Exploring the potential X-ray counterpart of the puzzling TeV gamma-ray source HESS J1507−622 with new Suzaku observations Free

Abstract

1 INTRODUCTION

Multiwavelength (MWL) observations are powerful tools to investigate the nature of unidentified very high energy (VHE, E > 100 GeV) gamma-ray sources detected by ground-based imaging atmospheric Cherenkov telescopes such as HESS, VERITAS or MAGIC (for a review, see e.g. Hinton & Hofmann 2009). In particular, non-thermal X-ray sources are excellent observational tracers for highly energetic particles, and are in many cases linked to VHE gamma-ray sources. Prominent examples of such cases are the supernova remnants (SNRs) Tycho (Decourchelle et al. 2001) and RX J1713.7−3946 (Cassam-Chenaï et al. 2004) as well as pulsar wind nebulae (PWNe; for a review, see Gaensler & Slane 2006), such as MSH 15-52 (Trussoni et al. 1996). Two well-known Galactic VHE gamma-ray sources related to strong X-ray emitters are the PWNe HESS J1825−137 (Aharonian et al. 2005b, 2006b) and Vela X (Aharonian et al. 2006a).

One, typically unknown, key quantity of unidentified gamma-ray sources is their distance. For Galactic sources, observations of their X-ray counterpart may help to constrain their location in the Galaxy. Soft X-rays are absorbed by atomic and molecular interstellar gas along the line of sight, and from the level of this attenuation the total column density (CD) of traversed gas (predominantly hydrogen) can be estimated (see e.g. Wilms, Allen & McCray 2000). If the distribution of atomic and molecular hydrogen in the Galaxy is known (e.g. Dickey & Lockman 1990; Ferrière 2001; Kalberla et al. 2005), constraints on the distance of the source can be placed, based on X-ray spectral measurements.

In the case where the level of soft X-ray attenuation is comparable to the total Galactic neutral hydrogen CD in this direction, the source is likely located at the far end or even outside of the Galaxy. This was the case for the VHE gamma-ray source HESS J1943+213, where the highly absorbed X-ray counterpart pointed towards an extragalactic origin (Abramowski et al. 2011). For sources with a significant angular offset from the Galactic plane, this method can also be used to constrain their physical distance from the disc.

In observations performed by the HESS array of imaging atmospheric Cherenkov telescopes an extended VHE gamma-ray source with an angular off-set of ∼3| $_{.}^{\circ}$|5 from the Galactic plane was discovered (HESS J1507−622; Acero et al. 2011). This source features the second largest angular offset (next to the nearby SNR SN 1006) from the Galactic plane among all VHE gamma-ray sources that or not clearly linked to known extragalactic objects. Despite detailed MWL analyses and theoretical considerations, the nature of this source has not been clearly identified yet. Interestingly, a faint, diffuse X-ray counterpart is likely connected to HESS J1507−622 (see Section 2). The current literature favours a PWN interpretation for this object where the X-ray and VHE gamma-ray emission is produced by ultrarelativistic electrons through synchrotron and inverse-Compton radiation, respectively (see Section 4). However, the distance to HESS J1507−622, and its potentially associated X-ray nebula, is still unknown, but constitutes a key component for all attempts to model the evolution of the source as well as the radiation mechanisms.

In this paper, we present detailed spectroscopic results for this X-ray counterpart, based on newly available Suzaku observations, which can provide a direct estimate of the total CD of hydrogen (N_H) in the line of sight towards this source and thus of its distance.

2 Previous X-ray observations with XMM–Newton and Chandra

Following up on its original detection by HESS, the region around HESS J1507−622 has been observed once with Chandra (Acero et al. 2011) and twice with XMM–Newton (Tibolla, Kaufmann & Kosack 2014).

In the 20 ks Chandra observation (ObsID: 9975), several point-like sources were detected but classified to be most likely unrelated to HESS J1507−622. This Chandra observation also revealed two extended sources. One of them (CXOU J150850.6−621018) is rather bright, with a flux of F_X(2–10 keV) = 7.0 ± 0.7 × 10⁻¹³ erg cm⁻² s⁻¹, but outside of the intrinsic VHE gamma-ray size of HESS J1507−622 (Acero et al. 2011). As briefly outlined by Tibolla et al. (2014), assuming the X-ray source CXOU J150850.6−621018 and the VHE gamma-ray source HESS J1507−622 are physically related, the lack of overlap between the two would be hard to explain, even in an aged PWN scenario. For such an object, despite a much more extended IC/TeV nebula compared to the synchrotron/X-ray nebula, one still would expect IC emission from freshly injected electrons to overlap with the X-ray nebula, which is not seen when comparing CXOU J150850.6−621018 and HESS J1507−622. Prime examples for such aged PWN scenarios are HESS J1825−137 (Aharonian et al. 2005b) and HESS J1303−631 (Aharonian et al. 2005a), where the TeV sources are much more extended but still partly overlap with the more compact X-ray nebulae (see Uchiyama et al. 2009 and Abramowski et al. 2012, respectively). A link between HESS J1507−622 and the offset X-ray source CXOU J150850.6−621018 might thus require an even more evolved, relic PWN scenario.

The second extended X-ray source (CXOU J150706.0−621443) was detected with a statistical significance of ∼7σ by Chandra and is spatially consistent with the VHE gamma-ray emission region. This source is rather faint, with a flux of F_X(2–10 keV) = |$1.1^{+0.3}_{-0.5}\times 10^{-13}$| erg cm⁻² s⁻¹, estimated from the Chandra count rate (Acero et al. 2011). Given the low flux, the limited statistics from the 20 ks Chandra observation did not allow for a detailed spectral study of CXOU J150706.0−621443. The extensions of the sources were estimated to be 20–25 arcsec for CXOU J150706.0−621443 and 35–40 arcsec for CXOU J150850.6−621018, respectively (Acero et al. 2011). Due to its positional coincidence and extended nature CXOU J150706.0−621443 was suggested as a potential counterpart to HESS J1507−622 (see Acero et al. 2011). To account for the very low X-ray flux compared to the flux in TeV gamma-rays, these authors suggested a relic PWN scenario with a very low magnetic field of ∼0.5 μG. Such a low magnetic field leads to a very faint X-ray nebula along with a large accumulation of highly energetic electrons radiating predominantly TeV gamma-rays via the IC mechanism.

Unfortunately, both observations performed with XMM–Newton (ObsIDs: 0556310201, 0651620101) suffered from long periods of strong background flaring activity, rendering a significant fraction of the exposure time unusable for scientific analyses (see Acero et al. 2011; Tibolla et al. 2014). Even the remaining observation time of ≲10 ks featured an increased background level which affected the overall sensitivity, particularly for extended sources. However, both of the above mentioned extended Chandra sources were also detected in the second XMM–Newton observation (Tibolla et al. 2014). CXOU J150706.0−621443, the fainter of the two, was barely above the detection threshold with a statistical significance of ∼4σ. Again, due to the limited number of counts no detailed spectral analysis was possible for CXOU J150706.0−621443. An additional source (XMMU J150835.7−621021; Tibolla et al. 2014), in close proximity (112 arcsec) to CXOU J150850.6−621018, was detected in the second XMM–Newton observation but not seen in the earlier Chandra data despite being above the detection threshold if assuming constant flux. Therefore, Tibolla et al. (2014) concluded that this source is probably variable and propose an X-ray binary or a flaring star as the most likely scenarios.

3 Suzaku data analysis

In 2012 september, the region of HESS J1507−622 was observed with Suzaku with two deep pointings, one of them (sequence no. 507025010, 79.8 ks) centred on HESS J1507−622 the other (sequence no. 507026010, 40.9 ks) centred on CXOU J150850.6−621018 (see Table 1 for more details). We analysed the XIS data using the most recent versions of the headas software package (v6.15.1; Blackburn 1995), and the HEASARC calibration data base. We extracted images and spectra with xselect from the cleaned event lists using the recommended selection criteria (STATUS<524287 && (STATUS%(2**17)<2**16)) to remove events from the ⁵⁵Fe calibration source.

We extracted count images from both front-illuminated detectors (XIS0 and XIS3) in the 1–10 keV energy range. To generate images of the contribution from the non-X-ray background (NXB) in this energy range, we used the tool xisnxbgen (Tawa et al. 2008). After NXB subtraction we corrected the images for the mirror vignetting using a simulated exposure map created with the tool xissim (Ishisaki et al. 2007) at an energy of 2.5 keV. Fig. 1 shows a combined mosaic image of the two Suzaku observations.

Figure 1.

Combined mosaic of Suzaku XIS0 and XIS3 count maps from the two observations in the energy range 1–10 keV, smoothed with a Gaussian kernel with a width of 0.5 arcmin. The dashed boxes (white) indicate the field of view of the two observations. The dashed circle (red) shows the position and intrinsic size of HESS J1507−622. The small boxes show the positions of sources detected with wavdetect. The cross (yellow) indicates the position of the faint diffuse source detected with Chandra and XMM–Newton, and the circle (green) shows the region used to extract the spectrum from CXOU J150706.0−621443. The colour scale is linear and adjusted such that the fainter sources are visible, but S1 is highly saturated.

Open in new tabDownload slide

To detect point-like and extended sources, we used the wavdetect tool from the ciao v4.6 software package (Fruscione et al. 2006), setting the chance probability for a false detection to 1 per cent. The positions of all detected sources are listed in Table 2 and indicated by small squares in Fig. 1. The table also lists likely counterparts from previous Chandra and XMM–Newton observations (Acero et al. 2011; Tibolla et al. 2014). With the exception of two sources (Suzaku J1506.7−6221 and Suzaku J1506.5−6229), all detected objects have counterparts from previous X-ray observations (see Table 2). However, due to their low count rates, no more detailed analyses concerning their spectra or extensions are possible.

Due to the significantly larger point spread function (PSF) of Suzaku two previously detected X-ray sources fall within the range of Suzaku J1508.8−6210: The bright extended source CXOU J150850.6−621018 (Acero et al. 2011), also detected by XMM–Newton with consistent flux (XMMU J150851.1−621017; Tibolla et al. 2014), and the much fainter but variable source XMMU J150835.7−621021 (Tibolla et al. 2014, see also Section 2). With XMM–Newton the flux of XMMU J150835.7−621021 was detected at ∼10 per cent of the flux from CXOU J150850.6−621018. However, due to the apparent variability, the relative flux contributions may be different in this Suzaku observation (for further discussion see below). Due to this potential issue of source confusion, we will use the new Suzaku name for this source whenever we refer to the current analysis.

To test for an extension beyond the Suzaku PSF of Suzaku J1508.8−6210, we extracted its radial profile from the unsmoothed XIS0+3 counts image (1–10 keV). This profile is shown in Fig. 2 and compared to the on-axis PSF at an energy of 4.5 keV as stored in the Suzaku calibration data base. To account for the diffuse astrophysical background component, we subtracted from the data the surface flux level measured at offsets larger than 0| $_{.}^{\circ}$|07 from the source position. The relative normalization between the data and PSF profiles was calculated such that they yield the same integral between 0° and 0| $_{.}^{\circ}$|06. As is evident from Fig. 2, the morphology of Suzaku J1508.8−6210 is incompatible with a point-like source, as already stated by Matsumoto et al. (2014) and Sakai (2013). However, due to the comparatively large PSF of Suzaku, it is unclear how much of this apparent extent is due to the intrinsic size of CXOU J150850.6−621018 or due to confusion with the offset source XMMU J150835.7−621021 (see previous paragraph). Therefore, we suggest to refer to the extension measurements of CXOU J150850.6−621018 with Chandra (Acero et al. 2011) and XMM–Newton (Tibolla et al. 2014) for a more reliable estimate of this parameter.

Figure 2.

Radial profile of Suzaku J1508.8−6210 (1–10 keV) from the XIS0+3 counts map (markers with error bars, blue), and the Suzaku PSF at 4.5 keV (dashed, red).

Open in new tabDownload slide

Of particular interest here is the faint extended source CXOU J150706.0−621443, previously detected with Chandra and XMM–Newton, which can be also clearly identified in the new Suzaku observation: Suzaku J1507.0−6214. The position of Suzaku J1507.0−6214 is slightly shifted towards the north-west compared to the Chandra and XMM–Newton positions of CXOU J150706.0−621443 (see Acero et al. 2011; Tibolla et al. 2014), as indicated by the yellow cross in Fig. 1. However, the high-resolution Chandra results do not show any other source towards the shifted direction and we thus rule out source confusion due to Suzaku's larger PSF as the origin of the shift. More likely, the relative offset originates from uncertainties in the absolute pointing position of ∼20 arcsec (see e.g. Uchiyama et al. 2008). We see similar offsets between the Suzaku and Chandra positions also for the other detected sources with Chandra counterparts in this observation. We therefore identify Suzaku J1507.0−6214 with CXOU J150706.0−621443 and use the original Chandra name for this object for all further discussion in this paper.

We extracted spectra from all sources which offered a sufficient number of counts (i.e. S1, S2, S4, S7, and S8) from all three detectors (XIS0, XIS1, XIS3). For all sources but Suzaku J1508.8−6210, we chose an extraction radius of 60 arcsec which is recommended for point-like source analyses, and a nearby source-free region for the extraction of the background spectrum. Due to its intrinsic size and brightness we chose a larger radius of 120 arcsec for Suzaku J1508.8−6210 and a large concentric annular region for the background (also see Table 2). These new deep Suzaku data offer the opportunity to study the spectrum of the faint extended source CXOU J150706.0−621443 which was not possible with previous observations by Chandra and XMM–Newton due to the limited count statistics.

We performed the spectral fits with xspec v12 (Arnaud 1996) and used an absorbed power law as well as a plasma (MEKAL) model to investigate both non-thermal and thermal radiation mechanisms. To model the photoelectric absorption, we used the tbabs model along with the Galactic metal abundances from Wilms et al. (2000). For each source, we fitted the spectra from all detectors simultaneously with linked spectral parameters. The spectral results are compiled in Table 3.

For four of these sources, flux estimates from previous X-ray observations with XMM–Newton and Chandra are available, three of them based on spectral fitting. For Suzaku J1507.9−6222, a previous flux estimate is available only from XMM–Newton where the flux was calculated from the count rate assuming a certain spectral shape (see Tibolla et al. 2014, for details). Here, the new Suzaku flux is a factor of ∼2 lower compared to XMM–Newton, but still barely within the 1σ statistical uncertainties. This discrepancy is probably due to differences between the assumed spectral parameters in the XMM–Newton measurement and the measured spectral shape in this work. Suzaku J1507.1−6216 has an XMM–Newton and Chandra counterpart with available spectral fitting results for XMM–Newton (see Tibolla et al. 2014). The fluxes are compatible within 1σ uncertainties, however, the best-fitting spectral slopes deviate significantly. This is most likely due to the fact that in the XMM–Newton analysis the absorption CD was fixed at the total Galactic value in this direction, whereas here we find that the source spectrum shows very little absorption with an upper limit on N_H well below the total Galactic value (see Table 3). Correlations between N_H and the spectral slope likely explain the discrepancies in spectral index.

Below, we discuss the results for the two extended sources Suzaku J1508.8−6210 and CXOU J150706.0−621443 in more detail. Fig. 3 shows the spectra for these two sources along with the best-fitting power-law models. Judging from the χ² values, the power-law model is clearly preferred for Suzaku J1508.8−6210 (≥5σ), whereas there is no clear preference for CXOU J150706.0−621443, given the current data. For both sources, the value of N_H appears to be large (see discussion in Section 4.2) and, in the case of Suzaku J1507.0−6214, rather independent of the assumed spectral shape.

Figure 3.

Suzaku spectra of the two extended sources Suzaku J1508.8−6210 (left) and CXOU J150706.0−621443 (right) shown together with the best-fitting absorbed power-law model (stepped lines). The colours indicate the spectra from the individual cameras: XIS-0 (black), XIS-1 (red), XIS-3 (blue).

Open in new tabDownload slide

Matsumoto et al. (2014) and Sakai (2013) already reported about results for these two extended sources based on the same Suzaku data set. These authors confirmed the extended nature of Suzaku J1508.8−6210. However, the lower count statistics from CXOU J150706.0−621443, coupled with its more compact size, make it point-like for Suzaku. Our spectral results are compatible in terms of photon index and flux with the ones obtained by the authors above. However, we find systematically larger values for the hydrogen CD N_H. Unfortunately, the description of the spectral analysis by these authors is not complete, and key information like the used absorption model and the assumed metal abundances are missing. We were able to reproduce the results of Matsumoto et al. (2014) and Sakai (2013) by changing from Galactic to solar metal abundances. Because of the higher metallicity of the sun compared to the average Galactic level, a lower equivalent hydrogen CD is needed for the same absorption effect, thus the lower N_H value in their fit. However, because these sources are very likely of extra-solar origin, we remain with Galactic abundances and deem our results to be more realistic.

As already discussed in the first paragraph of this section, Suzaku J1508.8−6210 may be composed of the two independent X-ray sources CXOU J150850.6−621018 and XMMU J150835.7−621021. To compare with the previous Chandra result (Acero et al. 2011), we calculated a flux of Suzaku J1508.8−6210 in the 2–10 keV band of (9.8 ± 0.3) × 10⁻¹³ erg cm⁻² s⁻¹ which is about 30 per cent higher than the Chandra measurement of CXOU J150850.6−621018. This difference might arise from an additional contribution from the fainter but variable source XMMU J150835.7−621021. In this case the latter source must have had an ∼3 times higher flux during the Suzaku observation than during the earlier XMM–Newton measurement (see Tibolla et al. 2014). Variability of this magnitude is very well plausible for flaring stars (see e.g. Liefke, Fuhrmeister & Schmitt 2010) and X-ray binaries (for a review, see van den Berg 2010), the two most plausible scenarios for XMMU J150835.7−621021 as suggested by Tibolla et al. (2014). Another contributing factor to the larger flux seen with Suzaku from Suzaku J1508.8−6210 compared to Chandra and XMM–Newton could arise from the larger region used for spectral extraction. However, this would only be the case if CXOU J150850.6−621018 is much more extended than estimated from the XMM–Newton and Chandra maps, with a surface flux in the tails below their detection limit.

4 DISCUSSION

Out of the two extended X-ray sources in the vicinity of HESS J1507−622, originally detected by Chandra, in the following discussion we deem CXOU J150706.0−621443 to be the more likely counterpart to the TeV emission due to its spatial coincidence with the intrinsic size of HESS J1507−622, which is not the case for CXOU J150850.6−621018 (Suzaku J1508.8−6210). Due to the lack of overlap with the VHE gamma-ray size of HESS J1507−622, the latter source was considered a less likely counterpart in a PWN scenario (see Acero et al. 2011; Tibolla et al. 2014, and also section 2).

After the discovery of HESS J1507−622, a leptonic scenario was favoured (Acero et al. 2011) while hadronic scenarios were considered unlikely (Domainko 2011). Acero et al. (2013) reported a point-like Fermi Large Area Telescope (Fermi-LAT) counterpart with soft spectrum. However, the nature of the source still remains elusive. On the one hand, several PWN models were successfully applied to describe the emission seen from HESS J1507−622: Tibolla et al. (2011) proposed an ancient PWN model (developed by de Jager et al. 2009), Tibolla et al. (2012) fitted a modified leaky-box model of (Zhang, Chen & Fang 2008), and Vorster et al. (2013) managed to describe HESS J1507−622 with a time-dependent PWN model. On the other hand, no pulsar has been detected in the vicinity of HESS J1507−622 (Acero et al. 2011).

The nature of the TeV gamma-ray source HESS J1507−622 is challenging to explain with established models for Galactic sources, such as young SNRs and PWNe, due to its large offset from the Galactic plane coupled with its apparent compactness. In a PWN scenario, the latter feature may indicate either that the source is nearby and still young, or that the source is more evolved and far away. The problems with the first scenario are the absence of a young (≲10⁴ yr) and powerful pulsar and the relatively low X-ray flux of the candidate synchrotron nebula CXOU J150706.0−621443. To explain the low X-ray flux, in the second scenario the PWN would be a more evolved system (e.g. Mattana et al. 2009; Balbo et al. 2010). In this framework, Acero et al. (2011) found a distance of HESS J1507−622 of >6 kpc by comparing its angular extent with the size of the nearby, evolved Geminga PWN. A distance of 6 kpc has also been adopted in previous PWN models (Tibolla et al. 2011, 2012; Vorster et al. 2013). If the extension of HESS J1507−622 is driven by diffusion of energetic electrons for a time-scale of 2 × 10⁴ yr, the distance to the object needs to be very large (≳10 kpc) to be compatible with the apparent compactness of the TeV source (Domainko & Ohm 2012). This in turn implies a large physical distance of HESS J1507−622 from the Galactic plane, in particular much larger than the scaleheight of the distribution of VHE gamma-ray emitting pulsars. A remedy for this situation would be an association with hypervelocity stars or HESS J1507−622 being a member of a new population of gamma-ray sources associated with older stellar populations. For a more detailed discussion of these scenarios see Domainko & Ohm (2012) and Domainko (2014).

As becomes evident from the above discussion, the most important, but yet unknown parameter is the distance of HESS J1507−622 from Earth. Also, the details of the spectral shape and the flux of the non-thermal X-ray counterpart to HESS J1507−622 are very valuable to constrain intrinsic source properties, such as magnetic field and age.

In the subsections below, we explore the implications of the measured spectral parameters of CXOU J150706.0−621443 on its distance (using N_H), and on the properties of the underlying relativistic particle population (using the spectral index and flux).

4.1 Modelling the spectral energy distribution

We use our spectral measurement of CXOU J150706.0−621443 for an updated modelling of the spectral energy distribution (SED) of HESS J1507−622. We assume that CXOU J150706.0−621443 is physically related to HESS J1507−622 because of its extended nature and its spatial coincidence with the VHE gamma-ray source. Fig. 4 shows the broad-band spectral measurements used for this study. Here, the X-ray data points are the unfolded (corrected for instrument response) data from the Suzaku observation, corrected for the foreground absorption as determined by the best-fitting power-law model, taking also parameter uncertainties into account (hence the large error bars of the lowest energy data point where the influence of N_H is strongest). In addition to the new Suzaku spectral data points of CXOU J150706.0−621443, we show in Fig. 4 also the uncertainty range of the previous best-fitting power-law model measured with Chandra (see Tibolla et al. 2014). The latter is only an approximation, calculated from the quoted uncertainties of the integral flux and photon index, assuming no parameter correlations (as the full covariance matrix of the Chandra fit has not been published). The uncertainties of the Chandra measurement are very large due to limited statistics and both spectra are compatible within errors. The slightly lower flux seen with Chandra may arise from the smaller size of the spectral extraction region compared to the new Suzaku analysis.

Figure 4.

Broad-band SED of HESS J1507−622, including the radio flux upper limit from Molonglo (Bock, Large & Sadler 1999; Domainko & Ohm 2012), the uncertainty band of the best-fitting power-law spectrum measured with Chandra (shaded area; Tibolla et al. 2014), the Suzaku data points for CXOU J150706.0−621443 from this work, Fermi measurements in the high-energy gamma-ray regime (Domainko & Ohm 2012) and HESS flux points at energies larger than 100 GeV (Acero et al. 2011). Additionally, the model SED is shown, assuming Synchrotron (dots) and IC emission (dashes) from a broken power-law electron distribution. See Section 4.1 for a detailed description of the model.

Open in new tabDownload slide

A case for a leptonic origin of HESS J1507−622 has been made in the past by various authors (Acero et al. 2011; Tibolla et al. 2011, 2012; Vorster et al. 2013) in the framework of a PWN scenario (see also Weiler & Panagia 1978; Gaensler & Slane 2006). One-component broken power-law electron spectra (Torres et al. 2014) and two-component electron spectra (Vorster et al. 2013) were adopted to fit the SED of several PWNe. For HESS J1507−622, using the flux measurement from the 2-year Fermi-LAT source catalogue (2FGL) below 100 GeV (Nolan et al. 2012), Vorster et al. (2013) were able to fit the SED of HESS J1507−622 with a two-component electron spectrum. Here, we adopt the result of the longer 34-month Fermi-LAT data set analysed by Domainko & Ohm (2012) for energies below 100 GeV, following the approach of Torres et al. (2014), and fit a one-component electron spectrum. This is mainly done to reduce the number of free parameters compared to a two-component injection spectrum.

We assume a broken power-law energy distribution for the radiating electrons with a spectral break of ΔΓ = 1, as expected for a cooling break for continuous injection over longer periods of time caused by synchrotron and inverse-Compton radiation in the Thomson regime. With these assumptions we find a best-fitting model with a spectral index below the break energy (E < E_break) of Γ = 2, for E > E_break a spectral index of Γ = 3, and E_break = 0.9 TeV. The total energy in electrons is 5 × 10⁴⁷(d/1 kpc)² erg with d being the distance to the source. The maximum energy of electrons in this model is E_max = 1 PeV, a value which is necessary to reproduce the highest energy X-ray data points.

From this model fit, constraints on the age of the radiating electrons can be obtained. For the case were the break at 0.9 TeV in the broken power law is introduced by inverse-Compton cooling on the CMB, this would point towards an age of the source of ≈10⁶ yr, i.e. a rather old source. This can be compared to age estimates based on the evolution of PWNe if such an origin is adopted for HESS J1507−622. Mattana et al. (2009) found that the ratio of the flux in VHE gamma-rays (1–10 TeV) F_γ and the flux in X-rays (2–10 keV) F_X strongly increases with PWN age. Applying this method for HESS J1507−622 with log(F_γ/F_X) ≈ 1.85, an age of about 3 × 10⁴ yr and thus a rather evolved PWN would be found. This is consistent with the fact that the size of the TeV source is much larger than the size of the X-ray source for evolved PWN (Kargaltsev & Pavlov 2010) which also appears to be the case for HESS J1507−622. An age of about 3 × 10⁴ yr would be in line with the PWN models of Tibolla et al. (2012) and Vorster et al. (2013). This age estimate is significantly smaller than the age of the Geminga PWN. Therefore, in the PWN scenario, HESS J1507−622 could be significantly less extended than the Geminga PWN and could thus be located closer than the 6 kpc as discussed above (note that the multi-kpc distance estimate in Section 4.4 is independent from spectral modelling and only based on the X-ray absorption measurement of the potential X-ray counterpart). The caveat for this interpretation is the smaller F_γ/F_X ratio of Geminga with respect to HESS J1507−622. Vorster et al. (2013) proposed the passage of the reverse shock to resolve this discrepancy. In this paper, we did not consider the effect of a past reverse shock passage on HESS J1507−622. Our modelling parameters are given for the present particle spectrum in this source. It has also to be noted that no SNR has been detected around HESS J1507−622. Additional constraints on the age of radiating electrons can be placed by the maximum particle energy. Electrons with an energy of about 1 PeV cool very fast via inverse-Compton radiation, and have to have been injected less than 1 kyr ago into the system. This may indicate the presence of a second spectral component at the highest energies and more generally may indicate that a two-component electron spectrum seems to be favoured over a one-component electron spectrum. To summarize, for a leptonic scenario the SED of HESS J1507−622 suggests a rather evolved system albeit with recent injection of highly energetic particles.

In this model, the magnetic field B is 0.47 μG. This value for B is consistent with the estimates of by Acero et al. (2011) but is smaller than the values found by Domainko & Ohm (2012, 1 μG) and Vorster et al. (2013, 1.7 μG). The reason for this difference is the fact that here the measured hard X-ray spectrum for CXOU J150706.0−621443 is used for the SED modelling, in contrast to the softer spectra assumed in the previous studies.

One caveat of such leptonic one-zone models is the significant difference of sizes of the emission regions in the X-ray and gamma-ray regimes. This may introduce a bias (i.e. underestimation of the synchrotron flux) in models where it is assumed that the X-ray and TeV fluxes are produced by the same population of electrons. With Chandra the size of CXOU J150706.0−621443 was estimated to be 20–25 arcsec (Acero et al. 2011), and with the new Suzaku data the associated source (Suzaku J1507.0−6214) appears to be point-like and its flux to be contained within a radius of ∼60 arcsec (which is also the suggested spectral extraction radius for point-like sources with Suzaku, see also Fig. 1).

However, a larger scale emission component comparable to the intrinsic size of HESS J1507−622 may not be detectable if the surface flux in X-rays is sufficiently low. For a very conservative estimate of the highest possible X-ray flux from the TeV emission region, we use the ratio between the intrinsic size of HESS J1507−622 and the area used for spectral extraction from CXOU J150706.0−621443 in this work (⁠|$\rho _\mathrm{area} = \left(A_\mathrm{TeV} / A_\mathrm{X{\rm -}ray}\right)\approx 80$|⁠), as well as the ratio of the X-ray surface flux from source-free regions and the region of CXOU J150706.0−621443 (⁠|$\rho _\mathrm{flux} = \left(F^\mathrm{surf}_\mathrm{source{\rm -}free} / F^\mathrm{surf}_\mathrm{Suzaku\,J1507.0-6214}\right)\approx 0.33$|⁠). Assuming that all the diffuse X-ray surface flux from source-free regions is related to HESS J1507−622, a scaling factor of ρ_area × ρ_flux ≈ 27 needs to be applied to the measured flux from CXOU J150706.0−621443. The implications of the increased X-ray flux on the model parameters would be an enhanced magnetic field (by a factor of |$\sqrt{27} \approx 5$|⁠, i.e. B ≈ 2.9 μG) and a cutoff of the electron spectrum decreased by the same factor, i.e. E_max ≈ 200 TeV (following Hinton & Hofmann 2009).

4.2 Comparison to measured values of the neutral hydrogen column

The values of N_H derived from the X-ray spectra in Section 3 can be compared to measured values of the total absorbing CD of neutral hydrogen in the direction of the sources. Soft X-rays are absorbed by interstellar gas, which predominantly consists of atomic (H i) and molecular hydrogen (H₂).

The H i distribution in the galaxy has been measured by Dickey & Lockman (1990) and Kalberla et al. (2005). For the direction of HESS J1507−622, Dickey & Lockman (1990) estimated a H i CD of 5.0 × 10²¹ cm⁻² and Kalberla et al. (2005) constrained it to 4.2 × 10²¹ cm⁻².¹ In comparison to these values, Suzaku J1508.8−6210 and CXOU J150706.0−621443 seem to be absorbed on a comparable or even higher level. This indicates that these sources are located at the edge or even outside the main H i distribution of the Galaxy. The excess absorption for the two sources in comparison to the values found in the surveys could in principle be attributed to Galactic molecular hydrogen that is not measured in the H i surveys.

Absorption by neutral hydrogen in soft X-rays is tightly correlated to dust absorption in the optical (Predehl & Schmitt 1995). Consequently, from the measured level of dust absorption (A_V), the total CD in cold hydrogen can be estimated. For CXOU J150706.0−621443 (located inside HESS J1507−622) we find A_V = 4.6 and for Suzaku J1508.8−6210 A_V = 3.7 (Schlafly & Finkbeiner 2011).² These measured values of A_V correspond to a cold hydrogen absorption of 8.2 × 10²¹ cm⁻² for CXOU J150706.0−621443 and 6.6 × 10²¹ cm⁻² for Suzaku J1508.8−6210 (Predehl & Schmitt 1995). These comparisons evidence that both sources experience soft X-ray absorptions consistent with the total Galactic CD of neutral hydrogen. However, it has to be noted that the measurement of dust absorption may not be precise at Galactic latitudes <5° (Schlafly & Finkbeiner 2011).

An alternative method to determine the total X-ray absorbing hydrogen CD is described by Willingale et al. (2013) who used 493 gamma-ray burst afterglows to calibrate estimates of N_H based on measurements of atomic hydrogen and dust extinction. Using their web tool³ to estimate the total (⁠|$N_\mathrm{H\,\small {i}} + N_\mathrm{H2}$|⁠) for the region of HESS J1507−622, we get a value of N_H ≅5.7 × 10²¹ cm⁻² and for Suzaku J1508.8−6210 we obtain N_H ≅5.9 × 10²¹ cm⁻². Both values are slightly lower than estimated from the dust absorption A_V directly (see previous paragraph).

4.3 Comparisons to models of the Galactic neutral hydrogen distribution

The highly absorbed X-ray spectrum of CXOU J150706.0−621443 may suggest that any associated object is located at a considerably large distance to Earth. In order to estimate the source distance, we use a large-scale 3D model for both atomic and molecular Galactic hydrogen gas and compare it to the CD derived from the Suzaku measurements. This model is based on the descriptions of the large-scale distribution of H i and H₂ gas in the Milky Way by various authors, summarized by Ferrière (2001). There, a galactocentric distance of the sun of R_⊙ = 8.5 kpc is assumed, and gas distributions that apply different values of R_⊙ are rescaled to the updated value by the author. We also adopt this value of R_⊙ in the following.

The large-scale distribution of each of the mentioned gas components is modelled as follows: First, the radial (i.e. with R, the distance to the Galactic Centre along the Galactic plane) distribution of the azimuth-averaged gas CD perpendicular to the Galactic plane (vertical CD) is obtained. Secondly, the radial profile of the gas layer thickness is determined. Lastly, a vertical density distribution function is assumed, and its normalization and width is determined by the two other distributions. As a result, for each set of coordinates {R, z}, where z is the vertical distance to the Galactic plane, a space-averaged number density of hydrogen nuclei can be calculated.

Furthermore, we allow for a large-scale Galactic spiral arm structure by modulating the gas densities towards the arms in a simple manner, as described in Section 4.3.3.

4.3.1 Atomic hydrogen (H i)

4.3.2 Molecular hydrogen (H₂)

4.3.3 Spiral arms

As an option, we also allow for the modulation of the gas density towards Galactic spiral arms. In the interarm regions Clemens et al. (1988) found values of the space-averaged density of molecular hydrogen lowered by a factor of ∼3.6 compared to the density inside the arms. A similar contrast of ∼4 is found for the H i surface density, see Kulkarni (1982).

The resulting gas densities, averaged perpendicular to the Galactic plane, are shown in Fig. 5.

Figure 5.

|$\mathrm{H\,\small {i}}+\mathrm{H}_\mathrm{2}$| atomic nucleus density, averaged along the perpendicular direction to the Galactic plane. Additionally, the line of sight towards HESS J1507−622 (dashed) and the Valée spirals (coloured) are shown.

Open in new tabDownload slide

4.4 The distance to HESS J1507−622

The model described in Section 4.3 allows us to derive the theoretically expected neutral hydrogen CD in the direction of CXOU J150706.0−621443 as a function of the distance to Earth, d. The result is shown in the left-hand panel of Fig. 6. For distances larger than ∼2 kpc, both models agree within errors with the Suzaku measurement. The spiral arm modulation results in overall lower values for d than those obtained for a ‘flat’ galaxy in the x-y plane. This is especially pronounced at distances smaller than 1 kpc, as the main contribution to the CD stems from the gas concentrated in the Carina arm (red line in Fig. 6, see also Fig. 5).

Figure 6.

Left: calculated CD of H i and H₂ gas in the direction of HESS J1507−622 as a function of the distance to Earth. The solid line shows the model expectation in the absence of any Galactic spiral arm gas modulation while the thick dashed line takes such a modulation into account. The thin dashed lines represent the contributions of the interarm (black) and the intraarm gas (coloured, see Fig. 5). Right: probability profile as a function of the distance to Earth for CXOU J150706.0−621443 being located at a distance r greater than d.

Open in new tabDownload slide

We use the modelled CD profile along d to calculate the probability P[r > d] of CXOU J150706.0−621443, and presumably HESS J1507−622, being located at a distance r > d. To that end, we treat the asymmetric error interval of the X-ray N_H measurement as a distorted Gaussian distribution, following Method 2 described in Barlow (2004). The resulting profile is shown in Fig. 6. As can be seen, the X-ray measurements and Galactic gas model places CXOU J150706.0−621443 with probabilities of 70 per cent (flat galaxy) or 75 per cent (arm-modulated) at distances larger than 25 kpc, which corresponds to the edge of the Galaxy. Interestingly, HESS J1507−622 is located very close to the super-galactic plane (SG-lon: 183.6°, SG-lat: 0.2°), a location where nearby extragalactic objects cluster (Lahav et al. 2000), and also where all non-blazar extragalactic TeV sources are concentrated (see TeVCat).⁵ Although interesting, these values, however, do not yet allow us to make any definitive statements on the exact location of HESS J1507−622, due to the relatively large uncertainties of the N_H measurement from the X-ray spectrum.

Our constraints on the distance to HESS J1507−622 can be compared to the distance estimates by Matsumoto et al. (2014) for Suzaku J1508.8−6210 of d = 1.7 ± 0.2 kpc and for CXOU J150706.0−621443 of d = 2.6 ± 1.7 kpc. We note here that also their measurements of the total N_H are compatible with the total Galactic CD within 1σ uncertainties, and thus the calculation of an upper bound of the distances to these two sources should not be possible, contrary to what is indicated by the symmetric distance uncertainties quoted by Matsumoto et al. (2014). In such cases, the only constraints on the distance can be given in terms of probabilities for d to be larger than a certain value r, as, e.g. derived from our model (see Fig. 6). Furthermore, the estimates by Matsumoto et al. (2014) are based on the assumption of an average Galactic ISM density of 1 cm⁻³, which is probably not appropriate for the specific case of HESS J1507−622, due to its large offset from the Galactic plane and hence significantly lower ambient densities, particularly for large distances. This assumption of a too dense medium systematically biases the resulting distance estimates towards lower values.

We note here that the total values of N_H derived from the Galaxy model are comparable or slightly larger than those estimated from measurements (see Section 4.2). Therefore, the derived probabilities of HESS J1507−622 being located at a distance r > d in this section can be viewed as conservative, since lower values of N_H in the model would result in even larger distances.

For comparison, we apply the same procedure to Suzaku J1508.8−6210. The result can be seen in Fig. 7. Both the unmodulated and spiral arm model predictions for the absolute neutral hydrogen CD in the direction of the source agree within errors with the Suzaku measurements. As the predicted total hydrogen CD in case of the modulated gas model almost perfectly coincides with the X-ray value, the estimated probability for Suzaku J1508.8−6210 to be extragalactic or at the edge of the Milky Way is ∼50 per cent in this case. With the unmodulated gas model, the total hydrogen CD is predicted at a higher value but still matches the upper 1σ-limit of the X-ray measurement, favouring a Galactic nature of the source with a probability of ∼84 per cent. Again, our model predictions on the total hydrogen CD are somewhat larger than the independent measurements presented in Section 4.2, which would result in lower probabilities of Suzaku J1508.8−6210 being a Galactic source.

Figure 7.

Same as in Fig. 6, but calculated for Suzaku J1508.8−6210.

Open in new tabDownload slide

4.5 Implications for a PWN scenario

As discussed in Section (4.4) the comparatively large value of N_H measured in X-rays might indicate a considerable line-of-sight distance of CXOU J150706.0−621443, and potentially HESS J1507−622, from Earth. Given the rather high Galactic latitude (b = −3| $_{.}^{\circ}$|49), this would also imply a large vertical distance to the Galactic plane, z. Also, the Galactic longitude (b = 317| $_{.}^{\circ}$|97) of HESS J1507−622 defines a minimal value of R, R_min = 5.7 kpc. These distances, together with a model for the distribution of PWNe in the Galaxy, are used in the following to estimate the probability of a PWN scenario for HESS J1507−622.

4.5.1 Un-recycled pulsars

First, we investigate the probability of HESS J1507−622 being the PWN associated with an un-recycled pulsar. This is the population of pulsars considered in the study of Mattana et al. (2009). We assume that PWNe follow the distribution of pulsars in the Milky Way. Corresponding radial and vertical distribution functions have been proposed by Yusifov & Küçük (2004) and Faucher-Giguère & Kaspi (2006), respectively, and are adopted in our model. The mentioned radial distribution describes a power law increase followed by an exponential decay, ρ(R) ∼ R^1.64exp {−4.01 × (R − R_⊙)/R_⊙}, and reaches its maximum at R = 3.2 kpc. For the distribution perpendicular to the Galactic plane, we assume an exponential profile with scaleheight z₀. The value of z₀ is estimated by a fit to the z-distribution of pulsars detected by the Fermi-LAT instrument (Abdo et al. 2013, the Second Fermi Large Area Telescope Catalog of Gamma-Ray Pulsars), which provides a homogeneous sky coverage. To fit the pulsar distribution we used all Galactic Fermi-detected un-recycled pulsars without any cut on age or spin-down power. This results in a sample of 51 pulsars. An exponential fit to the z-distribution (χ/ndf = 9.6/7) yields a scaleheight of z₀ = (74.0 ± 16.1) pc.

With this distribution function one can determine the probability τ to find a pulsar at any given coordinate {R > R_min, z(τ)}. For instance, assuming a value of z₀ = 74 pc, at a perpendicular distance to the Galactic plane of z(0.001 per cent) = 815 pc, the chance to find a pulsar is τ = 0.001 per cent. We calculated the CD towards HESS J1507−622 that corresponds to z(0.001 per cent) and compare it to the X-ray measurement of N_H, again treating the probability distribution of the latter as a distorted Gaussian (see Section 4.4). This comparison places the sources with a probability of 73 per cent (flat galaxy) and 76 per cent (spiral arms) at values z > z(0.001 per cent). Thus, at a confidence of 99.999 per cent, model and measurement exclude a PWN scenario for HESS J1507−622 with probabilities between 73 and 76 per cent. Probabilities for different values of τ are given in Table 4 and also shown in Fig. 8 for z₀ = 74 pc. Again, given the relatively large uncertainties of the X-ray measurement of N_H, these results do not allow us to disfavour a PWN scenario with high confidence.

Figure 8.

Same as in the left-hand panel of Fig. 6, but here as a function of z. Additionally, the z-values corresponding to τ = 0.1, 0.01, 0.001 per cent and assuming a vertical scaleheight of the pulsar distribution of z₀ = 74 pc are shown.

Open in new tabDownload slide

4.5.2 Millisecond pulsars

In this section, we discuss the implication of the rather large value of N_H for the case where HESS J1507−622 is the nebula associated with a millisecond pulsar (MSP). MSPs are commonly believed to be very old pulsars with characteristic ages of 10⁹ yr that were recycled by accretion from a companion star (e.g. Alpar et al. 1982; Grégoire & Knödlseder 2013). Following the discussion of the potentially old age of HESS J1507−622 in Section 4.1, an MSP may be a potential counterpart. Owing to their age their distribution features a large Galactic scaleheight of about 1 kpc and a Galactic scalelength of about 4 kpc (as measured from the Fermi-LAT detected MSPs; Grégoire & Knödlseder 2013). This scaleheight is larger than the scaleheights of cold hydrogen and therefore no constraints on the probability of HESS J1507−622 being a PWN associated with an MSP can be made. However, so far no detection of a PWN related to an MSP in VHE gamma-rays has been reported in the literature. However, this can in principle be a selection bias since MSPs do not strongly cluster around the Galactic plane, the region with the deepest large-scale exposure by VHE gamma-tray telescopes such as HESS Future observations of powerful MSPs are required to test the presence of VHE-gamma-ray emitting PWNe around these objects.

5 Summary and Outlook

In this paper, we present spectral results for the potential X-ray counterpart to the enigmatic TeV gamma-ray source HESS J1507−622. On the basis of this new measurement we provide interesting constraints on the parameters of the underlying population of relativistic particles, and also on the line-of-sight distance to the object. Even though this study represents a significant step forward in the identification of the origin of non-thermal emission from this puzzling object, the relatively low count statistics of the X-ray detection, and the resulting large uncertainties of the derived model parameters do not allow very strong conclusions. However, we see indications that the potential X-ray counterpart to HESS J1507−622 might be located at a considerable distance from Earth, which would strongly challenge established models for Galactic gamma-ray source populations, in particular PWNe.

Future deeper X-ray observation with high-throughput instruments providing higher statistical quality could greatly improve the strength and conclusiveness of this result.

This research has made use of data obtained from the Suzaku satellite, a collaborative mission between the space agencies of Japan (JAXA) and the USA (NASA). This research has made use of data and/or software provided by the High Energy Astrophysics Science Archive Research Center (HEASARC), which is a service of the Astrophysics Science Division at NASA/GSFC and the High Energy Astrophysics Division of the Smithsonian Astrophysical Observatory.

REFERENCES