Deep X-ray Observations of the Gamma-ray Bright Pulsar Wind Nebula HESS J1303-631

Introduction

Pulsar wind nebulae (PWNe) are efficient accelerators of relativistic electrons and positrons, and they are now one of the dominant identified classes of Galactic TeV sources (e.g., [7, 16]. Their multiwavelength appearance depends strongly on age and environment. Synchrotron-emitting X-ray particles cool rapidly and remain concentrated near the pulsar, whereas inverse-Compton-emitting particles can survive longer and fill much larger GeV and TeV emitting nebulae. In evolved systems, reverse-shock interaction with the host supernova remnant and pulsar proper motion can displace the relic particle population from the currently powered compact X-ray nebula (e.g., [8]).

In the crowded regions of the Galactic plane many TeV sources remain unidentified or ambiguously identified due to multiple possible (or a complete lack of) contributors of energetic particles. The known types of Galactic accelerators, in addition to pulsars with PWNe, include microquasars, high-mass gamma-ray binaries, supernova remnants, and, possibly, young stellar objects (e.g., [17]). On some occasions an AGN of a blazar type can be seen as a TeV source through the Galactic plane (e.g., [12, 2]). To further complicate matters, in many cases X-ray counterparts of these accelerators are faint or relatively faint in X-rays. Therefore, a wholistic analysis and assessment of multiple sources is often needed to make a judgment about the nature of the TeV source.

HESS J1303-631 was discovered with H.E.S.S. (Aharonian et al. [1]) as a bright, extended TeV source in the Galactic plane ($l = 304.2^\circ$, $b = -0.3^\circ$) with and integrated flux above 380 GeV of $(1.2 \pm0.2) \times10^{-11}\ \mathrm{cm}^{-2}\ \mathrm{s}^{-1}$, about 17% of the Crab Nebula flux, and a TeV photon index $\Gamma= 2.44 \pm0.05_{\mathrm{stat}} \pm0.2_{\mathrm{syst}}$. A short ($\approx5$ ks) archival Chandra ACIS-I observation did not reveal an obvious X-ray counterpart and only placed a limit on pulsar and PWN emission and diffuse X-ray emission ([20]). This made the source one of the early “dark accelerator” candidates.

The source was later studied in more detail by the H.E.S.S. Collaboration (H.E.S.S. Collaboration et al. [15]) who found that the morphology is well described by an asymmetric Gaussian with $\sigma_x = 0.194 \pm0.008^\circ$, $\sigma_y = 0.145 \pm0.006^\circ$, and position angle $147 \pm6^\circ$. The large-region TeV spectrum is not well represented by a simple power law, with a better fit found for a cutoff power-law (PL) model having $\Gamma= 1.5 \pm0.2$ and cutoff energy $E_{\mathrm{cut}} = 7.7 \pm2.2$ TeV ([15]). A young (spin-down age 11 kyrs), energetic ($\dot{E} = 1.7 \times10^{36}\ \mathrm{erg}\ \mathrm{s}^{-1}$) PSR J1301-6305 is located $\approx7'$ NW of the TeV source center. It has been proposed as a source of energetic particles powering the TeV emission. Additional support for the evolved PWN scenario comes from the energy-dependent morphology of the TeV source: lower-energy 0.84–2 TeV photons form a broad, displaced nebula, while the $> 10$ TeV emission is more compact and closer to the pulsar. The distance to the pulsar inferred from its dispersion measure (DM=374 $\mathrm{cm}^{-3}$ pc) varies greatly depending on the galactic electron density distribution model used: 10.7 kpc (YMW17; [24]), 6.6 kpc (cl02; [5]), and 4.7 kpc (OC26; [21]). We assume the latter value for the rest for the paper.

H.E.S.S. Collaboration et al. [15] also reported a faint XMM-Newton counterpart associated with PSR J1301-6305. The X-ray emission consists of a compact source near the pulsar plus a diffuse component extending roughly eastward toward the TeV nebula. The diffuse feature was detected at $\sim6.5\sigma$ in the archival XMM-Newton data and was modeled with an absorbed PL with poorly constrained $N_{\mathrm{H}} = 2.7^{+1.3}_{-1.1} \times10^{22}\ \mathrm{cm}^{-2}$, $\Gamma_{\mathrm{X}} = 2.0^{+0.6}_{-0.7}$, and unabsorbed $F_{2-10\ \mathrm{keV}} = 1.6^{+0.2}_{-0.4} \times10^{-13}\ \mathrm{erg}\ \mathrm{cm}^{-2}\ \mathrm{s}^{-1}$ ([15]). The X-ray nebula is much smaller than the TeV nebula, as expected for an evolved PWN in which the highest-energy synchrotron-emitting electrons cool rapidly. The one-zone leptonic model presented by H.E.S.S. Collaboration et al. [15] obtained a magnetic field of 1.4 $\mu$G, but the authors noted that the very different X-ray and TeV sizes require a more complex treatment.

At GeV energies, Fermi-LAT analyses detected a hard, extended counterpart 4FGL J1303.0-6312e. The 14 yr Fermi-LAT study by Zhou et al. [25] measured a 10–500 GeV photon index of $1.69 \pm0.09$ and a 2D Gaussian extension $R_{68} = 0.32 \pm0.04^\circ$, strengthening the evolved-PWN interpretation.

Dedicated ATCA observations did not detect an extended radio PWN at 5.5 or 7.5 GHz, and archival 1.384 and 2.368 GHz data also did not show a significant radio counterpart to the X-ray nebula ((Suschu et al. 2015)). However, ATCA data revealed a shell-like radio structure (likely a supernova remnant) 15′ east of the pulsar, but its association with PSR J1301-6305 is questionable because it would require a very large pulsar velocity, 1,800 km $\mathrm{s}^{-1}$ for the smallest $d = 4.7$ kpc and pulsar age of 11 kyrs. Even if the true pulsar’s age is a factor of two larger, the velocity would still be high enough to expect a formation of a strong bowshock around a high-Mach pulsar.

In order to better resolve the compact PWN from the pulsar additional Chandra observations (PI J. Eagle; ObsIDs 28056, 28509, 28510, 29079, 29080, 29192, 29204, 29205, 29206) were carried out. It was accompanied by an XMM-Newton observation (PI J. Eagle; ObsID 0931990101) to collect more photons for spectral analysis. In § 2 we provide details of these observations followed by the data analysis results in § 3. We discuss the results and provide multiwavelength context (including new imaging from MeerKAT GPS; Goedhart et al. [10]) in § 4. We conclude with a summary in § 5.

Observations

Ten recent Chandra ACIS observations of PSR J1301-6305 region are summarized in Table 1. Nine of these were taken between Nov 2023 and Jan 2024. The target was imaged on ACIS S3 detector near the aim-point close to optical axis of the telescope. One observation was taken in 2004 (ObsID: 6098), and is not included in the analysis because of the short exposure time (4.7 ks) and PSR J1301-6305 being imaged $\sim9'$ off-axis. It is possible that this exposure could have collected a comparable number of photons as the recent observations since the effective area at lower energies was much larger in the past. However, there are very few counts at the location of the pulsar, suggesting a large $N_{\mathrm{H}}$ or very weak/absent thermal component. XMM-Newton EPIC observed the HESS J1303-631 region on 3 occasions (see Table 1). The deepest observation aimed at PSR J1301-6305 was carried out on August 31, 2024. Table 1 summarizes X-ray observations of the HESS J1303-631 field.

ObsID	Instrument	Exposure (ks)	Observation Date	Notes
Chandra ACIS
6098	ACIS-I	4.65	2004 Sep 25	Not included in merged images or spectral analysis
28056	ACIS-S	9.94	2024 Jan 16
28509	ACIS-S	10.93	2023 Nov 26
28510	ACIS-S	10.12	2024 Jan 08
29079	ACIS-S	10.93	2024 Jan 12
29080	ACIS-S	15.87	2024 Dec 02
29192	ACIS-S	11.62	2024 Jan 12
29204	ACIS-S	11.52	2024 Jan 16
29205	ACIS-S	11.80	2024 Jan 16
29206	ACIS-S	10.64	2024 Jan 18
Total (ACIS)	$\cdots$	108.02	$\cdots$	$\cdots$
XMM-Newton EPIC (M1/M2/PN)
0303440101	EPIC	$\sim$30/30/28	2005 Jul 12	Archival; affected by chip-gap placement for the PWN region
0302340101	EPIC	$\sim$31/30/26	2005 Jul 14	Archival; significantly offset
0931990101	EPIC	$\sim$70/77/48	2024 Aug 31	New; exposure limited by significant flaring
Total (EPIC)	$\cdots$	131/137/102	$\cdots$	$\cdots$

Data Reduction and Analysis

Chandra imaging and spectroscopy of PSR J1301-6305 and its PWN.

The merged ACIS image of PSR J1301–6305 (hereafter J1301) is shown in Figure 1. On small scales (upper panel in the figure) it appears that the compact nebula (polygon region) is actually composed of multiple point sources. There does appear to be excess emission above the background surrounding the pulsar, indicating a possible compact nebula extending roughly to the east of the pulsar, but the exact nature of the morphology is unclear given its faint X-ray flux, $F_{0.5-7} = (4.47 \pm0.92) \times10^{-14}\ \mathrm{ergs}\ \mathrm{s}^{-1}\ \mathrm{cm}^{-2}$. To quantify the extension of this emission we compared the radial profile of the CXO data to that of a MARX point-source simulation (see Figure 2). We see that there is significant extended emission beyond the PSF ($\gtrsim2^{\prime\prime}$). Faint, diffuse emission is seen on larger scales (circular region, $r=20^{\prime\prime}$) extending to the NE of the pulsar. Unfortunately, even though we have a large baseline between the observations ($\sim20$ years) there are too few counts in the 2004 exposure to constrain the proper motion.

We performed a spectral analysis of J1301 and the surrounding extended emission. The CXO observations included in the fit are summarized in Table 1. The older observation (ObsID: 6098) is not included due to the short exposure time and large off-axis angle. Therefore, astrometric correction is unnecessary. The spectra were fit in XSPEC (v.4.15.0; Freeman et al. [6]) using their implementation of the Cash statistic (cstat).

The model fit to the pulsar and nebula was XSPEC’s absorbed power-law (PL) where the interstellar extinction is modeled with the tbabs model (absorption cross sections and abundances described in Wilms et al. [23]). The spectra from all CXO observations were jointly fit in the 2–10 keV energy range. The column density was frozen to $n_{\mathrm{H}} = 2.7 \times10^{22}\ \mathrm{cm}^{-2}$, obtained by H. E. S. S. Collaboration et al. [15] from a fit to the PWN region using the archival XMM Newton observations. The regions used for spectral extraction are shown in Figure 1, and the results are summarized below and in Table 2.

Region	$N_{\mathrm{H}}$	$\Gamma$	Norm	C-stat/dof	$F_{2-10\ \mathrm{keV}}$	$L_{2-10\ \mathrm{keV}}$
PSR	2.7 (fixed)	2.02 ± 1.04	$(6.15 \pm5.36) \times10^{-6}$	17.97 / 22	$(1.30^{+0.20}_{-0.98}) \times10^{-14}$	$3.44 \times10^{31}$
Compact PWN	2.7 (fixed)	2.98 ± 0.53	$(3.02 \pm1.98) \times10^{-5}$	108.43 / 109	$(1.57^{+0.14}_{-0.71}) \times10^{-14}$	$4.15 \times10^{31}$
Extended PWN	2.7 (fixed)	1.20 ± 1.25	$(2.03 \pm3.43) \times10^{-6}$	372.65 / 356	$(1.65^{+0.12}_{-1.65}) \times10^{-14}$	$4.36 \times10^{31}$

**Note—**All Chandra observations were taken in Timed Exposure (TE) mode with VFAINT telemetry. ObsID 6098 was imaged $\sim9'$ off-axis and was not included in the merged images or spectral analysis; the 108.02 ks ACIS total above includes it, while the images and spectra analyzed in this paper use only the nine 2023–2024 pointings (103.37 ks). XMM-Newton exposures are effective, background-flare-filtered exposure times for MOS1, MOS2, and pn, respectively, in ks.

Note— Spectra were fit with an absorbed power-law model using tbabs. Column densities are in units of $10^{22}\ \mathrm{cm}^{-2}$. Fluxes are unabsorbed and reported in units of $\mathrm{erg\ s^{-1}\ cm^{-2}}$. Luminosities are unabsorbed and assume a distance of 4.7 kpc. Uncertainties are quoted at the $1\sigma$ level.

The pulsar region has $30.00 \pm5.66$ net counts after background subtraction, and the fit yields a PL photon index of $\Gamma= 2.02 \pm1.04$, which is within the typical range seen for pulsars powering a PWN. The corresponding unabsorbed flux is $F_{2-10} = (1.30^{+0.20}_{-0.98}) \times10^{-14}\ \mathrm{ergs\ s^{-1}\ cm^{-2}}$ and the unabsorbed luminosity is $3.44 \times10^{31}\ \mathrm{erg\ s^{-1}}$ for a distance of 4.7 kpc.

The compact nebula with point sources excised has $67.77 \pm17.09$ net counts and the spectrum is softer with $\Gamma= 2.98 \pm0.53$, which is somewhat softer than typical values seen for PWN. However, this could be simply due to the large uncertainty in the measurement.

The unabsorbed flux in 2–10 keV range is $F_{2-10} = (1.57^{+0.14}_{-0.71}) \times10^{-14}\ \mathrm{ergs\ s^{-1}\ cm^{-2}}$ and an unabsorbed luminosity of $4.15 \times10^{31}\ \mathrm{erg\ s^{-1}}$.

Finally, the extended nebula with the compact nebula excised is much fainter with only $14.68 \pm32.38$ counts after background subtraction and the spectrum yields $\Gamma= 1.20 \pm1.25$, an unabsorbed flux of $F_{2-10} = (1.65^{+0.12}_{-1.65}) \times10^{-14}\ \mathrm{ergs\ s^{-1}\ cm^{-2}}$, and corresponding unabsorbed luminosity of $4.36 \times10^{31}\ \mathrm{erg\ s^{-1}}$ at $d = 4.7$ kpc.

XMM-Newton imaging and spectroscopy of PSR J1301-6305 and its PWN.

We used the images² script to reduce the XMM-Newton data and make a exposure corrected image of the field. The data were processed with SAS version 1.3 using the standard procedures. MOS1 and 2 events were selected with PATTERN $\le12$ and pn events with PATTERN $\le4$, We removed background particle flares for the PN and MOS1/2 using a count rate cut of 0.4 cts s$^{-1}$ and 0.35 cts s$^{-1}$, respectively. The images were created in the 0.5-10 keV band for each of the three ObsIDs individually and then merged using the emosiac SAS task. Figure 3 shows the merged EPIC images of the field.

XMM-Newton’s angular resolution does not resolve the pulsar from a nearby source of nearly equal (in CXO images) strengths rendering EPIC data useless for the pulsar spectral analysis. On the other hand, the EPIC background is higher than that of ACIS precluding the detection of fainter extended PWN emission and severely hampering spectral quality for the region where extended PWN emission is seen in the CXO images (see polygon region in bottom panel of Figure 1). We attempted to extract EPIC spectra from a comparable in size circular region (after excluding the pulsar and a nearby point source). We only obtained a crude measurement of $\Gamma= 1.6^{+0.8}_{-0.7}$ with $n_{\mathrm{H}}$ being poorly constrained and dependent on the choice of background region with best fit values varying between 0.9

and $1.9 \times10^{22}$ cm$^{-2}$ and large uncertainties. These values are compatible, albeit less accurate, with Chandra measurements (see Table 2).

Transient Source CXOU J130131.0-625919

Thanks to multiple CXO observations and also XMM-Newton observations of this field we discovered a flare from a transient X-ray source that appears only in Chandra ObsID 29080 (15.87 ks exposure). To determine the source position the image was binned by a factor of four and the CIAO tool wavdetect was used. The center of the source was found to be (RA, Decl.) = ($13^\mathrm{h}01^\mathrm{m}31.08^\mathrm{s} \pm0.03^\mathrm{s}$, $-62^\mathrm{h}59^\mathrm{m}19.7^\mathrm{s} \pm0.2^\mathrm{s}$). These uncertainties do not include the astrometric error from Chandra’s pointing uncertainty which is $\sim1.17^{\prime\prime}$ (90% confidence).³ The total number of detected source counts is 225 (99% of which are due to the source). The source is imaged $\sim6.5^\prime$ off-axis on the ACIS-S6 chip (see Figure 4). Thanks to its off-axis placement the pile-up is negligible even at the peak of the flare (see Figure 5). The lightcurve shown in the top panel was extracted following the CIAO thread⁴. The bottom panel, created using the ciao task dmcopy and binning events from the source region into energy bins of width 500 eV from 500 to 7000 eV and time bins of 1000 s, shows energy-resolved lightcurves as a heatmap. The spectrum of this source was extracted from the region shown in Figure 4. XSPEC’s implementation of the cstat statistic was used for the fit and counts were grouped to have at least one count per bin prior to fitting. The energy was restricted to 0.5–7 keV. Fitting with a PL model modified by absorption (tbabs) resulted in $N_{\mathrm{H}}=(0.6\pm0.5)\times10^{22}$ cm$^{-2}$, $\Gamma=1.4\pm0.3$, and $\mathcal{N}=(5.0\pm2.0)\times10^{-5}$ photons keV$^{-1}$ cm$^{-2}$ s$^{-1}$ at 1 keV with cstat/dof=114.69/167. The corresponding unabsorbed flux is $F_{0.5-7}=(3.10^{+0.09}_{-0.51})\times10^{-13}$ ergs cm$^{-2}$ s$^{-1}$, and, assuming the same spectral shape throughout, the peak flux was $F_{0.5-7}=(9.60^{+0.28}_{-1.58})\times10^{-13}$ ergs cm$^{-2}$ s$^{-1}$.

We also attempted to fit this source with an absorbed mekal model. Although the fit quality is acceptable (cstat/dof = 113.97/167) the resulting temperature can be arbitrarily high with a $3\sigma$ lower limit of $\approx8$ keV and best-fit value of $\approx23$ keV. The best-fit $N_{\mathrm{H}}=(0.6\pm0.3)\times10^{22}$ cm$^{-2}$ while the best-fit normalization corresponds to an emission measure of $\approx3\times10^{55}$ cm$^{-3}$ at fiducial distance $d=3$ kpc.

This X-ray source has a plausible NIR counterpart 2MASS 13013105-6259198 located only 0.27″ away (see Figure 7). We retrieved all available optical-IR photometry from Vizier and fitted it with a set of PARSEC 1.2S + COLIBRI stellar evolution models⁵ with solar-like abundances using spectra from Castelli–Kurucz atmosphere models ([4]). The best fit corresponds to a pre-main sequence star ($\lesssim1$ Myrs) with $A_V = 5.1 \pm0.5$, $d = 2 - 3.7$, $\log g = 3.7$, and $T \approx5{,}000$ K. The fit, however, shows systematic residuals with data exceeding the fit at $\lesssim600$ nm and in the reddest Spitzer IRAC band. The best-fit $A_V$ corresponds to $d = 4.9 - 5.4$ kpc from DECaPS 3D dust map ([26]), which is noticeably higher than that inferred from the stellar model fit. We also inspected IRAC images which show diffuse emission likely associated with star-forming activity mostly concentrated north of the Chandra transient (see Figure 7). We note that DECaPS 3D resolution is $1'$ while pockets of star-formation seen in IRAC images appear to be much more compact than that. This may explain discrepancy between the two distance estimates. Therefore, we below will use $d = 3$ kpc. We also note that the best-fit $N_H$ implies (following relation of Güver & Özel [11]) a factor of 2 lower $A_V$ than that inferred from the optical-NIR SED fitting.

Discussion

An evolved, synchrotron-underluminous PWN scenario for HESS J1303-631 has been previously suggested by (H.E.S.S. Collaboration et al. [15]). The small ($\lesssim1.5'$) and faint X-ray nebula near J1301 shows some extension to the east (toward the center of HESS J1303-631; see Figure 3), supporting the association. However, this factor alone, can hardly be taken as decisive evidence of an association with HESS J1303-631. The stronger support for the association comes from the hard, extended GeV counterpart (Zhou et al. [25]) and the energy-dependent TeV morphology (H.E.S.S. Collaboration et al. [15]) which shows spectral hardening toward PSR J1301-6305. The compact X-ray PWN radiative efficiency, $L_X/\dot{E} \approx10^{-4}$, at $d = 4.7$ kpc, is comparable to those of many other pulsars with similar $\dot{E}$. On the other hand, the TeV luminosity is relatively high compared to other PWNe detected in TeV (see Figure 6). The shape of the extended X-ray PWN component is surprisingly anisotropic and could be interpreted as a tail behind the pulsar moving in the western direction. However, the PWN emission is far too faint to say this with certainty. This example demonstrates the difficulty in understanding the nature of TeV sources and identifying their X-ray counterparts, particularly in those cases when the existence of a young radio pulsar have not yet been established through the detection of pulsations. Deep radio observations from modern surveys sensitive to extended structures (such as MeerKAT GPS) often show some extended emission, which, however, may not correlate well with TeV or X-ray emission. This seems to be the case for PSR J1301-6305 and HESS J1303-631 (see Figure 3). Although the extended radio emission is clearly seen around PSR J1301-6305 on $\sim$arcminute scales, it mostly extends west of the pulsar. It also shows a complex structure which could be attributed to the interaction with the reverse shock of the host SNR. However, unlike Vela X where both radio and TeV emission appear to trace the relic PWN displaced by the reverse shock, the TeV emission of HESS J1303-631 is displaced in the opposite direction from the radio emission. It may be that the arcmin-scale radio emission is not related to PWN. Polarization and spectral measurements (which is beyond the scope of this study) can help to establish its origin. There also appears to be a relatively bright, unresolved source imaged in MeerKAT GPS coincident with PSR J1301-6305. Alternatively, it could be that enough time has passed for the most energetic (TeV emitting) electrons to diffuse out to much larger distances than the size of the PWN swept and compressed by the reverse shock.

Other studies have faced similar challenges. For example, Kargaltsev et al. [18] and Hare et al. [14] analyzed a deep 60 ks Chandra observation of the young pulsar PSR J1617-5055, offset by about 10′ from the TeV source HESS J1616-508, to explore a possible association. Faint, diffuse emission surrounding the pulsar was resolved by Chandra with no preferential extension towards the TeV source center. It corresponded to a low radiative efficiency $L_X/\dot{E} \sim10^{-4}$. The authors could not confirm the association from the X-ray data alone, and archival radio/IR imaging indicated that the TeV emission was likely also powered, at least in part, by an unrelated SNR or star-forming region.

The transient source’s most likely nature appears to be a pre-main sequence star (PMS) flare, given its location near/within the compact region of star-forming activity seen in IRAC images and the young age inferred from optical-IR SED fitting. The duration of the flare is compatible with those of PMS stars which typically follow the fast-rise-slow-decay profile (see Figure 5). However, the X-ray luminosity of $L_X = 3.34 \times10^{32}$ erg s$^{-1}$

($L_{X,\mathrm{peak}} = 10.3 \times10^{32}$ erg s$^{-1}$), in 0.5-7 keV for a fiducial distance of 3 kpc and the high temperature $\geq8$ keV, are rather extreme even for a young star (cf. Figure 9 of Getman et al. [9]). The inferred emission measure is compatible with those PMS star flares but too large for main sequence star flares. An alternative interpretation could be an active binary (e.g., RS CVn) with young stars. It could also be another type of binary with a compact object, if the 2MASS source is a spurious association. The large offset between the TeV source and the transient makes it unlikely to be related to the TeV source under any of the above scenarios.

Summary

1. We present new, deep joint Chandra and XMM-Newton observations of the X-ray counterpart to the PWN HESS J1303−631, likely powered by PSR J1301−6305. The Chandra ACIS-S imaging,

totaling 108 ks, resolves the field into a compact pulsar source, a faint compact nebula, and an extended diffuse component, successfully separating these from a nearby, comparably bright confusing source that remains blended even in the new, deep XMM-Newton data – underscoring the need for Chandra’s angular resolution in this crowded field.

2. Joint spectral fitting of the Chandra data with an absorbed power-law model yields a pulsar photon index of $\Gamma= 2.02 \pm1.04$ and an unabsorbed 2–10 keV flux of $F_{2-10} = (1.30^{+0.20}_{-0.98}) \times10^{-14}\ \mathrm{erg\ s^{-1}\ cm^{-2}}$, consistent with typical values for pulsars powering PWNe. The compact nebula spectrum is somewhat softer ($\Gamma= 2.98 \pm0.53$), while the extended nebula component, with the compact nebula excised, is harder again ($\Gamma= 1.20 \pm1.25$); all three regions have comparable luminosities of a few $\times10^{31}\ \mathrm{erg\ s^{-1}}$ at $d = 4.7\ \mathrm{kpc}$.

3. Combined with the previously known hard, extended GeV counterpart (4FGL J1303.0–6312e), the energy-dependent TeV morphology – a broader, displaced nebula at 0.84–2 TeV that becomes more compact and closer to the pulsar above 10 TeV – and the absence of a bright radio PWN, these X-ray results reinforce the picture of HESS J1303–631 as an evolved, synchrotron-underluminous PWN whose X-ray nebula is far more compact than its TeV nebula, as expected from the rapid cooling of the highest-energy synchrotron-emitting particles.

4. Deep ATCA and SARAO MeerKAT GPS radio imaging reveal a shell-like SNR candidate 15′ east of the pulsar and arcminute-scale diffuse structure extending mostly west of it, but neither correlates well with the compact X-ray/TeV nebula. Unlike Vela X, where the radio and TeV emission trace a common relic nebula displaced by the reverse shock, the radio and TeV structures here are offset in opposite directions from the pulsar, suggesting the extended radio emission may be unrelated to the PWN.

5. We serendipitously discovered a transient X-ray source, CXOU J130131.0–625919, in one of nine new Chandra pointings. Its hard, fading spectrum ($\Gamma= 1.4 \pm0.3$) and luminosity of $\approx3 \times10^{32}\ \mathrm{erg\ s^{-1}}$ at a fiducial distance of 3 kpc, together with a plausible young stellar near-infrared counterpart, most closely match a pre-main-sequence star flare, though an active binary or a compact-object binary cannot be excluded. Its large angular offset from HESS J1303–631 rules out a physical association with the TeV source.

6. The anisotropic, tail-like shape of the extended X-ray nebula and its possible relation to the host SNR’s reverse shock, along with the mismatch between the X-ray/TeV and radio morphologies, remain open questions. Deeper radio polarization and spectral-index mapping, together with a search for X-ray pulsations, would help establish the physical relationship between these components.

We thank Kaleb Martin for his early exploration of this data during a NASA summer internship.

Support for this work was provided by the National Aeronautics and Space Administration through Chandra Award Number GO4-25039A issued by the Chandra X-ray Center, which is operated by the Smithsonian Astrophysical Observatory under contract NAS8-03060. J.H. acknowledges support from NASA under award number 80GSFC24M0006.

This research employs a list of Chandra datasets, obtained by the Chandra X-ray Observatory, contained in DOI: 10.25574.