{"id":"ddfbc920-4221-44b4-ac19-540edfd43e42","arxiv_id":"2608.05072","paper_version":1,"verdict":"CONDITIONAL","confidence":"MODERATE","novelty_score":5.0,"correctness_risk":"medium","formal_verification":"none","parameter_count":0,"one_line_summary":"Characterization of the first science-grade, two-octave, dual-polarization TiN MKID array for Prime-Cam's 850 GHz module shows high fabrication yield, photon-noise-limited sensitivity, and optical efficiency near 75% under test loading.","lead":"The CCAT collaboration reports first cryogenic tests of the science-grade detector array for its 850 GHz camera module, a wafer with more than 10,000 superconducting titanium-nitride microwave kinetic inductance detectors. The early results indicate the two-octave multiplexing scheme works without degrading detector sensitivity, a key step toward deploying the full 38,000-detector focal plane on FYST in 2027.","discovery_kind":"extension","skeptic_critique":{"model":"deepseek-v4-flash","headline":"The photon-noise-limited and 75% optical-efficiency claim rests on only four representative resonators and a modelled benchmark, with no reported uncertainties; the central claim is not yet established by the presented data.","rationale":"The reader's verdict of CONDITIONAL matches my reading. The paper is an honest, preliminary instrument characterization with a plausible architecture, but the central claim—that the two-octave design does not degrade sensitivity and that photon-noise-limited performance is confirmed—is supported by only four representative resonators and a modelled efficiency benchmark. No uncertainties are given, and the optical-efficiency match rests on an assumed model rather than an independent measurement. These are exactly the load-bearing weaknesses identified by the reader. The paper itself acknowledges the preliminary nature and limited sampling, so the appropriate verdict is conditional acceptance with a request for expanded statistics, error bars, and a direct comparison to the photon-noise limit. My independent review did not uncover an additional, more serious flaw; the consistency of the qualitative trends in Figures 6 and 8 supports the decoupling of the inductor architecture, but that is a weaker claim than the one in the conclusions.","tokens_in":8498,"tokens_out":1241,"duration_ms":18090,"concrete_test":"Provide the white NEP values, with uncertainties, for at least one full network or a statistically meaningful sample of detectors, together with the photon-noise-limited NEP computed from the blackbody loading and the stated 75% optical efficiency. If the measured NEPs scatter around the photon-noise prediction within the reported uncertainties over a large fraction of the array, the central claim is supported; if not, the claim remains unverified.","verdict_should_be":"CONDITIONAL","load_bearing_attack":"The paper's strongest claim—'this work confirms the expected optical efficiency and photon-noise limited performance of the array'—is not actually supported by the data shown. Figure 7 (right) presents the extracted white NEP for four representative resonators, with no error bars, no sample statistics, and no explicit derivation of the 75% optical efficiency beyond a stated agreement with a model. Without a quantitative comparison of the measured NEP to the photon-noise limit (e.g., NEP_photon = sqrt(2 hν P)), the claim that TLS noise is subdominant remains a qualitative inference from the low-frequency separation between loaded and unloaded responses. The text itself repeatedly emphasizes that the results are preliminary and that only a small series of regions were sampled; the conclusion that the entire array is photon-noise-limited and that the inductor architecture is decoupled from resonator response is therefore an extrapolation from a handful of devices. The weakness is acknowledged in the Discussion ('Future work will expand the testing across larger regions') but the conclusions still overstate the current evidence. This does not invalidate the architecture, but the specific confirmatory claim needs more than four resonators.","agreement_with_reader":"agree"},"referee_report":{"model":"deepseek-v4-flash","summary":"The paper reports the first cryogenic characterization of a science-grade MKID array fabricated for the CCAT Prime-Cam 850 GHz module. The array uses a two-octave frequency plan in which coarse tuning is provided by selectively shorted inductor geometries and fine tuning by interdigitated capacitor variations, with the goal of multiplexing ~38,000 detectors across three arrays. The authors present S21 sweeps of all 12 networks, LED-based resonator mapping, optical loading curves, quality factor measurements, and NEP estimates from a few representative resonators. The main claims are: (i) a fabrication yield above 90% (the abstract states 99%); (ii) the shorted and non-shorted inductor architectures are decoupled from resonator performance; and (iii) the measured optical efficiency matches the modelled 75% value and the detectors are photon-noise limited. The paper explicitly labels these as preliminary results, and the conclusions state that future work will expand testing to larger regions.","tokens_in":8638,"tokens_out":3030,"duration_ms":37449,"significance":"If the claims hold, the two-octave coarse-short tuning strategy would be a significant enabling step for the 38,000-detector 850 GHz module, increasing multiplexing density without sacrificing detector sensitivity. The paper's strength is that it characterizes a fully fabricated science-grade array, not just test structures: the S21 survey of 12 networks, the LED mapping, and the controlled optical loading measurements are valuable direct data. However, the load-bearing confirmatory claims—decoupling of the inductor architecture, optical efficiency, and photon-noise-limited performance—rest on four representative resonators and on comparison to the authors' own modelled efficiency, with no reported uncertainties. As a milestone report the paper is useful, but as a confirmation of the architecture it currently under-delivers relative to its conclusions.","major_comments":[{"comment":"The abstract states a \"fabrication yield of 99%\", while §4.1 explicitly says that after the loss of the ninth network due to a scratch on the feedline, \"a greater than 90% fabrication yield was achieved\". A 99% yield with one entirely nonfunctional network is internally inconsistent unless the definition of yield excludes whole networks. The yield claim is load-bearing for the fabrication-success narrative, so the definition and the exact number must be reconciled.","section":"Abstract vs. §4.1"},{"comment":"The optical-efficiency confirmation is not quantitatively supported. The right panel of Fig. 7 shows extracted white NEP values for four representative resonators with no error bars, no sample statistics, and no explicit comparison to the photon-noise limit (e.g., NEP_photon = sqrt(2 hν P) or an equivalent expression). The statement that the measured NEP \"matches the expected modelled optical efficiency of the array of 75%\" uses the authors' own model as the benchmark, so it is not an independent test of the optical design. A quantitative derivation of optical efficiency from NEP, with an uncertainty budget, is needed before the claim \"This result confirms both the detectors and feedhorns work as expected\" can be accepted.","section":"§4.2, Fig. 7 (right)"},{"comment":"The central conclusions overstate the evidence. The claim that \"the inductor architecture is decoupled from the response of the resonator\" is supported by four representative resonators in Fig. 6 and Fig. 8, with no quantitative comparison of resonance depth, linewidth, Qi, or NEP distributions between shorted and non-shorted populations. Likewise, \"this work confirms the expected optical efficiency and photon-noise limited performance of the array\" is an extrapolation from four resonators to the entire array, even though the text acknowledges in §5 that \"Future work will expand the testing across larger regions\". At minimum, the sample size, the number of networks represented, and the scatter across those resonators must be reported and reflected in a more cautious conclusion.","section":"§4.1, §4.2, §6"},{"comment":"The performance limitations reported in §5 are in tension with the photon-noise-limited claim. The text states that Qi drops below the desired value of 15,000 even before reaching 70 pW of loading and that Qc is 1.5 times higher than expected; Fig. 6 shows that under loading the resonators become shallow, making the readout more susceptible to amplifier noise. If the resonators are not in the expected matched-coupling regime, the white NEP measured on four resonators may not be representative of the array's operating condition, and the photon-noise-limited assertion needs to account for the amplifier noise contribution under the achieved Qi and Qc values.","section":"§5, Fig. 8, Fig. 6"}],"minor_comments":[{"comment":"The heading \"Two-Octave F requency Planning\" contains an extra space in \"F requency\"; please correct the typo.","section":"§2.2 heading"},{"comment":"The caption contains an incomplete sentence: \"The 850 GHz in a testing configuration with the blackbody and filters defines the incident power on a given sample region of the array.\" Please rephrase for clarity.","section":"Fig. 4 caption"},{"comment":"The left panel of Fig. 7 shows only two MKIDs, one shorted and one non-shorted, and the claim that TLS noise is subdominant is based on the visual separation between loaded and unloaded curves at low frequency. Please add a quantitative fit or bound for the TLS contribution, or state the statistical basis for this inference.","section":"§4.2, Fig. 7"},{"comment":"The text refers to \"a small series of regions\" sampled due to blackbody size limits, but it is not stated how many regions or how many networks are represented in Figs. 6–8. Please report the sampled region count and the network coverage explicitly.","section":"§3, §4.1"},{"comment":"The statement that \"more than 10,000 submillimeter MKIDs were successfully produced on a single silicon wafer, the most to date\" lacks a supporting reference or quantitative comparison to earlier arrays. Please either provide a citation or soften the claim.","section":"§6"}],"recommendation":"major_revision","confidential_remarks":"This is a proceedings-style early report with useful direct measurements of a science-grade array. The main issue is that the abstract and conclusions claim confirmations (yield, optical efficiency, photon-noise-limited performance, architecture decoupling) that the data do not yet quantitatively support: only four representative resonators are used, no uncertainties are given, and the optical-efficiency reference is the authors' own model. I would encourage the authors to present the current results as a preliminary milestone and to adjust the conclusions accordingly, or to add the missing statistical and quantitative analysis. The yield discrepancy between abstract and text should also be resolved. These are fixable within the manuscript's scope."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"Hi [Name],\n\nBottom line: this is a genuine milestone for the CCAT 850 GHz detector program — first full science-grade array, more than 10,000 MKIDs on a wafer, LED mapping, and a two-octave coarse-short architecture working at scale. The central physics claim, that shorted-inductor resonators behave like non-shorted ones, is plausible and is supported by consistent resonance shapes, Qi-under-loading curves, and NEP for a handful of devices. That is real progress and worth a careful read by anyone building large MKID arrays.\n\nThe soft spots are the usual ones for a first-look instrument paper, but they matter here because the conclusions overstate what the evidence shows. The abstract says 99% yield; the text reports \"greater than 90%\" after losing network 9 to a feedline scratch. Pick one and explain the discrepancy. Optical efficiency is confirmed against the team's own 75% model, not an independent measurement, and the \"confirms both the detectors and feedhorns work as expected\" rests on four resonators without error bars. The paper explicitly says the results are preliminary and that testing will expand, so the strong \"confirms\" language in the conclusions goes beyond what is shown. That's fixable: add sample statistics, propagate uncertainties, and soften the confirmations to \"consistent with\" until the larger sample is in.\n\nAlso, the two-octave architecture itself is from prior SPIE/IEEE work; the new thing is the science-grade wafer and the first cryogenic readout of it. The citation pattern is appropriate — they cite their own prior design papers and the NIST fabrication work, and that's fair.\n\nWho should read this: detector people and CCAT collaborators. It is not a broad-audience paper, but it is exactly the kind of incremental, necessary characterization that the submillimeter community builds on. I would send it to a serious referee; it needs a moderate revision, not a desk reject. If the revision adds a few more resonators, puts error bars on the NEP and Qi plots, fixes the yield number, and pulls the conclusions back to what the data actually supports, this will be a useful reference for the field.\n\nRecommendation: accept after minor-to-moderate revision, with a referee who knows MKID noise and is asked to check the photon-noise accounting and the statistical basis for the 75% efficiency claim.","headline":"First science-grade 850 GHz MKID array shows promising yield and a plausible two-octave architecture, but the confirming claims outrun the data (four resonators, no error bars, model-referenced efficiency).","tokens_in":9296,"tokens_out":2337,"would_cite":false,"duration_ms":29707,"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":"The first science-grade MKID array for the Prime-Cam 850 GHz module meets its design targets, showing that its two-octave frequency-tuning scheme decouples frequency placement from detector response.","keywords":["microwave kinetic inductance detectors","MKID arrays","submillimeter astronomy","two-octave frequency multiplexing","titanium nitride detectors","detector array characterization","optical efficiency","photon-noise-limited sensitivity"],"falsifier":"Measure the same sampled detector regions under two or more well-separated blackbody temperatures and compute the optical efficiency from the change in loading without assuming the modelled passband; if the inferred efficiency disagrees with 75 percent by more than the model uncertainty, the optical-efficiency claim fails. Alternatively, a full-array scan in which shorted and non-shorted resonators separate in internal quality factor or noise under 70 pW loading would directly falsify the decoupling claim.","tokens_in":8253,"feed_emoji":"🔭","tokens_out":8801,"duration_ms":98484,"temperature":0.7,"pith_summary":"This paper reports the first characterization of a science-grade detector array for the 850 GHz module of Prime-Cam, the highest-frequency camera module for a 6-meter submillimeter telescope at a 5,600-meter site. The array is built from dual-polarization titanium-nitride microwave kinetic inductance detectors (MKIDs), with more than 10,000 resonators on one wafer, and uses a two-octave frequency plan: coarse tuning by shorting inductor lines plus fine tuning with interdigital capacitors. The central claim is that this frequency-expanding architecture is decoupled from detector response, so shorted and non-shorted detectors show the same optical efficiency and photon-noise-limited sensitivity. If that holds, the planned 38,000-detector focal plane can be read out through practical RFSoC-based electronics, and the module remains on track for 2027 deployment.","feed_headline":"Prime-Cam 850 GHz array passes first science-grade tests","feed_subtitle":"Two-octave readout shows no sensitivity penalty; measured noise matches the 75% optical-efficiency target.","key_machinery":"The device that carries the argument is the two-octave resonator frequency architecture, a scheme for spreading MKID resonance frequencies over a roughly two-octave band instead of a single narrow range. Coarse tuning is handled by shorting pairs of inductor lines, which lowers the inductance and moves the resonance into one of several bands while leaving the absorber geometry intact; fine tuning is handled by adjusting the interdigital capacitor. This lets each readout network hold more than 1,000 resonators without changing the optical design. The proof strategy is comparative: if the four variants formed by crossing two capacitor sizes with shorted and non-shorted inductors behave identically under loading, then frequency placement is not affecting detector physics.","core_discovery":"On the paper's own terms, the first science-grade array works as designed. Sweeps of all 12 readout networks place more than 1,000 resonators per network across two octaves, with a fabrication yield above 90 percent despite a scratch on one network's feedline. Under optical loading stepped up to the expected 70 pW, the four detector variants—two capacitor sizes crossed with shorted and non-shorted inductors—shift in frequency together, retain comparable resonance depth and linewidth, and follow a single internal-quality-factor trend, which the authors take as confirmation that the inductor architecture is decoupled from the resonator response. The measured white noise equivalent power matches the modelled 75 percent optical efficiency of the test setup, and the low-frequency noise shows photon noise dominating two-level-system noise. The conclusion is that both the detectors and feedhorns perform as expected, clearing the way for the full module.","pith_inferences":["Beyond the paper: if the decoupling holds across the full wafer, the same coarse/fine tuning strategy could likely be pushed beyond two octaves, further cutting the number of cryogenic radio-frequency chains needed for even larger future arrays.","The observed coupling quality factor being 1.5 times higher than expected suggests that post-fabrication trimming, which the paper already plans, could restore loaded quality factors above the 15,000 target and bring shallow-resonator operation back into a comfortable readout regime.","A more direct test of the optical-efficiency claim than matching one modelled number would be measuring the NEP at two or more blackbody temperatures and deriving efficiency from the slope of loading versus response, removing reliance on a single passband model.","The paper's sampled regions cover only part of the 12 networks; extending the same LED mapping and optical tests to the full array would reveal whether the scratched feedline and any other wafer-level defects affect yield and uniformity estimates."],"forward_implications":["The full Prime-Cam 850 GHz focal plane, with roughly 38,000 detectors across three arrays, can be read out with only 12 networks per array because each network holds more than 1,000 resonators.","The RFSoC readout chain, designed for the two-octave band, does not need detector-performance compromises to achieve its multiplexing factor.","The test-setup optical efficiency of 75 percent, together with the module's different filter passband, projects to greater than 90 percent efficiency at the deployment band.","All four detector architectures can be calibrated and operated with a single common readout and analysis treatment, since their response and noise are statistically indistinguishable.","If the array's performance is representative, the module is on schedule to begin observing in 2027."],"supporting_citations":[{"why":"Defines the densely packed dual-polarization pixel and two-octave geometry that the science-grade array implements.","marker":"[13]"},{"why":"Demonstrates the high-density photon-noise-limited multi-octave detector concept that this array scales to full fabrication.","marker":"[16]"},{"why":"Describes the two-octave RFSoC readout that the 1,000-plus detectors-per-network multiplexing plan depends on.","marker":"[15]"},{"why":"Provides the expected on-site loading of 70 pW and the atmospheric transmission used to define the optical test conditions.","marker":"[8]"},{"why":"Establishes the proximity-coupled Ti/TiN multilayer deposition process used to fabricate the detector stack.","marker":"[24]"},{"why":"Supplies the LED mapping method that assigns physical positions to resonance frequencies so optical tests target beam-filled detectors.","marker":"[26]"},{"why":"Identifies two-level-system noise as the low-frequency noise source that must be shown subdominant to photon noise.","marker":"[27]"},{"why":"Provides the internal-quality-factor model used to fit detector loss versus optical loading.","marker":"[28]"}],"fun_headline_variants":["First science-grade MKID array hits 99% yield for 850 GHz","Two-octave MKID array passes noise test for 850 GHz","Prime-Cam 850 GHz array: 99% yield, noise on target","850 GHz MKIDs: two-octave readout, on-target noise","First 850 GHz MKID array ready for 2027 deployment"],"cache_read_input_tokens":3200,"weakest_assumption_plain":"The claim that the array has the expected optical efficiency rests on matching the measured white noise to a modelled 75 percent efficiency for the test setup, so an inaccurate optical model, blackbody, or filter passband would leave the claim without independent support; the paper also assumes the few sampled regions represent all 12 networks.","fun_headline_variants_meta":{"raw":{"variants":["First science-grade MKID array hits 99% yield for 850 GHz","Two-octave MKID array passes noise test for 850 GHz","Prime-Cam 850 GHz array: 99% yield, noise on target","850 GHz MKIDs: two-octave readout, on-target noise","First 850 GHz MKID array ready for 2027 deployment"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.000712,"raw_usage":{"total_tokens":3252,"prompt_tokens":1042,"completion_tokens":2210,"prompt_tokens_details":{"cached_tokens":384},"prompt_cache_hit_tokens":384,"prompt_cache_miss_tokens":658,"completion_tokens_details":{"reasoning_tokens":2111}},"tokens_in":658,"tokens_out":2210,"duration_ms":19160,"temperature":1.0,"reasoning_tokens":2111,"cache_read_input_tokens":384,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-06T05:56:05.030782+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"Measure the same sampled detector regions under two or more well-separated blackbody temperatures and compute the optical efficiency from the change in loading without assuming the modelled passband; if the inferred efficiency disagrees with 75 percent by more than the model uncertainty, the optical-efficiency claim fails. Alternatively, a full-array scan in which shorted and non-shorted resonators separate in internal quality factor or noise under 70 pW loading would directly falsify the decoupling claim.","supporting_citations":[{"cited_title":"CCAT: design and performance of densely packed, high-frequency, dual-polarization kinetic inductance detectors for the Prime-Cam 850 GHz module,","cited_arxiv_id":null,"evidence_quote":"Defines the densely packed dual-polarization pixel and two-octave geometry that the science-grade array implements."},{"cited_title":"High-density photon-noise-limited multi-octave submillimeter kinetic inductance de- tectors for the prime-cam 850 ghz module,","cited_arxiv_id":null,"evidence_quote":"Demonstrates the high-density photon-noise-limited multi-octave detector concept that this array scales to full fabrication."},{"cited_title":"CCAT: a two-octave 1.024-GHz-KID readout featuring an overlap-channel polyphase syn- thesis filter bank on RFSoC","cited_arxiv_id":null,"evidence_quote":"Describes the two-octave RFSoC readout that the 1,000-plus detectors-per-network multiplexing plan depends on."},{"cited_title":"CCAT-prime Collaboration: Science Goals and Forecasts with Prime-Cam on the Fred Young Submillimeter Telescope,","cited_arxiv_id":null,"evidence_quote":"Provides the expected on-site loading of 70 pW and the atmospheric transmission used to define the optical test conditions."},{"cited_title":"Proximity-coupled ti/tin multilayers for use in kinetic inductance detectors,","cited_arxiv_id":null,"evidence_quote":"Establishes the proximity-coupled Ti/TiN multilayer deposition process used to fabricate the detector stack."},{"cited_title":"CCAT: LED mapping of the 850GHz instrument module for Prime-Cam on FYST","cited_arxiv_id":null,"evidence_quote":"Supplies the LED mapping method that assigns physical positions to resonance frequencies so optical tests target beam-filled detectors."},{"cited_title":"Two-level system noise reduction for Microwave Kinetic Inductance Detectors,","cited_arxiv_id":null,"evidence_quote":"Identifies two-level-system noise as the low-frequency noise source that must be shown subdominant to photon noise."},{"cited_title":"PhD thesis, California Institute of Tech- nology (2008)","cited_arxiv_id":null,"evidence_quote":"Provides the internal-quality-factor model used to fit detector loss versus optical loading."}],"review_version":1}