Astrophysical transients are events that appear and disappear on human-observable timescales, and are produced in a wide variety of physical processes. Longer-duration transients, on timescales of hours to decades, like fading supernovae, can emit incoherently from thermal electrons. Short-duration transients, however, with emission on timescales of seconds or less, are necessarily coherent in nature since the emission is too bright to be explained by individual electrons emitting separately. Whereas variable sources are characterized by occasional brightening and fading, often superimposed on a stable flux source, transients are often one-off events that fade when the emission mechanism turns off. The processes that produce both fast and slow transients are some of the most energetic in the Universe. The collapse of a massive star (Smith, 2014), or the collision of two neutron stars (Abbott et al., 2017a), injects massive amounts of energy and material into the surrounding environment, producing heavy elements and seeding further star formation in galaxies. These violent processes emit across the electromagnetic spectrum on various timescales – from a few seconds of coherent gamma-ray emission from gamma-ray bursts (GRBs; Gehrels et al., 2009) to the sometimes years-long incoherent thermal radio emission from expanding material after a supernova explosion or GRB (Chandra and Frail, 2012). Binary neutron star mergers can now also be observed through gravitational radiation (Abbott et al., 2017b). The energetic remnants of stellar explosions such as neutron stars are also known to produce millisecond-duration radio pulses (Hewish et al., 1968). Studies of fast transients can provide new windows on the processes that fuel galaxy evolution (Abbott et al., 2017b), and the compact stellar remnants left behind (Hamilton et al., 1985, Lyne et al., 2001). Within this context, it is no surprise that the discovery of fast radio bursts (FRBs), bright and seemingly extragalactic radio pulses, in 2007 (Lorimer et al., 2007) presented a tantalizing opportunity to the astronomical community as a potential new window on energetic extragalactic processes.
FRBs are one of the most exciting new mysteries of astrophysics. They are bright (50 mJy – 100 Jy) pulses of emission at radio frequencies, with durations of order milliseconds or less. FRB emission has so far been detected between 400 MHz and 8 GHz. The origins of FRBs are still unknown and at present the source class is only defined observationally. In the following we provide some background on the FRB phenomenon, compare the observed population to other types of known transients, and describe our motivation for this review and its contents.
1.1. A brief history
The existence of coherent, short-duration radio pulses was predicted at least as early as the 1970s – both from expanding supernova shells combing surrounding material in other galaxies (Colgate and Noerdlinger, 1971, Colgate, 1975) and from small annihilating black holes (Rees, 1977). These theories motivated early searches by, e.g., Phinney and Taylor in 1979, who re-purposed data from the Arecibo telescope to search for pulses as short as 16 ms. Although limited in bandwidth and time resolution, these data represented one of the first sensitive high-time-resolution searches for extragalactic radio pulses. No astrophysical radio pulses were detected in this search, but they placed some of the first sensitive upper limits on short-duration radio pulses from other galaxies.
Several decades later, the first detections of FRBs (Lorimer et al., 2007) were made in surveys for radio pulsars, rapidly rotating neutron stars that emit beams of radio emission from the open magnetic field lines at their magnetic poles (see Lorimer and Kramer, 2012, for more details). The stable but extreme magnetic fields associated with radio pulsars make them natural and long-lived particle accelerators that produce coherent radio emission through an as-yet poorly understood process (Melrose, 2017). As the neutron star rotates, the beams at the magnetic poles sweep across the sky and are observed as periodic radio pulses, each pulse lasting approximately 0.1–1000 ms. The radio pulses from pulsars also experience a frequency-dependent time delay through the ionized interstellar medium (ISM), which is quantified by a dispersion measure (DM) that is proportional to the number of free electrons along the line of sight (see Section 2.1 and Section 3 for more details). This is useful for measuring the ionized content of the ISM as well as for estimating the source distance. In addition to ‘canonical’ radio pulsar emission, some pulsars are also known to produce sporadic ‘giant pulses’ (GPs), which can be much shorter duration and have much higher peak luminosity. Pulsar GPs can be as short as a few nanoseconds (Hankins et al., 2003) and have been attributed to focused coherent emission by bunches of charged particles in the pulsar beam or magnetosphere (Eilek and Hankins, 2016).
The first pulsars were found through their bright, single pulses at the Mullard Radio Observatory in 1967 (Hewish et al., 1968), and for the first few years after their discovery, single-pulse studies allowed for further understanding of the pulsar phenomenon (Backer, 1970a, Backer, 1970b, Backer, 1970c, Backer, 1975). However, given the highly periodic nature of pulsar signals, searches were soon optimized to take advantage of this property. As early as 1969, only two years after the discovery of the first pulsar, Fast Fourier Transforms (FFTs) and Fast Folding Algorithms (FFAs) were recognized as more efficient for discovering periodic signals appearing at multiple harmonics in the frequency domain — resulting in the discovery of a larger number of Galactic pulsars, with diverse properties (Burns and Clark, 1969). These searches allowed for the discovery of fainter periodic signals, pulsars with millisecond rotational periods (Backer et al., 1982), and pulsars in relativistic binary systems (Hulse and Taylor, 1975). Periodicity searches have been highly successful, increasing the total pulsar population from a few tens in the first few years (Taylor, 1969) to over 2600 sources in 2018 1.
Modern surveys search for pulsars via their periodic emission as well as their sporadic, bright single pulses. These searches are also well suited to FRB discovery due to their large time on sky and high time resolution, both of which are necessary for finding new and potentially rapidly rotating pulsars. The drive to find more millisecond pulsars (MSPs) pushed instrumentation towards the narrower frequency channels and higher time resolution required to find their signatures in the data. Improved frequency resolution in pulsar surveys also allowed more sensitive single-pulse searches up to higher DM values, including to DMs much larger than expected from the Galactic column of free electrons. Throughout the past 50 years, each new pulsar search has attempted to expand the phase space in which we search for new pulsars, expanding coverage along the axes of pulse duration, DM, duty cycle, spectrum, and acceleration in the case of pulsars in binary orbits.
As many new pulsar searches focused on finding stable periodic sources, the parameter space of short-duration single event transients remained relatively unexplored. The study of the single pulses of known pulsars continued as an active area of research (for a review, see Rankin and Wright, 2003). However, blind searches for new pulsars through their single pulses tapered off. Following a successful search for single pulses in archival Arecibo data by Nice (1999), a return to the single pulse search space was motivated by Cordes and McLaughlin (2003) and McLaughlin and Cordes (2003). In an effort to explore this parameter space within the Galaxy, McLaughlin et al. (2006) discovered 11 new sources identified through their bright, millisecond-duration radio pulses. These rotating radio transients (RRATs) were believed to be a subset of the radio pulsar population. Although RRATs had underlying periodicity, they were more readily discovered through single pulse searches, rather than through FFTs. Current observations probe only the tip of the pulse energy distribution (Weltevrede et al., 2006) and some sources could be extreme examples of pulsars that exhibit various types of variable emission such as nulling, mode changing, and intermittency, as well as GPs. The first RRATs implied that a large population of bright single pulses might be hiding in existing radio survey data (Keane et al., 2011).
Single-pulse searches in archival data targeting the Small Magellanic Cloud (SMC), and taken with the Parkes telescope in 2001, revealed a single pulsar-like pulse, so bright it saturated the primary detection beam of the receiver and was originally estimated to have a peak flux density of > 30 Jy (Fig. 1; Lorimer et al., 2007). This pulse, which soon became known as the ‘Lorimer burst’, was remarkable not only for its incredible brightness but also for its implied distance (see Section 5.1 for more details). The pulse's large dispersive delay was estimated to be roughly eight times greater than could be produced by the free electrons in the Milky Way (along this line of sight) or even in the circum-galactic medium occupying the space between the Milky Way and the SMC. Upon its discovery, the Lorimer burst suggested the existence of a population of bright, extragalactic radio pulses (Lorimer et al., 2007).
Figure 1. The Lorimer burst (Lorimer et al., 2007, now also known as FRB 010724), as seen in the beam of the Parkes multibeam receiver where it appeared brightest. These data have been one-bit digitized and contain 96 frequency channels sampled every millisecond. The burst has a DM of 375 cm−3 pc. The pulse was so bright that it saturated the detector, causing a dip below the nominal baseline of the noise right after the pulse occurred. This signal was also detected in 3 other beams of the receiver. The top panel shows the burst as summed across all recorded frequencies. The bottom panel is the burst as a function of frequency and time (a ‘dynamic spectrum’). The red horizontal lines are frequency channels that have been excised because they are corrupted by RFI.
For several years after its discovery the Lorimer Burst remained the only known signal of its kind. A new pulse of potentially similar nature was discovered in 2011 by Keane et al. (2011); however, this source was along a sight-line in the Galactic plane and thus a Galactic origin (like a RRAT) was also considered possible (see Section 5.2, and Bannister and Madsen, 2014). Strong support in favor of the Lorimer burst as an astrophysical phenomenon came from Thornton et al. (2013), who presented four high-DM pulses discovered in the High Time Resolution Universe survey at the Parkes telescope (HTRU; Keith et al., 2010). The discoveries by Thornton et al. (2013) had similar characteristics to the Lorimer burst, and implied an all-sky population of extragalactic radio pulses, which they termed ‘Fast Radio Bursts’, or FRBs.
FRBs were immediately considered of great interest due to their large implied distances and the energies necessary to produce such bright pulses. As discussed further in Section 2, from the DMs of the four new FRB sources discovered by Thornton et al. the bursts were estimated to have originated at distances as great as z = 0.96 (luminosity distance 6 Gpc). With peak flux densities of approximately 1 Jy, this implied an isotropic energy of 1032 J (1039 erg) in a few milliseconds or a total power of 1035 J s−1 (1042 erg s−1). The implied energies of these new FRBs were within a few orders of magnitude of those estimated for prompt emission from GRBs and supernova explosions, thereby leading to theories of cataclysmic and extreme progenitor mechanisms (see Section 9).
The excitement around the discovery by Thornton et al. led to increased searches through new and archival data not just at the Parkes telescope (Burke-Spolaor and Bannister, 2014, Ravi et al., 2015, Champion et al., 2016), but also at other telescopes around the world, resulting in FRB discoveries at the Arecibo Observatory (Spitler et al., 2014), the Green Bank Telescope (Masui et al., 2015), the Upgraded Molonglo Synthesis Telescope (UTMOST, Caleb et al., 2016b), the Australian Square Kilometre Array Pathfinder (ASKAP, Bannister et al., 2017, Shannon et al., 2018), and the Canadian Hydrogen Intensity Mapping Experiment (CHIME, Boyle and CHIME/FRB Collaboration, 2018, CHIME/FRB Collaboration et al., 2019b). Since 2013, the discovery rate of FRBs has increased each year, with a doubling of the known population in the last 12-month period alone (Shannon et al., 2018, CHIME/FRB Collaboration et al., 2019b).
Highlights from these discoveries have included the first two (so far) repeating FRB sources, FRB 121102 (Spitler et al., 2016, Scholz et al., 2016, Chatterjee et al., 2017) and FRB 180814.J0422+73 (CHIME/FRB Collaboration et al., 2019a), detections with interferometric techniques (Caleb et al., 2016b, Bannister et al., 2017, Chatterjee et al., 2017, Marcote et al., 2017), and FRBs with measured polarization profiles (Petroff et al., 2015a, Masui et al., 2015, Ravi et al., 2016, Petroff et al., 2017a, Michilli et al., 2018a, Caleb et al., 2018b).
Searches through archival data in 2011 also revealed a peculiar class of artificial signal at Parkes that mimicked the dispersive sweep of a genuine astrophysical signal, but through multi-beam coincidence was thought to be local in origin (Burke-Spolaor et al., 2011). These signals, dubbed ‘Perytons’, remained a curiosity and source of controversy in the field of FRBs for several years. Because of the Perytons, some astronomers speculated that perhaps all FRBs were artificial in origin. Further investigation of the Peryton phenomenon with a larger population of events and upgraded RFI monitoring at the Parkes telescope subsequently pinpointed their source to microwave ovens being operated at the site (Petroff et al., 2015c). Their identification as spurious RFI put the Peryton mystery to bed and allowed for further progress on the study of genuine astrophysical FRBs.
The discovery of FRBs as an observational class has also prompted re-examination of previously published transients surveys such as the reported discovery of highly dispersed radio pulses from M87 in the Virgo cluster in 1980 (Linscott and Erkes, 1980) and the 1989 sky survey with the Molonglo Observatory Synthesis Telescope by Amy et al. (1989), which discovered an excess of non-terrestrial short-duration bursts (1 s to 1 ms) in 4000 hours of observations. These unexplained bursts showed no clustering in time or position and were not associated with known Galactic sources. Building on the searches by Phinney and Taylor (1979), these may have been the first reported detections of FRBs; however, the limited bandwidth and time resolution of these instruments hampered further classification of the events.
1.2. The FRB population
Currently, the research community has no strict and standard formalism for defining an FRB, although attempts to formalize FRB classification are ongoing (Foster et al., 2018). In practice, we identify a signal as an FRB if it matches a set of loosely defined criteria. These criteria include the pulse duration, brightness, and broadbandedness, and in particular whether the DM is larger than expected for a Galactic source. For signals where the DM is close to the expected maximum Galactic contribution along the line of sight there is ambiguity as to whether the source is a Galactic pulsar/RRAT or an extragalactic FRB (Fig. 2).
Figure 2. The dispersion measures (DMs) of Galactic radio pulsars, Galactic rotating radio transients (RRATs), radio pulsars in the Small and Large Magellanic Clouds (SMC & LMC), and published FRBs, relative to the modeled maximum Galactic DM along the line of sight from the NE2001 model (Cordes and Lazio, 2002). Sources with DM / DMmax > 1 are thought to originate at extragalactic distances and accrue additional DM from the intergalactic medium and their host galaxy. This figure is based on an earlier version presented in Spitler et al. (2014).
As a population, FRBs have not yet been linked to any specific progenitors, although dozens of theories exist (see Platts et al. (2018) and Section 9). As of the writing of this review, the known population of FRBs consists of more than 60 independent sources detected at 10 telescopes and arrays around the world 2 (Petroff et al., 2016). The observed population spans a large range in DM, pulse duration, and peak flux density, as well as detected radio frequency. Two sources have been found to repeat (Spitler et al., 2016, CHIME/FRB Collaboration et al., 2019a) and over 10 have now been discovered in real-time and followed up across the electromagnetic spectrum (Petroff et al., 2015a, Keane et al., 2016, Petroff et al., 2017a, Bhandari et al., 2018). The properties of the observed FRB population are discussed in Section 6.
The estimated rate is roughly ≳ 103 FRBs detectable over the whole sky every day with large radio facilities (e.g. Champion et al., 2016). Even for a cosmological distribution, if FRBs are generated in one-off cataclysmic events their sources must be relatively common and abundant. The redshift distribution is poorly known; however, the rate is higher than some sub-classes of supernovae, although lower than the overall core-collapse supernova (CCSN) rate by two orders of magnitude. A more detailed discussion of the FRB rate is presented in Section 7.
At the time of this review the progenitor(s) of FRBs remain unknown. Many theories link FRBs to known transient populations or to new phenomena not observable at other wavelengths. Emission and progenitor theories are discussed in Section 8 and Section 9 (see also Platts et al., 2018, for a living catalog of theories).
1.3. Motivation for this review
Because of the rapid expansion of the research related to FRBs, and the many new discoveries reported each year, we feel that now is the ideal time for a review that covers these topics. The growing population of FRBs is also expected to bring a larger population of researchers to the field. We intend this review as a resource for researchers entering the field, as well as its growing list of practitioners.
The timing of this review is such that we hope to encapsulate the field as it stands at the beginning of 2019, with close to a hundred sources discovered but many questions left unanswered. It is our hope that many questions related to the origins and physics of FRBs will be understood as a larger population is discovered in the next few years with large instruments like CHIME, FAST, ASKAP, APERTIF, UTMOST and MeerKAT. These and many other telescopes are expected to cumulatively find hundreds of FRBs per year.
The outline of the remainder of this review is as follows: in Section 2 we introduce the observed and derived properties of FRBs. In Section 3 we detail the propagation effects that act on an FRB as it travels through the intervening magnetized and ionized medium. In Section 4 we summarize the current observational techniques used for finding FRBs, including search pipelines and single dish and interferometric methods. Section 5 discusses some of the landmark FRB discoveries from the past decade. Section 6 discusses the FRB population in terms of the distributions of observed parameters such as width, DM, and sky position. In Section 7 we extrapolate these observed distributions and speculate as to the intrinsic population distributions. Section 8 details some of the proposed mechanisms for generating FRB emission, and Section 9 more generally discusses the progenitor theories proposed for FRBs. We summarize the review in Section 10 and conclude with predictions for the next five years in Section 11.
1 All published pulsars are available through the pulsar catalogue: http://www.atnf.csiro.au/people/pulsar/psrcat/ (Manchester et al., 2005) Back.
2 All published FRBs are available via the FRB Catalogue (FRBCAT) www.frbcat.org. Back.