A MINI X-RAY SURVEY OF SUB-DAMPED LYMAN-ALPHA ABSORPTION SYSTEMS: SEARCHING FOR ACTIVE GALACTIC NUCLEI FORMED IN PROTOGALAXIES

The galaxy formation process is thought to be hierarchical, with smaller dark matter halos coalescing to form larger ones (e.g., for a review see Springel et al. 2006 and references therein). According to this theory, when the baryonic matter cools down enough to form molecular hydrogen it falls into these halos, resulting in the birth of the first stars. Eventually, clusters of stars may form in the centers of these halos. The merger of these halos may produce the first protogalaxies.

The damped and sub-damped Lyman-alpha (sub-DLA) absorption line systems found in quasar and gamma-ray burst spectra are thought to be produced by intervening protogalaxies or young galaxies (e.g., Wolfe et al. 2005 and references therein). The DLA absorbers have neutral hydrogen column densities log N_H i ⩾ 20.3, while sub-DLAs have 19.0 ⩽ log N_H i < 20.3. Together, DLAs and sub-DLAs constitute most of the neutral hydrogen in galaxies at high redshifts. The nature of the protogalaxies associated with DLAs and sub-DLAs is very uncertain. To understand where DLA/sub-DLAs fit in the big picture of galaxy evolution, it is important that we understand the nature of the galaxies associated with DLA/sub-DLA absorbers.

Recent observations of element abundances show most DLAs to be metal-poor, while a substantial fraction of sub-DLAs appear to be metal-rich, i.e., near-solar or even super-solar (see, e.g., Kulkarni et al. 2005, 2007, 2010; Prochaska et al. 2003, 2006; Péroux et al. 2006; Meiring et al. 2008, 2009). A small fraction (about 5%) of DLAs are also found to be metal-strong (Herbert-Fort et al. 2006). How did these metal-rich absorbers get so enriched >7–10 Gyr ago? Supersolar metallicities are also found in active galactic nuclei (AGNs) at both low and high redshifts using emission-line observations (e.g., Storchi-Bergmann et al. 1998; Hamann et al. 2002; Dietrich et al. 2003; Nagao et al. 2006; Groves et al. 2006). The presence of AGNs with supersolar metallicity at even z ∼ 5 may seem surprising given that galaxies at high redshifts are expected to be more gas-rich, and less enriched compared to low-redshift galaxies (e.g., Kauffmann & Haehnelt 2000). The first major star formation episode in high-redshift metal-rich AGNs appears to have happened <1 Gyr after the big bang. In any case, the observation of high metallicities in AGNs at high and low redshifts, together with the existence of metal-rich sub-DLA absorbers, raises the question of whether some of the metal-rich absorbers may be associated with galaxies possessing AGNs powered by supermassive black holes (SMBHs).

Indeed, the process by which SMBHs form and grow at the centers of galaxies is also very uncertain. Proposed models of SMBH formation and growth (e.g., Kauffmann & Haehnelt 2000; Wyithe & Loeb 2003; Volonteri et al. 2003, 2012; Hopkins et al. 2006; Croton et al. 2006) include: (1) the direct collapse from dense cold gas clouds, (2) contraction of dense cold gas into a supermassive star that after fragmentation or multiple formation events may form black hole binaries that spiral together via gravitational radiation to form a SMBH, (3) runaway growth of a massive black hole by accretion, (4) mergers of black holes, and (5) more realistically, a combination of the above models.

It is unknown at what exact time during a galaxy's formation period the central massive black hole accretes enough gas to become active. A rough estimate of the duration of the active phase can be obtained from the observed fraction of galaxies that contain active nuclei. Specifically, surveys of galaxies indicate that about one in every hundred contain active nuclei, implying AGN lifetimes of the order of 10⁸ yr. It is thought that an AGN's lifetime will depend on the amount of time it takes for the accretion rate to drop below a critical level.

Typical quasars and Seyfert galaxies require accretion rates of the order of a few M_☉ yr⁻¹ to be active, whereas, the accretion rate onto a non-active nucleus such as SgrA* is ∼10⁻⁹–10⁻⁸ M_☉ yr⁻¹ (e.g., Dexter et al. 2009; Shcherbakov et al. 2012). There are various proposed processes that can lead to the reduction of the available gas reservoir that feeds the central source. These include the ejection of gas from the galaxy by powerful AGNs and starburst winds, the formation of stars, and the infall of gas into the central black hole.

In protogalaxies, these processes have not had enough time to deplete the gas reservoir and protogalaxies may contain enough gas to fuel a possible central massive black hole. If this is the case, then we might expect to find a large fraction of protogalaxies with active nuclei. The protogalaxies sampled by metal-rich DLAs and sub-DLAs could thus be associated with SMBHs with undepleted gas reservoirs. Examples of AGNs in local analogues to protogalaxies are the actively accreting massive black hole detected in the dwarf starburst galaxy Henize2-10 (Reines et al. 2011; Reines & Deller 2012) and AGNs detected in local galaxies with properties that are very similar to distant Lyman break galaxies (Jia et al. 2011).

In Section 2, we present our sample of sub-DLAs and the analysis of X-ray observations; in Section 3, we discuss the possible detection of X-rays near background quasars that are known to contain sub-DLAs; and in Section 4, we present a discussion of our results and prospects for expanding the sample and improving the confidence of the detections. Throughout this paper, we adopt a flat Λ cosmology with H₀ = 70 h km s⁻¹ Mpc⁻¹, Ω_Λ = 0.7, and Ω_M = 0.3.

To test the plausibility of sub-DLAs being associated with SMBHs, we performed an exploratory search for X-ray emission associated with a subset of the sub-DLAs presented in a survey of z > 4 quasars (Guimarães et al. 2009). We also included five sub-DLAs detected in additional studies (Péroux et al. 2011; Prochaska et al. 2005; Som et al. 2013).

These sub-DLAs were chosen because X-ray spectra of their background quasars have been obtained with the Chandra X-ray Observatory (hereafter Chandra) and are available in the Chandra X-ray Center (CXC) archives. Our search was restricted to Chandra observations since only Chandra has a spatial resolution of ∼05; high spatial resolution is desirable to resolve X-ray emission from the candidate protogalaxies associated with the sub-DLAs. As a result of this search, we found 21 quasars with sub-DLAs that had been observed with the Advanced CCD Imaging Spectrometer (ACIS; Garmire et al. 2003) on board Chandra. A log of the X-ray observations that includes observation dates, observational identification numbers, exposure times, ACIS frame time, and the observed 0.2–10 keV counts is presented in Table 1.

The Chandra observations of the sub-DLAs of our mini-sample were analyzed using the standard software CIAO 4.5 provided by the CXC. We used standard CXC threads to screen the data for status, grade, and time intervals of acceptable aspect solution and background levels. To improve the spatial resolution, we employed the sub-pixel resolution technique developed by Li et al. (2004) and incorporated via the Energy-Dependent Subpixel Event Repositioning algorithm into the tool acis_process_events of CIAO 4.5. For comparison, we also employed the sub-pixel resolution technique developed by Tsunemi et al. (2001). In 17 out of 21 cases, the background quasar is detected with Chandra. In 6 out of the 21 Chandra observations, we detected X-ray emission (resolved from the quasar itself) within ∼1'' of the background quasar that is known to contain a sub-DLA in its optical/UV spectrum. One exciting possibility is that this X-ray emission originates from an AGN near the center of a protogalaxy. In Figure 1, we show the Chandra images of the fields around quasars PSS 0121+0347, PSS 0133+0400, PSS 0955+5940, SDSS J095744.46+330820.7, Q 1323−0021, and PSS 2322+1944, for which we detected X-ray emission nearby, possibly associated with the intervening systems. The Chandra-ACIS point-spread function (PSF) has a FWHM of ∼05, and is indicated by the solid circles centered on the optical positions of the quasars. We detect X-rays centered on the background quasars, however, we also detect a significant number of counts asymmetrically distributed and located about 05–1'' from the optical positions of the quasars. Regions containing ≳3 events with a distance of greater than 0.5 arcsec from the background quasar and clustered within a 0.5 arcsec radius circle were considered to be marginally detected X-ray sources. The dashed circles are centered on the marginally detected X-ray emission within ∼1'' of the background quasars. In Table 2, we list the properties of the background quasars and their sub-DLAs.

Zoom In Zoom Out Reset image size

Table 2. Properties of Background Quasars and Sub-DLAs

Object Name	zem	zabs	log NG, Ha	log Nabs,H ib	Lqsoc	LsubDLAd
			(cm−2)	(cm−2)	(1045 h−2 erg s−1)	(1044 h−2 erg s−1)
PSS 0121+0347	4.130	2.977	20.544	19.50 ± 0.15	7.1(8.3)	2.1
PSS 0133+0400e	4.154	3.995	20.491	20.15 ± 0.10	3.2(4.2)	2.5
PSS 0209+0517e	4.174	3.862	20.668	20.30 ± 0.10	2.2(2.9)	<4.0
SDSS J074749.74+443417.1	4.432	3.139	20.719	20.00 ± 0.20	0.84(1.7)	<4.0
SDSS075618+410408	5.09	4.360	20.685	20.15 ± 0.10	1.9(1.8)	<3.8
PSS 0955+5940	4.336	3.843	20.134	20.00 ± 0.15	2.1(1.7)	4.2
		4.044		20.10 ± 0.15		4.7
SDSS J095744.46+330820.7	4.227	3.043	20.179	19.65 ± 0.15	2.6(2.4)	3.6
		3.364		19.70 ± 0.15		4.4
		3.900		19.50 ± 0.10		5.8
PSS 1057+4555	4.137	2.909	20.061	20.05 ± 0.10	5.5(6.4)	<5.1
		3.058		19.80 ± 0.15		<5.6
		3.164		19.50 ± 0.20		<6.0
		3.317		20.15 ± 0.10		<6.6
SDSS J122556.61+003535.1	1.226	0.773	20.283	21.38 ± 0.12	0.18(0.37)	<0.36
SDSS J124942.12+334953.8	4.897	4.572	20.097	19.80 ± 0.10	4.3(4.2)	<11
SDSS J131743.12+353131.8	4.381	3.461	19.998	19.90 ± 0.05	1.1(2.1)	<9.3
Q1323-0021	1.388	0.716	20.270	20.21 ± 0.20	2.2(2.2)	0.5(S1), 0.5(S2)
SDSS J132512.49+112329.7	4.400	3.723	20.281	19.50 ± 0.20	4.9(4.1)	<5.7
		4.133		19.50 ± 0.20		<6.9
SDSS J132611.85+074358.4e	4.123	2.919	20.303	19.95 ± 0.10	6.7(6.9)	<2.5
		3.425		19.90 ± 0.15		<3.5
SDSS J132611.85+074358.4f	4.123	2.919	20.303	19.95 ± 0.10	6.5(5.8)	<3.9
		3.425		19.90 ± 0.15		<5.3
PSS 1435+3057	4.350	3.267	20.068	20.05 ± 0.10	<1.3	<7.5
		3.516		20.20 ± 0.10		<8.7
		3.778		19.85 ± 0.10		<10.0
PSS 1646+5514	4.084	2.932	20.382	19.50 ± 0.10	0.95(0.78)	<5.1
	4.084	4.029		19.80 ± 0.15		<9.3
SDSS 173744+582829	4.940	4.152	20.550	19.85 ± 0.15	<8.1	<66.0
SDSS J2123-0050	2.261	2.058	20.667	19.35 ± 0.10	1.2(1.4)	<2.2
PSS 2322+1944g	4.118	2.888	20.655	19.95 ± 0.10	$2.0\mu ^{-1}_{X}(1.8\mu ^{-1}_{X})$	<3.0
		2.975		19.80 ± 0.10		<3.2
PSS 2322+1944h	4.118	2.888	20.655	19.95 ± 0.10	$2.7\mu ^{-1}_{X}(2.6\mu ^{-1}_{X})$	1.1(S1), 0.89(S2)
		2.975		19.80 ± 0.10		1.2(S1), 0.96(S2)
PSS J2344+0342	4.239	3.884	20.735	19.80 ± 0.10	<0.9	<0.8
SDSS J235253.51-002850.4	1.628	0.873	20.543	19.18 ± 0.10	<0.09	<0.2
		1.032		19.81 ± 0.13		<0.3
		1.247		19.60 ± 0.25		<0.5

Notes. ^aTotal Galactic hydrogen column density from Dickey & Lockman (1990). ^bH i column density. ^c0.2–10 keV luminosity of background quasar assuming a spectral model consisting of a power-law with a photon index fixed at Γ = 1.8 and Galactic absorption. The values listed in parentheses represent 0.2–10 keV luminosities obtained from spectral fits assuming the same model where the photon index was allowed to vary. ^dUpper limit of the 0.2–10 keV luminosity of the sub-DLA assuming a 5 count detection within a circle of radius 05. In cases with detected nearby sources, we list the observed 0.2–10 keV luminosities of the subDLA's. In cases where two sources are detected near the background quasar, the luminosity of each source is followed by its label shown in Figure 1. ^eThe estimated luminosities correspond to the 2002 January 10 observation of SDSS J132611.85+074358.4. ^fThe estimated luminosities correspond to the 2011 March 7 observation of SDSS J132611.85+074358.4. ^gThe estimated luminosities correspond to the 2002 August 20 observation of PSS 2322+1944. We have assumed that the model-predicted magnification in the optical band of μ_opt ≈ 4.7 (Riechers et al. 2008) is equal to the X-ray magnification, μ_X. ^hThe estimated luminosities correspond to the 2005 August 10 observation of PSS 2322+1944.

Download table as:  ASCIITypeset image

The spatial distributions of the excess X-ray counts near the quasars are consistent with being produced by point sources in the sense that these counts are clustered within circles of radii of less than 05 (ACIS PSF FWHM ∼ 05). X-ray emission from the host galaxy is too weak to be detected at the redshift of the quasar. PSS 2322+1944 is a gravitationally lensed quasar with the second lensed image represented by a circle near label B in Figure 1. We find a marginal detection of X-ray counts near the location of the second lensed image. We note that the background in these Chandra observations is very low, as is evident from the very few counts detected away from these sources. In Figure 2, we also show the observed surface brightness profiles centered on the quasars Q 1323-0021 and PSS 2322+1944 and compare them to simulated ones for a point source. The other four candidates have too few counts to produce surface brightness profiles. We employed the MARX tool to simulate PSFs. The observed surface brightness profiles deviate from those simulated for a single point source and the presence of additional sources within 1'' of the quasar is consistent with our image analysis. We note that these quasars were centered on-axis at the aim-point of ACIS S3.

Zoom In Zoom Out Reset image size

In the case of the moderate signal-to-noise ratio (S/N) image of Q 1323-0021, the detected counts were sufficient for a two-dimensional fit to both the background quasar and nearby X-ray sources with simulated PSFs. The image was binned with a bin size of 00246 bins (compared to the 0491 ACIS pixel scale) and fit by minimizing the C statistic (Cash 1979) between the observed and model images. In Figure 3, we show the best-fit PSF model of the Chandra observation of Q 1323-0021. The positions of the PSFs were left as free parameters in the fit and the best-fit values of the positions of the nearby sources, labeled as S1 and S2, are listed in Table 3. S1 and S2 lie at angular separations of 047 and 043,  respectively, from the background quasar. Neither S1 nor S2 is seen in the K'-band adaptive optics (AO) image with FWHM 008 obtained by Chun et al. (2010). However, we note that a massive candidate galaxy was reported by Chun et al. (2010) to be located 125 away from the quasar. It is possible that S1 and S2 are not bright enough in near-IR, or that they are lost in the artifacts of the AO PSF in the image of Chun et al. (2010). We note that the intervening absorber of Q 1323-0021 is one of the highest metallicity absorbers known. Specifically, Péroux et al. (2006) reported metallicities of [Zn/H] = +0.61 ± 0.20 and [Fe/H] = −0.51 ± 0.20.

Zoom In Zoom Out Reset image size

Our exploratory mini-survey of sub-DLAs contained mostly short snapshot Chandra observations with exposure times ranging between 2 and 19 ks. In 17 of the 21 Chandra observations of the sub-DLA fields in our sample, we detect X-rays from the background quasars that are known to contain the sub-DLAs. In the Chandra observations of six quasars from our sample, we also have marginal detections of X-ray sources within 1'' of the background quasars. These X-ray sources may be associated with the sub-DLAs. We briefly discuss the possible origins of these nearby X-ray sources.

• 1.  
  X-ray emission from a nearby X-ray source instead of the sub-DLA. We estimated the probability of detecting a background X-ray source within a circle with a radius of 1'' centered on the quasar by scaling the cumulative number counts per square degree found in the 4 Ms Chandra Deep Field-South survey (Lehmer et al. 2012) by the ratio of the areas. In Table 3, we list the 0.5–8 keV fluxes of the candidate sub-DLA sources and the probability of finding by chance a second source with a flux greater than or equal to the one detected. We conclude that it is unlikely that our six candidate X-ray sources are random background X-ray sources in the sky. We are being conservative in these estimates since the probability of detecting a faint source near a bright quasar is even smaller than our estimated values.
• 2.  
  X-ray emission from normal late-type galaxies instead of the sub-DLA. X-rays from normal galaxies are thought to be produced by supernovae, supernova remnants, stellar outflows of hot gas, the hot interstellar medium, high and low-mass X-ray binaries, and young stars (e.g., Fabbiano 2006). An analysis of normal late-type galaxies detected in the Chandra Deep Field survey by Lehmer et al. (2012) indicates that their 0.5–8 keV luminosities increase on average from ∼3 × 10³⁹ h⁻² erg s⁻¹ at z = 0.1 to ∼2 × 10⁴² h⁻² erg s⁻¹ at z = 1.4 and their mean X-ray luminosity to star formation rate (SFR) is found to be constant within this redshift span. An analysis of normal galaxies in the Great Origins Deep Survey fields by Ptak et al. (2007) shows that their X-ray evolution can be expressed as L*(z) = (1 + z)^pL*(z = 0), where p = 1.6 for early-type and p = 2.3 for late-type galaxies, which is consistent with what is found in the independent survey of Lehmer et al. (2012). Assuming that the nearby X-ray sources detected in our mini-survey are located at the redshifts of the sub-DLAs, their estimated X-ray luminosities (see Figure 4) are significantly larger than what would be estimated if they were originating from normal late-type galaxies at those redshifts.
• 3.  
  X-ray emission from AGNs associated with the sub-DLAs. One exciting possibility is that the X-ray emission of the nearby sources of our mini-survey originates from an AGN located near the center of the protogalaxy. Assuming that the X-ray emission of the nearby sources originates at the redshifts of the intervening sub-DLAs, we estimate their 0.2–10 keV luminosities to range between 0.8 × 10⁴⁴ h⁻² and 4.2 × 10⁴⁴ h⁻² erg s⁻¹, which is consistent with that of a low-luminosity AGN. For estimating the luminosities, we assumed a power-law model with a photon index of Γ = 1.8 modified by Galactic absorption. The angular separations of these X-ray sources from the background quasars correspond to projected linear separations at the redshift of the sub-DLA in the range of 3–7 h⁻¹ kpc. These separations are consistent with the expected impact parameters of the background quasars from the center of the sub-DLAs at the redshift of the foreground absorbers. In Table 3, we provide several properties of the candidate X-ray sources, assuming they are at the redshift of the sub-DLAs.
• 4.  
  X-ray emission from an additional image produced by gravitational lensing of the background quasar. The probability of a quasar being gravitationally lensed into multiple images depends primarily on the comoving number density of lenses and the lensing cross section of each lens and is of the order of 0.1% (e.g., Turner et al. 1984; Comerford et al. 2002). The fraction of background quasars in our sample with a detected candidate nearby X-ray source (6/21) is significantly larger than the expected number of gravitationally lensed quasars in our sample. Therefore, the probability that any of the quasars in our sample is gravitationally lensed (in addition to PSS 2322, which is a known gravitationally lensed quasar) is negligible. We note that gravitationally lensed quasars are often included in surveys of intervening absorption systems because of their advantage of having multiple lines of sight through intervening absorbers. We searched catalogs of known gravitationally lensed systems (e.g., Master Lens database of the Orphan Lenses Project) and found that none of our candidate X-ray sources are listed as lensed quasars, other than PSS 2322.
• 5.  
  X-ray emission from intense star formation associated with the sub-DLAs or the host galaxies of the background quasars. Lehmer et al. (2010) determine the relation between SFR and 2–10 keV luminosity (L$^{{\rm Gal}}_{\rm X}$) for a sample of galaxies composed of normal galaxies, luminous infrared galaxies, and ultraluminous infrared galaxies to be log(L$^{{\rm Gal}}_{\rm X}$) = α+βlog(SFR), where α = 39.49 ± 0.21, β = 0.74 ± 0.12, and SFR is in units of M_☉ yr⁻¹. We find that the observed 2–10 keV luminosities (listed in Table 3) of the possible X-ray sources near the background quasars are several orders of magnitude larger than the inferred L$^{{\rm Gal}}_{\rm X}$ values, assuming that the nearby X-ray emission originates from intense star formation associated with the sub-DLAs or the host galaxies of the background quasars.

Zoom In Zoom Out Reset image size

Follow-up observations with longer exposures are required to confirm or refute the discovery of a possible new class of X-ray sources, in particular, those that may be associated with active protogalactic nuclei. Specifically, spectral and variability analyses of deeper X-ray observations of these sources will increase the S/N of these detections and provide insight into their nature. Follow-up deep spectroscopic and imaging optical and UV observations may also provide the redshifts and images of possible galaxies at the locations of the X-ray sources.

We acknowledge financial support from NASA via the Smithsonian Institution grant SAO AR0-11019X. V.P.K. acknowledges partial support from the National Science Foundation grant AST/1108830.