Bremsstrahlung (Free-Free Emission)

Bremsstrahlung — German for "braking radiation" — is the light emitted whenever a charged particle is accelerated. In astrophysical plasmas the dominant case is a light, fast electron whipping past the near-static Coulomb field of a heavier ion: the electron's path curves, it accelerates, and any accelerating charge radiates. Because both the initial and final electron states are unbound, the process is called free-free emission, and unlike bound-bound line transitions or bound-free recombination it produces a smooth continuum rather than sharp lines. The electron surrenders a small slice of its kinetic energy as a photon in each hyperbolic encounter, and the collective glow of a whole thermal electron population is the thermal bremsstrahlung spectrum. The theory is built in four moves. The Larmor formula fixes the instantaneous power radiated by an accelerating charge. A Fourier analysis of the acceleration in one flyby gives the energy radiated per collision as a function of impact parameter $b$ and speed $v$ . Integrating over impact parameters introduces a Coulomb logarithm whose divergences are cut off by quantum mechanics and absorbed into the order-unity Gaunt factor. Finally, averaging over a Maxwell–Boltzmann distribution at temperature $T$ yields the thermal free-free emissivity. The result is strikingly simple. The spectrum is nearly flat below the thermal energy and cuts off exponentially as $e^{-h\nu/k_BT}$ above it, so the position of the knee reads off the plasma temperature. The frequency-integrated power scales as $n_e n_i Z^2 T^{1/2}$ , so hotter, denser gas cools faster. In LTE, Kirchhoff's law converts the emissivity into the free-free (Kramers') opacity that governs stellar interiors. Bremsstrahlung is thus a master diagnostic underlying the X-ray glow of galaxy clusters, the radio continuum of HII regions, and stellar-structure opacity.

Key takeaways

Bremsstrahlung ("braking radiation") is the free-free continuum emitted when a free electron is accelerated in the Coulomb field of an ion, radiating without changing any bound state.
A single encounter radiates by the Larmor formula; integrating over impact parameters and a Maxwellian electron distribution gives the thermal free-free emissivity.
The spectrum is flat up to $h\nu\approx k_BT$ and then cuts off as $e^{-h\nu/k_BT}$ , so the exponential knee measures the gas temperature directly.
The frequency-integrated cooling rate scales as $\epsilon^{ff}\propto n_e n_i Z^2 T^{1/2}$ — the signature square-root temperature law of thin, hot plasmas.
Classical divergences at small impact parameter are regularised by the dimensionless Gaunt factor $\bar{g}_{ff}\sim1$ ; in LTE, Kirchhoff's law gives Kramers' free-free opacity $\bar\kappa\propto\rho T^{-7/2}$ .
Galaxy clusters ( $T\sim10^7$ – $10^8$ K) glow in thermal X-ray bremsstrahlung, HII regions emit a free-free radio continuum, and free-free absorption sets stellar opacity.
Key idea: every accelerated charge radiates (Larmor); a thermal electron gas scattering off ions therefore emits a smooth continuum, flat below $h\nu\approx k_BT$ and exponentially cut off above it, with total power $\propto n_e n_i Z^2 T^{1/2}$ .

The core insight

Every accelerated charge radiates (Larmor); a thermal electron gas scattering off ions therefore emits a smooth continuum whose spectrum is flat below $h\nu\approx k_BT$ and exponentially cut off above it, with total power $\propto n_e n_i Z^2 T^{1/2}$ .

Where this lives today

Bremsstrahlung is the diagnostic backbone of modern hot-plasma astrophysics. X-ray observatories — Chandra, XMM-Newton, and the eROSITA all-sky survey — fit the exponential cutoff of the intracluster medium to map cluster temperatures and, through the emission-measure normalisation, gas masses and hence total (dark-matter-dominated) masses via hydrostatic equilibrium. In cosmology, free-free emission from ionized gas is a named foreground that Planck and CMB experiments must separate from the primordial signal. In the radio, the ngVLA and SKA will use optically thin free-free emission as a dust-free tracer of star-formation rates and resolve the optically thick turnover in compact HII regions. The same physics is a laboratory diagnostic: bremsstrahlung from tokamak plasmas (JET, ITER) measures effective ion charge $Z_{\rm eff}$ and electron temperature. And a standing open problem — the "cooling-flow problem" of why cluster cores do not cool as fast as their bremsstrahlung luminosity predicts — links radiative cooling to AGN feedback.

The physical picture

Picture a fast electron whistling past a heavy positive ion. The Coulomb attraction bends its path into a hyperbola; while it is being bent it is accelerating, and any accelerating charge radiates. The closer the flyby, the harder the kick and the higher the frequency of the emitted photon. A whole plasma is a swarm of such encounters over every impact parameter and speed, so the emission is a smooth continuum, not lines.

Figure

A single Coulomb encounter: a fast electron of speed v is deflected by an ion of charge +Ze at impact parameter b, and the acceleration a = Ze²/(m_e r²) makes it radiate a photon hν — Fourier-analysing the kick gives the flat single-collision spectrum dW/dω ∝ Z²e⁶/(v²b²)

The exponential cutoff has a one-line origin: an electron cannot radiate a photon carrying more energy than it possesses, and in a thermal gas the fraction of electrons with kinetic energy above $k_BT$ falls off as $e^{-E/k_BT}$ , so the photon spectrum inherits that tail. Below the cutoff nearly every electron can contribute, so the spectrum is flat. The second figure is what an X-ray astronomer actually measures — the spectrum of a real cluster plasma, whose exponential knee is a thermometer.

Figure

Thermal bremsstrahlung spectrum of a T = 10⁸ K cluster plasma (kT = 8.6 keV): the emissivity εν is nearly flat at low photon energy and falls as e^(−hν/kT) above the knee at hν ≈ kT, so the exponential cutoff reads the gas temperature directly

Why $\propto T^{1/2}$ rather than the blackbody $T^4$ ? Blackbody radiation counts photons in equilibrium with matter; bremsstrahlung counts collisions. The collision rate per electron is $\propto n_i v\propto n_i T^{1/2}$ , and each collision radiates, so total power scales as $n_e n_i T^{1/2}$ . It is a thin-plasma, non-equilibrium mechanism — which is exactly why it lets us see straight through a cluster and weigh the gas.

The core equations

Everything begins with the Larmor formula: the instantaneous power radiated by a nonrelativistic charge is set by the square of its acceleration. This is the seed from which the whole free-free apparatus grows.

P=\frac{2}{3}\frac{q^2 a^2}{c^3}=\frac{q^2 a^2}{6\pi\varepsilon_0 c^3} \qquad (1)

where: $P$ — instantaneous radiated power; $q$ — charge; $a$ — magnitude of acceleration; $c$ — speed of light; $\varepsilon_0$ — vacuum permittivity (SI form). Reading: because the Coulomb acceleration $a\propto1/m$ , the power $\propto a^2\propto1/m^2$ — the electron, being $\sim1836$ times lighter than a proton, outradiates the ion by that factor squared, so bremsstrahlung is an electron process and the ion merely supplies the field. Fourier-analysing one flyby in the dipole approximation turns Larmor into a spectrum. For a straight-line pass at impact parameter $b$ and speed $v$ , the acceleration is a pulse of duration $\sim b/v$ , whose transform is flat up to $\omega\sim v/b$ and falls off above it.

\frac{\mathrm{d}W}{\mathrm{d}\omega}(b,v)\approx\frac{8Z^2e^6}{3\pi c^3 m_e^2 v^2 b^2}\qquad(\omega\ll v/b) \qquad (2)

where: $\mathrm{d}W/\mathrm{d}\omega$ — energy radiated per unit angular frequency per encounter; $b$ — impact parameter; $v$ — electron speed; $Z$ — ion charge number; $e$ — elementary charge (Gaussian units). Reading: the spectrum from a single collision is flat, and its strong $1/b^2$ weighting means close encounters dominate — which is what forces the impact-parameter integral, and its logarithm, in the thermal average. Averaging equation (2) over impact parameters and a Maxwellian speed distribution gives the thermal free-free emissivity. Two features survive the averaging: the $T^{-1/2}$ prefactor from the $\langle1/v\rangle$ collision weighting, and the $e^{-h\nu/k_BT}$ cutoff from the requirement $\tfrac12 m_ev^2\ge h\nu$ .

\epsilon_\nu^{ff}=\frac{32\pi e^6}{3 m_e c^3}\left(\frac{2\pi}{3 k_B m_e}\right)^{1/2}T^{-1/2}\,Z^2 n_e n_i\,e^{-h\nu/k_BT}\,\bar{g}_{ff} \qquad (3)

where: $\epsilon_\nu^{ff}$ — power per unit volume, per unit frequency, integrated over solid angle; $n_e,n_i$ — electron and ion number densities; $T$ — electron temperature; $k_B$ — Boltzmann constant; $\bar{g}_{ff}$ — velocity-averaged Gaunt factor ( $\sim1$ ); $h$ — Planck constant. Reading: the whole spectral shape is a flat plateau multiplied by an exponential knee, with normalisation set by the emission measure $n_en_i$ . Collecting the constants in CGS gives the working formula X-ray astronomers keep at hand.

\epsilon_\nu^{ff}\approx 6.8\times10^{-38}\,Z^2\,n_e n_i\,T^{-1/2}\,e^{-h\nu/k_BT}\,\bar{g}_{ff}\quad[\text{erg s}^{-1}\text{cm}^{-3}\text{Hz}^{-1}] \qquad (4)

where: $n_e,n_i$ in cm $^{-3}$ ; $T$ in K; $\epsilon_\nu^{ff}$ in erg s $^{-1}$ cm $^{-3}$ Hz $^{-1}$ . Reading: plug in densities and temperature and you have the specific emissivity at any frequency below the cutoff — the quantity that, integrated along a line of sight, becomes an observed surface brightness. Integrating equation (4) over all frequencies collapses the exponential to a factor $k_BT/h$ and produces the total volumetric cooling rate — the celebrated square-root law.

\epsilon^{ff}=\int_0^\infty\epsilon_\nu^{ff}\,\mathrm{d}\nu\approx 1.4\times10^{-27}\,T^{1/2}\,Z^2\,n_e n_i\,\bar{g}_B\quad[\text{erg s}^{-1}\text{cm}^{-3}] \qquad (5)

where: $\epsilon^{ff}$ — total radiated power per unit volume; $\bar{g}_B$ — frequency-averaged Gaunt factor $\approx1.1$ – $1.5$ . Reading: the $T^{-1/2}$ prefactor times the $T^{+1}$ from the frequency integral gives the net $T^{+1/2}$ ; this is what sets the cooling time of every hot diffuse plasma in the universe. In local thermodynamic equilibrium, Kirchhoff's law ties emission to absorption through the Planck function, converting the emissivity into the free-free absorption coefficient.

\alpha_\nu^{ff}=\frac{\epsilon_\nu^{ff}}{4\pi B_\nu(T)}\ \propto\ Z^2 n_e n_i\,T^{-1/2}\,\nu^{-3}\left(1-e^{-h\nu/k_BT}\right)\bar{g}_{ff} \qquad (6)

where: $\alpha_\nu^{ff}$ — free-free absorption coefficient (per unit length); $B_\nu(T)$ — Planck function; $(1-e^{-h\nu/k_BT})$ — the stimulated-emission correction. Reading: at radio frequencies the bracket $\to h\nu/k_BT\propto\nu$ , so $\alpha_\nu^{ff}\propto\nu^{-2}$ — opacity grows steeply toward low frequency, which is why HII regions turn optically thick in the radio and their spectra bend from $\nu^{-0.1}$ to $\nu^{2}$ . Two derived quantities close the toolkit: the Rosseland-mean opacity of stellar structure, and the emission measure that observers fit.

\bar{\kappa}^{ff}_{\rm Ross}\propto\rho\,T^{-7/2},\qquad \mathrm{EM}\equiv\int n_e n_i\,\mathrm{d}\ell \qquad (7)

where: $\bar\kappa^{ff}_{\rm Ross}$ — Rosseland-mean free-free (Kramers') opacity per unit mass; $\rho$ — mass density; $\mathrm{EM}$ — emission measure, the line-of-sight integral of $n_en_i$ . Reading: Kramers' $\rho T^{-7/2}$ enters the radiative-diffusion equation deep in stars, while the observed flux $F_\nu\propto\mathrm{EM}\,T^{-1/2}e^{-h\nu/k_BT}\bar g_{ff}$ makes the emission measure the single number linking spectrum to gas content.

Derivation ASingle-encounter spectrum from Larmor and Fourier analysis

The goal is to turn the instantaneous Larmor power into the energy radiated per unit frequency in one electron–ion flyby, and to show why that per-collision spectrum is flat up to a cutoff $\omega\sim v/b$ . We treat the trajectory as nearly straight (small deflection, valid for all but the closest passes), Fourier-transform the acceleration, and invoke Parseval's theorem to convert a time integral of $|a(t)|^2$ into a frequency integral that defines $\mathrm{d}W/\mathrm{d}\omega$ . This is the classical backbone of every free-free result that follows.

Step 1. Start from Larmor in the time domain, $\mathrm{d}W/\mathrm{d}t=(2e^2/3c^3)|a(t)|^2$ (Gaussian units, $q=e$ ), so the total energy radiated in the encounter is $W=\int|a(t)|^2\,(2e^2/3c^3)\,\mathrm{d}t$ . Why: all the frequency content of the emission is carried by the time profile of the acceleration, so the spectrum must come from transforming $a(t)$ .

Step 2. Apply Parseval's theorem, $\int|a(t)|^2\mathrm{d}t=2\int_0^\infty|\hat a(\omega)|^2\mathrm{d}\omega$ , to rewrite the energy as a frequency integral and read off the spectral density

\frac{\mathrm{d}W}{\mathrm{d}\omega}=\frac{2e^2}{3\pi c^3}\,|\hat a(\omega)|^2 .

Why: Parseval is what legitimises "energy per unit frequency" — it guarantees the frequency integral of $\mathrm{d}W/\mathrm{d}\omega$ returns the total radiated energy.

Step 3. Note the acceleration pulse has duration $\tau\sim b/v$ (the time spent near closest approach), so its transform $\hat a(\omega)$ is essentially constant for $\omega\ll v/b$ and collapses for $\omega\gg v/b$ . Why: a pulse of width $\tau$ has a flat Fourier spectrum out to $\omega\sim1/\tau$ — this is the origin of the flat single-collision spectrum and its sharp high-frequency knee.

Step 4. Estimate $|\hat a(\omega)|^2$ from the Coulomb kick $a\sim Ze^2/(m_e b^2)$ acting for time $\tau\sim b/v$ , giving $\mathrm{d}W/\mathrm{d}\omega\approx 8Z^2e^6/(3\pi c^3 m_e^2 v^2 b^2)$ for $\omega\ll v/b$ . Why: this recovers equation (2), exposes the crucial $1/b^2$ weighting, and shows the emission is set by $Z^2$ and by the electron mass squared in the denominator.

Go deeper · Derivation B · 1 minFrom single collisions to the thermal emissivity

Derivation BFrom single collisions to the thermal emissivity

The goal is to assemble the plasma emissivity $\epsilon_\nu^{ff}$ by summing single-collision spectra over all impact parameters and all electron speeds in a Maxwellian gas, and to see the two signatures — the $T^{-1/2}$ prefactor and the $e^{-h\nu/k_BT}$ cutoff — emerge from the averaging. We build the collision rate, integrate over $b$ to get the Coulomb logarithm, then average over the Maxwell–Boltzmann distribution with the correct lower velocity limit set by energy conservation.

Step 1. For one electron of speed $v$ , the rate of encounters with ions in the impact-parameter ring $[b,b+\mathrm{d}b]$ is $\mathrm{d}R=n_i\,v\,2\pi b\,\mathrm{d}b$ . Why: multiplying the per-collision spectrum by this rate and by $n_e$ converts single-encounter energy into power per unit volume.

Step 2. The product $(\mathrm{d}W/\mathrm{d}\omega)\times b\,\mathrm{d}b\propto b^{-2}\cdot b\,\mathrm{d}b=\mathrm{d}b/b$ integrates to $\ln(b_{max}/b_{min})$ , the Coulomb logarithm. The upper cutoff $b_{max}\sim v/\omega$ comes from the spectral falloff; the lower cutoff $b_{min}$ from quantum uncertainty or large-angle scattering — both are packaged into $\bar g_{ff}$ . Why: the classical integral would diverge without cutoffs, and the Gaunt factor is precisely the finite, slowly varying number that replaces the ill-defined logarithm.

Step 3. Average over the Maxwellian $f(v)\propto v^2 e^{-m_ev^2/2k_BT}$ , but only electrons with $\tfrac12 m_ev^2\ge h\nu$ can emit a photon $h\nu$ , so the speed integral starts at $v_{min}=\sqrt{2h\nu/m_e}$ . Why: this energy-conservation floor is the physical seed of the exponential cutoff — no electron radiates a photon it cannot pay for.

Step 4. The per-collision spectrum carries $\propto1/v$ , so the average is $\int_{v_{min}}^\infty v^{-1}v^2 e^{-m_ev^2/2k_BT}\mathrm{d}v\propto T^{1/2}e^{-h\nu/k_BT}$ ; folding in the overall normalisation leaves the net $T^{-1/2}e^{-h\nu/k_BT}$ of equation (3), with CGS coefficient $6.8\times10^{-38}$ . Why: the lower limit produces the Boltzmann factor while the $1/v$ weighting flips one power of $T$ , giving the counterintuitive $T^{-1/2}$ spectral prefactor.

Go deeper · Derivation C · 1 minThe T to the one-half cooling law and Kramers opacity

Derivation CThe T to the one-half cooling law and Kramers opacity

The goal is to integrate the spectral emissivity over frequency to obtain the total cooling rate, showing why the spectral $T^{-1/2}$ becomes a total $T^{+1/2}$ , and then to use Kirchhoff's law in LTE to turn the same emissivity into the free-free absorption coefficient and the Rosseland-mean Kramers opacity that governs stellar interiors. This links the observable cooling of cluster gas to the opacity that shapes stars.

Step 1. Integrate equation (4) over frequency. The only strongly $\nu$ -dependent factor is the exponential, and $\int_0^\infty e^{-h\nu/k_BT}\mathrm{d}\nu=k_BT/h$ . Why: the flat plateau times an exponential of width $k_BT/h$ has an area set entirely by that width, which is proportional to $T$ .

Step 2. Combine the scalings: $\epsilon^{ff}\propto T^{-1/2}\times T\times Z^2n_en_i=T^{1/2}Z^2n_en_i$ , giving the coefficient $1.4\times10^{-27}$ (CGS). The cooling time then follows as $t_{cool}\sim\tfrac{3}{2}(n_e+n_i)k_BT/\epsilon^{ff}\propto T^{1/2}/n$ . Why: this is the master result — hotter and denser plasmas cool faster, setting the clock for cluster cooling flows and post-shock gas.

Step 3. In LTE the source function equals the Planck function, so Kirchhoff's law $\epsilon_\nu^{ff}/4\pi=\alpha_\nu^{ff}B_\nu(T)$ gives $\alpha_\nu^{ff}=\epsilon_\nu^{ff}/(4\pi B_\nu)$ . Dividing the $\nu$ -independent emissivity by $B_\nu\propto\nu^3/(e^{h\nu/k_BT}-1)$ produces the $\nu^{-3}(1-e^{-h\nu/k_BT})$ shape of equation (6). Why: absorption is not an independent quantity in LTE — it is fixed by the emissivity and the temperature, which is what makes the free-free opacity computable.

Step 4. Form the Rosseland harmonic mean of $1/\alpha_\nu^{ff}$ over the Planck-weighted spectrum. The $\nu^{-3}$ frequency dependence and the extra factor of $T$ from the weighting yield the Kramers law $\bar\kappa^{ff}_{\rm Ross}\propto\rho\,T^{-7/2}$ . Why: this steep temperature dependence dominates the opacity deep in stellar envelopes and helps set the temperature gradient and the mass–luminosity relation of intermediate-mass stars.

Limiting cases

Case	Result	Meaning
Low frequency ( $h\nu\ll k_BT$ )	$\epsilon_\nu^{ff}\approx\text{const}\times\bar g_{ff}(\nu,T)$	Nearly flat spectrum; the only frequency dependence is the slowly varying Gaunt factor. The radio free-free continuum lives here.
High frequency ( $h\nu\gtrsim k_BT$ )	$\epsilon_\nu^{ff}\propto e^{-h\nu/k_BT}$	Exponential cutoff; the position of the knee measures the plasma temperature (X-ray cluster spectroscopy).
Optically thick, low $\nu$	$S_\nu=B_\nu(T)\Rightarrow I_\nu\propto\nu^{2}$	Where $\alpha_\nu^{ff}\propto\nu^{-2}$ grows large, HII regions become optically thick and radiate the Rayleigh–Jeans blackbody form.
Relativistic electrons	$E_e\gtrsim m_ec^2$	Relativistic corrections change the angular pattern and cross section; distinct from the nonrelativistic thermal case treated here.
Non-thermal electrons	power-law $N(E)\propto E^{-p}$	A power-law electron distribution gives a power-law free-free continuum, not the Maxwellian shape — seen in some flares and jets.

Common trapThe classical rate diverges, so bremsstrahlung must be "unphysical"

The classical single-collision spectrum grows without bound as the impact parameter $b\to0$ , and the impact-parameter integral $\int\mathrm{d}b/b$ diverges at both ends. This is not a physical divergence. At small $b$ the electron's quantum wavelength forbids arbitrarily close, well-defined encounters; at large $b$ the finite emission frequency sets a natural cutoff $b_{max}\sim v/\omega$ . Both effects are bundled into the dimensionless Gaunt factor $\bar g_{ff}$ , an order-unity number computed once and for all from the exact quantum cross section (Karzas and Latter). The practical rule: never quote a bremsstrahlung rate "from first principles" without its Gaunt factor, and never apply the thermal formulae to a plasma that is not Maxwellian — a non-thermal electron population radiates a completely different continuum.

Worked examples

1 · Why cluster gas does not cool (fully worked)

A galaxy cluster's intracluster medium has $T=10^8$ K, $n_e=n_i=10^{-3}$ cm $^{-3}$ , $Z=1$ , and $\bar g_B=1.2$ . Find the volumetric cooling rate and the cooling time, and say what it implies.

Step 1. Convert the temperature to a photon energy: $k_BT=(8.617\times10^{-5}\ \text{eV K}^{-1})(10^8\ \text{K})=8.6\times10^3$ eV $=\mathbf{8.6\ keV}$ — the exponential knee sits in the hard-X-ray band, exactly where Chandra and XMM observe.

Step 2. Apply equation (5): $\epsilon^{ff}=1.4\times10^{-27}\times(10^8)^{1/2}\times1\times(10^{-3})(10^{-3})\times1.2$ . With $(10^8)^{1/2}=10^4$ , this is $1.4\times10^{-27}\times10^4\times10^{-6}\times1.2=\mathbf{1.7\times10^{-29}\ erg\ s^{-1}\ cm^{-3}}$ .

Step 3. Compute the thermal energy density of the fully ionized gas: $u=\tfrac32(n_e+n_i)k_BT=1.5\times(2\times10^{-3})\times(1.38\times10^{-16})\times10^{8}=\mathbf{4.1\times10^{-11}\ erg\ cm^{-3}}$ .

Step 4. The cooling time is $t_{cool}=u/\epsilon^{ff}=4.1\times10^{-11}/1.7\times10^{-29}=2.5\times10^{18}$ s $=\mathbf{78\ Gyr}$ . This is far longer than the age of the universe, so the diffuse bulk of a cluster cannot cool radiatively. Only in dense cores, where $n_e\sim0.1$ cm $^{-3}$ , does $t_{cool}\propto1/n$ drop to $\sim\mathbf{0.8\ Gyr}$ — short enough to drive the classic cooling flows seen at cluster centres.

2 · Weighing the hot gas from its X-ray glow (fully worked)

A cluster shows an exponential X-ray cutoff with e-folding energy $\approx6$ keV, an X-ray luminosity $L_X=10^{45}$ erg s $^{-1}$ , and an emitting region of radius $R=1$ Mpc. Estimate the gas density and total gas mass ( $Z=1$ , $\bar g_B=1.2$ ).

Step 1. The e-folding energy equals $k_BT$ , so $T=6000\ \text{eV}/(8.617\times10^{-5}\ \text{eV K}^{-1})=\mathbf{7.0\times10^{7}\ K}$ .

Step 2. The emitting volume is $V=\tfrac43\pi R^3$ with $R=1\ \text{Mpc}=3.09\times10^{24}$ cm, giving $V=\mathbf{1.2\times10^{74}\ cm^3}$ .

Step 3. The mean emissivity is $\epsilon^{ff}=L_X/V=10^{45}/1.23\times10^{74}=\mathbf{8.1\times10^{-30}\ erg\ s^{-1}\ cm^{-3}}$ .

Step 4. Invert equation (5) with $n_i=n_e$ : $n_e^2=\epsilon^{ff}/(1.4\times10^{-27}\,T^{1/2}\,\bar g_B)$ . With $T^{1/2}=8.4\times10^{3}$ , the denominator is $1.4\times10^{-27}\times8.4\times10^{3}\times1.2=1.4\times10^{-23}$ , so $n_e^2=5.8\times10^{-7}$ and $n_e=\mathbf{7.6\times10^{-4}\ cm^{-3}}$ — a textbook cluster density.

Step 5. The gas mass is $M_{gas}\approx n_e m_p V=(7.6\times10^{-4})(1.67\times10^{-24})(1.23\times10^{74})=1.6\times10^{47}$ g $=\mathbf{7.9\times10^{13}\ M_\odot}$ . About $10^{14}$ solar masses of hot gas — inferred from nothing but the brightness and the shape of a bremsstrahlung spectrum.

Physics of the portrait

The quantitative theory of free-free radiation and its opacity is the work of Hendrik Kramers, who in 1923 turned the continuous X-ray spectrum into a calculable emission and absorption process and handed stellar astrophysics its master opacity law.

Origins · 1 minHendrik Kramers and how this was discovered

Hendrik Kramers

1894–1952 · quantum theory of X-ray opacity and free-free radiation

Hendrik Anthony Kramers was Niels Bohr's closest collaborator in the founding years of quantum theory and one of the great calculators of his generation. Born in Rotterdam, he made his name on the dispersion of light — the Kramers–Heisenberg formula and the causality-based Kramers–Kronig relations that tie the real and imaginary parts of any response function still bear his name. In 1923, in a landmark paper on the continuous X-ray spectrum, he derived the classical theory of emission and absorption by free electrons in the Coulomb fields of ions, obtaining the free-free (and bound-free) opacity that scales as $\rho T^{-7/2}$ — Kramers' opacity, the workhorse of stellar-structure theory. His treatment exposed the Coulomb-logarithm divergence that quantum mechanics would later regularise, motivating J. A. Gaunt's 1930 computation of the correction factor now called the Gaunt factor. Kramers went on to fundamental work on the interaction of matter and radiation and to renormalisation ideas that anticipated modern quantum electrodynamics, holding chairs at Utrecht and Leiden before his death in 1952.

History

1895

— Röntgen discovers X-rays, later recognised as bremsstrahlung from electrons braking in the anode

1923

— Kramers derives the classical free-free emission and absorption, giving the Coulomb-logarithm opacity

1930

— Gaunt computes the quantum correction factor that tames the classical divergence

1934

— Bethe and Heitler give the relativistic bremsstrahlung cross section

1961

— Karzas and Latter tabulate the free-free Gaunt factors still in standard use

Going further

Texts. Rybicki and Lightman, Radiative Processes in Astrophysics, Ch. 5 (Bremsstrahlung) · Carroll and Ostlie, An Introduction to Modern Astrophysics, 2nd ed., §9.5 and §10.4 (opacity, free-free) · Longair, High Energy Astrophysics, ch. on bremsstrahlung · Mo, van den Bosch and White, Galaxy Formation and Evolution, §8 (the intracluster medium) · Dodelson, Modern Cosmology (plasma processes and recombination) · Binney and Tremaine, Galactic Dynamics (context for hot-gas dynamics). Key results to keep in mind: the Larmor formula and the flat single-collision spectrum with cutoff $\omega\sim v/b$ ; the thermal-emissivity coefficient $6.8\times10^{-38}$ (CGS, per Hz) from the Maxwellian and impact-parameter integrals; the total-emissivity coefficient $1.4\times10^{-27}$ (CGS) from the frequency integral $\int e^{-h\nu/k_BT}\mathrm{d}\nu=k_BT/h$ ; the $T^{1/2}$ cooling law and the Kramers $\rho T^{-7/2}$ Rosseland-mean opacity; and the optically-thick radio slope $S_\nu\propto\nu^2$ via Kirchhoff's law and the Rayleigh–Jeans limit. The worked-example numbers ( $k_BT=8.6$ keV at $10^8$ K, $t_{cool}\approx78$ Gyr, $M_{gas}\approx8\times10^{13}\,M_\odot$ ) follow directly from these formulae. Gaunt factors are order-unity (Karzas and Latter).

Prerequisites. Specific Intensity and Radiative Flux · The Equation of Radiative Transfer · Kirchhoff's Laws and Thermal Emission · Blackbody Radiation and the Planck Law · Local Thermodynamic Equilibrium. Leads to. Opacity and Absorption Coefficients · Optical Depth · Emission and Source Function · Stefan-Boltzmann and Wien Displacement Laws.

Ask the tutor: Ask me to walk through why a $10^8$ K cluster glows in X-rays but takes longer than the age of the universe to cool, and we'll build the whole free-free story from one electron flyby.

Why is it called "free-free" emission?

The names classify radiative transitions by the electron's bound state before and after. In a bound-bound transition the electron jumps between two atomic levels, emitting a line; in bound-free (recombination) it starts bound and ends free, or vice versa, giving a continuum with an edge. In free-free emission the electron is unbound both before and after — it merely changes speed and direction in an ion's Coulomb field. Because there is no quantised level to fix the photon energy, the electron can radiate any fraction of its kinetic energy, and a whole population radiates a smooth continuum. "Bremsstrahlung" (German for "braking radiation") and "free-free emission" are the same process under two names.

Why do the electrons radiate and not the ions?

Both are accelerated in the encounter — Newton's third law guarantees equal and opposite forces — but the Larmor power depends on acceleration squared, and acceleration is force divided by mass. For the same Coulomb force, the electron's acceleration is larger than the ion's by the mass ratio $m_i/m_e$ , which is at least $\sim1836$ for hydrogen. Since power scales as acceleration squared, the electron outradiates the ion by a factor $(m_i/m_e)^2\gtrsim3\times10^6$ . So the ion's own radiation is utterly negligible; it serves only to provide the near-static Coulomb field that deflects the electron. This is why bremsstrahlung is always described as an electron process.

What physically sets the exponential cutoff at hν ≈ kT?

Energy conservation, filtered through the Maxwell–Boltzmann distribution. An electron can only emit a photon of energy $h\nu$ if it carries at least that much kinetic energy, $\tfrac12 m_ev^2\ge h\nu$ . In a thermal gas the fraction of electrons above any energy $E$ falls off as $e^{-E/k_BT}$ , so the number able to make a photon of energy $h\nu$ drops exponentially once $h\nu$ exceeds the typical thermal energy $k_BT$ . The photon spectrum therefore inherits an $e^{-h\nu/k_BT}$ tail. Below the cutoff essentially every electron can contribute, so the spectrum is flat; the transition between the two regimes is the "knee" that observers fit to measure temperature.

Is the Gaunt factor just a fudge to hide our ignorance?

No — it is a precisely defined and tabulated quantum-mechanical correction. The purely classical calculation gives a logarithmically divergent impact-parameter integral with ill-defined upper and lower limits. Quantum mechanics replaces those hand-picked limits with definite physics: the electron's de Broglie wavelength sets the minimum meaningful impact parameter, and the finite emission frequency sets the maximum. Solving the full quantum problem (the Born approximation, or exact Coulomb wavefunctions) yields a dimensionless factor $\bar g_{ff}$ , of order unity and varying slowly with frequency and temperature, that multiplies the classical rate. Karzas and Latter tabulated it in 1961, and those tables are still used. It is a rigorous number, not a fudge.

How is bremsstrahlung different from blackbody radiation?

Blackbody radiation is the equilibrium spectrum of photons in perfect thermal contact with matter; its shape is universal (Planck) and its total power scales as $T^4$ . Thermal bremsstrahlung is optically thin, non-equilibrium radiation: photons escape freely rather than thermalising, the spectrum is flat-then-exponential rather than Planckian, and the total power scales as $n_en_iT^{1/2}$ — it depends on density because it counts collisions. The two meet in a limit: if the free-free optical depth becomes large (as in HII regions at low radio frequency), the source function reaches $B_\nu(T)$ and the emergent spectrum turns into the Rayleigh–Jeans tail of a blackbody, $I_\nu\propto\nu^2$ . So optically thick bremsstrahlung is a blackbody; optically thin bremsstrahlung is the diagnostic continuum.

Why does the total cooling rate rise with temperature while the spectral emissivity falls?

This looks paradoxical but is just bookkeeping. The spectral emissivity carries a $T^{-1/2}$ prefactor, so at any fixed frequency a hotter plasma is slightly fainter per hertz. But raising the temperature also stretches the exponential cutoff to higher frequency: the spectrum stays flat over a wider band, out to $\nu\sim k_BT/h$ . Integrating the flat plateau over that widening band contributes a factor $\int e^{-h\nu/k_BT}\mathrm{d}\nu=k_BT/h\propto T$ . Multiplying the $T^{-1/2}$ height by the $T^{+1}$ width gives a net $T^{+1/2}$ for the total power. The spectral and total scalings are consistent — one is per unit frequency, the other integrated over all of them.

How do astronomers turn a bremsstrahlung spectrum into a gas mass?

In two steps, both visible in the worked examples. First, the shape gives the temperature: the exponential cutoff falls as $e^{-h\nu/k_BT}$ , so the e-folding energy of the high-energy tail equals $k_BT$ . Second, the brightness gives the density: the emissivity is $\propto n_e n_i T^{1/2}$ , so once $T$ is known the observed luminosity fixes the emission measure $\int n_en_i\,\mathrm{d}V$ and hence the electron density. Multiplying density by the emitting volume and the proton mass gives the gas mass. Because the emission goes as density squared, the densest regions dominate, so careful deprojection is needed for accurate masses — but the principle is exactly the two-parameter fit shown above.