Formation of High-Mass stars in an isolated environment in the Large Magellanic Cloud

TL;DR

This study investigates isolated high-mass star formation in the Large Magellanic Cloud, revealing that compact molecular clouds can efficiently form massive stars even outside typical giant molecular cloud environments.

Contribution

It demonstrates that high-mass stars can form in small, isolated molecular clouds with high efficiency, challenging the traditional view of star formation in large GMCs.

Findings

High-mass YSOs are associated with compact molecular clouds of a few thousand solar masses.

Star formation efficiency in these clouds can reach up to 40%.

Enhanced turbulence may facilitate efficient star formation in low-metallicity environments.

Abstract

The aim of this study is to characterize the distribution and basic properties of the natal gas associated with high-mass young stellar objects (YSOs) in isolated environments in the Large Magellanic Cloud (LMC). High-mass stars usually form in Giant Molecular Clouds (GMCs) as part of a young stellar cluster, but some OB stars are observed far from GMCs. By examining the spatial coincidence between the high-mass YSOs and 12CO (J = 1-0) emission detected by NANTEN and Mopra observations, we selected ten high-mass YSOs that are located away from any of the NANTEN clouds but are detected by the Mopra pointed observations. The ALMA observations revealed that a compact molecular cloud whose mass is a few thousand solar masses or smaller is associated with the high-mass YSOs, which indicates that these compact clouds are the sites of high-mass star formation. The high-density and high-temperature throughout the clouds are explained by the severe photodissociation of CO due to the lower metallicity than in the Galaxy. The star formation efficiency ranges from several to as high as ~ 40%, indicating efficient star formation in these environments. The enhanced turbulence may be a cause of the efficient star formation therein, as judged from the gas velocity information and the association with the lower density gas.

Equations

$${\tau }_V{(}^{13}\mathrm{CO})=-\ln \left(1-\frac{T_V{(}^{13}\mathrm{CO})/5.3}{1/(\exp [\frac{5.3}{T_{\mathrm{ex}}}]-1)-0.16}\right)$$

$$N{(}^{13}\mathrm{CO})=3.0\times 10^{14}\frac{T_{\mathrm{ex}}\int {\tau }_V^{13}dV}{1-\exp [\frac{-5.3}{T_{\mathrm{ex}}}]},$$

$$N({\mathrm{H}}_2)=\frac{50}{1.6\times 10^{-5}}N{(}^{13}\mathrm{CO}).$$

$$M_{\mathrm{clump}}=\mu m({\mathrm{H}}_2)\sum [D^2\Omega N({\mathrm{H}}_2)],$$

$$M_{\mathrm{vir}}[\text{MO}]=190\Delta V_{\mathrm{clump}}^2[\mathrm{km}{\mathrm{s}}^{-1}]R_{\mathrm{deconv}}[\mathrm{pc}],$$

$$SFE=\frac{M_{\mathrm{ecl}}}{M_{\mathrm{ecl}}+M_{\mathrm{cloud}}},$$

Full text

1 Introduction

High-mass stars strongly influence physically and chemically the interstellar matter and thus galactic evolution. They emit a significant amount of ultraviolet emission, ionizing and heating the ambient gas. They also generate a stellar wind, and it compresses the surrounding interstellar medium. Furthermore, a supernova explosion occurs at the end of their evolution, releasing massive energy into interstellar space. The interstellar matter is strongly influenced by the released energy from the high-mass star. It is, therefore, extremely important to investigate the formation mechanism of high-mass stars to study galaxy evolution. Our current theory of star formation proposes that: 1) most stars are formed in giant molecular clouds (GMCs) as part of a cluster or group of stars, 2) the evolutionary timescale of high-mass young stellar objects (YSOs) is short compared to the disruption timescale of their parent molecular cloud, and 3) star formation is inefficient (i.e., a low fraction of the molecular gas mass is ultimately converted into stars) (c.f., [Zinnecker & Yorke (2007)] and the references therein). Assuming a normal stellar initial mass function, the birth of a single high-mass star should therefore always be coincident with the formation of many low-mass YSOs. The formation mechanism of a high-mass star, however, is not straightforward; high-mass stars are considered to be formed by the collapse of a highly turbulent core (core accretion models; e.g., [McKee & Tan (2002), McKee & Tan (2003)]) or by Bondi-Hoyle accretion onto small protostars (competitive accretion model; e.g., [Zinnecker (1982)], [Bonnell et al. (1997)]). Molecular observations toward the natal clouds of high-mass YSOs are of vital importance to resolve the issue although high-mass star-forming regions are usually embedded in a complex environment in the plane of the galaxy.

  Although most OB stars are believed to be formed as associations/clusters in GMCs, there exists a class of OB stars that are distant from such complexes. Many are considered to be runaway OB stars that are ejected from their birthplace; however, some cannot be assigned to a probable progenitor cloud and the origin of these $``$isolated$"$ OB stars located far from molecular cloud complexes remains unknown (e.g., [de Wit et al. (2004), de Wit et al. (2005)]; [Zinnecker & Yorke (2007)] and the references therein).

  The majority of Milky Way YSOs are located along the Galactic plane, and therefore distance ambiguity and contamination from unrelated emission sources along the same line-of-sight complicates the study of Galactic high-mass YSOs. With the high resolution and sensitivity of ALMA, we can now study CO gas in nearby galaxies in great detail. The Large Magellanic Cloud (LMC) is located close to us ($\sim$ 50 kpc; [Schaefer (2008)]; [de Grijs et al. (2014)]) at high Galactic latitude ($b$ $\sim$ $-\timeform{30D}$) with face-on orientation ($i$ $\sim$ $\timeform{38D}$; [Balbinot et al. (2015)]). Therefore, toward the LMC, we have the clearest view of the distribution of molecular clouds and young stars in any galaxy, including our own, making the LMC one of the best places to investigate the origin of high-mass YSOs.

  The CO clouds in the LMC have been studied extensively as reviewed by Fukui & Kawamura (2010) (see references therein). Spatially-resolved observations of the GMCs at 40 pc resolution in the whole LMC disk were initiated with NANTEN by Fukui et al. (1999), and followed up at higher sensitivity by Fukui et al. (2008) and Kawamura et al. (2009). Subsequently, higher resolution CO observations with Mopra telescope by Wong et al. (2011) revealed the CO distribution at 11 pc resolution toward the NANTEN GMCs and their physical properties were studied in detail (Hughes et al. (2010)). These preceding works focused on massive and active OB star-forming regions with clusters, whereas it was not explored how star formation is taking place at smaller scales, involving only a single OB star. It is possible that such small-scale star formation is important in galactic evolution if they are numerous in the whole disk. It is also important to learn the implications of single OB star formation in theories of high mass star formation, which assume massive aggregates as the precursor of O star clusters (e.g., see for a review Tan et al. (2014); also Ascenso (2018)).

  In this paper, we present the results of ALMA observations of compact molecular clouds associated with $``$isolated$"$ high-mass YSOs in the LMC that are located away from any GMCs to show the physical properties of the natal clouds. Since high-mass star formation involves many diverse processes, the study of their formation can be facilitated by characterizing the molecular gas of more isolated clouds. This study specifically targets high-mass YSOs that are encompassed by relatively isolated clouds. We characterize these clouds in detail based on ALMA and ancillary observations, and make comparisons to known Galactic star-forming regions.

2 Source selection and observations

As part of the Surveying the Agents of a Galaxy’s Evolution (SAGE) Spitzer Legacy program (Meixner et al., 2006), $\sim$ 1,800 unique YSO candidates were cataloged in the LMC (Whitney et al. (2008); see also Gruendl & Chu (2009)). The YSOs are selected by their excess of infrared emission, indicative of an early evolutionary stage when the YSOs are still surrounded by disks and/or infalling envelopes. We identified high-mass YSOs in the LMC that appear to be isolated, i.e., not associated with CO emission by examining the spatial coincidence between the YSOs and 12CO ($J$ = 1–0) emission detected by the NANTEN survey (half-power beam width, HPBW \sim\;$$\timeform{2^{\prime}.6}, Fukui et al. (2008)) and MAGMA CO survey (Wong et al. (2011)). A further set of single-pointing CO observations with the Mopra Telescope (HPBW \sim\;$$\timeform{30"}; Proposal ID: M579, PI: T. Wong) targeted 76 $``$definite$"$ and $``$probable$"$ YSOs from Gruendl & Chu (2009) with [8.0]$\;<\;$8.0 mag that were outside the MAGMA-LMC survey coverage. Single pointing observations toward the sources were carried out, with 80% of sources detected in 12CO ($J$ = 1–0). Follow-up Mopra CO mapping observations (Proposal ID: M2013B34, PI: T. Onishi) showed that CO emission was spatially extended toward some of the YSOs, although CO emission was only detected precisely at the YSO positions in some cases. We note that a small fraction of the isolated YSOs with Mopra CO detections are located in regions with H i column densities of $\lesssim$ 1021 cm*-2*, where equilibrium models of H2 formation and dissociation (e.g., Krumholz et al. (2009)) predict that there should be no molecular gas.

  For this study, we targeted a typical sample, regarding their association with neutral gas tracers, of the isolated high-mass YSOs in the LMC. All of the targets are separated by more than 200 pc from NANTEN CO clouds to exclude potential $``$runaway$"$ sources, which are stars that have been ejected from their progenitor gas cloud at high speed. This minimum spatial separation is obtained by assuming that these runaway YSOs have a maximum speed of 200 km s*-1* (e.g., Perets & Šubr (2012)) and a conservative age of 106 yrs. Included in the above samples are YSOs where the H i column density is less than 2 $\times$ 1021 cm*-2* with an angular resolution of 1 arcmin (Kim et al., 2003), which corresponds to the spatial resolution of 14.4 pc. Our final target sample (see table 2) is spatially distributed across the LMC gas disk (see figure 1), and consists of:

1. (a)

Three YSOs with CO emission that were unresolved by Mopra. The molecular clouds associated with these high mass YSOs must have sizes smaller than 7 pc and masses less than a few thousand solar masses. 2. (b)

Seven YSOs where CO emission was detected by Mopra; CO mapping observations were not attempted for these sources, but the morphology of the 350 $\mu$m emission suggests that the molecular clouds are likely to be spatially compact. These CO clouds were not detected by NANTEN, which puts an upper limit on their mass of $\sim$ 10,000 solar masses (Fukui et al., 2008).

In order to determine the physical parameters of these YSOs, we used the YSO model grid of \authorciteRobitaille06 (\yearciteRobitaille06, \yearciteRobitaille07) and the spectral energy distribution (SED) fitter for the available photometry data including Spitzer and Herschel fluxes (1.2–500$\;\mu$m). Figure 1 shows the positions of the sources overlaid on the H i intensities (Kim et al., 2003). The basic properties are listed in table 2. The luminosities range from 0.3 to 19 $\times$ 104 $\LO$, and the fitted masses from 10 to 17 $M_{\odot}$, which means that they are B stars earlier than B2 (e.g., Hohle et al. (2010)). We note that these models do not include multiplicity even though most high-mass stars are multiples.

3 ALMA observations

We carried out ALMA Cycle 2 Band 3 (86–116 GHz) and Band 6 (211–275 GHz) observations toward 10 isolated YSOs (Project 2013.1.00287.S, PI: Toshikazu Onishi). The targeted molecular lines were 13CO ($J$ = 1–0), C18O ($J$ = 1–0), CS ($J$ = 2–1), 12CO ($J$ = 2–1), 13CO ($J$ = 2–1) and C18O($J$ = 2–1) with a frequency resolution of 61 kHz, corresponding to velocity resolutions of $\sim\;$0.17 km s*-1* for Band 3 and $\sim\;$0.08 km s*-1* for Band 6, both with 1920 channels. We also observed the continuum emission with the wider bandwidth of 1875 MHz. The H40$\alpha$ radio recombination line was included in the wider bandwidth observations with a frequency resolution of 976 kHz ($\times$ 1920 channels). The ALMA Band 6 observations were carried out in June 2014 and September 2015. They used 34 antennas, and the projected baseline length of the 12 m array ranges from 13 m to 392 m. The ALMA Band 3 observations were carried out in December 2014. They used 34 antennas and the projected baseline length of the 12 m array ranges from 13 m to 247 m. The projected baselines of the Atacama Compact Array (ACA; a.k.a. Morita Array) observation range from 8 m to 36 m. For the Band 6 data, we combined the 12 m array, 7 m array, and TP array data to recover the extended emission except for Target 9, toward which we have no TP array observations due to the failed observations. For Band 3, we didn’t include the ACA observations because the maximum recoverable scale of $\sim$ $\timeform{26"}$, corresponding to $\sim$ 6.2 pc, is considered to be large enough to investigate the gas distribution of the compact clouds. The ALMA beam sizes and sensitivities in the present observations are listed in table 3.

4 Results

4.1 Compact clouds associated with isolated YSOs

We show the ALMA images toward the targets both in Band 3 and Band 6 in figure 2. CO and 13CO emission are detected from all the targets, and compact molecular clouds are associated with all the candidate isolated YSOs. This confirms that they are indeed YSOs as they still have their natal clouds around them. The figures show that the molecular distribution of 12CO ($J$ = 2–1) and 13CO ($J$ = 1–0) toward Target 1, 2, and 3 is compact having a size of $\sim$ 1 pc, which is consistent with the past Mopra observations, and each YSO is located at the peak of the molecular emission. The 8$\;\mu$m emission by Spitzer also shows a good correspondence with the molecular distribution. We detected CS ($J$ = 2–1) emission toward Target 2, 5, 6, 8, and 10, indicating the association of dense gas toward these YSOs. Two observation fields, Target 5 and 6, include multiple YSO candidates. The observation field of Target 6 includes 3 YSOs, two of which are associated with molecular gas centered at the position of each YSO (6–1 and 6–2 in figure 2). The other source of Target 6 and one of the two sources of Target 5 are clearly not associated with molecular emission, indicating that they are more evolved YSOs or not YSOs. Mopra single pointing observations toward Target 4–10 were not able to reveal the size of the CO clouds. The present ALMA observations show a very compact cloud associated with Target 10, and some extended emission with a size of $\sim$ 5 pc is associated with Target 4, 5, and 8. For Target 4 and 7, the YSOs are located at the holes of the molecular distribution and H$\alpha$ emission with a size of $\sim$ 1 pc is associated with these sources (see figure 3). This fact suggests that significant dissipation of the associated gas has started toward these two sources. We detected Band 3 100 GHz continuum emission as shown in figure 3 toward Target 5 and 6, and detected the H40$\alpha$ radio recombination line toward Target 6.

4.2 Velocity structure of associated molecular clouds

Figures 4 and 5 show the kinematic structure of the clouds associated with the YSOs. The first moment maps show that there is no significant velocity shift toward the compact clouds such as Target 1, 2, and 10. Some extended clouds show a complex velocity field (Target 5, 6, and 8), and Target 8 shows a velocity shift by 5 km s*-1* from North-West to the South-East of the source. For linewidth maps, the linewidths tend to be larger toward the position of the YSOs. Figure 6 shows the averaged radial distributions of the linewidth centered at the position of the YSOs: we see a gradual increase of the linewidth towards the central YSO, rather than a sharp jump at small radii. Especially toward Target 8, the linewidth enhancement is obvious toward the source. Figure 7 shows the 12CO ($J$ = 2–1) spectrum toward Target 8, and there is a clear sign of blue-shifted velocity wing due to a molecular outflow. Even though this is the second most luminous target (see table 2), there is no H$\alpha$ emission detected, indicating that this source is at a very early stage of star formation. This is supported by the fact that free-free emission is undetected toward this source, based on ATCA 3 and 6 cm observations (Stephens, 2013). Figure 7 (a) shows the spatial distribution of the wing emission on the 13CO ($J$ = 2–1) integrated intensity map. The wing is very compact with a size of $<$ 0.5 pc, located a little bit north of the YSO by 0.1 pc and coincides with the molecular peak. This fact indicates that the source is very young and the YSO is dissipating the molecular gas south of the YSO, which is consistent with the non-detection of red-shifted outflow wing. We could not detect outflow wings toward other sources. The reason may be that Target 8 is the youngest among the sources, having the highest luminosity among the sources without H$\alpha$ or free-free emission, and therefore, the other sources may be more evolved, or the intrinsic outflow activity is lower. Further high spatial resolution observations would be needed to detect the outflow activities for the other sources.

4.3 LTE mass

The 13CO ($J$ = 1–0) line intensities give a better estimate of column densities than 12CO line ones whose optical depth are typically thick; the 13CO ($J$ = 1–0) line is considered to be optically thin toward typical molecular clouds. We thus assume the local thermodynamic equilibrium (LTE) to derive the column density from the 13CO luminosity. To characterize the gas condensations, we defined a clump as the region where the integrated intensity of 13CO ($J$ = 2–1) emission is stronger than the 3$\sigma$ level (see figure 2) around each YSO because 13CO ($J$ = 2–1) observations have better angular resolutions than 13CO ($J$ = 1–0) observations. The physical properties are derived for these clumps. We derived 13CO ($J$ = 1–0) column densities, $N$(13CO) [cm*-2*], by using the following relations by assuming the LTE (Wilson et al., 2014):

[TABLE]

[TABLE]

where $T_{V}(^{13}\mathrm{CO})$ is the 13CO ($J$ = 1–0) brightness temperature at the velocity. The excitation temperatures, $T_{\mathrm{ex}}$, was assumed to be 20 K (e.g., Nishimura et al. (2015)). We used the same abundance ratios of [12CO/H2] and [12CO/13CO] as Mizuno et al. (2010) for their N159 study in the LMC as 1.6 $\times$ 10*-5* and 50, respectively, to derive the column density of molecular hydrogen from $N(^{13}$CO) as follows:

[TABLE]

The masses of the clump, $M_{\mathrm{clump}}$, are calculated from the following equation:

[TABLE]

where $\mu$ is the mean molecular weight per hydrogen molecule, $m$(H2) is the H2 molecular mass, $D$ is the distance to the object of 50 kpc, and $\Omega$ is the solid angle of a pixel element. The masses of the clumps associated with the isolated high-mass YSOs range from $\sim$ 300 to $\sim$ 6,000 $M_{\odot}$ (see table 4.3).

4.4 Virial mass

The mass can be also estimated from the dynamical information of the size of the clumps and the linewidth of CO lines by assuming the virial equilibrium. We derived the virial masses from the 13CO ($J$ = 1–0) line using the procedure described by Fujii et al. (2014) as follows. The virial mass is estimated as

[TABLE]

assuming the clumps are spherical with density profiles of $\rho$ $\propto$ $r^{-1}$, where $\rho$ is the number density, and $r$ is the distance from the cloud center (MacLaren et al., 1988). Deconvolved clump sizes, $R_{\mathrm{deconv}}$, are defined as [$R^{2}_{\mathrm{observed}}$ - ($\theta_{\mathrm{HPBW}}$/2)2]1/2. We calculated the effective radius as ($A/\pi)^{1/2}$ where $A$ is the observed total cloud surface area. The virial ratio is derived by dividing the virial mass by the LTE mass. The virial ratio of the clumps associated with the isolated high-mass YSOs range from 0.6 to 2.7 (see table 4.3).

4.5 Excitation analyses using multiple transitions

The kinetic temperature and number density of the molecular gas can be estimated by observing multiple lines having different critical densities and optical depths. In order to derive the properties, we performed a large velocity gradient analysis (LVG analysis; Goldreich & Kwan (1974); Scoville & Solomon (1974)) for our CO line observations. The assumption of the LVG analysis is that the molecular cloud is spherically symmetric with uniform density and temperature, and having a spherically symmetric velocity gradient proportional to the radius. The model requires three independent parameters: the kinetic temperature, $T_{\mathrm{kin}}$, the density of molecular hydrogen, $n$(H2), and the fractional abundance of CO divided by the velocity gradient in the cloud, X(CO)/($dv/dr$). We use the abundance ratios of [12CO/H2] and [12CO/13CO] described in section 4.3. The mean velocity gradient is estimated as $dv/dr$ (km s*-1* / pc) = ${\Delta}V_{\mathrm{clump}}\;/\;2R_{\mathrm{deconv}}$, and we adopt $dv/dr$ = 2 from the typical size and linewidth of the clumps. According to the analysis using the same molecular lines by Nishimura et al. (2015), the derived density is inversely proportional to the square root of the X(CO)/($dv/dr$).

  The distributions of the kinetic temperatures and densities are shown in figures 8 and 9, and the averaged values are described in table 4.5. The density is in the range of 5.4–8.8 $\times$ 103 cm*-3*, and the kinetic temperatures are 19–30 K, which is consistent with the dust temperatures derived by Seale et al. (2014). We note that the kinetic temperatures do not show clear local enhancement toward the position of the YSOs. This implies that the heating by the newly formed YSOs is not yet significant.

5 Discussion

5.1 Clump dynamics

Table 4.3 shows that the virial ratio, $M_{\mathrm{LTE}}$/$M_{\mathrm{vir}}$, is distributed around unity. We assumed the abundance ratio of 13CO to be a typical value in the LMC as described in section 4.5, which is $\sim$ 1/3 of the Galactic value. If the abundance ratio is correct, the virial ratios indicate that the clumps are roughly gravitationally bound. Each cloud has a high-mass YSO inside so that the observation of gravitational boundedness seems to be appropriate, although gas dissipation may be starting for some of the clumps.

5.2 Photodissociation of CO

The densities of the CO clumps range from 5.4 to 8.8 $\times$ 103 cm*-3* based on the excitation analysis (see table 4.5). We also estimated the densities from the virial mass for the circularly shaped clouds, i.e., Target 1, 2, 3, and 10, under the assumption of the spherical morphology; the densities are estimated to be 3,500, 7,600, 3,800, and 16,500 cm*-3*, respectively. These values are roughly consistent with the densities derived from the excitation analysis, and thus the density estimation seems to be roughly appropriate. These densities are higher than the typical ones toward GMCs in the Galaxy by a factor of a few or more if we observe them in the same probe; for Orion molecular cloud, the average density is estimated to be 2 $\times$ 103 cm*-3* (e.g., Nishimura et al. (2015)). This may be due to the fact that in the LMC, the gas-to-dust ratio is higher by a factor of $\sim\;$2–3 and that the UV radiation is stronger than in the Galaxy, and thus CO can be more strongly photodissociated than in the Galaxy, which results in the photodissociation of lower density gas in the LMC. The ALMA observations of N83C in the SMC showed that the densities measured in 12CO are $\sim$ 104 cm*-3*, and the kinetic temperature is estimated to be 30–50 K throughout the cloud (Muraoka et al., 2017), and they argued that this is due to the effect of photodissociation in a low metallicity environment. Although the metallicity at the SMC is below the LMC (0.2$\;Z_{\odot}$; Dufour (1984); Kurt et al. (1999); Pagel (2003)), this fact indicates that CO observations can miss molecular mass in the low-metallicity/strong-UV environments in the LMC unlike the case in the Galaxy. In this case, [C i] (1–0) may be another proper probe for the molecular mass (e.g., Papadopoulos & Greve (2004)).

5.3 Compact clumps associated with the isolated high-mass YSOs

The masses of the clumps associated with the isolated high-mass YSOs range from $\sim$ 300 to $\sim$ 6,000 $M_{\odot}$ and the radii from 0.5 to 2.0 pc (see table 4.3), which is similar to those of typical nearby dark clouds where low-mass stars are forming, and much smaller than the typical high-mass star-forming clouds in the Galaxy, such as the Orion molecular cloud, which has a mass of $\sim$ 105 $M_{\odot}$ (Nishimura et al., 2015). The NANTEN and Mopra observations show that these clumps are not associated with extensive molecular gas. These facts indicate that the isolated high-mass YSOs are formed in compact and low-mass clumps. Most of the present target YSOs are located at the peak of the clumps, indicating that the YSOs were formed in the associated clumps, and the significant dissipation of the gas has not started yet toward the clumps. From the LVG analysis in section 4.5, we found that there is no kinetic temperature enhancement toward the YSOs, which is consistent with the fact that the YSO is very young and the effect of the YSO to the natal gas is not significant.

5.4 Star Formation Efficiency

As described in the previous section, the isolated high-mass YSOs in this study appear to have been formed in compact and low-mass molecular clumps, which are much smaller than the typical GMCs. In this section, we estimate the star formation efficiency of these clumps. If we assume a normal IMF, the formation of a single high-mass star should be associated with the formation of a number of low-mass stars. With Spitzer observations alone, it is hard to identify the low-mass YSOs, but we have to take the mass of the low-mass YSOs into account to estimate the star formation efficiency. One of the targets, Target 9, was observed by HST (Stephens et al., 2017), and in figure 10 we show the comparison with our CO distribution. There are a number of low-mass YSOs around the high-mass YSO, and there is a high-concentration of the low-mass YSOs toward the CO clump. Weidner et al. (2013) studied the correlation between the star cluster mass, $M_{\mathrm{ecl}}$, and the mass of the most high-mass star in a star cluster, $m_{\mathrm{max}}$. HST observations by Stephens et al. (2017) found that relatively isolated high-mass YSOs generally follow the $m_{\mathrm{max}}$–$M_{\mathrm{ecl}}$ relation. If we use this correlation, the masses of the star cluster associated with the isolated high-mass YSOs range from 110 to 310 $M_{\odot}$ (see table 5.4). From this information, we calculated the star formation efficiency with the following equation:

[TABLE]

where $M_{\mathrm{ecl}}$ is the mass of the star cluster derived from the $m_{\mathrm{max}}$–$M_{\mathrm{ecl}}$ relation (Weidner et al., 2013) and $M_{\mathrm{cloud}}$ is the LTE mass derived from 13CO ($J$ = 1–0). The star formation efficiencies are shown in table 5.4 and they range from 4 to 43%. We note that this SFE is an upper limit because a part of the natal clouds may have been already dissipated away and because some molecular gas may be CO-dark (see discussion in section 5.2). The efficiencies are high especially for the compact clumps, such as for Target 1, 2, 3, and 10. Although the high efficiency is seen toward Target 8, the gas dissipation seems to be significantly on-going for this clump (see section 4.2), which makes the mass of the associated gas small. In the case of the Orion molecular cloud, the efficiency was estimated to be $\sim$ 4.5% (Nishimura et al., 2015), which means that the star formation is efficiently on-going in the compact molecular clumps we observed in the present paper.

5.5 Enhancement of velocity width toward YSOs

As described in section 4.2, linewidths increase toward the positions of the YSOs or the peak of the CO intensities. Here, we discuss the cause of the linewidth enhancement toward the YSOs, and in particular whether it is due to the YSO activity or to the clump properties prior to star formation. The outflow from YSOs can inject momentum into the natal molecular cloud, which results in the enhancement of the linewidth of molecular clouds (e.g., Saito et al. (2007)). In the case of Target 8, the outflow wing is the apparent cause of the linewidth enhancement. For the other sources, apparent wing feature cannot be observed at our spatial resolution and sensitivity, which could mean that the current outflow activities for these YSOs are not strong enough to be detected.

  Here, we assume that the linewidth enhancement is $\sim\,$1.5 km s*-1* at the peak and located within $\sim\;$1 pc of the YSO peak, which is typical for the larger clumps in figure 6. The central $\sim\;$1 pc corresponds to a mass of $\sim\,$1,500 $M_{\odot}$ by assuming the average column density of 2$\,\times\,$1022 cm*-2*. If we assume that the linewidth enhancement is caused by the momentum input by the outflow, the momentum is calculated to be $\sim\,$1,100$\,M_{\odot}$ km s*-1* (= 1/2$\,\times\,$1,500$\,M_{\odot}\,\times\,$1.5 km s*-1*) under the assumption that the linewidth enhancement is linearly increasing toward the peak and that the column density is inversely proportional to the distance from the center. Actually, the total momentum contributing to the linewidth enhancement, the difference from 2.2 km s*-1* at 1 pc, of Target 6–1 is measured to be $\sim\,$2,200$\,M_{\odot}$ km s*-1* by summing up all the excess momenta within 1 pc of the YSO where the average column density is 4$\,\times\,$1022 cm*-2*. Nevertheless, the momentum that can be supplied by outflows is estimated to be 1–300$\,M_{\odot}\,$km s*-1* for young stars with a luminosity of 102–104 $L_{\odot}$ (e.g., Zhang et al. (2005)). This means that, even if the outflow has a momentum of the maximum of 300 $M_{\odot}$ km s*-1* and all the outflow momentum turns into that of the clump, the momentum of the outflow may not contribute to the linewidth enhancement fully. Therefore, there is a high possibility that the high-mass YSOs were formed in the turbulent clump region prior to the star formation. We discuss the interaction between multiple velocity gas components as a mechanism to realize the turbulent environment in the next section.

5.6 The cause of the high-mass star formation in an isolated environment

The importance of cloud collisions as a mechanism of the formation of high-mass stars has recently been discussed by many observational and theoretical studies. Recent mm-submm observations revealed supersonic collisions between multiple molecular clouds having different velocities toward the super star clusters in the Galaxy, e.g., Westerlund 2, RCW 38, and NGC 3603, and the observational studies suggested that the collision is followed by strong shock compression of the molecular gas that leads to the formation of massive clusters in less than 1 Myr (see Furukawa et al. (2009); Ohama et al. (2010); Fukui et al. \yearciteFukui14, \yearciteFukui16). Recent ALMA observations in the LMC also revealed highly filamentary molecular structure, which is explained by the effect of the cloud-cloud collisions (Fukui et al. (2015); Saigo et al. (2017); Fukui et al. (2018) submitted; Tokuda et al. (2018) submitted). The magnetohydrodynamical numerical simulations by Inoue & Fukui (2013) showed that compression excites turbulence and amplifies field strength, leading to high-mass star formation due to the enhancement of the Jeans mass by the collision (see also Inoue et al. (2018)).

  Recent Galactic observations have indicated that high-mass stars may form via cloud-cloud collisions between intermediate mass molecular clouds. One of the best examples is the Trifid Nebula, M20. The total molecular mass of this system is only $\sim$ 103 $M_{\odot}$, but it contains a single O star along with hundreds of low-mass stars. With a spatial resolution of 1 pc, \authorciteTorii11 (\yearciteTorii11, \yearciteTorii17) identified two molecular gas components with different radial velocities toward M20 with NANTEN2, Mopra and ASTE, and proposed that a recent collision between two clouds increased the effective Jeans mass and led to the formation of the O star.

  In the present sample, we detected relatively complex velocity field toward a few sources. Figure 11 shows the CO intensity maps integrated with different velocity ranges. For Target 5, there are two filamentary components having different velocities separated by $\sim$ 3 km s*-1*, and the high-mass YSO is formed at the position where the two filamentary clouds are merged. A similar configuration is also seen toward Target 6. At least for these clouds, there is a possibility that the interaction of molecular clouds enhanced the star formation efficiency there, and high-mass YSOs are formed therein.

For the other targets, multiple velocity components or significant velocity gradients are not observed, and there is no evidence of the molecular gas interaction for these sources. However, the high-mass YSOs are formed in isolated compact clouds. Then, what is the mechanism to initiate the high-mass star formation in such low-mass CO clouds? Figure 12 shows the Herschel 350 $\mu$m images toward all the sources (Meixner et al., 2013). It is clear that the associated 350 $\mu$m emission is much more extended than the CO distribution, and the high-mass YSOs tend to be located at the edge of the 350 $\mu$m clouds. Most parts of the 350 $\mu$m cloud are not seen in CO, which indicates that the area is of low-density, possibly H i gas. In this case, H i observations would be quite important to investigate the dynamical information of the surrounding gas around the present targets. The spatial resolution of the current H i observations (Kim et al., 2003) is not high enough to spatially resolve the structures associated with the compact CO clouds. The association with the edge of the 350 $\mu$m clouds may indicate the existence of an effect from external sources. Figure 1 shows that many of the targets are seem to be associated with H i supershells (Dawson et al., 2013). These facts imply that the past explosive events are affecting star formation around the targets, and the future high-resolution H i observations with an angular resolution of $\lesssim$ 10 arcsec, such as ASKAP (Duffy et al., 2012) are highly anticipated to investigate the gas dynamics.

6 Summary

We have carried out ALMA observations toward ten high-mass YSO candidates in isolated environments in the LMC at the highest angular resolution of 1.4 arcsecs, corresponding to a spatial resolution of 0.35 pc at the distance of the LMC. The targets are at least 200 pc away from any of the NANTEN clouds whose detection limit is a few $\times$ 104 $M_{\odot}$, and are detected by the pointed Mopra observations with a spatial resolution of 7 pc. The aim of our observations was to investigate high-mass star formation in a simple environment away from any molecular complexes, avoiding external energetic events and contamination along the same line of sight. We have obtained the following results:

Compact molecular clouds are associated with all the targets, indicating that these targets are bona fide YSOs formed in the compact clouds. The masses of the clouds range from 2.8 $\times$ 102 to 6.5 $\times$ 103 $M_{\odot}$ and the radii from 0.5 to 2.0 pc. They seem to be gravitationally bound. 2. 2.

The excitation analyses using 13CO ($J$ = 1–0), 12CO ($J$ = 2–1) and 13CO ($J$ = 2–1) show that the density is $\sim$ 6 $\times$ 103 cm*-3* and the temperature is $\sim$ 20 K. The high density and temperature compared with high-mass star-forming clouds in the Galaxy may indicate that the photodissociation of CO is more severe due to the LMC’s lower metallicity. There is no apparent enhancement of the temperature toward the position of the YSOs, indicating that the YSOs are very young and the effect of the YSO on the gas is not significant. 3. 3.

The star formation efficiency is calculated by assuming a standard IMF to derive the total stellar mass and is estimated to range from several percent to as high as $\sim$ 40%, which means that the star formation is efficiently on-going in at least some of the observed compact molecular clumps. 4. 4.

Toward two targets, there is a possibility that interaction of molecular clouds enhanced the star formation activity as suggested by entangled filaments with different velocities. While the other sources have no clear indication of the interaction of the CO gas, the interaction of lower density gas may be a cause of the high star formation activity as shown by the fact that most of the high-mass YSOs are located at the edge of the 350 $\mu$m clouds.

{ack}

This paper makes use of the following ALMA data: ADS/ JAO.ALMA#2013.1.00287.S. ALMA is a partnership of the ESO, NSF, NINS, NRC, NSC, and ASIAA. The Joint ALMA Observatory is operated by the ESO, AUI/NRAO, and NAOJ. This work was supported by NAOJ ALMA Scientific Research Grant Numbers 2016-03B and JSPS KAKENHI (Grant No. 22244014, 23403001, 26247026, and 18K13582). The Mopra Telescope is part of the Australia Telescope, which is funded by the Commonwealth of Australia for operation as a National Facility managed by CSIRO. Cerro Tololo Inter-American Observatory (CTIO) is operated by the Association of Universities for Research in Astronomy Inc. (AURA), under a cooperative agreement with the National Science Foundation (NSF) as part of the National Optical Astronomy Observatories (NOAO). M. Meixner and O. Nayak were supported by NSF grant AST-1312902.