factor (CO → H₂) \[ N_{\rm H_2} = X_{\rm CO}\, W_{\rm CO}, \quad X_{\rm CO} \approx 2\times10^{20}\ \frac{\rm cm^{-2}}{\rm K\,km\,s^{-1}} \]
H₂ is invisible in cold clouds (no dipole, first line needs 500 K), so the field counts CO photons instead and multiplies by a calibration constant. That the optically thick CO line counts mass at all works only because line width tracks virial mass — a fortunate conspiracy.
 W_CO = integrated CO(1–0) intensity; X_CO ×2–10 higher at low metallicity; ~5× lower in starburst centers
The conversion behind every molecular-gas mass in the literature — Galactic cloud catalogs, the Kennicutt–Schmidt law, high-z gas fractions. Its systematic uncertainties (metallicity, "CO-dark" envelopes) propagate into all of them.
Key referencesDickman (1978); Solomon et al. (1987); Bolatto, Wolfire & Leroy (2013); Grenier et al. (2005).
Critical Density \[ n_{\rm crit} = \frac{A_{ul}}{\gamma_{ul}} \]
Below n_crit, collisions can't keep up with radiative decay and a line barely emits; above it, the level thermalizes and the line saturates as a tracer. Every molecule is a density-selective filter — choose the transition, choose the gas you see.
 A_ul = Einstein coefficient; γ_ul = collision rate coefficient; radiative trapping lowers the effective value ~10×
The organizing principle of molecular spectroscopy: CO(1–0) (n_crit ~ 2×10³, effective ~10²) maps whole clouds; HCN(1–0) (~few ×10⁵ effective) picks out dense cores — and the HCN/CO ratio has become the standard "dense-gas fraction" in galaxies.
Key referencesEvans (1999); Gao & Solomon (2004); Shirley (2015); Leroy et al. (2017).
Cosmic-Ray Ionization \[ \zeta_{\rm H_2} \approx 2\!-\!5\times10^{-16}\ {\rm s^{-1}} \ ({\rm diffuse});\ \sim\!10^{-17}\ ({\rm dense}) \]
Starlight cannot reach cloud interiors, but ~GeV cosmic rays can: they maintain a trace ionization of ~10⁻⁷–10⁻⁸ that couples the gas to magnetic fields and ignites all ion–neutral chemistry. The faintest touch of high-energy physics runs the coldest chemistry in the universe.
 ζ = ionization rate per H₂; measured via H₃⁺ absorption (diffuse) and DCO⁺/HCO⁺ (dense)
Controls the ionization fraction that sets ambipolar diffusion (Section X), the chemistry clock, and cloud heating at depth; the order-of-magnitude drop from diffuse to dense gas (CR exclusion or energy loss) is itself a probe of low-energy cosmic rays no direct detector can see.
Key referencesHayakawa et al. (1961); Herbst & Klemperer (1973); McCall et al. (2003); Indriolo & McCall (2012).
Ion–Molecule Chemistry \[ k_{\rm L} = 2\pi e\sqrt{\alpha_{\rm pol}/\mu}\ \approx 10^{-9}\ {\rm cm^3\,s^{-1}} \]
Neutral–neutral reactions freeze out at 10 K (activation barriers), but an ion polarizes and captures a passing molecule with no barrier at all — the Langevin rate, temperature-independent. Cold interstellar chemistry runs on charge.
 α_pol = polarizability; μ = reduced mass; network seeded by ζ (previous row) via H₃⁺
The backbone rate of astrochemical networks (UMIST, KIDA: ~10⁴ reactions): H₃⁺ + CO → HCO⁺, then dissociative recombination, builds most observed species; deuterium fractionation through H₂D⁺ turns the same chemistry into a cloud thermometer and clock.
Key referencesLangevin (1905); Watson (1973); Herbst & Klemperer (1973); Caselli & Ceccarelli (2012).
Freeze-Out onto Grains \[ t_{\rm fo} \approx \frac{3\times10^{9}}{n/{\rm cm^{-3}}}\ {\rm yr} \]
Every molecule that strikes a 10 K grain sticks. In dense cores the gas plates itself onto the dust in under a Myr — CO vanishes from the gas, ice mantles grow, and the grain surfaces become the chemical factories (H₂O, CH₃OH, complex organics).
 n = density; desorption (thermal, CR spot heating, UV) opposes it; mantles released intact near protostars
Explains CO "depletion holes" in prestellar cores (why N₂H⁺ and NH₃ map what CO cannot), the composition of cometary ices, and — via hot-core sublimation — the complex-organic inventory JWST reads in protostellar spectra.
Key referencesLéger (1983); Caselli et al. (1999); Bergin & Tafalla (2007); McClure et al. (2023).
Open unknowns · Molecular Clouds & Astrochemistry
CO-Dark Gas
A third to a half of the Galaxy's H₂ has no CO. What is its structure and role?
γ-rays, dust, and [C II] all reveal molecular gas outside CO's reach, but its distribution, lifetime, and star-forming relevance are open — and every extragalactic gas mass built on X_CO inherits the blind spot at unknown scale.
Cloud Lifetimes
Do molecular clouds live 1 free-fall time or 10?
Rapid-cycling pictures (clouds as transient turbulent structures) and slow, magnetically regulated pictures differ by an order of magnitude in lifetime. PHANGS-style CO/Hα offset statistics favor ~10–30 Myr — short, but the Galactic consensus is still forming.
The Origin of Complex Organics
How far does interstellar chemistry climb toward prebiotic molecules?
Glycolaldehyde, urea, and branched molecules are detected; amino acids are not (in the gas), yet arrive in meteorites. Whether the ISM delivers prebiotic inventory to planets, or planets start over, connects this section directly to the habitability questions on the planets sheet.
VII

Interstellar Turbulence

5 equations

Every ISM velocity field ever measured is turbulent — and supersonically so in the cold phases, where random motions outrun sound tenfold. Turbulence broadens every line, structures every cloud, resists gravity on large scales while creating the seeds of collapse on small ones. It is the ISM's default state of motion, continuously refueled by feedback and instabilities.

NameEquationVariablesUse in Research
The Turbulent Cascade \[ E(k) \propto k^{-5/3}\ ({\rm Kolmogorov}) \;\to\; k^{-2}\ ({\rm Burgers}) \]
Energy injected at large scales cascades to small ones where viscosity finally eats it, filling the inertial range with a power law: −5/3 for incompressible eddies, steepening toward −2 when supersonic shocks do the transferring. Structure at every scale, from one law.
 k = wavenumber; injection at ~10–100 pc (SNe, instabilities), dissipation at ~milli-pc (ambipolar/viscous)
The framework for interpreting all ISM power spectra; the observed electron-density spectrum follows −5/3 over 10 decades of scale (the "Big Power Law in the Sky") — among the largest dynamic ranges of any measured physical law.
Key referencesKolmogorov (1941); Armstrong, Rickett & Spangler (1995); Chepurnov & Lazarian (2010); Federrath (2013).
Larson's Relations \[ \sigma_v \approx 1\,\Big(\frac{L}{\rm pc}\Big)^{0.5}\ {\rm km\,s^{-1}} \]
Bigger clouds move faster inside — line width grows as the square root of size, exactly what a supersonic cascade predicts for velocity differences across a scale L. Clouds are not objects with temperatures; they are eddies with sizes.
 σ_v = 1D velocity dispersion; coefficient rises with surface density (Heyer relation: σ²/L ∝ Σ)
The empirical spine of cloud physics: with the size–mass relation it implies clouds sit near virial balance, and its Σ-dependence (Heyer) is the diagnostic separating turbulence-supported from gravity-dominated regions across Galactic and extragalactic surveys.
Key referencesLarson (1981); Solomon et al. (1987); Heyer et al. (2009); Goodman et al. (1998).
Density Variance–Mach Relation \[ \sigma^2_{\ln\rho} = \ln\!\left(1 + b^2 \mathcal{M}^2\right) \]
Supersonic turbulence shocks gas into a lognormal density distribution whose width grows with Mach number — solenoidal stirring (b ≈ 1/3) makes narrow distributions, compressive driving (b ≈ 1) broad ones. Turbulence doesn't just stir clouds; it manufactures their density contrast.
 ℳ = sonic Mach number; b = driving parameter; magnetic fields reduce the variance
The statistical bridge from turbulence to star formation: analytic SF theories (Krumholz–McKee, Padoan–Nordlund, Hennebelle–Chabrier) integrate this PDF above a collapse threshold to predict star-formation rates — and possibly the IMF's lognormal body.
Key referencesVazquez-Semadeni (1994); Padoan, Nordlund & Jones (1997); Federrath et al. (2008); Kainulainen et al. (2009).
Virial Parameter \[ \alpha_{\rm vir} = \frac{5\,\sigma_v^2\, R}{G\,M} \]
Twice kinetic over gravitational energy: above ~2, motions can disperse the cloud; near 1, gravity has caught it. One dimensionless number triages every structure in the ISM into bound and unbound.
 σ_v = 1D dispersion; R, M = radius, mass; magnetic terms add support (Section VIII)
The standard boundedness diagnostic of cloud and core catalogs (from Galactic-plane CO surveys to ALMA core censuses); its distribution — most clouds marginally bound, most small structures unbound — frames the "what fraction of a cloud actually collapses" question.
Key referencesBertoldi & McKee (1992); Kauffmann, Pillai & Goldsmith (2013); Miville-Deschênes et al. (2017).
Turbulent Decay \[ t_{\rm decay} \approx \frac{L_{\rm drive}}{\sigma_v} \approx t_{\rm cross} \]
Supersonic turbulence dies in one crossing time — shocks are efficient dissipators, and magnetic fields do not save it (MHD simulations killed that hope in 1998). Whatever stirs the ISM must keep stirring, forever.
 L_drive = driving scale; t_cross ~ 10 Myr for a 20-pc cloud at 2 km/s
The result that reframed cloud physics: turbulent support is temporary, so clouds either re-energize (feedback, accretion), collapse, or disperse within ~10 Myr — the theoretical foundation of the short-cloud-lifetime picture and of feedback-regulated star formation generally.
Key referencesMac Low et al. (1998); Stone, Ostriker & Gammie (1998); Mac Low & Klessen (2004); Elmegreen & Scalo (2004).
Open unknowns · Turbulence
What Drives It?
Supernovae, spiral shear, accretion, gravity itself — in what proportions, at what scales?
The cascade is measured; its energy sources are apportioned only by inference. Outer-disk turbulence persists where supernovae are rare (accretion-driven? magnetorotational?), and whether cloud-internal turbulence is driven from outside or generated by collapse changes the meaning of every α_vir.
Where Does It Die?
How and where is turbulent energy finally dissipated — and does it leave a chemical signature?
Intermittent dissipation in shocks and vortex sheets should locally heat gas to ~10³ K, driving endothermic chemistry (CH⁺, SH⁺) that cold clouds can't explain. The observed CH⁺ abundance problem may be turbulence's dissipation made visible — a 40-year-old clue still not fully cashed in.
Turbulence and the IMF
Does the turbulent density spectrum set the stellar initial mass function?
Competing analytic theories derive core mass functions from the lognormal-plus-tail PDF and each reproduces the Salpeter slope under its own assumptions. Whether the IMF is inherited from turbulence, set by accretion, or regulated by feedback remains one of star formation's central questions (see the IMF gap node on the master map).
VIII

Magnetic Fields & Interstellar Plasma

6 equations

The ISM is a plasma threaded by a few-microgauss field whose energy density rivals everything else in the medium. Invisible directly, it is measured through five different back doors — Zeeman splitting, Faraday rotation, dispersion, polarized dust, and synchrotron — each returning a different projection of the same field.

NameEquationVariablesUse in Research
Flux Freezing \[ \frac{\partial \vec B}{\partial t} = \nabla\times(\vec v \times \vec B) \]
In a conducting fluid the field lines move with the gas — compress the gas, compress the field. Even the ISM's wisp of ionization (10⁻⁴–10⁻⁷) is enough: matter and field are handcuffed nearly everywhere, and the exceptions (Section X) are where stars get made.
 Ideal MHD induction equation; breaks down via ambipolar diffusion and reconnection at small scales
The zeroth law of ISM magnetism: predicts B–n scaling under compression (B ∝ n²ᐟ³ isotropic, B ∝ n along-field flows) — the framework against which every Zeeman survey is compared, and the origin of the magnetic-flux problem of star formation.
Key referencesAlfvén (1942); Mestel & Spitzer (1956); Crutcher (2012).
Alfvén Speed \[ v_{\rm A} = \frac{B}{\sqrt{4\pi\rho}} \]
The speed at which magnetic news travels — waves running along field lines like plucked strings. Where v_A exceeds the sound speed, magnetic forces, not pressure, carry the signals; in the cold ISM that is everywhere.
 B = field strength; ρ = mass density (neutrals included, if coupled)
Sets the magnetosonic speeds that classify every ISM shock (Section IX), the Alfvén Mach number of turbulence (super- vs sub-Alfvénic cascades differ qualitatively), and cosmic-ray scattering rates.
Key referencesAlfvén (1942); Heiles & Troland (2005); Federrath (2016).
Zeeman Splitting \[ \Delta\nu_Z = \frac{b_Z\, B_\parallel}{2\pi}\quad (2.8\ {\rm Hz/\mu G\ for\ HI}) \]
A magnetic field splits spectral lines into polarized components — for interstellar µG fields, by hertz, buried a thousandfold beneath the line width. Detected via circular polarization differencing: the only direct measurement of interstellar field strength that exists.
 B_∥ = line-of-sight field; species: HI (21 cm), OH, CN, masers for dense gas
The calibration anchor of all ISM magnetism: the B–n relation, mass-to-flux ratios (Section X), and the field strengths quoted for CNM (~6 µG median) all trace to a few hundred painstaking Zeeman detections.
Key referencesZeeman (1897); Verschuur (1968); Heiles & Troland (2005); Crutcher (2012).
Dispersion & Rotation Measure \[ {\rm DM} = \int n_e\, dl; \qquad {\rm RM} = 0.81\!\int n_e B_\parallel\, dl \]
Radio pulses arrive later at lower frequencies in proportion to the electrons passed (DM), and their polarization angle corkscrews in proportion to electrons-times-field (RM). Every pulsar and fast radio burst is an unpaid survey probe of the magnetoionic medium.
 DM in pc cm⁻³; RM in rad m⁻²; RM/DM → electron-weighted mean B_∥
The tomographic backbone of Galactic magnetism (pulsar RM grids map the spiral field and its reversals) and of electron-density models (NE2001, YMW16) that convert every new pulsar's DM into a distance.
Key referencesHewish et al. (1968); Manchester (1972); Cordes & Lazio (2002); Macquart et al. (2020).
Chandrasekhar–Fermi Method \[ B_{\rm pos} \approx \sqrt{4\pi\rho}\; \frac{\sigma_v}{\sigma_\theta} \]
Turbulence shakes field lines; the shake shows up as scatter in polarization angles. Stiff fields scatter little — so the straighter the polarization map, the stronger the field. Strength inferred from tidiness.
 σ_θ = polarization-angle dispersion; σ_v = velocity dispersion; calibration factor ~0.5 from simulations
The only field-strength method that works from dust polarization alone — now applied from Planck's diffuse ISM through ALMA maps of protostellar cores, wherever Zeeman is too slow or too insensitive.
Key referencesChandrasekhar & Fermi (1953); Ostriker, Stone & Gammie (2001); Pattle et al. (2023).
Plasma Beta & Synchrotron Emission \[ \beta = \frac{8\pi P_{\rm th}}{B^2}; \qquad \epsilon_{\rm sync} \propto n_{\rm CR}\, B^{1+\alpha} \]
β compares gas pressure to magnetic pressure — below one, the field bosses the gas, and the cold ISM lives there. The field also lights itself up: cosmic-ray electrons spiraling in it radiate synchrotron, making the magnetized ISM glow at radio wavelengths.
 β ≈ 0.3–1 in CNM, ≪1 in molecular gas, >1 in HIM; α = CR electron spectral index
β classifies which phases are magnetically dominated; synchrotron (with equipartition assumptions) maps B in the Milky Way and external galaxies, feeds CMB-foreground models, and made the radio sky the field's first all-sky magnetogram.
Key referencesJansky (1933); Ginzburg & Syrovatskii (1965); Beck (2015); Han (2017).
Open unknowns · Magnetic Fields
Origin of the Galactic Field
Primordial seed amplified by dynamo — or something faster?
The α–Ω dynamo needs ~Gyr to build µG fields from tiny seeds, yet µG fields are observed in galaxies at z > 2. Turbulent (small-scale) dynamos amplify faster but produce tangled fields; the observed large-scale spiral ordering with reversals fits no complete story.
Reconnection in the ISM
How does a nearly ideal plasma change its field topology as fast as it evidently does?
Classical (Sweet–Parker) reconnection is absurdly slow at ISM conductivities, yet field topology must evolve for dynamos and turbulence to work. Turbulence-mediated fast reconnection is the leading answer — with direct observational tests still scarce.
Does B Matter for Star Formation?
Regulator, spectator, or something in between?
Zeeman statistics say clouds are (marginally) supercritical — gravity wins — yet simulations show even weak fields change fragmentation, disk formation, and outflows qualitatively. Thirty years after the ambipolar-diffusion paradigm fell, the field's true job description is still being written.
IX

Shocks & Supernova Remnants

6 equations

Supersonic turbulence, expanding H II regions, stellar winds, and above all supernovae mean the ISM is constantly being shocked — irreversibly heated, compressed, and chemically reset at moving discontinuities. Supernova remnants are the flagship case: 10⁵¹ erg working its way through every phase of the medium, accelerating cosmic rays as it goes.

NameEquationVariablesUse in Research
Shock Jump Conditions \[ \frac{\rho_2}{\rho_1} \to 4; \qquad T_2 = \frac{3\,\mu m_{\rm H}}{16\, k_{\rm B}}\, v_s^2 \]
Mass, momentum, and energy conservation across a discontinuity (Rankine–Hugoniot) fix everything: a strong adiabatic shock compresses gas fourfold — no more — and converts a fixed fraction of ram energy to heat. Measure a postshock temperature and you have measured the shock speed.
 v_s = shock speed; γ = 5/3; radiative shocks compress far beyond 4 once cooling deflates them
The entry point of all shock diagnostics: X-ray temperatures of SNR rims give blast speeds, optical line ratios classify radiative shocks, and the ×4 ceiling explains why magnetic fields (compressed with the gas) become dynamically important behind strong shocks.
Key referencesRankine (1870); Hugoniot (1889); Zel'dovich & Raizer (1967); Ghavamian et al. (2013).
Sedov–Taylor Blast Wave \[ R(t) = 1.15\left(\frac{E\,t^2}{\rho_0}\right)^{1/5} \]
Once a blast has swept up more mass than it ejected, only E and the ambient density matter — the expansion forgets the explosion's details and follows a universal t²ᐟ⁵. Dimensional analysis alone fixes the law; the physics only supplies the 1.15.
 E ≈ 10⁵¹ erg; ρ₀ = ambient density; valid between free expansion and the radiative phase
The standard clock and energy-meter of observed SNRs: radius + temperature → age and E, the consistency check applied to every X-ray remnant. Taylor famously derived the yield of the Trinity test from declassified fireball photos using exactly this law.
Key referencesTaylor (1950); Sedov (1959); Ostriker & McKee (1988); Truelove & McKee (1999).
Radiative Phase & Momentum Injection \[ p_{\rm SN} \approx 3\times10^{5}\ M_\odot\,{\rm km\,s^{-1}} \ \ ({\rm per\ SN}) \]
When the postshock gas cools faster than it expands (~0.2 Myr in average gas), the energy-conserving phase ends — thermal energy radiates away, but momentum cannot. The surviving momentum, remarkably insensitive to environment, is feedback's durable currency.
 Snowplow phase; onset at T ~ 10⁶ K, R ~ 20 pc for n = 1; p_SN varies only ~×2 across realistic media
The number modern galaxy-formation simulations actually inject per supernova: momentum-driven feedback regulates disk star formation, launches winds, and its near-universality (from high-resolution SNR simulations) is why sub-grid feedback converged in the 2010s.
Key referencesCioffi, McKee & Bertschinger (1988); Kim & Ostriker (2015); Walch & Naab (2015).
J-Shocks vs. C-Shocks \[ v_s \lesssim v_{\rm A,ions}\ \Rightarrow\ {\rm continuous\ (C)\ shock} \]
In weakly ionized molecular gas, magnetic signals travel through the ion fluid faster than the shock itself — the field warns the neutrals, which are then heated gently by ion–neutral friction rather than a discontinuous jump. Some shocks arrive with a cushion.
 C-shocks: v_s ≲ 40–50 km/s in molecular gas, T peaks ~10³ K, molecules survive; J-shocks dissociate and reform them
The interpretive fork for every shocked-molecular-gas spectrum: H₂ rotational lines, SiO (sputtered from grains — the standard shock tracer), and H₂O from protostellar outflows all require the C/J distinction to convert line fluxes into shock conditions.
Key referencesMullan (1971); Draine (1980); Draine & McKee (1993); Gusdorf et al. (2008).
Diffusive Shock Acceleration \[ N(E) \propto E^{-2} \quad ({\rm strong\ shock}) \]
Particles scattering back and forth across a shock gain energy on every crossing; the escape probability per cycle turns the ladder into a power law whose slope depends only on the compression ratio. Fermi's mechanism, sharpened: shocks are the Galaxy's particle accelerators.
 N(E) = source spectrum; propagation steepens E⁻² to the observed ~E⁻²·⁷; efficiency ~10% of shock energy
The standard theory of Galactic cosmic-ray origin: SNR shocks accelerating to the "knee" (~3×10¹⁵ eV). Its predictions — X-ray synchrotron rims, amplified fields, γ-rays from π⁰ decay — turned SNR astronomy into cosmic-ray physics.
Key referencesFermi (1949); Bell (1978); Blandford & Ostriker (1978); Ackermann et al. (2013).
Cloud–Shock Interaction \[ t_{\rm cc} = \chi^{1/2}\, \frac{a_{\rm cloud}}{v_s} \]
A blast hitting a dense cloud drives a slower internal shock (slowed by the density contrast χ), then shreds the cloud in a few "cloud-crushing" times via instabilities. Dense obstacles don't stop shocks; they get disassembled by them — unless something (fields, cooling) intervenes.
 χ = ρ_cloud/ρ_ambient; a = cloud radius; magnetic draping and radiative cooling extend survival
The microphysics of multiphase feedback: governs whether SNRs destroy or merely displace clouds, how cold gas survives inside hot galactic winds (it shouldn't, yet it's observed — the "entrainment problem"), and the mixing that seeds new-generation abundances.
Key referencesKlein, McKee & Colella (1994); Scannapieco & Brüggen (2015); Gronke & Oh (2018).
Open unknowns · Shocks & SNRs
The Knee and Beyond
Can SNRs really accelerate to 3×10¹⁵ eV — and what takes over above it?
Standard DSA in SNRs struggles to reach the knee without strong self-amplified fields; no Galactic PeVatron census yet matches the requirement (LHAASO has candidates — including non-SNR ones). The CR spectrum's structure between knee and ankle is unassigned territory.
Electron–Ion Equilibration
Behind a collisionless shock, who gets the heat?
Coulomb collisions are too slow to share energy at SNR densities; plasma processes must do it, and measured T_e/T_i behind Balmer-dominated shocks varies with shock speed in ways no first-principles model fully predicts. The same ignorance propagates into galaxy-cluster shock physics.
Cold Gas in Hot Winds
Why do galactic winds carry molecular gas that shock-shredding should have destroyed?
M82, NGC 253, and high-z outflows all show cold/molecular phases moving at wind speeds. Radiative mixing layers that condense rather than ablate ("cloud growth") are the current answer — a sign convention flip on a 30-year-old problem, still being tested against ALMA data.
X

From Gas to Stars

6 equations

Everything on this sheet converges here: whether, where, and how fast the ISM turns into stars. The Jeans criterion itself lives on the stellar sheet; these are the ISM-side controls — the timescale gravity sets, the support magnetic fields and turbulence provide, and the strange slowness of the outcome.

NameEquationVariablesUse in Research
Free-Fall Time \[ t_{\rm ff} = \sqrt{\frac{3\pi}{32\,G\rho}} \approx \frac{1.4\ {\rm Myr}}{\sqrt{n_{\rm H}/10^3\,{\rm cm^{-3}}}} \]
The collapse time of a pressureless cloud — density's own clock, independent of size or mass. Every star-formation rate, efficiency, and delay is quoted in these units because gravity offers no faster schedule.
 ρ = mass density; n_H = hydrogen-nucleus density; ~4×10⁷ yr at n_H = 1, ~4×10⁴ yr at 10⁶
The normalization of the field's central puzzle: if clouds collapsed in t_ff, the Galaxy would form ~500 M☉/yr of stars. It forms ~2. Everything else in this section is about that factor of ~100 (Zuckerman–Evans argument).
Key referencesJeans (1902); Zuckerman & Evans (1974); Krumholz (2014).
Bonnor–Ebert Mass \[ M_{\rm BE} = 1.18\, \frac{c_s^4}{\sqrt{G^3 P_{\rm ext}}} \]
The heaviest isothermal sphere that external pressure and self-gravity allow to sit in equilibrium — the pressure-bounded refinement of Jeans. Push a core past it (by accretion, cooling, or a pressure squeeze) and there is no equilibrium left to find.
 c_s = isothermal sound speed; P_ext = surface pressure; ≈ 1–2 M☉ for 10 K cores at typical cloud pressures
The stability standard for observed prestellar cores: measured density profiles are routinely fit with Bonnor–Ebert spheres, and the near-coincidence of M_BE with the IMF's characteristic mass (~0.2–1 M☉) is either the origin of the stellar mass scale or a deep red herring.
Key referencesEbert (1955); Bonnor (1956); Alves, Lada & Lada (2001); Könyves et al. (2015).
Mass-to-Flux Ratio \[ \lambda \equiv \frac{(M/\Phi)}{(M/\Phi)_{\rm crit}}, \quad (M/\Phi)_{\rm crit} = \frac{1}{2\pi\sqrt{G}} \]
Frozen-in flux means a magnetized cloud cannot collapse unless its mass exceeds what its field can carry — a threshold no amount of compression changes, since M and Φ compress together. Below λ = 1 gravity cannot win, ever, until the field escapes.
 Φ = magnetic flux; λ < 1 subcritical (supported), λ > 1 supercritical (collapse allowed)
The dividing line of magnetized star-formation theory: the classical picture held clouds subcritical (collapse gated by ambipolar diffusion, next row); Zeeman surveys now find cores supercritical at λ ≈ 2–3 — a paradigm-scale reversal in slow motion.
Key referencesMestel & Spitzer (1956); Mouschovias & Spitzer (1976); Crutcher (2012).
Ambipolar Diffusion \[ t_{\rm AD} \approx 25\, t_{\rm ff} \times \frac{x_e}{10^{-7}}\Big/ \sqrt{\frac{n}{10^4}} \ \ {\rm (scaling)} \]
Only the ions feel the field; the neutrals feel the ions through friction. In weakly ionized gas the neutrals slowly drift through the field lines — gravity leaking matter past the magnetic guard, at a rate set by the ionization fraction (which cosmic rays control, Section VI).
 x_e = ionization fraction; classical t_AD ~ 10 t_ff for quiescent cores; turbulence accelerates it
The classical gatekeeper of star formation and still the dissipation scale of molecular-cloud turbulence; even dethroned as the regulator, ambipolar drift controls protostellar disk formation (the "magnetic braking catastrophe" and its resolution).
Key referencesMestel & Spitzer (1956); Shu, Adams & Lizano (1987); Mouschovias (1991); Zhao et al. (2020).
Star-Formation Efficiency per Free-Fall Time \[ \epsilon_{\rm ff} = \frac{\dot M_*\, t_{\rm ff}}{M_{\rm gas}} \approx 0.01 \]
Per free-fall time, clouds convert about one percent of themselves into stars — a number startlingly constant from local clouds to starburst galaxies. Star formation is not a dam bursting; it is a faucet, dripping at 1% everywhere.
 ε_ff measured from YSO counts (local) or L_IR/M_dense (galaxies); scatter ~0.5 dex, systematics contested
The quantitative statement of the Zuckerman–Evans problem and the target every SF theory must hit: turbulence-regulated, magnetically assisted, and feedback-limited models all produce ~1% for different reasons — degeneracy at the heart of the field.
Key referencesZuckerman & Evans (1974); Krumholz & McKee (2005); Krumholz, Dekel & McKee (2012); Evans et al. (2014); Utomo et al. (2018).
Gas Depletion Time \[ t_{\rm dep} = \frac{M_{\rm gas}}{\dot M_*} \approx 1\!-\!2\ {\rm Gyr\ (molecular)} \]
At current rates, the Galaxy's molecular reservoir lasts ~2 Gyr — far less than its 10-Gyr star-forming history. The ISM is not an endowment but a flow: continuously drained into stars and refilled from the halo, or star formation would have ended long ago.
 t_dep = 1/ε_ff × t_ff at cloud scales; ~2 Gyr for molecular gas across nearby disks (Kennicutt–Schmidt, on the galaxies sheet)
The bridge from ISM physics to galaxy evolution: its near-constancy across nearby spirals underlies the "molecular star-formation law," and its shortness demands ongoing accretion — connecting cloud-scale physics to the circumgalactic medium and cosmological gas supply.
Key referencesKennicutt (1998); Bigiel et al. (2011); Putman, Peek & Joung (2012); Fox et al. (2019).
Open unknowns · From Gas to Stars
Why 1%?
What sets ε_ff ≈ 0.01 — turbulence, magnetism, feedback, or all three conspiring?
Three theory families reproduce the same number from different physics, and observations haven't broken the tie: the scatter and environmental dependence of ε_ff are the discriminating measurements, and they are still systematics-limited. This is the ISM's contribution to the master map's IMF/star-formation gap nodes.
Cores to Stars
Is the stellar IMF just the core mass function shifted by a constant efficiency?
Herschel and ALMA core catalogs resemble the IMF scaled by ~×3 in mass — suggestive of a one-to-one mapping. But cores fragment, accrete, and feed from beyond their boundaries; whether the resemblance is causal or coincidental splits the community.
The Galaxy's Fuel Line
Where is the accretion that must be feeding the disk?
Star formation has burned several times the present gas reservoir over cosmic time. Counted inflow (HVCs, Magellanic Stream, fountain return) falls short of ~1 M☉/yr; ionized, slow, or well-mixed accretion could hide the rest. The circumgalactic medium holds the answer — and most of the baryons.
ISM reference values: Milky Way gas ≈ 7×10⁹ M☉ (H₂ ~1.2, HI ~5, WIM ~1, HIM ~0.1); P/k ≈ 3000–4000 cm⁻³ K; phases: H₂ 10–20 K / CNM ~80 K / WNM ~8000 K / WIM ~8000 K / HIM ~10⁶ K; 21-cm: N_HI = 1.823×10¹⁸ ∫T_B dv; N_H/E(B−V) = 5.8×10²¹ cm⁻² mag⁻¹; R_V = 3.1; A_V = 1.086 τ_V; T_dust ≈ 19–20 K (β ≈ 1.6); X_CO ≈ 2×10²⁰ cm⁻² (K km/s)⁻¹; ζ ≈ 2–5×10⁻¹⁶ s⁻¹ (diffuse); α_B(10⁴ K) = 2.6×10⁻¹³ cm³/s; CNM B ≈ 6 µG; v_A(6 µG, 30 cm⁻³) ≈ 2 km/s; SN: 10⁵¹ erg, p ≈ 3×10⁵ M☉ km/s; Sedov R(10⁴ yr, n=1) ≈ 12 pc; t_ff(10³ cm⁻³) ≈ 1.4 Myr; M_BE(10 K) ≈ 1–2 M☉; ε_ff ≈ 0.01; t_dep(H₂) ≈ 1–2 Gyr; MW SFR ≈ 2 M☉/yr; ~3 SNe/century.