5. LANDMARK FRB DISCOVERIES

In the following, we discuss some of the most influential FRB discoveries of the past 10 years. These include FRBs that extend the parameter space in one or more ways, as well as FRBs that have been the center of extended discussion in the literature.

5.1. FRB 010724 − The Lorimer Burst

FRB 010724, also known as ‘the Lorimer burst’, is considered to be the first FRB discovery (Lorimer et al., 2007). It was discovered before the term ‘fast radio burst’ was even coined (the term was introduced by Thornton et al., 2013), and remains one of the brightest FRBs yet to be detected. The burst was initially reported as detected in three beams of the Parkes multi-beam receiver – implying a location between the beams, which required an extremely high inferred peak flux density. The burst saturated the primary detection beam and was initially estimated to have a peak flux density of 30 Jy and a fluence of 200 Jy ms (Lorimer et al., 2007). Subsequent re-analysis of the data by Burke-Spolaor et al. (2011) detected the FRB signal weakly in a fourth beam of the receiver. Based on new beam pattern models of the Parkes multi-beam receiver, a revised analysis of FRB 010724 by Ravi (2019) localized FRB 010724 to a region of a few square arcminutes within the primary beam and the true fluence was estimated to be 800 ± 400 Jy ms, further solidifying the Lorimer burst as one of the most luminous known FRBs.

While FRB 010724 remains an outlier in the Parkes FRB population, several FRBs in the ASKAP sample appear to have similar fluences (Shannon et al., 2018). This is perhaps unsurprising given that the ASKAP surveys provided much larger sky coverage, but at lower sensitivity compared with Parkes. Recent studies of the ensemble properties of FRBs have remarked that the Lorimer burst strongly affects the slope of the source counts and other statistics related to the brightness distribution of FRBs. Macquart and Ekers (2018) have argued that FRB 010724 should be excluded from statistical analyses of the FRB population and that it is subject to discovery bias – i.e. the idea that the first-discovered source in a new class may be easier to detect and have exceptional properties compared to the rest of the underlying population.

5.2. FRB 010621 − The Keane Burst

FRB 010621, also known as ‘the Keane burst’ was the second candidate FRB reported in the literature. Presented in Keane et al. (2011), and further discussed in Keane et al. (2012), the burst was discovered in a search of the Parkes Multibeam Pulsar Survey (PMPS; Manchester et al., 2001) for single pulses from RRATs and Lorimer-type bursts. The single bright pulse was reported with a DM of 745 ± 10 cm⁻³ pc along a sightline where the modeled DM contribution from the Galaxy is 523 cm⁻³ pc according to the NE2001 model (although the line-of-sight DM_MW is only estimated to be 320 cm⁻³ pc in the YMW16 model). The small fractional DM excess of the pulse relative to the expected DM of the Galaxy in that direction made it unclear whether the pulse was extragalactic in origin or from a Galactic source located along an overdense sightline through the Galactic plane. Bannister and Madsen (2014) studied the sightline of FRB 010621 in Hα and Hβ emission to determine a more precise electron density measurement and concluded with 90% confidence that the burst was from a Galactic source along an overdense sightline. Unless repeating pulses, allowing a precise localization and a host galaxy association, are detected in the future, the true distance will remain uncertain. FRB 010621 is thus considered a marginal case between the FRB and Galactic pulse source classes.

5.3. FRB 140514

FRB 140514, also known as ‘the Petroff burst’, was discovered in a targeted search of the locations of previously detected FRBs, where the motivation was to search for repeating pulses from these sources (Petroff et al., 2015a). It was found in the field of the previously reported bright FRB 110220 (Thornton et al., 2013) in a receiver beam pointed 9′ away from the reported location of the previous FRB. Despite the similar sky location, the two FRBs were discovered with markedly different DMs: 944.38 ± 0.05 cm⁻³ pc for FRB 110220, and 562.7 ± 0.6 cm⁻³ pc for FRB 140514. Petroff et al. (2015a) thus concluded that the bursts were not related and estimated a 32% probability of finding two positionally similar but physically unrelated FRB in the survey as a whole. However, Maoz et al. (2015), using the argument that FRB 140514 occurred in the receiver beam pointed to the field of FRB 110220, concluded that the two bursts must be from the same source with 99% confidence. Ultimately, the difference in statistical analyses between the two teams come from considering only a single beam of the Parkes multi-beam receiver or the entire receiver field of view (see further discussion in Chapter 6, Petroff, 2016).

If FRB 110220 and FRB 140514 are indeed two bursts from the same source separated by 3 years, Piro and Burke-Spolaor (2017) argue that the source could be a neutron star embedded in a dense supernova remnant and the large change in DM could be explained by a shell of material expanding radially outward. In order to produce such a large fractional change they estimate that the supernova would have to have occurred within ∼10.2 years of FRB 110220.

FRB 140514 was also the first discovery by a newly commissioned real-time search pipeline in operation at the Parkes telescope, which enabled the full polarimetric properties of the FRB to be recorded. The burst was found to be 20% circularly polarized, with no detection of linear polarization. See Section 6.1 for a more detailed discussion.

5.4. FRB 121102

Discovered using the 305-m Arecibo telescope in Puerto Rico, FRB 121102, also known as ‘the Spitler burst’, was the first FRB to be detected with a telescope other than Parkes. As such, it added credence to the astrophysical interpretation of the phenomenon in the early days of the field. Spitler et al. (2014) discovered the burst in a single-pulse search of archival data from the PALFA Galactic plane survey (Cordes et al., 2006, Lazarus et al., 2015). It was the only burst seen in a 180-s observation, and no additional bursts were seen in a second survey scan coincidentally taken 2 days later. FRB 121102 is in the Galactic anti-center at l = −0.2^∘, b = 175^∘. The DM = 557 cm⁻³ pc is 300% larger than that predicted by the NE2001 model (Cordes and Lazio, 2002), which suggested an extragalactic origin despite the low Galactic latitude of the source. Curiously, the spectrum of the burst is inverted, following approximately S_ν ∝ ν⁷. This led Spitler et al. (2014) to hypothesize that the burst was detected in a side lobe of the ALFA 7-beam receiver.

At the time of discovery, it was unclear whether FRB 121102 was a genuine extragalactic burst, a RRAT with an anomalously high DM, or some type of pernicious RFI. While initial follow-up observations detected no additional bursts (Spitler et al., 2014), a deeper campaign was planned to better establish whether FRB 121102 was truly a one-off event. Deep follow-up of the Lorimer and Keane bursts had made no additional detections and similar follow-up of other Parkes FRBs yielded no repeating pulses (Petroff et al., 2015b). Thus it came as a surprise when Arecibo observations in May 2015 detected the first repeat bursts from FRB 121102 (Spitler et al., 2016). These additional follow-up observations used the 7-beam Arecibo ALFA receiver to grid a large area around the original detection position. Perhaps most surprising was how active FRB 121102 suddenly was. Of the 10 new bursts detected by Spitler et al. (2016), 6 were discovered within a 10-minute observation and some were substantially brighter compared with the first-detected burst. The new detections showed that the original FRB 121102 burst had been detected in the sidelobe of one of the telescope beams; however, each new burst had a different spectrum, sometimes poorly modeled by a power-law and peaking within the observing band. The strange spectrum was therefore something characteristic to the signal itself and not an instrumental artifact.

In terms of constraining theory, the detection of repetition provides a clear constraint: the FRB cannot come from a cataclysmic event and whatever is producing the bursts must be able to sustain this activity over a period of at least 7 years – 2012 to present day. The repeating pulses made it possible to study the source in greater detail and perform multi-wavelength measurements. Most importantly, it became possible to precisely localize the source using a radio interferometer, without having to do this in real-time using the initial discovery burst.

Scholz et al. (2016) presented additional detections of FRB 121102 using Arecibo and the GBT. They also performed a multi-wavelength study of the field around FRB 121102 and showed that it was unlikely that the source's high DM was produced by a Galactic Hii region along the line-of-sight.

At the same time, the VLA and European VLBI Network (EVN) were used to obtain a precision localization. After tens of hours of observations with the VLA, 9 bursts were detected using high-time-resolution (5 ms) visibility dumps (Chatterjee et al., 2017), which localized FRB 121102 to ∼ 100 mas precision (Fig. 10, left). This allowed Chatterjee et al. (2017) to see that FRB 121102 is coincident with both persistent radio and optical sources (Fig. 10, right). Very-long-baseline radio interferometric observations using the EVN and Very Long Baseline Array (VLBA) showed that the radio source is compact on milli-arcsecond scales. Archival optical images from the Keck telescope suggested that the optical source was slightly extended.

Figure 10. Left: Using the VLA, repeating bursts from FRB 121102 were localized to sub-arcsecond precision using interferometric techniques. Right: The localization allowed for the identification of the host galaxy at radio and optical (inset) wavelengths. Figures 1 and 2 from Chatterjee et al. (2017).

Marcote et al. (2017) managed to detect additional bursts using EVN data, providing another step in localization precision. FRB 121102 and the persistent source were found to be coincident to within ∼ 12 mas. In parallel, Tendulkar et al. (2017) acquired Gemini North spectroscopic data that detected the optical source and measured its redshift: z = 0.193, which corresponds to a luminosity distance of ∼ 1 Gpc. The extragalactic origin and host galaxy of FRB 121102 were thus conclusively established.

FRB 121102's host galaxy turned out to be a low-metallicity, low-mass dwarf (Tendulkar et al., 2017). Given that such galaxies are also known to be the common hosts of superluminous supernovae (SLSNe) and long gamma-ray bursts (LGRBs), this presented a tantalizing possible link between FRBs and these other types of extreme astrophysical transients (Metzger et al., 2017). Deeper observations of the host using the Hubble Space Telescope (HST) revealed that FRB 121102 is coincident with an intense star-forming region (Bassa et al., 2017b). The EVN radio position is offset from the optical centroid of the star-forming region by 55 mas, statistically significant, but within the half-light radius.

Multi-wavelength observations also searched for prompt optical, X-ray and γ-ray flashes associated with the radio bursts. No optical pulses were found in a campaign where the 2.4-m Thai National Telescope was shadowed by Effelsberg and 13 radio bursts were detected (Hardy et al., 2017). Similarly, despite the detections of multiple radio bursts, no prompt X-ray or γ-ray bursts were found in observations with simultaneous radio and Chandra, XMM-Newton, Swift, and Fermi coverage. Nor is there any persistent X-ray/γ-ray emission detected (Scholz et al., 2016, Scholz et al., 2017).

In the absence of high-energy bursts, the radio bursts themselves become even more important for interpreting FRB 121102. The precision localization has allowed for observations at higher radio frequencies (> 2 GHz) and using higher time and frequency resolution. Given that the DM of the source is known, on-line coherent dedispersion can be used in order to avoid intra-channel dispersive smearing. The earliest high-frequency burst detections were made at 5 GHz with Effelsberg (Spitler et al., 2018) and at 3 GHz with the VLA (Law et al., 2017). Gajjar et al. (2018) detected over a dozen bursts in only a 30-minute observing window using an ultra-wideband recording system from 4−8 GHz. Arecibo observations from 4−5 GHz also detected over a dozen bursts, and the full Stokes recording mode allowed polarization to be detected for the first time. The bursts were found to be ∼ 100% linearly polarized with a rotation measure of 1.46 × 10⁵ rad m⁻² that decreased to 1.33 × 10⁵ rad m⁻² within 7 months (in the source frame; Michilli et al., 2018a). This demonstrated that FRB 121102 is in an extreme and dynamic magneto-ionic environment. It also distinguished the first repeater in a new way: no other FRB had been shown to have such a large RM.

Most recently, Hessels et al. (2018) used a sample of high-S/N, coherently dedispersed bursts to demonstrate complex time-frequency patterns in the signals from FRB 121102. This is discussed in more detail in Section 8, and it may represent a means to observationally separate repeating and non-repeating FRBs. Gourdji et al. (2019) studied a sample of low-S/N, low-energy (10³⁷⁻³⁸ erg/s) FRB 121102 bursts and showed that their typically narrow-band spectra (∼ 200 MHz at 1400 MHz) are a significant impediment to detection when using standard search methods. It is certain that current methods are sub-optimal and that bursts are being missed; one can speculate that this is true not only for FRB 121102, but for FRBs in general.

Table 2 summarizes the observational properties of FRB 121102 and its host galaxy.

5.5. FRB 180814.J0422+73

In January, 2019 it was reported by the CHIME/FRB collaboration that a second repeating FRB was discovered in the pre-commissioning data from the telescope. This repeating burst source, FRB 180814.J0422+73 (also referred to colloquially as ‘R2’, whereas FRB 121102 is ‘R1’) was found at a very low dispersion measure DM = 189 cm⁻³ pc. Bursts were detected at 6 epochs between August and October 2018 CHIME/FRB Collaboration et al. (2019a). The FRB source was found in a circumpolar region of the sky, meaning that it was visible to the CHIME telescope in both ‘upper’ and ‘lower’ transits. Using all detections, FRB 180814.J0422+73 was published with an estimated position: RA = 04:22:22, Dec = +73:40 with uncertainties of ± 4′ in RA and ± 10′ in Dec.

Interestingly, at least two bursts from R2 show complex time-frequency structure. These bursts show multiple sub-bursts, each with finite frequency bandwidth with earlier sub-bursts peaking in brightness at higher frequencies. The descending time-frequency structure within a total burst envelope is similar to structure seen in some pulses from FRB 121102 (Hessels et al., 2018). That this structure is seen in some pulses from both repeaters (Fig. 11) may indicate that the origin is intrinsic to the emission mechanism rather than an extrinsic propagation effect that requires a particular geometry, such as plasma lensing.

Figure 11. De-dispersed spectra of individual bursts from (a) the repeating FRB 121102 at 1.4 GHz using Arecibo, and (b) the repeating FRB 180814.J0422+73 discovered with CHIME at 700 MHz. Both repeating sources have some bursts that show distinct sub-burst structure with descending center frequencies over time. Horizontal bands in both spectra are due to narrow-band RFI excision in the data. FRB 121102 data from Hessels et al. (2018). FRB 180814.J0422+73 data from CHIME/FRB Collaboration et al. (2019a).