A Search for Coincident Neutrino Emission from Fast Radio Bursts with Seven Years of IceCube Cascade Events

The IceCube Neutrino Observatory, located at the geographic South Pole, is the largest neutrino detector in the world. Encompassing a cubic kilometer of instrumented ice, IceCube is comprised of 5160 digital optical modules (DOMs) situated on 86 read-out and support cables or "strings" to detect the Cherenkov radiation from charged particles created by neutrino interactions in the Antarctic ice (Abbasi et al. 2009; Aartsen et al. 2017). IceCube's instrumentation density is optimized for the detection of neutrinos with energies from 100 GeV to 10 PeV. It contains a higher density subvolume, enabling the detection of neutrinos down to 10 GeV (Abbasi et al. 2012). IceCube has observed a diffuse flux of high-energy astrophysical neutrinos (Aartsen et al. 2013, 2014a, 2015). A study of data collected between 2008 April 6 and 2018 July 10 revealed a 3.3σ inconsistency with background expectations for four sources that include NGC 1068 and TXS 0506+056 (Aartsen et al. 2020b). In 2022, IceCube found further evidence of neutrino emission from NGC 1068 at a significance of 4.2σ (Abbasi et al. 2022). Despite this growing evidence, the origin of the majority of the diffuse astrophysical flux remains unexplained. In this paper, we present a search for time-dependent neutrino emission from various fast radio bursts (FRBs) using seven years of IceCube cascade events and set upper limits on the associated neutrino flux for flares of varying duration.

Transient astrophysical objects are among the primary candidates for producing the astrophysical neutrino flux (Murase & Bartos 2019). The first evidence of neutrino emission from a flaring object came from the blazar TXS 0506+056 (Aartsen et al. 2018a, 2018b). FRBs are a class of transient astrophysical objects that could contribute to the diffuse neutrino flux (Metzger et al. 2020). FRBs are periodic or nonperiodic transient radio bursts that have Galactic or extragalactic origins. To date, hundreds of FRBs across the entire sky have been detected. Recently, FRB 200448 was localized to the Galactic magnetar SGR 1935+2154, suggesting magnetars may be a source of FRBs (Anderson et al. 2020). This has been further supported by the polarization of some FRBs (Wang et al. 2021). While the underlying mechanism that creates FRBs is unknown, it is predicted that this coherent radio emission is the result of an ultrarelativistic shock that propagates into a baryon-filled medium. A by-product of this scenario are TeV–PeV neutrinos that are produced by photohadronic interactions on timescales of varying duration after the FRB (Metzger et al. 2020; Qu & Zhang 2022). Here, we perform a time-dependent stacking search to test if two catalogs of FRBs, one repeating and the other nonrepeating, are producing a statistically significant number of neutrinos.

The majority of events detected by IceCube can be divided into two topological classes: cascades and tracks. Cascade signatures are produced by charged-current electron neutrino and tau neutrino interactions as well as all-flavor neutral-current interactions. These interactions typically produce electromagnetic and hadronic showers that have a range of 20 m or less; this range scales with energy (Aartsen et al. 2014b). Due to the average DOM spacing and light scattering within the ice, these particle showers lead to cascade events having angular resolutions ∼10°–15° (Aartsen et al. 2019). In contrast, track events are the result of muons that are produced by charged-current muon neutrino interactions and have an angular resolution of less than 1° at TeV–PeV energies (Abbasi et al. 2021a). Track events are subject to larger background rates that stem from atmospheric muons. To reduce this, track events below a certain energy threshold are filtered out (Abbasi et al. 2021a). The distinctive topology of cascade events allows our selections to include events that have energies down to hundreds of GeV (Aartsen et al. 2019).

To search for correlations between FRBs and cascade events, the energy, spatial, and temporal information of each cascade event is used with the spatial and temporal information of each FRB. This method is similar to that of IceCube's previous FRB analyses, referred to as Six Year Southern Tracks (Aartsen et al. 2018c) and All-Sky Tracks (Aartsen et al. 2020c), which analysed track data for correlations.

2.1. Data Set and FRB Catalogs

This analysis uses IceCube's Medium Energy Starting Cascade (MESC) data set. A starting event is classified as an event that contains a neutrino interaction vertex in the detector. This data set was originally developed to search for all-sky signatures of potential neutrino sources such as in the Galactic plane (Aartsen et al. 2019). It contains 1,980 cascade events detected from 2010 May to 2017 May. This corresponds to 2428 days of IceCube live time. The first year of data was taken with a 79-string detector configuration and the remaining six years use the complete 86-string configuration. The events in this data set have energies that range between 270 GeV and 1.6 PeV.

This analysis focuses on 48 unique bursts that were detected between 2010 May and 2017 May; the bursts are separated into two catalogs (Petroff et al. 2016). As shown in Figure 1, the first catalog contains 22 nonrepeating FRBs that are located at a decl. of six degrees or below. The second catalog contains one repeating FRB (FRB 121102) that produced 26 bursts within the live time of the MESC data set. Since FRBs are an observational class of astrophysical phenomena, nonrepeating and repeating FRBs could be associated with different classes of astrophysical objects. As the underlying physical processes may differ, we treat them in this analysis as independent phenomena. These catalogs are a subset of the FRBs that were analysed in the Six Year Southern Tracks and All-Sky Tracks analyses. This is because the MESC data set has a live time period that does not fully overlap with the track data sets. The catalogs of nonrepeating and repeating FRBs are independently analysed through two time-dependent stacking analyses.

Zoom In Zoom Out Reset image size

2.2. Analysis Methods

This analysis uses an unbinned extended maximum likelihood method, similar to the previous analyses (Aartsen et al. 2018c). The likelihood is comprised of spatial, temporal, and energy probability density functions (PDFs). These PDFs, similar to those defined in Section 3.1 of Braun et al. (2010), characterize how signal-like or background-like a given cascade is. To test if a cascade and FRB are temporally correlated, a search time window, ΔT, is constructed in the interval [t_FRB, t_FRB + ΔT], where t_FRB denotes when the FRB was detected. Motivated by Metzger et al. (2020), we test eight different ΔT that are logarithmically spaced from [10⁻², 10⁷] s after the FRB has occurred. ⁶⁸  The spatial PDF accounts for the angular distance between a given FRB and the reconstructed cascade direction. A 2D Gaussian spatial PDF is used for cascades with an angular resolution less than 7°, while for a resolution greater than 7° a Kent distribution is used. At larger angular resolutions, a Kent distribution characterizes a probability distribution in curved space and cannot be approximated by a Gaussian distribution in flat space. The energy PDF characterizes the neutrino flux in terms of a power law, E^−γ , where γ is the spectral index. We assume a single power law for the energy flux. Using the likelihood ratio test described in Wilks (1938), we construct a test statistic (TS) that encompasses our null and alternative hypotheses,

$$\begin{eqnarray}&&{\rm{TS}}=-2\mathrm{log}\displaystyle \frac{{ \mathcal L }({n}_{s}=0)}{{ \mathcal L }({n}_{s},\gamma )}.\end{eqnarray} \tag{ 1 }$$

The null hypothesis represents background expectations (n_s = 0) whereas the alternative hypothesis fits for two parameters: the number of signal events, n_s , and γ. The number of events in a given time window can vary according to Poisson statistics. In each ΔT, a small number of events are required to observe a statistically significant correlation. To perform a stacking search, the TS in Equation (1) is additionally summed over each FRB in the catalog.

We observed no significant emission in this analysis resulting in TS = 0 for every time window. Upper limits are calculated for both catalogs at the 90% confidence level for the time-integrated flux per FRB for every ΔT (Figure 2). The upper limits assume a flavor ratio (ν_e :ν_μ :ν_τ ) of 1:1:1 with equal parts from ν and $\overline{\nu }$. From 10⁻² to approximately 10⁶ s, the upper limits are relatively constant due to a low background rate. The background saturation occurs 10⁶ s after the FRB.

Zoom In Zoom Out Reset image size

For the E⁻² (Figure 2, left) and E⁻³ (Figure 2, right) spectra, we find that the stacked upper limits with cascade events are comparable to the Six Year Southern Tracks and All-Sky Tracks results (Figure 2). For E⁻³, this analysis offers improved upper limits with respect to the Six Year Tracks. For E⁻² this analysis is not quite as sensitive as the Six Year Tracks when searching for neutrino emission in ΔT less than 10⁴ s. The flux allowed by the All-Sky Tracks analysis increases by an order of magnitude between ΔT that have durations of subseconds to days. This is due to the rapidly increasing background rate, which requires more signal events to see a statistically significant correlation. As noted above, this is in contrast with cascade events. Finally, we find that for harder spectra, the catalog of nonrepeating FRBs offers slightly more constraining upper limits when compared with FRB 121102. This is in contrast with our upper limits for soft spectra where FRB 121102 has slightly stronger upper limits.

These results are extended to E^−2.53 to draw comparisons to the measured diffuse flux that has been observed in IceCube's cascade analyses (Aartsen et al. 2020a). The diffuse flux is an all-sky high-energy neutrino flux. Since FRBs are an all-sky phenomenon, these results are relevant. Figure 3 compares the per-burst upper limits for each catalog to the diffuse cascade astrophysical flux by integrating the diffuse flux over both the duration of each ΔT and the entire sky then dividing by the Canadian Hydrogen Intensity Mapping Experiment's (CHIME's) estimated FRB all-sky rate of 820 FRBs per day (Amiri et al. 2021). This assumes that each burst contributes equally to the diffuse all-sky cascades. This flux per FRB is then compared to the E^−2.53 upper limits from this analysis. Using this method of comparison, we see that the measured diffuse cascade flux establishes the most stringent limit on neutrino emission from FRBs.

Zoom In Zoom Out Reset image size

In two independent searches for neutrino emission from 22 unique, nonrepeating FRBs and 26 unique bursts from FRB 121102, no significant correlation was found. We provide upper limits on the time-integrated neutrino flux at the 90% confidence level for various spectral indices, as shown in Figures 2 and 3. We also provide estimates on the neutrino flux for FRB catalogs of various sizes (Figure 4). These estimates show that for nonperiodic FRBs that are isotropically distributed throughout the southern sky, the neutrino flux per FRB decreases as the catalog size increases. In recent years more FRB observatories, such as CHIME, have come online. This has increased the detection rate of both repeating and nonrepeating FRBs. However, we were only able to analyse a subset of FRBs that overlapped with our data set. New data sets are in preparation that would overlap with 100–1000 FRBs and potentially offer more stringent upper limits (Petroff et al. 2016; Amiri et al. 2021).

Zoom In Zoom Out Reset image size

Overall, this analysis shows that cascades offer sensitive upper limits when performing transient stacking searches. Given that cascades have a low background rate and are considered independent of track events, IceCube's future analyses would benefit from combining track and cascade events when searching for time-dependent neutrino emission from FRBs. This approach can be extended to IceCube's potential searches that aim to observe transient neutrino emission in real time as well as gamma-ray bursts (Aartsen et al. 2016; Abbasi et al. 2021b). In addition, the next generation of IceCube, IceCube-Gen2, will provide the opportunity to conduct more sensitive searches for neutrino emission from transient sources (Aartsen et al. 2021). In turn, this will allow us to uncover the origin of high-energy astrophysical neutrinos.

The IceCube collaboration acknowledges the significant contributions to this manuscript from Michael Kovacevich. USA—U.S. National Science Foundation-Office of Polar Programs, U.S. National Science Foundation-Physics Division, U.S. National Science Foundation-EPSCoR, Wisconsin Alumni Research Foundation, Center for High Throughput Computing (CHTC) at the University of Wisconsin–Madison, Open Science Grid (OSG), Advanced Cyberinfrastructure Coordination Ecosystem: Services & Support (ACCESS), Frontera computing project at the Texas Advanced Computing Center, U.S. Department of Energy-National Energy Research Scientific Computing Center, Particle astrophysics research computing center at the University of Maryland, Institute for Cyber-Enabled Research at Michigan State University, and Astroparticle physics computational facility at Marquette University; Belgium—Funds for Scientific Research (FRS-FNRS and FWO), FWO Odysseus and Big Science programmes, and Belgian Federal Science Policy Office (Belspo); Germany—Bundesministerium für Bildung und Forschung (BMBF), Deutsche Forschungsgemeinschaft (DFG), Helmholtz Alliance for Astroparticle Physics (HAP), Initiative and Networking Fund of the Helmholtz Association, Deutsches Elektronen Synchrotron (DESY), and High Performance Computing cluster of the RWTH Aachen; Sweden—Swedish Research Council, Swedish Polar Research Secretariat, Swedish National Infrastructure for Computing (SNIC), and Knut and Alice Wallenberg Foundation; European Union—EGI Advanced Computing for research; Australia—Australian Research Council; Canada—Natural Sciences and Engineering Research Council of Canada, Calcul Québec, Compute Ontario, Canada Foundation for Innovation, WestGrid, and Compute Canada; Denmark—Villum Fonden, Carlsberg Foundation, and European Commission; New Zealand—Marsden Fund; Japan—Japan Society for Promotion of Science (JSPS) and Institute for Global Prominent Research (IGPR) of Chiba University; Korea—National Research Foundation of Korea (NRF); Switzerland—Swiss National Science Foundation (SNSF); United Kingdom—Department of Physics, University of Oxford.