{"id":"91b53da7-8bc9-4eed-8a56-c34b2f695fa0","arxiv_id":"2509.09133","paper_version":1,"verdict":"CONDITIONAL","confidence":"MODERATE","novelty_score":6.0,"correctness_risk":"medium","formal_verification":"none","parameter_count":4,"one_line_summary":"ULXs show suppressed reprocessed X-ray emission: iron line equivalent widths are below 11 to 20 eV, no eclipse hardness ratio change is seen, and the pattern points to metal-poor or highly ionized environments.","lead":"Using archival XMM-Newton and Chandra data, the authors set the tightest upper limits yet on iron K-alpha emission in eleven ultraluminous X-ray sources, and find no hardness ratio change during eclipses in five eclipsing ULXs. They interpret the weak reprocessed emission as evidence that ULX surroundings are metal-poor or highly ionized.","discovery_kind":"new_application","skeptic_critique":{"model":"deepseek-v4-flash","headline":"The 11-20 eV combined Fe Kα EW upper limit is obtained after excluding the only ULX epoch with a detected iron line (NGC 300 ULX1 2010, EW=343±93 eV); without a robustness test that includes it, the headline constraint may be a selection artifact.","rationale":"I agree with the reader's CONDITIONAL verdict: the iron-line constraints are well-executed but the environmental interpretation needs additional work. However, I identify a different load-bearing weakness than the reader's NH-baseline concern. The NH issue primarily affects the speculative metal-poor/highly-ionized discussion; the exclusion of the 2010 NGC 300 ULX1 observation directly threatens the paper's most quantitative claim, the 11-20 eV combined EW upper limit. The authors flag the exclusion explicitly (§2), but they do not test its impact. A simple re-stacking with that epoch included would settle whether the central result is robust. My verdict remains CONDITIONAL (UNCHANGED relative to the reader) because the concern is addressable and the rest of the analysis (simulations, individual upper limits) appears sound. I mark agreement as partial because I do not adopt the reader's weakest_assumption as the primary issue, though it is a legitimate secondary concern.","tokens_in":23735,"tokens_out":17505,"duration_ms":218778,"concrete_test":"Re-run the §4.1 stacking analysis including the 2010 NGC 300 ULX1 EPIC-pn spectrum (using the same tbabs*(powerlaw+gauss) model, fixed NH=0.15e22, Gaussian at 6.4 keV, 5-8 keV range) along with all other spectra, and report the combined EW and 90% upper limit/detection. Also fit the 2010 spectrum alone to confirm the input EW. If the combined result remains <20 eV, the exclusion is benign; if it yields a detection or an upper limit several times larger, the paper's headline constraint does not hold for ULX sources as a class.","verdict_should_be":"UNCHANGED","load_bearing_attack":"The paper's strongest quantitative claim is the combined upper limit EW<11-20 eV (Table 4, §4.1). In §2, the authors exclude the 2010 XMM-Newton observation of NGC 300 ULX1, the sole ULX epoch with a prominent 6.4 keV line (EW=343±93 eV; Carpano et al. 2018), because the spectrum was 'remarkably different' and no pulsations were detected. This is a selection cut on the very signal being measured: NGC 300 ULX1 is a ULXP and still a ULX in that epoch, and the absence of detected pulsations does not remove it from the ULX population. The combined 5-8 keV stack is therefore an upper limit for the sample after deleting the only known detection. If that epoch is included, the combined spectrum may show a line or the upper limit may rise well above 20 eV, invalidating the 'most stringent constraint' and weakening the claim that reprocessed iron emission is universally meagre in ULXs. The manuscript does not report a stacking test with the 2010 spectrum included, so the central result is vulnerable to a single excluded observation.","agreement_with_reader":"partial"},"referee_report":{"model":"deepseek-v4-flash","summary":"This paper investigates reprocessed X-ray emission in ULXs/ULXPs using three diagnostics: (i) a search for Fe Kα lines in combined XMM-Newton EPIC-pn spectra of six bright ULXs and five ULXPs, (ii) hardness-ratio evolution across eclipses in five eclipsing ULXs compared with the Galactic HMXBs LMC X-4 and SMC X-1, and (iii) the eclipse-to-out-of-eclipse flux ratio in the same eclipsing systems. The authors report 90% upper limits of 11–20 eV on the equivalent width of a narrow 6.4 keV line in the combined spectrum, no significant hardness-ratio changes during ingress/egress, and higher reprocessed flux fractions than in most Galactic HMXBs. They interpret the combined results as evidence that ULX environments are either metal-poor or highly ionized, supported by a curve-of-growth comparison and a CAK radiation-driven wind model.","tokens_in":24090,"tokens_out":6825,"duration_ms":81493,"significance":"If the claims hold, the paper would establish that ULX environments are not simply scaled-up HMXB winds, a relevant constraint for understanding super-Eddington accretion and donor-wind reprocessing. The main strengths are the large archival sample, the clean narrow-band spectral fitting, the injection tests calibrating the combined upper limit, and the explicit comparison with Galactic eclipsing systems. However, the headline upper limit is obtained after deleting the only ULX epoch with a detected Fe Kα line, and the environmental interpretation depends on comparing literature line-of-sight N_H values with a wind model without subtracting foreground interstellar absorption. The result is a useful observational constraint, but the universal, population-level conclusion is not yet robust.","major_comments":[{"comment":"The combined upper limit of 11–20 eV is computed after excluding the 2010 XMM-Newton observation of NGC 300 ULX1, the only ULX epoch in the sample with a detected Fe Kα line (EW = 343 ± 93 eV; Carpano et al. 2018). The stated justification—'remarkably different spectrum' and no detected pulsations—does not remove the epoch from the ULX/ULXP population; it is a selection on the very signal being measured. Please report a stacking test that includes the 2010 spectrum, or explicitly state that the constraint applies only to the non-detection epochs. If the limit degrades above ~20 eV when the detection is included, the 'most stringent constraint' and the inference of universal suppression require revision.","section":"§2, §4.1, Table 4"},{"comment":"The curve-of-growth comparison (Fig. 5) and the wind-model comparison (Fig. 6) treat the literature N_H values in Table 1 as the intrinsic line-of-sight column density of the ULX binary environment. These values are total absorption columns from spectral fits and almost certainly contain significant Milky Way and host-galaxy interstellar foreground. Without subtracting that foreground, the fact that typical ULX N_H is 0.06–0.7×10^22 cm^-2 does not imply that the donor-wind column is small; the apparent deficit against the CAK model predictions (10^22–10^25 cm^-2) could disappear. Please propagate foreground-subtracted intrinsic columns with uncertainties, or substantially downgrade the environmental claim. The wind-model calculation also adopts M82 X-2 binary parameters as representative of all ULX donors without exploring parameter dependence.","section":"§5, Table 1, Fig. 6"},{"comment":"The non-detection of hardness-ratio changes during eclipse ingress/egress in the ULX sources is reported without any sensitivity estimate. The ULX light curves are binned at 1200–3000 s with low count rates, so the lack of a visible hardness-ratio variation may be a photon-statistics effect. Please quantify, via simulations, the smallest hardness-ratio change excluded at 90% confidence for each source, and compare that threshold with the excursions seen in LMC X-4 and SMC X-1. Without this, the qualitative contrast with Galactic HMXBs is not strongly supported.","section":"§4.2, Fig. 3"},{"comment":"The OOE-to-eclipse flux ratios for the ULXs (e.g., 4.5–5.2 for M51 S1, 3.0–4.4 for M51 S2, 23.4 for CG X-1, and the wide ranges for M51 ULX1/2) overlap substantially with the LMXB range (6–54) and the lower end of the HMXB range (8–237) quoted in the introduction. The statement that 'the reprocessing efficiency is larger in ULXs compared to Galactic HMXBs' requires a formal statistical comparison of the distributions; as presented it appears to invert the OOE/eclipse ratio without testing significance. Please provide such a comparison or soften the claim.","section":"§4.3, Table 5"}],"minor_comments":[{"comment":"The abstract says 'reprocessing efficiency (eclipse to OOE flux ratio)' while §4.3 defines the OOE-to-eclipse ratio; please make the reciprocal relationship explicit to avoid confusion.","section":"Abstract / §4.3"},{"comment":"The 2010 NGC 300 ULX1 observation is not listed in Table A1. Please add the ObsID/MJD and state the exact exclusion criterion, including the reason why 'no pulsations' removes a ULXP from the sample.","section":"§2, Table A1"},{"comment":"The Gaussian line is fixed at 6.4 keV with σ = 0.1 keV. Many ULX spectra show ionized lines at 6.7–6.9 keV; please either search with a free centroid or at 6.7 keV, or clearly state that the quoted limits apply only to a narrow neutral Fe Kα line.","section":"§4.1"},{"comment":"The injection tests only inject lines at 6.4 keV with a fixed width; they do not validate recovery of broad or ionized lines. This limitation should be stated.","section":"§4.1, Fig. 2"},{"comment":"The adopted N_H values are quoted without uncertainties or a specification of whether they include Galactic foreground absorption. Adding both would improve the transparency of Fig. 5.","section":"Table 1"},{"comment":"In Eq. (3), κ is not defined; please define the conversion factor and the integration geometry. In Fig. 6, the gray 'measured' band appears to represent the full range of Table 1 values; overlaying the per-source measurements would be more informative.","section":"Eq. (3), Fig. 6"}],"recommendation":"major_revision","confidential_remarks":"This is a useful archival study with a clear set of diagnostics, and the injection tests are a good check on the line upper limits. The main risk is overclaiming a universal constraint when the sample excludes the one epoch with a detected Fe Kα line. I would not reject on the current text, but the authors need to make the sample-selection choice transparent and show how the 2010 NGC 300 ULX1 spectrum affects the combined limit. The N_H foreground-subtraction issue also needs to be addressed or the environmental interpretation must be softened. If those points are handled in revision, the paper could become acceptable."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"The paper gives a useful new upper limit on Fe Kα emission in ULXs: from a stack of 11 sources, the 90% limit on the equivalent width is 11–20 eV, a factor of a few better than Walton et al.'s per-source limits. The spectral fits look honest, and the injection tests show they recover artificial lines with EW 15, 20 and 30 eV. The eclipse hardness-ratio and OOE/eclipse flux-ratio analyses are new territory for ULXs, even if the sample is small.\n\nThe main problem is the selection. The combined limit is obtained after excluding the only ULX epoch that shows a strong iron line: NGC 300 ULX1 in 2010, EW around 343 eV, from Carpano et al. The authors justify the exclusion by saying the spectrum was remarkably different and no pulsations were detected. But that epoch is still a ULX, and deleting it from the stack removes the one positive signal. Without a robustness test that includes the 2010 data (or at least quantifies how the combined limit would change), the claim of a 11–20 eV constraint on ULXs in general is not supported. It's a constraint on the chosen sample.\n\nSecond soft spot: the hardness-ratio non-detection during ULX eclipses is presented as a contrast to Galactic HMXBs, but there is no sensitivity estimate. The ULX light curves are sparse, with 1200–3000 s bins; a null result at that cadence is weak evidence.\n\nThird: the wind-model comparison (Fig. 6) and the curve of growth (Fig. 5) use literature NH values that almost certainly include interstellar absorption along the line of sight. If those are ISM-dominated, the comparison is not measuring the binary wind column. The conclusion that ULXs are metal-poor or highly ionized may be true (there is independent supporting literature), but this paper's data don't nail it.\n\nThe core measurement—weak Fe Kα in most ULXs—is probably right. But the paper overreaches with the environmental interpretation. It deserves peer review, but needs major revision: include the 2010 epoch in a stacking robustness check, add a sensitivity analysis for the hardness ratio, and recompute the NH comparison with foreground subtraction or an explicit caveat.\n\nWorth a reading group discussion, especially about selection effects in stacked spectra.","headline":"The combined Fe Kα EW limit is a real improvement, but the headline number excludes the only ULX epoch with a detected line, and the environmental interpretation leans on a NH comparison that ignores interstellar absorption.","tokens_in":24555,"tokens_out":7009,"would_cite":false,"duration_ms":79088,"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":"Reprocessed X-ray emission is suppressed in ultraluminous X-ray sources","keywords":["ultraluminous X-ray sources","X-ray reprocessing","iron K-alpha line","equivalent width","eclipsing X-ray binaries","hardness ratio","super-Eddington accretion","stellar wind"],"falsifier":"A stacked ULX spectrum with sensitivity below 10 eV that recovers an iron K-alpha line above 20 eV equivalent width, or an eclipse transition in any ULX showing a hardness-ratio increase like those in Galactic HMXBs, would break the central claim. The authors' own simulations show that their method can recover injected lines, so a future detection at the few-tens-of-eV level would be decisive.","tokens_in":23619,"feed_emoji":"🛰️","tokens_out":4451,"duration_ms":47984,"temperature":0.7,"pith_summary":"The paper sets out to show that, despite accreting at super-Eddington rates, ultraluminous X-ray sources (ULXs) show strikingly little X-ray reprocessing compared with Galactic X-ray binaries. It combines three diagnostics: stacking spectra to hunt for the iron K-alpha fluorescence line, tracking hardness-ratio changes at eclipse ingress and egress, and comparing eclipse to out-of-eclipse fluxes. The result is an upper limit of 11-20 eV on the iron-line equivalent width, no eclipse-transition hardening, and a larger eclipse-to-out-of-eclipse flux ratio than in Galactic binaries. The authors read this as evidence that ULX surroundings are metal-poor or highly ionized, which would explain why the expected reprocessing signatures are missing. This matters because it points to a fundamental environmental difference in how super-Eddington accretors illuminate their surroundings.","feed_headline":"ULXs show almost no reprocessed X-ray emission","feed_subtitle":"Iron-line upper limits of 11-20 eV and flat eclipse hardness ratios point to metal-poor or highly ionized gas.","key_machinery":"Three diagnostics carry the argument. First, a stacked spectral analysis with an absorbed power-law plus a Gaussian line, fixed at 6.4 keV, converts non-detections into equivalent-width upper limits. Second, a hardness ratio HR=(H-S)/(H+S) is tracked across eclipse transitions and compared with known Galactic eclipsing HMXBs. Third, spectral fits in and out of eclipse give the eclipse-to-out-of-eclipse flux ratio. A curve of growth comparing ULXs against Galactic source classes, together with a radiation-driven wind model that predicts much larger column densities than observed, supplies the interpretive step: the column is there but the iron fluorescence is missing, implying low metallicity","core_discovery":"The paper's central claim is that the material around ULXs re-emits far less of the incident X-ray flux than standard binary wind reprocessing predicts. In eleven ULX and ultraluminous X-ray pulsar systems, no iron K-alpha line is detected, and stacking all spectra gives a 90% confidence upper limit of 11-20 eV on its equivalent width—several times tighter than earlier limits. In five eclipsing ULXs, the hardness ratio does not rise during ingress or egress, whereas Galactic eclipsing high-mass X-ray binaries show clear hardening from wind absorption. The eclipse-to-out-of-eclipse flux ratios are also smaller (3-23 times) than typical Galactic values, meaning the residual eclipse light is re","pith_inferences":["If the environment is highly ionized rather than simply metal-poor, the reprocessed flux may emerge as recombination lines of hydrogen and helium instead of iron K-alpha; future ULX spectra could search for those lines as a discriminant.","The wind-model comparison assumes the literature column densities are intrinsic to the ULX binary. If much of that column is interstellar, the predicted-vs-observed deficit shrinks, and the metal-poor conclusion would need host-galaxy absorption maps to separate components.","A testable corollary is that ULXs in higher-metallicity host regions, like NGC 925 ULX-3, should show stronger iron lines or eclipse hardening if metallicity is the controlling variable; existing data could be re-examined for such source-by-source differences.","The same suppression logic could apply to other super-Eddington accretors, such as tidal disruption events or hyperluminous pulsars, suggesting that reprocessing diagnostics may systematically underestimate the gas content of extreme accretion flows."],"forward_implications":["The tight iron-line upper limits (11-20 eV) mean that a ULX sample stacked from many observations still shows no fluorescence, so the absence is an environmental property, not a data limitation.","Eclipsing ULXs do not harden during ingress/egress like Galactic HMXBs, indicating their donor winds absorb soft X-rays far less.","Residual eclipse flux is a larger fraction of the out-of-eclipse flux in ULXs than in Galactic binaries, so scattering or reprocessing by an extended ionized medium contributes more.","If the interpretation holds, the gas around most ULXs is either metal-poor or highly ionized, with consequences for how super-Eddington outflows are diagnosed in other galaxies."],"fun_headline_variants":["ULXs reemit almost none of their X-rays","Iron lines vanish in ULXs: reprocessing is minimal","Flat eclipse hardness hints at ionized gas around ULXs","Tightest iron-line limits yet: ULXs reprocess <20 eV"],"cache_read_input_tokens":2304,"weakest_assumption_plain":"The literature column densities used as the intrinsic ULX line-of-sight absorption really do trace the ULX donor wind rather than being dominated by interstellar gas in the Milky Way or the host galaxy.","fun_headline_variants_meta":{"raw":{"variants":["ULXs reemit almost none of their X-rays","Iron lines vanish in ULXs: reprocessing is minimal","Flat eclipse hardness hints at ionized gas around ULXs","Tightest iron-line limits yet: ULXs reprocess <20 eV"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.000222,"raw_usage":{"total_tokens":1337,"prompt_tokens":839,"completion_tokens":498,"prompt_tokens_details":{"cached_tokens":256},"prompt_cache_hit_tokens":256,"prompt_cache_miss_tokens":583,"completion_tokens_details":{"reasoning_tokens":428}},"tokens_in":583,"tokens_out":498,"duration_ms":6234,"temperature":1.0,"reasoning_tokens":428,"cache_read_input_tokens":256,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-04T19:38:10.466167+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"A stacked ULX spectrum with sensitivity below 10 eV that recovers an iron K-alpha line above 20 eV equivalent width, or an eclipse transition in any ULX showing a hardness-ratio increase like those in Galactic HMXBs, would break the central claim. The authors' own simulations show that their method can recover injected lines, so a future detection at the few-tens-of-eV level would be decisive.","supporting_citations":[],"review_version":1}