Online material

Appendix A: Comments on individual sources

Appendix A.1: 16061−5048c4 and 16435−4515c3

16061−5048c4 is included in the SFS due to the presence of a water maser. No clear mid-IR emission can be seen in the Spitzer images, suggesting a clump in very early stages of evolution or a very high extinction.

For example, Breen & Ellingsen (2011) suggest that a maser is one of the first sign of active star formation to appear. This is because the maser is likely excited by the outflow, the onset of which happens in a phase where the mid-IR emission may still be too weak to be detected. This seems to be consistent with the position of this source in the M-L plot. This is, in fact, one of the objects with the lowest kinetic temperature in the SFS and with a very low luminosity.

However, 16061−5048c4, in addition to the maser, shows a clear detection in the radio-continuum emission. Visual inspection of the radio image shows an elongated morphology of the emission at this wavelength, following closely the shape of the mm-clump, and displaced towards South-East. We conclude that this is not a compact Hii region produced by the central YSO, but an ionization front generated by a nearby massive star (most likely the corresponding IRAS source), rather than by the YSO.

A similar situation is observed in 16435−4515c3. This clump shows no signs of active star formation, apart from radio emission. It could be that also in this case this emission comes from an ionization front outside the clump. In any case, this source was excluded from the analysis because the NH₃(2, 2) was not detected.

Appendix A.2: 17355−3241c1

As evolution proceeds, the massive ZAMS star delivers huge quantities of ionizing photons and energetic particles into the parent molecular cloud. This process disperses the clump, leaving only remnants of the original cloud, around an Hii region.

17355−3241c1 is the most evolved source in our sample, as can be seen in the mass-luminosity plot. This is the only Type 3 source in our sample, with still a detectable 1.2 mm flux, but with dominant IR-emission.

The properties of this clump appear to confirm the evolutionary scenario: it has the lowest mass, column and surface density of the entire sample, and a high temperature and luminosity. Moreover, the L/M ratio is the highest among all the observed clumps, above . The virial parameter α (~3.5) is the highest. This could be due to the destructive action of the central ZAMS star, dispersing the clump from which it was formed. On the other hand, this object does not show compact radio-continuum emission. The M − L plot shows that the mass of the central ZAMS star should be M ~ 6 M_⊙, corresponding to a B5 star, comparing its position in the M − L plot with the evolutionary tracks. The output of Lyman continuum photons of such a star is low, and below the detection limit of the radio observations (cf. Thompson 1984; Sánchez-Monge et al. 2013a). However the uncertainty on the stellar mass is quite large. The SED fit with Robitaille models suggests a stellar mass of ~8 M_⊙. Another possibility to explain the lack of radio continuum emission is that it is extended and filtered out by the interferometer.

Finally, we remind that in both cases the stellar mass is an upper limit, under the assumption that the luminosity is dominated by the most massive object.

The SED of this object is different from that typical of the other objects with appreciable emission in the mid-IR, being almost flat in terms of energy up to 3.6 μm. Also the Robitaille model indicates that this is an older object.

Appendix A.3: Undetected sources

A number of sources do not have detections in ammonia, especially in the (2, 2) transition. Among the SFS, they are weak also in (1, 1), suggesting low beam-averaged NH₃ column densities. 16164−4929c2 and 14166−6118c1 have a clear detection at 1.2 mm, but no ammonia counterpart. Near-IR images show that these two objects contain a star cluster. The SED, the morphology of the IR emission and radio-continuum observations of these two clumps show that they host a massive ZAMS star, that has already developed an Hii region, different from the nearby clumps detected in ammonia. 13563−6109c1 also shows signs of the presence of a recognizable embedded cluster in near-IR images, even if not as clear as 16164−4929c2 and 14166−6118c1.

On the other hand, four more sources with similar properties are found in the QS, but without clear signs of an embedded cluster. Two of these have quite strong NH₃(1, 1) detections, indicating very low gas temperatures. These clumps might be just too cold to be detected in (2, 2) or simply not massive enough to form stars with M > 8 M_⊙.

Finally, also 15454−5335c2 was not detected in ammonia. This source was identified at the edge of the SEST map. No IR source was found to be associated with it. The non-detection in NH₃ and visual inspection of the SEST map suggest that this might be just a noise spike in the SEST 1.2 mm continuum map.

Appendix B: SED fit

The modified black-body fit was done with a simple Bayesian approach, considering Gaussian uncertainties on the fluxes, taking into account the rms of the image and the calibration uncertainty at each wavelength (15% for SEST and SPIRE, 20% for PACS red and 10% for PACS blue fluxes: Beltrán et al. 2006; Swinyard et al. 2010; Poglitsch et al. 2010). We adopted a modified black body as a model, comparing the observed and predicted fluxes at each wavelength, and multiplying the probability for each point, to obtain the total probability of the model. We took a constant prior on the mass, while we used a Gaussian prior on β, with mean value μ = 2 and standard deviation σ = 2, and on the dust temperature T_d, with μ = 20 K and σ = 15 K. We considered temperatures between 5 and 50 K in 200 equal linear steps, masses between 10 and 10   000 M_⊙ in 200 equal logarithmic steps, and β between 0.1 and 4, in 40 equal linear steps. To test the dependence of the fit on the choice of prior, we also tried to use a constant prior on all the parameters of the fit. The results show that T_d, M and β are not sensitive to this choice, within the uncertainties. To derive the probability distribution, and thus the uncertainty on the single parameters we integrated over the other two parameters.

Appendix C: Tables and figures

© ESO, 2013