FEEDBACK: a SOFIA Legacy Program to Study Stellar Feedback in Regions of Massive Star Formation

Infrared (IR)- and FIR-surveys of the Milky Way, carried out with Spitzer, WISE, and Herschel, and radio surveys, have revealed many thousand H ii regions (Paladini et al. 2003; Churchwell et al. 2006; Deharveng et al. 2010; Anderson et al. 2011, 2014). Anderson et al. (2011) have shown that approximately half of all H ii regions have a ring-like morphology, commonly referred to as "bubbles." For the other H ii regions, their geometry varies between bipolar structures and more complex regions with fragmented shells, pillars, and globules.¹⁵ Within FEEDBACK, we observe template regions for these different geometries over a large range of size scales (from a few pc to tens of pc).

In the classical view of Spitzer (1968), expanding H ii regions, excited by massive stars, sculpt ionized spherical cavities in their surrounding molecular clouds. Because stellar radiation both ionizes and heats gas up to ∼10⁴ K, it creates a strong overpressure that expands the gas into the surrounding cold molecular cloud. For homogeneous initial conditions, this expansion can be described by a simple analytical solution (Spitzer 1968). For very massive stars, stellar winds are important as well since they drive shocks into the surrounding medium, creating XDRs with temperatures up to 10⁴ K in regions with high gas column density and—in the fully ionized phase—a hot, but very tenuous, plasma with a typical density of n ∼ 1 cm⁻³ and temperatures up to 10⁶ K. The resulting overpressure of this plasma leads to the formation of dense swept-up shells on the surfaces of surrounding clouds (Weaver et al. 1977). The inside of this shell is ionized by the stellar EUV photons, while the shell forms a dense PDR. Energy conduction from the hot plasma to the dense shell across the contact discontinuity that separates the two, results in additional cooling, as well as mass loading of the plasma. While this process is not well understood, it likely controls the dynamics of the shell. Finally, radiation pressure on the dust and gas assists the expansion, or even dominates under extreme conditions (Krumholz & Matzner 2009; Draine 2011; Haid et al. 2016). Note that stellar wind is not the only way to have a shock because the expansion velocity of the ionized gas into the surrounding medium is supersonic. The shock front precedes the ionization front as the expansion continues and the surrounding material is locked into a layer (between the shock front and the ionization front). In simulations (Dale et al. 2014), the wind-blown H ii regions resemble better observed H ii region morphologies than those created by ionization alone. More specific, the relative effects of stellar winds are strongest at early times when the massive star(s) are embedded in dense gas from the molecular cloud. Once this gas has been cleared away, the expanding H ii region dominates the dynamical and morphological evolution of the cloud. It is one of the main objectives of FEEDBACK to discern if H ii regions are wind-driven or ionization-driven, or both, depending on physical parameters such as density, radiation field and spectral type(s) and ages(s) of the exciting star(s).

Each of these mechanisms leads to the creation of a dense shell of gas and the density profile along a cut orthogonal to the ionization front will be asymmetric, with a sharp gradient toward the H ii region. Such profiles were indeed observed, using for example Herschel dust column density maps (Peretto et al. 2011; Schneider et al. 2013; Tremblin et al. 2013, 2014). The PDR is typically trapped in this shell and its contribution will dominate the [C ii] line emission. Hence, we can assess the distribution and the dynamics of the warm PDR gas via the [C ii] line, while low- to mid-J CO lines trace the molecular gas further away from the ionization front as well as the surrounding unperturbed cold molecular cloud material, see, e.g., Peng et al. (2012), Mookerjea et al. (2012), Stock et al. (2015), Nagy et al. (2017), Joblin et al. (2018), Schneider et al. (2018). The high angular and spectral resolution data of the [C ii] line will reveal the distribution and velocity field of the PDR material. Some of the [C ii] emission may originate in the the H ii region itself and this will have to be distinguished through various diagnostics. First, the [C ii] emission distribution, velocity field and line width need to be compared to other data sets, i.e., with diffuse X-ray emission, molecular line data, dust column density and temperature maps. Position-velocity (pv) diagrams and channel-maps of [C ii] will show signatures of near and far molecular walls if the bubbles are 3D shells. Evaluating the energetics, [C ii] observations will also show whether such a [C ii] bubble expands due to the impact of stellar winds or the pressure-driven expansion of the H ii region. In addition, the contribution of strong stellar winds can be assessed by X-ray emission. Several of our targets have been observed by Chandra in the MOXC surveys and show diffuse X-ray emission, tracing stellar wind activity (Townsley et al. 2014, 2018, 2019).

Accompanying observations of CO and/or other molecular lines tracing the molecular cloud will then show how such an expansion continues in the cold and denser gas phase. However, disentangling the primordial cloud structure and the effects of winds and photoionization carving into the clouds will be a challenge (see also Section 2.2). Line profiles of [C ii] are fitted using multicomponent Gaussians to determine surface brightness and identify coherent velocity and cloud structures/components from the channel and pv maps. Because of the high sensitivity of our data, we are able to detect the [¹³C ii] hfs emission in many sources. This allows by comparison of the line profiles study of whether the [C ii] line is affected by self-absorption and thus might mimic unreal kinematics.

A recent [C ii] study of the Orion molecular cloud (Pabst et al. 2019, 2020) displays clear observational signatures of several expanding structures. The Veil nebula dominates the large scale appearance of the region. Over the last ∼100,000 yr, the stellar wind from the massive O7 star θ¹ Orionis C has blown a 4 pc diameter half sphere containing some 2000 M_⊙ of gas expanding at 15 km s⁻¹ toward the low pressure region in front of the OMC1 core. The cavity of the shell is filled by hot plasma from the shocked stellar wind (Güdel et al. 2008). The total kinetic energy of this expanding half-shell is comparable to the total mechanical energy delivered by the wind. The [C ii] observations of the Orion Molecular cloud also reveal expanding shells surrounding M43 and NGC1977. In these two cases, the overpressure in the ionized gas, created by these B0.5 stars, gives rise to a Spitzer-type expansion (Pabst et al. 2020).

The [C ii] observations form the basis for a study on the dynamical response of newly formed stellar clusters and the disruption of their nascent clouds. Dense gas condensations, i.e., the locations of future stars, might form as a result of fragmentation of the neutral material due to the gravitational instability during the expansion of the H ii region. These cold, dense cores in the H ii region/molecular cloud interface of the FEEDBACK sources can be traced by submm dust continuum observations, either already performed by Herschel (Liu et al. 2017; Tigé et al. 2017; Zhang et al. 2020), or using ground-based instruments such as ArTéMiS (Hill et al. 2012a; André et al. 2016; Zavagno et al. 2020) or LABOCA (Massi et al. 2019) and—for detailed studies—ALMA and NOEMA.

So far, an overdensity of young stellar objects is observed at the edges of H ii regions, and up to 25% of the ionized regions show high-mass star formation triggered on their edges (Deharveng et al. 2010; Thompson et al. 2012). In the picture of sequential triggered star formation (Elmegreen & Lada 1977), there should be an age gradient observed in the spatial distribution of young sources in the surrounding molecular cloud (Martins et al. 2010). Furthermore, radiative feedback will impact gas and dust temperatures within the molecular cloud, modifying the initial conditions for collapse and affecting proto- and young stellar objects in different evolutionary phases. However, observational studies of this issue are inconclusive. Taking the example of the Rosette Molecular cloud, Balog et al. (2007) reveal an increase of the average NIR excess fraction for stellar clusters with distance to the cluster center, and Williams et al. (1994) show that star formation activity is more intense in the H ii region/molecular cloud interface region than in the molecular cloud center. On the other hand, Roman-Zuniga et al. (2008) and Cambrésy et al. (2013) found that the relative age differences of the clusters are not consistent with a sequential triggered star formation scenario. Schneider et al. (2012a) argue that star formation takes place in filaments and filament mergers that arose from the primordial turbulent structure but can be locally induced in the direct interaction zone between an expanding H ii region and the molecular cloud.

The [C ii] data, combined with molecular observations, will provide physical conditions, and in particular pressures in these regions and will help to evaluate if triggered star formation can occur in dense, swept up shells (Section 2.3). This will be a difficult task but we will for example investigate the variation of the [C ii] and molecular line velocities and line widths from the interface region into the surrounding cloud. Clumpiness of the cloud needs to be taken into account because it enables UV radiation to penetrate deeper into the cloud, giving rise to [C ii] emission from many PDRs along the line-of-sight. Herschel data provide the distribution of the dust temperature that will be compared to the gas temperature, derived from CO.

The amount of energy and momentum that drive the dynamics of H ii regions need to be evaluated for each source in order to assess which mechanism is the dominant process for each source. Pellegrini et al. (2007, 2011), Lopez et al. (2011, 2014) for example provide a comprehensive overview how the pressure (or energy) terms for direct radiation, dust-processed IR radiation, warm ionized gas, and hot, shock-heated by stellar winds are determined observationally. We here focus on how [C ii] and [O i] observations, together with complementary data, can be used for estimating the energy budget.

The kinetic energy input into the ISM by massive stars will be investigated as a function of star formation activity (cluster size, spectral type, stellar wind) and evolutionary stage of the region both in terms of the generation of large scale motions as well as the local injection of turbulence into molecular clouds. The [C ii] observations reveal the morphology of expanding structures, and their masses are determined from the [C ii] column density and compared to the molecular gas mass, estimated from Herschel dust column density maps and CO maps. The [C ii] line can be affected by high optical depth, so that mass estimates are lower limits, but observations of the optically thin [¹³C ii] hfs emission can be employed. The integrated intensity of the [¹³C ii] hfs emission is expected to be only a few percent of the [¹²C ii] emission (typically 50 times lower), depending on the elemental ¹²C/¹³C abundance ratio and the optical depth of the emission in the main isotope. However, averaging over a large area, the Signal-to-Noise ratio (S/N) of the [¹³C ii] hfs emission can be sufficiently increased to derive the optical depth and hence a lower limit on the mass of emitting gas. For shells expanding toward us, extinction measurements of the stars in the ionized gas can also provide an estimate of the column density through the shell and, hence, its mass. Combined with the measured expansion velocity, this yields the kinetic energy of the gas, which can be directly compared to the thermal energy of any hot plasma present (determined from X-ray observations) and of the ionized gas. The ionizing photon flux and the total luminosity of the region are estimated from optical, radio, and IR observations. The spectral types of the ionizing stars in each cluster have been determined directly through infrared spectroscopy. Hence, we can observationally link the kinetics of the shell to the stellar characteristics (wind mechanical energy, luminosity).

We can also quantify the radiative coupling of PDR gas to the FUV photons. [C ii] line intensities are directly compared via correlation plots to CO emission, IRAC 8 μm and WISE 12 μm emission due to PAHs, and PACS 70, 160 μm (and 100 μm if available) far-IR dust emission, all convolved to the same beam size. The [C ii] line is the dominant cooling line of the gas and by comparing the integrated line intensity to the total infrared dust and PAH emission we empirically derive the heating efficiency in terms of the [C ii]/FIR (or [C ii]/TIR¹⁶ ), [C ii]/CO and [C ii]/PAH emission ratios (see Okada et al. 2013; Pabst et al. 2017; Anderson et al. 2019). Theoretically, the heating efficiency depends on the ionization parameter, $\gamma ={{\rm{G}}}_{0}{T}^{1/2}/{n}_{e}$ with G₀ the intensity of the radiation field in terms of the average interstellar radiation field, T the gas temperature and n_e the electron density (Bakes & Tielens 1994). As PAHs and grains charge up (large γ), the ionization potential increases and fewer photons can further ionize the PAHs/grains. Moreover, the photoelectron has to overcome an increased Coulomb potential, diminishing the energy delivered to the gas. Hence, in a stronger radiation field or lower density region, PAHs and very small grains charge up and the heating efficiency drops. Presently, there is limited data available (Okada et al. 2013; Pabst et al. 2017), showing a decrease with ionization parameter as theory predicts. However, overall, theory seems to overpredict the heating efficiency. Preliminary analysis of the Orion data reveals a large spread in heating efficiency at any given ionization parameter. These variations seem to be linked to the spatial location in the region, indicating a dependence on local conditions not caught by the ionization parameter, or on the past evolution of the region.

FEEDBACK also has the potential for follow-up studies of γ-ray emission from highly dynamic and ionized regions of the ISM. Several of our targets show energetic γ-ray emission, detected with Fermi and/or HESS. These are Cygnus X, observed with Fermi (Ackermann et al. 2011), RCW49 with HESS (Aharonian et al. 2007; HESS Collaboration et al. 2011), and W43 with Fermi (Lemoine-Goumard et al. 2011) and HESS (Chaves et al. 2008). M16, M17, and RCW36 are listed in the Fermi Large Area Telescope Fourth Source Catalog (4FGL, Abdollahi et al. 2020).

The coupling of FUV photons to the gas is a key process that sets the structure and physical conditions in PDRs and their emission characteristics (Hollenbach & Tielens 1999), and regulates the phase structure of the ISM (Wolfire et al. 1995, 2003). The FEEDBACK program allows in depth studies of the structure and characteristics of PDRs on small (∼1 pc) and large (tens of pc) physical scales. The [C ii] line is also an important coolant of warm neutral atomic and CO-dark (Wolfire et al. 2010) molecular gas in PDRs and the cloud selection spans a wide range of physical conditions that are well probed by [C ii]. The (undersampled) [O i] 63 μm maps will be bright in the densest regions illuminated by strong FUV fields. We are also obtaining complementary data in other PDR tracers, namely the [C i] line at 490 GHz, using the 4GREAT receiver on SOFIA and the APEX telescope, and mid- to high-J CO lines using APEX. We sample a wide range of physical conditions including the radiation field (G₀ ranging up to a few times 10⁵), (electron) density (typically from 10² to 10⁶ particles cm⁻³), gas temperature (from 10² to 10³ K), and spectral type of the illuminating star (O9 to O4 and WR).

In order to enable a more efficient way to compare [C ii] and [O i] maps with PDR models, we are upgrading the PDR Toolbox (Kaufman et al. 2006; Pound & Wolfire 2008) at http://dustem.astro.umd.edu/pdrt. The PDR code used to generate its underlying database of line intensities has improved physics and chemistry. Critical updates include those discussed in Neufeld & Wolfire (2016), plus photorates from Heays et al. (2017), oxygen chemistry rates from Kovalenko et al. (2018), and Tran et al. (2018), and carbon chemistry rates from Dagdigian (2019). We have also implemented new collisional excitation rates for [O i] from Lique et al. (2018) (and Lique private communication) and have included ¹³C chemistry along with the emitted line intensities for [¹³C ii] and ¹³CO. The new PDR Toolbox covers many more spectral lines and metallicities and allows map-based analysis so users can quickly compute spatial images of density n and radiation field G₀ from map data. It has been rewritten in Python and provides Jupyter notebooks for data analysis. It also can support other PDR model codes such as KOSMA-τ (Röllig et al. 2006), enabling comparison of derived properties between codes.

KOSMA-τ differs from all other numerical PDR codes in that it is simulating the PDR emission from a spherical model cloud. This allows to model a wider range of astrophysical scenarios compared to plane-parallel geometries. A superposition of many small clumps can represent the inhomogeneous structure of the interstellar medium, while the limit of large clumps approaches the plane-parallel picture. Stutzki et al. (1998) showed that the fractal nature of the interstellar medium can be represented by an ensemble of clumps following a well-defined size distribution. This allows us to successfully model the highly structured PDRs in our observational sample by the superposition of individual spherical model clouds. KOSMA-τ was compared against other models in a dedicated benchmark study in 2007 (Röllig et al. 2007) and the code is continuously updated and improved as new data (e.g., reaction rates, atomic and molecular data) become available. Recent updates particularly refined the chemistry module of the code, including:

• 1.  
  Flexible inclusion of isotopomeric chemistry.
• 2.  
  Improved treatment of linear and cyclic isomers of a given molecule.
• 3.  
  Inclusion of exothermal reaction energies as heating terms in the local energy balance.
• 4.  
  New solution algorithms in order to improve numerical stability.
• 5.  
  Introduction of full surface chemistry.
• 6.  
  Inclusion of time-dependent numerical solvers.

A detailed description of the current state of the code is in preparation (M. Röllig et al. 2020, in preparation). Presently we are working to incorporate non-local transport terms such as diffusion and advection into the code framework. Moreover, we expand the three-dimensional modeling capabilities of KOSMA-τ-3D (Cubick et al. 2008; Andree-Labsch et al. 2017) and include the full continuum radiative transfer in our simulations.

Within the FEEDBACK project, clumpy KOSMA-τ results are compared to the plane-parallel results from the PDR Toolbox to test the diagnostic power of the models given the spatially and spectrally resolved data available.

Energy can be simultaneously injected in radiative (FUV photons) and kinetic (shocks) form. In the regions with evidence for shocked gas emission (for example large observed linewidths of [C ii] and/or [O i]), we will compare our data to the results of dedicated modeling of irradiated shocks (Godard et al. 2019; Lee et al. 2019). These models include the effects of FUV photons on the physical conditions and chemistry of the gas. For moderate-velocity molecular shocks (up to a shock velocity of about 30 km s⁻¹), the most recent reference model was presented in Godard et al. (2019). Given the high number of input parameters, the comparison with observations is only meaningful if the number of observables (emission lines) is sufficient. In FEEDBACK regions, where this is the case and the observations in terms of linewidths justify it, our models will enable us to quantify the effects of kinetic versus radiative energy (see e.g., Figure 12 of Godard et al. 2019). A large grid of models has been run in preparation for the interpretation of FEEDBACK data, thoroughly exploring input parameters such as the pre-shock density, the shock velocity, the external radiation field, the magnetic field strength, and the PAH abundance in the observed regions. A study is in preparation to extend this model toward higher-velocity shocks (up to 60 km s⁻¹). In such shocks, additional FUV photons are generated by the shock itself (see the earlier works of Hollenbach & McKee 1989).

Summarizing, our project will investigate issues such as the relative importance of the [C ii] and [O i] for gas cooling in regions of different physical conditions, the role of self-absorption in [C ii] and [O i] emission and its effect on the analysis of low resolution, Galactic and extragalactic observations, and the origin of the [C ii] deficit in (Ultra)Luminous IR Galaxies.

Massive stars have a profound influence on the overall structure of the ISM and the proposed observations can address such key questions as: Are dense filaments formed due to the interaction of expanding, compressed shells around H ii regions (see Section 2.2)? How is stellar feedback impacting an inhomogeneous molecular cloud or cloud surface? Early studies using the [C ii] line (Stutzki et al. 1988; Meixner et al. 1992; Schneider et al. 1998) have revealed that FUV radiation can penetrate deep (>10 pc) into clouds with high clump to interclump density contrast, thus inducing emission from multiple PDRs along the line of sight. Simulations have also shown that geometry plays an important role. Walch et al. (2012, 2013, 2015) demonstrated that for low fractal dimensions of the cloud, the border of the H ii region is dominated by shells that break up into massive high-density clumps ("shell-dominated" region) while high fractal dimensions lead to the formation of many pillars and cometary globules, containing compact dense clumps ("pillar-dominated" region). Tremblin et al. (2012b, 2012a) showed that UV radiation creates a dense shell compressed between an ionization front and a shock ahead and that density modulations in the interface produce a curved shock that collapses on itself, leading to pillar-like structures that can evolve into globules.

With the [C ii] and [O i] mapping, we are able to study on a small scale (<1 pc up to a few pc) all structures in our sample that are produced under the influence of radiation such as pillars, globules, proplyds, and evaporating gaseous globules (EGGs). This includes the famous "Pillars of Creation" in M16 (Hester et al. 1996) as well as less known, yet equally interesting, features in other regions. The velocity resolved extended [C ii] and [O i] maps are of use to investigate in more detail the dynamics of pillars and globules and allow the detection of velocity gradients, high-velocity outflowing gas and rotation. The spatial emission distributions trace the external PDR surfaces as well as internal heating sources, similar to what was found by Schneider et al. (2012b) for a globule in the Cygnus X region. The observed line intensities and ratios will be applied to the KOSMA-τ model (Röllig et al. 2006) and the PDR Toolbox for modeling the FUV-irradiated regions of the pillars and globules in order to derive the physical conditions. These can then be compared to dynamical models of pillar and globule formation (Williams et al. 2001; Mizuta et al. 2005; Gritschneder et al. 2009; Miao et al. 2009; Gritschneder et al. 2010; Tremblin et al. 2012b, 2012a).

We have selected a sample of sources that were included in the Spitzer/GLIMPSE and Herschel HOBYS (Motte et al. 2010), Gould Belt (André et al. 2010) and Hi-Gal (Molinari et al. 2010) Legacy/Key Programs. Northern sources are observed with flights originating from the Armstrong Flight Research Center in Palmdale, California, and southern sources with flights from Christchurch, New Zealand. Tables 1 and 2 and Figure 1 give an overview of the sample. In the selection we ensured that a wide range in star formation activity is covered, from regions dominated by single O stars, by small groups of O stars, by compact clusters, by super star clusters, and by mini starbursts.¹⁷ Another consideration was to include regions that are probably at different stages of their evolution (see individual source descriptions below). Morphology was an important criterion (see Section 2.1) and we incorporated H ii regions that are (almost) perfect spherical bubbles, multi-bubbles or broken bubbles, bipolar structures, all with or without pillars and globules. Regions with more dispersed star formation activity and clouds with many filaments and massive ridges were also considered. Furthermore, the selected sources include regions dominated by the thermal expansion of ionized gas (Spitzer-type expansion), by stellar wind driven flows, by radiation pressure, by the concerted interaction of multiple expanding H ii regions, by the presence of nearby rich OB associations, and by the action of converging flows associated with the large scale spiral arm structure of the Milky Way. This allows us to study feedback on a wide range of scales both in size (from sub-pc to tens of pc) and energy (energy of a single O star to that of clusters of stars containing tens to hundreds of OB stars).

Zoom In Zoom Out Reset image size

The sample comprises well-known sources dominated by these different dynamical processes (thermal, wind, radiation pressure) and covers the parameter space of star formation activities and evolutionary stages.

Much ancillary data is available for all regions as they have been studied in depth in a variety of Herschel and Spitzer programs. These surveys provide detailed spectral energy distributions, luminosities, dust temperatures, and column densities at spatial scales that are comparable to those that are obtained by the upGREAT mapping of [C ii] and [O i]. Source catalogs from various instruments are available for a census of ongoing star formation activity (pre- or protostellar cores, protostars).

We use existing molecular line data (mostly CO) from publically available large surveys (e.g., the FUGIN project from the Nobeyama telescope for M16, M17, and Aquila, and dedicated molecular line and atomic carbon observations using the APEX telescope (Güsten et al. 2006). We already obtained ¹²CO and ¹³CO 3 → 2 maps for some sources, performed recently with the new LAsMA array on APEX (RCW120 was among these sources and spectra are shown in Section 5.2). LAsMA is a 7-pixel single polarization heterodyne array that allows simultaneous observations of the two isotopomers in the upper (¹²CO) and lower (¹³CO) sideband of the receiver, respectively. The array is arranged in a hexagonal configuration around a central pixel with a spacing of about two beam widths (the beamwidth is 182 at 345.8 GHz) between the pixels. It uses a K mirror as de-rotator. The backends are advanced Fast Fourier Transform Spectrometers (Klein et al. 2012) with a bandwidth of 2 × 4 GHz and a native spectral resolution of 61 kHz. APEX will also be used for observations of the atomic carbon line at 490 GHz and the CO 6 → 5 line.

The [C i] line at 490 GHz was already mapped with the 4GREAT instrument on SOFIA for NGC7538 and Cygnus (in parallel, 3 high-J CO lines were observed). More observations of [C i] and CO lines and of the [O i] 145 μm line for the FEEDBACK sources are planned.

We also recently finished a program to observe radio-recombination lines of hydrogen, helium, and carbon in M17, M16, W40, Cygnus X, and NGC7538 at the Green Bank Telescope.

Information on the individual sources (and of the program in general) is found on the webpages of the FEEDBACK program: https://feedback.astro.umd.edu and https://astro.uni-koeln.de/18620.html. We here give a short summary for each source (see also Tables 1 and 2).

Table 2.  FEEDBACK Sources Physical Properties

Cloud	$\langle \mathrm{FUV}\rangle$	FUV-range	SpT	vlsr	Mass	Cloud Geometry	H ii Region Geometry
	G0	G0		(km s−1)	(103 M⊙)
	(1)	(2)	(3)	(4)	(5)	(6)	(7)
Cygnus X	290	200-7.6 × 104	∼50 O, 3 WR	−3	200	ridge/filaments	bubbles and compact
M16	300	270–8.7 × 103	1 O4, ∼10 late O	25	87	radiation sculpted interfaces	irregular
M17	1295	300–4.6 × 104	2 O4, ∼10 late O	22	483	clumpy	irregular
NGC6334	580	500–8.9 × 104	5 O5–8, 8 B	8	239	ridge/filaments	many bubbles
NGC7538	904	103–3.9 × 105	1 O3	−55	130	clumpy, ring	evolved bubble
RCW49	555	200–5 × 103	2 WR, 12 early O	0	170+	fragmented	irregular/evolved bubble
RCW79	140	35–5 × 103	2 O4, ∼10 late O	−50	146	fragmented	evolved bubble
RCW120	375	100–1.5 × 104	1 O8	−10	20	ring/shell	bubble
RCW36	413	300–1.5 × 104	1 O8, B-cluster	5	58	ridge	bipolar
W40	237	150–8.2 × 103	1 O, 2 B	5	26	clumps/filaments	bipolar
W43	741	700–5.9 × 104	OB, WR cluster	100	6000	ridge/filaments	bubbles and compact

Note. (1) Spatially averaged FUV field in Habing units, determined from Herschel FIR-fluxes at 70 and 160 μm (Schneider et al. 2016). (2) FUV range in Habing units within the [C ii] mapping area, determined from Herschel FIR-fluxes at 70 and 160 μm. (3) Spectral type(s) and number of dominating star(s). (4) Approximate local-standard-of-rest velocity of the source. (5) Approximate mass of the associated molecular cloud derived from Herschel dust column density maps (N. Schneider et al. 2020, in preparation).⁺Mass determined from CO 1 → 0 data (Furukawa et al. 2009). (6) Type/geometry of the molecular cloud region and interface. (7) Type/geometry of the H ii region.

Download table as:  ASCIITypeset image

Cygnus X

The Cygnus X region (Reipurth & Schneider 2008) is one of the richest star formation sites in the Galaxy, mainly excited by the Cygnus OB2 association that contains well over 100 OB stars (Comerón et al. 2002; Wright et al. 2015). Most of the molecular clouds are located at a distance of 1.4 kpc (Rygl et al. 2012), the total molecular gas mass is a few 10⁶ M_⊙ with an average density of ∼60 cm⁻³ (Schneider et al. 2006) and densities >10⁵ cm⁻³ in clumps and cores associated with ongoing star formation. The northern part of Cygnus X contains the prominent DR21 and W75N regions that are located within dense filamentary structures, called "ridges" (Schneider et al. 2010; Hennemann et al. 2012). The average UV-field is high (∼300 G_◦) and reaches peak values up to ∼10⁵ G_◦ in PDRs close to Cyg OB2 (Schneider et al. 2016). The interaction of UV radiation with the molecular clouds created a wealth of structures such as pillars, globules, EGGs, and proplyd-like features (proplyds are evaporating circumstellar disks, but the sources found by Wright et al. (2012) in the immediate environment of the Cyg OB2 association are much larger than typical proplyds). Based on this classification using Herschel data, Schneider et al. (2016) proposed an evolutionary scheme in which pillars can evolve into globules, which in turn then evolve into EGGs, condensations and proplyd-like objects.

We study the northern Cygnus X region because it shows many sequential stages of star formation and—as a large complex—is a key to understanding the role that massive star complexes play in external galaxies. The Cygnus X North region is exposed to a high overall radiation field, mostly arising from Cyg OB2, but also from many local H ii regions. For FEEDBACK, we focus on mapping the DR21 ridge and the "Diamond Ring" (Marston et al. 2004), an H ii region southwest of DR21.

M16

The Eagle Nebula (M16) is a young (1–3 × 10⁶ yr), active high-mass SF region in Serpens. Guarcello et al. (2010) proposed that the stars in the northwest part of the H ii region are younger than the stars in the southeast part and that a 200 pc shell triggered the formation of both M16 and M17 3 Myr ago on much larger scales. Responsible for heating and ionizing the Nebula is the young open star cluster NGC6611, containing four early-type O stars, leading to a high average UV-field of ∼300 G_◦. The transition between H ii region and dense, cold gas is rather sharp with many UV-illuminated features. In particular Hubble Space Telescope imaging of the "Pillars of Creation" has made M16 iconic.

M16 is well suited to study the mechanical (stellar winds) and radiative energy (UV field) incident on the molecular cloud complex, shaping the gas into pillars and other features. In contrast to the Cygnus X region, M16 does not show free-floating objects such as globules or proplyds. We thus focus on pillar structures that are a common morphological phenomenon, appearing on the boundaries of many evolved H ii regions (Dent et al. 2009; Pound et al. 2003; Xu et al. 2019). Herschel studies (Hill et al. 2012b) showed that the cluster affects the temperature within the molecular cloud, modifying the initial conditions for collapse and affecting the evolutionary criteria of protostars, for example increasing the bolometric temperature and the ${L}_{\mathrm{submm}}/{L}_{\mathrm{bol}}$ ratio. The brightest area of M16 is smaller than other massive SF regions and can be fully covered in the [C ii] line, including the Pillars, the Spire, and the Arch.

M17

M17 is located at a distance of 1.98 kpc (Wu et al. 2019) and associated with the highly obscured (A_V > 10) cluster NGC6618 with more than 100 OB stars. The edge-on geometry allows to study feedback of the rich cluster with the molecular cloud all along the interface and into the highly clumpy molecular cloud. The total mass of the complex (from CO observations) is 2 × 10⁴ M_⊙ and can be split into M17 North (M17–N) and M17 Southwest (M17–SW). For M17–N (M17–SW), the density ranges between 10⁴–10⁵ cm⁻³ (10⁶ cm⁻³) (Stutzki & Güsten 1990; Meixner et al. 1992; Wilson et al. 2003). M17–SW has been studied extensively in many different tracers and frequencies. A recent SOFIA [¹³C II] study reveals unexpectedly large columns of warm and cold [C ii](Guevara et al. 2020).

NGC6334

NGC 6334 is a very active star-forming region (see e.g., Persi & Tapia 2008) with a remarkably large number of H ii regions spread across the complex (7 compact and optical H ii regions per square degree) at a distance of 1.3 kpc (Chibueze et al. 2014). Most of the H ii regions have a bubble-like morphology in the IR, FIR and optical, but there are also examples of an expanding wind shell-like H ii region (GUM61) or champagne flow (GUM64b), see e.g., Russeil et al. (2016). The associated molecular cloud consists of a ∼10 pc long, dense filament that is associated with strong extinction and is embedded in a larger ∼50 pc-long, less dense filamentary cloud (Zernickel et al. 2013; Russeil et al. 2016). A number of active star formation sites exist within the dense filament (Brogan et al. 2016; Juarez et al. 2017; Tigé et al. 2017; Sadaghiani et al. 2020) with compact H ii regions detectable at radio wavelengths such as NGC 6334-I and NGC 6334-E being driven by a rich embedded cluster of B-type stars. The average UV-field (∼580 G₀) is above the median within the selected sample.

The gas dynamics in NGC 6334 are dominated by the large number of H ii regions (Russeil et al. 2016), in particular for the dense 10 pc long filament. This one is also undergoing longitudinal collapse (Zernickel et al. 2013) and is also likely compressed by the expansion of the two large H ii region bubbles to the north and south (Russeil et al. 2013). Structure analysis based on the Δ-variance method applied to the Herschel H₂ column density maps identified characteristic scales that can be caused by the injection of energy due to expanding H ii regions (Russeil et al. 2013).

NGC7538

NGC7538 is a small H ii region, which is part of the Cas OB2 complex at a distance of 2.65 kpc (Moscadelli et al. 2009). It is illuminated by a group of O stars of spectral type O6-O9. The dominant ionizing star of NGC7538 has been classified as an O3 or an O5 star (Ohja et al. 2004; Puga et al. 2010), the radio data is consistent with an O5 star (Luisi et al. 2016).

NGC7538 is expanding into a massive star-forming cloud south and southeast of the H ii region and has most likely triggered star formation in the cloud. Especially the extremely young O star IRS1, which is located at the boundary between the expanding H ii and the molecular cloud, is still heavily accreting and surrounded by a cluster of mm-continuum sources, which are probably all young pre-main-sequence sources (Frau et al. 2014).

RCW49

RCW49 is one of the most luminous and massive H ii regions of the southern Galaxy. From recent GAIA observations, its distance is determined to be 4.21 kpc (Cantat-Gaudin et al. 2018), slightly further away than the 4.16 kpc determined by Vargas-Alvarez et al. (2013) from spectroscopy. It hosts the Westerlund 2 (Wd2) cluster, comprising a dozen of OB stars and 30 OB star candidates around it (Tsujimoto et al. 2007; Rauw et al. 2011; Zeidler et al. 2015). A binary Wolf–Rayet star (WR20a), perhaps the most massive binary in the Galaxy, is associated with the cluster and its presence indicates that the cluster is a few 10⁶ yr old. Beyond the cluster core are at least two very massive stars, which include an O4 or O5 star and another Wolf–Rayet star (WR20b). More than 3000 X-Ray point sources were found centered on Wd2 and toward its west lies a pulsar wind nebula surrounded by diffuse X-Ray emission (Townsley et al. 2019). Stellar winds from Wd2 and from the surrounding stars play an important role in the star formation occurring in RCW49.

The velocity dispersion of the associated molecular gas suggests that collision between two CO molecular clouds in the velocity ranges −11 to 9 km s⁻¹ and 11 to 21 km s⁻¹ also contributed to the formation of the stellar cluster Furukawa et al. (2009). Whiteoak & Uchida (1997) reported the presence of two wind driven shells at the center of RCW49. We assume that both the stellar winds and cloud-cloud collision are responsible for the morphological evolution of the bubbles associated with RCW49.

RCW79

The RCW79 bubble is ionized by a cluster of a dozen O stars, the two most massive of which have a spectral type O4–6V/III (Martins et al. 2010). The ionizing luminosity of the ionizing stars was estimated to be 10³ times higher than the mechanical luminosity of their stellar winds, indicating a radiation driven H ii region. RCW79 is spatially encompassed by an almost complete dust ring, with a diameter of 12', corresponding to ∼12 pc at a distance of 4.2 kpc (Russeil et al. 1998). Liu et al. (2017) suggested that triggered star formation might occur around this bubble. Based on Herschel data for RCW79 (Liu et al. 2017), more than 50 compact sources (Class 0 and I) were found in the ionization-compressed layer of cold and dense gas from which 12 are candidate massive dense cores that may form high-mass stars. The core formation efficiency (CFE) shows an increase with increasing density, suggesting that the denser the condensation, the higher the fraction of its mass transformation into dense cores.

RCW120

This source is a well-studied, bubble-shaped H ii region of ∼4.5 pc diameter, excited by an O8V star, CD −38°11636, at a distance of 1.68 kpc (Kuhn et al. 2019). RCW 120 has become sort of a poster-child (Zavagno et al. 2010; Anderson et al. 2015) of the myriad of bubbles discovered by the GLIMPSE survey. The H ii region is surrounded by a dense shell of gas and dust, observed in dust (Deharveng et al. 2009; Zavagno et al. 2010; Anderson et al. 2015; Figueira et al. 2017; Zavagno et al. 2020) and molecular lines (Torii et al. 2015; Kirsanova et al. 2019). Remarkable is an arc of emission visible at 24 μm south of the ionizing star because it may represent the upstream boundary between the wind bubble and photoionized gas (Mackey et al. 2015). There is an ongoing discussion whether the ring-shape appearance of RCW120 is due to an expanding H ii region (Deharveng et al. 2009; Zavagno et al. 2010) or a cloud-cloud collision (Torii et al. 2015). Using Herschel observations Anderson et al. (2012) found that 20% of the total FIR emission of bubble H ii regions comes from the direction of bubble "interiors," the locations inside the PDRs, which suggests a three-dimensional morphology.

RCW36

RCW36 is among the closest H ii regions to the Sun, at a distance of 950 pc (Massi et al. 2019) within the Vela C molecular cloud. The H ii region has a bipolar shape and is surrounded by molecular gas with a dust lane that crosses the bipolar cavity. The embedded cluster (∼350 stars) with the most massive star being a type O8 or O9 is located within the cavity (Baba et al. 2004). The cluster extends over a radius of 0.5 pc, with a stellar surface number density of 3000 stars pc⁻² within the central 0.1 pc. Herschel studies suggest that the bipolar morphology has evolved from its filamentary beginnings under the impact of ionization (Minier et al. 2013).

The bipolar H ii region RCW36 is also an example of a region for the interplay between ionization and structures (bright rims and pillars) around an H ii region. Comparing the dynamics of [C ii] and CO emission (Fissel et al. 2019) will address the question whether filamentary structures can be the location of very dynamical phenomena inducing the formation of dense clumps at the edge of H ii regions. Moreover the [C ii] mapping of this region will lead to a better understanding of the formation of bipolar nebulae as a consequence of the expansion of an H ii region into a molecular ridge or an interstellar filament.

W40

The W40 complex is a nearby (260 pc, Massi et al. 2019) site of high-mass star formation associated with a cold molecular cloud (∼10⁴ M_⊙ ) and includes a blistered H ii region powered by an OB association and two interconnected cavities, forming an hour-glass shape on large scales (a few pc). The main cluster is located just northwest of the narrow waist where the two cavities are joined. The bright-rimmed clouds at the cavity walls show clearly that dense clumps and pillars are illuminated from inside by the cluster. The OB association is comprised of IRS/OS1a (O9.5), IRS/OS2b (B4) and IRS/OS3a (B3) and an associated stellar cluster of pre-main-sequence stars (Shuping et al. 2012).

The W40 molecular cloud/H ii region is one of the few nearby regions with active high mass star formation, hosting an embedded cluster (Könyves et al. 2015). The [C ii] mapping will focus on the bipolar cavity walls illuminated by the cluster to study the feedback of the OB cluster on the surrounding molecular cloud.

W43

W43 is one of the most active star-forming regions in the Galaxy. Its position in the Galactic plane and its radial velocity place it at the junction point of the Galactic long bar and the Scutum spiral arm at 5.5 kpc distance to the Sun. It contains two of the most massive cloud groups of the first Galactic quadrant (W43 Main and South) with a total mass of ∼6 × 10⁶ M_⊙ (Carlhoff et al. 2013; Motte et al. 2014). W43 Main is heated by a cluster of Wolf–Rayet and OB stars (∼3.5 × 10⁶ L_⊙ ) and is considered to be a Galactic mini starburst region (Motte et al. 2003; Bally et al. 2010) since it is undergoing a remarkably efficient episode of high-mass star formation (∼15 high-mass protoclusters and a star formation efficiency (SFE) of ∼0.1 M_⊙ yr⁻¹).

The formation of the associated ionizing cluster was likely supported by the expansion of an older H ii region to the south that triggered the formation and gravitational collapse of the GMC that evolved into W43. Subsequently, UV radiation from the central O+WR cluster compressed the parent cloud toward both low- and high Galactic longitudes, triggering the formation of additional massive stars. Its mini starburst activity is continuously fueled by converging flows, owing to its position at the junction point of the Scutum-Centaurus (or Scutum-Crux) Galactic arm and the Bar (Beuther et al. 2012; Motte et al. 2014).

Its position in the Galaxy makes W43 a very interesting object for studying the formation of molecular clouds and the feedback in a dynamically extreme environment. Despite its distance, it is possible to analyze the details of this cloud, due to its large spatial scale of 150 pc and the large amount of gas at high density.

We have estimated the [C ii] 158 μm surface brightness from 8 μm PAH emission maps (convolved to the upGREAT beam) because the latter are signposts of bright PDR gas and [C ii] and 8 μm brightness are very well correlated with each other (Pabst et al. 2017, 2019; Anderson et al. 2019). The map extent was defined by a certain expected level of [C ii] line integrated intensity (varying between 100 and 250 K km s⁻¹, depending on source) and compared to maps of the far-UV field, derived from 70 and 160 μm Herschel/PACS flux maps (see procedure described in Schneider et al. 2016). Typically, the mapping area starts above 100 G_◦. Figure 1 shows the area to be covered by the tiles of the [C ii] mapping overlaid on IRAC 8 μm maps in color. The emission free reference positions for each map are listed in Table 1.

All sources are mapped with the dual-frequency heterodyne array receiver upGREAT¹⁸ heterodyne receiver (Risacher et al. 2018) with the [C ii] 158 μm (1.9 THz) tuned in the LFA 2 × 7 pixel array and the [O i] 63 μm (4.7 THz) line tuned in the HFA 7 pixel array. The LFA has an array size of 726, and a pixel spacing of 318 while the HFA has an array size of 272 and a pixel spacing of 136 (see Figure 2 for a footprint of the arrays on the sky during one scan). The half-power beam widths are 141) (1.9 THz) and 63 (4.7 THz), determined by the instrument and telescope optics, and confirmed by observations of planets. The receiver noise temperatures are 2000 K for the LFA and 2500 K for the HFA (see Table 1 in Risacher et al. 2018). The backends for both channels are Fast Fourier Transform Spectrometer (FFTS) with 4 GHz instantaneous bandwidth (Klein et al. 2012). The frequency resolution of the raw [C ii] and [O i] data is hardware selected to 0.244 MHz, giving a velocity resolution of 0.04 km s⁻¹ and 0.015 km s⁻¹, respectively. The data cubes provided to the SOFIA science center and used by the FEEDBACK consortium have a resampled velocity resolution of 0.2 km s⁻¹.

Zoom In Zoom Out Reset image size

Procedures to determine the instrument alignment and telescope efficiencies, antenna temperature and atmospheric transmission calibration, as well as the spectrometers used, are described in Risacher et al. (2016, 2018) and Guan et al. (2012). For each flight series, the main-beam efficiencies (η_mb) for each pixel for the LFA and HFA channels are determined, typical values are η_mb = 0.65 (0.69) for [C ii] ([O i]). The forward efficiency is η_f = 0.97.

The telluric [O i] line, originating from the mesosphere, can contribute as a narrow feature in the observed band. The half-intensity width of this non-Gaussian line is typically 1 km s⁻¹. We try to schedule each source at an observing time of the year when the mesospheric line does not appear at the velocity of the bulk emission. However, because of flight planning constraints and because the [C ii] line is the main science driver, this is not always possible. We will endeavor to maximize the scientific return around the atmospheric line, but the line can be rather opaque at the line center. We will need to fine tune the atmospheric model layering of the mesophere to determine the correct transmission on the wings of the atmospheric [O i] line profile. Atomic oxygen is now included in the latest version of the "am" atmospheric code (Paine 2019) used to calibrate the upGREAT data. The large FEEDBACK data set should give us ample data to constrain the distribution of atomic oxygen in the atmosphere (H.-W. Hübers et al. 2020, in preparation).

We will also consider to apply a correction for the absorption following the procedure described in Leurini et al. (2015) and Schneider et al. (2018), assuming that the profile can be characterized by a Gaussian. In summary, for the opacity correction, the absorption strength was adjusted in a way to achieve an adequate interpolation between adjacent unaffected spectral channels.

The thus corrected [O i] data will be delivered later as a Level 4 data product.

Mapping was performed in an optimized array-on-the-fly mapping mode. Each region was split into multiple square "tiles" with 4356 on one side and each square was covered 4 times. The OTF scan speed was selected to attain Nyquist sampling of the LFA beam (dump every 5.2'') while the HFA is undersampled. The total time for one OTF line is then 25.2 s. This is, together with the OFF observation, within the measured Allan variance stability time of the system. The Allan variance of the LFA system is of the order of 80–100 s under stable ambient temperature conditions but can decrease to 30–35 s if there are temperature instabilities or drifts onboard the SOFIA aircraft (Risacher et al. 2018). The first two coverages are done once horizontally and vertically with the array rotated 19° against the scan direction, so that scans by 7 pixels are equally spaced. The second two coverages are then shifted by 36'' in both directions to achieve the best possible coverage for the [O i] line in the LFA array mapping mode (Figure 3). Each tile takes ≈50 minutes to complete. This mode of operation is a slightly modified version of the mode used in the [C ii] survey of the Orion molecular cloud (Pabst et al. 2019) and R. D. Higgins et al. 2020, in preparation, where the second coverage was performed with rotating the hexagonal array by 60° instead of shifting the mapping positions. The mode used in the FEEDBACK project achieves a higher degree of redundancy as each pixel maps a different strip of sky during each scan. Most importantly, though, this mode returns a better, though not fully sampled, coverage of [O i]. The only disadvantage is that the [C ii] maps may show edge effects since some LFA coverage is sacrificed toward 2 sides of the map edges. However, there is still a uniform noise coverage for the LFA. At the beginning of each flight leg for the project, a short calibration observation of a single point in each source is performed to monitor the line intensity if the observations extend over several flights.

Zoom In Zoom Out Reset image size

Spectra are presented on a main beam brightness temperature scale T_mb. Recommended main beam coupling efficiencies are presented in the notes "Project Overview" in the SOFIA science archive. All pixels have been calibrated individually and there are slight efficiency variations depending on the Cycle when the sources were observed. For the sources observed in Cycle 7 and 8, the main-beam effciencies are typically 0.65 for [C ii] and 0.69 for [O i]. The calibrated [C ii] and [O i] spectra were further reduced and analyzed with the GILDAS¹⁹ software. From the spectra, a 3rd order baseline was removed and the spectra were then gridded with 1/σ² weighting (average baseline noise).

The example data we show in this paper (Figures 4 and 5) were obtained during one flight from Christchurch, New Zealand on 10th of June 2019 (flight F579). Three out of the four planned tiles were observed in the mapping scheme described in Section 4.3 (the last tile was not completed so that 72% of the map is finished). The data were reduced as outlined in Section 4.4. The [O i] mesospheric line at ∼8 km s⁻¹ was not much affecting the astronomical [O i] line at around −8 km s⁻¹. We here show preliminary data that were obtained by removing a third order baseline, and excluding the velocity range of the [O i] telluric line.

Zoom In Zoom Out Reset image size

As of April 2020, observations for 32% of the 96h of the observing time of the FEEDBACK program have been performed. The state of observations is constantly updated and is shown on the Cologne FEEDBACK webpage. The most advanced map is that of RCW120 (Section 5.2) with 72% completed. The H ii bubble RCW49 is 65% complete, and RCW36 in Vela 51%. For all other sources, the completeness is less than 20%.

We selected the RCW120 region as an example for the mapping capabilities of upGREAT on SOFIA because these observations are the most progressed. The maps and spectra shown in this subsection are preliminary and intended to give a first impression of the nature and the quality of the data that the FEEDBACK program will deliver. It is not our objective here to discuss in more detail the science, which will be done in M. Luisi et al. (2020, in preparation) and other upcoming publications. Figure 4 shows the line integrated [C ii] map, smoothed to 15'' angular resolution (the [C ii] beam is 141)) and on a 35 grid. The velocity range for integration was −30 to +10 km s⁻¹ which covers all relevant velocity ranges of [C ii] emission. The overall structure is shell-like with two emission peaks in the southwest (up to ∼300 K km s⁻¹) and southeast (up to ∼200 K km s⁻¹), and more diffuse emission from the bubble center. These two peaks correspond to condensations 1 and 2 labeled by Zavagno et al. (2010).

Figure 5 displays an averaged [C ii] spectrum over the southwestern clump, indicated in Figure 4 as a dashed polygon, together with spectra of [O i] 63 μm, and ¹²CO 3 → 2 and ¹³CO 3 → 2 data from APEX, averaged over the same area. Note that the [O i] line shows some baseline structures because we conservatively only removed a baseline order 3 for this preliminary data reduction. The [C ii] spectrum shows a prominent blue tail, reaching up to −30 km s⁻¹ which is not apparent in the other lines. All spectra have a double-peak structure which is most prominent for the ¹²CO 3 → 2 and the [O i] line with a deep dip at ∼−8 km s⁻¹. Because the [C ii] line and the ¹³CO 3 → 2 line (which is commonly considered to be optically thin) also show a dip at that velocity—though not as deep as the one for [O i] and ¹²CO—the interpretation of this profile can be twofold. Either all 4 lines are affected by self-absorption, or there are individual velocity components. Self-absorption is the most likely explanation, though it requires a complex layering of various gas components at different densities and temperatures along the line of sight because the excitation conditions for all lines are distinct (Section 1). While the [C ii] line is thermally excited at temperatures around 100 K and densities of a few 10³ cm⁻³, the CO lines require higher densities (a few 10⁴ cm⁻³), and the [O i] line has the highest critical density of 10⁵ cm⁻³ and necessitates temperatures >200 K. A temperature gradient clearly exists, with higher values in the PDR facing the inside of the shell. Herschel dust observations (Anderson et al. 2010) show temperatures >30 K in the interior of RCW120 (with a hot dust component around 100 K), ∼20 K in the PDR, and ∼10 K in the cool molecular shell. In addition, YSOs in different evolutionary phases from Class 0 to Class I were detected in the southwestern clump (Figueira et al. 2017) and could contribute as internal heating sources. Clumpiness of the shell would lead to higher density, cool clumps embedded in a lower density, warm interclump medium. Self-absorption in the observed CO lines requires cool, low-density gas. Kirsanova et al. (2019) fitted their ¹³CO 3 → 2 and 2 → 1 observations with a foreground cloud with a density of about 50 cm⁻³ and a temperature <60 K. This gas component would also lead to self-absorption in the [C ii] line. A more detailed investigation is out of the scope of this paper, but will be done in a furthcoming study (S. Kabanovic et al. 2020, in preparation). We will then also use observations of the [¹³C ii] line to determine the optical depth of the [¹²C ii] line and employ models of several emission and absorption layers, similar as it was done in Guevara et al. (2020) for M17.

Zoom In Zoom Out Reset image size