Soft Gamma Repeater

Discovery: the 5 March 1979 event

On 5 March 1979, an unprecedented γ-ray flash swept the inner solar system. Nine spacecraft of the Inter-Planetary Network (IPN) — among them Venera 11 and 12, Pioneer Venus Orbiter, Helios 2, ISEE-3, the German-American HEAO-1, and Vela satellites positioned across the heliosphere — recorded the same event with arrival-time delays that triangulated its source to a tiny error box inside the supernova remnant N49 in the Large Magellanic Cloud, 50 kiloparsecs away. The Soviet group led by Evgeny Mazets at the Konus instruments reported a fast 0.2-second initial spike followed by a hundreds-of-seconds pulsating tail, repeating at a period of 8 seconds — the rotation period of a neutron star.

The peak flux at Earth was so high — roughly 10⁵ erg/cm²/s of soft γ-rays for a fifth of a second — that the implied isotropic luminosity at the LMC distance was 10⁴⁵ erg/s, briefly exceeding the integrated luminosity of every star in the Milky Way. Over the following months the same source produced weaker, shorter bursts. The "5 March" event had a repeating, soft-spectrum source: a new class of object. The name "soft gamma repeater" (SGR) was coined to distinguish it from the hard-spectrum, non-repeating, cosmological γ-ray bursts (GRBs) that had been mystifying X-ray astronomers since 1973.

The magnetar model

Through the 1980s the central engine remained obscure. Two further SGRs were identified (SGR 1900+14 and SGR 1806-20), each repeating but quiescent for years between active phases. The breakthrough came in 1992 from Robert Duncan and Christopher Thompson, who pointed out that newborn neutron stars rotating with periods shorter than the convective-overturn timescale (~10 ms) can amplify their internal magnetic field by α-Ω dynamo action up to ~10¹⁵ G — far above the ~10¹² G of ordinary pulsars. They predicted a population of slowly spinning neutron stars whose X-ray emission is powered not by rotation but by the decay of the trapped field itself. They called these objects magnetars.

The model made specific predictions. The magnetic luminosity (~B²/8π × volume × decay time) must dwarf the rotational luminosity. The crust must be subject to magnetic-stress fracturing, producing short bursts and rare giant events. Magnetars must spin down rapidly because of their enormous dipole moments. They must have brief active lifetimes (~10⁴ yr) and be associated with young supernova remnants. Every one of these predictions has been confirmed observationally. The 2002 detection of cyclotron-like features in the X-ray spectra of magnetars, and the direct measurement of magnetic field strengths from spin-down (P, Ṗ), pinned down B ~ 10¹⁴–10¹⁵ G as the SGR norm.

Three types of burst

SGR activity comes in three distinct intensities, separated by orders of magnitude in duration and energy. The table below summarises a population of hundreds of observed bursts.

Class	Duration	Total energy	Rate	Spectrum	Notes
Short burst	~10⁻² – 10⁻¹ s	10⁴⁰ – 10⁴¹ erg	hundreds per active episode	thermal, kT ~ 5–10 keV	Recurrent — defines the SGR class
Intermediate flare	~1 – 100 s	10⁴¹ – 10⁴³ erg	several per year (active source)	two-component, soft tail	Often modulated at spin period
Giant flare	0.1–0.5 s spike + ~300 s tail	10⁴⁴ – 10⁴⁶ erg	~ 1 per source per century	spike hard (> 100 keV), tail soft	Three ever recorded

Short bursts arrive in storms: an SGR can emit dozens to hundreds in a single day during an active episode, then go quiet for months or years. Intermediate flares often initiate or terminate such episodes. Giant flares are the rare extremity — only three have ever been observed in 47 years of monitoring.

The three giant flares

• SGR 0526-66, 5 March 1979. The original event in the LMC. Initial 0.2-s spike ~5 × 10⁴⁴ erg, total energy ~10⁴⁵ erg. Hundreds-of-seconds pulsating tail at the 8-second spin period — the first measurement of magnetar rotation.
• SGR 1900+14, 27 August 1998. Triggered by Ulysses and RXTE. Initial spike ~10⁴⁴ erg, located in a young remnant in Aquila. The flare's pulsating tail showed clear 5.16-second modulation. Hit Earth's ionosphere measurably enough to register on RHESSI's predecessors.
• SGR 1806-20, 27 December 2004. The largest ever recorded. Peak luminosity ~10⁴⁷ erg/s, total energy 2 × 10⁴⁶ erg in the 0.2-s initial spike alone — 100× brighter than the 1979 event. Saturated every operating γ-ray detector (RHESSI, Swift, Wind, Konus, INTEGRAL); ionised Earth's day-side ionosphere from 50,000 ly away.

The 2004 event has a special importance: it lets us put concrete numbers on the upper limit of magnetar energy release. The total emitted energy was about 4% of the magnetic energy stored in a uniformly magnetised 10¹⁵-G neutron star — a sanity check that the magnetar model can in principle power what we see, but only by tapping a substantial fraction of the global field.

Why the crust cracks

The neutron-star crust is a Coulomb lattice of nuclei (mostly Fe-56 in the outer crust, more neutron-rich nuclei deeper) immersed in a sea of relativistic electrons. It has an enormous breaking strain — laboratory analogues and ab initio calculations both give a yield strain ε_max ~ 10⁻¹, roughly 10⁸ times stiffer than terrestrial steel. But the stresses are correspondingly enormous. The magnetic pressure at the surface scales as

P_B = B² / (8π)
    ≈ 4 × 10²⁸ erg/cm³   at B = 10¹⁵ G
    ≈ 4 × 10²⁸ dyne/cm²

The shear modulus of the crust is μ ~ 10³⁰ erg/cm³ — only ~30× larger than this magnetic pressure. As the internal field slowly diffuses through the crust on a Hall/Ohmic timescale of 10³–10⁴ yr, it twists the lattice. When the integrated shear strain locally exceeds ε_max, the lattice fails. The footpoints of the magnetic flux tubes embedded in the crust jump suddenly to a new location, leaving the magnetosphere above with a twisted, non-potential field that releases its excess energy via reconnection. The result, on a millisecond–to–second timescale, is a pair-photon plasma fireball trapped on the closed field lines, which radiates the SGR burst as it cools.

Worked example: the December 2004 fluence

SGR 1806-20 lies at distance d ≈ 8.7 kpc ≈ 2.7 × 10²² cm. The peak fluence in the 0.2-s spike at Earth was F ≈ 1.4 erg/cm². For isotropic emission, the total isotropic-equivalent energy is

E_iso = 4π d² F
      = 4π × (2.7 × 10²²)² × 1.4
      ≈ 1.3 × 10⁴⁶ erg

This 1.3 × 10⁴⁶ erg corresponds to converting the mass-energy of about 1.4 × 10²⁵ g of matter into radiation in 0.2 seconds — about 7 × 10⁻⁹ of a solar mass — and a peak isotropic luminosity

L_peak = E_iso / Δt = 1.3 × 10⁴⁶ / 0.2 ≈ 7 × 10⁴⁶ erg/s

which exceeds the integrated luminosity of the entire Milky Way (about 4 × 10⁴³ erg/s in total stellar light) by more than three orders of magnitude. The magnetic-energy budget of the source is

E_B = (B²/8π) × (4/3 π R³)
    = (10¹⁵)²/(8π) × 4/3 π × (10⁶)³
    ≈ 1.7 × 10⁴⁷ erg     (R = 10 km neutron star)

So the 27 December 2004 flare released roughly 4 percent of the total magnetic energy of an internally uniform 10¹⁵-G magnetar in a single 0.2-second event. The model survives the energetics test — barely.

The FRB 200428 link

Until 2020 every confirmed fast radio burst (FRB) was extragalactic, identifiable only by its dispersion measure (the integrated free-electron column along the line of sight, which placed sources at megaparsec-to-gigaparsec distances). The progenitors were unknown. On 28 April 2020 the Galactic magnetar SGR 1935+2154 emitted a millisecond radio burst, detected simultaneously by the CHIME and STARE2 radio facilities, that was 1000× brighter than any pulsar-style emission ever seen from a Galactic neutron star. The radio burst was coincident, within milliseconds, with a hard X-ray burst seen by INTEGRAL and other instruments.

The radio fluence, scaled to extragalactic distances (~100 Mpc), would have looked like a typical (low-luminosity) FRB. For the first time, a single object had been seen producing both an SGR burst and an FRB. Several more radio bursts from SGR 1935+2154 have been recorded since, and a handful of confirmed extragalactic FRB hosts are now consistent with magnetar progenitors. The link is not yet exclusive — the brightest extragalactic FRBs may still need a different mechanism — but at least one channel for FRB production is now empirically nailed to a magnetar.

The Galactic population

The McGill Online Magnetar Catalogue currently lists about 30 confirmed and candidate magnetars. Roughly half were discovered as soft gamma repeaters (defined by recurrent bursts), and half as anomalous X-ray pulsars (defined by persistent ~10³⁵ erg/s X-ray emission with spin periods of 2–12 s and no binary companion). Several "SGRs" have been seen as quiescent AXPs, and several "AXPs" have been caught bursting; the two historical classes are now considered overlapping subsets of the same underlying magnetar population.

Object	Spin P	B (dipole)	Notable event	Distance
SGR 0526-66	8.0 s	5.6 × 10¹⁴ G	5 March 1979 giant flare	50 kpc (LMC)
SGR 1900+14	5.2 s	7 × 10¹⁴ G	27 August 1998 giant flare	~12.5 kpc
SGR 1806-20	7.6 s	2 × 10¹⁵ G	27 December 2004 giant flare	~8.7 kpc
SGR 1935+2154	3.2 s	2.2 × 10¹⁴ G	FRB 200428	~9 kpc
1E 1547.0-5408	2.07 s	3.2 × 10¹⁴ G	Fastest spinning magnetar; 2009 burst storm	~4.5 kpc
4U 0142+61	8.7 s	1.3 × 10¹⁴ G	Prototypical AXP	~3.6 kpc
Swift J1818.0-1607	1.36 s	2.7 × 10¹⁴ G	Fastest spinning known magnetar	~5 kpc

Most magnetars lie in or near the Galactic plane, two are in the Magellanic Clouds. The implied Galactic birthrate from the observed population and ~10⁴-yr active lifetime is one new magnetar every several centuries — roughly 10% of all neutron-star births. The total Galactic population is therefore in the thousands, most of them quiescent and below current detection sensitivity.

SGRs vs. AXPs

Anomalous X-ray pulsars share every essential property of SGRs: slow spin (P = 2–12 s), large Ṗ (and therefore B ~ 10¹⁴–10¹⁵ G), persistent ~10³⁵ erg/s X-ray emission that exceeds rotational losses, association with young supernova remnants, and a young kinematic age. The historical observational distinction was simply that AXPs had been discovered by their steady X-ray emission while SGRs had been discovered by their bursts. As monitoring improved through the 2000s, several AXPs (1E 1048.1−5937, 1E 2259+586, XTE J1810−197) were caught bursting, and several SGRs were re-classified as having long quiescent stretches with steady X-ray output. The two classes are now considered different observational windows on the same magnetar population.

Open questions

• What fraction of FRBs are magnetars? FRB 200428 was a low-luminosity FRB. The brightest extragalactic FRBs (those with isotropic energies above ~10⁴² erg) may require either a more efficient radio-emission mechanism or a different progenitor.
• What sets the giant-flare rate? Only three giant flares in 47 years across ~30 known sources gives one giant flare per source per ~500 years. Whether this is set by global magnetic-energy reservoirs or by stochastic crustal-fault dynamics is unresolved.
• Can a Galactic giant flare be hazardous? A 10⁴⁶-erg giant flare at the distance of the nearest known magnetar (1E 1048.1−5937, ~3 kpc) would deposit ~10² erg/cm² in Earth's upper atmosphere — enough to substantially deplete the ozone layer. Estimates suggest such an event happens at Earth-relevant distances roughly once per 10⁵–10⁶ years.
• Are short GRBs related? A few of the historical "short GRBs" with hard initial spikes and soft pulsating tails have been re-attributed to giant flares from extragalactic magnetars. The dividing line between an extragalactic giant flare and a low-luminosity merger short-GRB is still being mapped.
• How do magnetars form? Whether the extreme field is dynamo-amplified during a fast-rotating proto-neutron-star phase (Duncan-Thompson) or fossilised from a strongly magnetised progenitor (Ferrario-Wickramasinghe) is still actively debated. Stellar-progenitor mass distributions and supernova-remnant associations are starting to discriminate.

Common pitfalls

• Conflating SGR bursts with classical GRBs. Classical (long and short) GRBs are extragalactic, peak in hard γ-rays (> 100 keV), and last seconds to minutes. SGR bursts are Galactic, soft-spectrum (~10–30 keV), and last a tenth of a second to a few minutes. Giant flares can mimic short GRBs when seen from outside the Milky Way, which is exactly how the historical 5 March 1979 event was first mis-classified.
• Calling SGR a "type of GRB". They are distinct phenomena with different progenitors. A giant flare seen from outside its host galaxy can look like a short GRB, but the underlying object is a magnetar in its own galaxy, not a merger or collapsar.
• Assuming B-field decay powers all the X-ray emission. A substantial fraction of magnetar persistent flux is non-thermal magnetospheric Comptonisation, not direct surface thermal emission. Modelling pulse profiles without a corona-like component gives biased temperatures.
• Using a thin-shell fireball model for the giant-flare tail. The 300-second pulsating tail is generally interpreted as a trapped fireball on closed field lines, not a freely-expanding shell. Optically thin shell models give qualitatively wrong cooling timescales.
• Treating SGRs and AXPs as separate populations. Modern usage treats them as the same population seen through different observational windows. Cataloguing tools (McGill Online Magnetar Catalogue) list them jointly.