Late-time Radio and Millimeter Observations of Superluminous Supernovae and Long Gamma-Ray Bursts: Implications for Central Engines, Fast Radio Bursts, and Obscured Star Formation

The SLSNe observed with the Karl G. Jansky Very Large Array (VLA) represent a complete (as of 2017 February) volume-limited sample to a distance twice that of FRB 121102 (z ≲ 0.35), and with a timescale of ≳ 5 yr post-explosion, leading to 15 SLSNe. Of these, three events were previously observed in the radio at early time, leading to nondetections: PTF09cnd (Chandra et al. 2009, 2010), SN2012il (Chomiuk et al. 2012), and SN2015bn (Nicholl et al. 2016b, 2018); see Table 1. Observations of PTF10hgi obtained as part of this work were presented in Eftekhari et al. (2019). We also observed seven LGRBs with the VLA, of which two were detected at early times: (i) GRB 020903 had a 5 GHz flux density of ≈ 100 μJy at 0.5 yr post-explosion, declining as F ∝ t^−1.1 (Soderberg et al. 2004); and (ii) GRB 030329 was last detected 4.9 yr post-explosion with a 5 GHz flux density of 70 μJy and a decline rate of t^−1.3 (Mesler et al. 2012). The contribution from an afterglow at the time of our observations (δ t ≈ 5490 and δ t ≈ 5323 days for GRB 020903 and GRB 030329, respectively) is expected to be negligible.

We additionally observed 29 SLSNe with the Atacama Large Millimeter/submillimeter Array (ALMA), seven of which overlap with our VLA sample. These events were observed as part of two observing campaigns. Events observed during the first campaign (2017.1.00280.S; PI: Berger) constitute all sources accessible to ALMA (as of March 2017) ¹⁸ with a timescale of ≳3 yr post-explosion and a distance of z ≲ 0.5. The second campaign (2019.1.01663.S; PI: Eftekhari) consists of all events with decl. < 30° and z < 0.4, with the exception of SN2017egm, which has been observed as part of a dedicated ALMA campaign; the source is also observed at early times in the radio (Bright et al. 2017; Bose et al. 2018; Coppejans et al. 2018).

We obtained 6 GHz (C-band) radio observations of our sample with the VLA in configurations A and B with a total on-source integration time of ≈ 40 minutes per target. The details of the observations are summarized in Table 2. All observations obtained in 2017 utilized the 8 bit samplers with ∼2 GHz bandwidth (excluding excision of edge channels and radio frequency interference (RFI)), while our 2019 observations were configured using the 3 bit samplers, providing the full 4 GHz of bandwidth across the observing band.

We processed the data within the Common Astronomy Software Application (CASA) software package (McMullin et al. 2007). We performed bandpass and flux density calibration using 3C286 and 3C48. Complex gain calibrators for individual sources are listed in Table 2. We imaged each field using 2048 × 2048 pixels at a scale of 0.07 − 0.2 arcsec per pixel using multifrequency synthesis (MFS; Sault & Wieringa 1994) and w-projection with 128 planes (Cornwell et al. 2008). Flux densities at the source positions (and image rms values) were extracted using the imtool program as part of the pwkit ¹⁹ package (Williams et al. 2017) (Table 2).

In addition to the detection of PTF10hgi, with F_ν = 47 ± 7 μJy, reported in Eftekhari et al. (2019), we detect radio emission near the location of the SLSN PTF12dam with a flux density F_ν = 117 ± 12 μJy; see Figure 1. This is comparable to the 3 GHz detection presented in Hatsukade et al. (2018) with F_ν = 141.5 ± 5.1 μJy. The radio emission is offset from the SN position by ∼1 arcsecond and is instead centered on, and traces, the optical emission from the host galaxy. Thus the emission is most likely related to star formation in the host galaxy (see discussion in Section 3). Indeed, the emission is resolved out in our second-epoch, A-configuration observation with a decreased flux density F_ν = 61.5 ± 10.8 μJy, indicative of an extended source. No clear radio emission (signal-to-noise ratio, S/N> 5σ) is detected from the remainder of the SLSNe in our sample, or from the location of any of the LGRBs in our sample, with typical rms values of 5–10 μJy.

Zoom In Zoom Out Reset image size

We report marginal detections near the positions of SN2009jh and GRB 050826. In the case of SN2009jh, the source significance is 3.6σ (F_ν = 23.3 ± 6.4 μJy) but with an offset of 16 ± 03 from the SN position, and we therefore consider it unlikely to be related to the SN. Similarly, for GRB 050826, the source significance is only 2.5σ (F_ν = 23 ± 9 μJy) with an offset of 10 ± 05 from the GRB position.

In addition to standard continuum observations, we also obtained phased-array observations with the VLA to search for FRBs from the observed sources. Similar searches for FRBs from the locations of SLSNe and LGRBs were presented in Law et al. (2019) and Hilmarsson et al. (2020), yielding no detections. We present here the results of our initial VLA sample of events observed in 2017. The phased-array data were recorded with 256 μs time resolution and 2 MHz channels with 2 GHz total bandwidth. Each raw filterbank file is divided into a channelized time series with 1 GHz bandwidth centered at 5 and 7 GHz. We performed a standard RFI search using the rfifind routine in PRESTO (Ransom 2001) with 2 s integration times. The RFI-excised maps are applied to the data for subsequent processing. For each source, we incoherently de-dispersed the data at 1000 trial dispersion measures (DMs) ranging up to DM = 5000 pc cm⁻³ with a step size of 5, corresponding to a maximum redshift z ∼ 7, well beyond the max redshift of our sample (z ∼ 0.57), and indeed well above the maximum observed DM for any FRB observed to date. ²⁰ This additionally allows for a substantial host galaxy DM contribution. Following de-dispersion, we normalize the time series by performing a standard red noise removal. Individual scans are then searched for FRBs using the matched-filtering algorithm single_pulse_search.py (Ransom 2001).

Following the prescription of Cordes & McLaughlin (2003), the minimum detectable flux density for an FRB is given by:

$$\begin{eqnarray}&&{S}_{\min }=\displaystyle \frac{{\left({\rm{S}}/{\rm{N}}\right)}_{\min }\mathrm{SEFD}}{\sqrt{{n}_{\mathrm{pol}}{\rm{\Delta }}\nu W}}\end{eqnarray} \tag{ 1 }$$

where n_pol is the number of summed polarizations, Δν is the bandwidth, W is the intrinsic pulse width, ${\left({\rm{S}}/{\rm{N}}\right)}_{\min }$ is the minimum signal-to-noise threshold, and SEFD refers to the system equivalent flux density. Assuming a phasing efficiency factor of 0.9 and a nominal 10 ms pulse width, we find a minimum detectable flux density of ${S}_{\min }\approx 16$ mJy (7σ) for our observations.

We do not detect any FRBs from our sample. We place limits on the maximum energy of an FRB emitted from each source following the prescription of Cordes & Chatterjee (2019), in which the burst energy is given by:

$$\begin{eqnarray}&&{E}_{{\rm{b}},\max }=4\pi {S}_{\nu }W{\rm{\Delta }}\nu {d}_{L}{f}_{b},\end{eqnarray} \tag{ 2 }$$

where we assume a beaming fraction f_b = 0.5. For the range of redshifts probed by our initial VLA sample of SLSNe (z ∼ 0.101 − 0.376), we find ${E}_{{\rm{b}},\max }\lesssim 2\,\times \,{10}^{37}\mbox{--}4\,\times \,{10}^{38}$ erg. This is comparable to the range of inferred burst energies for the lowest-energy bursts from FRB 121102, with E_b ∼ 10³⁷–10³⁸ erg (Gourdji et al. 2019). Thus, while we are sensitive to the low-energy range of FRBs (from FRB 121102), given the intermittent nature of FRB 121102, the short duration of our observations ($\approx 40\,\min$ per source) is prohibitively small. A significant time investment with the Green Bank Telescope (GBT), Arecibo, or the Five-hundred-meter Aperture Spherical Telescope (FAST) may reveal bursts from these sources.

We obtained millimeter observations with ALMA in Band 3 (∼100 GHz) of 28 SLSNe. Flux density and complex gain calibrators are listed in Table 3. In all cases, we use the ALMA data products, which utilize standard imaging techniques within CASA. Each field is imaged using MFS, Briggs weighting with a robust parameter of 0.5, and a standard gridding convolution function. Image scales span 0.04 − 0.2 arcseconds per pixel, with typical image sizes of about 1500 × 1500 pixels. Flux densities and image rms values were obtained using the imtool program as part of the pwkit package.

We do not detect millimeter emission from any of our sources with the exception of a 3σ detection near the location of SN2007bi, with a flux density F_ν = 55.3 ± 18.8 μJy (R.A. = 13^h19^m20226, decl. = $+08^\circ 55^{\prime} 43\buildrel{\prime\prime}\over{.} 85;$ J2000). To compare the position of the millimeter source relative to the SN position, we use archival images of SN2007bi from the Liverpool Telescope (Young et al. 2010), which are first matched to an absolute astrometric system. We find that the millimeter source is offset from the SN position by 1'' ± 06 and is thus unlikely to be related to the SN. We similarly compare the position of this source to that of the host galaxy of SN2007bi using a wide-field i-band image from the Panoramic Survey Telescope and Rapid Response System (Pan-STARRS1) 3π survey. We first perform relative astrometry between the PS1/3π image and the Liverpool image and find an astrometric tie uncertainty of σ_host ≈ 06. Comparing the position of the millimeter source to this image, we find that the source is offset from the host galaxy position by 15 ± 06. We thus conclude that the weak ALMA detection is unlikely to be related to the SN or the host galaxy.

In Figure 2 we plot the 6 GHz radio luminosity upper limits of our sample of SLSNe and LGRBs, compared to the 6–10 GHz light curves of other classes of transients, including relativistic SNe, normal SNe Ib/c, fast blue optical transients, and tidal disruption events (TDEs). We also include observations of LGRBs and SLSNe from the literature.

Zoom In Zoom Out Reset image size

We find that the late-time limits presented here probe a largely unexplored region of parameter space with timescales of δ t ≳ 10³ days and luminosities L_ν ∼ 10²⁸ − 10²⁹ erg s⁻¹ Hz⁻¹. With the exception of a few TDEs, most transients have been observed in the radio at earlier times. The radio limits for SLSNe are significantly fainter than the vast majority of cosmological GRBs, which are observed at earlier times (δ t ∼ 1 − 10³ days) with luminosities L_ν ≳ 10³⁰ erg s⁻¹ Hz⁻¹. Conversely, the limits do not reach the levels of the least luminous SNe Ib/c with L_ν ∼ 10²⁶ erg s⁻¹ Hz⁻¹. We explore the implications of these radio upper limits in the subsequent sections.

In the context of a shock interaction from the nonrelativistic, quasi-spherical SN ejecta, we place limits on the shock velocity and mass-loss rate for each SLSN following the prescription of Chevalier (1998) and assuming that the date of each observation corresponds to the peak luminosity; see Figure 4. We also consider the possibility that the emission from each event may have peaked at earlier times, by extrapolating the peak luminosity back in time assuming L_ν,pk ∝ t⁻¹ (Berger et al. 2002), and assuming that the emission is optically thin at the time of observations. This allows us to rule out the region of parameter space above each curve.

Zoom In Zoom Out Reset image size

We find that in this context, our limits at late times do not probe the range of mass-loss rates and ejecta velocities inferred for SNe Ib/c, which have mostly been detected at early times, with typical inferred values of v_ej ∼ 0.1c and A ∼ 1–100 A^* (e.g., Berger et al. 2002; Soderberg et al. 2005, 2012). On the other hand, a small number of SLSNe limits from the literature at earlier times probe the typical ejecta velocities and mass-loss rates inferred for SNe Ib/c, and thus suggest that comparable outflows are not ubiquitous in SLSNe. One possible exception is PTF10hgi, in which the radio detection would imply an ejecta velocity and progenitor mass-loss rate a few times larger than for SNe Ib/c. However, our interpretation for this event is that the radio emission is due to central engine activity.

For the LGRBs in our sample, we place constraints on emission from associated supernovae following the prescriptions of Barniol Duran & Giannios (2015) and Kathirgamaraju et al. (2016). In this scenario, radio emission is produced by a spherical outflow from the accompanying supernova, which peaks at the deceleration time (t_dec), when the SN has swept up a mass comparable to the initial ejected mass. During this phase, the light curve rises as ²¹ ∝t³ and decays after the deceleration time as ∝ t^−3(1+p)/10. We assume typical values of _e = _B = 0.1 and p = 2.5. We also include in our analysis late-time observations of three LGRBs from Peters et al. (2019).

In Figure 5, we plot representative light curves for external number densities of n = 10 and n = 100 cm⁻³ and a fiducial supernova energy of 10⁵² erg, typical of broad-line Ic SNe associated with GRBs (Mazzali et al. 2014). For GRB 020903 and GRB 030329, we plot individual curves corresponding to the inferred densities (n = 100 and n = 1.8 cm⁻³, respectively) from modeling of the GRB afterglows (Berger et al. 2003; Soderberg et al. 2004). We further fix the ejecta velocity to 8 × 10³ km s⁻¹, consistent with the median observed nebular-phase velocity for several broad-line Ic SNe (e.g., Mazzali et al. 2001, 2007b, 2007a). We note that this is in contrast to Peters et al. (2019), in which the authors use the SN ejecta velocity at ∼10 days after explosion, corresponding to velocities three to four times larger than the observed nebular velocities. This has a dramatic effect on the resulting light curves, which scale as ${\left({v}_{\mathrm{ej}}/c\right)}^{(11+p)/2}$ at t < t_dec. Indeed, we find that the light-curve luminosities are roughly an order of magnitude fainter than our LGRB limits on the same timescale assuming n = 100 cm⁻³, with the exception of deep limits for GRB 060218 and GRB 980425 from Peters et al. (2019), which rule out these models, but cannot rule out densities ≲ 10 cm⁻³. Furthermore, imposing the lower inferred nebular-phase velocities shifts the peak of the light curves as ${t}_{\mathrm{dec}}\propto {\left({v}_{\mathrm{ej}}/c\right)}^{-5/3}$ to t_dec ≳ 50 yr, well beyond the timescale of our observations at δ t ∼ 10 yr. Finally, we note that even for a more promising choice of SN parameters (large ejecta velocity and energy) our radio nondetections may still be unconstraining due to inhibition of the early (≲ decades) SN light curve that is predicted to be caused by interaction of the GRB jet with the interstellar medium (Margalit & Piran 2020).

Zoom In Zoom Out Reset image size

We thus conclude that we cannot place strong constraints on associated SNe in our sample of LGRBs, but note that such sources may be detectable on timescales of a few decades if they occur in high-density (n ≈ 100 cm⁻³) environments. Lower densities will require deep radio observations on century timescales.

To constrain the presence of off-axis jets, we produce a grid of afterglow models for a range of jet energies and CSM densities using the two-dimensional relativistic hydrodynamical code Boxfit v2 (van Eerten et al. 2012). We generate afterglow light curves at 6 and 100 GHz and compare these to our upper limits to determine the allowed region of parameter space for each event. We model the light curves assuming a jet opening angle of θ_j = 10° and viewing angles of θ_obs = 30°, 60°, and 90°. We further assume a constant-density CSM and jet microphysical parameters of _e = 0.1, _B = 0.01, and p = 2.5, following previous studies of LGRBs (e.g., Curran et al. 2010; Laskar et al. 2013; Wang et al. 2015; Laskar et al. 2016; Alexander et al. 2017a).

The results are summarized in Figure 6, where individual curves trace the jet energies and CSM densities corresponding to the 3σ flux density limit for each SLSN. In each case, the region of parameter space to the right of the curve (higher energies) is ruled out by the 6 and 100 GHz nondetections. We find that the radio limits largely preclude the presence of jets with E_iso ≳ 10⁵⁴ erg (n ∼ 10⁻³–10² cm⁻³) for θ_obs = 30° and 60°. For θ_obs = 90°, we cannot rule out jets with E_iso ≲ 10⁵⁴ erg and n ∼ 10⁻³–10⁻² cm⁻³.

Zoom In Zoom Out Reset image size

In Figure 7, we compare three representative off-axis jet light curves at 6 GHz to the limits for SLSNe, where we include all observations of SLSNe at 5–9 GHz from the literature (see Table 1). Individual curves depict jet energies of E_iso = 1.5 − 2.5 × 10⁵³ erg and CSM densities n = 1–100 cm⁻³, corresponding to a range of afterglow models that are consistent with the 6 GHz detection of PTF10hgi at δ t ∼ 7.5 yr (Eftekhari et al. 2019). We find that previous observations of SLSNe on timescales of a few hundred days post-explosion rule out the presence of jets with an energy scale similar to that inferred for PTF10hgi. Conversely, for the sample of SLSNe presented here, we generally cannot rule out the presence of similar jets, primarily because most of the SLSNe are located at larger distances than PTF10hgi.

Zoom In Zoom Out Reset image size

We also plot off-axis jet light curves at 100 GHz and compare to our ALMA upper limits in the right-hand panel of Figure 7. Because the peak of the afterglow emission is generically at GHz frequencies on such long timescales, our millimeter limits can accommodate higher jet energies of E_iso ≳ 10⁵⁴ erg. This can also be seen in Figure 6, where the ALMA limits trace out generally higher energies than the 6 GHz limits. We note that the two most constraining limits in the entire sample correspond to the lowest redshift sources in our sample, SN2018bsz and SN2018hti, which rule out off-axis jets with E_iso ≳ 10⁵³ erg.

We further rule out a population of off-axis jets with an energy scale significantly larger than is seen in typical LGRBs (Figure 6). In this context, observations at earlier times (δ t ≲ 1 yr) are more constraining, as these probe the peak of the afterglow emission.

As noted in Margalit et al. (2018b), a successful jet will break out of the SN ejecta only if the jet head velocity exceeds the ejecta velocity, and if the kink stability criterion is satisfied, i.e., that magnetic energy is not dissipated due to kink instabilities (Bromberg & Tchekhovskoy 2016). These conditions lead to a minimum threshold energy for successful jet breakout given by:

$$\begin{eqnarray}&&{E}_{\min }\simeq 0.195\,{E}_{K}{\left(\displaystyle \frac{{\gamma }_{j}}{2}\right)}^{-4}{f}_{j}^{-1},\end{eqnarray} \tag{ 4 }$$

where E_K is the initial kinetic energy of the SN, γ_j = 1/5θ_j as per Mizuta & Ioka (2013), and f_j is the fraction of energy contributing to a collimated jet (Margalit et al. 2018b). We assume f_j = 0.5 and adopt the inferred E_K values for each source based on fits to the optical light curves. For each SLSN in our sample, ²² we therefore estimate the minimal energy for jet breakout to determine whether our nondetections can constrain the theoretical predictions. The results are listed in Table 5.

Table 5. SLSN Magnetar Parameters and CLOUDY Results

Source	P	B	Mej	EK	vej	tff, 6 GHz	tff, 100 GHz a	${E}_{\min ,\mathrm{iso}}$ b
	(ms)	(1014 G)	(M⊙)	(1051 erg)	(103 km s−1)	(yr)	(yr)	(1051 erg)
SN2005ap	1.28	1.71	3.57	8.85	15.22	2.43	< 0.16	1108
SN2006oz	2.70	0.32	2.97	2.66	9.46	2.05	< 0.19	333
SN2007bi	3.92	0.35	3.80	2.37	7.90	5.94	1.94	297
SN2009jh	1.74	0.27	6.98	6.78	9.11	4.52	1.70	849
PTF09cnd	1.46	0.10	5.16	3.29	8.56	1.33	< 0.16	412
PTF10hgi	4.78	2.03	2.19	0.55	5.12	4.85	1.39	69
SN2010gx	3.66	0.59	2.39	3.78	12.65	3.63	1.24	473
SN2010kd c	3.51	0.57	10.51	3.64	5.90	16.13	4.43	456
SN2011ke	0.78	3.88	7.64	5.22	8.15	2.90	< 0.13	653
SN2011kf	1.48	0.70	4.57	6.72	11.46	3.11	< 0.15	841
SN2011kg	2.07	2.88	6.54	7.97	12.11	6.98	1.86	998
SN2012il	2.35	2.24	3.14	1.94	7.93	3.95	< 0.14	243
PTF12dam	2.28	0.18	6.27	3.03	7.01	4.28	1.81	379
LSQ12dlf	2.82	1.20	3.68	2.54	8.28	5.72	1.66	318
SSS120810	3.00	1.93	2.22	2.82	11.13	3.42	< 0.14	353
SN2013dg	3.50	1.56	2.75	1.85	8.38	5.19	1.50	232
LSQ14bdq	0.98	0.49	33.71	25.06	8.71	16.67	4.57	3137
LSQ14mo	4.97	1.01	2.10	2.43	10.74	4.77	1.44	304
SN2015bn	2.16	0.31	11.73	3.45	5.46	12.00	3.49	432
OGLE15sd	2.16	1.74	1.91	6.29	9.33	9.17	2.53	787
iPTF16bad	3.73	2.62	2.22	1.12	7.11	4.55	1.29	140
SN2016ard d	0.93	1.55	16.6	33.0	14.2	7.80	<2.11	4131
SN2016els c	0.92	5.38	11.83	28.0	15.43	5.50	1.39	3505
SN2017gci c	1.26	3.46	11.74	7.16	7.83	6.86	1.81	896
SN2017jan c	3.08	0.34	7.14	4.58	8.03	10.84	3.26	573
SN2018hti c	1.25	2.59	31.04	11.3	6.06	15.67	3.80	1415
SN2018ibb c	0.74	0.16	43.47	45.6	10.27	12.12	3.67	5708
SN2018lfe e	2.85	2.2	3.80	3.82	10.05	6.04	1.71	478

Notes. Magnetar parameters are from Nicholl et al. (2017c) except where specified. All times refer to the observer frame.

^a We report upper limits where t_ff,100GHz < 1 yr(1 + z), as these results are based on extrapolations to early times. ^b Minimum energy required for a 10° jet to break out of the SN ejecta. ^c Parameters derived in this work. ^d From Blanchard et al. (2018). ^e From Y. Yin et al. (2021, in preparation).

Download table as:  ASCIITypeset image

Among our VLA sources, we find that four SLSNe (SN2010kd, SN2011ke, SN2011kf, and SN2011kg) have radio limits that preclude the presence of a relativistic jet. That is, for the afterglow model parameters specified above, the radio limits probe jet energies below ${E}_{\min }$. Thus, a successful jet produced by one of these sources would be readily detected in our observations. The remaining VLA sources have jet breakout energies that are not ruled out by our observations, and thus may harbor jets that are below the sensitivity of our observations, but with ${E}_{\mathrm{iso}}\gtrsim {E}_{\min }$.

Similarly, among our ALMA sources, we can rule out the presence of off-axis jets at an observer viewing angle θ_obs ≲ 30° for 13 sources (SSS120810, SN2013dg, LSQ14bdq, LSQ14mo, OGLE15sd, SN2015bn, SN2016ard, SN2016els, SN2017gci, SN2017jan, SN2018hti, SN2018ibb, and SN2018lfe) given that the jet breakout energy is ruled out by our 3σ radio limits. For SN2016ard, we further rule out all jets with θ_obs ≲ 60°, as well as jets with θ_obs ≲ 90° and n ≳ 10⁻² cm⁻³. Similarly, for LSQ14bdq, SN2016els, and SN2018ibb, we can rule out θ_obs ≲ 60° and n ≳ 10⁻² cm⁻³ and θ_obs ≲ 90° and n ≳ 10⁻¹ cm⁻³. For SN2017gci and SN2018hti, we can rule out all jets with θ_obs ≲ 60° and jets with θ_obs ≲ 90° and n ≳ 10⁻¹ cm⁻³. Thus, for SN2016ard, LSQ14bdq, SN2016els, SN2018ibb, SN2017gci, and SN2018hti, the allowed phase-space for a successful jet that is not ruled out by our observations is prohibitively small.

Finally, we note that the range of jet breakout energies (${E}_{\min ,\mathrm{iso}}\approx 7\,\times \,{10}^{52}\mbox{--}6\,\times \,{10}^{54}$ erg) for the SLSNe in our sample suggests that we do not expect a population of off-axis jets with energies comparable to the lowest-energy LGRBs, e.g., E_iso ≲ 10⁵³ erg, given that such jets will fail to break out of the SN ejecta.

Margalit & Metzger (2018) proposed a magnetized ion–electron wind from a young magnetar to explain the observed properties of FRB 121102, namely the size and flux of the persistent radio counterpart, as well as the large and decreasing RM of the bursts. The model assumes a one-zone nebula in which the magnetar's magnetic energy leads to the injection of relativistic electrons and ions that are thermalized at the termination shock of the magnetar wind. The model is characterized by the magnetic energy of the magnetar (${E}_{{B}_{* }}$), the nebula velocity (v_n ), the rate of energy input into the nebula (t^−α ) following the onset of the active period (t₀), the magnetization of the outflow (σ), and the mean energy per particle (χ) assuming a proton–electron composition. An ion–electron plasma composition is invoked given that the RM contribution of an electron–positron plasma is zero and thus inconsistent with the large observed RM (Michilli et al. 2018).

In Figure 8, we plot models A and B from Margalit & Metzger (2018), corresponding to inferred source ages of t_age ≈ 12.4 and 38 yr, respectively. In each case, the emission is expected to peak in the millimeter regime on timescales of a few years, and in the gigahertz regime on timescales of ≳ 10 yr. Moreover, the models predict 6 GHz luminosities of ∼ 10²⁹ erg s⁻¹ Hz⁻¹, consistent with the luminosity of the FRB 121102 persistent radio source (Chatterjee et al. 2017). We find that our limits at both 6 and 100 GHz are sufficient to rule out these models, with the same parameters as for FRB 121102.

Zoom In Zoom Out Reset image size

We further compare our limits to the models used to describe the 6 GHz radio detection and 100 GHz upper limit of PTF10hgi from Eftekhari et al. (2019; see Figure 8). The first model is identical to model A for FRB 121102 from Margalit & Metzger (2018) with the magnetic energy scaled down by a factor of ≈ 20, while the second model explores a scenario in which the 6 GHz emission is marginally self-absorbed. In Eftekhari et al. (2019), we showed that a fully self-absorbed nebula (as constrained by the 100 GHz non-detection) is also consistent with the data. However, recent results presented by Mondal et al. (2020) constrain the self-absorption frequency to ∼1 GHz based on a steep drop in flux density between 0.6 and 1 GHz. Indeed, we find that the predicted light-curve evolution of a marginally self-absorbed nebula is more consistent with the observed evolution of the 6 GHz luminosity.

We find that the majority of our limits are consistent with the model light curves invoked for PTF10hgi, although several deep limits at 100 GHz, most notably SN2018bsz with L_ν = 7.6 × 10²⁶ erg s⁻¹ Hz⁻¹, strongly rule out the presence of such nebulae. We note that SN2018bsz exhibits unusual properties, including a pre-maximum plateau in the optical light curve in addition to strong carbon lines, which suggest that it may be an atypical event among SLSNe (Anderson et al. 2018). Although the timescale of our observations probes the predicted peak of emission, the sensitivity of our VLA and ALMA observations is not sufficient to rule out emission at the level of L_ν ≲ 10²⁸ erg s⁻¹ Hz⁻¹. Thus, if ion–electron nebulae similar to the one inferred for PTF10hgi are ubiquitous in SLSNe, such sources will require deep radio observations, and may be more readily detected with future instruments such as the Square Kilometer Array.

It is worth noting however that PTF10hgi is among the nearest sources in our sample with z = 0.0987. Thus our nondetections for the remainder of the events may be indicative of their distance. Only two sources are located at lower redshifts: SN2018hti and SN2018bsz with z = 0.063 and z = 0.027, respectively, both of which are observed at 100 GHz. In the case of SN2018hti, we find that the SN ejecta is optically thick to 100 GHz emission at the time of our observations, while magnetar parameters have not been derived for SN2018bsz. In this context, a non-detection for SN2018hti is thus unsurprising, while nondetections for the remaining events may simply point to their greater distance relative to PTF10hgi.

Finally, the fact that the SLSNe radio limits preclude FRB 121102 models suggests that there may be physical differences (e.g., magnetic field strength) between the magnetar engines responsible for powering SLSNe and that inferred for FRB 121102.

Here we consider a modified version of the PWN model presented in Omand et al. (2018; see also Murase et al. 2020), following the prescription of Murase et al. (2016) for quasi-steady radio emission from magnetar engines in SLSNe and FRBs. An electron–positron wind nebulae has also been considered to explain the observed properties of FRB 121102 (Murase et al. 2016; Kashiyama & Murase 2017). In this work, we modify the spin-down formula and neutron star masses, and ignore the effects of ejecta feedback for consistency with the MOSFiT models. The radio models solve the Boltzmann equation for photons and electron/positrons in the PWN over all photon frequencies and electron energies (Murase et al. 2015), allowing for a self-consistent calculation of pair cascades, Compton and inverse Compton scattering, adiabatic cooling, and both internal and external attenuation. The electron–positron injection spectrum is assumed to be a broken power law with injection spectral indices of q₁ = 1.5 and q₂ = 2.5 and a peak Lorentz factor of γ_b = 10⁵, which is consistent with Galactic PWNe (e.g., Tanaka & Takahara 2010, 2013) such as the Crab PWN, as well as the inferred nebula for PTF10hgi with q₁ = 1.3 (Mondal et al. 2020). Free–free absorption in the ejecta is calculated assuming a singly ionized oxygen ejecta, and we do not consider absorption outside the ejecta, as in Omand et al. (2018) and Law et al. (2019).

The results of the models are shown in Figure 9 where we plot the predicted light curves at 6 and 100 GHz for the SLSNe in our sample with existing MOSFiT parameters (Table 5). The models generically predict emission that peaks initially in the millimeter regime with an increased flux density on timescales of ∼1000 days and cascades to lower frequencies and lower flux densities at later times. In this context, the timescales of our ALMA observations lead to less constraining limits, as they do not probe the peak of the emission, which is expected at earlier times. In a few cases, we also include models at 3 GHz for comparison to limits and models presented in Law et al. (2019). We generate new 3 GHz models using the formalism described above for self-consistency.

Zoom In Zoom Out Reset image size

For the majority of our sources, we find that the nondetections are consistent with the predicted light curves. At 6 GHz, a number of sources are expected to peak at later times relative to the timescale of our observations, and hence may be detected in the future with continued monitoring. We note that five sources exhibit predicted emission at or above the level of our 3σ limits at the time of our observations. This includes SN2006oz, PTF09cnd, and SN2011kf at 6 GHz, SN2010gx at 3 and 6 GHz, and SN2007bi at 3, 6, and 100 GHz. For an additional six sources (SN2009jh, SN2010kd, SN2012il, SN2017jan, SN2018ibb, and SN2018hti), the limits exclude models without absorption but cannot rule out models with absorption.

In Figure 10, we plot the predicted peak luminosity for each SLSNe (assuming maximal absorption) as a function of the inferred SLSN magnetic field and spin period. We find a strong anticorrelation between the peak luminosity and magnetic field, i.e., the more luminous events correspond to lower magnetic fields. This is consistent with the fact that the assumed pulsar spin-down luminosity L_em ∝ B⁻² as per numerical simulations (Gruzinov 2005). Indeed, the five sources with predicted emission at or above the level of our 3σ limits correspond to exclusively low-B (B ≲ 10¹⁴ G) events. Interestingly, from Figure 10, we also find that our VLA and ALMA observations are sensitive to a large fraction of low-B events, and thus indicate that continued monitoring on the appropriate timescales may reveal emission from these sources. Conversely, we find no clear correlation between the peak luminosity and the spin period.

Zoom In Zoom Out Reset image size

For PTF10hgi, we plot the light curves over the range 0.6–100 GHz in Figure 11. We find that the observations at 6 GHz at δ t ≈ 8 yr (Eftekhari et al. 2019) are consistent with the model with no absorption, while the same models underpredict the data at δ t ≈ 10 yr (Mondal et al. 2020). At 3 GHz, the models slightly underpredict the data at δ t ≈ 8 yr (Law et al. 2019); however, this difference is negligible in light of systematic uncertainties. The 3 GHz light curve is predicted to rise between δ t ≈ 8–10 yr, consistent with the shape of the observed data, but with a scale significantly below the observations. Nondetections at both 0.6 and 100 GHz are consistent with the models. Conversely, the data at 1.2 and 15 GHz lie well above the predicted light curves.

Zoom In Zoom Out Reset image size

In Law et al. (2019), the authors use unique magnetar parameters for PTF10hgi (B = 1.4 × 10¹⁴ G, P = 1 ms, M_ej = 15 M_⊙ based on by-eye estimates), and find that a model in which ∼ 40% of the ejecta is singly ionized is consistent with the observed data. However, such a model fails to accurately predict the latest SED obtained in Mondal et al. (2020). Namely, the models slightly underpredict the data at 1.2–6 GHz, but vastly underpredict the data at 15 GHz. This is in contrast to the models presented here in which the low-frequency model at 1.2 GHz drastically underpredicts the observed flux value. This suggests that the true magnetar parameters may lie somewhere between the inferred values in both models (e.g., P ≈ 2–4 ms, M_ej ≈ 6–9 M_⊙). On the other hand, the flat shape of the SED at 1–15 GHz may indicate a steeper injection spectrum, or an ionization fraction that evolves as a function of time. Continued broadband radio monitoring will provide a test of these scenarios.

In Figure 12 we plot the same data and models, but with a peak Lorentz factor of γ_b = 10⁴ instead of 10⁵; we find that these light curves are more consistent with the data. Specifically, we find that the models are below the non-detection upper limits at 0.6 and 100 GHz, and are consistent with the 3 GHz observations assuming no absorption and with the 6 GHz observations assuming partial absorption at δ t ≈ 6 yr and almost maximal absorption at δ t ≈ 8 yr. The models are consistent with the data at 15 GHz regardless of absorption (the ejecta are predicted to be optically thin for free–free absorption at 15 GHz at the relevant timescale). Conversely, the model vastly underpredicts the 1.2 GHz data, which are predicted to peak at later times.

Zoom In Zoom Out Reset image size

One interesting physical implication of decreasing γ_b is that the pair multiplicity in PTF10hgi is significantly higher than that of the Crab pulsar or other Galactic PWNe, which may indicate that the pair formation or acceleration processes in the nebulae of supernovae driven by highly magnetic millisecond pulsars are qualitatively different than those of Galactic PWNe. Nevertheless, it is unclear whether this is unique to SLSNe nebulae due to the luminosity of the nebula and strength of the pulsar field, or if SLSNe nebulae eventually evolve to have a lower multiplicity.

SN1999as (z ≈ 0.127) was discovered on 1999 February 18 by the Supernova Cosmology Project (Knop et al. 1999). From Knop et al. (1999), the peak time is MJD 51242. Given its similarity to SN2007bi, we assume a 62 day rest-frame rise time, corresponding to an explosion date of MJD 51172. Spectroscopic data are given in Hatano et al. (2001). Observations of the host galaxy are given in Leloudas et al. (2015), Angus et al. (2016), and Schulze et al. (2018).

SN2005ap (z ≈ 0.28) was discovered on 2005 March 3 in images taken with the Robotic Optical Transient Search Experiment Telescope (ROTSE-IIIb; Akerlof et al. 2003) as part of the Texas Supernova Search (Quimby 2006). SN2005ap exhibited a 1–3 week rise to peak and a relatively rapid decay (Quimby et al. 2007). We assume a rest-frame rise time of 12 days and a peak time of MJD 53439 (Quimby et al. 2011b), corresponding to an explosion date of 53424. Radio and optical observations of the host are presented in Schulze et al. (2018). Additional host galaxy observations are given in Lunnan et al. (2014).

SN2006oz (z ≈ 0.38) was discovered on 2006 October 20 by the Sloan Digital Sky Survey II. We assume an explosion date of MJD 54025. Light curves and spectroscopy are given in Leloudas et al. (2012). Host galaxy observations are given in Lunnan et al. (2014), Leloudas et al. (2015), and Schulze et al. (2018).

SN2007bi (z ≈ 0.13) was discovered on 2007 April 6 by the Nearby Supernova Factory (Nugent 2007). The SN exhibited a slow rise time of 77 days with a peak time of MJD 54152, corresponding to an explosion date of MJD 54075 (Gal-Yam et al. 2009). Host galaxy observations are given in Lunnan et al. (2014) and Schulze et al. (2018). The relatively slow decay time and the large inferred mass of radioactive ⁵⁶Ni have been used to argue for a pair-instability supernova explosion (Gal-Yam et al. 2009), although Nicholl et al. (2013) have shown that the event properties are well matched to a magnetar central engine with a modest ejecta mass.

SN2009jh (=PTF09cwl=CSS090802:144910+292510; z ≈ 0.35) was first discovered in the Catalina Sky Survey on 2009 August 2 (Drake et al. 2009) and independently during commissioning of the PTF system (Quimby et al. 2011b). From Quimby et al. (2011b), the peak time is MJD 55081 and the rest-frame rise time is ∼50 days, corresponding to an explosion date of 55014. Light curves and spectroscopy are given in De Cia et al. (2018) and Quimby et al. (2018). Observations of the host galaxy are presented in Leloudas et al. (2015), Angus et al. (2016), Perley et al. (2016), and Schulze et al. (2018). Extensive X-ray observations with Swift-XRT spanning δ t = 48–1961 days reveal no X-ray source at the location of the SN (Levan et al. 2013; Margutti et al. 2018).

PTF09cnd (z ≈ 0.26) was first detected on 2009 July 13 by the Palomar Transient Factory with the 1.2 m Samuel Oschin Telescope during commissioning of the PTF system (Quimby et al. 2011b). From Inserra et al. (2013), the peak time is MJD 55069.145 and the rest-frame rise time is ∼50 days. Thus the explosion date is MJD 55006. Spectra for PTF09cnd are given in Quimby et al. (2018). Observations and properties of the host galaxy are presented in Neill et al. (2011), Leloudas et al. (2015), and Perley et al. (2016). A previous search for FRBs from this event was conducted in Hilmarsson et al. (2020). X-ray nondetections place limits on the unabsorbed fluxes between 10⁻¹³–10⁻¹⁵ erg cm⁻² s⁻¹ (Levan et al. 2013; Margutti et al. 2018).

SN2010kd (z ≈ 0.23) was discovered on 2010 November 14 by ROTSE-IIIb (Akerlof et al. 2003). From Vinko et al. (2012), we assume a peak time of MJD 55554 and a rest-frame rise time of 50 days, corresponding to an explosion date of MJD 55499. Host galaxy observations are given in Leloudas et al. (2015) and Schulze et al. (2018). X-ray observations of the source span δ t ∼ 120–1964 days post-explosion (rest frame) and correspond to limits on the unabsorbed X-ray flux of 10⁻¹³–10¹⁵ erg cm⁻² s⁻¹ (Levan et al. 2013; Margutti et al. 2018).

SN2010gx (=PTF10cwr=CSS100313:112547-084941; z ≈ 0.23) was discovered by the Catalina Real-time Transient Survey on 2010 March 13 (Mahabal et al. 2010). From Inserra et al. (2013), the peak time is MJD 55279, and the rest-frame rise time is ∼23 days; the explosion date is therefore MJD 55251. Light curves and spectra are given in Quimby et al. (2011b) and Inserra et al. (2013). Host galaxy observations are given in Chen et al. (2013), Lunnan et al. (2014), Leloudas et al. (2015), Perley et al. (2016), and Schulze et al. (2018). A previous search for FRBs from this event was conducted in Hilmarsson et al. (2020). Swift-XRT observations spanning δ t = 19–659 days reveal no X-ray source at the location of the SN (Levan et al. 2013; Margutti et al. 2018).

SN2011ke (=PTF11dij=CSS110406:135058+261642=PS1-11xk; z ≈ 0.14) was discovered by the Catalina Real-time Transient Survey on 2011 April 4 (Drake et al. 2011) and independently by the Palomar Transient Factory on 2011 March 30 (Quimby et al. 2011c). A non-detection of the event one day prior to the 2011 March 30 detection constrains the explosion date to MJD 55650.65 (Inserra et al. 2013). Host galaxy observations are given in Lunnan et al. (2014), Leloudas et al. (2015), Perley et al. (2016), and Schulze et al. (2018). Swift-XRT observations span δ t ∼ 40 − 1604 days post-explosion and reveal no X-ray source at the transient position (Levan et al. 2013; Margutti et al. 2018).

SN2011kf (=CSS111230:143658+163057; z ≈ 0.25) was discovered on 2011 December 30 by the Catalina Real-time Transient Survey (Drake et al. 2012). Additional spectra were obtained by Prieto et al. (2012). From Inserra et al. (2013), the inferred explosion date is MJD 55920.65. Host galaxy observations are given in Lunnan et al. (2014), Leloudas et al. (2015), and Schulze et al. (2018).

SN2011kg (=PTF11rks; z ≈ 0.19) was discovered on 2011 December 21 by the Palomar Transient Factory (Quimby et al. 2011a). From Inserra et al. (2013), the explosion date is MJD 55912.1. Host galaxy photometry and spectroscopy are given in Lunnan et al. (2014) and Perley et al. (2016). Swift-XRT observations reveal no X-ray source at the position of the SN over the time period δ t ∼ 11–25 days (Levan et al. 2013; Margutti et al. 2018).

SN2012il (=PS1-12fo=CSS120121:094613+195028; z ≈ 0.18) was discovered by the Pan-STARRS1 3Pi Faint Galaxy Supernova Survey on 2012 January 19 (Smartt et al. 2012a) and independently by the Catalina Real-time Transient Survey on 2012 January 21 (Drake et al. 2012). From Inserra et al. (2013), the explosion date is MJD 55918.56. Observations of the host galaxy are presented in Lunnan et al. (2014), Leloudas et al. (2015), and Schulze et al. (2018). No X-ray emission is detected at the location of the source down to ∼ 10⁻¹⁶ erg s⁻¹ cm⁻² (Levan et al. 2013; Margutti et al. 2018).

PTF12dam (z ≈ 0.11) was discovered by the Palomar Transient Factory on 2012 April 10 (Quimby et al. 2012), and is among the subset of slowly evolving SLSNe (Nicholl et al. 2013; Inserra et al. 2017; Vreeswijk et al. 2017; Quimby et al. 2018). From Nicholl et al. (2013), the peak date is MJD 56088, and the rest-frame rise time is ∼60 days; the inferred explosion date is thus MJD 56022. Extensive studies of the host galaxy are presented in the literature (Lunnan et al. 2014; Chen et al. 2015; Leloudas et al. 2015; Thone et al. 2015; Perley et al. 2016; Hatsukade et al. 2018). The source is not detected in Swift-XRT observations over the timescale δ t ∼ 43 − 900 days down to a limiting flux of F_X ∼ 5 × 10⁻¹⁴ erg s⁻¹ cm⁻² (Margutti et al. 2018). Chandra X-ray observations over the range δ t ∼ 60–68 days reveal an X-ray source with an unabsorbed flux F_x = (7.3 ± 2.9) × 10⁻¹⁶ erg s⁻¹ cm⁻² (0.3–10 keV; Margutti et al. 2018). A previous search for FRBs from this event was conducted in Hilmarsson et al. (2020).

LSQ12dlf (z ≈ 0.26) was discovered by the La Silla-Quest survey (LSQ; Baltay et al. 2013) on 2010 July 10 and subsequently classified as an SLSN by the Public ESO Spectroscopic Survey of Transient Objects (PESSTO; Smartt et al. 2012b). From Nicholl et al. (2014), the peak time is MJD 56,149 and the rest-frame rise time is approximately ∼24 days. The inferred explosion date is therefore MJD 56119. A previous search for FRBs from this event was conducted in Hilmarsson et al. (2020). Host galaxy observations are given in Schulze et al. (2018).

SSS120810 (z ≈ 0.16) was discovered on 2012 August 10 by the Catalina Real-time Transient Survey and subsequently classified as an SLSN by PESSTO (Wright et al. 2012). From Nicholl et al. (2014), the peak time is MJD 56146. Assuming a rest-frame rise time of 26 days, the explosion date is MJD 56116. Observations of the host galaxy are given in Leloudas et al. (2015) and Schulze et al. (2018).

SN2013dg (=CSS130530:131841-070443=MLS130517:131841-07044; z ≈ 0.27) was discovered on 2013 May 17 by the Mount Lemmon Survey and independently by the Catalina Sky Survey on 2018 May 30 (Drake et al. 2013). From Nicholl et al. (2014), the peak time is MJD 56449 and the rest-frame rise time is approximately 24 days. The inferred explosion date is therefore MJD 56419. Host galaxy observations are given in Schulze et al. (2018).

LSQ14bdq (z ≈ 0.35) was discovered by the LSQ on 2014 April 5 and subsequently classified by PESSTO (Benitez et al. 2014). The light curve exhibits an initial peak lasting ∼15 days followed by a slower rise to maximum light (Nicholl et al. 2015). Nondetections prior to the initial peak constrain the explosion date to MJD 56721. Host galaxy observations are given in Schulze et al. (2018).

LSQ14mo (z ≈ 0.25) was discovered by the LSQ on 2014 January 30 and subsequently classified by PESSTO as an SLSN on 2014 January 31 (Leloudas et al. 2014). From Leloudas et al. (2015), the peak time is MJD 56699. Assuming a rest-frame rise time of 50 days, the inferred explosion date is MJD 56636. Host galaxy observations are given in Chen et al. (2017) and Schulze et al. (2018). Swift-XRT observations over the timescale δ t ∼ 52–774 days rest-frame post-explosion do not show evidence for X-ray emission (Margutti et al. 2018).

LSQ14an (z ≈ 0.16) was discovered by the LSQ on 2014 January 3 and classified as an SLSN by PESSTO (Leget et al. 2014). The light curve indicates that LSQ14an is part of the class of slowly evolving SLSN (Inserra et al. 2017). From Jerkstrand et al. (2017), the peak time is MJD 56595. Assuming a rest-frame rise time of ∼70 days, the inferred explosion date is MJD 56513. Host galaxy observations are given in Schulze et al. (2018). No X-ray emission is detected at the location of the source based on Swift-XRT observations spanning δ t ∼ 64–234 days rest-frame post-explosion (Margutti et al. 2018).

LSQ14fxj (z ≈ 0.36) was discovered by the LSQ on 2014 October 12 and classified as an SLSN I by PESSTO (Smith et al. 2014). From Smith et al. (2014), on MJD 56940, the source was 4−5 weeks post-maximum light. We assume a rest-frame rise time of 50 days, as in Margutti et al. (2018). The explosion date is MJD 56882. Host galaxy observations are given in Schulze et al. (2018). Swift-XRT observations of the source span δ t ∼ 64–234 days rest-frame post-explosion, and do not reveal any X-ray emission at the location of the source (Inserra et al. 2017; Margutti et al. 2018).

CSS140925 (z ≈ 0.46) was discovered on 2014 September 25 by the Catalina Sky Survey and classified by PESSTO (Campbell 2014). From the CRTS catalog (http://nesssi.cacr.caltech.edu/catalina/AllSN.html), the explosion date is MJD 56900. No X-ray emission is detected at the SN location over the timescale δ t ∼ 29–186 days rest-frame post-explosion (Margutti et al. 2018). Host galaxy observations are given in Schulze et al. (2018).

SN2015bn (=PS15ae=CSS141223-113342+004332=MLS150211-113342 +004333; z ≈ 0.11) was initially discovered by the Catalina Sky Survey on 2014 December 23 (Drake et al. 2009). It was subsequently discovered by the Mount Lemmon Survey on 2015 February 11 and the Pan-STARRS Survey for Transients on 2015 February 15 (Huber et al. 2015). Host galaxy observations are presented in Schulze et al. (2018). Spectroscopy and UV to near-IR photometry spanning −50 to +250 days from optical maximum revealed a slowly fading source with undulations in the light curve on a timescale of 30–50 days (Nicholl et al. 2016b). Late-time, nebular-phase observations show evidence for a central engine (Nicholl et al. 2016a; Jerkstrand et al. 2017). Radio emission is not detected at 320–335 days post-explosion (Nicholl et al. 2016b). X-ray observations correspond to upper limits on the unabsorbed flux in the range 10⁻¹⁴–10⁻¹³ erg cm⁻² s⁻¹ (Margutti et al. 2018).

OGLE15sd (z ≈ 0.57) was discovered on 2015 November 7 by the Optical Gravitational Lensing Experiment (OGLE; Wyrzykowski et al. 2015). From the OGLE-IV Transient Detection System, the explosion date is MJD 57295 (http://ogle.astrouw.edu.pl/ogle4/transients/transients.html). No X-ray emission is detected at the SN location over the timescale δ t ∼ 34–212 days rest-frame post-explosion (Margutti et al. 2018).

iPTF16bad (z ≈ 0.2467) was discovered on 2016 May 31 by the Intermediate Palomar Transient Factory (iPTF; Yan et al. 2017). The event is among a small subset of SLSNe displaying Hα emission at late times. We constrain the explosion date to MJD 57513 by fitting the optical light curve to a magnetar model using the Modular Open-Source Fitter for Transients (MOSFit).

SN2016ard (=PS16aqv=CSS160216:141045-100935; z ≈ 0.16) was discovered on 2016 February 20 by the Pan-STARRS near-Earth object survey (Chambers et al. 2016) and classified by Chornock et al. (2016). From Blanchard et al. (2018), the peak time is MJD 57453 and the rest-frame rise time is 17 days, corresponding to an explosion date of MJD 57433.4. Swift-XRT observations span δ t ∼ 53–131 days rest-frame since explosion and do not show evidence for X-ray emission at the SN location (Margutti et al. 2018).

SN2016els (=AT2016els=PS16dnq; z ≈ 0.217) was discovered on 2016 July 29 by PESSTO (Mattila et al. 2016). From the Open Supernova Catalog, the peak date is MJD 57604. We assume a 50 day rest-frame rise time for an explosion date of MJD 57543.

SN2017gci (=AT2017gci=Gaia17cbp; z ≈ 0.09) was discovered on 2017 August 12 by the Gaia Photometric Survey (Delgado et al. 2017) and subsequently classified as an SLSN by ePESSTO (Lyman et al. 2017). We constrain the explosion date to MJD 57939 by fitting the optical light curve to a magnetar model using MOSFit.

SN2017jan (=AT2017jan=OGLE17jan; z ≈ 0.396) was discovered on 2017 December 18 by OGLE (Wyrzykowski & Gromadzki 2017) and subsequently classified by ePESSTO (Angus et al. 2017). We constrain the explosion date to MJD 57986 by fitting the optical light curve to a magnetar model using MOSFit.

SN2018bsz (=ASASSN-18km=AT2018bsz=ATLAS18pny; z ≈ 0.0267) was discovered on 2018 May 17 by the All Sky Automated Survey for SuperNovae (Brimacombe et al. 2018) and detected independently by the Asteroid Terrestrial-impact Last Alert System (ATLAS) on 2018 May 21. From Anderson et al. (2018), the explosion date is estimated as the mid-point between the last non-detection epoch and the discovery date and is given by MJD 58202.5. Properties of the host galaxy are also given in Anderson et al. (2018).

SN2018gft (=AT2018gf=ZTF18abshezu=ATLAS18uymActions; z ≈ 0.23) was discovered on 2018 September 2 by the Zwicky Transient Facility (ZTF; Fremling 2018a). We constrain the explosion date to MJD 58355 by fitting the optical light curve to a magnetar model using MOSFit.

SN2018ffj (=AT2018ffj=ATLAS18tec; z ≈ 0.234) was discovered on 2018 August 7 by ATLAS (Tonry et al. 2018a) and classified as an SLSN by ePESSTO on 2018 August 19 (Kostrzewa-rutkowska et al. 2018). From the Open Supernova Catalog, the peak date is MJD 58337. We assume a 50 day rest-frame rise time, corresponding to an explosion date of MJD 58275.

SN2018ffs (=AT2018ffs=ATLAS18txu=ZTF18ablwafp; z ≈ 0.142) was discovered on 2018 August 13 by ATLAS (Tonry et al. 2018a) and classified by ePESSTO on 2018 September 16 (Gromadzki et al. 2018). We constrain the explosion date to MJD 58324 by fitting the optical light curve to a magnetar model using MOSFit.

SN2018hti (=AT2018hti=ATLAS18yff=MLS181110:034054+114637=Gaia19amt; z ≈ 0.063) was discovered on 2018 November 2 by ATLAS (Tonry et al. 2018b) and classified by the Global Supernova Project (Burke et al. 2018). We constrain the explosion date to MJD 58415 by fitting the optical light curve to a magnetar model using MOSFit.

SN2018ibb (=AT2018ibb=ATLAS18unu=ZTF18acenqto=Gaia19cvo; z ≈ 0.16) was discovered on 2018 September 10 by ATLAS (Tonry et al. 2018c) and subsequently classified by ZTF (Fremling et al. 2018). From the Open Supernova Catalog, the peak date is MJD 58465. We assume a 50 day rest-frame rise time, corresponding to an explosion date of MJD 58407.

SN2018jfo (=AT2018jfo=ZTF18achdidy=MLS181220:112339+255952; z ≈ 0.163) was discovered on 2018 November 10 by ZTF (Fremling 2018b) and subsequently classified as an SLSN on 2019 January 5 (Fremling et al. 2019). We constrain the explosion date to MJD 58411 by fitting the optical light curve to a magnetar model using MOSFit.

SN2018lfe (=AT2018lfe=PS18cpp=ZTF18acqyvag; z ≈ 0.35) was discovered on 2018 December 31 by the Pan-STARRS Survey for Transients (Chambers et al. 2019) and subsequently classified as an SLSN on 2019 February 6 (Gomez et al. 2019). We constrain the explosion date to MJD 58424 by fitting the optical light curve to a magnetar model using MOSFit.

GRB 020903 (=XRF020903; z ≈ 0.251) was discovered on 2002 September 3 by the Wide-Field X-Ray monitor and the Soft X-Ray Camera on the High Energy Transient Explorer-2 (HETE-II; Soderberg et al. 2004). Observations of the radio and optical afterglow are presented in Soderberg et al. (2004). Host galaxy observations are given in Michałowski et al. (2012) and Greiner et al. (2016).

GRB 030329 (z ≈ 0.168) was discovered by the HETE-II satellite 2003 March 29 (Vanderspek et al. 2003). Radio observations of the afterglow, which is observable to δ t ∼ 5 yr post-explosion, are presented in Mesler et al. (2012) and van der Horst et al. (2008). Optical and radio observations of the host galaxy are presented in Niino et al. (2017) and Michałowski et al. (2012), respectively.

GRB 050826 (z ≈ 0.296) was discovered on 2005 August 26 by the Swift Burst Alert Telescope (BAT; Mangano et al. 2005). The event is categorized as a sub-luminous, sub-energetic event (Mirabal et al. 2007). A previous search for FRBs from this event was conducted in Hilmarsson et al. (2020). Host galaxy observations are presented in Levesque et al. (2010) and Niino et al. (2017).

GRB 061021 (z ≈ 0.3463) was discovered by the Swift-BAT on 2006 October 21 (Moretti et al. 2006). The event is one of a small subset of GRBs that exhibit excess X-ray emission at early times, which has been attributed to thermal emission from the shock breakout of a supernova (Sparre & Starling 2012). Optical and radio observations of the host galaxy are presented in Michałowski et al. (2012), Perley et al. (2015), and Greiner et al. (2016).

GRB 090417B (z ≈ 0.345) was discovered on 2009 April 17 by the Swift-BAT (Sbarufatti et al. 2009). The burst is classified as a "dark" GRB, a small subset of events lacking optical afterglows. Optical and radio observations of the host galaxy are given in Perley & Perley (2013) and Perley et al. (2013), respectively.

GRB 112225A (z ≈ 0.297) was discovered on 2011 December 25 by the Swift-BAT (Siegel et al. 2011). A previous search for FRBs from this event was conducted in Hilmarsson et al. (2020). Host galaxy observations are not reported in the literature for this event.

GRB 120422A (z ≈ 0.283) was discovered and localized by the Swift-BAT on 2012 April 22 (Troja et al. 2012). The event is classified as a transition object between low- and high-luminosity GRBs (Schulze et al. 2014). Host galaxy observations are presented in Niino et al. (2017).