Free Access
Issue
A&A
Volume 647, March 2021
Article Number L7
Number of page(s) 8
Section Letters to the Editor
DOI https://doi.org/10.1051/0004-6361/202140517
Published online 16 March 2021

© ESO 2021

1. Introduction

Ultraviolet (UV) line absorption studies show that the gas-phase abundance of certain elements (Si, Mg, Fe, Ca, Ti, etc.) are highly depleted in diffuse (AV ≲ 1) and translucent (AV ≳ 1) low-density clouds of the interstellar medium (ISM)1, with D(X) ranging from −1 to −3 dex (Savage & Sembach 1996; Sofia 2004; Jenkins 2009). This is consistent with their incorporation into (mainly) silicate grains (Mathis 1990; Draine 2003). The inferred depletion factors of much more volatile elements, such as carbon, are lower but still significant, D(C) ≳ −0.5 dex.

Sulfur is on the top ten list of the most abundant elements, with [S] = (1.4 ± 0.1)⋅10−5, but it is an unusual element in the sense that almost all of the sulfur in diffuse clouds remains in the gas phase (Fitzpatrick & Spitzer 1994; Howk et al. 2006). However, the relevant UV S II lines saturate, and thus the determination of [S] in neutral atomic H I and translucent molecular clouds may be uncertain (Federman et al. 1993; Sofia 2004).

Sulfur plays an important role in stellar nucleosynthesis; it is mainly produced in massive stars (e.g., Perdigon et al. 2021), in star formation, and in astrochemistry (e.g., Fuente et al. 2017; Shingledecker et al. 2020). Inside star-forming clouds, an undetermined fraction of sulfur gradually converts into molecules and ice mantles (e.g., Goicoechea et al. 2021), and yet the initial [S] in these dense molecular clouds is poorly constrained.

Radio recombination lines (RRLs) provide a powerful tool for studying star-forming regions independently of dust obscuration. Hydrogen and He RRLs are extensively used to constrain the morphology and physical conditions (Te and ne) of fully ionized H II regions (e.g., Churchwell et al. 1978). However, only stellar far-UV (FUV) photons, with energies below 13.6 eV, permeate the rims of molecular clouds, so-called photodissociation regions (PDRs; Hollenbach & Tielens 1999). In the first layers of a PDR, the dominant state of elements with ionization potential (IP) below H is singly ionized: C+ (11.3 eV), S+ (10.4 eV), Si+ (8.2 eV), Fe+ (7.9 eV), or Mg+ (7.6 eV). Indeed, the narrower carbon RRLs probe these neutral PDRs adjacent to H II regions (Ball et al. 1970; Natta et al. 1994; Wyrowski et al. 1997; Salas et al. 2019; Cuadrado et al. 2019). Carbon RRLs have historically been detected in the centimeter (cm) domain: ∼8.6 GHz for the C91α line (where α stands for Δn = 1 transitions). Compared to the fine-structure 2P3/22P1/2 line of C+, the very important far-infrared [C II] 158 μm cooling line (Hollenbach & Tielens 1999), carbon RRLs are optically thin and their intensity is proportional to . That is, they have different excitation properties than the [C II] 158 μm line (e.g., Natta et al. 1994; Salas et al. 2019).

The 4S ground-electronic state of sulfur ions does not have low-lying fine-structure splittings. Hence, S+ in neutral gas cannot be detected2 by far-infrared observations. Instead, sulfur RRLs, although much less studied, can be detected, typically after long-time integrations, and probe the “S+ layer” of a PDR3. Early observations with cm-wave single-dish radio telescopes, at several-arcminute resolution, allowed the detection of emission features at the high-frequency shoulder of carbon RRLs. These features were associated with sulfur RRLs but can correspond to a superposition of RRLs from low-IP elements heavier than carbon (Chaisson et al. 1972; Chaisson & Lada 1974; Pankonin & Walmsley 1978; Falgarone et al. 1978; Silverglate 1984; Vallee 1989; Smirnov et al. 1995).

In this Letter we present the first detection of sulfur RRLs in the Orion Bar (e.g., Tielens et al. 1993; Wyrowski et al. 1997; Walmsley et al. 2000; Goicoechea et al. 2016). This prototypical dense PDR (see Fig. 1) is a nearly edge-on interface of the Huygens H II region that is photoionized by the Trapezium massive stars θ1 Ori (O’Dell 2001) and the Orion Molecular Cloud (OMC) – the closest region to host ongoing massive-star formation (e.g., Genzel & Stutzki 1989; Pabst et al. 2020).

thumbnail Fig. 1.

Multiple FUV-illuminated edges in OMC-1 (dense PDRs). The colored map shows the integrated emission (I) of the [13C II] (2P3/22P1/2F = 2–1) line at 1900.466 GHz (Goicoechea et al. 2015). This emission traces the C+ layers of the molecular cloud. Cyan contours show a map of the SH+ 10 − 01F = 1/2–3/2 emission at 345.944 GHz (Goicoechea et al. 2021), a proxy of the S+ layer (both maps at θHPBW ≃ 20″). Black contours show the O I 1.317 μm fluorescent emission that delineates the IF of the Bar (Walmsley et al. 2000). Dashed blue circles show the smallest and largest θHPBW of our Yebes 40 m observations toward the DF position: 36″ and 54″, respectively.

Open with DEXTER

2. Observations

We used the Yebes 40 m radio telescope at Yebes Observatory (Guadalajara, Spain) to observe the Bar. The observed position encompasses the dissociation front (DF), hereafter the DF position, at α2000 = 05h 35m 20.8s, δ2000 = − 05° 25′17.0″. We observed the complete Q band, from 31.1 GHz to 50.4 GHz, using the new high-electron-mobility transistor (HEMT) Q band receiver and fast Fourier transform spectrometers. These cover 18 GHz of instantaneous bandwidth per polarization at a spectral resolution of 38 kHz (∼0.3 km s−1 at ∼40 GHz). We employed two frequency setups of slightly different central frequencies to identify any spurious line. The main beam efficiency varies from 0.6 at 32 GHz to 0.43 at 48 GHz, and the half power beam width (HPBW) at these frequencies ranges from θFHWM∼ 54″ to 36″ (Tercero et al. 2021). We used the position switching observing mode, with a reference position at an offset (−600″, 0″). The total integration time was ∼8 h under winter conditions (∼8 mm of precipitable water vapor). We reduced and analyzed the data using the CLASS software of GILDAS. The achieved root mean square (rms) noise ranges from ∼3 mK to ∼10 mK per velocity channel in antenna temperature units (). Because of the relatively compact angular size of the C+- and S+-emitting layers (compared to the secondary beam lobes), we converted the scale to main beam temperature, ηF/ηMB, where ηF is the forward efficiency and ηMB the main beam efficiency. We adopted the values tabulated in Tercero et al. (2021): for example, ηF ≃ 0.92 and ηMB ≃ 0.52 at ∼40 GHz. We identified nine Cnα RRLs (Table B.1) that show a fainter feature at slightly higher frequency, which we assigned to Snα RRLs. Figure B.1 shows the individual spectra.

3. The C+ and S+ layers in the Bar PDR

Figure 1 shows the environment around the targeted position in the Bar. The colored map shows the 2P3/22P1/2 fine-structure line emission (F = 2–1 hyperfine component) from the 13C+ isotope, hereafter the [13C II] 158 μm line. While the [12C II] 158 μm line is moderately optically thick in dense PDRs (e.g., Ossenkopf et al. 2013), [13C II] lines are optically thin and trace the distribution of dense PDRs at the FUV-illuminated rims of OMC-1 (Goicoechea et al. 2015). The “C+ layer” (where most of the gas-phase carbon is in the form of C+) extends from the ionization front (IF; the H+–H transition at the H II region–PDR interface) to the PDR layers, where most of the gas-phase carbon is converted first into C atoms and then into CO, at a few magnitudes of visual extinction into the cloud (AV). The distance between the IF and AV(CO), perpendicular to the Bar and in the UV-illumination direction, is ∼15″ (∼0.03 pc; Tielens et al. 1993; Goicoechea et al. 2016). Interestingly, the distribution of the C91α emission, mapped with the Very Large Array (VLA; Wyrowski et al. 1997), is remarkably similar to the ν = 1–0 S(1) H2 emission that traces the H–H2 transition at the DF.

Although S has a lower IP than C, the extent of the S+ layer depends on the wavelength- and AV-dependent attenuation of the FUV field (Goicoechea & Le Bourlot 2007) as well as on details of the S photoionization and S+ radiative recombination (RR) and resonant dielectronic recombination (DR) processes. Paradoxically, a modern treatment of S+ recombination, especially at low Te, is still lacking (Badnell 1991; Bryans et al. 2009). Our observations encompass the C+ and S+ layers of the Bar. Indeed, Fig. 1 shows that the observed DF position coincides with the position of the SH+ emission peak (cyan contours). This trace molecular ion forms via gas reactions of S+ with vibrationally excited H2 molecules (e.g., Goicoechea et al. 2021).

4. Results: Detection of Snα RRLs

We detected new Cnα and Henα RRLs (from n = 51 to 59; Fig. B.1). Helium RRLs arise from the adjacent Huygens H II region (IPHe = 24.6 eV) and from the different foreground layers of ionized gas in Orion’s Veil (O’Dell 2001). The high electron temperature and pressure in the hot and fully ionized gas produce the broad line width (Δv ≃ 20 km s−1) of He RRLs. Consistent with ionized gas that flows toward the observer, He RRLs are blueshifted by ∼15 km s−1 with respect to OMC emission.

A visual inspection of each individual Cnα spectrum shows the presence of a narrow emission feature at a slightly higher frequency. Unless produced by a molecular line4, these features imply the presence of an RRL from an element heavier than 12C (12 amu). To determine their origin, we calculated the nα RRL frequencies of 32S (31.972 amu) using the Rydberg equation and assuming that the recombined electron in level n moves on the field of a nucleus of effective charge Zeff ≃ 1 = 16 (p+)−15 (inner e). This means that RRLs of (many-electron) neutral atoms possess frequencies close to hydrogen RRLs but, as the mass of the nucleus increases, shifted to slightly higher frequencies (Table B.1). Likewise, we computed the RRL frequencies of other neutral atoms with low IP (< 13.6 eV) and with solar abundances [X] > 10−5: 24Mg (23.985 amu), 28Si (27.977 amu), and 56Fe (55.935 amu).

In order to increase the signal-to-noise ratio (S/N) of the detection, and because RRLs of similar n and Δn have similar emission properties, we stacked all Cnα spectra from n = 54 to 59 (those with the highest S/N). We first converted each spectrum to the local standard of rest (LSR) velocity scale, then resampled them to the same velocity channel resolution, and finally stacked them. The resulting spectrum in Fig. 2 displays three line features that can be fitted with three Gaussians reasonably well: a broad line from Henα RRLs and two narrow lines, one from Cnα RRLs and another from the heavier element. According to the atomic mass differences between carbon and S, Mg, Si, and Fe, the velocity separation of their nα RRLs from Cnα lines is: δv(12C−24Mg) = −6.9 km s−1, δv(12C−28Si) = −7.8 km s−1, δv(12C−32S) = −8.6 km s−1, and δv(12C−56Fe) = −10.8 km s−1. The observed (peak-to-peak) velocity separation between the Cnα and the higher-frequency feature in the Bar is δv = −8.4 ± 0.3 km s−1. Therefore, the third feature is largely produced by Snα RRLs. This conclusion is supported by the severe depletions of Mg, Si, and Fe inferred in diffuse neutral clouds (e.g., Jenkins 2009). The typical depletion1 factors in these clouds are D(Mg) ≃ −1.3, D(Si) ≃ −1.4, and D(Fe) ≃ −2.2 dex (Savage & Sembach 1996). This implies [S] > [Mg] ≃ [Si] > [Fe] (in the gas phase). It also means that in interstellar PDRs, RRLs of these more condensable elements are much fainter than those of S. From our observations we can only conclude that Mg, Si, and Fe are depleted by more than −0.6 dex relative to S.

thumbnail Fig. 2.

Stacked spectrum (filled gray histogram) of several nα RRLs (n = 54–59). The spectrum is centered at the frequencies of Snα lines and shows the expected position of Mg, Si, and Fe nα RRLs. The black curve is a three-Gaussian line fit: Snα (blue), Cnα (red), and Henα (green). Vertical dashed lines mark the LSR velocity of the Bar PDR.

Open with DEXTER

The stacked Snα line shows a similar Gaussian profile, line width (Δv ≃ 2.6 ± 0.2 km s−1; mostly microturbulent broadening), and velocity centroid (vLSR ≃ 10.8 ± 0.1 km s−1) as those of C54α, [13C II], and SH+ lines – all observed toward the DF position at θHPBW ≃ 45″ (see Fig. C.1). The narrow line profiles (see Table C.1) demonstrate that Snα RRLs arise from PDR gas and not from the fully ionized H II region.

5. Discussion

To guide our interpretation, Fig. 3 shows the [C+], [C], [CO], [S+], [S], [H2S-ice], and [e] abundance profiles predicted by a homogeneous PDR model5 run with the Meudon code (Le Petit et al. 2006). We adopted the gas-phase carbon abundance, [C]Ori = (1.4 ± 0.6)⋅10−4, measured from UV absorption-line observations of the translucent neutral gas toward the Trapezium star θ1 Ori B (Sofia et al. 2004). This abundance implies a depletion factor1 of D(C) ≃ −0.3 dex (carbon that is incorporated into carbonaceous grains, polycyclic aromatic hydrocarbons, and macromolecules such as fullerenes; e.g., Berné & Tielens 2012). We assumed that this is precisely the initial gas-phase carbon abundance in the Bar (i.e., [C+]IF = [C]Ori). Figure 3 shows that the C+ abundance remains constant in the C+ layer, from the IF (AV = 0) to AV(C I) ≃ 3. The S+ layer (where most of the sulfur is in the form of S+) extends a bit deeper, up to AV ≃ 4. However, one would need observations at high angular resolution (≲5″) to spatially resolve the different sizes of the C+ and S+ layers; they are small at the moderate gas densities of the Bar. In these intermediate PDR layers, AV ≃ 3.5 to 5.5, photoionization of S atoms becomes the major source of e, and thus of the ionization fraction. This is one of the reasons why determining [S] in dense gas is relevant.

thumbnail Fig. 3.

Model5 of the C+ layer with [C]/[S] = 10. Lower panel: abundance profiles as a function of AV along the UV-illumination direction: from the IF to the cloud interior (lower axis) and the equivalent angular distance perpendicular to the Bar (upper axis). Upper panel: normalized emissivities of the [C II] 158 μm, S54α, and C54α lines. For the RRLs, we adopted an LTE model (see Appendix A).

Open with DEXTER

Because the HPBW of the 40 m telescope encompasses the C+ and S+ layers of the Bar, and assuming that the C and S level populations of fairly high principal quantum numbers n depart from local thermodynamic equilibrium (LTE) in a similar fashion (e.g., Dupree 1974; Salgado et al. 2017), we can determine the gas-phase carbon-to-sulfur abundance ratio, [C]/[S] = nC+/ nS+, directly from the Cnα/Snα = Rnα line intensity ratio. From the stacked spectrum we obtain Rnα = 10.4 ± 0.6. This is a factor of two lower than the [C]/[S] ratio. As we adopted [C+]IF = [C]Ori, the Rnα ratio leads to [S+]IF = [S]Ori = (1.4 ± 0.4)⋅10−5.

Under the above assumptions, the initial [S] in the OMC matches the solar abundance. Interestingly, it also coincides with the stellar [S] abundances measured in young B-type stars of the Ori OB1 association (Daflon et al. 2009). This result implies very little depletion of S in dense clouds, D(S) = 0.0 ± 0.2. In the unlikely case that the [C+]IF in the Bar is significantly lower than the value measured by Sofia et al. (2004), our estimated [S]Ori abundance would be an upper limit and D(S) would decrease accordingly. Still, an LTE excitation model with N(S+) ≃ 7⋅1017 cm−2 (Appendix A) is consistent with the absolute intensities of the observed sulfur RRLs, with [S]Ori = 1.4⋅10−5, and with N(C+)/N(S+) ≃ 10.

The ratio R(54 − 59)α observed in the Bar is higher than that determined at several-arcminute resolution in the pioneering cm-wave observations of the ρ Oph dark cloud (R158α ≃ 3.5 and R110α ≃ 6.2; Pankonin & Walmsley 1978) and the W3A and Orion B clouds (R158α ≃ 3.1 and 7.1, respectively; Chaisson et al. 1972). These variations can be produced by (i) the different beam filling factors of the C+- and S+-emitting regions in these clouds, such that their RRL emission is not spatially resolved and the large beam mixes various PDRs, (ii) very different excitation properties of Snα versus Cnα RRLs, or (iii) intrinsically lower [C]/[S] ratios. We currently favor possibility (i). Indeed, our higher angular resolution observations nearly spatially resolve the Bar PDR. If this were not the case, and if the S+ layer were significantly more extended than the C+ layer, then the intrinsic R(54 − 59)α ratios would be larger. In the future, more sensitive observations and more details on the RR and DR of S+ ions will allow us to model the non-LTE excitation of sulfur RRLs individually, as well as to constrain ne and Te as a function of AV (e.g., RRLs do not arise from the same [C II] 158 μm-emitting layer; see Fig. 3). In the meantime, ours is probably the most direct estimation of [S] in a dense cloud.

Since stellar sulfur abundances increase with increasing metallicity (e.g., Perdigon et al. 2021), and the estimated [S] in stars and H II regions decreases with galactocentric radius (RGC) as −0.04 dex kpc−1 (e.g., Rudolph et al. 2006; Daflon et al. 2009; Arellano-Córdova et al. 2020), we anticipate shallow variations of [S] in molecular clouds of different RGC. A related open question (e.g., Sofia 2004) is how to reconcile the low sulfur depletion in the ISM with the existence of interplanetary dust particles containing sulfur (Bradley 1994) and with the existence of solid MgS in evolved stars. The latter seems only viable if MgS is present in the outer coating surfaces of circumstellar grains and not in their cores (Lombaert et al. 2012; Sloan et al. 2014). These outer surfaces may be more easily destroyed in the harsh ISM. Observations of Snα RRLs from a larger sample of star-forming regions will improve our understanding of interstellar sulfur.


1

We refer to the gas-phase abundance “[X]” with respect to H nuclei as the gas column density ratio [X] = N(X)/NH, where NH = N(H) + 2N(H2) in molecular clouds. We denote the “solar” or “bulk” abundance of element X in the Sun (i.e., the current photospheric abundances corrected for diffusion) as [X], with [C] = 2.9⋅10−4, [Fe] = 3.5⋅10−5, [Mg] = 4.4⋅10−5, and [Si] = 3.6⋅10−5 (Asplund et al. 2009). We took these [X] as the cosmic abundances and used the logarithmic depletion factor defined as D(X) = log [X] − log [X].

2

In H II regions and IFs, S+ can be detected through the infrared 2P2D lines at ∼1.03 μm (Walmsley et al. 2000) and in the visible through the 2D4S lines at 6718, 6733 Å (Pellegrini et al. 2009). The estimation of [S] from these lines is sensitive to Te gradients, extinction, and ionization corrections (Rudolph et al. 2006). The ground 4S and excited 2D electronic states are separated by ∼21 400 K. In PDRs and cold molecular clouds, S+ largely exists in the 4S state.

3

Carbon and sulfur RRLs were identified in the early 1970s. These detections implied the presence of what at that time was called “C II and S II regions” in star-forming clouds. The term “PDR” started to be used in the late 1980s to model these FUV-illuminated regions.

4

None of the Snα line frequencies correspond to molecules present in the Bar (Cuadrado et al. 2015, 2017) or in molecular-line catalogs.

5

We adopted a uniform gas density nH = n(H) + 2n(H2) = 5⋅104 cm−3, an FUV radiation field of χFUV = 104 times the Draine’s field, and RV = AV/EB − V = 5.5, appropriate for Orion. This is sometimes called the “interclump” medium of the Bar (e.g., Young Owl et al. 2000). This model is valid for the outer C+ layer and reproduces the observed IF–AV(CO) angular separation. The chemical network includes updated gas-phase and grain-surface reactions (Goicoechea et al. 2021).

Acknowledgments

We thank J. H. Black for reminding us that our spectra of the Orion Bar may contain sulfur RRLs and for useful remarks on how S+ dielectronic capture may proceed. We thank B. Tercero and the Yebes staff for their help with these observations. We thank the Spanish MCIYU for funding support under grants PID2019-106110GB-I00 and AYA2016-75066-C2-1-P.

References

  1. Arellano-Córdova, K. Z., Esteban, C., García-Rojas, J., & Méndez-Delgado, J. E. 2020, MNRAS, 496, 1051 [Google Scholar]
  2. Asplund, M., Grevesse, N., Sauval, A. J., & Scott, P. 2009, ARA&A, 47, 481 [Google Scholar]
  3. Badnell, N. R. 1991, ApJ, 379, 356 [Google Scholar]
  4. Ball, J. A., Cesarsky, D., Dupree, A. K., Goldberg, L., & Lilley, A. E. 1970, ApJ, 162, L25 [Google Scholar]
  5. Berné, O., & Tielens, A. G. G. M. 2012, Proc. Nat. Acad. Sci., 109, 401 [Google Scholar]
  6. Bradley, J. P. 1994, Science, 265, 925 [Google Scholar]
  7. Brocklehurst, M., & Seaton, M. J. 1971, Astrophys. Lett., 9, 139 [Google Scholar]
  8. Bryans, P., Kreckel, H., Roueff, E., Wakelam, V., & Savin, D. W. 2009, ApJ, 694, 286 [Google Scholar]
  9. Chaisson, E. J., & Lada, C. J. 1974, ApJ, 189, 227 [Google Scholar]
  10. Chaisson, E. J., Black, J. H., Dupree, A. K., & Cesarsky, D. A. 1972, ApJ, 173, L131 [Google Scholar]
  11. Churchwell, E., Smith, L. F., Mathis, J., Mezger, P. G., & Huchtmeier, W. 1978, A&A, 70, 719 [NASA ADS] [Google Scholar]
  12. Cuadrado, S., Goicoechea, J. R., Pilleri, P., et al. 2015, A&A, 575, A82 [NASA ADS] [CrossRef] [EDP Sciences] [Google Scholar]
  13. Cuadrado, S., Goicoechea, J. R., Cernicharo, J., et al. 2017, A&A, 603, A124 [NASA ADS] [CrossRef] [EDP Sciences] [Google Scholar]
  14. Cuadrado, S., Salas, P., Goicoechea, J. R., et al. 2019, A&A, 625, L3 [CrossRef] [EDP Sciences] [Google Scholar]
  15. Daflon, S., Cunha, K., de la Reza, R., Holtzman, J., & Chiappini, C. 2009, AJ, 138, 1577 [Google Scholar]
  16. Draine, B. T. 2003, ARA&A, 41, 241 [Google Scholar]
  17. Dupree, A. K. 1974, ApJ, 187, 25 [Google Scholar]
  18. Falgarone, E., Cesarsky, D. A., Encrenaz, P. J., & Lucas, R. 1978, A&A, 65, L13 [NASA ADS] [Google Scholar]
  19. Federman, S. R., Sheffer, Y., Lambert, D. L., & Gilliland, R. L. 1993, ApJ, 413, L51 [Google Scholar]
  20. Fitzpatrick, E. L., & Spitzer, L. J. 1994, ApJ, 427, 232 [Google Scholar]
  21. Fuente, A., Goicoechea, J. R., Pety, J., et al. 2017, ApJ, 851, L49 [Google Scholar]
  22. Genzel, R., & Stutzki, J. 1989, ARA&A, 27, 41 [Google Scholar]
  23. Goicoechea, J. R., & Le Bourlot, J. 2007, A&A, 467, 1 [NASA ADS] [CrossRef] [EDP Sciences] [Google Scholar]
  24. Goicoechea, J. R., Teyssier, D., Etxaluze, M., et al. 2015, ApJ, 812, 75 [Google Scholar]
  25. Goicoechea, J. R., Pety, J., Cuadrado, S., et al. 2016, Nature, 537, 207 [Google Scholar]
  26. Goicoechea, J. R., Aguado, A., Cuadrado, S., et al. 2021, A&A, 647, A10 [EDP Sciences] [Google Scholar]
  27. Hollenbach, D. J., & Tielens, A. G. G. M. 1999, Rev. Mod. Phys., 71, 173 [Google Scholar]
  28. Howk, J. C., Sembach, K. R., & Savage, B. D. 2006, ApJ, 637, 333 [Google Scholar]
  29. Jenkins, E. B. 2009, ApJ, 700, 1299 [Google Scholar]
  30. Le Petit, F., Nehmé, C., Le Bourlot, J., & Roueff, E. 2006, ApJS, 164, 506 [Google Scholar]
  31. Lombaert, R., de Vries, B. L., de Koter, A., et al. 2012, A&A, 544, L18 [NASA ADS] [CrossRef] [EDP Sciences] [Google Scholar]
  32. Mathis, J. S. 1990, ARA&A, 28, 37 [Google Scholar]
  33. Natta, A., Walmsley, C. M., & Tielens, A. G. G. M. 1994, ApJ, 428, 209 [Google Scholar]
  34. O’Dell, C. R. 2001, ARA&A, 39, 99 [Google Scholar]
  35. Ossenkopf, V., Röllig, M., Neufeld, D. A., et al. 2013, A&A, 550, A57 [NASA ADS] [CrossRef] [EDP Sciences] [Google Scholar]
  36. Pabst, C. H. M., Goicoechea, J. R., Teyssier, D., et al. 2020, A&A, 639, A2 [CrossRef] [EDP Sciences] [Google Scholar]
  37. Pankonin, V., & Walmsley, C. M. 1978, A&A, 64, 333 [NASA ADS] [Google Scholar]
  38. Pellegrini, E. W., Baldwin, J. A., Ferland, G. J., Shaw, G., & Heathcote, S. 2009, ApJ, 693, 285 [Google Scholar]
  39. Perdigon, J., de Laverny, P., Recio-Blanco, A., et al. 2021, A&A, in press, https://doi.org/10.1051/0004-6361/202040147 [Google Scholar]
  40. Rudolph, A. L., Fich, M., Bell, G. R., et al. 2006, ApJS, 162, 346 [Google Scholar]
  41. Salas, P., Oonk, J. B. R., Emig, K. L., et al. 2019, A&A, 626, A70 [CrossRef] [EDP Sciences] [Google Scholar]
  42. Salgado, F., Morabito, L. K., Oonk, J. B. R., et al. 2017, ApJ, 837, 141 [Google Scholar]
  43. Savage, B. D., & Sembach, K. R. 1996, ARA&A, 34, 279 [Google Scholar]
  44. Shingledecker, C. N., Lamberts, T., Laas, J. C., et al. 2020, ApJ, 888, 52 [Google Scholar]
  45. Silverglate, P. R. 1984, ApJ, 278, 604 [Google Scholar]
  46. Sloan, G. C., Lagadec, E., Zijlstra, A. A., et al. 2014, ApJ, 791, 28 [Google Scholar]
  47. Smirnov, G. T., Sorochenko, R. L., & Walmsley, C. M. 1995, A&A, 300, 923 [NASA ADS] [Google Scholar]
  48. Sofia, U. J. 2004, in Astrophysics of Dust, eds. A. N. Witt, G. C. Clayton, & B. T. Draine, ASP Conf. Ser., 309, 393 [Google Scholar]
  49. Sofia, U. J., Lauroesch, J. T., Meyer, D. M., & Cartledge, S. I. B. 2004, ApJ, 605, 272 [Google Scholar]
  50. Tercero, F., López-Pérez, J. A., Gallego, J. D., et al. 2021, A&A, 645, A37 [EDP Sciences] [Google Scholar]
  51. Tielens, A. G. G. M., Meixner, M. M., van der Werf, P. P., et al. 1993, Science, 262, 86 [Google Scholar]
  52. Vallee, J. P. 1989, ApJ, 341, 238 [Google Scholar]
  53. Walmsley, C. M., Natta, A., Oliva, E., & Testi, L. 2000, A&A, 364, 301 [NASA ADS] [Google Scholar]
  54. Wyrowski, F., Schilke, P., Hofner, P., & Walmsley, C. M. 1997, ApJ, 487, L171 [Google Scholar]
  55. Young Owl, R. C., Meixner, M. M., Wolfire, M., Tielens, A. G. G. M., & Tauber, J. 2000, ApJ, 540, 886 [Google Scholar]

Appendix A: Absolute intensities of sulfur RRLs

The level populations of the recombined S atom can be written as N(n) = bnN(n)LTE, where n is the principal quantum number, N(n)LTE are the LTE populations given by the Saha-Boltzmann equation, and bn are the so-called departure coefficients. We consider that the S54α RRL emission is dominated by spontaneous emission, with little contribution from stimulated emission. Hence, we neglected amplification of any background continuum emission (Tc). This is a reasonable assumption for these RRLs in Orion (e.g., Natta et al. 1994). Indeed, the 7 mm continuum emission is produced by the cosmic microwave background, the long-wave tail of the dust thermal emission, and the free-free Bremsstrahlung emission. In Orion, the last component largely arises from the Huygens H II region, but this nebular emission is in the foreground relative to the Bar. In addition, owing to the low electron densities in PDRs, we estimate that the free-free opacity is very low, τff(7 mm) ∼ 10−7. Hence, the continuum opacity (τc = τff + τd) is dominated by the opacity of warm (Td ≃ 50 K) dust grains, which is also low, τd(7 mm) ∼ 10−5. The observed sulfur RRLs are optically thin (τRRL of a few 10−4) and are in the Rayleigh-Jeans regime (with hν/k ≪ Te and TcTe). Hence, we can write the S54α integrated line intensity (IS54α in K km s−1) as:

(A.1)

where χn = 15.79⋅104n−2 and EMS+ = nenS+L (in pc cm−6) is the S+ emission measure along a slab of thickness L (along the line-of-sight; L should not be confused with the cloud depth in the UV-illumination direction). The factor ≃ 15 is specific to 54α RRLs and includes the oscillator strength ΔnMn) ≃ 0.1908 for Δn = 1 (Brocklehurst & Seaton 1971; Dupree 1974; Salas et al. 2019). When excitation conditions are close to LTE and as the principal quantum number n increases, the departure coefficients approach bn ≃ 1. Coefficients for hydrogen and carbon RRLs have been estimated in the literature. At the relatively low electron temperatures of a PDR, these coefficients are typically 0.3 < bn < 1, with values up to bn ≃1.5 when the details of resonant dielectronic capture for non-hydrogenic atoms are included (e.g., Salgado et al. 2017). In this study we implicitly assumed that any substantial difference between S+ and C+ recombination will have little effect on the relative intensities of their moderately high-n recombination spectra. Nonetheless, we point out again that the low Te RR and DR of S+ have not been properly studied (Bryans et al. 2009). Here, we simply adopted bn = 1 and Te = Tgas (i.e., LTE) to estimate the RRL emissivities (see the PDR model in Fig. 3).

We then calculated the expected intensity of the S54α RRL using Eq. (A.1) and compared it with the observational value in the Orion Bar PDR, IS54αIC54α/R(54 − 59)α, where R(54 − 59)α is the C(54–59)α/S(54–59)α integrated line intensity ratio of the stacked spectrum (see Table C.1). We obtain IS54α (observations) = 0.407/10.4 = 0.04 K km s−1. Figure A.1 shows a grid of single-slab models appropriate for the two possible sources of RRL emission discussed in the literature of the Bar: the “interclump” medium model – with ne ≃ 10 cm−3 and L ≃ 0.2 pc – and the denser “clump” or “high-pressure” model – with ne ≃ 100 cm−3 and L ≃ 0.025 pc. Assuming a beam filling factor of one, both models reproduce the observed IS54α intensities with a similar column N(S+) ≃ 7⋅1017 cm−2 but different (LTE) electron temperatures: Te(interclump) ≃ 125 K and Te(clump) ≃ 500 K. Cuadrado et al. (2019) recently reported the detection of 3 mm wave carbon RRLs in the Orion Bar: α lines from n = 42 to 38, βn = 2) lines, and γn = 3) lines. Although they estimated, from single-slab models, that these lines arise from PDR layers with ne ≃ 60–100 cm−3 and Te ≃ 500–600 K, the clump versus interclump dichotomy is not yet fully solved (Tielens et al. 1993; Goicoechea et al. 2016). Since PDRs have steep temperature and density gradients, we suspect that only higher angular resolution observations will clarify the exact small spatial-scale origin of these lines.

thumbnail Fig. A.1.

S54α intensity as a function of sulfur abundance and Te. Upper panel: interclump model of the Bar. Lower panel: model that assumes that RRLs arise from denser clumps or high-pressure gas (see text). The horizontal dashed green line marks the observational value.

Open with DEXTER

The modeled column density of S+, beam-averaged over the mean angular resolution of the Yebes 40 m observations (∼45″), is a factor of ten lower than the N(C+) independently inferred from maps of the [13C II] 158 μm line emission toward the same angular area of the Bar: N(C+) ≃ 7⋅1018 cm−2 (see Table 2 of Ossenkopf et al. 2013). Therefore, the observed S(54–59)α line intensities are consistent with both N(C+)/N(S+) ≃ 10 and with [S+] ≃ 1.4⋅10−5. More refined estimations of Te and ne will require a proper non-LTE model, higher angular resolution observations, and higher S/N detections of several RRLs.

Appendix B: Carbon RRL observational parameters and sulfur RRL frequencies

thumbnail Fig. B.1.

Sulfur, carbon, and helium RRLs observed with the Yebes 40 m radio telescope between 31 GHz and 48 GHz (at 38 kHz resolution) toward the Orion Bar. Dashed lines indicate the position of Snα RRLs at vLSR ≃ 10.7 km s−1, the emission of the neutral PDR gas.

Open with DEXTER

Table B.1.

Cnα line spectroscopic parameters obtained from a two-Gaussian line fit to C and He RRLs.

Appendix C: Line parameters of the stacked spectrum and comparison with other lines

thumbnail Fig. C.1.

Similar line profiles, velocity centroids, and line widths of the stacked Snα line (blue), C54α (hatched red), [13C II] 158 μm (F = 2–1; solid gray), and SH+ (hatched cyan) lines toward the DF position (all observed or extracted at θHPBW ≃ 45″ and ≃ 0.3 km s−1 resolutions). The dashed line corresponds to the LSR velocity of the PDR, vLSR = 10.7 km s−1.

Open with DEXTER

Table C.1.

Line spectroscopic parameters obtained from Gaussian fits toward the Orion Bar PDR.

All Tables

Table B.1.

Cnα line spectroscopic parameters obtained from a two-Gaussian line fit to C and He RRLs.

Table C.1.

Line spectroscopic parameters obtained from Gaussian fits toward the Orion Bar PDR.

All Figures

thumbnail Fig. 1.

Multiple FUV-illuminated edges in OMC-1 (dense PDRs). The colored map shows the integrated emission (I) of the [13C II] (2P3/22P1/2F = 2–1) line at 1900.466 GHz (Goicoechea et al. 2015). This emission traces the C+ layers of the molecular cloud. Cyan contours show a map of the SH+ 10 − 01F = 1/2–3/2 emission at 345.944 GHz (Goicoechea et al. 2021), a proxy of the S+ layer (both maps at θHPBW ≃ 20″). Black contours show the O I 1.317 μm fluorescent emission that delineates the IF of the Bar (Walmsley et al. 2000). Dashed blue circles show the smallest and largest θHPBW of our Yebes 40 m observations toward the DF position: 36″ and 54″, respectively.

Open with DEXTER
In the text
thumbnail Fig. 2.

Stacked spectrum (filled gray histogram) of several nα RRLs (n = 54–59). The spectrum is centered at the frequencies of Snα lines and shows the expected position of Mg, Si, and Fe nα RRLs. The black curve is a three-Gaussian line fit: Snα (blue), Cnα (red), and Henα (green). Vertical dashed lines mark the LSR velocity of the Bar PDR.

Open with DEXTER
In the text
thumbnail Fig. 3.

Model5 of the C+ layer with [C]/[S] = 10. Lower panel: abundance profiles as a function of AV along the UV-illumination direction: from the IF to the cloud interior (lower axis) and the equivalent angular distance perpendicular to the Bar (upper axis). Upper panel: normalized emissivities of the [C II] 158 μm, S54α, and C54α lines. For the RRLs, we adopted an LTE model (see Appendix A).

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

S54α intensity as a function of sulfur abundance and Te. Upper panel: interclump model of the Bar. Lower panel: model that assumes that RRLs arise from denser clumps or high-pressure gas (see text). The horizontal dashed green line marks the observational value.

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

Sulfur, carbon, and helium RRLs observed with the Yebes 40 m radio telescope between 31 GHz and 48 GHz (at 38 kHz resolution) toward the Orion Bar. Dashed lines indicate the position of Snα RRLs at vLSR ≃ 10.7 km s−1, the emission of the neutral PDR gas.

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

Similar line profiles, velocity centroids, and line widths of the stacked Snα line (blue), C54α (hatched red), [13C II] 158 μm (F = 2–1; solid gray), and SH+ (hatched cyan) lines toward the DF position (all observed or extracted at θHPBW ≃ 45″ and ≃ 0.3 km s−1 resolutions). The dashed line corresponds to the LSR velocity of the PDR, vLSR = 10.7 km s−1.

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.