NuSTAR AND MULTIFREQUENCY STUDY OF THE TWO HIGH-REDSHIFT BLAZARS S5 0836+710 AND PKS 2149–306

Blazars are a subclass of active galactic nuclei whose emission is dominated by a relativistic jet pointing toward us. Their emission extends from the radio band to the γ-ray and TeV band and is dominated by nonthermal processes. Their spectral energy distribution (SED) is dominated by two humps. The first, spanning the infrared to the X-ray band, is usually attributed to synchrotron emission, while the second, located from the X-ray to the γ-ray band, is attributed to the inverse Compton scattering process. This second component can be produced by energetic electrons scattering only their own synchrotron photons (synchrotron self-Compton, SSC for short), or also scattering radiation produced externally to the jet (i.e., external Compton, EC for short). The latter process is likely to be important especially in those sources for which the second component is largely dominating the SED. This usually occurs for flat-spectrum radio quasars (FSRQs), which are among the most luminous persistent sources of the universe. As such, they can be detected in almost all bands also at high redshifts, well above $z\gt 5$ (Romani et al. 2004; Sbarrato et al. 2012).

The most powerful FSRQs are therefore very bright hard X-ray and γ-ray sources, with the hard X-ray band (20–150 keV) more effective in selecting bright FSRQs at $z\gt 4$ (Ajello et al. 2009; Ghisellini et al. 2010a), while the γ-ray band (0.1–10 GeV) is very effective up to $z=$ 2–3 (Ghisellini et al. 2011; Giommi et al. 2013; Shaw et al. 2013). Hard X-ray and γ-ray surveys recently obtained with the Burst Alert Telescope (BAT, on board the Swift satellite; Gehrels et al. 2004) and the Large Area Telescope (LAT, on board Fermi; Atwood et al. 2009) instruments (Ajello et al. 2009, 2014; Cusumano et al. 2010; Nolan et al. 2012) provide us the opportunity to see whether the jet properties evolve with cosmic time (Ghisellini et al. 2011; 2013; Volonteri et al. 2011). Moreover, the nonthermal SED of these powerful sources leaves the UV band unhidden (the synchrotron hump peaks at much smaller frequencies, and the high-energy hump at much larger energies; Sbarrato et al. 2013), and at these frequencies the thermal accretion flux can emerge and be detected. This allows us to study the relation between the jet power and the accretion luminosity.

It is clear from the above that the hard X-ray and γ-ray bands are crucial to characterize the properties of these sources. In particular, the hard X-ray band (10–100 keV) is where the Compton component is rapidly rising, but it is also the band in which up to now we did not have very sensitive instruments (e.g., compared to those at softer X-rays). Thanks to the advent of the Nuclear Spectroscopic Telescope Array (NuSTAR) satellite (Harrison et al. 2013), it is now possible to obtain detailed X-ray spectra in the 3–79 keV band for moderately bright X-ray sources. Therefore, we selected two bright blazars, PKS 2149–306 at redshift z = 2.345 and S5 0836+710 at z = 2.172, and organized simultaneous observing campaigns with the NuSTAR and Swift satellites, plus on-ground optical/near-infrared (NIR) observations for PKS 2149–306. Both are detected in the γ-ray band by Fermi-LAT.

Both sources are high-redshift, bright blazars with an SED dominated by the Compton component. PKS 2149–306 was detected up to 100 keV with the PDS detector on board the Beppo SAX satellite with a very flat hard X-ray spectrum (energy spectral energy index $\alpha \simeq 0.4$; Elvis et al. 2000). A flat hard X-ray spectrum up to 100 keV was later also detected with INTEGRAL (Bianchin et al. 2009) and Swift/BAT (Sambruna et al. 2007). This source is detected in the γ-ray band by the Fermi-LAT detector (Nolan et al. 2012). S5 0836+710 has been detected up to 100 keV with the PDS detector on board the Beppo SAX satellite with a very flat hard X-ray spectrum (energy index $\alpha \simeq 0.4$, Tavecchio et al. 2000). The source has also been detected both by INTEGRAL (Beckmann et al. 2006) and Swift/BAT (Sambruna et al. 2007). In particular, the BAT detection shows a steeper spectrum, $\alpha \simeq 0.8$, and a flux a factor of ∼5 weaker than the one recorded with Beppo SAX (Sambruna et al. 2007). This blazar is a bright and variable γ-ray source already detected by the Energetic Gamma-ray Experiment Telescope (onboard the Compton Gamma Ray Observatory; Hartman et al. 1999), with flare activity recently seen with the Fermi-LAT detector (Akyuz et al. 2013).

Here we present the results of simultaneous observations obtained in the X-ray band with NuSTAR and with the X-ray Telescope (XRT; Burrows et al. 2005) on board Swift for the two blazars. For both sources we repeated these observations over a timescale of a few months to check for source variability. The X-ray data are complemented with simultaneous optical/UV data taken with the Ultraviolet/Optical Telescope (UVOT; Roming et al. 2005) on board Swift and, for PKS 2149–306, with optical/NIR data from the robotic Rapid Eye Mount (REM) telescope (Zerbi et al. 2004) located in La Silla, Chile. For both sources we also analyze the Fermi-LAT data over a 1 yr timescale centered on our NuSTAR and Swift observations. While we were in the process of submitting this work, a paper appeared on astro-ph presenting these data of S5 0836+710 (Paliya 2015). That paper is more concentrated on the γ-ray variability, but it also analyzes and discusses the SED of this source. We will briefly compare our results with the one of that paper in the discussion.

In this work, we adopt a flat cosmology with ${H}_{0}=70$ km s⁻¹ Mpc⁻¹ and ${{\rm{\Omega }}}_{M}=0.3$.

The NuSTAR and Swift (and for PKS 2149–306, also REM) observations are integrated with data from the Wide-field Infrared Survey Explorer (WISE ¹⁴ ) satellite (Wright et al. 2010) and with archival data from the NASA/IPAC Extragalactic Database (NED) and the ASI Science Data Center (ASDC).

2.1. NuSTAR Observations

2.2.  Swift Observations

2.3. REM Observations

2.4.  Fermi-LAT Observations

Both sources are also bright γ-ray emitters and are regularly detected by Fermi-LAT. We analyzed the data collected for 1 yr between 2013 June 01 and 2014 June 01 (MJD 56,444–56,809) following the standard procedure,¹⁸ using the Fermi-LAT analysis software ScienceTools v9r34p1 with the P7REP_SOURCE_V15 instrument response functions. Events in the energy range 100 MeV–300 GeV were extracted within a 15° acceptance cone of the Region of Interest (ROI) centered on the location of each source. Gamma-ray fluxes and spectra were determined by an unbinned maximum likelihood fit with gtlike. The background model included all known γ-ray sources within the ROI from the second Fermi-LAT catalog (Nolan et al. 2012).¹⁹ Additionally, the model included the isotropic and Galactic diffuse emission components. Flux normalization for the diffuse and background sources were left free in the fitting procedure.

The Fermi-LAT light curves above 100 MeV binned over a timescale of 7 days are shown in Figure 1 for both sources. We mark with vertical dotted lines the dates of the NuSTAR observations. For bins with test statistic TS < 4, 95% confidence level upper limits are plotted (for the meaning of TS see Mattox et al. 1996). We fitted the 1 yr Fermi-LAT spectrum for each source with a power-law model, finding a best-fit photon spectral index of ${\rm{\Gamma }}=2.89\pm 0.09$ (TS = 210) and ${\rm{\Gamma }}=2.69\pm 0.10$ (TS = 284) for PKS 2149–306 and S5 0836+710, respectively (errors are at the 68% confidence interval for one parameter of interest). The corresponding average 1 yr fluxes are $F(\gt 100$ MeV)$\;=\;(8.4\pm 0.9)\times {10}^{-8}$ photons cm⁻² s⁻¹ and $F(\gt 100$ MeV)$\;=\;(5.6\pm 0.8)\times {10}^{-8}$ photons cm⁻² s⁻¹.

Zoom In Zoom Out Reset image size

In the case of S5 0836+710 we have enough statistics to extract a 1 month (2014 January 1–February 1) source spectrum centered around the second NuSTAR observation. The fitting result yielded ${\rm{\Gamma }}=2.7\pm 0.2$ (TS = 46) with an average flux above 100 MeV of $(8.7\pm 1.9)\times {10}^{-8}$ photons cm⁻² s⁻¹. From Figure 1 it is also apparent that a couple of months later the source became much more active in the Fermi-LAT band.

2.5. X-Ray Spectral Analysis

For all observations of both sources simultaneous fits of the XRT and NuSTAR spectra were performed using the XSPEC package. In both cases a broken power-law model with an absorption hydrogen-equivalent column density fixed to the Galactic value was adopted (${N}_{{\rm{H}}}=1.6\times {10}^{20}$ cm⁻² and ${N}_{{\rm{H}}}=2.8\times {10}^{20}$ cm⁻² for PKS 2149–306 and S5 0836+710, respectively; Kalberla et al. 2005). To allow for the cross-calibration uncertainties between the three telescopes (two NuSTAR and one Swift), a multiplicative constant factor has been included, kept equal to 1 for the FPMA spectra and free to vary for the FPMB and XRT spectra. In the case of FPMB the difference is in the range of 2%–4%, while for XRT it is somewhat larger but always less than 10%. This is consistent with the values usually found for other sources. We find that this model provides a good description of the observed spectra in the 0.3–79 keV energy band. The results of the spectral fits are in Table 3, while Figure 2 shows an example of the XRT and NuSTAR spectra, together with the best-fit model for the observation of PKS 2149–306 on 2014 April 18. To test the robustness of the spectral curvature, we estimated the instrumental cross-calibration factors by fitting the data in a common energy band (3–9 keV), adopting a power-law model. We found fully consistent best-fit values of the cross-calibration factors. We then verified that the broken power-law best-fit spectral results were unchanged using these values. Besides using a broken power-law model, we tried a simpler power-law model, but this did not provide a good fit to the data for either source. In fact, in all cases the reduced ${\chi }_{r}^{2}$ was larger than 1.1–1.2 with more than 900 degrees of freedom. The F-test shows that the improvement of these values obtained with the broken power-law model is significant, with a probability smaller than 10⁻³⁰ in three cases and smaller than 10⁻¹¹ in one case. Moreover, when using the power-law model, the constant value included to take into account the cross-calibration uncertainties was too large or too low (e.g., 20%–30% difference) with respect to the values usually found, another indication that the single power-law model is not correct.

Zoom In Zoom Out Reset image size

Table 3.  Parameters of the X-ray Spectral Analysis for the Simultaneous Fit of the NuSTAR and Swift/XRT Data

PKS 2149–306
Date	${{\rm{\Gamma }}}_{1}$	${{\rm{\Gamma }}}_{2}$	Break	${F}_{2-10\;\mathrm{kev}}$	${F}_{10-40\;\mathrm{kev}}$	${\chi }^{2}$/dof
			(keV)	(erg cm−2 s−1)	(erg cm−2 s−1)
2013 Dec 17	${0.94}_{-0.08}^{+0.07}$	${1.35}_{-0.01}^{+0.02}$	${2.61}_{-0.46}^{+0.67}$	$2.2\times {10}^{-11}$	$5.0\times {10}^{-11}$	1095.8/1108
2014 Apr 18	${0.97}_{-0.09}^{+0.07}$	${1.46}_{-0.02}^{+0.02}$	${3.25}_{-0.63}^{+1.36}$	$1.9\times {10}^{-11}$	$3.7\times {10}^{-11}$	936.2/952
S5 0836+710
Date	${{\rm{\Gamma }}}_{1}$	${{\rm{\Gamma }}}_{2}$	Break	${F}_{2-10\;\mathrm{kev}}$	${F}_{10-40\;\mathrm{kev}}$	${\chi }^{2}$/dof
			(keV)	(erg cm−2 s−1)	(erg cm−2 s−1)
2013 Dec 15	${1.03}_{-0.32}^{+0.20}$	${1.66}_{-0.02}^{+0.02}$	${1.73}_{-0.48}^{+1.27}$	$1.6\times {10}^{-11}$	$2.3\times {10}^{-11}$	642.6/611
2014 Jan 18	${1.18}_{-0.10}^{+0.08}$	${1.66}_{-0.01}^{+0.02}$	${2.84}_{-0.62}^{+1.03}$	$2.5\times {10}^{-11}$	$3.6\times {10}^{-11}$	896.4/892

Note. The errors are at the 90% level of confidence for one parameter of interest. Fluxes are corrected for the Galactic absorption.

Download table as:  ASCIITypeset image

In Table 3 we also provide the fluxes in the two bands 2–10 keV and 10–40 keV for the four observations. Both sources have varied between the two observations, spanning a timescale of 1 month for S5 0836+710 and of 4 months for PKS 2149–306. No fast variability is observed during the single observations. However, while for S5 0836+710 there is an increase by a factor of ∼1.5 over the full 0.3–79 keV band, with a hint of lower variability below ∼1 keV, for PKS 2149–306 a variability of ∼30% is present essentially only above 10 keV. To better show this, we plot all XRT and NuSTAR spectra in Figure 3.

Zoom In Zoom Out Reset image size

Figure 4 shows the X-ray data of both sources and compares them with archival observations. For S5 0836+710 the two observations correspond to different flux levels and slightly different spectra. The XRT spectrum smoothly joins the NuSTAR data points in both observations. The two data sets differ mainly around 10 keV, by a factor of ∼1.6, with fewer variations at both ends of the observed X-ray band. On the contrary, for PKS 2149–306 the variability amplitude monotonically increases at higher frequencies, while the two spectra are very similar in the soft X-ray band. The archival data (orange symbols) indicate that both sources (but especially S5 0836+710) can vary in X-rays by about an order of magnitude. Note also the much softer archival spectrum of PKS 2149–306 by XMM (and by INTEGRAL).

Zoom In Zoom Out Reset image size

Figure 5 shows the overall SED of the two sources. Both of them show, besides the typical double-peak SED of blazars, a third narrow peak in the IR–UV band, which we identify as thermal emission from the accretion disk. Fitting it with a standard Shakura & Sunyaev (1973) disk, we find both the disk luminosity ${L}_{{\rm{d}}}$ and the black hole mass M (as listed in Tables 4 and 5; see Calderone et al. 2013 for a full discussion of this method). For a given efficiency η (defined through ${L}_{{\rm{d}}}=\eta \dot{M}{c}^{2}$), and when the peak of the disk emission is visible, this method returns a value of the black hole mass with a relatively small uncertainty (factor of ∼1.5), better than the virial method (factor of 3–4; Vestergaard & Peterson 2006; Park et al. 2012). Adopting $\eta =0.08$, we find a black hole mass of $5\times {10}^{9}{M}_{\odot }$ for S5 0836+710 and $3.5\times {10}^{9}{M}_{\odot }$ for PKS 2149–306.²⁰ These values differ slightly from what we reported previously using the same method (in Ghisellini et al. 2010a, hereafter GG10). This is due to the better coverage of the IR part of the spectrum now available from WISE and REM. With these values of the black hole mass, the disks of both sources emit at ∼20%–30% of the Eddington luminosity. The blazar S5 0836+710 has also been monitored for reverberation mapping by Kaspi et al. (2007). Using the observed lag between the continuum, the C iv FWHM, and Equation (5) of Kaspi et al. (2000), they obtained a black hole mass $M=2.6\times {10}^{9}{M}_{\odot }$. Instead, using the UV luminosity, the C iv FWHM, and Equation (7) of Vestergaard & Peterson (2006), they derive $M=1.8\times {10}^{10}{M}_{\odot }$.

Zoom In Zoom Out Reset image size

Table 4.  Input Parameters Used to Model the SED

Name	$M$	Γ	${\theta }_{{\rm{v}}}$	Rdiss	R	${R}_{\mathrm{BLR}}$	${P}_{{\rm{i}}}^{\prime }$	B	${\gamma }_{{\rm{b}}}$	${\gamma }_{\mathrm{max}}$	s1	s2	${\gamma }_{{\rm{c}}}$
(1)	(2)	(3)	(4)	(5)	(6)	(7)	(8)	(9)	(10)	(11)	(12)	(13)
0836+710 H	5e9	16	3	1950	195	1500	0.11	1.11	250	5e3	1.7	3.2	5.7
0836+710 L	5e9	16	3	2100	210	1500	0.09	1.04	190	4e3	1.6	3.2	9.8
0836+710 GG10	3e9	14	3	540	54	1500	0.22	3.28	90	2e3	–1	3.6	2.1
2149–306 H	3.5e9	14	3	1365	137	1212	0.2	1.05	75	4e3	0.5	3.3	2.9
2149–306 L	3.5e9	14	3	1365	137	1212	0.1	1.05	50	4e3	1	3.0	2.9
2149–306 GG10	5e9	15	3	1200	120	1224	0.18	1.12	60	3e3	0	3.3	1

Note. "L" and "H" stand for low and high state, respectively; GG10 stands for Ghisellini et al. (2010a), where the two sources were also studied. Column [1]: source name and state/observation; column [2]: black hole mass in solar mass units; column [3]: bulk Lorentz factor; column [4]: viewing angle (degrees); column [5]: distance of the blob from the black hole in units of 10¹⁵ cm; column [6]: source size in units of 10¹⁵ cm; column [7]: size of the broad-line region in units of 10¹⁵ cm; column [8]: power injected in the blob calculated in the comoving frame, in units of 10⁴⁵ erg s⁻¹; column [9]: magnetic field in G; column [10] and [11]: minimum and maximum random Lorentz factors of the injected electrons; column [12] and [13]: slopes of the injected electron distribution [$Q(\gamma )$] below and above ${\gamma }_{{\rm{b}}}$; column [13]: value of the minimum random Lorentz factor of the electrons cooling in $R/c$. The spectral shape of the corona is assumed to be $\propto \;{\nu }^{-1}\mathrm{exp}(-h\nu /150\;\mathrm{keV})$.

Download table as:  ASCIITypeset image

Table 5.  Accretion and Jet Powers

Name	$\mathrm{log}{L}_{{\rm{d}}}$	${L}_{{\rm{d}}}/{L}_{\mathrm{Edd}}$	$\mathrm{log}{P}_{{\rm{r}}}$	$\mathrm{log}{P}_{{\rm{B}}}$	$\mathrm{log}{P}_{{\rm{e}}}$	$\mathrm{log}{P}_{{\rm{p}}}$
(1)	(2)	(3)	(4)	(5)	(6)	(7)
0836+710 H	47.4	0.3	46.8	46.7	45.9	48.6
0836+710 L	47.4	0.3	46.7	46.7	45.9	48.5
0836+710 GG10	47.4	0.5	46.6	46.4	45.5	48.0
2149–306 H	47.2	0.3	47.1	46.2	45.7	48.3
2149–306 L	47.2	0.3	46.7	46.2	45.6	48.3
2149–306 GG10	47.2	0.2	46.6	46.2	45.3	48.0

Note. Column [1]: source name and state/observation; column [2]: logarithm of the accretion disk luminosity (units are erg s⁻¹); column [3]: accretion disk luminosity in Eddington units; column [4]–[7]: logarithm of the jet power in the form of radiation (${P}_{{\rm{r}}}$), poynting flux (${P}_{{\rm{B}}}$), bulk motion of electrons (${P}_{{\rm{e}}}$), and protons (${P}_{{\rm{p}}}$) (assuming one cold proton per emitting electron). Units are erg s⁻¹. These values refer to one jet only.

Download table as:  ASCIITypeset image

As discussed also in GG10, these sources are weak in the γ-ray band. This is due to two effects. The first is a k-correction effect: for increasing redshifts, the observed hard X-ray spectrum is closer to the high-energy peak and thus brighter. The second effect is due to the intrinsic shift of the high-energy peak frequency as the bolometric nonthermal luminosity is increased. This implies that the most powerful jetted sources can be better found in hard X-ray surveys rather than in γ-ray surveys (GG10). For S5 0836+710 Fermi-LAT detected the source in 1-month integration times around the second NuSTAR observation (filled red points with black circles), while PKS 2149–306 is detected using 1 yr integration times (from 2013 June 1 to 2014 June 1). Both γ-ray spectra are steep, as usual for powerful FSRQs (see, e.g., Nolan et al. 2012). The fact that the synchrotron far-mm-IR spectrum is steep is consistent with the assumption that both the low- and high-energy nonthermal peaks of the SED are produced by the same population of electrons.

We applied a one-zone leptonic model, fully described in Ghisellini & Tavecchio (2009), to interpret the SED of the two sources and to explore the variability that both sources experienced between the two Swift+NuSTAR observations. In brief, the model assumes that the emitting source is a homogeneous sphere located at a distance R_diss from the black hole, in a conical jet of semi-aperture angle ϕ = 0.1 rad. The broad-line region (BLR) is assumed to be a spherical shell of radius ${R}_{\mathrm{BLR}}={10}^{17}{L}_{{\rm{d}},45}^{1/2}$ cm, and the size of the IR torus is assumed to be ${R}_{\mathrm{torus}}=2\times {10}^{18}{L}_{{\rm{d}},45}^{1/2}$ cm (here ${L}_{{\rm{d}},45}$ is the disk luminosity in units of 10⁴⁵ erg s⁻¹). The source is assumed to move with a bulk Lorentz factor Γ in a direction making an angle ${\theta }_{{\rm{v}}}$ with the line of sight. The magnetic field B (as measured in the comoving frame) depends on the distance from the black hole and the bulk Lorentz factor, to yield a constant Poynting flux ${P}_{{\rm{B}}}\propto {B}^{2}{R}_{\mathrm{diss}}^{2}{{\rm{\Gamma }}}^{2}$.

The energy distribution of the emitting particles is found through the continuity equation, accounting for continuous injection, radiative cooling, and electron-positron pair production. Electrons (with random Lorentz factors $1\lt \gamma \lt {\gamma }_{\mathrm{max}}$) are assumed to be injected with a total power ${P}_{{\rm{i}}}^{\prime }$ (in the comoving frame) throughout the source with a broken power-law distribution, $\propto \;{\gamma }^{-{s}_{1}}$ and $\propto \;{\gamma }^{-{s}_{2}}$ below and above ${\gamma }_{{\rm{b}}}$, respectively. The radiative processes are SSC and EC with both photons from the BLR and the torus. We also consider the presence of an X-ray corona, emitting 30% of ${L}_{{\rm{d}}}$ with a spectrum ${F}_{\nu }\propto {\nu }^{-1}\mathrm{exp}(-h\nu /150\;\;\mathrm{keV}$).

Owing to the compactness of the source, required to account for the fast variability, the synchrotron emission is self-absorbed at radio frequencies, up to hundreds of GHz (observer frame). Therefore, these models cannot reproduce the radio data.

Figure 5 shows the models corresponding to the two states of each source, and Table 4 lists the model parameters. It also reports the parameters used in GG10 to fit another data set. To model the two observed states, we changed a minimum number of parameters. Both sources have ${R}_{\mathrm{torus}}\gt {R}_{\mathrm{diss}}\gt {R}_{\mathrm{BLR}}$. This choice is preferred (with respect to R_diss within the BLR) because of the small value of the high-energy peak, requiring seed photons of frequency smaller than the hydrogen Lyα photons (which dominate the BLR emission). In this respect, NuSTAR is crucial because the hardness of its spectrum, together with the extrapolation from the γ-ray energies, greatly helps in pinpointing the peak frequency of the high-energy hump. This is the major difference with respect to the parameters reported by GG10 for S5 0836+710, which assumed R_diss < R_BLR.

The observed variability of S5 0836+710 is ascribed to a small difference in the position of the emitting region, in turn implying a small change in the magnetic field. Furthermore, the injected power changes by 10%, and the break energy of the injected electron distribution changes by 20%.

For PKS 2149–306 the variability can be explained by changing the injected power by a factor of 2, as well as by a 50% change in the break energy of the injected electron distribution.

Comparing these changes with the distribution of the parameters in many Fermi-LAT blazars (as listed, e.g., in Ghisellini et al. 2010b), we conclude that the changes required to explain the observed (factor of ∼1.5) variability are very small.

The simultaneous observations of NuSTAR and Swift/XRT revealed that the X-ray spectra from ∼0.3 to 60 keV of both sources are well described by a broken power-law model. Both indices are very hard, with no requirement of absorption in excess of the Galactic one. This broken power-law behavior is well reproduced by the model as a result of a combination of two effects. First, the electron energy distribution, below the cooling energy ${\gamma }_{{\rm{c}}}$, retains the slope of the injected electrons (i.e., ${\gamma }^{-{s}_{1}}$), which is very hard (see Table 4). Second, what we observe as low-energy X-ray emission at the frequency ${\nu }_{x}$ is seen, in the comoving frame, as ultraviolet emission (i.e., ${\nu }_{x}^{\prime }={\nu }_{x}(1+z)/\delta$), produced by the scattering between low-energy electrons and seed photons with frequencies smaller than ${\nu }_{x}^{\prime }$. In both sources, the main contributions to the seed photons come from the IR torus, the BLR, and the disk. All these components have a corresponding radiation energy density peaking at some frequency, which is seen between the optical and the UV in the comoving frame of the jet. Inverse Compton scattered photons at, for example, $\nu \prime \sim {10}^{15}$ Hz can be produced using seed photons of frequencies smaller than 10¹⁵ Hz, which do not correspond to the full energy density of the seed photons. Scattered photons of higher energies, instead, can use the entire seed photon distribution. In other words, there is a paucity of seed photons at low energies, making the inverse Compton spectrum harder below a few keV. What is observed in the two blazars studied here can then be regarded as a proof of the "EC" process, where the major contribution to the seed photons comes from radiation produced externally to the jet. Furthermore, the spectrum between 0.3 and 60 keV is smooth, indicating the absence of other emitting components.

Both blazars show variability, but not extreme variability. This agrees with a relatively large source size and a correspondingly relatively large distance from the black hole. The observed light-crossing time predicted by our model is ${t}_{\mathrm{var}}^{\mathrm{obs}}\equiv (R/c)(1+z)/\delta$, corresponding to ∼13 days for S5 0836+710 and ∼10 days for PKS 2149–306. This agrees with no variability observed within a single-epoch observation and is consistent with the emission site being between the BLR and the torus (${R}_{\mathrm{BLR}}\lt {R}_{\mathrm{diss}}\lt {R}_{\mathrm{torus}}$). We note that, contrary to what we have observed, significant variability on shorter timescales has been reported for S5 0836+710 by Akyuz et al. (2013). In the framework of our modeling this is explained by assuming that the component dominating the emission at a given time is not always located in the same place along the jet. Sometimes it can be at a position closer to the black hole and be more compact, giving rise to variability on shorter timescales and to a slightly different SED.

We have shown that the observed moderate variability can be produced by a rather small change in the injected power, break energy of the electron distribution, and the location of the emitting region. The largest change is needed for PKS 2149–306, requiring a factor of 2 change in the injected power.

The NuSTAR data at high energies, coupled with the γ-ray data, allow us to determine where the high-energy hump of the SED peaks. This is at ∼1 MeV (observed, thus ∼3 MeV, rest frame) for both sources. As already mentioned, this helps to determine the location of the emitting region, making us prefer an emitting region located between the BLR and the torus. We stress that this may not necessarily always be the case since the emitting site can change, and sometimes it can be within the BLR. For a given electron distribution, the high-frequency peak would shift to higher values, in this case. The data of S5 0836+710 have been analyzed also by Paliya (2015), who fit the data using the same single-zone leptonic emission model. We find consistent results, but our black hole mass is slightly larger, and our emitting region is just outside the BLR, while in Paliya (2015) it is inside. We find that while the parameters of Paliya (2015) indeed satisfactorily fit the overall SED and the average NuSTAR spectrum, our fit accounts for the two states of the source in a better way, and in particular the two sets of Swift/XRT+NuSTAR data.

Both blazars have a black hole with mass exceeding ${10}^{9}{M}_{\odot }$ and accretion disks emitting at about one-third of the Eddington rate. The power that the jet expends for producing the radiation we see (${P}_{{\rm{r}}}$ in Table 5) is of the same order as ${L}_{{\rm{d}}}$ (note that the listed values of ${P}_{{\rm{r}}}$ refer to one jet only and should be doubled). ${P}_{{\rm{r}}}$ is a lower limit of the true jet power ${P}_{\mathrm{jet}}$. An estimate of ${P}_{\mathrm{jet}}$ can be derived assuming, following Nemmen et al. (2012), that ${P}_{\mathrm{jet}}\sim 10{P}_{{\rm{r}}}$. Another estimate can be found assuming that there is one proton for each emitting electron (this is the quantity ${P}_{{\rm{p}}}$ in Table 5). Either way, we are forced to conclude that the jet power exceeds the luminosity of the accretion disk, thus suggesting that the jet is powered not only by accretion but also by the rotational energy of a spinning black hole, as found by Ghisellini et al. (2014) for a large sample of blazars.

Finally, given the very hard spectra detected by NuSTAR and their high-energy peak at ∼1–10 MeV, one can wonder whether powerful blazars can significantly contribute to the X-ray background above its ∼30 keV peak. While a complete study of this issue is still missing, there are preliminary results by Draper & Ballantyne (2009), Giommi (2011), and Comastri & Gilli (2011) suggesting that blazars can contribute at the 10% level at ∼100 keV, before becoming the dominant contributors to the γ-ray background (Abdo et al. 2010; Ajello et al. 2015).

We acknowledge financial support from the ASI-INAF grant I/037/12/0. This work was supported under NASA Contract No. NNG08FD60C and made use of data from the NuSTAR mission, a project led by the California Institute of Technology, managed by the Jet Propulsion Laboratory, and funded by NASA. We thank the NuSTAR Operations, Software and Calibration teams for support with the execution and analysis of these observations. We also thank the Swift team for quickly approving and executing the requested ToO observations. This research has made use of the NuSTAR Data Analysis Software (NuSTARDAS) jointly developed by the ASI Science Data Center (ASDC, Italy) and the California Institute of Technology (Caltech, USA). 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 online services provided by the ASI Data Center (ASDC).