Chandra Observations of the Field Containing HESS J1616–508

The High Energy Stereoscopic System (HESS) is an array of atmospheric Cherenkov telescopes sensitive to photons above 100 GeV, which has discovered >75 TeV sources in the Galactic plane (Aharonian et al. 2005).⁵ Among these sources, there are a few that have not been associated with any known objects (Aharonian et al. 2008; Kargaltsev et al. 2013). The most straightforward way to understand what type of source is producing the high energy emission is to use observations at longer wavelengths. However, for some of the TeV sources, attempts to find their X-ray counterparts with Chandra or XMM-Newton has either failed or produced an ambiguous result, causing these sources to be dubbed "dark accelerators" (Aharonian et al. 2005). Observations of these fields with Fermi-LAT have revealed possible GeV counterparts for many TeV sources (Nolan et al. 2012; Acero et al. 2013; Acero et al. 2015), but the angular resolution of the LAT is often not sufficient to establish a firm association.

Spatially extended TeV sources are often associated with energetic pulsars located nearby (10'–15' from the center of the TeV emission; see Kargaltsev et al. 2012 and Kargaltsev et al. 2013 for reviews). Pulsars eject relativistic particles into their surroundings and can fill large bubbles (plerions) with these particles during their lifetime. These particles can produce TeV photons by inverse Compton (IC) upscattering of background photons. The large observed offsets between pulsar position and TeV source could, in some cases, be explained by the crushed plerion scenario, which assumes that the asymmetric reverse shock from the host supernova remnant (SNR) has displaced the plerion from the pulsar (Blondin et al. 2001). The offset extended TeV source would then represent a relic plerion filled with aged electrons. These same pulsars can also power compact X-ray pulsar wind nebulae (PWNe), whose synchrotron emission is produced by "younger," more energetic electrons. In some cases, such nebulae are detected before the pulsar is discovered (see Table 1 in Kargaltsev et al. 2013), revealing the nature of the TeV source.

HESS J1616–508 (hereafter HESS J1616) is one of the brightest unidentified extended (∼16' rms size) TeV sources (Aharonian et al. 2006). The source has a 1–10 TeV flux F_1–10TeV ≈ 1.7 × 10⁻¹¹ erg s⁻¹ cm⁻², and its spectrum can be fit by a power-law model with Γ = 2.34 ± 0.06. HESS J1616's center is located about 10' east of the young ($\tau \equiv P/2\dot{P}=8.13$ kyr) and energetic ($\dot{E}=1.6\times {10}^{37}$ erg s⁻¹) pulsar, PSR J1617–5055 (PSR J1617, hereafter; Kaspi et al. 1998). PSR J1617 has a period P = 69 ms, a magnetic field B = 3.1 × 10¹² G, and is estimated to be at a distance of 6.5 kpc. This pulsar is energetic enough to power the TeV source, however, no other links between the PSR J1617 and HESS J1616 have been found (Kargaltsev et al. 2009; K+09 hereafter).

The field of HESS J1616 has been observed in X-rays multiple times. Matsumoto et al. (2007) observed the center of the HESS J1616 field with Suzaku XIS but found no counterpart, to a limiting flux of 3 × 10⁻¹³ erg s⁻¹ cm⁻² in the 2–10 keV energy range. The field was also observed with INTEGRAL/IBIS/ISGRI and Swift XRT. The data were analyzed by Landi et al. (2007); however, due to the lack of any other obvious X-ray counterpart, they concluded that the most likely counterpart to the TeV source was PSR J1617.

K+09 analyzed a dedicated 60 ks Chandra X-ray Observatory (CXO) observation of PSR J1617, as well as the available archival data (including XMM-Newton and Chandra observations of nearby SNRs). K+09 found that the under-luminous X-ray PWN (${L}_{0.5\mbox{--}8\mathrm{keV},\mathrm{PWN}}\sim 3.5\times {10}^{33}$ erg ${{\rm{s}}}^{-1}\,\sim 0.2{L}_{0.5\mbox{--}8\mathrm{keV},\mathrm{PSR}}\sim 2\times {10}^{-4}\,\dot{E}$, at d = 6.5 kpc) does not extend in the direction toward the center of the TeV source (unlike, e.g., the X-ray PWN of PSR B1823–13 and HESS J1825–137; Pavlov et al. 2008). K+09 suggested that even if PSR J1617 is responsible for a fraction of the TeV emission, HESS J1616 could be a double or multiple source, which might include an unknown SNR or PWN. K+09 also reported an X-ray source, dubbed "Source X," with a flux F_{0.5–8 keV} ∼ 2 × 10⁻¹³ erg cm⁻² s⁻¹ in the 0.5–8 keV band located close to the center of the TeV source. However, this source was imaged 16' off-axis (in an archival Chandra observation), making it impossible to distinguish between a single extended source or several point sources.

There is also an extended (σ ∼ 19') source reported at the exact same coordinates in both the Fermi-LAT 3FGL and 2FHL source catalogs (Acero et al. 2015; Ackermann et al. 2016), named 3FGL J1616.2–5054e and 2FHL J1616.2–5054e, respectively. This source is spatially coincident with the TeV source (i.e., the center of this Fermi-LAT source is only ∼36'' from the center of HESS J1616). The LAT source has fluxes F = (1.56 ± 0.08) × 10⁻¹⁰ erg cm⁻² s⁻¹ in the 1–100 GeV band, F = (7.3 ± 1.7) × 10⁻¹¹ erg cm⁻² s⁻¹ in the 50 GeV–2 TeV band and spectral indices Γ = 2.14 ± 0.03, Γ = 1.74 ± 0.18 in the 3FGL and 2FHL catalogs, respectively. The spectrum of this source shows no sign of curvature, which is typically seen in the Fermi-LAT spectra of pulsars⁶ (Abdo et al. 2013).

Another potential counterpart of HESS J1616 is PSR J1614–5048 (PSR J1614 hereafter; Johnston et al. 1992), located 22' west of the TeV source center. PSR J1614 is a young pulsar, based on its spin-down age τ = 7.4 kyr. However, in some cases, this age may be a poor indicator of the true age. The pulsar is estimated to be at d ∼ 7 kpc, which was determined from its dispersion measure DM = 586 pc cm⁻³ (Taylor & Cordes 1993). The pulsar has a period P = 232 ms, a rather low spin-down power for a young pulsar $\dot{E}=1.6\times {10}^{36}$ erg s⁻¹, and a high surface dipole magnetic field B = 1.1 × 10¹³ G (Wang et al. 2000). This pulsar is also interesting because it has experienced two of the largest glitches ever observed and is also one of the noisiest pulsars (Wang et al. 2000; Yu et al. 2013).

Here we report the results of our analysis of three new Chandra observations, two of them covering the center of HESS J1616 (ObsIDs 16956 and 17676) and another one (ObsID 16955) covering the field of PSR J1614 (Section 2). We also search for multi-wavelength (MW) counterparts to the detected X-ray sources (Section 3), use the MW data to classify the X-ray sources detected in these observations, and discuss whether any of these sources could be a counterpart to HESS J1616 (Section 4). We find that PSR J1614 is not detected in X-rays (Section 5), and summarize our findings in Section 6.

We have analyzed three Chandra observations. The first two, ObsIDs 16956 and 17676, observed on 2015 June 18 and 19, covered the central region of HESS J1616 (Field 1 hereafter; 24.7 ks and 33.6 ks exposures, respectively), while the third observation, ObsID 16955, observed on 2015 June 15, covered the field containing PSR J1614 (Field 2, hereafter; 59.3 ks exposure). Fields 1 and 2 are shown in Figure 1. All three observations were obtained with the ACIS-I instrument operated in timed exposure mode using the Very Faint telemetry format. The data were processed with the Chandra Interactive Analysis of Observations (CIAO⁷ ) software version 4.6 and the Chandra Calibration Database (CALDB) version 4.6.3. We restricted our analysis to the 0.5–7 keV range, unless otherwise specified. The spectra were fit using XSPEC version 12.8.2.

Zoom In Zoom Out Reset image size

CIAO's source detection routine wavdetect (Freeman et al. 2002) was used to detect X-ray point sources and determine their coordinates in the Chandra images (see Table 1 and Figure 2). In order to detect fainter point sources, the data for ObsIDs 17676 and 16956 (Field 1) were merged⁸ prior to running wavdetect. The srcflux CIAO tool was then run individually on each observation (using the coordinates found by running wavdetect on Field 1) and the fluxes were averaged by weighting the flux in each observation with its corresponding exposure time. There were 26 and 21 sources detected with a wavdetect reported significance >6 for Field 1 and Field 2, respectively (see Figure 2). A cut on significance is needed to ensure that meaningful hardness-ratio values can be obtained. Several off-axis sources (5 in Field 1 and 4 in Field 2) had wavdetect significance values <6 due to a higher background in their immediate vicinity caused by nearby very faint sources or hints of extended emission. We have added these sources to our source list by hand and marked them with green labels in Figure 2.

Zoom In Zoom Out Reset image size

The X-ray spectra for the five brightest sources (≥100 counts) have been created by the srcflux CIAO tool. For sources with >200 counts, we have binned the X-ray spectra to have a minimum of 10 counts per bin. Otherwise, the spectra were not binned and were fit using C-statistics (Cash 1979). For Field 1, the spectra were extracted separately for each observation and combined using the combine_spectra CIAO tool.⁹ All fits used the XSPEC tbabs model (Wilms et al. 2000) for interstellar absorption.

Along with calculating X-ray properties for each source (i.e., the 0.5–2 keV flux, 2–7 keV flux, and two hardness ratios), we have also searched MW catalogs for potential counterparts¹⁰ to these sources and compiled a set of MW parameters for each of the sources with counterparts. The MW photometry was taken in optical from the USNO-B (Monet et al. 2003) catalog, the near-infrared (NIR) photometry was taken from the Two Micron All Sky Survey (2MASS; Skrutskie et al. 2006), and the IR photometry was taken from the Wide-field Infrared Survey Explorer (WISE; Cutri et al. 2012). We found that 16 out of 31 and 14 out of 25 of the X-ray sources had MW counterparts in Field 1 and Field 2, respectively. MW sources were considered potential counterparts if they were within the error radius of each X-ray source. This error radius was calculated using the empirical relations provided by the Chandra Multi-wavelength Project (see Equation (12) in Kim et al. 2007). Then, we followed the Chandra Source Catalog (Evans et al. 2010), and added an additional 2σ astrometric error of 039 in quadrature (see Table 1). The average 2σ error circle for all of the sources in both fields is 12. The chance coincidence probabilities for each catalog were calculated by finding the average source density in these fields, ρ, and calculating the probability of having one or more sources within a randomly placed circle, $P=1-\exp (-\rho \pi {r}^{2})$. The source densities for each of the catalogs in these fields are ρ_USNOB = 0.00342 arcsec⁻², ρ_2MASS = 0.00895 arcsec⁻², and ρ_WISE = 0.00168 arcsec⁻², leading to chance coincidence probabilities of 1.5%, 4.0%, and 0.8%, respectively, for r = 12. The 2MASS catalog has the largest chance coincidence probability (of the cross-matched catalogs), which implies that up to ∼2 2MASS sources may be spurious cross-matches with an X-ray source for both fields combined. None of the X-ray sources had more than one WISE or 2MASS counterpart; however, two X-ray sources (19 and 26 in Field 2) had two USNO-B counterparts. For these cases, the closest optical source to the X-ray source position was taken to be the optical counterpart. The MW magnitudes for each counterpart are given in Table 2.

The magnitudes¹¹ for the MW counterparts together with the X-ray properties (fluxes in two bands and two hardness ratios), add up to 19 features (or parameters), which are then used to classify these sources with our machine-learning pipeline (see the appendix of Hare et al. 2016). One minor change to the pipeline used in Hare et al. (2016) is that we no longer use the See5 implementation of the C5 decision tree algorithm (Quinlan 1993) and solely rely on a Random Forest classifier (Breiman 2001) implemented in python¹² (Pedregosa et al. 2012). We have also reddened the AGNs in the training data set using the total Galactic absorption column in the direction of our observations (N_H = 2.4 × 10²² cm⁻²; Dickey & Lockman 1990) to make them similar to the background AGNs in this field.

To search for possible radio counterparts we used radio data from The Sydney University Molonglo Sky Survey (SUMMS; Mauch et al. 2003) taken at 843 MHz, which is shown in the top panel of Figure 1. The bottom panel of Figure 1 shows a false color (red–8.0 μm; green–5.8 μm; blue–3.6 μm) Spitzer IRAC image. There is a bright radio source seen toward the top of CXO Field 1, which can also be seen in the Spitzer image. This is a known young massive stellar (Roman-Lopes & Abraham 2004) with a lower limit on the bolometric luminosity of $8.7\times {10}^{4}\,{L}_{\odot }$; it contains a number of O and B-type stars (35 cluster member candidates detected down to the completeness limit). Field 2 partially covers the SNRs Kes 32 and G332.0+0.2, with Kes 32 being partially seen in the CXO image. G332.0+0.2 is not detected in the Chandra observation.

Lastly, we have also searched the INTEGRAL/IBIS 7-year source catalog (Krivonos et al. 2010), which has been observing the field of HESS J1616 regularly since 2013 February. We found only one source (associated with the nearby pulsar PSR J1617) within 30' of the center of each field. Therefore, we conclude that there are no INTEGRAL counterparts to any other sources in these fields, down to a limiting flux of 3.7 × 10⁻¹² erg s⁻¹ cm⁻² in the 17–60 keV energy band (Krivonos et al. 2010).

Out of the 56 sources listed in Table 1 only 5 had enough counts (≥100) for meaningful spectral fitting. The count rate spectra of these sources have been created using the specextract CIAO tool called by the srcflux CIAO script. This script creates circles with differing¹³ radii that enclose 90% of the PSF at 1 keV at a given location on the chip. The background is also automatically extracted by the srcflux script from adjacent annular regions. We fit each of the five spectra with blackbody (BB), power law (PL), and APEC (Smith et al. 2001) models and report the results of the best-fit models in Table 3. We show the best-fit spectra in Figure 4. Due to the low number of counts for sources 18 and 25 in Field 1, their spectra were left unbinned and fit using C-statistics.¹⁴

Table 3.  Best-fit Spectral Parameters for Sources with ≥100 Counts

Field	Source	Modela	Normb	NH cm−2	Γ/kT	χ2 or C-stat (C)/dof
			×10−5	×1022	.../keV
2	1	PL	${2.0}_{-0.8}^{+1.4}$	${3.6}_{-0.8}^{+0.9}$	Γ = 0.2 ± 0.3	64.16/58
1	6	AP	${2.4}_{-0.5}^{+0.6}$	0.2+0.1	$\mathrm{kT}={0.96}_{-0.07}^{+0.06}$	114.01/91 (C)
1	13	PL	${4.7}_{-1.2}^{+1.6}$	0.4 ± 0.2	Γ = 2.3 ± 0.3	32.34/32
1	13	AP	${8.1}_{-0.9}^{+1.4}$	${0.2}_{-0.1}^{+0.2}$	$\mathrm{kT}={4.1}_{-1.0}^{+1.1}$	32.99/32
1	18	PL	${1.0}_{-0.4}^{+0.9}$	0.5 ± 0.4	Γ = 1.8 ± 0.5	89.96/90 (C)
1	25	BB	${570}_{-270}^{+560}$	${0.6}_{-0.4}^{+0.5}$	kT = 0.9 ± 0.2	62.76/94 (C)
1	25	PL	${3.0}_{-1.8}^{+5.1}$	${2.0}_{-0.8}^{+0.9}$	Γ = 2.3 ± 0.7	62.94/94 (C)
1	25	AP	${5.5}_{-1.7}^{+2.7}$	${1.8}_{-0.6}^{+0.7}$	$\mathrm{kT}={3.7}_{-1.4}^{+5.3}$	62.28/94 (C)

Notes. All uncertainties are at 68% confidence for each single interesting parameter. Sources whose spectra are well fit by multiple models have all acceptable fits listed in the table.

^aThe models are abbreviated as follows: power law (PL), APEC (AP), and blackbody (BB). ^bThe normalizations are defined differently for each model. The BB normalization is defined as $N={R}_{\mathrm{km}}^{2}/{D}_{10\mathrm{kpc}}^{2}$, where R is the source radius in units of km and D is the distance to the source scaled to 10 kpc. The PL normalization N is given in units of photons keV⁻¹ cm⁻² s⁻¹ at 1 keV. The normalization of the AP model is defined by the integral $N={10}^{-14}/(4\pi [{D}_{A}{(1+z)}^{2}])\int {n}_{{\rm{e}}}{n}_{{\rm{H}}}{dV}$, with D_A defined as the distance to the source in cm, n_e is the electron density in cm⁻³, and n_H is the hydrogen density in cm⁻³.

Download table as:  ASCIITypeset image

4.1. Brighter Sources in Fields 1 and 2

4.2. Field 1 Source 6 (CXOU J161723.7–505152)

4.3. Field 1 Source 13 (CXOU J161551.0–504925)

Source 13's spectrum was extracted from an r = 7'' circle, while the background was taken from a 7'' < r < 34'' annulus. This source has 388 net counts and is the second brightest source detected in either of the Chandra fields. It should be noted that Source 13 is the closest source with >100 counts to the center of the TeV emission (∼7' offset). There is a very bright (B = 14.42, J = 9.88, W1 = 8.80) counterpart detected in USNO-B, 2MASS, and WISE, which is offset from the X-ray position by ∼05. It is also found in The Second-Generation Guide Star Catalog (Lasker et al. 2008). This object has a measured proper motion in the UCAC4 catalog of μ_α cos δ = (−7.7 ± 1.8) mas yr⁻¹, μ_δ = (−8.5 ± 1.8) mas yr⁻¹ (Zacharias et al. 2013), ruling out the possibility that it is an AGN. It is possible, although unlikely, that the optical/IR source is accidentally projected at the X-ray source position. However, the X-ray hardness ratios¹⁷ also rule out the possibility that this source is an obscured AGN undetected in the optical/NIR (see Figure 3). The X-ray spectrum is well fit by either an absorbed APEC or PL model. The APEC model produces a temperature of ${kT}={4.1}_{-1.0}^{+1.1}$ keV and absorption column ${N}_{{\rm{H}}}=({2}_{-0.1}^{+0.2})\times {10}^{21}$ cm⁻², while the best-fit absorbed PL model has a photon index Γ = 2.3 ± 0.3 and absorption column N_H = (4 ± 2) × 10²¹ cm⁻² (see Figure 4 for the spectrum and best-fit PL model).

Zoom In Zoom Out Reset image size

Zoom In Zoom Out Reset image size

The large amount of MW data provided for this source allowed us to fit for a photometric spectral classification. We used the SEDfitter¹⁸ (Robitaille et al. 2007) code to fit the photometry of this source to stellar models from Castelli & Kurucz (2004). The best-fit models for this source are those of M0V–M3V M-dwarf stars. The X-ray-to-B-band luminosity ratio $\mathrm{log}({L}_{0.5\mbox{--}2\mathrm{keV}}/{L}_{B})=-3.1$ is also consistent with a late-type star. Our classification pipeline also classified this source as a star with a confidence of 72% (the Guide Star Catalog also classifies this source as a star). The distance to this source, using the source's V-band magnitude V = 13.04 (Zacharias et al. 2013), is 74–29 pc, giving an X-ray luminosity L_{0.5–7 keV} ∼ 8 − 1 × 10²⁸ erg s⁻¹ for M0V and M3V spectral types, respectively. This X-ray luminosity is within the typical range for M-type stars (see e.g., Güdel 2004 and Preibisch & Feigelson 2005). We have also searched for variability in the light curve of this source but none was found. The classification as a late-type dwarf star makes this source an unlikely counterpart to the TeV source.

4.4. Field 1 Source 18 (CXOU J161643.5–504604)

4.5. Field 1 Source 25 (CXOU J161703.6–504705)

4.6. Remaining Sources

PSR J1614 was imaged ∼08 away from the optical axis in Field 2. We find 5 photons with energies between 0.5 and 8 keV¹⁹ within the source region defined as an r = 21 circle centered on the interferometric radio position, α = 16^h14^m116(8), δ = −50°48019(1) (Stappers et al. 1999). The local background estimated from the annular, 21 < r < 35'', region centered on J1614 is 0.085 ± 0.005 counts arcsec⁻². On average, ∼1 count within the source region is expected to be due to the background, implying a detection with only ∼2σ source significance. Therefore, we calculate a 90% confidence upper limit of 0.15 counts ks⁻¹ on the count rate from PSR J1614 and its possible compact PWN using Table 1 in Gehrels (1986). To calculate the upper limit on the unabsorbed flux in the 0.5–8 keV band, we used PIMMS and assumed an absorbed PL model with Γ = 2 and N_H = 1.75 × 10²² cm⁻² derived from the pulsar's DM = 582.8 cm⁻³ pc according to Equation (2) from He et al. (2013). The estimated upper limit of 5.0 × 10⁻¹⁵ erg cm⁻² s⁻¹ corresponds to an X-ray luminosity L_{0.5–8 keV} = 2.9 × 10³¹ erg s⁻¹ at an assumed distance d = 7 kpc. This makes PSR J1614 one of the least efficient X-ray pulsars known, with an upper limit on its X-ray efficiency of ${\eta }_{0.5\mbox{--}8\mathrm{keV}}\equiv {L}_{0.5\mbox{--}8\mathrm{keV}}/\dot{E}\sim 2\times {10}^{-5}$ (see Figure 5, top panel).

Zoom In Zoom Out Reset image size

PSR J1614 is not detected at GeV energies, which is similar to some other pulsars with small X-ray efficiencies (e.g., PSR J1906+0746, PSR J1913+1011). One exception is the Vela pulsar (Pavlov et al. 2001), which has a small X-ray efficiency and is detected at GeV energies. However, Vela is much closer (∼300 pc; Dodson et al. 2003) than the aforementioned pulsars. To understand how PSR J1614 compares to other similar pulsars, we have looked at the nearest pulsars on the P–$\dot{P}$ diagram²⁰ (see Figure 5, bottom panel) with comparably high magnetic fields (i.e., ≳10¹³ G). Unlike PSR J1614, neither of the two neighboring pulsars in the P–$\dot{P}$ diagram (i.e., J1640–4631 and J0007+7303) are detected at radio wavelengths, which could be due to a low intrinsic radio efficiency or a misalignment of their radio beams with the line of sight.

Interestingly, PSR J1640–4631 has been associated with the TeV source HESS J1640–465 and the corresponding 1FHL source J1640.5–4634 (Gotthelf et al. 2014). This pulsar has been detected by XMM-Newton, pulsations were discovered by NUSTAR (Gotthelf et al. 2014), and a PWN was resolved with Chandra (Lemiere et al. 2009). This pulsar is a factor of two younger than PSR J1614 (based on the spin-down ages) but it may be a factor of two more distant (d = 12 kpc based on the DM; Taylor & Cordes 1993). However, the younger age and slightly higher spin-down luminosity $\dot{E}=4.4\times {10}^{36}$ erg s⁻¹ (a factor of 2.8 larger than that of PSR J1614) can hardly explain the 100 times higher X-ray efficiency (η_{0.5–8 keV} ≈ 2 × 10⁻³).

PSR J0007+7303 also has both an extended TeV counterpart detected by VERITAS (likely the PWN; Aliu et al. 2013) and a Fermi-LAT counterpart (Abdo et al. 2012) with pulsations detected, as well as extended emission, likely from the PWN. This pulsar and its wind nebula have also been detected by Chandra (Halpern et al. 2004). PSR J0007+7303 is a factor of two older than PSR J1614, with a characteristic age τ = 13.9 kyr, and it has a lower spin-down luminosity $\dot{E}=4.5\times {10}^{35}$ erg s⁻¹. It has a low X-ray efficiency η_{0.5–10 keV} ≈ 10⁻⁵, comparable to that of PSR J1614.

Overall, it is interesting to note that several of the pulsars that are nearest to PSR J1614 on the P–$\dot{P}$ diagram have TeV sources associated with their PWNe, implying that it may be possible for PSR J1614 to power a relic PWN that is responsible for the TeV emission. However, if we assume that PSR J1614 is responsible for the TeV emission, then the lower limit on the TeV-to-X-ray luminosity ratio of its undetected PWN, L_1–10TeV/L_{0.5–8 keV} > 3400, is larger than any other known pulsar with associated TeV emission (see Figure 4 in Kargaltsev et al. 2013). We can also compare the lower limit on the GeV-to-X-ray luminosity ${L}_{1\mbox{--}100\mathrm{GeV}}/{L}_{0.5\mbox{--}7\mathrm{keV}}\gtrsim 11500$, which is larger than all but two pulsars (PSR J1836–5925 and PSR J2021+4026) in the Fermi second pulsar catalog (Abdo et al. 2013). The fact that PSR J1614 is an outlier in the TeV and GeV-to-X-ray flux ratios (assuming it is the counterpart to these sources), the lack of a PWN, and the large offset from the TeV source, make it an unlikely counterpart to HESS J1616.

TeV emission can be produced by a number of different source types, including blazars, SNRs, PWNe, and γ-ray binaries (see, e.g., Holder 2012). However, only PWNe and SNRs (or SNRs interacting with molecular clouds) are known to produce extended TeV emission. There are three SNRs detected at radio frequencies surrounding the center of HESS J1616, including Kes 32, RCW103, and G332.0+0.2, but their rather large offsets (∼14', 15', and 29', respectively) and difference in morphology compared to HESS J1616 make them unlikely counterparts to the TeV emission. Therefore, among the known types of TeV emitters, the only other candidates are relic PWNe, which can be dim in X-rays. The relic plerion scenario can also account for the offset between the known pulsars, within the extent of HESS J1616, and the TeV emission peak. However, until an X-ray or radio PWN with a preferential extension toward the TeV source is detected, this scenario will remain unconfirmed. Observations with the SKA may be able to detect synchrotron emission from such relic plerions.

The Fermi source, 3FGL J1616.2–5054e, is coincident with the TeV emission. This source is extended and similar in size to HESS J1616. The spectra of the GeV and TeV sources smoothly connect to each other (see Figure 6.8 in Lande 2014). The hard spectral index Γ = 1.74 listed in the 2FHL catalog (Ackermann et al. 2016) is typical of other PWNe detected by Fermi (Abdo et al. 2013). We have analyzed three Chandra observations of the field containing HESS J1616 and searched among the 56 detected X-ray sources for a possible counterpart. Of these, 30 were classified with ≥70% confidence, primarily as either AGNs or non-degenerate stars. Stars are not known to produce TeV emission, while AGNs seen at TeV energies do not produce extended emission with the size of HESS J1616, so any of these sources can be ruled out as sole counterparts to HESS J1616. Five sources were bright enough (≥100 counts) to extract and fit their spectra. Of these, only Source 18 in Field 1 may be of interest because we cannot rule out an isolated pulsar interpretation. If this source is a nearby pulsar, it could host a relic PWN that could be responsible for the TeV emission. The TeV-to-X-ray luminosity ratio is compatible with other pulsars detected at TeV energies, but there are no hints of extended X-ray emission. Better quality X-ray data and deeper MW limits for a possible counterpart of this source are necessary to classify it.

We did not detect PSR J1614 and set an upper limit on its X-ray flux (F_{0.5–8 keV} < 5 × 10⁻¹⁵ erg cm⁻² s⁻¹ at a 90% confidence level). This makes PSR J1614 one of the least efficient X-ray pulsars known, with an upper limit on its X-ray efficiency η_{0.5–8 keV} ≲ 2 × 10⁻⁵. While, the lack of an X-ray PWN does not preclude the existence of a relic TeV PWN, it does not allow us to look for any PWN asymmetry, which could lend support to the association. Lastly, the large offset between PSR J1614 and HESS J1616 of ∼22' (45 pc at d = 7 kpc) makes the association doubtful, so we consider PSR J1614 an unlikely counterpart to HESS J1616.

The only other known pulsar energetic enough to power a relic PWN in this region is PSR J1617. The offset from HESS J1616 is smaller for PSR J1617 (10', is 19 pc at d = 6.5 kpc). The faint extended X-ray PWN of PSR J1617 was detected by CXO, but it does not extend in the direction toward the TeV source center, making the association ambiguous (see K+09 for a discussion). If we assume that the PWN of PSR J1617 is responsible for the TeV emission, then the TeV-to-X-ray luminosity ratio is L_1–10TeV/L_{0.5–8 keV} ∼ 23 and the GeV-to-X-ray luminosity ratio is L_1–100GeV/L_{0.5–8 keV} ∼ 91. These ratios are consistent with other PWNe observed at TeV energies and pulsars detected by Fermi (Abdo et al. 2013; Kargaltsev et al. 2013). Therefore, PSR J1617 remains a possible counterpart of the TeV source, given that the associations of the other X-ray sources in this field appear to be less likely.

Future observations with CTA and SKA will be able to shed more light on this source's nature, in particular by testing whether it is a single extended or multiple blended source, and whether there is any spatially dependent spectral structure or other morphological connection to PSR J1617.

Support for this work was provided by the National Aeronautics and Space Administration through Chandra Awards G05-16075X and AR3-14017X issued by the Chandra X-ray Observatory Center, which is operated by the Smithsonian Astrophysical Observatory for and on behalf of the National Aeronautics Space Administration under contract NAS8-03060. J.H. would like to thank George Younes for useful discussions regarding this paper. We would like to thank the anonymous referee for a careful reading of the paper and useful comments.