XMM-Newton Observations of Young and Energetic Pulsar J2022+3842
aa r X i v : . [ a s t r o - ph . H E ] J un Draft version August 23, 2018
Preprint typeset using L A TEX style emulateapj v. 08/22/09
XMM-NEWTON
OBSERVATIONS OF YOUNG AND ENERGETIC PULSAR J2022+3842
P. Arumugasamy
Department of Astronomy & Astrophysics, Pennsylvania State University, 525 Davey Lab,University Park, PA 16802, USA
G. G. Pavlov
Department of Astronomy & Astrophysics, Pennsylvania State University, 525 Davey Lab,University Park, PA 16802, USA
O. Kargaltsev
Department of Physics, The George Washington University, Washington, DC 20052, USA
Draft version August 23, 2018
ABSTRACTWe report on
XMM-Newton
EPIC observations of the young pulsar J2022+3842, with a character-istic age of 8.9 kyr. We detected X-ray pulsations and found the pulsation period P ≈ . P ≈ . × − , twice larger than the previously reported values. The pulsar exhibits twovery narrow (FWHM ∼ . ≈ .
48 of the period, witha pulsed fraction of ≈ .
8. Using the correct values of P and ˙ P , we calculate the pulsar’s spin-downpower ˙ E = 3 . × erg s − and magnetic field B = 2 . × G. The pulsar spectrum is wellmodeled with a hard power-law (PL) model (photon index Γ = 0 . ± .
1, hydrogen column density n H = (2 . ± . × cm − ). We detect a weak off-pulse emission which can be modeled with a softerPL (Γ ≈ . ± . η PSR = L PSR / ˙ E = 2 × − ( D/
10 kpc) ,is similar to those of other pulsars. The XMM-Newton observation did not detect extended emis-sion around the pulsar. Our re-analysis of
Chandra
X-ray observatory archival data shows a hard,Γ ≈ . ± .
5, spectrum and a low efficiency, η PWN ∼ × − ( D/
10 kpc) , for the compact pulsarwind nebula, unresolved in the XMM-Newton images.
Subject headings: pulsars: individual (PSR J2022+3842) — stars: neutron — X-rays: stars INTRODUCTION
Nonthermal emission of rotation-powered pulsars(RPPs), observable from the radio to γ -rays, is poweredby the loss of their rotational energy. X-ray observationsof RPPs allow one to understand the origin and mecha-nisms of the nonthermal emission from the pulsar mag-netosphere and thermal emission from the neutron star(NS) surface. If the pulsar is young enough, X-ray obser-vations can also detect the pulsar wind nebula (PWN),whose synchrotron emission is generated by relativisticparticles outflowing from the pulsar magnetosphere, andthe supernova remnant (SNR), formed by the same su-pernova explosion as the pulsar. They are particularlyuseful for pulsars that have been observed at other wave-lengths, in which case the multi-wavelength data analy-sis helps to understand the properties of the emittingparticles, the locations of the emitting regions, and themechanisms involved in the multi-wavelength emission.PSR J2022+3842 is a young, energetic pulsar, dis-covered by Arzoumanian et al. (2011) (henceforth re-ferred to as A+11) in a 54 ks Chandra
X-ray observa-tory (CXO) observation of the radio SNR G76.9+1.0(Landecker et al. 1993). Although A+11 found no ev-idence for G76.9+1.0 in the CXO data, they did finda point source CXOU J202221.68+384214.8, surroundedby a faint nebulosity, at the center of the radio SNR,which they interpreted as a pulsar and its PWN. A+11
Electronic address: [email protected] fit an absorbed power-law (PL) model to the pulsar spec-trum and found a hydrogen column density n H, ≡ n H / (10 cm − ) = 1 . ± . . ± .
2. From an absorbed PL fit of the PWN spec-trum, they obtained an unusually low absorbed flux ra-tio F PWN /F PSR ≈ .
08 in the 2–10 keV band (assumingfixed n H, = 1 . . Rossi X-ray Timing Explorer ( RXTE ), A+11 found apulsation period P = 24 ms with a spin-down rate ˙ P ≈ . × − s s − (MJD 54957–55469), and a spin glitch ofmagnitude ∆ P/P ≃ . × − (between MJD 54400 and54957). They derived the pulsar’s dispersion measureDM = 429 . ± . − , which formally correspondsto very large distances, D >
50 kpc in the NE2001 Galac-tic electron distribution model (Cordes & Lazio 2002).However, the authors noted that a likely overdensity offree electrons in the Cygnus region, along the line of site,may account for the higher-than-expected DM, so theactual distance remains uncertain.The pulsar’s 2–20 keV X-ray pulse profile, obtainedwith the GBT/
RXTE ephemeris, shows a single narrowpulse (FWHM = 0.06 of full cycle) with a 91% – 100%pulsed fraction (A+11). The authors fit a Γ = 1 . ± . n H, =1 .
6. They derived the pulsar’s spin-down power ˙ E =1 . × erg s − , and estimated the pulsar’s 0.5 – 8keV X-ray efficiency η X ≡ L X / ˙ E = 5 . × − D , where D is the distance to the pulsar in units of 10 kpc. Insummary, A+11 characterized this distant pulsar as themost rapidly rotating non-recycled pulsar and the secondmost energetic Galactic pulsar known (after the Crabpulsar), but far less efficient at generating a PWN andconverting the spin-down power to X-rays.The pulsar has not been detected in the γ -rays, per-haps due to its location amidst a particularly crowded re-gion in the γ -ray sky. An unidentified Fermi source 2FGLJ2022.8+3843c is listed in the Second
Fermi
Catalog, andgiven a tentative association with the SNR G079.6+01.0(Nolan et al. 2012). Abdo et al. (2013) discuss a possi-ble pulsar counterpart ∼ . ◦
06 from the pulsar position,which they claim to show a PL spectrum with exponen-tial cut-off, but still without any pulsations.To study the pulsar’s phase-resolved X-ray spec-trum and further investigate its unusually faint PWN,we carried out a 110 ks
XMM-Newton observation ofJ2022+3842. In this deep observation we searched forX-ray counterpart of the radio SNR, an extended PWNand the pulsar’s off-pulse emission. We also performedX-ray timing of the pulsar and phase-resolved spectralanalysis. OBSERVATION AND DATA ANALYSIS
Pulsar J2022+3842 was observed with the Euro-pean Photon Imaging Camera (EPIC) of the
XMM-Newton observatory (obsid 0652770101) on 2011 April14 (MJD 55665) for about 110 ks. EPIC-pn chip
XMM-Newton
ScienceAnalysis System (SAS) 12.0.0 , applying standard tasks.The observations were partly affected by soft-protonflares. These flaring events are characterized by periodsof significantly higher background and rapid variability.Periods of strong flaring are better identified using lightcurves of single pixel events (Pattern = 0) with energies >
10 keV, henceforth referred to as flaring light curves .In Figure 1, we show the EPIC-pn (chip emldetect on theMOS1 image (Figure 2), we determined the target sourcecoordinates, α = 20 h m . s , δ = +38 ◦ ′ . ′′
61, witha statistical 1 σ uncertainty of 0 . ′′
18. This position differsfrom the CXO position by 1 . ′′
08, which is consistent withthe
XMM-Newton ’s systematic position uncertainty of ≈ ′′ (Watson et al. 2009). http://xmm.esac.esa.int/sas http://xmm.esac.esa.int/sas/current/documentation/threads/ C oun t r a t e ( c / s ) Time (ks)0.12 cps C oun t r a t e ( c / s ) Time (ks)1.0 cps
Figure 1.
Flaring light curves in EPIC-pn (top) and MOS1 (bot-tom) for the entire observation duration. Optimal GTI cut-off ratesare shown by dashed blue lines. ° , ″ , ″ , , ″ , , ″ , h m s s s s s s s D ec l . ( J ) R. A. (J2000)
Figure 2.
Binned and smoothed MOS1 image of the field aroundPSR J2022+3842 (center) in the 0.5–10 keV band.
Timing Analysis
In the EPIC–pn (PN hereafter) timing mode, theevents collected over the entire chip µ s at the expense of positional information along thecoordinate axis RAWY. In Figure 3 (top-right panel),we show the GTI-filtered 0.5–12 keV PN data by plot-ting the events’ RAWX positions against their times ofarrival (TOAs). Note that this representation is differentfrom the conventional RAWX versus RAWY plot. Theplotted time coordinate represents elapsed time since thestart of observation, and the horizontal gaps in the plotrepresent flaring intervals from which data has been dis-carded; the initial 25 ks of the filtered flaring interval isomitted from the plot (see Figure 1, top panel).Since positional information is available only along onecoordinate for all events in PN, we located the targetand other sources in the field by analyzing the MOS1imaging mode data. By identifying the PN timing-chip’sfield-of-view (FOV) on the MOS1 image (Figure 3, top-left panel), we found two potential contaminant sources,C1 and C2 (Figure 2), with the projected RAWX sep-arations from the target of about 9 ′′ and 8 ′′ ( ≈ . ′′ × . ′′ barycen to trans-form the X-ray event times from spacecraft TerrestrialTime (TT) to Barycenter Dynamical Time (TDB). Wefound the previously reported 41 Hz pulsations (A+11)using Z test (e.g., Buccheri et al. 1983). However, sub-sequent phase folding over twice longer period revealstwo distinct pulses with markedly unequal fluxes. Weconclude that the pulsar has a twice smaller pulsationfrequency, about 20.5 Hz, with two narrow peaks perperiod (main pulse and interpulse).To measure the frequency more precisely and estimatethe frequency derivative, we switched to Z n tests ( n > ν - ˙ ν space in the box 20 .
584 Hz < ν < .
586 Hz, − . × − Hz s − < ˙ ν < − . × − Hz s − , withstep sizes of 1 × − Hz and 1 × − Hz s − , andfound Z , max = 1515 for ν = 20 . ν = − . × − Hz s − (Figure 4, bottom panel) at thereference epoch 55666.23783581 (MJD TDB). Here andbelow the numbers in parentheses are 1 σ uncertainties forthe corresponding last significant digit(s) of the measuredquantity.We show the results of Z n tests, for n = 1–17, in thetop panel of Figure 4. The H test (de Jager et al. 1989)fails to find a reasonable value for the number of statis-tically significant harmonics, as the H-statistic is an in-creasing function of n even beyond n = 30. Adopting Z n test with multiple harmonics ( n ≥ ν XMM = 20 . ° , ″ , , ″ , , ″ h m s s s s s D ec l . ( J ) R. A. (J2000)PulsarC1C2 2535455565758595105 0 10 20 30 40 50 60 T I M E ( k s ) EPIC-pn RAWX C oun t s EPIC-pn RAWX B ac kg r ound S ou r ce B ac kg r ound Figure 3.
Top-left: Combined MOS1 and MOS2 (imaging modechips) image, cropped to the PN timing-chip FOV, with the tar-get pulsar and contaminant sources C1 and C2. Top-right: GTIfiltered, 0.5–12 keV PN timing data; time origin reset to the ob-servation start time. Bottom: PN events histogram along RAWX.The shaded areas show the events extraction ranges for spectralanalysis for the pulsar (RAWX = 38–42, grey) and background(RAWX = 28–33 and 48–53, yellow).
Hz, ˙ ν XMM = − . × − Hz s − .The corrected pulsar ephemeris at the RXTE ob-servation reference epoch of 55227.00000027, ν RXTE =20 . , ˙ ν RXTE = − . × − Hzs − , is straight-forwardly inferred from the values re-ported by A+11. From this ephemeris, the expectedfrequency at the reference epoch of the XMM-Newton observation is 20.5851193(30) Hz, which coincides withthe measured ν XMM at a 0 . σ level. Conversely, usingthe frequency values at the RXTE and
XMM-Newton epochs, we calculate the long-term frequency derivative˙ ν XMM − RXTE = ( ν XMM − ν RXTE ) / ∆ T = − . × − Hz s − (where ∆ T = 439 .
238 days is the dif-ference between the epochs). Being more precise than˙ ν RXTE due to the much longer time span, this estimateis consistent with ˙ ν RXTE at the 0 . σ level, which sug-gests that there were no glitches between the RXTE and
XMM-Newton observations. It, however, differs byabout 3 σ from ˙ ν XMM . Given the excellent agreement be-tween ˙ ν RXTE and ˙ ν XMM − RXTE , and the relatively short Z n2 ( ν - ν c ) x 10 (Hz) ν c = 20.58511983 Hz 0 200 400 600 800 1000 1200 1400 1600 20.5846 20.5848 20.585 20.5852 20.5854 20.5856 Z Frequency ν (Hz)-- 20.58511979 Hz Figure 4.
Bottom: Pulsation frequency search using the Z test. Top: The Z n statistics around the central frequency ν c =20 . n = 1–17 at ˙ ν = − . × − Hzs − . time span of the XMM-Newton observation (82 ks versus691 ks for the
RXTE observation), we consider ˙ ν RXTE (or˙ ν XMM − RXTE if we believe there were no glitches betweenthe two observations) more reliable than ˙ ν XMM . Thetiming solution and derived pulsar properties are listedin Table 1.The 0.5–12 keV folded ( ν = 20 . ν = − . × − Hz s − , zero phase epoch =55666.23783581) and binned (250 equal bins) X-ray pulseprofile is shown in the top panel of Figure 5. In orderto determine the pulse phase and pulse separation accu-rately, we first smooth the data using an adaptive kerneldensity estimation (KDE) technique (Feigelson & Babu2012). We assign Gaussian kernels to each event with abandwidth adapted to the number density of events atits phase. The smoothed and area-normalized pulse pro-file is shown in the bottom panel of Figure 5. The mainpulse and interpulse peak at phases φ main = 0 . ± . . ± . φ inter = 0 . ± . ± . ± . ≈ .
074 and ≈ . ≈ . Table 1
Timing solution and derived parameters for PSR J2022+3842.Parameter ValuePeriod P (ms) 48.578779636(24)Period derivative ˙ P × − Epoch (MJD TDB) 55666.23783581Main Pulse (FWHM) 0 . ± . . ± . . ± . E (erg s − ) 3 . × Characteristic age τ (kyr) 8.9Surface dipole magnetic field B s (G) 2 . × tion p = 0 . ± .
02, defined as the ratio of background-subtracted counts in the two pulses ( N pulsed = 2703)to the background-subtracted net source counts ( N net =3488, N bgd = 6267). The intrinsic pulsed fraction p int of the pulsar radiation is higher because of some contri-bution from the unresolved PWN. Using the PWN fluxmeasured from the CXO ACIS data (see Section 2.2), weestimate p int = 0 . ± .
03. The 1 σ uncertainties for thepulse profile parameters quoted above are found throughMonte-Carlo estimations with non-parametric bootstrapre-sampling of our data (Feigelson & Babu 2012). C oun t s p e r b i n Phase 0 1 2 3 4 5 6 7 8 0 0.2 0.4 0.6 0.8 1 N o r m a li ze d C oun t s Phase
Figure 5.
Top: Phase-folded and binned pulse profile with av-eraged background count rate (red line), average off-pulse countrate (black, dotted line), and count rate cut-off used to establishpulse base width (dashed, green line). Bottom: KDE smoothed,normalized pulse profile.
We performed a similar analysis of the MOS2 tim-ing mode data. The MOS2 CCD has a lower sensitiv-ity than PN and a considerably lower time resolution of1.5 ms. We achieved highest S/N for 2820 total countsextracted in the 1.1–8 keV range, over 109.7 ks of the ob-servation, of which only 646 ±
35 were from the source.The Z test returned a high statistic Z , max = 862 for ν = 20 . ν = − . × − Hzs − , at the reference epoch 55666.23783581, consistentwith the PN timing ephemeris. We, however, found thephase-folded pulse profile to be noisy due to low sourcecounts, with the pulses broadened due to the poorer timeresolution of MOS2. We also found an absolute timingerror of ≈ +6 . , we exclude the MOS2 data from further analy-sis. Spectral Analysis
We use XSPEC v.12.7.1 for X-ray spectral anal-ysis. We model absorption by the interstellarmedium (ISM) using the T¨ubingen-Boulder model(Wilms et al. 2000) through its XPEC implementation tbabs , setting the abundance table to wilm (Wilms et al.2000) and photoelectric cross-section table to bcmc (Balucinska-Church & McCammon 1992), with new Hecross-section based on Yan et al. (1998). We performchi-square fitting of the spectra (C-statistic for contam-inant C2), and quote the 90% confidence uncertaintiesfor the model parameters evaluated for single interestingparameter.Prior to pulsar spectral analysis, we modeled the spec-tra of the contaminating sources C1 and C2 (see Figure3), using the MOS1 and archival (ObsID α = 20 h m . s δ = +38 ◦ ′ . ′′
55 is offset by 75 ′′ fromthe pulsar, but this separation projected onto the one-dimesional (1D) PN image is just 9 ′′ . For spectral fit-ting, we extracted events from a 12 ′′ radius circle aroundthe source in MOS1 (308 net source counts in the 0.2–10 keV band) and from a 4 . ′′ D ≈ .
15 kpc in Cygnus .We fit the stellar spectrum with a two-component APECmodel (calculated using ATOMDB code v2.0.1 ) whichdescribes the emission from shocked, collisionally-ionizedwinds seen in such early-type stars. For the best fit( χ ν = 0 . kT = 0 . +0 . − . keV and kT = 2 . ± . n H = 1 . +0 . − . × cm − . The absorbed flux is F abs0 . − = 2 . +0 . − . × − erg cm − s − .Contaminant C2 at coordinates α = 20 h m . s δ =+38 ◦ ′ . ′′
46 is offset by 103 ′′ from the pulsar but has aprojected separation of 8 ′′ in the 1D PN image. It is an http://xmm2.esac.esa.int/docs/documents/CAL-TN-0082.pdf http://heasarc.gsfc.nasa.gov/docs/xanadu/xspec http://simbad.u-strasbg.fr/simbad/ http://atomdb.org −6 −5 −4 −3 C oun t s s − k e V − χ ν = 0.95 1 100.2 0.5 2 5−202 Δ χ Energy (keV)
Figure 6.
ACIS-S (red) and MOS1 (black) spectra of contam-inant C1 fitted with two APEC components. Individual modelcomponents for MOS1 are shown with black curves. The bottompanel shows residuals of the χ fit in units of sigma. unidentified soft X-ray point source. For spectral fitting,we extracted events from a 10 ′′ radius circle around thesource in MOS1 (133 net counts in 0.4–10 keV band),and from a 3 . ′′ kT = 2 . +0 . − . keV, n H = 3 . +1 . − . × cm − , and F abs0 . − = 1.7 +0 . − . × − erg cm − s − (see Figure 7). −6 −5 −4 −3 C oun t s s − k e V − −3 −10 −3 −3 −3 R e s i dua l s Energy (keV)
Figure 7.
ACIS-S (red) and MOS1 (black) spectra of contaminantC2 fitted with APEC model. The bottom panel shows C-statisticfit residuals.
The low-energy part of the pulsar’s emission ( E . . b = 0 . ◦ . +0 . − . fits the phase-integrated spectra very well (Table 3,Figure 8). Inclusion of the contamination-free MOS1and ACIS-S spectra with a lower energy cut-off allowedus to constrain the hydrogen column density, n H, =2 . +0 . − . . The two parameter confidence contours forthis PL fit are shown in Figure 9.The photon index we measured is consistent with thatobtained by A+11 from the ACIS-S data, but the hy-drogen column density is substantially different from n H, = 1 . ± . n H, = 2 . ± . n H values is due to the differ-ent absorption model ( phabs with abundance table angr ;Anders & Grevesse 1989) used by A+11.To examine the dependence of spectral parameters onpulsation phase, we divided the pulse profile into mainpulse, interpulse and off-pulse regions and analyzed theirspectra individually. The main pulse contributes ≈ ∼
10% phase interval.This allowed us to extract a high-quality (S / N > . ′′ ≈ . n H value for the pulsar at n H, = 2 .
32, ob-tained from the phase-integrated fit. The procedure out-lined above provides the best constraints on the photonindex for the interpulse and off-pulse emission. The ex-traction parameters are listed in Table 2, and the fittingparameter values are listed in Table 3. The best spec-tral fit and residuals for interpulse emission are shown inFigure 11, and for off-pulse emission in Figure 12. Theinterpulse PL slope is close to that of the main pulsewhile for the off-pulsar emission the spectrum appearsto be softer, but with a large uncertainty in its slope.Our search for a PWN in the
XMM-Newton data didnot yield positive results. We fit the ACIS-S PWN spec-trum with a PL model at fixed n H, = 2 .
32 and ob-tained Γ = 0 . ± .
5, which is marginally consistent withΓ = 1 . . ′′ . ′′ F abs0 . − = 5 +2 − × − erg cm − s − ,is consistent with that estimated by A+11. SUMMARY AND DISCUSSION
We did not detect any prominent extended emissionin the 105 ks MOS1 exposure of the region aroundPSR J2022+3842. The presence of a large number of X-ray point sources around the pulsar hinders quantitativespatial analysis for assigning restrictive upper limits on −4 −3 C oun t s s − k e V − Γ = 0.93 Power law χ ν = 0.851 100.5 2 5−202 ∆ χ Energy (keV)
Figure 8.
Absorbed PL fit and its residuals (in units of sigma)for the phase-integrated pulsar spectra from ACIS-S (black), MOS1(green) and PN (red). + 0.75 0.8 0.85 0.9 0.95 1 1.05 1.1 1.15Photon Index Γ n H , P L N o r m . N - Photon Index Γ + . . . Figure 9.
Top: n H – Γ confidence contours at the 68%, 90%,and 99% levels for the phase-integrated spectral fit. Bottom: N − – Γ confidence contours at the 68%, 90%, and 99% levels for thephase-integrated spectral fit. N − is the PL normalization in unitsof 10 − photons cm − s − keV − at 1 keV. the extended emission from either the SNR or the PWN.Our timing analysis has shown that the true pulsar Table 2
Extraction parameters for phase-integrated, main pulse, interpulse and off-pulse spectraIntegrated Main pulse Interpulse Off-pulse PWNMOS1 PN ACIS-S PN ACIS-SPhase Range b a ′′
38 – 42 2 . ′′ . ′′ × . ′′ c ±
42 2777 ±
91 1183 ±
35 1320 ±
41 1130 ±
45 1383 ±
109 96 ± a PN extraction region specified in RAWX coordinate, in pixels (1 pixel = 4 . ′′ b Pulsed to off-pulse transitional phases are omitted to obtain better constraints on fit parameters. c σ uncertainties assuming Poisson statistics. Table 3
Fitting parameters with 90% confidence uncertainties for PSR J2022+3842 and its PWN.Phase range n H, Γ PL. norm. a χ ν /d.o.f. F abs0 . − F unabs0 . − Integrated (100%) c . +0 . − . . +0 . − . . +0 . − . . +0 . − . . +0 . − . Main pulse (9%) 2 . +0 . − . . +0 . − . . +6 . − . . +1 . − . . +2 . − . Interpulse (8%) 2.32 (fixed) 0 . +0 . − . . +2 . − . . +1 . − . . +1 . − . Off-pulse (76%) 2.32 (fixed) 1 . +0 . − . . +2 . − . . +0 . − . . +0 . − . PWN 2.32 (fixed) 0 . +0 . − . . +0 . − . d /6 0 . +0 . − . . +0 . − . PL normalization in units of 10 − photons cm − s − keV − at 1 keV. b F abs0 . − and F unabs0 . − are absorbed and unabsorbed fluxes, respectively, in units of 10 − erg cm − s − . c Percentages in parentheses denote the fraction of total period included. d Best-fit value of C-statistic. −3 C oun t s s − k e V − Γ = 0.90 Power law χ ν = 0.991 100.5 2 5−202 ∆ χ Energy (keV)
Figure 10.
Absorbed PL fit and its residuals for the main pulsespectrum. period, P ≈ . P ≈ . × − , are twice larger than those reported by A+11, andthe phase-folded light curve has two peaks per period, themain pulse and the interpulse, separated by ≈ .
48 of theperiod. Using these P and ˙ P values, we re-evaluated thepulsar’s spin-down power, ˙ E = 3 . × erg s − , andmagnetic field strength B = 2 . × G.The X-ray pulses are very narrow compared to mostof the pulsars with known X-ray pulse profiles. How-ever, a young ( τ = 5 . P = 65 . −3 −3 −3 C oun t s s − k e V − Contamination Γ = 0.90 Power law χ ν = 1.11 100.5 2 5−202 ∆ χ Energy (keV)
Figure 11.
Absorbed PL fit and its residuals for interpulse spec-trum. The black solid and red dashed lines show the pulsar’s andcombined C1+C2 contributions, respectively. ms) PSR J0205+6449 shows a similar X-ray pulse profileand spectral characteristics (Kuiper et al. 2010). Thevery narrow X-ray pulse profiles and hard X-ray spec-tra of these pulsars indicate that the X-ray emissionoriginates from the pulsar magnetosphere. The double-peaked profile, with separations of ≈ .
48 and no dis-cernible bridge emission, indicate emission from dia-metrically opposite sites in the pulsar magnetosphere.Gamma-ray light curves possessing similar characteris-tics favor a high magnetic obliquity (large angle α be- − − − . C oun t s s − k e V − Contamination Γ = 1.7 Power law χ ν = 0.851 100.5 2 5 − ∆ χ Energy (keV)
Figure 12.
Absorbed PL fit and its residuals for the off-pulsespectrum. The solid and dashed curves show the contributionsfrom the pulsar and the C1+C2 contaminants, respectively. tween the rotation and magnetic axes) for the pulsar(Watters et al. 2009). From radio and γ -ray light curvemodeling of PSR J0205+6449, Pierbattista et al. (2014)estimate α ≈ ◦ for the pulsar. If the similarities toPSR J0205+6449 do extend to the γ -ray regime, PSRJ2022+3842 could be established as a nearly orthogonalrotator. This can be further tested through the γ -raylight curve modeling (if γ -ray emission is detected in fu-ture), or through the radio polarization measurements.Our estimate of the total hydrogen column den-sity (neutral, ionized and molecular) towards PSRJ2022+3842, n H, = 2 . +0 . − . , obtained using the tbabs model with wilm elemental abundances, is sig-nificantly higher than the previous estimate, n H, =1 . ± . phabs model with angr abundances. We conclude that estimating hydro-gen column densities through X-ray spectral modeling ofemission from heavily obscured targets is highly sensitiveto the ISM absorption model and abundance table used.The phase-integrated pulsar spectrum fits a hard PLmodel with Γ = 0 . ± .
1. The main pulse and the in-terpulse contribute ∼
80% of the total emission. Theoff-pulse spectrum is poorly constrained due to contami-nation and an inherently weak signal. A possible sourceof the off-pulse emission could be the compact PWN,which cannot be resolved by
XMM-Newton because ofits broad point spread function. Comparing the PN off-pulse spectrum with the ACIS-S PWN spectrum (see Ta-ble 3), we find different best-fit values of photon indexand flux, but the uncertainties are too large to claim thedistinction between the two spectra to be statisticallysignificant. From re-analysis of the ACIS-S data, we alsofound the PWN spectrum to be harder than previouslyassumed, with Γ = 0 . +0 . − . . This result is consistentwith the empirical correlation between the PWN photonindex and its 2–10 keV luminosity (and, more tightly,the PWN X-ray efficiency (see Figure 1 and Figure 7 inLi et al. 2008).We have assessed that the pulsar has a factor of 4lower spin-down power and a slightly higher X-ray fluxthan reported by A+11. As a result, our pulsar X-rayefficiency estimate is a factor of 4 higher, η PSR0 . − = L PSR0 . − / ˙ E = 2 . × − D . As shown in Figure 13(top panel), the X-ray efficiency of PSR J2022+3842 is comparable to those of other young, energetic pul-sars for the adopted distance of 10 kpc (for illustrativepurposes, we assign 25% uncertainty to J2022+3842’sdistance). In contrast, the associated PWN efficiency, η PWN0 . − ∼ × − D , is the lowest among youngpulsars with comparable values of ˙ E (Figure 13, bottompanel). A low magnetic obliquity might in principle ex-plain a weak PWN, but is disfavored by the observedX-ray light curve. The reason for so low PWN efficiencyremains to be understood.We thank Michael Freyberg and Bettina Posselt for dis-cussions and clarifications regarding EPIC-MOS2 timinganalysis. We also thank Eric Feigelson for valuable dis-cussion and suggestion on statistical techniques, and thereferee for useful comments. This work was partly sup-ported by NASA Astrophysics Data Analysis Programaward NNX13AF21G.
30 31 32 33 34 35 36 37 L og L p s r ( . - k e V ) η = - η = - J0537-6910B0531+21B0540-69J2229+6114J1617-5055J1124-5916B0833-45J2021+3651B1706-44J1016-5857J1747-2958B1046-58J1301-6305J0729-1448J1740+1000B1957+20J0538+2817B0355+54J0633+1746J2124-3358B1929+10B2224+65 J1833-1034J0205+6449B1509-58J1420-6048J1846-0258J1811-1925J1357-6429B1757-24J1119-6127J1718-3825J1509-5850B1853+01J1702-4128
J2022+3842
29 30 31 32 33 34 35 36 37 33 34 35 36 37 38 39 L og L p w n ( . - k e V ) Log E. η = - η = - J0537-6910B0531+21B0540-69J2229+6114J1617-5055J1124-5916B0833-45J2021+3651B1706-44J1016-5857J1747-2958B1046-58J1301-6305J1740+1000B1957+20J0538+2817B0355+54J0633+1746B1929+10 J1833-1034J0205+6449B1509-58J1420-6048J1846-0258J1811-1925J1357-6429B1757-24J1119-6127J1718-3825J1509-5850B1853+01J1702-4128 J1301-6305B1957+20
J2022+3842
Figure 13.
Comparison of pulsar and PWN efficiencies for non-recycled pulsars with spin-down power in the 10 − erg s − range.The dashed straight lines are lines of constant efficiency. PSR J2022+3842 and its PWN are marked by blue asterisks. These graphs areadapted from Kargaltsev & Pavlov (2008) (Tables 1 and 2, and Figure 5).0