Follow-up of the Neutron Star Bearing Gravitational-wave Candidate Events S190425z and S190426c with MMT and SOAR

The joint detection of gravitational waves (GWs) and electromagnetic (EM) radiation from the binary neutron star (BNS) merger GW170817 was a watershed event. The merger was accompanied by a weak short gamma-ray burst (SGRB), by ultraviolet (UV)/optical/near-infrared (NIR) emission due to a kilonova (during the first month), and by radio, X-ray, and long-term optical emission due to an off-axis jet (LIGO Scientific Collaboration & Virgo Collaboration 2017, 2017a; LIGO Scientific Collaboration & Virgo Collaboration 2017b). GW170817 was localized to a region of about 30 deg² and to a distance of 40 ± 8 kpc, which enabled both galaxy-targeted and wide-field searches to rapidly identify the EM counterpart, within about 11 hr of merger (Arcavi et al. 2017; Coulter et al. 2017; Lipunov et al. 2017; Soares-Santos et al. 2017; Tanvir et al. 2017; Valenti et al. 2017).

Observing run 3 (O3) of the Advanced Laser Interferometer Gravitational-Wave Observatory (LIGO) and Advanced Virgo (ALV) commenced on 2019 April 1, with a 50% increase in sensitivity compared to Observing Runs 1 and 2. The resulting BNS merger detection distances in O3 are on average about 140 Mpc for LIGO Livingston, 110 Mpc for LIGO Hanford, and 50 Mpc for Virgo. Given the volumetric merger rate inferred from GW170817, 110–3840 Gpc⁻³ yr⁻¹ (LIGO Scientific Collaboration & Virgo Collaboration 2018), the expected number of BNS merger detections in the year-long O3 is ∼1–20 (assuming 70% duty cycle for LIGO). For neutron star (NS)–black hole (BH) mergers the upper bound on the rate based on non-detections in O1 and O2 is ≲600 Gpc⁻³ yr⁻¹ (LIGO Scientific Collaboration & Virgo Collaboration 2018); however, given their larger detection volume relative to BNS mergers the observed NS–BH merger rate in O3 may exceed that of BNS mergers.

On 2019 April 25 at 08:18:05.017 UTC ALV detected a GW candidate event, designated S190425z, with a false alarm rate (FAR) of 1 in 7 × 10⁴ yr, a probability of being a BNS merger of >99%, and a luminosity distance of 155 ± 45 Mpc (LIGO Scientific Collaboration & Virgo Collaboration 2019a). Because the event was detected only by LIGO Livingston and marginally by Virgo (LIGO Hanford was offline at the time of detection), the localization region had an initial area of about 10⁴ deg² (90% confidence; LIGO Scientific Collaboration & Virgo Collaboration 2019a), which was refined to 7460 deg² about 31 hr post merger (LIGO Scientific Collaboration & Virgo Collaboration 2019b). The most up-to-date sky localization region and distance estimate is shown in Figure 1.

Zoom In Zoom Out Reset image size

Soon after, on 2019 April 26 at 15:21:55.337 UTC, ALV detected another GW candidate event, designated S190426c, with an FAR of 1 in 1.7 yr, a probability of containing a NS of >99%, a luminosity distance of 375 ± 108 Mpc, and an initial localization region of about 1260 deg² (90% confidence; LIGO Scientific Collaboration & Virgo Collaboration 2019c), which was refined to 1130 deg² about 20 hr post merger (LIGO Scientific Collaboration & Virgo Collaboration 2019d). Given the relatively high FAR, we cannot be certain that the event was astrophysical. However, under the assumption that it was, the latest parameter estimation gives a 60% probability that the more massive binary component was >5 M_⊙, a 25% probability that it was 3–5 M_⊙, and a 15% probability that it was <3 M_⊙, suggesting that the NS–BH classification is most likely (LIGO Scientific Collaboration & Virgo Collaboration 2019e). The most up-to-date sky localization region and distance estimate is shown in Figure 1.

Here we report our optical follow-up of both events, using the MMT 6.5 m telescope to target galaxies within their localization volumes. In Section 2 we present our MMT observations. In Section 3 we collate searches reported publicly via the GRB Coordinates Network (GCN) circulars to explore a few aspects of the follow-up and announced candidates, as well as our spectroscopic follow-up of two candidates. In Sections 4 and 5 we compare the results to the kilonova emission of GW170817, to theoretical kilonova models, and to on-axis and slightly off-axis SGRB afterglow models. In Section 6 we collate γ-ray and X-ray searches and compare to the same SGRB models. We summarize and draw some initial conclusions in Section 7.

Since the beginning of O3 we have been using the MMT 6.5 m telescope at Fred L. Whipple Observatory in Arizona to carry out follow-up observations of galaxies within the localization volumes of GW alerts. Upon receipt of an alert, our automated software generates a list of galaxies in the Galaxy List for the Advanced Detector Era (GLADE) catalog (Dálya et al. 2018) that are located within the 90% confidence volume, ranked by probability within the volume. The software also downloads reference images and catalogs from the Pan-STARRS1 (PS1) 3π database (Chambers et al. 2016) and collates the locations of all previously reported transients and moving objects from the Zwicky Transient Facility Public Survey (Bellm et al. 2019; Graham et al. 2019; Masci et al. 2019, via the MARS broker¹⁹ ), all other public time-domain surveys (via the Transient Name Server²⁰ ), and the Minor Planet Center (via the SkyBoT service; Berthier et al. 2006). A custom data reduction pipeline processes each image as it is read out and performs image subtraction (using PyZOGY; Zackay et al. 2016; Guevel & Hosseinzadeh 2017). The reference, science, and subtracted images are then inspected for new transients using a custom web interface based on Flask and JS9 (Mandel & Vikhlinin 2018). An example from our search in the localization region of S190425z is shown in Figure 2.

Zoom In Zoom Out Reset image size

For S190425z, we commenced observations using the MMTCam imager on 2019 April 25 at 11:39:23 UT, 3.4 hr post merger, and continued until morning twilight, with our last exposure ending at 12:06:46 UT (Hosseinzadeh et al. 2019b). We obtained 30-s g-band exposures of 17 galaxies.²¹ On the following night, from 08:30:29 to 10:55:00 UT (24.2–26.6 hr post merger), we imaged 50 additional galaxies in i-band, to minimize moonlight contamination (Hosseinzadeh et al. 2019a). No transient sources were uncovered in these observations to median 3σ limiting magnitudes of g = 22.0 and i = 22.5. We provide the information for all of the individual galaxies in Table 1.

Table 1.  Log of MMT Follow-up Observations

Name	R.A.	Decl.	Date	Time	Filter	Limiting Mag.a
S190425z
2MASX J16545364–1657072	16h54m5365	−16d57m073	2019 Apr 25	11:39:23	g	22.3
2MASX J16530485–1617273	16h53m0486	−16d17m274	2019 Apr 25	11:40:50	g	22.4
2MASX J16571426–0613510	16h57m1426	−06d13m510	2019 Apr 25	11:42:30	g	22.4
2MASX J16520774–1703135	16h52m0775	−17d03m135	2019 Apr 25	11:44:22	g	22.5
PGC 58929	16h45m5464	−23d27m060	2019 Apr 25	11:47:33	g	22.0
NGC 6234	16h51m5734	+04d23m008	2019 Apr 25	11:49:28	g	22.4
PGC 59064	16h49m3316	+06d00m585	2019 Apr 25	11:51:21	g	22.2
PGC 59201	16h53m2409	+04d14m105	2019 Apr 25	11:52:39	g	22.3
2MASX J16564688–0142052	16h56m4689	−01d42m053	2019 Apr 25	11:54:12	g	22.0
2MASX J16504669+0436170	16h50m4670	+04d36m170	2019 Apr 25	11:55:46	g	22.2
UGC 10426	16h30m5016	+16d15m025	2019 Apr 25	11:57:27	g	21.9
PGC 90265	16h57m2682	−10d11m279	2019 Apr 25	11:59:14	g	21.7
NGC 6225	16h48m2157	+06d13m220	2019 Apr 25	12:00:51	g	21.7
2MASX J16462248+0902154	16h46m2249	+09d02m155	2019 Apr 25	12:02:05	g	21.6
PGC 58705	16h39m2639	+11d12m377	2019 Apr 25	12:03:15	g	21.8
2MASX J16552449–0715255	16h55m2450	−07d15m255	2019 Apr 25	12:04:57	g	21.7
2MASX J16580128–0149216	16h58m0129	−01d49m217	2019 Apr 25	12:06:16	g	21.4
2MASX J16590728–0544311	16h59m0728	−05d44m312	2019 Apr 26	08:30:29	i	22.2
PGC 58987	16h47m2448	−20d08m303	2019 Apr 26	08:32:14	i	22.2
NGC 6224	16h48m1855	+06d18m439	2019 Apr 26	08:34:11	i	22.5
NGC 6051	16h04m5670	+23d55m583	2019 Apr 26	08:36:37	i	21.8
IC 4572	15h41m5420	+28d08m027	2019 Apr 26	08:42:22	i	22.0
UGC 10320	16h18m0732	+21d03m590	2019 Apr 26	08:44:30	i	22.5
IC 4569	15h40m4835	+28d17m314	2019 Apr 26	08:47:02	i	22.1
PGC 57607	16h14m5784	+21d56m179	2019 Apr 26	08:50:11	i	22.6
PGC 57472	16h12m2034	+23d00m070	2019 Apr 26	08:51:41	i	22.6
IC 1219	16h24m2744	+19d28m573	2019 Apr 26	08:53:28	i	22.6
UGC 10412	16h29m3614	+15d39m304	2019 Apr 26	08:55:24	i	22.6
NGC 6001	15h47m4596	+28d38m307	2019 Apr 26	08:57:41	i	22.1
PGC 55883	15h43m4604	+28d24m546	2019 Apr 26	08:59:39	i	22.3
PGC 57645	16h15m4215	+19d38m150	2019 Apr 26	09:02:13	i	22.6
PGC 59121	16h51m2166	+07d51m443	2019 Apr 26	09:04:08	i	22.1
PGC 56949	16h04m3557	+25d11m234	2019 Apr 26	09:06:30	i	22.2
PGC 58860	16h44m0928	+07d26m430	2019 Apr 26	09:08:30	i	22.0
PGC 57542	16h13m4601	+22d55m080	2019 Apr 26	09:10:44	i	22.4
UGC 10224	16h08m5024	+22d02m335	2019 Apr 26	09:13:36	i	22.6
PGC 58097	16h25m3808	+16d27m180	2019 Apr 26	09:16:57	i	22.6
PGC 1717114	16h04m1628	+24d48m444	2019 Apr 26	09:18:57	i	22.6
PGC 59239	16h54m2403	−09d53m213	2019 Apr 26	09:23:22	i	22.5
UGC 10260	16h11m5788	+20d55m245	2019 Apr 26	09:25:28	i	22.7
IC 4570	15h41m2256	+28d13m473	2019 Apr 26	09:38:18	i	22.7
UGC 10035	15h47m3635	+26d03m492	2019 Apr 26	09:44:32	i	22.3
PGC 57692	16h16m4571	+19d31m169	2019 Apr 26	09:46:51	i	22.4
2MASX J16505342–1500143	16h50m5342	−15d00m143	2019 Apr 26	09:49:19	i	21.9
NGC 6240	16h52m5886	+02d24m035	2019 Apr 26	09:51:27	i	22.3
UGC 10360	16h23m1134	+16d55m574	2019 Apr 26	09:53:38	i	22.6
PGC 55774	15h40m3664	+28d30m448	2019 Apr 26	09:56:12	i	22.6
PGC 58735	16h40m4022	+14d21m053	2019 Apr 26	09:58:50	i	22.5
IC 4621	16h50m5119	+08d47m019	2019 Apr 26	10:00:50	i	22.5
NGC 6075	16h11m2257	+23d57m545	2019 Apr 26	10:07:05	i	22.5
2MASX J16582619–0319463	16h58m2620	−03d19m464	2019 Apr 26	10:11:42	i	22.5
2MASX J16073961+2220315	16h07m3962	+22d20m315	2019 Apr 26	10:15:40	i	22.5
PGC 58028	16h24m1515	+20d11m010	2019 Apr 26	10:17:30	i	22.6
PGC 54895	15h22m4491	+29d46m110	2019 Apr 26	10:19:52	i	22.3
IC 4505	14h46m3338	+33d24m312	2019 Apr 26	10:21:52	i	22.7
PGC 58768	16h41m2090	+08d54m326	2019 Apr 26	10:24:48	i	22.5
PGC 55373	15h32m4654	+28d22m015	2019 Apr 26	10:27:25	i	22.7
IC 4587	15h59m5161	+25d56m264	2019 Apr 26	10:29:22	i	22.7
UGC 08145	13h02m1828	+32d53m268	2019 Apr 26	10:32:45	i	22.4
PGC 57293	16h09m0646	+24d52m131	2019 Apr 26	10:36:12	i	22.5
2MASX J16540875–0738073	16h54m0876	−07d38m073	2019 Apr 26	10:38:59	i	22.3
PGC 52138	14h35m1842	+35d07m077	2019 Apr 26	10:41:50	i	22.2
2MASX J16153554+1927123	16h15m3554	+19d27m124	2019 Apr 26	10:44:25	i	23.0
PGC 59338	16h58m0576	−21d16m268	2019 Apr 26	10:46:47	i	22.2
UGC 09233	14h24m3503	+35d16m474	2019 Apr 26	10:50:02	i	23.4
PGC 56421	15h56m0387	+24d26m527	2019 Apr 26	10:52:12	i	22.6
PGC 1484188	16h28m5242	+15d25m148	2019 Apr 26	10:54:31	i	22.5
S190426c
2MASX J18191810+8807285	18h19m1811	+88d07m286	2019 Apr 27	08:38:51	i	22.0
2MASX J18215068+8642223	18h21m5068	+86d42m224	2019 Apr 27	08:40:44	i	22.0
2MASX J18242867+8642139	18h24m2867	+86d42m140	2019 Apr 27	08:42:02	i	22.1
2MASX J19301513+8540516	19h30m1513	+85d40m516	2019 Apr 27	08:44:01	i	22.2
2MASX J20412914+8626330	20h41m2914	+86d26m331	2019 Apr 27	08:47:16	i	22.3
2MASX J20441724+8654219	20h44m1724	+86d54m220	2019 Apr 27	08:48:45	i	21.4
PGC 3085923	20h52m2972	+86d11m119	2019 Apr 27	08:50:07	i	22.4
2MASX J20452666+8620428	20h45m2666	+86d20m429	2019 Apr 27	08:53:22	i	22.2
2MASX J20592695+8454369	20h59m2695	+84d54m370	2019 Apr 27	08:55:00	i	22.3
2MASX J20110295+4637149	20h11m0296	+46d37m149	2019 Apr 27	09:00:56	i	22.3
2MASX J20113931+4550035	20h11m3932	+45d50m036	2019 Apr 27	09:03:03	i	22.3
2MASX J20114858+4657335	20h11m4858	+46d57m336	2019 Apr 27	09:04:39	i	22.2
2MASX J20132761+4630313	20h13m2762	+46d30m313	2019 Apr 27	09:06:17	i	22.3
2MASX J20134502+4726333	20h13m4503	+47d26m334	2019 Apr 27	09:07:51	i	22.3
2MASX J20152058+4555282	20h15m2059	+45d55m283	2019 Apr 27	09:10:17	i	22.3
2MASX J20201548+4720364	20h20m1548	+47d20m364	2019 Apr 27	09:12:40	i	22.2
2MASX J20242781+4900526	20h24m2781	+49d00m527	2019 Apr 27	09:14:19	i	22.1
2MASX J20354336+4953165	20h35m4337	+49d53m166	2019 Apr 27	09:15:47	i	22.4
2MASX J20224302+5636145	20h22m4303	+56d36m146	2019 Apr 27	09:17:53	i	22.4
2MASX J20244336+5245430	20h24m4337	+52d45m430	2019 Apr 27	09:19:32	i	22.3
2MASX J20245359+5610264	20h24m5360	+56d10m264	2019 Apr 27	09:20:53	i	22.2
2MASX J20260256+5552523	20h26m0256	+55d52m524	2019 Apr 27	09:22:19	i	22.3
2MASX J20273404+5015483	20h27m3404	+50d15m483	2019 Apr 27	09:23:52	i	22.3
2MASX J20273859+5353393	20h27m3859	+53d53m393	2019 Apr 27	09:25:24	i	22.4
2MASX J20281516+5641284	20h28m1517	+56d41m284	2019 Apr 27	09:26:51	i	22.3
2MASX J20290191+5817016	20h29m0192	+58d17m016	2019 Apr 27	09:28:14	i	22.3
2MASX J20291160+5219510	20h29m1160	+52d19m510	2019 Apr 27	09:29:56	i	22.3
2MASX J20300804+5415120	20h30m0804	+54d15m121	2019 Apr 27	09:31:48	i	22.2
2MASX J20304675+6259395	20h30m4676	+62d59m395	2019 Apr 27	09:37:44	i	22.4
2MASX J20322187+5812031	20h32m2188	+58d12m032	2019 Apr 27	09:39:26	i	22.3
2MASX J20334424+5403120	20h33m4425	+54d03m120	2019 Apr 27	09:41:06	i	22.5
2MASX J20334533+6254178	20h33m4534	+62d54m179	2019 Apr 27	09:42:45	i	22.4
2MASX J20352447+5759548	20h35m2447	+57d59m549	2019 Apr 27	09:44:34	i	22.4
2MASX J20354995+6208172	20h35m4996	+62d08m172	2019 Apr 27	09:47:17	i	22.4
2MASX J20355212+5549587	20h35m5213	+55d49m588	2019 Apr 27	09:49:53	i	22.3
2MASX J20370886+5756538	20h37m0887	+57d56m538	2019 Apr 27	09:51:24	i	22.5
2MASX J20373532+5628217	20h37m3532	+56d28m217	2019 Apr 27	09:52:51	i	22.3
2MASX J20384775+6128473	20h38m4775	+61d28m474	2019 Apr 27	09:54:43	i	22.3
2MASX J20385946+5351220	20h38m5947	+53d51m220	2019 Apr 27	09:56:30	i	22.3
2MASX J20391360+6454369	20h39m1360	+64d54m370	2019 Apr 27	09:58:12	i	22.5
2MASX J20404068+6437003	20h40m4069	+64d37m003	2019 Apr 27	09:59:56	i	22.2
2MASX J20431546+5424171	20h43m1547	+54d24m171	2019 Apr 27	10:02:36	i	22.3
2MASX J20450027+6441410	20h45m0028	+64d41m410	2019 Apr 27	10:04:16	i	22.5
2MASX J20452857+6332249	20h45m2858	+63d32m250	2019 Apr 27	10:05:41	i	22.3
2MASX J20453196+6157519	20h45m3196	+61d57m520	2019 Apr 27	10:07:12	i	22.1
2MASX J20482612+6414178	20h48m2613	+64d14m178	2019 Apr 27	10:08:31	i	22.4
2MASX J20492307+6412331	20h49m2308	+64d12m332	2019 Apr 27	10:10:38	i	22.4
2MASX J20495907+6207478	20h49m5908	+62d07m479	2019 Apr 27	10:12:07	i	22.3
2MASX J20515127+6309235	20h51m5128	+63d09m235	2019 Apr 27	10:13:43	i	22.4
2MASX J20581565+6217178	20h58m1566	+62d17m178	2019 Apr 27	10:15:28	i	22.3

Note.

^aThese limiting magnitudes correspond to 3 times the sky noise within an aperture of 2.5 times the FWHM, where the sky noise is estimated to be 1.48 times the median absolute deviation of the difference image.

Download table as:  ASCIITypeset images: 1 2

For S190426c, we imaged 50 galaxies with 30-s i-band exposures on 2019 April 27 at 08:38:51 to 10:15:57 UT (17.3–18.9 hr post merger; Hosseinzadeh et al. 2019c). No transient sources were uncovered in these observations to a median 3σ limiting magnitude of i = 22.3 (see Table 1).

Multiple teams reported UV, optical, and NIR follow-up imaging of the sky regions of S190425z and S190426c. In Table 2 (S190425z) and Table 3 (S190426c) we collate the available information, and summarize the timing of the observations relative to the merger time, the filter(s) used and limiting magnitudes, the sky area covered or number of galaxies targeted, and the relevant GCN circular references. In Figure 1 we map the searches that reported their telescope pointing coordinates relative to the GW localization regions.

3.1. S190425z

In all, 24 telescopes were reported to follow up S190425z, whose 90% confidence localization volume is about 8 × 10⁶ Mpc³. The galaxy-targeted searches observed a combined total of 418 galaxies in this volume, corresponding to about 1% of the total number of galaxies in the GLADE catalog within the volume. Most searches observed galaxies more luminous than M_B ≈ −19 mag (373 galaxies). Integrating the galaxy luminosity function down to this limit indicates 2.5 × 10⁴ galaxies within the localization volume, confirming that the GLADE catalog is effectively complete at this luminosity. The fraction of observed bright galaxies was thus about 1.5%. To further quantify the effective coverage of the galaxy-targeted searches, we consider only the galaxies that were imaged to a sufficient depth to detect a GW170817-like kilonova (M ≈ −16 mag; Villar et al. 2017) at the distance of each galaxy. We find that 304 out of the 373 galaxies satisfy this criterion, leading to an effective coverage of about 1.2%. In the UV, Swift/Ultra-Violet/Optical Telescope (UVOT) observed 389 galaxies that are both in the GLADE catalog and have a redshift that would make it possible to detect a GW170817-like kilonova at the depth of the observations; this corresponds to an effective coverage of about 1.5%.

Similarly, for the wide-field searches we determine the effective fractional volume coverage using the distance to which each search would have detected a GW170817-like kilonova at the reported limiting magnitude, and combine this effective distance with the reported areal coverage. We find that most of the wide-field searches had an effective fractional volume coverage of about 0%–8%, while ZTF had a fractional volume coverage of about 40%; the values of zero correspond to searches that reported limiting magnitudes too shallow to have detected a GW170817-like kilonova at the lower distance limit of ≈115 Mpc.

Naturally, the fractional coverage of the galaxy-targeted and wide-field searches would be smaller (larger) for a dimmer (brighter) counterpart than in GW170817. We also note that our calculation is simplified, and likely errs on the side of being too optimistic. For example, we are not taking into account variations in Galactic extinction, moon illumination, and other differential observational effects that would generally serve to reduce the efficiency of the searches. On the other hand, other groups may have conducted follow-up campaigns that have not (yet) been publicly reported, which may increase the overall efficiency of the community effort. Lastly, the numbers in Tables 2 and 3 are uncertain due to possible human errors in real-time GCN composition (as ours had), but we assume these uncertainties are small compared to the uncertainty in the kilonova models.

The various searches returned 69 candidate optical counterparts, reported by ZTF, ATLAS, Pan-STARRS, Swift/UVOT, and Gaia, with candidates ranging in brightness from about 14 to 21.5 mag (Anand et al. 2019; Breeveld et al. 2019; Kasliwal et al. 2019; Kostrzewa-Rutkowska et al. 2019a, 2019b; McBrien et al. 2019; Smith et al. 2019); see Figure 3 for the brightness distribution. Of these, 18 candidates were followed up with at least one targeted observation, including seven events that were spectroscopically classified.

Zoom In Zoom Out Reset image size

Two of the classifications were obtained through our spectroscopic follow-up program using the 4.1 m SOAR telescope (Nicholl et al. 2019a). ZTF19aasckkq was selected based on a known spectroscopic redshift of z = 0.0528 for its host galaxy (Anand et al. 2019), which is within the 2σ contour of the the GW localization distance. The absolute magnitude of the source at this distance was −16.3 mag, comparable to GW170817 at peak. We obtained a 1900 s exposure beginning at 2019 April 28 05:13:50 UT using the Goodman High Throughput Spectrograph (Clemens et al. 2004) with the 400 lines/mm grating and a central wavelength of 5750 Å. The data were processed and the one-dimensional spectrum extracted using a custom Iraf pipeline (for details, see Margutti et al. 2019). The spectrum shows broad H i lines at the redshift of the host galaxy, as well as a strong absorption consistent with He i λ5876. Classification using the Supernova Identification code (SNID; Blondin & Tonry 2007) indicates that this transient is likely a young Type IIb supernova (SN IIb; Figure 4). Our second SOAR target was ZTf19aasckwd. This event did not have a reported redshift, but the apparent magnitude of 20.15 was a good match to the brightness of GW170817 at ∼155 Mpc. Moreover, a search of the coordinates in PS1 3π data (Flewelling et al. 2016) showed a likely host galaxy. We obtained a 1500 s exposure in the same setting as above, beginning at 2019 April 28 04:41:03 UT. The spectrum revealed a SN Ia at z = 0.145 (Figure 4). Both spectra are publicly available via the Transient Name Server (ZTF19aasckkq = SN2019eff, ZTf19aasckwd =SN2019eib).

Zoom In Zoom Out Reset image size

The other five candidates classified by the community also turned out to be normal SNe: one SN Ia, three SNe II, and one SN Ib/c (Anand et al. 2019; Buckley et al. 2019; Carini et al. 2019; Castro-Tirado et al. 2019; Dichiara et al. 2019; Dimitriadis et al. 2019; Izzo et al. 2019a; Jencson et al. 2019; Jonker et al. 2019; McCully et al. 2019; Morokuma et al. 2019; Nicholl et al. 2019b, 2019c; Pavana et al. 2019; Perley et al. 2019a; Short et al. 2019a, 2019b; Wiersema et al. 2019). Additionally, a UV candidate uncovered by Swift/UVOT (Breeveld et al. 2019) was shown to be an M dwarf flare (Bloom et al. 2019; Lipunov et al. 2019a). No NIR candidates were announced.

As shown in Figure 3, the bulk of reported candidates overlapped the optical brightness of a GW170817-like kilonova in the 90% confidence distance range (≈19–21 mag). We further find that the subset of events followed up photometrically and/or spectroscopically similarly span the same magnitude range. In terms of the choice of targets for spectroscopic follow-up, four of the seven classified transients (and four of the 11 transients with only photometric follow-up) were selected based on probable associations with galaxies that have secure distance measurements compatible with the GW distance (Anand et al. 2019; Kasliwal et al. 2019; Smith et al. 2019).

Conversely, six of of the 11 sources with only photometric follow-up were announced as apparently host-less (or "orphan") transients (Anand et al. 2019; McBrien et al. 2019). None of these sources were recovered in follow-up imaging (Ahumada et al. 2019a; Nicholl et al. 2019b; Perley & Copperwheat 2019), suggesting that they were potential image artifacts or due to stellar variability. Follow-up of host-less transients therefore appears to be a somewhat risky strategy, although we note that some mergers may occur at large offsets from their hosts: based on the distribution of SGRB offsets (Fong & Berger 2013), about 10% of BNS and/or NS–BH mergers may have offsets of tens of arcseconds from their hosts at the ALV detection distances.

3.2. S190426c

In Figure 5 (left panel) we compare the limiting magnitudes of the various searches (galaxy-targeted and wide-field) to the model optical/NIR light curves of GW170817 (from Villar et al. 2017 ²³ ) shifted to the 90% distance ranges of S190425z and S190426c. We also show the shock cooling model of Piro & Kollmeier (2018), as it predicts potentially brighter emission in the first few hours post merger (which were missed in the case of GW170817). For both events some of the searches reached sufficient depth to detect a GW170817-like kilonova, although this was more challenging for S190426c.

Zoom In Zoom Out Reset image size

In the right panel of Figure 5 we compare the searches to several other models of early optical/NIR emission. We consider a kilonova that lacks the blue (lanthanide-poor) component and has an ejecta mass of 0.01 M_⊙ for the red (lanthanide-rich) component (i.e., about 4 times lower than in GW170817); this model represents a more pessimistic possibility, but one that is supported by binary merger simulations (e.g., Hotokezaka et al. 2011). Both models were generated with MOSFiT (Guillochon et al. 2018). We also show a model for a blue precursor powered by the decay of free neutrons from the shock-heated interface between the NSs (Metzger et al. 2015). In the context of these models, which peak at ≈22 mag for S190425z and ≈24 mag for S190426c, we find that few (if any) observations reached sufficient depth to place meaningful constraints.

Another source of early UV/optical/NIR emission is an on-axis or slightly off-axis relativistic jet, as observed in SGRBs (Berger 2014). In the absence of information about the binary inclination from GW data we cannot directly assess the viewing angle of a potential jet, but we note that based on jet opening angle measurements in SGRBs (Fong et al. 2015) we expect at most a few percent of GW mergers to exhibit on-axis jets (Metzger & Berger 2012). Conversely, for substantial off-axis angles (as was the case for GW170817 with θ_obs ≈ 30°; Alexander et al. 2017, 2018; Margutti et al. 2017, 2018) the optical emission is significantly delayed and exceedingly dim, making this scenario irrelevant for the rapid searches considered here.

We therefore consider two afterglow models: on-axis (θ_obs = 0°) and slightly off-axis (θ_obs = 15°). We generate light curves using the BOXFIT code (v2; van Eerten & MacFadyen 2011) for "top hat" jets²⁴ using median values for cosmological SGRBs (Fong et al. 2015): jet opening angle of θ_j = 10°, isotropic kinetic energy of E_K,iso = 2 × 10⁵¹ erg, circumburst density of n = 4 × 10⁻³ cm⁻³, electron energy power-law index p = 2.4, and fractional post-shock energies in the relativistic electrons and magnetic fields of _E = 0.1 and _B = 0.01, respectively. We note that this model is comparable to the inferred properties of the relativistic jet in GW170817.

The resulting model light curves, scaled to the distances of S190425z and S190426c, are shown in Figure 5 (right panels). We find that for S190425z, the on-axis afterglow model remains brighter than about 20 mag for the first day, exceeding the expected brightness of a GW170817-like kilonova. A substantial fraction of the searches reached sufficient depth to detect such an on-axis jet. In the case of S190426c, however, such an on-axis afterglow would have declined below 20 mag within about 0.3 days, although it would have still been detectable by at least some of the searches in the first day. We stress again that the probability of an on-axis merger is at most a few percent, so even in the case when the observations are sufficiently deep and cover the entire GW localization region, such a detection is unlikely. For the slightly off-axis afterglow model, the peak brightness is about 23 mag for S190425z and about 25 mag for S190426c. These brightness levels are well below the limiting magnitudes of the majority of follow-up observations, indicating the challenge of detecting off-axis afterglows at these distances.

Multiple γ-ray and X-ray space missions reacted to the GW alerts and/or were observing parts of the relevant sky area at the time of merger, including the Monitor of All-sky X-ray Image (MAXI), the Neil Gehrels Swift Observatory, the INTErnational Gamma-Ray Astrophysics Laboratory (INTEGRAL), the High-Altitude Water Cherenkov Observatory (HAWC), the Fermi Gamma-ray Space Telescope, the AstroRivelatore Gamma a Immagini Leggero (AGILE), the Hard X-ray Modulation Telescope (Insight-HXMT) and the CALorimetric Electron Telescope (CALET). No high-energy counterpart was identified with high statistical significance for either S190425z or S190426c (Axelsson et al. 2019a, 2019b; Barthelmy et al. 2019; Casentini et al. 2019; Chelovekov et al. 2019; Evans et al. 2019; Fletcher 2019a, 2019b; Guan et al. 2019; HAWC Collaboration 2019a, 2019b; Minaev et al. 2019; Piano et al. 2019a, 2019b; Sakamoto et al. 2019; Shimizu et al. 2019; Sugita et al. 2019; Sugizaki et al. 2019; Tamura et al. 2019; Xiao et al. 2019; Yi et al. 2019). The fractional localization coverage at the time of the GW detection varies widely, from a few percent to nearly 100%.

For S190425z there are four measurements of interest: (i) Fletcher (2019a) reported a Fermi-GBM 3σ flux limit in the range ${F}_{10-1000\mathrm{keV}}\lt (0.1-3)\times {10}^{-6}\,\mathrm{erg}\,{{\rm{s}}}^{-1}\,{\mathrm{cm}}^{-2}$ (depending on the assumed spectral model) for observations obtained at ±30 s relative to the merger time using a 1 s integration time. This corresponds to a luminosity limit of ${L}_{1-{10}^{4}\mathrm{keV}}\lt (0.03-6)\,\times {10}^{49}\,\mathrm{erg}\,{{\rm{s}}}^{-1}$ at 155 Mpc. These observations covered 51% of the initial probability map. (ii) Konus–Wind was observing the entire sky at the time of the GW trigger. Svinkin et al. (2019a) report a flux limit of $2.7\times {10}^{-7}\,\mathrm{erg}\,{{\rm{s}}}^{-1}\,{\mathrm{cm}}^{-2}$ (20–1500 keV) for a SGRB-like spectrum. (iii) Martin-Carillo et al. (2019) and Savchenko et al. (2019) reported the presence of a possible excess of γ-ray emission with limited significance in INTEGRAL data acquired ∼6 s after the GW detection. However, this excess is most probably due to background fluctuations. Under this preferred assumption, V. Savchenko et al. (2019, in preparation) estimate a typical 3σ upper limit on the 75–2000 keV fluence within 50% GW probability containment region of $(2-6)\times {10}^{-7}\,\mathrm{erg}\,{\mathrm{cm}}^{-2}$ depending on the sky location and for a burst lasting less than 1 s with a typical SGRB spectrum. (iv) In the time interval of 1054–5520 s post merger MAXI observed an area of the sky corresponding to 81% of the probability map, with a 1σ flux limit of ${F}_{4-10\mathrm{keV}}\lt 2\times {10}^{-10}\,\mathrm{erg}\,{\mathrm{cm}}^{-2}\,{{\rm{s}}}^{-1}$ corresponding to ${L}_{4-10\mathrm{keV}}\lt 6\times {10}^{44}\,\mathrm{erg}\,{{\rm{s}}}^{-1}$ (Sugizaki et al. 2019).

Similarly, for S190426c there are four measurements of interest: (i) Fletcher (2019b) reported a Fermi-GBM 3σ flux limit in the range ${F}_{10-1000\mathrm{keV}}\lt (1-9)\times {10}^{-7}\,\mathrm{erg}\,{{\rm{s}}}^{-1}\,{\mathrm{cm}}^{-2}$ at ±30 s relative to merger for a 1 s integration time, corresponding to ${L}_{1-{10}^{4}\mathrm{keV}}\lt (0.1-10)\times {10}^{49}\,\mathrm{erg}\,{{\rm{s}}}^{-1}$ at 377 Mpc. These observations covered about 100% of the initial probability region. (ii) From Konus–Wind observations covering the entire sky, Svinkin et al. (2019b) reported a flux limit of $7.3\,\times {10}^{-7}\,\mathrm{erg}\,{{\rm{s}}}^{-1}\,{\mathrm{cm}}^{-2}$ (20–1500 keV) for an SGRB-like spectrum. (iii) Swift/BAT observations covering 95% of the initial probability region obtained at ±100 s relative to merger indicate ${F}_{15-350\mathrm{keV}}\,\lt {10}^{-6}\,\mathrm{erg}\,{\mathrm{cm}}^{-2}\,{{\rm{s}}}^{-1}$ for a 1 s integration time, corresponding to ${L}_{15-350\mathrm{keV}}\lt 2\times {10}^{49}\,\mathrm{erg}\,{{\rm{s}}}^{-1}$ (Barthelmy et al. 2019). (iv) MAXI covered 76% of the probability region at 750–4488 s post merger to a 1σ limit of ${F}_{4-10\mathrm{keV}}\,\lt 2\times {10}^{-10}\,\mathrm{erg}\,{\mathrm{cm}}^{-2}\,{{\rm{s}}}^{-1}$, corresponding to ${L}_{4-10\mathrm{keV}}\lt 3\,\times {10}^{45}\,\mathrm{erg}\,{{\rm{s}}}^{-1}$ at d = 377 Mpc (Shimizu et al. 2019).

As shown in Figure 6, the limits on the prompt γ-ray emission for both events can rule out an on-axis SGRB comparable to the bulk of the energetic cosmological population, which have E_γ,iso ≈ 10⁵¹–10⁵² erg. Song et al. (2019) and Saleem et al. (2019) reached a similar conclusion. As indicated in Section 5, however, an on-axis orientation is expected in at most a few percent of the cases.

Zoom In Zoom Out Reset image size

In Figure 7 we compare the limits from MAXI to the observed X-ray afterglows of cosmological SGRBs, and find that they similarly rule out about half of the observed population, for the fractional areal coverage of each MAXI search. The same caveat about the rarity of on-axis events applies to these limits as well. On the other hand, if we compare the MAXI limits to the same off-axis model described in Section 5, we find that such a model cannot be constrained. We similarly find that Swift/XRT observations carried out for both events (Evans et al. 2019; Tohuvavohu et al. 2019a, 2019b) cannot constrain off-axis jets (Figure 7).

Zoom In Zoom Out Reset image size

We presented MMT follow-up observations of the first two ALV candidate events in O3 that appear to contain NSs, and are therefore capable of generating EM emission. Our MMT search targeted 67 and 50 galaxies for S190425z and S190426c, respectively, and did not yield potential counterparts to a limiting magnitude of i ≈ 22.5. We further presented our spectroscopic follow-up with SOAR of two candidate optical counterparts from other searches, which revealed unrelated SNe Ia and IIb. For comparison we further collated information available from the GCN circulars about other galaxy-targeted and wide-field searches. Due to the large localization areas and volumes of both events all searches were far from complete. Still, a combined total of nearly 100 optical candidates were announced for the two events, and 14 were followed up spectroscopically, revealing normal SNe.

Parameterizing the efficacy of the searches relative to the brightness of the kilonova associated with GW170817, we find maximal volume coverage of about about 40% for S190425z (ZTF) and about 60% for S190426c (ZTF plus DECam). Relative to a dimmer kilonova model (0.01 M_⊙ of lanthanide-rich ejecta), a neutron precursor, or a slightly off-axis SGRB, we find that only a few searches (including our MMT observations) reached sufficient depth. On the other hand, comparing to an on-axis SGRB we find that most searches would have been able to detect such emission, but this is expected in at most a few percent of mergers.

We end with a few general thoughts. First, the open rapid alerts implemented by the LIGO/Virgo Collaboration in O3 work remarkably well in providing rapid access to sky maps, distance estimates, and rudimentary information about the detections (e.g., FAR). Second, the events considered here indicate that due to duty cycle limitations and the larger detection distances, the localization regions (and volumes) for most events will be much larger than for GW170817, and this is likely to reduce the efficiency of counterpart identification. Third, despite the larger distances of the GW events at least some searches are capable of reaching the depth necessary to detect GW170817-like kilonovae (if those are common). Finally, we note that the robustness of the GW detections, as well as the actual properties of the binaries, are difficult to assess with the partial information provided by the LIGO/Virgo Collaboration at the present. In particular, we advocate breaking down the FAR by detector and by search pipeline, which would provide more guidance about the significance of a given event. Furthermore, the chirp mass and the individual component masses potentially provide critical insight about the expected EM signatures (Margalit & Metzger 2019). Early release of this additional information is particularly important as the number of detections increases in order to prioritize follow-up and tailor it to the properties of the transient.

We thank Dallan Porter and Sean Moran for software development that facilitated our rapid follow-up program at MMT, as well as Ben Kunk for operating the MMT during our observations. We also thank Joey Chatelain for suggesting the use of SkyBoT in our analysis software, and Cristiano Guidorzi for help with the BOXFIT simulations. The Berger Time-Domain Group is supported in part by NSF grant AST-1714498 and NASA grant NNX15AE50G. G.H. thanks the LSSTC Data Science Fellowship Program, which is funded by LSSTC, NSF Cybertraining Grant #1829740, the Brinson Foundation, and the Moore Foundation; his participation in the program has benefited this work. P.S.C. is grateful for support provided by NASA through the NASA Hubble Fellowship grant #HST-HF2-51404.001-A awarded by the Space Telescope Science Institute, which is operated by the Association of Universities for Research in Astronomy, Inc., for NASA, under contract NAS 5-26555. M.N. is supported by a Royal Astronomical Society Research Fellowship. W.F. and K.P. acknowledge support by the National Science Foundation under award No. AST-1814782. I.P. acknowledges funding by the Deutsche Forschungsgemeinschaft under grant GE2506/12-1. R.M. acknowledges support for this work provided by the National Aeronautics and Space Administration through Chandra award No. DD8-19101A and DDT-18096A issued by the Chandra X-ray Center, which is operated by the Smithsonian Astrophysical Observatory for and on behalf of the National Aeronautics Space Administration under contract NAS8-03060. K.D.A. is grateful for support provided by NASA through the NASA Hubble Fellowship grant #HST-HF2-51403.001-A awarded by the Space Telescope Science Institute, which is operated by the Association of Universities for Research in Astronomy, Inc., for NASA, under contract NAS 5-26555. Development of the Boxfit code was supported in part by NASA through grant NNX10AF62G issued through the Astrophysics Theory Program and by the NSF through grant AST-1009863. This research was supported in part through the computational resources and staff contributions provided for the Quest high-performance computing facility at Northwestern University, which is jointly supported by the Office of the Provost, the Office for Research, and Northwestern University Information Technology. Observations reported here were obtained at the MMT Observatory, a joint facility of the Smithsonian Institution and the University of Arizona, and the Southern Astrophysical Research (SOAR) telescope, which is a joint project of the Ministério da Ciência, Tecnologia, Inovações e Comunicações (MCTIC) do Brasil, the U.S. National Optical Astronomy Observatory (NOAO), the University of North Carolina at Chapel Hill (UNC), and Michigan State University (MSU).

Facilities: ADS - , MMT (MMTCam) - , SOAR (Goodman). -

Software: Astropy (Astropy Collaboration 2018), BOXFIT (van Eerten & MacFadyen 2011), Flask, healpy (Zonca et al. 2019), JS9 (Mandel & Vikhlinin 2018), ligo.skymap (Singer 2019b), Matplotlib (Hunter 2007), MOSFiT (Guillochon et al. 2018), NumPy (van der Walt et al. 2011), Photutils (Bradley et al. 2019), PyGCN (Singer 2019a), PyZOGY (Guevel & Hosseinzadeh 2017), SciPy (Oliphant 2007), SEP (Bertin & Arnouts 1996; Barbary 2016), SNID (Blondin & Tonry 2007), SQLAlchemy.