An Extremely Luminous Flare Recorded from a Supermassive Black Hole

Abstract

Since their discovery more than 60 years ago, accreting supermassive black holes in active galactic nuclei (AGN) were recognized as highly variable sources, requiring an extremely compact, dynamic environment. Their variability traces to multiple phenomena, including changing accretion rates, temperature changes, foreground absorbers, and structural changes to the accretion disk. Spurred by a new generation of time-domain surveys, the extremes of black hole variability are now being probed. We report the discovery of an extreme flare by the AGN J224554.84+374326.5, which brightened by more than a factor of 40 in 2018. The source has slowly faded since then. The total emitted UV/optical energy to date is $\sim10^{54}$ erg, i.e., the complete conversion of approximately one solar mass into electromagnetic radiation. This flare is 30 times more powerful than the previous most powerful AGN transient. Very few physical events in the Universe can liberate this much electromagnetic energy. We discuss potential mechanisms, including the tidal disruption of a high mass $(>30\, M_\odot)$ star, gravitational lensing of an AGN flare or supernova, or a supermassive (pair instability) supernova in the accretion disk of an AGN. We favor the tidal disruption of a massive star in a prograde orbit in an AGN disk.

Figures (12)

• Figure 1: Photometric data for J2245$+$3743 from PS1, CRTS, ZTF, ATLAS, and WISE surveys, as indicated (see text for details). The dashed vertical lines indicate when spectra were obtained (see Supplementary Figure 3). Data are presented as observed values with one sigma measurement errors.
• Figure 2: Peak absolute magnitude vs. restframe timescale to fade to 50% peak flux. This shows the overlapping regions from Hinkle24 as well as the locations of the three ENTs published to date, and the brightest classical TDE to date (see text). The dotted boundary indicates a putative region containing intermediate mass TDEs. Data are presented as observed values with standard deviation uncertainties.
• Figure 3: A comparison of the normalized light curve of J2245$+$3743 with ENTs, including AT2021lwx. The ENT light curves show Gaussian process fits to the photometric data with a Matérn-3/2 kernel. The dotted and dashed lines show the best-fit power law fits to the declining part of the light curve for -5/3 and -2/3 exponents, respectively. J2245$+$3743 shows a similar fast rise, like the superluminous TDE-He AT2023vto, but then a much slower decline, similar to ENTs. Data are presented as observed values with one sigma measurement errors.
• Figure 4: Forced photometric data in flux space for the flare in J2245$+$3743 from PS1, CRTS, ZTF, and ATLAS. CRTS and ATLAS $o$-band data have been transformed to PS1 $r$-band for direct comparison. Host contributions have been removed from the CRTS and PS1 data. The right panel highlights the region defined by the dashed lines in magnitude space. The dotted line indicates a best fit model to the combined data for the flare. The model Mendoza22 describes the time profile of energy release in the flare and features a Gaussian rise to account for impulsive heating and a double exponential to account for rapid and gradual cooling phases. Data are presented as observed values +/- one sigma measurement errors.
• Figure 5: Multiband photometric data for J2245$+$3743. (Top) The apparent magnitude evolution of J2245$+$3743 from ZTF $g$- (green) and $r$-bands (red) and ATLAS $c$- (cyan) and $o$-bands (orange). The black points show the combined data set calibrated to the $r$-band. The color bands indicate Gaussian process regression fits to the individual filter light curves and their variance. The vertical dotted line indicates the peak of the flare at MJD = 58295 days and the grayed region indicates the pre-peak time period when there is insufficient data for the Gaussian process and the fit is extrapolated. (Bottom) The color evolution of J2245$+$3743. Minimal color evolution (i.e., statistically consistent with zero evolution) from the flare during its cooling phase. Data are presented as observed values +/- one sigma measurement errors.