next previous
Up: A survey of SiO


6 Discussion

The SiO detection rate is the highest towards those H2O masers and UC HII regions which are associated with IRAS sources having large integrated FIR flux densities and luminosities. Since SiO emission is indicative of the presence of shocks, this tendency might be understood in terms of the correlation discovered by Felli et al. (1992), between the mechanical luminosities of outflows associated with H2O masers and IRAS source FIR luminosities.

On the other hand, the detection rate was found to be higher in the galactic longitude range $300^\circ \leq l < 60^\circ$ than elsewhere. By limiting the sample to objects which are associated with luminous IRAS point sources, it was shown that the higher detection rate in this part of the sky is not caused barely by the fact that the selected sources there are associated with FIR sources brighter than the average. Most sources in this longitude range are located between the galactocentric radii 3 and 7 kpc, which region is sometimes referred to as the "molecular ring''. It seems therefore possible that the conditions of the dense molecular cores in the inner 7 kpc of the Galaxy, where most of the massive star formation is taking place, are particularly favourable for gas-phase SiO production.

According to Schilke et al. (1997) shock velocities in excess of 10 km s-1 are needed for grain mantle evaporation, whereas the silicate cores are destroyed at velocities greater than about 25 km s-1. The majority of the SiO lines detected in the present survey have full-widths greater than 10 km s-1 indicating that high-velocity shocks indeed are an important production channel for SiO also in high-mass star forming regions. The line shapes suggest that SiO emission at large velocities arise from turbulent wakes behind bow-shocks, i.e. post-shock gas. The derived excitation temperatures of the SiO transitions and the (3-2)/(2-1) line ratios suggest moderate average densities, supporting the notion of filled-in cavities. In this gas component SiO is probably rather short-lived due to further oxidation to SiO2 (Schilke et al. 1997). Therefore high-velocity SiO emission is a sign of recent outflow activity.

Several of our target sources have been observed with the IRAM 30-m telescope in the transitions J=2-1, J=3-2 and J=5-4 of SiO by Acord et al. (1998). A comparison between the results shows that while the integrated line intensities and the FWHMs of the lines are comparable, the full widths observed at IRAM are systematically larger than those observed at SEST and Onsala. This is probably due to a combination of two effects. Firstly, high velocity emission regions are likely to be compact and better traced with the 30-m telescope, which has an effective aperture 3.4 times larger than the SEST at 3 mm. Second, in a quick survey like the present the spectral noise cuts the lowest intensity line wings.

As pointed out by Acord et al. (1998), due to the fact that the mass distribution as a function of velocity, m(v), approximately follows the power law $m(v) \propto v^{-\gamma}$, where $\gamma$ is close to 2 (Masson & Chernin 1992, 1994), the spectra of outflow sources often bear a resemblance of Lorentzian profiles characteristic of pressure broadening. In the latter case the observed full width depends heavily on the intensity threshold where it is determined. This arouses the suspicion that the distribution of FWs (full widths above two sigma) in Fig. 7 does not reflect true changes in the velocity extent of the SiO lines, but is determined by the S/N ratios. We have examined the dependence of the full widths on the S/N ratio in Fig. 13, where the ratios $FW/\Delta
v_{1/2}$ vs. $T_{\rm A}^{\rm peak}/T_{\rm A}^{\rm min}$ for the SiO(J=2-1) lines are plotted. $\Delta v_{1/2}$ is the FWHM of the line, $T_{\rm A}^{\rm peak}$ is the peak antenna temperature and $T_{\rm A}^{\rm min}$ is the intensity threshold used for estimating the FW. The ratio $T_{\rm A}^{\rm peak}/T_{\rm A}^{\rm min}$ represents here the S/N ratio of the smoothed spectra. For a Gaussian profile,

\begin{displaymath}
\frac{T_{\rm A}^{\rm peak}}{T_{\rm A}^{\rm min}} = 
\exp\lef...
 ...ln{2} \, \left(\frac{FW}{\Delta v_{1/2}}\right)^2 \right\} \; .\end{displaymath}

The ratio $FW/\Delta
v_{1/2}$ is thus proportional to the square root of the logarithm of the S/N ratio. The relation expected from Gaussian line shapes is plotted as a dotted curve in Fig. 8. For the sake of comparison we have also plotted the expected relation for Lorentzian shapes as a dashed curve. For the latter

\begin{displaymath}
\frac{T_A^{\rm peak}}{T_A^{\rm min}} = 
1 \, + \, \left(\frac{FW}{\Delta v_{1/2}}\right)^2 \; , \end{displaymath}

and the ratio $FW/\Delta
v_{1/2}$ is roughly proportional to the square root the S/N ratio. In Fig. 13 it can be seen that several spectra follow the Gaussian curve which slopes only gently beyond a S/N ratio of about 10. In these cases the determined full widths give reasonable estimates of the FWZIs (full widths at zero intensity). On the other hand, the observations above the Lorentzian curve have relatively low S/N ratios, and probably a longer integration would have revealed extensive line wings. Nevertheless, Fig. 13 bears evidence of intrinsically narrow lines and confirms that the histogram in Fig. 2 depicts underlying changes of the FWZIs for the emission regions covered by the SEST beam.

  
\begin{figure}
\psfig {file=fig13.eps,width=8.8cm,bbllx=10pt,bblly=70pt,bburx=550pt,bbury=800pt,angle=270}

~\end{figure} Figure 13: The ratio $FW/\Delta
v_{1/2}$ as a function of $T_{\rm A}^{\rm peak}/T_{\rm A}^{\rm min}$ for the SiO(J=2-1) lines detected with SEST (filled squares) and Onsala (open squares). $T_{\rm A}^{\rm min}$ is the intensity threshold used for the determination of FW (see text), and $T_{\rm A}^{\rm peak}/T_{\rm A}^{\rm min}$ thus represents the S/N ratio. The dotted curve represents the expected relation in the case of a Gaussian line shape and the dashed curve represents a Lorentzian

The detection of narrow, Gaussian SiO lines and the fact that none of the discussed kinematical models can satisfactorily explain the observed distribution of the line profiles make it plausible, however, that the observed lines have a substantial contribution from quiescent gas, where SiO production has not occured by means of high-velocity outflows. In the chemistry model of McKay (1995, 1996) the main reservoir of Si in grain mantles is SiH4, which can evaporate into the gas phase in warm conditions. To form SiO, SiH4 needs first to be deprived of the hydrogen atoms, which can take place through reactions with neutral hydrogen or, with a much higher efficiency, with ionized carbon C+ (McKay 1996).

Enhanced abundances of C+ and other ions required to speed up the formation of SiO can be the result of the X-ray, EUV, and FUV radiation from shocks. According to the model of Wolfire & Königl (1993) high-energy photons from shocks can cause an overabundance of ions and produce electronic heating in ambient dense clumps. In this way they explain the enhanced HCO+ emission ahead the shock fronts observed towards several Herbig-Haro objects. Another possible source of energetic radiation in star-forming regions is the magnetic field activity close to the protostellar surface (Montmerle et al. 1993).

The high SiO detection rate between the galactocentric radii 3 and 7 kpc for a luminosity limited sample can be due to some additional mechanism apart from protostellar shocks which sustain elevated temperatures also in the relatively quiescent dense gas in the "molecular ring''. There are several alternativies. Cloud-cloud collisions and shocking by the spiral density waves are supposed to be particularly effective in the inner Galaxy (e.g. Sakamoto et al. 1997). The heating might be also caused by increased interstellar radiation field or by embedded OB stars (Bronfman 1992 and references therein).

SiO emission may originate from regions with very different characteristics: 1) dense gas heated by the radiation from a bow-shock (Wolfire & Königl 1993) or by some of the "external'' mechanism mentioned above; and 2) post-shock gas filling the cavity behind the bow-shock (Raga & Cabrit 1993). Does SiO then serve our original purpose i.e. the study of shock fronts associated with maser emission? We think so since firstly, with the aid of models like that of Raga & Cabrit (1993) "shocked'' profiles can be used for estimating the jet velocity and thereby the momentum input to the ambient gas. Second, with the aid of narrow SiO emission features detected towards some sources the dense gas ahead of shock fronts can be traced. It is clear, however that observations of higher J-transitions combined with model calculations like those performed by Schilke et al. (1997) are required for determining the physical characteristics of the densest regions providing favourable environment for maser excitation.


next previous
Up: A survey of SiO

Copyright The European Southern Observatory (ESO)