JWST Observations of Starbursts: Polycyclic Aromatic Hydrocarbon Emission at the Base of the M82 Galactic Wind

Abstract

We present new observations of the central 1 kpc of the M82 starburst obtained with the James Webb Space Telescope near-infrared camera instrument at a resolution θ ∼ 0.″05–0.″1 (∼1–2 pc). The data comprises images in three mostly continuum filters (F140M, F250M, and F360M), and filters that contain [Fe ii ] (F164N), H 2 v = 1 → 0 (F212N), and the 3.3 μ m polycyclic aromatic hydrocarbon (PAH) feature (F335M). We find prominent plumes of PAH emission extending outward from the central starburst region, together with a network of complex filamentary substructures and edge-brightened bubble-like features. The structure of the PAH emission closely resembles that of the ionized gas, as revealed in Paschen α and free–free radio emission. We discuss the origin of the structure, and suggest the PAHs are embedded in a combination of neutral, molecular, and photoionized gas.

Full text

JWST Observations of Starbursts: Polycyclic Aromatic Hydrocarbon Emission at the
Base of the M82 Galactic Wind
Alberto D. Bolatto
1,2
, Rebecca C. Levy
3,34
, Elizabeth Tarantino
4
, Martha L. Boyer
4
, Deanne B. Fisher
5,6
,
Serena A. Cronin
1
, Adam K. Leroy
7
, Ralf S. Klessen
8,9
, J. D. Smith
10
, Danielle A. Berg
11
, Torsten Böker
12
,
Leindert A. Boogaard
13
, Eve C. Ostriker
14
, Todd A. Thompson
7,15,16
, Juergen Ott
17
, Laura Lenkić
18,19
,
Laura A. Lopez
7,15
, Daniel A. Dale
20
, Sylvain Veilleux
1,2
, Paul P. van der Werf
21
, Simon C. O. Glover
8
,
Karin M. Sandstrom
22
, Evan D. Skillman
23
, John Chisholm
11
, Vicente Villanueva
1,24
, Thomas S.-Y. Lai
25
,
Sebastian Lopez
7,15
, Elisabeth A. C. Mills
26
, Kimberly L. Emig
27,35
, Lee Armus
25
, Divakara Mayya
28
,
David S. Meier
17,29
, Ilse De Looze
30
, Rodrigo Herrera-Camus
24
, Fabian Walter
13
, Mónica Relaño
31
,
Hannah B. Koziol
22
, Joshua Marvil
17
, María J. Jiménez-Donaire
32,33
, and Paul Martini
7,15
1
Department of Astronomy, University of Maryland, College Park, MD 20742, USA; bolatto@umd.edu
2
Joint Space-Science Institute, University of Maryland, College Park, MD 20742, USA
3
Steward Observatory, University of Arizona, Tucson, AZ 85721, USA
4
Space Telescope Science Institute, 3700 San Martin Drive, Baltimore, MD 21218, USA
5
Centre for Astrophysics and Supercomputing, Swinburne University of Technology, Hawthorn, VIC 3122, Australia
6
ARC Centre of Excellence for All Sky Astrophysics in 3 Dimensions (ASTRO 3D), Hawthorn, VIC 3122, Australia
7
Department of Astronomy, The Ohio State University, Columbus, OH 43210, USA
8
Universität Heidelberg, Zentrum für Astronomie, Institut für Theoretische Astrophysik, Albert-Ueberle-Str. 2, D-69120 Heidelberg, Germany
9
Universität Heidelberg, Interdisziplinäres Zentrum für Wissenschaftliches Rechnen, Im Neuenheimer Feld 205, D-69120 Heidelberg, Germany
10
Ritter Astrophysical Research Center, University of Toledo, Toledo, OH 43606, USA
11
Department of Astronomy, The University of Texas at Austin, 2515 Speedway, Stop C1400, Austin, TX 78712, USA
12
European Space Agency, c/o STScI, 3700 San Martin Drive, Baltimore, MD 21218, USA
13
Max-Planck-Institut für Astronomie, Königstuhl 17, 69120 Heidelberg, Germany
14
Department of Astrophysical Sciences, Princeton University, Princeton, NJ 08544, USA
15
Center for Cosmology and Astro-Particle Physics, The Ohio State University, Columbus, OH 43210, USA
16
Department of Physics, The Ohio State University, Columbus, OH 43210, USA
17
National Radio Astronomy Observatory, P.O. Box O, 1003 Lopezville Road, Socorro, NM 87801, USA
18
Stratospheric Observatory for Infrared Astronomy, NASA Ames Research Center, Mail Stop 204-14, Moffett Field, CA 94035, USA
19
Jet Propulsion Laboratory, California Institute of Technology, 4800 Oak Grove Drive, Pasadena, CA 91109, USA
20
Department of Physics & Astronomy, University of Wyoming, Laramie, WY 82071, USA
21
Leiden Observatory, Leiden University, P.O. Box 9513, 2300 RA Leiden, The Netherlands
22
Department of Astronomy & Astrophysics, University of California, San Diego, La Jolla, CA 92093, USA
23
Minnesota Institute for Astrophysics, University of Minnesota, Minneapolis, MN 55455, USA
24
Departamento de Astronomía, Universidad de Concepción, Barrio Universitario, Concepción, Chile
25
IPAC, California Institute of Technology, 1200 East California Boulevard, Pasadena, CA 91125, USA
26
Department of Physics and Astronomy, University of Kansas, 1251 Wescoe Hall Drive, Lawrence, KS 66045, USA
27
National Radio Astronomy Observatory, 520 Edgemont Road, Charlottesville, VA 22903, USA
28
Instituto Nacional de Astrofísica, Óptica y Electrónica, Luis Enrique Erro 1, Tonantzintla 72840, Puebla, Mexico
29
New Mexico Institute of Mining and Technology, 801 Leroy Place, Socorro, NM 87801, USA
30
Sterrenkundig Observatorium, Ghent University, Krijgslaan 281—S9, B-9000 Gent, Belgium
31
Dept. Física Teórica y del Cosmos, Universidad de Granada, 18071, Granada, Spain
32
Observatorio Astronomico Nacional (IGN),C/Alfonso XII, 3, E-28014 Madrid, Spain
33
Centro de Desarrollos Tecnologicos, Observatorio de Yebes (IGN), 19141 Yebes, Guadalajara, Spain
Received 2024 January 21; revised 2024 February 28; accepted 2024 March 7; published 2024 May 17
Abstract
We present new observations of the central 1 kpc of the M82 starburst obtained with the James Webb Space
Telescope near-infrared camera instrument at a resolution θ∼005–0 1 (∼1–2pc). The data comprises images in
three mostly continuum filters (F140M, F250M, and F360M), and filters that contain [Fe II](F164N),H
2
v=1→0
(F212N), and the 3.3 μm polycyclic aromatic hydrocarbon (PAH)feature (F335M).Wefind prominent plumes of
PAH emission extending outward from the central starburst region, together with a network of complex filamentary
substructures and edge-brightened bubble-like features. The structure of the PAH emission closely resembles that
of the ionized gas, as revealed in Paschen αand free–free radio emission. We discuss the origin of the structure,
and suggest the PAHs are embedded in a combination of neutral, molecular, and photoionized gas.
Unified Astronomy Thesaurus concepts: Luminous infrared galaxies (946);Galaxy winds (626);Starburst galaxies
(1570);Dust physics (2229);Interstellar medium (847)
1. Introduction
Local starburst galaxies are characterized by fast star
formation, causing rapid consumption of their gas reservoirs
on typical timescales of ∼100 Myr. This high-activity mode
resembles the situation found at cosmic noon (z∼2), where
The Astrophysical Journal, 967:63 (15pp), 2024 May 20 https://doi.org/10.3847/1538-4357/ad33c8
© 2024. The Author(s). Published by the American Astronomical Society.
34
NSF Astronomy and Astrophysics Postdoctoral Fellow.
35
Jansky Fellow of the National Radio Astronomy Observatory.
Original content from this work may be used under the terms
of the Creative Commons Attribution 4.0 licence. Any further
distribution of this work must maintain attribution to the author(s)and the title
of the work, journal citation and DOI.
1

gas-rich, strongly star-forming galaxies dominate the cosmic
star formation budget. Therefore, it is important to understand
the physics that drive this mode of star formation, something
best done in nearby galaxies which can be studied in detail.
Differences in the physical conditions of the interstellar
medium (ISM)during starburst phases have been invoked as
drivers for variations in the stellar initial mass function (IMF,
e.g., Conroy & van Dokkum 2012), while concentrated star
formation events can also drive multiphase winds that
drastically affect the evolution of a galaxy (Heckman et al.
1990; Veilleux et al. 2005,2020), limiting gas available to fuel
supermassive black holes and grow galaxies, and playing a key
role in determining the galaxy mass function.
The higher rate of star formation per unit molecular gas mass
observed in starbursts strongly suggests actual physical
changes in the process of forming stars, not just a one-to-one
scaling of the activity with the gas reservoir. Starbursts occur
due to the concentration of large masses of (molecular)gas,
which when combined with shorter freefall timescales, produce
high star formation rates (e.g., Krumholz et al. 2012; Leroy
et al. 2015a). High gas surface densities in starbursts imply that
very high gas consumption rates are required for star formation
feedback to prevent the collapse of their gas disks (Ostriker &
Shetty 2011).
These conditions alter how star formation happens, causing
the fraction of stars formed in clusters to likely increase with
the surface density of star formation (e.g., Kruijssen 2012;
Krumholz et al. 2019). Vigorous starbursts produce super star
clusters (SSCs, e.g., Herrera et al. 2012; Whitmore et al. 2014;
Turner et al. 2015; Leroy et al. 2018; Emig et al. 2020);
massive (M
å
∼10
5
M
e
), and compact (R∼1pc)concentra-
tions of stars that are the younger cousins to globular clusters
(Whitmore et al. 2005; Portegies Zwart et al. 2010; Linden
et al. 2021). The formation of SSCs is a particularly efficient
way of converting gas into stars (Krumholz et al. 2019), likely
characteristic of starburst environments.
This highly concentrated star formation activity also results
in massive ejections of material from the galaxy into the
circumgalactic medium, which eventually quenches the
starburst and curtails the growth of the galaxy (Heckman
et al. 1990). Galactic winds are thought to regulate the growth
of galaxies and thus shape the galaxy mass function, in part
because they eject cool gas into the circumgalactic medium that
is then no longer immediately available for star formation
(although it may be reaccreted on long timescales), in part by
preventing the accretion of cosmic web gas into the galaxy halo
(e.g., Mitchell & Schaye 2022). The concentrated starburst and
the launching of a wind are intimately related, and this may
leave an imprint in their respective structures. For example,
starburst rings may produce fast biconical hot outflows
collimated along the ring axis sheathed in cooler material
(Nguyen & Thompson 2022). The mass-loss rate of the wind—
key to understanding its effect on galaxy evolution—is highly
uncertain because it requires measuring the contribution of its
different phases, especially the cooler (T10
4
K)phases
thought to carry the bulk of the mass.
The key observations we present in this study concern the
imaging of the base of the M82 wind in polycyclic aromatic
hydrocarbon (PAH)emission. PAHs are complex molecules
and/or very small dust grains that pervade the cool ISM,
absorbing UV and optical photons and reemitting their energy
in relatively broad spectral features (bands)throughout the mid-
infrared (mid-IR)spectrum. They play a key role in the heating
of the neutral gas and possibly in the formation of H
2
. The
3.3 μm PAH band is thought to be mostly due to small, neutral
PAHs (Draine & Li 2001; Maragkoudakis et al. 2020; Draine
et al. 2021). PAHs are extraordinarily resilient, being observed
in close proximity to active galactic nuclei (e.g., García-
Bernete et al. 2022). They are destroyed by shocks (Micelotta
et al. 2010a)or X-rays associated with luminous active galactic
nuclei (Xie & Ho 2022; Lai et al. 2023)and T Tauri stars
(Siebenmorgen & Krügel 2010), however, and are rapidly
destroyed in hot gas (e.g., Micelotta et al. 2010b); therefore,
they trace the cooler phases of the ISM and their spectrum
provides key information on the physical conditions in those
phases.
M82 is part of the M81 group located 3.6 Mpc away
(Freedman et al. 1994; Dalcanton et al. 2009), and represents
the prototypical example of an interaction-driven global starburst
(Yun et al. 1994; Mayya et al. 2006; de Blok et al. 2018)in a
dwarf galaxy (the mass of M82 is approximately 10
10
M
e
,Greco
et al. 2012), affording a uniquely detailed view of the
astrophysical processes at the core of the starburst phenomenon.
Modeling of near- to mid-IR spatially resolved spectroscopy
suggests that M82 experienced a first starburst episode in its
central 500 pc between 8 and 15 Myr ago, peaking at
160 M
e
yr
−1
, followed by a second, somewhat more compact
burst occurring in a circumnuclear ring and stellar bar 4–6Myr
ago and peaking at 40 M
e
yr
−1
(Colbertetal.1999;Förster
Schreiber et al. 2003). The far-infrared (far-IR)luminosity of
4×10
10
L
e
(Herrera-Camus et al. 2018)corresponds to a current
star formation rate of SFR ;12 M
e
yr
−1
, broadly consistent with
other recent estimates (Förster Schreiber et al. 2003).
M82 not only hosts a large number of compact, massive
clusters that represent a significant fraction of the star formation
activity (O’Connell et al. 1995; McCrady et al. 2003;Meloetal.
2005; Smith et al. 2006; McCrady & Graham 2007; Mayya et al.
2008), but it also has a powerful multiphase wind. This wind is
visible in X-rays (Strickland et al. 1997; Lopez et al. 2020),
hydrogen recombination (Shopbell & Bland-Hawthorn 1998;
Lokhorst et al. 2022), far-IR dust and line emission (Engelbracht
et al. 2006; Contursi et al. 2013; Beirão et al. 2015;Levyetal.
2023), warm molecular hydrogen (Veilleux et al. 2009; Beirão
et al. 2015), and cold molecular and neutral atomic gas (Walter
et al. 2002; Leroy et al. 2015b; Martini et al. 2018; Krieger et al.
2021). The wind geometry corresponds to a truncated bicone
with its southern lobe approaching us, and an opening angle of
Ω
w
∼0.8πsr (Xu et al. 2023). The main ionization mechanism
of the wind is photoionization close to the starburst, with
increasing shock ionization 1–2 kpc away from it (Shopbell &
Bland-Hawthorn 1998). The wind extends at least 11 kpc away
from the galaxy (Devine & Bally 1999; Lehnert et al. 1999),and
it may reach out to 40 kpc (Lokhorst et al. 2022). Polarization
work shows the existence of a large reflection nebula in Hα
(Scarrott et al. 1991), with a polarization fraction that increases
to 30% outward, at 4 kpc from the nucleus (Yoshida et al. 2019).
Kinematic analysis suggests that some of the dust causing the
reflection 4kpc away from the galaxy center is moving at
velocities of 300–450 km s
−1
, faster than the molecular gas
velocities observed closer in Leroy et al. (2015b). The radiation
reflected appears to have been emitted by a combination of two
sources; one is the nucleus of M82 as identified at 2.2 μm(which
dominates at large distances), the other coincident with the 3 mm
continuum peak west of the nucleus (Yoshida et al. 2019).
2
The Astrophysical Journal, 967:63 (15pp), 2024 May 20 Bolatto et al.

In this manuscript we present new observations of the central
region of the M82 starburst obtained with the James Webb
Space Telescope (JWST)near-infrared camera (NIRCam)
instrument (Figure 1), at a resolution (θ∼005–0 1 ∼1–2pc)
and sensitivity unavailable before. These observations are part of
Cycle 1 GO program 1701, which also images NGC 253.
Pointing JWST to a very bright source, like a nearby starburst
nucleus, carries the risk of saturation. The observations presented
here were designed to avoid saturation in the nuclear region and
complement larger, deeper mosaics, which at the time of writing
were not yet observed. As a new, very high-resolution view of
the archetypal starburst in the near-infrared (near-IR),they
present a unique opportunity for science. We discuss the
observations and their processing in Section 2, and we present
the results on PAH emission at the base of the outflow in
Section 3, and the conclusions in Section 4. Companion papers
analyze the detailed correlations between outflow tracers (D. B.
Fisher et al. 2024, in preparation), and the identification and
properties of massive star clusters (R. C. Levy et al. 2024, in
preparation)based on these same data.
2. Observations and Data Processing
2.1. Observations
Observations were acquired by JWST as part of the Cycle 1
GO project 1701 (PI: Alberto Bolatto)using the NIRCam
instrument in a 1.5 hr long visit starting on 2022 October 17,
23:20 UT. To minimize the chance of saturation, the
observations were set up using the SUB640 subarray with the
RAPID readout pattern. The mosaics,
36
shown in Figure 2,
have a linear extent of 50″(∼870 pc)and a resolution of
005–0 1 (∼1–2pc).
Data were obtained in six NIRCam filters: F140M, F164N,
and F212N on the blue side (short wavelength channel),and
F250M, F335M, and F360M on the red side (long wavelength
channel). Color combinations of most of these filters are
presented in Figures 3and 4. The F164N and F212N narrowband
filters contain a bright fine-structure transition from [Fe II]and the
v=1–0 vibrational transition of H
2,
respectively, while the
F335M filter is dominated by a bright 3.3 μmPAHfeature.The
mosaic consisted of four primary dithers in INTRAMODULE-
BOX mode and four small dithers in SMALL-GRID-DITHER
mode, with a depth of six groups per integration and a total
exposure of just under 470 s per filter.
2.2. Pipeline Processing
We processed the NIRCam uncalibrated data products with
the JWST pipeline version 1.9.6 (Bushouse et al. 2023)and
CRDS context jwst_1077.pmap. For stage 1 of the pipeline, we
used default parameters except for the following: we used
frame0 to recover emission from sources that are saturated in
the first integration group in the ramp fitting step (suppres-
s_one_group=False). In the jump step, we set expan-
d_large_events=True in order to flag the “snowball”
features that occur due to large cosmic-ray events (e.g., Rieke
et al. 2023). We employed all the default parameters for stage 2
processing. In between stage 2 and 3, we applied an additional
correction to the
*
_cal.fits files for the two narrowband
filters, F212N and F164N, which have the least signal and a
comparatively larger contribution from 1/fnoise from the
NIRCam detectors that appears as bright striping features. We
adapted the algorithm presented in Willott et al. (2022)for use
with the SUB640 subarray and removed the 1/fnoise by
Figure 1. Advanced Camera for Surveys image of M82 (Mutchler et al. 2007)with the JWST NIRCam central mosaic footprint overlaid for reference. We use F658N
(Hα), F555W, and F435W for red, green, and blue, respectively. These JWST observations cover the central region of the starburst and the base of the wind bicone.
They are part of a larger mosaic imaging the wind and the starburst, which will be acquired in 2024.
36
Automatic pipeline processed MAST mosaics can be found at
doi:10.17909/cwtn-nh63.
3
The Astrophysical Journal, 967:63 (15pp), 2024 May 20 Bolatto et al.

masking bright sources and subtracting the median of each
column.
Aligning the small NIRCam mosaics for M82 was challen-
ging, due to the lack of stars cataloged by the Gaia observatory
within the field of view and the small number of bright sources
outside of the plane of the galaxy. We used the JWST/HST
Alignment Tool (JHAT; Rest et al. 2023)to align the individual
exposures. First, we aligned the F250M exposures to the stars
available in Gaia DR3 using JHAT. This filter has the least
amount of distortion issues because it uses the larger field of
view LW detector and contains little to no contamination from
PAH emission seen in the F335M and F360M filters. Then, we
created a catalog of the stars from the F250M image by
selecting pixels with a brightness between 1 and 250 MJy sr
−1
(the pixel size is 0 042). This range ensures that the catalog
includes the fainter stars outside of the plane of the galaxy,
which allows better alignment of the smaller SW detector. We
then aligned the rest of the mosaics using this catalog with
JHAT. We applied the stage 3 step in the pipeline to the aligned
files with the tweakreg step turned off, which yielded the
final mosaics used in this work. The 3σscatter of the relative
positional alignment for all other filters relative to F250M
reported by JHAT is 0 03 or better, and the relative alignment
between any two filters is typically better than 0 015. The
absolute astrometry is based on five Gaia stars which were
found to be in common with sources in F250M, so it is not
possible to judge its quality based on the scatter reported by
JHAT. In Section 3.1, however, we report the cross-identifica-
tion of supernova remnants (SNRs)in the JWST images
against very large baseline interferometry (VLBI)positions,
where we find correspondences within 0 2. Inspection of the
images suggests our absolute astrometry is of order 0 1or
better. The combination of these images with our upcoming
large-scale, deeper mosaic will enable more extensive cross-
matching against Gaia and a better estimate of the absolute
astrometric accuracy of the images.
2.3. Point-spread Function Matching
Matching the point-spread function (PSF)of the different
filters is necessary for a clean continuum subtraction. We use
the WebbPSF
37
tool version 1.2.1 to compute the PSFs of the
F250M, F335M, and F360M images. We oversampled the
PSFs to obtain the desired pixel size of 0 008. To match the
F250M and F335M PSFs to F360M, we generated convolution
kernels using pypher (Boucaud et al. 2016). We used the
default regularization parameter of 10
−4
, which stabilizes the
kernel solution by penalizing noisy high frequencies. We
experimented with changing this parameter, and it did not
affect our results in any significant manner.
Figure 2. NIRCam mosaics of the center of M82, with focal plane orientation. The F140M, F250M, and F360M are primarily continuum bands. The F164N, F212N,
and F335M are a combination of continuum and emission: F164N covers the 1.64 μm[Fe II]fine-structure transition, F212N covers H
2
vibrational emission at
2.12 μm, and the diffuse emission from the 3.3 μm aromatic feature is evident in F335M. Contamination by PAH-associated emission can also be seen in F360M.
Position offsets are referred to as 09
h
55
m
51 60, 69 40 45.
6
+¢(J2000).
37
https://webbpsf.readthedocs.io
4
The Astrophysical Journal, 967:63 (15pp), 2024 May 20 Bolatto et al.

To quantitatively check the performance of these kernels, we
adopted the Aniano et al. (2011)figure of merit (W
−
),which
focuses on negative values (see their Equation (21)).Inessence,
the lower the W
−
value, the less aggressive the kernel. Aniano
et al. (2011)find that kernels with W
−
1 are generally safe to
use, though kernels with W
−
≈1, while reasonable, are mildly
aggressive. The authors caution against using a kernel with
W
−
>1.2, which would be aggressive enough to result in
artifacts. For the F250M and F335M matching kernels, we found
W
−
=0.45 and 0.98, respectively. These values indicate that
both kernels are reasonably well behaved and safe to use, with
the F335M kernel being only moderately aggressive.
After regridding the images to a common pixel size of 0 008,
we performed a fast Fourier transform convolution of the F250M
and F335M images with the corresponding kernels in order to
match the PSFs of the 2.50, 3.35 μm images to the PSF of the
3.60 μmimage,whichhasanFWHMresolutionof0118
(2.06 pc). The clean continuum-subtracted image produced (see
next section)validates this PSF matching procedure.
2.4. Continuum Subtraction
To produce an image of the PAH emission band at 3.3 μm,
caused by the C−H stretching mode in the aromatic (i.e.,
benzene-like)rings of PAH molecules, we need to remove the
underlying continuum from the F335M image. The 3.3 μm
feature is the brightest of a complex of PAH-related spectral
features that extends redward, including a 3.4 μm feature
thought to be due to C−H vibrational modes in aliphatic
(chain-like)hydrocarbons (Joblin et al. 1996; Lai et al. 2020),
and a broad plateau at 3.47 μm, which is observed to be well
correlated with the 3.3 μm intensity and thought to be also due
to aromatic material (Hammonds et al. 2015). The intensities of
the 3.3 μm PAH, 3.4 μm aliphatic, and the 3.47 μm plateau are
observed to be tightly correlated (Lai et al. 2020). The longer-
wavelength features are also in the passband of the F360M
filter, which therefore cannot be directly used to estimate the
continuum (Sandstrom et al. 2023). In addition, the 3.3 μm
feature sits on the shoulder of a broad water ice band at
3.05 μm, which sometimes can be seen in absorption in highly
extinguished systems, such as luminous and ultraluminous IR
galaxies (e.g., Imanishi et al. 2010; Lai et al. 2020). In contrast
to F360M, we expect the F250M filter to be a relatively clean
measure of the continuum according to the existing spectrosc-
opy of M82 (Sturm et al. 2000; Förster Schreiber et al. 2001).
Assuming that the contamination of the F360M filter by PAH-
related emission is tightly correlated with the PAH-dominated
Figure 3. Three-color image with F212N, F164N, and F140M for red, green, and blue, respectively. Some dust entrained in the galactic wind is seen as faint,
elongated dark streaks against the bright background. Compact green sources are SNRs (see below). Red sources are either highly extinguished or emitting in the
vibrationally excited v=1→0H
2
, a tracer of shocks and strong UV fields (image credit A. Pagan).
5
The Astrophysical Journal, 967:63 (15pp), 2024 May 20 Bolatto et al.

emission detected in the F335M filter (see, e.g., Sandstrom et al.
2023), we can set up an iterative method to separate the PAH
and continuum contributions to the signal measured in F335M.
In essence, we apply the following algorithm:
1. Compute the continuum signal at 3.60 μm(C
360
)by
removing from the F360M measurement a fraction qof
the 3.3 μm PAH intensity where that intensity is positive.
2. Compute the continuum signal at 3.35 μm(C
335
)by
using a weighted combination of the F250M filter and the
3.60 μm continuum.
3. Compute the 3.3 μm PAH intensity by removing the 3.35
μm continuum from the F335M filter.
4. Repeat until convergence.
This corresponds to the equations,
()CqF360M PAH 1
360 335
=-´
() ()CCRF250M F250M 2
335 360
=+- ´
()CPAH F335M , 3
335 335
=-
where qand Rare constants, PAH
335
is the PAH component in
F335M, C
335
and C
360
are the continuum component for the
F335M and F360M filters, and q=0ifPAH 0. The iteration
starts from PAH
335
=0 and produces values for PAH
335
,C
335
,
and C
360
.
Here we will adopt single values for q(the ratio between the
contamination in F360M due to aromatic or aliphatic compo-
nents, and the PAH emission in F335M)and R(yielding the
continuum in F335M as a linear combination of the continuum in
F250M and F360M)appropriate for our region of interest, the
extraplanar emission. Note, however, that in principle, these are
local quantities with values that change depending on PAH
properties and the color of continuum sources. The typical value
of q, as determined using the AKARI spectroscopy of entire
(unresolved)galaxies in Lai et al. (2020),isq=0.36 (T. Lai
2024, private communication)when including all the features
discussed above that contribute to F360M as well as extended
wings of bright PAH lines such as the 6.2 μm feature (just the 3.3
μm PAH wing by itself would yield q=0.09). Our data are well
resolved spatially, and the extraplanar region of M82 contributes
little to the integrated light of the galaxy. Its conditions will be
substantially different from those in the central regions that
dominate the integrated light, so the AKARI result does not
necessarily represent it well. The wispy structure visible in
F360M (which is completely absent in the continuum-dominated
Figure 4. Three-color image with F335M, F250M, and F164N for red, green, and blue respectively. Dust throughout the base of the galactic wind lights up in the
3.35 μmfilter (red), due to PAH emission. The bright F335M emission appears to outline a truncated bicone corresponding to the edge of the outflow base,
particularly on the left-side wall. In this scenario, the elongated streaks correspond to the front and back surfaces of the bicone, seen in projection against the middle
regions north and south of the starburst (image credit A. Pagan).
6
The Astrophysical Journal, 967:63 (15pp), 2024 May 20 Bolatto et al.

F250M), however, implies that the PAH contribution is large.
This is similar to the findings of Sandstrom et al. (2023)in PAH-
dominated regions. The factor R, on the other hand, captures the
fractional contribution of C
360
to the C
335
continuum when
predicting it via a linear combination of C
250
and C
360
:forR=0,
C
335
=C
250
, while for R=1, C
335
=C
360
. If the slope of the
continuum spectral energy distribution is constant between 2.50
and 3.60 μm, we expect Rto be close to the ratio of central
wavelength differences, (3.35 −2.50)/(3.60 −2.50)≈0.77,
modulo color correction factors.
Fortunately, it is possible to arrive at optimal values for q
and Rby testing a range of values and inspecting the resulting
images. The best qfor our purposes should result in a C
360
continuum without the wispy extraplanar structures visible in
F360M (Figure 2), which are caused by PAH-related emission
in this filter, while not introducing negative wispy artifacts
caused by oversubstraction from using too large a value for q.
Similarly, the best value of Rminimizes the small-scale star
subtraction residuals in the continuum for the extraplanar
regions. Using these considerations, we obtain q=0.45 ±0.05
and R=0.55 ±0.05 (Figure 5). The value of qwe find is
comparable to the q=0.36 for entire galaxies from AKARI
spectroscopy discussed above, and somewhat smaller than
q∼0.66 obtained for very PAH-dominated lines of sight in
highly resolved galaxy observations (Sandstrom et al. 2023).
We compare our image with that obtained from the continuum
subtraction method used by Sandstrom et al. (2023),andfind
both methods produce results largely consistent with each other.
We adopt their slope of B
PAH
=1.6 and offset of A
PAH
=−0.2
for colors in PAH-dominated regions (see their Equation (4)).
While Sandstrom et al. (2023)calculate this slope using the
F300M, F335M, and F360M filters, we must substitute the
F250M filter for F300M due to availability. This substitution
should not drastically change the results, as the Sandstrom et al.
(2023)slope appears reasonable in our F335M/F250M–
F360M/F250M plane. We find a ∼10% larger intensity in the
PAH
335
than in the previous method toward the central regions
of the map, increasing to ∼30% in the outer regions. Most
importantly, the overall morphology of the extended PAH
features remains unchanged when adopting the Sandstrom et al.
(2023)approach.
2.5. Ancillary Data
A 6 GHz image from J. Marvil et al. (2024, in preparation)is
derived from data taken with the Karl G. Jansky Very Large
Array (VLA)between 2011 February and 2012 March in
antenna configurations A (5hr),B(10 hr)and C (3hr)under
project codes TDEM0010, 10C-199 and 12A-457, respec-
tively. The data were processed in CASA (CASA Team et al.
2022), using the flux-density scale of Perley & Butler (2013)
for calibration and the multiscale, multifrequency synthesis
algorithm (Rau & Cornwell 2011)for imaging. The 6 GHz
continuum image combines 2048 ×2 MHz channels spanning
the entire 4–8 GHz frequency response of the C-band receiver.
The synthesized beam has an FWHM of 0 36, and the rms
noise level of the image is 3.5 μJy beam
−1
. Although the image
includes short spacing information from the compact Jansky
VLA configurations, the galaxy emission sits on a mild
negative bowl of order ∼5μJy beam
−1
that suggests some
extended emission is missing, but this does not affect our
calculations.
The Paschen αobservations are from the Mikulski Archive
for Space Telescopes (MAST)and were obtained by the
Hubble Space Telescope (HST)with the Near Infrared Camera
and Multi-Object Spectrometer with two filters: F187N (PID
7919, PI: W. Sparks), and F166N (PID 7218, PI: M. Rieke).
38
Figure 5. Decomposition of the F335M filter into continuum (left panel)and PAH emission (right panel)following the procedure described in Section 2.4. The
extraplanar tendrils of PAH emission are highly structured, and likely represent a view of the cold material entrained in the galactic wind. Position offsets are referred
to as 09
h
55
m
51 60, 69 40 45.
0
+¢(J2000), the often-used center of M82 based on 2 μm ground observations is located at offsets of +3.9, +0.8 (Lester et al. 1990).
38
doi:10.17909/jdx7-qg88
7
The Astrophysical Journal, 967:63 (15pp), 2024 May 20 Bolatto et al.

The continuum subtraction for the Paschen αimage uses
F166N and is described by D. B. Fisher et al. (2024, in
preparation). To briefly summarize, the subtraction followed
the procedure outlined in Böker et al. (1999), in which the flux
in F166N is scaled to the flux of F187N, with a scaling
determined in regions avoiding the bright midplane of the
galaxy. For a given F166N flux, there is a well-defined
minimum of the F187N flux. Pixels with values near the
minimum F187N/F166N trace the ratio of stellar continuum in
the two filters, while pixels with larger F187N/F166N are
dominated by Paschen αemission. To determine the con-
tinuum, we fit a linear correlation between F187N and F166N
for those pixels in the lowest 0.5% of the ratio F187N/F166N,
which is then subtracted from all pixels in the F187N image.
The CO J=1−0 image from Krieger et al. (2021)is part of
a large mosaic obtained using the Institut de Radioastronomie
Millimétrique (IRAM)Northern Extended Millimeter Array
interferometer (NOEMA). It includes short spacings from the
IRAM 30 m telescope, so in principle, it recovers all the source
flux. It has a synthesized beam of 2 08 ×1 65 (i.e., a circular
beam equivalent θ≈185), and an rms sensitivity of 138 mK
(5.15 mJy beam
−1
)in a 5 km s
−1
channel.
3. Results and Discussion
3.1. Multiwavelength Images of the Starburst
The wavelength progression in Figure 2shows clearly the
dramatic reduction in extinction at longer wavelengths. One of
the most striking features of these mosaics is the highly
structured extraplanar tendrils of emission seen in the F335M
and F360M filters. These are associated with PAH emission
features around 3.3 and 3.4 μm, as discussed in Section 2.4.
A three-color combination of near-IR data in Figure 3shows
images from our three filters on the blue side of NIRCam (1.40,
1.64, and 2.12 μm), maximizing the image resolution. This
high resolution allows us to identify resolved stellar sources
corresponding to clusters with stellar masses 10
4
M
e
and
larger, which are described in a companion paper (R. C. Levy
et al. 2024, in preparation). The highly obscured starburst
region is partially visible at these wavelengths, along with dark
extraplanar filaments emanating from the central region due to
dust entrained in the galaxy wind. Figure 4uses the
combination of the 1.64, 2.50, and 3.35 μmfilters to show
the PAH emission associated with the entrained dust. We
discuss the structure of the emission, which we separate from
the continuum as described in Section 2.4 (Figure 5)and
Section 3.2. Note that the bright extraplanar structures seen in
PAH emission in Figure 4have corresponding extinction (dark)
features due to dust lanes in Figure 3, seen against the blue
background of stars. In particular, the left-side wall of PAH
emission and some of the prominent PAH plumes on the north
side show clear corresponding extinction features.
The bright green compact sources in Figure 3are regions
with enhanced 1.64 μm emission. They mostly correspond to
SNRs, visible as an excess in our F164N image due to the
contribution from the [Fe II]transition at 1.644 μm, due to
shock-induced destruction of dust grains and the subsequent
release of iron atoms into a gaseous phase. The first
extragalactic source detected in the 1.644 μm[Fe II]transition
was indeed M82 (Rieke et al. 1980). SNRs also emit brightly in
the radio, due to synchrotron emission powered by relativistic
electrons accelerated by shocks (Dubner & Giacani 2015).We
can thus compare our 1.64 μm excess positions with radio
cataloged SNRs, in particular, the catalog from 1.7 GHz
observations with the Multi-Element Radio Linked Interfe-
rometer Network and VLBI observed by (Fenech et al. 2010,
Table 5).
Figure 6shows a zoomed area in detail, illustrating the very
good agreement in several supernovae (SNe)between the
[Fe II]emission and radio data (black circles). The SNRs
visible at 1.64 μm are within 0 2 of their radio positions,
providing an independent check of the absolute astrometric
accuracy. Not all radio SNRs have clear associated [Fe II]
emission. Some of them are located in regions that are very
heavily extinguished at 2 μm, and that may be the cause.
Conversely, not all regions with excess in F164N correspond to
cataloged radio SNRs. Some of these regions may correspond
to unidentified SNRs or extended shocks associated with the
galactic wind, a matter of future research.
3.2. Extraplanar PAH Emission and the Structure of the Wind
The combined data in Figure 4shows that material entrained in
the wind emits in the PAH near-IR bands. We produce an image
of the 3.3 μm PAH emission using the procedure described in
Section 2.4 to separate the continuum and the PAH contributions
to the F335M image (Figure 5). The separation works very well,
with no wispy features or artifacts present in the continuum
image. The 3.3 μm PAH emission has almost no stellar residuals,
although there are negative (black)areas in the central regions of
the starburst. These areas tend to correspond to highly reddened,
highly extinguished dark clouds, and they are mostly mildly
negative (∼−30 MJy sr
−1
)except in the area at the very center of
the starburst (near offsets +2, +2), where it becomes deeply
negative (∼−1000 MJy sr
−1
). Although the mildly negative
regions depend on the details of the continuum subtraction (
i.e., the local value of q)the central deeply negative region does
not: the F335M filter there is considerably below the baseline
determined by the continuum at 2.50 and 3.60 μm. As pointed
out above, the F335M filter is on the shoulder of a broad water
ice feature centered at 3.05 μm, so the simplest way to make it
underluminous is to have a deep ice absorption feature. But it is
also possible that the relative mix of aliphatic (which contributes
mostly to the F360M filter)and aromatic components changes
drastically in some of these regions. A large growth in the
aliphatic component (expected to be higher in dense regions, Rau
et al. 2019)at these locations may raise preferentially the
emission in the F360M filter. We hypothesize that very negative
regions are mostly due to water ice absorption along heavily
extinguished lines of sight, but definitive confirmation will
require spectroscopy.
The PAH image shows prominent plumes or pillars of PAH
emission extending outward from the central starburst region,
together with a network of complex filamentary substructures
and edge-brightened bubble-like features. An example of one
of these plumes (which appear to be bundles of filaments)is the
structure north of the midplane at R.A. +10, which extends
roughly north to south of the edge of the image and forms the
left segment of a V-shaped pattern with a neighboring plume.
Extended emission from dust grains and PAHs associated with
the wind has been detected before in M82, over scales of 6 kpc
(Engelbracht et al. 2006; Beirão et al. 2015). Our imaging
extends to ∼0.4 kpc from the central region, representing only
the innermost part of the material in the wind but providing a
8
The Astrophysical Journal, 967:63 (15pp), 2024 May 20 Bolatto et al.

much sharper view than was possible with the Spitzer Space
Telescope.
We can appreciate better the detailed structure of the PAH
emission in Figure 7, where the bottom panel presents an
unsharp-masked image that brings up the contrast on the small
scales. To obtain this image, we performed unsharp masking in
logarithmic space through the following steps: (1)clip the
original image I(top panel)below zero, (2)take the ()Ilog 1 +,
(3)convolve with a 5″(87 pc)FWHM Gaussian kernel to
produce I
S
(this scale was chosen so that it preserves reasonably
well the large-scale structure of the plumes, while enhancing
scales smaller than 100 pc),(4)compute the final image
[( ) ]
I
I1101
UM IS
=+ -
, which is displayed on a scale of
−0.5–1. The insets in Figure 7provide a zoom into some of the
wind substructure. The image shows a network of intertwined
filaments, which appear as coherent structures with lengths
∼100 pc in many cases, and thicknesses of 5–9 pc approaching
the 1.9 pc resolution of the image. Qualitatively similar
structures are created in simulations by the destruction of
dense clumps of material immersed in a wind (e.g., Schneider
et al. 2020; Abruzzo et al. 2022; Fielding & Bryan 2022),
where cold gas is ablated from clumps and extended into
filamentary structures that survive due to stabilization caused
by fast cooling.
Particularly striking are structures with the morphology of
elongated bubble walls, forming a ridge north of the midplane
and parallel to it, on the west side of the image (e.g., Figure 7
panel (F)). The larger bubbles on this part of the image extend
to ∼150 pc from the midplane. The north and south sides of the
outflow look somewhat different, with the north (the receding
side)having more defined large-scale structures than the south.
The bubble-like structures seen in PAH emission are also
apparent in high-resolution Hαimages, particularly in the less
extinguished south (approaching)side of the wind cone
(Figure 1). They may arise from super-bubbles pumped by
individual SNe events ejecting material. Another possibility is
that they are due to wind instabilities. For example, if the wind
in the warm gas is driven by the pressure of streaming cosmic
rays, the characteristic growth timescale of instabilities would
be τ
grow
∼H/c
cr
, where ()
c
P23
cr
2cr
r
=for P
cr
the cosmic-
ray pressure, ρthe density of the gas, and His a pressure scale
height (Quataert et al. 2022). The production of structures by
the instability requires that the growth time is shorter than the
flow timescale determined by the wind speed, v
w
, so that
τ
flow
∼H/v
w
, requiring c
cr
>v
w
, which in turn may be
attainable for moderate P
cr
if ρis not large.
In Figure 8, we compare the JWST PAH emission with the
highest resolution available images of (1)6 GHz radio
continuum from the VLA (J. Marvil et al. 2024, in preparation),
(2)ionized gas Paschen αrecombination emission from HST
(D. B. Fisher et al 2024, in preparation), and (3)molecular gas
from NOEMA (Krieger et al. 2021). See Section 2.5 for more
details on these ancillary data. The similarity of the extraplanar
filamentary features present in both the 6 GHz and Paschen α
images is particularly striking, and the same broad streams of
filaments are present in the PAH 3.3 μm emission. There is a
recombination transition at 3.297 μm(Pfund δ)that could
contribute to the line flux in F335M, and directly trace ionized
emission potentially confusing the interpretation of the
observations. Calculation shows, however, that it can only
account for ∼0.5 MJy sr
−1
(out of the several tens of
megajanskys per steradian present in the PAH filamentary
structures)using the observed Paschen αflux and the relative
intensity coefficients for case B recombination at low density
by Hummer & Storey (1987).
Figure 6. A11 5×9″(200 ×160 pc)area near the west corner of the mosaic, with radio SNR positions overlaid (the orientation is the same as Figure 2), The red,
green, and blue NIRCam mosaic uses the F212N, F164N, and F140M filters for red, green, and blue respectively. The circles (radius 0 2, or 3.5 pc)and labels
correspond to the radio SNRs (Fenech et al. 2010, Table 5). Green regions show an excess of emission at 1.64 μm likely associated with the [Fe II]line due to dust
destruction in supernovae shocks. This excess is visible on the radio SNRs 39.10+57.3, 40.68+55.1, 39.64+53.3, 38.76+53.5, and 40.32+55.2, although some
regions not associated with known radio SNRs also show a possible F164N excess, perhaps due to local or large-scale shocks. Very red sources are either heavily
extinguished at 2 μm or associated with vibrational emission from H
2
, usually caused by shocks.
9
The Astrophysical Journal, 967:63 (15pp), 2024 May 20 Bolatto et al.

For comparison, in the Milky Way center, the longest radio
filaments (which can be as long as ∼100 pc)are caused by
nonthermal synchrotron emission at 1.28 GHz, and appear to
be related to wind from SgrA
*
(Yusef-Zadeh et al. 2023).In
M82, however, the ratio of the 6 GHz continuum to
recombination line emission we measure for gas off the plane
of the galaxy is T
b
(mK)/I
Hα
(Rayleigh)≈0.1–0.2, assuming
case B recombination to convert Paschen αto Hα(Osterbrock
& Ferland 2006). This approximately corresponds to the
expected ratio for thermal free–free emission from plasma at
about T
e
≈5000 K (T
b
(mK)/I
Hα
(Rayleigh)≈0.18, Gaustad
et al. (2001), Equation (1)), assuming obscuration is
unimportant for Paschen α(an extinction correction would
drive the ratio even more into the thermal regime for the 6 GHz
emission, and a reflection component would act in the opposite
direction). This suggests that the elongated filamentary
structures are not due to a very strong magnetic field, which
presumably would cause the radio emission to be dominated by
synchrotron radiation even at 6 GHz and yield a nonthermal
ratio. Could the emission from these structures be dominated
by synchrotrons at lower frequencies near 1 GHz, where such
measurements are typically made? For a synchrotron spectral
index of ν
−0.7
the emission would grow by a factor of 3
between 6 and 1.3 GHz. This suggests that if synchrotrons were
Figure 7. Small-scale structure in the PAH emission. The top panel shows the same image as Figure 5with a grid superimposed. The bottom panel shows the result of
removing a version of the image smoothed on 5″(87 pc)scales, a technique known as unsharp masking, to bring up the structure below those scales. This highlights
the network of fine filaments and the bright walls of bubble-like structures present in the PAH image. The panels to the right show 7″×7″zooms into the top panel,
highlighting the rich substructure in the outflow. Their grayscale is linear and adjusted separately for each panel, as indicated by the accompanying color bar (in
MJy sr
−1
). Reference position is as in Figure 5.
10
The Astrophysical Journal, 967:63 (15pp), 2024 May 20 Bolatto et al.

contributing just under 40% of the radio emission at 6 GHz, it
could become just over 60% of the emission at 1.3 GHz.
Therefore, our measurements at 6 GHz place quite a strong
constraint on the emission contributed by nonthermal electrons
gyrating in magnetic fields associated with the base of the M82
outflow.
By contrast, even after accounting for the different resolutions,
there is just a weak correlation between the extraplanar structures
in the PAH emission and those in the molecular gas image. This
is also true when comparing with an integrated intensity map,
instead of the peak intensity shown in Figure 8.Itishardto
interferometrically image faint emission in the vicinity of a
bright source, especially with a small number of baselines. We
tested the quality of the faint extraplanar CO emission by
comparing the CO 1−0 mosaic with a very high-resolution CO 3
−2 mosaic obtained by the Submillimeter Array (SMA; M. J.
Figure 8. Comparison of different tracers of extraplanar material. A 5″scale corresponds to 87.3 pc at the distance of M82. Top left: Jansky VLA 6 GHz continuum at
036 (6.3 pc)resolution (square root stretch, units MJy sr
−1
). Top right: HST archival continuum-subtracted Paschen αat 0 25 (4.4 pc)resolution (square root
stretch, units erg s
−1
cm
−2
sr
−1
). Bottom left: NOEMA CO 1−0 peak intensity from Krieger et al. (2021)at 1 85 (32.3 pc)resolution (linear stretch, units K). Bottom
right: JWST 3.3 μm PAH emission at 0 11 (1.9 pc)resolution (square root stretch, units MJy sr
−1
). The ratio of 6 GHz to Paschen αimplies the extraplanar
filamentary radio continuum emission is thermal (see Section 3.2)The PAH emission shows very good detailed correspondence with the ionized gas traced by Paschen
αand the radio continuum. A faint grid is superimposed to facilitate comparison. Reference position is as in Figure 5.
11
The Astrophysical Journal, 967:63 (15pp), 2024 May 20 Bolatto et al.

Jiménez-Donaire et al. 2024, in preparation)and found reason-
ably good correspondence between the faint emission in both
CO maps, suggesting that the imaging quality for the faint
extraplanar gas is not the cause of the lesser degree of correlation
between PAH and CO emission.
The correlation between PAH and CO emission is poor, but not
completely absent. Since the resolution of the CO data is 17 times
lower than that of the PAH image, we only expect large-scale
features to be correlated, and indeed some of them are present
both in the CO and the PAH emission. For example, the wall of
CO emission that starts at +10″R.A., and larger offsets on the
northeast corner of the image, match the location of the plume of
PAH emission discussed previously. Figure 9shows the matched
resolution PAH image with CO-integrated intensity contours
overlaid to facilitate comparison. The base of some of the PAH
plumes, in both the northern and southern outflows, clearly
correlates with CO emission. Note that the lowest CO contour still
represents a fairly high column density, N(H
2
)∼2×10
22
cm
−2
(computed assuming X
CO
∼1×10
20
cm
−2
(Kkms
−1
)
−1
),and
this may be at least in part responsible for the poor correlation.
These are column densities of giant molecular clouds in the Milky
Way, so what is seen in CO are substantial clouds. Quantitative
comparison of spatial profiles of emission at different heights
from the plane is presented in D. B. Fisher et al. (2024, in
preparation).
3.3. Origin of the PAH Emission
Presumably, the PAH emission is tracing the cooler gas
entrained by the hot outflow since small dust grains in hot,
X-ray emitting gas should be very quickly destroyed by
sputtering caused by collisions with fast-moving electrons
(Micelotta et al. 2010b; Hu et al. 2019). This is more so for
PAHs emitting at 3.3 μm, a feature that is thought to be
dominated by the smallest PAHs (Draine et al. 2021), which
are those most quickly ablated by sputtering and fragmented by
photodestruction processes.
The timescale to travel from the midplane to half the vertical
extent of the extraplanar emission in our mosaic (a distance of
∼200 pc)at a typical velocity for a neutral or molecular gas
wind (v
w
∼200–300 km s
−1
, Leroy et al. 2015b; Yoshida et al.
2019)is τ
dyn
∼1 Myr, and the shortest possible τ
dyn
∼0.2 Myr
would correspond to gas moving at the velocity of the hot
phase (v
w
∼1000 km s
−1
). By comparison, the expected time
for sputtering a 50 atom PAH immersed in T≈6×10
6
K gas
with density n≈10
−2
cm
−3
is τ
sputter
∼10
−3
Myr, and even
200-atom PAHs have predicted lifetimes τ
sputter
10
−2
Myr
(Micelotta et al. 2010b). The central region of M82 may have
even hotter (T≈3×10
7
K)and denser (n≈0.1 cm
−3
)
conditions (Lopez et al. 2020), which would only accelerate
the PAH destruction. Therefore, it appears impossible for the
PAHs responsible for the observed emission to be mixed with
hot, X-ray-emitting gas.
As a consequence, we expect the PAHs to be associated with
colder phases, either neutral atomic/molecular gas or possibly
photoionized gas with T∼3×10
3
–10
4
K. The latter represents a
gas phase in which PAHs are not usually observed. PAHs appear
to be faint or absent in the ionized gas of H II regions (Chastenet
et al. 2019,2023; Egorov et al. 2023). High-resolution studies of
the Orion Bar photodissociation region (PDR)show a sharp
increase in the brightness of the PAH emission going from the
HII region into the molecular cloud and the neutral atomic layer
surrounding it at the location of the ionization front (Peeters et al.
2024). Although PAH emission is seen in the ionized gas region,
it is attributed to a background PDR. Observations of the
Horsehead nebula suggest a lower limit of 5 ×10
3
yr for the
lifetime of PAHs in ionized gas (Compiègne et al. 2007),butin
general, their destruction timescale is highly uncertain. The
collisional destruction calculations by Micelotta et al. (2010b)
indicate that PAHs embedded in high-density (n∼10
4
cm
−3
)
photoionized gas are destroyed mostly by impacts from He
nuclei, and their expected lifetime is τ
sputter
∼10 Myr. Since the
destruction rate is proportional to the frequency of collisions,
τ
sputter
∝n
−1
. Therefore, the PAH lifetime to collisional
destruction will be even longer at a more reasonable density of
n10
3
cm
−3
for the base of the outflow (measurements indicate
n∼200 cm
−3
500 pc away from the midplane, increasing
inward; Xu et al. 2023). These timescales are well in excess of
the dynamical time τ
dyn
needed to travel over the extent of our
image, which makes it plausible that some fraction of the PAH
emission originates from photoionized gas if collisions are the
primary destruction mechanism. Photodestruction (e.g., Jochims
et al. 1999; Tielens 2008), which preferentially removes the
smallest PAHs (Allain et al. 1996;LePageetal.2003)may be,
however, the ultimate mechanism limiting their lifetime as it
likely occurs for H II regions. Neutral or molecular gas advected
from high-density regions in the walls of the outflow and
recently exposed to ionizing radiation could act as a source of
PAHs constantly replenishing the ionized gas. Also, it is possible
that at least part of the PAHs observed are related to large dust
grain shattering in shocks, which can tilt the dust grain size
distribution toward smaller sizes (Jones et al. 1996;Seoketal.
2014), and provide a PAH source. Note that PAHs are rapidly
destroyed in shocks with velocities 100 km s
−1
(Micelotta et al.
2010a); therefore, a net PAH generation would require slower
velocities. Shocks associated with the interface between hot and
Figure 9. PAH and CO emission compared at the same resolution. The JWST
3.3 μm PAH emission (grayscale)has been convolved to the 1 85 resolution
of the NOEMA CO, presented here as the integrated intensity map (color
contours). The contour values are 190, 320, 550, 930, and 1580 K km s
−1
. The
lowest contour corresponds to a column density N(H
2
)∼2×10
22
cm
−2
.
Partial correspondence is observed between the base of some of the PAH
plumes and extraplanar CO emission. Reference position is as in Figure 5.
12
The Astrophysical Journal, 967:63 (15pp), 2024 May 20 Bolatto et al.

warm gas (Fielding & Bryan 2022)would be very inhospitable
to PAHs.
Neutral gas harboring PAHs in the extraplanar region of M82
close to the starburst is exposed to a very powerful photoionizing
radiation field. A starburst with SFR ∼12 M
e
yr
−1
generates
Q
ion
∼10
54
s
−1
ionizing photons according to Starburst99
(Leitherer et al. 1999). Only a fraction of this radiation will
escape the central regions, due to absorption by neutral gas or
extinction by dust. It is hard to know the precise escape fraction,
f
esc
, but ionizing radiation likely escapes from the central ionized
bubble along low-density, fully ionized channels. The ratio of
Paschen αin the extended extraplanar emission to that centrally
concentrated in the starburst suggests f
esc
∼0.25, but that is
likely an overestimate since the starburst region is heavily
extinguished even at 1.875 μm. Assuming that 90% of the
photons are lost to intervening extinction or neutral material (i.e.,
f
esc
=0.1)the ionizing photon flux in a region 200 pc (the
halfway travel distance in our mosaic discussed above, and
coincidentally also the approximate radius of the starburst
region)away from a compact starburst will be F
ion
∼2×
10
10
(f
esc
/0.1)s
−1
cm
−2
(comparable to the ionizing radiation
field a few parsecs away from an O7 star).
For neutral material to survive while immersed in such a
radiation field, it must have a minimum density. For a region
with a size of L≈03–0 5 (5–9pc), the typical width of many
of the filamentary features, balancing available photon flux
against recombination requires a threshold density of at least
()
n
FL f50 0.1
Bthr ion esc
a»~ cm
−3
to stay neutral
(α
B
∼4×10
−13
cm
3
s
−1
is the recombination rate of H).
Radio recombination line measurements over large scales point
to a high-density ionized gas component, which could be, in
part, newly ionized gas (Seaquist et al. 1996). Given the
ionizing flux computed above, the photoionization timescale
is ()F0.
3
ion ion ion 1
ts=~
-(f
esc
/0.1)
−1
yr for neutral gas
(where σ
ion
is the photoionization cross section of H),
and the recombination timescale is

()n
Brec 1
ta=-
()n1, 600 50 cm 3-yr for ionized gas. Since these timescales
are short compared to the time to traverse a few parsecs at the
flow velocity, it is likely that ionization equilibrium is satisfied,
although there may be large local variations in ionization state
due to lines of sight strongly attenuated by patchy overdense
gas. On average, we expect most neutral gas to have densities
larger than n
thr
or be otherwise quickly ionized, although this
will be strongly dependent on f
esc
and distance to ionizing
sources.
Neutral gas that survives the gauntlet (by having high
enough volume and column density to self-shield)and moves
away from the central source is likely to stay neutral for a long
time (the timescale to thermal conduction evaporation is long,
τ
evap
10 Myr, depending on density and mass; Micelotta
et al. 2010b), and pockets of dense, cool gas may even act as
condensation seeds and accrete from their hot surroundings
(Gronke & Oh 2018,2020). Measurements in the outflow find
column densities of neutral atomic gas N
H
∼3–10 ×10
20
cm
−2
(and similar values for molecular gas)on angular scales of 20″
close to M82 (Leroy et al. 2015b; Martini et al. 2018). This
neutral gas component of the outflowing material is highly
clumped on sub-resolution scales, as suggested by the large-
scale CO interferometer mapping (Krieger et al. 2021),
reaching much higher column density than measured above.
The picture that emerges from these considerations is that the
observed PAH emission likely comes from a mix of neutral and
photoionized material. Part of the PAHs are associated with
(and protected by)fairly dense and thus likely molecular gas.
Not all of this molecular gas may be emitted in CO since
protecting CO molecules requires higher column densities than
it does for H
2
in order to build up shielding to the
photodissociating UV radiation (see, e.g., Bolatto et al.
2013). Also, some of this molecular gas may be below the
sensitivity limit of the existing CO interferometric observations
(Krieger et al. 2021). Mid-IR spectroscopy of H
2
transitions
would be necessary to definitively establish the correlation
between 3.3 μm and molecular emission on small scales.
Because of the morphological similarity between the PAH
and ionized gas images, we suggest that another part of the
PAH emission arises from photoionized gas, which is easily
maintained given the high luminosity of the starburst. Since the
intensity of recombination and free–free emission depend
sharply on the gas density, bright emission indicates large
ionized gas densities, likely occurring at the interface between
neutral material at the photodissociation region and ionized
material in its vicinity. Projection effects are likely important.
Part of the PAH-emitting material may be located at the walls
of the outflow cone around the edges of the starburst, where
neutral and molecular gas are advected into the flow, and seen
in projection against the central regions of the outflow.
4. Conclusions
We present new, very high-resolution IR images obtained by
JWST of the central starburst region and the base of the galactic
outflow in M82 (Figures 3,4). These images reveal a highly
textured network of filamentary and bubble-like structures in
the dust component traced by 3.35 μm PAH emission
(Figures 5,7), as the material is entrained into the wind at
the base of the galactic outflow present in M82. The observed
filaments have thicknesses of 5–9 pc, approaching the JWST
resolution of ∼2 pc, lengths that can reach ∼100 pc, and
associations of these filaments form prominent plumes of
emission. The bubble-like structures also reach sizes of
∼100 pc. We suggest the latter arise from super-bubbles
pumped by individual SNe events ejecting material, but it is
also possible they are due to wind instabilities, for example,
due to cosmic rays (Section 3.2). The filaments of parsec-scale
width observed in PAH emission (Figure 7)are likely related to
ablation and shredding of dense clumps of material by the
galactic wind (Schneider et al. 2020; Abruzzo et al. 2022;
Fielding & Bryan 2022), possibly located in the walls of the
ionized outflow, although the precise mechanisms leading to
their structure remain to be determined.
The structure of the PAH emission closely resembles that of
the ionized gas, revealed by recombination in Paschen αand
free–free emission at 6 GHz, and has some correspondence
with the CO emission (Figure 8). The 6 GHz to Paschen αratio
in the ionized gas filaments indicates that their 6 GHz emission
is thermal, and hence they do not appear likely to be due to
very strong magnetic fields. We discuss the conditions in the
material associated with the PAH emission, finding that the
most likely scenario is a combination of PAHs embedded in
neutral/molecular gas and photoionized gas, likely associated
with photodissociation regions. Companion papers explore the
quantitative correlations between PAHs and other gas tracers
(D. B. Fisher et al. 2024, in preparation)and the statistics and
properties of massive star clusters in the starburst (R. C. Levy
et al. 2024, in preparation).
13
The Astrophysical Journal, 967:63 (15pp), 2024 May 20 Bolatto et al.

Acknowledgments
We acknowledge the comments from Bruce Draine, Brandon
Hensley, and the anonymous referee, which helped improve this
manuscript. This work is based on observations made with
NASA/ESA/CSA JWST. The data were obtained from MAST
at the Space Telescope Science Institute, which is operated by
the Association of Universities for Research in Astronomy, Inc.,
under NASA contract NAS 5-03127 for JWST. These
observations are associated with program JWST-GO-01701.
Support for program JWST-GO-01701 is provided by NASA
through a grant from the Space Telescope Science Institute,
which is operated by the Association of Universities for
Research in Astronomy, Inc., under NASA contract NAS
5-03127. We gratefully acknowledge Alyssa Pagan at the Space
Telescope Science Institute for her beautiful multicolor rendition
of the data used in Figures 3and 4. A.D.B. and S.A.C.
acknowledge support from the NSF under award AST-2108140.
R.C.L. acknowledges support for this work provided by an NSF
Astronomy and Astrophysics postdoctoral fellowship under
award AST-2102625. R.H.-C. thanks the Max Planck Society
for its support under the Partner Group project “The Baryon
Cycle in Galaxies”between the Max Planck Institute for
Extraterrestrial Physics and the Universidad de Concepción. R.
H-C. also gratefully acknowledges financial support from ANID
BASAL project FB210003. R.S.K. and S.C.O.G. acknowledge
funding from the European Research Council via the Synergy
Grant “ECOGAL”(project ID 855130), from the German
Excellence Strategy via the Heidelberg Cluster of Excellence
(EXC 2181-390900948)“STRUCTURES,”and from the Ger-
man Ministry for Economic Affairs and Climate Action in
project “MAINN”(funding ID 50OO2206). V.V. acknowledges
support from the scholarship ANID-FULBRIGHT BIO 2016-
56160020, funding from NRAO Student Observing Support
(SOS)—SOSPADA-015, and funding from the ALMA-ANID
Postdoctoral Fellowship under the award ASTRO21-0062. I.D.
L. acknowledges funding support from the European Research
Council (ERC)under the European Unionʼs Horizon 2020
research and innovation program DustOrigin (ERC-2019-StG-
851622)and funding support from the Belgian Science Policy
Office (BELSPO)through the PRODEX project “JWST/MIRI
Science exploitation”(C4000142239). L.L. acknowledges that a
portion of their research was carried out at the Jet Propulsion
Laboratory, California Institute of Technology, under a contract
with the National Aeronautics and Space Administration
(80NM0018D0004). The work of E.C.O. was supported in part
by grant 510940 from the Simons Foundation.
Facilities: JWST, VLA, HST, IRAM:NOEMA, IRAM:30m.
Software: JWST Calibration Pipeline (Bushouse et al. 2023),
JHAT (Rest et al. 2023), Matlab.
ORCID iDs
Alberto D. Bolatto https://orcid.org/0000-0002-5480-5686
Rebecca C. Levy https://orcid.org/0000-0003-2508-2586
Elizabeth Tarantino https://orcid.org/0000-0003-1356-1096
Martha L. Boyer https://orcid.org/0000-0003-4850-9589
Deanne B. Fisher https://orcid.org/0000-0003-0645-5260
Serena A. Cronin https://orcid.org/0000-0002-9511-1330
Adam K. Leroy https://orcid.org/0000-0002-2545-1700
Ralf S. Klessen https://orcid.org/0000-0002-0560-3172
J. D. Smith https://orcid.org/0000-0003-1545-5078
Danielle A. Berg https://orcid.org/0000-0002-4153-053X
Torsten Böker https://orcid.org/0000-0002-5666-7782
Leindert A. Boogaard https://orcid.org/0000-0002-
3952-8588
Eve C. Ostriker https://orcid.org/0000-0002-0509-9113
Todd A. Thompson https://orcid.org/0000-0003-2377-9574
Juergen Ott https://orcid.org/0000-0001-8224-1956
Laura Lenkićhttps://orcid.org/0000-0003-4023-8657
Laura A. Lopez https://orcid.org/0000-0002-1790-3148
Daniel A. Dale https://orcid.org/0000-0002-5782-9093
Sylvain Veilleux https://orcid.org/0000-0002-3158-6820
Paul P. van der Werf https://orcid.org/0000-0001-
5434-5942
Simon C. O. Glover https://orcid.org/0000-0001-6708-1317
Karin M. Sandstrom https://orcid.org/0000-0002-
4378-8534
Evan D. Skillman https://orcid.org/0000-0003-0605-8732
John Chisholm https://orcid.org/0000-0002-0302-2577
Vicente Villanueva https://orcid.org/0000-0002-5877-379X
Thomas S.-Y. Lai https://orcid.org/0000-0001-8490-6632
Sebastian Lopez https://orcid.org/0000-0002-2644-0077
Elisabeth A. C. Mills https://orcid.org/0000-0001-
8782-1992
Kimberly L. Emig https://orcid.org/0000-0001-6527-6954
Lee Armus https://orcid.org/0000-0003-3498-2973
Divakara Mayya https://orcid.org/0000-0002-4677-0516
David S. Meier https://orcid.org/0000-0001-9436-9471
Ilse De Looze https://orcid.org/0000-0001-9419-6355
Rodrigo Herrera-Camus https://orcid.org/0000-0002-
2775-0595
Fabian Walter https://orcid.org/0000-0003-4793-7880
Mónica Relaño https://orcid.org/0000-0003-1682-1148
Hannah B. Koziol https://orcid.org/0009-0001-5949-1524
Joshua Marvil https://orcid.org/0000-0003-1111-8066
María J. Jiménez-Donaire https://orcid.org/0000-0002-
9165-8080
Paul Martini https://orcid.org/0000-0002-0194-4017
References
Abruzzo, M. W., Bryan, G. L., & Fielding, D. B. 2022, ApJ,925, 199
Allain, T., Leach, S., & Sedlmayr, E. 1996, A&A, 305, 602
Aniano, G., Draine, B. T., Gordon, K. D., & Sandstrom, K. 2011, PASP,
123, 1218
Beirão, P., Armus, L., Lehnert, M. D., et al. 2015, MNRAS,451, 2640
Böker, T., Calzetti, D., Sparks, W., et al. 1999, ApJS,124, 95
Bolatto, A. D., Wolfire, M., & Leroy, A. K. 2013, ARA&A,51, 207
Boucaud, A., Bocchio, M., Abergel, A., et al. 2016, A&A,596, A63
Bushouse, H., Eisenhamer, J., Dencheva, N., et al. 2023, JWST Calibration
Pipeline, v1.12.5, Zenodo, doi:10.5281/zenodo.10022973
CASA Team, Bean, B., Bhatnagar, S., et al. 2022, PASP,134, 114501
Chastenet, J., Sandstrom, K., Chiang, I.-D., et al. 2019, ApJ,876, 62
Chastenet, J., Sutter, J., Sandstrom, K., et al. 2023, ApJL,944, L11
Colbert, J. W., Malkan, M. A., Clegg, P. E., et al. 1999, ApJ,511, 721
Compiègne, M., Abergel, A., Verstraete, L., et al. 2007, A&A,471, 205
Conroy, C., & van Dokkum, P. G. 2012, ApJ,760, 71
Contursi, A., Poglitsch, A., Graciá Carpio, J., et al. 2013, A&A,549, A118
Dalcanton, J. J., Williams, B. F., Seth, A. C., et al. 2009, ApJS,183, 67
de Blok, W. J. G., Walter, F., Ferguson, A. M. N., et al. 2018, ApJ,865, 26
Devine, D., & Bally, J. 1999, ApJ,510, 197
Draine, B. T., & Li, A. 2001, ApJ,551, 807
Draine, B. T., Li, A., Hensley, B. S., et al. 2021, ApJ,917, 3
Dubner, G., & Giacani, E. 2015, A&ARv,23, 3
Egorov, O. V., Kreckel, K., Sandstrom, K. M., et al. 2023, ApJL,944, L16
Emig, K. L., Bolatto, A. D., Leroy, A. K., et al. 2020, ApJ,903, 50
Engelbracht, C. W., Kundurthy, P., Gordon, K. D., et al. 2006, ApJL,
642, L127
Fenech, D., Beswick, R., Muxlow, T. W. B., Pedlar, A., & Argo, M. K. 2010,
MNRAS,408, 607
14
The Astrophysical Journal, 967:63 (15pp), 2024 May 20 Bolatto et al.

Fielding, D. B., & Bryan, G. L. 2022, ApJ,924, 82
Förster Schreiber, N. M., Genzel, R., Lutz, D., Kunze, D., & Sternberg, A.
2001, ApJ,552, 544
Förster Schreiber, N. M., Genzel, R., Lutz, D., & Sternberg, A. 2003, ApJ,
599, 193
Freedman, W. L., Hughes, S. M., Madore, B. F., et al. 1994, ApJ,427, 628
García-Bernete, I., Rigopoulou, D., Alonso-Herrero, A., et al. 2022, MNRAS,
509, 4256
Gaustad, J. E., McCullough, P. R., Rosing, W., & Van Buren, D. 2001, PASP,
113, 1326
Greco, J. P., Martini, P., & Thompson, T. A. 2012, ApJ,757, 24
Gronke, M., & Oh, S. P. 2018, MNRAS,480, L111
Gronke, M., & Oh, S. P. 2020, MNRAS,492, 1970
Hammonds, M., Mori, T., Usui, F., & Onaka, T. 2015, P&SS,116, 73
Heckman, T. M., Armus, L., & Miley, G. K. 1990, ApJS,74, 833
Herrera, C. N., Boulanger, F., Nesvadba, N. P. H., & Falgarone, E. 2012,
A&A,538, L9
Herrera-Camus, R., Sturm, E., Graciá-Carpio, J., et al. 2018, ApJ,861, 94
Hu, C.-Y., Zhukovska, S., Somerville, R. S., & Naab, T. 2019, MNRAS,
487, 3252
Hummer, D. G., & Storey, P. J. 1987, MNRAS,224, 801
Imanishi, M., Nakagawa, T., Shirahata, M., Ohyama, Y., & Onaka, T. 2010,
ApJ,721, 1233
Joblin, C., Tielens, A. G. G. M., Allamandola, L. J., & Geballe, T. R. 1996,
ApJ,458, 610
Jochims, H. W., Baumgärtel, H., & Leach, S. 1999, ApJ,512, 500
Jones, A. P., Tielens, A. G. G. M., & Hollenbach, D. J. 1996, ApJ,469, 740
Krieger, N., Walter, F., Bolatto, A. D., et al. 2021, ApJL,915, L3
Kruijssen, J. M. D. 2012, MNRAS,426, 3008
Krumholz, M. R., Dekel, A., & McKee, C. F. 2012, ApJ,745, 69
Krumholz, M. R., McKee, C. F., & Bland-Hawthorn, J. 2019, ARA&A,
57, 227
Lai, T. S. Y., Armus, L., Bianchin, M., et al. 2023, ApJL,957, L26
Lai, T. S. Y., Smith, J. D. T., Baba, S., Spoon, H. W. W., & Imanishi, M. 2020,
ApJ,905, 55
Le Page, V., Snow, T. P., & Bierbaum, V. M. 2003, ApJ,584, 316
Lehnert, M. D., Heckman, T. M., & Weaver, K. A. 1999, ApJ,523, 575
Leitherer, C., Schaerer, D., Goldader, J. D., et al. 1999, ApJS,123, 3
Leroy, A. K., Bolatto, A. D., Ostriker, E. C., et al. 2015a, ApJ,801, 25
Leroy, A. K., Bolatto, A. D., Ostriker, E. C., et al. 2018, ApJ,869, 126
Leroy, A. K., Walter, F., Martini, P., et al. 2015b, ApJ,814, 83
Lester, D. F., Carr, J. S., Joy, M., & Gaffney, N. 1990, ApJ,352, 544
Levy, R. C., Bolatto, A. D., Tarantino, E., et al. 2023, ApJ,958, 109
Linden, S. T., Evans, A. S., Larson, K., et al. 2021, ApJ,923, 278
Lokhorst, D., Abraham, R., Pasha, I., et al. 2022, ApJ,927, 136
Lopez, L. A., Mathur, S., Nguyen, D. D., Thompson, T. A., & Olivier, G. M.
2020, ApJ,904, 152
Maragkoudakis, A., Peeters, E., & Ricca, A. 2020, MNRAS,494, 642
Martini, P., Leroy, A. K., Mangum, J. G., et al. 2018, ApJ,856, 61
Mayya, Y. D., Bressan, A., Carrasco, L., & Hernandez-Martinez, L. 2006, ApJ,
649, 172
Mayya, Y. D., Romano, R., Rodríguez-Merino, L. H., et al. 2008, ApJ,
679, 404
McCrady, N., Gilbert, A. M., & Graham, J. R. 2003, ApJ,596, 240
McCrady, N., & Graham, J. R. 2007, ApJ,663, 844
Melo, V. P., Muñoz-Tuñón, C., Maíz-Apellániz, J., & Tenorio-Tagle, G. 2005,
ApJ,619, 270
Micelotta, E. R., Jones, A. P., & Tielens, A. G. G. M. 2010a, A&A,510, A36
Micelotta, E. R., Jones, A. P., & Tielens, A. G. G. M. 2010b, A&A,510, A37
Mitchell, P. D., & Schaye, J. 2022, MNRAS,511, 2948
Mutchler, M., Bond, H. E., Christian, C. A., et al. 2007, PASP,119, 1
Nguyen, D. D., & Thompson, T. A. 2022, ApJL,935, L24
O’Connell, R. W., Gallagher, J. S. I., Hunter, D. A., & Colley, W. N. 1995,
ApJL,446, L1
Osterbrock, D. E., & Ferland, G. J. 2006, Astrophysics of Gaseous Nebulae
and Active Galactic Nuclei (Mill Valley, CA: Univ. Science Books)
Ostriker, E. C., & Shetty, R. 2011, ApJ,731, 41
Peeters, E., Habart, E., Berne, O., et al. 2024, A&A, submitted (arXiv:2310.
08720)
Perley, R. A., & Butler, B. J. 2013, ApJS,204, 19
Portegies Zwart, S. F., McMillan, S. L. W., & Gieles, M. 2010, ARA&A,
48, 431
Quataert, E., Jiang, Y.-F., & Thompson, T. A. 2022, MNRAS,510, 920
Rau, S.-J., Hirashita, H., & Murga, M. 2019, MNRAS,489, 5218
Rau, U., & Cornwell, T. J. 2011, A&A,532, A71
Rest, A., Pierel, J., Correnti, M., et al. 2023, arminrest/jhat: The JWST HST
Alignment Tool (JHAT), v2, Zenodo, doi:10.5281/zenodo.7892935
Rieke, G. H., Lebofsky, M. J., Thompson, R. I., Low, F. J., & Tokunaga, A. T.
1980, ApJ,238, 24
Rieke, M. J., Kelly, D. M., Misselt, K., et al. 2023, PASP,135, 028001
Sandstrom, K. M., Chastenet, J., Sutter, J., et al. 2023, ApJL,944, L7
Scarrott, S. M., Eaton, N., & Axon, D. J. 1991, MNRAS,252, 12P
Schneider, E. E., Ostriker, E. C., Robertson, B. E., & Thompson, T. A. 2020,
ApJ,895, 43
Seaquist, E. R., Carlstrom, J. E., Bryant, P. M., & Bell, M. B. 1996, ApJ,
465, 691
Seok, J. Y., Hirashita, H., & Asano, R. S. 2014, MNRAS,439, 2186
Shopbell, P. L., & Bland-Hawthorn, J. 1998, ApJ,493, 129
Siebenmorgen, R., & Krügel, E. 2010, A&A,511, A6
Smith, L. J., Westmoquette, M. S., Gallagher, J. S., et al. 2006, MNRAS,
370, 513
Strickland, D. K., Ponman, T. J., & Stevens, I. R. 1997, A&A,320, 378
Sturm, E., Lutz, D., Tran, D., et al. 2000, A&A,358, 481
Tielens, A. G. G. M. 2008, ARA&A,46, 289
Turner, J. L., Beck, S. C., Benford, D. J., et al. 2015, Natur,519, 331
Veilleux, S., Cecil, G., & Bland-Hawthorn, J. 2005, ARA&A,43, 769
Veilleux, S., Maiolino, R., Bolatto, A. D., & Aalto, S. 2020, A&ARv,28, 2
Veilleux, S., Rupke, D. S. N., & Swaters, R. 2009, ApJL,700, L149
Walter, F., Weiss, A., & Scoville, N. 2002, ApJL,580, L21
Whitmore, B. C., Brogan, C., Chandar, R., et al. 2014, ApJ,795, 156
Whitmore, B. C., Gilmore, D., Leitherer, C., et al. 2005, AJ,130, 2104
Willott, C. J., Doyon, R., Albert, L., et al. 2022, PASP,134, 025002
Xie, Y., & Ho, L. C. 2022, ApJ,925, 218
Xu, X., Heckman, T., Yoshida, M., Henry, A., & Ohyama, Y. 2023, ApJ,
956, 142
Yoshida, M., Kawabata, K. S., Ohyama, Y., Itoh, R., & Hattori, T. 2019, PASJ,
71, 87
Yun, M. S., Ho, P. T. P., & Lo, K. Y. 1994, Natur,372, 530
Yusef-Zadeh, F., Arendt, R. G., Wardle, M., & Heywood, I. 2023, ApJL,
949, L31
15
The Astrophysical Journal, 967:63 (15pp), 2024 May 20 Bolatto et al.