Pulsar glitches and their impact on neutron-star astrophysics
aa r X i v : . [ a s t r o - ph . H E ] J a n Pulsar Astrophysics - The Next 50 YearsProceedings IAU Symposium No. 337, 2017P. Weltevrede, B.B.P. Perera, L. Levin Preston & S. Sanidas, eds. c (cid:13) Pulsar Glitches
R. N. Manchester
CSIRO Astronomy and Space Science, Sydney, Australiaemail: [email protected]
Abstract.
The first known pulsar glitch was discovered in the Vela pulsar at both Parkes andGoldstone in March 1969. Since then the number of known glitches has grown enormously, withmore than 520 glitches now known in more than 180 pulsars. Details of glitch parameters andpost-glitch recoveries are described and some implications for the physics of neutron stars arediscussed.
Keywords. pulsars: general, stars: neutron
1. The discovery of glitches and the starquake model
In late 1968 and early 1969, I was at Parkes and, among other things, helping Radhakr-ishnan with the observations of the Vela pulsar that ultimately led to the “rotating vectormodel” for pulsar polarisation (Radhakrishnan & Cooke 1969). In mid-March, 1969, weset up the signal-averager to fold the Vela pulsar data at the predicted topocentric periodand noticed that the pulse was slowly drifting backwards on the screen, indicating thatthe folding period was not quite correct. After exhaustively checking the equipment andobserving other pulsars, we concluded that everything was working correctly and, con-sequently, that the pulsar period P was not as predicted. The implied period decrease∆ P was 196 ns, corresponding to a relative change ∆ ν/ν = − ∆ P/P ∼ . × − ,where the pulsar rotation frequency ν = 1 /P . We contacted Paul Reichley and GeorgeDowns, whom we knew were timing the Vela pulsar using the Goldstone antenna of theJet Propulsion Laboratory (JPL) in California, and confirmed that they also had seenthe glitch (Figure 1). † Back-to-back papers reporting the discovery were published inthe April 19 Nature (Radhakrishnan & Manchester 1969; Reichley & Downs 1969). TheJPL observations limited the glitch epoch to between February 24 and March 3, and alsorevealed a change in the slow-down rate ∆ ˙ ν/ ˙ ν = ∆ ˙ P / ˙ P ∼ − . Both groups suggesteda sudden decrease in the neutron-star moment of inertia, which could account for thechanges in both ν and ˙ ν . The required effective change in the radius of the neutron starwas about 1 cm.Within a few months, Baym et al. (1969) had refined this “starquake” model, suggest-ing that the change in moment of inertia was due to relaxation of the neutron-star crustto the current equilibrium oblateness, which of course changes because of the gradualslow-down in rotation. They also predicted that the observed increase in | ˙ ν | would de-cay on a timescale of years because of the weak frictional coupling of a more rapidlyrotating superfluid interior, earlier predicted to exist by Russian theorists (Migdal 1960;Ginzburg & Kirzhnits 1965), and the neutron-star crust to which the emission beams arelocked.A prediction of this model was that it would be at least 300 years before the stressesdue to the oblateness differential built up sufficiently to cause another starquake. The † The term “glitch” was not initally used to describe these events. The first published use ofthe term appears to be by Rees et al. (1971). ν/ν ∼ . × − (Reichley & Downs1971). This of course ruled out relaxation of crustal oblateness as a mechanism forthe glitch trigger. Alternative models quickly followed, with “corequakes” suggested byPines et al. (1972) and the sudden unpinning of interior superfluid vortices, with a con-sequent transfer of angular momentum to the crust, first suggested by Anderson & Itoh(1975). As will be discussed in Section 3 below, the latter idea forms the basis for mostsubsequent interpretations of the glitch phenomenon.Both Radhakrishnan & Manchester (1969) and Reichley & Downs (1969) pointed outin their concluding remarks that glitches could be expected in the Crab pulsar period.Sure enough, about six months later, Boynton et al. (1969) and Richards et al. (1969)announced the discovery of a glitch in the Crab pulsar period. The relative glitch size,∆ ν/ν ∼ × − , was about 300 times smaller than for the Vela glitches suggesting adifferent mechanism. Observations over the next few years at Jodrell Bank and opticalobservatories (e.g., Lohsen 1981; Lyne et al. 1993) revealed several glitches, includinglarger ones in 1975 and 1989. These observations also showed that the post-glitch be-haviour in the Crab pulsar was quite different to that for Vela, being dominated by apersistent increase in the slow-down rate | ˙ ν | .
2. The glitch population
Tables of observed glitches are maintained by Jodrell Bank Observatory (JBO) ∗ andas part of the ATNF Pulsar Catalogue ‡ . While these two tables broadly contain the sameinformation, there is some information in one, but not the other. In particular, the JBOtable contains details of about 80 otherwise unpublished glitches. Collating the data fromthe two tables gives a total of 520 known glitches in 180 different pulsars. Figure 2 showsthe distribution of glitching pulsars on the P – ˙ P diagram. This figure shows that glitchesin young pulsars, including magnetars, are generally large with ∆ ν/ν ∼ − . However,the youngest pulsars, e.g., the Crab pulsar, PSR J0537 − − ∗ ‡ Figure 1.
Changes in the Vela pulsar period in late 1968 and early 1969 showing the first detectionsof a pulsar glitch. The left panel shows the Parkes observations (Radhakrishnan & Manchester1969) and the right panel shows the JPL Goldstone observations (Reichley & Downs 1969). ulsar glitches and their impact on neutron-star astrophysics ν/ν ∼ − or 10 − . The mostfrequently glitching pulsar is PSR J0537 − −
24A and J0613 − ν/ν ∼ − (Cognard & Backer2004; McKee et al. 2016). It is interesting that the Hulse – Taylor binary pulsar, PSRB1913+16, also had a small glitch in 2003 of about the same magnitude (Weisberg et al.2010).Large glitches are relatively common in magnetars. While their magnitudes are similarto those in other young pulsars, they have a number of distinguishing features. Forexample, they are sometimes accompanied by radiative changes, either X-ray bursts orassociated changes in the pulse profile (see, e.g., Dib & Kaspi 2014). Such X-ray burstsand profile changes are very common in magnetars and are generally accompanied bytiming irregularities, but only a small proportion of them are associated with glitches.Since magnetar glitch properties are broadly similar to those in other young pulsars, itseems plausible that the glitch mechanism is the same or similar, i.e., related to changesin the superfluid interior of the star. In this case, the radiative associations suggest someconnection between the stellar interior and the magnetosphere of the star.However, Archibald et al. (2013) reported the detection of an “anti-glitch”, i.e., anabrupt spin-down, in the period of the magnetar 1E 2259+586 (PSR J2301+5852) ofrelative magnitude about 3 . × − . Several large glitches, one with ∆ ν/ν ∼ . × − ,have also been seen in this pulsar (Dib et al. 2009). The anti-glitch was associated with Figure 2.
Distribution of the 180 glitching pulsars on the P – ˙ P diagram. The symbol sizeis proportional to the logarithm of the largest fractional glitch size ∆ ν/ν observed for eachpulsar. Glitching pulsars and AXPs/SGRs (magnetars) and non-glitching pulsars are markedwith different symbols as labelled. R. N. Manchestera short-duration hard X-ray burst and an increase in the soft X-ray flux which decayedover 100 days or so. These properties suggest that the anti-glitch was magnetospheric inorigin.Although glitches in normal young pulsars are generally not associated with radia-tive changes, glitches observed in the very young pulsar PSR J1119 − ν/ν ∼ . × − ), intermittent strong pulses and a second profile component wereobserved. An even larger subsequent glitch (∆ ν/ν ∼ . × − ) was observed byArchibald et al. (2016) to be accompanied by X-ray bursts and X-ray pulsations. It seemsas though this pulsar is half-way to being a magnetar.
3. Glitch properties and their interpretation
Glitch activity in pulsars can be quantified by the relation A g = 1 T Σ(∆ ν g ) ν (3.1)where T is the total data span for glitch monitoring and Σ(∆ ν g ) is the sum of all observedglitch frequency jumps. These jumps reverse some fraction of the regular slow-down due toelectromagnetic and wind torques. Many studies (see, e.g., Espinoza et al. 2011; Yu et al.2013) have shown that for pulsars with characteristic ages τ c = P/ (2 ˙ P ) between about10 and 10 years, about 1% of the slowdown is reversed by glitches. For the Crab andother very young pulsars, glitches have about two orders of magnitude less effect on theslow-down. In the two-component superfluid models (e.g., Alpar et al. 1981), the glitchresults from the sudden unpinning and then repinning of vortex lines transferring angularmomentum from the more rapidly spinning interior superfluid to the crust. The momentof intertia of the pinning/unpinning superfluid I s is related to the glitch activity by the“coupling parameter” G = 2 τ c A g = ˙ ν g | ˙ ν | ∼ I s I (3.2)where I is the total moment of inertia of the neutron star.Many different types of post-glitch behaviour are observed. In some pulsars, most orall of the initial frequency jump decays, whereas in other cases the glitch is like a stepjump in frequency with little or no change in ˙ ν (e.g., Espinoza et al. 2011; Yu et al. 2013).Figure 3 shows the observed long-term variations in ˙ ν for the Crab and a number of otheryoung pulsars. Glitches are marked by a sudden decrease in ˙ ν . The fractional increasein slow-down rate | ˙ ν | is typically about 1% although for some pulsars the increase ismuch smaller, even not detectable. For most pulsars, much of this initial increase decaysexponentially on a timescale of 10 – 100 days. For the Crab pulsar, about 90% of theincrease quickly decays, but the remaining 10% persists as a long-term increase in theslow-down rate. For large glitches in other pulsars, typically about half of the initialincrease decays exponentially. Following that, there is a basically linear increase in ˙ ν until the next glitch.The simple two-component model of Baym et al. (1969) cannot account for these differ-ent post-glitch behaviours. In a series of papers, Ali Alpar and his colleagues have devel-oped this model with different regions within the neutron star having different propertiesto account for the different post-glitch behaviours (e.g., Alpar et al. 1981, 1993, 1996).For example, in regions with weakly pinned vortices, vortex creep can occur, whereasin strongly pinned regions, there is no creep following the repinning of vortices after aglitch. Weakly pinned regions have a linear dynamical response and can give the observed ulsar glitches and their impact on neutron-star astrophysics ν . More strongly pinned regions have a non-linear dynamicalresponse that can result in a long-term linear increase in ˙ ν as observed for Vela and otheryoung pulsars.While the Alpar et al. models have been broadly successful in accounting for the proper-ties of pulsar glitches, they depend on many assumptions about poorly known propertiesof neutron star interiors. Various authors have proposed alternative views about some ofthese assumptions. For example, Chamel (2012) has argued that “entrainment” of theneutron superfluid by the crystalline structure of the crust greatly reduces its mobil-ity. Consequently, unpinning of the crustal superfluid is insufficient to account for largeglitches, and other mechanisms, e.g., unpinning of core superfluid neutrons, are required.However, in a recent paper, Watanabe & Pethick (2017) argue that entrainment is nota significant issue and there is no need to invoke core superfluid. In another recent pa-per, Link (2014) has argued that the “linear-response” regime invoked by Alpar et al. isstrongly suppressed by a high vortex activation energy. If this is the case, the interpreta-tion of the exponential post-glitch decays invoked by, e.g., Alpar et al. (1993) would notbe viable.Various authors have used observed glitch properties as a probe to investigate themass of neutron stars. For example, Pizzochero et al. (2017) used the largest observedglitch in a given pulsar to limit the neutron-star mass on the assumption that therewas complete unpinning of vortices at a glitch, transferring the entire excess angularmomentum to the crust. They gave mass estimates for all pulsars in which at least twolarge glitches had been observed. For Vela and PSR J0537 − . ± .
08 M ⊙ and 1 . ± .
06 M ⊙ , respectively. In contrast,Ho et al. (2017) used detailed modelling of neutron-star thermal evolution combined withnuclear equation-of-state and superfluid models to interpret the size and rate of observedglitches as a function of the neutron-star mass. For the same two pulsars, Ho et al. Figure 3.
Variations of spin-down rate ˙ ν for the Crab pulsar (left, Lyne et al. 2015) and a sampleof young pulsars (right, Espinoza et al. 2017)). For the Crab pulsar, the lower subpanel shows the˙ ν variations after subtraction of the linear trend evident in the upper subpanel. R. N. Manchester(2017) obtained mass estimates of 1 . ± .
04 M ⊙ and 1 . ± .
04 M ⊙ , respectively.The differences between these derived masses clearly indicate that uncertainties in thephysical properties of neutron-star interiors lead to large systematic offsets in estimatedneutron-star masses. A positive aspect of this is that pulsar glitches can provide usefulinput into the determination of these properties. Acknowlegements
I thank the organisers of this Symposium for inviting me to review this topic and mycolleagues with whom I have had many fruitful discussions and collaborations over thepast five decades or so.
References
Alpar, M. A., Anderson, P. W., Pines, D., & Shaham, J. 1981,
Astrophys. J. , 249, L29Alpar, M. A. & Baykal, A. 2006,
MNRAS , 372, 489Alpar, M. A., Chau, H. F., Cheng, K. S., & Pines, D. 1993,
Astrophys. J. , 409, 345—. 1996,
Astrophys. J. , 459, 706Anderson, P. W. & Itoh, N. 1975,
Nature , 256, 25Archibald, R. F., Kaspi, V. M., Ng, C.-Y., et al. 2013,
Nature , 497, 591Archibald, R. F., Kaspi, V. M., Tendulkar, S. P., & Scholz, P. 2016,
Astrophys. J. , 829, L21Baym, G., Pethick, C., Pines, D., & Ruderman, M. 1969,
Nature , 224, 872Boynton, P. E., Groth, E. J., Partridge, R. B., & Wilkinson, D. T. 1969,
IAU Circ.
Phys. Rev. C , 85, 035801Cognard, I. & Backer, D. C. 2004,
Astrophys. J. , 612, L125Dib, R. & Kaspi, V. M. 2014,
Astrophys. J. , 784, 37Dib, R., Kaspi, V. M., & Gavriil, F. P. 2009,
Astrophys. J. , 702, 614Espinoza, C. M., Lyne, A. G., & Stappers, B. W. 2017,
MNRAS , 466, 147Espinoza, C. M., Lyne, A. G., Stappers, B. W., & Kramer, M. 2011,
MNRAS , 414, 1679Ginzburg, V. L. & Kirzhnits, D. A. 1965,
Soviet Physics JEPT , 20, 1346Ho, W. C. G., Espinoza, C. M., Antonopoulou, D., & Andersson, N. 2017, , 010805Link, B. 2014,
Astrophys. J. , 789, 141Link, B., Franco, L. M., & Epstein, R. I. 1998,
Astrophys. J. , 508, 838Lohsen, E. H. G. 1981,
Astron. Astrophys. Suppl. , 44, 1Lyne, A. G., Jordan, C. A., Graham-Smith, F., Espinoza, C. M., et al. 2015,
MNRAS , 446, 857Lyne, A. G., Pritchard, R. S., & Smith, F. G. 1993,
MNRAS , 265, 1003Marshall, F. E., Gotthelf, E. V., Middleditch, J., et al. 2004,
Astrophys. J. , 603, 682McKee, J. W., Janssen, G. H., Stappers, B. W., et al. 2016,
MNRAS , 461, 2809Migdal, A. B. 1960,
Soviet Physics JEPT , 10, 176Pines, D., Shaham, J., & Ruderman, M. A. 1972,
Nature Phys. Sci. , 237, 83Pizzochero, P. M., Antonelli, M., Haskell, B., & Seveso, S. 2017,
Nature Astronomy , 1, 0134Radhakrishnan, V. & Cooke, D. J. 1969,
Astrophys. Lett. , 3, 225Radhakrishnan, V. & Manchester, R. N. 1969,
Nature , 222, 228Reichley, P. E. & Downs, G. S. 1969,
Nature , 222, 229—. 1971,
Nature Phys. Sci. , 234, 48Richards, D. W., Pettengill, G. H., Roberts, J. A., et al. 1969,
IAU Circ.
Nature , 229, 395Watanabe, G. & Pethick, C. J. 2017,
Phys. Rev. Lett. , 119, 062701Weisberg, J. M., Nice, D. J., & Taylor, J. H. 2010,
Astrophys. J. , 722, 1030Weltevrede, P., Johnston, S., & Espinoza, C. M. 2011,
MNRAS , 411, 1917Yu, M., Manchester, R. N., Hobbs, G., et al. 2013,