© ESO, 2011

1. Introduction

The present generation of space observatories for high-energy astrophysics is characterised by large area and wide-field instrumentation. This is the case for the Burst Alert Telescope (BAT, Barthelmy et al. 2005) and the Large Area Telescope (LAT, Atwood et al. 2009) onboard Swift and Fermi-GST, respectively. One of the main throughputs of these instruments is the discovery of a thousand hard X-ray and γ-ray sources and the possibility of performing investigations into the population properties and evolution that are much more accurate than in the past.

Fig. 1 Details of the BAT 15–150 keV all-sky map of the detection significance σ at the position of a) 1FGL J1103.7−2329 (left panel), b) 1FGL J2056.7+4938 (middle panel), and c) 1FGL J0238.3−6132 (right panel). A dashed line represents the position of the 1FGL (black ellipse) and the 2PBC (white circle) sources with its uncertainty. For each source, the position of the associated counterpart is indicated by a cross of the corresponding colour. Each square map is  ~60′ × 60′; the colour scale is optimised for each hard X-ray source.

The first catalogue of γ-ray point sources (1FGL, Abdo et al. 2010a), based on data obtained in the 11 months after the beginning of scientific operation (2008 August 4) contains 1451 entries, and a fraction of about 40% of them do not have reliable counterparts. A rather similar situation is also occurring with the hard X-ray sources detected by BAT: the second Palermo BAT catalogue (Cusumano et al. 2010b, hereafter 2PBC), which covers the observation period from November 2004 to May 2009 and is the richest one in this energy range, also includes 177 unassociated sources, over a total number of 1256. A multifrequency approach based on the cross-correlation of catalogues in different energy bands from the radio to the γ rays can be very useful, if not essential, for unravelling the nature of unidentified sources.

In this paper we use the 1FGL and the 2PBC catalogues to compare the populations of hard X-ray and γ-ray sources and to search for possible correspondences among them. The paper is organised as follows. We report the details of the cross-correlation between the two catalogues in Sect. 2 and of the cross-correlation of the 1FGL catalogue with the BAT 15−150 keV all-sky map of the detection significance σ in Sect. 3. We discuss our results in Sect. 4 giving particular emphasis to the blazar content of the collected group of sources. In Sect. 5 we describe the properties of two unidentified objects in the 1FGL catalogue at low Galactic latitude, with significant emission in the hard X-ray band. We have collected all the available data in the literature and analyse their broad band spectral energy distribution (SED) in detail to support their classification as blazars. The main results of our analysis are summarised in Sect. 6.

2. Correspondences between 1FGL and 2PBC sources

The 2PBC catalogue was obtained from the reduction of the BAT data collected over the 54 months since the launch of the Swift mission using the dedicated software batimager  (Segreto et al. 2010). It contains 1256 hard X-ray sources detected at a significance level higher than σ_T = 4.8, and their coordinates are given with a 95% confidence level radius r_BAT. A counterpart was associated to 1079 sources (~86%); for 26 of them, a double association was found, and in two cases three possible counterparts were proposed. There were 177 sources without any associated counterpart.

We matched the 2PBC catalogue with the 1FGL catalogue to search for sources emitting in both the energy bands surveyed by the Swift-BAT and the Fermi-LAT telescopes. The positions of the 1451 1FGL γ-ray sources, determined by means of a maximum likelihood algorithm, are given with an uncertainty ellipse corresponding to the 95% probability of locating the source, and 58 objects are reported without any positional uncertainty. We adopted, when available, the mean value of the two axes of the 95% confidence ellipse as the radius r_LAT of the error circle for a 1FGL source. We calculated the angular distance d among the centroids of 1FGL and 2PBC sources and established a correspondence when the two error circles overlap, adopting the condition d ≤ (r_BAT + r_LAT). After applying this criterion we obtain 77 correspondences. We define a simple parameter to estimate the quality of the correspondence and use the expression Q = d/r_BL, where r_BL is the higher value between r_BAT and r_LAT. The parameter Q < 1 implies that the wider error circle includes both centroids. This value occurs for 64 correspondences, while in four cases we find Q > 1.4.

As a further step, we consider the possibility that the choice of the mean value of the axes of the 95% uncertainty ellipse, adopted to estimate the 1FGL uncertainty region, may be not firmly reliable. This aspect arises particularly in cases of high eccentricity values, in which the orientation of the 1FGL ellipse with respect to the 2PBC circle must be taken into account. To gain confidence in the established correspondence between sources, we computed the axial ratio of the uncertainty ellipse and verify in the sky map the 25 cases (~1/3) in which we find values higher than 1.2. In all but four cases, not only is the centroid of the 2PBC source inside the 1FGL ellipse, but also the associated counterpart.

Further correspondences have been established for nine of the 58 1FGL sources with no positional uncertainty, since the position of the 1FGL source is within the error circle of the 2PBC source. The remaining 49 cases were all unambiguously discarded, because no 2PBC source was found in the proximity of these 1FGL sources. The final list includes 86 1FGL-2PBC correspondences for a total of 84 1FGL sources: two of them in fact have a possible correspondence with two different 2PBC sources. This confusion is due to crowded fields, in one case close to the Galactic centre direction. Moreover, we find a correspondence in which two possible counterparts, both high-mass X-ray binaries, have been associated to the same 2PBC source in the Small Magellanic Cloud. The list of correspondences has been split in two parts reported in Tables 1 and 2 according to the Galactic latitude, with 63 correspondences at |b| > 10° and 23 at |b| < 10°. For each correspondence, we report the 1FGL and 2PBC identifiers, the associated counterparts, and their classification. For a few AGNs the counterpart, missing in the 1FGL catalogue, has been found in the first catalogue of active galactic nuclei detected by the Fermi Large Area Telescope (Abdo et al. 2010b), and these cases are marked with an asterisk (^∗) in Table 1. The values of the Q parameter and the agreement between the corresponding counterparts are reported in the last two columns. We mark with a colon the Q values of the four correspondences for which the 2PBC counterparts are outside the 1FGL uncertainty ellipse. In 62 cases (marked with “y”) the counterpart is the same, while in 11 cases (marked with “n”) the counterparts associated with the high-energy sources are different: in these cases the correspondence is supposed to be due to chance. We also find 13 correspondences for which both sources, or just one of them, lack the associated counterpart. Assuming that the 1FGL-2PBC correspondence is the result of high-energy emission from a single source, we can tentatively associate a counterpart to these 1FGL sources. All these correspondences, with only one exception, are characterised by a value of the Q parameter lower than unity.

We plot in Fig. 1 the details of the BAT 15–150 keV all-sky map of the detection significance σ for some of these correspondences. In the left hand panel of Fig. 1 we report the case in which the 1FGL and the 2PBC sources correspond to the same object, the BL Lac 1H 1100−230. In the middle panel we report the case of the source 1FGL J2056.7+4938 that has not been identified in the 1FGL catalogue. The correspondence with the 2PBC source suggests the X-ray source RXJ2056.6+4940 as a possible counterpart. The properties of this source have been investigated in greater detail and are discussed in Sect. 5.2. Finally, in the right hand panel of Fig. 1 we report the case of a correspondence probably due to chance: two close objects, the galaxy IRAS F02374−6130 and the flat spectrum radio quasar PKS 0235−618, are responsible for the hard X-ray and the γ-ray emission, respectively.

We compare our results with those reported in Abdo et al. (2010b, their Table 7) and find that all the 50 sources provided in their list are included in Table 1, with the exception of 1FGL J1938.2−3957. Hard X-ray emission in this direction was revealed by INTEGRAL and reported in the fourth IBIS catalogue (Bird et al. 2010). The counterpart associated to this 1FGL source, PKS 1933−400, is also reported in the BZCAT as a blazar with uncertain classification. The value σ ~ 3.5 that we find in the BAT 15–150 keV all-sky map at the position of PKS 1933−400 is compatible with hard X-ray emission from this source, but it is lower than σ_T = 4.8.

3. Correspondences of 1FGL sources with the 54-month hard X-ray maps

The 2PBC catalogue includes sources detected at a significance level higher than σ_T = 4.8. A remarkable number of fainter objects are presumably imaged in the 54-month BAT all-sky maps at a lower significance level. As an effort towards increasing the list of objects emitting both in the BAT and in the LAT energy ranges, we consider the adoption of a lower threshold of the detection significance. Taking into account the results obtained by Maselli et al. (2010) in the cross-correlation of the BZCAT Blazar Catalogue (Massaro et al. 2009) with the 39-month 15–150 keV BAT map (Cusumano et al. 2010a) we adopt ${\mathit{\sigma }}_{\mathrm{T}}^{\mathit{\star }}\mathrm{=}\mathrm{3}$. From their analysis, carried out at |b| > 10°, Maselli et al. (2010) find that the adoption of ${\mathit{\sigma }}_{\mathrm{T}}^{\mathit{\star }}$ follows in a fraction of  ~ 3% of related spurious associations, but this value is presumably underestimated at |b| < 10° owing to the higher density of sources.

A total of 1043 and 408 objects are found in the 1FGL catalogue at high (| b| > 10°) and low Galactic latitude, respectively. We extracted the value of the detection significance in the 54-month 15–150 keV BAT all-sky map at their positions and find $\mathit{\sigma }\mathrm{\ge }{\mathit{\sigma }}_{\mathrm{T}}^{\mathit{\star }}$ for 80 objects at |b| > 10° and 49 objects at |b| < 10°. In a few of these objects, the σ value may be strongly biased by some close, very bright hard X-ray sources. Following the same criterion as adopted in Maselli et al. (2010), we excluded the 1FGL sources whose position is found within 36′ from these bright BAT sources and no overlap, even marginal, is found between the corresponding error regions. After this screening we obtain a sample of 75 (| b| > 10°) and 29 (|b| < 10°) 1FGL sources. All the sources listed in Tables 1 and 2 are included in this sample with nine exceptions, all of them at high Galactic latitude. They concern 1FGL sources for which the 95% confidence ellipse is particularly wide and where the 2PBC source is at a considerable distance from the centre of the ellipse. The final list of additional sources with respect to those obtained from the cross-correlation between the 1FGL and 2PBC catalogues has been split according to the Galactic latitude and reported in Tables 3 and 4. For each source the association and the classification type indicated in the 1FGL catalogue is reported, together with the BZCAT blazar classification.

4. Properties of the resulting samples of sources

Fig. 2 Histograms of the redshift distributions for BL Lac objects (left panel) and flat spectrum radio quasars (right panel). The contribution of 1FGL sources detected with a significance down to ${\mathit{\sigma }}_{\mathrm{T}}^{\mathit{\star }}\mathrm{=}\mathrm{3}$ in the BAT 15−150 keV all-sky map (filled columns) is compared with that of the whole corresponding population reported in the BZCAT (empty columns). The contribution of sources included in the 2PBC catalogue is emphasised by an unspotted colour.

The firm correspondences that we find by cross-correlating the 1FGL and 2PBC catalogues and verifying the agreement between the identified counterparts lead to 62 sources. This number rises to 104 (~7% of all the 1FGL objects) when considering all the sources detected with a significance down to ${\mathit{\sigma }}_{\mathrm{T}}^{\mathit{\star }}\mathrm{=}\mathrm{3}$, therefore the number of sources emitting in both the energy ranges covered by the BAT and the LAT instruments aboard Swift and Fermi, respectively, is small. This is a clear indication that the emission in the hard X-ray and in the γ-ray sky is dominated by sources of a different nature. This result is expected, because among extragalactic sources the main contribution in the 1FGL catalogue is given by blazars, while in the 2PBC catalogue it is given by Seyfert galaxies. Considering the 1FGL extragalactic sources provided with an association, the blazar contribution is given by 295 BL Lacs and 274 FSRQs: active galaxies with uncertain classification are 92, while non-blazar active galaxies are only 28. Conversely, in the 2PBC catalogue there are 307 Seyfert 1 and 165 Seyfert 2 galaxies. Even if considerable, the blazar contribution (97 objects) is less relevant.

The group of 104 objects that we have collected is made up of 82 extragalactic and 15 Galactic sources, while seven objects are unidentified. The largest number of extragalactic sources are given by blazars with only a very few exceptions: the Seyfert 1.2 galaxy ESO 323−77, the FR II radiogalaxy 3C 111, the flat spectrum radio source PKS 0336−177 and the starburst galaxies M 82 and NGC 4945. At a low Galactic latitude, we find six pulsars (including Crab and Vela), four high-mass X-ray binaries (including Cyg X-3), a low-mass X-ray binary, the supernova remnant Cas A, two cataclysmic variable stars, and the peculiar object Eta Carinae. We briefly report the results obtained after considering the even lower value of the significance threshold σ_T = 2. We find 74 more sources, the largest fraction of which are blazars (16 BZB, 18 BZQ, and 9 BZU), followed by other AGNs (among which is the well known FR I radiogalaxy M87), the starburst galaxy NGC 253, and a few pulsars; 14 sources are unidentified. We note that, according to Maselli et al. (2010), the number of spurious correspondences for such a low σ_T value is  ~20%.

We focused on the blazar content of the 1FGL-2PBC sources and investigate their distance. We followed the classification of the second edition of the BZCAT catalogue (Massaro et al. 2010), which includes 1164 BL Lac objects (indicated in the following with the suffix “B”), 1660 flat spectrum radio quasars (suffix “Q”), and 262 blazars with uncertain classification (suffix “U”). Note that a z value is available for 531 BL Lacs and therefore the others were discarded in this analysis. We computed the mean redshift value of the different classes of 1FGL-2PBC blazars: the results are reported in Table 5. The values  ⟨ z₁ ⟩  refer to the groups of blazars obtained from the cross-correlation of the 1FGL and 2PBC catalogues, while  ⟨ z₂ ⟩  refer to the larger groups obtained by adopting ${\mathit{\sigma }}_{\mathrm{T}}^{\mathit{\star }}\mathrm{=}\mathrm{3}$ as a threshold for the detection significance in the BAT 15−150 keV all-sky map. Values in brackets refer to sources characterised by redshift estimates that are firmly established and different from zero. We note that five among the 24 BL Lacs detected above ${\mathit{\sigma }}_{\mathrm{T}}^{\mathit{\star }}$ do not have any redshift estimate, two of which are included in the 2PBC catalogue. We compared these results with the values  ⟨ z ⟩  that characterise the blazar subclasses as a whole, computed by considering the totality of sources classified in the BZCAT.

We plot in Fig. 2 the histograms of the redshift distributions for BL Lacs (left panel) and flat spectrum radio quasars (right panel) for these groups of sources. Our results show that the subsample of blazars emitting both in the BAT and in the LAT energy bands is made up of sources that are relatively closer than average to the observer. In fact,  ⟨ z₂ ⟩  is lower than the mean redshift value  ⟨ z ⟩  computed for all the different blazar subclasses, as reported in Table 5. The analysis of the plot in the left hand panel of Fig. 2 shows that the largest part of BL Lac objects coming from the cross-correlation of 1FGL and 2PBC catalogues has a redshift z_B < 0.2, while the modal value of the BL Lac distribution is in the range 0.2 < z_B < 0.3. Adding sources with a detection significance down to ${\mathit{\sigma }}_{\mathrm{T}}^{\mathit{\star }}\mathrm{=}\mathrm{3}$ confirms this result, where most have z_B < 0.1, with the remaining sources more or less equally distributed at higher redshift up to z_B = 0.7. An analogous result is found for flat-spectrum radio quasars (Fig. 2, right panel): high-energy sources emitting both in the BAT and in the LAT energy bands have redshift z_Q < 3.2, with a peak in the range 0.5 < z_Q < 1. The addition of sources with a detection significance down to ${\mathit{\sigma }}_{\mathrm{T}}^{\mathit{\star }}\mathrm{=}\mathrm{3}$ adds seven more sources to this peak, and none with a redshift higher than z_Q = 2. Conversely, the modal value of the distribution of all the FSRQs catalogued in the BZCAT is higher than this peak, in the range 1.25 < z_Q < 1.5.

5. New possible associations of sources in the BAT and LAT surveys

The high-energy emission revealed by the BAT and the LAT telescopes is very helpful in classifying already known sources more correctly. Moreover, it can lead to the discovery of new blazars in regions where their identification is complicated by the belt surrounding the Galactic plane intercepting the line of sight towards them. We focused our attention on two 1FGL unidentified sources, 1FGL J0137.8+5814 and 1FGL J2056.7+4938, detected at low Galactic latitude and included in Tables 2 and 4. For each source we built the SED by adding the data obtained from our reduction of Swift and XMM-Newton pointed observations to all the data that we found in the literature.

Fig. 3 The BAT 15–150 keV all-sky map (60′ × 60′) of the detection significance in the field of 1FGL J0137.8+5814. The positions of the Fermi-LAT and the INTEGRAL-IBIS detections with the corresponding uncertainty regions are plotted with black and a white dashed lines, respectively. The positions of the soft X-ray detections (ROSAT: yellow cross; XMM-Newton: green cross; Swift-XRT: blue cross) are very close to each other and practically coincident to that of the optical counterpart.

Swift-XRT observations, carried out using the most sensitive photon-counting readout mode (see Hill et al. 2004, for a description of readout modes), are available for both 1FGL J0137.8+5814 and 1FGL J2056.7+4938. The highest detected count rate for each source is, in all cases, lower than the pile-up threshold (0.5 cts s^-1). We reduced the data with the HEASOFT 6.8 package distributed by the NASA High Energy Astrophysics Archive Research Center (HEASARC). All the files necessary for the spectral analysis were obtained using the xrtpipeline and the xrtproducts tasks. For each source two Swift observations, characterised by very similar values for the count rate, are available; therefore, we decided to sum the two event files using the xselect task and the exposure maps to correct for vignetting and for CCD hot and damaged pixels. We used the xrtcentroid task to detect the centroid of the source and the corresponding positional error. A nearby source-free region was chosen for extracting the background spectrum. Spectral data of sources were extracted in circular regions surrounding the centroid, adopting a radius of 20 pixels (1 pixel = 2.36′′) for the source spectrum and of 50 pixels for the background spectrum.

A pointed observation from XMM-Newton is available for 1FGL J0137.8+5814. We restricted our analyses of this observation to MOS1 and MOS2, discarding PN data because the source is located at the edge of the instrument’s FoV. We used the standard analysis software SAS 10.0 to extract high-level science products from the ODF files. The source spectrum is extracted by selecting all the events with PATTERN  ≤ 12 (restricting the patterns to single and doubles) and FLAG  = 0 within a circle of 40′′ radius centred on the source. Similarly, the background was extracted by collecting all the counts within an annulus, centred on the source, with inner and outer radii of 45′′ and 85′′, respectively. We combined the instrumental channels in the spectral files to include at least 20 counts in each new energy bin for the Swift observation and 25 counts in the case of XMM-Newton. The spectral analysis was carried out using XSPEC 12.5.1n.

The flux density of both sources in the hard X-ray domain (20 keV) was obtained converting their count rate in the 15−30 keV map of the 54-month BAT survey. The conversion factor was calculated from the count rate of Crab and its spectrum used for calibration purposes, as explained in the BAT calibration status report¹.

The measurements available for these objects at different frequencies are mostly non-simultaneous, and there are not enough of them to allow the possibility of analysing their intrinsic variability at different epochs in detail. We evaluated their general emission properties by selecting opportune energy ranges and, whenever possible, by fitting the corresponding data with analytical models to estimate some relevant parameters. The data in the radio band are well fitted by a power-law model S(ν)    ∝    ν^−α_r to derive their spectral index α_r. We adopted a log-parabolic model to fit the curvature of both synchrotron and inverse Compton components that characterise the SED of blazars. This model, expressed by the analytical formula $\mathit{\nu F}\mathrm{\left(}\mathit{\nu }\mathrm{\right)}\mathrm{=}{\mathit{\nu }}_{\mathrm{p}}\hspace{0.17em}\mathit{F}\mathrm{\left(}{\mathit{\nu }}_{\mathrm{p}}\mathrm{\right)}\mathrm{\times }{\mathrm{10}}^{\mathrm{-}\mathit{\beta }\hspace{0.17em}\mathrm{(}\mathrm{Log} \mathit{\nu }\mathit{/}{\mathit{\nu }}_{\mathrm{p}}{\mathrm{)}}^{\mathrm{2}}}\mathit{,}$(1)provides the peak frequency ν_p of the component and the parameter β that describes its curvature at the peak. This law reproduces curved spectra with a small number of parameters and has been verified to fit the broad band spectra of blazars well (Landau et al. 1986; see also Massaro et al. 2004a,b, for an interpretation in terms of statistical acceleration).

We note that the Galactic latitude of these sources is very low. For this reason the flux density in the optical band may be severely affected from uncertainties of the Galactic extinction values, which are not supposed to be fully reliable. As regards the γ-ray data, derived from the 1FGL catalogue, we consider the possibility of contamination due not only to the γ-ray background but also to the occasional presence of neighbour sources in the field as for the case of 1FGL J0137.8+5814, with a pulsar at an angular distance of  ~12′.

5.1. 1FGL J0137.8+5814

We assume from the 1FGL catalogue the position of 1FGL J0137.8+5814 (; ) with an error radius , and the Galactic latitude is . The analysis of the 54-month BAT all-sky map of the detection significance shows a considerable hard X-ray emission at this position, but no source has been included in the 2PBC catalogue (Cusumano et al. 2010b) because its detection significance (3.4σ) is below the threshold σ_T = 4.8 established in this catalogue. In this case, determining the position of the source is a delicate task, so we report the coordinates of the pixel with the locally higher value of the detection significance (3.5σ): and . A hard X-ray detection in this region of the sky by INTEGRAL was first included in the all-sky survey by Krivonos et al. (2007). It is also reported in the 4th IBIS/ISGRI catalogue (Bird et al. 2010) with , and with an error radius of 5′, thus shrinking the region needed to search for a counterpart to the Fermi γ-ray source. The variability of this object in the 20−40 keV band has been recently assessed by Telezhinsky et al. (2010), where it is still classified as “unidentified” and as a transient source detected on the intrinsic variance map but not on the significance map.

At lower energies, a source in the LAT and IBIS error boxes has been reported in the 1RXS catalogue (Voges et al. 1999) at , with an error radius of 9′′. Later on, an X-ray observation performed by XMM-Newton on January 16, 2003 and aimed at observing the pulsar PSR 0136+57 revealed a serendipitous source that was included in the 2XMMi catalogue (Watson et al. 2009) with coordinates , Dec = +58°14′10′′ and an error radius of 1′′. We analysed this XMM-Newton observation, characterised by a long exposure time t_exp = 8439 s, and considered the results obtained by the combined analysis of MOS1 and MOS2 detectors. We fit the spectrum with a power-law model and obtain a photon index Γ = (2.31 ± 0.06). The value of the hydrogen column density N_H = (4.96 ± 0.18) × 10²¹ cm^-2 that we obtain when leaving this parameter free to vary is very similar to the Galactic one (N_H = 4.01 × 10²¹ cm^-2) as reported in the Leiden/Argentine/Bonn (LAB) Survey (Kalberla et al. 2005). The source flux is 1.04 × 10^-11 erg cm^-2 s^-1 in the 2−10 keV band and is 1.5 × 10^-11 erg cm^-2 s^-1 in the 0.2−12 keV band. We note the lower value of the flux 1.8 × 10^-12 erg cm^-2 s^-1 in the 2−10 keV band reported by Stephen et al. (2010) and attribute it to the different value of the photon index Γ = 1.7 that they adopted in their fit.

Two Swift observations have been recently obtained in this region: the XRT exposure of the first observation (2010 September 4) is 1153 s, while a longer exposure of 3393 s is available for the second observation (2010 October 22). Across this period the source has not shown appreciable variations in activity, with a stable count rate around  ~ 1.7 × 10^-1 cts s^-1. The position of the XRT source is and with an error radius 3.61′. We fit the obtained spectrum with a power-law model and find ${\mathit{\chi }}_{\mathit{r}}^{\mathrm{2}}$ / d.o.f. = 0.96 / 35. As for the XMM-Newton observation, the obtained value for the hydrogen column density N_H = (4.75 ± 0.49) × 10²¹ cm^-2 can be considered consistent with the Galactic one. The value of the photon spectral index is Γ = (2.21 ± 0.14), while the obtained value for the 2–10 keV flux is 5.67 × 10^-12 erg cm^-2 s^-1, which is nearly half the value measured by XMM-Newton in 2003.

Possibly interesting radio counterparts are in the 87GB and NVSS (Condon et al. 1998) catalogues (; ) with flux densities of F_5   GHz = 136 mJy and F_1.4   GHz = 170 mJy, respectively, from which we derive a spectral index α_r    ≃ 0.28. The image in the NVSS survey shows a marginally extended source with a very compact core.

Optical observations (Bikmaev et al. 2008) aimed at identifying the five INTEGRAL sources reported by Krivonos et al. (2007), performed with the Russian-Turkish 1.5-m RTT-150 and the 6-m BTA telescopes, led to the discovery of an object having a continuum without emission or absorption lines, located at and , fully compatible with the radio and X-ray position and at a separation of  ~0.8′ from the γ-ray centroid. Photometric data of the object in the field are available from the Sloan Digital Sky Survey (SDSS-DR8). The morphological classification is that of a starlike source and the magnitudes are r = (18.12 ± 0.01) mag, g = (19.05 ± 0.01) mag, and u = (20.04 ± 0.05) mag, with a reddening in the r band equal to 1.47 mag. This implies a correction to the u−r colour index of  ~1.2 mag, and the intrinsic colour index would be  ~0.7 mag, corresponding to a spectral distribution with a strong excess at blue wavelengths, typical of BL Lac objects.

Fig. 4 The SED of 1FGL J0137.8+5814. From lower to higher frequencies we report radio data up to 5 GHz (red circles) and data from 2MASS (orange circles), SDSS-DR8 (yellow circles), Swift-XRT (light green circles), XMM-MOS (dark green circles), Swift-BAT (cyan square), Integral-IBIS (blue squares), and Fermi-LAT (magenta squares). Data are fitted with a power-law model in the radio band; a log-parabolic model is used to emphasise the synchrotron and inverse Compton component of this high-energy synchrotron peak (HSP) BL Lac object.

Considering all these available data, we obtain the SED of this source shown in Fig. 4. It looks very similar to an SED for an LAT blazar (Abdo et al. 2010c) with the two broad bumps associated with the synchrotron and inverse Compton emission. We estimate the peak frequency ν_p and the curvature parameter β of both components by fitting a log-parabolic law to the data. The observations of the X-ray telescopes, which monitor the region close to the synchrotron peak in two different epochs, have shown relevant variability in emission from this source. The results of the log-parabolic fit that includes Swift data are ν_S = 1.66 × 10 ^+ 16 Hz for the peak frequency with a corresponding maximum of synchrotron emission ν_S F(ν_S) = 6.25 × 10^-12 erg cm^-2 s^-1, and the estimated value of the curvature parameter is β_S1 = 0.08. The log-parabolic fit that includes XMM-Newton data provides a higher value (ν_S = 6.69 × 10 ^+ 16 Hz) for the synchrotron peak frequency and a corresponding maximum of synchrotron emission ν_S F(ν_S) = 8.30 × 10^-12 erg cm^-2 s^-1. A very similar curvature (β_S2 = 0.07) was also found in this case. As regards the inverse Compton component, we obtain ν_IC = 1.78 × 10 ^+ 21 Hz with a corresponding ν_IC F(ν_IC) = 7.78 × 10^-12 erg cm^-2 s^-1, and the curvature parameter is β_IC = 0.07. Both the flux at the peak frequency and the curvature parameter that we derive from our log-parabolic models are therefore very similar for the synchrotron and the inverse Compton components.

A BL Lac classification for this source, which was first suggested by Bikmaev et al. (2008), has been recently confirmed by Stephen et al. (2010). Following the classification scheme reported in Abdo et al. (2010c) we can conclude from our analysis that this object is a high-energy synchrotron peak (HSP) BL Lac object with a ν_S > 10 ^+ 16 Hz in different detected states of the source’s activity.

5.2. 1FGL J2056.7+4938

The position of 1FGL J2056.7+4938 (, , corresponding to the Galactic latitude ), can be assumed with an error radius from the 1FGL catalogue. In the 2PBC there is the source 2PBC J2056.5+4938 close to this location at , with an error radius  ~ at the 95% confidence level. The significance of the detection is 5.5σ in the 15−150 keV map and 6.3σ in the 15–30 keV map. Earlier than Swift-BAT, a detection in the hard X-ray band in the proximity of 1FGL J2056.7+4938 was obtained by INTEGRAL-IBIS. The source, first reported by Krivonos et al. (2007) with the name IGR J20569-4940, was left without classification, and the detection was later confirmed by Bird et al. (2010) in their 4th IBIS/ISGRI soft gamma-ray survey catalogue.

The counterpart of this high-energy emission can be searched by shrinking the region of the sky with circles centred on the positions of Fermi-LAT, Swift-BAT and INTEGRAL-IBIS centroids with a radius proportional to their errors, respectively. The radio source 4C +49.35, classified as symmetric double in NED, is found within the intersection of these circles. Radio measurements at 1.4 GHz reported in the NVSS catalogue (Condon et al. 1998) resolve two components separated by  ~3′: a north-east component (, ) with a flux density F_NE = 167 mJy and a south-west component (, ) with F_SW = 124 mJy.

A soft X-ray emission was detected in this region for the first time by the Ariel V satellite and reported by Warwick et al. (1981) with the name 3A 2056+493. Later detections by ROSAT and, more recently, by Swift and XMM-Newton are characterised by adequate precision to associate the soft X-ray emission with the north-east component. Two detections from XMM-Newton have been reported in the slew survey clean source catalogue (Saxton et al. 2008). Their positions are very similar and separated by  ~2.5′′ from each other: , for the first one and ; for the second one. The error on the position is 8′′ at the 1σ confidence. Even though the two slews were carried out along the same day (November 03, 2007) flux variations in the 0.2−12 keV band have been reported in the catalogue, dropping from (1.85 ± 0.31) to (0.83 ± 0.25) × 10^-11 erg cm^-2 s^-1 in a few hours.

Two soft X-ray pointed observations were performed by Swift on 2006 February 26 and 2009 March 03. The first one has a much longer exposure (8377 s) than the other (1439 s). We stack the two observations and obtain ; for the position of the X-ray detection, with a precision of r = 3.54′. We carried out a spectral analysis that adopts a power-law model. Leaving all the parameters free to vary, we obtain N_H = (1.62 ± 0.09) × 10²² cm^-2, Γ = (2.43 ± 0.08), and ${\mathit{\chi }}_{\mathit{r}}^{\mathrm{2}}$/d.o.f. = 0.94/97. The 2−10 keV flux is 1.19 × 10^-11 erg cm^-2 s^-1. We note that, unlike our result, Landi et al. (2010) report a value of the hydrogen column density N_H = (0.53 ± 0.18) × 10²² cm^-2 lower than the Galactic contribution (N_H = 1.0 × 10²² cm^-2) quoted in the LAB Survey (Kalberla et al. 2005). We repeated the fit and fixed the N_H parameter to the Galactic one, but we do not find an acceptable variation in the χ² value (${\mathit{\chi }}_{\mathit{r}}^{\mathrm{2}}$/d.o.f. = 1.61/96). A broad band spectral fit of data from the radio to the X rays with a log parabola provides a curvature parameter β_S = 0.07 (Fig. 5). Therefore we repeated the fit of the X-ray spectrum with a log-parabolic model by fixing β = 0.07 and also the hydrogen column density at a moderately higher value (N_H = 1.2 × 10²² cm^-2) than the Galactic one; in this way, we obtain the acceptable result${\mathit{\chi }}_{\mathit{r}}^{\mathrm{2}}$ / d.o.f. = 1.14/94.

Fig. 5 The SED of 1FGL J2056.7+4938. From lower to higher frequencies we report radio data up to 22 GHz (red circles) and data from 2MASS (orange circles), Swift-XRT (light green circles), Swift-BAT (cyan square), Integral-IBIS (blue squares) and Fermi-LAT (magenta squares). Data are fitted with a power-law model in the radio band; a log-parabolic model is used to emphasise the synchrotron and inverse Compton component of this high-energy synchrotron peak (HSP) BL Lac object.

The association between the radio and the soft X-ray emission can be considered firmly established by the short distance between the Swift and the XMM-Newton centroids, on the one hand, and the north-east component of 4C +49.35, on the other. These are very close to the bright star SAO 50269 (; ; R₂ = 8.39 mag in USNO-B1, Monet et al. 2003). The UVOT image in the UV-W2 filter does not show any evidence of sources in the proximity of the star, for which we obtain a magnitude W2 = (13.12 ± 0.05) mag. Both images acquired in the U filter are saturated by the star. In a study of the low-latitude sample area in Cygnus of the ROSAT Galactic plane survey, Motch et al. (1997) associated the X-ray source RX J2056.6+4940 with SAO 50269 with a probability for a random association of about 10%. The distance between the X-ray and the star position is  ~13′′ within the estimated ROSAT accuracy that is  ~25′′. The origin of the X-ray emission was related to an active corona surrounding this star, as for the 85% of sources with a PSPC count rate higher than 3 × 10^-2 cts s^-1 belonging to their sample. Recently, Haakonsen & Rutledge (2009) report a low probability of any association between 1RXS J205644.3+494011 and the star in their statistical cross-association of the ROSAT Bright Source Catalogue (Voges et al. 1999) and the 2MASS Point Source Catalogue (Cutri et al. 2003). As reported by Landi et al. (2010), the counterpart of this high-energy source is, with high probability, an object located at and at only from SAO 50269. It has been detected in the 2MASS infrared images retrieved from the NASA/IPAC infrared Science Archive. The magnitude in the J filter is 13.69 mag and is uncertain, whereas in the H and K filters they are (14.30 ± 0.10) mag and (13.74 ± 0.08) mag, respectively. The emission of this object in the optical as well as in the UV band is most likely overwhelmed by that originating from the star. This source is separated from the γ-ray centroid by .

Paredes et al. (2002) cited the ROSAT source 1RXS J205644.3+494011 in a search for microquasar candidates at low Galactic latitudes, which was obtained by means of a cross-identification between the ROSAT bright source catalogue (RBSC, Voges et al. 1999) and the NVSS catalogue. These authors gave a priority to sources for which the offset between the X-ray and the radio position was within the 1σ RBSC position error; unfortunately, this was not the case for 1RXS J205644.3+494011. Recently, the hypothesis that the nature of this source might be extragalactic has been considered by several authors (Landi et al. 2010; Voss & Ajello 2010; Stephen et al. 2010). We collected for the first time all the available data in the literature to build the SED of this object, which is reported in Fig. 5, to investigate the broad-band emission properties of 1FGL J2056.7+4938. We exclude the radio measurements reported by surveys with a low resolution to avoid the risk of including a spurious contribution from the SW component of 4C +49.35. We estimate a radio spectral index α_r    ~    0.24 from the fit of data up to 22 GHz (Petrov et al. 2007), and this value of α_r is indeed typical of flat spectrum radio sources. An evaluation of correct fluxes both at optical and at infrared wavelengths is difficult not only for the near star but also for the high value of the reddening E(B−V) = 2.85 mag (Schlegel et al. 1998), which may be uncertain for such a low value of the Galactic latitude. We fit the data from radio to X-ray frequencies with a log-parabola and obtain the peak frequency ν_S = 3.86 × 10 ^+ 16 Hz, the corresponding maximum of synchrotron emission ν_S F(ν_S) = 1.28 × 10^-11 erg cm^-2 s^-1 and the curvature parameter β_S = 0.07. In carrying out our fit, we force the log-parabola profile to agree with the power-law fit to the radio data and assume E(B−V) = 1 mag, consistently lower than the value quoted by Schlegel et al. (1998). The uncertain infrared measurement in the J filter was omitted. We also carried out a fit of high-energy data from the hard X-ray to the γ-ray band and obtain ν_IC = 6.46 × 10 ^+ 21 Hz with a corresponding ν_IC F(ν_IC) = 8.32 × 10^-12 erg cm^-2 s^-1 and a curvature parameter β_IC = 0.04.

More precise measurements are needed to characterise both emission components of this source. Nevertheless, the analysis of its broad-band properties, such as the radio spectral index, the low Compton dominance, and the synchrotron peak frequency higher than 10¹⁶ Hz, support the classification of 1FGL J2056.7+4938 as a blazar, and in particular as an HSP BL Lac object, but only the analysis of the optical spectrum can confirm this interpretation.

6. Conclusions

We have reported the results of our analysis devoted to research of the sources that show high-energy emission in both the Swift-BAT and the Fermi-LAT telescopes, according to the data collected by LAT over the first 11 months of operation and by BAT over a much longer period of 54 months. As expected, we only found very few sources: only 7% of those included in the 1FGL catalogue are characterised by any significant emission (σ > 3) as shown by the 54-month BAT 15–150 keV map. The larger fraction of them include extragalactic objects, and the dominant part is represented by blazars. We investigated their redshift distribution in greater detail making a distinction among its subclasses. Comparison with the distribution of the whole population classified in the second edition of the BZCAT catalogue shows that they are mainly closer than average to the observer.

Driven by the detection from both the BAT and the LAT instruments, we focused on a couple of objects with very low Galactic latitude. Our detailed analysis of their broad-band SED supports the classification of these objects as blazars, and in particular as high synchrotron peaked (HSP) BL Lac Objects. Analyses of the same kind can be carried out with success for other sources that are expected in the near future with the release of the second Fermi-LAT point source catalogue, based on two years of γ-ray data, and also a new Palermo BAT catalogue from analysing six years of BAT data.

Acknowledgments

The authors are grateful to the referee for the suggestions and comments that helped to improve the manuscript. They acknowledge financial support by ASI/INAF through contract I/011/07/0. Part of this work is based on archival data, software, or online services provided by the ASI Science Data Center (ASDC) and by the SIMBAD database operated at the CDS, Strasbourg, France.

References

All Tables

All Figures

Fig. 1 Details of the BAT 15–150 keV all-sky map of the detection significance σ at the position of a) 1FGL J1103.7−2329 (left panel), b) 1FGL J2056.7+4938 (middle panel), and c) 1FGL J0238.3−6132 (right panel). A dashed line represents the position of the 1FGL (black ellipse) and the 2PBC (white circle) sources with its uncertainty. For each source, the position of the associated counterpart is indicated by a cross of the corresponding colour. Each square map is  ~60′ × 60′; the colour scale is optimised for each hard X-ray source.

In the text

Fig. 2 Histograms of the redshift distributions for BL Lac objects (left panel) and flat spectrum radio quasars (right panel). The contribution of 1FGL sources detected with a significance down to ${\mathit{\sigma }}_{\mathrm{T}}^{\mathit{\star }}\mathrm{=}\mathrm{3}$ in the BAT 15−150 keV all-sky map (filled columns) is compared with that of the whole corresponding population reported in the BZCAT (empty columns). The contribution of sources included in the 2PBC catalogue is emphasised by an unspotted colour.

In the text

Fig. 3 The BAT 15–150 keV all-sky map (60′ × 60′) of the detection significance in the field of 1FGL J0137.8+5814. The positions of the Fermi-LAT and the INTEGRAL-IBIS detections with the corresponding uncertainty regions are plotted with black and a white dashed lines, respectively. The positions of the soft X-ray detections (ROSAT: yellow cross; XMM-Newton: green cross; Swift-XRT: blue cross) are very close to each other and practically coincident to that of the optical counterpart.

In the text

Fig. 4 The SED of 1FGL J0137.8+5814. From lower to higher frequencies we report radio data up to 5 GHz (red circles) and data from 2MASS (orange circles), SDSS-DR8 (yellow circles), Swift-XRT (light green circles), XMM-MOS (dark green circles), Swift-BAT (cyan square), Integral-IBIS (blue squares), and Fermi-LAT (magenta squares). Data are fitted with a power-law model in the radio band; a log-parabolic model is used to emphasise the synchrotron and inverse Compton component of this high-energy synchrotron peak (HSP) BL Lac object.

In the text

Fig. 5 The SED of 1FGL J2056.7+4938. From lower to higher frequencies we report radio data up to 22 GHz (red circles) and data from 2MASS (orange circles), Swift-XRT (light green circles), Swift-BAT (cyan square), Integral-IBIS (blue squares) and Fermi-LAT (magenta squares). Data are fitted with a power-law model in the radio band; a log-parabolic model is used to emphasise the synchrotron and inverse Compton component of this high-energy synchrotron peak (HSP) BL Lac object.

In the text