Free Access
Issue
A&A
Volume 627, July 2019
Article Number A72
Number of page(s) 10
Section Extragalactic astronomy
DOI https://doi.org/10.1051/0004-6361/201935750
Published online 03 July 2019

© ESO 2019

1. Introduction

Blazars are radio-loud AGNs whose relativistic jet points directly at us, that is, with a viewing angle θv ≲ 1/Γ with respect to the jet axis, where Γ is the jet bulk Lorentz factor. The jet emission is greatly boosted by relativistic beaming, making blazars highly visible even at high cosmic distances.

The beamed nonthermal spectral energy distribution (SED) of powerful blazars is characterized by two broad distinctive humps. Most of the electromagnetic output of very powerful blazars is in the megaelectronvolt band, in the exact position for which there is no sensitive instrument to make observations. We can detect blazars however in the adjacent bands, through Fermi/LAT (> 100 MeV) or in the hard X-rays, through INTEGRAL, Swift/BAT and NuSTAR. Only NuSTAR has the spectral resolution (through pointed observations) to accurately find, together with the LAT data (detections and upper limits), the peak frequency and luminosity of the blazar emission. We claimed (Ghisellini et al. 2010, hereafter G10) that the trend of lower intrinsic peak frequency with larger luminosity observed in blazars of low and intermediate power continues to be valid also at the extremely high-power end of the population. This was based on blazars detected by BAT, but not by LAT. Instead, considering blazars detected by both instruments, Ajello et al. (2009) claimed that no trend was visible. In addition to this controversial intrinsic property, the K-correction nevertheless favors the detection in the hard X-ray band of blazars at high redshifts. Therefore, the most powerful persistent objects of the Universe should be found in the hard X-ray band. Looking for these extreme objects, we proposed to use NuSTAR to observe a few blazars at z >  2 that have already been detected by Fermi/LAT, but not by Swift/BAT, hoping to shed light on the intrinsic properties of these sources, and in particular on the possible relation between the peak frequency of the high-energy component of the SED and its luminosity.

Another key question in modern cosmology pertains to how supermassive black holes (SMBH) gained most of their mass, especially at the highest redshifts probed by current observations. Most high-z searches of SMBHs concern radio-quiet objects, but a very promising alternative approach concerns radio-loud ones, and specifically blazars. Beaming makes blazars a unique tool in assessing the number density of radio-loud SMBHs at high redshift. In fact, for any confirmed high-redshift blazar there must exist other 2Γ2 = 450(Γ/15)2 sources sharing the same intrinsic properties, but whose jets are not pointing at us. Some SMBHs with masses in excess of 109M were already in place when the Universe was only ≃700 Myrs old (e.g., ULAS J1120+0641 at z = 7.08, Mortlock et al. 2011; ULAS J1342+0928 at z = 7.5, Bañados et al. 2018). Their very existence is difficult to reconcile with black hole growth at the Eddington rate starting from stellar-sized seeds (e.g., Volonteri et al. 2010).

To the three blazars observed for the first time by NuSTAR, we have added all other blazars with z >  2 observed by NuSTAR in order to better understand their common properties. We show that all of them belong to the group of the most powerful blazars both in their jet and in their accretion disk properties, fully confirming the fact that the jet power is proportional to the accretion luminosity and our expectations that the hard X-ray selection of high-redshift blazars picks up the most powerful sources.

We use a flat ΛCDM cosmology with h = ΩΛ = 0.7.

2. Data analysis

Table 1 lists the three blazars observed by NuSTAR, selected among all blazars at z >  2 already detected by Fermi/LAT (Atwood et al. 2009) with a [0.3–10 keV] flux larger than 10−12 erg cm−2 s−1 and not already observed by NuSTAR, or by Swift/BAT. This table also reports the redshift, the flux at 5 GHz, the optical magnitude in the R band, and the estimate of the black hole mass obtained through the virial estimate, when available.

Table 1.

Selected targets.

2.1. NuSTAR

The NuSTAR satellite (Harrison et al. 2013) observed PKS 0123+25 on 2018 January 03 (obsID 60367001002), PKS 0227–369 on 2017 August 10 (obsID 60367002002), and TXS 0458–020 on 2018 April 26 (obsID 60367003001). The total net exposure times were 19.9, 23.3, and 20.7 ks, respectively.

The Focal Plane Modules A and B (FPMA and FPMB) data sets were processed with the NuSTARDAS software package (v.1.8.0) developed by the Space Science Data Center of the Italian Space Agency (SSDC-ASI, Italy) in collaboration with the California Institute of Technology (Caltech, USA). Calibrated and cleaned event files were produced with the nupipeline task using the version 20170705 of the NuSTAR Calibration Database (CALDB).

The three sources were all well detected above the background by the two NuSTAR hard X-ray telescopes up to 30 keV. The FPMA and FPMB energy spectra of the three sources were extracted from the cleaned and calibrated event files using a circular spatial region with a radius of 12 pixels (∼30 arcseconds) centered on the target, while the background was extracted from nearby circular regions of 50 pixel radius. The ancillary response files were generated with the nuproducts task, applying corrections for the point spread function (PSF) losses, exposure maps, and telescope vignetting.

For all three observations the spectral analysis of the NuSTAR data was performed using the XSPEC package adopting a single power-law model with an absorption hydrogen-equivalent column density fixed to the Galactic values given by Kalberla et al. (2005), that is, NH = 6.8 × 1020 cm−2 for PKS 0123+25, NH = 2.4 × 1020 cm−2 for PKS 0227–369 and NH = 6.0 × 1020 cm−2 for TXS 0458–02. All spectra were binned to ensure a minimum of 30 counts per bin and energy channels below 3.0 keV and above 30.0 keV were excluded. A multiplicative constant factor was included to take into account cross-calibration uncertainties between the two telescopes (NuSTAR FPMA and FPMB). We found that this model fitted the spectral data very well for all three sources in the considered energy band. The results of the spectral fits are given in Table 2.

Table 2.

Parameters of the X-ray spectral analysis of the NuSTAR data.

2.2. Swift/XRT

The Neil Gehrels Swift Observatory (Gehrels et al. 2004) observed PKS 0123+25 with the X-ray Telescope (XRT, Burrows et al. 2005) simultaneously with NuSTAR, namely on 2018 January 03 and January 4 (obsIDs 00088100001, 00088100002), for a total net exposure time of 2.0 ks.

The XRT observations were carried out with the Photon Counting (PC) readout mode. The XRT data were first processed using the XRT Data Analysis Software (XRTDAS, v.3.4.1), which was developed by SSDC-ASI. Standard calibration and cleaning processing steps were applied using the xrtpipeline software module and using version 20180710 of the Swift/XRT Calibration Database (CALDB).

Source events for the spectral analysis were extracted in the 0.3–10 keV energy band using a circular spatial extraction region with a radius of 20 pixels (∼47 arcseconds). The background was estimated using a nearby source-free circular region with a radius of 50 pixels. Corrections to the ancillary response files for PSF losses, CCD defects, and telescope vignetting were calculated and applied using the xrtmkarf software module.

For the spectral analysis the energy spectrum was grouped to ensure at least 20 counts in each bin. We adopted an emission model described by a single power law with an absorption hydrogen-equivalent column density fixed to the Galactic value of NH = 6.8 × 1020 cm−2 (Kalberla et al. 2005). The results of the spectral fit were found to be consistent in slope and normalization with those derived from the NuSTAR observation, thus extending the observed spectral slope down to 0.3 keV, with a best-fit photon index of .

For the two blazars PKS 0227–369 and TXS 0458–020 no simultaneous observations with NuSTAR were carried out by the Neil Gehrels Swift Observatory.

2.3. Fermi/LAT

We analyzed the Fermi/LAT data around the NuSTAR pointings using the Pass-8 data version and the public Fermi Science Tools version v11r5p3.

First we looked for nearly simultaneous data, with several choices of exposure time, until we derived a detection. The blazar TXS 0458–02 was in a bright state, and an integration time of just 2 days (±1 day around the NuSTAR pointing) was enough for a detection of ∼11σ. The other two objects instead require years of integration for a detection. We therefore considered two exposures, a short one of 30 days (±15 days around the NuSTAR pointing) to derive a meaningful upper limit at the same epoch as NuSTAR, and a long one of years, in order to measure the average spectrum. The long exposure is 4 years for PKS 0123+25 (from May 24, 2014 to May 24, 2018) and 2 years for PKS 0227–369 (from May 24, 2016 to May 24). The results are reported in Table 3.

Table 3.

Parameters of the power-law fits to the Fermi/LAT data.

Gamma-ray events were selected from a region of interest (ROI) of 15° using standard quality criteria, as recommended by the Fermi Science Support Center (FSSC). We performed the likelihood analysis in two steps. In the first step the XML model included all the sources in the preliminary LAT 8 year point-source list (FL8Y). We then performed a second likelihood fit using the XML model from the first step, optimized by dropping all sources with a TS <  1. The analysis was performed with the NEWMINUIT optimizer, using an unbinned likelihood for the short datasets and a binned likelihood for the long exposures, with bins of 0.1°. Furthermore, each decade of energy was split in 10 bins.

The LAT data points for the SED were obtained by binning the spectrum with two bins per decade in energy, in the 0.1–100 GeV range, and performing a likelihood analysis in each single energy bin. In the XML model all parameters were kept fixed to the best-fit values, except for the normalization of the target and of the two backgrounds (isotropic and Galactic). A binned or unbinned likelihood was used if the total number of counts in the bin was higher or lower than 15 000, respectively. A Bayesian upper limit was calculated if in that bin the target had a TS <  9 or npred <  3. The light curves were obtained by performing an unbinned likelihood analysis in each time bin of 7 days, leaving the parameters of the brightest or variable FL8Y sources in the ROI to vary freely within an 8° radius of the target.

3. Modeling

We interpret the overall SEDs of our sources with a leptonic, one-zone jet emission model plus the contribution from an accretion disk, its X-ray corona, and a molecular torus, that is absorbing and re-emitting in the infrared a fraction of the disk radiation. The details of the model are given in Ghisellini & Tavecchio (2009, 2015) and here we summarize its main features.

– The emitting region producing the non-thermal radiation is assumed to be spherical, with radius R and at a distance Rdiss from the central black hole. The jet is assumed conical, with semi-aperture angle ψ. Although ψΓ ∼ 1 is borne out by numerical simulations of jet acceleration, jets could have a parabolic shape while accelerating, becoming conical when coasting (e.g., Marscher 1980; Komissarov et al. 2007). They could also re-collimate at large distances, making the relation between the transverse radius r and the distance Rdiss uncertain. We assume, for simplicity, ψ = 0.1, corresponding to 5.7° and ψ ≈ 1/Γ. The emitting plasma is assumed to move with a bulk motion of velocity βc and Lorentz factor Γ at a viewing angle θv from the line of sight. The Doppler factor is δ = 1/[Γ(1 − βcosθv)].

– Throughout the emitting region relativistic electrons are continuously injected at a rate Q(γ) [cm−3 s−1] for a time equal to the light-crossing time R/c. The shape of Q(γ) is assumed to be a smoothly broken power law with a break at γb:

(1)

– The power injected in the form of relativistic electrons is

(2)

This is calculated in the comoving frame. We solve the continuity equation to find the energy distribution N(γ) [cm−3] of the emitting particles at the particular time R/c, when the injection process is assumed to end. We account for synchrotron and inverse Compton cooling and e± pair production and reprocessing; although in our sources, e± pairs are not important.

– The magnetic field B is tangled and uniform throughout the emitting region.

– There are several sources of radiation external to the jet:

1. The broad line region (BLR) is assumed to re-emit 10% of the accretion luminosity from a shell-like distribution of clouds located at a distance cm;

2. The IR emission from a dusty torus, located at a distance cm;

3. The direct emission from the accretion disk, including its X-ray corona;

4. The starlight contribution from the inner region of the host galaxy, and the cosmic background radiation.

All these contributions are evaluated in the blob comoving frame, where we calculate the corresponding inverse Compton radiation from all these contributions, and then transform this into the observer frame.

– The numerical code we use is not time dependent: it gives a “snapshot” of the predicted SED at the time R/c, when the particle distribution N(γ) and consequently the produced flux are at their maximum.

– For powerful sources, the radiative cooling is efficient and the cooling timescale can be shorter than R/c even for low-energy particles. This implies that γpeak, the random Lorentz factor of the electron emitting most of the radiation, is close to γb.

– The size of the emitting region is rather compact, as indicated by the short variability timescales observed in blazars. As a consequence, the synchrotron flux is self-absorbed at high frequencies, in the submillimeter band. Therefore the model cannot account for the radio emission at lower frequencies, which must be produced by more extended regions of the jet.

– To calculate the flux produced by the accretion disk, we adopt a standard Shakura & Sunyaev (1973) disk (see Ghisellini & Tavecchio 2009). This model depends mainly on the accretion rate (regulating the total disk luminosity) and on black hole mass (regulating the location of the peak of the emission). This allows us to also fit the thermal radiation seen in the optical-UV range, and to estimate the accretion rate and the black hole mass.

– The disk luminosity is independent of the adopted accretion model (e.g., standard Shakura & Sunyaev, with zero spin, or an accretion disk around a Kerr black hole). Instead the estimate of the mass does depend on the assumed accretion model (see e.g., Calderone et al. 2013; see also Campitiello et al. 2018 who studied how the black hole spin and the special and general relativistic effects impact on the determination of the mass of the black hole).

– The total jet power is the sum of the power carried by particles (we assumed one cold proton per emitting electron), magnetic field, and radiation. Therefore, the estimate of the magnetic and particle power is model dependent because the particle number and the value of the magnetic field depend on which model we are using to interpret the data (leptonic or hadronic, molti or one-zone, and so on). This is calculated at the dissipation region, through

(3)

where the subscript “i” can stand for protons, electrons, magnetic field, or radiation, and U is the corresponding energy density, as calculated in the comoving frame. The power in radiation is instead model independent. It can be calculated with the equation above, that can be re-written as (for viewing angles θv ∼ 1/Γ):

(4)

where is the bolometric observed luminosity produced by the jet. This is an observable. Therefore, only the knowledge of Γ enters this estimate. This makes Pr almost model independent. It is a lower limit of the jet power. Pjet is the sum of the different components.

– The uniqueness of the parameter values was discussed in some detail in Ghisellini & Tavecchio (2015). We stressed there that in the framework of our leptonic, one-zone model, it is possible to find a unique solution for fitting the SED, but only if the data are of sufficient quality. One would need simultaneous data from the millimeter to the γ-rays, and this is possible only in a few cases. We are therefore constrained to assume that the nonsimultaneous data we have collected are a reasonably good representation of the SED. We tried to constrain the γ-ray flux and slope as best we could by analyzing the Fermi/LAT data as close as possible to the NuSTAR observations. In addition, when possible we compare the resulting SED with the SED corresponding to other states of the sources in search of the possible causes of variations.

4. Results

We show the overall SEDs of the three blazars analyzed in this paper in Figs. 1, 4, and 6. The SEDs of PKS 0123+25 and PKS 0227–369 show the presence of a thermal component at optical-UV frequencies, that we interpret as being due to a standard accretion disk. Perhaps more surprising, this thermal emission is not clearly visible in TXS 0458–02, most probably because it is hidden by the dominating synchrotron spectrum. Besides showing our data, the figures report the archival data from the ASI/SSDC database1.

thumbnail Fig. 1.

Overall SEDs of PKS 0123+25. Besides our data (red points), we show the archival data collected from the ASI/SSDC database. We have indicated in green the XMM-Newton data, taken in January 2009, and in blue the Swift/XRT data taken simultaneously with the NuSTAR observation. The blue arrows in the γ-ray band are upper limits obtained integrating over 30 days (15 days before and 15 days after the NuSTAR observation). Red γ-ray points and arrows correspond to the average flux during the last 4 years. The lines are the result of the modeling (see text).

Open with DEXTER

4.1. PKS 0123+25

The NuSTAR data of this source lie on the extrapolation of the lower-energy X-ray data taken by XMM-Newton January 8, 2009, and the Swift/XRT data taken simultaneously with NuSTAR. Integrating the Fermi/LAT data 15 days before and 15 day after the NuSTAR observation, the source was not detected. The corresponding 95% upper limits are shown in Fig. 1 together with the Fermi/LAT spectrum integrating over the last 4 years. The upper limits are consistent with the spectrum obtained with the long exposure, indicating no flares during the NuSTAR observation.

The optical spectrum can be well fitted by a standard accretion-disk model, and we find a black hole mass of M = 1.5 × 109M and a disk luminosity of Ld = 5.85 × 1046 erg s−1, corresponding to 30% of the Eddington luminosity. This value agrees with the observed broad-line luminosities, as observed by the SDSS spectrum (DR13). We used the template of Francis et al. (1991) and assumed that Ld = 10LBLR. In this way we derived LBLR = 1046 erg s−1 (using the CIV line); LBLR = 7.3 × 1045 erg s−1 (CIII] line) and LBLR = 1.6 × 1045 erg s−1 (MgII line). The contribution of both the torus and the jet emission can be found in the infrared band. In order to disentangle the two, we have assumed that the time-averaged γ-ray spectrum is indicative of the high-energy emission during the NuSTAR observation. In Fig. 2 we show the model SED assuming there is no torus: if we fit the high-energy SED, we under-reproduce the near-IR. We therefore assume that the near-IR flux is produced by the torus, and this helps us to find the peak of the high-energy SED and its dominance with respect to the synchrotron component. This information helps to constrain the magnetic field and γpeak allowing us to find a robust solution for the model parameters (assuming that the archival data are indicative of the real SED). Figure 3 compares the models assuming two different values for the aperture angle of the jet: ψ = 0.1 = 5.7° (blue lines) and ψ = 0.023 = 1.3° (red lines). The latter value corresponds to the average value of Fermi/LAT blazars derived by Pushkarev et al. (2017). Both models represent the data well, and are indistinguishable. The model with the smaller ψ requires a larger Rdiss (factor 3) and a larger jet power (factor 3). For homogeneity with the blazars fitted previously, in the rest of the paper we use ψ = 0.1. The parameters are listed in Table 4.

thumbnail Fig. 2.

Lines resulting from the modeling assuming that there is no torus, and assuming both a small Rdiss(=2.25 × 1017 cm) and a large Rdiss(=3.6 × 1018 cm). Parameters are listed in Table 4. If we fit the high-energy emission, the model underproduces the near-IR flux. The long dashed line corresponds to the first- and the second-order Compton SSC contributions.

Open with DEXTER

thumbnail Fig. 3.

Comparison between the best models assuming ψ = 0.1 = 5.7° and ψ = 0.023 = 1.3°, as labeled. The long-dashed lines are the SSC contribution. Parameters are listed in Table 4.

Open with DEXTER

Table 4.

Parameters for the models shown in Figs. 1, 2, 3, 4, and 6.

Table 4.

continued.

4.2. PKS 0227–369

The X-ray flux was significantly lower during the NuSTAR observations with respect to an earlier Swift/XRT observation carried out in November 2008 (Ghisellini et al. 2009a,b). The shown γ-ray data (red symbols) refer to the last 2 years, and indicate a low state both with respect to the archival data and to an older flaring state. The slopes of both the X-ray and the γ-ray data are instead the same as the ones derived by the archival data. Unfortunately, during the NuSTAR observations, the source was not observed by Swift, meaning that we cannot check if any change occurred also in the optical-UV bands. However, we do not expect any strong flux variability in these bands, since they are produced by the accretion disk, whose emission is usually much more stable than that of the jet. Applying our standard disk model we derive M = 2 × 109M and Ld = 1.8 × 1046 erg s−1, corresponding to 7% of the Eddington luminosity. We did not find any published optical spectra reporting the luminosity of the broad lines. However, the disk emission is clearly visible in this source and the accretion disk luminosity we found is therefore reliable. As in PKS 0123+25, the IR flux is dominated by the jet synchrotron emission. As a consequence, the torus component is somewhat uncertain: in Fig. 4 we show a torus reprocessing half of the disk luminosity.

thumbnail Fig. 4.

Overall SEDs of PKS 0227–369. The X-ray flux was significantly lower during the NuSTAR observations with respect to an earlier Swift/XRT+UVOT observation carried out in November 2008. The red points in the Fermi/LAT band correspond to integrating the last two years of observations. This shows that the source was in a low state during this period of time.

Open with DEXTER

To model the source, we assumed that the radio-to-optical archival data give a good representation of the SED in this frequency range, and we tried to explain the change of the SED by changing the minimum number of parameters.

We find that the observed variability can be explained by changing the power of the relativistic electrons injected throughout the source that are responsible for the emission. The models shown differ by a factor of four in Pinj. Furthermore, the lower NuSTAR state is characterized by a slightly larger dissipation region, with a slightly smaller magnetic field and a larger value of the energy of the electrons emitting at the peaks of the SED. The total jet power is a factor three smaller than in the high state.

4.3. TXS 0458–020

Figure 5 reports the Fermi/LAT light curve of the last 3 years, in order to show the variable behavior of this source. The dashed vertical line indicates the day of the NuSTAR observation.

thumbnail Fig. 5.

γ-ray light curve of TXS 0458–02. Blue triangles are 95% upper limits, calculated assuming a power law with photon spectral index Γ = 2. The dashed vertical line corresponds to the NuSTAR observation epoch, when the source was in a very high γ-ray state.

Open with DEXTER

Figure 6 shows the overall SED of the source, which is characterized by a relatively harder γ-ray spectrum with respect to the other two sources, as suggested by the nearly simultaneous Fermi/LAT data (red points). In this case the flux was high enough to allow the detection and some spectral determination integrating for one week around the NuSTAR observation.

thumbnail Fig. 6.

Overall SEDs of TXS 0458–02, showing the changes in the high-energy emission due to its strong variability. Since unfortunately there are no low-frequency (mm-optical) data simultaneous to the varying high-energy flux, the models shown assume a quasi-constant flux at these frequencies. This illustrates how the model parameters would change in this case.

Open with DEXTER

Since the synchrotron jet emission hides the accretion disk component, we cannot directly fit the disk. We can derive a rough estimate of the accretion disk luminosity by the observation of the broad lines, which are seen in this source even if the continuum is dominated by the synchrotron emission. The CIV broad line has a flux FCIV = 2.6 × 10−15 erg cm−2 s−1, corresponding to a luminosity of LCIV = 1.1 × 1044 erg s−1. According to the template of Francis et al. (1991) this should correspond to a BLR total luminosity of LBLR = 9.7 × 1044 erg s−1 and to a disk luminosity ten times larger: Ld ∼ 1046 erg s−1.

For the black hole mass, we must consider that smaller masses, for a given Ld, correspond to a disk spectrum peaking at larger frequencies. Therefore we can derive a lower limit to the black hole mass requiring that the disk emission does not over-contribute to the optical-UV flux. We obtain an upper limit to the mass, requiring that the disk that is emitting is geometrically thin and optically thick, and therefore has a luminosity larger than 0.01LEdd. We chose Ld = 0.1LEdd for Ld ∼ 2 × 1046 erg s−1, deriving M = 8 × 108M. These values are only indicative, and uncertain by at least a factor of two.

To explain the observed different states, we assumed that the archival data are representative of the quiescent state, while during the NuSTAR observation the source was in a high state. In March 2014 there was a Fermi/LAT flare almost brighter than in 2018, but unfortunately with no other observations at other frequencies. We show a possible fit for this flare, but only to illustrate the change of the parameters if the source were ever to resemble the proposed theoretical SED.

As usual, we look for a solution involving the smallest change of the minimum number of parameters to explain the observed variability. For the “NuSTAR state” the power injected in relativistic electrons is 4 times larger than in the quiescent state, but the magnetic field is ∼2.5 times smaller. The slopes of the injected electron distribution are slightly harder and the total jet power in the NuSTAR state is twice as much as in quiescence. The “high” state would require more power in the injected electrons (more than ten times that in quiescence) and a still-smaller magnetic field, and the total jet power would be approximately 13 times larger. All these estimates are calculated assuming that the synchrotron part of the spectrum is well represented by the quiescent state, in turn shown by the archival data. This source was studied also in Ghisellini et al. (2011), where simultaneous Swift (UVOT and XRT) and Fermi/LAT observations are reported. They correspond to the black symbols in Fig. 6.

Recently, Lister et al. (2016) measured the apparent speed of a superluminal knot in this source, deriving an apparent speed βapp ∼ 6. Although this is a lower limit to the value of the bulk Lorentz factor, and therefore consistent with the values used in Fig. 6, it is interesting to compare these models with the one using a smaller value of Γ. This is done in Fig. 7, which compares the models with Γ = 14 and Γ = 7, as labeled. The latter slightly underestimates the NuSTAR data, but can accurately reproduce the rest of the SED. The parameters listed in Table 4 indicate (for the Γ = 7 case) that the jet power and the magnetic field are slightly smaller, and the electron energies are larger. Overall, we note that the parameters are not vastly different.

thumbnail Fig. 7.

Comparison of the models adopting Γ = 14 and Γ = 7, as labeled. Parameters are listed in Table 4. The model with Γ = 7 slightly underestimates the NuSTAR data.

Open with DEXTER

5. Discussion

5.1. Comparison with other z >  2 NuSTAR blazars

Table 5 reports the list of all blazars at z >  2 observed by NuSTAR; there are 11 sources. The table reports their redshift and the reference to the papers discussing the NuSTAR X-ray data. All 11 sources are FSRQs, and their SEDs are shown in Fig. 8, in the νLν versus ν (rest frame) representation. In this way we can compare the rest frame SED of the sources. Most of the data come from archives (mostly ASI/SSDC) and the figure shows how similar the sources are in the radio–millimeter band, while they become different (and varying with a very large amplitude) at greater frequencies. We note the source S5 0014+813, the most luminous in the optical-UV, due to its extraordinary luminous accretion disk (Ghisellini et al. 2009a,b), and S5 0836+710, the most luminous in X-rays and γ-rays, where it reached a luminosity of ∼1050 erg s−1 during a flare observed on August 2, 2015 (Ciprini 2015).

Table 5.

Entire sample of z >  2 blazars observed by NuSTAR.

thumbnail Fig. 8.

SED of all 11 blazars at z >  2 observed so far by NuSTAR. It can be noted that (1) the synchrotron hump is remarkably similar; (2) for several sources the accretion disk sticks out in the optical-UV band; (3) 0014+813 has an exceptionally powerful accretion disk; and (4) the X-ray and γ-ray emission is more dispersed and variable.

Open with DEXTER

The reason for the smaller dispersion of data points in the radio with respect to the other wavelengths is probably the lower amplitude variability in the radio band. Another reason for having less dispersion in the radio–millimeter band is that the Doppler amplification of the synchrotron flux scales as F(ν)∝δ3 + α ∼ δ3 (for flat spectral indices α = 0), while the amplification factor for the inverse Compton process, with photons produced externally to the jet, scales as F(ν)∝δ4 + 2α ∼ δ5 (for X-ray spectral indices α ∼ 0.5) as pointed out by Dermer (1995) and illustrated in Fig. 5 of Ghisellini (2015).

We note that all sources show no sign of changing slope at the lowest radio frequencies, an indication that the jet emission is extremely strong and hides any contribution of the extended radio structure, which should have a steep (i.e., increasing at lower frequencies) spectrum. On the other hand, for almost all sources we do see the contribution of the accretion disk in the optical-UV. The accompanying X-ray coronal emission is absent in these sources, completely overwhelmed by the beamed X-rays from the jet. As a consequence, there is no sign of the presence of the iron fluorescence line at 6.4 keV (rest frame) for any of the sources.

The hardness of the X-ray spectrum coupled with the steepness of the γ-ray one indicates a spectral peak around ∼10 MeV. We can try to be more precise by extrapolating the X-ray and γ-ray spectra of each source and find out the matching frequency. The result is shown in Fig. 9: the γ-ray luminosity Lγ is plotted against the peak frequency. For Lγ we chose an average state, not the extreme flaring state. It should be borne in mind that this result can be affected by systematic errors, since the spectral shape around the peak is likely to be curved and not accurately described by a broken power law. The figure in any case suggests a trend (smaller νpeak for larger Lγ) and an outlier (TXS 0458–020 in the high state).

thumbnail Fig. 9.

Peak luminosity of the high-energy component as a function of its peak frequency. The dashed line connects three different states of TXS 0458–020. Error bars correspond to factor 3 uncertainties in νpeak and factor 2 in Lγ. There is a weak trend of smaller luminosities for larger peak frequencies, with the exception of TXS 0458–020 when in the high state.

Open with DEXTER

5.2. Seed photons from the BLR or the torus?

The peak frequency νC of the high-energy hump of blazars depends on the frequency of the seed photons, the energy of the relevant electrons contributing to the peak, the bulk Lorentz factor Γ, and the beaming factor δ. For our sources, which are all very powerful, we can assume that δ ≈ Γ, implying that the viewing angle θv ≈ 1/Γ. If the emitting region is inside the broad-line region (i.e., Rdiss <  RBLR) the most important seed photons are the Lyα ones. Therefore, we expect

(5)

If RBLR <  Rdiss <  Rtorus, the most important seed photons are the ones produced by the torus. These have a frequency related to the torus temperature, which has to be lower than ∼2000 K to avoid sublimation.

(6)

The ratio of the two νC frequencies is ∼40 (103 K/Ttorus). If the emitting region is at a distance that is greater than but close to RBLR, both types of seed photons are important, and we have an intermediate peak frequency as long as γpeak is the same. In general, one would expect that the radiative cooling time is affected by the nature of the seed photons: inside the BLR, the BLR radiation energy density is larger than that produced by the torus. Cooling is more severe, and this could favor smaller γpeak. This compensates the larger seed photon energy. On the other hand, we calculate the particle distribution at the end of the injection, which lasts for a time R/c. We also assume that the jet is conical, and therefore R ∝ Rdiss: if the emitting region is beyond RBLR, it is larger than if it is inside. This means that emission (and cooling) operate for a longer time, and this has the effect of decreasing γpeak. Therefore, it is not obvious that sources dissipating beyond RBLR should be “bluer” than the others. In any case, we have tried to see how many blazars studied previously by our group require Rdiss >  RBLR.

Figure 10 shows Rdiss as a function of RBLR for the sample of blazars studied in Ghisellini et al. (2014) and for the high-redshift NuSTAR FSRQs studied here. The figure shows that there is a small (∼12%) fraction of sources with Rdiss ≳ RBLR and that there is an overall trend for Rdiss increasing more than linearly with RBLR. The NuSTAR blazars require the largest Rdiss and RBLR and nearly half of them dissipate beyond RBLR. We also consider the possibility that the Rdiss/RBLR ratio could be a function of the black hole mass. We do expect some dependence, because RBLR depends on the black hole mass only through Ld (and we do expect a more luminous disk for larger black hole masses), while Rdiss should scale linearly with the mass if dissipation occurs at the same distance measured in units of the Schwarzschild radius. Therefore we expect a dependence (albeit weak) for larger Rdiss/RBLR ratios for larger masses. Figure 11 shows this weak trend.

thumbnail Fig. 10.

Distance Rdiss at which most of the luminosity is produced as a function of the size of the broad-line region, RBLR. Blue (“BL Lacs”, but with broad emission lines; see text) and red (FSRQ) data points are from Ghisellini et al. (2014). Green diamonds are our NuSTAR blazars. Different states of the same source are connected by a segment. For about 12% of all sources the dissipation region is located beyond the BLR (Rdiss >  RBLR).

Open with DEXTER

thumbnail Fig. 11.

Ratio Rdiss/RBLR as a function of the black hole mass. Blue (“BL Lacs”) and red (FSRQ) from Ghisellini et al. (2014). Green diamonds are our NuSTAR blazars. Different states of the same source are connected by a segment.

Open with DEXTER

5.3. The γpeak − U′ relation

We now consider the relation between the electron random Lorentz factor γpeak of the electrons emitting at the peaks of the SED (both synchrotron and IC) and the magnetic plus radiation energy density in the comoving frame of the emitting region. This is shown in Fig. 12, which compares our high-zNuSTAR blazars with the samples of blazars studied by Celotti & Ghisellini (2008) and Ghisellini et al. (2014). If considered altogether, there is a clear trend of decreasing γpeak for increasing energy density. On the other hand, the number of NuSTAR blazars is too small to derive any conclusions: they are, as are all the other powerful FSRQs, at the extreme of the distribution.

thumbnail Fig. 12.

Random Lorentz factor of the electrons emitting at the synchrotron and IC peaks vs the radiation+magnetic energy density as measured in the comoving frame. Grey filled circles: sources studied in Celotti & Ghisellini (2008); empty red circles and blue circles: FSRQs and “BL Lacs” from Ghisellini et al. (2014); green diamonds: the sample of z >  2 blazars observed by NuSTAR. Segments connect different states of the same source.

Open with DEXTER

5.4. Jet power and disk luminosity

Finally, in Fig. 13, we consider the jet power as a function of the disk luminosity. The blue circles are labeled “BL Lacs”, as was done by Ghisellini et al. (2014); they come from the sample of Sbarrato et al. (2013), containing 475 sources. Of these, Ghisellini et al. (2014) selected the few (26) objects with broad emission lines. Therefore, these “BL Lacs” should be considered as the low-disk-luminosity tail of the FSRQs. The relation between Pjet and Ld remains significant even after accounting for the common dependence upon redshift, with a probability P <  10−8 of being random (Ghisellini et al. 2014). This figure clearly shows that the NuSTAR blazars studied in this paper are the most powerful. This remains true even if we consider the lower limit to the jet power given by Pr, which is almost model independent. PKS 0836+710 has the most powerful jet, and S5 0014+81 has the most powerful accretion disk. They extend the almost linear correlation between the two quantities found in Ghisellini et al. (2014), and confirm that active blazars have jets that are often more powerful than their accretion disks.

thumbnail Fig. 13.

Jet power as a function of disk luminosity of FSRQs (red) and “BL Lacs” (blue) considered in Ghisellini et al. (2014) compared with the NuSTAR blazars considered here. We also show the blazars with z >  4 and z >  5, considered in Sbarrato et al. (2016) and in Ghisellini et al. (2015). Segments connect different states of the same source. The NuSTAR blazars are among the most powerful, both in terms of their disk luminosity and jet power, with PKS 0836+710 having the most powerful jet, and S5 0014+81 having the most powerful accretion disk. We note that the BL Lacs shown here were the only BL Lacs observed by Sbarrato et al. (2013) with broad emission lines. They must be considered the low disk luminosity tail of FSRQ.

Open with DEXTER

6. Conclusions

We report the results of our NuSTAR observations of three blazars at redshifts greater than 2, and discuss the properties of all blazars at z >  2 observed by NuSTAR and whose data are public. These objects form a sample of 11 sources. The main conclusions of our study are:

– Selection in the hard X-rays allows one to find the most powerful blazar jets and the most luminous accretion disks.

– PKS 0227–369 and TXS 0458–020 show significant variability in hard X-rays with respect to previous observations. This variability can be explained mainly by a change of power of the injected electrons and in part by a change of the magnetic field.

– All the high-zNuSTAR blazars observed so far belong to the class of very powerful FSRQs and have large black hole masses and accretion disks emitting well above the 0.01 LEdd rate.

– The high-zNuSTAR blazars extend and confirm the relation between jet power and accretion-disk luminosity.


Acknowledgments

We acknowledge the ASI–NuSTAR grant ASI 1.05.04.95 and the grant ASI-INAF n. 2017–14–H.0 for funding.

References

  1. Ajello, M., Costamante, L., Sambruna, R. M., et al. 2009, ApJ, 699, 603 [NASA ADS] [CrossRef] [Google Scholar]
  2. Atwood, W. B., Abdo, A. A., Ackermann, M., et al. 2009, ApJ, 697, 1071 [NASA ADS] [CrossRef] [Google Scholar]
  3. Bañados, E., Venemans, B. P., Mazzucchelli, C., et al. 2018, Nature, 553, 473 [NASA ADS] [CrossRef] [Google Scholar]
  4. Bhatta, G., Mohorian, M., & Bilinsky, I. 2018, A&A, 619, A93 [NASA ADS] [CrossRef] [EDP Sciences] [Google Scholar]
  5. Burrows, D., Hill, J., Nousek, J., et al. 2005, Space Sci. Rev., 120, 165 [NASA ADS] [CrossRef] [Google Scholar]
  6. Calderone, G., Ghisellini, G., Colpi, M., & Dotti, M. 2013, MNRAS, 431, 210 [NASA ADS] [CrossRef] [Google Scholar]
  7. Campitiello, S., Ghisellini, G., Sbarrato, T., & Calderone, G. 2018, A&A, 612, A59 [NASA ADS] [CrossRef] [EDP Sciences] [Google Scholar]
  8. Celotti, A., & Ghisellini, G. 2008, MNRAS, 385, 283 [NASA ADS] [CrossRef] [Google Scholar]
  9. Ciprini, S. 2015, ATel, 7870 [Google Scholar]
  10. D’Ammando, F., & Orienti, M. 2016, MNRAS, 455, 1881 [NASA ADS] [CrossRef] [Google Scholar]
  11. Dermer, C. D. 1995, ApJ, 360, 197 [NASA ADS] [CrossRef] [Google Scholar]
  12. Fan, Z. H., & Cao, X. 2004, ApJ, 602, 103 [NASA ADS] [CrossRef] [Google Scholar]
  13. Francis, P. J., Hwett, P. C., Foltz, C. B., et al. 1991, ApJ, 373, 465 [NASA ADS] [CrossRef] [Google Scholar]
  14. Gehrels, N., Chincarini, G., Giommi, P., et al. 2004, ApJ, 611, 1005 [NASA ADS] [CrossRef] [Google Scholar]
  15. Ghisellini, G. 2015, JHEAp, 7, 163 [NASA ADS] [Google Scholar]
  16. Ghisellini, G., & Tavecchio, F. 2009, MNRAS, 397, 985 [NASA ADS] [CrossRef] [Google Scholar]
  17. Ghisellini, G., & Tavecchio, F. 2015, MNRAS, 448, 1060 [NASA ADS] [CrossRef] [Google Scholar]
  18. Ghisellini, G., Foschini, L., Volonteri, M., et al. 2009a, MNRAS, 399, L24 [NASA ADS] [CrossRef] [Google Scholar]
  19. Ghisellini, G., Tavecchio, F., & Ghirlanda, G. 2009b, MNRAS, 399, 2041 [NASA ADS] [CrossRef] [Google Scholar]
  20. Ghisellini, G., Ceca, R. D., Volonteri, M., et al. 2010, MNRAS, 405, 387 [NASA ADS] [Google Scholar]
  21. Ghisellini, G., Tagliaferri, G., Foschini, L., et al. 2011, MNRAS, 411, 901 [NASA ADS] [CrossRef] [Google Scholar]
  22. Ghisellini, G., Sbarrato, T., Tagliaferri, G., et al. 2014, MNRAS, 440, L111 [NASA ADS] [CrossRef] [Google Scholar]
  23. Ghisellini, G., Haardt, F., Ciardi, B., et al. 2015, MNRAS, 452, 3457 [NASA ADS] [CrossRef] [Google Scholar]
  24. Harrison, F. A., Craig, W. W., Christensen, F. E., et al. 2013, ApJ, 770, 103 [NASA ADS] [CrossRef] [Google Scholar]
  25. Kalberla, P. M. W., Burton, W. B., Hartmann, D., et al. 2005, A&A, 440, 775 [NASA ADS] [CrossRef] [EDP Sciences] [Google Scholar]
  26. Kelly, B. C., & Bechtold, J. 2007, ApJS, 168, 1 [NASA ADS] [CrossRef] [Google Scholar]
  27. Komissarov, S. S., Barkov, M. V., Vlahakis, N., & Königl, A. 2007, MNRAS, 380, 51 [NASA ADS] [CrossRef] [Google Scholar]
  28. Lister, M. L., Aller, M. F., Aller, H. D., et al. 2016, AJ, 151, 133 [NASA ADS] [CrossRef] [Google Scholar]
  29. Marcotulli, L., Paliya, V. S., Ajello, M., et al. 2017, ApJ, 839, 96 [NASA ADS] [CrossRef] [Google Scholar]
  30. Marscher, A. P. 1980, ApJ, 235, 386 [NASA ADS] [CrossRef] [Google Scholar]
  31. Mattox, J. R., Bertsch, D. L., Chiang, J., et al. 1996, ApJ, 461, 396 [NASA ADS] [CrossRef] [Google Scholar]
  32. Mortlock, D. J., Warren, S. J., Venemans, B. P., et al. 2011, Nature, 474, 616 [NASA ADS] [CrossRef] [PubMed] [Google Scholar]
  33. Paliya, V. S. 2015, ApJ, 804, 74 [NASA ADS] [CrossRef] [Google Scholar]
  34. Pushkarev, A. B., Kovalev, Y. Y., Lister, M. L., & Savolainen, T. 2017, MNRAS, 468, 4992 [NASA ADS] [CrossRef] [Google Scholar]
  35. Sbarrato, T., Ghisellini, G., Nardini, M., et al. 2013, MNRAS, 433, 2182 [NASA ADS] [CrossRef] [Google Scholar]
  36. Sbarrato, T., Ghisellini, G., Nardini, M., et al. 2016, MNRAS, 462, 1542 [NASA ADS] [CrossRef] [Google Scholar]
  37. Shakura, N. I., & Sunyaev, R. A. 1973, A&A, 24, 337 [NASA ADS] [Google Scholar]
  38. Shaw, M. S., Romani, R. W., Cotteret, G., et al. 2012, ApJ, 748, 49 [NASA ADS] [CrossRef] [Google Scholar]
  39. Shen, Y., Richards, G. T., Strausset, M. A., et al. 2011, ApJS, 194, 45 [NASA ADS] [CrossRef] [Google Scholar]
  40. Tagliaferri, G., Ghisellini, G., Perri, M., et al. 2015, ApJ, 807, 167 [NASA ADS] [CrossRef] [Google Scholar]
  41. Volonteri, M. 2010, A&ARv, 18, 1049 [NASA ADS] [CrossRef] [Google Scholar]

All Tables

Table 1.

Selected targets.

Table 2.

Parameters of the X-ray spectral analysis of the NuSTAR data.

Table 3.

Parameters of the power-law fits to the Fermi/LAT data.

Table 4.

Parameters for the models shown in Figs. 1, 2, 3, 4, and 6.

Table 4.

continued.

Table 5.

Entire sample of z >  2 blazars observed by NuSTAR.

All Figures

thumbnail Fig. 1.

Overall SEDs of PKS 0123+25. Besides our data (red points), we show the archival data collected from the ASI/SSDC database. We have indicated in green the XMM-Newton data, taken in January 2009, and in blue the Swift/XRT data taken simultaneously with the NuSTAR observation. The blue arrows in the γ-ray band are upper limits obtained integrating over 30 days (15 days before and 15 days after the NuSTAR observation). Red γ-ray points and arrows correspond to the average flux during the last 4 years. The lines are the result of the modeling (see text).

Open with DEXTER
In the text
thumbnail Fig. 2.

Lines resulting from the modeling assuming that there is no torus, and assuming both a small Rdiss(=2.25 × 1017 cm) and a large Rdiss(=3.6 × 1018 cm). Parameters are listed in Table 4. If we fit the high-energy emission, the model underproduces the near-IR flux. The long dashed line corresponds to the first- and the second-order Compton SSC contributions.

Open with DEXTER
In the text
thumbnail Fig. 3.

Comparison between the best models assuming ψ = 0.1 = 5.7° and ψ = 0.023 = 1.3°, as labeled. The long-dashed lines are the SSC contribution. Parameters are listed in Table 4.

Open with DEXTER
In the text
thumbnail Fig. 4.

Overall SEDs of PKS 0227–369. The X-ray flux was significantly lower during the NuSTAR observations with respect to an earlier Swift/XRT+UVOT observation carried out in November 2008. The red points in the Fermi/LAT band correspond to integrating the last two years of observations. This shows that the source was in a low state during this period of time.

Open with DEXTER
In the text
thumbnail Fig. 5.

γ-ray light curve of TXS 0458–02. Blue triangles are 95% upper limits, calculated assuming a power law with photon spectral index Γ = 2. The dashed vertical line corresponds to the NuSTAR observation epoch, when the source was in a very high γ-ray state.

Open with DEXTER
In the text
thumbnail Fig. 6.

Overall SEDs of TXS 0458–02, showing the changes in the high-energy emission due to its strong variability. Since unfortunately there are no low-frequency (mm-optical) data simultaneous to the varying high-energy flux, the models shown assume a quasi-constant flux at these frequencies. This illustrates how the model parameters would change in this case.

Open with DEXTER
In the text
thumbnail Fig. 7.

Comparison of the models adopting Γ = 14 and Γ = 7, as labeled. Parameters are listed in Table 4. The model with Γ = 7 slightly underestimates the NuSTAR data.

Open with DEXTER
In the text
thumbnail Fig. 8.

SED of all 11 blazars at z >  2 observed so far by NuSTAR. It can be noted that (1) the synchrotron hump is remarkably similar; (2) for several sources the accretion disk sticks out in the optical-UV band; (3) 0014+813 has an exceptionally powerful accretion disk; and (4) the X-ray and γ-ray emission is more dispersed and variable.

Open with DEXTER
In the text
thumbnail Fig. 9.

Peak luminosity of the high-energy component as a function of its peak frequency. The dashed line connects three different states of TXS 0458–020. Error bars correspond to factor 3 uncertainties in νpeak and factor 2 in Lγ. There is a weak trend of smaller luminosities for larger peak frequencies, with the exception of TXS 0458–020 when in the high state.

Open with DEXTER
In the text
thumbnail Fig. 10.

Distance Rdiss at which most of the luminosity is produced as a function of the size of the broad-line region, RBLR. Blue (“BL Lacs”, but with broad emission lines; see text) and red (FSRQ) data points are from Ghisellini et al. (2014). Green diamonds are our NuSTAR blazars. Different states of the same source are connected by a segment. For about 12% of all sources the dissipation region is located beyond the BLR (Rdiss >  RBLR).

Open with DEXTER
In the text
thumbnail Fig. 11.

Ratio Rdiss/RBLR as a function of the black hole mass. Blue (“BL Lacs”) and red (FSRQ) from Ghisellini et al. (2014). Green diamonds are our NuSTAR blazars. Different states of the same source are connected by a segment.

Open with DEXTER
In the text
thumbnail Fig. 12.

Random Lorentz factor of the electrons emitting at the synchrotron and IC peaks vs the radiation+magnetic energy density as measured in the comoving frame. Grey filled circles: sources studied in Celotti & Ghisellini (2008); empty red circles and blue circles: FSRQs and “BL Lacs” from Ghisellini et al. (2014); green diamonds: the sample of z >  2 blazars observed by NuSTAR. Segments connect different states of the same source.

Open with DEXTER
In the text
thumbnail Fig. 13.

Jet power as a function of disk luminosity of FSRQs (red) and “BL Lacs” (blue) considered in Ghisellini et al. (2014) compared with the NuSTAR blazars considered here. We also show the blazars with z >  4 and z >  5, considered in Sbarrato et al. (2016) and in Ghisellini et al. (2015). Segments connect different states of the same source. The NuSTAR blazars are among the most powerful, both in terms of their disk luminosity and jet power, with PKS 0836+710 having the most powerful jet, and S5 0014+81 having the most powerful accretion disk. We note that the BL Lacs shown here were the only BL Lacs observed by Sbarrato et al. (2013) with broad emission lines. They must be considered the low disk luminosity tail of FSRQ.

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.