Free Access
Issue
A&A
Volume 574, February 2015
Article Number A27
Number of page(s) 10
Section Interstellar and circumstellar matter
DOI https://doi.org/10.1051/0004-6361/201322694
Published online 20 January 2015

© ESO, 2015

1. Introduction

During the 20052006 Galactic Plane Survey in the very-high-energy (VHE; E> 100 GeV) range, the High Energy Stereoscopic System (H.E.S.S.), an array of imaging atmospheric Cherenkov telescopes, revealed more than a dozen new sources (Aharonian et al. 2005a, 2006b). Extensions of the survey and deeper observations of the Galactic plane led to the detection of more than 80 sources (Carrigan et al. 2013). While many Galactic TeV γ-ray sources can be identified with counterparts at other wavelengths, such as the binary LS 5039 (Aharonian et al. 2005b, 2006c), the supernova remnant (SNR) RX J17133946 (Aharonian et al. 2004), or HESS J1356645 (Abramowski et al. 2011), associated with a pulsar wind nebula (PWN), more than 20 γ-ray sources remain unidentified to date. Pulsar wind nebulae are the dominant class of Galactic TeV sources with at least 27 identifications up to now out of a total of more than 35 sources considered to be potentially due to PWN emission. The presence of an energetic pulsar close to the position of the source is an important clue in the identification process.

HESS J1834−087 is one such unidentified source (Aharonian et al. 2006b) that was later detected by the MAGIC telescope (Albert et al. 2006). It features bright and extended (intrinsic Gaussian width σ = 0.2°) VHE emission spatially coincident with the SNR G23.30.3 (W41). Additionally, high-energy (HE; E> 100 MeV) emission spatially coincident with W41 was detected with the Fermi Large Area Telescope (LAT) and listed in the first- and second-year catalogs as 1FGL J1834.30842c (Abdo et al. 2010a) and 2FGL J1834.30848 (Nolan et al. 2012).

The SNR W41 shows an incomplete shell of ~33′ diameter in the radio domain (Tian et al. 2007). Several HII regions are spatially coincident with the shell but unrelated to the remnant as explained by Leahy et al. (2008), who also estimated the kinematic distance of W41 to be between 3.9 and 4.5 kpc based on HI and CO observations. The associated cloud has a radial velocity in the local standard of rest (lsr) of VCllsr = 77 km s-1. Frail et al. (2013) detected OH (1720 MHz) maser line emission, which demonstrates the physical association between the SNR and the molecular cloud (MC). The maser is located near the center of the remnant, and its radial velocity (VOHlsr ≃ 74 km s-1) coincides with the MC velocity. Tian et al. (2007) estimated the ambient density n ~ 6 cm-3, and, by applying a Sedov model (Cox 1972), derived an age of 6 × 104 yr. However, the observed radius of the shock front is larger than the radius of the shell merged with the surrounding medium (called complete cooling in Cox 1972). The age determined by Tian et al. (2007) assuming a rapid cooling of the blast wave and a completed shell formation is ~2 × 105 yr.

With the XMM-Newton observatory, Mukherjee et al. (2009) detected a central compact object (CCO; XMMU J183435.3084443) at the center of the remnant (l = 23.236°,b = −0.270°) in X-rays, as well as a faint tail-like emission. The spectral parameters of this diffuse emission, which could be a PWN, were extracted in a region of 0.1° radius. The integrated energy flux and the spectral index are F2 − 10 keV = (4.2 ± 2.2) × 10-13 erg cm-2 s-1 and Γ = 1.9 ± 1.0. Perhaps because of a better resolution than available with the European Photon Imaging Camera (EPIC) onboard XMM-Newton and a short observation time (~40 ks), Misanovic et al. (2011) did not report the offset emission in Chandra data but a compact nonthermal X-ray nebula surrounding the CCO (CXOU J183434.9084443), which could be a dust halo or a PWN. Despite the nondetection of X-ray pulsations from the CCO, the existence of the nonthermal X-ray nebula suggests that the compact object could be a pulsar.

No radio emission from the putative pulsar or its nebula has been detected so far. Radio upper limits were derived from the Very Large Array (VLA) observations at 20 cm (Helfand et al. 2006) and the 1.1 mm Bolocam (Aguirre et al. 2011) Galactic Plane Survey images within a 0.1° radius around the position of the point-like X-ray source.

Two scenarios have been proposed to explain the TeV emission: the interaction between the SNR W41 and a nearby MC (Li & Chen 2012; Castro et al. 2013) and a PWN powered by the compact object CXOU J183434.9084443 (Misanovic et al. 2011).

The analyses presented here were performed using data taken with the H.E.S.S. telescopes and Fermi LAT. H.E.S.S. is an array of four identical 13 m diameter imaging telescopes located in the Khomas Highland of Namibia at an altitude of 1800 m. A larger telescope with 28 m diameter (H.E.S.S. Phase II) has been built in the center of the array and saw its first light in July 2012. No data from H.E.S.S. Phase II were used in this work. H.E.S.S. is sensitive to γ-rays at energies above 100 GeV up to several tens of TeV and has a large field-of-view (FoV) of 5° diameter. Requiring at least two telescopes triggered by an air shower, the system yields a very good angular resolution (r68% ~ 0.1°), energy resolution (ΔE/E ~ 15%), and an efficient background rejection (Aharonian et al. 2006a). These characteristics, together with its geographic location, make H.E.S.S. ideally suited for discoveries and studies of extended sources in the Galactic plane.

The LAT onboard the Fermi Gamma-ray Space Telescope, launched in June 2008, is a pair-conversion telescope sensitive to photons in a broad energy band from ~20 MeV to more than 300 GeV. It has a wide FoV of 2.4 sr and a good angular resolution (r68% ≲ 1° above 1 GeV). For a detailed description of the instrument, see Atwood et al. (2009).

Deeper observations of the region of SNR W41 with H.E.S.S. and Fermi LAT since the discovery of the γ-ray source allow morphological and spectral analyses over a very wide energy range of a few GeV to several tens of TeV, providing new insights into the origin of the γ-ray emission coincident with SNR W41.

The H.E.S.S. observations, data analysis, and characteristics of HESS J1834087 are provided in Sect. 2, while the results from the analysis of the Fermi LAT data are presented in Sect. 3. A discussion on possible γ-ray emission mechanisms is given in Sect. 4.

2. H.E.S.S. observations and data analysis

H.E.S.S. observations of 28 min duration were taken in wobble mode centered on the position of the SNR W41. The pointing alternates between offsets of ±0.5° in declination and right ascension with respect to the nominal target position. After standard quality selection (Aharonian et al. 2006a) to remove data affected by unstable weather conditions or hardware-related problems, the total live-time of the data set, including data presented in the discovery paper (Aharonian et al. 2006b), is 61 h. The zenith angle of the observations ranges from 10° to 45°, with a median value of 20°.

Table 1

Centroid (Galactic coordinates) and extension fits to the H.E.S.S. events spatial distribution in the W41 region.

The data were analyzed with the Model Analysis method, implemented in the ParisAnalysis software (version 0-8-22), which compares the raw atmospheric shower images in the Cherenkov camera with the prediction of a semi-analytical model (de Naurois & Rolland 2009). Two different event selection cuts were applied. Standard cuts, which require a minimum intensity in the camera image of 60 photo-electrons (p.e.), were used to compute energy spectra and sky maps. To improve the angular resolution and obtain better background rejection, hard cuts (higher minimum intensity of 120 p.e.) were used for morphological studies. The energy thresholds for standard and hard cuts are 177 GeV and 217 GeV, respectively. A second analysis (HAP, Aharonian et al. 2006a), which included independent calibration of pixel amplitude and event reconstructions and discrimination based on Hillas moments of the cleaned shower image, was used for cross-checks and provides compatible results.

thumbnail Fig. 1

H.E.S.S. image of the VHE γ-ray excess in the direction of SNR W41 after standard cut analysis. The image was smoothed with a Gaussian kernel of 0.068° corresponding to the instrument’s point-spread function for this data set, which is shown in the inset. Green contours represent the GRS 13CO data around W41 integrated in the velocity range 74 km s-1 to 82 km s-1 and smoothed with the same kernel. White contours of the VLA radio data at 20 cm show the shell-type SNR. The position of CXOU J183434.9084443 and the brightest OH (1720 MHz) maser are indicated with a triangle and a cross, respectively. The linear color scale is in units of counts per smoothing Gaussian width.

Open with DEXTER

2.1. Morphological analysis

Figure 1 presents the H.E.S.S. excess map obtained after the analysis with standard cuts and smoothed with a Gaussian kernel of 0.068°, which corresponds to the mean point-spread function (PSF) for this data set. Cloud density traced by 13CO from the Galactic Ring Survey (GRS; Jackson et al. 2006) smoothed with the same Gaussian kernel, radio 20 cm contours (White et al. 2005) from W41, and the position of the CCO CXOU J183434.9084443 are shown as well.

The detection of the TeV source is clear, with a significance of 27σ within an integration radius of 0.3° centered on the candidate pulsar position. The following morphological results were obtained with the tools included in the ParisAnalysis software. The best-fit position assuming a point-like model (model (A) in Table 1) can be considered compatible with the pulsar position (0.01° away from it). Using a symmetrical Gaussian model (model (B) in Table 1) significantly improves the likelihood of the fit with respect to the previous model. The best-fit position moves 0.04° away from the putative pulsar but is still compatible with it. The sum of a point-like and a symmetrical Gaussian convolved with the PSF (model (C) in Table 1) again improves (5σ) the fit with respect to the single-Gaussian model. No significant extension of the central component was found. As can be seen in Fig. 2, the radial profiles of the unsmoothed best-fit model (model (C)) and the excess map centered on the position of the putative pulsar agree well. For comparison, the radial profile of the single-Gaussian model (model (B)) is displayed. Table 1 summarizes the tested models and Fig. 4 represents the best-fit morphology.

thumbnail Fig. 2

H.E.S.S. radial profile of the uncorrelated excess map centered on the position of the putative pulsar. The solid dark blue line shows the result of the best-fit model (labeled (C) in Table 1) estimated in Sect. 2.1. The dark blue and light blue dashed lines represent the point-like and Gaussian components of this model. The profile from the single Gaussian model (labeled (B) in Table 1) is plotted in orange.

Open with DEXTER

Table 2

H.E.S.S. best-fit spectral parameters assuming a power-law model (Eq. (1)) between 0.2 and 30 TeV.

The wide energy range covered by H.E.S.S. allows a morphological study in two energy bands: we used data below and above 1 TeV to obtain equivalent statistics in each band. Slices perpendicular to the Galactic plane do not show any significant variation of the morphology, as seen in Fig. 3. The same conclusion was obtained along other angles. The corresponding composite image (Fig. 3, inset) highlights that the brightest emission remains centered on the putative pulsar position.

thumbnail Fig. 3

H.E.S.S. slices in the uncorrelated excess map along the direction perpendicular to the Galactic plane (white arrow in the inset) in two independent energy bands (E< 1 TeV: blue and E> 1 TeV: orange). The slices are centered on the candidate pulsar position (marked by a triangle in the inset). The composite color image (inset) shows the box (height = 1.0°,width = 0.22°) used for the profile.

Open with DEXTER

thumbnail Fig. 4

20 cm radio continuum map (Helfand et al. 2006) with superimposed best-fit positions as vertical crosses (their sizes indicate the statistical errors) and intrinsic extensions as rings. Magenta and dashed blue represent the Gaussian model of the Fermi LAT source and the extended component of the two-component model of the H.E.S.S. source. The inner and outer radius represents ± 1σ errors on the fitted extension. The candidate pulsar (triangle, brown) and OH maser (cross, yellow) positions are also reported.

Open with DEXTER

2.2. Spectral analysis

Spectral information was obtained between 0.2 and 30 TeV in two distinct regions corresponding to the best morphological model. The central and annular emissions were analyzed within a circular region of 0.1° radius and between rmin = 0.1° and rmax = 0.3° respectively. The reflected region background technique (Berge et al. 2007) was used to evaluate the background in the FoV. The spectrum of the central region is well fitted by a pure power law (equivalent χ2/ndf = 22/18) defined as: (1)where the decorrelation energy Edec is the energy at which the correlation between the flux normalization (Φ) and spectral index Γ is zero and where the statistical errors are the smallest. The best-fit parameters, listed in Table 2, yield an integrated flux I1−30 TeV = (0.40 ± 0.05stat ± 0.08syst) × 10-12 cm-2 s-1, which corresponds to ~2% of the Crab nebula flux in the same energy range. The use of an exponential cutoff power-law model improves the fit at a 2.4σ level, slightly lower than the improvement level reported in Méhault et al. (2011), but still in agreement with it. In the following, the pure power-law shape is considered.

The spectrum extracted from the annular region follows a power-law shape (χ2/ndf = 37/32) with the best-fit parameters reported in Table 2. The integrated flux I1−30 TeV = (1.13 ± 0.11stat ± 0.23syst) × 10-12 cm-2 s-1 corresponds to ~5% of the Crab nebula. No significant difference between the spectral indices in the central and annular regions was detected. The sum of the integral fluxes calculated in each region agrees with the flux found in the total emission, which is well fitted by a power-law model (parameters listed in Table 2). Figure 5 shows the H.E.S.S. spectral points for the annular and central regions and 1σ statistical errors. The bowties represent the 1σ error contours of the spectra. The measured fluxes and spectral indices for these regions given in Table 2 and Fig. 5 are reliable, because these regions are larger than the H.E.S.S. PSF, and the H.E.S.S. PSF shows little variation with energy.

As explained in Sect. 2.1, the VHE γ-ray emission was modeled by the sum of two components: one point-like and one extended. To estimate the intrinsic flux of each component and to model the spectral energy distribution (SED), the spectral points extracted in the central and annular regions (see Fig. 5) were corrected for the contributions of each component in each region. To normalize only their flux, a constant spectral index at any distance from the putative pulsar was assumed, in agreement with the spectral fits (see Table 2). From the best morphological model, the contributions of the point-like and extended components to the total flux were calculated to be 64% ± 4% and 36% ± 4% of the central region flux (r < 0.1°), respectively. The point-like component contribution to the total flux emission in the annular region is 17% ± 2%.

thumbnail Fig. 5

Spectral points and bowties of the LAT source (squares - statistical and systematic uncertainties added in quadrature in green and only statistical errors in black) and the H.E.S.S. annular (red diamonds) and central (blue points) regions. Bowties represent the 1σ confidence level error on spectra. The upper limits are calculated for a 3σ confidence level.

Open with DEXTER

3. Fermi LAT observations and analysis

The following analysis was performed using 47 months of data acquired from 2008 August 4 to 2012 June 30 within a 15°× 15° square around the position of the SNR W41. Only γ-ray events with reconstructed zenith angles smaller than 100° were selected to reduce contamination from γ rays from cosmic-ray interactions in the upper atmosphere. In addition, time intervals were excluded when the rocking angle was wider than 52° and when the Fermi satellite was in the South Atlantic Anomaly. Events were selected from 1.8 GeV to 100 GeV in the Pass 7 Source class (Ackermann et al. 2012). The P7_SOURCE_V6 instrument response functions (IRFs) were used. This selection is motivated by the morphological stability of the source, the improved angular resolution at high energy, and the difficulties in modeling the Galactic diffuse emission, that implies large systematic errors at low energy. These effects can be seen in the low-energy part (E< 1 GeV) of the spectrum in Castro et al. (2013). Two different tools using the maximum-likelihood technique (Mattox et al. 1996) were applied to perform the spatial and spectral analysis: pointlike (Lande et al. 2012) and Fermi LAT Science Tools1. These tools fit a source model to the data along with models for the residual charged particles, diffuse γ-ray emission, and the LAT sources in the region of interest listed in the second LAT catalog (Nolan et al. 2012). To describe the Galactic diffuse emission, the ring-hybrid model gal_2yearp7v6_v0.fits was used. The instrumental background and the extragalactic radiation are described by a single isotropic component with the spectral shape in the tabulated model iso_p7v6source.txt. These models are available from the Fermi Science Support Center2.

All 2FGL sources within 20° radius around W41 were added to the spectral-spatial model of the region. Lande et al. (2012) studied the region surrounding HESS J1837069 (l ~ 25.3°;b ~ −0.3°) in detail based on the LAT catalog. The best-fit model (including spectrum) for this region was used. The source 2FGL J1837.30700c was modeled with a symmetrical Gaussian (σ ≃ 0.33°). The two closest sources, 2FGL J1834.70705c and 2FGL J1836.80623c, were relocalized although their positions did not significantly change.

3.1. Search for pulsed emission

The Fermi LAT is the first instrument sensitive enough to discover new pulsars based on their γ-ray pulsations alone (Abdo et al. 2009a): 41 new γ-ray pulsars have been detected to date in this way. To check for the presence of a γ-ray pulsar in W41, a search for pulsations in Fermi LAT data using the hierarchical blind-search technique described in Pletsch et al. (2012) was performed. To improve the signal-to-noise ratio (S/N) of the putative pulsar in W41, each photon was assigned a probability that it originates from the target source, using the spectral model obtained from the analysis described in Sect. 3.3. While the faintest pulsars detected in blind searches have S/N ~ 15, in the case of W41, S/N ≤ 1: no significant pulsations were found.

thumbnail Fig. 6

Fermi LAT TS map of the W41 region, above 1.8 GeV. The TS was evaluated by placing a point-source at the center of each pixel, Galactic diffuse emission and nearby 2FGL sources are included in the background model. H.E.S.S. significance contours (3, 4, 5, 6, 7 and 8σ) are overlaid as black solid lines. The triangle and the cross indicate the position of the putative pulsar and the OH maser.

Open with DEXTER

3.2. Morphological analysis

Figure 6 shows the test statistic (TS) map between 1.8 and 100 GeV, with H.E.S.S. significance contours (from 3 to 8σ) overlaid. The TS is defined as twice the difference between the log-likelihood L1 obtained by fitting a source model plus the background model to the data, and the log-likelihood L0 obtained by fitting the background model alone, that is, TS = 2ln(L1L0). This skymap contains the TS value for a point source at each map location, thus giving a measure of the statistical significance for the detection of a γ-ray source in excess of the background. The source is detected with TSmax = 111 above 1.8 GeV. During the fit, the spectral parameters of sources closer than 5° to W41 were allowed to vary in the likelihood fit as well as the normalization of the diffuse models, while the parameters of all other sources were fixed at the values of the 2FGL catalog. The source extension was determined using pointlike with a Gaussian hypothesis (compared with a point-source hypothesis). The difference in TS between the Gaussian and the point-source hypothesis is 20, which converts to a significance of ~ 4.5σ. This suggests that the source is extended with respect to the LAT PSF. The intrinsic size is σGeV = (0.15° ± 0.03°) and the best-fit position (l = 23.20° ± 0.03°, b = −0.28° ± 0.03°) is compatible with the candidate pulsar position. This morphology is shown in Fig. 4. An asymmetric Gaussian distribution was also considered but did not improve the fit with respect to the Gaussian hypothesis (ΔTS = 1.5).

3.3. Spectral analysis

The spectrum of the Fermi LAT source was estimated with gtlike taking into account the sky model and the shape obtained with pointlike. The spectrum is well described by a power law (see Eq. (1)). The best-fit parameters are Φ(Edec) = (4.98 ± 0.20stat ± 0.22syst) × 10-13 cm-2 s-1 MeV-1, Edec = 4088 MeV, and Γ = 2.15 ± 0.12stat ± 0.16syst. These results are compatible with the parameters found by Castro et al. (2013).

Two main sources of systematic errors were considered: imperfect modeling of the Galactic diffuse emission and uncertainties in the effective area. The first one was estimated by artificially changing the normalization of the Galactic diffuse model by ± 6%, as in Abdo et al. (2010e). The second one was estimated by using modified IRFs whose effective areas bracket the nominal IRFs. These bracketing IRFs are defined by envelopes above and below the nominal energy dependence of the effective area by linearly connecting differences of (10%, 5%, 20%) at log(E/MeV) of (2, 2.75, 4), respectively (Ackermann et al. 2012).

The Fermi LAT spectral points were obtained by dividing the energy range (1.8100 GeV) into six logarithmically spaced energy bins. A Bayesian upper limit (Loparco et al. 2011) at a 3σ confidence level was calculated if the TS value in the energy bin was lower than 9. The LAT spectral points and their uncertainties are reported in Fig. 5.

The scenario where the emission seen with Fermi LAT is produced by two components was investigated by fixing the best-fit extension and position found in GeV and adding a point-like source at the putative pulsar position. In this case, the point-like source is not significant (TS ≃ 7) and a Bayesian upper limit on its integrated flux was calculated: S(1.8−100 GeV) < 1.0 × 10-12 erg cm-2 s-1 at a 3σ confidence level.

4. Discussion

Follow-up observations of the previously discovered source HESS J1834087 as described here have led to the discovery of a two-component morphology: a central point-like and an extended component. Fermi LAT observations were also analyzed in the GeV energy band, leading to the detection of a γ-ray source (2FGL J1834.30848) coincident with the TeV source and with a similar extension. Figure 4 highlights that the positions of both the TeV and GeV sources are compatible with each other and also with SNR W41. This new information is very useful to constrain the nature of the γ-ray source.

In addition to the extended emission, no significant point-like component was detected using Fermi LAT data, in contrast with H.E.S.S. (see Sect. 2.1). With Fermi LAT observations, Abdo et al. (2013) have found that pulsar spectra usually follow a power law with an exponential cutoff at energies between 1 and 10 GeV. However, such a cutoff is not apparent in the spectrum derived in Sect. 3.3. Moreover, no γ-ray pulsations from the putative pulsar at the center of the remnant were detected. These arguments and the fact that the γ-ray emission observed with Fermi LAT is significantly extended strongly suggest that the signal does not arise from an unseen pulsar.

It is clear from Fig. 1 that the VHE γ-ray emission does not correlate with the SNR radio shell, but comes from the center of the remnant, as shown by the VHE radial profile in Fig. 2. This disfavors a scenario in which the signal originates in the SNR shell, in contrast with the case of RX J1713.73946 (Aharonian et al. 2004) for example.

Table 3

Fitted parameters of the SED modeling shown in Fig. 7 assuming HESS J1834087 is a single physical source as observed from the GeV to the TeV range.

Therefore, only two plausible scenarios remain: either the γ-ray emission is produced by a PWN powered by the pulsar candidate CXOU J183434.9084443 detected in X-rays, or it is caused by the interaction of the SNR with a nearby MC.

Unless otherwise specified, an assumed distance of d = 4.2 kpc was considered in the following sections.

4.1. Pulsar wind nebula candidate scenario

The increasing number of detected PWNe at VHE motivated us to investigate a scenario where the GeV and TeV γ-ray emission is produced by the wind of the putative pulsar CXOU J183434.9084443.

4.1.1. TeV source HESS J1834–087 as a single source powered by the putative pulsar

The spatial coincidence between the centroid of the GeV and the TeV emissions and the pulsar candidate strengthens this hypothesis, as also stated by Misanovic et al. (2011). Interestingly, assuming that the extended component detected in X-rays is a PWN, Misanovic et al. (2011) found that the ratio of the extended source and pulsar candidate X-ray luminosities, LPWNX/LPSRX ~ 1.8, is typical of a PWN/pulsar system. Furthermore, the PWN scenario is reasonable from an energetic point of view. Indeed, assuming a distance of d = 4.2 kpc, Misanovic et al. (2011) derived an X-ray luminosity of LX(2−10 keV) ≃ 1 × 1033 erg s-1. Then, assuming a termination shock radius rs,17 = rs/ (1017 cm) scaled to a plausible value (corresponding to an angular size of 2.7′′ at the assumed distance) and an ambient pressure pamb, −9 = pamb/ (10-9 dyne cm-2) inside the SNR in the Sedov expansion phase, the authors derived a spin-down luminosity of erg s-1, which is typical for Vela-like pulsars and supports the assumption of a young PWN. This value is consistent with the estimates of [1036,1037] erg s-1 that can be derived from Kargaltsev & Pavlov (2008). Pulsars associated with several TeV PWNe have a similar spin-down power (Gallant et al. 2008).

From the data analyses described in the previous sections, the GeV and TeV luminosities are estimated to be LGeV(1−100 GeV) ≃ 1.6 × 1035 erg s-1 and LTeV(1−30 TeV) ≃ 1.1 × 1034 erg s-1, the latter being calculated for the entire emitting region. Mattana et al. (2009) used a large sample of PWNe observed in X-rays and γ-rays to demonstrate the correlation between the luminosity ratio LTeV/LX and the characteristic age τc of the powering pulsar. Using the luminosities derived from XMM-Newton and H.E.S.S. observations, the characteristic age is in the range τc ∈ [3 × 103,6 × 104] yr, which is younger than the estimated age of the remnant (tentative remnant dating: tSNR ∈ [6 × 104,2 × 105] yr ; see Sect. 1).

thumbnail Fig. 7

Spectral energy distribution (SED) for three different assumptions about the nature of the source. Solid line: The association between a single TeV source (violet diamonds) and the GeV source (green squares) has been assumed. A leptonic scenario has been tested where a low-energy Maxwellian distribution replaces the low-energy part of the electron spectrum. The fitted parameters of this model can be found in Table 3. Dashed line: a simple leptonic model has been tested assuming the TeV point-like source (blue points) is associated with a hypothetical GeV point-like source superimposed on the extended one (yellow upper limits). The blue TeV spectral points have been obtained after normalizing the points computed in the central extracted region (see Sect. 2.2 for details). Dash-dotted line: the hadronic model studied by Li & Chen (2012) has been superimposed to the new data points obtained in this work. The brown X-ray spectral point has been derived from the XMM-Newton spectrum given by Mukherjee et al. (2009), the black radio upper limits from VLA 20 cm (Helfand et al. 2006) and Bolocam 1.1 mm (Aguirre et al. 2011) images.

Open with DEXTER

At the assumed distance of 4.2 kpc, the radius of the nebula in the TeV range is r ≃ 12 pc. A corresponding age of [6 × 103,2 × 104] yr is obtained by modeling the PWN evolution inside the SNR (Reynolds & Chevalier 1984) with a mass of the progenitor M ~ 8 M, an explosion energy E0 ~ 1051 erg, a spin-down luminosity of Ė ∈ [1036,1037] erg s-1, and a density inside the shell n = 0.1 cm-3. By using standard values, this age agrees with the estimated characteristic age of the putative pulsar described above, but is still younger than the interval proposed for the remnant.

Therefore, if HESS J1834087 is a PWN powered by the putative pulsar CXOU J183434.9084443, this would imply that the SNR W41 might be younger than 6 × 104 yr, or that W41 is located at a distance smaller than the assumed value, or that the pulsar and the TeV source are simply not related to W41. These conclusions are strongly dependent on the input parameters and were derived assuming an evolutionary phase before the crushing by the reverse shock supported by the X-ray observations described above. Therefore, they must be considered as indicators only.

An SED modeling has been performed assuming that the whole TeV source is associated with the GeV emission. However, unlike most PWNe detected by Fermi LAT, which present a hard spectrum with an average spectral index Γ ~ 1.8 (Acero et al. 2013), the Fermi LAT spectrum of HESS J1834087 is rather soft. In this sense, this γ-ray source resembles HESS J1640465, for which Slane et al. (2010) inferred a distinct population of low-energy electrons based on the particle-in-cell simulations performed by Spitkovsky (2008). We used a similar reasoning by modeling the low-energy part of the electron spectrum by an additional Maxwellian distribution. The solid black curve in Fig. 7 represents the obtained fitted leptonic model (χ2/ndf = 10.5/7). The output parameters are listed in Table 3 and are discussed below. The cosmic microwave background and the Galactic interstellar radiation field at the location of the SNR W41, used to calculate the inverse Compton emission, were derived from the estimates of Porter & Strong (2005). The energy densities for the IR and optical photon fields are 1.6 eV cm-3 and 2.2 eV cm-3, respectively. The characteristic age of the pulsar was assumed to be t = 10 kyr, close to the mean of the ranges previously estimated. The best-fit magnetic field value (B ≃ 14 μG) is similar to the values found for other TeV PWNe (de Jager & Djannati-Ataï 2009). The derived total energy in electrons is We ≃ 2.4 × 1048 erg. Assuming a constant spin-down power over the entire lifetime of the pulsar, this energy corresponds to ~700% or ~70% of the total energy injected by the pulsar for a spin-down power of 1036 erg s-1 or 1037 erg s-1, respectively. In this context, if the whole GeV and TeV emissions correspond to the same PWN, the spin-down power of the pulsar must be close to (or even higher than) ~1037 erg s-1 at present, assuming a constant energy-loss rate, or was much higher than 1036 erg s-1 in the past and decreased with time. A precise modeling of the evolution of the spin-down power and magnetic field with time is beyond the scope of this paper, since most parameters for such a model are poorly constrained.

4.1.2. Only the TeV point-like component is powered by the putative pulsar

In an alternative PWN scenario, we assumed that only the TeV point-like component (spatially coincident with the X-ray point-like source) is produced by the PWN powered by the putative pulsar. Under this assumption, the γ-ray efficiency is estimated to be LTeV/Ė ∈ [0.1%,1%]. This is consistent with the range of efficiencies (0.01–10%) found for several VHE PWNe candidates (Gallant et al. 2008).

A simple one-zone leptonic model was considered and is displayed in Fig. 7 (dashed line). The TeV data were computed from the normalized data found in the central extraction region that are displayed in Fig. 5. The calculation of the normalization is detailed in Sect. 2.2. Despite the large spectral uncertainties, an SED modeling was performed to estimate the physical parameters of the source. The model assumes a distribution of accelerated electrons that cool radiatively by means of synchrotron radiation and inverse-Compton scattering, a distance of 4.2 kpc, and an age of 104 yr for the pulsar. This age is younger than that estimated for W41, but still in the estimated range for τc. A good representation of the radio upper limits and X-ray data together with the GeV point-like source upper limits and TeV central component is obtained for an injection spectral index Γinj = −2.0 and a magnetic field of B ≃ 15 μG. The total energy injected to leptons (We ≃ 3.0 × 1047 erg) represents ~10% (or 100%) of the rotational energy that can be injected by the pulsar over its entire lifetime, assuming a constant rotational power of 1037 erg s-1 (or 1036 erg s-1, respectively), which again favors a spin-down power higher than 1036 erg s-1 for the powering pulsar.

In this scenario, the TeV point-like source is assumed to be unrelated to the TeV extended component. Thus, the extended structure seen by Fermi LAT and H.E.S.S. may have two possible origins. Electrons accelerated in the past and affected by radiative losses could create a relic nebula that emits in γ-rays. This ancient PWN would be too faint to be detected in X-rays. An energy-dependent morphology of the γ-ray emission from 1.8 GeV up to 100 TeV might suggest electron cooling in the PWN. However, without any significant variation of the intrinsic sizes of the γ-ray source in this energy range or variation of the spectral slope at VHE energy, this scenario tends to be disfavored. The other possibility suggests that the emission from the interaction of the SNR with the MC is in the line of sight of the PWN γ-ray emission. But the chance probability of detecting two TeV sources in spatial coincidence in the part of the Galactic plane covered by H.E.S.S. (140° × 6°) was estimated to be p ≃ 9.4 × 10-4 from the formula in Akujor (1987). Nevertheless, this configuration may occur, as in the case of W51C, where the TeV source seems to be powered by a PWN and an SNR/MC interaction (Aleksić et al. 2012; Feinstein et al. 2009).

4.2. SNR-MC interaction

Using high-resolution CO observations, Tian et al. (2007) suggested that a MC is located close to the SNR W41 and could even be interacting with the SNR, as proposed by Albert et al. (2006). The γ-ray emission detected by H.E.S.S. (as seen in Fig. 1) is spatially coincident with the 13CO high-density region extracted from GRS data (Jackson et al. 2006), as observed for other interacting SNRs at TeV energy. Direct evidence that the shock wave of W41 is interacting with a MC has been discovered recently. Frail et al. (2013) found two regions with OH maser emission close to the center of the remnant. Only the brightest maser at l,b = (23.26°, −0.31°) is displayed in Figs. 1, 4, and 6.

Fermi LAT observations have already led to the detection of γ-ray emission from several SNRs. Interestingly, SNRs interacting with MCs constitute the dominant class of γ-ray luminous SNRs, and several evolved interacting remnants (i.e., W51C, W44, IC443, and W28) are associated with extended GeV emission (Abdo et al. 2009b, 2010b,c,d). For W41, the estimated luminosity L(0.1−100 GeV) assuming d = 4.2 kpc is similar to that of other interacting SNRs detected by Fermi, as shown in Fig. 8 (Thompson et al. 2012). In this context, Li & Chen (2012) proposed an accumulated diffusion model in which the MC size is taken into account and where the diffusion distance of the injected protons varies as the remnant expands. This illuminated cloud model reproduces the GeV and TeV emission using a reasonable energy budget: the total energy injected up to Emax ≃ 50 TeV is Wp ≃ 1050 erg. The set of parameters is summarized in Table 3 and reproduced in Fig. 7 (dash-dotted line). This model was obtained from the spectrum published in Méhault et al. (2011). The flat spectrum seen by Fermi up to several tens of GeV implies a break at ~100 GeV, which is unusual for interacting SNRs. This led Li & Chen (2012) to use a cloud as large as the shock wave radius and a diffusion coefficient lower than the average Galactic value. This may occur if the interstellar medium is highly ionized by cosmic rays.

thumbnail Fig. 8

GeV luminosity (0.1100 GeV) as a function of diameter squared for each of the SNRs detected with Fermi LAT (Thompson et al. 2012). Filled circles correspond to the SNRs interacting with molecular clouds. Rectangles represent young SNRs. Puppis A, a middle-aged SNR without clear molecular interaction, is shown as an open circle. The source coincident with the SNR W41 is marked by a red triangle.

Open with DEXTER

5. Conclusion

Follow-up observations have been performed from 1.8 GeV to up to 100 TeV in the direction of the SNR W41 to provide new clues on the origin of the γ-ray emission. In the VHE domain, the H.E.S.S. data exhibit a two-component morphology: a central point-like source and an extended component. The detailed study of Fermi LAT HE γ-ray observations revealed extended emission with an intrinsic size (0.15°) similar to that of the TeV extended component (0.17°). The GeV and TeV emissions are spatially coincident with a candidate pulsar located at the center of the remnant. However, no pulsations in radio, X-rays, or HE γ-rays were detected. Different scenarios were investigated to explain the γ-ray emission.

First, the spatial coincidence of the best-fit position of the γ-ray sources with an X-ray CCO possibly surrounded by a PWN strengthens the PWN scenario. The potential CCO is able to power the whole γ-ray emission only if its current spin-down is close to (or even higher than) 1037 erg s-1. Then, the flat GeV spectrum can only be reproduced assuming a second low-energy electron population. These two aspects are very similar to the case of the PWN candidate HESS J1640-465 (Slane et al. 2010).

Second, the extended GeV and the whole TeV sources are coincident with a high-density region traced by 13CO. A model based on the assumption of an interaction between W41 and the nearby MC reproduces the spectral data with a reasonable energy budget. Because of the flat Fermi LAT spectrum, an unusually high-energy spectral break at ~100 GeV is required. However, the interacting scenario is reinforced by the recent discovery of OH (1720 MHz) maser lines coincident with the remnant and compatible with the associated cloud velocity.

The last possibility would be that the TeV emission has two different sources. The point-like component would originate from the PWN powered by the putative pulsar, while the extended TeV component that matches the GeV source would be created by the interaction between the remnant and the cloud. In this case, the spatial coincidence between the point-like and the extended VHE γ-ray emission would occur by chance, but the probability is very low.

Nevertheless, the hadronic scenario seems to explain both GeV and TeV extended emission, as argued in this work. This origin is supported by the recent maser detection and theoretical studies of this γ-ray source.

A clear identification of the extended component in X-rays could help to investigate the hypothesis of a PWN, but given the low-energy flux expected from typical relic PWNe, this appears to be a serious observational challenge. Deeper VHE observations with the Cherenkov Telescope Array (CTA) might also allow higher energy resolution studies and a firm identification of the origin of the VHE emission, in particular through morphological analysis. Concerning the Fermi LAT analysis, the next improvement of the Galactic diffuse emission model and the IRFs might ameliorate the study of the GeV γ-ray emission at lower energies in order to better constrain the modeling of the spectral energy distribution.


Acknowledgments

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 UK 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. The Fermi LAT Collaboration acknowledges generous ongoing support from a number of agencies and institutes that have supported both the development and the operation of the LAT as well as scientific data analysis. These include the National Aeronautics and Space Administration and the Department of Energy in the United States, the Commissariat à l’Énergie Atomique and the Centre National de la Recherche Scientifique/Institut National de Physique Nucléaire et de Physique des Particules in France, the Agenzia Spaziale Italiana and the Istituto Nazionale di Fisica Nucleare in Italy, the Ministry of Education, Culture, Sports, Science and Technology (MEXT), High Energy Accelerator Research Organization (KEK) and Japan Aerospace Exploration Agency (JAXA) in Japan, and the K. A. Wallenberg Foundation, the Swedish Research Council and the Swedish National Space Board in Sweden. Additional support for science analysis during the operations phase is gratefully acknowledged from the Istituto Nazionale di Astrofisica in Italy and the Centre National d’Études Spatiales in France.

References

All Tables

Table 1

Centroid (Galactic coordinates) and extension fits to the H.E.S.S. events spatial distribution in the W41 region.

Table 2

H.E.S.S. best-fit spectral parameters assuming a power-law model (Eq. (1)) between 0.2 and 30 TeV.

Table 3

Fitted parameters of the SED modeling shown in Fig. 7 assuming HESS J1834087 is a single physical source as observed from the GeV to the TeV range.

All Figures

thumbnail Fig. 1

H.E.S.S. image of the VHE γ-ray excess in the direction of SNR W41 after standard cut analysis. The image was smoothed with a Gaussian kernel of 0.068° corresponding to the instrument’s point-spread function for this data set, which is shown in the inset. Green contours represent the GRS 13CO data around W41 integrated in the velocity range 74 km s-1 to 82 km s-1 and smoothed with the same kernel. White contours of the VLA radio data at 20 cm show the shell-type SNR. The position of CXOU J183434.9084443 and the brightest OH (1720 MHz) maser are indicated with a triangle and a cross, respectively. The linear color scale is in units of counts per smoothing Gaussian width.

Open with DEXTER
In the text
thumbnail Fig. 2

H.E.S.S. radial profile of the uncorrelated excess map centered on the position of the putative pulsar. The solid dark blue line shows the result of the best-fit model (labeled (C) in Table 1) estimated in Sect. 2.1. The dark blue and light blue dashed lines represent the point-like and Gaussian components of this model. The profile from the single Gaussian model (labeled (B) in Table 1) is plotted in orange.

Open with DEXTER
In the text
thumbnail Fig. 3

H.E.S.S. slices in the uncorrelated excess map along the direction perpendicular to the Galactic plane (white arrow in the inset) in two independent energy bands (E< 1 TeV: blue and E> 1 TeV: orange). The slices are centered on the candidate pulsar position (marked by a triangle in the inset). The composite color image (inset) shows the box (height = 1.0°,width = 0.22°) used for the profile.

Open with DEXTER
In the text
thumbnail Fig. 4

20 cm radio continuum map (Helfand et al. 2006) with superimposed best-fit positions as vertical crosses (their sizes indicate the statistical errors) and intrinsic extensions as rings. Magenta and dashed blue represent the Gaussian model of the Fermi LAT source and the extended component of the two-component model of the H.E.S.S. source. The inner and outer radius represents ± 1σ errors on the fitted extension. The candidate pulsar (triangle, brown) and OH maser (cross, yellow) positions are also reported.

Open with DEXTER
In the text
thumbnail Fig. 5

Spectral points and bowties of the LAT source (squares - statistical and systematic uncertainties added in quadrature in green and only statistical errors in black) and the H.E.S.S. annular (red diamonds) and central (blue points) regions. Bowties represent the 1σ confidence level error on spectra. The upper limits are calculated for a 3σ confidence level.

Open with DEXTER
In the text
thumbnail Fig. 6

Fermi LAT TS map of the W41 region, above 1.8 GeV. The TS was evaluated by placing a point-source at the center of each pixel, Galactic diffuse emission and nearby 2FGL sources are included in the background model. H.E.S.S. significance contours (3, 4, 5, 6, 7 and 8σ) are overlaid as black solid lines. The triangle and the cross indicate the position of the putative pulsar and the OH maser.

Open with DEXTER
In the text
thumbnail Fig. 7

Spectral energy distribution (SED) for three different assumptions about the nature of the source. Solid line: The association between a single TeV source (violet diamonds) and the GeV source (green squares) has been assumed. A leptonic scenario has been tested where a low-energy Maxwellian distribution replaces the low-energy part of the electron spectrum. The fitted parameters of this model can be found in Table 3. Dashed line: a simple leptonic model has been tested assuming the TeV point-like source (blue points) is associated with a hypothetical GeV point-like source superimposed on the extended one (yellow upper limits). The blue TeV spectral points have been obtained after normalizing the points computed in the central extracted region (see Sect. 2.2 for details). Dash-dotted line: the hadronic model studied by Li & Chen (2012) has been superimposed to the new data points obtained in this work. The brown X-ray spectral point has been derived from the XMM-Newton spectrum given by Mukherjee et al. (2009), the black radio upper limits from VLA 20 cm (Helfand et al. 2006) and Bolocam 1.1 mm (Aguirre et al. 2011) images.

Open with DEXTER
In the text
thumbnail Fig. 8

GeV luminosity (0.1100 GeV) as a function of diameter squared for each of the SNRs detected with Fermi LAT (Thompson et al. 2012). Filled circles correspond to the SNRs interacting with molecular clouds. Rectangles represent young SNRs. Puppis A, a middle-aged SNR without clear molecular interaction, is shown as an open circle. The source coincident with the SNR W41 is marked by a red triangle.

Open with DEXTER
In the text

Current usage metrics show cumulative count of Article Views (full-text article views including HTML views, PDF and ePub downloads, according to the available data) and Abstracts Views on Vision4Press platform.

Data correspond to usage on the plateform after 2015. The current usage metrics is available 48-96 hours after online publication and is updated daily on week days.

Initial download of the metrics may take a while.