{"id":"3da94bee-25cc-484e-bc3c-e8c647492d05","arxiv_id":"2507.15482","paper_version":1,"verdict":"CONDITIONAL","confidence":"MODERATE","novelty_score":6.0,"correctness_risk":"medium","formal_verification":"none","parameter_count":5,"one_line_summary":"X-ray observations of the tidal disruption event AT2020afhd reveal an ultrafast outflow that appears after day 74, slows from 0.19c to 0.01c, and disappears after day 215.","lead":"This paper reports X-ray absorption features in the tidal disruption event AT2020afhd that appear only between days 172 and 215, then vanish, with the outflow slowing from about 0.19c to 0.01c in roughly ten days. If correct, it is the first time a tidal disruption event's X-ray outflow has been seen to turn on, strengthen, and turn off.","discovery_kind":"extension","skeptic_critique":{"model":"deepseek-v4-flash","headline":"Epoch 1 and 4 non-detections only constrain fixed line energies; without an energy/velocity grid search, the claimed turn-on/turn-off and 'full evolutionary sequence' are not established.","rationale":"The paper's central claim is that AT2020afhd shows the first full evolutionary sequence of a TDE X-ray outflow, including delayed appearance and disappearance. For that to hold, the non-detections in epochs 1 and 4 must be physical. The paper's own procedure for setting upper limits (Tab. B.2) fixes the line centroid, so the non-detections are only valid at the assumed energies; the appendix's residual search covers a limited band. Because the line energy demonstrably evolves within epoch 3, this is a real blind spot. If a grid search finds absorption at an unexamined energy in epoch 1 or 4, the central timeline collapses to 'a UFO was seen for ~40 days,' which is not the claimed evolutionary sequence. The seasonal gap between days 74 and 172 was acknowledged by the authors and further weakens the 'delayed launch' phrasing, but even with the gap, a delayed appearance relative to the continuum could hold if the epoch 1 non-detection is robust; hence the grid search is the decisive test. The velocity-drop and anti-correlation claims involve an additional pion-model degeneracy and large uncertainties (Tab. B.1), but those are secondary to the presence/absence timeline that defines the 'full evolutionary sequence.' The reader's CONDITIONAL verdict already captures the need for this test, so no change to the verdict is warranted.","tokens_in":17857,"tokens_out":12347,"duration_ms":136802,"concrete_test":"Perform a blind grid search on the co-added Swift/XRT spectra of epochs 1 and 4: model the continuum as in Tab. B.2, then add a Gaussian absorption component with centroid energy scanned from 0.5 to 1.2 keV in steps of 0.02 keV, leaving the normalization free (and, in a second pass, also the width free), and record the maximum DeltaCstat and the 3sigma upper-limit surface. Repeat with a pion component allowing vout in [0, 0.3c] and log xi in [1, 4]. If any grid point in either epoch yields DeltaCstat > 9 relative to the continuum-only fit, then the non-detection is not physical and the claimed turn-on/turn-off sequence is not established.","verdict_should_be":"UNCHANGED","load_bearing_attack":"The full evolutionary-sequence claim ('appeared no earlier than day 74... disappeared after day 215'; Abstract, Sec. 4.1) rests on the non-detections in Swift epochs 1 and 4. Tab. B.2 shows these upper limits were derived by fixing Eline and Esigma to values taken from the detection epochs (0.93 keV for epoch 1, 0.78 keV for epoch 4). The upper limits therefore only test absorption at those two specific energies. The Appendix A residual search, used to argue that the non-detections are not low-SNR artifacts, is restricted to the 0.8-1.1 keV band and does not cover energies below 0.8 keV where a faster wind (v >~ 0.15c) would appear. Within epoch 3 itself the centroid varies by ~0.17 keV over ~10 days (Fig. 2), so there is no reason the early or late wind should be at the assumed energy. Moreover, the epoch 4 Swift 3sigma upper limit (Enorm < 4.2e-5) is only slightly below the epoch 3 detection (5.0 +/- 1.2e-5), and because the epoch 4 continuum flux is ~4 times lower, an absorber with the same physical column as in epoch 3 would produce a line ~4 times weaker and is therefore not excluded. The seasonal gap (74-172 d) further leaves the onset unconstrained. The delayed-launch and disappearance conclusions are thus not yet established: a wind present in epochs 1 or 4 at a different velocity, ionization, or covering fraction would be missed.","agreement_with_reader":"agree"},"referee_report":{"model":"deepseek-v4-flash","summary":"The paper reports the discovery of blueshifted X-ray absorption features in the tidal disruption event AT2020afhd, detected in NICER and XMM-Newton observations taken between 194 and 212 days after discovery. The features are statistically significant in six individual observations (ΔCstat = 24–66 for 3 dof). Using photoionization modeling with the pion model, the authors derive outflow velocities apparently declining from ~0.19c to ~0.0097c and an anti-correlation between outflow velocity and ionization parameter. Because the features are absent in combined Swift spectra before day 74 and after day 215, the authors claim to have observed the full evolutionary sequence of an ultrafast outflow in a TDE for the first time, with delayed launch, strengthening, and disappearance. They argue that the behavior is inconsistent with radiation-pressure-driven winds and favor magnetic driving.","tokens_in":18181,"tokens_out":6995,"duration_ms":70334,"significance":"If the full evolutionary sequence is real, this would be an important result: it would demonstrate that TDE disk winds can turn on and off on ~100-day timescales, decelerate dramatically, and challenge radiation-pressure wind models. The core detections themselves are robust and well documented, with multiple instruments and consistent continuum treatment. The statistical significance of the six detections is high and the spectral fitting tables are detailed. However, the evolutionary-sequence claim rests on non-detections that are not yet demonstrated to be physical, and the velocity/ionization trend is not shown to be free of parameter degeneracy. The paper's physical conclusions therefore need substantial additional analysis before they can be accepted.","major_comments":[{"comment":"The conclusion that the outflow appeared only after day 74 and disappeared after day 215 is based on non-detections in the Swift/XRT epoch spectra. These upper limits were derived by fixing the Gaussian line energy and width to the values from the detection epochs (Eline = 0.93 keV for epoch 1 and 0.78 keV for epoch 4). This tests only absorption at those specific energies, yet Fig. 2 shows that the line centroid varies by ~0.17 keV across epoch 3, so a wind with a different velocity would be missed. The Appendix A residual search is restricted to 0.8–1.1 keV and therefore does not constrain the range below 0.8 keV where a faster wind (v ≳ 0.15c) would appear. In addition, the epoch 4 upper limit (Enorm < 4.2×10^-5) is comparable to the epoch 3 detection (5.0 ± 1.2×10^-5), and because the epoch 4 continuum flux is ~4 times lower, an absorber with the same physical column as in epoch 3 would produce a line ~4 times weaker and would not be excluded. The 'turn-on' and 'turn-off' are thus not established; a grid search over line energy/velocity and column should be performed on the epoch 1 and 4 spectra before claiming a full evolutionary sequence.","section":"Sec. 3 and Table B.2"},{"comment":"The claimed 'dramatic deceleration from ~0.19c to ~0.0097c over approximately 10 days' is not a monotonic trend in the fitted values. The two high velocities (-0.186 and -0.190 c) occur only on days 198.0 and 199.1; the day 194.3 observation gives vout = -0.015 (+0.012, -0.119) and the day 201.1 observation gives vout = -0.036 (+0.013, -0.304), both consistent with the low-velocity end. The apparent deceleration is therefore driven by only two epochs and is not a smooth 10-day decline. The authors should present the full confidence contours for vout at each epoch and discuss the non-monotonic behavior, or temper the deceleration claim accordingly.","section":"Sec. 3, Table B.1 and Fig. 4"},{"comment":"The inverse correlation between outflow velocity and ionization parameter is central to the argument against radiation-pressure-driven winds. However, vout and log ξ are both fitted parameters of the same pion model, and the broad absorption feature is a blend of the Fe M-shell UTA and O K-edge; such models often have strong degeneracies between velocity, column, and ionization. The paper does not show confidence contours or a grid search over these parameters. Given the large error bars (e.g., the first point in Fig. 5a has vout consistent with zero), the anti-correlation may not be robust. The authors should demonstrate that the anti-correlation persists when the degeneracy is marginalized, for example by producing ΔCstat contours in the (vout, log ξ) plane for representative epochs.","section":"Sec. 4.2, Fig. 5"},{"comment":"The statement that AT2020afhd is 'the first instance in which the full evolutionary process of outflows has been observed in a TDE' is not yet supported. The seasonal gap between days 74 and 172 leaves the launch time unconstrained, as acknowledged in Sec. 4, and the non-detections in epochs 1 and 4 are based on fixed line energies, as discussed above. Please revise the abstract and conclusions to state the constraints more accurately, e.g., 'absorption is detected only between days 172 and 212; non-detections before day 74 and after day 215 are currently consistent with sensitivity or visibility effects.'","section":"Sec. 4.1 and Abstract"}],"minor_comments":[{"comment":"The title has a stray space in 'A T2020afhd'; it should read 'AT2020afhd'.","section":"Title"},{"comment":"The word 'Morever' is a typo and should be 'Moreover'.","section":"Sec. 5"},{"comment":"The paper repeatedly refers to 'Wang et al. (2025, submitted)' for classification, black hole mass, radio emission, and the X-ray light curve; this unpublished work is not in the reference list, making it impossible for the reader to verify these inputs. Please provide a preprint identifier or include the relevant data in the paper.","section":"References and Sec. 1"},{"comment":"The dashed line marking 0.95 keV is used for all four epochs, but the absorption feature in epoch 3 evolves to 0.78 keV in later observations (Table B.1); the figure would be clearer if it indicated the energy range or used the epoch-specific centroid.","section":"Fig. 3"},{"comment":"The Gaussian line width Eσ is fixed at different values for different epochs (0.05 keV for epochs 2 and 3, 0.12 keV for epoch 4 and the XMM spectra); the justification for this choice should be stated explicitly, as the resulting upper limits are sensitive to the assumed width.","section":"Sec. 2 and Table B.2"}],"recommendation":"major_revision","confidential_remarks":"The manuscript depends heavily on a companion paper (Wang et al., submitted) for the classification, black hole mass, radio data, and even the X-ray light curve. While not disqualifying, the editor should be aware that those inputs are not independently verified in this manuscript. The central interpretive claims (delayed launch, anti-correlation, deceleration) rest on non-detections and parameter degeneracies that will require substantial additional analysis. I recommend major revision rather than acceptance."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"The paper has a real detection: a statistically significant, variable absorption feature in six X-ray observations of AT2020afhd, with Delta C 24-66 for 3 dof. That is worth taking seriously. What is new is the claim that this is the first full evolutionary sequence of a TDE UFO, with delayed appearance, strengthening, and disappearance, plus a rapid deceleration from 0.19c to 0.0097c. The data set is well-sampled, using NICER, Swift, and XMM, and the reduction is standard. The mass-loss estimate is honest and useful as a lower limit for the missing-energy discussion.\n\nBut the headline claims are stronger than the evidence. The non-detections in epochs 1 and 4 are upper limits at fixed line energies (0.93 keV and 0.78 keV), so they only rule out absorption at those specific energies. A wind at a different velocity or ionization state could be missed. The Appendix A residual search covers only 0.8-1.1 keV, so a line outside that band would not be caught. The authors acknowledge the seasonal gap between days 74 and 172, which alone leaves the onset time unconstrained. So the 'delayed launch' and 'disappearance' are consistent with the data but not established by it.\n\nThe velocity evolution also rests on a weak point. The first NICER observation at day 194.3 has vout/c = -0.015 with asymmetric errors spanning -0.13 to -0.003; the next two points also have very large uncertainties. The dramatic deceleration is driven by one or two poorly constrained measurements, not by a clean sequence. The paper does not show confidence contours for vout and log xi, so the claimed inverse correlation could be partly a degeneracy of the pion model on the blended Fe UTA and O K-edge.\n\nThe abstract and conclusions overstate what the data demonstrate. A revision that includes an energy/velocity grid search for the non-detection epochs, confidence contours for the key pion parameters, and more cautious language about the evolutionary sequence would make the paper solid. As it stands, treat the strong claims as tentative.\n\nThis paper deserves a serious referee: the detection is credible, the variability is genuinely interesting, and the small sample of TDE UFOs needs more cases. I would bring it to a reading group working on TDEs or outflows, with a warning about the interpretation.","headline":"Credible detection of a variable absorption feature in a TDE, but the 'full evolutionary sequence' and rapid deceleration claims go beyond what the non-detection constraints and velocity uncertainties support.","tokens_in":18773,"tokens_out":7013,"would_cite":true,"duration_ms":77517,"reading_group":"maybe","serious_thinker":"yes","would_accept_peer_review":true},"rs_alignment":null,"lean_confirmation":null,"pith_extraction":{"msc":[],"pacs":[],"model":"deepseek-v4-flash","headline":"AT2020afhd provides the first observed full life cycle of an X-ray ultrafast outflow in a tidal disruption event.","keywords":["AT2020afhd","tidal disruption events","ultrafast outflows","X-ray absorption spectroscopy","photoionization modeling","disk winds","magnetic driving","super-Eddington accretion"],"falsifier":"A deeper co-added spectrum from days 0-74 and days 220-302, with enough counts to detect the same iron and oxygen absorption at the epoch-(3) amount of absorbing gas and ionization state, would settle whether the turn-on and turn-off are physical; if the line still cannot be found, the delayed-launch claim stands, and if it can, the evolutionary sequence is a sampling artifact. Fitting those epochs with line energy and width free, and marginalizing over covering fraction, would provide the same test.","tokens_in":17637,"feed_emoji":"🌀","tokens_out":9028,"duration_ms":87218,"temperature":0.7,"pith_summary":"The paper reports that the tidal disruption event AT2020afhd displayed X-ray absorption from an ultrafast outflow only between roughly days 74 and 215 after discovery, making it the first TDE for which the full birth-to-death evolution of such an outflow has been seen. Across about ten days the outflow velocity fell from 0.19c to 0.0097c, and the velocity and ionization parameter moved in opposite directions, the reverse of what radiation-pressure-driven wind models predict. The authors conclude that neither radiation pressure nor thermal launching can explain the wind, leaving magnetic driving as the most viable mechanism, and propose that the delayed appearance reflects a widening wind opening angle and/or gradual iron and oxygen enrichment as the stellar core is accreted. If the sequence is real, TDE disk winds can turn on, strengthen, decelerate, and vanish on month-long timescales, a behavior that future X-ray monitoring of bright TDEs can test directly.","feed_headline":"Ultrafast outflow appears, decelerates, and vanishes in one TDE","feed_subtitle":"X-ray wind in AT2020afhd dropped from 0.19c to 0.0097c in 10 days, challenging radiation-pressure models.","key_machinery":"The machinery is time-resolved X-ray spectroscopy of the broad blueshifted absorption feature near 0.95 keV, identified as iron M-shell unresolved transition array and oxygen K-edge absorption and modeled with the pion photoionization model in SPEX, a self-consistent treatment of a photoionized absorber. The four-epoch grouping of Swift/XRT spectra supplies the absence-appearance-strengthening-disappearance timeline, while individual NICER and XMM-Newton spectra track the rapid centroid shift that sets the deceleration.","core_discovery":"The paper's central claim is that AT2020afhd is the first tidal disruption event in which the entire evolutionary track of an X-ray ultrafast outflow is observed. Blueshifted absorption from the iron M-shell unresolved transition array and the oxygen K-edge is absent in the combined spectrum of the first 74 days, appears weakly by days 172-194, strengthens through days 194-215, and is absent again after day 215. During this active window the line centroid shifts from 0.95 keV to 0.78 keV in about ten days, which the photoionization fits translate into a velocity drop from roughly 0.19c to 0.0097c, while the ionization parameter rises as the velocity falls. The authors read this inverse correlation, together with the delayed onset at roughly constant luminosity, as inconsistent with radiation-pressure-driven winds and as evidence for magnetic launching; they attribute the delay to an expanding wind opening angle and/or metal enrichment during disk formation.","pith_inferences":["Beyond the paper: if delayed onset is a generic disk-formation effect rather than a quirk of this event, high-cadence X-ray monitoring of future TDEs should catch the same turn-on pattern within the first few months, letting observers time the disk formation phase directly.","Beyond the paper: the velocity-ionization anticorrelation could serve as a cheap diagnostic, since measuring both quantities in a sample of TDEs may separate magnetic from radiation-pressure winds without requiring high-resolution line profiles.","Beyond the paper: the seasonal gap means the true launch time could be as early as day 74 or as late as day 172, so scheduled observations across that window in comparable bright TDEs would pin down whether the delay scales with black hole mass or fallback timescale."],"forward_implications":["If the sequence is real, TDE disk winds can switch on and off on roughly 100-day timescales, so single-epoch detections or non-detections in other TDEs cannot be taken as permanent states.","The deceleration from 0.19c to 0.0097c at roughly constant luminosity rules out a declining mass fallback rate as the cause, the explanation previously applied to the slower UV wind in AT2019qiz.","The inverse velocity-ionization correlation and the absence of an ionization-luminosity correlation contradict radiation-pressure driving and shift the balance of evidence toward magnetically launched winds.","Even with dense multi-wavelength coverage, the inferred wind and jet mass loss is only a small fraction of the disrupted star's mass, so outflow observations alone are unlikely to resolve the TDE missing-energy problem."],"supporting_citations":[{"why":"Supplies the pion photoionization model used to fit the absorption features and provides the ASASSN-14li UFO as a comparison object.","marker":"Miller et al. 2015"},{"why":"Detected an early UFO in ASASSN-14li, the key comparison that makes AT2020afhd's delayed onset notable.","marker":"Kara et al. 2018"},{"why":"Detected an early UFO in ASASSN-20qc, another contrast for the delayed-onset claim.","marker":"Pasham et al. 2024"},{"why":"Found a late UFO in 3XMM J152130.7+074916, showing that late outflows exist without a full evolutionary sequence.","marker":"Lin et al. 2015"},{"why":"Found a UFO in AT2020ksf that remains detectable past day 770, providing the longer-lived counterpart to this transient wind.","marker":"Wevers et al. 2023"},{"why":"Invoked magnetic driving for the outflow in AT2021ehb, the mechanism the paper favors for AT2020afhd.","marker":"Xiang et al. 2024"},{"why":"Simulated radiation-pressure-driven TDE winds whose predicted velocity-ionization correlation is contradicted here.","marker":"Dai et al. 2018"},{"why":"Supplies hydrodynamical wind simulations used to estimate the launching radius and to rule out thermal driving.","marker":"Thomsen et al. 2022"},{"why":"Provides the mass-loss-rate relation used to estimate the wind's contribution to the missing-energy problem.","marker":"Nardini et al. 2015"}],"fun_headline_variants":["First full X-ray outflow evolution observed in a TDE","TDE's ultrafast wind: delayed launch, sharp deceleration, fade","X-ray outflow in TDE appears late, then drops from 0.19c to ~0.01c","Delayed X-ray wind in TDE challenges radiation-pressure models","TDE shows first complete track of an ultrafast outflow"],"cache_read_input_tokens":3200,"weakest_assumption_plain":"The load-bearing premise is that the non-detections before day 74 and after day 215 mean the outflow really was absent, rather than too faint to detect in Swift spectra that contain only roughly 800-1100 counts and in which the line energy and width were fixed to the detection epochs.","fun_headline_variants_meta":{"raw":{"variants":["First full X-ray outflow evolution observed in a TDE","TDE's ultrafast wind: delayed launch, sharp deceleration, fade","X-ray outflow in TDE appears late, then drops from 0.19c to ~0.01c","Delayed X-ray wind in TDE challenges radiation-pressure models","TDE shows first complete track of an ultrafast outflow"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.00056,"raw_usage":{"total_tokens":2671,"prompt_tokens":968,"completion_tokens":1703,"prompt_tokens_details":{"cached_tokens":384},"prompt_cache_hit_tokens":384,"prompt_cache_miss_tokens":584,"completion_tokens_details":{"reasoning_tokens":1605}},"tokens_in":584,"tokens_out":1703,"duration_ms":14691,"temperature":1.0,"reasoning_tokens":1605,"cache_read_input_tokens":384,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-06T15:32:02.279694+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"A deeper co-added spectrum from days 0-74 and days 220-302, with enough counts to detect the same iron and oxygen absorption at the epoch-(3) amount of absorbing gas and ionization state, would settle whether the turn-on and turn-off are physical; if the line still cannot be found, the delayed-launch claim stands, and if it can, the evolutionary sequence is a sampling artifact. Fitting those epochs with line energy and width free, and marginalizing over covering fraction, would provide the same test.","supporting_citations":[{"cited_title":"W., Irwin , J","cited_arxiv_id":null,"evidence_quote":"Found a late UFO in 3XMM J152130.7+074916, showing that late outflows exist without a full evolutionary sequence."},{"cited_title":"M., Zoghbi , A., et al","cited_arxiv_id":null,"evidence_quote":"Invoked magnetic driving for the outflow in AT2021ehb, the mechanism the paper favors for AT2020afhd."}],"review_version":1}