Search for high energy $\gamma$-rays from the direction of the candidate electromagnetic counterpart to the binary black hole merger gravitational-wave event S190521g

The gravitational-wave event S190521g -- a likely binary black hole merger in the accretion disk of an active galactic nucleus -- was accompanied by an optical counterpart. Such dense environments around luminous energy release regions are favourable for high energy $\gamma$-ray production. We report on a search for high energy $\gamma$-rays from the direction of the candidate electromagnetic counterpart to the S190521g event using publicly-available data of the Fermi-LAT space $\gamma$-ray telescope. No significant signal was found. We present upper limits on the spectral energy distribution of the source in the 100 MeV - 300 GeV energy range. We discuss the importance of studying S190521g-like transients in the context of cosmic ray acceleration, $\gamma$-ray and neutrino production in such sources.


Introduction
Very recently, the first detection of a plausible optical electromagnetic counterpart to a candidate binary black hole merger S190521g was reported in [1]. Namely, an optical flare with the duration of ∼ 50 days was detected with the Zwicky Transient Facility, indicating that this merger occured inside the accretion disk of J124942.3 + 344929 1 -an active galactic nucleus situated at the redshift of z = 0.438. This observation is interesting in the context of γ-ray astronomy for the following two reasons: 1) a very high value of the estimated bolometric luminosity of the flare L bol ∼ 10 45 erg/s 2 , 2) a relatively dense environment around the energy release region -both in terms of matter 3 and background photon density. Indeed, the Eddington ratio for J1249 + 3449 is ∼0.02-0.2, typical for quasars (e.g. [2]) that usually have dense accretion flows [3] as well as broad line regions (BLRs), filled with photon fields reflected from BLR clouds. These circumstances make the region around * Corresponding author * * Corresponding author Email addresses: podlesnyi.ei14@physics.msu.ru (Egor Podlesnyi), timur1606@gmail.com (Timur Dzhatdoev) 1 hereafter called J1249 + 3449 for simplicity 2 given that the estimated mass of the final black hole is ∼ 100 solar masses [1] 3 the estimated gas density in the accretion disk is ∼ 10 −10 g/cm 3 [1] the merger favourable for particle acceleration and subsequent γ-ray and neutrino production.
For the apparent isotropic bolometric luminosity L bol ∼ 10 45 erg/s ∼ 10 57 eV/s and the luminosity distance to the source d L ∼ 3 · 10 3 Mpc ∼ 10 28 cm the expected energy flux 4 could be as high as F E ∼ L bol /(4πd 2 L ) ∼ 1 eV/(cm 2 s), if the power transferred to high energy (HE, E > 100 MeV) particles is comparable with L bol . F E ∼ 1 eV/(cm 2 s) is still within the capabilities of the Fermi-LAT space γ-ray telescope [4], motivating the search for HE γrays from this source.
In the present paper we perform a search for HE γ-rays from the direction of J1249 + 3449 on a month -year timescale using publicly-available data of the Fermi-LAT space γ-ray telescope. We note that no significant signal was found with Fermi-LAT over a short time period of 10 ks [5].

Fermi-LAT data analysis
Here we derive upper limits on the observable spectral energy distribution (SED = E 2 dN/dE) of J1249 + 3449. We select Fermi-LAT data within two time windows: 1) 2019, May 19, 00:00:01 UTC -2019, September 1, 00:00:01 UTC; 2) 2019, May 19, 00:00:01 UTC -2020, June 30, 00:00:01 UTC. The region of interest (ROI) is We performed an unbinned likelihood analysis of the selected data. We constructed a model of the observed emission including the following sources that could contribute to the detected γ-ray counts inside the ROI: 1) J1249 + 3449 itself, modelled as a point-like source with a power-law spectrum, situated at the center of the ROI, 2) all sources from the Fermi 8-Year Point Source Catalog (4FGL) [6] located within 20 • from the center of the ROI, 3) galactic and isotropic diffuse γ-ray backgrounds using models provided by the Fermi-LAT Collaboration. For J1249 + 3449 we set both spectral index and normalization as free parameters; for point-like sources within 5 • from the center of the ROI and the diffuse backgrounds only the normalizations were left free, while the spectral shapes were fixed; for point-like sources beyond 5 • from the center of the ROI both normalizations and shapes were fixed according to the catalog values.
Using this model of the observed emission, we perform the maximization of the likelihood. We calculate the value of the test statistic T S corresponding to the hypothesis of the J1249 + 3449 emission being present in the dataset against the null hypothesis of it being absent. We obtain T S ≪ 1, i.e. no significant γ-ray flux was detected from this object. Then we derive upper limits (95% confidence level) on the SED, using likeSED.py [7]. These upper limits for both considered time intervals are shown in Fig. 1. We also performed an independent binned analysis using the fermipy package [8] and derived upper limits with the gta.sed method implemented in this package, but the obtained limits are weaker than for the case of the unbinned analysis.

Discussion
There are many possible explanations for the negative results of the search for HE γ-rays reported above, including the following ones: 1) the bolometric luminosity L bol and/or the power transferred to HE particles could have been significantly overestimated; 2) HE γ-ray production efficiency could be significantly lower than unity (especially for the case of primary protons or nuclei); 3) HE γ-rays could have been beamed out away from the line-of-sight; 4) HE γ-rays could have been absorbed by the material of the accretion disk, or photon fields of the accretion disk corona and/or thermal photon field created by hot gas around the merger. The absorption of γ-rays on photon fields of the BLR is usually significant at E > 10 − 30 GeV (e.g. [9]). This might impair the prospects of detecting S190521g-like transients with imaging atmospheric Cherenkov telescopes such as H.E.S.S. [10,11], MAGIC [12,13], VERITAS [14,15], or CTA [16,17].
A detailed study of these effects is underway and will be published elsewhere. X-ray data may be helpful in constraining some models, especially those that include the process of electromagnetic cascade development in the source (both in matter and photon fields) with subsequent synchrotron emission of cascade electrons.
Of course, other multiwavelenght/multimessenger data could also be helpful, in particular, very high energy (VHE, E > 100 GeV) neutrino searches from the direction of J1249 + 3449. S190521g-like transients could be copious sources of two components of VHE neutrinos: 1) "hadronuclear" neutrinos coming from interactions of accelerated protons or nuclei with the material of the accretion disk and 2) "photohadronic" neutrinos from interactions of these protons or nuclei with photon fields in the source. Such neutrinos could contribute to the IceCube diffuse neutrino flux [18,19]. A characteristic signature of this twocomponent neutrino flux is an "ankle" connecting relatively hard photohadronic component and a softer hadronuclear component. We believe that the γ-ray limits presented in this paper could prove to be helpful in constraining models of γ-ray, cosmic-ray, and neutrino production in S190521g-like transients.
them. E. I. Podlesnyi thanks the Foundation for the Advancement of Theoretical Physics and Mathematics "BA-SIS" for the support in participation at the aforementioned school (travel-grant no. 19-28-030) and for the student scholarship (agreement no. 19-2-6-195-1).