Star-forming cores embedded in a massive cold clump:
Fragmentation, collapse and energetic outflows
The fate of massive cold clumps, their internal structure and collapse need to be characterised to understand the initial conditions for the formation of high-mass stars, stellar systems, and the origin of associations and clusters. We explore the onset of star formation in the 75 M SMM1 clump in the region ISOSS J18364-0221 using infrared and (sub-)millimetre observations including interferometry. This contracting clump has fragmented into two compact cores SMM1 North and South of 0.05 pc radius, having masses of 15 and 10 M, and luminosities of 20 L and 180 L. SMM1 South harbours a source traced at 24 and 70 , drives an energetic molecular outflow, and appears supersonically turbulent at the core centre. SMM1 North has no infrared counterparts and shows lower levels of turbulence, but also drives an outflow. Both outflows appear collimated and parsec-scale near-infrared features probably trace the outflow-powering jets. We derived mass outflow rates of at least M yr and outflow timescales of less than 10 yr. Our HCN(1-0) modelling for SMM1 South yielded an infall velocity of 0.14 km s and an estimated mass infall rate of M yr. Both cores may harbour seeds of intermediate- or high-mass stars. We compare the derived core properties with recent simulations of massive core collapse. They are consistent with the very early stages dominated by accretion luminosity.
Subject headings:dust, extinction — ISM: clouds — ISM: individual (ISOSS J18364-0221) — ISM: jets and outflows — ISM: kinematics and dynamics — stars: formation
1.1. Star formation in massive cold clumps
The early phases of star formation occur in the densest and coldest regions within molecular clouds, i.e. dense cold cores. Detailed studies of nearby star-forming molecular clouds have led to a conception of the evolution of individual low-mass stars from prestellar cores to pre-main-sequence stars (e.g. André & Montmerle, 1994; di Francesco et al., 2007; Ward-Thompson et al., 2007). The formation of high-mass stars (M M) is less understood, in particular its earliest stages and the link to the origin of associations and clusters (Zinnecker & Yorke, 2007; Beuther et al., 2008). Previous studies, mostly aimed towards luminous infrared sources, have provided evidence for collimated outflows (recently e.g. Zhang et al., 2007; Beuther & Shepherd, 2005, and ref. therein) and large-scale rotating structures around forming high-mass objects (Zhang, 2005, and ref. therein). This indicates that they also grow via accretion through a disk and that the high radiation pressure could be released in outflow cavities (Krumholz et al., 2005). In the low-mass regime, accretion rates are found to be of the order 10 M yr (Whitney et al., 1997). Much higher values of 10 M yr are expected during the build-up of high-mass stars (McKee & Tan, 2003; Banerjee & Pudritz, 2007) and corresponding increased outflow energetics have been observed (e.g. see compilation by Wu et al., 2004, and ref. therein). Recently, simulations of protostars accreting at such high rates predicted a bloating by several tens of solar radii or more and the accompanying high luminosity at low surface temperature before they reach the main sequence (Yorke & Bodenheimer, 2008; Hosokawa & Omukai, 2008). This gives further indication that the formation process differs from a scaled-up version of the low-mass case.
It is presumed that the onset of high-mass star formation occurs in cores (diameters of 0.1 pc and less) within massive cold clumps (diameters of around 0.5 pc). Surveys in the far-infrared and at millimetre wavelengths are well suited to search for such objects, but so far the earliest, i.e. prestellar, stages have eluded discovery. Motte et al. (2007) derive statistical lifetimes of less than 10 yr for the former, and of 10 yr for massive dense cores that are forming stars but exhibit low infrared emission. Interferometric observations provide the essential spatial resolution to study the substructure of clumps, and to infer the properties of embedded cores further requires a good coverage of their spectral energy distribution. In this work we present such a detailed study of one clump. In particular, we investigate the amount of fragments in the clump and their properties, the arising outflow activity, and the indications for collapse. The clump is located in the region (catalog ISOSS J18364-0221) that has been identified using the ISOPHOT Serendipity Survey (ISOSS, Lemke et al. (1996); Stickel et al. (2007)) at 170 to establish far-infrared colour temperatures. Krause (2003) and Krause et al. (2004) selected sources containing a large fraction of cold dust, and subsequent studies revealed a population of cold clumps (Krause et al., 2003; Hennemann et al., 2008) and a collapsing massive core (Birkmann et al., 2007).
1.2. The ISOSS J18364-0221 star-forming region and the SMM1 clump
The star-forming region (catalog ISOSS J18364-0221) (R.A. , Decl. -02°21′49″[J2000]) was initially studied by Birkmann et al. (2006) and is located at a distance of about 2.2 kpc. Derived from the near-infrared extinction over a field ( 100 pc), the cloud complex associated with the star-forming region comprises a mass M M. In the same manner, a lower mass limit M M was found for the central region ( 1 pc), while the far-infrared and submillimetre measurements give a luminosity L L, an average dust temperature of about 15 K and a mass of M. This region contains two clumps detected in the submillimetre continuum named SMM1 and SMM2 that are separated by about along the east-west direction. The SMM1 clump is subject of this publication. Its effective radius is about 0.2 pc, and from the thermal emission a characteristic dust temperature of K and a clump mass of M M were found. The molecular line observations reported in Birkmann et al. (2006) show red-shifted self-absorption that is interpreted as signature of collapse motions. Furthermore, at least one outflow is present. The outflow properties resulting from the single-dish observations, i.e. the outflow mass of about 18 M and the mass outflow rate of 10 M yr, are comparable to values derived for outflows from presumed high-mass star precursors. These results render the SMM1 clump a promising object to study the early phases of collapse and fragmentation occurring in massive cold cores.
The detailed study of this source we present here made use of recently collected infrared and (sub-)millimetre data described in the next section. They reveal two star-forming cores and molecular outflows (Section 3). In Section 4, we discuss the core properties in the context of early-stage star formation and present a comparison of the observed HCN spectra with simple radiative transfer models. Finally, the results are summarised in Section 5.
2. Observations and data reduction
2.1. Near-infrared observations
Imaging observations in the J, H, and Ks band towards (catalog ISOSS J18364-0221) have been obtained and are described in Birkmann et al. (2006). Additional near-infrared images in the H = 1-0 S(1) line ( 2.122 ) were taken with the Calar Alto 3.5 m telescope in October 2005 using the prime-focus wide-field camera Omega-2000 (Baumeister et al., 2003). Omega-2000 features a field of view (FOV) of with a pixel scale of . The exposures were dithered on source to allow for sky subtraction. The data reduction was done using IRAF.
2.2. Mid- and far-infrared observations
IRAC (Fazio et al., 2004) imaging in all four photometric bands, MIPS (Rieke et al., 2004) imaging at 24 and 70 and MIPS spectral energy distribution (SED) mode observations were undertaken with Spitzer (Werner et al., 2004). The basic flux calibrated imaging data of the Spitzer Science Center (SSC) pipeline were used for further data reduction and analysis. Cosmetic corrections and astrometric refinement were performed with the MOPEX software (Makovoz & Marleau, 2005), and the final images were combined using scripts in IRAF. Aperture photometry and PSF fitting was done with the aperture corrections given in the IRAC data handbook and on the SSC website111http://ssc.spitzer.caltech.edu/mips/apercorr/. The MIPS SED mode calibration is based on a spectrum of Boo (Low et al., 2005) and the measured MIPS 70 fluxes. The calibration uncertainties are about 2% (IRAC, Reach et al. (2005)), 4% (MIPS 24, Engelbracht et al. (2007)), and 10% (MIPS 70, Gordon et al. (2007)). The resulting photometric accuracy is estimated to 5% (IRAC), 10% (MIPS 24), and 20% (MIPS 70 and SED).
2.3. Submillimetre observations
The submillimetre continuum observations with SCUBA at the JCMT are outlined in Birkmann et al. (2006). In light of the results from the interferometric observations that are described in the next section, we reanalysed the jiggle maps. The ORAC-DR (Jenness & Economou, 1999) and SURF (Jenness & Lightfoot, 1998) software were used for data reduction and the photometric calibration based on maps of Uranus acquired shortly before and after the observations. Further analysis as described in Sandell & Weintraub (2001) used the MIRIAD software (Sault et al., 1995). The deviations of the JCMT beam from a single Gaussian have been considered by using the Uranus maps to construct symmetric beam models and deconvolve the maps of the target regions. The derived beam sizes are 8.2 at 450 and 14.8 at 850 . The maps were restored with Gaussian beams of 8 and 14, respectively, and fluxes as well as deconvolved source sizes have been derived by fitting Gaussian components. The noise levels (1) in the restored maps are 100 mJy beam at 450 and 23 mJy beam at 850 . The photometric accuracy obtained is estimated to 30% at 450 and 20% at 850 . For a large aperture covering the SMM1 clump, the photometric results reproduce those of Birkmann et al. (2006) within the uncertainty ranges.
2.4. Millimetre observations
We have carried out millimetre observations using the IRAM 30m and Plateau de Bure Interferometer (PdBI). The molecular lines CO(2-1), HCO(1-0) and HCN(1-0) have been observed together with the continuum at 1.3 mm and 3.4 mm. The line frequencies are 230.538 GHz for CO(2-1) and 89.188526 GHz for HCO(1-0). The HCN(1-0) transition includes three hyperfine components within 4 MHz at 88.630416 GHz (F), 88.631847 GHz (F), and 88.633936 GHz (F)222NIST Recommended Rest Frequencies for Observed Interstellar Molecular Microwave Transitions: http://physics.nist.gov/restfreq. The PdBI configurations C and D were utilised; D was observed in September 2006 and C in April 2007 with the new generation facility receivers. Spectral resolutions of 40 kHz (3 mm lines), 160 kHz (CO(2-1)), 1.25 MHz (3.4 mm continuum, MHz bandwidth), and 2.5 MHz (1.3 mm continuum, MHz bandwidth) were used. Phase calibrators were 1749-096 and 1741-038, additional amplitude calibrators were MWC349 and 3C273. Corresponding short-spacing observations were accomplished in February 2007 at the IRAM 30m as On-the-fly maps using the single pixel heterodyne receivers with the VESPA correlator and spectral resolutions of 80 kHz (1 mm lines, 80 MHz bandwidth) and 20 kHz (3 mm lines, 40 MHz bandwidth). The data were reduced and calibrated with the GILDAS333http://www.iram.fr/IRAMFR/GILDAS software. GILDAS was also used to combine the short-spacing and interferometric data for the lines. In the case of CO, we chose to convert the measured visibilities to maps using a weighting scheme that achieves the highest spatial resolution (synthesised beam size of ) at the expense of sensitivity for extended emission. Therefore the combined CO map does not recover the complete flux. Compared to the single-dish spectra, the line wing fluxes in the combined map are lower by a factor of . For HCN, the comparison of the extracted spectra shows that the flux measured in the single-dish data is reproduced in the combined map. The synthesised continuum beam sizes are (PA 11.6°) at 1.3 mm and (PA 6.1°) at 3.4 mm. We reach noise levels (1) in the cleaned maps of 0.6 mJy beam (1.3 mm continuum), 0.13 mJy beam (3.4 mm continuum), 0.5 Jy beam km s (CO(2-1)), 0.07 Jy beam km s (HCO(1-0)), and 0.1 Jy beam km s (HCN(1-0)). Assuming a dust temperature of 20 K, we are sensitive to total masses of about 0.2 M (3 in the 1.3 mm continuum). The noise level (1) in the individual HCN(1-0) spectra used for the modelling is 0.1 K. The 1.2 mm continuum was observed with the MAMBO-2 bolometer array in March 2007. The MOPSIC software was used for data reduction and the noise level (1) in the resulting map is 10 mJy beam. The photometric accuracy is estimated to 20%.
3.1. Multiwavelength maps of SMM1
In Figs. 1 and 2 we show multiwavelength maps observed towards (catalog ISOSS J18364-0221) SMM1. In the interferometric millimetre continuum observations SMM1 is resolved into two cores SMM1 North and SMM1 South (Fig. 1 lower panels), separated by (ca. 21000 AU). In the near-infrared (Fig. 1 upper left), noticeable extinction and reddening is present in the surrounding of SMM1. There are no near-infrared sources detected towards the (sub-)millimetre peaks. The mid-infrared measured with IRAC (Fig. 1 upper right) is dominated by extended emission in the 8 band along the outer rim of SMM1. Towards the centre of SMM1, a filamentary structure is observed in absorption, and no stellar objects are associated. At 24 (Fig. 1 lower left) a point source at SMM1 South shows up. At this wavelength, the extended emission around SMM1 is also visible, and in the north-west a dip remains, but the absorption feature is superposed by the PSF of the southern source. No obvious 24 source towards SMM1 North is detected. Similarly, SMM1 South emits at 70 (Fig. 1 lower right). The wide PSF overlays possible extended emission at the rim of SMM1, and no obvious emission at SMM1 North is detected.
In the interferometric 1.3 mm continuum map (Fig. 2 top row), both SMM1 North and South appear slightly extended to the north, but the beam sidelobes may affect the morphology. At 3.4 mm they appear unresolved (Fig. 2 bottom row), and none of the extensions are traced. The contours in Fig. 2 derived from the emission in the line wings of the CO(2-1) and HCO(1-0) transitions reveal two molecular outflows. SMM1 South constitutes the origin of a north-east-to-south-west (PA ca. 50°) outflow that also gives rise to the mid-infrared features. Towards SMM1 North an outflow in east-to-west direction (PA ca. -80°) is found. Its red lobe blends into the red lobe of the SMM1 South outflow. Most prominent in CO(2-1) is the SMM1 South blue outflow lobe that appears collimated. In HCO(1-0), the blue lobe of the SMM1 North outflow is strongest and it also shows a rather low outflow opening angle, but in both cases the outflows are traced only close to the cores and we could not derive an accurate quantitative measure.
In Fig. 3 we show the near-infrared H = 1-0 S(1) ( 2.122 ) line emission map. This map has been derived by subtracting the scaled K band image from the narrow-band image so that stars cancelled out. Stars that were not properly subtracted have been masked afterwards for clarity. As shown in the inlay of Fig. 3, towards the blue and red lobes of the molecular outflow driven by SMM1 South there is patchy near-infrared H emission. Such features trace protostellar jets, and the emission probably arises from collisionally excited H in shocked gas (Elias, 1980; Reipurth & Bally, 2001, and ref. therein). This is supported by faint filamentary emission in the 4.5 and 5.8 bands we detected lateral to SMM1 South along north-east-to-south-west and in the vicinity of SMM1 North (yellowish features in Fig. 1 top right). These bands contain several H lines (Noriega-Crespo et al., 2004; Smith & Rosen, 2005). Only a weak near-infrared feature is detected close to the blue lobe of the SMM1 North outflow. On the large-scale map we found additional filaments of H emission. Though they are not exactly aligned with the molecular outflows, they can be assigned to individual lobes which would make it necessary that the proposed jets show bending resulting in S-shapes. The outermost features in the North-East and South-West are located at a projected distance of about 6.7 pc from each other, and the South-West feature is at a projected distance of about 4.6 pc from SMM1 South. Among the sources detected in the vicinity of the features, we do not find other candidates for jet-driving young stellar objects.
3.2. SEDs and core sizes
In the near-infrared and at 3.6 no emission is coinciding with SMM1 North or South and we derived upper flux limits from our maps. Close to SMM1 North, the emission in the 4.5 and 5.8 IRAC bands were measured on levels of 71 and 155 Jy. In the 4.5 band there is extended emission overlapping with the SMM1 South position, where the flux is about 1.2 mJy. At 8 no counterparts corresponding to SMM1 North or South are detected.
The source towards SMM1 South clearly dominates the emission at 24 and 70 . Due to the separation of about 9 between SMM1 North and South, the derivation of fluxes at these wavelengths is impaired towards SMM1 North (PSF FWHM of about 6 and 17). We accomplished PSF subtraction to remove the emission of SMM1 South. At both 24 and 70 the residual maps do not reveal a second compact source at SMM1 North. The 3 upper limits for SMM1 North were derived from the residual scatter at its position. We ended up with 24 fluxes of 199 mJy and 15.7 Jy for SMM1 South and upper flux limits of 5.3 mJy and 177 mJy for SMM1 North at 24 and 70 respectively.
To derive submillimetre fluxes for the two internal SMM1 components we fitted a Gaussian to each peak on the 450 map and an extended Gaussian to account for the surrounding clump emission. The same sizes convolved with the 14 Gaussian were then used as fixed parameters to extract 850 fluxes. The clump background is well fit with a deconvolved FWHM size of 68 and total fluxes of 12.3 Jy (450 ) and 2.02 Jy (850 ). For SMM1 South we got a deconvolved FWHM size of 5.9 and fluxes of 4.42 Jy (450 ) and 622 mJy (850 ), and for SMM1 North a size of 7.0 and fluxes of 2.66 Jy (450 ) and 476 mJy (850 ). The total fluxes we derived are higher than those given in Birkmann et al. (2006) because we allowed for a larger extent of the background component. As in the 850 map (Fig. 1), SMM1 North and South blend into a single elongated maximum in the single-dish 1.2 mm map. Therefore we extracted fluxes in the same way for both sources, using the positions from the submillimetre and allowing for a pointing offset. The derived fluxes are 151 mJy for SMM1 South and 156 mJy for SMM1 North.
The continuum fluxes for both cores are listed in Table 1. Fig. 4 shows the spectral energy distributions of the continuum emission towards the two sources including the jet features in the IRAC bands. The curves fitted to the data are described in the next subsection.
From the interferometric continuum maps we also derived fluxes for both sources by fitting Gaussian components. In the case of SMM1 South, 80 mJy (1.3 mm) and 4.1 mJy (3.4 mm) and FWHM dimensions of (PA -50°) were measured, corresponding to a core size of 9000 AU. For SMM1 North the fluxes are 49 mJy (1.3 mm) and 3.9 mJy (3.4 mm) and FWHM sizes of (PA 34°) were measured, corresponding to a more elliptical core of AU.
We also estimated the bolometric luminosities from the two SEDs. SMM1 South has a luminosity L L, and for SMM1 North we got L L using the upper limit fluxes in the infrared.
3.3. Dust temperatures and masses
Ossenkopf & Henning (1994) have derived dust opacities for dense protostellar cores and for our analysis we used the model with thin ice mantles and coagulation at a gas density of n cm (OH5, = 0.9 cm g). To estimate the temperature uncertainties we also applied the model without ice mantles (OH2) and the initial, non-coagulated opacities (OH1). In Fig. 4 we show the OH5 fit to the emission of SMM1 South using two modified Planck components assuming optically thin emission. The long-wavelength SED is matched by thermal dust emission at K (solid line). From the fit we derived a dust mass of M M. The fluxes at shorter wavelengths are reproduced by a dust component at 48 K with a dust mass of M M (dashed line). For the SMM1 North source we used the upper flux limit at 70 to derive an upper limit dust temperature of K from a single component fit also shown in Fig. 4. The corresponding dust mass is M M. Assuming a canonical gas-to-dust mass ratio of 100, the masses of the cores are ca. 12 M (SMM1 South) and 18 M (SMM1 North), with an uncertainty of about a factor two (see 4.2).
Preserving the dust temperatures and masses, the expected continuum fluxes at 3.4 mm are 4.9 mJy (SMM1 South) and 4.8 mJy (SMM1 North), thus the measured fluxes trace about 10 M for SMM1 South and about 15 M for SMM1 North as compact components. The differences stem from the filtering in the interferometric observations. Using these masses and the core FWHM sizes of 9000 AU and 10000 AU, we calculated volume-averaged densities n = 3M/(4 FWHM) which are cm for SMM1 South and cm for SMM1 North. The remaining mass of the SMM1 clump is located in a more extended envelope. From the dust temperatures and peak fluxes in the interferometric 1.3 mm continuum map we derived peak column densities of cm (SMM1 North) and cm (SMM1 South). Table 1 summarises the properties of the two detected cores.
|Parameter||SMM1 North||SMM1 South|
|24 flux (mJy)||5.3||199|
|70 flux (mJy)||177||15700|
|450 flux (mJy)||2660||4420|
|850 flux (mJy)||476||622|
|30m 1.2 mm flux (mJy)||156||151|
|PdBI 1.3 mm flux (mJy)||49||80|
|PdBI 3.4 mm flux (mJy)||3.9||4.1|
|FWHM size (AU)||140007000 (PA 34°)||100008000 (PA -50°)|
|Dust temperatureaaDerived from the single-dish continuum flux measurements/upper limits between 70 and 1200 . (K)||15||22|
|NbbDerived from the interferometric 1.3 mm continuum map and the given dust temperatures. (cm)|
|MassccDerived from the interferometric 3.4 mm continuum map and the given dust temperatures. (M)||15||10|
3.4. Properties of the molecular outflows
From our CO(2-1) map we inferred the properties of the outflows. For this purpose it is necessary to convert the measured CO emission into molecular hydrogen column densities N, and we used the relation N cm K km s T dv from Osterloh et al. (1997). The mass was then calculated as M md N, where is the ratio of gas mass to hydrogen mass (taken to be 1.36), m is the mass of the hydrogen molecule, and is the solid angle covered by N. The integrated emission in the CO map towards SMM1 traces 4.8 M, compared to a mass of 75 30 M derived from the dust continuum emission (Birkmann et al., 2006). This discrepancy probably stems from the uncertainty of the above relation in combination with the missing flux in the CO map (see 2.4). Apparent also in the integrated CO(2-1) spectrum in Fig. 2, the optical depth in the line is significant. Other outflow studies have found CO(2-1) optical depths on the order of 10 referring to a comparison with CO emission as described in Choi et al. (1993). In the case of SMM1 South, the fact that near-infrared H emission is observed towards both molecular outflow lobes indicates that the difference in extinction is not very large. This suggests a high inclination of the outflow axis with respect to the line-of-sight. Because we cannot further constrain the inclination, we assume for both outflows in the following, corresponding to the mean of a random distribution of outflow orientations. In Table 2 we list the parameters derived as described in Henning et al. (2000a) for the different outflow lobes, corresponding to line emission above 3. Mean outflow velocities were derived from the mechanical momenta P and the masses using vP/M, and halves of the lobe extents in outflow direction were taken as travelled distance to calculate the dynamical timescales. We interpret the given masses as lower limits, and as described in 2.4, the true masses are probably higher by roughly a factor of five. The same applies to the mechanical momenta and the kinetic energies. While the mass derivation may be precise within factors of 2 to 4 (cf. Cabrit & Bertout, 1990), the dynamical parameters are less certain and can be considered as order of magnitude estimates. Besides, the blending of the two red outflow lobes hampered our parameter derivation, so we regard the values derived for the blue lobes as more precise.
|Outflow Parameter||South Red Lobe||South Blue Lobe||North Red Lobe||North Blue Lobe|
|Proj. Velocities (km s)||40…50||20…30||40…50||20…30|
|Solid Angle (as)||86||195||27||119|
|MassaaMasses are lower limits because of missing flux. (M)||0.066||0.15||0.023||0.12|
|Mech. MomentumbbAssuming an inclination of the outflow axes. (M km s)||0.85||-2.3||0.30||-1.8|
|Kin. EnergybbAssuming an inclination of the outflow axes. (J)||1.710||2.510||1.810||2.410|
|Dyn. TimescalebbAssuming an inclination of the outflow axes. (yr)||7.710||7.010||3.810||5.810|
|Mass Outflow RatebbAssuming an inclination of the outflow axes. (M yr)||8.610||2.110||6.010||2.110|
|Mech. LuminositybbAssuming an inclination of the outflow axes. (L)||1.8||2.9||3.9||3.4|
3.5. HCN emission towards SMM1
The three HCN(1-0) hyperfine components F, F, and F lie within 4 MHz and the expected line ratios are 3:5:1 in the optically thin case for local thermal equilibrium (LTE). The HCN(1-0) map integrated over all three components (Fig. 5 left) shows a pronounced maximum towards SMM1 South, whereas only weak emission is detected at SMM1 North. It is possible that the minor peaks that are offset from SMM1 North stem from the molecular outflows, because it has been shown recently that HCN can be present there as well (Zhang et al., 2007). The HCN(1-0) spectra towards the centre positions of SMM1 South and North are shown in the middle panels of Fig. 5.
In the case of SMM1 North, the signal-to-noise ratio of HCN(1-0) is rather low ( 6) and the line shapes are uncertain. As indicated in the plot, the positions of the peaks are consistent with a systemic velocity v km s. The observed line ratios (neglecting the noise) depart from the expected optically thin LTE ratios: F/F and F/F compared to 0.2 and 0.6 in the LTE case. Such hyperfine “anomalies” have also been observed towards star-forming clouds in the past (Walmsley et al., 1982). Gonzalez-Alfonso & Cernicharo (1993) have investigated this numerically and found that the velocity structure in the emitting cloud core could strongly affect the line ratios. In particular, they found that an increased F/F ratio indicates inward motion in the core. Line ratios similar to the SMM1 North spectrum were predicted for a subset of their models.
In the case of SMM1 South the HCN(1-0) spectrum is detected with a higher signal-to-noise ratio of up to 40. By fitting the line wings with a symmetric profile, we get v km s. The hyperfine components are characterised by complex line shapes and cannot be described within the optically thin LTE approximation. The main features of the line profiles are strong dips near the line centres. In essence, such dips can be interpreted: 1) as an artifact of the data combination, i.e. the short-spacing flux is not recovered when combining the interferometric and single-dish data; 2) as a physical effect, namely, as a result of the self-absorption of the internal radiation in the envelope. To illustrate that this is not an artifact of data reduction we show the single-dish and interferometric HCN(1-0) spectra (integrated over ) before their combination in the right panels of Fig. 5. If the dips in the interferometric spectrum are a result of the missing short-spacing flux, then we should see prominent emission near the line centres in the single-dish data. However, both interferometric and single-dish spectra have strong dips that indicate that these line features are real.
4.1. The fragmentation and column density of the SMM1 clump
Our observations reveal two major fragments towards the SMM1 clump. From its mean FWHM extent of 30000 AU (Birkmann et al., 2006) we calculate a volume-averaged density of n cm for SMM1. The dust temperature of 16.5 K gives a Jeans length of about 17000 AU and a Jeans mass of 0.7 M (Stahler & Palla, 2005). This indicates that the two cores with masses more than one magnitude higher are not the direct result of thermal fragmentation, because one would then expect a number of cores with about one Jeans mass to form (e.g. Jappsen et al., 2005; Bonnell et al., 2006). The projected distance of the two cores is 21000 AU. It is rather large compared to the radius of influence of radiative feedback from protostellar objects. The simulations of Krumholz et al. (2007) show that the latter is around 1000 AU in the first 20000 yr of core collapse. This suggests that the two cores evolve individually in terms of radiative feedback, however a kinematic influence is not excluded. Krumholz & McKee (2008) proposed a lower column density threshold of 1 g cm (N cm) for the formation of high-mass stars. For the SMM1 clump, we derived a peak column density of cm from the single-dish 1.2 mm map peak flux. It does not reach the proposed limit. However, the core peak column densities lie beyond the threshold (see Table 1). This shows that observations with high spatial resolution are required to evaluate the core properties.
4.2. Properties of the two millimetre cores
The two detected cores SMM1 North and South appear quite similar at long wavelengths because they are of nearly the same size and exhibit comparable continuum emission. However, the derived core masses depend crucially on the assigned dust temperatures, and these are determined by the flux levels in the far-infrared. At wavelengths below 100 the SMM1 clump of about 75 M is expected to become optically thick for continuum emission, and our Planck component fit does not take this into account. We therefore neglect the fitted warm component for SMM1 South in the rest of the discussion. Also for the cold components, we may have underestimated the characteristic dust temperature. However, because of its mass the far-infrared continuum optical depth of SMM1 is expected to be of the order of unity only. This means that a significantly higher dust temperature, corresponding to a lower mass and in consequence to a lower optical depth, would not be consistent with the far-infrared fluxes. Thus, the resulting uncertainty of the masses are about a factor of two, but the general averaging along the line-of-sight could introduce larger errors. Furthermore, our mass derivation relies on the used dust opacities. While our choice of the OH5 model is rather conservative and the results can be compared to other studies, the true opacities may well differ. In the following, we take the derived masses at face value.
Both SMM1 North and South are compact and appear to drive individual outflows. Therefore we suspect them to form individual stars or stellar systems of a few. With about 10 and 15 M, SMM1 South and North represent cores of intermediate mass. From the comparison of core and stellar mass distributions, there are tentative indications that the star-forming efficiency on the scales of individual cores is between 25 and 50% (see e.g. Krumholz, 2008, and ref. therein). Under this assumption, one would expect both cores to form intermediate-mass protostars or protostellar systems in case none of the surrounding matter enters in the core collapse. However, the latter is probable because of the large-scale collapse motions that have been observed towards SMM1. In comparison to samples of low-mass dense cores, SMM1 North and South are denser by a factor of 10 (cf. Motte et al., 2007). We regard the peak column density values we derived from the interferometric 1.3 mm peak fluxes as lower limits because of the filtering in the interferometric map. For low-mass dense cores, peak column densities of cm or lower were reported (Ward-Thompson et al., 1999), so both cores lie above. They rather resemble the objects found in IRDCs by Rathborne et al. (2007). While their sample of four clumps (we use a different nomenclature here) have masses of 100 M and greater, the small scale cores are several thousand AU in size and their masses are 2 to 21 M with the exception of one hot molecular core candidate. As in the case of SMM1 South, 24 sources are associated with those cores. Similar findings have been reported by Beuther & Steinacker (2007) for the more massive core in IRDC 18223-3, in particular the remarkable low luminosity.
The SED of SMM1 South at wavelengths below about 80 (Fig. 4) does not resemble the single thermal emission component reproducing the submillimetre and millimetre fluxes. A similar result was also derived in the study of a sample of 12 clumps detected in five other ISOSS star-forming regions (Hennemann et al., 2008) and for one core in the ISOSS J23053+5953 region (Birkmann et al., 2007). Warm and hot dust components are required to explain the emission down to 24 , and also emission from very small grains may contribute in this wavelength regime. The fact that SMM1 South appears as point source at 24 and 70 shows that this core contains a compact region of heated dust. Assuming half of the PSF FWHM (6 at 24 ) as an upper limit, the emitting region has a size of less than 7000 AU. This is consistent with an embedded young stellar precursor that constitutes the driving source of the SMM1 South outflow.
SMM1 North lacks emission in the infrared. The only sign of star formation is the SMM1 North outflow, indicating that this core is also further evolved than a supposed prestellar stage. The characteristic dust temperature is constrained to an upper limit, and therefore in this case the derived core mass represents only a lower limit.
Regarding the motion of the two cores with respect to each other, their systemic velocities derived from the HCN spectra differ by about 1.3 km s and SMM1 North appears to be receding. The estimated uncertainty is 0.5 km s. Nevertheless, the value is relatively high compared to the velocity dispersions of up to 0.5 km s measured for the core-to-core motions e.g. in NGC 1333 and Perseus (Walsh et al., 2007; Kirk et al., 2007). It corresponds to roughly 0.3 AU yr, and on a timescale of about 10 yr the cores would cover their projected distance.
4.3. The star-forming process within the cores
The SEDs of both cores are dominated by the emission at long wavelengths arising from cold dust. This emission does not disclose much information about the internal structure of the cores. The lack of mid-infrared emission and the low spatial resolution in the far-infrared, when compared to the expected scales of emitting regions, do also prevent clarification of the properties of the embedded source in the case of SMM1 South. SMM1 North is very likely less evolved, but we cannot exclude projection effects. More observations are needed to further constrain the star formation process in the detected cores. The core emission morphology in the far-infrared will be explored with the ESA Herschel mission. In the case of SMM1 South, high spatial resolution imaging beyond 20 combined with high sensitivity, possible with the MIRI instrument on JWST, will severely constrain the properties of the embedded protostellar object. In the following, we discuss implications given by the derived outflow properties and our modelling of the HCN spectra.
4.3.1 Outflow activity
Collimated outflows have been observed for low-mass cores and, more recently, also for young high-mass objects, supporting the disk accretion scenario (see Beuther & Shepherd, 2005, and ref. therein). In the former case, a high degree of collimation is linked to early stages and a widening of the outflow opening angle follows during further evolution (Arce & Sargent, 2006). According to some observational evidence, this holds for high-mass sources as well and may provide a basis for their evolutionary classification.
The fact that we observed rather collimated outflows close to their proposed origins SMM1 North and South is consistent with the idea that the embedded driving sources have formed and launched the outflows recently. This is supported by the estimated dynamical timescales of less than yr.
The outflow energetics change by orders of magnitude from low-mass cases to luminous high-mass driving sources (e.g. Wu et al., 2004; Henning et al., 2000b), and the span of outflow kinetic energies that have been derived is J E J. For the sample of Beuther et al. (2002b) with L L, they got values on the order of E J. With E J, the outflows of the SMM1 cores reach the same order of magnitude, in particular because the adopted masses are lower limits. They exceed those of low-mass cores.
So far we have only considered the molecular outflows close to the cores. The traced molecular gas has probably been entrained by collimated jets. If we assume that the H emission features at larger distances stem from these jets, which is supported by the rough alignment with the molecular outflows, we can derive a second timescale for the outflow activity of the cores. However, we cannot constrain the jet velocities from observations. Beuther et al. (2002b) used a ratio of jet velocity to molecular outflow velocity of 20, and for the SMM1 cores the resulting jet velocities are about 300 km s. This is in the range of velocities that have been observed for Herbig-Haro flows from low-luminosity sources (Reipurth & Bally, 2001). From the measured offsets of the outermost H feature in the south-west we got a timescale of approximately yr for the SMM1 South jet. This is above the molecular outflow timescale of approximately yr. For the SMM1 North jet, the outermost feature in the west gives a timescale of about yr, which agrees with the molecular outflow timescale. Accounting for the overall uncertainties, these values are consistent with the proposed relation of jets and molecular flows, which of course does not exclude other scenarios.
Parsec-scale jets and outflows have been found towards many young low-mass stellar objects, mostly traced by optical or near-infrared line emission (Reipurth & Bally, 2001). The more energetic outflows in the high-mass regime are therefore expected to also extend to these sizes, although few observations were reported to date (e.g. Bally, 2008; Beuther et al., 2002a). The presumed jets from SMM1 support this idea. The filamentary and patchy structure observed in H can be caused by different factors. Besides the varying line-of-sight extinction, the local H abundance may play an important role in determining where the shock-excited emission arises, and the kinematics of internal shocks is presumably influenced by the penetrated medium. An interesting possibility is the connection to the mass ejection history of the driving source, which is linked to the accretion history for disk-driven outflows. The several features we detected indicate a varying mass-loss rate for the SMM1 cores. Such a burst mode of accretion may result from disk instability driven by infall (Vorobyov & Basu, 2006).
The S-shape that we observe for the presumed SMM1 South jet can be interpreted as a precession of the outflow axis. The outermost feature we found in the south-west approximately lies on the projected outflow axis with PA we derived from the molecular outflow lobes close to the core, while the features in between are offset to lower PA. This means the core rotation axis is not along the line-of-sight, but rather has a high inclination.
4.3.2 Modelling of the HCN emission
HCN appears to be more resistant to freeze-out in cold dense cores compared to CO-related molecules as HCO (Redman et al., 2008). Assuming that the HCN emission traces the dense gas in the cores, we investigate in the following if the observed HCN(1-0) spectra can be, in principle, reproduced by simple models for the core. First, we considered the “one-layer” model where the core is spherically symmetric and homogeneous. We assumed that the observed spectrum is obtained towards the centre of the core as the antenna beam size is little smaller than the spatial extent of the core. The parameters of the model are: the hydrogen density n, the temperature T, the molecular column density N, the turbulent velocity v, and the regular velocity v which characterises the radial expansion or contraction of the core. Note that we do not specify the radius of the core as well as the actual abundance of HCN since the emergent spectrum for such a core depends on their product (cf. Pavlyuchenkov et al., 2008). Therefore, in the frame of this model we cannot independently constrain the relative HCN abundance, the radius of the core, or its mass.
Having specified the above parameters, we generated the model core and performed a line radiative transfer (LRT) simulation with the non-LTE code of Pavlyuchenkov & Shustov (2004) using molecular line data from Schöier et al. (2005). A synthetic spectrum of HCN(1-0) was calculated from the LRT simulation. In principle, the systemic velocity v is an additional parameter. We have assumed the values of 36.1 km s (SMM1 North) and 34.8 km s (SMM1 South) derived from the positions of the spectral features. The consistency between calculated and observed spectra was evaluated with a -criterion and took into account all the velocity channels of the combined HCN(1-0) spectrum.
To search for the best set of the model parameters, we used Powell’s minimisation algorithm (Press et al., 1992). We had to specify parameter ranges for the search for the best-fit model. In our simulations we used the ranges given in the Table 3 to derive the best-fit parameter sets. The ranges resulted from several runs where we specified parts of the adjacent parameter space and did not find good reproductions of the observed spectra. Though unlikely, we cannot exclude that additional compatible sets of parameters exist. In order to assure that the minimisation routine was not trapped in a local minimum, we repeated the calculations starting with several sets of initial parameters. The best-fit spectra obtained for the one-layer model with the corresponding parameters are shown in Fig. 6 in the top row.
In the case of SMM1 North, the observed line ratios were reproduced fairly well. We found a degeneracy between the hydrogen density, the kinetic temperature and the HCN column density and therefore we set n cm. This resulted in T K and N cm. Furthermore, given the assumption that the observed HCN(1-0) components are indeed single-peaked and relatively narrow, the effect of the regular and turbulent velocities cannot be distinguished. Therefore we set the regular velocity to zero. The derived value of the turbulent velocity is 0.4 km s. Note that in contrast to the study of Gonzalez-Alfonso & Cernicharo (1993), we did not have to introduce any regular velocity to explain the “anomalous” line ratios. The obtained ratios naturally appeared as a result of the non-LTE excitation in the subcritical density.
In the case of SMM1 South, the one-layer model could not reproduce the maximal intensities and the strength of the self-absorption dips at the same time. We show one representative spectrum from the one-layer model in Fig. 6.
A natural way to reproduce the observed spectrum of SMM1 South is to include a low-density envelope in the model which should lead to self-absorption. We checked this by constructing a “two-layer” model with 5+5 parameters, and chose a configuration where a static core (v) is surrounded by a spherically symmetric, homogeneous envelope. The increased number of parameters makes it difficult to identify a best set through the minimisation routine. Therefore we fixed some of the parameters based on the following arguments. First, we assumed that the hydrogen density in the core is higher while in the envelope it is lower than the critical density for HCN(1-0). Given this assumption, we expect the formation of strong self-absorption dips if the HCN column density in the envelope is high enough. We set the densities to n cm and n cm. We also fixed the temperature in the envelope T K assuming that it is not heated by the inner (most probably warmer) part of the core, for which we chose T K. Changes in the core temperature do not affect the derived model spectra significantly. Thus, we varied the five parameters N, v, N, v, v, and constrained them using the minimisation routine.
The best-fit spectrum obtained from the two-layer model is presented in Fig. 6 (bottom). The two-layer model could reproduce the intensity, the strong self-absorption dips and asymmetry of the observed spectrum quite well. The line wings may stem from outflowing gas, or also be due to a more complex velocity profile not incorporated in the model. The important outcome of this model can be summarised as follows: First, the strong self-absorption in the line profiles can be reproduced by the high-density core together with the low-density envelope. Second, the asymmetry of the line profiles can be explained by the infalling envelope (we got v km s), given the relatively high turbulent velocities in the core (v km s) and in the envelope (v km s). However, this configuration is not the only way to fit the line asymmetries. In particular, we were able to reproduce the line asymmetries by the combination of an expanding core and a static envelope because of the similar relative velocities. We consider this scenario implausible, though, because of the short outflow timescales.
Finally, we stress that the considered simple models did not reliably constrain the other parameters of the cores, but they illustrated that the observed spectra can indeed be explained theoretically. In order to get more reliable information about the core properties one has to consider a set of molecular lines and transitions together with more physical models of the source. Unfortunately, the available data, in particular the HCO(1-0) measurements, are not sufficient to enable this. Indeed, if we used HCO(1-0) in addition to HCN(1-0) in our fitting routine, we would have to incorporate two more parameters for the HCO abundance (in the core and in the envelope), while HCO(1-0) alone will not allow us to further constrain the excitation conditions. For the latter, at least a few more transitions of HCO or its isotopomers were necessary. On the other hand, a more physical model of the core (e.g. with smooth distributions) has more parameters to fit which makes such a model even more uncertain. Thus, the usage of more sophisticated models is required when detailed spectral maps of the sources are available.
|Parameter||SMM1 North||SMM1 South|
|v (km s)|
|v (km s)|
|Two-layer Model for SMM1 South|
|v (km s)|
|v (km s)|
4.3.3 Collapse of SMM1 South
The modelling of the HCN emission showed that the spectrum of SMM1 South can be explained with a collapse of the core. From the resulting velocity we calculated estimates for the mass infall rate using dM/dt M/t M v/r. For SMM1 South, v km s is the regular velocity of the envelope in the two-layer model. Together with the mass and radius from Table 1 this gave an infall rate of M yr. This is consistent with the mass outflow rates, but is only a crude estimate.
4.3.4 Internal turbulence
Additional parameters we obtained from the HCN emission are the internal turbulent velocities of the cores that reproduced the observed line widths. However, in both cases rotational motions, more complex velocity profiles, and also outflows may contribute to the line widths. Such effects were not considered in our models. In the case of SMM1 South, the fitted model shows a very turbulent core and less turbulence in the envelope. When compared to the isothermal sound speed of around 0.2 km s for cold cores (T K), the core turbulence of 1.7 km s is highly supersonic. The turbulent velocity in the envelope of 0.5 km s is on the scale of the sound speed but still in the supersonic regime. The SMM1 North HCN spectrum indicates a more quiescent state of this core. The linewidths are reproduced with a turbulent velocity of 0.4 km s.
5. Summary and conclusions
We have investigated the onset of star formation in the massive cold clump (catalog ISOSS J18364-0221) SMM1. The derived results regarding the clump fragmentation, the core and outflow properties, and the collapse indications are summarises as follows:
Two compact, embedded cores are the only objects we identified. Thus, this clump shows little fragmentation, but seems to rather produce a low number of compact objects.
The cores dubbed SMM1 North and South are of about 10000 and 9000 AU radius. The dust temperatures are 15 and 22 K and they have masses of about 15 and 10 M, respectively. SMM1 South harbours an infrared source not detected at 8, but at 24 and 70 . The luminosity of SMM1 South is about 180 L and a molecular outflow is detected in its vicinity. SMM1 North lacks infrared emission, but a second molecular outflow is emerging from this core. The luminosity of SMM1 North is approximately 20 L. Both outflows appear collimated and we derived lower outflow mass limits of about 0.2 and 0.3 M of presumably entrained gas, and mass outflow rates of about M yr. In Fig. 7 we show a schematic illustration of these results.
We detected filaments of shock-excited H emission at 2.122 towards the lobes of the SMM1 South molecular outflow, and also at distances up to 4.6 pc from SMM1. Each filament is roughly aligned with one of the outflow axes. They are possibly tracing the outflow-generating jets, and the dynamical timescales we derived are consistent with those of the molecular outflows considering the uncertain jet velocities. For both molecular outflows, ages of less than yr were found.
SMM1 South is bright in HCN(1-0) and our modelling results support that the core is collapsing. We got an infall velocity of 0.14 km s. This results in a mass infall rate estimate of M yr. For both cores the spectra can be interpreted with turbulent internal motions, and in the case of SMM1 South the innermost part appears highly supersonic.
Both SMM1 North and South are more massive than typical low-mass dense cores and embedded in the presumably contracting 75 M SMM1 clump. This suggests that the two cores harbour protostellar seeds that may become intermediate- to high-mass stars. The presence of a 24 source, a very turbulent central region and the jet features at large distances to the core support that SMM1 South is more evolved than SMM1 North. We interpret it as having an embedded young protostar that constitutes the driving source of the outflow. The outflows as well as the infall indicate that it is accreting. The associated outflow indicates that SMM1 North is also a star-forming core, and the outflow energetics are similar. However, the forming object remains undetected, and the relatively low level of turbulence implies that it has only marginally affected the surrounding core. The presumed difference in evolution of the forming objects contrast with the similar sizes, densities and outflow properties of the cores. These findings suggest a rapid evolution of the luminosity as predicted for high accretion rates (Yorke, 2008).
In comparison to the core collapse simulations of Krumholz et al. (2007) we find that the luminosity of SMM1 South would be reached at a stage where the embedded protostellar mass is around 0.3 M. Accretion onto the protostar is the dominant luminosity source, and they find an accretion rate on the order of 10 M yr from directly after the formation of the protostar until this stage. The resulting timescale of about 3000 yr is consistent with the outflow timescales. Also the mass outflow rate as well as the infall rate estimate may be consistent with the high accretion rate. SMM1 North is less luminous, and hence the protostellar mass would be even lower, but may still be consistent with a high accretion rate in their model. However, the formation of outflows is not incorporated therein.
- André & Montmerle (1994) André, P., & Montmerle, T. 1994, ApJ, 420, 837
- Arce & Sargent (2006) Arce, H. G., & Sargent, A. I. 2006, ApJ, 646, 1070
- Bally (2008) Bally, J. 2008, in ASP Conf. Ser. 387, Massive Star Formation: Observations Confront Theory, ed. H. Beuther, H. Linz, & Th. Henning (San Francisco: ASP), 158
- Banerjee & Pudritz (2007) Banerjee, R., & Pudritz, R. E. 2007, ApJ, 660, 479
- Baumeister et al. (2003) Baumeister, H., Bizenberger, P., Bayler-Jones, C. A. L., Kovács, Z., Röser, H.-J., & Rohloff, R.-R. 2003, in Proc. SPIE 4841, Instrument Design and Performance for Optical/Infrared Ground-based Telescopes, ed. M. Iye & A. F. M. Moorwood, 343
- Beuther et al. (2008) Beuther, H., Linz, H., & Henning, Th., 2008, ASP Conf. Ser. 387, Massive Star Formation: Observations Confront Theory (San Francisco: ASP)
- Beuther et al. (2002a) Beuther, H., Schilke, P., Gueth, F., McCaughrean, M., Andersen, M., Sridharan, T. K., & Menten, K. M. 2002a, A&A, 387, 931
- Beuther et al. (2002b) Beuther, H., Schilke, P., Sridharan, T. K., Menten, K. M., Walmsley, C. M., & Wyrowski, F. 2002b, A&A, 383, 892
- Beuther & Shepherd (2005) Beuther, H., & Shepherd, D. 2005, Reprinted from Astrophysics and Space Science Library 324, Cores to Clusters: Star Formation with Next Generation Telescopes, ed. M. S. N. Kumar, M. Tafalla, & P. Caselli (New York: Springer), 105
- Beuther & Steinacker (2007) Beuther, H., & Steinacker, J. 2007, ApJ, 656, L85
- Birkmann et al. (2007) Birkmann, S. M., Krause, O., Hennemann, M., Henning, Th., Steinacker, J., & Lemke, D. 2007, A&A, 474, 883
- Birkmann et al. (2006) Birkmann, S. M., Krause, O., & Lemke, D. 2006, ApJ, 637, 380
- Bonnell et al. (2006) Bonnell, I. A., Clarke, C. J., & Bate, M. R. 2006, MNRAS, 368, 1296
- Cabrit & Bertout (1990) Cabrit, S., & Bertout, C. 1990, ApJ, 348, 530
- Choi et al. (1993) Choi, M., Evans, II, N. J., & Jaffe, D. T. 1993, ApJ, 417, 624
- di Francesco et al. (2007) di Francesco, J., Evans, II, N. J., Caselli, P., Myers, P. C., Shirley, Y., Aikawa, Y., & Tafalla, M. 2007, in Protostars and Planets V, ed. B. Reipurth, D. Jewitt, & K. Keil (Tucson: Univ. Arizona Press), 17
- Elias (1980) Elias, J. H. 1980, ApJ, 241, 728
- Engelbracht et al. (2007) Engelbracht, C. W. et al. 2007, PASP, 119, 994
- Fazio et al. (2004) Fazio, G. G. et al. 2004, ApJS, 154, 10
- Gonzalez-Alfonso & Cernicharo (1993) Gonzalez-Alfonso, E., & Cernicharo, J. 1993, A&A, 279, 506
- Gordon et al. (2007) Gordon, K. D. et al. 2007, PASP, 119, 1019
- Hennemann et al. (2008) Hennemann, M., Birkmann, S. M., Krause, O., & Lemke, D. 2008, A&A, 485, 753
- Henning et al. (2000a) Henning, Th., Lapinov, A., Schreyer, K., Stecklum, B., & Zinchenko, I. 2000a, A&A, 364, 613
- Henning et al. (2000b) Henning, Th., Schreyer, K., Launhardt, R., & Burkert, A. 2000b, A&A, 353, 211
- Hosokawa & Omukai (2008) Hosokawa, T., & Omukai, K. 2008, in ASP Conf. Ser. 387, Massive Star Formation: Observations Confront Theory, ed. H. Beuther, H. Linz, & Th. Henning (San Francisco: ASP), 255
- Jappsen et al. (2005) Jappsen, A.-K., Klessen, R. S., Larson, R. B., Li, Y., & Mac Low, M.-M. 2005, A&A, 435, 611
- Jenness & Economou (1999) Jenness, T., & Economou, F. 1999, in ASP Conf. Ser. 172, Astronomical Data Analysis Software and Systems VIII, ed. D. M. Mehringer, R. L. Plante, & D. A. Roberts (San Francisco: ASP), 171
- Jenness & Lightfoot (1998) Jenness, T., & Lightfoot, J. F. 1998, in ASP Conf. Ser. 145, Astronomical Data Analysis Software and Systems VII, ed. R. Albrecht, R. N. Hook, & H. A. Bushouse (San Francisco: ASP), 216
- Kirk et al. (2007) Kirk, H., Johnstone, D., & Tafalla, M. 2007, ApJ, 668, 1042
- Krause (2003) Krause, O. 2003, Ph.D. Thesis, Combined Faculties for the Natural Sciences and for Mathematics of the Ruperto-Carola University of Heidelberg, Germany
- Krause et al. (2003) Krause, O., Lemke, D., Tóth, L. V., Klaas, U., Haas, M., Stickel, M., & Vavrek, R. 2003, A&A, 398, 1007
- Krause et al. (2004) Krause, O., Vavrek, R., Birkmann, S., Klaas, U., Stickel, M., Tóth, L. V., & Lemke, D. 2004, BaltA, 13, 407
- Krumholz (2008) Krumholz, M. R. 2008, in ASP Conf. Ser. 387, Massive Star Formation: Observations Confront Theory, ed. H. Beuther, H. Linz, & Th. Henning (San Francisco: ASP), 200
- Krumholz et al. (2007) Krumholz, M. R., Klein, R. I., & McKee, C. F. 2007, ApJ, 656, 959
- Krumholz & McKee (2008) Krumholz, M. R., & McKee, C. F. 2008, Nature, 451, 1082
- Krumholz et al. (2005) Krumholz, M. R., McKee, C. F., & Klein, R. I. 2005, ApJ, 618, L33
- Lemke et al. (1996) Lemke, D. et al. 1996, A&A, 315, L64
- Low et al. (2005) Low, F. J., Smith, P. S., Werner, M., Chen, C., Krause, V., Jura, M., & Hines, D. C. 2005, ApJ, 631, 1170
- Makovoz & Marleau (2005) Makovoz, D., & Marleau, F. R. 2005, PASP, 117, 1113
- McKee & Tan (2003) McKee, C. F., & Tan, J. C. 2003, ApJ, 585, 850
- Motte et al. (2007) Motte, F., Bontemps, S., Schilke, P., Schneider, N., Menten, K. M., & Broguière, D. 2007, A&A, 476, 1243
- Noriega-Crespo et al. (2004) Noriega-Crespo, A. et al. 2004, ApJS, 154, 352
- Ossenkopf & Henning (1994) Ossenkopf, V., & Henning, Th. 1994, A&A, 291, 943
- Osterloh et al. (1997) Osterloh, M., Henning, Th., & Launhardt, R. 1997, ApJS, 110, 71
- Pavlyuchenkov et al. (2008) Pavlyuchenkov, Ya., Wiebe, D., Shustov, B., Henning, Th., Launhardt, R., & Semenov, D. 2008, ApJ, in press (arXiv:0808.2375)
- Pavlyuchenkov & Shustov (2004) Pavlyuchenkov, Y. N., & Shustov, B. M. 2004, Astronomy Reports, 48, 315
- Press et al. (1992) Press, W. H., Teukolsky, S. A., Vetterling, W. T., & Flannery, B. P. 1992, Numerical recipes in FORTRAN. The art of scientific computing (2nd ed.; Cambridge: University Press)
- Rathborne et al. (2007) Rathborne, J. M., Simon, R., & Jackson, J. M. 2007, ApJ, 662, 1082
- Reach et al. (2005) Reach, W. T. et al. 2005, PASP, 117, 978
- Redman et al. (2008) Redman, M. P., Khanzadyan, T., Loughnane, R. M., & Carolan, P. B. 2008, in ASP Conf. Ser. 387, Massive Star Formation: Observations Confront Theory, ed. H. Beuther, H. Linz, & Th. Henning (San Francisco: ASP), 38
- Reipurth & Bally (2001) Reipurth, B., & Bally, J. 2001, ARA&A, 39, 403
- Rieke et al. (2004) Rieke, G. H. et al. 2004, ApJS, 154, 25
- Sandell & Weintraub (2001) Sandell, G., & Weintraub, D. A. 2001, ApJS, 134, 115
- Sault et al. (1995) Sault, R. J., Teuben, P. J., & Wright, M. C. H. 1995, in ASP Conf. Ser. 77, Astronomical Data Analysis Software and Systems IV, ed. R. A. Shaw, H. E. Payne, & J. J. E. Hayes (San Francisco: ASP), 433
- Schöier et al. (2005) Schöier, F. L., van der Tak, F. F. S., van Dishoeck, E. F., & Black, J. H. 2005, A&A, 432, 369
- Smith & Rosen (2005) Smith, M. D., & Rosen, A. 2005, MNRAS, 357, 1370
- Stahler & Palla (2005) Stahler, S. W., & Palla, F. 2005, The Formation of Stars (Weinheim: Wiley-VCH)
- Stickel et al. (2007) Stickel, M., Krause, O., Klaas, U., & Lemke, D. 2007, A&A, 466, 1205
- Vorobyov & Basu (2006) Vorobyov, E. I., & Basu, S. 2006, ApJ, 650, 956
- Walmsley et al. (1982) Walmsley, C. M., Churchwell, E., Nash, A., & Fitzpatrick, E. 1982, ApJ, 258, L75
- Walsh et al. (2007) Walsh, A. J., Myers, P. C., Di Francesco, J., Mohanty, S., Bourke, T. L., Gutermuth, R., & Wilner, D. 2007, ApJ, 655, 958
- Ward-Thompson et al. (2007) Ward-Thompson, D., André, P., Crutcher, R., Johnstone, D., Onishi, T., & Wilson, C. 2007, in Protostars and Planets V, ed. B. Reipurth, D. Jewitt, & K. Keil (Tucson: Univ. Arizona Press), 33
- Ward-Thompson et al. (1999) Ward-Thompson, D., Motte, F., & Andre, P. 1999, MNRAS, 305, 143
- Werner et al. (2004) Werner, M. W. et al. 2004, ApJS, 154, 1
- Whitney et al. (1997) Whitney, B. A., Kenyon, S. J., & Gomez, M. 1997, ApJ, 485, 703
- Wu et al. (2004) Wu, Y., Wei, Y., Zhao, M., Shi, Y., Yu, W., Qin, S., & Huang, M. 2004, A&A, 426, 503
- Yorke (2008) Yorke, H. W. 2008, presented at EPoS 2008: The Early Phase of Star Formation – The future of the field, ed. J. Steinacker & A. Bacman (Ringberg Castle Rottach-Egern: Max-Planck-Gesellschaft), http://www.mpia-hd.mpg.de/homes/stein/EPoS2008/index.html
- Yorke & Bodenheimer (2008) Yorke, H. W., & Bodenheimer, P. 2008, in ASP Conf. Ser. 387, Massive Star Formation: Observations Confront Theory, ed. H. Beuther, H. Linz, & Th. Henning (San Francisco: ASP), 189
- Zhang (2005) Zhang, Q. 2005, in IAU Symp. 227, Massive Star Birth: A Crossroads of Astrophysics, ed. R. Cesaroni, M. Felli, E. Churchwell, & M. Walmsley (Cambridge: University Press), 135
- Zhang et al. (2007) Zhang, Q., Sridharan, T. K., Hunter, T. R., Chen, Y., Beuther, H., & Wyrowski, F. 2007, A&A, 470, 269
- Zinnecker & Yorke (2007) Zinnecker, H., & Yorke, H. W. 2007, ARA&A, 45, 481