Prompt X-Ray Emission from Fast Radio Bursts—Upper Limits with AstroSat

Fast radio bursts (FRBs) are bright (∼Jy), spatially unresolved and short (∼ms duration) transients in the radio regime (frequency range of 400 MHz to 8 GHz). These are characterized by their high observed dispersion measures (DMs)—often an order of magnitude higher than the total Galactic electron column density along the line of sight (LOS; Cordes & Lazio 2002; Yao et al. 2017)—indicating that the progenitor is extragalactic. The millisecond duration of the pulse constrains the emission region of the source to r ≲ ct_pulse ∼ 300 km, not considering any relativistic effects in the source frame.

A total of 85 FRB detections have been publicly reported until 2019 September (Petroff et al. 2016). Of these, 11 FRBs (Spitler et al. 2016; CHIME/FRB Collaboration et al. 2019a; Kumar et al. 2019; The CHIME/FRB Collaboration et al. 2019) are found to be repeating, but no periodicity or pattern has been found in its repetition (Spitler et al. 2016; Scholz et al. 2016). Unlike other FRBs, FRB 121102 has been localized, to milliarcsecond precision, to a dense star-forming region of a low-metallicity dwarf galaxy with redshift z = 0.193 colocated within a projected transverse distance of 40 pc to an unresolved radio source (Chatterjee et al. 2017; Marcote et al. 2017; Tendulkar et al. 2017). This localization and redshift measurement has led to a detailed study of its energetics, host environment (Bassa et al. 2017; Kokubo et al. 2017) and possible links to long Gamma-ray bursts (GRBs) and hydrogen-poor superluminous supernovae (Metzger et al. 2017; Margalit et al. 2018; Margalit & Metzger 2018).

Until now, no clear physical picture of either the mechanism for an FRB emission or the progenitor has emerged. A wide range of models have been proposed, many of which invoke neutron stars and strong magnetic fields to explain the short duration and high brightness temperatures of FRBs. The astrophysical scenarios hypothesized for the origin of FRBs include Crab-like giant pulses from neutron stars (Cordes & Wasserman 2016; Katz 2017), magnetar giant flares (Lyutikov 2002; Popov & Postnov 2013; Kulkarni et al. 2014; Pen & Connor 2015), binary neutron star mergers (Totani 2013; Paschalidis & Ruiz 2019), collisions between asteroids and neutron stars (Geng & Huang 2015), and the collapse of a neutron star into a black hole (Falcke & Rezzolla 2014). There are also non-neutron-star models that hypothesize FRBs arising from processes such as Dicke's superradiance (Houde et al. 2018), axion decay in a strong magnetic field (van Waerbeke & Zhitnitsky 2019), and compact explosions of macroscopic magnetic dipoles (Thompson 2017). See Platts et al. (2019),⁴ (Popov et al. 2018) for a recent review of FRB observations and models.

The presence or absence of a prompt emission corresponding to FRBs in different wavebands can constrain the emission mechanisms. In models invoking curvature radiation, photons are emitted along the direction of electron motion and the scope for inverse Compton scattering to higher wavebands is small (Kumar et al. 2017; Ghisellini & Locatelli 2018). If at all, such models predict possible prompt counterparts to FRBs in the THz—optical/infrared regime, but not at X-ray energies. Synchrotron emission models allow for possible inverse Compton upscattering of radio photons to X-ray energies, suggesting prompt X-ray/γ-ray counterparts for FRBs. Similarly, astrophysical scenarios such as binary neutron star mergers may also lead to the ejection of a GRB jet, which if aligned along our LOS, will produce a short γ-ray burst. Radio observations, combined with X-ray, gamma-ray, and gravitational wave observations will allow us to constrain the emission mechanisms of FRBs as well as to possibly discover the astrophysical scenarios that lead to them. Totani (2013) states that in particular, if some FRBs are linked to binary neutron star mergers and short GRBs, this may allow us to increase the detection horizon of LIGO and other gravitational wave observatories.

To date high-energy limits on FRB counterparts have been unconstraining due to the relative insensitivity of X-ray/γ-ray telescopes. Tendulkar et al. (2016) set limits on the fluence ratio in the γ-ray to radio bands of F_γ/F_{1.4 GHz} ≲ 10⁻⁷–10⁻⁹ erg cm⁻² Jy⁻¹ ms⁻¹ corresponding to F_γ/F_{1.4 GHz} ≲ 10⁸–10¹⁰ for a bandwidth of 1 GHz.⁵ For most FRBs, these limits are inconsistent with the observational limits for radio emission from the 2007 giant flare of SGR 1806−20 (Tendulkar et al. 2016). For FRB 121102, Scholz et al. (2017) set deep limits of F_γ/F_{1.4 GHz} < 10⁶–10⁸ using simultaneous radio and X-ray observations. We caution that observed values (or limits) of the γ-ray to radio fluence may significantly differ from intrinsic (source-frame) values due to beaming effects.

There have been a number of efforts at low to intermediate radio frequencies (∼GHz) searching for prompt counterparts to GRBs, also without significant success. Bannister et al. (2012) searched for short, dispersed radio transients from nine GRBs and detected a few >6σ candidates from two of them with delays of a few hundred seconds compared to the gamma-ray emission. However, the possibility of these being radio frequency interference could not be ruled out. Palaniswamy et al. (2014) searched for radio transients within 140 s of the occurrence of five GRBs but did not detect any significant candidate counterparts. The response time is limited by how fast a large radio dish can slew. In the future, these efforts may be improved by the chances of detecting GRBs in the fields of view of wide-field radio telescopes or using software beamforming radio telescopes.

From synchrotron emission, typical expected X-ray to radio ratios are ∼10⁴ (Lyutikov 2002) to 10⁶ (Lyubarsky 2014). Pulsars have typical γ-ray to radio ratios of 10⁴–10⁸. Other theories like those of Katz (2016) and Falcke & Rezzolla (2014) also predict prompt X-ray/γ-ray emission, though the emission ratios are not well-quantified. These constraints still allow for the possibility of FRBs originating from pulsars, magnetars, or other astrophysical scenarios. However, as we discover greater numbers of FRBs, it is important to have a framework to easily search for counterparts and set limits. After we amass a large population of FRBs, especially radio-bright ones, with X-ray nondetections, we can use them to constrain FRB models.

The Cadmium Zinc Telluride Imager (CZTI; Bhalerao et al. 2017a) on board AstroSat (Singh et al. 2014) is a coded aperture mask instrument with a 46 × 46 imaging field of view in the hard X-ray band from 20 keV to 200 keV. The instrument structure is designed to be nearly transparent to photons with energies ≳100 keV, making CZTI sensitive to transients from the entire sky, barring ∼30% occulted by the Earth. The four identical, independent quadrants of CZTI allow for rejection of spurious signals to a high degree. CZTI has detected over 200 GRBs.⁶ For transients with a clear detection, CZTI data can yield spectra (Rao et al. 2016), polarization (Vadawale et al. 2015; Chattopadhyay et al. 2017), and localization (Rao et al. 2016; Bhalerao et al. 2017b).

Here we present a framework for burst searches, the hard X-ray limits on known FRBs and a plan for a near-automated future pipeline.

The standard procedure for CZTI transient searches differs from the usual analysis for steady sources. We outline the transient search method (used, for instance, in Anumarlapudi et al. 2018a, 2018b) in this section. In Section 3 we expand the sample with searches in archival data.

2.1. Data Reduction

2.2. Qualitative Searches

After the data are prepared in this manner, we first undertake a qualitative search for FRB counterparts. We select a t_search = 20 s window centered on the dedispersed time of the FRB. We note that the uncertainties in the reported times of the FRBs are much smaller than our search window. For instance, uncertainty in time corresponding to DM error of 0.1 pc cm⁻³ at a frequency of 800 MHz (bottom of the band for UTMOST) will be 0.65 ms, while that due to AstroSat's positional error of 1° will be 0.78 ms.

Each of the four independent, identical quadrants of CZTI are treated as a separate instrument. We create two-dimensional spectrograms (time-energy plots) for the data of each quadrant by binning the data in 20 keV bins in the nominal CZTI energy range of 20–200 keV, and in temporal bins of width t_bin (Figure 1, top). The spectrograms are plotted and visually inspected for excess emission in the t_search window. In this work, we conducted the searches at three timescales: t_bin = 10 ms, 100 ms, and 1 s. We do not search at 1 ms timescale: given the effective area of CZTI, the count rate required to ensure at most a single false positive in the t_search window will be an order of magnitude higher than the count rate observed in the succeeding t_bin (10 ms) which means the source has to be extremely bright to be detectable at 1 ms timescale. These spectrograms are dominated by the typical "background" spectrum of CZTI, which in our case also includes photons from the on-axis source at the instant of the search. Hence, we also create median-subtracted spectrograms by calculating a median count rate for each energy bin and subtracting it from the instantaneous counts at that energy in each time bin (Figure 1, middle). As a last step, we enhance outliers in each energy bin by dividing the light curve at that energy with its standard deviation calculated over the entire t_search window (Figure 1, bottom). We visually inspect these spectrograms for any signs of enhanced X-ray emission in the t_search window. For all publicly available FRBs where coincident CZTI data were available (Table 2), we did not detect any X-ray candidates.

Zoom In Zoom Out Reset image size

Table 2.  Observed Parameters of Radio Bursts

Name	Time	Coordinates (J2000)		Radio Telescope	Central Frequency	Bandwidth	Sradioa	FWHM	Fradioa
	UTC	R.A.	Decl.		(MHz)	(MHz)	(Jy)	(ms)	(Jy ms)
FRB 190806	17:07:58.0	00:02:21.38	−07:34:54.6	UTMOST	835.0	31.25	3.91	11.96	46.8
FRB 190714	05:37:12.901	12:15.9	−13:00	ASKAP	1297.0	336.0	4.7	1.7	8.0
FRB 190711	01:53:41.100	21:56	−80:23	ASKAP	1297.0	336.0	4.11	9.0	28.0
FRB 190523	06:05:55.815	13:48:15.6	+72:28:11	DSA−10	1405.0	125.0	666.67	0.42	280.0
FRB 190322	07:00:12.3	04:46:14.45	−66:55:27.8	UTMOST	835.0	31.25	11.85	1.35	16.0
FRB 181228	13:48:50.100	06:09:23.64	−45:58:02.4	UTMOST	835.0	31.25	19.23	1.24	23.85
FRB 181017	10:24:37.400	22:05:54.82	−08:50:34.22	UTMOST	835.0	31.25	161	0.32	51.52
FRB 180817.J1533+42	01:49:20.202	15:33	+42:12	CHIME/FRB	600.0	400.0	70.27	0.37	26.0
FRB 180814.J1554+74	14:20:14.440	15:54	+74:01	CHIME/FRB	600.0	400.0	138.89	0.18	25.0
FRB 180812.J0112+80	11:45:32.872	01:12	+80:47	CHIME/FRB	600.0	400.0	14.4	1.25	18.0
FRB 180810.J1159+83	22:40:42.493	11:59	+83:07	CHIME/FRB	600.0	400.0	60.71	0.28	17.0
FRB 180806.J1515+75	14:13:03.107	15:15	+75:38	CHIME/FRB	600.0	400.0	34.78	0.69	24.0
FRB 180801.J2130+72	08:47:14.793	21:30	+72:43	CHIME/FRB	600.0	400.0	54.9	0.51	28.0
FRB 180730.J0353+87	03:37:25.937	03:53	+87:12	CHIME/FRB	600.0	400.0	119.05	0.42	50.0
FRB 180729.J0558+56	17:28:18.258	05:58	+56:30	CHIME/FRB	600.0	400.0	112.5	0.08	9.0
FRB 180727.J1311+26	00:52:04.474	13:11	+26:26	CHIME/FRB	600.0	400.0	17.95	0.78	14.0
FRB 180725.J0613+67	17:59:32.813	06:13	+67:04	CHIME/FRB	600.0	400.0	38.71	0.31	12.0
FRB 180528	04:24:00.9	06:38:48.7	−49:53:59	UTMOST	835.0	32	15.75	2.0	32
FRB 180525	15:19:06.515	14:40	−02:12	ASKAP	1297.0	336.0	78.9	3.8	299.82
FRB 180430	10:00:35.700	06:51	−09:57	ASKAP	1297.0	336.0	147.5	1.2	177.0
FRB 180324	09:31:46.706	06:16	−34:47	ASKAP	1297.0	336.0	16.5	4.3	70.95
FRB 180315	05:05:30.985	19:35	−26:50	ASKAP	1297.0	336.0	23.3	2.4	55.92
FRB 180311	04:11:54.800	21:31:33.42	−57:44:26.7	Parkes	1352.0	338.381	0.2	12.0	2.4
FRB 180301	07:34:19.760	06:12:43.4	04:33:44.8	Parkes	1352.0	338.381	0.5	3.0	1.5
FRB 180212	23:45:04.399	14:21	−03:35	ASKAP	1297.0	336.0	53.0	1.81	95.93
FRB 180130	04:55:29.993	21:52.2	−38:34	ASKAP	1297.0	336.0	23.1	4.1	94.71
FRB 180119	12:24:40.747	03:29.3	−12:44	ASKAP	1297.0	336.0	40.7	2.7	109.89
FRB 180110	07:34:34.959	21:53.0	−35:27	ASKAP	1297.0	336.0	128.1	3.2	409.92
FRB 171213	14:22:40.467	03:39	−10:56	ASKAP	1297.0	336.0	88.6	1.5	132.9
FRB 171209	20:34:23.500	15:50:25	−46:10:20	Parkes	1352.0	338.381	0.92	2.5	2.3
FRB 171020	10:27:58.598	22:15	−19:40	ASKAP	1297.0	336.0	117.6	3.2	376.32
FRB 171019	13:26:40.097	22:17.5	−08:40	ASKAP	1297.0	336.0	40.5	5.4	218.7
FRB 171003	04:07:23.781	12:29.5	−14:07	ASKAP	1297.0	336.0	40.5	2.0	81.0
FRB 170906	13:06:56.488	21:59.8	−19:57	ASKAP	1297.0	336.0	29.6	2.5	74.0
FRB 170827	16:20:18.000	00:49:18.66	−65:33:02.3	UTMOST	835.0	32	50.3	0.4	19.87
FRB 170707	06:17:34.354	02:59	−57:16	ASKAP	1297.0	336.0	14.8	3.5	51.8
FRB 170606	10:03:27.000	5:34:0.0	41:45:0.0	Pushchino	111.0	2.5	0.54	3300.0	1782.0
FRB 170428	18:02:34.700	21:47	−41:51	ASKAP	1320.0	336.0	7.7	4.4	33.88
FRB 170416	23:11:12.799	22:13	−55:02	ASKAP	1320.0	336.0	19.4	5.0	97.0
FRB 160608	03:53:01.088	07:36:42	−40:47:52	UTMOST	843.0	16b	4.3	9.0	38.7
FRB 151230	16:15:46.525	09:40:50	−03:27:05	Parkes	1352.0	338.381	0.42	4.4	1.9

Notes.

^aS_radio is radio flux and F_radio is radio fluence of the burst. ^bThe value reported is assumed to be 16 on the basis of previous detection (FRB 160317) since the actual value is missing from FRBCAT.

Only a portion of this table is shown here to demonstrate its form and content. A machine-readable version of the full table is available.

Download table as:  DataTypeset image

2.3. Flux Limits

For this work, we limit ourselves to the time period from 2015 October 6 (the day CZTI became operational) to 2018 August 31. FRBCAT (Petroff et al. 2016) lists 64 FRBs in this period—of which 16 were occulted by Earth in the CZTI frame, two were very close to the Earth limb and are ignored, while five occurred while CZTI was nonfunctional due to a passage of AstroSat through the South Atlantic Anomaly (SAA). Table 2 lists the properties of the remaining 41 FRBs that were visible to CZTI. The radio fluences for the Parkes bursts are defined assuming the burst was in the center of the beam, and hence are lower limits. We searched the CZTI data for prompt emission in the 20–200 keV band from the remaining 41 FRBs at three timescales, and did not obtain any detection. Corresponding upper limits on X-ray fluence (erg cm⁻²) assuming a simple power-law spectrum with photon index Γ = −1 are reported in Table 3. The table also reports upper limits on the X-ray to radio fluence ratios (η) by self-consistently choosing a power-law slope as discussed in Section 2.3.

Table 3.  CZTI Fluence Limits on X-Rays from FRBs

Name	Radio Flux Density	Radio Fluence	tbin	Γmax	X-Ray fluence		$\eta /{10}^{9}=\tfrac{{F}_{X \mbox{-} {ray}}}{{F}_{\mathrm{Radio}}}/{10}^{9}$
(Reference to original detection)	(Jy)	(Jy ms)	(s)		(erg  cm−2)
					Γ = −1	Γ = Γmax	Γ = −1	Γ = Γmax
FRB 190806	3.91	46.8	0.01	−1.19	1.6e−07	1.65e−07	0.34	0.35
(Gupta et al. 2019a)			0.1	−1.25	3.67e−07	3.84e−07	0.78	0.82
			1.0	−1.33	5.69e−07	6.03e−07	1.21	1.29
FRB 190714	4.7	8.0	0.01	−1.24	7.38e−08	7.47e−08	0.92	0.93
(Bhandari et al. 2019)			0.1	−1.3	1.67e−07	1.69e−07	2.08	2.11
			1.0	−1.38	2.72e−07	2.76e−07	3.4	3.45
FRB 190711	4.1	28.0	0.01	−1.16	4.33e−07	4.44e−07	1.55	1.59
(Shannon et al. 2019)			0.1	−1.22	9.72e−07	1.01e−06	3.47	3.6
			1.0	−1.3	1.55e−06	1.64e−06	5.55	5.85
FRB 190523	666.7	280.0	0.01	−1.42	1.53e−07	1.61e−07	0.05	0.06
(Ravi et al. 2019)			0.1	−1.48	3.51e−07	3.74e−07	0.13	0.13
			1.0	−1.56	5.56e−07	6.02e−07	0.2	0.21
FRB 190322	11.8	16.0	0.01	−1.23	2.15e−07	2.23e−07	1.35	1.39
(Gupta et al. 2019b)			0.1	−1.29	4.83e−07	5.07e−07	3.02	3.17
			1.0	−1.37	7.53e−07	8.02e−07	4.7	5.01
FRB 181228	19.2	23.8	0.01	−1.28	9.74e−08	9.6e−08	0.41	0.4
Farah et al. (2019)			0.1	−1.36	1.38e−07	1.36e−07	0.58	0.57
			1.0	−1.44	2.08e−07	2.04e−07	0.87	0.85
FRB 181017	161.0	51.5	0.01	−1.39	6.8e−08	6.51e−08	0.13	0.13
Farah et al. (2019)			0.1	−1.47	9.7e−08	9.2e−08	0.19	0.18
			1.0	−1.55	1.45e−07	1.36e−07	0.28	0.26
FRB 180817.J1533+42	70.3	26.0	0.01	−1.3	2.17e−07	2.31e−07	0.83	0.89
(CHIME/FRB Collaboration et al. 2019b)			0.1	−1.36	4.84e−07	5.22e−07	1.86	2.01
			1.0	−1.43	7.57e−07	8.33e−07	2.91	3.2
FRB 180814.J1554+74	138.9	25.0	0.01	−1.35	1.18e−07	1.21e−07	0.47	0.48
(CHIME/FRB Collaboration et al. 2019b)			0.1	−1.41	2.71e−07	2.78e−07	1.08	1.11
			1.0	−1.49	4.31e−07	4.46e−07	1.72	1.78
FRB 180812.J0112+80	14.4	18.0	0.01	−1.25	1.26e−07	1.29e−07	0.7	0.72
(CHIME/FRB Collaboration et al. 2019b)			0.1	−1.31	2.91e−07	3.01e−07	1.62	1.67
			1.0	−1.39	4.41e−07	4.62e−07	2.45	2.57
FRB 180810.J1159+83	60.7	17.0	0.01	−1.2	1.85e−06	2e−06	10.89	11.74
(CHIME/FRB Collaboration et al. 2019b)			0.1	−1.26	4.29e−06	4.74e−06	25.24	27.89
			1.0	−1.33	6.91e−06	7.89e−06	40.66	46.39
FRB 180806.J1515+75	34.8	24.0	0.01	−1.27	2.22e−07	2.30e−07	0.93	0.96
(CHIME/FRB Collaboration et al. 2019b)			0.1	−1.33	5.1e−07	5.34e−07	2.13	2.23
			1.0	−1.4	8.1e−07	8.6e−07	3.37	3.58
FRB 180801.J2130+72	54.9	28.0	0.01	−1.29	2.01e−07	2.10e−07	0.72	0.75
(CHIME/FRB Collaboration et al. 2019b)			0.1	−1.35	4.58e−07	4.85e−07	1.64	1.73
			1.0	−1.43	7.21e−07	7.75e−07	2.57	2.77
FRB 180730.J0353+87	119.0	50.0	0.01	−1.31	2.80e−07	2.875e−07	0.56	0.58
(CHIME/FRB Collaboration et al. 2019b)			0.1	−1.37	6.42e−07	6.64e−07	1.28	1.33
			1.0	−1.44	9.96e−07	1.04e−06	1.99	2.08
FRB 180729.J0558+56	112.5	9.0	0.01	−1.3	3.34e−07	3.52e−07	3.71	3.91
(CHIME/FRB Collaboration et al. 2019b)			0.1	−1.36	7.47e−07	7.98e−07	8.3	8.86
			1.0	−1.43	1.22e−06	1.33e−06	13.59	14.76
FRB 180727.J1311+26	18.0	14.0	0.01	−1.21	4.93e−07	5.08e−07	3.52	3.63
(CHIME/FRB Collaboration et al. 2019b)			0.1	−1.27	1.09e−06	1.14e−06	7.82	8.15
			1.0	−1.34	1.72e−06	1.82e−06	12.32	13.02
FRB 180725.J0613+67	38.7	12.0	0.01	−1.25	3.73e−07	3.96e−07	3.11	3.3
(CHIME/FRB Collaboration et al. 2019b)			0.1	−1.31	8.73e−07	9.43e−07	7.28	7.86
			1.0	−1.38	1.39e−06	1.53e−06	11.57	12.77
FRB 180528	15.8	32.0	0.01	−1.24	1.21e−06	1.31e−06	3.77	4.1
Farah et al. (2019)			0.1	−1.3	2.752e−06	3.07e−06	8.6	9.6
			1.0	−1.38	4.32e−06	4.97e−06	13.49	15.55
FRB 180525	78.9	299.8	0.01	−1.35	8.14e−08	8.23e−08	0.03	0.03
(Macquart et al. 2019)			0.1	−1.42	1.85e−07	1.87e−07	0.06	0.06
			1.0	−1.5	2.94e−07	2.99e−07	0.1	0.1
FRB 180430	147.5	177.0	0.01	−1.32	2.85e−07	3.065e−07	0.16	0.17
(Qiu et al. 2019)			0.1	−1.39	6.43e−07	7.03e−07	0.36	0.4
			1.0	−1.46	1.04e−06	1.16e−06	0.59	0.65
FRB 180324	16.5	71.0	0.01	−1.26	1.59e−07	1.64e−07	0.22	0.23
(Macquart et al. 2019)			0.1	−1.32	3.57e−07	3.72e−07	0.5	0.52
			1.0	−1.4	5.46e−07	5.76e−07	0.77	0.81
FRB 180315	23.3	55.9	0.01	−1.2	9.84e−07	1.03e−06	1.76	1.84
(Macquart et al. 2019)			0.1	−1.26	2.25e−06	2.39e−06	4.03	4.28
			1.0	−1.34	3.48e−06	3.77e−06	6.22	6.74
FRB 180311	0.2	2.4	0.01	−1.02	5.435e−07	5.46e−07	22.64	22.76
(Osłowski et al. 2019)			0.1	−1.08	1.20e−06	1.23e−06	50.17	51.25
			1.0	−1.16	2.01e−06	2.095e−06	83.68	87.28
FRB 180301	0.5	1.5	0.01	−1.11	1.72e−07	1.74e−07	11.46	11.63
(Price et al. 2018)			0.1	−1.17	3.86e−07	3.95e−07	25.71	26.33
			1.0	−1.25	6.21e−07	6.44e−07	41.39	42.96
FRB 180212	53.0	95.9	0.01	−1.34	8.71e−08	8.75e−08	0.09	0.09
(Shannon et al. 2018)			0.1	−1.4	2.03e−07	2.05e−07	0.21	0.21
			1.0	−1.47	3.44e−07	3.48e−07	0.36	0.36
FRB 180130	23.1	94.7	0.01	−1.18	1.25e−06	1.35e−06	1.32	1.42
(Shannon et al. 2018)			0.1	−1.25	2.86e−06	3.14e−06	3.01	3.32
			1.0	−1.32	4.79e−06	5.45e−06	5.06	5.76
FRB 180119	40.7	109.9	0.01	−1.31	1.07e−07	1.10e−07	0.1	0.1
(Shannon et al. 2018)			0.1	−1.38	2.36e−07	2.44e−07	0.21	0.22
			1.0	−1.46	3.85e−07	4.02e−07	0.35	0.37
FRB 180110	128.1	409.9	0.01	−1.33	2.21e−07	2.36e−07	0.05	0.06
(Shannon et al. 2018)			0.1	−1.39	5.1e−07	5.54e−07	0.12	0.14
			1.0	−1.47	7.88e−07	8.74e−07	0.19	0.21
FRB 171213	88.6	132.9	0.01	−1.33	1.68e−07	1.77e−07	0.13	0.13
(Shannon et al. 2018)			0.1	−1.39	3.82e−07	4.08e−07	0.29	0.31
			1.0	−1.47	6.14e−07	6.67e−07	0.46	0.5
FRB 171209	0.92	2.3	0.01	−1.28	1.9e−07	1.79e−07	8.26	7.78
(Osłowski et al. 2019)			0.1	−1.3	2.8e−07	2.98e−07	12.17	12.96
			1.0	−1.32	4.5e−07	4.77e−07	19.57	20.74
FRB 171020	117.6	376.3	0.01	−1.37	9e−08	9.05e−08	0.02	0.02
(Shannon et al. 2018)			0.1	−1.43	1.99e−07	2.01e−07	0.05	0.05
			1.0	−1.51	3.36e−07	3.4e−07	0.09	0.09
FRB 171019	40.5	218.7	0.01	−1.34	5.93e−08	5.92e−08	0.03	0.03
(Shannon et al. 2018)			0.1	−1.4	1.3e−07	1.3e−07	0.06	0.06
			1.0	−1.48	2.22e−07	2.23e−07	0.1	0.1
FRB 171003	40.5	81.0	0.01	−1.25	4.42e−07	4.50e−07	0.55	0.56
(Shannon et al. 2018)			0.1	−1.32	1.01e−06	1.03e−06	1.24	1.27
			1.0	−1.39	1.72e−06	1.78e−06	2.13	2.2
FRB 170906	29.6	74.0	0.01	−1.28	1.62e−07	1.68e−07	0.22	0.23
(Shannon et al. 2018)			0.1	−1.34	3.68e−07	3.86e−07	0.5	0.52
			1.0	−1.42	6.01e−07	6.39e−07	0.81	0.86
FRB 170827	50.3	19.9	0.01	−1.32	1.10e−07	1.15e−07	0.55	0.58
(Farah et al. 2017)			0.1	−1.38	2.44e−07	2.58e−07	1.23	1.3
			1.0	−1.45	3.97e−07	4.25e−07	2.0	2.14
FRB 170707	14.8	51.8	0.01	−1.26	1.42e−07	1.46e−07	0.27	0.28
(Shannon et al. 2018)			0.1	−1.32	3.13e−07	3.25e−07	0.6	0.63
			1.0	−1.4	5.05e−07	5.32e−07	0.98	1.03
FRB 170606	0.5	1782.0	0.01	−1.02	1.38e−06	1.39e−06	0.08	0.08
Rodin & Fedorova (2018)			0.1	−1.08	3.13e−06	3.21e−06	0.18	0.18
			1.0	−1.15	4.75e−06	5.01e−06	0.27	0.28
FRB 170428	7.7	33.9	0.01	−1.2	2.715e−07	2.80e−07	0.8	0.83
(Shannon et al. 2018)			0.1	−1.27	6.13e−07	6.38e−07	1.81	1.88
			1.0	−1.34	9.71e−07	1.03e−06	2.86	3.03
FRB 170416	19.4	97.0	0.01	−1.22	4.525e−07	4.79e−07	0.47	0.49
(Shannon et al. 2018)			0.1	−1.28	1.0e−06	1.085e−06	1.04	1.12
			1.0	−1.36	1.56e−06	1.73e−06	1.61	1.78
FRB 160608	4.3	38.7	0.01	−1.11	1.88e−06	1.29e−06	4.85	3.34
Caleb et al. (2018)			0.1	−1.17	4.25e−06	3e−06	10.99	7.75
			1.0	−1.25	6.69e−06	4.87e−06	17.29	12.57
FRB 151230	0.4	1.9	0.01	−1.1	1.78e−07	1.80e−07	9.36	9.49
(Bhandari et al. 2018)			0.1	−1.16	4e−07	4.09e−07	21.05	21.54
			1.0	−1.24	6.76e−07	7e−07	35.57	36.87

Only a portion of this table is shown here to demonstrate its form and content. A machine-readable version of the full table is available.

Download table as:  DataTypeset images: 1 1a 1b

We have set limits on the X-ray to radio fluence ratios that vary between ∼10⁷ and ∼10¹⁰ depending on the intrinsic brightness of the FRB, its location in the radio telescope beam, and the location relative to the CZTI field of view. This approaches the range of X-ray to radio fluence ratios expected from theory (∼10⁴, Lyutikov 2002 to 10⁶, Lyubarsky 2014) and also the range of observed values for pulsars.

DeLaunay et al. (2016) searched the Swift-BAT data for any possible γ-ray emission in the energy range 15–150 KeV and obtained a fluence limit of F_γ ≲ 10⁻⁶ erg cm⁻² in an interval of 300 s for a total of four FRBs. This corresponds to η (F_γ/F_{1.4 GHz}) of ∼10¹¹. We note that the claimed γ-ray transient detection corresponding to FRB 131104 is extremely marginal (illuminating only 2.9% of the Swift-BAT detector). Furthermore, the search was conducted at much longer timescales (T₉₀ = 100 s) as compared to our study, hence the claimed detection is not at odds with our limits. Similar searches carried out for a possible gamma-ray counterpart by Martone et al. (2019) and Guidorzi et al. (2019) resulted in non-detections, placing upper limits on gamma-ray to radio fluence.

However, despite the strong upper limits from nondetections, it is challenging to constrain theories directly based on individual FRB observations because the intrinsic X-ray to radio fluence ratio may be significantly different from the observed ratio due to beaming effects. For instance, some models of binary neutron star mergers suggest that the X-ray/γ-ray emission would be strongly beamed (as in the case of relativistic jets from GRBs) while the radio emission is relatively isotropic (see, for instance, Totani 2013, and references therein.) In such cases, the lack of observed high-energy emission in an individual case can be dismissed. If the jets are highly relativistic, the emission will be strongly beamed and visible to <1% of all observers—as happens for gamma-ray bursts (Berger 2014). We need statistical limits on the X-ray to radio fluence ratios on hundreds of FRBs to help us constrain their emission models.

Conversely, if we expect that the radio emission is beamed while the X-ray emission is nearly isotropic (as in the case of a magnetar giant flare), it will be significantly more challenging to verify emission mechanisms. However, in terms of energetics, high-energy emission powered by compact objects with magnetar-like magnetic field strengths cannot be detected at gigaparsec distances unless they are relativistically beamed (Murase et al. 2017).

AstroSat CZTI is one of the most sensitive instruments for detecting short duration high-energy transients (see, for example, Bhalerao et al. 2017b). Our limits on 41 bursts out of 64 that occurred in our search period are consistent with the operational expectations of being sensitive to about half the events, the rest being lost to Earth-occultation and SAA transits. AstroSat continuously records time-tagged photon data which can be used to search for FRB counterparts in ground processing. AstroSat also has the advantage of being sensitive to the entire sky not obstructed by Earth: similar to Fermi-LAT but several times larger than the field of view of Swift-BAT. This wide-field hard X-ray sensitivity of CZTI will be very useful in the future as the rate of FRB detections increases with facilities such as CHIME (The CHIME/FRB Collaboration et al. 2018), ASKAP (Johnston et al. 2009), HIRAX (Newburgh et al. 2016), and SKA (Maartens et al. 2015).

We thank E. Aarthy for helpful discussions in flux calculations. This publication also uses the data from the AstroSat mission of the Indian Space Research Organisation (ISRO), archived at the Indian Space Science Data Centre (ISSDC). The CZT–Imager is built by a consortium of Institutes across India. The Tata Institute of Fundamental Research, Mumbai, led the effort with instrument design and development. Vikram Sarabhai Space Centre, Thiruvananthapuram provided the electronic design, assembly, and testing. ISRO Satellite Centre (ISAC), Bengaluru, provided the mechanical design, quality consultation, and project management. The Inter University Centre for Astronomy and Astrophysics (IUCAA), Pune, did the Coded Mask design, instrument calibration, and Payload Operation Centre. Space Application Centre (SAC) at Ahmedabad provided the analysis software. A vast number of industries participated in the fabrication and the University sector pitched in by participating in the test and evaluation of the payload. The Indian Space Research Organisation funded, managed and facilitated the project. This work utilized various software including Python, IDL, FTOOLS, C, and C++.

Facility: AstroSat(CZTI). -

Software: AstroPy (Astropy Collaboration et al. 2013), Python, IDL, FTOOLS (Blackburn 1995), C, C++.

Here we describe our method for detrending the data and calculating the cutoff rate. To calculate the cutoff count rate, we choose five consecutive orbits each, both before and after the event orbit, excluding the orbit of interest. Figure 2 (top panel, left) shows the light curve of an entire orbit of one of the quadrants (Quad D) binned at 1 s. The slow variation in the counts reflects the motion of satellite in its orbit while the data gap is due to the satellite's passage through SAA. We use a second-order Savitzky–Golay filter to estimate this background variation and detrend the light curve. The red solid curve in Figure 2 (top panel, left) shows the estimation of this slow variation using a Savitzky–Golay filter. Figure 2 (top panel, right) depicts the 1 s binned light curve after subtracting this slow variation. The average histogram of the count rates of all 10 neighboring orbits is used to estimate the cutoff rate, which is quantified by the parameter confidence. Given the time span of the search interval t_search, the binning time t_bin, and the false alarm rate FAR, the chance of detecting a false positive is 1 in (t_search/t_bin/FAR); hence confidence is 1−(t_search/t_bin/FAR). The cutoff rate is chosen based on this required confidence, from the average histogram and is independently estimated for each binning time. It can be noted that the false alarm rate chosen for this analysis is 0.1 per quadrant for a time span of 20 s (since the transient is short-lived ∼ms). The four quadrants of CZTI are independent and so the probability of getting a temporally coincident false positive in all the quadrants is 10⁻⁴ in the search interval (20 s). The orbital period of AstroSat is ∼6000 s; hence there can be ∼30 false positives per quadrant in an orbit (Figure 2 bottom panel, left). However, the requirement of temporal coincidence across quadrants removes such false positives. Figure 2 (bottom panel, right) shows the light curve in a 20 s window around the arrival time of FRB, with cutoff rates for individual quadrants marked. We see no evidence for any temporally coincident prompt emission in all the quadrants above the background level.

Zoom In Zoom Out Reset image size