The Red Dead Redemption Survey of Circumgalactic Gas About Massive Galaxies. I. Mass and Metallicity of the Cool Phase
We present a search for HI in the circumgalactic medium (CGM) of 21 massive (), luminous red galaxies (LRGs) at . Using UV spectroscopy of QSO sightlines projected within 500 kpc () of these galaxies, we detect HI absorption in 11/21 sightlines, including two partial Lyman limit systems and two Lyman limit systems. The covering factor of gas within the virial radius of these LRGs is , while for optically-thick gas () it is . Combining this sample of massive galaxies with previous galaxy-selected CGM studies, we find no strong dependence of the HI covering factor on galaxy mass, although star-forming galaxies show marginally higher covering factors. There is no evidence for a critical mass above which dense, cold ( K) gas is suppressed in the CGM of galaxies (spanning stellar masses ). The metallicity distribution in LRGs is indistinguishable from those found about lower-mass star-forming galaxies, and we find low-metallicity gas with (1.5% solar) and below about massive galaxies. While the high-metallicity cold gas seen in LRGs could plausibly result from condensation from a corona, the low-metallicity gas is inconsistent with this interpretation.
0000-0002-8518-6638]Michelle A. Berg \move@AU\move@AF\@affiliationDepartment of Physics, University of Notre Dame, Notre Dame, IN 46556
0000-0002-2591-3792]J. Christopher Howk \move@AU\move@AF\@affiliationDepartment of Physics, University of Notre Dame, Notre Dame, IN 46556
0000-0001-9158-0829]Nicolas Lehner \move@AU\move@AF\@affiliationDepartment of Physics, University of Notre Dame, Notre Dame, IN 46556
0000-0001-6923-978X]Christopher B. Wotta \move@AU\move@AF\@affiliationDepartment of Physics, University of Notre Dame, Notre Dame, IN 46556
0000-0002-7893-1054]John M. O’Meara \move@AU\move@AF\@affiliationDepartment of Chemistry and Physics, Saint Michael’s College, Colchester, VT, 05439 \move@AU\move@AF\@affiliationW. M. Keck Observatory 65-1120 Mamalahoa Hwy. Kamuela, HI 96743
0000-0002-5668-0397]David V. Bowen \move@AU\move@AF\@affiliationPrinceton University Observatory, Princeton, NJ 08544
0000-0002-1979-2197]Joseph N. Burchett \move@AU\move@AF\@affiliationDepartment of Astronomy and Astrophysics, UCO/Lick Observatory, University of California, Santa Cruz, Santa Cruz, CA 95064
0000-0003-1455-8788]Molly S. Peeples \move@AU\move@AF\@affiliationSpace Telescope Science Institute, Baltimore, MD, 21218 \move@AU\move@AF\@affiliationDepartment of Physics and Astronomy, Johns Hopkins University, Baltimore, MD 21218
0000-0002-1883-4252]Nicolas Tejos \move@AU\move@AF\@affiliationInstituto de Física, Pontificia Universidad Católica de Valparaíso, Valparaíso, Chile
Many galaxies have experienced some process(es) that “quenched” their ability to transform gas into stars. For more massive “red-and-dead” galaxies, this does not mean that they lack a supply of gas: massive and/or elliptical galaxies contain significant reservoirs of hot gas (e.g., Anderson et al., 2013; Singh et al., 2018) and even some cool gas in both their interstellar medium (Serra et al., 2012; O’Sullivan et al., 2018) and the surrounding circumgalactic medium (CGM, Thom et al., 2012), albeit at a lower mass fraction than typical star-forming galaxies. Indeed, as these galaxies continue to accrete mass from their surroundings, they will accrete new gas as well. Massive galaxies thus contain significant gas mass, but they are not able to efficiently access that gas to form new stars at a high rate. This is in large part due to the physical conditions of their gas: it is predominantly hot, with long cooling times.
Simulations and theoretical work have suggested that the transition to quiescence is perhaps associated with different modes for gas accretion with galaxy mass (see the general CGM review by Tumlinson, Peeples, & Werk 2017 and additional reviews on accretion in Fox & Davé 2017). Low-mass galaxies may be able to accrete cold matter directly from the intergalactic medium (IGM); if that gas can stay cool as it falls to the center of the halo, it may fuel star formation relatively directly (e.g., Birnboim & Dekel, 2003; Kereš et al., 2005; Dekel & Birnboim, 2006; Stewart et al., 2011). (In this context “cold” or “cool” is used to describe gas at , although we will typically use it to describe dense, photoionized gas of order K.) However, cool material accreting onto higher-mass galaxies is thought to encounter strong accretion shocks, heating the gas to (Dekel & Birnboim, 2006; Ocvirk et al., 2008; Crain et al., 2010; Correa et al., 2018). Fueling star formation with matter accreted in this way requires the gas be able to cool (likely through a cooling instability) and survive a fall into the central regions of a galaxy to form stars (Maller & Bullock, 2004; McCourt et al., 2012; Voit et al., 2015). The prevailing picture is that this “hot-mode accretion” involving cooling from the hot CGM is relatively inefficient, especially when confronted by processes that suppress the cooling such as feedback from supermassive blackholes in these galaxies and ionization by the ultraviolet background (Oppenheimer & Schaye, 2013; Nelson et al., 2015). It is this inefficiency that ultimately keeps the star formation rate in massive galaxies low.
Measuring the properties of cold gas in the CGM about massive galaxies offers a way to directly assess several of the assumptions inherent in this picture. Cold gas in massive halos has been traditionally expected to be relatively rare. Such gas could arise from expulsion or stripping of gas from satellites or from cooling instabilities in the hot CGM itself. Both scenarios would produce relatively metal-rich cold gas and be found preferentially in the inner CGM, where the gas is denser and metallicities higher (the latter important for the cooling process). Generally, any pristine gas accreted onto massive halos should be shock-heated and unable to cool until it is mixed with more metal-rich material (Wiersma et al., 2009). Thus, we expect very little in the way of cold, metal-poor gas in the CGM of massive galaxies, in contrast to what is found more generally (e.g., Lehner et al., 2013, 2016; Wotta et al., 2016, 2018).
There are hints that this picture is not so simple (see Chen 2017 for a recent review of cold gas in the CGM of massive galaxies). For example, Thom et al. (2012) found no discernible difference between the cold HI content in the CGM of quiescent galaxies and star-forming galaxies in the COS-Halos survey (for a recent update, see Prochaska et al., 2017, hereafter P17). Thus, it appears that the CGM of these quiescent galaxies has a significant component of cold HI that is not being used to fuel star formation (Tumlinson et al., 2013).
Several recent studies have probed the cold gas content of massive luminous red galaxies (LRGs) at (e.g., Gauthier et al., 2009; Lundgren et al., 2009; Chen et al., 2010a; Bowen & Chelouche, 2011; Zhu et al., 2014; Pérez-Ràfols et al., 2015; Huang et al., 2016; Chen et al., 2018; Smailagić et al., 2018; Zahedy et al., 2018). Given their importance for cosmological studies (e.g., Eisenstein et al., 2005; Xu et al., 2013; Slepian et al., 2017), there are large samples of LRGs with both SDSS photometry and spectroscopy (Dawson et al., 2013, 2016; Prakash et al., 2016; Albareti et al., 2017). LRGs are selected to have stellar masses above ; most are passive and have been quiescent since (Banerji et al., 2010). Several studies have found significant metal line absorption from the CGM of LRGs, hinting at a prevalence of metal-enriched cool gas. Bouché et al. (2004) found a high cross-correlation amplitude between MgII absorbers seen against QSOs and LRGs, implying LRG halos house cold, metal-enriched gas. This result was confirmed by Gauthier et al. (2009) and Lundgren et al. (2009) with larger samples. Using stacked QSO absorption lines to study weaker MgII absorption, Zhu et al. (2014) and Pérez-Ràfols et al. (2015) again found MgII absorbers are correlated with LRGs out to 10 Mpc (a result confirmed recently by Lan & Mo, 2018a). Thus, there is a clear statistical correlation of metal-enriched cold gas with LRG halos.
Searches for individual strong MgII absorbers associated with individual LRGs have yielded high covering factor estimates of cold, metal-enriched gas around LRGs using absorption lines toward background QSOs (Chen et al., 2010b; Bowen & Chelouche, 2011; Gauthier & Chen, 2011; Huang et al., 2016). For example, Huang et al. (2016) found a covering factor () for impact parameters kpc of , a value that falls to for kpc. All of these surveys indicate an abundance of cool, metal-enriched gas in the CGM of LRGs. However, they have limited diagnostic power. In order to obtain larger samples of galaxies with QSOs projected at small impact parameters, these works utilize low-resolution spectroscopy of MgII, giving sensitivity only to the strongest absorbers (typically at the level of Å). While the presence of strong MgII tells us there is metal-enriched gas about LRGs, it gives us no information on whether metal-poor gas that may trace new accretion can survive deep in the halos of these massive galaxies.
In order to address these shortcomings, we study the HI content within kpc () of a sample of massive LRGs using archival ultraviolet (UV) spectroscopy from the Hubble Space Telescope (HST). The galaxies targeted by our Red Dead Redemption (RDR) survey all have masses in excess of the predicted critical masses marking the transition between galaxies thought to obtain their gas via “cold-mode” accretion and those acquiring their gas via “hot-mode” accretion (e.g., Birnboim & Dekel, 2003; Dekel & Birnboim, 2006; Ocvirk et al., 2008; Nelson et al., 2018). Here we aim to test whether LRGs and other massive galaxies show a deficit of cold, dense HI compared with lower-mass, star-forming galaxies and to assess the frequency of gas that may represent recent accretion (assessed through its metallicity) in these halos. During the preparation of our survey, a complementary work by Chen et al. (2018, hereafter C18) became available. Their work focused on cool gas in the inner regions of a sample of massive galaxy halos (their sightlines probe kpc while ours extend to kpc). Their findings of a significant covering factor of optically-thick HI are in agreement with ours, and we combine the samples in later sections of this paper.
Our paper is organized as follows. We describe our selection of the RDR sample of LRGs with HST UV spectroscopy in § 2 and summarize our data reduction and spectral analysis in § 3. We consider the column density and covering factors of HI absorbers with impact parameters about the RDR LRGs in § 4. We measure the metallicity of the cool absorbers to constrain their origins in § 5. We explore the implications of our results in § 6. Our main results are summarized in § 7. A companion paper, Howk et al. (2018, hereafter Paper II) focuses on the highly-ionized phase of the CGM about the RDR LRGs as traced by OVI.
Throughout this paper we adopt the cosmology from Planck Collaboration et al. (2016), notably km s Mpc, , and . We adopt a naming scheme for objects as follows: target galaxies are labelled as “LRG” followed by their SDSS coordinate designation; QSO targets are listed by their SDSS coordinate name. We will be studying absorption from partial Lyman limit systems (pLLSs) and Lyman limit systems (LLSs) in this work. We adopt definitions of pLLSs as absorbers with HI column densities ; LLSs are defined by . We generally refer to absorption systems with as “strong HI absorbers.”
2 Sample Selection and Galaxy Properties
2.1 Sample Selection
In this work we compile a sample of QSO-LRG pairs for which the QSOs have been observed in the UV by HST. We selected the LRGs from the SDSS DR13 (Albareti et al., 2017) LOWZ and CMASS samples of the BOSS and eBOSS surveys (Dawson et al., 2013, 2016, with additional information on the eBOSS target selection in Prakash et al. 2016). The LOWZ sample targets massive galaxies in the range , whereas the CMASS sample focuses on (Dawson et al., 2013). Hereafter, we refer to the galaxies as “LRGs,” even if the sample is broader than the traditional LRG definition of Eisenstein et al. (2001). This includes the late-type galaxies that may make up of the sample (Masters et al., 2011). The most important characteristic of these samples is that they select high-mass galaxies, with a mean stellar mass of (e.g., Maraston et al., 2013).333Unless otherwise stated, all masses in this work are given in physical solar masses, , with all cosmological corrections included. We adopt throughout.
The LRGs from which we draw our sample all have spectroscopic redshifts available. We further restrict our LRG selection to the redshift range . The minimum redshift is adopted to place the Lyman break in the UV spectral range accessible to HST; the Lyman break is usually key to accurately assessing the total HI column density, , in high column density systems that are not damped. The maximum redshift is chosen so that a UV selection of QSOs can be made based on GALEX FUV magnitudes (see below). The typical redshift errors from the SDSS pipeline in this sample are , which corresponds to 30 km s.
We drew QSOs from the DR7 QSO catalog of SDSS (Schneider et al., 2010) cross-matched with GALEX UV sources (Martin et al., 2005). We selected QSOs projected within 500 kpc of the LRG sample, requiring and . The low-redshift cut-off simply ensures the QSOs are at redshifts higher than our target LRGs. The high-redshift cut is made in order to avoid the confusing influence of the dense Lyman- forest at higher redshifts. The redshift separation constraint minimizes the contamination from unrelated absorption that might contaminate the signature of the LRGs’ CGM. We also apply a constraint to the GALEX photometry of the QSOs, in order to ensure good S/N in even the low-resolution observations. These constraints yield QSO-LRG pairs projected within kpc.
We used this sample as the basis of a search of the Mikulski Archive for Space Telescopes (MAST) for HST UV spectra of the background QSOs, searching for data from the Goddard High Resolution Spectrograph, the Faint Object Spectrograph, the Space Telescope Imaging Spectrograph, and the Cosmic Origins Spectrograph. Because we are interested in an unbiased search for HI (in particular) associated with LRGs, we excluded any data from program 14171 (PI: Zhu) that targeted (some) LRGs on the basis of previously-identified MgII absorption. This search yielded 24 sight lines with UV spectral coverage of the Lyman series lines at the redshift of the LRGs. Three of these LRGs were later removed from our sample. In one case, the QSO was initially targeted on the basis of a strong foreground absorber that resides at the redshift of the LRG. In the second, the galaxy was an outlier in stellar mass from the rest of the sample (with ), and the data for the third sightline were low enough resolution that we could not definitively determine if the absorption feature is associated with the LRG.
The remaining 21 QSO-LRG pairs listed in Table The Red Dead Redemption Survey of Circumgalactic Gas About Massive Galaxies. I. Mass and Metallicity of the Cool Phase form the basis of our study. The 21 LRGs along these sightlines lie in the range . The median sample redshift is . To the best of our knowledge, based on a reading of the publicly-available program abstracts in MAST, none of the background QSOs were targeted due to the presence of the foreground LRGs in our sample. (Several of the QSOs were targeted to study other foreground galaxies, but this does not affect our results.) Even with this selection, three sightlines do not have coverage of the break due to the choice of grating. These LRGs are all probed by Lyman series lines that give us the ability to place strong constraints on the absorption, so we have included them in our sample.
2.2 Galaxy Properties
Table The Red Dead Redemption Survey of Circumgalactic Gas About Massive Galaxies. I. Mass and Metallicity of the Cool Phase summarizes the properties of the LRGs in the RDR sample. While there are galaxies in our final sample potentially hosting star formation ( show detectable emission), the defining characteristics of the galaxies are their redshift range and large masses. The redshifts in Table The Red Dead Redemption Survey of Circumgalactic Gas About Massive Galaxies. I. Mass and Metallicity of the Cool Phase are adopted from the SDSS pipeline for each LRG. The median redshift of our sample is , while the median impact parameter is kpc. The distributions of redshifts and several other properties of our sample (described below) are shown in Figures 2.2 and 2.2.
For each target galaxy we estimate stellar masses and absolute -band magnitudes using the kcorrect code (v4_3)444Available through http://kcorrect.org or http://github.com/blanton144/kcorrect. of Blanton & Roweis (2007). These SED fits make use of spectral templates from Bruzual & Charlot (2003) and assume the initial mass function of Chabrier (2003). We used Galactic extinction-corrected (Schlegel et al., 1998) MODEL magnitudes with the pipeline redshifts as inputs to kcorrect to assess the -corrections and mass-to-light ratios of the galaxies. The results are given in Table The Red Dead Redemption Survey of Circumgalactic Gas About Massive Galaxies. I. Mass and Metallicity of the Cool Phase. The output masses from the code are given in , which we convert to masses with units in Table The Red Dead Redemption Survey of Circumgalactic Gas About Massive Galaxies. I. Mass and Metallicity of the Cool Phase and all that follows. These masses likely have an uncertainty of at least dex, if not somewhat larger (Conroy, 2013). Although we have not specifically fit the full spectra for all of these galaxies, the redshifts are fixed and the galaxy templates are well determined for these systems. We show the distribution of stellar masses for our sample in Figure 2.2 (bottom right panel). The galaxies in our sample have a median stellar mass (in ) with a standard deviation of 0.2 dex, both consistent with the sample studied by Huang et al. (2016).
We estimate the halo mass, , of each LRG using the stellar mass–halo mass (SMHM) relationship of Rodríguez-Puebla et al. (2017), who combine a large number of observational studies of the SMHM scalings. In particular, they provide a prescription (their Equation 66) for assessing the mean halo mass at a given stellar mass, (which is different than the inverse of the stellar mass-to-halo mass ratios given the asymmetries in the scatter about the mean relationship; Behroozi et al., 2010). Recently Tinker et al. (2017) have assessed the SMHM relationship in a sample of CMASS galaxies drawn in much the same way as our sample, focusing on , though they extend their fits to somewhat lower masses. Their predictions for yield halo masses lower by roughly 0.3 dex (i.e., a factor of ) at a given stellar mass than those of Rodríguez-Puebla et al. (2017). There are a number of other recent studies on the SMHM relation at high stellar masses (e.g., Velander et al., 2014; Shan et al., 2017). We adopt the Rodríguez-Puebla et al. (2017) results both because they account for the asymmetric scatter in the relationship in order to properly calculate and because their analysis allows us to calculate the masses of the RDR LRGs and lower-mass galaxies against which we will compare the LRGs with a consistent treatment. Our final halo mass estimates are given in Table The Red Dead Redemption Survey of Circumgalactic Gas About Massive Galaxies. I. Mass and Metallicity of the Cool Phase; the median halo mass derived for our sample is (in ).
For comparison with other galaxy absorber studies, it can be useful to consider the impact parameters relative to the virial radii of each galaxy. We define the “virial radius” as , the radius enclosing the halo mass for a mean density that of the critical density at the redshift of each LRG. That is, , where and is the critical density. We refer to this scale simply as throughout this paper. Our estimates of are given in Table The Red Dead Redemption Survey of Circumgalactic Gas About Massive Galaxies. I. Mass and Metallicity of the Cool Phase. Given the dependence of these values, they are not strongly affected by our choice of SMHM relation. Our calculations give a median virial radius kpc; the median normalized impact parameter of our sightlines is .
To limit the star formation rates (SFRs) of the RDR sample, we use 3727+3729 emission as an indicator of star formation in these galaxies. These results should be considered upper limits given the potential for non-star formation contributions to these lines (Huang et al., 2016). We adopt the calibration of the -SFR relationship from Moustakas et al. (2006) for the highest-luminosity galaxies (which are appropriate given the range in absolute magnitudes of our sample; see Table The Red Dead Redemption Survey of Circumgalactic Gas About Massive Galaxies. I. Mass and Metallicity of the Cool Phase). Thus we assume . We adopt the SDSS pipeline fits to emission from the LRGs, correcting for Milky Way foreground extinction assuming Schlegel et al. (1998) extinction with and an extinction curve following Cardelli et al. (1989). We do not correct for internal extinction, as neither do Moustakas et al. (2006) in their calibration, which partly causes the luminosity dependence in their -SFR relationship. Where galaxies do not have detectable emission, we adopt the upper limits on the flux.
In two cases the SDSS fits gave uncertainties well above ( higher) those typical of the sample. In one case, it was not clear why the fits yielded such high uncertainties; in the other, the doublet was coincident with a poorly-subtracted sky emission line. In these two cases, we used the pPXF spectral fitting code of Cappellari (2017) to refit the SDSS spectra. Our fit for LRG SDSSJ125859.98+413128.3 yielded tighter limits on the emission. For LRG SDSSJ075217.92+273835.6, where the line is contaminated, we had good measurements for the fluxes of H and H. The limit on H is stronger (as H is in the forest of sky lines in the far red). For only this galaxy we use the H line as an indicator of the galaxy’s SFR, adopting the H-SFR calibration of from Moustakas et al. (2006).
The specific star formation rates, , for the RDR galaxies are given in Table The Red Dead Redemption Survey of Circumgalactic Gas About Massive Galaxies. I. Mass and Metallicity of the Cool Phase. Because the SDSS fibers used to limit the SFR do not encompass all of the stellar light from the galaxy, we compare our derived star formation rates to the stellar masses contained within the fiber. We use the ratio of the SDSS FIBER2FLUX to the MODELFLUX in the -band to scale down our derived stellar masses for this comparison. In most cases, non-detections of do not set stringent constraints on the sSFR. In Figure 2.2 we show the sSFRs calculated using the SFR derived from a luminosity-weighted median stack of the LRG spectra showing no detectable emission, . Thus even the stacked spectrum of these galaxies gives only a marginal detection of emission.
Even amongst the galaxies with detectable emission, it is not necessarily connected to star formation. The and emission of LRGs generally resembles that of low-ionization nuclear emission-line region (LINER)-like galaxies (Huang et al., 2016), where the emission is suspected to come from either activity from active galactic nuclei (AGN) or ionization from post-asymptotic giant branch stars (e.g., Ho 2008; Yan & Blanton 2012). Belfiore et al. (2016) have shown more generally that LINER emission does not necessarily originate from the nucleus, but from spatially extended regions in both star-forming and passive galaxies. Furthermore, stacking of FIRST radio images suggests nearly all LRGs house an AGN (Hodge et al., 2008, 2009). In light of these considerations, all sSFRs should be considered upper limits.
For the most part we do not have extensive data on the environments of these specific galaxies. We expect all will be surrounded by an extensive suite of lower-mass satellites. At lower redshift (), of LRGs are central galaxies in clusters with richness parameter in the redMaPPer cluster catalog (Rykoff et al. 2014, 2016; Hoshino et al. 2015). We cross-matched our RDR sample of galaxies with the redMaPPer cluster catalog version 6.3555Available through http://risa.stanford.edu/redmapper/. and found five matches. The LRGs SDSSJ124307.36+353926.3, SDSSJ141307.39+091956.7, SDSSJ171651.46+302649.0, SDSSJ075217.92+273835.6, and SDSSJ132457.98+271742.6 are all within of the cluster center, indicating that they are either the central galaxy or the brightest cluster galaxy. LRGs SDSSJ124307.36+353926.3, SDSSJ141307.39+091956.7, and SDSSJ075217.92+273835.6 are found in rich clusters (), while the other two are found in clusters with a smaller number of member galaxies (). The fact that the other galaxies are not specifically identified with overdensities like those in the redMaPPer catalog does not mean they lack satellites.
3 Observations and Data Analysis
Our RDR sample comprises 21 QSOs with HST UV spectroscopy that pierce the CGM of 21 LRGs within kpc. The use of UV spectroscopy allows us to identify the prevalence of HI and associated metal ions in the CGM of these galaxies. We supplement these UV data with optical spectroscopy to measure MgII associated with the LRGs. Below we describe the datasets we have used and our approach to measuring absorption from the LRG CGM.
3.1 Ultraviolet Spectroscopy from HST
A summary of the QSO observations is provided in Table The Red Dead Redemption Survey of Circumgalactic Gas About Massive Galaxies. I. Mass and Metallicity of the Cool Phase. Seventeen of the QSOs were observed with the medium resolution mode () of COS using the G130M and/or G160M gratings. The other five observations were obtained with low resolution () UV spectra from COS and FOS. All of the COS spectra were retrieved from the HST Spectroscopic Legacy Archive (HSLA, Peeples et al. 2017) except for SDSSJ112756.76+115427.1, SDSSJ124307.57+353907.1, and SDSSJ150527.60+294718.3, which were not available in the April 2017 data release. For these spectra, we use the routines in COS Tools (Danforth et al., 2010, 2016) to coadd the data. Differences exist between the two data reduction processes, but they do not impact our column density measurements (Lehner et al., 2018a). For the objects observed with FOS, we use the reduced spectra from Ribaudo et al. (2011).666Available through Vizier. One object, SDSSJ125224.99+291321.1, was observed with both COS and FOS, with a gap of 140 Å between the two spectral regions. The spectrum and analysis of SDSSJ171654.20+302701.4 is described in Wotta et al. (2016). We make use of their data and column densities for this sightline.
3.2 Supporting Ground-based Observations
Previous surveys have used MgII to understand the presence of metal-enriched gas in LRG halos (e.g., Bowen & Chelouche 2011; Zhu et al. 2014; Huang et al. 2016). We also measure MgII for several of our sightlines through LRG halos because MgII often provides the best metallicity constraints for pLLSs and LLSs (Wotta et al., 2016, 2018). High resolution spectroscopy yields the best constraints on the column density; one QSO has spectroscopy from the High Resolution Echelle Spectrometer (HIRES) on Keck I in the Keck Observatory Archive. This spectrum is reduced following Lehner et al. (2016) and is reported in Wotta et al. (2016). A second sightline was observed with the Multi-Object Double Spectrograph (MODS) on the Large Binocular Telescope (LBT). Wotta et al. (2016) details the reduction procedure for this instrument. We also analyze the SDSS spectra of the QSOs when available (four sightlines do not have MgII coverage, Abolfathi et al. 2018). Due to the limited sensitivity and resolution of SDSS (and the low HI column densities along most of the sightlines), these columns are almost all upper limits.
3.3 Absorption Line Search Methodology
To find absorbing gas associated with the RDR LRGs, we conduct a search for hydrogen and metal-ion absorption in the QSO spectrum at the redshift of the LRG using a search window of km s. Our choice of this large velocity search window follows from the escape velocity of the median LRG halo from our survey (930 km s assuming a Navarro-Frank-White potential Navarro et al. 1996, 1997) and the MgII survey in the CGM of LRGs by Huang et al. (2016, see their Figure 4). They found only 12% of their absorbers lie at a velocity offset greater than km s when considering impact parameters kpc. However, they also found a few systems at km s, so we adopt the same window to ensure we consider all the gas that could be associated with each LRG.
When HI absorption is detected within km s of the redshift of the LRG, we set the redshift of the absorbing gas using the Lyman series transitions. If a break in the QSO spectrum is observed, the redshift of the pLLS/LLS is determined from the centroids of the higher Lyman series lines. If a break is not present (due to spectral coverage or lower HI column), we use Ly or Ly. In cases where no absorption is detected, we use the redshift of the LRG to set the center of the velocity integration window to estimate upper limits on the column densities.
3.4 Column Density Measurements
Our measurements of HI and ion column densities are made with different techniques or assumptions depending on the column density of the system and the quality of the data. Due to the heterogeneous mix of resolution and signal-to-noise (S/N), our limits on HI or metal absorption are not uniform across the sample. However, our focus in this work is on understanding the amount of high HI column density gas present about LRGs. For this purpose, the mixture of data properties is not an issue because the Lyman break is covered in almost every sightline. Two of the three sightlines that do not have Lyman break coverage do not exhibit high HI column absorbers; the final sightline can only be given a lower limit for the HI column density.
Four of the LRGs show strong HI systems (). Of these four systems, three have observable breaks in the QSO spectrum at the Lyman limit. The HI column density can be determined from the break using , where is the optical depth at the Lyman limit, and is the absorption cross section of the hydrogen atom at the Lyman limit (Spitzer, 1978). A composite QSO model is scaled to fit the continuum of the spectrum (Telfer et al., 2002); a break is then added to match the level of the absorbed continuum giving a measure of the optical depth and hence . To estimate the errors on we follow the methodology described in Wotta et al. (2016). In short, the errors are estimated by offsetting the model to a point in which the model and data are clearly inconsistent. This column density value is taken to be the value.
We determine the HI column density for the absorbers associated with LRGs SDSSJ111132.33+554712.8 and SDSSJ171651.46+302649.0 (reported in Wotta et al. 2016) from the optical depth at the break, but we can only estimate a lower limit for the absorber associated with LRG SDSSJ111508.24+023752.7 from this method due to the low level of flux recovery blueward of the Lyman limit. To further constrain the column density of the latter absorber, we perform a Voigt profile fit on the Lyman series lines using an updated version of the profile fitting code from Fitzpatrick & Spitzer (1997, see for more detail). This technique models the absorption line (with the ability of adding more components) and determines the best-fit model through minimization. Using an initial guess for the column density, velocity dispersion (Doppler parameter), and velocity centroid, it returns the best-fit values of the model profile for these three parameters and can be iterated on. We fit the lines HI 1025, 972, 937, 926, 923, 920, 919, 918, 917, 916 in this manner (see Appendix A). We use two components to model the HI profiles since the three stronger transitions show an additional component with and km s, shifted by km s from the redshift of the LLS. The LLS is reasonably well fit with and km s, which is consistent with the lower limit derived just using the break at the Lyman limit, (see also Lehner et al. 2018a).
No strong break in the QSO spectrum is observed for the other high HI column density system. The pLLS associated with LRG SDSSJ141540.01+163336.4 is on the low end of the pLLS column density range, exhibiting a small partial break in the QSO spectrum. We measure the HI column density from the higher Lyman series lines using the apparent optical depth method (AODM, Savage & Sembach 1991). The AODM converts an absorption line to apparent column density per unit velocity interval. The total apparent column density is then determined by direct integration of the apparent column density per unit velocity interval: . The column density is assumed to be as long as there is no evidence of saturation (see below for our treatment of saturation). We also measure from the break in the QSO spectrum at the Lyman limit for this absorber and find the column densities from the two methods are consistent. The value measured from the break exhibits larger errors, so we adopt the value measured from the Lyman series lines.
An absorber is detected in the spectrum of QSO SDSSJ125224.99+291321.1, which has been observed with COS and FOS. Unfortunately, these data do not cover the Lyman limit and higher Lyman series lines associated with this absorber. With only Ly and Ly, a curve-of-growth analysis did not provide any robust results. Using the AODM on Ly observed with FOS, we could place only a lower limit on ().
For the remaining absorbers, the absorption is much weaker, and we use only the AODM to estimate the column densities or limits on the column densities of the HI and metal lines. The metal ion absorption line integration limits are set by the width of the associated hydrogen lines unless the metal ion profile is broader, contaminated, or shifted. If a line is contaminated, we shorten the integration limits. When no absorption is detected (for hydrogen or metals) we measure upper limits using the AODM by integrating over the section of the spectrum where the line would be located. As stated above, the redshift of the LRG is used as the center of our integration window in these cases, and the velocity integration limits are set by the smallest line width (v) that can still be detected in the respective grating. For the low-resolution observations, this equates to nine pixels for FOS/G190H and 12 pixels for COS/G140L.
To check for saturation in lines that do not reach zero flux but have significant peak optical depth, we employ two procedures. If the transition is part of a doublet, we compare the column density values of the stronger line to the weaker line. When the weaker line has a higher column density, the Savage & Sembach (1991) saturation correction is applied. For other transitions where there are no other lines in the same ionization state of a species to use for comparison, other species in the same state are considered (e.g., CIII and SiIII). If one of these lines is saturated, we also mark the other transitions as saturated if the peak optical depth is similar or larger.
We include in Appendix A plots of the HI and ionic absorption profiles for each sightline to show the velocity range over which the transitions are integrated. Table A1 details the apparent column densities measured for each sightline. If an absorber shows more than one component in the profiles, we determine the column densities for each component, in an effort to compare the component metallicities. Since we will show there is no significant difference between the component metallicities in these cases (see Appendix C), we quote total column densities throughout the paper. The three spectra in which we observe a break at the Lyman limit of the pLLS/LLS are also provided in Appendix A, along with the Voigt profile fits to the Lyman series lines for the absorber that could only be given a lower limit for from the break in the QSO spectrum.
4 Characteristics of Strong HI Absorption in the CGM About LRGs
4.1 Hydrogen and Metal-Line Absorption
Out of 21 sightlines, we find 11 detections of absorbing gas within km s of the LRG central redshift. The hydrogen column densities of the detected absorption lines span a broad range from to , but the distribution is not continuous. The absorber properties are summarized in Figures 4.1 and 4.1. We show in Figure 4.1 the HI column density as a function of impact parameter for the RDR sample. We find four strong HI absorbers (two pLLSs and two LLSs) in our sample with ; four of the other six HI detections are mostly at (likely Ly forest interlopers; see below). It seems the LRGs exhibit either high HI column density gas or not much at all. Most of the strong absorbers lie at small impact parameters, 0.5 , consistent with results for other samples of galaxies (e.g., Tumlinson et al. 2013; Heckman et al. 2017; Keeney et al. 2017; P17).
The low HI column density absorbers in Figure 4.1 are likely dominated by Ly forest contamination. As shown in Figure 4.1, a higher proportion of the weak absorbers exhibit large velocity offsets from the redshift of the LRGs () compared with the strong absorbers, suggesting they are not directly associated with the LRGs. We estimate the expected number of Ly forest interlopers in our sample, , by integrating the differential column density distribution, , for each sightline that has sensitivity to absorption. We use the fit to from Lehner et al. (2007, their equation 6), extrapolating below their sensitivity limit of and using the corrected parameter values from their erratum for the column density range with km s. We perform the integration for each sightline from to cm, where is the sensitivity to HI column for each spectrum. Once has been calculated for each sightline, we total the values to determine for our survey. We expect 5 Ly forest interlopers in our spectra, equal to the number of systems we detect (Figures 4.1 and 4.1 show only four as SDSSJ125901.67+413055.8 exhibits two systems that we sum for display in the figures; see Appendix A). Thus, it is very likely that all of the low HI column density systems are Ly forest interlopers unrelated to the RDR galaxies themselves. We expect to find zero random interlopers for assuming the parameterization from Shull et al. (2017). Thus, the high column density systems are likely associated with the LRGs.
LRGs exhibiting high HI column densities () in our sample have associated metal-line absorption. We do not have uniform spectral coverage for each sightline, so the ions we probe vary (see Appendix A). For several of our sightlines we cover OVI and can estimate its column density, which is the focus of Paper II. For pLLSs/LLSs, CIII is often one of the most prominent ions. We find absorption limits or detections of CIII spanning the range , with detections in four RDR LRGs. We calculate the covering factor (see below) for CIII to be and (68% confidence interval). We find no inconsistencies between our MgII equivalent width distribution and that presented by Huang et al. (2016), although we have a limited sample. In no cases do we find metal-line absorption having the strength or velocity breadth of those seen by Smailagić et al. (2018).
4.2 Covering Factor of Strong HI About LRGs
Covering factors can be used to characterize the CGM about different classes of galaxies allowing for quantitative comparisons. In addition, simulations often report covering factors of different ions, and their treatment of the CGM can be assessed by comparing to observational results. We split the sightlines into bins of impact parameter and determine the ratio of absorbers detected, for a limiting , to the total number of sightlines in each bin. An absorber that meets this criterion is referred to as a “hit,” while those below this limit are “misses.” We calculate the covering factor, , of strong HI assuming a binomial distribution for the covering factor where the posterior distribution is described by a beta function. This follows from the discussion in Cameron (2011) who describes the posterior distribution of values in a Bayesian treatment of binomial-distributed data. The beta function also provides estimates of the Bayesian confidence intervals, which we utilize throughout this paper.
We summarize our covering factor estimates in Table The Red Dead Redemption Survey of Circumgalactic Gas About Massive Galaxies. I. Mass and Metallicity of the Cool Phase, giving results for three limiting column densities: , 16.0, and 17.2. When imposing these limits we consider only the central value of our HI measurements. Except for the covering factor calculation with , we do not include the absorber for which we could only derive a lower limit for in all other covering factor calculations throughout this paper. Table The Red Dead Redemption Survey of Circumgalactic Gas About Massive Galaxies. I. Mass and Metallicity of the Cool Phase lists the impact parameter range for each bin, the mean impact parameter of sightlines within the bin, the covering factor with 68% and 95% confidence intervals, the number of sightlines probed, and the number of detected absorbers for all three HI column density limits. We plot the covering factors as a function of impact parameter and normalized impact parameter for in Figure 4.2. The horizontal error bars show the width of the bins, while the vertical error bars show the 68% confidence interval for the distribution. The covering factor within the inner 250 kpc of our sample of RDR LRGs is substantial at . The covering factor for within the virial radius for our sample is . For LLS absorption with , we find .
The covering factor of high column density gas around LRGs is non-negligible, with 35% of LRGs showing strong HI at these columns within 0.5 (the median = 516 kpc for our sample). A similar result is found around the quiescent COS-Halos galaxies (Thom et al., 2012, see also P17). We calculate a covering factor within 160 kpc (0.75 ) for the quiescent COS-Halos galaxies of at the same cutoff. C18 recently reported a covering factor within 165 kpc (0.33 ) of their sample of massive galaxies for , consistent with our results in Figure 4.2. While no direct simulation analogs of LRGs exist in the literature, we can extrapolate the results from Rahmati et al. (2015) to lower . They use the EAGLE simulation suite to characterize the HI distribution around galaxies from –1. Though they do not extend the analysis to , they simulate galaxies up to . In their Figure 5 they show a covering factor for of for the LRG halo mass range. Our covering factor at the same HI cutoff is consistent with their result. Rahmati et al. (2015) report a differential centered at for all galaxies with (their Figure 7); over a similar normalized impact parameter () we find for the LRGs. For optically-thick gas, the LRG observations match those from the simulations at . Future simulation work should consider the pLLS column density range at lower redshifts for covering factor calculations to aid in comparing with observations.
On the whole, LRGs do not make a strong contribution to the total population of pLLSs/LLSs. To assess their contribution to we start with the absorber number per path for uniformly-distributed galaxies of cross section and number density :
where we assume covering factors from our measurements. Connecting absorber distance to redshift path,
(Padmanabhan, 2002; Prochaska et al., 2010; Ribaudo et al., 2011). In these equations is the speed of light, is the covering factor, is the median virial radius for our sample, is the number density of LRGs, is the Hubble parameter, and is the median redshift of our sample. All other variables are defined in § 1. The physical number density of LRGs at is listed in Table 1 of Almeida et al. (2008), and we adopt the fiducial value Mpc. The contribution of LRGs to the combined pLLS and LLS population () is . The distribution calculated in Shull et al. (2017) that includes pLLSs has a value of = 2.24 at . For LLS absorption (), we find for the LRG contribution. The line density of LLSs calculated by Ribaudo et al. (2011) has a value of at . Thus, LRGs contribute only a few percent of the population of strong HI absorbers.
4.3 An Extended Sample of Galaxy-Selected CGM Measurements
To understand the nature of strong CGM absorption systems, we compare and combine the RDR sample with the galaxy-selected COS-Halos survey (Tumlinson et al. 2013; P17) and the recent C18 LRG study. We show in Figure 4.3 the distribution of stellar mass as a function of normalized impact parameter at which each of the galaxies is probed in this combined sample (we have recalculated the virial radii and normalized impact parameters for the COS-Halos and C18 samples using the same assumptions as for our sample, starting with the stellar masses given in those works). The COS-Halos sample comprises 44 star-forming and quiescent galaxies, with –11.5, and a median galaxy redshift of (Werk et al., 2012). As this sample is galaxy-selected, it has well-constrained host galaxy properties, and the sightlines extend to an impact parameter of 160 kpc (0.75 for their mean galaxy). The COS-Halos galaxies were selected to be relatively isolated, but this did not always turn out to be the case (Werk et al., 2012). The COS-Halos sample does not lie in the same redshift range as the RDR LRGs, so temporal evolution may affect our comparison. The C18 sample is another galaxy-selected survey comprised of 16 galaxies with stellar masses , a median redshift , and impact parameters kpc. Our sample includes five galaxies in common with the C18 sample, and the COS-Halos sample has five galaxies in common with C18. This leaves a total of six unique galaxies from C18 to include in the extended sample. In this section we focus on the COS-Halos sample, and in § 4.4 we include the C18 results.
The measurements for much of the COS-Halos sample are well constrained; however, several sightlines have HI absorption that can only be constrained by upper and lower bounds (P17). In what follows we remove three absorbers in the COS-Halos sample that have upper limits of , 16.65, and 17.01. These upper limits are not constraining for our covering factor calculations for absorbers having . Following Tumlinson et al. 2011, we cut the COS-Halos galaxies into two subsamples, defining quiescent and star forming galaxies as those with and , respectively.
The way in which we assess covering factors for the COS-Halos sample is a bit different than the approach in § 4.2. For sightlines with only HI bounds, the probability distribution function (PDF) of is described by a top-hat distribution between the two limits. This occurs for sightlines where the break at the Lyman limit is saturated (taking the QSO spectrum to zero flux and giving only a lower limit) and there are no detectable damping wings on Ly (giving an upper limit). There is also one sightline where the lower limit on comes from saturated Ly and Ly; in this case, we impose an upper bound of due to the lack of damping wings (J.X. Prochaska 2018, priv. comm.). When determining if these absorbers are hits or misses, we must consider the entire range of allowed values for these bounded systems. If the value at the low end of the interval is above 16.0, we count this absorber as a hit. However, the column density interval for several of these systems includes . In these cases, if less than 50% of the interval is above , we count the absorber as a miss; if it is , we count it as a hit. All other absorbers are treated as discussed in § 4.2. We note that Howk et al. (2017) performed a similar covering factor analysis of the COS-Halos results using a Monte Carlo approach; our approach does not produce significantly different results from theirs.
We compare the covering factors of our RDR sample with the quiescent and star-forming COS-Halos samples in Figure 4.3, and Table The Red Dead Redemption Survey of Circumgalactic Gas About Massive Galaxies. I. Mass and Metallicity of the Cool Phase summarizes our derivation of covering factors for the COS-Halos galaxies as in Table The Red Dead Redemption Survey of Circumgalactic Gas About Massive Galaxies. I. Mass and Metallicity of the Cool Phase (we include a complementary figure and table in Appendix B for results using optically-thick absorption). We use normalized impact parameter, , in all comparisons. We note that the sample of star-forming galaxies is almost twice as large as the RDR sample for the inner bin. Within 0.5 , the star-forming galaxies have a significantly higher covering factor of high HI column gas than the LRGs with . However, both the COS-Halos quiescent galaxies and RDR LRGs have large covering factors. All three samples are consistent at larger impact parameters, .
Figure 4.3 shows as a function of limiting HI column for our RDR sample compared with the COS-Halos samples. We calculate the covering factors as discussed above, but using the limiting HI column density given along the horizontal axis. The RDR LRGs do not have many detections of gas within , and the covering factor distribution is relatively flat for lower limiting columns. The COS-Halos quiescent galaxies have several detections at lower columns and a few detections of LLSs with column densities , while the highest column in the RDR sample is . Among these three samples, the LRGs are unique in the lack of gas with (a result that can also be seen in the C18 sample), while the passive galaxies in COS-Halos do show intermediate HI column density gas (). This difference may be in part due to the differing mass distributions between the two samples or the difference in redshift. It could also be the environment of these galaxies plays a role: most of the COS-Halos galaxies are selected (imperfectly) to be isolated, while the RDR LRGs are either in clusters or dense groups (see § 2.1).
4.4 The Mass Dependence of Strong HI Absorption
In this section we combine the RDR sample with the COS-Halos and C18 samples to assess the CGM properties across this combined sample of 71 galaxies. All of the sightlines probed were selected based on galaxy properties. We break the entire sample into bins of stellar mass and halo mass to understand whether the properties of CGM HI absorption depend on galaxy mass. We first consider the covering factors for the combined sample in three bins of stellar mass: 9.0–10.5), [10.5–11.3), [11.3–12.0]. We choose one of the boundaries at because this is the stellar mass at which early theoretical arguments suggested galaxies transition from cold-mode accretion to hot-mode accretion (Cattaneo et al. 2006; Dekel & Birnboim 2006, but see Nelson et al. 2015). We split the sample at in an effort to have a similar number of galaxies in the two higher mass bins. Figure 4.4 displays the covering factors for these mass bins, and Table The Red Dead Redemption Survey of Circumgalactic Gas About Massive Galaxies. I. Mass and Metallicity of the Cool Phase summarizes these results (we include a complementary figure and table in Appendix B for optically-thick absorption). There is no clear difference in the covering factors across the predicted critical value of stellar mass at (even comparing all masses above with those below). Figure 4.4 shows the covering factors as a function of limiting for each mass bin, as we showed in Figure 4.3. This figure shows that the lack of lower HI column absorption is limited to the highest-mass galaxies. The good agreement in the curves for the two lower-mass bins suggests that any transition in CGM HI properties occurs at masses above . Indeed, galaxies with show higher covering factors at than the lower-mass galaxies.
Figure 4.4 shows the covering factors in bins of halo mass (using the scalings from stellar mass in Rodríguez-Puebla et al. 2017). We bin the combined RDR, COS-Halos, and C18 sample by 11.0–12.0), [12.0–13.0), [13.0–14.0]. Table The Red Dead Redemption Survey of Circumgalactic Gas About Massive Galaxies. I. Mass and Metallicity of the Cool Phase gives a summary of the information from the figure (we include a complementary figure and table in Appendix B for optically-thick absorption). All three halo mass subsamples show consistent covering factors for . The lowest halo mass sample almost exclusively comprises star-forming galaxies, so this high covering factor result is consistent with Figure 4.3. There are several galaxies in the [12.0–13.0) halo mass range that are still forming a modest amount of stars, while those in the [13.0–14.0] halo mass range are all considered quiescent. These covering factors show there is a significant amount of cool gas in galaxy halos of all masses probed in this combined sample.
We can extend to even higher halo masses using galaxy clusters, the largest (often) virialized structures in the universe. These structures are roughly a factor of 10 higher in mass than the LRGs and extend to Mpc scales for . Clusters are dominated by hot X-ray emitting gas in the intra-cluster medium (ICM, the same phase that should dominate in the LRG CGM), and many satellite galaxies are undergoing gas stripping in these clusters. Yoon et al. (2012) and Yoon & Putman (2017) studied the distribution of Ly absorbers in the Virgo and Coma clusters. They found no detections of strong or optically-thick HI within when considering absorption within of the cluster velocity; most of their detections lie below . Their detections at are more numerous, yet still below what we consider strong HI gas.777There are two detections of strong HI gas outside of the Virgo cluster virial radius and within of the cluster velocity. Both detections are in areas of un-virialized substructure that may represent a recent merger event (Yoon & Putman, 2017). The covering factors within for the Virgo and Coma clusters at our adopted limit of are and (68% confidence interval). Burchett et al. (2018) also conducted a survey of absorption from cool gas in the ICM of five X-ray detected clusters. As with the two other cluster surveys, no strong HI was detected. We calculate the covering factor of within for the X-ray clusters to be . These results, coupled with that of Figure 4.4, suggest that the increase in mass between LRGs and clusters creates conditions that disfavor strong HI. A similar result is found in the COS-Clusters survey (Tejos et al. in prep.).
In these low-redshift clusters, sightlines passing through the ICM do not exhibit strong HI absorption (but see Muzahid et al. 2017). It may be that the physical properties of the ICM are preventing the formation of cool, dense gas that would give rise to pLLSs and LLSs (e.g., as discussed in Yoon & Putman 2013 and Burchett et al. 2018), perhaps in an analogous manner to the suppression of gas in LRGs. Having said that, we note that one of the pLLSs in our sample (associated with LRG SDSSJ171651.46+302649.0) is in a redMaPPer cluster with (see § 2.1), so the suppression of such gas is not complete.
5 Metallicities of Strong HI Absorbers about LRGs
5.1 Metallicity Methodology
In the previous section, we demonstrated that the massive RDR galaxies have substantial HI gas in their CGM. By determining the metal-enrichment levels of the gas, we can start to differentiate between the plausible origins of the gas. The gas probed by pLLSs and LLSs is predominantly ionized, and large ionization corrections are required to derive the metallicity since we compare metal-ions with HI (e.g., Lehner et al., 2013; Fumagalli et al., 2016; Lehner et al., 2018a; Wotta et al., 2018). To derive the metallicity, we follow the methodology of the COS CGM Compendium (CCC) where Wotta et al. (2018) derived the metallicities of 82 pLLSs and 29 LLSs using Bayesian techniques and Markov-chain Monte Carlo (MCMC) sampling of a grid of photoionization models (see Fumagalli et al., 2016). The grid of photoionization models is made using Cloudy (version C13.02, see Ferland et al., 2013), assuming a uniform slab geometry in thermal and ionization equilibrium. The slab is illuminated with a Haardt–Madau extragalactic ultravoilet background (EUVB) radiation field from quasars and galaxies (we adopt HM05 as implemented in Cloudy, see Haardt & Madau 1996).888We also undertook models using the HM12 EUVB and found the metallicities systematically increase by 0.1–0.5 dex, similar to the findings of Wotta et al. 2018. The grid parameters are summarized in Table The Red Dead Redemption Survey of Circumgalactic Gas About Massive Galaxies. I. Mass and Metallicity of the Cool Phase.
Most of the diagnostics available for the pLLSs and LLSs at probe the elements (O, Mg, Si) as well as carbon. The elements are produced by similar nucleosynthesis processes and, since dust depletion is negligible for the densities of the pLLSs and LLSs, their relative abundances are expected to be approximately solar (Lehner et al., 2013, 2018a). To describe the metallicity of the gas, we use the standard notation , where X here represents an -process element (e.g., O, Mg, Si). For the pLLSs and LLSs, [C/] is on average solar (Lehner et al., 2018a; Wotta et al., 2018), but this ratio can depart from a solar value with (Wotta et al., 2018). In our photoionization models, we therefore use a flat prior on [C/] following Wotta et al. (2018). In certain cases, the metallicity and density of an absorber cannot be constrained from the observations without additional assumptions due to either not enough available metal ion measurements or the dominance of upper limits. Wotta et al. (2018, and see also and ) showed that the ionization parameter has a normal distribution for low-redshift pLLSs and LLSs, which can be used as a prior for the absorbers that do not have sufficient constraints; this method provides what we describe as “low-resolution” metallicities (Wotta et al., 2016).
We run the MCMC analysis for the four strong HI absorbers associated with the LRGs SDSSJ111508.24+023752.7, SDSSJ111132.33+554712.8, SDSSJ171651.46+302649.0, and SDSSJ141540.01+163336.4. Metallicity estimates could not be determined for the weak HI absorbers () because they have only upper limits for their metal column densities or metal ion detections that will not give reliable metallicities. This result is unsurprising, as COS is not sensitive to metal lines in absorbers with (Lehner et al., 2018a, b).
5.2 Metallicities of LRG CGM Absorbers
The metallicities for the four pLLSs/LLSs detected about RDR LRGs are summarized in Table The Red Dead Redemption Survey of Circumgalactic Gas About Massive Galaxies. I. Mass and Metallicity of the Cool Phase. For each system the metallicity from our Cloudy/MCMC modeling is listed as the median value of the PDF with uncertainties reflecting the 68% confidence interval, along with the bounds of the 95% confidence interval. There are three high-metallicity absorbers with [X/H] among the four RDR systems. The last system, associated with LRG SDSSJ171651.46+302649.0, is a low-metallicity absorber with [X/H] (“low-resolution” method applied, see also Wotta et al. 2016). “Corner” plots summarizing the constraints on the properties of the four pLLSs/LLSs associated with the RDR LRGs and the modeling details for each absorber are included in Appendix C.
We plot the metallicity of the RDR absorbers versus normalized impact parameter in Figure 5.2; we also show two absorbers from the C18 sample with metallicities from the CCC and the metallicity measurements for absorbers associated with the COS-Halos galaxies (P17). Out of 17 quiescent galaxies in the COS-Halos sample (defined by ), only nine have and sufficient data to calculate an absorber metallicity. Out of 27 star-forming galaxies in the COS-Halos sample, only 16 have sufficient data and meet this criterion. The COS-Halos metallicities in P17 were originally estimated using the HM12 EUVB.999J1009+0713_170_9 was originally reported with one value of , but we report and [X/H] for each of the two components. This metallicity solution for J1241+5721_199_6 is not consistent with all of the observed column densities. For the SLLSs J0925+4004_196_22, J0928+6025_110_35, and J1435+3604_68_12 we use the column densities reported in Battisti et al. (2012) from their Voigt profile fits. We could not reach a satisfactory solution for J0925+4004_196_22 and instead adopt the reported [O/H] metallicity from Battisti et al. (2012) for this absorber. To remove the systematic offset introduced when using different ionizing radiation fields to derive metallicities, all the metallicities for the COS-Halos absorbers have been recalculated with the HM05 EUVB and are reported in Wotta et al. (2018) and Lehner et al. (2018b).
Figure 5.2 shows both passive and star-forming galaxies have a similar broad range of metallicities. The COS-Halos sample includes a low-metallicity absorber with [X/H] in the CGM of a massive quiescent galaxy. Like the [X/H] system associated with our RDR LRG, this absorber is projected deep within its halo. These systems suggest low-metallicity absorbers might not be rare around quiescent galaxies. This result is also bolstered by a low-metallicity LLS () reported around an LRG in the C18 sample. In fact, using additional constraints from Keck/HIRES detections of MgII, Wotta et al. (2018) report the metallicity of this absorber to be extremely low, . This absorber is not pristine, despite the lack of metal-line absorption in the COS spectrum analyzed by C18 (see Lehner et al. 2018a).
Figure 5.2 shows the metallicity versus for our sample of galaxy-selected clouds compared with HI absorption-selected systems at . The HI absorption-selected systems are shown in grey and are from the CCC (Lehner et al., 2018b; Wotta et al., 2018), while the galaxy-selected absorbers (RDR LRGs, C18, and COS-Halos) are shown in color. We also plot the and metallicity histograms of the galaxy-selected absorbers of star-forming galaxies (blue) and passive galaxies (red, which includes the RDR LRGs, C18, and COS-Halos quiescent galaxies) in the top and right sub-panels respectively.
Overall, the LRG absorbers do not seem to be fundamentally different in metallicity from the other galaxy-selected pLLSs/LLSs in spite of the characteristics of their host galaxies. There is also no appreciable difference between the metallicity distributions of the star-forming and passive samples; most of the metallicity values lie between solar and a tenth solar, which can be explained by enrichment from stellar outflows or tidal material (Lehner et al., 2013), and both samples have a similar fraction of low-metallicity absorbers ([X/H] ). There is also no appreciable difference between the distributions of star-forming galaxies split by stellar mass. Using a two-sample Kolmogorov-Smirnov test and K-sample Anderson-Darling test, we cannot reject the null hypothesis that the star-forming and passive metallicity distributions are drawn from the same population at better than 0.4 significance. We cannot reject the null hypothesis that the values are drawn from the same distribution either. Thus, there is no evidence the metallicity of the cool CGM—at least for absorbers in the range —is correlated with the galaxies’ star-forming properties. The LRG absorbers also have metallicities and HI column densities consistent with the distributions in pure HI absorption-selected systems. We find no strong difference between the absorption-selected and galaxy-selected absorbers.
Using a sample of 21 LRGs with UV observations of background QSOs with kpc, we have mapped the distribution of cool,101010We calculate the temperature for the absorber associated with LRG SDSSJ111508.24+023752.7 using the value from the Voigt profile fit and find K. neutral hydrogen absorption in the CGM of these massive galaxies. We have measured the covering factor and metallicity of gas with in the CGM of LRGs. We find substantial covering factors, especially in the inner regions of the CGM. The covering factor within for our sample is . The metallicities of the absorbers are mixed, with both metal-rich systems at [X/H] and metal poor with metallicities as low as [X/H] . The covering factors observed about the RDR LRGs are about 1/2 of those seen in lower-mass star-forming galaxies; the metallicity distributions of strong HI absorbers about LRGs are consistent with those observed about lower-mass galaxies.
6.1 Mass of Cool Gas in the LRG CGM
To assess the relative importance of this cool HI to the overall mass budget of the LRGs, we calculate the mass of cool gas associated with the measured HI as:
where is a function of the impact parameter . We do not have good constraints on the functional distribution of or the covering factor for , so we instead break this integral up into two discrete bins , = [0, 0.5 ] and (0.5 , ], assuming constant and in each bin. The sum of these two bins gives the typical total HI mass per LRG, . We estimate the total cool gas mass as , where the average ionization fraction of HI for the pLLSs/LLSs comes from the best-fit Cloudy models, and to account for helium. The typical LRG has a total cool CGM mass of in its halo, assuming the average value from our Cloudy calculations (our absorbers exhibit a range consistent with Lehner et al. 2018b). Compared to the expected baryonic mass for these galaxies, , only 1% of the baryons is in the cool CGM, while 6% is in stars. If the hot gas in the LRG halo dominates the remaining baryon mass, it could represent .
The total cool CGM mass has also been calculated around * galaxies. The COS-Halos sample, treated as one halo with many sightlines, has within kpc (P17, but see Bregman et al. 2018). Keeney et al. (2017) supplemented the COS-Halos sample with measurements extending to ; they adopted different assumptions for the structure and ionization of the gas to determine . The cool CGM mass of M31 is estimated to be within assuming a solar metallicity for the gas (Lehner et al., 2015).111111The erratum for Lehner et al. (2015) lists this updated cool CGM mass estimate. However, while the cool CGM mass in LRGs is comparable to that for lower-mass galaxies, it is a much smaller fraction of the total baryons. For the RDR sample, the cool CGM is 10% of the stellar mass of the galaxy. M31 shows a similar fraction, while the COS-Halos galaxies (and the extended Keeney et al. 2017 results) have almost twice as much mass in the cool CGM as the stellar disk. That said, the cool CGM gas in LRGs is not unimportant as we show below.
6.2 Origins and Fate of Cool Gas in LRG Halos
Cool gas is clearly found in the halos of LRGs and other massive, quiescent galaxies (e.g., Thom et al. 2012; \al@prochaska2017,chen2018; \al@prochaska2017,chen2018; this work). Several models predict cool gas can form through cooling instabilities in the hot corona and fall onto the central galaxy in high-mass halos (e.g., Maller & Bullock 2004; Voit et al. 2017; Correa et al. 2018). Metal-rich gas can more efficiently cool and condense out of the hot halo due to the increased number of cooling channels. Coupled with the shorter cooling time associated with denser gas, we expect condensation is most prevalent at smaller radii (e.g., see the definitions of the cooling or precipitation radii in Voit et al., 2015; Correa et al., 2018). For metal-rich gas with , the ratio of the cooling to free-fall time is —a characteristic criterion for the formation of thermal instabilities in galaxy halos (Sharma et al., 2012)—at 0.5 (assuming a gas profile following Mathews & Prochaska 2017, cooling losses from Gnat & Sternberg 2007, and a Navarro et al. 1996 mass profile). For metallicities below , gas within 0.25 typically meets these criteria. Certainly the absorbers we find around LRGs have higher column densities on average at smaller impact parameters, and the highest metallicity gas is seen within 0.25 in Figure 5.2. Thus, the high metallicity absorbers could plausibly represent gas precipitated from the corona.
However, the lowest metallicity gas in our combined sample of massive galaxies is also seen in the inner CGM. The [X/H] absorber reported here is found at 0.4 , while the [X/H] (our analysis of this absorber drawn from Wotta et al., 2018) system drawn from the C18 sample is at an impact parameter 0.3 . It is difficult to imagine such low-metallicity gas could have condensed out of the halo, as the typical corona about galaxies with masses similar to our RDR sample are expected to be significantly more metal rich (e.g., the ICM of clusters is often quite metal rich, Mernier et al. 2018, while the Milky Way corona is thought to have , Miller & Bregman 2013). As such, this low-metallicity material could be material accreted from the IGM or stripped from a satellite galaxy. As we discuss below, cold gas like this would be susceptible to relatively rapid evaporation if embedded within a diffuse, hot corona, which makes the direct accretion scenario problematic. On the other hand, extremely-metal-poor dwarf galaxies with are rare enough (Izotov et al., 2018) that it is also difficult to explain the prevalence of very-metal-poor pLLSs/LLSs seen at low redshift through satellite stripping (Lehner et al., 2013; Wotta et al., 2018). Given these difficulties, the origins of the gas are not yet clear. While it is possible to explain the more metal-rich gas as material directly condensed from a hot corona (similar to hot-mode accretion or precipitation models) or through satellite stripping, this cannot universally explain the cold, dense gas we see in the CGM of LRGs.
The ultimate fate of the cool gas we detect is clearly not surviving the fall into the center of the galaxy and fueling star formation. If all of the mass of the cool CGM were to collapse to the central galaxy within a free-fall time (see below) from , , then the galaxy would have , assuming total efficiency. We do not see sSFRs this high for the LRGs frequently (and the emission seen in these galaxies may not trace star formation). It appears that much of the cool CGM gas does not actually make it to the galaxy cold, or it is somehow kept from forming stars if it does reach the central galaxy.
It is not clear that we expect the gas to survive such a fall through the corona of the LRGs. Here we consider the evaporative destruction timescales for dense, cool gas in a hot LRG corona, comparing it with typical free-fall times. We calculate the expected evaporation timescale for a spherical cloud in the halo of an LRG following Maller & Bullock (2004) and the discussion in Gauthier & Chen (2011). Specifically, we evaluate equation 35 from Maller & Bullock (2004) as
We calculate the virial temperature of the median LRG halo (e.g., equation 1 of Oppenheimer et al., 2016) K. We adopt for the metallicity scaling of the cooling function (Appendix A in Maller & Bullock 2004), which is appropriate for gas with 0.1 Z, close to the mean metallicity of the LRG absorbers in our study (0.15 Z). We assume , the halo formation time, is the age of the universe at . We adopt the average density for the absorbers calculated from the best-fit Cloudy models, cm; for the radius of the cloud we use half of the average length scale of the absorbers, kpc. These values are comparable to the median values seen in the Lehner et al. (2018b) sample of pLLSs/LLSs. These assumptions give an evaporation timescale of Myr.
For these clouds to survive to accrete into the center of the galaxy, the evaporation timescale should be greater than the free-fall timescale, . We assess at and 0.5 for an LRG halo, evaluating the expression
where the average density . For the calculation at we adopt the median of our sample for . This yields Gyr. For we integrate the Navarro-Frank-White density profile (Navarro et al., 1996, 1997) for a dark matter halo to determine the mass enclosed within a radius . Using a concentration parameter, , from Shan et al. (2017, their Equation 8 and the Table 2 parameterizations for ) and our values for , we find Myr.
These considerations suggest that if a cloud falls into the halo from the IGM it is unlikely to survive to the center of the LRG before evaporating. As discussed above, however, there is not another obvious way to produce cool gas with metallicities as low as we observe deep in the halos of quiescent, massive galaxies (but see Peeples et al. 2018). More metal-rich gas that condenses from the corona within may roughly remain intact to the centers of the LRGs. However, even if the cool gas from the inner CGM makes it to the center of the galaxy, it evidently does not fuel significant star formation.
AGN feedback could plausibly play a role in this. While LRGs do not house many active AGN (Sadler et al., 2007; Hodge et al., 2008, 2009), the duty cycle of such AGN is not well-known. Most LRG AGNs could be in a quiescent state awaiting the build-up of a cool gas reservoir to fuel its ignition. We note that these calculations have, of course, been done in the absence of magnetic fields, which could significantly affect the conductive evaporation timescales (e.g., Lan & Mo 2018b). However, even if the timescale is increased and the gas is able to survive into the halo core, we still do not see the gas being transformed into stars due to the low SFRs.
There may be evidence in the distribution of HI column densities in the most massive halos that evaporation or some other destructive mechanism is at work. The paucity of HI absorption systems below about the most massive galaxies may be due to the destruction of the clouds that would lead to such absorption. There is no lack of such absorption in lower-mass galaxies (e.g., Keeney et al. 2017, P17). This difference can be seen in Figure 4.4, where the covering factor continues to rise towards lower column densities in the two lower bins of galaxy mass (), while it is consistent with a flat profile for the highest-mass galaxies. If the lower column density gas is on average lower density, it may be more susceptible to evaporative destruction in the dense, hot coronae about the most massive galaxies. Radiative and other feedback effects, especially from AGN, could also play a role in suppressing the amount of low column density gas in the halos of massive galaxies, although this effect should also be present in lower-mass systems.
7 Summary and Concluding Remarks
This paper presents the first results from the RDR survey. We analyze the CGM of 21 LRGs at using UV spectroscopy of background QSOs projected within kpc. We measure the covering factor of HI about the RDR LRGs and determine the metallicity of the strong HI absorbers. Our main results are as follows.
We detect HI absorption in 11/21 sightlines in the CGM of LRGs. We detect four strong HI absorbers: two pLLSs and two LLSs. There is a dearth of absorption between , and we show that any absorbers with most likely arise in the IGM. The covering factor of strong HI absorption about the RDR LRGs at small impact parameters is that seen about lower-mass star-forming galaxies. The covering factor for about LRGs is .
Combining our data with previous galaxy-selected samples (\al@prochaska2017,chen2018; \al@prochaska2017,chen2018), we estimate the covering factors of HI as a function of galaxy mass. The covering factor of gas with in the highest-mass galaxies () is marginally smaller than that of lower-mass galaxies (differences at only the level). If there is a transition between hot- and cold-mode accretion with mass, it shows little signature in strong HI systems. Lower-mass galaxies have significantly higher covering factors for gas with , which is generally not seen in the highest-mass galaxies.
We estimate the metallicity of the four strong HI absorbers about the RDR LRGs. Three of these are metal-rich with [X/H] , while the fourth absorber is metal-poor at [X/H] . Other identifications of metal-poor absorbers in the inner halos of massive galaxies have been reported in the literature. Although the sample is still small, these metal-poor absorbers do not appear to be rare in the CGM of LRGs. The frequency of low-metallicity systems is also similar between star-forming and quiescent galaxies. There is no apparent statistically significant difference in the metallicity distributions of cool CGM gas as a function of stellar mass.
Our observations represent tests of our understanding of the gaseous environments of massive galaxies, including our prevailing theories of how galaxies accrete gas. We have found plentiful dense, cold gas surviving deep in the halos of very massive galaxies. The mass of this gas is similar to that found in lower-mass star-forming galaxies, although it represents a lower fraction of the total baryonic mass in the high-mass halos.
The metal-rich cool gas we observe could plausibly have condensed from the dense halos in these galaxies. If the coronae of these galaxies contain a majority of their halo’s baryons, the inner regions may have conditions in which sustained formation of cool clouds via thermal instabilities is possible. The high column density (high density) gas we observe is concentrated toward the inner regions (at least in projection), as expected if the gas is condensing from the highest-density (and perhaps highest-metallicity) regions of the corona, where the cooling times are shortest. The relatively small velocity offsets between the cold gas and the central galaxies is also consistent with this origin, as has been noted previously for more massive, quiescent galaxies (e.g., Thom et al., 2012; Tumlinson et al., 2013).
At the same time, cooling from the corona cannot be the entire story. The metal-poor absorbers seen in our RDR sample and others (\al@prochaska2017,chen2018; \al@prochaska2017,chen2018), with metallicities below , cannot be explained in this way, as the coronae of these massive galaxies should have metallicities in excess of , if not substantially higher (Miller & Bregman, 2013; Mernier et al., 2018). While in principle such gas could arise as matter being stripped from very-low-mass dwarf satellites, we argue that this is unlikely given their frequency and the small impact parameters at which we see such gas (and it would require the galaxy to maintain its gas to small impact parameters against ram-pressure stripping and probably the dwarf to be on its first pass close to the central halo). Alternatively, the low-metallicity cold gas may have arisen in the IGM and remained cold through a putative accretion shock, surviving against evaporation as it fell. This is not entirely satisfying either, unless we have misunderstood the importance of these phenomena.
If there is a critical mass above which massive cold clouds are unable to exist for a significant amount of time, it is not clear from our measurements. We find similar covering factors with mass and find metal-poor absorbers in high-mass galaxies with roughly the same frequency as seen in lower-mass systems. The only clear distinction between clouds in the highest-mass halos and those about lower-mass galaxies is at column densities , which are essentially absent in the highest-mass galaxies (). The lack of OVI in these halos (see Paper II) also indicates there is little cool gas at very low densities, (e.g., Roca-Fàbrega et al., 2018; Stern et al., 2018), which would give rise to photoionized OVI. This may imply gas below some critical density is readily evaporated, heated, and ionized to a point where it is not detected in our HI selection, while the higher density gas can survive. However, the survival of cold gas at small impact parameters may be telling us that the evaporation within a diffuse hot corona must consider other mitigating factors, such as magnetic fields.
Even though the LRGs have a supply of cool gas at low impact parameters, some process must be at work keeping it from reaching the central galaxy and forming stars. The presence of these clouds coupled with the lack of significant star formation in the central regions implies there is a mechanism actively quenching the formation of stars from this material.
In the second paper in this series (Paper II), we focus on the high-ionization phase of the CGM traced by OVI for the RDR sample of LRGs (see also Zahedy et al., 2018). We are also conducting a survey to observe the CGM of 50 LRGs within with HST/COS. We will employ these data to further delineate the metallicity distribution function of CGM absorbers about LRGs.
Support for this research was provided by NASA through grants HST-AR-12854 and HST-GO-15075 from the Space Telescope Science Institute, which is operated by the Association of Universities for Research in Astronomy, Incorporated, under NASA contract NAS5-26555. Support for the KODIAQ project that enabled some of this work was provided by the NSF through grant AST-1517353 and by NASA SMD through grant NNX16AF52G. This work makes use of data collected at the LBT, an international collaboration among institutions in the United States, Italy and Germany. LBT Corporation partners are: The University of Arizona on behalf of the Arizona Board of Regents; Istituto Nazionale di Astrofisica, Italy; LBT Beteiligungsgesellschaft, Germany, representing the Max-Planck Society, The Leibniz Institute for Astrophysics Potsdam, and Heidelberg University; The Ohio State University, and The Research Corporation, on behalf of The University of Notre Dame, University of Minnesota and University of Virginia. Funding for the Sloan Digital Sky Survey IV has been provided by the Alfred P. Sloan Foundation, the U.S. Department of Energy Office of Science, and the Participating Institutions. SDSS-IV acknowledges support and resources from the Center for High-Performance Computing at the University of Utah. The SDSS web site is www.sdss.org. SDSS-IV is managed by the Astrophysical Research Consortium for the Participating Institutions of the SDSS Collaboration including the Brazilian Participation Group, the Carnegie Institution for Science, Carnegie Mellon University, the Chilean Participation Group, the French Participation Group, Harvard-Smithsonian Center for Astrophysics, Instituto de Astrofísica de Canarias, The Johns Hopkins University, Kavli Institute for the Physics and Mathematics of the Universe (IPMU) / University of Tokyo, the Korean Participation Group, Lawrence Berkeley National Laboratory, Leibniz Institut für Astrophysik Potsdam (AIP), Max-Planck-Institut für Astronomie (MPIA Heidelberg), Max-Planck-Institut für Astrophysik (MPA Garching), Max-Planck-Institut für Extraterrestrische Physik (MPE), National Astronomical Observatories of China, New Mexico State University, New York University, University of Notre Dame, Observatário Nacional / MCTI, The Ohio State University, Pennsylvania State University, Shanghai Astronomical Observatory, United Kingdom Participation Group, Universidad Nacional Autónoma de México, University of Arizona, University of Colorado Boulder, University of Oxford, University of Portsmouth, University of Utah, University of Virginia, University of Washington, University of Wisconsin, Vanderbilt University, and Yale University. Some of the data presented herein were obtained at the W. M. Keck Observatory, which is operated as a scientific partnership among the California Institute of Technology, the University of California and the National Aeronautics and Space Administration. The Observatory was made possible by the generous financial support of the W. M. Keck Foundation. The authors wish to recognize and acknowledge the very significant cultural role and reverence that the summit of Maunakea has always had within the indigenous Hawaiian community. We are most fortunate to have the opportunity to conduct observations from this mountain. This research was supported in part by the Notre Dame Center for Research Computing through the Grid Engine software and, together with the Notre Dame Cooperative Computing Lab, through the HTCondor software.
Facilities: HST (COS, FOS), Keck:I (HIRES), LBT (MODS), Sloan
- Abolfathi et al. (2018) Abolfathi, B., Aguado, D. S., Aguilar, G., et al. 2018, ApJS, 235, 42
- Albareti et al. (2017) Albareti, F. D., Allende Prieto, C., Almeida, A., et al. 2017, ApJS, 233, 25
- Almeida et al. (2008) Almeida, C., Baugh, C. M., Wake, D. A., et al. 2008, MNRAS, 386, 2145
- Anderson et al. (2013) Anderson, M. E., Bregman, J. N., & Dai, X. 2013, ApJ, 762, 106
- Astropy Collaboration et al. (2013) Astropy Collaboration, Robitaille, T. P., Tollerud, E. J., et al. 2013, A&A, 558, A33
- Banerji et al. (2010) Banerji, M., Ferreras, I., Abdalla, F. B., Hewett, P., & Lahav, O. 2010, MNRAS, 402, 2264
- Battisti et al. (2012) Battisti, A. J., Meiring, J. D., Tripp, T. M., et al. 2012, ApJ, 744, 93
- Behroozi et al. (2010) Behroozi, P. S., Conroy, C., & Wechsler, R. H. 2010, The Astrophysical Journal, 717, 379
- Belfiore et al. (2016) Belfiore, F., Maiolino, R., Maraston, C., et al. 2016, MNRAS, 461, 3111
- Birnboim & Dekel (2003) Birnboim, Y., & Dekel, A. 2003, MNRAS, 345, 349
- Blanton & Roweis (2007) Blanton, M. R., & Roweis, S. 2007, AJ, 133, 734
- Bouché et al. (2004) Bouché, N., Murphy, M. T., & Péroux, C. 2004, MNRAS, 354, L25
- Bowen & Chelouche (2011) Bowen, D. V., & Chelouche, D. 2011, ApJ, 727, 47
- Bregman et al. (2018) Bregman, J. N., Anderson, M. E., Miller, M. J., et al. 2018, ApJ, 862, 3
- Bruzual & Charlot (2003) Bruzual, G., & Charlot, S. 2003, MNRAS, 344, 1000
- Burchett et al. (2018) Burchett, J. N., Tripp, T. M., Wang, Q. D., et al. 2018, MNRAS, 475, 2067
- Cameron (2011) Cameron, E. 2011, PASA, 28, 128
- Cappellari (2017) Cappellari, M. 2017, MNRAS, 466, 798
- Cardelli et al. (1989) Cardelli, J. A., Clayton, G. C., & Mathis, J. S. 1989, ApJ, 345, 245
- Cattaneo et al. (2006) Cattaneo, A., Dekel, A., Devriendt, J., Guiderdoni, B., & Blaizot, J. 2006, MNRAS, 370, 1651
- Chabrier (2003) Chabrier, G. 2003, PASP, 115, 763
- Chen (2017) Chen, H.-W. 2017, in Astrophysics and Space Science Library, Vol. 430, Gas Accretion onto Galaxies, ed. A. Fox & R. Davé, 167
- Chen et al. (2010a) Chen, H.-W., Helsby, J. E., Gauthier, J.-R., et al. 2010a, ApJ, 714, 1521
- Chen et al. (2010b) Chen, H.-W., Wild, V., Tinker, J. L., et al. 2010b, ApJL, 724, L176
- Chen et al. (2018) Chen, H.-W., Zahedy, F. S., Johnson, S. D., et al. 2018, MNRAS, 479, 2547
- Conroy (2013) Conroy, C. 2013, ARA&A, 51, 393
- Correa et al. (2018) Correa, C. A., Schaye, J., van de Voort, F., Duffy, A. R., & Wyithe, J. S. B. 2018, MNRAS, 478, 255
- Crain et al. (2010) Crain, R. A., McCarthy, I. G., Frenk, C. S., Theuns, T., & Schaye, J. 2010, MNRAS, 407, 1403
- Danforth et al. (2010) Danforth, C. W., Keeney, B. A., Stocke, J. T., Shull, J. M., & Yao, Y. 2010, ApJ, 720, 976
- Danforth et al. (2016) Danforth, C. W., Keeney, B. A., Tilton, E. M., et al. 2016, ApJ, 817, 111
- Dawson et al. (2013) Dawson, K. S., Schlegel, D. J., Ahn, C. P., et al. 2013, The Astronomical Journal, 145, 10
- Dawson et al. (2016) Dawson, K. S., Kneib, J.-P., Percival, W. J., et al. 2016, The Astronomical Journal, 151, 44
- Dekel & Birnboim (2006) Dekel, A., & Birnboim, Y. 2006, MNRAS, 368, 2
- Eisenstein et al. (2001) Eisenstein, D. J., Annis, J., Gunn, J. E., et al. 2001, AJ, 122, 2267
- Eisenstein et al. (2005) Eisenstein, D. J., Zehavi, I., Hogg, D. W., et al. 2005, ApJ, 633, 560
- Ferland et al. (2013) Ferland, G. J., Porter, R. L., van Hoof, P. A. M., et al. 2013, RMxAA, 49, 137
- Fitzpatrick & Spitzer (1997) Fitzpatrick, E. L., & Spitzer, Jr., L. 1997, ApJ, 475, 623
- Fox & Davé (2017) Fox, A., & Davé, R., eds. 2017, Astrophysics and Space Science Library, Vol. 430, Gas Accretion onto Galaxies
- Fumagalli et al. (2016) Fumagalli, M., O’Meara, J. M., & Prochaska, J. X. 2016, MNRAS, 455, 4100
- Gauthier & Chen (2011) Gauthier, J.-R., & Chen, H.-W. 2011, MNRAS, 418, 2730
- Gauthier et al. (2009) Gauthier, J.-R., Chen, H.-W., & Tinker, J. L. 2009, The Astrophysical Journal, 702, 50
- Gnat & Sternberg (2007) Gnat, O., & Sternberg, A. 2007, The Astrophysical Journal Supplement Series, 168, 213
- Haardt & Madau (1996) Haardt, F., & Madau, P. 1996, ApJ, 461, 20
- Haardt & Madau (2012) —. 2012, ApJ, 746, 125
- Heckman et al. (2017) Heckman, T., Borthakur, S., Wild, V., Schiminovich, D., & Bordoloi, R. 2017, ApJ, 846, 151
- Ho (2008) Ho, L. C. 2008, ARA&A, 46, 475
- Hodge et al. (2008) Hodge, J. A., Becker, R. H., White, R. L., & de Vries, W. H. 2008, AJ, 136, 1097
- Hodge et al. (2009) Hodge, J. A., Zeimann, G. R., Becker, R. H., & White, R. L. 2009, AJ, 138, 900
- Hoshino et al. (2015) Hoshino, H., Leauthaud, A., Lackner, C., et al. 2015, MNRAS, 452, 998
- Howk et al. (2017) Howk, J. C., Wotta, C. B., Berg, M. A., et al. 2017, ApJ, 846, 141
- Howk et al. (2018) Howk, J. C., Berg, M. A., Lehner, N., et al. 2018, in preparation
- Huang et al. (2016) Huang, Y.-H., Chen, H.-W., Johnson, S. D., & Weiner, B. J. 2016, MNRAS, 455, 1713
- Hunter (2007) Hunter, J. D. 2007, Computing In Science & Engineering, 9, 90
- Izotov et al. (2018) Izotov, Y. I., Thuan, T. X., Guseva, N. G., & Liss, S. E. 2018, MNRAS, 473, 1956
- Keeney et al. (2017) Keeney, B. A., Stocke, J. T., Danforth, C. W., et al. 2017, ApJS, 230, 6
- Kereš et al. (2005) Kereš, D., Katz, N., Weinberg, D. H., & Davé, R. 2005, MNRAS, 363, 2
- Lan & Mo (2018a) Lan, T.-W., & Mo, H. 2018a, ApJ, 866, 36
- Lan & Mo (2018b) —. 2018b, ArXiv e-prints, arXiv:1810.11771
- Lehner et al. (2015) Lehner, N., Howk, J. C., & Wakker, B. P. 2015, ApJ, 804, 79 (LHW15)
- Lehner et al. (2016) Lehner, N., O’Meara, J. M., Howk, J. C., Prochaska, J. X., & Fumagalli, M. 2016, ApJ, 833, 283
- Lehner et al. (2007) Lehner, N., Savage, B. D., Richter, P., et al. 2007, ApJ, 658, 680
- Lehner et al. (2018a) Lehner, N., Wotta, C. B., Howk, J. C., et al. 2018a, ApJ, 866, 33
- Lehner et al. (2018b) —. 2018b, in preparation
- Lehner et al. (2013) Lehner, N., Howk, J. C., Tripp, T. M., et al. 2013, ApJ, 770, 138
- Lundgren et al. (2009) Lundgren, B. F., Brunner, R. J., York, D. G., et al. 2009, ApJ, 698, 819
- Maller & Bullock (2004) Maller, A. H., & Bullock, J. S. 2004, MNRAS, 355, 694
- Maraston et al. (2013) Maraston, C., Pforr, J., Henriques, B. M., et al. 2013, Monthly Notices of the Royal Astronomical Society, 435, 2764
- Martin et al. (2005) Martin, D. C., Fanson, J., Schiminovich, D., et al. 2005, ApJL, 619, L1
- Masters et al. (2011) Masters, K. L., Maraston, C., Nichol, R. C., et al. 2011, MNRAS, 418, 1055
- Mathews & Prochaska (2017) Mathews, W. G., & Prochaska, J. X. 2017, ApJ, 846, L24
- McCourt et al. (2012) McCourt, M., Sharma, P., Quataert, E., & Parrish, I. J. 2012, MNRAS, 419, 3319
- Mernier et al. (2018) Mernier, F., Biffi, V., Yamaguchi, H., et al. 2018, ArXiv e-prints, arXiv:1811.01967
- Miller & Bregman (2013) Miller, M. J., & Bregman, J. N. 2013, ApJ, 770, 118
- Moustakas et al. (2006) Moustakas, J., Kennicutt, Jr., R. C., & Tremonti, C. A. 2006, ApJ, 642, 775
- Muzahid et al. (2017) Muzahid, S., Charlton, J., Nagai, D., Schaye, J., & Srianand, R. 2017, ApJL, 846, L8
- Navarro et al. (1996) Navarro, J. F., Frenk, C. S., & White, S. D. M. 1996, ApJ, 462, 563
- Navarro et al. (1997) —. 1997, ApJ, 490, 493
- Nelson et al. (2015) Nelson, D., Genel, S., Vogelsberger, M., et al. 2015, MNRAS, 448, 59
- Nelson et al. (2018) Nelson, D., Pillepich, A., Springel, V., et al. 2018, MNRAS, 475, 624
- Ocvirk et al. (2008) Ocvirk, P., Pichon, C., & Teyssier, R. 2008, MNRAS, 390, 1326
- Oppenheimer & Schaye (2013) Oppenheimer, B. D., & Schaye, J. 2013, MNRAS, 434, 1043
- Oppenheimer et al. (2016) Oppenheimer, B. D., Crain, R. A., Schaye, J., et al. 2016, MNRAS, 460, 2157
- O’Sullivan et al. (2018) O’Sullivan, E., Combes, F., Salomé, P., et al. 2018, A&A, 618, A126
- Padmanabhan (2002) Padmanabhan, T. 2002, Theoretical Astrophysics - Volume 3, Galaxies and Cosmology, 638, doi:10.2277/0521562422
- Peeples et al. (2017) Peeples, M., Tumlinson, J., Fox, A., et al. 2017, The Hubble Spectroscopic Legacy Archive, Tech. rep.
- Peeples et al. (2018) Peeples, M. S., Corlies, L., Tumlinson, J., et al. 2018, ArXiv e-prints, arXiv:1810.06566
- Pérez-Ràfols et al. (2015) Pérez-Ràfols, I., Miralda-Escudé, J., Lundgren, B., et al. 2015, MNRAS, 447, 2784
- Planck Collaboration et al. (2016) Planck Collaboration, Ade, P. A. R., Aghanim, N., et al. 2016, A&A, 594, A13
- Prakash et al. (2016) Prakash, A., Licquia, T. C., Newman, J. A., et al. 2016, The Astrophysical Journal Supplement Series, 224, 34
- Prochaska et al. (2010) Prochaska, J. X., O’Meara, J. M., & Worseck, G. 2010, ApJ, 718, 392
- Prochaska et al. (2017) Prochaska, J. X., Werk, J. K., Worseck, G., et al. 2017, ApJ, 837, 169
- Rahmati et al. (2015) Rahmati, A., Schaye, J., Bower, R. G., et al. 2015, MNRAS, 452, 2034
- Ribaudo et al. (2011) Ribaudo, J., Lehner, N., & Howk, J. C. 2011, ApJ, 736, 42
- Roca-Fàbrega et al. (2018) Roca-Fàbrega, S., Dekel, A., Faerman, Y., et al. 2018, ArXiv e-prints, arXiv:1808.09973
- Rodríguez-Puebla et al. (2017) Rodríguez-Puebla, A., Primack, J. R., Avila-Reese, V., & Faber, S. M. 2017, MNRAS, 470, 651
- Rykoff et al. (2014) Rykoff, E. S., Rozo, E., Busha, M. T., et al. 2014, ApJ, 785, 104
- Rykoff et al. (2016) Rykoff, E. S., Rozo, E., Hollowood, D., et al. 2016, ApJS, 224, 1
- Sadler et al. (2007) Sadler, E. M., Cannon, R. D., Mauch, T., et al. 2007, MNRAS, 381, 211
- Savage & Sembach (1991) Savage, B. D., & Sembach, K. R. 1991, ApJ, 379, 245
- Schlegel et al. (1998) Schlegel, D. J., Finkbeiner, D. P., & Davis, M. 1998, ApJ, 500, 525
- Schneider et al. (2010) Schneider, D. P., Richards, G. T., Hall, P. B., et al. 2010, AJ, 139, 2360
- Serra et al. (2012) Serra, P., Oosterloo, T., Morganti, R., et al. 2012, MNRAS, 422, 1835
- Shan et al. (2017) Shan, H., Kneib, J.-P., Li, R., et al. 2017, ApJ, 840, 104
- Sharma et al. (2012) Sharma, P., McCourt, M., Quataert, E., & Parrish, I. J. 2012, MNRAS, 420, 3174
- Shull et al. (2017) Shull, J. M., Danforth, C. W., Tilton, E. M., Moloney, J., & Stevans, M. L. 2017, ApJ, 849, 106
- Singh et al. (2018) Singh, P., Majumdar, S., Nath, B. B., & Silk, J. 2018, MNRAS, 478, 2909
- Slepian et al. (2017) Slepian, Z., Eisenstein, D. J., Brownstein, J. R., et al. 2017, MNRAS, 469, 1738
- Smailagić et al. (2018) Smailagić, M., Prochaska, J. X., Burchett, J., Zhu, G., & Ménard, B. 2018, The Astrophysical Journal, 867, 106
- Spitzer (1978) Spitzer, L. 1978, Physical processes in the interstellar medium, doi:10.1002/9783527617722
- Stern et al. (2018) Stern, J., Faucher-Giguère, C.-A., Hennawi, J. F., et al. 2018, ApJ, 865, 91
- Stewart et al. (2011) Stewart, K. R., Kaufmann, T., Bullock, J. S., et al. 2011, ApJ, 738, 39
- Telfer et al. (2002) Telfer, R. C., Zheng, W., Kriss, G. A., & Davidsen, A. F. 2002, ApJ, 565, 773
- The Astropy Collaboration et al. (2018) The Astropy Collaboration, Price-Whelan, A. M., Sipőcz, B. M., et al. 2018, The Astronomical Journal, 156, 123
- Thom et al. (2012) Thom, C., Tumlinson, J., Werk, J. K., et al. 2012, ApJL, 758, L41
- Tinker et al. (2017) Tinker, J. L., Brownstein, J. R., Guo, H., et al. 2017, ApJ, 839, 121
- Tumlinson et al. (2017) Tumlinson, J., Peeples, M. S., & Werk, J. K. 2017, ARA&A, 55, 389
- Tumlinson et al. (2011) Tumlinson, J., Thom, C., Werk, J. K., et al. 2011, Science, 334, 948
- Tumlinson et al. (2013) —. 2013, ApJ, 777, 59
- Velander et al. (2014) Velander, M., van Uitert, E., Hoekstra, H., et al. 2014, MNRAS, 437, 2111
- Voit et al. (2015) Voit, G. M., Bryan, G. L., O’Shea, B. W., & Donahue, M. 2015, ApJL, 808, L30
- Voit et al. (2017) Voit, G. M., Meece, G., Li, Y., et al. 2017, ApJ, 845, 80
- Werk et al. (2012) Werk, J. K., Prochaska, J. X., Thom, C., et al. 2012, ApJS, 198, 3
- Wiersma et al. (2009) Wiersma, R. P. C., Schaye, J., & Smith, B. D. 2009, MNRAS, 393, 99
- Wotta et al. (2018) Wotta, C. B., Lehner, N., Howk, J. C., et al. 2018, ApJ, submitted
- Wotta et al. (2016) Wotta, C. B., Lehner, N., Howk, J. C., O’Meara, J. M., & Prochaska, J. X. 2016, ApJ, 831, 95
- Xu et al. (2013) Xu, X., Cuesta, A. J., Padmanabhan, N., Eisenstein, D. J., & McBride, C. K. 2013, MNRAS, 431, 2834
- Yan & Blanton (2012) Yan, R., & Blanton, M. R. 2012, ApJ, 747, 61
- Yoon & Putman (2013) Yoon, J. H., & Putman, M. E. 2013, ApJL, 772, L29
- Yoon & Putman (2017) —. 2017, ApJ, 839, 117
- Yoon et al. (2012) Yoon, J. H., Putman, M. E., Thom, C., Chen, H.-W., & Bryan, G. L. 2012, ApJ, 754, 84
- Zahedy et al. (2018) Zahedy, F. S., Chen, H.-W., Johnson, S. D., et al. 2018, ArXiv e-prints, arXiv:1809.05115
- Zhu et al. (2014) Zhu, G., Ménard, B., Bizyaev, D., et al. 2014, MNRAS, 439, 3139