Digging into the Interior of Hot Cores with ALMA (DIHCA). I. Dissecting the High-mass Star-Forming Core G335.579-0.292 MM1
Fernando A. Olguin, Patricio Sanhueza, Andrés E. Guzmán, Xing Lu, Kazuya Saigo, Qizhou Zhang, Andrea Silva, Huei-Ru Vivien Chen, Shanghuo Li, Satoshi Ohashi, Fumitaka Nakamura, Takeshi Sakai, Benjamin Wu
DDraft version January 22, 2021
Typeset using L A TEX twocolumn style in AASTeX63
Digging into the Interior of Hot Cores with ALMA (DIHCA).I. Dissecting the High-mass Star-Forming Core G335.579–0.292 MM1
Fernando A. Olguin , Patricio Sanhueza ,
2, 3
Andr´es E. Guzm´an , Xing Lu , Kazuya Saigo, Qizhou Zhang , Andrea Silva , Huei-Ru Vivien Chen , Shanghuo Li , Satoshi Ohashi , Fumitaka Nakamura ,
2, 3
Takeshi Sakai , and Benjamin Wu
9, 2 Institute of Astronomy and Department of physics, National Tsing Hua University, Hsinchu 30013, Taiwan National Astronomical Observatory of Japan, National Institutes of Natural Sciences, 2-21-1 Osawa, Mitaka, Tokyo 181-8588, Japan Department of Astronomical Science, SOKENDAI (The Graduate University for Advanced Studies), 2-21-1 Osawa, Mitaka, Tokyo181-8588, Japan Center for Astrophysics | Harvard & Smithsonian, 60 Garden Street, Cambridge, MA 02138, USA Institute of Astronomy and Department of Physics, National Tsing Hua University, Hsinchu 30013, Taiwan Korea Astronomy and Space Science Institute, 776 Daedeokdae-ro, Yuseong-gu, Daejeon 34055, Republic of Korea RIKEN Cluster for Pioneering Research, 2-1, Hirosawa, Wako-shi, Saitama 351-0198, Japan Graduate School of Informatics and Engineering, The University of Electro-Communications, Chofu, Tokyo 182-8585, Japan. NVIDIA, 2788 San Tomas Expressway, Santa Clara, CA 95051, USA (Received XXX, 2020; Revised YYY, 2020; Accepted January 22, 2021)
Submitted to ApJABSTRACTWe observed the high-mass star-forming region G335.579–0.292 with the Atacama Large Millime-ter/submillimeter Array (ALMA) at 226 GHz with an angular resolution of 0 . (cid:48)(cid:48) ∼ CN emission. In addition, hot gasexpansion in the innermost region is unveiled by a red-shifted spot in the first moment map of HDCOand (CH ) CO (both with E u > II region, while infall and rotation motions originate in the outer regions. ALMA3 shows clear signsof rotation, with a rotation axis inclination with respect to the line of sight close to 90 ◦ , and a systemmass (disk + star) in the range of 10–30 M (cid:12) . Keywords: stars:formation – stars: massive – ISM: kinematics and dynamics – ISM: individual objects(G335.579–0.272) INTRODUCTIONHigh-mass stars form in dense “hot cores” (sizes offew 1000 au), within massive molecular clumps that havebeen recently found to be preponderantly under globalcollapse (Jackson et al. 2019). Observational evidenceshows that many “hot cores” are located in the inter-
Corresponding author: Fernando [email protected] section of large pc-scale filaments, the so-called hubs,which are fed through filaments (e.g., Lu et al. 2014,2018; Trevi˜no-Morales et al. 2019; Chen et al. 2019).Similarly, at smaller scales ( < a r X i v : . [ a s t r o - ph . GA ] J a n Olguin et al. zation of the kinematics of the gas at core and smallerscales is key to understanding how high-mass stars insingle or multiple systems form, in addition to providingbetter constraints for numerical simulations. In particu-lar, cores inside infrared dark clouds (IRDCs; Rathborneet al. 2006; Chambers et al. 2009; Sanhueza et al. 2012,2019; Li et al. 2019, 2020) are more likely to be youngenough that ionization plays a minor role, thus limit-ing the number of physical processes to account for in adetailed study.The IRDC G335.579–0.292 is a massive cloud with amass of 5 . × M (cid:12) at a distance of 3.25 kpc (Perettoet al. 2013). Peretto et al. (2013) identified two cores us-ing 3 mm Atacama Large Millimeter/submillimeter Ar-ray (ALMA) observations which are massive enoughto form high-mass stars. The most massive core,G335–MM1, is one of the most massive cores in theGalaxy with a mass of 545 M (cid:12) (comparable to onlyfew other cores, e.g., Stephens et al. 2015). G335–MM1 is being fed by large scale infalling gas at a rateof ∼ − M (cid:12) yr − (Peretto et al. 2013). Avison et al.(2015) estimated a bolometric luminosity of 1 . − . × L (cid:12) for G335–MM1. Their radio cm wavelength ob-servations revealed that G335–MM1 further fragmentsinto at least two sources (MM1a and MM1b) with spec-tral indices consistent with hyper-compact (HC) H II re-gions, and with free-free emission equivalent to those ofzero age main sequence (ZAMS) B-type stars (9–10 M (cid:12) ).Class II methanol, which are exclusively associated withhigh-mass star formation, and water masers have alsobeen observed towards both regions within G335–MM1(Breen et al. 2010; Caswell et al. 2011; Avison et al.2015). Regardless of these efforts, there are no studiesso far of the gas kinematics at the size scales directlyrelated to the star formation process ( ∼ § §
3. Finally, our discussion and conclusions are presentedin § § OBSERVATIONS We observed G335–MM1 in band 6 (226.2 GHz,1.33 mm) with ALMA during November 2016, cycle 4(Project ID: 2016.1.01036.S; PI: Sanhueza). Observa-tions were performed with the 12 m array using 41 an-tennas in a configuration similar to C40-5, with mini-mum and maximum baselines of 18.6 and 1100 m, re-spectively. With this configuration, the observationsachieved a resolution of ∼ . (cid:48)(cid:48) ∼ . (cid:48)(cid:48) ∼ ∼ . − ) anda bandwidth of 1.875 GHz. These windows coveredthe frequency ranges between 233.5–235.5 GHz, 231.0–233.0 GHz, 216.9–218.7 GHz and 219.0–221.0 GHz.The data were calibrated using the CASA 4.7 reduc-tion pipeline (version r38377; McMullin et al. 2007).According to the ALMA Proposer’s User Guide, the es-timated error in the absolute flux is 10 %. The data werethen self-calibrated, and a continuum map from line-freechannels and continuum subtracted data cubes were pro-duced. The channel identification procedure is detailedin Appendix A. The script and continuum subtractionpipeline are available on GitHub under a MIT License.Version 2.0 of the script is archived in Zenodo (Olguin &Sanhueza 2020). We used the tclean task with Briggsweighting and robust parameter of 0.5 to image the data,resulting in a continuum sensitivity of 0.4 mJy beam − (vs. expected thermal noise level of 0.15 mJy beam − ).The cleaned continuum map is shown in Figure 1. Thesynthesized beam of the continuum map is 0 . (cid:48)(cid:48) × . (cid:48)(cid:48) ◦ .We used the automatic masking procedure YCLEAN(Contreras et al. 2018; Contreras 2018) to CLEAN thedata cubes for each spectral window. YCLEAN iter-ates between CLEANing steps with increasingly deeperthresholds, and mask generating steps. The emissionthat is added to the masks and the CLEANing thresh-olds are based on the the image rms, residuals and beamsecondary lobe at each step. A noise level per channelbetween 4–6 mJy beam − was achieved. RESULTSIn this section, we describe the G335–MM1 high-massstar-forming region in 1.3 mm dust continuum emissionand selected molecular lines. Those will be good tracersto investigate the morphology and kinematic structuresof high-mass star forming cores.3.1.
Morphology and Line Identification GoContinuum: https://github.com/folguinch/GoContinuum. RA (ICRS) D e c ( I C R S ) ALMA1ALMA2 ALMA3ALMA4ALMA5 MM1a MM1b5000 au
Intensity (mJy beam ) Figure 1.
ALMA continuum map of G335–MM1 at 1.3 mmin color scale and contours. The grey contours levels are5 , , , , , , × σ cont with σ cont = 0 . − .The green crosses and labels mark the peak position ofthe sources identified within this region and are labeled bybrightness. The position of the radio continuum sources fromAvison et al. (2015) are marked with light blue triangles. Thebeam size is shown in the lower left corner and correspondsto a scale of ∼ The single G335–MM1 region identified at 3 mm ob-servations ( ∼ (cid:48)(cid:48) resolution) by Peretto et al. (2013) frag-ments into at least five continuum sources (Figure 1).These sources were identified by visual inspection. Thepositions and 1.3 mm fluxes of these sources were derivedfrom a 2-D Gaussian fit to the continuum emission andare listed in Table 1. The fit indicates that the sourcesare elongated with ratios between their axes in the 1.2–2.0 range. The brightest of the five 1.3 mm sources iden-tified by us (ALMA1) matches with the position of oneof the two radio continuum sources identified by Avisonet al. (2015), MM1a. The second radio source, MM1b, islocated in between of two of the 1.3 mm sources (ALMA3and ALMA4).We have identified two sources potentially associatedwith MM1b (ALMA3 and ALMA4). Avison et al. (2015)argue that the spectral indices from radio emission forMM1b are consistent with a collimated jet or a compactH II region. They suggested that this emission is notassociated with MM1a, but rather with another sourcedue to the presence of CH OH maser emission closer to MM1b. They discarded the jet hypothesis due to themisalignment of MM1b with the HNC emission tracingthe molecular outflow associated with MM1a. Given theoffset position of MM1b to the ALMA sources, we ar-gue that the radio emission is possibly tracing the lobeof an ionized jet likely associated with the high-masssource ALMA3. However, higher resolution observa-tions at wavelengths λ ≥ II regions (e.g., Hatchellet al. 1998). A lower number of species is detectedtowards ALMA3. The lines detected in ALMA3 areweaker than those of ALMA1. We focus our analysis ina handful of lines to explore the kinematics of these twocores. These lines and their properties are summarizedin Table 2. The molecular lines detected in other sourcescorrespond to transitions tracing more extended mate-rial (e.g., CO, C O), and thus cannot be assigned toa source in particular. The lack of line emission towardsthe other sources may indicate that they are in an earlierevolutionary stage, likely in the prestellar phase.We estimated the local standard of rest (LSR) veloc-ity ( v LSR ) of ALMA1 from the CH CN J = 12 − K = 7 and 8 transitions. For ALMA3, we estimatedthe v LSR from the centroid of the K = 7 line, which isequivalent to the velocity at the position of the sourcefrom its first moment map. The velocities are listed inTable 1. Figure 2 shows the continuum of ALMA1 andexample CH CN spectra from two locations marked onthe continuum image. The saturated CH CN line emis-sion in Figure 2b indicates that the lower K transitionshave become optically thick towards the center positionof ALMA1. These transitions even show self-absorbedprofiles, except for K = 7 and 8, which are opticallythinner. The spectrum shown toward the south-east ofthe peak position of ALMA1 (Figure 2c) is typical of op-tically thin spectra around the central region, display-ing Gaussian-like profiles and not self-absorption fea-tures. Although the gas traced by high K -ladder tran-sitions is hotter than low K transitions and may thuscorrespond to a different velocity component, the de-rived LSR velocity is consistent with other species andprevious estimations of the source v LSR at larger scales( v LSR = − . − ; Bronfman et al. 1996). To de-termine the v LSR of ALMA1, the lines were fitted witha Gaussian profile and the adopted velocity correspondto the average v LSR of both K = 7 and 8 transitions. Olguin et al.
Table 1.
Continuum source propertiesALMA RA Dec I peak F total Deconvolved Size v LSR
Source [ h : m : s ] [ ◦ : (cid:48) : (cid:48)(cid:48) ] (mJy beam − ) (mJy) ( (cid:48)(cid:48) ) (km s − )1 16:30:58.761 -48:43:54.10 209.8 566 ±
28 0 . × .
39 –46.92 16:30:58.703 -48:43:52.60 60.2 114 ±
12 0 . × .
26 –3 16:30:58.631 -48:43:51.28 34.1 88 ± . × .
39 –47.64 16:30:58.676 -48:43:51.78 27.8 48 ± . × .
21 –5 16:30:58.889 -48:43:55.18 12.6 48 ± . × .
51 –
Note —Positions in ICRS coordinate standard.
Table 2.
Summary of lines analyzedMolecule Transition Frequency E u Ref.(GHz) (K)CH CN J K = 12 − CN J K = 12 − CN J K = 12 − CO J = 2 − J = 5 − CO J K a ,K c = 3 , − , J K a ,K c = 28 , − , ) CO J K a ,K c = 54 , − , AE 231.6868319 1148 (1)
References —(1) Jet Propulsion Laboratory (JPL, Pickett et al. 1998);(2) Cologne Database for Molecular Spectroscopy (CDMS, M¨uller et al.2005).
Physical Properties
From the 2-D Gaussian fit to the dust continuumimage, we estimate the size of the sources. The ra-dius is defined as the geometric mean of the decon-volved semi-major and minor axes (see Table 1). Theresults are listed in Table 3. All sources are compact( ∼ −
800 au) with radii typical of inner envelope/discscales ( < CN J = 12 − K = 0 to 7 transitions towardsthe continuum peak of the sources with local thermalequilibrium (LTE) 1-D model. However, the line emis-sion towards ALMA1 becomes quickly saturated, hencewe only fit the temperature towards ALMA3. The fitwas performed with the Markov Chain Monte Carlomethod within XCLASS (M¨oller et al. 2017). We obtaina gas temperature at the continuum peak of ALMA3of 290 K. For ALMA1, we estimate a dust mass usinga temperature of 100 K and 300 K. The former corre- Table 3.
Source physical propertiesALMA
R T M d N H N CH CN Source (au) (K) (M (cid:12) ) (10 cm − ) (cm − )1 a
710 100 19 10.4 –300 6.2 3.3 –2 b > <
24 18.7 –3 730 290 1.0 0.6 10 b > <
10 8.7 –5 b > <
10 3.9 –
Note —The radius, R , corresponds to half of the geo-metric mean of the deconvolved sizes (FWHM) fromTable 1. a The dust temperatures are an approximation (see Sec-tion 3.2). b CH CN emission is not detected towards thesesources, hence the temperature is a lower limit, andthe gas masses and column densities are upper limits. spond to roughly the brightness temperature at whichthe lower K transition lines saturate. As the emissionis becoming optically thick, the brightness temperatureshould converge to the kinetic temperature. However,the beam filling factor is likely less than one (e.g., Suet al. 2009), hence this provides a temperature lowerlimit and in turn a dust mass upper limit. The lattertemperature is based on the result for ALMA3.Assuming that the dust and gas temperatures are inequilibrium, we estimate the gas mass as M d = F ν d R gd κ ν B ν ( T d ) (1)with F . the flux density from Table 1, d = 3 .
25 kpcthe source distance, the gas-to-dust mass ratio R gd =100, the dust opacity κ . = 1 cm gr − (Ossenkopf& Henning 1994), and B ν the Planck blackbody func- RA (ICRS) D e c ( I C R S ) centerSE arcfeature (a) Continuum 2500 au Frequency (GHz) I n t e n s i t y ( J y b e a m ) (b) Center Frequency (GHz) I n t e n s i t y ( J y b e a m ) (c) SE Figure 2.
G335 ALMA1 continuum map and CH CN J = 12 − K = 0 to 8 line emission examples. (a) Lo-cations of the example spectra marked over the continuummap. The light blue dashed line shows the arc shapedstructure. Contour level corresponds to 20 × σ cont with σ cont = 0 . − . (b) Line emission towards con-tinuum peak, which is marked with a green cross on (a). (c)Line emission towards the south-east of the dust peak posi-tion, in a less dense region marked with a blue triangle in(a). Dashed gray vertical lines mark the rest frequency ofthe K transitions (0–8 from left to right) corrected by thesystemic velocity ( v LSR = − . − ). tion. The H column density is estimated from the dustemission as N H = I ν R gd B ν ( T d ) κ ν µ H m H (2)with I . the peak intensity from Table 1, µ H = 2 . m H the atomic hydrogen mass.The results are listed in Table 3. We estimate that thecontribution from free-free emission to the 1.3 mm fluxdensities is less than 5 mJy for ALMA1 and less than1 mJy for ALMA3 from the fit to the radio cm observa-tions in Avison et al. (2015), and are thus negligible.The 1.3 mm flux density of G335–MM1 is 1.4 Jy fromaperture photometry and 0.9 Jy by summing the valuesin Table 1. The expected flux at 1.3 mm from the spec-tral energy distribution in Avison et al. (2015), obtainedby interpolating the data points at 870 µ m and 3.2 mm,is ∼ . ∼
10% of the total mass estimated by Peretto et al.(2013) from 3.2 mm observations. The discrepancy inmass can be explained primarily by the temperature es-timates, followed by extended emission not included inthe 2-D Gaussian fit measurements, and finally emissionfiltered out by the interferometer.The mass of the gas reservoir of ALMA3 is 1 M (cid:12) (Ta-ble 3). However this value is likely a lower limit, becausecooler regions of the envelope also contribute to the dustemission. Emission from the region may have also beenfiltered out by the interferometer.3.3.
Kinematics
Line emission from CH CN transitions, a commonlyused tracer of gas rotation, is detected only towards twosources (ALMA1 and ALMA3). In ALMA1, transitions K = 0 to 5 are saturated and blended with other molec-ular lines (e.g., Figure 2). We therefore calculated themoments 0, 1, and 2 for transition K = 7, which arepresented in Figure 3 (line width is displayed instead ofvelocity dispersion). For ALMA3, the transition K = 4is used in the moments shown in Figure 4 as there is lesscontamination from other molecular lines. First and sec-ond moments were calculated only with data over 5 σ rms with σ rms = 5 mJy beam − . Line emission from CH CNis only detected in ALMA1 and ALMA3 as shown bythe contours in Figures 3a and 4a. The deconvolvedsize of the emission in the contours of Figure 3a is 0 . (cid:48)(cid:48) ∼ CO J = 2 − J = 5 − Olguin et al. -48°43'53"54"55"56" (a) CH CN K = 7 - moment 0 and 12500 au V e l o c i t y ( k m s ) RA (ICRS) D e c ( I C R S ) (b) moment 2 L i n e w i d t h ( k m s ) Figure 3.
G335 ALMA1 moment maps from CH CN J =12 − K = 7 line emission. (a) First moment map with ze-roth moment shown in green contours. Zero systemic velocitycorresponds to the v LSR (–46.9 km s − ). Contour levels are5, 10, 20, 40, 80 × σ rms with σ rms = 31 mJy beam − km s − .The blue arrow shows the direction of the blue-shifted out-flow lobe (see Section 3.3). (b) Second moment map. Thepink crosses mark the position of the continuum sources, andthe beam size is shown in the lower left corners. The physicalscale is shown in the lower right corner of panel (a). of the flows (Figure 5). The blue and red windows areseparated ± .
25 km s − from the v LSR and have widthsof ∼ .
75 and 13 km s − for CO and SiO, respectively.A clear molecular flow is detected towards ALMA1 inthe NE-SW direction. From the blue-shifted emission,we estimate a position angle PA ∼ ◦ . Additionally,the arc shaped structure observed in the dust emissiontowards the west of ALMA1 (Figure 2a) is likely asso-ciated with emission from the base of the outflow cav-ity as seen in CO. This may be the result of a wideangle wind interacting with the envelope (e.g., Kuiperet al. 2015), as observed in the diffuse dust emission andoutflow emission of G16.64+0.16 (Maud et al. 2018).Towards ALMA3, CO seems to be tracing the enve-lope of the source with the blue- and red-shifted orien-tation consistent with the rotation pattern observed inCH CN. The red-shifted SiO emission towards the NW -48°43'51"52" (a) CH CN K = 4 - moment 0 and 12000 au V e l o c i t y ( k m s ) RA (ICRS) D e c ( I C R S ) (b) moment 2 L i n e w i d t h ( k m s ) Figure 4.
G335 ALMA3 moment maps of CH CN J = 12 − K = 4 line emission. (a) First moment map with zerothmoment shown in green contours. Zero systemic velocity cor-responds to the source v LSR (–47.6 km s − ). Contour levelsare 5, 10, 20, 40 × σ rms with σ rms = 51 mJy beam − km s − .The arrow shows the derived direction of the rotation axis(following the right hand rule; see Section 3.3). (b) Secondmoment map. The pink crosses mark the position of thecontinuum sources, and the beam size is shown in the lowerleft corners. The physical scale is shown in the lower rightcorner of panel (a). shows structures that may be associated with ALMA3(SE-NW direction, P . A . ∼ − ◦ ) and ALMA4 (SW-NEdirection, P . A . ∼ ◦ ). However, since ALMA3 seems tobe closer to edge-on given the symmetry of the CH CNfirst moment map and the origin of the SiO emissiondoes not coincide with ALMA3, we cannot rule out thatthe SiO emission is related to an unresolved source.We have identified two lines with upper energy levelshigher than 1000 K: HDCO (28 , − , ) with E up equal to 1460 K and (CH ) CO (54 , − , ) AEwith E up equal to 1140 K. Their moment maps are dis-played in Figure 6. The emission of these two lines iscompact, but resolved towards ALMA1. Both lines arelikely tracing the innermost regions of the hot core giventhe high temperatures required to be excited. The firstmoment maps of these lines show a consistent velocity -48°43'48"51"54"57" (a) CO (2 1) 10000 au
RA (ICRS) D e c ( I C R S ) (b) SiO (5 4) 10000 au Figure 5.
Blue- and red-shifted zeroth order moment from CO J = 2 − J = 5 −
4. Blue and red con-tours show the blue- and red-shifted emission, respectively.Contour levels are 3, 6, 12, 24, 48 and 96 × σ rms with σ rms =47 mJy beam − km s − for CO and 38 mJy beam − km s − for SiO. Gray scale map corresponds to the continuum emis-sion and green crosses mark the position of the ALMAsources. The beam is shown in the lower left corner. gradient with that of CH CN, but remarkably, the linesare red-shifted rather than blue-shifted towards the cen-ter of the source. DISCUSSIONHere, we discuss the results focused primarily on thekinematics of ALMA1 and ALMA3.4.1.
ALMA1
Large Scale Infall
The velocity field in ALMA1 presents an overall gra-dient from east to west (Figure 3a). However this gra- dient is likely masked as a result of a combination ofgas motions. Possible reasons that make velocity gradi-ents difficult to identify are (e.g., Silva et al. 2017): (i)CH CN is possibly tracing outflowing motions, for in-stance, from gas removed from a disk surface by stellarwinds, together with rotation and infall (e.g., Beutheret al. 2017) making the interpretation of the velocitygradients less straightforward; (ii) the observed veloc-ity field is potentially produced by the combination ofgas motions due to the presence of unresolved sources;(iii) the orientation of the protostellar disk is close toface-on; and/or (iv) infall or expansion motions.The first moment map of the K = 7 transition pre-sented in Figure 3a shows a spot of velocities close tozero at the center of the core enclosed mainly by the80 × σ rms level of the zeroth moment contour. Thesevelocities are bluer than expected for a smooth tran-sition of velocity from east to west, assuming CH CN istracing the rotation of the core envelope/disk. Hence-forth we refer to this region as the blue-shifted spot. Asshown in Figure 2, the CH CN lines up to K = 4 showan absorption feature at the center of the line, resem-bling the blue-asymmetry characteristic of infall motions(e.g., Zhang et al. 1998) as is the blue-shifted spot (e.g.,Estalella et al. 2019). This blue-asymmetric profile isnot evident in larger, likely optically thinner, K transi-tions, but the effect of the collapse can still be detectedthrough the first moment map. The first moment mapspot of blue-shifted velocities (with respect to the v LSR )is thus likely produced by line profiles that are blue-skewed. To further investigate the infall hypothesis, wesearch for blue-shifted emission coupled with red-shiftedabsorption against the continuum source, also known asan “inverse P-Cygni profile”. Such features provide un-ambiguous evidence for infall toward the central proto-star (e.g., Evans et al. 2015).Among the myriad of lines included in the spectralsetup, those that trace a more extended gas distribu-tion display absorption against the continuum only atthe center position of ALMA1 (and not anywhere else).Figure 7 shows the spectrum of CO (2 −
1) and H CO(3 , − , ) at the peak position of the dust continuum ofALMA1. The CO displays an absorption against thecontinuum that is red-shifted with respect to the v LSR of –46.9 km s − by 1.7 km s − . On the other hand, theH CO line in Figure 7b shows a dip consistent with thesource v LSR . In order to be excited, CO requires lowerdensities and temperatures than H CO. We thereforesuggest that the absorption against the continuum thatis red-shifted with respect to the v LSR in the CO lineis tracing a large scale infall, consistent with the findingsat much lower resolution of 5 (cid:48)(cid:48) ( ∼ Olguin et al. -48°43'53"54"55"56" (a) HDCO - moment 0 2500 au I n t e n s i t y ( J y k m b e a m s ) (b) moment 1 V e l o c i t y ( k m s ) (c) moment 2 L i n e w i d t h ( k m s ) RA (ICRS) D e c ( I C R S ) (d) (CH ) CO - moment 02500 au I n t e n s i t y ( J y k m b e a m s ) (e) moment 1 V e l o c i t y ( k m s ) (f) moment 2 L i n e w i d t h ( k m s ) Figure 6.
Moment maps of hot lines ( E u > , − , ) transition ( E u = 1460 K). (d)–(f) Same as upper row but for (CH ) CO (54 , − , ) AE transition( E u = 1148 K). Zero systemic velocity in (b) and (e) corresponds to the source LSR velocity, v LSR = − . − , and thecolor scales are shifted to highlight the velocity structure around the continuum source. Blue arrows in (b) and (e) show thedirection of the blue-shifted outflow. et al. (2013). This is also supported by the positionvelocity (pv) map of CO in Figure 8, where the mor-phology of the emission resembles the “C” shape char-acteristic of infall motions (Zhang & Ho 1997). H COemission may be produced closer to the core center (orin the outflows) making harder the detection of infallsigns (although the absorption dip is slightly red-shiftedwith respect to the v LSR ).4.1.2.
Small Scale Expansion
The spot at the core center in the first moment map(Figure 6) of the hot lines HDCO and (CH ) CO is red-shifted with respect to the v LSR (and not blue-shiftedas in CH CN). Moment maps have so far been madeby using windows of ± . ± . − symmetricfrom the v LSR . The HDCO line profile towards the con-tinuum source position presented in Figure 9 is singlepeaked with a red wing at higher velocities (shadowedregion). In order to discard the red wing as the cause ofthe red-shifted spot, we have derived the velocity struc-ture in ALMA1 by first finding the peak at each pixel(defining a “local” pixel v LSR ) and then assuming dif-ferent windows to make the first moment. The velocitymaps are shown in Figure 10. This approach would be more robust to strong velocity gradients and we have noneed to adopt large windows for making moment mapsthat can introduce noise or contamination from neigh-boring spectral lines. Figure 10 shows that HDCO isred-shifted towards the center of the continuum sourceand increasingly blue-shifted further out, while towardsthe arc shaped structure the emission is blue-shifted asobserved in CH CN. The area of the red-shifted regionincreases when more channels are included and shifts to-wards the south (cf. Figure 10c and d), indicating thatthe lines to the south are skewed towards the red due tohigh-velocity line wings. Overall, the velocity gradientis consistent with what is seen in CH CN and as thewindow for making the moment map increases, the red-shifted spot becomes more evident. Applying a similarreasoning for the origin of the blue-asymmetric/skewedlines (e.g., Zhou et al. 1993, see also the expansion lineprofiles in Keto et al. 2006), this pattern likely indi-cates gas expansion at the core center. Anglada et al.(1987) noted that an inverted profile to that of blue-asymmetric profiles is expected for expansion motionsif the temperature increases towards the central regionand the velocities decrease outwards. The first condition CH CN J = 12 11 K = 7(a) CO (2 1)
40 30 20 10 0 10 20 30 40
Velocity (km s ) I n t e n s i t y ( J y b e a m ) (b) H CO (3
0, 3
0, 2 ) Figure 7.
Examples of absorption line profiles towardsALMA1. The black lines in (a) and (b) present the CO(2 −
1) and H CO (3 , − , ) line emission, respectively. Thegreen lines show the optically thin CH CN J = 12 − K = 7transition. is likely satisfied since a high-mass young stellar objectmay have already been formed and started to ionize thecircumstellar gas. They also assumed in their analysisthat the Sobolev approximation is valid, but the expan-sion velocity of HDCO (see Section 4.3) is comparableto the line width. Hence the second condition may notnecessarily be satisfied.Using cm observations, Avison et al. (2015) deter-mined that an HC H II region, unresolved at their1 . (cid:48)(cid:48) II region is an important evolution-ary stage in the life of high-mass stars. HC H II regionstend to be smaller than 0.03 pc (6000 au) and the likelyculprit of the ionized flow is the photo-evaporation of anaccretion disk surface (Kurtz 2005). While at this stage,it has been suggested that the star can continue grow-ing into earlier types by non-spherical accretion flows(Keto 2007). Eventually it is believed that HC H II re- Figure 8.
Position velocity (pv) map of the CO lineemission towards ALMA1. The position angle of the pvcut is 90 ◦ , i.e. close to perpendicular with respect tothe outflow emission in Figure 5. The red contours corre-spond to − . , − . , . , . , , , × σ rms with σ rms =6 . − . The dashed green line highlights the “C”shape produced by infalling motions.
10 5 0 5 10 15 20
Velocity (km s ) I n t e n s i t y ( J y b e a m ) HDCO 28
5, 23
3, 26
Figure 9.
HDCO (28 , − , ) line profile towards thecontinuum peak position G335–MM1 ALMA1. The dash-dotted lines show the limits used for the moments in Fig-ure 6. The red shadowed region highlights the red high-velocity wing. gions will expand into ultra compact (UC) H II regionsand then into classical H II regions. We suggest thatin ALMA1 we are witnessing the early expansion of theionized gas that is pushing outward the hot moleculargas. The effect of the expansion is more clear in thetransitions tracing the hot, inner molecular core (e.g.,HDCO). The picture we propose for ALMA1 and itsimmediate surroundings is sketched in Figure 11. Thespatial distribution of the emission from the hot transi-tion lines (Figure 6a) is closer to perpendicular to the0 Olguin et al. -48°43'53"54"55" (a) Velocity at peak 2500 au V e l o c i t y ( k m s ) (b) 1 2500 au V e l o c i t y ( k m s ) RA (ICRS) D e c ( I C R S ) (c) 2 2500 au V e l o c i t y ( k m s ) (d) 5 2500 au V e l o c i t y ( k m s ) Figure 10.
HDCO (28 (5 , − (3 , ) velocity maps of G335–MM1 ALMA1. (a) Velocity at the line peak. (b)–(d) Firstmoment map integrated over ± , , × δ from the line peak, respectively. The value of δ is defined as the median line standarddeviation over the region, δ = 2 channels. The green cross marks the position of the continuum source. outflow (P . A . = 210 ◦ ), with position angles between125 ◦ and 130 ◦ as measured from 2-D Gaussian fitted tothe zeroth order moments. Hence, it is likely comingfrom molecular gas in a putative disk surface.4.2. ALMA3
The K = 4 transition first moment map in Figure 4ashows clear signs of rotation towards source ALMA3.We estimate an average position angle of the rotationaxis from the velocity gradient within a region of ra-dius 0.16 (cid:48)(cid:48) (i.e., one beam) centered on the source ofP . A . rot = 292 ± ◦ from the first moment map of the K = 4 transition. A similar value is obtained fromthe K = 7 transition but with a slightly higher errordue to its smaller angular extent (P . A . rot = 290 ± ◦ ).We estimate the kinetic mass of source ALMA3 fromthe velocity extremes from the rotation axis. Assumingthat the source is edge-on, the source mass is in the 10–30 M (cid:12) range (hereafter kinetic mass). Lower inclinationangles, i.e., towards a face-on configuration, would im- CH CN (E up <600K)blue-asymmetry CO (E up =16K)Inverse P-Cygni r<2000 au HDCO (E up =1460K) red-shiftCH CN op. thick i =45º (model) Plane of sky PA=210 º Figure 11.
Diagram of the motions within G335–MM1ALMA1 revealed by the observations and modeling. ply an even larger mass. This is one order of magnitudelarger than the 1 M (cid:12) derived from the dust emission(Table 3). Note that the kinetic mass includes the con- > . M d ∝ T − d ). Different dust opacity laws wouldexplain discrepancies of a factor ∼
2. The lower num-ber of lines detected and lack of radio emission suggestthat this source is younger than ALMA1 and still deeplyembedded. 4.3.
SIMPLE MODELING
In this section, we provide additional support for theinterpretation of the moment maps by comparing qual-itatively the data with simple LTE radiative transfermodels. In these models, the molecular line emissionarises from a spherically symmetric core with a radius of a = 2000 au, comparable to the extent of source ALMA1shown in Figures 3, 6, and 10. We assume that this coreis characterized by a density ρ ( r ) ∝ r − / , and a thermalgradient with temperature T ( r ) ∝ r − . , characteristicof radiative equilibrium under optically thin dust condi-tions (Adams & Shu 1985). We calculated models withsolid body rotation at each radius combined with in-fall or expansion motions. The azimuthal velocity fieldsof the core vary as V φ ∝ r − / . The radial velocitiesare proportional to ± r − / , with positive velocities forexpansion motions and vice versa. This model is a sim-plified version of a pressure-less free-falling core solu-tion dominated by the gravity of a central object (Ulrich1976; Mendoza et al. 2004, 2009). Note that the modelsdo not combine infall and expansion motions, thus theycan only explain the features of one line at a time. Tofit the observations, we optimize numerically the central v LSR and the line width. The remaining parameters arefine-tuned by visual inspection.The blue asymmetry, characteristic of infalling mo-tions, requires a combination of partially optically thickemission and internal heating. Figures 12a and 12bshows the first moment map of the CH CN J = 12 − K = 4 line towards ALMA1 and the one derived fromthe model. We assume that the rotation axis of the coreis inclined with respect to the line of sight by i = 45 ◦ andit forms a P.A.= 5 ◦ , as indicated by the approximate di- rection of the velocity gradient. Note that the angularvelocity of the core and the inclination angle are degen-erate parameters. We are able to reproduce the mainfeatures observed in ALMA1 with an infall velocity atthe external radius a of V in ( a ) = − .
15 km s − , an angu-lar velocity Ω( a ) sin( i ) = 3 . × − s − , a temperature T ( a ) = 33 K, and a line absorption coefficient given by κ v ( a ) = 3 . × − φ v cm − . The line profile φ v is as-sumed Gaussian with a FWHM of ∆ v = 2 km s − . Wenote that, in agreement with infall, the rotation of thecore is not enough to maintain the core in equilibriumat the assumed inclination.The gas mass within a radius R from the model isgiven by M ( < R ) = 8 π ρ ( a )( Ra ) / . (3)From the line absorption coefficient and assuming aCH CN abundance of 10 − (e.g., Hern´andez-Hern´andezet al. 2014), we obtain ρ ( a ) = 2 . × − g cm − . Thegas mass of the model at R = 710 au (same size derivedfrom the dust continuum, see Table 3) is M ( <
710 au) =6 . (cid:12) . We note however that the abundance of CH CNis very uncertain and can vary by one order of magnitude(Hern´andez-Hern´andez et al. 2014), precluding a directcomparison with the mass derived from dust continuumemission.Figures 12c and 12d show the maps of the HDCOvelocity at peak line intensity obtained from the ob-servations and a rotating and expanding core model.The expansion velocity is V exp ( a ) = +0 . − . Twofeatures describe the shape of the HDCO line profile:1) a relatively slow expansion which red-shifts the linepeak in (cid:46) − in the directions of highest opacity,that is, toward the center of the core (see Figure 10a);2) a distinct red-shifted wing, typical of outflows (seeFigure 9). A combination of these two effects producethe red-shifted velocities shown in the first moment map(Figure 10d). We find that a model with an absorptioncoefficient 4 times less than that of CH CN and a linewidth ∆ v = 2 . − is able to reproduce fairly wellthese features of the HDCO line, particularly the red-shifting of the peak toward the center of the core. Thisrotation-expansion model also has different P.A.= 20 ◦ compared to the CH CN model, which is suggested bythe velocity gradient shown in Figure 10. Models withthe same P.A. produce fits that are not significantlyworse.As discussed in Section 4.1, the expansion of the gasis likely caused by the expansion of the HC H II (seeFigure 11). This may have already grown beyond thegravitational radius of the young star (Sartorio et al.2019), which is 54 au for a 10 M (cid:12) HMYSO, at in the2
Olguin et al. -48°43'53"54"55" (a) CH CN K = 4 Observed
Model V e l o c i t y ( k m s ) RA (ICRS) D e c ( I C R S ) (c) HDCO 2500 au (d) Expansion + rotation2500 au L i n e p e a k v e l o c i t y ( k m s ) Figure 12.
Modeling of ALMA1 CH CN J = 12 − K = 4 and HDCO observations. (a) and (c) observed CH3CN K = 4moment 1 and line peak intensity of HDCO. (b) Model of CH CN K = 4 combining rotation and infall of a spherically symmetriccore. (d) Model of the line peak intensity of HDCO by a spherically symmetric core with expansion and rotation gas motions.The beam is shown in the lower left corner of panel (a). τ = 1 cross the coremuch closer to its center compared to those of CH CN.Indeed, while for CH CN the lines of sight with impactparameter of 130 au are optically thick, for HDCO, opti-cally thick lines of sight are those that pass within 30 auof the center. This difference in opacities explains whywe are not able to see the expansion signature in theCH CN profiles and why we cannot discern the infallsignature in the HDCO line. In the former case, the ex-pansion is hidden within a very small radius associatedwith a large opacity. On the other hand, HDCO emis-sion from gas in expansion is associated with opticallythin emission, and therefore the blue asymmetry doesnot arise. CONCLUSIONSWe observed the high-mass star-forming regionG335.579–0.292, in particular the core G335–MM1, at0 . (cid:48)(cid:48) ∼ − (cid:48)(cid:48) resolution, while the other radio continuumsource towards this region possibly arises from a jet asso-ciated with ALMA3. From the study of the kinematicsin these two sources, we conclude that they are likely toform or have already formed at least one high-mass star.Line emission was not detected in the remaining threecontinuum sources, and were thus not studied in detail.ALMA1 has a complex kinematic structure. We ob-serve large scale infalling motions from CO inverse P-Cygni profiles, while CH CN blue asymmetric profilesindicate infall motions at smaller scales. The overallCH CN velocity gradient may be indicative of rotationof the circumstellar material. This velocity gradient isroughly perpendicular to the outflow direction observedfrom CO and SiO emission. Finally, lines tracing hotmolecular gas, HDCO and (CH ) CO, show an expan-sion velocity pattern in their moment 1 map, which maybe the result of photoevaporation of the surface of amolecular disk due to the ionizing radiation of the HCH II region. To support these hypotheses, we model therotating infall and expansion motions with a sphericallysymmetric envelope. We conclude that the expansionmotion observed in hot lines is indicative of reversing ofthe accretion flow in a region smaller than the one tracedby the rotating infall seen in CH CN. Higher angularresolution observations will reveal the scales at whichexpansion dominates and whether or not the complexkinematics towards this source can also be the result ofunresolved sources. In ALMA3, CH CN line emission shows clear evidenceof rotation, which imply a total mass for the source of10–30M (cid:12) assuming Keplerian rotation.The nature of the remaining sources (ALMA2,ALMA3, and ALMA5) remains unclear due to their lackof line emission. They likely are prestellar cores thatmay or may not form high-mass stars.ACKNOWLEDGMENTSThe authors would like to thank the anonymousreferee for the insightful comments. F.O. and H.-R.V.C. acknowledge the support of the Ministry ofScience and Technology of Taiwan, project no. 109-2112-M-007-008-. P.S. was partially supported bya Grant-in-Aid for Scientific Research (KAKENHINumber 18H01259) of Japan Society for the Promo-tion of Science (JSPS). Data analysis was in partcarried out on the Multi-wavelength Data Analy-sis System operated by the Astronomy Data Center(ADC), National Astronomical Observatory of Japan.This paper makes use of the following ALMA data:ADS/JAO.ALMA
Facility:
ALMA
Software: astropy (Astropy Collaboration et al.2013; Price-Whelan et al. 2018), CASA (McMullin et al.2007), GoContinuum (Olguin & Sanhueza 2020), mat-plotlib (Hunter 2007), numpy (Harris et al. 2020), scipy(Virtanen et al. 2020), YCLEAN (Contreras et al. 2018;Contreras 2018)4
Olguin et al.
APPENDIX A. CONTINUUM SUBTRACTIONTo obtain the line-free channels, we:1. Compute a dirty data cube for each spectral window in the observations.2. Compute a map with the maximum value along the spectral axis (hereafter maximum map) for each spectralwindow.3. Obtain the peak position of the maximum map for each spectral window.4. Average the values of the peak positions. We rejected the position which is further from the centroid to avoidoutliers.5. Obtain the spectrum at the averaged peak position from each spectral window data cube.6. Use asymmetric sigma clipping to obtain channels free of line contamination.7. Recover bands of channels rejected by the asymmetric sigma clipping that span less than two channels (onespectral resolution).In addition, 10 channels at each end of the spectral windows were not used for continuum calculations as those channelstend to be noisier.A corrected symmetric sigma clipping has been implemented by S´anchez-Monge et al. (2018) to subtract the con-tinuum from data cubes. Their correction of the symmetric sigma clipping is based on the image noise, which cannotbe calculated without having to CLEAN the data cube first. Based on their approach, we found that an asymmetricsigma clipping can obtain similar results without applying their correction to the symmetric one. Considering thatour spectra are dominated by line emission we found that an acceptance range of − . . σ i , with σ i the standarddeviation at iteration i of the clipping algorithm, is enough to obtain similar results than in S´anchez-Monge et al.(2018, their default range is ± . σ i ). What our method provides at the end is the list of channels, in CASA format,to be used for making both the continuum subtracted line cubes and the continuum image free of line contamination.REFERENCES Adams, F. C., & Shu, F. H. 1985, ApJ, 296, 655,doi: 10.1086/163483Ahmadi, A., Kuiper, R., & Beuther, H. 2019, A&A, 632,A50, doi: 10.1051/0004-6361/201935783Anglada, G., Rodriguez, L. F., Canto, J., Estalella, R., &Lopez, R. 1987, A&A, 186, 280Astropy Collaboration, Robitaille, T. P., Tollerud, E. J.,et al. 2013, A&A, 558, A33,doi: 10.1051/0004-6361/201322068Avison, A., Peretto, N., Fuller, G. A., et al. 2015, A&A,577, A30, doi: 10.1051/0004-6361/201425041Beuther, H., Walsh, A. J., Johnston, K. G., et al. 2017,A&A, 603, A10, doi: 10.1051/0004-6361/201630126Breen, S. L., Caswell, J. L., Ellingsen, S. P., & Phillips,C. J. 2010, MNRAS, 406, 1487,doi: 10.1111/j.1365-2966.2010.16791.x Bronfman, L., Nyman, L. A., & May, J. 1996, A&AS, 115,81Caswell, J. L., Fuller, G. A., Green, J. A., et al. 2011,MNRAS, 417, 1964,doi: 10.1111/j.1365-2966.2011.19383.xChambers, E. T., Jackson, J. M., Rathborne, J. M., &Simon, R. 2009, ApJS, 181, 360,doi: 10.1088/0067-0049/181/2/360Chen, H.-R. V., Zhang, Q., Wright, M. C. H., et al. 2019,ApJ, 875, 24, doi: 10.3847/1538-4357/ab0f3eContreras, Y. 2018, Automatic Line Clean, 1.0, Zenodo,doi: 10.5281/zenodo.1216881Contreras, Y., Sanhueza, P., Jackson, J. M., et al. 2018,ApJ, 861, 14, doi: 10.3847/1538-4357/aac2ecEstalella, R., Anglada, G., D´ıaz-Rodr´ıguez, A. K., &Mayen-Gijon, J. M. 2019, A&A, 626, A84,doi: 10.1051/0004-6361/20183499815