The Astrophysical Journal Letters, 939:L21 (10pp), 2022 November 10 https://doi.org/10.

3847/2041-8213/ac982a
© 2022. The Author(s). Published by the American Astronomical Society.

Populations of Magnetized Filaments in the Intracluster Medium and the Galactic Center
F. Yusef-Zadeh1 , R. G. Arendt2,3 , and M. Wardle4
1
Department of Physics and Astronomy, CIERA, Northwestern University, 2145 Sheridan Road, Evanston , IL 60207, USA; [email protected]
2
Code 665, NASA/GSFC, 8800 Greenbelt Road, Greenbelt, MD 20771, USA; [email protected]
3
UMBC/CRESST 2
4
School of Mathematical and Physical Sciences, Research Centre for Astronomy, Astrophysics and Astrophotonics, Macquarie University, Sydney NSW 2109,
Australia; [email protected]
Received 2022 September 9; revised 2022 October 6; accepted 2022 October 6; published 2022 November 2

Abstract
Magnetized radio filaments are found in abundance in the inner few hundred parsecs of our galaxy. Progress in
understanding this population of filaments has been slow over the last few decades, in part due to a lack of
detection elsewhere in the galaxy or in external galaxies. Recent highly sensitive radio continuum observations of
radio galaxies in galaxy clusters have revealed remarkable isolated filamentary structures in the intracluster
medium (ICM) that are linked to radio jets, tails, and lobes. The origin of this class of filaments is not understood
either. Here, we argue that the underlying physical mechanisms responsible for the creation of the two populations
are the same because of their similarities in morphology, spacing between the filaments, aspect ratio, and magnetic
energy densities to the thermal pressure of the medium and that both populations have undergone synchrotron
aging. These similarities provide an opportunity to investigate the physical processes in the interstellar medium
(ISM) and ICM for the first time. We consider that the origin of the filaments in both the Galactic center and ICM is
a result of the interaction of a large-scale wind with clouds, or the filaments arise through the stretching and
collection of field lines by turbulence in a weakly magnetized medium. We examine these ideas in four radio
galaxy filaments associated with four radio galaxies—IC 40B, IC 4496, J1333–3141, ESO 137–006—and argue
that much can be understood in the future by comparing these two populations of filaments.
Unified Astronomy Thesaurus concepts: Fanaroff-Riley radio galaxies (526); Galactic center (565); Cosmic rays
(329); Galaxy clusters (584); Intracluster medium (858); Interstellar magnetic fields (845)

1. Introduction polarized synchrotron emission tracing cosmic-ray activity

throughout the inner few hundred parsecs of the galaxy
In the last few years, sensitive high-resolution radio continuum
(Yusef-Zadeh et al. 1984, 2004; Liszt 1985; Bally & Yusef-
observations of galaxy clusters have begun to reveal a population
Zadeh 1989; Gray et al. 1991; Haynes et al. 1992; Staguhn et al.
of magnetized filamentary structures in their intracluster medium
1998, 2019; Lang et al. 1999; LaRosa et al. 2001, 2004; Nord
(ICM) (Shimwell et al. 2016; Ramatsoku et al. 2020; Condon
et al. 2004; Law et al. 2008; Pound & Yusef-Zadeh 2018;
et al. 2021; Fanaroff et al. 2021; van Weeren et al. 2021;
Arendt et al. 2019). MeerKAT observations have provided a
Brienza et al. 2022; Giacintucci et al. 2022; Knowles et al. 2022;
remarkable mosaic of the inner few degrees of the Galactic
Rudnick et al. 2022). Discoveries of new structures in the ICM
center revealing hundreds of radio filaments, housed within a
surrounding radio galaxies (RGs) are in large part due to recent
∼400 pc bipolar radio bubble filled with thermal X-ray gas that
advances in broadband correlators installed on radio telescopes,
surrounds the Galactic center (Heywood et al. 2019, 2022;
providing remarkable brightness sensitivity and dynamic range
Yusef-Zadeh et al. 2022a, 2022b). H+ 3 measurements toward the
with exquisite details at ∼6″ spatial resolution. A striking aspect
Galactic center have quantified the cosmic-ray ionization rate to
of the appearance of these long and narrow isolated filaments is
be 102 to 103 times greater than across the rest of the galaxy
that they are not located within the radio lobes, the tails of radio
(Oka et al. 2005, 2019). This high ionization rate led to the
galaxies, or associated with the radio core of the galaxy (Lane
interpretation that cosmic rays drive the wind, producing the
et al. 2002; Hardcastle et al. 2019). The filaments are detected
radio bubble (Yusef-Zadeh & Wardle 2019).
outside radio galaxies but linked to radio lobes and jets with a
In spite of decades of studying the GC filaments, a good
wide range of angles to the orientation of radio jets or the lobes.
understanding of these structures is still lacking. The emerging
The nature of this new phenomenon, which appears to be
population of extragalactic radio filaments appears to have
common in high-dynamic-range images of radio galaxies, had not
remarkably similar properties to the radio filaments in the GC.
been appreciated or predicted in previous studies due to the lack
The comparison of the generic properties of both populations
of sufficient surface brightness sensitivity and angular resolution,
provides an opportunity to gain insight into the origin of both
particularly at lower frequencies.
populations. Here, side-by-side images of four examples of GC
Closer to home, in the vastly different environment of the
and ICM filaments are displayed in Section 2 to illustrate their
Galactic center (GC), radio observations have uncovered a
similar morphologies, followed by a discussion of the proper-
population of magnetized radio filaments with linearly
ties that are common in two populations of GC and ICM
filaments in Section 3. We then compare the physical
Original content from this work may be used under the terms
of the Creative Commons Attribution 4.0 licence. Any further parameters of the two populations of filaments in Section 4
distribution of this work must maintain attribution to the author(s) and the title and discuss their implications for models of their structure and
of the work, journal citation and DOI. origin in Section 5.

1
The Astrophysical Journal Letters, 939:L21 (10pp), 2022 November 10 Yusef-Zadeh, Arendt, & Wardle

Figure 1. Left: a grayscale MeerKAT image of the radio galaxy 3C 40B and its long filaments (Rudnick et al. 2022). The filaments are connected to a portion of the jet
shown at the top of the panel. Right: the three panels to the right display three groups of GC radio filaments based on a 4″ resolution MeerKAT mosaic of the GC at
1.28 GHz (Heywood et al. 2022), discussed in detail in Yusef-Zadeh et al. (2022b). The intensity I (Jy beam−1) scales are, for 3C 40B, -4 < log (I ) < -1.4 for the
bright emission, and −1 ´ 10−4 < I < 7.8 ´ 10−4 for the fainter emission; Snake, -4 < log (I + 3 ´ 10-4) < -2.33; Flamingo, -4.3 < log (I + 3 ´ 10-4) < -2.3;
and Eyebrow, -4.3 < log (I + 5 ´ 10-4) < -3.

2. GC versus ICM Filaments: Four Examples below the morphology and properties of filaments associated
with four RGs and compare them with seven GC filaments.
The GC filaments are rich in their morphological variety and
details as well as being abundant in the inner few hundred
parsecs of the Galactic center. ICM filaments are being 2.1. C40B in A194 versus G359.132–0.296 (the Snake),
revealed in high-dynamic-range images of radio galaxies and G359.808+0.130 (the Flamingo), and G359.717+0.228 (the
their surrounds. We have selected a subset of GC filaments Eyebrow)
based on their similar morphology to ICM filaments. The Figure 1 displays a comparison of the filamentary structures
comparison of a limited number of ICM filaments with a large associated with the radio jet of 3C 40B and three morpholo-
number of GC filaments offers an opportunity to advance our gically similar groups of filaments identified in the GC (Yusef-
understanding of the origin of the filaments and to probe the Zadeh et al. 2022b). 3C 40B is a radio galaxy at a distance of
magneto-ionized media of the GC and ICM. We examine ∼75.5 Mpc, embedded in the poor cluster environment Abell

2
The Astrophysical Journal Letters, 939:L21 (10pp), 2022 November 10 Yusef-Zadeh, Arendt, & Wardle

Figure 2. Left: a grayscale MeerKAT image of the core and its radio jet (top of the panel) and the filaments (center of the panel) of IC 4296, an elliptical radio galaxy,
is displayed (Condon et al. 2021). Because of the strong surface brightness of the core and the jet compared to the filaments, different color tables were applied. Right:
a MeerKAT filtered image of the comet tail as discussed in Yusef-Zadeh et al. (2022b). The intensity I (Jy beam−1) scales are, for IC 4296, -4 < log (I ) < -1.2
where I > 3.6 × 10−4 and −4 × 10−5 < I < 3.6 × 10−4 for the fainter emission, and the comet tail, -5 < log (I + 3 ´ 10-4) < -4.

194 (A194; Knowles et al. 2022), which has recently been cosmic-ray particles over such a large distance (Rudnick et al.
studied in detail (Rudnick et al. 2022). Two remarkable 2022), and yet they show similar physical characteristics.
filaments emerge from a deviated region of the jet and run The ranges of the spectral indices are α = [−0.60, −0.79]
parallel to each other as they bend together with two different for the Snake, α = [−0.80, −1.05] for the Flamingo,
curvatures leading to diffuse ends. The angular spacing α = [−0.51, −0.76] for the Eyebrow (Yusef-Zadeh et al.
between the filaments is ∼30″ (∼11 kpc), their widths range 2022b). These estimates of the spectral index are made along
between 0.85 and 2.8 kpc, and both have an extent of ∼220 kpc individual filaments. The GC spectral index values are all flatter
(Rudnick et al. 2022). The filaments show signs of physical than the spectral indices of the filaments in 3C 40B, which have
interaction with a region of the jet that is highly bent, a mean value of α ∼ − 2 and steepen from −1.3 to −2.5 with
distance from the jet.
suggesting a scenario in which a cloud is crossing the jet and
The equipartition magnetic field of GC filaments ranges
the filaments are then dragged from the interaction site
between 20 and 100 μG (for a cosmic-ray proton to electron
(Rudnick et al. 2022). ratio of p/e = 1), which is much higher than that of 3C 40B
The Snake has a length of ∼30′ (∼70 pc), is one of the with the magnetic field of 2–7 μG. The equipartition magnetic
longest in the Galactic center, and is characterized by two kinks field of the Snake has a maximum of ∼0.1 mG and decreases
and three different curvatures along its length. Another two away from the Galactic plane (Yusef-Zadeh et al. 2022a).
examples of GC filaments are the groups known as the Furthermore, the direction of the magnetic field runs along the
Flamingo and the Eyebrow. The Flamingo shows multiple 3C 40B and the Snake filaments. (The magnetic field directions
filaments with similar curvatures and the Eyebrow consists of of the Flamingo and Eyebrow are unknown).
two parallel filaments with the same curvature suggesting that
they are associated with each other.
The morphology of the four filament groups shown in 2.2. Radio Galaxy IC 4296 versus G359.992–0.572 (the
Figure 1 is similar to each other to the extent that it is difficult Comet Tail)
to distinguish which one is in a galaxy cluster based on IC 4296 is a low-luminosity elliptical galaxy in cluster
appearance alone. The systems of filaments in the GC and 3C A3566 at a distance of 50 Mpc (Condon et al. 2021), as shown
40B have similar morphologies, surface brightnesses, length- in Figure 2. A sensitive radio image of this galaxy shows a
to-width ratios, and angular spacings. What is remarkable is remarkable group of three filaments with an angular spacing of
that the ICM filaments in 3C 40B are ∼7000 times longer than ∼20″ (called threads in Condon et al. 2021) that appear to
nonthermal filaments (a.k.a. NTFs) in the GC with a coherent converge to a point (Figure 2, left), as showin in Figure 2. The
magnetic structure, steepening spectral index, and streaming projected lengths and widths of the filaments are ∼50 and

3
The Astrophysical Journal Letters, 939:L21 (10pp), 2022 November 10 Yusef-Zadeh, Arendt, & Wardle

∼2 kpc, respectively, with a surface brightness of ∼0.1 mJy per diffuse and filamentary structures are present. The filaments
∼7 6 beam (Condon et al. 2021; Figure 2, left panel). This bridging the lobes, with similar morphology to the above three
group of filaments lies to the south of the western radio jet examples, arise from a region where the jet is expanding into
where the jet shows a wiggle, thought to be due to Kelvin– the lobe.
Helmholtz (KH) instability (Condon et al. 2021). These authors G359.484+0.1122 consists of five regularly spaced fila-
suggest that relativistic electrons escape from the jet where ments with the appearance of a bent harp (Thomas et al. 2020;
there is KH instability along the radio jets (Condon et al. 2021). Yusef-Zadeh et al. 2022b). A trend was noted between the
There is also a single isolated filament with a different position length of the filaments and their brightness as well as with the
angle that arises from the eastern jet. One puzzle is that the steepening of the spectral index in the range between [−1.3,
spectral index of the filaments steepens from α ∼ −1 to −1.5 −1.0] (Yusef-Zadeh et al. 2022b).
closer to the injection point of the radio jet. The longest and straightest filament in ESO 137–006
The GC grouping of the comet tail consists of six linear (bridging the radio lobes) appears like many straight GC
structures that resemble a fragmented comet tail (Figure 2, right filaments running perpendicular to the Galactic plane (Yusef-
panel). Some of the filaments branch out into fainter Zadeh et al. 2022b). The Bent Harp is one group of GC
components. The spectral index of individual filaments varies filaments with a resemblance to the ICM structures in ESO
and shows steep spectra in the range [−1, −1.8]. There is a 137–006. Unlike the filaments of ESO 137+006, the Bent Harp
trend where the GC filaments with steeper spectral indices lie at is not connected to any obvious sources of cosmic rays.
high latitudes (Yusef-Zadeh et al. 2022a). The comparison
indicates similar morphology and steep spectral indices in both
the comet tail and IC 4296 filaments. 3. General Physical Properties
Galactic and extragalactic filaments are produced in totally
2.3. The Core of the A3562 Galaxy Cluster versus different environments, namely, the nucleus of a normal galaxy
G359.411–0.709 (the Feather) and the intracluster medium. Their length scales, energy
densities, and other characteristic parameters differ by orders
Figure 3(a) shows a radio image of the core of the A3562
of magnitude. In spite of these differences, we argue that
galaxy cluster at a distance of ∼200 Mpc (Giacintucci et al.
similar processes operate in both systems motivated by their
2022). A linearly polarized filament with a length of 50 kpc similar morphology and the similar dimensionless ratios of
runs perpendicular to the end of the tail of the radio galaxy
their physical parameters. We will examine and compare four
J1333–3141. The filament that arises from the region where the
examples of GC and RGs in the next section, following the
jet terminates is distorted before it splits into two components
common physical characteristics, as described below.
(Giacintucci et al. 2022). The splitting occurs 60 kpc away
from the radio tail. The mean spectral index of the filament 1. There is a remarkable similarity in the morphological
is α ∼ −1.5. details of the GC and ICM filaments despite their
For comparison we show in Figure 3(b), a striking GC distances differing by a factor of ∼103–104. Both classes
filament, the Feather, which splits into a two-pronged forked of filaments are narrow and long with typical aspect ratios
filament with the junction at the location of a compact source ranging between 10 and 50, assuming that the widths of
G359.416–0.706 with a possible stellar counterpart detected at the filaments are resolved. The filaments show bending
IR (Yusef-Zadeh et al. 2022a, 2022b). The compact source is and wiggles along their lengths and have similar surface
suggested to be an obstacle that splits the bright filament into brightness roughly ranging between 0.01 and 0.1 mJy
two parallel, fainter components with steeper spectral index beam−1.
values. The comparison noted in Figure 3 provides another 2. Both classes of filaments appear either as single isolated
example that radio filaments in both classes of Galactic and filaments or multiple filaments grouped together. Some of
extragalactic filaments break up into multiple components, the apparently single filaments of the GC are found to
suggestive of a flow along the filaments that splits into two consist of bundles of filaments in higher ∼1″ resolution
components as it runs into an obstacle. VLA observations. Many filaments in groupings are
approximately equally separated from each other and run
parallel to each other, giving a harp-like appearance.
2.4. ESO 137–006 versus G359.484+0.1122 (the Bent Harp)
Filaments in some groupings converge to a point or to an
Figure 4(a) shows a remarkable group of filaments linked to extended source. Groups of filaments shift sideways
the radio lobes of ESO 137–006, a luminous radio galaxy in the together, changing direction coherently, implying that
Norma galaxy cluster at 70 Mpc from us (Ramatsoku et al. they are parts of the same system of filaments with a
2020). The wide-angle tail morphology of this radio galaxy similar origin.
includes multiple straight and bent filaments extending from 3. The mean spacings between parallel filaments in a given
the eastern lobe toward the western lobe. The longest straight GC grouping peaks at ∼16″ (Yusef-Zadeh et al. 2022b).
filaments act as a bridge connecting the two radio lobes. In There is a limited number of groups of ICM filaments but
addition, a number of semiregularly spaced filaments are noted they show regular spacing between parallel filaments, as
to the south of the lobes. The length and width of the longest noted in numerous groups of GC filaments. In the case of
filament are 80 kpc and 1.2 kpc, respectively, with an aspect 3C 40B, the spacing between two parallel filaments is
ratio (length:width) of 67 (Ramatsoku et al. 2020). The mean ∼30″ (Rudnick et al. 2022). We expect the spacing-to-
spectral index of the filaments, α ; −2, is similar to the length ratio of ICM and GC filament to be similar to each
spectral indices of the lobe from which they emerge. The other if similar processes operate in both systems.
western half of the galaxy is completely surrounded by diffuse 4. Both systems of GC and ICM filaments trace synchrotron
X-ray emission. The western lobe is also highly distorted where emission from cosmic-ray electrons. They both show a

4
The Astrophysical Journal Letters, 939:L21 (10pp), 2022 November 10 Yusef-Zadeh, Arendt, & Wardle

Figure 3. Top (a): a MeerKAT image of the galaxy cluster A3562 (Giacintucci et al. 2022) convolved with a Gaussian kernel with FWHM ∼10 3. Bottom (b): a
portion of the Feather filament where the filament splits into two fainter components at a compact source (Yusef-Zadeh et al. 2022b). The intensity I (Jy beam−1)
scales are listed as follows: Feather: -4.7 < log (I + 3 ´ 10-4) < -4.3; A3562: -4 < log (I ) < -2 where I > 7× 10−5 and −2 × 10−5 < I < 7 × 10−5 for fainter
regions.

high percentage of polarization, ∼50%, revealing ordered on the assumed ratio of cosmic-ray protons to electrons
magnetic fields (B) running along the filaments. The (p/e). The magnetic field of ICM filaments is typically
orientation of the field is expected from velocity shear at estimated to be a few microgauss, 50–100 times smaller.
the interface between the ICM filaments and the ambient The low magnetic field of ICM filaments is expected
gas in the ICM (Condon et al. 2021). because the filaments have a large depth L and old if steep
5. The mean magnetic field strengths along the GC spectral index, where flux density Sν ∝ ν α, is an
filaments range between ∼100 and 400 μG depending indication of the age of cosmic-ray electrons. The weaker

5
The Astrophysical Journal Letters, 939:L21 (10pp), 2022 November 10 Yusef-Zadeh, Arendt, & Wardle

Figure 4. Top: a MeerKAT image of ESO 137–006 galaxy in the Norma galaxy cluster (Ramatsoku et al. 2020). The beam FWHM is ∼10″. 7 62 × 7 29. Bottom: a
portion of the Bent Harp filament showing multiple parallel filaments (Yusef-Zadeh et al. 2022b). The intensity I (Jy beam−1) scale for the Bent Harp
is -3.8 < log (I + 3 ´ 10-4) < -3.

magnetic field in the ICM filaments implies long cooling spectral index (Heywood et al. 2022; Yusef-Zadeh et al.
timescales, which scale as B−3/2 for observations at a 2022a).
fixed frequency. 7. In some ICM filaments, the radio lobes and tails of radio
6. The mean spectral indices of the GC filaments are steeper galaxies appear to provide an injection of cosmic-ray
than supernova remnants (SNRs; α ∼ −0.62) with a particles to illuminate radio filaments. However, the
value of α ∼ −0.8. Furthermore, a trend is noted in the sources of particle injection in the GC filaments have not
steepening of the spectral index as a function of absolute been well established. Recent MeerKAT measurements
Galactic latitude but not necessarily for typical GC indicate a number of compact radio sources located at one
filaments (LaRosa et al. 2000; Yusef-Zadeh et al. 2022a). or another end of the GC filaments. This suggests that
With a limited number of ICM filaments, the typical compact radio sources are potentially responsible for the
spectral indices appear steeper than GC filaments, which injection of cosmic-ray particles into radio filaments or
is not surprising given the vastly longer timescales. In acting as interfaces to collimate cosmic-ray-driven winds.
addition, the GC filaments show a number of filaments, 8. Diffuse thermal X-ray gas and diffuse nonthermal radio
such as the radio Arc near l ∼ 0°. 2, with a relatively flat emission in the interstellar medium (ISM) of the GC and

6
The Astrophysical Journal Letters, 939:L21 (10pp), 2022 November 10 Yusef-Zadeh, Arendt, & Wardle

Table 1
The Physical Characteristics of the Galactic Center and Radio Galaxy Filaments

Physical Parameters Galactic Center Filaments Radio Galaxy Filaments

length (pc) [4, 60] 103–5
typical width (pc) 0.5 few×103
typical aspect ratio [10, 100] [10, 70]
magnetic field strength (mG) [0.1, 0.6] few×10−3
spectral index (α) ∼−0.8 (mean) [−2, 0] [−2, −1]
surface brightness (mJy beam−1) [0.01–10] [0.01, 10]
spacing between filaments (pc) ∼0.7 mean [0.4, 1.2] [10–20]×103

ICM of RGs suggest another similarity in the physical multiple parallel filaments, splitting into a two-pronged fork,
conditions of these environments. In addition, there is and bending and spacing between the filaments. The filaments
evidence of a high cosmic-ray ionization rate from H+3 appear in parallel bundles or to converge toward a point in both
studies of the Galactic center region. populations with a spacing of a few filament widths. These
similarities are hardly surprising, given that the motivation
behind making this comparison is that structures show a
4. The Origin of Two Populations of Filaments
remarkable resemblance in the two environments, as shown in
The previous sections described the overall physical proper- Figures 1–4.
ties of a subset of ICM and GC filaments as members of two Second, the filaments’ magnetic fields are estimated to be of
populations of magnetized radio filaments. There are hundreds order 100 and 3 μG in the GC and ICM, respectively. Their
of radio filaments found in the GC at a similar distance. A corresponding magnetic energy densities, B2/8π ≈ 3 × 106 and
statistical study characterizing the mean properties of the 3 × 103 cm−3 K, are comparable with or lower than the thermal
spectral index, the equipartition magnetic field, and the spacing pressure of the medium in which they are embedded. These
between the filaments and surface brightness is based on characteristics suggest that the filaments are confined by the
MeerKAT data at 1.28 GHz (Yusef-Zadeh et al. 2022a, 2022b). thermal pressure of the external medium. This is supported in the
The physical parameters of GC and ICM filaments are listed in case of the filaments in A194 and IC 4296 (Condon et al. 2021;
columns 2 and 3, respectively, of Table 1. In this table, the Rudnick et al. 2022), where the X-ray surface brightness of the
range and mean values of the length, width, magnetic field intracluster medium is reduced along the filament, consistent
strength, surface brightness, and spacing between the filaments
with the exclusion of hot gas from the filament’s interior. In the
are given. The lengths of the GC filaments follow a power-law
case of GC filaments, high external cosmic-ray energy density
distribution (F. Yusef-Zadeh et al. 2022, in preparation). At
present, only a handful of ICM filaments are known, thus there inferred from H+ 3 measurements contributes to the confinement
are not any sufficient measurements to do statistics beyond of the GC filaments (Yusef-Zadeh & Wardle 2019).
simple means. This is because ICM filaments require high- Third, the spectral index values are steeper for ICM filaments
dynamic-range images to uncover faint magnetized filaments, than the mean spectral index of the GC filaments, as given in
and only limited observations are presently available. The GC Table 1. Both systems of filaments also show steepening of the
filaments have a wide range of surface brightnesses with many spectral index along individual filaments. The relativistic
more fainter filaments. The faint end is cut off by limited electron populations responsible for the observed synchrotron
sensitivity. We recognize that the ICM filaments detected so far emission at 1.2 GHz have steep spectral indices and show a
quite likely are the most prominent members of a larger high percentage of linear polarization based on sensitive
population, with many fainter filaments yet to be detected. observations for both the GC and ICM filament populations.
For the equipartition field strengths mentioned above, the
4.1. Physical Parameters of GC and ICM Filaments characteristic energy of the electrons dominating the emission
at 1.2 GHz is E = (4pme cn 3eB)1 2 me c 2 » 0.9 and 5 GeV,
We compare the measurements and estimates of the physical respectively. The synchrotron power radiated by an ultra-
characteristics of the two populations of filaments at the GC relativistic electron of energy E = γmec2 is E = 2sT cg 2B2 4p ,
and in the ICM of galaxy clusters, as tabulated in Table 1, to where for simplicity we have assumed a 90° pitch angle. Then,
explore the implications for the formation of the magnetic
the synchrotron loss timescale ts = E E » 0.9 or 200 Myr for
structures and the source of the ultrarelativistic electrons
the GC or IC filaments, respectively. As electron acceleration
responsible for the observed radio emission. While the filament
length scales and magnetic field strengths differ by orders of mechanisms tend to produce electrons with an E−2 energy
magnitude, it is their respective dimensionless ratios of spectrum and hence a ν−0.5 synchrotron spectrum, the
timescales and length scales that should be compared. This steepness of the observed spectra implies that the filament
allows us to examine whether the underlying physical ages are longer than the synchrotron lifetimes given above. In
mechanisms responsible for the creation of the two populations addition, the ages cannot be too much more than this because
are similar. In particular, we are interested in the mechanisms the filaments would rapidly become faint. Thus, the synchro-
responsible for creating magnetized filaments and for accel- tron lifetimes give a rough idea of the age of the filaments and
erating the relativistic electrons. impose a requirement on the cosmic-ray electron transport
First, the morphological parameters of the two populations mechanism that must be able to distribute relativistic electrons
of filaments are similar: the aspect ratios are ∼30:1, there are along the observed filament lengths on this timescale.

7
The Astrophysical Journal Letters, 939:L21 (10pp), 2022 November 10 Yusef-Zadeh, Arendt, & Wardle

4.2. Transport of Cosmic-Ray Electrons: Diffusion Versus another model, cosmic-ray electrons associated with two
Streaming parallel filaments of 3C 40B are injected from a region where
the jet is highly distorted (Rudnick et al. 2022). In particular, an
The disparate synchrotron loss timescales need to be placed
impact between a moving dense cloud and the jet is expected to
in context. For source models involving acceleration or
distort the shape of the jet as the magnetic field lines are
injection of relativistic electrons at a localized site, they impose
dragged away from the site of the interacting jet (Rudnick et al.
a requirement to be able to transport the electrons via diffusion
2022). Another remarkable object is source C in A2256 with its
or streaming from the injection site throughout the observed
extremely narrow and long tail. This unusual source is
filament lengths L ∼ 30 pc and 100 kpc for the GC and ICM
considered to be a one-sided radio jet. However, it has a very
populations, respectively.
similar morphology to GC radio filaments. Lastly, the
If transport is by diffusion, then the diffusion timescale
collimated synchrotron filaments in ESO 137–006 could be
L2/κ  ts so the diffusion coefficient κ  L2ts ∼ 3 × 1026 and
due to the interaction of the magnetic fields of the radio lobes
2 × 1031 cm2 s−1, respectively, implying mean free paths
with the magneto-ionic plasma of the intracluster medium
l = 3c/κ ∼ 0.01 and 570 pc. The required diffusion is plausible
(Ramatsoku et al. 2020).
for GC filaments with L  30 pc but seems rather large in the
case of the ICM filaments (Rudnick et al. 2022) as well as GC
filaments, such as the Snake, with lengths L  30 pc. (see 4.4. Scenarios for the Origin of the ICM and GC Filaments
Figure 1).
In GC and ICM populations, there are two broad scenarios
The alternative to diffusion is that the cosmic-ray pressure is
that could potentially explain how magnetic filaments are
sufficient to allow them to stream along the magnetic field
formed. Both scenarios posit the amplification and organization
lines, in which case scattering by self-generated waves limits
of an initially weak magnetic field in the surrounding medium.
the streaming to the Alfvén speed, vA. The requirement to
In the first scenario, the filaments are formed by an initially
populate the entire filament length then implies that L/vA  ts,
weak field that is amplified and structured by subsonic
i.e., vA  L/ts ∼ 30 and ∼600 km s−1 for the GC and ICM
turbulent motions in a weakly magnetized medium, a scenario
filaments, respectively. The density of the ionized thermal
that has been proposed both for GC (Boldyrev & Yusef-
plasma present within the filaments is unknown, but we can
Zadeh 2006) and the ICM environments (Porter et al. 2015;
usefully estimate the maximum density able to yield the
Vazza et al. 2018; Rudnick et al. 2022). The challenge in this
required Alfvén speeds, using the characteristic field magnetic
picture is that simulations show the overall orientation of most
strengths B ∼ 100 and 3 μG for the two populations. This
filaments is random (Porter et al. 2015; Vazza et al. 2018). The
implies that the thermal electron density ne  3 × 1011 and
GC filaments are mainly directed in the direction away from the
8 × 105 cm−3 for the GC and ICM filaments, respectively.
Galactic plane. It is possible that the alignment of GC filaments
These upper limits clearly exceed any plausible value in each of
is generated by cosmic-ray-driven wind running away from the
the environments by orders of magnitude, so we can conclude
Galactic plane, thus creating a velocity shear in the turbulent
another similarity in parameter space between the two
medium that aligns the filaments.
populations that streaming would be able to transport electrons
In the second scenario, the filaments are created by the
from their injection site along the entire length of the observed
subsonic flow of a weakly magnetized medium past an obstacle.
filaments.
The passing flow drapes the obstacle with field lines that are then
compressed and stretched back into a comet-like tail, creating a
magnetic filament with magnetic pressure of order the thermal
4.3. ICM and GC Filament Models
ram pressure in the medium. The obstacle could be any relatively
Models of ICM filaments benefit from the fact that there is dense ionized clump that creates a bow shock in the flow such as
an obvious reservoir of cosmic-ray particles in jets, lobes, and a stellar wind bubble, a red giant, an expanding H II region, wind
mini radio halos (Brunetti & Jones 2014) that can be injected tubes associated with red giants in AGN relativistic bubbles
into filaments. For example, the filaments of the radio galaxy (Chugai et al. 2011), or a disturbed region along jets or lobes
IC 4296 are thought to originate where helical KH instability showing size scales similar to the separation between parallel
disrupts the flow of the radio jet followed by the relativistic filaments. In this picture, the magnetized filaments are produced
particles feeding the filaments (Condon et al. 2021). One at the interaction sites of a large-scale driven wind outflowing
puzzle is why the spectrum flattens instead of steepening while away from the Galactic plane and embedded stellar wind bubbles
moving away from the presumed injection point at the end of (Rosner & Bodo 1996; Shore & LaRosa 1999; Yusef-Zadeh &
the cometary structure’s tail toward the convergence point of Wardle 2019). The large-scale radio bubble filled with X-ray gas
the filaments. One possible explanation for this is that the in the GC is considered to be arising from cosmic-ray-driven
injected particles’ pitch angle would tend to increase to outflow (Yusef-Zadeh & Wardle 2019). In this picture, the
conserve the magnetic moment p⊥/B as they spiral along the obstacle sets the length scale of the separation between the
converging field lines. This mechanism relies on the initial filaments. This is analogous to the interaction of fast-moving
pitch angle of the injected particles being anticorrelated with mass-losing stars whose winds interact with the ISM and create
energy and the pitch-angle scattering to be ineffective in cometary tails (Martin et al. 2007). In addition to the cometary
isotropizing the pitch-angle distribution. model, numerous other models have been proposed in the past to
In the case of the A3562 galaxy cluster, it has been explain the origin of the GC filaments and the sources in which
suggested that the tail of the radio galaxy is interacting with a cosmic-ray electrons are accelerated. Unlike ICM filaments, it
tangential wind blown in the direction along the filaments is not clear what feeds cosmic-ray electrons into
(Giacintucci et al. 2022). An alternative to this picture is a filaments (Nicholls & Le Strange 1995; Rosner & Bodo 1996;
scenario in which the cosmic-ray electrons injected from the Shore & LaRosa 1999; Bicknell & Li 2001; Dahlburg et al.
tail feed cluster magnetic field lines (Giacintucci et al. 2022). In 2002; Yusef-Zadeh 2003; Boldyrev & Yusef-Zadeh 2006;

8
The Astrophysical Journal Letters, 939:L21 (10pp), 2022 November 10 Yusef-Zadeh, Arendt, & Wardle

Ferrière 2009; Banda-Barragán et al. 2016; Yusef-Zadeh & In both cases, the filaments are in rough pressure equilibrium
Wardle 2019; Sofue 2020; Thomas et al. 2020; Coughlin et al. with their surroundings but are more strongly magnetized,
2021). consistent with scenarios in which they are formed by
In the context of the ICM filaments, the relative motion with dynamical processes at work in their surroundings. In one
the external medium is created by the orbital motion of the scenario, which has previously been discussed in both the GC
obstacle or a head–tail radio galaxy through the ICM around and ICM contexts, the filaments arise through the stretching
the center of the cluster. An additional contribution for slow- and collection of field lines by turbulence in a weakly
moving galaxies close to the center of the cluster could be magnetized medium. An alternative scenario that has been
moving through denser ICM (Toniazzo & Schindler 2001; proposed for the GC filaments is the collection and draping of
Douglass et al. 2008). Other sources of external pressure could field lines by a moving stellar wind source or similar obstacle
be due to “weather” motion within the ICM (Brunetti & with respect to the medium. It is not clear that this mechanism
Jones 2014). On the other hand, in the Galactic center ISM, can create the large aspect ratios of the ICM filaments because
there is a contribution due to the large-scale nuclear wind any orbital motion is mildly transonic at best and so is unlikely
generated by high cosmic-ray pressure (Yusef-Zadeh & to lead to a tight magnetized tail or lead to sufficient particle
Wardle 2019). acceleration to energize it.
One interesting aspect of this picture when applied to ICM The source of the relativistic particles still remains to be
filaments is that the filaments are linked to jets or lobes of radio established. In the GC, a few filaments have associated
galaxies. The locations where the filaments cross radio galaxies compact radio sources, and in the ICM the filaments are
are highly distorted and cosmic-ray particles are likely to associated with radio jets and lobes, but the exact nature of any
escape from the jet (Condon et al. 2021). This supports the connection in both cases remains unclear. Both filament
picture in which there is a concentration of cosmic-ray particles populations have steep synchrotron spectra indicating an aged
in the jet or the lobe that injects relativistic plasma into the ICM cosmic-ray electron population, so it is possible that any
filaments. As described in Section 4.3, the origin of filamenta- injection event happened long ago and the link to the injecting
tion is interaction with an obstacle and splitting of the filament source is no longer apparent.
into multiple filaments. Another possibility for the origin of the The striking similarities between these enigmatic popula-
filamentation is synchrotron cooling instability formed as a tions, despite their very different environments, suggest the
result of the interaction of cosmic rays generated by the winds possibility that one or the other may be amenable to
(e.g., Simon & Axford 1967). The mean spacing and the mean observational probes that shed light on the physical processes
width of the filaments are expected to be similar to each other at work in both populations. Some examples include (i) more
(Yusef-Zadeh et al. 2022a). In the case of ICM filaments, the sensitive MeerKAT observations of ICM sources to look for
spacing is estimated to be the product of the cooling timescale fainter as well as shorter (young and bright) filaments, (ii) high-
and the Alfvén speed which is of the order of kiloparsecs. resolution 1″ VLA study of ICM and GC filaments, and (iii)
Assuming that the GC and ICM filaments are formed with X-ray studies searching for depressions in X-ray surface
the same mechanism in the context of an origin involving an brightness overlapping ICM and GC filaments.
obstacle collimating the magnetized wind, the lack of compact
radio sources associated with some of the GC filaments is likely Work by R.G.A. was supported by NASA under award No.
to be related to the lifetime of the obstacle. In particular, if an 80GSFC21M0002. We are grateful to W. Cotton, S. Giacin-
obstacle is an expanding H II region or a planetary nebula, then tucci, M. Ramatsoku, and L. Rudnick for providing the FITS
it may have already dissipated as simulations of cloud–wind images of their published data. The MeerKAT telescope is
interactions indicate (Banda-Barragán et al. 2016). Another operated by the South African Radio Astronomy Observatory,
possibility is that the source of injection has a very steep which is a facility of the National Research Foundation, an
spectrum and would be visible only at very low frequencies. agency of the Department of Science and Innovation. The
The above cometary and turbulent models are technically National Radio Astronomy Observatory is a facility of the
different than a scenario in which there is a strong pre-existing National Science Foundation operated under a cooperative
organized field dominating the region, with only some field agreement by Associated Universities, Inc.
lines being lit up by an injected population of cosmic-ray
electrons. This model was one of the first models proposed for ORCID iDs
the GC, which was originally thought to be threaded by a
milligauss poloidal field (Morris 2007). However, the diffuse F. Yusef-Zadeh https://orcid.org/0000-0001-8551-9220
radio emission from the inner few hundred parsecs of the R. G. Arendt https://orcid.org/0000-0001-8403-8548
galaxy suggests a much weaker global field, ∼10–20 μG M. Wardle https://orcid.org/0000-0002-1737-0871
(Yusef-Zadeh et al. 2013). This scenario is also inapplicable to
the ICM environment in which the field strengths are much less References
than several microgauss inferred for the magnetized filaments.
Arendt, R. G., Staguhn, J., Dwek, E., et al. 2019, ApJ, 885, 71
Bally, J., & Yusef-Zadeh, F. 1989, ApJ, 336, 173
5. Summary Banda-Barragán, W. E., Parkin, E. R., Federrath, C., Crocker, R. M., &
Bicknell, G. V. 2016, MNRAS, 455, 1309
We compared the two populations of magnetized filaments Bicknell, G. V., & Li, J. 2001, ApJL, 548, L69
in the GC and ICM. They differ vastly in their physical Boldyrev, S., & Yusef-Zadeh, F. 2006, ApJL, 637, L101
properties, such as lengths, widths, and magnetic field Brienza, M., Lovisari, L., Rajpurohit, K., et al. 2022, A&A, 661, A92
Brunetti, G., & Jones, T. W. 2014, IJMPD, 23, 1430007
strengths. However, we argued that, nevertheless, they are Chugai, N. N., Churazov, E. M., & Sunyaev, R. A. 2011, MNRAS, 414, 879
analogous to each other as might be anticipated based on their Condon, J. J., Cotton, W. D., White, S. V., et al. 2021, ApJ, 917, 18
similar morphologies. Coughlin, E. R, Nixon, C. J., & Ginsburg, A. 2021, MNRAS, 501, 1868

9
The Astrophysical Journal Letters, 939:L21 (10pp), 2022 November 10 Yusef-Zadeh, Arendt, & Wardle

Dahlburg, R. B., Einaudi, G., LaRosa, T. N., & Shore, S. N. 2002, ApJ, Oka, T., Geballe, T. R., Goto, M., et al. 2019, ApJ, 883, 54
568, 220 Oka, T., Geballe, T. R., Goto, M., Usuda, T., & McCall, B. J. 2005, ApJ,
Douglass, E. M., Blanton, E. L., Clarke, T. E., Sarazin, C. L., & Wise, M. 632, 882
2008, ApJ, 673, 763 Porter, D. H., Jones, T. W., & Ryu, D. 2015, ApJ, 810, 93
Fanaroff, B., Lal, D. V., Venturi, T., et al. 2021, MNRAS, 505, 6003 Pound, M. W., & Yusef-Zadeh, F. 2018, MNRAS, 473, 2899
Ferrière, K. 2009, A&A, 505, 1183 Ramatsoku, M., Murgia, M., Vacca, V., et al. 2020, A&A, 636, L1
Giacintucci, S., Venturi, T., Markevitch, M., et al. 2022, ApJ, 934, 49 Rosner, R., & Bodo, G. 1996, ApJL, 470, L49
Gray, A. D., Cram, L. E., Ekers, R. D., & Goss, W. M. 1991, Natur, 353, 237 Rudnick, L., Brüggen, M., Brunetti, G., et al. 2022, ApJ, 935, 168
Hardcastle, M. J., Croston, J. H., Shimwell, T. W., et al. 2019, MNRAS, Shimwell, T. W., Luckin, J., Bruggen, M., et al. 2016, MNRAS, 459, 277
488, 3416 Shore, S. N., & LaRosa, T. N. 1999, ApJ, 521, 587
Haynes, R. F., Stewart, R. T., Gray, A. D., et al. 1992, A&A, 264, 500 Simon, M., & Axford, W. I. 1967, ApJ, 150, 105
Heywood, I., Camilo, F., Cotton, W. D., et al. 2019, Natur, 573, 235 Sofue, Y. 2020, PASJ, 72, L4
Heywood, I., Rammala, I., Camilo, F., et al. 2022, ApJ, 925, 165 Staguhn, J., Arendt, R. G., Dwek, E., et al. 2019, ApJ, 885, 72
Knowles, K., Cotton, W. D., Rudnick, L., et al. 2022, A&A, 657, A56 Staguhn, J., Stutzki, J., Uchida, K. I., & Yusef-Zadeh, F. 1998, A&A, 336,
Lane, W. M., Kassim, N. E., Ensslin, T. A., Harris, D. E., & Perley, R. A. 290
2002, AJ, 123, 2985 Thomas, T., Pfrommer, C., & Enßlin, T. 2020, ApJL, 890, L18
Lang, C. C., Morris, M., & Echevarria, L. 1999, ApJ, 526, 727 Toniazzo, T., & Schindler, S. 2001, MNRAS, 325, 509
LaRosa, T. N., Kassim, N. E., Lazio, T. J. W., & Hyman, S. D. 2000, AJ, van Weeren, R. J., Shimwell, T. W., Botteon, A., et al. 2021, A&A, 651, A115
119, 207 Vazza, F., Brunetti, G., Bruggen, M., & Bonafede, A. 2018, MNRAS,
LaRosa, T. N., Lazio, T. J. W., & Kassim, N. E. 2001, ApJ, 563, 163 474, 1672
LaRosa, T. N., Nord, M. E., Joseph, T., Lazio, W., & Kassim, N. E. 2004, ApJ, Yusef-Zadeh, F. 2003, ApJ, 598, 325
607, 302 Yusef-Zadeh, F., Arendt, R. G., Wardle, M., et al. 2022a, ApJL, 925, L18
Law, C. J., Yusef-Zadeh, F., & Cotton, W. D. 2008, ApJS, 177, 515 Yusef-Zadeh, F., Arendt, R. G., Wardle, M., et al. 2022b, MNRAS,
Liszt, H. S. 1985, ApJL, 293, L65 515, 3059
Martin, D. C., Seibert, M., Neill, J. D., et al. 2007, Natur, 488, 780 Yusef-Zadeh, F., Hewitt, J. W., & Cotton, W. 2004, ApJS, 155, 421
Morris, M. 2007, arXiv:astro-ph/0701050 Yusef-Zadeh, F., Hewitt, J. W., Wardle, M., et al. 2013, ApJ, 762, 33
Nicholls, J., & Le Strange, E. T. 1995, ApJ, 443, 638 Yusef-Zadeh, F., Morris, M., & Chance, D. 1984, Natur, 310, 557
Nord, M. E., Lazio, T. J. W., Kassim, N. E., et al. 2004, AJ, 128, 1646 Yusef-Zadeh, F., & Wardle, M. 2019, MNRAS, 490, L1

10