Draft version September 24, 2018 Preprint typeset using L A T E X style emulateapj v. 5/2/11 THE KECK + MAGELLAN SURVEY FOR LYMAN LIMIT ABSORPTION III: SAMPLE DEFINITION AND COLUMN DENSITY MEASUREMENTS J. Xavier Prochaska 1 , John M. O’Meara 2 , Michele Fumagalli 3,4 , Rebecca A. Bernstein 4 , Scott M. Burles 6 Draft version September 24, 2018 ABSTRACT We present an absorption-line survey of optically thick gas clouds – Lyman Limit Systems (LLSs) – observed at high dispersion with spectrometers on the Keck and Magellan telescopes. We measure column densities of neutral hydrogen N HI and associated metal-line transitions for 157 LLSs at z LLS = 1.76 - 4.39 restricted to 10 17.3 cm -2 ≤ N HI < 10 20.3 cm -2 . An empirical analysis of ionic ratios indicates an increasing ionization state of the gas with decreasing N HI and that the majority of LLSs are highly ionized, confirming previous expectations. The Si + /H 0 ratio spans nearly four orders-of- magnitude, implying a large dispersion in the gas metallicity. Fewer than 5% of these LLSs have no positive detection of a metal transition; by z ∼ 3, nearly all gas that is dense enough to exhibit a very high Lyman limit opacity has previously been polluted by heavy elements. We add new measurements to the small subset of LLS (≈ 5 - 10%) that may have super-solar abundances. High Si + /Fe + ratios suggest an α-enhanced medium whereas the Si + /C + ratios do not exhibit the super-solar enhancement inferred previously for the Lyα forest. Subject headings: absorption lines – intergalactic medium – Lyman limit systems 1. INTRODUCTION As a packet of ionizing radiation (hν ≥ 1 Ryd) tra- verses the universe, it has a high probability of encoun- tering a slab of optically thick, H I gas. For sources in the z ∼ 4 universe the mean free path is only ≈ 30 Mpc (physical; Worseck et al. 2014), i.e. less than 2% of the event horizon. Observationally, researchers refer to this optically thick gas as Lyman limit systems (LLSs) owing to their unmistakable signature of continuum opacity at the Lyman limit (≈ 912 ˚ A) in the system restframe. A fraction of this gas lies within the dense, neutral interstel- lar medium (ISM) of galaxies, yet the majority of opac- ity must arise from gas outside the ISM (e.g. Fumagalli et al. 2011b; Ribaudo et al. 2011). Indeed, the interplay between galaxies and the LLS is a highly active area of research which includes studies of the so-called circum- galactic medium (CGM; e.g. Steidel et al. 2010; Werk et al. 2013; Prochaska et al. 2014a). For many decades, LLS have been surveyed in quasar spectra (e.g. Tytler 1982; Sargent et al. 1989; Storrie- Lombardi et al. 1994), albeit often from heterogeneous samples. These works established the high incidence of LLSs which evolves rapidly with redshift. With the realization of massive spectral datasets, a renaissance of LLS surveys has followed yielding statistically ro- bust measurements from homogenous and well-selected quasar samples (Prochaska et al. 2010; Songaila & Cowie 1 Department of Astronomy and Astrophysics, UCO/Lick Ob- servatory, University of California, 1156 High Street, Santa Cruz, CA 95064, USA 2 Department of Chemistry and Physics, Saint Michael’s Col- lege. One Winooski Park, Colchester, VT 05439, USA 3 Institute for Computational Cosmology, Department of Physics, Durham University, South Road, Durham, DH1 3LE, UK 4 Observatories of the Carnegie Institution for Science, 813 Santa Barbara Street, Pasadena, CA 91101, USA 6 Cutler Group, LP., 101 Montgomery St., San Francisco, CA 94104, USA 2010; Ribaudo et al. 2011; O’Meara et al. 2013; Fuma- galli et al. 2013). Analysis of these hundreds of systems reveals an incidence of approximately 1.2 systems per unit redshift at z ∼ 3 that evolves steeply with redshift ‘(z) ∝ (1+z) 1.5 for z ≈ 1 -5 (Ribaudo et al. 2011; Fuma- galli et al. 2013). With these same spectra, researchers have further measured the mean free path of ionizing ra- diation (λ 912 mfp ; Prochaska et al. 2009; O’Meara et al. 2013; Fumagalli et al. 2013; Worseck et al. 2014), which sets the intensity and shape of the extragalactic UV background. Following the redshift evolution of the LLS incidence, λ 912 mfp also evolves steeply with the expanding universe, implying a more highly ionized universe with advancing cosmic time (Worseck et al. 2014). The preponderance of LLSs bespeaks a major reservoir of baryons. In particular, given the apparent paucity of heavy elements within galaxies (e.g. Bouch´ e et al. 2006; Peeples et al. 2014), the LLSs may present the dominant reservoir of metals in the universe (e.g. Prochaska et al. 2006). However, a precise calculation of the heavy ele- ments within LLSs and their contribution to the cosmic budget has not yet been achieved. Despite our success at surveying hundreds of LLSs, there have been few studies resolving their physical properties and these have gener- ally examined a few individual cases (e.g. Steidel 1990; Prochaska 1999) or composite spectra (Fumagalli et al. 2013). This reflects both the challenges related to data acquisition and analysis together with a historical focus in the community towards the ISM of galaxies (probed by DLAs) and the more diffuse intergalactic medium (IGM). At z> 2, a few works have examined the set of LLSs with high H I column density (N HI ≥ 10 19 cm -2 ), gener- ally termed the super-LLSs or sub-damped Lyα systems. Their N HI frequency distribution f (N HI ,X) and chemi- cal abundances have been analyzed from a modestly sized sample (Dessauges-Zavadsky et al. 2003; P´ eroux et al. 2005; O’Meara et al. 2007; Zafar et al. 2013; Som et al. 2013). Ignoring ionization corrections, which may not arXiv:1506.08863v1 [astro-ph.CO] 29 Jun 2015
19
Embed
z arXiv:1506.08863v1 [astro-ph.CO] 29 Jun 2015 · 2018-09-24 · lege. One Winooski Park, Colchester, VT 05439, USA 3 Institute for Computational Cosmology, Department of Physics,
This document is posted to help you gain knowledge. Please leave a comment to let me know what you think about it! Share it to your friends and learn new things together.
Transcript
Draft version September 24, 2018Preprint typeset using LATEX style emulateapj v. 5/2/11
THE KECK + MAGELLAN SURVEY FOR LYMAN LIMIT ABSORPTION III: SAMPLE DEFINITION ANDCOLUMN DENSITY MEASUREMENTS
J. Xavier Prochaska1, John M. O’Meara2, Michele Fumagalli3,4, Rebecca A. Bernstein4, Scott M. Burles6
Draft version September 24, 2018
ABSTRACT
We present an absorption-line survey of optically thick gas clouds – Lyman Limit Systems (LLSs)– observed at high dispersion with spectrometers on the Keck and Magellan telescopes. We measurecolumn densities of neutral hydrogen NHI and associated metal-line transitions for 157 LLSs at zLLS =1.76 − 4.39 restricted to 1017.3 cm−2 ≤ NHI < 1020.3 cm−2. An empirical analysis of ionic ratiosindicates an increasing ionization state of the gas with decreasing NHI and that the majority of LLSsare highly ionized, confirming previous expectations. The Si+/H0 ratio spans nearly four orders-of-magnitude, implying a large dispersion in the gas metallicity. Fewer than 5% of these LLSs have nopositive detection of a metal transition; by z ∼ 3, nearly all gas that is dense enough to exhibit a veryhigh Lyman limit opacity has previously been polluted by heavy elements. We add new measurementsto the small subset of LLS (≈ 5− 10%) that may have super-solar abundances. High Si+/Fe+ ratiossuggest an α-enhanced medium whereas the Si+/C+ ratios do not exhibit the super-solar enhancementinferred previously for the Lyα forest.Subject headings: absorption lines – intergalactic medium – Lyman limit systems
1. INTRODUCTION
As a packet of ionizing radiation (hν ≥ 1 Ryd) tra-verses the universe, it has a high probability of encoun-tering a slab of optically thick, H I gas. For sources inthe z ∼ 4 universe the mean free path is only ≈ 30 Mpc(physical; Worseck et al. 2014), i.e. less than 2% of theevent horizon. Observationally, researchers refer to thisoptically thick gas as Lyman limit systems (LLSs) owingto their unmistakable signature of continuum opacity atthe Lyman limit (≈ 912A) in the system restframe. Afraction of this gas lies within the dense, neutral interstel-lar medium (ISM) of galaxies, yet the majority of opac-ity must arise from gas outside the ISM (e.g. Fumagalliet al. 2011b; Ribaudo et al. 2011). Indeed, the interplaybetween galaxies and the LLS is a highly active area ofresearch which includes studies of the so-called circum-galactic medium (CGM; e.g. Steidel et al. 2010; Werket al. 2013; Prochaska et al. 2014a).
For many decades, LLS have been surveyed in quasarspectra (e.g. Tytler 1982; Sargent et al. 1989; Storrie-Lombardi et al. 1994), albeit often from heterogeneoussamples. These works established the high incidenceof LLSs which evolves rapidly with redshift. With therealization of massive spectral datasets, a renaissanceof LLS surveys has followed yielding statistically ro-bust measurements from homogenous and well-selectedquasar samples (Prochaska et al. 2010; Songaila & Cowie
1 Department of Astronomy and Astrophysics, UCO/Lick Ob-servatory, University of California, 1156 High Street, Santa Cruz,CA 95064, USA
2 Department of Chemistry and Physics, Saint Michael’s Col-lege. One Winooski Park, Colchester, VT 05439, USA
3 Institute for Computational Cosmology, Department ofPhysics, Durham University, South Road, Durham, DH1 3LE,UK
4 Observatories of the Carnegie Institution for Science, 813Santa Barbara Street, Pasadena, CA 91101, USA
6 Cutler Group, LP., 101 Montgomery St., San Francisco, CA94104, USA
2010; Ribaudo et al. 2011; O’Meara et al. 2013; Fuma-galli et al. 2013). Analysis of these hundreds of systemsreveals an incidence of approximately 1.2 systems perunit redshift at z ∼ 3 that evolves steeply with redshift`(z) ∝ (1+z)1.5 for z ≈ 1−5 (Ribaudo et al. 2011; Fuma-galli et al. 2013). With these same spectra, researchershave further measured the mean free path of ionizing ra-diation (λ912
mfp; Prochaska et al. 2009; O’Meara et al. 2013;
Fumagalli et al. 2013; Worseck et al. 2014), which sets theintensity and shape of the extragalactic UV background.Following the redshift evolution of the LLS incidence,λ912
mfp also evolves steeply with the expanding universe,implying a more highly ionized universe with advancingcosmic time (Worseck et al. 2014).
The preponderance of LLSs bespeaks a major reservoirof baryons. In particular, given the apparent paucity ofheavy elements within galaxies (e.g. Bouche et al. 2006;Peeples et al. 2014), the LLSs may present the dominantreservoir of metals in the universe (e.g. Prochaska et al.2006). However, a precise calculation of the heavy ele-ments within LLSs and their contribution to the cosmicbudget has not yet been achieved. Despite our success atsurveying hundreds of LLSs, there have been few studiesresolving their physical properties and these have gener-ally examined a few individual cases (e.g. Steidel 1990;Prochaska 1999) or composite spectra (Fumagalli et al.2013). This reflects both the challenges related to dataacquisition and analysis together with a historical focusin the community towards the ISM of galaxies (probed byDLAs) and the more diffuse intergalactic medium (IGM).
At z > 2, a few works have examined the set of LLSswith high H I column density (NHI ≥ 1019 cm−2), gener-ally termed the super-LLSs or sub-damped Lyα systems.Their NHI frequency distribution f(NHI, X) and chemi-cal abundances have been analyzed from a modestly sizedsample (Dessauges-Zavadsky et al. 2003; Peroux et al.2005; O’Meara et al. 2007; Zafar et al. 2013; Som et al.2013). Ignoring ionization corrections, which may not
arX
iv:1
506.
0886
3v1
[as
tro-
ph.C
O]
29
Jun
2015
2
be justified, these SLLSs exhibit metallicities of approx-imately 1/10 solar, comparable to the enrichment levelof the higher-NHI, damped Lyα systems (DLAs; Rafelskiet al. 2012). In addition, a few LLSs have received spe-cial attention owing to their peculiar metal-enrichment(Prochaska et al. 2006; Fumagalli et al. 2011a) and/orthe detection of D for studies of Big Bang Nucleosyn-thesis (e.g. Burles & Tytler 1998; O’Meara et al. 2006).Most recently, a sample of 15 LLSs has been surveyedfor highly ionized O VI absorption (Lehner et al. 2014),which is present at a high rate. A comprehensive studyof the absorption-line properties of the LLSs at high red-shift, however, has not yet been performed.
Scientifically, we have two primary motivations to sur-vey the LLSs at z > 2. First and foremost, we aim todissect the physical nature of the gas that dominates theopacity to ionizing radiation in the universe. One sus-pects that these LLSs trace a diverse set of overdensestructures ranging from galactic gas to the densest fila-ments of the cosmic web. Such diversity may manifestin an wide distribution of observed properties (e.g. metalenrichment, ionization state, kinematics). Second, mod-ern theories of galaxy formation predict that the gas fu-eling star formation accretes onto galaxies in cool, densestreams (e.g. Keres et al. 2005; Dekel et al. 2009). Ra-diative transfer analysis of hydrodynamic simulations ofthis process predict a relatively high cross-section of op-tically thick gas around galaxies (e.g. Faucher-Giguere& Keres 2011; Fumagalli et al. 2011b, 2014; Faucher-Giguere et al. 2014). Indeed, an optically thick CGM en-velops the massive galaxies hosting z ∼ 2 quasars (Hen-nawi & Prochaska 2007; Prochaska et al. 2013), LLSs areobserved near Lyman break galaxies (Rudie et al. 2012),and such gas persists around present-day L∗ galaxies (e.g.Chen et al. 2010; Werk et al. 2013). The latter has in-spired, in part, surveys of the LLSs at z < 1 with ul-traviolet spectroscopy (e.g. Ribaudo et al. 2011; Lehneret al. 2013).
Thus motivated, we have obtained a large dataset ofhigh-dispersion spectroscopy on z > 3 quasars at theKeck and Las Campanas Observatories. We have sup-plemented this program with additional spectra obtainedto survey the damped Lyα systems (e.g. Prochaska et al.2007; Berg et al. 2014) and the intergalactic medium (e.g.Faucher-Giguere et al. 2008). In this paper, we presentthe comprehensive dataset of column density measure-ments on over 150 LLSs. Future manuscripts will ex-amine the metallicity, chemical abundances, kinematics,and ionization state of this gas. This manuscript is out-lined as follows: Section 2 describes the dataset analyzedincluding a summary of the observations and proceduresfor generating calibrated spectra. We define an LLS inSection 3 and detail the procedures followed to estimatethe H I column densities in Section 4. Section 5 presentsmeasurements of the ionic column densities and the pri-mary results of an empirical assessment of these data aregiven in Section 6. A summary in Section 7 concludesthe paper.
2. DATA
This section describes the steps taken to generate alarge dataset of high-dispersion, calibrated spectra ofhigh redshift LLSs.
2.1. Our Survey
The sample presented in this manuscript is intendedto be a nearly, all-inclusive set of LLSs discovered in thehigh-dispersion (echelle or echellette; R > 5, 000) spec-tra that we have gathered at the Keck and Magellantelescopes. Regarding Keck, we have examined all ofthe data obtained by Principal Investigators (PIs) A.M.Wolfe and J.X. Prochaska at the W.M. Keck Observa-tory through April 2012, and from PIs Burles, O’Meara,Bernstein, and Fumagalli at Magellan through July 2012.We also include the Keck spectra analyzed by Penpraseet al. (2010).
Each spectrum was visually inspected for the presenceof damped Lyα absorption and/or a continuum break atwavelengths λ < 912A in the quasar rest-frame. Thecomplex combination of spectral S/N, wavelength cover-age, and quasar emission redshift zem leads to a varyingsensitivity to an LLS. No attempt is made here to definea statistical sample, e.g. to assess the random incidenceof LLSs nor their NHI frequency distribution f(NHI, X).We refer the reader to previous manuscripts on this topic(Prochaska et al. 2010; Fumagalli et al. 2013). Becauseour selection is based solely on H I absorption, however,we believe the sample is largely unbiased with respectto other properties of the gas, e.g. metal-line absorption,kinematics, ionization state.
The sample was limited during the survey by: (1)generally ignoring LLSs with absorption redshifts within3000 km s−1 of the reported quasar redshift zem, so-calledproximate LLS or PLLS; and (2) generally ignoring LLSswith NHI < 1017.3 cm−2, especially when the S/N waspoor near the Lyman limit. We note further that manyof the Keck spectra were obtained to study damped Lyαsystems (DLAs) at z > 2 (e.g. Prochaska et al. 2001,2007; Rafelski et al. 2012; Neeleman et al. 2013; Berget al. 2014). We have ignored systems targeted as DLAsand also absorbers within ≈ 1500km s−1 of these DLAsbecause the DLA system complicates analysis of the H ILyman series and metal-line transitions of any nearbyLLS. In § 3, we offer a strict definition for an LLS todefine our sample of 157 systems.
2.2. Observations
We present data obtained at the W.M. Keck and LasCampanas Observatories using the twin 10 m Keck I andKeck II telescopes and the twin 6.5 m Baade and Claytelescopes. Altogether, we used four spectrometers: (1)the High Resolution Echelle Spectrometer (HIRES; Vogtet al. 1994); (2) the Echellette Spectrograph and Imager(ESI; Sheinis et al. 2002); (3) the Magellan Inamori Ky-ocera Echelle (MIKE; Bernstein et al. 2003); and (4) theMagellan Echellette Spectrograph (MagE; Marshall et al.2008).
The MagE spectra were presented in Fumagalli et al.(2013) and we refer the reader to that manuscript fordetails on the observations and data reduction. Similarlythe ESI observations have been published previously in aseries of papers (Prochaska et al. 2003b, 2007; O’Mearaet al. 2007; Rafelski et al. 2012).
Observing logs for the HIRES and MIKE spectra areprovided in Tables 1 and 2. A significant fraction of thesedata have been analyzed previously (e.g. O’Meara et al.2007; Faucher-Giguere et al. 2008; Neeleman et al. 2013),
3
TABLE 1JOURNAL OF HIRES OBSERVATIONS
QSO Alt. Name RA DEC r/V a zem Date Slitb Mode Exp S/Nc
a Magnitude from the SDSS database (r-band) or as listed in the SIMBAD Astronomical Database (V -band).b Decker employed.c Median signal-to-noise per 3.0km s−1 pixel in the quasar continuum at ≈ 5000A for the old HIRES detector (HIRESO), ≈ 3400/4000A forHIRESb, and ≈ 6000/8000A for HIRESr. An “X” indicates no wavelength coverage or that the S/N was compromised by an LLS.
4
TABLE 2JOURNAL OF MIKE OBSERVATIONS
QSO Alt. Name RA DEC r/V a zem Date Slitb Exp S/Ncblue S/Ndred(J2000) (J2000) (mag) (UT) (′′) (s) (pix−1) (pix−1)
a Magnitude from the SDSS database (r-band) or as listed in the SIMBAD Astronomical Database (V -band).b Slit width employed. For the blue (red) side, a 1′′ slit yields a FWHM resolution of 10.7 (13.6) km s−1 for a source that fills the slit.c Median signal-to-noise per 3.0km s−1 pixel in the quasar continuum at ≈ 3400/4000A. An X designates no flux.d Median signal-to-noise per 4.2km s−1 pixel in the quasar continuum at ≈ 6000/8000A. An X designates no flux.
5
0.0
0.2
0.4
0.6
0.8
1.0
Norm
alize
d Fl
uxJ010355.3-300946 z=2.908
Lyα
1000 0 1000Relative Velocity (km/s)
0.0
0.2
0.4
0.6
0.8
1.0
Norm
alize
d Fl
ux
Lyβ
0.0
0.2
0.4
0.6
0.8
1.0
Norm
alize
d Fl
ux
J034402.85-065300.6 z=3.843
Lyα
1000 0 1000Relative Velocity (km/s)
0.0
0.2
0.4
0.6
0.8
1.0No
rmali
zed
Flux
Lyβ
0.0
0.2
0.4
0.6
0.8
1.0
Norm
alize
d Fl
ux
J145907.18+002401.2 z=2.767
Lyα
2000 1000 0 1000 2000Relative Velocity (km/s)
0.0
0.2
0.4
0.6
0.8
1.0
Norm
alize
d Fl
ux
Lyβ
Fig. 1.— H I Lyα (top row) and Lyβ (bottom row) profiles for three LLSs with NHI ≥ 1019 cm−2. The damping wings of Lyα are well-resolved in these SLLSs and the blue curves indicate for the best-estimates and uncertainty of the NHI values. These are well constrained,even in poor S/N data.
but not for a comprehensive LLS survey.
2.3. Data Reduction
The HIRES spectra were reduced with the HIRedux6
software package, primarily as part of the KODIAQproject (Lehner et al. 2014). Briefly, each spectral im-age was bias-subtracted, flat-fielded with pixel flats, andwavelength-calibrated with corresponding ThAr frames.The echelle orders were traced using a traditional flat-field spectral image. The sky background was subtractedwith a b-spline algorithm (e.g. Bochanski et al. 2009),and the quasar flux was further traced and optimally ex-tracted with standard techniques. These spectra wereflux normalized with a high-order Legendre polynomialand co-added after weighting by the median S/N of eachorder. This yields an individual, wavelength-calibratedspectrum for each night of observation in the vacuumand heliocentric frame. When possible, we then com-bined spectra from quasars observed on multiple nightswith the same instrument configuration.
Processing of the MIKE spectra used the MIReduxpackage now bundled within the XIDL software pack-age7. This pipeline uses algorithms similar to HIRedux.The primary difference is that the flux is estimated to-gether with the sky using a set of b-spline models which isdemanded by the short 5′′ slits employed with MIKE. Inaddition, these data were fluxed prior to coaddition usingthe reduced spectrum of a spectrophotometric standard
(taken from the same night in most cases). Therefore,we provide both fluxed and normalized spectra from thisinstrument.
Details on the data reduction of ESI and MagE spectraare provided in previous publications (Prochaska et al.2003b; Fumagalli et al. 2013).
All of the reduced and calibrated spectra are availableon the project’s website8. The Keck/HIRES spectra willalso be provided in the first data release of the KODIAQproject (Lehner et al. 2015).
3. LLS DEFINITION
Before proceeding to analysis of the sample, we strictlydefine the Lyman limit system. There are three aspectsto the definition:
1. The velocity interval analyzed, which also corre-sponds to a finite redshift window.
2. The NHI value of the system.
3. The spatial proximity of the LLS to other astro-physical objects (e.g. the background quasar or aforeground DLA).
Of these three, the first has received the least attentionby the community yet may be the most important. Es-tablishing a precise definition, however, is largely arbi-trary despite the fact that it may significantly impactthe studies that follow. This includes the assessment
Fig. 2.— Plots of the Lyman limit (left) and the Lyα (right) profile for the LLS at z = 2.917 toward J0924–1104. We observed thissystem with both the MIKE (top) and MagE (bottom) spectrometers at LCO. The spectra show a break in the flux at λ ≈ 3585A butresidual flux down to λ ≈ 3500A where the Lyman limit from a lower redshift absorber occurs. By modeling the flux decrement of thez = 2.917 LLS, we establish a precise estimate of its total NHI value. It is also evident that the H I Lyα profile alone offers very littleconstraint on NHI.
of gas kinematics (Prochaska & Wolfe 1997), metallic-ity (Prochter et al. 2010; Fumagalli et al. 2011a), andeven the contribution of LLSs to the cosmic mean freepath (Prochaska et al. 2014b). In this paper (and futurepublications), we adopt an observationally-motivated ve-locity interval of ±500km s−1 centered on the peak op-
tical depth zpeakLLS of the H I Lyman series. Frequently,
we estimate zpeakLLS from the peak optical depth of a low-
ion, metal-line transition. An LLS, then, is all of the
optically thick gas at |δv| < 500km s−1 from zpeakLLS . In
practice, we have not simply summed the H I columndensities of all Lyα absorbers within this interval. In-stead, we have adopted the integrated NHI estimate fromthe Lyman limit decrement or adopt NHI from the anal-ysis of damping in the Lyα profiles (see the next sectionfor more detail). As an example of the latter, we treatthe two absorbers at z = 3.1878 and z = 3.1917 towardsPKS2000-330 (Prochter et al. 2010) as a single LLS. Sim-ilarly, we sum metal-line absorption identified within theinterval although it rarely is detected in intervals thatexceeds 200km s−1. Moreover, this window was adjustedfurther to exclude absorption from unrelated (e.g higheror lower redshift) systems. While this is an observation-ally driven definition, we note that it should also captureeven the largest peculiar motions within dark matter ha-los at z ∼ 2.
With |δv| < 500km s−1 as the first criterion, we de-fine an LLS as any combination of systems with NHI ≥
1017.3 cm−2 within that interval; this yields an integratedoptical depth at the Lyman limit τ912 ≥ 1. In prac-tice, we distinguish the LLS from DLAs by requiringthat NHI < 1020.3 cm−2. Systems with 1016 < NHI <1017.3 cm−2 are referred to as partial LLSs or pLLSs, andare excluded from analysis in this manuscript. Last, werefer to an LLS within 3000 km s−1 of the backgroundquasar as a proximate LLS or PLLS (Prochaska et al.2008b). There are 5 PLLSs within our sample satisfyingthis definition, all with velocity separations of at least2000km s−1 from the reported quasar redshifts. Alto-gether, we present measurements for 157 LLSs at red-shifts zLLS = 1.76 − 4.39 and with NHI = 1017.3 −1020.25 cm−2. Here and in future papers we refer tothis dataset as the high-dispersion LLS sample (HD-LLSSample). We will augment this sample in the years tofollow via our web site.
4. NHI ANALYSIS
Although the continuum opacity of the Lyman limitgenerates an unambiguous signature in a quasar spec-trum, it is generally challenging to precisely estimate theH I column density NHI for a given LLS system. Thisfollows simply from the fact that exp(−τ912) � 1 forNHI > 1018 cm−2 and all of the H I Lyman series linesare on the saturated portion of the curve-of-growth forNHI < 1019 cm−2. Furthermore, the damping of Lyαis difficult to measure for NHI � 1020 cm−2, especiallyin low S/N spectra or at z > 3 where IGM blending is
Fig. 3.— Two examples of LLSs with a strong Lyman limit break (τ912 > 1) yet weak or absent damping of H I Lyα. In these examples,we only estimate bounds on the NHI values which can span an order-of-magnitude uncertainty.
substantial.Our approach to identifying LLS and estimating their
NHI values involved two relatively distinct procedures.For LLS with large NHI values (> 1019 cm−2), wesearched each spectrum for absorption features with largeequivalent widthsWλ characteristic of a damped Lyα line(i.e. Wλ � 1A). We then considered whether these can-didates could be related to a broad absorption line (BAL)system associated to the background quasar or Lyβ asso-ciated to a higher redshift DLA. Any such coincidenceswere eliminated. For the remaining candidates, we per-formed an analysis of the Lyα profile by overplotting aseries of Voigt profiles with NHI > 1018.5 cm−2, adjustingthe local continuum by-eye as warranted. When low-ionmetal absorption was detected near the approximate cen-troid of Lyα, we centered the model to its peak opticaldepth and refined the NHI value accordingly. We didnot, however, require the positive detection of metal-lineabsorption. In all cases, the Doppler parameter of themodel Lyα line was set to 30 km s−1.
For cases were the S/N was deemed sufficient and line-blending not too severe, we estimated (by visual inspec-tion) a ‘best’ NHI value and corresponding 1σ uncertain-ties. Although this procedure is rife with human inter-action, we maintain that it offers the most robust as-sessment (to date) for NHI estimation. This is becausethe dominant uncertainties are systematic (e.g. contin-uum placement and line-blending), which are difficult toestimate statistically. Figure 1 shows three examples ofLLSs with damped Lyα lines giving precisely estimatedNHI values. Such systems are commonly referred to as
super LLSs (SLLSs) or sub-DLAs.We provide the adopted NHI values and error estimates
of our SLLS sample (99 systems with NHI ≥ 1019 cm−2)in Table 3. These represent roughly 2/3 of the totalHD-LLS Sample. This high fraction occurs because weonly require coverage of H I Lyα to identify and ana-lyze these SLLS. This implies a much larger survey path-length than for the lower NHI LLS. In addition, one mayidentify and analyze multiple SLLSs along a given sight-line whereas one is restricted to a single LLS when theLyman Limit is central to the analysis. Because theseSLLSs tend to span nearly the entire ±500km s−1 win-dow that defines an LLS, it is possible that there is ad-ditional, optically thick gas not included in our NHI esti-mate. This will be rare, however, and the underestimateof NHI should generally be much less than 10%.
In parallel with the search for LLSs having strong Lyαlines, we inspected each spectrum for a Lyman limitbreak. For those the LLSs that exhibit non-negligibleflux at the Lyman limit, i.e. τ912 < 3, a precise NHI
estimation may be recovered. In practice, such analy-sis is hampered by poor sky subtraction and associatedIGM absorption that stochastically reduces the quasarflux through the spectral region near the Lyman limitand affects continuum placement. In the following, wehave been conservative regarding the systems with NHI
measurements from the Lyman limit flux decrement. Weare currently acquiring additional, low-dispersion spectrato confirm the flux at the Lyman limit for a set of theHD-LLS Sample. Figure 2 shows one example of a pLLSobserved with both the MIKE and MagE spectrometers.
Note. — All column densities are log10. The flag in the final columnindicates the quality of the measurement. A flgHI = 1 correspondsto a more precisely measured value and one may assume a GaussianPDF with the errors reported taken as 1σ uncertainties. A flgHI = 2corresponds to a less precisely measured value, and we recommend oneadopt a uniform prior for NHI within the error interval reported. Seetext for further details.a Redshift estimates for the peak H I opacity from metals and Lymanseries absorption.
16 17 18 19 20 21log NHI
0
5
10
15
20
25
30
35
Num
ber
1 2 3 4 5zLLS
17.0
17.5
18.0
18.5
19.0
19.5
20.0
20.5
logN
HI
Fig. 4.— (left) Histogram of NHI values for the HD-LLS Sample.The ligher bars indicate NHI values with large uncertainty (1σ >0.4 dex). Nearly two-thirds of the sample have NHI ≥ 1019 cm−2
which is a consequence of the spectral coverage required to estimateNHI for an LLS (see text). (right) Scatter plot of NHI vs. zLLS forthe sample. Again, the lighter points indicate LLSs with poorerconstraints on the NHI values (flg=2 in Table 3). Systems withNHI < 1019 cm−2 all have zLLS > 2.6 as coverage of the Lymanlimit is required for the H I analysis.
The flux decrement is obvious and one also appreciatesthe value of higher spectral resolution (with high S/N)for resolving IGM absorption.
For the remainder of the systems identified on thebasis of a Lyman limit break, we adopt conservativebounds (i.e. upper and lower limits) to the NHI values.These are based primarily on analysis of the Lyα lineand the flux at the Lyman limit. The absence of strongdamping in the former provides a strict upper limit toNHI while the latter sets a firm lower limit. Thesebounds are provided in Table 3, and Figure 3 showstwo examples of these ‘ambiguous’ cases. In practice,the bounds are often an order-of-magnitude apart, e.g.1017.7 cm−2 < NHI < 1018.9 cm−2. Furthermore, it isdifficult to estimate the probability distribution function(PDF) of NHI within these bounds. One should not, forexample, assume a Gaussian PDF centered within thebounds with a dispersion of half the interval. In fact,we expect that the PDF is much closer to uniform, i.e.equal probability for any NHI value within the bounds.This expectation is motivated by current estimations ofthe NHI frequency distribution f(NHI, X) which arguefor a uniform distribution of NHI values for randomly se-lected systems with NHI ≈ 1018 cm−2 (Prochaska et al.2010; O’Meara et al. 2013). Going forward, we advocateadopting a uniform PDF.
As a cross-check on the analysis, 50 of the sightlineswere re-analyzed by a second author to identify LLSs andestimate their NHI values. With two exceptions, the val-ues between the two authors agree within the estimateduncertainty and we identify no obvious systematic bias9.These two exceptions have> 0.5 dex difference due to dif-fering definitions used by the two authors and we haveadopted the values corresponding to the strict definitionprovided in § 3. This exercise confirms that the uncer-tainties are dominated by systematic effects, not S/N northe analysis procedures.
Table 3 lists the adopted NHI value, errors on thisvalue, the bounds on NHI, and a flag indicating whetherone would assume a normal or uniform PDF. Figure 4
9 In fact the formal reduced χ2 for the comparison is significantlyless than unity, but this is because the estimated uncertainties in-clude systematic error and because each author analyzed the samedata.
9
0.0
0.2
0.4
0.6
0.8
1.0
1.2
Norm
alize
d Fl
ux
J034402.85-065300.6 z=3.843
SiII1526
400 200 0 200 400Relative Velocity (km/s)
0.0
0.2
0.4
0.6
0.8
1.0
1.2
Norm
alize
d Fl
ux
SiIV1393
logNHI =19.550.0
0.2
0.4
0.6
0.8
1.0
1.2
Norm
alize
d Fl
ux
J100428.43+001825.6 z=2.746
SiII1526
400 200 0 200 400Relative Velocity (km/s)
0.0
0.2
0.4
0.6
0.8
1.0
1.2No
rmali
zed
Flux
SiIV1393
logNHI =19.800.0
0.2
0.4
0.6
0.8
1.0
1.2
Norm
alize
d Fl
ux
J092705.92+562114.1 z=1.775
SiII1526
400 200 0 200 400Relative Velocity (km/s)
0.0
0.2
0.4
0.6
0.8
1.0
1.2
Norm
alize
d Fl
ux
SiIV1393
logNHI =19.00
Fig. 5.— Si II and Si IV transitions for three SLLS representative of the full HD-LLS Sample. Note the large diversity in metal-linestrength despite the comparable NHI values. These systems have a tendency to show both low and high-ion absorption indicative ofpartially ionized gas. The gray dotted line in each panel indicates the estimated 1σ error array.
Note. — [The complete version of this table is in the electronic edition of the Journal. The printed edition contains only a sample.]
10
0.0
0.2
0.4
0.6
0.8
1.0
1.2
Norm
alize
d Fl
ux
J155103.39+090849.2 z=2.700
CII1334
400 200 0 200 400Relative Velocity (km/s)
0.0
0.2
0.4
0.6
0.8
1.0
1.2
Norm
alize
d Fl
ux
CIV1548
logNHI =17.500.0
0.2
0.4
0.6
0.8
1.0
1.2
Norm
alize
d Fl
ux
J004049.5-402514 z=2.816
CII1334
400 200 0 200 400Relative Velocity (km/s)
0.0
0.2
0.4
0.6
0.8
1.0
1.2No
rmali
zed
Flux
CIV1548
logNHI =17.550.0
0.2
0.4
0.6
0.8
1.0
1.2
Norm
alize
d Fl
ux
J233446.4-090812.3 z=3.226
CII1334
400 200 0 200 400Relative Velocity (km/s)
0.0
0.2
0.4
0.6
0.8
1.0
1.2
Norm
alize
d Fl
ux
CIV1548
logNHI =17.70
Fig. 6.— C II and C IV transitions for LLSs with low NHI values. Unlike the SLLSs from Figure 5, these LLSs have metal absorptionthat is dominated by high-ions. There only a few cases of LLSs with NHI < 1018 cm−2 and a positive low-ion detection.
Note. — [The complete version of this table is in the electronic edition of the Journal. The printed edition contains only a sample.]
11
shows a histogram of the adopted NHI values and a scat-ter plot against zLLS. It is evident that the HD-LLSSample is weighted towards higher NHI values and z ∼ 3.
5. IONIC COLUMN DENSITIES
For each of the HD-LLS Sample, we inspected thespectra for associated metal-line absorption. Empha-sis was placed on transitions with observed wavelengthsredward of the Lyα forest. A velocity interval wasestimated for the column density measurements basedon the cohort of transitions detected. Velocity plotswere generated and inspected to search for line-blending.Severely blended lines were eliminated from analysis andintermediate/weak cases were measured but reportedas upper limits. All of these assignments were vet-ted by JXP, MF, and JMO. Figures 5 and 6 show theSi II 1526/Si IV 1393 transitions for three representa-tive SLLSs and the C II 1334/C IV 1548 transitions forthree LLSs with NHI . 1017.7 cm−2. These data indicatea great diversity of line-strengths for these transitionswithin the SLLS sample. We also conclude that metal-absorption is dominated by high-ions in the lower NHI
systems.Column densities were measured using the apparent
optical depth method (AODM; Savage & Sembach 1991)which gives accurate results for unsaturated line profiles.On the latter point, the echelle data (MIKE,HIRES)have sufficiently high resolution to directly assess line-saturation, i.e. only profiles with minimum normalizedflux fmin less than 0.1 may be saturated. For the echel-lette data (MagE, ESI), however, line-saturation is a con-cern (e.g. Prochaska et al. 2003b). In general, we haveproceeded conservatively by treating most lines as satu-rated when fmin < 0.5. For many of the ions analyzedin these LLSs, we observe multiple transitions with dif-fering oscillator strengths and have further assessed line-saturation from the cohort of measurements.
Uncertainties were estimated from standard propaga-tion of error, which does not include error from contin-uum placement. To be conservative, we adopt a mini-mum uncertainty of 0.05 dex to the measurements froma given transition. When multiple transitions from thesame ion were measured (e.g. Si II 1304 and Si II 1526for Si+), we calculate the ionic column density from theweighted mean. Otherwise, we adopt the measurementfrom the single transition or a limit from the cohort em-phasizing positive detections.
A complete set of tables and figures for the metal-linetransitions analyzed for each LLS are given online. Ta-bles 4 and 5 summarize the results for Al+, Al++, Fe+,C+, C+3, O0, Si+, and Si+3. A listing of all the measure-ments from this manuscript is provided in the Appendix.Figures 7 and 8 show the column density distributionsfor a set of Al, Fe, Si, C, and O atoms/ions as a func-tion of the LLS NHI value. Not surprisingly, the lowerionization states show an obvious correlation10 with H Icolumn density although there is a large scatter at all val-ues. The near absence of positive detections for O I (i.e.N(O0) < 1014 cm−2) at low NHI is also notable. Thissuggests a rarity of high metallicity gas in systems withτ912 < 10. The high ions are also positively correlated
10 Taking limits as values, all of these ions have a Spearmanrank test probability of less than 0.0001.
12
13
14
15
logN
(C+
3)
17 18 19 20log NHI
12
13
14
15
logN
(C+
)
12
13
14
15
logN
(Si+
3)
17 18 19 20log NHI
12
13
14
15
16
logN
(Si+
)
Fig. 7.— Scatter plot of Si and C ionic column densities forthe HD-LLS Sample. Circles indicate measured values; their un-certainties are generally less than 0.1 dex. Triangles indicate lim-its to the values with the open symbols indicating upper limits.Lighter points mark LLSs with a poorly constrained NHI value.Gray boxes encompass 50-percent of the measurements in threelogarithmic NHI intervals: [17.3, 18.0), [18.0, 19.0), [19.0, 20.3).At all column densities, there is a large dispersion in the mea-surements. Nevertheless, the low-ions (C+, Si+) exhibit a strong,positive correlation with NHI value. A Spearman rank test rulesout the null hypothesis at > 99.99% c.l.
13
14
15
16
logN
(O0
)
17 18 19 20log NHI
13
14
15
logN
(Fe
+)
12
13
14
logN
(Al+
)
17 18 19 20log NHI
12
13
14
logN
(Al+
+)
Fig. 8.— Same as Figure 7 but for four additional ions.
with the neutral column but with yet larger scatter andmuch smaller correlation coefficients.
6. RESULTS
In the following, we present a set of results derivedfrom the column density measurements of the previoussections. For this manuscript, we restrict the analysisto an empirical investigation. Future studies will intro-duce additional models and analysis (e.g. photoionizationmodeling) to interpret the data. We further restrict thediscussion to ionic abundances and defer the analysis ofkinematics to future work.
12
1.51.00.50.00.51.01.52.0
logN
(C+
3) -
logN
(C+
)
17 18 19 20log NHI
1.5
1.0
0.5
0.0
0.5
1.0
logN
(Si+
3) -
logN
(Si+
)
1.0
0.5
0.0
0.5
1.0
logN
(Al+
+) -
logN
(Al+
)
17 18 19 20log NHI
2.0
1.5
1.0
0.5
0.0
0.5
logN
(Si+
) - lo
gN
(O0
)
Fig. 9.— Scatter plots of four ionic ratios that diagnose theionization state of the LLSs. Gray boxes encompass 50-percent ofthe measurements in two logarithmic NHI intervals: [17.3, 19.0),[19.0, 20.3). The C, Si, and Al ratios show strong evidence that thegas is more highly ionized at low NHI values. Similarly, the set ofSi+/O0 values exceeding −0.7 dex are indicative of a highly ionizedgas. We further emphasize that the log(Si+/Si+3) ≈ −0.5 dexvalues at high NHI suggests that this gas is also partially ionized.
6.1. Ionization State
As noted above, a full treatment of the ionization stateof the gas including the comparison to models will be pre-sented in a future manuscript. We may, however, explorethe state of the gas empirically through the examinationof ionic ratios that are sensitive to the ionization state ofthe gas. Figure 9 presents four such ionic ratios againstNHI. These primarily compare ions of the same element(e.g. C+3/C+) to eliminate offsets due to differing intrin-sic chemical abundances (i.e. varying abundance ratios).In this analysis, we have taken the integrated columndensity across the entire LLS. While there is evidencefor variations in these ratios within individual compo-nents, these tend to be small (e.g. Prochter et al. 2010,Figure 5). Therefore, the trends apparent in Figure 9reflect the gross properties of the LLS sample.
All of the C+3/C+, Si+3/Si+, and Al++/Al+ ratiosexhibit a strong anti-correlation with NHI indicating anincreasing neutral fraction with increasing H I opacity.Taking limits as values, the Spearman rank test yieldsa probability of less than 10−3 for the null hypothesis,in each case. For all of these ions, the upper ionizationstate is dominant for NHI . 1018.5 cm−2 and vice-versafor higher NHI values. We emphasize, however, that evenat NHI ≈ 1020 cm−2 the observed ratios are frequentlylarge, e.g. log(Si+3/Si+) ≈ −0.5 dex. This suggests thatthe gas is predominantly ionized even at these larger totalH I opacities.
This inference is further supported by the Si+/O0.Ignoring differential depletion, which we expect to bemodest in LLSs, the Si+/O0 ratio should trend towardsthe solar abundance ratio (εSi/εO = −1.2 dex) in aneutral gas given that Si and O are both produced inmassive stars and are observed to trace each other inastrophysical systems (e.g. stellar atmospheres). Weidentify, however, a significant sample of systems with
3
2
1
0
1
{ C+/H
0}
C II
17 18 19 20log NHI
3
2
1
0
1
{ Si+/H
0}
Si II
3
2
1
0
1
{ Al+/H
0}
Al II
17 18 19 20log NHI
3
2
1
0
1
{ Fe
+/H
0}
Fe II
Fig. 10.— Scatter plots of low-ion column densities relativeto H I, normalized to the solar abundance {Xi/H0} and plottedagainst the LLS NHI value. Gray boxes encompass 50-percent ofthe measurements in two logarithmic NHI intervals: [17.3, 19.0),[19.0, 20.3). If ionization corrections are small, {Xi/H0} providesan estimate of the logarithmic metal abundance relative to so-lar. The measurements appear to indicate a declining trend ofgas metallicity with increasing NHI. We argue, however, that thisapparent trend is driven by ionization effects and the set of up-per/lower limits at low/high NHI values. Furthermore, given thelarge scatter at all NHI values, it will be challenging to establishany trend between enrichment and NHI value in the LLSs.
NHI ≈ 1018.5 cm−2 that have log(Si+/O0) > −1 dex. Be-cause the majority of ionization processes (e.g. photoion-ization, collisional ionization) predict Si+/O0 > Si/O(e.g. Prochaska & Hennawi 2009), these measurementsoffer further evidence that LLSs are highly ionized.
6.2. Metallicity
A principal diagnostic of the LLSs is the gas metallic-ity, i.e. the enrichment of the gas in heavy elements. Thisquantity is generally characterized relative to the chemi-cal abundances observed for the Sun. For the following,we adopt the solar abundance scale compiled by Asplundet al. (2009), taking meteoritic values when possible.
Because the LLSs are significantly ionized, the ob-served ionic abundances reflect only a fraction of the totalabundances of Si, O, H, etc. Therefore, a full treatmentrequires ionization modeling. We may, however, offer in-sight into the problem by examining several ions relativeto H0. To minimize ionization corrections, one restrictsthe analysis to ionization states dominant in a highlyoptically thick (i.e. neutral) medium.
The results for four low-ions are presented in Fig-ure 10, normalized to the solar abundance. We have in-troduced here a new quantity and notation: {Xi/Yj} ≡log(N(Xi)/N(Yj))− εX + εY, where εX is the solar abun-dance on the logarithmic scale for element X. This quan-tity explicitly ignores ionization corrections and shouldnot be considered a proper estimate of the chemicalabundance ratio, traditionally expressed as [X/Y]. In thecases where ionization corrections are negligible, how-ever, {Xi/H0} = [X/H] and this quantity represents themetallicity.
A cursory inspection of the plots suggests a signifi-
13
17 18 19 20log NHI
3
2
1
0[O
/H]
O I
0 5 10 15
Fig. 11.— Estimations of the oxygen metallicity in the LLSswhere we have assumed that [O/H]={O0/H0}, i.e. that ionizationcorrections are small for this ionic ratio (see the text). Gray boxesencompass 50-percent of the measurements in three logarithmicNHI intervals: [17.3, 18.0), [18.0, 19.0), [19.0, 20.3). Similar to theresults for other low-ions (Figure 10), there is an apparent decline in[O/H] with NHI. This apparent trend, however, is primarily drivenby the high incidence of upper/lower limits at low/high NHI values.In fact, the underlying trend may well be the opposite. Comparethe set of low [O/H] values and upper limits at NHI ≈ 1018 cm−2
with the large set of lower limits at NHI ≈ 1019.7 cm−2.
cant decline in metal content with increasing NHI. Thisapparent anti-correlation, however, is driven by at leasttwo factors. First, larger NHI implies larger metal col-umn densities such that the transitions saturate yieldinga preponderance of lower limits. By the same token, atlow NHI values the transitions are often undetected yield-ing upper limits to the ionic ratios. Second, we have ar-gued from Figure 9 that the gas is increasingly ionizedwith decreasing NHI. For Si+, C+, and Al+, the ion-ization corrections for {Xi/H0} are likely negative (e.g.Prochaska 1999; Fumagalli et al. 2011a), and would lowerthe metallicity one infers from such ratios. We believethese factors dominate the trends apparent in Figure 10.
In fact, it is even possible that the true distribution ex-hibits the opposite trend. Figure 11 shows [O/H] againstNHI for the LLSs where we have assumed no ioniza-tion corrections, i.e. [O/H] = {O0/H0}. This approx-imation is justified by the fact that O0 and H0 havevery similar ionization potentials and their neutral statesare coupled by charge-exchange reactions. This assump-tion may break down at low NHI values in the pres-ence of a hard radiation field (Sofia & Jenkins 1998;Prochter et al. 2010), but the corrections are still likelyto be modest (several tenths dex). Unfortunately, themeasurements are dominated by limits: upper limits atNHI < 1018.5 cm−2 and lower limits at NHI > 1019 cm−2.Nevertheless, the data require that [O/H] > −1.7 forthe SLLSs and indicate [O/H] < −1.3 dex for LLSs withNHI ≈ 1018 cm−2. We tentatively infer that the me-dian metallicity is approximately flat with NHI and pos-sibly increasing; more strictly, we rule out a steeply de-clining O/H metallicity with increasing NHI. A similarconclusion may be drawn from the {Si+/H0} measure-
3.02.52.01.51.00.50.00.51.0
{ C+/H
0}
NHI<1019 cm−2
3.02.52.01.51.00.50.00.51.0
{ C+/H
0}
NHI>1019 cm−2
2.0 2.5 3.0 3.5 4.0 4.5zLLS
3.02.52.01.51.00.50.00.51.0
{ Si+/H
0}
2.0 2.5 3.0 3.5 4.0 4.5zLLS
3.02.52.01.51.00.50.00.51.0
{ Si+/H
0}
Fig. 12.— Comparison of {Xi/H0}measurements for C+ and Si+
against absorption redshift. Left-hand panels are for the LLSs withNHI < 1019 cm−2 and the right-hand panels are for the SLLS pop-ulation. All of the data show evidence for a declining enrichmentwith increasing redshift, although this assertion is statistically sig-nificant (> 99% c.l.) only for the SLLS sample. The absence oflow values at z ≈ 2 implies a reduced incidence of near-pristine gaswith high H I columns at that epoch.
ments which scatter less from line-saturation. The LLSswith NHI ≈ 1018 cm−2 show very few positive detec-tions and have a median {Si+/H0} < −1 dex. In con-trast, the LLSs at NHI ≈ 1019.5 cm−2 frequently exhibit{Si+/H0} > −1 dex.
Another result apparent from Figure 10 is the largedispersion in measurements at every NHI value. Thisis most notable for Si+ which has multiple transitionsthat permit measurements of the column density overa larger dynamic range. At the largest NHI values,the values/limits of {Si+/H0} span nearly four ordersof magnitude! And although the measurements forLLSs with NHI ≈ 1017.5 − 1019 cm−2 include many up-per limits, one identifies values and upper limits with{Si+/H0} > −0.5 dex together with upper limits having{Si+/H0} < −2. Clearly, any underlying trend of en-richment with NHI will be diluted by the large intrinsicscatter within the LLSs. One may even argue that ifsuch a dispersion is indicative of multiple astrophysicalsystems, then defining a mean of the LLS population haslimited scientific value.
Despite the large dispersion, we emphasize that veryfew of the LLS in the HD-LLS Sample are “metal-free”,i.e. exhibiting no metal-line absorption and therefore con-sistent with primordial abundances. Of the nLLS LLSs,only 25 have no low-ion detections outside the Lyα for-est and 18 of these exhibit a positive detection in ahigher-ion. For the other 7, one has been previously beenidentified as consistent with primordial (Fumagalli et al.2011a). The remainder have a diversity of S/N and spec-tral coverage and therefore are generally less sensitive tomeasuring a low metallicity. Several will be examinedin greater detail in a future manuscript. Nevertheless,we may conclude that the incidence of very low metal-licity gas (< 1/1000 solar) is rare in the LLS popula-
14
tion (< 5%). Furthermore, none of the 82 LLSs withNHI > 1019.2 cm−2 are metal-free11. By z ∼ 3, gas thatis dense enough to exhibit a very high Lyman limit opac-ity has previously been polluted by heavy elements.
At the opposite end of the enrichment distribution, weidentify 13 systems with a positive {Si+/H0} measure-ment that exceeds 0 dex. This includes four extreme ex-amples with {Si+/H0} > +0.5 dex. Because these fourLLSs also have NHI ≥ 1019 cm−2 we expect that cor-rections for ionization are modest (see Prochaska et al.2006) and that these are truly super-solar abundances.The others, however, have uncertainties consistent withthe gas being sub-solar even before accounting for ioniza-tion. We conclude, subject to additional future analysis,that super-solar enrichment is also rare in the LLSs.
In Figure 12, we examine {Si+/H0} and {C+/H0} val-ues as a function of redshift, splitting the LLS sample atNHI = 1019 cm−2. The values for the lower NHI systemssuggest a declining trend with increasing redshift, e.g. incontrast to the lower redshift systems, none of the z > 3.5LLSs have a positive detection of {C+/H0} > −0.5 dex.Even if we restrict analysis to positive detections, how-ever, an anti-correlation is not statistically significant.
Turning to the SLLS population, the {Xi/H0} distribu-tions show obvious trends with redshift (limits not with-standing). Treating all of the positive detections at theirplotted values, a Spearman’s rank correlation test rulesout the null hypothesis at > 99% c.l. We interpret thisanti-correlation as lower average enrichment within theSLLS at higher redshift. This conclusion relies on the as-sumption that ionization corrections will not evolve sig-nificantly with redshift, which will be investigated in afuture work. A similar decline in metallicity has beenestablished in the DLA population (e.g. Prochaska et al.2003a; Rafelski et al. 2012) and has been interpeted asresulting from the ongoing enrichment of galactic ISMwith cosmic time. Future work will perform a quantita-tive comparison between the two populations and explorethe implications for the evolving enrichment of opticallythick gas at z > 2. In passing, we emphasize the absenceof low {Xi/H0} values at z ≈ 2 which implies a reducedincidence of near-pristine gas with high H I columns atthat epoch.
6.3. Nucleosynthetic Patterns
It is the conventional wisdom that LLSs primarily tracegas outside of the ISM of galaxies, e.g. within their darkmatter halos (aka CGM) or at yet greater distances (Fu-magalli et al. 2011b; Prochaska et al. 2013). Despite theirseparation from galaxies, we have demonstrated that theLLSs are generally enriched in heavy elements and pro-vided evidence that their metallicity frequently reaches∼ 1/10 solar abundance. Therefore, a non-negligiblefraction of this optically thick medium has been pro-cessed through the furnaces of a stellar interior and pre-sumably was transported from a galaxy via one morephysical processes. One plausible transport process isan explosive event, e.g. a supernovae that expelled thegas shortly after enriching it. In this case, the gas mayexhibit a distinct nucleosynthetic pattern from those ob-
11 There is the possibility of a slight bias against our identifyingmetal-free SLLS but we have been as careful as possible to selectsystems based solely on the Lyα profile.
1.0
0.5
0.0
0.5
1.0
{ Si+/Fe
+}
Ia II PI D
17 18 19 20log NHI
1.0
0.5
0.0
0.5
1.0
{ O0/F
e+}
Ia II PI D
1.0
0.5
0.0
0.5
1.0
{ Si+/C
+}
Ia II PI D
17 18 19 20log NHI
1.0
0.5
0.0
0.5
1.0
{ Al+/C
+}
Ia II PI D
Fig. 13.— Scatter plots of a series of ionic ratios, normalized tothe solar relative abundances, against the NHI values of the LLSs.The left two panels show ratios related to the α/Fe abundance. De-spite the preponderance of lower limits (especially for {O0/Fe+}),and concerns on the ionization corrections, we tentatively concludethat the LLSs exhibit super-solar α/Fe ratios, especially at largeNHI values. The measurements in the upper-right panel indicatethat Si/C is possibly enhanced by a few 0.1 dex relative to solaralthough the majority of the sample is consistent with [Si/C] =0. Similarly, the {Al+/C+} measurements are roughly consistentwith the solar relative abundance or possibly sub-solar. In eachpanel, we indicate the expected offsets to the measurements thatwould be due to Type Ia (Ia) nucleosynthesis, Type II (II) nu-cleosynthesis, photoionization (PI), and differential depletion (D).Circles indicate a small or unknown impact.
served for galactic ISM, i.e. if the supernovae ejecta didnot mix prior to escaping the system. Additionally, theLLSs may couple the metal production within galaxiesto the enrichment of the diffuse IGM (e.g. Aguirre et al.2001; Schaye et al. 2003; Steidel et al. 2010). This moti-vates comparison of the abundances for these two diffuseand ionized phases.
We may explore several ionic ratios that trace differentnucleosynthesis channels. As with metallicity, one mustaccount for ionization effects when interpreting the re-sults. Figure 13 plots four pairs of ions from the dataset,again represented as {Xi/Yj} with ionization correctionsexplicitly ignored. The figure also indicates the proba-ble offsets to the ratios if ionization effects were impor-tant, as estimated from photoionization calculations (e.g.Prochaska 1999). Similarly, we indicate the likely offsetsfrom differential depletion and the dominant nucleosyn-thesis channels (Type Ia and Type II enrichment).
The left-hand panels show two measures of α/Fe, akey diagnostic of the relative contributions of Type Iaand Type II SNe nucleosynthesis (Tinsley 1979). Unfor-tunately, the {O0/Fe+} ratios are dominated by lowerlimits due to the saturation of O I 1302 and the non-detection of Fe II transitions. The values are nearly con-sistent with a solar abundance although there are at leasttwo systems with {O0/Fe+} > +0.3 dex suggesting anα-enhanced gas. Correcting for photoionization effectswould only strengthen this conclusion. These two LLSsalso exhibit a low metallicity ([O/H] ≈ −2) such thattheir chemical signature is very similar to that of metal-
15
poor Galactic stars (McWilliam 1997).Turning to {Si+/Fe+}, the sample is dominated by
measurements exceeding the solar abundance. Thisincludes a non-negligible set of measurements with{Si+/Fe+} > +0.5 dex, and one may speculate that thisrepresents the metal-enriched ejecta of Type II SNe. The{Si+/Fe+} ratio, however, is likely to require an ioniza-tion correction to accurately estimate Si/Fe. This couldexplain, in part, the positive {Si+/Fe+} values in Fig-ure 13. On the other hand, the highest {Si+/Fe+} valuesoccur in LLSs with high NHI values where one expectsionization effects to be minimal12. We conclude, there-fore, that at least a subset of the LLS population exhibitssuper-solar α/Fe ratios indicative of Type II enrichment,even in higher metallicity gas.
Previous studies of gas in the IGM at z ∼ 2 havereported an enhanced Si/C abundance (Aguirre et al.2004). This result was derived statistically from the pixeloptical depth method and is sensitive to the assumedmodel of the extragalactic UV background (EUVB; seealso Simcoe 2011). The results for LLSs offer a mixedpicture (Figure 13). There are a handful of positive{Si+/C+} values up to +0.5 dex, with the highest mea-surements at low metallicity. On the other hand, thesample is dominated by upper limits (from C II 1334saturation) and over half of these have {Si+/C+} <+0.3 dex. Once again, photoionization corrections wouldonly strengthen this result. As such, the LLS observa-tions do not appear to exhibit a high enrichment of Si/Cthan that previously inferred for the IGM.
Figure 13 also presents the set of {Al+/C+} measure-ments that are not fully compromised by line-saturation.These data are consistent with the lighter element ratiosin LLSs having solar relative abundances. The prepon-derance of upper limits, however, allows that Al could beunder-abundant relative to C.
6.4. Comparisons
We have restricted the HD-LLS Sample to systemswith NHI < 1020.3 cm−2 to exclude the DLAs. This waspartly motivated by the expectation that the majority ofLLSs are predominantly ionized and therefore physicallydistinct from the neutral gas comprising DLAs. It wasalso motivated by the desire to examine this opticallythick gas separately from the decades of research on theDLAs. Nonetheless, the NHI = 1020.3 cm−2 criterion isprimarily an observationally defined boundary and onemay gain insight into the nature of the LLSs through acombined comparison. Such analysis has been performedpreviously for the SLLS by ?Som et al. (2013).
We consider two such comparisons here. Figure 14presents the {Si+/H0} and {Fe+/H0} measurements forthe HD-LLS Sample together with measurements fromthe sample of DLAs of Rafelski et al. (2012). For bothdatasets, we have restricted to zabs = [1.6, 3.3] to mini-mize trends related to redshift evolution. To zeroth or-der, the DLA measurements extend in a roughly continu-
12 Such gas may also experience differential depletion, i.e. ele-vated Si/Fe ratios in the gas phase to the refractory nature of theseelements (e.g. Jenkins 2009). If the gas is predominantly ionized,however, the depletion levels may be modest and this effect wouldbe small.
3
2
1
0
{ Si+/H
0}
Si II
17 18 19 20 21log NHI
3
2
1
0
{ Fe
+/H
0}
Fe II
0 10 20 30 40
Fig. 14.— Comparison of the {Si+/H0} and {Fe+/H0} mea-surements for the LLSs and DLAs. The latter are drawn from theabundance compilation of Rafelski et al. (2012). For both ions, theDLAs show a continuous extension of the measurements observedin the LLSs. The right hand panels compare the distributions ofthe DLAs (darker) against those for the SLLSs (lighter), and em-phasize the commonality between the two datasets. In each case,we have treated upper and lower limit estimates as values.
ous manner from the measurements of the LLSs. Indeed,comparing the samples of DLA measurements with theSLLSs (taking limits at their values), one observes over-lapping distributions with similar median values. Theonly notable distinction, perhaps, is the small set of LLSswith NHI ≈ 1019 cm−2 and high {Xi/H0} values (exceed-ing 0 dex for Si+). This suggests a higher incidence ofhighly enriched gas in the LLS, although we caution itcould be partly an effect of ionization. The dispersionin the measurements is also larger for the LLSs, and islikely higher than suggested by the Figure given the pre-ponderance of upper/lower limits for the LLS/DLA.
Turning to the higher ionization states, Figure 15presents the C+3 and Si+3 column densities from theLLSs and DLAs. Once again, the DLA distribution ex-tends in a nearly continuous manner from the upper endof the LLS data and the column density distributions forthe SLLSs and DLAs are similar. Together, Figures 14and 15 lend support to scenarios that envision LLSs asthe outer layers of gas surrounding DLAs, i.e. these sys-tems frequently sample the same structures. Such physi-cal associations may be examined by studying DLAs andLLSs along pairs of quasar sightlines, an active area ofresearch (Ellison et al. 2007; Fumagalli et al. 2014; Rubinet al. 2014).
Examining the high-ion comparison further, there is atleast one import distinction: the LLSs and especially theSLLSs show a much higher incidence of low N(C+3) andN(Si+3) values. This is unexpected given that (i) theLLSs trace highly ionized gas; (ii) the DLAs trace pre-dominantly neutral gas that is physically distinct fromthe high-ions (Wolfe & Prochaska 2000; Prochaska et al.2008a). The results presented here indicate that the gaslayers giving rise to DLAs are embedded in a reservoirof highly ionized gas that frequently exceeds the typi-cal surface density in LLSs. This follows previous work
16
12
13
14
15
logN
(C+
3)
17 18 19 20 21log NHI
12
13
14
15
logN
(Si+
3)
0 10 20 30
Fig. 15.— Comparison of the high-ion column densities mea-sured for the LLS and a representative set of DLAs (drawn fromRafelski et al. 2012; Neeleman et al. 2013). Similar to the low-ionabundances, the measurements show a continuous transition fromthe LLS regime to higher NHI values. As such, there is substantialoverlap in the distribution of measurements and limits (right-handpanels compare the values for DLAs [darker] and SLLSs [lighter]).One notable and surprising difference is that the SLLSs show ahigher incidence of low columns of C+3 and Si+3. Despite trac-ing predominantly neutral gas, the DLAs also mark a reservoir ofhighly ionized gas that frequently exceeds the medium encompass-ing the LLSs.
which has inferred high quantities of both neutral andionized gas for the DLAs (Fox et al. 2007; Lehner et al.2014). It further suggests that high-ions more closelytrace higher density regions in the universe and/or mayreflect a difference in the masses of the dark matter haloshosting LLSs and DLAs.
Lastly, we have compared our measurements againstthe small set of literature values for SLLSs at z > 2(Dessauges-Zavadsky et al. 2003; Som et al. 2013). Thelow-ion column densities are typically larger in those pub-lications, consistent with the higher NHI values of theSLLSs that were sampled.
7. SUMMARY
We have constructed a sample of 157 LLSs at z ∼ 2−4observed at high-dispersion with spectrometers on theKeck and Magellan telescopes which constitute the HD-LLS Sample. In this manuscript, we present the completesample and present column density measurements of H Iand associated metal absorption. For the latter, analysiswas restricted to transitions redward of the Lyα forestand has focused on commonly detected species. Thesemeasurements and the associated spectra are made avail-able online with this publication13. This constitutes, byroughly an order of magnitude, the largest high redshiftsample of LLS analyzed in this manner.
We have explored empirical trends in the column den-sity measurements and report statistically significant(> 99.99%) correlations between the low-ion (e.g. Si+,C+) columns and NHI. High-ion species (Si+3, C+3) aredetected in nearly all LLSs and their column densities
13 http://www.ucolick.org/ ∼xavier/LLS
also correlated with NHI. Examining ionic ratios sensi-tive to the ionization state (e.g. C+3/C+, Si+3/Si+), weconclude that the LLSs are predominantly ionized withmore highly ionized gas in lower NHI systems.
Ratios of low-ion column densities to NHI indicate awide spread in metal-enrichment within the LLSs, likelyspanning four orders of magnitude. Only a small sub-set (. 5%) of the HD-LLS Sample have no positive de-tections of associated metals, consistent with primordialabundances. None of the LLSs with NHI ≥ 1019.2 cm−2
are ‘metal-free’. We conclude that a very high percent-age of high-density gas at z ∼ 3 was previously enrichedto & 1/1000 solar abundance. The HD-LLS Sample alsoexhibits a small subset (∼ 10%) of LLSs that have so-lar or super-solar enrichment. These likely represent themost enriched gas reservoirs in the high redshift universe.
Lastly, we have examined several ionic ratios that aresensitive to the nucleosynthetic enrichment history of thegas. The preponderance of elevated Si+/Fe+ and O0/Fe+
measurements suggest the LLSs have an α-enhancementcharacteristic of Type II nucleosynthesis. In contrast,the Si+/C+ and Al+/C+ ratios are consistent with solarrelative abundances.
Future manuscripts on the HD-LLS Sample will: (i)study the metallicity distribution of the LLSs accountingfor ionization effects and will estimate the contributionof optically thick gas to the cosmic metal budget; (ii)examine the kinematic characteristics to constrain thephysical origin of the gas; (iii) offer constraints on theNHI frequency distribution for optically thick gas.
J. X. P. was supported by NSF grants AST-1010004and AST-1412981. MF acknowledges support by theScience and Technology Facilities Council, grant num-ber ST/L00075X/1. We thank Claude-Andre Faucher-Giguerre for kindly providing his continuum fits to MIKEspectra. We acknowledge the contributions of Wal Sar-gent and Brian Penprase in collecting a portion of theESI data and Arthur M. Wolfe, Marcel Neeleman, andMarc Rafelski for their contributions to the Keck obser-vations.
Much of the data presented herein were obtained at theW.M. Keck Observatory, which is operated as a scientificpartnership among the California Institute of Technol-ogy, the University of California, and the National Aero-nautics and Space Administration. The Observatory wasmade possible by the generous financial support of theW.M. Keck Foundation. Some of the Keck data wereobtained through the NSF Telescope System Instrumen-tation Program (TSIP), supported by AURA throughthe NSF under AURA Cooperative Agreement AST 01-32798 as amended.
17
APPENDIX
APPENDIX: Measurements for Individual LLSTable 6 lists measurements for all of the metal-line transitions analyzed in this manuscript and Figures showing
velocity plots are provided in the on-line materials (Figure 16 shows one example). The analysis was restricted tolines outside the Lyα forest and those lines that are not severely blended with another feature or compromised bysky-subtraction residuals.
REFERENCES
Aguirre, A., Hernquist, L., Schaye, J., Weinberg, D. H., Katz, N.,& Gardner, J. 2001, ApJ, 560, 599
Aguirre, A., Schaye, J., Kim, T.-S., Theuns, T., Rauch, M., &Sargent, W. L. W. 2004, ApJ, 602, 38
Asplund, M., Grevesse, N., Sauval, A. J., & Scott, P. 2009, ArXive-prints
Berg, T. A. M., Neeleman, M., Prochaska, J. X., Ellison, S. L., &Wolfe, A. M. 2014, ArXiv e-prints
Bernstein, R., Shectman, S. A., Gunnels, S. M., Mochnacki, S., &Athey, A. E. 2003, in Instrument Design and Performance forOptical/Infrared Ground-based Telescopes. Edited by Iye,Masanori; Moorwood, Alan F. M. Proceedings of the SPIE,Volume 4841, pp. 1694-1704 (2003)., 1694–1704
Bochanski, J. J., et al. 2009, PASP, 121, 1409Bouche, N., Lehnert, M. D., & Peroux, C. 2006, MNRAS, 367,
L16Burles, S., & Tytler, D. 1998, ApJ, 499, 699Chen, H., Helsby, J. E., Gauthier, J., Shectman, S. A.,
Thompson, I. B., & Tinker, J. L. 2010, ApJ, 714, 1521Dekel, A., et al. 2009, Nature, 457, 451Dessauges-Zavadsky, M., Peroux, C., Kim, T.-S., D’Odorico, S.,
& McMahon, R. G. 2003, MNRAS, 345, 447Ellison, S. L., Hennawi, J. F., Martin, C. L., & Sommer-Larsen,
J. 2007, MNRAS, 378, 801Faucher-Giguere, C.-A., Hopkins, P. F., Keres, D., Muratov,
A. L., Quataert, E., & Murray, N. 2014, ArXiv e-printsFaucher-Giguere, C.-A., & Keres, D. 2011, MNRAS, 412, L118Faucher-Giguere, C.-A., Prochaska, J. X., Lidz, A., Hernquist, L.,
& Zaldarriaga, M. 2008, ApJ, 681, 831Fox, A. J., Petitjean, P., Ledoux, C., & Srianand, R. 2007, A&A,
465, 171Fumagalli, M., Hennawi, J. F., Prochaska, J. X., Kasen, D.,
Dekel, A., Ceverino, D., & Primack, J. 2014, ApJ, 780, 74Fumagalli, M., O’Meara, J. M., & Prochaska, J. X. 2011a,
Science, 334, 1245Fumagalli, M., O’Meara, J. M., Prochaska, J. X., & Worseck, G.
2013, ApJ, 775, 78Fumagalli, M., Prochaska, J. X., Kasen, D., Dekel, A., Ceverino,
D., & Primack, J. R. 2011b, MNRAS, 418, 1796Hennawi, J. F., & Prochaska, J. X. 2007, ApJ, 655, 735Jenkins, E. B. 2009, ApJ, 700, 1299Keres, D., Katz, N., Weinberg, D. H., & Dave, R. 2005, MNRAS,
363, 2Lehner, N., et al. 2013, ApJ, 770, 138Lehner, N., O’Meara, J. M., Fox, A. J., Howk, J. C., Prochaska,
J. X., Burns, V., & Armstrong, A. A. 2014, ApJ, 788, 119Marshall, J. L., et al. 2008, in Society of Photo-Optical
Instrumentation Engineers (SPIE) Conference Series, Vol. 7014,Society of Photo-Optical Instrumentation Engineers (SPIE)Conference Series
McWilliam, A. 1997, ARA&A, 35, 503Neeleman, M., Wolfe, A. M., Prochaska, J. X., & Rafelski, M.
2013, ApJ, 769, 54O’Meara, J. M., Burles, S., Prochaska, J. X., Prochter, G. E.,
Bernstein, R. A., & Burgess, K. M. 2006, ApJ, 649, L61O’Meara, J. M., Prochaska, J. X., Burles, S., Prochter, G.,
Bernstein, R. A., & Burgess, K. M. 2007, ApJ, 656, 666O’Meara, J. M., Prochaska, J. X., Worseck, G., Chen, H.-W., &
Madau, P. 2013, ApJ, 765, 137Peeples, M. S., Werk, J. K., Tumlinson, J., Oppenheimer, B. D.,
Prochaska, J. X., Katz, N., & Weinberg, D. H. 2014, ApJ, 786,54
Penprase, B. E., Prochaska, J. X., Sargent, W. L. W.,Toro-Martinez, I., & Beeler, D. J. 2010, ApJ, 721, 1
Peroux, C., Dessauges-Zavadsky, M., D’Odorico, S., Sun Kim, T.,& McMahon, R. G. 2005, MNRAS, 363, 479
Prochaska, J. X. 1999, ApJ, 511, L71Prochaska, J. X., Chen, H.-W., Wolfe, A. M.,
Dessauges-Zavadsky, M., & Bloom, J. S. 2008a, ApJ, 672, 59Prochaska, J. X., Gawiser, E., Wolfe, A. M., Castro, S., &
Djorgovski, S. G. 2003a, ApJ, 595, L9Prochaska, J. X., Gawiser, E., Wolfe, A. M., Cooke, J., & Gelino,
D. 2003b, ApJS, 147, 227Prochaska, J. X., & Hennawi, J. F. 2009, ApJ, 690, 1558Prochaska, J. X., Hennawi, J. F., & Herbert-Fort, S. 2008b, ApJ,
675, 1002Prochaska, J. X., et al. 2013, ApJ, 776, 136Prochaska, J. X., Lau, M. W., & Hennawi, J. F. 2014a, ApJ, 796,
140Prochaska, J. X., Madau, P., O’Meara, J. M., & Fumagalli, M.
2014b, MNRAS, 438, 476Prochaska, J. X., O’Meara, J. M., Herbert-Fort, S., Burles, S.,
Prochter, G. E., & Bernstein, R. A. 2006, ApJ, 648, L97Prochaska, J. X., O’Meara, J. M., & Worseck, G. 2010, ApJ, 718,
392Prochaska, J. X., & Wolfe, A. M. 1997, ApJ, 487, 73Prochaska, J. X., Wolfe, A. M., Howk, J. C., Gawiser, E., Burles,
S. M., & Cooke, J. 2007, ApJS, 171, 29Prochaska, J. X., et al. 2001, ApJS, 137, 21Prochaska, J. X., Worseck, G., & O’Meara, J. M. 2009, ApJ, 705,
L113Prochter, G. E., Prochaska, J. X., O’Meara, J. M., Burles, S., &
Bernstein, R. A. 2010, ApJ, 708, 1221Rafelski, M., Wolfe, A. M., Prochaska, J. X., Neeleman, M., &
Mendez, A. J. 2012, ApJ, 755, 89Ribaudo, J., Lehner, N., & Howk, J. C. 2011, ApJ, 736, 42Rubin, K. H. R., Hennawi, J. F., Prochaska, J. X., Simcoe, R. A.,
Myers, A., & Wingyee Lau, M. 2014, ArXiv e-printsRudie, G. C., et al. 2012, ApJ, 750, 67Sargent, W. L. W., Steidel, C. C., & Boksenberg, A. 1989, ApJS,
69, 703Savage, B. D., & Sembach, K. R. 1991, ApJ, 379, 245Schaye, J., Aguirre, A., Kim, T.-S., Theuns, T., Rauch, M., &
Sargent, W. L. W. 2003, ApJ, 596, 768Sheinis, A. I., Bolte, M., Epps, H. W., Kibrick, R. I., Miller, J. S.,
Radovan, M. V., Bigelow, B. C., & Sutin, B. M. 2002, PASP,114, 851
Simcoe, R. A. 2011, ApJ, 738, 159Sofia, U. J., & Jenkins, E. B. 1998, ApJ, 499, 951Som, D., Kulkarni, V. P., Meiring, J., York, D. G., Peroux, C.,
Khare, P., & Lauroesch, J. T. 2013, MNRAS, 435, 1469Songaila, A., & Cowie, L. L. 2010, ApJ, 721, 1448Steidel, C. C. 1990, ApJS, 74, 37Steidel, C. C., Erb, D. K., Shapley, A. E., Pettini, M., Reddy, N.,
Bogosavljevic, M., Rudie, G. C., & Rakic, O. 2010, ApJ, 717,289
Storrie-Lombardi, L. J., McMahon, R. G., Irwin, M. J., &Hazard, C. 1994, ApJ, 427, L13
Tinsley, B. M. 1979, ApJ, 229, 1046Tytler, D. 1982, Nature, 298, 427Vogt, S. S., et al. 1994, in Proc. SPIE Instrumentation in
Astronomy VIII, David L. Crawford; Eric R. Craine; Eds.,Volume 2198, p. 362, 362–+
Werk, J. K., Prochaska, J. X., Thom, C., Tumlinson, J., Tripp,T. M., O’Meara, J. M., & Peeples, M. S. 2013, ApJS, 204, 17
18
TABLE 6IONIC COLUMN DENSITIES
Quasar RA DEC zabs NHI λ vlim flg Nλ σ(N) Ion flg Nion σ(N)(J2000) (J2000) (A) (km/s)