{"id":"9435e3c4-2d98-4213-b409-db386c84e05a","arxiv_id":"2501.15930","paper_version":1,"verdict":"ACCEPT","confidence":"HIGH","novelty_score":6.0,"correctness_risk":"low","formal_verification":"none","parameter_count":1,"one_line_summary":"Six of 13 free-floating planetary-mass objects in IC348 have infrared excess from disks, giving a disk fraction of 46%, similar to heavier brown dwarfs.","lead":"By stacking 38 archived Spitzer images of the young cluster IC348, astronomers measured infrared colors of 13 free-floating objects near the planet/brown dwarf boundary and found that about 46 percent of them show signs of dusty disks. This adds to the evidence that objects as light as giant planets can form disks and keep them for several million years, possibly forming planets around themselves.","discovery_kind":"extension","skeptic_critique":{"model":"deepseek-v4-flash","headline":"The 46% disk fraction is computed from 13 of 23 M9+ members, with no check that the 10 objects lacking two-band IRAC coverage are representative; if they are not, the population estimate could shift from ~26% to ~70%.","rationale":"The photometric core of the paper is solid: the stacking is straightforward, the K−IRAC2 excess is the standard disk diagnostic, the external photospheric templates are appropriate, three of the six disk candidates have independent 5.8/8.0 um detections, and the individual SEDs are shown and discussed. The marginal case LRL5231 could lower the fraction to 5/13 = 38%, but that is within the quoted uncertainty and does not by itself threaten the conclusion. The load-bearing weakness is the leap from 13 measured objects to a population disk fraction for 23 known M9+ members. The paper never checks whether the 10 excluded objects differ in extinction, brightness, position, or disk content; the reader's weakest-assumption statement identifies the same issue. Because the central claim is explicitly population-level and is used to build the age trend in Figure 5, this selection uncertainty should be resolved before the 46% value is adopted as a secure datapoint. I therefore recommend CONDITIONAL rather than REJECT: the data and method support the disk detections, but the headline fraction should carry an explicit representativeness caveat or a completeness-corrected value.","tokens_in":12955,"tokens_out":7994,"duration_ms":84290,"concrete_test":"As a single decisive check, use Luhman et al. (2016) to compile the full 23-object M9+ census with K-band magnitude, spectral type, A_J, and the published disk classification for every object. Split the sample into the 13 objects with new two-band IRAC photometry and the 10 objects without, and run two-sample KS tests on K, A_J, and spectral type, plus a comparison of disk fractions using the Luhman classifications. If the two groups differ in disk fraction or in K/A_J/spectral type at more than about 1σ, the 6/13 estimate is not representative and the population disk fraction should be re-derived with completeness weights or explicitly restricted to the measured subset. If the groups are statistically indistinguishable, the representativeness concern is resolved.","verdict_should_be":"CONDITIONAL","load_bearing_attack":"The central claim is a population-level disk fraction: 6/13 = 46% for planetary-mass objects in IC348 (Section 3.4), and the comparison in Figure 5 uses this point to argue that PMO disks survive 3–4 Myr. The denominator is not the full M9+ census. Section 2.2 reports 23 M9+ members, 19 covered by the stacked IRAC images, 17 with valid measurements, and only 13 with both IRAC1 and IRAC2; four are IRAC1-only and six are not measured in both bands. Section 3.4 then computes 6/13 without a selection function or a comparison of the measured and unmeasured subgroups. This is not a purely cosmetic limitation: disk-bearing objects are systematically brighter at 4.5 um, so flux-dependent losses would bias the ratio upward, and the spatial offset between the IRAC1 and IRAC2 footprints could correlate with local disk fraction. The sensitivity is large: if the 10 missing objects were all disk-free, the population fraction would be 6/23 = 26%; if all had disks, it would be 16/23 = 70%. The quoted 34–59% interval accounts only for binomial sampling within the 13, not for this selection uncertainty. The paper states in Sections 2.2 and 4 that the core sample is the 13 measured objects, but it does not test representativeness, so the population-level conclusion is not yet secured.","agreement_with_reader":"agree"},"referee_report":{"model":"deepseek-v4-flash","summary":"The paper stacks 38 epochs of Spitzer/IRAC 1 and 2 time-series imaging of IC348 to produce deep 3.6 and 4.5 micron images, measures photometry for a sample of 13 planetary-mass candidate members (spectral type M9 or later) selected from the Luhman et al. (2016) census, dereddens the measurements using Wang & Chen (2019), and identifies infrared excess relative to BT-Settl photospheric templates. Six of the 13 objects are classified as disk-bearing, yielding a disk fraction of 6/13 = 46% with a binomial 1-sigma interval of 34-59%. The paper then compares this fraction with literature values for IC348 and other star-forming regions and concludes that free-floating planetary-mass objects retain disks for several million years, at rates comparable to brown dwarfs and low-mass stars.","tokens_in":13251,"tokens_out":5845,"duration_ms":57818,"significance":"If the measurement is taken at face value, it is a useful new data point in a sparsely sampled regime: disk fractions around objects below or near the deuterium-burning limit. The photometry is carefully described, the dereddening is explicitly documented, the SED fitting uses external templates, and the authors state that individual images were inspected for neighbors and edge effects. The paper also benefits from being a direct follow-up to Scholz et al. (2023), providing a consistency check across two clusters. The main caveats are the small sample size, the fact that one disk classification (LRL5231) is marginal, and the lack of a demonstrated representative relation between the 13 measured objects and the full 23-object planetary-mass census. Because the central claim is a population-level disk fraction and an age-evolution comparison, the selection issue is not cosmetic.","major_comments":[{"comment":"The population-level disk fraction is computed as 6/13 from the 13 M9+ objects with both IRAC1 and IRAC2 measurements, but this is a subsample of the 23 M9+ members in the Luhman et al. (2016) census. The paper does not test whether the 13 are representative of the full planetary-mass population. Excluded objects include four that are only covered in IRAC1 and sources outside the stacked footprints or near edges. The quoted 34-59% interval is binomial sampling within the 13 and does not include this selection uncertainty. The sensitivity is large: if the 10 missing objects are all disk-free, the population fraction would be 6/23 = 26%; if all have disks, it would be 16/23 = 70%. Please compare the included and excluded subsamples in K magnitude, A_J, and spatial position, and either restrict the population-level claim to the 13-object sample or propagate the selection uncertainty into the quoted disk fraction.","section":"Sections 2.2 and 3.4"},{"comment":"The classification of LRL5231 as disk-bearing rests on marginal excess in IRAC1 and substantial excess in IRAC2 relative to a 2000 K template, with photometric errors of 0.22-0.23 mag, and it is the one object for which the authors' classification disagrees with Luhman et al. (2016). Removing this object changes the central result from 6/13 to 5/13, i.e., from 46% to 38%. Please quantify the significance of the excess for this object (for example, the measured color relative to the template color plus its uncertainty) and state explicitly how the disk fraction changes if LRL5231 is excluded.","section":"Section 3.3, LRL5231"}],"minor_comments":[{"comment":"The abstract reports the disk fraction as both '46% (34-59%)' and '46±13 12%' in different places; please unify the notation for the uncertainties.","section":"Abstract and Section 3.4"},{"comment":"The table lists two entries with the same coordinates J034449.33+320949.4, one with spectral type M9 and no LRL number and one with LRL40023 and M9.5; please verify whether these are indeed two distinct sources or whether one coordinate is a typo.","section":"Table 1"},{"comment":"The x-axis label appears as 'log( ( m))' in the provided version; it should read as wavelength in microns, e.g., log(lambda/micron).","section":"Figure 3"},{"comment":"There are several typographical errors, including 'Adadditional' in Section 2.3, 'An trend' in Section 3.4, and the header 'MNRAS000, 1–7 (20252024)' with mismatched years; these should be corrected.","section":"Throughout"},{"comment":"The statement that the disk identification outcome does not depend on the specific choice of model temperature is plausible but not demonstrated; please state the range of effective temperatures tested and the resulting spread in the number of disk classifications.","section":"Section 3.2"}],"recommendation":"major_revision","confidential_remarks":"The paper is within the scope of the journal and the data-processing effort is solid. The main issue is the gap between the 13-object measured sample and the 23-object population-level claim. This is fixable with a representativeness check and a more cautious statement of scope, so I do not recommend rejection. The marginal LRL5231 classification should also be addressed because it is the single object that moves the headline fraction by 8 percentage points."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"This is a solid, modest paper that gives the first homogeneous disk census for M9+ planetary-mass objects in IC348, using deep stacked Spitzer IRAC data. The photometry is carefully described, errors are quantified, dereddening is documented, and the SED fits are checked against external templates (Sanghi et al. 2023, BT-Settl models). The comparison with literature disk fractions is transparent. The central measurement—6/13 = 46%, with a 34–59% binomial interval—is defensible for the sample actually measured. Three of those disk detections were already flagged by Luhman et al., but the census itself is new, and the PMO disk fraction vs. age sequence in Figure 5 is the first of its kind.\n\nThe main soft spot is the one the stress-test flagged: the 46% is computed for 13 of the 23 known M9+ members, and the paper does not test whether those 13 are representative. The loss is mostly spatial—IRAC1/IRAC2 field offset and edge effects—so it is not obviously biased, but it is not demonstrated. If the missing 10 objects were all disk-free the population fraction would be 26%; if all had disks, 70%. The paper should at least compare the measured and unmeasured subsamples in spectral type, extinction, and spatial distribution, or explicitly discuss why selection is unlikely to correlate with disk presence. This is a moderate caveat, not a fatal flaw, because the measurement for the 13-object sample is still internally valid and the comparison points in Figure 5 come from similar small samples.\n\nTwo minor issues: LRL5231 is classified as a disk based on marginal excess and is the one case where the authors disagree with Luhman et al. That is not unreasonable, but it deserves a closer look. Also, the abstract says \"plus three which may or may not have a disk,\" while the body concludes those three have SEDs largely consistent with a photosphere; the abstract overstates the ambiguity.\n\nThe central argument holds: PMO disk fractions at a few Myr are comparable to brown dwarfs and low-mass stars, and disks persist for several Myr. The paper is honest about the core sample and the limitations it sees. It is a useful incremental result for the substellar disk community and a good target list for JWST/ALMA follow-up.\n\nRecommendation: send it to peer review. It deserves a serious referee, and with minor revisions—mainly adding a representativeness check or a clearer statement of selection effects—it is a solid publication.","headline":"A careful, small-sample stacking analysis gives a 6/13 disk fraction for planetary-mass objects in IC348; the result is plausible and useful, though the paper should address how representative those 13 objects are of the full M9+ census.","tokens_in":13816,"tokens_out":2838,"would_cite":true,"duration_ms":27146,"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":"Six of 13 free-floating planetary-mass objects in IC348 host disks.","keywords":["free-floating planetary-mass objects","protoplanetary disks","disk fraction","IC348","brown dwarfs","infrared excess","Spitzer IRAC","disk lifetime"],"falsifier":"A complete infrared census of all 23 M9-or-later members of IC348, for example with JWST, that measures a disk fraction outside the 34-59% range would show the 13-object subsample was biased.","tokens_in":12741,"feed_emoji":"🪐","tokens_out":7706,"duration_ms":65880,"temperature":0.7,"pith_summary":"The paper uses stacked Spitzer infrared images of the young cluster IC348 to measure fluxes at 3.6 and 4.5 microns for 13 free-floating planetary-mass objects with spectral type M9 or later. By comparing their dereddened infrared colors and full spectral energy distributions with photospheric templates, the authors identify six objects with circumstellar disks, a disk fraction of 6/13 or 46% with a 1-sigma range of 34-59%. Combining this measurement with disk fractions in other young regions, the paper argues that planetary-mass objects keep their disks for at least 3-4 Myr, comparable to more massive brown dwarfs and low-mass stars. If correct, this implies that objects with masses comparable to giant planets form and evolve along the same disk-bearing path as stars.","feed_headline":"46% of free-floating planet-mass objects have disks","feed_subtitle":"Ultradeep Spitzer stacks of IC348 show these disks survive several million years, like brown dwarfs'.","key_machinery":"The method that carries the argument is the stacking of 38 archival Spitzer/IRAC time-series images per band, reaching a combined exposure of 456 seconds per filter and a depth about two magnitudes better than a single epoch. Disk identification uses the dereddened K-[4.5] colour compared with photospheric colour relations for late M and early L dwarfs, followed by spectral energy distribution fitting with BT-Settl model atmospheres. An object counts as having a disk only when the excess appears in at least two bands, with longer-wavelength detections at 5.8 and 8.0 microns serving as confirmation where available.","core_discovery":"The central claim is that free-floating planetary-mass objects in IC348, at an age of roughly 3-5 Myr, host disks at a rate of about 46%, statistically indistinguishable from the disk fractions of more massive brown dwarfs and low-mass stars in the same cluster. Six of the 13 objects with valid IRAC measurements show infrared excess in the K-[4.5] colour and in multi-band SED fits; three more are ambiguous and treated as non-detections. The paper concludes that disk fractions for planetary-mass objects stay in the 40-50% range for at least 3-4 Myr before dropping, so the disk lifetime at planetary masses is not strongly mass-dependent.","pith_inferences":["A natural test the paper does not run is to search for accretion signatures in the six disk hosts; if those disks accrete at measurable rates, they are actively evolving rather than passive remnant rings.","The representativeness of the 13 measured objects for the full 23-member census could be checked with a complete JWST survey; if the uncovered members differ systematically in extinction or binarity, the 46% value could be off.","If the disk fraction is truly flat across two orders of magnitude in mass, the clock for disk dissipation may be set more by the cluster environment than by the central object, a hypothesis the current data cannot yet distinguish."],"forward_implications":["The six disk-bearing planetary-mass objects in IC348 become specific targets for higher-resolution infrared and submillimetre follow-up aimed at measuring disk masses and structures at the lowest mass end.","The comparison across five star-forming regions implies that disk dissipation timescales are roughly constant from low-mass stars down to a few Jupiter masses.","A disk fraction near 50% in a 3-5 Myr cluster supports the picture in which many free-floating planetary-mass objects form like stars, rather than as ejected planets stripped of their disks.","Long-lived disks around planetary-mass objects raise the possibility that bodies resembling moons or sub-planets could assemble around objects that are themselves comparable to giant planets."],"supporting_citations":[{"why":"Supplies the cluster census, spectral types, extinction values, and K-band photometry that define the planetary-mass sample.","marker":"Luhman et al. (2016)"},{"why":"Establishes the stacking and disk-identification approach that this paper directly applies to IC348.","marker":"Scholz et al. (2023)"},{"why":"Provides the photospheric colour-magnitude relations used to predict the K-[4.5] colours of late M and L dwarfs.","marker":"Sanghi et al. (2023)"},{"why":"Offers the BT-Settl model atmospheres used as templates in the SED fitting.","marker":"Allard et al. (2007)"},{"why":"The C2D catalogue supplies the 5.8 and 8.0 micron fluxes that confirm disks for three objects.","marker":"Evans et al. (2009)"},{"why":"Provides the photometric catalogue used to calibrate the stacked IRAC magnitudes.","marker":"Gutermuth et al. (2009)"},{"why":"Gives earlier disk fraction measurements in IC348 for M-type brown dwarfs, serving as a comparison baseline.","marker":"Luhman et al. (2005)"}],"fun_headline_variants":["Half of free-floating planet-mass objects keep disks","IC348: 46% of planetary-mass free floaters host disks","Spitzer finds disks around half of planet-mass free floaters","Disk fraction ~46% for free-floating planets in IC348","Young free-floating planets keep disks as long as stars"],"cache_read_input_tokens":3200,"weakest_assumption_plain":"The 13 planetary-mass objects with valid measurements are assumed to be representative of all 23 M9-or-later members of IC348; if the ten objects without clean measurements preferentially have or lack disks, the 46% fraction would not be the population value.","fun_headline_variants_meta":{"raw":{"variants":["Half of free-floating planet-mass objects keep disks","IC348: 46% of planetary-mass free floaters host disks","Spitzer finds disks around half of planet-mass free floaters","Disk fraction ~46% for free-floating planets in IC348","Young free-floating planets keep disks as long as stars"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.000174,"raw_usage":{"total_tokens":1244,"prompt_tokens":866,"completion_tokens":378,"prompt_tokens_details":{"cached_tokens":384},"prompt_cache_hit_tokens":384,"prompt_cache_miss_tokens":482,"completion_tokens_details":{"reasoning_tokens":302}},"tokens_in":482,"tokens_out":378,"duration_ms":4182,"temperature":1.0,"reasoning_tokens":302,"cache_read_input_tokens":384,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-10T13:49:47.218418+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"A complete infrared census of all 23 M9-or-later members of IC348, for example with JWST, that measures a disk fraction outside the 34-59% range would show the 13-object subsample was biased.","supporting_citations":[],"review_version":1}