Molecular envelope around the HII region RCW 120

TL;DR

This study analyzes molecular emission data of RCW 120, revealing a ring-like structure and surrounding diffuse cloud, which refines understanding of its geometry and gas dynamics in the interstellar medium.

Contribution

It provides a detailed 2D model of RCW 120's molecular structure and confirms the presence of a surrounding diffuse molecular cloud, clarifying its geometry.

Findings

RCW 120 has a ring-like face-on molecular structure.

The C18O(3-2) line profile indicates stalled expansion.

A diffuse molecular cloud surrounds RCW 120, confirmed by extinction maps.

Abstract

The H II region RCW120 is a well-known object, which is often considered as a target to verify theoretical models of gas and dust dynamics in the interstellar medium. However, the exact geometry of RCW120 is still a matter of debate. In this work, we analyse observational data on molecular emission in RCW120 and show that 13CO(2-1) and C18O(2-1) lines are fitted by a 2D model representing a ring-like face-on structure. The changing of the C18O(3-2) line profile from double-peaked to single-peaked from the dense molecular Condensation 1 might be a signature of stalled expansion in this direction. In order to explain a self-absorption dip of the 13CO(2-1) and 13CO(3-2) lines, we suggest that RCW120 is surrounded by a diffuse molecular cloud, and find confirmation of this cloud on a map of interstellar extinction. Optically thick 13CO(2-1) emission and the infrared 8 um PAH band form a neutral envelope of the H II region resembling a ring, while the envelope breaks into separate clumps on images made with optically thin C18O(2-1) line and far-infrared dust emission.

Tables

Table 1. Table 1: Basic observational parameters.

Lines	Reciever	Freq. range	Beam	rms
		(GHz)	(″)	(K)
13CO (2–1)	APEX-1	220.0–221.0	29	0.1
C18O (3–2)	APEX-2	328.9–329.9	19	0.2
13CO (3–2)	APEX-2	330.1–331.1	19	0.2

Full text

††pubyear: 2015††pagerange: Molecular envelope around the HII region RCW 120–Molecular envelope around the HII region RCW 120

1 Introduction

Spitzer and WISE maps of the Milky Way Galaxy have revealed more than five thousand ring-like structures, which are commonly referred to as infrared (IR) bubbles, around H ii regions excited by young massive stars, e.g. Churchwell et al. (2006, 2007); Deharveng et al. (2010); Anderson et al. (2011); Anderson et al. (2014); Simpson et al. (2012), and also recent studies by Topchieva et al. (2017, 2018a). Spitzer 8 $\mu$m and WISE 11 $\mu$m emission is commonly attributed to polycyclic aromatic hydrocarbon (PAH) molecules, which are excited by the absorption of ultraviolet photons from massive stars (see, for example, classical studies by Sellgren, 1984; Leger & Puget, 1984) in the photo-dissociation regions (PDRs) on the border between the H ii region and surrounding neutral material.

While the word ‘bubble’ implies a 3D structure, the Spitzer bubbles look like rings in the plane of the sky, so their actual geometry might be different. This is an important issue, as it may shed light on the detailed structure of matter distribution in star-forming regions, on the interaction of massive stars with their parent molecular clouds, on the dust evolution in the vicinity of massive stars, and on the possibility and extent of triggered star formation. But the problem of the actual geometry of IR ring nebulae is still far from being solved.

An important study of this issue was performed by Beaumont & Williams (2010). They utilised observations of CO(3–2) and HCO+(4–3) molecular lines to study gas morphology in the vicinity of 43 Spitzer bubbles, and found that the distribution of neutral material around these bubbles lacks convincing signatures of near and far molecular walls, which would be expected if the bubbles were 3D shells. In addition, in molecular emission, the inner regions of many bubbles in their sample are fainter than in the surrounding material. Finally, the kinematic structure of the bubbles does not show any signs of expansion. Beaumont & Williams (2010) therefore concluded that the bubbles and H ii regions in their study are enclosed in molecular rings, rather than spherical shells, indicating that the parent molecular clouds are flattened. While a flattened shape of a molecular cloud is evident, when an H ii region is observed edge-on as a bipolar structure (e.g. Samal et al., 2018), a ring-shaped region requires a more thorough analysis. Despite this fact, ring-like nebulae are often implicitly assumed to have a 3D geometry (e.g., Everett & Churchwell, 2010; Ochsendorf et al., 2014).

In this paper we consider one of the most well-known examples of an infrared bubble, RCW 120. The H ii region RCW 120 is one of the closest to the Sun, at a distance of 1.34 kpc Russeil (2003). The parallax of the ionizing star of RCW 120, CD-38 11636, measured by Gaia DR2 (Gaia Collaboration, 2018), corresponds to a distance of $0.5-1.1$ kpc. However, this estimate has large uncertainties, as the astrometric solution for the star is poor. Therefore, in this work, we adopt 1.3 kpc as a distance to RCW 120 to keep consistency with our previous studies (Pavlyuchenkov et al., 2013; Akimkin et al., 2015). An effective temperature of CD-38 11636 is 37000 K, as found by Martins et al. (2010).

RCW 120 has a nearly perfect ring-like shape in Spitzer and Herschel infrared images (Deharveng et al., 2009; Anderson et al., 2012). The infrared ring is claimed to be a projection of neutral material collected during the period of expansion of the H ii region. Tremblin et al. (2014) analysed the column density maps from Herschel observations of the neutral environment of RCW 120, and found that it consists of a compression zone induced by the expansion of the ionized gas and surrounding turbulent medium. This apparent simplicity has made RCW 120 a template object to study various aspects of the interaction of massive stars with their parent molecular clouds, including triggered star formation.

Signatures of triggering in RCW 120 are seen in the fine region structure: the shell of neutral material breaks down into up to eight or nine dense condensations, seen in continuum observations at 1.2 mm and 870 $\rm\mu$m by (Zavagno et al., 2007) and (Deharveng et al., 2009), respectively, who argue that the condensations might have been formed as a result of fragmentation of the neutral material due to the gravitational instability during the expansion of the H ii region. There are numerous young stellar objects embedded in the condensations, including a massive compact source in Condensation 1, adjacent to the ionization front of RCW 120 (see also Zavagno et al., 2010). Figueira et al. (2017) recently re-examined the Herschel data and found that 80% of the massive compact sources in RCW 120 belong to Condensation 1. ALMA observations by Figueira et al. (2018) reveal the importance of turbulence or magnetic fields to explain the low observed fragmentation, which is inconsistent with a pure thermal Jeans model of gravitational instability.

The geometry of RCW 120 remains a matter of debate. It has been considered as a bubble viewed in projection in series of works by Deharveng et al. (2005); Zavagno et al. (2007); Deharveng et al. (2009); Zavagno et al. (2010). However, Anderson et al. (2010) found a lack of emission in the interior of RCW 120 at wavelengths $\leq 250$ $\rm\mu$m, which suggests a two dimensional ring, rather than a three dimensional bubble.

Pavlyuchenkov et al. (2013) tried to fit the Herschel observations using a 1D spherical model of an expanding H ii region, and showed that the 1D model predicts appreciable emission at 100 $\rm\mu$m from the central part of the object and the ionizing star (the front and back molecular walls), while this emission is essentially absent in the observations. They concluded that dust envelope of RCW 120 may not be spherical, but instead has cylindrical symmetry viewed along the axis of the cylinder. In particular, the 1D model of an H ii region expanding into a neutral molecular cloud predicts that the maximum velocity difference should be observed between the near and far sides, centred at the ionizing star.

Zavagno et al. (2007) analysed the H i emission from the Southern Galactic Plane Survey (SGPS, McClure-Griffiths et al., 2005) and found its presence in a broad area around RCW 120. They looked for the near and far expanding sides of the neutral environment of the H ii region and did not find it. Anderson et al. (2015) studied the expansion of the neutral material around the H ii region with the $J=1-0$ transition of 12CO, 13CO and C18O molecules using the 22-m Mopra telescope. They made no claims about the geometry, because they found no velocity difference between the suggested near and far sides. They concluded that if the H ii region is expanding, then the expansion velocity ($V_{\rm exp}$) should be $\leq 1$km s*-1*. We note that in the series of simulations by Pavlyuchenkov et al. (2013); Akimkin et al. (2015, 2017), where the parameters of RCW 120 were used as a basic model, the value of $V_{\rm exp}$ was indeed found to be about 1–1.5 km s*-1*. Recently, Marsh & Whitworth (2019) re-examined Herschel data using a Bayesian procedure and surprisingly found that spatial distribution of warm dust in the H ii region is consistent with the projection of a spherical shell, while colder peripheral dust has a ragged clumpy structure.

The aim of this paper is to use available data about molecular emission around RCW 120 and a 1D spherically symmetric simulation of the H ii region, developed by Pavlyuchenkov et al. (2013); Akimkin et al. (2015, 2017), to find a more reliable model for the RCW 120 geometry.

2 Observations, archival data, and numerical modelling

2.1 Molecular emission

We observed CO molecular lines in RCW 120 at several locations in Condensation 1 around the peak of 870 $\rm\mu$m emission found by Deharveng et al. (2009) ($\alpha_{2000}=17^{\rm h}12^{\rm m}08^{\rm s}$ $\delta_{2000}=-38^{\circ}$30′ 45″) with the APEX telescope during 5 hours from 7th to 9th July of 2009 in service mode (project number is O-083.F-9311A-2009) using SHeFI recievers (Belitsky et al., 2006; Vassilev et al., 2008). We used position switching mode with OFF position at ($\Delta\alpha,\Delta\delta$)=(1800″,0″) relative to the reference position. Observations were done in good weather conditions, with the amount of precipitable water vapour in between 0.4 and 0.7 mm. Our choice of APEX instrumentation and line selection is shown in Table 1. The Fast Fourier Transform Spectrometer with 8192 spectral channels provided a spectral resolution about 0.17 km s*-1*. Reduction and calibration of the observations were performed by the APEX staff.

We also use APEX archival data at a frequency of 220 GHz to reveal the large-scale distribution of molecular gas around RCW 120. The data were obtained as a part of the SEDIGISM project (E-193.C-0584A-2014, PI: F. Schuller Schuller et al., 2017). The region around RCW 120 (source G348.25+0.38) was observed twice (in 2014 and 2015). We use calibrated data from the observational session in 2015 and in the present study concentrate mainly on 13CO(2–1) at 220.3987 GHz and C18O(2–1) at 219.5604 GHz. We smoothed and regridded the original archival data at 220 GHz to obtain 1-$\sigma$ noise level of 1 K on $T_{\rm mb}$ scale at 0.2 km s*-1*, i.e. an order of magnitude higher than in our data from 2009 (see Table 1). Since only a portion of the sky with galactic latitude $b\leq 0.5^{\circ}$ was covered by Schuller et al. (2017), there are no data for the north-west part of RCW 120.

Analysis of the data cubes was performed using GILDAS111http://www.iram.fr/IRAMFR/GILDAS.

2.2 Optical depth and column density

We calculate the 13CO column density ($N_{\rm{}^{13}CO}$) using a standard LTE analysis described by Mangum & Shirley (2015) (Eq. 80) and an optical depth correction factor introduced by Goldsmith & Langer (1999). This procedure implicitly assumes the same, constant excitation temperature for 13CO and C18O along the line of sight. The 13CO-to-C18O abundance ratio is assumed to be 7.5, based on the isotopic ratios of 13C/12C and 16O/18O from Wilson (1999). To convert $N_{\rm{}^{13}CO}$ to molecular hydrogen column density we use a relative 13C abundance of $2\cdot 10^{-6}$.

2.3 Dust extinction

We use extinction maps based on the 2MASS near-infrared photometry from Juvela & Montillaud (2016) to study the large-scale dust distribution around RCW 120. We downloaded $A_{\rm J}$ extinction maps with a spatial resolution of 3.0″, produced by the NICEST method. Parameters for $R_{\rm V}=3.1$ and 5.5 given in Cardelli et al. (1989) were used to calculate extinction in $A_{\rm V}$ from the $A_{\rm J}$ data.

2.4 Numerical model

Observations provide us with information on the large-scale distributions of the 13CO(2–1) and C18O(2–1) emission, and this allows us to probe the geometry and kinematics of the molecular envelope around RCW 120. For this purpose, we adopt the model of an expanding H ii region MARION, which was developed by Kirsanova et al. (2009) and further modified by Pavlyuchenkov et al. (2013); Akimkin et al. (2015, 2017). While many other models for RCW 120 have been proposed, invoking various physical effects such as stellar wind (Sánchez-Cruces et al., 2018), a moving star (Mackey et al., 2016), or a radiation-driven dust wave (Ochsendorf et al., 2014), we consider the ‘classical’ picture of an expanding H ii region, relying on our experience in reproducing the observed intensities of dust emission from 8 to 500 $\rm\mu$m with a detailed radiative transfer model (see Pavlyuchenkov et al. (2013) and Pavlyuchenkov et al., in prep.).

We performed new simulations of the expanding H ii region with the MARION code, using the same spherically symmetric model as Akimkin et al. (2017), but with modified initial parameters. We updated the stellar radiation field using the calibrations of Martins et al. (2005) with the following star parameters: $T_{\rm eff}=37000$ K and $\log Q_{0}=48.8$ s*-1*. The initial hydrogen number density was raised from $n_{\rm gas}=3\cdot 10^{3}$ cm*-1* to $10^{4}$ cm*-1*, since the latter value allows us to get a better fit to the observed intensities of dust emission at 100–500 $\rm\mu$m (Pavlyuchenkov et al., in prep.). An age of 590 kyr is adopted as a the age of the RCW 120 region, as this time is required for the dense molecular layer to reach the 1.2 pc radius, which is the minimum projected separation between the ionizing star and the molecular envelope, given the distance to RCW 120.

We use the two-dimensional non-LTE code URAN(IA) (Pavlyuchenkov & Shustov, 2004; Pavlyuchenkov et al., 2008) to calculate the emergent line profiles. The code is designed to solve the system of radiative transfer and balance equations with the accelerated-iterations method, where the mean intensity is calculated using a Monte-Carlo approach. One of the important parameters which affects the width and shape of the emission profiles is the non-thermal velocity $V_{\rm nth}$, which is associated with an uncertain turbulent field. To study the effect of $V_{\rm nth}$, we assumed a uniform non-thermal velocity distribution (micro-turbulent velocity) over the entire region and tested two values, namely, 0.3 km s*-1* and 1.0 km s*-1*.

3 Observational perspective on neutral material around RCW 120

3.1 13CO and C18O emission around the HII region

Maps of the 13CO(2–1) line emission around RCW 120 are shown in Fig. 1. A ring-like molecular envelope surrounds the H ii region from the west, south and east. The envelope contains several condensations initially distinguished by Zavagno et al. (2007). The two most prominent condensations, Condensation 1 and Condensation 2, are designated in Fig. 1. Emission at 870 $\mu$m from the ATLASGAL archive (Schuller et al., 2009) is over-plotted on the 13CO(2–1) map to show the condensations more clearly. The integrated intensity of the 13CO(2–1) emission is 10 times higher at the position of the envelope, compared to the position of the ionizing star. 13CO(2–1) emission is observed within the H ii region itself only in the north-eastern part of RCW 120. The spatial distribution of the C18O(2–1) emission is similar to that of 13CO(2–1); however, while the integrated line intensities are almost the same in both condensations, in Condensation 2 the peak of the 13CO(2–1) emission is 1.5 times higher and the lines are narrower.

The 843 MHz radio continuum map from the Sydney University Molonglo Sky Survey (SUMSS, Bock et al., 1999) is also shown in Fig. 1. The signal of the radio continuum map at about a 10-$\sigma$ level has an ellipse-like appearance, filling the 13CO(2–1) shell. The radio continuum emission is concentrated to the south of the ionizing star. The peak of the 1.4 GHz radio continuum emission from the NRAO VLA Sky Survey (NVSS, Condon et al., 1998) is also shifted to the south of the star (not shown in Fig. 1).

The compact sources identified by Figueira et al. (2017) as reliable are shown in Fig. 1 as black circles. The region with the highest 13CO(2–1) integrated intensity in Condensation 1 corresponds to the concentration of the sources, but there is also a region in Condensation 1 with bright 13CO(2–1) emission and no sources with reliable SEDs. Only sources with unconstrained SEDs are observed in this part of the shell (see Fig. 7 in Figueira et al., 2017).

3.2 Gas kinematics

According to Anderson et al. (2015), the difference between the velocity peaks corresponding to the near and far sides of the expanding material does not exceed 2 km s*-1*. This implies that the expansion velocity of RCW 120 cannot be higher than 1 km s*-1*. To isolate a dense molecular shell formed due to the H ii region expansion, we separated all the velocity channels into several intervals, looking for an interval with the most uniform spatial distribution of line emission around the H ii region. Channel maps of the 13CO(2–1) emission are shown in Fig. 2 for three different intervals. We found that the most uniform ring-like distribution is demonstrated by the integrated emission between $-8\leq V_{\rm lsr}<-6$ km s*-1*, especially to the south-east and north-east of RCW 120. Blue-shifted ($V_{\rm lsr}<-8$ km s*-1*) and red-shifted ($V_{\rm lsr}>-6$ km s*-1*) 13CO(2–1) emission shows up only locally in the shell. The velocities and 13CO(2–1) line profiles are in general more regular in Condensation 2 than in Condensation 1. There are two bright areas on the red-shifted channel map in Fig. 2 around $\phi=60^{\circ}$ and $\phi=270^{\circ}$, respectively, where the azimuthal angle $\phi$ is measured from south to west, with the ionizing star at the origin. Emission of 13CO(2–1) in the blue-shifted velocity range is observed to the south of the molecular envelope. High-mass compact sources with constrained SEDs from Figueira et al. (2017) in Condensations 1 and 2 are absent from the region of the blue-shifted 13CO(2–1).

We present position-velocity (PV) diagrams for various directions from the ionizing star from the periphery of the H ii region, both with and without compact sources, in Fig. 3 (see Fig. 2 for the locations of the cuts with different $\phi$ for the PV diagrams). To find out whether compact sources are located inside the collected molecular layer, we plot the projected distance between the ionizing star and the compact sources on the diagrams. Lines of 13CO(2–1) in Condensation 1 (rows 1–3 in Fig. 3) clearly have double-peaked profiles, with a velocity difference of $1-3$ km s*-1* between the peaks. Single-peaked profiles symmetric or skewed to the red are observed in Condensation 2 (rows 4–5 in Fig. 3). The C18O(2–1) lines have either single- or double-peaked profiles. When they are double-peaked, the velocity difference between the C18O(2–1) peaks is about 1 km s*-1*. The C18O(2–1) peaks fall in the intermediate velocity interval mentioned above. The self-absorption effect is more significant for 13CO(2–1): the C18O(2–1) peaks correspond to the 13CO(2–1) dips.

Compact sources 2, 9, 10 and 39 in Condensation 1 (source index number in Table 5 in Figueira et al., 2017) are projected closer to the ionizing star than the dense molecular shell, but source 1 is projected farther from the star than the bright 13CO(2–1) and C18O(2–1) emission. All compact sources in Condensation 2 are projected closer to the star, which could mean that source 1 formed independently from the molecular shell collected by the expansion of the H ii region.

3.3 Detailed study of Condensation 1

In order to demonstrate possible complications that may arise in the analysis of the molecular shell kinematics, we consider line observations separately in Condensation 1. Emission of 13CO(2–1), 13CO(3–2) and C18O(3–2) was found at all observed positions in Condensation 1. Examples of spectra in selected positions are shown in Fig. 4. The 13CO lines have double-peaked profiles at almost all observed positions (confirming results obtained with lower S/N ratio from the APEX archive) which are shown in Fig. 3, but sometimes demonstrate more complicated profiles (spectra b and c in Fig. 4, black curves). Both 13CO(2–1) and 13CO(3–2) double-peaked lines have red asymmetry at the majority of the observed positions, including the compact source 2 (spectrum e). The asymmetry is less prominent at the position of compact source 1 (spectrum f). The brightness of the 13CO(2–1) and (3–2) emission is almost the same, which implies a significant optical depth of the 13CO lines in a diagonal strip from the south-east to north-west along the molecular envelope of the H ii region. The C18O(3–2) lines are single-peaked in almost all observed positions, but we find red and blue C18O(3–2) line wings from the compact source 2 (spectrum e). The peaks of the C18O(3–2) line coincide with the 13CO(2–1) and 13CO(3–2) dips, confirming self-absorption in the 13CO emission. There is a velocity gradient from –8.2 km s*-1* at the south-east to –7 km s*-1* at the north-west, which agrees with the gradient demonstrated on the channel maps in Fig. 2.

Double-peaked C18O(3–2) line profiles are observed in positions near the border between the molecular shell and the PDR (e.g. spectrum a in Fig. 4). The C18O(3–2) dips coincide with the 13CO(3–2) dips in these directions with the velocity difference up to 2–3km s*-1* between the peaks. We suggest that the velocity difference might be caused by the projected relative motion of different parts of the expanding shell. The C18O(3–2) line becomes single-peaked farther from the PDR border, which means that the proposed expansion of the shell does not involve the dense material of Condensation 1.

We find blue asymmetry of the double-peaked 13CO lines in the eastern part of Condensation 1 (spectrum d) in agreement with the appearance of significant emission in the blue-shifted velocity interval ($V_{\rm lsr}<-8$ km s*-1*) visible in Fig. 2. The C18O(3–2) line has a single-peaked profile, with the peak corresponding to the 13CO self-absorption dip. Poor spatial coverage of Condensation 1 by single-point observations from 2008 session does not allow us to infer a reason for the change in line asymmetry. Weak C18O(3–2) emission is observed in the outer part of Condensation 1 (spectra g and h), but the 13CO lines still show self-absorption features.

Overall, we conclude that local phenomena significantly impact the gas kinematics in Condensation 1 and hamper the analysis of the global kinematics of the entire region.

3.4 Dust extinction around RCW 120

Beaumont & Williams (2010) noticed that many infrared bubbles are embedded in extended molecular envelopes. To verify if this is the case for RCW 120, we checked available infrared images of RCW 120, looking for additional surrounding neutral material. We inspected 2MASS images at 1.24, 1.66 and 2.16 $\rm\mu$m and WISE images at 3.6 $\rm\mu$m (see the images online through Aladin at http://cdsportal.u-strasbg.fr/?target=RCW%20120), and indeed found obscuration of background stars in a broad area around RCW 120. As we move to longer wavelengths, obscuration becomes less significant and more patchy. Instead of continuous absorption, we see a number of isolated cloudlets in the 12 $\rm\mu$m WISE image, delineating the border of the absorbing cloud. Sharp linear absorption across the face of the H ii region is visible at optical wavelengths (see also H$\alpha$ images by Zavagno et al., 2007; Anderson et al., 2015) might be a part of the absorbing cloud. The extent of the cloud is 30′ in the plane of the sky, as can be seen in the extinction map shown in Fig. 5 in $A_{\rm J}$ units. The minimum $A_{\rm J}=3-4$ mag is observed towards the ionizing star, which corresponds to $A_{\rm V}=8-12$ mag (depending on the extinction law used). The angular size of the cloud corresponds to 11.3 pc at a distance of 1.3 kpc. RCW 120 is isolated from other star-forming regions, and situated at some distance from the galactic plane. Therefore, we propose that the absorbing cloud is physically related to the H ii region, but not just projected at the same line of sight. Visual extinction $A_{\rm V}=4.36$ mag determined by Zavagno et al. (2007) using optical and infrared photometry data of the ionizing star is less than our estimated value, suggesting that the absorbing cloud is partly in front of and partly behind the H ii region.

4 Simulated line emission in RCW 120

The results of a simulation of RCW 120 with the MARION code are shown in Fig. 6. Ionized gas with $n_{\rm H^{+}}\approx 10^{2}$ cm*-2* is surrounded by a dense shell of molecular hydrogen with $n_{\rm H_{2}}\approx 10^{5}$ cm*-2*. The shell is truncated by a shock at $\approx 1.25$ pc preceding the ionization front, located at $\approx 1.15$ pc. The dense shell is surrounded by an undisturbed molecular gas envelope with $n_{\rm H_{2}}=0.5\cdot 10^{4}$ cm*-2* and an outer radius of 1.6 pc. We adopt this value for the outer radius as we see molecular emission up to 300″ offset from the ionizing star. The maximum velocity of the moving gas is about 1 km s*-1*. We note that within the H ii region the carbon is double-ionized, while near the inner boundary of the dense shell it resides mainly in the form of C+, and in the densest part of the shell carbon is bound to CO. Thus, the CO-bearing region is characterized by $T_{\rm gas}<60$ K and $10^{4}<n_{\rm gas}<10^{6}$ cm*-3*.

4.1 Line emission in a spherically-symmetric molecular shell around HII region

We start with the assumption of a 3D expanding bubble and present results for a spherically symmetric model. In Fig. 7, we show PV diagrams for the 13CO(2–1), C18O(2–1) and C18O(3–2) emission along the radial direction. Dashed vertical lines on the plots show the location of the dense shell compressed due to expansion of the H ii region into molecular gas. The simulated intensity of 13CO(2–1) from the dense shell is close to the observed values (see Fig. 3), while the C18O(2–1) and C18O(3–2) intensities are about 2 times higher than those observed (see also Fig. 4). The major discrepancy between the observed and simulated PV diagrams is the presence of the bright CO emission over the entire extent of the modelled H ii region (corresponding offsets are $<200$″), which is not observed in RCW 120. The simulated 13CO(2–1), C18O(2–1), and C18O(3–2) profiles from the inner parts of the region are strong and double-peaked, which is also not observed.

One may suppose that this discrepancy is related to a value of $n_{\rm gas}$ too high in the shell. Indeed, an overestimated initial density would result in an over-dense molecular layer, making both 13CO and C18O emission optically thick over the entire extent of the H ii region. We checked if this is possible to resolve the inconsistency by artificially decreasing the 13CO and C18O abundances. PV diagrams for the abundances decreased by a factor of 8 are shown in Fig. 7. As expected, the intensities of all CO transitions become lower, which brings the C18O(2–1) and C18O(3–2) peak temperatures much closer to the observed values. However, the extended double-peaked emission from the H ii region is still significant in all three maps. Further reducing the 13CO and C18O abundance does not make sense, since the model intensities would become much lower than those observed. Therefore, we conclude that a spherically symmetric model cannot reproduce the observed PV diagrams, both in terms of line shapes and line intensities, for any reasonable CO abundance in the shell.

4.2 Position-velocity diagrams for a phenomenological 2D model

In principle, the absence of a dense anterior part of the shell can be inferred from the presence of optical emission coming from the interior of the H ii region. Thus, a possible way to resolve the inconsistency between the simulated and observed PV diagrams is to assume that RCW 120 is not a spherically symmetric object. The absence of strong 13CO and C18O emission from the inner parts of the region can be explained if RCW 120 is an oblate structure seen face-on. Such a structure may arise naturally as a result of expansion of an H ii region in a flattened molecular cloud. The best proof of this concept would be to present synthetic PV-diagrams based on a self-consistent 2D/3D hydrodynamical model, however, here we illustrate the associated geometrical effects by manually modifying our simpler spherically symmetric model.

In Fig. 8, we present PV diagrams for the same model as in Fig. 7 (with unaltered 13CO and C18O abundances), but with conical parts of the dense layer removed, as shown in the top left panel of Fig. 8. The line of sight is parallel to the polar axis. As expected, the synthetic PV diagrams do not reveal strong molecular emission within a circular area with a radius $200$″. Both C18O(2–1) and 13CO(3–2) profiles are double-peaked, while C18O(2–1) and C18O(3–2) are single-peaked towards the dense shell, as is indeed observed.

While these synthetic PV diagrams better reproduce the observed maps, the observed profiles are generally broader than modelled profiles (see Fig. 3). So we calculated another set of synthetic PV diagrams additionally increasing the micro-turbulent velocity in the input model from $V_{\rm nth}=0.3$ km s*-1* to 1 km s*-1*, as the observed CO lines have widths of several km s*-1* (Sec. 3). The corresponding PV diagrams are shown in the bottom row in Fig. 8, and the spectra extracted from the PV diagrams in Fig. 4. The extracted spectra are shown for the direction of the intensity peak in the dense compressed shell (spectrum i for the offset 200″, at Fig. 8) and in the outer part of the undisturbed molecular gas envelope (spectrum j at the offset 250″). As a result, the width of the synthetic 13CO(2–1) lines gets closer to the observed values, while the synthetic C18O(2–1) profiles become broader than those observed in the direction with $\varphi=285^{\circ}$ in Condensation 2, corresponding to the molecular envelope with no embedded compact sources.

The presented 2D model does not include the extended rarefied molecular cloud discussed in Sec. 3.4. We testified only one model of the cloud in order to see how the simulated line profiles changes. Specifically, the cloud encompassing the H ii region is assumed to have a thickness of 1 pc, gas density of 50 cm*-3*, temperature ($T_{\rm gas}$) of 10 K, and the same CO abundance as in the undisturbed envelope. Such a rarefied cloud produces weak extended emission over the considered maps. The signal-to-noise ratio of the large-scale 13CO(2–1) and C18O(2–1) data presented in Fig. 1 and 3 is too low to detect this emission. The self-absorption in the cloud additionally splits the 13CO(2–1) profile. To study how warm the absorbing extended envelope might be to produce a self-absorption effect on the line profiles, we calculated the excitation temperature $T_{\rm ex}$ of the 13CO(2–1) lines using the RADEX code (van der Tak et al., 2007), with $10\leq T_{\rm gas}\leq 2000$ K (the latter value gives the pressure equilibrium between the extended rarefied cloud and the undisturbed envelope) and $N_{\rm{}^{13}CO}$ from 1014 (corresponding to the front of the rarefied cloud) to 1017 cm*-2*. We found that $T_{\rm ex}$ is always less than the observed value of $T_{\rm mb}=14-16$ K for 13CO(2–1). So, using higher $T_{\rm gas}$ values we are still on the safe side regarding to the self-absorption effect.

Comparing the observed PV diagrams in Fig. 3 with those in Figs 7 and 8, we make the following conclusions. An expanding shell produces double-peaked line profiles with an enhanced red peak due to self-absorption effects, or single-peak profiles skewed to the red, depending on $V_{\rm exp}$ and the micro-turbulent velocity of the molecular gas. A static (non-expanding) optically thick shell would give symmetric double-peaked line profiles. On the one hand, neither in the case of a spherical expanding bubble nor in the case of an expanding torus do we see any noticeable asymmetry in the theoretical line profiles. This means that the expansion velocity in our model is too small to produce line asymmetry, even at the centre of the spherical bubble. This is even more true for the torus model, where the line of sight (presumably) passes through the molecular envelope nearly perpendicular to the expansion velocity. On another hand, the observed double-peaked line profiles are markedly asymmetrical towards Condensation 1, which could mean that even in the relatively simple case of RCW 120 that the actual line profiles are strongly affected by local kinematics. If we neglect these local effects, we can argue that the theoretical PV diagrams resemble the observed ones for the torus model with $V_{\rm nth}=1$ km s*-1*, (compare with the directions with no embedded compact sources in Fig. 3).

Our model is consistent with the absence of front and back walls of the molecular shell, but we still need some molecular gas in the foreground of the H ii region in order to quantitatively reproduce the depth of the self-absorption dip in the double-peaked lines. Moreover, Anderson et al. (2015) detected CO(1–0) and 13CO(1–0) emission at the central part of the H ii region, which could not be reproduced by the face-on torus envelope alone. Given the complex density and velocity structure of the shell on small spatial scales, one may argue that this self-absorption can also be local, arising not in the associated foreground cloud, but in the clumps and condensations themselves. The cloud hypothesis seems to be preferable, as we observe self-absorbed lines both within and beyond dense condensations, and we also see signs of some foreground absorbing layer in the extinction maps.

We note that the goal of these exercises is to show the most promising way to develop consistent hydrodynamical models, rather than to describe all aspects of the observations. Given this, we conclude that such a model should take into account not only the non-spherical geometry of the source, but also the extended cloud and kinematic inhomogeneities.

5 Gas as a proxy of dust and vice versa

So far, our results are in line with the references mentioned in Sec. 1, even though we have not yet constructed a real physical model of the RCW 120 and its surroundings. Anyway, RCW 120 is definitely not a classical spherical expanding H ii region surrounded by dense molecular gas which was collected and compressed by a shock wave preceding the ionization front as was theoretically analysed by Spitzer (1978) and Elmegreen & Lada (1977) (in the context of triggered star formation), see also Hosokawa & Inutsuka (2006); Raga et al. (2012); Bisbas et al. (2015).

Beaumont & Williams (2010) found that the thickness of the dense shells around infrared Spitzer bubbles is $\leq 20$% of their radius. Our torus model corresponds to the thickness of the molecular shell to be 16% of the radius of the H ii region (Fig. 6) in the plane of the sky, but its extent along the line of sight varies from 50% to 90% of the radius, depending on the impact parameter (Fig. 8). While the PV diagrams for RCW 120 indeed suggest that this object is an example of an H ii region in a flattened molecular cloud, the extended surrounding cloud, which we need to introduce into the model and is seen in the dust extinction map, is actually quite extended in the direction perpendicular to the plane of the sky. We also do not attempt to make any conclusions regarding the shape of the torus structure and the diffuse cloud from Fig. 5, because they both have a clumpy structure.

It is interesting to note how the clumpy molecular shell around the H ii region transforms into a more or less continuous ring with rising optical depth of the dust and gas emission. Using equations from Sec. 2.2, we find $\tau_{\rm{}^{13}CO}\approx 2-7$ in the molecular envelope of RCW 120. The corresponding optical depth for the rare isotopologue $\tau_{\rm C^{18}O}$ is $\approx 0.3-1$. The molecular hydrogen column density calculated in Sec. 2.2 agrees within a factor of a few with the calculations made by Anderson et al. (2012) using Herschel data and a canonical dust-to-gas mass ratio of 1:100. Thus, the dust is well mixed with the gas in the molecular envelope of RCW 120. Simulations of Akimkin et al. (2015, 2017) show how charged dust can be expelled from an H ii region by radiation pressure, and estimate that the dust-to-gas mass ratio for the interior of an H ii region like RCW 120 can be less than 50% of the canonical value. However, the drifting dust is stopped inside the collected dense molecular envelope of the H ii region due to high gas density, and dust-to-gas mass ratio there remains near the canonical of 1:100.

We find agreement between the gas column densities found by CO emission and by dust emission even along positions with $\phi{}\approx 0^{\circ}$ and $\phi{}\approx 270^{\circ}$, where Anderson et al. (2012) find discontinuities in the neutral envelope of RCW 120. Anderson et al. (2010) found stellar radiation leaking out the shell through these discontinuities, and dust heating outside the PDR. We note also that because the 13CO and C18O lines have more negative radial velocities along these directions, compared to other regions of the envelope (see Fig. 2), the radiation might leak into the line of sight of the observer in these discontinuities.

On the other hand, the bright 13CO(2–1) integrated intensity does not completely coincide with the bright ATLASGAL 870 $\mu$m emission. In particular, there is bright 13CO(2–1) towards the south and east discontinuities in the neutral envelope of the H ii region. Comparing the gas emission with dust emission in different bands observed by the Spitzer and Herschel telescopes, we conclude that the neutral envelope of the H ii region looks different in optically thin and optically thick regimes. The 13CO(2–1) line emission spatially coincides with the 8 $\mu$m band emission of polycyclic aromatic hydrocarbons (PAHs) around the H ii region, as shown in Fig. 9. These tracers of gas and dust have significant optical depth at the south and east discontinuities ($\tau_{\rm{}^{13}CO}\approx 3-4$, $\tau_{\rm 8\mu{\rm m}}\approx 1-3$; Ya. Pavlyuchenkov et al., in prep.; see also Robitaille et al. (2012)). The distribution of optically thin dust emission in the far infrared (160-1100 $\mu$m) agrees with the C18O(2–1) line emission, which is also optically thin. The optically thin tracers can reveal discontinuities in the neutral envelope, where strong ultraviolet radiation leaks the ionized gas volume. Therefore, gas is a good proxy of dust, and vice versa, if we consider these tracers in the same (either thin or thick) optical regimes.

Analysing a large set of objects from a catalogue of IR ring-like nebulae around H ii regions embedded into molecular clouds (Topchieva et al., 2017), and also large sample of AKARI data of IR bubbles (Hanaoka et al., 2019), we conclude that the images of the H ii regions at 8 $\mu$m appear as quite uniform bubbles due to the significant optical depths, while their shape at longer wavelengths represents a number of clumps around massive stars. Topchieva et al. (2018b) constructed an evolutionary sequences of H ii regions and checked how fluxes at 8, 24, and 160 $\mu$m change during their expansion. While they found that the electron number density decreases along with increasing size of the H ii regions in agreement with theoretical expectations, there was not any particular trend for the IR fluxes. They proposed that an absence of continuous growth in the masses of the neutral envelopes around the H ii regions with time is related to non-uniform gas distribution in the parental molecular clouds, followed by leaks of the ionizing photons through the discontinuities. Recently, Topchieva et al. (2019) showed using a sample of 32 embedded H ii regions, that the morphology of the PDRs around them is likely 3D (spherical) rather than 2D (plane-like). The scenario proposed agrees with 3D simulations of expanding H ii regions in turbulent magnetized media (e.g. by Arthur et al., 2011) or hydrodynamical simulations by Walch et al. (2015), for H ii regions in fractal molecular clouds. Recent analysis of the Herschel images of RCW 120 by Marsh & Whitworth (2019) demonstrates significant differences in the spatial distributions of warm and cold dust. The former is localized near the ionizing star in a confined volume, but cold dust is distributed in a clumpy manner in a broad area around the H ii region.

We conclude the paper with an idea that due to different optical depth of dust emission the bubbling galactic disk found by Spitzer becomes a disk full of clumps organized around young massive stars on the Herschel and APEX ATLASGAL images.

6 Conclusions

• •

In this work, we presented observations of RCW 120 in lines of CO isotopologues, and analysed these observations (along with archival data) to put constraints on the geometry of RCW 120. Using results of numerical modelling and radiative transfer simulations, we show that the integrated intensities of 13CO(2–1) and C18O(2–1) lines in the H ii region and its environment cannot be reproduced within the framework of a spherical shell model. However, the observational data are consistent with a model of a toroidal dense molecular shell around RCW 120 seen face-on.

• •

The 13CO(2–1) and 13CO(3–2) lines show deep self-absorption, which cannot be explained even by the torus model, and require additional absorption in foreground cold gas. Extinction maps at 2 $\mu$m, and also infrared images at longer wavelengths, show that RCW 120 is a part of a large-scale absorbing cloud 11.3 pc in size in the plane of the sky. An extended foreground cloud with a gas number density of about 50 cm*-3* can fit the self-absorption profiles of the 13CO lines.

• •

There is a spatial coincidence of optically thick gas and dust tracers, such as 13CO(2–1) line emission and 8 $\mu$m emission of PAHs. The neutral envelope of RCW 120 looks almost continuous in these tracers. The same is true for a pair of optically thin dust emission in the far infrared (160–1100 $\mu$m) and with C18O(2–1) line emission. The optically thin emission demonstrates discontinuities in the neutral envelope, where strong ultraviolet radiation leaks the ionized gas volume.

Acknowledgements

We are thankful to D. A. Kovaleva, V. V. Akimkin for fruitful discussions of RCW 120 and also to staff of Onsala Space Observatory for their care for technical details of the O-083.F-9311A-2009 project.

M. S. Kirsanova and Ya. N. Pavlyuchenkov were partly funded by the Russian Foundation for Basic Research, research project 18-32-20049.

S. V. Salii and A. M. Sobolev work was supported in part by the Ministry of Education and Science (the basic part of the State assignment, RK no. AAAA-A17-117030310283-7) and by the Act no. 211 of the Government of the Russian Federation, agreement no. 02.A03.21.0006.

This research is based on observations with the Atacama Pathfinder EXperiment (APEX) telescope. APEX is a collaboration between the Max Planck Institute for Radio Astronomy, the European Southern Observatory, and the Onsala Space Observatory.

The ATLASGAL project is a collaboration between the Max-Planck-Gesellschaft, the European Southern Observatory (ESO) and the Universidad de Chile. It includes projects E-181.C-0885, E-078.F-9040(A), M-079.C-9501(A), M-081.C-9501(A) plus Chilean data.

This research has made use of NASA’s Astrophysics Data System Bibliographic Services; SIMBAD database, operated at CDS, Strasbourg, France (Wenger et al., 2000); Aladin web page (Bonnarel et al., 2000); Astropy, a community-developed core Python package for Astronomy (Astropy Collaboration et al., 2013); APLpy, an open-source plotting package for Python ((http://aplpy.github.com) Robitaille & Bressert, 2012).