Sources of the Radio Background Considered
aa r X i v : . [ a s t r o - ph . C O ] A ug Mon. Not. R. Astron. Soc. , 1– ?? (2010) Printed 2 November 2018 (MN L A TEX style file v2.2)
Sources of the Radio Background Considered
J. Singal ⋆ , L. Stawarz , , , A. Lawrence † , V. Petrosian ‡ Kavli Institute for Particle Astrophysics and Cosmology, SLAC National Accelerator Laboratory and Stanford University,382 Via Pueblo Mall, Stanford, CA 94305-4060, USA Institute of Space and Astronautical Science, Japan Aerospace Exploration Agency,3-1-1 Yoshinodai, Chuo-ku, Sagamihara, Kanagawa 252-5510, Japan Astronomical Observatory of the Jagiellonian University, ul. Orla 171, 30-244 Krak´ow, Poland University of Edinburgh Institute for Astronomy, Scottish Universities Physics Alliance, Royal Observatory,Blackford Hill, Edinburgh UK
In press version, 2010 August 16
ABSTRACT
We investigate different scenarios for the origin of the extragalactic radio background.The surface brightness of the background, as reported by the ARCADE 2 collabo-ration, is several times higher than that which would result from currently observedradio sources. We consider contributions to the background from diffuse synchrotronemission from clusters and the intergalactic medium, previously unrecognized fluxfrom low surface brightness regions of radio sources, and faint point sources belowthe flux limit of existing surveys. By examining radio source counts available in theliterature, we conclude that most of the radio background is produced by radio pointsources that dominate at sub µ Jy fluxes. We show that a truly diffuse backgroundproduced by electrons far from galaxes is ruled out because such energetic electronswould overproduce the obserevd X-ray/ γ -ray background through inverse Comptonscattering of the other photon fields. Unrecognized flux from low surface brightnessregions of extended radio sources, or moderate flux sources missed entirely by radiosource count surveys, cannot explain the bulk of the observed background, but maycontribute as much as 10 per cent. We consider both radio supernovae and radio quietquasars as candidate sources for the background, and show that both fail to produce itat the observed level because of insufficient number of objects and total flux, althoughradio quiet quasars contribute at the level of at least a few percent. We concludethat if the radio background is at the level reported, a majority of the total surfacebrightness would have to be produced by ordinary starforming galaxies above redshift1 characterized by an evolving radio far-infrared correlation, which changes towardthe radio loud with redshift. Key words: cosmology: diffuse radiation — radio continuum: general, ISM
Recent results from the ARCADE 2 (Absolute Radiome-ter for Cosmology, Astrophysics, and Diffuse Emission)project suggest a radio background several times brighterthan can be produced by currently observed radio sources(Fixsen et al. 2010; Seiffert et al. 2010; Singal et al. 2010).The extragalactic background, detected from 22 MHz to 8GHz, has a power law spectral index of α ≃ . S ν ∝ ν − α , where S ν is the spectral flux density) and ⋆ E-mail: [email protected] † Visiting physicist at KIPAC/SLAC National Accelerator Lab-oratory and Stanford University, USA ‡ Also Departments of Physics and Applied Physics, StanfordUniversity, Stanford, CA 94305, USA a brightness temperature of 1 .
17 K at 1 GHz, correspondingto a cosmic radio background (CRB) intensity of 7 × − W Hz − m − sr − = 3 . × Jy/sr. A significant contri-bution to the background from free-free emission has beenruled out based on the spectral shape (Seiffert et al. 2010).It is unlikely that the radio signal reported by the AR-CADE 2 collaboration is Galactic or local in origin. As dis-cussed in Kogut et al. (2010), the extragalactic componentis separated from the diffuse Galactic foreground by twoindependent robust indicators, a correlation of radio withCII emission and a cosecant dependence on Galactic lati-tude. The inferred residual extragalactic component is sev-eral times brighter than the high latitude Galactic level.Comparisons with observed edge-on galaxies indicate thatour Galaxy would be quite anomolous to support a radio c (cid:13) J. Singal et al. emitting halo of the intensity needed to explain the signal .Additionally, assuming a diffuse Galactic origin for the mea-sured radio signal would require a source of radio emissionthat does not follow the correlation with far-infrared emis-sion observed in local galaxies, again making our Galaxyanomalous. Isotropic radio emission from a local region onthe scale of the local bubble accounting for the level observedwould manifest an all-sky quadrupole polarization patternat a level visible in WMAP 23 GHz data, but such a pat-tern is not seen. These lines of evidence are summarized inKogut et al. (2010). Furthermore, the considerations of § µ Jy level.At high fluxes, say between 1 mJy and 1 Jy at 1.4 GHz, ra-dio loud active galactic nuclei (AGN) dominate radio sourcecounts (e.g., Condon 2007; Windhorst et al. 1993), withtheir differential counts following a power law dN/dS ∝ S − γ with γ < . . One may add here that this is true even takinginto account presence of ‘anomalous large-scale radiocontinuum features’ present in some edge-on spirals(Hummel, van Gorkom, & Kotanyi 1983; Elmouttie et al.1995). the measured diffuse X-ray/ γ -ray background through in-verse compton (IC) scattering of the cosmic microwave back-ground (CMB) and other background photon fields by thesame electrons. Additionally, the source spectra must beconsistent with the CRB power-law spectrum of index 0 . § § γ -ray background tolimit the contribution from more diffuse emission, clusters,and the intergalactic medium, and in § § § § § Figure 1 shows a schematic but fairly accurate depiction ofradio source counts available in the literature (see Table 1).In the top panel of the figure, we plot the S . dN/dS distri-butions given in different radio surveys (note that the differ-ential count is per steradian per unit flux). The observationswere conducted at a range of frequencies (0 . − . . .
75, the canonical value of the ex-tragalactic radio sources. Note that even though the surveysconsidered have a wide range of angular resolution (spanningfrom 1 to 300 arcseconds) there is not only a clear agreementamong the counts at
S >
S < S ( dN/dS ), divided by the observed CRB brightness asreported by the ARCADE collaboration in Fixsen et al.(2010), i.e. S ( dN/dS ) B CRB ( ν ) (cid:12)(cid:12)(cid:12)(cid:12) . ≡ S ( dN/dS )2 kν c − T CRB ( ν ) (cid:12)(cid:12)(cid:12)(cid:12) . , (1)where T CRB ( ν ) = 1 . × (cid:16) ν GHz (cid:17) − . K . (2)The true surface brightness B counts ( S ) = Z ∞ s S dNdS dS = 1( γ − S dNdS (3)for a power law dN/dS ≡ S − γ , is equal to the area under the S ( dN/dS ) curve, and thus the bottom pannel of Figure 1provides directly a minimum fractional contribution to theCRB from the resolved objects. We estimate this fractionto be ≃
26 per cent. We note that the high-flux populationis dominated by bright radio-loud AGN, while the low-flux c (cid:13) , 1– ?? ources of Radio Background - - - @ S (cid:144) Jy D l og @ H S . d N (cid:144) d S L (cid:144) J y . s r - D - - - - - - - - @ S (cid:144) Jy D l og @ H S d N (cid:144) d S L (cid:144) B CRB D Figure 1. TOP:
Scheamatic description of the observed S . dN/dS distribution from different radio surveys. Sourcecounts from 8 . . . . .
75 spectral index.
BOTTOM:
The S dN/dS distribution representing the surfacebrightness per logarithmic flux bin of the same surveys, dividedby the surface brightness of CRB at 1.4 GHz reported by the AR-CADE collaboration. The source counts tell roughly a consistentstory over a wide range of frequencies and resolutions. The peakicontribution at ∼
100 mJy comes from bright radio galaxies. Thecontribution decreases with decreasing flux density, but begins torise again below ∼ ∼
26 per cent of the total surface brightness of theCRB at 1 . one is generally thought to be dominated by starforminggalaxies (see Ballantyne 2009, and references therein), withan additional possible contribution from radio-quiet andlow-radio power AGN (e.g. Ibar et al. 2009; Padovani et al.2009). Surveys probing fluxes as low as a few µ Jy are con-sistent with the two population model (e.g., Fomalont et al.2002; Biggs & Ivison 2006).
The integrated (fractional) contribution of the high flux pop-ulation (
S > S ≃ R S dS S ( dN/dS ) /B CRB , is about ≃
16 per cent at 1 . N ( > S ) = R S dS ( dN/dS ) ≃ . × sr − . One couldask if a significant portion of the flux from the high flux pop-ulation has been missed by radio surveys. If this were thecase, then the contribution of the high flux population to themeasured background would have been underestimated. Onepossible source of the missing flux could be extended low sur-face brightness sources. However, as mentioned above, thesurveys included in Figure 1, in spite of the fact that theyspan a wide range of resolutions and are obtained from dif-ferent interferometer arrays, show very good agreement onthe integrated contribution to the background. We will re-turn to the contribution of low surface brightness sourcesand quantify the possible missed flux further in § Unlike the high-flux population, the total contribution ofthe low-flux population is not fixed by existing surveys, sincethese do not constrain the low flux peak on the S ( dN/dS ) − S plot. Even though at low fluxes ( S < S ) source countsindicate a power-law distribution dN/dS ∝ S − γ , the mea-sured faint end index γ varies substantially between differentsurveys, ranging from γ = 2 .
11 claimed by Fomalont et al.(2002) and γ = 2 .
61 claimed by Owen & Morrison (2008).The value of γ = 2 . dNdS = k (cid:16) SS (cid:17) − γ for S min < S < S , (4)one can calculate the required value of S min as a function of γ such that the sources in this range provide the rest of theCBR. It is easy to show that for γ > S min = S (cid:20) (1 − H ) ( γ − B CRB k S + 1 (cid:21) / (2 − γ ) , (5)where the factor H is the estimated fractional contributionfrom sources above S ≃ .
16. The top panel of Figure 2 shows the variation of S min with γ at ν = 1 . N ( > S min ) − N ( > S ) = k S − γ γ − (cid:20)(cid:16) S min S (cid:17) − γ − (cid:21) , (6)whose dependence on γ is shown on the bottom panel ofFigure 2. Thus, if the background is to be made primarilyfrom low flux sources, their faint end index below 1 mJyshould be close to γ = 2 . µ Jy and have asurface density of ( & sr − ), but not much higher thanthis value.A modeling of the dN/dS distribution of the low-flux c (cid:13) , 1–, 1–
16. The top panel of Figure 2 shows the variation of S min with γ at ν = 1 . N ( > S min ) − N ( > S ) = k S − γ γ − (cid:20)(cid:16) S min S (cid:17) − γ − (cid:21) , (6)whose dependence on γ is shown on the bottom panel ofFigure 2. Thus, if the background is to be made primarilyfrom low flux sources, their faint end index below 1 mJyshould be close to γ = 2 . µ Jy and have asurface density of ( & sr − ), but not much higher thanthis value.A modeling of the dN/dS distribution of the low-flux c (cid:13) , 1–, 1– ?? J. Singal et al.
Figure 2. TOP:
The lower flux density of sources S min at 1.4GHz required so that the extrapolated source counts from S ≃ γ , can account for observed CRBas reported by the ARCADE 2 collaboration, including a 16 percent (solid lines) or 25 per cent (dashed line; see §
4) contributionto the background from S > S sources. The upper and lower solidcurves are for the two extreme values of the 1 mJy normalizationsn of the surveys we consider. For γ ∼ S min ∼ . µ Jy is needed, while the needed lower flux densityfalls rapidly with decreasing faint end index.
BOTTOM:
Sameas the top panel, but for the minimum number of sources belowS needed to produce the observed CRB. ( < mJy) population in terms of a single power law, althoughcustomary, is more than likely an over-simplification. A morerealistic description would involve a smoothly curved func-tion with the integrated contribution per log flux bin (Figure1, lower panel) peaking around S min and then falling off, butincluding some contribution from objects with fluxes lowerthan S min . The large population of dwarf galaxies may bethe sources that increasingly become relevant at fluxes below S min . In this section we consider the possibility that the CRB re-sults from truly diffuse emission associated with the largescale structure of the Universe, such as the intracluster medium (ICM), intergalactic medium (IGM), or the fil-aments connecting the clusters containing warm-hot gas.First we consider very general constraints, and then lookat specific possibilities.
The simplest constraint on any population of electrons pro-ducing the CRB is that it must have a relatively flat energyspectrum, in order to produce the observed spectral indexof α ∼ .
6. Less obviously, we show below that it must beassociated with a magnetic field of at least 1 µ G. This isbecause otherwise the magnetic field energy density in suchsystems is much less than the energy density of the cosmicmicrowave background (CMB) and other background radi-ations. As a result, the relativistic electrons responsible forthe radio emission would lose most of their energy produc-ing hard X-ray and gamma radiation via inverse Comptonscattering of the other background fields. We now derive thislimit more carefully.Given the observed power law spectrum, the energydensity of the radio background per frequency dex at ν r is given by[ ν r U ν r ] = 4 πc [ ν r B CRB ( ν r )] = 1 .
17 8 πk B ν ⋆ c (cid:16) ν r ν ⋆ (cid:17) . , (7)where ν ⋆ = 1 GHz. We are considering the background span-ning frequencies from ν ∼ ν ∼
10 GHz.For a density of ultrarelativistic electrons with a power-law energy spectrum n e ( γ e ) = k e γ − se for γ e γ e γ e , (8)where k e is a normalization constant in units of cm − , and γ e and γ e correspond here to the Lorentz factors of theelectrons producing radiation primarily around ν and ν ,respectively, the synchrotron emissivity may be approxi-mated as (see e.g. Rybicki & Lightman 1979)[ ν r j ν r ] ≃ cσ T U B π k e (cid:16) ν r ν cr (cid:17) − s , (9)where U B ≡ B / π is the energy density of the magneticfield, and ν cr ≡ (3 eB/ πmc ) γ e ≃ . B/µ G) γ e Hz is thecritical (radio) synchrotron frequency for a given γ e .For production of the observed radio background weneed s = 2 . γ e = 5 × ( B/µ G) − / , and γ e =5 × ( B/µ G) − / , with the value of k e being determinedfrom the following relation between the emissivity and theobserved energy density.We relate the (radio) synchrotron energy density to theemissivity with[ ν r U ν r ] = 4 πc Z dVdz dz πd L ( z ) [˜ ν r j ˜ ν r ] == [ ν r j ν r ] 4 πH Z F syn ( z ) dz (1 + z ) ( s +1) / E ( z ) , (10)where ˜ ν r = ν r (1 + z ), H = 70 km − Mpc − is the Hubbleconstant, and F syn ( z ) describes the evolution of the product U B × k e . Here E ( z ) ≡ p Ω M (1 + z ) + Ω Λ for the assumedflat cosmology, and the comoving volume element is dVdz ≡ c πH E ( z ) (cid:20)Z z dz ′ E ( z ′ ) (cid:21) = 4 πc d L H (1 + z ) E ( z ) . (11) c (cid:13) , 1– ?? ources of Radio Background Similarly, we can get the inverse Compton (IC) emis-sivity resulting from the same population of electrons as [ ν ic j ν ic ] = 3 chσ T π Z dν Z dγ e n ( ν ) n e ( γ e ) ν ic f T γ e ν , (12)(Blumenthal & Gould 1970), where n ( ν ) ≡ [ ν U ν ] tot /hν is the total spectral number density of the extragalacticbackground photon field (including radio, CMB, infrared,optical/UV, X-ray, etc), f T = 2 q T ln q T + q T + 1 − q T ,q T ≡ ν ν γ e and 14 γ e q T . (13)The IC emissivity can be related to the IC energy den-sity by an equation similar to equation 10 with F syn ( z ) re-placed by F ic ( z ) which now describes the evolution of theproduct n ( ν ) × k e . Ignoring the differences between thesetwo evolutions (i.e. assuming that the ratio of the integralsover redshift involving F syn ( z ) and F ic ( z ) is of order unity)and by eliminating k e we can express the IC energy densityin terms of the synchrotron energy density as[ ν ic U ν ic ] ≃ [ ν r U ν r ] B / π (cid:18) . B/µ G) ν ⋆ (cid:19) . ν ×× Z γ e ( B ) γ e ( B ) dγ e Z ν max ν min dν ν − γ − . e [ ν U ν ] tot f T , (14)with ν max = min[ ν , ν, m e c / hγ e ] and ν min =max[ ν , ν/ γ e ]. We note that because the IC emissivity isdominated by upscattering of the CMB photon field, forwhich the energy density increases with redshift, the pre-sented evaluation of [ ν ic U ν ic ] with the cosmological evolu-tion neglected corresponds strictly to a lower limit.Figure 3 shows the energy density of the observed ex-tragalactic background light (thick curve), and the expectedIC energy density resulting from upscattering of these back-ground photons by electrons producing the radio back-ground as given by equation 14, for different magnetic fields B = 0 . .
01, 0 .
1, and 1 µ G (dotted, dashed, dot-dashed,and solid curves, respectively). Spectral energy densities ofthe IR/optical, X-ray, and γ -ray cosmic photon fields wereconstructed to be in agreement with the background lev-els provided by Franceschini et al. (2008), Gilli et al. (2007),and Sreekumar et al. (1998), respectively. Clearly any mag-netic field weaker than 1 µ G would result in X-ray/ γ -rayemission exceeding the observed background, and regionswith such magnetic fields may be excluded as significantsources of the CRB.Importantly, this consideration excludes our own Galac-tic halo as the origin of the bulk of the isotropic radio signal.Taylor, Stil, & Sunstrum. (2009) use rotation measure mea-surements of 37,000 polarized extragalactic radio sources todetermine the intensity of the magnetic field in the Galactichalo, concluding with a value of approximately 1 µ G. In our The following expressions are valid in the Thompson regime andare a good approximation for scattering of photons with hν ∼ γ . For scattering of photonsabove this energy one must use the Klein Nishina cross section.Few relevant photons lie above this range so in what follows weapproximate the Klein-Nishina suppression by a sharp cutoff.
10 15 20 25 - - - - - -
10 log Ν @ Hz D l og Ν U Ν @ e r g c m - D Figure 3.
The thick black curve shows the measured energydensity of the radio, microwave, infrared, optical, ultraviolet,X-ray, and γ -ray extragalactic backgrounds. The other curvesshow the energy density produced by inverse Compton scatter-ing of the photon backgrounds by electrons necessary to producethe radio background reported by the ARCADE 2 collabora-tion via synchrotron emission. The dotted, dashed, dot-dashedand solid curves are for a 1 nG, 10 nG, l
00 nG, and 1 µ G levelaverage magnetic field. Because the intergalactic magnetic fieldis known to be µ G, the observed level of the X-ray back-ground rules out a significant portion of the radio backgroundbeing produced by electrons far from galaxies. Spectral energydensities of the IR/optical, X-ray, and γ -ray cosmic photon fieldswere constructed to be in agreement with the background lev-els provided by Franceschini et al. (2008), Gilli et al. (2007), andSreekumar et al. (1998), respectively. Galactic halo, the level of the ambient optical and infraredphoton fields will be even higher than that considered inthe calculation here, by an amount depending on the dis-tance from the Galactic plane, predicting an X-ray back-ground many times larger than that observed. There areGalactic and solar system components to the observed dif-fuse X-ray background, but these are significant only below 1keV (Hickox & Markevitch 2006). The observed level of theX-ray background therefore strongly disfavours a Galacticorigin for the observed isotropic radio signal.
The CRB could in principle result from a population of rela-tivistic electrons pervading the IGM as as a whole, perhapsresulting from many generations of AGN. Large scale radiosources can easily expand to sizes of many Mpc on gigayeartimescales, and so overlap. Almost certainly, however, adi-abatic expansion losses would result in a very low energydensity, and synchrotron losses would lead to a steep en-ergy spectrum. Furthermore a variety of arguments suggestthat the magnetic field in the IGM is likely to be very weak, B . µ G (see Vall´ee (2004) and references therein). Dif-fuse emission from the IGM therefore seems unlikely to bethe solution.
There is evidence for diffuse radio emission in some butnot all clusters of galaxies (e.g. Feretti & Giovannini 2008). c (cid:13) , 1– ?? J. Singal et al.
Ever since discovery of this emission from the Coma cluster(Large et al. 1959), there have been improved observationsof this system (Wilson 1970; Schlickeiser et al. 1987), as wellas more extensive surveys by Giovannini & Feretti (2000)who have identified more than 40 clusters with diffuse ra-dio emission. These are classified as halos or relics, and areoften associated with dynamically disturbed (merging) clus-ters. The origin of the relativistic electrons is controversial.They may be injected by AGN jets (McNamara & Nulsen2007, and references therein) or produced as secondarypairs from ultrarelativistic cosmic-ray hadrons scattering ofICM protons. Finally, non-thermal electrons within IGMmay be also picked up directly from the thermal pool andaccelerated to ultrarelativistic energies by ICM shocks orturbulence (‘primary’ electrons; see e.g., Petrosian 2001;Brunetti & Lazarian 2006). We note that the most recentobservations of few clusters at γ -ray energies seem to ex-clude the ‘secondary’ nature of the radio halo electrons (e.g.,Aleksi´c et al. 2010).The fraction of clusters with radio emission increasesrapidly with increasing soft X-ray luminosity, indicating that strong diffuse radio emission is not a common property ofall clusters. However, deeper radio observations, especiallyat low frequencies, suggest that the presence of extended,low surface-brightness radio structures in galaxy clusters isrelatively common (e.g. Rudnick & Lemmerman 2009) butat lower flux levels. Possibly then the integrated effect ofa large number of such weakly emitting clusters could bea significant contribution to the CRB. We are not able topredict the level of this integrated emission, but we can checkthe cluster hypothesis against our two general constraints –magnetic field and spectral index.There are estimates of the magnetic fields in someclusters via Faraday rotation measurements, which indicate(line-of-sight averaged) magnetic fields of 1 to 10 µ G (e.g.,Kim & Kronberg 1990; Taylor et al. 2002). In particular, inComa, Kronberg et al. (2007) measure a line-of-sight aver-aged intergalactic magnetic field in the range 0 . − . µ G.However, the actual magnetic field will be larger by a factorof (
R/H B ) / if it is tangled or chaotic on a distance scale H B ) that is smaller than cluster size R . For H B ∼
10 kpc onecan get fields as high as few µ G. Overall, it is quite plausiblethat some cluster emission passes the magnetic field test.However, the observed radio spectra are relatively steep(index α >
1; see Liang et al. (2000)) which does notmatch the index of the CRB. It therefore seems that dif-fuse ICM/IGM emission, while possibly widespread, is nota dominant contributor to the cosmic radio background.
It is possible that the calculated strength of the CRB basedon source counts is underestimated in two ways, due to thesurface brightness limits of surveys. Firstly, the fluxes ofsome extended objects may have been undercharacterized,if their sizes are large, or they have extended low surfacebrightness components. Secondly, if there is a wide range ofsource surface brightnesses extending to low values, theremay exist sources which will not be detected. In order to discuss this issue in more detail, it is useful to consider highand low flux sources separately.
At high fluxes, source counts are dominated by radio galax-ies and radio-loud quasars that very often have large scalemorphology. It is useful to distinguish the two well knownFanaroff-Riley morphological types (Fanaroff & Riley 1974).The low-power FR Is have surface brightness profiles whichfade outwards with no clear end, and therefore part of theirtotal flux will be undetected in interferometric radio surveys.However, these outer components typically have steep spec-tra (e.g., Hardcastle 1999; Liang et al. 2006, 2008), whichis inconsistent with being the dominant component of theCRB. Moreover, they have only a modest cosmological evo-lution (Willott et al. 2001; Jamrozy 2004; Rigby et al. 2008;Smolˇci´c et al. 2009) and so although significant at sub-mJyfluxes (see Rigby et al. 2008; Padovani et al. 2009), are un-likely to be the dominant low-flux population. ‘Classical doubles’ (FR IIs), on the other hand, haveclearly defined boundaries of their extended radio lobes, butare expected to decrease in surface brightness as they ageand grow in size (Kaiser et al. 1997). Such extended struc-tures, which can reach Mpc-scale sizes, may to fall belowtypical survey surface brightness detection limits, especiallyat high redshifts. Even the cores may remain undetected,as they often have very low levels of central (AGN-like)activity (Machalski et al. 2001, 2006; Dwarakanath & Kale2009). However, the known low-surface brightness giant ra-dio galaxies tend to have steep spectra (see Jamrozy et al.2008; Konar et al. 2008), which is not consistent with thatof the CRB.A new survey, the Australia Telescope Low BrightnessSurvey (Subrahmanyan et al. 2010, 2009) has addressed thequestion of the number of potentially missed low-surfacebrightness radio galaxies or their extended components, us-ing an 8 . survey down to mJy fluxes, with a surfacebrightness threshold a factor of 5 lower than previous compa-rable surveys. Subrahmanyan et al. (2010) state that 30 percent of their sources have at least half of their flux between5 arcsec and 30 arcsec, and that 10 per cent of their sourceshave a size at least 1.5 times their beamsize of 50 arcsec, in-dicating that there is a significant, but not dominant, com-ponent of extended flux. This survey discovered a few newgiant radio galaxies. The authors have not yet publishedsource counts, but their initial analysis finds 500 sources in8 . to a completeness limit of 1 mJy at 1 . . × sr − .This is not very different from our estimate based on othersurveys discussed in § This conclusion is in agreement with the modest estimated con-tribution of FR I sources to the extragalactic X-ray and γ -raybackgrounds (Celotti & Fabian 2004; Stawarz et al. 2006, respec-tively) c (cid:13) , 1– ?? ources of Radio Background this is a very large effect; at most a 50 per cent correction,which can increase the maximum integrated contribution ofthe high flux population to the CRB to ∼
25 per cent.
Below 1 mJy, optical spectra of identified sources indi-cate that source counts become dominated by star form-ing galaxies (e.g., Benn et al. 1993) and the characteristicradio source size changes from ∼
10 arcsec to ∼ f ∼ . z = 1) would instead re-quire f ∼ .
7. This effect is therefore unlikely to be verylarge for the star forming galaxy population.One could imagine a population of large and low surfacebrightness radio-emitting galaxies whose fluxes are badlyunderestimated. Then for median source sizes of 10 and50 arcsec, the correction factors would be f ∼ . f ∼ .
9, respectively, the later being nearly sufficient toexplain the observed CRB. However, a completely missingpopulation at S ∼ . µ Jy levels which had previously beensurveyed at the same frequency with a beam size of 19 arcsecby Condon & Mitchell (1984). Of 159 sources found in thefirst (large beam) survey, only 8 were not seen in the later(small beam) survey.Summarizing the situation for low flux sources, it seemsunlikely that significant extended flux has been missed (i.e.not already corrected for).
The above results indicate that a class of radio sources notpreviously considered is needed to produce the bulk of theradio background. In particular, the background cannot beformed from diffuse emission or relatively few high lumi-nosity sources, but rather many lower luminosity sources.As discussed in § ∼ µ Jylimit of current radio surveys down to the ∼ − µ Jy level.In addition, the sources should have high magnetic fields,
B > µ G, and should be produced by a relatively flat-spectrum of relativistic electrons. In this section we considersome possible candidates.
Radio supernovae (RSNe)produce significant radio flux andcould be a contributor to the background. RSNe associ-ated with Type II supernovae have a mean spectral in-dex roughly comparable to the observed CRB (Weiler et al.1986). Colina et al. (2001) discovered a RSN in a radio mon-itoring campaign of the galaxy NGC 7469 with a 8 . ≃ . × W Hz − , which is more than a thou-sand times more luminous than Cassiopeia A, the brightestradio supernova remnant in the Milky Way, and which wouldcorrespond to flux densities in the range µ Jy < S < mJy be-tween z = 0 . z = 2. This object would not have beenfound in conventional optical searches.Are RSNe of this luminosity numerous enough to ac-count for the background? As discussed in Weiler et al.(2004) significant radio emission is only seen in core collapsesupernovae, and approximately 10 per cent of core collapsesupernovae exhibit radio loudness. Madau et al. (1998) es-timate the number of core collapse supernovae integratingacross all redshifts to be 3 arcmin − yr − . Mannucci et al.(2007) update this with the number of supernovae obscuredoptically by starbursts, limiting the correction to a factorof two at most, giving × yr − core collapse super-novae over the whole sky. With the period of radio loudnesslasting on the order of a year (Weiler et al. 2004), we haveonly × RSNe on the sky at any time. Noting that z = 2 is the peak era of supernova activity and assuminga mean RSNe 8.4 GHz luminosity of 1 × W Hz − , thiswould give a contribution to the background intensity at thelevel of .
10 Jy/sr, falling short of the observed CRB by afactor of > × W Hz − , the contribution will be even less.Note also that a fraction of the RSNe population at this lu-minosity located at significantly lower redshifts would havesufficiently high flux to be included in the radio surveys con-sidered in § − arcmin − yr − , which is much less than regularcore collapse supernovae.Finally, if RSNe were significant contributors to the ra-dio fluxes of spiral galaxies, they would be seen as compactsources in high-resolution images of nearby spiral galaxies.However, they are not. Young RSNe can sometimes be seenas individual sources, but they contribute only a tiny frac-tion of the total flux. We conclude that RSNe are not asignificant component of the radio background. Radio quiet (RQ) quasars, as is well known, are not radiosilent, and thus could potentially account for a missing pop-ulation of CRB sources. They are typically unresolved onsub arcsecond scales, with 1 . − erg s − (Blundell & Kuncic 2007; White et al.2007; Elvis et al. 1994). The radio emission of higher lumi- c (cid:13) , 1– ?? J. Singal et al. nosity RQ quasars is most often produced by mildly/non-relativistic electrons in nuclear outflows originating in theinner parts of accretion disks, though large-scale relativis-tic jets are sometimes (surprisingly) detected in such sys-tems (Blundell et al. 2003). The expected strong magneti-zation of disk winds ( B ≫ µ G) requires low Lorentz fac-tors and low energies of radio-emitting electrons which re-move the IC problem discussed in §
3. We note that the ra-dio emission from most low-luminosity ( < W/Hz) RQquasars may not originate in nuclear outflows, but ratherfrom star-forming host galaxies (Kellerman et al. 1994), buttheir space density is much lower than that of comparablyluminous sources in normal starforming galaxies. The ob-served radio continua of RQ quasars are flat, possibly inagreement with the CRB spectrum (Blundell, priv.com.).RQ quasars are also numerous, though underrepresented inexisting radio surveys due to the flux limits of the surveys.In addition, White et al. (2007) have demonstrated thatstacking seemingly empty radio images of optically identi-fied quasars which are individually below the noise level ofthe radio survey leads to a composite image of significantradio flux. Finally, Simpson et al. (2006), Ibar et al. (2009)and Padovani et al. (2009) have claimed recently that radioquiet AGN constitute a significant fraction of the sub-mJyradio source population.On the other hand, the idea that RQ quasars producethe bulk of the CRB presents several problems. From Fig-ure 2, even with a faint end index of 2 . µ Jy, thenumber of sources needed to make the background is per-haps only a factor of 10 lower than reasonable estimatesof the total number of non-dwarf galaxies in the observableUniverse, and as such is much higher than the expected num-ber of quasars, i.e. high-accretion rate objects. Also, opticalquasar number counts (e.g. Richards et al. 2006) point to apeak in the contribution to S dN/dS occurring at ∼ µ Jyin i band. Converting this to a radio flux using a radio loud-ness parameter (the ratio of the 1.4 GHz radio to 2500 ˚Aoptical luminosity) on the order of 1 for RQ quasars, andadjusting the optical flux between i band and 2500 ˚A ac-cording to standard spectral models, indicates the peak of S dN/d log( S ) at 1 . µ Jy. Below thispeak flux the faint end index would drop to a sub-Euclideanvalue. This turn over would occur at least one order of mag-nitude higher than that necessitated by Figure 2.Let us now estimate the expected contribution of RQquasars to the measured CRB. Knowing the radio lumi-nosity function (LF) of RQ quasars, ψ ( L R , z ), where L R ≡ [ ν r L ν r ] is the radio luminosity at frequency ν r (in what fol-lows we will use ν r = 5 GHz), their contribution to the CRBenergy density can be evaluated in a manner equivalent tothat of equation (10) with the emissivity replaced by the LFmultiplied by L or L R ψ ( L R , z ), namely[ ν r U ν r ] = 1 c Z dz Z dL R dVdz ψ ( L R , z ) L R (1 + z ) α − π d L == Z dz Z dL R ψ ( L R , z ) L R H (1 + z ) α E ( z ) , (15)where α = d ln L ν r /d ln ν r is the radio spectral index.We do not have direct knowledge of the radio LF but wecan estimate as described below by relating it to the knownoptical bolometric LF of quasars ψ ( L bol , z ) which may evolve as a pure luminosity evolution, pure density evolution, orluminosity-dependent density evolution as constrained byHopkins et al. (2007). Assuming similar evolution for theradio LF, then by definition, ψ ( L R , z ) dL R = ψ ( L bol , z ) dL bol . (16)To evaluate the above integral we need the relation be-tween the bolometric and radio luminosities. White et al.(2007) give a relation between the radio luminosity at 5 GHzand the absolute 2500 ˚A magnitude M UV , which can bewritten as log (cid:2) L / (erg s − Hz − ) (cid:3) ≃ . − . M UV or log (cid:2) L R / (erg s − ) (cid:3) = 31 . − . M UV . We convertthe 2500 ˚A magnitude to the bolometric quasar luminos-ity using the approximate relations L bol ≃ × ν UV L ν UV , m UV − M UV = 5 log[ d L / pc] − K , where K stands for the K -correction factor, and m UV = − . S UV / (cid:18) L R erg s − (cid:19) ≃ . × (cid:18) L bol erg s − (cid:19) . , (17)where the luminosities L bol , ν UV Lν UV , and L R are ex-pressed in the units of erg s − .After integrating equation 15 over the redshift range z = 0 − L R = 10 − erg s − (corresponding to the bolometric luminosityrange L ≃ − erg s − ), for the fit parametersregarding the quasar luminosity function as provided inHopkins et al. (2007), we find the expected contribution ofRQ quasars to the extragalactic radio background at 5 GHzto be at the level of ≃ While some high accretion rate objects may be present beforez=6, the number must necessarliy fall far short of the number ofobjects needed to make the background outlined in § (cid:13) , 1– ?? ources of Radio Background Given the large ( & sr − ) number of sources needed toproduce the CRB, the constraints on extended and non-galactic emission discussed in sections 4 and 3, the requiredstrong magnetic fields ( B > . µ G) and flat spectra, andthe shortcomings of radio supernovae and RQ quasars ( § § α = 0 . B & µ G instar forming regions circumvents the IC problem discussedin § The radio emission originating in local star forming regionsis observed to be correlated with the far-infrared (FIR) flux(e.g. Dwek & Barker 2002; Ibar et al. 2008, and referencestherein), which combined with the observed level of theinfrared background, constrains the contribution of thesesources to the CRB. Dwek & Barker (2002) give a relationbetween the 1.4 GHz radio power P . and the FIR lumi-nosity, L F IR , in the wavelength range 10 − µ m, whichcan be written as L r ≡ [ νP ( ν )] . ∼ × − L F IR . There-fore, if the radio background is produced in the star formingregions of Seyferts, spirals, and ULIRGs, then from a naiveapplication of this ratio one should expect the observed FIRbackground at the level of[ νU ν ] F IR ≃ [ νU ν ] . ≃ × − erg cm − , which is higher than that observed (Marsden et al. 2010) bya factor of >
10. This is in agreement with Dwek & Barker(2002) result showing that the expected surface brightnessof the sky at 178 MHz for different models of the cosmo-logically evolving star formation rate to be between 3 Kand 30 K; about 3 to 30 per cent of the ARCADE result T CRB (178 MHz) ≃
104 K (Fixsen et al. 2010).An alternative way to formulate this problem is as fol-lows. Appleton et al. (2004) find that individual starforminggalaxies locally show a ratio between FIR and radio fluxesgiven by the value q ≡ log( S µ m /S . ) = 2 .
15. If wetake the observed CRB and assume it is made by star form-ing galaxies following this ratio, then, ignoring for now anyK-correction issues, we predict a FIR surface brightness at70 µ m of 178 nW m − sr − , which is about 25 times the ob-served level reported in Dole et al. (2006). Thus we can con-clude that the contribution of systems that obey the localradio FIR correlation to the radio background is on the levelof approximately . Since star formation in the Universe has evolved and therate was much higher at redshifts of 1 and above (e.g., Madau et al. 1998), the bulk of both the radio and infraredbackgrounds are produced at these redshifts. Therefore, anevolution in the observed FIR to radio flux ratio towardsgreater radio loudness with redshift is necessary to have theradio and FIR backgrounds produced at the observed levelsby starforming galaxies.It is important to note that the ratio of the relative con-tribution to the FIR background and radio background sur-face brightnesses from a galaxy (or class of galaxies) locatedat a particular redshift is given by the non-K-corrected FIRto radio flux ratio, rather than the K-corrected ratio that isoften reported for higher redshifts. Relating an observed fluxratio or ratio evolution to a K-corrected one, or vice-versa, isdependent on the source SED assumed, and the wavelengthsin question. For observations at 70 µ m and 1 . q (where q x is the log of the flux ratio at infrared frequency x to 1.4 GHzradio) from a galaxy at z ∼ q ∼ . µ m backgrounds shown in § q (observed) ∼ − . q (intrinsic) ∼ − . q with redshift at either MIR, FIR or submm wave-lengths. These determinations are complicated by selec-tion effects, and in the case of reports of the K-correctedcorrelation, template spectra. Although some papers con-clude that there is no evidence for a change in intrinsic q (eg Ibar et al. (2008), Sargent et al. (2010), Ibar et al.(2008)) others do find possible evolution in intrinsic q (egVlahakis et al. (2007), Beswick et al. (2008), Seymour et al.(2009), Bourne et al. (2010)), although it is not clear thatthis is large enough to explain the radio background. Recentdata from the BLAST instrument (Ivison et al. 2010a) alsoshow the intrinsic correlation evolving, and new results fromthe Herschel instrument (Ivison et al. 2010b) are consistentwith this as well. The most suitable reported results for com-parison are those of Sargent et al. (2010) and Bourne et al.(2010) who use Spitzer 70 µ m data and present the evolutionin observed q versus redshift (Fig 12 in Sargent et al, Fig 9in Bourne et al). These figures agree in showing that by z ∼
2a change of ∆ q (observed) ∼ − . q (observed) ∼ − .
4. We note again that determina-tions of the value of the correlation at higher redshifts arecomplicated by selection biases and contamination by AGNemission.An evolving radio to FIR luminosity ratio toward theradio loud with increasing redshift would suggest that one ormore of the following is true at higher redshifts: 1) a largerportion of the energy of star formation goes into in relativis-tic particles, for some reason possibly related to a differentstructure of the interstellar medium shaping the cosmic-rayacceleration efficiency at shocks driven by supernova, 2) alarger portion of the stars are high mass, resulting in moresupernovae as well as an enhanced cosmic-ray accelerationat shocks driven by winds from massive stars, 3) synchrotron c (cid:13) , 1– ?? J. Singal et al. emissivity is enhanced by higher interstellar magnetic fields,or 4) AGN activity in ordinary galaxies is proportionallymore important.There is no concrete information on item 1, while item2 requires evolution of the initial mass function, which iftrue would alter the interpretation of the data on the starformation rate. The third possibility can be neither rejectednor supported by current data.Here let us only comment on the possibility thatthe cosmological evolution of supermassive black holes inthe galactic centers results in enhanced AGN activity inlate-type galaxies at high redshifts, and thus in their en-hanced radio emission relative to infrared. Note that inthis context that it is now established that a large frac-tion of nearby spiral galaxies show weak AGN activ-ity (Ho, Filippenko, & Sargent 1997; Ho 2008). In addi-tion, several authors have argued that supermassive blackholes (SMBHs) were spinning more rapidly at early epochs(Wang et al. 2009). For spiral-hosted SMBHs the effect maybe even stronger than for the elliptical hosted SMBHsthat give rise to quasars, due to the different charac-ter of the dominant accretion events which determine theblack hole spin evolution (see the discussion in Sikora et al.2007; Volonteri et al. 2007). In the framework of the spinparadigm for jet production, critically re-examined recentlyby Sikora et al. (2007), one could therefore expect morepowerful jets, and therefore more AGN-related radio emis-sion (for the same accretion luminosity) at higher redshiftsin these systems.If the radio background is indeed formed from ∼ . µ Jysources, that corresponds to about ∼ W Hz − radio(1 . z = 2, which would be between10 and 10 L/L ⊙ at 70 µ m with an evolving (K-corected) q . This infrared luminosity is below the ’knee’ of the in-frared luminosity function, i.e. the radio background wouldbe made by relatively normal spiral galaxies, in contrast tothe infrared background, which appears to be dominatedby luminous and ultraluminous IR galaxies (Magnelli et al.2009). We note the difficulty that locally the peak contri-bution to radio emission is from relatively luminous galax-ies with 1.4 GHz radio luminosities of around 10 W Hz − (Condon, Cotton, & Broderick 2002), while the peak contri-bution to the radio background, from galaxies beyond red-shift 1, is at 1.4 GHz luminosities of around 10 W Hz − . We have considered several mechanisms as the origin of the’missing’ radio flux necessary to account for the radio back-ground, assuming it is at the level reported by the ARCADE2 collaboration. As shown in §
3, diffuse, low surface bright-ness synchrotron emisssion from large scale structures (IGM,ICM and WHIM), and from our own Galactic halo, is limitedby the observed level of the X-ray/ γ -ray background. As dis-cussed in § § ∼ µ Jy contribute another ∼
10 per cent, leaving between 60 per cent and 75 per cent of the cosmicradio background unaccounted for.We find it difficult to explain the level of the radio back-ground reported by ARCADE 2 without a new populationof low flux radio sources. These sources should be numer-ous and faint enough to dominate source counts below the ∼ µ Jy limit of current radio surveys and must extend tothe ∼ − µ Jy (at 1.4 GHz) level. Moreover, they shouldhave an observed ratio of radio to infrared output a factorof 5 above what is observed in local galaxies. As discussedin § § z > ACKNOWLEDGMENTS
We thank Philip Best, Katherine Blundell, Chi C. Chueng,Sebastian Jester, Dan Marrone, Matthew Turk, and TomAbel for their input. JS thanks the ARCADE team, S.Kahn, and R. Schindler for their encouragement. LS wassupported by the Polish Ministry of Science and Higher Ed-ucation through the project N N203 380336, and also by thescandinavian NORDITA program on ‘Physics of RelativisticFlows’.
REFERENCES
Aleksi´c J., et al., 2010, ApJ, 710, 634Appleton P. et al., 2004, ApJS, 154, 147Ballantyne D.R., 2009, preprint (arXiv:0904.0996)Benn C., Rowan-Robinson M., McMahon R., BroadhurstT., Lawrence A., 1993, MNRAS, 263, 98Beswick R., Muxlow T., Thrall, H. Richards A., GarringtonS., 2008, MNRAS, 385, 1143Biggs A., Ivison R., 2006, MNRAS, 371, 963Blumenthal G.R., Gould R. J., 1970, Reviews of ModernPhysics, 42, 237Blundell K. M., 2003, New Astronomy Review, 47, 593Blundell K., Kuncic Z., 2007, ApJ, 668, 103Blundell K.M., Beasley A. J., Bicknell G. V., 2003, ApJL,591, L103Bondi M. et al., 2003, A&A, 403, 857Bondi M. et al., 2007, A&A, 463, 519Bourne L., Dunne L., Ivison R., Maddox S., Dickin-son M., Frayer D., 2010, MNRAS, submitted, preprint(arXiv:1005.3115)Brunetti G., Lazarian A., 2006, IAUJD, 1, 13Celotti A., Fabian A.C., 2004, MNRAS, 353, 523Ciliegi P., Zamorani G., Hasinger G., Legmann I, SzokolyG., Wilson G., 2003, A&A, 398, 801Coleman P., Condon J., 1985, AJ, 90, 1431Colina L., Alberdi A., Torrelles J., Panagia N., Wilson S.,2001, ApJL, 553, L19Condon J., Cotton W., Broderick J., 2002, AJ, 124, 675 c (cid:13) , 1– ?? ources of Radio Background Condon J., 2007, ASP conference series, volume 80, eds:Afonso, J. Ferguson, C. Mobasher, & B. Norris, R. p. 189Condon J., Mitchell K., 1984, AJ, 87, 1429Dole H., et al. 2006, A&A, 451, 417Dwarakanath K.S., Kale R., 2009, ApJL, 698, 163Dwek E., Barker M., 2002, ApJ, 575, 7DElmouttie E., Haynes R. F., Jones K. L., Ehle M., Beck R.,Wielebinski R., 1995, MNRAS, 275, L53Elvis M. et al., 1994, APJS, 95, 1Fanaroff B.L., Riley J.M., 1974, MNRAS, 167, 31PFeretti L., Giovannini G. 2008, in Pilonis, M., Lopez-Cruz,O., Hughes, D., eds, Lecture Notes in Physics vol. 470,A Pan-Chromatic View of Clusters of Galaxies and theLarge-Scale Structure, Springer-Verlag, Berlin, p. 143Fixsen D. et al., 2010, ApJ, submitted, preprint(arXiv:0901.0559)Fomalont E., Kellerman K., Partridge R., Windhorst R.,Richards E., 2002, ApJ, 123, 2402Fomalont E., et al., 2006, ApJS, 167, 103Franceschini A., Rodighiero G., Vaccari M., 2008, a&A,487, 837Gervasi M., Tartari A., Zannoni M., Boella G., Sironi G.,2008, ApJ, 682, 223Gilli R., Comastri A., Hasinger G., 2007, A&A, 463, 79Giovannini G., Feretti L., 2000, New Astronomy Review,5, 335Gruppioni C., Zamorani G, de Ruiter H., Parma P., MignoliM, Lari C., 1997, MNRAS, 286, 470Grueff G., 1988, A&A, 193, 40Hales S., Baldwin J., Warner P., 1988, MNRAS, 234, 919Hardcastle M., 1999, A&A, 349, 381Hickox R., Markevitch M., 2006, ApJ, 645, 95Ho L., 2008, ARAA, 46, 475Ho L., Filippenko A., Sargent W., 1997, Proceedings of theIAU colloquium No. 159, eds. Peterson, B. Cheng, F.- Z.and Wilson, A. p. 429Hopkins A., Alfonso J., Chan B., Cram L., Georgakakis A.,Mobasher R., 2003, AJ, 125, 465Hopkins P.F., Richards G.T., Hernquist L., 2007, ApJ, 654,731Hummel E., van Gorkom J. H., Kotanyi C. G., 1983, ApJ,267, L5Ibar E., et al., 2008, MNRAS, 386, 953Ibar E., Ivison R., Biggs A., Lal D., Best P., Green D.,2009, MNRAS, 397, 281Ivison R. et al., 2010a, MNRAS, 402, 245Ivison B., et al. 2010b, A&A, 518, 31Jamrozy M., 2004, A&A, 419, 63Jamrozy M., Konar C., Machalski J., Saikia D.J., 2008,MNRAS, 385, 1286Kaiser C., Dennett-Thorpe J., Alexander P. 1997, MNRAS,292, 723Katgert-Merkelijn J., Robertson J., Windhorst R., KatgertP., 1985, A&AS, 61, 517Kellerman K., Sramek R., Schmidt M., Green R., ShafferD. 1994, AJ, 108, 1163Kim K., Kromberg P., 1990, ApJ, 355, 29Kogut A. et al., 2010, ApJ, submitted, preprint(arXiv:0901.0562)Konar C., Jamrozy M., Saikia D.J., Machalski, J., 2008,MNRAS, 383, 525Kronberg P., Kothes R., Salter C., Perillat P., 2007, ApJ, 659, 257Liang H., Hunstead R., Birkinshaw M., Andreani P., 2000,ApJ, 544, 686Liang R.A., Canvin J.R., Cotton W.D., Bridle, A.H., 2006,MNRAS, 368, 48Liang R.A., Bridle A.H., Parma P., Feretti L., GiovanniniG., Murgia M., Perley R.A., 2008, MNRAS, 386, 657Large M., Mathewson D., Haslam C., 1959, Nature, 183,1663LMachalski J., Jamrozy M., Zola S., 2001, A&A, 371, 445Machalski J., Jamrozy M., Zola S., Koziel D., 2006, A&A,454, 85Madau P., Della Valle M., Panagia N., 1998, MNRAS, 297,L17Magnelli B., Elbaz D., Chary R., Dickinson M., LeBorgneD., Frayer D. Willmer C., 2009, A&A, 496, 57Mannucci F., Della Valle M., Panagia N., 2007, MNRAS,377, 1229Marsden G. et al., 2010, ApJ, 707, 1729McGilchrist M., Baldwin J., Riley J., Titterington D., Wal-dram E., Warner P., 1990, MNRAS, 246, 110McNamara B.R., Nulsen, P.E.J., 2007, ARAA, 45, 117Owen F., Morrison G., 2008, ApJ, 136, 1889Padovani P., Mainieri V., Tozzi P., Kellermann K.I., Foma-lont E.B., Miller N., Rosati P., Shaver P., 2009, ApJ, 694,235Pearson T. Kus A., 1978, MNRAS, 182, 273Petrosian V., 2001, ApJ, 557, 560Prandoni L., Parma P., Wieringa M., de Ruiter H., Gre-gorini L., Mignano A., Vettolani G., Ekers R., 2006, A&A,457, 517Richards E., 2000, ApJ, 533, 611Richards E. et al.,, 2006, AJ, 131, 2766Rigby E.E., Best P.N., Snellen I.A.G., 2008, MNRAS, 385,310Rudnick L., Lemmerman J.A., 2009, ApJ, 697, 1341Rybicki G.B., Lightman A.P., 1979, ‘Radiative Processesin Astrophysics’ , New York: Wiley 1979Sargent M., et al., 2010, ApJ, 186, 341 submitted, preprint(arXiv:1005.1072)Schlickeiser R., Sievers A., Thiemann H., 1987, A&A, 182,21Seiffert M. et al., 2010, submitted, preprint(arXiv:0901.0559)Seymour N. et al. 2009, MNRAS, 2009, 398, 1573Sikora M., Stawarz L., & Lasota J.P., 2007, ApJ, 658, 815Singal J. et al. 2010, ApJ, accepted, preprint(arXiv:0901.0546)Simpson C. et al. 2006, MNRAS, 372, 741Smolˇci´c V. et al., 2009, ApJ, 696, 24Sreekumar P., et al., 1998, ApJ, 494, 523Stawarz L., Kneiske T. M., Kataoka J., 2006, ApJ, 637, 693Subrahmanyan R., Ekers R., Saripalli L., Sadler M., 2010,MNRAS, 402, 2792Subrahmanyan R. et al., in prepTaylor G., Fabian A., Allen S., 2002, MNRAS, 334, 769Taylor A., Stil J., Sunstrum C., 2009, ApJ, 702, 1230Vall´ee J.P., 2004, New Astronomy Review, 48, 763Vallee J., Roger R., 1989, A&AS, 77, 31Vlahakis C., Eales S., Dunne L., 2007, MNRAS, 379, 1042Volonteri M., Sikora M., Lasota J., 2007, ApJ, 667, 704Wang J. et al., 2009, ApJL, 697, 704 c (cid:13) , 1– ?? J. Singal et al.
Weiler K., Sramek R., Panagia N., van der Hulst J., SalvatiM., 1986, ApJ, 301, 790Weiler K., van Dyk S., Sramek R., Panagia N., 2004, newastro. reviews, 09.017White R., Helfand D., Becker R., Glickman E., De VriesW., 2007, ApJ, 654, 99Willott C.J., Rawlings S., Blundell K.M., Lacy M., EalesS.A., 2001, MNRAS, 322, 536Wilson M., 1970, MNRAS, 151, 1Wilson A., Vallee J., 1982, A&AS, 47, 601Windhorst R., va Heerde G., Katgert P., 1984, A&A, 58,1WWindhorst R., Miley G., Owen F., Kron R., Koo D., 1985,ApJ, 289, 494Windhorst R., Mathis D., Neuschaefer L., 1990, Astron.Soc. Pacific, 10, 398Windhorst R., Fomalont E., Partridge R., Lowenthal J.,1993, ApJ, 405, 9498Wise J., Abel T., 2005, ApJ, 629, 615 c (cid:13) , 1– ?? ources of Radio Background Table 1.
Radio source count data plotted in Figure 1Survey Frequency Facility a Resolution b S limitc γ d k d S maxe (GHz) (arcsec) ( µ Jy) (mJy)Fomalont et al. (2002) f ± .13 g ± .2 g g h h ± .13 g ×
12 60 1.83 ( i - 10001.89 ( > g > i
341 10000Windhorst et al. (1985) 1.4 VLA 14.7 225 2.1 ( h > g h × h f h h h × h × h h ×
57 9000 1.94 h g ×
276 70000 1.9 ± .1 g ×
288 70000 2.28 h h h > ×
100 80000 2.07 ( h > a VLA= Very Large Array, ATCA= Australia Telescope Compact Array, WSRT=Westerbrook SynthesisRadio Telescope, PSRT=Penticon Synthesis Radio Telescope, NCTRT=Northern Cross Transit RadioTelescope, CMBR= Cambridge Radio Telescope Survey. b The resolution quoted here is the full width at half max of the composite image. If the beam is not round,values for two axes are given. c The low flux limit of the survey is a factor determined by the survey authors (usually 5) times the RMSnoise. d From fits to dN/dS of the form dN/dS = k × S − γ , where the dN/dS distribution is expressed in sr − Jy − , and the flux S in Jy. e The highest flux object observed, or the highest flux listed in a dN/dS table or fit to dN/dS . f Separate results for different fields reported. g Source counts power law directly given. h dN/dS given in table and power law determined with fit. i Source counts given by equation in S .c (cid:13) , 1–, 1–