Pulsar Kick: How a Lopsided Supernova Launches a Neutron Star

A corpse that outruns its parent star

Neutron stars are among the fastest-moving objects in the Galaxy. Radio timing and interferometric astrometry of hundreds of pulsars give a three-dimensional space velocity that averages around 400 km/s, with the distribution roughly Maxwellian and a one-dimensional spread of σ ≈ 265 km/s (Hobbs et al. 2005). The tail of that distribution is startling: the fastest well-measured pulsars move at more than 1000 km/s — fast enough to cross the distance from the Earth to the Moon in about six minutes, or to traverse the entire ~100,000 light-year Milky Way in a few tens of millions of years.

Here is the puzzle. The massive O and B stars that end their lives as neutron stars are sluggish by comparison, drifting through their birth clouds at only about 10–30 km/s. Whatever happens at death does not merely inherit the parent star's motion — it adds a brand-new velocity an order of magnitude or two larger. That extra velocity is the natal kick. We even see its statistical fingerprint in Galactic geography: pulsars are found at much larger heights above the Milky Way's disk (a scale height of several hundred parsecs) than the thin, ~50–80 pc layer of OB stars they came from. Something launched them out of the plane.

Crucially, the kick is cheap. A 1.4-solar-mass neutron star (≈ 2.8×10³³ g) moving at 400 km/s carries a momentum of only ~1.1×10⁴¹ g·cm/s and a kinetic energy of a few ×10⁴⁸ erg — a fraction of a percent of the ~10⁵¹ erg released as explosion kinetic energy (rising toward ~1% only for the very fastest pulsars). The supernova barely notices donating this energy; the neutron star feels it for the rest of its existence.

The mechanism: a one-sided explosion and a gravitational tug-boat

A core-collapse supernova is not the tidy, spherical blast of textbook cartoons. In the crucial first second, the collapsing iron core rebounds into a proto-neutron star, and a shock wave stalls a couple hundred kilometers out, held back by infalling matter. What revives that shock — and blows the star apart — is a violently turbulent, asymmetric process. Two hydrodynamic instabilities dominate: neutrino-driven convection in the heated "gain region" just below the shock, and the standing accretion shock instability (SASI), a large-scale sloshing and spiral mode that can pump the shock into a lopsided, dipolar (ℓ = 1) shape. The explosion therefore emerges stronger on one side than the other.

Momentum must balance. If more ejecta — and more momentum — leaves toward one hemisphere, the neutron star recoils toward the other. But the decisive insight from modern three-dimensional simulations (Scheck, Janka, Wongwathanarat, Müller, Nordhaus, Burrows and collaborators, from ~2006 onward) is subtler than "blow gas one way, recoil the other." The neutron star is accelerated mainly by gravity, not by the pressure of the gas hitting it. The over-dense, slow-moving clump of ejecta left behind on one side lingers near the neutron star and tugs it gravitationally for a second or more, like a tug-boat towing a ship. This is the aptly named gravitational tug-boat mechanism.

The arithmetic works comfortably. The ejecta carry a total momentum of order several ×10⁴² g·cm/s (a few solar masses moving at thousands of km/s). A dipolar asymmetry of just a few percent in that ejecta momentum supplies the ~10⁴¹ g·cm/s the neutron star needs to reach several hundred km/s. And because the tug is gravitational and long-lived, the kick keeps building over ~1–3 seconds after collapse — long after the neutrinos have done most of their work. Simulations that follow this process naturally produce kicks from tens up to ~700 km/s or more, matching the bulk of the observed population.

The neutrino channel and the energy budget

There is a second way to push a neutron star, and it exploits an enormous reservoir. A collapsing core releases roughly 3×10⁵³ erg as neutrinos over about 10 seconds — some 99% of the total gravitational energy of the collapse, dwarfing the ~10⁵¹ erg of visible explosion. Neutrinos carry momentum (p = E/c), so the total neutrino momentum is about 3×10⁵³ erg ÷ c ≈ 10⁴³ g·cm/s. Compare that to the ~10⁴¹ g·cm/s the neutron star needs: a mere ~1% front-to-back asymmetry in the neutrino emission would deliver a ~300 km/s kick all by itself.

That sounds easy, but generating and sustaining a coherent 1% (let alone several-percent) neutrino anisotropy is hard. Random turbulent fluctuations tend to average out. The most promising route to a large neutrino kick invokes strong magnetic fields (≳ 10¹⁵–10¹⁶ G) inside the proto-neutron star, which break the symmetry of weak-interaction cross sections and make neutrinos preferentially escape along the field — a natural way to also explain why some fast pulsars are strongly magnetized. In most current models, though, the neutrino channel is a contributor at the level of a hundred or a few hundred km/s, while the hydrodynamic tug-boat accounts for the biggest kicks.

Either way, the energetics underline how lopsided a supernova only needs to be. The neutron star's final kinetic energy — around 10⁴⁹ erg even for a 1000 km/s monster — is a rounding error against the neutrino luminosity and just ~1% of the explosion. Nature does not have to try hard to fling a city-sized object across the Galaxy; it only has to be slightly asymmetric about the right axis.

How we clock a kick: proper motions, bow shocks, and offset pulsars

We measure kicks three complementary ways, and the best cases combine all three.

• Astrometric proper motion + distance. Very Long Baseline Interferometry (VLBI, e.g. the VLBA) tracks a pulsar's position to sub-milliarcsecond precision year over year, while a parallax gives its distance. Multiply the angular drift by the distance and you get the transverse velocity. This is how PSR B1508+55 was clocked at ~1083 km/s (Chatterjee et al. 2005), one of the fastest reliably measured neutron stars.
• Cometary bow shocks. A neutron star moving supersonically through the interstellar medium piles up gas into a glowing bow shock. The most famous is the Guitar Nebula, an Hα bow shock trailing pulsar PSR B2224+65, whose transverse velocity is of order 1000 km/s (higher in some distance estimates). The shape and standoff distance of the shock encode both the speed and the ambient gas density.
• Offset from a supernova remnant. If a pulsar sits away from the geometric center of the remnant it was born in, the offset divided by the remnant's age gives the velocity — and points back along the direction of launch. The "cosmic cannonball" PSR J0002+6216 is the poster child: imaged with the VLA in 2019, it trails a ~13-light-year tail that points straight back to the center of supernova remnant CTB 1, implying a velocity of about 1130 km/s. Similarly, the "cannonball" neutron star RX J0822−4300 in the Puppis A remnant has been clocked with Chandra at several hundred km/s.

Because these methods anchor to real geometry — a tail, a bow, an expanding shell — they do more than give a number; they show the kick as an arrow on the sky.

Spin-kick alignment: a clue to the stopwatch

One of the most revealing findings is that, for several young pulsars, the kick direction lines up with the spin axis. The Vela pulsar shows this beautifully: Chandra images its pulsar wind nebula as a symmetric X-ray torus with jets, which fix the spin axis, and the pulsar's proper motion aligns with that axis to within a few degrees (Ng & Romani 2004). The Crab pulsar shows a comparable, if looser, alignment (within ~10–30°).

Why should a kick "know" about the spin axis? The answer is a stopwatch. If the kick were delivered as a single instantaneous shove in a random direction while the proto-neutron star was already spinning quickly, the direction would be arbitrary. To end up parallel to the spin, the accelerating force must either be tied to the rotation axis or, more simply, must be averaged over many rotation periods — a push that persists for much longer than the initial spin period smears out any sideways component, leaving a net thrust along the pole. Alignment therefore argues for a kick that builds over a relatively long time (tens of milliseconds to seconds), consistent with the slow, gravitationally driven tug-boat picture and with rapid initial rotation. It is one of the cleaner observational constraints we have on when and how long the kick acts.

Consequences: exiled pulsars, shattered binaries, and open questions

A few hundred km/s changes everything about a neutron star's fate. Open clusters and OB associations have escape velocities of only a few to ~10 km/s, so essentially every kicked neutron star is immediately ejected from its birthplace and joins a fast, disk-crossing population — the reason pulsars are scattered so far from the star-forming plane.

The effect on binaries is even more dramatic. Most massive stars are born with a companion, and a kick comparable to or larger than the orbital velocity unbinds the system, flinging out a lone pulsar and a runaway star in opposite directions. Only a minority of binaries survive both supernovae to become the double neutron stars we prize for tests of gravity — and even then, the kicks leave scars: the eccentric orbits and misaligned spins of systems like the Hulse–Taylor binary (PSR B1913+16, e ≈ 0.62) are direct evidence that the second-born neutron star got a real jolt. Kick statistics also feed directly into how many neutron-star mergers, and thus gravitational-wave sources and kilonovae, the Universe produces.

Not every neutron star is kicked hard, and that too is informative. The velocity distribution looks bimodal: alongside the fast σ ≈ 300+ km/s population there is a low-velocity component (σ ≈ 75 km/s; Verbunt, Igoshev & Cator 2017). The gentle kicks are naturally produced by electron-capture supernovae and ultra-stripped supernovae — explosions with little asymmetric ejecta — which explains how some neutron stars are retained in globular clusters (escape speeds of only tens of km/s) rather than being blasted free.

Open questions remain sharp. What sets the exact split between the hydrodynamic and neutrino channels, and do the very fastest pulsars (>1000 km/s) demand something special? Do black holes receive kicks, and are they weaker because the larger mass dilutes the same momentum (and because fallback may symmetrize the explosion)? Is there a correlation between kick velocity and a neutron star's spin, magnetic field, or mass? Each new astrometric measurement and each new 3D supernova simulation is, in effect, another data point on how asymmetric the death of a massive star really is.

Proposed mechanisms for a neutron star's natal kick (and one look-alike that isn't a kick)

Mechanism	What carries the momentum	Typical magnitude	Current verdict
Gravitational tug-boat (hydrodynamic)	Slow, dense, one-sided ejecta gravitationally pulls the neutron star for seconds	up to ~700–1000+ km/s	Leading mechanism; reproduced in 3D core-collapse simulations
Anisotropic neutrino emission	A ~1% dipole asymmetry in the ~3×10⁵³ erg neutrino burst	~100s of km/s (larger needs strong B-fields)	Contributes; large kicks debated, may need magnetic effects
Electromagnetic rocket (Harrison–Tademaru)	Off-center rotating magnetic dipole radiates asymmetrically	≲ tens of km/s for most pulsars	Usually subdominant; naturally aligns kick with spin axis
Blaauw recoil (a look-alike, not a natal kick)	Sudden symmetric mass loss in a binary; the surviving star recoils	Set by mass lost and orbital speed	A real but distinct binary effect — no intrinsic push on the star