CANDELS: Constraining the AGN-Merger Connection with Host Morphologies at z~2
Dale D. Kocevski, S.M. Faber, Mark Mozena, Anton M. Koekemoer, Kirpal Nandra, Cyprian Rangel, Elise S. Laird, Marcella Brusa, Stijn Wuyts, Jonathan R. Trump, David C. Koo, Rachel S. Somerville, Eric F. Bell, Jennifer M. Lotz, David M Alexander, Frederic Bournaud, Christopher J. Conselice, Tomas Dahlen, Avashi Dekel, Jennifer L. Donley, James S. Dunlop, Alexis Finoguenov, Antonis Georgakakis, Mauro Giavalisco, Yicheng Guo, Norman A. Grogin, Nimish P. Hathi, Stephanie Juneau, Jeyhan S. Kartaltepe, Ray A. Lucas, Elizabeth J. McGrath, Daniel H. McIntosh, Bahram Mobasher, Aday R. Robaina, David Rosario, Amber N. Straughn, Arjen van der Wel, Carolin Villforth
aa r X i v : . [ a s t r o - ph . C O ] S e p Draft version September 3, 2018
Preprint typeset using L A TEX style emulateapj v. 11/10/09
CANDELS: CONSTRAINING THE AGN-MERGER CONNECTION WITH HOST MORPHOLOGIES AT z ∼ Dale D. Kocevski , S. M. Faber , Mark Mozena , Anton M. Koekemoer , Kirpal Nandra , Cyprian Rangel ,Elise S. Laird , Marcella Brusa , Stijn Wuyts , Jonathan R. Trump , David C. Koo , Rachel S. Somerville ,Eric F. Bell , Jennifer M. Lotz , David M Alexander , Frederic Bournaud , Christopher J. Conselice , TomasDahlen , Avashi Dekel , Jennifer L. Donley , James S. Dunlop , Alexis Finoguenov , Antonis Georgakakis Mauro Giavalisco , Yicheng Guo , Norman A. Grogin , Nimish P. Hathi , St´ephanie Juneau , JeyhanS. Kartaltepe , Ray A. Lucas , Elizabeth J. McGrath , Daniel H. McIntosh , Bahram Mobasher , AdayR. Robaina , David Rosario , Amber N. Straughn , Arjen van der Wel , Carolin Villforth University of California Observatories/Lick Observatory, University of California, Santa Cruz, CA 95064
Draft version September 3, 2018
ABSTRACTUsing
HST /WFC3 imaging taken as part of the Cosmic Assembly Near-infrared Deep ExtragalacticLegacy Survey (CANDELS), we examine the role that major galaxy mergers play in triggering activegalactic nuclei (AGN) activity at z ∼
2. Our sample consists of 72 moderate-luminosity ( L X ∼ − erg s − ) AGN at 1 . < z < . Chandra observations in the
Chandra
Deep Field South, the deepest X-ray observations to date. Employing visual classifications, we haveanalyzed the rest-frame optical morphologies of the AGN host galaxies and compared them to a mass-matched control sample of 216 non-active galaxies at the same redshift. We find that most of theAGN reside in disk galaxies (51 . +5 . − . %), while a smaller percentage are found in spheroids (27 . +5 . − . %).Roughly 16 . +5 . − . % of the AGN hosts have highly disturbed morphologies and appear to be involvedin a major merger or interaction, while most of the hosts (55 . +5 . − . %) appear relatively relaxed andundisturbed. These fractions are statistically consistent with the fraction of control galaxies that showsimilar morphological disturbances. These results suggest that the hosts of moderate-luminosity AGNare no more likely to be involved in an ongoing merger or interaction relative to non-active galaxiesof similar mass at z ∼
2. The high disk fraction observed among the AGN hosts also appears to beat odds with predictions that merger-driven accretion should be the dominant AGN fueling mode at z ∼
2, even at moderate X-ray luminosities. Although we cannot rule out that minor mergers areresponsible for triggering these systems, the presence of a large population of relatively undisturbeddisk-like hosts suggests that secular processes play a greater role in fueling AGN activity at z ∼ Subject headings: galaxies: active — galaxies: evolution — X-rays: galaxies INTRODUCTION
Although it has been established that super-massiveblack holes (SMBHs) lie at the center of most, if not all,massive galaxies (Magorrian et al. 1998), the primarymechanism that turns quiescent black holes into activegalactic nuclei (AGN) is still debated. Galaxy mergershave long been espoused as a possible fueling mechanism [email protected] University of California, Santa Cruz Space Telescope Science Institute Max-Planck-Institut f¨ur Extraterrestrische Physik Imperial College of Science Johns Hopkins University University of Michigan Durham University Laboratoire AIM Paris-Saclay University of Nottingham The Hebrew University Royal Observatory, University of Edinburgh University of Maryland National Observatory of Athens University of Massachusetts Observatories of the Carnegie Institution of Washington Steward Observatory, University of Arizona National Optical Astronomy Observatories University of Missouri University of California, Riverside Institut de Ciencies del Cosmos NASA Goddard Space Flight Center given their effectiveness in dissipating angular momen-tum and funneling gas to the center of galaxies (Barnes& Hernquist 1991; Mihos & Hernquist 1996). This candrive both accretion onto the SMBH and growth of thestellar bulge, which would help explain the tight cor-relations observed between the two (e.g., Gebhardt etal. 2000; Ferrarese & Merritt 2000; Marconi & Hunt 2003;H¨aring & Rix 2004). In fact, recent galaxy merger simu-lations that incorporate a prescription for self-regulatedblack hole growth have successfully reproduced many ob-served properties of AGN and their host galaxies. Thisincludes the correlation between SMBH mass and bulgevelocity dispersion (Di Matteo et al. 2007, Robertson etal. 2006), the quasar luminosity function (Hopkins etal. 2005, 2006a), and the luminosity function of post-quenched red galaxies (Hopkins et al. 2006b). Such sim-ulations have shown that, when coupled with recent AGNfeedback scenarios, major galaxy mergers provide an at-tractive mechanism to both trigger AGN activity andhelp explain the coevolution observed between SMBHsand their hosts (Hopkins et al. 2008).Thus far, however, efforts to detect an AGN-mergerconnection have produced mixed results. At low redshiftsluminous quasi-stellar objects (QSOs) have long beentied to ongoing or past merger activity (Stockton 1982; Kocevski et al.
Fig. 1.—
Redshift versus restframe X-ray luminosity (0.5-8 keV)for sources detected in the 4 Msec observations of the CDF-S andwhich fall in the CANDELS and ERS H -band imaging. The ver-tical lines denote the target redshift range of 1 . < z < .
5. Thedotted horizontal line denotes a luminosity of L x = 10 erg s − ,the maximum X-ray luminosity attributable to starburst galaxies. Canalizo & Stockton 2001; Bennert et al. 2008). HoweverDunlop et al. (2003) find that QSOs at z ∼ . Hubble Space Telescope ( HST ) to ex-amine the host morphologies of X-ray selected AGN outto z ∼ .
3. Grogin et al. (2005) and Pierce et al. (2007),using data from the Great Observatories Origins DeepSurvey (GOODS, Giavalisco et al. 2004) and the All-wavelength Extended Groth strip International Survey(AEGIS, Davis et al. 2007), respectively, find that hostgalaxies at z ∼ z ∼ z ∼ z ∼ z ∼ Fig. 2.—
Mass distribution of FIREWORKS galaxies in GOODS-S with K s < . black ) and AGN hosts ( red ) in the redshiftrange 1 . < z < .
5. The distribution of our mass-matched controlsample of non-active galaxies is shown in blue. ing taken with
HST /WFC3 as part of the Cosmic As-sembly Near-infrared Deep Extragalactic Legacy Survey(CANDELS, Grogin et al. 2011; Koekemoer et al. 2011)with the 4 Msec
Chandra observations of the
Chandra
Deep Field South (CDF-S; Xue et al. 2011), the deepestX-ray observations obtained to date. While
HST /ACSobservations have characterized the restframe ultravio-let structure of galaxies at z > . HST /WFC3 observations move beyond theBalmer break ( λ rest ≥ z ∼ Chandra ob-servations in the CDF-S provides greater sensitivity tomore obscured (and hence fainter) AGN that may havebeen missed in previous studies.In this study we report on the visual classification ofgalaxies hosting X-ray selected AGN in the CDF-S us-ing
HST /WFC3 imaging in the H -band. We examinewhether AGN hosts exhibit an enhancement of merger orinteraction signatures relative to a mass-matched controlsample at the same redshift. Our analysis is presented inthe following manner: § § § § §
6. Whennecessary the following cosmological parameters areused: H = 70kms − Mpc − ; Ω tot , Ω Λ , Ω m = 1 , . , . OBSERVATIONS AND DATA DESCRIPTION
Near-Infrared Imaging
The near-infrared imaging used for this study consistsof WFC3/IR observations of the GOODS-S field in theF125W ( J ) and F160W ( H ) bands obtained as part ofthe Early Release Science program (ERS, Windhorst etal. 2011) and the ongoing CANDELS Multi-Cycle Trea-sury Program. The ERS dataset consists of 10 WFC3/IRtiles with exposure times of 5000s (each bands) coveringthe northern 10 ′ × ′ region of the GOODS-South field.The CANDELS dataset consists of 15 WFC3/IR tilesGN Host Morphologies at z ∼ ∼ ′ × ′ region just south of andadjacent to the ERS imaging. At the time of writing,each tile in the CANDELS dataset has a total exposuretime of 3000s in both the J and H bands.The publicly available ERS data and the CANDELSimaging were reduced as described in Koekemoer etal. (2011). Due to the difference in the exposure timesof the observations, they were combined into separatemosaics with matching pixel scales of 0 . ′′ H -band selected catalog of objectsin each region using the Source Extractor (SExtractor)software (Bertin & Arnouts 1996). Further details of theCANDELS observations and data reduction can be foundin Grogin et al. (2011) and Koekemoer et al. (2011). X-ray Observations
The 4 Msec
Chandra imaging of the CDF-S was ob-tained in 54 observations (obsIDs) over the course of 3
Chandra observing cycles in 2000, 2007 and 2010 us-ing the Advanced CCD Imaging Spectrometer imagingarray (ACIS-I; Garmire et al. 2003). The data werereduced using CIAO v4.2 according to the basic proce-dure described in Laird et al (2009). Before combin-ing the observations (and differing from Laird et al), theastrometry of each obsID was registered to that of theGOODS-MUSYC survey (Gawiser et al 2006) by match-ing the positions of bright X-ray sources to H-band se-lected sources, using the tool reproject aspect . Sourcedetection was carried out according to the method de-scribed in Laird et al (2009). A total of 569 sources weredetected to a Poisson probability limit (i.e. the wavdetectfalse-positive threshold) of 4e-6. SAMPLE SELECTION
Our AGN selection is based on X-ray detections in the
Chandra z ∼ ∼ − Chandra sensitivitywindow, which peaks at 1.5 keV, potentially allowing usto detect even heavily obscured SMBHs. At the targetredshift of z ∼
2, the luminosity limit of these observa-tions is roughly L x ∼ erg s − , the maximum X-rayluminosity thought to be attributable to star formationprocesses (Bauer et al. 2002). This allows us to probethe entire population of moderate-luminosity AGN athigh redshifts for the first time, while ensuring that thesample is not significantly contaminated by star forminggalaxies without active nuclei.To identify H -band counterparts to the X-ray sourceswe employed a maximum likelihood technique introducedby Sutherland & Saunders (1992) and described in Brusaet al. (2005). A total of 350/569 X-ray sources in the4 Msec catalog fall in the area covered by the ERS andCANDELS imaging. Of these, 322 were reliably matchedto an H -band detected object. Four spurious associa-tions are expected among these matches. Redshifts forthe matched hosts were obtained from a variety of pub-licly available spectroscopic redshift catalogs (the bulkof these redshifts come from Silverman et al. 2010) andthe photometric redshifts in the FIREWORKS catalog(Wuyts et al. 2008). The latter includes photometric redshift information for galaxies with K s < . . < z < .
5, ofwhich 22 are new detections not found in the shallower 2Msec observations of the CDF-S (Luo et al. 2008). Theredshift-luminosity distribution of these sources is shownin Figure 1. The sample has a median luminosity of7 . × erg s − in the 0.5-8 keV band (rest frame).Five sources have luminosities in excess of 10 erg s − ,while the faintest source has a luminosity of 1 . × erg s − . Of the 72 H -band counterparts, 22 fall withinthe ERS region, while 50 are located in the CANDELSregion. These 72 galaxies comprise the primary sampleof AGN hosts studied in this analysis. Mass-Matched Control Sample
To properly compare the morphologies of the AGNhosts to similar non-active galaxies, we constructed acontrol sample consisting of galaxies with masses sim-ilar to those of the AGN hosts. Masses for both theactive and non-active populations were obtained fromthe FIREWORKS catalog, where they were estimatedby modeling each galaxy’s observed spectral energy dis-tribution (SED) with Bruzual & Charlot (2003) stellarpopulation synthesis models. Non-thermal nuclear emis-sion has previously been shown to not significantly con-taminate the rest-frame optical and infrared emissionof galaxies hosting moderate-luminosity AGN such asthose in our sample (e.g. Barger et al. 2005; Bundy etal. 2008). An examination of the AGN host SEDs con-firms this finding for >
90% of the sample. Therefore weproceed under the assumption that the FIREWORKS-derived redshifts and masses for the bulk of the AGNsample are not systematically biased compared to thecontrol sample.For each AGN host, we randomly selected threeunique, non-active galaxies from the FIREWORKS cata-log whose masses are within a factor of 2 of the AGN hostmass (i.e., M AGN / ≤ M gal ≤ M AGN ). Because of thedifference in depth between the ERS and CANDELS ob-servations, the control sample is selected separately forAGN in each region, using only non-active galaxies inthe corresponding region. The number of control galax-ies was restricted to three per AGN due to the limitednumber of massive galaxies available in each field, espe-cially the smaller ERS region.Constructing a mass-matched control sample for thisanalysis is vital, as the bulk of the galaxy populationat z ∼ K s -selected FIREWORKSgalaxies (with K s < . . < z < .
5, as well as the masses of the AGN hostsand their corresponding control galaxies. Without tak-ing mass into consideration, any control sample selectedat z ∼ VISUAL CLASSIFICATION
To determine if the AGN host galaxies exhibit mergeror interaction signatures more often than non-active Kocevski et al.
Fig. 3.—
Examples of AGN host galaxies that were classified as having spheroid and disk morphologies, as well as two galaxies experiencingdisruptive interactions. Thumbnails on the top row are WFC3/IR images taken in the F160W ( H ) band (rest-frame optical), while those onthe bottom row are from ACS/WFC in the F775W ( i ) band (rest-frame ultraviolet). These images demonstrate that accurately classifyingthe morphology of these galaxies at z ∼ H -band imaging. galaxies of similar mass, we have visually classified themorphologies of both populations. Compared to moreautomated classification techniques, this type of visualinspection allows us to better pick up low surface bright-ness features and faint signatures of past interactionsthat can be missed using quantitative measures such asconcentration and asymmetry (Kartaltepe et al. 2010).Classifiers were asked to determine the predominant mor-phological type of each galaxy and the degree to which itwas disturbed. These inspections were performed blind(i.e., the AGN hosts and control galaxies were mixed)and done primarily in the H -band, although classifierswere also given supplemental V , z , and J -band imagesof each galaxy in order to provide additional color infor-mation. The inspectors had access to FITS images in allfour bands so as to manipulate the contrast and stretchof an image if needed. Because of the difference in depthof the ERS and CANDELS imaging, the classificationswere carried out separately for the subsamples in eachregion.The classification of each galaxy was split into twocategories, with the first being a general morphologicalclassification and the second a disturbance classification.For the former, inspectors were asked to choose amongthe following broad morphologies: Disk, Spheroid, Ir-regular/Peculiar, Point-Like . Classifiers were allowed tochoose as many as were applicable to a given galaxy. Forexample, a spiral galaxy with a substantial bulge wouldbe classified as having both a disk and spheroid compo-nent.The second category is meant to gauge the degree towhich a galaxy is distorted or disturbed, presumably asa result of a recent interaction. The disturbance classeswithin this category were designed to pick up not onlymajor mergers, but also weak interactions that may leadto only minor disturbances or asymmetries, as well asfaint signatures of past merger activity. Classifiers wereasked to choose one of the following disturbance classes: • Merger:
Highly disturbed with multiple nucleiand/or strong distortions in a single coalescing sys-tem. • Interaction:
Two distinct galaxies showing distor-tions and interaction features such as tidal arms. • Distorted:
Single asymmetric or distorted galaxy with no visible interacting companion. • Double Nuclei:
Multiple nuclei in a single coalescedsystem. • Close Pair:
Near-neighbor pair in which both areundisturbed. • Undisturbed:
None of the above.While the
Merger and
Interaction classes togetheridentify galaxies in the various stages of a highly disrup-tive interaction, the
Distorted class is meant to flag moresubtle signatures of minor interactions where the com-panion galaxy may not be visible. To improve our statis-tics, we have grouped several of the interaction classesinto the following categories according to the severity ofthe observed disturbance: • Disturbed I:
Highly disturbed systems. Includesgalaxies in the
Merger and
Interaction classes. • Disturbed II:
High to moderately disturbed systems.Includes all galaxies in the
Disturbed I class, as wellas systems with distorted or asymmetric morpholo-gies (those in the
Distorted class), and galaxies with
Double Nuclei . • Companion:
Includes all galaxies that have a neigh-bor within 1 . ′′ z ∼ Close Pair class.Each galaxy in the sample of 72 AGN hosts and 216control galaxies was examined by 12 independent clas-sifiers. The individual classifications were combinedon a galaxy-by-galaxy basis using the consensus of thegroup. For example, a galaxy was assigned a morpholog-ical classification only if a majority (greater than half)of the inspectors agreed that that morphological com-ponent was present. Multiple morphological classifica-tions per galaxy were possible. Since the disturbanceclasses are mutually exclusive, disturbance classificationswere based on the class most often chosen for a givengalaxy. In cases where the inspectors were evenly split,the galaxy was assigned the more disturbed interpreta-tion. For example, a galaxy flagged as an
Interaction andas
Disturbed by an equal number of inspectors would beassigned the
Interaction class. These combined classifi-cations were then used to calculate the fraction of AGNand control galaxies with a given morphology or distur-GN Host Morphologies at z ∼ n , and the visual morphologies. Only a hand-ful of sources show signs of point source contamination,as evidenced by best-fit Sersic profiles that are steeperthan a de Vaucouleurs profile ( n >
4; de Vaucouleurs1948). These sources were predominantly the most lu-minous AGN in our sample and were easily identified vi-sually as extended spheroids, despite the added nuclearemission. For these reasons we do not believe that pointsource contamination has biased or strongly affecting thevisual classification of the bulk of the AGN host sample. RESULTS
Sample WFC3 H -band images of AGN host galaxiesthat exhibit a range of morphologies and disturbancesare shown in Figure 3, while the combined results of thevisual analysis of the AGN hosts and control galaxies areshown in Figure 4 and listed in Table 1. In the followingsections we first present the morphological breakdown ofthese galaxies and later discuss the frequency of distur-bances observed among them.A brief note regarding nomenclature: in the followingsections the term disks (or all disks ) refers to all galaxieswith a visible disk, including those with and without adiscernible central bulge. We will use the term pure disks to refer to disk-like galaxies where no bulge is discernible.On the other hand, spheroids (or pure spheroids ) refers tospheroidal galaxies with no discernible disk component.At times we will use the term all spheroids to refer toboth pure spheroids and the bulge component of galaxieswith a visible disk. Host Morphologies
Shown on the left side of Figure 4 is the fraction ofAGN hosts and control galaxies that were classified ashaving disk, spheroid, point-like, or irregular morpholo-gies. The error bars on each fraction reflect the 68.3%binomial confidence limits given the number of sources ineach category, calculated using the method of Cameronet al. (2010). For the disk fraction, we show both thefraction of AGN found in pure disks (i.e., those with nodiscernible bulge) and the fraction hosted by any diskygalaxy (i.e., disks with and without a discernible bulge).If AGN activity at z ∼ . +5 . − . %) and consistent with the fraction observedamong the control sample (18 . +2 . − . %). This is the firstindication that AGN are not found in substantially dis-turbed galaxies more often than their non-active coun-terparts at this redshift. In fact, a high fraction of the AGN are found in galaxies with a visible disk, a compo-nent which is unlikely to have survived a major mergerin the recent past. We find disks to be the most commonsingle morphology assigned to the AGN hosts, makingup 51 . +5 . − . % of the entire sample. Two-thirds of thesegalaxies (67.6%) also exhibit a prominent bulge compo-nent, while 32.4% of the disk galaxies show no discerniblecentral bulge. Of the remaining hosts, pure spheroidscomprise 27 . +5 . − . % of the entire sample, while point-likesources constitute 9 . +4 . − . %.Despite the prevalence of disks among the AGN hosts,we find that the active galaxies are more often associ-ated with spheroid morphologies than their non-activecounterparts. Pure spheroids make up 27 . +5 . − . % of theAGN hosts versus only 16 . +2 . − . % of the massive con-trol galaxies. Disks, on the other hand, are more com-mon among the control sample, comprising 69 . +2 . − . % ofthe non-active galaxies, but only 51 . +5 . − . % of the AGNhosts. The smaller disk fraction among the active galax-ies compared to the control galaxies is significant at the99.6% level, assuming a binomial error distribution.The fact that the AGN tend to favor more spheroid-dominated hosts is further illustrated in Figure 5, wherewe show the fraction of active and control galaxies clas-sified as pure disks, disks with central bulges and purespheroids. The AGN host morphologies are skewed to-ward more spheroid-dominated systems as they show anexcess of pure spheroid morphologies relative to the con-trol sample and a deficit of pure disk morphologies. Wefind that bulgeless, pure disks constitute 30 . +3 . − . % ofthe control population while making up only 16 . +5 . − . %of the AGN host galaxies. These findings suggests thatthe trend observed at lower redshifts, that AGN hostsare more spheroid-dominated relative to similarly mas-sive non-active galaxies (e.g., Grogin et al. 2005; Pierceet al. 2007), continues to some extent out to z ∼ L X = 10 ergs − . These subsamples includes 32 and 40 AGN, re-spectively, out of the full sample of 72 at z ∼
2. Themorphological breakdown of these subsamples is listedin Table 1. We find no increase in the irregular frac-tion among the more luminous AGN, but we do observea dramatic reversal in the spheroid and disk fractions:spheroids constitute 40 . +9 . − . % of the galaxies hosting themore X-ray luminous AGN, while the disk fraction dropsto 34 . +9 . − . %. This is compared to a spheroid and diskfraction of 18 . +7 . − . % and 68 . +6 . − . %, respectively, for thelower luminosity sample. This finding agrees with localhost properties, where luminous AGN and QSOs tend tobe more often associated with early-type hosts (Kauff-mann et al. 2003). Interaction Signatures
The fraction of active and non-active galaxies that ex-hibit various levels of disturbance in their morphologies isshown on the right side of Figure 4 and the bottom partof Table 1. We find that 16 . +5 . − . % of the AGN hosts Kocevski et al. Fig. 4.—
Fraction of AGN hosts ( red triangles ) and control galaxies ( blue squares ) at 1 . < z < . Pure Disk class includes only disks without a central bulge. The
Pure Disk class is a subsample of the
AllDisks class, which includes disks with and without a central bulge. Similarly, the
Pure Spheroid class includes only spheroids with nodiscernible disk component. The
All Spheroids class includes both
Pure Spheroids and disk galaxies with a central bulge. The
DisturbedI class is limited to heavily disturbed galaxies in a clear merger or interaction. The
Disturbed II class includes galaxies in the
Disturbed I class, as well as those showing even minor asymmetries in their morphologies. See text for details. are involved in highly disruptive mergers or interactionsand fall in the
Disturbed I category. This percentageis statistically no different than the fraction of similarlydisturbed non-active galaxies (15 . +2 . − . %). If we includegalaxies showing minor asymmetries in their morpholo-gies and those with double nuclei (i.e., Disturbed II sys-tems), the fraction of disturbed active galaxies increasesto 44 . +5 . − . %. This is below the percentage of controlgalaxies that fall in the same category (48 . +3 . − . %), butthe fractions are again statistically equivalent. In fact,for all of the distortion classes we considered, the proper-ties of the AGN hosts are not significantly different thanthe non-active sample. This includes the fraction show-ing clear merger and interaction signatures (see Table1), those with only a minor asymmetry in their mor-phology (30 . +4 . − . % for AGNs versus 32 . +3 . − . % for thecontrol sample, respectively), and the frequency of com-panions within 1 . ′′ . +5 . − . % versus19 . +3 . − . %).The most common disturbance class assigned to boththe AGN hosts and control galaxies is Undisturbed , mak-ing up 55 . +5 . − . % and 52 . +3 . − . % of each population, re-spectively. This suggests that a majority of the AGNat z ∼ H -band imag-ing. Significantly, these results do not change when welimit our analysis to the more luminous AGN in the sam-ple: the Undisturbed fraction is still 46 . +8 . − . % even whenonly AGN with L X > erg s − are considered. Fur-thermore, the fraction of hosts in the Disturbed I and
Disturbed II categories are 18 . +8 . − . % and 53 . +8 . − . %, re-spectively, and in rough agreement with the percentagesfound for the full sample. DISCUSSION
In summary, our primary findings are: i) moderateluminosity, X-ray selected AGN at z ∼ . +5 . − . %) of theAGN reside in galaxies with discernible disks. Therefore,based on our visual classifications, we do not find a strongconnection between highly disruptive major mergers andmoderate luminosity AGN activity at z ∼ ∼ few 100 Myr), the most obvious signatures ofmorphological disturbance will have faded by the time itis identified as an X-ray bright AGN (Lotz et al. 2010).Hydrodynamic simulations of SMBH growth in galaxymergers do predict such a delay (Hopkins et al. 2006),and there is observational evidence for a delay of order ∼
250 Myr between starburst and AGN activity (Wild,Heckman, & Charlot 2010).While these two effects can help explain the lack of ob-vious merger signatures among the AGN hosts, the highGN Host Morphologies at z ∼ TABLE 1Visual Classification Results
AGN Control AGN AGNClassification Hosts Galaxies L X < erg s − L X > erg s − Pure Disk 16 . +5 . − . % 30 . +3 . − . % 21 . +8 . − . % 12 . +8 . − . %All Disks 51 . +5 . − . % 69 . +2 . − . % 68 . +6 . − . % 34 . +9 . − . %Pure Spheroid 27 . +5 . − . % 16 . +2 . − . % 18 . +7 . − . % 40 . +9 . − . %All Spheroids 62 . +5 . − . % 55 . +3 . − . % 65 . +6 . − . % 62 . +7 . − . %Irregular 16 . +5 . − . % 18 . +2 . − . % 21 . +8 . − . % 06 . +7 . − . %Point-like 09 . +4 . − . % 05 . +2 . − . % 02 . +5 . − . % 18 . +8 . − . %Disturbed I 16 . +5 . − . % 15 . +2 . − . % 15 . +7 . − . % 18 . +8 . − . %Disturbed II 44 . +5 . − . % 48 . +3 . − . % 36 . +8 . − . % 53 . +8 . − . %Companion 19 . +5 . − . % 19 . +3 . − . % 18 . +7 . − . % 21 . +8 . − . %Undisturbed 55 . +5 . − . % 52 . +3 . − . % 63 . +7 . − . % 46 . +8 . − . % Note . — The
Pure Disk and
Pure Spheroid classes are included in the
All Disks and
All Spheroids classes, respectively. Likewise, the
DisturbedI class is a subset of the
Disturbed II class. disk fraction we observe is harder to reconcile with themerger picture of AGN fueling. This finding that disk-like morphologies are prevalent among active galaxies atthis redshift agrees with the recent findings of Schaw-inski et al. (2011), who examined a smaller sample ofAGN at z ∼ major galaxy-galaxy inter-action (e.g. Bournaud et al. 2011), making it highly un-likely that their nuclear activity is being fueled by a ma-jor merger-driven process. It is more likely that the nu-clear activity in these disk galaxies is being fueled by thestochastic accretion of cold gas, possibly triggered by adisk instability or minor interaction that did not sub-stantially perturb the large-scale structure of the galaxy.Despite the prevalence of disk-like morphologies, wealso find the AGN are more often associated withspheroids than their non-active counterparts. Whilethe connection between spheroid-dominated galaxies andAGN has been well established at lower redshifts, thisis the first such finding at z ∼
2, where there is ev-idence that the canonical mass-morphology relationshipappears to break down (McGrath et al. 2008; van der Welet al. 2011). Even in an era where massive galaxies pre-dominately have disk-like morphologies, our observationssuggest SMBH continue to be preferentially embedded inspheroidal systems.At higher X-ray luminosities ( L X > erg s − ) wealso observe a shift in the overall AGN host populationtoward spheroids. Although many of these galaxies ap-pear undisturbed, they may have been triggered by amerger event in the recent past and have since suffi-ciently relaxed. In the evolutionary sequence of a merger-triggered QSO presented in Hopkins et al. (2007), thesesystems would be in the post-blowout phase, when nu-clear activity and X-ray luminosity are in gradual decline.Our finding that more luminous AGN are more oftenassociated with spheroids generally agrees with the lumi-nosity and redshift dependent AGN fueling model pre- All DisksControlAGNPure Disks Disk + Spheroid(Disk with Bulge) Pure SpheroidsAll Spheroids
Fig. 5.—
Fraction of AGN hosts and control galaxies at 1 . < z < . sented by Hopkins & Hernquist (2006; hereafter HH06),which proposes that luminous AGN and QSOs are largelytriggered by major mergers, while lower luminosity AGNare fueled by the random accretion of gas via internal,secular processes. However this model also predicts thatby z ∼ L x ∼ erg s − ). This is because merger-driven accretion in this model is tied to the cosmolog-ical galaxy merger rate, which increases rapidly withredshift (Conselice et al. 2003; Kartaltepe et al. 2007)whereas quiescent accretion is related to the mass func-tion and gas fraction of late-type galaxies, which evolvemore slowly.The HH06 model predicts that at z = 2 the numberdensity of quiescently accreting AGN will not equal thatof merger-fueled AGN until roughly two orders of magni-tude below the knee in the AGN luminosity function. Inthe hard X-ray band this knee occurs at a luminosity ofroughly L X ∼ erg s − (Aird et al. 2010). This meansthe predicted X-ray luminosity at which an equal frac-tion of AGN are fueled by quiescent and merger-triggeredaccretion at z = 2 is roughly L X ∼ erg s − . If we as-sume that disk-like hosts are fueling their AGN via inter-nal processes and have not experienced a major merger Kocevski et al.in the recent past, then this prediction is at odds with thehigh disk fraction we observe at L X ∼ erg s − , an or-der of magnitude above this luminosity. We find that theluminosity at which an equal fraction of AGN are hostedby disk and spheroid galaxies is roughly L X ∼ ergs − . This finding suggests that the stochastic fueling ofSMBHs is far more prevalent at moderate luminositiesthan predicted by the HH06 model.This apparent disagreement with the HH06 fuelingmodel was previously reported at lower redshifts byGeorgakakis et al. (2009), who found that the contri-bution to the X-ray luminosity function at z ∼ L X > erg s − ) at z ∼ z ∼
1, and by Schawinski et al. (2011) at z ∼
2, basedon the large disk fraction found among AGN hosts. It hasalso been proposed by Mullaney et al. (2011) based onthe average specific star formation rates (SSFR) of AGNhosts out to z ∼
3. They find that a vast majority ofhosts have SSFRs consistent with the star forming mainsequence (Noeske et al. 2007) and that less than 10%appear to be undergoing a starburst phase. From thisthey conclude that the nuclear activity in these galaxiesis being fueled by internal mechanisms rather than vio-lent mergers. A similar conclusion was also reached byAllevato et al. (2011) based on the projected clusteringof AGN in the COSMOS field out to z ∼ . z > ∼ M ⊙ yr − ; Cacciato & Dekel 2011), potentially fuel-ing increased AGN activity in disk galaxies without theneed for galaxy-galaxy mergers (Bournaurd et al. 2011,in prep). Of course, the disagreement between the high disk frac-tion we observe and the merger-dominated fueling modelis predicated on the assumption that disk-like hosts havenot experienced a merger in the recent past. The twocan be reconciled if these disks have instead survived orreformed following a merger event. Numerical simula-tions have shown that disks can reform after a mergerif the interacting systems are gas rich (Robertson etal. 2006; Bundy et al. 2010), although it has been arguedthat such interactions are not conducive to the fuelingof SMBHs (Hopkins & Hernquist 2009). Alternatively,minor mergers provide a means to trigger AGN-activitywithin galaxies without entirely destroying their pre-existing morphology. Semi-analytic cosmological galaxyformation models in which all AGN activity is assumedto be triggered by mergers (Somerville et al. 2008) dopredict that the average merger event that triggers anAGN with L X > erg s − has a mass ratio of 1:8as opposed to the more disruptive 1:1 or 1:2 mergers(Somerville et al. 2011). Coupled with a time delay be-tween the merger and the visibility of the AGN, the sig-natures of these mergers could prove difficult to detect.Therefore, since we cannot rule out such interactions,minor mergers would seem to be one of the remainingways to reconcile the merger-dominated fueling modelwith the high disk fraction and lack of disturbed mor-phologies that we observe. CONCLUSIONS
To explore if major galaxy mergers are the primarymechanism fueling AGN activity at z ∼
2, we have used
HST /WFC3 imaging to examine the rest-frame opticalmorphologies of galaxies hosting moderate-luminosity, X-ray selected AGN at z = 1 . − .
5. Employing visualclassifications, we have determined both the predominantmorphological type of these galaxies and the frequencyat which they exhibit morphological disturbances indica-tive of recent interactions. To determine if the AGNhosts show merger or interaction signatures more oftenthan similar non-active galaxies, we have also classifieda sample of mass-matched control galaxies at the sameredshift.First, we find that just over half of the AGN reside indisk galaxies (51 . +5 . − . %), while a smaller percentage arefound in spheroids (27 . +5 . − . %) and systems with irregu-lar morphologies (16 . +5 . − . %). This high disk fraction isalso observed among the control galaxies (69 . +2 . − . % ofwhich are disks), which indicates disk-like morphologiesare prevalent among all massive galaxies at this redshift,regardless of whether they host an active nucleus. Infact, we find the AGN to be more often associated withspheroidal galaxies compared to their non-active coun-terparts. Pure spheroids account for 27 . +5 . − . % of theactive galaxies, while comprising only 16 . +2 . − . % of thecontrol galaxies. At X-ray luminosities above 10 ergs − we observe a reversal in the morphological make-upof the AGN hosts, with spheroids and disks comprising40 . +9 . − . % and 34 . +9 . − . % of the sample, respectively.Second, we find that 16 . +5 . − . % of the AGN hosts havehighly disturbed morphologies and appear to be involvedin a major merger or interaction, while 44 . +5 . − . % showat least some disturbance, including minor asymmetriesGN Host Morphologies at z ∼ . +5 . − . % and 52 . +3 . − . %, respectively). Theseresults suggest the AGN hosts are no more likely to beinvolved in an ongoing major merger or interaction thannon-active galaxies of similar mass.Finally, the high disk fraction observed among theAGN hosts appears to be at odds with predictions thatmajor merger-driven accretion should be the dominantAGN fueling mode at z ∼
2. The presence of a largepopulation of relatively undisturbed disk-like hosts sug-gests that either secular evolution, disk instability-drivenaccretion, minor mergers, or a combination of the three, play a greater role in triggering AGN activity at theseredshifts than previously thought. In a forthcoming pa-per we plan to calculate the fraction of the AGN X-ray lu-minosity function attributable to these disk-hosted AGNin order to quantify any discrepancy between our obser-vations and merger-triggered AGN fueling models.Support for Program number HST-GO-12060 was pro-vided by NASA through a grant from the Space Tele-scope Science Institute, which is operated by the Asso-ciation of Universities for Research in Astronomy, Incor-porated, under NASA contract NAS5-26555. Further-more, D.K. is funded in part by the NSF under GrantNo. AST-0808133.
REFERENCESAird, J., Nandra, K., Laird, E. S., Georgakakis, A., Ashby,M. L. N., Barmby, P., Coil, A. L., Huang, J.-S., et al. 2010,MNRAS, 401, 2531Allevato, V., Finoguenov, A., Cappelluti, N., Miyaji, T., Hasinger,G., Salvato, M., Brusa, M., Gilli, R., et al. 2011, ApJ, 736, 99Barnes, J. E. & Hernquist, L. E. 1991, ApJ, 370, L65Bauer, F. E., Alexander, D. M., Brandt, W. N., Hornschemeier,A. E., Vignali, C., Garmire, G. P., & Schneider, D. P. 2002, AJ,124, 2351Bennert, N., Canalizo, G., Jungwiert, B., Stockton, A., Schweizer,F., Peng, C. Y., & Lacy, M. 2008, ApJ, 677, 846Bertin, E. & Arnouts, S. 1996, A&AS, 117, 393Bournaud, F., Chapon, D., Teyssier, R., Powell, L. C., Elmegreen,B. G., Elmegreen, D. M., Duc, P.-A., Contini, T., et al. 2011,ApJ, 730, 4Brusa, M., Comastri, A., Daddi, E., Pozzetti, L., Zamorani, G.,Vignali, C., Cimatti, A., Fiore, F., et al. 2005, A&A, 432, 69Bruzual, G. & Charlot, S. 2003, MNRAS, 344, 1000Bundy, K., Scarlata, C., Carollo, C. M., Ellis, R. S., Drory, N.,Hopkins, P., Salvato, M., Leauthaud, A., et al. 2010, ApJ, 719,1969Cacciato, M. & Dekel, A. 2011, in prepCameron, E. 2010, ArXiv e-prints 1012.0566Canalizo, G. & Stockton, A. 2001, ApJ, 555, 719Conselice, C. J., Bershady, M. A., Dickinson, M., & Papovich, C.2003, AJ, 126, 1183Davis, M., Guhathakurta, P., Konidaris, N. P., Newman, J. A.,Ashby, M. L. N., Biggs, A. D., Barmby, P., Bundy, K., et al.2007, ApJ, 660, L1de Vaucouleurs, G. 1948, Annales d’Astrophysique, 11, 247Dekel, A., Sari, R., & Ceverino, D. 2009, ApJ, 703, 785Di Matteo, P., Combes, F., Melchior, A.-L., & Semelin, B. 2007,A&A, 468, 61Dunlop, J. S., McLure, R. J., Kukula, M. J., Baum, S. A., O’Dea,C. P., & Hughes, D. H. 2003, MNRAS, 340, 1095Elmegreen, B. G., Elmegreen, D. M., Vollbach, D. R., Foster,E. R., & Ferguson, T. E. 2005, ApJ, 634, 101Ferrarese, L. & Merritt, D. 2000, ApJ, 539, L9Gabor, J. M., Impey, C. D., Jahnke, K., Simmons, B. D., Trump,J. R., Koekemoer, A. M., Brusa, M., Cappelluti, N., et al. 2009,ApJ, 691, 705Garmire, G. P., Bautz, M. W., Ford, P. G., Nousek, J. A., &Ricker, Jr., G. R. 2003, in X-Ray and Gamma-Ray Telescopesand Instruments for Astronomy. Edited by Joachim E.Truemper, Harvey D. Tananbaum. Proceedings of the SPIE,Volume 4851, pp. 28-44 (2003)., ed. J. E. Truemper & H. D.Tananbaum, 28–44Gawiser, E., van Dokkum, P. G., Herrera, D., Maza, J.,Castander, F. J., Infante, L., Lira, P., Quadri, R., et al. 2006,ApJS, 162, 1Gebhardt, K., Bender, R., Bower, G., Dressler, A., Faber, S. M.,Filippenko, A. V., Green, R., Grillmair, C., et al. 2000, ApJ,539, L13 Genzel, R., Tacconi, L. J., Eisenhauer, F., F¨orster Schreiber,N. M., Cimatti, A., Daddi, E., Bouch´e, N., Davies, R., et al.2006, Nature, 442, 786Georgakakis, A., Coil, A. L., Laird, E. S., Griffith, R. L., Nandra,K., Lotz, J. M., Pierce, C. M., Cooper, M. C., et al. 2009,MNRAS, 397, 623Giavalisco, M., Ferguson, H. C., Koekemoer, A. M., Dickinson,M., Alexander, D. M., Bauer, F. E., Bergeron, J., Biagetti, C.,et al. 2004, ApJ, 600, L93Grogin, N. A., Conselice, C. J., Chatzichristou, E., Alexander,D. M., Bauer, F. E., Hornschemeier, A. E., Jogee, S.,Koekemoer, A. M., et al. 2005, ApJ, 627, L97Grogin, N. A., Kocevski, D. D., Faber, S. M., Ferguson, H. C.,Koekemoer, A. M., Riess, A. G., Acquaviva, V., Alexander,D. M., et al. 2011, ArXiv e-prints, 1105.3753H¨aring, N. & Rix, H.-W. 2004, ApJ, 604, L89Hopkins, P. F., Bundy, K., Hernquist, L., & Ellis, R. S. 2007,ApJ, 659, 976Hopkins, P. F. & Hernquist, L. 2006, ApJS, 166, 1—. 2009, ApJ, 694, 599Hopkins, P. F., Hernquist, L., Cox, T. J., Di Matteo, T.,Robertson, B., & Springel, V. 2005, ApJ, 630, 716Hopkins, P. F., Hernquist, L., Cox, T. J., & Kereˇs, D. 2008,ApJS, 175, 356Hopkins, P. F., Hernquist, L., Cox, T. J., Robertson, B., DiMatteo, T., & Springel, V. 2006a, ApJ, 639, 700Hopkins, P. F., Hernquist, L., Cox, T. J., Robertson, B., &Springel, V. 2006b, ApJS, 163, 50Jahnke, K., S´anchez, S. F., Wisotzki, L., Barden, M., Beckwith,S. V. W., Bell, E. F., Borch, A., Caldwell, J. A. R., et al. 2004,ApJ, 614, 568Kartaltepe, J. S., Sanders, D. B., Le Floc’h, E., Frayer, D. T.,Aussel, H., Arnouts, S., Ilbert, O., Salvato, M., et al. 2010,ApJ, 721, 98Kartaltepe, J. S., Sanders, D. B., Scoville, N. Z., Calzetti, D.,Capak, P., Koekemoer, A., Mobasher, B., Murayama, T., et al.2007, ApJS, 172, 320Kartaltepe, J. S., et al. 2011, in prepKauffmann, G., Heckman, T. M., Tremonti, C., Brinchmann, J.,Charlot, S., White, S. D. M., Ridgway, S. E., Brinkmann, J., etal. 2003, MNRAS, 346, 1055Koekemoer, A. M., Faber, S. M., Ferguson, H. C., Grogin, N. A.,Kocevski, D. D., Koo, D. C., Lai, K., Lotz, J. M., et al. 2011,ArXiv e-prints, 1105.3754Koekemoer, A. M., Fruchter, A. S., Hook, R. N., & Hack, W.2002, in The 2002 HST Calibration Workshop : Hubble afterthe Installation of the ACS and the NICMOS Cooling System,ed. S. Arribas, A. Koekemoer, & B. Whitmore, 337–+Krumholz, M. & Burkert, A. 2010, ApJ, 724, 895Laird, E. S., Nandra, K., Georgakakis, A., Aird, J. A., Barmby,P., Conselice, C. J., Coil, A. L., Davis, M., et al. 2009, ApJS,180, 102Lotz, J. M., Jonsson, P., Cox, T. J., & Primack, J. R. 2010,MNRAS, 404, 5750 Kocevski et al.