Gaia-assisted selection of a quasar reddened by dust in an extremely-strong Damped Lyman-{\alpha} Absorber at z=2.226

TL;DR

This study reports the discovery of a dust-reddened quasar at z=2.60 behind an extremely-strong DLA at z=2.226, using Gaia data combined with optical and infrared photometry, revealing a potentially underrepresented population of dusty absorbers.

Contribution

First demonstration of selecting dust-reddened quasars using Gaia astrometry combined with multi-wavelength photometry, uncovering rare dusty DLA systems.

Findings

Discovered a z=2.60 quasar reddened by dust in an extremely-strong DLA at z=2.226.

Measured a high neutral hydrogen column density of log N(HI)=21.95.

Identified dust extinction with A_V=0.82 mag and LMC-like extinction curve.

Abstract

Damped Lyman-{\alpha} Absorbers (DLAs) as a class of QSO absorption-line systems are currently our most important source of detailed information on the cosmic chemical evolution of galaxies. However, the degree to which this information is biased by dust remains to be understood. One strategy is to specifically search for QSOs reddened by metal-rich and dusty foreground absorbers. In this Letter we present the discovery of a z=2.60 QSO strongly reddened by dust in an intervening extremely-strong DLA at z=2.226. This QSO was identified through a novel selection combining the astrometric measurements from ESA's Gaia satellite with extent optical and near/mid-infrared photometry. We infer a total neutral atomic-hydrogen column density of log N(HI)=21.95{\pm}0.15 and a lower limit on the gas-phase metallicity of [Zn/H]>-0.96. This DLA is also remarkable in exhibiting shielded neutral gas witnessed in CI and tentative detections of CO molecular bands. The Spectral Energy Distribution (SED) of the QSO is well-accounted for by a normal QSO-SED reddened by dust from a DLA with a 10%-of-Solar metallicity, dust extinction of A_V=0.82{\pm}0.02mag, and LMC-like extinction curve including the characteristic 2175{\AA} extinction feature. Such QSO absorption-line systems have shown to be very rare in previous surveys, which have mostly revealed sight-lines with low extinction. The present case therefore suggests that previous samples have under-represented the fraction of dusty absorbers. Building a complete sample of such systems is needed to assess the significance of this effect.

Tables

Table 1. Table 1: Log of OSIRIS observations.

Date (UT)	Grism	Resolving	Exp. time	Airmass
		Power	[sec]
09/12/2018	R1000B	500	2$\times$1000	1.32–1.22
10/12/2018	R2500V	1680	1$\times$1500	1.92–1.68
11/12/2018	R2500V	1680	3$\times$1500	1.36–1.12
17/12/2018	R2500R	1510	2$\times$1350	1.20–1.12

Table 2. Table 2: Absorption lines from the z = 2.2261 ± 0.0004 𝑧 plus-or-minus 2.2261 0.0004 z=2.2261\pm 0.0004 DLA detected in the GTC R2500V and R2500R spectra.

Transition	EWr	Redshift
	[Å]
Si iv $\lambda$ 1393	$0.41\pm 0.11$	2.2262
Si iv $\lambda$ 1402a	$0.53\pm 0.09$	2.2262
CO $\lambda$ 1447	$0.20\pm 0.11$	2.2265
CO $\lambda$ 1477	$0.09\pm 0.06$	2.2265
CO $\lambda$ 1509	$0.09\pm 0.07$	2.2265
Si ii $\lambda$ 1526	$1.21\pm 0.07$	2.2257
C iv $\lambda$ 1548	$0.48\pm 0.05$	2.2266
C iv $\lambda$ 1550	$0.33\pm 0.05$	2.2255
C iv $\lambda$ 1548	$0.46\pm 0.04$	2.2474
C i $\lambda$ 1560a	$0.54\pm 0.04$	2.2265
Fe ii $\lambda$ 1608	$1.10\pm 0.06$	2.2257
Fe ii $\lambda$ 1611	$0.11\pm 0.04$	2.2256
C i $\lambda$ 1656	$0.58\pm 0.06$	2.2265
Al ii $\lambda$ 1670	$1.43\pm 0.05$	2.2260
Ni ii $\lambda$ 1709	$0.15\pm 0.03$	2.2259
Ni ii $\lambda$ 1741	$0.22\pm 0.03$	2.2260
Ni ii $\lambda$ 1751	$0.19\pm 0.04$	2.2263
Si ii $\lambda$ 1808	$0.67\pm 0.04$	2.2260
Al iii $\lambda$ 1854	$0.35\pm 0.05$	2.2262
Al iii $\lambda$ 1862	$0.20\pm 0.05$	2.2266
Ti ii $\lambda$ 1910	$0.13\pm 0.04$	2.2261
Zn/Cr ii $\lambda$ 2026	$0.51\pm 0.04$	2.2268
Cr ii $\lambda$ 2056	$0.23\pm 0.04$	2.2259
Zn/Cr ii $\lambda$ 2062	$0.45\pm 0.04$	2.2257
Cr ii $\lambda$ 2066	$0.14\pm 0.04$	2.2265
Fe ii $\lambda$ 2249	$0.17\pm 0.04$	2.2261
Fe ii $\lambda$ 2260	$0.36\pm 0.04$	2.2265
Fe ii $\lambda$ 2344	$1.93\pm 0.05$	2.2261

Full text

1 Introduction

Damped Lyman-$\alpha$ Absorbers (DLAs) have been used to study the cosmic chemical evolution of galaxies for almost 30 years (e.g., Pettini et al., 1990; De Cia et al., 2018). Huge progress has been made and large samples of well-studied systems are now available (e.g., Berg et al., 2016, and references therein). From the beginning of this line of research, it was clear that despite the enormous advantage of the method in obtaining detailed information on the properties of otherwise nearly invisible galaxies, the presence of dust will bias the cosmic chemical evolution history derived this way. This is because DLAs are identified in QSO spectra and the presence of dust inevitably affects the likelihood of identifying QSOs in most types of QSO selection (Fall & Pei, 1989; Pei et al., 1991, 1999).

The study of red and/or dust-obscured QSOs also has a long and intricate history (e.g., Ostriker & Heisler, 1984; Webster et al., 1995; Benn et al., 1998; Warren et al., 2000; Glikman et al., 2006, 2013; Fynbo et al., 2013; Banerji et al., 2015; Hamann et al., 2017; Glikman et al., 2018). Despite some disagreement among studies about the exact fraction of QSOs missed due to reddening, it is clear that a substantial number of QSOs are missed when selecting them only by their optical colours.

A range of studies has attempted to gauge the importance of dust bias in DLA samples. Direct measurements of the effect of dust reddening have typically found very small amounts of excess reddening in samples of QSOs with intervening DLAs (Murphy & Bernet, 2016, and references therein). These studies however used to analyse optically-selected QSO samples. Radio-selected samples (which are free from dust bias as long as the optical follow-up is 100% complete) have been built (Ellison et al., 2001, 2005; Jorgenson et al., 2006), but these are small and there is room for substantial amounts of missing metals due to dust-bias (Jorgenson et al., 2006; Pontzen & Pettini, 2009).

Our group has specifically aimed at finding examples of QSOs reddened by intervening dust-rich DLAs using new QSO-selection techniques (Fynbo et al., 2013; Krogager et al., 2015, 2016a). We have found several such systems, which shows that dust bias exists and that our methods succeeds in locating examples of DLAs that have dropped out of optically-selected QSO samples (Krogager et al., 2016b; Fynbo et al., 2017; Heintz et al., 2018b). Our ambition is to build complete samples of quasars and, within such samples, identify examples of very metal-rich sight-lines, which are interesting for deep spectroscopic follow-up studies. This is challenging because of the overlap with cool stars for point sources which are red in optical bands. This issue can be alleviated by constructing a complete and (in terms of colour) unbiased sample based on astrometric data, specifically from the 2nd Gaia data-release (Gaia-DR2; Gaia Collaboration et al., 2018), removing sources with significant proper motions, which are not extragalactic (Heintz et al., 2015, 2018a).

This Letter is organized as follows. In Sect. 2, we present the spectroscopic observations of a newly-discovered $z=2.60$ QSO with celestial coordinates: RA $=$ 12:18:30.10, Dec $=$ 08:32:15.5 (J2000.0), called GQ 1218$+$0832, reddened by dust in a foreground DLA. In Sect. 3, we present our results on the characteristics of this DLA, constraining its metallicity and induced QSO extinction. We draw our conclusions in Sect. 4.

2 Target selection and observations

We have built our parent QSO catalog using a novel combination (Heintz et al., 2018a) of astrometry from Gaia-DR2 and photometry from the optical Sloan Digital Sky Survey data release 12 (SDSS-DR12, Eisenstein et al., 2011), the near-infrared UKIDSS (Warren et al., 2007), and the mid-infrared all-sky WISE mission (Cutri et al., 2013). We confined the sample to a region of the sky of Galactic latitude $b>60$ deg, to limit the contamination by Galactic sources. We only considered targets listed as point sources in both SDSS and UKIDSS and then only included sources with total proper motions $\mu=\sqrt{\mu_{\rm\alpha*}^{2}+\mu_{\rm\delta}^{2}}$ consistent with zero within 2$\sigma$, i.e., S/N${}_{\mu}=\mu/\mu_{\rm err}<2$ (following Heintz et al., 2018a, see Fig. 1). To be as complete as possible, we initially only imposed the zero-proper motion criterion. However, as a significant number of stars were still present, we imposed the additional near/mid-infrared criteria shown in the bottom panel of Fig. 1 to exclude the large majority of star contaminants. Following Heintz et al. (2018b), we also imposed the optical colour criteria $u-g>1$ and $r-z>0.5$ to specifically target dust-reddened QSOs at $z>2$.

The object discussed in this Letter, designated GQ 1218+0832, was observed as part of a larger spectroscopic campaign. It has $r=20.35$, $g-r=1.07$, and $J-K=0.42$ (AB mag) and is marked with a red square in Fig. 1.

Spectroscopic observations were obtained with the OSIRIS instrument at the Gran Telescopio Canarias (GTC). We gathered spectra using three different grisms in order to better constrain the QSO spectral energy distribution, and intervening metal and H i Ly$\alpha$ lines. The listed spectral resolution values are based on the measured widths of sky-emission lines. The observing log is given in Table 1.

The spectroscopic data were reduced using standard IRAF111IRAF is distributed by the National Optical Astronomy Observatory, which is operated by the Association of Universities for Research in Astronomy (AURA) under a cooperative agreement with the National Science Foundation. procedures. The spectra were flux-calibrated using observations of spectro-photometric standard stars observed on each night. Both the spectra and the photometric data points were corrected for Galactic extinction using the dust maps from Schlafly & Finkbeiner (2011). All reported wavelengths are in vacuum and corrected to the heliocentric rest-frame. To improve the absolute flux calibration and correct for slit-loss, we scaled the observed spectra to the $r$-band photometry from SDSS.

3 Results

We display the GTC R1000B spectrum in Fig. 2, together with the optical and near/mid-infrared photometry from SDSS, UKIDSS, and WISE. We identify an extremely strong DLA (hereafter ES-DLA; Noterdaeme et al., 2014) at a redshift of $z=2.2261$ based on a range of associated metal lines (see Table LABEL:tab:EW).

3.1 Extinction

The spectrum of the background QSO is inconspicuous except for its reddening. The fact that the QSO looks ”normal” makes it well-suited for the determination of the extinction curve of dust in the intervening ES-DLA.

To derive the amount of extinction, we use the composite QSO template from Selsing et al. (2016) including a range of different extinction models. We follow the parametrization of Gordon et al. (2003) and fit the extinguished spectrum assuming a dust composition similar to the Small and Large Magellanic Clouds (SMC and LMC). We find the best-fit to be an LMC-type extinction curve with a visual extinction of $A_{V}=0.82\pm 0.02$ mag, including the 2175 Å dust extinction feature. We measure a bump strength of $A_{\rm bump}=0.38$ mag (following the definition in Gordon et al., 2003).

3.2 Absorption lines

In the inset of Fig. 2, we show a zoom-in on the damped Lyman-$\alpha$ absorption line. To derive the H i column density of the absorber, we included a DLA in the dust-reddened QSO model (using the approximation by Tepper-García, 2006). From the best-fit model, we find $\log N$(H i) = $21.95\pm 0.15$.

We also detect a range of metal lines associated with the DLA, as well as another intervening system at $z=2.247$ seen in C iv and Si iv. In Table LABEL:tab:EW, we list the equivalent width (EW) and redshift measurements drawn from the R2500V and R2500R spectra. The average redshift of the system inferred from the low-ionization lines listed in the table is $z=2.2261\pm 0.0004$.

We have fitted Voigt profiles to the low-ionization metal lines and derived the following lower limits on the abundances: [Zn/H] ¿ $-0.96$, [Si/H] ¿ $-1.23$, [Fe/H] ¿ $-1.89$, and [Cr/H] ¿ $-1.64$, using the Solar photospheric abundances from Asplund et al. (2009). Because of the limited resolution of the spectra, these abundances should be considered cautiously and only treated as lower limits due to saturation. Taken at face value, our measurements indicate an overall metallicity of at least 10% of Solar and significant depletion of refractory elements, e.g., [Fe/Zn$]\sim-1$.

We detect clear absorption signatures of C i but, due to the limited spectral resolution, the transitions from the ground state and fine-structure energy levels are blended. Remarkably, there are consistent features also at the expected locations of CO molecular bands (see Fig. 3). A deeper, higher resolution spectrum is needed to confirm the presence of these lines.

4 Discussion and conclusions

In Fig. 4, we compare the metallicity and H i column density of the presently-studied system to the general population of QSO-DLAs from De Cia et al. (2018), GRB-DLAs from Bolmer et al. (2018), the extreme case of GRB 080607 from Prochaska et al. 2009, and the compiled sample of dusty ($A_{V}>0.1$ mag) DLAs from Heintz et al. (2018b), including the recent discovery by Ranjan et al. (2018). Compared to other QSO-DLAs, the absorption system towards GQ 1218+0832 is remarkable in several ways. Firstly, such a high H i column density is very rare among the general population of QSO absorbers (Bird et al., 2017). Even within the population of extremely strong QSO-DLAs, this system has higher-than-average metal-line strengths, depletion factors, and H i column density (Noterdaeme et al., 2014). Its H i column density is more similar to the column densities found in GRB-DLAs (see also Tanvir et al., 2019). This DLA is also unusual given the presence of C i and possibly CO absorption. The incidence of C i absorbers with $W_{r}({\lambda 1560})>0.4\leavevmode\nobreak\ {\AA}$ is about one-hundred times less than that of QSO-DLAs (Ledoux et al., 2015). The presence of strong C i lines is an indicator of high metallicity, diffuse molecular gas, and Noterdaeme et al. (2018) found that all such systems present H2 and a significant fraction of them exhibit CO at a detectable level. This suggests that CO lines are likely to be detected in this system as well, in agreement with the presence of consistent dips in the spectrum. This system provides further evidence that ES-DLAs (with $N$(H i) ¿ $10^{21.7}$ cm*-2*) lie in the transition regime where H i is being converted into H2 (Schaye, 2001; Noterdaeme et al., 2015; Balashev & Noterdaeme, 2018).

This case, in terms of extinction and molecular content, is the most remarkable example so far of a QSO missed by traditional optical-colour selections due to dust in an intervening galaxy as predicted by Fall & Pei (1989).

How many quasars would be missed if they had a DLA in their line-of-sight with the same level of extinction as in the present case? The answer to this question is complex, as the effect of dust depends both on the redshift of the quasar, the redshift of the DLA, and the colour distribution of stellar populations. In this survey, we find that the fraction of candidate quasars with the same red optical colours as the system presented here is $\sim 1\%$, compared to the full set of quasars identified based on the astrometric and near/mid-infrared selection criteria detailed in Sect. 2. Krogager et al. (2019) have carried out a comprehensive analysis of the effect of dust in DLAs and found that whereas the fraction of quasars dropping out of selections is small, the effect on the derived mass density of neutral hydrogen and metals is much larger, up to a factor of 2 and 5 at $z\sim 2.2$, respectively (see also Pontzen & Pettini, 2009).

There is little doubt that more such systems are yet to be found. An important objective of future work will be to identify them. Another major endeavour will be to use the information of their incidence rate to build up a more complete picture of cosmic chemical evolution.

Acknowledgements.

We thank the anonymous referee for a constructive report that helped improve the manuscript on important points. This work is based on observations carried out on the island of La Palma with the Gran Telescopio Canarias (GTC), installed at the Spanish Observatorio del Roque de los Muchachos belonging to the Instituto de Astrofísica de Canarias. The Cosmic Dawn Center is funded by the DNRF. KEH and PJ acknowledge support from a Project Grant (162948–051) of The Icelandic Research Fund.