© ESO, 2015

1. Introduction

The freezeout and desorption of volatiles are amongst the key factors that determine the gas-phase abundances of chemical species in interstellar gas, in particular in protostellar cores (e.g. Bergin & Langer 1997; Caselli & Ceccarelli 2012). The classical example is CO, which is strongly depleted in prestellar cores, but returns to the gas in the warm protostellar stages Class 0 and I (e.g. Caselli et al. 1999; Bacmann et al. 2002; Jørgensen et al. 2005). Similar behaviour, with quantitative differences, is seen or expected for many other species. Here, we investigate the freezeout and main reservoirs of chlorine in a protostellar core.

An ice mantle on a dust grain in a protostellar core mainly consists of a few dominant ice species (e.g. H₂O, CO, CO₂, CH₃OH). However, the ices also contain a large number of minor species, accreted from the gas and formed in the ice matrix. Such minor species very likely include HF and HCl, which are the main molecular carriers of the halogen elements fluorine and chlorine, in well-shielded molecular gas. With increasing temperature, such as seen during infall towards a central protostar, the ice mantle desorbs. For any given ice species, the desorption may have multiple stages, with details influenced by its abundance and interaction with the other species in the ice (Collings et al. 2004; Lattelais et al. 2011). In the hot cores of protostars, where ice mantles largely desorb, most volatile species are expected to be back in the gas phase.

We study the depletion of volatile chlorine towards the OMC-2 FIR 4 protostellar core, using HCl and H₂Cl⁺ as proxies. We employed the Herschel Space Observatory¹ and the Caltech Submillimeter Observatory (CSO) telescopes and made use of both an updated gas-grain chemical network for chlorine and new calculations of the HCl-H₂ hyperfine collisional excitation rate coefficients. In Sect. 2, we review the interstellar chemistry of chlorine. The source is described in Sect. 3 and the observations in Sect. 4. The analysis separately covers HCl (Sect. 5) and H₂Cl⁺ (Sect. 6), and the results are discussed in Sect. 7. We present our conclusions in Sect. 8.

Fig. 1 Observed spectra of HCl from Herschel/HIFI and CSO, velocity-corrected and centred on the strongest hyperfine component. Red dashed lines indicate the hyperfine components of the rotational transitions, while blue dashed lines show CH3OH lines. The CSO data have been binned to δv = 3.7 km s-1.

2. Interstellar chlorine chemistry

Chlorine has a simple and relatively well characterized interstellar chemistry (e.g. Jura 1974; Dalgarno et al. 1974; Blake et al. 1986; Schilke et al. 1995; Neufeld & Wolfire 2009). In dense molecular gas, the gas-phase formation of HCl begins with the reaction $\mathrm{Cl}\mathrm{+}{\mathrm{H}}_{\mathrm{3}}^{\mathrm{+}}\mathrm{\to }{\mathrm{H}}_{\mathrm{2}}\mathrm{C}{\mathrm{l}}^{\mathrm{+}}\mathrm{+}\mathrm{H}\mathit{.}$(1)Dissociative recombination of H₂Cl⁺ with e⁻ then produces HCl. A recent study of D₂Cl⁺ has confirmed the branching ratio of this recombination to be ~10% into HCl, ~90% into Cl (Novotnỳ et al. 2012). Below ~100 K, the resulting fraction of chlorine in HCl is ~0.1 to 0.3 (Dalgarno et al. 1974; Blake et al. 1986; Schilke et al. 1995; Neufeld & Wolfire 2009). At temperatures of ≳350 K, all chlorine can be converted into HCl in dense (n ≳ 10⁴ cm^-3) molecular gas by the reaction $\mathrm{Cl}\mathrm{+}{\mathrm{H}}_{\mathrm{2}}\mathrm{\to }\mathrm{HCl}\mathrm{+}\mathrm{H}\mathit{,}$(2)on a timescale of ≲10⁴ yr. At colder temperatures, this reaction is much slower.

In photon-dominated regions (PDRs), a layer can form where the reactions Cl⁺ + H₂ → HCl⁺ + H and HCl⁺ + H₂ → H₂Cl⁺ + H lead again via dissociative recombination to HCl. In diffuse PDRs (n_H ~ 10³ cm^-3), the HCl and H₂Cl⁺ column density ratio is ~1, while for 10⁷ cm^-3 it is ~10² (Neufeld & Wolfire 2009). At A_V ≲ 1 however, the ratio can be ~0.01, while atomic Cl is the dominant gas-phase reservoir of chlorine. At low densities and A_V ≪ 1, Cl⁺ dominates.

The abundance of Cl with respect to atomic hydrogen in the solar photosphere is [Cl]/[H] = 3.16 × 10^-7 (Asplund et al. 2009). The meteoritic abundance is 1.8 × 10^-7 (Lodders 2003), while in the diffuse interstellar medium (ISM) it is 10^-7, indicating a factor of two depletion into refractory grains or volatile ices (Moomey et al. 2012). For our modelling, we refer to the abundance with respect to molecular hydrogen, X(Cl) = [Cl]/[H₂], and adopt a reference value X(Cl) = 10^-7.

The main stable isotopes of chlorine are ³⁵Cl and ³⁷Cl. Their ratio in the solar system is 3.1 (Lodders 2003).

3. The FIR 4 protostellar core

Our target is a nearby intermediate-mass protostellar core, OMC-2 FIR 4² (hereafter FIR 4), located in Orion, at a distance of d ≈ 420 pc (Hirota et al. 2007; Menten et al. 2007). Its envelope mass is ~30  M_⊙ and its luminosity within ~20′′ is ~100  L_⊙, while the ~1′ scale envelope has an estimated luminosity of ~ 400–1000 L_⊙ (Mezger et al. 1990; Furlan et al. 2014; Crimier et al. 2009, the latter is hereafter referred to as C09). Class 0 suggests an age of ≲10⁵ yr, and Furlan et al. (2014) propose that FIR 4 is amongst the youngest protostellar cores known. In the far-infrared to millimetre regimes, FIR 4 is undergoing intense study in the Herschel key programmes CHESS (Ceccarelli et al. 2010; Kama et al. 2010, 2013; López-Sepulcre et al. 2013a,b; Ceccarelli et al. 2014) and HOPS (Adams et al. 2012; Manoj et al. 2013; Furlan et al. 2014), as well as in a number of ground-based projects.

The main physical components of FIR 4 are the warm, clumpy inner envelope and the cold, extended outer one; a proposed outflow; and a tenuous, heavily irradiated foreground cloud (Crimier et al. 2009; Kama et al. 2013; López-Sepulcre et al. 2013a,b; Furlan et al. 2014). The inner envelope has been resolved into continuum peaks with different luminosities (Shimajiri et al. 2008; Adams et al. 2012; López-Sepulcre et al. 2013b; Furlan et al. 2014). These sources appear to share a large envelope. The line profiles and excitation of CO and H₂O show that FIR 4 harbours a compact outflow (Kama et al. 2013; Furlan et al. 2014). There are also two PDRs to consider: the dense outermost envelope (n_H2= 6 × 10⁵ cm^-3, Crimier et al. 2009), and the tenuous foreground cloud (n_H2≈ 10² cm^-3, López-Sepulcre et al. 2013a). Whether or not they are physically connected is unclear. For simplicity, we treat them as separate entities, even though there may be a smooth transition in physical conditions between the two.

4. Observations

The main neutral and ionized molecular carriers of chlorine in the ISM – such as HCl, HCl⁺ and H₂Cl⁺ – have strong radiative transitions, although their observations from the ground are hampered by atmospheric water absorption. The observations of HCl and H₂Cl⁺ towards FIR 4, summarized in Table 1 and shown in Figs. 1 (HCl) and 2 (H₂Cl⁺), were carried out with the Herschel Space Observatory and the CSO.

4.1. Herschel/HIFI

As part of the CHESS key programme (Ceccarelli et al. 2010), FIR 4 was observed with the Heterodyne Instrument for the Far-Infrared (HIFI) wide band spectrometer (de Graauw et al. 2010) on the Herschel Space Observatory (Pilbratt et al. 2010), in Dual Beam Switch (DBS) mode at a resolution of dν = 1.1 MHz (R ~ 10⁶). Multiple transitions of HCl and H₂Cl⁺ were covered. The data quality and reduction, carried out with the HIPE 8.0.1 software (Ott 2010), are presented in Kama et al. (2013).

4.2. Caltech submillimeter observatory

Ground-based observations of HCl J = 1–0 were carried out in DBS mode with a chopper throw of 240′′, using the 690 GHz facility heterodyne receiver of the CSO, on Mauna Kea, Hawaii, on February 6th and 11th, 2013. The atmosphere was characterized by a 225 GHz zenith opacity of 0.04–0.06 or 1 mm of precipitable water vapour. Typical single sideband system temperatures were 5000–7000 K. The backend was the high-resolution FFT spectrometer, with 4095 channels over 1 GHz of IF bandwidth. The on-source integration time was 51 min, resulting in an RMS noise of ~0.15 K at a resolution of 3.7 km s^-1. Jupiter was used for pointing and calibration. The beam efficiency was 40%, assuming a 148.5 K brightness temperature for Jupiter.

4.3. Overview of the data

Both HCl and H₂Cl⁺, as well as their isotopologs, are detected with Herschel/HIFI. The HCl 1−0 transition is also detected with the CSO. The measured line fluxes, integrated over the hyperfine components, are summarized in Table 1.

The HCl lines show evidence for a broad and a narrow component. Both components are present in the HIFI HCl 1–0 data, which also shows a roughly optically thin hyperfine component ratio (2:3:1, in order of increasing ν) for the narrow component. The signal to noise of the HCl 2–1 and the CSO 1–0 data are insufficient to make firm conclusions about the relative importance of the broad and narrow components, however the broad component seems to contribute substantially to both observations. Unfortunately, the H³⁷Cl 2–1 transition is contaminated by a CH₃OH transition and is therefore excluded from our analysis.

5. Analysis of HCl

Here, we first disentangle the kinematical components of the hydrogen chloride lines. Then, we use radiative transfer and chemical modelling to determine the gas-phase HCl abundance. The outermost envelope of FIR 4 is strongly externally irradiated, and is considered separately in Sect. 5.4.

Fig. 2 Observed spectra of H2Cl+ from Herschel/HIFI. Only the lines formally detected in absorption are shown, and are velocity-corrected and centred on the strongest hyperfine component. Red dashed lines indicate the hyperfine components of the rotational transitions. The continuum has not been subtracted. See also Fig. B.1.

Fig. 3 Kinematical properties of HCl and H2Cl+ compared to other species in FIR 4, each of which is annotated with the main physical component it traces. The label “PDR” refers here to the tenuous foreground cloud discovered by López-Sepulcre et al. (2013a). For the comparison species, the mean Gaussian fit parameters from Kama et al. (2013) are plotted.

Fig. 4 Line width map of C34S 3–2 (colour map), obtained with the Plateau de Bure Interferometer (López-Sepulcre et al. 2013b). The channel width of the data is 4 km s-1. Overplotted are continuum contours (black lines), which start at 3σ with a 3σ step, with 1σ = 0.14 Jy km s-1beam-1. North is up and east is left.

5.1. Kinematics

A two-component model of the hyperfine structure in the HCl 1–0 observations strongly constrains the kinematic parameters of the broad and narrow components, given in Table 2.

In Fig. 3, we compare the kinematic components of HCl to those of other species from the FIR 4 HIFI spectrum of Kama et al. (2013). The properties of the narrow component match the large-scale quiescent envelope tracers, such as C¹⁸O. The broad component parameters lie between those of the envelope and the outflow tracers, and match the mean properties of CS.

Fig. 5 Top and middle panels: large-scale source structure of FIR 4 (ngas at top, Tgas at middle) from Crimier et al. (2009, C09), corresponding to an integrated luminosity of 1000 L⊙. Bottom panel: relative populations of the lowest HCl rotational states, showing J = 0 (black), J = 1 (blue), and J = 2 (red) for the C09 source structure without modification (solid lines) and with an added external irradiation of G0 = 415 ISRF (dashed lines; Sect. 5.4). The level populations were modelled with Ratran using the new hyperfine collision rate coefficients presented in Appendix A.

Contrary to HCl, for which exceptional observing conditions or space observatories are required, isotopologs of CS are readily observed with ground-based instruments. In Fig. 4, we show a map of the C³⁴S 3–2 line width, based on Plateau de Bure Interferometer data from López-Sepulcre et al. (2013b). A spatially resolved region ~10′′ across (2100 AU radius) and detected at a 3σ confidence at F_C³⁴S ≥ 0.4 Jy km s^-1 beam^-1 has a line width of ≳6 km s^-1. The spatial resolution of the data is ~5′′, the channel width is 4 km s^-1, and the flux loss compared to single-dish data is ~60%, so we can only conclude that if the C³⁴S broad component is equivalent to that of HCl, the latter is centrally peaked and extended on several thousand AU scales.

Based on the arguments presented above, we attribute the narrow and broad HCl line profile component to the outer envelope and some compact yet dynamic inner region, respectively. Given the large difference in line width between the broad HCl component and the outflow tracers CO and H₂O, it is unlikely that they trace the same volumes of gas.

5.2. The gas-phase HCl abundance in FIR 4

To obtain an abundance profile, we modelled the HCl excitation and radiative transfer in FIR 4 using the Monte Carlo code, Ratran³ (Hogerheijde & van der Tak 2000) as described below.

5.2.1. New HCl-H${\mathrm{2}}_{}$ collisional excitation rates

We modelled the HCl emission using new hyperfine-resolved collisional excitation rate coefficients, presented in detail in Appendix A. These are based upon the recent potential energy surface and rotational excitation rate coefficients of Lanza et al. (2014b,a). In Fig. 6, we compare the new rate coefficients to the HCl-He ones from Neufeld & Green (1994) and Lanza & Lique (2012). The latter, scaled to H₂ collisions by a mass correction factor of 1.38, differ from the new coefficients by a factor of a few at T_kin> 50 K, and by around a factor of ten at T_kin< 50 K. A similar difference has been found for HF (Guillon & Stoecklin 2012). We discuss the impact of the new excitation rates on HCl depletion estimates in Sect. 7.5.

5.2.2. The source model

As the basis of our modelling, we adopted the spherically symmetric large-scale source structure with no enhancement of the external irradiation field (G₀ = 1 interstellar radiation fields, ISRF) from C09. The density and temperature profiles, as well as the relative populations of the relevant HCl rotational levels, are shown in Fig. 5. Due to the limited spatial resolution of the continuum maps it is based on, the source model is not well constrained on scales ≲2000 AU. Thus, we interpret it as the spherically-averaged large-scale structure of the source. The total H₂ column density in a pencil beam through the centre of the source model is N(H₂) = 4.6 × 10²³ cm^-2.

According to the relative level populations shown in Fig. 5, the J_u = 1 state is most relevant within ~4000 AU, with a factor of five to ten decrease at larger radii, while the J_u = 2 state is mostly populated in the inner ~2000 AU and plays no role in the outer envelope. If an external irradiation field is added (Sect. 5.4), the gas temperature reaches ~1000 K in a thin outer layer, and the J = 2 and higher level population gains in importance. Thus, the HCl 2–1 line constrains the abundance within 2000 AU and more weakly in a thin outer layer, while the 1–0 transition constrains it in the bulk of the envelope.

Fig. 6 Rate coefficients for the collisional excitation HCl J = 0 → 1 by H2. The previously available HCl-He rate coefficients, scaled by 1.38 to account for the reduced mass difference, are from Neufeld & Green (1994, orange circles) and Lanza & Lique (2012, blue circles). The new HCl-ortho-H2 and -para-H2 rate coefficients (Appendix A) are shown by black squares and diamonds, respectively.

5.2.3. Fitting a constant abundance profile

As a first guess, we assume a constant HCl abundance in the source. Assuming a fixed source structure, the model has only two free parameters: X_c, the gas-phase HCl abundance; and R_bn, the radius where the line width switches from broad to narrow. We performed a χ² minimization on the HCl 1–0 and 2–1 line profiles from HIFI, and the 1–0 from CSO. The best fit parameters, with a reduced χ² = 1.45, are X_c = 9 × 10^-11 and R_bn = 7000 AU. In Fig. 7, we show the data, the best-fit model and the range of models within reduced χ² = 3. The log ₁₀(χ²) surface is shown in Fig. 8.

5.3. A full chemical model

We modelled the chlorine chemistry at each radial location in FIR 4 with the Nautilus gas-grain chemical code. Nautilus time-dependently computes the gas and grain chemistry including freezeout, surface chemistry, and desorption due to thermal and indirect processes. The grain surface reactions are described in Hersant et al. (2009) and Semenov et al. (2010). The binding energy of Cl on H₂O ice is 1100 K, for HCl we adopted 5174 K from Olanrewaju et al. (2011). The gas-phase network is based on kida.uva.2011, from Wakelam et al. (2012), and was updated with data from Neufeld & Wolfire (2009). The full network contains 8335 reactions, 684 species and 13 elements. The H₂ and CO self shielding are computed following Lee et al. (1996), as described in Wakelam et al. (2012). We adopted a volatile chlorine abundance of 10^-7. The chemistry is first evolved to steady state in dark cloud conditions. This yields the initial abundances for the time-dependent chemistry in FIR 4, using the C09 source structure for the density and temperature.

Fig. 7 HCl 1–0 and 2–1 transitions as observed with HIFI and CSO (black). Overplotted are the best-fit constant HCl abundance model (red), and the reduced χ2< 3 range (shaded orange) of the models. The strongest feature in the CSO data, at −20 km s-1, is a CH3OH line.

Fig. 8 log 10(reduced χ2) surface for the two constant HCl abundance model parameters: Xc and Rbn.

Dark cloud stage. The initial conditions for the FIR 4 calculation are computed for dense, cold cloud conditions: a temperature of 10 K, n_H = 2 × 10⁴ cm^-3, A_v = 10 mag, and a cosmic-ray ionization rate of ζ = 10^-17 s^-1. The model is evolved to 10⁶ yr. The species are initially all atomic, except for H₂, with most abundances from Hincelin et al. (2011). The abundance of oxygen is set to 3.3 × 10^-4, and of chlorine to 10^-7. At the end of the dark cloud stage, about 93% of elemental chlorine is in HCl ice, the remaining 7% is almost entirely in gas-phase atomic Cl. The temperature, the density and the age of the cloud influence the fraction of Cl and HCl in the gas versus in the ices, but do not affect the HCl/Cl ratio. This ratio, both in the gas and ices, is mostly influenced by the cosmic ray ionization rate ζ. The resulting abundances are used as the initial conditions for FIR 4.

Protostellar stage. For the protostellar stage, we use the C09 density and temperature structure without enhanced external irradiation. The outermost envelope is treated in more detail in Sect. 5.4. In Fig. 9, we show the gas and ice abundances of HCl and atomic Cl in FIR 4, as modelled with Nautilus at ages 10⁴ and 10⁵ yr. The trial cosmic ray ionization rates were 10^-16 s^-1 (left panel; a foreground cloud value from López-Sepulcre et al. 2013a) and 10^-14 s^-1 (right panel; recently inferred for FIR 4 by Ceccarelli et al. 2014).

As seen from the blue lines in Fig. 9, HCl ice is the main reservoir of elemental chlorine outside of ~5000 AU at both times in both models. It stores 90 to 98% of all elemental chlorine. The second reservoir is gaseous atomic chlorine, at 1 to 10% of the elemental total, typically an order of magnitude above the gas-phase HCl abundance. Inside 5000 AU, HCl ice contains 90% (ζ = 10^-16 s^-1) or 20% (ζ = 10^-14 s^-1) of the chlorine. The HCl ice is built up over the 10⁶ year cold dark cloud stage. The gas-phase HCl abundance slowly decreases with time in most of the source, due to continuing freezeout.

Above, the physical structure switched instantly from the dark cloud structure to the protostellar core. Adding time dependency to the density structure would decrease the abundance changes in the inner envelope, which would be replenished with pristine material from the outer envelope. Additionally, using a higher prestellar core density would lead to even stronger and more rapid freezeout of chlorine into HCl ice, increasing the depletion.

The gas-phase abundance of HCl from Nautilus is consistent with the observational constraints on X(HCl)_gas, as reported in Sect. 5.2, within a factor of a few. This is with the exception of the high-ionization model, where X(HCl)_gas exceeds the observed limit by two orders of magnitude. The gas-phase HCl abundance of ~10^-10 – seen in most of the source in both models – is the tip of the iceberg, as 90% of all chlorine is frozen onto grain surfaces as HCl and likely in a water ice matrix.

Fig. 9 HCl and Cl gas- and ice-phase abundances by 104 (solid) and 105 yr (dashed) after the dark cloud stage, as modelled with Nautilus. The cosmic ray ionization rate is ζ = 10-16 s-1 (left panel) and ζ = 10-14 s-1 (right panel).

5.4. The outermost envelope PDR

The dense and heavily irradiated photon dominated region in the outermost envelope requires a specialized treatment. We employ the Meudon⁴ PDR code (Le Petit et al. 2006) for this. The physical structure is a slab with a constant density of n_H = 6 × 10⁵ – the outermost density in the C09 source model – extending to A_V = 20 mag. We used two cosmic ray ionization rates, as before: ζ = 10^-16 s^-1 and 10^-14 s^-1.

The external irradiation of FIR 4 has previously been inferred to be G₀ ~ 500 interstellar radiation field units, based on [CII] 158 μm emission (Herrmann et al. 1997). We re-evaluated G₀ using the Herschel/HIFI [CII] line flux from Kama et al. (2013), which was obtained at higher spatial resolution and yields G₀ ~ 300. We further checked G₀ based on an archival Spitzer 8 μm map, which has a substantial contribution from PAH emission, and this directly counts excitation events by ultraviolet photons. The measured flux density of ~250 MJy/sr gives G₀ = 415, using Eq. (1) of Vicente et al. (2013) with a standard [ C ]/[ H ] ~ 10^-4 and 4% of elemental carbon locked in PAHs (Tielens 2005). All estimates are consistent with each other, and with irradiation by the Trapezium cluster at a projected separation of ~2 pc. We adopt the middle ground: G₀ = 415.

Fig. 10 Meudon models of the chlorine chemistry in the two PDR layers seen towards FIR 4, showing abundances relative to ngas. Left panel: dense PDR in the outermost envelope of FIR 4, with ngas= 6 × 105 cm-3, G0 = 415 ISRF. The cosmic ray ionization rate is ζ = 10-16 s-1 (solid lines) and ζ = 10-14 s-1 (dotted). The position is given outside-in from the right. Right panel: tenuous PDR in the foreground of FIR 4, with ngas= 102 cm-3, G0 = 1500 ISRF and ζ = 3 × 10-16 s-1. The position is given from the strongly irradiated side on the right.

The resulting abundance profiles of gas-phase HCl, Cl, Cl⁺ and H₂Cl⁺ are shown in the left-hand panel of Fig. 10. For both ζ values, HCl peaks at ~10^-8 at the H/H₂ transition, then decreases deeper into the PDR. By A_V ≈ 1, X(HCl)_gas drops below 10^-10. The HCl abundance in these models is different from the Nautilus results because of different initial conditions and because the Meudon code solves for steady state, which the time-dependent Nautilus models do not reach by 10⁵ yr. The rapid changes around 0.1 to 0.2 AU are due to numerical issues related to H₂ formation and the heating-cooling balance. The effect on HCl is limited to ≲40% of the column density, and does not substantially affect our conclusions. The temperature in this region is between ~100 and ~1000 K, and the total column density in the outer PDR is N(HCl) ≈ 1 × 10¹² cm^-2.

Using the X(HCl) profile shown in Fig. 10 for the outermost envelope, we re-fitted the HCl abundance model from Sect. 5.2. The two X(HCl) profiles were joined at T_kin= 25 K. The impact of the PDR on the HCl emission is small. The best-fit combined model – X_c = 7 × 10^-11, R_bn = 9000 AU – has a reduced χ² = 1.55. The model matches a tentative narrow peak in the HCl 2–1 line, which the constant abundance model does not, although this does not improve the global χ².

In the outer envelope PDR, elemental chlorine is not depleted from the gas in the outermost 300 AU. The PDR model is consistent with the observed HCl emission, and still requires most of the narrow and broad HCl flux to originate deeper in FIR 4. It is also consistent with the upper limit on H₂Cl⁺ absorption at 11.4 km s^-1. The H₂Cl⁺ data is analysed next, in Sect. 6.

6. Analysis of H₂Cl⁺

The H₂Cl⁺ lines, shown in Fig. B.1, appear in absorption against the weak continuum of the source. As shown in Fig. 3, they are blueshifted by 2 km s^-1 with respect to the systemic velocity of FIR 4, which is 11.4 km s^-1. Thus, the H₂Cl⁺ absorption does not originate in the dense outer envelope PDR. Instead, the radial velocity matches that of the tenuous foreground PDR layer recently identified towards FIR 4 by López-Sepulcre et al. (2013a). The distance and relation between the foreground layer and FIR 4 are poorly constrained.

We used the CASSIS⁵ software to perform Markov Chain Monte Carlo fitting of local thermodynamic equilibrium models to the H₂Cl⁺ and H$\begin{array}{c}\mathrm{37}\\ \mathrm{2}\end{array}$Cl⁺ lines. Not all the lines are detected, but are still useful as constraints. To minimize baseline issues, we added the continuum fit of Kama et al. (2013) to baseline subtracted spectra. We find a column density of N(H₂Cl⁺) = 1.3 × 10¹⁴ and an isotopic ratio of [³⁵Cl]/[³⁷Cl] = (4.3 ± 0.8). The main fitting results are given in Table 3.

To compare the observed H₂Cl⁺ column density with that expected from the chlorine chemical network, we again used the Meudon code (Le Petit et al. 2006). The foreground PDR was determined by López-Sepulcre et al. (2013a) to have a density of n = 10² cm^-3, an FUV irradiation of ~1500 G₀ and a cosmic ray ionization rate of 3 × 10^-16 s^-1. The resulting abundance profiles of Cl, Cl⁺, HCl and H₂Cl⁺ are shown in the right-hand panel of Fig. 10. The PDR chemistry model predicts N(H₂Cl⁺) = 1.1 × 10¹³ cm^-2, an order of magnitude below the observed value.

An order of magnitude excess of the observed H₂Cl⁺ column density over the chemical model predictions, such as we find for FIR 4, has been noted since the recent first identification of the species towards NGC 6334I and Sgr B2(S) by Lis et al. (2010), who suggested viewing geometry as a possible explanation – PDR model column densities are commonly given for a face-on viewing angle. The discrepancy was discussed in detail for several sources by Neufeld et al. (2012), but no satisfactory explanation surfaced, although suggestions for future work were given by the authors in their Sect. 6. Thus, while our data is consistent with no depletion of chlorine from the gas in the tenuous foreground layer, evidently our understanding either of chlorine in diffuse PDRs or of the geometry and conditions in these regions, is not yet complete.

7. Discussion

7.1. Depletion of chlorine

We find a ratio of elemental volatile chlorine to gas-phase HCl, X(Cl)_tot/X(HCl)_gas, of ~1000 in FIR 4. Previous studies of the HCl abundance in molecular gas have found ratios in the range 50 to 640, typically ~400 (e.g. Schilke et al. 1995; Zmuidzinas et al. 1995; Salez et al. 1996; Neufeld & Green 1994; Peng et al. 2010). A depletion factor of ~100 was found for the outflow-shocked region L1157-B1 (Codella et al. 2012).

Our modelling suggests that hydrogen chloride ice is the main chlorine reservoir in protostellar core conditions, containing 90 to 98% of the elemental volatile chlorine. Gas-phase atomic Cl contains most of the remaining 10% of the chlorine.

The above result is independent of our choice of cosmic ray ionization rate, ζ, although inside of ~5000 AU, the gas-phase abundance of HCl (a minor, but key part in the chlorine budget) can vary substantially depending on this parameter. For ζ = 10^-16 s^-1, the gaseous HCl abundance stays similar to that of the outer envelope (~10^-10), however for ζ = 10^-14 s^-1 it is 10^-8 up to 10⁵ yr. This is, at face value, not consistent with our observations, as it would cause extremely strong HCl emission which is not observed. On the other hand, evolving the static ζ = 10^-14 s^-1 model beyond 10⁵ yr leads to a decreasing abundance of HCl in the gas, and it eventually falls below 10^-10 in the inner envelope.

Another possibility is that infall influences the abundances in the inner envelope, keeping more chlorine in HCl ice than is seen in our static models. This may require quite rapid infall (within a few thousand years on 5000 AU scales) in order to prevent the buildup of a high gas-phase abundance. The poorly known physical structure on ≲2000 AU scales also impacts our modelling of the chemistry as well as the excitation of HCl, leading to further uncertainty about the HCl abundance on small scales. Another possibility is that HCl is liberated from grain mantles more slowly than the bulk ice. This might also explain the low gas-phase HCl abundance found in the L1157-B1 shock by (Codella et al. 2012), and would point to some – as yet unknown – relatively refractory reservoirs of chlorine on grains.

The dominance of HCl ice as a reservoir of volatile Cl warrants a discussion of the relevant grain surface processes. Much chlorine arrives on ice mantles in atomic form, rather than already in HCl, and then subsequently reacts with H to form HCl. Chlorine may also react with H₂ also present on an interstellar grain surface, as the barrier for reaction with H₂ is measured to be only 2300 K in the gas phase. On a grain surface, H₂ can tunnel through a barrier of up to 4700 K (Tielens & Hagen 1982). In this way, Cl will act similarly to OH, which also has a low barrier for reaction with H₂  and theory and experiments have shown that that reaction is key to interstellar H₂O formation (Tielens & Hagen 1982; Oba et al. 2012).

Analogous to water solutions of hydrochloric acid, and given the low elemental abundance of chlorine, adsorbed HCl can solvate as a trace ion pair (Cl⁻ and H₃O⁺, e.g. Horn et al. 1992). Theoretical studies suggest this process is energetically allowed on an ice surface and proceeds rapidly by tunnelling at 190 K (Robertson & Clary 1995). Experimental studies show that HCl adsorbs dissociatively at sub-monolayer coverages onto the surface of dense amorphous solid water at temperatures as low as 20 K (Ayotte et al. 2011). As dangling OH bonds are involved – which will be omnipresent on growing interstellar ice surfaces – and in view of the long interstellar timescales, we consider solvation likely on a 10 K icy interstellar grain. Observationally, it is well established that ion-solvation is a key aspect of interstellar ices (Demyk et al. 1998) and experiments have shown that ion-solvation can occur at low temperatures and is promoted by the presence of strong bases such as NH₃, leading to trapped Cl⁻–NH$\begin{array}{c}\mathrm{+}\\ \mathrm{4}\end{array}$ ion pairs (Grim et al. 1989). Dipole alignment in ice mantles can further assist in ion-solvation (Balog et al. 2011).

Upon warmup, HCl will evaporate close to the H₂O evaporation temperature. This likely involves the relaxation of the water ice matrix, followed by the recombination and evaporation of HCl (Olanrewaju et al. 2011). The observed depth of depletion outside of the hot core in FIR 4 is consistent with such a codesorption scenario, although with the present data we can only place an upper limit of 10^-8 on the gas-phase HCl abundance in the ~500 AU size hot core (or 10^-7 within 100 AU).

7.2. Uncertainty in the source luminosity

The 1000 L_⊙ luminosity of the FIR 4 source model from C09 exceeds the protostellar luminosity of 100 L_⊙ found by Furlan et al. (2014, F14). The latter authors, as well as López-Sepulcre et al. (2013b), attribute this to the lower spatial resolution data and larger photometric annuli used by C09. We assess here the potential impact on our results.

While the far-infrared fluxes of C09 were likely contaminated by the nearby source, FIR 3, the (sub-)mm continuum maps were spatially resolved on a scale comparable to that studied by F14 and must be reproduced by any model. For a luminosity and thus cooling rate decrease of 1 dex, the corresponding dust temperature decrease is a factor of 1.5 or ~30%. The fixed mm flux demands a matching increase in the density of the emission region. Recalculating the HCl 1–0 excitation for these conditions implies a change in our derived HCl abundance by at most a factor of two.

The impact of the changed source luminosity on the chemistry is also expected to be small. Lowered temperatures would most of all assist in keeping chlorine locked in ices.

7.3. The broad component

The broad component has a combination of relatively large line width (~10 km s^-1) and large spatial extent (thousands of AU), while embedded in, or at least projected over, a dense protostellar core. As we discuss below, it is not obvious what the nature of this component is. Kinematical evidence relates the broad HCl component to the CS molecule, mapping of which in turn suggests that this component is extended on a scale of ≳2100 AU. Our constant HCl abundance models suggest a best-fit spherically symmetric radial extent of 7000 AU. At first glance, such a large line width and large spatial scale make hypotheses other than an outflow seem unlikely.

There is indeed mounting evidence from kinematical and excitation considerations that FIR 4 indeed hosts a compact outflow (Kama et al. 2013, and in prep.; Furlan et al. 2014). Spatially and spectrally resolved data, to be presented in a companion paper, suggest that the outflow axis runs roughly north to south, with lobe sizes of at a few thousand AU. However, the C³⁴S map in Fig. 4 shows a significant east-west elongation, which seems difficult to explain with such an outflow, unless outflow-driven gas is spilling over the protostellar core surface where the outflow cone breaks out. The C³⁴S velocity map shows a slow rotation around the north-south axis, with a typical velocity an order of magnitude below the linewidth.

It has been proposed that a larger outflow from the nearby Class I source, OMC-2 FIR 3, impacts and shocks the FIR 4 core (Shimajiri et al. 2008). It seems unlikely, however, that the FIR 3 outflow is responsible for the broad line emission in FIR 4, because the broad C³⁴S emission peaks on-source, and because the high-velocity wings of the CO and H₂O lines in FIR 4 are perfectly symmetric around the local v_lsr. Spatially resolved studies of the high velocity wings of CO are needed to clarify the issue. Previous interferometric observations have had insufficient sensitivity to probe the outflow gas at several tens of km s^-1.

7.4. The chlorine isotopic ratio

Studies of this ratio throughout the Galaxy have typically been consistent with the solar system value of 3.1 (Lodders 2003), within large error bars (e.g. Salez et al. 1996; Peng et al. 2010; Cernicharo et al. 2010).

Because of the very small Cl isotope mass difference (6%), minimal chemical fractionation is expected, and we determined the isotope ratio via the HCl and H₂Cl⁺ isotopolog ratios. For HCl, we find a line flux ratio of 3.2 ± 0.1, which is a robust isotopolog ratio indicator, given the low optical depth of the lines suggested by the hyperfine component ratios of the narrow HCl emission (see also Cernicharo et al. 2010). For H₂Cl⁺, we found a ratio of 4.3 ± 0.8. The results, summarized in Table 4, are consistent with the solar system value and with values measured elsewhere in the Orion star forming region.

7.5. Impact of the new HCl-H₂ excitation rates

That the difference between the HCl-H₂ excitation rate coefficients and the scaled HCl-He ones should impact abundance determinations was noted already by Lanza et al. (2014b). As shown in Fig. 6, the new HCl-H₂ hyperfine-resolved collisional excitation rates are roughly a factor of five to ten larger than the previously used, mass-scaled HCl-He rates from Neufeld & Green (1994) and Lanza & Lique (2012). This suggests that previous estimates of the gas-phase HCl abundance in molecular gas must be re-evaluated to be up to an order of magnitude lower, and correspondingly the typical fraction of elemental chlorine in gas-phase HCl must be around a factor of 10^-3 (a depletion factor of ~1000). Modelling results in Sect. 5.3 show that the strong depletion can be well understood in a framework where elemental chlorine is sequestered into HCl ice, where it remains at least as strongly bound as H₂O itself.

Based on the new excitation rates, the critical density of the HCl 1–0 transition is ~10⁷ cm^-3. This is accurate within a factor of a few in the temperature range of the new rate coefficients (up to 300 K).

8. Conclusions

We carried out a study of chlorine towards the OMC-2 FIR 4 protostellar core, using Herschel and CSO observations of HCl and H₂Cl⁺. Our main findings are listed below.

• 1.

  We detect the HCl 1−0 and 2−1 transitions in emissionwith Herschel and CSO, andH₂Cl⁺ in absorption with Herschel.

• 2.

  The narrow HCl component (FWHM = 2.0 km s^-1) traces the outer envelope, and the broad one (FWHM = 10.5 km s^-1) a compact central region, possibly outflow-driven gas.

• 3.

  The HCl data are well modelled with a constant abundance of X(HCl)_gas = 9 × 10^-11 in FIR 4, corresponding to ~10^-3 of the ISM abundance of elemental chlorine.

• 4.

  Chemical models show that HCl ice contains ~90 to 98% of all volatile chlorine in the source. The second largest reservoir is gas-phase atomic Cl, up to 10% of the total. All other species have much lower abundances. In the inner 100 AU, HCl gas may hold up to 100% of volatile chlorine.

• 5.

  The external irradiation of the FIR 4 envelope is G₀ = 415 ISRF. Elemental chlorine is undepleted in the outermost 300 AU of the resulting dense PDR. Including this PDR in the source model gives a best-fit X(HCl)_gas = 7 × 10^-11 in the rest of the source.

• 6.

  H₂Cl⁺ traces a recently discovered diffuse, blueshifted foreground PDR. The observed H₂Cl⁺ column density is 1.3 × 10¹⁴ cm^-2, an order of magnitude above the model prediction of 1.1 × 10¹³ cm^-2.

• 7.

  Our best estimate of the [³⁵Cl]/[³⁷Cl] isotope ratio in OMC-2 FIR 4 is 3.2 ± 0.1 (1σ), consistent with other measurements in the solar system and in the Orion region.

• 8.

  Newly calculated HCl-H₂ hyperfine-resolved collisional excitation rate coefficients exceed previous HCl-He scaled values by up to an order of magnitude at protostellar core temperatures, suggesting that previous estimates of chlorine depletion from the gas should be revisited.

Online material

Appendix A: Hyperfine excitation of HCl by H${\mathrm{2}}_{}$

Rate coefficients for rotational excitation of HCl(${}^{\mathrm{1}}\mathrm{\Sigma }^{\mathrm{+}}$) by collisions with H₂ molecules have been computed by Lanza et al. (2014a) for temperatures ranging from 5 to 300 K. The rate coefficients were derived from extensive quantum calculations using a new accurate potential energy surface obtained from highly correlated ab initio approaches (Lanza et al. 2014b).

However, in these calculations, the hyperfine structure of HCl was neglected. To model the spectrally resolved HCl emission from molecular clouds, hyperfine resolved rate coefficients are needed. In this appendix, we present the calculations of HCl-H₂ hyperfine resolved rate coefficients from the rotational rate coefficients of Lanza et al. (2014a). Note that, for rotational levels, we use here the lowercase j instead of the astronomical J notation used in the main body of the paper.

Appendix A.1: Methods

In HCl, the coupling between the nuclear spin (I₁ = 3 / 2) of the chlorine atom and the molecular rotation results in a weak splitting of each rotational level j₁ into 4 hyperfine levels (except for the j₁ = 0 level which is split into only 1 level and for the j₁ = 1 level which is split into only 3 levels). Each hyperfine level is designated by a quantum number F₁ (F₁ = I₁ + j₁) varying between | I₁ − j₁ | and I₁ + j₁. In the following, j₂ designates the rotational momentum of the H₂ molecule.

In order to get HCl–H₂ hyperfine resolved rate coefficients, we extend the Infinite Order Sudden (IOS) approach for diatom-atom collisions (Faure & Lique 2012) to the case of diatom-diatom collisions.

Within the IOS approximation, inelastic rotational rate coefficients ${\mathit{k}}_{{\mathit{j}}_{\mathrm{1}}\mathit{,}{\mathit{j}}_{\mathrm{2}}\mathrm{\to }{\mathit{j}}_{\mathrm{1}}^{\mathrm{\prime }}\mathit{,}{\mathit{j}}_{\mathrm{2}}^{\mathrm{\prime }}}^{\mathrm{IOS}}\mathrm{\left(}\mathit{T}\mathrm{\right)}$ can be calculated from the “fundamental” rates (those out of the lowest j₁ = 0,j₂ = 0 channel) as follows (e.g. Alexander 1979): $\begin{array}{ccc}{\mathit{k}}_{{\mathit{j}}_{\mathrm{1}}\mathit{,}{\mathit{j}}_{\mathrm{2}}\mathrm{\to }{\mathit{j}}_{\mathrm{1}}^{\mathrm{\prime }}\mathit{,}{\mathit{j}}_{\mathrm{2}}^{\mathrm{\prime }}}^{\mathrm{IOS}}\mathrm{\left(}\mathit{T}\mathrm{\right)}& \mathrm{=}& \sum _{{\mathit{L}}_{\mathrm{1}}\mathit{,}{\mathit{L}}_{\mathrm{2}}}\mathrm{(}\mathrm{2}{\mathit{j}}_{\mathrm{2}}^{\mathrm{\prime }}\mathrm{+}\mathrm{1}\mathrm{)}\mathrm{(}\mathrm{2}{\mathit{j}}_{\mathrm{1}}^{\mathrm{\prime }}\mathrm{+}\mathrm{1}\mathrm{)}{\left(\begin{array}{c}\\ & & \\ & & \end{array}\right)}^{\mathrm{2}}\\ & & \end{array}$(A.1)Similarly, IOS rate coefficients amongst hyperfine structure levels can be obtained from the ${\mathit{k}}_{\mathrm{0}\mathit{,}\mathrm{0}\mathrm{\to }{\mathit{L}}_{\mathrm{1}}\mathit{,}{\mathit{L}}_{\mathrm{2}}}^{\mathrm{IOS}}\mathrm{\left(}\mathit{T}\mathrm{\right)}$ rate coefficients using the following formula: $\begin{array}{ccc}{\mathit{k}}_{{\mathit{j}}_{\mathrm{1}}\mathit{,}\hspace{0.17em}{\mathit{F}}_{\mathrm{1}}\mathit{,}\hspace{0.17em}{\mathit{j}}_{\mathrm{2}}\mathrm{\to }{\mathit{j}}_{\mathrm{1}}^{\mathrm{\prime }}\mathit{,}\hspace{0.17em}{\mathit{F}}_{\mathrm{1}}^{\mathrm{\prime }}\mathit{,}\hspace{0.17em}{\mathit{j}}_{\mathrm{2}}^{\mathrm{\prime }}}^{\mathrm{IOS}}\mathrm{\left(}\mathit{T}\mathrm{\right)}& \mathrm{=}& \sum _{{\mathit{L}}_{\mathrm{1}}\mathit{,}\hspace{0.17em}{\mathit{L}}_{\mathrm{2}}}\mathrm{(}\mathrm{2}{\mathit{j}}_{\mathrm{2}}^{\mathrm{\prime }}\mathrm{+}\mathrm{1}\mathrm{)}\mathrm{(}\mathrm{2}{\mathit{j}}_{\mathrm{1}}\mathrm{+}\mathrm{1}\mathrm{)}\mathrm{(}\mathrm{2}{\mathit{j}}_{\mathrm{1}}^{\mathrm{\prime }}\mathrm{+}\mathrm{1}\mathrm{)}\\ & & \mathrm{\times }\mathrm{(}\mathrm{2}{\mathit{F}}_{\mathrm{1}}^{\mathrm{\prime }}\mathrm{+}\mathrm{1}\mathrm{)}{\left(\begin{array}{c}\\ & & \\ & & \end{array}\right)}^{\mathrm{2}}{\left(\begin{array}{c}\\ & & \\ & & \end{array}\right)}^{\mathrm{2}}\\ & & \mathrm{\times }{\left\{\begin{array}{c}\\ & & \\ & & \end{array}\right\}}^{\mathrm{2}}{\mathit{k}}_{\mathrm{0}\mathit{,}\hspace{0.17em}\mathrm{0}\mathrm{\to }{\mathit{L}}_{\mathrm{1}}\mathit{,}\hspace{0.17em}{\mathit{L}}_{\mathrm{2}}}^{\mathrm{IOS}}\mathrm{\left(}\mathit{T}\mathrm{\right)}\end{array}$(A.2)where ${}^{\mathrm{\right(}} ^{\mathrm{\left)}}$ and { } are respectively the “3 − j” and “6 − j” Wigner symbols.

Fig. A.1 Temperature dependence of the hyperfine resolved HCl–para-H2 (upper panel) and HCl–ortho-H2 (lower panel) rate coefficients for HCl(${\mathit{j}}_{\mathrm{1}}\mathrm{=}\mathrm{2}\mathit{,F}\mathrm{\to }{\mathit{j}}_{\mathrm{1}}^{\mathrm{\prime }}\mathrm{=}\mathrm{1}\mathit{,}{\mathit{F}}^{\mathrm{\prime }}$) transitions.

The IOS approximation is expected to be moderately accurate at low temperature. As suggested by Neufeld & Green (1994), we could improve the accuracy by computing the hyperfine rate coefficients as: ${\mathit{k}}_{{\mathit{j}}_{\mathrm{1}}\mathit{,}\hspace{0.17em}{\mathit{F}}_{\mathrm{1}}\mathit{,}\hspace{0.17em}{\mathit{j}}_{\mathrm{2}}\mathrm{\to }{\mathit{j}}_{\mathrm{1}}^{\mathrm{\prime }}{\mathit{F}}_{\mathrm{1}}^{\mathrm{\prime }}\mathit{,}\hspace{0.17em}{\mathit{j}}_{\mathrm{2}}^{\mathrm{\prime }}}^{\mathrm{SIOS}}\mathrm{\left(}\mathit{T}\mathrm{\right)}\mathrm{=}\frac{{\mathit{k}}_{{\mathit{j}}_{\mathrm{1}}\mathit{,}\hspace{0.17em}{\mathit{F}}_{\mathrm{1}}\mathit{,}\hspace{0.17em}{\mathit{j}}_{\mathrm{2}}\mathrm{\to }{\mathit{j}}_{\mathrm{1}}^{\mathrm{\prime }}\mathit{,}\hspace{0.17em}{\mathit{F}}_{\mathrm{1}}^{\mathrm{\prime }}\mathit{,}\hspace{0.17em}{\mathit{j}}_{\mathrm{2}}^{\mathrm{\prime }}}^{\mathrm{IOS}}\mathrm{\left(}\mathit{T}\mathrm{\right)}}{{\mathit{k}}_{{\mathit{j}}_{\mathrm{1}}\mathit{,}\hspace{0.17em}{\mathit{j}}_{\mathrm{2}}\mathrm{\to }{\mathit{j}}_{\mathrm{1}}^{\mathrm{\prime }}\mathit{,}\hspace{0.17em}{\mathit{j}}_{\mathrm{2}}^{\mathrm{\prime }}}^{\mathrm{IOS}}\mathrm{\left(}\mathit{T}\mathrm{\right)}}{\mathit{k}}_{{\mathit{j}}_{\mathrm{1}}\mathit{,}\hspace{0.17em}{\mathit{j}}_{\mathrm{2}}\mathrm{\to }{\mathit{j}}_{\mathrm{1}}^{\mathrm{\prime }}\mathit{,}\hspace{0.17em}{\mathit{j}}_{\mathrm{2}}^{\mathrm{\prime }}}^{\mathrm{CC}}\mathrm{\left(}\mathit{T}\mathrm{\right)}$(A.3)using the CC rate coefficients ${\mathit{k}}_{\mathrm{0}\mathit{,}\hspace{0.17em}\mathrm{0}\mathrm{\to }{\mathit{L}}_{\mathrm{1}}\mathit{,}\hspace{0.17em}{\mathit{L}}_{\mathrm{2}}}^{\mathrm{CC}}\mathrm{\left(}\mathit{T}\mathrm{\right)}$ of Lanza et al. (2014a) for the IOS “fundamental” rates in Eqs. (A.1)–(A.2). ${\mathit{k}}_{{\mathit{j}}_{\mathrm{1}}\mathit{,}\hspace{0.17em}{\mathit{j}}_{\mathrm{2}}\mathrm{\to }{\mathit{j}}_{\mathrm{1}}^{\mathrm{\prime }}\mathit{,}\hspace{0.17em}{\mathit{j}}_{\mathrm{2}}^{\mathrm{\prime }}}^{\mathrm{CC}}\mathrm{\left(}\mathit{T}\mathrm{\right)}$ are the rotational rate coefficients also taken from Lanza et al. (2014a). We named the method “SIOS” for scaled IOS.

In addition, fundamental excitation rates ${\mathit{k}}_{\mathrm{0}\mathit{,}\hspace{0.17em}\mathrm{0}\mathrm{\to }{\mathit{L}}_{\mathrm{1}}\mathit{,}\hspace{0.17em}{\mathit{L}}_{\mathrm{2}}}^{\mathrm{CC}}\mathrm{\left(}\mathit{T}\mathrm{\right)}$ were replaced by the de-excitation fundamental rates using the detailed balance relation: ${\mathit{k}}_{\mathrm{0}\mathit{,}\mathrm{0}\mathrm{\to }{\mathit{L}}_{\mathrm{1}}\mathit{,}\hspace{0.17em}{\mathit{L}}_{\mathrm{2}}}^{\mathrm{CC}}\mathrm{\left(}\mathit{T}\mathrm{\right)}\mathrm{=}\mathrm{(}\mathrm{2}{\mathit{L}}_{\mathrm{1}}\mathrm{+}\mathrm{1}\mathrm{)}\mathrm{(}\mathrm{2}{\mathit{L}}_{\mathrm{2}}\mathrm{+}\mathrm{1}\mathrm{)}{\mathit{k}}_{{\mathit{L}}_{\mathrm{1}}\mathit{,}\hspace{0.17em}{\mathit{L}}_{\mathrm{2}}\mathrm{\to }\mathrm{0}\mathit{,}\hspace{0.17em}\mathrm{0}}^{\mathrm{IOS}}\mathrm{\left(}\mathit{T}\mathrm{\right)}\mathit{.}$(A.4)This procedure is found to significantly improve the results at low temperature due to important threshold effects.

Hence, we have determined hyperfine HCl–H₂ rate coefficients using the computational scheme described above for temperature ranging from 5 to 300 K. We considered transitions between the 28 first hyperfine levels of HCl (j, j′ ≤ 7) due to collisions with para-H₂(j₂ = 0) and ortho-H₂(j₂ = 1). The present approach has been shown to be accurate, even at low temperature, and has also been shown to induce almost no inaccuracies in radiative transfer modeling compared to more exact calculations of the rate coefficients (Faure & Lique 2012).

Appendix A.2: Results

The complete set of (de)excitation rate coefficients is available on-line from the LAMDA⁶ (Schöier et al. 2005) and BASECOL⁷ (Dubernet et al. 2013) websites. For illustration, Fig. A.1 depicts the evolution of para- and ortho-H₂ rate coefficients as a function of temperature for HCl(j = 2,F → j′ = 1,F′) transitions.

First of all and as already discussed in Lanza et al. (2014a), para- and ortho-H₂ rate coefficients differ significantly, the rate coefficients being larger for ortho-H₂ collisions. One can also clearly see that there is a strong propensity in favour of Δj₁ = ΔF₁ transitions for both collisions with para- and ortho-H₂. This trend is the usual trend for such a molecule (Roueff & Lique 2013).

Finally, we compare in Table A.1 our new hyperfine HCl–H₂ rate coefficients with the HCl–He ones calculated by Lanza & Lique (2012) which are scaled by a factor 1.38 to account for the mass difference (see Fig. 6 for a visual comparison).

Indeed, collisions with helium are often used to model collisions with para-H₂. It is generally assumed that rate coefficients with para-H₂(j₂ = 0) should be larger than He rate coefficients owing to the smaller collisional reduced mass.

As one can see, the scaling factor is clearly different from 1.38. The ratio varies with the transition considered and also with the temperature for a given transition. The ratio may be larger than a factor 10. This comparison indicates that accurate rate coefficients with para-H₂ (j₂ = 0) and ortho-H₂ (j₂ = 1) could not be obtained from He rate coefficients. HCl molecular emission analysis performed with HCl-He rate coefficients result in large inaccuracies in the HCl abundance determination.

Appendix B: CASSIS fitting of H₂Cl⁺

In Fig. B.1, we show the H₂Cl⁺ lines used in the LTE fitting with the CASSIS software and the best-fit model resulting from the χ² minimization. The hyperfine components and other lines detected nearby are also shown. The data are all from the HIFI spectral survey of FIR 4 (Kama et al. 2013).

Fig. B.1 The four H2Cl+ lines (black) used in the LTE model fitting with CASSIS, and the best-fit model (red). All spectra are corrected for the foreground PDR velocity of 9.4 km s-1. The dashed red lines show the hyperfine components of H2Cl+ transitions, while dashed blue lines in the top right panel indicate the native frequencies of C2H transitions. The other H2Cl+ transitions do not have any lines nearby that were listed as detections in the spectral survey of Kama et al. (2013).

Acknowledgments

We would like to thank the anonymous referee for constructive comments that helped to improve the manuscript. We also thank Catherine Walsh, Alexandre Faure, Yulia Kalugina, Laurent Wiesenfeld and Ewine van Dishoeck for useful discussions; Charlotte Vastel for help with molecular data; and Evelyne Roueff for support with the Meudon code. Astrochemistry in Leiden is supported by the Netherlands Research School for Astronomy (NOVA), by a Royal Netherlands Academy of Arts and Sciences (KNAW) professor prize, and by the European Union A-ERC grant 291141 CHEMPLAN. V.W. acknowledges funding by the ERC Starting Grant 3DICE (grant agreement 336474). F.L. and M.L. acknowledge support by the Agence Nationale de la Recherche (ANR-HYDRIDES), contract ANR-12-BS05-0011-01, by the CNRS national program “Physique et Chimie du Milieu Interstellaire” and by the CPER Haute-Normandie/CNRT/Énergie, Électronique, Matériaux. Support for this work was provided by NASA (Herschel OT funding) through an award issued by JPL/Caltech. We gratefully acknowledge Göran Pilbratt for granting Herschel Director’s Discretionary Time that greatly improved the HIFI data sensitivity. 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 Astronómico 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. The Caltech Submillimeter Observatory is operated by the California Institute of Technology under cooperative agreement with the National Science Foundation (AST-0838261). Based on analysis carried out with the CASSIS software. CASSIS has been developed by IRAP-UPS/CNRS.

References

All Tables

All Figures

	Fig. 1 Observed spectra of HCl from Herschel/HIFI and CSO, velocity-corrected and centred on the strongest hyperfine component. Red dashed lines indicate the hyperfine components of the rotational transitions, while blue dashed lines show CH3OH lines. The CSO data have been binned to δv = 3.7 km s-1.
In the text

	Fig. 2 Observed spectra of H2Cl+ from Herschel/HIFI. Only the lines formally detected in absorption are shown, and are velocity-corrected and centred on the strongest hyperfine component. Red dashed lines indicate the hyperfine components of the rotational transitions. The continuum has not been subtracted. See also Fig. B.1.
In the text

	Fig. 3 Kinematical properties of HCl and H2Cl+ compared to other species in FIR 4, each of which is annotated with the main physical component it traces. The label “PDR” refers here to the tenuous foreground cloud discovered by López-Sepulcre et al. (2013a). For the comparison species, the mean Gaussian fit parameters from Kama et al. (2013) are plotted.
In the text

	Fig. 4 Line width map of C34S 3–2 (colour map), obtained with the Plateau de Bure Interferometer (López-Sepulcre et al. 2013b). The channel width of the data is 4 km s-1. Overplotted are continuum contours (black lines), which start at 3σ with a 3σ step, with 1σ = 0.14 Jy km s-1beam-1. North is up and east is left.
In the text

Fig. 5 Top and middle panels: large-scale source structure of FIR 4 (ngas at top, Tgas at middle) from Crimier et al. (2009, C09), corresponding to an integrated luminosity of 1000 L⊙. Bottom panel: relative populations of the lowest HCl rotational states, showing J = 0 (black), J = 1 (blue), and J = 2 (red) for the C09 source structure without modification (solid lines) and with an added external irradiation of G0 = 415 ISRF (dashed lines; Sect. 5.4). The level populations were modelled with Ratran using the new hyperfine collision rate coefficients presented in Appendix A.

In the text

	Fig. 6 Rate coefficients for the collisional excitation HCl J = 0 → 1 by H2. The previously available HCl-He rate coefficients, scaled by 1.38 to account for the reduced mass difference, are from Neufeld & Green (1994, orange circles) and Lanza & Lique (2012, blue circles). The new HCl-ortho-H2 and -para-H2 rate coefficients (Appendix A) are shown by black squares and diamonds, respectively.
In the text

	Fig. 7 HCl 1–0 and 2–1 transitions as observed with HIFI and CSO (black). Overplotted are the best-fit constant HCl abundance model (red), and the reduced χ2< 3 range (shaded orange) of the models. The strongest feature in the CSO data, at −20 km s-1, is a CH3OH line.
In the text

	Fig. 8 log 10(reduced χ2) surface for the two constant HCl abundance model parameters: Xc and Rbn.
In the text

	Fig. 9 HCl and Cl gas- and ice-phase abundances by 104 (solid) and 105 yr (dashed) after the dark cloud stage, as modelled with Nautilus. The cosmic ray ionization rate is ζ = 10-16 s-1 (left panel) and ζ = 10-14 s-1 (right panel).
In the text

Fig. 10 Meudon models of the chlorine chemistry in the two PDR layers seen towards FIR 4, showing abundances relative to ngas. Left panel: dense PDR in the outermost envelope of FIR 4, with ngas= 6 × 105 cm-3, G0 = 415 ISRF. The cosmic ray ionization rate is ζ = 10-16 s-1 (solid lines) and ζ = 10-14 s-1 (dotted). The position is given outside-in from the right. Right panel: tenuous PDR in the foreground of FIR 4, with ngas= 102 cm-3, G0 = 1500 ISRF and ζ = 3 × 10-16 s-1. The position is given from the strongly irradiated side on the right.

In the text

	Fig. A.1 Temperature dependence of the hyperfine resolved HCl–para-H2 (upper panel) and HCl–ortho-H2 (lower panel) rate coefficients for HCl(${\mathit{j}}_{\mathrm{1}}\mathrm{=}\mathrm{2}\mathit{,F}\mathrm{\to }{\mathit{j}}_{\mathrm{1}}^{\mathrm{\prime }}\mathrm{=}\mathrm{1}\mathit{,}{\mathit{F}}^{\mathrm{\prime }}$) transitions.
In the text

Fig. B.1 The four H2Cl+ lines (black) used in the LTE model fitting with CASSIS, and the best-fit model (red). All spectra are corrected for the foreground PDR velocity of 9.4 km s-1. The dashed red lines show the hyperfine components of H2Cl+ transitions, while dashed blue lines in the top right panel indicate the native frequencies of C2H transitions. The other H2Cl+ transitions do not have any lines nearby that were listed as detections in the spectral survey of Kama et al. (2013).

In the text