Free Access
Issue
A&A
Volume 575, March 2015
Article Number L8
Number of page(s) 4
Section Letters
DOI https://doi.org/10.1051/0004-6361/201425553
Published online 20 February 2015

© ESO, 2015

1. Introduction

It has been shown that the global star-formation rate in the Universe gradually increases from z ~ 10 to z ~ 2–3 and then steeply decreases until the present epoch, z = 0 (see e.g. Dunlop, 2011, and references therein). Because metals are produced by stars, the determination of metal abundance in the gas provides complementary information about star-formation history (Rafelski et al., 2012). This can be done using damped Lyman-α absorption systems (DLAs) that represent the main reservoir of neutral gas at high redshift (Prochaska & Wolfe, 2009; Noterdaeme et al., 2009) and are likely to be located in galaxies or in their close environment (e.g. Krogager et al., 2012). These systems arise mostly in the warm neutral medium (e.g. Petitjean et al., 2000; Kanekar et al., 2014) and have a multicomponent velocity structure, with metal absorption lines spread typically over 100–500 km s-1 (Ledoux et al., 1998). In a small fraction of DLAs, the line of sight intercepts cold gas, as traced by molecular hydrogen (e.g. Noterdaeme et al., 2008, 2011; Balashev et al., 2014) and/or 21 cm absorption (e.g. Srianand et al., 2012). Important progress has been made towards understanding the properties of the gas, for example through deriving physical conditions (Srianand et al. 2005; Noterdaeme et al. 2007a; Jorgenson et al. 2009) and physical extent (Balashev et al. 2011), and the incidence of cold gas in DLAs has been related to other properties such as the metallicity (Petitjean et al., 2006) or the dust content (Ledoux et al., 2003; Noterdaeme et al., 2008). However, owing to the strong saturation of H i Lyman series lines, it remains impossible to directly determine the H i column density associated with the individual cold gas components traced by H2 absorption. Even for metals, whose absorption lines are not saturated, it is very difficult to determine what fraction originates from the cold phase. Difficulties arise as well with the 21 cm absorption that does not always exactly coincide with H2 absorption (Srianand et al., 2013) although it could be due to the different structures of the optical and radio emitting regions of the background quasars. Out of all the metals, chlorine shows a unique behaviour in the presence of H2. Because the ionization potential of chlorine (12.97 eV) is less than that of atomic hydrogen, chlorine is easily ionized in the diffuse neutral medium. However, this species reacts exothermically with H2 at a very high rate rapidly converting Cl+ into HCl+, which subsequently releases neutral chlorine through several channels (Jura, 1974; Neufeld & Wolfire, 2009). This process is so efficient that chlorine is completely neutral in the presence of a small amount of H2. In our Galaxy the fact that chlorine abundance anti-correlates with the average number density along the line of sight (Harris & Bromage, 1984; Jenkins et al., 1986) has been interpreted as chlorine depletion. However, models predict and observations indicate that gas with moderate dust content presents negligible depletion of chlorine (e.g. Neufeld & Wolfire, 2009; Savage & Sembach, 1996; Jenkins, 2009). Observationally, a tight relation is indeed found between Cl i and H2 in the local ISM (Jura, 1974; Sonnentrucker et al., 2006; Moomey et al., 2012). In this Letter, we present the first study of this relation at high redshift and over a wide range of column densities.

2. Data sample and measurements

Since the first detection by Levshakov & Varshalovich (1985), about two dozen H2 absorption systems have been detected at high redshifts in quasar spectra. The detection limit of the strongest Cl i absorption line (1347 Å, f = 0.0153, Schectman et al. 1993) in a high quality spectrum (S/N ~ 50, R ~ 50 000) corresponds to N(Cl i) ~ 1012 cm-2. The solar abundance of chlorine is 10-6.5 that of hydrogen (Asplund et al., 2009) and given previous measurements of N(Cl i)/2N(H2) (e.g. Moomey et al., 2012) we conservatively limit our study to systems with N(H2) ≳ 1017 cm-2.

thumbnail Fig. 1

Voigt profile fits to the newly detected H2 absorption lines from rotational levels J = 0 to J = 5 at z = 3.091485 towards J 21000641. The column densities are indicated (in log (cm-2)) in the top right corner of each panel.

Open with DEXTER

Redshifts, H i and H2 column densities, and metallicities were mainly taken from the literature and are based on VLT/UVES, Keck/HIRES, or HST/STIS data. We refitted H2 absorption systems towards Q 21230050 and Q J23400053 to take into account the positions of the detected Cl i components. We also detect a new H2 absorption system in the z = 3.09 DLA towards J 21000641 in which Jorgenson et al. (2010) have reported the presence of neutral carbon. Indeed, C i is known to be an excellent indicator of the presence of molecules (e.g. Srianand et al., 2005). We used the MAKEE package (Burles) to reduce archival data from this quasar obtained in 2005, 2006, and 2007 under programs U17H (PI: Prochaska), G400H (PI: Ellison), and U149Hr (PI: Wolfe). We have found strong H2 absorption lines from rotational levels up to J = 5 (see Fig. 1) at z = 3.091485 with a total column density of log N(H2) = 18.76 ± 0.03.

For all systems we retrieved data from the VLT/UVES or the Keck/HIRES archives. We reduced the data and fitted the lines using profile fitting. Neutral chlorine is detected in nine DLAs (Fig. 2). Four detections were already reported in the literature: Q 1232+082 (Balashev et al., 2011), Q 08123208 (Prochaska et al., 2003), Q 1237+0647 (Noterdaeme et al., 2010), and Q 2140-0321 (Noterdaeme et al. 2015). The remaining five are new detections. We measured upper-limits of N(Cl i) for the remaining nine systems. We used mainly the 1347 Å Cl i line. Whenever possible, we also used Cl i lines at 1088 Å, 1188 Å, 1084 Å, 1094 Å, and 1085 Å, with oscillator strength from Schectman et al. (1993), Morton (2003), Sonnentrucker et al. (2006), and Oliveira & Hébrard (2006), respectively.

Table 1 summarizes the results of Cl i measurements. We have kept all components with log N(H2) > 17. We did not use two known H2 absorption systems towards Q 00130029 and J 091826.16+163609.0 since H2 column densities in these systems are not well defined.

thumbnail Fig. 2

Voigt profile fits to Cl iλ1347 absorption lines associated with high-z strong H2 absorption systems.

Open with DEXTER

Table 1

Measurements of Cl i in strong H2 absorption systems at high redshift.

3. Results

Figure 3 shows the Cl i column density, N(Cl i), versus N(H2) and compares our high-z measurements to those obtained in the local ISM using the Copernicus satellite (Moomey et al., 2012). As can be seen, our high-z measurements extend the relation to H2 column densities ten times smaller than those measured in the local ISM. A very close correlation (r = 0.95) between Cl i and H2 has been found over the entire N(H2) range. It is striking that measurements at high and low redshifts are indistinguishable in the overlapping regime (log N(H2) ~ 19–20.2). The correlation is seen over about three orders of magnitude in column density with a dispersion of only about 0.2 dex. A least-squares bisector linear fit provides a slope of 0.83 and 0.87 for the high-z and z = 0 data, respectively, with an almost equal normalization (log N(Cl i) ≈ 13.7 at log N(H2) = 20). We note that the upper limits on N(Cl i) lie mostly at the low N(H2) end and are least constraining since they are compatible with the values expected from the above relation. For this reason, we will not consider them further in the discussion but still include them in the figures for completeness. The slopes are less than one, meaning that the Cl i/H2 ratio slightly decreases with increasing N(H2). This is unlikely to be due to conversion of Cl into H2Cl+ and/or HCl, since chlorine chemistry models (Neufeld & Wolfire, 2009) and measurements (e.g. towards Sgr B2(S), Lis et al. 2010) show that in diffuse molecular clouds only ~1% of chlorine is in the molecular form. A< 1 slope could in principle be due to dust depletion. However, there is no trend for increasing Cl depletion with increasing H2 or Cl i column densities in Galactic clouds (see Moomey et al. 2012). In addition, for high redshift measurements, elemental abundance patterns (Noterdaeme et al., 2008) as well as direct measurements (e.g. Noterdaeme et al., 2010) indicate Av< 0.2 when modelling of chlorine chemistry (Neufeld & Wolfire, 2009) shows that for such low extinction (Av< 1) all chlorine is in the gas phase.

A possibility is thus that the molecular fraction in the gas probed by Cl i is slightly increasing with increasing N(H2). It can be expected since H2 self-shielding increases while Cl i is already completely in the neutral form. Finally, the similarity between our Galaxy and high-z measurements at log N(H2) ≥ 19 might indicate that the chemical and physical conditions in the cold gas can be similar, otherwise fine tuning would be required between the different factors that affect the N(Cl i)-to-N(H2) ratio (e.g. number density, metallicity, dust content, and UV flux).

thumbnail Fig. 3

Column densities of Cl i versus that of H2. The red and blue points indicate the measurements at high redshift (this work) and in our Galaxy (Moomey et al., 2012), respectively. The straight dashed and dotted lines show the respective least-squares bisector fits to the data.

Open with DEXTER

Before continuing further, we note that in H2-bearing gas, chlorine is found exclusively in the neutral form (i.e. N(Cl)= N(Cl i)) (e.g. Jura, 1974). Since we expect that all chlorine is in gas-phase1, the abundance of chlorine, [Cl/H], in H2-bearing gas can be expressed as (1)where (2)and f = 2N(H2)/(2N(H2) + N(H i)) is the molecular fraction. Therefore, the ratio [Cl i/H2] gives a direct constraint on the chlorine-based metallicity of H2-bearing gas provided the molecular fraction of H2-bearing gas is known. Conversely, if a constraint can be put on the actual chlorine abundance, [Cl i/H2] can provide an estimate of the amount of H i present in H2-bearing gas. If Cl is depleted onto dust grains then the mentioned estimates of metallicity and molecular fraction will have to be corrected from the Cl depletion factor.

Figure 4 shows [Cl i/H2] as a function of the overall metallicity for DLAs (given in Table 1) at high redshift or as a function of [Cl/H] for clouds in our Galaxy. Since f ≤ 1, [Cl i/H2] gives an upper limit on the metallicity in H2-bearing gas, which is found to be roughly equal to or less than solar metallicity. For 13 out of 21 Cl i bearing clouds in our Galaxy, associated Cl ii was measured (Moomey et al., 2012). Therefore, we have estimated the overall chlorine abundance [Cl/H] of these clouds as (N(Cl i) + N(Cl ii))/(N(H i) + 2N(H2)) (see Fig. 4). Unfortunately, for high redshift DLAs, not only is Cl ii not detected, but DLAs also contain several H i clouds so that chlorine abundance of the very cloud of interest cannot be measured. Therefore, we consider the overall metallicity (averaged over velocity components) measured using another non-depleted element (usually Zn or S, see Table 1). In Fig. 4 it can be seen that the [Cl i/H2] ratio is likely not correlated with the overall metallicity of the DLA (Pearson correlation coefficient 0.3 at 0.3 significance level). For Milky Way clouds it can be seen that [Cl/H] is typically one third solar, which can be interpreted as evidence for chlorine depletion (Moomey et al., 2012).

The large difference between [Cl i/H2] and [X/H]DLA for the high redshift clouds can be explained by a molecular fraction f< 1 in H2-bearing clouds, by a higher metallicity in the H2-bearing gas compared to the overall DLA metallicity, or by both effects. If we assume that the metallicity in the H2-bearing gas is equal to the overall DLA metallicity we find (using Eq. (1)) that the molecular fraction in the H2-bearing gas is typically an order of magnitude higher than the overall inferred DLA molecular fraction. Interestingly, two systems sitting close to the one-to-one relation are those where CO molecules have been detected (Q 1439+1118 and Q 1237+0647). In such systems, the H2 component is probably fully molecularized and its metallicity is close to the overall metallicity of the DLA.

thumbnail Fig. 4

[Cl i/H2] as a function of the overall metallicity for high-z DLAs (red points) and the chlorine-based metallicity for Milky Way clouds (blue points). The dashed line represents the one-to-one relation.

Open with DEXTER

4. Conclusion

We have studied the neutral chlorine abundance in high redshift (z ~ 2–4) strong H2-bearing DLAs with log  N(H2) > 17.3. These systems arise in the cold neutral medium of galaxies in the early Universe. We have used 17 systems from the literature and also present a new H2 detection at z = 3.09145 in the spectrum of J 21000641. We have detected Cl i absorption lines in half of these systems, including five new detections. The derived upper limits for N(Cl i) for the remaining systems are shown to be consistent with the behaviour of the overall population. Our measurements extend the Cl i-H2 relation to lower column densities than measurements towards nearby stars. We show that there is a 5σ correlation between the column densities of both species over the range 18.1 < log N(H2) < 20.1 with indistinguishable behaviour between high and zero redshift systems. This suggests that at a given N(H2) the physical conditions are likely similar in our Galaxy and high-z gas, in spite of possible differences in the dust depletion levels. As we expect the Cl to be depleted less in the high-z absorbers studied here, we use the abundance of chlorine with respect to H2 to constrain the molecular fraction and the metallicity in H2-bearing gas. Our results suggest that the molecular fraction and/or the metallicity in the H2- and Cl i-bearing components could be much higher than the mean value measured over the whole DLA system. This implies that a large fraction of H i is unrelated to the cold phase traced by H2. Finally, our understanding of the formation of H2 onto dust grains, self-shielding, and lifetime of cold diffuse gas would certainly benefit from further observations of chlorine and molecular hydrogen in different environments and over a wide range of column densities.


1

The presence of Cl ii in the outer envelope of the H2 cloud is not excluded, but it does not influence our derivation.

Acknowledgments

S.B. and V.K. thank RF President Program (grant MK-4861.2013.2) and “Leading Scientific Schools of Russian Federation” (grant NSh-294.2014.2). R.S. and P.P. gratefully acknowledge support from the Indo-French Centre for the Promotion of Advanced Research (Centre Franco-Indien pour la Promotion de la Recherche Avancée) under contract No. 4304-2.

References

All Tables

Table 1

Measurements of Cl i in strong H2 absorption systems at high redshift.

All Figures

thumbnail Fig. 1

Voigt profile fits to the newly detected H2 absorption lines from rotational levels J = 0 to J = 5 at z = 3.091485 towards J 21000641. The column densities are indicated (in log (cm-2)) in the top right corner of each panel.

Open with DEXTER
In the text
thumbnail Fig. 2

Voigt profile fits to Cl iλ1347 absorption lines associated with high-z strong H2 absorption systems.

Open with DEXTER
In the text
thumbnail Fig. 3

Column densities of Cl i versus that of H2. The red and blue points indicate the measurements at high redshift (this work) and in our Galaxy (Moomey et al., 2012), respectively. The straight dashed and dotted lines show the respective least-squares bisector fits to the data.

Open with DEXTER
In the text
thumbnail Fig. 4

[Cl i/H2] as a function of the overall metallicity for high-z DLAs (red points) and the chlorine-based metallicity for Milky Way clouds (blue points). The dashed line represents the one-to-one relation.

Open with DEXTER
In the text

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

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

Initial download of the metrics may take a while.