[C ii] Spectral Mapping of the Galactic Wind and Starbursting Disk of M82 with SOFIA

M82 is an archetypal starburst galaxy in the local Universe. The central burst of star formation, thought to be triggered by M82's interaction with other members in the M81 group, is driving a multiphase galaxy-scale wind away from the plane of the disk that has been studied across the electromagnetic spectrum. Here, we present new velocity-resolved observations of the [C ii] 158 μm line in the central disk and the southern outflow of M82 using the upGREAT instrument on board SOFIA. We also report the first detections of velocity-resolved (ΔV = 10 km s−1) [C ii] emission in the outflow of M82 at projected distances of ≈1–2 kpc south of the galaxy center. We compare the [C ii] line profiles to observations of CO and H i and find that likely the majority (>55%) of the [C ii] emission in the outflow is associated with the neutral atomic medium. We find that the fraction of [C ii] actually outflowing from M82 is small compared to the bulk gas outside the midplane (which may be in a halo or tidal streamers), which has important implications for observations of [C ii] outflows at higher redshift. Finally, by comparing the observed ratio of the [C ii] and CO intensities to models of photodissociation regions, we estimate that the far-ultraviolet (FUV) radiation field in the disk is ∼103.5 G 0, in agreement with previous estimates. In the outflow, however, the FUV radiation field is 2–3 orders of magnitudes lower, which may explain the high fraction of [C ii] arising from the neutral medium in the wind.


INTRODUCTION
Stars inject energy and momentum into their surrounding environments over their entire lifetimes.This so-called stellar feedback can result in massive outflows of gas and dust from the centers of galaxies, where there are high concentrations of stars and star clusters.In these dense environments, energy and momentum injected by supernova explosions and winds from massive stars that are clustered in space and time push material out of the midplane of a galaxy, leading to the observed biconical outflows and superwinds (e.g., Heckman et al. 1990;Veilleux et al. 2005Veilleux et al. , 2020, and references therein).These outflows are inherently multiphase, with the cool neutral phases of the winds potentially carrying away the bulk of the outflowing mass and the fuel for future star formation (e.g., Veilleux et al. 2020).
The [Cii] 158µm line in the far-infrared (FIR) is one of the brightest lines in star forming galaxies.It is a major cooling channel of the neutral interstellar medium (ISM) and can contribute 0.1-1% of the total FIR emission from a galaxy (e.g., Crawford et al. 1985;Stacey et al. 1991).Owing to its low ionization potential (11.2 eV), neutral carbon can be singly ionized in a range of conditions and ISM phases, making the [Cii] 158µm line an excellent tracer of multiphase gas (e.g., Madden et al. 1993;Goldsmith et al. 2012;Pineda et al. 2013).In the disks and centers of nearby star forming galaxies, [Cii] tends to be most closely associated with the neutral atomic component, but a substantial fraction can also arise from the molecular component (Mookerjea et al. 2016;Röllig et al. 2016;Fahrion et al. 2017;Tarantino et al. 2021).It is unknown, however, how the origin and distribution of [Cii] may change in a starburst-driven superwind.M 82 is an archetypal starburst galaxy, located at a distance of 3.63 ± 0.34 Mpc (Freedman et al. 1994) in the M 81 group.M 82 is interacting with the other group members (Yun et al. 1994;de Blok et al. 2018).It is thought that this tidal interaction triggered a central burst of star formation 10 Myr ago, followed by a second bar-driven burst 5 Myr ago (Förster Schreiber et al. 2003).The central starburst has launched a multiphase, galaxy-scale wind, which has been studied across the electromagnetic spectrum (e.g., Lynds & Sandage 1963;Heckman et al. 1990;Strickland et al. 1997;Walter et al. 2002;Engelbracht et al. 2006;Strickland & Heckman 2009;Veilleux et al. 2009;Yoshida et al. 2011;Yamagishi et al. 2012;Contursi et al. 2013;Beirão et al. 2015;Leroy et al. 2015;Martini et al. 2018;Yoshida et al. 2019;Krieger et al. 2021).It is unclear whether the material in the wind has sufficient energy to escape into the intergalactic medium or whether it will fall back onto the galaxy as a fountain (e.g., Leroy et al. 2015;Martini et al. 2018;Yuan et al. 2023).
Because M 82 is such an important laboratory for studying the multiphase ISM in a starburst environment, its center has previously been observed in the [Cii] 158µm line by the Kuiper Airborne Observatory (KAO; Stacey et al. 1991) and Herschel HIFI and PACS (Loenen et al. 2010;Contursi et al. 2013;Herrera-Camus et al. 2018).While the PACS data extend into the base of the outflow, [Cii] has not been detected in the outflow at distances greater than ∼ 1 kpc from the midplane.Moreover, because of the low velocity resolution of PACS, the [Cii] data from that instrument are not velocity-resolved.
In this paper, we present new observations of the [Cii] 158µm line taken with the upgraded German REceiver for Astronomy at Terahertz Frequencies (upGREAT; Risacher et al. 2018) on board the Stratospheric Observatory for Infrared Astronomy (SOFIA; Temi et al. 2018).These velocity-resolved observations measured [Cii] at 10 km s −1 velocity resolution in the inner disk of M 82 and the southern side of the superwind at distances of 1 − 2 kpc from the midplane.
This paper is organized as follows.We describe the data used in this study in Section 2. In Section 3, we discuss the main results from the [Cii] detections in the outflow of M 82.The main results from the [Cii] map in the disk are discussed in Section 4. A discussion of the disk and wind together is presented in Section 5. We discuss constraints on the far-ultraviolet (FUV) radiation field in both the disk and outflow in Section 6.We summarize our main conclusions in Section 7.
Throughout, we use CO to refer to 12 C 16 O(J = 1−0).We use (α, δ) = (09 h 55 m 52.72 s , +69 • 40 m 45.8 s ) as the J2000 right ascension and declination of the center of M 82 (Martini et al. 2018).We assume a central recessional velocity of 210 km s −1 LSRK (Krieger et al. 2021).We assume that the major axis position angle 1 (PA) of M 82 is 67 • ; therefore, the PA of the southern side of the wind is 157 • (e.g., Martini et al. 2018).We adopt an inclination of 80 • for the disk (e.g., Lynds & Sandage 1963;McKeith et al. 1993;Martini et al. 2018;Krieger et al. 2021).The data and code for the analysis presented here are available online2 .

[CII] Data from upGREAT
These observations of the [Cii] 158µm line in M 82 were taken using the low frequency array (LFA) of the upGREAT3 instrument (Risacher et al. 2018) onboard SOFIA (Temi et al. 2018).These data were taken in cycle 8 as part of project 08_0225 (PI: R. Levy) on 2021 February 19,February 23,February 25,March 10,and March 11.The upGREAT LFA consists of a seven-pixel hexagonal array for each polarization; the polarizations were averaged for these data.It was tuned to the [Cii] line at 158µm (1.9005 THz).At this frequency, each up-GREAT LFA pixel has a half-power beam width of 14.1 ′′ (≈ 250 pc).The bandwidth of the observations ranged from -150−430 km s −1 (1.8996−1.9033THz, ∆ν = 3.6 GHz), with a native velocity resolution of 0.04 km s −1 (R ≈ 7.5 × 10 6 ).

Levy et al.
The observing time was split between making an onthe-fly (OTF) map of the inner 3 ′ ×1 ′ of the disk and a single pointing of the upGREAT LFA along the southern outflow.Both observing strategies used a dual-beamswitching mode to measure the OFF positions to maximize baseline stability.More details pertaining to the OTF map and single pointing are given in Sections 2.1.1 and 2.1.2.The footprints of the OTF map and single pointing are shown in Figure 1.
All observations were pipeline calibrated (in particular: correction for atmospheric transmission) with the GREAT kalibrate software (Guan et al. 2012) by the upGREAT team and further processed to level-2 data using the class software in gildas.As part of the calibration, a first-order baseline was removed, the final spectra were smoothed to 10 km s −1 channels, and the spectra were converted to a main-beam temperature (T mb ) scale, where η mb ≈ 0.67.

Disk Map
To map the inner region of the disk, the LFA was scanned across the central 4.3 ′ ×2.3 ′ , centered on the galaxy center given in Section 1, at an angle of -20 • so that the long-axis of the map is aligned with the galaxy's major axis.With this strategy, the fully-sampled region of the map covers the inner 3.1 ′ ×1.1 ′ (3 kpc × 1 kpc) as shown in Figure 1.Two tunings (centered at 250 km s −1 and 290 km s −1 LSRK) were used to capture the full extent of the [Cii] emission in the disk.The spectra were weighted by 1/σ 2 rms , where σ rms is the root-mean-square (rms) noise of the baseline.The 3D data cube of the disk has an rms noise of 307 mK in 10 km s −1 channels away from the emission.Spatial pixels are 7 ′′ ×7 ′′ and hence oversample the beam by a factor of ≈ 3 (in area).
We make 2D maps of the [Cii] peak intensity, integrated intensity, mean velocity, and linewidth using moments4 .We restrict the velocity range to −100 − 400 km s −1 in the calculation of these quantities.For the disk maps, we mask out elements in the cube where the intensity is less than 2× rms noise (2 × 307 = 614 mK).These moment maps are shown in Figure 2, with the integrated intensity map also being shown in Figure 1.
As noted in Section 1, the [Cii] 158µm line has been previously observed in the center of M 82 using the KAO (Stacey et al. 1991) and Herschel PACS (Contursi et al. 2013;Herrera-Camus et al. 2018) and HIFI (Loenen et al. 2010).In this article, we compare the upGREAT map to that from PACS (Contursi et al. 2013) 5 .To robustly compare these measurements, we convolve the PACS cube to the 14 ′′ Gaussian beam of upGREAT using the kernels provided by Aniano et al. (2011).In single beams (not maps), Tarantino et al. (2021) found that not properly accounting for the different point-spread-function (PSF) shapes can lead to 40% discrepancies between fluxes measured by PACS and up-GREAT.We compute the integrated intensity using a moment analysis over the same velocity range as the up-GREAT data.We measure the flux from the matched upGREAT and PACS maps in a 1 ′ × 1 ′ box (rotated by -20 • ) centered on the center of M 82 (Martini et al. 2018).This box optimizes the overlapping regions of the upGREAT and PACS maps.Over this region, the integrated [Cii] line intensity measured from the upGREAT map is 2.0 × 10 −13 W m −2 .The matched PACS map yields 1.9 × 10 −13 W m −2 .These two values agree to within 5%, well within the 30% PACS calibration uncertainty (Contursi et al. 2013).

Outflow Pointings
To measure [Cii] in the outflow, a single pointing of the upGREAT LFA was used.A location on the southern (brighter) side of the outflow was chosen such that the central pixel of the upGREAT array (Pixel 0) was located 1.5 kpc from the galaxy center along the outflow (i.e., the galaxy minor axis with PA = 157 • ).The LFA array was rotated by 70 • to maximize the extent along the southern outflow.The footprint of the LFA is shown is Figures 1 and 2. The spectra in the outflow, shown in Figure 3, have rms noise of 7 − 12 mK in 10 km s −1 channels.
We calculate moments of the spectra to measure the peak and integrated intensity, the mean velocity, and the linewidth.As with the disk map, we restrict the velocity range to −100 − 400 km s −1 in the calculation of these quantities.We report these values in Table 1 and show these quantities in Figure 2.
As a check, we compare the flux we measure in outflow Pixels 1 and 6 to the Herschel PACS map (Contursi et al. 2013) convolved to a 14 ′′ Gaussian PSF as described in Section 2.1.1.Pixels 1 and 6 are the only upGREAT pointings that overlap with the PACS coverage.From the PACS map, we measure a total [Cii] integrated line intensity of 4.8 K km s −1 (12.0 K km s −1 ) in Pixel 1 (6).This is lower than the [Cii] integrated intensity Note-The R.A. and Decl. of the center of each pixel are given.The other columns show the moments of the spectra including the peak intensity (I peak ), the integrated intensity (moment 0; Iint), the mean velocity (moment 1; V0), the linewidth (moment 2, σV), and corresponding uncertainties.The moments are calculated over a fixed velocity range spanning -100 − 400 km s −1 .rms is the root-mean-square noise of the spectrum calculated outside of the velocity range used for the moments.If the moments cannot be calculated because the line is not detected, then the rms is calculated over the entire bandpass.All temperatures refer to T mb .M C + is the mass in C + inferred from the spectrum considering only collisions with the atomic gas; see Appendix A for details of this calculation.
of 21.2 K km s −1 (23.3 K km s −1 ) measured in the up-GREAT map (Table 1).We note, however, that these pixels are at the edges of the PACS map and there may be substantial flux loss due to edge effects and undersampling.When these PACS maps were reprocessed by Herrera-Camus et al. (2018), for example, these edge regions were excluded.

Ancillary Data
Because [Cii] can be excited in many conditions, we compare the [Cii] emission with the molecular and atomic ISM components.We use ancillary CO and Hi data as tracers of those respective components.

CO(1-0) Tracing the Molecular Gas
The CO data used in this study is from the Institut de Radioastronomie Millimétrique (IRAM) 30-m telescope.These data were presented by Krieger et al. (2021), and we direct the reader to that paper for details on the observations, calibration, and imaging.We note that while the data presented by Krieger et al. (2021) include both interferometric and single-dish observations, we find that there are significant interferometric artifacts (absorption) in the spectra that hinder the comparison to the [Cii] data in this work.Therefore, we use only the single-dish data, which do not show these spectral artifacts.The final CO cube has a spatial resolution of 22 ′′ (≈ 385 pc) and a velocity resolution of 5 km s −1 .For this work, we smooth the CO cube to a velocity resolution of 10 km s −1 to match the [Cii] data.We show the CO integrated intensity in Figure 1.

HI Tracing the Atomic Gas
The Hi data used in this study is from the Karl G. Jansky Very Large Array (VLA) which has been combined with single-dish data from the Robert C. Byrd Green Bank Telescope (GBT).These data were presented by Martini et al. (2018), and we direct the reader to that paper for details on the observations, calibration, and imaging.The Hi cube has a native velocity resolution of 5 km s −1 .We spectrally smooth this data to a velocity resolution of 10 km s −1 to match the [Cii] data.The Hi data used here have a spatial resolution of 17 ′′ (≈ 300 pc), slightly higher than that presented by Martini et al. (2018).In the center of M 82, the Hi is heavily absorbed against the bright continuum (Martini et al. 2018), so the Hi cannot be compared to the [Cii] in the disk; this is not a problem in the outflow pointings.
To produce the integrated intensity map (Figure 1), we create a mask to exclude the absorption and to exclude channels where the Hi intensity < 2.3 K (SNR< 3).We note that our analysis is performed on the Hi cube itself, and the Hi integrated intensity map is to aid in the visualization.

r-band Image
For visualization purposes, we show an r-band image of M 82 in Figure 1.This image was taken with as part of the SINGS Survey (Kennicutt et al. 2003;SINGS Team 2020) and was downloaded from the NASA/IPAC Extragalactic Database.Note-To form the ratios, all integrated intensities are in K km s −1 units and all luminosities are in L⊙ units.

Matching the Datasets
In the outflow, we extract the CO and Hi from the central pixel of each upGREAT pointing.As noted above, the CO and Hi data are smoothed to a velocity resolution of 10 km s −1 , to match the [Cii].To make the most accurate comparison, the data should be convolved to the 22 ′′ resolution of the CO data.Because, however, the upGREAT data in the outflow are single pointings (not a map), they cannot be convolved to lower resolution.Therefore, we do not match the beam sizes of the CO, Hi, and [Cii] data in the outflow.As we will discuss in Section 4, we do match the resolutions and pixel scales of the [Cii] and CO datasets in the disk.
We show the [Cii], CO, and Hi spectra in Figure 3, where the CO and Hi are simply normalized to the peak intensity of the [Cii].We derive the CO and Hi integrated intensities (moment 0) in the same way as for the [Cii] spectra and over the same velocity range.In Table 2, we give the ratios of the integrated intensities of Hi/[Cii] and CO/ [Cii], where all integrated intensities are in K km s −1 units.

[CII] IN THE WIND OF M 82
We robustly detect the [Cii] 158µm line in five of the seven LFA pixels in the southern outflow of M 82, as shown in Figure 3.As described in Section 2.1.2,we calculate moments of these spectra which are listed in Table 1 and shown in Figure 2. The ratios of the peak CO to [Cii] intensities (in K brightness temperature units) are ∼ 1, in agreement with previous work (Stacey et al. 1991).We present the integrated intensity ratios (on a K km s −1 intensity scale) in each LFA pixel in Table 2.
From the [Cii] spectra in the outflow, we calculate the column density and mass of C + in each outflow pointing.These calculations are detailed in Appendix A. Briefly, the C + column density as a function of ve-locity (N C + ) is calculated the following Equation A5, assuming that the [Cii] is only excited in the cold neutral medium (CNM; e.g., Pineda et al. 2013;Fahrion et al. 2017;Herrera-Camus et al. 2017).For this calculation, we must assume a kinetic temperature, T , and a gas density, n.Our assumed temperature and density come from inspecting the spatially-resolved photodissociation region (PDR) modeling results presented by Contursi et al. (2013).Though they do not probe out as far into the outflow as our measurements, they find temperature of T ∼ 200 − 300 K and densities of n ∼ 10 − 150 cm −3 along the southern outflow away from the disk (uncorrected for the effects of the ionized gas; see their Figures 15 and 16).We note that these values differ from those presented in their Table 1.The "southern outflow" macro region they define is likely contaminated by the starburst (see their Figure 18) and hence the average temperature and density reported in that table are likely too high to apply to the part of the outflow we are studying.From these results, we take a representative temperature of T = 250 ± 50 K and a representative density of n = 100 ± 50 cm −3 .
As we discuss in more detail below, our choice to assume that the [Cii] is primarily excited in the CNM is well-justified.We sum this column density over velocity and multiply by the area of each LFA pixel to measure the mass in C + (M C + ; Equation A11).We list M C + in each pointing in Table 1.The total C + mass in the part of the outflow covered by these pointings is 2.4×10 4 M ⊙ (excluding Pixels 4 and 5 where the [Cii] line is not detected).

Atomic Gas in the Outflow
From N C + , we can estimate the effective CNM column density needed to produce the observed [Cii] spectra (see details in Section A.1).To make this conversion, we divide N C + by a C/H abundance ratio (C/H = 1.5 × 10 −4 ; Gerin et al. 2015).We show these effective CNM column density profiles (per unit velocity, i.e., divided by the channel width of 10 km s −1 ) based on the [Cii] (N CNM ) in teal in Figure 4.In order to determine the ISM phase from which most of the [Cii] emission originates, we compare our N 1 max(N CO H2 ) max(N [CII]   H2 ) = 0.42 ± 0.07 0 250 500 6 max(N CO H2 ) max(N [CII]   H2 ) = 0.20 ± 0.04 H2 ) = 0.30 ± 0.05 H 2 ) assuming collisions with only H2, T = 250 K, n = 100 cm −3 , and C/H2 = 3 × 10 −4 .The gray curves show the column density of H2 measured from the CO data (N CO H 2 ) assuming XCO = 0.5 × 10 20 cm −2 (K km s −1 ) −1 .The number in the top right corner of each panel is the ratio of the peaks of N CO H 2 and N [CII] H 2 over the shaded portions of the profiles.The uncertainties come from a Monte Carlo over the range of temperature and densities considered.
which assumes optically thin emission.These column density profiles (per unit velocity) are shown in magenta in Figure 4.For all LFA pixels, N HI > N [CII] CNM , meaning there is sufficient Hi to fully explain the [Cii] emission without the need to invoke optical depth effects.
A somewhat more direct comparison of the column densities can be made by assuming some CNM fraction (f CNM ) of the atomic component, since the [Cii] emission is thought to only arise from the cold phase (e.g., Pineda et al. 2013;Fahrion et al. 2017;Herrera-Camus et al. 2017).In Figure 4, we show N HI CNM ≡f CNM N HI for f CNM = 0.5 (thin magenta lines).In all LFA pixels, CNM , meaning that there is enough CNM to fully explain all of the [Cii] emission.The ratio of N HI CNM to N [CII] CNM at the peak of the Hi profile is given in the upper right of each panel in Figure 4.By varying f CNM , we find that the [Cii] emission can be fully explained by collisions with the atomic gas for f CNM > 0.2.
To assess the impact of uncertainties in our assumed temperature and density, we perform a Monte Carlo over both quantities.Using 500 trials, we allow both T and n to vary uniformly over the uncertainty range when calculating the column density.We define the uncertainty on the column density as the standard deviation of the trials.We propagate this uncertainty through to the ratio shown in the upper right of each panel in Figure 4.
There is evidence of a warm atomic phase in some of the spectra.Focusing specifically on Pixel 1, the [Cii] profile is more "flat-topped" than the Hi (both in Figure 3 and Figure 4).This peak in the Hi profile around V LSRK = 200 km s −1 may be indicative of an appreciable WNM component.A similar WNM component is seen in Pixel 2 at a similar velocity.In Pixel 6, however, the atomic gas appears to be dominated by a cold component.Thus there are variations in the overall density and temperature of the atomic component within the outflow of M 82, likely because the entrained material is clumpy.

Molecular Gas in the Outflow
As the [Cii] emission can also arise from the molecular phase, we also calculate N C + assuming collisions with only H 2 (see details in Section A.2). We assume the same temperature and density as for collisions with the atomic gas and we assume an abundance ratio of C/H 2 = 2 × C/H = 3.0 × 10 −4 .We show these effective H 2 column density profiles (per unit velocity) based on the [Cii] (N H2 ) in teal in Figure 5.We repeat the Monte Carlo analysis over the uncertainties in T and n as above.
We show these molecular gas column density profiles in gray in Figure 5, assuming our fiducial X CO = 0.5 × 10 20 cm −2 (K km s −1 ) −1 .We note that adopting the conversion factors used by Leroy et al. (2015) or Yuan et al. (2023) produce negligible changes in the N CO H2 profiles.In all LFA pixels, N CO H2 ∼ 15−40% N [CII] H2 , with most pixels ∼ 20%, meaning that the molecular gas (traced by CO) is not the dominant contributor to the overall [Cii] emission.In summary, because only ∼ 20% of the [Cii] can be excited by the molecular ISM, we conclude that the majority of the [Cii] arises from the CNM.This result has important connections to studies of [Cii] in dust star-forming galaxies at higher redshifts, where the [Cii] 158µm line is visible with interferometers such as ALMA.While it is tempting to use the [Cii] lines as a tracer of molecular gas and/or hence star formation rate (e.g., Herrera-Camus et al. 2015;Zanella et al. 2018;Dessauges-Zavadsky et al. 2020), caution should be used as these results in M 82 suggest that the majority of the [Cii] does not arise from the molecular, star-forming material.

CO-dark Molecular Gas
CO does not perfectly trace the total molecular gas content of the ISM, and this component of missed gas is called CO-dark molecular gas (e.g., Grenier et al. 2005;Langer et al. 2010;Wolfire et al. 2010).By its nature, this phase is difficult to study.In the Milky Way, the fraction of molecular gas not traced by CO varies with galactocentric radius and cloud density, dropping below 20% within 4 kpc of the center and in dense clouds, but reaching nearly 80% in the diffuse ISM and beyond 10 kpc (Pineda et al. 2013;Langer et al. 2014).From their models, Wolfire et al. (2010) found that the fraction of CO-dark gas in PDRs is relatively constant with the ambient radiation field and that the main driver is the cloud's visual extinction, A V -a measure of the dust shielding -where the fraction of CO-dark molecular gas decreases steeply with increasing A V .The fraction of CO-dark molecular gas increases at lower metallicity as the CO molecule is more easily dissociated due to a lack of shielding dust (e.g., Wolfire et al. 2010;Bolatto et al. 2013).
Applying these previous results to the starburstdriven outflow of M 82, we would expect a low CO-dark molecular gas fraction.The metallicity in this region is solar (or slightly supersolar; e.g., Lopez et al. 2020).At the resolution of these upGREAT observations, A V will vary substantially within a beam.Over their entire field-of-view, which mainly covers the central starburst, Förster Schreiber et al. (2001) found A V = 36 ± 16 mag.Extending the models from Wolfire et al. (2010) would imply a CO-dark gas fraction of ∼ 10 − 25%.Because the shielding by dust in the outflow is likely lower than in the nucleus, we estimate that ∼ 25% of the molecular gas may be in a CO-dark phase.Combining this with the fraction of the [Cii] line attributed to the CO-emitting molecular gas (∼ 20%), we would expect a ∼ 5% contribution to the total [Cii] line from CO-dark molecular gas.

Ionized Gas in the Outflow
The [Cii] line can also be excited in the ionized phase of the ISM.It is thought that, when [Cii] is excited in ionized conditions, the majority of the [Cii] emission arises from regions of diffuse ionized gas rather than Hii regions (e.g., Nagao et al. 2011;Contursi et al. 2013).In the KINGFISH sample, Croxall et al. (2017) found that 25 ± 8% of the [Cii] is associated with the ionized gas.In star-forming regions in the center of the Milky Way, Harris et al. (2021) found that PDRs and Hii regions contribute roughly equally to the [Cii] flux.Therefore, while it is likely that the fraction of [Cii] associated with the ionized gas changes with environment, the ionized ISM is not the dominant contributor to the [Cii] emission.Contursi et al. (2013) found that the ionized gas traced by Hα in the outflow of M 82 is kinematically decoupled from the neutral (atomic and molecular) phases and from ionized gas traced by [Oiii] 88µm emission.Moreover, they found that the ionized and neutral phases may not be co-spatial in the outflow.They proposed a scenario where the Hα-emitting ionized gas is more extended, is confined to the walls of the biconical outflow, has a higher outflow velocity (∼600 km s −1 ), and is ionized (at least in part) by shocks between the outflowing X-ray-emitting plasma and the galaxy halo.The [Oiii]emitting ionized gas, on the other hand, is more colli-mated (even narrower than the neutral material), has a slower outflow velocity (∼75 km s −1 ), and is primarily photoionized by the starburst.
For the most part, the [Cii] lines we measure have similar spectral shapes to the Hi and/or CO (e.g., Figures 4 and 5), indicating that the [Cii]-emitting gas is likely coupled to the neutral material and photoionized by the starburst.We do not see evidence for components of the [Cii] line that do not correspond to either the Hi or CO, except perhaps the most redshifted edge of the [Cii] profile in Pixel 1 (Figures 3, 4, and 4).
Another method to determine the fraction of [Cii] associated with the ionized gas is to compare the [Cii] intensity to that of the [Nii] 205µm line.The [Nii] 205µm line is only excited in the ionized ISM and it has a similar critical density as the [Cii] 158µm line.Using this method, Tarantino et al. (2021) found that the fraction of [Cii] arising from the ionized medium is < 12% in the disks of two normally star-forming galaxies.Unfortunately, there are no observations of the [Nii] 205µm line from either Herschel or SOFIA in M 82.Contursi et al. (2013) used the [Nii] 122µm line to place limits on the fraction of [Cii] arising from the ionized medium in the outflow at distances < 1 kpc from the disk.Because the [Cii] and [Nii] 122µm lines do not have similar critical densities, these results are dependent on the electron density, which is largely unknown in the outflow of M 82 (e.g., Shopbell & Bland-Hawthorn 1998;Yoshida et al. 2011).Nevertheless, Contursi et al. (2013) were able to constrain that ∼ 11 − 40% of [Cii] emission arises from the ionized gas in the southern outflow of M 82 (at distances < 1 kpc from the midplane).They note that these estimates will be even more uncertain when applied to regions with a complex mix of different ISM components, which is certainly the case in the outflow of M 82.We can conclude, however, that the ionized gas is not the dominant contributor to the [Cii] emission in the outflow of M 82.

Synthesis
We summarize the contributions to the [Cii] emission in the outflow of M 82 from various ISM phases as follows.From our comparison of the H 2 column density measured from CO compared to that predicted from [Cii], we find that ∼ 20% of the [Cii] emission may arise from CO-emitting molecular gas (Figure 5).We attempt to quantify the contribution of CO-dark molecular gas, finding ∼ 4% of the total [Cii] emission may arise from this phase.Therefore, ∼ 24% of the [Cii] emission in the outflow of M 82 may arise from the molecular ISM.For the contribution of ionized gas, we rely on measurements by Contursi et al. (2013) and estimate that ∼ 20% of the total [Cii] emission may arise from the ionized ISM.Finally, we compare the Hi column density measured from Hi data to that predicted from [Cii] and find that there is sufficient Hi to fully explain the [Cii] emission (Figure 4).Therefore, we can attribute the remaining ∼ 55% of the [Cii] emission to the atomic ISM.

[CII] IN THE DISK OF M 82
As shown in Figures 1 and 2, we detect the [Cii] line at high significance in the disk of M 82.We calculate a total [Cii] mass in the disk of 4 × 10 6 M ⊙ , calculated where SNR > 2 (Figure 2) following Equations A5 and A11.
In Figure 6, we compare representative [Cii] and CO spectra in the disk of M 82 at the center, at the peak of the [Cii] emission, and on the redshifted and blueshifted sides of the galaxy.For this figure, we have spectrally smoothed the CO data to 10 km s −1 to match the [Cii], convolved the [Cii] data to the larger 22 ′′ CO beam, and matched the pixel sizes.Both the CO and [Cii] are reported in main-beam brightness temperature units.We calculate the integrated intensity (moment 0) of the [Cii] and CO on these matched data sets, as shown in Figure 6.Comparisons with the Hi spectra are not possible because of deep absorption features in the Hi over the region where the [Cii] is detected in the disk (Section 2.2.2, Figure 1, and Martini et al. 2018).In general, we find that the CO and [Cii] intensities and spectral shapes are quite similar, in agreement with previous work in this galaxy at lower resolution (e.g., Stacey et al. 1991).In Figure 7, we make a position-velocity (PV) diagram of the matched-resolution CO and [Cii] in the disk of M 82.The velocities for both datasets are reported in the radio velocity convention and in the LSRK frame.We extract the [Cii] and CO PV slices over the entire fully-sampled region of the [Cii] map, covering the central 3 kpc × 1 kpc of the galaxy along its major axis (Section 2.1.1).We collapse the PV slice along the minor axis by taking an average weighted by the intensity of each spaxel.Spaxels with intensities less than the rms noise of the cube (in areas away from emission) are removed.
As shown in Figure 7, the kinematics of the [Cii] and CO generally agree in the inner regions of M 82.In detail, however, there are some differences.The [Cii] velocity appears to rise faster in the center compared to the CO.This disagreement is worse on the redshifted (i.e.eastern) side of the galaxy.The CO on this side of the galaxy appears more kinematically disturbed (e.g., Figure 2 of Leroy et al. 2015 and Figure 2 of Krieger et al. 2021).The Hi on the other hand appears less kinematically disturbed on this side of the galaxy (e.g., Figure 1 of Martini et al. 2018).Therefore, if the [Cii] in the disk primarily arises from the atomic gas (as it does in the outflow) then perhaps this could explain the kinematic differences we see in the PV diagrams, though this is somewhat speculative.The CO and upGREAT [Cii] luminosities in the disk (green-blue filled contours) and outflow (teal circles) of M 82.The teal numbers indicate the pixel number of each outflow pointing.We compare our measurements to those from PACS (Contursi et al. 2013) in the grayscale contours.For all contours, the colors reflect the density of points.We also compare to integrated measurements from the xCOLD GASS survey of star-forming galaxies (red stars; Accurso et al. 2017).

[CII] and CO Luminosity Ratios
A somewhat different angle to assess the contribution of the molecular gas to the [Cii] emission than presented in Section 3.2 is to compare the luminosity ratios of [Cii] and CO.In principle, this ratio is sensitive to the FUV radiation field and the ability of CO to self-shield via dust from the FUV radiation (e.g., Accurso et al. 2017).To calculate the luminosities from the integrated intensities, we follow Equation 1 of Solomon et al. (1997): (2) where ν is the line rest frequency, z is the redshift, and D L is the luminosity distance.Because M 82 is very nearby, we take D L = 3.63 Mpc and calculate z = V sys /c with V sys = 210 km s −1 as the recessional velocity.We show these luminosities in Figure 8 for both the disk (blue-green filled contours) and outflow (teal circles; Table 2) of M 82.
We compare the L [CII] we infer from our measurements to those measured by Contursi et al. (2013).The region mapped by PACS extends farther into the outflow (∼ 1 kpc) compared to the upGREAT map (∼ 0.5 kpc), but not as far as the upGREAT outflow pointings.We match the PACS [Cii] data to the IRAM 30-m CO data by first convolving the PACS cube to a 22 ′′ Gaussian beam using the kernel provided by Aniano et al. (2011).We then match the pixel scales of the [Cii] and CO cubes and re-derive the integrated intensities.We show the PACS [Cii] and matched CO luminosities in Figure 8 as the grayscale contours.The PACS and upGREAT data agree well (as expected from the analysis in Section 2.1.1).
We compare the ratios we derive for M 82 to a sample of 24 normally star-forming galaxies from the xCOLD GASS survey from Accurso et al. (2017) in Figure 8.In the disk of M 82 we find somewhat lower L [CII] /L CO ratios than for normally star-forming galaxies (the median for M 82 is 0.4 dex lower).We note, however, that the measurements from Accurso et al. (2017) are galaxyintegrated, whereas the measurements from the disk of M 82 are in ≈ (125 pc) 2 regions (one pixel).
Another important caveat is that the galaxies studied by Accurso et al. (2017) tend to have metallicities less than solar, whereas M 82 has solar (or slightly supersolar) metallicity (e.g., Lopez et al. 2020).Accurso et al. (2017) found the strongest trend in L [CII] /L CO with metallicity (in 12+log(O/H) units), where higher metallicity systems have lower L [CII] /L CO .They also found a strong trend with the hardness of the radiation field (defined as the ratio of the FUV to near UV (NUV) flux), where harder radiation environments have lower L [CII] /L CO .Both of these trends may help explain why the disk of M 82 has lower L [CII] /L CO than the starforming galaxies analyzed by Accurso et al. (2017).

Fraction of [CII] in the Outflow of M 82
At z ∼ 3 − 7, recent observations with ALMA have revealed extended [Cii] halos around some galaxies extending up to ∼ 10 kpc from the center and containing ∼ 50% of the [Cii] emission (e.g., Rybak et al. 2019;Fujimoto et al. 2019Fujimoto et al. , 2020;;Ginolfi et al. 2020;Meyer et al. 2022, though see also Novak et al. 2020 for counterexamples).We note that these studies are a mix of detections from individual systems and stacks, as well as spatially unresolved and marginally-resolved studies.Simulations have shown that these extended [Cii] halos can be powered by supernova-driven cooling outflows (e.g., Pizzati et al. 2020).
As M 82 is sometimes used as a local anchor for highz star-forming galaxies, it is interesting to constrain the fraction of [Cii] emission in the outflow compared to the disk.The CO in the outflow of M 82 only extends for ∼ 2 − 3 kpc above and below the midplane (Walter et al. 2002;Salak et al. 2013;Leroy et al. 2015;Krieger et al. 2021).Martini et al. (2018) found that the Hi is significantly more extended, reaching ∼ 5 kpc above the midplane and ∼ 10 kpc below (in the direction of M 81 with which M 82 is interacting; see also Yun et al. 1994).
Based on lower resolution CO data, Walter et al. (2002) measured 3.3 × 10 8 M ⊙ of molecular gas in the halo and outflow of M 82, 2.3 × 10 8 M ⊙ in the disk, and 8.0 × 10 8 M ⊙ in the tidal streamers, for a total molecular gas mass of 1.3 × 10 9 M ⊙ .We note that this total molecular gas mass agrees with other more recent measurements (Salak et al. 2013;Leroy et al. 2015;Krieger et al. 2021).Overall, Walter et al. (2002) found that while > 70% of the molecular material resides outside of the disk of M 82, only ∼ 25% of the total molecular gas mass is swept up in the outflow/halo component with the rest being in the tidal streamers.
For the measurements of the [Cii] halos at high-z, outflow/halo and streamer components would be mixed together.However, as we know from M 82, not all of this mass is outflowing so attributing all of the extended [Cii] emission to the outflow can significantly overestimate the [Cii] mass outflow rates in these high-z systems.This rough comparison assumes that the [Cii] and CO masses in each component track one another.
Because our upGREAT observations do not cover the full extent of the outflow of M 82, we cannot directly measure the total [Cii] extent, mass, or flux in the outflow relative to the disk.We will instead extrapolate our [Cii] measurements to infer the total fraction of [Cii] we might expect based on the CO.In the outflow, the average ratio of the peak brightness of the CO and [Cii] line (where [Cii] is detected) is 2.0 ± 0.8, where the uncertainty is the standard deviation (see Figure 3).In the central disk, the peak brightness ratio is nearly the same, with an average and standard deviation of 2.1±1.1 (see e.g., Figure 6).We note that Walter et al. (2002) define the M 82 disk as the inner 1 kpc, which is very similar to the region of the disk where we robustly detect [Cii] emission (e.g., Figure 2).Therefore, since the ratio of the intensities in the central disk and outflow are roughly the same, we might also expect the relative mass ratios to be the same as well.This means that we would expect to find ∼ 25% of the total [Cii] in the outflow, ∼ 18% in the inner disk, with the remaining [Cii] distributed in the streamers.Given that we measure a [Cii] mass of 4 × 10 6 M ⊙ in the disk (Section 4), we would predict ∼ 5.5 × 10 6 M ⊙ of [Cii] in the entire outflow of M 82.This mass corresponds to a total integrated intensity of ∼ 3.5 × 10 4 K km s −1 (Equations A11 and A5) and L [CII] ∼ 1.7 × 10 8 L ⊙ (Equation 2).
Indeed, some observational studies of [Cii] halos at high-z do find evidence of an extended component but without a broad [Cii] profile that would indicate an outflow (e.g., Novak et al. 2020;Spilker et al. 2020;Meyer et al. 2022).In particular, Spilker et al. (2020) stud-Levy et al.
ied molecular outflows in a sample of lensed dusty starforming galaxies at z > 4. They found that > 70% of the galaxies in their sample had clear evidence for a molecular outflow based on OH 119µm absorption.However, none of these galaxies with confirmed molecular outflows had broad [Cii] emission line wings.This suggests that, at least in this population of z > 4 highly star-forming galaxies, that [Cii] is not a robust tracer of outflowing molecular gas at high redshift.
In summary, in M 82 we clearly detect [Cii] in the starburst-driven outflow, though we expect that the outflowing [Cii] accounts for only ∼ 25% of the total [Cii] of the system.This is somewhat different than is observed for high redshift [Cii] outflows, where there is evidence for molecular outflows in broad emission lines but that lack robust [Cii] (e.g., Novak et al. 2020;Spilker et al. 2020;Meyer et al. 2022).

FUV RADIATION FIELD
Within PDRs, photoelectric heating of small dust grains efficiently heats the region and governs the chemistry, and this heating is primarily governed by the density (n) and the far ultraviolet (FUV) radiation field strength (G/G 07 ; e.g., Tielens & Hollenbach 1985;Wolfire et al. 1990).The intensities of lines emitted in a PDR are, therefore, sensitive to these properties as well, and line ratios of FIR fine structure lines and CO can be used to constrain n and G (e.g., Wolfire et al. 1990;Kaufman et al. 1999).In the center of M 82, Kaufman et al. (1999) applied their PDR model to integrated measurements of FIR lines, finding n ∼ 10 4 cm −3 and G ∼ 10 3.5 G 0 .Contursi et al. ( 2013) also used the Kaufman et al. (1999) PDR model to measure the FUV radiation field in the central starburst and southern outflow of M 82 (within 1 kpc of the midplane), finding G = 10 3.1−3.4G 0 in the starburst and G ∼ 10 2.1 G 0 in the outflow.

Radiation Field Constraints from CO and [CII]
From their PDR model, Kaufman et al. (1999) find that, while the ratio of the [Cii] to CO integrated intensities (where both quantities are in units of erg s −1 cm −2 sr −1 ) is mostly sensitive to the column density of C + and the temperature, this ratio does also depend on n and G (see their Figure 9).We, therefore, use the line ratios of [Cii] and CO that we measure in the disk and outflow of M 82 to place new constraints on the FUV radiation field in this region.
Figure 9 (left) shows the integrated intensity ratio of [Cii] to CO.Like has been found previously (e.g., Stacey et al. 1991;Kaufman et al. 1999), ratios range from ∼ 3 − 7 × 10 3 in the disk of M 82.Ratios in the outflow are substantially lower, with Pixels 0, 1, 2, and 6 (purple colors in Figure 9 left) having ratios of 300 − 400.
Using the PDR model developed by Kaufman et al. (1999), we place limits on the strength of the FUV radiation field assuming some density of the material.We interpolate the predictions of the Kaufman et al. (1999) PDR model (their Figure 9) to estimate G for a given [Cii]-to-CO intensity ratio.This estimation of G is shown in Figure 9 (right).For a fiducial density of n = 10 4 cm −3 (following the results of Kaufman et al. 1999), the observed [Cii]/CO ratios in the disk can be explained by G ∼ 10 3−5 G 0 , in agreement (though somewhat circularly) with Kaufman et al. (1999) and Contursi et al. (2013).
In the outflow, we estimate a much lower FUV radiation field, with Pixels 0, 1, 2, and 6 (blue colors in Figure 9 right) having G ∼ 6 G 0 .Although the outflow is less dense than the disk (e.g., Yuan et al. 2023), the result of a much smaller FUV radiation field holds (see Section 6.2 for more discussion on the effect of uncertainties in the assumed density).

Uncertainties on G/G 0 Due to the Assumed Density
A major source of uncertainty in these calculations is the assumed density (n).While we assume a fiducial n = 10 4 cm −3 motivated by the results of Kaufman et al. (1999) in the center of M 82, it is unlikely that the outflow is this dense (as discussed in Section 3 and in Contursi et al. 2013).
In the disk, we allow the assumed density to vary by 0.5 dex (i.e., n = 10 3.5−4.5 cm −3 ).Because the PDR models are not monotonic with density (see Figure 9 of Kaufman et al. 1999), we perform a grid search in steps of 0.1 dex in density and find the minimum and maximum values of G/G 0 at each disk pixel.For the disk, the uncertainty at each pixel is roughly the same (log 10 (G/G 0 ) +0.5 −0.6 ).In the outflow, the density is almost certainly much lower than 10 4 cm −3 .We employ the same grid search as described above over a range of n = 10 2−4 cm −3 .The lower limit encompasses the density limits determined by Contursi et al. (2013) and used in Section 3.1.We assume that our fiducial density is the maximum density in the outflow.The lower uncertainty on G/G 0 in the outflow comes from allowing the assumed density to be 2 dex lower than the fiducial assumption (i.e., n = 10 2−4 cm −3 ).Accounting for the density uncertainties, we find G/G 0 ≈ 10 1.8−5.2 in the disk and G/G 0 ≈ 10 0.1−1.0 in the outflow (excluding each green point in the disk and outflow).Therefore, even considering the uncertainties from the density assumptions, the radiation field in the outflow is substantially lower than in the disk.
7. SUMMARY M 82 is an archetypal example of a starburst-driven outflow and is an ideal laboratory to study the detailed physics of superwinds.Here, we present new velocityresolved observations of the [Cii] 158µm emission line towards the center and southern outflow of M 82, enabled by upGREAT onboard SOFIA.With upGREAT, we mapped the central 3 kpc × 1 kpc of the disk of M 82.In the southern outflow, we use one pointing of the seven-pixel upGREAT array to measure [Cii] at distances of 1 − 2 kpc from the midplane (Figure 1).Below we summarize the main results of this analysis, indicating the relevant figures and/or tables: 1. We detect the [Cii] line out to 2 kpc from the midplane along the southern outflow at 10 km s −1 velocity resolution (Figure 3).This twice as far as previously probed by Herschel PACS (Contursi et al. 2013;Herrera-Camus et al. 2018).
2. We compare the column densities of the atomic medium measured from the Hi data to the CNM column density measured from the [Cii] spectra (Figure 4).Similarly, we compare the column densities of the molecular medium measured from the CO data to the H 2 column density measured from the [Cii] spectra (Figure 5).From these comparisons, we find that the majority (> 55%) of the [Cii] arises from the atomic component.It is likely that the molecular gas (including an estimate of the CO-dark molecular gas) contributes ∼ 25% and that the ionized gas contributes ∼ 20% of the [Cii] emission.
3. We are able to extend the results from Walter et al. (2002) from CO to estimate the total fraction of [Cii] in the outflow of M 82.While the bulk of the [Cii] emitting gas is likely outside of the main disk, only a small fraction is actually outflowing (with the rest located in tidal streamers, for example).This may help inform observations of [Cii] halos at higher redshifts, which sometimes lack outflow signatures.
4. We estimate the strength of the FUV radiation field in the disk and outflow of M 82 using the PDR model developed by Kaufman et al. (1999).In the disk of M 82, we find G ∼ 10 3−5 G 0 , in agreement with previous measurements (Figure 9; Kaufman et al. 1999;Contursi et al. 2013).The FUV radiation field we measure ∼ 1.5 kpc away from the disk in the outflow, however, is 2-3 orders of magnitude lower than in the disk.
Owing to the sensitivity and wavelength coverage of ALMA, the [Cii] 158µm emission line is routinely observed in galaxies at z ≳ 2. Because this line is bright, it is a more attractive tracer of molecular gas than CO in these systems.However, it is crucial to understand the contribution of the various ISM phases to the [Cii] line in order to use this line as a tracer of molecular gas and star formation.The galaxy systems studied so far at z ≳ 2 tend to have high star formation rates, so under-standing the behavior of the [Cii] line in this starburst environment is critical to inform these high-z measurements.Unfortunately, with the end of the SOFIA mission, observations of [Cii] and other FIR lines in the local Universe will be possible only with balloon missions for at least the next few decades.Future facilities in space are needed to more completely understand how the various ISM phases contribute to the [Cii] 158µm line as a function of spatial resolution, environment, and ISM conditions.

A.3. C + Mass Estimate
From the C + column density calculated from Equation A5 (since most of the [Cii] is excited through collisions with atomic gas), we calculate the total C + mass (M C + ) in each upGREAT pointing in the outflow.In each LFA pixel, ) where A pix is the area of each LFA pixel (≈ 156 ′′2 ), d is the distance to the galaxy in Mpc, and v N C + is the sum of the C + column density over all of the channels.We report M C + in each LFA pixel in the pointing along the outflow in Table 1.The total C + mass measured in these observations of the outflow is 4.2 × 10 4 M ⊙ (excluding Pixels 4 and 5 where the [Cii] line is not detected).M C + changes by a factor of ∼ 2 for a factor of 2 change in either the assumed CNM temperature (T ) or density (n CNM ).

Figure 1 .
Figure 1.Composite maps of M 82.In all panels, the teal rectangle shows the fully-sampled region of the upGREAT [Cii] map of the disk (3 kpc × 1 kpc).The footprints of the upGREAT single pointing along the southern outflow are shown as the teal circles, where the size of each circle is the upGREAT beam size (14.1 ′′ ).Top left: An r-band image of M 82 from the SINGS Survey (Kennicutt et al. 2003).Top right: The Hi integrated intensity is shown in orange-purple filled contours, with values linearly spanning 4 − 1500 K km s −1 (Martini et al. 2018).The missing Hi data in the center reflects where accurate Hi information cannot be obtained due to strong absorption against the continuum; this region fully covers the region detected with SNR > 2 in the [Cii] map.The CO integrated intensity is shown in grayscale contours, with values logarithmically spanning 20 − 600 K km s −1 (Krieger et al. 2021).The FWHM beam sizes of the Hi and CO data are shown in the lower right corner in magenta and black respectively.Bottom left: A zoom in, showing the [Cii] integrated intensity in green-blue, with values logarithmically spanning 200 − 1200 K km s −1 .The Hi integrated intensity is shown as the orange-purple filled contours over the same range of values as in the top right panel.The FWHM beam sizes of the Hi and [Cii] data are shown in the lower right corner in magenta and teal respectively.The numbers in each of the teal circles give the pixel numbers of the LFA.The white × marks the galaxy center.Bottom right: The same as the bottom left, but showing the CO integrated intensity (in grayscale contours over the same range of values as in the top right panel) instead of the Hi.

9Figure 2 .Figure 3 .
Figure 2. The upGREAT [Cii] maps of the central 3 kpc × 1 kpc of M 82, showing the [Cii] peak intensity (top left), integrated intensity (Moment 0; top right), mean velocity (Moment 1; bottom left), and linewidth (Moment 2; bottom right).Spaxels with SNR < 2 (relative to the rms noise of the cube) are masked out of the 3D cube before the 2D maps are made.The black rectangle shows the boundaries of the fully-sampled map.The black-outlined circles below the maps show the same values for the single pointings along the outflow.The numbers in each of the circles in the upper left panel give the pixel numbers of the LFA.The solid black circles in the lower left corners show the 14.1 ′′ (≈ 250 pc) beam.The black × marks the galaxy center (Martini et al. 2018).In the lower left panel, the gray horizontal line in the color bar marks the systemic velocity (Vsys) of M 82 (210 km s −1 ; Krieger et al. 2021).
profiles to the measured total Hi column density profiles (N HI ).Modulo optical depth effects, the Hi profiles (measured directly from the Hi data) probe the total atomic gas content from both the CNM and WNM.FollowingMartini et al. (2018), we convert the Hi intensity to a column density where N HI = 1.823 × 10 18 I HI K km s −1 cm s −2(1)

Figure 4 .
Figure 4.The Hi and CNM column densities (per unit velocity) in the outflow of M 82.The teal curves show the CNM column density inferred from the [Cii] spectrum (N [CII] CNM ) assuming collisions with only the atomic gas, T = 250 K, n = 100 cm −3 , and C/H = 1.5 × 10 −4 .The thick magenta curves show NHI measured directly from the Hi data.The thin magenta curves show the CNM column density inferred from the Hi (N HI CNM ) assuming fCNM =0.5.The LFA pixels are oriented as in Figure 3, and the pixel numbers are given in the top left corner of each panel.The number in the top right corner of each panel is the ratio of the peaks of N HI CNM and N [CII]CNM over the shaded portions of the profiles.The uncertainties come from a Monte Carlo over the range of temperature and densities considered.

Figure 5 .
Figure 5. Similar to Figure 4, but showing the H2 column densities (per unit velocity) in the outflow of M 82.The teal curves show the H2 column density inferred from the [Cii] spectrum (N [CII]

Figure 6 .
Figure 6.A comparison of the [Cii] and CO spectra in the disk of M 82.The grayscale filled contours show the CO integrated intensity.The blue-green open contours show the [Cii] integrated intensities (see the lines in the color bar).The black rectangle marks the fully-sampled region of the [Cii] map.The subplots show the spectra of the CO (gray) and [Cii] (teal) extracted at the galaxy center (yellow), at the peak of the [Cii] emission (cyan), and on the redshifted and blueshifted sides of the galaxy (red, blue).Both the CO and [Cii] intensities are reported as main-beam brightness temperatures, and the data sets have been matched to the same spatial and spectral gridding and convolved to the CO beam size (22 ′′ ).The black vertical lines in the panels show the Vsys = 210 km s −1 .Comparisons with Hi are not possible to due strong absorption over this region (see Section 2.2.2).

Figure 7 .
Figure 7.A PV diagram of the [Cii] (filled colored contours)and CO (open grayscale contours) in the center of M 82.This slice is extracted along the galaxy major axis and covers the entire fully-sampled region of the [Cii] map (3 × 1 kpc).TheCO and [Cii]  data have the same beam size, pixel scale, and velocity resolution.The x-axis shows the offset along the galaxy major axis with respect to the center.The y-axis shows the velocity relative to Vsys = 210 km s −1(Krieger et al. 2021).In general, the kinematics of the CO and [Cii] agree, though the [Cii] velocity rises faster than the CO in the center.
Figure8.The CO and upGREAT[Cii]  luminosities in the disk (green-blue filled contours) and outflow (teal circles) of M 82.The teal numbers indicate the pixel number of each outflow pointing.We compare our measurements to those from PACS(Contursi et al. 2013) in the grayscale contours.For all contours, the colors reflect the density of points.We also compare to integrated measurements from the xCOLD GASS survey of star-forming galaxies (red stars;Accurso et al. 2017).

Figure 9 .
Figure9.Left: The ratio of the[Cii]  to CO integrated intensities (in units of erg s −1 cm −2 sr −1 and K km s −1 ), in the same style as Figure2.In the disk, the [Cii] and CO data cubes, from which the integrated intensities are calculated, have the same beam size, pixel scale, and velocity resolution (see Section 4).Right: The estimated FUV radiation field strength based on the PDR model byKaufman et al. (1999) using the ratios from the left plot and assuming a density of 10 4 cm −3 .
R.C.L. acknowledges support for this work provided by NASA through award number 08-0225 issued by the Universities Space Research Association, Inc. (USRA) and by a National Science Foundation (NSF) Astronomy and Astrophysics Postdoctoral Fellowship under award AST-2102625.A.D.B. acknowledges support from the NSF under award AST-2108140.R.H.-C.thanks the Max Planck Society for support under the Partner Group project "The Baryon Cycle in Galaxies" between the Max Planck for Extraterrestrial Physics and the Universidad de Concepción.R.H-C.also gratefully acknowledge financial support from Millenium Nucleus NCN19058 (TITANs), and ANID BASAL projects ACE210002 and FB210003.This work is based on observations made with the NASA/DLR Stratospheric Observatory for Infrared Astronomy (SOFIA), project #08_0225.SOFIA is jointly operated by USRA, under NASA contract NNA17BF53C, and the Deutsches SOFIA Institut (DSI) under DLR contract 50 OK 2002 to the University of Stuttgart.This work is based on observations carried out under project No. 107-19 with IRAM 30 m telescope.IRAM is supported by INSU/CNRS (France), MPG (Germany), and IGN (Spain).The National Radio Astronomy Observatory is a facility of the National Science Foundation operated under cooperative agreement by Associated Universities, Inc.This research has made use of the NASA/IPAC Extragalactic Database (NED), which is funded by the National Aeronautics and Space Administration and operated by the California Institute of Technology.This research has made use of NASA's Astrophysics Data System Bibliographic Services.

Table 1 .
Properties of [Cii] spectra in the outflow