A&A 376, L5-L8 (2001)
DOI: 10.1051/0004-6361:20011008

Gas-phase SO $_\mathsf{2}$ in absorption towards massive protostars[*]

J. V. Keane1 - A. M. S. Boonman2 - A. G. G. M. Tielens1,3 - E. F. van Dishoeck2

1 - Kapteyn Institute, PO Box 800, 9700 AV Groningen, The Netherlands
2 - Leiden Observatory, PO Box 9513, 2300 RA Leiden, The Netherlands
3 - SRON, PO Box 800, 9700 AV Groningen, The Netherlands

Received 18 June 2001 / Accepted 11 July 2001

We present the first detection of the ${\nu_{\rm 3}}$ ro-vibrational band of gas-phase ${\rm SO_2}$ in absorption in the mid-infrared spectral region around 7.3 ${\rm ~\mu m}$of a sample of deeply embedded massive protostars. Comparison with model spectra shows that the derived excitation temperatures correlate with previous C2H2 and HCN studies, indicating that the same warm gas component is probed. The ${\rm SO_2}$ column densities are similar along all lines of sight suggesting that the ${\rm SO_2}$ formation has saturated, but not destroyed, and the absolute abundances of ${\rm SO_2}$ are high ($\sim$10-7). Both the high temperature and the high abundance of the detected ${\rm SO_2}$ are not easily explained by standard hot core chemistry models. Likewise, indicators of shock induced chemistry are lacking.

Key words: star-formation: gas-phase molecules - ISM: abundances - ISM: molecules

1 Introduction

Observations with the Infrared Space Observatory (ISO) have dramatically increased our knowledge of the active chemistry occurring within star-forming regions. Extensive studies have revealed a vast richness of solid-state molecules embedded in icy grain mantles (Ehrenfreund & Schutte 2000) which highlight the crucial role of grain surface chemistry in molecule formation. Paralleling this, the infrared and submillimeter (Boonman et al. 2000; Lahuis & van Dishoeck 2000; van der Tak et al. 2000a) observations of gas-phase molecules directly probe the chemical and physical conditions of the star-formation process. By combining the solid-state and gas-phase observations, a detailed picture of the evolving chemistry emerges which can serve as a stringent test of proposed chemical models of star-forming regions.

Sulphur-bearing species are particularly interesting to study as they were originally proposed as tracers of shocks since the increased availability of OH radicals will lead to enhanced abundances of specific molecular species (Hartquist et al. 1980). The chemistry of sulphur-bearing molecules in warm gas is essentially governed by neutral-neutral reactions involving ${\rm H_2S}$ formed on grain surfaces and subsequently evaporated into the gas-phase (cf. Charnley 1997). The destruction of ${\rm H_2S}$ frees atomic sulphur, which can then readily react with OH and O2 to produce SO. ${\rm SO_2}$ is easily formed through the conversion of SO by OH. Above $\sim$200-300 K, the OH radicals are driven into H2O and the formation of ${\rm SO_2}$ is halted. Except for the detection of the ${\nu_{\rm 3}}$ band of gas-phase ${\rm SO_2}$ in emission towards Orion (van Dishoeck et al. 1998), only purely rotational lines of ${\rm SO_2}$ in the submillimeter have been detected toward massive star-forming regions. Abundances of $\sim$10-9 (Schreyer et al. 1997) are typically derived which are much lower than model predictions (Charnley 1997). ${\rm SO_2}$ abundances up to $\rm 10^{-7}$ are only found in Orion-KL in the so-called plateau gas associated with the low-velocity outflow (e.g. Blake et al. 1987; Sutton et al. 1995). This gas is known to contain high abundances of OH (Melnick et al. 1987), and hence, the formation of ${\rm SO_2}$ is intimately connected with the availability of reactive OH.

Here we present the first detection of infrared gas-phase ${\rm SO_2}$ in absorption in the ${\nu_{\rm 3}}$ ro-vibrational band towards a sample of embedded massive protostars.

\par\includegraphics[angle=-270,width=13.6cm,height=8.1cm,clip]{DF185F1.ps} \end{figure} Figure 1: ISO-SWS AOT 6 spectra towards six massive protostars. The thin solid lines indicate the locally defined 4th order polynomials adopted as the continua. Some of the sources were offset for clarity by a constant factor indicated in the brackets.
Open with DEXTER

2 Observations and reduction

High resolution AOT6 ( ${\lambda/\Delta\lambda~\rm\ge 1600}$) grating mode observations of the massive protostars presented in this article were made with the Short Wavelength Spectrometer (SWS) on-board ISO. The ${\nu_{\rm 3}}$ ro-vibrational mode of   ${\rm SO_2}$ lies within Band 2C which suffers from instrumental fringing of varying severity. In order to extract unimpeded ${\rm SO_2}$ ${\nu_{\rm 3}}$ absorption profiles, the fringes were corrected for by dividing the observed fluxes by cosine functions fitted to the data in wavenumber space (Lahuis & van Dishoeck 2000). The data were flat-fielded to the average level and then rebinned to the wavelength grid with a constant binsize of 0.003 ${\rm ~\mu m}$which corresponds to ${\lambda/\Delta\lambda~\rm\sim 2500.}$ The fully reduced 7-8 ${\rm ~\mu m}$spectra are shown in Fig. 1, where the noise level is approximately 4-5% for 3$\sigma$ significance.

3 Absorption features

The 7-8 ${\rm ~\mu m}$spectra (Fig. 1), towards all lines of sight, display a richness of broad and narrow absorption features attributable to solid-state and gas-phase molecular species. The region is dominated by the red wing of the 6.85 ${\rm ~\mu m}$feature and the blue wing of the 10 ${\rm ~\mu m}$silicate feature (Keane et al. 2001). The feature near 7.6 ${\rm ~\mu m}$is well fitted by gas-phase and/or solid CH4 (Boogert et al. 1997; Dartois et al. 1998). The spectra in Fig. 1 show evidence for weak features between 7.2 ${\rm ~\mu m}$and 7.4 $\mu$m. The spectrum of MonR2:IRS3 is particularly revealing in that it shows a narrow absorption feature at 7.342 ${\rm ~\mu m}$flanked by broader red- and blue-shifted bands. This structure is reminiscent of the P, Q, and R branch structure of gaseous molecules. The other sources show similar structure albeit less pronounced due to the presence of gas-phase H2O absorption lines (Boonman et al. 2000). A weak broad feature has been seen toward W33A centered at 7.25 ${\rm ~\mu m}$and has been attributed to solid HCOOH (Keane et al. in prep.). However, this feature is easily distinguished from the spectral structure observed here as it is shifted to the blue and cannot explain the observed Q- and P-branch structure. We attribute the spectral structure between 7.2 ${\rm ~\mu m}$and 7.4 ${\rm ~\mu m}$to gas-phase ${\rm SO_2}$.

\par\includegraphics[angle=0,width=8.1cm,height=8.8cm,clip]{DF185F2.ps} \end{figure} Figure 2: Synthetic gas-phase ${\rm SO_2}$ spectra calculated at various temperatures and column densities for a spectral resolution $R \sim 2000$. For all panels the Doppler parameter is 3 kms-1. The dashed line shows the shift in the R-branch as a function of increasing excitation temperature.
Open with DEXTER

\par\includegraphics[angle=-270,width=13cm,height=8.2cm,clip]{DF185F3.ps} \par\end{figure} Figure 3: The continuum divided spectra upon which the best fitting models (grey) are superimposed. Also shown are the H2O model spectra (offset) used for the modeling except in the case of NGC7538:IRS1. The position of the gas-phase CH4 band is indicated for the sources where it was included in the modeling along with the SO2 Q, R, and P branches (thick solid lines).
Open with DEXTER

Gas-phase SO $_\mathsf{2}$

The modeling of the spectra has been performed using synthetic spectra from Helmich (1996) combined with the molecular line data from the HITRAN 2000 database (http://www.hitran.com). Following the same analysis as in Lahuis & van Dishoeck (2000) and Boonman et al. (2000) a homogeneous source has been assumed with a single temperature $T_{\rm {ex}}$ and column density N. Since the SO2 models are not sensitive to the linewidth, a Doppler b parameter of 3 kms-1 is adopted here, corresponding to the mean value of the submillimeter SO2 lines. Figure 2 illustrates the expected spectral structure of the ${\nu_{\rm 3}}$ ro-vibrational band of gas-phase ${\rm SO_2}$ for different column densities and excitation temperatures. A global comparison with the observations shows that the observed features imply typically a column of a few times 1016 cm-2 of warm ( $\mathrel{\mathchoice {\vcenter{\offinterlineskip\halign{\hfil
$\displaystyle ...200 K) ${\rm SO_2}$.

Using these models, we have made detailed fits to the observed absorption features. Half of the sources show the presence of strong gas-phase H2O absorption in the $\nu_2$ ro-vibrational band extending well into the 7.2-7.5 ${\rm ~\mu m}$region. Therefore the H2O model fits of Boonman et al. (2000) have been included in the modeling of gas-phase SO2. In the sources GL2591, NGC7538:IRS1, and GL2136 gas-phase CH4 is also present and this has been included in the models, although it affects the SO2 band only moderately ( $\mathrel{\mathchoice {\vcenter{\offinterlineskip\halign{\hfil
$\displaystyle ...2%). The best fitting models have been determined using the reduced $\chi_{\nu}^2$-method and are shown in Fig. 2. The corresponding excitation temperatures and column densities are listed in Table 1.


Table 1: ${\rm SO_2}$ excitation temperatures and column densities in comparison with other gas-phase molecules.
Source $T_{\rm {ex}}$(K) N(1016 cm-2) $\frac{\rm {\it N}(SO_{2})}{\rm {\it N}(H_{2})^{{\it c}}}$
  ${\rm SO_2}$ H $_{2}{\rm O}\,^a$ CO $_{\rm hot}^b$ HCNc ${\rm C_2 H_2}$c ${\rm SO_2}$ H $_{2}{\rm O}\,^a$ CO $_{\rm total}^b$ CO $_{\rm hot}^b$ HCNc ${\rm C_2 H_2}$c 10-7
MonR2:IRS3 225 ${\rm ^{+50}_{-70}}$ 300 310d -- -- 4 ${\rm\pm}$ 0.8 60 980d 444d -- -- 8.2
W3:IRS5 450 ${\rm ^{+100}_{-100}}$ 400 577 400 500 5 ${\rm\pm}$ 0.8 40 2580 1260 0.5 0.3 3.8
GL2136 350 ${\rm ^{+100}_{-50}}$ 500 580 600 800 6 ${\rm\pm}$ 0.8 150 2200 1500 3.5 1.5 5.5
GL2591 750 ${\rm ^{+70}_{-100}}$ 450 1010 600 900 6 ${\rm\pm}$ 0.4 350 1280 558 4 2 6.3
GL4176 350 ${\rm ^{+175}_{-75}}$ 400 $\mathrel{\mathchoice {\vcenter{\offinterlineskip\halign{\hfil
$\displaystyle ...500c 500 700 4 ${\rm\pm}$ 1.0 150 1600c 800c 2 1 5.0
NGC7538:IRS1 700 ${\rm ^{+300}_{-400}}$ 176b 176 600 800 4 ${\rm\pm}$ 1.0 <20 1740 840 1 0.8 4.7
a Boonman et al. (2000) unless otherwise noted; b Mitchell et al. (1990) unless otherwise noted, using 13CO and assuming
12CO/13CO = 60; c Lahuis & van Dishoeck (2000); d Giannakopoulou et al. (1997).

5 Discussion

Molecular abundances can serve as a direct means of probing the chemical history of star-formation. The derived ${\rm SO_2}$ excitation temperatures range from 200-700 K and are in good agreement with those inferred for HCN and ${\rm C_2 H_2}$ (Table 1), which are good tracers of warm gas (Lahuis & van Dishoeck 2000). The ${\rm SO_2}$ column densities show little variation from source to source with typical abundances of 4- $8 \times 10^{-7}$ relative to the total H2. The infrared ${\rm SO_2}$ abundances are roughly consistent with the ${\rm SO_2}$ abundances of $\sim$10-7 observed in the submillimeter towards the Plateau, the Compact Ridge, and the Hot Core in Orion (Sutton et al. 1995). The relative constancy of the Orion ${\rm SO_2}$ abundances is striking given the physical differences that exist between the afore mentioned regions in Orion. On the other hand, the derived ${\rm SO_2}$ abundances in hot cores are much higher than ${\rm SO_2}$ abundances in dark clouds (Irvine et al. 1983). More recently, Hatchell et al. (1998) have found gas-phase ${\rm SO_2}$ abundances of $5 \times 10^{-10}$ to $2 \times 10^{-8}$ in hot core regions, which are a factor of $\ge$10 less than what is derived here. Some of this difference may well reflect the beam dilution suffered by the submillimeter observations. Thus, given the Orion template, there are two possible origins for the high abundance of gaseous ${\rm SO_2}$:-hot core chemistry or shock induced chemistry.

In hot core chemistry, the ${\rm SO_2}$ originates from oxidation of sulphur bearing species by OH (Charnley et al. 1997). This limits the ${\rm SO_2}$ to gas with temperatures in the range $\sim$100 to $\sim$200 K. The observed ${\rm SO_2}$ temperature is well above this, though it is possible that this high temperature reflects radiative pumping by the dust. Moreover, HCN and C2H2 appear in gas with similar temperatures to that of the ${\rm SO_2}$. This is difficult to reconcile since the formation of ${\rm SO_2}$ becomes very inefficient for temperature above $\sim$230 K, whereas the route to HCN is greatly enhanced (Charnley 1997; Boonman et al. 2001; Rodgers & Charnley 2001). Thus, ${\rm SO_2}$cannot be abundant in gas which has become enriched in HCN through the removal of OH. The ${\rm SO_2}$ and HCN abundances must therefore peak at different radii from the protostar in order for standard hot core models to be compatible. Alternatively, the ${\rm SO_2}$ may be formed on grains surfaces and then be released into the gas by evaporation. Grain surface chemistry models predict an abundance of $\sim$ $3 \times 10^{-3}$ relative to H2O on the ice (Tielens & Hagen 1982; Tielens private communication), which is a factor $\mathrel{\mathchoice {\vcenter{\offinterlineskip\halign{\hfil
$\displaystyle ...10 less than what is derived here. Another aspect is that the ${\rm SO_2}$ column density does not vary between the sources, whereas H2O shows large variation. The constancy of the ${\rm SO_2}$ column density can be explained by the fact that the ${\rm SO_2}$ is only abundant in a narrow zone between $\sim$90 K (the ice evaporation temperature) and $\sim$230-300 K (the OH $\to$ H2O transition), whose mass does not vary much in spite of the different total masses of the sources (Doty et al., in prep.).

The presence of SO and ${\rm SO_2}$ has often been quoted as evidence for shocks (Hartquist et al. 1980). The degree to which the ${\rm SO_2}$ abundance is enhanced depends on whether most of the sulphur is initially in atomic form ($\sim$10-7, Pineau des Forêts et al. 1993) or locked up in stable molecules ($\sim$10-8, Leen & Graff 1988). A good aspect of the shock induced chemistry hypothesis is that the ${\rm SO_2}$ and HCN may be colocated in gas which contains freshly (i.e., $\mathrel{\mathchoice {\vcenter{\offinterlineskip\halign{\hfil
$\displaystyle ...3. $4 \times 10^4$ yr) released grain mantle molecules. However, the lack of increased ${\rm SO_2}$ line widths at submillimeter wavelengths would seem to indicate the decay of shock activity within the region sampled by the submillimeter observations. In addition, the presence of fragile molecules sensitive to destruction by shocks (e.g. H2CO; van der Tak et al. 2000b) makes shock-induced chemistry less likely as the source of ${\rm SO_2}$ for these sources.

In general, the gaseous ${\rm SO_2}$/ ${\rm H_2S}$ ratio serves as a sensitive chemical clock for the star formation process and searches for these molecules at high spectral resolution are needed to help resolve the issue of the origin of the gas-phase ${\rm SO_2}$.

The authors are grateful to F. Helmich for setting up the ${\rm SO_2}$ synthetic spectra and to D. Kester for insightful discussions on fringe removal from ISO data. This works was supported by the Netherlands Organization for Scientific Research (NWO) through grant 614-041-003.


Copyright ESO 2001