Foreground Free-Free Absorption in Radio Astronomy

• Foreground free-free absorption is defined as the attenuation of radio waves by ionized gas acting as an effective foreground screen, producing characteristic low-frequency spectral turnovers.
• Modeling techniques use parameters like electron temperature, emission measure, and frequency to quantify the optical depth and separate thermal from synchrotron components.
• Observations of sources such as Sgr A, M82, and AGN illustrate how f-FFA influences spectral morphology and spatial emission profiles in diverse astrophysical environments.

Searching arXiv for the cited literature on foreground free-free absorption and closely related modeling papers. {"query":"Foreground free-free absorption arXiv (Jurusik et al., 2013, Clemens et al., 2010, Wajima et al., 2020, 1404.5800, Kundu et al., 2021, Marr et al., 2013, Baskin et al., 2021)", "max_results": 10} Foreground free-free absorption (f-FFA) is the attenuation of radio radiation by ionized gas located in front of a radiating source, or treated observationally as an effective foreground screen. In radio astronomy it is the absorptive inverse of thermal bremsstrahlung, and its optical depth increases strongly toward low frequency and with emission measure. The process is used to explain low-frequency turnovers, spectral flattening, inversions, and spatial depressions in synchrotron-dominated sources, and it has been modeled in environments ranging from the Sgr A Complex and M82 to luminous starbursts, compact symmetric objects, 3C 84, supernovae, pulsars, colliding-wind binaries, AGN circumnuclear gas, FRB shocks, and Galactic foregrounds for 21-cm cosmology (Jurusik et al., 2013, Clemens et al., 2010, Wajima et al., 2020).

1. Physical basis and transfer formalism

In the standard radio-astronomy treatment, free-free absorption is parameterized by an optical depth that depends on electron temperature, observing frequency, and emission measure. A commonly used form is

$$\tau_{\nu,ij} = 0.0824 \left(\frac{T}{K}\right)^{-1.35} \left(\frac{\nu}{GHz}\right)^{-2.1} EM,$$

with

$EM = \int_0^l N_e^2\, f_{\nu}\, dl.$

Here $EM$ is the line-of-sight emission measure, and $f_\nu$ is a filling factor that accounts for the fact that the ionized medium need not fill the entire volume. The same dependence is written in turnover-frequency form as $\tau_\nu = (\nu/\nu_t)^{-2.1}$, where $\tau_{\nu_t}=1$ defines the turnover frequency (Jurusik et al., 2013, Clemens et al., 2010).

The foreground-screen interpretation is the simplest radiative-transfer limit. In that geometry, synchrotron radiation emitted behind the plasma is attenuated as $S_\nu^{\rm obs}=S_\nu^{\rm int}e^{-\tau_\nu}$, while the absorbing layer also emits thermal free-free radiation. Jurusik et al. write the cell-based form

$$S_{ij} = \left[ S_{sync,ij}\,\nu^{-\alpha}\, \exp(-\tau_{\nu,ij}) + S_{th,ij}\,\left(1-\exp(-\tau_{\nu,ij})\right) \right]\Omega_{ij},$$

which is the slab solution

$$I_\nu^{\rm obs} = I_\nu^{\rm sync} e^{-\tau_\nu} + B_\nu^{\rm ff}(1-e^{-\tau_\nu})$$

applied on a spatial grid (Jurusik et al., 2013). Across AGN, supernova, and relativistic-shock applications, alternate normalizations are used, but they retain the same core scaling $\tau_{\rm ff}\propto EM\,T^{-1.5}\,\nu^{-2}$ or $EM = \int_0^l N_e^2\, f_{\nu}\, dl.$0 (Wajima et al., 2020, 1404.5800, Kundu et al., 2021).

Foreground and internal absorption are distinct geometries. In f-FFA, the ionized gas is external to the synchrotron source and acts as a line-of-sight screen. In internal or mixed absorption, the synchrotron-emitting and absorbing plasma are co-spatial, and the low-frequency spectrum need not show an exponential cutoff. Several later applications explicitly contrast these two cases, especially in colliding-wind binaries and starbursts (Clemens et al., 2010, Tasseroul et al., 14 Aug 2025).

2. Spectral and morphological signatures

The most common spectral signature of f-FFA is a low-frequency turnover or flattening relative to a high-frequency synchrotron power law. Jurusik et al. emphasize that “the radio spectra of galaxies can be flattened towards low frequencies,” and their Sgr A model produces a turnover near $EM = \int_0^l N_e^2\, f_{\nu}\, dl.$1 GHz. In M82, the global spectrum also turns over at low frequency, and at $EM = \int_0^l N_e^2\, f_{\nu}\, dl.$2 MHz the starburst core shows a depression in radio emission on a scale of about $EM = \int_0^l N_e^2\, f_{\nu}\, dl.$3 pc caused by free-free absorption in a giant H II region (Jurusik et al., 2013).

In luminous IRAS galaxies, the observed spectra below $EM = \int_0^l N_e^2\, f_{\nu}\, dl.$4 GHz are “rarely a simple power-law extrapolation” of the higher-frequency synchrotron emission. Clemens et al. report turnovers, bends, strong flattening, and in some systems two absorbed components with different turnover frequencies, including Arp 220 and Arp 299. They explicitly argue that synchrotron self-absorption is disfavored for most of the sample because it would require extremely high brightness temperatures, whereas the observed rapid spectral changes and inversions are naturally produced by free-free absorption with $EM = \int_0^l N_e^2\, f_{\nu}\, dl.$5 (Clemens et al., 2010).

Spatial morphology is often as diagnostic as the integrated spectrum. In compact symmetric objects, optical-depth maps can distinguish foreground absorption from self-absorption. For J1324+4048, the optical-depth maps are “strikingly smooth,” which is suggestive of a foreground screen of absorbing gas. The spectra at the intensity peaks fit a simple free-free absorption model with reduced $EM = \int_0^l N_e^2\, f_{\nu}\, dl.$6, better than a simple synchrotron self-absorption model with reduced $EM = \int_0^l N_e^2\, f_{\nu}\, dl.$7. In J0029+3457, the optical-depth maps show structure, but the morphology does not correlate with that in the intensity maps, so free-free absorption is likely but not definitive (Marr et al., 2013).

Side-dependent absorption is another hallmark of foreground geometry. In 3C 84, the northern counter-jet component N1 remains inverted between $EM = \int_0^l N_e^2\, f_{\nu}\, dl.$8 and $EM = \int_0^l N_e^2\, f_{\nu}\, dl.$9 GHz with $EM$0, while the approaching lobe has $EM$1. The counter-jet is much fainter than the approaching side, and the asymmetry decreases with increasing frequency, which is consistent with a foreground free-free screen whose optical depth declines toward high frequency (Wajima et al., 2020).

For pulsars, the signature can be written in normalized form as an exponential turnover with $EM$2 dependence. In the sample of $EM$3 pulsars assembled by Abadi et al., the turnover frequencies cluster around $EM$4 MHz, indicating thermal free-free absorption along the line of sight rather than intrinsic curvature alone (Abadi et al., 20 Nov 2025).

3. Modeling strategies and parameter inference

A central modeling issue is the decomposition of thermal and non-thermal emission in the presence of frequency-dependent absorption. Jurusik et al. model Sgr A and M82 on an $EM$5 grid, assigning synchrotron and thermal components cell by cell and then applying free-free absorption. For Sgr A, the modeled turnover near $EM$6 GHz requires emission measures $EM$7 and a filling factor $EM$8 of about $EM$9 in the thermal component. For M82, they conclude that more than $f_\nu$0 of the galaxy must be significantly contaminated by thermal gas, and that a single small emitting region with extremely high density is insufficient to reproduce the global turnover (Jurusik et al., 2013).

Clemens et al. extend this logic to integrated starburst spectra by fitting either a single absorbed synchrotron-plus-thermal component or two absorbed components with different turnover frequencies. Across their luminous IRAS sample, the implied emission measures span $f_\nu$1 to $f_\nu$2. Their derived “H II region lifetimes” of $f_\nu$3–$f_\nu$4 Myr are much longer than plausible lifetimes for individual H II regions, and they therefore interpret them as ionized-gas survival times along the line of sight, consistent with younger H II regions obscuring older star-forming regions in a shell-like, foreground-screen geometry (Clemens et al., 2010).

Spatially resolved inference can be performed directly from optical-depth maps. In the VLBI analysis of compact symmetric objects, high-frequency maps are extrapolated to lower frequencies using the local high-frequency spectral index, and the optical depth is then recovered from $f_\nu$5. In 3C 84, a simpler asymmetry method is used: $f_\nu$6 which yields $f_\nu$7 and $f_\nu$8. The inferred frequency exponent $f_\nu$9, much flatter than $\tau_\nu = (\nu/\nu_t)^{-2.1}$0, indicates a non-uniform absorber rather than a uniform slab (Marr et al., 2013, Wajima et al., 2020).

Other environments use additional observables to break parameter degeneracies. In pulsars, the combination of $\tau_\nu = (\nu/\nu_t)^{-2.1}$1 from the spectral turnover and $\tau_\nu = (\nu/\nu_t)^{-2.1}$2 from pulse dispersion allows the absorber size $\tau_\nu = (\nu/\nu_t)^{-2.1}$3 and density $\tau_\nu = (\nu/\nu_t)^{-2.1}$4 to be estimated, revealing compact, dense structures. In SN 1993J, the time dependence of $\tau_\nu = (\nu/\nu_t)^{-2.1}$5 is used to test whether free-free heating sets the circumstellar temperature and to infer a progenitor mass-loss rate of $\tau_\nu = (\nu/\nu_t)^{-2.1}$6 for a wind velocity of $\tau_\nu = (\nu/\nu_t)^{-2.1}$7 (Abadi et al., 20 Nov 2025, 1404.5800).

4. Galaxies and star-forming systems

In galaxies and star-forming complexes, f-FFA is usually associated with H II regions, ionized halos, and dense starburst gas that lie in front of synchrotron-emitting supernova remnants or diffuse cosmic-ray plasma. The Sgr A Complex and M82 remain canonical examples, because both show that the observed spectrum depends not only on density and path length but also on filling factor and angular resolution (Jurusik et al., 2013).

System or class	Foreground ionized medium	Reported outcome
Sgr A Complex	Sgr A West and its ionized halo	Turnover near $\tau_\nu = (\nu/\nu_t)^{-2.1}$8 GHz; $\tau_\nu = (\nu/\nu_t)^{-2.1}$9; $\tau_{\nu_t}=1$0
M82	Giant H II region and widespread thermal gas	Central depression at $\tau_{\nu_t}=1$1 MHz; $\tau_{\nu_t}=1$2 thermally contaminated
Luminous IRAS galaxies	Starburst ionized gas with multiple $\tau_{\nu_t}=1$3 components	Turnovers, bends, and two-component absorbed spectra

For Sgr A, the low-frequency depression occurs where Sgr A West is bright at higher frequencies, so the thermal gas is interpreted as lying in front of non-thermal emission from Sgr A East and related background components. For M82, the depression around the central supernova remnant at $\tau_{\nu_t}=1$4 MHz becomes progressively less visible at poorer resolution and disappears completely at $\tau_{\nu_t}=1$5, showing that unresolved absorption can be washed out by beam averaging (Jurusik et al., 2013).

The luminous and ultra-luminous IRAS galaxies observed at $\tau_{\nu_t}=1$6 and $\tau_{\nu_t}=1$7 MHz generalize this picture to merger-driven starbursts. Arp 220, Arp 299, and IRAS 03359+1523 require two absorbed components with widely separated turnover frequencies, implying regions with very different emission measures along different sightlines. Clemens et al. argue that the strongly absorbed component traces supernovae still embedded in dense ionized gas, whereas the less absorbed component traces older supernovae outside those regions. Their preferred interpretation is that galaxies with a significant strongly absorbed radio component are systems in which the star formation rate is still increasing with time (Clemens et al., 2010).

5. AGN and circumnuclear media

In AGN, f-FFA is most often attributed to circumnuclear disks, tori, narrow-line-region gas, or radiation-pressure-compressed photoionized layers. The 3C 84 counter-jet provides a pc-scale example: from the counter-jet/approaching-lobe asymmetry, the average density of the foreground absorber is constrained to $\tau_{\nu_t}=1$8, consistent with dense ionized gas in a circumnuclear disk and/or an assembly of clumpy clouds within the central $\tau_{\nu_t}=1$9 pc (Wajima et al., 2020).

VLBI optical-depth mapping provides complementary evidence in compact symmetric objects. J1324+4048 is a strong case for foreground free-free absorption because the optical-depth morphology is smooth and the FFA fit outperforms SSA at the lobe peaks. J0029+3457 is more ambiguous, but even there the lack of correlation between optical-depth and intensity structure favors an external absorber over a purely intrinsic synchrotron turnover. One component in J0029+3457 remains inverted even at the highest frequencies, and it is interpreted as the core; the corresponding magnetic field upper limit is of order $S_\nu^{\rm obs}=S_\nu^{\rm int}e^{-\tau_\nu}$0 Gauss at a radius of order $S_\nu^{\rm obs}=S_\nu^{\rm int}e^{-\tau_\nu}$1 pc (Marr et al., 2013).

A broader radial framework is supplied by radiation-pressure-compression models of AGN photoionized gas. In that treatment, the free-free absorption frequency increases from $S_\nu^{\rm obs}=S_\nu^{\rm int}e^{-\tau_\nu}$2 MHz on the kpc scale to $S_\nu^{\rm obs}=S_\nu^{\rm int}e^{-\tau_\nu}$3 GHz on the sub-pc scale, and the free-free emission at $S_\nu^{\rm obs}=S_\nu^{\rm int}e^{-\tau_\nu}$4 GHz yields a radio loudness $S_\nu^{\rm obs}=S_\nu^{\rm int}e^{-\tau_\nu}$5, below the typical $S_\nu^{\rm obs}=S_\nu^{\rm int}e^{-\tau_\nu}$6 of radio-quiet AGN. The same models predict that the flat free-free continuum may become dominant above $S_\nu^{\rm obs}=S_\nu^{\rm int}e^{-\tau_\nu}$7 GHz. In NGC 1068, optically thin free-free emission on the sub-pc scale is excluded because the brightness temperature is too high, whereas excess ALMA emission above $S_\nu^{\rm obs}=S_\nu^{\rm int}e^{-\tau_\nu}$8 GHz is consistent with predicted free-free emission from gas just outside the broad-line region (Baskin et al., 2021).

6. Stellar, binary, pulsar, and transient environments

In core-collapse supernovae, f-FFA is produced by ionized circumstellar material outside the synchrotron-emitting shock. Björnsson and Lundqvist analyze the regime in which the same absorption also heats the circumstellar medium, which makes the temperature a self-consistent output rather than an externally imposed parameter. For SN 1993J they argue that the observed free-free absorption is consistent with heating from free-free absorption, and they derive a progenitor mass-loss rate approximately $S_\nu^{\rm obs}=S_\nu^{\rm int}e^{-\tau_\nu}$9 for a wind velocity of $$S_{ij} = \left[ S_{sync,ij}\,\nu^{-\alpha}\, \exp(-\tau_{\nu,ij}) + S_{th,ij}\,\left(1-\exp(-\tau_{\nu,ij})\right) \right]\Omega_{ij},$$0 (1404.5800).

Pulsars offer a different diagnostic because the absorber can be constrained by both continuum turnover and pulse dispersion. In the sample of $$S_{ij} = \left[ S_{sync,ij}\,\nu^{-\alpha}\, \exp(-\tau_{\nu,ij}) + S_{th,ij}\,\left(1-\exp(-\tau_{\nu,ij})\right) \right]\Omega_{ij},$$1 pulsars with $$S_{ij} = \left[ S_{sync,ij}\,\nu^{-\alpha}\, \exp(-\tau_{\nu,ij}) + S_{th,ij}\,\left(1-\exp(-\tau_{\nu,ij})\right) \right]\Omega_{ij},$$2 exponential turnovers, the characteristic turnover frequencies cluster around $$S_{ij} = \left[ S_{sync,ij}\,\nu^{-\alpha}\, \exp(-\tau_{\nu,ij}) + S_{th,ij}\,\left(1-\exp(-\tau_{\nu,ij})\right) \right]\Omega_{ij},$$3 MHz, implying $$S_{ij} = \left[ S_{sync,ij}\,\nu^{-\alpha}\, \exp(-\tau_{\nu,ij}) + S_{th,ij}\,\left(1-\exp(-\tau_{\nu,ij})\right) \right]\Omega_{ij},$$4 for an absorbing medium with $$S_{ij} = \left[ S_{sync,ij}\,\nu^{-\alpha}\, \exp(-\tau_{\nu,ij}) + S_{th,ij}\,\left(1-\exp(-\tau_{\nu,ij})\right) \right]\Omega_{ij},$$5 K. Combining these $$S_{ij} = \left[ S_{sync,ij}\,\nu^{-\alpha}\, \exp(-\tau_{\nu,ij}) + S_{th,ij}\,\left(1-\exp(-\tau_{\nu,ij})\right) \right]\Omega_{ij},$$6 values with dispersion measures reveals a discrete near-in population of absorbers with $$S_{ij} = \left[ S_{sync,ij}\,\nu^{-\alpha}\, \exp(-\tau_{\nu,ij}) + S_{th,ij}\,\left(1-\exp(-\tau_{\nu,ij})\right) \right]\Omega_{ij},$$7 and $$S_{ij} = \left[ S_{sync,ij}\,\nu^{-\alpha}\, \exp(-\tau_{\nu,ij}) + S_{th,ij}\,\left(1-\exp(-\tau_{\nu,ij})\right) \right]\Omega_{ij},$$8, together with a clear size-density anticorrelation reminiscent of Galactic and extragalactic H II regions (Abadi et al., 20 Nov 2025).

Foreground geometry can also fail, which is itself diagnostic. In the colliding-wind binary WR 147, a classical f-FFA model does not reproduce the low-frequency spectral energy distribution, whereas an internal-FFA model accounts for the spectral-index change down to $$S_{ij} = \left[ S_{sync,ij}\,\nu^{-\alpha}\, \exp(-\tau_{\nu,ij}) + S_{th,ij}\,\left(1-\exp(-\tau_{\nu,ij})\right) \right]\Omega_{ij},$$9 MHz without the exponential drop predicted by foreground absorption. At the same time, the upper limit at $$I_\nu^{\rm obs} = I_\nu^{\rm sync} e^{-\tau_\nu} + B_\nu^{\rm ff}(1-e^{-\tau_\nu})$$0 MHz suggests that two turnovers occur in the radio spectrum, so both internal and foreground free-free absorption may be present in different frequency ranges (Tasseroul et al., 14 Aug 2025).

A more extreme transient application appears in FRB models with hot relativistic shocks. There the shocked shell can function as a foreground absorber for radio waves produced within or interior to the shock. Kundu and Zhang show that free-free absorption in such a shell can produce a negative frequency drift, with $$I_\nu^{\rm obs} = I_\nu^{\rm sync} e^{-\tau_\nu} + B_\nu^{\rm ff}(1-e^{-\tau_\nu})$$1; for an internal shock with Lorentz factor $$I_\nu^{\rm obs} = I_\nu^{\rm sync} e^{-\tau_\nu} + B_\nu^{\rm ff}(1-e^{-\tau_\nu})$$2, the normalized drift rate is $$I_\nu^{\rm obs} = I_\nu^{\rm sync} e^{-\tau_\nu} + B_\nu^{\rm ff}(1-e^{-\tau_\nu})$$3 per ms, and the implied shell radius is $$I_\nu^{\rm obs} = I_\nu^{\rm sync} e^{-\tau_\nu} + B_\nu^{\rm ff}(1-e^{-\tau_\nu})$$4–$$I_\nu^{\rm obs} = I_\nu^{\rm sync} e^{-\tau_\nu} + B_\nu^{\rm ff}(1-e^{-\tau_\nu})$$5 cm (Kundu et al., 2021).

7. Survey foregrounds, interpretive limits, and broader significance

At survey scale, the same free-free physics becomes a foreground problem in low-frequency cosmology. Two SKA/EoR studies model Galactic free-free emission from H$$I_\nu^{\rm obs} = I_\nu^{\rm sync} e^{-\tau_\nu} + B_\nu^{\rm ff}(1-e^{-\tau_\nu})$$6-derived emission measures and emphasize that the same $$I_\nu^{\rm obs} = I_\nu^{\rm sync} e^{-\tau_\nu} + B_\nu^{\rm ff}(1-e^{-\tau_\nu})$$7 formalism governs absorption. In Monte Carlo foreground simulations, Galactic free-free emission is $$I_\nu^{\rm obs} = I_\nu^{\rm sync} e^{-\tau_\nu} + B_\nu^{\rm ff}(1-e^{-\tau_\nu})$$8–$$I_\nu^{\rm obs} = I_\nu^{\rm sync} e^{-\tau_\nu} + B_\nu^{\rm ff}(1-e^{-\tau_\nu})$$9, $\tau_{\rm ff}\propto EM\,T^{-1.5}\,\nu^{-2}$0–$\tau_{\rm ff}\propto EM\,T^{-1.5}\,\nu^{-2}$1, and $\tau_{\rm ff}\propto EM\,T^{-1.5}\,\nu^{-2}$2–$\tau_{\rm ff}\propto EM\,T^{-1.5}\,\nu^{-2}$3 times more luminous than the EoR signal on $\tau_{\rm ff}\propto EM\,T^{-1.5}\,\nu^{-2}$4 in the $\tau_{\rm ff}\propto EM\,T^{-1.5}\,\nu^{-2}$5–$\tau_{\rm ff}\propto EM\,T^{-1.5}\,\nu^{-2}$6, $\tau_{\rm ff}\propto EM\,T^{-1.5}\,\nu^{-2}$7–$\tau_{\rm ff}\propto EM\,T^{-1.5}\,\nu^{-2}$8, and $\tau_{\rm ff}\propto EM\,T^{-1.5}\,\nu^{-2}$9–$EM = \int_0^l N_e^2\, f_{\nu}\, dl.$00 MHz bands, respectively, and the leaked power inside the EoR window can still reach $EM = \int_0^l N_e^2\, f_{\nu}\, dl.$01–$EM = \int_0^l N_e^2\, f_{\nu}\, dl.$02, $EM = \int_0^l N_e^2\, f_{\nu}\, dl.$03–$EM = \int_0^l N_e^2\, f_{\nu}\, dl.$04, and $EM = \int_0^l N_e^2\, f_{\nu}\, dl.$05–$EM = \int_0^l N_e^2\, f_{\nu}\, dl.$06 on $EM = \int_0^l N_e^2\, f_{\nu}\, dl.$07 (Lian et al., 2020). A related H$EM = \int_0^l N_e^2\, f_{\nu}\, dl.$08-based study finds $EM = \int_0^l N_e^2\, f_{\nu}\, dl.$09–$EM = \int_0^l N_e^2\, f_{\nu}\, dl.$10, $EM = \int_0^l N_e^2\, f_{\nu}\, dl.$11–$EM = \int_0^l N_e^2\, f_{\nu}\, dl.$12, and $EM = \int_0^l N_e^2\, f_{\nu}\, dl.$13–$EM = \int_0^l N_e^2\, f_{\nu}\, dl.$14 excess in the same three bands, with leaked power ratios reaching about $EM = \int_0^l N_e^2\, f_{\nu}\, dl.$15, $EM = \int_0^l N_e^2\, f_{\nu}\, dl.$16, and $EM = \int_0^l N_e^2\, f_{\nu}\, dl.$17 on scales of $EM = \int_0^l N_e^2\, f_{\nu}\, dl.$18 (Lian et al., 2020).

Several interpretive cautions recur across the literature. First, a turnover does not by itself establish foreground free-free absorption; synchrotron self-absorption, Razin suppression, ionization losses, and mixed-medium radiative transfer can mimic curvature in unresolved data. Second, finite angular resolution can erase the spatial depressions that most clearly reveal f-FFA, as in M82. Third, parameters such as “H II region lifetimes” inferred from multicomponent fits need not be literal cloud lifetimes; in Clemens et al. they are interpreted as ionized-gas survival times along the line of sight (Jurusik et al., 2013, Clemens et al., 2010).

These limitations do not diminish the diagnostic value of f-FFA. Rather, they define the conditions under which it is most informative: broad frequency coverage, separation of thermal and non-thermal components, adequate angular resolution, and a geometry that is constrained by independent observables such as asymmetry, dispersion measure, or time evolution. Under those conditions, foreground free-free absorption becomes a quantitative probe of emission measure, clumpiness, covering fraction, circumnuclear structure, and source-environment coupling across a wide range of radio astrophysics (Wajima et al., 2020, 1404.5800, Baskin et al., 2021).