© ESO, 2012

1. Introduction

Blazars are the most powerful of all active galactic nuclei (AGNs); their continuously emitting radiation covers the entire electromagnetic spectrum from radio to gamma-ray frequencies. Because of their enormous luminosities, blazars are visible to very far distances/high redshifts. In the widely adopted scenario of blazars, a single population of high-energy electrons in a relativistic jet radiates from the radio/far-infrared (FIR) to the ultraviolet (UV)/soft-X-ray through the synchrotron process and at higher frequencies through inverse Compton (IC), which scatters soft-target photons that are present either in the jet (synchrotron self-Compton [SSC] model), in the surrounding material (external Compton [EC] model), or in both (Ghisellini et al. 1998, and references therein). Therefore a strong signature of the blazar nature of a source is a double peaked structure in the spectral energy distribution (SED), with the synchrotron component peaking anywhere from infrared to X-rays and the Compton component extending up to GeV or even TeV gamma-rays. To explain all the different SEDs observed in blazars, Fossati et al. (1998) proposed the blazar sequence, which claims an inverse relation between peak energies and source luminosity: the more luminous sources have both synchrotron and Compton peaks at lower energies than their fainter (and generally lower redshift) counterparts. Within the blazar population, high-redshift objects, which belong to the class of flat-spectrum radio quasars (FSRQ), tend to be the most luminous ones. Recently, Ghisellini et al. (2010) showed that the hard X-ray selected blazars at high redshifts have most powerful jets, the most luminous accretion disks and the largest black hole masses. In other words, they are among the most extreme blazars, even more extreme than those selected in other wavebands like the gamma-ray region explored by Fermi/LAT.

Theoretically, this can be understood on the basis of the blazar sequence and indeed can be considered a proof of its validity: the high-energy peak in the SED of the most powerful blazars is in the hard X-ray/MeV range. As a consequence, these objects are more luminous in the 20 − 100 keV band than above 100 MeV, and thus become detectable in the hard X-ray surveys even if they are undetected in the gamma-ray ones. Considered from another point of view, this also means that a survey in hard X-rays is more efficient in finding the most powerful blazars, which lie at the highest redshifts.

Within the Swift/BAT sample, there are ten blazars (all FSRQs) at redshift greater than 2, of which five have a redshift between 3 and 4 (Ghisellini et al. 2010). Within the INTEGRAL/IBIS surveys we have eight blazars at z ≥ 2 of which two are located between 3 and 4 (Malizia et al., in prep.).

Here, we report the discovery with INTEGRAL of yet another powerful blazar, IGR J12319–0749, which is associated with a quasi-stellar object (QSO) at z = 3.12; we present a set of follow-up observations with Swift/XRT/UVOT, provide the first combined X/gamma-ray spectrum of the source and discuss its non-simultaneous SED. This is likely the second-most distant blazar ever detected with INTEGRAL.

2. INTEGRAL data

IGR J12319–0749 was reported for the first time as a high-energy emitter by Bird et al. (2010) in the 4th INTEGRAL/IBIS catalogue. The source is detected with a significance of ~ 4.8σ at a position corresponding to RA(J2000) = ${\mathrm{12}}^{\mathrm{h}}{\mathrm{31}}^{\mathrm{m}}\mathrm{54}\stackrel{\mathrm{s}}{.}\mathrm{5}$ and Dec(J2000) = $\mathrm{-}{\mathrm{07}}^{\mathrm{◦}}{\mathrm{48}}^{\mathrm{\prime }}\mathrm{57}\stackrel{\mathrm{″}}{.}\mathrm{6}$ with an associated positional uncertainty of 4.7′ (90% confidence level); the source location is 54.8 degrees above the Galactic plane, which strongly suggests an extragalactic nature.

The IBIS spectrum was extracted using the standard off-line scientific analysis (OSA version 7.0) software released by the Integral Scientific Data Centre. Here and in the following, spectral analysis was performed with XSPEC v.12.6.0 package and errors are quoted at 90% confidence level for one interesting parameter (Δχ² = 2.71). A simple power law provides a good fit to the IBIS data (χ² = 10.4 for 8 d.o.f.) with a photon index Γ = 2.80$\begin{array}{}\mathrm{+}\mathrm{1.35}\\ -0.94\end{array}$ and an observed 20 − 100 keV flux of 8.3 × 10^-12 erg cm^-2 s^-1.

Within the IBIS positional uncertainty, we found the faint ROSAT source 1RXS J123158.3–074705, located with an uncertainty of 15′′ (Voges et al. 2000). This object probably coincides with the radio source NVSS J123157–074717 (Condon et al. 1998), which is located $\mathrm{15}\stackrel{\mathrm{″}}{.}\mathrm{8}$ away, i.e. just outside the ROSAT error circle (see Fig. 1); this radio source has recently been optically classified by us and was found to be a QSO at z = 3.12 (Masetti et al. 2012).

Although the association between IGR J12319–0749 and this radio/X-ray high-redshift QSO is intriguing, it cannot be considered secure because 1) there are other potential candidates in the IBIS error circle, for example at radio frequencies, which could emerge at higher X-ray energies and 2) it is surprising to detect such a distant object in a relatively short IBIS exposure (1256 ks).

For this reason, we have requested and obtained follow-up observations of IGR J12319–0749 with the X-ray telescope (Burrows et al. 2005) on board the Swift satellite (Gehrels et al. 2004).

Fig. 1 XRT 0.3–10 keV image of the region surrounding IGR J12319 − 0749. The big circle corresponds to the IBIS positional uncertainty, while the cross indicates the location of the radio source NVSS J123157–074717; the small circle represents the positional uncertainty of the faint ROSAT source.

3. Swift observations and data reduction

During the period March 27 and July 14, 2011, XRT carried out seven observations of the sky region containing IGR J12319−0749 for a total exposure of ~4 ks (see Table 1 for details on individual measurements). XRT data reduction was performed using the XRTDAS standard data pipeline package (xrtpipeline v. 0.12.6), to produce screened event files. All data were extracted only in the photon counting (PC) mode (Hill et al. 2004), adopting the standard grade filtering (0−12 for PC) according to the XRT nomenclature.

Events for spectral analysis were extracted within a circular region of radius 20′′ (which encloses about 90% of the PSF at 1.5 keV, Moretti et al. 2004) centred on the source position. The background was extracted from various source-free regions close to the X-ray source using both circular/annular regions with different radii to ensure an evenly sampled background. In all cases, the spectra were extracted from the corresponding event files using XSELECT software and binned using grppha in an appropriate way, so that the χ² statistic could be applied. We used the latest version (v.011) of the response matrices and created individual ancillary response files (ARF) using xrtmkarf v.0.5.8. In all our fitting procedures we used a Galactic column density that in the direction of IGR J12319−0749 is 1.7 × 10²⁰ cm^-2 (Kalberla et al. 2005).

The XRT images extracted in the 0.3 − 10 keV band were checked for significant excesses (above 2.5 sigma level) falling within the IBIS 90% confidence error circle; only one source was clearly detected in each individual observation and in the combined image (see Fig. 1). This source is located at RA(J2000) = ${\mathrm{12}}^{\mathrm{h}}{\mathrm{31}}^{\mathrm{m}}\mathrm{57}\stackrel{\mathrm{s}}{.}\mathrm{67}$ and Dec(J2000) = $\mathrm{-}{\mathrm{07}}^{\mathrm{◦}}{\mathrm{47}}^{\mathrm{\prime }}\mathrm{19}\stackrel{\mathrm{″}}{.}\mathrm{2}$ with a positional uncertainty of 3.8′′ (90% confidence level), which makes this X-ray source compatible with the faint detection reported by ROSAT. It also coincides with the NVSS/QSO object mentioned above; since this is the only source detected in X-rays, we assume that it is the true counterpart of the IBIS object.

Table 1 reports for each X-ray measurement the observation date, the relative exposure, the net count rate in the 0.3−10 keV energy band, and the detection significance. There is some evidence that the source flux underwent some changes over the XRT monitoring with the most evident variation occurring between observation #2 and #7, i.e. over a period of a few months. Variability in a QSO is not unexpected and can be used to characterise this new gamma-ray source in more detail.

The UVOT instrument (Roming et al. 2005) observed IGR J12319–0749 in conjunction with the XRT for all the observations listed in Table 1. Each observation consists of one or more exposures with a single UVOT filter. The data were analysed using the latest HEASoft tools and calibration products following the methods described by Poole et al. (2008) and Breeveld et al. (2010). A 3′′ source aperture centred on the X-ray position¹ and a surrounding annulus background region were used for all exposures; the usual corrections including dead time, coincidence loss, and aperture loss were made. Magnitudes were obtained from the observed count rates using the latest zero points (Breeveld et al. 2011) and are reported in Table 1 for each observation. To convert to AB magnitudes, 1.02, 1.51, and 1.69 should be added to the reported u, uvw1, and uvw2 magnitudes, respectively (Breeveld et al. 2011). The source was only detected in the u filter in observation #4 and #7 while 90% confidence upper limits for either the uvw1 or uvw2 filters are listed for the other measurements. Count rates were also converted to fluxes using the prescription of Poole et al. (2008).

4. X-ray spectral analysis and results

The spectral analysis of the XRT data indicated that the soft X-ray spectrum during observation #1 and #4, i.e. the two with the best statistical significance, is hard and bright: a simple power law fit to these two measurements provides $\mathrm{\Gamma }\mathrm{=}\mathrm{0.9}{\mathrm{6}}_{-0.34}^{\mathrm{+}\mathrm{0.32}}$ and Γ = 1.53 ± 0.28, together with a 2−10 keV flux of  ~5.7 × 10^-12 and  ~2.6 × 10^-12 erg cm^-2 s^-1, respectively. Note that for simplicity, here and in the following all modelling has been carried out in the observer’s rest frame.

The 0.1–2.4 keV band flux is about 1.4−1.5 ×  10^-12 erg cm^-2 s^-1, to be compared with the lower ROSAT flux, which is in the range 0.4−1 ×  10^-12 erg cm^-2 s^-1, depending on the source spectrum²; this comparison provides another indication that the source might be variable in soft X-rays.

To better characterise the QSO’s X-ray spectrum and because the IBIS detection was made over a few revolutions, i.e. provides an average flux, we also combined all seven XRT spectra and repeated the spectral analysis. The fit to the average XRT spectrum provides a photon index Γ = 1.32 ± 0.14 and a 2 − 10 keV flux of  ~3.3 × 10^-12 erg cm^-2 s^-1.

Finally, a joint spectral fit to the average XRT and IBIS data was also attempted. To account for a possible calibration mismatch between the two instruments and/or variability between the two datasets, we introduced a constant C, which was left free to vary. A simple power law provides an acceptable fit with Γ = 1.35 ± 0.14, $\mathit{C}\mathrm{=}\mathrm{0.}{\mathrm{8}}_{-0.38}^{\mathrm{+}\mathrm{0.67}}$ and χ² = 31.4 for 22 d.o.f. Since there is some evidence for a high-energy cut-off (E_C) in the data-to-model ratio, we also fitted the source broad-band spectrum with a cut-off power law: in this case, $\mathrm{\Gamma }\mathrm{=}\mathrm{1.2}{\mathrm{4}}_{-0.19}^{\mathrm{+}\mathrm{0.17}}$, $\mathit{C}\mathrm{=}\mathrm{2.5}{\mathrm{3}}_{-1.55}^{\mathrm{+}\mathrm{5.24}}$, E_C = $\mathrm{24.}{\mathrm{5}}_{-14.6}^{\mathrm{70.6}}$, and the χ² = 25.7 for 21 d.o.f. (see Fig. 2). With respect to a simple power law the Δχ² is 5.7 for one extra parameter, which implies a fit improvement at a confidence level of about 96%, not highly significant, but still possibly suggesting that a spectral change might be present in the spectrum of IGR J12319–0749. In the source rest frame this change would be located at about 100 keV.

Assuming that this broad-band fit represents the average state of the source, we obtain observer frame luminosities³ of 2.8 × 10⁴⁷ erg s^-1 in the X-ray (2 − 10 keV) band and 7.3 × 10⁴⁷ erg s^-1 at hard X-rays (20 − 100 keV), i.e. IGR J12319–0749 has a high-energy luminosity similar to those observed in the Ghisellini et al. (2010) sample of high-redshifts blazars.

Fig. 2 X-ray spectrum of IGR J12319–0749 fitted with a cut-off power law.

5. Is IGR J12319-0749 another powerful blazar?

The source is reported in the NED database⁴ as NVSS J123157–074717, but with no reference, while it is not contained in the SIMBAD archive; despite being potentially quite a powerful AGN, it has never been studied before. Nonetheless, the sparse information available is sufficient to broadly characterise this new gamma-ray source.

Fig. 3 Left: non-simultaneous SED (source rest frame) of IGR J12319–0749; data and references are discussed in the text. Right: SED (source rest frame) of two high-z QSOs (225155+2217 at z = 3.668, and 0347–221 at z = 2.944). (Reproduced by permission of G. Ghisellini (Ghisellini 2011).)

The NVSS (NRAO VLA Sky Survey, Condon et al. 1998) image of IGR J12319–0749 shows the source to be core-dominated with no extended radio features and a 1.4 GHz flux of 59.8 ± 1.8 mJy. The source is also listed in the FIRST survey with a similar 1.4 GHz flux (60.99 ± 0.16 mJy, White et al. 1997). No other detection is reported at radio frequencies, preventing a measurement of the source spectrum and an estimate of the radio loudness defined as RL = L_{5  GHz}/L_B. Despite this, we can make some guesses on the source spectral shape at radio frequencies considering that IGR J12319–0749 has not been detected in the Australian Telescope 20 GHz (AT20G) survey, which poses a limit of 40 mJy to its flux at this frequency (Murphy et al. 2010). Assuming S_ν ∝ ν^α and considering the available information, we estimate that α is ≤ − 0.15; this makes the source compatible with being a flat-spectrum radio QSO⁵. Furthermore, by evaluating the source B magnitude flux we should also be able to put a limit on the source radio loudness. IGR J12319–0749 is listed in the USNO-B1.0 catalogue (Monet et al. 2003) with a B magnitude in the range 19.7 − 19.8, while the R and I magnitudes are ~ 19.7 and ~ 18.2, respectively. Using the standard photometric system conversion from magnitude to flux (Zombeck 1990), we estimate a B flux in the range 0.05 − 0.06 mJy; extrapolating the 1.4 GHz flux to 5 GHz using the limit obtained on the radio spectral index, it is possible to estimate loosely that the radio loudness is ≤ 1000, sufficiently high to suggest that the source may be a radio-loud AGN⁶.

IGR J12319–0749 is not reported in the near infrared 2MASS survey, with upper limits to the J, H, K flux of ~ 17.1,  ~ 16.4, and  ~ 15.3 mag, respectively (Skrutskie et al. 2006).

Optically, the source is classified as a broad emission line AGN (FWHM = 5600 km s^-1) by Masetti et al. (2012); these authors furthermore estimate the mass of the black hole at the centre of the QSO to be 2.8 × 10⁹ solar masses, i.e. quite a massive black hole, similar to those found in other high-redshift blazars that were discovered in hard X-ray surveys (Ghisellini et al. 2010; De Rosa et al., in prep.).

At high energies, the source is bright with a hard X-ray spectral shape (Γ = 1.2 − 1.3) and some indication of flux variability. There is marginal evidence in the combined XRT/IBIS data for a spectral break at about 100 keV (source restframe), which could be interpreted as a peak in the spectral energy distribution.

Taken all together, the observed properties suggest that IGR J12319–0749 could be a bright blazar, in which the emission is relativistically beamed and the SED is double-peaked. In Fig. 3 (left side), we construct the non-simultaneous (source rest frame) SED of this object by combining all data gathered in this work; the optical UVOT data are not corrected for Galactic reddening, which is, however, low in the source direction (less than 10% in B). To cover as many frequencies as possible, we also used upper limits obtained from the latest Fermi survey (Abdo et al. 2012). We adopted the cut-off power law model to describe the combined Swift/XRT and INTEGRAL/IBIS spectrum. Although the data are sparse and not simultaneous, the source SED resembles that of a blazar with the synchrotron peak located between radio and near-infrared frequencies and the Compton peak in the hard X-ray band. For comparison, we show on the right-hand of Fig. 3 a plot taken from Ghisellini (2011), which describes the SED (also source rest frame) of two high-z blazars together with a single leptonic model fit to their data: one source (225155+2217 at z = 3.668) has been detected by INTEGRAL/IBIS and Swift/BAT, but not by Fermi/LAT, the other (0347 − 221 at z = 2.944) has been seen by LAT, but not by IBIS or BAT. IGR J12319–0749 is more similar to 225155+2217 (see also Lanzuisi et al. 2012) than to 0347 − 221. It shows a similar overall brightness, Compton peak location and is not detected by Fermi/LAT.

From the available optical/UV data and reported black hole mass, it is possible to evaluate that the source accretion disk is emitting at 10% or more of the Eddington luminosity (see Fig. 3, right panel). This is similar to other high-z blazars discovered by hard X-rays surveys, which tend to have black holes with M greater than 10⁹ solar masses and disks emitting close to, or above 10% of the Eddington limit.

6. Conclusions

Through X-ray follow-up observations with Swift/XRT, we have been able to provide a likely identification of the newly discovered INTEGRAL source, IGR J12319–0749, with a radio source coincident with an AGN at z = 3.12. This would make IGR J12319–0749 the second-most distant object so far detected by INTEGRAL.

The source probably belongs to the class of flat-spectrum radio QSO: it is a broad line AGN with a huge black hole at its centre; it is likely radio-loud, and it shows variability at X-ray energies. The source SED is also similar to another high-z source (225155+2217) discovered by means of hard X-ray observations: it has a similar brightness and disk luminosity, a Compton peak location in the hard X-ray soft gamma-ray domain and is not detected by Fermi/LAT. We conclude that this is likely another example of an extreme blazar, a group of sources that shows the most powerful jets, the most luminous accretion disks and the largest black hole masses. This finding additionally supports the claim that the hard X-ray waveband is the most efficient to discover these powerful AGN.

Acknowledgments

We acknowledge finacial support from ASI under contract ASI I/033/10/0. This research has made use of the NED NASA/IPAC Extragalactic Database (NED) operated by JPL (Caltech)aboratory and of the HEASARC archive provided by NASA’s Goddard Space Flight Center.

References

All Tables

All Figures

	Fig. 1 XRT 0.3–10 keV image of the region surrounding IGR J12319 − 0749. The big circle corresponds to the IBIS positional uncertainty, while the cross indicates the location of the radio source NVSS J123157–074717; the small circle represents the positional uncertainty of the faint ROSAT source.
In the text

	Fig. 2 X-ray spectrum of IGR J12319–0749 fitted with a cut-off power law.
In the text

	Fig. 3 Left: non-simultaneous SED (source rest frame) of IGR J12319–0749; data and references are discussed in the text. Right: SED (source rest frame) of two high-z QSOs (225155+2217 at z = 3.668, and 0347–221 at z = 2.944). (Reproduced by permission of G. Ghisellini (Ghisellini 2011).)
In the text