EDP Sciences
Home All issues Volume 515 (June 2010) A&A, 515 (2010) A55 Full HTML
Free Access
Issue
A&A
Volume 515, June 2010
Article Number A55
Number of page(s) 18
Section Interstellar and circumstellar matter
DOI https://doi.org/10.1051/0004-6361/200913632
Published online 09 June 2010
A&A 515, A55 (2010)

The earliest phases of high-mass star formation: the NGC 6334-NGC 6357 complex[*],[*],[*]

D. Russeil1 - A. Zavagno1 - F. Motte2 - N. Schneider2 - S. Bontemps3 - A. J. Walsh4

1 - Laboratoire d'Astrophysique de Marseille - UMR 6110, CNRS - Université de Provence, 13388 Marseille Cedex 13, France
2 - Laboratoire AIM, CEA/DSM - INSU/CNRS - Université Paris Diderot, IRFU/Service d'Astrophysique, CEA-Saclay, 91191 Gif-sur-Yvette Cedex, France
3 - Laboratoire d'Astrophysiaue de Bordeaux, OASU - UMR 5804, CNRS - Université de Bordeaux 1, 2 rue de l'Observatoire, BP 89, 33270 Floirac, France
4 - Centre for Astronomy, School of Engineering and Physical Sciences, James Cook University, Townsville, QLD, 4811, Australia

Received 10 November 2009 / Accepted 11 February 2010

Abstract
Context. Our knowledge of high-mass star formation has been mainly based on follow-up studies of bright sources found by IRAS, and has thus been incomplete for its earliest phases, which are inconspicuous at infrared wavelengths. With a new generation of powerful bolometer arrays, unbiased large-scale surveys of nearby high-mass star-forming complexes now search for the high-mass analog of low-mass cores and class 0 protostars.
Aims. Following the pioneering study of Cygnus X, we investigate the star-forming region NGC 6334-NGC 6357 ($\sim$1.7 kpc).
Methods. We study the complex NGC 6334-NGC 6357 in an homogeneous way following the previous work of Motte and collaborators. We used the same method to extract the densest cores which are the most likely sites for high-mass star formation. We analyzed the SIMBA/SEST 1.2 mm data presented in Munoz and coworkers, which covers all high-column density areas ( $A _{v} \ge 15$ mag) of the NGC 6334-NGC 6357 complex and extracted dense cores following the method used for Cygnus X. We constrain the properties of the most massive dense cores (M > 100 $M_\odot $) using new molecular line observations (as SiO, N2H+,H13CO+, HCO+ (1-0) and CH3CN) with Mopra and a complete cross-correlation with infrared databases (MSX, GLIMPSE, MIPSGAL) and literature.
Results. We extracted 163 massive dense cores of which 16 are more massive than 200 $M_\odot $. These high-mass dense cores have a typical FWHM size of 0.37 pc, an average mass of $M \sim 600$ $M_\odot $, and a volume-averaged density of $\sim$ $1.5 \times 10 ^{5}$ cm-3. Among these massive dense cores, 6 are good candidates for hosting high-mass infrared-quiet protostars, 9 cores are classified as high-luminosity infrared protostars, and we find only one high-mass starless clump ($\sim$0.3 pc, $\sim$ $4 \times 10^{4}$ cm-3) that is gravitationally bound.
Conclusions. Since our sample is derived from a single molecular complex and covers every embedded phase of high-mass star formation, it provides a statistical estimate of the lifetime of massive stars. In contrast to what is found for low-mass class 0 and class I phases, the infrared-quiet protostellar phase of high-mass stars may last as long as their more well known high-luminosity infrared phase. As in Cygnus X, the statistical lifetime of high-mass protostars is shorter than found for nearby, low-mass star-forming regions which implies that high-mass pre-stellar and protostellar cores are in a dynamic state, as expected in a molecular cloud where turbulent and/or dynamical processes dominate.

Key words: dust, extinction - H II regions - stars: formation - radio continuum: ISM - submillimeter: ISM - radio lines: general

1 Introduction

High-mass (O- or B-type) stars play a major role in the energy budget and enrichment of galaxies, but their formation remains poorly understood. High-mass stars probably form in massive dense cores by the powerful accretion of gas onto a protostellar embryo (e.g. Beuther & Schilke 2004). However, the physical origin of these high accretion rates remains unclear with multiple mechanisms proposed including a high degree of turbulence (McKee & Tan 2002), converging flows (Heitsch et al. 2008), cloud collisions (Bonnell & Bate 2002), and competitive accretion on large scales (Bonnell et al. 2006).

From a purely observational point of view, the evolutionary sequence leading from clouds to OB stars is far from being well constrained. For instance, the existence and lifetime of the infrared (IR)-quiet phase analog to low-mass class 0 protostars and pre-stellar cores for high-mass stars is still a matter of debate (Motte et al. 2007). Moreover, the exact ordering and overlap of the different phases/diagnostics (pre-stellar, cold/infrared-quiet protostar, hot core, OH/H2O/CH3OH masers, warm/infrared-bright sources, HMPOs, hypercompact H II, UCH II regions) needs to be fully determined. Therefore, it is of crucial importance to build representative and unbiased samples of high-mass pre-stellar and protostellar objects, in large, nearby, high-mass star formation complexes.

(Sub)millimeter continuum mapping is the perfect tool for systematically searching the earliest phases of star formation since dust emission is mostly optically thin and directly traces cold, high-mass dense cores on the verge of collapse, and young protostars already collapsing. Dust continuum surveys in the submm range are efficient for recognizing high-mass young stellar objects if the spatial resolution is high enough to discriminate high-mass dense cores from their surroundings. The typical size of dense cores is 0.1-0.2 parsec (e.g. Bergin & Tafalla 2007; Zinnecker & Yorke 2007), and the typical highest spatial resolution achievable on millimeter telescopes is of the order of 10 $\hbox{$^{\prime\prime}$ }$, which translates into 0.15 pc at 3 kpc. We thus propose that the massive complexes within 3 kpc offer a unique opportunity to study the earliest phases of high-mass stars.

The twin molecular complex NGC 6334-NGC 6357 (distance 1.7 kpc, size $\sim$68 pc $\times$ $\sim$80 pc) is one of the most prominent of these massive complexes since it includes high-mass star formation at different evolutionary stages (cores, embedded compact H II regions, evolved optical H II regions). The high-column density parts of NGC 6334-NGC 6357 were delineated using near-IR extinction maps produced from 2MASS and CO data (Schneider et al. 2010). The extinction map was derived from the publicly available 2MASS point source catalog by calculating the average reddening of stars with a method adapted from those described in Lada et al. (1994), Lombardi & Alves (2001), and Cambrésy et al. (2006). The extinction was derived from the reddening of both [J-H] and [H-K] colors. From the stellar population model of Robin et al. (2003), a predicted density of foreground stars was obtained at the complex distance. For each 2 arcmin size pixel of the map, this expected number of foreground stars was removed from the least reddened 2MASS sources before deriving the average reddening. Figure 1 (up) shows a global view of this complex in an H$\alpha $ image from the AAO/UKST H$\alpha $ survey of the southern Galactic plane (Parker et al. 2005) with the extinction map (angular resolution of 2') superimposed. Near the H II regions, the extinction caused by dust corresponds well to the extinction zones seen in H$\alpha $ (for example, the ``elephant trunk'' features in NGC 6334 seen at the border of the H$\alpha $ extinction zone). Figure 1 (down) reproduces the 1.2 mm continuum map (angular resolution 24'') obtained by Muñoz et al. (2007), where it becomes obvious that the highest column density regions ( Av > 30 mag equivalent to a hydrogen column density larger than 3$\times$1022 cm-2) correspond to peak emission in dust continuum. Figure 2 presents zooms at 8 $\mu $m (mosaics of GLIMPSE residual images produced by the GLIMPSE team) of NGC 6334 and NGC 6357. The 8 $\mu $m channel of the Spitzer IRAC instrument is dominated by PAH emission, excited by nearby ultra-violet radiation. In these images, many 1.2 mm continuum emission peaks clearly appear in the direction to IR extinction patches but only a few correspond to 8 $\mu $m emission. For the central part of NGC 6334, Burton et al. (2000) detected evidence of dark lanes parallel to the ridges and loops of 3.3 $\mu $m PAH emission. Because of the lower resolution of our extinction map, we cannot compare it directly with, but, if these dark lanes correspond well to optical extinction, only those aligned along the main ridge can be related to the 1.2 mm dust emission.

NGC 6334 is a well studied region (as reviewed by Persi & Tapia (2008). At 1.2 mm, the central part of NGC 6334 consists of a 10 pc long filament associated with large extinction. At least seven sites of high-mass star formation are observed (e.g. Loghran et al. 1986), recognizable in terms of water masers, H II regions (e.g. Carral et al. 2002), and molecular outflows. Previous (sub)millimeter continuum studies of NGC 6334 focused mainly on the northern portion of the filament containing sources I and I(N) (e.g. Sandell 2000). These sources exhibit outflows and have masses of between 200 and 400 $M_\odot $. The molecular emission associated with NGC 6334 has a mean velocity of -4 km s-1 with a velocity gradient (Kraemer & Jackson 1999) from the center ( $V_{\rm LSR} \sim 0$ km s-1) to the edge of the ridge ( $V_{\rm LSR} \sim -11$ km s-1). In addition to the far-IR sources, NGC 6334 is a grouping of the well-known H II regions GUM 61, GUM 62, GUM 63, and GUM 64. On the basis of the distance modulus determined by Persi & Tapia (2008), a mean distance of $1.63 \pm 0.3$ kpc is obtained for the exciting stars of these H II regions.

\begin{figure} \par\includegraphics[width=8.5cm,clip]{13632fg1.ps} \includegraphics[width=8.5cm,clip]{13632fg2.ps} \end{figure} Figure 1:

Top: H$\alpha $ image (UKST H$\alpha $ survey, Parker et al. 2005) of NGC 6334 and NGC 6357 with extinction isocontours overlaid on it. Red contours correspond to Av between 18 mag and 30 mag (by step of 2 mag) the blue contour corresponds to Av = 16.5 mag. Bottom: the 1.2 mm continuum emission from Muñoz et al. (2007) with extinction isocontours overlaid.

Open with DEXTER

NGC 6357 is a large H II region exhibiting an annular morphology in the radio and optical (e.g. Lortet et al. 1984). Far-IR continuum data detected several luminous ( $L \sim 10^5~L_{\odot}$) embedded sources coinciding with 12CO(1-0) and radio continuum emission peaks (McBreen et al. 1983). In contrast to NGC 6334, almost no water maser emission is found in NGC 6357 (Healy et al. 2004). The brightest H II region (G353.2+0.9) has a sharp boundary facing the massive open cluster Pismis 24. The distance of NGC 6357 is usually taken to be that of Pismis 24: a distance of 1.7 kpc was determined by Neckel (1978) and Lortet et al. (1984), a distance of 1.1 kpc is obtained by Conti & Vacca (1984) for a Wolf-Rayet star belonging to the cluster, and Massey et al. (2001) give a distance of 2.5 kpc.

The kinematics of NGC 6357 is around -4 km s-1 (ionized gas, Caswell & Haynes 1987), which is similar to the mean velocity of NGC 6334 and strongly suggests that both regions are at the same distance (1.7 kpc). In addition, the extinction map (Fig. 1) and the morphology of the 1.2 mm SIMBA emission tend to indicate that NGC 6334 and NGC 6357 are connected by a filamentary structure, thus again suggesting that both regions belong to the same complex. We therefore adopt a common distance for NGC 6334 and NGC 6357 of 1.7 kpc.

In this paper, we follow the approach of Motte et al. (2007) to characterize the star formation content of this high-mass star-forming complex and constrain the evolutionary sequence of high-mass star formation. Motte et al. (2007) applied a specific extraction method to a 1.2 mm continuum map of the Cygnus X complex to extract dense cores. They complemented these data with SiO(2-1) follow-up observations of the most likely progenitors of high-mass stars, and determined the main characteristics of the millimeter sources by searching for signposts of protostellar activity including SiO emission (a tracer of outflow activity). In Cygnus X, they identified 33 high-mass dense cores of mean size, mass, and density of 0.13 pc, 91 $M_\odot $, and $1.9 \times 10 ^{5}$ cm-3, respectively. Seventeen dense cores were found to harbor high-mass protostars in their IR-quiet phase. Their unbiased survey of the high-mass young stellar objects in Cygnus X demonstrates that high-mass IR-quiet protostars do exist, and that their lifetimes should be comparable to those of more evolved high-luminosity IR protostars. By comparing the number of high-mass protostars and OB stars across the entire Cygnus X complex, a statistical lifetime of $3 \times 10^4$ yr for high-mass protostars was estimated, which is one order of magnitude smaller than the lifetime of nearby low-mass protostars, and in agreement with the free-fall time of Cygnus X dense cores.

\begin{figure} \par\includegraphics[width=8.5cm,clip]{13632fg3.ps} \includegraphics[width=8.5cm,clip]{13632fg4.ps} \end{figure} Figure 2:

GLIMPSE 8 $\mu $m residual (corrected for point sources) mosaics (produced and delivered by the GLIMPSE team) of NGC 6334 (top) and NGC 6357 ( bottom). The 1.2 mm continuum emission isocontours are overlaid.

Open with DEXTER

2 The data

2.1 Dust continuum data at 1.2 mm

The 1.2 mm (250 GHz) continuum observations of NGC 6334 and NGC 6357 and the data reduction were performed by Muñoz et al. (2007). The observations were completed using the 37-channel SEST Imaging Bolometer Array (SIMBA) in the fast-mapping mode. The angular resolution is 24''(full width at half maximum) corresponding to a spatial resolution of 0.2 pc at 1.7 kpc. The typical pointing accuracy of the SEST telescope is 3''-5'' and SIMBA observations usually have a relative flux uncertainty of 20% (Faúndez et al. 2004). The 1.2 mm mosaic has a relatively homogeneous rms noise of $\sim$25  $\mbox{mJy}~\mbox{beam}^{-1}$. The filamentary morphology of the emission becomes evident in Fig. 3. The strongest emission peaks are associated with NGC 6334 and NGC 6357. A filamentary structure, which we call ``inter-region filament'', is identifiable between both regions.

\begin{figure} \par\includegraphics[angle=-90]{13632fg5.eps} \end{figure} Figure 3:

Greyscale image of the 1.2 mm continuum emission toward NGC 6334 and NGC 6357 observed by Muñoz et al. (2007) with SIMBA (SEST). The white crosses indicate the 42 dense cores we have extracted that are more massive than 100 $M_\odot $. The core number is indicated as well, referring to the fragment number given Table 6.

Open with DEXTER

2.2 Molecular line observations

We performed pointed observations toward the 42 most massive dense cores (M > 100 $M_\odot $) in the SiO (v = 0, J=2-1) transition using the 22 m Mopra telescope. The observations were performed with the 3mm receiver and the Mopra spectrometer (MOPS) in the ``zoom mode'' that allows to observe simultaneously up to 16 different frequencies. The pointing was regularly checked by observing SiO masers, typical corrections required were smaller than 5''. The typical system temperature was 190 K. The velocity resolution is 0.12 km s-1, the beam width at 86 GHz is 36'', and a main beam efficiency $\tau_{\rm MB}$ of 0.49 (Ladd et al. 2005) was adopted. All observations were performed in position switching mode with the off-position a few arcminutes away. A total integrated time of 11 min on-source was used to achieve a rms of $\sim$0.05 K. Initial spectral processing (base-removal and calibration onto a T*A scale) was performed with the ASAP software[*]. In addition to the SiO line, other molecular species (N2H+ (1-0), H13CO+ (1-0), HCN (1-0), HNC (1-0), 13CS (2-1), HCO+ (1-0), and CH3CN (5-4)) were observed simultaneously (example spectra are presented in Fig. 4).

\begin{figure} \par\mbox{\includegraphics[width=6cm,clip,angle=-90]{13632fg6.ps}... ...fg8.ps}\includegraphics[width=6cm,clip,angle=-90]{13632fg9.ps} } \end{figure} Figure 4:

Example of spectra obtained for the dense core 61 (the spectra for the other dense cores are given in the Appendix). All profiles (except for N2H+) have been smoothed to a velocity resolution of 0.3 km s-1. Upper panels: SiO ( left) and N2H+ ( right). Lower panels: on the left HCO+ superimposed on H13CO+ (thick line) and on the right HNC superimposed on 13CS (thick line).

Open with DEXTER

3 Analysis and results

3.1 The compact dense cores at 1.2 mm

Muñoz et al. (2007) identified a total of 181 clumps with sizes ranging from 0.1 to 1 pc, and a median of 0.36 pc, and masses ranging from 3 to 6000 $M_\odot $ with a mean value of 170 $M_\odot $. They extracted these clumps from NGC 6334 and the ``inter-region filament'' using the Clumpfind algorithm (Williams et al. 1994). Clumpfind defines a source as an emission peak that extends to a defined contour level (in Muñoz et al. 2007, 3 $\sigma=75~\mbox{$\mbox{mJy}~\mbox{beam}^{-1}$ }$) and separates it from other peaks/sources at saddle points located below a given level (here again 3$\sigma$) from the peaks. However, this approach is not well suited to identifying compact and dense sources and generally identifies cloud structures with a large variety of sizes, which could thus be progenitors of single stars, small groups, or even clusters of stars.

To build a more homogeneous sample of cloud fragments that would be compact enough to be called dense cores ($\sim$0.1 pc, see the terminology given in e.g. Motte et al. 2007), we use the compact source extraction method developed by Motte et al. (2003).

Our reason for excluding diffuse molecular cloud structures from our analysis is to focus on the most likely sites of current intermediate- to high-mass star formation (here called massive dense cores). The procedure described in more detail in Motte etal. (2007) uses a multi-resolution analysis (Starck & Murtagh 2006) and the Gaussclumps program (Stuzki & Gusten 1990; Kramer et al. 1998). The multi-resolution technique is based on wavelet transformations and allows us to set a cutoff length-scale that separates sources observed on different spatial scales of the 1.2 mm map. We choose to filter out spatial scales larger than 1 pc, corresponding to ``clumps'' according to the terminology of Williams et al. (2000) and Motte et al. (2007). The compact ($\le$1 pc) fragments are then identified above the $6\sigma$ ($\sim$ $150~\mbox{$\mbox{mJy}~\mbox{beam}^{-1}$ }$) level in the filtered 1.2 mm map by a 2D-Gaussian fitting. As a consequence, all sources are on average compact and thus dense. We extracted 163 dense cores in total, their sizes and 1.2 mm fluxes being given in Table 6. For each core, the total mass (dust + gas) was derived by assuming that the 1.2 mm emission consists of thermal dust emission that is largely optically thin

\begin{displaymath}M_{1.2~{\rm mm}} = \frac{{S^{\rm int}_{1.2~{\rm mm}}} d^2}{\k... ...2~{\rm mm}} B_{1.2~{\rm mm}}(\mbox{$T_{\mbox{\tiny dust}}$ })} \end{displaymath} (1)

where $\kappa_{1.2~\rm mm}$ is the dust opacity per unit mass column density at 1.2 mm and $B_{\rm 1.2~mm}(\mbox{$T_{\mbox{\tiny dust}}$ })$ is the Planck function for a dust temperature $T_{\mbox{\tiny dust}}$. The dust mass opacity (including dust properties and gas-to-dust mass ratio) is likely to vary with density, temperature, and the evolutionary state of the emitting medium (Henning et al. 1995). Models of dust in low-mass protostellar cores (e.g. Ossenkopf & Henning 1994) suggest that a value of $\kappa_{1.2~{\rm mm}} = 0.01$ cm2 g-1 is well suited to cool (10-30 K) and high-density ( $n_{H_{2}} \geq 10^{5}$ cm-3) cloud fragments. We therefore choose a dust opacity per unit (gas + dust) mass column density of $\kappa_{1.2~{\rm mm}} = 0.01$ cm2 g-1, although we note that this value is uncertain by a factor of 2.

The temperature to be used in Eq. (1) is the mass-weighted dust temperature of the cloud fragments, whose value could be determined from gray-body fitting of their spectral energy distributions. This measurement has been performed only for NGC 6334I(N) by Sandell (2000), who found 30 K. In addition, Matthews et al. (2008) inferred a mean temperature (derived from integrated flux density ratio 450 $\mu $m/850 $\mu $m) of $\sim$25 K for clumps in NGC 6334. This agrees with Motte et al. (2007), who demonstrate that the temperature generally measured in dense fragments forming high-mass stars is in the range of 15-25 K. We then assume $\mbox{$T_{\mbox{\tiny dust}}$ }=20~\mbox{K}$ in Eq. (1). The mass estimate is correct to within a factor of 2 due to the uncertainty in the dust opacity, while an uncertainty of 30% is implied by a temperature change from 15 to 25 K.

The volume-averaged densities are then estimated to be

\begin{displaymath}\langle n_{\rm H_{2}}\rangle = \frac{M_{1.2~{\rm mm}}}{{\frac{4 \pi}{3}} {FWHM}^{3}} \end{displaymath} (2)

where $M_{1.2~{\rm mm}}$ is the mass derived by Eq. (1) and FWHM is the geometric mean of full widths at half maximum, determined by Gaussian fits. Using a radius equal to the FWHM in Eq. (2), we can accurately determine the volume-averaged density because the flux (and thus the mass) measured within such an aperture corresponds to >$98\%$ of the integrated flux (respectively total mass) of Gaussian cloud structures. The often-used, beam-averaged peak density would of course be higher, but is less relevant when estimating physical constraints such as the free-fall time.

As no radio continuum data of sufficiently high resolution around 5 GHz is available, the contribution of the free-free emission from the gas to the mass calculation has been estimated from both the NVSS 1.4 GHz (beam size $45'' \times 45''$) image (Condon et al. 1998) and the 1.6 GHz (beam size $26'' \times 20''$) image (Muñoz et al. 2007). Since the 1.6 GHz image unfortunately covers only NGC 6334, we used the NVSS 1.4 GHz for NGC 6357. We then measured the radio flux in an aperture, convolved to the beam of the image, of the cores. We extrapolated this flux to 1.2 mm by assuming a power law dependence of $S_{\nu} \sim \nu^{-0.1}$, subtracted this flux from $S^{\rm int}_{1.2~{\rm mm}}$, and recalculated the mass and density (Table 6) from the corrected value of $S^{\rm int}_{1.2~{\rm mm}}$. Since the NVSS 1.4 GHz survey also covers NGC 6334, we compared the mass obtained from the 1.4 and 1.6 GHz images and found good agreement (the linear regression gives a slope of $0.95 \pm 0.0087$ and constant term of $1.07 \pm 1.9$), which justifies our use of the 1.4 GHz data for NGC 6357. The mass correction due to the free-free emission represents on average 6% of the $M_{1.2~{\rm mm}}$. However, at 1.4 GHz and 1.6 GHz the emission is not always in the optically thin regime. This implies that our free-free correction certainly underestimates the free-free contribution at 1.2 mm and that in turn, the corrected masses are overestimated.

Table 6 summarises the properties of the extracted cores: core number (Col. 1), core coordinates (Col. 2), $S^{\rm peak}_{1.2~{\rm mm}}$ (Col. 3), the deconvolved FWHM size (Col. 4), $S^{\rm int}_{1.2~{\rm mm}}$ (Col. 5), free-free corrected mass (Col. 6) and free-free corrected density (Col. 7).

Amid the 181 clumps extracted by Muñoz et al. (2007), 68 have associations with one or several of our cores (only 11 clumps of Muñoz et al. are associated with several cores). Our cores are all found in the most massive and densest clumps of Muñoz et al. (2007): the mean mass and density of clumps with associated cores are 451 $M_\odot $  and $\sim$ $1.4 \times 10^{4}$ cm-3, respectively while for clumps without associated cores they are 50 $M_\odot $ and $\sim$ $8 \times 10^{3}$ cm-3, respectively.

3.2 Molecular lines towards high-mass 1.2 mm dense cores

N2H+ is a good tracer of the highest density and cold regions of clouds because it appears to be mostly optically thin and less depleted onto dust grain surfaces than CO and other molecular species (Tafalla et al. 2002, 2004, 2006). The isolated 101-012 line width of the hyperfine structure is used to estimate the virial mass and help us to quantify infall motions by comparison with optically thick lines such as HCO+ and HNC.

The opacity and excitation temperature of the N2H+ 1$\to$0 line was determined using the known hyperfine (HFS) structure pattern of this transition. The 6 relative distances and intensities of HFS components of the $1 \to 0$ line were given as input to a simultaneous Gaussian fit to all components, assuming equal excitation temperature. For most of the sources, one velocity component, and thus a single Gaussian HFS fit, was performed, using the ``Method HFS'' feature of GILDAS[*]. Few sources were found to have two velocity components and only one to have three components (core 66). However, some of the densest and most massive cores (cores 60, 61, and 63) exhibit a non-LTE HFS pattern (the relative intensities have not been correctly fitted) and rather broad linewidths. These sources probably consist of several cloud fragments along the line of sight which thus blending their N2H+ emission lines. In Table 2, we indicate the excitation temperature (equal for all components by definition), the line center velocity of the molecular gas bulk emission determined from the 123-013 (93.173.809 MHz) component of N2H+(the receiver was tuned to this frequency), the main beam temperature, the full line width, and the opacity of the isolated 101-012 line component. The typical uncertainty estimated for these determined parameters is 15%.

Table 2:   Fitted parameters for the isolated N2H+ 101-012 line component using the ``method HFS'' from the GILDAS software

Both HCO+ and H13CO+ are usually used to probe the kinematics of the extended envelope and hence the bulk motion of the gas in the region. Because they are usually optically thin, 13CS and H13CO+, are often used to establish the systemic velocity of the dense cores. All the 42 high-mass dense cores observed (Table 3) were found to have velocities in agreement with the kinematics of NGC 6334 and NGC 6357.

The asymmetric rotator CH3CN is a good tracer of the conditions in ``hot cores'' owing to its favorable abundance and excitation in warm ($\ge$100 K) and dense ($\ge$105 cm-3) regions. It traces objects that are internally heated and its emission is more intense and more commonly detected towards ultra-compact H II regions than towards isolated maser sources (Purcell et al. 2006). Assuming local thermal equilibrium and optically thin lines, the relative intensities of the K components yield a direct measure of the kinetic temperature (Purcell et al. 2006). Amid the 42 massive cores, 8 have well detected CH3CN, while 13 have no detectable CH3CN and 21 have barely detectable CH3CN. From CH3CN rotational diagrams, we can establish the temperature of the hot component within the 8 dense cores for which several K components of CH3CN are observed to be: 72.5 K for core 62; 62 K for core 54; 47 K for core 60; $\sim$42 K for cores 63, 35, and 29; 31 K for core 3; and 20 K for core 61. All these dense cores are associated with stellar activity. The temperature of the hot core inferred from CH3CN (between 20 and 70 K) is normally higher than the dust temperature mass-averaged over 0.2 pc. This does not contradict our assumption that the mass-averaged temperature over cloud structures is 20 K on average and probably slightly higher only for the above 8 sources.

4 Origin and characteristics of high-mass dense cores

The NGC 6334 - NGC 6357 complex has already formed generations of high-mass stars since it contains OB stars and H II regions. Here we focus on its ability to form high-mass stars in the (near) future by making a census of high-mass prestellar and protostellar dense cores and comparing with the high mass star formation in Cygnus X. For Cygnus X, a lower mass limit of 40 $M_\odot $ was adopted for dense cores to have a high probability of forming 10-20 $M_\odot $ of stars, including at least one high-mass star (Motte et al. 2007). From Table 6, we note that the dense cores of NGC 6334-NGC 6357 are on average $\sim$3 times larger than the dense cores found in Cygnus X by Motte et al. (2007). This is mainly because the physical resolution is a factor of two poorer in NGC 6334-NGC 6357 SIMBA images than in Cygnus X MAMBO2 ones, which needs to be taken into account when adopting a low-mass limit. If we assume[*] that the cores have a $\rho(r) \propto r^{-2}$ density gradient, for high-mass stars to be formed with a similar probability, the lower mass limit for the fragments extracted in NGC 6334-NGC 6357 needs to be $\sim$3 times higher than the Cygnus X one. We therefore choose a lower mass limit of 100 $M_\odot $ to select a sample of high-mass dense cores that are expected to be good candidate progenitors of high-mass stars.

Table 3:   Turbulence support, SiO outflow, and gravitational infall of the most massive dense cores of NGC 6334 - NGC 6357.

The 42 compact cloud fragments identified in NGC 6334-NGC 6357 with masses higher than 100 $M_\odot $ have sizes ranging from 0.16 to 0.63 pc with a mean size of $\sim$0.36 pc (see Table 6) and mean volume-averaged density of $\sim$ $7 \times 10^{4}$ cm-3. After correcting for the free-free contamination, the mass range of our sample is between 101 $M_\odot $ and 1951 $M_\odot $. The cloud structures extracted in NGC 6334-NGC 6357 are thus slightly smaller in size and 10 times denser than typical HMPOs (high-mass protostellar objects, Beuther et al. 2002) and IRDC (infrared dark cloud, Rathborne et al. 2006) clumps (HMPOs and IRDCs have typical sizes of 0.5 pc and their respective typical mass and density are 290 $M_\odot $  and 150 $M_\odot $  and $8.5 \times 10 ^{3}$ cm-3 and $5.9 \times 10 ^{3}$ cm-3; see Table 4 of Motte et al. 2007). Therefore, the NGC 6334 - NGC 6357 dense cores are on average, more likely host precursors of high-mass stars.

4.1 Correlation of 1.2 mm dense cores with signposts of stellar activity

To determine the origin of the massive cores detected at 1.2 mm, we studied their spatial association with the following signposts of stellar activity.

4.1.1 Association with Spitzer/GLIMPSE

The position of sources in the [3.6]-[4.5] versus [5.8]-[8] diagram is related to the presence of circumstellar dust. The principal classification scheme for low-mass star formation is the class 0-I-II-III system, which notably characterizes objects in terms of their IR excesses or SEDs (e.g. Adams et al. 1987; André et al. 1993, 2000). Class 0 and I objects are understood to be protostars surrounded by dusty infalling envelopes, which would explain both the relatively strong far-IR emission and significant near-IR extinction from their envelopes. They are deeply embedded objects with a spectral energy distribution (SED) that peaks in the submillimeter or the far-IR, indicating that the source of emission is cold dust. Class II systems are optically visible stars with disks, and thus exhibit a smaller IR excess and near-IR extinction (unless observed edge-on). Class III objects are essentially stars without significant amounts of circumstellar dust.

To establish an association of dense cores with class I and class II objects (see Table A.1), we used the IRAC/GLIMPSE point source catalogue (http://irsa.ipac.caltech.edu/data/SPITZER/GLIMPSE/). The class is defined using [5.8]-[8.0] and [3.6]-[4.5] colors and based on models of disk or envelopes or both, Allen et al. (2004) defined the criteria:

class I: [5.8]-[8.0] $\geq$ 0.35 and [3.6]-[4.5] > 0.4, 
class II: [5.8]-[8.0] $\geq$ 0.35 and [3.6]-[4.5] $\leq$ 0.4.

The distribution of class II sources in the [3.6]-[4.5] color diagram depends mainly on the accretion rate, while the disk inclination and the grain properties explain the spread in the [5.8]-[8.0] color. The larger distribution spread in both colors, [3.6]-[4.5] and [5.8]-[8.0], for class I objects is due to the higher temperature and density of the envelope. However, an extinction of 30 mag causes a shift of about 0.4 mag toward redder [3.6]-[4.5] indices, implying that class II sources can fall in the class I area. Based on the mean extinction found for the dense cores (estimated from the extinction map), we can consider a source as a true class I, if its [3.6]-[4.5] color is above 0.8. Hartmann et al. (2005) confirm the criteria of Allen et al. (2004) on the basis of observations of the Taurus pre-main sequence stars.

We establish a physical association with Spitzer/GLIMPSE point sources when the Spitzer/GLIMPSE point source(s) fall within 6''(i.e. upper limit to the sum of the maximum pointing error 5'' and the typical GLIMPSE position accuracy 0.3'') of the dense core center. This is justified because high-mass protostars are expected to form at the center of the dense core. In this way, only 10 dense cores have such an association. They are all associated with true class I objects except core 152, which is associated with a class II object.

Table 4:   High-mass young stellar objects in NGC 6334-NGC 6357 (M > 200 $M_\odot $) compared to Cygnus X cores and HMPOs clumps.

The Spitzer/GLIMPSE point sources sample may be contaminated by galaxies. Chavarria et al. (2008) established a criterion to separate young stellar objects from galaxies, which is roughly that [4.5]< 14.5. Assuming that it can be applied to NGC 6334-NGC 6357, all of the Spitzer/GLIMPSE point sources associated with the dense massive cores have [4.5] between 6.21 and 11.44 and are thus most probably young stellar objects.

4.1.2 Association with radio sources and masers

We used the SIMBAD database[*] to look for additional signposts of stellar activity provided by centimeter free-free emission and OH, H2O, and CH3OH masers. We define an association when the object is located within the FWHM size of the core.

An association with a source of free-free emission was established by using the 1.4 GHz SGPS survey (Haverkorn et al. 2006) and the 843 MHz MOST survey (Green et al. 1999), which were both correlated with the NVSS catalog given by Condon et al. (1998) and with White (2005) catalog. We cannot check the nature of the detected cm-emission with the current database, but it is probably caused by the emission of H II regions. Since a systematic search for masers in the area studied here does not exist, we checked the dense core association with known masers in the literature. We found a correlation with maser emission for 8 sources (e.g. Pestalozzi et al. 2005, Caswell et al. 2008, Moran et al. 1980; Caswell & Phillips 2008; Val'tts et al. 1999), 3 of which are also radio sources. The systemic velocity of the associated maser(s) is in good agreement with the general kinematics of NGC 6334-6357, except for core 163 (methanol and H2O masers, $V_{\rm LSR} = -50$ km s-1). Either these masers are caused by an outflow or core 163 is not associated with NGC 6334 - 6357.

4.1.3 Association with 24 ${\sf\mu }$m sources

Correlation with mid-IR sources was determined by using MSX-21 $\mu $m, and Spitzer/MIPSGAL-24 $\mu $m (Carey et al. 2009) data. For MSX we used the point-source catalogue (Egan et al. 1999), while for Spitzer/MIPSGAL we used aperture photometry from the post-basic calibrated data available at IPAC server[*]. As high-mass protostars are expected to form at the center of the dense core and because little Spitzer/MIPSGAL 24 $\mu $m or MSX-21 $\mu $m emission can be extended, we define a physical association when the source falls within the delimitation of the core but not farther than 23'' from the dense core center. This value is adopted on the basis of the association of core 51 with MSX 21 $\mu $m extended emission for which association is morphologicaly clear, while the MSX point-source catalogue position is at 23'' from the SIMBA peak.

The 24 $\mu $m flux was measured using SAOImage DS9 with ``funtools'', through both a 6.5'' and 13'' circular aperture (depending on the source size). The background correction is estimated from a 7''-13'', 20-30'', or 40-50'' annulus, depending on the background structure and crowding. An aperture correction is applied depending on the size of the background annulus (Engelbracht et al. 2007). The typical uncertainty in the aperture measurement deduced from the ``funtools'' errors is 2%. Since the uncertainty in the MIPS flux is, however, dominated by the background, we estimated a typical uncertainty of 20% from different local background measurements. Owing to the sensitivity of the MIPSGAL survey, the brightest areas are saturated. For cores in these areas, we adopted the MSX-21 $\mu $m flux. To scale these fluxes to the 24 $\mu $m, fluxes we measured the 24 $\mu $m flux of about 20 known MSX-21 $\mu $m sources and derived the best-fit linear regression (slope = 1.13 constant term=-0.64). Assuming that the relation is applicable to both high fluxes and the spectral distribution of our objects, we then multiplied all 21 $\mu $m flux by 1.13. In this way, 51 dense cores have a 24 $\mu $m flux (see Table A.1).

For a large sample of red sources, Robitaille et al. (2008) show that young stellar objects and AGB stars can generally be separated using simple color-magnitude criteria, sources with [4.5]>7.8 and [8.0]-[24.0] $\geq$ 2.5 probably being young stellar objects. All cores associated with Spitzer/GLIMPSE and 24 $\mu $m fluxes (2 cores have Spitzer/GLIMPSE but no Spitzer/MIPSGAL detection and thus cannot be classified ) follow these criteria and are probably young stellar objects. The other cores with 24 $\mu $m flux detections should only be labeled as having a high probability of being young objects since contamination by planetary nebulae or background galaxies represents at most 3% of all red sources (Robitaille et al. 2008).

4.1.4 High-luminosity and infrared-quiet cores

Dense cores that are luminous at IR wavelengths are usually considered to be good high-mass protostars or UCH II candidates (e.g. Wood & Churchwell 1989). Following Motte et al. (2007), we qualify as ''high-luminosity IR sources'' those dense cores of bolometric luminosity higher than 10 $^3~L_{\odot}$, which corresponds to that of a B3 star on the main sequence. This luminosity converts into a MIPS $24~\mu$m flux of $\sim$15 Jy, this flux being estimated in the same way as described in Motte et al. (2007), assuming that the luminosity of high-mass protostars is dominated by their mid- to far-IR luminosity and that their average colors are as defined by Wood & Churchwell (1989, see their Table 1). Following this definition, we have 11 high-luminosity and high mass (M > 100 $M_\odot $) dense cores, 9 of which are associated with star formation activity (Fig. 5). In contrast, only 3 out of 30 high mass IR-quiet dense cores exhibit stellar activity. These 27 high-mass IR-quiet dense cores are probably high-mass pre-stellar dense cores.

4.2 Turbulence level and kinematics

From molecular lines we investigate here the turbulence level and the kinematics (virial mass, infall, outflow) of the high-mass cores.

4.2.1 Turbulence

The width of emission lines in star-forming clouds are indicative the gravitational boundness of a molecular region. According to Goldsmith (1987), sites of high-mass star formation are characterized by large line widths. In contrast, dark clouds which are isolated sites of low-mass star formation, have considerably smaller line widths. Caselli et al. (2002) derived typical line widths of 0.33 km s-1 for clumps in which no IRAS source is detected. In our sample the line widths of the N2H+ isolated line vary from 1.13 to 5.38 km s-1 (see Table 1) and are thus much larger than those in the Caselli et al. (2002) study. In addition, the N2H+ line width is much higher than the thermal width (0.62 km s-1 for T=20 K), emphasizing the importance of turbulence and other non-thermal motions such as outflow, infall, and rotation. However the relative contributions of these systematic motions to the line width appear to be small compared to the turbulent component (Mardones et al. 1997). Following Kirk et al. (2007), we calculate the non-thermal component $\sigma_{NT}$ of the velocity dispersion from the width of the isolated N2H+ line and derive the level of internal turbulence $f_{\rm turb}$ defined by Kirk et al. (2007) to be the ratio of $\sigma_{NT}$ to the mean thermal velocity dispersion of the gas (sound speed of 0.23 km s-1). We note that all cores have $f_{\rm turb} > 1$ and that on average $f_{\rm turb}$ is slightly larger for NGC 6334 ( $\langle f_{\rm turb} \rangle = 4.15 \pm 1.72$) than for NGC 6357 ( $\langle f _{\rm turb}\rangle = 3.15 \pm 0.56$).

\begin{figure} \par\includegraphics[width=6.5cm,angle=-90]{13632fg10.ps} \end{figure} Figure 5:

Separating the high-luminosity sources from IR-quiet dense cores on the basis of their 24 $\mu $m flux (limit set to 15 Jy). The dense cores (open triangles) can be associated with masers (open squares), class I/II sources (filled squares), or radio sources (black diamonds)

Open with DEXTER

4.2.2 The virial mass

For the 42 high-mass dense cores, the virial mass was calculated to be (e.g. Walsh et al. 2007)

\begin{displaymath}M_{\rm vir} (M_{\odot}) = 210~R \langle \delta v \rangle^{2} \end{displaymath} (3)

where R is the core radius (in pc) and $\delta~v$ is the linewidth of the isolated line of the N2H+hyperfine structure component in km s-1. We assume that the velocity dispersion measured in the pointed observations represents the value that would be present across the entire core, as demonstrated to be the case by e.g. Kirk et al. (2007). This expression is valid for homogeneous, spherically symmetric objects with no external pressure and no magnetic field. If the density of the cores were to decrease outwards, the virial mass should be multiplied by $\frac{3}{5} \frac{(5 -2p)}{(3-p)}$, where p is the power law index of the density radial profile. For $p \sim 2$ (Motte & André 2001; Shirley et al. 2002), this factor is $\sim$0.6 implying that the calculated virial masses are upper limits. The virial mass is the minimum mass required for a cloud to be gravitationally bound, i.e. $\frac{M}{M_{\rm vir}} > 0.5$. (e.g. Pound & Blitz 1993). However, since the submillimeter masses are uncertain by a factor of 2 and the virial masses are upper limits, Motte et al. (2003) estimated that the $\frac{M}{M_{\rm vir}}$ ratio must be larger than 0.2 for gravitational boundness. In this case, only core 51 is not gravitationally bounded.

\begin{figure} \par\includegraphics[width=6.5cm,angle=-90,clip]{13632fg11.ps} \end{figure} Figure 6:

Integrated intensity of SiO (2-1) detected toward the dense cores with $M_{1.2~{\rm mm}} > 100$ $M_\odot $ as a function of their mass. The symbols are: IR-quiet souces (open triangles), H II region (open circles), maser association (open squares), high-luminosity dense core (filled triangles).

Open with DEXTER

4.2.3 SiO outflow

The next step is to establish the proto-stellar status on the basis of the SiO outflow detection. From Table 3, we note that the SiO outflow intensity and velocity are similar to that detected in Cygnus X and for a large sample of high-mass objects (Harju et al. 1998). The detection rates are 67% for NGC 6334 and 25% for NGC 6357. The global detection rate is thus 49% (58% and 47% for high-luminosity and IR-quiet objects, respectively), clearly different from the Cygnus X value of 93%. This is probably not caused by different instrumental sensitivities because we designed our SiO observations to reach similar detection limits as the observation performed for Cygnus X. In addition, the faintest Cygnus X SiO peak intensity detected is above the faintest SiO detection of our sample. For a sample of H2O and OH masers sources and ultracompact H II regions, Harju et al. (1998) measured a detection rate that decreases with Galactocentric distance and increases with FIR luminosity. They found a rate of 58% for Galactocentric distances of between 6 and 8 kpc (the Galactocentric distance of NGC 6334 - NGC 6357 is $\sim$6.83 kpc) in agreement with our value.

On average, we also note that the high-luminosity and IR-quiet dense cores with SiO outflows are, respectively, 15 and 4 times denser than those without a SiO outflow. We can estimate that the typical density that a core should have to ensure an SiO outflow is $\sim$ $6.5 \times 10^{4}$ cm-3. This may explain why the detection rate is higher in Cygnus X, because the Cygnus X dense cores are on average less massive but denser ( $ \langle n_{H_{2}} \rangle = 1.1 \times 10 ^{5}$ cm-3) than the NGC 6334 - NGC 6357 dense cores ( $ \langle n_{\rm H_{2}} \rangle = 2.7 \times 10 ^{4}$ cm-3). Finally, in contrast to Motte et al. (2007) the high-mass IR-quiet dense cores are not found to clearly exhibit brighter SiO emission than IR-bright dense cores.

4.2.4 Infall motions

To complete our census of high-mass prestellar and protostellar dense cores, we now search for signposts of infall motions. Infall motions can be studied by investigating the profiles of optically thick molecular lines (here we use HCO+ and HNC) that have a blue asymmetric structure, i.e. a double peak with a brighter blue peak, or a skewed single blue peak (e.g. Myers et al. 1996; Mardones et al. 1997). To exclude the possibility that the profile is caused by two velocity components along the line of sight, an optically thin line needs to peak close to the velocity of the self-absorption dip of the optically thick line (e.g. Zhou et al. 2003; Choi et al. 1995; Wu & Evans 2003). By examining the line profiles of the optically thin H13CO+ line, we identified dense cores with these multiple emitting regions along the line of sight and find that most of them are composed of a single broad line with a self-absorption dip. The optically thick HNC and HCO+ lines (see Appendix B), however, often show non-Gaussian, broad lines that are caused by several clumps along the line-of-sight and/or bulk motions of the envelope gas (e.g. cores 9, 28, 35, 37, 52) and outflow emission (e.g. cores 35, 37, 46, 52, 54). This is expected since the beam of around 40''is rather large. To extract more distinctive infall profiles, higher angular resolution is required. However, as a first order approximation, we can still quantify the blue asymmetry of a line by using the asymmetry parameter $\delta $V defined by Mardones et al. (1997) $\delta V= \frac{V({\rm thick})- V({\rm thin})}{{\rm d}V({\rm thin})}$. This is the difference between the peak velocities of an optically thick line $V({\rm thick})$ and an optically thin line V(thin) in units of the optically thin line full width at half-maximum (FWHM) dV(thin). We adopt the criterion of Mardones et al. (1997) for blue ( $\delta V\ <$ -0.25) and red asymmetry ( $\delta V > 0.25$). We assume HNC and HCO+ to be optically thick lines and the isolated component of the N2H+ hyperfine structure to be the optically thin line used to determine $\delta $V (see Table 3).

Another approach to characterizing infall motion in double-peak spectra is to measure the ratio of the blue to red peak (Wu & Evans 2003). A ``blue profile'' fulfills the criterion $\frac{T_{B}}{T_{R}} > 1$. The results are given in Table 3 and line asymmetry measured from HCO+ and HNC are presented in Fig. 7.

We now have four criteria to quantify infall and establish that infall has been detected when two of them are fulfilled. In this way, apart from the cores already associated with outflow emission and those suspected to be due to two components along the line of sight, only cores 18 and 23 have a high probability of exclusively exhibiting infall motions. The core 18 - due to its 24 $\mu $m counterpart and its infall motion - has a high probability of being proto-stellar in nature. For core 23, on the basis of only infall motion, no clear decision can be made about its proto-stellar nature.

\begin{figure} \par\includegraphics[width=6.5cm,angle=-90,clip]{13632fg12.ps} \end{figure} Figure 7:

Comparison of the measured asymetry in HCO+ and HNC. The dashed lines mark $\mid \delta V \mid = 0.25$. $\delta $V < -0.25 indicates blue asymmetry while $\delta $V > 0.25 indicates red asymmetry.

Open with DEXTER

5 Lifetime and massive star formation in NGC 6334 - NGC 6357

Table 3 summarizes the characteristics of the most massive denses cores (M > 100 $M_\odot $). The cores 62 and 63 were classified as a UCH II region and IR-quiet protostar, respectively, because they are, respectively, the well studied regions NGC 6334I and NGC 6334I(N), and we can compare with the literature. This agrees with the results of Walsh et al. (2010) and Thorwirth et al. (2003), who provide evidence, on the basis of a multi-line (around 3 mm) analysis, that NGC 6334I appears more evolved than NGC 6334I(N).

Since our sample is complete for embedded high-mass young stellar objects, we can statistically estimate the relative lifetime of high-mass IR-quiet and high-luminosity protostars as well as prestellar sources. To achieve this, we assume a constant star-formation rate over the past 1-2 Myr, which is justifiable because the complex exhibits sources at all stages of evolution distributed not too far apart from each other within the cloud. The lifetime estimated here is close to that found by the study in Cygnus X for which no burst of star formation had been quantitatively established (Motte et al. 2007). We also need to estimate the content of massive stars (earlier than B3) in NGC 6334 and NGC 6357, since the statistical lifetime is measured relative to the known age of OB stars. Absolute lifetimes of the different high-mass phases may also be estimated from their free-fall dynamical timescale.

In NGC 6357, two open clusters are referenced in the literature, the well known Pismis 24 and AH03J1726-34.4 (Dias et al. 2002). For Pismis 24, Wang et al. (2008) identified from X-ray studies 34 O-B3 stars, while Damke et al. (2006) estimate a density of 40 stars per arcmin2 for AH03J1726-34.4 (size 2.6 arcmin), i.e., approximately 200 stars for this cluster leading to about 60 O-B3 stars.

For NGC 6334, Neckel (1978) found 14 O-B3 stars from optical photometry and Bica et al. (2003) listed 7 embedded clusters/groups associated with radio sources. Tapia et al. (1996) estimated that the clusters towards NGC 6334I and NGC 6334E contain 93 and 12 O-B3 stars, respectively. In parallel, Bochum 13, a cluster at the north-west border of NGC 6334 contains 5 O-B3 stars (McSwain 2005), and UBV data from a field comprising NGC 6334, NGC 6357 and the inter-region filament provide an estimate of 40 O-B3 stars (Russeil et al. in preparation). We thus estimate 300 massive stars (earlier than B3) for the entire complex.

For Cygnus X and its total of 120 O stars, the expected number of massive stars (earlier than B3) can be inferred to be 660 using the mass function slope obtained and both the spectral type and mass conversion used in Knödlseder (2000).

From Fig. 5, we note that most of the cores with radio and/or maser counterparts have M > 200 $M_\odot $. For Cygnus X, observed with a 0.09 pc resolution, the mass selection of M > 40 $M_\odot $ allowed us to select dense cores with high-mass star activity (H II region, strong infrared counterpart, masers, or strong SiO outflow). This difference can be explained by our lower spatial resolution than the observations for Cygnus X, which implies we select less dense cores than in Cygnus X. In the case of NGC 6334 - NGC 6357, the average density ( $2.4 \times 10^{4}$ cm-3, while the average density is $1.4 \times 10 ^{5}$ cm-3 for M > 200 $M_\odot $ cores) and the weak stellar activity found towards starless 100 $M_\odot $  < M < 200 $M_\odot $  cores suggest they are probably forming intermediate- to low-mass stars. A large part of the IR-quiet massive cores with mass between 100 and 200 $M_\odot $ may therefore harbor low-mass to intermediate-mass protostars that are not detected here, while others could be low-mass to intermediate-mass pre-stellar cores. This result indirectly shows that starless clumps probably have density profiles flatter than r-2 and confirms that resolutions of 0.1 pc are necessary to focus on sites of high-mass star formation. After selecting cores with M > 200 $M_\odot $ , the number of high-mass progenitors is 16 (6 IR-quiet massive cores, 9 high-luminosity protostars, 1 starless clump). In addition, this new mass selection brings the SiO detection rate of the NGC 6334-NGC 6357 massive dense cores yet closer to that of Cygnus X (80% versus 93%, see Table 4).

Table 5:   High-mass young objects in NGC 6334 - NGC 6357 (M > 200 $M_\odot $) at various stages of the high-mass star formation process.

Table 5 summarizes the number, characteristics, and lifetimes of our massive dense core sample (M > 200 $M_\odot $) at each evolutionary stage and compares them to those found in Cygnus X and nearby low-mass star-forming regions. Following the definition of Motte et al. (2007), a starless clump is defined as a high-mass core without any indication of stellar activity, an IR-quiet high-mass protostar is a high-mass core with stellar activity (H II region, masers, or SiO outflow) but $24~\mu$m flux lower than 15 Jy (see Sect. 4.1.4), and a high-luminosity IR core is a high-mass core with stellar activity and a $24~\mu$m flux higher than 15 Jy. Table 4 compares the characteristics of the high-mass protostellar dense cores (IR-quiet and high-luminosity IR protostellar cores) of NGC 6334-NGC 6357 to those in Cygnus X and to HMPOs. In both tables, values given for Cygnus X and HMPOs are taken from Motte et al. (2007).

On average, the size of the cores in NGC 6334-NGC 6357 are similar whatever the evolutionary stage of the core, while the mean mass, hence the mean density, increases from starless clumps to high-luminosity cores. This is consistent with the material of dense cores concentrating itself towards its center during the star formation process.

Evans et al. (2009) find median lifetimes of $4.6 \times 10 ^{5}$ yr, $1.6 \times 10^{5}$ yr, and $5.4 \times 10^{5}$ yr, respectively, for low-mass pre-stellar cores, class 0, and class I stars. However, substantial variation in lifetime estimates from cloud to cloud are observed. For example, Wilking et al. (1989) in Ophiuchus and Kenyon et al. (1990) in the Taurus-Auriga region measure for class I sources a lifetime of 2- $4 \times 10^{5}$ yr and $1.2 \times 10^{5}$ yr respectively. In Ophiuchus, André & Montmerle (1994) obtain $5 \times 10^{3}$- $1 \times 10^{4}$ yr for class 0 and 105 yr for class I. In this framework, the NGC 6334-NGC 6357 protostellar phase lifetime ( $1.5 \times 10 ^{5}$ yr) is similar to the typical lifetime of nearby low-mass class 0 sources, but younger than that of class I stars. This suggests high-mass stellar formation proceeds more rapidly than for low-mass stars.

A high-mass pre-stellar core can be defined as a starless clump with a size of $\sim$0.1 pc and a volume-averaged density of $\sim$105 cm-3, which is gravitationally bound (Motte et al. 2007). In Cygnus X, no such pre-stellar dense core was found. For NGC 6334-NGC 6357, the low spatial resolution of our data has prevented us from detecting these pre-stellar dense cores. We detect only one starless clump with a mean size of 0.29 pc and a mean density of $4.1 \times 10^{4}$ cm-3, which is smaller and denser than those found in Cygnus X ($\sim$0.8 pc size and $\sim$ $7 \times 10^{3}$ cm-3 density). That we have detected only one starless core agrees with Hatchell & Fuller (2008). They show that the ratio of protostellar cores to starless cores increases with mass, there being ultimately no starless cores at all at the highest masses (above 12 $M_\odot $). They suggest that this implies that either more massive cores have relatively short pre-stellar lifetimes or the masses may continue to increase well into the protostellar phase. The absence of high-mass pre-stellar cores and their short lifetimes ($\le$103 yr) are discussed by Motte et al. (2007), who suggest that high-mass pre-stellar cores are dynamically evolving into protostars.

In addition, the statistical lifetime of the starless clump is shorter than the estimated free-fall time. It is also shorter than the lifetimes observed in nearby regions that form mostly low-mass stars, since starless cloud structures with volume-averaged densities of $\sim$103 cm-3 and $\sim$104 cm-3 have pre-stellar lifetimes of $\sim$106 yr and $\sim$105 yr (see Fig. 11 of Kirk et al. 2007, and references therein). Therefore, high-mass pre-stellar cores seem to be short-lived or even transient features compared to both nearby low-mass pre-stellar cores and high-mass protostars. This suggests that a supersonic dynamical process should be acting to create pre-stellar condensations from starless clumps. Short lifetimes are theoretically expected in molecular clouds, where high levels of turbulence dominate and pre-stellar cores are magnetically supercritical (e.g. Vásquez-Semadeni et al. 2005), in agreement with the turbulence levels observed within these cores. These dynamical processes are also necessary for the protostellar lifetime to last for only one free-fall time and, consequently, for the accretion to be strong enough ($\sim$ $10^{-3}~M_\odot$ yr-1) to overcome the radiation pressure and form a high-mass star.

When comparing protostellar cores in NGC 6334 and NGC 6357, we note that they have different mean densities ($\sim$ $1.7 \times 10^{4}$ cm-3 and $\sim$ $1.9 \times 10 ^{5}$ cm-3 for NGC 6357 and NGC 6334, respectively), which suggests that an environmental and/or external process is needed to explain this difference.

The comparison of the NGC 6334 - NGC 6357 and Cygnus X proto-stellar cores (Tables 4 and 5) show that both regions have similar cores densities and similar protostellar phase timescales, while the mass and size of NGC 6334 - NGC 6357 cores are higher and larger than those of Cygnus X. In addition to the lower SiO detection rate,this implies that the cores in NGC 6334 - NGC 6357 are statistically more evolved than those in Cygnus X.



6 Conclusions

To improve our knowledge of high-mass star formation, Motte et al. (2007), by studying Cygnus X, started an unbiased study of its earliest phases, i.e. the high-mass analog of low-mass pre-stellar cores and class 0 protostars. We have applied the same strategy to the star-forming complex NGC 6334 - NGC 6357. We have performed SiO(2-1) follow-up observations of 42 high-mass dense cores ($\ge$100 $M_\odot $), the most likely progenitors of intermediate-mass to high-mass stars, detected in a 1.2 mm continuum map.

Our results can be summarized as follows:

1.
We have used the MSX-21 $\mu $m and Spitzer/MIPSGAL-24 $\mu $m fluxes of our 1.2 mm high-mass dense cores to identify high-luminosity (> $10^3~L_\odot$) high-mass young stellar objects. More than half of the dense cores are considered to be good candidate precursors of high-mass stars and are found to be IR-quiet (i.e. 33 dense cores more massive than $100~M_\odot$ have $F_{\rm 24~\mu m} < 15$ Jy).

2.
We have surveyed our sample of high-mass ($\ge$100 $M_\odot $) dense cores in SiO(2-1) to search for shocked gas in molecular outflows and/or hot cores. In contrast to Cygnus X, we do not find any association of high-velocity SiO emission with all high-mass IR-quiet cores. In this way, 15 cores have been classified as starless clumps.

3.
A $M \ge 200$ $M_\odot $ selection appears more appropriate for selecting cores with high-mass star activity in NGC 6334 - NGC 6357, while for Cygnus X the selection for smaller dense cores was $M \ge 40$ $M_\odot $.

4.
Our unbiased survey of the high-mass young stellar objects ($M \ge 200$ $M_\odot $) confirms the results of the previous study of Cygnus X: high-mass IR-quiet protostars do exist and their lifetime is comparable to that of more evolved high-luminosity IR protostars. We have estimated the statistical lifetime of high-mass protostars to be $1.8\times10^5$ yr, similar to the statistical lifetime ( $1.3 \times 10^5$ yr) of high-mass protostars in Cygnus X.

5.
Only one starless clump is observed, suggesting that starless clumps rapidly concentrate and collapse to form high-mass protostars. The starless clump that we observe is smaller, denser and less massive ($\sim$0.3 pc, $\sim$ $4 \times 10^{4}$ cm-3, 242 $M_\odot $) than those of Cygnus X ($\sim$0.8 pc, $\sim$ $7 \times 10^{3}$ cm-3, 800 $M_\odot $).

6.
Lifetime measurements of the pre-stellar and protostellar IR-quiet phases of high-mass stars in NGC 6334 - NGC 6357 infer that a dynamical process is regulating their evolution. Highly turbulent processes throughout the molecular cloud complex would be necessary in such a dynamical picture of the high-mass star formation process.
To observe the high-mass star formation process during its earliest phases, far-IR to sub-millimeter continuum imaging of the entire NGC 6334 - NGC 6357 star-forming complex (such as that proposed with Herschel by Motte, Zavagno, Bontemps et al.: the HOBYS Key Programme) will be very useful. In particular, the 75/110/170 $\mu $m PACS and 250/350/500 $\mu $m SPIRE images of the HOBYS project will provide an unbiased census of both massive pre-stellar cores and massive Class 0-like protostars, and will trace the large-scale emission surrounding the cores. For the first time, this will provide accurate far-infrared photometry, which is essential for deriving good luminosity and mass estimates (from spectral energy distributions) of our high-mass dense cores sample. In addition, high angular resolution molecular line data will be required to trace in more detail the infall and outflow signatures of the protostellar sources.

Acknowledgements
The authors thanks the PNPS for financial support for the Mopra Observations. This paper is part of the ANR PROBES scientific framework. 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. The University of New South Wales Digital Filter Bank used for the observations with the Mopra Telescope was provided with support from the Australian Research Council. This research has made use of the SIMBAD database, operated at CDS, Strasbourg, France. A large Part of this work was done thanks to Spitzer (GLIMPSE and MIPSGAL) data.

References

Online Material

Table 1:   Properties of dense cores detected in NGC 6334-NGC 6357.

Appendix A: Table

Table A.1:   24 $\mu $m flux and signpost of stellar activity.

Footnotes

... complex[*]
Based on observations made with Mopra telescope. 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.
...[*]
Table 1 and Appendix are only available in electronic form at http://www.aanda.org
...[*]
Profiles as FITS files are only available in electronic form at the CDS via anonymous ftp to cdsarc.u-strasbg.fr (130.79.128.5) or via http://cdsweb.u-strasbg.fr/cgi-bin/qcat?J/A+A/515/A55
... software[*]
http://www.atnf.csiro.au/computing/software/
... GILDAS[*]
Grenoble Imaging and Line Data Analysis software
... assume[*]
The mass limit will be considerably higher in the case of cores with a flatter density profile.
... database[*]
http://simbad.u-strasbg.fr/simbad/
... server[*]
http://irsa.ipac.caltech.edu/applications/Spitzer/Spitzer

All Tables

Table 2:   Fitted parameters for the isolated N2H+ 101-012 line component using the ``method HFS'' from the GILDAS software

Table 3:   Turbulence support, SiO outflow, and gravitational infall of the most massive dense cores of NGC 6334 - NGC 6357.

Table 4:   High-mass young stellar objects in NGC 6334-NGC 6357 (M > 200 $M_\odot $) compared to Cygnus X cores and HMPOs clumps.

Table 5:   High-mass young objects in NGC 6334 - NGC 6357 (M > 200 $M_\odot $) at various stages of the high-mass star formation process.

Table 1:   Properties of dense cores detected in NGC 6334-NGC 6357.

Table A.1:   24 $\mu $m flux and signpost of stellar activity.

All Figures

  \begin{figure} \par\includegraphics[width=8.5cm,clip]{13632fg1.ps} \includegraphics[width=8.5cm,clip]{13632fg2.ps} \end{figure} Figure 1:

Top: H$\alpha $ image (UKST H$\alpha $ survey, Parker et al. 2005) of NGC 6334 and NGC 6357 with extinction isocontours overlaid on it. Red contours correspond to Av between 18 mag and 30 mag (by step of 2 mag) the blue contour corresponds to Av = 16.5 mag. Bottom: the 1.2 mm continuum emission from Muñoz et al. (2007) with extinction isocontours overlaid.

Open with DEXTER
In the text

  \begin{figure} \par\includegraphics[width=8.5cm,clip]{13632fg3.ps} \includegraphics[width=8.5cm,clip]{13632fg4.ps} \end{figure} Figure 2:

GLIMPSE 8 $\mu $m residual (corrected for point sources) mosaics (produced and delivered by the GLIMPSE team) of NGC 6334 (top) and NGC 6357 ( bottom). The 1.2 mm continuum emission isocontours are overlaid.

Open with DEXTER
In the text

  \begin{figure} \par\includegraphics[angle=-90]{13632fg5.eps} \end{figure} Figure 3:

Greyscale image of the 1.2 mm continuum emission toward NGC 6334 and NGC 6357 observed by Muñoz et al. (2007) with SIMBA (SEST). The white crosses indicate the 42 dense cores we have extracted that are more massive than 100 $M_\odot $. The core number is indicated as well, referring to the fragment number given Table 6.

Open with DEXTER
In the text

  \begin{figure} \par\mbox{\includegraphics[width=6cm,clip,angle=-90]{13632fg6.ps}... ...fg8.ps}\includegraphics[width=6cm,clip,angle=-90]{13632fg9.ps} } \end{figure} Figure 4:

Example of spectra obtained for the dense core 61 (the spectra for the other dense cores are given in the Appendix). All profiles (except for N2H+) have been smoothed to a velocity resolution of 0.3 km s-1. Upper panels: SiO ( left) and N2H+ ( right). Lower panels: on the left HCO+ superimposed on H13CO+ (thick line) and on the right HNC superimposed on 13CS (thick line).

Open with DEXTER
In the text

  \begin{figure} \par\includegraphics[width=6.5cm,angle=-90]{13632fg10.ps} \end{figure} Figure 5:

Separating the high-luminosity sources from IR-quiet dense cores on the basis of their 24 $\mu $m flux (limit set to 15 Jy). The dense cores (open triangles) can be associated with masers (open squares), class I/II sources (filled squares), or radio sources (black diamonds)

Open with DEXTER
In the text

  \begin{figure} \par\includegraphics[width=6.5cm,angle=-90,clip]{13632fg11.ps} \end{figure} Figure 6:

Integrated intensity of SiO (2-1) detected toward the dense cores with $M_{1.2~{\rm mm}} > 100$ $M_\odot $ as a function of their mass. The symbols are: IR-quiet souces (open triangles), H II region (open circles), maser association (open squares), high-luminosity dense core (filled triangles).

Open with DEXTER
In the text

  \begin{figure} \par\includegraphics[width=6.5cm,angle=-90,clip]{13632fg12.ps} \end{figure} Figure 7:

Comparison of the measured asymetry in HCO+ and HNC. The dashed lines mark $\mid \delta V \mid = 0.25$. $\delta $V < -0.25 indicates blue asymmetry while $\delta $V > 0.25 indicates red asymmetry.

Open with DEXTER
In the text

Copyright ESO 2010

Current usage metrics show cumulative count of Article Views (full-text article views including HTML views, PDF and ePub downloads, according to the available data) and Abstracts Views on Vision4Press platform.

Data correspond to usage on the plateform after 2015. The current usage metrics is available 48-96 hours after online publication and is updated daily on week days.

Initial download of the metrics may take a while.