Draft version February 24, 2025

Typeset using LATEX twocolumn style in AASTeX631

SOFIA/HAWC+ Far-Infrared Polarimetric Large Area CMZ Exploration Survey. V. The Magnetic
Field Strength and Morphology in the Sagittarius C Complex
Roy J. Zhao ,1, 2, 3 Mark R. Morris ,3 David T. Chuss ,4 Dylan M. Paré ,4 Jordan A. Guerra ,5
Natalie O. Butterfield ,6 Edward J. Wollack ,7 and Kaitlyn Karpovich 8, 9, 4
1 Kavli Institute for Cosmological Physics, The University of Chicago, 5640 S Ellis Ave., Chicago, IL 60637, USA
arXiv:2502.14961v1 [astro-ph.GA] 20 Feb 2025

2 Department of Physics, The University of Chicago, 5720 S Ellis Ave., Chicago, IL 60637, USA
3 Department of Physics & Astronomy, University of California, Los Angeles, 475 Portola Pl., Los Angeles, CA 90095, USA
4 Department of Physics, Villanova University, 800 E. Lancaster Ave., Villanova, PA 19085, USA
5 Cooperative Institute for Research in Environmental Sciences, University of Colorado, Boulder, CO 80309 USA
6 National Radio Astronomy Observatory, 520 Edgemont Rd., Charlottesville, VA 22903, USA
7 NASA Goddard Space Flight Center, Mail Code: 665, Greenbelt, MD 20771, USA
8 Kavli Institute for Particle Astrophysics & Cosmology, P.O. Box 2450, Stanford University, Stanford, CA 94305, USA
9 Department of Physics, Stanford University, Stanford, CA 94305, USA

ABSTRACT
We present an analysis of the magnetic field strength and morphology in the Sagittarius C complex
(Sgr C; G359.43-0.09) in the Milky Way Galaxy’s Central Molecular Zone (CMZ) using the 214 µm po-
larimetry data acquired with the High-Angular-Resolution Wideband Camera+ (HAWC+) instrument
aboard the Stratospheric Observatory for Infrared Astronomy (SOFIA). We use several hundred mag-
netic field pseudovectors in the Sgr C region to trace the projected magnetic field orientation within
cold molecular gas clouds, and as is the trend throughout the CMZ, they show a higher polariza-
tion fraction toward the periphery of the clouds. We conduct a modified Davis-Chandrasekhar-Fermi
(DCF) analysis of individual clouds and find that the sky-plane magnetic field strength varies from
highly turbulent regions having inferred strengths of ∼ 30 µG to regions of relatively uniform field
orientation having strengths of ∼ 1 mG. The magnetic field orientations suggest that outflows from
active star-forming regions, such as the extended green object (EGO) G359.43-0.10 and the protostel-
lar source FIR-4 (G359.43+0.02), cause high turbulence in their vicinity. The magnetic field direction
is found to be tangential to the surface of the Sgr C H II region, as well as two [C II] emission cavities
around this region. Several other features in the vicinity of Sgr C, especially numerous non-thermal
filaments (NTFs) and a diffuse source of X-ray emission towards the southwest of the H II region, are
discussed with regard to the observed magnetic field pseudovectors.

Keywords: (565) Galactic Center — (847) Interstellar medium — (786) Infrared astronomy — (994)
Magnetic Field — (1278) Dust Polarimetry

1. INTRODUCTION Interstellar Medium (ISM) properties, a strong mag-

The central half kiloparsec of the Milky Way disk, netic field has also been found to permeate the CMZ
commonly referred to as the Central Molecular Zone (Tsuboi et al. 1985; LaRosa et al. 2005; Morris 2006;
(CMZ), is an ideal laboratory to study astrophysics in Ferrière 2009; Morris 2015; Akshaya & Hoang 2024; Paré
extreme environments (Morris & Serabyn 1996; Ferrière et al. 2024; Tress et al. 2024), especially in dense clouds
et al. 2007; Bryant & Krabbe 2021; Henshaw et al. 2023). (Yusef-Zadeh et al. 1999; Pillai et al. 2015; Hsieh et al.
Apart from the extensively studied star formation and 2018; Butterfield et al. 2024a; Lu et al. 2024). The rel-
atively strong magnetic field (0.1 – 1 mG) is said to
counteract the gravitational collapse of clouds and con-
Corresponding author: Roy J. Zhao tributes to the low star-formation rate in this region
[email protected]
2

(Morris 1993; Chuss et al. 2003; Kruijssen et al. 2014; electrons diffuse along the magnetic field lines in the
Sormani et al. 2020; Morris 2023). inter-cloud ISM (Serabyn & Morris 1994; Morris 1996;
In this paper, we use the results of the Far-Infrared Po- Chuss et al. 2003). In addition, many other hypotheses
larimetric Large Area CMZ Exploration survey (FIRE- have been offered to account for the NTFs (references
PLACE; PI: D. Chuss), a high-resolution dust polarime- in Morris 1996; Boldyrev & Yusef-Zadeh 2006; Yusef-
try survey conducted at 214 µm, to study the magnetic Zadeh & Königl 2004; Banda-Barragán et al. 2016, 2018;
field in the CMZ. As a common experimental probe of Yusef-Zadeh & Wardle 2019). For example, an alterna-
the sky-plane component of the magnetic field, dust po- tive hypothesis for generating relativistic electrons that
larimetry exploits the property that the spin axes of does not directly involve magnetic fields invokes diffu-
dust grains tend to be aligned with the magnetic field sive shock acceleration where stellar winds from massive
(Andersson et al. 2015). The measurement of polarized young stars impact relatively dense interstellar material
dust emission reveals the alignment of dust grains within (Rosner & Bodo 1996).
these clouds, thereby indicating the sky-plane magnetic Previous studies of the GC magnetic field, including
field component integrated along the line of sight (Chuss those using the Submillimeter Polarimeter for Antarc-
et al. 2003; Andersson et al. 2015). tic Remote Observations (SPARO; Novak et al. 2003),
There are several possible mechanisms that could be Planck (Planck Collaboration et al. 2015), the Polar-
responsible for the field orientations deduced from the ized Instrument for the Long-wavelength Observation of
FIREPLACE measurements. As the magnetic field lines the Tenuous ISM (PILOT) (Mangilli et al. 2019), and
are anchored inside the molecular clouds, the field lines the Atacama Cosmology Telescope (ACT; Guan et al.
could be sheared by the clouds’ internal motions (Parker 2021), have focused on the large-scale magnetic field,
1979). In a differentially orbiting medium like the CMZ, with typical resolutions ≳ 1′ . These works reported a
the shear stretches the magnetic field into a primarily largely uniform magnetic field that permeates the CMZ,
toroidal geometry (Werner et al. 1988; Wardle & Konigl and is oriented at a ∼ 20◦ angle with respect to the
1990). The same effect can also be indicative of the di- Galactic plane. However, conclusions made from these
rection of the internal shear within clouds (Aitken et al. surveys remain on a global level, since the spatial res-
1991; Morris et al. 1992; Aitken et al. 1998; Novak et al. olution of these surveys does not generally resolve the
2000; Chuss et al. 2003; Morris 2015). Other candidates, internal structure of individual molecular clouds. High-
such as cloud compression, which occurs at the interface resolution surveys, such as those reported by Chuss et al.
of expanding H II regions or supernova blast waves, tend (2003) and Lu et al. (2024) found a correlation between
to shape the field to be parallel to the compressive in- the magnetic field orientation and molecular cloud den-
terface. sity, though they do not have enough coverage to cover
Indeed, a high-resolution magnetic field study, like the a sufficiently large portion of the CMZ to draw global
FIREPLACE survey, can benefit our understanding of conclusions. Along the same line, Bally et al. (2024)
many aspects of the CMZ, such as the magnetic field studied the magnetic field inside the prominent H II re-
morphology within molecular clouds, and within por- gion using the JWST-NIRCam images. With the FIRE-
tions of low-density ISM occupied by the NTFs (Yusef- PLACE survey dataset, a recent study by Paré et al.
Zadeh et al. 1984; Morris & Yusef-Zadeh 1985; Heywood (2025) (or FIREPLACE IV) demonstrated the advan-
et al. 2022; Yusef-Zadeh et al. 2022a). These NTFs are tage of studying the magnetic field orientation in the
manifested as spatially extended synchrotron emission CMZ using datasets with both high resolution and large
from relativistic electrons constrained to diffuse along coverage. In this paper, we focus on studying the mag-
a magnetic flux tube. The NTFs within the CMZ are netic field within one of its sub-regions, the Sgr C com-
generally oriented perpendicular to the Galactic plane, plex.
which implies a large-scale vertical magnetic field per- Sgr C was initially reported as one of the most promi-
vading the Galactic Centre (GC), thereby pointing to nent H II regions in the GC (Downes & Maxwell 1966),
the existence of an organized inter-cloud magnetic field and it was later detailed by Liszt (1985) and Liszt &
(Morris & Serabyn 1996; Heywood et al. 2022). Further- Spiker (1995), who noted the presence of a prominent
more, Serabyn & Morris (1994) argued that turbulent NTF apparently connected to the H II region. Fig-
instabilities at the H2 /H II interface could mix two or- ure 1 portrays this region at multiple wavelengths over-
thogonal magnetic field systems inside and outside that laid with a line integral contour image (LIC; Cabral &
interface, which facilitates magnetic field line reconnec- Leedom 1993) to represent the magnetic field direction
tion, which in turn can boost free electrons in the ISM made by measurements reported by Paré et al. (2024)
to relativistic velocities, thereby forming NTFs as those (or FIREPLACE III). To highlight important struc-
 3

FIR-4

Sgr C NTF
NTF Part A

NTF Part B
Source C

100 pc ring

Sgr C HII Region

G359.44-0.10 Faint NTF

Head-tail EGO
Structure

Figure 1. A tri-colour view of the Sgr C complex with the

Figure 2. Schematic diagram of the main features within
Spitzer/IRAC 8 µm image in cyan (Stolovy et al. 2006), the
the Sgr C complex on the sky-plane, including FIR-4 (blue
Herschel/SPIRE 250 µm image in magenta (Molinari et al.
ellipse), the major molecular clouds detected in far-IR (pur-
2010), and the MeerKAT/L-Band 20 cm image in yellow
ple clouds), the Sgr C H II region and NTF detected in radio
(Heywood et al. 2022), overlaid with a LIC rendition of the
continuum (yellow circle and long, vertical rectangle, respec-
magnetic field measurements reported by Paré et al. (2024).
tively), the G359.44-0.10 EGO (Kendrew et al. 2013; green
The main structures, including the Sgr C H II region, the
star), and the radio and mid-IR structure source C (yellow
Sgr C NTF, FIR-4, and the 100 pc ring, are highlighted in
cloud). The velocities of the molecular clouds depicted in
Figure 2.
magenta are obtained using the molecular line observation
of the 3 mm Mopra survey reported by Jones et al. (2012),
tures, Figure 2 shows a simplified schematic diagram detailed in Section 4.2.
of Sgr C.
The large-scale astrogeography of Sgr C is also of par- tion about the large-scale turbulence of dense clouds, a
ticular interest and, may be crucial in reaching a full un- thorough study of the magnetic field morphology within
derstanding of the global structure of the CMZ. Many the Sgr C complex may provide improved characteriza-
studies found that the Sgr C complex resides on the tion of a region subjected to large-scale shocks and how
projected western vertex of a dust ring associated with star formation enters the picture in such a region.
the X2 orbit due to the Galactic bar potential (Molinari In this paper, we examine the magnetic field struc-
et al. 2011; Krumholz & Kruijssen 2015; Sormani et al. ture of the Sgr C complex using the FIREPLACE sur-
2018; Salas et al. 2020). Through the lens of simulations, vey. The paper is organized as follows. In Section 2,
Sormani et al. (2024) and Tress et al. (2024) alterna- we briefly describe the survey data used. In Section 3,
tively described the formation of the ring as a result of we present analyses of the Sgr C magnetic field. In Sec-
the particular gas dynamics in the Galactic bar, where tion 4, we use multi-wavelength observations to investi-
shocks and magnetic field are also considered in these gate numerous intriguing structures in the Sgr C region
computations. Therefore, this theory places Sgr C and where the magnetic field has apparently become man-
its positive-longitude counterpart, Sgr B2, at the two ifest. In Section 5, we present a detailed examination
interaction sites of X1 -X2 orbits, located at both pro- of several features in which the magnetic field is likely
jected ends of the nuclear stellar disk. Meanwhile, the playing an important role. We finally summarize our
Mopra survey (Jones et al. 2012) revealed a significant conclusions in Section 6. Throughout this paper, we
amount of shock tracers (SiO and HNCO) within the adopt 8.2 kpc as the distance to the GC (GRAVITY
Sgr C complex. The widely distributed shock tracers are Collaboration et al. 2019).
presumably induced by the X1 -X2 orbit collision. The
exact role of this collision is yet unclear since it can ei- 2. OBSERVATIONS AND DATA
ther compress the clouds and provoke star formation, or The 214 µm intensity and polarization results pre-
increase turbulence and thereby inhibit star formation. sented in this paper utilize the second data release
As magnetic field measurements could reveal informa- (DR2) of the FIREPLACE survey described in Paré
4

et al. (2024). The data were acquired using HAWC+ 2024a). However, in our field of view, our conservative
(Harper et al. 2018), an imaging polarimeter aboard selection criteria have largely mitigated such issues.
the SOFIA telescope. The complete survey, including Throughout this paper, we compare our 214 µm mag-
the first data release (DR1; Butterfield et al. 2024a, or netic field orientations to the 850 µm JCMT magnetic
FIREPLACE I) covers the inner 1.5 degrees of the CMZ field orientations reported by Lu et al. (2024), depicted
with a spatial resolution of 19.6′′ . We refer interested as orange in Figure 3. The magnetic field measurements
readers to Paré et al. (2024) for full details of the obser- of Lu et al. (2024) were obtained with a spatial resolu-
vations and data reduction. tion similar to that in Paré et al. (2024) and were pro-
In this paper, the inferred magnetic field orientations duced under a similar set of cutoff thresholds. From
represent the component of the magnetic field projected Figure 3, the 850 µm magnetic field orientations do not
onto the plane of the sky and integrated along the line generally agree with those at 214 µm. It is likely that,
of sight. Additionally, we assume that the B-RAT dust given the large difference in the survey wavelengths, the
grain alignment mechanism dominates in the Sgr C com- two studies sampled different temperature regimes along
plex. Therefore, the magnetic field orientations are ob- the line of sight.
tained by rotating the polarization orientations by 90 As was pointed out in Butterfield et al. (2024a) and
degrees (Andersson et al. 2015). For a full visual repre- Paré et al. (2024) for other portions of the CMZ, the po-
sentation, the length of the magnetic field pseudovectors larization fraction exhibits an anti-correlation with 214
presented in this paper is scaled to be proportional to µm intensity. That is, the polarization fraction tends to
the debiased polarization fraction, defined by be relatively higher in low-intensity regions and lower
q in high-intensity regions. This can either be due to the
p = p2m − σp2 , (1) loss of grain alignment in denser regions, or due to the
presence of multiple independent domains of different
where pm is the measured polarization fraction and σp magnetic field orientations along the line of sight in the
is the associated uncertainty. high-column density regions, therefore decreasing the
Throughout this paper, we consider only those mea- integrated polarization fraction in high-column-density
surements that meet the conservative SOFIA/HAWC+ (and thus high-intensity) regions. The faint regions, typ-
selection criteria: a polarization signal-to-noise ratio of ically present in the periphery of molecular clouds, ap-
p/σp > 3; a polarization fraction of p < 50%; and a pear more uniform in magnetic field orientation. The
signal-to-noise ratio of the total intensity of I/σI > 200 region where this anti-correlation is most obvious is the
(Gordon et al. 2018). head-tail cloud, extending from the G359.44-0.10 EGO,
where both 214 µm and 850 µm magnetic field detec-
3. MAGNETIC FIELD IN THE SGR C COMPLEX tions exhibit extremely low polarization fractions, while
3.1. Overview the intensity is high at both of these wavelengths.
In particular, we observe an elongated far-IR-dark re-
Figure 3 presents the 214 µm magnetic field measure-
gion along l = 0.08◦ in Figure 3. This region corre-
ments reported by Paré et al. (2024) as pseudovectors
sponds to the Galactic plane and is home to many com-
superposed on the Stokes I intensity image in the Sgr
pact mid-IR sources and short and curved radio filamen-
C region. A similar pseudovector representation of the
tary structures (see Panel #2 of Figure 7; Liszt & Spiker
magnetic field measurements is adopted throughout this
1995; Carey et al. 2009; Yusef-Zadeh et al. 2009; Han-
paper. The 214 µm emission is dominated by ther-
kins et al. 2020; Cotera et al. 2024). The low 214 µm
mal emission from dust in molecular clouds, including
intensity and polarization non-detections there indicate
the −90 km s−1 cloud, the −65 km s−1 cloud in the
a relative absence of cold dust, where it has either been
south, and FIR-4. Throughout the extended region sur-
heated by the local radiation field, expelled by stellar
rounding Sgr C, the 214 µm polarization measurements
winds and expanding H II regions, or there is simply a
that satisfy the selection criteria are clearly present with
dearth of material there.
high significance throughout the far-IR-bright emission
region. Measurements in faint regions do not commonly
pass the selection criteria, owing either to low signal-to-
noise ratio in either or both total intensity or polariza- 3.2. DCF Analysis
tion fractions, as outlined in Section 2. Pseudovectors in We adopt a modified DCF method (Davis 1951; Chan-
faint regions can be contaminated by residual systematic drasekhar & Fermi 1953) to calculate the sky-plane mag-
effects and/or confused by unrelated emissions originat- netic field strength using the polarization measurements
ing outside the CMZ (Chuss et al. 2019; Butterfield et al. from Paré et al. (2024). The sky-plane magnetic field
 5

Figure 3. The FIREPLACE magnetic field pseudovectors in the Sgr C complex, displayed with the 214 µm intensity image
in the background, rendered using measurements of Paré et al. (2024). The yellow dashes are the 214 µm magnetic field
pseudovectors made using Nyquist sampling at the HAWC+ beamsize, that fulfill the selection criteria described in Section 2.
The length of the vectors is scaled by their polarization fraction. The legend in the lower right corner denotes the length of a
5% polarized pseudovector. The magenta contours are made from the MeerKAT/L-band observation of Heywood et al. (2022)
at levels 0.0004 and 0.002 Jy/beam, with the brighter contour indicating higher intensity level. The orange dashes are the
JCMT/POL2 magnetic field detections at 850 µm (Lu et al. 2024). For visual clarity, we display all 850 µm magnetic field
detections with polarization fraction exceeding p = 0.08 with a length equivalent to an 8% polarized pseudovector.
6

strength, following the DCF method, is expressed as from large and small scales. We write the sky-plane
p σv magnetic field as a superposition BPOS = B0 +Bt , where
|BPOS | = 4πρm , (2) B0 is the ordered, large-scale field, and Bt is the turbu-
σϕ
lent, small-scale field. The angular dispersion can then
where ρm is the mass density, σv is the velocity disper- be approximated as the turbulent-to-ordered magnetic
sion of the gas content, and σϕ is the angular dispersion field energy ratio (Houde et al. 2009):
of the magnetic field pseudovectors. The DCF method
is based on the assumption that the energy density of ⟨B2t ⟩
σϕ2 ≈ . (3)
the turbulent or random gas motions is equal to the ⟨B20 ⟩
energy density of the fluctuations of the magnetic field
To determine this ratio, we adopt a two-component
about the local mean field, where its velocity dispersion
structure function, again separately accounting for the
is assumed to be isotropic. Though the general principle
large- and small-scale magnetic field systems. As a
of the DCF method is commonly agreed upon, specific
statistical measure of the polarization angle difference
measurements of the input parameters of Equation 2 are
∆ϕ(l) ≡ ϕ(r + l) − ϕ(r) at a distance scale l, the struc-
under active refinement and debate, especially in the
ture function 1 − ⟨cos [∆ϕ(l)]⟩ can be computed using
context of a dynamically complex region like the GC
our polarization measurements. As such, fitting our
(Guerra et al. 2023; Butterfield et al. 2024b; Lu et al.
two-component function to the empirical results (i.e.
2024). We address our methodology for estimating the
the FIREPLACE measurements) can constrain the con-
inputs ρm , σv , and σϕ in the following subsections. In
tribution to the overall magnetic field strength from
our DCF analysis, we do not consider the effect of shear
both scales, thereby yielding the required turbulent-to-
on the sky-plane magnetic field strength |BPOS | (e.g.
ordered field ratio to calculate angular dispersion using
the shear-flow approximation considered by Guerra et al.
Equation 3.
2023). Therefore, there is a caveat that our analysis can
The two-component structure function is written as
overestimate |BPOS | as the shear could lower the an-
(Houde et al. 2009)
gular dispersion. However, due to the highly turbulent
2
/2(δ 2 +2W 2 )
nature of the Sgr C region, shear does not play a signif- 1 − e−l 2
icant role in the estimated magnetic field strength, as it 1 − ⟨cos [∆ϕ(l)]⟩ = h i−1 + a2 l , (4)
⟨B2t ⟩
is only significant in clouds with uniform motion. 1+N ⟨B20 ⟩

3.2.1. Angular Dispersion (σϕ ) where δ is the correlation length for the turbulent field,
Past decades of research have shown that the GC W is the beam radius, and a2 is the large-scale contribu-
magnetic field exhibits different behaviours viewed from tion coefficient. Moreover, N is the number of turbulent
different observational scales: a pervasive and ordered cells along the line of sight defined by Houde et al. (2009)
magnetic field in large-scale (Novak et al. 2003; Planck as
(δ 2 + 2W 2 )∆′
Collaboration et al. 2015; Mangilli et al. 2019; Guan N = √ , (5)
2πδ 3
et al. 2021), a turbulent magnetic field driven by local
environments in molecular clouds in small-scale (Chuss computed for a given region, where ∆′ is the cloud’s
et al. 2003; Butterfield et al. 2024a,b; Lu et al. 2024; effective depth. In Equation 4, the first term on the
Paré et al. 2025), and a vertical inter-cloud field signi- right-hand side describes small-scale turbulence, and the
fied by the NTFs (Serabyn & Morris 1994). However, second describes the large-scale contribution.
in a dynamically complex region like the CMZ, the rel- Before fitting Equation 4, we follow Houde et al.
ative contribution of magnetic field systems on different (2009) and infer ∆′ using the shape of the autocorre-
scales is still under investigation. To this end, we follow lation function
the methods developed by Hildebrand et al. (2009) and ⟨P̄ 2 (l)⟩ ≡ ⟨P̄ (r)P̄ (r + l)⟩, (6)
Houde et al. (2009), which have been implemented in
numerous SOFIA/HAWC+ studies (Chuss et al. 2019; where P̄ (r) is the beam-integrated polarized flux at
Guerra et al. 2021; Butterfield et al. 2024a,b). We di- coordinate r. Instead of using the default array-by-
rect interested readers to Houde et al. (2009, 2016) for array approach taken by the polBpy package, we use a
a detailed account of this method, and note that the Fast Fourier Transform to compute ⟨P̄ 2 (l)⟩. Intuitively,
analysis in this subsection was implemented using the Equation 6 is a statistical measure of the average polar-
polBpy library (Guerra et al. 2024). ized flux within a distance l in the sky plane, thereby
We now detail the specific procedure, which separately giving the approximate spatial extent of a cloud’s pro-
accounts for the sky-plane magnetic field contribution jection in the sky-plane. Assuming the cloud is isotropic,
 7

Mackey et al. 2013) to generate MCMC posterior distri-

butions of parameters δ and a2 , but do so for ⟨B2t ⟩/⟨B20 ⟩
through a similar quantity ∆′ (⟨B2t ⟩/⟨B20 ⟩)−1 . All pa-
rameters are constrained to be positive with no upper
bounds, though we do note that δ cannot be greater
than the diameter of the dispersion function’s circular
kernel. The fit was performed on the structure func-
tion with l ≤ 1.25′ to ensure only local dispersion is
#1 sampled. Furthermore, to improve the efficiency of the
MCMC fits, we calculate and fit the dispersion functions
using all available polarization measurements within one
#2 #4 region shown in Figure 4. The best-fit results are then
used as the initial condition of the pixel-by-pixel MCMC
fit to accelerate convergence. The final results were ob-
#3 tained with 300 iterations for each pixel, at which point
we manually inspected the MCMC fit at numerous lo-
cations to ensure convergence. We found that the fit
results do not evolve significantly after 150 iterations.
The resulting images of the variables are shown in Fig-
Figure 4. Individual regions used in the DCF analysis when
ure 5.
fitting Equation 4, overlaid with the 214 µm intensity image
and pseudovector representation. The magenta circle marks
Globally, the a2 and ⟨B2t ⟩/⟨B20 ⟩ values are positively
the extent of the Sgr C H II region. We note that the −65 km correlated, and they are negatively correlated with δ.
s−1 cloud is separated into four numbered regions (number- Given the high column density in many clouds in Sgr
labelled green polygons). C, an increase in turbulence (which entails a decrease in
turbulent correlation length) likely decreases the polar-
we posit that the distribution of polarized flux along the ization fraction integrated along a line of sight. There-
cloud’s depth is similar. We follow Houde et al. (2009) fore, the large-scale contribution B0 naturally becomes
and use the half-width at half-maximum (HWHM) value the dominant term relative to the turbulent contribution
as the effective cloud depth, as it serves as a good ap- Bt in highly turbulent regions.
proximation of the spatial extent of a cloud that contains Among the three high-density molecular clouds, we
the bulk of the polarized flux. However, because the note that FIR-4 and the −90 km s−1 cloud exhibit more
Sgr C complex consists of several regions identifiable by diverging magnetic field directions than the −65 km s−1
their spatial, radial velocity separation, or by the overall cloud, as can be observed in the cyan and pink regions in
trends in their magnetic field orientations, the dispersion Figure 4. In particular, the region immediately north of
calculation only remains meaningful if applied to indi- the Sgr C H II region has a high turbulent-to-ordered ra-
vidual regions with their own identifiable magnetic field tio, which is likely induced by the compression from the
system. The regions that we have thereby identified, ongoing expansion of the H II region. On the contrary,
along with their spatial extents, are shown in Figure 4. the relatively uniform −65 km s−1 cloud and the source
We list their computed effective cloud depths computed C cloud do not exhibit a significant spatial variation for
from Equation 6 in Table 1. The individual clouds in Sgr these parameters, except for the area of the head-tail
C have a typical effective cloud depth of ∼ 1′ or 2.38 pc. feature, where the polarization fraction is extremely low
However, given the known line-of-sight overlap of clouds (see Figure 3).
in Sgr C (Lang et al. 2010), the composite depths of this
region could be much larger than interpreted on the level 3.2.2. Mass Density (ρm )
of individual clouds. More details regarding the auto- We infer the mass density by deriving a column den-
correlation function and effective cloud depth of each sity image using the Hi-GAL survey data (Molinari et al.
region can be found in Appendix A.1. 2010, 2016). Our procedure largely follows the spectral
Having estimated the effective cloud depth, we per- energy distribution (SED) fitting process described in
form a Markov Chain Monte Carlo (MCMC) fit of section 3.1.3 of Chuss et al. (2019) and section 2.2 of
Equation 4 for variables δ, a2 , and most importantly, Paré et al. (2025). Our SED fitting includes the Hi-
⟨B2t ⟩/⟨B20 ⟩ for all pixels in the field of view. The GAL survey images from 160 µm, 250 µm, and 350 µm.
polBpy package uses the emcee package (Foreman- We did not include the 70 µm image due to its uncer-
8

Cloud Colour/Number ∆′ [arcmin] ∆′ [pc]

−65 km s−1 cloud #1 Green #1 0.64±0.15′ 1.45±0.36
−65 km s−1 cloud #2 Green #2 1.43±0.20′ 3.24±0.36
−65 km s−1 cloud #3 Green #3 1.03±0.08′ 2.45±0.19
−65 km s−1 cloud #4 Green #4 1.04±0.13′ 2.48±0.31
FIR-4 Pink 0.84±0.13′ 2.00±0.31
Galactic plane Red 1.39±0.22′ 3.32±0.52
Source C Orange 1.41±0.14′ 3.36±0.33
Far Northeast White 1.36±0.07′ 3.24±0.17
−90 km s−1 cloud Cyan 1.09±0.12′ 2.60±0.29
Table 1. The effective cloud depths for individual regions shown in Figure 4.

tain extinction correction (Compiègne et al. 2010; Bat- Due to the turbulent nature of Sgr C, we adopt a
tersby et al. 2011, 2024). The 500 µm was also excluded constant velocity dispersion of 12 km s−1 by calculating
due to its low resolution. The images are convolved the second spectral moment map of the HCN spectral
with a Gaussian kernel to match the 24′′ resolution of cube provided by Jones et al. (2012) using the Common
the SPIRE 350 µm image. Finally, we perform a pixel- Astronomy Software Applications (CASA; CASA Team
by-pixel SED fitting to the single-temperature modified et al. 2022) package and obtain the dispersion by taking
blackbody modelled by Vaillancourt (2002) its square root.

Iν (ν, T ) = (1 − e−τ (ν) )Bν (T ), (7)

3.2.4. Magnetic Field Strength Results
where Bν is the Planck function. The optical depth τ (ν) The sky-plane component of the magnetic field
is defined as strength calculated from the above-described inputs us-
 β
ν ing Equation 5 is displayed in the lower right panel of
τ (ν) = ϵ , (8)
ν0 Figure 5. The field strength exhibits a large range from
∼ 30 µG to ∼ 400 µG. This range agrees with many
where ϵ is a constant proportional to column density
past attempts to infer the CMZ magnetic field strength
and ν0 is the frequency normalization constant adopted
(Crocker et al. 2010; Chuss et al. 2003; Kruijssen et al.
from Sadavoy et al. (2013). For the fit, we leave ϵ and
2014; Sormani et al. 2020; Tress et al. 2024). We find
T as free parameters while fixing β = 2. Finally, the
that the morphology of the angular or velocity disper-
column density at each pixel is computed as
sion image has a limited effect on the inferred |BPOS |,
ϵ but the local turbulence of a region is anti-correlated
N (H2 ) = , (9)
κν0 µmH with |BPOS | (shown in the bottom panels of Figure 5).
Therefore, the dominant factor affecting |BPOS | is the
where κν0 is the reference dust opacity, µ is the mean local shear and compression. On the scale of individual
molecular weight per hydrogen atom, and mH is the molecular clouds, this conclusion suggests that the most
atomic mass of hydrogen. We adopt ν0 = 1000 GHz, turbulent clouds (FIR-4, the −90 km s−1 cloud, and the
κν0 (1000 GHz) = 0.1 cm2 g−1 and µ = 2.8. The de- −65 km s−1 cloud) exhibit the lowest |BPOS | ∼ 100 µG.
rived column density map is shown in Figure 6. As The head-tail cloud (which extends from the southern
expected, the cold molecular clouds, such as FIR-4, the portion of the Sgr C H II region, see Figure 2) displays
−65 km s−1 cloud, and the −90 km s−1 cloud, which intriguing features from our DCF analysis. The low po-
dominates the far-IR emission, are the most prominent larization fraction “tail” detected in both 214 and 850
in this map. We compare our column density result with µm surveys extending from the G359.45-0.10 “head” dis-
that of Battersby et al. (2024), which uses a slightly dif- plays a consistently lower |BPOS |, creating a sharp con-
ferent method and also excludes the 70 µm image, as trast with the high field strength regions in clouds with
well as Paré et al. (2025), which used the same method uniform field direction. Once more, the low |BPOS | in
with this work but included the 70 µm Hi-GAL image. the head-tail cloud likely signifies its high turbulence
The column density values are similar across these works and complex composition (i.e. several clouds likely su-
upon comparison. perposed along the same line of sight due to the high
column density and low polarization fraction detected
3.2.3. Velocity Dispersion (σv )
in both 214 µm and 850 µm).
 9

a2 δ
0.05◦ 0.05◦
40
140
35
0.00◦ 0.00◦
120
30

a2 [10−3 arcmin−2]
100
−0.05◦ −0.05◦ 25

δ [arcsec]
80
20
−0.10◦ 60 −0.10◦
15
40
10
Galactic Latitude

−0.15◦ −0.15◦
20
5
0
359.55◦ 359.50◦ 359.45◦ 359.40◦ 359.35◦ 359.55◦ 359.50◦ 359.45◦ 359.40◦ 359.35◦

hB2t i/hB20i |BPOS|

0.05◦ 0.05◦
0.8 400

0.7 350
0.00◦ 0.00◦
0.6 300

0.5

|BPOS| [µG]
−0.05◦ −0.05◦ 250
hBt2i/hB02i

0.4 200
◦ ◦
−0.10 0.3 −0.10 150

0.2 100
−0.15◦ −0.15◦
0.1 50

0.0
◦ ◦ ◦ ◦ ◦
359.55 359.50 359.45 359.40 359.35 359.55◦ 359.50◦ 359.45◦ 359.40◦ 359.35◦
Galactic Longitude

Figure 5. Images of the best-fit structure function variables a2 (top left), δ (top right), ⟨B2t ⟩/⟨B20 ⟩ (lower left), and the sky
plane magnetic field strength |BPOS | (lower right) in Sgr C. The blue circle and rectangle denote the approximate location of
the Sgr C H II region and NTF, respectively. We make a 2σ cut for the fitting variables prior to displaying these images. A
small number of pixels with |BPOS | > 10 mG was manually removed to eliminate unphysical field strength.

We also highlight the morphology of the field near ducted in a single wavelength. The 214 µm filter tends
the Sgr C H II region. In the north of the H II re- to sample clouds with colder temperatures (colder com-
gion around l = 359.43◦ , b = −0.07◦ , a half-shell of pared to the warmer Sgr C H II region; Carey et al.
low field strength is apparent. Simultaneously as a high 2009, Hankins et al. 2020), which in some regions may
turbulent-to-ordered ratio region, we speculate that the only be one of multiple clouds along the line of sight.
ionization front of the H II region may be the reason be- The most obvious example in Sgr C is a cloud centered
hind the high turbulence. Although the molecular cloud at l = 359.48◦ , b = −0.17◦ with a broad range of ve-
there has a high column density, the turbulence induced locities (from -92.6 km s−1 to -34.3 km s−1 ) detected in
by the shock and ionization front driving the expansion the HCN line analysis presented in Section 4.2 (see Fig-
presumably leads to an overall low |BPOS |. ure 8). The 850 µm measurements from Lu et al. (2024)
We note a few caveats in our DCF analysis. One nicely trace the morphology of this cloud but disagree
should be cautious that the FIREPLACE survey is con- with the 214 µm magnetic field orientations in this re-
10

features nearly identical to those in the JCMT/SCUBA

1023
850 µm survey (Pierce-Price et al. 2000).
0.00◦ The Galactic plane appears bright in the IR regime
out to 70 µm and in radio, with the peak emission at
Galactic Latitude

N (H2) [cm−2]
these wavelengths arising near the Sgr C H II region. In
mid-IR and radio images (Panels #2–4, and #7 in Fig-
−0.10◦ ure 7), the Sgr C H II region exhibits a similar elliptical
morphology elongated along the Galactic plane. We ob-
serve an absorption “tail” originating from inside the
Sgr C H II region pointing towards the southeast, where
−0.20◦
the EGO is located (G359.44-0.10; Kendrew et al. 2013).
1022
This low-intensity “valley” is observed at 8, 24, and 70
359.60◦ 359.50◦ 359.40◦ µm and is a result of foreground absorption by the head-
Galactic Longitude tail cloud, which turns to emission in far-IR wavelengths
beyond 70 µm. The mid-IR intensity images tend to be-
Figure 6. The column density map derived from fitting come more diffuse at longer wavelengths, which is par-
Equation 7 to the Hi-GAL 160, 250, and 350 µm images. tially due to the trend that lower-temperature dust is
The blue circle and rectangle indicate the location of the Sgr generally the most extended, and partially due to de-
C H II region and the Sgr C NTF.
creased resolution at longer wavelengths.
The 8 µm emission traces the polycyclic aromatic
gion, which may indicate that the 850 µm measurements hydrocarbons (PAHs) ionized by stellar UV radiation
are sampling a different cloud along that line of sight, (Stolovy et al. 2006; Narayanan et al. 2023), thereby
which may naturally possess a different magnetic field showing the extent of the elliptical region impacted by
orientation. As such, inferring |BPOS | using only one UV emission and star formation around the Sgr C H II
wavelength introduces some uncertainty, since different region. In the 8 to 70 µm interval (Panels #2–4 in Fig-
clouds, or even the same cloud with different tempera- ure 7), the emission also traces warm dust heated by
ture domains, along a given line of sight can have in- stellar emission. In this band, one could also observe
dependent field directions that then lead to a reduction the “head” of the head-tail structure as an extremely
of the polarization fraction. Another possible source of compact and luminous source (classified as an EGO by
error comes from the column density estimation, as we Kendrew et al. 2013) to the immediate south of the H II
assume a single-temperature blackbody, having multi- region, also becoming more diffuse towards longer wave-
ple clouds (possibly with different temperatures) along lengths.
the same line of sight could cause an error in our col- In the far-IR regime (214–850 µm, Panels #5-7 in Fig-
umn density estimation, which also affects the |BPOS | ure 7), the images are dominated by thermal emission
calculation. In regions dominated by higher tempera- from cold dust residing in molecular clouds. The Galac-
tures, the DCF analysis cannot be reliably performed tic plane and the H II region, by contrast, appear dark
since there is an absence of valid 214 µm magnetic field in the far-IR regime. The “head” of the head-tail struc-
detections. ture is larger in size at far-IR wavelengths than in the
4. MULTI-WAVELENGTH OBSERVATIONS
mid-IR. Though this decrease in size could be a result
of cold dust being more diffuse, it can again be a result
This section provides an overview of the relevant fea- of lower spatial resolution in far-IR wavelengths. The
tures of Sgr C by comparing existing multi-wavelength “tail” also appears bright at these wavelengths, reflect-
studies with the FIREPLACE 214 µm magnetic field ing a relatively high column density of cold dust. Indeed,
measurements. we have an inferred column density of ∼ 1023 cm−2 in
4.1. Assembly of Observations this region, shown in Figure 6.
A noteworthy structure is the vertical “stripe” extend-
Figure 7 shows nine intensity images covering a broad
ing northwards from l = 359.52◦ , b = −0.05◦ that ap-
range of wavelengths, from 1 nm to 20 cm, to highlight
pears in the 450 and 850 µm SCUBA images (Panels
the significant structures in the Sgr C region. Of the
#6–7 in Figure 7), though it is most noticeable at 450
published surveys of the Sgr C region, the images pre-
µm and is much more diffuse at 850 µm. With the caveat
sented in Figure 7 have been chosen to avoid repetition.
that this could be an artifact due to filtering, we note
For example, the 870 µm APEX/ATLASGAL survey
that the northern end of this “stripe” coincides with the
(Schuller et al. 2009) is not included because it exhibits
 11

Figure 7. Selected multi-wavelength images of the Sgr C complex. The region has been surveyed in 1 − 4 keV X-rays by
Chandra (Wang 2021), 8 µm by Spitzer/IRAC (Stolovy et al. 2006), 24 µm by Spitzer/MIPSGAL (Carey et al. 2009), 70 µm by
Herschel/PACS (Molinari et al. 2010), 214 µm by SOFIA/HAWC+ (Paré et al. 2024), 450 µm and 850 µm by JCMT/SCUBA
(Pierce-Price et al. 2000), 2 mm by IRAM/GISMO (Arendt et al. 2019), and at 20 cm by MeerKAT (Heywood et al. 2022). For
visual comparison between the multi-wavelength images and the magnetic field detections, we add the 214 µm magnetic field
pseudovectors from Paré et al. (2024) in Panel #5 as yellow segments. The corresponding beam sizes are displayed as white
circles in the lower right corner of each image. All images are displayed with a logarithmic stretch.
12

mid-IR ridge that connects the extended thermal source structures. Here, we appeal to the 3 mm Mopra spectro-
AFGL5376 and the Galactic plane (Uchida et al. 1990, scopic survey conducted by Jones et al. (2012) and the
1994). The 214 µm magnetic field orientations do not H I absorption analysis by Lang et al. (2010) for velocity
bear any obvious relationship to the morphology of this constraints and line-of-sight placements.
stripe. We go no further than noting the above corre-
spondence, as a discussion of AFGL5376 is beyond the 4.2.1. Structure of Clouds in Sgr C
scope of this paper.
We select the HCN J=1-0 line to deduce the dynam-
As a transition between far-IR and radio wavelengths,
ical structure of the clouds in Sgr C to aid our un-
the GISMO image at 2 mm displays characteristics from
derstanding of the FIREPLACE magnetic field orienta-
both regimes, showing a combination of thermal emis-
tions. The HCN J=1-0 line is one of the most prominent
sion from cold dust and free-free synchrotron emission
lines in the GC, arising in relatively dense gas (Jones
from ionized gas. As such, we see not only the 100 pc
et al. 2012). Figure 8 shows the velocity channel maps
ring dominated by cold dust, but also the ionized gas in
of this line in the velocity range from −92.6 km s−1 to
the Sgr C H II region.
−34.3 km s−1 . We note that a foreground cloud often
The radio continuum emission (Panel #9 in Figure 7)
referred to as the “3 kpc arm” causes significant absorp-
is produced by free-free emission, primarily from the
tion around −50 km s−1 (Lang et al. 2010). The channel
Sgr C H II region and by synchrotron emission from the
maps in Figure 8 show a coherently decreasing radial ve-
nearby NTFs (Heywood et al. 2022; Yusef-Zadeh et al.
locity from the north of the Sgr C H II region (≳ −90
2022a,c), the most prominent of which is the Sgr C NTF
km s−1 , which is constituted of FIR-4 and the −90 km
bundle that appears to originate near the Sgr C H II re-
s−1 cloud) to the southeast (∼ −65 km s−1 , which is
gion. Although the peak emission from the Sgr C NTF
constituted of the head-tail cloud and the −65 km s−1
is at b ≳ −0.10◦ and is most obviously seen to extend
cloud). These velocities follow the 100 pc dust ring re-
towards the north, there is a faint extension towards the
vealed by Molinari et al. (2011). The radial velocity of
south which ends near the relatively compact elliptical
FIR-4 (G359.45+0.02) is similar to that of the nearby
or bipolar radio source G359.46-0.17. We hypothesize
−90 km s−1 cloud, which is a part of the 100 pc ring.
that the abrupt intensity change in the NTF is caused
FIR-4 is therefore likely physically related to the 100 pc
by its interaction at that point with the high-density
ring.
gas in the head-tail structure (detailed in Section 5.1.1).
The Sgr C H II region is largely devoid of HCN emis-
At that location, the diffusion of relativistic electrons to
sion, except for the intruding head-tail cloud. Both
the south along the field lines is largely, but apparently
the radio recombination line measurements reported by
not entirely, blocked by the presence of the dense gas.
Liszt & Spiker (1995) and H I absorption analysis by
Furthermore, if the interaction with the cloud is taking
Lang et al. (2010) determined the velocity of the H II
place along the ionization front of the Sgr C H II region,
region to be ∼ −60 km s−1 . Given the velocity struc-
the acceleration of the electrons to relativistic energies
ture observed in Figure 8, we infer that the −90 km s−1
could be the result of magnetic field line reconnection
cloud, the −65 km s−1 cloud, and the Sgr C H II region
there, as proposed by Serabyn & Morris (1994) or by
constitute a continuous structure linked to the 100 pc
diffusive shock acceleration if a strong stellar wind cre-
ring, and are likely intrinsically related. We direct the
ates a shock front there (Rosner & Bodo 1996).
reader to more details of the line-of-sight placement of
There are also abundant short and curved radio fila-
structures in Sgr C to Section 3.6.3 of Lang et al. (2010).
mentary structures to the east of the Sgr C H II region
In the southern portion of the −65 km s−1 cloud,
along the Galactic plane, as noted in Liszt & Spiker
there is a bright bulk of HCN emission at l = 359.47◦ ,
(1995). Some of these radio structures can be non-
b = −0.17◦ visible at all displayed velocities. In this re-
thermal, as inferred by the absence of IR counterparts.
gion, we do not find any morphological relationship be-
Their relative faintness could simply be interpreted in
tween the HCN emission and the 214 µm magnetic field
terms of a smaller population of relativistic electrons.
orientations. However, many 850 µm magnetic field de-
We also do not observe any correspondence between the
tections were reported in this cloud by Lu et al. (2024),
FIREPLACE magnetic field orientations and these short
and they do not generally agree with the 214 µm results
radio filaments.
of Paré et al. (2024). The HCN emission offers several
possible explanations for this discrepancy. It could be
4.2. Spectral Line Analyses that this is a particularly disturbed and turbulent cloud
To better understand the dynamics of Sgr C, we re- with multiple temperature components, or there could
quire the relative position and motion of the observed be multiple clouds superposed along the same line of
 13

Figure 8. Velocity channel maps of the HCN J=1-0 line measured by the 3 mm Mopra survey (Jones et al. 2012). The survey
has spatial and spectral resolutions of 36′′ and 1.84 km s−1 , respectively. The mean velocity of each channel is indicated in the
top left corner. These channel maps cover the velocity range between −92.6 and −34.3 km s−1 in steps of 7.3 km s−1 . The
grey-scale contours show the peak emission of HCN integrated through all observed velocities drawn at 0.75, 1.25, 1.75 K km
s−1 , with the brighter curves indicating higher intensity. The yellow schematics indicate the approximate placement and extent
of the Sgr C H II region and the Sgr C NTF. The green star in the central panel denotes the location of the G359.44-0.10 EGO,
which is addressed in detail in Section 5.3. The global low intensity in channels near −50 km s−1 is due to absorption in the
foreground “3 kpc arm” (Lang et al. 2010). The 214 µm magnetic field pseudovectors in this figure are rendered using Nyquist
sampling at the Mopra beam size.
14

sight. Both can cause the magnetic field measured from mately leading to a low polarization fraction, a higher
different wavelengths to exhibit different orientations. It turbulent-to-ordered magnetic field strength ratio, and
is also likely that this cloud has a radial velocity gradient relatively low inferred sky-plane magnetic field strength,
along the line of sight, given that this is where the ro- as shown in Figure 5.
tating 100 pc ring reverses direction in projection. The
observation of this particularly broad-linewidth cloud re- 5. INDIVIDUAL FEATURES IN THE SGR C
affirms our caveat for the DCF analysis in Section 3.2.4 COMPLEX
regarding the incomplete accounting for multiple line-of- In this section, we use the 214 µm magnetic field
sight clouds having different velocities in the derivation measurements from Paré et al. (2024) along with multi-
of the magnetic field strengths. wavelength images to draw inferences about individual
The previously mentioned 214 µm magnetic field mea- features in the Sgr C complex.
surement cavity at l = 359.44◦ , b = −0.13◦ between
the −65 km s−1 cloud and Cloud #4 (centered at 5.1. Associated NTFs & Radio Features
l = 359.43◦ , b = −0.17◦ ; see Figure 4) is of interest As has previously been widely discussed, NTFs are
to be studied. In Figure 8, the −65 km s−1 cloud and valuable probes of the large-scale magnetic field in the
Cloud #4 appear to be portions of a continuous struc- inter-cloud volume of the central few hundred parsecs
ture so they are likely physically related. We hypothe- of the Galaxy (Serabyn & Morris 1994; Morris 2006;
size that they started as one cloud and were later sep- Thomas et al. 2020). The following subsections examine
arated into the currently observed shape by an outside the magnetic field behaviour around the NTFs and radio
force that created the magnetic field measurement cavity structures observed in the Sgr C region, keeping in mind
at l = 359.44◦ , b = −0.13◦ . We note a potentially rele- that the FIREPLACE detections primarily probe the
vant X-ray association with this cavity in Section 5.2.2. magnetic field within clouds.

4.2.2. Shock Tracer in the Twisted 100 pc Ring 5.1.1. The Sgr C NTF
Figure 10 shows the FIREPLACE 214 µm magnetic
As previously mentioned, Sgr C is located on the west-
field pseudovectors superimposed upon five selected im-
ern vertex of a dust ring (Molinari et al. 2011), and is
ages covering the Sgr C NTF. This extended NTF ap-
postulated to be a collision site of X1 -X2 orbits due
pears luminous and compact below l = −0.03◦ and it
to the bar potential (Sormani et al. 2018, 2020), which
splits into two parallel filaments and becomes fainter
likely induces shocks and star formation. Meanwhile,
and more diffuse above this latitude. This NTF over-
the expanding Sgr C H II region can also induce a com-
laps on the sky-plane with three distinct magnetic field
pression front that collides with surrounding molecular
intervals:
clouds (Liszt & Spiker 1995) and add to the effect of
the large-scale shocks. We employ the many 214 µm • (i) Below b = −0.10◦ , where the faint end of this
magnetic field detections toward the molecular clouds in NTF disappears in the head-tail cloud and the −65
the Sgr C region to examine how such shocks might be km s−1 cloud (detailed in Section 5.1.5)
manifested in the magnetic field orientations. Figure 9
shows the velocity channel maps of the SiO J=2-1 line, • (ii) −0.08◦ ≲ b ≲ −0.02◦ , where the FIREPLACE
a known shock tracer also from the Mopra survey (Jones magnetic field detections arise from the −90 km
et al. 2012). The SiO channel maps exhibit a similar ve- s−1 cloud.
locity structure to that of HCN, indicating that the SiO • (iii) Above b = 0◦ , where this NTF overlaps with
emissions are from within the 100 pc ring. However, the a magnetic field system dominated by FIR-4 (de-
SiO emission appears much less spatially extended than tailed in Section 5.1.2).
that of HCN and more concentrated around the Sgr C
H II region. This local enhancement of SiO indicates Interval (ii) above is roughly the extent of Part A as
that the portions of the 100 pc ring surrounding the Sgr defined in Roy (2003) and Lang et al. (2010). We note
C H II region are particularly subjected to shocks, likely additionally that the Sgr C NTF has a faint end towards
due to a combination of 1) the impact of the dust lane the south, detailed in Section 5.1.5.
inflowing along the Galactic bar upon the CMZ material In the radio continuum below l = −0.02◦ , the NTF
in the 100 pc ring, and 2) the collision between ambient has the same width but Part A is more luminous than
gas and the expanding shell surrounding the H II region Part B (the part of the Sgr C NTF below l = −0.08◦
(Riquelme et al. In-preparation). The collision likely labelled in Figure 10; Roy 2003). The radio spectral
increases the turbulence within the 100 pc ring, ulti- index analysis of Yusef-Zadeh et al. (2022a) indicates
 15

Figure 9. Velocity channel maps of SiO J=2-1 emission between −110 and −49 km s−1 , measured by the Mopra survey (Jones
et al. 2012). The 214 µm magnetic field pseudovectors are superimposed as white line segments, as in Figure 3. The yellow
schematic indicates the approximate location and extent of the Sgr C H II region and the Sgr C NTF.
16

Figure 10. The 214 µm magnetic field pseudovectors near the Sgr C NTF and G359.43+0.02 YSO superimposed on emission
at 20 cm (Heywood et al. 2022), 8 µm (Stolovy et al. 2006), 24 µm (Carey et al. 2009), 250 µm (Molinari et al. 2010), and 850
µm (Pierce-Price et al. 2000). The yellow segments are the 214 µm magnetic field pseudovectors from Paré et al. (2024) and
the orange segments are the 850 µm magnetic field pseudovectors from Lu et al. (2024).
 17

that Part A has an average spectral index of ∼ −1.02 FIR-4 intersects with the Sgr C NTF in projection on
and Part B has an average spectral index of ∼ −0.65. the plane of the sky. Lang et al. (2010) suggested that
In Part A, the magnetic field orientation near the NTF FIR-4 could have the same line-of-sight placement as the
roughly follows its direction. As our magnetic field mea- Sgr C NTF. Examining the MeerKAT image in detail,
surements sample the field from relatively cold clouds, we find that the radio luminosity at the intersection of
the direction of the FIREPLACE magnetic field in the the east-west ridge of FIR-4 with a few of the two west-
−90 km s−1 cloud is consistent with the orientation of ernmost sub-filaments of the NTF is ∼20% higher than
the Sgr C NTF, as are the 850 µm measurements of Lu the sum of individual luminosities of the neighbouring
et al. (2024). Moreover, the DCF analysis from Fig- regions of the superposed features. Moreover, the po-
ure 5 shows that there is a “tunnel” of low turbulent- tential interaction site displays an increased north-south
to-ordered ratio and high |BPOS | along the Sgr C NTF. radio emission width (which is much larger than the
However, such alignment ceases beyond l > −0.02◦ due MeerKAT beam size; seen in filament #13 in Heywood
to an absence of magnetic field detections outside the et al. 2022), and the NTF sub-filaments brighten as they
molecular cloud. Additionally, the NTF becomes much approach the intersection site from both the north and
more diffuse after departing from the molecular cloud south. These characteristics lead us to suggest that FIR-
toward the north. Above l = 0◦ , the magnetic field 4 is interacting directly with the Sgr C NTF.
orientation and the Sgr C NTF are no longer aligned, We note that there is an absence of magnetic field
which could be attributed to strong far-IR emission from detections in the far-IR dark region that lies north of
FIR-4 dominating the magnetic field detections in that the Sgr C H II region at latitude -0.02◦ and south of
region. the cluster of YSOs. This region has faint emission at
both 8 and 24 µm, but gradually becomes negligible
towards far-IR and radio frequencies (c.f., Figure 10).
5.1.2. FIR-4: G359.43+0.02 This indicates that this region contains mainly diffuse
warm gas, possibly heated by the nearby YSOs.
At the northern portion of the Sgr C NTF where it
splits into two filaments and begins to fade out toward
5.1.3. The G359.33-0.15 Filaments
the north, lies a short, linear radio feature oriented par-
allel to the Galactic plane along b ≈ 0.01◦ . It was origi- A fainter bundle of parallel radio filaments, referred to
nally found and denoted as FIR-4 by Odenwald & Fazio as the “Hummingbird” by Yusef-Zadeh et al. (2022a),
(1984) in the course of their far-IR survey of the GC. is located to the southwest of the Sgr C H II region.
FIR-4 and its counterparts are displayed in Figure 10. The northeastern end is coincident with an extended X-
The mid-IR counterpart of FIR-4 is the G359.43+0.02 ray source noted at energies < 6 keV (which we detail
YSO Cluster reported by Yusef-Zadeh et al. (2009) and in Section 5.2.2). Figure 11 shows the MeerKAT/L-
prominently displayed in the central panel of Figure 10. Band image of the G359.33-0.15 filament bundle and
Given its moderately strong far-IR brightness, many its possible 8–70 µm counterparts. Such counterparts
significant 214 µm magnetic field detections were made would be more characteristic of a thermal source rather
in FIR-4. Overall, the magnetic field exhibits a diver- than an NTF. However, the MeerKAT spectral index
gent morphology with the magnetic field pointing in the map (Yusef-Zadeh et al. 2022a; Bally et al. 2024) in-
radial direction away from the centre of the source. The dicates that the spectrum of at least a portion of the
polarization fraction in the centre and near the radio G359.33-0.15 filaments is too steep to be thermal, al-
thread is mostly small, but increases towards the pe- though the mean spectral index is somewhat ambigu-
riphery. The |BPOS | in the centre is also lower than ous in this regard. Furthermore, the multiplicity of the
that in the periphery (c.f., Figure 5). We hypothesize sub-filaments is a characteristic commonly found among
that this morphology likely traces the outflow driven by NTFs. Therefore, the apparent infrared counterparts
protostellar winds, which induce high turbulence within to the G359.33-0.15 filaments could be coincidental and
the cluster but a collective, ordered outflow on the pe- unrelated. No far-IR emission beyond 70 µm has been
riphery. In particular, the radio counterpart of FIR-4 detected toward G359.33-0.15, and consequently, there
coincides with the region along l = 0.01◦ with the low- are no 214 µm magnetic field detections to report along
est polarization fraction. The radio and mid-IR emission the G359.33-0.15 filaments.
from FIR-4, arrayed as they are along the base of the Unlike most NTFs in the CMZ, which are perpendic-
embedded stellar cluster, suggest that UV emission from ular to the Galactic plane, the G359.33-0.15 filaments
nearby YSOs has heated the local dust and has produced are much closer to being parallel to the Galactic plane.
an ionization front, which appears linear in projection. Another of its notable characteristics is that it is appar-
18

Figure 11. The G359.33-0.15 NTF towards the southwest of the Sgr C H II region shown at 20 cm (Heywood et al. 2022;
Yusef-Zadeh et al. 2022a), 8 µm (Stolovy et al. 2006), 24 µm (Carey et al. 2009), and 70 µm (Molinari et al. 2010), overlaid
with 214 µm magnetic field pseudovectors from Paré et al. (2024). The orange segments in the lower right panel are the 850
µm magnetic field detections from Lu et al. (2024).

ently interacting with an infrared dark cloud (IRDC) tion fraction. The magnetic field detections imply a low
at its southwestern end (compare the top two panels in turbulent-to-ordered ratio and a largely uniform |BPOS |,
Figure 11 at near l = 359.32◦ , b = −0.17◦ ), and that as shown in Figure 5.
interaction has led to a strong morphological bunching Source C contains a short radio filament with a bright
and brightening of the radio emission at the interface southern end that ends abruptly at l = 359.390◦ , b =
(see Fig. 3k of Yusef-Zadeh et al. (2022b) for additional −0.077◦ in the MeerKAT/L-Band image. To the north-
radio detail). The interaction has also led to the pro- east of that point, there are two adjacent and paral-
duction of a slightly extended X-ray source at or very lel filaments, the upper one of which appears to un-
near that location (Yusef-Zadeh et al. 2007, their desig- dergo a sudden northward deflection at l = 359.405◦ ,
nation: G359.32-0.16), raising the possibility that this b = −0.06◦ , above which its brightness increases. The
location is where the relativistic electrons that illumi- deflection point of this filament is located within a far-IR
nate the G359.33-0.15 filament are produced, by anal- emission cavity centered at l = 359.405◦ , b = −0.060◦ ,
ogy with G359.1-0.2 (Uchida et al. 1996) or G0.18-0.04, where we find no significant magnetic field detection,
and the Sickle (Serabyn & Morris 1994). This inter- but north of the cavity, the magnetic field orientation is
esting structure clearly warrants further investigation. approximately aligned with the filament. If the filament
Unfortunately, no magnetic field detections have been is a magnetic NTF, its “deflection” of the top filament
measured along the G359.33-0.15 NTF and the IRDC could occur where it transitions going northward from
with which it is interacting due to a low far-IR luminos- the far-IR cavity into the local molecular cloud, thereby
ity (Paré et al. 2024). reflecting the differing field orientations in the two me-
5.1.4. Source C: G359.39-0.08 dia.
Another notable feature in Source C is the curvature of
Source C is a radio source associated with a diffuse H II the magnetic field around the peak of the 24 µm emission
region to the immediate west of the Sgr C H II region, at l = 359.37◦ , b = −0.08◦ . This could perhaps be
where thermal counterparts have been reported (Liszt attributed to the expansion of the ionized gas in this
& Spiker 1995; Hankins et al. 2020). Figure 12 displays region.
multi-wavelength structures of Source C overlaid with
the 214 µm magnetic field pseudovectors, which exhibit
a uniformly vertical structure with a similar polariza-
 19

Figure 12. The FIREPLACE 214 µm magnetic field pseudovectors around Source C, superimposed on emission maps at 20
cm (Heywood et al. 2022), 8 µm (Stolovy et al. 2006), 24 µm (Carey et al. 2009), and 250 µm (Molinari et al. 2010). The yellow
segments are the 214 µm magnetic field pseudovectors reported by Paré et al. (2024). (No 850 µm magnetic field detections
from Lu et al. (2024) were reported near Source C.)
20

5.1.5. The Faint Southern Tail of the Sgr C NTF & ing shell of the H II region, presumably driven by stellar
Source G359.467-0.17 winds and the advancing ionization front, compresses
The prominent radio NTF extending north from Sgr the tangential component of the field in the surrounding
C has an apparent faint southern “tail” that extends clouds to the point where it has become the dominant
onto the −65 km s−1 cloud, in projection, starting at component.
l = 359.468◦ , b = −0.12◦ . The top left panel of Fig- However, the magnetic field orientation near the head-
ure 13 shows the full extent of the Sgr C NTF, includ- tail cloud and the G359.44-0.10 EGO displays a different
ing the faint southern extension. It is apparent that morphology. Near the EGO, the magnetic field is mainly
there is a sharp decline in the brightness of the Sgr C influenced by activity in the EGO (protostellar winds
NTF precisely where it overlaps with the head-tail cloud and jets) instead of the ionization front of the Sgr C
(depicted in yellow contours in the top left panel of Fig- H II region, which is behind the head-tail cloud (Lang
ure 13) on the sky-plane, suggesting that the NTF is et al. 2010). We return to a more detailed discussion of
interacting with the head-tail cloud. A zoomed-in view the magnetic field morphology of the head-tail cloud in
of the southern extension is shown in the top right panel Section 5.3.
of Figure 13. This tail is parallel to a bundle of other
radio filaments having a more diffuse nature, shown in 5.2.1. [C II] Emission: A “Shell” around the H II Region
the top right panel of Figure 13 (see also Figure 12 of To shed more light on the properties of the Sgr C
Yusef-Zadeh et al. 2022a). Additionally, we note the H II region, we use the [C II] line analysis conducted
obvious spatial anti-correlation between this bundle of by Riquelme et al. (In-preparation) using the upgraded
filaments with the PAH emission shown in 8 µm. German Receiver for Astronomy at Terahertz Frequen-
The NTF tail persists until l = 359.47◦ , b = −0.25◦ . cies (upGREAT) instrument aboard the SOFIA tele-
Along its trajectory, the tail is tangent in projection to scope. The [C II] line traces the warm ionized and par-
a compact radio source, G359.467-0.17 (shown in de- tially ionized media exposed to UV photons from the
tail in the bottom panel of Figure 13), which has the HII region. Figure 14 displays the C II velocity channel
morphology of a tilted bipolar nebula, along with clear image at −41 km s−1 (Riquelme et al. In-preparation)
IR counterparts that indicate a centrally heated, dusty overlaid by the FIREPLACE 214 µm magnetic field
nebula. Because the brightness of the NTF tail does not pseudovectors for comparison with the morphology of
vary as a function of its proximity to G359.467-0.17, it the C II shell. In the immediate vicinity of the Sgr C
seems unlikely that the two structures are interacting. H II region, the 850 µm magnetic field measurements
reported by Lu et al. (2024) are largely consistent with
5.2. The Sgr C H II Region the 214 µm ones shown here.
The 19.6′′ resolution of the FIREPLACE survey (Paré Figure 14 reveals what appears to be a double-cavity
et al. 2024) allows us to closely examine the magnetic morphology associated with the Sgr C H II region, with
field morphology near the Sgr C H II region. From Fig- the western cavity being larger in size. The [C II] shell
ure 3, we observe a cavity within the Sgr C H II region, in surrounds and overlaps with the outer perimeter of the
which resides relatively high-temperature dust, as man- H II region appearing in radio emission (see Panel #9
ifested in mid-IR maps (Panels #2-4 in Figure 7). In in Figure 7), presumably identifiable as the ionization
this paper, we focus on magnetic field detections in cold front. The 214 µm magnetic field orientation is mostly
molecular clouds surrounding the H II region. Because tangential to the cavities, except in the southeast di-
such warm regions are not well sampled by our 214 µm rection because of the dense foreground head-tail cloud.
survey, the field there would best be studied with fu- The [C II] emission in this shell is found to deviate some-
ture mid-IR polarimetry; indeed, a recent study by Bally what from circular symmetry, with the strongest emis-
et al. (2024) also inferred the existence of strong mag- sion in the northwest and near the G359.44-0.10 EGO.
netic fields in this H II region, in the wavelength range Together with the tangential magnetic field morphology,
1-5 µm with JWST. these observations resonate with the conclusion found in
The magnetic field orientation in the northern hemi- the simulations of Arthur et al. (2011), that the expan-
sphere of the Sgr C H II region is mostly tangential to the sion of an H II region creates a tangential magnetic field
radio shell shown in both Figure 3 and Panel #9 of Fig- orientation around its border. Moreover, the [C II] cav-
ure 7. This tangential morphology along the northern ities are separated by a “bridge” of faint [C II] emission
surface of the H II region is consistent with the results centered at l = 359.455◦ , b = −0.090◦ . Only two sig-
of magnetohydrodynamic simulations (Krumholz et al. nificant 214 µm pseudovectors are present toward the
2007; Arthur et al. 2011). We posit that the expand- “bridge,” and both are parallel to it.
 21

Figure 13. Top left: The full extent of the Sgr C NTF in radio continuum (Heywood et al. 2022). The orange contours are
the FIREPLACE 214 µm emission at (3, 5.25, 7.5, 9.75, 12) Jy/pixel from Paré et al. (2024). Note the faint southern extension
of the bright NTF bundle to the north. Top right: The faint southern “tail” of the Sgr C NTF, and source G359.467-0.17,
displayed at wavelengths of 20 cm (Heywood et al. 2022), 8 µm (Stolovy et al. 2006), 24 µm (Carey et al. 2009), and 70 µm
(Molinari et al. 2010), overlaid with the FIREPLACE 214 µm magnetic field pseudovectors from Paré et al. (2024). Bottom:
The extended 20 cm radio emission towards the south of the Sgr C H II region, from Heywood et al. (2022), 8 µm (Stolovy
et al. 2006), 250 µm (Molinari et al. 2010), and 850 µm (Pierce-Price et al. 2000) overlaid with the 214 µm magnetic field
pseudovectors from Paré et al. (2024).
 22

-0.06°
-0.06°
-0.08°
Galactic Latitude

Galactic Latitude
-0.08°
-0.10°

-0.10° -0.12°

-0.12° -0.14°
p = 5% p = 5%
-0.16°
359.48° 359.46° 359.44° 359.42° 359.48° 359.46° 359.44° 359.42° 359.40°
Galactic Longitude Galactic Longitude
Figure 14. The CMZ in [C II] Survey image of the −41 Figure 15. The 2.5-4 keV Chandra X-ray image reported by
km s−1 velocity channel observed using SOFIA/upGREAT Wang (2021). The white contour outlines the [C II] emission
(Riquelme et al. In-preparation). White segments are the 214 in the −41 km s−1 velocity channel at the intensity of 13 K
µm pseudovectors made from data reported by Paré et al. (Riquelme et al. In-preparation). The yellow segments are
(2024). Pink stars indicate the positions of two Wolf-Rayet the 214 µm magnetic field pseudovectors made from Paré
stars reported by Geballe et al. (2019). The cyan schematics et al. (2024). The pink-filled star symbols mark the locations
mark the approximate location and extent of the Sgr C H II of two Wolf-Rayet stars (Geballe et al. 2019). The purple
region and the Sgr C NTF. lines denote the location of the Sgr C H II region and the Sgr
C NTF, and the green triangle indicates the location of the
EGO at the “head” of the head-tail cloud.
The polarization fraction in high [C II] intensity re-
gions appears low due to their high column density
and high turbulent-to-ordered ratio, as shown in Sec- towards the southeast in the bands between 1 – 6 keV
tion 3.2.1. Indeed, for the same region, the inferred can perhaps be attributed to absorption by the head-tail
|BPOS | near the cavity is low. cloud situated in front of the southeast portion of the
Geballe et al. (2019) reported two dusty Wolf-Rayet H II region.
stars that are located in the Sgr C H II region (pink- A somewhat stronger region of extended X-ray emis-
filled stars in Figure 14). The western Wolf-Rayet star sion lies outside and to the southwest of the Sgr C H II re-
is positioned at the centre of the western cavity, and gion at l = 359.42◦ , b = −0.12◦ . It is possibly connected
is therefore a strong candidate for contributing to the to the X-ray emission within the [C II] cavity through
formation of this cavity and to the ionization of the Sgr a “bridge” at l = 359.425◦ , b = −0.103◦ , although it
C H II region. Its winds are potentially responsible for could also be a slightly overlapping, independent source
the tangential magnetic field morphology around the rim at a different distance. This X-ray feature, clearly ex-
of the Sgr C H II region. ternal to the Sgr C H II region, exhibits a strong spatial
anti-correlation with far-IR emission and with the pres-
ence of significant 214 µm magnetic field detections (see
5.2.2. X-rays in the Sgr C H II region
Figure 3).
Additional morphological correspondence is observed We note again that the location of the X-ray emission
in the X-ray images of Wang (2021). Figure 15 overlays to the southwest of the Sgr C HII region coincides in
the 2.5-4 keV Chandra X-ray image from Wang (2021), projection with the northeast ends of the G359.33-0.15
with the 214 µm pseudovectors. Toward the Sgr C H II radio filaments (displayed in Figure 11 and detailed in
region and its [C II] shell, much of the X-ray emission ap- Section 5.1.3), though it is not yet clear whether there
pears to arise from within the H II region. The emission is any physical association between these two features.
becomes fainter in the [C II] shell marked by the white
contours in Figure 15. If the X-rays arise from within 5.3. The Head-tail Cloud (G359.44-0.10) and the −65
the HII region, the relative paucity of X-ray emission km s−1 cloud
 23

To the southeast of the Sgr C H II region lies the head- region, and describe potential relationships with struc-
tail cloud. The head-tail cloud consists of two parts: an tures observed at other wavelengths:
exceptionally bright compact source (the G359.43-0.10
EGO or the “head”; Kendrew et al. 2013; Crowe et al. • Cloud #1. This cloud, centered at l = 359.495◦ ,
2024) and a diffuse and cometary ridge of high far-IR b = −0.11◦ , lies between the Galactic plane to
luminosity, high column density (see Figure 6), and low the north (the far-IR-dark region near the top of
polarization fraction extending towards the southeast Figure 16) and the high turbulent-to-ordered ra-
(the “tail” addressed in Section 5.1.5). This cloud is also tio “ridge” (the green contours in Figure 16) in
found to exhibit a consistently high turbulent-to-ordered the south. Lu et al. (2024) argue that the field
ratio (⟨B2t ⟩/⟨B20 ⟩), as detailed in Section 3.2.1.Figure 16 morphology in this region, which is parallel to
shows the magnetic field pseudovectors of the head-tail the Galactic plane, is due to its interaction with
cloud and the −65 km s−1 cloud overlaid with various numerous H II regions along the Galactic plane
background images. Initially discovered by Forster & (Carey et al. 2009; Hankins et al. 2020).
Caswell (2000) and Yusef-Zadeh et al. (2009), Kendrew • Cloud #2. This cloud, centered at l = 359.47◦ ,
et al. (2013) confirm the existence of early-stage star for- b = −0.15◦ , is located to the south of Cloud #1,
mation within the EGO, driving multiple jet-like proto- bordered by the high turbulent-to-ordered ratio
stellar outflows, which is detailed Crowe et al. (2024). “ridge” to the north and the 2.5-4 keV X-ray emis-
These magnetic field orientations can be due to the sion source to the west (orange contours in Fig-
cloud being sheared along its long dimension. Alter- ure 17). The magnetic field within Cloud #2 is
natively, the head-tail cloud is moving as a unit with uniformly perpendicular to the Galactic plane and
constant velocity, and the magnetic field lines threading to the magnetic field orientation observed in Cloud
through that cloud, originally roughly perpendicular to #1. However, the magnetic field transition from
the cloud, can be deformed and dragged into their ob- Cloud #1 to Cloud #2 takes place toward the high
served orientations by the bulk motion of the cloud, into turbulent-to-ordered ratio “ridge,” thus obscuring
which the field lines are frozen. Towards the southeast of the transition from our perspective. We note ad-
the EGO, the tail forms a coherent structure continuous ditionally that the SiO emission in Figure 9 is ex-
with the −65 km s−1 cloud, as depicted in Figure 8. clusively found in Cloud #2, likely supporting the
The JCMT/POL2 polarimetric study surveyed this re- case that the magnetic field systems in these listed
gion at 850 µm (Lu et al. 2024; shown in orange in the clouds are segregated.
lower left panel of Figure 16). We note that the 850
µm magnetic field pseudovector orientations near the • Cloud #4. Centered at l = 359.425◦ , b =
“head” generally agree with those of Paré et al. (2024). −0.17◦ , Cloud #4 is separated from Cloud #2
However, in the body of the −65 km s−1 cloud near above b = −0.15◦ but merges with Cloud #2 be-
l = 359.47◦ , b = −0.16◦ , there are several 850 µm mag- low this latitude. We note that the far-IR “gulf”
netic field detections which appear roughly parallel to approximately located at l = 359.43◦ , b = −0.14◦
the Galactic plane and perpendicular to those detected exhibits an intriguing spatial anti-correlation with
at 214 µm. This misalignment could occur if the po- the 2.5-4 keV X-ray emission (orange contours in
larimetric surveys at different wavelengths are detecting Figure 17). Paré et al. (2024) argues that the
emission from different temperature regimes within the curved magnetic field morphology of Cloud #4
clouds or different components of the same cloud along can possibly be ascribed to winds associated with
a given line of sight. winds associated with star formation in the afore-
Furthermore, we outline a few intriguing morphologi- mentioned 870 µm source AGAL 359.474-0.152
cal associations between the 214 µm magnetic field pseu- (Contreras et al. 2013).
dovectors and several observed structures in the −65
We note that the observation we provided in this sec-
km s−1 cloud, shown in Figure 17. The magnetic field
tion only pertains to the Paré et al. (2024) 214 µm mag-
within the −65 km s−1 cloud is separable into three dis-
netic field measurements. For instance, we do not take
tinct parts, whose spatial extent is well represented by
into account the observed JCMT/POL2 pseudovectors
regions #1, #2, and #4 of the −65 km s−1 cloud in
in Cloud #2 (Lu et al. 2024). However, we note again
Figure 4. In this subsection, we continue to adopt this
that the JCMT/POL2 measurements may be sampling
naming scheme for our discussion, and in the following,
from a different temperature regime or another cloud
we summarize the magnetic field morphology of each
along the same line of sight. A multi-wavelength dust
polarimetry measurement that covers a larger area is
24

Figure 16. The magnetic field pseudovector representation from FIREPLACE coinciding with the head-tail cloud and the
−65 km s−1 cloud, displayed at wavelengths of 8 µm (Stolovy et al. 2006), 250 µm (Molinari et al. 2010), 850 µm (Pierce-Price
et al. 2000), and 20 cm (Heywood et al. 2022). The orange pseudovectors in the lower left panel are the 850 µm magnetic
field pseudovectors from Lu et al. (2024). The purple rectangle in the lower right panel denotes the extent of source AGAL
359.474-0.152 detected in the 870 µm ATLASGAL survey (Contreras et al. 2013).

warranted to properly interpret the observational differ- • Through a modified DCF analysis of individual
ences at the two wavelengths. clouds, we found the sky-plane magnetic field
strength to be 30 µG−1 mG in Sgr C. Regions with
6. CONCLUSION higher turbulence typically have weaker magnetic
This paper presents a detailed analysis of the 214 field strength, such as the head-tail cloud and near
µm magnetic field pseudovectors from the Paré et al. the Sgr C H II region. The dominant determinant
(2024) dust polarimetry results and their relationship of the plane-of-sky magnetic field strength is the
with prominent structures in the Sgr C complex. local turbulence.
We summarize our findings as follows:
• The expansion of the Sgr C H II region causes
• As was reported in the earlier FIREPLACE pa- extensive turbulent behaviour in its vicinity, as
pers (Butterfield et al. 2024a; Paré et al. 2024, shown by our DCF analysis. We found that the
2025), We find that the polarization fraction ex- H II region’s interior has been cleared of cold dust
hibits an anti-correlation with 214 µm intensity. content since no far-IR emission is seen there,
As such, the polarization fraction on the boundary thereby creating a cavity devoid of significant 214
of clouds is typically higher than in the centre of µm magnetic field pseudovectors. Additionally,
clouds. The higher polarization fraction is likely a we found that the 214 magnetic field orientation
result of a combination of two effects: 1) decreased is largely tangential to the [C II] shell surround-
turbulence towards the boundary of clouds, as in- ing the H II region found by Riquelme et al. (In-
dicated by our angular dispersion analysis and 2) preparation), and briefly shown in Figure 14. The
by the likelihood that the multiplicity of different angular dispersion analysis again confirms the el-
magnetic domains within the cloud is likely to be evated turbulence in this shell, accompanied by
smaller toward the edges of clouds. existing chemical shock tracers. We note an in-
 25

Figure 17. Various morphological features within the −65 km s−1 cloud. The yellow segments are the 214 µm magnetic field
pseudovectors from Paré et al. (2024), superposed upon the Herschel/SPIRE 250 µm intensity image. The green contours are
the ⟨B2t ⟩/⟨B20 ⟩ ratio at values of (0.1, 0.6, 1.2). The red contours are the 2.5-4 keV X-ray emission from Wang (2021) at (5000,
7000) counts s−1 arcmin−2 . The cyan outlines indicate the extent of the Sgr C H II region and the Sgr C NTF. The purple
square denotes the extent of source AGAL 359.474-0.152 detected in the 870 µm ATLASGAL survey (Contreras et al. 2013).

triguing diffuse source of X-ray emission to the coupled with detailed molecular line surveys, are
southwest of the Sgr C H II region, which war- needed to elucidate the dynamics of these struc-
rants further study as potentially resulting from tures and how the magnetic field and the gas are
past activity in the Sgr C region. interacting.

• The magnetic field morphology near the base

7. ACKNOWLEDGEMENTS
of the Sgr C NTF is consistent with the hy-
potheses that either magnetic field line reconnec- This work is primarily based on observations made
tion or diffusive shock acceleration at the ioniza- with the NASA/DLR Stratospheric Observatory for In-
tion/compression front of the H II region are re- frared Astronomy (SOFIA). SOFIA was jointly oper-
sponsible for the relativistic electrons that illu- ated by the Universities Space Research Association,
minate the NTF with their synchrotron emission. Inc. (USRA), under NASA contract NNA17BF53C, and
The field direction in the −90 km s−1 cloud is also the Deutsches SOFIA Institut (DSI) under DLR con-
found to be parallel to the filament. tract 50 OK 2002 to the University of Stuttgart. Fi-
nancial support for this work was provided by NASA
• We have examined the morphological correspon- through award #09-0054 through USRA. This work has
dences between the 214 and 850 µm magnetic field made extensive use of NASA’s Astrophysics Data Sys-
pseudovectors and numerous radio features, such tem (http://ui.adsabs.harvard.edu/) and the arXiv e-
as Source C, FIR-4, and the G359.33-0.15 fila- Print service (http://arxiv.org).
ments. From these comparisons, we have con- RJZ and MRM are supported by NASA through
cluded that many radio structures in the Sgr award #09-0054 administered as a subgrant to UCLA.
C region exhibit morphological correspondences RJZ thanks Keila Cunha e Silva and Joan-Emma Shea
with the observed magnetic field direction. Fu- for generously sharing their computing resources, and Q.
ture multi-wavelength dust polarimetry surveys, Daniel Wang for kindly providing his data. Finally, RJZ
26

wishes to acknowledge all community members of Mead- Sioux people. We value the opportunity to learn, live,
owridge School, located in British Columbia, Canada, and share research and educational experiences on these
for their home-like hospitality and helpful conversations traditional lands.
while much of this work was completed.
Facility: SOFIA, Chandra, Herschel, JCMT,
The authors at UCLA Department of Physics &
MeerKAT, Mopra, Spitzer
Astronomy acknowledge our presence on the tra-
ditional, ancestral, and unceded territory of the Software: Astropy (Astropy Collaboration et al.
Gabrielino/Tongva peoples. Much of this work is com- 2013, 2018), CASA (CASA Team et al. 2022), cmo-
pleted at Meadowridge School, located on the ances- cean (Thyng et al. 2016), emcee (Foreman-Mackey
tral and unceded territories of the Katzie, Kwantlen, et al. 2013), Matplotlib (Hunter 2007), Numpy (Har-
and Coast Salish Peoples, as well as the University of ris et al. 2020), Pandas (pandas development team
Chicago, located on the ancestral and unceded territo- 2020), polBpy (Guerra et al. 2024), Scipy (Virtanen
ries of the Kickapoo, Peoria, Potawatomi, Miami, and et al. 2020)

APPENDIX

A.1. AUTOCORRELATION FUNCTION OF SGR C CLOUDS

In this section, we present the details of the autocorrelation function computation and effective cloud depth. We
show the shapes of the autocorrelation function calculated for the individual Sgr C clouds in Figure A.1. We refer
the readers to Figure 4 for the extent of individual regions. The cloud depth is positively correlated with the size and
uniformity of a region. The autocorrelation functions typically exhibit a uniform decline, except for FIR-4 where a
“bump” near l ≈ 1.4′ is observed (central panel). The shape of the FIR-4 autocorrelation function is likely a result of
its size. The vertical extent of FIR-4 is roughly 1.4′ or 0.02◦ , therefore the autocorrelation function likely displays the
increase of polarized flux towards the periphery of FIR-4.

REFERENCES
Aitken, D. K., Gezari, D., Smith, C. H., McCaughrean, M., Banda-Barragán, W. E., Federrath, C., Crocker, R. M., &
& Roche, P. F. 1991, ApJ, 380, 419, doi: 10.1086/170600 Bicknell, G. V. 2018, MNRAS, 473, 3454
Aitken, D. K., Smith, C. H., Moore, T. J. T., & Roche, Banda-Barragán, W. E., Parkin, E. R., Federrath, C.,
P. F. 1998, MNRAS, 299, 743, Crocker, R. M., & Bicknell, G. V. 2016, MNRAS, 455,
doi: 10.1046/j.1365-8711.1998.01807.x 1309
Akshaya, M. S., & Hoang, T. 2024, MNRAS, Battersby, C., Bally, J., Ginsburg, A., et al. 2011, A&A,
doi: 10.1093/mnras/stae1464 535, A128, doi: 10.1051/0004-6361/201116559
Andersson, B. G., Lazarian, A., & Vaillancourt, J. E. 2015, Battersby, C., Walker, D. L., Barnes, A., et al. 2024, arXiv
ARA&A, 53, 501, e-prints, arXiv:2410.17334,
doi: 10.1146/annurev-astro-082214-122414 doi: 10.48550/arXiv.2410.17334
Arendt, R. G., Staguhn, J., Dwek, E., et al. 2019, ApJ, 885, Boldyrev, S., & Yusef-Zadeh, F. 2006, ApJL, 637, L101
71, doi: 10.3847/1538-4357/ab451c Bryant, A., & Krabbe, A. 2021, NewAR, 93, 101630,
Arthur, S. J., Henney, W. J., Mellema, G., de Colle, F., & doi: 10.1016/j.newar.2021.101630
Vázquez-Semadeni, E. 2011, MNRAS, 414, 1747, Butterfield, N. O., Chuss, D. T., Guerra, J. A., et al. 2024a,
doi: 10.1111/j.1365-2966.2011.18507.x ApJ, 963, 130, doi: 10.3847/1538-4357/ad12b9
Astropy Collaboration, Robitaille, T. P., Tollerud, E. J., Butterfield, N. O., Guerra, J. A., Chuss, D. T., et al. 2024b,
et al. 2013, A&A, 558, A33, ApJ, 968, 63, doi: 10.3847/1538-4357/ad402c
doi: 10.1051/0004-6361/201322068 Cabral, B., & Leedom, L. C. 1993, in Proceedings of the
Astropy Collaboration, Price-Whelan, A. M., Sipőcz, B. M., 20th annual conference on Computer graphics and
et al. 2018, AJ, 156, 123, doi: 10.3847/1538-3881/aabc4f interactive techniques, ACM, 263–270
Bally, J., Crowe, S., Fedriani, R., et al. 2024, arXiv e-prints, Carey, S. J., Noriega-Crespo, A., Mizuno, D. R., et al.
arXiv:2412.10983, doi: 10.48550/arXiv.2412.10983 2009, PASP, 121, 76, doi: 10.1086/596581
 27

1.0
−65 km s−1 Cloud #1 (Green #1) −65 km s−1 Cloud #2 (Green #2) −65 km s−1 Cloud #3 (Green #3)
0.8 ∆0 = 0.64 ± 0.15’ ∆0 = 1.43 ± 0.20’ ∆0 = 1.03 ± 0.08’

0.6

0.4

0.2

0.0
1.0
−65 km s−1 Cloud #4 (Green #4) FIR-4 (Pink) Galactic Plane (Red)
0.8 ∆0 = 1.04 ± 0.13’ ∆0 = 0.84 ± 0.13’ ∆0 = 1.39 ± 0.22’
hP̄ 2(l)i

0.6

0.4

0.2

0.0
1.0
Source C (Yellow) Far Northeast (White) −90 km s−1 Cloud (Cyan)
0 0
0.8 ∆ = 1.41 ± 0.14’ ∆ = 1.36 ± 0.07’ ∆0 = 1.09 ± 0.12’

0.6

0.4

0.2

0.0
0 2 4 0 2 4 0 2 4
` [arcmin]

Figure A.1. The autocorrelation functions computed from individual Sgr C clouds, as shown in Figure 4. The blue curves are
the autocorrelation functions (⟨P̄ 2 (l)⟩), and the shaded region indicate the standard error. The purple horizontal line indicates
where ⟨P̄ 2 (l)⟩ = 0.5. The burgundy vertical line indicates the HWHM value of the autocorrelation function, which is equivalent
to the effective cloud depth.
28

CASA Team, Bean, B., Bhatnagar, S., et al. 2022, PASP, Hankins, M. J., Lau, R. M., Radomski, J. T., et al. 2020,
134, 114501, doi: 10.1088/1538-3873/ac9642 ApJ, 894, 55, doi: 10.3847/1538-4357/ab7c5d
Chandrasekhar, S., & Fermi, E. 1953, ApJ, 118, 113, Harper, D. A., Runyan, M. C., Dowell, C. D., et al. 2018,
doi: 10.1086/145731 Journal of Astronomical Instrumentation, 7, 1840008,
Chuss, D. T., Davidson, J. A., Dotson, J. L., et al. 2003, doi: 10.1142/S2251171718400081
ApJ, 599, 1116, doi: 10.1086/379538 Harris, C. R., Millman, K. J., van der Walt, S. J., et al.
Chuss, D. T., Andersson, B. G., Bally, J., et al. 2019, ApJ, 2020, Nature, 585, 357, doi: 10.1038/s41586-020-2649-2
872, 187, doi: 10.3847/1538-4357/aafd37 Henshaw, J. D., Barnes, A. T., Battersby, C., et al. 2023, in
Compiègne, M., Flagey, N., Noriega-Crespo, A., et al. 2010, Astronomical Society of the Pacific Conference Series,
ApJL, 724, L44, doi: 10.1088/2041-8205/724/1/L44 Vol. 534, Protostars and Planets VII, ed. S. Inutsuka,
Contreras, Y., Schuller, F., Urquhart, J. S., et al. 2013, Y. Aikawa, T. Muto, K. Tomida, & M. Tamura, 83,
A&A, 549, A45, doi: 10.1051/0004-6361/201220155 doi: 10.48550/arXiv.2203.11223
Cotera, A. S., Hankins, M. J., Bally, J., et al. 2024, ApJ, Heywood, I., Rammala, I., Camilo, F., et al. 2022, ApJ,
973, 110, doi: 10.3847/1538-4357/ad55f2 925, 165, doi: 10.3847/1538-4357/ac449a
Crocker, R. M., Jones, D. I., Melia, F., Ott, J., & Hildebrand, R. H., Kirby, L., Dotson, J. L., Houde, M., &
Protheroe, R. J. 2010, Nature, 463, 65 Vaillancourt, J. E. 2009, ApJ, 696, 567,
Crowe, S., Fedriani, R., Tan, J. C., et al. 2024, arXiv doi: 10.1088/0004-637X/696/1/567
e-prints, arXiv:2410.09253, Houde, M., Hull, C. L. H., Plambeck, R. L., Vaillancourt,
doi: 10.48550/arXiv.2410.09253 J. E., & Hildebrand, R. H. 2016, ApJ, 820, 38,
Davis, L. 1951, Physical Review, 81, 890,
doi: 10.3847/0004-637X/820/1/38
doi: 10.1103/PhysRev.81.890.2
Houde, M., Vaillancourt, J. E., Hildebrand, R. H.,
Downes, D., & Maxwell, A. 1966, ApJ, 146, 653,
Chitsazzadeh, S., & Kirby, L. 2009, ApJ, 706, 1504,
doi: 10.1086/148943
doi: 10.1088/0004-637X/706/2/1504
Ferrière, K. 2009, A&A, 505, 1183,
Hsieh, P.-Y., Koch, P. M., Kim, W.-T., et al. 2018, ApJ,
doi: 10.1051/0004-6361/200912617
862, 150, doi: 10.3847/1538-4357/aacb27
Ferrière, K., Gillard, W., & Jean, P. 2007, A&A, 467, 611,
Hunter, J. D. 2007, Computing in Science & Engineering, 9,
doi: 10.1051/0004-6361:20066992
90, doi: 10.1109/MCSE.2007.55
Foreman-Mackey, D., Conley, A., Meierjurgen Farr, W.,
Jones, P. A., Burton, M. G., Cunningham, M. R., et al.
et al. 2013, emcee: The MCMC Hammer, Astrophysics
2012, MNRAS, 419, 2961,
Source Code Library, record ascl:1303.002
doi: 10.1111/j.1365-2966.2011.19941.x
Forster, J. R., & Caswell, J. L. 2000, ApJ, 530, 371,
Kendrew, S., Ginsburg, A., Johnston, K., et al. 2013, ApJL,
doi: 10.1086/308347
775, L50, doi: 10.1088/2041-8205/775/2/L50
Geballe, T. R., Lambrides, E., Schlegelmilch, B., et al.
Kruijssen, J. M. D., Longmore, S. N., Elmegreen, B. G.,
2019, ApJ, 872, 103, doi: 10.3847/1538-4357/aafedc
et al. 2014, MNRAS, 440, 3370
Gordon, M. S., Lopez-Rodriguez, E., Andersson, B. G.,
et al. 2018, arXiv e-prints, arXiv:1811.03100, Krumholz, M. R., & Kruijssen, J. M. D. 2015, MNRAS,
doi: 10.48550/arXiv.1811.03100 453, 739, doi: 10.1093/mnras/stv1670
GRAVITY Collaboration, Abuter, R., Amorim, A., et al. Krumholz, M. R., Stone, J. M., & Gardiner, T. A. 2007,
2019, A&A, 625, L10, doi: 10.1051/0004-6361/201935656 ApJ, 671, 518, doi: 10.1086/522665
Guan, Y., Clark, S. E., Hensley, B. S., et al. 2021, ApJ, Lang, C. C., Goss, W. M., Cyganowski, C., & Clubb, K. I.
920, 6, doi: 10.3847/1538-4357/ac133f 2010, ApJS, 191, 275, doi: 10.1088/0067-0049/191/2/275
Guerra, J. A., Chuss, D. T., Dowell, C. D., et al. 2021, LaRosa, T. N., Brogan, C. L., Shore, S. N., et al. 2005,
ApJ, 908, 98, doi: 10.3847/1538-4357/abd6f0 ApJL, 626, L23, doi: 10.1086/431647
Guerra, J. A., Chuss, D. T., & Paré, D. 2024, polBpy: a Liszt, H. S. 1985, ApJL, 293, L65, doi: 10.1086/184492
Python package for the analysis of dust polarimetric Liszt, H. S., & Spiker, R. W. 1995, ApJS, 98, 259,
observations, latest, Zenodo, doi: 10.1086/192160
doi: 10.5281/zenodo.11414008.2024 Lu, X., Liu, J., Pillai, T., et al. 2024, ApJ, 962, 39,
Guerra, J. A., Lopez-Rodriguez, E., Chuss, D. T., doi: 10.3847/1538-4357/ad1395
Butterfield, N. O., & Schmelz, J. T. 2023, AJ, 166, 37, Mangilli, A., Aumont, J., Bernard, J. P., et al. 2019, A&A,
doi: 10.3847/1538-3881/acdacd 630, A74, doi: 10.1051/0004-6361/201935072
 29

Molinari, S., Swinyard, B., Bally, J., et al. 2010, PASP, Rosner, R., & Bodo, G. 1996, ApJL, 470, L49,
122, 314, doi: 10.1086/651314 doi: 10.1086/310286
Molinari, S., Bally, J., Noriega-Crespo, A., et al. 2011, Roy, S. 2003, A&A, 403, 917,
ApJL, 735, L33, doi: 10.1088/2041-8205/735/2/L33 doi: 10.1051/0004-6361:20030460
Molinari, S., Schisano, E., Elia, D., et al. 2016, A&A, 591, Sadavoy, S. I., Di Francesco, J., Johnstone, D., et al. 2013,
A149, doi: 10.1051/0004-6361/201526380 ApJ, 767, 126, doi: 10.1088/0004-637X/767/2/126
Morris, M. 1993, ApJ, 408, 496, doi: 10.1086/172607 Salas, J. M., Naoz, S., & Morris, M. R. 2020, arXiv e-prints,
Morris, M. 1996, in Unsolved Problems of the Milky Way, arXiv:2010.04170, doi: 10.48550/arXiv.2010.04170
ed. L. Blitz & P. J. Teuben, Vol. 169, 247 Schuller, F., Menten, K. M., Contreras, Y., et al. 2009,
Morris, M. 2006, in Journal of Physics Conference Series, A&A, 504, 415, doi: 10.1051/0004-6361/200811568
Vol. 54, Journal of Physics Conference Series, 1–9, Serabyn, E., & Morris, M. 1994, ApJL, 424, L91,
doi: 10.1088/1742-6596/54/1/001 doi: 10.1086/187282
Morris, M., Davidson, J. A., Werner, M., et al. 1992, ApJL, Sormani, M. C., Sobacchi, E., Fragkoudi, F., et al. 2018,
399, L63, doi: 10.1086/186607 MNRAS, 481, 2, doi: 10.1093/mnras/sty2246
Morris, M., & Serabyn, E. 1996, ARA&A, 34, 645, Sormani, M. C., Sobacchi, E., & Sanders, J. L. 2024,
doi: 10.1146/annurev.astro.34.1.645 MNRAS, 528, 5742, doi: 10.1093/mnras/stae082
Morris, M., & Yusef-Zadeh, F. 1985, AJ, 90, 2511, Sormani, M. C., Tress, R. G., Glover, S. C. O., et al. 2020,
MNRAS, 497, 5024, doi: 10.1093/mnras/staa1999
doi: 10.1086/113955
Stolovy, S., Ramirez, S., Arendt, R. G., et al. 2006, in
Morris, M. R. 2015, in Lessons from the Local Group: A
Journal of Physics Conference Series, Vol. 54, Journal of
Conference in honor of David Block and Bruce
Physics Conference Series, 176–182,
Elmegreen, 391, doi: 10.1007/978-3-319-10614-4 32
doi: 10.1088/1742-6596/54/1/030
Morris, M. R. 2023, in Physics and Chemistry of Star
Thomas, T., Pfrommer, C., & Enßlin, T. 2020, ApJL, 890,
Formation: The Dynamical ISM Across Time and Spatial
L18
Scales, ed. V. Ossenkopf-Okada, R. Schaaf, I. Breloy, &
Thyng, K. M., Greene, C. A., Hetland, R. D., Zimmerle,
J. Stutzki, 49, doi: 10.48550/arXiv.2301.13469
H. M., & DiMarco, S. F. 2016, Oceanography, 29,
Narayanan, D., Smith, J. D. T., Hensley, B. S., et al. 2023,
doi: 10.5670/oceanog.2016.66
ApJ,, 951, 100, doi: 10.3847/1538-4357/accf8d
Tress, R. G., Sormani, M. C., Girichidis, P., et al. 2024,
Novak, G., Dotson, J. L., Dowell, C. D., et al. 2000, ApJ,
arXiv e-prints, arXiv:2403.13048,
529, 241, doi: 10.1086/308231
doi: 10.48550/arXiv.2403.13048
Novak, G., Chuss, D. T., Renbarger, T., et al. 2003, ApJL,
Tsuboi, M., Inoue, M., Handa, T., Tabara, H., & Kato, T.
583, L83, doi: 10.1086/368156
1985, PASJ, 37, 359
Odenwald, S. F., & Fazio, G. G. 1984, ApJ, 283, 601,
Uchida, K., Morris, M., & Serabyn, E. 1990, ApJ, 351, 443,
doi: 10.1086/162345
doi: 10.1086/168481
pandas development team, T. 2020, pandas-dev/pandas: Uchida, K. I., Morris, M., Serabyn, E., & Guesten, R. 1996,
Pandas, latest, Zenodo, doi: 10.5281/zenodo.3509134 ApJ, 462, 768
Paré, D., Butterfield, N. O., Chuss, D. T., et al. 2024, ApJ, Uchida, K. I., Morris, M. R., Serabyn, E., & Bally, J. 1994,
969, 150, doi: 10.3847/1538-4357/ad4462 ApJ, 421, 505, doi: 10.1086/173667
Paré, D. M., Chuss, D. T., Karpovich, K., et al. 2025, ApJ, Vaillancourt, J. E. 2002, ApJS, 142, 53, doi: 10.1086/341398
978, 28, doi: 10.3847/1538-4357/ad9586 Virtanen, P., Gommers, R., Oliphant, T. E., et al. 2020,
Parker, E. N. 1979, Cosmical magnetic fields. Their origin Nature Methods, 17, 261, doi: 10.1038/s41592-019-0686-2
and their activity Wang, Q. D. 2021, MNRAS, 504, 1609,
Pierce-Price, D., Richer, J. S., Greaves, J. S., et al. 2000, doi: 10.1093/mnras/stab801
ApJL, 545, L121, doi: 10.1086/317884 Wardle, M., & Konigl, A. 1990, ApJ, 362, 120
Pillai, T., Kauffmann, J., Tan, J. C., et al. 2015, ApJ, 799, Werner, M. W., Davidson, J. A., Morris, M., et al. 1988,
74, doi: 10.1088/0004-637X/799/1/74 ApJ, 333, 729, doi: 10.1086/166781
Planck Collaboration, Ade, P. A. R., Aghanim, N., et al. Yusef-Zadeh, F., Arendt, R. G., Wardle, M., et al. 2022a,
2015, A&A, 576, A104, MNRAS, 515, 3059, doi: 10.1093/mnras/stac1696
doi: 10.1051/0004-6361/201424082 Yusef-Zadeh, F., Arendt, R. G., Wardle, M., Heywood, I.,
Riquelme, D., Güsten, R., Harris, A., et al. In-preparation & Cotton, W. 2022b, MNRAS, 517, 294
30

Yusef-Zadeh, F., Arendt, R. G., Wardle, M., et al. 2022c, Yusef-Zadeh, F., Muno, M., Wardle, M., & Lis, D. C. 2007,
ApJ, 656, 847
ApJL, 925, L18, doi: 10.3847/2041-8213/ac4802
Yusef-Zadeh, F., Roberts, D. A., Goss, W. M., Frail, D. A.,
Yusef-Zadeh, F., & Königl, A. 2004, ASPC, 322, 201 & Green, A. J. 1999, ApJ, 512, 230, doi: 10.1086/306769
Yusef-Zadeh, F., & Wardle, M. 2019, MNRAS, 490, L1
Yusef-Zadeh, F., Morris, M., & Chance, D. 1984, Nature, Yusef-Zadeh, F., Hewitt, J. W., Arendt, R. G., et al. 2009,
310, 557, doi: 10.1038/310557a0 ApJ, 702, 178, doi: 10.1088/0004-637X/702/1/178