Limits on the Hard X-Ray Emission From the Periodic Fast Radio Burst FRB 180916.J0158+65

Fast radio bursts (FRBs) are bright, millisecond duration radio outbursts whose origin is still not clearly understood. To date, most FRBs possess dispersion measures larger than that of our own galaxy implying distances consistent with an extragalactic origin. In a few cases the host galaxies of these FRBs have been localized, e.g., FRB 121102 (Tendulkar et al. 2017), where the authors have found that it is located within a dwarf galaxy at a redshift of z ∼ 0.192 and there is a faint persistent radio source of unknown origin at the FRB location (Chatterjee et al. 2017). Most FRBs have not been found to repeat. But deeper sky surveys such as CHIME/FRB are revealing an increasing number of repeater FRBs (CHIME/FRB Collaboration et al. 2019b; Fonseca et al. 2020; Amiri et al. 2021). It has been proposed that most FRBs that we observe are actually repeaters because their all-sky rates largely exceed the all-sky rates of known cataclysmic events (see, e.g., Nicholl et al. 2017; Ravi 2019; Zhang 2020, and references therein).

FRB 180916 is one of the most well-known repeaters and also one of the most nearby FRBs detected to date. In 2019 December, the CHIME/FRB Collaboration reported the discovery of eight new repeating FRBs including FRB 180916, which was localized to a star-forming region of a nearby massive spiral galaxy at a redshift of z = 0.0337 ± 0.0002 (Marcote et al. 2020). FRB 180916 exhibited ∼38 radio bursts during 2019 September–2020 October, showing a clear periodicity of 16.35 ± 0.15 days, the first for any FRB. All the bursts arrived in a 5 day phase window and almost 50% of the bursts arrived in a 0.6 day phase window; these are the most active FRBs in the published CHIME/FRB sample (CHIME/FRB Collaboration et al. 2019a). The periodicity suggests a mechanism that periodically modulates the observed emission itself, or a periodic burst, or something else. Hence, this source is the focus of intense multiwavelength campaigns.

High-energy emission from magnetars has long been thought to be one of the important mechanisms by which FRBs may also be generated (see Margalit et al. 2020, and references therein). The recent discovery of the connection between FRB 200428 and SGR 1935 (CHIME/FRB Collaboration et al. 2020; Margalit et al. 2020; Mereghetti et al. 2020; Younes et al. 2020; Li et al. 2021; Kirsten et al. 2021; Younes et al. 2021) confirmed this fact that at least some FRBs are produced by magnetar short bursts. The typical radio to X-ray fluence measured was η ≡ E_R/E_X ∼ 10⁻⁵, implying that the radio emission is weak and contributes little to the overall burst energy budget. Moreover the radio fluence of FRB 200428 was somewhat lower compared to its extragalactic counterparts (the STARE2 burst energetics however have a strictly lower limit since some of the burst appears to be out of band). Variable high-energy emission at different timescales (milliseconds to years) is one of the key features of magnetars, and may have a relation to FRB emission (Palmer et al. 2005; Younes et al. 2015; Coti Zelati et al. 2018; Younes et al. 2020).

Here we discuss the three most common types of emission phenomenology from magnetars. First, the flares associated with the short bursts, which can be bright (L_X ≤ 10⁴² erg s⁻¹) and last a few hundreds of milliseconds; these short bursts may happen in isolation, or during a burst storm when hundreds of them are detected, continuing for days. Second, magnetars can exhibit giant flares, which consist of a very bright pulse (L_X ≤ 10⁴⁷ erg s⁻¹) at submillisecond level, followed by a softer pulsating tail lasting for several minutes. These events have been detected in three Galactic magnetars (e.g., SGR 1806; Mazets et al. 1979; Hurley et al. 1999; Palmer et al. 2005). Third, on longer timescales, magnetars recurrently enter outburst phases when their persistent flux level increases by orders of magnitude accompanied by random spectral and temporal variability. These phases often persist for a few months to years, after which they recover to their pre-outburst levels. In this work we focus on detecting the second and third kinds of emission from a putative extragalactic magnetar powering FRB 180916. Swift Burst Alert Telescope (BAT) survey observations, which cover the energy range 14–195 keV and have a time integration of a few hundreds of seconds, can constrain to models of young magnetars undergoing giant flares.

Several works studying the X-rays to γ-rays have put upper limits on the possible nondetection from the sky location of FRB 180916. Verrecchia et al. (2021) using AGILE in the energy range 0.4 MeV–30 GeV could put an upper limit on the total hard X-ray energy of ∼10⁴⁶ erg s⁻¹ comparable to that of SGR 1806-20. Although the extragalactic FRBs emit radio pulses of energies that are significantly larger than that detected from SGR 1935, there are no simultaneous or contemporaneous detections of keV, MeV, or GeV photons to date. Contemporaneous X-ray and radio observations of FRB 180916 were carried out by Scholz et al. (2020), with CHIME and the Chandra X-ray Observatory. They detected no X-ray events in excess of the background, and the nondetections imply a 5σ fluence limit of < 5 × 10⁻¹⁰ erg cm⁻² in the 0.5–10 keV energy range during the prompt emission and < 1.3 × 10⁻⁹ erg cm⁻² at any time during the observations. Given the cosmological distance of this FRB, these relate to the total isotropic energies < 1.6 × 10⁴⁵ erg and < 4 × 10⁴⁵ erg, respectively. In the 10–100 keV Fermi Gamma Ray Burst Monitor (GBM) observations, the search for prompt emission (Scholz et al. 2020) led to a fluence upper limit of 9 × 10⁻⁹ erg cm⁻². Tavani et al. (2020) searched for contemporaneous γ-ray emission from FRB 180916 using AGILE and X-ray emission using the X-Ray Telescope (XRT) on board The Neil Gehrels Swift Observatory (Swift from now on), during the bursting phases of the source in 2020 February and 2020 March. They did not detect any hard X-ray or γ-ray photons, and could provide a persistent flux upper limit of 5 × 10⁻¹⁴ erg cm⁻² s⁻¹, which translates to an isotropic luminosity of L_X ∼ 1.5 × 10⁴¹ erg s⁻¹. Guidorzi et al. (2020) using the Insight-Hard X-ray Modulation Telescope (Insight-HXMT) could provide a prompt emission upper limit in hard X-rays (1–100 keV) of <10⁴⁶ erg over a timescale of Δt < 0.1 s. They could rule out giant flares similar to the ones that were observed in Galactic magnetars.

FRB sources like magnetars may emit the majority of their flux at photon energies >10 keV (Woods & Thompson 2006; Turolla et al. 2015; Scholz et al. 2017; Margalit et al. 2020). To study the possible hard X-ray emission in the 14–195 keV energy band, we analyze all the available survey mode observations from BAT on board Swift. BAT's well-suited energy range (14–195 keV) coupled with hundreds of pointing on the source over a period extending from 2020 February–2022 January, serves as a unique opportunity for us to capture any hard X-ray emission from FRB 180916 and/or put upper limits on the short-term (hundreds of seconds) or long-term (∼year) persistent X-ray emission from the source.

The paper is arranged as follows: Section 2 discusses the observations and data analysis. Section 3 discusses the results, followed by conclusions in Section 4.

FRB 180916 has been observed by Swift BAT (Gehrels et al. 2004; Barthelmy et al. 2005) ∼220 times from 2020 February–2022 January in survey mode. We checked in the CHIME/FRB catalog (Amiri et al. 2021) for the time stamps of the ∼90 radio bursts detected to date from FRB 180916. Unfortunately, none of the CHIME-reported burst times were coincident with our BAT survey observations. We also note that there are no time-tagged event data available in the BAT archive where FRB 180916 is in the field of view of BAT. For out-of-field-of-view sources, the response matrix of BAT is highly uncertain and hence any measurement of flux is associated with large errors, hence we did not use them in this work. Below we describe the methods used to obtain the survey data sets, and to reprocess and analyze them. We have extensively used bat-tools from HEASOFT.

The BAT survey mode data are also known as detector plane histograms or DPHs. As opposed to event data, BAT survey data are accumulated in histograms on board the spacecraft, with typical integration times of 300 s or more. An 80-channel binned spectrum is recorded for each of the active detectors and saved in the DPH files. In this work we use all the available BAT data sets and we explain below the steps we have taken to reprocess and analyze them (see Table 1 for a full list of observations).

The BAT survey data sets were downloaded from HEASARC, and every survey observation has one or more pointings and hence different integration times. The task batsurvey performs basic analysis of BAT survey data and reduces a set of "raw" observed DPHs. Most importantly, it performs data screening that the BAT team has found vital for obtaining good quality results. It produces sky images and source fluxes for each independent "snapshot," corresponding to a single pointed visit by BAT. We chose a set of eight independent energy bins, and batsurvey recorded the images and fluxes in each of those bands separately. B atsurvey operates on a single BAT observation. For multiple observations, we used batsurvey once for each observation. We provided the source R.A. and decl. in the catalog, which were used as an input to batsurvey.

The output from batsurvey gives us exposure-specific results. For example, one of the important files produced for each pointing is the flux catalog. This catalog contains sources listed in the input catalog (which in this case includes FRB 180916) as well as sources detected by a blind search. From this file, we create the lightcurve for the source FRB 180916 using the command "batsurvey-catmux." We then extract the Tstart, and ΔT from the lightcurve and obtain the time-integrated spectrum with eight energy channels.

In the first round, to generate the source spectrum we choose "CENT-RATE" as the column for rate array, and use the background variance (BKG-VAR) for the error on the rate. We note that the source spectrum is already background subtracted due to the coded mask technique; hence sometimes when the background variance is larger than the "net source" counts, then the count rate may become negative. Once the spectrum is obtained we obtain the response matrix using the task "batdrmgen". We used the model cflux*powerlaw in XSPEC notation to estimate the flux in the 14–195 keV energy band. We froze the power-law normalization to a value of 1e − 03 and kept the power-law slope (Γ) free, in order to calculate the flux using the model above. We did not detect any excess emission above the background (at a >5σ confidence) in any of the pointings of this source.

In the second round, we reextracted the spectrum for every pointing, now using the 5 × BKG-VAR as the rate array. Here we use a simple powerlaw model with Γ = 1 fixed, to estimate the flux in the 14–195 keV energy range. Note that this is the 5σ upper limit on the background variance and hence the 5σ upper limit on the flux. In Table 1 we quote the 5σ upper limit on the 14–195 keV flux obtained for all the available BAT pointings. To carry out all the above steps in a coherent way, we developed a user-friendly pipeline in python, which will be published in the near future.

We have carried out a hard X-ray counterpart search of FRB 180916 using survey mode observations of BAT on board Swift. We have analyzed all available observations as of 2022 January 15 using standard analysis methods. We do not detect any source signal in excess of the background with >5σ significance. The 5σ upper limit on the background flux range from 4.5 × 10⁻¹⁰–7.6 × 10⁻⁹ erg cm⁻² s⁻¹ for different pointings of BAT survey observations. At the source distance this relates to a 5σ upper limit on a luminosity of 5.08 × 10⁴⁴–8.5 × 10⁴⁵ erg s⁻¹. Below we discuss our results in context to previous estimates of X-ray limits on this source and also discuss the possible progenitor scenarios. We note that the CHIME-measured fluences for this FRB vary considerably, and one representative value is ∼27 Jy ms (for burst ID: 181222 in the CHIME catalog). Unfortunately, we did not have any simultaneous CHIME and BAT (in field-of-view) observations.

Observations with the AGILE telescope (Tavani et al. 2021), working in the energy range 400 keV–100 MeV, could constrain the bursting X-ray energy of FRB 180916 to ∼10⁴⁶ erg, which is smaller than that observed from giant flares from Galactic magnetars. Similar upper limits of ∼10⁴⁶ erg on the bursting energy of FRB 180916 were obtained using Insight-HMXT in the energy range 1–100 keV (Guidorzi et al. 2020). On the soft X-ray range 0.3–10 keV strong limits on persistent luminosity (<2 × 10⁴⁰ erg s⁻¹) and prompt emission fluence (<5 × 10⁻¹⁰ erg cm⁻²) could be obtained during the bursting phases of FRB 180916 using observations from Chandra (Scholz et al. 2020). Similarly, Tavani et al. (2020) could put a cumulative upper limit on the X-ray flux using Swift XRT observations F < 5 × 10⁻¹⁴ erg cm⁻² s⁻¹, obtained during the 5 day active period of the source. These clearly rule out magnetar giant flares and/or SGR 1935 type events.

The Swift BAT flux upper limits in 14–195 keV obtained in this work, which are of a few ×10⁻⁹ erg cm⁻² s⁻¹, correspond to a luminosity limit of about ∼10⁴⁵ erg s⁻¹ at the source location (that is at 149 Mpc). We discuss the following three progenitor scenarios:

• 1.  
  Magnetar Giant Flares (MGFs). Typical magnetar giant flares with a luminosity ∼10⁴⁷ erg s⁻¹, lasting for ∼0.1 s (Burns et al. 2021), would lead to an average luminosity of ∼10⁴³ erg s⁻¹, when averaged over 1000 s of a typical BAT survey observation. This would correspond to a flux of ∼4 × 10⁻¹² erg cm⁻² s⁻¹. This flux level is well below the flux level that BAT is sensitive to, and hence we cannot rule out an MGF event.
• 2.  
  Pulsating Tails of MGFs. The pulsating tails have a typical luminosity of 10⁴² erg s⁻¹, and their peak energy lies in the BAT energy range, and they extend for several seconds (Younes et al. 2020; Burns et al. 2021). The corresponding flux emitted by such pulsating tails would be ∼10⁻¹³ erg cm⁻² s⁻¹, which is well below the flux level that BAT is sensitive to.
• 3.  
  An SGR 1935 Type Event. For such an event happening at the location of FRB 180916 the expected 20–200 keV flux is about a few ×10⁻¹⁴ erg cm⁻² s⁻¹ (Mereghetti et al. 2020; Li et al. 2021). This flux level is also well below the flux level that BAT is sensitive to, and hence we cannot rule out an event like SGR 1935.

Our limits exclude any persistent emission or transient activity with the source position down to ∼10⁴⁵ erg s⁻¹. The possible progenitor of FRB 180916 might be older than canonical magnetars, based on the offset reported in Tendulkar et al. (2021). The authors suggest that the spatial offset and hence the timescale (of travel from its presumed birth site) points more toward high-mass X-ray binaries and gamma-ray binaries, rather than magnetars. Given the unpredictable nature of FRBs and magnetars, such multiband long-term monitoring snapshots are extremely useful. Future opportunities with prompt downlink and analysis of Swift BAT time-tagged event data of FRBs (GUANO; Tohuvavohu et al. 2020; DeLaunay & Tohuvavohu 2021), contemporaneous with radio bursts, will open up new avenues of hard X-ray follow-up of these hitherto unknown phenomena.

We searched for high-energy transients in Swift BAT survey mode data associated with FRB 180916 from 2020 February to 2022 January. We did not detect any significant emission in the 14–195 keV energy band in any of our observational snapshots when FRB 180916 was in the field of view of BAT. The upper limits on the flux estimated in these observations exclude any persistent emission or transient activity with the source position down to ∼10⁴⁵ erg s⁻¹, for all the snapshots. Our results confirm and corroborate previous limits of high-energy transients by Fermi GBM, AGILE, HXMT-Insight, and XMM-Newton.

The material is based upon work supported by NASA under award No. 80GSFC21M0002. E.T. acknowledges support from NASA grant 80NSSC18K0429. M.N. is supported by the European Research Council (ERC) under the European Unions Horizon 2020 research and innovation program (grant agreement No. 948381) and by a Fellowship from the Alan Turing Institute.