{"id":"8f5e95ab-6d41-41a9-891a-c65f73908e7e","arxiv_id":"2601.16977","paper_version":3,"verdict":"CONDITIONAL","confidence":"MODERATE","novelty_score":7.0,"correctness_risk":"medium","formal_verification":"none","parameter_count":3,"one_line_summary":"JWST infrared spectra of eight low-luminosity AGN reveal a transition near 1/3000 of the Eddington rate below which ultraviolet ionizing photons drop sharply and molecular gas is roughly 500 K hotter than in brighter AGN.","lead":"Eight faint galactic nuclei observed by JWST show a sharp drop in ultraviolet ionizing power when their accretion rate falls below about 1/3000 of the Eddington limit, along with unusually warm molecular gas. The findings give a new infrared-based view of how low-luminosity active galactic nuclei (LLAGN) differ from brighter quasars and Seyferts.","discovery_kind":"new_application","skeptic_critique":{"model":"deepseek-v4-flash","headline":"The -3.5 Eddington break rests on X-ray bolometric corrections calibrated for standard AGN, yet the paper reports a 0.54 dex offset for its own LLAGN; this is a circular basis for a transition in SED shape.","rationale":"I focused on the conditional linear model in Fig. 6 / Eq. 11 because it is the quantitative basis for the -3.5 threshold. The bolometric correction is load-bearing because it defines both the breakpoint's x-coordinate and the high/low membership. The 0.54 dex offset is not a minor detail; it is comparable to the full dynamic range of the low-branch separation (up to an order of magnitude in [NeV]) and to the breakpoint uncertainty. The line-ratio evidence for UV-deficient SEDs is independent and somewhat supportive, but it does not establish the threshold; NGC 4395 is the only clear high-Eddington source and its ratios match the disk model, while NGC 4258 is an outlier. I also considered jet contamination of the 2–10 keV luminosities; this strengthens the concern but is downstream of the KX issue. The test—recompute with SED-based or LLAGN-calibrated Lbol and refit—would settle whether the threshold is real or calibration-induced. No independent formal verification exists, but the data are public, so the check is feasible.","tokens_in":36132,"tokens_out":6623,"duration_ms":79325,"concrete_test":"Recompute all Eddington ratios in Table 1 using SED-integrated Lbol for the four galaxies with direct values (Prieto et al. 2010/2014; Fernández-Ontiveros et al. 2023) and, for the remaining targets, a LINER/LLAGN-specific bolometric correction (e.g., Eracleous et al. 2010 or Nemmen et al. 2014) instead of Eq. (2). Refit the conditional linear model (Eq. 11) to the [NeV]–Lx data and record the best-fit z0 and ΔBIC. If z0 shifts by more than its quoted +0.6/−1.0 dex, or if ΔBIC drops below ~10, the transition is not robust. As a secondary check, refit the split using log(L2–10keV/MBH) directly, dropping KX entirely, and compare the high/low offset.","verdict_should_be":"UNCHANGED","load_bearing_attack":"The paper's central assertion (Abstract; §5.1.2) is a transition at log(Lbol/LEdd) ≈ -3.5, below which the ionizing UV output drops. The independent variable in that split is Lbol/LEdd, with Lbol obtained from Eq. (2), the Duras et al. (2020) KX(L2–10 keV) relation calibrated on ~1000 Type 1/2 AGN with standard SEDs. Section 2.3 acknowledges a median absolute difference of 0.54 dex between this X-ray-based Lbol and direct SED-integrated Lbol for NGC 1052, M87, Cen A, and Sombrero—precisely the sources at the low-Eddington end of Fig. 6. A KX that depends only on Lx cannot encode the very change in SED shape that the paper claims; using it to define z therefore embeds the standard-SED assumption in the test. If the true LLAGN bolometric correction is lower (consistent with UV-deficient SEDs), the low-branch points shift to lower z and the fitted z0 may move by more than its quoted +0.6/−1.0 dex or vanish; if jet-dominated X-ray emission makes KX invalid in the other direction, membership in the high/low bins changes. The claim that the offset in Fig. 6 is larger than the ~0.1 dex spread in kX refers to scatter in Duras-type corrections, not to the 0.54 dex systematic offset the authors themselves document. Thus the breakpoint, which is already only a modest ΔBIC≈9 preference, rests on an unvalidated calibration.","agreement_with_reader":"agree"},"referee_report":{"model":"deepseek-v4-flash","summary":"The paper presents JWST NIRSpec and MIRI/MRS spectra of eight low-luminosity AGN (LLAGN), including Cen A from archival data. The authors measure a large set of infrared emission lines, including high-ionization species such as [Ne V] 14.3 µm and [O IV] 26 µm, and use them to argue that below log(L_bol/L_Edd) ≈ -3.5 the ionizing continuum becomes increasingly deficient in UV photons. The main evidence is a conditional linear model fit to the [Ne V]–L_2-10 keV relation, which prefers a break at that Eddington ratio (ΔBIC ≈ 9), and a set of diagnostic line-ratio diagrams in which the LLAGN are better matched by photoionization models with a power-law, UV-bump-free SED than by accretion-disk SEDs or shocks. Additional results include elevated rotational H2 excitation temperatures, H2/PAH ratios consistent with AGN, and compact nuclear silicate emission in most targets.","tokens_in":36539,"tokens_out":5205,"duration_ms":59649,"significance":"If the claimed transition is correct, it provides a direct, IR-based diagnostic for the switch from a standard thin disk to a radiatively inefficient accretion flow in LLAGN, a regime that is otherwise difficult to probe. The observational strengths are substantial: high-quality JWST data, a careful line-fitting pipeline with Monte Carlo uncertainties, explicit spatial-concentration checks against point-source encircled-energy profiles, and public release of reduced cubes and extracted spectra. The H2 aperture correction is tested against sky subtraction, and the [Ne V]–[O IV] correlation is extended to lower luminosities than previous samples. The weaknesses concern the robustness of the Eddington-ratio axis and the statistical weight of the break, both of which are load-bearing for the headline claim.","major_comments":[{"comment":"The Eddington-ratio split that defines the claimed -3.5 transition is computed from the Duras et al. (2020) KX(L2-10 keV) bolometric correction, calibrated on ~1000 luminous Type 1/2 AGN with standard SEDs. The paper itself reports a median absolute difference of 0.54 dex between this proxy and direct SED-integrated L_bol for the four LLAGN with available SEDs (NGC 1052, M87, Cen A, Sombrero) — exactly the sources that occupy the low-Eddington branch. Because z = log(KX) + log(LX) - log(L_Edd), any systematic error in KX for UV-deficient LLAGN directly shifts the split variable. If the true KX is lower, the low-branch points shift to lower z and the fitted z0 could move by more than the quoted +0.6/-1.0 dex, or the break could vanish. Please test robustness: recompute the conditional fit with SED-based L_bol for the four sources, with an alternative bolometric correction, or with z defin","section":"§2.3, Eq. (2); §5.1.2, Fig. 6"},{"comment":"The statistical evidence for a break is modest: ΔBIC ≈ 9 with a threshold plus separate slopes/intercepts (2 extra parameters). The two branch slopes are consistent with each other within 1σ (0.957±0.109 vs. 0.845±0.190), so the claimed effect is essentially an offset, not a change in slope. The paper does not state how many sources lie in each branch; from Table 1, only a handful (≈5–6) lie below -3.5. The quoted threshold uncertainty is also asymmetric and large. Please report the number of sources per branch, show residual plots, and run a leave-one-out or bootstrap analysis to confirm that the offset is not driven by one or two extreme low-LX points. The abstract's 'order of magnitude' decrease should be quantified with the actual model offset and its uncertainty.","section":"§5.1.2, Fig. 6"},{"comment":"The 'power-law' photoionization grid that the LLAGN ratios agree with is constructed from the Sombrero SED (Fernández-Ontiveros et al. 2023), i.e., from a UV-deficient LLAGN that is itself part of the sample. This does not by itself invalidate the comparison, but it weakens the claim that the data are 'consistent with a UV-deficient SED' because the grid was built from a UV-deficient source. The argument would be much stronger if the UV-bump amplitude were a free parameter (e.g., varying the fraction of thermal disk emission) and the models showed that the data prefer a reduced or zero bump. Also, the full model description is deferred to 'Acharya et al., in prep.'; the paper should include enough detail (ionizing SED shape, metallicity, density, column, and the exact CLOUDY parameter ranges used) to allow reproduction, since this is the primary evidence for the central 'UV-deficient SED","section":"§5.1.1, Fig. 5"}],"minor_comments":[{"comment":"The caption says sources with log(L_Bol/L_Edd) less than -3 are plotted as circles and higher as triangles, while the text and abstract use -3.5. Reconcile this inconsistency.","section":"Fig. 6 caption"},{"comment":"The threshold is quoted as -3.5 +0.6/-1.0 in Section 5.1.2 but -3.5 +0.8/-0.7 in the Conclusions. Which uncertainty is correct?","section":"Abstract and §5.1.2 vs. §6"},{"comment":"Typo: 'Given this offest' should be 'offset'. Also, the statement that the observed offset is larger than the ~0.1 dex spread in X-ray bolometric corrections (kX ~16–20) is unclear: kX is not defined, and the relationship between kX scatter and L_bol scatter should be shown explicitly.","section":"§2.3"},{"comment":"The last two columns list L_bol and log(L_Bol/L_Edd) without separation, e.g., '42.71−4.21' reads ambiguously. Use separate columns or a clear format such as 'log L_bol = 42.71, log(L/L_Edd) = −4.21'.","section":"Table 1"},{"comment":"The description of the H2 aperture correction ('scaling each measured flux by the ratio of its value in the native aperture to that measured in a larger 1.2'' aperture, under the assumption that the spatial distribution of the emission is roughly constant') is somewhat confusing — if the distribution is roughly constant, the correction could be area-based; if it is not, the ratio itself is the empirical correction. Clarify the procedure and its validation.","section":"§3.1"}],"recommendation":"major_revision","confidential_remarks":"The paper presents a valuable dataset and careful measurements, and the central idea — an Eddington-ratio-dependent change in the ionizing continuum of LLAGN — is plausible and worth publishing. However, the headline break at -3.5 is currently supported by a modest ΔBIC on a split variable that is constructed from the same X-ray luminosity used on the x-axis, and the photoionization grid used to infer UV deficiency is built from one of the sample sources. These are fixable with additional robustness tests and model presentation, but they are load-bearing. I also noticed inconsistencies in the quoted threshold and its uncertainty between sections, which should be reconciled. I recommend major revision."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"You should know this paper is the first JWST mid-IR census of LLAGN nuclei and it delivers a 131-line catalog with detections of [Ne V] and [O IV] down to luminosities no one had reached. That part is real and will be used for years.\n\nWhat is new: they extend the [Ne V]-[O IV] correlation by orders of magnitude, show the emission is spatially concentrated, find elevated H2 excitation temperatures, and propose an empirical break in the [Ne V]-Lx relation at log Lbol/LEdd ~ -3.5, below which [Ne V] is weak by ~10x. They also show line ratios fit power-law photoionization models (no UV bump) better than disk models for all but NGC 4395. The measurements look careful: Monte Carlo uncertainties, PSF-matched apertures, WICKED for NIRSpec wiggles, sky subtraction checked, data public.\n\nThe central claim rests on Eddington ratios computed from Duras et al. X-ray bolometric correction, which they acknowledge gives 0.54 dex median offset against SED-integrated values for four LLAGN. If the true KX is lower for UV-deficient sources, the low-z points shift and the break location—already only ΔBIC≈9—could move by more than the quoted uncertainty. This doesn't kill the paper, but it means the break is provisional until bolometric luminosities are measured from actual SEDs.\n\nAlso the comparison sample is modest (43 objects), and the low branch is anchored by just a few objects; the H2 temperature claims are intriguing but rely on literature comparisons with different apertures, and the silicate strengths are quoted without errors.\n\nOverall, the paper is honest and the data will be a resource. The -3.5 transition is a hypothesis, not a discovery, until the bolometric correction issue is addressed. I'd send it to a good referee—the dataset alone justifies the review—and I'd ask the referee to check the Eddington-ratio dependence and the statistics of the break. I'd cite it and bring it to reading group.","headline":"A valuable JWST dataset and a plausible but not yet nailed transition at log L/LEdd ~ -3.5, with the main caveat being the bolometric correction used to compute Eddington ratios.","tokens_in":37195,"tokens_out":2298,"would_cite":true,"duration_ms":25441,"reading_group":"yes","serious_thinker":"yes","would_accept_peer_review":true},"rs_alignment":null,"lean_confirmation":null,"pith_extraction":{"msc":[],"pacs":[],"model":"deepseek-v4-flash","headline":"JWST spectra of eight low-luminosity AGN reveal a transition at an Eddington ratio near 10^-3.5 where the ionizing ultraviolet output drops by up to an order of magnitude, indicating a switch to a radiatively inefficient accretion flow.","keywords":["low-luminosity AGN","JWST spectroscopy","emission-line diagnostics","Eddington ratio","radiatively inefficient accretion flow","molecular hydrogen excitation","silicate emission","neon lines"],"falsifier":"A direct measurement of bolometric luminosity from multi-wavelength SEDs for a sample of LLAGN spanning the critical Eddington ratio, followed by a recomputation of the [Ne V]–X-ray relation, would settle the claim. If the break vanishes or moves by more than the quoted uncertainty when using SED-based L_bol, the claimed accretion-state transition would be falsified. Alternatively, finding a low-Eddington LLAGN with strong [Ne V] at fixed X-ray luminosity (i.e., no suppression) would already contradict the observed relation.","tokens_in":35990,"feed_emoji":"🔭","tokens_out":4811,"duration_ms":49379,"temperature":0.7,"pith_summary":"The paper uses JWST near- and mid-infrared spectroscopy to isolate the nuclear emission of eight low-luminosity active galactic nuclei (LLAGN) plus Cen A, detecting high-ionization lines an order of magnitude fainter than previously possible. It claims that emission-line diagnostics show a transition at log(L_bol/L_Edd) ~ -3.5, below which the spectral energy distribution becomes increasingly deficient in ultraviolet photons. The evidence is a conditional linear fit to [Ne V] 14.3 µm versus 2-10 keV X-ray luminosity, finding a break at that Eddington ratio with [Ne V] suppressed by up to an order of magnitude at fixed X-ray luminosity in the low-Eddington branch. Diagnostic ratios like [Ne V]/[Ne II], [Ar III]/[Ar II], and [S IV]/[S III] are best matched by photoionization with a power-law continuum lacking a thermal UV bump. The paper also finds molecular hydrogen excitation temperatures ~500 K hotter than in luminous AGN or star-forming galaxies, pointing to additional heating from shocks or X-ray-dominated regions.","feed_headline":"JWST locates the accretion rate where AGN ultraviolet light drops","feed_subtitle":"Emission lines from eight low-luminosity nuclei reveal a radiatively inefficient regime with warmer molecular gas and weaker high-energy ion","key_machinery":"The argument rests on two main tools. First, a conditional linear model fits separate slopes and intercepts to the [Ne V] 14.3 µm versus 2-10 keV X-ray luminosity relation above and below a threshold Eddington ratio; the best fit (by BIC) places the break at log(L_bol/L_Edd) = -3.5 (+0.6/-1.0). Second, diagnostic line ratios ([Ne V]/[Ne II], [Ar III]/[Ar II], [S IV]/[S III]) are compared with grids from shock and photoionization models; the data systematically favor power-law photoionization models (without a UV bump) over disk models or shocks alone. These tools together localize the change in ionizing photon production to the accretion state of the black hole.","core_discovery":"The central discovery is that low-luminosity AGN below log(L_bol/L_Edd) ~ -3.5 have a UV-deficient ionizing continuum. Using [Ne V] 14.3 µm as a proxy for photons above 97 eV, the authors find that the relation between [Ne V] and 2-10 keV X-ray luminosity breaks at this Eddington ratio: at fixed X-ray luminosity, [Ne V] emission is suppressed by up to an order of magnitude in the low-Eddington regime. The observed ratios of high- to low-ionization lines are consistent with photoionization by a power-law SED without a thermal UV bump, rather than by the standard accretion-disk SED of higher-luminosity AGN. Together, these results indicate a genuine change in the ionizing spectrum near this ac","pith_inferences":["An extension not explored in the paper: the same break might appear in other high-ionization lines like [O IV] 26 µm or [Ne VI] 7.6 µm; if so, it would strengthen the claim that the UV deficit is a general property of the ionizing continuum, not a peculiarity of the [Ne V] line.","A testable prediction: the location of the break at log(L_bol/L_Edd) ~ -3.5 is close to theoretical expectations for the transition to radiatively inefficient accretion; direct SED-based L_bol measurements for the same sources would show whether the break remains at the same Eddington ratio or moves, which would distinguish between an intrinsic accretion-state change and an artifact of the adopted","If the UV-deficient SED is real, one might expect the optical/UV continuum of these LLAGN to lack the characteristic 'big blue bump'; this could be tested with spatially resolved UV spectroscopy or by searching for a soft X-ray excess in high-quality X-ray data."],"forward_implications":["If correct, LLAGN below log(L_bol/L_Edd) ~ -3.5 are powered by radiatively inefficient accretion flows with little or no standard thin-disk UV emission, meaning mechanical feedback (jets/outflows) is the dominant energy output in this regime.","The suppression of [Ne V] at low Eddington ratios implies that using this line as a bolometric indicator for low-luminosity AGN will systematically underestimate the accretion power unless the Eddington ratio is taken into account.","The elevated H2 excitation temperatures and H2/PAH ratios indicate that kinetic-mode feedback, via shocks or X-ray irradiation, heats the molecular interstellar medium in LLAGN, which may affect star formation in galaxy centers.","The detection of nuclear silicate emission in most targets suggests that optically thin dust survives at parsec scales around low-luminosity black holes, informing models of the torus and dust distribution in the absence of a classic torus."],"fun_headline_variants":["JWST finds AGN UV deficit below Eddington ratio -3.5","Accretion rate shift removes UV bump in LLAGN","Faint AGN lose high-energy photons at a critical rate","JWST reveals a UV-poor regime in low-luminosity AGN"],"cache_read_input_tokens":2304,"weakest_assumption_plain":"The paper's division between low- and high-Eddington sources relies on bolometric luminosities derived from a single X-ray bolometric correction calibrated on more luminous AGN; if that correction is wrong for LLAGN, the claimed transition at log(L_bol/L_Edd) ~ -3.5 could shift or disappear.","fun_headline_variants_meta":{"raw":{"variants":["JWST finds AGN UV deficit below Eddington ratio -3.5","Accretion rate shift removes UV bump in LLAGN","Faint AGN lose high-energy photons at a critical rate","JWST reveals a UV-poor regime in low-luminosity AGN"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.000402,"raw_usage":{"total_tokens":1986,"prompt_tokens":851,"completion_tokens":1135,"prompt_tokens_details":{"cached_tokens":256},"prompt_cache_hit_tokens":256,"prompt_cache_miss_tokens":595,"completion_tokens_details":{"reasoning_tokens":1057}},"tokens_in":595,"tokens_out":1135,"duration_ms":10214,"temperature":1.0,"reasoning_tokens":1057,"cache_read_input_tokens":256,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-03T08:25:16.566193+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"A direct measurement of bolometric luminosity from multi-wavelength SEDs for a sample of LLAGN spanning the critical Eddington ratio, followed by a recomputation of the [Ne V]–X-ray relation, would settle the claim. If the break vanishes or moves by more than the quoted uncertainty when using SED-based L_bol, the claimed accretion-state transition would be falsified. Alternatively, finding a low-Eddington LLAGN with strong [Ne V] at fixed X-ray luminosity (i.e., no suppression) would already contradict the observed relation.","supporting_citations":[],"review_version":1}