CO observations toward the isolated mid-infrared bubble S44: External triggering of O-star formation by a cloud–cloud collision

Abstract

1 Introduction

1.1 H ii regions and Spitzer mid-infrared bubbles in the Milky Way

H ii regions are formed around the high-mass stars, which are mainly distributed in the spiral arms of the Milky Way. They ionize and destroy the parent molecular clouds and create cavity-like structures around the exciting stars via stellar winds and ultraviolet radiation (e.g., Whitworth 1979). These cavity-like structures in the interstellar medium (ISM) are often called “interstellar bubbles” (e.g., Castor et al. 1975; Weaver et al. 1977).

The Spitzer mid-infrared bubbles were identified by Churchwell et al. (2006, 2007) from the Galactic Legacy Infrared Mid-Plane Survey Extraordinaire (GLIMPSE) data. They cataloged about 600 bubbles in the northern and southern hemispheres in the Galactic plane (|l| ≤ 65°, |b| ≤ 1°). The authors suggested that most bubbles are H ii regions that contain embedded OB-type stars or star clusters.

Several formation scenarios have been discussed for the mid-infrared bubbles, including radiation-driven implosions (RDIs: Sandford et al. 1982; Lefloch & Lazareff 1994) and the collect-and-collapse process (C & C: Elmegreen & Lada 1977; Whitworth et al. 1994). The first process involves the compression of pre-existing clouds by the pressure of the ionized gas, while the second consists of the sweeping-up of the diffuse ISM inside the wall of the expanding shell that is undergoing gravitational collapse. These scenarios can explain the star formation at the edge of bubbles, which is triggered by the expanding H ii regions (e.g., Deharveng et al. 2005, 2008, 2009, 2010; Zavagno et al. 2006, 2007). On the other hand, from their numerical simulations, Dale, Haworth, and Bressert (2015) have pointed out that it is difficult to distinguish between triggered and spontaneous star formation around H ii regions solely from their observational morphologies.

Recently, Torii et al. (2015) carried out CO observations toward the mid-infrared bubble RCW 120. They showed that a cloud–cloud collision scenario can explain the morphologies of the bubbles and the formation of ionizing O stars, based on numerical simulations of a head-on collision between different-sized clouds (Habe & Ohta 1992; Anathpindika 2010). In this scenario, high-mass stars are formed in the compressed layer created by the collision of two clouds. These stars then ionize the parent clouds, leading to the formation of bubble-like H ii regions (see, figure 12 of Torii et al. 2015). This scenario has also been suggested as the formation mechanism for many mid-infrared bubbles (e.g., N 37, Baug et al. 2016; Sh2-48, Torii et al. 2021; RCW 79, Ohama et al. 2018a; RCW 166, Ohama et al. 2018b; S116-117-118, Fukui et al. 2018; N 35, Torii et al. 2018; N 49, Dewangan et al. 2017; N 4, Fujita et al. 2019).

Thus many different star-formation scenarios may be related to the bubbles, but the dominant process is not clear.

1.2 S44 as an isolated mid-infrared bubble

[CPA2006] S44 (hereafter S44) is an isolated mid-infrared bubble located at |$(l,b)= (334{^{\circ}_{.}}523,\, 0{^{\circ}_{.}}823)$|⁠. It was cataloged by Churchwell et al. (2006, see their figure 2c), and corresponds to the closed bubble identified from the AKARI 9 μm emission (Hanaoka et al. 2018). Simpson et al. (2012) also identified S44 as MWP1 G334525+008255 by visual inspection as a part of the “Milky Way project” (see also Kendrew et al. 2012). Caswell and Haynes (1987) carried out a radio-recombination-line survey toward southern H ii regions, deriving the radio recombination line velocity of −77 km s⁻¹ of G334.529+0.825. From table 4 of Churchwell et al. (2006), the distance to S44 is estimated to be either ∼4.6 ± 0.5 kpc (the nearer estimate) or ∼10.8 ± 0.5 kpc (the more distant estimate), based on this velocity and the kinematic-distance model of Brand and Blitz (1993). Churchwell et al. (2006) suggested that the nearer kinematic distance is the more correct choice toward bubbles, because it is more likely to be detected in continuum emission than at the greater distance in the Galactic plane. Following previous studies, in this paper we choose 4.6 kpc as the distance to S44, which suggests that it is located in the Norma spiral arm in the Milky Way (Brown et al. 2014). Hattori et al. (2016) estimated the total infrared luminosity of S44 to be 1.86 × 10⁵(10^5.27) L_⊙ by fitting the spectral energy distribution (SED) with polycyclic aromatic hydrocarbons (PAHs), warm-dust, and cold-dust components.

Figure 1 shows a three-color composite image of the Spitzer space telescope observations (GLIMPSE: Benjamin et al. 2003, Churchwell et al. 2009; MIPSGAL: Carey et al. 2009), where blue, green, and red correspond to the 3.6, 8, and 24 μm emissions, respectively. The 3.6 μm emission mainly traces thermal emission from the stars, while the 8 μm emission traces the PAH features in the photo-dissociation region (e.g., Draine 2003; Draine & Li 2007; Churchwell et al. 2004). The 8 μm emission has a ring-like structure, with bright emissions at the southern edge of S44, and the diameter of the bubble is about 5 pc in the extended 8 μm emission. The 24 μm emission traces hot dust grains heated by high-mass stars in the H ii region (e.g., Carey et al. 2009), and it has an arc-like distribution. A bright infrared source exists at |$(l,b)=({334{^{\circ}_{.}}46}, {0{^{\circ}_{.}}88})$| on the western side of S44, which corresponds to the OH/IR star (OH 334.458+0.877) identified from the OH maser survey at 1612.231 MHz using the Australia Telescope Compact Array (ATCA: Sevenster et al. 1997). The relationship of this infrared source to S44 is not clear. We also find pillar-like structures in the 8 μm emissions at the edge of the bubble that are elongated in the direction of the center, with lengths of ∼1 pc (the pink dotted arrows in figure 1b).

Fig. 1.

(a) Large-scale three-color composite images of S44. Blue, green, and red show the Spitzer/IRAC 3.6 μm, Spitzer/IRAC 8 μm, and Spitzer/MIPS 24 μm results. The jagged white line along the Galactic latitude at |$b\sim {0{^{\circ}_{.}}95}$| shows the observing limit of Spitzer/MIPS 24 μm. (b) Close-up image of panel (a). The colors are the same as in (a). The pink dotted arrows indicate the pillar-like structures. (Color online)

Open in new tabDownload slide

S44 is an isolated mid-infrared bubble, for which the three-dimensional spatial distribution, velocity structures, and physical properties of the associated molecular gas have not yet been determined. In this paper, we carried out new CO observations in the direction of S44 using the NANTEN2, Mopra, and ASTE (Atacama Sub-millimeter Telescope Experiment) radio telescopes. This paper is organized as follows: section 2 describes the observational information; section 3 presents the observed cloud properties and comparisons with observations at other wavelengths; in section 4 we discuss the formation mechanism of S44 based on a star-formation scenario; in section 5 we conclude this paper.

2 Observations

2.1 NANTEN2 ¹²CO J = 1–0 observations

We carried out ¹²CO J = 1–0 (115.271 GHz) observations with the NANTEN2 4 m millimeter/sub-millimeter radio telescope located in Chile and operated by Nagoya University. The observations were made using the on-the-fly (OTF) mode from 2012 May to December as a part of a Galactic plane survey. The half-power beam width (HPBW) is 2|${^{\prime}_{.}}$|7 at 115 GHz. This corresponds to 3.6 pc at the distance of 4.6 kpc. The front end was a cooled 4 K superconductor-insulator-superconductor (SIS) mixer receiver. The system temperature including the atmosphere was ∼250 K in the double-side band (DSB) toward the zenith. The backend was a digital-Fourier-transform spectrometer (DFS) with 16384 channels, each with a 1 GHz bandwidth. The velocity coverage and resolution at 115 GHz were ∼2600 km s⁻¹ and 0.16 km s⁻¹, respectively. We confirmed that the pointing accuracy was better than |${15^{\prime \prime }}$| from daily observations toward IRC +10216 and the Sun. We used the chopper-wheel method to calibrate the antenna temperature |$T_{\rm a}^*$| (Penzias & Burrus 1973; Ulich & Haas 1976; Kutner & Ulich 1981). We calibrated the absolute intensity fluctuation using daily observations of IRAS 16293 −2422 [|$\alpha _{\rm J2000.0} = {16^{\rm h}32^{\rm m}23{^{\rm s}_{.}}3}$|⁠, |$\delta _{\rm J2000.0} = {-24^{\circ}28^{\prime }39^{\prime \prime }\!.2}$|], and we converted the intensity scale into the main-beam temperature T_mb by assuming its peak to be T_mb = 18 K (Ridge et al. 2006). The typical intensity uncertainty of the data sets is ∼20% from the NANTEN2 errors and reference data. We smoothed the data cube with a Gaussian kernel of |${90^{\prime \prime }}$|⁠, and the final beam resolution was |${180^{\prime \prime }}$| (FWHM). The typical root-mean-square (rms) noise level was ∼1.2 K after smoothing and velocity-channel binning down to 0.43 km s⁻¹.

2.2 Mopra ¹²CO and ¹³CO J = 1–0 observations

In 2014 July, we observed a |${12^{\prime }}\times {12^{\prime }}$| area toward S44 using the ¹²CO and ¹³CO J = 1–0 transitions (115.271 and 110.201 GHz) using the Mopra 22 m telescope in the OTF mapping mode at the Australia Telescope National Facility (ATNF) located in the Warrumbungle Mountains. The HPBW was ∼33″ ± 2″ at 115 GHz, as measured from planetary observations (Ladd et al. 2005). This corresponds to 0.7 pc at the distance of 4.6 kpc. The front end was a high-electron-mobility-transistor (HEMT) receiver covering the 3 mm band. The typical system noise temperature T_sys was ∼600 K in the single-side band (SSB). The back-end system of the Mopra Spectrometer (MOPS) has 4096 channels across 137.5 MHz in each of the two polarizations. The velocity resolution was 0.088 km s⁻¹, and the velocity coverage was 360 km s⁻¹ at 115 GHz. That the pointing accuracy would be within |${5^{\prime \prime }}$| was checked by observing the SiO 86 GHz (3 mm) maser sources in AH Sco [|$\alpha _{\rm J2000.0} ={17^{\rm h}11^{\rm m}17{^{\rm s}_{.}}16},\ \delta _{\rm J2000.0} = {-32^{\circ}19^{\prime }30^{\prime \prime }\!.72}$|]. We checked the intensity variation by observing M 17 SW [|$\alpha _{\rm J2000.0} ={18^{\rm h}20^{\rm m}23{^{\rm s}_{.}}1},\ \delta _{\rm J2000.0} = {-16^{\circ}11^{\prime }37^{\prime \prime }\!.2}$|] and the peak of RCW 79 [|$(l,b) = ({308{^{\circ}_{.}}760},\ {0{^{\circ}_{.}}546})$|]. Typical data fluctuations in the integrated intensity were about ∼20%. We smoothed the data cube with a Gaussian kernel of |${30^{\prime \prime }}$|⁠, and the final resolution was |${45^{\prime \prime }}$| (FWHM). We converted the intensity from antenna temperature (⁠|$T_{\rm a}^*$|⁠) to the main-beam temperatures (T_mb) by applying the relation |$T_{\rm mb}=T_{\rm a}^*/ \eta _{\rm mb}$|⁠, assuming a main-beam efficiency (η_mb) of 0.42 for 115 GHz (Ladd et al. 2005). The rms noise level was ∼0.76 K for ¹²CO J = 1–0 and ∼0.53 K for ¹³CO J = 1–0, with a velocity resolution of 0.43 km s⁻¹.

2.3 ASTE ¹²CO J = 3–2 observations

In 2014 June, we performed ¹²CO J = 3–2 (345.796 GHz) observations toward an area of |${10^{\prime }}\times {10^{\prime }}$| around S44 using the ASTE (Ezawa et al. 2004, 2008; Kohno et al. 2004) located in Chile. The HPBW was |$\sim {22^{\prime \prime }}$| at 345 GHz. This corresponds to 0.5 pc at a distance of 4.6 kpc. The front end was the two-side band (2SB) SIS mixer called “CATS 345” (cartridge-type-sideband-separating receiver at 345 GHz, Inoue et al. 2008). The typical system temperature was 300 K at 345 GHz in the SSB mode. The back-end was the XF-type digital spectro-correlator, “MAC,” with 1024 channels in each of 128 MHz width. The velocity resolution was 0.11 km s⁻¹, and the velocity coverage was 111 km s⁻¹ at 345 GHz (Sorai et al. 2000). We checked that the pointing accuracy would be within |${2^{\prime \prime }}$| by observing RAFGL 4202 [|$\alpha _{\rm J2000.0} ={14^{\rm h}52^{\rm m}23{^{\rm s}_{.}}82},\ \delta _{\rm J2000.0} = {-62^{\circ}04^{\prime }19^{\prime \prime }\!.2}$|]. We checked the intensity calibration by observing W 44 [|$\alpha _{\rm J2000.0} ={18^{\rm h}50^{\rm m}46{^{\rm s}_{.}}1}$|⁠, |$\delta _{\rm J2000.0} = {1^{\circ}11^{\prime }11^{\prime \prime }\!.0}$|]. Typical data fluctuations in the peak intensity were about ∼ 10%. We convolved the intensity scale into T_mb by assuming the W 44 peak to be T_mb = 35.5 K (Wang et al. 1994). The rms noise level was ∼0.3 K for ¹²CO J = 3–2, with a velocity resolution of 0.43 km s⁻¹.

2.4 Archived data sets

We used the following archived data sets to compare with the CO data, i.e., to the near- and mid-infrared data from the Spitzer space telescope (GLIMPSE at 3.6 μm and 8.0 μm: Benjamin et al. 2003, Churchwell et al. 2009; MIPSGAL at 24 μm: Carey et al. 2009). We obtained the 870 μm radio-continuum-emission data with the Atacama Pathfinder Experiment (APEX) Telescope Large Area Survey of the GALaxy (ATLASGAL: Schuller et al. 2009), and the 843 MHz (36 cm) radio continuum emission data from the Sydney University Molonglo Sky Survey (SUMSS: Bock et al. 1999) observed with the Molonglo Observatory Synthesis Telescope (MOST). We summarize the observational properties and archival information in table 1.

3 Results

3.1 CO distributions and velocity structures toward S44

Figure 2a shows a large-scale ¹²CO J = 1–0 integrated intensity map obtained by NANTEN2. The CO cloud is distributed over about 50 pc from the northern side to southern side of the bubble with a peak at |$(l,b)=({334{^{\circ}_{.}}52}, {0{^{\circ}_{.}}77})$|⁠. Figure 2b presents a detailed CO distribution obtained with Mopra. The cloud distribution delineates the 8 μm emission at the southern side of the bubble, and the intensity is depressed inside the bubble. In order to investigate the detailed morphologies of the parent clouds, we focused on the molecular gas around the bubble using the high-spatial-resolution data sets obtained by Mopra. Figures 3 and 4 show velocity-channel maps of ¹²CO J = 1–0 and ¹³CO J = 1–0, respectively. The cloud distribution outlines the bubble shape of the 8 μm emission, and the southern part of the bubble (figures 3d, 3e, 4d, and 4e) is more intense than the northern part (figures 3b, 3c, 4b, and 4c). The ¹³CO emission coincides with the most intense parts of the ¹²CO J = 1–0 emission. We obtain the velocity difference between the northern and southern parts of the bubble from the velocity range of −86.4 to −77.8 km s⁻¹ (figures 3b–3e and 4b–4e).

Fig. 2.

(a) Integrated map of the ¹²CO J = 1–0 emission in the velocity range of −88.5 to −69.2 km s⁻¹. The contours show the Spitzer/IRAC 8 μm result, where the region used for the 8 μm emission is indicated by the black box. The lowest contour and contour intervals are 70 MJy beam⁻¹ and 80 MJy beam⁻¹, respectively. The final beam size after convolution is |${180^{\prime \prime }}$|⁠. (b) Integrated intensity map of the ¹²CO J = 1–0 emission with Mopra. The final beam size after convolution is 45″. (Color online)

Open in new tabDownload slide

Fig. 3.

Velocity-channel map of the ¹²CO J = 1–0 emission, with a velocity step of 2.15 km s⁻¹, obtained with Mopra. The contours show the 8 μm emission from Spitzer/IRAC. The final beam size after convolution is |${45^{\prime \prime }}$|⁠. The 1 σ noise level is ∼0.7 K km s⁻¹ for the velocity interval of 2.15 km s⁻¹. (Color online)

Open in new tabDownload slide

Fig. 4.

Velocity-channel map of the ¹³CO J = 1–0 emission, with a velocity step of 2.15 km s⁻¹, obtained with Mopra. The contours show the 8 μm emission from Spitzer/IRAC. The final beam size after convolution is |${45^{\prime \prime }}$|⁠. The 1 σ noise level is ∼0.5 K km s⁻¹ for the velocity interval of 2.15 km s⁻¹. (Color online)

Open in new tabDownload slide

Figure 5 shows the first-moment map of ¹³CO J = 1–0 and several spectra. The ¹³CO map is useful for investigating velocity gradients because it delineates the denser regions of the parent clouds. From the first-moment map (figure 5a) and the spectra (figures 5b, 5c, and 5d), we can see that two velocity components exist around the bubble. Focusing on the eastern side of the bubble, we can clearly identify both velocity components in the same region (figure 5c). We therefore suggest that these velocity differences do not represent a velocity gradient of a single cloud but, instead, are two independent components around the bubble. Because of their velocities, we hereafter designate these two clouds as the “−84 km s⁻¹ cloud” and the “−79 km s⁻¹ cloud.”

Fig. 5.

(a) The first-moment map of the ¹³CO J = 1–0 emission, which we created for the velocity range of −87.64 to −77.33 km s⁻¹using the volume voxels with intensities greater than 2.1 K (4 σ). The lowest contour and contour intervals were 70 MJy beam⁻¹ and 80 MJy beam⁻¹ for the Spitzer/IRAC 8 μm result. The boxes show the averaging areas for each profile. (b), (c), and (d) Averaged spectra for ¹²CO, ¹³CO J = 1–0, and ¹²CO J = 3–2. The dotted lines indicate the two velocity components at −84 km s⁻¹ and −79 km s⁻¹. The size of the averaging box is |${35^{\prime \prime }} \times {35^{\prime \prime }}$|⁠. (Color online)

Open in new tabDownload slide

Figure 6a shows the two ¹³CO J = 1–0 clouds superposed on the Spitzer 8 μm image. The −84 km s⁻¹ and −79 km s⁻¹ clouds overlap on the eastern side of the bubble. Figures 6b and 6c present position–velocity diagrams for which the integration ranges focus on the overlapping parts of the two clouds. The velocity of the radio recombination line (−77 km s⁻¹ from Caswell & Haynes 1987) is also shown in the position–velocity diagram (figures 6b and 6c). We note the cavity-like structure around the bubble, and we find that the two clouds are connected to each other by a bridging feature at intermediate velocities. The cavity-like structures (a few km s⁻¹) around the radio-recombination-line velocity may be caused by ionization from the exciting star(s), and the bridging feature indicates that these two clouds may be interacting with each other around the bubble.

Fig. 6.

(a) Integrated intensity map of ¹³CO J = 1–0 obtained by Mopra for the −84 km s⁻¹ cloud (blue contours) and the −79 km s⁻¹ cloud (red contours) superposed on the Spitzer 8 μm emission. The dashed yellow lines show the integration ranges in latitude and longitude. (b) Galactic latitude–velocity diagram integrated over the longitude range of from |${334{^{\circ}_{.}}54}$| to |${334{^{\circ}_{.}}57}$|⁠. The 1σ noise level is ∼0.004 K degree for the longitude interval of |${0{^{\circ}_{.}}03}$| . (c) Galactic longitude–velocity diagram integrated over the latitude range of from |${0{^{\circ}_{.}}81}$| to |${0{^{\circ}_{.}}84}$|⁠. The dashed lines represent the radio-recombination-line velocity (−77 km s⁻¹) from Caswell and Haynes (1987). The spatial and velocity resolutions are smoothed to |${52^{\prime \prime }}$| and 0.18 km s⁻¹, respectively. The 1σ noise level is ∼0.004 K degree for the latitude interval of |${0{^{\circ}_{.}}03}$|⁠. (Color online)

Open in new tabDownload slide

3.2 Physical properties of the two clouds

3.3 ¹²CO J = 3–2 distributions and ¹²CO J = 3–2/1–0 intensity ratios

Figure 7 shows the velocity-channel map for ¹²CO J = 3–2 obtained with ASTE. The ¹²CO J = 3–2 distribution more clearly shows the ring features associated with the 8 μm emission from the bubble. We also find two clumps, at |$(l,b)=({334{^{\circ}_{.}}495},\ {0{^{\circ}_{.}}835})$| and |$(l,b)=({334{^{\circ}_{.}}505},\ {0{^{\circ}_{.}}845})$|⁠, on the western side of the bubble in the velocity range of −84.2 to −82.1 km s⁻¹ (figure 7c), and a clumpy structure at |$(l,b)=({334{^{\circ}_{.}}53},\ {0{^{\circ}_{.}}80})$| in figure 7d. Figure 8 shows the ¹²CO J = 3–2/¹²CO J = 1–0 (R_3−2/1−0) intensity ratio maps for the −84 km s⁻¹ cloud in panel (a) and for the −79 km s⁻¹ cloud in panel (b). We convolved the ¹²CO J = 3–2 map with the Gaussian kernel of |${45^{\prime \prime }}$|⁠, which is the final beam size of the ¹²CO J = 1–0 Mopra data. The 5σ (∼1.5 K km s⁻¹) clipping level we adopted is shown by the white dotted contour of ¹²CO J = 3–2. The intensity ratio between the different rotational transition levels of the CO lines provides the excitation condition in the CO gas, and a high intensity ratio is a good indicator of a physical association of the CO gas with the H ii region. The −84 km s⁻¹ cloud has high ratios (R_3−2/1−0∼1.0–1.2) on the western edge |$(l,b)=({334{^{\circ}_{.}}490}, {0{^{\circ}_{.}}830})$| and the northeastern edge |$(l,b)=({334{^{\circ}_{.}}555}, {0{^{\circ}_{.}}825})$| of the bubble (figure 8a). The −79 km s⁻¹ cloud also has enhanced R_3−2/1−0 ∼ 1.0–1.2, but around the southern and western edges of the bubble.

Fig. 7.

Velocity-channel map of the ¹²CO J = 3–2 emission, with a velocity step of 2.15 km s⁻¹, obtained with ASTE. The contours show the 8 μm emission from Spitzer/IRAC. The final beam size is |${22^{\prime \prime }}$|⁠. The 1σ noise level is ∼0.3 K km s⁻¹ for the velocity interval of 2.15 km s⁻¹. (Color online)

Open in new tabDownload slide

Fig. 8.

Intensity ratio map of ¹²CO J = 3–2/¹²CO J = 1–0 from ASTE and Mopra for the −84 km s⁻¹ cloud (a) and the −79 km s⁻¹ cloud (b). The final beam size after convolution is |$\sim {45^{\prime \prime }}$|⁠. The 5σ (∼1.5 K km s⁻¹) clipping levels adopted are shown by the dotted white contour of ¹²CO J = 3–2. The lowest yellow contour and intervals are 70 MJy beam⁻¹ and 80 MJy beam⁻¹ for the Spitzer/IRAC 8 μm result. (Color online)

Open in new tabDownload slide

The distributions of the intensity ratio R_3−2/1−0 delineate the 8 μm ring structure, showing a steep increase of temperature inside the bubble. These results indicate that the two clouds are likely to be physically associated with the bubble.

3.4 Comparing with the ionized gas and cold dust emissions

Figures 9a and 9b show comparisons of the −84 km s⁻¹ and −79 km s⁻¹ clouds, respectively, with SUMSS 843 MHz (36 cm) continuum images. The continuum image traces the free–free emission from the ionized gas heated by the high-mass stars. The 843 MHz intensity is enhanced at the southern side of the bubble, having an arc-like structure. We note that the ionized gas is not spread uniformly inside the 8 μm shell structure, which is different from other bubbles (e.g., N 10 and N 21, Watson et al. 2008). The −84 km s⁻¹ cloud is distributed along the northern and western edge of the bubble (figure 9a). The −79 km s⁻¹ cloud surrounds the ionized gas at the southern side of the bubble (figure 9b). Figures 9c and 9d show comparisons of the −84 km s⁻¹ and −79 km s⁻¹ clouds, respectively, and the 870 μm continuum images obtained with APEX. The 870 μm continuum image shows the distribution of the thermal emission from the cold dust (Schuller et al. 2009). The distribution of the cold dust outlines the shape of the bubble, with some peaks coinciding with the peaks of the −84 km s⁻¹ and −79 km s⁻¹ clouds. The −84 km s⁻¹ cloud is distributed along the edge of the cold-dust emission at |$(l,b)\sim ({334{^{\circ}_{.}}555},\ {0{^{\circ}_{.}}825})$| (figure 9c). The peak in the 870 μm emission at |$(l,b)=({334{^{\circ}_{.}}52},\ {0{^{\circ}_{.}}77})$| on the southern side of the bubble corresponds to the compact source AGAL G334.521+00.769 cataloged by the ATLASGAL survey (Contreras et al. 2013; Urquhart et al. 2014). Figure 9d shows that AGAL G334.521+00.769 is embedded in one of the peaks of the −79 km s⁻¹ cloud. We note that the peaks of the radio continuum, CO, and 870 μm emissions have ordered distributions moving to lower galactic latitudes.

Fig. 9.

(a), (b) Integrated intensity map of ¹²CO J = 3–2 (contours) obtained with ASTE superposed on the MOST 843 MHz continuum image. (c), (d) Integrated intensity map of ¹²CO J = 3–2 (contours) obtained by ASTE superposed on the APEX 870 μm continuum image. The blue and red contours represent the −84 km s⁻¹ cloud and the −79 km s⁻¹ cloud, respectively. (Color online)

Open in new tabDownload slide

3.5 Estimation of the spectral type of the exciting star(s)

3.6 Color–color diagram of the 24 μm sources

We constructed a color–color diagram to distinguish between young stellar objects (YSOs) and other objects toward the 24 μm sources cataloged around this bubble by Gutermuth and Heyer (2015). They obtained a radius of 7 pc (⁠|$= {5.25^{\prime }}$|⁠) around the geometric center position at |$(l,b)=({334{^{\circ}_{.}}523},\ {0{^{\circ}_{.}}823})$|⁠. Figure 10 shows the result of the [3.6] − [5.8] versus [8.0] − [24] diagram toward the 24 μm sources. We adopted the YSO criteria from Muzerolle et al. (2004), who carried out this classification toward the embedded star-forming region NGC 7129 in the Milky Way. We identified the #1 source in S44 as a Class II YSO at the southern edge of the bubble (figures 11c and 11d). Taking account of its error bars, the #2 source located outside the bubble is also possibly a Class II YSO. The photometric parameters of these sources (#1 and #2) are summarized in table 3. Many of the other 24 μm sources are categorized as Class III/stellar objects. We will not argue these Class III/stellar objects in this paper, because it is not clear whether they are physically related to the bubble or not.

Fig. 10.

Color–color diagram [3.6] − [5.8] versus [8.0] − [24] for the 24 μm sources cataloged by Gutermuth and Heyer (2015) around the bubble. The dotted boxes show the classification of YSOs from Muzerolle et al. (2004). (Color online)

Open in new tabDownload slide

Fig. 11.

(a) APEX 870 μm image (contours) superposed on the H₂ column density image from ¹³CO J = 1–0 with the velocity range of −88.5 to −69.2 km s⁻¹. (b) APEX 870 μm image (contours) superposed on the Spitzer 24 μm emission. The red dotted square shows the sources detected in the 24 μm image embedded in the cold dust condensation. Close-up (c) 8 μm and (d) 24 μm images at the southern edge of the bubble. The YSO is indicated by a red arrow in panels (c) and (d). (Color online)

Open in new tabDownload slide

4 Discussion

4.1 Star formation around the bubble and the origin of the isolated O star(s)

From previous studies of bubbles, the C&C and/or RDI processes have been provided for the mechanisms of star formation at the edge of a bubble created by an expanding H ii region (e.g., Deharveng et al. 2010; Zavagno et al. 2006, 2007). In the case of S44, we find cold-dust condensation at 870 μm at the southern edge of the bubble (AGAL G334.521+00.769). Figure 11a shows the H₂ column density derived from ¹³CO J = 1–0, together with the contours of cold-dust emission from 870 μm. There is clearly a good spatial correspondence between the high molecular column densities and the peaks of cold-dust emission, which suggests that star formation around the bubble is likely to be happenig at the southern edge, while such cold- and dense-dust condensations are not detected at the northern side of the bubble. These observational results are common to other bubbles (e.g., RCW 120: Deharveng et al. 2009; Figueira et al. 2017; RCW 79: Zavagno et al. 2006, Liu et al. 2017).

Our observations show two velocity components associated with the bubble at the northern and southern sides of the 8 μm emission (figure 6a), together with a bridging feature connecting the two clouds (figures 6b and 6c). These signatures suggest that the two clouds are interacting with each other in the bridging feature, and they are similar to the properties of other bubbles that are formed by cloud–cloud collisions (RCW 120, Torii et al. 2015; RCW 79, Ohama et al. 2018a; N 4, Fujita et al. 2019). Numerical simulations of a cloud–cloud collision reproduce the broad-line bridging feature at the interface between the two clouds in the position–velocity diagram (Haworth et al. 2015a, 2015b; see also the review by Haworth et al. 2018) based on the model from Takahira, Tasker, and Habe (2014), Takahira et al. (2018), and Shima et al. (2018). From synthetic CO observations, they showed that the bridging feature is caused by turbulent motions in the compressed layer between the two colliding clouds.

Stellar feedback from the exciting star may be an alternative explanation for the bridging feature. If expanding motions dominate the kinematics of the bubble, we expect to observe a ring-like velocity distribution in the position–velocity diagram. However, in the position–velocity diagram toward the center of the bubble, we do not find expanding velocity structures from the CO data corresponding to the sound speed (∼10 km s⁻¹) of the ionized gas (figure 6c). This suggests that any acceleration caused by stellar feedback is limited, which is consistent with the case of RCW 120 (Anderson et al. 2015; Torii et al. 2015). We note that S44 is an isolated bubble, because we do not find extended infrared emission (figure 1a), even though the molecular clouds are distributed up to 50 pc beyond the northern and southern sides of the bubble (figure 2a). Hence, isolated O star(s) are unlikely to be formed by stellar feedback from other high-mass stars. The formation of massive, dense cores from an O star or a star cluster require the external shock compression (e.g., Zinncker & Yorke 2007). Some numerical magnetohydrodynamical simulations show that a cloud–cloud collision process satisfies the initial conditions for O-star formation (e.g., Inoue & Fukui 2013; Wu et al. 2015, 2017a, 2017b). We therefore hypothesize that the two clouds collided with each other and that the collision triggered the formation of an isolated, massive exciting star.

4.2 A cloud–cloud collision scenario

Based on our observational results, in this section we propose a scenario in which star formation is triggered by a cloud–cloud collision. From the similar mid-infrared bubble RCW 120 (Torii et al. 2015), and based on the numerical simulation of Habe and Ohta (1992), our proposing scenario is in the following (see our schematic picture in figure 12):

• The −84 km s⁻¹ diffuse cloud, enclosing a dense core, and the −79 km s⁻¹ cloud approach each other (figure 12, stage I).

• The two clouds collide with each other, creating a compressed layer in the dense part of the −84 km s⁻¹ cloud at the interface between the two clouds and forming a cavity in the −79 km s⁻¹ cloud. The two clouds mix at the boundary and form the intermediate velocity component (figure 12, stage II).

• A high-mass star are formed in the compressed layer at the interface between the two colliding clouds. The parent cloud and the surrounding interstellar medium are then ionized, leading to the formation of a bubble-like structure (figure 12, stage III).

Figure 11b shows the hot-dust distribution at 24 μm, together with contours of the cold-dust emission at 870 μm. The 24 μm hot-dust emission has an asymmetric distribution at the southern side of the bubble, that is more intense than that at the northern side. We also find a class II YSO at 24 μm (red arrows in figures 11c and 11d) that is embedded in the cold-dust condensation producing the 870 μm emission. This is similar to the distribution of YSOs emitting at 24 μm embedded in “condensation 1” at the edge of RCW 120 (Deharveng et al. 2009, their figure 10). We note that it is not clear whether a cloud–cloud collision caused the formation of this class II YSO or not, because it may be a pre-existing star.

Torii et al. (2015) and Ohama et al. (2018a) showed that remnants of the colliding clouds exist outside the openings of the bubbles in RCW 120 and RCW 79. In the case of S44, the opening (broken) bubble seen in 8 μm emission is not clearly comparable to RCW 120 or RCW 79. We suggest that this difference between the closed and broken 8 μm emission bubbles can be explained by the projection effect toward the bubble. Based on this assumption, we choose 45° as a projection angle toward S44 in subsection 4.3 below.

Fig. 12.

Schematic image of a cloud–cloud collision scenario based on Habe and Ohta (1992) and Torii et al. (2015, 2017). Stage I: 3D image of the initial condition of the two clouds. Stage II: 2D image of the two clouds in the x–z plane at the time when the two clouds collide with each other. Stage III: 2D image of the bubble in the x–z plane. The y-axis corresponds to the line of sight. The final panel shows the observational result for the two clouds superposed on the 24 μm image. (Color online)

Open in new tabDownload slide

4.3 The timescale for star formation

If we assume an inclination angle of 45°, the collisional timescale is about |$20\:\mbox{pc}/ (5 \times \sqrt{2})\:$|km s⁻¹ ∼ 3 Myr from the extended cloud size and velocity difference. On the other hand, if we assume the mass-accretion rate of high-mass stars to be 2 × 10⁻⁴ M_⊙ yr⁻¹ from the numerical simulation of Inoue et al. (2018), the timescale for high-mass star formation in S44 is (18–19)M_⊙/(2 × 10⁻⁴ M_⊙ yr⁻¹) ∼ 0.1 Myr. We thus suggest that the event of a cloud–cloud collision happens on a long time scale (∼ a few Myr), because the small cloud is decelerated by conserving the momentum through the collision, whereas O-star formation has a short timescale of ∼0.1 Myr. This is similar to the case of the super star cluster NGC 3603, except for the number of O stars and the high H₂ column density (Kudryavtseva et al. 2012; Fukui et al. 2014). We propose that S44 may be a miniature version of a super star cluster.

5 Conclusions

We summarize the conclusions of the present study in the following.

1. We made new CO observations toward the mid-infrared bubble S44 using NANTEN2, Mopra, and ASTE. We identified two clouds, at −84 km s⁻¹ and −79 km s⁻¹, in the direction of the bubble.

2. The −84 km s⁻¹ cloud shows diffuse CO emission that extends outside of the bubble, with R_3−2/1−0 greater than 0.6 on the northern side of the bubble. From the Mopra and ASTE data sets, the −79 km s⁻¹ cloud corresponds morphologically to the 8 μm emission, with R_3−2/1−0 greater than 0.8 around the bubble. The ionized-gas and cold-dust images have a spatial correlation with the bubble.

3. We estimate the spectral type of the exciting star to be O8.5–9 (∼20 M_⊙), from the SUMSS 843 MHz (36 cm) radio-continuum flux, if we assume a single object.

4. The two clouds are connected by a bridging feature at intermediate velocities that overlap on the eastern side of the bubble. These observational signatures are interpreted as being due to the interaction between the two clouds.

5. We hypothesize that the two clouds collided with each other 3 Myr ago, triggering the formation of the O star(s) and the isolated bubble. A cloud–cloud collision scenario can explain the morphology of the two clouds and the origin of the isolated O-star.

Acknowledgments

We are grateful to the referee, Dr. Christer Watson, for thoughtful comments on the paper. We are also grateful to Mr. Kazuki Okawa for a useful discussion. NANTEN2 is an international collaboration of ten universities: Nagoya University, Osaka Prefecture University, University of Cologne, University of Bonn, Seoul National University, University of Chile, University of New South Wales, Macquarie University, University of Sydney, and Zurich Technical University. The work is financially supported by Grants-in-Aid for Scientific Research (KAKENHI Nos. 15K17607, 15H05694) from MEXT (the Ministry of Education, Culture, Sports, Science and Technology) and JSPS (Japan Society for the Promotion of Science). The ASTE telescope is operated by National Astronomical Observatory of Japan (NAOJ). This work is based on observations made with the Spitzer Space Telescope, which is operated by the Jet Propulsion Laboratory, California Institute of Technology under a contract with NASA. Support for this work was provided by NASA. The authors would like to thank Enago (http://www.enago.jp) for the English language review. The Mopra radio telescope is part of the Australia Telescope National Facility which is funded by the Australian Government for operation as a National Facility managed by CSIRO. The University of New South Wales Digital Filter Bank used for the observations with the Mopra Telescope was provided with support from the Australian Research Council. 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.

References