DISCOVERY OF AN ENERGETIC PULSAR ASSOCIATED WITH SNR G76.9+1.0

The past decade has been witness to a highly productive back-and-forth relationship between X-ray and radio studies of supernova remnants (SNRs). Their nonthermal radio emissions, serving as tracers of particle acceleration, have motivated follow-up X-ray observations, most notably with the Chandra telescope. In many instances, X-ray imaging has revealed point-like sources surrounded by extended synchrotron structures, typically axisymmetric; these pulsar wind nebulae (PWNe) arise from the shocked outflows of relativistic particle winds driven by a central engine, the powerful time-varying fields in a neutron star magnetosphere. Targeted radio periodicity searches of PWNe have then uncovered previously unseen pulsations, providing the spin periods and, eventually, estimated ages, magnetic field strengths, and spin-down luminosities of the neutron stars, without which a meaningful physical understanding of the surrounding phenomena would be beyond reach. A few specific examples of this general picture of radio/X-ray symbiosis have been the discoveries of the pulsars in the remnants 3C58, G54.1+0.3, G292.0+1.8, and in the "Mouse" (see, e.g., Camilo 2004).

The Crab Nebula has been viewed for decades as a prototype, but there is growing evidence of PWNe with substantially different properties. At least three objects, among them the Vela SNR, have nonthermal radio spectra that are steep in comparison to the Crab (i.e., with spectral indices α ≈ 0.6, where flux S_ν∝ν^−α at frequency ν) and radio morphologies characterized by a pair of emission lobes straddling a central depression. While the nature of one such candidate PWN, DA 495, was once in question, X-ray imaging observations led to the discovery of an X-ray nebula and a presumed neutron star, both residing near the radio "hole" (Arzoumanian et al. 2008, hereafter, ASL+08). Radio and X-ray pulsation searches have not, however, detected a pulsar signature, so a key ingredient is lacking from our understanding of DA 495. In this paper, we describe a similar series of observations directed toward understanding the third member of this group of unusual SNRs, G76.9+1.0.

SNR G76.9+1.0 was discovered in a 408 MHz Dominion Radio Astrophysical Observatory survey of the Cygnus X region and initially characterized as a "possible galactic object" (Wendker et al. 1991). Follow-up three-band radio observations using the Very Large Array (VLA) by Landecker et al. (1993) resolved a two-lobed structure 4' in size, with a connecting bridge of emission, embedded in a faint, roughly circular emitting region 9' × 12' across. Based on its steep, nonthermal spectrum, polarization, and morphological similarities to center-filled SNRs (in important aspects such as the lack of an outer boundary and radially decreasing flux away from the central depression), these authors favored an SNR interpretation. In this picture, the radio lobes trace the Crab-like synchrotron emission of an evolved pulsar wind nebula.

Here we present imaging, spectral, and timing studies of G76.9+1.0 in X-rays and radio, culminating in the discovery of PSR J2022+3842 and of its associated X-ray PWN. In Section 2, we describe our analysis of a Chandra observation that yielded the first identification of the SNR's central engine, a neutron star with its surrounding nonthermal nebula. In Section 3, we describe the discovery of the radio pulsar J2022+3842 and follow-up radio timing observations with the Green Bank Telescope (GBT) which have provided its spin-down rate and radio pulse properties. In Section 4, we present the results of a deep Rossi X-ray Timing Explorer (RXTE) observation: we show that the pulsed X-ray flux is consistent with the Chandra point-source flux, cementing the pulsar's association with the SNR.

For reasons developed below, we adopt a distance to G76.9+1.0 of 10 kpc throughout this work.

SNR G76.9+1.0 was observed for 54 ks by the Chandra X-ray Observatory on 2005 August 1 UT using the Advanced CCD Imaging Spectrometer (ACIS; Burke et al. 1997) operating in the full-frame TIMED/VFAINT exposure mode (ObsID 5586). This detector is sensitive to X-rays in the 0.3–12 keV energy range with a resolution of ΔE/E ∼ 0.06 FWHM at 1 keV. The CCD pixel scale is 05, comparable to the telescope's on-axis spatial resolution. Low-level data were reprocessed with the latest calibrations using the CIAO script chandra_repro. The data analysis made use of the CIAO (v4.3), FTOOLS (v6.10), CALDB (v4.4.2), and XSPEC (v12.6.0q) software packages. A total of 53.7 ks of live-time was accumulated. No additional time filtering was necessary as the background rate was stable over the course of the observation. We followed the CIAO online science threads to create ACIS spectra and images. To allow for imaging with the best available angular resolution, we used the EDSER reprocessing option, applying an energy-dependent sub-pixel event-repositioning algorithm to each photon (Li et al. 2004).

The left-hand panel of Figure 1 presents the exposure-corrected image, in the 0.5–7 keV band, acquired by the ACIS-S3 and S2 CCDs, which together covered the entire extent of the radio SNR. No morphological evidence of G76.9+1.0 itself is apparent nor is there any hint of resolved diffuse emission in images created in the soft (<2 keV) or hard (>2 keV) energy bands; instead, several faint point-like sources are detected. The brightest of these, enclosed within a box in the figure, is located close to the optical axis and contains N = 1190 photons. This source is not associated with any known stellar counterpart or object in the Two Micron All Sky Survey (2MASS) image of the field. Its identification as a possible neutron star associated with G76.9+1.0 is reinforced by the presence of faint but distinct X-ray nebulosity surrounding a brighter, unresolved core. To highlight the diffuse emission, we show as insets in Figure 1 both the raw and smoothed 0.5–7 keV count-rate maps for the boxed 1' × 1' region centered on the source. The smoothed image was produced using a circular convolution kernel of varying size, requiring a minimum of five counts within the kernel diameter. Although the putative nebula is faint, there is a clear excess of emission elongated on a line with position angle ≃ 26° north through east. This and the relative faintness of the point source argue against identification of the diffuse emission as a dust-scattering halo; these are normally circular and found around much brighter sources.

Zoom In Zoom Out Reset image size

The 1.49 GHz VLA radio image shown in the right-hand panel of Figure 1 places the X-ray nebula in the context of the double-lobed radio structure. The unambiguous alignment of the long axis of the diffuse X-ray emission with the radio ridge connecting the lobes suggests that the radiation processes underlying the emissions in both bands are related, such as in the canonical picture of synchrotron emission from relativistic particles originating in the pulsar's magnetosphere.

To refine the location of the putative pulsar, the X-ray image was registered using the coordinates of eight 2MASS infrared counterparts obtained from the NASA/IPAC Infrared Science Archive. The updated centroid is located at (J2000.0) R.A. = $20^{\rm h}22^{\rm m}21\mbox{$.\!\!^{\mathrm s}$}689$, decl. = +38°42'1482 with a 1σ uncertainty of (009, 007) in the two coordinates, respectively. The required shift in R.A. and decl. was (−0012, −0162). This source, CXOU J202221.68+384214.8, is hereafter referred to by the shortened name CXOU J202221.

We extracted a spectrum for CXOU J202221, and generated an appropriate response matrix, with the specextract CIAO script, using a circular 25 radius aperture (representing >90% encircled energy, depending on photon energy) centered on the source. The diffuse emission and cosmic and detector backgrounds contribute negligibly (∼10 counts) in this region. With a maximum count rate in a pixel of <4 × 10⁻³ s⁻¹, photon pile-up could safely be ignored. The 0.5–10 keV spectrum was grouped with a minimum of 15 counts per spectral channel and fitted using XSPEC to an absorbed power-law model, a common property of young, rotation-powered pulsars. (Blackbody models are formally allowed, but produce unnaturally high temperatures.) An excellent fit, with χ² = 69.5 for 70 degrees of freedom (dof) was obtained for an absorbing column N_H = (1.6 ± 0.3) × 10²² cm⁻² and photon index Γ = 1.0 ± 0.2 (90% confidence intervals are used throughout). The absorbed 2–10 keV source flux is F_PSR = 5.3 × 10⁻¹³ erg cm⁻² s⁻¹, implying an isotropic luminosity of L_X = 7.0 × 10³³ erg s⁻¹ at 10 kpc distance. As described in Section 4.2, this Chandra spectrum was also used in joint fits with RXTE data.

To investigate the spectrum of the nebula, we used an elliptical extraction region (Figure 1), with semimajor and semiminor axes of length 83 × 51, centered on the pulsar, with the point-source region (r < 25) excluded. For the local background, we extracted photons from a concentric annular region well outside the nebular extent (20'' < r < 40''). We also performed a MARX simulation to determine the small (2.6%) flux contribution within the nebular region from point-source photons scattered into the point-spread function wings; this simulation took into account the measured pulsar spectrum. The tally of background-subtracted counts from the nebula is then N = 76 ± 13. The faintness of the nebula does not allow a well-constrained spectral fit; instead we fixed the column density to the best-fit value obtained for the unresolved source and assumed a nominal absorbed power-law model with Γ = 1.4, a value derived from the empirical trend relating the spectral indices of energetic pulsars and their nebulae (Gotthelf 2004). The absorbed 2–10 keV flux for the putative PWN is F_PWN ≈ 4 × 10⁻¹⁴ erg cm⁻² s⁻¹. The flux ratio F_PWN/F_PSR ≈ 0.08 is well below that expected for similarly energetic pulsars (Gotthelf 2004). The uncertainty in this result is dominated by the small number of nebula photons and not the choice of Γ.

Motivated by the Chandra discovery of the point source and possible PWN at the heart of SNR G76.9+1.0, we searched for radio pulsations with the GBT. The Spectrometer SPIGOT data-acquisition back end (Kaplan et al. 2005) was used for 5 hr on each of two days, with 600 MHz of useful bandwidth centered at 1.95 GHz. The instrument provided 768 channels with 16 bit sampling of summed polarizations at 81.92 μs intervals. The PRESTO software (Ransom et al. 2002) was used to search for pulsations, and a candidate 24 ms periodicity was detected on both days with formal significances of 7σ on MJD 54397 and 12σ on MJD 54400. Confirmation observations were carried out on 2009 May 6 (MJD 54957) using the GUPPI data-acquisition system configured similarly to the discovery instrumentation, but with the addition of full-Stokes sensitivity. The difference in the pulse periods derived at these epochs gave the first indication of the pulsar's spin-down rate and thence its canonical age, magnetic field strength, and spin-down power (see below). A search for single "giant" pulses yielded no compelling candidates, despite the fact that the pulsar's estimated magnetic field strength at the light cylinder, suggested to be relevant in the production of giant pulses (Cognard et al. 1996), is approximately 70% that of both the Crab pulsar and PSR B1937+21, and comparable to that of PSR B1821−24, all of which exhibit giant pulses.

Additional radio observations, also with the GUPPI system, were carried out at roughly three-month intervals to constrain the pulsar's long-term timing behavior. Notably, the pulse periods measured during and after the confirmation observation differed substantially from those detected during the two discovery observations made 1.5 years earlier, implying that a spin "glitch" of magnitude ΔP/P ≃ 1.9 × 10⁻⁶ had occurred at an unknown epoch in this interval. The spacing of the radio timing observations together with apparent "timing noise" instability in the pulsar's rotation on similar timescales precluded derivation of a phase-connected timing solution. We therefore determined the long-term average spin-down rate, 4.3192(27) × 10⁻¹⁴, through a least-squares fit to the multi-epoch period measurements (Figure 2), a result consistent with the short-term X-ray-derived ephemeris shown in Table 1 (see Section 4.1).

Zoom In Zoom Out Reset image size

Table 1. Properties of PSR J2022+3842

Parameter	Value
X-ray Measurements
R.A. (J2000)a......................	$20^{\rm h}22^{\rm m}21\mbox{$.\!\!^{\mathrm s}$}689(6)$
Decl. (J2000)a.....................	+38°42'1482(7)
Epoch (MJD TDB)b.................	55227.00000027
Period, P (ms)b.....................	24.2877561082(84)
Period derivative, $\dot{P}$b................	4.3064(93) × 10−14
Range of timing solution (MJD)b.......	55223–55231
Radio Measurements
Long-term average $\dot{P}$................	4.3192(27) × 10−14
Range of $\dot{P}$ fit (MJD)................	54957–55469
Dispersion measure, DM (pc cm−3).....	429.1 ± 0.5
Scatter-broadening timescale, τscatt1 GHz (ms)	55(7)
Flux density, S2 GHz (μJy).............	60
Derived Parameters
Characteristic age, τc (kyr)............	8.9
Spin-down luminosity, $\dot{E}$ (erg s−1)......	1.19 × 1038
Surface dipole magnetic field, Bs (G)....	1.03 × 1012

Notes. 1σ uncertainties given in parentheses. ^aChandra ACIS-S3 position registered using 2MASS objects (see the text). ^bRXTE phase-connected ephemeris.

Download table as:  ASCIITypeset image

The pulsar's flux density at 2 GHz, S_2 GHz, is 60 μJy; its 4.8 GHz flux density of ≈45 μJy suggests that PSR J2022+3842 has an unusually flat spectrum, α ≃ −0.33. The implied radio pseudo-luminosity S_1.4 GHzd² ≃ 7 D²₁₀ mJy kpc² is unremarkable, low in comparison to the majority of known pulsars, but an order of magnitude higher than the luminosities of other young, faint pulsars discovered in deep searches of SNRs (Camilo 2004). Even with a long cumulative exposure, our best average pulse profile (Figure 2) retains significant statistical noise; despite this faintness, a number of relevant pulse shape and polarization results are available.

The radio pulsations of PSR J2022+3842 are significantly affected by dispersion and scattering due to propagation in the interstellar medium, which are typically important at frequencies ≲ 1 GHz, even in our 2 GHz observations. To estimate dispersion measure (DM) and the pulse-broadening timescale τ^scatt, we formed a pulse shape model assuming a Gaussian-shaped intrinsic profile, a one-sided exponential-decay impulse response function for the scatter broadening, and a ν⁻⁴ frequency dependence for the width of the exponential. We fit the S-band (i.e., 2 GHz) data to this model in four evenly spaced frequency sub-bands, allowing the width of the intrinsic Gaussian profile to vary and also accounting for a DM bias caused by the frequency-dependent pulse broadening due to scattering. This simple model fit the data well, but we caution that the assumptions inherent in the model may introduce significant systematic errors, perhaps as large as twice the statistical uncertainties quoted below.

The NE2001 Galactic electron distribution model (Cordes & Lazio 2002) fails to accommodate the derived DM = 429.1  ±  0.5 pc cm⁻³ in the direction of PSR J2022+3842—formally, the model suggests a lower bound on the distance to the pulsar of 50 kpc. Instead, a likely overdensity of free electrons in the Cygnus region, along the line of sight, accounts for the higher-than-expected dispersion—a similar explanation has been advanced for the unexpectedly large DM of the nearby PSR J2021+3651 (Roberts et al. 2002).

In this direction, the Perseus Arm lies at a distance of approximately 6 kpc, and the Outer Arm at ≳ 10 kpc. For convenience in scaling distance-dependent parameters and because the larger distance brings X-ray efficiencies more approximately in line with those of other young pulsars (see Section 5), we adopt a distance d = 10 D₁₀ kpc. Independent supporting evidence for the large distance is derived from H i absorption and X-ray studies of the PWN CTB 87 (Kothes et al. 2003), roughly 2° from G76.9+1.0 on the sky. At a distance of 6.1 kpc, CTB 87 lies in the Perseus arm and its X-ray-derived absorbing column is N_H = (1.4 ± 0.2) × 10²² cm⁻² at 90% confidence (Safi-Harb et al. 2011); although their uncertainties are large, the nominal H i column for G76.9+1.0 is somewhat larger than that for CTB 87. A radio-bright extragalactic point source adjacent to CTB 87 exhibits a deep absorption component at a velocity of about −85 km s⁻¹ not seen in the absorption profile of CTB 87. This feature could account for the larger foreground absorption toward G76.9+1.0 if the latter is located at a significantly larger distance, such as the Outer Arm or beyond it.

The NE2001 model also predicts a timescale for pulse broadening due to interstellar scattering of 0.3 ms at 1 GHz; instead, we find τ^scatt_1 GHz = 55 ± 7 ms, which scales to 3.8 ms at the center of our observing band but varies strongly across it. The long decay of the pulse, here averaged over the 700 MHz band, can be seen in the upper half of Figure 2. Consistent with this result, an exploratory observation centered at 800 MHz sky frequency failed to detect the pulsar; similar attempts at 1.4 GHz and 5 GHz yielded weak detections. At 5 GHz, the pulse was unexpectedly broad, consistent with the results of the frequency-subband fit to the pulse shape: in addition to the importance of scatter broadening at the lower frequencies, the pulse appears to be intrinsically broad, roughly 30% of the pulse period, FWHM. The well-established trend of pulse widths increasing as P^−0.5 underestimates this result by a factor of 2–3 (e.g., for the outermost conal component of Mitra & Deshpande 1999).

Polarization calibration and analysis using the PSRCHIVE software package (Hotan et al. 2004) reveals that the radio pulse is substantially linearly polarized, despite depolarization resulting from multipath propagation in the scattering tail, incomplete removal of Faraday rotation (due to the low signal-to-noise ratio), and other effects. The true polarized fraction in the pulse is likely to be significantly larger than we have observed. The variation of the polarization vector's position angle with rotational phase, shown in the topmost panel of Figure 2, is found to be fairly flat, inconsistent with the canonical rotating-vector model (Radhakrishnan & Cooke 1969), but similar to behavior seen in the Crab pulsar (especially its high-frequency components; Moffett & Hankins 1999) and several millisecond-period "recycled" pulsars (Ord et al. 2004). Perhaps not coincidentally, many of these same pulsars exhibit nonthermal magnetospheric pulsations in X-rays. Finally, we note that the pulsar's rotation measure is RM ≃ +270 rad m⁻².

The field containing CXOU J202221 was observed by RXTE for a total of 99 ks over the eight-day span 2010 January 27–February 4 UT (observation P95316). The data were collected with the Proportional Counter Array (PCA; Jahoda et al. 1996) in the GoodXenon mode with an average of 1.8 of the 5 proportional counter units (PCUs) active. In this mode, photons are time-tagged to 0.9 μs and have an absolute uncertainty better than 100 μs (Rots et al. 1998). The effective area of the five combined detectors is approximately 6500 cm² at 10 keV with a roughly circular field-of-view of ∼1° FWHM. Spectral information is available in the 2–60 keV energy band with a resolution of ∼16% at 6 keV.

Production PCA data for this observation were obtained from NASA's HEASARC archive and time-filtered using the standard criteria. This rejected intervals of South Atlantic Anomaly passages, Earth occultations, and other periods of high particle activity, resulting in a total of 94.2 ks of useful data. The photon arrival times were transformed to the solar system barycenter in Barycentric Dynamical Time (TDB) using the JPL DE200 ephemeris and the Chandra-derived coordinates of CXOU J202221.

4.1. Timing Analysis

The timing analysis was restricted to PCA channels 2–50, corresponding approximately to the 2–20 keV energy range, and from the top Xenon layer of each PCU, to optimize the signal-to-noise for a typical pulsar spectrum. Because of the long observation span, photon detection times were accumulated in 4 ms bins and a 2²⁸ point accelerated fast Fourier transform was performed to search for a periodic signal, testing a range of plausible frequency derivatives. This immediately revealed a signal at 24 ms of sufficient strength to analyze short intervals of data and build up a fully phase-coherent timing solution using the time of arrival (TOA) method, as follows.

The RXTE observations of CXOU J202221 distributed over the eight days were clustered into 10 groups of two, three, or four adjacent 96 minute orbits, with large gaps in between. For each of these groups, we were able to extract the period and phase of the signal with sufficiently small uncertainties to maintain cycle counts between fitted epochs. Pulse profiles were generated using the optimum period derived from the Z²_n test (Buccheri et al. 1983) with n = 10 harmonics to allow for the sharp profile. The resulting profiles were cross correlated, shifted, and summed to generate a master pulse profile template. Individual profiles were then cross-correlated with the template to determine the arrival time (time of phase zero) and its uncertainty at each epoch.

The derived TOAs were iteratively fitted to a quadratic ephemeris using the TEMPO software. We started by fitting TOAs from the first three most closely spaced TOAs to a linear solution, and then iteratively added the next TOA. At each step we found that the new TOA would match to <0.02 cycles the predicted phase derived from the previous set. The resulting ephemeris referenced to the mid-point of the observation is presented in Table 1 and the phase residuals are shown in Figure 3. The residuals are all less than 0.02 cycles and appear to be random within their statistical uncertainties, with the exception of a single 3σ data point (Δϕ = 0.025 ± 0.01) which an investigation finds no reason to exclude.

Zoom In Zoom Out Reset image size

Figure 3 displays the pulse profile using all of the 2–20 keV data folded on the final ephemeris. It has a single symmetric peak that is triangular in shape and narrow, with an FWHM of 0.06 of a full cycle. The measured pulsed emission corresponds to 0.91% of the total PCU count rate; however, as will be shown below, the intrinsic signal is nearly 100% pulsed compared to the flux derived from the spectrum of CXOU J202221. We see no energy dependence of the pulse profile when subdividing the 2–20 keV RXTE energy range.

4.2. RXTE Spectral Analysis

The pulsed-flux spectrum of PSR J2022+3842 can be isolated through phase-resolved spectroscopy. We used the FTOOLS fasebin software to construct phase-dependent spectra based on the ephemeris of Table 1. We combined spectra from all observation intervals using only data recorded in the top Xenon layer of PCU2; this PCU was (uniquely) used for all observation intervals over the eight days. Similarly, we averaged the standard PCA responses generated for each epoch. In fitting the pulsed flux, the unpulsed emission provides a near perfect background estimate. The spectra were accumulated in two phase intervals representing the off-peak (0.1 < ϕ_off ⩽ 0.88) and on-peak (0.88 < ϕ_on ⩽ 1.1) emission and fitted using XSPEC. The fits were performed in the 2–16 keV range, above which the background dominates.

Based on the Chandra spectrum, we fitted an absorbed power-law model with the interstellar absorption held fixed at N_H = 1.6 × 10²² cm⁻²; leaving N_H unconstrained results in a larger uncertainty in Γ. The resulting best-fit photon index is Γ = 1.1  ±  0.2 with χ² = 27.9 for 30 dof. The absorbed 2–10 keV flux for the pulsed emission is 5.4 × 10⁻¹³ erg cm⁻² s⁻¹, which represents all of the point-source flux measured from CXOU J202221, indicating that its intrinsic pulsed fraction is essentially 100%. In Figure 4, we show the result of a joint fit to the RXTE pulsed emission from PSR J2022+3842 and CXOU J202221 in the 0.5–16 keV band, leaving only their normalizations free. The best fit parameters are N_H = (1.7 ± 0.3) × 10²² cm⁻² and Γ = 1.0 ± 0.2 with χ² = 99.3 for 100 dof. There is no indication of spectral curvature in the fitted energy range. The independently measured absorbed 2–10 keV flux is the same for both CXOU J202221 and the pulsed emission from PSR J2022+3842, namely F_X = 5.3 × 10⁻¹³ erg cm⁻² s⁻¹. We adopt the results of the joint spectral fit as our final reported values.

Zoom In Zoom Out Reset image size

There can be little doubt that CXOU J202221 and PSR J2022+3842 are one and the same: the nonthermal X-ray source lies at the center of G76.9+1.0, lacks an optical counterpart, and anchors nebular emission consistent with a PWN. Most importantly, the pulsed flux detected using RXTE accounts for all of the Chandra emission and no more. These properties are consistent with pulsars producing broadband magnetospheric emission powered by rotational energy losses. Chandra astrometry thus locates the pulsar to sub-arcsecond precision.

PSR J2022+3842 is outstanding in several respects. It is the second-most energetic Galactic pulsar known, after the Crab pulsar, and the fourth overall, taking into account two LMC pulsars, in N157B (PSR J0537−6910, with $\dot{E} = 4.9\times 10^{38}$ erg s⁻¹ and P = 16 ms) and in SNR 0540−69 (PSR J0540−6919, with $\dot{E} = 1.5\times 10^{38}$ erg s⁻¹ and P = 50 ms). However, it is among the least efficient at converting its spin-down luminosity into X-rays,⁷ with $\eta \equiv L_{\rm 2\hbox{--}10\,keV} /\dot{E} = 5.9\times 10^{-5}\, D^2_{10}$. The flat pulsed spectrum also sets it apart from the typical young, energetic pulsar, deviating significantly from the observed trend $\Gamma \propto 1/\sqrt{\dot{E}}$ (Gotthelf 2003; 2–10 keV), which predicts an index of Γ = 1.8. Flat power laws have primarily been found among pulsars with H.E.S.S. counterparts (e.g., PSR J1838−0655; Gotthelf & Halpern 2008).

PSR J2022+3842 is also the second-most rapidly rotating young pulsar after the LMC pulsar J0537−6910, and the shortest-period radio-bright pulsar known. Its radio properties are thus potentially interesting probes of the elusive radio emission mechanism in pulsars, but detailed studies will be hampered by this object's faintness and the interstellar propagation effects imposed by its great distance. The short rotation period also engenders uncertainty in the pulsar's (and thus the SNR's) age: the characteristic age $\tau _c = P/2\dot{P} = 8.9$ kyr approximates the true age τ only when P₀ ≪ P, where P₀ is the spin period at birth,

$$\begin{equation} \tau = \frac{P}{(n-1)\dot{P}} \left[1-\left(\frac{P_0}{P}\right)^{n-1}\right], \end{equation} \tag{ 1 }$$

and n is the so-called braking index. For a birth period of 16 ms, the shortest known among young pulsars, the true age decreases to ≈5 kyr, assuming spin-down due to magnetic dipole braking (n = 3). A countering effect, however, is spin-down that deviates from the standard dipole assumption: the measured braking indices for several young pulsars are found to lie within the range 2.0 ≲ n < 3.0 (see, e.g., Melatos 1997). For instance, as shown in Figure 5, P₀ = 10 ms for n = 2 would imply an age of 11 kyr. For any assumed value of n, an upper limit to the age of the pulsar, and thus the remnant, is provided in the limit P₀ → 0. A braking index well below n = 1.5 would be highly unusual, so that ∼40 kyr is a reasonable upper limit on the age of G76.9+1.0.

Zoom In Zoom Out Reset image size

PSR J2022+3842 lacks the bright X-ray PWN expected from an energetic pulsar: it is only the second example (of some 20) of a pulsar with $\dot{E} \gtrsim 4\times 10^{36}$ erg s⁻¹ unaccompanied by a PWN of comparable brightness, F_PWN/F_PSR ≳ 1, defying the trend presented in Gotthelf (2004). As such, it is similar to PSR J1617−5055 (Torii et al. 1998), the 69 ms pulsar with $\dot{E} = 1.6 \times 10^{37}$ erg s⁻¹, B_s = 3.1 × 10¹² G, and τ_c = 8.1 kyr. The underluminous PWNe in these cases remain unexplained. In other respects, however, the X-ray PWN around PSR J2022+3842 may not be so unusual. The semimajor axis of the elliptical elongation around the unresolved source is approximately 6'' in length, representing a physical dimension of 9 × 10¹⁷ D₁₀ cm. For comparison, the Vela PWN's X-ray "outer arc" lies at a distance 1 × 10¹⁷ cm from the pulsar in the geometric model of Helfand et al. (2001).

If the extended emission around PSR J2022+3842 does arise from axisymmetric features, the Chandra image constrains—in analogy with the Crab, Vela, and others—the orientation of the pulsar's spin axis: in inclination, to ≃ 50°, and on the sky, to the symmetry axis at position angle −64° north through east, orthogonal to the long direction of the diffuse X-ray emission and the ridge linking the radio lobes. The apparent alignment of spin axes and proper motion vectors in some of these same pulsars offers a testable prediction for the motion of PSR J2022+3842. The spin inclination relative to the line of sight, meanwhile, may help elucidate the role of viewing geometry (e.g., relative to a narrow magnetospheric emission beam) in (1) simultaneously producing a narrow X-ray pulse but an unexpectedly broad radio pulse, a reversal of their typical relationship, and (2) determining whether observed neutron star spectra are primarily thermal or nonthermal: the flat spectrum of PSR J2022+3842 departs markedly from the dominant thermal emissions of the central stars in the Vela and DA 495 SNRs.

We have already alluded to some of the notable similarities between G76.9+1.0 and DA 495, both of which can now unambiguously be characterized as radio PWNe. Both are very bright radio sources without a well-defined outer boundary. Both manifest a bipolar structure and have a steep radio spectrum, unusual for a PWN. DA 495 has a break in its radio spectrum at roughly 1 GHz, and G76.9+1.0 may have one below that as well. Both show very faint X-ray emission compared to their radio emission, and a small X-ray nebula compared to the radio nebula. The conclusions arrived at by ASL+08 for DA 495 may thus also apply to G76.9+1.0. For example, based on the radio/X-ray PWN size ratio of DA 495, ≳ 25, and other lines of argument (see also Kothes et al. 2008), ASL+08 suggested that the pulsar wind energizing DA 495 has a high magnetization factor—i.e., it carries electromagnetic flux and particle flux at roughly comparable levels—in contrast with the Crab, which has a strongly particle-dominated wind, but similar to independent assessments of the wind properties of the Vela pulsar. Moreover, ASL+08 speculated that, because particle-dominated winds are necessary for efficient conversion of wind luminosity to synchrotron luminosity, PWNe in which Poynting flux is an important wind component may be those with dim X-ray PWNe.

The PWN size ratio for G76.9+1.0 is ≃ 20, comparable to DA 495 and nearly two orders of magnitude greater than that for the Crab Nebula. (At a distance of 10 kpc, G76.9+1.0 may be the largest PWN known in our Galaxy, with a physical size of 29 × 35 pc, bigger than or comparable to MSH 15−57, which is at least 25 pc in diameter, Gaensler et al. 2000, and so far believed to be the largest Galactic PWN.) If our prior conclusions hold, the wind from PSR J2022+3842 should be highly magnetized well beyond its termination shock, contributing to the low X-ray conversion efficiency of its PWN. Why this might be so is an open question.

The overarching conclusion of ASL+08 and Kothes et al. (2008) was that DA 495 is likely to be a PWN of advanced age (∼20 kyr), but having evolved without significant interaction with the ambient medium. All of the characteristics we find for G76.9+1.0 similarly indicate a rather old object, yet PSR J2022+3842 has all the characteristics of a young object: a spin-down age of 20 kyr would require both n < 2 and very high spin at birth, P₀ ≲ 5 ms. We do not have a ready explanation for this apparent discrepancy, which highlights the importance of having uncovered the spin and energetic properties of the central pulsar.

G76.9+1.0 and its pulsar, J2022+3842, are clearly unusual and require further investigation. Additional multi-wavelength observations (e.g., an X-ray search for the SNR shell, to aid in constraining the system's age) may be fruitful in providing important components for their understanding. Based on its spin-down luminosity, and given the sub-arcsec localization and available timing information, PSR J2022+3842 is a good candidate for a search for gamma-ray pulsations using Fermi data. The absence of a bright PWN, however, makes it an unlikely TeV target, even though the spectrum of the pulsed X-ray emission suggests otherwise.

Support for this work was provided by the National Aeronautics and Space Administration through Chandra Award Number GO5-6077Z issued by the Chandra X-ray Observatory Center, which is operated by the Smithsonian Astrophysical Observatory for and on behalf of NASA under contract NAS8-03060. The National Radio Astronomy Observatory is a facility of the National Science Foundation operated under cooperative agreement by Associated Universities, Inc. We have also made use of RXTE data provided by the High Energy Astrophysics Science Archive Research Center at NASA's Goddard Space Flight Center, as well as data products from the Two Micron All Sky Survey, a joint project of the University of Massachusetts and the Infrared Processing and Analysis Center/Caltech, funded by NASA and the National Science Foundation. S.S.H. acknowledges support by the Natural Sciences and Engineering Research Council of Canada (NSERC) and the Canada Research Chairs program.