© ESO 2019

1. Introduction

Chlorine is a minor constituent of the Universe, with a solar abundance of only ∼3 × 10⁻⁷ relative to hydrogen (Asplund et al. 2009), that is orders of magnitude smaller than the elemental abundances of carbon, nitrogen, and oxygen. Nevertheless, chlorine has a strong chemical tendency to form hydrides (e.g., H₂Cl⁺, HCl) whose abundances can be as high as those of CH and H₂O, for example, in diffuse molecular clouds (Neufeld et al. 2010; Sonnentrucker et al. 2010; Lis et al. 2010). This is because of its unique thermo-chemistry properties: first, chlorine has an ionization potential of 12.97 eV, which is lower than that of hydrogen, such that it is predominantly in its ionized stage, Cl⁺, in the atomic phase of the interstellar medium (ISM), and second, Cl⁺ can react exothermically with H₂ triggering an active chemistry.

In fact, the interstellar chemical network of chlorine is thought to be relatively simple, dominated by a handful of hydrides (Neufeld & Wolfire 2009), though the vast majority of interstellar chlorine will be in its neutral form in the presence of even small amounts of H₂. The reaction of Cl⁺ with H₂ readily forms HCl⁺, which has been detected in absorption in Galactic diffuse clouds (De Luca et al. 2012) harboring a significant amount (3–5%) of the gas-phase chlorine. In turn, HCl⁺ can further react exothermically with H₂ to form chloronium (H₂Cl⁺; first observed by Lis et al. 2010). Chloronium itself can react with free electrons to either form hydrogen chloride (HCl; historically the first chlorine-bearing molecule detected in the ISM; Blake et al. 1985) or release neutral chlorine. There are several pathways to destroy HCl through photodissociation, photoionization, or reactions with He⁺, ${\text{H}}_3^{+}$, and C⁺. Methyl chloride, CH₃Cl, was also recently identified in the young stellar object IRAS 16293 − 2422 and the gaseous coma of comet 67P/Churyumov–Gerasimenko (Fayolle et al. 2017), suggesting that chlorine chemistry in space also extends to more complex species (see, e.g., Acharyya & Herbst 2017 for gas-grain chemical models of chlorine chemistry).

From the nucleosynthetic point of view, elemental chlorine is produced on both short timescales in core-collapse supernovae, and long timescales in asymptotic giant branch (AGB) stars and Type Ia supernovae (p.164, Clayton 2003). Chlorine is produced during oxygen burning, from fast reactions with the more abundant alpha elements, in stars massive enough to ignite it. In core-collapse supernovae excess neutrons cause a great increase in chlorine abundance, and hence they produce the majority of chlorine in the universe through explosive nucleosynthesis. Type Ia supernovae produce smaller amounts of chlorine through explosive oxygen burning.

Chlorine can be found in two stable isotopes, ³⁵Cl and ³⁷Cl, with different sources producing different relative abundances. Low metallicity core-collapse supernovae, like the ones that likely enriched the pre-Solar nebula, produce a ³⁵Cl/³⁷Cl ratio of ∼3 as measured in the Solar system (Asplund et al. 2009). Higher metallicity core-collapse supernovae can produce lower ratios (Kobayashi et al. 2006, 2011). The isotopic ratios produced in AGB stars also depend heavily on metallicity, with low metallicities producing lower ³⁵Cl/³⁷Cl values (Cristallo et al. 2011).

Molecular absorbers at intermediate redshift (z ∼ 1) are powerful probes of the physico-chemical properties of distant galaxies. They offer a cosmological perspective on the chemical evolution of the Universe, for example with measurements of isotopic ratios at look-back times on the order of half or more of the present age of the Universe, but unfortunately, only a handful of such absorbers have been identified so far (e.g., Combes 2008; Wiklind et al. 2018). In this paper, we collect observations of chlorine-bearing species in the two best studied and most molecule-rich redshifted radio molecular absorbers, with the goal of investigating their chemical properties and measuring the ³⁵Cl/³⁷Cl isotopic ratio in different absorbing sightlines.

The molecular absorber located in front of the quasar PKS 1830−211 is a face-on spiral galaxy at a redshift z = 0.88582 (hereafter labeled MA0.89). The intervening galaxy lenses the quasar into two bright and compact images, leading to two independent sightlines in the southwest and northeast where absorption is detected (hereafter MA0.89 SW and MA0.89 NE), with impact parameters of ∼2 kpc and 4 kpc, respectively, in the absorber. These two compact, lensed images are embedded in a faint structure, reminiscent of an Einstein ring, seen at low radio frequencies (Jauncey et al. 1991). A wealth of molecules and their rare isotopologs have been detected in this absorber, especially in MA0.89 SW, which is characterized by a large H₂ column density (∼2 × 10²² cm⁻²) and moderate gas density ∼10³ cm⁻³ (e.g. Wiklind & Combes 1996; Muller et al. 2011, 2013, 2014a). MA0.89 NE is characterized by a lower H₂ column density (∼1 × 10²¹ cm⁻²) and more diffuse gas composition as shown by the enhancement of the relative abundances of hydrides such as OH⁺, H₂O⁺, and H₂Cl⁺ (e.g., Muller et al. 2016a).

The second molecular absorber (hereafter labeled MA0.68), located in front of the quasar B 0218 + 357, shares similar properties with MA0.89. It is a nearly face-on spiral galaxy at an intermediate redshift z = 0.68466, also lensing the background quasar into two main compact images and an Einstein ring, seen at low radio frequencies and centered on the faintest image. Molecular absorption has only been detected in one line of sight (toward the brightest image), at an impact parameter of ∼2 kpc from the absorber’s center (Wiklind & Combes 1995; Muller et al. 2007). A number of molecules and their isotopologs have also been observed in MA0.68 (e.g. Wallström et al. 2016), which has an H₂ column density (∼8 × 10²¹ cm⁻²) that is intermediate between MA0.89 SW and MA0.89 NE.

With these two absorbers, we are thus exploring three independent lines of sight, with different redshifts, galactocentric distances, H₂ column densities, visual extinctions, metallicities, local radiation fields, cosmic-ray ionization rates, etc. Our observations and results are presented in Sects. 2 and 3. We discuss our findings in terms of chemical properties derived from chlorine-bearing species and ³⁵Cl/³⁷Cl isotopic ratios in Sect. 4.

2. Observations

Observations of the two absorbers were obtained with the Atacama Large Millimeter/submillimeter Array (ALMA) between 2014 and 2017. A summary of these observations is given in Table 1. Details of the observational setups and data analysis are described below for each absorber separately.

MA0.68. Observations of H₂Cl⁺ were obtained on 2016 October 22 and 2017 May 02. During the first run, the precipitable water vapor content was ∼2.4 mm. The array was in a configuration where the longest baseline was 1.7 km. The bandpass calibrator was J 0237+2848 (∼2.2 Jy at 100 GHz) and the gain calibrator was J 0220+3241 (∼0.2 Jy at 100 GHz). During the second run, the precipitable water vapor content was < 1 mm. The array’s longest baseline was 1.1 km. The bandpass calibrator was J0510+1800 (∼2.6 Jy at 100 GHz) and the gain calibrator J 0205+3212 (∼0.9 Jy at 100 GHz). The data calibration was done with the CASA package following standard procedures. We further improved the data quality with a step of self-calibration of the phase, using a model of two point-sources for MA0.68. When combined, the two datasets result in a synthesized beam of ∼1.0″ × 0.6″ (P.A. = −176°) at 100 GHz, slightly larger than the separation of 0.3″ between the two lensed images of the quasar. However, we extracted the spectra along each line of sight by directly fitting each visibility set separately using the CASA-Python task UVMULTIFIT (Martí-Vidal et al. 2014), with a model of two point-sources. This is possible in the Fourier plane, thanks to the high signal-to-noise ratio of the data and the known and simple geometry of the source. The final spectrum was the weighted average of the individual spectra.

MA0.89. We complemented previous observations of the ortho and para forms of H₂Cl⁺ toward MA0.89 presented by Muller et al. (2014a) and Le Gal et al. (2017) with new observations of HCl obtained with ALMA during three observing runs: in 2014 May 03, 2014 May 06, and 2014 Jun 30. Both HCl isotopologs were observed in the same spectral window, and within the same receiver tuning as the para-H₂O⁺ data presented by Muller et al. (2017). The data reduction followed the same method as for MA0.68 data described above (self-calibration on the quasar, extraction of the spectra from visibility fitting, and weighted-averaged final spectrum).

3. Results

The spectra observed along the three lines of sight are shown in Fig. 1. H₂Cl⁺ is detected in all sightlines, unsurprisingly since it is ubiquitous in the Galactic diffuse and translucent gas (Lis et al. 2010; Neufeld et al. 2012). On the other hand, HCl is detected in only one of the two sightlines toward PKS 1830−211 (MA0.89 SW), with a stringent upper limit toward the other. It was not observed toward MA0.68.

Fig. 1. Absorption spectra of chlorine-bearing species observed toward MA0.89 SW (top three panels), MA0.89 NE (middle three panels), and MA0.68 (bottom panel). Hyperfine structure is indicated in green for 35Cl-isotopologs of each line toward MA0.89 SW. Hyperfine structure was deconvolved by fitting multiple Gaussian velocity components; these fits are shown in red on top of spectra, with residuals shown in blue. On the right, deconvolved opacity profiles (assuming optically thin lines with fc = 1, as explained in Sect. 3), normalized to their peak opacity, are shown, together with opacity profiles of some other species for comparison. The profile of p-H2O+ is of the strongest hyperfine component of the 110–101 transition, at rest frequency of 607 GHz. The profile of o-H2S is the 110–101 transition, at rest frequency of 168.8 GHz, which was observed simultaneously with o-H2Cl+ in the same tuning.

The H₂Cl⁺ absorptions reach at most a few percent of the continuum level for all sightlines, so we must determine whether the absorption is optically thin and whether the source covering factor, f_c, is less than unity. In MA0.89 SW, the ground state transition of species like CH⁺ and H₂O is strongly saturated and reaches a depth of nearly 100% of the continuum level, implying that f_c ≈ 1 for these species (Muller et al. 2014a, 2017). In MA0.89 NE, the situation is not so clear, but the deepest absorption reaches a depth of several tens of percent of the continuum level (Muller et al. 2017). For MA0.68, the covering factor of the 557 GHz water line is known to be large and close to unity (Combes & Wiklind 1997), as confirmed by recent ALMA data (Muller, priv. comm.). The HCl absorption reaches a depth of ∼10% of the continuum level toward MA0.89 SW, so without further evidence we consider f_c = 1 for all species and all sightlines – hence an optically thin line regime – and our derived column densities (Table 2) can be viewed, strictly speaking, as lower limits.

When detected, both H₂Cl⁺ and HCl show both their ³⁵Cl- and ³⁷Cl-isotopologs, observed simultaneously in the same spectral window (see a summary of the spectroscopic parameters of both species in Table A.1). Considering optically thin lines and given the very small difference in their relative weight, the abundance ratio of the ³⁵Cl- and ³⁷Cl-isotopologs should reflect the ³⁵Cl/³⁷Cl isotopic ratio (see Sect. 4.2).

In MA0.89, the ground-state transitions of both the ortho and para forms of H₂Cl⁺ were observed. It should be noted that, from the same data, Le Gal et al. (2017) determined ortho-to-para ratios (OPR) in agreement with the spin statistical weight 3:1 for both sightlines, within the uncertainties. In MA0.68, only ortho-H₂Cl⁺ was observed. We adopted an OPR = 3 for the calculations of H₂Cl⁺ total column densities in all sightlines.

Finally, both H₂Cl⁺ and HCl harbor a hyperfine structure that needs to be deconvolved before comparing intrinsic absorption profiles. This was done by fitting an intrinsic normalized profile, composed by the sum of individual Gaussian velocity components and convolved with the hyperfine structure (see Table A.1), using χ² minimization and taking the errors to be 1σ. Both the ³⁵Cl- and ³⁷Cl-isotopologs were treated in a common fit, with the ³⁵Cl/³⁷Cl ratio as a free parameter. In order to assess whether the hyperfine structure deconvolution was robust, we fitted the ortho and para lines of H₂Cl⁺ separately for MA0.89 SW. We obtained a good solution, with residuals at the noise level, with four Gaussian velocity components for this line of sight. For MA0.89 NE and MA0.68 the detections have lower signal-to-noise ratios, so we limited the number of Gaussian components to one for sake of simplicity. These fits still yield residuals consistent with the noise level. All the velocity profiles in Fig. 1 were normalized to the peak opacity to enable a straightforward comparison.

There is a striking difference in the opacity profiles of H₂Cl⁺ and HCl toward MA0.89 SW: the former has prominent line wings, at velocities |v| = 10 − 50 km s⁻¹, while the latter is mostly a narrow component centered at v = 0 km s⁻¹ with FWHM ∼ 20 km s⁻¹. In addition to the chlorine-bearing species of interest here, we also show in Fig. 1 the opacity profiles for some other species observed at the same epoch (to prevent effects from time variations of the absorption profile, see Muller & Guélin 2008; Muller et al. 2014a). In MA0.89 SW, we find that the profile of H₂Cl⁺ best matches that of H₂O⁺, while the profile of HCl best matches that of H₂S. Similarly, in MA0.68, we observe that the FWHM of the H₂Cl⁺ opacity is noticeably larger than that of H₂S, suggesting again that the two species trace different gas components.

Using the column density ratios between the two lines of sight of MA0.89, γ_SW/NE, Muller et al. (2016a); Muller et al. (2017) could classify the different species in two simple categories: those with low γ_SW/NE (≲5), for example ArH⁺, OH⁺, H₂Cl⁺, and H₂O⁺, which are known to trace gas with low molecular hydrogen fraction, f_H2¹, of 1–10% (see, e.g., Neufeld & Wolfire 2016), and those with γ_SW/NE above 20, for example CH, HF, HCO⁺, HCN, and H₂S, tracing gas with much higher molecular hydrogen fraction, > 30% and up to 100%. Since MA0.89 NE is dominated by gas with low f_H2 the non-detection of HCl in this line of sight, with a large γ_SW/NE >  90, suggests that HCl is a tracer of high-f_H2 gas, in agreement with chemical predictions (see Sect. 4). The line of sight towards MA0.68 is not yet fully characterized, but the estimated column density of H₂ and relative abundance of H₂Cl⁺ suggest that it is intermediate between MA0.89 NE and SW in term of molecular hydrogen fraction.

4. Discussion

4.1. Chemistry

Given the lack of observational constraints other than molecular abundances in the lines of sight through MA0.68 and MA0.89, and the fact that they sample different clouds and potentially different gas components, we have limited ourselves to a simple analytical model to explore the chlorine chemistry and how it varies with a few key parameters. For diffuse cloud conditions, HCl and H₂Cl⁺ are linked in a simple way, as HCl results from one specific channel of the dissociative recombination of H₂Cl⁺ and is principally destroyed by photodissociation:

$$\begin{array}{cc}& {\mathrm{H}}_2{\mathrm{Cl}}^{+}+{\mathrm{e}}^{-}\to \mathrm{HCl}+\mathrm{H}\hfill \end{array}$$(1)

$$\begin{array}{cc}& \mathrm{HCl}+\gamma \to \mathrm{H}+\mathrm{Cl}\hfill \end{array}$$(2)

with reaction rates k₁  =  7.5  ×  10⁻⁸(T/300)^−0.5 cm³ s⁻¹ (see Appendix B) and k₂ = 1.8 × 10⁻⁹ G₀exp(−2.9A_V) s⁻¹, where G₀ is the scaling factor of the interstellar UV radiation field (ISRF), expressed in Draine units (Heays et al. 2017). Under these conditions HCl is almost completely destroyed.

From these two reactions, we can write, at steady-state:

$$\begin{array}{c}\hfill [{\mathrm{H}}_2{\mathrm{Cl}}^{+}]/[\mathrm{HCl}]=\frac{k_2}{k_1{n}_{e^{-}}}\end{array}$$(3)

and, introducing the fractional ionization x_e = n(e⁻)/n_H and neglecting the A_V dependence of the photodissociation rate as in a diffuse line of sight A_V ≪ 1, we get

$$\begin{array}{c}\hfill \frac{[{\mathrm{H}}_2{\mathrm{Cl}}^{+}]}{[\mathrm{HCl}]}\sim 1.4\times {10}^{-2}\frac{G_0}{n_{\mathrm{H}}{x}_e}{\left(\frac{T}{100}\right)}^{0.5}·\end{array}$$(4)

A large [H₂Cl⁺] / [HCl] ratio is obtained for large values of G₀/n_H as suggested by Neufeld & Wolfire (2009).

We can perform a similar analysis for the [OH⁺] / [H₂O⁺] ratio by considering the main formation process of H₂O⁺ via the OH⁺ + H₂ reaction and its destruction by dissociative recombination and reaction with molecular hydrogen:

$$\begin{array}{cc}& {\mathrm{OH}}^{+}+{\mathrm{H}}_2\to {\mathrm{H}}_2{\mathrm{O}}^{+}+\mathrm{H}\hfill \end{array}$$(5)

$$\begin{array}{cc}& {\mathrm{H}}_2{\mathrm{O}}^{+}+e^{-}\to \mathrm{products}\hfill \end{array}$$(6)

$$\begin{array}{cc}& {\mathrm{H}}_2{\mathrm{O}}^{+}+{\mathrm{H}}_2\to {\mathrm{H}}_3{\mathrm{O}}^{+}+\mathrm{H}\hfill \end{array}$$(7)

with reaction rates k₅  =  1.7  ×  10⁻⁹ cm³ s⁻¹, k₆  =  4.3  ×  10⁻⁷(T/300)^−0.5 cm³ s⁻¹, and k₇  =  6.4  ×  10⁻¹⁰ cm³ s⁻¹. Substituting the fractional ionization x_e and the molecular fraction f_H2 we get

$$\begin{array}{cc}\hfill \frac{[{\mathrm{OH}}^{+}]}{[{\mathrm{H}}_2{\mathrm{O}}^{+}]}& =\frac{k_6[e^{-}]+k_7[{\mathrm{H}}_2]}{k_5[{\mathrm{H}}_2]}=\frac{2k_6{x}_e+k_7{f}_{{\mathrm{H}}_2}}{k_5{f}_{{\mathrm{H}}_2}},\hfill \\ \hfill & \sim 8.8\times {10}^2\frac{x_e}{f_{{\mathrm{H}}_2}}{\left(\frac{100}{T}\right)}^{0.5}+0.38.\hfill \end{array}$$(8)

Using the derived molecular fractions from Muller et al. (2016a), we can estimate the fractional ionization from the [OH⁺] / [H₂O⁺] ratio in the two MA0.89 lines of sight, assuming T = 100 K. The [H₂Cl⁺] / [HCl] ratio then allows for the derivation of G₀, again following Muller et al. (2016a) in the assumption of n_H = 35 cm⁻³.

The calculated values are reported in Table 3. As the derivation does not consider any dependence on the visual extinction, the derived values of the radiation field correspond to lower limits. We note that the SW line of sight has a significantly higher molecular fraction, and hence a larger total visual extinction (A_V; see Sect. 4.1.1), than the NE line of sight, so its ISRF is likely to be underestimated in our reasoning. The results depend strongly on the value of A_V, but as an example: for a cloud of A_V = 1 in MA0.89 SW, we calculate G₀ = 9. This suggests that both MA0.89 SW and MA0.89 NE have more intense radiation fields than in the Solar neighborhood.

Our simple model is broadly consistent with the results of Neufeld & Wolfire (2009), though we note that the chlorine chemical network still has some uncertainties, as discussed in Appendix B, which can significantly impact the model results. Neufeld & Wolfire (2009) find a column density ratio [H₂Cl⁺] / [HCl] ∼1, as we measure in MA0.89, for a two-sided slab model with G₀ = 10 and n_H = 315 cm⁻³ at A_V ∼ 1. However, the predicted column densities of these molecules are significantly smaller than the observed ∼10¹³ cm⁻² in our lines of sight.

4.1.1. Properties of each sightline

Here we discuss more specifically our results towards MA0.68 and MA0.89 and compare the physical and chemical properties in the different sightlines.

MA0.89 SW. This sightline is rich in molecular species, with almost 50 different species detected so far (e.g., Muller et al. 2011, 2014a), and the H₂ column density is relatively high, at ∼10²² cm⁻². We can estimate the total visual extinction A_V, tot in MA0.89 SW using Bohlin’s law (Bohlin et al. 1978) and the H and H₂ column densities (Chengalur et al. 1999; Muller et al. 2014a). We find A_V, tot ≈ 20 mag, which is shared among a number of individual clouds or kinematical structures. Obviously, there is at least one, most likely a few, cloud(s) with A_V high enough (∼1) in the line of sight for HCl to be detectable. Indeed, the line profile of HCl indicates two peaks close to v ∼ 0 km s⁻¹. Those two velocity components were previously noticed as two distinct clouds with different chemical setups, as they show opposite peaks in their CF⁺ vs. CH₃OH absorption (Muller et al. 2016b), suggesting chemical segregation.

The multi-phase composition of the absorbing gas in this sightline has been clearly revealed by the recent observations of hydrides with ALMA (Muller et al. 2014a, 2016b, 2017). Accordingly, the absorption from H₂Cl⁺ and HCl can be simply understood as tracing gas with different molecular fraction. In Fig. 2, we see a clear difference between the core and wings of the line profiles: in the line center the gas is denser or better shielded and HCl dominates, with a [H₂Cl⁺] / [HCl] ratio of ∼0.5, while in the wings the ratio increases to ≳1 and H₂Cl⁺ dominates. This is consistent with different gas properties (i.e., f_H2) between the core and wings.

Fig. 2. Abundance ratio of H2Cl+ (derived from ortho and para forms, shown in top and bottom panels, respectively) over HCl across absorption profile toward MA0.89 SW. We use measured ortho-to-para ratio of three to obtain total H2Cl+ column density. Abundance ratio is calculated only for channels where signal-to-noise ratio of normalized profile of both H2Cl+ and HCl is higher than five.

MA0.89 NE. Only H₂Cl⁺ is detected in this line of sight, which is characterized by gas at low f_H2. The [H₂Cl⁺] / [HCl] abundance ratio has a lower limit of ∼17, more than one order of magnitude higher than in the SW line of sight. Following the method outlined above, we estimate a total A_V ≈ 2 mag. Based on a statistical analysis, Muller & Guélin (2008) argued that the large time variations of the opacity profiles² imply a relatively small number of individual clouds in the line of sight, ≲5, and that the bulk of the absorption should arise from clouds not much smaller than the background continuum source, ∼1 pc. ALMA observations of the CH⁺ absorption (Muller et al. 2017) reveal a complex velocity profile which can be fitted with ∼10 different Gaussian velocity components, some with widths as small as a few km s⁻¹, spanning a total velocity interval between −300 km s⁻¹ and −100 km s⁻¹. Taking a simplified model of ∼10 individual clouds along the line of sight results in each having an average A_V of ∼0.2. From the chemical predictions, we indeed expect some production of H₂Cl⁺ at this level of A_V but very little HCl, consistent with the non-detection of HCl.

MA0.68. Here, we estimate a total A_V ≈ 8 magnitudes. The overall absorption profile is narrow compared to that of MA0.89 and we estimate the number of clouds to be ∼3 based on the number of distinct velocity features in the profile (e.g. Wallström et al. 2016), suggesting characteristics intermediate between MA0.89 SW and MA0.89 NE. The abundance of v relative to H₂ is also intermediate between the MA0.89 sightlines.

4.1.2. Elemental chlorine and abundances of chlorine-bearing molecules

Models of chlorine chemistry (e.g., Neufeld & Wolfire 2009) find that in the presence of H₂ most chlorine resides in the form of the neutral atom Cl. Observationally, a strong correlation is found between the column densities of Cl and H₂, in both the local ISM (Jura 1974; Sonnentrucker 2006; Moomey et al. 2012) and in high-redshift damped Lyman-α absorption systems (Balashev et al. 2015) over three orders of magnitude in H₂ column densities, with a relation³

$$\begin{array}{c}\hfill \mathrm{l}og[N_{\mathit{Cl}}]=(0.79\pm 0.06)\times \mathrm{l}og[N_{H_2}]-(2.13\pm 1.15).\end{array}$$(9)

This trend is found to be independent of the overall gas metallicity, suggesting that neutral chlorine could be an excellent tracer of H₂.

Assuming that the correlation in Eq. (9) also holds for our molecular absorbers at intermediate redshifts, we can infer the column density of neutral chlorine corresponding to the column density of H₂ in the different sightlines. With the observed column densities of H₂Cl⁺ and HCl, we can then estimate the fraction of chlorine in these molecules (assuming neutral chlorine is vastly dominant and summing the ³⁵Cl and ³⁷Cl-isotopologs). Accordingly, we estimate that about 0.7%, 1.8%, and 0.9% of chlorine is in the form of H₂Cl⁺ for MA0.89 SW, MA0.89 NE, and MA0.68, respectively. For MA0.89 SW, we also estimate that about 0.9% of Cl is in HCl. These values are uncertain by about an order of magnitude (due to dispersion of the N_Cl − N_H2 correlation and uncertainty in the H₂ column density), and averaged over the whole sightline.

4.1.3. HCl⁺ and CH₃Cl

Besides H₂Cl⁺ and HCl, we here discuss briefly some other chlorine-bearing species, already detected or potentially present in the ISM in our intermediate-redshift absorbers. As previously mentioned, HCl⁺ is the first chlorine-bearing species to be formed in the chlorine-chemistry network from the reaction of Cl⁺ and H₂, in gas with a low enough f_H2 to maintain some Cl⁺, and as such, HCl⁺ is an excellent tracer of the diffuse gas component. HCl⁺ has a complex spectrum, with multiple ground-state hyperfine transitions around 1.44 THz (De Luca et al. 2012). Unfortunately, those transitions fall out of ALMA bands for a redshift of 0.89 and HCl⁺ cannot be observed toward PKS 1830−211. It could in principle be observed toward B 0218+357 in ALMA Band 10, although with the observational challenge of a weak background continuum source at low elevation.

Another chlorine-bearing molecule, recently detected in the ISM, is methyl chloride (CH₃Cl). It has several hyperfine transitions which were covered in a deep spectral scan in the 7 mm band with the Australia Telescope Compact Array toward PKS 1830−211 (Muller et al. 2011). No features related to CH₃Cl (rest frequency near 79.8 GHz) are detected down to a few per mil of the continuum level. Simulated absorption spectra for the conditions appropriate for the MA0.89 absorber yield a peak optical depth of τ = 1.95 × 10⁻³ for the hyperfine-structure blend, with an upper limit on the column density of N(CH₃Cl) = 5 × 10¹² cm⁻². CH₃Cl was recently identified in the young stellar object IRAS 16293 − 2422 by Fayolle et al. (2017), who infer a CH₃Cl column density from these high-resolution ALMA data under the assumption of a fixed excitation temperature of 102 K of 4.6 × 10¹⁴ cm⁻² over a $0\stackrel{″}{.}5$ beam. A previous observation of HCl toward the same source by Peng et al. (2010) yielded well-constrained optical depths in the J = 1 − 0 transition but referred to a much larger beam size of $13\stackrel{″}{.}5$. If the HCl result is re-interpreted for a fixed excitation temperature of 102 K, the resulting column density of HCl would be 8.9 × 10¹⁴ cm⁻², and if the ratio of column densities is simply scaled by the ratio of beam areas, then the abundance ratio would be [CH₃Cl] / [HCl] = 7 × 10⁻⁴. This is likely to be an underestimate, because the HCl-emitting source probably does not fill the $13\stackrel{″}{.}5$ beam. Even so, the [CH₃Cl] / [HCl] ratio in IRAS 16293 − 2422 is not inconsistent with the value (4 ± 2)×10⁻³ determined from the Rosetta Orbiter Spectrometer for Ion and Neutral Analysis (ROSINA) measurements of the gaseous coma of comet 67P/Churyumov–Gerasimenko (Fayolle et al. 2017). This [CH₃Cl] / [HCl] ratio is consistent with our upper limit on CH₃Cl in MA0.89 SW.

4.2. ³⁵Cl/³⁷Cl isotopic ratio

The ³⁵Cl/³⁷Cl isotopic ratio has been studied in a range of Galactic sources ranging from circumstellar envelopes to molecular clouds and star-forming regions. The measured ratios vary between ∼1 and 5, though most fall around 2.5 (Maas & Pilachowski 2018) with uncertainties large enough to be consistent with the solar system ratio of 3.13 (Lodders et al. 2009). This value is measured from meteorites, and hence is indicative of the conditions at the formation of the solar system. Galactic chemical evolution models by Kobayashi et al. (2011) predict that the chlorine isotopic ratio in the solar system neighborhood has decreased, with increasing metallicity, since its formation to about 1.8 at the present day and Solar metallicity.

From H₂Cl⁺ absorption observed in 2012 in MA0.89 SW, Muller et al. (2014b) found a ratio ³⁵Cl/³⁷Cl = ${3.1}_{-0.2}^{0.3}$, consistent with the Solar system value. With the better quality of these new 2014 data, we revise this ratio (2.99 ± 0.05) and we are also able to measure it in MA0.89 NE (3.3 ± 0.3). These two measurements, at galactocentric radii of 2 and 4 kpc, respectively, are consistent each other, suggesting there are only small effects from metallicity or stellar population differences in the disk of MA0.89. In addition, we obtain another measurement from the HCl isotopologs in MA0.89 SW, 3.28 ± 0.08, relatively consistent with previous values and suggesting a good mixing between the different gas components with low and high f_H2.

In MA0.68, we find a ³⁵Cl/³⁷Cl ratio of 2.2 ± 0.3, from H₂Cl⁺ isotopologs. Wallström et al. (2016) found overall very similar C, N, O, and S isotopic ratios between MA0.68 and MA0.89 SW, which show clear evolution effects compared to the ratios found in the Solar neighborhood. The ³⁵Cl/³⁷Cl ratio is the first to deviate between MA0.89 SW and MA0.68, and it would be interesting to confirm this difference with further observations.

A ³⁵Cl/³⁷Cl ratio around three reflects nucleosynthesis products mainly from massive stars and core-collapse supernovae (Kobayashi et al. 2011). On the other hand, a ³⁵Cl/³⁷Cl ratio of ∼2 requires nucleosynthesis by either low-metallicity AGB stars or higher-metallicity Type II supernovae. As AGB stars affect their environment on long timescales, the first possibility requires MA0.68 to be a relatively old galaxy. For the second possibility, higher-metallicity Type II supernovae require short timescales and an intrinsically more metal-rich galaxy. The simplest explanation of the difference in chlorine isotopic ratios between MA0.89 and MA0.68 might be that MA0.68 has an intrinsically higher metallicity.

5. Summary and conclusions

We investigate the absorption of chlorine-bearing molecules in two molecular absorbers (three independent lines of sight with different properties) at intermediate redshift, MA0.89 located toward PKS 1830−211 (z_abs  =  0.89) and MA0.68 towards B 0218+357 (z_abs  =  0.68). H₂Cl⁺ was observed, and detected, toward all sightlines. HCl was observed only toward PKS 1830−211, but detected only in one of the two sightlines.

Our results and conclusions are summarized as follows:

• The comparison of the absorption spectra of chlorine-bearing species between the different sightlines and with that of other species, namely H₂O⁺ and H₂S, provides us with a simple classification based on molecular fraction (f_H2). H₂Cl⁺ and HCl trace the gas component with low and high f_H2 gas, respectively. This picture is consistent with predictions from chemical modeling, where HCl requires higher visual extinction (A_V ∼ 1) than H₂Cl⁺ to form.

• These two chlorine-bearing species can hence be used together to characterize the column of absorbing gas. In particular, toward MA0.89 SW, we find that the [H₂Cl⁺] / [HCl] abundance ratio varies from ∼0.5 at the line center, to ∼2 in the line wings, reflecting the multi-phase composition of the gas along the line of sight. The same ratio has a lower limit more than one order of magnitude higher toward MA0.89 NE.

• The detection of the ³⁵Cl- and ³⁷Cl-isotopologs allows us to measure the ³⁵Cl/³⁷Cl isotopic ratio at look-back times of about half the present age of the Universe. We find basically the same values at galactocentric radii of 2 kpc (³⁵Cl/³⁷Cl = 2.99 ± 0.05 from H₂Cl⁺ and 3.28 ± 0.08 from HCl) and 4 kpc (³⁵Cl/³⁷Cl = 3.3 ± 0.3) in the disk of MA0.89 and in the Solar neighborhood (³⁵Cl/³⁷Cl ∼ 3), while we find a lower ratio (³⁵Cl/³⁷Cl = 2.2 ± 0.3) in MA0.68. This could be interpreted as MA0.68 having an intrinsically higher metallicity. It would be interesting to obtain more measurements of the ³⁵Cl/³⁷Cl isotopic ratio, including at higher redshifts, to investigate whether this isotopic ratio is a useful tracer of evolution.

• The comparison of the observed abundances for H₂Cl⁺ and HCl with that predicted from chemical modeling suggests the need for a stronger interstellar radiation field in the disk of these absorbers than in the Solar neighborhood. Evidence for a slightly increased cosmic-ray ionization rate of atomic hydrogen was also found in MA0.89 (Muller et al. 2016a), potentially related to the higher star formation activity at these intermediate redshifts.

In short, the two chlorine-bearing species H₂Cl⁺ and HCl are sensible probes of the ISM in redshifted absorbers. They offer diagnostics of the molecular fraction, gas composition, and nucleosynthesis enrichment via the ³⁵Cl/³⁷Cl ratio. Their ground-state transitions are readily observable, for instance with ALMA, in a wide range of redshifts (z ∼ 0 − 6), making them interesting tools to probe physico-chemical properties and evolution effects in the young Universe.

Acknowledgments

This paper uses the following ALMA data: ADS/JAO.ALMA#2012.1.00056.S ADS/JAO.ALMA#2013.1.00020.S ADS/JAO.ALMA#2016.1.00031.S. ALMA is a partnership of ESO (representing its member states), NSF (USA) and NINS (Japan), together with NRC (Canada) and NSC and ASIAA (Taiwan) and KASI (Republic of Korea), in cooperation with the Republic of Chile. The Joint ALMA Observatory is operated by ESO, AUI/NRAO and NAOJ. S.H.J.W. acknowledges support by the Ministry of Science and Technology of Taiwan under grants MOST104-2628-M-001-004-MY3 and MOST107-2119-M-001-031-MY3, and from Academia Sinica under AS-IA-106-M03.

References

Appendix A: Spectroscopic data

Appendix B: Considerations on the chemical network of chlorine-bearing species

Neufeld & Wolfire (2009) have investigated the chlorine chemistry in different photodissociation region models where the thermal balance and chemical equilibrium are solved simultaneously for different densities, illumination conditions and visual magnitudes, and for two different values of the cosmic ionization rate: 1.8 × 10⁻¹⁷ s⁻¹ and 1.8 × 10⁻¹⁶ s⁻¹. That study allows for the derivation of different chlorine-chemistry regimes which affect the [H₂Cl⁺] / [HCl] ratio, where a large value is obtained for high radiation field and rather low density conditions.

However, the dissociative recombination (DR) of H₂Cl⁺ induces a major uncertainty in the chlorine chemical network, as discussed by Kawaguchi et al. (2016), which severely impacts the abundances of H₂Cl⁺ and HCl, and their ratio, as shown below. Three exothermic channels are available in the reaction:

$$\begin{array}{cc}\hfill {\mathrm{H}}_2{\mathrm{Cl}}^{+}+e^{-}& \to \text{H}+\text{H}+\mathrm{Cl}\hfill \end{array}$$(B.1)

$$\begin{array}{cc}\hfill {\mathrm{H}}_2{\mathrm{Cl}}^{+}+e^{-}& \to {\mathrm{H}}_2+\mathrm{Cl}\hfill \end{array}$$(B.2)

$$\begin{array}{cc}\hfill {\mathrm{H}}_2{\mathrm{Cl}}^{+}+e^{-}& \to \text{H}+\mathrm{HCl}\hfill \end{array}$$(B.3)

The values of the different reaction rate coefficients used in the Neufeld & Wolfire (2009) studies were unpublished results where the branching ratio towards HCl is taken as 10% and the total DR rate coefficient is taken as $1.2\times {10}^{-7}\times {(\frac{T}{300})}^{-0.85}$ cm³ s⁻¹. That value was challenged by Kawaguchi et al. (2016) who report a total DR rate coefficient of (0.46 ± 0.05)×10⁻⁷ cm³ s⁻¹ at 209 K from infrared time-resolved spectroscopy experiments. Last, but not least, Novotný et al. (2018) studied the dissociative recombination of the deuterated counterpart D₂Cl⁺ in an ion storage ring experiment and obtained a total DR rate coefficient of $4.3\times {10}^{-7}\times {(\frac{300}{T})}^{0.5}$ cm³ s⁻¹. The derived coefficient is claimed to be most reliable at low temperatures relevant for astrochemical applications, in contrast to other experiments performed at room temperature. In addition, the branching ratio towards DCl was found to be 8.7%. The dissociative recombination rate of H₂Cl⁺ is expected to be larger according to previous experiments on other hydrogenated or deuterated molecular ions, which is also anticipated from theoretical considerations (Jungen & Pratt 2009). Indeed, Wiens et al. (2016) obtain a DR rate coefficient of (1.1  ±  0.3)  ×  10⁻⁷ cm³ s⁻¹ for D₂Cl⁺ and (2.6  ±  0.8)  ×  10⁻⁷ cm³ s⁻¹ for H₂Cl⁺ in an afterglow measurement at 300 K. Novotný et al. (2018) emphasize that their measurement of D₂Cl⁺ DR at 300 K is within the error bar of the storage ring experimental result but question the corresponding temperature extrapolation at low temperature. Summing up all these facts, we assume that the DR rate coefficient of H₂Cl⁺ can be evaluated from the D₂Cl⁺ measurement of Novotný et al. (2018), assuming the same [D₂Cl⁺] / [H₂Cl⁺] ratio than that obtained by Wiens et al. (2016) at 300 K. The corresponding values are reported in Table B.1.

These values correspond to a total destruction channel of H₂Cl⁺ by electrons which is about seven times more efficient than that taken in Neufeld & Wolfire (2009) at 300 K whereas the branching ratio to HCl is similar. The ratio between the two estimates reduces to a factor of four at 50 K and a factor of two at 10 K, due to the different slopes assumed for the energy dependence of the DR cross sections. Given these uncertainties, a detailed chemical modeling appears superfluous.

All Tables

All Figures

Fig. 1. Absorption spectra of chlorine-bearing species observed toward MA0.89 SW (top three panels), MA0.89 NE (middle three panels), and MA0.68 (bottom panel). Hyperfine structure is indicated in green for 35Cl-isotopologs of each line toward MA0.89 SW. Hyperfine structure was deconvolved by fitting multiple Gaussian velocity components; these fits are shown in red on top of spectra, with residuals shown in blue. On the right, deconvolved opacity profiles (assuming optically thin lines with fc = 1, as explained in Sect. 3), normalized to their peak opacity, are shown, together with opacity profiles of some other species for comparison. The profile of p-H2O+ is of the strongest hyperfine component of the 110–101 transition, at rest frequency of 607 GHz. The profile of o-H2S is the 110–101 transition, at rest frequency of 168.8 GHz, which was observed simultaneously with o-H2Cl+ in the same tuning.

In the text

	Fig. 2. Abundance ratio of H2Cl+ (derived from ortho and para forms, shown in top and bottom panels, respectively) over HCl across absorption profile toward MA0.89 SW. We use measured ortho-to-para ratio of three to obtain total H2Cl+ column density. Abundance ratio is calculated only for channels where signal-to-noise ratio of normalized profile of both H2Cl+ and HCl is higher than five.
In the text