Kinematics and star formation toward W33: a central hub as a hub--filament system
Xiao-Lan Liu, Jin-Long Xu, Jun-Jie Wang, Nai-Ping Yu, Chuan-Peng Zhang, Nan Li, Guo-Yin Zhang
AAstronomy & Astrophysics manuscript no. 35035corr © ESO 2021February 23, 2021
Kinematics and star formation toward W33: a central hub as ahub–filament system
Xiao-Lan Liu , , Jin-Long Xu , , Jun-Jie Wang , , Nai-Ping Yu , , Chuan-Peng Zhang , , Nan Li , , Guo-YinZhang , National Astronomical Observatories, Chinese Academy of Sciences, Beijing 100101, Chinae-mail: [email protected] CAS Key Laboratory of FAST, National Astronomical Observatories, Chinese Academy of Sciences, Beijing 100101, ChinaReceived / Accepted
ABSTRACT
Aims.
We investigate the gas kinematics and physical properties toward the W33 complex and its surrounding filaments. We studyclump formation and star formation in a hub–filament system.
Methods.
We performed a large-scale mapping observation toward the W33 complex and its surroundings, covering an area of1 . ◦ × . ◦ , in CO (1-0), CO (1-0), and C O (1-0) lines from the Purple Mountain Observatory (PMO). Infrared archival datawere obtained from the Galactic Legacy Infrared Mid-Plane Survey Extraordinaire (GLIMPSE), the Multi-band Imaging PhotometerSurvey of the Galaxy (MIPSGAL), and the Herschel Infrared Galactic Plane Survey (Hi-GAL). We distinguished the dense clumpsfrom the ATLASGAL survey. We used the GLIMPSE I catalogue to extract young stellar objects.
Results.
We found a new hub–filament system ranging from 30 to 38.5 km s − located at the W33 complex. Three supercriticalfilaments are directly converging into the central hub W33. Velocity gradients are detected along the filaments and the accretion ratesare in order of 10 − M (cid:12) yr − . The central hub W33 has a total mass of ∼ . × M (cid:12) , accounting for ∼
60% of the mass of the hub–filament system. This indicates that the central hub is the mass reservoir of the hub-filament system. Furthermore, 49 ATLASGALclumps are associated with the hub–filament system. We find 57% of the clumps to be situated in the central hub W33 and clusteredat the intersections between the filaments and the W33 complex. Moreover, the distribution of Class I young stellar objects (YSOs)forms a structure resembling the hub–filament system and peaks at where the clumps group; it seems to suggest that the mechanismsof clump formation and star formation in this region are correlated.
Conclusions.
Gas flows along the filaments are likely to feed the materials into the intersections and lead to the clustering andformation of the clumps in the hub–filament system W33. The star formation in the intersections between the filaments and the W33complex might be triggered by the motion of gas converging into the intersections.
Key words. stars: formation - stars: kinematics and dynamics - stars: massive - (ISM:) HII regions - ISM: lines and bands - ISM:Molecules
1. Introduction
High-mass stars ( > (cid:12) ) play a vital role in the evolution ofinterstellar medium (ISM) and their host galaxies. They can en-rich the chemical components in the ISM by throwing heavymetal elements into the ISM and therefore promote the chem-ical evolution of galaxies. However, the formation mechanismof high-mass stars remains unknown. Recently, filaments havefound to be prevalent as the main sites of star formation (e.g. My-ers 2009; André et al. 2010, 2014; Wang et al. 2015). Particularlyin the hub–filament systems, active high-mass star formation isfrequently reported in hubs, which host a number of clumps indi ff erent phases (e.g. Schneider et al. 2012; Peretto et al. 2013;Hennemann et al. 2014; Friesen et al. 2016; Yuan et al. 2018).In addition, theoretical studies suggest that filaments in the hub–filament systems may act as tributaries, converging mass intothe protocluster clumps in the central hub and therefore result-ing in vigorous star formation there (Myers 2009; Liu et al. 2012;Gómez & Vázquez-Semadeni 2014). Therefore, identifying andinvestigating the hub–filament systems can help us to understandthe formation of clumps and massive stars.The W33 complex is a massive star-forming region locatedat l ∼ ◦ . b ∼ − . ◦ in the Galactic plane. The parallactic distance of the W33 complex is 2 . + . − . kpc (Immer et al. 2013).The W33 complex contains six massive dust clumps (W33 Main,W33A, W33B, W33 Main1, W33A1, and W33B1), which wereidentified by Immer et al. (2014) using the APEX TelescopeLarge Area Survey of the Galaxy (ATLASGAL) (Schuller et al.2009). The calculations toward these clumps suggest a total massof ∼ (0 . − × M (cid:12) and an integrated IR luminosity of ∼ × L (cid:12) (Immer et al. 2014; Kohno et al. 2018). More-over, molecular line observations toward the W33 complex de-tect a complex velocity field (e.g. Gardner & Whiteoak 1972;Goldsmith & Mao 1983; Immer et al. 2013; Kohno et al. 2018).As for W33A and W33 Main, emission and absorption peakswere observed at a radial velocity of ∼
35 km s − , while thespectra toward W33B were seen to peak at ∼
60 km s − (Gard-ner & Whiteoak 1972; Goldsmith & Mao 1983). Immer et al.(2013) proved that these two velocity components are locatedat the same parallaxial distance using the water maser emission.Apart from the 35 km s − and 58 km s − velocity componentsassociated with the W33 complex, Kohno et al. (2018) detectedand identified another velocity component at 45 km s − using theCO data from the NANTEN2 and Nobeyama 45 m telescopes.Furthermore, the 21 cm absorption line data suggest that this ve- Article number, page 1 of 21 a r X i v : . [ a s t r o - ph . GA ] F e b & A proofs: manuscript no. 35035corr locity component is likely partly associated with the W33 com-plex (Kohno et al. 2018). Kohno et al. (2018) propose that thecloud–cloud collision scenario between the 35 km s − and 58km s − clouds can explain the observed properties.Visualisation of the W33 complex on a large scale with thethree-colour composite of 8, 24, and 70 µ m in Fig. 1 suggeststhat the W33 complex is similar to a central hub surrounded bya set of infrared-dark filaments. In order to highlight the kine-matics in the W33 complex and the e ff ects that the surround-ings have on the W33 complex, we performed a large-scale COmolecular line observation towards the W33 complex. In addi-tion, the archival infrared data are adopted to probe star forma-tion activities. In Section 2, we describe the data sets, and insection 3 the results. We present our analysis and discussions inSection 4. Finally, we summarise our main results in Section 5.
2. Observations and data reduction
The mapping observations were made towards the W33 complexand its adjacent regions in CO (1-0), CO (1-0), and C O (1-0) lines (the rest frequencies of 115.10, 110.20, and 109.78 GHz)using the PMO 13.7 m radio telescope at De Ling Ha in westernChina at an altitude of 3200 m during October 2017. The totalmapping extent was approximately 80 (cid:48) × (cid:48) for all the lines. Thehalf-power beam width (HPBW) at 115 GHz is ∼ (cid:48)(cid:48) . Thenew nine-beam array receiver system in single-sideband mode(SSB) was used as a front end. Fast Fourier transform spectrom-eters were used as a back end with a total bandwidth of 1 GHzand 16384 channels. CO (1-0) was observed at upper sidebandwith a system noise temperature ( T sys ) in range of 150 −
300 K,while CO (1-0) and C O (1-0) were observed simultaneouslyat lower sideband. The velocity resolution for CO (1-0) is ∼ − and for CO (1-0) / C O (1-0) is ∼ − . Thepointing accuracy of the telescope was better than 4 (cid:48)(cid:48) . The phasecentre of the observations was ( l , b ) ∼ (13 ◦ .222, -0 ◦ .219) withthe position-switch on-the-fly (OTF) mode. The o ff -position was( l , b ) ∼ (12 ◦ .375, -2 ◦ .124), devoid of CO emission. The standardchopper wheel calibration technique was used to measure theantenna temperature T ∗ A corrected for the atmospheric absorp-tion. The final data were recorded in a brightness temperaturescale of T mb (K). The data were reduced and regridded using thesoftware GILDAS (Pety 2005). The pixel sizes of the CO FITScubes were 30 (cid:48)(cid:48) × (cid:48)(cid:48) . The CO (1-0) data are merely used toestimate the excitation temperatures (see Appendix A), becausethe velocity components in the CO (1-0) spectra cannot be dis-entangled.
The Galactic Legacy Infrared Mid-Plane Survey Extraordinaire(GLIMPSE) observed the Galactic plane (10 ◦ < | l | < ◦ , | b | < ◦ ) with the Infrared Array Camera (IRAC) instrument thatcan obtain simultaneous broadband images at 3.6, 4.5, 5.8, and 8 µ m (Fazio et al. 2004). We downloaded the mid-infrared (MIR)image at 8 µ m of this region from the Spitzer
GLIMPSE (Ben-jamin et al. 2003). The archived angular resolutions of IRAC 3.6,4.5, 5.8, and 8.0 µ m are 1 . (cid:48)(cid:48) , 1 . (cid:48)(cid:48) , 1 . (cid:48)(cid:48) , and 2 . (cid:48)(cid:48) , respectively.In addition, we also extracted the MIPSGAL 24 µ m image cov-ering the same region with a resolution of ∼ (cid:48)(cid:48) (Carey et al. http: // english.dlh.pmo.cas.cn / fs / Csengeri et al. (2014) extracted ∼ σ from the maps of the AT-LASGAL Survey (Schuller et al. 2009). The ATLASGAL Sur-vey was conducted by the Large APEX Bolometer Camera(LABOCA) at the 12m diameter APEX telescope at 870 µ m(Güsten et al. 2006a,b), dominated by the cold dust emissionfrom dense clumps. It was the first systematic survey of the in-ner Galactic plane. It has a beam size of ∼ (cid:48)(cid:48) . − ◦ < l < ◦ .
3. Results
To fit the spectrum of each pixel in the CO (1-0) and the C O(1-0) fits cubes, we use Behind the Spectrum (BST) (Clarkeet al. 2018), which is a fully automated routine. A detailed de-scription of BST is given by Clarke et al. (2018). For each com-ponent, the BST returns the amplitude (i.e. peak intensity), cen-troid velocity ( υ c ), velocity dispersion ( σ c ), and the reduced χ values of each spectrum. For accuracy, we check all the fits, andmanually remake the Gaussian fits to each spectrum. If it is fittednot well, we will exclude it. And the returned amplitudes haveto be over 1 K (5 σ ) for C O (1-0) and 3 K (6 σ ) for CO (1-0). The velocity dispersions σ c have to be more than 0.51 kms − (3 channels). Moreover any two fits from the same spectrumhave a υ c separation greater than 0.51 km s − . Figure 2 showsthe returned υ c histograms of the C O (1-0) and CO (1-0) fits,respectively. From the υ c distributions, we identify six velocitycomponents along the line of sight toward the W33 complex.These are in the ranges of 2-15 km s − , 15-21 km s − , 21-28km s − , 30-44 km s − , 44-48 km s − , and 48-60 km s − , markedwith the black dashed boxes in Fig. 2. In addition, we estimatethe excitation temperatures and the optical depths for the fits oftwo lines via equations A.1-A.2. The H column densities for theC O fits are also calculated using equation A.3. The distributiondiagrams of these parameters are shown in Figs. A.1-A.5. Fromthe figures, we can see that the CO lines are really opticallythick in the dense regions with N(H ) (cid:38) cm − . Here, we usethe optically thin C O (1-0) line to determine which velocitycomponent is associated with the W33 complex.Based on the above six velocity ranges, we make the C O(1-0) integrated intensity maps overlaid on the
Spitzer µ memission, as shown in Fig. 3. The C O (1-0) emission revealscompletely di ff erent structures for the six velocity components.Also, from the morphologies of the C O (1-0) emission, we findthat the velocity components of 30-44 km s − and 48-60 km s − seem associated with the W33 complex and W33B, respectively.This result is consistent with those of Immer et al. (2013) andKohno et al. (2018). On a larger scale, our CO velocity compo-nents reveal a number of other structures besides the W33 com-plex. In particular, the 30-44 km s − velocity component in Fig.3d is closely associated with a set of infrared dark filaments,referred to here as f1-f5, which seem to surround the W33 com-plex. Below, we further use the position-position-velocity (PPV) https: // github.com / SeamusClarke / BTSArticle number, page 2 of 21iu et al.: CO OBSERVATION TOWARD W33 cubes to discern the velocity structures associated with the W33complex in the 30-44 km s − range. To highlight the velocity structures in 30-44 km s − , we makethe C O (1-0) PPV space using the BST obtained fits cube, asshown in Fig. 4. According to the values of the centroid veloc-ities υ c , we mark the colour for each point. Meanwhile, the 3Dprojections of all the points are plotted on each of the axes. Oneach projection plane in Fig. 4, we identified three small velocityranges, namely 30-38.5 km s − , 38.5-42 km s − , and 42-44 kms − . Figure 5 shows the centroid velocity distribution diagramsfor the above three velocity ranges. From Fig. 5, we find thatthe C O (1-0) emission presents di ff erent structures in the threeranges. The structure of the 38.5-42 km s − velocity componentin Fig. 5b mainly presents an east-to-west large-scale filamentcontaining filaments f4 and f5. From Fig. 5c, we see that thevelocity component of 42-44 km s − is mainly coincident withsome isolated structures. While the velocity component of 30-38.5 km s − is closely associated with the W33 complex and fil-aments f1-f3; below we focus on this component. In Fig. 5a, anellipse represents the W33 complex. To avoid confusion, we re-name the filaments adjacent to the W33 complex as hf1, hf2, andhf3. The three filaments traced by the black lines gather in thedirection of the W33 complex. Similar to the structure of NGC2264 (Kumar et al. 2020), the 30-38.5 km s − velocity compo-nent appears to show a hub–filament system, whose central hubis the W33 complex. Furthermore, we estimate the physical parameters of the hub–filament system, which are all listed in Table 1. From column12 of Table 1, we see that the central hub W33 has an equiv-alent radius of 6.0 pc. The filament hf2 is the longest with alength up to 15.7 pc, while hf1 is the shortest with a length of7.5 pc. The widths of the filaments hf1, hf2, and hf3 are 4.6 pc,3.8 pc, and 3.0 pc, respectively. Considering the sensitivity lim-its of C O (1-0) (Amplitude > σ ) as well as the inclinations,the measured lengths and widths of the filaments should be alower limit. Column 4 of Table 1 presents the velocity rangesspanned by these structures. The minimum velocity interval is ∼ − in the filament hf3, further demonstrating the com-plicated velocity structures in the whole hub–filament system. Inaddition, column 7 in Table 1 shows the minimum ratios of thenon-thermal velocity dispersions to the sound speeds ( σ NT / c s ) inthese structures. We can see that all the ratios σ NT / c s are greaterthan 1, suggesting that the whole hub–filament system is super-sonic, probably dominated by the turbulence (Liu et al. 2018).The hub–filament system has a mean excitation temperature of ∼ . ∼ . col-umn density in magnitude of 10 cm − and a mean H numberdensity on the order of 10 cm − . In addition, the W33 complexhas a total mass of 1 . × M (cid:12) , which is consistent with pre-vious findings (e.g. Immer et al. 2014) and accounts for 60%of the mass of the hub–filament system. The masses of the fila-ments hf1, hf2, and hf3 are in the range of (2 . − . × M (cid:12) . These three filaments account for ∼
35% of the mass of the hub–filament system, illustrating that the mass of the hub–filamentsystem is mainly concentrated in the central hub W33. µ m clumps in the hub–filament system The ATLASGAL 870 µ m emission traces the distribution ofcold dust (Beuther et al. 2012). From the catalogue of the AT-LASGAL 870 µ m clumps (Urquhart et al. 2018), we extract49 clumps associated with the hub–filament system based ontheir V LSR and distances. The physical parameters of these 49clumps are presented in Table 2. All the clumps are found at adistance of 2.6 kpc, which is consistent with the parallax dis-tance of 2 . + . − . kpc (Immer et al. 2013). In addition, the distri-bution of the clumps with the same distance in the hub–filamentsystem provides further evidence that the W33 complex and thesurrounding filaments constitute a complete system. Also, their V LSR are in the range of 32 . − . − . We overlay theclumps on the C O (1-0) centroid velocity distribution map inFig. 5a, which reveals that 28 of them are located in the cen-tral hub W33. The other clumps are mainly distributed along thespines of the filaments hf3 (10) and hf2 (6). We also made his-tograms of radius, dust temperature, mass, and H column den-sity of the clumps in the W33 complex and in the filaments; seeFig. 6. We find that the clumps are warmer and denser with largersizes in the W33 complex, while the mean values of the mass arethe same for the clumps in the W33 complex and in the filament.Furthermore, Urquhart et al. (2018) classified ∼ µ m clumps into four evolutionary sequences: qui-escent (70 µ m weak), protostellar (MIR-dark but FIR-bright),young stellar objects (YSOs; MIR-bright), and MSF (associatedwith massive star formation tracers, such as radio-bright HII re-gions and methanol masers). From column 13 of Table 2, wesee that the extracted 49 clumps consist of 12 ( ∼ ∼ ∼ ∼ ∼
64% of the protostellar clumps are in the filaments. Thepercentages of the quiescent clumps are equal in the central huband the filaments.To investigate the capability of the clumps in the hub–filament system to form massive stars, we consider the relation-ship between mass and size, as shown in Fig. 7. The yellowshaded region represents a parameter space devoid of massivestar formation, as determined by Kau ff mann & Pillai (2010), of M ( r ) ≥ (cid:12) ( r / pc) . . From Fig. 7, we find that ∼
46% of theclumps in the central hub W33 and ∼
57% of the clumps in thefilaments lie above the threshold, suggesting significant poten-tial to form massive stars. Therefore, the hub–filament system islikely to be a suitable environment to form massive stars.
To study the star formation activity in the hub–filament system,we search for YSOs using the GLIMPSE I Spring’07 catalogue.In total, 47787 near-infrared (NIR) sources with 3.6, 4.5, 5.8, and8.0 µ m are selected in the observation zoom. The IRAC [5.8]- Article number, page 3 of 21 & A proofs: manuscript no. 35035corr [8.0] versus [3.6]-[4.5] colour-colour (CC) diagram is a use-ful tool for identifying YSOs with infrared excess (Allen et al.2004). Based on the criteria of Allen et al. (2004), these NIRsources are classified into three evolutionary stages, as shown inFig. 8. Here, 811 NIR sources are identified as Class I sources,which are protostars with circumstellar envelopes and have anage of ∼ yr, and 1252 sources are Class II sources, which aredisc-dominated objects with an age of ∼ yr. The remainingNIR sources are other sources (such as classical T Tauri, HerbigAe / Be). Here, Class I and Class II sources are selected as theYSOs.As the identified Class II YSOs are almost uniformly dis-tributed in the hub–filament system, we only show the distribu-tion of all the identified Class I sources overlaid on the C O(1-0) emission map in Fig. 9a. We note that Class I sources arepreferentially situated in the regions where clumps are clusteredin the hub–filament system. The clustering of YSOs in a givenarea can help us identify the active star-forming regions. To high-light the clustering behaviour of YSOs within the hub–filamentsystem, we analysed the surface density distribution of the iden-tified Class I YSOs in Fig. 9b. The lowest level of the contoursis ∼ − to get rid of the e ff ect of the foreground and back-ground objects. In Fig. 9b, we find that the surface density dis-tribution of the Class I YSOs resembles the structure of the hub–filament system. Moreover, the peaks of the Class I YSO densitydistribution are located where the clumps group, which appearsto suggest that the mechanism of clump and YSO formation inthe hub–filament system might be the same.
4. Discussion
In order to explore the clump and YSO formation in the hub–filament system, we need to analyse the stability of the cen-tral hub W33, which can be calculated using the expression α vir = σ v2 R / GM , where R is the equivalent radius of the cen-tral hub W33, and σ V = (cid:113) σ + c , the mean velocity dis-persion. The derived α vir from the optically thin C O (1-0)line is ∼ .
04. When α vir is less than 1, the central hub W33is likely gravitationally bound and perhaps collapsing. At thesame time, for the filaments hf1, hf2, and hf3, the di ff erencebetween the critical mass to length ratio ( M / L ) crit and the lin-ear mass M / L can tell us the stability of the filaments (Jacksonet al. 2010). When M / L > ( M / L ) crit , the filaments are dominatedby gravity and may be collapsing. The ( M / L ) crit can be derivedfrom ( M / L ) crit = σ v2 / G , which is mainly caused by turbulence(Jackson et al. 2010), and the M / L are calculated using the totalmasses M and lengths L of the filaments hf1, hf2, and hf3. Thederived M / L are the upper limits due to inclination and projec-tion e ff ects. Our computed results are listed in columns 14-15 ofTable 1. They indicate that filaments hf1, hf2, and hf3 are prob-ably collapsing globally.Figure 10 shows the di ff erence between the velocities of thefilaments and the junctions as a function of distance to junction.The points in Fig. 10 are extracted along the spine of each fil-ament. From Fig. 10, we can see that the filament hf1 shows atransition on the tail within the last 2 pc, but this might be inac-curate because of possible confusion with other clouds. Indeed,we detect a ∼
39 km s − filament there (see Sect. 3.2.1 and Fig.5b). If ignoring the decreasing trend of hf1, the filaments hf1,hf2, and hf3 all present monotonically increasing profiles. Theresultant velocity gradients are consistent with the ones reportedin other collapsing filaments (Peretto et al. 2014; Liu et al. 2016;Yuan et al. 2018). The velocity gradients for these filaments are estimated to be 0.32, 0.11, and 0.10 km s − pc − , respectively.The values are comparable to those detected in previous studies(Peretto et al. 2014; Yuan et al. 2018).Both cloud rotation and accretion flows along the filamentcan result in this kind of velocity gradient (Veena et al. 2018).To examine the rotational feature in the hub–filament system,we constructed position–velocity (PV) diagrams of C O (1-0)along the cuttings in Fig. 3e, as shown in Fig. 11. The whitedashed lines mark the location of the W33 complex. The PVplots in Fig. 11 reveal the velocity gradients between the fila-ments and the W33 complex as well as along the filaments, butno signs of a Keplerian rotation signature are observed. There-fore, we propose that the hub–filament system is not rotating.On the other hand, accretion flow along the filament couldalso be causing the smoothed velocity gradient (Kirk et al. 2013,and reference therein). Following Kirk et al. (2013), we derivedthe gas accretion rate ˙ M for a simple cylindrical cloud,˙ M = ∇ V M tan( α ) , (1)in which α is assumed as an inclination angle of 45 ◦ withoutconsidering the projection e ff ect (Yuan et al. 2018; Veena et al.2018). The accretion rates of the filaments are calculated as8 . × − , 5 . × − , and 3 . × − M (cid:12) yr − for the fila-ments hf1, hf2, and hf3, respectively, while residual e ff ects fromthe velocity coherence widely detected in Giant Molecular Fil-aments (GMFs) cannot be ruled out (Ragan et al. 2014; Wanget al. 2015). From Fig. 5a, we propose that the central hub W33is probably being fed by the filaments hf1, hf2, and hf3 directly.The total accretion rate is ∼ . × − M (cid:12) yr − , two orders ofmagnitude greater than rates found by other studies (Yuan et al.2018; Chen et al. 2019; Treviño-Morales et al. 2019). Given thatour velocity gradients are comparable to those in other stud-ies (Yuan et al. 2018; Chen et al. 2019; Treviño-Morales et al.2019), this larger accretion rate could be caused by the two or-ders of magnitude greater masses of the filaments hf1, hf2, andhf3. Furthermore, this higher accretion rate on the larger scaleprobably leads to the much larger mass in the W33 complex andprobably promotes proto-cluster formation in the W33 complex(Messineo et al. 2015). Consequently, the W33 complex is pos-sibly accumulating about 1 . × M (cid:12) / Myr as an upper limit.The central hub W33 has a total mass of ∼ . × M (cid:12) , ac-counting for ∼
60% of the mass of the hub–filament system. Thisindicates that the central hub is the mass reservoir of the hub–filament system. The clumps in Fig. 5a are clustered in the inter-section between filaments hf1, hf2, and hf3 and the central hubW33, indicating that gas flows along the filaments are probablychannelling mass to the junction and promoting clump formationin the intersection. Moreover, the other clumps are distributed inthe filaments and the W33 complex might be the result of globalcollapse. Global collapse could be causing the accretion of ma-terials radially into the central zones, providing mass for clumpformation.
5. Summary and conclusions
We present the large-scale molecular CO (1-0), CO (1-0), and C O (1-0) lines, and infrared observations toward theW33 complex and its surroundings. Our main findings are sum-marised as follows:1. According to the centroid velocity distributions of theGaussian fits to all the CO (1-0) and C O (1-0) lines, six ve-locity components are distinguished along the line of sight to-ward the W33 complex from 0 to 70 km s − , namely 2-15 km Article number, page 4 of 21iu et al.: CO OBSERVATION TOWARD W33 s − , 15-21 km s − , 21-28 km s − , 30-44 km s − , 44-48 km s − , and 48-60 km s − . The velocity component 30-44 km s − onlarge scale is likely to be associated with the W33 complex fromthe CO emission in Fig. 3d.2. The PPV space of C O (1-0) reveals that the 30-44 km s − velocity component consists of three di ff erent velocity compo-nents in 30-38.5 km s − , 38.5-42 km s − , and 42-44 km s − . The30-38.5 km s − velocity component shows a real hub–filamentsystem, whose central hub is the W33 complex. Three filamentshf1, hf2, and hf3 are spatially adjacent to the W33 complex indi ff erent directions.3. The central hub W33 has a mass of 1 . × M (cid:12) , account-ing for about 60 % of the mass of the hub–filament system, andthe masses of the filaments hf1, hf2, and hf3 are in the rangeof (2 . − . × M (cid:12) . The mass of the hub–filament systemis concentrated in the W33 complex and the adjacent filaments( ∼ α vir which is lower than 1. Thefilaments hf1, hf2, and hf3 are supercritical with mass per unitlength ranging from 2214 to 3467 M (cid:12) pc − , indicating that theyare also globally collapsing. Regular velocity gradients alongthe filaments hf1, hf2, and hf3 suggest that these filaments arechannelling materials into the junctions with an accretion rate of ∼ . × − M (cid:12) yr − .4. In total, 49 dense clumps are identified to be associatedwith the hub–filament system, and over 57% of the clumps (28)are located at the central hub W33, including 3 MSF clumps.The other clumps are situated in the filaments without the MSFclumps. Also, 70% of the YSO clumps of the system are found inthe central hub. We find that ∼
46% of the clumps in the centralhub W33 and ∼
57% of the clumps in the filaments lie abovethe mass threshold permitting massive star formation, suggestingsignificant potential to form massive stars in those regions.5. The surface density distribution of Class I YSOs in thissystem shows a similar structure to the hub–filament system, andthe peaks in this distribution are at the junctions between thefilaments and the W33 complex, where there are concentrationsof clumps. Gas flows along the filaments possibly provide themass for clump formation and probably create the star formationin the intersections.Based on the above findings, we conclude that the W33 com-plex is likely to be a central hub of a hub–filament system. Also,the gas flows along the tributary filaments in the system probablypromote the proto-cluster formation in the W33 complex.
Acknowledgements.
We are grateful to the sta ff at the Qinghai Station of PMOfor their assistance during the observations. Thanks for the Key Laboratoryfor Radio Astronomy, CAS, for partly supporting the telescope operation. Thiswork is partly supported by the National Natural Science Foundation of China11703040 and the National Natural Science Foundation of China 11933011. Thiswork has made used of data from the NASA / IPAC Infrared Science Archive,which is operated by the Jet Propulsion Laboratory, California Institute ofTechnology, under contract with the National Aeronautics and Space Admin-istration. The ATLASGAL project is a collaboration between the Max-Planck-Gesellschaft, the European Southern Observatory (ESO) and the Universidad deChile. 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. C.-P.Z. acknowledges supports from theNAOC Nebula Talents Program, and the Cultivation Project for FAST ScientificPayo ff and Research Achievement of CAMS-CAS. References
Allen, L. E., Calvet, N., D’Alessio, P., et al. 2004, ApJS, 154, 363André, P., Di Francesco, J., Ward-Thompson, D., et al. 2014, in Protostars andPlanets VI, ed. H. Beuther, R. S. Klessen, C. P. Dullemond, & T. Henning, 27André, P., Men’shchikov, A., Bontemps, S., et al. 2010, A&A, 518, L102Benjamin, R. A., Churchwell, E., Babler, B. L., et al. 2003, PASP, 115, 953 Beuther, H., Tackenberg, J., Linz, H., et al. 2012, ApJ, 747, 43Carey, S. J., Noriega-Crespo, A., Mizuno, D. R., et al. 2009, PASP, 121, 76Chen, H.-R. V., Zhang, Q., Wright, M. C. H., et al. 2019, ApJ, 875, 24Clarke, S. D., Whitworth, A. P., Spowage, R. L., et al. 2018, MNRAS, 479, 1722Csengeri, T., Urquhart, J. S., Schuller, F., et al. 2014, A&A, 565, A75Fazio, G. G., Hora, J. L., Allen, L. E., et al. 2004, ApJS, 154, 10Frerking, M. A., Langer, W. D., & Wilson, R. W. 1982, ApJ, 262, 590Friesen, R. K., Bourke, T. L., Di Francesco, J., Gutermuth, R., & Myers, P. C.2016, ApJ, 833, 204Garden91, R. P., Hayashi, M., Gatley, I., Hasegawa, T., & Kaifu, N. 1991, ApJ,374, 540Gardner, F. F. & Whiteoak, J. B. 1972, Astrophys. Lett., 12, 107Goldsmith, P. F. & Mao, X. J. 1983, ApJ, 265, 791Gómez, G. C. & Vázquez-Semadeni, E. 2014, ApJ, 791, 124Güsten, R., Booth, R. S., Cesarsky, C., et al. 2006a, in Society of Photo-OpticalInstrumentation Engineers (SPIE) Conference Series, Vol. 6267, Society ofPhoto-Optical Instrumentation Engineers (SPIE) Conference Series, ed. L. M.Stepp, 626714Güsten, R., Nyman, L. Å., Schilke, P., et al. 2006b, A&A, 454, L13Guzmán, A. E., Sanhueza, P., Contreras, Y., et al. 2015, ApJ, 815, 130Hennemann, M., Motte, F., & Schneider, N. 2014, in The Labyrinth of Star For-mation, Vol. 36, 271Immer, K., Galván-Madrid, R., König, C., Liu, H. B., & Menten, K. M. 2014,A&A, 572, A63Immer, K., Reid, M. J., Menten, K. M., Brunthaler, A., & Dame, T. M. 2013,A&A, 553, A117Jackson, J. M., Finn, S. C., Chambers, E. T., Rathborne, J. M., & Simon, R. 2010,ApJ, 719, L185Kau ff mann, J., Bertoldi, F., Bourke, T. L., Evans, N. J., I., & Lee, C. W. 2008,A&A, 487, 993Kau ff mann, J. & Pillai, T. 2010, ApJ, 723, L7Kirk, H., Myers, P. C., Bourke, T. L., et al. 2013, ApJ, 766, 115Kohno, M., Torii, K., Tachihara, K., et al. 2018, PASJ, 70, S50Kumar, M. S. N., Palmeirim, P., Arzoumanian, D., & Inutsuka, S. I. 2020, A&A,642, A87Liu, H. B., Hsieh, P.-Y., Ho, P. T. P., et al. 2012, ApJ, 756, 195Liu, T., Zhang, Q., Kim, K.-T., et al. 2016, ApJ, 824, 31Liu, X.-L., Xu, J.-L., Ning, C.-C., Zhang, C.-P., & Liu, X.-T. 2018, Research inAstronomy and Astrophysics, 18, 004Messineo, M., Clark, J. S., Figer, D. F., et al. 2015, ApJ, 805, 110Morgan, L. K., Figura, C. C., Urquhart, J. S., & Thompson, M. A. 2010, MN-RAS, 408, 157Müller, H. S. P., Schlöder, F., Stutzki, J., & Winnewisser, G. 2005, Journal ofMolecular Structure, 742, 215Müller, H. S. P., Thorwirth, S., Roth, D. A., & Winnewisser, G. 2001, A&A, 370,L49Myers, P. C. 2009, ApJ, 700, 1609Peretto, N., Fuller, G. A., André, P., et al. 2014, A&A, 561, A83Peretto, N., Fuller, G. A., Duarte-Cabral, A., et al. 2013, A&A, 555, A112Peretto, N., Lenfestey, C., Fuller, G. A., et al. 2016, A&A, 590, A72Pety, J. 2005, in SF2A-2005: Semaine de l’Astrophysique Francaise, ed. F. Ca-soli, T. Contini, J. M. Hameury, & L. Pagani, 721Pilbratt, G. L., Riedinger, J. R., Passvogel, T., et al. 2010, A&A, 518, L1Pineda, J. E., Caselli, P., & Goodman, A. A. 2008, ApJ, 679, 481Ragan, S. E., Henning, T., Tackenberg, J., et al. 2014, A&A, 568, A73Sanhueza, P., Jackson, J. M., Foster, J. B., et al. 2012, ApJ, 756, 60Schneider, N., Csengeri, T., Hennemann, M., et al. 2012, A&A, 540, L11Schuller, F., Menten, K. M., Contreras, Y., et al. 2009, A&A, 504, 415Treviño-Morales, S. P., Fuente, A., Sánchez-Monge, Á., et al. 2019, A&A, 629,A81Urquhart, J. S., König, C., Giannetti, A., et al. 2018, MNRAS, 473, 1059Veena, V. S., Vig, S., Mookerjea, B., et al. 2018, ApJ, 852, 93Wang, K., Testi, L., Ginsburg, A., et al. 2015, MNRAS, 450, 4043Yuan, J., Li, J.-Z., Wu, Y., et al. 2018, ApJ, 852, 12 Article number, page 5 of 21 & A proofs: manuscript no. 35035corr T a b l e . M a i np r op e r ti e s o f t h e fi l a m e n t s a nd t h e W c o m p l e x i n t h e hub - fi l a m e n t s y s t e m . N a m e < l >< b > V e l o c it y r a ng e < T e x >< σ >< σ N T / c s >< l og ( N ( H )) >< l og ( n ( H )) > M a ss L e ng t h W i d t h M / L ( M / L )c r it [ d e g ][ d e g ][ k m s − ][ K ][ k m s − ][ c m − ][ c m − ] M (cid:12) [ p c ][ p c ][ M (cid:12) p c − ][ M (cid:12) p c − ]( )( )( )( )( )( )( )( )( )( )( )( )( )( ) t h e hub - fi l a m e n t s y s t e m . - . . - . . . . . . . × ———— t h e W c o m p l e x12 . - . . - . . . . . . . × . . —— h f . - . . - . . . . . . . × . . f . - . . - . . . . . . . × . . f . - . . - . . . . . . . × . . N o t e s . C o l u m n1 : C e n t r a l hub W a nd t h e fi l a m e n t ss ho w n i n F i g . a . C o l u m n s - : M ea n c oo r d i n a t e s . C o l u m n4 : S p a nn e dv e l o c it y r a ng e s . C o l u m n5 : M ea n e x c it a ti on t e m p e r a t u r e s . C o l u m n6 : M ea nv e l o c it yd i s p e r s i on s . C o l u m n7 : M ea nnon - t h e r m a l v e l o c it yd i s p e r s i on s i n s ound s p ee dun it s . C o l u m n8 : M ea n H c o l u m nd e n s iti e s . C o l u m n9 : M ea n H nu m b e r d e n s iti e s . C o l u m n10 : T o t a l m a ss . C o l u m n11 : L e ng t h s o f t h e fi l a m e n t s . C o l u m n12 : W i d t h s o f t h e fi l a m e n t s . C o l u m n13 : L i n ea r m a ss . C o l u m n14 : C r iti ca l m a ss t o l e ng t h r a ti o s . Article number, page 6 of 21iu et al.: CO OBSERVATION TOWARD W33 T a b l e . S u mm a r yo f t h e p a r a m e t e r s f o r t h e A TL A S GA L c l u m p s i n t h e hub - fi l a m e n t s y s t e m . N a m e l b V r e f d i s t a n ce r a d i u s S p ea k S i n t T du s t l og ( M a ss ) l og ( N ( H )) c l a ss i fi ca ti on [ d e g ][ d e g ][ k m s − ][ kp c ][ p c ][ J yb ea m − ][ J y ][ K ][ M (cid:12) ][ c m − ]( )( )( )( )( )( )( )( )( )( )( )( ) AGA L . - . . - . . . . . . . . . P r o t o s t e ll a r AGA L . + . . . . . . . . . . . P r o t o s t e ll a r AGA L . - . . - . . . . . . . . . P r o t o s t e ll a r AGA L . - . . - . . . . . . . . . Y S OAGA L . - . . - . . . . . . . . . P r o t o s t e ll a r AGA L . - . . - . . . . . . . . . Y S OAGA L . - . . - . . . . . . . . . Y S OAGA L . - . . - . . . . . . . . . Y S OAGA L . - . . - . . . . . . . . . Q u i e s ce n t AGA L . - . . - . . . . . . . . . P r o t o s t e ll a r AGA L . - . . - . . . . . . . a . a . a M SF AGA L . - . . - . . . . . . . . . Y S OAGA L . - . . - . . . . . . . . . Y S OAGA L . - . . - . . . . . . . . . P r o t o s t e ll a r AGA L . - . . - . . . . . . . . . Y S OAGA L . - . . - . . . . . . . . . Q u i e s ce n t AGA L . - . . - . . . . . . . . . M SF AGA L . - . . - . . . . . . . . . Q u i e s ce n t AGA L . - . . - . . . . . . . . . Y S OAGA L . - . . - . . . . . . . . . P r o t o s t e ll a r AGA L . - . . - . . . . . . . a . a . a Y S OAGA L . - . . - . . . . . . . . . Y S OAGA L . - . . - . . . . . . . . . P r o t o s t e ll a r AGA L . - . . - . . . . . . . . . M SF AGA L . - . . - . . . . . . . . . Y S OAGA L . - . . - . . . . . . . . . Q u i e s ce n t AGA L . - . . - . . . . . . . . . Y S OAGA L . - . . - . . . . . . . . . Y S OAGA L . - . . - . . . . . . . . . Q u i e s ce n t AGA L . - . . - . . . . . . . . . Y S OAGA L . - . . - . . . . . . . . . Q u i e s ce n t AGA L . - . . - . . . . . . . . . Y S OAGA L . - . . - . . . . . . . . . Q u i e s ce n t AGA L . - . . - . . . . . . . . . P r o t o s t e ll a r AGA L . - . . - . . . . . . . . . Q u i e s ce n t AGA L . - . . - . . . . . . . . . Q u i e s ce n t AGA L . - . . - . . . . . . . . . Y S OAGA L . - . . - . . . . . . . . . Y S OAGA L . - . . - . . . . . . . . . P r o t o s t e ll a r AGA L . - . . - . . . . . . . . . P r o t o s t e ll a r AGA L . - . . - . . . . . . . . . Y S OAGA L . - . . - . . . . . . . . . Y S OAGA L . - . . - . . . . . . . . . Q u i e s ce n t AGA L . - . . - . . . . . . . . . Q u i e s ce n t AGA L . - . . - . . . . . . . . . Y S OAGA L . - . . - . . . . . . . . . P r o t o s t e ll a r AGA L . - . . - . . . . . . . . . P r o t o s t e ll a r AGA L . - . . - . . . . . . . . . P r o t o s t e ll a r AGA L . - . . - . . . . . . . . . Q u i e s ce n t N o t e s : C o l u m n1 : N a m e o f t h ec l u m p s . C o l u m n s - : P ea k c oo r d i n a t e s . C o l u m n4 : L S R v e l o c iti e s fr o m U r quh a r t e t a l . ( ) . C o l u m n5 : D i s t a n ce s . C o l u m n6 : E qu i v a l e n t r a d ii . C o l u m n7 : P ea k fl ux e s a t µ m . C o l u m n8 : I n t e g r a t e d fl ux e s a t µ m . C o l u m n9 : D u s tt e m p e r a t u r e s . C o l u m n10 : M a ss e s . C o l u m n11 : H c o l u m nd e n s iti e s . C o l u m n12 : C l a ss i fi ca ti on i nd i ff e r e n t e vo l u ti on a l s t a g e s . a v a l u e s fr o m G u z m á n e t a l . ( ) Article number, page 7 of 21 & A proofs: manuscript no. 35035corr
Table 3.
Percentages of the clumps in each stage in di ff erent places.Name Quiescent clumps Protostellar clumps YSO clumps MSF clumpsThe W33 complex 50%(6) 36%(5) 70%(14) 100%(3)Filaments 50%(6) 64%(9) 30%(6) 0(0)) G a l a c t i c L a t i t u d e W33BW33B1W33 MainW33 Main1W33A1W33A
70 m/24 m/8 m
Fig. 1.
Three-olour composite image towards the W33 complex and its surroundings with blue, green, and red corresponding to Hi-GAL 70 µ m(Pilbratt et al. 2010), GLIMPSE 8.0 µ m (Benjamin et al. 2003), and MIPSGAL 24 µ m (Carey et al. 2009), respectively. The ‘ Xs ’ symbols representthe massive clumps W33 Main, W33A, W33B, W33 Main1, W33A1, and W33B1 identified by Immer et al. (2014) with the ATLASGAL surveyat 870 µ m (Schuller et al. 2009). The plus symbols mark the locations of IRDCs identified by Peretto et al. (2016), and the white dashed ellipserepresents the W33 complex. υ c (km s − )0200400600800 N u m b e r (a) C O 10 20 30 40 50 60 υ c (km s − )050010001500 (b) CO Fig. 2.
Histograms showing the distributions of the velocity centroid of the C O (1-0) and CO (1-0) fits, respectively.Article number, page 8 of 21iu et al.: CO OBSERVATION TOWARD W33 (f) G a l a c t i c L a t i t u d e (e) (d) f3f4f2 f5 f121-28 km s (c) (b)
50 80 120 200 300 450Myr sr (a) Fig. 3. (a)-(f) Velocity-integrated C O (1-0) contours in the velocity ranges of 2-15 km s − , 15-21 km s − , 21-28 km s − , 30-44 km s − , 44-48km s − , and 48-60 km s − respectively, overlaid on the Spitzer µ m image. The white dashed lines in (d) show the filaments detected in 30-44 kms − , and the magenta dashed lines in (e) represent the cutting directions of Fig. 10. The colour bar represents the flux at 8 µ m in units of Myr sr − .Article number, page 9 of 21 & A proofs: manuscript no. 35035corr L ong it ud e . . . . . . . L a tit ud e − . − . − . − . − . − . . . . V e l o c it y k m s − Fig. 4.
The C O (1-0) PPV space in the velocity interval of 30-44 km s − . The projections on the three axes are presented. The colour of eachpoint represents the centroid velocity at that point, corresponding to the colour bar shown to the right of the panel. The symbols ‘ Xs ’ are similar tothose in Fig. 1, and the W33 complex is marked by the black dashed lines or the dashed ellipses on the three projections.Article number, page 10 of 21iu et al.: CO OBSERVATION TOWARD W33 − . − . − . . . . − (a) hf2hf3 hf1 3032343638404244 − . − . − . . . . − (b) f5f4 12 . . . . . .
75 Longitude [degree] − . − . − . . . . L a tit ud e [ d e g r ee ] − (c) Fig. 5.
Plots showing the velocity distributions in 30-38.5 km s − (a), 38.5-42 km s − (b), and 42-44 km s − (c) respectively. The points are colouredwith the velocity values shown in the right colour bar. The asterisks represent the MSF clumps, ‘ (cid:53) ’ mark the YSO clumps, ‘ (cid:52) ’ flag the protostellarclumps, and the squares are for the quiescent clumps. The black lines denote the spines of the filaments hf1, hf2, and hf3. The clumps are codedwith di ff erent colours in the W33 complex (yellow) and in the filaments (red), respectively. Article number, page 11 of 21 & A proofs: manuscript no. 35035corr0 1 2 3radius (pc)0 . . . . . . N u m b e r Mean: 0.9W335 10 15 20 25 30 35T d (K)012345 N u m b e r Mean: 19.121 . . . . . . . ) (cm − )]0246810 N u m b e r Mean: 22.61 . . . . . . . . ⊙ )]02468 N u m b e r Mean: 2.5 0 1 2 3radius (pc)02468 N u m b e r Mean: 0.8Filaments5 10 15 20 25 30 35T d (K)02468 N u m b e r Mean: 14.621 . . . . . . . ) (cm − )]02468 N u m b e r Mean: 22.41 . . . . . . . . ⊙ )]02468 N u m b e r Mean: 2.5
Fig. 6.
Comparisons of the clump physical parameter distributions in the central hub W33 and in the filaments.Article number, page 12 of 21iu et al.: CO OBSERVATION TOWARD W33 − r ( pc ) M ( M (cid:12) ) High-mass star formationLow-mass star formationFilamentsW33
Fig. 7.
Mass–size relationship of the clumps with masses determined. The yellow shaded region represents the parameter space devoid of massivestar formation, where M / M (cid:12) =
580 (R pc − ) . (Kau ff mann & Pillai 2010). The red squares represent the clumps in the central hub W33 and theblue circlesmark the clumps in the filaments. The errors on the masses and radii of the clumps are ∼
20% and ∼ -1.0 -0.5 0.0 0.5 1.0 1.5 2.0 2.5 3.0-0.50.00.51.01.52.02.53.0 [ . ] - [ . ] [5.8]-[8.0]class Iclass II Fig. 8.
IRAC colour-colour plot for the sources in our observational region. The regions indicated the stellar evolutionary stage as defined by Allenet al. (2004). Class I sources represent protostars with circumstellar envelopes and Class II are disc-dominated objects.Article number, page 13 of 21 & A proofs: manuscript no. 35035corr . . . . . .
75 Longitude [degree] − . − . − . . . . L a tit ud e [ d e g r ee ] (a) Class I YSOs 12 . . . . . .
75 Longitude [degree] − . − . − . . . . L a tit ud e [ d e g r ee ] (b) Class I density Fig. 9. (a) Class I YSOs marked by plus symbols overlaid on the C O (1-0) emission. (b) Corresponding overlay of Class I YSOs surface densitycontours on the C O (1-0) emission. The contour levels start from 0.5 pc − to 1.7 pc − (magenta contours) in steps of 0.13 pc − . The di ff erentsymbols represent the clumps in di ff erent stages, similar to those in Fig. 5a. In addition, the clumps marked with the red colour are in the velocityrange of 38.5-42 km s − . . . . . . | υ − υ i n j ec ti on | [ k m s − ] hf1y = 0.32*x+0.16 0 5 10 15offset from junctions [pc]0 . . . . . | υ − υ i n j ec ti on | [ k m s − ] hf2y = 0.12*x+0.11 0 5 10 15offset from junctions [pc]0 . . . . . . . . | υ − υ i n j ec ti on | [ k m s − ] hf3y = 0.10*x-0.26 Fig. 10.
Line-of-sight velocity di ff erence of C O (1-0) as a function of position from junctions.Article number, page 14 of 21iu et al.: CO OBSERVATION TOWARD W33
Fig. 11.
Position–velocity diagrams of C O (1-0) along the cuts in Fig. 3e. The contour levels start from 5 σ to the peak integrated intensity by 3 σ ( σ = . & A proofs: manuscript no. 35035corr
Appendix A: Calculations to derive the physical parameters of each fit
In general, the CO (1-0) emission is optically thick. Therefore, we can estimate the excitation temperature T ex for each fit of CO(1-0) and C O (1-0) respectively via the flowing formula (Garden91 et al. 1991; Pineda et al. 2008) T ex = . ln [1 + . / ( T mb ( CO) + . , (A.1)where T mb ( CO) is the CO (1-0) peak intensity in the velocity range ( υ c − . σ c , υ c + . σ c ) of each fit from the CO (1-0)and C O (1-0) spectra, respectively. Because we check and correct every Gaussian fit from the BST code, the median errors forthe peak intensity, the centroid velocity υ c , and the velocity dispersions σ c are about 1%. Therefore, the derived median excitationtemperatures for the CO (1-0) fits and C O (1-0) fits are about 1 . σ NT and the ratio of σ NT to the sound speed c s . The non-thermalvelocity dispersion σ NT of each fit is calculated using the following equation: σ NT = (cid:113) σ c2 − k B T k m gas , where σ c is the dispersion of eachfit, k B is the Boltzmann constant, T k is the kinetic temperature (here T k ≈ T ex under the assumption of local thermodynamic equilib-rium (LTE) (Morgan et al. 2010)), m gas is the mass of the molecule, and the sound speed c s can be obtained as c s = (cid:112) k B T k /µ H m H ,in which m H is the mass of a hydrogen atom and µ H = . ff mann et al.2008). Thereby, the median errors of σ NT and σ NT / c s are ∼ .
1% and ∼ . CO (1-0) line and the C O (1-0) lines, we can estimatethe optical depth τ of the C O (1-0) lines, the H column density, the H number density, and the mass of each fit. The optical depth τ of the C O (1-0) lines can be derived from the equation as follows (Garden91 et al. 1991; Pineda et al. 2008): τ = − ln[1 − T mb T ( J ( T ex ) − J ( T bg )) ] , (A.2)where T mb is the amplitude (i.e. peak intensity) of each fit, T = h ν/ k , h is the Planck constant, k is the Boltzmann constant, and ν is the transition frequency of the optically thin line. Also, J ( T ) is defined by J ( T ) = / (exp( T / T ) − N = k π µ D hB T ex + hB / k ( J + exp( E u / kT ex )exp( h ν/ kT ex ) − J ( T ex ) − J ( T bg ) τ − exp( − τ ) W , (A.3)where B is the rotational constant of the molecule, J is rotational quantum number of the lower state, µ D is the permanent dipolemoment of the molecule, T bg = .
73 is the background temperature, and E u is the energy of the upper level. The values of these canbe obtained from the Cologne Database for Molecular Spectroscopy (CDMS; Müller et al. 2001, 2005). Also, W is the integratedintensity, which can be estimated using FWHW × amplitude. Next, the H column density of each component can be calculated by N (H ) / N (C O) ≈ × for dense regions (Frerking et al. 1982). The median uncertainty of the H column density is estimatedto be 2% using the median errors of the excitation temperature, the optical depth, the centroid velocity, the velocity dispersion, andthe peak intensity. The H number density and the mass have the same uncertainties as the H column density with a median valueof 2%.Finally, we calculate the H number density and mass of each component by assuming the point as a rectangle shape. Theequations are as follows: n (H ) = N (H ) / R pixel , (A.4) M = µ H m H N (H )( R pixel ) , (A.5)where R pixel is the size of a pixel, µ H = . H is the mass of a hydrogenatom. https: // cdms.astro.uni-koeln.de / cdms / portal / Article number, page 16 of 21iu et al.: CO OBSERVATION TOWARD W3312 . . . . . .
75 Longitude [degree] − . − . − . . . . L a tit ud e [ d e g r ee ] − . . . Median: 0.4 − . . . Median:0.5 − . . . . . . . . . . . Median: 0.4 − . . . Median: 0.4 − . . . Median: 0.5 − . . . Median: 0.5
Fig. A.1.
Distributions of the optical depth of the CO (1-0) line in di ff erent velocity components. The velocity interval of each panel is shown inthe left-top corner. The dashed ellipse in each panel marks the W33 complex (Immer et al. 2014; Kohno et al. 2018) and the ‘ Xs ’ symbols are thesame as those in Fig. 1. Article number, page 17 of 21 & A proofs: manuscript no. 35035corr12 . . . . . .
75 Longitude [degree] − . − . − . . . . L a tit ud e [ d e g r ee ] − .
25 0 . Median: 0.1 − .
25 0 . Median:0.2 − . . . . . . . .
25 0 . Median: 0.1 − .
25 0 . Median: 0.1 − .
25 0 . Median: 0.1 − .
25 0 . Median: 0.2
Fig. A.2.
Distributions of the optical depth of the C O (1-0) line in di ff erent velocity components. The velocity interval of each panel is shown inthe left-top corner. The dashed ellipse in each panel marks the W33 complex (Immer et al. 2014; Kohno et al. 2018) and the ‘ Xs ’ symbols are thesame as those in Fig. 1.Article number, page 18 of 21iu et al.: CO OBSERVATION TOWARD W3312 . . . . . .
75 Longitude [degree] − . − . − . . . . L a tit ud e [ d e g r ee ] −
10 20 300100
Median: 17.6 −
10 20 300100
Median:10.6 − Median: 16.9 −
10 20 300100
Median: 13.8 −
10 20 3002500
Median: 15.4 −
10 20 3001000
Median: 12.8
Fig. A.3.
Distributions of the excitation temperature in di ff erent velocity components. The velocity interval of each panel is shown in the left-topcorner. The dashed ellipse in each panel marks the W33 complex (Immer et al. 2014; Kohno et al. 2018) and the ‘ Xs ’ symbols are the same asthose in Fig. 1. The colour bar is in units of K. Article number, page 19 of 21 & A proofs: manuscript no. 35035corr12 . . . . . .
75 Longitude [degree] − . − . − . . . . L a tit ud e [ d e g r ee ] − Median: 0.7 − Median:0.8 − . . . .
01 20250
Median: 0.7 − Median: 0.7 − Median: 0.8 − Median: 0.8
Fig. A.4.
Distributions of the C O (1-0) velocity dispersion in di ff erent velocity components. The velocity interval of each panel is shown in theleft-top corner. The dashed ellipse in each panel marks the W33 complex (Immer et al. 2014; Kohno et al. 2018) and the ‘ Xs ’ symbols are the sameas those in Fig. 1. The colour bar is in units of km s − .Article number, page 20 of 21iu et al.: CO OBSERVATION TOWARD W3312 . . . . . .
75 Longitude [degree] − . − . − . . . . L a tit ud e [ d e g r ee ] −
22 23050
Median: 22.2 −
22 23050
Median:22.1 − . . . .
522 230250
Median: 22.1 −
22 230100
Median: 22.1 −
22 2301000
Median: 22.2 −
22 230500
Median: 22.2
Fig. A.5.
Distributions of the H column density in di ff erent velocity components. The velocity interval of each panel is shown in the left-topcorner. The dashed ellipse in each panel marks the W33 complex (Immer et al. 2014; Kohno et al. 2018) and the ‘ Xs ’ symbols are the same asthose in Fig. 1. The colour bar is logarithmic with units of cm −2