🔭 astrophysics

Lost and Found - A gallery of overlooked optical nuclear transients from the ZTF archive

This paper presents a catalog of 19 optical nuclear transients discovered through the Fink alert broker's analysis of the ZTF archive, including new candidates for repeated TDEs, extreme nuclear transients, and an exceptionally long-lived event, thereby demonstrating the value of archival searches while highlighting necessary improvements for future detection systems.

Original authors: E. Quintin, E. Russeil, M. Llamas Lanza, S. Karpov, E. E. O. Ishida, J. Peloton, M. V. Pruzhinskaya, A. Möller, M. Giustini, G. Miniutti, R. S. Saxton, P. Sánchez-Sáez, S. Zheltoukhov, A. Dodin, A. Be

Published 2026-06-24

📖 5 min read🧠 Deep dive

CC BY 4.0

Original authors: E. Quintin, E. Russeil, M. Llamas Lanza, S. Karpov, E. E. O. Ishida, J. Peloton, M. V. Pruzhinskaya, A. Möller, M. Giustini, G. Miniutti, R. S. Saxton, P. Sánchez-Sáez, S. Zheltoukhov, A. Dodin, A. Belinski

Original paper licensed under CC BY 4.0 (http://creativecommons.org/licenses/by/4.0/). ✨ This is an AI-generated explanation of the paper below. It is not written or endorsed by the authors. For technical accuracy, refer to the original paper. Read full disclaimer

Imagine the universe as a giant, bustling city. Most of the time, the "nucleus" (the center) of a galaxy is quiet, like a library that's closed for the night. But sometimes, something dramatic happens in the center: a star gets too close to a supermassive black hole and gets ripped apart. This event, called a Tidal Disruption Event (TDE), is like a cosmic explosion that lights up the night sky for months or even years.

This paper is essentially a "Lost and Found" report. The authors built a new digital tool (a smart filter) to catch these explosions as they happen in real-time data from the Zwicky Transient Facility (ZTF), a telescope that takes pictures of the sky every night. But while they were building this tool, they also went digging through the "trash bins" of old data (the archive) and found 19 interesting events that had been overlooked or mislabeled by previous systems.

Here is a breakdown of their discoveries using simple analogies:

1. The "Long-Lived" Ghost

Most of these cosmic explosions fade away quickly, like a firework that burns out in seconds. However, they found one object, AT2020ukj, that has been glowing for over five years and shows no sign of stopping.

The Analogy: Imagine a firework that, instead of fizzling out after a few seconds, keeps burning with the same intensity for five years.
The Mystery: It happened in a galaxy that was previously "quiet" (passive). The authors aren't sure if this is an incredibly slow-motion star destruction or if the quiet galaxy suddenly "woke up" and started feeding its black hole (turning into an Active Galactic Nucleus). It's like finding a sleeping bear that suddenly starts roaring for five years straight.

2. The "Double-Act" and the "Imposter"

Some of these events happen twice.

The Real Double-Act: They found AT2023adr, which flared up, went quiet, and then flared up again about a year later. This suggests the star didn't die the first time; it got a "partial haircut" from the black hole, survived, and then came back for a second, smaller haircut.
The Imposter: They also found a case that looked like a double event (AT2019agc). It flared, went quiet, and flared again. But when they looked closely at the exact location of the flashes, they realized it wasn't the same star twice. It was actually two different supernovae (exploding stars) happening in the same galaxy, two and a half years apart, at slightly different spots.
The Lesson: It's like seeing a car crash in a parking lot, then seeing another crash in the same lot a year later. Your brain might think it's the same car crashing twice, but it's actually two different cars. The authors' new tool needs to be smart enough to tell the difference.

3. The "Super-Bright" Giants

They found several events that were incredibly bright, much brighter than a normal star explosion.

The Analogy: If a normal supernova is a lightbulb, these are like searchlights or lighthouses.
The Discovery: Three of these events were so bright they are classified as "Extreme Nuclear Transients." They happened in galaxies that were already active (like a busy highway). The authors suggest these might be the result of a star being torn apart inside the busy traffic of an active galaxy, creating a massive, chaotic flare.

4. The "Faint" Whispers

Not all explosions are loud. They found AT2023npi, which was very dim but lasted a long time.

The Analogy: This is like finding a tiny, flickering candle in a dark room that somehow stays lit for years.
The Mystery: It was so faint that standard models say it shouldn't exist. It might be a very small star being eaten, or it might be hidden behind a thick cloud of dust that makes it look dimmer than it really is.

5. The "Late Bloomers"

Some of these events had a weird habit: they started to fade, and then suddenly got brighter again or stopped fading.

The Analogy: Imagine a candle that is burning down, and then someone pours a little more wax on it, making the flame jump up again.
The Implication: This suggests the black hole might be "re-fueling" or that the debris from the destroyed star is swirling around in a complex way before finally disappearing.

Why Does This Matter?

The authors argue that our current "filters" for finding these events are too strict. They are like bouncers at a club who only let in people who look exactly like the standard "TDE" guest list. Because of this, we are missing the weird, long, faint, or repeated events that are actually the most interesting for understanding how black holes work.

The Main Takeaway:
By digging through old data and using a more open-minded search tool, the authors found a "gallery" of overlooked cosmic events. They are urging future telescopes (like the upcoming LSST) to be less picky. Instead of just looking for the "perfect" explosion, we need to look for the weird, the long, the faint, and the repeated ones, because those are the ones that will teach us the most about the physics of the universe. They also emphasize that we need to check if a "repeat" event is actually the same object or just two different things happening in the same neighborhood.

Technical Summary: Lost and Found — A gallery of overlooked optical nuclear transients from the ZTF archive

Problem and Context
The diversity of transient astrophysical sources has expanded significantly with large-scale surveys, yet the boundaries between classes of optical nuclear transients remain fluid. While Tidal Disruption Events (TDEs) are well-studied, the current sample is limited, preventing strong constraints on physical models. Furthermore, existing detection classifiers often prioritize purity over completeness, potentially rejecting outliers or events in active galactic nuclei (AGNs) that do not fit standard templates. This paper addresses the need to recover overlooked transients from archival data to better understand the population of optical nuclear transients, including TDEs, Ambiguous Nuclear Transients (ANTs), Extreme Nuclear Transients (ENTs), and supernovae (SNe).

Methodology
The authors utilized the Zwicky Transient Facility (ZTF) data stream, processed through the Fink alert broker. The study employed a dual approach to candidate identification:

Systematic Search: A dedicated early TDE detection module (Llamas Lanza et al. 2025) was applied to ZTF data from late 2019 to mid-2025. This module uses a physically-motivated parametric fit (the Rainbow model) on rising lightcurve segments followed by a machine learning classifier. The system was designed for high completeness (recall ~76%) rather than high purity, intentionally flagging outliers and events superimposed on AGN variability.
Serendipitous Discovery: Additional transients were identified manually during the development of the detection tool.

For the resulting sample of 19 optical nuclear transients, the authors conducted a multi-wavelength analysis:

Photometry: Forced photometry from ZTF ( $g, r, i$ bands), Catalina Real-time Transient Survey (CRTS, $V$ -band), and NeoWISE ( $W1, W2$ bands) to detect infrared echoes.
Host Characterization: Determination of host galaxy activity (passive vs. active) using spectroscopic data (SDSS, DESI), Gaia DR3 classifications, and infrared color criteria ( $W1-W2 > 0.8$ ).
Modeling: Lightcurves were fitted using two approaches:
- Phenomenological: A Gaussian rise followed by a power-law decay ( $t^{-5/3}$ ) to extract timing parameters.
- Physically-motivated: The tde_fallback model within the Redback Python package, using Bayesian inference (MCMC via dynesty) to estimate physical parameters such as black hole mass, stellar mass, viscosity timescales, and accretion efficiency.

Key Contributions and Results
The paper reports on 19 optical nuclear transients, categorized as follows:

TDE Candidates in Passive Galaxies (9 objects):
- AT2020ukj: An exceptionally long-lived transient (>5 years) in a passive galaxy, showing similarities to ZTF19acnskyy ("Ansky"). Its slow decay suggests extreme physical parameters (high viscosity, low efficiency) or a "turn-on" AGN scenario.
- AT2023adr: A candidate for a repeated partial TDE, displaying a second peak ~400 days after the first with a lower amplitude. Spectroscopy during the second peak showed TDE-consistent features (e.g., [O II], [O III], N III/He II) but no SN decay lines.
- AT2023npi: An extremely faint TDE candidate in a nearby galaxy, potentially the faintest optical TDE to date, with a strong infrared echo.
- AT2020pno: Exhibits a late-time plateau in its decay, a feature consistent with TDE models but often missed by standard classifiers.
- The remaining candidates (AT2020aexc, AT2020afap, AT2023jag, ZTF20aatpzog, AT2021ovg) display standard TDE characteristics (blue color, slow decay, lack of cooling).
ANT/ENT Candidates in Active Galaxies (8 objects):
- ZTF23abjvojy: A repeated flare in a Narrow-Line Seyfert 1 (NLSy1) galaxy, with a second event occurring ~16 years after a previous CRTS detection. This represents the longest baseline for a repeated nuclear transient in an AGN.
- ENT Candidates: AT2022yhf, AT2024nxp, and AT2023rav are exceptionally bright (absolute $g$ -band magnitude $<-24$ ), qualifying them as Extreme Nuclear Transients. AT2023rav notably shows a permanent decrease in quiescent flux post-transient, suggesting depletion of the accretion flow.
- AT2024gzn: A QSO flare showing late-time rebrightening, potentially indicative of a sudden increase in accretion or a complex TDE/ANT decay.
Supernova Candidates (4 objects):
- AT2024ljd: Confirmed as a Type Ia SN via spectroscopy.
- AT2019agc: Initially identified as a repeated transient, high-precision positional analysis revealed it consists of two independent supernovae occurring in the same galaxy but at distinct locations (~0.4" offset) separated by 2.5 years. This highlights limitations in current position-based transient identification schemes.
- AT2021lnu and AT2020mvg: Candidates for Superluminous Supernovae (SLSNe).

Significance and Implications
The authors claim that this sample demonstrates the value of archival searches and high-completeness detection strategies in uncovering outliers that challenge existing models. Key takeaways include:

Relevance of AGN Hosts: A significant fraction of the transients occurred in previously active galaxies. The paper argues that classifiers should not blindly exclude AGN hosts, as TDEs can occur in AGNs, and distinguishing between AGN flares and ANTs/ENTs requires nuanced analysis.
Repeated Events: The discovery of new repeated candidates (AT2023adr, ZTF23abjvojy) suggests that repeated nuclear transients are not exceptionally rare (approx. 10% of this sample). The authors caution against using simple two-peak intervals to predict third peaks, as the timing may vary or the events may be independent.
Model Constraints: The physical parameters derived for AT2020ukj and AT2023adr provide constraints on TDE models, particularly regarding partial disruptions and extreme viscosity/efficiency regimes.
Future Survey Design: The paper emphasizes that future surveys like LSST must balance purity with completeness. Overly strict classifiers risk missing rare, physically interesting outliers (e.g., long-duration events, faint TDEs, or events in AGNs). The authors advocate for systematic checks on positional consistency for repeated transients to avoid misidentifying independent events (like AT2019agc) as single rebrightening sources.

In conclusion, the study underscores that the "zoo" of optical nuclear transients is more complex than current classifications suggest, and that expanding the sample through archival mining is essential for refining the physical understanding of black hole accretion and stellar disruption.