Detection of a hot molecular core in the Large Magellanic Cloud with ALMA
Takashi Shimonishi, Takashi Onaka, Akiko Kawamura, Yuri Aikawa
aa r X i v : . [ a s t r o - ph . GA ] J un Detection of a hot molecular core in the Large Magellanic Cloud with ALMA
Takashi Shimonishi , , ⋆ Takashi Onaka Akiko Kawamura and Yuri Aikawa ABSTRACT
We report the first detection of a hot molecular core outside our Galaxy based on radio observationswith ALMA toward a high-mass young stellar object (YSO) in a nearby low metallicity galaxy, the LargeMagellanic Cloud (LMC). Molecular emission lines of CO, C O, HCO + , H CO + , H CO, NO, SiO,H CS, SO, SO , SO , and SO are detected from a compact region ( ∼ and SO lines. The compact source size, warm gas temperature, highdensity, and rich molecular lines around a high-mass protostar suggest that ST11 is associated with a hotmolecular core. We find that the molecular abundances of the LMC hot core are significantly di ff erentfrom those of Galactic hot cores. The abundances of CH OH, H CO, and HNCO are remarkably lowercompared with Galactic hot cores by at least 1–3 orders of magnitude. We suggest that these abundancesare characterized by the deficiency of molecules whose formation requires the hydrogenation of CO ongrain surfaces. In contrast, NO shows a high abundance in ST11 despite the notably low abundance ofnitrogen in the LMC. A multitude of SO and its isotopologue line detections in ST11 imply that SO can be a key molecular tracer of hot core chemistry in metal-poor environments. Furthermore, we findmolecular outflows around the hot core, which is the second detection of an extragalactic protostellaroutflow. In this paper, we discuss physical and chemical characteristics of a hot molecular core in the lowmetallicity environment. Subject headings: astrochemistry – ISM: abundances – ISM: molecules – circumstellar matter – Magellanic Clouds– radio lines: ISM Frontier Research Institute for Interdisciplinary Sciences, To-hoku University, Aramakiazaaoba 6-3, Aoba-ku, Sendai, Miyagi,980-8578, Japan Astronomical Institute, Tohoku University, Aramakiazaaoba 6-3, Aoba-ku, Sendai, Miyagi, 980-8578, Japan Department of Astronomy, Graduate School of Science, TheUniversity of Tokyo, 7-3-1 Hongo, Bunkyo-ku, Tokyo 113-0033,Japan National Astronomical Observatory of Japan, 2-21-1 Osawa,Mitaka, Tokyo, 181-8588, Japan Center for Computational Sciences, The University of Tsukuba,
1. Introduction
Because cosmic metallicity is increasing in timewith the evolution of our universe, interstellar chem-istry in low metallicity environments is crucial to un-derstand chemical processes in the past universe. Forthis purpose, observations of chemically-rich objectsin nearby low metallicity galaxies and comparative ⋆ [email protected] ≤ ≥ cm − ), and warm gas / dust temperature( ≥
100 K) (e.g., Kurtz et al. 2000; van der Tak 2004).Chemistry of hot cores is characterized by sublimationof ice mantles, which accumulated in the course ofstar formation. In cold molecular clouds and prestellarcores, gaseous molecules and atoms are frozen ontodust grains and hydrogenated. As the core is heatedby star-formation activities, reaction among heavyspecies become active on grain surfaces to form largermolecules. In addition, sublimated molecules, suchas CH OH and NH , are subject to further gas-phasereactions (e.g., Garrod & Herbst 2006; Garrod et al.2008; Herbst & van Dishoeck 2009). As a result, hotcores show a wealth of molecular spectral lines in in-frared and radio wavelengths. Thus detailed studiesof chemical properties of hot cores are crucial to un-derstand complex chemical processes triggered by starformation.The Large Magellanic Cloud (LMC) is an excel-lent target to study interstellar and circumstellar chem-istry in di ff erent metallicity environments owing to itsproximity (49.97 ± ff ective in the LMC than in our Galaxy(Israel et al. 1986). Furthermore, according to gamma-ray observations, the cosmic-ray density in the LMC isestimated to be lower than the Galactic typical valuesby a factor of four (Abdo et al. 2010). It is thereforehighly anticipated that these environmental di ff erencesshould a ff ect chemical processes, and hot cores in theLMC should provide us key information to understandchemistry, particularly those of complex molecules, inlow metallicity environments. However, so far, ob-servations of hot cores have been limited to Galacticsources due to lack of spatial resolution and sensitivityof radio telescopes.Most of radio studies on chemical compositionsof molecular gas in the LMC have been performedwith single-dish telescopes. Early studies by theSEST 15 m telescope conducted multiline obser- vations toward H II regions in the LMC and de-tected molecular species as large as CH OH, C H and SO (Johansson et al. 1994; Chin et al. 1997;Heikkil¨a et al. 1999; Wang et al. 2009). Submillime-ter observations of relatively dense molecular gas instar-forming regions in the LMC are also reported(Paron et al. 2014, 2016). Nishimura et al. (2016) re-cently conducted deep and unbiased spectral line sur-veys in the 3 mm window toward a number of molec-ular clouds in the LMC using the Mopra telescope.They reported the low abundances of nitrogen-bearingmolecules, the deficiency of CH OH, and the highabundance of C H in the LMC compared with Galac-tic molecular clouds.Characteristic interstellar chemistry in the LMC isalso suggested from previous infrared observations ofices around embedded young stellar objects (YSOs) inthe LMC. Shimonishi et al. (2008, 2010) reported thatthe CO / H O ice ratio of high-mass YSOs in the LMCis systematically higher than those of Galactic high-mass YSOs based on infrared observations with the
AKARI satellite. Recently, Shimonishi et al. (2016) re-ported that the CH OH ice around high-mass YSOsin the LMC is less abundant compared with Galac-tic counterparts based on infrared observations withthe Very Large Telescope. The authors suggest thatwarm ice chemistry (grain surface reactions at a rel-atively high dust temperature) is responsible for theobserved characteristics of ice chemical compositionsin the LMC. Furthermore, detailed studies of the 15.2 µ m CO ice band toward LMC high-mass YSOs with Spitzer suggest a higher degree of thermal processingof ices in the LMC than in our Galaxy (Oliveira et al.2009; Seale et al. 2011). Since gas-grain chemistry isbelieved to play an important role in hot cores, thecharacteristic ice chemistry in the LMC would implydiverse hot core chemistry in extragalactic environ-ments according to their metallicities.High-spatial resolution interferometry observa-tions toward star-forming regions in the LMC havebeen reported with the Australia Telescope Com-pact Array (e.g., Wong et al. 2006; Ott et al. 2008;Seale et al. 2012; Anderson et al. 2014) and recentlywith the Atacama Large Millimeter / submillimeter Ar-ray (ALMA) (e.g., Indebetouw et al. 2013; Fukui et al.2015). These observations resolve star-forming re-gions in the LMC down to parsec- or sub-parsec-scaleand investigate physical properties of dense moleculargas. However, chemical properties of warm and densemolecular gas associated with a single high-mass YSO2n the LMC still remain to be investigated.In this paper, we report the detection of a hot molec-ular core in the LMC based on submillimeter inter-ferometric observations with ALMA. In § §
3. Analysis of spectral line data and deriva-tion of physical quantities of molecular gas and dustare described in §
4. Physical and chemical propertiesof the observed source are discussed in §
5. Finally,conclusions of this paper are summarized in §
2. Observations and data reduction2.1. Target
The target of the present observations is a high-mass YSO, 2MASS J05264658-6848469 or ST11(hereafter ST11), located in the LMC. The sourceis spectroscopically identified to be a high-massYSO in previous infrared studies (Seale et al. 2009;Shimonishi et al. 2010). Detailed YSO properties ofST11 are revisited in this work and discussed in § Observations were carried out with ALMA be-tween November 2013 and February 2014 as a partof the Cycle 1 high priority program 2012.1.01108.S(PI T. Shimonishi). The telescopes were pointedto RA = h m . s
63 and Dec = -68 ◦ ′ . ′′ ′′ at these frequencies,which corresponds to the field-of-view (FoV) of imag-ing data. The maximum recoverable angular scale isabout 7 ′′ . The velocity resolution of the original datais 0.4 km / s for the spectral bands and 26 km / s for thecontinuum band. Raw interferometric data is processed using the
Common Astronomy Software Applications (CASA)package. The calibration is done by CASA 4.1.0 andimaging as well as spectral extraction are done byCASA 4.3.0. The flux calibrator is J0519-454 and thephase calibrators are J0601-7036 and J0635-7516. Thesynthesized beam size in the 338 GHz region is ap-proximately 0.5 ′′ × ′′ , which corresponds to 0.12pc at the assumed distance to the LMC (49.97 kpc,Pietrzy´nski et al. 2013). The primary beam correctionis done by the impbcor task in CASA, but the correc-tion have little e ff ect on the extracted spectra since thetarget source is located at the center of FoV and verycompact.The spectra as well as continuum flux are extractedfrom the circular region with a diameter of 0.5 ′′ cen-tered at RA = h m . s
60 and Dec = -68 ◦ ′ . ′′ µ m) continuum emission ofST11 measured in this study, and the diameter corre-sponds to the beam size. The continuum emission issubtracted from the spectral data using the uvcontsubtask in CASA. Several channels are concatenated dur-ing the clean process to increase S / N and the channelspacing of the reduced data is 1.5 km / s (1.73 MHz) ex-cept for CO(3–2). For the CO(3–2) line, the spectralregion is not binned since the line is su ffi ciently strong,and the channel spacing is 0.4 km / s (0.46 MHz).
3. Results3.1. Observed spectra
Figures 1–2 show the spectra extracted from the0.5 ′′ diameter region centered at ST11. In the fig-ures the sky frequency is converted to the rest fre-quency using the LSR velocity of 250.5 km / s, which isthe typical radial velocity of molecular lines detectedtoward the source. Spectral lines are identified withthe aid of the Cologne Database for Molecular Spec-troscopy (CDMS, M¨uller et al. 2001, 2005) and themolecular database of the Jet Propulsion Laboratory (JPL, Pickett et al. 1998). Molecular emission linesdue to CO, C O, HCO + , H CO + , H CO, NO, SiO,H CS, SO, SO , SO , and SO are detectedfrom a compact region associated with ST11. A num- https: // / cdms http: // spec.jpl.nasa.gov m b [ K ] ! T m b [ K ] ! T m b [ K ] ! T m b [ K ] ! Rest Frequency [GHz] ! S O ! H C O + ! C O ! S O ! S O ! S O ! " S O ! " S O ! U ! S O ! " S O ! " S O ! " S O ! " S O ! " S O ! " S O ! C O ! S O ! S i O ! U ! " S O ! U ! U ! " S O ! H C S ! Fig. 1.— ALMA band 7 spectra of ST11 extracted from the central region with a radius of 0.5 ′′ . Detected emissionlines are labeled. Tentative detections are indicated by “ ? ” and unidentified lines are by “ U ”. The adopted sourcevelocity is 250.5 km / s. 4 m b [ K ] ! T m b [ K ] ! T m b [ K ] ! Rest Frequency [GHz] ! N O ! N O ! S O ! " S O ! " S O ! U ! S O ! H C O ! S O ! S O ! S O ! S O ! D C O ? ! H C O + ! S O ! S O ! S O ! S O ! U ! U ! U ! S O ν ? ! U ! H N C O ? ! Fig. 2.—
Continued ber of high excitation lines ( E u >
100 K) are detectedfor SO and its isotopologues. The emission lines ofthe above molecules in the 345 GHz band are for thefirst time detected toward the LMC source except forCO and HCO + . Emission lines of CH OH, HNCO,CS, HC N, and complex organic molecules are not de-tected. Unidentified lines are labeled in the figures, butsome of them may be spurious signals.
Figures 3–4 show synthesized images of continuumand molecular emission lines observed toward ST11.For SO and SO , we only show representative linesbecause distributions of emission are similar in di ff er-ent transitions. The lines of SO ( ν = SO , andH CS are not included in the figure because they aretoo weak to visualize the brightness distribution. Theimages are constructed by integrating each spectrum5ig. 3.— Flux distribution of the ALMA 840 µ m continuum data tracing cold dust (left) and the Gemini / T-ReCSmid-infrared 10 µ m image tracing warm dust (right).in the velocity range where the emission line is seen,typically between 240 km / s and 260 km / s. For CO, thespectrum is integrated between 230 km / s and 285 km / sbecause the line is very broad.A high spatial resolution mid-infrared image ofST11 is also shown in Figure 3. The image is obtainedby T-ReCS at the Gemini South telescope (ProgramID: S10B-120, PI: T. Shimonishi) and a broad-bandfilter in the N-band (7.70–12.97 µ m, centered at 10.36 µ m) is used for the observation. The ALMA 840 µ mcontinuum traces distribution of cold dust, while themid-infrared image traces warm dust.The source is compact in general and peak positionsof each emission line coincide with the region of dustcontinuum emission. We estimate a source size (Gaus-sian FWHM, θ ) for continuum and relatively strongemission lines by a two-dimensional Gaussian fit. Theestimated θ in the 840 µ m continuum image is about0.6 ′′ , which is close to the beam size. The mid-infraredemission shows θ = ′′ , which is as compact as the840 µ m continuum. The molecular lines of NO, SiO, SO, SO , and SO typically have θ ∼ ′′ , which isindistinguishable from the beam size. Since the distri-bution is as compact as the beam size, these emissionsare considered to be a point source with the presentspatial resolution. The H CO + and H CO lines areslightly extended compared with the above lines andhave θ ∼ ′′ . The lines of CO, C O and HCO + aremore extended and θ is 1.1 ′′ –1.4 ′′ . Because the sourcesize is su ffi ciently smaller than the maximum recover-able angular scale of ∼ ′′ , the emission from ST11 isalmost recovered by the present interferometric obser-vations.
4. Analysis4.1. Spectral fitting
The line parameters are measured by fitting a sin-gle Gaussian profile to the observed lines. For the NOlines at 351.0435 GHz and 351.0515 GHz, we fit adouble Gaussian because they are partially blended. Insome cases we subtract a local baseline, which is es-timated from adjacent line-free regions to correct forweak baseline ripples. For the SO (10 , –10 , ) line,we subtracted the HCO + (4–3) line upon fitting sincethey are partly blended. In general, good fits are ob-tained with Gaussian profiles except for the CO(3–2)line, which deviates from a Gaussian. We estimate apeak main-beam brightness temperature, a FWHM, aLSR velocity, and an integrated intensity for each lineon the basis of the fitting. For CO, instead, we estimatethe peak brightness temperature and the FWHM byvisual inspection, and the integrated intensity is esti-mated by integrating the spectrum in the velocity rangebetween 230 km / s and 285 km / s. The spectra and theresults of the gaussian fitting are shown Figures 5–8.The figures also show the spectral regions of severalimportant non-detection lines. The measured line pa-rameters are summarized in Table 1.Multiple hyperfine components sometime existwithin fitted profile according to the spectroscopic cat-alogues (CDMS or JPL), but these are not resolved dueto the low spectral resolution of the present spectra.When the blended lines have comparable upper stateenergies, we split the measured flux according to thetheir S µ values and the upper state degeneracy (see § S and µ ). Using this method,6ig. 4.— Integrated intensity distributions of CO, C O, HCO + , H CO + , H CO, SiO, NO, SO, SO , and SO lines. Contours represent the distribution of the 840 µ m continuum, and the contour levels are 25 %, 50 % and 75 % ofthe peak flux. The synthesized beam size (0.5 ′′ , 0.12 pc at the LMC) is shown by the gray filled circle in each panel.we estimate the line parameters of the strongest hyper-fine line, which is used in the subsequent analysis ofcolumn densities and rotational temperatures. , SO and SO Since we detect multiple emission lines with dif-ferent excitation energies for SO , SO and SO ,we perform the rotation diagram analysis assuming anoptically thin condition and an local thermodynamicequilibrium (LTE). A column density of molecules inthe upper energy level, N thinu , is derived by the follow-ing equation for optically thin lines (e.g., Sutton et al. 1995; Goldsmith & Langer 1999), N thinu g u = k R T mb dV π ν S µ , (1)where g u is the degeneracy of the upper level, k is theBoltzmann constant, R T mb dV is the integrated inten-sity as estimated from the observations, ν is the tran-sition frequency, S is the line strength, and µ is thedipole moment. Under the LTE condition, the totalcolumn density, N total , is given by N thinu g u = N total Q ( T rot ) e − E u / kT rot , (2)7ig. 5.— Spectra of CO, C O, HCO + , H CO + , H CO, SiO, NO, H CS, and SO emission lines extracted from the0.5 ′′ diameter region centered at ST11. The blue lines represent Gaussian profiles fitted to the observed spectra. Thespectral regions of important non-detection lines including CH OH, HNCO, C S, CH OCH , C H OH, HC N, andHCOOCH are also shown. 8ig. 6.— Spectra of SO emission lines extracted from the 0.5 ′′ diameter region centered at ST11. The blue linesrepresent a Gaussian profile fitted to the observed spectra. The spectra are sorted in ascending order of the upper stateenergy (The emission line with the lowest upper state energy is shown in the upper left panel and that with the highestenergy is in the lower right panel). The bottom panel is for the SO ( ν =
1) line.where Q ( T rot ) is the partition function, T rot is the ro-tational temperature, and E u is the upper state energy.This equation is rearranged as follows.log N thinu g u ! = − log eT rot ! (cid:18) E u k (cid:19) + log N total Q ( T rot ) ! (3)When N thinu / g u is plotted against E u / k and datapoints are fitted by a straight-line, the slope and the in-tercept correspond to T rot and N total , respectively. Thuswe can simultaneously determine the rotational tem-perature and the total column density. All the spectro- scopic parameters required in the above analysis areextracted from the CDMS or the JPL database. For thepartition function, we interpolate the data given in thedatabases and estimate the appropriate Q ( T rot ) at thederived rotational temperature.The constructed rotation diagrams for SO , SO and SO are shown in Figure 9 and the derived tem-peratures and column densities are summarized in Ta-ble 2. Uncertainties in the table are of 2 σ level and donot include systematic errors due to spectroscopic pa-rameters extracted from the CDMS and JPL databases.9ig. 7.— Spectra of SO emission lines observed toward ST11 as in Figure 6. The spectra are sorted in ascendingorder of the upper state energy from the upper left to the lower right.For a comparison purpose, we also perform therotation diagram analysis for the Orion hot core us-ing the lines with the similar spectroscopic propertieswith those used in the analysis of ST11. The Oriondata is obtained with the 20 ′′ beam size ( ∼ and SO measuredfor W3 (H O) and G34.3 + Column densities of other molecular species thanSO and its isotopologues are derived by solving Eq.2 for N total under assumption of the LTE and opti-cally thin condition. We here assume that the observedmolecular species are located in the same region as10ig. 8.— Spectra of SO emission lines observed toward ST11 as in Figure 6. The spectra are sorted in ascendingorder of the upper state energy from the upper left to the lower right. The 17 , –17 , line is a tentative detection.SO and its isotopologues, and thus have the similarrotational temperatures; T rot is assumed to be 100 K,which is roughly the average rotational temperature ofSO , SO and SO .The similar spatial distributions of emission and thesimilar radial velocities of SO and other moleculessupport the validity of this assumption, but more rig-orous excitation analysis with further spectral data isabsolutely necessary in the future.NO shows multiple transitions with the same upperstate energy in the 350–351 GHz region. We estimatethe column density for each transition, and the aver-age value is adopted as the final NO column density.The scatter of column densities estimated from di ff er-ent lines is less than 20 %.We also estimate upper limits on column densi-ties of important non-detection lines such as CH OH,HNCO, C S and some complex molecules (see § . / s to estimate theupper limits on integrated intensities. This FWHMis consistent with the typical velocity width of othermolecular lines except for CO and HCO + .The derived column densities and upper limits (2 σ level) are summarized in Table 2. and total gas mass A column density of molecular hydrogen, whichusually dominate the total mass of embedded sources,is estimated using the dust continuum emission dataobtained in our observations. The flux density of dustcontinuum, F ν , at the optically thin frequency, ν , is ex-pressed as F ν = Ω τ ν B ν ( T d ) , (4)where Ω is a beam solid angle, τ ν is an optical depth, B ν ( T d ) is the Planck function and T d is a dust temper-ature (Whittet 1992). The optical depth is expressedas τ ν = ρ d κ ν L , (5)where ρ d is a mass density of dust, κ ν is a mass absorp-tion coe ffi cient, and L is a path length. We use the massabsorption coe ffi cient of dust grains coated by thin icemantles as presented in Ossenkopf & Henning (1994).Using the dust-to-gas mass ratio, Z , the mass densityof dust is expressed as ρ d = Z µρ H = Z µ N H m H / L , (6)where µ is a mean atomic mass per hydrogen, ρ H isa mass density of hydrogen, N H is a column densityof hydrogen, and m H is a hydrogen mass. We hereassume µ to be 1.41 according to Cox (2000). Weassume that the dust-to-gas mass ratio in the LMC is11 able ine P arameters Molecule Transition Eu / k Frequency T mb ∆ V R T mb dV V LSR
RMS I timea Note(K) (GHz) (K) (km / s) (K km / s) (km / s) (K) (s)CO 3–2 33 345.7960 ∼ ∼ ∼ ∼
252 0.42 ‡ O 3–2 32 337.0611 2.30 ± ± † HCO + ± ± ‡ H CO + ± ± ‡ NO 7 / / / / f
36 350.6895 1.63 ± ± † / / / / f
36 350.69087 / / / / f
36 350.6948NO 7 / / / / e
36 351.0435 0.85 ± ± † NO 7 / / / / e
36 351.0515 0.71 ± ± † / / / / e
36 351.0517H CO 5 , –4 ,
62 351.7686 2.48 ± ± † CH OH 7 –6 A +
65 338.4087 < · · · < · · · † HNCO 16 , –15 ,
143 351.6333 < · · · < · · · † SiO 8–7 75 347.3306 1.64 ± ± ‡ C S 7–6 65 337.3965 < · · · < · · · † H CS 10 , –9 ,
102 338.0832 0.44 ± ± † CH OCH , –7 , AE 55 356.5753 < · · · < · · · ‡ , –7 , EE 55 356.5760C H OH 10 , –9 ,
66 357.0674 < · · · < · · · ‡ SO 8 –7
81 337.1986 2.84 ± ± † N 38–37 323 345.6090 < · · · < · · · ‡ HCOOCH , –8 , A 80 345.7187 < · · · < · · · ‡ , –8 , A 80 345.7187SO , –17 ,
245 336.6696 1.87 ± ± † SO , –18 ,
197 338.3060 5.25 ± ± † SO , –19 ,
199 338.6118 6.39 ± ± † SO , –27 ,
521 345.4490 0.76 ± ± ‡ SO , –18 ,
168 346.6522 7.99 ± ± ‡ SO , –11 ,
139 350.8628 2.43 ± ± † SO , –14 ,
136 351.8739 6.62 ± ± † SO , –10 ,
90 356.7552 6.19 ± ± ‡ , –13 ,
123 357.1654 6.03 ± ± ‡ SO ν = , –4 ,
781 357.0872 0.50 ± ± ‡ SO , –4 ,
35 346.5901 0.57 ± ± ‡ SO , –13 ,
122 350.7881 0.42 ± ± † SO , –15 ,
149 350.9146 0.40 ± ± † SO , –11 ,
99 350.9951 0.64 ± ± † SO , –5 ,
52 351.6351 0.48 ± ± † SO , –17 ,
179 351.7449 < · · · < · · · † SO , –12 ,
92 338.3204 0.87 ± ± † SO , –14 ,
134 338.7857 0.98 ± ± † SO , –7 ,
64 345.5197 1.01 ± ± ‡ SO , –6 ,
57 345.5531 0.67 ± ± ‡ SO , –5 ,
52 345.6513 0.61 ± ± ‡ SO , –4 ,
47 345.6788 0.64 ± ± ‡ SO , –17 ,
179 345.9293 0.59 ± ± ‡ SO , –28 ,
391 347.4831 0.23 ± ± ‡ SO , –21 ,
250 352.0829 0.68 ± ± † SO , –19 ,
185 357.1022 1.32 ± ± ‡ N ote .— Uncertainties and upper limits are of 2 σ level and do not include systematic errors due to baseline subtraction and adopted spectroscopicconstants. a Total on-source integration time, where † represents 575 seconds and ‡ + (4–3), which issubtracted when fitting this SO line. (6) Blend with seven hyperfine components. (7) Blend with ten hyperfine components. SO (upper left), SO (upper right), and SO (lower left) for ST11. Thefilled squares (black) are for ST11 and the open squares (green) are for the Orion hot core. The downward trianglesrepresent upper limits. The straight-lines fitted to the ST11 and the Orion data points are shown by the black and greensolid lines, respectively. The derived rotational temperatures are shown in Table 3.13 able stimated column densities Molecule T rot N (X)(K) (cm − )SO ∗ ± × ∗ SO ∗ ± × ∗ SO ∗ ± × ∗ H – 4.5 × CO 100 ∼ × C O 100 1.2 ± × HCO +
100 1.4 ± × H CO +
100 6.9 ± × H CO 100 1.0 ± × CH OH 100 < × NO 100 9.1 ± × HNCO 100 < × HC N 100 < × SiO 100 1.5 ± × C S 100 < × H CS 100 2.8 ± × SO 100 2.7 ± × CH OCH < × HCOOCH < × C H OH 100 < × N ote .—Uncertainties and upper limits areof 2 σ level and do not include systematic er-rors due to adopted spectroscopic constants. ∗ Derived based on the rotation diagram anal-ysis (see § T able omparison of rotational temperatures T rot [K]Molecule ST11 a Orion-KL b W3 (H O) c G34.3 + d SO ± ± ∗
184 108, 284 ∗∗ SO ± ± ∗
179 131 SO ±
14 104 (73 ± ∗ · · · · · · N ote .—Uncertainties are of 2 σ level. ∗ Numbers in parentheses in the Orion data are derived in this work (see § ∗∗ Two temperature components.References. — a This work; b Schilke et al. (1997); c Helmich & van Dishoeck (1997); d MacDonald et al. (1996) Z = ( N H = N H / N H = F ν / Ω κ ν B ν ( T d ) Z µ m H . (7)We measure the flux density per beam solid an-gle, F ν / Ω , using the 840 µ m dust continuum image ofST11 shown in Figure 3. With the aperture size of 0.5 ′′ in diameter, the F ν / Ω is measured to be 0.34 Jy / beam.For dust temperature, we assume T d =
40 K, whichis a typical dust temperature of high-mass YSOs inthe LMC estimated based on far-infrared observations(van Loon et al. 2010). The assumed dust temperatureis consistent with the far-infrared SED peak of ST11(see § N H = × cm − . This H column density corresponds to thetotal gas mass of 115 M ⊙ in the line of sight. The esti-mated dust opacity at 840 µ m is τ µ m = A v) of ∼
150 mag.We also estimate a lower limit on the gas densityaround ST11. We assume that gas is spherically dis-tributed around a protostar with a radius which is thesame as the beam size. With this assumption, the totalgas mass around ST11 derived above corresponds tothe H number density of 2 × cm − . We empha-size that this is a lower limit because the H densityincreases inversely with the assumed source size. The SED of ST11 is shown in Figure 10. Details ofthe collected data are summarized in Table 4. Most ofthe energy is emitted in the mid- to far-infrared wave-length regions and the peak of the SED is between 60 µ m and 70 µ m, which is consistent with the character-istics of high-mass YSOs.The bolometric luminosity of ST11 is estimatedto be 5 × L ⊙ , which is derived by integrat-ing the interpolated SED from 1 µ m to 1000 µ m.About 60 % of the total luminosity is emitted inin the 30–100 µ m wavelength region. The stellarmass of ST11 is estimated using the Online SEDFitter (Robitaille et al. 2007). As the input data http: // caravan.astro.wisc.edu / protostars / sedfitter.php of the SED fit, we use 2–840 µ m photometric andspectroscopic data, which are obtained from theIRSF / SIRIUS,
Spitzer
SAGE,
AKARI
LSLMC, and
Herschel
HERITAGE databases (Kato et al. 2007;Meixner et al. 2006; Kemper et al. 2010; Kato et al.2012; Shimonishi et al. 2010; Meixner et al. 2013).We also use the results of our mid-infrared narrow-band filter photometry (centered at 7.73, 8.74, 9,69,10.38, 11.66, and 12.33 µ m) and spectroscopy (8–12 µ m) conducted with T-ReCS at the Gemini South tele-scope. The 840 µ m flux is estimated using the presentALMA data (see § µ m and 500 µ m band data in the fit because they arepossibly contaminated by di ff use emission around theYSO due to their large point spread function (about27 ′′ and 41 ′′ in FWHM, respectively). The distance toST11 is assumed to be the same as the distance to theLMC.The estimated stellar mass of ST11 in the best-fitmodel is 50 M ⊙ . The visual extinction derived in thebest-fit model is Av ∼
80 mag, which di ff ers by a fac-tor of two from the value estimated based on the dustcontinuum data presented in § ff erence mayarise from the assumption of optical properties of dustand / or the existence of a dust temperature gradient inthe line of sight, though this does not a ff ect the mainconclusion of this paper. The result of the SED fit isplotted in Figure 10.
5. Discussions5.1. Hot molecular core associated with ST11
Hot cores are one of the early stages of high-mass star formation. It is suggested that the hotcores are in a transitional evolutionary stage betweena deeply embedded protostellar object and a zero-age main-sequence star with a compact H II region(e.g., Zinnecker & Yorke 2007). Physical propertiesof hot cores are characterized by a small source size( ≤ ≥ cm − ), and warmgas / dust temperature ( ≥
100 K) (e.g., Kurtz et al. 2000;van der Tak 2004). In this section, we discuss thepresence of a hot core around ST11 on the basis of itsphysical and spectral properties.
A source size of ST11, which is measured by de-tected emission lines, is consistent with those of Galac-tic hot cores. The molecular lines of NO, SiO, SO,15ig. 10.— The spectral energy distribution of ST11. The plotted data are based on IRSF / SIRIUS photometry (pluses,black),
AKARI / IRC spectroscopy (solid line, blue),
AKARI / IRC photometry (open squares, light blue),
Spitzer / MIPSspectroscopy (solid line, green),
Spitzer / IRAC and MIPS photometry (open diamonds, light green), Gemini / T-ReCSnarrow-band photometry (open circles, light brown), Gemini / T-ReCS N-band spectroscopy (solid line, brown),
Her-schel / PACS and SPIRE photometry (filled diamonds, orange), ALMA 840 µ m continuum (filled star,red), and the bestfitted SED model (dashed line, grey). See section § able hotometric and spectroscopic data of ST11
Instrument Wavelength Flux FWHM a Ref.( µ m) (mJy) ( ′′ )IRSF / SIRIUS J 1.23 0.50 ± / SIRIUS H 1.66 0.82 ± / SIRIUS K s ± AKARI / IRC NG 2.5-5 spectroscopy 7.3 2 b AKARI / IRC N3 3.2 66.0 ± Spitzer / IRAC Band 1 3.6 88.1 ± Spitzer / IRAC Band 2 4.5 252.2 ± Spitzer / IRAC Band 3 5.7 577.8 ± Spitzer / IRAC Band 4 7.9 1090.0 ± AKARI / IRC S11 11 885.6 ± / T-ReCS Si1 7.7 1329 ±
29 0.6 5Gemini-S / T-ReCS Si2 8.7 940 ±
53 0.5 5Gemini-S / T-ReCS Si3 9.7 523 ±
10 0.6 5Gemini-S / T-ReCS Si4 10.4 950 ±
10 0.8 5Gemini-S / T-ReCS Si5 11.7 2297 ±
12 0.8 5Gemini-S / T-ReCS Si6 12.3 3226 ±
28 0.5 5Gemini-S / T-ReCS Lo-Res 8–12 spectroscopy 0.7 c Spitzer / IRS SL, SH, LH 5–36 spectroscopy 4-11 c Spitzer / MIPS 70 70 83810 ± Herschel / PACS 100 100 47180 ± Herschel / PACS 160 160 26800 ± Herschel / SPIRE 250 250 9650 ±
589 18 7
Herschel / SPIRE 350 350 4359 ±
226 27 7
Herschel / SPIRE 500 500 2110 ±
151 41 7ALMA Band 7 837 23.2 ± ote .— a FWHM of the point spread function. b Extraction width of the slitlessspectroscopic data. c Slit width.References. — (1) Kato et al. (2007); (2) Shimonishi et al. (2010); (3) Kato et al.(2012); (4) Meixner et al. (2006); (5) This work; (6) Kemper et al. (2010); (7)Meixner et al. (2013) , SO , and SO as well as dust continuum showa compact source size, which is equal to or lower thanthe beam size of ∼ O,HCO + , H CO + and H CO show somewhat extendeddistributions, but the dominant emission come fromthe compact region associated with a high-mass YSO.The extended components of these emission lines mayarise from their relatively low critical densities and / ore ffi cient formation routes in gas-phase reactions. Pro-tostellar outflows, which is discussed in § ff ect the spatial distribution of these moleculargas.The spatial association of the molecular emissionlines to the infrared source is another key to the natureof detected warm and dense molecular gas. Separationbetween a hot core region and an infrared source is re-ported to be smaller than ∼ The H gas density around ST11 is estimated to be > × cm − in § ( n cr ∼ cm − ,Williams & Viti 2013) imply an even higher gas den-sity around ST11. Galactic hot cores typically showgas densities between 10 –10 cm − (estimated fromTab.1 presented in Kurtz et al. 2000). The gas densityaround ST11 is thus consistent with those of knownhot cores. Rotational temperatures of molecular gas aroundST11 are compared with those of Galactic hot coresin Table 3. The temperature of SO , T rot = ± SO , T rot = ± , but this may be at-tributed to the combined e ff ect of the lack of availablehigher excitation lines of SO and slightly larger op-tical thickness of SO lines. Note that the temperatureof SO is even lower, but this is probably becausethe detected lines are biased to those with low exci-tation temperatures ( E u <
200 K) due to the limited frequency coverage of the data. The rotation analy-sis of SO and SO for the Orion data using thesimilar set of lines with those used in the analysis ofST11 results in T rot = ± SO and T rot = ± SO , which are lower than the valuesderived by Schilke et al. (1997) using a larger num-ber of lines with a wide range of upper state energies(Tab. 3). Therefore we conclude that the actual tem-perature of SO gas is in the order of ∼
100 K, whichcan trigger hot core chemistry via sublimation of icemantles. Sublimation of ice mantles around ST11 isalso suggested from the infrared observations of icesin Shimonishi et al. (2010), which reported that ST11shows the second weakest ice absorption bands among12 high-mass YSOs in the LMC.
As estimated in § × L ⊙ ) and stellar mass (50 M ⊙ ) of ST11 are con-sistent with the properties of high-mass YSOs. Thisindicates that ST11 can form a hot core region withthe help of its intense radiation. In addition, the redSED of ST11 (Fig. 10) indicates that it is still in anearly evolutionary stage, which is consistent with theproperties of hot core sources. Velocity widths of emission lines from ST11 aretypically 4–7 km / s except for CO and HCO + . Thisis in good agreement with velocity widths of emissionlines from Galactic hot core regions (typically 4–10km / s, e.g., Helmich & van Dishoeck 1997). Systemicvelocities of the lines are in a very narrow range, typ-ically 250–251 km / s, which suggests that the detectedmolecular species spatially coexist in a small regionaround a high-mass YSO.SO is often detected in hot core sources and oneof the useful tracers of warm and dense gas arounda high-mass YSO (e.g., Beuther et al. 2009). How-ever, sometimes warm SO gas is also prominentlydetected in deeply embedded and even younger pro-tostellar objects in which hot cores are not yet formed(e.g., W3 IRS5 in Helmich et al. 1994). The reasonof such SO enhancement is suggested to be due toshock dominated chemistry triggered by protostellaroutflows. These deeply embedded sources often showa deep self-absorption profile in their CO or HCO + emission lines due to the presence of a significantamount of cold gas in the envelope. As shown in18igure 5, emission from ST11 does not show such adeep self-absorption profile. This suggests that emis-sion from ST11 is mostly dominated by warm gas andthat ST11 has already reached the hot core phase.On the other hand, relatively weak strengths of hy-drogen recombination lines seen in the near-infraredspectrum of ST11 suggests that a prominent H II re-gion is not yet formed around the source (see Fig.1 inShimonishi et al. 2010). Such a transitional evolution-ary phase is well consistent with the properties of hotcores.In summary, the compact size of the emittingsource, the presence of warm and dense moleculargas around a high-mass protostar, the sublimation ofice mantles, and the detections of rich molecular linessuggest that ST11 is associated with a hot molecularcore. This is the first detection of an extragalactic hotmolecular core. Note that CH OH and complex or-ganic molecules, which are often detected in Galactichot cores, are not detected in ST11. The reason forthe lack of these molecular species is discussed in thenext section in the context of characteristic hot corechemistry in low metallicity.
Hot cores play a key role in the chemical com-plexity of interstellar and circumstellar molecules. Inthis section, we compare the chemical compositionsof ST11 with those of Galactic hot cores and discussthe impact of metallicity on chemical processes in hotcores.Fractional abundances of molecules around ST11are shown in Table 5. We use the following isotopeabundances for ST11; C / C = O / O = S / S =
15, according to Wang et al. (2009) inwhich isotope ratios for the star-forming region N113in the LMC are reported. For S, we use S / S = § O),and G34.3 + C / C = O / O = S / S =
22, and S / S = ∼ O) andG34.3 + O), and G34.3 + CO, CH OH, HNCO, and CS show significantlylower abundances, which can not be simply explainedby the low abundances of heavy elements in the LMC.On the other hand, NO shows the higher abundance inST11 than in Galactic sources despite the notably lownitrogen abundance in the LMC. Characteristics of in-dividual molecules are discussed below.
Carbon monoxide is detected by the CO(3–2) andC O(3–2) lines in ST11. The line intensity ratio ofCO(3–2) / C O(3–2) is about 28, and using O / O = ∼
120 and ∼ O(3–2), respectively. The CO(3–2) line is com-pletely optically thick, while the C O(3–2) line isoptically thin as in most of Galactic hot cores (e.g.,Helmich & van Dishoeck 1997). The abundance ofCO in ST11 estimated from C O is 9.1 × − , whichis very similar to the average CO abundance of 1.0 × − for three Galactic hot cores in Table 5. Elementalcarbon and oxygen are about three times less abundantin the LMC; [C / H] LMC = × − and [O / H] LMC = × − (Korn et al. 2002), while [C / H] ⊙ = × − and [O / H] ⊙ = × − (Grevesse & Sauval1998). Given the low abundances of carbon and oxy-gen in the LMC, the CO around ST11 is slightly over-produced as compared with Galactic hot cores.19 able ractional abundances N (X) / N (H )Molecule ST11 a Orion b W3 (H O) c G34.3 + d NoteCO 9.1 ± × − × − × − > × − + ± × − × − × − > × − CO 2.2 ± × − × − × − > × − CH OH < × − × − × − × − NO 2.0 ± × − × − · · · > × − HNCO < × − × − × − > × − HC N < × − × − × − > × − SiO 3.3 ± × − × − × − > × − CS < × − × − × − > × − CS 6.2 ± × − × − × − × − SO 2.4 ± × − × − × − > × − ± × − × − × − × − OCH < × − × − × − · · · HCOOCH < × − × − × − × − C H OH < × − × − · · · × − N ote .—Uncertainties and upper limits are of 2 σ level and do not include systematicerrors due to adopted spectroscopic constants. See § O. (2) Estimated from H CO + except for Orion, which is estimated fromHC O + . (3) Estimated from C S. (4) Estimated from SO, except for ST11, which isestimated from SO. (5) Estimated from SO .References. — a This work; b Ziurys et al. (1991); Sutton et al. (1995); Schilke et al.(1997), also see § c Helmich & van Dishoeck (1997); d MacDonald et al. (1996) + , H CO, CH OH, NO, HNCO, HC N, SiO, CS, H CS, SO, andSO for ST11 in the LMC (red) and Galactic hot cores, Orion hot core (cyan), W3 H O (orange), and G34.3 + + Formyl ion is detected via the HCO + (4–3) andH CO + (4–3) lines. The line intensity ratio of HCO + (4–3) / H CO + (4–3) is about 19, and using C / C =
49 in the LMC (Wang et al. 2009) we estimate op-tical depths of ∼ ∼ + (4–3) andH CO + (4–3), respectively. The HCO + (4–3) line ismoderately optically thick, while the H CO + (4–3)line is optically thin. The abundance of HCO + inST11 estimated from H CO + is 7.5 × − , whichis lower by a factor of ∼ + abundance of 2.0 × − for three Galactic hotcores in Table 5.We here discuss the connection between the ob-served HCO + abundance and the cosmic-ray ioniza-tion rate in the LMC. One of the possible pathways toform HCO + in molecular clouds is the gas-phase reac-tion: H + + CO → HCO + + H , (8)in which H + is formed by ionization of H via cos-mic rays and subsequent reaction with H (e.g.,Caselli et al. 1998). We here simply assume that HCO + is a major cationand electron is a major anion in dense clouds. Giventhat production of cations and electrons by the cosmic-ray ionization balances with their recombination, theequilibrium is formulated as ζ n H = k rec n HCO + n e , (9)where ζ is a cosmic-ray ionization rate, k rec = × − (300 / T ) . cm s − (5.1 × − at T =
100 K) is adissociative recombination rate of HCO + and electron(Mitchell 1990), n indicates a number density of eachspecies. With the assumption that cations and elec-trons have the the same abundance in neutral clouds,this equation is reformatted as X HCO + = X e = p ζ/ n H k rec , (10)where X indicates a fractional abundance relative toH . This equation indicates that the abundance ofHCO + is proportional to ζ . under the above simpleassumption.The cosmic-ray density in the LMC is reportedto be 20–30% of the typical Galactic value of ζ = × − s − based on gamma-ray observations21Abdo et al. 2010). The low cosmic-ray densityleads to the low cosmic-ray ionization rate (e.g.,Spitzer & Tomasko 1968). According to Eq. 10, theHCO + abundance in the LMC is expected to be lowerthan the Galactic value by a factor of ∼ density and gas temperature. This factor is closeto the observed di ff erence of the HCO + abundancebetween ST11 and Galactic sources. Therefore, wespeculate that the low cosmic-ray ionization rate in theLMC is one of the factors that contribute to the slightlylow abundance of HCO + in ST11.Note that the above assumption is very simple; i.e.,the cosmic-ray ionization rate may vary within theLMC and the electron abundance may di ff er from thatof HCO + due to the presence of other ions. Further-more, in the above discussion, we consider only one ofthe possible mechanisms to form HCO + . In a circum-stellar environment of hot cores, however, outflowsand stellar radiation would a ff ect the ionization con-dition. Previous studies actually argue that molecularoutflows and shocks contribute to the enhancement ofHCO + in star-forming cores (e.g., Girart et al. 1999;Rawlings et al. 2004; Arce & Sargent 2006). Futureobservations of other molecular ions and shock tracersare highly required for comprehensive understandingof ionization conditions in the LMC. CO Formaldehyde (H CO) is detected by the 5 , –4 , transition at 351.7686 GHz ( E u =
62 K). The abun-dance of H CO in ST11 is estimated to be 2.2 × − ,while Galactic hot cores show the abundances between10 − to 10 − as in Table 5, except for G34.3 + CO is less abun-dant in ST11 than Galactic sources by 1–2 orders ofmagnitude.Both gas-phase and grain surface reaction path-ways are suggested for the formation of H CO indense ISM. The grain surface pathway for the H COformation requires the hydrogenation of CO (e.g.,Hama & Watanabe 2013, and references therein).However, this pathways is suppressed in the LMCdue to higher dust temperatures (see detailed discus-sion presented in § CO ice on grain sur-faces contributes to the low abundance of H CO gas inST11. OH Any of methanol (CH OH) lines is not detectedin ST11. We estimate an upper limit on the frac-tional abundance to be < × − using the spec-trum at the frequency of the CH OH(7 –6 A + ) tran-sition at 338.4087 GHz ( E u =
65 K). Abundancesof CH OH in Galactic hot cores are typically be-tween ∼ − to ∼ − as in Table 5, but sometimesthe abundance reaches ∼ − (e.g., G5.89-0.39 inThompson & MacDonald 1999). In ST11, CH OHis depleted by at least 2–3 orders of magnitude ascompared with Galactic hot cores.The low abundance of CH OH in the LMC has beensuggested in previous studies. Nishimura et al. (2016)reported, based on spectral line surveys toward molec-ular clouds in the LMC, that thermal emission lines ofCH OH gas are significantly weak in the LMC com-pared to our Galaxy. Searches for maser emissionin the LMC reported the underabundance of CH OHmasers in the LMC (e.g., Ellingsen et al. 2010). Fur-thermore, Shimonishi et al. (2016) reported that theCH OH ice is less abundant in the LMC than in ourGalaxy based on infrared observations. They reportedthat all of the observed ten high-mass YSOs in theLMC show CH OH ice abundances less than 5–8 %relative to water ice, while about one-third of Galac-tic high-mass YSOs show CH OH ice abundances be-tween 10 % and 40 %. The authors suggested thatwarm ice chemistry is responsible for the low abun-dance of solid CH OH in the LMC; i.e., high dusttemperatures in the LMC suppress the hydrogenationof CO on the grain surface, which leads to ine ffi -cient production of the CH OH ice. A decrease inthe e ffi ciency of CO hydrogenation at an elevated tem-perature is measured by laboratory experiments (e.g.,Watanabe et al. 2003). In addition, the reduced forma-tion of CH OH in relatively warm molecular cloudsis confirmed by numerical simulations of grain sur-face chemistry dedicated to the LMC environment(Acharyya & Herbst 2015).The warm gas temperature around ST11 suggeststhat ice mantles are mostly sublimated. The low ele-mental abundances of carbon and oxygen in the LMCshould partly contribute to the observed low CH OHabundance. However, we need additional explanationto account for the CH OH deficiency by several or-ders of magnitude, and this should be related to grainsurface chemistry by which CH OH is mainly formed.We suggest that suppressed production of CH OH ice22ue to warm ice chemistry at the molecular cloud stageor the deeply embedded YSO stage is responsible forthe deficiency of CH OH gas in this LMC hot core.The above interpretation on the low CH OH abun-dance can also be applied to the low abundance ofH CO in ST11 since a major grain surface pathwayof both species requires the CO hydrogenation. How-ever, the di ff erence is that H CO has possible forma-tion routes both in the gas-phase and in the solid-phase,while CH OH does not have an e ffi cient formationroute in the gas-phase under typical molecular cloudconditions. We speculate that the H CO observed inST11 is mainly produced by gas-phase reactions. Aslightly extended distribution of the H CO emission(Fig. 4) supports this idea, because the species subli-mated from ice mantles often show the compact distri-bution compared to those produced in the gas phase.Owing to the lack of a possible gas-phase formationroute, the degree of depletion around ST11 is actuallylarger in CH OH as compared to H CO.
Nitric oxide (NO) is one of the interesting moleculesthat show a peculiar abundance in ST11. The abun-dance of NO in ST11 is estimated to be 2.0 × − .On the other hand, the average abundance and standarddeviation of NO in six Galactic sources in Ziurys et al.(1991) including hot cores, high-mass protostellar ob-jects, and Galactic center objects is 8.2 ± × − .In the LMC, nitrogen is even more depleted than othermajor elements such as carbon and oxygen; the ele-mental abundance of nitrogen in the LMC is [N / H] LMC = × − (Korn et al. 2002), while [N / H] ⊙ = × − for the Sun (Grevesse & Sauval 1998). Despitethe lower nitrogen abundance by a factor of 8 in theLMC, the abundance of NO in ST11 is higher thanGalactic star-forming regions by a factor of 2–3.Ammonia (NH ) gas, which is often the most abun-dant nitrogen-bearing species in star-forming regions,is reported to be deficient in the LMC by 1.5–2 or-ders of magnitude as compared with Galactic star-forming regions (Ott et al. 2010). A previous infraredstudy suggests that the NH ice around an embeddedhigh-mass YSO in the LMC is possibly less abun-dant as compared with Galactic high-mass YSOs(Shimonishi et al. 2016). Although the NH abun-dance in ST11 is unknown, the relatively high abun-dance of NO implies that NO could play an importantrole in nitrogen chemistry around ST11. It is suggested that NO is formed by a neutral-neutral reaction in the gas-phase (e.g., Herbst & Klemperer1973; Pineau des Forets et al. 1990):N + OH → NO + H , (11)and destroyed byNO + N → N + O . (12)A numerical simulation of nitrogen chemistry in warmmolecular gas suggests that the resultant abundance ofNO through the above reactions increases as the tem-perature of molecular gas increases (Pineau des Forets et al.1990). However, the enhancement of the NO abun-dance is suggested to be only a factor of 1.5 as thegas temperature increases from 70 K to 250 K. Thuswe speculate that pure gas-phase chemistry make onlya limited contribution on the enhanced abundance ofNO in ST11.A compact distribution of the NO emission ob-served in ST11 may hint at the possible origin in icesublimation. Laboratory experiments argue that bom-bardment of energetic ions to interstellar ice analoguescontaining H O, O , and N or CO and N produceNO in ice mixtures (Boduch et al. 2012; Sicilia et al.2012). However, the low cosmic-ray density in theLMC (see § ffi cient aroundST11. A reaction pathway though di ff usive grain sur-face chemistry, N + O → NO, is also suggested in theliterature (Tielens & Hagen 1982), but uncertainty re-main regarding the e ffi ciency of the reaction in actualgrain surfaces.The reason of the enhanced abundance of NO inST11 despite the low elemental abundance of nitro-gen in the LMC remains to be investigated. Detailedmodeling of hot core chemistry in metal-poor environ-ments is obviously necessary to interpret the enhancedabundance of NO in ST11. Isocyanic acid (HNCO) is not detected in ST11. Weestimate an upper limit on the fractional abundance ofHNCO to be < × − based on the non-detection ofthe 16 , –15 , transition at 351.6333 GHz ( E u = , –15 , and 16 , –15 , ,351.4168 GHz) line with V LS R = / s. How-ever, the high upper state energy of this transition ( E u =
518 K) and the non-detection of the 16 , –15 , − to 10 − (Bisschop et al. 2007), which is consis-tent with the average abundance of 2.3 × − for threeGalactic hot cores in Table 5. Thus the HNCO abun-dance in ST11 is lower than Galactic counterparts byat least 1–2 orders of magnitude.Both gas-phase and grain surface reaction pathwaysare suggested for the formation of HNCO in denseISM (e.g., Tielens & Hagen 1982; Turner et al. 1999).The grain surface pathway requires the hydrogenationof CO to form HCO and subsequent reaction of HCO + N → HNCO. However, as discussed in § ffi cientin the LMC due to high dust temperatures, which pos-sibly suppress the grain surface formation of HNCO.This may contribute to the low abundance of HNCOin ST11 along with the low elemental abundance ofnitrogen in the LMC. N Cyanoacetylene (HC N), an unsaturated carbon-chain molecule often detected in star-forming regions,is not detected in ST11. We estimate an upper limiton the fractional abundance of HC N to be < × − based on the non-detection of the 38–37 transition at345.6090 GHz ( E u =
323 K). Abundances of HC Nin Galactic hot cores vary from ∼ − to ∼ − as inTable 5. The estimated upper limit on the HC N abun-dance in ST11 is thus at the lower end of the GalacticHC N abundance.
Silicon monoxide (SiO) is detected by the J = E u =
75 K). The abun-dance of SiO for ST11 is 3.3 × − , which seemsto be low compared to Galactic sources, but the dis-persion in the Galactic SiO abundances is significantlylarge as shown in Figure 11. The average SiO abun-dance for the three Galactic hot cores in Table 5 is1.8 × − , but their abundances range from ∼ − to ∼ − .The elemental abundance of silicon in the LMCis about three times lower than the solar abundance;[Si / H] LMC = × − (Korn et al. 2002), while[Si / H] ⊙ = × − (Grevesse & Sauval 1998). Thusthe elemental abundance in the LMC partly contributesto the observed low SiO abundance, but additional pro- cesses may be necessary to account for the 1–2 ordersof magnitude lower SiO abundance.SiO is often linked with the presence of ener-getic protostellar outflows since destruction of silicon-bearing dust (e.g., MgSiO ) by shock is believed to bethe origin of SiO (e.g., Blake et al. 1996; Wright et al.1996; Tercero et al. 2011, and references therein). De-struction of dust grains and release of silicon can oc-cur around ST11 since it shows a sign of outflows asdescribed in § ff ect the SiO abundance. The C S(7–6) transition at 337.3965 GHz ( E u =
65 K) is not detected toward ST11. We estimate anupper limit on the abundance of carbon monosulfide(CS) to be < × − , while Galactic hot cores andhigh-mass protostellar objects typically show the CSabundances between 10 − and 10 − (Tab. 5, see alsovan der Tak et al. 2003). CS is significantly less abun-dant by at least 1–2 orders of magnitude in ST11 com-pared to Galactic sources, and the low elemental abun-dances of carbon and sulfur in the LMC can not by it-self explain the depletion of CS. Given the relativelyhigh abundance of SO in ST11 as discussed in § due to di ff erent circumstellarchemistry around ST11. CS We detect the emission line of thioformaldehyde(H CS) at 338.0832 GHz (10 , –9 , , E u =
65 K), butthe S / N of the line is relatively poor. The H CS abun-dance is estimated to be 6.2 × − in ST11, whileGalactic hot cores show the abundance of ∼ × − on average with a relatively low dispersion (Tab. 5).H CS seems to be less abundant in the LMC hot corethan in Galactic hot cores as well as CS, and only asmall fraction of gas-phase sulfur is incorporated intoH CS in ST11. Since our H CS abundance is esti-mated using a single line with poor S / N, further multi-line observations of H CS transitions are necessary foraccurate determination of the H CS abundance in the24MC.
The abundance of sulfur monoxide (SO) is es-timated to be 2.4 × − in ST11, while Galactichot cores and high-mass protostellar objects typicallyshow the SO abundances between 10 − and 10 − (Tab.5, see also van der Tak et al. 2003). The Orion hotcore shows an exceptionally high SO abundance of2.0 × − . The results suggest that ST11 shows aslightly higher abundance of SO than typical Galacticcounterparts despite the low metallicity of the LMC.If we use the solar S / S ratio of 127, then the SOabundance in ST11 is 7.6 × − , which is even higherthan typical Galactic abundances.We emphasize, however, that our SO abundance en-tails considerable uncertainty. The abundance is esti-mated using a single SO line at 337.1986 GHz ( E u =
81 K), which contains a number of unresolved hy-perfine structures, while Galactic SO abundances areestimated from SO lines. Multi-line observations ofSO and its isotopologues are necessary for further dis-cussion. Sulfur dioxide (SO ) is in this study a key moleculefor which we detect the largest number of transitions;we detect nine SO lines, one SO ( ν =
1) line, ten SO lines, and five SO lines. The fractional abun-dance of SO is estimated to be 2.1 × − in ST11based on rotation diagram analysis of SO lines. Thecolumn density ratio of SO and SO is about 14,suggesting that the optically thin assumption is mostlyvalid for the observed SO lines because the S / Sratio is reported to be 15 in the LMC. If we assume thesolar isotope ratio of S / S =
22, the SO lines couldbe moderately optically thick. In either case, we canreasonably assume that SO and SO lines are op-tically thin because these isotopologues are much lessabundant than SO . The isotope abundance of Sin ST11 based on the present results is S / S = S / S = abun-dance from ∼ − to ∼ − , while even younger em-bedded high-mass protostellar objects show the abun-dance of ∼ − (van der Tak et al. 2003). ST11 showsa factor of ∼ abundance as compared withthe average abundance of 7.0 × − for the threeGalactic hot cores in Table 5. The elemental abun- dance of sulfur in the LMC is reported to be [S / H] LMC = × − , while the solar abundance is [S / H] ⊙ = × − (Russell & Dopita 1992). Sulfur is lessabundant by a factor of ∼ abundancein ST11 is well explained by the elemental abundanceof sulfur in the LMC. This would suggest that hotcore chemistry of SO is dependent on elemental abun-dances of the host galaxy. A multitude of SO and itsisotopologue line detections in ST11 imply that SO can be a key molecular tracer to test hot core chem-istry in metal-poor environments.It should be noted that sulfur chemistry in hotcore regions is highly time-dependent (e.g., Charnley1997). Hence, both the age and the interstellar en-vironment should be taken into account to interpretthe chemical compositions of sulfur-bearing speciesaround ST11. Numerical simulations of hot corechemistry dedicated to low metallicity environmentsare thus highly required. OCH , HCOOCH , C H OH Spectral lines from complex organic molecules arenot detected in this work. We estimate upper limits onfractional abundances of three molecules whose rel-atively strong transitions are covered in the presentdata; [CH OCH / H ] < × − , [HCOOCH / H ] < × − , and [C H OH / H ] < × − . Aver-age abundances of these molecules for Galactic hotcores in Table 5 are [CH OCH / H ] = × − ,[HCOOCH / H ] = × − , and [C H OH / H ] = × − . These estimates suggest that CH OCH is less abundant by at least an order of magnitude inST11 compared with Galactic hot cores, while upperlimit abundances of HCOOCH and C H OH are com-parable to the average abundances of Galactic sources.Although the present data do not provide conclusiveupper limits on abundances of a majority of complexorganic molecules in the LMC, the possibly lowerabundance of CH OCH in ST11 suggests the for-mation of large molecules may be less e ffi cient in theLMC.It is suggested by theoretical studies that CH OH inice mantles plays an important role in the formation ofcomplex organic molecules (e.g., Nomura & Millar2004; Garrod 2008; Herbst & van Dishoeck 2009).Shimonishi et al. (2016) argue that formation of com-plex organic molecules from methanol-derived speciescould be less e ffi cient in the LMC due to the low25ig. 12.— (a) Comparison of spectral line profiles of CO(3–2) (thick solid line, black) and C O(3–2) (thin solid line,gray) observed toward ST11. The spectra are arbitrarily scaled and the horizontal axis is the LSR velocity. The COline shows a broad line width and a prominent red-shifted component, which indicate the presence of outflows in theline of sight. The blue and red vertical dashed lines represent the velocity range of the blue-shifted and red-shiftedhigh-velocity components, which are visualized in the right panel. (b) Spatial distributions of integrated intensities ofhigh-velocity wings. The blue-shifted and red-shifted wing components are shown by the blue and red contours. Thecontour levels are 20 %, 40 %, 60 %, and 80 % of the peak intensity. The background is the 840 µ m continuum. Thesynthesized beam size is shown by the gray filled circle at the lower left.abundance CH OH ice around high-mass YSOs inthe LMC. We confirm the significant deficiency ofCH OH gas around ST11 in this work. We thus spec-ulate that the low abundance of CH OH contributes tothe low production e ffi ciency of CH OCH and possi-bly other complex organic molecules in the LMC.Note that unknown rotational temperatures and lim-ited frequency coverages produce considerable uncer-tainties on the abundance estimate. Furthermore, theformation of large molecules from other parent speciessuch as H CO, C H, c-C H , and NO, which aredetected in the LMC, should be taken into accountfor comprehensive understanding of complex chem-istry around protostars. Hence, further observationswith broader frequency coverage and higher sensitiv-ity are critically needed in conjunction with theoreticaland experimental e ff ort to understand complex organicchemistry in metal-poor environments. In this section, we discuss the second detectionof extragalactic protostellar outflows after Fukui et al.(2015), which reported the detection of protostellaroutflows in the star-forming region N159 in the LMC with ALMA. Evidence of protostellar outflows is seenin the CO emission line of ST11. Figure 12(a) showsthe spectral profile of the CO(3–2) line extracted fromthe 0.5 ′′ diameter region centered at ST11. The profileof the optically thin C O(3–2) line is also shown fora comparison purpose. The CO(3–2) profile shows anapparently broader velocity width compared to otherlines detected in ST11. The central velocity of the COline is nearly consistent with those of other lines ( V LS R ∼ / s to 280 km / s. In addition, a promi-nent red-shifted component is seen around V LS R ∼ / s. We suggest that these high-velocity wing com-ponents are due to protostellar outflows from ST11 be-cause the high-velocity CO gas is well spatially asso-ciated with a high-mass YSO as described in the nextparagraph. The observed outflow velocity of ∼ / s is consistent with molecular outflow velocitiesobserved in Galactic high-mass star-forming regions(e.g., Lada 1985). A high-velocity component is notobviously seen in other lines besides CO(3–2).Spatial distributions of high-velocity wings areshown in Figure 12(b). We here define the velocityrange of the blue-shifted wing to be 230.5 km / s to2640.5 km / s and the red-shifted wing to be 265.5 km / sto 285.5 km / s. Both high-velocity wings are spatiallyassociated with a central protostar which is traced thethe continuum emission. The complex structures seenin the distribution of high-velocity gas imply that thecircumstellar environment of ST11 is dynamically ac-tive. If we assume the spatial extent of high-velocitygas to be 0.24 pc ( ∼ ′′ ) and the outflow velocity tobe 20 km / s according to the distribution and spectrumof the red wing component, we can roughly estimatean upper limit on the dynamical timescale of the out-flow, which is about 10 years. This timescale is lowerthan a typical formation time of high-mass stars ( ∼ years, Zinnecker & Yorke 2007) and thus consistentwith high-mass star-formation scenarios.We also make a rough estimate of several outflowparameters, namely, the outflow mass, the mass en-trainment rate, the mechanical force, and the energy.The outflow mass is estimated by adding blue and redwing components within the region where outflow gasemission is detected with the S / N ratio higher than six.To convert integrated intensities of CO(3–2) to the to-tal gas mass, we use a conversion factor of 8.8 M ⊙ (Kkm / s) − pc − , which is derived from the present obser-vations toward the ST11 center. The derived outflowmass is M out =
74 M ⊙ , where the blue-shifted compo-nent contains 13 M ⊙ ( M blue ) and the red-shifted com-ponent contains 61 M ⊙ ( M red ). This outflow mass cor-responds to the mass entrainment rate of ˙ M = M out / t = × − M ⊙ yr − , where t is the dynamical time scaleof outflows discussed above (10 yr). The mechanicalforce ( F ) and the energy ( E ) of outflows are derived bythe following equations: F = ( M blue V blue + M red V red ) / t and E = ( M blue V + M red V ) / V blue and V red aremean velocities of blue and red wing components, andwe use V blue =
13 km / s and V red =
20 km / s, respec-tively. Consequently, the derived outflow force is F = ⊙ km / s yr − and the outflow energy is E = × erg. The above outflow parameters for ST11are roughly consistent with those observed in Galactichigh-mass YSOs that have similar luminosities withST11 (e.g., Beuther et al. 2002).The present detection increases the number of ex-tragalactic protostellar outflow samples, which shouldhelp understand the dynamical processes of high-mass star-formation in di ff erent metallicity environ-ments. Systematic observations of extragalactic out-flow sources are necessary for statistical comparison ofGalactic and extragalactic protostellar outflows. Fur-ther detailed analysis of outflows around ST11 is be- yond the scope of this paper and will be presented in afuture work.
6. Summary
We report the first detection of an extragalactic hotmolecular core based on radio interferometric observa-tions toward ST11, a high-mass YSO in the LMC, withALMA. The high spatial resolution (0.12 pc) ALMABand 7 (345 GHz) spectral and continuum band ob-servation data are presented. We discuss the physicaland chemical properties of the source and obtained thefollowing conclusions.1. Molecular emission lines of CO, C O, HCO + ,H CO + , H CO, NO, SiO, H CS, SO, SO , SO , and SO are detected from a com-pact region ( ∼ E u >
100 K) of SO and its iso-topologues are detected. On the other hand,CH OH, HNCO, CS, HC N, and complex or-ganic molecules are not detected.2. Physical properties of ST11 are derived usingthe obtained data. The H gas density around thesource is estimated to be at least 2 × cm − based on the dust continuum data. The temper-ature of molecular gas is estimated to be higherthan 100 K based on rotation diagram analysisof SO and SO lines. The SED analysis in the1–1000 µ m range suggests that ST11 is a high-mass YSO with the luminosity of 5 × L ⊙ andthe stellar mass of 50 M ⊙ .3. The compact size of the emitting source, warmgas temperature, high density, and rich molecu-lar lines around a high-mass protostar suggestthat ST11 is associated with a hot molecularcore.4. We find that the molecular abundances of thehot core in the LMC are significantly di ff erentfrom those of Galactic hot cores. The abun-dances of CH OH, H CO, and HNCO are re-markably lower compared with Galactic sourcesby at least 1–3 orders of magnitude, althoughthe gas temperature is warm enough for thesublimation of ice mantles. The deficiency ofCH OH gas in a warm and dense region is con-sistent with the previously reported low abun-dance of the CH OH ice in the LMC. We sug-27est that the chemical compositions of ST11are characterized by the deficiency of moleculeswhose formation requires the hydrogenation ofCO on grain surfaces.5. It is interesting that NO shows a higher abun-dance in ST11 than in Galactic sources despitethe notably low abundance of nitrogen in theLMC. This is in contrast to low abundances ofnitrogen-bearing molecules such as NH , HCN,HNC in the LMC reported by previous studies.The reason of the enhanced abundance of NOremains to be investigated.6. Slightly lower abundance of SO in ST11 thanin Galactic hot cores is well explained by thelow abundance of elemental sulfur in the LMC.CS and H CS are less abundant than Galactichot cores by at least 1–2 orders of magnitude.The abundance of SO is possibly high in theLMC, but the estimate based on a single SOline should be taken with caution. The largenumber of SO and its isotopologue line detec-tions in the LMC hot core imply that SO can bea key molecular species to test hot core chem-istry in metal-poor environments.7. We find molecular outflows around ST11, whichis the second detection of an extragalactic pro-tostellar outflow. An apparently broad veloc-ity width is seen in the spectral profile of theCO(3–2) line, which ranges in the LSR veloc-ity from 230 km / s to 280 km / s. A prominenthigh-velocity component is also seen in the red-shifted wing. We estimate an upper limit on thedynamical timescale of outflows to be 10 years,which is consistent with the timescale of high-mass star formation. Several outflow parametersare also estimated based on the present results. This paper makes use of the following ALMA data: ADS / JAO.ALMA / NRAO andNAOJ. This work has made extensive use of the Cologne Databasefor Molecular Spectroscopy and the molecular database of the JetPropulsion Laboratory. Partly based on data obtained at the Gem-ini Observatory via the time exchange program between Gemini and the Subaru Telescope (Program ID: S10B-120). The Gemini Obser-vatory is operated by the Association of Universities for Researchin Astronomy, Inc., under a cooperative agreement with the NSFon behalf of the Gemini partnership: the National Science Foun-dation (United States), the National Research Council (Canada),CONICYT (Chile), Ministerio de Ciencia, Tecnolog´ıa e Innovaci´onProductiva (Argentina), and Minist´erio da Ciˆencia, Tecnologia eInovac¸ ˜ao (Brazil). The authors are grateful to Satoshi Yamamoto forhis useful comment on spectral data. Takashi Shimonishi was sup-ported by the ALMA Japan Research Grant of NAOJ Chile Observa-tory, NAOJ-ALMA-0061. This work is supported by a Grant-in-Aidfrom the Japan Society for the Promotion of Science (15K17612).Finally, we would like to thank an anonymous referee, whose sug-gestions greatly improved this paper.
REFERENCES