AT2018hyz: TDE with Delayed Radio Flare

• AT2018hyz is a tidal disruption event (TDE) marked by an initial optical/UV peak followed by an unusually delayed, bright radio flare that challenges standard models.
• Its multi-wavelength observations reveal evolving emission-line profiles and a late-time onset of mildly relativistic outflow, supporting interpretations like delayed jet launch or outflow-cloud interaction.
• Detailed spectroscopic and radio analyses of AT2018hyz provide critical insights into black hole mass estimates, accretion physics, and jet versus non-jet outflow scenarios in TDEs.

AT2018hyz, also designated ASASSN-18zj, is a tidal disruption event (TDE) in the nucleus of a quiescent E+A galaxy at redshift $z=0.04573$. It was initially characterized as a luminous optical/UV TDE with a blue continuum, broad hydrogen and helium emission lines, and an early bolometric evolution broadly consistent with the canonical optical-TDE phenomenology. Its later development made it unusual: after multiple radio non-detections extending to $\sim 700$ days, it produced a rapidly rising radio flare first detected at $\sim 972$ days, with subsequent monitoring showing continued brightening to luminosities approaching the jetted-TDE regime. AT2018hyz has therefore become a central case in debates over delayed outflows, off-axis relativistic jets, jet breakout after alignment, and late interaction with circumnuclear structure (Gomez et al., 2020, Short et al., 2020, Cendes et al., 2022, Cendes et al., 11 Jul 2025).

1. Discovery, host galaxy, and basic parameters

AT2018hyz was discovered by ASAS-SN on 2018 Oct 14. It is spatially consistent with the host-galaxy nucleus, with an offset of $0.2\pm0.8''$, and the host is identified as a quiescent Balmer-strong or post-starburst E+A galaxy. The transient was reported at an apparent magnitude near $g=16.4$ mag in one spectroscopic study, while the photometric study gives an ASAS-SN detection at $g = 17.08 \pm 0.22$. Using the host distance implied by $z=0.04573$, the event had an absolute optical magnitude of about $-20.2$ mag at discovery, and the luminosity distance is given as about $205$ Mpc (Gomez et al., 2020, Short et al., 2020).

The host-galaxy environment is notable because TDEs are overrepresented in Balmer-strong galaxies. Archival spectroscopy shows negligible current star formation, and ALMA non-detections imply $$\mathrm{SFR}<8.6\times10^{-6}\,M_\odot\,\mathrm{yr^{-1}}$$. This host context reduced ambiguity with core-collapse transients and supported the TDE interpretation from early on (Gomez et al., 2020).

The central black-hole mass is not uniquely determined. From stellar-velocity-dispersion measurements, one spectroscopic analysis adopted $\sim 700$0 with factor-of-$\sim 700$1 uncertainty, while MOSFiT light-curve modeling gave $\sim 700$2, corresponding to $\sim 700$3. This mass uncertainty propagates into inferences about emitting radii, pericenter scale, and disruption depth (Gomez et al., 2020, Short et al., 2020).

2. Optical, UV, and X-ray evolution

In bolometric terms, AT2018hyz was comparable to other optical TDEs. The peak bolometric luminosity is given as $\sim 700$4, peaking at MJD 58429, and the early decline is described as consistent with $\sim 700$5 for roughly the first $\sim 700$6 days after bolometric peak. Integrating the observed light curve from phase $\sim 700$7 to $\sim 700$8 days gives $\sim 700$9 erg, and the total radiated energy including the model-estimated pre-peak contribution is $\sim 972$0 erg (Gomez et al., 2020).

The thermal continuum evolved non-monotonically. Blackbody fits gave a temperature near $\sim 972$1 K around peak, cooling to $\sim 972$2 K by $\sim 972$3 days and reheating to $\sim 972$4 K by $\sim 972$5 days. The photospheric radius was about $\sim 972$6 cm near peak and later contracted to roughly $\sim 972$7 cm. A UV-bright bump appeared around $\sim 972$8 days after bolometric peak, lasted for at least $\sim 972$9 days, and was followed by flattening beyond roughly $0.2\pm0.8''$0 days to at least $0.2\pm0.8''$1 days. The late UV luminosity above $0.2\pm0.8''$2 nm is quoted as $0.2\pm0.8''$3, consistent with persistent accretion-disk UV emission (Gomez et al., 2020).

X-ray behavior remained ambiguous. Swift/XRT detected a source out to $0.2\pm0.8''$4 days with no significant spectral evolution and with luminosity consistent with a roughly constant level rather than a decline. The preferred spectral model for the combined $0.2\pm0.8''$5-10 keV data at $0.2\pm0.8''$6 d was an absorbed blackbody plus power law, with blackbody temperature $0.2\pm0.8''$7 keV, photon index $0.2\pm0.8''$8, total unabsorbed flux $0.2\pm0.8''$9, and luminosity of roughly $g=16.4$0. Because this luminosity and the [O III]/X-ray ratio are consistent with a weak AGN, the X-ray source was not uniquely assigned to the TDE. At the same time, the radio non-detection at early times favored inverse-Compton scattering rather than a jet for the non-thermal X-ray component (Gomez et al., 2020).

3. Spectroscopic phenomenology and line-formation physics

The optical spectra showed a blue continuum and broad Balmer emission with He I $g=16.4$1 at early times, consistent with the H-rich part of the optical-TDE population. High-cadence spectroscopy then revealed unusually strong profile evolution. The Balmer lines changed from smooth broad profiles to a boxy, asymmetric, double-peaked or “double-horned” phase around $g=16.4$2–80 days after discovery, and later returned to smoother profiles by roughly $g=16.4$3 days after discovery. He II $g=16.4$4 was absent from the earliest spectra and appeared only after roughly $g=16.4$5–$g=16.4$6 days after peak (Short et al., 2020).

The kinematics were broad. The H$g=16.4$7 FWHM narrowed from about $g=16.4$8 at early times to about $g=16.4$9 later in the main sequence. Once He II was clearly visible and fitted with a single Gaussian, it had $g = 17.08 \pm 0.22$0. If the late $g = 17.08 \pm 0.22$1 day feature is decomposed into N III and He II, the fitted widths become $g = 17.08 \pm 0.22$2 for N III and $g = 17.08 \pm 0.22$3 for He II, implying a more extended He II-emitting region under a virial interpretation (Short et al., 2020).

A central spectroscopic result was the very flat Balmer decrement, with $g = 17.08 \pm 0.22$4, remaining roughly constant through the monitoring interval and consistently below $g = 17.08 \pm 0.22$5. This is unlike a typical AGN broad-line region and inconsistent with standard Case B expectations. The preferred interpretation was that the Balmer lines are collisionally excited rather than primarily photoionized, with the line-emitting gas originating in a disc chromosphere analogous to those seen in cataclysmic variables. In that picture, the continuum arises in a hotter disc or reprocessing layer, while the Balmer lines form in a cooler chromospheric structure (Short et al., 2020).

The double-peaked Balmer profiles were successfully fit with standard relativistic disc-line prescriptions. For H$g = 17.08 \pm 0.22$6 at $g = 17.08 \pm 0.22$7 days, one fit gave $g = 17.08 \pm 0.22$8, $g = 17.08 \pm 0.22$9, $z=0.04573$0, and $z=0.04573$1, with $z=0.04573$2 fixed at $z=0.04573$3. This supported the view that accretion-disc emission was directly visible for part of the event. However, this was not treated as a unique solution. Outflow-based interpretations, or a disc plus Gaussian or wind component, were also discussed, and the later smoothing of the lines was interpreted as possible obscuration growth and/or increasing wind dominance rather than disappearance of the disc itself (Short et al., 2020).

4. Radio emergence and the late-time flare

The radio behavior is the defining late-time property of AT2018hyz. Early searches produced non-detections extending from $z=0.04573$4 days to about $z=0.04573$5 days after optical discovery, including a final non-detection at $z=0.04573$6 days. The source was then detected in the radio only at $z=0.04573$7 days, already implying a dramatic delayed turn-on (Cendes et al., 11 Jul 2025).

The first detailed late-time radio study presented detections at $z=0.04573$8–$z=0.04573$9 days across $-20.2$0–$-20.2$1 GHz, together with optical/UV and X-ray measurements. Relative to the time of optical discovery, the radio rise was steeper than $-20.2$2. That study argued that such a steep rise could not be explained in any reasonable scenario of an outflow launched at the time of disruption, including an off-axis jet or a sudden increase in ambient density, and therefore pointed to a delayed launch. Using multi-frequency data, it directly inferred a radio-emitting outflow launched $-20.2$3 days after optical discovery, with mildly relativistic velocity: $-20.2$4 for a spherical geometry and $-20.2$5 for a $-20.2$6 jet geometry. The corresponding minimum kinetic energies were $-20.2$7 erg and $-20.2$8 erg, respectively. This was described as the first definitive evidence for a delayed mildly relativistic outflow in a TDE (Cendes et al., 2022).

Continued monitoring later extended the radio/mm baseline to $-20.2$9–$205$0 days and $205$1–$205$2 GHz. Over this interval the source continued to brighten at essentially all observed frequencies. At C band, for example, the flux density increased from about $205$3 mJy at $205$4 days to $205$5 mJy at $205$6 days. The rise was not described by a single temporal index: between about $205$7 and $205$8 days it followed $205$9, while later it followed $$\mathrm{SFR}<8.6\times10^{-6}\,M_\odot\,\mathrm{yr^{-1}}$$0, summarized as about $$\mathrm{SFR}<8.6\times10^{-6}\,M_\odot\,\mathrm{yr^{-1}}$$1 during the later monitoring. By $$\mathrm{SFR}<8.6\times10^{-6}\,M_\odot\,\mathrm{yr^{-1}}$$2 days the radio luminosity had reached $$\mathrm{SFR}<8.6\times10^{-6}\,M_\odot\,\mathrm{yr^{-1}}$$3, only about a factor of $$\mathrm{SFR}<8.6\times10^{-6}\,M_\odot\,\mathrm{yr^{-1}}$$4 below Swift J1644+57 at similar timescales (Cendes et al., 11 Jul 2025).

The broadband spectral evolution was also unusual. Over a $$\mathrm{SFR}<8.6\times10^{-6}\,M_\odot\,\mathrm{yr^{-1}}$$5-day span, the spectral peak remained near a nearly constant frequency while the peak flux density increased by about an order of magnitude. The mean fitted values were $$\mathrm{SFR}<8.6\times10^{-6}\,M_\odot\,\mathrm{yr^{-1}}$$6 GHz and $$\mathrm{SFR}<8.6\times10^{-6}\,M_\odot\,\mathrm{yr^{-1}}$$7, while the peak flux density rose from $$\mathrm{SFR}<8.6\times10^{-6}\,M_\odot\,\mathrm{yr^{-1}}$$8 mJy at $$\mathrm{SFR}<8.6\times10^{-6}\,M_\odot\,\mathrm{yr^{-1}}$$9 days to $\sim 700$00 mJy at $\sim 700$01 days. A later spherical equipartition analysis, incorporating a cooling break near $\sim 700$02 GHz around $\sim 700$03 days and adopting $\sim 700$04, inferred $\sim 700$05–$\sim 700$06 cm, a launch date $\sim 700$07 days after disruption, $\sim 700$08, and late-time kinetic energy of order $\sim 700$09 erg (Cendes et al., 11 Jul 2025).

5. Competing physical interpretations and decisive tests

The late-time flare has generated several physically distinct interpretations. A key point in the subsequent literature is that unresolved radio light curves and spectra alone do not cleanly distinguish them (Matsumoto, 25 Dec 2025).

Scenario	Representative AT2018hyz parameters	Distinguishing implication
Delayed spherical outflow	$\sim 700$10 d, $\sim 700$11, $\sim 700$12 erg (Cendes et al., 11 Jul 2025)	Centroid motion remains non-relativistic (Matsumoto, 25 Dec 2025)
Off-axis relativistic jet	$\sim 700$13-$\sim 700$14, $\sim 700$15, $\sim 700$16 erg (Cendes et al., 11 Jul 2025)	Apparent superluminal centroid motion would be a smoking-gun signature (Matsumoto, 25 Dec 2025)
Outflow-cloud interaction	$\sim 700$17, $\sim 700$18, $\sim 700$19 d, $\sim 700$20 pc, $\sim 700$21 pc (Zhuang et al., 2024)	Sharp flare from collision with a dense cloud; diffuse CNM must satisfy $\sim 700$22 (Zhuang et al., 2024)
Delayed escaping jet after alignment	$\sim 700$23 d, $\sim 700$24 erg, $\sim 700$25 (Teboul et al., 2023)	Radio zero-point is the late breakout or alignment time, not the disruption time (Teboul et al., 2023)

In the VLBI-diagnostics framework, the most robust discriminator is emission-centroid motion. In the delayed-outflow interpretation, the centroid remains stationary or moves only by a clearly non-relativistic amount. In the off-axis jet interpretation, the centroid can show apparent superluminal motion, and that is identified as the decisive observational signature. Morphology alone is less secure, because at early times an off-axis jet image can still look deceptively disk-like (Matsumoto, 25 Dec 2025).

A distinct alternative replaces both delayed-launch and off-axis-jet explanations with interaction between an early-launched outflow and a dense circumnuclear cloud. In that model, the long delay reflects the travel time to $\sim 700$26 pc, the steep rise reflects the shell-width crossing time of the cloud, and the early radio upper limit at $\sim 700$27 days requires a very dilute diffuse medium, parameterized as $\sim 700$28. The same framework was argued to reproduce the spectral evolution and to be consistent with a later X-ray luminosity of $\sim 700$29 at $\sim 700$30 days (Zhuang et al., 2024).

Another interpretation embeds AT2018hyz in a broader theory of misaligned TDE jets. In that picture, a weak jet is launched early but cannot escape while precessing through wind ejecta; only after slow hydrodynamic alignment with the SMBH spin does breakout occur, yielding a delayed mildly relativistic radio flare. For AT2018hyz this model used $\sim 700$31 days and explicitly identified the source with a “delayed escaping mildly relativistic jet” class (Teboul et al., 2023).

6. Position within the TDE population and broader implications

AT2018hyz occupies an intermediate position between past non-relativistic radio TDEs such as ASASSN-14li and AT2019dsg, and the relativistic TDE Swift J1644+57. In the original delayed-outflow interpretation, its energy and velocity were already described as intermediate between those classes, with the suggestion that such delayed outflows may be common in TDEs (Cendes et al., 2022). Later monitoring reinforced that it does not fit comfortably into a simple division between “ordinary thermal TDEs with slow outflows” and “rare relativistic jetted TDEs,” because by $\sim 700$32 days its radio luminosity had reached $\sim 700$33, encroaching directly on the jetted-TDE regime (Cendes et al., 11 Jul 2025).

AT2018hyz has also been used as an off-axis-jet benchmark in ultra-high-energy cosmic-ray work. In that context it is treated as a normal optically discovered TDE at $\sim 700$34 that later revealed a powerful jet through radio brightening. One such analysis adopted a jet opening angle of about $\sim 700$35, viewing angle of about $\sim 700$36, kinetic energy $\sim 700$37, and a best-fit collimation-corrected kinetic energy $\sim 700$38. On that basis the event was argued to satisfy the Hillas condition for UHECR acceleration, and the inferred AT2018hyz-like jetted-TDE rate was used to derive an energy injection rate $\sim 700$39, exceeding the quoted observed UHECR luminosity density of $\sim 700$40. This interpretation is explicitly model-dependent, since later radio work also admits nearly edge-on and non-jet solutions (Piran et al., 2023, Cendes et al., 11 Jul 2025).

A persistent misconception is that the late radio flare already has a unique explanation. The literature summarized here does not support that conclusion. Delayed outflow, highly off-axis jet, outflow-cloud interaction, and delayed jet escape after alignment all remain viable in published analyses. The consistent point across these studies is methodological rather than interpretive: continued broadband radio monitoring, measurement of the eventual turnover, and VLBI astrometry or imaging are the decisive next observations. In that sense, AT2018hyz has become a reference event for the late-time diversity of TDE outflows and for the observational problem of separating true delayed launching from delayed visibility (Matsumoto, 25 Dec 2025, Cendes et al., 11 Jul 2025).