Thermal Dust Emission and Infrared Radiation

Interstellar dust is barely one percent of the mass of the interstellar medium, yet it intercepts and reprocesses a third to a half of all the starlight a galaxy makes. The mechanism is disarmingly simple. A grain in the ambient radiation field absorbs photons — most greedily in the ultraviolet and optical, where its cross-section is largest — and turns that energy into heat. In steady state it radiates exactly the power it soaks up, but because a grain is cold, only $T_d\sim 15$ – $40\,\mathrm{K}$ , Wien's law drops that re-emission deep into the far-infrared, peaking near $100\,\mu\mathrm{m}$ . That glow is thermal dust emission, the single most direct handle astronomers have on the cold, star-forming medium. The spectrum is not a clean blackbody. A grain is a poor antenna for wavelengths larger than itself, so its emission efficiency $Q_\nu$ — equivalently the mass opacity $\kappa_\nu$ — collapses toward long wavelengths as a power law $\kappa_\nu\propto\nu^\beta$ . The observed intensity is that opacity times the Planck function at the grain temperature: a modified blackbody, or greybody, whose emissivity index $\beta\approx1.5$ – $2$ carries the fingerprint of grain composition and is notoriously degenerate with the fitted temperature. This one process spans the discipline. In the far-infrared the medium is transparent to its own emission, so the brightness reads off dust column — and, through a gas-to-dust ratio, total gas column — where CO has frozen onto grains. Integrated over a galaxy it measures obscured star formation. And on the largest scales the diffuse Galactic greybody is the dominant polarized foreground that must be scrubbed from the microwave sky before a primordial B-mode signal from inflation can be believed. A heated speck of soot reaches from a prestellar core to the edge of cosmology.

Key takeaways

Dust grains absorb ultraviolet and optical starlight, equilibrate at $T_d\sim10$ – $50\,\mathrm{K}$ , and re-radiate the absorbed power as far-infrared and submillimeter continuum.
The emergent spectrum is a modified blackbody: $I_\nu\propto\kappa_\nu B_\nu(T_d)$ with opacity rising as $\kappa_\nu\propto\nu^\beta$ , emissivity index $\beta\approx1.5$ – $2$ .
Radiative balance makes the grain a stiff thermostat: absorbed power $\propto G$ , emitted power $\propto T_d^{4+\beta}$ , so $T_d\propto G^{1/(4+\beta)}\approx G^{1/6}$ — nearly isothermal across enormous ranges of environment.
In the optically thin Rayleigh–Jeans regime the flux is $S_\nu\propto\kappa_0\,\nu^{2+\beta}\,T_d\,M_d\,D^{-2}$ , making submillimeter photometry a near-linear scale for dust — and hence total — mass.
The same greybody, emitted by the diffuse Milky Way at $T_d\approx19$ – $20\,\mathrm{K}$ , is the leading polarized foreground for the cosmic microwave background above $\sim100\,\mathrm{GHz}$ .
Very small grains and PAHs are stochastically heated by single photons, spike to hundreds of kelvin, and break the single-temperature picture with mid-infrared bands.
Key idea: a dust grain is a thermostat — it absorbs starlight and re-emits an equal power as a modified blackbody $I_\nu\propto\kappa_\nu B_\nu(T_d)$ , so its far-infrared spectrum fixes the grain temperature from the peak and, in the optically thin limit, the dust column and mass from the amplitude.

The core insight

A dust grain is a thermostat: it absorbs starlight and re-emits an equal power as a modified blackbody $I_\nu\propto\kappa_\nu B_\nu(T_d)$ , so its far-infrared spectrum simultaneously fixes the grain temperature (from the peak) and, in the optically thin limit, the dust column and mass (from the amplitude).

Where this lives today

Thermal dust emission is an active frontier at both ends of astrophysics. On the small end, ALMA and JCMT/SCUBA-2 resolve prestellar cores at $850\,\mu\mathrm{m}$ and $1.3\,\mathrm{mm}$ , converting continuum into mass and column maps precisely where CO freezes out, and those masses feed the Jeans and virial analyses that decide which cores collapse. Herschel photometry from $70$ to $500\,\mu\mathrm{m}$ produced the coupled temperature-and-column maps that revealed the filamentary skeleton of molecular clouds, and JWST now dissects the stochastically heated PAH bands the equilibrium greybody cannot describe. The sharpest open problem is cosmological: the diffuse Galactic greybody ( $T_d\approx19$ – $20\,\mathrm{K}$ , $\beta\approx1.5$ ) is polarized because aligned grains radiate anisotropically, and it dominates the microwave sky above $\sim100\,\mathrm{GHz}$ . After the 2014 BICEP2 episode — a claimed primordial B-mode that proved consistent with unmodeled dust — the next generation (Simons Observatory, CMB-S4, LiteBIRD) is built around measuring this greybody in many channels precisely enough to subtract it. Whether inflation left a gravitational-wave imprint on the CMB may be decided by how well we model a heated speck of interstellar soot.

The physical picture

Leave a black pebble in sunlight and it warms until it radiates exactly the power it absorbs; the temperature it settles at is fixed by how bright the light is. A dust grain does the same, but its "sunlight" is dilute starlight, so it equilibrates at a chilly $\sim20\,\mathrm{K}$ and its glow is invisible far-infrared heat rather than a red shine. Two facts complicate the analogy, both load-bearing. First, the grain is a poor emitter at wavelengths larger than itself, so it radiates inefficiently in the far-infrared even though it absorbed efficiently in the ultraviolet — that mismatch, $Q_\nu\propto\nu^\beta$ , is what "modifies" the blackbody. Second, because the far-infrared opacity is tiny, a whole cloud is transparent to its own emission, so a detector counts grains directly and the brightness measures column. The first figure fixes the energy budget and the thermostat law it implies.

Figure

A dust grain in the interstellar radiation field: it absorbs ultraviolet and optical starlight with efficiency near unity and re-radiates the same power inefficiently in the far-infrared, Qν ∝ νᵝ. Balancing P_abs ∝ G against P_emit ∝ Td⁴⁺ᵝ gives the stiff thermostat Td ∝ G^(1/(4+β)) ≈ G^(1/6), pinning diffuse-ISM dust near 18–20 K

Because the opacity climbs with frequency, the emitted spectrum is tilted bluewards of a Planck curve at the same temperature — the factor $\kappa_\nu\propto\nu^\beta$ pushes the peak to shorter wavelength and steepens the submillimeter slope to $\nu^{2+\beta}$ . That distortion is why we call it a modified blackbody, and it is the origin of the temperature–index degeneracy that plagues fitting. The second figure is the payoff — a greybody and a blackbody at the identical $20\,\mathrm{K}$ , so the peak shift reads straight off the axis.

Figure

Modified blackbody versus a true Planck curve at the same Td = 20 K. The opacity factor κν ∝ ν² multiplies Bν and shifts the peak from λ ≈ 255 μm (pure blackbody) to λ ≈ 145 μm (greybody, β = 2) — a factor 1.76 bluer. Fitting the greybody as a pure blackbody therefore overestimates the temperature by the same 1.76

Finally, the picture has a hard limit. Shrink a grain until a single ultraviolet photon carries more energy than its heat capacity can hold at $20\,\mathrm{K}$ , and the thermostat breaks: the grain spikes to hundreds of kelvin, flashes a burst of mid-infrared, and cools before the next photon arrives. Such stochastically heated very small grains and PAHs never hold a steady temperature; they sample a broad distribution and paint the mid-infrared with emission bands rather than a greybody. Equilibrium dust and stochastic dust are two different populations sharing one sky.

The core equations

The starting point is the cross-section of a single grain: geometric area $\pi a^2$ reduced by a dimensionless efficiency near unity in the ultraviolet but collapsing in the far-infrared.

C_\mathrm{abs}(\nu) = \pi a^2\, Q_\mathrm{abs}(\nu, a) \qquad (1)

where: $C_\mathrm{abs}$ — absorption cross-section (area); $a$ — grain radius; $Q_\mathrm{abs}$ — dimensionless efficiency, $\lesssim1$ in the UV and $\propto\nu^\beta$ in the far-IR; $\nu$ — frequency. Reading: the grain is a leaky absorber, and by Kirchhoff's law the same $Q_\mathrm{abs}$ governs its emission. Steady state demands that the power absorbed equals the power re-emitted. Integrating the cross-section against the incident field on the left and the grain's own Planck emission on the right fixes $T_d$ .

\int_0^\infty C_\mathrm{abs}(\nu)\,\frac{c\,u_\nu}{4\pi}\,\mathrm{d}\nu \;=\; \int_0^\infty C_\mathrm{abs}(\nu)\,\pi B_\nu(T_d)\,\mathrm{d}\nu \qquad (2)

where: $u_\nu$ — spectral energy density of the ambient field; $c$ — speed of light; $B_\nu(T_d)$ — Planck function at the grain temperature; $T_d$ — equilibrium dust temperature. Reading: absorbed equals emitted, one equation with one unknown, so it defines $T_d$ . The emission is coarse-grained into an opacity per unit dust mass, whose frequency dependence is the power law carrying the grain physics.

\kappa_\nu = \kappa_0\left(\frac{\nu}{\nu_0}\right)^{\beta} \qquad (3)

where: $\kappa_\nu$ — opacity per unit dust mass ( $\mathrm{cm^2\,g^{-1}}$ ); $\kappa_0$ — reference opacity at $\nu_0$ (order $0.1\,\mathrm{cm^2\,g^{-1}}$ at $250\,\mu\mathrm{m}$ , model dependent); $\beta$ — emissivity index, $\approx1.5$ – $2$ ; $\nu_0$ — reference frequency. Reading: $\beta$ is the opacity slope, and $\kappa_0$ is the largest single systematic in any dust-mass estimate. Radiative transfer through a slab of this dust gives the emergent intensity; the greybody used in fitting is its optically thin limit.

I_\nu = B_\nu(T_d)\left(1 - e^{-\tau_\nu}\right),\qquad \tau_\nu = \kappa_\nu\,\Sigma_d \qquad (4)

where: $I_\nu$ — specific intensity; $\tau_\nu$ — dust optical depth; $\Sigma_d$ — dust mass column density. Reading: when $\tau_\nu\ll1$ the bracket becomes $\tau_\nu$ and $I_\nu\to\kappa_\nu\Sigma_d B_\nu(T_d)$ ; when $\tau_\nu\gg1$ it saturates to $B_\nu(T_d)$ . The Planck function supplies the temperature dependence — steep on its Wien tail, linear on its Rayleigh–Jeans tail.

B_\nu(T) = \frac{2h\nu^3}{c^2}\,\frac{1}{e^{h\nu/k_B T}-1} \qquad (5)

where: $h$ — Planck constant; $k_B$ — Boltzmann constant; $T$ — temperature. Reading: for $h\nu\ll k_BT$ it flattens to $2\nu^2 k_BT/c^2$ (linear in $T$ ); for $h\nu\gg k_BT$ it plunges as $e^{-h\nu/k_BT}$ . Inverting the optically thin intensity over the source turns a flux into a dust mass — the workhorse of submillimeter astronomy.

M_d = \frac{S_\nu\,D^2}{\kappa_\nu\,B_\nu(T_d)} \qquad (6)

where: $M_d$ — total dust mass; $S_\nu$ — observed flux density; $D$ — distance to the source. Reading: mass is linear in flux, so once $\kappa_\nu$ and $T_d$ are fixed, submillimeter photometry weighs cold gas — total gas mass following from gas-to-dust $\sim100$ . Integrating over all frequencies recovers a Stefan–Boltzmann-like law, with the opacity exponent added to the power.

F_\mathrm{IR} = \int_0^\infty \kappa_\nu\,\Sigma_d\,\pi B_\nu(T_d)\,\mathrm{d}\nu \;\propto\; \Sigma_d\,T_d^{\,4+\beta} \qquad (7)

where: $F_\mathrm{IR}$ — bolometric far-infrared surface brightness. Reading: the emitted power scales as $T_d^{4+\beta}$ rather than $T_d^{4}$ — the exponent that makes the grain thermostat so stiff in the first derivation below.

Derivation AWhy interstellar dust is nearly isothermal — the grain thermostat

The goal of this derivation is to show that a grain's equilibrium temperature depends only very weakly on the strength of the surrounding radiation field, obeying $T_d\propto G^{1/(4+\beta)}$ , and to make explicit where the exponent $4+\beta$ comes from. This is the physical reason diffuse-ISM dust everywhere clusters near $18$ – $20\,\mathrm{K}$ even though the interstellar field varies by orders of magnitude between a cirrus cloud and an H II region. We balance absorbed against emitted power, evaluate the emission integral, and cancel every grain-specific factor so only the scaling survives.

Step 1. Write the absorbed power for a grain in a field scaled to $G$ times the local value, $u_\nu = G\,u_\nu^\mathrm{ISRF}$ , so $P_\mathrm{abs}=\int \pi a^2 Q_\mathrm{abs}(\nu)\,c\,u_\nu\,\mathrm{d}\nu \propto G$ . Why: essentially all absorption happens in the ultraviolet and optical where $Q_\mathrm{abs}$ is flat and near unity, so absorbed power tracks the field brightness with no dependence on grain temperature.

Step 2. Write the emitted power using the same efficiency by Kirchhoff's law, $P_\mathrm{emit}=\int 4\pi a^2 Q_\mathrm{abs}(\nu)\,\pi B_\nu(T_d)\,\mathrm{d}\nu$ , and insert $Q_\mathrm{abs}\propto\nu^\beta$ . Why: the grain emits only where it is warm enough to radiate — the far-infrared — and there the steeply falling efficiency, not the ultraviolet plateau, sets the integrand.

Step 3. Evaluate the integral by substituting $x=h\nu/k_BT_d$ , giving $\int_0^\infty \nu^\beta B_\nu(T_d)\,\mathrm{d}\nu \propto T_d^{4+\beta}\int_0^\infty x^{3+\beta}/(e^x-1)\,\mathrm{d}x$ , the last factor a pure number. Why: the temperature scaling is kinematic — it comes from pulling factors of $T_d$ out of the substitution; the shape integral never touches the exponent.

Step 4. Set $P_\mathrm{abs}=P_\mathrm{emit}$ ; every $\pi a^2$ cancels, leaving $G\propto T_d^{4+\beta}$ and hence $T_d\propto G^{1/(4+\beta)}$ . Why: the grain size drops out because it multiplies absorption and emission identically, so $T_d$ is a property of the field and opacity slope, not of the grain.

Step 5. Read off the numbers: for $\beta=2$ the exponent is $1/6$ , so a hundredfold brighter field raises $T_d$ by only $100^{1/6}\approx2.15$ , and $\beta=0$ recovers Stefan–Boltzmann $T\propto G^{1/4}$ . Why: the thermostat quantified — dust resists heating so strongly that a single temperature describes most of a galaxy's diffuse medium. *Verified: reduces to $T\propto G^{1/4}$ at $\beta=0$ ; matches the Draine (2011) Ch. 24 scaling; the arithmetic $100^{1/6}=2.15$ was checked numerically.*

Go deeper · Derivation B · 1 minWeighing a cloud from its submillimeter glow

Derivation BWeighing a cloud from its submillimeter glow

The goal here is to turn an observed submillimeter flux density into a dust mass, and thence a gas mass, showing precisely which assumptions enter and which quantities dominate the error budget. This is the relation used to weigh prestellar cores where CO has frozen out and dusty high-redshift galaxies where the ultraviolet is entirely absorbed, so its systematics matter as much as the formula itself. We start from the optically thin greybody, integrate over the source, take the Rayleigh–Jeans limit, and read off the uncertainties.

Step 1. In the optically thin limit $I_\nu=\kappa_\nu\Sigma_d B_\nu(T_d)$ ; integrate over the source solid angle and use $\int\Sigma_d\,\mathrm{d}A=M_d$ with $\mathrm{d}A=D^2\,\mathrm{d}\Omega$ to get $S_\nu=\kappa_\nu M_d B_\nu(T_d)/D^2$ . Why: because the medium is transparent to its own radiation, every grain contributes independently, so the flux is a simple sum — the linearity that makes dust a clean mass tracer.

Step 2. Invert to $M_d=S_\nu D^2/[\kappa_\nu B_\nu(T_d)]$ . Why: the mass is linear in the observed flux, so one submillimeter band fixes it up to the two calibrating quantities $\kappa_\nu$ and $T_d$ .

Step 3. At submillimeter frequencies $h\nu\ll k_BT_d$ (at $850\,\mu\mathrm{m}$ , $h\nu/k_B\approx17\,\mathrm{K}\lesssim T_d$ ), so take $B_\nu(T_d)\approx2\nu^2 k_BT_d/c^2$ and insert $\kappa_\nu=\kappa_0(\nu/\nu_0)^\beta$ , giving $M_d=S_\nu D^2 c^2/[2\kappa_0(\nu/\nu_0)^\beta\nu^2 k_BT_d]$ . Why: on the Rayleigh–Jeans tail the Planck function is linear in $T_d$ , so the exponential temperature sensitivity of the mid-infrared is gone and the mass depends on $T_d$ merely as $1/T_d$ .

Step 4. Read the error budget: $M_d$ is inversely linear in $T_d$ and directly linear in $\kappa_0$ , uncertain by a factor of a few between grain models — these dominate over photometric noise. Why: the systematic floor is set by the opacity normalization and assumed temperature, keeping submillimeter masses honest to a factor of two rather than three-digit precision.

Step 5. Convert to gas with $M_\mathrm{gas}\approx100\,M_d$ . Why: it is gas mass that enters the Jeans and virial analyses of collapse, and the ratio is metallicity-dependent, adding one more assumption for external galaxies. *Verified: recovers $S_\nu\propto\nu^{3+\beta}e^{-h\nu/k_BT}$ before the Rayleigh–Jeans step and $S_\nu\propto\nu^{2+\beta}$ after; consistent with the Hildebrand (1983) formalism and Draine (2011) Ch. 23.*

Go deeper · Derivation C · 1 minThe blueshifted peak — why a greybody is not a blackbody

Derivation CThe blueshifted peak — why a greybody is not a blackbody

The goal of this short derivation is to prove that the modified-blackbody spectrum peaks at a higher frequency than a Planck curve of the same temperature, and to compute the exact shift for $\beta=2$ , because this is the origin of the most common practical error in dust work — mistaking a greybody peak for a blackbody peak and overestimating the temperature. We maximize the greybody intensity, obtain a transcendental peak condition, and compare to the blackbody case.

Step 1. Write the greybody intensity up to constants as $I_\nu\propto\nu^{3+\beta}/(e^x-1)$ with $x=h\nu/k_BT_d$ , combining the opacity factor $\nu^\beta$ with the $\nu^3$ of the Planck function. Why: the opacity multiplies the Planck curve, reweighting it toward high frequency — the peak must move blueward before any arithmetic.

Step 2. Set $\mathrm{d}I_\nu/\mathrm{d}x=0$ ; clearing $x^{2+\beta}$ gives the peak condition $(3+\beta)(1-e^{-x})=x$ . Why: extremizing in the dimensionless $x$ absorbs temperature and constants, so one equation locates the peak for any $\beta$ .

Step 3. Solve numerically: $\beta=0$ (pure blackbody in $\nu$ ) gives $x=2.821$ , while $\beta=2$ gives $x=4.965$ . Why: the peak $x$ rises with $\beta$ ; $2.821$ is just the standard Wien result recovered as the $\beta=0$ case.

Step 4. Translate to wavelength at $T_d=20\,\mathrm{K}$ : the blackbody peaks at $\lambda\approx255\,\mu\mathrm{m}$ , the $\beta=2$ greybody at $\lambda\approx145\,\mu\mathrm{m}$ , a factor $4.965/2.821\approx1.76$ bluer. Why: fitting the greybody peak with a Planck function returns $T_\mathrm{fit}=1.76\,T_d$ — the exact trap the misconception box warns against. *Verified: the peak condition, the roots $2.821$ and $4.965$ , and the wavelength ratio $1.76$ were all checked numerically.*

Limiting cases

The optically thin modified blackbody $I_\nu=\kappa_\nu\Sigma_d B_\nu(T_d)$ collapses to a different simple law in each regime, and knowing which one you are in decides what the data can tell you.

Case	Result	Meaning
Rayleigh–Jeans (submm, $h\nu\ll k_BT_d$ )	$S_\nu \propto \kappa_0\,\nu^{2+\beta}\,T_d\,M_d\,D^{-2}$	Flux linear in dust mass and temperature; the $\nu^{2+\beta}$ slope diagnoses $\beta$ . The basis of submillimeter mass estimates.
Wien tail (mid-IR, $h\nu\gg k_BT_d$ )	$S_\nu \propto \nu^{3+\beta}\,e^{-h\nu/k_BT_d}$	Exponentially sensitive to $T_d$ ; a trace of warm dust dominates the mid-infrared and biases single- $T$ fits high.
Optically thick (dense cores, $\tau_\nu\gg1$ )	$I_\nu \to B_\nu(T_d)$	Emission saturates to a blackbody; column information is lost and only $T_d$ is recoverable.
Stochastic heating (very small grains, PAHs)	$P(T)\,\mathrm{d}T \neq \delta(T-T_d)$	No single equilibrium temperature; a distribution produces broad mid-infrared bands rather than a smooth greybody.

Common trapDo not read a dust SED peak as if it were a blackbody

The opacity factor $\kappa_\nu\propto\nu^\beta$ multiplies the Planck function, so the modified-blackbody peak sits at higher frequency than $B_\nu(T_d)$ alone — a factor $1.76$ bluer for $\beta=2$ at $20\,\mathrm{K}$ , as the third derivation shows. Fit a pure Planck function to that peak and you overestimate $T_d$ by the same factor. Two traps compound it. First, a single-temperature fit to a sight line mixing warm and cold grains is biased warm, because the Wien tail $\nu^{3+\beta}e^{-h\nu/k_BT}$ lets a little warm dust dominate the short-wavelength flux while the cold, mass-dominant dust is underweighted. Second, in noisy data $T_d$ and $\beta$ are strongly anti-correlated — raising one and lowering the other leaves the curve almost unchanged — so never quote a fitted temperature or index without its covariance.

Worked examples

1 · The stiff thermostat (confidence builder)

Diffuse-ISM dust ( $G=1$ ) sits at $T_d\approx18\,\mathrm{K}$ . Estimate its temperature at the edge of an H II region where the field is $10^4$ times stronger, taking $\beta=2$ .

Step 1. The thermostat law is $T_d\propto G^{1/(4+\beta)}$ ; for $\beta=2$ the exponent is $1/6$ , so the ratio is $(10^4)^{1/6}=10^{4/6}\approx\mathbf{4.64}$ .

Step 2. Scale the reference: $T_d(G=10^4)\approx18\,\mathrm{K}\times4.64\approx\mathbf{84\,\mathrm{K}}$ .

Step 3. Interpret it: a four-order-of-magnitude jump in the field raises the temperature by less than a factor of five, and warm dust near $\sim80$ – $100\,\mathrm{K}$ is just what is seen bordering luminous star-forming regions. Had the grains been grey ( $\beta=0$ ), the exponent would be $1/4$ and the same field would give $180\,\mathrm{K}$ — the opacity slope is what keeps real dust cool. $T_d\approx84\,\mathrm{K}$ , a factor $4.6$ warmer for a $10^4$ brighter field.

2 · Weighing a prestellar core from one submillimeter band (fully worked)

A prestellar core in Taurus ( $D=140\,\mathrm{pc}$ ) is measured at $S_\nu=1\,\mathrm{Jy}$ at $850\,\mu\mathrm{m}$ . Taking $T_d=15\,\mathrm{K}$ , $\kappa_{850}=0.02\,\mathrm{cm^2\,g^{-1}}$ per gram of dust, and gas-to-dust $100$ , find the dust and gas masses.

Step 1. Frequency and Planck function. At $850\,\mu\mathrm{m}$ , $\nu=3.53\times10^{11}\,\mathrm{Hz}$ , so $h\nu/k_B=16.9\,\mathrm{K}$ and $x=16.9/15=1.13$ . Then $B_\nu(T_d)=(2h\nu^3/c^2)/(e^{x}-1)=(6.47\times10^{-16})/2.09=\mathbf{3.10\times10^{-16}\,W\,m^{-2}\,Hz^{-1}\,sr^{-1}}$ .

Step 2. Convert to SI. $S_\nu=1\,\mathrm{Jy}=10^{-26}\,\mathrm{W\,m^{-2}\,Hz^{-1}}$ ; $D=140\,\mathrm{pc}=4.32\times10^{18}\,\mathrm{m}$ so $D^2=1.87\times10^{37}\,\mathrm{m^2}$ ; $\kappa_{850}=0.02\,\mathrm{cm^2\,g^{-1}}=2.0\times10^{-3}\,\mathrm{m^2\,kg^{-1}}$ (since $1\,\mathrm{cm^2\,g^{-1}}=0.1\,\mathrm{m^2\,kg^{-1}}$ ).

Step 3. Apply $M_d=S_\nu D^2/[\kappa_\nu B_\nu(T_d)]$ . Numerator $=10^{-26}\times1.87\times10^{37}=1.87\times10^{11}$ ; denominator $=2.0\times10^{-3}\times3.10\times10^{-16}=6.2\times10^{-19}$ ; hence $M_d=3.0\times10^{29}\,\mathrm{kg}=\mathbf{0.15\,M_\odot}$ .

Step 4. Gas mass. $M_\mathrm{gas}\approx100\,M_d=\mathbf{15\,M_\odot}$ , a genuine prestellar clump whose comparison against the local Jeans mass decides whether it collapses. Every digit past the first is illusory: halving the assumed $T_d$ roughly doubles $M_d$ , and the factor-of-a-few uncertainty in $\kappa_0$ dominates the rest — the honest answer is " $\sim15\,M_\odot$ , good to a factor of two."

Physics of the portrait

The modified blackbody is the concept the whole subject turns on, a hybrid of nineteenth-century thermodynamics and twentieth-century solid-state astrophysics. Kirchhoff supplied the principle that a good absorber is a good emitter, Planck supplied the spectrum, and a lineage of ISM physicists supplied the frequency-dependent efficiency that turns the Planck curve into the greybody we observe.

Origins · 1 minmodified blackbody and how this was discovered

modified blackbody

the greybody law Iν = κν Bν(Td) with κν ∝ νᵝ · thermal emission from interstellar dust

The modified blackbody is not a person but a law, assembled over more than a century. Its deepest root is Gustav Kirchhoff's 1860 statement that, in thermal equilibrium, a body's emissivity equals its absorptivity at every frequency — the reason the same $Q_\nu$ that governs how a grain absorbs also governs how it radiates. Max Planck's 1900 derivation supplied the function $B_\nu(T)$ a perfect emitter would follow. The astrophysical twist is that a small grain is a bad emitter at wavelengths larger than itself, so its spectrum is the Planck curve tilted by $Q_\nu\propto\nu^\beta$ — the "modification". Robert Hildebrand crystallized the practical form in 1983, writing the optically thin flux as $\kappa_\nu M_d B_\nu(T_d)/D^2$ and giving observers a recipe to weigh clouds from their glow; that formula, opacity uncertainty and all, remains the field standard. Bruce Draine and collaborators grounded the opacity in laboratory grain physics, and the ESA Planck mission measured the diffuse Galactic greybody across the sky, pinning $T_d\approx19$ – $20\,\mathrm{K}$ , $\beta\approx1.5$ — and exposing it as the foreground between us and the primordial B-mode. A thermodynamic identity is now a per-pixel fitting function drawn across the microwave sky.

History

1860

— Kirchhoff states that a good absorber is a good emitter (emissivity equals absorptivity)

1900

— Planck derives the blackbody radiation law Bν(T)

1983

— Hildebrand sets out the optically thin submillimeter dust-mass method

1984

— IRAS surveys the far-infrared sky and reveals diffuse dust cirrus

2014

— Planck maps polarized Galactic dust as the leading CMB B-mode foreground

Going further

Texts. Carroll and Ostlie, An Introduction to Modern Astrophysics, 2nd ed., Ch. 12 (the interstellar medium and star formation) · Draine, Physics of the Interstellar and Intergalactic Medium (2011), Ch. 23–24 (dust emission and grain temperature) · Binney and Tremaine, Galactic Dynamics, 2nd ed. (ISM context) · Ryden, Introduction to Cosmology, 2nd ed. (blackbody radiation and CMB foregrounds) · Dodelson, Modern Cosmology (foreground contamination of the CMB).

Provenance. The thermostat scaling $T_d\propto G^{1/(4+\beta)}$ and the emission integral $\int\nu^\beta B_\nu\,\mathrm{d}\nu\propto T_d^{4+\beta}$ are re-derived from radiative balance (check against Draine Ch. 24; the $\beta=0$ limit recovers Stefan–Boltzmann $T\propto G^{1/4}$ ). The submillimeter mass relation follows Hildebrand (1983) and Draine Ch. 23; the worked $M_d\approx0.15\,M_\odot$ ( $M_\mathrm{gas}\approx15\,M_\odot$ ) for $1\,\mathrm{Jy}$ at $850\,\mu\mathrm{m}$ , $140\,\mathrm{pc}$ , $15\,\mathrm{K}$ , $\kappa_{850}=0.02\,\mathrm{cm^2\,g^{-1}}$ was computed explicitly. The peak condition $(3+\beta)(1-e^{-x})=x$ with roots $2.821$ ( $\beta=0$ ) and $4.965$ ( $\beta=2$ ), and the $1.76$ blueshift at $20\,\mathrm{K}$ ( $255$ vs $145\,\mu\mathrm{m}$ ), were solved numerically. Standard textbook and Planck values are used for $T_d\sim10$ – $50\,\mathrm{K}$ , $\beta\approx1.5$ – $2$ , gas-to-dust $\sim100$ , and Galactic $T_d\approx19$ – $20\,\mathrm{K}$ ; $\kappa_0$ is model-dependent and quoted to order of magnitude only.

Prerequisites. Interstellar Dust Grains · Interstellar Extinction and Reddening · Interstellar Gas Heating and Cooling Processes · Phases of the Interstellar Medium. Leads to. Molecular Clouds and Giant Molecular Clouds · CO as a Molecular Gas Tracer and the X-Factor · Interstellar Polarization and Magnetic Fields · The Jeans Instability and Jeans Mass.

Ask the tutor: Ask me to walk through weighing a real cloud from its submillimeter flux, or to show exactly how the $T_d$ – $\beta$ degeneracy deforms an SED fit.

Why is the emission in the far-infrared if the dust absorbs ultraviolet light?

Because absorption and emission are set by different spectra. The grain absorbs where the photons are — mostly the ultraviolet and optical, where starlight is bright and the cross-section is large. But once heated it radiates at its own temperature, only $\sim20\,\mathrm{K}$ , and a $20\,\mathrm{K}$ body peaks by Wien's law near $\lambda\sim2.9\times10^{3}/20\approx145\,\mu\mathrm{m}$ — deep in the far-infrared. Energy is conserved but redistributed from a few ultraviolet photons into a great many far-infrared ones. The grain is a photon down-converter.

What exactly is the emissivity index β, and why does it matter?

It is the power-law slope of the dust opacity, $\kappa_\nu\propto\nu^\beta$ , encoding grain composition: silicate and carbonaceous grains give $\beta\approx1.5$ – $2$ . It does two jobs — it sets the submillimeter slope $S_\nu\propto\nu^{2+\beta}$ that you fit, and it controls the blueshift of the SED peak. The catch is the $T_d$ – $\beta$ degeneracy: in noisy data, raising $\beta$ and lowering $T_d$ deform the curve almost identically, so the two are anti-correlated and neither should be quoted without the fit covariance.

Why is dust such a good gas-mass tracer when CO is so widely used?

CO is a molecule, and in cold dense cores ( $T\lesssim15\,\mathrm{K}$ , $n\gtrsim10^5\,\mathrm{cm^{-3}}$ ) it freezes onto grain mantles and vanishes from the gas phase, so CO underestimates the mass exactly where the star-forming action is. Dust does not freeze out — it is the grain. In the optically thin far-infrared the continuum brightness is a near-linear sum of every grain along the sight line, measuring dust column and, via the gas-to-dust ratio, total gas column. The price is assuming $\kappa_0$ and a temperature, whereas CO's price is a variable conversion factor; the two are complementary.

How can a submillimeter flux at one frequency give a mass at all?

Because in the optically thin Rayleigh–Jeans regime the flux collapses to $S_\nu\approx\kappa_\nu M_d B_\nu(T_d)/D^2$ with $B_\nu\approx2\nu^2k_BT_d/c^2$ , so once you fix $D$ , $T_d$ , and $\kappa_\nu$ , the only unknown is $M_d$ — linear in the measured flux. One band suffices for the mass; a second only breaks the $T_d$ – $\beta$ degeneracy. The answer is good to a factor of two at best, since $\kappa_0$ is uncertain by a few and $M_d\propto1/T_d$ .

Why does Galactic dust threaten measurements of the cosmic microwave background?

Because the same $\sim20\,\mathrm{K}$ greybody that traces our Galaxy's dust glows all over the microwave sky, and above $\sim100\,\mathrm{GHz}$ it outshines the CMB. Worse, it is polarized: elongated grains align with the Galactic magnetic field and radiate anisotropically, mimicking the polarization a primordial gravitational-wave background would leave. The 2014 BICEP2 claim of an inflationary B-mode proved consistent with this unmodeled dust. Every next-generation experiment therefore observes in many channels, using dust's $\nu^{2+\beta}$ spectrum to separate and subtract it — now a leading systematic in the search for inflation.

Is the dust really all at one temperature along a line of sight?

Almost never, and pretending otherwise is the classic bias. A real sight line threads warm dust near heating sources and cold dust in shielded interiors, and because the Wien tail scales as $\nu^{3+\beta}e^{-h\nu/k_BT}$ , even a little warm dust dominates the short-wavelength flux and pulls a single-temperature fit warm. Since the cold component carries most of the mass, that warm bias underestimates the mass. This is why careful analyses use multi-band photometry, sometimes two-temperature models, and quote $T_d$ as a luminosity-weighted average.