Interstellar Dust Grains

Interstellar dust is the condensed, solid component of the interstellar medium: micron-and-smaller particles of amorphous silicate, carbonaceous matter, and ice mantles, built from the refractory elements — silicon, magnesium, iron, carbon — depleted out of the gas and locked into grains. Depletion studies show iron and silicon missing from the gas at the 90–99% level, so most of those atoms are literally solid. Dust carries only about 1% of the ISM mass (a dust-to-gas ratio $\sim0.01$ ), yet its enormous surface-area-to-mass ratio makes it the dominant source of continuum opacity from the ultraviolet through the infrared. The whole subject turns on one dimensionless number. A grain of radius $a$ meeting light of wavelength $\lambda$ responds according to the size parameter $x=2\pi a/\lambda$ — the grain compared to the wave. When $x\ll1$ the grain is a tiny dipole scattering as $\lambda^{-4}$ ; when $x\sim1$ the interaction is resonant and extinction peaks; when $x\gg1$ the grain is a geometric obstacle and extinction goes grey. This one ratio explains why blue starlight is removed more than red (reddening), why the extinction curve climbs steeply into the UV with its $2175\,\text{Å}$ bump, and why cold grains re-radiate as an inefficient modified blackbody in the far-infrared, reprocessing roughly a third of a galaxy's starlight. The same grains catalyse $\mathrm{H_2}$ , eject photoelectrons that heat the diffuse ISM, and shield molecules deep in clouds — the quiet engine of interstellar chemistry as well as its opacity.

Key takeaways

Interstellar dust grains are solid particles — silicates, carbonaceous material and PAHs, ice mantles — from a few ångström to $\sim0.25\,\mu\mathrm{m}$ , carrying $\sim1\%$ of the ISM mass but dominating its optical opacity.
Grains absorb and scatter starlight (extinction and reddening), re-radiate the absorbed energy in the far-IR as a modified blackbody, and align with the magnetic field to polarise transmitted light.
The mean extinction curve is fit by a power-law size distribution $\mathrm{d}n/\mathrm{d}a\propto a^{-3.5}$ (Mathis–Rumpl–Nordsieck) of silicate and graphite grains, with PAHs added for the mid-IR bands.
Diffuse-ISM grain temperatures are $T_d\sim15$ – $20\,\mathrm{K}$ from balancing absorbed starlight against emission $\propto\nu^{\beta}B_\nu(T_d)$ ( $\beta\approx1.5$ – $2$ ); $T_d$ scales only as $u^{1/(4+\beta)}$ .
Dust drives ISM thermodynamics: photoelectric heating of the diffuse gas, gas–grain coupling in dense cores, and surface catalysis of $\mathrm{H_2}$ the gas phase cannot achieve.
Key idea: A grain's response to light is governed by the size parameter $x=2\pi a/\lambda$ — the Rayleigh regime ( $x\ll1$ , $Q\propto\lambda^{-4}$ ), the resonant regime ( $x\sim1$ , extinction peaks), and the geometric regime ( $x\gg1$ , grey extinction) — so the UV rise of the extinction curve and the far-IR emission are two faces of the same size-versus-wavelength physics.

The core insight

A dust grain's interaction with light is governed by the size parameter $x=2\pi a/\lambda$ (grain radius $a$ compared to wavelength $\lambda$ ): this one ratio sets whether the grain is a Rayleigh scatterer ( $x\ll1$ , efficiency $\propto\lambda^{-4}$ ), a resonant absorber ( $x\sim1$ , extinction peaks), or a grey geometric obstacle ( $x\gg1$ ), so that reddening, the UV extinction rise, and far-infrared emission are all consequences of one number.

Where this lives today

Dust is now a precision tool and, increasingly, a precision nuisance. Three-dimensional dust maps built from the parallaxes and photometry of more than a billion Gaia stars turn $E(B-V)$ along every sightline into a tomographic map of the Local Bubble, the spiral arms, and nearby cloud complexes. At the other extreme, thermal dust modelled as a modified blackbody with $T_d\approx19\,\mathrm{K}$ and $\beta\approx1.5$ – $1.6$ (the Planck values) is the dominant Galactic foreground above $\sim100\,\mathrm{GHz}$ , and its polarised component from aligned grains is the principal contaminant in the hunt for primordial B-mode polarisation — the reason the BICEP/Keck and Planck teams had to disentangle a claimed inflationary signal from dust in 2014–2015. Meanwhile JWST resolves the PAH bands and silicate features in nearby galaxies and protoplanetary disks, and the surprising abundance of dust at redshift $z>6$ is an open problem, demanding rapid grain formation in supernovae or AGB winds faster than standard budgets allow.

The physical picture

Picture a grain as a tiny antenna whose response depends on how its size compares with the wavelength. When the wavelength dwarfs the grain — red and infrared light — the grain barely feels the passing wave and scatters weakly, with a cross section falling as $\lambda^{-4}$ (the Rayleigh physics that blues the daytime sky). When the wavelength matches the grain — blue and UV light — the interaction is resonant and extinction is strong. So along any dusty sightline blue photons are stripped out preferentially and the transmitted star looks redder than it is. That selective removal, not uniform dimming, is the signature of dust; the first figure follows a beam of starlight through a cloud.

Figure

A star seen through a dust cloud: short-wavelength blue photons are scattered out of the line of sight with a cross section that grows as λ⁻⁴, while long-wavelength red photons pass through, so the observer sees a reddened star and a positive colour excess E(B−V) > 0

The absorbed energy has to go somewhere. A grain soaks up UV and optical photons over its geometric cross section and can cool only by radiating. But because it is small compared with far-infrared wavelengths it is an inefficient radiator there — its emissivity is suppressed by $Q\propto\nu^{\beta}$ — so it must warm until the reduced emissivity times the Planck function balances the input, settling near $15$ – $20\,\mathrm{K}$ . The second figure is the object every dust model must reproduce: the mean Galactic extinction versus inverse wavelength, rising into the UV, carrying the $2175\,\text{Å}$ bump, anchored at $R_V\approx3.1$ .

Figure

The mean Galactic extinction curve A_λ/A_V versus inverse wavelength 1/λ in μm⁻¹ for R_V = 3.1: extinction rises steeply toward the ultraviolet, carries the broad 2175 Å (4.6 μm⁻¹) bump from small carbonaceous grains, and climbs again in the far-UV — the near-IR end approaches zero because large grains cannot remove long waves

Two footnotes complete the physics. A grain surface is a two-dimensional catalyst: an adsorbed H atom wanders, meets a partner, and recombines, dumping the excess energy into the lattice — something two free atoms in vacuum can never do, since they cannot conserve both energy and momentum. Nearly all interstellar $\mathrm{H_2}$ is made this way. And the tiniest grains break the equilibrium picture: a PAH of a few hundred atoms has so little heat capacity that one UV photon spikes it to hundreds of kelvin, so it flashes out a few IR photons before the next arrives — stochastic heating that makes the aromatic bands, not a continuum.

The core equations

Extinction begins with a column integral: each grain presents a cross section, and the optical depth is the running sum of those cross sections — geometric area times a dimensionless efficiency that carries all the size-parameter physics.

\tau_\lambda=\int_0^L n_d\,\sigma_{\mathrm{ext}}(\lambda)\,\mathrm{d}s,\qquad \sigma_{\mathrm{ext}}=\pi a^2\,Q_{\mathrm{ext}}(a,\lambda) \qquad (1)

where: $\tau_\lambda$ — optical depth at $\lambda$ ; $n_d$ — grain number density ( $\mathrm{cm^{-3}}$ ); $\sigma_{\mathrm{ext}}$ — extinction cross section per grain; $a$ — grain radius; $Q_{\mathrm{ext}}=Q_{\mathrm{abs}}+Q_{\mathrm{sca}}$ — extinction efficiency; $L$ — path length. Reading it: extinction both absorbs and scatters photons out of the beam; all wavelength dependence lives in $Q_{\mathrm{ext}}$ . Observers measure magnitudes, not $\tau$ ; the two are joined by a fixed factor from the definition of the magnitude scale.

A_\lambda = 1.086\,\tau_\lambda \qquad (2)

where: $A_\lambda$ — extinction in magnitudes at $\lambda$ ; $\tau_\lambda$ — optical depth. Reading it: $\tau=1$ dims a source by $1.086$ mag, the factor being just $2.5\log_{10}e$ (Derivation 1). Because extinction rises toward the blue, the difference between two bands measures the reddening, and the total-to-selective ratio encodes the typical grain size.

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

where: $E(B-V)$ — colour excess between B and V; $A_B,A_V$ — extinction in the B and V bands; $R_V$ — total-to-selective ratio, $\approx3.1$ diffuse and $4$ – $6$ in dense clouds where grains have grown. Reading it: a larger $R_V$ means bigger grains and a flatter, greyer extinction law. The optical response averages over the whole population, which the classic Mathis–Rumpl–Nordsieck model represents as a steep power law in radius, normalised per species.

\frac{\mathrm{d}n}{\mathrm{d}a}=C\,n_{\mathrm H}\,a^{-3.5},\qquad 0.005\,\mu\mathrm{m}\lesssim a\lesssim0.25\,\mu\mathrm{m} \qquad (4)

where: $\mathrm{d}n/\mathrm{d}a$ — grains per volume per radius; $C$ — normalisation (per species); $n_{\mathrm H}$ — hydrogen density; $a$ — grain radius. Reading it: with slope $-3.5$ the small grains dominate number and surface area while the large grains dominate the mass — the balance shaping the extinction curve. A grain's temperature is fixed by radiative balance: absorbed UV/optical power equals emitted far-IR power, the emissivity following $Q_{\mathrm{abs}}\propto\nu^{\beta}$ . The emission integral gives a generalised Stefan–Boltzmann law and a temperature barely sensitive to the field.

\pi a^2\langle Q_{\mathrm{abs}}\rangle_{\mathrm{UV}}\,c\,u_{\mathrm{rad}}=4\pi a^2\!\int\!\pi Q_0\Big(\tfrac{\nu}{\nu_0}\Big)^{\!\beta}B_\nu(T_d)\,\mathrm{d}\nu\ \ \Longrightarrow\ \ T_d\propto u_{\mathrm{rad}}^{1/(4+\beta)} \qquad (5)

where: $Q_{\mathrm{abs}}$ — absorption efficiency (equals emissivity, by Kirchhoff); $u_{\mathrm{rad}}$ — radiation-field energy density; $B_\nu(T_d)$ — Planck function; $\beta\approx1.5$ – $2$ — emissivity index; $T_d$ — grain temperature. Reading it: with $\beta=2$ the exponent is $1/6$ , so a factor-10 stronger field raises $T_d$ by only $\sim47\%$ . The re-emitted light is a modified blackbody: the Planck function weighted by a frequency-dependent optical depth, thin in the far-IR so the emergent intensity is simply $\tau_\nu B_\nu$ .

I_\nu=B_\nu(T_d)\,(1-e^{-\tau_\nu})\ \approx\ \tau_\nu\,B_\nu(T_d),\qquad \tau_\nu\propto\nu^{\beta} \qquad (6)

where: $I_\nu$ — observed specific intensity; $B_\nu(T_d)$ — Planck function; $\tau_\nu$ — dust optical depth (thin in the far-IR); $\beta$ — emissivity index. Reading it: fitting this two-parameter form ( $T_d,\beta$ ) is how Planck and Herschel map dust temperatures. Finally, dust runs the gas thermodynamics and chemistry: far-UV photons eject photoelectrons that heat the diffuse medium, and grain surfaces catalyse $\mathrm{H_2}$ formation at a rate set by how fast H atoms strike, stick, and recombine.

\Gamma_{\mathrm{pe}}=10^{-24}\,\epsilon\,G_0\,n_{\mathrm H}\ \ \mathrm{erg\,cm^{-3}\,s^{-1}},\qquad R_{\mathrm{H_2}}=\tfrac12\,n_{\mathrm H}\,v_{\mathrm{th}}\,n_d\sigma_d\,S\,\eta \qquad (7)

where: $\Gamma_{\mathrm{pe}}$ — photoelectric heating rate; $\epsilon$ — efficiency (a few percent); $G_0$ — far-UV field in Habing units; $R_{\mathrm{H_2}}$ — $\mathrm{H_2}$ formation rate per volume; $v_{\mathrm{th}}$ — thermal speed of H; $n_d\sigma_d$ — grain cross section per volume; $S$ — sticking coefficient; $\eta$ — recombination efficiency. Reading it: both scale with grain surface area, so the small grains that dominate the surface dominate both.

Derivation AWhy optical depth becomes magnitudes and why extinction reddens

The goal is to connect the optical depth $\tau_\lambda$ to the observer's extinction $A_\lambda$ in magnitudes, pin the factor $1.086$ to the definition of the magnitude scale, and show that because $\tau$ rises toward the blue the difference of two bands is always positive — the colour excess $E(B-V)$ that makes a star look redder than it is. We start from the attenuation law, pass through the magnitude definition, and take a colour difference.

Step 1. Along a dusty path the flux is attenuated exponentially, $F=F_0\,e^{-\tau_\lambda}$ , with $F_0$ the unextinguished flux. Why: each column element removes a fixed fraction of photons, so the survival probability compounds into an exponential — Beer–Lambert bookkeeping.

Step 2. Extinction is defined as the added magnitudes, $A_\lambda\equiv m-m_0=-2.5\log_{10}(F/F_0)$ . Substitute the attenuation law: $A_\lambda=-2.5\log_{10}(e^{-\tau_\lambda})=2.5\,\tau_\lambda\log_{10}e$ . Why: the magnitude scale is logarithmic by construction, so an exponential dimming turns into a linear one in magnitudes.

Step 3. Evaluate the constant: $\log_{10}e=0.4343$ , so $A_\lambda=2.5\times0.4343\,\tau_\lambda=1.086\,\tau_\lambda$ . An optical depth of one is $1.086$ mag of dimming. Why: the factor is pure arithmetic from the base change between physics' natural log and astronomy's base-10 magnitude — no dust physics enters.

Step 4. Form a colour. Writing $m_\lambda=m_{\lambda,0}+A_\lambda$ in two bands and subtracting, $(B-V)_{\mathrm{obs}}=(B-V)_0+(A_B-A_V)$ . Since $Q_{\mathrm{ext}}$ rises toward shorter $\lambda$ , $\tau_B>\tau_V$ , hence $A_B>A_V$ and $E(B-V)>0$ . Why: the wavelength dependence of $Q_{\mathrm{ext}}$ breaks the colour symmetry — a grey opacity dims without reddening, a rising one always reddens.

Step 5. Divide total by selective extinction: $R_V=A_V/E(B-V)\approx3.1$ for diffuse dust. Why: $R_V$ is a shape parameter for how steeply extinction rises, hence the characteristic grain size — growing to $4$ – $6$ where grains coagulate in dense clouds. *Note: $2.5\log_{10}e=1.0857$ gives the factor $1.086$ in $A_\lambda=1.086\,\tau_\lambda$ ; the standard diffuse-ISM value $R_V=3.1$ implies $A_V\approx3.1\,E(B-V)$ .*

Go deeper · Derivation B · 1 minThe equilibrium temperature of a dust grain

Derivation BThe equilibrium temperature of a dust grain

The goal is to derive how a grain's temperature is set and explain why grains across the diffuse Galaxy cluster in a narrow $15$ – $20\,\mathrm{K}$ band despite huge variations in the local starlight. The key is that absorption and emission occur in different parts of the spectrum, and the far-IR emissivity's power-law rise turns emitted power into a very high power of temperature — so $T_d$ barely has to move to rebalance the books.

Step 1. Write the absorbed power. Starlight peaks in the UV/optical where $Q_{\mathrm{abs}}\sim1$ , so $P_{\mathrm{abs}}=\pi a^2\langle Q_{\mathrm{abs}}\rangle_{\mathrm{UV}}\,c\,u_{\mathrm{rad}}$ , set by the geometric cross section and the radiation energy density. Why: a small grain intercepts starlight over its projected area and absorbs the UV efficiently, so the intercepted power is essentially geometric.

Step 2. Write the emitted power with a power-law emissivity. By Kirchhoff's law emissivity equals $Q_{\mathrm{abs}}$ , and in the far-IR $Q_{\mathrm{abs}}(\nu)=Q_0(\nu/\nu_0)^{\beta}$ , so $P_{\mathrm{emit}}=4\pi a^2\!\int_0^\infty\!\pi Q_0(\nu/\nu_0)^{\beta}B_\nu(T_d)\,\mathrm{d}\nu$ . Why: the grain radiates over $4\pi a^2$ but is a poor long-wavelength emitter — that inefficiency is the $\nu^{\beta}$ suppression.

Step 3. Evaluate the integral. Substituting $x=h\nu/kT_d$ turns $\int\nu^{\beta}B_\nu\,\mathrm{d}\nu$ into a constant times $T_d^{\,4+\beta}$ — a generalised Stefan–Boltzmann law. Why: the usual $T^4$ gains an extra $T^{\beta}$ because each power of $\nu$ in the emissivity pulls another power of $T$ from the Planck integral.

Step 4. Balance and solve. Setting $P_{\mathrm{abs}}=P_{\mathrm{emit}}$ cancels $\pi a^2$ and gives $c\langle Q_{\mathrm{abs}}\rangle_{\mathrm{UV}}u_{\mathrm{rad}}\propto T_d^{\,4+\beta}$ , hence $T_d\propto u_{\mathrm{rad}}^{1/(4+\beta)}$ . Why: the geometric size drops out, so grains of very different radii reach the same temperature — it depends on the field, not the grain.

Step 5. Read the exponent. With $\beta\approx2$ , $T_d\propto u_{\mathrm{rad}}^{1/6}$ : a factor-10 change in field shifts $T_d$ by only $10^{1/6}\approx1.47$ . Why: a one-sixth power is extraordinarily weak — which is why Galactic cirrus everywhere sits near $\sim18\,\mathrm{K}$ . *Note: $10^{1/6}=1.468$ , confirming that a tenfold increase in radiation field raises $T_d$ by only $\sim47\%$ — consistent with the narrow far-IR colour temperature of Galactic cirrus.*

Go deeper · Derivation C · 1 minThe Rayleigh scattering limit for grains much smaller than the wavelength

Derivation CThe Rayleigh scattering limit for grains much smaller than the wavelength

The goal is to derive the $\lambda^{-4}$ scattering law for grains far smaller than the wavelength — the physics that reddens starlight and blues reflection nebulae — from nothing more than a polarisable sphere driven by an oscillating field. We build the induced dipole, radiate it with the Larmor formula, divide by the incident flux for a cross section, and normalise by the geometric area for the efficiency.

Step 1. Induced dipole. A dielectric sphere of complex index $m$ in a uniform field acquires $p=\alpha E_0$ with polarisability $\alpha=a^3\,(m^2-1)/(m^2+2)$ , the Clausius–Mossotti form. Why: for a grain much smaller than the wave the field across it is uniform at each instant, so electrostatics fixes the dipole and the wave enters only through time.

Step 2. Radiated power. An oscillating dipole radiates $P_{\mathrm{sca}}=\omega^4|p|^2/(12\pi\epsilon_0c^3)$ by the Larmor formula. Why: this $\omega^4$ is the whole origin of the wavelength dependence — a harmonic dipole's acceleration carries two powers of $\omega$ into the amplitude, four into the power.

Step 3. Cross section. Divide radiated power by the incident flux $c\epsilon_0E_0^2/2$ : $\sigma_{\mathrm{sca}}=\tfrac{8\pi}{3}k^4a^6\,|(m^2-1)/(m^2+2)|^2$ with $k=2\pi/\lambda$ . Why: a cross section is power removed per unit incident flux — an effective collecting area.

Step 4. Efficiency factor. Divide by $\pi a^2$ : $Q_{\mathrm{sca}}=\tfrac{8}{3}(ka)^4\,|(m^2-1)/(m^2+2)|^2=\tfrac{8}{3}x^4\,|(m^2-1)/(m^2+2)|^2\propto\lambda^{-4}$ . Why: through $x=ka=2\pi a/\lambda$ the efficiency depends only on size parameter and material — small grains scatter blue far more than red, reddening transmitted light and bluing nebulae.

Limiting cases

The extinction efficiency $Q_{\mathrm{ext}}(x)$ falls into three regimes in the size parameter $x=2\pi a/\lambda$ , and the whole extinction curve is a walk through them as $\lambda$ sweeps the grain population.

Case	Result	Meaning
Rayleigh ( $x\ll1$ , IR/radio)	$Q_{\mathrm{abs}}\propto x\propto\lambda^{-1}$ , $\ Q_{\mathrm{sca}}\propto x^4\propto\lambda^{-4}$	Grain far smaller than the wave: scattering negligible, weak absorption dominates. The far-IR is nearly transparent and the sub-mm emissivity scales as $\nu^{\beta}$ .
Resonant ( $x\sim1$ , optical/UV)	$Q_{\mathrm{ext}}$ peaks, $\ Q_{\mathrm{ext}}\gtrsim1$	Grain comparable to the wave: strong, structured extinction that shapes the optical curve and its UV rise for the sub-micron grains carrying the mass.
Geometric ( $x\gg1$ , X-ray/large grains)	$Q_{\mathrm{ext}}\to2$	Grain much larger than the wave: extinction approaches twice the geometric area (the extinction paradox), goes grey, and produces no reddening.

Common trapExtinction is absorption, and dust shares the gas temperature

Two conflations to avoid. First, extinction is not absorption: it is absorption plus scattering, and scattered photons are removed from the sightline but not destroyed (they reappear as reflection nebulosity), so a point source is dimmed more than the grains are heated — only the absorbed fraction becomes grain heat and far-IR emission. Second, dust and gas are not at a common temperature: diffuse-ISM grains sit at $T_d\sim15$ – $20\,\mathrm{K}$ while the gas can be far hotter, and only in dense cores ( $n\gtrsim10^{4.5}\,\mathrm{cm^{-3}}$ ) do collisions couple them. And a grain radiates a clean modified blackbody only if it holds a steady temperature — the smallest grains and PAHs are stochastically heated and never reach equilibrium.

Worked examples

1 · Reading the size parameter (confidence builder)

Take a typical sub-micron grain of radius $a=0.1\,\mu\mathrm{m}$ . Compute its size parameter $x=2\pi a/\lambda$ in the V band (550 nm), in the far-UV (150 nm), and in the far-IR (100 μm), and classify each regime.

Step 1. V band: $x=2\pi(0.1)/0.55=\mathbf{1.14}$ . The grain is comparable to the wavelength — the resonant regime, where extinction is strong and structured. This is why $0.1\,\mu\mathrm{m}$ grains dominate optical extinction.

Step 2. Far-UV: $x=2\pi(0.1)/0.15=\mathbf{4.19}$ . Several wavelengths across, moving toward the geometric regime where $Q_{\mathrm{ext}}\to2$ — this is the UV rise in the second figure.

Step 3. Far-IR: $x=2\pi(0.1)/100=\mathbf{6.3\times10^{-3}}$ . Deep in the Rayleigh regime, scattering $\propto\lambda^{-4}$ is negligible and only weak $\nu^{\beta}$ absorption survives, so the grain is nearly invisible — which is why the far-IR probes the whole dust column. One 0.1 μm grain is resonant in the optical, geometric in the UV, and Rayleigh in the far-IR.

2 · From reddening to gas column (fully worked)

A star behind a diffuse cloud has a measured colour excess $E(B-V)=0.30\,\mathrm{mag}$ . Assuming standard diffuse dust ( $R_V=3.1$ ), find the visual extinction $A_V$ , the optical depth $\tau_V$ , and the total hydrogen column $N_{\mathrm H}$ using $N_{\mathrm H}/E(B-V)=5.8\times10^{21}\,\mathrm{cm^{-2}\,mag^{-1}}$ .

Step 1. Visual extinction from $R_V$ : $A_V=R_V\,E(B-V)=3.1\times0.30=\mathbf{0.93\ mag}$ . The star is dimmed by nearly a magnitude in V — a factor $10^{0.93/2.5}\approx2.4$ in flux.

Step 2. Optical depth from equation (2): $\tau_V=A_V/1.086=0.93/1.086=\mathbf{0.86}$ . The sightline is marginally optically thin at V.

Step 3. Hydrogen column from the Bohlin–Savage–Drake relation: $N_{\mathrm H}=5.8\times10^{21}\times0.30=\mathbf{1.7\times10^{21}\ cm^{-2}}$ . Dust extinction has become a gas-mass tracer. E(B−V) = 0.30 gives A_V = 0.93 mag, τ_V = 0.86, and N_H = 1.7×10²¹ cm⁻².

3 · How stiff is the grain thermostat?

Diffuse-ISM grains sit at $T_d\approx18\,\mathrm{K}$ . A grain drifts into a region near an OB association where the radiation field is $100\times$ stronger. Estimate its new temperature, taking $\beta=2$ .

Step 1. Use the scaling from equation (5): $T_d\propto u_{\mathrm{rad}}^{1/(4+\beta)}=u_{\mathrm{rad}}^{1/6}$ for $\beta=2$ . The one-sixth power is the whole story.

Step 2. Apply the factor: $T_d'=18\times100^{1/6}=18\times2.15=\mathbf{39\ K}$ . A hundredfold jump in starlight only doubles the temperature.

Step 3. Interpret. A $100\times$ field raises $T_d$ by barely a factor $2.2$ — which is why the diffuse Galaxy glows at a nearly uniform far-IR colour temperature and why dust makes such a clean CMB foreground template. A 100× field takes an 18 K grain only to ~39 K — the grain thermostat is extremely stiff.

Physics of the portrait

Interstellar dust was first inferred not from what it emitted but from what it stole: distant stars too faint and too red for their distances. The portrait belongs to the astronomer who turned that suspicion into proof.

Origins · 1 minRobert Trumpler and how this was discovered

Robert Trumpler

1886–1956 · proved interstellar absorption in 1930 · Galactic structure and the interstellar medium

For most of the nineteenth century the space between the stars was assumed transparent, and Herschel's dark lanes were read as genuine holes empty of stars. Robert Julius Trumpler, a Swiss-born astronomer at Lick Observatory, overturned this in 1930 with a beautifully simple argument. Studying open clusters, he estimated each cluster's distance two independent ways: from the apparent brightness of its stars, and from its angular diameter assuming clusters share a common true size. The two disagreed systematically — the brightness distances were always too large, and the gap grew with distance. The only consistent explanation was that starlight is progressively dimmed by an absorbing medium in the Galactic plane, about $0.7$ mag per kiloparsec. Crucially, Trumpler also found the distant stars reddened, telling him the absorber was not gas but fine solid particles — dust — scattering blue more than red. In one stroke he established the existence, amount, and reddening signature of interstellar dust, and corrected the entire Galactic distance scale, which had over-estimated sizes by ignoring the dimming. The grain models of van de Hulst, the MRN size distribution, and the PAHs of Weingartner and Draine all rest on the extinction curve whose reality Trumpler first proved.

History

1785

— Herschel records dark "holes in the heavens" among the star fields of the Milky Way

1930

— Trumpler proves interstellar absorption and reddening from the distances of open clusters

1949

— Hall and Hiltner discover the polarisation of starlight by aligned dust grains

1977

— Mathis, Rumpl and Nordsieck fit the extinction curve with an a⁻³·⁵ grain-size distribution

2001

— Weingartner and Draine add PAHs to reproduce the full UV-to-IR extinction and emission

Going further

Texts. Carroll & Ostlie, An Introduction to Modern Astrophysics, 2nd ed., ch. 12 · Draine, Physics of the Interstellar and Intergalactic Medium (2011), chs. 21–24 · Spitzer, Physical Processes in the Interstellar Medium (1978), ch. 7 · Bohren & Huffman, Absorption and Scattering of Light by Small Particles (1983) · Binney & Tremaine, Galactic Dynamics, 2nd ed., §8. Key numerical checks: $A_\lambda=1.086\,\tau_\lambda$ follows from $2.5\log_{10}e=1.0857$ ; the Rayleigh efficiency is $Q_{\mathrm{sca}}=\tfrac{8}{3}x^4|(m^2-1)/(m^2+2)|^2$ via the Larmor route; the grain-temperature scaling $T_d\propto u_{\mathrm{rad}}^{1/(4+\beta)}$ comes from $\int\nu^{\beta}B_\nu\,\mathrm{d}\nu\propto T^{4+\beta}$ , with $10^{1/6}=1.468$ and $100^{1/6}=2.154$ ; the worked sightline gives $A_V=0.93$ , $\tau_V=0.86$ , $N_{\mathrm H}=1.7\times10^{21}\,\mathrm{cm^{-2}}$ . Standard reference values: MRN slope $-3.5$ ; $R_V\approx3.1$ and the $2175\,\text{Å}$ bump; $N_{\mathrm H}/E(B-V)\approx5.8\times10^{21}\,\mathrm{cm^{-2}\,mag^{-1}}$ ; diffuse $T_d\approx15$ – $20\,\mathrm{K}$ , $\beta\approx1.5$ – $2$ .

Prerequisites. Phases of the Interstellar Medium · Blackbody Radiation and the Planck Function · Electromagnetic Scattering and Mie Theory. Leads to. Interstellar Extinction and Reddening · Thermal Dust Emission and Infrared Radiation · Interstellar Polarisation and Magnetic Fields · Interstellar Gas Heating and Cooling · Molecular Clouds.

Ask the tutor: Ask me to walk through any step — from why $2.5\log_{10}e=1.086$ , to how the $\nu^{\beta}$ emissivity makes the grain thermostat so stiff, to why the $2175\,\text{Å}$ bump fingerprints carbonaceous grains.

Is a dust grain made of one material, or a mixture?

The mean extinction curve is fit by two separate populations: amorphous silicates (magnesium-iron pyroxenes and olivines, which make the $9.7\,\mu\mathrm{m}$ Si–O feature) and carbonaceous grains (graphite, amorphous carbon, and PAHs), with the $2175\,\text{Å}$ bump assigned to the small carbonaceous grains. Real grains are probably neither pure — many are aggregates with silicate cores, carbon coatings, and, in cold dense clouds, icy mantles of water, CO and $\mathrm{CO_2}$ . The two-population MRN picture is a modelling convenience that reproduces the observations remarkably well.

Why is the size distribution a power law with slope −3.5?

The MRN slope $-3.5$ was found empirically by fitting the extinction curve, but it is close to the steady state of a collisional cascade — grains repeatedly shattering in low-velocity collisions, the same physics that governs asteroid and debris-disk size spectra. With that slope the small grains dominate surface area and total extinction cross section while the large grains hold most of the mass. Modern models (Weingartner & Draine 2001) keep the power-law shape but add a log-normal population of very small grains and PAHs to match the mid-IR emission.

If dust is only 1% of the mass, why does it matter so much?

Because opacity scales with surface area, not mass, and grinding the same mass into micron particles multiplies its surface enormously — a gram in $0.1\,\mu\mathrm{m}$ grains presents roughly a million times the area of a gram in a single pebble. That is why a dust-to-gas ratio of $0.01$ still makes dust the dominant continuum opacity from UV to IR, the dominant far-IR emitter, the main heating agent of the diffuse gas, and the catalytic surface for essentially all interstellar $\mathrm{H_2}$ . Mass is the wrong currency for dust; area is the right one.

How does a grain absorb UV yet emit far-IR without breaking Kirchhoff's law?

Kirchhoff's law says emissivity equals absorptivity at each frequency, $\varepsilon_\nu=Q_{\mathrm{abs}}(\nu)$ — not that a body must emit where it absorbs. The grain absorbs where the incident radiation is strong (the UV/optical, where starlight peaks and $Q_{\mathrm{abs}}\sim1$ ), but emits according to $Q_{\mathrm{abs}}(\nu)B_\nu(T_d)$ , and because it is cold ( $\sim18\,\mathrm{K}$ ) the Planck factor peaks in the far-IR. Same $Q_{\mathrm{abs}}$ , two weighting spectra: absorption weighted by starlight, emission by the grain's own thermal curve. No contradiction.

What is the 2175 Å bump and why do we care?

It is a broad, strong absorption feature centred at $2175\,\text{Å}$ ( $4.6\,\mu\mathrm{m^{-1}}$ ) riding on the rising UV extinction curve — the most prominent spectral feature in Galactic extinction. Its central wavelength is remarkably constant across sightlines while its strength varies, pointing to a specific carrier: small carbonaceous grains (PAHs or graphitic material) undergoing a $\pi\!\to\!\pi^*$ electronic resonance. It matters as a fingerprint of grain composition — its shift or absence in other galaxies diagnoses different dust chemistry — and it must be modelled to de-redden UV spectra of stars and quasars.

Does interstellar dust relate to the dust that builds planets?

It is the same material lineage, coarsened. Interstellar grains are the seed solids a molecular cloud carries into a collapsing core and then into the protoplanetary disk, where they coagulate — sticking collisions grow them from sub-micron ISM grains through millimetre pebbles toward kilometre planetesimals, the first rung of planet formation. The composition is inherited: ISM silicates and carbon become the refractory backbone of rocky planets, and the icy mantles become the volatiles of comets. The $R_V$ growth seen in dense clouds — grains getting bigger — is the opening move of that process.