AT 2020zso: Nuclear Tidal Disruption Event

• AT 2020zso is a nuclear tidal disruption event characterized by a rapidly formed, highly eccentric, nearly edge-on accretion disk and evolving UV/optical emission.
• Optical spectroscopy documents time-dependent broad line profiles (He II, Hα, Bowen) that unveil disk formation processes and reprocessing effects in a compact envelope.
• Radio monitoring identifies two distinct, non-relativistic outflows, linking episodic accretion states to multi-phase mass ejection in an AGN host.

AT 2020zso is a nuclear tidal disruption event (TDE) in the galaxy SDSS J222217.13-071558.9 at $z = 0.0563$, identified from its UV/optical blackbody evolution, broad He/H/Bowen emission-line phenomenology, and time-dependent spectral development. It is notable for two distinct observational results: optical spectroscopy indicates a newly formed, highly eccentric, nearly edge-on accretion disk, while late-time radio monitoring reveals at least two physically separate non-relativistic outflows. The event therefore connects prompt post-disruption disk formation, strong viewing-angle effects, and multi-episode outflow launching in a single system (Wevers et al., 2022, Christy et al., 17 Sep 2025).

1. Discovery, host environment, and TDE classification

AT 2020zso was discovered in 2020 by wide-field optical surveys as ZTF20acqoiyt, and was also seen by ATLAS and Gaia. Gaia astrometry places the transient at a nuclear offset of $46 \pm 60$ pc, consistent with the galaxy nucleus. The host is an elliptical galaxy, and late-time narrow-line diagnostics classify the nucleus as a Seyfert AGN. That AGN context is central to the interpretation: the transient occurred in a nucleus already showing narrow-line evidence for activity, yet its broad-line and continuum evolution were argued to be inconsistent with ordinary AGN variability (Wevers et al., 2022).

The TDE classification rests on the combination of a hot, blue continuum, evolving broad He/H/Bowen emission features, and UV/optical blackbody evolution unlike standard AGN fluctuations. The event is X-ray faint, with a combined Swift/XRT upper limit of $L_X \lesssim 3.8\times10^{41}\ {\rm erg\ s^{-1}}$ in 0.3–10 keV for a 75 eV blackbody assumption, or $\lesssim 4.3\times10^{41}\ {\rm erg\ s^{-1}}$ in 3–20 keV for an AGN-like power law. At the same time, it shows Bowen fluorescence lines, implying an ionizing EUV/soft X-ray source hidden from direct view. This observational combination places AT 2020zso naturally within the viewing-angle-dependent picture often invoked for optical/UV TDEs in which the intrinsic ionizing source is reprocessed and partially obscured (Wevers et al., 2022).

2. UV/optical light curve and blackbody evolution

The UV/optical light curves show three phases in the ZTF $g$ and $r$ bands: a very steep early rise, a break to a slower rise, and a turnover into decline. The early rise was fit with

$L = a + b \times (t-t_0)^{\alpha},$

yielding an initial behavior consistent with $L \propto t^2$, with $\alpha = 1.9 \pm 0.4$. After a break around phase $\sim -12$ d, the rise flattens to $46 \pm 60$0. The optical bands peak later than the UV bands, indicating strong color evolution (Wevers et al., 2022).

Blackbody fits to the UV+optical spectral energy distributions show that the source cooled by about $46 \pm 60$1 K during the first part of the light curve. Around peak, the characteristic color temperature is $46 \pm 60$2 K. The blackbody radius initially expands approximately linearly, corresponding to a photospheric expansion speed of $46 \pm 60$3, reaches a maximum near $46 \pm 60$4, plateaus near peak, and later recedes to $46 \pm 60$5. For the favored black hole mass range, these scales correspond to $46 \pm 60$6–$46 \pm 60$7 at peak and $46 \pm 60$8–$46 \pm 60$9 at late times. The peak bolometric UV/optical luminosity is $L_X \lesssim 3.8\times10^{41}\ {\rm erg\ s^{-1}}$0, approximately Eddington for a black hole of $L_X \lesssim 3.8\times10^{41}\ {\rm erg\ s^{-1}}$1, or about $L_X \lesssim 3.8\times10^{41}\ {\rm erg\ s^{-1}}$2 for $L_X \lesssim 3.8\times10^{41}\ {\rm erg\ s^{-1}}$3 (Wevers et al., 2022).

Several black hole mass estimators were considered. MOSFit gives $L_X \lesssim 3.8\times10^{41}\ {\rm erg\ s^{-1}}$4–6.1, with a representative value $L_X \lesssim 3.8\times10^{41}\ {\rm erg\ s^{-1}}$5, and prefers a low stellar mass with $L_X \lesssim 3.8\times10^{41}\ {\rm erg\ s^{-1}}$6, suggestive of a partial disruption, though with large systematic uncertainties. TDEMASS yields a mass consistent with $L_X \lesssim 3.8\times10^{41}\ {\rm erg\ s^{-1}}$7 and a stellar mass near solar. Independently, the host-galaxy stellar velocity dispersion is $L_X \lesssim 3.8\times10^{41}\ {\rm erg\ s^{-1}}$8, corresponding to $L_X \lesssim 3.8\times10^{41}\ {\rm erg\ s^{-1}}$9 using McConnell & Ma and $\lesssim 4.3\times10^{41}\ {\rm erg\ s^{-1}}$0 using Kormendy & Ho. A practical adopted range is therefore $\lesssim 4.3\times10^{41}\ {\rm erg\ s^{-1}}$1. From the radius extrapolation, the first observations were taken about 15 days after expansion began, implying a disruption date around MJD 59149 and a rise time of about 35 days to peak; MOSFit instead gives MJD $\lesssim 4.3\times10^{41}\ {\rm erg\ s^{-1}}$2 (Wevers et al., 2022).

3. Emission-line evolution and Bowen deblending

The optical spectra show transient broad emission associated with He II $\lesssim 4.3\times10^{41}\ {\rm erg\ s^{-1}}$3, H$\lesssim 4.3\times10^{41}\ {\rm erg\ s^{-1}}$4, He I $\lesssim 4.3\times10^{41}\ {\rm erg\ s^{-1}}$5, and Bowen N III at $\lesssim 4.3\times10^{41}\ {\rm erg\ s^{-1}}$6. Their evolution is highly structured. At the earliest epochs, $\lesssim 4.3\times10^{41}\ {\rm erg\ s^{-1}}$7 and $\lesssim 4.3\times10^{41}\ {\rm erg\ s^{-1}}$8 d, the broad features are better described by single broad Gaussians than by disk profiles, with line centroids consistent with rest velocity within uncertainties. In the $\lesssim 4.3\times10^{41}\ {\rm erg\ s^{-1}}$9 d X-shooter spectrum, the measured Gaussian FWHM values are $g$0 for H$g$1 and $g$2 for He II, with velocities near systemic: $g$3 for H$g$4 and $g$5 for He II. The He II/H$g$6 equivalent-width ratio is initially very high, dropping from about 11 at $g$7 d to 7 at $g$8 d (Wevers et al., 2022).

By $g$9 d the line profiles are clearly non-Gaussian and become double-peaked. The complex near 4700 Å is interpreted mainly as He II $r$0 with contamination from Bowen N III $r$1, while the 6563 Å feature is identified as H$r$2. The red peaks of He II and H$r$3 occur at similar velocities, around $r$4, and the red wings extend to about $r$5. By $r$6 d, H$r$7 develops a triple-peaked structure consisting of a pronounced red peak, a blue peak, and a central component near systemic velocity; the red peak reaches around $r$8. Similar structure appears in He I $r$9 and in the feature around 4100 Å, which is identified not as H$L = a + b \times (t-t_0)^{\alpha},$0 but as Bowen N III $L = a + b \times (t-t_0)^{\alpha},$1. The interpretation implies double-peaked Bowen lines, which were argued to be the first such profiles clearly seen in a TDE. Meanwhile, the He II/H$L = a + b \times (t-t_0)^{\alpha},$2 equivalent-width ratio falls rapidly and then stabilizes near $L = a + b \times (t-t_0)^{\alpha},$3 after about $L = a + b \times (t-t_0)^{\alpha},$4 d (Wevers et al., 2022).

A key methodological issue is deblending He II from N III. A stable multi-Gaussian decomposition of the 4600–4700 Å complex was not obtained. Instead, the independently visible N III $L = a + b \times (t-t_0)^{\alpha},$5 feature at $L = a + b \times (t-t_0)^{\alpha},$6 d was used, assuming that N III 4100 and 4640 have roughly a 1:1 flux ratio and share the same velocity structure. Subtracting the observed N III 4100 profile from the He II complex yields an inferred He II profile that is remarkably similar to H$L = a + b \times (t-t_0)^{\alpha},$7, supporting the conclusion that He II, H$L = a + b \times (t-t_0)^{\alpha},$8, and Bowen N III arise from the same disk-like region (Wevers et al., 2022).

4. Relativistic elliptical accretion-disk inference

The line profiles at $L = a + b \times (t-t_0)^{\alpha},$9 and $L \propto t^2$0 d were modeled with the relativistic elliptical disk prescription of Eracleous et al. (1995), applied to H$L \propto t^2$1 and the deblended He II line. The earliest $L \propto t^2$2 d spectrum was not fit with the disk model because the lines were interpreted as arising primarily in an outflowing envelope. The model uses seven main physical parameters: the emissivity power-law index $L \propto t^2$3, local line broadening $L \propto t^2$4, disk inclination $L \propto t^2$5, eccentricity $L \propto t^2$6, orientation angle $L \propto t^2$7, inner pericentre radius $L \propto t^2$8, and outer pericentre radius $L \propto t^2$9. Broad priors were adopted: $\alpha = 1.9 \pm 0.4$0–3, $\alpha = 1.9 \pm 0.4$1–4500 km s$\alpha = 1.9 \pm 0.4$2, $\alpha = 1.9 \pm 0.4$3–$\alpha = 1.9 \pm 0.4$4, $\alpha = 1.9 \pm 0.4$5–1, $\alpha = 1.9 \pm 0.4$6–$\alpha = 1.9 \pm 0.4$7, $\alpha = 1.9 \pm 0.4$8–550 $\alpha = 1.9 \pm 0.4$9, and $\sim -12$0–4750 $\sim -12$1. A weak extra Gaussian component was also included to represent a possible outflow/wind or central component (Wevers et al., 2022).

The best-fit solutions are mutually consistent in implying a highly inclined, highly eccentric, relatively compact accretion flow. For H$\sim -12$2 at $\sim -12$3 d, the fit gives $\sim -12$4, $\sim -12$5, $\sim -12$6, $\sim -12$7, $\sim -12$8, $\sim -12$9, and $46 \pm 60$00. For He II at $46 \pm 60$01 d, the fit gives $46 \pm 60$02, $46 \pm 60$03, $46 \pm 60$04, $46 \pm 60$05, $46 \pm 60$06, $46 \pm 60$07, and $46 \pm 60$08. For H$46 \pm 60$09 at $46 \pm 60$10 d, the fit gives $46 \pm 60$11, $46 \pm 60$12, $46 \pm 60$13, $46 \pm 60$14, $46 \pm 60$15, $46 \pm 60$16, and $46 \pm 60$17. For He II at $46 \pm 60$18 d, the fit gives $46 \pm 60$19, $46 \pm 60$20, $46 \pm 60$21, $46 \pm 60$22, $46 \pm 60$23, $46 \pm 60$24, and $46 \pm 60$25 (Wevers et al., 2022).

Averaging across lines and epochs yields $46 \pm 60$26, $46 \pm 60$27, $46 \pm 60$28–$46 \pm 60$29 with an average near $46 \pm 60$30, an inner radius of several hundred $46 \pm 60$31, an outer radius of several thousand $46 \pm 60$32, local broadening of $46 \pm 60$33–$46 \pm 60$34, and emissivity slope $46 \pm 60$35–3. The line asymmetry is essential to the eccentricity inference: in a standard relativistic circular disk, the blue peak is usually at least as strong as the red peak because of Doppler boosting, whereas in AT 2020zso the red peak is stronger. That strongly favors a non-axisymmetric eccentric disk over a circular one (Wevers et al., 2022).

5. Envelope, viewing angle, and spin constraints

The preferred physical picture is a two-component structure consisting of an optically thick outflowing/reprocessing envelope at early times and an accretion disk that becomes spectroscopically visible near peak. Before peak, the photosphere expands and the continuum plus lines are dominated by a broad, quasi-spherical reprocessing region. The very broad, nearly Gaussian He II seen at $46 \pm 60$36 to $46 \pm 60$37 d is attributed to this envelope rather than to the disk. The rapid drop in the He II/H$46 \pm 60$38 ratio is interpreted as an optical-depth effect: when the envelope is compact and dense, H$46 \pm 60$39 is more self-absorbed than He II, making the spectrum appear He-rich; as the envelope expands and cools, H$46 \pm 60$40 becomes less suppressed (Wevers et al., 2022).

Around peak, the envelope reaches $46 \pm 60$41, corresponding to $46 \pm 60$42–$46 \pm 60$43 for $46 \pm 60$44, and then becomes optically thin. At that point, disk emission dominates the broad-line profiles. The late-time blackbody radius of $46 \pm 60$45 overlaps the spectroscopically inferred disk scale, suggesting that the UV continuum may become directly disk-dominated at late epochs. The inferred eccentricity is also close to the characteristic value for returning tidal debris,

$46 \pm 60$46

which supports the interpretation that the debris has not yet circularized strongly. This suggests that an accretion disk can form within about a month after disruption while remaining highly eccentric (Wevers et al., 2022).

AT 2020zso also bears directly on TDE unification arguments. It is Bowen-strong and X-ray dim, yet line-profile modeling gives $46 \pm 60$47. The event was therefore presented as the first quantitative confirmation of a high-inclination geometry in a Bowen TDE, consistent with models in which viewing angle largely determines whether the EUV/soft X-ray source is directly observed. Alternative explanations were examined and disfavored. A binary SMBH interpretation was rejected because binary-driven eccentric AGN disk scenarios evolve on thousands-of-years timescales, whereas AT 2020zso shows major line-profile changes on weeks timescales; the narrow-line region also appears more consistent with rotation-dominated gas than with a disturbed dual-AGN system. Turn-on or changing-look AGN, bipolar outflows, and spiral-arm interpretations were also considered less satisfactory than a fresh TDE disk (Wevers et al., 2022).

A further inference concerns black hole spin. The modeled disk parameters, particularly inclination and orientation, do not change significantly over a baseline of about 15 days, so the empirical constraint $46 \pm 60$48 was adopted. The only explicit timescale formula quoted in this context is

$46 \pm 60$49

which gives roughly 5–15 days for AT 2020zso. Using published alignment and precession calculations, the absence of measurable precession-driven inclination change was taken to rule out high spins, with a quoted upper limit $46 \pm 60$50 for $46 \pm 60$51; in the abstract this is phrased as excluding high spin values for disk viscosity $46 \pm 60$52. This is an indirect, theory-informed constraint rather than a direct measurement of precession frequency (Wevers et al., 2022).

6. Long-lived radio emission and multiple outflows

Multi-year radio monitoring later showed that AT 2020zso also produced unusually rich long-lived radio emission. The campaign extended for more than three years, primarily with the VLA and supplemented by GMRT. VLA coverage spans $46 \pm 60$53 bands, roughly 1.3–17.4 GHz, while GMRT adds 0.65 and 1.26 GHz. The first VLA detection occurred at $46 \pm 60$54 d with $46 \pm 60$55. The radio luminosity peaks at $46 \pm 60$56 erg s$46 \pm 60$57, well below the $46 \pm 60$58 erg s$46 \pm 60$59 characteristic of on-axis relativistic TDE jets. High-frequency emission rises and fades earlier than low-frequency emission, as expected for an expanding self-absorbed synchrotron source whose peak frequency moves downward with time (Christy et al., 17 Sep 2025).

The radio light curves require two flares. The first rises soon after optical peak, reaches maximum around $46 \pm 60$60 yr, and then fades. A second flare begins around $46 \pm 60$61 d and becomes dominant after $46 \pm 60$62 d. Broken-power-law fits give rise and decay indices $46 \pm 60$63 and $46 \pm 60$64 for the first flare, and $46 \pm 60$65 and $46 \pm 60$66 for the second. The decisive evidence for two physical components is spectral: at $46 \pm 60$67 d and especially $46 \pm 60$68 d, the radio spectral energy distribution shows two distinct peaks. At $46 \pm 60$69 d, the first component has $46 \pm 60$70 and $46 \pm 60$71, while the second has $46 \pm 60$72 and $46 \pm 60$73. At $46 \pm 60$74 d, the first is constrained by $46 \pm 60$75 and $46 \pm 60$76, while the second has $46 \pm 60$77 and $46 \pm 60$78. The $46 \pm 60$79 d spectrum was stated to be incompatible with any single-region synchrotron-break ordering of the Granot & Sari type, so the emission was interpreted as the sum of two independent synchrotron sources rather than a refreshed shock or simple energy injection into one blast wave (Christy et al., 17 Sep 2025).

The radio analysis adopts a non-relativistic synchrotron-emitting outflow with

$46 \pm 60$80

and spectral ordering $46 \pm 60$81. MCMC fitting yields $46 \pm 60$82 and $46 \pm 60$83. Physical parameters are then inferred using the equipartition/minimum-energy formalism of Barniol Duran et al. for a non-relativistic spherical blast wave, with an emitting shell of thickness $46 \pm 60$84, filling factors $46 \pm 60$85 and $46 \pm 60$86, and microphysical assumptions $46 \pm 60$87 and $46 \pm 60$88 (Christy et al., 17 Sep 2025).

For the first flare, the inferred launch date is $46 \pm 60$89 relative to optical discovery, about 56 d after the bolometric peak. The outflow remains non-relativistic, with $46 \pm 60$90 and epoch-by-epoch $46 \pm 60$91 values around 0.10–0.12. The radius evolves from $46 \pm 60$92 at $46 \pm 60$93 d to $46 \pm 60$94 at $46 \pm 60$95 d, with $46 \pm 60$96 cm at $46 \pm 60$97 d. The internal energy rises and then plateaus around $46 \pm 60$98, more specifically $46 \pm 60$99–49.53 at late epochs. The magnetic field declines from $L_X \lesssim 3.8\times10^{41}\ {\rm erg\ s^{-1}}$00 to $L_X \lesssim 3.8\times10^{41}\ {\rm erg\ s^{-1}}$01, and the ambient density from $L_X \lesssim 3.8\times10^{41}\ {\rm erg\ s^{-1}}$02 to $L_X \lesssim 3.8\times10^{41}\ {\rm erg\ s^{-1}}$03. The kinetic mass estimate is $L_X \lesssim 3.8\times10^{41}\ {\rm erg\ s^{-1}}$04 (Christy et al., 17 Sep 2025).

For the second flare, the inferred launch date is $L_X \lesssim 3.8\times10^{41}\ {\rm erg\ s^{-1}}$05 relative to optical discovery. This second outflow is likewise non-relativistic and approximately freely coasting, with characteristic velocity $L_X \lesssim 3.8\times10^{41}\ {\rm erg\ s^{-1}}$06 and epoch-by-epoch $L_X \lesssim 3.8\times10^{41}\ {\rm erg\ s^{-1}}$07–0.23. Its radius grows from $L_X \lesssim 3.8\times10^{41}\ {\rm erg\ s^{-1}}$08 at $L_X \lesssim 3.8\times10^{41}\ {\rm erg\ s^{-1}}$09 d to $L_X \lesssim 3.8\times10^{41}\ {\rm erg\ s^{-1}}$10 at $L_X \lesssim 3.8\times10^{41}\ {\rm erg\ s^{-1}}$11 d, with $L_X \lesssim 3.8\times10^{41}\ {\rm erg\ s^{-1}}$12 cm at $L_X \lesssim 3.8\times10^{41}\ {\rm erg\ s^{-1}}$13 d. The energy rises to $L_X \lesssim 3.8\times10^{41}\ {\rm erg\ s^{-1}}$14, with $L_X \lesssim 3.8\times10^{41}\ {\rm erg\ s^{-1}}$15 increasing from 48.34 at $L_X \lesssim 3.8\times10^{41}\ {\rm erg\ s^{-1}}$16 d to 49.65 at $L_X \lesssim 3.8\times10^{41}\ {\rm erg\ s^{-1}}$17 d and $L_X \lesssim 3.8\times10^{41}\ {\rm erg\ s^{-1}}$18 at $L_X \lesssim 3.8\times10^{41}\ {\rm erg\ s^{-1}}$19 d. The magnetic field falls from $L_X \lesssim 3.8\times10^{41}\ {\rm erg\ s^{-1}}$20 to $L_X \lesssim 3.8\times10^{41}\ {\rm erg\ s^{-1}}$21, and the ambient density from $L_X \lesssim 3.8\times10^{41}\ {\rm erg\ s^{-1}}$22 to $L_X \lesssim 3.8\times10^{41}\ {\rm erg\ s^{-1}}$23. Its kinetic mass is also of order $L_X \lesssim 3.8\times10^{41}\ {\rm erg\ s^{-1}}$24 (Christy et al., 17 Sep 2025).

The external environment inferred from the radio data is consistent with a Bondi-like circumnuclear medium, $L_X \lesssim 3.8\times10^{41}\ {\rm erg\ s^{-1}}$25, with $L_X \lesssim 3.8\times10^{41}\ {\rm erg\ s^{-1}}$26 for the first outflow and $L_X \lesssim 3.8\times10^{41}\ {\rm erg\ s^{-1}}$27 for the second, close to the $L_X \lesssim 3.8\times10^{41}\ {\rm erg\ s^{-1}}$28 expectation for spherical Bondi accretion. Using $L_X \lesssim 3.8\times10^{41}\ {\rm erg\ s^{-1}}$29, $L_X \lesssim 3.8\times10^{41}\ {\rm erg\ s^{-1}}$30 K, $L_X \lesssim 3.8\times10^{41}\ {\rm erg\ s^{-1}}$31, and $L_X \lesssim 3.8\times10^{41}\ {\rm erg\ s^{-1}}$32, the Bondi radius is estimated as $L_X \lesssim 3.8\times10^{41}\ {\rm erg\ s^{-1}}$33. No density break near $L_X \lesssim 3.8\times10^{41}\ {\rm erg\ s^{-1}}$34 is required to explain the second flare, although magnetic-field inhomogeneities were suggested by an unusually shallow optically thick slope $L_X \lesssim 3.8\times10^{41}\ {\rm erg\ s^{-1}}$35 at $L_X \lesssim 3.8\times10^{41}\ {\rm erg\ s^{-1}}$36 d, compared with the canonical self-absorbed synchrotron value $L_X \lesssim 3.8\times10^{41}\ {\rm erg\ s^{-1}}$37 (Christy et al., 17 Sep 2025).

The preferred physical interpretation links the two radio outflows to two accretion states. The first is favored to be accretion-driven, launched while the source was still accreting at a relatively high Eddington fraction; the optical luminosity implies $L_X \lesssim 3.8\times10^{41}\ {\rm erg\ s^{-1}}$38, consistent with a slim disk capable of driving winds. The second is associated with a later transition to an advection-dominated accretion flow, after an intermediate regime in which strong outflow production is not expected below $L_X \lesssim 3.8\times10^{41}\ {\rm erg\ s^{-1}}$39, and then may resume when $L_X \lesssim 3.8\times10^{41}\ {\rm erg\ s^{-1}}$40. Off-axis relativistic jet and delayed-accretion explanations were considered less compelling, and “energy injection into Outflow 1” was explicitly rejected because two simultaneous spectral peaks isolate two emitting regions (Christy et al., 17 Sep 2025).

7. Uncertainties, alternatives, and significance

Several limitations qualify the interpretation. On the optical side, the MOSFit light-curve fit is formally poor because the event shows rapid temperature evolution and a broken rise that the model does not capture well, so disruption parameters and stellar-mass estimates carry substantial systematics. The He II profile is blended with Bowen N III, and the deblending assumes a 1:1 flux ratio between N III 4100 and 4640 and identical velocity structure, which likely introduces additional systematics; for this reason the H$L_X \lesssim 3.8\times10^{41}\ {\rm erg\ s^{-1}}$41 fits were treated as primary. The disk model is idealized, not a full radiative-transfer or hydrodynamical calculation, and some parameters—especially the inner and outer radii—remain degenerate across lines and epochs. The spin constraint is indirect, relying on comparison to published precession and alignment theory rather than on a direct measurement of precession (Wevers et al., 2022).

The radio interpretation likewise depends on modeling assumptions. The inferred $L_X \lesssim 3.8\times10^{41}\ {\rm erg\ s^{-1}}$42, $L_X \lesssim 3.8\times10^{41}\ {\rm erg\ s^{-1}}$43, $L_X \lesssim 3.8\times10^{41}\ {\rm erg\ s^{-1}}$44, and $L_X \lesssim 3.8\times10^{41}\ {\rm erg\ s^{-1}}$45 values rely on equipartition and quasi-spherical shell geometry, and the launch dates depend on the assumption of free expansion; if the outflows decelerate, the inferred $L_X \lesssim 3.8\times10^{41}\ {\rm erg\ s^{-1}}$46 values would shift. The first outflow’s launch time disfavors unbound debris and pure collisionally induced outflow scenarios, but does not absolutely exclude some contribution from such processes. The similar density profiles inferred for the two outflows are somewhat surprising if both were quasi-spherical, because the second would be expected to encounter gas modified by the first; this suggests, plausibly, that the two ejections may have different geometries, such as one more polar and one more toroidal. A repeated partial-TDE interpretation was also briefly considered, since optical modeling suggested a partial disruption and the separation between the two radio outflows is comparable to recurrence times in repeating partial TDE candidates, but the lack of an observed second optical flare—during a sun-constrained interval—left that possibility unresolved (Christy et al., 17 Sep 2025).

Within these caveats, AT 2020zso is significant for several reasons. Optical spectroscopy provides unusually direct evidence for a compact, edge-on, highly eccentric TDE accretion flow, closer to the expected geometry of newly returned debris than the near-circular disks inferred in some earlier events. The event also supports the unification picture in which Bowen-strong, X-ray-faint TDEs are preferentially seen at high inclination. Finally, the radio data show that a thermal TDE can launch at least two distinct non-relativistic outflows over a multi-year interval, with the separation of components demonstrated spectrally rather than inferred solely from late-time rebrightening. AT 2020zso therefore serves as an important observational link between TDE disk formation, incomplete circularization, anisotropic reprocessing, and episodic accretion-driven outflows in an AGN host (Wevers et al., 2022, Christy et al., 17 Sep 2025).