Pulsar Glitch

A clock that hiccups

A pulsar is a rapidly rotating neutron star whose beams of radio emission sweep across Earth once per rotation, like a lighthouse. The pulses arrive with such regularity that pulsars rival atomic clocks: their rotation slows down under magnetic braking at a rate so smooth and predictable that timing models can forecast the arrival of individual pulses years into the future to within microseconds. This staggering steadiness is what makes the exceptions so revealing.

Every so often the clock hiccups. The rotation frequency, instead of continuing its slow decline, jumps upward — abruptly, in seconds or less — and then slowly drifts back over the following days, weeks, and years. This sudden spin-up is a pulsar glitch. The fractional jump in frequency, written Δν/ν, ranges from about 10⁻⁹ for the smallest detectable events to a few × 10⁻⁶ for the largest. That sounds tiny, and it is: a 10⁻⁶ glitch on an 11 Hz pulsar is a frequency jump of about 11 microhertz. But it is also a violation of the otherwise perfect spin-down, and it is not random noise. The glitch is a window — one of the only ones we have — into the physics of matter at densities beyond anything reachable in a laboratory.

How a glitch works: a superfluid that refuses to slow down

The standard model for large, recurrent glitches lives inside the star. A neutron star is not a uniform ball. Beneath a thin solid outer crust, in a region a few hundred metres deep called the inner crust, the density is high enough that neutrons drip out of nuclei and form a quantum fluid. Below about 10⁹ K — and the interiors of most observed pulsars are far colder than that — these neutrons pair up and become a superfluid: a fluid with exactly zero viscosity.

A superfluid cannot rotate the way a normal fluid does. Quantum mechanics forbids it from spinning as a smooth solid body; instead it carries all of its rotation in a dense array of microscopic quantized vortices, each a tiny tornado of circulating superfluid threading the star. The areal density of vortices is proportional to the rotation rate, so to spin down, the superfluid must let vortices migrate outward and disappear at the boundary. Here is the catch: those vortices pin to the nuclei of the crustal lattice — they get stuck on the nuclear clusters like a flag snagged on a fence.

Meanwhile, the visible crust — the part we time via the radio beam, electromagnetically coupled to the rest of the charged star — is braked by its magnetic dipole and steadily slows down. The pinned superfluid cannot follow, because its vortices are stuck. So a rotational lag develops: the superfluid keeps spinning at its old, faster rate while the crust falls behind. The lag stores angular momentum, like winding a spring.

When the lag grows large enough that the Magnus force on the vortices exceeds the maximum pinning force, the vortices break free. And because vortices interact, the unpinning is not gentle — it is a catastrophic avalanche. A vast population of vortices unpins almost simultaneously, migrates outward, and dumps its stored angular momentum into the crust. The crust speeds up. That is the glitch: the spin-up you observe is the superfluid suddenly sharing its hoarded rotation with the part of the star you can see.

Worked example: the Vela pulsar

The Vela pulsar, PSR B0833−45, is the textbook case. Born in a supernova about 11,000 years ago, it spins at ν ≈ 11.19 Hz (a period of about 89 ms) and is one of the most intensively monitored pulsars in the southern sky. Since its discovery in 1968 it has produced more than twenty large glitches, with a characteristic fractional size Δν/ν ≈ 2 × 10⁻⁶ and a quasi-periodic recurrence of roughly one large glitch every three years.

Take a canonical Vela glitch and put numbers on it:

Spin frequency:        ν   ≈ 11.19 Hz
Fractional spin-up:    Δν/ν ≈ 2 × 10⁻⁶
Frequency jump:        Δν  = ν × (Δν/ν) ≈ 11.19 × 2×10⁻⁶ ≈ 2.2 × 10⁻⁵ Hz (22 µHz)
Spin-up timescale:     τ_rise < 1 minute   (Vela 2016, caught live)
Recurrence:            ≈ 3 years between large glitches

Now estimate the angular momentum involved. A neutron star has a moment of inertia of order I ≈ 10⁴⁵ g·cm² = 10³⁸ kg·m². The angular momentum delivered to the crust in the glitch is

ΔL = I_crust × Δω = I_crust × 2π Δν
   ≈ 10³⁸ × 2π × 2.2×10⁻⁵
   ≈ 1.4 × 10³⁴  kg·m²/s

Where did that angular momentum come from? From the superfluid reservoir that had been lagging behind. If a reservoir of moment of inertia I_sf carries the lag, conservation of total angular momentum during the avalanche gives I_sf Δω_sf ≈ I_crust Δω_crust. Because the reservoir is only a few percent of the total moment of inertia, it must have built a substantial lag over the preceding three years of spin-down to accumulate enough stored angular momentum. The Vela 2016 event, caught with sub-second cadence, showed the spin-up complete in under a minute, an unresolved instantaneous jump within the data, followed by a brief overshoot before relaxation — exactly the signature of vortices avalanching and then the components re-equilibrating.

Postglitch recovery: reading the relaxation

The spin-up is only half the story. After the jump, the frequency does not stay put. It partially decays back toward — but never fully to — the pre-glitch extrapolation, over a hierarchy of timescales. A typical recovery is modelled as a sum of exponentials plus a long-term linear term:

ν(t) = ν_0(t) + Δν_p + Σ_i Δν_i exp(−t/τ_i) − Δν̇ · t

  Δν_p   permanent frequency step  (never recovers)
  Δν_i   recovering amplitudes with timescales τ_i (hours → weeks)
  Δν̇     persistent change in spin-down rate

The fraction of the jump that recovers, often denoted Q (the "healing parameter"), varies from near 0 (Crab glitches barely recover, leaving an almost permanent step) to a large fraction for some events. The recovery timescales τ_i — ranging from hours to weeks to a long linear term over months — encode how strongly the various superfluid components are coupled to the crust through mutual friction. Weakly coupled components share angular momentum slowly, giving long τ; strongly coupled ones recover fast. Fitting the recovery curve is therefore a way to weigh and time-tag the internal components of a star you can never see inside.

Variants and regimes across the pulsar population

Glitches are not one phenomenon but a family, and where a pulsar sits in age and magnetic field shapes what it does.

• Vela-type giant glitches. Middle-aged pulsars (~10⁴–10⁵ yr) like Vela produce large (Δν/ν ≈ 10⁻⁶), quasi-periodic glitches. These are the canonical superfluid-reservoir events and the ones the unpinning model was built to explain.
• Crab-type small glitches. The young Crab pulsar (PSR B0531+21) glitches more often but with small sizes (Δν/ν ≈ 10⁻⁹–10⁻⁸) and a characteristic persistent increase in spin-down rate after each event. Crab glitches barely recover, hinting at different reservoir physics in a hotter, younger star.
• Millisecond pulsars. Old, recycled millisecond pulsars glitch extremely rarely and weakly; their interiors are warm-equilibrated and only modestly out of corotation, so the lag rarely builds to threshold. This is why they remain the best clocks for pulsar timing arrays.
• Magnetars. In magnetars, the enormous magnetic field drives crustal failure and magnetospheric activity. They show large glitches often correlated with X-ray outbursts, and, uniquely, occasional anti-glitches — sudden spin-downs — which pure vortex unpinning cannot easily produce.
• Starquake glitches. The original 1969 mechanism: spin-down shrinks the centrifugal bulge, crustal stress builds, and a crack reduces the moment of inertia, spinning the star up. Quakes can supply small, rare glitches in slow pulsars but cannot recharge fast enough to power Vela's giant events every three years.

Quantitative analysis: the glitch activity constraint

One of the most powerful uses of glitches is a statistical bookkeeping argument. Over a long monitoring baseline, sum up all the frequency the glitches add back and compare it to all the frequency the star lost to spin-down. The glitch "activity parameter" is

A_g = (1/T) Σ (Δν/ν)        (cumulative fractional spin-up per unit time)

The two-component picture says the angular momentum dumped in glitches was previously stored as lag in the superfluid reservoir. Requiring the reservoir to supply the observed long-term glitch activity sets a lower bound on the fractional moment of inertia of the participating superfluid:

I_sf / I  ≥  A_g × (ν / |ν̇|)        (a "coupling" / moment-of-inertia bound)

For Vela this works out to roughly I_sf/I ≳ 0.016, i.e. a few percent — historically a beautiful match to the moment of inertia attributed to the inner-crust superfluid. The story was tidy until detailed calculations of crustal entrainment intervened: the dripped neutrons are Bragg-scattered by the lattice, raising their effective mass by a large factor (up to ~4–5) and reducing the mobile superfluid available to drive glitches. That made the inner crust alone insufficient by a factor of several — the "entrainment crisis" — and forced the conclusion that the core superfluid must also participate, or that the equation of state must give thicker crusts and lower neutron-star masses. Glitch bookkeeping thus directly constrains the dense-matter equation of state, which is why glitches matter far beyond timing curiosities.

Observational status and applications

More than 600 glitches have now been catalogued across some 200+ pulsars (the Jodrell Bank and ATNF glitch catalogues are the standard references). Wide-field monitoring at facilities like the MeerKAT, Parkes (Murriyang), and the FAST telescope keeps long timing baselines on the most active glitchers. The headline observational achievements include:

• Catching a glitch live. The 2016 Vela glitch was observed pulse-by-pulse with single-dish timing, constraining the spin-up rise to under a minute and revealing a transient overshoot — direct evidence for distinct superfluid components recoupling on different timescales.
• Weighing the interior. Population-wide glitch activity, combined with entrainment theory, places quantitative bounds on the superfluid moment of inertia and feeds into multimessenger equation-of-state constraints alongside NICER radius measurements and gravitational-wave tidal deformabilities from neutron-star mergers.
• Pulsar timing array hygiene. Glitches and post-glitch recovery are a noise source that must be modelled and excised when using millisecond pulsars to search for nanohertz gravitational waves. Knowing which pulsars glitch, and how, protects the array.
• Glitch precursors and emission changes. A handful of events show changes in pulse shape or radio brightness around the glitch epoch, tentatively linking interior dynamics to magnetospheric behaviour — an active research frontier.

Famous and instructive glitches

Pulsar	Spin freq ν (Hz)	Typical Δν/ν	Recurrence	Character
Vela (B0833−45)	11.19	~ 2 × 10⁻⁶	~ 3 yr	Giant, quasi-periodic — the prototype
Crab (B0531+21)	29.6	~ 10⁻⁹–10⁻⁸	~ 1–2 yr	Small, frequent, persistent spin-down step
PSR J0537−6910	62	~ 10⁻⁷–10⁻⁶	~ 100 days	Most frequent giant glitcher known
PSR B1737−30	5.0	~ 10⁻⁷	variable	Many medium glitches over decades
SGR 1806−20 (magnetar)	0.13	~ 10⁻⁶	burst-linked	Glitch tied to X-ray outburst
1E 2259+586 (magnetar)	0.14	~ −10⁻⁶ (anti)	rare	Anti-glitch: a sudden spin-down
Typical MSP	100–700	≲ 10⁻¹⁰	essentially never	Quiet; ideal timing-array clock

The bottom rows are deliberately contrasting cases. PSR J0537−6910 in the Large Magellanic Cloud glitches so often and so quasi-periodically that the next glitch can be roughly forecast from the time since the last — a remarkable predictability. At the opposite extreme, millisecond pulsars almost never glitch, which is precisely why they are the workhorses of nanohertz gravitational-wave detection.

Common pitfalls and misconceptions

• "A glitch means the pulsar gains energy from nowhere." No — total angular momentum is conserved. The crust speeds up because the superfluid, which had been spinning faster all along, hands over angular momentum it had been hoarding. Nothing is created; an internal reservoir is tapped.
• "The spin-up is the whole story." The recovery is at least as informative. The relaxation timescales and the fraction that heals (Q) carry the physics of internal coupling; ignoring them throws away most of the signal.
• "Glitches are random glitches in the instrument." The word is unfortunate. A pulsar glitch is a real, astrophysical, repeatable rotational event, not measurement noise. It is distinguished from timing noise by its step-like, persistent character.
• "All glitches are starquakes." Starquakes were the first idea and may produce small, rare events, but they cannot recharge fast enough to power Vela's giant glitches every three years. The recurrent giant glitches require the rechargeable superfluid reservoir.
• "Glitch sizes just smoothly fill 10⁻⁹ to 10⁻⁶." The population is closer to bimodal: a continuum of small glitches plus a fairly distinct class of large, ~10⁻⁶ giant glitches. The two classes likely reflect different trigger physics.
• "The crust superfluid alone explains the angular momentum." It did, until entrainment was computed. The entrainment crisis shows the simple crustal-reservoir budget falls short by a factor of several, implicating the core superfluid — a live constraint on the equation of state.