Draftversion January13,2010 PreprinttypesetusingLATEXstyleemulateapjv.11/10/09 EVIDENCE FOR INFRARED-FAINT RADIO SOURCES AS Z >1 RADIO-LOUD AGN Minh T. Huynh InfraredProcessingandAnalysisCenter,,MS220-6,CaliforniaInstitute ofTechnology, PasadenaCA91125,USA. [email protected] Ray P. Norris AustraliaTelescopeNational Facility,CSIRO,EppingNSW1710,Australia 0 1 0 Brian Siana 2 CaliforniaInstitute ofTechnology, MS105-24,Pasadena,CA91125,USA n a Enno Middelberg J AstronomischesInstitut, Ruhr-UniversitatBochum,Universitatsstr. 150,44801Bochum,Germany 3 Draft version January 13, 2010 1 ABSTRACT ] O Infrared-Faint Radio Sources (IFRSs) are a class of radio objects found in the Australia Telescope LargeArea Survey(ATLAS) whichhaveno observablemid-infraredcounterpartinthe Spitzer Wide- C area InfraredExtragalactic(SWIRE) survey. The extended Chandra Deep Field South now has even . deeper Spitzer imaging (3.6 to 70 µm) froma number of Legacy surveys. We reportthe detections of h two IFRS sources in IRAC images. The non-detection of two other IFRSs allows us to constrain the p - source type. Detailed modeling of the SED of these objects shows that they are consistent with high o redshift (z &1) AGN. r t Subject headings: galaxies: formation — galaxies: evolution — galaxies: starburst s a [ 1. INTRODUCTION threshold. 1 Infrared-Faint Radio Sources (IFRSs) were recently Recently, Middelberg et al. (2008b) and Norris et al. v discovered in the Australia Telescope Large Area Sur- (2007) targeted6 IFRSs withthe AustralianLongBase- 4 line Array (LBA) and successfully detected two of the vey (ATLAS) by Norris et al. (2006). These are ra- 8 sources. The Norris et al. (2007) LBA detection con- dio sources brighter than a few hundred µJy at 1.4 0 strained the source size to less than 0.03 arcsec, sug- GHz which have no observable infrared counterpart 2 gesting a compact radio core powered by an AGN. in the Spitzer Wide-area Infrared Extragalactic Sur- . Middelberg et al. (2008b) found the size and radio lu- 1 vey (SWIRE) (Lonsdale et al. 2004). They may be re- minosity of their LBA-detected source to be consistent 0 lated to the optically invisible radio sources found by 0 Higdon et al. (2005, 2008), which are compact radio with a high redshift (z > 1) Compact Steep Spectrum 1 sources with no optical counterpart to R∼25.7. Source. The VLBI detections rule out the possibility : Norris et al.(2006)andMiddelberg et al.(2008a)have that these particular IFRSs are simply the radio lobes v of unidentified radio galaxies. Garn & Alexander (2008) i identified53suchsourcesoutof2002radiosourcesinthe X ATLASsurvey. Mosthavefluxdensitiesofafewhundred stacked IFRS sources in the Spitzer First Look Survey at infrared wavelengths as well as at 610 MHz. They r µJy at 1.4 GHz, but some are as bright as 20 mJy. The a IFRS sources were unexpected as SWIRE was thought find that the IFRS sources can be modelled as compact FanaroffRiley type II (FRII) radio galaxies at high red- tobe deepenoughtodetectallradiosourcesinthe local shift(z ∼4). Thusthe evidencesuggeststhatIFRSs are universe, regardless of whether star formation or Active predominately high redshift radio-loud AGN. GalacticNuclei(AGN)poweredtheradioemission. Pos- Ultra-deepSpitzerimagingisnowavailableoverthe30 sibleexplanationsarethatthesesourcesarei)extremely × 30 arcmin2 region of the Chandra Deep Field South high-redshift (z > 3) radio-loud AGN, ii) very obscured SWIREfield. FourIFRSslieinthisregionandinthispa- radio galaxies at more moderate redshifts (1 < z < 2), per we report on the constraints on the nature of IFRSs iii)lobes ofnearbybutunidentified radiogalaxies,oriv) derived from the new Spitzer data. We assume a Hub- an unknown type of object. The nature of IFRSs has been hard to determine be- ble constantof71kms−1Mpc−1,andmatter andcosmo- cause they have only been detected in the radio. Spec- logical constant density parameters of ΩM = 0.27 and troscopy is difficult because the hosts are optically faint ΩΛ =0.73 in this paper. andtheradiopositionscanalsohaveuncertaintiesonthe order of a few arcseconds. Norris et al. (2006) stacked 2. OBSERVATIONS,DATAANDSAMPLESELECTION the positions of 22 IFRSs in the Spitzer 3.6 µm IRAC The extended Chandra Deep Field South is centered images and found no detection in the averaged image, at 3h32m28s, −27◦ 48′ 30′′. It overlaps the Great Ob- showing that they are well below the SWIRE detection servatories Origins Deep Survey (GOODS) South field, 2 which is one of the best studied regions of the sky. The one of the IFRS sources (S415) lies within the X-Ray ATLAS survey (Norris et al. 2006) consists of deep ra- coverageand there is no detection. dio observations of a 3.7 degree2 field surrounding the The photometry of the IFRSs was fitted with the fol- eCDFSwhichiscoincidentwithSWIRE(Lonsdale et al. lowing templates: 1) a prototypical starburst ULIRG 2004). The ATLAS 1.4 GHz observations reach 20 µJy (Arp220, Silva et al. 1998), 2) a hot dusty AGN dom- to60µJyrms,withthedeepestregioncoveringthe30× inated ULIRG (Mrk231, photometry from NED1), 3) a 30 arcmin2 eCDFS. ATCA 2.4 GHz observations of the starforming M82-like galaxy (Chary & Elbaz 2001), 4) SWIRE region were obtained by the ATLAS team over a radio-loud galaxy with an extreme radio-optical ratio the last two years and the current image reaches ∼ 0.1 (3C273, photometry from NED), and 5) an old stellar mJy rms (Middelberg et al. in preparation). population from Maraston (2005). The old stellar tem- The SWIRE survey reaches 5σ depths of 3.5, 7.0, plates were fitted only to the two IFRS sources with 41, and 49 µJy , respectively, in the four IRAC bands IRACdetections(S446andS506). TheMaraston(2005) (Lonsdale et al. 2004). In the MIPS bands at 24 and templates explored have a metallicity of 0.67, assume a 70 µm the SWIRE 5σ depths are 189 µJy and 16 mJy. SalpeterIMF,andhaveagesof0.3,1and5Gyr. There- To these depths, 22 radio sources in the full SWIRE sultsofthe SpectralEnergyDistribution (SED)analysis Chandra Deep Field South region were undetected, and are described in the next section. classed as IFRS (Norris et al. 2006). This paper focuses on four IFRS sources which lie in the 30 × 30 arcmin2 3. DISCUSSION region of the extended Chandra Deep Field South. 3.1. S283 The eCDFS sub-regionof SWIRE has been the target This 0.287 mJy radio source lies just outside of the of two separate Spitzer Legacy proposals since SWIRE, SIMPLE ultra-deep IRAC imaging and FIDEL 70 µm and hence there are now deeper infrared data. The imaging. There are no SWIRE optical data either, but “SpitzerIRACMUSYCPublicLegacyinE-CDFS(SIM- the shallower SWIRE IRAC 3.6 and 4.5 µm imaging PLE) Legacy project (Damen et al. in prep) provides shows the radio source is positioned between two IRAC ultra-deep imaging of eCDFS in all four IRAC bands, sources (Figure 1), which are 3.5 and 3.7 arcsec offset, and the “Far Infrared Deep Extragalactic Legacy (FI- respectively. The offsets are large (>5 times the uncer- DEL, PI: Dickinson) Legacy project adds ultra-deep 24 taintyof the radioposition) so wedo not considereither and 70 µm observations. The 5σ depths of SIMPLE are IRAC source to be a plausible counterpart. The FIDEL about 0.8, 1.2, 6.3 and 6.6 µJy in IRAC channels 1 to 4, 24 µm imaging shows some possible flux at the source respectively. FIDEL achieves 5σ depths of 50 µJy and 3 position, but it is likely that this faint flux is confusion mJyat24and70µm,respectively,intheareaswithmost from the two nearby IRAC sources. To verify that there coverage. We searchedforcounterpartstothefourIFRS isnodetectionoftheradiosourceintheIRACandMIPS sources in the new Spitzer data. A matching radius of 24µmbandsthetwopointsourcesweresubtractedfrom up to 3 arcsec was used for the IRAC and MIPS 24 µm the images and the residual image is shown in Figure 2. data, and 8 arcsec for MIPS 70 µm. All four positions The residual flux at the radio position is 2.4±1.8 µJy were examined by eye in all Spitzer bands for any obvi- at 3.6 µm and 13±7 µJy at 24 µm. This is consistent ous failures (see Figure 1). We find two of the sources with the conservative upper limits we have assumed of (S446andS506)weredetectedintheSIMPLEultra-deep 3.5 µJy at 3.6 µm and 100 µJy at 24 µm. IRAC imaging (see Figure 1) and we have improved IR Ifthis sourcewasatypicalULIRGlikeArp220orMrk constraintsforthe rest. Table1summarisesthe datafor 231,evenforz >2,itwouldbe detected byIRAC imag- the four IFRSs. ing (see Figure 3). For z < 2 the 24 µm flux density We have also searched for counterparts in existing expected is >> 100 µJy so it would be clearly detected deepopticalandnear-infrared(NIR) data ofthe eCDFS in MIPS. A starforming galaxy such as M82 would also fromGOODS(Giavalisco et al.2004),GEMS(Rix et al. have clearly been detected by Spitzer. While a redshift 2004)andMUSYC(Gawiser et al.2006),andfindanop- less than 1 is allowed by the 3C273 SED, that would tical/NIRdetectionofonlyoneIFRSsource(S506). The imply a low mass object. The stellar mass estimate at GOODS ACS imaging reaches approximately B = 28.1 , V = 28.9 , I =28.3 and z = 27.4. GAEBMS z = 1 is M ∼ 6×109M⊙ assuming the K band lumi- AB AB AB nosity from the 3C273 SED fit and a mass to light ratio comprises two band ACS imaging and reaches depths of of 1 (e.g. Bell & de Jong 2001). This is comparable to V =28.25andz =27.1. TheMUSYCsurveyhasa AB AB the most massive dwarf galaxies, which are not known 5σ sensitivity of about J ∼ 22.5 and K ∼ 22.0 AB AB to be strong radio emitters or contain AGN. The IRAC (Taylor et al. 2009). Two IFRSs, S415 and S506, lie within the deep optical/NIRimaging area and only one, limits alone suggest a stellar mass of M . 6×1010M⊙ at z = 1. Radio-loud AGN with SEDs similar to 3C273 S506, is detected as a very faint source. The lack of a detection in these optical and near-IR images for S415 are found in massive galaxies with M > 1011M⊙. The IRAC non-detection therefore suggests S283 is a radio- stronglyimplies that it is athigh redshift and/or is very loud AGN at z > 1. This is supported by the galaxy’s obscured. MIR-radiocorrelation,q =log(S /S )<−0.46, A2MsX-RayexposureofGOODSSouthwasobtained 24µm 1.4GHz which is well in the range expected for radio-loud AGN with the Chandra X-ray Observatory (Giacconi et al. (e.g. Boyle et al. 2007). 2002; Luo et al. 2008). The on-axis full band sensitiv- ity reaches 7.1 × 10−17erg s−1 cm−2 at the average aim 1 http://nedwww.ipac.caltech.edu/. The NASA/IPAC Extra- pointandthe minimumfullbandsensitivityisabout3.3 galacticDatabase (NED)isoperatedbytheJetPropulsionLabo- × 10−16erg s−1 cm−2 over the GOODS-S region. Only ratory,CaliforniaInstituteofTechnology,undercontractwiththe NationalAeronauticsandSpaceAdministration. 3 3.2. S415 1986). The P-statistic is only a rough estimate because it does not does not take into account the individual This IFRS has a flux density of 1.21 mJy at 1.4 GHz, positional uncertainties and assumes a random distribu- but is undetected in all Spitzer bands. Since it lies in tion of the background population, whereas astronomi- the GOODS-S proper, where the data is most sensi- cal sources are clustered. Nevertheless, for the SIMPLE tive, it has by far the most extreme flux density ratios IRAC source density of 116700/deg2 the P-statistic sug- of the sources discussed here. It is as extreme, if not gests that the chances of an IRAC source lying closer moreso,thantheopticallyfaintbutsubmmbrightHDF than 2.2 arcsec is 13%, so the counterpart is reasonably 850.1 (Dunlop et al. 2004; Cowie et al. 2009). This ra- reliable. The source is 6.6 ± 0.3 µJy and 5.7 ± 0.5 µJy diosourcehasnoACScounterpartandwemeasuredthe at 3.6 and 4.5 µm, respectively, and not detected in the 3σ ACS limits at the source position to be B &28.1. AB other IRAC bands. There is no detection in the 70 and V & 28.9, I & 28.3, and z & 27.4, using a 0.6” AB AB AB 24 µm images, but limits are hard to quantify because diameteraperture. Thereisnonear-infraredcounterpart thesourcefallsinbetween2brightersources. Weassume in the MUSYC or VLT ISAAC observations, giving 3σ thenoisehereistwicethatofthelocalnoiseintheMIPS limits of J &25.5, H &25.8 and Ks &25.5. AB AB AB imaging for the purposes of the SED fitting. If this source has an SED similar to M82, Arp220 or Similar to the other IFRS sources, the non-detection Mrk231itwouldbedetectedinallSpitzerbands,even70 in the two longer wavelength IRAC bands, and MIPS µm, for z <4. If the radio-loud 3C273 SED is adopted, 24 and 70 µm bands, rules out M82, Arp220 and Mrk this source must lie at z &3 for it to remain undetected 231 SEDs for this source, at a wide range of redshift in the IRAC and ACS bands (assuming no obscuration, (z < 6). A 3C273 SED would not produce the MIR see Figure 4). At redshift z = 1, the IRAC 3.6 and 4.5 emissiondetectedintheIRACbands(seeFigure5),soa µm limits are a factor of 2 to 3 below the 3C273 SED, possible explanation for this source is a radio-loudAGN which implies an extinction of 0.8 to 1.1 mags would which dominates the radio emission but with a stellar berequiredforthissourcetobe undetectedintheIRAC component which is seen in the MIR. channels. Thislevelofextinctionisseeninsomeextreme The Maraston (2005) old stellar population fit to the ULIRGs (e.g. Genzel et al. 1998; Murphy et al. 2001), IRAC data constrains the redshift of this galaxy to 1 but this source is not luminous in the infrared. This – 1.5, and the best fit is a 1 Gyr old model at z = 1.5. source therefore lies at redshift z >> 1. Even at these AtthisredshifttheradioluminosityofS446isP = redshifts obscurationof a few magnitudes is requiredfor 1.4GHz 3.7×1024WHz−1,whichwouldplaceitatthelowpower a 3C273 object to be undetectable in the optical/NIR end of radio-loud AGN. bands (see Figure 4). At z & 1 the radio luminosity of S415 is P > 5×1024 W Hz−1. The Spitzer non- Using Maraston (2005) SEDs with a different metal- 1.4GHz licity, including a different library of SEDs (e.g. detection of S415 therefore suggests that this source is a Bruzual & Charlot 2003), or adding reddening as a free distant (z >>1) obscured radio-loud AGN. parameterwouldgiveadifferentbest-fitredshift,butthis The X-Ray non-detection implies that this source has zan=X2-Ranady Llu0m.5−in8okseiVty.L60.×5−18k0e4V2 e.rg2s−×11a0t4z2 =erg3,sw−h1earet ldueanvtleiaklpeoolyfincttoosm.bLpealescxtailutyyswecdeanbnoyntetohttehbahetottehxdepulIosRtrAecdComwdpietothenceotnniotlnyosftawarnoe the Luo et al. (2008) limit assumes an X-Ray power law AGNtorus. TheIRACdetectionsimplytheMIRpeakis of Γ=1.4. This suggests that, for these redshifts, either shorter than 3.6 µm, so the dust temperatures would be the source has a very highcolumn density ofneutral hy- greater than ∼1600 K at z = 1 and greater than ∼2400 drogenobscuringtheX-Rayemission,whichisconsistent K at z = 2, using Wien’s law. Silicate grains sublimate with the optical/NIR non-detection, or the source is in- atabout1000K,whilegraphitegrainssublimatearound trinsically X-Ray faint for an AGN. 1500 K, so these temperatures are too high for an AGN A preliminary analysis of ATCA 2.4 GHz data (Mid- torus. Instead,AGNtorusmodelstypicallyshowacooler delbergetal. inpreparation)findsS415hasasignificant MIRemissionpeakof7to10µm(e.g. Schartmann et al. detection (> 7σ) of 0.67 mJy. This implies a steep ra- 2005), which is longwardof IRAC channels 1 and 2. dio spectral slope of α = −1.1 2, which can not be pro- ducedbystarformingprocesses. S415hasaslopesimilar 3.4. S506 to some ultra steep spectrum sources (Rottgering et al. This source is similar to S446. This 0.170 mJy ra- 1997; De Breuck et al. 2004), which have been linked to dio source has a GEMS ACS V and z band detection massive high-redshift (z >2) radio galaxies. 1.4” south of the radio position that is 26.27 and 25.62 AB magnitudes, respectively (see Figure 1). The likely 3.3. S446 counterpartin the IRAC 3.6 and 4.5 µm channels is less IFRS S446 is a 0.338 mJy radio source which lies just than 1” away from the ACS source, roughly 2.3” south outside the GEMS ACS coverage and MUSYC near- oftheradioposition. ThisACScounterpart,while faint, infrared imaging. There is a faint source in the IRAC is a 7.0 and 5.1 sigma detection in the V and z bands, 3.6 and 4.5 µm channels just 2.2 arcsec north of the ra- respectively. dio position (see Figure 1) and we assume this is the The P-statistic suggests the chance of this IRAC counterpart to S446. The probability that one or more source being a random coincidence is about 14%. The IRAC sources lies randomly within a distance θ of a ra- source is 5.5 ± 0.3 µJy and 5.5 ± 0.4 µJy at 3.6 and 4.5 dio source is P = 1−exp(−πnθ2), for an IRAC source µm, respectively, and not detected in the other IRAC densityn(oftencalledtheP-statistic;e.g. Downes et al. bands. There is no MIPS 70 or 24 µm detection in the FIDELimagesofthissource. TheFIDELimageshowsa 2 S∝να possible 24 µm excess but this has a flux density of only 4 7.3 ± 4.2 µJy and it is not coincident with the IRAC We now consider whether IFRSs may be galactic ob- source. There is also no detection in the MUSYC NIR jects such as pulsars or radio stars. The majority of imaging (J & 22.5 and K & 22.0). Again, M82, known pulsars are young objects with a low spin rate AB AB Arp220 and Mrk 231 type SEDs are ruled out for this andthese aredistributed acrossthe Galactic disc, asex- source by the non-detection in the MIPS 24 and 70 µm pected for objects that have originated fairly recently bands. The MIR peak is between 3.6 and 4.5 µm and from massive stellar supernovae. They are rare objects if hot dust is responsible for this then Wien’s law im- athighgalacticlatitude. Forexample,theParkesMulti- plieshotdusttemperaturesof∼1400Katredshift1and beam Pulsar Survey, with a 1.4 GHz flux density limit ∼2200Katredshift2. SimilartoS446,thisrulesoutthe of 0.15 mJy for long period pulsars, finds a density of IRAC detection of a hot AGN tori. The most likely ex- about 1/deg2 for |b| < 1 (Camilo et al. 2000). The den- planationforthis sourceisa radio-loudAGNat(z >2) sity of these pulsars drops to <0.25/deg2 by |b|=4 deg. withastellarcomponentdominatingthe IRACchannels Applying the P-statistic from Section 3.3, we find the (Figure 6). probabilityofoneormorepulsarswithin15arcmin(half Using the IRAC data alone, we find the best fit the eCDFS size) of a radio position is less than 5%, as- Maraston (2005) old stellar population model to the suming a pulsar densityof <0.25/deg2. The eCDFS has IRACdatais1Gyroldandplacesthisgalaxyatz =2.5. agalacticlatitudeofabout-55degrees,sothespaceden- The 5 Gyr model has a redder SED that would be de- sityofpulsarsinthe eCDFSfieldismuchlessthanthat. tected by IRAC 5.8µm imaging atz < 2,and the 1 Gyr Thereforethe0.25degreeeCDFSfieldisunlikelytocon- model becomes a bad fit for z > 2.6. However the ACS tain a pulsar. detection suggests a blue excess from star formation ac- Shorter period pulsars, so-called millisecond pulsars, tivity. ThebestfitstellarpopulationmodeltobothACS are thought to have been spun up to high rotation rates andIRACdetectionsisa0.3GyrMaraston(2005)model asamemberofalow-massX-raybinary(LMXB)system. at z =2.0. As for S446, the small number of datapoints These objects are usually found in globular clusters (e.g doesnotwarrantexploringallthe freeparametersinthe Manchester et al. 2001). The eCDFS is not a globular redshift fitting, such as metallicity and reddening. How- cluster field and so is unlikely to contain a LMXB. ever, SEDs younger than 0.3 Gyr were also explored for The optical non-detection of the radio sources gives S506 and these young SEDs have a blue excess inconsis- us another clue about whether the sources are pulsars. tent with the faintness of S506 in the optical bands. At Neutron stars have been detected in the optical, and thebestfitredshiftofz =2.0,S506hasaradioluminos- we can take two examples: the Crab pulsar and the ityP1.4GHz =3.6×1024WHz−1,which,similartoS446, Vela pulsar. The Crab pulsar’s optical counterpart is would place it at the low power end of radio-loud AGN. known as Baade’s star and it is relatively bright in the The ACS counterpart is made up of two clumps (see optical, with V = 16.6. The Crab pulsar lies at 2 kpc Figure1),oneofwhichisextendedandhasfaintemission (Manchester et al.2005),soat10kpcitwouldhaveV= over 0.3 arcsec, which is 2.5 kpc at redshift z = 2. The 20.1 and V = 25.1 at 100 kpc. Thus, if these IFRSs are dualnatureofS506couldbeexplainedbyonecomponent asbrightastheCrabpulsarinthe opticaltheywouldbe havinganAGNwhileone(orboth)hassomerecentstar easilydetectableinthe opticalimaging. The Velapulsar formation. is much fainter however, V = 23.6 (Mignani & Caraveo 2001), and lies at a distance of 294 pc (Caraveo et al. 3.5. IR-radio Correlation 2001). An object such as this would be too faint for TheIR-radiocorrelation(e.g. Yun et al.2001)isanin- eCDFS HST detection if it lies further than about 2.3 dicatorof the dominantemission mechanismin a galaxy kpc. because both IR and radio emission are thought to be X-Ray emission from the neutron star surface or from strongly linked to star formation. Any deviation from thethepulsarwindnebulamaybedetectableiftheIFRS thecorrelationisasignofanAGN.Galaxieswithexcess is a pulsar. The one source with X-Ray data, S415, has radioemissionprobablycontainaradio-loudAGN,while an X-Ray luminosity limit that is 100 times less than IR-excess sources are likely to be radio-quiet AGN with some of the faintest X-Ray emitting pulsars known (e.g. hot dust dominating in the MIR (for z > 1). The ob- Kargaltsev & Pavlov2009). X-Rayemissionfromaneu- servedMIPS to radio flux density ratio limits are shown tron star may not be orientated the same way as the in Figure 7. Some of the xFLS galaxies (Frayer et al. radio emission, and the X-Ray-radio observations were 2006) have ratios consistent with the IFRSs but all are not simultaneous, so this is not conclusive evidence that easily detected at optical and infrared wavelengths. The S415 is not a pulsar. It is however consistent with the S /S limitsforall4IFRSsshowthattheyhave idea that S415 is an extragalactic source. 24µm 1.4GHz excess radio emission for z . 2. The limits from 70 µm CouldIFRSs be mainsequencestars? Ultracooldwarf are not as stringent as that from 24 µm, but they do stars, a class thought to be radio active, were observed show that the IFRSs have as much excess radio emis- at 4.8 GHz, but only one was detected in a survey of sion as the AGN-dominated ULIRG Mrk231. The ex- eight(Antonova et al. 2008). This star lies at a distance treme source S415 has a ratio that is consistent with a of 12.2 pc and has a 4.8 GHz flux density of 0.286 mJy. radio-loud AGN at z & 4. The IFRS IR-radio ratios Berger(2006) observed90M andbrowndwarfstarsand areconsistentwithasampleofhighredshiftradiogalax- found 8.5 GHz flux densities wellbelow 1 mJy, although ies (HzRGs, Seymour et al. 2007), which host luminous flarescanincreasethe flux density by atleasta factorof radio-loud AGN. a few. The stars in this sample are all closer than about 13 pc and have J and K magnitudes brighter than 18 3.6. Could IFRSs be galactic objects? mag. ThetypicalM-dwarfhasanabsolutemagnitude of 5 8-17M (e.g. Kaler1997),soatadistanceof1kpc(10 IFRSs maybe the lower luminosity analogs of HzRGs. V kpc) M-dwarfs have an observed magnitude of V = 18 There are only a few hundred known HzRGs across the - 27 (23.1 - 32.1). Therefore the weakest of these would full sky, so they are even rarer than IFRSs. be undetected in the optical images at large distances, buttheradiostudiessuggestM-dwarfstarscloseenough 4. SUMMARY to be detected in the radio would be seen in the optical We have searched for infrared counterparts to four images. Infrared Faint Radio Sources in the extended Chandra DeepFieldSouthfieldusingrecentlyavailableultra-deep 3.7. Comparison to other galaxy populations Spitzer observations. No IFRS is detected in ultra-deep We can compare the space density of our sample of 24 µm imaging, implying that the sources do not fol- IFRSswiththatofotherhighredshiftz ∼2samples. For low the IR-radio correlation and star formation can not example,muchworkhasbeendoneontheBzK selection produce allthe radioemissionobservedin these objects. technique (Daddi et al. 2004) to select star-forming and We find IRAC detections for two of the sources,and the passive galaxies at z ∼ 2. The space density of BzKs non-detections of the other two provides constraints on is ∼ 1 per arcmin2 (Daddi et al. 2007), more than 200 thesourceSEDs. TypicalULIRGSEDs,suchasMrk231 times greater than that of the IFRSs in this work (∼16 andArp220,andL∗ galaxiesareruledoutbytheSpitzer per deg2). data. The most likely explanation for these sources is that they are radio-loud AGN, with radio-to-optical ra- AnotherhighredshiftsampleisDust-ObscuredGalax- tios similar to 3C273, at redshifts z & 1. The most ex- ies(DOGs),whichareselectedusingacombinationofred treme source (S415) lies at z >> 1 and requires several colors(R−[24]>14,inVegamagnitudes)andbrightflux magnitudes of obscurationin the optical/NIR to remain densities in infrared (S > 0.3 mJy) (Dey et al. 2008). 24 undetected by deep imaging. It is very unlikely that These criteria proved remarkably efficient for selecting z the IFRSs are Galactic sources,but current data cannot ∼ 2 galaxies. DOGs contribute ∼ 26% of the total IR conclusively rule this out in the case of the undetected luminosity density at z = 2 and 60% of the total from ULIRGs. Theyhaveaspacedensityof∼0.1perarcmin2 sources. or 2.82±0.05×10−5 h3 Mpc−3, similar to submillime- For the two sources with IRAC detections we find the SED can be described by a radio-loud 3C273-like SED tergalaxiesandabout20times morenumerousthanthe combined with a stellar population. The stellar popu- IFRSs. lation SED fits to the IRAC MIR data, combined with The paucity of IFRSs is not a surprise as they were theavailableopticallimits,suggestthatthetwodetected selected to have extreme radio to infrared ratios, and as IFRS sources have redshifts of z ∼1.5 and z ∼ 2.0. For such are rare compared to other populations selected in one source,S506,the ACS detection suggeststhe galaxy the optical and infrared. How do AGN samples selected may have a stellar component that is only 0.3 Gyr in in the radio compare? By matching Faint Images of the age. Thesetwosourcesaresimilartothe IRACdetected RadioSkyatTwenty-cm(FIRST)1.4GHzradiosources OIRSs which are posited to be “red and dead” radio with the Sloan Digital Sky Survey (SDSS), Ivezi´c et al. galaxies at z >1 (Higdon et al. 2008). (2002) find about 3 radio selected AGN per deg2 with Theevidenceismountingthatasignificantproportion, S > 1 mJy and i <21. The four IFRSs in this study, 1.4 ifnotall,oftheIFRSsourcesareradio-loudAGNathigh likely radio-loud AGN at z > 1, have a source density redshift (z &1), and not merely lobes of an unidentified ∼16 per deg2. This suggests that IFRSs are a popula- radio galaxy. The source density in eCDFS is ∼16 per tion of radio-loud AGN which have been unstudied by deg2 for S > 0.1 mJy. Garn & Alexander (2008) find previous radio work. 1.4 a source density of ∼3.5 per deg2 for S > 0.5 mJy. S446 and S506 lie at redshifts 1 to 2 and have radio 1.4 powers (P ) of ∼ 1024 W Hz−1. The local luminos- While these sources are rare, they point to a population 1.4GHz ofAGNathighredshiftthathasbeenundiscovereduntil ityfunctionofradioAGN(Best et al.2005;Sadler et al. recently. 2002) suggests that there are 40 to 60 AGN per deg2 with radio powers of 1024 - 1025 W Hz−1 in the volume between redshifts 1 to 2, assuming no evolution. If lu- This work is based in part on observations made with minosityevolution(e.g. Donoso et al.2009;Sadler et al. the Spitzer Space Telescope, which is operated by the 2007) is applied the number of AGN with radio pow- Jet Propulsion Laboratory, California Institute of Tech- ers of 1024 -1025 W Hz−1 increases to 110 to 140 per nology under a contract with NASA. Support for this deg2. Therefore, the IFRSs presented in this work make work was provided by NASA through an award issued up only a small portion (. 1%) of the total AGN pop- by JPL/Caltech. This research has made use of the ulation with similar radio powers at that epoch. The NASA/IPACExtragalacticDatabase(NED)whichisop- remainder are presumably bright enough to be detected erated by the Jet Propulsion Laboratory, California In- in the Spitzer bands. stitute of Technology, under contract with the National TheIFRSscanalsobecomparedtohighredshiftradio Aeronautics and Space Administration. galaxies(HzRGs). A representative sample of 69 HzRGs at z > 1 were observed with Spitzer (Seymour et al. 2007)andthesemassivegalaxies(1011 -1011.5 M⊙)have the mid-IR luminosities of LIRGs or ULIRGs. While HzRGs have much greater radio flux densities than IFRSs, they have extreme MIR-radio ratios consistent with the available data on IFRSs (Figure 7). Hence, 6 REFERENCES Antonova, A.,Doyle,J.G.,Hallinan,G.,Bourke,S.,&Golden, Higdon,J.L.,etal.2005,ApJ,626,58 A.2008, A&A,487,317 Higdon,J.L.,Higdon,S.J.U.,Willner,S.P.,Brown,M.J.I., Bell,E.F.,&deJong,R.S.2001, ApJ,550,212 Stern,D.,LeFloc’h,E.,&Eisenhardt,P.2008,ApJ,688,885 Berger,E.2006,ApJ,648,629 Ivezi´c,Zˇ.,etal.2002, AJ,124,2364 Best,P.N.,Kauffmann,G.,Heckman,T.M.,&Ivezi´c,Zˇ.2005, Kaler,J.B.1997,StarsandtheirSpectra MNRAS,362,9 Kargaltsev,O.,&Pavlov,G.G.2009, ApJ,702,433 Boyle,B.J.,Cornwell,T.J.,Middelberg,E.,Norris,R.P., Lonsdale,C.,etal.2004,ApJS,154,54 Appleton, P.N.,&Smail,I.2007,MNRAS,376,1182 Luo,B.,etal.2008, ApJS,179,19 Bruzual,G.,&Charlot,S.2003, MNRAS,344,1000 Manchester,R.N.,Hobbs,G.B.,Teoh,A.,&Hobbs,M.2005, Camilo,F.,etal.2000,inAstronomicalSocietyofthePacific AJ,129,1993 ConferenceSeries,Vol.202,IAUColloq.177:Pulsar Manchester,R.N.,etal.2001,MNRAS,328,17 Astronomy-2000andBeyond, ed.M.Kramer,N.Wex,& Maraston,C.2005, MNRAS,362,799 R.Wielebinski,3 Middelberg,E.,etal.2008a,AJ,135,1276 Caraveo,P.A.,DeLuca,A.,Mignani,R.P.,&Bignami,G.F. Middelberg,E.,Norris,R.P.,Tingay,S.,Mao,M.Y.,Phillips, 2001, ApJ,561,930 C.J.,&Hotan,A.W.2008b, A&A,491,435 Chary,R.,&Elbaz,D.2001,ApJ,556,562 Mignani,R.P.,&Caraveo,P.A.2001,A&A,376,213 Cowie,L.L.,Barger,A.J.,Wang,W.-H.,&Williams,J.P.2009, Murphy,T.W.,Jr.,Soifer,B.T.,Matthews,K.,Armus,L.,& ApJ,697,L122 Kiger,J.R.2001, AJ,121,97 Daddi,E.,Cimatti,A.,Renzini,A.,Fontana, A.,Mignoli,M., Norris,R.P.,etal.2006,AJ,132,2409 Pozzetti, L.,Tozzi,P.,&Zamorani,G.2004,ApJ,617,746 Norris,R.P.,Tingay,S.,Phillips,C.,Middelberg,E.,Deller,A., Daddi,E.,etal.2007,ApJ,670,156 &Appleton,P.N.2007,MNRAS,378,1434 DeBreuck,C.,Hunstead, R.W.,Sadler,E.M., Rix,H.-W.,etal.2004,ApJS,152,163 Rocca-Volmerange, B.,&Klamer,I.2004,MNRAS,347,837 Rottgering,H.J.A.,vanOjik,R.,Miley,G.K.,Chambers,K.C., Dey,A.,etal.2008, ApJ,677,943 vanBreugel,W.J.M.,&deKoff,S.1997, A&A,326,505 Donoso,E.,Best,P.N.,&Kauffmann, G.2009,MNRAS,392, Sadler,E.M.,etal.2007,MNRAS,381,211 617 Sadler,E.M.,etal.2002,MNRAS,329,227 Downes,A.J.B.,Peacock, J.A.,Savage,A.,&Carrie,D.R. Schartmann,M.,Meisenheimer,K.,Camenzind,M.,Wolf,S.,& 1986, MNRAS,218,31 Henning,T.2005,A&A,437,861 Dunlop,J.S.,etal.2004,MNRAS,350,769 Seymour,N.,etal.2007, ApJS,171,353 Frayer,D.T.,etal.2006,AJ,131,250 Silva,L.,Granato,G.L.,Bressan,A.,&Danese,L.1998,ApJ, Garn,T.,&Alexander,P.2008, MNRAS,391,1000 509,103 Gawiser,E.,etal.2006,ApJS,162,1 Taylor,E.N.,etal.2009,ApJS,183,295 Genzel,R.,etal.1998,ApJ,498,579 Yun,M.S.,Reddy,N.A.,&Condon,J.J.2001,ApJ,554,803 Giacconi,R.,etal.2002,ApJS,139,369 Giavalisco,M.,etal.2004, ApJ,600,L93 S283 S415 S446 S506 RA(J2000) 3:30:48.686 3:32:13.077 3:32:31.540 3:33:11.486 Dec(J2000) -27:44:45.32 -27:43:51.07 -28:04:33.53 -28:03:19.09 radio1.4GHz 0.287mJy 1.21mJy 0.338mJy 0.170mJy radio2.4GHz <0.40mJy 0.67mJy <0.45mJy <0.45mJy α21..54GGHHzz (S∝να) <0.6 -1.1±0.13 <0.5 <1.8 ACSBmag ··· >28.1 ··· ··· ACSVmag ··· >28.9 ··· 26.27 ACSImag ··· >28.3 ··· ··· ACSzmag ··· >27.4 ··· 25.62 SWIRE g’mag ··· - >25.3 - SWIRE r’mag ··· - >24.8 - SWIRE i’mag ··· - >23.7 - Jmag ··· >25.5 ··· >22.5 Hmag ··· >25.8 ··· ··· Kmag ··· >25.5 ··· >22.0 IRAC3.6µm <3.5µJy <0.8µJy 6.6±0.3µJy 5.5±0.3µJy IRAC4.5µm <7.0µJy <1.2µJy 5.7±0.5µJy 5.5±0.4µJy IRAC5.8µm <41µJy <6.3µJy <6.3µJy <6.3µJy IRAC8.0µm <49µJy <6.6µJy <6.6µJy <6.6µJy MIPS24µm <100µJy <50µJy <80µJy <80µJy MIPS70µm <16mJy <3mJy <5mJy <5mJy TABLE 1 Summaryof the observed propertiesof the fourInfraredFaint RadioSourcesin the extended ChandraDeep Field South. 7 3.6µm 4.5µm 24µm ACSV S283 S415 S446 S506 Fig.1.—GreyscaleSpitzer andACSimages ofS283, S415, S446 andS506. Fromleftto right,imagesareIRAC3.6,4.5µm,MIPS24 µm, and ACS V band. The ACSgrey scaleis inverted for clarity. The green contours arethe radio 1.4GHz imagewith levels set at 50, 100,150,200µJy. Theredcrossmarkstheradioposition. TheSpitzerimagesare25×25arcsecandtheACSimageis12.5×12.5arcsec. The IRAC counterpart to S446 can be seen 2.2 arcsec north of the radio position. The red circle in the S506 ACS image mark the very faintoptical counterpart, whichisconsistentwiththeproposedIRACcounterpart 2arcsecsouthoftheradioposition. Fig.2.— The residual image for S283 with the two point sources near the radio center removed, for IRAC 3.6 µm (left) and MIPS 24 µm (right). As for Figure 1, The green contours are the radio 1.4 GHz image with levels set at 50, 100, 150, 200 µJy and the red cross markstheradioposition. ThereisverylittleIRACorMIPSfluxattheradioposition. 8 Fig. 3.— The SED of S283, fit by Arp220, Mrk231, M82 and 3C273. Points in black assume the galaxy is at z = 2, and blue are the data forz=1. Thisgalaxy isnotwell fitbyM82,Arp220or Mrk231,as itwouldbedetected byIRAC and24µm imaging. TheSED of radio-loudgalaxy3C273 atz>1isconsistentwiththeobservations. TheSEDsarescaledupinluminosityby90,12and2000 timesfor Arp220,Mrk231andM82respectively,and3C273isscaleddownbyafactorof1800. Fig. 4.— The SED of S415, fit by Arp220, Mrk231, M82 and 3C273. Points in black assume the galaxy is at z = 4, and blue are the dataforz=3. ThisgalaxyisnotwellfitbyM82,Arp220orMrk231,asitwouldbedetected bySpitzerimaging. TheSEDofradio-loud galaxy3C273 atz>>1isconsistent withtheobservations, ifobscuration atoptical wavelengths isassumed. TheSEDs arescaled upin luminosityby2.7×106, 3.9× 105 and6.0×107 times forArp220, Mrk231andM82respectively, and3C273 isscaleddownbyafactor of30. 9 Fig. 5.—Left: TheSEDofS446,fitbyArp220Mrk231,M82and3C273. TheSEDsarescaledupinluminosityby750,110and17000 times for Arp220, Mrk231 and M82 respectively, and 3C273 is scaled down by a factor of 550. An old stellar population (red line) from Maraston (2005) is fit to the IRAC data. Points in black assume the galaxy is at z =4, and blue are the data for z =1.5. This galaxy is not well fit by M82, Arp220 or Mrk231, as it would be detected by 24 µm imaging. The SED of radio loud galaxy 3C273 at z ∼2.0 combined with the old stellar population is consistent with the observations. Right: A zoom of the optical/NIR region. The Maraston (2005) stellar population models are shown at the best fit redshift of 1.5. The 0.3 and 5 Gyr models are satisfactory fits, but the 1 Gyr modelprovidesthebestfit. Theopticallimitsareg,r andilimitsfromSWIRE. Fig.6.— Left: The SED of S506, fit by Arp220, Mrk231, M82 and 3C273. The SEDs arescaled up inluminosity by 380, 55 and 8500 times for Arp220, Mrk231 and M82 respectively, and 3C273 is scaled down by a factor of 530. An old stellar population (red line) from Maraston (2005) is fit to the IRAC data. Points in black assume the galaxy is at z = 4, and blue are the data for z =2.0. This galaxy is not well fit by M82, Arp220 or Mrk231, as it would be detected by 24 µm imaging. The SED of radio loud galaxy 3C273 at z ∼ 2.0 combined with the old stellar population is consistent with the observations. Right: A zoom of the optical/NIR region. The Maraston (2005)stellarpopulationmodelsareshownatthebestfitredshiftof2.0. The1and5GyrmodelsfittheIRACdata,butonlythe0.3Gyr model(blackline)canreproducebothIRACandACSdetections. TheJandKlimitsarefromMUSYC. 10 Fig.7.—TheratioofSpitzer70(left)and24(right)µmto1.4GHzfluxdensity,q,asafunctionofredshift. ThefourIFRSsareshown asblackarrowsatz=0.06,aswehavelimitsonly. RedcirclesarexFLSgalaxies(Frayer etal.2006),whicharepredominatelystarforming galaxies. Blueboxesorarrowsshowratiosfromasampleofhighredshiftradiogalaxies(Seymour etal.2007),whichareradio-loudAGN. The black SED tracks are Arp220 (dotted), Mrk231 (dashed) and M82 (dot dashed). Blue solid lines show tracks from Chary&Elbaz (2001)withbolometricIRluminositiesrangingfrom108 to1013 L⊙.