High-mass star formation possibly triggered by cloud–cloud collision in the H ii region RCW 34

Abstract

We report on the possibility that the high-mass star located in the H ii region RCW 34 was formed by a triggering induced by a collision of molecular clouds. Molecular gas distributions of the ¹²CO and ¹³CO J = 2–1 and ¹²CO J = 3–2 lines in the direction of RCW 34 were measured using the NANTEN2 and ASTE telescopes. We found two clouds with velocity ranges of 0–10 km s⁻¹ and 10–14 km s⁻¹. Whereas the former cloud is as massive as ∼1.4 × 10⁴ M_⊙ and has a morphology similar to the ring-like structure observed in the infrared wavelengths, the latter cloud, with a mass of ∼600 M_⊙, which has not been recognized by previous observations, is distributed to just cover the bubble enclosed by the other cloud. The high-mass star with a spectral type of O8.5V is located near the boundary of the two clouds. The line intensity ratio of ¹²CO J = 3–2/J = 2–1 yields high values (≳1.0), suggesting that these clouds are associated with the massive star. We also confirm that the obtained position–velocity diagram shows a similar distribution to that derived by a numerical simulation of the supersonic collision of two clouds. Using the relative velocity between the two clouds (∼5 km s⁻¹), the collisional time scale is estimated to be ∼0.2 Myr with the assumption of a distance of 2.5 kpc. These results suggest that the high-mass star in RCW 34 was formed rapidly within a time scale of ∼0.2 Myr via a triggering of a cloud–cloud collision.

1 Introduction

1.1 High-mass star formation

High-mass stars strongly affect the environment of the interstellar medium (ISM) by injecting a large amount of energy via ultraviolet (UV) photons, stellar winds, and supernova explosions. Materials around the massive stars are ionized by the UV radiation and are swept up by the stellar winds or supernova explosions, leading to the formation of an H ii region. Heavy elements produced in the massive stars also have a great influence on the chemical evolution of the ISM. Investigating the formation process of massive stars and the environment around them are essential to elucidate the structure and evolution of galaxies. However, their remarkably short lives and small population create difficult observational constraints. Theoretical studies have remained controversial due to the lack of observational guidance.

In general, “core accretion” and “competitive accretion” are thought be promising pictures to explain the formation process of high-mass stars (e.g., Zinnecker & Yorke 2007; Tan et al. 2014). Neither of these models, however, has been established due to the lack of convincing confrontation between the theories and observations. Elmegreen and Lada (1977) proposed the model that the massive stars are formed by the ambient material accumulated by the expanding motions of the ionized gas: high pressure in a shock wave driven by the expansion accumulates the surrounding material and forms a dense molecular layer, which becomes gravitationally unstable and collapses to form the next generation of stars (“collect & collapse”). There are several candidates for high-mass star-forming regions suggesting the scenario of collect & collapse (e.g., Deharveng et al. 2008; Zavagno et al. 2010).

On the other hand, recent observational studies of the molecular clouds distributed in star-forming regions suggest the possibility that high-mass stars could be formed by triggers of supersonic collisions between the interstellar clouds. A large velocity difference (∼10–30 km s⁻¹) between these clouds is too large to be gravitationally bound, but they are likely physically associated as verified by the morphological correspondence and bridging features connecting two clouds in velocity space. The excited gas distribution around the massive stars suggests physical associations between the clouds and massive stars. Molecular clouds having these properties are found in several massive star clusters—Westerlund 2, NGC 3603, RCW 38, and the Orion Nebula Cluster (Furukawa et al. 2009; Ohama et al. 2010; Fukui et al. 2014, 2016, 2017)—as well as H ii regions harboring a single O star—RCW 120 and M 20 (Torii et al. 2015, 2017). ALMA observations also found similar diagnostics in the high-mass star-forming regions in the Large Magellanic Cloud, N 159 West, N 159 East, and R 136 (Fukui et al. 2015; Saigo et al. 2017). The typical time scale of the collision is estimated as ∼0.1–1 Myr, suggesting that the high-mass stars are formed rapidly within a few 10⁵ yr.

Theoretical studies also showed that the cloud collision can be a trigger of the formation of massive stars (Habe & Ohta 1992; Anathpindika 2010; Takahira et al. 2014; Haworth et al. 2015a, 2015b). Magnetohydrodynamical simulations have derived that the collision induces the formation of dense self-gravitating clumps at the compressed region, leading to a large mass accretion rate (10⁻⁴ to 10⁻³ M_⊙ yr⁻¹), enough to create massive stars (Inoue & Fukui 2013). Numerical simulations have also shown that a collision between different sizes of clouds creates a cavity in the larger cloud, forming a ring-like gas distribution (e.g., figure 12 in Torii et al. 2017). After forming the massive stars, the surrounding materials are heated by the UV radiation, which is observed well in the infrared wavelengths (e.g., Torii et al. 2015). The cloud–cloud collision is an alternative view to give a unified explanation for the formation of massive stars and the surrounding structures such as the infrared ring.

1.2 RCW 34 in the Vela Molecular Ridge

RCW 34 is one of the H ii regions located in the Vela Molecular Ridge, a famous star-forming region on the southern galactic plane consisting of several giant molecular clouds (e.g., Yamaguchi et al. 1999). It is characterized by a ring-like structure extended over 4 pc × 7 pc at a distance of 2.5 kpc (Bik et al. 2010) accompanying a single O star, vdBH 25a (O8.5V; Heydari-Malayeri 1988; Bik et al. 2010), and two early B-type stars (B0.5V and B3V; Bik et al. 2010) located near the peak of the infrared emission. Figure 1 shows a composite color image of RCW 34, in which the positions of the O and early B stars and the young stellar objects (YSOs) classified into class 0/I, class I/II, and class II (Bik et al. 2010) are overlaid. One can see a large bubble enclosed by the infrared ring observed by Spitzer/IRAC 8 μm (green) and Herschel/SPIRE 250 μm (red), which mainly traces the polycyclic aromatic hydrocarbon (PAH) and cold dust, respectively. The WISE 22 μm (blue) emission probably tracing hot dust is extended around the massive stars. While the massive stars and the class 0/I and I/II objects are located within the 8 μm ring, the class II objects are distributed extensively in the bubble.

Heydari-Malayeri (1988) conducted a detailed optical study, identifying the spectral type of the O star, and suggested high velocity flow (champagne flow) from the exciting star toward the southern direction by a measurement of H α. As a complementary study, CO observations of the molecular clouds toward RCW 34 were carried out by Pagani, Heydari-Malayeri, and Castets (1993). They revealed gas distribution partly in front of the ionized gas, supporting the large visual extinction at the northern part of the bubble. By using radio and near-infrared data obtained with VLT spectroscopic and Spitzer photometric observations, Bik et al. (2010) found that RCW 34 consists of three different regions according to distributions of the YSOs. Van der Walt, de Villiers, and Czanik (2012) revealed more detailed stellar distribution across the entire H ii region including fainter low-mass stars traced by near-infrared observation with the IRSF (InfraRed Survey Facility) telescope.

Bik et al. (2010) proposed a scenario for the formation of RCW 34: a sequential star formation in the bubble from the south to north triggers the formation of the massive stars, which promotes current star formations in the dense molecular clouds. However, the class II objects in the bubble show spatially extended distribution rather than clustering in one zone (see figure 1). It is difficult to find the age gradient from these stellar objects with the current data accuracy. As suggested by van der Walt, de Villiers, and Czanik (2012), other star-forming scenarios should be considered if the distributions of the low-mass YSOs are found to be spread out almost uniformly across the H ii region. Revealing the initial condition triggering the sequential star formation is also a remaining issue.

In the Vela Molecular Ridge, the youngest super star cluster RCW 38 is suited at ∼5° away from RCW 34. Recent observational studies of molecular clouds found that the formation of massive stars in RCW 38 can be explained by a triggering induced by cloud–cloud collision (Fukui et al. 2016). The Vela Molecular Ridge has several H ii regions showing collisional diagnostics which may form the massive stars (Sano et al. 2018; Enokiya et al. 2018). RCW 34, characterized by the ring-like structure with a single O star (and two early B-type stars), might also be explained by a cloud collision, as suggested by other star-forming regions, RCW 120 and M 20 having similar numbers of massive stars and cloud mass (Torii et al. 2015, 2017). In addition, detailed studies of molecular clouds in RCW 34 using recent millimeter/sub-millimeter observatories have not yet been reported. With the molecular transition lines of ¹²CO and ¹³CO J = 2–1 data obtained by NANTEN2, and ¹²CO J = 3–2 data provided by the ASTE observation, we investigate the possibility of the formation of the high-mass star and the surrounding structure via cloud–cloud collision.

This paper is organized as follows. In section 2, we describe observations of the molecular clouds in RCW 34 by the NANTEN2 and ASTE telescopes. Section 3 shows the results obtained from the ¹²CO, ¹³CO J = 2–1, and ¹²CO J = 3–2 data. In section 4, we discuss the possibility of cloud–cloud collision by comparison with simulation results and studies of the other star-forming regions. We give the conclusion of our study in section 5. In this paper, the minimum contour level is set to be 5 σ unless otherwise noted.

2 Observation

2.1 NANTEN2 ¹²CO and ¹³CO J = 2–1 observations

With the NANTEN2 4 m millimeter/sub-millimeter telescope located in Chile, a simultaneous ¹²CO and ¹³CO J = 2–1 observation toward RCW 34 was conducted from 2015 October 31 to November 2. The OTF (on-the-fly) mode was used for the mapping observation with orthogonal directions to reduce the scanning effect, which provides cubic data covering a scale of |${15^{\prime }}$| × |${15^{\prime }}$|⁠. The system temperature including atmosphere was ∼130 K in the double side band. A Fourier digital spectrometer installed on the backend of the beam transmission system provides data resolved into 16384 channels at 1 GHz bandwidth. The velocity range covers 650 km s⁻¹ at 230 GHz, giving a frequency resolution of 61 kHz, which corresponds to a velocity resolution of 0.08 km s⁻¹. The pointing accuracy was measured to be ≲|${10^{\prime \prime }}$| by observing IRC +10216 located at (RA, Dec) = (⁠|${09^{\rm h}13^{\rm m}57{^{\rm s}_{.}}40}$|⁠, |${+13^{\circ}16^{\prime }04{^{\prime\prime}_{.}}5}$|⁠). The absolute intensity was estimated by calibration using the CO data from Orion-KL at (RA, Dec) = (⁠|${05^{\rm h}35^{\rm m}14{^{\rm s}_{.}}48}$|⁠, |${-5^{\circ}22^{\prime }27{^{\prime\prime}_{.}}55}$|⁠), which were obtained by the large-scale survey of the Orion molecular clouds performed by Nishimura et al. (2015). These data were smoothed with a Gaussian of σ = |${45^{\prime \prime }}$|⁠, which gives a beam size of |${90^{\prime \prime }}$| on the spatial map, and were smoothed to be 0.5 km s⁻¹ in the velocity axis. The typical rms noise levels for the ¹²CO and ¹³CO J = 2–1 data at the smoothed velocity resolution of 0.5 km s⁻¹ were ∼0.3 K per channel and ∼0.5 K per channel, respectively.

2.2 ASTE ¹²CO J = 3–2 observation

Observations of the ¹²CO J = 3–2 rotational transition line toward RCW 34 were carried out in 2015 November using the ASTE 10 m sub-millimeter telescope situated in Chile (Ezawa et al. 2004, 2008). The OTF mapping observation covered the |${7{^{\prime}_{.}}5}$| × |${7{^{\prime}_{.}}5}$| field of view arranged almost centered on the exciting star, vdBH 25a. The typical system temperature including the atmosphere was measured to be ∼350–550 K in the single side band. The digital spectrometer “MAC”¹ with a total bandwidth of 128 MHz and 0.125 MHz resolution at 350 GHz provided the data covering a velocity range of 111 km s⁻¹ with a velocity resolution of 0.11 km s⁻¹. The pointing accuracy was checked by observing RAFGL 4078 located at (RA, Dec) = (⁠|${07^{\rm h}45^{\rm m}02{^{\rm s}_{.}}41}$|⁠, |${-71^{\circ}19^{\prime }45{^{\prime\prime}_{.}}7}$|⁠) every ∼1 hr to achieve an offset better than |${3^{\prime \prime }}$|⁠. The absolute intensity was calibrated by using the measurements of the CO line from a standard source, IRC + 10216.² Finally, these data were smoothed to be 0.3 km s⁻¹ in the velocity axis and |${24^{\prime \prime }}$| on the spatial map. The typical rms noise level was ∼0.3 K per channel at the smoothed velocity resolution of 0.3 km s⁻¹.

3 Results

3.1 Two clouds at different velocities

Figures 2a and 2b indicate the intensity distributions of ¹²CO and ¹³CO J = 2–1 lines, respectively, with the integrated velocity ranges of 0 to 14 km s⁻¹. We found similar gas distributions to the infrared 8 μm ring shown by the yellow line. The cloud in the northern area shows an arc-shaped morphology elongated to the east–west direction, which contains the main peak at (l, b) ∼ (⁠|${264.\!\!^{\circ}29}$|⁠, |${1.\!\!^{\circ}48}$|⁠). Another peak is detected at (l, b) ∼ (⁠|${264.\!\!^{\circ}4}$|⁠, |${1.\!\!^{\circ}43}$|⁠) in a cometary-like cloud distributed in the southern area. The bubble-like structure enclosed by the infrared ring (see figure 1) is also confirmed in the CO distribution. The exciting star, vdBH 25a, with the spectral type O8.5V, is located slightly south of the highest CO peak. The CO intensity around the O star is drastically changed, as represented by the white contours with high contrast. From the ¹²CO data, we found low-intensity diffuse gas enclosed by the infrared ring, whereas the ¹³CO data did not show significant intensities (even for a 3 σ detection level). To see the gas distribution in narrower velocity ranges, we made a velocity channel map separated into every 2 km s⁻¹ with V_LSR (velocity of the local standard of rest) from 0 km s⁻¹ to 14 km s⁻¹ for the ¹²CO data as shown in figure 3. We found that the gas distribution is mainly separated into two components with different velocities, V_LSR = 0–10 km s⁻¹ and 10–14 km s⁻¹ (hereafter denoted by “blue cloud” and “red cloud,” respectively). The blue cloud exhibits the ring-like gas distribution which includes the highest intensity peak at V_LSR ∼ 7 km s⁻¹, and low signifiant intensity gas extended in the bubble, such as the diffuse component at V_LSR ∼ 7 km s⁻¹ and small clumpy structures of size ∼1 pc at (l, b) ∼ (⁠|${264.\!\!^{\circ}35}$|⁠, |${1.\!\!^{\circ}475}$|⁠) and (⁠|${264.\!\!^{\circ}34}$|⁠, |${1.\!\!^{\circ}42}$|⁠) at V_LSR ∼ 9 km s⁻¹. The red cloud with the low-intensity emission is distributed to just cover the northern part of the bubble, which has not been recognized by previous molecular observations, probably due to low detection sensitivity.

In figures 4a and 4b we show the integrated intensity distributions of the blue and red clouds, respectively. Figure 4c compares the positions of the two clouds represented by the contours (red cloud) overlaid on the image (blue cloud). According to the peak velocities of each cloud (at V_LSR ∼ 7 km s⁻¹ and ∼12 km s⁻¹ for the blue and red clouds, respectively), these clouds have a relative velocity of ∼5 km s⁻¹. The exciting star is located between the two clouds, where the blue cloud shows a large gradient of intensity, while the red cloud shows low significant emission with an intensity of ∼2 K km s⁻¹ (∼5 σ). The northern part of the red cloud is enclosed by the blue cloud. The complementary distribution of the two clouds, which harbors high-mass stars near the boundary between them, is a characteristic structure recently found in several high-mass star-forming regions (e.g., Torii et al. 2015, 2017; Fukui et al. 2017). Figure 4d presents a longitude–velocity diagram with the integrated latitude from b = |${1.\!\!^{\circ}425}$| to |${1.\!\!^{\circ}5}$| as shown by the yellow dashed lines in figure 4c. In addition to the two intensity peaks included in the blue cloud (V_LSR < 10 km s⁻¹), we confirmed the low significant (∼5–15 σ) velocity structure corresponding to the red cloud in V_LSR > 10 km s⁻¹. The low-intensity gas extended between the two peaks in the blue cloud corresponds to the bubble structure inside the infrared ring. We found a wide velocity feature near the position of the O star as shown by the white dashed line, which may suggest outflows due to the YSOs around the massive stars.

We estimated the molecular gas masses and the column densities of the blue and red clouds assuming that the intensity ratio between the ¹²CO J = 2–1 and J = 1–0 lines is 0.6, a typical intensity ratio obtained by the galactic plane survey performed by Yoda et al. (2010), which gives an almost consistent result obtained by a recent theoretical study based on a hydrodynamical simulation (Peñaloza et al. 2017). With the assumptions of the distance of 2.5 kpc toward RCW 34 (Bik et al. 2010) and the CO-to-H₂ conversion factor X_CO = 1.0 × 10²⁰ H₂ molecule cm⁻² K⁻¹ km⁻¹ s (e.g., Okamoto et al. 2017), the molecular masses for the blue and red clouds are derived to be ∼1.4 × |$10^{4}{\,\,}M_{\odot }$| and |$\sim 600{\,\,}M_{\odot }$|⁠, respectively. The peak values of the gas column density (⁠|$N_{\rm H_{2}}$|⁠) for the blue and red clouds are 1.1 × 10²² cm⁻² and 1 × 10²¹ cm⁻², respectively. We also calculated the column density of the blue cloud to take into account the self-absorption effect often observed in the optically thick ¹²CO lines in high-density regions. Based on the model formula assuming local thermodynamic equilibrium (Nishimura et al. 2015), we estimated the optical depth of the ¹³CO J = 2–1 line with the excitation temperature (T_ex) obtained from the ¹²CO J = 2–1 peak and derived the total molecular column density. The typical optical depth of the ¹³CO J = 2–1 line is ∼0.1 and the peak column density for the blue cloud is estimated to be ∼2.1 × 10²² cm⁻², which is consistent within the uncertainty by a factor of two compared to the value obtained with the ¹²CO J = 2–1/J = 1–0 intensity ratio. The low CO intensity of the red cloud would suggest less self-absorption effect, and hence the column density of the red cloud would not change significantly. Detailed studies of the column density and mass estimate are beyond the scope of this paper. Hereafter we use the column density and molecular gas mass estimated from the ¹²CO J = 2–1/J = 1–0 intensity ratio.

3.2 ¹²CO J = 3–2 results

The ¹²CO J = 3–2 data obtained with ASTE are useful for tracing more excited gas properties in high-density clouds. Figure 5 indicates the velocity channel distributions of the J = 3–2 data separated into every 2 km s⁻¹ with V_LSR from 0 to 14 km s⁻¹. Although the gas distribution is almost the same as the J = 2–1 data (see the velocity channel map in figure 3), the J = 3–2 data with the higher angular resolution trace more tiny gas structures. Like the molecular gas distribution obtained by H₂ line observations (see figure 4 in Bik et al. 2010), the gas distributions in V_LSR from 2 km s⁻¹ to 8 km s⁻¹ for the blue cloud have mildly curved shape at the northern part to the O star, where the intensity peak at V_LSR ∼7 km s⁻¹ is separated into two elongated structures. As traced by the J = 2–1 velocity channel map in figure 3, low significant diffuse emission (∼5 σ) is observed at V_LSR ∼ 7 km s⁻¹ as well as the clumpy gas structures at (l, b) ∼ (⁠|${264.\!\!^{\circ}35}$|⁠, |${1.\!\!^{\circ}475}$|⁠) and (⁠|${264.\!\!^{\circ}34}$|⁠, |${1.\!\!^{\circ}42}$|⁠) with sizes of ∼1 pc at V_LSR ∼ 9 km s⁻¹ distributed in the bubble. While the J = 2–1 data indicate that the red cloud (V_LSR > 10 km s⁻¹) keeps a cloud morphology with almost symmetric structure at a significant detection level, the J = 3–2 data in the low-density areas of the southeast part of the O star do not show significant J = 3–2 emission (e.g., (l, b) ∼ (⁠|${264.\!\!^{\circ}31}$|⁠, |${1.\!\!^{\circ}47}$|⁠) at V_LSR ∼ 11 km s⁻¹). The non-detection of the J = 3–2 line might be due to low temperature or low density in the area compared with the critical density required for the J = 3–2 transition. The large beam size of the NANTEN2 J = 2–1 data or insufficient sensitivity of the ASTE J = 3–2 data possibly causes a difference in the CO distribution for the red cloud.

3.3 ¹²CO J = 3–2/J = 2–1 intensity ratio

Figure 6 indicates the velocity channel distributions representing the intensity ratio of ¹²CO J = 3–2 to J = 2–1 (hereafter denoted by |$R_{\rm 3-2/2-1}$|⁠). Taking intensity ratios between the different rotational transitions is useful to infer the gas properties of the exciting temperature or density. We need to carefully evaluate these ratios since the optically thick lines such as ¹²CO lower the intensity, especially in opaque regions. To compare the intensities between the two lines, the J = 3–2 data were smoothed to the NANTEN2 beam size of |${90^{\prime \prime }}$|⁠. In the ratio map, we mask the three pixels from the edges, whose sizes are equivalent to the the beam diameter, because the correction of the beam size generates ambiguous uncertainties for the outermost pixels. In each velocity range, we also overlay contours to represent the integrated intensity of the J = 3–2 line. The obtained |$R_{\rm 3-2/2-1}$| distribution shows that most of the molecular gas has a high ratio (≳1) throughout the velocity ranges. Using the three CO transition lines, we performed a Large Velocity Gradient analysis (Goldreich & Kwan 1974) with assumptions of the abundance ratios of [¹²CO]/[H₂] = 10⁻⁴ (e.g., Frerking et al. 1982; Leung et al. 1984) and [¹²C]/[¹³C] = 77 (Wilson & Rood 1994), and a typical velocity gradient roughly estimated from our ¹²CO J = 2–1 observation, dv/dr = (4 km s⁻¹)/(2 pc). The results give a lower limit of T_ex ∼ 200 K. We confirmed |$R_{\rm 3-2/2-1}$| ≳ 1 in T_ex > 200 K. The optical depth of the ¹²CO lines is 2–8 within the typical H₂ density of the molecular clouds, 10²–10⁴ cm⁻³, and the J = 2–1 and 3–2 transitions have T_ex lower than the kinetic temperature T_k (i.e., sub-thermal excitation), indicating that the observed brightness temperature (T_b) for these lines is weaker than T_k. Even if we take into account the possible large optical depth in ¹²CO, |$R_{\rm 3-2/2-1}$| exhibits high values, supporting our result of the high intensity ratio shown in figure 6.

To compare the line spectra among the different areas, we show the ¹²CO J = 2–1, J = 3–2, and ¹³CO J = 2–1 spectra in figure 7 for the areas of A–I (shown in the bottom-right panel in figure 6). These spectra represent the sum of T_b within the 3 × 3 pixels. The intensity peaks at V_LSR ∼ 6–8 km s⁻¹ in the blue cloud and V_LSR ∼ 10–12 km s⁻¹ in the red cloud indicate high |$R_{\rm 3-2/2-1}$|⁠, as in the following examples shown by the shaded areas: (i) The spectrum in area E containing the O-star position exhibits a similar spectral shape but different intensities among the three lines, which gives a high |$R_{\rm 3-2/2-1}$| (∼1.5). (ii) Area G, located ∼1–3 pc away from the O star, in which weak line emission from the red cloud is detected, also shows a high |$R_{\rm 3-2/2-1}$| comparable to area E. (iii) The high-density region in the blue cloud, area C, located opposite area G around the O star, yields high |$R_{\rm 3-2/2-1}$| values up to ∼1.0. As shown by the integrated intensity map in figure 2b, significant ¹³CO J = 2–1 emission from area G is not detected.

4 Discussion

4.1 Properties of molecular clouds associated with the high-mass star

In the previous section, we found two molecular clouds separated in velocity with ranges of 0–10 km s⁻¹ (blue cloud) and 10–14 km s⁻¹ (red cloud), whose masses are ∼1.4 × 10⁴ M_⊙ and ∼600 M_⊙, respectively. We summarize several pieces of evidence of physical associations between the two clouds and the massive stars:

1. The blue and red clouds show a complementary distribution. While the blue cloud with an arc-shaped structure distributes along the infrared 8 μm ring, the red cloud is distributed to cover the northernmost part of the bubble, which consists of the diffuse gas emission and clumpy gas structures with sizes of ∼1 pc. The O / early B stars are located near the boundary between the two clouds, where the blue cloud is possibly highly compressed due to the feedback from the massive stars, as suggested by its mildly curved shape in the neighborhood of the O star. The diffuse gas component near the massive stars in the red cloud is probably disrupted by the strong UV radiation.

2. From the position–velocity diagram, we confirmed the velocity of each of the two clouds, as well as the bubble structure sandwiched by the two peaks of the blue cloud. We will discuss this velocity feature in a comparison with simulation results in subsection 4.3.

3. The line intensity ratio of ¹²CO J = 3–2/J = 2–1 in the blue and red clouds shows high values (≳1.0), possibly due to heating by the UV radiation from the massive stars, suggesting physical associations between the two clouds and the high-mass stars.

In figure 8, we compare the distribution of the ¹²CO J = 3–2 line with that of the infrared 8 μm in a velocity channel map. The 8 μm emission mainly traces the PAH, which is probably created by the photodissociation induced by the UV radiation from the massive stars. As shown by the white circles in the figure, we found good correspondence between the clumpy gas structures and the 8 μm distributions for both the blue and red clouds. In each clump, the 8 μm emission on the side facing the exciting stars tends to be higher as compared to the opposite side. These results indicate that the molecular clouds distributed in the bubble are strongly affected by the UV radiation at a far distance, at least ∼3 pc, supporting physical associations between the molecular clouds and the massive stars.

4.2 Possible scenarios to form the gas and stellar distributions in RCW 34

A triggered mechanism of massive-star formation in the H ii region is generally explained by the collect & collapse scenario (Elmegreen & Lada 1977). The expanding motion of the ionized gas due to the feedback from the massive star sweeps up the surrounding material between the ionization front and the shock front (e.g., Hosokawa & Inutsuka 2006). The collected material forming a dense molecular layer becomes gravitationally unstable and collapses to form the next generation of stars in individual fragmentations (e.g., Whitworth et al. 1994). There are several candidates for the high-mass star-formation site on the borders of the H ii region, which can be explained by collect & collapse (e.g., Deharveng et al. 2008; Zavagno et al. 2010). Assuming an ideal uniform medium to form a massive star via collect & collapse, star-formation processes should be observed regardless of the directions, which yields a symmetric gas distribution (e.g., figure 1 in Deharveng et al. 2010 and figure 8 in Torii et al. 2015) to form next-generation stars. In practice, such a concentric gas distribution may not be observed because of gas disruption due to the radiative heating and the projection effect along the line of sight. The initial conditions to trigger forming massive stars can also change the observed gas distributions. However, as suggested by Bik et al. (2010), more clumps should be observed in the southern area of the O star if collect & collapse can be applied to RCW 34. Young stellar objects such as the class 0/I and I/II objects corrected to the gas distribution would be found in the inner bubble (see figure 1).

Bik et al. (2010) proposed another scenario, sequential star formation from south to north, which eventually gives rise to the formation of massive stars including the YSOs such as the class 0/I objects located on the northern part of the bubble. However, it is unclear whether there is an age gradient across the whole bubble because the formation time scale of the class II objects distributed in the bubble has large uncertainty [a few Myr; Bik et al. (2010) and references therein]. In addition, the class 0/I and I/II objects located around (l, b) ∼ (⁠|${264.\!\!^{\circ}33}$|⁠, |${1.\!\!^{\circ}43}$|⁠) are hard to explain with the scenario of sequential star formation from south to north. The class II objects are likely to be spread out in the bubble extensively rather than clustering in one zone. We infer that the formation process of the class II objects in the bubble are not strongly related to the past formation history of the massive stars and the ring-like gas distribution.

As an alternative interpretation, we propose a scenario for the formation of the massive star and ring-like structure in RCW 34 via cloud–cloud collision, which can explain the properties of the molecular clouds summarized in subsection 4.1. An accidental supersonic collision between the blue and red clouds induces the formation of cloud cores near the boundary of the clouds as a consequence of the shock compression, giving a trigger of a high accretion rate to form high-mass stars. In the center of the blue cloud, a large cavity is created by the collisional effect, resulting in the complementary distribution of the two clouds. The molecular gas around the O star is heated by the strong UV radiation and ultimately disrupted, where the CO emission becomes weak or cannot be observed. Molecular clouds accompanying high-mass stars have been found recently in several super star clusters and H ii regions, which can be explained by cloud–cloud collision (Furukawa et al. 2009; Ohama et al. 2010; Fukui et al. 2014, 2016, 2017; Torii et al. 2015, 2017). The cloud–cloud collision model can also be applied to the other high-mass star-forming regions in the Vela Molecular Ridge (RCW 38: Fukui et al. 2016; RCW 32: Enokiya et al. 2018; RCW 36: Sano et al. 2018). Among these candidates summarized in Fukui et al. (2017), RCW 34 with a single O star is characterized by the two clouds with relatively lower mass.

In establishing the model of high-mass star formation, the constraint on the initial condition to trigger the star-forming process is a significant subject. Although the major theoretical models “core accretion” and “competitive accretion” considering the turbulent effect are commonly thought to be promising pictures to explain the high accretion rate to form massive stars (e.g., McKee & Tan 2003), the gas column density in RCW 34 obtained in this study (∼4 × 10⁻² g cm⁻² for the blue cloud) does not satisfy the threshold of the column density to halt fragmentation (1 g cm⁻²; Krumholz & McKee 2008). It is difficult to convince these models observationally in terms of the required initial condition. The model of collect & collapse including the sequential star formation suggested by Bik et al. (2010) described above does not give observational constraints on the triggering process on the onset of the sequential star formation. On the other hand, the model of cloud–cloud collision based on the gravitationally unbound system can satisfactorily explain in RCW 34 the triggering process to form the massive stars by likely amplifying the magnetic field and the turbulent velocity for high mass accretion rate (Inoue & Fukui 2013).

The velocity channel map in figure 3 exhibits lobe-like gas distributions with the highly blue-shifted (V_lsr ∼ 0–4 km s⁻¹) and red-shifted (V_lsr ∼ 10–14 km s⁻¹) components toward the northern and southern directions, respectively, indicating the possibility that the gas distribution in RCW 34 is formed by molecular outflow from the star-forming region around the O star instead of the collision of the two clouds. This is supported by the wide velocity feature found in the position–velocity diagram in figure 4d. As shown in figure 7, most of the ¹²CO spectra indicate a possible blue-shifted lobe (for areas A, B, D, E, F, H, and I), whereas the spectrum for area G does not have significant emission of the blue-shifted lobe. The spectrum in area G shows a significant red-shifted lobe, whereas another emission peak at V_lsr ∼ 6 km s⁻¹ is detected. Such a spectral feature with a velocity jump is difficult to understand with the molecular outflow from the massive star-forming region around the O star; the significant emission connecting the red-shifted lobe (included in the red cloud) and the ∼6 km s⁻¹ peak (included in the blue cloud) can rather be attributed to a bridging feature often observed in candidates for cloud–cloud collision. Hereafter we discuss the possibility that the formations of the O star and the ring-like gas distribution in RCW 34 are via a collision between the two clouds.

4.3 Velocity structure compared with simulation results

Figure 9a shows the longitude–velocity diagram of the ¹²CO J = 2–1 data (same as figure 4d), in which the two lines connecting the intensity peaks in the blue and red clouds are overlaid. A velocity gradient characterized by an inverted v-shaped structure emerges, implying the possibility that the central cavity was formed by crashing the red cloud into the blue cloud with an angle in the direction of the lower Galactic longitude to the line of sight.

We compared this velocity feature with a numerical simulation of cloud–cloud collision, in which clouds with different sizes of Bonnor–Ebert spheres (Bonnor 1956) are assumed (Takahira et al. 2014). In this comparison, we do not compare the ¹²CO J = 3–2 data, because the J = 3–2 observation does not cover the whole clouds. In the model calculation, a turbulent effect is given to the clouds, although any radiative feedback from the massive stars is not considered. Assuming an initial collision velocity of 10 km s⁻¹ with a viewing angle of 45° to the line of sight, we analyzed data obtained by a synthetic ¹²CO J = 1–0 observation (Haworth et al. 2015a, 2015b) based on the collisional model provided by Takahira, Tasker, and Habe (2014). We note that the physical conditions of the colliding clouds in this simulation do not match our observational study, but a comparison with the simulation result is useful to investigate qualitatively how the velocity distribution is observed by the collisional effect. Figure 9b shows a snapshot of the position–velocity diagram at 1.6 Myr after the collision started, when the relative speed was decreased to 7 km s⁻¹.³ We confirmed the gas distribution having inverted v-shaped and cavity-like structures similar to our observational result, which suggests a past collisional event in RCW 34 between the blue and red clouds. The simulation result traces a highly compressed region shown by the elongated high-intensity feature, which corresponds to the bridging site connecting the two clouds in velocity. Turbulent gas motions are expected there, possibly leading to the high-mass star formation. Our observational results, however, cannot verify tiny gas structures due to the poor spatial and velocity resolutions. Future follow-up observations by ASTE toward the southern area will provide the velocity feature of the whole clouds and ALMA observations of the colliding area are expected to resolve filamentary gas structures associated with the turbulent gas motions suggested by numerical simulation (Inoue & Fukui 2013; Inoue et al. 2017), which will be useful to finally probe whether cloud–cloud collision is a relevant process in the formation of high-mass stars.

4.4 Collisional time scale

Figure 10 indicates the integrated intensity distribution of the ¹²CO J = 2–1 data for the blue cloud (same as figures 4a and 4c), overlaying the intensity of the red cloud by the white contours. The dashed contours indicate the intensities for the red cloud applying a displacement of 1 pc toward the positive longitude direction. The displacement makes the red cloud cover the overall bubble, showing a further morphological correspondence between the blue and red clouds. If the cloud–cloud collision model can be applied to these clouds, the existence of a cavity whose size is the red cloud is probable, and movement of the red cloud toward the north direction is expected. The morphological correspondence with the displacement implies that the collision started from there and the formation of the O star was triggered by a continuous high compression at the shock front. We therefore adopted the size of displacement of 1 pc to estimate the collisional time scale. If we apply tentatively 45° for the viewing angle of the collision to the line of sight, the distance between the two clouds is D = 1/sin(45°) (pc) and the velocity difference estimated from the observed relative velocity is V = 5/cos(45°) (km s⁻¹). This inclination angle roughly matches the collisional direction suggested from the longitude–velocity diagram discussed in subsection 4.3. The collisional time scale can be estimated to be D/V ∼ 1.5 Myr. If we take into account variable viewing angles of the collision to the line of sight from 30° to 60°, the collisional time scale is derived to be 0.1–0.35 Myr. We note that actual collisional time scale will be shorter since the initial velocity is expected to be higher and reduced to the current velocity due to the deceleration effect. After the collision started, continuous high pressure had been driven between the clouds, which gave a trigger to form the high-mass star; the time scale for forming the O star is inferred to be ≲0.2 Myr.

The collisional time scale of ∼0.2 Myr to form the H ii region RCW 34 is much shorter than the age of 2 ± 1 Myr estimated by the population of pre-main-sequence (PMS) stars distributed in high-density regions around the massive stars (Bik et al. 2010). Meanwhile, the author noted the possibility that these PMS stars are likely much younger because outflows are detected, possibly originating from the YSOs such as the class 0/I objects distributed in the neighborhood of the massive stars (see figure 10). In general, the time scale of the class I objects is thought to be ∼a few 10⁵ years (e.g., Whitney et al. 2003), which is almost consistent with that of O-star formation inferred by the cloud collision. The formation of class 0/I and I/II objects located in the southern area is difficult to explain by the sequential star formation from the south to north suggested by Bik et al. (2010). If the cloud–cloud collision model can be applied to the low-mass star formation (Nakamura et al. 2012), the existence of local class 0/I and I/II objects is understood with the same formation time scale as the other class 0/I and I/II objects. In terms of the formation time scale of the class II objects (a few Myr; Bik et al. (2010) and references therein) and their distributions extensively spread out in the whole bubble, these YSOs might have been formed prior to the cloud collision.

4.5 Comparison with other star-forming regions

Finally, we compared properties of molecular clouds in RCW 34 with those of other candidates for cloud–cloud collision, RCW 120 and M 20 (Torii et al. 2015, 2017), which accompany a single O star in their H ii regions, and the super star cluster, NGC 3603 and RCW 38, having a few tens of massive stars (Fukui et al. 2014, 2016). Table 1 summarizes the physical parameters of these candidates. We note that the uncertainties due to utilizing different CO lines as the gas tracer, angular resolutions, and the assumptions of X_CO among these high-mass star-forming regions do not affect this discussion of the scale of column density in the order of magnitude. Fukui et al. (2017) suggested that the molecular column density of the colliding clouds can be an important parameter to determine the number of O stars. Whereas NGC 3603 and RCW 38 have high column density up to 10²³ cm⁻², other candidates to form a single O star show a lower column density with the range of 10²¹ to ≲10²² cm⁻². The present study of RCW 34 also shows a similar diagnostic that the clouds with column density ∼10²² cm⁻² can form a single O star. Compared to the other two candidates of RCW 120 and M 20 accompanying a single O star, RCW 34 has a large difference in the column density between the two clouds. The formation of an O star might be triggered if one of the parental clouds has a sufficiently large (≳10²² cm⁻²) column density. We also note that the time scales for forming high-mass stars are almost consistent between RCW 34 and RCW 38 in spite of the very different number of massive stars, which may indicate physically associated star-formation history via cloud–cloud collision in the Vela Molecular Ridge. The similar time scale of the cloud collision to trigger the formation of O stars derived in RCW 36 (Sano et al. 2018) supports this indication. We need molecular observations toward other star-forming regions to reveal the properties of clouds related to the massive stars, which may establish the scenario for the formation of the high-mass stars triggered by cloud–cloud collision.

5 Conclusion

Using the ¹²CO and ¹³CO J = 2–1 and ¹²CO J = 3–2 data obtained by the NANTEN2 and ASTE telescopes, we have studied molecular gas distribution toward the H ii region RCW 34, in which a single O star and two early B-type stars are located. The main conclusions are summarized below.

1. We found two clouds separated in velocity with ranges of 0–10 km s⁻¹ (blue cloud) and 10–14 km s⁻¹ (red cloud), whose masses are ∼1.4 × 10⁴ M_⊙ and ∼600 M_⊙, respectively. The blue cloud is distributed to trace the ring-like structure observed in the infrared wavelengths. The red cloud, which has not been recognized in previous observations, consists of diffuse gas emission and clumpy gas structures, covering the northernmost part of the bubble. The blue and red clouds show a complementary distribution. The high-mass stars with spectral types of O8.5V and early B are located near the boundary between the two clouds.

2. The blue cloud is likely to be compressed by the feedback from the O star. The weak CO emission in the red cloud indicates disruptions by the ionization effect from the massive stars. The line intensity ratio of ¹²CO J = 3–2/2–1 shows high values (≳1.0), possibly due to heating by the UV radiation. The infrared 8 μm emission shows good spatial correspondence with the molecular clouds. These results suggest physical associations between the two clouds and the high-mass stars.

3. The position–velocity diagram of ¹²CO J = 2–1 data shows an inverted v-shaped structure, which is similar to the distribution obtained by a synthetic observation with a numerical simulation. We need follow-up observations by ASTE and ALMA to discuss the velocity feature at more finely resolved scales in the whole clouds. If we apply a 1 pc displacement to the red cloud, the morphological correspondence between the two clouds becomes more pronounced. Assuming a collision at a viewing angle 45° to the line of sight, the collisional time scale is estimated to be ∼0.2 Myr, which indicates that the high-mass stars and the surrounding structure in RCW 34 were formed in a much shorter time scale, ≲0.2 Myr. Compared to other massive star-forming regions identified as candidates for cloud–cloud collision, RCW 34 is characterized by high contrast in the molecular column density (or mass) between the two clouds. The number of O stars formed by triggering via cloud–cloud collision may relate to their column densities.

Acknowledgements

We would like to thank Arjan Bik for providing information on the YSOs distributed in RCW 34. 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 ASTE telescope is operated by the National Astronomical Observatory of Japan (NAOJ).

Footnotes

References