ALMA Observations of Giant Molecular Clouds in M33. I. Resolving Star Formation Activities in the Giant Molecular Filaments Possibly Formed by a Spiral Shock

Figure 1 shows the molecular gas distributions in ¹²CO and ¹³CO(J = 2–1) in GMC-16. The missing flux of the 12 m + 7 m data in ¹²CO toward the southern region where the decl. angle is lower than +30°48'50'' is as small as 10%, indicating that the gas distributions are dominated by compact rather than diffuse gas. This is consistent with an early result that active star-forming GMCs in the LMC are highly structured (Sawada et al. 2018). The previous single-dish studies in CO lines by Tosaki et al. (2011), Miura et al. (2012), and Druard et al. (2014) marginally resolved this GMC into two peaks. The ALMA observations clearly reveal multiple filamentary structures with the length scale of ≳50 pc elongated in the north–south direction. The filaments are not randomly distributed but exhibit ordered direction and there are roughly two main filamentary components (Filaments A and B) as indicated in Figure 1(a). In addition to these filaments, there is a relatively "extended gas" (see the final paragraph of this subsection) and a dense clump associated with the 1.3 mm continuum emission (hereafter, MMS, millimeter source) and multiple small filaments (see Section 3.2) at the southern part of the GMC. We estimated the total gas mass and relatively dense gas mass from the ¹²CO and ¹³CO data, respectively. The first one is derived from the ¹²CO data assuming the X_CO factor in M33, 2.0 × 10²⁰ cm⁻²(K km s⁻¹)⁻¹ (Rosolowsky et al. 2007), and CO(2–1)/CO(1–0) ratio of 0.7 (Tosaki et al. 2011; Druard et al. 2014). The mass of higher density regions traced by ¹³CO is estimated by the local thermodynamical equilibrium calculation applying the excitation temperature derived from the ¹²CO data and the relative abundance of [H₂]/[¹³CO] of 1.4 × 10⁶, which is close to an intermediate value adapted in the Galaxy (e.g., Frerking et al. 1982) and the LMC studies (e.g., Fujii et al. 2014; Fukui et al. 2019; Tokuda et al. 2019). The total mass (¹²CO mass) and dense gas mass (¹³CO mass) are 8 × 10⁵ M_⊙ and 2 × 10⁵ M_⊙, respectively. The fraction of ¹³CO/¹²CO mass is ∼0.2, which is consistent with that in the Galactic plane (Torii et al. 2019). The total mass, median (FWHM) width, and length of Filament A/B in the ¹²CO map are measured to be ∼1 × 10⁵ M_⊙, ∼5–6 pc, and ∼50–70 pc, respectively. These characteristics are comparable to those of GMFs, e.g., the Nessie cloud (Jackson et al. 2010), in the Galaxy.

Figure 2(a) shows that GMC-16 is located at an optical spiral arm in the northern part of M33. Based on the morphology of the spiral arm, the moving direction of the gas (i.e., galactic rotation) is considered to be from east to west. Figure 2(b) shows the comparison between the ¹²CO and Hα images. There are several bright H ii regions and some of them are outside of the present field coverage with ALMA. Two GMFs (Filaments A and B) are located at the eastern side of the H ii regions. Although the previous single-dish studies with a spatial resolution of a few ×10 pc identified these H ii regions inside the molecular cloud, the present observations reveal a clear position offset between ionized and CO gas with a separation of ∼10–20 pc.

Zoom In Zoom Out Reset image size

We made channel maps with a velocity bin of 4 km s⁻¹ of the ¹²CO(J = 2–1) data to investigate the velocity structure of the GMC (Figure 3(a)). As one can see in the channel maps, there is a velocity gradient along the north–south direction from blueshifted to redshifted velocity. To further illustrate the difference between the individual components, i.e., Filament A, extended gas, and some of the filaments, we made a position–velocity (PV) diagram of ¹²CO (Figure 3(b)) along the dotted lines shown in Figure 1(a). We manually selected the analyzed region to avoid the complex hub-filamentary structure (Section 3.2). The velocity width (FWHM) of the filamentary structures is measured as ∼3 km s⁻¹, in contrast to this, the extended gas shows much wider velocity width, ∼6 km s⁻¹. The centroid velocity of the extended gas is ∼−240 km s⁻¹, which is apparently redshifted compared to that of Filament A. This difference may be relevant to turbulent dissipation and deceleration by the galactic spiral shock. We discuss this possibility in Section 4.1.

Zoom In Zoom Out Reset image size

Figure 4 shows a zoomed in view of the ¹²CO brightest peak clump in GMC-16. Panel (a) shows the ¹³CO distribution and there are several filamentary structures connecting to the central peak. This type of multiple filamentary structure, which is referred to as the "hub filament" (Myers 2009), is found in high-mass star-forming regions in the Galaxy and the LMC (e.g., Peretto et al. 2013; Williams et al. 2018; Fukui et al. 2019; Tokuda et al. 2019). The hub filament in GMC-16 shows an asymmetric distribution, which is extended in a fan shape toward the east direction. The moment 1 map in Figure 4(b) shows that the northern filament has a central velocity of ∼−248 km s⁻¹, which is different from those of the other filaments (∼−245 km s⁻¹). Note that the hub-filamentary complex is connected to the extended gas (see the moment 0 map (Figure 1) in Section 3.1), but their central velocities are different from each other.

Zoom In Zoom Out Reset image size

We detected the 1.3 mm continuum source (MMS) as well as the C¹⁸O emission at the intersection of a few filaments, i.e., the ¹³CO peak (Figures 4(a) and 4(d)). If there is a protostellar source inside MMS, the source may contribute to the enhancement of the 1.3 mm flux. This effect is considered to be small, and the continuum flux is mainly arising from the cold dust emission, because the gas mass estimated from the 1.3 mm continuum and that from the C¹⁸O emission are similar to each other (see the next paragraph). For the same reason, it is likely that the free–free contamination from the H ii region next to MMS is also negligible. In fact, the peak position of the hub filament as well as MMS is shifted from that of the Hα emission as shown in Figure 4(c). The Spitzer observations found 24 μm emission around this source with the intensity of 82 mJy, which corresponds to the total infrared luminosity of ∼2 × 10⁶ L_⊙ (Verley et al. 2007). Although the brightness of the source suggests the presence of early-O-type star(s) (Martins et al. 2005; Zinnecker & Yorke 2007), the luminosity is mainly arising from the high-mass stars inside the neighboring H ii region.

The deconvolved (FWHM) size of MMS is (053 ± 009) × (027 ± 006), corresponding to ∼2.2 pc×1.1 pc. The total mass of MMS (${M}_{\mathrm{total}}^{\mathrm{MMS}}$) above the 3σ detection is estimated to be ∼2 × 10⁴ M_⊙ assuming the dust opacity of ${\kappa }_{1.3\mathrm{mm}}$ of 1 cm² g⁻¹ for protostellar envelopes (e.g., Ossenkopf & Henning 1994), a dust-to-gas ratio of ∼3 × 10⁻³ (for the LMC-like metallicity; Gordon et al. 2014), and the dust temperature of 20 K. The velocity width (FWHM) of the C¹⁸O emission in MMS is 6.8 km s⁻¹. With a radius of 1.6 pc (=geometric mean of the major and minor axis) the resultant virial mass, ∼2 × 10⁴ M_⊙, is consistent with the ${M}_{\mathrm{total}}^{\mathrm{MMS}}$, indicating that the dense clump is gravitationally bound. Assuming a spherical geometry, the average density of MMS is calculated to be ∼2 × 10⁴ cm⁻³. We note that we could not find any other continuum sources as well as C¹⁸O emission in the observed field, indicating that the high-density region is localized within a few parsecs with respect to the entire molecular cloud of GMC-16.

Figure 4(d) shows spatial distributions of the high-velocity ¹²CO gas toward MMS. The spectra of ¹²CO, ¹³CO, and C¹⁸O are shown in panels (e) and (f). The ¹²CO profile is slightly asymmetric with a peak velocity of ∼−245 km s⁻¹. As shown in the ¹³CO moment map in Figure 4(b), there are two velocity components toward MMS, and the ¹²CO intensity of redshifted gas is stronger than that of the blueshifted gas. The double (or multiple) peaked profile in C¹⁸O is insignificant at the present sensitivity. Since the ¹³CO profile shows a relatively symmetric shape, we use it to estimate the systemic velocity of MMS using Gaussian fitting. The systemic velocity judged from the ¹³CO profile is ∼−246 km s⁻¹ and the maximum relative velocity of the ¹²CO red/blueshifted wing components is ∼20 km s⁻¹. CO observations in the Galaxy and the LMC often found this type of high-velocity gas toward young stellar objects (e.g., Beuther et al. 2002; Fukui et al. 2015, 2019; Shimonishi et al. 2016; Harada et al. 2019; Tokuda et al. 2019). The presence of the gravitationally bound dense material toward MMS strongly indicates that there is at least one embedded high-mass protostar and the ¹²CO high-velocity components originated from its bipolar outflow. This is the first strong candidate for protostellar outflow in M33 as well as the external disk galaxies. We further discuss the reliability of the protostellar outflow in Section 4.2.

We discuss the formation mechanism of the GMFs in GMC-16. Early molecular gas surveys in the Galaxy speculated that this type of long filamentary structure is supposed to be formed by galactic spiral shocks (e.g., Burton & Shane 1970; Jackson et al. 2010). Alternative ideas have also been proposed; for example, the streamer in W51 can be regarded as a part of the expansion ring based on its velocity structure (Moon & Park 1998). Our extragalactic view of GMC-16 with a resolution similar to that of the Galactic single-dish studies allows us to address the origin of the GMFs. The morphology of the GMFs in GMC-16 shows a relatively straight shape, and the sizes of the H ii regions are small compared to those of the GMFs. Based on their morphologies, it is unlikely that the GMFs are part of the ring-like structure associated with the H ii regions.

As shown in Section 3.1, we find clear position discrepancies between the GMFs and the H ii regions. This type of spatial offset is sometimes seen in nearby grand design spiral galaxies (Hirota et al. 2011; Colombo et al. 2014), although the separations are more than a few ×10 pc, which is much larger than the present case. The authors suggest that high-mass star formation activities are triggered by the density-wave driven galactic spiral shocks. The sequence of the GMFs and H ii regions in GMC-16 is consistent with the propagation direction of the spiral arm. According to the steady-state density-wave theory (Lin & Shu 1964), the interstellar medium is decelerated and compressed at the bottom of the stellar potential, i.e., the shock front (Fujimoto 1968; Roberts 1969). If we adopt a rotation velocity of the interstellar medium, ∼80 km s⁻¹ at 2 kpc from the galactic center (Corbelli et al. 2014), and a pattern speed, Ω_p, of ∼25 km s⁻¹ kpc⁻¹ (Newton 1980), the timescale of the propagation of the H ii regions from the possible shock front traced by the CO filaments is calculated to be ∼1 Myr (=20 pc/ (80–50) km s⁻¹) assuming that the GMFs and the H ii regions are located at the same height in the galactic disk. This is consistent with the age of the H ii regions judged from their sizes based on the Galactic studies (Tremblin et al. 2014). This means that the two major H ii regions inside the observed field (Figure 2(b)) might be formed around the present locations of Filament A/B, although the galactic shock alone may not be enough to trigger the high-mass star formation (see Section 4.3).

We found a central velocity difference between the extended gas component (∼−240 km s⁻¹) at the eastern (upstream) side of GMC-16 and the filamentary structures (∼−245 km s⁻¹), as shown in Figure 3 and Section 3.1. Based on the ∼40 pc resolution measurements in CO and H i by Gratier et al. (2010), the velocity gradient along the north–south direction of the GMC is ≲0.1 km s⁻¹ pc⁻¹, which is mostly arising from the galactic rotation. The rotational motion alone may not simply explain the velocity offset (∼5 km s⁻¹) in the PV diagram (Figure 3(b)), which is extracted perpendicular to the large-scale gradient direction. In addition to this, the velocity width of the extended gas is larger than that of the filamentary clouds. These velocity features can be qualitatively explained by the dissipation of turbulent energy and the deceleration of the gas at the shock front. The idea that filamentary structure can be seen as a consequence of turbulent dissipation has been proposed in various scales of interstellar medium (e.g., Padoan et al. 2001; Inoue et al. 2018; Tokuda et al. 2018). In addition to this, our observations toward GMC-8, which is a quiescent GMC, show fewer filamentary structures and a large velocity dispersion compared to the GMFs in GMC-16 (forthcoming paper). In summary, our indirect evidence suggests that the galactic spiral shock compresses a turbulent diffuse cloud to form filamentary structures.

In flocculent spiral galaxies like M33, it may be hard to realize strong spiral shocks compared to grand design galaxies (Egusa et al. 2011, 2017; Hirota et al. 2011; Colombo et al. 2014). Wada et al. (2011) proposed that the complex spiral structures found in M33 are explained by the nonsteady stellar arms rather than the conventional density-wave picture. The galactic shear motion can elongate GMCs over an ∼100 pc scale (Wada 2008; Miyamoto et al. 2014). The nonsteady stellar arm model predicts that there is no clear spatial offset between the gas spiral arm and young stars across the galaxy. Further statistical studies to investigate the position discrepancies between molecular clouds and H ii regions based on observations toward similar targets in M33 with a similar angular resolution will allow us to draw a comprehensive picture of the relation between the galactic kinematics and GMF formation.

We found the redshifted and blueshifted wing features at MMS (Section 3.2). According to the Galactic studies of high-mass star-forming regions (e.g., Beuther et al. 2002), such high-velocity wings with a maximum velocity of ∼±20 km s⁻¹ or even much smaller velocities are normally interpreted as protostellar outflows. The typical size of outflow lobes in the Galaxy (e.g., Beuther et al. 2002) and the LMC (e.g., Fukui et al. 2019; Tokuda et al. 2019) is less than 1 pc. To demonstrate compactness of the possible outflow, we made the velocity profile maps around the MMS's location in ¹²CO and ¹³CO (Figure A1 in the Appendix) with a grid size of 093 × 061, which corresponds to ∼3.7 pc × ∼2.5 pc. We could not find similar high-velocity wing emission except in the middle panel, i.e., at the location of MMS. These observational results strongly suggest that the high-velocity ¹²CO emission at MMS is a protostellar outflow.

We briefly mention the limitations of the present observations. As seen in Figure 4, there is no significant spatial offset between the red and blue velocity components. Because protostellar outflows are as compact as ≲1 pc as mentioned in the previous paragraph, this feature is mostly due to the lack of spatial resolution. For example, a low-resolution study toward the Galactic high-mass protostellar object AFGL 2591 identified the molecular outflow without a clear offset between the red- and blueshifted components (Lada et al. 1984). Subsequently, Mitchell et al. (1991) clearly revealed its bipolar nature with a finer resolution measurement. A pole-on configuration is another possibility to interpret our data at MMS, in which case the accretion disk may be observable as face-on. However, the sizes of such disks associated with high-mass protostars are as small as ≲1000 au (e.g., Motogi et al. 2019), which is much smaller than outflow lobes, and thus it is also impossible to resolve it with this measurement. The present elongation of MMS in the north–south direction is arising from the combination of the beam swelling effect and its filamentary nature.

Although we need much higher angular resolution observations to reveal the actual distribution of the outflow and its age, the current data tell us that the dynamical time (=size/velocity) is as young as ∼(5–8) × 10⁴ yr depending on the assumption of the inclination angle (45°–60°). Considering the fact that the high-mass stars are frequently formed in binary or multiple manners (e.g., Massey 2003), MMS may be composed of unresolved multiple protostellar sources. Future high-resolution infrared and long-baseline ALMA observations will elucidate further details regarding the nature of the source.

In Section 4.1, we suggest that the galactic shock may be a plausible mechanism to explain the GMF formation. However, the high-mass star formation activity itself is localized within a few parsec scales rather than the entire ∼50 pc scale filament. Although Filament B has a ¹³CO peak at the edge of the filament (Figure 1), we could not detect any dense gas tracers, C¹⁸O, and 1.3 mm continuum so far. With respect to the H ii regions, which represent the recent star formation activities (Figure 2(b)), their distributions are discrete rather than continuous. These facts indicate that an additional factor may also be needed to trigger the high-mass star formation.

One possible mechanism to trigger the high-mass star formation in MMS is "collect and collapse" (Elmegreen & Lada 1977; Dale et al. 2007) driven by the expansion of the nearby H ii region (Figure 4(c)). However, the velocity gradient traced by the ¹³CO emission (Figure 4(b)) does not follow the direction of the expanding motion judged from the distribution of the H ii region. It is unlikely that the second generation star formation in MMS of GMC-16 was triggered by the expanding motion.

The moment 1 map shows a drastic velocity change at the boundary between the northern and eastern filaments as mentioned in Section 3.2. This type of velocity difference is often seen in high-mass star-forming filamentary clouds in the LMC at a subparsec resolution (Fukui et al. 2015; Saigo et al. 2017; Harada et al. 2019; Nayak et al. 2019), and its interpretation is that collision between two or several filaments triggers high-mass star formation. More recently, the follow-up observations toward some of the targets with a spatial resolution of ≲0.1 pc resolved further complex substructures composed of many filaments, which are difficult to explain by a coalescence process of individual components (Fukui et al. 2019; Tokuda et al. 2019). Nevertheless, because the orientation of filaments in the two different high-mass star-forming regions separated by over 50 pc is roughly aligned, they concluded that a tidally driven galactic scale colliding flow (see Fukui et al. 2017; Tsuge et al. 2019) induced the formation of the fan-shaped hub filaments as precursors of high-mass stars following the propagation direction of the flow. Such a hub filament is also reproduced by numerical simulations of cloud–cloud collision (Inoue et al. 2018).

In GMC-16, as shown in Figures 4(a)–(c), the hub-filamentary structure extends toward the eastern direction, which is considered to be the upstream side in the spiral shock. Although the large-scale spiral shock may form the GMFs with a length of ≳50 pc as discussed in the Section 4.1, the formation of the small-scale hub filament may not be explained by the spiral shock alone. An additional factor, such as a collision between the GMF and a preexisting small cloud, is needed to interpret the localization of the hub-filamentary dense clump leading to high-mass star formation. We note that a similar hub-like molecular complex is also found in NGC 604-GMC (K. Muraoka et al. 2020, in preparation). According to the galactic scale numerical simulations, cloud–cloud collisions frequently occur around the galactic potential where interstellar media are concentrated (e.g., Dobbs & Baba 2014). Sano et al. (2019) found a high-mass star-forming clump possibly formed by a collision between two clouds in the central region in M33, suggesting that cloud–cloud collision is not a rare event in M33 (see also, Tachihara et al. 2018; K. Muraoka et al. 2020, in preparation). The Galactic high-mass star-forming regions NGC 6334 and NGC 6357 (Persi & Tapia 2008) are also good counterparts to consider the high-mass star formation activities analogous to GMC-16. The regions contain several bright infrared sources with a total luminosity of >10⁵ L_⊙ along the molecular filament with a length of ∼100 pc. Fukui et al. (2018) concluded that a cloud–cloud collision promoted over a 100 pc scale mini-starburst in the NGC 6334 and NGC 6357 regions. Our GMC-16 studies may be equivalent to observations of such Galactic high-mass star-forming regions from outside the Galaxy.