4. Results and derived properties
4.1. General survey statistics
Out of the 35 globules observed, all globules were detected in the 12CO(2-1) line (3 detection limit = 0.3 K), 24 globules were detected in the CS(2-1) line (detection rate 69%, detection limit 0.2 K), and 18 globules were detected in the 1.3 mm continuum emission (51%, detection limit 40 mJy/beam). Due to the differences in nature and distances of the sources, we analyse the results separately for the following object groups: local goup 1 sources (d 500 pc, = 300 pc; 9 objects), local group 2 sources (9 objects), sources in the near Carina arm (0.7 kpc d 1.3 kpc, = 0.9 kpc; 10 objects), and sources in the far Carina arm (2.2 kpc d 3.7 kpc, = 3.2 kpc; 6 objects). The objects in the Carina arm belong mostly to group 1. Three objects could not be classified and one globule (DC 313.3-0.3) contains three IRAS sources (all group 1).
The ASURV 1 software package was used for the statistical analysis of the results. This package performs a "survival analysis" which takes upper limits into account. Parameter distributions and mean values were derived by the cumulative Kaplan-Meier estimator. Two-sample tests were performed in order to check the significance of differences or similarities in the parameter distributions of the globule groups.
4.2. Millimetre continuum data
4.2.1. Detection statistics
In Table 3, we compile the measured 1.3 mm continuum flux densities per beam together with the r.m.s. noise and the IRAS point source fluxes (detected objects are printed boldface). For non-detections, the 3 detection limits are listed. Five of the detected sources were meanwhile mapped with the bolometer at 1.3 mm. The maps were already or will be discussed in more detail in other papers (Bourke et al. 1997; Launhardt et al. 1998b).
Table 3. Results - IRAS and 1.3 mm fluxes, luminosities, and masses
The overall detection rate of 51% for the 1.3 mm continuum radiation in this survey is higher than the rate of 35% obtained for the northern globule sample (Paper I). The reason for this discrepancy is that due to our selection criteria the southern sample investigated here is much more biased towards globules with embedded protostars than the northern sample (cf. Sect. 2.1). Excluding the far Carina sources in the southern sample and counting only group 1 and 2 sources in the northern sample, the overall detection rate of the southern sample amounts to 50% vs. 70% for the northern sample. The reason for the lower detection rate in this southern survey compared to the northern survey is probably the lower sensitivity of the SEST observations (3 detection limit of 40 mJy/beam) compared to the IRAM observations of the northern globules (17 mJy/beam).
Since the uncertainty of the IRAS positions (major axis of the position error ellipse) is often as large as the SEST beam at 1.3 mm ( to with a mean of vs. a beam size of ), the measured mm fluxes probably do not always correspond to the emission maximum. This could especially apply to the group 2 sources which were not detected in the shorter IRAS bands. On the other hand, we do not expect the group 2 sources to have point-like mm emission, so that the detection rates should not be biased (cf. Sect. 5.5).
Despite of this uncertainty, the detection rates for the two globule groups differ significantly from each other, viz. 77% vs. 17% for group 1 and 2, respectively. For the local sources, the detection rates are 78% (gr. 1) and 22% (gr. 2). These detection rates compare well to rates obtained for the northern sample (Paper I; 94% for group 1, 20% for group 2). Although the far Carina sources have much higher masses and luminosities than the local globules and are, therefore, assumed to be of different nature (cf. Sects. 4.2and 5.2), we found no significant difference in the detection rates between the local and the Carina sources.
4.2.2. Masses and column densities
Table 3 also lists the total gas mass per beam for the sources detected at 1.3 mm. The masses were derived from the 1.3 mm continuum flux densities by assuming optically thin configurations and an isothermal, uniformly distributed population of dust grains (see Paper I). To avoid confusion, we applied a uniform set of parameters for the derivation of mass to all objects: a dust opacity of 0.8 cm2 per gram of dust (Ossenkopf & Henning 1994), a total gas-to-dust mass ratio of 150 (see Paper I), and an average dust temperature of 25 K. The dust temperature was derived from greybody fits to the sub-mm/mm SEDs of ten northern globule cores with similar IRAS characteristics (Launhardt et al. 1997; nine sources were group 1, one source group 2) and is consistent with the FIR colour temperature of these objects (cf. Sect. 2.1).
All masses were derived under the assumption of optically thin dust emission which certainly holds for most objects since massive disks (optically thick at 1.3 mm) were shown to be rare (Terebey et al. 1993). The assumption that the mm continuum emission arises mainly from optically thin circumstellar envelopes than from optically thick disks is supported by the correlation between (1.3 mm) and T(CS) (Fig. 6) since disks emit comparatively little line emission. The relative uncertainty of the derived masses is estimated to be a factor of 2, due to the imprecisely known values of the mass opacity and the dust temperature (see Henning et al. 1995, or Gordon 1995 for a discussion of the uncertainties in masses derived from submillimetre continuum data). To approximately correct the derived masses for the actual dust temperature, the mass has to be scaled with (adopted)/. This also means that the derived masses are not very sensitive to the adopted temperature.
With the above input parameters, we derive masses of 0.04 to 2.3 /beam for the local group 1 sources (d 500 pc), with a mean value of 0.60.25 /beam. The average mass of the local group 2 sources is 0.050.01 /beam, with the only two detected objects having masses of 0.1 /beam. Adopting a more realistic dust temperature of 20 K and a lower dust opacity value of 0.5 cm2 g-1 for the group 2 sources, the derived masses would increase by a factor of two. The average 3 detection limit for the local sources is 0.2 /beam. The masses of the Carina sources (mainly group 1) range from 1 to 37 /beam with average values of 1.70.4 /beam and 296 /beam for the near and far Carina sources, respectively. The mean beam-averaged column densities of the (local) group 1 and 2 sources are = (55) 1022cm-2 and = (10.5) 1022 cm-2, respectively. The significance of the difference in the detection rates and masses between the two groups will be discussed in more detail in Sect. 5.2.1.
At a distance of 300 pc, the HPBW of corresponds to a linear diameter of 0.033 pc or 7000 AU, respectively, which is smaller than the diameter of an infalling protostellar envelope at an infall age of some yrs (sound speed of 0.2 km s-1, see Sect. 5.4). For the local sources, the measured continuum fluxes can, therefore, be assigned to "circumstellar" material. Since the sources may be more extended than the beam, the derived masses are, however, lower limits to the total circumstellar masses. The masses derived for the more distant (Carina) sources may include more extended material which is probably not "circumstellar".
The circumstellar masses can also be derived from the 100 µm IRAS fluxes. Here, we use the same effective dust temperature of 25 K and a dust opacity of 80 cm2 per gram of dust (Ossenkopf & Henning 1994). The assumption of optically thin emission holds at this wavelength for all sources when assuming that the bulk of the emission does not originate from a compact disk. We find a good correlation (r = 0.91) between the masses derived from the 1.3 mm continuum emission and from the 100 µm IRAS point source fluxes, although the two mass estimates are not exactly equal, but differ systematically. While the mm masses of the local group 1 sources are, on average, 2.5 times higher than the 100 µm masses, the mm masses of the (detected) local group 2 sources are only half as large as the 100 µm masses. The systematically lower ratio of the group 2 sources compared to the group 1 sources suggests that these objects are more extended than the group 1 sources. Obviously, the derived mm masses per beam underestimate the total masses of the group 2 sources stronger than those of the group 1 sources. In general, the relative good agreement between the 1.3 mm masses (HPBW = ) and the 100 µm IRAS masses suggests that the bulk of the dust continuum emission from these sources arises from a core which is more compact than . This applies estecially to the group 1 sources. There is no evidence that the bulk of the dust in these sources is too cold to be detected by IRAS at 100µm.
At the current stage of investigation we have no information about the morphology and the total masses of the globule cores. However, the higher mm flux densities (and masses) per beam and the stronger CS lines (cf. Sect. 4.3.4) suggest that the group 1 sources are more centrally condensed than the group 2 sources and have, thus, higher central densities. Considering only the local globules ( pc), the beam-averaged volume densities (derived from the continuum emission under the assumption of spherical sources with diameters corresponding to the beam size) of the group 1 sources range from 5 104 to 5 106 cm-3, with a mean value of cm-3. In case of an internal density gradient, which we expect to be present in such star-forming cores, the local densities in the core centres may be much higher.
In contrast, the group 2 sources have a mean beam-averaged volume density of cm-3, with the two sources detected at 1.3 mm continuum (DC 249.4-5.1 and 289.3-2.8) having a mean density of 1 105 cm-3. In spite of the larger beam size, these densities compare well to the values derived for the northern globule sample (Paper I). The critical density of the CS (2-1) line is 3 105 cm-3 which is just between the mean beam-averaged densities of the two object groups. Hence, the densities derived from the dust continuum emission are consistent with the finding that group 1 globules have in general CS cores while group 2 globules mostly don't have CS cores (cf. Sect. 4.3.1). The beam-averaged densities of the Carina sources are by more than an order of magnitude lower than those of the local globules which results from the large projected beam size.
Table 3 also lists the bolometric luminosities and the 1.3 mm luminosities . The 1.3 mm luminosity was calculated with a bandwidth of = 50 GHz. The bolometric luminosities were obtained by integrating under the NIR fluxes taken from Persi et al. (1990), the IRAS point source fluxes, and the 1.3 mm fluxes. If a source was not detected in the lower IRAS bands, a value of one half of the upper limit was adopted for the last non-detected band. The flux densities of all other non-detected bands were set to zero. For the wavelength region from 135 to 1300m, we integrated under greybody curves fitted to the 60, 100, and 1300m fluxes. If no NIR photometry was available for a source, we list the luminosity integrated between 7 and 1300 µm. The luminosity of these young objects comes mainly from the FIR spectral region where the SEDs have their maximum (between 100 and 300 µm). Note, however, that for group 1 sources the NIR contribution to can be as high as 30% (e.g., DC 267.7-7.4). For the group 2 sources which are assumed not to have embedded YSOs and of which none was found to have an optical counterpart (star or nebulosity), the contribution from wavelengths shorter than 7 µm is negligible.
The local objects span a luminosity range between 0.3 and 17 , with the mean values of the group 1 and 2 sources being 6.72.0 and 1.50.4 , respectively. The average luminosities of the near and far Carina sources are 5015 (without DC 313.3-0.3) and 800400 (without DC 296.2-3.6), respectively. The far Carina source DC 296.2-3.6 has an exceptional high luminosity ( ). There are no significant differences between group 1 and 2 sources in the Carina arm.
One has to keep in mind that more than one object may contribute to the IRAS luminosity due to the large beam of IRAS . This could especially apply to the more distant objects. The derived luminosity may also be different from the bolometric luminosity if the source is not spherically symmetric (see, e.g., Men'shchikov & Henning 1997). The assumption of spherical symmetry (isotropic emission) is, however, assumed to be fulfilled for deeply embedded sources without optical or NIR counterparts. In spite of these uncertainties, there is the general tendency that group 1 objects are more luminous than group 2 objects and that objects with higher luminosities are associated with more massive globule cores (cf. Fig. 11).
4.3. Molecular line data
4.3.1. Detection statistics
The results of the CO and CS line observations are summarized in Table 4. We list the centre velocity , the main beam temperature , the line width (FWHM), and the integrated line intensity for both lines. The given temperatures are the observed peak line temperatures corrected for the main beam efficiency. The line widths and center velocities were obtained from Gaussian fits to the lines after masking out line wings, self absorption features, and line components at other but nearby velocities. The integrated line intensities include the emission from line wings. If a line was contaminated by a second component (e.g. DC 275.9+1.9), this second line was modeled with a Gaussian profile and subtracted before integrating under the line of interest.
Table 4. Results - CO and CS line parameters
All sources were detected in the CO line. If more than one CO line was detected, the CS emission was used to discriminate between the globule and foreground or background CO emission. The detection rates for the CS (2-1) line emission are similar to those for the dust continuum emission. While the overall detection rate in the CS line is 69% (detection limit = 0.2 K), the detection rates for the (local) group 1 and 2 sources are 90% and 40%, respectively. As for the mm continuum emission, we found no significant difference in the CS detection rates between the local and the Carina sources. Fig. 6 (left panel) shows that there is a clear correlation between the mm continuum flux densities and the CS line peak temperatures. The objects which were not detected in one of the two tracers are mostly close to the detection limit in the other tracer. For the sake of comparison, we included the northern globule sources (only group 1, Paper I) in the diagram. The northern sources tend to have stronger CS lines compared to the southern sources. Note, however, that the IRAM observations of the northern globules were performed with a smaller beam than the SEST observations of the southern globules. The higher beam filling factor of the dense gas in the smaller beam might explain the stronger CS lines of the northern sources. Although no distance effects nor the different beam sizes of the continuum and CS observations were considered here, this correlation indicates that the optically thin CS line keeps growing with column density (cf. discussion in Launhardt et al. 1998a; see also Paper I).
We also evaluated the ammonia survey of Bourke et al. (1995b). Out of the 20 globules of our sample which were observed in NH3 by these authors, 12 globules belong to group 1 and 8 globules to group 2. The NH3 detection rates for the two globule groups are very similar to those for the CS line or for the mm continuum, viz. 83% for group 1 and 37% for group 2. Fig. 6 (right panel) shows the clear correlation between the peak temperatures of both lines. It can also be seen that group 1 sources have, on average, much stronger lines than group 2 sources. The only exceptions from this rule are DC 267.2-7.2 - a local group 2 source with strong lines (cf. Sect. 5.1), and DC 276.2-10.6 and 354.2+3.2 - two local group 1 sources with very weak lines. All three objects were not detected at 1.3 mm continuum.
Alltogether, these results show that the 1.3 mm dust continuum, CS, and NH3 are all roughly equivalent tracers of dense, star-forming globule cores.
4.3.2. CO(2-1) line characteristics
The 12CO (2-1) line temperatures () of the local globules range from 1.9 K (DC 275.9+1.9) to 8.9 K (DC 297.7-2.8), with mean values 5.90.8 K and 5.30.6 K for group 1 and 2 sources, respectively. Assuming that the CO line is optically thick and thermalized, the mean radiation temperatures translate into kinetic gas temperatures of 11 K and 10 K for group 1 and 2 sources, respectively. These values compare well to the mean kinetic temperature derived for the northern globules ( 10 K; CB88; CYH91; Lemme et al. 1996; see also Paper I). For six of the southern group 1 globules, Bourke et al. (1995b) derived an average kinetic gas temperature of K (from NH3, HPBW = ).
The peak line temperatures of the two globule groups differ only marginally, indicating high optical depths and similar kinetic gas temperatures. The linewidths and, hence, the integrated intensities differ more significantly. Here, we consider only the Gaussian cores of the lines but not the line wings. The mean linewidth and integrated intensity for group 1 (local objects) are 2.30.25 km s-1 and 20.44.6 K km s-1 vs. 1.60.13 km s-1 and 9.41.7 K km s-1 for group 2. Although the line broadening can have many reasons (e.g., turbulence, outflows, infall, rotation) which do not necessarily have something to do with star-formation activity, the broader CO lines of the group 1 sources indicate that these globules have a more complex gas dynamics than the group 2 globules. The average CO linewidths compare well to the linewidths found for the northern globules (CYH91; see Paper I).
No significant difference in the CO line parameters was found between the local globules and the near Carina sources. However, the far Carina sources (d 2 kpc, mostly group 1) have higher CO line temperatures (8.41.2 K), broader lines (3.40.3 km s-1), and larger integrated intensities (323.5 K km s-1) than the local globules. Taking into account a lower beam filling factor for these distant sources, their mean kinetic gas temperature must be higher than 15 K (see discussion in Sect. 5.6).
In 12 globules CO line wings were found indicating the presence of molecular outflows (DC numbers printed boldface in Table 4). Five of the objects with CO line wings have also wings in the CS line (also marked in Table 4). The CO(2-1) spectra of the 12 outflow candidates are shown in Fig. 7. All spectra were taken towards the positions listed in Table 1. In addition, these objects are marked in the colour-colour diagram (Fig. 8). It can be clearly seen that most of the sources with strong millimetre continuum emission show outflow activity and vice versa (cf. Cabrit & André 1991; Bontemps et al. 1996).
Of the 12 sources showing line wings, four objects were already known to have outflows: DC 253.3-1.6, 267.4-7.5, 297.7-2.8, and 303.8-14.2 (for references see Sect. 2.1). All four objects are local group 1 sources. The first three of these objects are those with the strongest line wings in our sample. These three sources have also high mm continuum fluxes corresponding to high circumstellar masses ( , see Table 3) indicating that they are very young (cf. discussion in Sect. 5.4). In case of 8 globules, we detected CO line wings for the first time, which increases the known number of globules with outflows considerably. This shows that the outflow phenomenon is quite widespread in this kind of globules and ultimatively proofs that we see evidence for star formation in these objects.
Out of the 12 outflow candidates 11 objects belong to the group 1 and one object (DC 356.5-4.5) belongs to group 2. Excluding again all Carina arm sources, 6 out of 9 group 1 globules and one out of 9 group 2 globules have outflows. This gives an outflow rate of 67% for group 1 which is nearly the same value as the rate of 65% derived for the northern group 1 globules (Yun & Clemens 1992; Paper I). All group 1 outflow sources were also detected at 1.3 mm continuum. DC 356.5-4.5 (group 2) is the only source with a CO line wing which was not detected at 1.3 mm continuum nor in the CS line. Since the detected (red) line wing is very weak, we cannot exclude that this line wing is caused by an independent velocity component and has possibly nothing to do with an outflow. Very weak line wings were further found in the CO spectra of DC 267.7-7.4 (local globule, gr. 1) and 289.9-3.2 (far Carina arm).
Eight of the 11 outflow candidates remaining (excluding DC 356.5-4.5) were found to have both blue and red line wings indicating the presence of bipolar outflows. The red wing in the spectrum of DC 275.9+1.9 is probably blended by a second line. Two globules (DC 267.7-7.4, and 344.6-4.3) have only blue line wings and one globule (DC 289.9-3.2) has only a red line wing. Summarizing these results, we find that most of the detected outflows are bipolar and that neither blue nor red monopolar outflows are dominating the sample. Typical outflow velocities (maximum velocity extent from the line center at the 0.2 K level) are 7 km s-1 ( km s-1).
4.3.4. CS(2-1) line characteristics
The CS (2-1) peak line temperatures () of the detected sources vary between 0.2 K (DC 294.9+0.1) and 2.5 K (DC 253.3-1.6) with a mean value of 0.70.2 K, a mean linewidth of 1.30.1 km s-1, and a mean integrated intensity of 1.30.3 K km s-1. The average 3 detection limit for the non-detected sources is 0.20 K. In contrast to CO, the CS linewidths of group 1 and 2 sources (if detected) are very similar.
But, as expected from the different CS detection rates (Sect. 4.3.1), the line temperatures and, hence, the integrated intensities differ significantly between the two groups. The mean peak line temperature and integrated intensity for local group 1 sources are 1.10.3 K and 1.60.4 K km s-1 vs. 0.40.1 K and 0.70.3 K km s-1 for group 2, respectively. Considering only the detected (local) group 2 sources, the mean peak line temperature increases to 0.6 K which is still considerably lower than the mean value of the group 1 sources. This shows that dense cores in group 2 globules are not only less frequent than in group 1 globules, but that the cores are, if present, also less dense. The Carina sources have, on average, lower CS peak line temperatures than the local globules which comes probably due to the smaller beam filling factor of the dense gas in these distant objects.
© European Southern Observatory (ESO) 1998
Online publication: September 8, 1998