Rate of Repeating Tidal Disruption Events with 5--19 years interval constrained by CRTS and ZTF

TL;DR

The study targets long-interval repeating tidal disruption events (rTDEs) with $5$--$19$ year separations by cross-matching CRTS archival flares with a well-defined ZTF BTS TDE sample ($z<0.05$). Two strong long-interval rTDE candidates, AT 2019azh and AT 2024pvu, are identified with rest-frame cadences of $13.2$ and $17.1$ years, and UV/optical analysis (including GALEX data) supports a TDE origin over SNe. The authors show the inferred TDE rate for this window is $2$--$3$ dex higher than the average, and, accounting for possible missed events, suggest that $25$%--$60$% of optical TDEs may be rTDEs for intervals $<20$ years. They argue in favor of repeating partial TDEs (rpTDE) as a plausible origin, though independent TDEs cannot be entirely ruled out; future flares and larger samples are needed to test periodicity and distinguish between scenarios. The work highlights that long-interval rTDEs, previously overlooked, can significantly bias TDE statistics and offers testable predictions for upcoming observations.

Abstract

Statistics on tidal disruption events (TDEs) may be contaminated by repeating TDEs (rTDEs), which have been extensively discovered recently. However, the origin of rTDEs remains unclear. In addition, no statistical research on rTDEs with time intervals $>5$ years has been made yet. In this work, we searched for rTDEs with time intervals of 5--19 years using CRTS data in a sample of 16 ZTF BTS TDEs at $z<0.05$. We found 2 rTDE candidates, AT 2019azh and AT 2024pvu, with time intervals of 13.2 and 17.1 years, respectively. The peak luminosities of CRTS flares are close to those of ZTF flares. For the CRTS flare of AT 2024pvu, using GALEX UV observations near the peak, we measured a blackbody temperature of $\sim19500$ K, consistent with TDEs and higher than SNe. Moreover, we estimated the expected number of SNe in the sample to be $\lesssim0.08$, and hence the probability that both CRTS flares are SNe is only 0.3\%. Therefore, the possibility that both CRTS flares are SNe can be ruled out, and it is likely that both are TDEs. Using the two rTDEs, we inferred that the TDE rate is 2--3 orders of magnitude higher than the average over 5--19 years prior to TDE detection. Considering another two rTDEs with intervals of $\sim$2 years in the sample and possible rTDEs missed by CRTS, rTDEs with intervals of $<20$ years may account for 25\%--60\% of the TDE sample. We prefer to explain rTDEs as repeating partial TDEs, but the possibility of independent TDEs cannot be ruled out and requires future observational tests.

Figures (4)

• Figure 1: Host-subtracted optical light curves of AT 2019azh and AT 2024pvu. The insert panels show the partial enlargement of the CRTS flares and the GP models that fit the data (MCMC). For AT 2024pvu, we labeled the observation time of GALEX using a green dashed line.
• Figure 2: (a): The UV to MIR SED of AT 2024pvu. Blue points are the host galaxy's SED in the quiescent state, and black line is the best-fitting model from CIGALE with the minimum $\chi^2$. We show the GALEX and Swift UV photometries using red pentagrams and red triangles, respectively. The latter is consistent with the CIGALE's prediction (black open circle and the error bar), whereas the former shows a significant excess. (b): The SED of the CRTS flare at the time of GALEX observation, and the blackbody models that fit it (MCMC).
• Figure 3: (a): An example using AT 2020vdq of how to calculate the EMD. We show the 5$\sigma$ upper-limit of each CRTS data point using black triangles, and the maximum $L_{\rm V}$ allowed by the observations as a function of $t_{\rm peak}$ using the green line. We converted the observational time to the phase before the ZTF TDE in the rest-frame. For a given $L_V$ (we show an example for $L_V\sim10^{43.14}$ erg/s, corresponding to $M_V\sim-19.4$), the time ranges when the maximum allowed $L_V$ is below this luminosity (red) are considered to have been effectively monitored, while others (blue) are not. (b): The resultant EMDs as functions of peak $L_V$ for the sample. We show the average EMD using the black dashed line, and the typical $L_{\rm peak}$ range of SN Ia ($-18<M_V<-19.4$) using the red shade.
• Figure 4: (a): The same example as Figure \ref{['fig:AT2020vdq_SNIa']}(a), but for TDEs. (b): The TDE LF in our sample, with EMD calculated assuming different time scales (blue, green, and red for fast, intermediate and slow types, respectively). We present the best estimates in solid lines, and the $1\sigma$ upper limit for the fast type and the $1\sigma$ lower limit for the slow type, both shown in dashed lines. We also show the TDE LF from Yao2023 in grey shade for comparison.