LOFAR constraints on the repetition & environments of CHIME FRBs

TL;DR

This paper addresses the gap in understanding FRB emission at frequencies below 400 MHz by combining archival LOTAAS data and dedicated LOFAR follow-ups of CHIME-detected repeaters. Using a PRESTO-based FRB search pipeline and FETCH classifications, the study derives population-level limits on the spectral index of FRB activity, showing $α_s > -0.9$ for repeaters and $α_s > -1.2$ for non-repeaters, with a strong per-source constraint of $α_s > 0.55$ for FRB 20201124A, indicating a suppression of low-frequency activity for this source. The work also constrains circumburst environments through a free-free absorption analysis, finding consistency with a dense HII-region or a very young supernova remnant for FRB 20201124A. Predictions for LOFAR2.0 suggest 0.3–9 FRB detections per week possible in commensal operation, enabling robust cosmological constraints if high-redshift FRBs are detected. Overall, the results demonstrate that even non-detections at low frequency provide meaningful constraints on FRB repetition, spectra, and environments and inform strategies for future low-frequency FRB surveys and instrument design.

Abstract

The behaviour of fast radio bursts (FRBs) at radio frequencies <400 MHz is not well understood due to very few detections, with only two known sources detected below 300 MHz. Characterising low-frequency emission of FRBs is vital for understanding FRB emission mechanisms and circumburst environments. We robustly characterise the 150 MHz activity CHIME-detected FRB sources relative to their 600 MHz activity -- using their non-detection in 473 h of archival observations from the Low Frequency Array (LOFAR) Tied-Array All-Sky Survey (LOTAAS), and 252 h of LOFAR observations of 14 repeating FRB sources, the largest sub-300 MHz targeted FRB campaign to date. In the LOTAAS data, we search for repeat bursts from 33 CHIME/FRB repeaters, 10 candidate repeaters and 430 apparent non-repeaters. Their non-detection yields a population-level constraint on the statistical spectral index $α_{s, 150MHz/600MHz}>-0.9$, indicating that FRB spectral indices are, on average, flatter than known spectral indices from pulsars. From the targeted campaign, we find that the prolific repeater FRB 20201124A shows a positive $α_s>0.55$, implying reduced low-frequency activity, unlike the typically negative $α_{s}$ seen from FRBs at higher frequency bands. We explore free-free absorption in the circumburst environment as a cause of the non-detection at 150 MHz. The non-detection of FRB 20201124A is consistent with either a very young $\sim10$ yr old supernova remnant, or a typical HII region. Our simulations indicate that LOFAR2.0 can detect 0.3-9 FRBs per week, and up to 4 FRBs at redshifts in the range $1<z<3$. Such detections will provide robust constraints on cosmological parameters due to their clean environments. Our results guide future low-frequency FRB searches by showing how even non-detections can place meaningful constraints on the repetition rates and circumburst environments of FRBs.

Figures (7)

• Figure 1: Examples of typical localization regions from CHIME baseband data (top) and real-time header information (bottom), and their overlap with the LOTAAS tied-array beams (TABs). We plot the LOTAAS TABs that overlap with the localisation regions of FRB 20190430C and FRB 20180907A and were searched for repeat bursts in this work. The localisation regions are plotted in green, with the positional uncertainty for FRB 20190430C increased by a factor of 5 for visual clarity. TABs which did not meet the overlap criterion (see §\ref{['sec:lotaas']}) were excluded from the search and are marked with crosses. The marker colours indicate the survey pass that the TABs were observed in. The grid lines are spaced by 1$^\circ$.
• Figure 2: A timeline of targeted LOFAR exposures of CHIME/FRB repeaters, along with the CHIME/FRB burst detection times. Most individual LOFAR observations lasted 1--2 h on a single day. Each vertical line marker shows the time of a LOFAR observation session but the height of the line is arbitrary. The histogram bin heights show the exposure (in hours) but the bin widths are arbitrary. The figure emphasises how closely in time LOFAR observations were conducted relative to the most recent CHIME/FRB burst for each source. Detection times are marked up to one burst beyond the last LOFAR observation for each source. For the last panel which depicts observations of six repeaters, we prioritise summarising the LOFAR observations (histograms/vertical lines of different colours). Therefore, the first plotted CHIME burst detection (triangles of different colours) is an arbitrary amount of time before the first LOFAR observation of that source, and is not necessarily the first CHIME/FRB detection of the source. The CHIME/FRB detection data are taken from their public database (https://www.chime-frb.ca/), which also includes bursts not accounted for in the rates used in this work (although this is not meant to be a complete burst sample). We use rates from the catalogue of CHIME/FRB repeaters chime23, which only includes bursts detected up to May 1, 2021 (MJD 59335). For instance, the repeaters FRB 20220912A and FRB 20240114A were discovered after this cut-off date.
• Figure 3: Fluence threshold of LOFAR observations for the CHIME FRB sources plotted as a function of the expected scattering time at 135 MHz. The thresholds account for the fraction of fluence of scattered bursts that can be recovered with our search pipeline. This fraction, $f_\textrm{boxcar}$ (indicated by the marker colour), is derived by simulating a burst with a width and scattering time the same as that measured for the corresponding source. The inset illustrates that for a burst with a scattering timescale of 2 s, only 20% of the fluence (indicated by the shaded blue region) can be recovered with the widest boxcar in our search pipeline.
• Figure 4: Constraints on the statistical spectral index ($\alpha_\textrm{s}$) of repeating FRBs derived by comparing burst rates in different frequency ranges. The lower limits are derived by comparing rate upper limits in LOFAR observations (110--190 MHz) with the rate measured at higher frequencies. The frequency ranges corresponding to measurements by houben19 and chawla20 are indicated on the plot. The CHIME/FRB repetition rates are all scaled to a fluence threshold of 5 Jy ms, assuming the power-law index of the differential energy distribution, $\gamma_\textrm{src} = -2.5$. For sources with available constraints from both LOTAAS and targeted observations, the stronger of the two constraints is plotted here.
• Figure 5: Constraints on the average statistical spectral index ($\alpha_\textrm{s}$) for 33 repeating FRB sources and 371 (apparent) non-repeating FRBs. The constraints are derived by simulating $10^4$ realisations of our LOTAAS search, with the average number of expected detections in these realisations plotted on the y-axis. Based on our non-detection, values of $\alpha_\textrm{s}$ less than the plotted 90% confidence constraints can be ruled out as they predict LOTAAS FRB detections in $>90$% of the simulated realisations.