ebook img

Discovery of a single faint AGN in a large sample of z>5 Lyman break galaxies PDF

0.19 MB·English
Save to my drive
Quick download
Download
Most books are stored in the elastic cloud where traffic is expensive. For this reason, we have a limit on daily download.

Preview Discovery of a single faint AGN in a large sample of z>5 Lyman break galaxies

Mon.Not.R.Astron.Soc.000,1–7() Printed5February2008 (MNLATEXstylefilev2.2) Discovery of a single faint AGN in a large sample of z > 5 Lyman break galaxies Laura S. Douglas1, Malcolm N. Bremer1, Elizabeth R. Stanway1, Matthew D. Lehnert2 1H H Wills PhysicsLaboratory, Tyndall Avenue, Bristol, BS8 1TL, UK 2Laboratoire d’Etudes des Galaxies, Etoiles, Physique et Instrumentation GEPI, Observatoire de Paris, Meudon, France Accepted 2007January25.Received2007January24;inoriginalform2006December18 7 0 0 ABSTRACT 2 As part of a large spectroscopic survey of z > 5 Lyman break galaxies (LBGs), we have identified a single source which is clearly hosting an AGN. Out of a sample of n more than fifty spectroscopically-confirmed R−band dropout galaxies at z ∼ 5 and a J above, only J104048.6-115550.2 at z = 5.44 shows evidence for a high ionisation- potential emission-line indicating the presence of a hard ionising continuum from an 5 AGN. Like most objects in our sample the rest-frame-UV spectrum shows the UV 2 continuumbreakingacrossaLyαline.UniquelywithinthissampleofLBGs,emission 1 from NV is also detected, a clear signature of AGN photo-ionisation. The object is v spatially resolved in HST imaging. This, and the comparatively high Lyα/NV flux 4 ratio indicates that the majority of the Lyα (and the UV continuum longward of 2 it) originates from stellar photo-ionisation, a product of the ongoing starburst in the 7 Lyman break galaxy. Even without the AGN emission, this object would have been 1 photometrically-selected and spectroscopically-confirmed as a Lyman break in our 0 survey. The measured optical flux (IAB = 26.1) is therefore an upper limit to that 7 from the AGN and is of order 100 times fainter than the majority of known quasars 0 at these redshifts. The detection of a single object in our survey volume is consistent / h with the best current models of high redshift AGN luminosity function, providing a p substantialfractionofsuchAGNarefoundwithinluminousstarburstinggalaxies.We - discuss the cosmologicalimplications of this discovery. o r t Key words: Galaxies; High Redshift; Active; Individual: J104048.6-115550.2 s a : v i X 1 INTRODUCTION accumulation and merger process can be studied over cos- r mic time, feeding back intocosmological simulations. a Recentyearshaveseenrapidobservationalprogressatz >5. Despite the success of recent observational work, large ‘Dropout’surveys,searchingfortheopticalsignatureofab- gaps still remain in our understanding of galaxies and sorption by the dense high redshift intergalactic medium quasars at this epoch and in particular how they relate to (IGM), have revealed a population of star-forming galaxies each another. Neither the population of bright galaxies nor within 1Gyrof theBig Bang (e.g. Lehnert & Bremer 2003; faint AGN have been well constrained at z > 5. Not only Stanway et al. 2003; Dickinson et al 2004; Bouwens et al. arethepopulationsexpectedtooverlapatfaintmagnitudes 2004).Atthesametime,thehugebutshallowSloanDigital and at low space densities, both populations can be tar- Sky Survey (SDSS, Yorket al 2000) has revealed a popu- getedbythesamelarge-areadeepopticalsurveys.Acurrent lation of highly luminous quasars, extending as far back as topic of much debate is the influence of AGN activity on z=6.4 (Fan et al 2006). starformation within agalaxyandviceversa.Manygalaxy Observations of galaxies and Active Galactic Nuclei formation models invoke AGN feedback to quench star for- (AGN)atthesehighredshiftsgiveusauniquewindowinto mation and prevent the formation of many small galaxies the early stages of structure formation in the universe, and which are over predicted by dark matter simulations. The on the effects that the first stars and galaxies had on their details of theform of thisAGN feedback are still unknown. surroundings.Analysisoffossilpopulationsinnearbygalax- WhatworkthathasbeendoneontheAGNfractioninsam- ies and simulations of hierarchical galaxy formation both ples of z > 5 Lyman break galaxies (LBGs) has indicated predict that the galaxies observed at z > 5 are the pro- thatthisfractionissmall.ForexampleBremer et al.(2004) genitors that will eventually merge to form today’s massive showed that none of a sample of over 50 photometrically- galaxies and clusters e.g. Thomas et al. (2005). The mass selected z > 5 LBGs showed any evidence of X-ray emis- (cid:13)c RAS 2 L. S. Douglas et al. sion down to deep limits ruling out strong hidden AGN Table 1. Photometric properties of J104048.6-115550.2 and 2σ components in them. Stanway et al. (2003) obtained simi- photometricdepths ofimagesmeasuredin2′′ apertures lar null results for z ∼ 6 LBGs. A second X-ray luminos- ity contraint is from the stacking anyalsis of Lehmer et al. V R I z J Ks (2005) who used a larger sample of galaxies within the Aper >27.8 >27.5 25.9±0.3 25.8±0.3 >24.9 >23.9 same field as used by Bremer et al. (2004); Stanway et al. Corr >27.9 >27.6 26.1±0.3 26.0±0.3 >25.1 >24.1 (2003). Lehmer et al. (2005) place X-ray luminosity limits of 2.8x1041 and 7.1x1041ergs−1 for z > 5 and z > 6 LBGs Mlim 27.8 27.5 26.7 26.3 24.9 23.9 respectively. Row aper is the photometry of the object within 2′′ diame- This is consistent with the results of observations at ter apertures with 1σ errors. Row corr is the photometry after lowerredshifts.Richards et al.(2006a)andFan et al(2001) galactic extinction (< 0.15 magnitudes) and lensing from fore- have inferred a low number density and contribution to groundobjects (0.25magnitudes) correctionsusingthemodelof reionisation byAGNfrom thebright endslope ofthe z>4 Cloweet.al. (2006) were applied. The third row corresponds to AGN luminosity function. This result has been supported the 2σ photometric depths of images measured in 2′′ apertures byMahabal et al.(2005)whoreportedthedetectionofjust withoutextinction oflensingcorrections. onefainthighredshiftquasarinasamplesurveying2.5deg2 andselectingforflat-spectrumopticalsources.Nonetheless, opticalimages. Thephotometryforthisobjectwasthenre- none of these observations have probed below the knee in measuredwith IDLroutinesusing2′′ aperturescenteredon the tentative z > 5 quasar luminosity function and hence the I-band position and the results were indistinguishable the current observations of quasars with M1450 < −27 do from thosemeasuredbySextractor.Photometriclimits and not preclude the existence of faint AGN which could rea- uncertaintiesweremeasuredfrom thestandarddeviationof sonably contribute a significant fraction of the ionising and the background in apertures of the same size. The flux was ultraviolet luminosity density at high redshift. measuredinrandomlyplacedaperturesandbackgroundlev- In this paper we present the photometric and spectro- elsweresubtractedusingthecorrespondingsmoothedback- scopic properties of an AGN detected at extremely faint ground used by Sextractor in its calculations. This method magnitudes in a large multiwavelength survey. In section 2 exactly matches the photometric measurements, thus pro- we briefly describe the survey and present the source prop- viding depths which are closely related to the photometry. erties. In section 3 we consider the implications of this de- The photometric properties of J104048.6-115550.2 and 2σ tection for our understanding of high redshift surveys and depthsfor each band are quoted in Table 1. thequasarluminosityfunction.Finallyinsection4wesum- This object and other LBG candidates were spectro- marise ourconclusions. scopicallyobservedonFORS2VLT/UT1onthe1st and4th We adopt the standard Lambda cosmology, i.e. a May 2005. The candidates were observed in ‘MXU’ mode flat universe with ΩΛ = 0.7, ΩM = 0.3 and H0 = withthe300IgratingandtheOG590blockingfiltercentered 70h70kms−1Mpc−1.Allmagnitudesarequotedinthephys- on a wavelength of 7600˚A. With a slit width of 1′′, this set ically meaningful AB system (Oke & Gunn 1983) in which upproducedadispersion of3.2˚A perpixelacross thewave- a source flat in fν has zero colour. length scale and a spatial scale of 0.2′′ per pixel. Twenty spectroscopic frames were taken of the target, each for 650 seconds. The object was placed at a different pixel position alongtheslit foreachexposuretofacilitate skysubtraction 2 OBSERVATIONS AND OBJECT SELECTION and to avoid bad pixels. The data were then reduced fol- The source J104048.6-115550.2, was photometrically se- lowing the prescription in Lehnert & Bremer (2003). Each lected and spectroscopically followed-up as part of a large frame was bias subtracted and flat fielded. The sky was sample of z > 4.8 Lyman break galaxies. The sample was then subtracted by producing a sky frame from the three selected from multiband imaging with VLT/FORS2 (two observations closest in time. To improve the skyline resid- hoursineachofV,R,I andz bands)andNTT/SOFI(five uals a first-order polynomial fit to each column was also hoursinJandsixhoursinKs)oftenwidely-separatedfields subtracted. The spectra were then shifted and combined, covering a total sky area of about 430 sq arcminutes. All wavelength calibrated using a lamp observations and flux imagesexceptthez-bandweretakenaspartoftheEDisCS calibrated using the spectrophotometric standard Feige 56. survey (Whiteet al. 2005). The details of sample selection Of the dozens of spectra of LBGs obtained as part of will be presented elsewhere (Douglas et al. 2007, in prepa- the survey,some tens of spectra show a clear Lyα line cou- ration), we brieflynote here how thesource was selected. pledwith acontinuumbreakandnootherclearly identified The source was photometrically selected to be a z ∼5 spectralfeatures(thesespectrawillbepresentedinDouglas LBG candidate, with a non-detection in V, a colour cut of et al. 2007). Only J104048.6-115550.2 additionally shows a R−I >1.3andnon-detectioninJ andKs (J >24.9,Ks > clear, but faint NV line. Lines as faint as this have been 23.9). The latter near-IR requirement rejects interloping detected before in distant LBGs, but have been identified z ∼1 galaxies (which can have similar optical colours) and asLyα(Lehnert & Bremer 2003,Stanway et al.2007). The many cool stellar (or substellar) objects that can contami- line does not fall on any obvious strong sky features. nateLBGsamples.Infactthiscolourselectionwillnotonly The source was spectroscopically observed in a com- identify LBG candidates but also any unobscured AGN at parable way to the infrared spectroscopic technique of ob- similar redshifts given that they have similar colours. As serving in a ABBA repeated offset pattern resulting in a with all other candidates, photometry was initially carried final spectrum with a positive and negative signal. Instead outusingSextractor(Bertin & Arnouts1996)oneachofour of one negative signal our adapted technique results in nu- (cid:13)c RAS,MNRAS000,1–7 Discovery of a single faint AGN in a large sample of z > 5 Lyman break galaxies 3 component measurement of the Lyαline caused by the low Table 2.Spectroscopic linepropertiesofJ104048.6-115550.2 resolution of the spectrum we are unable to determine the Line λobs /˚A λrest /˚A Flux/ergscm2 s−1 W0 /˚A effectsofopeninganglewhichcouldpotentiallybrightenthe AGN. Lyα 7829 1216 3.23±0.65×10−17 127±45 Inadditiontothemultibandground-basedimagingthis NV 7989 1240 2.43±0.73×10−18 10±4 object has also been observed by the Hubble Space Tele- scope.ThefieldcontainingJ104048.6-115550.2wasobserved bytheACSintheF775W filterforfourorbitsinthecenter merous negative signals in the sky background at different ofthefieldandforoneorbitaroundtheedge.Thehalf-light spatial positions. This is because of the way thesky frames radiusmeasuredforJ104048.6-115550.2intheoneorbitarea are assembled from several data frames nodded along the was 0.095′′. To test whether this is consistent with the ob- slit and then subtracted from another of the data frames. ject being spatially resolved, the PSF, as measured from a NormallytheLyαandcontinuumemissionfromourobjects 23 magnitude star in the four orbit region, was faintened are sufficiently faint that the negative artifacts in the sky to IAB=26 and placed randomly on the HST image in ar- background do not affect the final object spectrum. How- eas onlycovered byoneorbit. It was foundthat only 5% of ever, the Lyα line in this object is comparatively strong recovered objects had a half-light radius greater than 0.09′′ andconsequentlythenegativefeaturesproducedbythesky- with a mean radius of 0.07′′, implying that this source is subtractionprocedureincreasestheuncertaintyindetermin- resolved. If a point source profile of the same brightness is ingtheLyαlinefluxandlineshape.Tominimisethiseffect added to the object image, the resulting half-light radius thedata were rereduced following a different perscription. drops to within the unresolved range showing that it is un- The bias subtracted and flat fielded images were each likely that this object has an AGN contributing more than skysubtractedbyfittingasecond orderpolynomial toeach 50% of thetotal flux. column (unique wavelength bin) of the data using an IDL AnalternativeexplanationofthehighLyα/NVratiois procedure. Cosmic rays were excluded using sigma clipping a lower metallicity within the nucleus of the galaxy. The (2σ). Each sky subtracted frame was shifted so the object expected Lyα/NV ratio is taken from bright QSOs with spectrumfellonthesamespatialpixel.Thefinalimagewas solar or supersolar metallicities in the nucleus. A metal- made from the average of these frames rejecting the two licity lower than solar concentrations would explain the highestandlowestvaluesoutof20samplesateachposition weak NV emission. However from studying figure six in thusrejectingcosmic raysandbadpixels.Figures1present Hamann & Ferland(1999),themetallicityofthecorewould thisreductionofthedatawhichiscomparabletotheinitial havetobeahundredthofsolartoaccountforthelineratio. reduction, apart from the improvement in the characterisa- Hamann & Ferland (1999) also show that from elemental tion of Lyα. The Lyα line has the truncated, asymmetric abundances in AGN at low and high redshift out to z > 4, shapeexpectedofahighredshift Lyαlinesufferingabsorp- thegasinAGNappearstohaveasolartosupersolarmetal- tiononthebluesidefromtheLyαforest.TheNVisclearly icity. Verma et al. (2006) found consistent fits to the SEDs visiblebut at alower fluxthantheLyα(fainter byafactor of ∼ 13). of z ∼ 5 LBGs assuming 0.2 Z⊙. In addition, Verma et al. (2006) suggest that the typical z ∼5 galaxy is young, only afew 10Myrs old,and drivevigorous starburst-drivenout- flows. Such constraints would also suggest relatively low 3 ANALYSIS AND DISCUSSION chemical abundance in the typical LBGs at z ∼ 5. These resultswouldmakethegalaxysignificantly moremetalrich 3.1 AGN vs Starburst Contribution thanthenucleus,asituationhighlyunlikelygiventhesmall The broad band colours of this source are consistent with mass in thenuclear region compared tothe galaxy. Assum- those of a starbursting host galaxy (flat in fν and mod- ingthatthehighLyα/NVratioisduetothecontributionof ified by IGM absorption), undetected in V and R bands star-formation suggests that the ‘narrow line regions’ (i.e., and flat in colour between the I and z bands. However, withinabout10pcoftheAGN)hasalreadyreachedsolaror given the indications of AGN activity, what fraction of the supersolar metallicities at z=5.4. light observed is from the faint AGN? The fact that NV The absolute magnitude of this object at 1400˚A is - has been observed means that at least several tens of per- 22.5.Howeverasthehostgalaxycontributesalargefraction cent of the light must be attributed to the AGN. This of the observed light the magnitude of the AGN is likely also agrees with the Lyα/NV ratio observed. Through the- to be between -21 and -22 making this the faintest AGN oretical calculations and observations it has been found seen at z > 5. Although the Sloan Digital Sky Survey and that in a typical quasar the Lyα/NV ratio is between four others(Stern et al.2000;Romani et al.2004;Mahabal et al. and seven (Osterbrock & Ferland 1989; VandenBerk et al. 2005) have found QSO at similar and higher redshifts they 2001; Dietrich et al. 2002; Kuraszkiewicz et al. 2004). This are all at least two magnitudes brighter. At this faint level object has a Lyα/NV ratio of 13.3, a factor of 2-3 higher in the rest frame UV, the host galaxy luminosity function than that expected. This can be explained if there is large is overlapping with the faint end of the QSO luminosity contribution to theLyαfluxfrom the starformation in the function. host galaxy. Assuming thetypicalquasar ratio, theAGN is Indetailedphotometricandspectroscopicworkatz∼3 contributing 30-50% of the total flux of the source making Steidelet al. (2003) found a AGN fraction of 3% in their it 0.5−2 magnitudes fainter than that quoted. Of course comparableLBGsample.Thiswasconfirmedbythestudyof thisisonlyanestimateduetouncertaintiesintheLyα/NV theX-ray properties of the same z∼3 sample (Laird et al. ratio measurements. Due to the uncertainties in the broad 2006). From the one AGN detected in this z > 5 sample (cid:13)c RAS,MNRAS000,1–7 4 L. S. Douglas et al. Rest Wavelength / A 1200 1220 1240 1260 c 3 e cs 2 r n / a 01 o ati -1 Ly-α N V ar -2 p e -3 S 7700 7800 7900 8000 8100 8200 Observed Wavelength / A Figure 1.FORS2/VLTtwodimensionalspectrumofJ104048.6-115550.2 clearlyshowingthetypicallyasymmetricprofileofaLyαline withasharpbluecutoffandaextended redwing,andthedetection ofNV. the AGN fraction of ∼ 2% agrees with that found at z ∼ Lyαemissioncomesfromanunresolvedstellar-photoionised 3. However, studies comparing the rest-frame optical and component,theAGNcomponentisbroader.Subtracting50 UV properties of LBGs observed at z ∼ 3 and z ∼ 5 have per cent of the line using an unresolved component broad- found significant differences between the two populations. ens the remaining flux to around 1000 km/s, although this By using the same selection and analysis used in the z ∼3 figureisindicativeratherthanpreciseasthestarformation sample, Vermaet al. (2006) have found that z ∼5 galaxies contribution is uncertain. are typically much younger and have lower stellar masses Although objects with strong, wider Lyα emission are thantheirz∼3counterparts,demonstratingaclearchange more likely to have AGN components contributing to the in properties of the LBG between z ∼ 3 and z ∼ 5. With flux,theothereightsourceswithsimilarLyαfluxandcom- evidence of evolution in stellar age and mass, changes in parableequivalentwidthsshownoevidenceofanAGNcom- other properties such as AGN fraction would be expected ponent,demonstrating that such large Lyαlines can beex- thoughsurprisingly not seen in thislarge spectroscopically- plained by star formation alone. Apart from the NV line, confirmed sample. J104048.6-115550.2 does not exhibit otherproperties which A comparable spectroscopic study by Vanzella et. al. are extremefor a z>5 LBG. (2006) in the GOODS CDFS field have not reported on This broad Lyα component and the presence of NV, theAGNfractiontodatealthoughtheyhavenotcompleted only seen through AGNactivity,confirmstheidentification theirstudysocannotyetaddtothestatisticspresentedhere. of a central AGN. As such this is the faintest z > 5 AGN They also spectroscopically followed-up photometrically- seen to date by at least two magnitudes. Out of the tensof selectedz>5LBGcandidates.Weencourageotherauthors spectroscopically confirmedLymanbreakandLyαemitting tocloselyinvestigatespectroscopicfollow-upofsuchobjects galaxiesinourtensurveyfieldsonlythisobject,J104048.6- andreportpositiveandnegativedetectionsofAGNactivity 115550.2, shows any NV emission suggesting that although to providemore comprehensive statistics in thisfield. theseobjectsarepresentatthisredshiftepochtheyarevery rare. It is likely that some of the other Lyα emitters may havevery weak AGN which are being swamped by thevig- 3.2 Velocity Profile of Line Emission orousongoingstarformationinthehostgalaxies.Wewould not havedetected NVin such objects. The presence of a clear NV line strongly suggests that this galaxy, selected through its star formation properties, also contains a significant AGN component. As Figure 2 illus- 3.3 Implications for the AGN Luminosity trates, the Lyα line in the AGN candidate is one of the Function strongest inthespectroscopicsampleandisthebroadest of all lines of comparable flux. The dashed line in this figure TheQSOluminosity function is fundamentaltocosmology. showstheaverageprofileofthe8Lyαemitterswithsimilar It is intimately linked to many aspects of cosmology that line fluxes (and luminosity) to the main object. All 8 Lyα have yet to be fully characterised, such as galaxy and su- line are unresolved by the spectrograph setup, some with permassive black hole formation, the accretion history of indications of an asymmetric red wing. The average mea- black holes and the role of galaxy mergers and interactions sured FWHMof the sample (omitting theAGN candidate) in galaxy formation and evolution. The Sloan Digital Sky isabout400km/s,theresolutionofthespectroscopicsetup. Survey has provided good constraints on the bright end of The emission line in J104048.6-115550.2 matches those theluminosity function butcan not probethelow luminos- fromstarforminggalaxiesintheblueslope(consistentwith ity end occupied by Seyfert type AGN at high redshift. To a sharp drop due to absorption), but has a clear red wing. probethesefaintlevelsaphotometricandspectroscopicsur- It is resolved, with a measured FWHM of about 500 km/s veyatlevelsmorenormallyusedtosearchforLBGs,suchas which, when deconvolved with the instrumental response, thework described here, is needed. The faint-end slope has relates to ∼ 300 km/s. However, if we assume much of the been the topic of some debate as there is growing evidence (cid:13)c RAS,MNRAS000,1–7 Discovery of a single faint AGN in a large sample of z > 5 Lyman break galaxies 5 1.0 -4 0.8 M=-22.5 - 0.3921 objects expected 0.6 x -6 u ve Fl 0.4 Mpc-3 Relati 0.2 M>M) UV N( -8 0.0 -0.2 -40 -20 0 20 40 Relative Wavelength (Angstroms) -10 -20 -22 -24 -26 -28 -30 Figure2.Lyαlineprofileforaveragelineemitters(dashedline) MUV and J104048.6-115550.2 (solid line). The X-axis is measures in observed wavelength relative to the peak of the line. Neither of Figure 3. Luminosity function of QSO using Richardsetal. thecurvesisdeconvolvedfromtheinstrumentalresponsebecause (2006b) and Hopkinsetal. (2006) for the faint-end slope. The theaverageoftheeightappeartobeunresolved. horizontal dotted line shows the number density corresponding tooneAGNdetected inoursurveyarea. that it is flattening with increasing redshift, seen in the X- accounts for less that 50% of theUV luminosity of this ob- ray by Uedaet al. (2003) and La Franca et al. (2005). The ject the AGN is intrinsically fainter than this by up to a high-zfaint-endslopeisimportant indeterminingtheearly magnitude. The modified luminosity function predicts one formation historyofblackholes andtheirimpact ongalaxy object inoursurveyvolumeatthisfainterluminosity.Con- evolution. sequentlyouridentificationofasingleAGNagreeswiththe Fromasurveycovering430arcmin2withover85%spec- theoretical predictions provided that all AGN at this red- troscopic completeness of the photometrically selected can- shift are in starbursting LBGs. didates plus additional spectra of less robust candidates, If thereare additional AGNat this lower fluxlevel not J104048.6-115550.2 was the only object to show evidence embeddedinaUVluminousLBGsoappearingasa‘naked’ for AGN activity. As the photometric selection criteria of nucleus,thesewould not havebeen selected within oursur- AGN and LBGS is the same at these redshifts the uncer- vey as they would be fainter than our I-band flux limit. If tainity of our one AGN detection out of 50 confirmed LBG thiswerethecasewewouldexpecttoseesimilar‘naked’nu- candidates is around a factor of two. At this flux level the cleiat brightermagnitudeswhichwould havebeenspectro- NV line could potentially have been detected in any of the scopically identified as AGN in our survey.Given the slope spectra. Therefore at M1400 ≈ −21 the surface density of oftheAGNluminosity function, ournondetection ofbright AGN observed in this surveyis 2.3 x 10−3 arcmin2. ‘naked’ AGN requires that a significant fraction of all faint TheQSOluminosityfunctionofRichards et al.(2006b) AGN are within luminous LBGs. modifiedbythetheoreticalfaint-endslopeofHopkinset al. TheprospectofidentifyingalargesampleoffaintAGN (2006) is shown in Figure 3, calculated for z = 5. inordertoconstrainthefaint-endoftheluminosityfunction Hopkinset al. (2006) show that the evolution of the faint- is daunting. Not only would such surveys require an area end slope seen at lower redshifts can be explained by of many square degrees at a comparable depth to ours in luminosity-dependent quasar lifetimes and the physics of opticalandIRbands,butdisentanglingAGNandstarburst quasar feedback, where the faint-end slope is explained by emission in such sources is limited by the sensitivity of the quasarswithapeakluminosityneartheluminosityfunction deepestspectroscopicobservationscurrentlypossible.Todo breakbutseen inaless luminousstageofevolution.Thisis so cleanly requires characterisation of multiple metal lines an alternative to assuming there are changes in the distri- inspectra.Thisislikelytorequirethecapabilities offuture bution of the peak luminosity of quasars and of the masses largetelescopeseitherspacebased(JWST)orontheground of the black holes. The detailed modelling of the faint-end (ELTs). slope by Hopkinset al. (2006) is motivated by its ability to give insight into the distributions of central black hole 3.4 Cosmological Implications masses,hostpropertiesandquasarfuellingmechanismsand accretion. Therefore the faint-end of the AGN luminosity ThedetectionofasingleAGNinasampleof50ormoreUV functionisasensitiveprobeoftheformationofstructureat starbursts indicates that faint AGN contribute an insignifi- high redshift. cantamountofthemeta-galacticionisingfluxatz >5.Itis At the luminosities probed by this survey, the faint- already knownthatbrighterSDSSquasars(Fan et al2006) end slope is beginning to significantly modify the expected contributelessionisingfluxthantheemsambleofstarburst- numberofobjectspredicted.Inoursurveywewouldexpect ing galaxies at z > 5. Until now there was the possibility to find 0.4 objects with AGN activity at M=-22.5, the to- that a large population of faint AGN could rival the star- tal magnitude of this object. Given that the AGN emission bursts for UV output. This is clearly ruled out. Indeed the (cid:13)c RAS,MNRAS000,1–7 6 L. S. Douglas et al. fraction of ionising flux provided by AGN decreases as a REFERENCES function of luminosity. With one source out of 50 contain- Bertin, E. and Arnouts,S., 1996, A&AS,117, 393 ing an AGN which provides at most 50% of its flux, only 1% of ionising photons emitted by MUV ∼−22 is provided Bouwens R. J., Illingworth G. D., Thompson R. I., Blakeslee J. P., Dickinson M. E., Broadhurst T. J. et al., byAGN.Given that weareobserving theseobjects just af- 2004, ApJ, 606, L25 terreionisation theobviousconclusionisthatAGNhaveno Bremer M. N., Lehnert M. D., Waddington I., Hardcastle global influenceon theionisation state of theIGM. The majority of LBGs at z ∼ 5 appear to be under- M. J., Boyce P.J., Phillipps S., 2004, MNRAS,347, L7 Clowe, D., Schneider, P.,Arag´on-Salamanca, A., Bremer, going their first generation or generations of star formation M., de Lucia, G. and Halliday, C., A&A,451, 395 typically with stellar populations younger than 100 million Dickinson M., Stern D., Giavalisco M., Ferguson H. C., years (Vermaet al. 2006). The stellar density of these sys- Tsvetanov Z., Chornock R. et al., 2004, ApJ,600, L99 tems is only matched at low redshift in the centers of the Chaffee, F., Foltz, C. B., Junkkarinen, V. T., 2002, ApJ, massivegalaxiesindicatingthatthesesourceswilleventually 581, 912 evolve into the central regions of such galaxies today. This Fan X., Narayanan V. K., Lupton R. H., Strauss M. A., is supported by analysis of the spectra of near-by galax- KnappG. R., Becker R.H. et al., 2001, AJ, 122, 2833 ies which indicates that the most massive and dense sys- temsformedtheirstarsattheseredshifts(e.g.Thomas et al. Fan X., Strauss M. A., Becker R. H., White R. L., Gunn J. E., Knapp G. R. et al., 2006, AJ, 132, 117 (2005); Heavenset al. (2004)). It is perhaps not surprising Fan X., Strauss M. A., Richards G. T., Hennawi J. F., that we have identified a massive black hole in at least one BeckerR. H., WhiteR. L.et al., 2006, AJ, 131, 1203 galaxy given that these objects are routinely found at the center of massive galaxies. In at least one instance we are Hamann, F. and Ferland, G., 1999, ARA&A,37, 487 seeingsimultaneousAGNandgalaxyformation activity,in- Heavens A., Panter B., Jimenez R., Dunlop J., 2004, Na- dicating at least at this luminosity AGN activity does not ture,428, 625 quenchstar formation. Hopkins, P. F. and Hernquist, L. and Cox, T. J. and Robertson, B. and Di Matteo, T. and Springel, V., 2006, ApJ,639, 700 Kuraszkiewicz,J.K.,Green,P.J.,Crenshaw,D.M.,Dunn, J., Forster, K., Vestergaard, M., Aldcroft, T. L., 2004, 4 CONCLUSIONS ApJS,150, 165 In a spectroscopic sample of LBG at z ∼ 5 we have found La Franca, F. and Fiore, F. and Comastri, A. and Perola, a single object showing clear signatures of AGN emission. G.C.andSacchi,N.andBrusa,M.etal.,2005,ApJ,635, AnalysisofitsspectrumandHSTimagingindicatesthatno 864 morethan50%oftheUVfluxarisesfromtheAGN,therest Laird, E. S., Nandra, K., Hobbs, A., Steidel, C. C., 2006, comes from thestarburst in thehost LBG. MNRAS,373, 217 This single detection is consistent with the predictions Lehnert M. D.,Bremer M., 2003, ApJ, 593, 630 ofthebestavailableluminosityfunctionforfaintAGNonly Lehmer, B. D., Brandt, W. N., Alexander, D. M., Bauer, ifasignificantfractionofallfaintAGNarefoundinstrongly F. E., Conselice, C. J., Dickinson, M. E. et al., 2005, AJ, starbursting galaxies. 129, 1 Disentangling the AGN emission from the starburst MahabalA.,SternD.,Bogosavljevi´c M.,DjorgovskiS.G., emissionwillbeamajortechnicalchallengeinfutureobser- Thompson D., 2005, ApJ, 634, L9 vational studies of the faint-end of the luminosity function McLure R. J., Cirasuolo M., Dunlop J. S., Sekiguchi K., at these redshifts. Almaini O., Foucaud S. et al., 2006, MNRAS,372, 357 AtthesefluxlevelsthecontributionofAGNtothetotal Osterbrock, D. E., Ferland, G. J., 1989, Astrophysics of ionising photon budget at z ∼ 5 is of order of 1%. Given gaseousnebulaeandactivegalacticnuclei,UniversitySci- the short time elasped since the end of reionisation, this enceBooks, USA implies that the role of AGN in the global reionisation of RichardsG.T., HaimanZ.,PindorB.,StraussM.A.,Fan theuniverse is small. X.,Eisenstein D. et al., 2006a, AJ, 131, 49 Richards,G.T.,Strauss,M.A.,Fan,X.,Hall,P.B.,Jester, S.,Schneider, D.P. et al., 2006b, AJ, 131, 2766 RomaniR.W.,Sowards-EmmerdD.,GreenhillL.,Michel- son P., 2004 ApJ,610, L9 ACKNOWLEDGMENTS Stanway E.R., BunkerA.J., McMahon R.G., 2003, MN- We thank the referee for their helpful comments. LSD and RAS,342, 439 ERSacknowledgesupportfromPPARC.Based onobserva- SteidelC.C.,AdelbergerK.L.,ShapleyA.E.,PettiniM., tionsmadewithESOTelescopesattheLaSillaandParanal Dickinson M., Giavalisco M., 2003, ApJ, 592, 728 Observatory under programme IDs 166.A-0162 and 175.A- Stern D., Spinrad H., Eisenhardt P., Bunker A., Dawson 0706. Based on observations made with the NASA/ESA S.,Stanford S. A., Elston R., 2002, ApJ, 533, L75 Hubble Space Telescope, obtained [from the Data Archive] Thomas D., Maraston C., Bender R., Mendes de Oliveria at the Space Telescope Science Institute, which is operated C., 2005, ApJ,621, 673 by the Association of Universities for Research in Astron- Ueda, Y., Akiyama, M., Ohta, K., Miyaji, T., 2003, ApJ, omy,Inc.,underNASAcontractNAS5-26555.Theseobser- 598, 886 vations are associated with program 9476. Vanden Berk, D. E., Richards, G. T., Bauer, A.,Strauss, (cid:13)c RAS,MNRAS000,1–7 Discovery of a single faint AGN in a large sample of z > 5 Lyman break galaxies 7 M.A.,Schneider,D.P.,Heckman,T.M.etal.,2001,AJ, 122, 549 Vanzella, E., Cristiani, S., Dickinson, M., Kuntschner, H., Nonino, M., Rettura, A., et. al., 2006, A&A,454, 423 VermaA.,LehnertM.D.,F¨orsterSchreiberN.M.,Bremer M. N, Douglas L., MNRASin press White,S.D.M.,Clowe, D.I.,Simard,L.,Rudnick,G.,de Lucia, G., Arag´on-Salamanca, A. et al., 2005, A&A,444, 365 York D. G., Adelman J., Anderson J. E., Anderson S. F., AnnisJ., Bahcall N.A. et al., 2000, AJ, 120, 1579 (cid:13)c RAS,MNRAS000,1–7

See more

The list of books you might like

Most books are stored in the elastic cloud where traffic is expensive. For this reason, we have a limit on daily download.