DISCOVERY OF THE HARD SPECTRUM VHE γ-RAY SOURCE HESS J1641−463

The large field of view (FoV) of the High Energy Stereoscopic System (H.E.S.S.), together with its stereoscopic observation strategy, allowed the discovery of tens of very high energy (VHE, ⩾ 0.1 TeV) γ-ray sources⁴⁴ by scanning a large fraction of the Galactic plane (Aharonian et al. 2005a; Carrigan et al. 2013). With deeper exposures, more VHE γ-ray sources are detected, although source confusion begins to be problematic. Complementing the spatial search for new sources, an investigation into energy bands can provide an additional powerful tool for new discoveries. In this work, it will be shown how this method allowed for the detection of a new object, HESS J1641−463 (hereafter J1641−463), previously hidden in the tails of the much brighter object HESS J1640−465. Interestingly, the newly discovered source exhibits one of the hardest spectra observed in VHE γ-rays, allowing its detection at higher energies, where the two sources are clearly separated. Hereafter, the observations and the analysis technique that led to the discovery of J1641−463 are described. Finally, a discussion of plausible counterparts of this source at other wavelengths is presented.

H.E.S.S. is an array of five imaging atmospheric Cherenkov telescopes located in the Khomas Highland of Namibia, 1800 m above sea level. In the initial phase of the H.E.S.S. project, during which the data described here were taken, the array was composed of four 13 m diameter telescopes. Extensive air showers are measured with an average energy resolution of 15% and an angular resolution better than 01 (Aharonian et al. 2006) for a typical energy of 1 TeV. The trigger energy threshold is about 100 GeV and increases with higher zenith angle (Funk et al. 2004).

J1641−463 remained unnoticed by the standard source detection techniques due to its low brightness and its proximity to the bright source HESS J1640−465 (Abramowski et al. 2014). During a study of a possible energy-dependent morphology of HESS J1640−465, a collection of images for events with energies above a set of energy thresholds (E > 1, 2, 3, 4, and 5 TeV) was created. J1641−463 was not visible in the original images of the HESS J1640−465 FoV, as those images included no energy cut in the events, and thus were dominated by the much more numerous low-energy events coming from the brighter HESS J1640−465. Thanks to the improved H.E.S.S. point-spread function (PSF) at higher energies and to its hard spectrum, J1641−463 was clearly visible in the highest energy sky maps, where the contamination from HESS J1640−465 was low. This discovery triggered further H.E.S.S. observations, allowing the firm establishment of a new VHE γ-ray source. The VHE γ-ray excess image obtained for E > 4 TeV is shown in Figure 1, where the background level is estimated following the ring background model (Berge et al. 2007).

Zoom In Zoom Out Reset image size

The observations of the FoV around J1641−463 were carried out from 2004 to 2011, corresponding to an acceptance-corrected live time of 72 hr, after quality selection criteria were applied as in Aharonian et al. (2006). The data were analyzed with the methods described in Aharonian et al. (2006).⁴⁵ The events were reconstructed using the Hillas parameter technique (Hillas 1995). The results were cross-checked using two independent analysis methods (Ohm et al. 2009; de Naurois & Rolland 2009).

The position of J1641−463 (together with the nearby HESS J1640−465) was determined by fitting a two-dimensional double-Gaussian model convolved with the H.E.S.S. PSF to the two-dimensional ON-source excess event distribution for E > 4 TeV, energies at which source confusion with HESS J1640−465 is mitigated. The centroid of the Gaussian corresponding to the location of J1641−463 was found to be α_J2000 = 16^h41^m21 ± 30_stat ± 19_sys, $\rm \delta _{J2000} = -46^{\circ } 18^{\prime } 13^{\prime \prime } \pm 35^{\prime \prime }_{\rm stat} \pm 20^{\prime \prime }_{\rm sys}$. The source is found to be point-like, but a slightly extended morphology up to a width of σ = 3 arcmin cannot be ruled out due to uncertainties in the H.E.S.S. PSF.

Figure 2 shows the projection of the excess events in the rectangular region shown in Figure 1 for different energy bands. An F-test (Martin 1971) was performed comparing the single Gaussian model fits with the double-Gaussian fits. For all the energy bands, the null hypothesis can be rejected at significance levels of 3.6–4.3σ, thus clearly favoring the double-Gaussian model.

Zoom In Zoom Out Reset image size

In order to minimize the contamination from HESS J1640−465, hard cuts were used, which imply a cut on θ² (the square of the angular difference between the reconstructed shower direction and the source position) of 0.01 deg², and on the individual image charge in photo-electrons of 200. The source is detected with a statistical significance of 8.5σ above 4 TeV, determined by using Equation (17) in Li & Ma (1983) after background suppression with the reflected background model (Berge et al. 2007).

The differential VHE γ-ray spectrum of J1641−463, derived using the forward-folding technique (Piron et al. 2001), is compatible with a power-law function dN/dE = ϕ₀ × (E/1 TeV)^−Γ with ϕ₀ = (3.91 ± 0.69_stat ± 0.78_sys) × 10⁻¹³ cm⁻² s⁻¹ TeV⁻¹ and Γ = 2.07 ± 0.11_stat ± 0.20_sys for the energy range from 0.64 to 100 TeV. The flux level is ϕ(E>1 TeV) = (3.64 ± 0.44_stat ± 0.73_sys) × 10⁻¹³ cm⁻² s⁻¹, corresponding to 1.8% of the Crab Nebula flux above the same energy. At those energies, the estimated total contamination from HESS J1640−465 is 15 ± 6%, reduced at higher energies (4 ± 3% at E > 4 TeV). A fit by a power law with an exponential cutoff is not statistically justified given the low flux level of J1641−463. A fit to a constant value of the period-by-period⁴⁶ light curve for energies above 0.64 TeV yields a χ²/d.o.f. = 11.7/14, with a p-value of 63%. No variability can be seen in other time binnings (from year-by-year to 28 minute exposures).

3.1. Radio Observations

3.2. X-Ray Observations

No candidate for an X-ray counterpart of J1641−463 was found in existing catalogs, even when extending the search radius to 01 away from the source. Two data sets from Chandra and one from XMM-Newton were thus inspected in order to search for an X-ray counterpart of J1641−463.

The Chandra ObsID 11008 partially covers J1641−463, with 40 ks of exposure, while ObsID 12508 fully encloses it with 19 ks. The data sets were processed with the CIAO package. The tool wavdetect was used to identify sources, providing 32 faint point-like or marginally extended candidates at distances smaller than 01 to the J1641−463 position. This sample was filtered by 2 criteria, reducing the sample to 12 candidates (see Figure 3): first, the sources with signal-to-noise ratios below 3 were rejected. Second, a cut on the hardness ratio as defined in Elvis et al. (2009) was applied, HR = (H−S)/(H+S), where H are the counts with 2−10 keV and S the counts with 0.3−2 keV. The sources with HR ⩽ 0 were excluded. Spectral fits using an absorbed power-law model were performed assuming a value of N_H of 2.0 × 10²² cm⁻², corresponding to the values reported by Kalberla et al. (2005) and Dickey & Lockman (1990), in good agreement with those derived with the NANTEN data. The estimated flux densities in the 0.3−10 keV energy band result in values from 7 × 10⁻¹⁵ erg cm⁻²s⁻¹ (src. B) to 1.5 × 10⁻¹³erg cm⁻² s⁻¹ (src. L). No evidence of variability was found for any of the sources after performing a one-sample Kolmogorov–Smirnov test: the probability P_KS for the hypothesis of a uniform flux was P_KS > 0.1. None these sources is an obvious counterpart of J1641−463 due to their low fluxes and the lack of any morphological feature that could point to such an association.

Zoom In Zoom Out Reset image size

The XMM-Newton ObsID 0302560201, covering the region of HESS J1640−465 (Funk et al. 2007), constitutes a partial 23.7 ks exposure of the source area. The data set was analyzed using the XMM SAS analysis task edetect_chain simultaneously in all three cameras and the five standard energy bands. In this manner, 27 sources were found, with only one consistent with the position and upper limit to the extension of J1641−463 (see Figure 3). This source was detected only in the pn camera and only in the energy band 0.5−1 keV with a significance of ≈4.6σ, and it is not detected in the Chandra data. The vignetting for this source is 0.35 in the pn camera, so the observation is very insensitive to the region of interest. Due to low statistics, calculating an HR or spectrum for this source was not possible, and it is unclear whether this may represent a counterpart.

3.3. HE Observations

Possible scenarios to explain the emission from J1641−463 include the emission from accelerated particles within an SNR, a molecular cloud illuminated by cosmic rays (CRs), a pulsar wind nebula (PWN), and a γ-ray binary. These scenarios are discussed below.

If G338.5+0.1 is a young SNR, it can accelerate particles up to hundreds of TeV. The left panel of Figure 4 shows the comparison between the H.E.S.S. spectrum and the spectrum produced by accelerated protons from G338.5+0.1 interacting with the ambient gas. The predicted spectra are calculated using the parameterization of Kelner et al. (2006), assuming a proton spectrum with a power-law slope of −2.1 and multiple cutoff energies. The profile of the log-likelihood ratio test statistic (Rolke et al. 2005) was used to estimate a confidence interval of the cutoff energies, while considering the spectral index and normalization as nuisance parameters and ignoring systematic errors. The 99% confidence level (CL) lower limit on the cutoff energy corresponds to 100 TeV. This proton spectrum is one of the hardest ever inferred to explain the emission from a γ-ray source and agrees well with the prediction by diffusive shock acceleration in young SNRs. Remarkably, the γ-ray spectrum of J1641−463 is harder than that observed from the young SNR RXJ1713−4936 at energies above few TeV, where a cutoff is seen (Aharonian et al. 2007). If the TeV luminosity measured by H.E.S.S. is produced by collisions of protons with the ambient gas, then the total energy of the supernova explosion converted into hadron acceleration is $W_p = L_{\gamma } \, t_{\pi ^{0}} ,\approx\, {10}^{50} n^{-1}$, where L_γ = 4 × 10³⁴ erg s⁻¹ is the total luminosity measured by H.E.S.S. above 0.64 TeV (at 11 kpc) and $t_{\pi ^{0}} \approx 5 \times 10^{15} (n/1\,\mathrm{cm}^{-3})^{-1}\, \mathrm{s}$ is the cooling time of protons through the channel of π⁰ production (Aharonian 2004). With a proton spectrum extending almost up to 1 PeV, J1641−463 may represent a source population contributing significantly to the galactic CR flux around the knee.

Zoom In Zoom Out Reset image size

If G338.5+0.1 is older (5−17 kyr; see Section 3.1), then VHE protons accelerated by the young SNR G338.3−0.0, positionally coincident with HESS J1640−465  (Abramowski et al. 2014), could have already reached the dense molecular cloud (MC) coincident with J1641−463. This would explain the relatively high brightness of J1641−463 in comparison with HESS J1640−465 at high energies, as shown in Figure 2 (Aharonian & Atoyan 1996; Gabici et al. 2009). In such a scenario, HESS J1640−465 would no longer look like a pevatron, as the highest energy CRs would have already left (Aharonian & Atoyan 1996). The much younger adjacent G338.3−0.0 would be in this scenario a major source of CRs.

Electrons of hundreds of TeV IC (inverse Compton) scattering off the cosmic microwave background photons (CMB) could explain the emission from J1641−463. These electrons would be accelerated either in G338.5+0.1 or in the PWN associated with the young energetic pulsar, PSR J1640−4631, discovered within the observational boundaries of HESS J1640−465 (Gotthelf et al. 2014). Even assuming a pure power law for the primary electron spectrum, the cross section for IC scattering decreases at high energies resulting in a break in the γ-ray spectrum at multi TeV energies. Such a break is not observed in the spectrum of J1641−463. The predicted IC radiation, shown in the right panel of Figure 4, was obtained by assuming that the electron cooled spectrum is a power law of spectral index −3.14 with different cutoff energies. The 99% CL lower limit on the cutoff energy, derived as in the case of the proton model using the exact Klein–Nishina expression for the IC emission, corresponds to 700 TeV. It is extremely difficult to accelerate electrons in SNRs to such energies, as hundred TeV electrons suffer severe synchrotron losses in the amplified magnetic fields of acceleration sites. Both the absence of a break in the γ-ray spectrum of J1641−463 and the derived lower limit on the cutoff energy of the electron spectrum strongly disfavor the leptonic scenario.

A γ-ray binary scenario could also be considered, given the point-like morphology of J1641−463 and that a similarly hard spectral index of −2.23 has been found in one of these systems (LS 5039; Aharonian et al. 2005b). An X-ray flux as low as ∼10⁻¹⁴ erg cm⁻² s⁻¹ is expected from a faint X-ray binary system similar to HESS J0632+057 (Hinton et al. 2009) assuming a distance of 11 kpc, where the lack of an obvious optical counterpart could be due to high optical extinction caused by the large distance and the position close to the Galactic plane.

Deeper exposures with H.E.S.S. together with a study of the emission in various energy bands made it possible to discover a new unique VHE source, showing one of the hardest γ-ray spectra ever found at these energies, extending up to at least 20 TeV without a break. In order to explain the observed VHE γ-ray spectrum, scenarios where protons are accelerated up to hundreds of TeV at either G338.5+0.1 or G338.3−0.0 and then interact with local gas or nearby massive MCs are the most compelling ones. Other possible scenarios, such as a PWN or a γ-ray binary, are disfavored but cannot be discarded. Deeper X-ray and VHE γ-ray observations, together with a better PSF for the latter, would allow for a better identification of the source.

The support of the Namibian authorities and of the University of Namibia in facilitating the construction and operation of H.E.S.S. is gratefully acknowledged, as is the support by the German Ministry for Education and Research (BMBF), the Max Planck Society, the French Ministry for Research, the CNRS-IN2P3, and the Astroparticle Interdisciplinary Programme of the CNRS, the U.K. Science and Technology Facilities Council (STFC), the IPNP of the Charles University, the Polish Ministry of Science and Higher Education, the South African Department of Science and Technology and National Research Foundation, and by the University of Namibia. We appreciate the excellent work of the technical support staff in Berlin, Durham, Hamburg, Heidelberg, Palaiseau, Paris, Saclay, and in Namibia in the construction and operation of the equipment. This research has made use of Chandra Archival data, as well as the Chandra Source Catalog, provided by the Chandra X-ray Center (CXC) as part of the Chandra Data Archive. This research has made use of software provided by the Chandra X-ray Center (CXC) in the application packages CIAO, ChIPS, and Sherpa. This research uses observations obtained with XMM-Newton, an ESA science mission with instruments and contributions directly funded by ESA Member States and NASA. This research has made use of the SIMBAD database, operated at CDS, Strasbourg, France. The work of J.M. was funded by contract ERC-StG-259391 from the European Community.

Facility: H.E.S.S. -