Planck early results
Free Access
Issue
A&A
Volume 536, December 2011
Planck early results
Article Number A15
Number of page(s) 56
Section Extragalactic astronomy
DOI https://doi.org/10.1051/0004-6361/201116466
Published online 01 December 2011

© ESO, 2011

1. Introduction

This paper is part of the first series of publications based on measurements made with the Planck1 satellite. Planck (Tauber et al. 2010; Planck Collaboration 2011a) is the third-generation space mission to measure the anisotropy of the cosmic microwave background (CMB). It observes the sky in nine frequency bands covering 30−857 GHz with high sensitivity and angular resolution from 31′ to 5′. The Low Frequency Instrument (LFI; Mandolesi et al. 2010; Bersanelli et al. 2010; Mennella et al. 2011) covers the 30, 44, and 70 GHz bands with amplifiers cooled to 20 K. The High Frequency Instrument (HFI; Lamarre et al. 2010; Planck HFI Core Team 2011a) covers the 100, 143, 217, 353, 545, and 857 GHz bands with bolometers cooled to 0.1 K. Polarization is measured in all but the highest two bands (Leahy et al. 2010; Rosset et al. 2010). A combination of radiative cooling and three mechanical coolers produces the temperatures needed for the detectors and optics (Planck Collaboration 2011b). Two data processing centres (DPCs) check and calibrate the data and make maps of the sky (Planck HFI Core Team 2011b; Zacchei et al. 2011). Planck’s sensitivity, angular resolution, and frequency coverage make it a powerful instrument for Galactic and extragalactic astrophysics as well as cosmology.

The paper uses data from the Planck Early Release Compact Source Catalogue (ERCSC; Planck Collaboration 2011c). The ERCSC provides positions and flux densities of compact sources found in each of the nine Planck frequency maps. The flux densities are calculated using aperture photometry, with careful modelling of Planck’s elliptical beams. The ERCSC includes data from the first all-sky survey, taken between 13 August 2009 and 6 June 2010. This unique dataset offers the first glimpse of the previously unmapped millimetre and sub-millimetre sky. It is used here to create spectral energy distributions (SEDs) of 104 radio-bright, northern active galactic nuclei (AGN), with the most complete coverage in the radio to sub-millimetre frequencies to date.

Radio-loud AGN host jets of relativistic matter emanating symmetrically from the core. These jets produce copious amounts of non-thermal radiation, which dominates the SEDs of such sources compared with any thermal emission from the nucleus, i.e., the accretion disk. The SEDs typically consist of two broad-band bumps, the one at lower frequencies attributed to synchrotron radiation, and the other at higher frequencies attributed to inverse Compton (IC) radiation. The peak frequencies of the two bumps vary from one object to another. The peak of the synchrotron component can be between the infrared and X-ray domains, and the IC peak can range from MeV to GeV energies. The sequence in the peak frequencies of the emitted energy, and the factors that create it, have been a hot topic in blazar research for more than a decade (e.g., Fossati et al. 1997; Ghisellini et al. 1998; Padovani 2007; Ghisellini & Tavecchio 2008; Nieppola et al. 2008; Ghisellini & Tavecchio 2008; Sambruna et al. 2010).

Research on blazar SEDs has concentrated on two approaches. The first is fitting the SED with a simple function, usually a second or third order polynomial, to obtain the pivotal peak frequencies and luminosities of the radiation components in a straightforward manner. This approach is typically used when studying large samples (Fossati et al. 1998; Nieppola et al. 2006; Sambruna et al. 2006; Nieppola et al. 2008; Abdo et al. 2010a). The second approach is detailed modelling of the SED, starting with the definition of initial parameters such as electron energy, magnetic field intensity, and Doppler factor. This method is more time-consuming and is used mostly for individual sources (e.g., Acciari et al. 2010; Collmar et al. 2010). The standard model is a leptonic, homogeneous, one-zone model, where the emission originates in a single component (for a review of the blazar emission models, see Böttcher 2010). One-zone models are useful as first-order approximations, but, in reality, AGN jets are rarely, if ever, dominated by a single source of radiation. The material in the jets flows through shocks in the jet, which locally enhance the radiation (the “shock-in-jet” model; Marscher & Gear 1985; Valtaoja et al. 1992). There can be several of these shocks in the jet simultaneously, and adding these to the emission of the quiescent jet, we have several radiation components. Therefore, ideally, the SEDs and radio spectra should be modelled with more than one component. Such modelling is also necessary for the proper identification of the high frequency (IC) emission sites.

In this paper we use the nearly complete SEDs and well-covered radio spectra provided by Planck to look for signs of these multiple components contributing to the total radiation that we observe. The paper is structured as follows. In Sects. 2 and 3 we introduce our sample and summarize the multifrequency data used in our study. The SEDs and radio spectra are discussed in Sects. 4 and 5, and their modelling is described in Sect. 6. In Sect. 7 we discuss the implications of our results for understanding the acceleration mechanisms in blazar jets, and in Sect. 8 we summarize our conclusions. Throughout the paper we adopt the sign convention for spectral index, αSν ∝ να. The errors of numerical values marked with a plus-minus sign correspond to one standard deviation.

2. The sample

The complete sample presented in this paper consists of 104 northern and equatorial radio-loud AGN. It includes all AGN with declination  ≥−10° that have a measured average radio flux density at 37 GHz exceeding 1 Jy. Most of the sample sources have been monitored at Metsähovi Radio Observatory for many years, and the brightest sources have been observed for up to 30 years. The sample can be divided into subclasses as follows: 33 high-polarization quasars (HPQs), 21 low-polarization quasars (LPQs), 21 BL Lacertae objects (BLOs), 19 normal quasars (QSOs), 9 radio galaxies (GALs), and one unclassified source (J184915+67064). (See, for example, Hovatta et al. 2008, 2009, for additional information on the classification.) By high-polarized quasars we mean objects which have a measured optical polarization  ≥3% at some epoch, while low-polarized quasars have a polarization  ≤3%. Normal quasars have no polarization measured, so they could be either HPQs or LPQs. Radio galaxies are non-quasar AGN. The full sample is listed in Table 1. Columns 1 and 2 give the name and J2000 name for the source, and for some sources an alternative name is given in Col. 3. The coordinates of the sources are given in Cols. 4 and 5. The start dates of the Planck scans are listed in Cols. 6 and 7. The average 37 GHz flux density from Metsähovi observations is given in Col. 8. For Col. 9 onwards, see Sect. 4.

3. Multifrequency data

The core of our data set is the Planck ERCSC. The construction and contents of the catalogue are described in Planck Collaboration (2011c). For most sources, the ERCSC flux density values are averages of two scans, separated by about six months. To enable extensive multifrequency studies with a simultaneous data set, the Planck Extragalactic Working Group has coordinated a programme in which ground-based and space-borne telescopes observe the sources in unison with Planck. In this paper we present SEDs based on the averaged ERCSC flux densities and on multifrequency supporting observations taken within two weeks of the Planck scans. In a later paper we will present single-epoch SEDs, constructed from all the available Planck and supplementary simultaneous multifrequency data. This collaborative programme includes 12 observatories around the world (Table 2). Archival data have been obtained from the literature and from the search tool at the ASI (Agenzia Spaziale Italiana) Science Data Center (ASDC) web page2.

Table 2

Optical and radio observatories that participated in the Planck multifrequency campaigns.

3.1. Radio and submillimetre data

Centimetre-band observations were obtained with the University of Michigan Radio Astronomy Observatory’s (UMRAO) 26-m prime focus paraboloid equipped with radiometers operating at central frequencies of 4.8, 8.0, and 14.5 GHz. Observations at all three frequencies employed rotating polarimeter systems permitting both total flux density and linear polarization to be measured. A typical measurement consisted of 8−16 individual measurements over a 20−40 min period. Frequent drift scans were made across stronger sources to verify the telescope pointing correction curves, and observations of programme sources were intermixed with observations of a grid of calibrator sources to correct for temporal changes in the antenna aperture efficiency. The flux scale was based on observations of Cassiopeia A (Baars et al. 1977). Details of the calibration and analysis techniques are described in Aller et al. (1985).

Table 3

Parameters for the RATAN-600 receivers.

Six-frequency broadband radio spectra were obtained with the RATAN-600 radio telescope in transit mode by observing simultaneously at 1.1, 2.3, 4.8, 7.7, 11.2, and 21.7 GHz (Parijskij 1993; Berlin & Friedman 1996). The parameters of the receivers are listed in Table 3, where νc is the central frequency, Δν is the bandwidth, ΔT is the sensitivity of the radiometer over a 1 s integration, Tphys is the physical temperature of the front-end amplifier, and Tsys is the noise temperature of the whole system at the given frequency. Data were reduced using the RATAN standard software FADPS (Flexible Astronomical Data Processing System) reduction package (Verkhodanov 1997). The flux density measurement procedure at RATAN-600 is described by Aliakberov et al. (1985).

The 37 GHz observations were made with the 13.7-m Metsähovi radio telescope using a 1 GHz bandwidth, dual-beam receiver centred at 36.8 GHz. The observations were ON–ON observations, alternating the source and the sky in each feed horn. The integration time used to obtain each flux density data point typically ranged from 1200 to 1400 s. The detection limit of the telescope at 37 GHz is  ~0.2 Jy under optimal conditions. Data points with a signal-to-noise ratio less than four were handled as non-detections. The flux density scale was set by observations of DR21. Sources NGC7027, 3C274, and 3C84 were used as secondary calibrators. A detailed description of the data reduction and analysis is given in Teräsranta et al. (1998). The error estimate in the flux density includes the contribution from the measurement rms and the uncertainty of the absolute calibration.

Quasi-simultaneous centimetre/millimetre radio spectra for a large number of Planck blazars have been obtained within the framework of a Fermi-GST related monitoring programme of gamma-ray blazars (the F-GAMMA programme, Fuhrmann et al. 2007; Angelakis et al. 2008). The frequency range spans 2.64 GHz to 142 GHz using the Effelsberg 100-m and IRAM 30-m telescopes. The Effelsberg measurements were conducted with the secondary focus heterodyne receivers at 2.64, 4.85, 8.35, 10.45, 14.60, 23.05, 32.00, and 43.00 GHz. The observations were performed quasi-simultaneously with cross-scans, i.e., slewing over the source position in the azimuth and elevation directions, with an adaptive number of sub-scans for reaching the desired sensitivity (for details, see Fuhrmann et al. 2008; Angelakis et al. 2008). Pointing offset correction, gain correction, atmospheric opacity correction, and sensitivity correction have been applied to the data.

The Institut de Radioastronomie Millimétrique (IRAM) observations were carried out with calibrated cross-scans using the EMIR horizontal and vertical polarisation receivers operating at 86.2 and 142.3 GHz. The opacity-corrected intensities were converted to the standard temperature scale and corrected for small remaining pointing offsets and systematic gain-elevation effects. The conversion to the standard flux-density scale was done using the instantaneous conversion factors derived from frequently observed primary calibrators (Mars, Uranus) and secondary calibrators (W3(OH), K3-50A, NGC7027). From this programme, radio spectra measured quasi-simultaneously with the Planck observations have been collected for a total of 37 Planck blazars during the period August 2009 to June 2010.

Many of the sources in the sample were monitored at 15 GHz using the 40-m telescope of the Owens Valley Radio Observatory (OVRO) as part of a larger monitoring programme (Richards et al. 2011). The flux density of each source was measured approximately twice weekly, with occasional gaps due to poor weather or instrumental problems. The telescope is equipped with a cooled receiver installed at prime focus, with two symmetric off-axis corrugated horn feeds sensitive to left circular polarization. The telescope and receiver combination produces a pair of approximately Gaussian beams (157′′ FWHM), separated in azimuth by 12.′95. The receiver has a centre frequency of 15.0 GHz, a 3.0 GHz bandwidth, and a noise-equivalent reception bandwidth of 2.5 GHz. Measurements were made using a Dicke-switched dual-beam system, with a second level of switching in azimuth to alternate source and sky in each of the two horns. Calibration was referred to 3C286, for which a flux density of 3.44 Jy at 15 GHz is assumed (Baars et al. 1977). Details of the observations, calibration, and analysis are given by Richards et al. (2011).

The Very Large Array (VLA) and (since spring 2010) the Expanded VLA also observed a subset of the sources as simultaneously as possible. Most of the VLA and EVLA runs were brief 1−2 h blocks of time. In a one-hour block of time, in addition to flux calibrators and phase calibrators, typically 5−8 Planck sources were observed. In many cases, VLA flux density and phase calibrators were themselves of interest, since they were bright enough to be detected by Planck. For these bright sources, the integration times could be extremely short. At 4.86 GHz and 8.46 GHz, each target was observed for approximately 30 s. The integration times were typically 100 s at 22.46 GHz and 120 s at 43.34 GHz. All VLA/EVLA flux density measurements were calibrated using standard values for one or both of the primary calibrator sources used by NRAO (3C48 or 3C286), and the uv data were flagged, calibrated and imaged using standard NRAO software, AIPS or CASA. The VLA and EVLA were in different configurations at different times in the several months duration of the observations. As a consequence, the angular resolution changed. In addition, for a given configuration, the resolution was much finer at higher frequencies. For that reason, sources that showed signs of resolution in any VLA configuration or at any VLA frequency have been carefully flagged.

The Planck-ATCA Co-eval Observations (PACO) project (Massardi et al. 2011a) consists of observations of a compilation of sources selected in the Australia Telescope 20 GHz survey (AT20G; Massardi et al. 2011b) taken with the Australia Telescope Compact Array (ATCA) in the frequency range 4.5−40 GHz. The observations were carried out at several epochs close in time with Planck satellite observations covering July 2009−August 2010. The PACO sample is a complete, flux-density limited, and spectrally-selected sample over the whole southern sky, with the exception of the region with Galactic latitude |b| < 5°.

The Simultaneous Medicina Planck Experiment (SiMPlE; Procopio et al. 2011) uses the 32-m Medicina single-dish antenna at 5 and 8.3 GHz to observe the 263 sources of the New Extragalactic WMAP Point Source (NEWPS) sample (Massardi et al. 2009) with δ > 0°, and partially overlapping with the PACO observations for −10° < δ < 0°. The project began in June 2010. Because of the lack of simultaneity with the ERCSC, the data have been used here to match the PACO observations in the overlapping region and to add information for δ > 70°, a region so far poorly covered at 5 GHz.

Twelve sources from our sample were observed in the submillimetre domain with the 12-m Atacama Pathfinder EXperiment (APEX) in Chile. The observations were made using the LABOCA bolometer array centred at 345 GHz. Data were taken at two epochs in 2009: September 3−4 2009 and November 12 2009. The data were reduced using the script package minicrush3, version 30-Oct.-2009, with Uranus used as calibrator.

3.2. Optical data

The optical observations were made with the 35-cm KVA (Kungliga Vetenskapsakademien) telescope on La Palma, Canary islands. All observations were made through the R-band filter (λeff = 640 nm) using a Santa Barbara ST-8 CCD camera with a gain factor of 2.3 e/ADU and readout noise of 14 electrons. The binning of pixels by 2  ×  2 pixels resulted in a plate scale of  per pixel. We obtained three to six exposures of 180 s per target. The images were reduced in the standard way of subtracting the bias and dark frames and dividing by twilight flat-fields.

The flux densities of the target and three to five stars in the target field were measured with aperture photometry and the magnitude difference between the target and a primary reference star in the same field was determined. Using differential mode makes the observations insensitive to variations in atmospheric transparency and accurate measurements can be obtained even in partially cloudy conditions. The R-band magnitude of the primary reference star was determined from observations made on photometric nights, using comparison stars in known blazar fields as calibrators (Fiorucci & Tosti 1996; Fiorucci et al. 1998; Raiteri et al. 1998; Villata et al. 1998; Nilsson et al. 2007) and taking into account the color term of the R-band filter employed. After the R-band magnitude of the primary reference star was determined, the object magnitudes were computed from the magnitude differences. At this phase we assumed V − R = 0.5 for the targets. Several stars in the field were used to check the quality of the photometry and stability of the primary reference. The uncertainties in the magnitudes include the contribution from both measurement and calibration errors.

The monitoring at Xinglong Station, National Astronomical Observatories of China, was performed with a 60/90-cm f/3 Schmidt telescope. The telescope is equipped with a 4096  ×  4096 E2V CCD, which has a pixel size of 12   μm and a spatial resolution of  per pixel. The observations were made with an i filter. Its central wavelength and passband width are 668.5 and 51.4 nm, respectively. The exposure times are mostly 120 s but can range from 60 to 180 s, depending on weather and moon phase.

3.3. Swift

The Swift Gamma-Ray-Burst (GRB) Explorer (Gehrels et al. 2004) is a multi-frequency space observatory devoted to the discovery and rapid follow up of GRBs. There are three instruments on board the spacecraft: the UV and Optical Telescope (UVOT, Roming et al. 2005), the X-Ray Telescope (XRT, Burrows et al. 2005) sensitive in the 0.3−10.0 keV band, and the Burst Alert Telescope (BAT, Barthelmy et al. 2005) sensitive in the 15−150 keV band. Although the primary scientific goal of the satellite is the observation of GRBs, the wide frequency coverage is useful for AGN studies because it covers the region where the transition between synchrotron and inverse Compton emission usually occurs.

When not engaged in GRB observations, Swift is available for target of opportunity (ToO) requests. The Swift team decided to devote an average of three ToO per week to the simultaneous observations of Planck AGN. The simultaneity of Swift observations within two weeks to the Planck first or second scan, or both, is shown in Table 4, marked sim_1st, sim_2nd, or sim_1st_2nd, respectively.

3.3.1. UVOT

Swift UVOT observations were carried out using the “filter of the day”, i.e., one of the lenticular filters (V, B, U, UVW1, UVM2, and UVW2), unless otherwise specified in the ToO request. Therefore images are not always available for all filters.

The photometry analysis of all our sources was performed using the standard UVOT software distributed within the HEAsoft 6.8.0 package, and the calibration included in the latest release of the “Calibration Database” (CALDB). A specific procedure has been developed to process all the ToO observations requested by the Planck project. Counts were extracted from an aperture of 5′′ radius for all filters and converted to flux densities using the standard zero points (Poole et al. 2008). The flux densities were then de-reddened using the appropriate values of E(B − V) for each source taken from Schlegel et al. (1998), with AλE(B − V) ratios calculated for UVOT filters using the mean galactic interstellar extinction curve from Fitzpatrick (1999). No variability was detected within single exposures in any filter.

3.3.2. XRT

The XRT is usually operated in “Auto State” mode which automatically adjusts the CCD read-out mode to the source brightness, in an attempt to avoid pile-up (see Burrows et al. 2005; Hill et al. 2004, for details of the XRT observing modes). As a consequence, part of the data sample was collected using the most sensitive Photon Counting (PC) mode while Windowed Timing (WT) mode was used for bright sources.

The XRT data were processed with the XRTDAS software package (v. 2.5.1, Capalbi et al. 2005) developed at the ASDC and distributed by the NASA High Energy Astrophysics Archive Research Center (HEASARC) within the HEASoft package (v. 6.9). Event files were calibrated and cleaned with standard filtering criteria with the xrtpipeline task, using the latest calibration files available in the Swift CALDB. Events in the energy range 0.3−10 keV with grades 0−12 (PC mode) and 0−2 (WT mode) were used for the analysis.

Events for the spectral analysis were selected within a circle of a 20-pixel (~47″) radius, centred on the source position, which encloses about 90% of the point spread function (PSF) at 1.5 keV (Moretti et al. 2005). For PC mode data, when the source count rate is above  ~0.5 counts/s data are significantly affected by pile-up in the inner part of the PSF. For such cases, after comparing the observed PSF profile with the analytical model derived by Moretti et al. (2005), we removed pile-up effects by excluding events detected within a certain inner radius around the source position, and used an outer radius of 30 pixels. The value of the inner radius was evaluated individually for each observation affected by pile-up, depending on the observed source count rate.

Ancillary response files were generated with the xrtmkarf task applying corrections for the PSF losses and CCD defects. Source spectra were binned to ensure a minimum of 20 counts per bin to allow the χ2 minimization fitting technique to be used. We fitted the spectra adopting an absorbed power-law model with photon index Γx. When deviations from a single power-law model were found, we adopted a log-parabolic law of the form F(E) = KE−(a + blog (E)) (Massaro et al. 2004) which has been shown to fit the X-ray spectrum of blazars well (e.g., Giommi et al. 2005; Tramacere et al. 2009). This spectral model is described by only two parameters: a, the photon index at 1 keV, and b, the curvature of the parabola. For both models the amount of hydrogen-equivalent column density (NH) was fixed to the Galactic value along the line of sight (Kalberla et al. 2005).

3.4. Fermi-LAT observations and data analysis

The Large Area Telescope (LAT) onboard Fermi is an electron-positron pair-conversion telescope sensitive to gamma-rays of energies from 20 MeV to above 300 GeV. The Fermi-LAT consists of a high-resolution silicon microstrip tracker, a CsI hodoscopic electromagnetic calorimeter, and an anticoincidence detector for charged particles background identification. A full description of the instrument and its performance can be found in Atwood et al. (2009). The large field of view (~2.4 sr) allows the LAT to observe the full sky in survey mode every 3 h. The LAT PSF strongly depends on both the energy and the conversion point in the tracker, but less on the incidence angle. For 1 GeV normal incidence conversions in the upper section of the tracker, the PSF 68% containment radius is 0.6°.

The Fermi-LAT data considered for this analysis cover the period from 4 August 2008 to 4 November 2010. They have been analyzed using the standard Fermi-LAT ScienceTools software package4 (version v9r16) and selecting for each source only photons above 100 MeV, belonging to the diffuse class (Pass6 V3 IRF, Atwood et al. 2009) which have the lowest background contamination. For each source, we selected only photons within a 15° region of interest (RoI) centred around the source itself. In order to avoid background contamination from the bright Earth limb, time intervals where the Earth entered the LAT Field of View (FoV) were excluded from the data sample. In addition, we have excluded observations in which the source was viewed at zenith angles larger than 105°, where Earth’s albedo gamma-rays increase the background contamination. The data were analyzed with a binned maximum likelihood technique (Mattox et al. 1996) using the analysis software (gtlike) developed by the LAT team5. A model accounting for the diffuse emission as well as the nearby gamma-ray sources is included in the fit.

The diffuse foreground, including Galactic interstellar emission, extragalactic gamma-ray emission, and residual cosmic ray (CR) background, has been modelled using the models 6gll iem v02 for the Galactic diffuse emission and isotropic iem v02 for the extra-galactic isotropic emission. Each source has been fit with a power law function (1)where both the normalization factor N and the photon index γ are left free in the model fit. The model also includes all the sources within a 20° RoI included in the Fermi-LAT one year catalogue (Abdo et al. 2010b), modelled using power law functions. If a source included in the model is a pulsar belonging to the Fermi pulsar catalog (Abdo et al. 2010d), we have modelled the source using a power-law with exponential cut-off and the spectral parameters from the pulsar catalogue. For the evaluation of the gamma-ray SEDs, the whole energy range from 100 MeV to 300 GeV is divided into two equal logarithmically spaced bins per decade. In each energy bin the standard gtlike binned analysis has been applied assuming for all the point sources in the model a power law spectrum with photon index fixed to −2.0. Assuming that in each energy bin the spectral shape can be approximated by a power law, the flux density of the source in all selected energy bins was evaluated, requiring in each energy bin a test statistic (TS)7 greater than 10. If the TS is lower than 10, an upper limit is evaluated in that energy bin. Only statistical errors for the fit parameters are shown in the plots. Systematic effects are mainly based on uncertainties in the LAT effective area derived from the on-orbit estimations, and are  <5% near 1 GeV, 10% below 0.10 GeV, and 20% above 10 GeV.

The LAT gamma-ray spectra of all AGN detected by Fermi are studied in Abdo et al. (2010c), based on 11 months of Fermi-LAT data. For this paper, we derived the gamma-ray SEDs of the 104 sources in the sample in three time intervals, presented in Table 4. In the first interval the Fermi-LAT observations are simultaneous to the Planck first or second scan, or both, within two weeks (marked sim_1st, sim_2nd, or sim_1st_2nd in Table 4). The Fermi-LAT data have in this case been integrated over two weeks. In the second interval the gamma-ray data are quasi-simultaneous to the Planck first or second scan, or both, having been integrated over two months (2M_1st, 2M_2nd, or 2M_1st_2nd in Table 4). In the third interval are sources for which Fermi-LAT data have been averaged over 27 months due to their faintness (27 m in Table 4).

Note that eight sources (0306+102, 0355+508, 0804+499, 0945+408, J1130+3815, 1413+135, 1928+738, 2005+403) that appear as significant gamma-ray detections in the 27 month dataset presented in this work, have not been included in any Fermi catalogue published so far (Abdo et al. 2010b). They can be considered new gamma-ray emitting sources, taking into account the possibility that their association with the radio source is purely spatial.

4. Spectral energy distributions

The SEDs for the whole sample are shown in Figs. 18121 (in panels on the left). The Planck measurements and the simultaneous auxiliary data are indicated by red circles. The grey points represent archival data obtained from the literature and using the ASDC search tool. The measurements were regarded as simultaneous if they were taken within two weeks (for radio frequencies) or five days (for optical and higher energies) of the Planck measurement. Previous studies of the radio variability of blazars (Hovatta et al. 2008; Nieppola et al. 2009) have shown that the two-week simultaneity limit is appropriate, as large-scale radio flux density changes on that timescale are quite rare. The SEDs were fitted with second and third degree polynomials, namely (2)and (3)where ci (i = 1...7) are fit parameters. We calculated the peak frequencies of the components from the third degree fit by default and used the second degree fit only if it seemed by visual inspection to be more accurate. Typically, in these cases, the third degree fit would not yield a sensible result at all. In the SEDs in Figs. 18121, third degree fits are marked with a dashed line, second degree fits with a dotted line.

The fits were divided into two classes, A and B, according to the fit quality, A being superior. We emphasize that this division is the subjective view of the authors, and is only meant to illustrate the varying quality of the SED fits. The classes have not been used in calculating the correlation coefficients (see below). We were able to obtain meaningful synchrotron fits for 60 sources (15 class A fits), and IC fits for 30 sources (10 class A fits). For 21 sources we were able to fit both components. The synchrotron and IC peak frequencies are listed in Table 1. Columns 9 and 10 give the peak frequency and the fit quality for the synchrotron component, and Cols. 11 and 12 give the peak frequency and the fit quality for the inverse Compton component.

thumbnail Fig.1

The distributions of the synchrotron (top panel) and IC (bottom panel) peak frequencies.

Open with DEXTER

Figure 1 shows the distributions of the logarithms of the peak frequencies. The synchrotron peak frequencies, , are typically very low, as we are dealing with bright radio sources. They range over two orders of magnitude, from 12.2 to 14.3, the average of the distribution being 13.2  ±  0.5. The source with the highest  in the near-ultraviolet domain is 0716+714. The IC peak frequencies, , range over three orders of magnitude, from MeV to the high GeV gamma-ray region. The distribution average is 22.0  ±  0.9. The source with the highest IC peak frequency is 1156+295, with . The source with the lowest IC peak frequency, with , is 0836+710.

We have plotted the interdependence of the component peaks in Fig. 2. The two peak frequencies seem to have no significant correlation, which is confirmed by the Spearman rank correlation test (ρ = 0.301, P = 0.092, both class fits included). There is, however, a tendency for the sources with high  to have a high . In the synchrotron-self-Compton (SSC) scenario, the separation of the component peaks depends on many factors, such as the electron Lorentz factors, magnetic field strengths, and particle densities. Therefore the lack of correlation as such does not allow any strong statements to be made about the origin of the IC radiation (SSC, or external Compton, EC).

We also compared our peak frequencies with those of Abdo et al. (2010a). These comparisons are illustrated in Fig. 3, where the one-to-one correspondence is shown with a dashed line. Both correlations are significant according to the Spearman test (ρ = 0.629, P = 0.003 for the synchrotron peaks and ρ = 0.660, P = 0.010 for the IC ones).

thumbnail Fig.2

The correlation of the peak frequencies of the synchrotron and the IC components. Solid circles denote class A fits, and open circles class B fits.

Open with DEXTER

thumbnail Fig.3

Comparison between our synchrotron and IC peak frequencies and those of Abdo et al. (2010a). The synchrotron peak frequencies are shown in the top panel and IC peak frequencies in the bottom panel. The dashed lines illustrate the one-to-one correspondence. Solid circles denote class A fits, and open circles class B fits.

Open with DEXTER

5. Radio spectra

In addition to the SEDs, we have plotted the standard radio spectra for the sample sources. The aim was to look for signs of multiple components contributing to the synchrotron emission. The spectra are shown in Figs. 18121 (in panels on the right). Red circles indicate low frequency (LF;  ≤ 70 GHz) data simultaneous to Planck, red stars indicate ERCSC LFI data, blue circles indicate high frequency (HF;  >70 GHz) data simultaneous to Planck, and blue stars indicate ERCSC HFI data. As can be seen, the Planck data and the supporting multifrequency data provide an excellent opportunity to find the signatures of possible multiple components or anomalous spectral shapes. A statistical study of extragalactic radio sources in the ERCSC is presented in Planck Collaboration (2011i), and SEDs of extreme radio sources, such as gigahertz-peaked spectrum (GPS) sources in Planck Collaboration (2011j).

In Figs. 4 and 5 we have plotted the distributions of the LF and HF spectral indices using only the Planck data for the latter. When plotting the HF α values, we have only taken into account those sources with three or more data points. This leaves us with 84 sources out of the 104. The sources with only one Planck scan are shown hatched in Figs. 4 and 5. Figure 6 shows αLF versus αHF. As expected, the LF indices are fairly flat, with an average of −0.06. Their distribution is narrow, with 91% of the indices being in the range αLF = −0.5−0.5. There are a couple of sources with remarkably steep LF spectra, namely 0552+398 and 2021+614, both having αLF ≤ −0.8. At the other end we have 0007+106 which has an inverted spectrum with αLF = 0.86. Also 1228+126 has a steep spectrum, although the fit value may be exaggerated in this case (see Fig. 75). For clarity, this source has been omitted from Figs. 4 and 6.

The HF indices are concentrated around −0.5, having an average of −0.56. Rather similar distributions for LF and HF spectral indices were found in Planck Collaboration (2011i), especially in their Fig. 7. However, a large part of the αHF distribution is in the flat domain with αHF ≥ −0.5. Extreme examples are 1413+135 (αHF = 0.02) and 0954+658 (αHF = 0.34). It is unexpected that the spectral indices of blazars at sub-millimetre and FIR frequencies should be this flat. Only 28 of the 84 sources with HF spectral fits shown in Figs. 18121 have α < −0.7, the “canonical” value for optically thin spectra, corresponding to an electron energy index s of about 2.4. This has two possible explanations: either the total HF spectra are defined by several underlying components, or the energy spectrum of the electron population producing the radiation is much harder than generally assumed. Although some HF spectra show evidence for several sub-components or even an upturn at the highest frequencies, many others appear to be perfectly straight and therefore perhaps originate from a single optically thin component. At the highest Planck frequencies the lifetimes of radiating electrons are short, and one would in most cases expect to see the optically thin spectra steepened by energy losses, with indices ranging from the original αthin to αthin−0.5 and beyond. This makes spectral indices  >−0.7 even more remarkable.

thumbnail Fig.4

Distribution of LF spectral indices for the whole sample (104 sources). The sources that have only been scanned once by Planck are shown hatched (46 sources).

Open with DEXTER

thumbnail Fig.5

Distribution of HF spectral indices for the 84 sources that had three or more data points. The sources that have only been scanned once by Planck are shown hatched (40 sources).

Open with DEXTER

To rule out contamination by dust in the host galaxies, we have estimated the possible contribution of an extremely bright submillimetre galaxy. We also estimated the flux densities of 2251+158 at several submillimetre bands in the case its luminosity were similar to the submillimetre galaxy. The comparison shows that if the dust component in 2251+158 were of the same brightness as in the extremely bright submillimetre galaxy, its emission would be completely swamped by the non-thermal emission from the jet. The LF and HF spectral indices have been calculated using ERCSC data which for most sources is an average of two Planck scans separated by about six months. However, the HF spectral index distribution for the sources that have been observed only once by Planck is quite similar to that of the whole sample. The implications of this result are discussed in more detail in Sect. 7.

thumbnail Fig.6

LF spectral indices αLF versus HF spectral indices αHF for the various AGN subclasses.

Open with DEXTER

thumbnail Fig.7

The 37 GHz lightcurve of 0234+285, measured at Metsähovi Radio Observatory, showing continuous variability. The period when Planck was observing the source is indicated by the hatched region in the lower panel, which is a blow-up of the last year.

Open with DEXTER

As Fig. 6 shows, BLOs, HPQs and LPQs have different αHF distributions. In Table 5 we have listed the average spectral indices for both frequency ranges for all subgroups. The LF spectra of radio galaxies are very steep on average, but the value is greatly influenced by the uncertain fit of 1228+126. If this source is left out of the calculations, the average αLF for galaxies is 0.10  ±  0.52. LPQs have the steepest spectra in the HF range. However, the standard deviations of the samples are substantial, which can also be seen as the large scatter in Fig. 6. To quantify the possible differences between the AGN classes, we ran the Kruskal-Wallis ANOVA test. For the LF indices we could find no significant differences. However, in the case of the HF indices, the distributions of the subgroups differ significantly (P < 0.001). Multiple-comparisons test with t distribution tells us that LPQs differ from all other subgroups (P ≤ 0.031). Also, BLO indices have a different distribution from all quasar subgroups (P ≤ 0.019).

Table 5

Average spectral indices for AGN subclasses.

6. Modelling the SEDs and spectra of blazars

Spectral energy distributions rapidly became one of the main tools for understanding blazar physics after the EGRET instrument aboard the Compton Gamma-Ray Observatory (CGRO) satellite discovered strong gamma-ray emission from 3C279 (Hartman et al. 1992) and, subsequently, from a large number of other radio-bright AGN. It was rapidly accepted that the double-peaked overall shape of the radio-to-gamma SED was due to synchrotron and inverse Compton radiation, but beyond this the agreement ended. The data simply lacked the sensitivity and the time coverage to sufficiently constrain the models. Even the most detailed and convincing effort for 3C279 (Hartman et al. 2001) could be criticized for unrealistic physical assumptions (Lindfors et al. 2005).

With the coming of the Fermi and Swift satellites, ground-based TeV telescopes, and other satellite data, the inverse Compton high-energy SED of a large number of AGN can now be determined with hugely improved accuracy and time coverage. This has initiated a new era of AGN modelling (e.g., Proceedings of the Workshop “Fermi meets Jansky: AGN in Radio and Gamma Rays”; Fermi meets Jansky 2010). However, since the IC spectrum is created by the relativistic electrons scattering either the synchrotron photons (SSC) or external photons (EC), accurate knowledge of the shape of the synchrotron component is essential for all realistic modelling. Here the Planck data are invaluable, since they provide the first simultaneous multifrequency coverage of the crucial radio-to-IR part of the synchrotron spectrum.

However, most of the ERCSC data are not suitable for detailed modelling, since they are averages of two spectra taken about six months apart. Considering the strong SED variability, accurate modelling can be done only with the final Planck datasets and their several individual scans. Here we limit ourselves to a short discussion of how Planck and supporting data can be used to improve our understanding of blazar physics.

The basic picture of a blazar is well established: a black hole surrounded by an accretion disk and a broad-line emitting region (BLR), and a relativistic jet which produces the lower-frequency bump of the SED through synchrotron emission. The jet is not totally stable; VLBI imaging shows how new components, identified as relativistic shocks in the jet, emerge from the radio core at intervals ranging from months to years. These growing and decaying shocks are also responsible for the total flux density variations in the radio-to-IR, and at least partly also in the optical regime. The important thing to note is that the synchrotron SED is rarely, if ever, the product of a single synchrotron-radiating component. The nature of the radio core is still being debated, as is the nature and the extent of the jet flow upstream of the radio core, between it and the black hole (e.g., Marscher & Jorstad 2010; Marscher et al. 2010). Shocked jet models (Marscher & Gear 1985; Hughes et al. 1989; Valtaoja et al. 1992) provide a good account of the multifrequency variations in the total flux density. In the particular case of 3C279, Lindfors et al. (2006) showed that the optical variations can also be moderately well explained as the higher-frequency part of the shock emission.

The origin of the IC component is a matter of intense debate. In the CGRO era, the favoured alternative was that the high-frequency radiation originates close to the black hole and the accretion disk, inside the BLR and well upstream of the radio core. With a strong radiation field external to the jet, external Compton is then likely to be the dominant emission mechanism. A minority view was that at least a part of the gamma-ray flaring comes from much farther out, at or downstream from the radio core. The evidence for this was that, statistically speaking, strong gamma-ray flaring tended to occur after the beginning of a millimetre flare (Valtaoja & Teräsranta 1996; Lähteenmäki & Valtaoja 2003) and after the ejection of a new VLBI component (Jorstad et al. 2001). Occurring well outside the BLR, SSC seems in this case to be the only viable emission mechanism, but on the other hand simple SSC models have generally failed to explain the gamma-ray emission, requiring unrealistically low magnetic field strengths in other than the extreme TeV BL Lac objects. With new Fermi data, the evidence for gamma-ray emission from the radio jet has strengthened (Kovalev et al. 2009; Pushkarev et al. 2010; León-Tavares et al. 2011; Nieppola et al. 2010; Tornikoski et al. 2010; Valtaoja et al. 2010), but the source of the upscattered photons remains a puzzle.

Most of the attempts to model the high-energy bump have adopted a “reverse engineering” method, starting from the secondary IC SED and not from the primary synchrotron SED. The original electron energy spectrum is considered to be rather freely adjustable, with parameters chosen to produce a good fit of the calculated IC spectrum to the observations within the assumed model. In most cases, the synchrotron spectrum is assumed to originate in a single homogeneous component, and often the spectrum is adjusted only to the optical data, without any attempt to explain the radio-to-IR part of the SED. There is not much observational or theoretical evidence for the actual physical existence of such an IR-to-optical component: it is simply assumed to be there, because it can produce the observed IC spectrum.

To some extent such an approach is justifiable if the gamma-rays originate upstream of the radio core, and are therefore not directly connected to the observed radio jet and shocks, and their lower-frequency synchrotron emission. However, even in this case physically more realistic models must take into account also the radio-to-IR part, for several reasons. First, whatever the exact mechanism, the radio shocks also contribute to the IC spectrum, and their contribution must be accounted for in the model fits. Assuming reasonable physical parameters, the shocks can quite often produce part or all of the observed X-ray emission through the SSC mechanism (e.g., Lindfors et al. 2006). Indeed, a standard method for estimating the physical parameters of VLBI components is to assume that the X-ray emission is produced through SSC (Ghisellini et al. 1993; Marscher 1987). Secondly, the components seen in the radio, especially in the Planck HFI regime, contribute to the optical synchrotron emission, and this contribution must be subtracted from the putative IR-to-optical component, which, as explained above, in many models generates the whole IC spectrum. For example, in the present sample, at least 0235+164, 0917+449, 0954+658, 2227−088, and especially 2251+158 show ongoing millimetre-to-submillimetre flaring, which leaves little room for any additional components between the highest Planck frequencies and the optical. This is the case in particular if the optically thin spectra are originally very flat, as we argue in Sects. 6.1 and 7.

On the other hand, if gamma-rays do come from the shocks at, or downstream from, the radio core, as increasing evidence indicates, the obvious first step is to model the shocks themselves and their spectra as accurately as possible using Planck and other radio data, and only after that attempt to calculate the IC radiation they produce.

Finally, it is clear that the more realistic multicomponent models, whatever their exact details, will give rather different physical parameter values than the single-component models. As an example, if the X-ray emission and the gamma-ray emission are not produced by a single component but by two (or even more) components, the derived electron energy spectral parameters will be quite different.

6.1. Some examples of different spectra and their implications

In this section we highlight five examples of different spectra of ERCSC sources which have been scanned only once by Planck and therefore have true snapshot spectra presented in Figs. 18121. As stated before, our aim at this stage is not to model these sources, but to draw attention to some features commonly seen in the SEDs, and to their implications for future multi-epoch modelling of the Planck sources. We have fitted the radio-to-optical spectra with one or several standard self-absorbed sychrotron components (based on a power-law assumption for the electron energy spectrum). These are not model fits from a numerical, physical code (except for Figs. 16 and 17), but rather are meant to guide the eye and to illustrate the general shape of the radio-to-optical spectra.

We also use support observations to address the behaviour of the five example sources. In Figs. 711 we have plotted the 37 GHz long-term flux density curves for the five sources. The long-term data have been taken at Metsähovi Radio Observatory.

thumbnail Fig.8

The 37 GHz lightcurve of 0235+164, details as in Fig. 7.

Open with DEXTER

Even without numerical modelling we can address two important topics with these sources. The first one is the spectral flatness of the optically thin radiation possibly hinting at non-standard acceleration processes. The second one concerns the origin of the IC component. As Figs. 1217 show there does not appear to be room for additional synchrotron components between the Planck and the optical frequencies. The IC radiation must therefore originate in the same shocks that also produce the radio emission, in contradiction to most theoretical models.

thumbnail Fig.9

The 37 GHz lightcurve of 1253−055 (3C279), details as in Fig. 7.

Open with DEXTER

thumbnail Fig.10

The 37 GHz lightcurve of 1413+135, details as in Fig. 7.

Open with DEXTER

thumbnail Fig.11

The 37 GHz lightcurve of 2251+158 (3C454.3), details as in Fig. 7.

Open with DEXTER

thumbnail Fig.12

A possible single-component spectrum for 0234+285 to illustrate the general shape. Red circles, LF data simultaneous to Planck; blue circles, HF data simultaneous to Planck.

Open with DEXTER

thumbnail Fig.13

A possible multiple-component spectrum for 0235+164 to illustrate the general shape. Symbols as in Fig. 12.

Open with DEXTER

thumbnail Fig.14

The radio spectrum of 1253−055 shows that there is no room for another component between radio and optical regimes. Symbols as in Fig. 12.

Open with DEXTER

0234+285 (Fig. 12). This source was in a very uneventful state during the Planck observing period (Fig. 7) and shows a spectrum typical of sources in a quiescent stage. The radio spectrum can be modelled with a single synchrotron component, having αthin = −0.61, which apparently steepens by Δα = −0.5 and smoothly joins the optical spectrum (see also Fig. 29). This synchrotron component can be identified with the latest significant flare (new shock), which peaked about two years earlier at 37 GHz (Fig. 7). During low flux levels, in the absence of recent flares/shocks, the underlying, relatively stable jet may also contribute significantly to the synchrotron emission. Another similar example is the source 0007+106 (Fig. 19). Lacking mid-IR data, we cannot prove that there is no additional synchrotron component, between the Planck frequencies and the optical, that might be the main source of the IC flux (Fig. 29). However, the most straightforward interpretation of the data is that the IC flux also comes from the synchrotron component outlined in Fig. 12: from a rather old shock far downstream the jet, and far beyond the BLR.

thumbnail Fig.15

The radio spectrum of 1413+135 is very flat and can be fitted with multiple components. Symbols as in Fig. 12.

Open with DEXTER

thumbnail Fig.16

The radio-optical SED of 2251+158 (3C454.3) fitted with one component and synchrotron emission from the jet (dashed line) using a model by Tammi et al. (in prep.). The radio spectrum is shown in Fig. 17.

Open with DEXTER

thumbnail Fig.17

A close-up of the radio spectrum and model fit of 2251+158 (3C454.3), details as in Fig. 16.

Open with DEXTER

0235+164 (Fig. 13). The bumpiness of the radio spectrum shows clear evidence of multiple components (shocks) contributing to the observed total flux density. We have outlined two possible synchrotron components, which together could explain most of the observed radio spectrum. Although 0235+164 was relatively quiescent during the Planck observation (Fig. 8), there were at least four strong flares during the previous five years, and they all are likely to be still contributing significantly to the total radio flux density, at least around 1010 Hz and below. Assuming reasonable physical parameters for these shocks, they can also produce significant amounts of X-rays through the SSC mechanism, which must be accounted for in a proper multicomponent modelling of the IC spectrum. As with 0234+285, the higher-turnover component with αthin = −0.4 can be joined smoothly to the optical spectrum assuming a steepening by about Δα = −0.5, leaving no room for additional mid-IR components. Fermi measured only upper limits for the gamma-ray flux, which may be due to the fact that even the most recent shock, peaking around 100 GHz, is already about two years old (Fig. 8).

1253055 (3C279). (Fig. 14). Again, the highest-peaking radio component with αthin = −0.6 can be made to smoothly join the steep optical spectrum. From Fig. 14 it is rather obvious that there can be no additional mid-IR components which might provide a significant contribution to the IC spectrum. If we assume that there is another synchrotron component turning over somewhere above the Planck frequencies and joining smoothly to the optical spectrum, Fig. 14 shows that its flux density cannot exceed the dashed line at any IR frequency between 1012 and 1014.5 Hz. This is the case even if we make the assumption that the spectrum of the component seen at the highest Planck frequencies steepens very rapidly above 857 GHz. Consequently, the Fermi gamma-rays must originate in the latest shock, which started a few months before the Planck observation (Fig. 9).

1413+135 (Fig. 15). This source exhibits a totally flat spectrum, with both the LF and the HF spectral indices approximately 0. Possible explanations are: an underlying electron energy index s ≈ 1, which is rather unlikely; the superposition of many self-absorbed synchrotron components (shocks of various ages) with different turnover frequencies; or a continuous jet dominating the flux and having an integrated spectral index close to zero over a wide frequency range. As Fig. 10 shows, the source has been rather inactive for over a decade, with no major ourbursts. The HFI spectrum may indicate an upturn, hinting at the presence of an IR component. We show a possible spectral decomposition, including an IR-peaking component. Further Planck observations are needed to ascertain the reality and the nature of this component, which may be either a synchrotron flare seen at an early developmental stage or, possibly, an infrared dust component. The correct alternative can be identified through variability information.

2251+158 (3C454.3). (Figs. 16 and 17). This source exhibits the most spectacular case of ongoing IR flaring, showing a very strong synchrotron component with a self-absorption turnover at about 80 GHz and a very flat (α = −0.2) optically thin spectrum. As Fig. 11 shows, a series of spectacular flaring events has occurred in 2251+158 during the last five years, with Planck observing the source during the early stages of the strongest outburst ever seen in the source. In Fig. 16 we show a model fit to the radio-to-optical data, using a preliminary version of the code that we are developing (Tammi et al., in prep.). A close-up of the fit in the radio regime is shown in Fig. 17. As Fig. 119 also shows, it is impossible to squeeze in another component between the strong radio-IR flare and the optical regime. The gamma-ray emission must come from the growing shock that we see both in the total flux density monitoring curve (Fig. 11) and in the SED (Fig. 119). Most of the theoretical model scenarios assume that gamma-rays originate close to the black hole and the accretion disk, well within the BLR and far upstream of the radio core. This is not a viable alternative for 2251+158, in which gamma-rays must come from the new shock component, parsecs downstream from the radio core. As Fig. 11 shows, at the time of the Planck observation the shock had already been growing for about half a year, which, changed into the source frame, translates into a distance L = D2c(1 + z)-1Δtobs (where D is the Doppler boosting factor, c the speed of light, z the redshift, and Δtobs the time elapsed from the onset of the flare), of at least 10 pc down the jet, assuming a modest value of D ≈ 10, or 90 pc using the value of D = 33.2 from Hovatta et al. (2009). This is the distance that the shock has moved downstream after emerging from the radio core.

7. The shape of the synchrotron spectrum: nonstandard acceleration mechanisms?

If the electron energy distribution is assumed to be a simple power-law with the form N(E) = KEs, the synchrotron spectrum of a single homogeneous source has an optically thick low-frequency part with a spectral index α =  +2.5, a turnover, and an optically thin part with a spectral index α related to the electron energy index by α = −(s − 1)/ 2. Energy losses eventually cause the spectrum to steepen to α − 0.5.

Since the earliest times of AGN research, the canonical value for α has been assumed to be about −0.75, corresponding to s  =  2.5. Such an electron energy index emerges naturally from the simplest first-order Fermi acceleration mechanism, and blazars also tend to show slopes around this value at the high-frequency (10 to 100 GHz) end of the radio spectrum, where the sources are typically assumed to be optically thin. However, Planck data now challenge this commonly accepted view. As Fig. 4 shows, the LF spectral slopes are, in the main, quite flat, and sometimes even rising. It is likely that in most cases this results from the superposition of several distinct synchrotron components, since the spectra are generally not smooth, and we also know from VLBI observations that there are nearly always several components (shocks) present in any source (see also Sect. 6). The HF spectra (Fig. 5), on the other hand, are often straight within the Planck flux density error bars, and give a definite impression of optically thin radiation. In some cases, such as 2251+158, the spectrum even shows a clear self-absorption turnover at high frequencies, followed by a flat optically thin part with α = −0.2 (Figs. 16 and 17).

The flatness of the HF spectra, shown in Fig. 5, is remarkable. Only a third of the sources have α steeper than −0.7, while 15 have α > −0.3. While the present data are not sufficient to exclude the possibility that multiple components produce the flatness, and some sources such as 0954+658 do show evidence of an additional IR component (see also, e.g., Raiteri et al. 1999), we consider such an explanation unlikely for the whole sample. The errors in αHF cannot explain the very flat spectral indices, since they generally are smaller than Δα ≈ 0.1.

Table 6

A list of the 10 sources scanned only once by Planck and with αHF > −0.5 + Δα.

In Table 6 we list the 10 sources that fullfill the following criteria. They have been scanned only once by Planck and they have αHF > −0.5 + Δα, where Δα is the HF spectral fit error shown in Figs. 18121. These form the most extreme end of the αHF distribution shown in Fig. 5. Three of them may show indications of multiple HF components in the spectra, but for the seven other sources the HF spectra appear to be straight within the Planck errors. If this is indeed the case, they cannot be explained with standard acceleration mechanisms with s ≥ 2.

The rise of a submillimetre flare is quite fast (see, for example, the JCMT monitoring data of Stevens et al. 1994); the self-absorption turnover of a fresh shock component passes through the Planck HFI frequencies rather rapidly, and one therefore sees it only rarely in the snapshot high frequency spectra. (In the present data only 2251+158 is clearly such a case, although 0235+164 and 1652+398 might be others.) Instead, in most cases we expect to see optically thin submillimetre spectra which have already experienced energy losses. Therefore the observed high-frequency spectral-index distribution of a sample of sources should show a range from αthin = −(s − 1)/ 2 to αthin − 0.5 and beyond, as the exponential steepening sets in, with spectral indices steeper than the initial αthin dominating. The observed distribution in Fig. 5 is thus incompatible with an initial αthin around −0.7, as we should then see a few values around α = −0.7 and a distribution peaking towards −1.2, reflecting the Δα = −0.5 steepening due to energy losses.

Instead, the spectral index distribution is compatible with an electron index s ≈ 1.5, resulting in a distribution of αthin having smallest values around −0.2 to −0.3 and a maximum around −0.7. Such a suggestion of a very hard original electron spectrum is not entirely new; Valtaoja et al. (1988) studied the shock spectra and concluded that α is around −0.2, and Hughes et al. (1991) also concluded that the flux density variations can best be modelled with very hard spectra. Gear et al. (1994) also found that submillimetre spectra in a sample of 48 sources tended to be flat. Indeed, the distribution they found (their Fig. 7) is quite similar to ours (Fig. 5).

As noted above, alternative explanations for the αHF distribution cannot be totally excluded. However, we consider sources such as 2251+158 (Figs. 16 and 17) to be compelling evidence for the existence of very hard non-standard electron spectra. The implications of this possibility are discussed below.

If the original αthin really is around −0.2, the distribution shown in Fig. 5 becomes easy to understand: in many slightly older HFI components we see the expected spectral steepening from −0.2 to −0.7 and beyond. One way to test this hypothesis is to try to estimate the ages of the highest-frequency spectral components, either from their turnover frequencies or from the total flux density monitoring: the older the component, the lower the turnover frequency and the steeper the spectra. However, this analysis is beyond the scope of this paper and reliable conclusions require more Planck data than presently exist.

An electron energy spectral index as hard as 1.5 has major implications for the acceleration mechanisms dominating in blazar shocks. Particle acceleration efficiency is strongly governed by the compression ratio of the flow, which in turn is confined by well-known plasma physics, preventing the spectral index from becoming arbitrarily small. For a relativistic shock the traditional first-order Fermi mechanism usually produces spectral indices of 2.2 or so (and not 2.5, which follows from the nonrelativistic theory), but not smaller. There are a few options, however, for bypassing the s ≈ 2.2 limit. For example, power-law spectra with s ≳ 1 have been found in certain kinds of oblique shocks (e.g., Kirk & Heavens 1989), as well as in parallel shocks in the case of large-angle particle scattering due to very strong turbulence (e.g., Stecker et al. 2007), or when the compression ratio felt by the particles is higher than that felt by the plasma (Ellison et al. 1990; Virtanen & Vainio 2005). The last alternative is an especially interesting one, because the combination of low density and relatively strong magnetic field – conditions likely to be found in Poynting-flux-dominated AGN jets – quickly leads to non-negligible Alfvén speeds and enhancement of the compression ratio felt by the particles and, consequently, to very hard particle spectra (Tammi 2008a,b, and references therein). High Alfvén speeds and scattering centres that are not frozen into the plasma also enable second-order Fermi acceleration, which can produce power-law spectra even with s < 1 in relativistic parallel shocks (Virtanen & Vainio 2005) and can work on time-scales comparable to the fastest blazar flares (Tammi & Duffy 2009).

As a simple order-of-magnitude reality check we can estimate the conditions required by the standard first-order acceleration in a parallel step shock in the presence of an increased scattering-centre compression ratio due to turbulence transmission (Vainio et al. 2003). Following the analysis of Tammi (2008b), we can estimate that in order to achieve s ≈ 1.5 in a shock with Lorentz factor Γ = 10, magnetic field strength of the order of 0.02−2 Gauss (depending on the composition and the density of the plasma) is needed. This is in good agreement with the usual modelling parameters of these sources, although we emphasise that the analysis used here is a simple one and may be limited to parallel shocks and weak turbulence.

8. Conclusions

We have presented the averaged Planck ERCSC spectra together with supporting ground and satellite observations obtained simultaneously with the Planck scans. The 104 SEDs, supplemented with archival data, demonstrate the usefulness of Planck data in determining the SED peak frequencies and fluxes and in modelling the SEDs in greater detail. In particular, the data demonstrate that the synchrotron spectrum contains contributions from several physically distinct AGN components: the LF and HF spectra are rarely smooth, except at the highest radio frequencies where the source component with the highest turnover becomes optically thin. The LF spectral indices cluster around α = 0, also indicating the superposition of many components with different turnover frequencies. In modelling the synchrotron and inverse-Compton SEDs, a multicomponent approach is therefore necessary. While physical modelling is beyond the scope of this paper, which uses averaged, rather than single-epoch, Planck ERCSC flux densities, we have shown some examples of multicomponent decomposition of the SEDs and the conclusions that can be drawn from the data.

The sources display remarkable variability, which must also be taken into account in modelling the SEDs. We have presented some examples of total flux density monitoring to demonstrate how sources have been observed during Planck scans in different stages of activity.

The Planck HFI spectra are remarkably flat, with only a minority of sources having spectral indices steeper than the “canonical” α = −0.7, and many apparently having αthin around −0.2 to −0.4. We suggest that the most likely interpretation for the very flat and straight synchrotron spectra observed by Planck is a very hard original electron spectral energy index, clearly below s = 2. While this possiblity must be confirmed by future data, an index s = 1.5 is compatible with the HF spectral index distribution seen in our sample. In a number of sources, energy losses steepen the HF spectra by Δα = 0.5 from about −0.3 to about −0.8, and further to −1 and beyond, which can explain the observed distribution. The hardness of the electron spectrum would also mean that the synchrotron SED peak is in most cases related to the maximum electron Lorentz factor achieved in the acceleration process. The hard spectra also require a rethinking of the acceleration processes dominating in the relativistic jets and shocks.

For some sources the HF spectra are flat or even rising, indicating the presence of multiple synchrotron-emitting components peaking in the gigahertz to terahertz regime, and ongoing flaring at still higher frequencies. However, in many cases the Planck HFI spectrum appears to join the optical spectrum smoothly, leaving little room for additional still higher-frequency components. In these cases, the inverse-Compton gamma-rays must originate in the synchrotron (i.e., shock) components seen in the Planck data.


1

Planck (http://www.esa.int/planck) is a project of the European Space Agency (ESA) with instruments provided by two scientific consortia funded by ESA member states (in particular the lead countries France and Italy), with contributions from NASA (USA) and telescope reflectors provided by a collaboration between ESA and a scientific consortium led and funded by Denmark.

7

The test statistic (TS) is defined as TS = −2ln(L0L1) with L0 the likelihood of the null-hypothesis model as compared to the likelihood of a competitive model, L1.

Acknowledgments

The Planck Collaboration acknowledges the support of: ESA; CNES and CNRS/INSU-IN2P3-INP (France); ASI, CNR, and INAF (Italy); NASA and DoE (USA); STFC and UKSA (UK); CSIC, MICINN and JA (Spain); Tekes, AoF and CSC (Finland); DLR and MPG (Germany); CSA (Canada); DTU Space (Denmark); SER/SSO (Switzerland); RCN (Norway); SFI (Ireland); FCT/MCTES (Portugal); and DEISA (EU). A description of the Planck Collaboration and a list of its members, indicating which technical or scientific activities they have been involved in, can be found via http://www.rssd.esa.int/Planck. The Metsähovi and Tuorla observing projects are supported by the Academy of Finland (grant numbers 212656, 210338, 121148, 127740 and 122352). UMRAO is supported by a series of grants from the NSF and NASA, and by the University of Michigan. This publication is partly based on data acquired with the Atacama Pathfinder Experiment (APEX). APEX is a collaboration between the Max-Planck-Institut für Radioastronomie, the European Southern Observatory, and the Onsala Space Observatory. This research is partly based on observations with the 100-m telescope of the MPIfR (Max-Planck-Institut für Radioastronomie) at Effelsberg, the IRAM 30-m telescope, and the Medicina (Noto) telescope operated by INAF – Istituto di Radioastronomia. This paper makes use of observations obtained at the Very Large Array (VLA) which is an instrument of the National Radio Astronomy Observatory (NRAO). The NRAO is a facility of the National Science Foundation operated under cooperative agreement by Associated Universities, Inc. The observations at Xinglong station are supported by the Chinese National Natural Science Foundation grants 10633020, 10778714, and 11073032, and by the National Basic Research Program of China (973 Program) No. 2007CB815403. The OVRO 40-m monitoring program is supported in part by NASA. The Australia Telescope is funded by the Commonwealth of Australia for operation as a National Facility managed by CSIRO. The Fermi LAT Collaboration acknowledges generous ongoing support from a number of agencies and institutes that have supported both the development and the operation of the LAT as well as scientific data analysis. These include the National Aeronautics and Space Administration and the Department of Energy in the United States, the Commissariat à l’Energie Atomique and the Centre National de la Recherche Scientifique/Institut National de Physique Nucléaire et de Physique des Particules in France, the Agenzia Spaziale Italiana and the Istituto Nazionale di Fisica Nucleare in Italy, the Ministry of Education, Culture, Sports, Science and Technology (MEXT), High Energy Accelerator Research Organization (KEK) and Japan Aerospace Exploration Agency (JAXA) in Japan, and the K. A. Wallenberg Foundation, the Swedish Research Council and the Swedish National Space Board in Sweden. Additional support for science analysis during the operations phase is gratefully acknowledged from the Istituto Nazionale di Astrofisica in Italy and the Centre National d’Études Spatiales in France. Part of this work is based on archival data, software or on-line services provided by the ASI Science Data Center ASDC. We thank the Fermi LAT team reviewers, S. Ciprini and M. Giroletti, for their effort and valuable comments.

References

Online material

Table 1

The complete 1 Jy northern sample of AGN. Column 8 refers to the 37 GHz observations of Metsähovi Radio Observatory.

Table 4

Simultaneity of high energy data to Planck observations displayed in the SEDs.

thumbnail Fig.18

Left-panel: the SED of the source 0003−066. Grey circles show the historical data. The red circles show data simultaneous to the Planck observations. The dotted and dashed lines show the second and third degree polynomials, respectively, fitted to the synchrotron and IC bumps in the SED. Right-panel: the radio spectrum of 0003−066. Red circles, LF data simultaneous to Planck; red stars, ERCSC LFI data; blue circles, HF data simultaneous to Planck; blue stars, ERCSC HFI data. The dashed and dotted lines are fits to simultaneous LF and HF data.

Open with DEXTER

thumbnail Fig.23

J0125−0005.

Open with DEXTER

thumbnail Fig.66

TEX1040+244.

Open with DEXTER

thumbnail Fig.99

J184915+670.

Open with DEXTER

All Tables

Table 2

Optical and radio observatories that participated in the Planck multifrequency campaigns.

Table 3

Parameters for the RATAN-600 receivers.

Table 5

Average spectral indices for AGN subclasses.

Table 6

A list of the 10 sources scanned only once by Planck and with αHF > −0.5 + Δα.

Table 1

The complete 1 Jy northern sample of AGN. Column 8 refers to the 37 GHz observations of Metsähovi Radio Observatory.

Table 4

Simultaneity of high energy data to Planck observations displayed in the SEDs.

All Figures

thumbnail Fig.1

The distributions of the synchrotron (top panel) and IC (bottom panel) peak frequencies.

Open with DEXTER
In the text
thumbnail Fig.2

The correlation of the peak frequencies of the synchrotron and the IC components. Solid circles denote class A fits, and open circles class B fits.

Open with DEXTER
In the text
thumbnail Fig.3

Comparison between our synchrotron and IC peak frequencies and those of Abdo et al. (2010a). The synchrotron peak frequencies are shown in the top panel and IC peak frequencies in the bottom panel. The dashed lines illustrate the one-to-one correspondence. Solid circles denote class A fits, and open circles class B fits.

Open with DEXTER
In the text
thumbnail Fig.4

Distribution of LF spectral indices for the whole sample (104 sources). The sources that have only been scanned once by Planck are shown hatched (46 sources).

Open with DEXTER
In the text
thumbnail Fig.5

Distribution of HF spectral indices for the 84 sources that had three or more data points. The sources that have only been scanned once by Planck are shown hatched (40 sources).

Open with DEXTER
In the text
thumbnail Fig.6

LF spectral indices αLF versus HF spectral indices αHF for the various AGN subclasses.

Open with DEXTER
In the text
thumbnail Fig.7

The 37 GHz lightcurve of 0234+285, measured at Metsähovi Radio Observatory, showing continuous variability. The period when Planck was observing the source is indicated by the hatched region in the lower panel, which is a blow-up of the last year.

Open with DEXTER
In the text
thumbnail Fig.8

The 37 GHz lightcurve of 0235+164, details as in Fig. 7.

Open with DEXTER
In the text
thumbnail Fig.9

The 37 GHz lightcurve of 1253−055 (3C279), details as in Fig. 7.

Open with DEXTER
In the text
thumbnail Fig.10

The 37 GHz lightcurve of 1413+135, details as in Fig. 7.

Open with DEXTER
In the text
thumbnail Fig.11

The 37 GHz lightcurve of 2251+158 (3C454.3), details as in Fig. 7.

Open with DEXTER
In the text
thumbnail Fig.12

A possible single-component spectrum for 0234+285 to illustrate the general shape. Red circles, LF data simultaneous to Planck; blue circles, HF data simultaneous to Planck.

Open with DEXTER
In the text
thumbnail Fig.13

A possible multiple-component spectrum for 0235+164 to illustrate the general shape. Symbols as in Fig. 12.

Open with DEXTER
In the text
thumbnail Fig.14

The radio spectrum of 1253−055 shows that there is no room for another component between radio and optical regimes. Symbols as in Fig. 12.

Open with DEXTER
In the text
thumbnail Fig.15

The radio spectrum of 1413+135 is very flat and can be fitted with multiple components. Symbols as in Fig. 12.

Open with DEXTER
In the text
thumbnail Fig.16

The radio-optical SED of 2251+158 (3C454.3) fitted with one component and synchrotron emission from the jet (dashed line) using a model by Tammi et al. (in prep.). The radio spectrum is shown in Fig. 17.

Open with DEXTER
In the text
thumbnail Fig.17

A close-up of the radio spectrum and model fit of 2251+158 (3C454.3), details as in Fig. 16.

Open with DEXTER
In the text
thumbnail Fig.18

Left-panel: the SED of the source 0003−066. Grey circles show the historical data. The red circles show data simultaneous to the Planck observations. The dotted and dashed lines show the second and third degree polynomials, respectively, fitted to the synchrotron and IC bumps in the SED. Right-panel: the radio spectrum of 0003−066. Red circles, LF data simultaneous to Planck; red stars, ERCSC LFI data; blue circles, HF data simultaneous to Planck; blue stars, ERCSC HFI data. The dashed and dotted lines are fits to simultaneous LF and HF data.

Open with DEXTER
In the text

Current usage metrics show cumulative count of Article Views (full-text article views including HTML views, PDF and ePub downloads, according to the available data) and Abstracts Views on Vision4Press platform.

Data correspond to usage on the plateform after 2015. The current usage metrics is available 48-96 hours after online publication and is updated daily on week days.

Initial download of the metrics may take a while.