Current Issue

Journal of the Korean Astronomical Society - Vol. 53 , No. 3

[ Article ]
Journal of the Korean Astronomical Society - Vol. 53, No. 3, pp.77-85
Abbreviation: JKAS
ISSN: 1225-4614 (Print) 2288-890X (Online)
Print publication date 30 Jun 2020
Received 04 Dec 2019 Accepted 08 Jun 2020

1Korea Astronomy and Space Science Institute, 776 Daedeok-daero, Yuseong, Daejeon 34055, Korea (
2Academia Sinica Institute of Astronomy and Astrophysics, P.O. Box 23-141, Taipei 10617, Taiwan (
3Institute of Astronomy and Department of Physics, National Tsing Hua University, Hsinchu 30013, Taiwan (

Correspondence to : Y. C. Minh

Published under Creative Commons license CC BY-SA 4.0


In the molecular cloud G33.92+0.11A, massive stars are forming sequentially in dense cores, probably due to interaction with accreted gas. Cold dense gas, which is likely the pristine gas of the cloud, is traced by DCN line and dust continuum emission. Clear chemical differences were observed in different source locations and for different velocity components in the same line of sight. Several distinct gas components coexist in the cloud: the pristine cold gas, the accreted dense gas, and warm turbulent gas, in addition to the star-forming dense clumps. Filaments of accreted gas occur in the northern part of the A1 and A5 clumps, and the velocity gradient along these features suggests that the gas is falling toward the cloud and may have triggered the most recent star formation. The large concentration of turbulent gas in the A2 clump seems to have formed mainly through disturbances from the outside.

Keywords: ISM: molecules, radio lines: ISM, stars: formation


The molecular cloud G33.92+0.11, which contains an ultra-compact (UC) H II region, is forming massive stars (Lbol ∼ 2.5 × 105L) in its dense clumps (e.g., Watt & Mundy 1999; Liu et al. 2012, 2015; Minh et al. 2016). This cloud shows complicated physical and chemical features marked by various chemical tracers (Minh et al. 2016, 2018, 2019) that result mainly from the activity of newly formed stars which formed on a relatively short evolutionary time scale compared to low-mass stars. Since the star-forming cores are deeply embedded within the dense molecular clumps, their chemistry often provides the only probe for investigating the properties of star-forming cores. In this paper, we report improved observations of the transitions from 13CS, DCN, and H2S that were acquired at a higher spatial resolution θHPBW ∼ 0.2” than our previous observations (∼ 0.6” , Liu et al. 2015; Minh et al. 2016). Especially, we investigate the internal structure and chemical properties of the complex star-forming cores of this source using the chemical characteristics of the observed molecules.

G33.92+0.11 is located at a distance of 7.1 kpc (Fish et al. 2003) and is thought to be nearly faceon in projection (Liu et al. 2012). This orientation and our high-angular resolution observations enabled a detailed investigation of the physical and chemical structure associated with star-forming cores without significant confusion by the source structure. We focused on the central ∼1 pc area of G33.92+0.11 (source “A”) where the star-forming activity is concentrated. This area (G33.92+0.11A) hosts several dense clumps identified by their dust continuum emission at 1.3 mm (Liu et al. 2015). Each clump in turn comprises sub-pc-scale features with different chemical properties. Massive stars are forming in these clumps; the most recent star formation is occurring in the clump associated with dust continuum core 5 (Minh et al. 2016). The UC H II region near continuum core 1 is expanding and, together with the OB stars associated with it, is disrupting the surrounding gas components. The ambient gas is still being accreted to the cloud and is observed as filamentary and arm-like features connected to the dense cores of gas clumps (Liu et al. 2015). We suppose that the accreted gas triggers the star formation in this region.

In Section 2, we summarize our observations made at 220 GHz with the Atacama Large Millimeter and submillimeter Array (ALMA) in Chile, our data from previous observations, and the characteristics of the observed molecules 13CS, DCN, and H2S. Section 3 presents results and discussions of the observed molecular transitions, especially on the internal structure of the clumps; describes the different chemical signatures of the observed transitions; and provides insights on the early phase of massive star formation. Finally, the summary is given in Section 4.

2.1. ALMA Observations

We observed G33.92+0.11 with ALMA in a long baseline array configuration on 2017 August 19 in the 220 GHz band, covering uv distances of ∼13.5−2540 kλ. In addition, we retrieved archival ALMA observations obtained on 2014 April 29 which had failed the official quality assurance (QA2) test and were not published previously. The pointing and phase referencing position of these observations was (α, δ)J2000 = (18h52m50.s272,”º55'29."604). The observational and spectral setups were identical to those of our previous observations carried out on 2014 May (Liu et al. 2015). We used two 234.4 MHz wide spectral windows (channel spacing 61 kHz, ∼0.085 km s−1 ) centered at 231.220690 GHz and 220.679320 GHz and two 1875 MHz wide spectral windows (channel spacing 488 kHz, ∼0.65 km s−1 ) centered at 231.900928 GHz and 217.104980 GHz with a systemic velocity of 107.6 km s−1 . From the difference of the upper and lower sideband images, the rms noise was estimated to be 25 µJy beam−1 (∼41 mK). The peak intensity was 17 mJy beam−1 (∼28 K) in the primary-beam corrected image.

We analyzed the 13CS 5 − 4 (rest frequency ν=231.220690 GHz), DCN 3 − 2 (ν=217.10498 GHz), and H2S 22,0 − 21,1 (ν=216.71044 GHz) transitions. Maps were generated with a synthesized angular resolution of 0.18" × 0.13" for 13CS and 0.21" × 0.15" for DCN and H2S. We note that the spectral window providing a spectral resolution of 61 kHz centered at 231.220690 GHz was corrupted in the observations of May 2014. Therefore, our high spectral resolution image cube of 13CS was created from the data taken in April 2014 and August 2017. All 2014 and 2017 data were utilized when producing the image cubes for the DCN 3 − 2 and H2S 22,0 − 21,1 lines. The final data cubes use a channel spacing of ∼139 kHz (∼0.18 km s−1 ) for 13CS but ∼1 MHz (∼1.5 km s−1 ) for DCN and H2S, providing only eight spectral channels. Therefore we also used our previous data separately for the DCN and H2S lines to compare the spectral profiles to those obtained from our present 13CS data. When comparing the spectral data, we spatially smoothed our present 13CS data with a beam size of 0.6” to match the angular resolutions of our previous DCN and H2S data. We also included the dust continuum image from the ALMA 12 m + ACA array observations (spatial resolution 0.67" × 0.47"), corrected for free-free emission (Liu et al. 2015).

2.2. Observed Molecules

In this paper, we mainly discuss the emission from the 13CS 5−4, DCN 3−2, and H2S 22,0 −21,1 lines. The CS molecule traces dense gas and is one of the most abundant molecules after CO. The critical density of the CS 5 − 4 line is about ∼ 3 × 106 cm−3 at ≤ 50 K (from the Leiden University molecular data, Schöier et al. 2005), and 13CS is the 13C rare isotope substitute of CS. DCN, the deuterated hydrogen cyanide (HCN), shows a high deuteration level in warm dense gas (e.g., Hatchell et al. 1998; Lis et al. 2002; Leurini et al. 2006; Zinchenko et al. 2012). In general, grain mantle evaporation is responsible for high deuteration in warm gas (van Dishoeck et al. 1995; Rodgers & Millar 1996; Hatchell et al. 1998; Roberts & Millar 2000; Tiné et al. 2000; Das et al. 2016). DCN is thought to further fractionate through reactions with CH2D+ in the lukewarm environment (≥ 70 K) of the gas phase (Gerlich et al. 2002; Asvany et al. 2004).

Sulfur is largely depleted onto grain mantles in dense quiet gas, and H2S was expected as a major solid sulfur-containing species on grain mantles (e.g., Charnley 1997). However, it was not detected by Infrared Space Observatory (ISO) observations in solid form and is probably rare in the dust mantle (Minh et al. 2016, and references therein). Therefore, H2S is thought to be formed by reactions with gaseous sulfur-containing species, mainly in the turbulent or high-temperature gas phase, and it remains abundant before being accreted to grains again (Minh 2016). Since the observed species 13CS, DCN, and H2S are each expected to exist in unique physical and chemical environments, we expect that a comparison of the emissions from these species will provide important information on deeply embedded starforming regions and their associated features.

3.1. Different Velocity Components

Figure 1 shows the velocity-integrated intensity map of the 13CS 5 − 4 transition observed with an angular resolution θHPBW ≈ 0.15" × 0.2”. Clumpy and filamentary features associated with the dense cores are more prominent than in a previous map of the same transition (with θHPBW ≈ 0.6” , Liu et al. 2015; Minh et al. 2016). Figure 1 shows the locations where the sample spectra and position–velocity cuts shown in Figures 3 and 4 were obtained. These positions were selected to identify features specific to each clump. As the emission from the clumps is extended, the exact location of the reference points and lines does not affect our analysis. In Figure 2 we show the dust continuum map by Liu et al. (2015) with 13CS emission contours overlayed for comparison. We labeled the gas clumps containing the continuum cores 1, 2, 5, and 9 as the A1, A2, A5, and A9 clumps, respectively.

Figure 1. 
Integrated intensity map for the emission from the 13CS 5−4 transition. The intensity scale is shown at the top of the map. The synthesized beam (θHPBW ≈ 0.18" × 0.13" , position angle −90º ) is illustrated in the bottom left corner. Circles (labeled “SP”) mark the positions of the spectra shown in Figure 3; the line parameters derived from these spectra are listed in Table 1. The position-velocity maps in Figure 4 were obtained along the straight line cuts labeled “PV”.

Figure 2. 
Continuum map at 1.3 mm taken from Liu et al. (2015) with 13CS contours (black; levels are 1, 2, 3, 4 × 0.06 Jy beam−1 km s−1 ). The synthesized beam (θHPBW ≈ 0.67" ×0.47", position angle 81º) is shown in the bottom left corner. Numbers and arrows indicate the continuum cores and molecular “arms”, respectively, identified previously by Liu et al. (2015).

In general, the correlation between 13CS and dust continuum emission in G33.92+0.11A is weak. This can be seen in the filamentary features of the 13CS emission in the northern parts of the A1 and A5 clumps. The most obvious difference is observed in the region containing the dust continuum cores 4 and 6 in arm c3, where the 13CS emission is almost negligible. While 13CS traces warm dense gas, the dust emission traces the overall gas density weighted with temperature. Arguably, 13CS molecules are produced mainly during the warm-up process along with the increase in gas density, which does not need to be proportional to the total gas column density. The 13CS molecules in the filamentary features may have been produced relatively recently in the gas accreted onto the clump from the ambient medium. In contrast, the region containing the continuum cores 4 and 6 may still contain unprocessed, pristine cold gas before collapse. The chemical properties of this region will be further discussed below.

Figure 3 displays selected spectra taken toward the positions indicated in Figure 1. Several spectra show apparent double (or multiple) peaks, which suggests the existence of different velocity components separated from the main component which has a systemic velocity of ∼107.6 km s−1 . Even single-peak spectra often show asymmetric profiles which can be fitted well with two Gaussian components, as shown, for example, for the spectrum taken toward SP2 (top right of Figure 3). We also include previous data for the same transition of 13CS, DCN, and H2S in the spectra for comparison because the present DCN and H2S data (Section 2.1) have a low spectral resolution. Table 1 lists the line parameters for the spectra in Figure 3 and the column densities toward these positions. Column densities were estimated by assuming an optically thin emission and local thermodynamic equilibrium (LTE). We used a rotational temperature Trot = 20 − 50 K to derive the fractional population of the observed species. This temperature range matches the temperatures expected for the extended components of the cloud except the hot core regions which have Trot ∼ 65 K (Liu et al. 2012; Minh et al. 2016). Relative uncertainties of column densities were calculated by adding the 3σ rms values of the spectra and the uncertainties resulting from the Trot range in quadrature.

Figure 3. 
Spectra taken toward the positions marked by circles in Figure 1. The position and the corresponding gas clump are indicated in the upper left and upper right corner of each panel, respectively. Our new 13CS 5 − 4 line data (black) were smoothed spatially using a beam size of about 0.6” to ease comparison to previously obtained data. The two-component Gaussian fit to the SP2 spectrum is shown in the top right panel. This figure includes archival (Section 2.1) DCN 3 − 2 (blue) and H2S 22,0 − 21,1 (red) data. The velocity zero point is systemic velocity vlsr = 107.6 km s−1 .

Table 1 
Line parameters of the observed spectra in Figure 3 and total column densities.
Positiona Moleculeb vlsrc
(km s−1)
(km s−1)
(K km s−1)
SP1 13CS 0.79 6.2 1.62 9.2 0.474 3.1(13) 26
2.78 6.0 1.44 9.5 0.474 3.2(13) 23
DCN 1.14 6.0 3.34 20.8 0.053 3.0(15) 52
H2S 0.80 3.8 3.70 13.0 0.052 1.4(13) 15
SP2 13CS 1.33 6.9 1.99 14.4 0.539 4.8(13) 24
(e) (1.58) (6.3) (2.12)
3.50 2.7 0.99 2.7 0.539 8.9(12) 60
(e) (3.80) (2.2) (1.04)
DCN 1.81 3.0 3.38 11.1 0.060 1.6(15) 52
H2S 2.16 2.8 3.36 10.5 0.042 1.2(13) 15
SP3 13CS 0.07 4.5 1.08 5.8 0.534 1.9(13) 31
2.42 4.7 1.26 7.0 0.534 2.3(13) 30
DCN 1.81 1.0 3.04 3.3 0.057 4.8(14) 54
H2S 2.16 4.5 4.33 9.7 0.044 1.1(13) 16
SP4 13CS −1.55 5.3 2.52 12.1 0.545 4.0(13) 35
DCN −1.56 4.9 3.08 14.7 0.062 2.2(15) 52
H2S −1.23 6.3 2.72 16.3 0.051 1.8(13) 15
SP5 13CS 0.07 18.1 1.62 37.4 1.059 1.3(14) 16
DCN 0.46 7.9 2.25 23.2 0.082 3.4(15) 52
H2S 0.13 8.8 2.45 27.0 0.071 3.0(13) 15
SP6 13CS −0.83 5.3 1.59 8.2 0.498 2.7(13) 30
1.15 5.9 1.54 9.1 0.498 3.0(13) 26
DCN −0.21 2.0 1.69 3.3 0.083 4.8(14) 53
H2S −0.55 1.9 3.38 5.2 0.057 5.7(12) 18
SP7 13CS 0.558 ≤1.9(12)
DCN −0.21 4.9 2.03 11.4 0.062 1.7(15) 52
H2S 0.056 ≤6.1(10)
SP8 13CS 0.523 ≤1.7(12)
DCN 0.46 3.9 2.03 8.4 0.062 1.2(15) 52
H2S 1.48 0.9 1.69 1.5 0.057 1.6(12) 24
Parameter values given as “x(y)” are to be read as “x × 10y”.
aPositions indicated in Figure 1 with circles (diameter ≈ 0.6").
bObserved transitions are the 13CS 5 − 4, DCN 3 − 2, and H2S 22,0 − 21,1 transitions.
cVelocity relative to the systemic velocity vlsr = 107.6 km s−1.
dRelative uncertainty of the column density, estimated by adding in quadrature the observational 3σ rms values and uncertainties from Trot = 20−50 K.
eGaussian fitting results given in the top right panel of Figure 3.

Different velocity components along the same line of sight can show different relative intensities among the observed species. For example, two peaks at different velocities are observed in the spectra of SP1, 3, and 6 (Figure 3) which show different relative emission intensities between 13CS, DCN, and H2S. The gas clumps corresponding to these two peaks have rather similar internal physical conditions considering their 13CS intensities and line widths but their chemical conditions must be different as can be seen from the different emission strengths compared to other species. This is probably because of a different origin of the the two components, like the existing core and an accreted or interacting component. We further discuss the overall characteristics of chemical differences in Section 3.3.

Line widths (∆vHPW) are listed in the column 5 of Table 1. We expect ∆vHPW ≈ 1 − 2 km s−1 for individual components which is small compared to the values typically found in massive star-forming regions (several km s−1). This probably does not result from G33.92+0.11 being physically different from other starforming clouds but from the face-on projected view and the high angular resolution of our observations. Arguably, the observed line widths arise mainly from the internal thermal turbulence of gas with temperatures of a few tens of Kelvin. This suggests that the physical properties of the basic units of dense gas clumps remain similar even though the massive star-forming clouds experience vigorous kinematic activity over their entire volume.

3.2. Velocity Gradients along the Filamentary Features

Figure 4 shows selected position-velocity (p-v) diagrams obtained from the p-v cuts indicated in Figure 1. The p-v cuts, PV-1, 2, and 3, are along the filamentary features associated with the A1 and A5 clumps. These features reach from the dense cores toward the northwest and northeast outward regions of the A1 and A5 clumps, respectively. The diagrams clearly show separate velocity components at the dense cores and velocity gradients along the filamentary features. For example, we find a gradient of about 0.7 km s−1 arcsec−1 (≈ 22 km s−1 pc−1 ) along PV-1 from the A1 core outward. Since we observed the molecular cloud face on, we have no information about the motions in the cloud plane (≈ tangential plane); the observed velocities and velocity gradients might correspond to the line-of-sight component of the accelerated motion of the infalling gas. We note that these filamentary features preferentially occur in the northern parts of the A1 and A5 clumps which are the locations of the most recent star formation activity (Minh et al. 2016). The properties of these filamentary features will be discussed further in the following sections.

Figure 4. 
Position–velocity diagrams obtained along the cut lines shown in Figure 1. Contour levels are 90%, 70%, 50%, and 30% of the peak intensity given in the bottom left corner of each panel. Positions are given as offsets along the corresponding p-v line, with higher values corresponding to higher declinations. The velocity zero point is the systemic velocity vlsr = 107.6 km s−1 .

The A2 clump, which is the most massive clump in the cloud (Liu et al. 2012) with stars already formed in its center (Minh et al. 2016) does not show filamentary features associated with it. Instead, it seems to consist of several components of turbulent gas, which may have originated from the influence of the newly formed stars in the center, (previous) interactions with the UC H II region, and gas that is accreted from the south. The A9 clump shows an elongated morphology pointing south from the A1 clump. As can be seen from its spectrum (SP6 in Figure 3) and p-v diagram (PV-5 in Figure 4), it also seems to consist of various separate gas clumps. The stretched shape may have formed along the gas flow toward the A1 core, probably in the period of star formation in the UC H II region. In this region, emission from shock tracers such as the CN molecule is enhanced relative to other regions (Minh et al. 2019).

Roughly speaking, there is one component which is faster (by about 1–2 km s−1 ) than the systemic velocity in the northwest region (near the north end of the A1 and A5 clumps) and one slower (by about 1– 2 km s−1 ) component in the southeast region (A2 and A9 clumps). Liu et al. (2015) explained this observation with the presence of a velocity gradient along the southeast-northwest direction. Such a velocity gradient, with a change of velocity by about 2 km s−1 over the entire cloud, may result from separate gas components that are being accreted or have been accreted recently from the ambient gas.

3.3. Comparison of Emission Distributions

G33.92+0.11A appears to be undergoing disruption after the formation of the OB stars associated with the UC H II region near the A1 clump. While ambient gas is still accreted to the cloud (Minh et al. 2019), massive stars are continuously forming in the gas clumps. The gas clumps are expected to experience different physical and chemical conditions in the slightly different evolutionary phases that can be probed by specific chemical tracers. Figure 5 shows the DCN 3 − 2 and H2S 22,0 − 21,1 emission distributions, and Figure 6 shows the intensity differences between the transitions. The observed transitions of DCN, H2S, and 13CS have similar critical densities (a few ×106 ∼ 107 cm−1 ) over the expected kinetic temperature range. As we reported previously (Minh et al. 2016, 2018, 2019), intensity differences between observed species are thought to show differences in evolutionary history across the observed region. H2S and DCN also trace very different chemical environments as summarized in Section 2.2. DCN traces the total gas column density including both warm and cold gas, and it shows a strong correlation with the continuum emission (as also noted by Liu et al. 2015). We note the strong DCN emission toward SP7 and 8 (Figures 1 and 3), even though there is no appreciable emission from 13CS and H2S. This “DCN clump” is located in the area surrounded by A1, A2, and A5; its characteristics are discussed in Minh et al. (2018). As 13CS and H2S trace processed warm dense gas, the DCN clump probably largely consists of pristine gas that was present before the cloud collapse began.

Figure 5. 
Integrated intensity maps for the DCN 3 − 2 (left) and H2S 22,0 − 21,1 (right) transitions. Intensity color scales are given at the top of each panel. The DCN map includes the continuum contours (white) at levels of 1, 2, 5, and 10 × 0.003 Jy beam−1 km s−1 . The H2S map includes the same 13CS contours as in Figure 2. The synthesized beams (θHPBW ≈ 0.21" × 0.15", position angle −90º) are shown in the bottom left corners.

Figure 6. 
Difference maps for the integrated intensities between H2S 22,0 − 21,1 and DCN 3 − 2 (left), and between H2S 22,0 − 21,1 and 13CS 5 − 4 (right). The intensities were subtracted after normalizing the intensity of each map to unity peak intensity. The dynamic ranges of the difference maps are given as color bars above each panel. For comparison, the same continuum contours (black) as in Figure 5 are included. The data were smoothed with the θ ≈ 0.6” × 0.6” beam shown in the bottom left corners.

In Figure 6 we directly compare the spatial intensity distributions for the various species. Instead of deriving intensity ratios which show large uncertainties for distributions that show strong variations with position, we directly subtracted the observed intensities after normalizing each image to unity peak intensity. The difference maps thus obtained illuminate trends in chemical evolution in different regions of the molecular cloud. Figure 6(a) shows the difference between the DCN and H2S intensities. The blue and red colors indicate a relative enhancement of DCN and H2S, respectively. DCN is clearly enhanced toward the ridge comprising continuum cores 4, 6, and 11 (the DCN clump) and also toward a region between the A1 and A2 clumps. As discussed before, this suggests that these regions are still composed of pristine gas. Although DCN can also be abundant in warm dense gas, it is not observed there except for the continuum core 1. We suspect that the high density and temperature of this core may have enhanced the DCN abundance by the evaporation of ice mantles. As shown in Figure 6(a), the H2S emission is enhanced relative to DCN toward (1) the 13CS filaments extending from A1 outward, indicating a recent infall of gas; (2) the northern part of A2, which was probably influenced by the expansion of the H II region and also by the newly formed stars in the center of A2; (3) the southern part of A2, where the interaction with accreted gas may have produced turbulence along the boundary (Minh et al. 2016, 2019); (4) the filamentary feature connected to the A5 core where star formation is taking place (Minh et al. 2016); and (5) the A9 clump, another filamentary feature which may trace the infall of gas.

We also compared the normalized intensities of H2S and 13CS in Figure 6(b). In this map, the blue and red colors indicate the relative enhancement of 13CS and H2S, respectively. This map has to be interpreted with caution because the dynamic ranges of the emission strengths of the two transitions are different. Regions where 13CS is enhanced relative to H2S probably consist of unperturbed warm dense gas. Regions where the opposite is the case probably consist of dense turbulent gas. We found enhanced 13CS emission mostly in the west and north of G33.92+0.11A, with filamentary features. These filaments might indicate recent flows of accreted gas which could be responsible for the recent star formation in the A5 clump. We expect that gas accretion from the ambient medium will be intermittent, and the gas will interact with different locations of the molecular cloud at different times. The recent accretion event may have triggered new star formation episodes in the dense cores located in the interaction area. The overall cloud, however, still contains various gas components with different properties and origins.

3.4. Sequential Star Formation

Massive stars form sequentially in G33.92+0.11A, and the OB stars associated with the H II region near the continuum core 1 are probably the first generation that formed. Core 1 is a candidate site for further star formation, but a comparison of various molecular emission signals in this region does not show any evidence for star formation yet (Minh et al. 2016). The highly compressed warm and dense gas in this core appears to be too turbulent to further collapse and form stars. The accreted gas appears to approach core 1 from the northwest, as indicated by the 13CS filaments connected to this core. The accreted gas interacts with the northern part of the A1 clump, resulting in the strong SiO outflows associated with this region (Minh et al. 2016). Fractionation of subcores and their properties have been discussed in detail by Liu et al. (2019).

A second generation of stars may have formed in the center of the A2 clump, which is the most massive cloud in G33.92+0.11A (∼200–300 M, Liu et al. 2012). The strong SiO emission in the center (continuum core 2) together with other signs of star formation (Minh et al. 2016) indicates that massive stars have already formed in A2. It shows strong and complex H2S emission over the entire clump, indicating the presence of abundant turbulent gas (as discussed Section 3.2). A2 appears to be in the process of being disrupted, and the formation of H II regions in the future will further accelerate the disruption of this clump.

Most recently, star formation has occurred in the A5 clump (Minh et al. 2016), as indicated by unique signs of the very early stage of massive star formation such as methanol hot core emission, and strong SiO outflows toward core 5 (Minh et al. 2016). Since these signatures usually disappear right after the formation of a very compact H II region (≲ 104 years, e.g., Minh et al. 2012), we expect that star formation in this core is at a very early stage. The H2S emission shows a filamentary structure connected to the north of A5 (Figure 5), possibly tracing the most recent interaction between accreted gas and the dense core which triggered the star formation in A5. This core seems to not have enough mass for star formation at present (about 1/3 of the A1 or A2 clump masses, Liu et al. 2012), and it also does not have much turbulent gas, traced by H2S emission. Therefore, we expect that core 5 may have formed already at the beginning of the initial overall collapse, and that there was no large-scale collision between the accreting gas and the existing clump just before the star formation began. During the cloud dismantling phase, fractionated dense cores may have survived. Probably due to the influence of the accreted gas, star formation was triggered recently in core 5.

The time scales of the observed chemical signatures are similar to the free-fall time scale of the cores. We expect that the evolution in the very early phase of massive star formation can be probed efficiently using these chemical tracers, since the massive star forming clouds preserve distinct gas components without being blended, even after the beginning of star formation.


Using ALMA, we observed the 220 GHz band transitions of 13CS, DCN, and H2S with high spatial resolution (θHPBW ≈ 0.2”) and high spectral resolution (∼0.2 km s−1 for the 13CS line) toward the massive starforming region G33.92+0.11. While the overall gas cloud is being disrupted, massive stars are forming continuously in separate clumps within the cloud. Using the observed transitions as chemical tracers, we investigated the complicated internal structure of and characteristic differences between the star-forming clumps in the early stages of massive star formation. We found that distinct, unblended gas components coexist in G33.92+0.11A, such as pristine cold gas, accreted dense gas, and warm turbulent gas, in addition to warm dense clumps where stars are forming. Clear chemical differences were found, not only toward different locations within the molecular cloud, but also in different velocity components along the same line of sight.

Our observations reveal a complex hierarchy of star-forming cloud components. Overall, G33.92+0.11A seems to be in a dismantling phase. Even though, fractionated dense cores seem to have survived and stars are formed sequentially probably because of the interaction with accreted gas. The accreted gas is largely concentrated in filaments, especially in the northern parts of the A1 and A5 clumps where the most recent infall of gas from the ambient medium might have occurred. The velocity gradients along the filaments suggest that gas keeps falling toward the cloud and may have triggered the most recent episode of star formation in A5. The presence of turbulent gas and of the star-forming core of the A2 clump indicate that this clump was both influenced by newly formed stars in its center and by disturbances from the outside. In addition to the features associated with massive star formation, a substantial reservoir of pristine cold gas, traced by DCN and the dust continuum, is still present.


This paper makes use of the following ALMA data: ADS/JAO. ALMA#2012.1.00387.S, ADS/JAO.ALMA #2016.1.00362.S. ALMA is a partnership of ESO (representing its member states), NSF (USA) and NINS (Japan), together with NRC (Canada), MOST and ASIAA (Taiwan), and KASI (Republic of Korea), in cooperation with the Republic of Chile. The Joint ALMA Observatory is operated by ESO, AUI/NRAO and NAOJ.

1. Asvany, O., Schlemmer, S., & Gerlich, D. 2004, Deuteration of CHn+ (n = 3 − 5) in Collisions with HD Measured in a Low-Temperature Ion Trap, ApJ, 617, 685
2. Charnley, S. B. 1997, Sulfuretted Molecules in Hot Cores, ApJ, 481, 396
3. Das, A., Sahu, D., Majumdar, L., & Chakrabarti, S. K. 2016, MNRAS, Deuterium Enrichment of the Interstellar Grain Mantle, 455, 540
4. Fish, V. L., Reid, M. J., Wilner, D. J., & Churchwell, E. 2003, H I Absorption toward UC HII Regions: Distances and Galactic Structure, ApJ, 587, 701
5. Gerlich, D., Herbst, E., & Roueff, E. 2002, H3+ +HD→ H2D++H2: Low-temperature Laboratory Measurements and Interstellar Implications, Planet. Space Sci., 50, 1275
6. Hatchell, J., Thompson, M. A ., Millar, T. J., & Macdonald, G. H. 1998, Sulphur Chemistry and Evolution in Hot Cores, A&A, 338, 713
7. Leurini, S., Rolffs, R., Thorwirth, S., et al. 2006, APEX 1 mm Line Survey of the Orion Bar, A&A, 454, L47
8. Lis, D. C., Gerin, M., Phillips, T. G., & Motte, F. 2002, The Role of Outflows and C Shocks in the Strong Deuteration of L1689N, A&A, 569, 322
9. Liu, H. B., Jiménez-Serra, I., Ho, P. T. P., et al. 2012, Fragmentation and OB Star Formation in High-Mass Molecular Hub-Filament Systems, ApJ, 756, 10
10. Liu, H. B., Galvań-Madrid, R., Jimeńez-Serra, I., et al. 2015, ALMA Resolves the Spiraling Accretion Flow in the Luminous OB Cluster-forming Region G33.92+0.11, ApJ, 804, 37
11. Liu, H. B., Chen, H. V., Román-Zúñiga, C. G., et al. 2019, Investigating Fragmentation of Gas Structures in OB Cluster-forming Molecular Clump G33.92+0.11 with 1000 au Resolution Observations of ALMA, ApJ, 871, 185
12. Minh, Y. C., Chen, H.-R., Su, Y.-N., & Liu, S.-Y. 2012, SMA Observations of the Hot Cores of DR21(OH), JKAS, 45, 157
13. Minh, Y. C., Liu, H. B., & Galvań-Madrid, R. 2016, Chemical Diagnostics of the Massive Star Cluster-forming Cloud G33.92+0.11. I. 13CS, CH3OH, CH3CN, OCS, H2S, SO2, and SiO, ApJ, 824, 99
14. Minh, Y. C. 2016, Sulfur-bearing Molecules Observed in the Massive Star-forming Regions DR21(OH) and G33.92+0.11, J. Phys. Conf. Ser., 728, 052007
15. Minh, Y. C., Liu, H. B., Galvań-Madrid, R., et al. 2018, Chemical Diagnostics of the Massive Star Cluster-forming Cloud G33.92+0.11. II. HDCS and DCN, ApJ, 864, 102
16. Minh, Y. C. & Liu, H. B. 2019, Chemical Diagnostics of the Massive Star Cluster-Forming Cloud G33.92+0.11. III. 13CN and DCN, JKAS, 52, 83
17. Roberts, H. & Millar, T. J. 2000a, Modelling of Deuterium Chemistry and Its Application to Molecular Clouds, A&A, 361, 388
18. Rodgers, S. & Millar, T. 1996, The Chemistry of Deuterium in Hot Molecular Cores, MNRAS, 280, 1046
19. Rodgers, S. D. & Charnley, S. B. 2001, Chemical Differentiation in Regions of Massive Star Formation, ApJ, 546, 324
20. Schöier, F. L., van der Tak, F. F. S., van Dishoeck E. F., & Black, J. H. 2005, An Atomic and Molecular Database for Analysis of Submillimetre Line Observations, A&A, 432, 369
21. Tiné, S., Roueff, E., Falgarone, E., et al. 2000, Deuterium Fractionation in Dense Ammonia Cores, A&A, 356, 1039
22. van Dishoeck, E. F., Blake, G. A., Jansen, D. J., & Groesbeck, T. D. 1995, Molecular Abundances and Low-Mass Star Formation. II. Organic and Deuterated Species toward IRAS 16293-2422, ApJ, 447, 760
23. Watt, S. & Mundy, L. G. 1999, Molecular Environments of Young Massive Stars: G34.26+0.15, G11.94-0.62, G33.92+0.11, and IRAS 18511+0146, ApJS, 125, 143
24. Zinchenko, I., Liu, S.-Y., Su, Y.-N., et al. 2012, A Multiwavelength High-resolution study of the S255 Star-forming Region: General Structure and Kinematics, ApJ, 755, 177