An X-ray Nebula Associated with the Millisecond Pulsar B1957+20

a r X i v :a s t r o -p h /0302588v 1 28 F e

b 2003An X-ray Nebula Associated with the Millisecond

Pulsar B1957+20

B.W.Stappers,1,2?B.M.Gaensler,3V .M.Kaspi,4,5M.van der Klis,2W.H.G.Lewin,51Stichting ASTRON,7990Dwingeloo,The Netherlands 2Sterrenkundig Instituut “Anton Pannekoek”,1098SJ Amsterdam,The Netherlands 3Harvard-Smithsonian Center for Astrophysics,60Garden St,Cambridge,Massachusetts,USA 4Physics Department,McGill University,3600University Street,Montreal,Quebec,Canada 5Physics Department and Center for Space Research,Massachusetts Institute of Technology,70Vasser St.,Cambridge,Massachusetts,USA ?To whom correspondence should be addressed;E-mail:stappers@astron.nl We have detected an x-ray nebula around the binary millisecond pulsar B1957+20.A narrow tail,corresponding to the shocked pulsar wind,is seen interior to the known H αbow shock and proves the long-held assumption that the rotational energy of millisecond pulsars is dissipated through relativistic winds.Unre-

solved x-ray emission likely represents the shock where the winds of the pul-sar and its companion collide.This emission indicates that the ef?ciency with which relativistic particles are accelerated in the post shock ?ow is similar to that for young pulsars,despite the shock proximity and much weaker surface magnetic ?eld of this millisecond pulsar.

Millisecond pulsars are old neutron stars (typically ～3Gyr)that have been spun up to a

rapid rotation rate (<～

25ms)by accretion of material from a binary companion (1,2).After the

accretion phase,they appear as radio pulsars with surface magnetic?eld strengths of～108G, which,combined with their older ages and rapid rotation rates,means that they form a separate population from younger pulsars.

PSR B1957+20is the second fastest-spinning pulsar known,with a rotation period(3)of 1.6ms and rotational spin-down luminosity(4)˙E=1×1035ergs s?1.The pulsar is in a9.16-hour binary orbit with a low-mass companion star.The wind of the companion star eclipses the radio emission for～10%of every orbit.The PSR B1957+20binary system provides an excellent opportunity to study the wind of a weakly magnetized,recycled neutron star.The wind is ablating,and may eventually evaporate,the low-mass companion star.Ablation and heating of the companion star(5)are believed to be caused by x-orγ-rays generated in an intrabinary shock between the pulsar wind and that of the companion star.Meanwhile,the high space velocity of the pulsar(>220km s?1)(6),as it moves through the interstellar medium (ISM),generates suf?cient ram-pressure to con?ne the pulsar wind and results in the formation of a bow shock.Upstream of this bow shock Hαemission is generated where neutral material is swept up and collisionally excited;this Hαnebula absorbs1–10%of the spin-down energy(7).

A reverse or termination shock decelerates the pulsar wind and is located between the pulsar and the bow shock.

The intrabinary and termination shocks both provide diagnostics of the pulsar wind.Ob-servations of the interaction of young pulsar winds and the ISM have shown that they power x-ray emitting synchrotron nebulae which are typi?ed by the Crab nebula(8).High-resolution observations of such pulsar-wind nebulae have shown that the composition of the wind,char-acterized by the ratio of the Poynting?ux to particle energy?uxσ,is kinetic energy dominated (σ≈0.003).However,for the evolutionarily distinct millisecond pulsar population,the com-position,ef?ciency,and geometry of the pulsar wind remain unknown.ROSAT observations of PSR B1957+20provided the only previous constraint,and they have been interpreted as indi-

cating that its wind is different from that of the Crab pulsar(9).However,these data lacked the spatial resolution and sensitivity to determine the cause of the emission.

We have undertaken a43kilosecond(10)observation of the millisecond pulsar B1957+20 using the Chandra X-ray Observatory.The brightest x-ray source in the?eld(Fig.1)is coinci-dent with the pulsar position.A tail of x-ray emission is seen extending from the pulsar to the northeast by at least16′′with a position angle opposite to the pulsar’s proper motion direction of212?(6).

Striking con?rmation of the association with the pulsar and the ISM shock comes by com-paring this x-ray tail with the Hαbow-shock nebula(Fig.2).In bow-shock nebulae,material at the termination shock is swept back by the ram pressure and forms a cylindrical tube aligned with the proper motion direction and interior to the bow shock(13,14).The morphology,di-rection and location of the x-ray nebula(Fig.2)therefore indicate that it corresponds to the termination shock and,in combination with the enclosing Hαemission,demonstrates the ex-pected double-shock nature of the pulsar’s interaction with the ambient medium.

Young pulsar winds are thought to be relativistic,and correspondingly generate non-thermal emission by synchrotron and/or inverse Compton processes(15,16).However,nothing is known about millisecond pulsar winds except that the scale of the observed Hαnebulae are at least consistent with their having relativistic winds(17).A different wind might be expected because of the reduction and possible alteration of the nature of the magnetic?eld during the accretion phase.We must therefore also consider the possibility that they drive slower winds like that of the Sun,resulting in shock-heated thermal x-rays.If the spin-down luminosity of a millisecond pulsar is carried away as kinetic energy in an out?ow then˙E=0.5˙MV2w,where˙M is the mass loss rate and V w is the wind speed.Because the pulsar remains and is at least1Gyr old we know that˙M<1017g s?1,and thus V w>109cm s?1.The temperature that we measure(kT≈1keV, where k is the Boltzmann constant)by?tting a spectrum expected for shock-heated gas to the

energy distribution of the82counts recorded in the tail region is inconsistent with such high velocities(18).

The emission must therefore be from some form of non thermal process,either synchrotron or inverse Compton emission,as seen in wind nebulae around young pulsars.In either case, the x-ray emission requires a population of relativistic particles in the pulsar wind.Thus,the detection of a distinct x-ray tail provides direct evidence that millisecond pulsars lose their rotational energy through relativistic winds.

It is highly unlikely that diffuse shock acceleration(19)is the mechanism that generates the emitting particle population in this nebula,because it is dif?cult to accelerate particles in a relativistic?ow through this mechanism(20).As in other pulsar wind nebulae,the acceleration presumably takes place through some other mechanism possibly involving the role of heavy ions(21)

Because the pulsar and the companion are separated by only1.5×1011cm,the intrabinary shock,formed where the pulsar and companion winds interact,will be located in a strong mag-netic?eld(much stronger than that at the shock in the Crab nebula).This intrabinary shock is therefore a potential source of unresolved synchrotron emission at the location of the pulsar(22) and can be used to determine the wind characteristics at the shock front.The x-ray luminosity in the shock is dependent on both the post shock magnetic?eld strength andσ(8).We therefore consider two possibilities for the wind composition of PSR B1957+20:either dominated by ki-netic energy(σ=0.003as seen in the Crab nebula)or magnetically dominated(σ?1).We can describe the process by which the spin-down energy of the pulsar is converted into x-ray emission at the intrabinary shock by(9):

f b?εLε=f rad fγf geom˙E(1)

where Lεis the spectral x-ray luminosity in a band of width?ε≈1keV centered on a pho-

ton energyε=1keV.The geometric factor f geom de?nes how much of the wind interacts with the companion,fγis the fraction of the intercepted spin-down energy?ux that goes into accelerating electrons with Lorentz factorγcorresponding toε=1keV,and f rad is the ra-diative ef?ciency of the corresponding synchrotron emission.The fraction of the unresolved x-ray emission that is due to synchrotron emission produced at the intrabinary shock is f b,the remainder presumably being emission produced by the pulsar itself.

Table1:Properties of the PSR B1957+20system(6),and comparison with the Crab nebula and its central pulsar(23).The fraction of the wind of PSR B1957+20that interacts with the companion star wind is f geom>～r2e/(4a2)≈0.02,where a is the orbital separation.This fraction is a lower limit,because the pulsar wind is most likely focused into the equatorial plane(24,25) which is also probably the orbital plane of the binary system(2).The companion star would therefore intercept more of the spin-down energy from the pulsar than if it has a spherical wind.

Spin period(ms) 1.633.5

Surface magnetic?eld(108G) 1.43.8×104

Distance(kpc) 1.52

Age(yr)>2×109948

Distance to shock(cm)～1.5×1011～3×1017 The?ow immediately downstream of the intrabinary shock is expected to undergo Doppler

boosting as it passes around the companion(22).Thus,one expects the x-ray emission at orbital phases before and after eclipse to be enhanced by up to a factor2.2depending on the?ow speed and the degree of absorption and/or scattering by the companion wind(22).In contrast,the x-ray emission at the eclipse(orbital phase0.25)may be reduced because of obscuration of the shock by the companion star.If we can measure these variations then we can determine f b.The lowest and highest count rates in the folded light curve(Fig3.)are during and immediately after eclipse,respectively,with comparatively low probability of this variation being by chance (see caption of Fig3.).If this apparent orbital modulation is genuine and corresponds to a modulation of the x-ray?ux from the intrabinary shock by a factor of2.2,then we?nd that f b～0.5.

The post shock magnetic?eld strengths(8)corresponding to our two limiting values ofσ(0.003and?1)are listed in Table1;the corresponding Lorentz factors of the synchrotron emitting relativistic electrons for both cases areγ=2.4×105(ε/B)1/2≈105atε=1keV, where B is the post shock magnetic?eld strength.For each case the emitting region is taken to be the radius of the radio-eclipse region(6)r e=5×1010cm;forσ=0.003the?ow speed is v flow=c/3,whereas forσ?1we expect v flow=c(8).The corresponding residence times are given by t flow=v flow/r e(Table1),and the radiative lifetimes of the emitting electrons are t rad=5.1×108/(γB2).We can thus compute the radiative ef?ciency(26)in each case as f rad=(1+t rad/t flow)?1.With f geom>0.02(Table1)and f b～0.5,we can infer that fγ<0.09(σ=0.003)or fγ<0.02(σ?1)for Lorentz factorγ～105.

We directly compared the properties of the shocked wind of PSR B1957+20with those derived for the Crab nebula with the same Lorentz factor and?nd that fγ≈0.04for electrons withγ=105in the Crab,which,regardless of the assumed value ofσ,is similar to that which we derived for PSR B1957+20.The available evidence therefore suggests that despite being subject to a prolonged evolutionary process that has altered its magnetic?eld by many orders

of magnitude and having a shock that occurs much closer to the pulsar,this millisecond pulsar generates a wind for which the ef?ciency of relativistic particle acceleration in this post shock ?ow is similar to that seen for the winds of young pulsars.New Chandra data on this and other pulsars(27,28,29)are providing the?rst detailed observational input into studies of relativistic ?ows,particle acceleration,and magnetohydrodynamic shocks.

References and Notes

1.M.A.Alpar,A.F.Cheng,M.A.Ruderman,J.Shaham,Nature300,728(1982).

2.D.Bhattacharya,E.P.J.van den Heuvel,Phys.Rep.203,1(1991).

3.A.S.Fruchter,D.R.Stinebring,J.H.Taylor,Nature333,237(1988).

4.M.Toscano,et al.,Mon.Not.R.Astron.Soc.307,925(1999).

5.A.S.Fruchter,J.Bookbinder,C.Bailyn,Astrophys.J.443,L21(1995).

6.Z.Arzoumanian,A.S.Fruchter,J.H.Taylor,Astrophys.J.426,L85(1994).

7.S.R.Kulkarni,J.J.Hester,Nature335,801(1988).

8.C.F.Kennel,F.V.Coroniti,Astrophys.J.283,694(1984).

9.S.R.Kulkarni,E.S.Phinney,C.R.Evans,G.Hasinger,Nature359,300(1992).

10.The observation was scheduled so that two eclipses were observed during the43ksec.

11.E.H?g,et al.,Astron.Astrophys.355,L27(2000).

12.J.H.Taylor,J.M.Cordes,Astrophys.J.411,674–684(1993).

13.Q.D.Wang,Z.-Y.Li,M.C.Begelman,Nature364,127(1993).

14.N.Bucciantini,Astr.Astrophys.387,1066(2002).

15.J.Arons,ASP Conf.Ser.271:Neutron Stars in Supernova Remnants,P.O.Slane,B.M.

Gaensler,eds.(Astronomical Society of the Paci?c,San Francisco,2002),pp.71+.

16.O.C.de Jager,A.K.Harding,Astrophys.J.396,161(1992).

17.S.Chatterjee,J.M.Cordes,Astrophys.J.575,407(2002).

18.We note that this spectrum differs from that used in calculating the?ux in the tail region

discussed in the caption to Figure1.

19.R.Blandford,D.Eichler,Phys.Rep.154,1+(1987).

20.J.Arons,M.Tavani,Astrophys.J.Supp.Series90,797(1994).

21.M.Hoshino,J.Arons,Y.A.Gallant,https://www.wendangku.net/doc/b014468526.html,ngdon,Astrophys.J.390,454(1992).

22.J.Arons,M.Tavani,Astrophys.J.403,249(1993).

23.J.H.Taylor,R.N.Manchester,A.G.Lyne,Astrophys.J.Supp.Series88,529(1993).

24.J.J.Hester,et al.,Astrophys.J.448,240(1995).

25.F.C.Michel,Astrophys.J.431,397(1994).

26.V.L.Ginzburg,S.I.Syrovatskii,Ann.Rev.Astr.Ap.3,297(1965).

27.B.M.Gaensler,et al.,Astrophys.J.569,878(2002).

28.M.C.Weisskopf,et al.,Astrophys.J.536,L81(2000).

29.D.J.Helfand,E.V.Gotthelf,J.P.Halpern,Astrophys.J.556,380(2001).

30.We thank Jon Arons for useful discussions and Heath Jones for providing the Hαimage.

This work was supported in part by NASA through a Chandra X-ray Observatory Guest Observer grant,and by the Netherlands Organisation for Scienti?c Research(NWO).

1

in the energy range0.3–10.0keV.The source at(J2000)RA19h59m36.s75±0.s01;Dec+20?48′15.′′0±0.′′1 is coincident with the proper motion-corrected radio timing position of PSR B1957+20,RA 19h59m36.s75788±0.s00005;Dec+20?48′14.′′8482±0.′′0006(6).Four of the next brightest x-ray sources in the?eld are circled and the source labeled1is coincident to better than0.′′1 with the Tycho II star1628-01136-1(11).In an aperture of radius1.′′5centered on the pulsar position we detect a background corrected total of370±20counts from the pulsar in the en-ergy range0.3–10.0keV.Binning the data in energy such that each bin contains a minimum of 30counts,we?t an absorbed power law resulting in a best?t with hydrogen column density N H=(1.8±0.7)×1021cm?2,photon indexΓ=1.9±0.5,and an unabsorbed?ux density F x,c=6×10?14ergs s?1cm?2(0.5–7.0keV)corresponding to an isotropic x-ray luminosity of L x,c=4πD2F x,c=1.6×1031D1.5ergs s?1(0.5–7.0keV),where D1.5=D/(1.5kpc)is the pulsar distance as derived from its dispersion measure(12).Counts in the tail region were summed in a16′′×6′′box enclosing the tail and aligned with the proper motion direction.After background correction,we detect a total of82±9counts from the tail region.Again?tting an absorbed power law and assuming that the N H andΓare similar to the values above we derive an unabsorbed?ux density F x,t=9×10?15ergs s?1cm?2(0.5–7.0keV).

Figure2:The same Chandra data as shown in Figure1,smoothed to a resolution of5′′(con-tours)and overlaid on an Hαimage obtained from Taurus Tunable Filter service mode obser-vations on the Anglo Australian Telescope on2000Aug3.The images were aligned to within ～0.′′2through use of USNO2.0stars.The x-ray tail is located well inside the boundaries of the Hαemission and also close to its symmetry axis.The x-ray contour levels are shown at0.9, 1.2,5.3,35.0and78.8%of the peak x-ray surface brightness.The optical residuals correspond to incompletely subtracted stars.No optical counterpart was found to the x-ray source associ-ated with the contours located in the northeastern corner of the?eld on a Digitized Sky Survey image.

00.002

0.004

0.006

0.0080.01

0.01200.20.40.60.81 1.2 1.4 1.6 1.82

C o u n t r a t e (c t s /s )Binary phase

Figure 3:The light curve of the x-ray emission (0.3–10.0keV)from the point source associated with PSR B1957+20folded through use of the known orbital ephemeris (D.J.Nice,private communication,2001).The errorbars correspond to one-sigma Poissonian errors.Radio eclipse occurs at orbital phase 0.25.The mean count rate,excluding phases in the range 0.15–0.35where variations in the x-ray ?ux are expected,is (8.4±0.8)cts ksec ?1.Orbital phase bins 0.25and 0.35each correspond to a total observing time of 6600s,and thus we expect to measure 55counts in each of these phase https://www.wendangku.net/doc/b014468526.html,ing Poisson statistics,the 76counts detected at phase 0.35deviate from the expected 55counts with 99%con?dence.Scattering and/or absorption of the x-ray emission could reduce the degree of modulation at either phase 0.15or 0.35.Taking into account either possibility,the chance probability for the observed variation at phase 0.35is 2%.A decrease in the count rate is predicted at orbital phase 0.25and the 44counts we detect in this phase bin have a 4%chance probability of being drawn from a steady ?ux distribution.