Fast Radio Burst

A burst found in dusty data

In 2007, Duncan Lorimer asked his student David Narkevic to look through archival 2001 data from the 64-metre Parkes radio telescope, hunting for new pulsars. They found something else. A single bright pulse, ~ 30 Jy peak flux density at 1.4 GHz, lasting ~ 5 milliseconds, with a strong frequency-dependent delay characteristic of a propagating radio wave through ionised gas. The dispersion measure was 375 pc/cm³ — far larger than the maximum 25–50 pc/cm³ allowed by the Milky Way along that line of sight. The source was therefore extragalactic; the implied isotropic energy was 10³³ erg in a single millisecond, ten million times brighter than a typical pulsar pulse.

The "Lorimer burst" — formally FRB 010724 — was reported in Science, but the field stalled for years. A Parkes-specific class of artefacts called "perytons" (later traced to opening the door of a microwave oven before the heating cycle finished) muddied the discrimination of real FRBs from local interference. By 2014 several more bursts had been found at multiple telescopes. By 2015, FRB 121102 was caught repeating at Arecibo, ruling out cataclysmic single-shot models for at least that source. By 2017, ASKAP localised the first apparently non-repeating FRB to a host galaxy. By 2020, the field was on solid ground.

What an FRB looks like

The defining observable is a brief, dispersed radio pulse. Three quantities characterise each event:

Quantity	Typical range	Meaning
Pulse width τ	0.1 – 10 ms	Light-crossing time of emitting region (compact)
Peak flux density S_ν	0.1 – 100 Jy at 1 GHz	Apparent brightness
Fluence F	0.1 – 1000 Jy ms	Time-integrated flux density
Dispersion measure DM	100 – 3000 pc/cm³	Free-electron column density
Frequency range	110 MHz – 8 GHz observed	Spans low to high radio bands
Brightness temperature T_b	10²⁴ – 10³⁶ K	Far above incoherent limit; coherent emission
Polarisation	Often near 100% linear	Constraint on emission mechanism

The brightness temperature is the most physical of these. For a source of physical size R radiating power L_ν at frequency ν, the Rayleigh-Jeans equivalent temperature is T_b ~ L_ν c²/(2 k_B ν² R²). Plugging in 10⁴⁰ erg in a 1 ms pulse from a region R ≲ c × 1 ms = 300 km, one gets T_b ~ 10³⁵–10³⁶ K — far above the ~ 10¹² K limit at which incoherent synchrotron self-Comptons itself to oblivion. FRB emission must therefore be coherent: many particles radiating in phase, like a maser or curvature radiation in a dense magnetic-field bunch, not random thermal photons.

Dispersion measure: a cosmic ruler

When a radio pulse propagates through a plasma of free electrons, the lower-frequency components are slowed more than the higher frequencies. The delay between two frequencies ν₁ and ν₂ is

τ_DM = (e² / 2π m_e c) × DM × (ν₁⁻² − ν₂⁻²)
     ≈ 4.15 × 10⁹ ms × DM × (ν₁⁻² − ν₂⁻²)        [DM in pc/cm³, ν in MHz]
DM = ∫ n_e dl                                    (pc/cm³)

For a 1 ms pulse observed between 1.4 and 1.2 GHz with DM = 1000 pc/cm³, the lower-frequency component arrives ~ 1 second later than the higher-frequency component. That dispersion delay is sweeping, easily observed, and the smoking gun that an event is a propagating coherent radio pulse rather than instrumental.

For an extragalactic FRB, the total DM has three contributions:

DM_observed = DM_MW + DM_IGM(z) + DM_host(z)

The Milky Way contribution is constrained by Galactic electron-density models (NE2001, YMW16) to typically < 50–200 pc/cm³ for high-latitude lines of sight. The host-galaxy term depends on the source environment, ranging from a few pc/cm³ for clean halos to thousands for dense star-forming regions. The intergalactic-medium term, averaged, is approximately

⟨DM_IGM(z)⟩ ≈ 1000 × (Ω_b h / 0.05) × z   pc/cm³

This is the Macquart relation, named after the late Jean-Pierre Macquart who first articulated the program. It directly connects FRB DM to the cosmic baryon density. In 2020 a five-FRB ASKAP sample with optical-localised host redshifts confirmed Ω_b h² consistent with Big Bang nucleosynthesis and CMB measurements — closing the so-called "missing baryon problem" for the warm-hot intergalactic medium.

Worked example: distance from DM

FRB 121102's host is a dwarf star-forming galaxy at spectroscopic z = 0.193 (luminosity distance ≈ 972 Mpc, comoving distance ≈ 815 Mpc). Its average DM is ≈ 558 pc/cm³. Galactic contribution at this line of sight ~ 188 pc/cm³ (NE2001). Therefore the residual DM from IGM + host is

DM_residual = 558 − 188 = 370 pc/cm³

From the Macquart relation, the expected mean IGM contribution at z = 0.193 is

⟨DM_IGM⟩(0.193) ≈ 1000 × (0.05 × 0.7 / 0.05) × 0.193
                 ≈ 1000 × 0.7 × 0.193
                 ≈ 135 pc/cm³

That leaves ~ 235 pc/cm³ for the host-galaxy contribution — large but plausible for a dwarf galaxy with a dense star-forming HII region and a young magnetar embedded in supernova-remnant nebular gas. VLBI imaging of FRB 121102 indeed reveals a persistent radio nebula coincident with the burst position, of size ~ 0.7 pc and luminosity 10³⁸ erg/s. The DM contribution from this nebula plus a young supernova remnant accounts naturally for the 235 pc/cm³ excess.

Inverting the logic: for an FRB of unknown z, one estimates the host contribution and uses the Macquart relation to solve for z. Localisations confirm distances inferred this way to ~ 30% accuracy without spectroscopy — invaluable for the bulk of the CHIME catalogue, where optical follow-up is impractical.

The April 2020 Galactic burst

On 28 April 2020, the Galactic magnetar SGR 1935+2154, ~ 9 kpc away in the supernova remnant G57.2+0.8, entered an X-ray-burst-storm episode. At 14:34:24 UTC, both CHIME (in zenith Canada) and the STARE2 array (in California) caught a brief, dispersed radio pulse coincident in time and direction with the magnetar. Within 3 ms of the radio signal, NICER, INTEGRAL, AGILE, and Insight-HXMT registered a hard X-ray burst from the same source.

The radio fluence was ~ 1.5 × 10⁶ Jy ms — millions of Jy — and the implied isotropic energy at 9 kpc was ~ 3 × 10³⁴ erg. That is 30 times the brightness of the brightest Crab giant pulse and 10² × the dimmest extragalactic FRB ever localised. Crucially, the same isotropic-equivalent energy as known FRBs at the bright end of their luminosity function. So if a magnetar can produce a burst of this energy, at the brightest extragalactic FRB sources we are seeing the same physics, scaled up only modestly.

The April 2020 detection therefore established magnetars as at least one progenitor class for FRBs. It does not prove that all FRBs come from magnetars — and the 2022 detection of an FRB in an old globular cluster in M81 (FRB 20200120E) argues against a single channel, since standard young-magnetar formation is not active in 10-Gyr stellar populations. But the magnetar model is the canonical baseline for the FRB community, with multiple bursts now seen from SGR 1935+2154 in the years following.

Repeaters and one-offs

Of the > 1000 catalogued FRBs by 2025, ~ 5% have been observed to repeat. Some, like FRB 121102, fire hundreds of bursts in clustered episodes. FRB 180916.J0158+65 shows a ~ 16.35-day periodicity in its activity windows — the strongest known periodicity in the FRB sample, and a constraint that any progenitor model must accommodate (binary modulation? precession? slow magnetar rotation?).

The big open question is whether "one-off" FRBs are physically distinct from repeaters, or whether they are simply repeaters whose follow-up monitoring has not yet caught a second burst. Three pieces of evidence:

• Pulse morphology. Repeater bursts often show downward frequency drift (the "sad trombone effect"), while many one-offs show simpler, narrower spectra. The CHIME/FRB collaboration argues this is a true progenitor diagnostic.
• Host galaxy diversity. Repeaters concentrate in younger, star-forming hosts; one-offs span a wider variety of hosts, including older quiescent galaxies. This hints at multiple progenitor channels.
• Energy distribution. The cumulative energy distributions of repeaters and one-offs are inconsistent with a single power-law. At least two populations are statistically required.

The provisional working hypothesis is that there are at least two progenitor classes: "young magnetars" (most repeaters, born in core-collapse supernovae) and "everything else" (older systems, possibly including merger products or interactions in compact-object binaries).

The instruments that catch them

Instrument	Active	Frequency	FRB cadence	Localisation
Parkes (Murriyang)	2001–	1.4 GHz	~ few/year	Beam-only (degree)
Arecibo (decommissioned 2020)	2012–2020	1.4 GHz	FRB 121102 monitoring	Arc-minute
ASKAP	2017–	700–1800 MHz	~ 30/year (incoherent)	Sub-arcsec ICS
CHIME/FRB	2018–	400–800 MHz	~ 1000/year	~ 0.1° → arcsec with Outriggers
FAST	2018–	1.05–1.45 GHz	10⁴ bursts on FRB 121102	Arcsec
DSA-110 / DSA-2000	2022–	1.28 GHz	~ 100/year, growing	Sub-arcsec
STARE2	2017–	1.4 GHz	SGR 1935+2154 detection	Beam-only

CHIME is the dominant detector by raw rate, with its ~ 200° sky-scan FOV cataloguing ~ 1000+ FRBs per year. ASKAP and the new DSA-110 are the localisation workhorses, providing arcsecond positions for optical follow-up and host-galaxy identification. FAST, the world's largest single-dish radio telescope at 500 m diameter, provides exquisite sensitivity for repeater monitoring — the FAST monitoring of FRB 121102 catalogued more than 10⁴ bursts in late 2019 alone.

Variants and extensions

• Young-magnetar model. Newly-formed (≲ 10⁴ yr) magnetars in core-collapse supernovae undergo magnetic reconnection and crustal failure events that drive coherent radio emission via curvature radiation, antenna mechanisms, or maser emission in their wind. Predicts repeaters in dense star-forming environments — matches FRB 121102.
• Magnetar-flare maser model. Magnetar flares drive relativistic shocks into the surrounding nebula; synchrotron-maser instability at the shock generates coherent radio emission downstream. Naturally predicts the high brightness temperatures and high linear polarisation observed.
• Compact-object merger model. Coalescing binary neutron stars, NS-BH mergers, or WD-NS mergers can momentarily produce magnetar-like conditions. Could explain one-off FRBs in older hosts or in unusual environments. Late-time simultaneous gravitational-wave detection of an FRB would clinch this channel.
• Globular cluster mechanism. The 2022 discovery of FRB 20200120E in an M81 globular cluster argued for an old-stellar-population progenitor: perhaps a magnetar formed by accretion-induced collapse of a white dwarf, or a binary merger. The first piece of clean evidence that multiple progenitor channels operate.
• FRB cosmography. Localised FRBs are now used to weigh diffuse cosmic baryons, constrain the Hubble constant, and bound the photon mass. The Macquart relation (DM_IGM ∝ H_0 z) is the engine of this cosmography. Future LSST + DSA-2000 + ngVLA samples of 10⁴+ localised FRBs will turn the technique into precision cosmology.

Where FRBs show up — and what they tell us

• FRB 121102 (Spitler et al. 2014). First repeater. Localised to a star-forming dwarf galaxy at z = 0.193. Persistent compact radio nebula at the burst position; rotation measure 10⁵ rad/m² indicates a magneto-ionic environment plausibly from a young magnetar in a supernova remnant.
• FRB 200428 / SGR 1935+2154. Galactic FRB caught simultaneously by CHIME and STARE2 in April 2020. Coincident X-ray burst from the magnetar within 3 ms. Established magnetars as one progenitor channel.
• FRB 180916 (CHIME/FRB 2020). 16.35-day periodicity in burst activity. Localised to a spiral arm in a nearby (z = 0.0337) galaxy. The periodicity remains the single strongest predictive timescale in any FRB source.
• FRB 20200120E (Kirsten et al. 2022). Localised to an old globular cluster in M81 at 3.6 Mpc — by far the closest extragalactic FRB. Argues against a single young-magnetar channel.
• FRB 20220610A (Ryder et al. 2023). Most distant localised FRB to date, in a galaxy at z = 1.016. DM ~ 1458 pc/cm³, fluence ~ 45 Jy ms. Tests the Macquart relation in the deep-redshift regime.

Common pitfalls

• Confusing dispersion delay with intrinsic duration. The intrinsic burst duration is the de-dispersed pulse width (≲ ms). The frequency-swept arrival time across a band is the dispersion delay, which can be seconds even for a millisecond pulse. The two have different physical meanings — instrumental vs intrinsic.
• Confusing fluence with isotropic energy. Isotropic-equivalent energy is E_iso = 4π d² F where F is fluence and d is luminosity distance. Apparent burst luminosity assumes isotropic emission; if the FRB is beamed (likely for many models), true luminosity is lower by a beaming factor.
• Treating Galactic and extragalactic FRBs interchangeably. SGR 1935+2154's burst at ~ 10³⁴ erg is at the low-end of the FRB luminosity function; the brightest extragalactic FRBs reach 10⁴² erg, eight orders of magnitude higher. Whether the same mechanism scales over this range is debated.
• Assuming all FRBs are coherent curvature radiation. Multiple emission mechanisms remain viable: synchrotron-maser at relativistic shocks, antenna mechanisms in plasma cavities, coherent curvature radiation in magnetar magnetospheres. The polarisation and spectral data so far do not uniquely select one.
• Reading a non-detection as a non-event. CHIME's frequency band is 400–800 MHz; an FRB whose spectrum cuts off above that frequency would be invisible to other instruments at higher bands and vice versa. The frequency-dependent FRB rate is itself a constraint that has produced significant CHIME-vs-Parkes population debates.