---
title: Free-Free Absorption Overview
url: https://www.emergentmind.com/topics/free-free-absorption
type: topic
---

# Free-Free Absorption Overview

Free-free absorption (FFA), also called inverse bremsstrahlung or free-carrier absorption in some semiconductor contexts, is the absorption of electromagnetic radiation by unbound electrons interacting with ions or with the electronic structure of a material. In an ionized plasma, a free electron absorbs photon energy while being accelerated in the Coulomb field of an ion; the photon is not redirected as in ordinary scattering, but its energy is transferred to the plasma. In radio astronomy, FFA is primarily a propagation and radiative-transfer effect that attenuates synchrotron emission in ionized gas. In semiconductors and heterostructures, the related term free-carrier absorption (FCA) denotes intraband absorption by carriers already present in conduction- or valence-band states. The common physical feature is that no new bound electron–hole pair is created by the absorbing photon. The observable consequences range from low-frequency spectral turnovers in galaxies, supernova remnants, pulsars, active galactic nuclei, and colliding-wind binaries to THz and infrared optical losses, phonon-assisted gain, and opacity effects in warm dense matter.

## 1. Physical mechanism and fundamental scalings

In a fully ionized plasma, free electrons and ions interact through Coulomb forces. A photon can be absorbed during an electron–ion encounter, with the collision supplying the momentum required for energy and momentum conservation. The inverse process is thermal bremsstrahlung emission. The absorption coefficient depends on the electron and ion densities, electron temperature, photon frequency, ionic charge, and the free–free Gaunt factor.

In the radio limit, a commonly used approximation is

$$
\alpha_{\nu}^{\rm ff}\propto n_{\rm e}n_{\rm i}T_{\rm e}^{-3/2}\nu^{-2}g_{\rm ff},
$$

where $n_{\rm e}$ and $n_{\rm i}$ are the electron and ion densities, $T_{\rm e}$ is the electron temperature, and $g_{\rm ff}$ is the Gaunt factor. Over radio frequencies, the frequency dependence of the Gaunt factor modifies the effective scaling to approximately

$$
\alpha_{\nu}^{\rm ff}\propto T_{\rm e}^{-1.35}\nu^{-2.1}.
$$

The optical depth along a path is

$$
\tau_\nu=\int \alpha_\nu\,ds.
$$

For a uniform absorber with emission measure

$$
{\rm EM}=\int n_{\rm e}^2\,ds,
$$

or more generally ${\rm EM}=\int n_{\rm e}n_{\rm i}\,ds$, the optical depth is proportional to the emission measure. Dense substructure is therefore strongly weighted because the density enters quadratically.

For a foreground screen, the transmitted background flux is

$$
S_{\nu}=S_{\nu,0}e^{-\tau_\nu}.
$$

If the intrinsic synchrotron spectrum is $S_{\nu,0}\propto\nu^{-\alpha}$, the rapid increase of $\tau_\nu$ toward low frequency can transform a declining power law into a flattened, bent, or inverted spectrum before producing strong suppression. The turnover frequency $\nu_t$ is conventionally defined by $\tau_{\nu_t}=1$; higher emission measures shift $\nu_t$ to higher frequency.

Free-free emission and absorption are related by thermal radiative transfer but are not identical observational signatures. The absorbing plasma emits thermally while attenuating background radiation. In a uniform slab, the thermal contribution contains a factor $1-e^{-\tau_\nu}$, approaching the source-function value for $\tau_\nu\gg1$. This coupling between emission and absorption is important when inferring gas properties from integrated spectra.

FFA must be distinguished from synchrotron self-absorption (SSA). SSA occurs within the relativistic synchrotron-emitting plasma and generally produces a source spectrum with a low-frequency power law, approximately $S_\nu\propto\nu^{2.5}$ for a standard homogeneous synchrotron source. A foreground FFA screen instead produces an exponential attenuation, approximately $\exp[-(\nu_0/\nu)^2]$ or $\exp[-(\nu_0/\nu)^{2.1}]$. Spatially resolved optical-depth morphology provides an additional diagnostic: SSA opacity is expected to correlate with synchrotron brightness, whereas FFA opacity can trace an unrelated foreground medium [1311.5762].

## 2. Radio-galaxy, starburst, and supernova-remnant applications

FFA is common in dense star-forming environments. Low-frequency spectra of luminous and ultraluminous infrared galaxies are rarely simple extrapolations of their higher-frequency synchrotron spectra. GMRT observations at 244 and 610 MHz of 20 luminous and ultraluminous IRAS galaxies, combined with archival spectra extending to approximately 23 GHz, revealed low-frequency bends, flattening, and turnovers. The spectra are interpreted as synchrotron emission viewed through ionized starburst gas [1002.3334].

A single absorbed component can describe some systems. Examples include NGC 1614, with a fitted turnover near $0.24$ GHz, UGC 2369 near $0.50$ GHz, UGC 5101 near $1.1$ GHz, Arp 152 near $0.21$ GHz, and IRAS 15250+3609 near $2.2$ GHz. Other galaxies require two absorbed synchrotron components with distinct turnover frequencies. Representative pairs are approximately $2.0$ and $0.32$ GHz for Arp 220, $4.2$ and $0.37$ GHz for Arp 299, and $6.2$ and $0.34$ GHz for IRAS 03359+1523. Two components occur in 13 of 23 objects in the broader set with data below 1.4 GHz.

The inferred emission measures span approximately

$$
{\rm EM}\simeq 0.07-140\times10^6\ {\rm cm^{-6}\,pc},
$$

using $T_{\rm e}=10^4$ K. Arp 299, for example, has estimated emission measures of approximately $62\times10^6$ and $0.38\times10^6\ {\rm cm^{-6}\,pc}$ for its two components. These values indicate that synchrotron-emitting supernovae occupy regions with substantially different ionized-gas columns, densities, or path lengths. The components may correspond to distinct nuclei or starburst complexes, but spectral decomposition does not establish a one-to-one correspondence with individual nuclei.

The relative strengths of absorbed and less-absorbed components can be compared with a model for the supernova rate. Assuming a Salpeter initial mass function and an exponentially declining star-formation history, effective ionized-gas survival times of tens of megayears are obtained in several systems, including approximately 16 Myr for Arp 220, 39 Myr for Arp 299, 23 Myr for UGC 5101, and 33 Myr for IC 1623. These values exceed plausible lifetimes of individual compact H II regions by roughly an order of magnitude. The inferred times therefore describe the persistence of absorbing ionized gas along lines of sight, rather than the literal lifetime of one H II region. Foreground screens, shells, overlapping regions, and propagating star formation can produce such effective timescales. A substantial strongly absorbed component is interpreted as evidence that the star-formation rate is increasing or that star formation is propagating through dense gas.

FFA also shapes the integrated spectra of individual remnants. In Cassiopeia A, low-frequency emission is modeled with both internal absorption by unshocked ejecta and external absorption by ionized interstellar or circumstellar gas. The 2019 GURT spectrum from 16 to 72 MHz peaks near 22 MHz. The fitted internal absorber has approximately ${\rm EM}=37.36\ {\rm pc\,cm^{-6}}$, $T_{\rm int}\simeq100$ K, and an average ionic charge $Z_{\rm int}\simeq2.55$; the external absorber has ${\rm EM}\simeq0.17\ {\rm pc\,cm^{-6}}$ and $T_{\rm ISM}\simeq20$ K [2212.06104].

The internal absorbing material is associated primarily with the rear part of the radio shell, while the external absorber attenuates both front- and back-side emission. With an ejecta-sheet thickness of approximately $0.16$ pc, the inferred electron density is about $15.3\ {\rm cm^{-3}}$ for a homogeneous medium and $18.7\ {\rm cm^{-3}}$ for a clumping factor of $0.67$. The corresponding unshocked-ejecta mass is approximately $2.61\,M_\odot$ without clumping and $0.96\,M_\odot$ for the illustrative clumped case. These masses depend on the assumed geometry, composition, path length, and clumping factor.

In supernova circumstellar media, FFA can also heat the absorbing gas. For a steady wind with $\rho\propto R^{-2}$, the absorbed synchrotron radiation deposits energy into electrons immediately ahead of the shock. A self-consistent treatment gives characteristic temperatures of order $10^5$ K when the optical depth is appreciable. For SN 1993J, the observed optical-depth evolution is approximately compatible with FFA heating, and the inferred progenitor mass-loss rate is approximately

$$
\dot M\simeq(0.8-1.0)\times10^{-5}\,M_\odot\,{\rm yr^{-1}}
$$

for a wind velocity of $10\ {\rm km\,s^{-1}}$ [1404.5800].

## 3. Geometry, inhomogeneity, and resolved diagnostics

The observed spectral form depends strongly on whether the absorber is external to, mixed with, or spatially structured relative to the emitting region. A homogeneous foreground screen gives exponential attenuation. An internal absorber, in which synchrotron emissivity and free-free opacity are co-spatial, can produce a new power-law regime rather than an exponential disappearance.

For an internally absorbing region,

$$
I_\nu=\frac{J_\nu}{4\pi Q_\nu}\left(1-e^{-\tau_\nu}\right),
$$

where $J_\nu$ is the synchrotron emissivity and $Q_\nu$ is an effective absorption coefficient. If $J_\nu\propto\nu^{-\alpha_{\rm NT}}$ and $Q_\nu\propto\nu^{-\beta}$, then in the optically thick limit

$$
I_\nu\propto\nu^{-\alpha_{\rm NT}+\beta}.
$$

This produces gradual spectral-index changes rather than the abrupt low-frequency cutoff expected from a foreground screen.

WR 147 provides an example of this distinction in a particle-accelerating colliding-wind binary. Its unresolved radio emission combines thermal wind emission with synchrotron radiation from a colliding-wind region. A foreground FFA model describes the spectrum above roughly 1 GHz but fails when the 610 MHz measurement is included because it predicts excessive exponential suppression. An internal-FFA model reproduces a nearly flat low-frequency non-thermal residual, with a preferred effective $\beta\simeq0.89$ and an internal turnover near a few GHz. A second turnover between approximately 150 and 610 MHz is attributed primarily to weaker foreground absorption, although a Razin–Tsytovich contribution is not completely excluded [2508.10506].

The distinction between internal and foreground absorption also applies to spatially resolved AGN jets. In J1324+4048, optical-depth maps constructed from multi-frequency VLBI data show nearly uniform opacity across the radio lobes and little correlation with high-frequency synchrotron intensity. The FFA fits have reduced $\chi^2$ values of approximately 1.5–2.3, compared with 3.5–5.7 for homogeneous SSA fits. The inferred emission measure is approximately $1.2\times10^7\ {\rm cm^{-6}\,pc}$ for an electron temperature near $10^4$ K [1311.5762].

J0029+3457 shows structured opacity that also fails to correlate with high-frequency intensity. This favors a spatially varying foreground absorber over SSA, but its gradual turnover makes the fits relatively insensitive to model parameters. The evidence for FFA is therefore likely but not definitive in that source.

Spatial resolution is equally important in galactic applications. A cell-based model of Sgr A and M82 shows that localized FFA can be prominent in high-resolution maps but diluted in integrated or low-resolution observations. In M82, an approximately 100-pc-scale depression near a central radio source can disappear at approximately $10''$ resolution at 408 MHz. The global turnover requires thermal-emitting and absorbing gas distributed over a substantial fraction of the disk; a single compact, extremely dense region is insufficient [1305.6963].

Filling factor introduces a major degeneracy. For clumpy gas,

$$
{\rm EM}\simeq N_{\rm e}^2 f_\nu l,
$$

where $f_\nu$ is the fraction of the path or volume occupied by ionized gas. The same nominal density and geometrical path length can yield different optical depths if $f_\nu$ changes. For Sgr A, a thermal filling factor near $0.1$ is required in the model. Consequently, absorption measurements constrain emission measure more directly than density, mass, or path length separately.

## 4. FFA as a probe of Galactic and AGN environments

An optically thick Galactic H II region can be used as a foreground screen that removes synchrotron emission behind it. Comparing the absorbed sightline with the corresponding unabsorbed brightness isolates the synchrotron contribution from a limited line-of-sight segment. The measured emission remains a convolution of cosmic-ray electron density, magnetic-field strength, and path length; it does not independently determine either the cosmic-ray density or the magnetic field.

At 74 MHz, an updated catalog of 135 background H II-region measurements shows an approximately flat average synchrotron emissivity of $0.5$–$0.7\ {\rm K\,pc^{-1}}$ for long background paths, but an increase to approximately $1.2\ {\rm K\,pc^{-1}}$ near a 12-kpc background path length. Simulations using the Hammurabi line-of-sight code, three Galactic magnetic-field models, and GALPROP cosmic-ray distributions reproduce the broad trend at long path lengths but predict a downturn at short paths. The observed rise is not reproduced by the J13b, JF12b, or Sun10b magnetic-field models, nor by the z10LMPDE and 71Xvarh7S cosmic-ray configurations [2002.09204].

Possible explanations include incorrect far-distance assignments, an absorption-selection bias toward bright background regions, enhanced synchrotron emissivity in the far outer Galaxy, or diminished emissivity in the inner Galaxy. The catalog alone cannot distinguish cosmic-ray density from magnetic-field strength. Toy models indicate that the observed trend could require enhanced emissivity beyond approximately 16 kpc from the Sun in the surveyed longitude range, or a reduced-emissivity region extending over approximately the inner 8 kpc.

In AGN, radiation-pressure confinement (RPC) provides a model in which the absorbing gas density is determined primarily by the incident ionizing flux,

$$
F_{\rm ion}\sim\frac{L_{\rm bol}}{4\pi r^2}.
$$

Near the hydrogen ionization front, the gas temperature is approximately $10^4$ K and the RPC density scales approximately as $n\propto L_{\rm bol}r^{-2}T^{-1}$. The resulting free-free turnover frequency is approximately

$$
\nu_{\rm thick}\propto L_{\rm bol}^{1/2}r^{-1}
$$

at fixed physical radius, with modifications from dust, ionization structure, and temperature gradients. Cloudy RPC calculations predict turnover frequencies increasing from approximately 100 MHz on kiloparsec scales to approximately 100 GHz or higher on sub-parsec scales [2109.03313].

Dusty and dustless gas have different ionized columns. Dust absorbs much of the incident ionizing continuum and produces a shorter ionized layer, whereas dustless gas has a larger compression length and a higher emission measure. The fitted RPC relations are approximately

$$
\nu_{\rm thick}=6.03\times10^{11}
\left(\frac{r}{r_{\rm dust}}\right)^{-0.95}\ {\rm Hz}
$$

for dusty gas and

$$
\nu_{\rm thick}=5.01\times10^{12}
\left(\frac{r}{r_{\rm dust}}\right)^{-1.42}\ {\rm Hz}
$$

for dustless gas over the ranges considered.

Spatially resolved absorbers in NGC 1275, NGC 4151, NGC 4261, and NGC 1052 are broadly compatible with the RPC predictions. In NGC 1052, the observed turnover frequency increases from approximately 3 to 20 GHz as the projected distance decreases from 1.7 to 0.4 pc, consistent with increasing density and emission measure toward the AGN center. In NGC 1068, a compact component with brightness temperature approximately $4\times10^6$ K is too bright to be ordinary optically thin free-free emission and is instead interpreted as synchrotron radiation viewed through an FFA screen.

## 5. Free-carrier absorption in semiconductors and heterostructures

In semiconductor physics, free-carrier absorption denotes absorption by electrons or holes already present in conduction- or valence-band states. It is an intraband process and normally does not generate a new electron–hole pair. The term overlaps conceptually with free-free absorption but involves band states, crystal momentum, scattering, and collective carrier response.

A first-principles treatment of doped silicon decomposes FCA into four contributions: direct single-particle absorption between bands within the same carrier manifold, phonon-assisted single-particle absorption, charged-impurity-assisted absorption, and the collective resistive or Drude contribution. The first three are explicit quantum-mechanical transitions; the last is a collective intraband response determined by the conductivity and momentum-relaxation time [2205.02768].

Direct transitions conserve crystal momentum and are described by velocity matrix elements between occupied and unoccupied Bloch states. Phonon-assisted processes include two time orderings: photon absorption followed by carrier scattering, and carrier scattering followed by photon absorption. Charged impurities supply momentum through screened Coulomb scattering. The resistive term is obtained from the linearized Boltzmann equation in a relaxation-time approximation.

For n-type silicon at $n=10^{19}\ {\rm cm^{-3}}$, phonon-assisted absorption dominates at wavelengths below approximately $1\,\mu{\rm m}$, direct conduction-band transitions dominate between approximately 1 and $4\,\mu{\rm m}$, and the resistive term becomes increasingly important beyond approximately $4\,\mu{\rm m}$. A pronounced feature near $2\,\mu{\rm m}$ arises from transitions between the conduction-band minimum and the second-lowest conduction band. At long wavelength, the resistive contribution grows approximately as $\alpha\propto\lambda^2$. Charged-impurity absorption rises toward long wavelength but remains at least one order of magnitude below the dominant contributions in the analyzed range.

For p-type silicon, the spectrum lacks the sharp $2\,\mu{\rm m}$ feature because the valence-band degeneracy produces continuously varying energy separations away from the Brillouin-zone center. Direct, phonon-assisted, and resistive contributions remain comparable from approximately 1 to $10\,\mu{\rm m}$, with the resistive term dominating only beyond roughly $10\,\mu{\rm m}$.

The calculated FCA agrees well with experimental data over approximately 1–10 $\mu{\rm m}$ and broad carrier-density ranges. At high doping, FCA can become comparable to or larger than cross-gap absorption near 1000–1200 nm. In conventional solar cells it is usually a minor loss because highly doped emitters are thin, but it becomes important in infrared and thermal devices. Under ideal 1000 K black-body radiation for a $10\,\mu{\rm m}$ film, the predicted electron–hole-pair current reduction is approximately 10% at $10^{18}\ {\rm cm^{-3}}$ and 50% at $10^{19}\ {\rm cm^{-3}}$.

In quantum cascade lasers, the classical bulk Drude estimate can be inappropriate because the optical field is polarized along the growth direction while carriers are freely mobile primarily in the plane. Direct intra-subband absorption is therefore suppressed by both the vanishing diagonal $p_z$ matrix element and in-plane momentum conservation. FCA becomes a second-order, scattering-assisted process involving virtual coupling to another subband. Interface defects, ionized impurities, and phonons provide the required momentum [1112.1822].

For a representative THz QCL, interface-defect-induced intra-subband FCA is approximately $10^{-2}\ {\rm cm^{-1}}$ far from resonance, while oblique intersubband absorption is about an order of magnitude larger. At room temperature, LO-phonon interactions dominate the calculated FCA. The absorption is strongly structured by subband spacings, virtual-state denominators, scattering form factors, and the electron and lattice temperatures rather than following a universal $\omega^{-2}$ or $\omega^{-3}$ law.

A related theoretical treatment shows that intersubband absorption continuously evolves into Drude-like bulk FCA as the number of wells increases and barrier widths decrease. In finite multiple-quantum-well structures, discrete intersubband transitions form a quasi-continuum as the number of wells grows. Neighboring transitions move toward low frequency and become the intra-miniband response, while higher-energy transitions form inter-miniband absorption. The frequency-weighted oscillator strength obeys the Thomas–Reiche–Kuhn $f$-sum rule, so adding an independent phenomenological FCA term to a complete intersubband calculation risks double counting [1109.4504].

## 6. Nonequilibrium, warm-dense, and transient manifestations

FCA need not always be a net loss in a nonequilibrium system. In THz QCLs, LO-phonon absorption-assisted transitions can produce negative FCA when the electronic temperature exceeds the lattice temperature. The direct process absorbs a photon and an LO phonon, whereas the reverse process emits a photon and an LO phonon. The latter contains the bosonic factor $N_q+1$, compared with $N_q$ for phonon absorption, and can dominate under suitable population conditions [1611.09269].

The approximate condition for negative LO-phonon-assisted FCA is

$$
\hbar\omega<
\hbar\omega_{\rm LO}
\left(\frac{T_{\rm e}}{T_{\rm L}}-1\right).
$$

The effect requires suitable upper-state population, energy alignment, and population inversion; a hotter electron system alone does not guarantee gain. Designed structures produce an effective FCA gain of approximately 2–3 ${\rm cm^{-1}}$, or a few ${\rm cm^{-1}}$ at approximately 200 K. In the reported device, the calculated contribution was not experimentally separable from ordinary direct gain.

In warm dense matter, collisional free-free absorption is calculated by combining the quantum-mechanical Sommerfeld electron–ion bremsstrahlung cross-section with detailed balance and Fermi–Dirac occupation factors. The treatment includes Pauli blocking, stimulated emission, plasma propagation above the plasma frequency, and the full quantum Coulomb correction through a hypergeometric-function kernel. The effective collision frequency is not characterized by a universal Coulomb logarithm [1907.09782].

The formalism applies for photon frequencies above the plasma frequency and for nonrelativistic photon energies and electron temperatures. In the classical regime it recovers the Spitzer-like behavior, while in the quasi-classical regime it approaches the Kramers limit. In strongly degenerate matter, only electrons within approximately one photon energy of the Fermi surface can absorb because deeper states have occupied final states. Depending on the regime, the collision frequency scales approximately as $\omega^{-3/2}$ or $\omega^{-1}$, and the free-free opacity can scale as

$$
\kappa\propto\omega^{-7/2}.
$$

The density dependence is nonlinear because the Fermi temperature scales as $T_F\propto n^{2/3}$ and the plasma frequency as $\omega_p\propto n^{1/2}$. These effects make Spitzer and Kramers expressions useful asymptotic limits but potentially inaccurate in transitional warm-dense matter.

FFA can also produce time-dependent spectral structure in relativistic outflows. In a model for fast-radio-burst sources, radiation passes through a hot, expanding relativistic shell whose free-free optical depth decreases as the shell expands. In the hot limit, the comoving absorption coefficient scales approximately as

$$
\alpha_{\nu'}^{\prime\,{\rm ff}}\propto n'^2T'^{-1/2}\nu'^{-2}.
$$

The transparency time therefore scales as

$$
t_{\rm obs}\propto\nu_{\rm obs}^{-1/2}.
$$

Higher frequencies escape first, followed by progressively lower frequencies, producing a negative frequency drift without requiring intrinsic emission-frequency evolution. For representative internal shocks with $\Gamma\sim100$, the model gives a normalized drift rate of approximately $10^{-2}\ {\rm ms^{-1}}$ at characteristic radii of $10^{10}$–$10^{11}$ cm. The required isotropic kinetic energy is generally $E\gtrsim10^{44}$ erg, while the relativistic-shell mass must remain below a few $\times10^{-10}\,M_\odot$ [2107.12989].

The interpretation is conditional on a hot, ionized, low-mass shell with the required energy, geometry, and timing. It competes with magnetospheric radius-to-frequency mapping and intrinsic shock-frequency evolution. The proposed drift is therefore an opacity-selected escape effect rather than a universal explanation of fast-radio-burst substructure.

FFA and FCA are consequently best regarded as families of radiative-transfer and carrier-dissipation phenomena whose detailed behavior is controlled by geometry, density structure, temperature, scattering, quantum statistics, and the relationship between the emitting and absorbing populations. In astrophysical plasmas, the decisive observables are low-frequency spectral curvature, optical-depth maps, emission measures, thermal contamination, and time evolution. In condensed matter, the relevant diagnostics are polarization, subband or band structure, scattering-assisted matrix elements, phonon populations, conductivity, and sum-rule consistency. Across these settings, the central constraint is that the absorber must contain mobile charges capable of accepting photon energy through free–free or intraband interactions.

Source: https://www.emergentmind.com/topics/free-free-absorption