Free Access
Issue
A&A
Volume 531, July 2011
Article Number L16
Number of page(s) 5
Section Letters
DOI https://doi.org/10.1051/0004-6361/201116893
Published online 04 July 2011

© ESO, 2011

1. Introduction

Newly-formed stars inject large amounts of energy into the ambient interstellar material through shocks and radiation, heating the surrounding gas and dust. In these warm regions, H2O becomes one of the most abundant gas-phase molecules because water ice evaporates from the grain mantles and gas-phase formation routes become available. The hydroxyl radical (OH) is closely linked to both the H2O formation and destruction through OH + H2  ⇔  H2O + H as well as a byproduct of the H2O photodissociation process in the presence of UV photons. A significant fraction of the gas cooling occurs through line emission in the far-infrared, including lines of  [O i] ,  [C ii] , CO, H2O, and OH (Giannini et al. 2001; van Kempen et al. 2010).

Observations of the OH far-infrared transitions first became possible with the Kuiper Airborne Observatory (e.g. Storey et al. 1981; Betz & Boreiko 1989; Melnick et al. 1990, and references therein). OH was frequently detected in star-forming regions with ISO (e.g. Giannini et al. 2001; Goicoechea et al. 2006) and with PACS (Wampfler et al. 2010) on board the Herschel Space Observatory (Pilbratt et al. 2010). Previously, OH was mainly observed through maser lines at cm wavelengths (Gaume & Mutel 1987, for models see e.g. Cesaroni & Walmsley 1991). Observations carried out with ISO and PACS lack the spectral resolution needed to distinguish between OH emission from the quiescent envelope and the outflow, since no information on the line shape is obtained. Furthermore, the hyperfine structure of the OH lines remains unresolved. The Heterodyne Instrument for the Far-Infrared (HIFI, de Graauw et al. 2010) on Herschel offers an adequate spectral resolution to resolve the 1837.8 GHz triplet of OH. Thus, optical depths and line widths to constrain the temperature and density of the emitting gas can be derived. A blended OH triplet was previously detected with HIFI from Orion KL (Crockett et al. 2010).

This Letter presents the first hyperfine velocity resolved observations of OH obtained using HIFI from the well-studied high-mass star-forming region W3 IRS 5 located at a distance of 2 kpc (Hachisuka et al. 2006) with a total luminosity of  ~105   L (e.g. Helmich & van Dishoeck 1997; Boonman et al. 2003; van der Tak et al. 2005; Chavarría et al. 2010; Benz et al. 2010).

2. Observations and data reduction

The OH triplet at 1837.747, 1837.817, and 1837.837 GHz from W3 IRS 5 (, ) was observed with HIFI on Herschel. The observations were carried out as part of the “Water In Star-Forming Regions with Herschel” (WISH) key program (van Dishoeck et al. 2011). W3 IRS 5 was observed on July 29th 2010 (obsid 1342201666) in dual beam switch mode with an on-source integration time of 17 min. The beam size (HPBW) at 1837 GHz is about 12″. The average system temperature was 1245 K. The wide band spectrometer (WBS) offers a nominal resolution of 1.1 MHz, corresponding to a velocity resolution of about 0.18   km   s-1 at 1837 GHz. The OH triplet at 1834.7 GHz was not observed.

HIFI data were reduced using the Herschel interactive processing environment (HIPE v4.0.0, Ott 2010) and the GILDAS-CLASS1 software. The H and V polarizations were combined, reaching Trms ≈ 0.1   K. We subtracted a first order polynomial from the spectra and calibrated to Tmb scale using a main beam efficiency of 0.70. Molecular data (Table 1) are taken from the Leiden atomic and molecular database LAMDA2 (Schöier et al. 2005) with data from the JPL catalog3 (Pickett et al. 1998) and collisional rate coefficients by Offer et al. (1994).

The calibration uncertainty of band 7b is currently estimated to  ≲ 32% and is mainly caused by the unknown side band gain ratio. The uncertainty reduces to about 10% for a ratio close to unity. The errors on the integrated intensities are only fitting errors and do not include the calibration uncertainty. The velocity calibration is uncertain by  ~0.2   km   s-1 and could partially explain the  ~0.5   km   s-1 offsets of the line peaks from the source velocity of 3lsr =  −38.4   km   s-1 (van der Tak et al. 2000).

3. Results

HIFI clearly detects the OH triplet at around 1837.8 GHz (163.1   μm) towards W3 IRS 5 (Fig. 1). For the first time, the hyperfine components of the line triplet are spectrally resolved.

Two components dominate the line shape: the line profile shows narrow components residing on top of a broad emission feature, similar to that found by Chavarría et al. (2010) for H2O lines in W3 IRS 5 and Kristensen et al. (2010) in low-mass YSOs. The narrow components with full widths at half maximum (FWHM) of  ~4−5   km   s-1 are centered close to the positions expected from the hyperfine pattern. The lines are split by 20.2   MHz (3.3   km   s-1) towards the blue and 70.2   MHz (11.5   km   s-1) towards the red relative to the middle line. Therefore, the (F = 2 +  → 1 − ; Table 1) and (F = 1 +  → 0 −) transitions are blended, while the (F = 1 +  → 1 − ) transition is resolved. The underlying broad component with a FWHM of  ~30   km   s-1 consists of the three blended hyperfine components, each with Δ3 ~ 20 km s-1. The hyperfine pattern is unresolved in the broad component.

Gaussian fits to the narrow components are presented in the Appendix (Sect. A). The derived hyperfine intensity ratios (1:3  ±  0.5:1.5  ±  0.3) deviate from the prediction of LTE in the optically thin limit (~Aulgu, 1:5:2). Pumping by far-infrared dust continuum radiation, line overlap, and optical depth may all contribute to the observed hyperfine anomaly. Modeling results (Sect. 3.1) indicate that at least the 2 +  → 1 −  line is optically thick. Based on comparison with spherical envelope models (Sect. 3.2), we attribute the narrow components to OH emission from the hot parts of the envelope.

The broad component in OH is most likely associated with outflow emission. Its integrated intensity is around 37   K   km   s-1 (Sect. A). A similar component can also be identified in the line profiles of other species like CO and H2O (Sect. B) from recent Herschel results and ground-based observations (e.g. Hasegawa et al. 1994; Boonman et al. 2003; Chavarría et al. 2010).

thumbnail Fig. 1

HIFI spectrum of the OH triplet at 1837.8 GHz with half the dual-sideband continuum. The expected positions of the lines and the source velocity are labeled. The blue lines show the best fit from the slab models and the outflow component separately.

Open with DEXTER

Table 1

Molecular data of the observed OH transitions between the 3/2 and 1/2 excited states.

3.1. Slab models

As a first modeling step to derive OH column densities, the slab method outlined in Bruderer et al. (2010, Appendix B) is used. This method calculates the molecular spectrum from two slabs – one representing the envelope and one the outflow – in front of a continuum source, determined from the observed continuum flux. The slabs are assumed to cover the entire continuum source. The free parameters describing each slab are the OH column density NOH, the excitation temperature Tex (assumed to be equal for all hyperfine transitions), the line width Δ3, and the position of the line center 3lsr. Overlap between different hyperfine components of OH is taken into account. To constrain the free parameters (NOH, Tex, Δ3, and 3lsr), the χ2 between observation and model is minimized using an uncertainty of Trms ~ 0.1   K.

The resolved hyperfine triplet structure of the narrow (envelope) component allows us to constrain the column density and the excitation temperature of the emitting gas simultaneously. Results of the best fitting slab models for the narrow component with a reduced χ2 = 1.2 are presented in Table 2. A peak line opacity of τ = 1.8 is reached (2 +  → 1 −  transition) and the 1 +  → 0 −  and 1 +  → 1 −  peaks have optical depths of τ = 0.8 and τ = 0.4, respectively. Figure C.1 of the Appendix shows the 1σ and 2σ contours for the column density and excitation temperature. The OH column densities within the 1σ interval range from NOH = 2  ×  1015 − 1  ×  1017 cm-2, depending on Tex, and thus vary by almost two orders of magnitude.

A broad (outflow) component can be clearly identified in the spectrum, but its hyperfine components are not resolved. Thus, it is not possible to determine an excitation temperature of the broad emission from the spectrum. Assuming Tex = 100 K, a column density of NOH = 2.0  ×  1014 cm-2 is obtained by fitting the line wings only. A lower limit on the OH column density of NOH ≥ 7.2  ×  1013 cm-2 is derived by assuming that the excitation temperature is equal to the upper level energy (270 K) of the transition. Both fits yield a line width of Δ3 = 21.9 km s-1 and a line position 3lsr =  −35.7   km   s-1. The line optical depth is τ < 0.01 for these excitation temperatures. An optical depth of τ = 1 is reached for Tex < 34 K and no good fit can be obtained for Tex ≲ 32 K. In the following, Tex = 270   K is assumed.

Table 2

Beam-averaged column density NOH, excitation temperature Tex, line width Δ3, and line position 3lsr of the best fit slab model.

To convert the column density into an OH abundance in the outflow, the CO(3 − 2) observation of Hasegawa et al. (1994) is used to derive NH2 = 1.3  ×  1021   cm-2 with RADEX (van der Tak et al. 2007), assuming a CO/H2 abundance ratio of 10-4, a density of 105   cm-3, and a temperature of 60   K, as derived in their paper. This H2 column density converts the lower limit of NOH ≥ 7.2  ×  1013 cm-2 into a lower limit on the OH abundance of xOH = NOH/NH2 ≥ 5.5  ×  10-8.

3.2. Spherical model of the envelope

In addition to the slab modeling, the narrow emission component is compared to the results from full radiative transfer models using the “RATRAN” code (Hogerheijde & van der Tak 2000). The physical structure of W3 IRS 5 is taken from van der Tak et al. (2000) with a power-law density profile ranging from n = 105 − 108   cm-3 and temperatures of 50 ≲ T ≲ 950   K within the Herschel beam at 1837   GHz. Dust and gas temperatures are assumed to be equal. Far-infrared excitation by the dust continuum for Tdust is included. We also adopt their distance (2.2 kpc) for consistency. Dust opacities based on grains with thin ice mantles are assumed (Ossenkopf & Henning 1994,Table 1, Col. 5). The molecular abundance is assumed to be constant throughout the envelope, because this is the simplest structure yielding a good fit to the line profiles. Figure 2 shows the comparison between model and data, where the best fitting slab model for the broad component (cf. Sect. 3.1) is subtracted from the observation, leaving only the narrow component for a simpler comparison.

thumbnail Fig. 2

Resulting HIFI spectrum after subtraction of the best fit outflow component. Overplotted are the spherical envelope models for constant OH abundances of 1  ×  10-8 (blue dashed), 8  ×  10-9 (red solid with individual components red dotted) and 5  ×  10-9 (pink dashed). Hyperfine components were simply added (see text).

Open with DEXTER

Models assuming an OH abundance of xOH = (0.5−1)  ×  10-8 agree reasonably well with the observed narrow components. RATRAN does not treat line overlap and overlap effects can therefore not be treated accurately for optically thick lines. Thus, we have simply added the intensities of the components as would be appropriate in the optically thin limit.

The spherical envelope models are also consistent with the 79,84, and 119   μm lines being in absorption, as observed in the unpublished spectral scan obtained with PACS on Herschel. The 79 and 119   μm transitions are connected to the ground  level and therefore prone to absorption. A quantitative analysis is not possible because the central spatial pixel of the detector is saturated.

4. Discussion

Insight into the water chemistry can be gained by comparing OH and H2O abundances, because these species are linked through the OH + H2  ⇔  H2O + H reactions. Chavarría et al. (2010) identify broad and narrow components in H2O line profiles from W3 IRS 5 similar to OH. OH/H2O abundance ratios can be derived separately for gas of the envelope and the outflow.

From the spherical non-LTE envelope models, which do not consider overlap of hyperfine components, we find an OH abundance of xOH ≈ 8  ×  10-9 for W3 IRS 5. New calculations of the H2O abundance xH2O in the envelope, based on the work by Chavarría et al. (2010) but recalculated using the same physical structure as for OH, yield xH2O = 10-5 for T ≥ 100   K and xH2O = 10-8 for T < 100   K. This allows us to estimate the OH/H2O abundance ratios: H2O is about three orders of magnitude more abundant in the inner envelope (T ≥ 100   K), while the OH/H2O ratio is around unity in the outer part (T < 100   K). The same conclusion is reached in the best fit H2O abundance structure of Boonman et al. (2003) to ISO and SWAS data, which was derived with the same physical model as adopted here.

The inferred OH/H2O ratios in the envelope are consistent with the current picture of the water chemistry in dense clouds (for detailed discussion of processes see Hollenbach et al. 2009; van Dishoeck et al. 2011). Different paths can increase the gas-phase H2O abundance in the inner part of protostellar envelopes. At T ≥ 100   K, H2O starts to evaporate from the ice mantles of dust grains. When temperatures of T ≳ 230   K are reached, which are typical for the innermost parts of high-mass protostellar envelopes, H2O can be rapidly formed in the gas phase through OH + H2  ⇒  H2O + H. In this case, OH is only a transient step of the H2O formation process and thus OH/H2O  ≪  1 in the absence of UV photons. In the outer envelope, low-temperature gas-phase chemistry provides xOH ≈ 10-8 (Doty et al. 2002, Fig. 11). At the outer edge, H2O and OH can be released into the gas phase by photodesorption from grain mantles. Öberg et al. (2009) and Andersson & van Dishoeck (2008) find from laboratory data and theoretical calculations that roughly equal amounts of OH and H2O are released at low temperatures. Depending on the optical depth of the lines, this outer layer may dominate the emission.

The slab model method outlined in Sect. 3.1 is also used to estimate the H2O abundance in the outflow component. Availability of HIFI para-H2O 20,2 − 11,1 (from Chavarría et al. 2010) and unpublished 21,1 − 20,2 (Chavarria et al., in prep.) and 33,1 − 40,4 data allows us to constrain Np − H2O and the para-H2O excitation temperature in the outflow simultaneously. We find Np − H2O = 6.4  ×  1014 cm-2 and thus NH2O = 2.6  ×  1015 cm-2 for an assumed ortho- to para-H2O ratio of 3:1. The water abundance is calculated to be 2.1  ×  10-6 with NH2 = 1.25  ×  1021   cm-2. Under the assumption that the OH and H2O emission arise from the same gas, we can calculate a lower limit of 0.028 on the OH/H2O abundance ratio in the outflow of W3 IRS 5. This limit is consistent with the emission originating in either a fast (3 > 60   km   s-1), dissociative J-type shock (Neufeld & Dalgarno 1989) or a slower UV-irradiated C-type shock (3  ~  20−30 km s-1, Wardle 1999). Standard C-type shocks underproduce the observed column density ratio by one order of magnitude or more (Kristensen et al., in prep.). It is not possible to distinguish between the two types of shocks with the current data and availability of model results.

5. Conclusions

The OH lines at 1837.8   GHz from W3 IRS 5 detected with HIFI consist of two components: a broad component, where the triplet components are blended because of the large individual line widths, and a narrow component with velocity resolved hyperfine structure on top. These results indicate that OH emission observed at a lower spectral resolution, e.g. with PACS, can be a blend of envelope and outflow contributions. Low-mass sources reach lower temperatures than the high-mass counterparts at the projected distance of the Herschel beam. From RATRAN models of low-mass sources we thus find a lower excitation, resulting in a weaker envelope contribution to the OH 1837   GHz lines than from W3 IRS 5, consistent with the results for HH 46 in Wampfler et al. (2010). Because the broad component also appears in the line profiles of other species from W3 IRS 5, in particular those of CO and H2O, the outflow is the most likely origin of the emission. Comparison with H2O yields a lower limit on the OH/H2O abundance ratio of 0.028, consistent with an origin from a fast J-type shock or a slower UV-irradiated C-type shock. The narrow emission agrees well with spherical envelope models with a constant OH abundance of xOH ≈ 8  ×  10-9. Comparison with new H2O results based on Chavarría et al. (2010) give OH/H2O  ≈  1 for T < 100   K and OH/H2O  ≈  10-3 at T ≳ 100   K,

consistent with the current picture of the dense cloud chemistry with freeze-out.


Acknowledgments

The authors are grateful to an U. A. Yıldız for maintaining the WISH database and to C. Dedes, T. Giannini, G. Melnick, and R. Visser for useful discussions. We thank the anonymous referee for constructive comments. The work on star formation at ETH Zurich is partially funded by the Swiss National Science Foundation (grant No. 200020–113556). This program is made possible thanks to the Swiss HIFI guaranteed time program. HIFI has been designed and built by a consortium of institutes and university departments from across Europe, Canada and the United States under the leadership of SRON Netherlands Institute for Space Research, Groningen, The Netherlands and with major contributions from Germany, France and the US. Consortium members are: Canada: CSA, U. Waterloo; France: CESR, LAB, LERMA, IRAM; Germany: KOSMA, MPIfR, MPS; Ireland, NUI Maynooth; Italy: ASI, IFSI-INAF, Osservatorio Astrofisico di Arcetri-INAF; Netherlands: SRON, TUD; Poland: CAMK, CBK; Spain: Observatorio Astronomico Nacional (IGN), Centro de Astrobiología (CSIC-INTA). Sweden: Chalmers University of Technology – MC2, RSS & GARD; Onsala Space Observatory; Swedish National Space Board, Stockholm University – Stockholm Observatory; Switzerland: ETH Zurich, FHNW; USA: Caltech, JPL, NHSC.

References

Online material

Appendix A: Gaussian fitting

Table A.1

Gaussian fit results to the OH line components using the velocity range [− 100, 30] and first order baselines.

Appendix B: Line width comparison

The comparison of OH and H2O column densities in the outflow is based on the assumption that the emission arises from the same gas. Figure B.1 illustrates the similar widths of the broad components of OH (Δ3 ≈ 27   km   s-1), CO (Δ3 ≈ 29   km   s-1), and H2O (Δ3 ≈ 26   km   s-1). The broad component of OH is a blend of three hyperfine components with individual widths of Δ3 ≈ 22   km   s-1, derived from the best fit slab model (Sect. 3.1).

thumbnail Fig. B.1

Comparison of the broad component of OH (black), H2O 202 − 111 (red), and CO(10 − 9) (blue) observed with HIFI. The black dashed line indicates the source velocity (3lsr =  −38.4   km   s-1).

Open with DEXTER

Appendix C: Slab modeling details

The modeling of the OH line spectrum (cf. Sect. 3.1) is carried out with the slab model code presented in Appendix B of Bruderer et al. (2010). Both OH line components (narrow/envelope and broad/outflow) are represented by a slab in front of a continuum source. The continuum temperature Tcont is derived from the observed single-sideband continuum value and the source is assumed to be fully covered by both slabs. No geometry is included, except that the outflow slab is placed

in front of the envelope slab. Each slab has four free parameters per line: the OH column density NOH, the excitation temperature Tex, the line width Δ3, and the position (in velocity space) of the line center 3lsr, but the excitation temperature is assumed to be the same for all hyperfine transitions. The normalized level populations of the upper (xu) and lower level (xl) are therefore determined by the Boltzmann distribution at Tex, (C.1)with the statistical weights gu and gl of the upper and lower level, respectively, Boltzmann’s constant kB, Planck’s constant h, and the frequency ν0 of the transition. The radiation temperature  is derived from the solution of the radiative transfer equation, using the Rayleigh-Jeans approximation, as (C.2)with c being the speed of light, Bν0 the Planck function, τenv the optical depth of the envelope layer, and τofl the optical depth of the outflow layer. The line optical depth of every slab is the sum of the contributions of all M hyperfine components and can be calculated from (C.3)where Aul is the Einstein coefficient and the normalized line profile function of the transition, assumed to be a Gaussian of width Δ3 centered at 3lsr. This approach takes line overlap into account.

thumbnail Fig. C.1

1σ and 2σ contours for the fit of the narrow (envelope) component with slab models. The best fit is indicated by the red dot.

Open with DEXTER

All Tables

Table 1

Molecular data of the observed OH transitions between the 3/2 and 1/2 excited states.

Table 2

Beam-averaged column density NOH, excitation temperature Tex, line width Δ3, and line position 3lsr of the best fit slab model.

Table A.1

Gaussian fit results to the OH line components using the velocity range [− 100, 30] and first order baselines.

All Figures

thumbnail Fig. 1

HIFI spectrum of the OH triplet at 1837.8 GHz with half the dual-sideband continuum. The expected positions of the lines and the source velocity are labeled. The blue lines show the best fit from the slab models and the outflow component separately.

Open with DEXTER
In the text
thumbnail Fig. 2

Resulting HIFI spectrum after subtraction of the best fit outflow component. Overplotted are the spherical envelope models for constant OH abundances of 1  ×  10-8 (blue dashed), 8  ×  10-9 (red solid with individual components red dotted) and 5  ×  10-9 (pink dashed). Hyperfine components were simply added (see text).

Open with DEXTER
In the text
thumbnail Fig. B.1

Comparison of the broad component of OH (black), H2O 202 − 111 (red), and CO(10 − 9) (blue) observed with HIFI. The black dashed line indicates the source velocity (3lsr =  −38.4   km   s-1).

Open with DEXTER
In the text
thumbnail Fig. C.1

1σ and 2σ contours for the fit of the narrow (envelope) component with slab models. The best fit is indicated by the red dot.

Open with DEXTER
In the text

Current usage metrics show cumulative count of Article Views (full-text article views including HTML views, PDF and ePub downloads, according to the available data) and Abstracts Views on Vision4Press platform.

Data correspond to usage on the plateform after 2015. The current usage metrics is available 48-96 hours after online publication and is updated daily on week days.

Initial download of the metrics may take a while.