Astronomy&Astrophysicsmanuscriptno.ms (cid:13)c ESO2011 December6,2011 BL Lacertae objects beyond redshift 1.3 - UV-to-NIR photometry and photometric redshift for Fermi/LAT blazars A.Rau,1,P.Schady,1,J.Greiner1,M.Salvato2,3,M.Ajello4,5,E.Bottacini4,N.Gehrels6,P.M.J.Afonso1,7,J.Elliott1, R.Filgas1,D.A.Kann8,S.Klose8,T.Kru¨hler1,3,9,M.Nardini1,10,A.NicuesaGuelbenzu8,F.OlivaresE.1,A.Rossi8, V.Sudilovsky1,A.C.Updike11,andD.H.Hartmann12 1 Max-Planck-Institutfu¨rExtraterrestrischePhysik,Giessenbachstraße1,85748Garching,Germany e-mail:[email protected] 1 2 Max-Planck-Institutfu¨rPlasmaPhysikandExcellenceCluster,Boltzmannstrasse2,85748Garching,Germany 1 3 UniverseCluster,TechnischeUniversita¨tMu¨nchen,Boltzmannstraße2,85748Garching,Germany 0 4 W.W.HansenExperimentalPhysicsLaboratory&KavliInstituteforParticleAstrophysicsandCosmology,StanfordUniversity, 2 USA c 5 SLACNationalAcceleratorLaboratory,StanfordUniversity,Stanford,CA94305,USA e 6 NASA-GoddardSpaceFlightCenter,Greenbelt,Maryland20771,USA D 7 AmericanRiverCollege,Physics&AstronomyDpt.,4700CollegeOakDrive,Sacramento,CA95841 8 Thu¨ringerLandessternwarteTautenburg,Sternwarte5,07778Tautenburg,Germany 5 9 DarkCosmologyCentre,NielsBohrInstitute,UniversityofCopenhagen,JulianeMariesVej30,2100Copenhagen,Denmark 10 Universita´deglistudidiMilano-Bicocca,PiazzadellaScienza3,20126,Milano,Italy ] 11 DepartmentofPhysicsandAstronomyDickinsonCollegeCarlisle,PA17013,USA O 12 DepartmentofPhysicsandAstronomy,ClemsonUniversity,Clemson,SC29634,USA C ReceivedSeptember27,2011;acceptedNovember19,2011 . h p ABSTRACT - o Context.Observationsoftheγ-rayskywithFermiledtosignificantadvancestowardsunderstandingblazars,themostextremeclass r ofActiveGalacticNuclei.AlargefractionofthepopulationdetectedbyFermiisformedbyBLLacertae(BLLac)objects,whose t s samplehasalwayssufferedfromasevereredshiftincompletenessduetothequasi-featurelessopticalspectra. a Aims.Ourgoalistoprovideasignificantincreaseofthenumberofconfirmedhigh-redshiftBLLacobjectscontainedinthe2LAC [ Fermi/LATcataloge. Methods.For103Fermi/LATblazars,photometricredshiftsusingspectralenergydistributionfittinghavebeenobtained.Thepho- 2 tometryincludes13broad-bandfiltersfromthefarultraviolettothenear-IRobservedwithSwift/UVOTandthemulti-channelimager v GRONDattheMPG/ESO2.2mtelescope.Datahavebeentakenquasi-simultaneouslyandtheremainingsource-intrinsicvariability 5 hasbeencorrectedfor. 2 Results.WereleasetheUV-to-near-IR13-bandphotometryforall103sourcesandprovideredshiftconstraintsfor75sourceswithout 0 previouslyknownredshift.Outofthose,eighthavereliablephotometricredshiftsatz & 1.3,whilefortheother67sourceswepro- 0 videupperlimits.SixoftheformereightareBLLacobjects,whichquadruplesthesampleofconfirmedhigh-redshiftBLLac.This . 2 includesthreesourceswithredshiftshigherthanthepreviousrecordforBLLac,includingCRATESJ0402-2615, withthebest-fit 1 solutionatz≈1.9. 1 Keywords.Techniques:photometric,(Galaxies:)BLLacertaeobjects:general,Galaxies:distancesandredshifts 1 : v i X 1. Introduction hole) (e.g., Abdoetal. 2010d) to probes of the extra-galactic background light (EBL) through attenuation of γ-ray photons r a Since its launch in 2008, the Fermi Space Laboratoryhas dra- (e.g.,Abdoetal.2010c).Oneofthecrucialparametersforthese matically extended our view of the high-energy sky. The re- applicationsisthedistancetotheobjects,whichunfortunatelyis cently released 24-month catalog (2LAC) of Active Galactic noteasytoobtaininmostcases. Nuclei (AGN) detected by the Large Area Telescope (LAT; Atwoodetal. 2009) revealed 885 high-significance sources, the large majority of them being blazars (Ackermannetal. Twoclassesofblazarsdominatethe2LACpopulation.These 2011). The latter form the most extreme class of AGN with are the flat-spectrum radio quasars (FSRQs, 310 sources) and their observational characteristics governed by the small an- BL Lac objects (395), named after the prototype BL Lacertae gle between their relativistic jets and the observer’s sight (Hoffmeister1929).Whilefortheformerredshiftmeasurements line (Blandford&Rees 1978). The resulting Doppler boosting areroutinelyperformedusingtheirstrongemissionlinesatUV- makes blazars exceptionally bright sources at nearly all wave- optical wavelengths, the featureless, power-law optical spectra lengthsandthereforevisibleouttohighredshift. ofBLLacobjectshaveproventobeachallenge(e.g.,Shawetal. The scientific relevance of blazars is very broad, ranging 2009).Indeed,220ofthe395BLLacsinthe2LAC(55%)lack fromlaboratoriesforthephysicsandstructureofrelativisticjets redshift estimates. Until this incompleteness is resolved, con- (andthustheextractionofenergyfromthecentralmassiveblack clusionsabouttheEBL,theblazarsequence(e.g.,Fossatietal. 2 A.Rau,etal.:BLLacobjectsbeyondredshift1.3-UV-to-NIRphotometryandphotometricredshiftforFermi/LATblazars 1998; Ghisellinietal. 1998), and the blazar population in gen- 120 eralremaintentativeatbest. 100% SeveralmethodshavebeenexploitedtoincreasetheBLLac s 100 redshiftsample.Atlowdistance,onecanutilizetheremarkably e c uniformabsolute brightnessof the giant elliptical BL Lac host ur 90% o 80 galaxies (Sbarufattietal. 2005; Meisner&Romani 2010) and s very-high signal-to-noise optical spectroscopy to help to iden- of # 60 tify weak emission or absorption features in a few other cases e 50% (e.g., Shawetal. 2009). An alternative method, applicable to ativ 40 the moredistantsources, is the photometricredshifttechnique, ul whichconsistsoffittingspectralenergydistribution(SED)tem- m 25% u platestomulti-bandphotometry.Neutralhydrogenalongtheline c 20 10% ofsighttotheblazarwillimprintaclearattenuationsignatureat 0 theLymanlimitandthusallowsanaccurateestimateofthered- 0.01 0.10 1.00 10.00 100.00 shift of the absorber. Even thoughthe absorberwill be located ∆t [days] somewherealongthe line of sightand itsredshiftwillthusnot GROND-UVOT necessarilycorrepondtothatoftheblazar,photometricredshifts, Fig.1. Cumulative distribution of the (absolute) time between z ,willprovideareliablelowerlimitontheblazarredshift. phot the mid-points of the GROND and Swift/UVOT observations. In this paper we explore the use of ultra-violet to near- The median offset is 1.5d. For longer UVOT observations, infraredquasi-simultaneousphotometrytomeasuretheredshift whichwerespreadoverseveralsatellite orbits,themid-timeof of 103 2LAC blazars, 86 of them without previous redshift theuvw2-exposurewastakenasreference. constraints. The sample selection is described in Sect. 2 while the observations and data reduction are detailed in Sect. 3 and Sect.4.TheresultsandtheirdiscussionarepresentedinSect.5 TheUVOTpointingshavebeenobtainedaspartoftheSwift andSect.6,respectively. fill-in program from January 2010 to October 2011. With few exceptions,they comprise observationsin three optical(u,b,v) andthreeUV(uvw2,uvm2,uvw1)lenticularfilters,coveringthe 2. SampleSelection wavelength range between 160 nm and 600 nm (Pooleetal. The sample is selected mainly from the clean sample 2008). The exposure times vary from source to source de- of the 24-month catalog of AGN detected by Fermi/LAT pending on visibility and brightness of the optical counterpart. (Ackermannetal. 2011) which includes 885 high-significance Approximatevaluesare100seachinu,b,vand200,250,400s γ-raysourceswhicharestatistically associatedwithAGNsand in uvw1,uvm2,uvw2, respectively. For a few sources multiple locatedathighGalacticlatitudes(|b| > 10deg).Toaccomodate consecutiveorbits,andthuslongerexposures,wereobtained. thelocationofoneofourfollow-upinstrumentsinthesouthern WhileUVOTneedsto cyclethroughallfiltersinsequence, hemisphere,weappliedacutindeclination(δ < +25deg). GROND observesin it’sfouropticalg′,r′,i′,z′ andthreenear- J2000 Furthermore, the importance of the ultra-violet bands for the IRJ,H,Ks bandssimultaneously.Thiscapabilityisparticularly z estimatesleadustodiscardallsourceswithGalacticfore- importantforstudiesoffast-varyingsourcessuchasblazarsasit phot ground reddening of E > 0.2mag, as derived from the allowstheconstructionofareliablespectralenergydistribution B−V Schlegeletal.(1998)maps.Thefinalsampleiscomposedof80 (SED) withoutthe need to correctfor source-intrinsicvariabil- sourceswithoutknownredshiftbutwithopticalorradiocounter- ity. Unfortunately,schedulinglimitationsrarely allowed simul- partassociationsfromBayesianstatistics(Abdoetal.2010a,b), taneousobservationswith bothinstruments,UVOTfromspace Likelihood Ratio, or log N – log S methods (Ackermannetal. and GROND from the ground. However, in order to minimize 2011)andwasfurtherextendedbyeightsourcesthatdonotbe- theimpactofvariabilityonthecombinedSED,GRONDobser- longtotheclean2LACsample1.Inaddition,observationsfor16 vations were executed as close as possible in time to the Swift 2LACsourceswithexistingredshiftmeasurementsareincluded pointings.Exceptinafewcaseswherevisibilityorweathercon- for verification. We emphasize here that our sample is not sta- straintsleadtooffsetsofmorethan10d,observationsweretyp- tisticallycomplete,butbiasedbyselection.Thefinallistof104 icallyachievedwithin1-2dofeachother(seeFig.1).Thezphot targetsispresentedinTable1.Ofthose,82havebeenclassified uncertainties associated with these offsets will be discussed in asBLLacobjects,threeasaFSRQ,andtheremaining19areof Sect.4.3. unknowntype(Ackermannetal.2011). A typical GROND observation had an integration time of 2.4minin(g′,r′,i′,z′)and4.0minin(J,H,K ). s 3. Observations 4. Data&Analysis The observations were performed with the Ultraviolet and 4.1.Swift/UVOTandGROND Optical Telescope (UVOT; Romingetal. 2005) onboard the Swift satellite (Gehrelsetal. 2004) and the Gamma-ray UVOT photometry was carried out on pipeline processed sky Optical/Near-Infrared Detector (GROND; Greineretal. 2008) images downloadedfrom the Swift data center 2, followingthe mounted on the MPI/ESO 2.2m telescope at La Silla, Chile. standard UVOT procedure (Pooleetal. 2008). Source photo- Each source has been observed at least once with both instru- metric measurements were extracted from the UVOT imaging ments, although for some targets several epochs had been ac- datausingthetooluvotmaghist(v1.1)withacircularsourceex- quired. tractionregionthatrangedfrom3′.′5-5′′ radiusto maximisethe 1 flaggedduetoanalysisissuesbytheLATteam 2 http://www.swift.ac.uk/swift portal A.Rau,etal.:BLLacobjectsbeyondredshift1.3-UV-to-NIRphotometryandphotometricredshiftforFermi/LATblazars 3 Table1. TargetList(firstfiverows,restinJournalonly) Name FGLNamea α b δ b offset[′′]c E [mag]d J2000 J2000 B−V CRATESJ0001-0746 2FGLJ0000.9-0748 00:01:18.00 -07:46:26.9 0.2 0.03 CRATESJ0009+0628 2FGLJ0009.0+0632 00:09:03.93 06:28:21.3 1.1 0.07 CRATESJ0021-2550 2FGLJ0021.6-2551 00:21:32.56 -25:50:49.1 0.8 0.02 1RXSJ002209.2-185333 2FGLJ0022.2-1853 00:22:09.26 -18:53:34.9 3.1 0.03 RXJ0035.2+1515 2FGLJ0035.2+1515 00:35:14.71 15:15:04.2 3.4 0.07 .. .. .. .. .. .. a:Fermi/LATID. b:GRONDcoordinatesoftheopticalcounterpartwithtypicaluncertaintiesof0′.′3inbothdirections. c:OffsetofcounterpartlocationfromthosegiveninAckermannetal.(2011). d:FromSchlegeletal.(1998). signal-to-noise. In order to remain compatible with the effec- 30 tiveareacalibrations,whicharebasedon5”aperturephotome- try(Pooleetal.2008),anaperturecorrectionwasappliedwhere 25 necessary. This correction was at maximum 5–6% of the flux, dependingonthefilter. Forthe furtheranalysis, allmagnitudes s 20 wereconvertedintotheABsystemandarepresentedinTab.2. e c Typicallyachieved3-σlimitingmagnitudesareuvw2AB ≈ 22.4 ur andTbhAeB ≈GR19O.6N.D data (Tab. 3) were reduced and analyzed of so 15 with the standardtoolsand methodsdescribed in Kru¨hleretal. # 10 (2008). The g′,r′,i′,z′ photometry was obtained using point- spread-function(PSF)fittingandcalibratedagainstobservations 5 of fields covered by the SDSS Data Release 8 (Aiharaetal. 0 2011). Due to the undersampled PSF in the near-infrared, 10 12 14 16 18 20 22 24 the J,H,K photometry was measured from apertures with s r’-band Brightness [mag] the sizes corresponding to the Full-Width at Half Maximum (FWHM) of field stars and calibrated against selected 2MASS Fig.2.ObservedGRONDr′-bandmagnitudedistributionofthe stars (Skrutskieetal. 2006). This resulted in 1σ accuracies of opticalcounterparts. 0.04mag(g′,z′),0.03mag(r′,i′),0.05mag(J,H),and0.07mag (K )fortheabsolutecalibration.WhiletheSDSScalibrationdi- s rectlyprovidesmagnitudesin the ABsystem, the near-IRpho- tometryrequiredfurthertransformationfromthe2MASS-native offsetsforthe103sourceswithidentificationaregiveninTab.1. Vega system into AB. Typically achieved 3-σ limiting magni- Most of the identified optical sources are bright and point-like tudesarer′ ≈23.5andK ≈19.8. with typical magnitudes in the GROND r′-band of 16–18mag AB s,AB CorrectionforGalacticforegroundextinctionwasperformed (Figure2). followingtheproceduredescribedinCardellietal.(1989)with The source morphology can be used as a prior for the se- E fromSchlegeletal.(1998).FortheUVOTbands,thecor- lection of the SED template library(see Sect. 4.4) and redshift B−V rection factors presented in Kataokaetal. (2008) were used. range, i.e., clearly extended sources will have a contribution UncertaintiesinE (10%;Schlegeletal.1998)andinthered- fromthehostgalaxyemissiontotheobservedSEDandarelikely B−V deninglawproduceanadditionalcontributiontothephotometric located at low redshift. A simple morphological distinction in errorbudget.Asthissystematicuncertaintyiscoupledbetween point-likeandextendedhasbeenperformedontheGRONDim- thethephotometricbandsforagivenpositioninthesky,itsim- ages and will be used for discussing the reliability of the zphot pact on the SED fitting is smaller than the contribution to the estimates. photometryineachindividualfilter.Anexactcalculationisvery complex. Thus, we adopt a conservative 5% of the reddening 4.3.Variabilitycorrection value in each band which is added in quadrature to the photo- metricuncertainties. The optical emission of blazars is known to vary on a wide range of time scales. Brightness variations range from a few tenths of a magnitude within minutes to days (e.g., Racine 4.2.Counterpartselectionandmorphology 1970; Milleretal. 1989; Carinietal. 1991; Urryetal. 1993) Astrometricsolutionsfor the GROND opticalbandshave been to several magnitudes over weeks to years (e.g., Ciprinietal. obtained throughcomparison with USNO or, if available, with 2003; Raiterietal. 2005). This variability contributes signifi- SDSSmeasurements,achievingatypicalrmsof0′.′3inbothco- cantlytotheuncertaintieswhenconstructinganSEDfromnon- ordinates.Forthelargemajorityofourtargets(91)onlyasingle simultaneousmulti-band,multi-instrumentobservations. opticalsourcewasfoundwithin2′′ radiusofthelocationgiven AsGRONDobservesinallsevenbandsatthesametime,the in the 2LAC catalog. For an additional twelve objects isolated g′toK photometrycanbeconsideredasasnapshotoftheSED s candidate counterparts were detected at offsets of up to 6′′. In and is thusunaffectedby variabilityon time scales longerthan one case (CGRaBS 1407-4302),several blended sources made the exposure of an individual observation. This leaves two ar- areliableidentificationimpossible.Counterpartcoordinatesand easwhereapropertreatmentofvariabilityhastobeperformed, 4 A.Rau,etal.:BLLacobjectsbeyondredshift1.3-UV-to-NIRphotometryandphotometricredshiftforFermi/LATblazars Table2. Swift/UVOTphotometry(firstfiverows,restinJournalonly) Name UTDatea ABMagnitudeb ∆m c GR→UV uvw2 uvm2 uvw1 u b v [mag] CRATESJ0001... 2010/10/1410:40 18.99±0.08 18.74±0.10 18.58±0.09 18.18±0.09 17.88±0.10 17.90±0.19 0.01 CRATESJ0009... 2010/12/0708:55 20.16±0.15 19.71±0.19 19.69±0.18 19.24±0.19 18.51±0.18 18.14±0.24 -0.24 CRATESJ0021... 2010/11/1818:32 18.86±0.08 18.90±0.27 18.57±0.09 18.15±0.10 17.82±0.11 17.93±0.21 0.09 1RXSJ002209.2... 2010/11/1813:49 18.36±0.07 17.95±0.08 17.91±0.08 17.48±0.07 17.20±0.08 16.97±0.12 -0.08 RXJ0035.2... 2011/01/1401:08 18.36±0.08 17.95±0.08 17.91±0.08 17.48±0.07 17.20±0.08 16.97±0.12 -0.24 .. .. .. .. .. .. .. .. .. a:Approximatestarttimeofuvw2exposure. b:CorrrectedforGalacticforegroundreddening.Upperlimitsare3-σ. c:Variability-correctionfactortobeappliedtoGRONDphotometry(seetext). Table3.GRONDphotometry(firstfiverows,restinJournalonly) Name UTDatea ABMagnitudeb g′ r′ i′ z′ J H K s CRATESJ0001... 2010/10/1403:33 17.80±0.05 17.43±0.05 17.18±0.05 16.86±0.05 16.48±0.05 16.16±0.06 15.76±0.08 CRATESJ0009... 2010/12/1200:41 18.91±0.05 18.49±0.05 18.15±0.05 17.89±0.06 17.34±0.06 16.90±0.06 16.46±0.08 CRATESJ0021... 2010/11/2300:13 17.69±0.05 17.43±0.05 17.20±0.05 16.98±0.05 16.76±0.05 16.51±0.05 16.15±0.07 1RXSJ002209.2... 2010/11/2300:21 17.30±0.05 16.92±0.05 16.71±0.05 16.51±0.05 16.22±0.05 16.00±0.06 15.57±0.08 RXJ0035.2... 2011/01/1400:48 17.39±0.05 17.10±0.05 16.96±0.05 16.77±0.05 16.41±0.06 16.17±0.05 15.92±0.08 .. .. .. .. .. .. .. .. .. a:Exposurestarttime. b:CorrrectedforGalacticforegroundreddening.Notcorrectedforvariability.Upperlimitsare3-σ. i) between the exposuresin the individualUVOTfilters and ii) betweentheGRONDandUVOTpointings. 0.6 UVOT observes in each band separately and, for our pro- gram, typically cycled through its filters in a specific order (uvw1,u,b,uvw2,v,uvm2).Acompletesequenceinallsixbands took≈ 12min,duringwhichtimethetargetmayvaryinbright- 2 0.4 w ness.Withoutaprioriknowledgeofthespectralshape,thisvari- uv g abilitycanonlybeaccountedforstatistically, i.e.,byincluding a m an additional contribution to the systematic uncertainties. For ∆ 0.2 this purpose we analyzed the photometry for those blazars for whichmultipleUVOTexposuresinasinglebandwereavailable. Here, we focused on the bluest filter (uvw2) which would be 0.0 leastaffectedbyapotentialhostgalaxycontributionandshould thus give a good representationof the maximum variability.In 0.1 1.0 10.0 100.0 Fig.3thedistributionofvariabilityasfunctionoftimebetween ∆t [days] twouvw2exposuresisshownfor25targetswith accuratepho- tometry(δm < 0.1mag)ineachimage.Whilevariabilityas uvw2 Fig.3. Absolute brightness changes in the uvw2 filter as func- largeas∆m ≈0.4magcanbeseenforsomesourcesalsoon uvw2 tion of time between two UVOT pointings for all 25 sources timescalesof∆t < 0.1d,themedianvariationfor∆t < 50dis withmultipleobservations.Theblackcrossesshowthemedian ≈0.1mag,comparabletothephotometricuncertainty. values(andmedian absolutedeviation)and the greydots mark The shortest time sampled between two uvw2 exposuresin all174pairsofobservations.Whilemicro-variability(∆m ≈ thissampleisonesatelliteorbit(≈90min)andthereforelonger uvw2 0.1mag) dominates on time scales shorter than 50d, signif- than it takes to cycle throughthe six filters. However,it is rea- icantly larger variability (∆m > 0.3mag) is observed at sonable to assume that the median variation on a 12min time uvw2 ∆t>50d. scaleisnotmuchlargerthantheoneover96min.Thus,wecon- servativelyadopted∆m = 0.1magasanadditionalcontribution tothesystematicuncertaintiesforallUVOTfilters. tionoffset.Hereweusedthecalibrationwiththelargestspectral In order to correct between GROND and UVOT we made overlap,namelytheonebetweenb,g′,andr′: use of the spectral overlap provided by the two instruments (Kru¨hleretal. 2011). Under the assumption that the SED re- b−g′ =0.17(g′−r′)+0.03(g′−r′)2 (1) mainsunchangedandcanbeapproximatedbyapowerlawand thus,thattheGRONDandb,vphotometryfollowthesamespec- whichis validfor−1 ≤ (g′−r′) ≤ 2 (Kru¨hleretal. 2011). tral slope 3, color-terms can be used to derive the normaliza- Thesecorrections,∆magGR→UV (Tab2),werecalculatedforall sourceswithaccuratephotometryinb,g′,andr′(δm<0.2mag). 3 This isreasonable for redshift z < 3.5and for sources where the Theywerecomputedforeachobjectindividuallyandappliedto hostgalaxycontributiontotheemissionisnegligible.SeeSect.4.4for theGRONDphotometry,shiftingitinlinewithUVOT.Typically furtherdiscussion. values are of the order of ±0.1mag (Fig. 4). In cases where A.Rau,etal.:BLLacobjectsbeyondredshift1.3-UV-to-NIRphotometryandphotometricredshiftforFermi/LATblazars 5 models intrinsic extinction was again neglected, a different lu- 1.0 minositypriorof−8 < Mg′ < −30waschosentoallowforthe expectedfainterhostmagnitudes.Thethirdlibraryconsistsofa OT 0.8 widerangeofstellartemplates(Pickles1998;Bohlinetal.1995; V Chabrieretal.2000),includedtotestagainstpotentialfalseas- U → sociations.Foreachsourceinoursample,allthreelibrarieswere D 0.6 ON fitindependently. R G The most prominent spectral feature used to measure z g 0.4 phot a forblazarsiscausedbytheabsorptionthroughneutralhydrogen m ∆ along the line of sight. At z ≈ 0.8 the Lyman-limitmovesinto 0.2 thewavelengthcoveredbytheUVOTandstartstosuppressthe fluxintheuvw2filter.Whilethisisthetheoreticallowerlimiton 0.0 z forthecombinedGRONDandUVOTcoverage,uncertain- phot tiesassociatedwiththephotometryandmodelingwillshiftitto 0.1 1.0 10.0 100.0 a higherredshift. Usually one would use a spectroscopic train- ∆t [days] ingsampletoevaluatetheredshiftrangeandaccuracywhichis accessible with our SEDs. As the majority of the sources with Fig.4.Absolutebrightnessoffsetasfunctionoftimebetweenthe known redshift in our sample are located at z < 1, they alone GRONDandSwift/UVOTobservations.Theblackcrossesshow do not suffice to provide a statistically meaningful test of the themedianvalues(andmedianabsolutedeviation)andthegrey method.Insteadweassessthereliabilityinourphotometricred- dots mark all 90 sources for which the offset was determined shiftcomputationwithaMonte-Carloapproach. (excludingelevensourceswithlargephotometricuncertainties, Here,wesimulated27,000testpower-lawSEDswithspec- missing b, g′, or r′ photometry,or strong hostgalaxycontribu- tral slopes ranging from 0.5 to 2, redshifts from 0 to 4, and at tion). threeapparentmagnitudes,r′ =17,18,19mag.Foreachsource, thephotometryintheindividualbandswasscatteredaroundthe model magnitudes with a brightness-dependentstatistical con- multiple epochs were available, we used the closest pair of tributionandwithacomponentrepresentingthesystematicun- GRONDandUVOTobservations.ThefinalSEDforeachsource certaintiesfromthephotometriccalibrationandvariabilitycor- corresponds to the best possible reconstruction between 160– rection.TheresultingSEDswerefedbackintoLePhareandthe 2200nm at the time of the UVOT observation. No correction outputz wascomparedtotheinputvalue. phot was performed for sources where the photometric uncertainty AscanbeseeninFig.5,thereisagoodcorrespondencebe- (δm ≥ 0.2mag in b,g′, and/or r′) was larger than the typical tweentheinputandtherecoveredredshiftforz >1.2.Inpar- sim ∆magGR→UV. ticularforbrightsources(r′ <18mag)thenumberofoutliersη, For a few sources, the emission from the underlying definedasthenumberofsourceswith|∆z(1+z )|>0.15,drops sim host galaxy contributes significantly to the overall SED (see steeplyabovethisredshift.Forfaintersources(r′ ≈19mag),the Sect.4.4).Whiletheblazaremissioncanstillvary,thehostcon- increasinglyless constrainedUV photometrycausesthe outlier tributionnaturallyremainsunchanged,violatingtheassumption fractiontoremainhigh(≈ 20%)outtoz ≈ 2.Amorequan- sim ofaconstantSEDslope.Thus,novariabilitycorrectionwasper- titativeselectioncanbemadewhenusingP ,theintegralofthe z formedinthosecaseswheretheUVtoopticalSEDclearlyde- probabilitydistributionfunctionR f(z)dzatzphot±0.1(1+zphot), viatesfromapowerlaw. which describes the probability that the redshift of a source is within 0.1(1+z) of the best fit value. When choosing a cut at P > 90%,themajorityoftheoutliersatz > 1.2,aswellas 4.4.SEDfitting z phot nearlyallthesolutionswithz < 1.2disappear(insetFig.5). phot The photometric redshift computation was performed with the Thisconfimsthatforpower-lawsourceswithz < 1.2theavail- publically available LePhare code v.2.2 (Arnoutsetal. 1999; ablephotometrycanonlyplaceanupperlimit,andsuggeststhat Ilbertetal.2006).Theprogramusesthesimpleχ2fittingmethod P > 90%canbeusedasa criteriontoidentifyreliablephoto- z to differentiate between a theoretical and an observed photo- metric redshift solutions. Selecting a higher P threshold, e.g., z metric catalogue. We used three custom template libraries for 99%, can further reduce the outlier fraction but will also shift comparisonwithourGRONDandUVOTphotometry.Thefirst thelowerconstrainableredshiftboundtohighervalues.Inorder was composed of 40 power law SEDs of the form F ∝ λ−β toreachsuchaverytightlypeakedredshiftprobabilitydistribu- λ with β ranging from 0 to 2 in steps of 0.05 and describes pri- tionfunctionmorethanonephotometricbandhastobeaffected marily the central-engine dominated blazars. Here, the source- bytheLyman-limit,limitingtheredshiftrangetoz&1.5. intrinsic extinction was assumed to be zero and the luminos- ity prior was choosen to cover the expected range of absolute 5. Results magnitudes (−20 < M < −30; Ve´ron-Cetty&Ve´ron 2010). g′ SuchasimplemodelisagoodapproximationfortheUV-near- The fit results for all 103 sources with identified counterparts IRSEDsofBLLacs,whichformthebulkofoursample.Also arepresentedinTab.4. Herewe givethephotometricredshifts FSRQs, if their emission lines are not dominant, can be mod- (and their 90% confidence errors), the P values, and the best z eled with a power law to first order. The second library con- fit models for the power law and galaxy templates. No source tains templates of normal (inactive) galaxies and galaxy/AGN requiresastellartemplate. hybrids(Salvatoetal.2009,2011)andhasbeenimplementedin Asdiscussedintheprevioussection,P >90%canbeused z order to model host galaxydominated objects. If the host con- asa goodreliabilitycriterionforthe photometricredshiftsolu- tributionissignificant,the4000Åbreakcanemergeandpoten- tion. Applying this cut to the power law model fits results in tially be misinterpretedasa Lyman-limit.While forthe galaxy a sample of 15 sources with χ2 < 30. Next, we also require 6 A.Rau,etal.:BLLacobjectsbeyondredshift1.3-UV-to-NIRphotometryandphotometricredshiftforFermi/LATblazars Table4.Excerptoftheresultstable.Onlyentriesforthosesourceswithreliableredshiftsofz > 1.2areshown.(restinJournal phot only) Name z a z b powerlaw galaxy phot,best spec,img z c χ2 P d βe z c χ2 P c model phot z phot z RXJ0035.2+1515 1.28+0.14 – 1.28+0.14 2.6 90.1 0.90 0.79+0.14 8.7 75.5 I22491 60 TQSO1 40 −0.17 −0.17 −0.13 PKS0047+023 1.44+0.16 – 1.44+0.16 10.8 94.9 1.60 0.41+0.06 13.6 80.5 pl QSO DR2 029 t0 −0.19 −0.19 −0.08 PKS0055-328 1.37+0.14 – 1.37+0.14 17.7 95.2 1.25 0.47+0.23 11.2 38.4 I22491 50 TQSO1 50 −0.17 −0.17 −0.47 PKS0332-403 1.47+0.11 1.426 1.47+0.11 6.0 99.8 1.35 1.11+0.10 13.0 62.3 I22491 60 TQSO1 40 −0.12 −0.12 −0.33 CRATESJ0402-2615 1.92+0.12 – 1.92+0.12 8.6 100.0 1.25 1.28+0.22 9.5 48.9 I22491 60 TQSO1 40 −0.09 −0.09 −0.23 SUMSSJ053748571828 1.55+0.09 – 1.55+0.09 20.5 99.9 0.80 1.58+0.08 21.5 99.6 pl I22491 30 TQSO1 70 −0.13 −0.13 −0.07 PKS0600-749 1.54+0.14 – 1.54+0.14 6.9 98.1 1.20 0.46+0.24 7.5 58.1 I22491 60 TQSO1 40 −0.19 −0.19 −0.05 CRATESJ0630-2406 1.60+0.10 – 1.60+0.10 9.3 100.0 0.85 1.19+0.50 24.6 83.7 I22491 60 TQSO1 40 −0.05 −0.05 −0.12 OM235 1.72+0.13 1.549 1.72+0.13 18.2 99.9 1.10 1.54+0.10 11.9 94.9 pl I22491 30 TQSO1 70 −0.13 −0.13 −0.09 CRATESJ1312-2156 1.77+0.09 1.491 1.77+0.09 7.1 100.0 0.95 1.60+0.14 18.4 100.0 I22491 60 TQSO1 40 −0.11 −0.11 −0.09 CLASSJ2352+1749 1.45+0.21 – 1.45+0.21 7.5 91.8 1.15 0.55+0.16 7.1 64.2 pl I22491 20 TQSO1 80 −0.18 −0.18 −0.14 a:Bestphotometricredshift,seetext. b:Spectroscopicorimagingredshift(ifknown)fromAckermannetal.(2011). c:Photometricredshiftwith90%confidenceuncertainties. d:Redshiftprobabilitydensityatz ±0.1(1+z ). phot phot e:Spectralslopeforpowerlawmodeloftheformλ−β. f: Starburst/QSO hybrid templates with varying contributions of the two components. Template names reflect the relative contribution of the starburst(I22491 NN)andAGN(TQSO1 NN)modelstothehybrid.“pl”marksanadditionalpower-lawcomponent atshortwavelengths.See Salvatoetal.(2011)fordetailsonthetemplates. thatasourcecannotbefitwellwithagalaxy/AGN-hybridtem- viously been classified as BL Lac objects and are therefore plate at low redshift (z < 1.2). This step is required in order unlikely to show strong emission lines in their optical spec- to identify those sources for which a degeneracy in the pho- tra. Given these considerations, we do not regard their galaxy- tometric redshift solution does not allow a distinction between templatephotometricredshiftstobereliable. a low-redshift galaxy/AGN-hybrid and a blazar at higher red- There are 79 sources which are well-fit by power-law tem- shift.Inanalogywithourselectionforagoodredshiftconstraint plates (χ2 < 30) but have a low P . In other words, the spec- z (Pz > 90%), we consider Pz,gal < 90% as a reliable criterion tral slope is well defined but the photometric redshift cannot that no adequatelow-redshiftgalaxy/AGN-hybridsolution was be unambiguously determined. As expected, the best solutions found.Thisremovesfurthersevensourcesfromthesampleand are predominantly at low redshift (z < 1.2) with typical red- leaves eight candidates, all unresolved in the GROND images shift probability distributions that are flat down to z = 0. For andwithzphot,pl >1.2. thosewegivethe90%upperlimitofthephotometricredshift.. Forthreeadditionalsourcesthepowerlawandgalaxymod- Thissub-samplealsoincludeselevensourceswithapreviously els fit similarly well (P ∼ P > 90%) and give compara- knownredshiftfromspectroscopyorimaging,allin agreement z,pl z,gal blehigh-redshiftsolutions,indicatingthattheLyman-limitwas withourlimits. fit by power law and galaxy templates alike. Including those, We also provide the 90% upper limit for four sources for weendupwithasampleofelevensourceswithreliablephoto- which the power-law and galaxy libraries give similarly good metricredshiftcomputation,all,withz rangingfromz ≈ constraints (P ≈P > 90%) but with very different red- phot phot z,pl z,gal 1.28 to z ≈ 1.92. This encompasses eight sources without shift solutions. This degeneracy can occur, e.g., when a break phot previous redshift measurements and three blazars with known in the SED issimilarly wellfitby theLymanlimitof a power- redshift. For the latter, the photometric redshifts are within 3- lawmodelandwiththe4000Å breakofagalaxytemplate.No σ of the known value (see Fig. 5), indicating an accuracy of z is given for ten sources for which no satisfactory fit phot,best ∆z/(1+zspec)<0.12forourzphotmethoddescribedinSect.4.4. (χ2 >30)wasobtained. ThefitresultsandSEDsfortheelevensourcesarepresentedin Tab.4andFig.6,respectively. For21sources,thegalaxylibraryprovidesthebetterfitwith 6. Discussion&Conclusion P > 90%, P < 90%, and χ2 < 30. All of these are z,gal z,pl gal best described by starburst/QSO hybrid templates at z < 1.3 In this paper we presented redshift constraints for 103 blazars andnonerequireanellipticalgalaxy,i.e.,astrong4000Åbreak. from the 2LAC catalog using UV-to-near-IR multi-band pho- The most significant difference between the starburst/QSO hy- tometry obtained quasi-simultaneously with Swift/UVOT and bridtemplatesandthe power-lawmodelsisthattheformerex- GROND. We provided the first reliable redshift measurements hibitstrongemission linesoriginatingfromstar-formation(see foreightsourcesandnewupperlimitsforanadditional66tar- Salvatoetal. 2009, for more details on the templates). These gets. Of the eight sources with reliable redshift, seven are lo- lines can be fit to small deviations in the photometry from a catedatz > 1.3.Six of thosebelongto the BL Lac popula- phot power-lawandthusprovidelowχ2 valuesandapparentlywell- tion.Forcomparison,outofthe total395BL Lac in the2LAC constrained z solutions. However, as described in Sect. 4.3, sample only two sources have previously been known to lie at phot the variability correction is not reliable when applied to non- z > 1.3 (Fig. 7). Redshifts for these two, PKS 0332-403 and power-law sources, as the color terms, and thus the correction CRATES J1312-2156,have also been confirmedwith our pho- factors, will be erroneous. Also, 18 of these sources have pre- tometricobservations. A.Rau,etal.:BLLacobjectsbeyondredshift1.3-UV-to-NIRphotometryandphotometricredshiftforFermi/LATblazars 7 4 r’=17mag 3 rr’’==1198mmaagg 8 z<1.2 z=1.28 hot 2 zp 4 6 z=1.37 P > 90% Z 3 y) 1 zphot2 µJ z=1.44 x ( 1 g Flu µ(arbitrary) log Flux (Jy)4 z=1.45 0 00 1 2 3 4 o z=1.47 00..440 1 2 3 4 y) l z)z)simsim 00..22 bitrar z=1.54 ++ 00..00 ar z=1.55 (1(1 ( 2 ∆∆z/z/ --00..22 z=1.60 --00..44 z=1.72 00 11 22 33 44 z=1.77 %) 0 10 z=1.92 ( η 0.1 1.0 Wavelength (µm) 1 0 1 2 3 4 z Fig.6.SEDs forthe elevensourcesforwhichphotometricred- sim shifts larger than 1.2 were estimated (see Tab. 4) and one ex- ample for z < 1.2. From bottom to top: CRATES J0402- Fig.5.Recoveredbest-fitphotometricvsinputredshift(top),ac- phot 2615, CRATES J1312-2156, OM 235, CRATES J0630- curacy (middle), and outlier fraction (bottom) for 27,000 sim- ulated sources with spectral slope of 0.05 ≤ β ≤ 2 and op- 2406, SUMSS J053748-571828, PKS 0600-749, PKS 0332- tical brightness of r′ = 17,18,19mag. The pink circles indi- 403, CLASS J2352+1749, PKS 0047+023, PKS 0055-328, RXJ0035.2+1515,andBZBJ0543-5532. cate the location of three blazars in our sample with known redshift, PKS 0332-403 (z = 1.351, Bergeronetal. 2011), spec OM 235 (z = 1.549), and CRATES J1312-2156 (z = spec spec troscopyreliesonthedetectionofveryfaintemissionfeatures, 1.491, Ackermannetal. 2011), as well as that of PKS 0537- 286 (z = 3.104, Wrightetal. 1978), a LAT-non-detected mainly from the underlying host galaxy. It therefore requires spec very high signal-to-noise, often at the cost of long exposure blazarwithavailableGRONDandUVOTphotometry.Thegreen times. On the other hand, the photometric redshift method for areainthemiddlepanelcorrespondstotheforbiddenparameter space where z > 0. Systematics are visible at z < 0.8. phot phot The large majority of the fits have χ2 < 30 (for typically 10 d.o.f.)with the exceptionsshownin lightgrey.The insetin the 1144 top paneldisplays the results if only solutions with P > 90% z areselected. 1122 2LAC redshift sample 1100 c a The six new BL Lac redshifts at z > 1.3 represent a dra- L L 88 matic(fromtwotoeight)increaseoftheconfirmedhigh-zFermi B sampleandthusdemonstratetheopportunitythattheSEDtem- of 66 this paper platefittingtechniqueholdsforobtainingphotometricredshifts # 44 for BL Lac sources. For our sample, this was possible due to thedensely-coveredwidespectralrange(160–2200nm),neces- 22 saryfora reliableconstraintof thespectralslope,the excellent ultra-violetcoveragetomeasuretheLyman-limit,andthequasi- 00 simultaneity of the observations, important for minimizing the 00..55 11..00 11..55 22..00 impactofsource-intrinsicvariability.Themethodappliedinthis z workovercomessignificantchallengesinherentinotherredshift techniques,namelythesimplicityoftheopticalemissionofBL Fig.7.RedshiftdistributionforthenineBLLacobjectswithre- Lacs manifested as the power-law-shapedsynchrotron spectral liablephotometricredshifts(bluefilledhistogram)togetherwith component. This complicates spectroscopic redshift measure- thedistributionfortheBLLacinthe2LACcatalog(emptyhis- ments,whichareinmostcaseslimitedtotheopticalwavelength togram;Ackermannetal.2011).Sourceswithz<0.5havebeen regime and thus insensitive to the Lyman-limit. Instead, spec- omittedforclarity. 8 A.Rau,etal.:BLLacobjectsbeyondredshift1.3-UV-to-NIRphotometryandphotometricredshiftforFermi/LATblazars BLLacdoesnotrequiresignificanttelescopetime,inparticular Fossati, G., Maraschi, L., Celotti, A., Comastri, A., & Ghisellini, G. 1998, whendatafromefficientmulti-channelinstrumentslikeGROND MNRAS,299,433 are available. Its weakness, however, is that the method relies Gehrels,N.,Chincarini,G.,Giommi,P.,etal.2004,ApJ,611,1005 Ghisellini, G., Celotti, A., Fossati, G., Maraschi, L., & Comastri, A. 1998, on the detection of a particular spectral feature, which, albeit MNRAS,301,451 strong,islocatedintherest-framefar-ultra-violet.Forz.3,this Giommi, P., Padovani, P., Polenta, G., et al. 2011, MNRAS, in press, is only accessible with space-based ultra-violettelescopes, and arXiv:1110.4706 even then usually limited to redshifts above z ≈ 1.2, as shown Greiner,J.,Bornemann,W.,Clemens,C.,etal.2008,PASP,120,405 Hoffmeister,C.1929,AstronomischeNachrichten,236,233 above. Ilbert,O.,Arnouts,S.,McCracken,H.J.,etal.2006,A&A,457,841 The sensitivity of this method to high-redshiftsources also Kataoka,J.,Madejski,G.,Sikora,M.,etal.2008,ApJ,672,787 allowstheplacementofupperlimitsforthoseSEDsthatdonot Kru¨hler,T.,Ku¨pcu¨Yoldas¸,A.,Greiner,J.,etal.2008,ApJ,685,376 show the imprint of the Lyman-limit. Only for three targets is Kru¨hler,T.,Schady,P.,Greiner,J.,etal.2011,A&A,526,A153 thisboundaryabovez = 2.In onecase, CRATESJ0250+1708 Meisner,A.M.&Romani,R.W.2010,ApJ,712,14 Miller,H.R.,Carini,M.T.,&Goodrich,B.D.1989,Nature,337,627 (z < 3.1), the optical counterpart is too faint (r′ ≈ 20.7mag) Pickles,A.J.1998,PASP,110,863 andonlyupperlimitsinallsixUVOTbandscouldbeobtained. Poole,T.S.,Breeveld,A.A.,Page,M.J.,etal.2008,MNRAS,383,627 ThecounterpartofCRATESJ0705-4847isonlydetectedinthe Racine,R.1970,ApJ,159,L99 GRONDJ(≈19.7mag )andHbandsanditsSEDandredshift Raiteri,C.M.,Villata,M.,Ibrahimov,M.A.,etal.2005,A&A,438,39 AB Roming,P.W.A.,Kennedy,T.E.,Mason,K.O.,etal.2005,SpaceSci.Rev., arethuspoorlyconstrained.Finally,theSEDofATG20J0124- 120,95 0625 (z < 2.46) shows a significant break at z ∼ 1.9 ± 0.5. Salvato,M.,Hasinger,G.,Ilbert,O.,etal.2009,ApJ,690,1250 However, as the redshift probability distribution is broad, and Salvato,M.,Ilbert,O.,Hasinger,G.,etal.2011,arXiv:1108:6061 the χ2 of the powerlaw fit comparableto thatof a z ≈ 0.4 so- Sbarufatti,B.,Treves,A.,&Falomo,R.2005,ApJ,635,173 Schlegel,D.J.,Finkbeiner,D.P.,&Davis,M.1998,ApJ,500,525 lutionforagalaxytemplate,thephotometricredshiftisconsid- Shaw,M.S.,Romani,R.W.,Healey,S.E.,etal.2009,ApJ,704,477 eredtobeunreliable.Exceptforthosethree,noothersourcein Skrutskie,M.F.,Cutri,R.M.,Stiening,R.,etal.2006,AJ,131,1163 our sample has a best-fit photometric redshift at z > 2. Thus, Urry,C.M.,Maraschi,L.,Edelson,R.,etal.1993,ApJ,411,614 CRATES J0402-2615 can now be considered the most distant Ve´ron-Cetty,M.-P.&Ve´ron,P.2010,A&A,518,A10 knownBLLacwithameasuredredshiftofz≈1.92. Wright,A.E.,Peterson,B.A.,Jauncey,D.L.,&Condon,J.J.1978,ApJ,226, L61 Fig.7indicatesthatourhigh-redshiftfindingsarethenatural extensionoftheexisting2LACredshiftsample.However,dueto theincompletenessesofbothsamples,werefrainfromdrawing anyquantitativeconclusionsatthisstage.Wenote,however,the our result is generalagreementwith the theoreticalpredictions fromGiommietal.(2011).Amoredetailedphysicalinterpreta- tion of the establishmentof an increased fractionof high-zBL Lacsinthe2LACsamplewillbereportedseparately. Acknowledgements. Wethank the referee for the valuable comments. Part of thefundingforGROND(bothhardwareaswellaspersonnel)wasgenerously grantedfromtheLeibniz-PrizetoProf.G.Hasinger(DFGgrantHA1850/28-1). MSacknowledges supportbytheGermanDeutsche Forschungsgemeinschaft, DFGLeibnizPrize(FKZHA1850/28-1).TKacknowledgessupportbytheDFG cluster of excellence Origin and Structure of the Universe, by the European Commission under the Marie Curie Intra-European Fellowship Programme), as well as the DARK: The Dark Cosmology Centre, funded by the Danish NationalResearchFoundation.FOEacknowledgesfundingofhisPh.D.through theDeutscher AkademischerAustausch-Dienst (DAAD).SK,DAKandANG acknowledgesupportbyDFGgrantKl766/16-1.ARossiacknowledgessupport fromtheBLANCEFLORBoncompagni-Ludovisi,ne´Bildtfoundation.MNac- knowledgessupportbyDFGgrantSA2001/2-1.PSacknowledges supportby DFGgrantSA2001/1-1.ACU,ANG,DAKandARossiaregratefulfortravel fundingsupportthroughMPE.Wealsoacknowledge theuseoftheTOPCAT tool(Taylor2005). References Abdo,A.A.,Ackermann,M.,Ajello,M.,etal.2010a,ApJ,715,429 Abdo,A.A.,Ackermann,M.,Ajello,M.,etal.2010b,ApJS,188,405 Abdo,A.A.,Ackermann,M.,Ajello,M.,etal.2010c,ApJ,723,1082 Abdo,A.A.,Ackermann,M.,Ajello,M.,etal.2010d,ApJ,710,1271 Ackermann,M.,Ajello,M.,Allafort,A.,etal.2011,arXiv:1108.1420 Aihara,H.,AllendePrieto,C.,An,D.,etal.2011,ApJS,193,29 Arnouts,S.,Cristiani,S.,Moscardini,L.,etal.1999,MNRAS,310,540 Atwood,W.B.,Abdo,A.A.,Ackermann,M.,etal.2009,ApJ,697,1071 Bergeron,J.,Boisse´,P.,&Me´nard,B.2011,A&A,525,A51 Blandford,R.D.&Rees,M.J.1978,inBLLacObjects,ed.A.M.Wolfe,328– 341 Bohlin,R.C.,Colina,L.,&Finley,D.S.1995,AJ,110,1316 Cardelli,J.A.,Clayton,G.C.,&Mathis,J.S.1989,ApJ,345,245 Carini,M.T.,Miller,H.R.,Noble,J.C.,&Sadun,A.C.1991,AJ,101,1196 Chabrier,G.,Baraffe,I.,Allard,F.,&Hauschildt,P.2000,ApJ,542,464 Ciprini,S.,Tosti,G.,Raiteri,C.M.,etal.2003,A&A,400,487