ALMA Observations of Giant Molecular Clouds in M33. III. Spatially Resolved Features of the Star formation Inactive Million-solar-mass Cloud

Figure 2 shows the velocity-integrated intensity map (moment 0) of ¹²CO(J = 2–1) and ¹³CO(J = 2–1). One of the remarkable features in the ¹²CO map (panel (a)) is a widely extended and round-shaped structure with a diameter of ∼50 pc as indicated by the magenta dashed circle in the figure. We call this structure the "Main cloud" hereafter. It has a relatively sharp boundary at the western side, although it extends outward with several arm-like features in the other directions. The peak integrated intensity is ∼60 K km s⁻¹, corresponding to an H₂ column density of ∼2 × 10²² cm⁻² if we assume a constant ¹²CO(J = 2–1)/¹²CO(J = 1–0) ratio of ∼0.5, obtained from the single-dish data (Tosaki et al. 2011; Druard et al. 2014), and the X_CO factor of 4 × 10²⁰ cm⁻²(K km s⁻¹)⁻¹ (Druard et al. 2014). The total mass integrated within the observed field (hereafter ${M}_{{}^{12}\mathrm{CO}}$) is ∼4 × 10⁶ M_⊙, which is consistent with the previous studies (Druard et al. 2014).

Zoom In Zoom Out Reset image size

In contrast to ¹²CO, the ¹³CO moment 0 show a clumpy distribution throughout the observed field. As one can see, there are ∼20 spatially isolated components, and we call these objects "¹³CO clumps." The intensities of the ¹³CO clumps are ∼2–10 K km s⁻¹, which are at most one-tenth lower than that of the two active star-forming GMCs (Paper I for GMC-16, and Paper II for NGC 604 GMC). We selected five bright clumps whose peak ¹³CO intensity exceeds 5 K km s⁻¹, corresponding to the column density of ∼1.5 × 10²¹ cm⁻², as indicated in Figure 2(b). We derived the size of the ¹³CO clumps by 2D Gaussian fittings at each clump in the image plane within the regions at a more than 3σしぐま contour. Table 1 gives the physical properties of the ¹³CO clumps. The intensity and typical sizes are similar to Galactic low-mass star-forming regions, such as Taurus, Chamaeleon, and Aquila (Mizuno et al. 1995, 1999; Kawamura et al. 1999), and a Planck Cold Cloud in the LMC (Wong et al. 2017), traced by their ¹³CO observations with a similar angular resolution. To characterize these clumps, we estimated the H₂ column density assuming the local thermodynamic equilibrium with a uniform excitation temperature of ∼10 K and a relative [H₂]/[¹³CO] abundance of 1.4 × 10⁶ (see also Papers I, II). The mean column densities and masses (M_LTE) are ∼(3–5) × 10²¹ cm⁻² and ∼(2–10) × 10³ M_⊙.

Table 1. Properties in Each ¹³CO Clump

Source ID	R.A.	Decl.	Size	vcenter(1)	vFWHM(2)	${N}_{{{\rm{H}}}_{2}}^{\mathrm{mean}}$(3)	${N}_{{{\rm{H}}}_{2}}^{\mathrm{peak}}$(4)	Virial Mass(5)	MLTE(6)
	(J2000.0)	(J2000.0)	(pc)	(km s−1)	(km s−1)	(cm−2)	(cm−2)	(M⊙)	(M⊙)
clump-1	01 34 09.22	30 49 12.19	20 × 5.8	−247.7	5.7	3 × 1021	1 × 1022	4 × 104	1 × 104
clump-2	01 34 09.57	30 49 09.72	9.1 × 3.3	−247.8	5.1	4 × 1021	1 × 1022	2 × 104	4 × 103
clump-3	01 34 09.06	30 49 09.76	18 × 8.5	−242.4	4.7	3 × 1021	8 × 1021	3 × 104	5 × 103
clump-4	01 34 09.34	30 49 05.86	7.3 × 3.3	−257.3	4.4	4 × 1021	9 × 1021	1 × 104	2 × 103
clump-5	01 34 09.89	30 49 04.82	5.4 × 1.8	−251.7	2.3	5 × 1021	2 × 1022	2 × 103	3 × 103

Note. (1) Central velocity. (2) Velocity width of ¹³CO(J = 2–1) emission. (3) Mean of H₂ column density. (4) Peak of H₂ column density. (5) Virial mass calculated from the following formula, M_vir = 210ΔでるたV² R , where R is the geometric mean of the source size. (6) Total LTE mass within each clump.

Download table as:  ASCIITypeset image

We determined that the average spectra within the regions have significant emission above 3σしぐま in each of the selected ¹³CO clumps. Figure 3 shows the ¹²CO, ¹³CO, and C¹⁸O spectra. The subsequent Gaussian fitting to the ¹³CO spectra determined their central velocity (v_center ) and velocity width in FWHM (v_FWHM ). The v_center has a large variety from −257 to −242 km s⁻¹. Although the ¹³CO profiles have a Gaussian shape, some ¹²CO spectra show complex profiles, which likely have multiple velocity components. The averaged analysis allows us to search weak C¹⁸O emission at the clumps. At clump-3, we found significant C¹⁸O emission, whose peak velocity is similar to that of ¹³CO. The detection of C¹⁸O with more than 3σしぐま noise level of the average spectra implies the presence of dense molecular materials with the density of ∼10⁴ cm⁻³. At clump-5, we found marginal C¹⁸O emission and the highest ¹³CO integrated intensity in the observed field (see Figure 2). The (column) density peak presents at the region close to the GMC edge instead of an inner side of the Main cloud.

Zoom In Zoom Out Reset image size

The integrated total M_LTE in the observed field (hereafter ${M}_{{}^{13}\mathrm{CO}}$) is ∼6 × 10⁴ M_⊙. The mass ratio (${M}_{{}^{13}\mathrm{CO}}$/${M}_{{}^{12}\mathrm{CO}}$) is ∼2%, which is remarkably lower than that of the other star-forming GMCs in M33 (Table 2, see also Papers I, II) and the Galactic plane average of ∼24% (Torii et al. 2019). This result shows that the relatively dense gas fraction with respect to the total molecular gas is not high compared to the other targets, and consistent with nondetection of 1.3 mm and C¹⁸O emission tracing ≳10⁴ cm⁻³ dense gas at this resolution/sensitivity.

Table 2. The Comparison of Dense Gas Properties Among the Three GMC in M33

GMC Name	C18O Detection	1.3 mm Detection	${M}_{{}^{12}\mathrm{CO}}$(1)	${M}_{{}^{13}\mathrm{CO}}$(2)	${M}_{{}^{13}\mathrm{CO}}$/${M}_{{}^{12}\mathrm{CO}}$
			(M⊙)	(M⊙)
GMC-16	Y	Y	∼2 × 106	∼2 × 105	∼0.1
NGC 604 GMC	Y	Y	∼3 × 106	∼3 × 105	∼0.1
GMC-8	N(3)	N	∼4 × 106	∼6 × 104	∼0.02

Note. (1) Total molecular gas mass derived from the ¹²CO (J = 2–1) luminosity assuming the X_CO factor of 4 × 10²⁰ cm⁻²(K km s⁻¹)⁻¹. Based on the single-dish data, we applied ¹²CO (J = 2–1)/¹²CO (J = 1–0) ratios of 0.7 for GMC-16, 0.85 for NGC 604, and 0.5 for GMC-8. (2) LTE mass derived from the ¹³CO (J = 2–1) data. (3) Smoothing analysis toward some of the selected ¹³CO clumps shows significant C¹⁸O emission (see the text).

Download table as:  ASCIITypeset image

Figure 4 shows the velocity dispersion, σしぐま_v (moment 2), map and the peak brightness temperature map in ¹²CO. The GMC center shows a higher velocity dispersion than the outer parts. In the Main cloud, the FWHM line width of the average spectra is ∼11 km s⁻¹ derived by the Gaussian fitting. We will describe the velocity structure further in the next subsection (Section 3.2). The peak brightness temperature (panel (b)) ranges from a few to 8 K, suggesting that the excitation temperature is low (≲10 K). In contrast to this, the most active star-forming complex, NGC 604, shows much higher temperature of ∼40 K (Paper II), indicating that GMC-8 currently does not harbor any remarkable heating sources. Although the edge of the Main cloud has a higher brightness temperature in the field, these regions are apart from the faint 24 μみゅーm and Hαあるふぁ sources (Figure 2(b)). Note that the position accuracy of the two data sets, 3'' for 24 μみゅーm (Rieke et al. 2004) and ∼05 for Hαあるふぁ (see Hoopes & Walterbos 2000; Bailer-Jones et al. 2018), does indicate that they are outside of the Main cloud.

Zoom In Zoom Out Reset image size

The velocity channel maps of ¹²CO in Figure 5 show that the emission are continuously distributed over the velocity range of −268 to −236 km s⁻¹. Figure 6 shows the intensity-weighted velocity (moment1) map. Toward the Main cloud, panel (a) shows that there is no velocity jump larger than ∼5 km s⁻¹ (see also panel (b)). If there are more than two velocity components, the velocity maps show opposite results, i.e., a drastic velocity gap in moment 1 and a large dispersion in moment 2.

Zoom In Zoom Out Reset image size

Zoom In Zoom Out Reset image size

We made the ¹²CO position–velocity (PV) diagram in the horizontal direction along the Main cloud to explore the velocity structure further (Figure 6(b)). Although we see some velocity fluctuations, the Main cloud shows an almost single velocity component with the velocity width of ∼10–20 km s⁻¹. We thus conclude that the Main cloud is almost composed of a single velocity component. However, at the western and eastern side of the Main cloud, there are notable blueshifted components whose velocity is ∼−260 km s⁻¹, as appeared in Figure 6(b). We confirmed large differences at most ∼15 km s⁻¹ among the centroid velocities of ¹³CO clumps (Section 3.1). The ¹³CO contours on the PV diagram in Figure 6(b) demonstrate that clump-4 and -5 are at the edge of the ¹²CO emission rather than centrally concentrated along with the ¹²CO intense velocity range. We discuss these velocity features and the possibility that more than two velocity gas streams develop in the massive cloud in Section 4.2.

As mentioned in the 1, GMC-8 is one of the most massive GMCs in M33. The GMC belongs to a series of extremely massive clouds in the Local group of galaxies at least in terms of its total H₂ mass. This section summarizes previous studies of massive GMCs in the MW, LMC, and M33 from the literature, and then compares their physical properties and those of GMC-8. Table 3 shows the characteristics of massive GMCs based on the CO or its isotopologue observations with an angular resolution of 1–40 pc and a velocity resolution of 0.1–0.6 km s⁻¹. The massive GMCs, whose molecular gas mass exceeds ∼10⁶ M_⊙ in the MW, actively form high-mass stars almost exclusively. The Maddalena's cloud and Cygnus OB7 are rare clouds as quiescent GMCs, which are hard to find within ∼1 kpc from the Sun.

The sizes and velocity widths do not differ greatly among these GMCs except for a relatively narrow line width in Cygnus OB7. We found a marginal indication of a higher line width in GMC-8. If the observational bias artificially produces such a feature, this effect should be more pronounced in the MW targets due to its edge-on view. Therefore, we speculate that GMC-8 inherently has a large line width, which means that a high-level turbulent motion dominates the cloud or there is a superposition of multiple velocity components. For example, spatially/velocity resolved observations of the MW/LMC/M33 massive clouds obtained substantial evidence of cloud–cloud collisions (W43, Kohno et al. 2021; W49, Miyawaki et al. 2009; W51, Fujita et al. 2021a; Carina, Fujita et al. 2021b; LMC-N159, Fukui et al. 2019; Tokuda et al. 2019; GMC-37, Sano et al. 2021; NGC 604, Paper II; GMC-16, Paper I) and the authors suggest that such drastic events triggered high-mass star formation therein. Although Dobashi et al. (2014) reported colliding filaments with a few km s⁻¹ velocity difference in Cygnus OB7 at the dense cloud center, such local phenomena (i.e., filament collisions) does not largely change the GMC scale velocity width as a whole. Star formation active GMC tends to show highly developed structures composed of many spatially steep features, such as shells and filaments, while an extended diffuse component dominates less active clouds (see also a quantitative analysis in the LMC; Sawada et al. 2018). The extended gas feature of GMC-8 is qualitatively different from filamentary structures discovered in the other two GMCs in M33 (Papers I, II). The low-density nature and less-developed structure are consistent with those in the other star formation inactive clouds, but the velocity width is comparable to active star-forming regions in the Local group. We discuss the velocity structure and the formation origin of GMC-8 in Section 4.2.

Previous numerical simulations investigate molecular cloud formation and its mass supply by atomic gas (H i) supersonic flow (e.g., Hennebelle & Péraut 1999; Koyama & Inutsuka 2000, 2002; Vázquez-Semadeni et al. 2007; Inoue & Inutsuka 2012). However, some theoretical works (e.g., Kobayashi & Inutsuka 2017; Kobayashi et al. 2018) suggest that it takes ∼100 Myr to gain the H₂ mass up to ∼10⁶ M_⊙ in the case of recursive accretion flows from multiple directions with a density of ∼1 cm⁻³ (see Koda et al. 2009, for estimations considering a spherical configuration of the mass accretion, which also suggest a similar timescale of 100 Myr). Such a long timescale is significantly larger than the typical GMC lifetime, a few ×10 Myr (Fukui et al. 1999; Kawamura et al. 2009). Fukui et al. (2009) observationally derived an H i gas accretion rate onto GMCs of ∼0.05 M_⊙ yr⁻¹ by considering the three-dimensional relation between the H i envelope and molecular clouds in the LMC. In this case, although it is possible to grow up to ∼10⁶ M_⊙ within the GMC lifetime, the star formation inactive GMC is supposed to evolve into the subsequent state (i.e., inhering bright H ii regions) in a much shorter time, ∼6 Myr (Kawamura et al. 2009; Corbelli et al. 2017). Therefore, in order to obtain a large amount of molecular gas in a shorter time, it is necessary to supply materials by a drastic phenomenon rather than a steady-state accretion condition.

We revisit the PV diagram (Figure 6) to investigate whether or not the velocity structure preserves the remnants of a gas convergence. As we show in Section 3.2, the GMC is composed of an almost single velocity component with some local fluctuations. Among these features, clump-4 is relatively blueshifted in the Main cloud (see panel (b)). This characteristic is similar to a synthetic PV diagram, which is called as "V-shaped" feature, produced by cloud–cloud collision (Takahira et al. 2014; Haworth et al. 2015; Fukui et al. 2018). Figure 7 shows the blueshifted and redshifted CO components with complementary distributions with each other, which is also one of the pieces of evidence for gas collision (Fukui et al. 2021). Based on these results, we set up a hypothesis that two molecular clouds with a relative velocity difference of ∼10 km s⁻¹ (see the next paragraph) collided with each other, and the event realizes the total molecular mass of ∼10⁶ M_⊙ in this region.

Zoom In Zoom Out Reset image size

Notably, there are two independent blueshifted (∼−260 km s⁻¹) clouds at the east and west sides of the Main cloud (see Figure 6(b)). If the blueshifted colliding counterpart was larger than the GMC seed, the extended parts of blueshifted clouds could not interact with the Main cloud. We may witness the present blueshifted components as two remnant clouds that are not decelerated by the collision (see also Section 4.3). The average column density of the Main cloud and blueshifted clouds are ∼3 × 10²² cm⁻² and ∼1 × 10²² cm⁻², respectively. If the column density ratio represents the mass ratio between the current product (i.e., the Main Cloud) and the GMC seed before the collision, the extra blueshifted component would contribute at least 20%–30% as the mass reservoir to produce the million-solar-mass GMC.

Even though it is not easy to precisely estimate the collision's timescale, the expected duration is on the order of a few Myr (=a few ×10 pc/∼10 km s⁻¹), which is derived from the GMC size divided by the relative velocity. This is consistent with the statistically derived timescale of the star formation inactive phase without remarkable H ii regions. According to the compilation by Corbelli et al. (2017), the total number of such GMCs in M33 is 178 (=Type A + B clouds in their catalog). The fact that there are no other clouds with a comparable mass to GMC-8 at the same stage indicates that its lifetime is very short if we assume the ratio of the number of objects at each stage is proportional to that of the timescale. Possible fate of GMC-8 is to evolve into a less massive GMC immediately by (i) gas consumption/destruction through star formation and/or (ii) dissipating the molecular gas due to turbulence without star formation. The presence of the 24 μみゅーm source at the vicinity of clump-5 indicates that an embedded star formation is already on-going. Meanwhile, as shown in Table 1, the virial mass of the ¹³CO clumps is larger than their M_LTE except for clump-5, indicating that most of the dense materials are not self-gravitating. Despite the presence of the collisional signatures in this system, the inactive nature of high-mass star formation can be explained by the lower column density of the molecular gas that can produce at least one single O-star. According to the recent statistical investigations of colliding clouds among more than 50 samples by Enokiya et al. (2021), at least a single O-star formation requests higher H₂ column density more ∼10²² cm⁻², which is mostly higher than those of the ¹³CO clumps in GMC-8. Recent ALMA observations toward the interacting galaxies system, NGC 4567/4568, also found a star formation inactive molecular layer with a colliding velocity feature (Kaneko et al. 2018). In summary, the combination of the local star formation activity and gas dissipation due to turbulent motion may be going to reduce the molecular gas of this system in the near future.

We subsequently discuss the origin of the colliding gas whose relative velocity is ∼10 km s⁻¹ against the Main cloud. Supernova remnant searches in M33 (e.g., Garofali et al. 2017; Long et al. 2018) did not find any candidates within ∼300 pc around GMC-8, which means that supernova feedback is unlikely to be the origin of the high-velocity colliding flow. The location of the GMC is in the interarm region (see Section 1), and thus the galactic shock also cannot be appropriate, as suggested in our GMC-16 study (Paper I) whose location is near the stellar spiral arm.

We here investigate a large-scale H i gas distribution obtained with the Very Large Array (VLA) observations (Gratier et al. 2010). We used a velocity cube that we corrected to the the galactic rotation, and the shifted velocity frame is V_shift (see Tachihara et al. 2018). We overlay a blueshifted velocity H i component (V_shift = −214 to −184 km s⁻¹), which is the same as the NGC 604 study defined by Tachihara et al. (2018), on the CO gas (Figure 8). This velocity is consistent with that in the blueshifted (−260 km s⁻¹) CO gas (Figure 6) in the standard frame. The colliding H i gas is likely converted into a molecular state at the GMC position as a whole. Our approximate estimation of the blueshifted atomic gas mass at both the northern/southern sides of GMC-8 is ∼10⁶ M_⊙ within the same area as GMC-8 traced by the single-dish CO observation assuming the optically thin limit of the H i emission. If we assume such a large amount of the atomic gas as an initial condition of the colliding H i flow onto a GMC-8 seed, a certain fraction of hydrogen gas may be converted into H₂ during the collision. Figure 9 summarizes our suggestion for the mass growth process of GMC-8. There were a less massive GMC seed along the interarm region and a blueshifted H i gas component possibly containing a dense layer before the collision (panels (a) and (b)). Panels (c) and (d) mimic observed gas distribution in X–Y–Z and the PV diagram. The deceleration by the collision of the two clouds explains the lack of blueshifted components toward the Main cloud (see Section 4.2). The present circumstance around GMC-8 suggests that the large-scale (≳50 pc) H i flow is the promising phenomena to increase GMC mass up to ∼10⁶ M_⊙. This is consistent with the theoretical prediction that a dynamic mass accumulation is required to form such massive GMC rather than local converging flows (see the 1).

Zoom In Zoom Out Reset image size

Zoom In Zoom Out Reset image size

We finally remark on the origin of the H i flow. Tachihara et al. (2018) pointed out that a tidal interaction between M31 and M33 drove external H i flows onto the M33 disk based on the presence of a bridge feature connecting the two galaxies (Braun & Thilker 2004; Putman et al. 2009; Lockman et al. 2012). Grossi et al. (2008) found high-velocity H i clouds, which are supposed to be tidal debris, and suggested they are supply sources for high-mass star formation. Alternatively, we cannot exclude an internal origin of such high-velocity diffuse gas. Wada et al. (2011) proposed that the dynamic spiral theory, which involves nonsteady stellar arms (Dobbs & Baba 2014 and references therein), explain the presence of complex gas structures in flocculent spiral galaxies like M33. In this case, we predict that there are gas flows from multiple directions onto molecular clouds. There are some high-velocity components apart from the galactic rotation other than GMC-8 and NGC 604 (Figure 8). Though we cannot determine which external or internal origins dominate as such gas supply phenomena, further investigating the relation between H i and CO will give a better understanding of GMC evolution during the galactic lifecycle.