© ESO, 2015

1. Introduction

A key strategic objective of the Fermi mission is to perform a survey of the sky at various gamma-ray energies. Recently, the first Fermi/LAT catalogue of high-energy sources (1FHL) has been published (Ackermann et al. 2013), listing 514 objects detected above 10 GeV. The main motivation behind this catalogue was to find the hardest gamma-ray sources in the sky and to get a sample of objects that are good candidates for detection at TeV energies. The vast majority of the sources reported therein could be immediately associated with known objects (449 or 87% of the sample): approximately 75% with active galactic nuclei (AGNs), mostly blazars, while Galactic sources, like pulsars, pulsar wind nebulae (PWNs), supernova remnants (SNRs), high-mass binaries, and star-forming regions, collectively represent 10% of the sample. The percentage of unassociated sources is less than 14% corresponding to 71 objects. The third Fermi/LAT catalogue (3FGL, Acero et al. 2015), which was subsequently published, contains most of these unassociated 1FHL sources except for 13 objects that are missing. Considering the associations discussed in the 3FGL catalogue (six objects) and our follow-up work (26 objects analysed by Landi et al. 2015a,b), the set of sources for which no counterpart has been found found contains at this stage 39 entries.

Identification of these objects is a primary objective of the mission, but it has been made difficult by the relatively large positional uncertainty (usually around a few arcminutes) of the Fermi/LAT instrument. Analysis of X-ray data can be a useful tool for restricting the positional uncertainty of the Fermi objects and facilitating the identification process. The list of likely counterparts that are found can then be studied in detail (for example using multi-waveband data) to identify those highly unusual objects with the parameters that might be expected to produce gamma-rays. Optical spectroscopy can finally be performed to confirm the association and classify the gamma-ray source.

Here we use data collected with the XRT instrument onboard the satellite Swift (Gehrels et al. 2004) to study a number of unassociated Fermi 1FHL sources. We do this by cross-correlating the list of 39 1FHL objects that lack a counterpart, with all the XRT pointings covering the Fermi/LAT positional uncertainty. Analysis of these data allows us to propose the likely or possible counterparts to another 12 Fermi high-energy objects; the nature of each likely or possible counterpart is studied by means of a multi-waveband approach and is generally found to conform to the nature of blazars of the BL Lac type.

2. Swift/XRT follow-up observations

We took all 39 sources in the 1FHL catalogue that still lack an association and class. The source positions were then cross-correlated with the Swift/XRT archival data, and we only considered distances between the Fermi source and XRT pointing positions below 9 arcmin in order to optimise coverage of the high-energy error ellipses. This cross-correlation method led to selecting a sample of 23 1FHL sources for which XRT data are available. The log of all X-ray observations analysed in this work is given in Table 1, where we report for each individual Fermi source, the XRT observation ID, the date and the exposure of the XRT pointings available up to March 15, 2015.

XRT data were reduced using the XRTDAS standard data pipeline package (xrtpipeline v. 0.12.9) to produce PC-mode screened event files. For each Fermi source, we summed together all the available XRT pointings using XSELECT v. 2.4c to enhance the signal-to-noise ratio and thus allow the detection of possible counterparts. The XRT images in the 0.3–10 keV energy band were obtained and analysed by means of the software package XIMAGE v. 4.5.1; some of the XRT images are shown in figures from 1 to 4 and discussed in dedicated sections.

From the XRT data we did not find evidence of soft X-rays for ten sources in our sample and these are listed in Table 2, where we also quote an upper limit on the count rate in the 0.3–10 keV energy band and the Galactic neutral hydrogen column density in the source direction.

For each of the remaining objects, XRT detected one or more sources¹ in the region containing or surrounding the Fermi positional uncertainty. These cases are listed in Table 3 and discussed in the following sections. We also provide further X-ray information in Table 4: spectral parameters only for those sources for which the quality of the XRT data is good enough to perform a basic spectral analysis, i.e. a simple power law passing through Galactic absorption. In the remaining cases, we only estimate the 0.3–10 keV count rate and the Galactic neutral hydrogen column density in the source direction.

For the spectral analysis, source events were extracted within a circular region with a radius of 20 pixels (1 pixel ~ 2.36 arcsec) centred on the source position, while background events were extracted from a source-free region close to the X-ray source of interest. The spectra were obtained from the corresponding event files using the XSELECT v. 2.4c software and binned using grppha in an appropriate way, so that the χ² statistic could be applied. We used version v.014 of the response matrices and created individual ancillary response files arf using xrtmkarf v. 0.6.0.

Finally, we point out that only four objects from Table 3 (1FHL J0639.6–1244, 1FHL J1240.4–7150, 1FHL J1507.0–6223, and 1FHL J1856.9+0252) are located close to the Galactic plane (i.e. below 10 degrees in Galactic latitude), suggesting that the remaining nine sources are most likely extragalactic.

3. Individual source studies

In the following, we provide detailed information available in the literature and in various archives on each individual association with the Fermi high-energy sources found in this work. Besides the X-ray information, we also checked the radio, infrared (by exploiting the Wide-field Infrared Survey Explorer (WISE, Wright et al. 2010), and optical characteristics of each possible counterpart to understand its nature in more detail and to assess the likelihood of its association with the Fermi object. More specifically, in Table 3 we report, for each Fermi high-energy source for which we find one or more X-ray counterparts, the name, the number of X-ray sources detected with XRT in the corresponding fields (and next to it a reference number used in the WISE colour-colour plot shown in Fig. 5), their coordinates and error radius, the name of the WISE counterparts, its magnitudes (W1 = 3.4, W2 = 4.6, W3 = 12, W4 = 22 micron)², and the likelihood of association with the corresponding gamma-ray source. It is evident from Table 3 that the XRT location accuracy allows a significant reduction in the Fermi positional uncertainty, thus making the search for possible optical counterparts much easier.

In most cases, the soft X-ray error radius is lower than or equal to six arcseconds, i.e. sufficiently small to highlight only one optical counterpart. Also X-ray information as reported in Table 4 can be used to characterise each source, in particular the count rate, and the 2–10 keV flux can help to distinguish multiple counterparts. We used the WISE colours as discussed by Massaro et al. (2013a) to test the possible blazar nature of each source. These authors found that in the W2 − W3 versus W1 − W2 colour–colour plot the positions of gamma-ray emitting blazars are all within a well-defined region known as the “Blazar Strip”. Therefore, objects with IR colours compatible with this strip could be AGN of the blazar type, hence be an even more likely counterpart to the Fermi high-energy sources. For the radio information, we either use data collected at 20 cm by the NRAO VLA Sky Survey (or NVSS, Condon et al. 1998), at 36 cm by the Sydney University Molonglo Sky Survey (or SUMSS, Mauch et al. 2003) and at 92 cm by the Westerbork Northern Sky Survey (WENSS, Rengelink et al. 1997) or Westerbork In the Southern Hemishpere (WISH, De Breuck et al. 2002).

We also consulted the results of Schinzel et al. (2015), who performed an All Sky survey between 5 (6) and 9 (3.3) GHz (cm) of sky areas surrounding unidentified objects listed in the second Fermi/LAT catalogue (2FGL, Nolan et al. 2012) in the search for new AGN associations with compact radio sources. Nine of our regions were covered by this survey, but only five were found to have a radio detection compatible with our proposed X-ray association. They are discussed in each individual section.

In the following, sources will be divided into three groups: those with a single counterpart, those with multiple associations within or at the border of the Fermi positional uncertainty, and those with a counterpart outside the Fermi error ellipse.

3.1. Single counterpart sources

In this section, we discuss those cases in which we find only one X-ray source inside the Fermi positional uncertainty. This information is used with the overall properties of the X-ray detection found to evaluate the likelihood of the association with the Fermi high-energy object.

3.1.1. 1FHL J0644.2+6036

The XRT counterpart found in this work (source #1 in Fig. 1) is listed both in the NVSS (NVSS J064435+603849) and in the WENSS (B0640.0+6041) catalogues with 20 and 92 cm flux densities of ~33.5 and ~83 mJy, respectively (spectral index α = 0.6)³. The WISE colours are W2 − W3 = 1.97 and W1 − W2 = 0.64, making this object compatible with the blazar strip. The high-energy Fermi source is also reported in the second/third Fermi/LAT catalogues as 2FGL J0644.6+6034/3FGL J0644.6+6035, an unidentified source studied by various authors. Massaro et al. (2013b) propose WISE J064459.38+603131.7 as the possible IR counterpart to 2FGL J0644.6+6034, while Paggi et al. (2013) find that this WISE source is also detected by Swift/XRT. Recently, Paggi et al. (2014) have provided the optical spectrum of this possible counterpart, finding that it could be a weak emission line quasar at z = 0.3582. As is evident in Fig. 1, the positional uncertainties of 2FGL J0644.6+6034, 3FGL J0644.6+6035 and 1FHL J0644.2+6036 are partly overlapping. The error ellipse of 2FGL J0644.6+6034, apart from the source discussed by Massaro et al. (2013b) and Paggi et al. (2014) (source #2), contains also source #1 found in this work, while the error ellipse of 3FGL J0644.6+6035 contains only this XRT detection, which incidentally is also the brightest of the two X-ray emitters. The lack of any radio detection so far for source #2, combined with the above-mentioned characteristics of source #1, raise some questions about the reality of the association between WISE J064459.38+603131.7 and 2FGL J0644.6+6034, and suggest instead that source #1 is the only putative counterpart to the 2FGL, 3FGL, and 1FHL Fermi object. As a confirmation of our findings, we note that source #1 is also reported in the survey work by Schinzel et al. (2015) as a flat-spectrum compact radio source detected with a 6 cm flux density of ~8.2 mJy.

3.1.2. 1FHL J1115.0–0701

This high-energy Fermi source has a counterpart in the still unclassified object 3FGL J1115.0–0701. The only XRT detection found is inside and at the border of the 1FHL and 3FGL error ellipse, respectively. The source has no radio counterpart or WISE colours (W2 − W3 = 1.825 and W1 − W2 = 0.069) that locate it outside the blazar strip. Both indications cast doubts on the association between this XRT detection and the Fermi high-energy source, which therefore we consider at this stage uncertain until optical follow-up observations provide a reliable classification of this X-ray source.

3.1.3. 1FHL J1129.2–7759

This XRT source has a counterpart in the ROSAT Bright (1RXS J113029.0–780108, 7 arcsec error radius) source catalogue (Voges et al. 1999), as well as in a XMM-Newton Slew survey catalogue (Saxton et al. 2008) (XMMSL1 J113031.6–780112, 3 arcsec error radius); indeed, its X-ray flux, listed in Table 4, is relatively high.

This X-ray source is most likely associated with the radio source PMN J1130–7801. The PMN object has a 36 cm flux density of ~136 mJy and a flat (α = − 0.4) radio spectrum from 3.5 to 6 cm (McConnell et al. 2012). The source is also reported as a WISE object with the following IR colours: W2 − W3 = 1.93 and W1 − W2 = 0.6, which are fully compatible with the blazar strip. The source is discussed in a study of the Chamaeleon star-forming region (Alcala et al. 2005): it has been spectroscopically investigated, but unfortunately with no success. We also note that this X-ray/radio source falls within the positional uncertainty of the 3FGL counterpart (3FGL J1130.7–7800) to this 1FHL source.

3.1.4. 1FHL J1223.3+7953

This X-ray source is listed in the Two Micron All Sky Survey Extended sources catalogue (Skrutskie et al. 2006) as 2MASX J12235831+7953279 and was classified as a galaxy in NED. The source is also found in both the NVSS (NVSS J122358+795329) and WENSS (WN 1222.0+8010) catalogues with 20 and 92 cm flux densities of ~31.5 and ~43 mJy, respectively. A power-law fit between these two frequencies indicates a rather flat (α ~ 0.2) radio spectrum. This is confirmed by the findings of Schinzel et al. (2015), who report parsec scale radio emission with a flat spectrum for this X-ray source. The WISE colours (W2 − W3 = 1.92 and W1 − W2 = 0.48) place the source well inside the blazar strip. This object is also listed in the Galaxy Evolution Explorer All-Sky Survey (GALEX, Bianchi et al. 2011) as GALEXASC J122357.56+795328.7 with magnitudes of 24.6 and 22.3 in the far- and near-UV. Massaro et al. (2015) have recently identified it as a BL Lac of unknown redshift. We note that this 1FHL source is listed as 3FGL J1222.7+7952 in the third Fermi catalogue, and the XRT source lies at the border of its positional uncertainty confirming 2MASX J12235831+7953279 as the counterpart to both gamma-ray detections.

3.1.5. 1FHL J1240.4–6150

The 3FGL association for this source is 3FGL J1240.3–7149. The only XRT source found is inside the error ellipses of both 1FGH and 3FGL objects. It is associated with the XMM-Newton object XMMSL1 J124021.7–714854 (2.0 arcsec error radius) and displays a relatively high X-ray flux (see Table 4). It has a radio counterpart (MGPS J124021–714901) that shows a ~15.5 mJy flux density at 36 cm (Murphy et al. 2007) and a flat radio spectrum between 5 and 9 GHz (α = − 0.45, Petrov et al. 2013). In this case Schinzel et al. (2015) report parsec scale radio emission emission with 6 and 3.3 cm flux densities of ~14 and ~11.2 mJy, respectively (α = 0.3) for this X-ray source. The source is listed in the WISE catalogue with colours W2 − W3 = 1.34 and W1 − W2 = 0.52, which places it outside the blazar strip.

3.1.6. 1FHL J1315.7–0730

In this case, the XRT source is associated with a relatively weak radio source (NVSS J131552–073301), displaying a 20 cm flux density of ~23.8 mJy and a flat (α ~ − 0.1) radio spectrum (Petrov et al. 2013). Radio emission is also reported by Schinzel et al. (2015) with 6 and 3.3 cm flux densities of ~40.2 and ~38.4 mJy, respectively, confirming the flat radio spectrum. This source is WISE-detected with colours W2 − W3 = 2.27 and W1 − W2 = 0.88, which are again compatible with those of gamma-ray emitting blazars. In UV the source can be associated with a GALEX object (GALEXASC J131553.03–073302.4), which displays variability and high polarization (Bauer et al. 2009; Fujiwara et al. 2012). On the basis of these peculiarities, Massaro et al. (2013b) suggest that it could be a new gamma-ray blazar, while Hassan et al. (2013) conclude that it might be a BL Lac object. This 1FHL source is reported as 3FGL J1315.7–0732 in the third Fermi catalogue. Our XRT detection lies well within the 3FGL positional uncertainty, thus also confirming our proposed association for the lower energy gamma-ray emitter.

3.1.7. 1FHL J1507.0–6223

The only X-ray source detected by XRT lies on the border of the Fermi error ellipse and has no radio counterpart. Furthermore, the WISE colours W2 − W3 = 1.17 and W1 − W2 = − 0.011 place this source well outside the blazar strip. In the third Fermi catalogue, this source is reported as an unclassified source that is nevertheless associated with HESS J1507–622. The three error ellipses (1FHL, 3FGL, and HESS) overlap considerably, but the smaller positional uncertainty of the third Fermi catalogue excludes the XRT object as a possible counterpart. This sky region has been studied in depth at X-rays energies using Chandra, XMM-Newton, and Suzaku (Tibolla et al. 2014; Eger et al. 2015) data. These observations, which are by far more sensitive than ours, revealed several X-ray sources, but not the XRT one discussed here. This lack of detection may be due either to source variability or to different exposure times. Of the two sources that are compatible with the restricted 3FGL error ellipse, one (source X3, Fig. 1 in Tibolla et al. 2014) is a star and the other (source S6, Fig. 1 in Eger et al. 2015) is too faint for an in-depth analysis.

At the present stage the most accredited association with HESS 1507–622 is a faint extended X-ray source thought to be a PWN; although this source is at the border of the HESS and 1FHL positional uncertainties, it is well outside the 3FGL ellipse, which casts some doubt on its true association with the MeV/TeV object. This source clearly deserves further study, but on the basis of the above information, we consider the association between the XRT and the Fermi high-energy source unlikely.

3.2. Multiple counterpart sources

In the following cases, more than one single X-ray detection is found within the Fermi error ellipse. Although this makes things more difficult to handle, it is still possible to highlight a more likely association on the basis of a multi-waveband approach.

3.2.1. 1FHL J0746.3–0225

This is a case where there are two X-ray detections inside the Fermi error ellipse (see upper left panel of Fig. 2), but only one (source #1), which is the brightest, is a likely counterpart. It is in fact the only one with radio emission because it is listed in the NVSS survey as NVSS J074627–022549, with a 20 cm flux density of ~12.8 mJy. It has a WISE counterpart with colours W2 − W3 = 2.1 and W1 − W2 = 0.68, as is typical of gamma-ray blazars. This X-ray source is also UV-detected (GALEXASC J074627.06–022548.9) with far- and near-UV magnitude in the range 20.4–20.8. The other object is dimmer, has no interesting association in other catalogues, and is therefore a less likely association. In the latest Fermi catalogue, 1FHL J0746.3–0225 is associated with 3FGL J0746.4–0225; as is clear in Fig. 2, the 3FGL error ellipse confirms the association of the high-energy detection with source #1 and excludes source #2 as a possible counterpart.

3.2.2. 1FHL J1410.4+7408

Also in this case we find a couple of detections that are compatible with the 1FHL positional uncertainty (see upper right panel of Fig. 2). Both have WISE colours (W2 − W3 = 2.79 and W1 − W2 = 0.88 and W2 − W3 = 1.81 and W1 − W2 = 0.79 for #1 and #2, respectively), so just compatible with the blazar strip, but no detection in radio. We also note that within the 1FHL positional uncertainty, there is also a XMM-Newton Slew source XMMSL1 J141002.6+740744 (5 arcsec error radius), which is quite bright in X-rays (0.2–12 keV flux of 2.3 × 10^-12 erg cm^-2 s^-1); despite this, we could not find a radio, infrared, or optical counterpart, even when doubling its positional uncertainty. Only going to an 18 arcsec distance are we able to find an association with a WISE object. Its WISE magnitudes are W1 = 17.189, W2 = 16.771, W3 = 12.792 and W4 = 9.606, and its colours turn out to be W2 − W3 = 3.98 and W1 − W2 = 0.42, which locate the source outside the blazar strip. Also this counterpart is not reported at radio frequencies. That this XMM-Newton Slew source is detected in neither the combined XRT image nor the individual observations, indicates further that it is extremely variable in X-rays. Despite its potential interest, this XMM-Newton Slew detection is, however, outside the error ellipse of the 3FGL counterpart to this high-energy Fermi source. The other two X-ray sources are instead both well inside the 3FGL positional uncertainty, so are more likely candidates. Clearly, this is another complicated case, and only optical follow-up observations can help to identify which of these two (possibly three) X-ray detections is the true counterpart to this high-energy emitter.

3.2.3. 1FHL J2134.6–2130

In this case three XRT sources are found within the 1FHL error ellipse (see lower left panel of Fig. 2). Source #1 has a radio counterpart (NVSS J213430–213032) with a 20 cm flux density of ~22 mJy and a flat radio index of 0.23 (Petrov et al. 2013). It is also reported in the list of possible compact radio counterparts to 2FGL sources by Schinzel et al. (2015), where it is described as a flat spectrum source with 6 and 3.3 cm flux densities of ~92.8 and ~103.9 mJy, respectively. The WISE colours (W2 − W3 = 2.27 and W1 − W2 = 0.78) are just compatible with the blazar strip and the source is discussed by various authors as a possible counterpart to 2FGL J2134.6–2130; for example, Massaro et al. (2013b) and Hassan et al.(2013) suggest that it could be a blazar of the BL Lac type. Source #2 is listed as a galaxy in NED, but it has no detection in radio, and it does not display WISE colours typical of blazars, so we consider it an unlikely association. As for source #3, the lack of any radio counterpart, as well as of WISE colours (W2 − W3 = 3.213 and W1 − W2 = 0.481) that are not compatible with the blazar strip, suggests that this object is also an unlikely association. The third Fermi catalogue lists this object as 3FGL J2134.5–2131. Its restricted error ellipse excludes source #3 as a likely counterpart, but maintains #1 and #2 as possible associations. On the basis of our analysis, we consider only the first to be the most promising counterpart to this Fermi high-energy detection.

3.2.4. 1FHL J2257.9–3644

Only two soft X-ray counterparts are found within the Fermi error ellipse (see lower right panel of Fig. 2). This 1FHL source is listed in the 3FGL catalogue as 3FGL J2258.2–3645. Even though its positional uncertainty is smaller than in the 1FHL survey, it does not help to exclude one of the two X-ray detections. Source #1 is dim with no counterparts in radio catalogues and WISE colours W2 − W3 = 3.06 and W1 − W2 = 1.46, which locate the source at the upper border of the blazar strip. The other detection (source #2 in the figure) is brighter and has a radio detection in the NVSS (NVSS J225815–364433) and in the SUMSS (SUMSS J225816–364446) with 20 and 36 cm flux densities of ~10.6 and ~14.1 mJy, respectively. The radio spectrum is quite flat displaying an index of 0.23 (Petrov et al. 2013). The source is also reported in the WISE catalogue with colours W2 − W3 = 2.24 and W1 − W2 = 0.57, which are compatible with the blazar strip. It is also listed in the GALEX All-Sky survey (GALEXASC J225815.00–364434.6) with far- and near-UV magnitudes around 20–21. Source #2 is reported as a galaxy in NED and has recently been classified as a BL Lac by Landoni et al. (2015). Optical spectroscopy of source #1 could unambiguously exclude it as the counterpart to this Fermi high-energy emitter.

3.3. Objects with X-ray detection outside the Fermi positional uncertainty

In two cases (1FHL J0639.6–1244 and 1FHL J1856.9+0252), we find a bright X-ray source outside the Fermi error ellipse. Despite this, in each case the association may be considered at least interesting for a number of reasons: a) Fermi error ellipses are quoted at the 95% confidence level, which suggests that in 5% of the cases, the true counterpart could be located outside; b) the associations are with interesting objects, which may have the characteristics of gamma-ray emitting objects; c) there is nothing else inside the Fermi positional uncertainty that could be of potential interest.

3.3.1. 1FHL J0639.6–1244

The only X-ray detection lies around 11 arcmin away from the Fermi position (see Fig. 3). This object is associated with 2MASX J06400717–1253150 and with the flat spectrum radio object PMN J0640–1253 (α ~ − 0.4, Chhetri et al. 2013). This source is fairly bright in radio with a 20 cm flux density of ~225 mJy and is detected by WISE with colours (W2 − W3 = 1.74 and W1 − W2 = 0.457), which are once again fully compatible with the blazar strip. It is also quite bright in X-rays (see Table 4), thus strengthening its association with 1FHL J0639.6–1244.

It was proposed as a BL Lac candidate for TeV emission by Massaro et al. (2013b). The nearest 3FGL source is 3FGL J0640.0–1253, but the two Fermi detections are not associated in the 3FGL catalogue, given that their positional uncertainties do not overlap (see Fig. 3). As is evident from the figure, the 3FGL source is associated with PMN J0640–1253. Seeing as there is no other X-ray emitter in this zone and that the 3FGL source could be expected to radiate even above 10 GeV, it is possible that the 1FHL error circle is slightly underestimated and then the two Fermi objects could well be the same source. We therefore consider the association possible.

3.3.2. 1FHL J1856.9+0252

This source is not reported in the third Fermi catalogue. As displayed in Fig. 4, it is not far from the TeV emitter HESS J1857+026, which is generally classified as a PWN candidate (Acero et al. 2013); the morphology and overall spectral shape of the main emission zone, combined with the proximity to the pulsar PSR J1856+0245 (Hessels et al. 2008), support this classification.

Unfortunately, the two error ellipses (from Fermi and from TeV observations) do not overlap, which makes the possible association between the two wavebands uncertain. Nearby the Fermi error ellipse XRT detects source #1 whose WISE colours (W2 − W3 = 1.91 and W1 − W2 = 1.71) locate it well outside the blazar strip; furthermore, this object has no counterpart in radio surveys. A second source (#2 in Fig. 4) coincides with the pulsar PSR J1856+0245, which is thought to be responsible for the HESS emission. It is difficult at this stage to conclude which of the two sources is the true association with the 1FHL object, nor is it clear whether the HESS and Fermi detection are the same object, leaving this case unresolved. For the following, we consider source #1 as the possible association.

4. Discussion and conclusion

The first result of this work is that we have uncovered a number of likely X-ray counterparts to still unassociated 1FHL gamma-ray sources. The majority of these associations are likely to be extragalactic in nature, most probably blazars of some type. Only in the case of 1FHL J1115.0–0701 is the association considered to be unlikely. In four other cases, the proposed associations are to be taken as possible. More specifically, for 1FHL J1115.0–0701 and 1FHL J1410.4+7408, the X-ray detections are compatible with the 1FHL positional uncertainty, but the overall multi-wavelength properties prevent us from drawing any firm conclusions about their nature and only future optical follow-up observations can help to provide a reliable classification of these X-ray sources. In the other two cases (1FHL J0639.6–1244 and 1FHL J1856.9+0252), the X-ray detection lies outside the 1FHL error ellipse. For 1FHL J0639.6–1244, there is only one bright X-ray source detected in the region surrounding the 1FHL source, which coincides with PMN J0640–1253, a flat spectrum radio source that has been recently associated with the third Fermi catalogue source 3FGL J0640.0–1252. Although the 1FHL and 3FGL error ellipses do not overlap, we suggest that a likely association between PMN J0640–1253 and the 1FHL source cannot be ruled out. Also in the case of 1FHL J1856.9+0252 the overall multi-waveband properties of the only XRT detection do not allow us to associate firmly the X-ray source with the 1FHL object.

In the remaining cases, at least one of the proposed associations has a radio detection and WISE colours compatible with the blazar strip, in other words, it qualifies as beamed AGN. Indeed two objects (1FHL J1223.3+7953 and 1FHL J2257.9–3644) have been optically classified as BL Lac objects. As already noted by Stephen et al. (2010) and confirmed by Masetti et al. (2013), the association of Fermi sources with soft X-ray counterparts favours discovery of this type of blazar. The preference for finding BL Lac when using soft X-ray data is very likely related to the SED of these objects compared to flat spectrum radio quasars. In fact, the X-ray selection favours discovery of high synchrotron peaked or HSP blazars, which are also good candidates for TeV emission, because the Compton peak in these AGN is expected in this energy range owing to the presence of high-energy electrons.

As discussed by Masetti et al. (2013) and in the references therein, these objects populate a well-defined region of the WISE colour-colour diagram, i.e a square located in the lower part of the blazar strip. In fact, if we plot in the WISE colour–colour diagram (see Fig. 5), the most likely counterparts to 1FHL sources listed in Table 3 (i.e. considering only those sources flagged with L or P in the last column of the Table), we notice that most fall along the blazar strip or close to it but, most important, nine of them lie within the limits of the locus populated by TeV-emitting BL Lacs, while one is at its border. Of the two possible counterparts to 1FHL J1410.4+7408, one lies within the TeV square (8b), while the other is on the blazar strip (8a), thus suggesting that both are likely associations. Finally, Figure 5 confirms that the association proposed here for 1FHL J1856.9+0252 (#12 in the figure) remains highly uncertain. We therefore conclude once again that the association among unassociated high-energy Fermi sources and soft X-ray objects appears to select blazars, for a large percentage of the HSP BL Lac type, i.e. those that are good candidates for TeV emission. Clearly, only the optical spectroscopy of all soft X-ray counterparts discussed in this work can confirm this suggestion and provide further insight into gamma-ray selected blazars in general and BL Lac in particular.

---

Acknowledgments

This research made use of the NASA/IPAC Extragalactic Database (NED) operated by the Jet Propulsion Laboratory (California Institute of Technology) and of the HEASARC archive provided by NASA’s Goddard Space Flight Center. The authors also acknowledge the use of public data from the Swift data archive. R.L. acknowledges financial support under contract INTEGRAL ASI I/033/10/0. N.M. thanks Francesco Massaro for help with the preparation of Fig. 5.

References

All Tables

All Figures